Obesity modifies the prognostic value of serially measured laboratory values in a real-world HFpEF cohort

Aim Obesity has been shown to affect the relationship between NT-proBNP levels and outcomes in heart failure, and optimal risk prediction cutoffs vary across body mass index (BMI) categories. However, it remains unclear whether this applies to other blood biomarkers. Therefore, we examined whether obesity modifies the relationship between serially measured laboratory values and all-cause mortality in patients with heart failure with preserved ejection fraction (HFpEF). Methods We included ambulatory HFpEF patients who attended the outpatient clinic of an academic hospital between 2017 and 2022. The follow-up lasted until 2023. Data on 32 laboratory parameters, along with clinical characteristics and mortality, were extracted from electronic health records. Patients were categorized into two groups based on BMI: obese (BMI≥30) and non-obese (BMI<30). Associations of time-varying laboratory values with all-cause mortality were analyzed using joint modelling while interaction terms were used to assess effect modification by obesity status. Results In 416 patients, median (25th–75th percentile) age was 70 (60, 79) years, 225 (54%) were women, and 129 (31%) were obese. The median number of laboratory measurements per patient varied between 1 (IQR 1–3) and 6 (IQR 2–18), depending on the specific value assessed. During a median follow-up of 2.2 (1.2, 3.6) years, 120 (29%) patients died. After adjustment for age and sex, obesity status significantly modified the associations between serial measurements of creatinine, AST, ALT, ALP, total bilirubin, ferritin, transferrin, and HbA1c with all-cause mortality (Figure 1; p-interaction<0.05). In group-stratified analyses, serially measured NT-proBNP, troponin, creatinine, eGFR, ferritin, transferrin, and hematocrit were associated with all-cause-mortality only in the non-obese group (Figure 1; filled blue squares), although the interaction term did not reach statistical significance for NT-proBNP, troponin, eGFR, and hematocrit. Conversely, only in the obese group, AST, ALT, total bilirubin, and HbA1c were associated with all-cause mortality (Figure 1; filled blue circles). Conclusion Our findings support that obesity influences the associations of certain renal, hepatic, iron, and glycemic markers with mortality, with some reaching significance only in obese or non-obese patients. These insights could help refine risk prediction cutoffs in heart failure patients with obesity.