PDF
Citation
Print
INTRODUCTION
With total population of 51 million and average life expectancy of 82 years, Statics Korea reported 22,745 deaths were due to stroke and 78,863 deaths were due to cancer in South Korea in 2017 [1]. Cancer and stroke are among the three most common causes of death, both of which lead to enormous health and economic burdens.
Although cancer and stroke have distinct etiologies, correlation between them are being studied. In an autopsy study from 1985, ischemic stroke (IS) was the second most common brain lesion in cancer patients and 14.6% of cancer patients had pathologic evidence of cerebrovascular disease [2]. Many studies have clarified the relationship between cancer and IS but the specific mechanism and disease characteristics are still elusive.
In 1865, Trousseau described the clinical association between idiopathic venous thromboembolism (VTE) and malignancy [3]. In a previous study, the risk for thromboembolism was up to sevenfold higher in the cancer patients [4]. And also, thromboembolism has been proven to be one of the complications of anticancer treatment, such as chemotherapy. This thrombophilia in cancer patients seems to play an important role in the development of cancer-related stroke. It has been demonstrated that the high risk of VTE in cancer patients also increases the risk of arterial thrombosis, such as IS [5]. The adenocarcinoma, which often is the most common histologic type in stroke with underlying cancer, was also associated with the increased risk of thromboembolism [6].
The neuroimaging characteristics of stroke can provide important clues to a physician. Many studies have demonstrated its role in estimating the clinical outcomes and the etiology of stroke on top of diagnosing stroke itself. Neuroimaging features of stroke with underlying cancer has been described in previous studies and multiple vascular territory infarction pattern was often more commonly encountered in cancer related stroke [78].
The mortality outcome seems to be more severe in stroke patients with underlying cancer: the median survival was 4.5 months and one-fourth of the patients died within one month after stroke [9]. In an observational study, 1,542 (12.3%) of the total 12,565 patients had died within 1 year after hospitalization for the acute stroke [10].
However, the diagnostic, therapeutic consensus and the prognostic factors of cancer-related stroke has not been set yet. This study is aimed to investigate the incidence, mortality and characteristics of IS patients with underlying cancer.
MATERIALS AND METHODS
We retrospectively reviewed the medical records (clinical, pathologic and radiographic) of a single medical center between January 2012 and December 2016. The 40,047 patients were diagnosed with cancer and 150 patients were diagnosed with both IS and cancer diagnosis. Following patients were excluded: 1) patients who were diagnosed with IS prior to cancer diagnosis, 2) brain tumor patients, and 3) patients with omitted initial assessment records. Among the 150 patients, 63 patients were enrolled in the study: 87 were excluded due to the above reasons. The types of cancer that led to IS, cerebrovascular risk factors (diabetes mellitus, hypertension, dyslipidemia, arrhythmia, alcohol and smoking history), cancer-related information (types of therapy, presence of systemic metastasis, stage and histologic diagnosis), infarction patterns in brain magnetic resonance image (MRI), initial stroke severity using National Institute of Health Stroke Scale (NIHSS) and initial functional assessments using modified Rankin Scale (mRS) were reviewed in the enrolled patients. The 6-month survival status was reviewed in 52 patients whose medical records were available.
The initial brain MRI and diffusion weighted image were reviewed for infarction lesion pattern. The infarction patterns were classified as followed: 1) single lesion in one vascular territory, 2) scattered lesion in one vascular territory, 3) borderzone infarction, and 4) infarcts involving multiple vascular territories of the brain [11]. In order to rule out artery-to-artery embolism, scattered lesions in the same vascular territory were separately categorized from scattered lesions in the multiple vascular territory [12]. Multi-territorial infarction was defined as multiple acute infarctions either in bilateral vascular territory or in anterior and posterior circulations [13]. A retrospective medical record review at our institution was conducted after approval from the Institutional Review Board (IRB) for conducting research and registered under number 2019AN0184.
