Researchers at University College London have identified a biological process that helps the body shut down inflammation once it is no longer needed. Inflammation is an essential defense mechanism that protects us from infection and injury. Until now, scientists did not clearly understand how the body transitions from an active immune attack to a healing phase. These molecules help prevent the buildup of specific immune cells called intermediate monocytes*, which are associated with chronic inflammation -- linked to tissue damage, illness and disease progression. To explore this process, researchers conducted a carefully controlled experiment in healthy volunteers. Participants received a small injection of UV-killed E. coli bacteria in the forearm. This triggered a temporary inflammatory response -- pain, redness, heat and swelling -- similar to what occurs after infection or injury. This medication blocks an enzyme known as soluble epoxide hydrolase (sEH), which normally breaks down epoxy-oxylipins. They were treated two hours before inflammation began to test whether boosting epoxy-oxylipins early could prevent harmful immune changes. This approach reflected how treatment would occur in real world settings once symptoms appear. Notably, the medication did not meaningfully change visible symptoms such as redness or swelling. Further investigation showed that one specific epoxy-oxylipin, 12,13-EpOME, works by suppressing a protein signaling pathway known as p38 MAPK, which drives monocyte transformation. Laboratory experiments and additional testing in volunteers who received a p38 blocking drug confirmed this mechanism. First author Dr. Olivia Bracken (UCL Department of Ageing, Rheumatology and Regenerative Medicine) said: "Our findings reveal a natural pathway that limits harmful immune cell expansion and helps calm inflammation more quickly. "Targeting this mechanism could lead to safer treatments that restore immune balance without suppressing overall immunity. Corresponding author Professor Derek Gilroy (UCL Division of Medicine) said: "This is the first study to map epoxy-oxylipin activity in humans during inflammation. He added: "This was an entirely human-based study with direct relevance to autoimmune diseases, as we used a drug already suitable for human use -- one that could be repurposed to treat flares in chronic inflammatory conditions, an area currently bereft of effective therapies." Scientists chose to investigate epoxy-oxylipins because previous animal research suggested they can reduce inflammation and pain. However, their role in human biology had not been clearly defined. Unlike well known inflammatory signals such as histamine and cytokines, epoxy-oxylipins belong to a lesser studied pathway that researchers believed might help naturally quiet the immune system. The findings open the possibility of clinical trials to test sEH inhibitors as treatments for diseases such as rheumatoid arthritis and cardiovascular disease. sEH inhibitors could be trialled alongside existing medications to investigate if they can help prevent or slow down joint damage incurred by the condition." Dr. Caroline Aylott, Head of Research Delivery at Arthritis UK, said: "The pain of arthritis can affect how we move, think, sleep and feel, along with our ability to spend time with loved ones. "We are excited to see the results of this study which has found a natural process that could stop inflammation and pain. We hope in the future that this will lead to new pain management options for people with arthritis." *Intermediate monocytes are white blood cells that help fight infection and repair tissue. After Decades of Global Searching, Scientists Finally Create the Silicon Aromatic Once Thought Impossible Stay informed with ScienceDaily's free email newsletter, updated daily and weekly. Keep up to date with the latest news from ScienceDaily via social networks: Tell us what you think of ScienceDaily -- we welcome both positive and negative comments.

You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript. While we have previously shown that a first pregnancy changes women's brain structure and resting-state brain activity, it is currently unknown how a woman's brain is transformed when she undergoes another pregnancy. Therefore, we performed a prospective pre-conception cohort study involving 110 women, including women who became pregnant of their second (PRG2) or first child (PRG1) and nulliparous women. Multimodal MRI data were acquired and differential changes between PRG2 and PRG1 were observed in grey matter volume, white matter tracts and functional neural network organization. Together, these results show similar but less pronounced structural and functional changes in the default mode and frontoparietal network in PRG2, suggesting a primary adaptation of these networks in first-time mothers that is further fine-tuned across a second pregnancy. Furthermore, stronger alterations were found in PRG2 in the dorsal attention and somatomotor network including the corticospinal tract, pointing to an enhanced plasticity within these externally-oriented networks. Neurostructural changes in both groups related to mother-infant attachment and peripartum depression. These findings show that a second pregnancy uniquely changes a woman's brain, entailing both convergent and distinct neural transformations. Pregnancy represents a monumental phase in many women's lives, orchestrated by unparallelled physiological and neuroendocrine changes that influence all major bodily systems. We have previously shown that reproduction is associated with widespread changes in women's brain structure1,2, which has since been replicated in various other studies across the world3,4,5. Additionally, we have shown that pregnancy also affects functional neural network organization in the default mode network2. While research is beginning to uncover pregnancy-induced brain plasticity, all studies so far have focused solely on first-time mothers, leaving it unclear whether similar brain changes occur as a result of subsequent pregnancies. Animal studies point to an association between subsequent reproductive experiences and neuroanatomical variations in the brain. Hippocampal dendritic morphology, neurogenesis and protein levels have been shown to differ between primiparous and multiparous female rodents6,7,8. Additionally, parity has lasting effects on synaptic plasticity and inflammation in the hippocampus and the response to ovarian hormones in middle-aged rodents9,10,11,12. While these findings provide evidence for structural variations in the rodent brain on the basis of parity, research on this topic in human mothers is still in its infancy. Recent studies have suggested that parity may involve long-lasting neural changes in human mothers measured beyond the postpartum period. In middle-aged women, parity has been linked to brain age, with multiparous women having younger-looking brains compared to primiparous and nulliparous women13,14. Additionally, parity has been associated with cortical thickness, functional connectivity and gray matter volume in women during late life15,16,17. Parity may also modulate the risk for and effects of various brain disorders that may occur later in life, such as Alzheimer's Disease, stroke and traumatic brain injury18,19,20,21. Furthermore, links between parity and cognition have been shown22, characterized by a U-shaped effect of parity on cognition, with the highest scores on the Mini Mental State Examination in women with one to four pregnancies, compared to nulliparous women and women with grand parity23. Although these findings provide evidence for long-lasting differences in neural anatomy in relation to multiple childbirths in human mothers, it remains unclear whether a first and second pregnancy differentially affect the human brain. Therefore, we performed a prospective pre-conception cohort study involving second-time mothers, first-time mothers and nulliparous control women to investigate whether a second pregnancy is associated with structural and functional changes in the brain. We acquired multimodal 3 T magnetic resonance imaging (MRI) data, involving high-resolution anatomical MRI, resting-state functional MRI, diffusion-weighted MRI and magnetic resonance spectroscopy (MRS), before pregnancy and in the early and late postpartum period in 30 women undergoing a second pregnancy (PRG2; becoming multiparous between the scans) and in 40 women undergoing a first pregnancy (PRG1; becoming primiparous between the scans). We also included a control group of 40 nulliparous control women (CTR), who underwent MRI scanning at a similar time interval. We assessed the differential effects of a second pregnancy on brain structure, functional neural networks, white matter organization and neural metabolite concentrations. This study demonstrates that a second pregnancy changes a woman's brain and uniquely impacts its gray matter structure, neural network organization and white matter tracts. To investigate anatomical brain changes across a first and second pregnancy compared to control women, we performed vertex-wise longitudinal analyses of cortical volume, thickness and surface area in Freesurfer. Generalized linear models (GLMs) comparing the volume change from pre-pregnancy (PRE) to the early postpartum period (POST) between multiparous and control women demonstrated widespread volumetric decreases across a second pregnancy (Fig. The median percentage volume decrease across these significant vertices in the left and right hemisphere was 2.8%. Similarly, women who were pregnant for the first time showed significant volumetric decreases from PRE to POST compared to control women (Fig. The areas of significant change covered a 79% larger part of the brain than in second-time mothers, and the median percentage of change was 3.1% within these significant vertices. Effect sizes for these volumetric decreases across a first and second pregnancy were large (Cohen's D > 1.0; Supplementary Table 1). In both groups, there were no significant increases in cortical volume across pregnancy compared to control women. Similar results were found when investigating cortical thickness (Supplementary Fig. 1a) and surface area (Supplementary Fig. 1b), showing decreases in cortical thickness and surface area across a first and second pregnancy when compared to control women. Vertex-wise analyses showing brain areas in which the cortical volume decreased from pre-pregnancy (PRE) to early postpartum (POST) in second-time mothers (PRG2; n = 30; a) and in first-time mothers (PRG1; n = 40; c) versus control women (CTR; n = 40). Extracted total volumes of all the significant vertices are shown in b (PRG2-CTR) and d (PRG1-CTR), with *p < 0.001 for two-sided paired t-tests (PRG2-CTR) or two-sided Wilcoxon signed rank tests (PRG1-CTR) within groups. When directly comparing cortical volume changes across a first and second pregnancy using a permutation-based correction for multiple comparisons, we found several areas of significantly different volumetric change between PRG1 and PRG2 (Supplementary Fig. 2), although it should be noted that these effects do not surface when applying an FDR correction as in the comparisons with CTR women. Effect sizes for these differences between a first and second pregnancy were smaller than the very large effect size of a pregnancy itself, but the contrast for PRG1 vs. PRG2 still showed a moderate to high effect size (Cohen's D > 0.5; Supplementary Table 1). A visual inspection of the results also shows a divergence in both the extent and localization of brain changes between these groups (Fig. This divergence between brain changes across a first and second pregnancy was supported by a multivariate classification approach. We performed a multivariate pattern classification analysis in PRonTo24,25 (v3.0) on the gray matter volume difference maps resulting from the longitudinal symmetric diffeomorphic modeling pipeline implemented in SPM1226. This analysis demonstrated that, based on the gray matter volumetric changes in the brain, 80% of women could be correctly classified as becoming primiparous or multiparous in between the PRE and POST session (PRG1 vs. PRG2; p = 0.0001 after 10,000 permutations; Fig. When classifying women getting pregnant and women not getting pregnant in between these sessions, the accuracy is even higher (PRG2 vs. CTR: 87%, p = 0.0001, Fig. Using k-folds cross-validation with ten folds to reduce the risk of overfitting, classification accuracy dropped slightly, but was still significant for all three combinations of groups (PRG1 vs. PRG2: accuracy = 70%, p = 0.009; PRG2 vs. CTR: accuracy = 87%, p = 0.0001; PRG1 vs. CTR: accuracy = 91%, p = 0.0001). Adding age as a regressor to the models involving groups with age differences rendered highly similar results, both with leave-one-out cross-validation (PRG1 vs. PRG2: accuracy = 81.43%, p < 0.0001; PRG2 vs. CTR: accuracy = 84.29%, p < 0.0001) and with ten-folds cross-validation (PRG1 vs. PRG2: accuracy = 74.29%, p = 0.0008; PRG2 vs. CTR: accuracy = 82.86%, p < 0.0001.) Multivariate pattern classification analysis of women becoming pregnant for the first (n = 40) or second time (n = 30) based on brain volume difference maps (a). On the left side the scatterplot of the classification results (balanced accuracy 80%), showing function values; dashed line is the function value cut-off between groups, using a leave-one-out cross validation with permutation testing (10,000 permutations; p = 0.0001) to correct for multiple comparisons. Similarly, the multivariate classification between control women (n = 40) and women getting pregnant for the second time (b; balanced accuracy 87%; 10,000 permutations; p = 0.0001) and for the first time (c; balanced accuracy 94%; 10,000 permutations; p = 0.0001). In the affected regions, brain volumes slightly increased in the late postpartum period (POST1) but had not returned to pre-pregnancy levels in both multiparous and primiparous women (Supplementary Fig. Correlation analyses between the volumetric change across pregnancy and age of the first child at the PRE session in the PRG2 group suggested that time since birth of the first child does not influence the amount of volumetric brain changes across a second pregnancy (Supplementary Fig. We also tested whether there were baseline differences in cortical volumes between groups, showing no significant differences between PRG1, PRG2 and CTR women in cortical volumes at the pre-pregnancy baseline. Extracting gray matter volumes inside the area of change across a first pregnancy, suggests a partial volume reversal across the postpartum period that seems to continue after the first postpartum year (Supplementary Fig. We then assessed the relationship between the volumetric brain changes across a second or first pregnancy and maternal behavior and mental health status. When comparing PRG1 and PRG2, we did not detect any significant differences in maternal behavior and mental health status (Supplementary Table 3), although we found a trend towards more depressive feelings during pregnancy in the PRG2 group. Correlation analyses showed that maternal behavior during pregnancy and in the early postpartum period was negatively correlated with the amount of volumetric change across pregnancy, except for the postpartum bonding questionnaire (PBQ) which showed a positive correlation due to its reversed scoring where higher values represent lower maternal behavior (Supplementary Fig. 6), with stronger changes in brain structure thus being associated with higher levels of maternal behavior on the employed measures. These correlations between brain volumetric changes and maternal behavior seem to be more widespread across a first compared to a second pregnancy (Supplementary Fig. While the volumetric changes across a first pregnancy were associated with maternal-fetal attachment as measured with the maternal antenatal attachment score (MAAS) and prenatal attachment inventory (PAI), this was not evident in second-time mothers. Associations between volumetric changes and measures of postpartum mother-infant attachment and impairments in the mother-infant relationship as measured with the maternal postnatal attachment score (MPAS) and PBQ were observed in both PRG1 and PRG2, but these were more widespread in PRG1. Nesting behavior was similarly associated with mothers' brain changes across a first or a second pregnancy. Regarding maternal mental health, we found associations between the observed brain changes with peripartum depression, measured with the Edinburgh Postnatal Depression Score (EPDS) and psychological distress measured with the K10 in both PRG groups, with less pronounced brain changes linked to more depressive complaints for most significant vertices. However, more widespread correlations were observed between volumetric change and depression and psychological distress in the PRG2 group during pregnancy, but in the PRG1 group in the early postpartum period (Supplementary Fig. To further investigate the regional distribution of changes in gray matter structure occurring across a first and second pregnancy, we assessed the differences and similarities between the changes in these groups in relation to the brain's functional networks using a cluster-wise approach (for cluster-wise results, see Supplementary Tables 6–9). Therefore, a combination of clusters resulting from the two significant contrasts (PRG2 vs. CTR and PRG1 vs. CTR) were applied on the volumetric GLM results described above. This resulted in three different areas of interest: Areas that are affected in both first and second pregnancies (overlap), areas that are only affected during a first pregnancy (only CTR-PRG1) and areas that are only affected during a second pregnancy (only CTR-PRG2) (Fig. Combining the cluster-corrected vertex-wise analyses of the multiparous (PRG2; n = 30) vs. control women (CTR; n = 40) and primiparous (PRG1; n = 40) vs. CTR resulted in three areas of interest: areas affected in both first and second pregnancies (Overlap; yellow), areas only affected in a first pregnancy (Only CTR-PRG1, magenta) and areas only affected in a second pregnancy (Only CTRL-PRG2; green) (a). We localized these areas according to the seven resting-state networks of Yeo27 (b). After calculating the expected overlap based on a random distribution across the gray matter in the brain, we calculated the observed/expected ratio for each area for each functional network (c). The horizontal line represents an observed/expected overlap of 1, meaning that the observed intersection of the area is similar as when there would be a random distribution across the brain. Spider plots representing the distribution across the seven Yeo networks for the three areas of interest (overlap: yellow; Only CTR-PRG1: magenta; Only CTR-PRG2: green) (d). To acquire more information on the localization of these areas across the brain, we then determined their intersection with the seven resting-state networks of Yeo27 (Fig. These analyses showed that the overlapping areas in first and second pregnancies were mostly located in the default mode network (DMN), followed by the frontoparietal and ventral attention network. Interestingly, the additional areas that were specifically affected only across a first pregnancy were localized in those same networks (Fig. In comparison, the areas affected in a second pregnancy only were not localized in the default mode network but instead were mainly located in the somatomotor and dorsal attention network (Supplementary Table 12 and Fig. To study functional network organization, we also analyzed resting-state functional MRI data, but we did not find significant changes in within-network coherence between a first or second pregnancy in any of the identified resting-state networks. Subsequent analyses using a default mode network (DMN) region that represents the area of significant change in functional coherence across a first pregnancy2, showed a significant group*session interaction effect between PRG1 and PRG2 in the DMN (MNI coordinates (x y z) = 12 −84 15, T = 2.85, p = 0.006 FWE-corrected; Supplementary Fig. 8), characterized by an increase in DMN network coherence only across a first pregnancy. Results for between-network connectivity are presented in the supplement (Supplementary Tables 13–15). We also investigated the effects of a first and second pregnancy on the organization of white matter tracts in the brain, measured with diffusion-weighted MRI. Based on these diffusion-weighted images, we extracted mean fractional anisotropy (FA) and mean diffusivity (MD in 10−5 mm2/s) in 11 large white matter tracts across the brain (Fig. We investigated the white matter organization of eleven large white matter tracts in the brain (a). Extracted fractional anisotropy (FA) and mean diffusivity (MD; in 10−5 mm2/s) in white matter tracts that showed significant changes from pre-pregnancy (PRE) to the early postpartum period (POST) across first (PRG1; n = 40) and/or second (PRG2; n = 30) pregnancies and control women (CTR; n = 40) (b). Thal thalamic radiation, CST Corticospinal Tract, IFOF Inferior Fronto-Occipital Fasciculus, Cing Cingulum bundle, Cing Hipp Cingulum bundle (Hippocampal part), SLF Superior Longitudinal Fasciculus, SLFT Superior Longitudinal Fasciculus (Temporal part), ILF Inferior longitudinal fasciculus, Unc Uncinate Fasciculus. Boxplots show the median and interquartile range (IQR); whiskers extend to 1.5x IQR. Individual points (jittered) are overlaid for visualization and include all values. General linear models showed significant group*session interaction effects in the MD of the right corticospinal tract (CST) between PRG2 and CTR (F(68) = 6.28, p = 0.03, η2 = 0.15, 95% CI [0.02, 0.29]) and in the FA of the left superior longitudinal