Prevalence of Cancer Across Cardiovascular and Cardiometabolic Conditions: A Systematic Review and Meta‐Analysis

ABSTRACT

Background

To the best of our knowledge, no previous systematic review has comprehensively quantified cancer prevalence across the broad spectrum of cardiovascular and cardiometabolic conditions included in the present study. We aimed to estimate pooled prevalence of active cancer (primary outcome) and other cancer categories among patients with coronary artery disease (CAD), heart failure (HF), atrial fibrillation (AF), hypertension, Type 2 diabetes mellitus (DM), stroke, peripheral artery disease (PAD), and valvular heart disease (VHD).

Methods

PubMed, Web of Science, and Scopus were searched from 2010 to July 2024. Secondary outcomes included prevalence of any, previous, haematological, solid, and metastatic cancer. Prevalence estimates were calculated using one‐step generalized linear mixed models.

Results

A total of 676 studies comprising approximately 180 million participants were included. The primary analysis of active cancer included 59 studies (4,759,695 patients). Active cancer prevalence ranged from 4.22% (95% confidence interval [CI] 2.18–5.32) in Type 2 DM to 5.55% (95% CI 3.97–7.01) in AF. The prevalence of any cancer, a secondary outcome with more heterogeneous ascertainment, ranged from 7.04% (95% CI 6.05–8.03) in CAD to 14.10% (95% CI 12.20–15.99) in AF.

Conclusion

Cancer represents a substantial comorbidity across cardiovascular and cardiometabolic conditions. Although cancer ascertainment methods varied across studies, particularly for any cancer, the findings provide contemporary estimates of cancer burden that may inform future cardio‐oncology research, healthcare planning, risk stratification, and the design of cardiovascular and cardiometabolic clinical trials.

Trial Registration

PROSPERO CRD42023494764

This systematic review and meta‐analysis assessed cancer prevalence across major cardiovascular diseases (CVDs), including coronary artery disease, heart failure, atrial fibrillation, hypertension, Type 2 diabetes, stroke, peripheral arterial disease, and valvular heart disease. A total of 676 studies involving approximately 180 million participants were included. Active cancer prevalence ranged from 4.22% in Type 2 diabetes to 5.55% in atrial fibrillation, while any cancer prevalence ranged from 7.04% in coronary artery disease to 14.10% in atrial fibrillation. These findings highlight a substantial cancer burden among patients with CVDs.

1. Introduction

Cardiovascular diseases (CVDs) and cancer are the two leading causes of mortality worldwide, accounting for ~19 million CVD deaths annually (up from ~13 million in 1990) [1] and ~20.0 million new cancer cases with 9.7–9.8 million deaths in 2022 [2]. Although distinct clinical entities, they share common risk factors and biological pathways [3, 4, 5]. Patients with cancer have an increased risk of developing CVD due to both cancer biology and the cardiotoxic effects of oncological therapies [6, 7, 8]. Conversely, CVD may promote tumourigenesis through cytokine release, protumourigenic extracellular vesicles, immune dysregulation, and microbial dysbiosis [3, 9]. While substantial epidemiological evidence supports this bidirectional relationship [10, 11, 12, 13], data on the prevalence of cancer among patients with CVD remain limited. Although several large cohort studies have reported prevalence estimates [14, 15, 16], no systematic review has synthesized these data across CVD populations.

Quantifying cancer prevalence in patients with CVD is important for defining the burden of coexisting disease, informing surveillance and multidisciplinary management strategies, and guiding the design of future randomized controlled trials (RCTs), in which patients with cancer and other comorbidities are often underrepresented [17, 18]. These estimates may also support healthcare planning, including the development of integrated cardio‐oncology services, resource allocation, reimbursement policies, and early detection initiatives [19].

We therefore conducted a systematic review and meta‐analysis to estimate cancer prevalence across a prespecified spectrum of cardiovascular and cardiometabolic conditions (registered prospectively in PROSPERO), including coronary artery disease (CAD), heart failure (HF), atrial fibrillation (AF), hypertension, Type 2 diabetes mellitus (DM), stroke, peripheral artery disease (PAD), and valvular heart disease (VHD). Each condition was analysed separately to provide disease‐specific prevalence estimates. We also investigated potential determinants of between‐study variability using prespecified meta‐regression analyses.

