Predictors of Occult Cancer in Acute Ischemic Stroke Patients Dolores Cocho, MD, PhD, Jordi Gendre, MD, Anuncia Boltes, MD, Jordi Espinosa, MD, Ana C. Ricciardi, MD, Jordi Pons, MD, Marta Jimenez, MD, and Pilar Otermin, MD
Background: We investigated the prevalence of occult malignancy (OM) in acute ischemic stroke patients to evaluate if any biological marker could help to detect the presence of OM. Methods: We retrospectively reviewed all ischemic stroke patients during 48 months. We did not perform any screening for OM. Demographic data, vascular risk factors, routine blood chemistry with fibrinogen and C-reactive protein (CRP), National Institutes of Health Stroke Scale (NIHSS), and etiological subtype of stroke according to Trial of Org 10172 in Acute Stroke Treatment criteria were analyzed. The patients were divided into 2 groups (Non-OM versus OM). Results: We analyzed 631 patients with acute ischemic stroke. The mean age was 69.7 6 12.7 years, and 59% were men. The distribution of vascular risk factors, etiological subgroups, and NIHSS was comparable between both groups. We detected 13 cases (2.1%) with OM, and this percentage was higher in patients with stroke of undetermined etiology (5.3%). We detected significant higher levels of fibrinogen and CRP in patients with stroke of undetermined cause with OM. Receiver operating characteristic curves showed a sensitivity of 75% and specificity of 96% for levels of CRP more than 20 mg/L, and a sensitivity of 67% and specificity of 91% for fibrinogen levels greater than 600 mg/dL. Conclusions: OM was present in 2.1 % of overall patients, and 5.3% of patients with stroke of undetermined cause. Baseline levels of fibrinogen more than 600 mg/dL or CRP greater than 20 mg/L in patients with undetermined stroke might be good predictors of OM. Key Words: Stroke—cancer—occult malignancy—ischemic stroke. Ó 2015 by National Stroke Association
Introduction From the Department of Neurology, Hospital General de Granollers, Universitat Internacional de Catalunya, Barcelona, Spain. Received August 22, 2014; revision received February 1, 2015; accepted February 6, 2015. The authors report no conflicts of interest and nonfinancial disclosures. D.C. carried out the review of the literature, conception of the study, study design, collection and data analysis, writing of the article, and critical revision for intellectual content. J.G. performed the review of the literature, writing of the article, and critical revision for intellectual content. A.B., J.E., A.C.R., J.P., M.J., and P.O. carried out the data analysis and critical revision for intellectual content. Address correspondence to Dolores Cocho, MD, PhD, Unidad de Neurologıa, Servicio de Medicina Interna, Hospital General de Granollers, AV/ Francesc Ribas sn., Barcelona 08402, Spain. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2015 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2015.02.006
Cancer is the main acquired prothrombotic state, and the relationship between cancer and cerebrovascular disease has been documented in patients with a wide variety of tumors.1-3 Previous studies have described 5 etiologic categories of stroke events in patients with cancer: direct tumor effects (tumor seeding with embolism or mucin-secreting tumors), coagulation disorders, infections, complications of therapeutic or diagnostic procedures,1,4 and paraneoplastic causes.4 Cerebrovascular disease occurs in 15% of cancer patients based on autopsy studies, but only about half of them have experienced clinical symptoms of stroke.1 Occasionally stroke is the first manifestation of an occult malignancy (OM) (.4%-3%),5-7 and the identification and treatment of these patients could improve their prognosis. Although patients with systemic cancer usually have poor outcomes, their survival rate is increasing with the
Journal of Stroke and Cerebrovascular Diseases, Vol. -, No. - (---), 2015: pp 1-5
1
D. COCHO ET AL.
2
development of more effective cancer treatments. Early identification of stroke mechanisms may be important in cancer patients because these mechanisms may differ from those in stroke patients without cancer.8 However, the mechanisms underlying stroke in patients with cancer are controversial.9 Our aim was to analyze the prevalence of OM in ischemic stroke patients and its different mechanisms and to evaluate if any usual biological marker could help to detect the presence of OM in stroke patients.