fasciculus (temporal part; SLFT) between PRG1 and CTR (F(78) = 7.86, p = 0.02, η2 = 0.16, 95% CI [0.03, 0.30]) (Fig. These effects were also visible when directly comparing subjects undergoing a first or second pregnancy, although the FA effect in the left SLFT did not survive the FDR correction for multiple testing (MD CST: F(68) = 7.44, p = 0.02, η2 = 0.18, 95% CI [0.03, 0.32]; FA SLFT: F(68) = 4.09, p = 0.09, η2 = 0.10, 95% CI [0.001, 0.23]; Supplementary Tables 16 and 17). Subsequent paired t-tests/Wilcoxon signed rank tests revealed that these effects were driven by decreases in the MD of the right CST in the PRG2 group (V = 386.5, p = 0.02, r = 0.58, 95% CI [0.28, 0.78]) and decreases in the FA of the left SLFT in the PRG1 group (t(39) = 3.74, p = 0.01, d = 0.59, 95% CI [0.26, 0.93]). We did not observe any differences in the organization of white matter tracts at the PRE session between all three groups. To examine if these changes were maintained across the postpartum period, we also analyzed the MD in the subset of PRG2 participants of which we had complete PRE, POST and POST1 measurements (n = 14). These analyses showed that the MD in the right CST was still decreased 1 year postpartum compared to the pre-pregnancy baseline (t(13) = 4.39; p = 0.02, d = 1.17, 95% CI [0.49, 1.85]), although in this subgroup of women there was no significant difference between the pre-pregnancy baseline and the early postpartum period after correction for multiple testing (t(13) = 2.45, p = 0.18, d = 0.65, 95% CI [0.08, 1.23]). To assess the influence of a second pregnancy on neural metabolite concentrations, we measured the concentration of five major metabolites, in the precuneus/posterior cingulate cortex VOI, an area showing strong volume reductions across a first pregnancy1: total NAA (tNAA: N-acetylaspartate including contributions from N-acetylaspartylglutamate), total creatine (tCr: creatine and phosphocreatine), total choline (tCho: phosphorylcholine and glycerophosphorylcholine), Glu (Glutamate), and Ins (myo-Inositol) (Fig. At the baseline measurement, there were no significant differences in metabolite concentration between the three groups. Neural metabolites were measured in a voxel placed in the precuneus/posterior cingulate (a). Extracted changes in tCr, tNAA, tCho, myo-Inositol and Glutamate concentration (mM) from pre-pregnancy (PRE) to the early postpartum period (POST) across a first (PRG1; n = 39) and second (PRG2; n = 27) pregnancy and in control women (CTR; n = 37) (b). tCr total creatine (creatine and phosphocreatine), tNAA total N-acetylaspartate including contributions from N-acetylaspartylglutamate, tCho total choline (phosphorylcholine and glycerophosphorylcholine). Boxplots show the median and interquartile range (IQR); whiskers extend to 1.5x IQR. Individual points (jittered) are overlaid for visualization and include all values. When examining the changes in metabolite concentration between PRG2 and CTR with general linear models, we observed a group*session interaction effect for tCr only, although this effect did not survive the correction for multiple testing (F(62) = 4.24, p = 0.08, βstd = 0.31, 95% CI [0.01, 0.61]) (Fig. Subsequent one sample-tests showed that this effect was driven by significant increases in tCr across a second pregnancy (t(26) = −5.07, p < 0.0001, d = −0.98, 95% CI [−0.85, −0.36]). Results for the other metabolites can be found in Supplementary Table 18. Additionally, directly comparing PRG2 and PRG1 did not show any significant differences in metabolite concentration changes across a first and second pregnancy (Supplementary Table 19). Although we are starting to unravel the drastic neuroplasticity associated with a first pregnancy, it is currently unknown how a woman's brain is transformed by undergoing another pregnancy. Therefore, we performed a prospective pre-conception cohort study involving 110 women and acquired multimodal imaging data, including high-resolution anatomical MRI, resting-state functional MRI, diffusion-weighted MRI and magnetic resonance spectroscopy. This allowed us to investigate the effects of a second pregnancy on gray matter volume, resting-state brain activity, white matter tract organization and neural metabolite concentrations in comparison to women undergoing a first pregnancy and nulliparous control women. These results revealed that widespread reductions take place in cortical brain volume across a second pregnancy compared to control women, which are similar to the structural changes seen across a first pregnancy. Nevertheless, classification and cluster-wise analyses demonstrated that a second pregnancy differentially affects women's gray matter brain structure compared to a first pregnancy, identifying differences in the patterns of neural change and the degree to which different networks are affected. The observed divergence in the affected networks across a first and second pregnancy was also supported by our functional MRI data showing that functional coherence in the default mode network only increases across a first pregnancy. Furthermore, changes in white matter tract organization in different neural networks were observed across a first and second pregnancy. Together, these results demonstrate that a first and second pregnancy induce similar but also distinctive effects on a woman's brain. The widespread cortical volume reductions we demonstrated across a second pregnancy resemble what we and other independent research groups have shown as brain plasticity across a first pregnancy1,2,3,4,5. However, a multivariate pattern recognition approach showed that women could significantly be classified as having undergone a first or a second pregnancy based solely on their brain changes between sessions. Whereas a previous cross-sectional study in mothers did not show differences in brain volume between primiparous and multiparous women in the early postpartum period28, our longitudinal approach reveals that the brain is differently affected across a first and second pregnancy. Deviations between reproduction-related neuroplasticity in primiparous and multiparous mothers have also been demonstrated in animal studies showing different cellular and molecular signatures in the early postpartum period6,7,8. Whereas multiparity has been associated with lower levels of hippocampal amyloid precursor protein (a marker of neurodegeneration)8 and enhanced spine density in the CA1 region of the hippocampus6, primiparity has been associated with dendritic remodeling in the CA1 and CA3 subregions of the hippocampus6 and a decrease in cell survival in the dentate gyrus7. Although our MRI-based measures of neuroplasticity cannot provide information about the underlying cellular processes across a first and second pregnancy, our data show that, similar to rodents, reproductive experience also differently influences the brain of human mothers. Subsequent analyses examining differences in brain plasticity across a first and second pregnancy reveal both distinct and overlapping neural networks involved. The overlapping areas affected across a first and second pregnancy were mostly located in the default mode network (DMN), followed by the frontoparietal and ventral attention network. We know from previous research that these introspective and higher-order cognitive networks are strongly affected during a first pregnancy, both structurally1,2,3,5 and functionally2. Our findings suggest that these networks represent the main networks associated with pregnancy-induced brain plasticity, being it the first or a subsequent pregnancy. Interestingly, the additional areas that were specifically affected across a first pregnancy were localized in those same networks. This suggests a primary adaptation of these networks in women who become a mother for the first time, which is then further fine-tuned in a similar but more subtle way during a second pregnancy. Our resting-state functional connectivity analyses show a similar pattern. While an increase in DMN network coherence was found in the cuneus of the DMN network in a first pregnancy2, this functional change did not occur to the same degree across a second pregnancy, pointing to a primary structural and functional adaptation of this network when becoming a mother for the first time. The increase in DMN coherence opposes the effect of healthy aging on DMN connectivity, which is characterized by a decrease in within-network DMN connectivity with increasing age29 and associated with changes in cognitive functioning30. On the other hand, psychiatric disorders have shared and disorder-specific patterns of within-network DMN connectivity, characterized by increases and decreases across various diagnosis31. Therefore, more research is needed to fully understand the functional implications of increased DMN connectivity across a first pregnancy. It has been well-established that the default mode network plays an important role in introspection, self-perception and social cognition32,33,34, and the cuneus—the main area where we found the differential effect of a first and second pregnancy on functional network coherence—represents a core structure subserving the neural representation of the self32. Changes in the DMN across a first pregnancy have been interpreted as shifts in a mother's self-perception and her ability to understand her children's needs and feelings. Indeed, we have previously shown that pregnancy-related brain changes in the default mode network relate to a mother's neural, physiological and emotional reactions to her infant1,2,35,36 and the degree to which she differentiates her fetus from herself during pregnancy2. Similarly, our current results show that the volumetric changes across pregnancy are related to maternal behavior. These correlations are more widespread in a first compared to a second pregnancy, suggesting that brain changes in a first pregnancy more strongly contribute to the induction of maternal behavior, whereas in a second pregnancy these changes play a smaller role as maternal behavior was already developed across the first pregnancy. Furthermore, the volumetric brain changes were also associated with peripartum depression and psychological distress in both first-time and second-time mothers, suggesting that pregnancy-induced brain changes play a role in the development of disorders of maternal mental health. Interestingly, the volumetric brain changes were more prominently associated with mental health status during a second pregnancy, but with mental health status in the postpartum period after a first pregnancy. We can speculate that his may be due to higher stress levels during a second pregnancy since the mother needs to care for another child during her pregnancy, although we did not find significant differences in stress and depression across a first and second pregnancy. More research investigating the neural substrates of maternal mental health disorders is needed to further elucidate these findings. Similar to the DMN changes across a first pregnancy, we found a primary modulation of the frontoparietal network in first-time mothers, namely a decreasing FA in the temporal part of the superior longitudinal fasciculus, also known as part of the arcuate fasciculus, compared to control participants but not compared to women undergoing a second pregnancy. Although a matter of debate, lower FA values might indicate reduced integrity of the superior longitudinal fasciculus, as water diffusion is less directional, suggesting a less organized tract. This tract plays an important role in language processing, and enables communication of the frontoparietal regions and the temporal lobe37. Disruptions in white matter integrity have been associated with cognitive dysfunction in different disorders, such as an association between lower FA values in the superior longitudinal fasciculus and working memory deficits and speed of processing in Schizophrenia38. In accordance with the relatively pronounced cortical volume changes in first-time mothers in the frontoparietal network2,3, these results suggest that a first pregnancy more strongly transforms this higher-order cognitive neural network compared to a second pregnancy. In comparison, the areas affected specifically across a second pregnancy were not localized in the introspection-related default mode network and the cognitive frontoparietal network but instead were mainly located in networks involved with the responsiveness to external stimuli, goal-oriented attention and task demands, such as the somatomotor and dorsal attention networks. These changes can be speculated to prepare a woman for the increased demands associated with caring for multiple children at the same time. Indeed, a previous study showed different neural responses as an index of attention to both social and non-social visual stimuli during pregnancy in primiparous compared to multiparous women39. The changes in the sensorimotor network across a second pregnancy were also supported by our diffusion-weighted MRI findings. In women undergoing a second pregnancy, mean diffusivity (MD) in the right corticospinal tract reduced across a second pregnancy, compared to control participants and women pregnant of their first child. A reduction in MD in white matter indicates that water molecules are diffusing less freely in all directions within the tissue, potentially reflecting an increase in the structural integrity. The corticospinal tract is the main white matter tract conveying motor and sensory signals from and to the sensorimotor network40. It has been suggested that a change in MD may reflect synaptic plasticity, and may be a biomarker for microstructural changes associated with learning41,42. In comparison, aging is associated with increases in MD across the white matter43, suggesting that pregnancy may oppose this aging effect. However, further research is needed to elucidate the functional implications of these changes in brain structure in second-time mothers. When analyzing the magnetic resonance spectroscopy data measuring neural metabolite concentrations in the precuneus/posterior cingulate voxel, we did not find robust changes that survived a correction for multiple testing across a second pregnancy. This is in line with our previous study that found indications for changes in myo-inositol, total creatine and total choline concentrations across a first pregnancy, but these also did not survive the multiple testing correction2. These results suggest that there is no clear effect across a first or second pregnancy on metabolite concentrations in the precuneus/posterior cingulate cortex. Comparisons of the baseline MRI data revealed no significant group differences in any of our presented neural measures of vertex-wise brain volumes, resting-state functional connectivity, white matter tract organization and neural metabolites. This is in line with our findings of changed temporal coherence in the DMN, which reverts to baseline across the first year postpartum2. Previous findings showing changes in white matter organization also indicate that these effects are transient4. However, the changes in gray matter structure seem to be long-lasting. Although a partial reversal to pre-pregnancy baseline has been shown across the postpartum period1,2,3,4, structural alterations were still evident 21 and even 6 years44 after delivery. While the volumetric means at the PRE sess are lower in PRG2 women in brain areas that undergo reductions across a first pregnancy, these are not statistically significantly different. This lack of observed differences in gray matter structure in the pre-pregnancy session between the PRG2 and other groups in our study could be due to the reduced power and sensitivity of cross-sectional analyses, especially because the variation in gray matter volumes at the PRE timepoint is very large. Additionally, the postpartum recovery process is highly dynamic45, with volumetric increases shown across different time points postpartum46,47,48,49, but also volumetric decreases from 1 to 2 years postpartum50. The large range of postpartum time since the first pregnancy in our second-pregnancy group may have masked baseline differences between the PRG2 and other groups in our current study. Future studies examining brain changes across a first and second pregnancy within the same women may provide a deeper insight into the recovery of brain changes in between subsequent pregnancies and the shared and distinct neural effects of successive pregnancies. Although our manuscript focusses on relatively short-term effects of parity on the brain, studies in late life showing associations between parity and brain structure in middle-aged women suggest that traces of pregnancy-related neural changes may be present throughout the lifespan. Middle-aged women who had undergone multiple pregnancies showed younger-looking brains compared to primiparous and nulliparous women14. Additionally, cortical thickness and functional connectivity was related to parity in elderly women15,16, and reproductive experience may influence the risk for and effects of different diseases in later life, like Alzheimer's Disease and stroke18,23. Similarly, in middle-aged rodents, long-term effects of parity on the immune system and reduced brain aging associated with reproductive experience have been shown, characterized by more neurogenesis, higher levels of brain derived neurotrophic factor and more synaptic proteins in the hippocampus9,10,51. The differential neural effects of a first and second pregnancy we showed may contribute to these neural effects of parity in late life. Various limitations of our study need to be considered. First, due to local ethical constraints, we were not allowed by the ethical committee to acquire MRI scans during pregnancy. Therefore, the exact timing of the pregnancy-induced changes cannot be concluded from our analyses. Nevertheless, previous studies examining changes during pregnancy have consistently replicated the changes demonstrated in a pre-post pregnancy design3,4,5. This strong consistency of pregnancy-induced volumetric decreases measured at different time points across pregnancy, characterized by an inverted U-shape from pre to post-pregnancy with the lowest volumes during the third trimester of pregnancy3,4,5, supports the notion that the changes we observed are induced by pregnancy. Although we demonstrate clear effects on MRI-based measures of cortical volume and white matter tracts across pregnancy, our study cannot reveal any information about the cellular processes that are underlying these changes. Gray matter volume reductions may reflect neurodegeneration, although our previous research has demonstrated a high similarity between morphometric change in pregnancy and adolescence, suggesting that pregnancy-induced brain plasticity may rather reflect a fine-tuning process52,53. Additionally, the cellular processes underlying changes in diffusion-based metrics in white matter tracts are a matter of debate54, and changes in FA or MD could reflect different processes underlying neuroplasticity and learning, such as astrocyte swelling, dendritic spine changes, angiogenesis or synaptic changes41,55,56. Nevertheless, decreases in MD have also been shown during adolescence57,58, further supporting the notion that pregnancy-induced neuroplasticity resembles brain plasticity across adolescence. Results regarding white matter fractional anisotropy or quantitative anisotropy changes in white matter across pregnancy have not been conclusive, with our previous study showing no changes from pre-pregnancy to the early postpartum period2 and another study showing increasing quantitative anisotropy during pregnancy, returning to baseline postpartum in a single women undergoing her first pregnancy4. Different timings of research sessions and different analysis strategies may explain these discrepancies, and more research is needed to fully elucidate the effect of pregnancy on white matter tracts. Although our study includes a relatively large group of women, especially given the complicated nature and logistics of our longitudinal study design, the group sizes are still limited for classification analyses. To control for the risk of overfitting in small sample sizes, we performed a leave one subject out cross-validation scheme, keeping the training dataset as large as possible to train our classifier. Since it has been shown that the leave-one-out cross-validation can potentially lead to over-optimistic model performance estimations, we repeated our analyses with a k-fold cross-validation scheme with 10 folds, which resulted in a slightly lower accuracy but still significant classification between PRG1 and PRG2. Nevertheless, future research with larger group sizes would be beneficial to confirm our findings. Because of time constraints during data acquisition, we were able to collect only 5 min of resting-state fMRI data. These data are valuable to compare to our previously published results about resting-state functional connectivity changes in the DMN across a first pregnancy2 that was acquired with the same MRI protocol, but we acknowledge that according to newer standards these data may be limited and should be interpreted with caution, and are therefore presented in the supplement. Future studies acquiring more resting-state fMRI volumes are needed to confirm changes in resting-state functional connectivity across pregnancy. Lastly, although we tried to match our groups based on demographic variables, there was a significant difference in age between second-time mothers and first-time mothers and control women. Since age is also known to influence cortical volumes59, resting-state functional connectivity60,61 and diffusion-based structural connectivity61, we corrected for age in all analyses. Since these aging effects on functional