2. Methods

2.1. Search Strategy

This meta‐analysis was based on previously published literature and did not involve human participants; therefore, ethical approval was not required. The study protocol was prospectively registered in PROSPERO (CRD42023494764). This publication represents the first stage of the project focused on estimating cancer prevalence in real‐world settings. The review followed the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) recommendations [20].

A systematic search was conducted in PubMed, Web of Science, and Scopus. To obtain contemporary prevalence estimates and manage the large literature volume, we limited inclusion to full‐length articles published after 2010. Detailed search terms, keywords, and filters are provided in the Supplementary Methods. No language restrictions were applied. Additional sources, including study registries, high‐impact journal websites, and conference proceedings, were also screened to identify eligible studies. The search process flowchart was generated using a dedicated ShinyApp web tool [21].

2.2. Screening and Eligibility Criteria

Eligible studies included adults (> 18 years) with CAD (acute/chronic coronary syndromes, PCI), HF (acute/chronic; preserved or reduced ejection fraction), AF (including non‐valvular AF), hypertension, Type 2 DM, stroke (ischaemic or haemorrhagic), PAD, or VHD, according to original study definitions. The included cardiovascular and cardiometabolic conditions were prespecified in the registered protocol (PROSPERO CRD42023494764); no post hoc modifications to disease eligibility were made. Pregnancy was excluded.

Studies were required to report baseline cancer prevalence (number and/or proportion). Only patient‐level data were included; hospitalization‐level studies were excluded because repeated admissions could bias prevalence estimates [22, 23, 24]. Studies limited to in situ disease or benign neoplasms were excluded.

Eligible designs comprised prospective/retrospective cohort, case–control, and cross‐sectional observational studies from inpatient or outpatient settings. RCTs, reviews, editorials, correspondence, conference abstracts, and propensity score–based studies were excluded because pseudo‐randomization may distort real‐world prevalence estimates [25]. When multiple publications reported the same cohort, the most representative study was selected based on eligibility criteria, enrollment period, sample size, and baseline characteristics. Only studies with sample sizes ≥ 500 were included, consistent with recommendations for expected cancer prevalence of 3%–20% [26].

Reported outcomes included active cancer (defined as a diagnosis of cancer within the previous 6 months, recurrent, locally advanced or metastatic disease, or receipt of anticancer treatment within the previous 6 months) [27], history of cancer (≥ 6 months before enrolment, when reported), any cancer (undefined ascertainment), haematologic malignancies (lymphoma, leukaemia, and/or myeloma), solid cancers, and metastatic disease. Active cancer prevalence was considered the primary outcome because ascertainment methods, coding practices, and look‐back periods for other outcomes were inconsistently reported, limiting interpretability.

Title/abstract and full‐text screening were conducted in Rayyan [28]. Disagreements were resolved by consensus among all reviewers.

2.3. Data Extraction and Risk of Bias Assessment

Data extraction was performed using the Systematic Review Database Repository Plus platform [29]. Extracted variables included authors, publication year, study design, setting, enrolment period, country, eligibility criteria, baseline population characteristics (age, sex, race, CV risk factors, cardiovascular diseases, comorbidities), and cancer prevalence data. CV risk factors comprised DM, hypertension, dyslipidaemia, obesity, smoking, and alcohol use; cardiovascular diseases included CAD, stroke, HF, AF, PAD, and VHD; comorbidities included cerebrovascular disease, liver abnormalities, CKD, gout, asthma, COPD, thyroid disorders, and anaemia.

When only study arm–level statistics were reported, aggregate estimates for the target population were derived using Cochrane Handbook methods [30]. Means and standard deviations (including age) were calculated from medians and interquartile ranges when required, according to Wan et al. [31].

Risk of bias was assessed using the Joanna Briggs Institute (JBI) Prevalence Critical Appraisal Tool, which evaluates methodological quality of prevalence studies, including sampling, measurement reliability, statistical analysis, and response rates [32]. Data extraction and quality assessment were conducted independently by multiple reviewers, with discrepancies resolved by discussion among all authors.

2.4. Statistical Analyses

Analyses followed the prespecified protocol (CRD42023494764). To avoid bias associated with traditional two‐step arcsine‐based meta‐analytic methods, pooled prevalence was estimated using one‐step generalized linear mixed models [33, 34]. Both median proportions and population‐averaged prevalence estimates were calculated. Median estimates summarize included studies, whereas population‐averaged estimates are interpretable at the target population level; these estimates apply only to populations meeting our predefined inclusion criteria (individuals with CVD and cardiovascular risk factors) and should not be generalized to broader populations beyond this scope. Confidence intervals (CIs) for population‐averaged estimates were obtained by bootstrap resampling (1000 iterations) [33].

