The research, published in Annals of Clinical and Translational Neurology, evaluated two FDA-cleared capillary blood collection devices, TAPII and Tasso+, for measuring four key Alzheimer’s plasma biomarkers: the amyloid-beta 42/40 ratio, phosphorylated tau 217 (pTau217), neurofilament light chain (NfL) and glial fibrillary acidic protein (GFAP). These analytes tell complementary stories about the disease. The Aβ42/40 ratio reflects amyloid deposition, pTau217 is currently the strongest single plasma marker for discriminating autopsy-confirmed Alzheimer’s disease from other neurodegenerative conditions, NfL signals neuroaxonal injury, and GFAP marks astrocyte activation associated with the onset of amyloid pathology. The team, led by researchers at the Johns Hopkins Center for CSF Disorders, compared capillary plasma measurements to simultaneously collected venous plasma and cerebrospinal fluid from the same individuals.

The cohort comprised 192 patients examined between October 2021 and April 2024, with a mean age of 61 years and a clinically diverse mix of diagnoses. The group included 25 patients with Alzheimer’s disease, 42 with idiopathic normal pressure hydrocephalus, 28 with vascular dementia, 61 with other neurological conditions such as dementia with Lewy bodies, frontotemporal dementia, Parkinson’s disease and idiopathic intracranial hypertension, and 36 cognitively normal participants split into younger and older normal groups. Each patient who could provide samples underwent venipuncture on one arm and capillary collection on the other using either TAPII or Tasso+, both containing EDTA as an anticoagulant. Twenty patients provided samples with both devices. Cerebrospinal fluid was obtained by diagnostic large-volume lumbar puncture under sterile conditions.

The devices themselves are compact, single-use tools designed to collect small amounts of capillary whole blood painlessly from capillaries in the skin. TAPII, approved in 2019, uses an array of 16 microneedles, each 500 micrometers in length, and gentle vacuum to draw blood. Tasso+, cleared in 2022, uses a single lancet 5 millimeters long. Both are described by their manufacturers as suitable for use by a caregiver or the patient themselves, if able, from the comfort of home. In this study, Tasso+ collected a mean volume of 294.1 microliters and TAPII 266.94 microliters, with Tasso+ succeeding in collecting more than 50 microliters of whole blood in 90.3 percent of attempts and TAPII in 86 percent, differences that were not statistically significant. Both devices yielded sufficient blood for multi-biomarker immunoassays in more than 85 percent of collections.

All biomarker levels were measured in singlicate on a Simoa HD-X analyzer using commercially available kits, with inter-assay coefficients of variation below 12 percent for all analytes. The results for three of the four biomarkers were strikingly positive. Capillary concentrations of NfL, GFAP and Aβ42/40 from both devices correlated strongly with their venous plasma counterparts. TAPII-derived NfL showed a correlation coefficient of 0.86 with venous NfL, GFAP 0.84 and Aβ42/40 0.74, all highly significant. Tasso+ performed similarly, with NfL at 0.91, GFAP at 0.80 and Aβ42/40 at 0.75. Pairwise comparisons found no significant differences between TAPII, Tasso+ and venous measurements for these three analytes, and Bland-Altman analysis revealed minimal bias and tight agreement for Aβ42/40 across the range of values, with wider limits of agreement for NfL and particularly GFAP, which showed substantial variability and multiple outliers for both devices.

Then came the surprise. pTau217, the biomarker that has consistently outperformed other phosphorylated tau isoforms in distinguishing Alzheimer’s disease from other dementias and in predicting PET positivity, failed to replicate between capillary and venous samples. Tasso+ capillary pTau217 correlated only weakly with venous pTau217 (r = 0.23, p = 0.04), and TAPII pTau217 did not correlate significantly at all (p = 0.12). Bland-Altman analysis demonstrated a consistent positive bias for both devices, with mean biases of approximately 7 picograms per milliliter for TAPII and 3 picograms per milliliter for Tasso+, and this overestimation became more pronounced at higher pTau217 concentrations. Passing-Bablok regression confirmed substantial proportional bias, with slope estimates markedly deviating from one for both devices. The practical consequence was dramatic: over 95 percent of capillary pTau217 concentrations across all patient groups, even cognitively normal controls, exceeded the preliminary cutoff of 0.42 picograms per milliliter used in established literature to indicate significant brain amyloid pathology. In venous plasma, only the Alzheimer’s disease group and the mixed neurodegenerative Other group exceeded that threshold on average. The study reports a nearly 30-fold difference between TAPII and venous measurements, a 10-fold difference between Tasso+ and venous, and a 3-fold difference between the two capillary devices themselves.