Methods We retrospectively studied patients with ischemic stroke in our Stoke Unit from 2009 to 2012. Medical history, physical examination, routine blood chemistry, electrocardiogram (ECG), chest X-ray, cranial neuroimaging with computed tomography scan or magnetic resonance imaging, arterial study (carotid ultrasound, transcranial Doppler, or magnetic resonance angiography), and transthoracic echocardiography were systematically performed in all patients. Other studies, such as Holter-ECG monitoring or transesophageal echocardiogram, were performed according to clinical criteria. The baseline vascular risk factors were defined as follows: Smoking was considered if the patient reported smoking cigarettes. Hypertension was defined as a systolic blood pressure of 140 mm Hg or greater or diastolic pressure of 90 mm Hg or greater or current use of antihypertensive treatment. Diabetes mellitus was considered in patients with a history of fasting blood glucose of 126 mg/dL or more or current use of antidiabetic drugs. Hypercholesterolemia was defined as a total cholesterol level of 220 mg/dL or more or current use of lipidlowering agents. Stoke were classified etiologically according to Trial of Org 10172 in Acute Stroke Treatment (TOAST) criteria10 as because of large-artery occlusive disease (LA), smallvessel disease (SV), cardioembolism (CE), and other causes or undetermined cause (UND).
Transient ischemic attack was defined as acute onset of focal cerebral or monocular symptoms lasting less than 24 hours and of presumptive vascular origin. In the cases where an OM was suspected by findings in usual physical examination (eg, cervical adenopathies, tumoral oral lesions, or abdominal mass) or in the routine complementary studies (cervical adenopathy in carotid ultrasound, hematuria in urine, or X-ray lung mass), other complementary examinations were performed according to clinical criteria. All occult malignancies were confirmed by biopsy and cytologic study (Table 1). We do not systematically perform any screening for OM.
Statistical Analysis All patients were divided into 2 groups: No OM group (Non-OM) and the newly diagnosed malignancy (OM). All analyses were carried out using the SPSS 19.0 (IBM, Armonk, NY) software package. Categorical variables were compared between the 2 groups using the c2 or by the Fisher exact test. A student t test was used to evaluate continuous variables. A multivariate analysis was performed to detect differences between the group OM and non-OM. If there were significant differences in continuous variables between both groups, the optimal cutoff points of each variable to discriminate the groups were obtained by receiver operating characteristic curves. We considered statistically significant P less than .05.
Results We retrospectively reviewed 631 patients who had suffered a cerebrovascular ischemic event and were hospitalized in our Stroke Unit during 4 consecutive years. Of these events, 19% were transient ischemic attacks and 81% cerebral infarcts. The mean age was 69.7 6 12.7 years, and 59% were men. The median National Institutes of Health Stroke Scale (NIHSS) on hospital admission was 4 points (range: 0-38). The average hospital stay
Table 1. Pathologic findings in the physical examination or in the routine complementary examinations in OM patients Type of malignancy
Abnormal physical examination
Abnormal routine complementary examinations
2 Lung carcinoma 2 Lymphomas 1 Colon carcinoma 1 Colon carcinoma 1 Chronic lymphatic leukemia 1 Breast carcinoma 1 Pancreas carcinoma 1 Uterine carcinoma 1 Multiple myeloma 1 Mouth carcinoma 1 Peritoneal mesothelioma
No No Intestinal occlusion Pain and abdominal mass Cervical plexopathy symptoms Breast mass
Chest X-ray Cervical adenopathies in carotid duplex
Abbreviation: OM, occult malignancy.
