Introduction

Individuals diagnosed with cancer face increased short-term risks of cardiovascular events [1] which is often attributed to invasive procedures and chemotherapy. Conversely, analyses from large administrative databases have suggested an increased cancer incidence among individuals who experienced a myocardial infarction, especially during the first 6 months after hospital admission [2,3,4]. From these studies it remains unclear, however, to what extent this finding represents an underlying biological link, or can be explained by increased detection of pre-symptomatic cancers in a population receiving comprehensive clinical work-up (i.e. asymmetrical follow-up) and/or shared risk factors for both atherosclerosis and cancer (i.e. confounding) [2,3,4,5,6].

We aimed to study cancer incidence and its temporality among patients with thrombotic ST-segment-elevation myocardial infarction (STEMI). As a secondary objective, we sought to identify indicators of short-term cancer incidence among STEMI patients.

Methods

We reviewed all consecutive STEMI admissions between 2010 and 2017 at a single regional primary percutaneous coronary intervention center. Patients with rare non-thrombotic causes of STEMI and patients who died during the index hospitalization were excluded, resulting in 1809 patients for analysis. Cancer incidence in these patients was compared with that of an unselected prospective population-based cohort in a neighboring city (the Rotterdam Study; visits RS-I-3, RS-II-1, and RS-III-1; n = 10,052). The objectives and design of the Rotterdam Study have been described in detail previously [7]. For both the STEMI cohort and the Rotterdam Study cohort, individuals with active malignancy, malignancy diagnosed ≤ 5 years before baseline, or coronary intervention ≤ 90 days before baseline were excluded (Fig. 1) [8, 9].

Fig. 1
figure 1

Flowchart of the study population included

Full size image

Data on pathology-confirmed cancer diagnoses (not including non-melanoma skin cancers) were obtained through identical linkage of both cohorts with the Netherlands Cancer Registry of the of the Netherlands Comprehensive Cancer Organization [9]. Follow-up started at the date of STEMI hospitalization or Rotterdam Study research center visit, and was truncated at date of pathology confirmed-cancer diagnosis, death (14.6 per 1000 person-years among the STEMI patients and 12.1 per 1000 person-years among the Rotterdam Study population), loss to follow-up (0.9% of the STEMI patients and 0.3% of the Rotterdam Study population), linkage with the cancer registry, or 5 years of follow-up (to account for longer follow-up in the Rotterdam Study), whichever came first.

Following a pre-specified analysis plan, Cox models were used to obtain hazards ratios (HRs) for incident cancer among STEMI patients compared to the Rotterdam Study population. All analyses were adjusted for factors associated with both atherosclerotic cardiovascular disease and cancer: age, sex, smoking, diabetes, body mass index, and C-reactive protein [5, 6]. Based on prior data, [2,3,4] we stratified the 5-year follow-up period into pre-specified time strata of 0–3, 3–6, 6–12, 12–24, and 24–60 months.

Continuous candidate risk indicators for cancer diagnosis in the first year after STEMI hospitalization were standardized for comparison of corresponding age- and sex-adjusted HRs. However, if proven significant after Bonferroni correction, adjusted HRs for clinically meaningful thresholds were additionally presented in order to enhance clinical interpretation.

IBM SPSS Statistics version 25 was used for all statistical analyses.

Results

Baseline characteristics of the STEMI patients and Rotterdam Study general population sample are displayed in Table 1. A total of 26.0% of the STEMI patients were women, and the median age was 62.6 years. Among the Rotterdam Study participants 57.3% were women and median age was 63.1 years. Of the included STEMI patients, 1.7% presented with a resuscitated cardiac arrest. Nearly all STEMI patients (99.4%) underwent coronary angiography, of whom 95.2% underwent primary percutaneous coronary intervention and 2.5% underwent urgent coronary artery bypass grafting. Significant obstructive coronary disease in non-culprit coronary arteries was present in 38.5% of these patients.

Table 1 Baseline characteristics
Full size table

Over the entire 5-year period, 115 STEMI patients and 677 general population individuals were diagnosed with cancer (crude incidence rates 16.5 and 14.3 per 1000 person-years, respectively; Table 2). Over the full 5-year follow-up, there was no significant difference in the incidence of cancer between STEMI patients and the general population (Fig. 2; HR 0.96, 95% CI 0.78–1.19). However, the proportional hazards assumption was violated [P < 0.001 for Ln(follow-up time) interaction term], indicating that hazards were not stable over the 5-year follow-up period [10]. In the first 3 months after STEMI, cancer incidence was markedly higher among STEMI patients compared to the general population (HR 2.45, 95% CI 1.13–5.30). Differences in cancer incidence between STEMI patients and the general population gradually dissolved during follow-up (P-for-trend 0.004).

Table 2 Incident cancer diagnoses
Full size table
Fig. 2
figure 2

Relative risk of cancer after STEMI and risk indicators for cancer among STEMI patients

Full size image

Sensitivity analyses, including competing risk regression to account for non-cancer mortality, age- and sex-matched analysis, and complete-case analysis yielded similar estimates and patterns (data available upon request). No significant associations were observed with common cancer locations or the presence of disseminated disease at time of cancer diagnosis (Table 2).

Among STEMI patients, higher C-reactive protein, higher platelet counts, and lower hemoglobin were significantly associated with cancer incidence during the first year after STEMI (Fig. 2; age- and sex-adjusted HRs 2.93 for C-reactive protein > 10 mg/dL, 2.10 for platelet count > 300*109, and 3.92 for hemoglobin < 7.5 mmol/L). History of coronary heart disease and presence of non-culprit coronary stenosis were not significantly associated with cancer incidence in any of the follow-up time strata.

Discussion

During the initial period after hospitalization, STEMI patients appeared at increased risk of being diagnosed with (likely pre-existing) cancer. This might be explained by some STEMIs being of 'paraneoplastic' thrombotic origin, for instance due to a pro-thrombotic state induced by an asymptomatic undiagnosed cancer, as we postulated previously [11].

Our results are generally in line with findings from several administrative databases [2,3,4]. These previous studies were hampered by asymmetrical data collection among individuals with and without myocardial infarction, detection bias through clinical work-up of patients with myocardial infarction, or lack of available data on confounding shared risk factors.

In order to study the existence of paraneoplastic STEMI, we chose our reference population carefully to address several issues. First, we ensured identical follow-up using linkage to a nationwide cancer registry, hence ruling out asymmetrical data collection. Next, we aimed to reduce the likelihood of spurious associations by detecting asymptomatic cancers through routine clinical work-up in STEMI patients, since Rotterdam Study participants underwent comparable blood work and examinations. The only marked difference was the prevalence of thoracic imaging: > 90% of STEMI patients underwent chest radiography, while only 35.3% of the Rotterdam Study participants underwent non-enhanced cardiac CT as part of the study protocol. However, lung cancers accounted for only 13.5% of all diagnoses during the initial 3 months of follow-up. Next, in order to address shared etiology, we adjusted for cardiovascular risk factors that have been implicated in cancer development [5]. Nonetheless, residual confounding cannot be ruled out.

The lack of ethnic diversity in our study populations (95.8% of the Rotterdam Study is white) warrants replication in other populations. Also, absolute cancer risks are likely somewhat underestimated, since we only had data available on pathology-confirmed cases [9]. However, this was the case for both populations and hence is unlikely to have affected the presented relative risk estimates.

Conclusion

In conclusion, although rare, thrombotic STEMI might be a paraneoplastic manifestation of yet to be diagnosed cancer, and is hallmarked by a pro-inflammatory status and anemia.