Preoperative Senescence Biomarkers Predict Complications Following Elective Cardiac Surgery in Older Adults

In a prospective observational cohort study evaluating biological aging markers in cardiac surgery, circulating preoperative proteins associated with cellular senescence were found to independently predict postoperative complications in older individuals undergoing elective operations. Researchers at University Hospital Halle in Germany measured a panel of circulating senescence-associated secretory phenotype mediators in older adults prior to surgery, finding that two specific regulatory proteins, insulin-like growth factor-binding protein 1 and insulin-like growth factor-binding protein 4, significantly tracked the likelihood of developing multi-organ adverse outcomes. The findings indicate that molecular markers of biological senescence can capture an informative dimension of physiological reserve and vulnerability that is not registered by standard clinical risk scores.

Major cardiac surgery represents an acute and severe physiological challenge characterized by the use of cardiopulmonary bypass, ischemia-reperfusion stress, widespread systemic inflammatory activation, and significant hemodynamic instability. In older individuals undergoing elective procedures such as coronary artery bypass grafting, transcatheter aortic valve implantation, or combined operations, clinicians currently rely on established surgical models such as EuroSCORE II and the Society of Thoracic Surgeons Morbidity and Mortality Score to estimate procedural risk. While these conventional risk assessment tools integrate chronological age alongside anatomical and clinical variables, they do not incorporate direct biochemical measures of biological aging or cellular exhaustion, resulting in moderate discriminatory capacity across heterogeneous patient populations.

Cellular senescence is a permanent state of stable cell cycle arrest induced by intrinsic and extrinsic stressors including telomere attrition, oncogene activation, genotoxic insults, and oxidative damage. Although senescent states serve constructive roles in wound healing, embryonic tissue remodeling, and tumor suppression, the long-term accumulation of non-dividing senescent cells within aging tissues promotes chronic pathology. These persistent cells actively secrete a complex mixture of cytokines, chemokines, growth factors, lipids, and extracellular matrix regulators known as the senescence-associated secretory phenotype. This ongoing secretory program establishes a persistent, sterile, low-grade inflammatory microenvironment that impairs baseline tissue repair, promotes pathological vascular remodeling, and accelerates functional organ decline across cardiovascular structures.

To investigate whether this underlying biological burden informs surgical recovery, investigators enrolled 131 older patients aged 65 years and older who were scheduled for elective cardiac interventions between November 2024 and September 2025. The participant cohort exhibited a median age of 75.0 years, with 67.9 percent male participants and a high background prevalence of chronic comorbidities, including hypertension in 99.2 percent of patients, renal impairment in 71.8 percent, and diabetes mellitus in 53.4 percent. Within the cohort, 55.0 percent underwent coronary artery bypass grafting, 34.4 percent underwent transcatheter aortic valve implantation, and 10.7 percent underwent combined bypass and aortic valve replacement surgery.

Venous blood specimens were gathered within 48 hours prior to surgical intervention, processed into serum, and analyzed across a custom multiplex panel of 19 senescence-associated proteins using a Luminex discovery assay. After excluding interleukin-1 beta and interleukin-12p70 due to excessive out-of-range readings exceeding 20 percent, 17 log-transformed biomarkers were evaluated. The primary clinical outcome was an adapted multimorbidity score known as the Hofmann Score, defined as a binary composite endpoint encompassing seven clinical organ domains: pulmonary, cardiac, renal, neurological, reoperation, wound complications, and in-hospital or follow-up mortality. Patients experiencing two or more organ domain events met the composite threshold, which occurred in 28 patients, representing an overall event rate of 21.4 percent.

In univariable logistic regression screening across all 17 evaluated proteins, five candidate markers initially demonstrated nominal statistical significance with unadjusted p-values below 0.05: insulin-like growth factor-binding protein 1, insulin-like growth factor-binding protein 4, growth differentiation factor 15, tumor necrosis factor receptor superfamily member 1A, and tumor necrosis factor alpha. For each of these five proteins, higher preoperative circulating concentrations correlated with elevated odds of developing the composite adverse outcome. However, following application of the Benjamini-Hochberg false discovery rate correction for multiple testing across the panel, none of the individual markers retained statistical significance, reflecting the modest cohort size and exploratory nature of the preliminary screen.