The authors propose several hypotheses for this pTau217 anomaly. Phosphorylated tau species, including pTau217, as well as alpha-synuclein aggregates, have been shown to exist in peripheral nerve terminals of the skin, raising the plausible hypothesis that capillary collection devices may release peripheral pTau217 into the whole blood sample collected, thereby inflating the overall concentration. Matrix effects unique to the pTau217 assay may also be involved, since all the other analytes were run on a different assay kit. Differences in the dimensions and number of needles utilized by the two devices may explain variation in pTau217 concentration as well as hemolysis, which was significantly higher in capillary samples (TAPII 2.08 out of 4, Tasso+ 1.37 out of 4) compared with venous. Notably, hemolysis did not significantly alter the strong relationship between capillary and venous measurements of Aβ42/40, NfL and GFAP, suggesting the pTau217 problem is specific rather than a general quality issue with capillary specimens.

When the researchers compared capillary and venous plasma to cerebrospinal fluid, venous NfL, GFAP and pTau217 all correlated significantly with CSF analytes (NfL r = 0.73, GFAP r = 0.59, pTau217 r = 0.33), consistent with previous CSF-plasma studies. Capillary NfL and GFAP also correlated modestly with CSF measures, whereas capillary Aβ42/40 and pTau217 did not correlate significantly with CSF. Age-adjusted logistic regression models constructed separately for each sample type found comparable discrimination of Alzheimer’s disease status across venous and capillary measurements for all four biomarkers, with no statistically significant differences in AUC. Venous pTau217 demonstrated the highest individual biomarker performance (AUC 0.88, 95 percent CI 0.82-0.94), while GFAP and Aβ42/40 showed similar diagnostic performance across capillary and venous sample types.

Lasso-regularized logistic regression, used to account for the small number of clinically diagnosed Alzheimer’s cases (n = 25) and reduce overfitting, identified age and plasma GFAP as the most consistent predictors of Alzheimer’s disease across all sample types. Models incorporating these two variables yielded AUC values of 0.75 for TAPII-derived capillary plasma, 0.87 for Tasso+-derived capillary plasma and 0.83 for venous plasma. Remarkably, the Tasso+ model, built from just age and GFAP, exceeded both the TAPII and venous models and fell close to AUC values generated by plasma Alzheimer’s biomarker testing in larger studies. The authors suggest that capillary GFAP, particularly in combination with age or APOE genotype, could serve as a scalable screening tool for Alzheimer’s disease in primary care or decentralized research settings. NfL, while elevated in Alzheimer’s disease, is also elevated in many other neurodegenerative conditions present in the cohort, explaining its weak discriminatory power between Alzheimer’s and non-Alzheimer’s participants. The reduced predictive utility of capillary Aβ42/40 concurs with previous studies citing the plasma amyloid ratio’s greatest diagnostic power in preclinical and mild cognitive impairment stages, whereas this cohort primarily comprised clinical Alzheimer’s dementia.

Pain scores, measured on a Numeric Rating Scale from 0 to 10, were dramatically lower for capillary collections than for traditional venipuncture. Tasso+ scored a mean of 0.2 out of 10, TAPII 0.73 out of 10, and venous collection 2.24 out of 10, with all pairwise comparisons highly significant. The authors note that lower pain associated with capillary collection may improve patient compliance for blood draws, a finding consistent with other studies using capillary collection devices. The study was conducted in a supervised clinical setting, and the authors caution that findings may differ in unsupervised or at-home collection environments. The cohort was also not randomized between devices, which may introduce bias but reflects real-world clinical conditions, and the number of clinically diagnosed Alzheimer’s cases was relatively small, limiting statistical power.