Cholestasis in laboratory test Abdominal pain and leukorrhea Bicytopenia in laboratory test Abnormal mouth examination Abdominal pain and ascites
PREDICTORS OF OCCULT CANCER IN STROKE PATIENTS
was 9 days. According to Oxfordshire classification, the strokes could be divided into total anterior circulation infarcts (20%), partial anterior circulation infarcts (32%), lacunar infarcts (30%), and posterior circulation infarcts (18%). According to TOAST criteria, the strokes were classified as follows: LA (14%), SV (19%), CE (31%), other causes (7%), and UND (28%). In the remaining 1% of cases, the stroke was classified also as of UND because of 2 possible etiologies. Medical history, physical examination, complete blood count, coagulation tests, routine blood biochemistry including determination of vitamin B12, folic acid, C-reactive protein (CRP), and fibrinogen, ECG, chest X-ray, and cerebral computed tomography scan were systematically performed. Extracranial and intracranial arterial study was performed in all patients with the following studies; carotid duplex ultrasonography (66%), transcranial Doppler (64%), or magnetic resonance angiogram magnetic resonance angiography (70%). Other performed studies were transthoracic echocardiography (89%) and Holter-ECG monitoring (21%). Vascular risk factors were present in 80% of cases. The distribution was as follows: hypertension (75%), diabetes mellitus (35%), dyslipidemia (54%), atrial fibrillation (23%), alcohol abuse of more than 40 g/day (13%), smoking (22%), drug abuse (1.2%), ischemic heart disease (16%), and previous cerebrovascular disease (20.6%). The more frequent medical complications during admission were dysphagia (19%) and aspiration pneumonia (3.6%). The mortality during the hospitalization was 3.8%. The main reasons of mortality were neurologic (57%), respiratory (28%), cardiac (5%), and others (10%). Thrombolysis with recombinant tissue plasminogen activator was administered in 13.2% of cases. The patients were discharged with antiplatelet drugs in 65.8%, anticoagulants in 34%, antiplatelet associated to anticoagulant drug .2%, statins in 83%, and antihypertensive drugs in 90%. An OM was detected in 13 patients (2.1%): 2 lung carcinomas, 2 colon carcinomas, 2 lymphomas, 1 chronic lymphatic leukemia, 1 breast carcinoma, 1 pancreas carcinoma, 1 uterine carcinoma, 1 multiple myeloma, 1 floor of the mouth carcinoma, and 1 peritoneal mesothelioma. Of the 13 cases with OM, 4 (30%) died with a mean survival time of 15 days. The remaining 9 patients are currently with mean survival time of 29 months. Comparing the OM group (n 5 13) with the non-OM group (n 5 618) there were no significant statistical differences with respect to demographic data, neurologic severity measured with the NIHSS, clinical subtype according to Oxfordshire classification, incidence of vascular risk factors, percentage of dysphagia, aspiration pneumonia, or mortality during the hospital admission. However, we found significant differences in the etiological subtypes according to TOAST criteria. Most patients with OM had a stroke with UND (61.5% versus 24%, P 5 .003; Table 2)
3
Table 2. Results comparing the OM with non-OM patients Non-OM group OM group (n 5 13) (n 5 618) TOAST % (n) Large-artery occlusive disease Cardioembolic Small-vessel disease Other causes Undetermined cause Undetermined by 2 causes
P .003
15.4 (2)
15 (92)
15.4 (2) 7.7 (1) 0 61.5 (8) 0
33 (204) 20 (124) 7 (44) 24 (148) 1 (6)
Abbreviations: OM, occult malignancy; TOAST, Trial of Org 10172 in Acute Stroke Treatment.
Because most OM occurred in the patients with stroke of UND, we analyzed the laboratory data in this group (n 5 156). The patients with undetermined etiology because 2 possible causes were excluded. In a multivariate analysis, we do not detect significant differences with respect to demographic data, NIHSS, clinical subtype according to Oxfordshire classification, incidence of vascular risk factors, percentage of dysphagia, aspiration pneumonia, or mortality during the hospital admission between patients with OM compared with non-OM. However, we detected significant higher baselines levels of fibrinogen (572 mg/dL versus 457 mg/ dL, P 5 .045) and CRP (17.7 mg/L versus 4.8 mg/L, P 5 .001) in the patients with OM compared with nonOM (Table 3). The receiver operating characteristic curves to discriminate between both groups and to estimate its sensitivity and specificity, showed a sensitivity of 75% and specificity of 96% for levels of CRP more than 20 mg/L (area under the curve, 0.917), and a sensitivity of 67% and specificity of 91% for fibrinogen levels more than 600 mg/dL (area under the curve, 0.646).