and structural connectivity are in the opposite direction compared to the pregnancy-induced effects we have found, we do not expect age differences to solely underlie our results. Additionally, there may be other potential confounding factors related to pregnancy that influence the observed changes across a second pregnancy, such as age of first pregnancy, breastfeeding, type of delivery and other contributing factors, such as sleep disturbances, stress or social support. Although our previous findings suggest that several of these factors including sleep, stress, type of delivery and breastfeeding do not strongly contribute to pregnancy-induced brain changes2, it is likely that such factors are associated with relatively subtle effects that only surface in large samples. Studies involving a larger group of women becoming mothers may be able to acquire more insights into the influence of such factors on pregnancy-induced brain changes. In conclusion, we have demonstrated widespread volumetric decreases across a second pregnancy, which are highly similar to the effects we and others have shown across a first pregnancy1,2,3,4,5. Despite these similarities, women could be classified as having undergone a first or second pregnancy solely based on the changes in their brain structure, indicating that subsequent pregnancies are associated with distinct neural transformations. Both a first and second pregnancy particularly strongly impacted the introspective default mode network and the frontoparietal network and one of its major white matter trats, the superior longitudinal fasciculus. However, these changes were more prominent in a first pregnancy, suggesting a primary adaptation of this network in women who become mothers for the first time that is further fine-tuned during a second pregnancy. On the other hand, second-time mothers exhibited stronger structural alterations in the dorsal attention and somatomotor networks including the corticospinal tract, suggesting that a second pregnancy entails an enhanced plasticity within these externally-oriented networks. Correlation results revealed associations of volumetric brain changes across both a first and a second pregnancy with mother-infant attachment, but these were more widespread in first-time mothers. Furthermore, the structural brain changes across both a first and a second pregnancy were associated with maternal mental health, but these were more prominently associated with postpartum depression and psychological distress in the postpartum period in first-time mothers and with depression and psychological distress during pregnancy in second-time mothers. These findings demonstrate that the human brain is altered across a second pregnancy, involving changes in gray matter structure, white matter tracts and resting-state brain activity, and show that both a first and second pregnancy confer a unique mark on a woman's brain. This research was evaluated and approved by the Ethics Review Board of the Leiden University Medical Center and complies with all relevant ethical regulations. All participants signed the informed consent forms before any study-related measurement and received monetary compensation. We used a prospective pre-conception cohort study, in which we followed women with the intention to become pregnant of their second child in the following year (PRG2: n = 30). Only women (self-reported) were included in this study. Because of the COVID-19 pandemic, only a subset of women could participate in the POST1 session (n = 14). Due to this low sample size, we just included the POST1 timepoint in analyses where we found a significant pregnancy effect from PRE to POST in the PRG2 group, to study the long-term effects of pregnancy. The pregnancy session only included questionnaires, hormone sampling and cognitive testing, since MRI scanning was not allowed during pregnancy by the ethical committee. Additionally, this study included a primiparous group, including women who became pregnant of their first child during the course of the study (PRG1: n = 40, for the POST1 session n = 28) and nulliparous control women (CTR: n = 40), who underwent a similar study set-up (see Fig. All groups were scanned in the same time period, and individuals of different groups were scanned intertwined. For more information about the sample included, see Supplementary Table 20. We included second time mothers (multiparous; PRG2), first-time mothers (primiparous; PRG1) and control women (CTR) not getting pregnant in between the scans (CTR). We acquired structural T1-weighted, resting-state functional, diffusion-weighted MRI scans and a magnetic resonance spectroscopy (MRS) scan before pregnancy (Session 1: PRE), in the early postpartum period (Session 3; POST) and the late postpartum period (Session 4; POST1). The pregnancy session only included questionnaires, hormone sampling and cognitive testing, since MRI scanning was not allowed during pregnancy by the ethical committee. Icons in this figure were obtained from Font Awesome (https://fontawesome.com) and are used under the Creative Commons Attribution 4.0 International License (CC BY 4.0; https://creativecommons.org/licenses/by/4.0/). The icons were not adapted. The PRG2 group was significantly older (32.03 ± 2.33 years) than the PRG1 (29.35 ± 3.51 years; p = 0.0027) and the CTR group (29.33 ± 3.57 years; p = 0.0025), so age has been used as a covariate in all analyses. There were no significant differences in level of education, time between PRE and POST sessions, time between birth and POST session and time between birth and POST1 session (Supplementary Table 20). High-resolution 3D anatomical T1-weighted images were acquired in transverse orientation, with the following acquisition parameters: repetition time (TR) = 9.8 ms, echo time (TE) = 4.6 ms, Flip Angle = 8°, Field of View (FOV) = 178 × 224 × 168 mm, voxel size = 0.875 × 0.875 × 1 mm. All MRI scans were visually checked for quality control, and no scans had to be excluded. Additionally, we determined Freesurfer's Euler number as a measure of data quality for each T1w image at each time point (Supplementary Table 21), and did not find a significant group (PRG2, PRG1, CTR) * session (PRE, POST, POST1) interaction effect (F(3,254) = 0.42, p = 0.74). After processing in Freesurfer, all individual T1w images had Euler numbers of 2, which reflect an accurate surface reconstruction. Functional MRI scans were acquired in all participants during rest, while fixating at a crosshair at the screen to prevent them from falling asleep. A total of 137 T2*-weighted whole-brain echo-planar images (EPIs) and two dummy scans were acquired with the following parameters: TR = 2.2 s; TE = 30 ms, Flip Angle = 80°, FOV = 220 × 220 × 111.65 mm, voxel size = 2.75 × 2.75 mm, 37 descending slices. Two sets of transverse diffusion-weighted images (DWI) were acquired with reversed k-space encoding direction, to allow for distortion correction. For both DWI scans, the acquisition parameters were: TR = 7315 ms, TE = 69 ms, Flip Angle = 90°, FOV = 240 × 240 mm, acquisition matrix = 128 × 98, reconstruction matrix 128 × 128, 30 different diffusion directs with b-factor 1000 s/mm2, 5 B0 acquisitions, SENSE factor = 3, 75 slices of 2 mm, no slice gap, no cardiac gating. Magnetic resonance spectroscopy (MRS) was performed with single-voxel point-resolved spectroscopy (PRESS) localization (TR = 2000 ms; TE = 37 ms; 128 averages, 2 dummy scans, and 16 reference scans without water suppression). Shimming was performed with an automated second order projection-based algorithm. The volumes of interest (VOIs) were positioned in two regions that were found to undergo strong changes in brain structure in our previous study1. One was the precuneus/posterior cingulate cortex (PCC), centered between both hemispheres (Fig. 6a), which had a volume of 8 mL (20 × 20 × 20 mm3). The other VOI was positioned in the right superior temporal gyrus and had a volume of 12 mL (20 × 30 × 20 mm3), but due to low quality spectra, this VOI was excluded from all analyses. The anatomical MR images were processed with Freesurfer 7.2.0, using the longitudinal recon-all stream62. We performed vertex-wise analyses using cortical volume, thickness and surface area, for which the subject's cortical maps were registered to the fsaverage space and smoothed with a 10 mm full-width-at-half-maximum (FWHM) kernel. Cortical change maps were created for each measure, and generalized linear models (GLMs), including age as a covariate, were performed to determine differences in the vertex-wise change between each combination of groups (PRG2 vs. CTR, PRG1 vs. CTR and PRG2 vs. PRG1). These linear models were false discovery rate (FDR) corrected for multiple testing. We also determined Cohen's D effect sizes for the significant vertices (Supplementary Table 1). In order to allow for slightly less prominent effects to surface, we also used a more lenient approach to correct this contrast for multiple testing than the FDR-correction, using 1000 permutations and a vertex-wise threshold of p < 0.01, corrected for performing analyses in the left and right hemisphere. Cortical volumes were extracted from the total area as output of each GLM, and paired t-tests, or in case of non-normality Wilcoxon signed rank tests, were used to determine the direction of effect in the three groups. These paired t-tests/Wilcoxon signed rank tests were repeated in multiparous (n = 14) and primiparous (n = 28) women with complete datasets, including the late postpartum timepoint (POST1). Differences between groups at the pre pregnancy baseline were determined using linear models corrected for age. To further investigate differences in brain changes across subsequent pregnancies, we determined whether women undergoing a first or second pregnancy could be separated based on their anatomical MRI scans. To do so, we performed linear support vector machine classification in Pronto (PRoNTo v3.0)24,25 on the smoothed gray matter volume difference maps as output from longitudinal symmetric diffeomorphic modeling pipeline26 in SPM12 (http://www.fil.ion.uncl.ac.uk/spm/) implemented in Matlab 7.8 (Mathworks), using a leave-one-out cross-validation. Because of our relatively small sample size for classification analyses, the leave-one-out cross validation provides the largest training dataset as possible, but at the risk of overfitting and overestimating classifier accuracy. Therefore, we repeated the analyses using a k-fold cross-validation using 10 folds, which reduces the risk of overfitting although it's better suited for large sample sizes. Details about the longitudinal pipeline in SPM12 can be found in Hoekzema et al.2. Permutation testing using 10,000 permutations was used to determine the significance of the classification accuracy for all three combinations of groups (PRG2 vs. PRG1, PRG2 vs. CTR and PRG1 vs. CTR). Lastly, we performed a cluster-wise correction on the results from the above described GLMs with a vertex-wise cluster threshold of 0.001 and a cluster-wise p-threshold of 0.05 to determine clusters of significant volumetric differences. We combined these cluster maps of the results from PRG1-CTR and PRG2-CTR, to determine the similarity and difference in areas affected with a first or second pregnancy compared to control women. This resulted in three areas of interest: areas affected in both a first and second pregnancy (Overlap), areas affected only in a first pregnancy (Only CTR-PRG1) and areas affected only in a second pregnancy (Only CTR-PRG2). The resting state fMRI images were pre-processed using DPARSF (version 4.5)63, involving slice timing correction, realignment, and co-registration of the anatomical images to the mean functional images. The transformed anatomical images were then segmented64, and DARTEL65 was used to transform the images to MNI space, followed by the application of a 10 mm3 FWHM Gaussian kernel. To account for head motion, we applied the Friston 24-parameter model66 and subjects with any frame-wise displacements (FD) exceeding 2 mm (for translations) or 2° (for rotations) or with a mean FD exceeding 0.2 in any of the sessions were excluded67. Therefore, 3 participants of the PRG2 group and 4 participants of the CTR group had to be excluded. After these exclusions, there were no significant group (PRG1, PRG2 and CTRL) * session (PRE and POST) interaction effects in the mean FD or any of the other motion parameters (maximum translations and rotations in the x, y and z-directions) (Supplementary Table 22). We performed Group spatial independent component analyses (ICA) using the Group ICA for fMRI Toolbox in Matlab (GIFT v4.0b, http://mialab.mrn.org/software/gift), using default options, 20 components and the InfoMax algorithm. Components were selected through automated selection by spatial sorting with the components of Smith et al.68 who defined the major networks in the resting brain using a similar ICA-based approach, using a cutoff value of R > 0.25. After determination of the neural networks, we extracted the correlation between the functional networks using the Functional Network Connectivity toolbox (FNC Toolbox version 2.3, https://trendscenter.org/software/fnc/) for Matlab with a lag-shift algorithm using default options. FNC calculates a constrained maximal lag correlation between each pair of networks by calculating Pearson's correlations and constraining the lag between the time courses69. To examine whether there were differences between groups in the within-network coherence change across the PRE and POST sessions, we performed generalized linear models in SPM12 and investigated the group * session interaction effect using the PRG2 and PRG1 group, and the PRG2 and CTR group. As a follow-up analysis, for the model investigating the default mode network, a region of interest representing the significant within-network coherence change across a first pregnancy2 was applied as mask. Results were considered significant at an FWE-corrected statistical threshold of p < 0.05. When significant interaction results were obtained, these were followed by paired sample t-tests, or in case of non-normality the non-parametric Wilcoxon signed rank test, using the PRE and POST measurement within each group. Baseline differences between groups were determined using linear models corrected for age. For the between-network correlation changes, we performed repeated-measures general linear models in Rstudio (version 23.06.1). In case of significant group * session interaction effects, we performed paired t-tests, or in case of non-normality the non-parametric Wilcoxon signed rank test, using the PRE and POST measurement within each group. Results were considered to be significant at a statistical threshold of p < 0.05, corrected for multiple testing using an FDR correction. To create images, the statistical maps were projected onto the PALS surface provided in Caret software (http://brainvis.wustl.edu/wiki/index.php/Caret). Slice overlays were created using MRIcron (http://www.mccauslandcenter.sc.edu/mricro/mricron/). Diffusion-weighted images were processed using MRtriX370, unless otherwise stated. We denoised the data, preprocessed the data including eddy current-induced distortion correction, motion correction and susceptibility-induced distortion correction using FSL tools eddy & top-up, and corrected the data for B1 inhomogeneity using dwidenoise, dwifslpreproc and dwibiascorrect, respectively. Subsequently, we fitted the diffusion-tensor and calculated the fractional anisotropy (FA) and mean diffusivity (MD) maps. Diffusion tensors were fitted in two steps using dwi2tensor with default settings: first, using weighted least-squares with weights based on empirical signal intensities, and second, by further iterated WLS with weights determined by the signal predictions from the previous iteration. Mean B0 maps were linearly registered using FSL (version 6.0.3) to one individual B0 map, followed by taking the mean to create a study-specific template. Afterwards, individual mean B0 maps were non-linearly registered to the study-specific template, and subsequently to the JHU-ICBM atlas template71. Registrations were inverted, and masks of 11 large white matter tracts of interest were registered to individual subject space. ROI masks in individual space were masked with a fractional anisotropy (FA) mask >0.25, to ensure inclusion of white matter, and mean FA and MD were calculated for each subject at each time point. In case of significant group * session interaction effects, we performed paired t-tests, or in case of non-normality the non-parametric Wilcoxon signed rank test, using the PRE and POST measurement within each group. Baseline differences between groups were determined using linear models corrected for age. To correct for multiple testing across all white matter tracts per measure (FA or MD), we performed FDR corrections. Lastly, for individuals with a complete PRE-POST-POST1 dataset (PRG2: N = 14, PRG1: n = 28), we performed paired t-tests/Wilcoxon signed rank tests between the PRE and POST1 measurements to investigate the long-term effects of pregnancy. MRS data of the PRE and POST session were available of 27 women in the PRG2 group, since the spectroscopy had to be omitted in three participants due to a lack of time. The POST1 data was excluded from the MRS analyses because data was only available of four participants in this group. Additionally, MRS data was acquired from 39 PRG1 women and 37 CTR women at the same timepoints (PRE and POST). Metabolite concentrations were estimated with LCModel, (version 6.3-1 M), using a dataset containing seventeen metabolites. For this study, we considered the major metabolites tNAA (N-acetylaspartate including contributions from N-acetylaspartylglutamate), tCr (creatine and phosphocreatine), tCho (phosphorylcholine and glycerophosphorylcholine), Glu (Glutamate), Ins (myo-Inositol). These metabolites (or combinations thereof) can typically be measured with high precision (see next section about spectral quality). Concentrations were expressed using water scaling. Next, we corrected for partial volume contributions of GM, white matter and cerebrospinal fluid in the corresponding VOI, based on Sienax segmentation (FSL 5.0.10) of each subject's 3DT1 images. Spectral quality was examined based on the full-width half maximum (FWHM), signal-to-noise ratio (SNR), and the estimated CramerRao lower bounds of each metabolite. Spectra with FHWM > 0.1 ppm (12 Hz) and/or SNR < 5 were considered poor quality. All PCC spectra in the PRG2 group had high quality, with SNR mean ± sd of 24.83 ± 1.51, FWHM 4.32 ± 0.51 Hz, and Cramer Rao lower bounds of metabolites well below 10%: tCr 2.06 ± 0.23%, tNAA 2.17 ± 0.38%, Cho 4.50 ± 0.50%, Ins 6.63 ± 0.62%, and Glu 8.00 ± 0.34%. Information about spectra quality in the PRG1 and CTR group can be found in our previous paper2. There were no differences in FWHM and SNR ratio across groups and time points. The metabolite concentrations resulting from LCModel were analyzed in Rstudio (version 23.06.1), using repeated measures general linear models to assess whether the change in metabolite concentration from PRE to POST was different in PRG2 women, compared to PRG1 and CTR women. In case of significant group * session interaction effects, we performed paired t-tests, or in case of non-normality the non-parametric Wilcoxon signed rank test, using the PRE and POST measurement within each group. Baseline differences were determined using linear models corrected for age. These analyses were FDR corrected across the five metabolites as correction for multiple testing. To measure maternal behavior, we used several questionnaires during pregnancy (at the PREG session) and in the early postpartum period (at the POST session). Additionally, to examine nesting behavior, the preparational activities during pregnancy, women filled in the Nesting Behavior Questionnaire74. To assess maternal-child bonding and impairments in the mother-infant relationship in the postpartum period, we asked women to fill in the Maternal Postnatal Attachment Scale (MPAS75) and the Postpartum Bonding Questionnaire (PBQ76) respectively. We assessed mental health status during pregnancy and in the early postpartum period. To assess psychological distress, women in the PRG1 and PRG2 group filled in the K10 questionnaire77. Additionally, to measure signs of postpartum depression, we acquired data from the Edinburgh Postnatal Depression Scale (EPDS78). We determined whether there was a difference in maternal behavior or maternal mental health between the PRG1 and PRG2 group using ANOVA analyses. Additionally, we performed correlation analyses in Freesurfer, to assess whether the maternal behavior or mental health status was associated with the volumetric changes across a first or second pregnancy. We applied the significant areas of change across a first or second pregnancy (PRG1 vs. CTR or PRG2 vs. CTR) as mask for these analyses, to only use the area of significant volume decrease across a first or second pregnancy. Results were corrected for multiple testing using 1000 permutations, with a vertex-wise p value of 0.01, and corrected for performing the analyses across the left and right hemisphere. Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Source data for each figure are provided with this paper and in Figshare (https://doi.org/10.6084/m9.figshare.31144273). The raw MRI data and group/demographic information generated in this study for the participants who have provided permission to share their data have been deposited in the Open Science Framework depository under the following https://doi.org/10.17605/OSF.IO/G8DNR. The deposited data are available open access. Source data are provided with this paper. Hoekzema, E. et al. Pregnancy leads to long-lasting changes in human brain structure. Hoekzema, E. et al. Mapping the effects of pregnancy on resting state brain activity, white matter microstructure, neural metabolite concentrations and grey matter architecture. Servin-Barthet, C. et al. Pregnancy entails a U-shaped trajectory in human brain structure linked to hormones and maternal attachment. Pritschet, L. et al. Neuroanatomical changes observed over the course of a human pregnancy. Paternina-Die, M. et al. Women's neuroplasticity during gestation, childbirth and postpartum. Pawluski, J. L. & Galea, L. A. M. Hippocampal morphology is differentially affected by reproductive experience in the mother. Pawluski, J. L. & Galea, L. A. M. Reproductive experience alters hippocampal neurogenesis during the postpartum period in the dam. Motherhood mitigates aging-related decrements in learning and memory and positively affects brain aging in the rat. Barha, C. K., Lieblich, S. E., Chow, C. & Galea, L. A. M. Multiparity-induced enhancement of hippocampal neurogenesis and spatial memory depends on ovarian hormone status in middle age. Duarte-Guterman, P. et al. Cellular and molecular signatures of motherhood in the adult and ageing rat brain. Barha, C. K. & Galea, L. A. M. Motherhood alters the cellular response to estrogens in the hippocampus later in life. Workman, J. L., Crozier, T., Lieblich, S. E. & Galea, L. A. M. Reproductive experience does not persistently alter prefrontal cortical-dependent learning but does alter strategy use dependent on estrous phase. Voldsbekk, I. et al. A history of previous childbirths is linked to women's white matter brain age in midlife and older age. De Lange, A. M. G. et al. Population-based neuroimaging reveals traces of childbirth in the maternal brain. Relationship between parenthood and cortical thickness in late adulthood. Protective role of parenthood on age-related brain function in mid- to late-life. Long-term association of pregnancy and maternal brain structure: the Rotterdam Study. Ritzel, R. M. et al. Multiparity improves outcomes after cerebral ischemia in female mice despite features of increased metabovascular risk. Gilfarb, R. et al. Multiparity differentially affects specific aspects of the acute neuroinflammatory response to traumatic brain injury in female mice. Jung, J. H. et al. Multiparity, brain atrophy, and cognitive decline. Previous parity differentially influences cognition in later life depending on dementia status. Differential effects of completed and incomplete pregnancies on the risk of Alzheimer disease. Schrouff, J., Mourão-Miranda, J., Phillips, C. & Parvizi, J. Decoding intracranial EEG data with multiple kernel learning method. Schrouff, J. et al. PRoNTo: pattern recognition for neuroimaging toolbox. & Ridgway, G. R. Symmetric diffeomorphic modeling of longitudinal structural MRI. The organization of the human cerebral cortex estimated by intrinsic functional connectivity. The expectant brain-pregnancy leads to changes in brain morphology in the early postpartum period. Reorganization of brain networks and its association with general cognitive performance over the adult lifespan. Doucet, G. E. et al. Transdiagnostic and disease-specific abnormalities in the default-mode network hubs in psychiatric disorders: a meta-analysis of resting-state functional imaging studies. Andrews-Hanna, J. R. The brain's default network and its adaptive role in internal mentation. Becoming a mother entails anatomical changes in the ventral striatum of the human brain that facilitate its responsiveness to offspring cues. Spalek, K. et al. Pregnancy renders anatomical changes in hypothalamic substructures of the human brain that relate to aspects of maternal behavior. Nakajima, R., Kinoshita, M., Shinohara, H. & Nakada, M. The superior longitudinal fascicle: reconsidering the fronto-parietal neural network based on anatomy and function. Reduced white matter integrity associated with cognitive deficits in patients with drug-naive first-episode schizophrenia revealed by diffusion tensor imaging. Rutherford, H. J. V., Maupin, A. N. & Mayes, L. C. Parity and neural responses to social and non-social stimuli in pregnancy. Moreno-López, Y., Olivares-Moreno, R., Cordero-Erausquin, M. & Rojas-Piloni, G. Sensorimotor integration by corticospinal system. Zatorre, R. J., Fields, R. D. & Johansen-Berg, H. Plasticity in gray and white: neuroimaging changes in brain structure during learning. Tavor, I., Botvinik-Nezer, R., Bernstein-Eliav, M., Tsarfaty, G. & Assaf, Y. Short-term plasticity following motor sequence learning revealed by diffusion magnetic resonance imaging. Bennett, I. J., Madden, D. J., Vaidya, C. J., Howard, D. V. & Howard, J. H. Age-related differences in multiple measures of white matter integrity: a diffusion tensor imaging study of healthy aging. Martínez-García, M. et al. Do pregnancy-induced brain changes reverse? The brain of a mother six years after parturition. Martínez-García, M., Paternina-Die, M., Desco, M., Vilarroya, O. The plasticity of human maternal brain: longitudinal changes in brain anatomy during the early postpartum period. Nehls, S., Losse, E., Enzensberger, C., Frodl, T. & Chechko, N. Time-sensitive changes in the maternal brain and their influence on mother-child attachment. Luders, E. et al. From baby brain to mommy brain: widespread gray matter gain after giving birth. & Kühn, S. Postpartal neural plasticity of the maternal brain: early renormalization of pregnancy-related decreases? Brain structural plasticity associated with maternal caregiving in mothers: a voxel-and surface-based morphometry study. Puri, T. A., Richard, J. E. & Galea, L. A. M. Beyond sex differences: short- and long-term effects of pregnancy on the brain. Carmona, S. et al. Pregnancy and adolescence entail similar neuroanatomical adaptations: a comparative analysis of cerebral morphometric changes. & Lonstein, J. S. Less can be more: fine tuning the maternal brain. Jones, D. K., Knösche, T. R. & Turner, R. White matter integrity, fiber count, and other fallacies: the do's and don'ts of diffusion MRI. Johansen-Berg, H., Baptista, C. S. & Thomas, A. G. Human structural plasticity at record speed. Sagi, Y. et al. Learning in the fast lane: new insights into neuroplasticity. Lebel, C., Walker, L., Leemans, A., Phillips, L. & Beaulieu, C. Microstructural maturation of the human brain from childhood to adulthood. & Yuan, W. White matter development during adolescence as shown by diffusion MRI. Lemaitre, H. et al. Normal age-related brain morphometric changes. Ng, K. K., Lo, J. C., Lim, J. K. W., Chee, M. W. L. & Zhou, J. Reduced functional segregation between the default mode network and the executive control network in healthy older adults: a longitudinal study. Damoiseaux, J. S. Effects of aging on functional and structural brain connectivity. Reuter, M., Schmansky, N. J., Rosas, H. D. & Fischl, B. Within-subject template estimation for unbiased longitudinal image analysis. Yan, C. G., Wang, X., Di, Zuo, X. N. & Zang, Y. F. DPABI: data processing & analysis for (resting-state) brain imaging. Accurate automatic estimation of total intracranial volume: a nuisance variable with less nuisance. A fast diffeomorphic image registration algorithm. Büchel, C., Wise, R. J., Mummery, C. J., Poline, J. & Friston, K. J. Nonlinear regression in parametric activation studies. Jenkinson, M., Bannister, P., Brady, M. & Smith, S. Improved optimization for the robust and accurate linear registration and motion correction of brain images. Correspondence of the brain's functional architecture during activation and rest. von dem Hagen, E. A. H., Stoyanova, R. S., Baron-Cohen, S. & Calder, A. J. Reduced functional connectivity within and between ‘social' resting state networks in autism spectrum conditions. Tournier, J. D. et al. MRtrix3: a fast, flexible and open software framework for medical image processing and visualisation. Hua, K. et al. Tract probability maps in stereotaxic spaces: analyses of white matter anatomy and tract-specific quantification. Muller, M. E. Development of the prenatal attachment inventory. Condon, J. T. The assessment of antenatal emotional attachment: development of a questionnaire instrument. Anderson, M. V. & Rutherford, M. D. Evidence of a nesting psychology during human pregnancy. Brockington, I. F., Fraser, C. & Wilson, D. The Postpartum Bonding Questionnaire: a validation. Kessler, R. C. et al. Short screening scales to monitor population prevalences and trends in non-specific psychological distress. Cox, J. L., Holden, J. M. & Sagovsky, R. Detection of postnatal depression: development of the 10-item Edinburgh Postnatal Depression scale. We acknowledge the participants for their contribution to this study. We thank research assistants A. van der Geest, A. Glasbergen, S. Altikulac, A. van Steenbergen, R. van Dort, P. Berns en I. Langereis for coordinating the data collection for this project. This project was supported by an Innovational Research Incentives Scheme grant (Veni, 451-14-036, E.H.) by the Netherlands Organisation for Scientific Research (NWO), a NARSAD grant from the Brain and Behaviour Research Foundation, U.S.A. (grant number 25312, E.H.) and a grant of the Leiden University Fund / Elise Mathilde Fund (CWB 740s / 2t-03-2017 /EM) awarded to E.H. E.H. is currently supported by an ERC Starting Grant (948031, E.H.) provided by the European Research Council. Erasmus School of Social and Behavioural Sciences, Erasmus University Rotterdam, Rotterdam, The Netherlands Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar analyzed data and wrote the paper, S.H. supervised processing and analyses of the MRS data, E.C. contributed to the design and interpretation, E.H. designed the study, analyzed data, and contributed to the manuscript and interpretation. All authors evaluated the manuscript. Correspondence to M. Straathof or E. Hoekzema. The authors declare no competing interests. Nature Communications thanks Darby Saxbe and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The effects of a second pregnancy on women's brain structure and function. Anyone you share the following link with will be able to read this content: Sorry, a shareable link is not currently available for this article. Provided by the Springer Nature SharedIt content-sharing initiative Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript. Predicting not just if, but also when, cognitively unimpaired individuals are likely to develop onset of Alzheimerʼs disease (AD) symptoms would be useful to clinical trials and, eventually, clinical practice. Although clock models based on amyloid and tau positron emission tomography have shown promise in predicting the onset of AD symptoms, a model based on plasma biomarkers would be more accessible. Using longitudinal plasma %p-tau217 (the ratio of phosphorylated to non-phosphorylated tau at position 217) from two independent cohorts (n = 258 and n = 345), clock models were used to estimate the age at plasma %p-tau217 positivity. The estimated age at plasma %p-tau217 positivity was associated with the age at onset of AD symptoms (adjusted R2 of 0.337−0.612) with a median absolute error of 3.0−3.7 years. Notably, the time from %p-tau217 positivity to onset of AD symptoms was markedly shorter in older individuals. Similar models were constructed with data from one p-tau217/Aβ42 immunoassay and four plasma p-tau217 immunoassays. These findings suggest that the time until onset of AD symptoms can be estimated using a single blood test within a margin of error that is acceptable for use in clinical trials. AD is the most common cause of dementia and is characterized by amyloid plaques primarily comprised of amyloid-β 42 (Aβ42) and neurofibrillary tangles comprising tau1. The burden of amyloid plaques as quantified by amyloid positron emission tomography (PET) increases for about 10−20 years during the preclinical phase of AD when patients are cognitively unimpaired and then plateaus around the time of symptom onset2,3. By contrast, neurofibrillary tangles as measured by tau PET develop later and increase with symptom severity4,5. Although treatments for early symptomatic AD are now clinically available in some countries6,7, treating patients earlier during the preclinical phase of disease before major neurodegeneration has occurred may be more efficacious8,9. Although biomarkers can accurately identify individuals with AD brain pathology, predicting which individuals are likely to develop symptoms is more challenging, and novel predictive modeling approaches are needed. Interestingly, once amyloid plaques and neurofibrillary tangles start to accumulate, the burden of pathology follows a remarkably consistent trajectory across individuals2,10,11,12,13. These consistent trajectories enable the creation of clock models that relate levels of amyloid or tau PET signal to time and allow for estimation of when individuals developed amyloid or tau PET abnormality2,11,12,13,14,15. Unlike general biological aging clocks or categorical staging based on multiple biomarkers, clock models track disease progression with a specific biomarker, thereby providing granular and intuitive time-based staging1,10,13,16. Clock models allow alignment of trajectories to a reference point (for example, amyloid or tau PET positivity) that reduces heterogeneity compared to models using chronological age alone13,16. Furthermore, the age at amyloid or tau abnormality or positivity estimated by clock models can be used to estimate the age at AD symptom onset10,12,13,15,17. Recently developed blood-based biomarkers are much more accessible and less expensive20. The plasma ratio of amyloid-β peptide 42 to 40 (Aβ42/40) decreases very early during the preclinical phase of AD and then plateaus at moderate levels of amyloid plaque burden, resulting in poor correlations with AD symptoms13,16,21,22,23. However, plasma levels of tau phosphorylated at position 217 (p-tau217) and the ratio of phosphorylated to non-phosphorylated tau at position 217 (referred to as %p-tau217) increase throughout the preclinical and early symptomatic phases of AD13,16,22. Importantly, plasma measures of p-tau217 and %p-tau217 have high associations not only with amyloid PET22,24,25,26,27,28 but also with tau PET24,25,29, brain volumes23 and cognition23,28. Although the concentration of plasma p-tau217 was previously demonstrated to predict risk for cognitive decline and progression to AD dementia30,31, no published studies have used plasma p-tau217 or %p-tau217 to estimate when individuals will develop onset of AD symptoms. In the present study, we aimed to use measurements from a single plasma sample to estimate not only the probability of a cognitively unimpaired individual with positive AD biomarkers developing AD symptoms but also when they would be likely to develop symptoms. Primary analyses were performed with the C2N Diagnostics plasma %p-tau217 measure because large longitudinal datasets were available from two cohorts: the Knight Alzheimer's Disease Research Center (Knight ADRC) and the Alzheimer's Disease Neuroimaging Initiative (ADNI). Plasma %p-tau217 has high accuracy in classifying amyloid PET, tau PET and cognitive status; this assay is a component of C2N Diagnostics' clinically available PrecivityAD2 test and is used by some clinical trials23,25,26,27,32. Using longitudinal plasma %p-tau217 data, we created clock models that related %p-tau217 values to time and enabled estimation of the age at %p-tau217 positivity for each individual. Two approaches were used to create %p-tau217 clock models: Temporal Integration of Rate Accumulation (TIRA)10,13, which integrates the inverse of the modeled rate of change, and Sampled Iterative Local Approximation (SILA)12,17, which uses discrete rate sampling and Euler's method for numerical integration. Secondary analyses to assess the generalizability of this approach across plasma biomarkers were performed in the ADNI cohort with the Fujirebio Lumipulse p-tau217/Aβ42 measure that was recently cleared by the US Food and Drug Administration and with four commercially available p-tau217 assays: C2N Diagnostics, Janssen LucentAD Quanterix, ALZpath Quanterix and Fujirebio Lumipulse23. Data from participants with longitudinal plasma %p-tau217 measurements collected at least 1 year apart were considered for potential inclusion in developing the clock models (Supplementary Table 1). When the baseline plasma %p-tau217 sample was collected, 506 individuals from the Knight ADRC cohort had a median age of 67.7 years (interquartile range (IQR) 61.7–72.4 years); 54.2% were female, 35.8% were APOE ε4 carriers and 8.5% were cognitively impaired (defined by Clinical Dementia Rating (CDR) > 0). Knight ADRC participants had a median age of 7.1 years (IQR 5.0−11.0 years) from the first to last plasma collection. ADNI participants had a median age of 5.0 years (IQR 4.0-6.5 years) from the first to last plasma collection. The rate of change in plasma %p-tau217 as a function of the estimated %p-tau217 value at the midpoint of collected samples for each participant was modeled with generalized additive models (GAMs) separately in the two cohorts (Fig. As was described for similar models using amyloid and tau PET13, intervals with relatively low variance in the rate of change (that is, below the 90th percentile) were identified. These intervals were between plasma %p-tau217 values of 0.29% and 10.45% for the Knight ADRC cohort and between 1.06% and 10.59% for the ADNI cohort (Extended Data Fig. For the Knight ADRC (a) and ADNI (b) cohorts, each gray point represents the rate of change in plasma %p-tau217 for an individual as a function of the estimated %p-tau217 value at the midpoint of follow-up. Black solid lines represent GAMs fitting the individual data points, showing the predicted mean rate of change. Gray shading indicates 95% confidence intervals around the fitted GAM curves. Red dashed vertical lines represent the estimated %p-tau217 value at which variance in the rate of change was greater than the 90th percentile of variance in rates of change for the cohort, which identified a range of 1.06−10.45% over which %p-tau217 had relatively consistent change. Clock models relating time to %p-tau217 are shown for the Knight ADRC (c) and ADNI (d) cohorts and were created using two approaches: TIRA (green) and SILA (orange). The vertical black dashed line represents the plasma %p-tau217 threshold of 4.06%, which aligns with an amyloid PET Centiloid value of 20. The horizontal black dashed line represents the estimated time that an individual has a %p-tau217 value of 4.06%. For the development of clock models, the cohort was restricted to individuals with longitudinal data within the interval of consistent change (Extended Data Table 1). The Knight ADRC clock cohort included 258 individuals with a median interval between first and last plasma %p-tau217 values of 6.5 years (IQR 3.9−9.8 years). The ADNI clock cohort included 345 individuals with a median interval between first and last plasma %p-tau217 values of 4.5 years (IQR 4.0−6.3 years). Plasma %p-tau217 positivity was defined as more than 4.06% to align with an amyloid PET Centiloid value of 20 (ref. Two methods that have previously been used to create clock models with longitudinal amyloid and tau PET data were used to estimate the years from plasma %p-tau217 positivity: TIRA10,13 and SILA12,17. The clock models relating time to plasma %p-tau217 levels are shown for the Knight ADRC (Fig. 1d) cohorts; numerical values are provided in Extended Data Table 2 and can also be accessed via a web-based application: https://amyloid.shinyapps.io/plasma_ptau217_time/#. Similar clock models were constructed using all longitudinal data (Supplementary Fig. At very low %p-tau217 values (<1.06%), there were highly unstable TIRA time estimates in the ADNI cohort. Longitudinal data from individuals with high %p-tau217 values were sparse (Fig. 1a,b), providing less certain clock estimates for higher values. Furthermore, there were rapid increases in %p-tau217 at high values (>10.45%) that could make time estimates unstable. These features provided further rationale to use clocks constructed with the restricted range of %p-tau217 values (1.06−10.45%). Plasma %p-tau217 trajectories as a function of age are shown for individuals included in the clock models (Fig. 2a,d) and demonstrate considerable heterogeneity, reflecting that AD pathology begins at widely varying ages. The plasma %p-tau217 trajectories were then re-plotted by years from estimated %p-tau217 positivity (age at plasma collection minus the individual's estimated age at %p-tau217 positivity) for each cohort and clock model (Fig. The trajectories then demonstrated increased alignment, suggesting relatively consistent patterns of change in plasma %p-tau217 across individuals (Supplementary Video 1). Longitudinal plasma %p-tau217 data from the Knight ADRC (a−c) and ADNI (d−f) cohorts are shown as a function of age (a,d) or estimated years from %p-tau217 positivity by TIRA (b,e) or SILA (c,f) clock models. Thick black lines represent the clock models shown in Fig. 1c,d; red lines represent individuals with at least one plasma %p-tau217 > 4.06%; and gray lines represent individuals with no plasma %p-tau217 > 4.06%. Horizontal black dashed lines represent the plasma %p-tau217 threshold of 4.06%. Vertical black dashed lines represent the estimated time that an individual had a %p-tau217 value of 4.06%. Additionally, we performed cross-cohort comparison of the models and found that the estimated ages at plasma %p-tau217 positivity based on TIRA models implemented on either the Knight ADRC or ADNI cohorts were highly correlated (adjusted R2 of 0.978), and ages based on SILA models from the two cohorts were almost perfectly aligned (adjusted R2 of 0.999) (Supplementary Fig. The estimated ages at plasma %p-tau217 positivity based