Association between heart failure and cancer: is gender the answer?

This article refers to ‘The association between heart failure and incident cancer in women: an analysis of the Women's Health Initiative’ by D.J. Leedy et al., published in this issue on pages 1712–1721. With the ageing of the population and the increasing burden of comorbidities, the prevalence of chronic diseases is progressively growing. Heart failure (HF) and cancer are the two major causes of morbidity/mortality in Western countries.1, 2 Besides their detrimental impact on life expectancy, their prognosis has improved over the last decades due to the exceptional efforts made in the field of prevention, diagnosis and treatment.3, 4 The close relationship between HF and cancer goes far beyond, involving common pathophysiological pathways, shared risk factors and therefore frequent co-prevalence.4 The demonstration of the cardio-toxic effect of antineoplastic therapies generated progressive attention on the risk of cardiovascular events, including HF, in patients with cancer, leading to a new field in cardiovascular sciences: cardio-oncology. The link between HF and cancer has been considered unidirectional for long time, with cancer potentially leading to HF but not vice versa. However, the interaction between HF and cancer is rather bidirectional. Nevertheless, the link between HF and incident cancer has been often neglected, and individuals with cancer have been largely excluded from clinical trials on HF. Over the last years, the incidence, management, and course of cancer in patients with pre-existing HF have attracted increasing attention, with two position papers from the European Society of Cardiology acknowledging the limited evidence on the role of cancer in chronic HF and therefore claiming for dedicated research.4, 5 The incidence of cancer in HF has been estimated to range between 18.9 and 33.7 per 1000 person-years and previous studies consistently reported higher risk in patients with chronic HF or post-myocardial infarction HF compared with the general population.6-8 The excess in risk was reported to be up to 70%, and mortality was 56% higher in HF patients with incident cancer.7 An increasing mortality due to cancer in the HF population has been also reported over the last two decades in a Spanish cohort of outpatients with HF.3 The potential reasons underlying the observed higher incidence of cancer in HF are various: (i) surveillance bias, as HF patients are in general considered for a stricter follow-up and exposed to several diagnostic exams or procedures that may detect otherwise silent neoplastic diseases4, 5; (ii) competing risk, since disease-modifying therapies prolong survival in HF patients but potentially increase the exposure to non-cardiovascular illnesses9; (iii) therapies, as for instance anticoagulants, that may unmask malignancies; and (iv) overlaps in symptoms of cancer and HF.5, 10 Moreover, in the contemporary era of superimposing comorbidities, shared risk factors such as smoking, diabetes, obesity and hypercholesterolaemia, predisposing to both HF and cancer, are becoming more prevalent and may promote the interaction between HF and neoplastic diseases.11 Changes in lifestyle and diet after HF diagnosis could also be involved in increasing the risk of cancer. At a translational level, two intriguing hypotheses have aimed to correlate HF with cancer. The first proposed a common perturbing milieu that might characterize both HF and cancer,12, 13 i.e. a chronic inflammatory state and oxidative stress. The second highlighted a role for the persisting hyperactivation of the sympathetic nervous system and of the renin–angiotensin–aldosterone and natriuretic peptide axes, that characterize the natural progression of HF but might also promote oncogenesis.4, 12, 14 Despite the increasing amount of evidence, there are conflicting data on the relationship between HF and incident cancer. In the Physicians' Health Study, exclusively enrolling males, there was no increased risk of cancer in patients with HF, drawing another character, gender, into the already complex interaction between HF and cancer.15 Furthermore, the HF phenotype at presentation, namely HF with reduced (HFrEF) vs. preserved ejection fraction (HFpEF), was not assessed in previous studies, and remains a further gap that, notably, has a tight connection with gender-related differences in disease expression. In this issue of the Journal, Leedy et al.16 made the attempt to fill these gaps in knowledge, by analysing a large nationwide prospective cohort of >140 000 post-menopausal women followed up for >20 years, in order to compare the incidence of cancer in women with vs. without HF. Moreover, the availability of data on left ventricular