Heterogeneity was assessed using tau² statistics. Meta‐regression analyses were performed to explore heterogeneity and derive age‐adjusted prevalence estimates when ≥ 10 studies were available [30]. Study‐specific prevalences were transformed to the logit scale using the logit transformation before analysis. Random‐effects meta‐regression models were fitted using maximum likelihood estimation with the rma() function from the metafor package. Age was included as a continuous moderator. Regression coefficients and model fitting were performed on the logit scale. Predicted prevalences and corresponding 95% confidence intervals were obtained by applying the inverse logit transformation to the model estimates. Subgroup analyses were conducted according to United Nations Sustainable Development Goal geographic regions, with results reported only when ≥ 3 studies were available [35].

Although Doi plots and LFK indices were prespecified [36], publication bias was assessed using multiple complementary methods, prioritizing the Thompson and Sharp test due to concerns regarding LFK false positives under high heterogeneity [37]. Such assessment was considered supplementary, consistent with recommendations for meta‐analyses of proportions [38].

Analyses were conducted in R using code proposed by Lin et al. and the metafor and lme4 packages [33, 39, 40]. Certainty of evidence was graded according to GRADE Working Group guidance [41]. Patients and the public were not involved in study design, conduct, reporting, or dissemination. Ethical approval was not required because only published data from studies with participant informed consent were used.

3. Results

3.1. Study Characteristics

A total of 676 articles (686 cohorts; 180,875,531 patients) were identified (Supplementary References); the study selection process is shown in Figure 1. Of these, 59 studies including 4,759,695 patients reported active cancer prevalence and comprised the main analysis. Study characteristics by UN geographic region are summarized in Table 1. Most cohorts originated from Europe/Northern America (n = 41) and Eastern/South‐Eastern Asia (n = 11), whereas Oceania (n = 3) and Northern Africa/Western Asia (n = 1) were underrepresented. Mean participant age ranged from 63.6 years in Northern Africa/Western Asia to 77.4 years in Europe/Northern America, while female representation varied from 26.2% in Eastern/South‐Eastern Asia to 53.8% in Europe/Northern America.

FIGURE 1.

The flow‐diagram for the systematic search.

TABLE 1.

The baseline characteristics of studies with reported prevalence of active cancer across the world geographic regions.

Variable	Europe and Northern America	Eastern and South‐Eastern Asia	Oceania	Northern Africa and Western Asia	Multiregional	Total
N of cohorts	41	11	3	1	3	59
Sample size	4,440,754	246,217	55,067	3089	14,568	4,759,695
Age, years	77.36	70.39	64.14	63.59	70.66	76.83
Female, %	53.81	26.21	34.3		31.12	52.35
Study design
Cross‐sectional study	1 (2.4%)					1 (1.7%)
Prospective cohort	15 (36.6%)	6 (54.5%)	2 (66.7%)	1 (100%)	2 (66.7%)	26 (44.1%)
Retrospective cohort	25 (61%)	5 (45.5%)	1 (33.3%)		1 (33.3%)	32 (54.2%)
Inpatient/outpatient
Inpatient	14 (34.1%)	5 (45.5%)	1 (33.3%)	1 (100%)	1 (33.3%)	22/39 (56.4%)
Outpatient	6 (14.6%)					6/39 (15.4%)
Both	6 (14.6%)	3 (27.3%)	1 (33.3%)		1 (33.3%)	11/39 (28.2%)
Multicentre/single‐centre
Multicentre	33 (80.5%)	7 (63.6%)	3 (100%)	1 (100%)	3 (100%)	47 (79.7%)
Single‐centre	8 (19.5%)	4 (36.4%)				12 (20.3%)
Risk factors/comorbidities
Smoking	22.54	44.59				22.7
Active smoking	15.34	19.03	23.18	40.11		17.29
Past smoking	40.82	22	42	22.15		38.44
Dyslipidaemia	60.23	60.88	57.68	72.56	73.9	60.53
Obesity	21.47				34.23	21.73
Alcoholism	1.7	0.6			0.8	1.69
History of stroke	9.02	17.96	2.24	8.34	13.92	9.23
Cerebrovascular disease	5.28	6.58				5.52
Liver diseases	2.12	39.35	2		4.95	2.25
Cirrhosis	0.3	2.05	0.1			0.73
Chronic kidney disease	29.03	7.51	3.85	11.54	35.31	27.73
End‐stage kidney disease	3.9	5.57				3.97
Chronic obstructive pulmonary disease	19.62	6.44	3.09		19.79	17.75
Anaemia	55.87	34.92			25.6	55.03

