Background

With progressive aging of the general population, there is an increasing intersection of cardiovascular diseases (CVDs) with cancer, raising questions around possible cause-and-effect relations between these two diseases [1,2,3]. Heart failure (HF) is one of the most severe complications of many CVDs, and over the past decades we witnessed an exponential growth in its incidence [4].

Cancer patients may develop HF due to the well-known cardiotoxic effects of antineoplastic treatments [5]. Therefore, in order to provide better care to subjects experiencing cardiotoxicity, cardio-oncology was born and gained a relevant role in cardiology settings [5]. Whereas HF risk from cancer drugs has been actively investigated in numerous clinical trials, only few data are available in the setting of “reverse cardio-oncology”, which looks at cancer incidence in patients with pre-existing HF [6]. As a matter of fact, patients with HF are usually excluded from oncological clinical trials because of concerns about cardiotoxicity of cancer drugs, the high prevalence of comorbidities and worse prognosis compared to the general population [1,2,3]. Oncological trials cannot therefore tell us how many patients were affected by HF before cancer was diagnosed.

Over the last years there have been analyses of community-based cohorts suggesting a higher prevalence of cancer in subjects with heart failure (HF) compared with those without HF, even when the influence of shared risk factors, surveillance biases and cardiovascular medications were taken in due consideration [1,2,3]. The association of cancer with HF may have important implications [1,2,3, 6], fostering research on common pathophysiological mechanisms and setting the stage for preventive cancer screening in HF patients and identification of new targets for pharmacological interventions. However, the link between HF and cancer was not confirmed in all available studies [7]. We therefore aimed at performing a systematic review and meta-analysis of cancer incidence and mortality in patients diagnosed with HF.

Methods

This systematic review and meta-analysis was in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA). We conducted a systematic search of PubMed, Google Scholar, Embase and reference lists of relevant articles, from their inception through July 2022. To identify all potentially relevant papers a combination of the following terms was used: “heart failure”, “cancer or tumor or malignancy”, “incidence”, “mortality”. Studies deemed relevant to our search were downloaded and the full manuscripts reviewed. Studies were eligible for inclusion if they were aimed at evaluating cancer incidence in patients with a pre-existing diagnosis of HF and reported hazard ratios (HRs). Analyses including adjusted and unadjusted HRs were used to evaluate the incidence of primary and secondary outcomes of interest (cancer incidence and mortality, respectively). Both adjusted and unadjusted ratios were considered. Three investigators (MC, ML, JGC) independently reviewed study titles and abstracts. I2 testing was performed to evaluate the magnitude of heterogeneity between studies, which was considered substantial when > 50%. Fixed-effects model was used for low heterogeneity among studies (I2 < 50%), while random-effects model was used for high heterogeneity (I2 > 50%). Publication bias for small study effect appraisal was assessed by Egger test and Funnel plots. Computations were performed with the R statistical software (4.0.0 version) using “meta” package.

Results

Of the 1′213 papers examined, 9 met our research criteria and were included in the final analysis, providing data on 515′041 HF cases and 1′365’452 controls without HF [8,9,10,11,12,13,14,15,16]. Study characteristics are reported in Table 1. In the HR analysis, HF patients had a significantly increased cancer incidence compared to controls without HF (HR 1.45, 95% CI (1.31–1.61), p < 0.0001) (Fig. 1, Panel A). However, when papers reporting HRs for cancer mortality were analyzed, no significant difference between the two groups was observed (HR 2.03, 95% CI [0.93–4.43], p = 0.0736) (Fig. 1, Panel B). Having considered that two reports included in our analyses had recruited only male or female patients [11, 15], a sensitivity analysis that excluded these papers was conducted and shown to confirm a higher incidence of cancer in subjects with HF (HR 1.57, 95% CI [1.41.1.76], p < 0.0001) (Fig. 1, Panel C). The same analysis revealed a statistically significant increase also in mortality (HR 2.6824, 95% CI [1.41–5.08], p = 0.0025) (Fig. 1, Panel D).