Statistical analysis
Categorical variables between two groups were compared using χ2 or Fisher's exact test. The Mann-Whitney U test was used to analyze the differences between two independent groups with non-parametric variables. And the Student's t-test for parametric variables. Six-month survival status was analyzed with Kaplan-Meier survival analysis using the log-rank test and forward conditional (variable entry and removal with p < 0.05) Cox proportional hazards multi-variate regression models were used to calculate the hazard ratios (HRs) and 95% confidence intervals (CI). Statistical Package for the Social Sciences (SPSS) version 24.0 (IBM, Chicago, IL, USA) was used for the statistical analysis. A p value of 0.05 or less was considered statistically significant.
RESULTS
General characteristics of patients
A total of 63 patients were enrolled with diagnosis of IS with underlying malignancy. This accounts for 0.16% of total cancer patients from our center during the study period. The demographics, comorbidities and clinical characteristics of the cancer patients with IS are given in Table 1. The mean age was 74.9 years (± 9.2) and 33 patients (52.4%) were men. More than half of the patients (56.0%) had systemic metastasis and stage 4 cancer was most common (57.1%). The 32 patients (50.8%) had received chemotherapy prior to stroke and two-thirds of the patients were treated with the platinum-based agent as the latest agent before stroke. The neuroradiologic findings revealed 29 patients (44.6%) with single territory infarction, 7 patients (10.8%) with borderzone infarction and 6 patients (9.5%) for scattered territorial and 21 patients (32.3%) with multi-territorial infarction. The median initial NIHSS score was 3 (0–24), and the median initial mRS score was 3 (0–5).
Table 1
Clinical findings in cancer patients with ischemic stroke
Clinical characteristics of cancer and stroke
The three most frequent types of the primary cancer in the enrolled patients were lung (27.0%), gastric (14.3%), colorectal (12.7%) cancer followed by hepatobiliary (12.7%), genitourinary (11.1%), pancreatic (6.3%) and breast (1.6%) cancer. Most patients had solid tumors in the enrolled group. The types of cancer in total cancer patients from our center during the study period were in the order of thyroid (14.6%), colorectal (11.2%), gastric (9.0%), followed by breast (7.0%), liver (5.1%) and lung (4.2%) (Fig. 1).
Fig. 1
(A) Cancer types in the enrolled group (n = 63), (B) Cancer types in total cancer patients (n = 40,047), (C) Histological subtypes of the cancer in the enrolled group (n = 63).
Histologically, adenocarcinoma was the most common type with 47 patients (74.6%) followed by hepatocellular carcinoma, small cell lung cancer and lymphoma. Among the adenocarcinoma histology lung, gastric, biliary duct and pancreas and colorectal were the most common cancer types (Fig. 1). We compared the variables between adenocarcinoma or non-adenocarcinoma group and the results were shown in Table 1. The adenocarcinoma group showed lower rates of stroke related comorbidities (dyslipidemia and smoking history, p < 0.05) and multi-territorial infarctions were more common (p < 0.05). The D-dimer levels were studied in 59 patients among the 63 enrolled patients and 4 patients (3 patients in adenocarcinoma, 1 patient in non-adenocarcinoma group) were not evaluated with D-dimer as part of their acute stroke evaluation. In an analysis of 59 patients, the average D-dimer level in adenocarcinoma group was 5.51 ± 6.54 mg/L, whereas the non-adenocarcinoma group was 1.75 ± 1.64 mg/L (p < 0.05). No significant differences were noted in initial stroke severity measured by NIHSS and mRS.