Discussion The major findings of the present study are (1) on this subject, our contribution is unique in the extensive body Table 3. Baseline fibrinogen and CRP levels in undetermined stroke patients OM group (n 5 8) Fibrinogen mean (mg/dL) CRP mean (mg/L)
572 17.7
Non-OM group (n 5 148) 457 4.8
P .045 .001
Abbreviations: CRP, C-reactive protein; OM, occult malignancy.
4
of literature, (2) the 2.1% of acute stroke patients had an OM, and this percentage was higher in undetermined etiology (5.3%), and (3) baseline levels of fibrinogen and CRP could be good predictors of OM in stroke patients with undetermined etiology. Autopsies studies of cancer patients, found evidence of cerebral infarction in 15% of cases,1 and the most common mechanism of stroke was atherosclerosis, but most of these infarctions were asymptomatic. The pathophysiology of cerebrovascular ischemic events in cancer in the reported clinical series is controversial. Nonbacterial thrombotic endocarditis (27%), hypercoagulable states (24%), undetermined (39.8%), and atherosclerosis (14.5%)1,2,7,8,11-13 have been the more frequently detected mechanisms, and usually the stroke occurs at the terminal stage of malignancy,11,14 whereas in other one study8 the mechanism of stroke was not different in patients with cancer compared with stroke patients without cancer. In addition, stroke can also occur as the first sign of cancer, and the diagnosis of the cause of stroke is crucial for treatment and prevention.4 In previous studies with stroke patients, OM was newly diagnosed in .4%-3% of cases, and the most common etiologic mechanisms were atherosclerosis, cardioembolic, and hypercoagulable states, and nonpatients with nonbacterial endocarditis were detected.5 This differences might suggest that different mechanisms of stroke-cancer–related exist according to the advanced malignancy. The more frequently detected malignancies in stroke patients were gastrointestinal (19%-32%), lung (11%38%), gynecological (21%), brain (9%), and prostate (9%).2,5,11,15 However, most of these reports included patients with cancer treated before stroke onset, or fecal occult blood test was systematically performed,5 and therefore, this could explain the higher prevalence of gastrointestinal malignancy. In our study, we did not systematically perform any screening of malignancy, and a predominance for any type of cancer was not found. Also, we did not detect differences in the distribution of vascular risk factors between both groups (non-OM versus OM), which is a consistent finding in previous studies. In our study, the prevalence of OM in stroke patients was similar to previous studies (2.1%).5 However, if we analyze these results according to etiological subtypes, we detected that the patients with UND had more frequent OM (5.3%). Similarly, to previously published studies,5 we did not detect any patient with nonbacterial thrombotic endocarditis. Stroke’s etiology was found in only 5 of the 13 cases (38%) and were LA in 2, CE in 2 and SV in the other one. Different factors have been proposed in stroke patients as markers of OM: laboratory findings (D-dimer .1.3 ng/ dL and hemoglobin ,12.8 g/dL), embolic signal detected by transcranial Doppler, and diffusion-weighted mag-