on the TIRA or SILA models were also highly correlated in the Knight ADRC cohort (adjusted R2 of 0.942) but were less correlated in the ADNI cohort (adjusted R2 of 0.863) (Supplementary Fig. Overall, the clock models estimated similar ages at plasma %p-tau217 positivity regardless of the method or cohort and were consistent with observed conversion ages from %p-tau217 negative to positive. Cox proportional hazards models were used in the longitudinally assessed Knight ADRC cohort to estimate the probability of initially cognitively unimpaired individuals developing symptomatic AD as a function of age (Fig. Symptomatic AD was defined to align with the clinical diagnosis of symptomatic AD: cognitive impairment (CDR > 0) with an AD syndrome (clinical features consistent with cognitive impairment caused by AD, including amnestic, logopenic aphasia, posterior cortical dysfunction or dysexecutive presentations1,33,34,35) in the context of biomarkers, indicating the presence of AD pathology1,35. The onset of AD symptoms was defined as the first clinical assessment when initially cognitively unimpaired (CDR = 0) individuals with positive AD biomarkers (based on estimated %p-tau217) were found to be cognitively impaired (CDR > 0) with an AD syndrome. Furthermore, AD symptom onset was applied only to individuals who were cognitively impaired (CDR > 0) with an AD syndrome at their last assessment—that is, if an individual had transient cognitive impairment but returned to cognitively unimpaired or had a non-AD diagnosis at their last assessment, the earlier impairment was not considered to be the onset of AD symptoms. To account for variable time between clinical assessments, the time of AD symptom onset was interval-censored between the last cognitively unimpaired assessment and the first symptomatic AD assessment. For individuals in the Knight ADRC cohort who were cognitively unimpaired at baseline, a Cox model evaluated the probability of remaining cognitively unimpaired from AD as a function of age, stratified by the age at %p-tau217 positivity based on TIRA (a). The estimated time from %p-tau217 positivity until 50% of individuals would be expected to have symptomatic AD is shown as a function of estimated age at %p-tau217 positivity (b). For example, with TIRA-based estimates, participants who became plasma %p-tau217 positive at age 60 had a median time until symptom onset of 20.5 years, whereas participants who became positive at age 80 had a median time until symptom onset of only 11.4 years (Fig. Cox models for both the Knight ADRC and ADNI cohorts are shown in Supplementary Fig. The probability of symptomatic AD associated with a specific plasma %p-tau217 value and age can be accessed via a web-based application: https://amyloid.shinyapps.io/plasma_ptau217_time/. The discriminative ability of Cox models to rank individuals by their risk of developing symptoms was assessed with a concordance index (C-index), where 0.5 is random prediction, >0.7 is very good prediction, >0.8 is excellent prediction and 1.0 is perfect prediction. For the Knight ADRC cohort, the TIRA-based model yielded a bootstrapped C-index of 0.784 (95% confidence interval: 0.720−0.843), and the SILA-based model had a C-index of 0.790 (95% confidence interval: 0.728−0.847). For the ADNI cohort, the TIRA-based model yielded a C-index of 0.730 (95% confidence interval: 0.622−0.834), and the SILA-based model had a C-index of 0.750 (95% confidence interval: 0.636−0.853). Additional analyses incorporating left-censored participants for those with cognitive impairment at study enrollment confirmed these associations with better discrimination and addressed potential survivor bias (Supplementary Fig. Next, the estimated age of plasma %p-tau217 positivity was used to model the age at onset of AD symptoms. For the Knight ADRC cohort, models included 59 individuals using TIRA and 61 individuals using SILA clock models; for the ADNI cohort, models included 20 individuals using TIRA and 22 individuals using SILA (see Supplementary Table 2 for cohort characteristics). Comprehensive model diagnostics confirmed the appropriateness of linear modeling across all cohort−method combinations (Supplementary Table 4). Sensitivity analyses removing outliers demonstrated continued significance of the linear models. The estimated age at AD symptom onset associated with a specific plasma %p-tau217 value and age can be accessed via a web-based application: https://amyloid.shinyapps.io/plasma_ptau217_time/#. Consistent with findings from the Cox models, these models demonstrated that older individuals had a much shorter interval from plasma %p-tau217 positivity to symptom onset (Supplementary Fig. Individuals were included who were initially cognitively unimpaired but had a typical AD syndrome at their last assessment and developed symptoms after estimated plasma %p-tau217 positivity. Age at %p-tau217 positivity was estimated using TIRA (a) or SILA (b) models in the Knight ADRC (green points) or ADNI (black points) cohorts. Each point represents an individual participant. Linear regression lines represent the predicted mean age at symptom onset for each dataset, with shaded bands indicating 95% confidence intervals around the regression lines (green shading for Knight ADRC data and gray shading for ADNI data). Linear regression equations, adjusted R2 values, Spearmanʼs correlation coefficients (ρ) and sample sizes (N) are shown for each cohort. The effects of APOE ε4 carrier status, sex and years of education were also examined, but these variables were either not significant or had minimal effects and, thus, were not included in the models (Supplementary Table 5). Associations between estimated age at plasma %p-tau217 positivity and age at symptom onset were lower when individuals who developed cognitive impairment before %p-tau217 positivity were included, likely because these individuals had non-AD causes of cognitive impairment (Supplementary Fig. The error in models predicting the age at AD symptom onset based on the estimated age at plasma %p-tau217 positivity was examined (Extended Data Table 3). Within the same cohort, models predicting the age at AD symptom onset had a median absolute error (MdAE) that ranged from 3.0 years to 3.5 years and non-parametric concordance correlation coefficients (CCCs) ranging from 0.771 to 0.839 (CCC > 0.6 is considered good and CCC > 0.8 is considered excellent). For models created in the Knight ADRC cohort and applied to plasma %p-tau217 values in the ADNI cohort, there were moderate associations (adjusted R2 of 0.467 for TIRA and 0.463 for SILA) with an MdAE of 3.0−3.2 years and a CCC of 0.801−0.805. Conversely, for models created in the ADNI cohort and applied to plasma %p-tau217 values in the Knight ADRC cohort, there were also moderate associations (adjusted R2 of 0.509 for TIRA and 0.577 for SILA) with an MdAE of 3.6−3.7 years and a CCC of 0.808−0.820. The relationship between predicted AD symptom onset and longitudinal clinical diagnoses was examined. For initially cognitively unimpaired individuals in the Knight ADRC cohort with an estimated age at plasma %p-tau217 positivity by the TIRA-based model (Supplementary Table 6), individuals who became %p-tau217 positive at age 60 were estimated to develop symptomatic AD after 14.0 years, whereas individuals who became %p-tau217 positive at age 80 were estimated to develop symptomatic AD after only 6.2 years (Figs. Similar analyses were performed for individuals with an estimated age at plasma %p-tau217 positivity by TIRA or SILA in both the Knight ADRC and ADNI cohorts, including only those who were initially cognitively unimpaired (Supplementary Fig. Regardless of the clock model used to estimate the age at plasma %p-tau217 positivity, there was a markedly shorter time until symptom onset for individuals who developed %p-tau217 positivity at older ages in both the Knight ADRC and ADNI cohorts. Each row represents the longitudinal clinical diagnoses for one individual by estimated years from %p-tau217 positivity (x axis). Individuals are sorted vertically by estimated age at %p-tau217 positivity (y axis). The point color denotes the clinical diagnosis: blue represents cognitively unimpaired at the assessment; red (AD syndrome/biomarker positive) represents cognitively impaired at the assessment and a diagnosis of symptomatic AD at their last assessment with symptoms starting after %p-tau217 positivity; purple (AD syndrome/biomarker negative) represents cognitively impaired at the assessment and a diagnosis of symptomatic AD at their last assessment with symptoms starting before %p-tau217 positivity; and orange (non-AD syndrome) represents cognitively impaired and a non-AD diagnosis at their last assessment. Progression of initially cognitively unimpaired Knight ADRC participants with a positive plasma %p-tau217 value to cognitive impairment, with either an AD or a non-AD syndrome, demonstrated wide variation in the time until symptom onset (Fig. Progression of individuals to cognitive impairment as a function of years from estimated plasma %p-tau217 positivity (Fig. 6b) or estimated years from symptom onset (Fig. Consistent with the other models, binning individuals by estimated age at plasma %p-tau217 positivity demonstrates that older individuals develop symptoms sooner after %p-tau217 positivity (Fig. However, models estimating symptom onset based on age at plasma %p-tau217 adjust for this age effect (Fig. Similar analyses are shown for SILA-based models in the Knight ADRC dataset (Supplementary Fig. Three groups were examined: red (AD syndrome/biomarker positive) had a diagnosis of symptomatic AD at their last assessment with symptoms starting after %p-tau217 positivity; purple (AD syndrome/biomarker negative) had a diagnosis of symptomatic AD at their last assessment with symptoms starting before %p-tau217 positivity; and orange (non-AD syndrome) had a non-AD diagnosis at their last assessment. Kaplan−Meier curves show the probability for each group of remaining cognitively unimpaired individuals as a function of time from first positive %p-tau217 collection (a,d), estimated years from %p-tau217 positivity (b,e) or estimated years from symptom onset (c,f). Density plots and points beneath Kaplan−Meier curves (a−c) represent the onset of symptoms for individuals in each group. The alignment of the 2024 Alzheimer's Association biological stages with plasma %p-tau217, years since estimated plasma %p-tau217 positivity and estimated years from symptom onset based on plasma %p-tau217 were examined. Individuals with amyloid PET, tau PET and an estimated age at plasma %p-tau217 were classified according to the biological staging framework: stage A (normal biomarkers), stage B (AD pathologic change), stage C (AD) and stage D (advanced AD). Estimated years since symptom onset did not as clearly separate the biological stages (Extended Data Fig. 4); it is possible that biological stages may have different relationships with symptoms across age groups. To assess whether clock models could be generated using other measures of plasma p-tau217, the same approach was implemented in the ADNI cohort with the Fujirebio Lumipulse p-tau217/Aβ42 measure and with four commercially available p-tau217 assays. First, variance analysis was used to find a range of values where the measure had a relatively consistent rate of change: Fujirebio Lumipulse p-tau217/Aβ42, 0.00−0.02; Janssen LucentAD Quanterix p-tau217, 0.01−0.16 pg ml−1; ALZpath Quanterix p-tau217, 0.10−1.11 pg ml−1; C2N Diagnostics PrecivityAD2 p-tau217, 0.72−7.99 pg ml−1; and Fujirebio Lumipulse p-tau217, 0.02−0.51 pg ml−1 (Extended Data Fig. Previously published positivity thresholds that align with amyloid PET Centiloid 20 were used for the p-tau217 measures23; using the same methodology, the threshold for Fujirebio Lumipulse p-tau217/Aβ42 was 0.00631. Next, TIRA and SILA were implemented to create clock models that related time from plasma positivity. Plasma trajectories by age or estimated years from positivity are shown for each plasma measure (Extended Data Figs. Dynamic visualizations of trajectory alignment for each assay are provided in Supplementary Videos 2−6. The associations between age at symptom onset and age at plasma positivity were examined: Fujirebio Lumipulse p-tau217/Aβ42, TIRA adjusted R2 of 0.276 and SILA adjusted R2 of 0.584 (Extended Data Fig. In this study, we demonstrated that clock modeling methods can be used to estimate the age at plasma %p-tau217 positivity and align %p-tau217 trajectories, revealing a relatively consistent change across individuals in %p-tau217 during the preclinical and early symptomatic phase of AD. Furthermore, the estimated age at plasma %p-tau217 positivity was associated with the age at AD symptom onset, enabling prediction of the age at AD symptom onset with a single %p-tau217 value. We found similar results with plasma p-tau217/Aβ42 and p-tau217 measured by immunoassays, demonstrating the generalizability of our approach across different blood biomarker tests. The estimated years until AD symptom onset based on plasma %p-tau217 had an MdAE of 3−4 years, which would limit its utility for individual decision-making, but it could still be useful for group-level studies. Clock models could improve selection of participants for clinical trials who are more likely to develop symptoms within the trial period, increasing statistical power and reducing the time needed to demonstrate treatment efficacy. The variance in age at AD symptom onset explained by estimated age at %p-tau217 positivity ranges from 0.337 to 0.599, which is similar to that explained by parental age at onset for autosomal dominant Alzheimerʼs disease (ADAD) (R2 = 0.384 (ref. The relatively predictable age at symptom onset has been a major advantage of performing clinical trials in ADAD cohorts3,37. It is possible that clock models incorporating plasma %p-tau217 or p-tau217 with other biomarkers, such as eMTBR-tau243 (ref. 38) or biomarkers of cerebrovascular disease39, may enable greater precision in estimating time until AD symptom onset. Future investigations could also explore continuous cognitive measures that identify subtle cognitive changes that occur before the threshold for clinical diagnosis. More precise models may decrease the error in the estimated years until AD symptom onset to a level that it becomes relevant for individual decision-making, which could have considerable clinical and ethical implications40. AD biomarker testing of cognitively unimpaired individuals is currently not recommended outside of research studies or clinical trials due to uncertain benefits and potential risks19,41,42,43, and we discourage individuals from using these models to determine their personal estimated age at AD symptom onset. Notably, we found that older individuals have a markedly shorter time until AD symptom onset after developing plasma %p-tau217 positivity. This is consistent with our previous work predicting symptom onset with amyloid PET10, where we found that older individuals developed symptoms at a lower amyloid PET burden. Age-related brain changes, including age-related increases in the prevalence of co-pathologies that affect the relationships between clinical symptoms and AD pathology44, may underlie this effect. As age increases, co-pathologies become more common and may further complicate the interpretation of %p-tau217 levels in older individuals. This finding has major implications for clinical trials: individuals with the same plasma %p-tau217 values likely have very different risks of developing cognitive impairment over a 3−5-year period depending on their age. Although statistical models typically include age as a covariate, the relationship between plasma %p-tau217 levels and symptom onset is complex and may not be well captured by linear or nonlinear models, although age-stratified analyses may be helpful. Nonlinear mixed-effects models characterize population-level trajectories with individual random effects, but clock models explicitly convert biomarker levels into individualized estimates that are intuitive (for example, years since biomarker positivity) and may reveal important findings such as the marked effect of age at plasma %p-tau217 positivity on the age at AD symptom onset. Although our clock models use the single biomarker %p-tau217, its strong associations with amyloid and tau PET effectively integrate the pathological processes of amyloid plaques and neurofibrillary tangles into the models. Notably, %p-tau217 dynamics likely capture the intertwined progression of both amyloid and tau pathology. Furthermore, the shorter interval from %p-tau217 positivity to symptom onset observed in older individuals may, in part, reflect the influence of age-related co-pathologies, such as cerebrovascular disease and other neurodegenerative diseases44. Recognizing the impact of these additional pathologies is crucial, as they may modify clinical trajectories beyond the core AD pathology. Future work incorporating complementary biomarkers of amyloid, tau and other pathologies will be important for improving the accuracy and applicability of these models. Knight ADRC participants were younger, had a much lower rate of cognitive impairment at baseline and were more likely to be APOE ε4 carriers. Despite these differences, the TIRA and SILA models generally were aligned, and neither was clearly superior, indicating the robustness of the clock concept for modeling years until AD symptom onset. Still, there were some differences, such as TIRA estimating longer periods compared to SILA, particularly in the ADNI cohort. These findings suggest that it may be helpful to implement multiple approaches when developing clocks and to rigorously evaluate the model fit. We have shared code for the TIRA method (the code for the SILA method is already publicly available from the Betthauser laboratory at the University of Wisconsin), which will facilitate testing of these modeling approaches in additional cohorts and using other measures. We encourage interested investigators to further refine these and other approaches to improve prediction of AD symptom onset. We also recommend that clinical trialists use this code and data to create models tailored to their specific goals—for example, determining p-tau217 values for specific age groups that identify individuals at high risk of developing AD symptom onset within 3 years. Despite these promising findings, our approach has limitations. The clock models can only be used for values over which there is a consistent change in %p-tau217 (values between 1.06% and 10.45%). Values outside this range, such as a %p-tau217 of 12%, cannot be used to estimate years until AD symptom onset, although individuals with very high %p-tau217 values are likely at very high risk for developing symptomatic AD. Similarly, very low %p-tau217 values suggest a low likelihood of developing symptomatic AD for many years, but precise estimates of years until AD symptom onset cannot be made. Our analysis focused on participants with plasma %p-tau217 values within the interval of consistent change, which increases model reliability but may limit generalizability to individuals with values outside this range. Symptomatic AD was defined as cognitive impairment with an AD syndrome in the context of an estimated positive %p-tau217 value. The threshold for %p-tau217 positivity corresponds to an amyloid PET Centiloid value of 20, below which very few individuals have symptoms due to AD23,45, making it unlikely that individuals with an estimated negative %p-tau217 value have cognitive impairment due to AD pathology. However, occasional individuals may have discrepant %p-tau217 values. Interpretation of estimates for smaller subgroups or those with mixed clinical presentations may be affected by limited sample sizes and co-pathologies. Additional limitations include that participants in the study had a variety of clinical diagnoses that were grouped together for analyses, and the models do not reflect the full complexity of clinical symptoms. Participants in the study largely identified as non-Hispanic White, which may limit the generalizability of these models to other groups, especially groups with different rates of non-AD co-pathologies. Furthermore, like other longitudinal aging studies, our analysis did not explicitly model participant dropout or death, which could introduce survival bias if individuals who develop more rapid cognitive decline are more likely to discontinue participation. This potential for survival bias is an important consideration when interpreting our results, as it may lead to underestimation of decline among the most vulnerable individuals. In conclusion, our study demonstrates that a single plasma %p-tau217 value can be used to estimate years from onset of AD symptoms with an MdAE of 3−4 years. Models with this level of precision may assist in selecting participants for clinical trials targeting certain phases of preclinical AD. Further refinement of these models could potentially improve predictions, enabling shorter clinical trials and possible relevance for individual decision-making. The STROBE requirements for an observational study were followed. Research participants were included who had been enrolled in previously described studies of memory and aging at the Knight ADRC16 or the ADNI13. Both cohorts consisted of community-dwelling older adults, including participants with and without cognitive impairment, who were followed longitudinally with standardized clinical and biomarker assessments. Sex was self-reported by participants in both cohorts. The Knight ADRC cohort is focused on longitudinal characterization of preclinical AD and the transition to symptomatic AD. ADNI was initiated in 2003 and represents a collaborative effort between public