Abstract
Introduction: Patients with cancer are known to have an increased risk of ischemic stroke (IS) around the time of their diagnosis. However, there is a paucity of data in Asian populations, and as such, we aimed to determine cancer incidence rates and patterns in Asian IS patients as well as investigate the differences in vascular risk profile of IS patients with and without concomitant cancer. Methods: We conducted a retrospective cross-sectional study using data from the Singapore Stroke and Cancer registries. We defined cases as patients with IS and a cancer diagnosis 2 years before or after the index IS. Cancer incidence was determined using the same direct age-standardization method performed for the Singapore general population in the 2015 Singapore cancer report. Multivariable logistic regression was used to analyze differences in vascular risk factors. Results: Among 21,068 IS patients (mean age, 67.9 ± 13.3 years), 6.3% (1,330) were found to have concomitant cancer; 4.4% (935) had prior cancer while 1.8% (395) had cancer diagnoses within 2 years following IS. The cancer incidence among IS patients was 3,393 (95% confidence interval [CI], 1,937–4,849) per 100,000 person-years compared to 219–231 per 100,000 person-years in the general population. Older age (odds ratio [OR], 1.02 [95% CI, 1.01–1.02] per year), males (OR, 1.25 [95% CI, 1.11–1.41), Chinese ethnicity (OR, 1.61 [95% CI, 1.37–1.89]) and a lower prevalence of hypertension (OR, 0.84 [95% CI, 0.73–0.97)]), and hyperlipidemia (OR, 0.53 [95% CI, 0.45–0.62]) were independently associated with cancer-related IS. Conclusions: The age-standardized cancer incidence was 15 times higher in IS patients than the general population. IS patients with concomitant cancer were older and had a lower prevalence of vascular risk factors.
Introduction
Ischemic stroke (IS) and cancer are both top 10 causes of death globally. Associations between the two have been found, with cancer patients exhibiting a higher stroke risk [1] and IS patients having a higher prevalence of cancer compared to the general population [2‒4]. IS may also be the first manifestation of an occult malignancy [5‒8]. Efforts have been made to identify differences in profile of IS patients with and without cancer in order to develop clinical scores to predict active cancer in IS patients [9, 10].
The profile of cancer risk and survival is different amongst patients with African, Asian, and Caucasian ancestry, especially so within breast, prostate, and colorectal cancer [11]. Even within Asian cancer patients, specific ancestry has been known to affect subtypes of cancer [12]. Studies from Asia thus far have shown that there is an association between stroke and cancer [6, 13, 14] in specific cancer subtypes; however, these studies have been mainly in East Asian populations. In this study, we aimed to look at the association between IS and cancer in a multiethnic population.
Aims
Using data from the Singapore Stroke Registry (SSR) and Singapore Cancer Registry (SCR), we aimed to determine the cancer incidence and patterns in IS patients in our multiethnic cohort. We also investigated differences in vascular risk factors in IS patients with and without concomitant cancer.
Methods
Registry Data
We conducted a retrospective cross-sectional study using the data from the SSR and SCR, both which are maintained by the National Registry of Disease Office (NDRO) [15]. The SSR [16] was established in 2002 with data from two major public hospitals in Singapore but expanded to receive data from all public healthcare institutions in Singapore from 2005. Hospital admission data are reviewed by trained research nurses to record data on stroke characteristics, as well as stroke-related outcomes. The SSR reviews all public hospital admissions from MediClaims listings, discharge summaries, and from the death registry. The data fields of the SSR are available in the online supplement materials (online suppl. eMethods 1; for all online suppl. material, see https://doi.org/10.1159/000534267). The latest report from the SSR is from 2020 [17].
The SCR was established in 1968 and contains data on all reportable cancer cases (online suppl. eMethods 2) including those from public and private healthcare institutions. Data sources include notifications by medical profession, pathology records, hospital records, public insurance claims, and the death registry. Clinical data in the NRDO database were captured from medical records from relevant institutions and verified by registry coordinators and a visiting consultant pathologist. The latest available report from the SCR is from 2020 [18]. For the purposes of these analyses, borderline and benign tumors were excluded from a cancer diagnosis.