Multivariable logistic regression models were developed to evaluate the predictive capacity of the identified biomarkers after accounting for surgical procedure type. In the primary multivariable model, designated Model M3, insulin-like growth factor-binding protein 1 and insulin-like growth factor-binding protein 4 were retained alongside operative category based on optimal model parsimony and information criteria. Within this model, elevated insulin-like growth factor-binding protein 1 remained an independent predictor of the composite outcome, with an odds ratio of 1.68, while insulin-like growth factor-binding protein 4 was similarly predictive, with an odds ratio of 1.43. Surgical procedure type also exerted a marked effect, with transcatheter aortic valve implantation associated with substantially lower odds of complications compared to open coronary artery bypass grafting.

The discriminatory performance of the primary biomarker model yielded an apparent area under the receiver operating characteristic curve of 0.777. Internal validation conducted through 1000 bootstrap resamples demonstrated modest model optimism of 0.031, producing an optimism-corrected area under the curve of 0.746. This performance was numerically comparable to EuroSCORE II, which achieved an uncorrected area of 0.725 and a corrected area of 0.727, while exceeding the Society of Thoracic Surgeons score, which registered an area under the curve of 0.660. The Hosmer-Lemeshow goodness-of-fit test confirmed adequate calibration across the models, showing no statistically significant deviation between observed and predicted event probabilities.

When evaluating incremental prognostic value, the biomarker model did not generate statistically significant increases in total area under the curve or integrated discrimination improvement over the established clinical scores. However, the addition of the two binding proteins produced statistically significant net reclassification improvement. The model yielded a continuous net reclassification improvement of 0.528 over EuroSCORE II and 0.418 over the Society of Thoracic Surgeons score, demonstrating that the biological markers substantially refined risk categorization at the individual patient level by correctly shifting predicted risk upward for patients who experienced complications and downward for those who did not.

To evaluate whether small-sample bias affected estimation given an events-per-variable ratio below the standard benchmark of ten, the investigators performed a sensitivity analysis using Firth-penalized logistic regression. Under Firth penalization, both proteins remained statistically significant predictors with confidence intervals remaining above 1.0, showing odds ratios of 1.62 for insulin-like growth factor-binding protein 1 and 1.40 for insulin-like growth factor-binding protein 4. An unweighted composite Z-score encompassing all 17 markers showed a rising event rate across tertiles from 15.9 percent to 29.5 percent, but did not achieve independent statistical significance, likely due to signal dilution from uninformative analytes such as leptin.

Analyses of a secondary composite endpoint composed of major adverse cardiac and cerebrovascular events, which occurred in 36.6 percent of patients, revealed distinct biological patterns. While growth differentiation factor 15 and insulin-like growth factor-binding protein 4 were associated with both clinical endpoints, insulin-like growth factor-binding protein 1 and tumor necrosis factor markers were associated exclusively with the multi-organ Hofmann Score. Conversely, interleukin-15 was associated exclusively with major adverse cardiac and cerebrovascular events, supporting the existence of distinct perioperative vulnerability axes spanning metabolic multi-organ stress and acute vascular or immune activation.

The researchers emphasized several methodological limitations, noting that the single-center design, limited sample size of 131 participants, and modest event count restrict immediate generalizability and necessitate external validation in larger prospective cohorts. Furthermore, the circulating mediators were quantified at a single preoperative baseline time point rather than serially across recovery, and residual confounding from unmeasured nutritional variables or medications such as metformin and corticosteroids cannot be entirely excluded. Because current immunoassay platforms for these binding proteins remain designated for research use only, clinical translation into multidisciplinary preoperative heart team workflows will require the future development and validation of standardized diagnostic-grade assays.

Subject of Research: Preoperative cellular senescence biomarkers and surgical risk in older cardiac surgery patients

Article Title: Preoperative SASP burden predicts perioperative vulnerability in older patients undergoing elective cardiac surgery: a prospective cohort study

Article References: Eberl, A.-M., Stephan, T., Haggenmueller, L., Hofmann, B., Szabó, G., Wienke, A., Eickhoff, C., Simm, A., & Großkopf, A. (2026). Preoperative SASP burden predicts perioperative vulnerability in older patients undergoing elective cardiac surgery: a prospective cohort study. GeroScience. https://doi.org/10.1007/s11357-026-02626-9

Image Credits: AI Generated

DOI: 10.1007/s11357-026-02626-9

Keywords: cardiac surgery, cellular senescence, SASP, IGFBP-1, IGFBP-4, EuroSCORE II, STS Score, biological aging, perioperative risk, geroscience, Preoperative, burden