and private sectors, with Michael W. Weiner serving as the principal investigator (https://adni.loni.usc.edu/). The primary goal of the ADNI has been to test whether serial imaging scans, other biological markers and clinical and neuropsychological assessment can be combined to measure the progression of early AD. All protocols were approved by the Washington University in St. Louis institutional review board (Human Research Protection Office) and by the local institutional review boards at each participating ADNI site. Written informed consent was obtained from every participant or, when appropriate, from a legally authorized representative. Plasma was collected as previously described in the Knight ADRC16 and ADNI46,47 cohorts. Plasma %p-tau217 was measured by C2N Diagnostics with a liquid chromatography−mass spectrometry (LC−MS)-based assay32. The plasma %p-tau217 measure was calculated as p-tau217 concentration divided by non-phosphorylated tau217 concentration times 100 and is also described as the percent phosphorylation occupancy27. The Fujirebio Lumipulse G assay for p-tau217 and Aβ42 was run in singlicate with research-use-only commercially available kits on a Fujirebio Lumipulse G1200 analyzer at the Indiana University National Centralized Repository for Alzheimer's Disease and Related Dementias Biomarker Assay Laboratory (NCRAD-BAL)23. Additional details are included in the study methodology report, available in the ADNI database (https://adni.loni.usc.edu/). Individuals with CDR = 0 were categorized as ‘cognitively unimpaired'. Individuals with CDR > 0 were categorized as ‘cognitively impaired'; this group includes individuals with mild cognitive impairment and dementia. Individuals with clinical features consistent with cognitive impairment caused by AD (for example, most commonly, insidious onset, slowly progressive decline and early amnestic impairment but also including logopenic aphasia, posterior cortical dysfunction or dysexecutive presentations) were considered to have an AD syndrome1,33,34,35. Individuals with a primary clinical diagnosis that did not include AD (such as Parkinson disease dementia and vascular dementia) were considered to have a non-AD syndrome. The assessment of clinical syndrome was made by experienced clinicians who were blinded to biomarker results, and determinations were based solely on clinical presentation and established diagnostic criteria and were recorded as the primary clinical diagnosis1,33,34,35. Symptomatic AD was defined to align with the established guidelines for clinical diagnosis of symptomatic AD: cognitive impairment with an AD syndrome in the context of biomarkers, indicating the presence of AD pathology1,35. The onset of AD symptoms was defined as the first clinical assessment when initially cognitively unimpaired individuals with positive AD biomarkers (based on estimated %p-tau217) were found to be cognitively impaired with an AD syndrome. Furthermore, AD symptom onset was applied only to individuals who were cognitively impaired with an AD syndrome at their last assessment—that is, if an individual had transient cognitive impairment but returned to cognitively unimpaired or had a non-AD diagnosis at their last assessment, the earlier impairment was not considered to be the onset of AD symptoms. For longitudinal visualization and analysis, participants were categorized based on their cognitive status at each assessment and final diagnostic outcome relative to estimated %p-tau217 positivity timing: (1) cognitively unimpaired at the assessment; (2) AD syndrome/biomarker positive: cognitively impaired at the assessment with a diagnosis of symptomatic AD at their last assessment and symptoms starting after %p-tau217 positivity; (3) AD syndrome/biomarker negative: cognitively impaired at the assessment with a diagnosis of symptomatic AD at their last assessment but symptoms starting before %p-tau217 positivity; and (4) non-AD syndrome: cognitively impaired with a non-AD diagnosis at their last assessment. Amyloid and tau PET imaging was conducted as previously described23. A mesial-temporal meta-region of interest (ROI) that included the entorhinal, parahippocampal and amygdala regions was used to assess early tau pathology (Tearly) with a corresponding positivity threshold of 1.328 standardized uptake value ratio (SUVR) derived from Gaussian mixture modeling using the mean plus 2 s.d. A temporo-parietal meta-ROI that included the superior temporal, cuneus, inferior-superior parietal, inferior-middle-superior temporal, isthmus cingulate, lateral occipital, lingual, posterior cingulate, precuneus and superior marginal was used to assess late tau pathology (Tlate) with a corresponding positivity threshold of 1.224 SUVR, using a similar approach for identifying the threshold. For analysis of variance in the rate of change in plasma %p-tau217, participants were included who had two or more %p-tau217 values at least 1 year apart. For development of clock models, the cohort was restricted to individuals who had two or more plasma %p-tau217 values between 1.06% and 10.45% at least 1 year apart. For models of age at AD symptom onset, individuals were included who were (1) initially cognitively unimpaired (CDR = 0), (2) subsequently developed cognitive impairment (CDR > 0) with an AD syndrome after estimated plasma %p-tau217 positivity and (3) were cognitively impaired (CDR > 0) with an AD syndrome at their last assessment. For visualization of predicted AD symptom onset as a function of estimated age at plasma %p-tau217 positivity, all individuals or individuals who were cognitively unimpaired at baseline were included. Clock models refer to mathematical transformations that convert biomarker levels (for example, plasma %p-tau217) into disease time (estimated years since biomarker positivity), enabling temporal staging of AD pathology progression. This approach first identifies periods of consistent biomarker change and then aligns data relative to time since estimated biomarker positivity rather than chronological age, which complements traditional longitudinal modeling methods. This terminology should be distinguished from general biological aging clocks. Our clock models are developed using the single biomarker %p-tau217, which reflects pathological processes of both amyloid plaques and neurofibrillary tangles23. Although these pathologies typically evolve jointly during the progression of AD, our approach does not explicitly model their joint evolution. Instead, by leveraging the strong associations of %p-tau217 with both amyloid and tau PET, our models may capture an integrated measure of disease progression. To determine the range of plasma %p-tau217 values over which rates of change were consistent, we quantified prediction uncertainty using squared standard errors from GAMs, which represent the variance of model-estimated rates. As was previously described for models using amyloid and tau PET13, intervals with variance in rates of change below the 90th percentile were identified. Plasma %p-tau217 values within the interval of consistent change were used for developing clock models. The TIRA approach estimates individual plasma %p-tau217 rates of change using linear mixed-effects modeling with random slopes and intercepts10,16. The rates of change are used in GAMs with cubic splines to characterize nonlinear relationships between the rates of change and plasma %p-tau217 levels at the estimated midpoint of follow-up. The SILA algorithm models longitudinal biomarker trajectories through discrete rate sampling and numerical integration17. The method estimates the first-order relationship between biomarker accumulation rate and biomarker levels by sampling rates across evenly distributed values and then applies Euler's method to numerically integrate this relationship into a non-parametric biomarker versus time curve. Both clocks were centered so that time zero was a plasma %p-tau217 value of 4.06%, which aligns with an amyloid PET Centiloid value of 20 (ref. To obtain an estimated age of plasma %p-tau217 positivity, participants with at least one %p-tau217 value between 1.06% and 10.45% were included. For example, if an 80-year-old person had a plasma %p-tau217 value of 7.06%, which corresponds to 8.8 years from %p-tau217 positivity (based on the Knight ADRC TIRA clock), their estimated age at %p-tau217 positivity would be 71.2 years (80 years minus 8.8 years). For individuals with more than one plasma %p-tau217 value, their estimated age at %p-tau217 positivity was an average of estimates from all plasma samples. Scatter plots were generated with data points being color coded by cohort for visualization, to assess the concordance of estimated ages of plasma %p-tau217 positivity between different clock models (TIRA and SILA) and across cohorts (Knight ADRC and ADNI). Associations were evaluated using metrics of adjusted R2, Spearmanʼs r and CCC. The onset of AD symptoms was defined as the first clinical assessment when initially cognitively unimpaired (CDR = 0) individuals with positive AD biomarkers (based on %p-tau217) were found to be cognitively impaired (CDR > 0) with an AD clinical syndrome. For participants who were cognitively normal at baseline, we used interval-censored and right-censored Cox proportional hazards regression models to examine the association between estimated age at plasma %p-tau217 positivity and time to cognitive impairment, accounting for the uncertainty in exact onset timing inherent in longitudinal studies. For participants who remained cognitively unimpaired throughout follow-up, survival times were right-censored at their last assessment age. To account for variable time between assessments, the time of AD symptom onset was interval-censored between the last cognitively unimpaired assessment and the first symptomatic AD assessment. In a sensitivity analysis, participants with cognitive impairment at baseline were included in the models and were left-censored. The models were fitted using the icenReg package in R with semi-parametric baseline hazard estimation. Bootstrap resampling (n = 5,000 samples) was performed to obtain robust standard errors and confidence intervals. Model discrimination was assessed using C-indexes calculated specifically for interval-censored data. Survival curves were generated to visualize the probability of remaining cognitively unimpaired across different estimated plasma %p-tau217 positivity age groups. Survival models treated estimated age at %p-tau217 positivity as a fixed covariate from the clock model estimation. For individuals with AD symptom onset after plasma %p-tau217 positivity, linear models estimated AD symptom onset as a function of the estimated age of %p-tau217 positivity. Sensitivity analyses included individuals with AD symptom onset prior to plasma %p-tau217 positivity. Sex, years of education and APOE ε4 carrier status were considered as covariates in the models but not included in the final models due to not being significant predictors. Model diagnostics were conducted to ensure the appropriateness of linear modeling. Normality of residuals was assessed using Shapiro−Wilk tests; homoscedasticity was evaluated using Breusch−Pagan tests; and linearity was confirmed through Akaike information criterion (AIC)-based model comparison and F-tests comparing linear, quadratic and cubic polynomial specifications. Sensitivity analyses were performed by refitting models after excluding observations with Cook's distance > 4/n to assess model robustness to influential points. Participant data were visualized as raster plots with estimated age at plasma %p-tau217 positivity on the y axis and estimated years from %p-tau217 positivity on the x axis, with participants ordered by their estimated positivity age and color coded by clinical diagnosis category to illustrate the temporal relationship between estimated %p-tau217 positivity and symptom onset. The estimated age at symptom onset from the linear models was overlaid on the plots to aid visualization of how the timing of symptom onset varies as a function of estimated years since plasma %p-tau217 positivity. To examine cognitive impairment risk across different temporal frameworks, we used Kaplan−Meier curves with different timelines. The primary analysis used time from baseline plasma %p-tau217 collection to onset of cognitive impairment. We also used estimated years from plasma %p-tau217 positivity and estimated years from predicted symptom onset, where predicted symptom onset age was calculated using the linear models described above. Kaplan−Meier curves were combined with density plots showing the distribution of observed events to visualize both survival probabilities and the timing of actual cognitive impairment for each outcome. Statistical comparisons were performed to evaluate differences in plasma %p-tau217 levels and estimated years from %p-tau217 positivity across the four 2024 Alzheimer's Association biological stages: stage A (normal biomarkers; equivalent to A−Tearly−Tlate−), stage B (AD pathologic change; equivalent to A+Tearly−Tlate−), stage C (AD; equivalent to A+Tearly+Tlate−) and stage D (advanced AD; equivalent to A+Tearly+Tlate+). Pairwise comparisons between all groups were conducted using non-parametric Conover−Iman tests with Benjamini−Hochberg adjustment for multiple comparisons. The same methodology for developing clocks was implemented using Fujirebio Lumipulse p-tau217/Aβ42 and p-tau217, C2N Diagnostics p-tau217, Janssen LucentAD Quanterix p-tau217 and ALZpath Quanterix p-tau217. Thresholds for each were obtained from previously work23. Statistical annotations and advanced plot arrangements used the ‘ggpubr' package. Publication-ready plot themes were achieved using the ‘cowplot' package. Multiple plots were combined and arranged using the ‘patchwork' package. Specialized distribution visualization was implemented using the ‘ggdist' package. Project management utilities included the ‘here' package for project-relative file path management and the ‘conflicted' package for function conflict resolution. Parallel computing support was enabled by the ‘doParallel' package to enhance computational efficiency. Linear mixed-effects modeling was conducted using the ‘nlme' package. Generalized additive modeling was implemented using the ‘mgcv' package. Interval-censored regression modeling was implemented using the ‘icenReg' package. CCC analysis was conducted using the ‘DescTools' package. Statistical tests and post hoc comparisons were performed with the ‘rstatix' package. Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. The data that support the findings of this study are not publicly available due to privacy restrictions and participant consent agreements that require controlled access to protect research participant confidentiality. Data from the Knight ADRC can be requested by qualified investigators (https://knightadrc.wustl.edu/professionals-clinicians/request-center-resources/). Code developed by the authors for this study is available for download from GitHub: https://github.com/WashUFluidBiomarkers/plasma_ptau217_time. Code for implementing the SILA algorithm is available at https://github.com/Betthauser-Neuro-Lab/SILA-AD-Biomarker. Jack Jr, C. R. et al. Revised criteria for diagnosis and staging of Alzheimer's disease: Alzheimer's Association Workgroup. Brain β-amyloid load approaches a plateau. Bateman, R. J. et al. Clinical and biomarker changes in dominantly inherited Alzheimer's disease. A. et al. Tau PET in autosomal dominant Alzheimer's disease: relationship with cognition, dementia and other biomarkers. Ossenkoppele, R. et al. Amyloid and tau PET-positive cognitively unimpaired individuals are at high risk for future cognitive decline. van Dyck, C. H. et al. Lecanemab in early Alzheimer's disease. Sims, J. R. et al. Donanemab in early symptomatic Alzheimer disease: the TRAILBLAZER-ALZ 2 randomized clinical trial. Predicting symptom onset in sporadic Alzheimer disease with amyloid PET. Heston, M. B. et al. Factors associated with age at tau pathology onset and time from tau onset to dementia in Alzheimer's disease. Timing of changes in Alzheimer's disease plasma biomarkers as assessed by amyloid and tau PET clocks. Budgeon, C. A. et al. Constructing longitudinal disease progression curves using sparse, short-term individual data with an application to Alzheimer's disease. Koscik, R. L. et al. Amyloid duration is associated with preclinical cognitive decline and tau PET. Betthauser, T. J. et al. Multi-method investigation of factors influencing amyloid onset and impairment in three cohorts. The AHEAD 3-45 study: design of a prevention trial for Alzheimer's disease. Updated appropriate use criteria for amyloid and tau PET: a report from the Alzheimer's Association and Society for Nuclear Medicine and Molecular Imaging Workgroup. Hampel, H. et al. Blood-based biomarkers for Alzheimer's disease: current state and future use in a transformed global healthcare landscape. Predicting continuous amyloid PET values with CSF and plasma Aβ42/Aβ40. Plasma pTau217 predicts continuous brain amyloid levels in preclinical and early Alzheimer's disease. Head-to-head comparison of leading blood tests for Alzheimer's disease pathology. Highly accurate blood test for Alzheimer's disease is similar or superior to clinical cerebrospinal fluid tests. Warmenhoven, N. et al. A comprehensive head-to-head comparison of key plasma phosphorylated tau 217 biomarker tests. Meyer, M. R. et al. Clinical validation of the PrecivityAD2 blood test: a mass spectrometry-based test with algorithm combining %p-tau217 and Aβ42/40 ratio to identify presence of brain amyloid. Head-to-head comparison of 10 plasma phospho-tau assays in prodromal Alzheimer's disease. Plasma pTau217 ratio predicts continuous regional brain tau accumulation in amyloid-positive early Alzheimer's disease. Prediction of future Alzheimer's disease dementia using plasma phospho-tau combined with other accessible measures. Plasma p-tau217 and tau-PET predict future cognitive decline among cognitively unimpaired individuals: implications for clinical trials. Plasma Aβ42/Aβ40 and phospho-tau217 concentration ratios increase the accuracy of amyloid PET classification in preclinical Alzheimer's disease. The diagnosis of dementia due to Alzheimer's disease: recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease. The diagnosis of mild cognitive impairment due to Alzheimer's disease: recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease. Dubois, B. et al. Alzheimer disease as a clinical-biological construct—an international working group recommendation. Symptom onset in autosomal dominant Alzheimer disease: a systematic review and meta-analysis. The DIAN-TU Next Generation Alzheimer's prevention trial: adaptive design and disease progression model. Plasma MTBR-tau243 biomarker identifies tau tangle pathology in Alzheimer's disease. Du, L. et al. Onset ages of cerebrovascular disease and amyloid and effects on cognition in risk-enriched cohorts. Schindler, S. E. Predicting symptom onset in sporadic Alzheimer's disease: ‘How long do I have?' Appropriate use criteria for lumbar puncture and cerebrospinal fluid testing in the diagnosis of Alzheimer's disease. Palmqvist, S. et al. Alzheimer's Association Clinical Practice Guideline on the use of blood-based biomarkers in the diagnostic workup of suspected Alzheimer's disease within specialized care settings. & Rabinovici, G. D. Expert opinion on Centiloid thresholds suitable for initiating anti-amyloid therapy. Summary of discussion at the 2024 spring Alzheimer's Association Research Roundtable. Improved protocol for measurement of plasma β-amyloid in longitudinal evaluation of Alzheimer's Disease Neuroimaging Initiative study patients. Shaw, L. M. et al. ADNI Biomarker Core: a review of progress since 2004 and future challenges. Morris, J. C. The Clinical Dementia Rating (CDR): current version and scoring rules. This study represents results of the Foundation for the National Institutes of Health (FNIH; https://fnih.org/) Biomarkers Consortium ‘Biomarkers Consortium, Plasma Aβ and Phosphorylated Tau as Predictors of Amyloid and Tau Positivity in Alzheimer's Disease' project, which was made possible through a public−private partnership managed by the FNIH and funded by AbbVie Inc., the Alzheimer's Association, the Diagnostics Accelerator at the Alzheimer's Drug Discovery Foundation, Biogen, Janssen Research & Development LLC and Takeda. We are grateful for the contributions of the following project team members: A. Bannon (AbbVie), M. Baratta (Takeda), J. Coomaraswamy (Takeda), J. Dage (Indiana University), I. Dobler (Takeda), L. Du-Cuny (AbbVie), K. Ferber (Biogen), J. Hsiao (National Institute on Aging (NIA)), H. Kolb (formerly with Johnson & Johnson Innovative Medicine), E. Meyers (Alzheimer's Association), Y. Mordashova (AbbVie), W. Potter, M. Quinton (AbbVie), D. Raunig (Takeda), E. Rosenbaugh (FNIH), C. Rubel (Biogen), Z. Saad (Johnson & Johnson Innovative Medicine), M. Sabandal (FNIH), P. Saletti (Alzheimer's Drug Discovery Foundation), S. Schindler (Washington University in St. Louis), L. Shaw (University of Pennsylvania), G. Triana-Baltzer (Johnson & Johnson Innovative Medicine), C. Weber (Alzheimer's Association) and H. Zetterberg (University of Gothenburg). This study was also supported by NIA grants R01AG070941 (S.E.S. Data collection and sharing for this project was funded by the Alzheimer's Disease Neuroimaging Initiative (ADNI) (NIH grant U01 AG024904) and DOD ADNI (Department of Defense award number W81XWH-12-2-0012). The ADNI is funded by the NIA and the National Institute of Biomedical Imaging and Bioengineering and through generous contributions from the following: AbbVie, the Alzheimer's Association; the Alzheimer's Drug Discovery Foundation; Araclon Biotech; BioClinica