Study Population
In this study, we included IS patients from the SSR from years 2008 to 2012 who were aged ≥18 years and of Asian ethnicity. Patients who were identified as having IS secondary to primary brain tumors or metastases and venous strokes were excluded. We screened the SSR for IS patients and then overlapped records with the SCR from year 2006 to 2014 to determine presence of concomitant cancer. We defined cancer-IS patients as cancer diagnoses up to 2 years before or after IS onset in accordance with findings from previous studies [3]. In situations of multiple IS admissions, data from the first IS admission were extracted for non-cancer-IS patients while data from the IS admission within ±2 years of cancer diagnosis was extracted for cancer-IS patients. IS patients with cancer occurring before or after more than 2 years were excluded from this study.
Relevant Exposures
Age, sex, ethnicity (Chinese, Malay, Indian), and vascular risk factors (online suppl. eMethods 1) were obtained from the SSR. These included hypertension, diabetes mellitus, hyperlipidemia, transient ischemic attack, ischemic heart disease, atrial fibrillation/flutter, valvular heart disease, peripheral vascular disease, and smoking. These diagnoses were extracted from medical notes (general practitioner, emergency department, referral letters, hospital inpatient discharge records, and clinical notes of the index event). Smoking status, previous transient ischemic attack, known ischemic heart disease, valvular heart disease, or peripheral heart disease were fields with missing data (number of subjects with data available is indicated in Table 1). Cancer diagnosis date, primary site of malignancy, staging, histology, grade differentiation, and treatment data were obtained from the SCR. Staging in the SCR was in accordance to the American Joint Committee on Cancer Staging. Primary sites of malignancy were grouped into 15 main systems (online suppl. eMethods 3).
Demographics and vascular risk factors between cancer IS and non-cancer IS patients
Characteristicsa . | Whole cohort (n = 21,068) . | Cancer IS patients (n = 1,330) . | Non-cancer IS patients (n = 19,738) . | p valueb . |
---|---|---|---|---|
Demographics | ||||
Age, mean (SD), years | 67.9 (13.3) | 71.2 (11.6) | 67.7 (13.3) | <0.001 |
Age categories, years | ||||
≤65 | 9,068 (43.0) | 402 (30.2) | 8,666 (43.9) | <0.001 |
>65 | 12,000 (57.0) | 928 (69.8) | 11,072 (56.1) | |
Sex, male | 11,914 (56.6) | 776 (58.4) | 11,138 (56.4) | 0.172 |
Ethnicity | <0.001 | |||
Chinese | 16,304 (77.4) | 1,141 (85.8) | 15,163 (76.8) | |
Indian | 1,429 (6.8) | 55 (4.1) | 1,374 (7.0) | |
Malay | 3,335 (15.8) | 134 (10.1) | 3,201 (16.2) | |
Vascular risk factors | ||||
Hypertension | 17,305 (82.1) | 1,053 (79.2) | 16,252 (82.3) | 0.004 |
Diabetes mellitus | 9,584 (45.5) | 546 (41.1) | 9,038 (45.8) | <0.001 |
Hyperlipidemia | 18,907 (89.7) | 1,079 (81.1) | 17,828 (90.3) | <0.001 |
Atrial arrhythmia | 4,012 (19.0) | 260 (19.6) | 3,752 (19.0) | 0.627 |
Current or ex-smoker (n = 20,030) | 8,129 (40.6) | 563 (44.5) | 7,566 (40.3) | 0.003 |
Transient ischemic attack (n = 6,888) | 906 (13.2) | 61 (16.4) | 845 (13.0) | 0.057 |
Ischemic heart disease (n = 11,248) | 5,086 (45.2) | 324 (50.8) | 4,762 (44.9) | 0.004 |
Valvular heart disease (n = 5,423) | 456 (8.4) | 28 (10.2) | 428 (8.3) | 0.268 |
Peripheral vascular disease (n = 6,680) | 767 (11.5) | 46 (13.0) | 721 (11.4) | 0.370 |
Number of risk factorsc | <0.001 | |||
No risk factor | 335 (1.6) | 43 (3.2) | 292 (1.5) | |