D. COCHO ET AL.
netic resonance imaging abnormalities in multiple territories.5,6,15,16 In our study, we did not systematically determine D-dimer levels. However, we detected that baseline fibrinogen levels more than 600 mg/dL and CRP more than 20 mg/L might be good predictors for OM in acute stroke patients, with good values of sensitivity and specificity. Although CRP and fibrinogen are nonspecific acute phase reactants that may be elevated in patients with ischemic stroke, this study suggests that they may also be markers for an OM in these patients. An elevation in these 2 markers should prompt consideration for cancer screening particularly in patients in whom the etiology of the ischemic event is not known. Our study has several limitations. First, our study is retrospective, and we have not performed any cancer screening; and therefore, the frequency of OM might be greater than reported. Second, after hospital discharge, the patients without malignancy were not evaluated, and the frequency of malignancy might be greater than reported. The incidence of OM might not be different in the patients with undetermined etiology with regard to the rest of subgroups, and perhaps neither CRP nor fibrinogen levels. Although we do not have enough evidence to accept that the incidence of OM is greater in patients with acute ischemic stroke in whom the etiology is not known, the incidence of malignancy detected in our study was greater than that of the general population. The annual incidence rate of cancer in general population over 65 years is 215 cases/100,000 people; therefore, the incidence expected in our sample with 631 patients should be 1.35 cases, and we have detected 13, and most patients had a stroke of UND. Third, D-dimer is a good biomarker of hypercoagulability and could be a good predictor of OM in patients with stroke of UND, but in our study was not examined, and it is difficult to definitively attribute a stroke mechanism to hypercoagulability based on fibrinogen and CRP level alone. Fourth, the type of study design did not allow us to establish a causal relation between the OM and ischemic stroke; however, our main objective was to evaluate if any biological marker could help to detect the presence of OM. In conclusion, at least 2.1% of acute ischemic stroke patients in our series had an OM, and this percentage was higher in patients with stroke of UND (5.3%). Higher baseline levels of fibrinogen and CRP were detected in stroke patients with OM and could be good predictors of OM. We believe that stroke patients with undetermined etiology and baseline levels of fibrinogen greater than 600 mg/dL or CRP more than 20 mg/L, should be examined for OM. Further studies with more patients, regarding the precise molecular mechanism of cancerrelated coagulopathy can confirm our findings. Acknowledgments: We thank Aitziber Aleu for her assistance with the editing of the article.
PREDICTORS OF OCCULT CANCER IN STROKE PATIENTS
References 1. Graus F, Rogers LR, Posner JB. Cerebrovascular complications in patients with cancer. Medicine (Baltimore) 1985;64:16-35. 2. Chaturvedi S, Ansell J, Recht L. Should cerebral ischemic events in cancer patients be considered a manifestation of hypercoagulability? Stroke 1994;25:1215-1218. 3. Rogers LR. Cerebrovascular complications in cancer patients. Neurol Clin 2003;21:167-192. 4. Grisold W, Oberndorfer S, Struhal W. Stroke and cancer: a review. Acta Neurol Scand 2009;119:1-16. 5. Uemura J, Kimura K, Sibazaki K, et al. Acute stroke patients have occult malignancy more often than expected. Eur Neurol 2010;64:140-144. 6. Arboix A. [Cerebrovascular disease in the cancer patient]. Rev Neurol 2000;31:1250-1252. 7. Taccone FS, Jeangette SM, Blecic SA. First-ever stroke as initial presentation of systemic cancer. J Stroke Cerebrovasc Dis 2008;17:169-174. 8. Kim SG, Hong JM, Kim HY, et al. Ischemic stroke in cancer patients with and without conventional mechanisms: a multicenter study in Korea. Stroke 2010;41: 798-801.
5 9. Bick RL. Cancer-associated thrombosis. N Engl J Med 2003;349:109-111. 10. Adams HP Jr, Bendixen BH, Kappelle LJ, et al. Classification of subtype of acute ischemic stroke. Definitions for use in a multicenter clinical trial. TOAST. Trial of Org 10172 in Acute Stroke Treatment. Stroke 1993;24:35-41. 11. Cestari DM, Weine DM, Panageas KS, et al. Stroke in patients with cancer: incidence and etiology. Neurology 2004;62:2025-2030. 12. Lim SP, Hakim SZ, Schoenhoff DD. Nonbacterial thrombotic endocarditis. Two-dimensional echocardiographic features in an autopsy-proven case. Chest 1987;92: 176-177. 13. Markides V, Nihoyannopoulos P. Non-bacterial thrombotic endocarditis. Eur J Echocardiogr 2000;1:291-294. 14. Yamamoto Y, Georgiadis AL, Chang HM, et al. Posterior cerebral artery territory infarcts in the New England Medical Center Posterior Circulation Registry. Arch Neurol 1999;56:824-832. 15. Seok JM, Kim SG, Kim JW, et al. Coagulopathy and embolic signal in cancer patients with ischemic stroke. Ann Neurol 2010;68:213-219. 16. Kwon HM, Kang BS, Yoon BW. Stroke as the first manifestation of concealed cancer. J Neurol Sci 2007;258:80-83.