Inc.; Biogen; Bristol Myers Squibb Company; CereSpir Inc.; Cogstate; Eisai Inc.; Elan Pharmaceuticals Inc.; Eli Lilly and Company; EuroImmun; F. Hoffmann-La Roche Ltd. and its affiliated company, Genentech Inc.; Fujirebio; GE Healthcare; IXICO Ltd.; Janssen Alzheimer Immunotherapy Research & Development LLC; Johnson & Johnson Pharmaceutical Research & Development LLC; Lumosity; Lundbeck; Merck & Co. Inc.; Meso Scale Diagnostics LLC; NeuroRx Research; Neurotrack Technologies; Novartis Pharmaceuticals Corporation; Pfizer Inc.; Piramal Imaging; Servier; Takeda Pharmaceutical Company; and Transition Therapeutics. The Canadian Institutes of Health Research is providing funds to support ADNI clinical sites in Canada. Private sector contributions are facilitated by the FNIH. ADNI data are disseminated by the Laboratory for Neuro Imaging at the University of Southern California. A complete listing of ADNI investigators can be found at https://adni.loni.usc.edu/wp-content/uploads/how_to_apply/ADNI_Acknowledgement_List.pdf.0. A full list of members and their affiliations appears in the supplementary information. Kellen K. Petersen, Yan Li, Benjamin Saef, Eric McDade, David M. Holtzman, John C. Morris, Randall J. Bateman & Suzanne E. Schindler Northern California Institute for Research and Education, San Francisco, CA, USA Department of Radiology and Biomedical Imaging, University of California, San Francisco, San Francisco, CA, USA Wisconsin Alzheimer's Institute, University of Wisconsin-Madison School of Medicine and Public Health, Madison, WI, USA Rush Alzheimer's Disease Center, Chicago, IL, USA Department of Neurological Sciences, Rush University Medical Center, Chicago, IL, USA Knight Alzheimer Disease Research Center, St. Louis, MO, USA Chengjie Xiong, Carlos Cruchaga, Eric McDade, David M. Holtzman, John C. Morris, Randall J. Bateman & Suzanne E. Schindler AbbVie Deutschland GmbH & Co. KG, Ludwigshafen am Rhein, Germany Stark Neurosciences Research Institute, Indiana University School of Medicine, Indianapolis, IN, USA Banner Alzheimer's Institute, Phoenix, AZ, USA Clinical Neurochemistry Laboratory, Sahlgrenska University Hospital, Mölndal, Sweden Foundation for the National Institutes of Health, North Bethesda, MD, USA NeuroGenomics and Informatics Center, Washington University, St. Louis, MO, USA Hope Center for Neurologic Diseases, St. Louis, MO, USA Carlos Cruchaga, David M. Holtzman, Randall J. Bateman & Suzanne E. Schindler Tracy Family SILQ Center for Neurodegenerative Biology, St. Louis, MO, USA contributed to the conceptualization and study design. All authors edited and approved the final version for publication and agree to be accountable for all aspects of the work. has served as a consultant for Eli Lilly and Company. is the co-inventor of the technology ‘Novel Tau isoforms to predict onset of symptoms and dementia in Alzheimer's disease', which is in the process of licensing by C2N Diagnostics. are employed by Johnson & Johnson Innovative Medicine and may receive salary and stock for their employment. are employed by AbbVie Deutschland GmbH & Co. KG. are employees of and may own stock in Biogen. is an inventor on patents or patent applications of Eli Lilly and Company relating to the assays, methods, reagents and/or compositions of matter for P-tau assays and Aβ-targeting therapeutics. has served or is serving as a consultant or on advisory boards for Eisai, AbbVie, Genotix Biotechnologies Inc., Gates Ventures, Gate Neurosciences, Dolby Family Ventures, Karuna Therapeutics, AlzPath Inc., Cognito Therapeutics, Inc., Prevail Therapeutics, Neurogen Biomarking, Spear Bio, the University of Kentucky, Rush University, Tymora and Quanterix. has received research support from ADx Neurosciences, Fujirebio, Roche Diagnostics and Eli Lilly and Company in the past 2 years. is a founder of and advisor to Monument Biosciences. has stock or stock options in Eli Lilly and Company, Genotix Biotechnologies, AlzPath Inc., Neurogen Biomarking and Monument Biosciences. has received speaking fees from Eli Lilly, Biogen, Quanterix and Alamar Biosciences. has served on scientific advisory boards and/or as a consultant for AbbVie, Acumen, Alector, Alzinova, ALZpath, Amylyx, Annexon, Apellis, Artery Therapeutics, AZTherapies, Cognito Therapeutics, CogRx, Denali, Eisai, LabCorp, Merry Life, Nervgen, Novo Nordisk, Optoceutics, Passage Bio, Pinteon Therapeutics, Prothena, Red Abbey Labs, reMYND, Roche, Samumed, Siemens Healthineers, Triplet Therapeutics and Wave; has given lectures sponsored by Alzecure, BioArctic, Biogen, Cellectricon, Fujirebio, Eli Lilly, Novo Nordisk, Roche and WebMD; and is a co-founder of Brain Biomarker Solutions in Gothenburg AB (BBS), which is a part of the GU Ventures Incubator Program (outside submitted work). has received research support from GlaxoSmithKline and Eisai. is also a member of the scientific advisory board of Admit and has served on the scientific advisory boards of GlaxoSmithKline and Novo Nordisk. has had advisory board roles, consulting and data safety monitoring board (DSMB) relationships with Eli Lilly, Alnylam, Alector, Alzamend, Sanofi, AstraZeneca, F. Hoffmann-La Roche, Grifols and Merck. receive income from C2N Diagnostics for serving on the scientific advisory board. is a co-inventor on the following US patent applications: ‘Methods to detect novel tau species in CSF and use thereof to track tau neuropathology in Alzheimer's disease and other tauopathies' and ‘CSF phosphorylated tau and amyloid beta profiles as biomarkers of Tauopathies'. is a co-inventor on a non-provisional patent application: ‘Methods of diagnosing and treating based on site-specific tau phosphorylation'. is an unpaid scientific advisory board member of Roche and Biogen and receives research funding from Avid Radiopharmaceuticals, Janssen, Roche/Genentech, Eli Lilly, Eisai, Biogen, AbbVie, Bristol Myers Squibb and Novartis. receives salary and company stock as compensation for his employment with AbbVie. was previously employed by the National Institute of Mental Health and is a stockholder in Merck & Co. W.Z.P. is a Co-Chair Emeritus for the FNIH Biomarkers Consortium Neuroscience Steering Committee. serves as a consultant for Karuna, Neurocrine, Neumarker and Vaaji and receives grant support from the NIA along with stock options from Praxis Bioresearch. has served on scientific advisory boards on biomarker testing and education for Eisai and Novo Nordisk and has received speaking fees for presentations on biomarker testing from Eisai, Eli Lilly, Novo Nordisk, Medscape and PeerView. She has provided unpaid scientific advising to Eisai, Johnson & Johnson Innovative Medicine, Eli Lilly, Biogen, Acumen, Cognito Therapeutics and Danaher. Nature Medicine thanks Junhao Wen and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editor: Jerome Staal, in collaboration with the Nature Medicine team. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Variance analysis was performed to identify biomarker ranges with relatively consistent rates of change (below the 90th percentile) for five plasma p-tau217 assays: C2N PrecivityAD2 %p-tau217 (A), Fujirebio Lumipulse p-tau217/Aβ42 (B), Fujirebio Lumipulse p-tau217 (C), ALZpath Quanterix p-tau217 (D), and Janssen LucentAD Quanterix p-tau217 (E). The rate of change in each biomarker was modeled using Generalized Additive Models (GAMs), and variance in the rate of change was calculated across biomarker values. Red dashed vertical lines mark the boundaries of intervals where variance in rates of change was below the 90th percentile, defining the biomarker ranges suitable for clock model development. Blue points indicate intervals with variance below the 90th percentile; red points indicate intervals with variance above the 90th percentile. The estimated age at plasma %p-tau217 positivity based on clock models is shown as a function of the observed conversion age, which is the average of the age at last negative and first positive %p-tau217. Clock models in the Knight ADRC (A, B) and ADNI cohorts (C, D) were created with the TIRA (A, C) and SILA (B, D) approaches. Each point represents an individual participant. Blue lines represent linear regression fits, and gray shaded bands represent 95% confidence intervals for the regression line. Adjusted R² and Spearman correlation coefficients (ρ) are shown for each model. For individuals in the Knight ADRC (A, B) and ADNI (C, D) cohorts, age at plasma %p-tau217 positivity was estimated using either the TIRA (A, C) or SILA (B, D) models. Each row represents the longitudinal clinical diagnoses for one individual by estimated years from %p-tau217 positivity (x-axis). Individuals are sorted vertically by estimated age at %p-tau217 positivity (y-axis). The point color denotes the clinical diagnosis: blue was cognitively unimpaired at the assessment; red (AD syndrome/biomarker positive) was cognitively impaired at the assessment and had a diagnosis of symptomatic AD at their last assessment with symptoms starting after %p-tau217 positivity; purple (AD syndrome/biomarker negative) was cognitively impaired at the assessment and had a diagnosis of symptomatic AD at their last assessment with symptoms starting before %p-tau217 positivity; orange (non-AD syndrome) was cognitively impaired and had a non-AD diagnosis at their last assessment. Vertical dashed lines at 0 represent the estimated time of %p-tau217 positivity. Brown lines indicate the relationship between %p-tau217 positivity age and estimated symptom onset based on the Knight ADRC models in Fig. Using data from ADNI, raincloud plots show plasma %p-tau217 levels (A), estimated years since %p-tau217 positivity by TIRA (B) and SILA (C), and years since estimated symptom onset by TIRA (D) and SILA (E) stratified by the 2024 Alzheimer's Association biological staging framework: Stage A (normal biomarkers), Stage B (Alzheimer's disease pathologic change), Stage C (Alzheimer's disease), and Stage D (advanced Alzheimer's disease). The n values provided for each group indicate the number of individual human participants included in that biological stage; each participant contributes one data point to the analysis. Group means with 95% confidence intervals are shown. Statistical comparisons between all biological stages were conducted using two-sided non-parametric Conover-Iman tests with Benjamini-Hochberg adjustment for multiple comparisons. Longitudinal plasma p-tau217/Aβ42 data from ADNI is shown as a function of age (A) or estimated years from p-tau217/Aβ42 positivity by TIRA (B) or SILA clock models (C). Thick black lines represent the clock models, red lines represent individuals with at least one plasma p-tau217/Aβ42 > 0.006312, and grey lines represent individuals with no plasma p-tau217/Aβ42 > 0.006312. Models for age at symptom onset included individuals who were initially cognitively unimpaired but had a typical AD syndrome at their last assessment and developed symptoms after estimated plasma p-tau217/Aβ42 positivity. Age at p-tau217/Aβ42 positivity was estimated using TIRA (D) or SILA (E) models. Each point represents an individual participant. In D and E, black lines represent linear regression fits showing the predicted mean age at symptom onset, and gray shaded bands indicate 95% confidence intervals around the regression lines. Linear regression equations, adjusted R² values, Spearman correlation coefficients (ρ), and sample sizes (N) are shown. Longitudinal plasma p-tau217 data from ADNI is shown as a function of age (A) or estimated years from p-tau217 positivity by TIRA (B) or SILA clock models (C). Thick black lines represent the clock models; red lines represent individuals with at least one plasma p-tau217 > 0.0615 pg ml−1 and grey lines represent individuals with no plasma p-tau217 > 0.0615 pg ml−1. Models for age at symptom onset included individuals who were initially cognitively unimpaired but had a typical AD syndrome at their last assessment and developed symptoms after estimated plasma p-tau217 positivity. Age at p-tau217 positivity was estimated using TIRA (D) or SILA (E) models. In D and E, black lines represent linear regression fits showing the predicted mean age at symptom onset, and gray shaded bands indicate 95% confidence intervals around the regression lines. Linear regression equations, adjusted R² values, Spearman correlation coefficients (ρ), and sample sizes (N) are shown. Longitudinal plasma p-tau217 data from ADNI is shown as a function of age (A) or estimated years from p-tau217 positivity by TIRA (B) or SILA clock models (C). Thick black lines represent the clock models; red lines represent individuals with at least one plasma p-tau217 > 0.444 pg ml−1 and grey lines represent individuals with no plasma p-tau217 > 0.444 pg ml−1. Models for age at symptom onset included individuals who were initially cognitively unimpaired but had a typical AD syndrome at their last assessment and developed symptoms after estimated plasma p-tau217 positivity. Age at p-tau217 positivity was estimated using TIRA (D) or SILA (E) models. In D and E, black lines represent linear regression fits showing the predicted mean age at symptom onset, and gray shaded bands indicate 95% confidence intervals around the regression lines. Linear regression equations, adjusted R² values, Spearman correlation coefficients (ρ), and sample sizes (N) are shown. Animation of C2N Diagnostics' PrecivityAD2 plasma %p-tau217 trajectories transitioning between plots versus age and years since plasma %p-tau217 positivity. The animation shows longitudinal plasma %p-tau217 data plotted as a function of age and years since plasma positivity. Red lines represent individuals with at least one plasma %p-tau217 > 4.06%; gray lines represent individuals with no plasma %p-tau217 > 4.06%; and thick black lines represent the clock models. Animation of C2N Diagnostics' PrecivityAD2 plasma p-tau217 trajectories transitioning between plots versus age and years since plasma p-tau217 positivity. The animation shows longitudinal plasma p-tau217 data plotted as a function of age and years since plasma positivity. Red lines represent individuals with at least one plasma p-tau217 > 2.34 pg ml−1; gray lines represent individuals with no plasma p-tau217 > 2.34 pg ml−1; and thick black lines represent the clock models. Animation of Fujirebio Diagnostics' Lumipulse plasma p-tau217/Aβ42 trajectories transitioning between plots versus age and years since plasma p-tau217/Aβ42 positivity. The animation shows longitudinal plasma p-tau217/Aβ42 data plotted as a function of age and years since plasma positivity. Red lines represent individuals with at least one plasma p-tau217/Aβ42 > 0.006312; gray lines represent individuals with no plasma p-tau217/Aβ42 > 0.006312; and thick black lines represent the clock models. Animation of Fujirebio Diagnostics' Lumipulse plasma p-tau217 trajectories transitioning between plots versus age and years since plasma p-tau217 positivity. The animation shows longitudinal plasma p-tau217 data plotted as a function of age and years since plasma positivity. Red lines represent individuals with at least one plasma p-tau217 > 0.158 pg ml−1; gray lines represent individuals with no plasma p-tau217 > 0.158 pg ml−1; and thick black lines represent the clock models. Animation of Janssen's LucentAD Quanterix plasma p-tau217 trajectories transitioning between plots versus age and years since plasma positivity. The animation shows longitudinal plasma p-tau217 data plotted as a function of age and years since plasma positivity. Red lines represent individuals with at least one plasma p-tau217 > 0.0615 pg ml−1; gray lines represent individuals with no plasma p-tau217 > 0.0615 pg ml−1; and thick black lines represent the clock models. Animation of ALZpath's Quanterix plasma p-tau217 trajectories transitioning between plots versus age and years since plasma p-tau217 positivity. The animation shows longitudinal plasma p-tau217 data plotted as a function of age and years since plasma positivity. Red lines represent individuals with at least one plasma p-tau217 > 0.444 pg ml−1; gray lines represent individuals with no plasma p-tau217 > 0.444 pg ml−1; and thick black lines represent the clock models. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. Predicting onset of symptomatic Alzheimerʼs disease with plasma p-tau217 clocks. Anyone you share the following link with will be able to read this content: Sorry, a shareable link is not currently available for this article. Provided by the Springer Nature SharedIt content-sharing initiative Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Alzheimer's blood tests predict what age people will be when the disease may cause symptoms, study finds Tests that could reveal when Alzheimer's disease will emerge, while promising, are not ready for use in otherwise healthy people, scientists say Blood tests that detect a protein involved in Alzheimer's disease could help predict the age at which the disease may strike people long before they develop symptoms, according to a new study. But questions remain about the accuracy and uncertainty of these tests, and experts caution that the assays aren't ready for prime time. “While the results here are encouraging, they are not yet at the level of having significant clinical benefit for individual patients,” says Corey Bolton, a clinical neuropsychologist and an assistant professor of medicine at Vanderbilt University Medical Center, who was not involved in the new study. The study included more than 600 people aged 62 to 78 who were not cognitively impaired. They had blood tests to detect a protein called p-tau217, which accumulates in the brains of people with Alzheimer's. If you're enjoying this article, consider supporting our award-winning journalism by subscribing. “A key innovation was estimating when they're going to develop symptoms,” says Suzanne Schindler, an associate professor of neurology at the Washington University School of Medicine in St. Louis, who co-authored the study. Several of the study authors have consulted for or received funding from companies that make these Alzheimer's blood tests. More than seven million Americans are living with Alzheimer's disease, and there is no cure. The neurodegenerative condition is associated with the buildup of plaques of amyloid protein and tangles of tau protein in the brain, which can develop for a decade or more before visible symptoms such as memory loss or confusion arise. Blood tests are increasingly used to detect biological signs of the disease. They are much cheaper and easier to administer than traditional diagnostics such as spinal taps or positron-emission tomography (PET) scans. But these tests may not always accurately predict who will and won't develop Alzheimer's, experts say. Detecting Alzheimer's before symptoms show up, however, may be crucial to treating it: although there is no cure for the condition, two drugs have been approved that can slow the rate of progression in some people when the disease is caught early. And there are clinical trials of these drugs underway to determine whether treatment could head off the disease in people who have biological signs of the disease but no symptoms. They found that these blood, or plasma, “clocks” could predict how likely and when people would develop symptoms of the disease. It's important to note that the researchers are “not recommending this for people who are asymptomatic,” says Zaldy Tan, a memory and aging specialist at the Cedars-Sinai Medical Center in Los Angeles. And a three- to four-year error margin on either side of diagnosis is “a big window,” he notes, especially if you're using the knowledge to make decisions about retirement plans or finances. “Other medical conditions, such as chronic kidney disease and obesity, seem to have a large impact on the circulating levels of these proteins and can greatly influence results, leading to false positives or false negatives,” Bolton says. This study used a type of test that limits the effect of these conditions, he says, but “there are still many unanswered questions about how these blood tests perform in diverse populations.” Despite their limits, however, the tests are still valuable for diagnosis and planning treatment, Bolton says. People found to be at greater risk of developing the disease could still benefit from interventions such as exercise, a healthy diet and cognitive or social stimulation. Nathaniel Chin, a geriatrician and medical director at the Wisconsin Alzheimer's Disease Research Center in Madison, who was not involved in the study, is “impressed and excited” by its results. He hopes researchers will replicate the findings in other populations. Tanya Lewis is senior desk editor for health and medicine at Scientific American. Previously, she has written for outlets that include Insider, Wired, Science News and others. She has a degree in biomedical engineering from Brown University and one in science communication from the University of California, Santa Cruz. If you enjoyed this article, I'd like to ask for your support. Scientific American has served as an advocate for science and industry for 180 years, and right now may be the most critical moment in that two-century history. If you subscribe to Scientific American, you help ensure that our coverage is centered on meaningful research and discovery; that we have the resources to report on the decisions that threaten labs across the U.S.; and that we support both budding and working scientists at a time when the value of science itself too often goes unrecognized. In return, you get essential news, captivating podcasts, brilliant infographics, can't-miss newsletters, must-watch videos, challenging games, and the science world's best writing and reporting. There has never been a more important time for us to stand up and show why science matters.

