Baseline lipoprotein(a) and heart failure-related events in oncology patients: competing-risk analysis
A S Stephanus, J Barreto, A C Sposito, L S F Carvalho, Y Tu, A Campos StafficoAbstract
Background
Anthracycline-induced cardiotoxicity is a major cardiovascular (CV) complication in cancer care. Risk assessment in cardio-oncology occurs amid high, early cancer mortality that competes with CV outcomes, complicating biomarker interpretation unless competing events are explicitly modelled. Lipoprotein(a) [Lp(a)] is a genetically determined biomarker strongly linked to atherosclerotic disease, but its association with heart failure (HF) in cancer patients – where cancer death may preclude CV events – remains unclear.
Purpose
To evaluate the association between baseline Lp(a) and HF-related events in patients receiving chemotherapy for anthracycline-eligible malignancies, accounting for competing cancer death.
Methods
Exploratory observational cohort study using UK Biobank data. Participants receiving chemotherapy for anthracycline-eligible neoplasms were identified using validated codes. Baseline Lp(a) was measured before chemotherapy, and prevalent HF was excluded. Primary endpoint was incident HF, defined as composite of HF hospitalisation, HF death, or HF diagnosis. Associations were estimated using Fine–Gray competing-risk regression with cancer death as the competing event. Model 1 was unadjusted; Model 2 adjusted for age, sex, LDL cholesterol, chemotherapy cycles, and neoplasm type. Lp(a) was modelled continuously, with log-transformation assessed for efficiency. Adjusted cumulative incidence functions were estimated at representative Lp(a) values. Results reported as subdistribution hazard ratios (sHR) with 95% confidence intervals (CI).
Results
Among 3,383 patients, 112 HF-related events and 1,180 cancer deaths occurred over a median 10.2-year follow-up (interquartile range 5.4–14.8). In Model 1, log-transformed Lp(a) was inversely associated with the HF composite (sHR 0.83 [0.70–0.99]); this association persisted in Model 2 (sHR 0.83 [0.69–0.99]). Sensitivity analyses using raw and transformed Lp(a) yielded similar effect estimates. Adjusted cumulative incidence curves showed sustained separation for the HF composite across Lp(a) levels, whereas cancer-death curves largely overlapped across the Lp(a) distribution.
Conclusion
In this exploratory analysis of chemotherapy-treated patients with anthracycline-eligible malignancies, competing-risk modelling identified an inverse association between baseline Lp(a) and incident HF, with a 17% lower rate of HF-related events per higher Lp(a) level. This is unexpected given Lp(a)’s established atherothrombotic risk profile. These observational findings are not causal and may reflect underlying vulnerability, treatment selection, or competing-event dynamics. The higher HF incidence at very low Lp(a) levels is hypothesis-generating and may indicate Lp(a) as a marker of systemic vulnerability rather a protective factor.