Most studies were retrospective cohorts (54.2%), conducted in inpatient settings (56.4%) and across multiple centres (79.7%). The pooled study population had a high burden of traditional CV risk factors and comorbidities, including smoking (22.7%), dyslipidaemia (60.5%), obesity (21.7%), CKD (27.7%), COPD (17.8%), and prior stroke (9.2%) (Table 1). Study‐level baseline characteristics are provided in Tables S1 and S2.

3.2. Risk of Bias Assessment

Risk of bias results for the main analysis of active cancer prevalence are shown in Table S3. All studies were based on nationwide or large cohorts with all, consecutive, or random patient samples and were therefore judged as low risk of bias for sample representativeness, recruitment, and sample size. Although all studies reported baseline characteristics, 35/59 did not provide patient flow diagrams and/or losses to follow‐up, resulting in unclear risk regarding sample coverage. Measurement reliability was also frequently unclear, as most studies did not report validated or consensus‐based cancer ascertainment methods (50/59) or procedures for data quality assessment (55/59). To mitigate the risks related to cancer ascertainment methods, we utilized strong definitions for active cancer ascertainment based on the previous research [27]. Because none of the included studies were designed primarily to assess cancer prevalence, the final three JBI Prevalence Critical Appraisal Tool domains (statistical analyses for prevalence data, confounding, and subgroup considerations) were considered not applicable.

3.3. Cancer Prevalence in CAD and Its Subtypes

The main analysis of active cancer included 16 cohorts (819,468 participants), predominantly multicentre retrospective inpatient studies. Mean age was 67.9 ± 1.7 years, with 26.5% females (Table S4). The population had a high burden of CV risk factors (hypertension 61.0%, DM 32.2%, smoking 53.6%, dyslipidaemia 62.6%, obesity 42.8%) and comorbidities (CKD 10.5%, COPD 8.6%, anaemia 25.6%; Table S4).

Pooled prevalence estimates for active, any, and previous cancer were 4.61% (95% CI 2.83–6.97; Tau² = 0.64, Figure 2A), 7.04% (95% CI 6.05–8.03; Tau² = 0.70), and 9.72% (95% CI 6.97–14.17; Tau² = 0.87, Table S5), respectively. Prevalence rates were broadly consistent across geographic regions. Meta‐regression suggested an ecological association between higher active cancer prevalence in studies and greater PAD and COPD burden (Table S6). The model‐predicted prevalence of any cancer increased with age, reaching 10.11% (95% CI 6.25–15.94) in patients > 80 years (Table S7). Prevalence estimates for solid cancers, haematologic malignancies, and metastatic disease were 4.39% (95% CI 2.04–7.72), 0.62% (95% CI 0.39–0.90), and 1.39% (95% CI 0.81–2.23), respectively (Table S5).

FIGURE 2.

Pooled prevalence of active cancer in coronary artery disease (A) and atrial fibrillation (B). CI, confidence intervals.

Across CAD subtypes, active cancer prevalence was comparable in acute coronary syndrome, acute myocardial infarction, and PCI populations: 3.12% (95% CI 1.91–5.13), 2.41% (95% CI 1.56–3.03), and 4.53% (95% CI 2.52–7.39), respectively (Table S5; Figures S1–S3). Previous cancer prevalence was highest among PCI patients (13.93%, 95% CI 7.25–24.76; Tau² = 1.18, Table S5).

3.4. Cancer Prevalence in AF and Its Subtypes

The meta‐analysis of active cancer in AF included 18 cohorts (642,836 individuals; mean age 75.9 ± 5.2 years; 42.3% females; Table S4). Active cancer prevalence was 5.55% (95% CI 3.97–7.01; Tau² = 0.63; Figure 2B), with higher rates in Eastern/South‐Eastern Asia (8.88%, 95% CI 7.15–10.94; p = 0.04; 3 studies; Table S5). Prevalence estimates for solid, haematologic, metastatic, and any cancer were 8.43% (95% CI 1.62–15.24), 1.26% (95% CI 0.88–1.72), 2.74% (95% CI 1.68–4.14), and 14.10% (95% CI 12.20–15.99), respectively (Table S5). The exploratory study‐level meta‐regression identified an ecological association between prevalence estimates with greater burdens of HF, hypertension, and prior stroke (Table S6). The model‐predicted prevalence of active cancer increased with age, reaching 2.9% (95% CI 1.80–4.63) and 6.34% (95% CI 3.97–9.99) in populations aged > 70 and > 80 years, respectively (Table S7).