1 risk factor | 1,767 (8.4) | 153 (11.5) | 1,614 (8.2) | |
2 risk factors | 5,680 (27.0) | 368 (27.7) | 5,312 (26.9) | |
≥3 risk factors | 13,286 (63.1) | 766 (57.6) | 12,520 (63.4) |
Characteristicsa . | Whole cohort (n = 21,068) . | Cancer IS patients (n = 1,330) . | Non-cancer IS patients (n = 19,738) . | p valueb . |
---|---|---|---|---|
Demographics | ||||
Age, mean (SD), years | 67.9 (13.3) | 71.2 (11.6) | 67.7 (13.3) | <0.001 |
Age categories, years | ||||
≤65 | 9,068 (43.0) | 402 (30.2) | 8,666 (43.9) | <0.001 |
>65 | 12,000 (57.0) | 928 (69.8) | 11,072 (56.1) | |
Sex, male | 11,914 (56.6) | 776 (58.4) | 11,138 (56.4) | 0.172 |
Ethnicity | <0.001 | |||
Chinese | 16,304 (77.4) | 1,141 (85.8) | 15,163 (76.8) | |
Indian | 1,429 (6.8) | 55 (4.1) | 1,374 (7.0) | |
Malay | 3,335 (15.8) | 134 (10.1) | 3,201 (16.2) | |
Vascular risk factors | ||||
Hypertension | 17,305 (82.1) | 1,053 (79.2) | 16,252 (82.3) | 0.004 |
Diabetes mellitus | 9,584 (45.5) | 546 (41.1) | 9,038 (45.8) | <0.001 |
Hyperlipidemia | 18,907 (89.7) | 1,079 (81.1) | 17,828 (90.3) | <0.001 |
Atrial arrhythmia | 4,012 (19.0) | 260 (19.6) | 3,752 (19.0) | 0.627 |
Current or ex-smoker (n = 20,030) | 8,129 (40.6) | 563 (44.5) | 7,566 (40.3) | 0.003 |
Transient ischemic attack (n = 6,888) | 906 (13.2) | 61 (16.4) | 845 (13.0) | 0.057 |
Ischemic heart disease (n = 11,248) | 5,086 (45.2) | 324 (50.8) | 4,762 (44.9) | 0.004 |
Valvular heart disease (n = 5,423) | 456 (8.4) | 28 (10.2) | 428 (8.3) | 0.268 |
Peripheral vascular disease (n = 6,680) | 767 (11.5) | 46 (13.0) | 721 (11.4) | 0.370 |
Number of risk factorsc | <0.001 | |||
No risk factor | 335 (1.6) | 43 (3.2) | 292 (1.5) | |
1 risk factor | 1,767 (8.4) | 153 (11.5) | 1,614 (8.2) | |
2 risk factors | 5,680 (27.0) | 368 (27.7) | 5,312 (26.9) | |
≥3 risk factors | 13,286 (63.1) | 766 (57.6) | 12,520 (63.4) |
IS, ischemic stroke; SD, standard deviation.
aAll data represented in n (%) unless otherwise specified.
bχ2 or Fisher’s exact test for categorical variables; two-sample t test for continuous variables.
cIncludes age, hypertension, diabetes mellitus, hyperlipidemia, atrial arrhythmia and ex/current smoking.
Statistical Analysis
Demographic and clinical variables were compared using χ2 or Fisher’s exact test for categorical variables and two-sided two-sample t test for continuous variables as appropriate. Cancer incidence rate was determined using the same direct age-standardization method with Segi’s world population as the 2015 SCR report [19]. Complete case multivariable logistic regression was performed to identify differences in the vascular risk profile of cancer-IS and non-cancer-IS patients. The inclusion criteria for multivariable analysis were p < 0.1 on univariate analysis and no missing data. Statistical significance was set at p ≤ 0.05. Analyses were performed using SAS software v9.4 for Windows (SAS, Inc, Cary, NC, USA).
Results
The study diagram detailing the derivation of the study cohort is summarized in Figure 1. Among 21,068 IS patients analyzed in this study, 11,914 (56.6%) were males and the mean age was 67.9 ± 13.3 years. The risk factor profile for the overall cohort is shown in Table 1. Cancer-IS patients were significantly older and had a higher proportion of Chinese compared to non-Chinese.
Consolidated Standards of Reporting Trials (CONSORT) diagram of stages of patient screening. IS indicates ischemic stroke.
Consolidated Standards of Reporting Trials (CONSORT) diagram of stages of patient screening. IS indicates ischemic stroke.