The overall conclusion is a nuanced one. The findings support the feasibility of capillary biomarker testing and the use of capillary collection devices in decentralized Alzheimer’s clinical trials and studies, with minimal loss of diagnostic signal for NfL, GFAP and Aβ42/40 compared with venous plasma. But a clear need exists for specific guidelines and assay optimization for capillary biomarkers, particularly pTau217, due to significant technical or biological differences between TAPII, Tasso+ and venous plasma samples. The authors envision a future in which patients or caregivers could provide a small capillary sample at home or in a clinic, and clinicians could triage them for confirmatory CSF or PET testing based on algorithmic risk. Before that vision becomes reality, capillary blood biomarker tests will need to be thoroughly evaluated as CSF and venous plasma assays have been over the past decade, potentially necessitating separate cutoffs and guidelines for different plasma sample types to account for the measurement gap that this study has now exposed.

Subject of Research: Validation of two capillary blood collection devices for measuring Alzheimer's disease plasma biomarkers against venous plasma and cerebrospinal fluid

Article Title: Alzheimer's Disease Biomarkers Measured by Capillary Blood Collection Versus Venous Plasma and Cerebrospinal Fluid

Article References: Ho, S., Shen, Z., Lewis, A., Patel, M., Zheng, A., Proscino, G., Ho, T., Wang, J., Albert, M., & Moghekar, A. (2026). Alzheimer's Disease Biomarkers Measured by Capillary Blood Collection Versus Venous Plasma and Cerebrospinal Fluid. Annals of Clinical and Translational Neurology, Article acn3.70524. https://doi.org/10.1002/acn3.70524

Image Credits: AI Generated

DOI: 10.1002/acn3.70524

Keywords: Alzheimer's disease, blood biomarkers, capillary blood collection, pTau217, GFAP, neurofilament light chain, amyloid beta 42/40, cerebrospinal fluid, Tasso+, TAPII, decentralized clinical trials, plasma biomarkers

The study, published in Nature Communications, enrolled eighteen patients with suspected normal pressure hydrocephalus, a neurological condition that already requires neurosurgery. These patients were undergoing ventriculoperitoneal shunt placement, a procedure in which a tube is implanted to drain excess cerebrospinal fluid. Because the surgery already involves passing through the parieto-occipital cortex, the surgeons were able to take a tiny rice-grain-sized core of brain tissue at the same time, with the patients’ informed consent. Before the operation, each participant received an intravenous infusion of isotopically labelled leucine, an amino acid that crosses the blood-brain barrier and becomes incorporated into newly synthesized proteins. By measuring how much of this label appeared in tau and amyloid-beta proteins extracted from the biopsy, the researchers could determine when those molecules were made.

This technique, known as stable isotope labelling kinetics, has been used before to measure protein turnover in cerebrospinal fluid, but never before has it been combined with direct sampling of living human brain tissue in this way. The result is a kind of molecular timestamp. Proteins carrying the heavy carbon signature of the labelled leucine were manufactured during the hours or days before surgery; proteins without it are older. The approach gives researchers something that post-mortem studies fundamentally cannot: a snapshot of disease biology in real time, at the very earliest stages of the Alzheimer continuum, before symptoms have fully declared themselves.

What the team found in those tiny biopsies was striking. Half of the samples contained amyloid-beta pathology, visible as cored plaques, diffuse deposits and, in one case, cerebral amyloid angiopathy, where amyloid accumulates in blood vessel walls. Tau pathology, the second hallmark, was present but limited, appearing mostly as neuritic threads and tau associated with plaques rather than the mature neurofibrillary tangles seen in end-stage disease. Crucially, the researchers showed that phosphorylated tau, specifically the pTau217 form that has become one of the most accurate blood-based biomarkers for Alzheimer disease, was physically concentrated within and around the amyloid plaques themselves. The signal decreased progressively with distance from the plaque centre, providing direct visual evidence in living human tissue that tau phosphorylation and amyloid deposition are spatially coupled.

This co-localization matters because it speaks to one of the central unanswered questions in Alzheimer research: how amyloid and tau interact to drive neurodegeneration. Dystrophic neurites, the swollen and damaged neuronal processes that surround plaques, contain phosphorylated tau and are associated with synaptic dysfunction. The new findings suggest that this periplaque tau pool, rather than the aggregated tau inside mature tangles, may be the source of the pTau217 that appears in cerebrospinal fluid and blood. That would explain why fluid levels of this biomarker track so closely with plaque burden, and it offers a mechanistic explanation for why amyloid-targeting antibodies such as donanemab, which specifically clears pyroglutamate-modified amyloid-beta, also slow the accumulation of tau pathology even though they have no direct effect on tau metabolism.