For non‐valvular AF, 8 cohorts (279,408 patients; mean age 77.8 years; 40.6% females) showed an active cancer prevalence of 5.87% (95% CI 3.92–7.74; Tau² = 0.39; Table S4, Figure S4). Corresponding prevalence estimates for any, previous, solid, haematologic, and metastatic cancers were 14.03% (95% CI 12.19–15.73), 14.31% (95% CI 9.36–19.94), 11.06% (95% CI 4.97–15.60), 1.27% (95% CI 0.84–1.61), and 2.95% (95% CI 1.57–5.48), respectively (Table S5).

3.5. Cancer Prevalence in HF and Its Subtypes

The data for HF were obtained from 7 cohorts with a total population of 3,070,089 participants (mean age 79.3 ± 2.6 years, females 55.8%). The proportion of hypertension, DM, dyslipidaemia, CKD, and COPD was 74.9%, 48.7%, 22.6%, 34.4%, and 32.8%, respectively (Table S4).

The pooled prevalence of active malignancies was 4.43% (95% CI 2.78–6.38, Tau² = 0.36, Figure 3A) without any identified significant discrepancies across the regions (Table S5). Regarding other outcomes, the highest estimate was identified for history of cancer at 13.06% (95% CI 8.42–18.25), while the prevalence rates for any, solid cancers, haematologic malignancies, and metastases were 12.82% (95% CI 10.60–14.89), 9.46% (95% CI 5.08–12.59), 1.11% (95% CI 0.39–1.63), and 2.02% (95% CI 1.47–2.57), respectively (Table S5). The meta‐regression analysis showed a positive study‐level ecological linear association between age and prevalence of any cancer (Table S7).

FIGURE 3.

Pooled prevalence of active cancer in heart failure (A) and stroke (B). CI, confidence intervals.

For acute HF, the total prevalence of active cancer was 9.75% (95% CI 7.03–12.36, Figure S5) with other outcomes provided in Table S5.

3.6. Cancer Prevalence in Stroke and Its Subtypes

Data from 6 cohorts (84,810 participants; mean age 76.2 ± 5.3 years; 47.8% female) were included (Table S4). The pooled prevalence of active cancer in stroke was 4.60% (95% CI 1.72–8.13, Tau² = 1.15; Figure 3B, Table S5). The prevalence of any, previous, solid, haematologic, and metastatic cancers was 9.52% (95% CI 7.33–12.19), 13.81% (95% CI 10.90–16.60), 7.02% (95% CI 2.86–11.57), 0.65% (95% CI 0.33–0.94), and 1.19% (95% CI 0.80–1.57), respectively (Table S5). As shown by the study‐level meta‐regression, any cancer was most prevalent in patients aged > 70 and > 80 years (7.6%, 95% CI 5.72–10.05; 9.47%, 95% CI 6.53–13.53; Table S7). By subtype, active cancer prevalence was 5.26% (95% CI 2.40–8.91, Tau² = 0.61; Figure S6) in ischaemic stroke, while any cancer prevalence reached 11.49% (95% CI 6.49–16.97, Tau² = 0.41; Table S5) in haemorrhagic stroke.

3.7. Cancer Prevalence in PAD

Data on active cancer prevalence in PAD were insufficient for meta‐analysis. Analysis of 16 studies (6,246,746 participants; mean age 71.2 ± 2.5 years; 42.5% female) showed prevalences of HF, hypertension, DM, dyslipidaemia, prior stroke, and CKD of 21.9%, 59.6%, 44.1%, 63.1%, 12.9%, and 30.3%, respectively (Table S8). The pooled prevalence of any cancer was 13.20% (95% CI 9.11–17.61, Tau² = 0.78, I ² = 100%; Table S5, Figure 4A), driven mainly by European and North American studies. Metastatic cancer prevalence was 4.44% (95% CI 3.25–5.45; 3 studies, 27,662 patients; Table S5).

FIGURE 4.