The prevalence of concomitant cancer was 6.3% with 935 (4.4%) had cancer diagnoses up to 2 years prior to IS onset while 395 (1.9%) were diagnosed with cancer in the 2 years following IS onset. Direct age-standardization using Segi’s world population [20] revealed the age-standardized cancer incidence rate in our cohort of IS patients to be 3,392.9 (95% confidence interval [CI], 1,937.1–4,848.7) per 100,000 person-years. The age-standardized cancer incidence rate of IS prior to a cancer diagnosis was 4,102.5 (95% CI, 4,101.6–4,103.4), while the age-standardized cancer incidence rate of IS after a cancer diagnosis was 1,810.0 (95% CI, 1,809.4–1,810.6).
In multivariable analyses, older age (odds ratio [OR], 1.02 [95% CI, 1.01–1.02] per year), males (OR, 1.25 [95% CI, 1.11–1.41), Chinese ethnicity (OR, 1.61 [95% CI, 1.37–1.89]), and a lack of hypertension (OR, 0.84 [95% CI, 0.73–0.97]), or hyperlipidemia (OR, 0.53 [95% CI, 0.45–0.62]) remained independent associations of cancer-IS (Table 2).
Summaries of association analysis of cancer IS and non-cancer IS patients using logistic regression model
Variables . | Univariate . | Multivariable . | ||
---|---|---|---|---|
crude OR (95% CI) . | p value . | adjusted ORa (95% CI) . | p value . | |
Age, years | 1.02 (1.02–1.03) | <0.001 | 1.02 (1.01–1.02) | <0.001 |
Male (vs. female) | 1.09 (0.96–1.20) | 0.174 | 1.25 (1.11–1.41) | 0.000 |
Chinese (vs. non-Chinese) | 1.82 (1.55–2.13) | <0.001 | 1.61 (1.37–1.89) | <0.001 |
Vascular risk factors | ||||
Hypertension | 0.81 (0.71–0.93) | 0.003 | 0.84 (0.73–0.97) | 0.018 |
Diabetes mellitus | 0.82 (0.74–0.92) | 0.001 | 0.98 (0.87–1.11) | 0.776 |
Hyperlipidemia | 0.46 (0.40–0.53) | <0.001 | 0.53 (0.45–0.62) | <0.001 |
Atrial arrhythmia | 1.04 (0.90–1.19) | 0.610 | … | … |
Current or ex-smoker | 1.19 (1.06–1.33) | 0.003 | … | … |
Transient ischemic attack | 1.32 (1.00–1.76) | 0.051 | … | … |
Ischemic heart disease | 1.27 (1.08–1.49) | 0.004 | … | … |
Valvular heart disease | 1.27 (0.85–1.90) | 0.236 | … | … |
Peripheral vascular disease | 1.17 (0.85–1.60) | 0.341 | … | … |
<3 risk factorsb | 1.28 (1.14, 1.43) | <0.001 | … | … |
Variables . | Univariate . | Multivariable . | ||
---|---|---|---|---|
crude OR (95% CI) . | p value . | adjusted ORa (95% CI) . | p value . | |
Age, years | 1.02 (1.02–1.03) | <0.001 | 1.02 (1.01–1.02) | <0.001 |
Male (vs. female) | 1.09 (0.96–1.20) | 0.174 | 1.25 (1.11–1.41) | 0.000 |
Chinese (vs. non-Chinese) | 1.82 (1.55–2.13) | <0.001 | 1.61 (1.37–1.89) | <0.001 |
Vascular risk factors | ||||
Hypertension | 0.81 (0.71–0.93) | 0.003 | 0.84 (0.73–0.97) | 0.018 |
Diabetes mellitus | 0.82 (0.74–0.92) | 0.001 | 0.98 (0.87–1.11) | 0.776 |
Hyperlipidemia | 0.46 (0.40–0.53) | <0.001 | 0.53 (0.45–0.62) | <0.001 |
Atrial arrhythmia | 1.04 (0.90–1.19) | 0.610 | … | … |
Current or ex-smoker | 1.19 (1.06–1.33) | 0.003 | … | … |
Transient ischemic attack | 1.32 (1.00–1.76) | 0.051 | … | … |
Ischemic heart disease | 1.27 (1.08–1.49) | 0.004 | … | … |
Valvular heart disease | 1.27 (0.85–1.90) | 0.236 | … | … |
Peripheral vascular disease | 1.17 (0.85–1.60) | 0.341 | … | … |
<3 risk factorsb | 1.28 (1.14, 1.43) | <0.001 | … | … |
Blank cells represent variables that are not included in multivariable analysis due to missing numbers or p > 0.1 in univariate analysis.