Six-month survival rate of patients
Among the enrolled cohort, 52 out of 63 patients were able to be reviewed with survival status at 6-month after stroke. Eighteen patients (34.6%) died before 6-month post stroke and most common cancer type was lung cancer (52.9%). Systemic metastasis and multi-territorial infarction were significantly more common in the mortality group (p < 0.001). And also, the initial NIHSS (p < 0.05) and mRS scores were worse in mortality group (p < 0.05) (Table 2). Adenocarcinoma histology showed some relevance with 6-month mortality but did not meet the statistical significance (Table 2). The common causes of death in 6-month mortality group were infection (33.4%), cancer related (22.2%) and cardiorespiratory arrest (22.2%) (Table 3). Survival probability was significantly lower in patients with multi-territorial infarction (p < 0.001) and systemic metastasis (p < 0.001) (Fig. 2). In addition, systemic metastasis (adjusted HR, 6.28), multi-territorial infarction (adjusted HR, 4.35) and initial NIHSS score (adjusted HR, 1.09) were significantly associated with 6-month mortality by multivariate analysis (Table 4).
Table 2
Survival status at 6-month post-stroke in cancer patients with ischemic stroke
Table 3
Causes of mortality in 6-month mortality group
| Causes of death | No. (%) | |
|---|---|---|
| Infection | ||
| Pneumonia | 5 (27.8) | |
| Sepsis (gastrointestinal) | 1 (5.6) | |
| Cancer related | 4 (22.2) | |
| Cardiorespiratory arrest | 4 (22.2) | |
| Stoke related | 3 (16.7) | |
| Renal failure | 1 (5.6) | |
| Total | 18 | |
Fig. 2
Kaplan-Meier survival curves stratified by (A) multi-territorial infarction and (B) systemic metastasis within 6-months after stroke (Log-rank test p < 0.001 for both).
Table 4
HRs for 6-month mortality after stroke diagnosis.
Age, sex, systemic metastasis, adenocarcinoma histology, stroke risk factors, NIHSS, mRS and involvement of multiple vascular territories were adjusted for in the multivariate model using a forward conditional method.
NIHSS, National Institutes of Health Stroke Scale; HR, hazard ratio; CI, confidence interval; mRS, modified Rankin Scale.
DISCUSSION
Cancer-related stroke is known to differ from stroke without cancer in many aspects. Understanding its etiology and characteristics is important for prevention and effective management. Our study showed different distributions of common cancer types between cancer-related stroke and general cancer groups. In addition, adenocarcinoma patients showed less relevance with conventional stroke risk factors and the multi-territorial infarction patterns were more common in comparison with non-adenocarcinoma patients. The mortality outcome was poor and different features of stroke and underlying cancer influenced the 6-month mortality.
The three most frequent types of the primary cancer in the enrolled group were lung (27.0%), gastric (14.3%) and colorectal (12.7%). However, in total cancer patients, thyroid was the most common cancer type and lung cancer only accounted for 4%. That thyroid cancer is the most common type is not surprising given its highest incidence rate in the nationwide cancer statistics in 2015 [14]. This discrepancy in common cancer types between the cancer related stroke and total cancer group is consistent with previous studies and this may support the idea that certain types of cancer have stronger correlations with stroke [15]. In our study, lung, gastric, bile duct and pancreas and colorectal cancer, which are the common cancer types in enrolled group, were mostly comprised of adenocarcinoma histology (Fig. 1C). And the discrepancy in common cancer types may have resulted from the relationship between adenocarcinoma and IS that has been delineated in the previous studies.
Although still in controversy, the etiology of IS in cancer patients seems to be different from that of stroke without cancer. A multicenter study in Korea reported that nearly 40% of the cancer patients with IS were assigned to the cryptogenic origin, whereas only 18% were cryptogenic in conventional stroke group [16]. In addition, the cancer related cryptogenic stroke patients displayed different clinical features from non-cancer patients with fewer atherosclerotic risk factors, coagulation abnormalities such as D-dimer levels, and low nutritional status [17]. Mucin-producing malignancies (e.g., adenocarcinoma) is one of the cancer-related conditions that leads to hypercoagulability and eventually raises the risk of thromboembolism [18]. Despite of missing data of 4 patients, D-dimer levels which represents the measure of activated coagulation were relatively high among the adenocarcinoma group in our study. And the conventional stroke risk factors such a dyslipidemia and smoking were less frequent in adenocarcinoma group and hypertension also showed tendency to be less frequent. As in most studies, there is a direct relationship between cholesterol levels and IS [19]. And a strong dose-response relationship was found between smoking and the risk of stroke, even in young population [20]. Considering this tendency of elevated D-dimer level and having less conventional atherosclerotic risk factors, it may leads us to consider that mucin producing malignancies such as adenocarcinoma could influence the cancer related mechanisms in development of stroke.