Pooled prevalence of any cancer in peripheral artery disease (A) and active cancer in valvular heart disease (B). CI, confidence intervals.

3.8. Cancer Prevalence in VHD and Its Subtypes

Eight cohorts (151,839 participants; mean age 81.5 ± 0.9 years; 48.1% female) were analysed. Common comorbidities included hypertension (92.3%), DM (39.8%), dyslipidaemia (42.6%), prior stroke (16.6%), liver disease (3.8%), CKD (32.7%), COPD (37.4%), and anaemia (58.3%) (Table S4). Active cancer prevalence in the overall cohort was 4.90% (95% CI 3.84–6.37, Tau² = 0.12; Figure 4B, Table S5). The prevalence of prior, any, haematologic, and metastatic cancers was 15.69% (95% CI 12.31–18.65), 10.32% (95% CI 8.28–12.35), 1.48% (95% CI 0.99–1.83), and 1.56% (95% CI 0.37–2.94), respectively (Table S5). In aortic stenosis, active, prior, and any cancer prevalence was 4.42% (95% CI 3.64–5.21, Figure S7), 17.74% (95% CI 14.60–20.03), and 10.53% (95% CI 7.70–13.36), respectively (Table S5); corresponding estimates after TAVI were 4.87% (95% CI 4.11–5.59, Figure S8), 18.28% (95% CI 14.40–20.71), and 10.63% (95% CI 7.12–15.01) (Table S5). The age‐specific model‐predicted prevalence estimates of any cancer are provided in Table S7.

3.9. Cancer Prevalence in Hypertension and Type 2 DM

For Type 2 DM, active cancer data were derived from 3 cohorts (n = 117,866; mean age 72 ± 1.4 years; 40.1% females; Table S4). Active cancer prevalence was 4.22% (95% CI 2.18–5.32; Tau² = 0.99; Table S7, Figure 5A). Prevalence of any cancer, prior malignancy, and metastases was 9.52% (95% CI 8.25–11.02), 12.72% (95% CI 4.87–23.94), and 1.00% (95% CI 0.53–1.38), respectively (Table S5).

FIGURE 5.

Pooled prevalence of active cancer in Type 2 diabetes mellitus (A) and any cancer in hypertension (B). CI, confidence intervals.

For hypertension, meta‐analysis was limited to any cancer due to scarce data. One study (Alanaeme et al.) contributed weighted survey data (~97 million individuals), substantially influencing pooled estimates; sensitivity analyses were performed excluding this study. After exclusion, the pooled cohort comprised 721,074 individuals (mean age 69.2 ± 8.4 years), with a high prevalence of smoking (48.1%), dyslipidaemia (43.4%), DM (29.2%), cerebrovascular disease (19.6%), and asthma (16.7%) (Tables S9 and S10).

Pooled prevalence of any cancer in hypertension was 9.01% (95% CI 5.16–13.54; Tau² = 1.72; I ² = 100%; Figure 5B, Table S5; without Alanaeme et al.) and 9.30% (95% CI 5.50–13.33; Tau² = 1.67; I ² = 100%; Table S5). The age‐specific model‐predicted prevalence estimates are presented in Table S7.

3.10. The Assessment of Publication Bias

The results of publication bias assessment were valid for those analyses that included 10 or more studies. No statistically significant asymmetry was detected for meta‐analyses for prevalence of active cancer in patients with CAD, AF, and non‐valvular AF (Table S11). GRADE assessments are presented in Table S12. The certainty of evidence was rated as moderate for all active cancer prevalence outcomes (CAD, AF, HF, stroke, VHD, and Type 2 DM). For the secondary outcome of any cancer, the certainty of evidence was low for peripheral artery disease and hypertension because of indirectness, whereas the remaining GRADE domains were rated as moderate.

3.11. Cancer Prevalence in Total Study Population

The prevalence rates of active, any, prior, haematologic, metastatic, and solid cancers were 5.36% (95% CI 4.39–6.33), 10.27% (95% CI 9.62–10.92), 11.72% (10.10–13.49), 0.97% (95% CI 0.73–1.20), 1.73% (95% CI 1.38–2.13), and 7.15% (95% CI 4.77–9.41), respectively (Table S5).