Background: We investigated the prevalence of occult malignancy (OM) in acute ischemic stroke patients to evaluate if any biological marker could help to detect the presence of OM. Methods: We retrospectively reviewed all ischemic stroke patients during 48 months. We did not perform any screening for OM. Demographic data, vascular risk factors, routine blood chemistry with fibrinogen and C-reactive protein (CRP), National Institutes of Health Stroke Scale (NIHSS), and etiological subtype of stroke according to Trial of Org 10172 in Acute Stroke Treatment criteria were analyzed. The patients were divided into 2 groups (Non-OM versus OM). Results: We analyzed 631 patients with acute ischemic stroke. The mean age was 69.7 6 12.7 years, and 59% were men. The distribution of vascular risk factors, etiological subgroups, and NIHSS was comparable between both groups. We detected 13 cases (2.1%) with OM, and this percentage was higher in patients with stroke of undetermined etiology (5.3%). We detected significant higher levels of fibrinogen and CRP in patients with stroke of undetermined cause with OM. Receiver operating characteristic curves showed a sensitivity of 75% and specificity of 96% for levels of CRP more than 20 mg/L, and a sensitivity of 67% and specificity of 91% for fibrinogen levels greater than 600 mg/dL. Conclusions: OM was present in 2.1 % of overall patients, and 5.3% of patients with stroke of undetermined cause. Baseline levels of fibrinogen more than 600 mg/dL or CRP greater than 20 mg/L in patients with undetermined stroke might be good predictors of OM. Key Words: Stroke—cancer—occult malignancy—ischemic stroke. Ó 2015 by National Stroke Association
Introduction From the Department of Neurology, Hospital General de Granollers, Universitat Internacional de Catalunya, Barcelona, Spain. Received August 22, 2014; revision received February 1, 2015; accepted February 6, 2015. The authors report no conflicts of interest and nonfinancial disclosures. D.C. carried out the review of the literature, conception of the study, study design, collection and data analysis, writing of the article, and critical revision for intellectual content. J.G. performed the review of the literature, writing of the article, and critical revision for intellectual content. A.B., J.E., A.C.R., J.P., M.J., and P.O. carried out the data analysis and critical revision for intellectual content. Address correspondence to Dolores Cocho, MD, PhD, Unidad de Neurologıa, Servicio de Medicina Interna, Hospital General de Granollers, AV/ Francesc Ribas sn., Barcelona 08402, Spain. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2015 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2015.02.006
Cancer is the main acquired prothrombotic state, and the relationship between cancer and cerebrovascular disease has been documented in patients with a wide variety of tumors.1-3 Previous studies have described 5 etiologic categories of stroke events in patients with cancer: direct tumor effects (tumor seeding with embolism or mucin-secreting tumors), coagulation disorders, infections, complications of therapeutic or diagnostic procedures,1,4 and paraneoplastic causes.4 Cerebrovascular disease occurs in 15% of cancer patients based on autopsy studies, but only about half of them have experienced clinical symptoms of stroke.1 Occasionally stroke is the first manifestation of an occult malignancy (OM) (.4%-3%),5-7 and the identification and treatment of these patients could improve their prognosis. Although patients with systemic cancer usually have poor outcomes, their survival rate is increasing with the
Journal of Stroke and Cerebrovascular Diseases, Vol. -, No. - (---), 2015: pp 1-5
1
D. COCHO ET AL.
2
development of more effective cancer treatments. Early identification of stroke mechanisms may be important in cancer patients because these mechanisms may differ from those in stroke patients without cancer.8 However, the mechanisms underlying stroke in patients with cancer are controversial.9 Our aim was to analyze the prevalence of OM in ischemic stroke patients and its different mechanisms and to evaluate if any usual biological marker could help to detect the presence of OM in stroke patients.