CI, confidence interval; IS, ischemic stroke; OR, odds ratio.
aAdjusted for age, sex, ethnicity, hypertension, diabetes mellitus, hyperlipidemia in the multivariable logistic regression analysis.
bIncludes age, hypertension, diabetes mellitus, hyperlipidemia, atrial arrhythmia and ex/current smoking.
The most prevalent cancers occurring in cancer-IS patients were lower gastrointestinal (19.7%), lung (13.8%), genitourinary (11.5%), hepatopancreatobiliary (10.8%), and upper gastrointestinal (7.8%), of which majority were stage III or IV (Table 3). In male cancer-IS patients, the top cancers were lower gastrointestinal (19.1%), genitourinary (18.0%), lung (16.2%), hepatopancreatobiliary (16.2%), and upper gastrointestinal (8.1%). In female cancer-IS patients, the top cancers were lower gastrointestinal (20.6%), breast (16.6%), lung (10.4%), gynecological (10.1%), and hepatopancreatobiliary (9.6%) (online suppl. eTable 1).
Baseline characteristics of overall, pre-stroke, and post-stroke cancer IS patients and the top 5 most common cancers
Characteristicsa . | Overall (n = 1,330) . | Cancer diagnosis prior IS onset (n = 935) . | Cancer diagnosis after IS onset (n = 395) . | |||
---|---|---|---|---|---|---|
Demographics | ||||||
Age, mean (SD), years | 71.2 (11.6) | 71.4 (11.6) | 70.69 (11.7) | |||
Age categories, years | ||||||
≤65 | 402 (30.2) | 277 (29.6) | 125 (31.6) | |||
>65 | 928 (69.8) | 658 (70.4) | 270 (68.3) | |||
Sex, male | 776 (58.3) | 523 (55.9) | 253 (64.1) | |||
Ethnicity | ||||||
Chinese | 1,141 (85.8) | 807 (86.3) | 334 (84.6) | |||
Indian | 55 (4.1) | 38 (4.1) | 17 (4.3) | |||
Malay | 134 (10.1) | 90 (9.6) | 44 (11.1) | |||
Cancer treatment prior to stroke | ||||||
No treatment | 266 (28.4) | NA | ||||
Surgery | 252 (27.0) | NA | ||||
Chemotherapy | 69 (7.4) | NA | ||||
Radiotherapy | 114 (12.2) | NA | ||||
Othersb | 109 (11.7) | NA | ||||
Combinationc | 125 (13.4) | NA | ||||
Top cancers | ||||||
1st | Lower GI | 262 (19.7) | Lower GI | 196 (21.0) | Lower GI | 66 (16.7) |
2nd | Lung | 184 (13.8) | Lung | 119 (12.7) | Lung | 65 (16.5) |
3rd | Genitourinary | 153 (11.5) | HPB | 96 (10.3) | Genitourinary | 58 (14.7) |
4th | HPB | 144 (10.8) | Genitourinary | 95 (10.2) | HPB | 48 (12.2) |
5th | Upper GI | 104 (7.8) | Upper GI | 76 (8.1) | Upper GI | 28 (7.1) |
Others | 483 (36.3) | Others | 353 (37.8) | Others | 130 (32.9) |
Characteristicsa . | Overall (n = 1,330) . | Cancer diagnosis prior IS onset (n = 935) . | Cancer diagnosis after IS onset (n = 395) . | |||
---|---|---|---|---|---|---|
Demographics | ||||||
Age, mean (SD), years | 71.2 (11.6) | 71.4 (11.6) | 70.69 (11.7) | |||
Age categories, years | ||||||
≤65 | 402 (30.2) | 277 (29.6) | 125 (31.6) | |||
>65 | 928 (69.8) | 658 (70.4) | 270 (68.3) | |||
Sex, male | 776 (58.3) | 523 (55.9) | 253 (64.1) | |||
Ethnicity | ||||||
Chinese | 1,141 (85.8) | 807 (86.3) | 334 (84.6) | |||
Indian | 55 (4.1) | 38 (4.1) | 17 (4.3) | |||
Malay | 134 (10.1) | 90 (9.6) | 44 (11.1) | |||
Cancer treatment prior to stroke | ||||||
No treatment | 266 (28.4) | NA | ||||
Surgery | 252 (27.0) | NA | ||||
Chemotherapy | 69 (7.4) | NA | ||||
Radiotherapy | 114 (12.2) | NA | ||||
Othersb | 109 (11.7) | NA | ||||
Combinationc | 125 (13.4) | NA | ||||
Top cancers | ||||||
1st | Lower GI | 262 (19.7) | Lower GI | 196 (21.0) | Lower GI | 66 (16.7) |
2nd | Lung | 184 (13.8) | Lung | 119 (12.7) | Lung | 65 (16.5) |
3rd | Genitourinary | 153 (11.5) | HPB | 96 (10.3) | Genitourinary | 58 (14.7) |
4th | HPB | 144 (10.8) | Genitourinary | 95 (10.2) | HPB | 48 (12.2) |
5th | Upper GI | 104 (7.8) | Upper GI | 76 (8.1) | Upper GI | 28 (7.1) |
Others | 483 (36.3) | Others | 353 (37.8) | Others | 130 (32.9) |