Their work has produced a searchable resource containing 67,573 magnetic compounds, including 25 materials that had not previously been recognized as magnets capable of staying magnetic at high temperatures. "By accelerating the discovery of sustainable magnetic materials, we can reduce dependence on rare earth elements, lower the cost of electric vehicles and renewable-energy systems, and strengthen the U.S. manufacturing base," said Suman Itani, lead author and a doctoral student in physics. Magnets are critical components in smartphones, medical devices, power generators, electric vehicles, and many other everyday systems. However, today's most powerful magnets depend on rare earth elements that are costly, largely imported, and increasingly difficult to secure. Despite the large number of known magnetic compounds, no entirely new permanent magnet has been identified from this pool. Researchers have long understood that many magnetic materials likely remain undiscovered. Yet testing every possible combination of elements, which could number in the millions, would take enormous amounts of time and money in a laboratory setting. "We are tackling one of the most difficult challenges in materials science -- discovering sustainable alternatives to permanent magnets -- and we are optimistic that our experimental database and growing AI technologies will make this goal achievable," said Jiadong Zang, physics professor and co-author. The research team also includes co-author Yibo Zhang, a postdoctoral researcher in both physics and chemistry. Looking ahead, the scientists believe the large language model used in this project could serve purposes beyond building this database, especially in higher education. For example, the technology could convert images into modern rich text formats, helping update and preserve library collections. Stay informed with ScienceDaily's free email newsletter, updated daily and weekly. Keep up to date with the latest news from ScienceDaily via social networks: Tell us what you think of ScienceDaily -- we welcome both positive and negative comments.

Sodium-ion batteries are emerging as a promising option for cleaner, more sustainable energy storage. Researchers at the University of Surrey have identified a surprisingly simple way to improve their performance by keeping water inside a critical battery material instead of removing it. Even so, matching the performance of lithium-ion technology has been a major hurdle for sodium-ion systems. They discovered that allowing the material to retain its natural water content significantly enhances how it functions inside a battery. The compound, called nanostructured sodium vanadate hydrate (NVOH), delivered far stronger results when used in its hydrated form. This performance places it among the top cathodes reported so far for sodium-ion batteries. Sodium vanadium oxide has been around for years, and people usually heat-treat it to remove the water because it's thought to cause problems. The material showed much stronger performance and stability than expected and could even create exciting new possibilities for how these batteries are used in the future." "Being able to use sodium vanadate hydrate in salt water is a really exciting discovery, as it shows sodium-ion batteries could do more than just store energy -- they could also help remove salt from water. In the long term, that means we might be able to design systems that use seawater as a completely safe, free and abundant electrolyte, while also producing fresh water as part of the process." This advance could speed up the adoption of sodium-ion batteries as a practical alternative to lithium-based technology. By simplifying the production of high-performance sodium-ion batteries, the Surrey team's findings move commercially viable, sustainable energy storage one step closer to reality. Note: Content may be edited for style and length. Scientists Say Your Fingers Hold a Secret of Brain Evolution Man Survives 48 Hours Without Lungs in a Medical First Stay informed with ScienceDaily's free email newsletter, updated daily and weekly. Keep up to date with the latest news from ScienceDaily via social networks: Tell us what you think of ScienceDaily -- we welcome both positive and negative comments.

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript. (2026)Cite this article Metrics details Organic batteries using abundant and recyclable organic electrode materials provide a sustainable and environmentally friendly alternative to commercial lithium-ion batteries1,2,3,4,5, which rely on resource-limited mineral-derived inorganic electrode materials6,7,8. However, the practical use of organic batteries has been severely hindered by the intrinsic insulation and dissolution of organic electrode materials9,10. Here we report practical organic batteries using an n-type conducting polymer cathode, poly(benzodifurandione) (PBFDO), which exhibits excellent mixed ionic and electronic transport and low solubility. The PBFDO cathode maintains its n-doped state throughout the electrochemical processes and exhibits stable and reversible redox characteristics, high electrical conductivities and significant lithium-ion diffusion coefficients, without the need for additional conductive additives. Consequently, ultrahigh-mass-loading polymer cathodes, with mass loadings up to 206 mg cm−2, are realized, delivering a high areal capacity of 42 mAh cm−2 and demonstrating robust cycling stability. Furthermore, practical 2.5 Ah lithium–organic pouch cells were fabricated, achieving an impressive energy density of 255 Wh kg−1. Notably, the conducting polymer cathode operates efficiently over a wide temperature range from −70 °C to 80 °C and demonstrates excellent flexibility and safety, marking considerable potential for applications in extreme conditions and wearable electronics. This is a preview of subscription content, access via your institution Access Nature and 54 other Nature Portfolio journals Get Nature+, our best-value online-access subscription cancel any time Subscribe to this journal Receive 51 print issues and online access $199.00 per year only $3.90 per issue Buy this article Prices may be subject to local taxes which are calculated during checkout The data supporting the findings of this study are available in the paper and its Supplementary Information. The data of this study are available from the corresponding authors upon reasonable request. Nguyen, T. P. et al. Polypeptide organic radical batteries. Prospects of organic electrode materials for practical lithium batteries. Kim, J. et al. Organic batteries for a greener rechargeable world. Dai, H., Guan, L., Mao, M. & Wang, C. J. Evaluating the present and future of organic batteries. Clean Technol. Li, M. et al. Electrolytes in organic batteries. Li, M. & Lu, J. Cobalt in lithium-ion batteries. Deng, T. et al. Designing in-situ-formed interphases enables highly reversible cobalt-free LiNiO2 cathode for Li-ion and Li-metal batteries. Ogihara, N. et al. Direct capacity regeneration for spent Li-ion batteries. Bai, S. et al. Permselective metal–organic framework gel membrane enables long-life cycling of rechargeable organic batteries. Li, M. et al. Soluble organic cathodes enable long cycle life, high rate, and wide-temperature lithium-ion batteries. Chen, Z. et al. A nitroaromatic cathode with an ultrahigh energy density based on six-electron reaction per nitro group for lithium batteries. Schön, T. B., McAllister, B. T., Li, P.-F. & Seferos, D. S. The rise of organic electrode materials for energy storage. Article PubMed Lee, M. et al. High-performance sodium–organic battery by realizing four-sodium storage in disodium rhodizonate. Luo, C. et al. Azo compounds derived from electrochemical reduction of nitro compounds for high performance Li-ion batteries. Sang, P., Chen, Q., Wang, D.-Y., Guo, W. & Fu, Y. Organosulfur materials for rechargeable batteries: structure, mechanism, and application. Xiong, P. et al. Thiourea-based polyimide/RGO composite cathode: a comprehensive study of storage mechanism with alkali metal ions. China Mater. Guo, J. et al. Revealing hydrogen bond effect in rechargeable aqueous zinc-organic batteries. Cong, G., Wang, W., Lai, N.-C., Liang, Z. & Lu, Y.-C. A high-rate and long-life organic-oxygen battery. Chen, Z. et al. Anion chemistry enabled positive valence conversion to achieve a record high-voltage organic cathode for zinc batteries. Wang, J. et al. Conjugated sulfonamides as a class of organic lithium-ion positive electrodes. Suga, T., Ohshiro, H., Sugita, S., Oyaizu, K. & Nishide, H. Emerging n-type redox-active radical polymer for a totally organic polymer-based rechargeable battery. Li, Z. et al. A small molecular symmetric all-organic lithium-ion battery. Zhao, C. et al. In situ electropolymerization enables ultrafast long cycle life and high-voltage organic cathodes for lithium batteries. Yu, Z. et al. Redox-active donor-acceptor conjugated microporous polymer for high-voltage and high-rate symmetric all-organic lithium-ion battery. Today Energy 53, 101995 (2025). Song, Z. et al. Polyanthraquinone as a reliable organic electrode for stable and fast lithium storage. Deng, X. et al. Ultrafast charging of two-dimensional polymer cathodes enabled by cross-flow structure design. Luo, L. et al. A redox-active conjugated microporous polymer cathode for high-performance lithium/potassium-organic batteries. China Chem. Kolek, M. et al. Ultra-high cycling stability of poly(vinylphenothiazine) as a battery cathode material resulting from π–π interactions. Energy Environ. Liang, Y. et al. Heavily n-dopable π-conjugated redox polymers with ultrafast energy storage capability. Peng, C. et al. Reversible multi-electron redox chemistry of π-conjugated N-containing heteroaromatic molecule-based organic cathodes. Lu, D. et al. Ligand-channel-enabled ultrafast Li-ion conduction. Tang, H. et al. A solution-processed n-type conducting polymer with ultrahigh conductivity. Jin, Z. et al. Iterative synthesis of contorted macromolecular ladders for fast-charging and long-life lithium batteries. Qin, J. et al. A metal-free battery with pure ionic liquid electrolyte. Ke, Z. et al. Controlled de-doping and redoping of n-doped poly(benzodifurandione) (n-PBDF). Li, Z. et al. Electrolyte design enables rechargeable LiFePO4/graphite batteries from −80 °C to 80 °C. Dong, X., Guo, Z., Guo, Z., Wang, Y. & Xia, Y. Organic batteries operated at −70 °C. Asl, H. Y. & Manthiram, A. Reining in dissolved transition-metal ions. Feng, X., Ren, D., He, X. & Ouyang, M. Mitigating thermal runaway of lithium-ion batteries. Liu, D. et al. Controlled large-area lithium deposition to reduce swelling of high-energy lithium metal pouch cells in liquid electrolytes. Muench, S. et al. Polymer-based organic batteries. Tang, H. et al. Highly conductive alcohol-processable n-type conducting polymer enabled by finely tuned electrostatic interactions for green organic electronics. Neese, F. Software update: the ORCA program system—version 5.0. Lu, T. & Chen, F. Multiwfn: a multifunctional wavefunction analyzer. Wang, B. et al. Diffusion coefficients during regenerated cellulose fibers formation using ionic liquids as solvents: experimental investigation and molecular dynamics simulation. This work was financially supported by the Fundamental and Interdisciplinary Disciplines Breakthrough Plan of the Ministry of Education of China (JYB2025XDXM410), the National Natural Science Foundation of China (22579126, 22179092, 52433012 and 52303227), the Fundamental Research Funds for the Central Universities (2024ZYGXZR076) and the China Postdoctoral Science Foundation (2024T170286 and 2023M741201). acknowledges support from the New Cornerstone Science Foundation through the XPLORER PRIZE. These authors contributed equally: Zhenfei Li, Haoran Tang School of Materials Science and Engineering, State Key Laboratory of Advanced Materials for Intelligent Sensing, National Industry-Education Platform for Energy Storage, Tianjin University, Tianjin, China Zhenfei Li, Yuansheng Liu, Mengjie Li, Lanhua Ma, Hongpeng Chen, Yanhou Geng & Yunhua Xu Institute of Polymer Optoelectronic Materials and Devices, Guangdong Basic Research Center of Excellence for Energy & Information Polymer Materials, State Key Laboratory of Luminescent Materials and Devices, South China University of Technology, Guangzhou, China Haoran Tang, Yuanying Liang, Yining Wang, Shaohua Tong, Qinglin Jiang, Yuguang Ma, Yong Cao & Fei Huang Guangdong Artificial Intelligence and Digital Economy Laboratory (Guangzhou), Guangzhou, China Yuanying Liang School of Materials Science and Engineering, Zhejiang University, Hangzhou, China Xiaoyu Zhai & Jiangwei Wang Department of Materials Science and Engineering, Southern University of Science and Technology (SUSTech), Shenzhen, China Xianbin Wei Eastern Institute for Advanced Study, Eastern Institute of Technology, Ningbo, China Meng Danny Gu Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar conceived the project under the supervision of Y.C., Y.X. synthesized materials, performed characterizations and assembled batteries. Z.L., H.T., Y. Liang and Y.G. discussed and analysed the data. Y. Liu and H.C. conducted the DFT and molecular dynamics calculations. assisted with data curation and manuscript revision. assisted in materials synthesis and characterizations. performed the Hall effect measurements. carried out the HRTEM characterization. designed and executed the cryo-TEM experiments. wrote and revised the paper, and all authors read and approved the paper. Correspondence to Yunhua Xu or Fei Huang. The authors declare no competing interests. Nature thanks Matthieu Becuwe and Nagaraj Patil for their contribution to the peer review of this work. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. a, Flexibility test of the self-supporting PBFDO cathode. b, Digital microscopy and SEM images of the flexible PBFDO cathode at different bending states. c-d, Bending test of the flexible PBFDO cathode: photographs of the test process (c) and bending cycle life of 75,000 cycles (d). The cycle life was indicated by the stress variation with test time. e-f, Photographs of the flexible Li ||PBFDO pouch cell at different bending states. g, Cycling stability of the flexible Li ||PBFDO pouch cell under different bending conditions. a, Schematical illustration of the lithium storage processes. Owing to the complexity of conducting polymer resonance and lithium-ion storage mechanisms, coupled with the uncertainty regarding the positions of counter-ions, only a schematic representation of one possible resonance pathway was provided here. The pristine PBFDO, with an n-doping level of around 90%, was first discharged to 1.5 V by lithium-ion uptake. During the subsequent charge process, lithium ions were extracted, while approximately half of protons were also removed, leaving the remaining protons preserved within PBFDO, which also contributed to PBFDO maintaining its n-doped state. Afterward, PBFDO experienced reversible electrochemical reactions, with some protonated carbonyl groups persisting, leading to a reversible capacity of 230.4 mAh g−1. b, XPS spectra of the PBFDO cathode at different charge/discharge states. c, Voltage profiles of the PBFDO cathode at 50 mA g−1 with marked voltages for the FTIR tests. d, FTIR spectra of the PBFDO cathode at different charge/discharge states as indicated in c. e, In situ Raman spectra of the PBFDO cathodes during charge/discharge processes. This file contains Supplementary Sections A–X, including Supplementary Figs. 1–53 and Supplementary Tables 1–9. Live demonstration of the bending endurance test of the PBFDO cathode. The PBFDO cathode can withstand 75,000 stretch cycles. This video shows the remarkable mechanical stability of the PBFDO cathode, highlighting its potential for applications requiring high durability under repeated mechanical deformation. Live demonstration showcasing the flexibility of the PBFDO cathode in comparison to a commercial inorganic cathode. The flexibility test of commercial inorganic cathode is shown from 36″ to 4′13″, and PBFDO cathode is shown from 4′24″ to 7′16″. The video highlights the superior mechanical flexibility of the PBFDO cathode, highlighting its durability and potential for flexible batteries. Live footage of the nail penetration test conducted on a 2.5 Ah PBFDO pouch cell. The test demonstrates the safety and robustness of the PBFDO pouch cell. No deformation, gas production or explosion was observed during or after the penetration, highlighting the exceptional stability of the PBFDO pouch cell under extreme mechanical stress. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. Reprints and permissions Li, Z., Tang, H., Liang, Y. et al. Practical lithium–organic batteries enabled by an n-type conducting polymer. Accepted: 22 January 2026 Version of record: 18 February 2026 DOI: https://doi.org/10.1038/s41586-026-10174-7 Anyone you share the following link with will be able to read this content: Sorry, a shareable link is not currently available for this article. Provided by the Springer Nature SharedIt content-sharing initiative ISSN 1476-4687 (online) ISSN 0028-0836 (print) © 2026 Springer Nature Limited Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). Acanthamoeba polyphaga mimivirus (artist's impression) is a giant among viruses, both in physical size and because of the size of its genome. Scientists report that a type of giant virus multiplies furiously by hijacking its host's protein-making machinery1 — long-sought experimental evidence that viruses can co-opt a system typically associated with cellular life. Virologists had already suspected that viruses could perform such a feat, says Frederik Schulz, a computational biologist at the Lawrence Berkeley National Laboratory in California, who was not involved with the work. But the new findings, published in Cell on 17 February, are an important confirmation. Compared with other viruses, he says, this one “has a more powerful toolbox to really replace what the host is doing”. They tend to infect single-celled organisms called protists — a group that includes amoebae and protozoa — that “are all over the place”, says Eugene Koonin, an evolutionary biologist at the US National Center for Biotechnology Information in Bethesda, Maryland. Scientists glimpse oddball microbe that could help explain rise of complex life Scientists glimpse oddball microbe that could help explain rise of complex life The giant DNA virus used in this study, Acanthamoeba polyphaga mimivirus, has a genome that is about five times larger than those of poxviruses, which have the biggest genomes of any virus that infects humans. To understand whether the virus affects its host's protein-assembly line, the researchers isolated viral proteins that interact with host organelles called ribosomes. The scientists identified three viral proteins that seemed likely to be involved in hijacking host protein production. Squirrels could be a reservoir for the virus that causes mpox Apply now with your research proposal and CV for a Postdoc position in chronic lung diseases. An essential round-up of science news, opinion and analysis, delivered to your inbox every weekday. Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.