Methods We retrospectively studied patients with ischemic stroke in our Stoke Unit from 2009 to 2012. Medical history, physical examination, routine blood chemistry, electrocardiogram (ECG), chest X-ray, cranial neuroimaging with computed tomography scan or magnetic resonance imaging, arterial study (carotid ultrasound, transcranial Doppler, or magnetic resonance angiography), and transthoracic echocardiography were systematically performed in all patients. Other studies, such as Holter-ECG monitoring or transesophageal echocardiogram, were performed according to clinical criteria. The baseline vascular risk factors were defined as follows: Smoking was considered if the patient reported smoking cigarettes. Hypertension was defined as a systolic blood pressure of 140 mm Hg or greater or diastolic pressure of 90 mm Hg or greater or current use of antihypertensive treatment. Diabetes mellitus was considered in patients with a history of fasting blood glucose of 126 mg/dL or more or current use of antidiabetic drugs. Hypercholesterolemia was defined as a total cholesterol level of 220 mg/dL or more or current use of lipidlowering agents. Stoke were classified etiologically according to Trial of Org 10172 in Acute Stroke Treatment (TOAST) criteria10 as because of large-artery occlusive disease (LA), smallvessel disease (SV), cardioembolism (CE), and other causes or undetermined cause (UND).
Transient ischemic attack was defined as acute onset of focal cerebral or monocular symptoms lasting less than 24 hours and of presumptive vascular origin. In the cases where an OM was suspected by findings in usual physical examination (eg, cervical adenopathies, tumoral oral lesions, or abdominal mass) or in the routine complementary studies (cervical adenopathy in carotid ultrasound, hematuria in urine, or X-ray lung mass), other complementary examinations were performed according to clinical criteria. All occult malignancies were confirmed by biopsy and cytologic study (Table 1). We do not systematically perform any screening for OM.
Statistical Analysis All patients were divided into 2 groups: No OM group (Non-OM) and the newly diagnosed malignancy (OM). All analyses were carried out using the SPSS 19.0 (IBM, Armonk, NY) software package. Categorical variables were compared between the 2 groups using the c2 or by the Fisher exact test. A student t test was used to evaluate continuous variables. A multivariate analysis was performed to detect differences between the group OM and non-OM. If there were significant differences in continuous variables between both groups, the optimal cutoff points of each variable to discriminate the groups were obtained by receiver operating characteristic curves. We considered statistically significant P less than .05.
Results We retrospectively reviewed 631 patients who had suffered a cerebrovascular ischemic event and were hospitalized in our Stroke Unit during 4 consecutive years. Of these events, 19% were transient ischemic attacks and 81% cerebral infarcts. The mean age was 69.7 6 12.7 years, and 59% were men. The median National Institutes of Health Stroke Scale (NIHSS) on hospital admission was 4 points (range: 0-38). The average hospital stay
Table 1. Pathologic findings in the physical examination or in the routine complementary examinations in OM patients Type of malignancy
Abnormal physical examination
Abnormal routine complementary examinations
2 Lung carcinoma 2 Lymphomas 1 Colon carcinoma 1 Colon carcinoma 1 Chronic lymphatic leukemia 1 Breast carcinoma 1 Pancreas carcinoma 1 Uterine carcinoma 1 Multiple myeloma 1 Mouth carcinoma 1 Peritoneal mesothelioma
No No Intestinal occlusion Pain and abdominal mass Cervical plexopathy symptoms Breast mass
Chest X-ray Cervical adenopathies in carotid duplex
Abbreviation: OM, occult malignancy.
