Adults starting hemodialysis whose serum albumin stays low or falls over their first year of treatment face a markedly higher risk of death, according to a new analysis of nearly 39,000 patients. The study, published in BMC Nephrology, applied data-driven clustering to monthly measurements of albumin and systolic blood pressure, mapping the dynamic trajectories of these two biomarkers rather than relying on single snapshot readings.
The research team, led by Catherine Wang with colleagues including Peter Kotanko, analyzed records from 38,863 adults who initiated hemodialysis in Fresenius Kidney Care North America clinics between 2010 and 2014. To be included, patients needed at least nine months of albumin and systolic blood pressure measurements during their first year on dialysis. Monthly averages were modeled as functional data, allowing the investigators to treat each patient’s biomarker history as a continuous curve rather than a series of isolated points.
Serum albumin and systolic blood pressure are strong predictors of adverse outcomes in hemodialysis patients, yet clinical interpretation often relies on static measurements rather than longitudinal patterns. The researchers argue that functional data analysis enables characterization of dynamic biomarker trajectories and identification of patient phenotypes that may better capture underlying nutritional, inflammatory, and hemodynamic states than conventional thresholds alone.
Two complementary unsupervised clustering approaches were used. K-means clustering combined with functional principal component analysis, or FPCA, identified distinct trajectory phenotypes for each biomarker. For albumin, K-means clustering produced three phenotypes labeled low, medium, and high, while FPCA identified three trajectories: high, rising, and falling. The same framework was applied to pre-dialysis systolic blood pressure, revealing its own set of distinct patterns.
The mortality findings were striking. Compared with patients in the high-albumin group, those with persistently low albumin had a 2.63-fold higher adjusted risk of death from any cause, with a 95 percent confidence interval of 2.42 to 2.87. Patients whose albumin was falling fared almost as poorly, showing a hazard ratio of 2.40 with a confidence interval of 2.13 to 2.70. Low or declining albumin was consistently associated with the highest mortality risk across both clustering methods.
Blood pressure trajectories told a more nuanced story. K-means-derived low pre-dialysis systolic blood pressure was associated with increased mortality, with a hazard ratio of 1.26 and a confidence interval of 1.16 to 1.36, whereas medium- and high-SBP clusters demonstrated more favorable outcomes. When trajectories were characterized by FPCA, falling systolic blood pressure paths, but not rising ones, were independently associated with higher mortality risk after adjustment, with a hazard ratio of 1.15 and a confidence interval of 1.04 to 1.28.
The associations were not simple correlations between raw measurements. The investigators used Cox proportional hazards models to assess the link between trajectory clusters and all-cause mortality after the first year of dialysis, adjusting for demographics, diabetes, catheter use, interdialytic weight gain, erythropoietin dose, hemoglobin, and neutrophil-to-lymphocyte ratio. That the trajectory signals survived this extensive statistical adjustment suggests the shape of a patient’s biomarker curve carries information beyond what is captured by standard risk factors.
The authors conclude that longitudinal albumin and systolic blood pressure trajectories identify reproducible patient phenotypes with markedly different mortality risk. Persistently low or declining albumin was a strong predictor of death, and a low or falling systolic blood pressure was associated with poorer survival. They suggest that incorporating trajectory-based assessments may improve identification of high-risk patients and support more individualized management strategies in maintenance hemodialysis.
As with any observational cohort analysis, the findings describe associations rather than proven causal effects, and the study population was drawn from a single dialysis provider network during a defined period. The article was published as an open-access paper on 23 October 2026 and is being shared early as a peer-reviewed, accepted version subject to further edits before the final Version of Record.
Subject of Research: Trajectory-based clustering of serum albumin and systolic blood pressure to identify mortality risk phenotypes in hemodialysis patients
Article Title: Longitudinal albumin and blood pressure trajectories identify mortality risk phenotypes in hemodialysis
Article References: Wang, C., Ahmadi, A., Hussein, R., Zhang, H., Usvyat, L., Wang, Y., & Kotanko, P. (2026). Longitudinal albumin and blood pressure trajectories identify mortality risk phenotypes in hemodialysis. BMC Nephrology. https://doi.org/10.1186/s12882-026-05458-z
Image Credits: AI Generated
DOI: 10.1186/s12882-026-05458-z
Keywords: hemodialysis, serum albumin, systolic blood pressure, mortality risk, trajectory clustering, functional principal component analysis, K-means, Cox proportional hazards, patient phenotypes, BMC Nephrology, kidney disease, biomarkers
News Source: Jerry Hayes. (October 28, 2026). Albumin and Blood Pressure Trajectories Reveal Mortality Risk Phenotypes in Hemodialysis Patients. Scienmag.