GI, gastrointestinal; HPB, hepatopancreatobiliary; IS, ischemic stroke; NA, not applicable.
aAll data represented in n (%) unless otherwise specified.
bIncludes hormonal, biological, and unspecified therapy.
cIncludes any combination of surgery, chemotherapy, and radiotherapy.
Discussion
In this nationwide registry study of cancer-related IS in ethnic Asians, the age-standardized cancer incidence rate in IS patients was 3,392.9 per 100,000 person-years, about 15 times that reported in the general population (219.4–230.9 per 100,000 person-years) [19], supporting the association of IS with cancer. This higher stroke risk in cancer patients was also reflected in a recent meta-analysis of 20 cohort studies which reported that cancer survivors are 1.66 times more likely to sustain a stroke compared to non-cancer controls [21].
Concomitant cancer was found in 6.3% of the IS patients in our study. This was similar to the cancer prevalence of 7.5% found in the NAVIGATE ESUS trial despite only including a specific subset of IS patients with embolic stroke of undetermined source [22]. Data from the Bergen NORSTROKE study also showed a comparable prevalence of active cancer in 5% of IS patients, albeit within a shorter period of 1 year within stroke onset [10]. Of the cancer-IS patients, majority suffered a stroke subsequent to cancer diagnosis while an estimated one-third of them presented with a stroke prior to cancer diagnosis. Overall, 1.9% of IS patients were diagnosed with cancer in the 2 years following the index stroke. This was less than half of the 4.5% cancer prevalence reported in Lindvig et al. [3], the largest prospective study to date, despite sharing comparable follow-up durations. This disparity may be in part due to their inclusion of additional patients with transient ischemic attack, non-IS, and patients with brain tumors. Conversely, recent meta-analyses have also suggested that the incidence of IS after a new cancer diagnosis is approximately 1.4% [8].
Cancer-IS patients were found to be older, more likely to be male, and had a significantly lower prevalence of vascular risk factors compared to non-cancer-IS patients. Specifically, a lower prevalence of hypertension and hyperlipidemia were independently associated with cancer-IS patients. Our findings replicated that found in smaller Western [23] and Asian studies [24], however, were contrary to those from other larger scale studies.
Zhang et al. [21] found that cancer tends to promote development of stroke in females and a younger age group (age of cancer diagnosis <20 or 45) in their meta-analysis while other large studies failed to demonstrate differences in vascular risk profile [2, 9]. These differences are likely to due to patient selection such as inclusion of a younger population and hemorrhagic stroke patients in the meta-analysis of Zhang et al. [21].
In our study, the relative lack of conventional risk factors for stroke in cancer-IS patients suggests that additional cancer-specific mechanisms such as hypercoagulability, non-bacterial thrombotic endocarditis, and treatment complications [25] may contribute to stroke pathogenesis in cancer-IS patients. While this suggests the possibility of cancer screening in IS patients without hypertension or hyperlipidemia, none of the variables alone are suitable to identify IS patients at high risk of cancer as the differences in vascular risk factors between the two groups were not sufficiently marked. Smoking was the only vascular risk factor that was significantly higher in cancer-IS patients. This was expected given that smoking is a risk factor shared by cancer, especially those pertaining to the lung, pancreas, and bladder [26] which were amongst the top cancers found in this study.