The isotope labelling also yielded a remarkable measurement of tau biology. Labelled leucine was detectable in newly synthesized tau protein within just over three hours of the infusion starting, demonstrating that tau is translated rapidly in the human brain. By modelling the decay of the isotopic signal across the cohort, the researchers estimated that the half-life of tau in brain tissue is approximately thirty-four days, broadly consistent with the roughly twenty-three-day turnover previously measured in lumbar cerebrospinal fluid. When the researchers stratified their samples by plaque status, they found that tau labelling was significantly higher in biopsy samples that contained amyloid plaques compared to those that did not, providing ex vivo support for the idea that amyloid pathology drives increased tau production.

In sharp contrast to tau, the amyloid-beta peptides in the insoluble plaque fraction showed no detectable isotope labelling at all, despite the researchers using assays sensitive enough to detect enrichment as low as 1.87 percent and validating their findings across two independent mass spectrometry laboratories. This absence is not a technical failure but a biological finding in its own right. It suggests that amyloid plaques are remarkably stable structures, built up over years or decades, that do not dynamically incorporate significant amounts of new amyloid-beta within the few-hour window of the labelling period. The plaques in these early-stage samples were chemically immature, containing less than forty percent isomerised amyloid-beta species, compared with roughly ninety percent isomerisation in post-mortem Alzheimer tissue, where plaques have had years to accumulate molecular damage.

That isomerisation clock is one of the most conceptually powerful tools to emerge from this work. Aspartate residues in amyloid-beta spontaneously convert from their native L-form to non-native isomeric forms over time, at a rate that can be used to estimate the age of a plaque. Previous work has suggested that around fifty percent isomerisation corresponds to a mean plaque age of roughly eight months, while eighty-five percent indicates an age of about four years. By quantifying these isomers in living tissue, the researchers could distinguish freshly formed, immature plaques from older, more established deposits. This kind of molecular dating has never before been possible in living human brain, and it opens the door to staging the disease biochemically rather than relying on crude measures of plaque quantity, which notoriously correlate poorly with cognitive symptoms.

The team also used matrix-assisted laser desorption ionization mass spectrometry imaging to map the chemical architecture of individual plaques at ten-micrometre resolution. This revealed a striking degree of spatial organization within single deposits. Amyloid-beta 1-40 and pyroglutamate-modified amyloid-beta species were concentrated in the dense plaque core, while amyloid-beta 1-42 and related C-terminal species were distributed more homogeneously across both the core and the diffuse periphery. This peptide-specific spatial segregation, previously observed only in mouse models and post-mortem tissue, confirms that pyroglutamate-modified amyloid sits at the growing edge of plaques and may be a key driver of plaque expansion, making it a particularly relevant target for immunotherapies like donanemab that are designed to clear exactly this modified species.

One further finding carries practical implications for how Alzheimer biomarkers are interpreted. The researchers compared tau profiles in ventricular cerebrospinal fluid, collected during shunt surgery directly from the brain’s fluid spaces, with those in lumbar cerebrospinal fluid collected from the lower spine. Ventricular fluid closely mirrored the tau peptide profile seen in brain tissue itself, while lumbar fluid showed a distinctly different pattern with reduced C-terminal abundance. The team attributes this partly to iatrogenic contamination, as the shunt insertion procedure inevitably releases intracellular brain proteins into the ventricular fluid. Even so, the observation that ventricular cerebrospinal fluid can serve as a surrogate for brain tissue chemistry is significant, particularly for normal pressure hydrocephalus patients, in whom amyloid pathology is common and brain biopsy during shunt surgery is already practised as a diagnostic adjunct by many specialists.

The study is a proof of principle, and the authors are careful to acknowledge its limitations. The cohort is small, the participants were recruited for a hydrocephalus indication rather than Alzheimer disease itself, and the biopsies came from a single brain region that may not represent disease burden elsewhere. The tau half-life estimate rests on a statistical model with sparse timepoints around the peak of the labelling curve. Even so, the conceptual advance is hard to overstate. For the first time, researchers have captured the molecular birth of Alzheimer pathology in living human brains, dated plaques by their chemistry, watched tau being synthesized in real time, and visualized the physical interface between the disease’s two hallmark proteins. The methods developed here should be applicable to other neurodegenerative conditions, and they promise to reshape how clinical trials of the future are designed, allowing therapies to be matched to the specific molecular stage of disease rather than to symptoms alone.

