Common blood signals explain why chronic diseases cluster as we age

By Vijay Kumar MalesuReviewed by Susha Cheriyedath, M.Sc.Jan 5 2026

By linking routine and advanced blood biomarkers to disease patterns and progression across aging cohorts, this study shows how systemic metabolic stress may underlie the growing burden of multiple chronic conditions long before clinical decline becomes apparent.

Study: Shared and specific blood biomarkers for multimorbidity. Image Credit: pinkeyes / Shutterstock

In a recent study published in the journal Nature Medicine, researchers identified shared and pattern-specific blood biomarkers that track multimorbidity and the rate of disease accumulation using population-based cohorts and modern statistical learning methods.

Aging, Survival, and Shared Disease Pathways

Many adults aged 60 years or older live with multiple chronic conditions, resulting in loss of independence, increased family strain, and substantial financial burden. Advances in medical care have also enabled survival from previously fatal conditions such as myocardial infarction, cerebrovascular events, and malignancies, increasing the likelihood of developing multiple chronic diseases later in life.

These trends have contributed to multimorbidity as an unintended consequence of population aging and improved survival, with advancing age acting as a central driver of both disease burden and underlying biological change.

The biological mechanisms underlying these conditions include metabolic stress, chronic inflammation, vascular injury, and neurodegeneration. As a result, many chronic diseases may share common biological pathways rather than arising from independent causes. Identifying shared blood-based signals may help characterize biological vulnerability, although translating these findings into preventive clinical strategies remains an open challenge.

Population-Based Cohort and Clinical Assessments

This prospective observational analysis used data from the Swedish National Study on Aging and Care in Kungsholmen (SNAC-K), which enrolled community-dwelling adults aged 60 years or older. Of the 3,363 participants at baseline, 2,247 had complete serum biomarker data and were followed for up to 15 years.

Clinicians recorded information on 60 categories of chronic diseases using interviews, physical examinations, laboratory testing, medication records, and registry linkage. Diagnoses were coded using the ICD-10. Functional and cognitive status were assessed using activities of daily living (ADL) measures and the Mini-Mental State Examination (MMSE).

Biomarker Measurement and Analytical Methods

Blood biomarkers were quantified using multiplex Luminex and single-molecule array (Simoa) platforms, along with accredited clinical assays for hemoglobin, albumin, and gamma-glutamyl transferase (GGT). This approach distinguished routinely available clinical markers from research-grade multiplex measurements.

To identify multimorbidity patterns and their biological correlates, the researchers applied the Least Absolute Shrinkage and Selection Operator (LASSO) regression in cross-sectional analyses. Linear mixed-effects models estimated individual rates of disease accumulation over time, defined as the average increase in chronic conditions per year and reflecting population-level trends rather than individual prediction.

Gaussian LASSO regression identified biomarkers associated with faster disease accumulation, while principal component analysis (PCA) summarized correlated biomarker subprofiles. For external validation, LASSO coefficients derived from SNAC-K were applied to the Baltimore Longitudinal Study of Aging (BLSA) cohort.

Multimorbidity Patterns in the SNAC-K Cohort

Participants had a mean age of 72.7 years, and 61.5% were women. At baseline, individuals had an average of 3.9 chronic diseases and used 3.7 medications. Mean MMSE score was 28.3, and 4.3% reported at least one ADL limitation.

Latent class analysis identified five multimorbidity patterns among individuals with two or more diseases: Unspecific; Neuropsychiatric; Psychiatric and Respiratory; Sensory Impairment and Anemia; and Cardiometabolic, alongside a no-multimorbidity group. These patterns represent statistical groupings rather than discrete clinical diagnoses.

The Neuropsychiatric pattern was characterized by older age, greater disability, polypharmacy, and cognitive impairment. The Cardiometabolic pattern showed high cardiovascular risk and medication burden, while the Psychiatric and Respiratory pattern involved younger individuals with modest disability. The Sensory Impairment and Anemia pattern was associated with relatively mild functional impairment.

Shared and Pattern-Specific Biomarker Associations

Across the cohort, higher chronic disease counts were associated with cystatin C, hemoglobin A1c (HbA1c), growth differentiation factor 15 (GDF15), leptin, insulin, neurofilament light chain (NfL), creatinine, and C-peptide, while hemoglobin showed an inverse association.

Multinomial analyses revealed shared associations across all multimorbidity patterns for C-peptide, creatinine, cystatin C, GDF15, folic acid, HbA1c, insulin, leptin, and total cholesterol. The amyloid-β 42/40 ratio and hemoglobin were inversely associated overall, although hemoglobin showed a positive association within the Unspecific pattern.

Stronger pattern-specific links were observed between GDF15 and the Neuropsychiatric and Cardiometabolic patterns, and between cystatin C and creatinine with the Cardiometabolic and Sensory Impairment and Anemia patterns.

Biomarkers and Rates of Disease Accumulation

Faster rates of disease accumulation were directly associated with GDF15, HbA1c, cystatin C, leptin, GGT, and insulin, and inversely associated with albumin.

PCA identified a dominant biomarker axis characterized by GDF15 and cystatin C across multimorbidity measures, consistent with systemic metabolic and renal stress processes that intensify with aging and declining organ function.

Insulin loaded prominently on the second principal component for disease-accumulation rate, while GGT contributed most strongly to the third component. NfL contributed primarily within the Neuropsychiatric pattern, and N-cadherin within the Cardiometabolic pattern.

External Validation and Long-Term Outcomes

External validation in the BLSA cohort reproduced inter-biomarker correlations and demonstrated similar predictive accuracy, with square-root prediction errors of approximately 0.18–0.19 diseases per year.

Long-term outcomes aligned with multimorbidity patterns: incident dementia and recurrent depression were more frequent in the Neuropsychiatric pattern, while ischemic heart disease and heart failure were more common in the Cardiometabolic pattern. Fifteen-year mortality was higher across several multimorbidity patterns compared with individuals without multimorbidity.

Study Implications and Limitations

This population-based study demonstrates that blood biomarkers capture both shared and pattern-specific biological features of multimorbidity. GDF15, HbA1c, cystatin C, leptin, and insulin were consistently associated with higher disease burden, while GGT and albumin were associated with faster or slower rates of disease accumulation.

Although PCA highlighted a recurring axis related to mitochondrial and renal stress, these findings are observational and do not establish causality. Overall, the results suggest that age-related metabolic and systemic stress reflects biological vulnerability common to multiple chronic diseases, rather than providing immediate or individualized clinical risk prediction tools.