The top 5 cancers identified in this study were lower gastrointestinal followed by lung, genitourinary, hepatopancreatobiliary, and upper gastrointestinal, all of which were top cancers in cancer-associated stroke literature [6], including the meta-analysis of Zhang et al. [21], with the exception of hematologic and head and neck cancers. Our cohort most closely replicates the Japanese cohort of Uemura et al. [6]. Gastrointestinal, lung, and pancreatic cancers have consistently been ranked as top cancers across cancer-IS studies [4], and this is in line with findings on them being cancers with the highest incidence of thromboembolic events [27]. In addition, majority of these top cancers were found to be stage III and beyond in this study and studies have shown that cancers at advanced stages tend to be associated with a higher stroke risk [27]. Theorized links between malignancies and pro-thrombotic states include activation of endothelium, production of cancer pro-coagulants, inflammatory cytokines, and cancer therapy which perturb the balance in the Virchow’s triad, leading to thrombosis.
There have been studies which investigated laboratory and imaging parameters in addition to vascular risk factors to assess cancer risk in IS patients. Selvik et al. [10] proposed a clinical score of elevated D-dimer, lower hemoglobin, and previous or current smoking for predicting active cancer in IS patients <75 years old while Kassubek et al. [9] developed a model comprising of number of stroke territories, granulocyte percentage, C-reactive protein and serum lactate dehydrogenase activity to identify IS patients with occult solid malignancies. This study was limited by registry data which did not include laboratory and imaging variables. There is currently no basis for cancer screening guidelines in a specific profile of IS patients, and further research involving laboratory and imaging parameters is needed.
The strength of this study lies in it being a large multiethnic study on cancer-related IS and the utilization of reported registry data which are nationally representative. However, this study has limitations. The study is a cross-sectional study hence causality cannot be determined. There were missing data for certain vascular risk factors parameters which precluded them from being controlled for in the multivariate analysis. Reporting of vascular risk factors to the registry by clinicians was also not standardized. In addition, D-dimer levels, imaging parameters, IS etiology subtype, alcohol intake, and cellular typing of cancers that were unobtainable from the registries may have provided further insights. The effect of cancer treatment on IS risk was also not analyzed. Given that the stroke registry data used was from 2008 to 2012, the results of this study may not be generalizable to current data given the increase in both IS and cancer rates in Singapore from 2008 to present.
Conclusions
In this large multiethnic population, the burden of cancer-IS amounted to 6.3% and there is evidence of a 15-times increase in age-standardized cancer incidence in IS patients, supporting the association between stroke and cancer. Lower gastrointestinal cancer was the top cancer involved in cancer-related IS. Older age and a lower prevalence of vascular risk factors were independently associated with IS patients with concomitant cancer.
Acknowledgments
The authors would also like to thank the Singapore Cancer and Stroke Registries Advisory Committee Members, as well as staff from the National Registry of Diseases Office, Singapore for providing anonymized data from the Singapore Cancer and Stroke Registries.
Statement of Ethics
This study was approved by the SingHealth Centralized Institutional Review Board (approval no. 2017/2450) and was granted waiver of informed consent as this study utilized deidentified data.
Conflict of Interest Statement
The authors have no conflicts of interest to declare.
Funding Sources
This study was funded by AM-ETHOS Medical Student Fellowship 2018 (AM-ETHOS01/FY2017/17-A17).
Author Contributions
Kendra Jing Ying Tang: conceptualization (supporting), methodology (supporting), writing – original draft (lead), and writing – review and editing (equal). Seyed Ehsan Saffari: formal analysis (lead). Kaavya Narasimhalu: writing – review and editing (supporting). Kian Kheng Queck: conceptualization (equal) and methodology (supporting). Deidre Anne De Silva: conceptualization (lead), methodology (lead), writing – review and editing (equal), and supervision (lead).
Data Availability Statement
Data used in this analysis is kept by the National Registry of Diseases Office (NRDO) and restricted due to potentially identifiable participant information. These data cannot be made publicly available but qualified investigators who have been granted access by the NRDO can obtain deidentified data on request to [email protected]. Further inquiries can be directed to the corresponding author.