Subject of Research: Molecular characterization of amyloid and tau pathology in living human brain tissue using stable isotope labelling and mass spectrometry imaging

Article Title: Spatiotemporal characterization of Alzheimer disease pathology in living human brain tissue

Article References: Aslanyan, A., Mukherjee, S., Moncur, E. M., Szadziewska, A., He, Y., Kac, P. R., Coyle, R., Camporesi, E., Sato, C., Ghoshal, N., Barthélemy, N. R., Wood, J., Ge, J., Dulewicz, M., Koutarapu, S., Schwetye, K., Roberts, K. F., Murphy, E. C., Carletti, F., … Paterson, R. W. (2026). Spatiotemporal characterization of Alzheimer disease pathology in living human brain tissue. Nature Communications, 17(1), Article 9878. https://doi.org/10.1038/s41467-026-76494-4

Image Credits: AI Generated

DOI: 10.1038/s41467-026-76494-4

Keywords: Alzheimer disease, amyloid-beta, tau, stable isotope labelling, mass spectrometry imaging, normal pressure hydrocephalus, pTau217, pyroglutamate, protein turnover, biomarkers, brain biopsy, neurodegeneration

The research draws on the Gothenburg H70 Birth Cohort Study, one of the longest-running population studies of ageing in the world. In 2014 to 2016, researchers comprehensively examined 70-year-olds born in 1944 who were living in Gothenburg, achieving a 72 percent response rate. Participants underwent extensive interviews and clinical assessments covering education, income, childhood family economic circumstances, mental health, cardiovascular disease, anthropometry, and laboratory measures, and they provided blood samples for genotyping. A follow-up examination was conducted five to eight years later, with a mean follow-up of roughly six years and a response rate of 77.6 percent among survivors. After excluding participants with missing parental data and those whose parents died before age 41, likely from causes unrelated to biological ageing such as accidents, the final baseline sample comprised 1,126 individuals, 523 men and 603 women.

The researchers classified parental longevity, or PL, into three groups. Participants were considered to have high parental longevity if both parents survived to age 85, which applied to 17.2 percent of the sample; medium parental longevity if one parent reached 85, applying to 43.3 percent; and low parental longevity if neither parent did, applying to 39.4 percent. The age-85 threshold was chosen deliberately: the participants’ parents were born around 1920, when cohort life expectancy was only about 74 years, so reaching 85 represented genuinely exceptional survival for that generation. Alongside this family-history measure, the team constructed two polygenic longevity scores, or PGLSs, using genome-wide association summary statistics from a large international meta-analysis of longevity genes, combined with a Bayesian shrinkage method and a European ancestry reference panel. One score included variants in the APOE locus, a gene region strongly tied to age-related disease and mortality, while the other excluded it.

At baseline, both higher parental longevity and higher polygenic scores were associated with more favourable social circumstances. People with long-lived parents, and people with higher genetic longevity scores, had spent longer in education, reported better childhood family economic conditions, and enjoyed higher current household income. Both measures were also linked to vascular health: high parental longevity and higher scores on the APOE-inclusive polygenic score were associated with less hypertension, and higher polygenic scores were inversely related to myocardial infarction, a finding the authors report as novel. But the two measures diverged sharply beyond that point, and the divergence is where the study becomes genuinely provocative.

Parental longevity, unlike the genetic scores, was associated with a broad constellation of biological advantages. Participants with high parental longevity had better scores on the Mini-Mental State Examination, higher levels of total cholesterol, HDL cholesterol and LDL cholesterol, lower body mass index, lower homocysteine, and lower levels of the inflammatory markers interleukin-6 and C-reactive protein. They were also less likely to be current smokers. Participants with even one long-lived parent showed better cognition, higher cholesterol, lower homocysteine and less smoking than those with two short-lived parents. The polygenic scores, by contrast, captured mainly the socioeconomic and cardiovascular factors and little else. Strikingly, the two measures showed no statistical association with each other at all, whether parental longevity was analysed as groups, as the mean of both parents’ ages, or as mothers’ and fathers’ ages separately.