4. Discussion

To our knowledge, this is the first systematic review and meta‐analysis summarizing cancer prevalence across common cardiovascular (CV) diseases and risk factors globally. Pooled prevalence of active cancer was relatively consistent across conditions, ranging from 4.2% in Type 2 DM to 5.6% in AF. Prevalence of any cancer was highest in AF (14.1%), PAD (13.2%), and chronic HF (12.8%). We also report pooled estimates for prior malignancy, haematologic and solid cancers, advanced cancer, and demonstrate positive associations between cancer prevalence, increasing age, and comorbidity burden. Importantly, our inclusion of Type 2 DM should not be interpreted as an attempt to reassess its association with cancer risk. Previous systematic reviews and meta‐analyses have already established that Type 2 DM is associated with an increased incidence of multiple malignancies, and recent translational reviews have comprehensively summarized the biological mechanisms underlying this relationship [42, 43, 44]. In contrast, the present review addresses a different clinical question by quantifying the prevalence of cancer among patients with established cardiovascular and cardiometabolic diseases. These prevalence estimates complement existing incidence‐based evidence by characterizing the contemporary burden of cancer encountered in routine cardiovascular care, thereby providing information relevant to healthcare planning, cardio‐oncology services, risk stratification, and the design of future cardiovascular and cardiometabolic clinical trials. Although diabetes has disease‐specific biological mechanisms linking it to cancer [45], each disease category in our review was analysed independently, ensuring that prevalence estimates reflect the burden of cancer within each individual condition rather than comparisons of causal pathways across diseases.

Active cancer estimates are likely more clinically interpretable than those for any cancer because most studies explicitly defined active disease as current treatment, recent diagnosis, or life‐threatening malignancy. By contrast, definitions of any cancer were often poorly specified regarding look‐back periods and ascertainment methods, potentially encompassing both remote and recent malignancies. This variability likely contributed to substantial heterogeneity; therefore, estimates for any cancer should be interpreted cautiously and viewed as hypothesis‐generating.

Cancer and CVD share a complex bidirectional relationship [3, 46]. Common risk factors—including aging, smoking, obesity, dyslipidaemia, DM, hypertension, sedentary lifestyle, diet, and alcohol use—drive chronic inflammation, oxidative stress, endothelial dysfunction, metabolic derangements, immune dysregulation, DNA damage, and cellular proliferation, promoting both atherosclerosis and carcinogenesis [3, 4, 5, 46]. Cancer can promote CVD through systemic inflammation, prothrombotic states, tumour‐derived mediators, and treatment‐related cardiotoxicity [3, 4, 5, 46].

While the cardiovascular consequences of cancer and its therapies are well established [8, 47], growing translational and epidemiological evidence suggests that CVD itself may facilitate malignant transformation, giving rise to the emerging field of reverse cardio‐oncology [3, 9, 46, 48]. Systematic reviews and large registries consistently report higher cancer incidence among patients with CVD and CV risk factors than in the general population [10, 11, 12, 13]. Proposed mechanisms include haematopoietic reprogramming, protumourigenic cytokines and extracellular vesicles, gut microbiome dysbiosis, and immunosuppressive pathways enabling tumour immune escape [3, 9, 49, 50, 51, 52].

The biological relationship between Type 2 DM and cancer differs in important respects from that between cardiovascular disease and cancer. While CVD and cancer largely arise from shared upstream risk factors, including aging, obesity, smoking, hypertension, dyslipidaemia, chronic inflammation, and endothelial dysfunction [3], diabetes may additionally promote carcinogenesis through metabolic disturbances intrinsic to the diabetic state [45]. Chronic hyperglycaemia induces oxidative stress, advanced glycation end‐product formation, DNA damage, epigenetic remodelling, and activation of oncogenic signalling pathways such as PI3K/AKT/mTOR, Wnt/β‐catenin, O‐GlcNAc, and AMPK. Concurrent insulin resistance and compensatory hyperinsulinemia enhance insulin and insulin‐like growth factor‐1 signalling, stimulating cellular proliferation, inhibiting apoptosis, and facilitating tumour progression. Persistent low‐grade inflammation and adipokine dysregulation further contribute to a protumourigenic microenvironment [43, 44]. These mechanisms provide a biological explanation for the well‐established association between Type 2 DM and increased cancer incidence and distinguish diabetes from cardiovascular diseases, in which the association with cancer appears to be driven predominantly by shared risk factors and emerging mechanisms of reverse cardio‐oncology.