ejection fraction in more than one-third of the study cohort and the high prevalence of HFpEF in females allowed to provide important insights into the potential interaction between the HF phenotype and the risk of cancer. The primary results of this study support previous evidence on a significant association between HF and incident cancer. During the follow-up, about 12% of the study cohort developed cancer and 2% developed HF, with 18.8 per 1000 person-years developing HF prior to cancer. Overall, the observed incidence of cancer was 28% higher in patients with HF. The risk of cancer was higher for obesity-related and for some site-specific (colorectal and lung) cancers, but not for tobacco-related and breast cancers. Furthermore, when the subgroup with data on ejection fraction was separately analysed, HFpEF but not HFrEF was associated with the risk of developing cancer. The notable sample size, by far the largest compared with former studies, allowed for extensive statistical adjustments, although lifestyle and diet modifications after the diagnosis of HF remain potential residual confounders. One first, and perhaps, more important point to highlight about this study is the all-women design. At an initial glance, this might appear as limiting generalizability of the findings and, without a control cohort of males, it is not possible to explore gender-related differences. However, this study might be considered as the ideal counterpart of the previous, and discordant, study from Selvaraj et al.15 which included an all-male cohort of about 30 000 individuals with a similar length of follow-up. As typical for studies on HF, women were under-represented in the previous analyses on this topic and, despite sex-adjusted analyses, gender-related differences were not specifically explored.6-8 The possibility that gender could influence the interaction between HF and incident cancer must be necessarily considered due to the differences in results between this and Selvaraj et al. analysis.15, 16 Although the study cohort of the current analysis was partially extracted by previous clinical trials on hormone replacement therapy, adjustments for hormonal treatment were performed, limiting its potential confounding role in the results.16 Another important difference between this study and the former from Selvaraj et al. concerns the diagnosis of HF.15 Indeed, Leedy et al. appreciably adjudicated HF based on a documented hospitalization with HF as primary diagnosis, whereas Selvaraj et al. considered self-reported HF.15, 16 As correctly acknowledged by the authors, this might have underestimated the real incidence of HF, leading to consider only cases with severe HF in their study. The median interval between HF diagnosis and incident cancer in this study was 2.2 vs. 3.4 years in the study by Selvaraj et al., potentially reflecting a higher likelihood of surveillance bias in the former.15, 16 The demonstration that the increased risk of cancer was limited to patients with adjudicated HFpEF, both for the overall and site-specific (lung and colorectal) risk of malignancies, deserves careful consideration. The authors should be congratulated for their analyses on ejection fraction, which are novel and fulfil an important gap in the literature.6-8, 15 Indeed, although data on ejection fraction were limited to the medical records cohort sub-cohort of 41 503 women, the findings reported by Leedy et al. provide a step ahead in the understanding of the relationship between HF and cancer.16 HFpEF is highly prevalent in women and in elderly patients, and is characterized by a higher and more complex burden of comorbidities, including lung disease and anaemia that might be the prelude to incident cancer.17 In conclusion, the authors approached rigorously one of the emerging topics in the field of HF. The importance of the study is also enriched by the specific-gender design and by the availability of ejection fraction in a large subgroup of patients. Limitations, as acknowledged by the authors, must be interpreted as a call for unselected community-based studies and for large registry studies exploring in depth how gender, phenotypic presentation and concomitant comorbidities influence the risk of cancer in the HF population. At the same time, translational research in Cardiology and Oncology will aid to understand the biological and pathophysiological factors that link HF to malignancy. Conflict of interest: G.S. reports grants and personal fees from Vifor, AstraZeneca, grants and non-financial support from Boehringer Ingelheim, personal fees from Società Prodotti Antibiotici, Roche, Servier, GENESIS, Cytokinetics, Medtronic, grants from Novartis, Boston Scientific, outside the submitted work. D.S. has nothing to disclose.