Table 1 Main characteristics of the analyzed studies
Full size table
Fig. 1
figure 1

Forest plots for the incidence of the primary and secondary outcomes. Panel A cancer incidence in patients with and without Heart Failure (HF). Panel B cancer mortality according to previous diagnosis of HF. Panel C-D sensitivity analysis evaluating cancer incidence and mortality excluding papers with gender restriction

Full size image

Heterogeneity among studies was high (incidence: I2 = 98%, p < 0.01; mortality: I2 = 95%, p < 0.01). This was probably caused by differences in sample size and unbalanced case:control ratios across studies. Therefore, papers with matched 1:1 case:control design were analyzed in isolation. In this secondary analysis there was no detectable heterogeneity among studies and patients with HF showed a significantly higher incidence of cancer (HR 1.76, 95% CI (1.72–1.80), p < 0.0001) (Fig. 2). Cancer mortality was not explored because only two studies remained available for this analysis.

Fig. 2
figure 2

Forest plots for the incidence of the primary outcome (Panel A) in analyses including only papers with matched 1:1 case-control design

Full size image

Further scrutiny of studies with adjusted HRs (8–10;12–15) confirmed that cancer incidence was significantly increased in patients with HF (HR 1.48, 95% CI [1.18–1.87], p = 0.0008), as was cancer mortality as well [8, 10, 15] (HR 2.88, 95% CI [1.64–5.05], p = 0.0002). In both analyses, however, heterogeneity was high (Fig. 3, Panel A-B).

Fig. 3
figure 3

Forest plots for the incidence of the primary and secondary outcomes using adjusted Hazard Ratios. Panel A cancer incidence in patients with and without Heart Failure (HF). Panel B cancer mortality according to previous diagnosis of HF

Full size image

Discussion

To our knowledge, this is the most comprehensive meta-analysis of cancer incidence and mortality in patients diagnosed with HF. As many as nine cohorts were in fact investigated, and cancer incidence proved to be significantly higher in HF patients both in the primary analysis and in the secondary analysis that was performed to reduce heterogeneity among studies. These results extend those reported in another meta-analysis [17] that included only four studies [8,9,10,11].

There are solid pathophysiological foundations to explain the higher risk of cancer in HF patients. Preclinical data show that HF generates a pro-neoplastic milieu characterized by secretion of oncogenic factors and neuro-hormonal activation, which in turn promotes tumor development [1,2,3]. Moreover, HF is characterized by a low-grade systemic inflammation that predisposes to neoplastic transformation and progression [18]. Cancer and HF also share risk factors, such as advanced age, metabolic disorders and smoking habit [19]. Increased cancer risk could well be caused by one or more comorbidities usually observed in patients with HF. This is a rather complex scenario if one considers that e.g., diabetes, obesity, respiratory and renal diseases are almost universally observed in patients affected by HF with preserved ejection fraction [19]. Moreover, HF was shown to associate with enhanced tumor growth, which may be independent of hemodynamic impairment but probably is caused by factors of cardiac origin (e.g. serpinA3) [20].

In our report, cancer mortality in patients with or without HF could be investigated only in the primary analysis, in which differences between the two groups could not be confirmed. For the sake of clarity, we further evaluated cancer mortality when adjusted HRs were available, showing a significantly increased cancer mortality in the HF population compared to controls. The paucity of studies reporting HRs for cancer mortality does nonetheless introduce statistical constraints and precludes firm conclusions.

An overview of clinical trials performed over a timespan of 30 years shows that only 25% of them provided information on cancer mortality in HF patients, mainly because cancer was a protocol-prespecified exclusion criterion. More recently, a meta-analysis of 15 phase 3 clinical trials of HF patients showed that up to 67% of non-cardiovascular deaths were attributable to cancer [21], with the proportion of cancer-related deaths ranging from 6 to 7% and peaking at over 14% of all deaths [21]. Limited number of studies and high heterogeneity between them made our analyses insufficiently powered to detect an increased cancer-related mortality in HF patients. Moreover, active surveillance for HF may have resulted in earlier diagnosis and treatment of cancer, thus modifying mortality rates in HF patients.

Most of the reports included in our analyses were performed with administrative data. This approach presents both strengths and limitations for research purposes. Administrative datasets provide information about large or very large cohorts but this comes at the cost of unmeasured confounders and consequent biases. We therefore acknowledge that our results should be interpreted with due caution. Study limitations should nonetheless be weighed against the fact that our analyses included the most updated sources of cancer incidence in patients with HF.

Conclusions

Our report shows that HF syndrome portends a higher incidence of cancer, even when secondary analyses eliminated biases from heterogeneity among studies. These results may have implications for research developments and patients’ management. Differences in cancer mortality between HF and non-HF patients cannot be firmly established at this point in time, due to the paucity of studies with HRs for cancer-related death. These facts strongly support the inclusion of oncologic endpoints in clinical trials of HF.