4.1. Study Limitations

Several limitations should be acknowledged. First, the limited number of studies precluded age‐specific model‐predicted prevalence estimates for active cancer in most populations except AF, restricting comparisons with the general population. Second, heterogeneity in cancer ascertainment across countries, healthcare systems, databases, coding practices, and definitions of cancer status likely contributed to between‐study variability. Meanwhile, the majority of included registry‐ and administrative database‐based studies identified cancer using standardized ICD‐10 (or equivalent) diagnostic codes; differences in coding algorithms, ascertainment procedures, and look‐back periods remained unavoidable across healthcare systems. Although several analyses demonstrated substantial I ² values, this finding is expected in meta‐analyses of prevalence and should be interpreted cautiously [53]. Recent empirical evidence evaluating more than 130 prevalence meta‐analyses has shown that high I ² values are almost universal, increase with the number of included studies and extreme prevalence estimates, and do not necessarily reflect excessive clinical or methodological heterogeneity [54]. Accordingly, we applied the one‐step generalized linear mixed model with population‐averaged estimation proposed by Lin et al. [33], which was specifically developed for meta‐analyses of proportions and provides clinically interpretable prevalence estimates despite substantial between‐study variability. Third, limited data prevented subgroup analyses by cancer stage, sex, and race/ethnicity. Fourth, an individual participant data meta‐analysis would likely yield more precise estimates and permit more comprehensive adjustment for confounding than aggregate published data, but was not feasible. Finally, because the primary literature predominantly consisted of descriptive cohort studies without matched control populations, we were unable to quantitatively compare cancer prevalence with age‐, sex‐, and calendar period‐matched general populations. Such comparative analyses should be considered a priority for future research.

5. Conclusion

We estimated cancer prevalence across a broad spectrum of cardiovascular and cardiometabolic conditions. The findings demonstrate a substantial burden of concomitant cancer in these populations and provide contemporary evidence to inform future cardio‐oncology research, healthcare planning, risk stratification, and the design of cardiovascular and cardiometabolic clinical trials.

Author Contributions

M.A.M. is an author of the idea and conceptualized the study design. M.A.M. is a guarantor of the manuscript. A.G., M.A., B.S.S., A.M., M.A.M. designed the study and wrote the study protocol. A.G., E.B., L.C.L., A.B.A.Z., G.D., M.A., B.S.S., A.M. conducted the systematic search, study selection, extraction, and risk of bias assessment. A.G. performed statistical analyses. M.A.M., B.K., R.B. supervised statistical analyses. A.G. and M.A.M. drafted the manuscript. M.A.M., R.B., L.Y.S., C.H.N., M.N.M., S.Z., B.K. supervised the writing. All authors had full access to the data. All authors participated in the interpretation of the results, review and approval of the paper, and the decision to submit it for publication. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.

Funding

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1: Supporting Information.

Figure S1: Pooled prevalence of active cancer in acute coronary syndromes.

Figure S2: Pooled prevalence of active cancer in acute myocardial infarction.

Figure S3: Pooled prevalence of active cancer in percutaneous coronary intervention.

Figure S4: Pooled prevalence of active cancer in non‐valvular atrial fibrillation.

Figure S5: Pooled prevalence of active cancer in acute heart failure.

Figure S6: Pooled prevalence of active cancer in ischaemic stroke.

Figure S7: Pooled prevalence of active cancer in aortic stenosis.

Figure S8: Pooled prevalence of active cancer in trans‐catheter aortic valve intervention.

Figure S9: Pooled prevalence of active cancer in the total population.

Table S1: The baseline characteristics of studies with reported prevalence of active cancer.

Table S2: The baseline characteristics of total population across the world geographic regions with consideration of all studies.

Table S3: The risk of bias assessment.

Table S4: Baseline characteristics studies included in meta‐analysis of prevalence of active cancer categorized by cardiovascular diseases.

Table S5: The pooled prevalence rates of cancer obtained during the main and sensitivity analyses and subgroup analyses across different geographical regions.

Table S6: The results of meta‐regression analyses.

Table S7: The age‐specific model‐predicted prevalence estimates of cancer.

Table S8: Baseline characteristics studies included in meta‐analysis of prevalence of any cancer in patients with peripheral artery disease.

Table S9: Baseline characteristics studies included in meta‐analysis of prevalence of any cancer in patients with hypertension.

Table S10: Baseline characteristics studies included in meta‐analysis of prevalence of any cancer in patients with hypertension after exclusion of Alanaeme et al.'s study.

Table S11: The results of publication bias assessment.

Table S12: Certainty of the evidence table.

Data S2: Supporting Information.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.