Tumour biomarkers: association with heart failure outcomes

Abstract

Background: There is increasing recognition that heart failure (HF) and cancer are conditions with a number of shared characteristics.

Objectives: To explore the association between tumour biomarkers and HF outcomes.

Methods: In 2,079 patients of BIOSTAT-CHF cohort, we measured six established tumour biomarkers: CA125, CA15-3, CA19-9, CEA, CYFRA 21-1 and AFP.

Results: During a median follow-up of 21 months, 555 (27%) patients reached the primary end-point of all-cause mortality. CA125, CYFRA 21-1, CEA and CA19-9 levels were positively correlated with NT-proBNP quartiles (all P < 0.001, P for trend < 0.001) and were, respectively, associated with a hazard ratio of 1.17 (95% CI 1.12-1.23; P < 0.0001), 1.45 (95% CI 1.30-1.61; P < 0.0001), 1.19 (95% CI 1.09-1.30; P = 0.006) and 1.10 (95% CI 1.05-1.16; P < 0.001) for all-cause mortality after correction for BIOSTAT risk model (age, BUN, NT-proBNP, haemoglobin and beta blocker). All tumour biomarkers (except AFP) had significant associations with secondary end-points (composite of all-cause mortality and HF hospitalization, HF hospitalization, cardiovascular (CV) mortality and non-CV mortality). ROC curves showed the AUC of CYFRA 21-1 (0.64) had a noninferior AUC compared with NT-proBNP (0.68) for all-cause mortality (P = 0.08). A combination of CYFRA 21-1 and NT-proBNP (AUC = 0.71) improved the predictive value of the model for all-cause mortality (P = 0.0002 compared with NT-proBNP).

Conclusions: Several established tumour biomarkers showed independent associations with indices of severity of HF and independent prognostic value for HF outcomes. This demonstrates that pathophysiological pathways sensed by these tumour biomarkers are also dysregulated in HF.

Keywords: biomarkers; heart failure; natriuretic peptides; neoplasms; tumour.

Conflict of interest statement

The UMCG, which employs several of the authors, has received research grants and/or fees from AstraZeneca, Abbott, Bristol‐Myers Squibb, Novartis, Novo Nordisk and Roche. RAdB received personal fees from Abbott, AstraZeneca, Novartis, and Roche. SDA reports personal fees from Vifor Int, Bayer, Boehringer Ingelheim, Novartis, Servier, Respicardia and Impulse Dynamics outside the submitted work; and grants from Abbott Vascular and Vifor Int outside the submitted work. JGC reports personal fees from Amgen, AstraZeneca, Bayer, Bristol‐Myers Squibb, GlaxoSmithKline, Medtronic, MyoKardia, Novartis International AG, Philips, Sanofi, Servier, Stealth Biopharmaceuticals, Torrent Pharmaceuticals, Vifor and Abbott; grants from Amgen, Bayer, Bristol‐Myers Squibb, Medtronic, Novartis, Pharmacosmos, Pharma Nord, Stealth Biopharmaceuticals, Torrent Pharmaceuticals and Vifor; and nonfinancial support from Pharmacosmos, Pharma Nord, Vifor, Novartis International AG and Medtronic. CCL reports grants from AstraZeneca, Amgen, Merck & Co, Novartis and Servier, during the conduct of the study. PLvH is an employee of Hoffmann‐La Roche Ltd and holds stock in this company. MM reports grants from European Community during the conduct of the study and personal fees from Bayer, Novartis and Servier outside the submitted work. PA reports personal fees from Servier, Novartis, Boehringer Ingelheim, Daiichi Sankyo, Bayer, Pfizer, Menarini, Sanofi, MSD and GSK as well as an investigator‐initiated grant from Boehringer Ingelheim. AAV reports grants from European Commission during the conduct of the study and Roche Diagnostics outside the submitted work; personal fees from Amgen, Boehringer Ingelheim, AstraZeneca, Bayer, Cytokinetics, GlaxoSmithKline, MyoKardia, Roche Diagnostics, Novartis and Servier, outside the submitted work. All other authors declare no competing interests.

© 2020 The Authors. Journal of Internal Medicine published by John Wiley & Sons Ltd on behalf of Association for Publication of The Journal of Internal Medicine.

Figures

Fig. 1

Cox regression models, estimating the predictive value of tumour markers and NT‐proBNP for all‐cause mortality. (a) CA125. (b) CYFRA 21‐1. (c) CEA. (d) CA19‐9. (e) CA15‐3. (f) NT‐proBNP. The P‐values presented are Bonferroni‐adjusted. HR: hazard ratio. Numbers between brackets represent the 95% CI. All models are corrected for the BIOSTAT risk model (including age, BUN, NT‐proBNP (except for NT‐proBNP), haemoglobin and beta blocker use at baseline).

Fig. 2

(a) ROC curves for CA125, CYFRA 21‐1, CEA, CA19‐9, CA15‐3 and NT‐proBNP for all‐cause mortality. (b) ROC curve for NT‐proBNP, CYFRA 21‐1 and both biomarkers combined for all‐cause mortality. P‐value (Bonferroni‐corrected) refers to the comparison between NT‐proBNP and NT‐proBNP + CYFRA 21‐1.