Some of these findings challenge conventional assumptions. The higher total cholesterol and LDL cholesterol among offspring of long-lived parents may seem paradoxical, but they echo previous research, including studies suggesting that rising cholesterol in late life is associated with reduced dementia risk and lower mortality in older adults. Cholesterol is a precursor to steroid hormones essential for metabolism and immune function, and the authors caution that the so-called cholesterol paradox in late life should be interpreted carefully, noting that lipid-lowering therapy did not appear to explain the association and that frailty was rare in this relatively young-old population. Lower homocysteine among those with long-lived parents is also notable, since elevated homocysteine is linked to folate and vitamin B deficiency, endothelial dysfunction, atherosclerosis, cardiovascular events, and dementia.

Perhaps the most forward-looking result emerged at follow-up. Plasma phosphorylated-tau 217, or pTau217, is one of the most accurate blood biomarkers currently available for Alzheimer’s disease pathology, and its levels are known to rise steeply with age. In the longitudinal analyses, participants with high parental longevity showed significantly less increase in pTau217 over the six years than those whose parents had both died before 85. This suggests that Alzheimer’s pathology may establish itself later, or accumulate more slowly, in people with long-lived parents, potentially delaying the onset of dementia. Consistent with this, the high-parental-longevity group also had higher baseline MMSE scores. No such longitudinal associations were observed for the polygenic scores, and neither measure was associated with plasma neurofilament light, another age-sensitive neurodegeneration marker.

The authors are careful about causality. Lower interleukin-6 and C-reactive protein in offspring of long-lived parents may reflect healthier lifestyles and better overall health rather than a direct determinant of longevity, although low-grade inflammation, sometimes called inflammaging, is considered a central driver of accelerated ageing, dementia, cardiovascular disease and mortality. Sex also mattered: several associations, including those involving C-reactive protein, neurofilament light, creatinine and triglycerides, were significant only in women. The team also acknowledges limitations, including self-reported parental ages, possible selective survival bias, a predominantly white and age-homogeneous sample, and the smaller follow-up cohort, and they report both uncorrected and false-discovery-rate-corrected statistics, noting that some findings, including those on inflammatory markers and pTau217, did not survive multiple-comparison correction.

Nevertheless, the overall pattern is coherent and biologically plausible. Polygenic longevity scores, in their current form, appear to capture only a narrow slice of the biology of ageing, likely because longevity’s genetic architecture is highly heterogeneous and heavily environment-dependent. Parental longevity, by contrast, seems to summarise everything at once: inherited genetics, shared early environment, transgenerational social advantage, and accumulated lifestyle influences. The authors conclude that asking about parental lifespan could serve as a simple, inexpensive proxy for biological ageing, suitable for routine health assessments of older adults to flag individuals who might benefit from preventive screening for hypertension, cognitive decline, inflammation and emerging Alzheimer’s pathology. In an era of billion-dollar biomarker pipelines, the humble family history may still be one of medicine’s most underrated diagnostic instruments.

Subject of Research: Parental longevity and polygenic longevity scores in relation to ageing-related social, cardiovascular, inflammatory and neurodegenerative factors in 70-year-olds

Article Title: Parental longevity and polygenic longevity scores in relation to ageing-related factors in a population of 70-year-olds followed over six years: The Gothenburg H70 Birth Cohort Study

Article References: Seidu, N. M., Rydén, L., Skoog, J., Samuelsson, J., Kern, S., Waern, M., Zetterberg, H., Holstege, H., Erhag, H. F., Westman, E., & Skoog, I. (2026). Parental longevity and polygenic longevity scores in relation to ageing-related factors in a population of 70-year-olds followed over six years: The Gothenburg H70 Birth Cohort Study. GeroScience. https://doi.org/10.1007/s11357-026-02501-7

Image Credits: AI Generated

DOI: 10.1007/s11357-026-02501-7

Keywords: parental longevity, polygenic longevity score, ageing, GeroScience, APOE, pTau217, Alzheimer's disease, inflammation, cholesterol, cardiovascular risk, socioeconomic status, H70 cohort