Our results showed that multi-territorial infarction pattern was more common in adenocarcinoma patients. In general, multi-territorial infarctions are known to have embolic etiologies. In the case reports on cervical cancer, lung cancer and cholangiocarcinoma patients with episodes of multi-territorial infarction, cardio-embolic work-up was non-remarkable and it was suggested that cancer related thromboembolic events are likely to have caused the deteriorating event [212223]. And cancer related hypercoagulability accounted for 75% of multi-territorial infarction cases when the identifiable embolic source was absent [8]. This further supports the idea that adenocarcinoma is associated with thromboembolism and implicates that the mechanism of the cancer-related stroke is different from that of conventional stroke.
Although stroke mortality is gradually declining, it still remains high and it is the third major cause of death in South Korea [24]. In the large cohort study on IS, 78% of the patients were still alive during the 6-month period after the stroke diagnosis [25]. In comparison with this result, the 6-month mortality rate was higher (34.6%) in our study and it supports the previous literatures describing the poorer prognosis and higher mortality with concurrence of stroke and cancer [11]. It has been described in a recent retrospective study that metastasis and cancer types were significantly associated with 6-month post stroke mortality [26]. The 6-months period was specifically determined based on American Heart Association/American Stroke Association guidelines that recommend the treatment of stroke with malignancy only for patients with the life expectancy of 6-months or longer [27]. In our study, systemic metastasis, multi-territorial infarction and initial stroke severity were the risk factors for 6-month mortality after stroke. Systemic metastasis places cancer in stage 4, which may lead to low performance status and underlying disability before the diagnosis of stroke. And the patients with pre-existing disability had higher chances of unfavorable outcome in stroke and also high risk of pneumonia-related mortality [2829]. Most of the mortalities in our study were due to non-neurological causes such as infection (33.4%) followed by cancer related and cardiorespiratory causes. In a large cohort study in Japan with IS patients, cerebrovascular was the overall leading cause of death, while pneumonia was the leading cause of death in more disabled sub-group [30]. In line with pneumonia being the most common cause of death in our group, these findings may indicate that widespread advanced cancer status might have influenced mortality. But, since stroke related features such as multi-territorial infarction and stroke severity showed independent correlations with mortality outcomes [25], future studies are warranted with larger cohort to elaborate on the prognostic factors of stroke mortality outcomes.
There are some limitations of this study. First, this study was based on retrospective analysis of clinical data, thus several clinical and functional outcomes data were missing in many patients. Secondarily, the patient data was retrieved from a single medical center and the total number of subjects was small. Therefore, subgroup analysis between cancer and non-cancer group and between the cancer types in adenocarcinoma histology was limited. And the potential confounding factors from cancer itself and the characteristics of each cancer type on clinical characteristics of stroke were not taken into consideration. In the future, a multi-center based and well-controlled cohort with a larger patient group is required.
In conclusion, frequent cancer types were different between stroke with cancer and total cancer group and lung cancer was the most common type in the former group. Adenocarcinoma was the most common histologic type and the adenocarcinoma group was associated with having less conventional stroke risk factors and multi-territorial infarction patterns. Six-month mortality outcomes were poor, which was influenced by systemic metastasis, multi-territorial infarction and initial neurological severity.
XML Download