Cholestasis in laboratory test Abdominal pain and leukorrhea Bicytopenia in laboratory test Abnormal mouth examination Abdominal pain and ascites
PREDICTORS OF OCCULT CANCER IN STROKE PATIENTS
was 9 days. According to Oxfordshire classification, the strokes could be divided into total anterior circulation infarcts (20%), partial anterior circulation infarcts (32%), lacunar infarcts (30%), and posterior circulation infarcts (18%). According to TOAST criteria, the strokes were classified as follows: LA (14%), SV (19%), CE (31%), other causes (7%), and UND (28%). In the remaining 1% of cases, the stroke was classified also as of UND because of 2 possible etiologies. Medical history, physical examination, complete blood count, coagulation tests, routine blood biochemistry including determination of vitamin B12, folic acid, C-reactive protein (CRP), and fibrinogen, ECG, chest X-ray, and cerebral computed tomography scan were systematically performed. Extracranial and intracranial arterial study was performed in all patients with the following studies; carotid duplex ultrasonography (66%), transcranial Doppler (64%), or magnetic resonance angiogram magnetic resonance angiography (70%). Other performed studies were transthoracic echocardiography (89%) and Holter-ECG monitoring (21%). Vascular risk factors were present in 80% of cases. The distribution was as follows: hypertension (75%), diabetes mellitus (35%), dyslipidemia (54%), atrial fibrillation (23%), alcohol abuse of more than 40 g/day (13%), smoking (22%), drug abuse (1.2%), ischemic heart disease (16%), and previous cerebrovascular disease (20.6%). The more frequent medical complications during admission were dysphagia (19%) and aspiration pneumonia (3.6%). The mortality during the hospitalization was 3.8%. The main reasons of mortality were neurologic (57%), respiratory (28%), cardiac (5%), and others (10%). Thrombolysis with recombinant tissue plasminogen activator was administered in 13.2% of cases. The patients were discharged with antiplatelet drugs in 65.8%, anticoagulants in 34%, antiplatelet associated to anticoagulant drug .2%, statins in 83%, and antihypertensive drugs in 90%. An OM was detected in 13 patients (2.1%): 2 lung carcinomas, 2 colon carcinomas, 2 lymphomas, 1 chronic lymphatic leukemia, 1 breast carcinoma, 1 pancreas carcinoma, 1 uterine carcinoma, 1 multiple myeloma, 1 floor of the mouth carcinoma, and 1 peritoneal mesothelioma. Of the 13 cases with OM, 4 (30%) died with a mean survival time of 15 days. The remaining 9 patients are currently with mean survival time of 29 months. Comparing the OM group (n 5 13) with the non-OM group (n 5 618) there were no significant statistical differences with respect to demographic data, neurologic severity measured with the NIHSS, clinical subtype according to Oxfordshire classification, incidence of vascular risk factors, percentage of dysphagia, aspiration pneumonia, or mortality during the hospital admission. However, we found significant differences in the etiological subtypes according to TOAST criteria. Most patients with OM had a stroke with UND (61.5% versus 24%, P 5 .003; Table 2)
3
Table 2. Results comparing the OM with non-OM patients Non-OM group OM group (n 5 13) (n 5 618) TOAST % (n) Large-artery occlusive disease Cardioembolic Small-vessel disease Other causes Undetermined cause Undetermined by 2 causes
P .003
15.4 (2)
15 (92)
15.4 (2) 7.7 (1) 0 61.5 (8) 0
33 (204) 20 (124) 7 (44) 24 (148) 1 (6)
Abbreviations: OM, occult malignancy; TOAST, Trial of Org 10172 in Acute Stroke Treatment.
Because most OM occurred in the patients with stroke of UND, we analyzed the laboratory data in this group (n 5 156). The patients with undetermined etiology because 2 possible causes were excluded. In a multivariate analysis, we do not detect significant differences with respect to demographic data, NIHSS, clinical subtype according to Oxfordshire classification, incidence of vascular risk factors, percentage of dysphagia, aspiration pneumonia, or mortality during the hospital admission between patients with OM compared with non-OM. However, we detected significant higher baselines levels of fibrinogen (572 mg/dL versus 457 mg/ dL, P 5 .045) and CRP (17.7 mg/L versus 4.8 mg/L, P 5 .001) in the patients with OM compared with nonOM (Table 3). The receiver operating characteristic curves to discriminate between both groups and to estimate its sensitivity and specificity, showed a sensitivity of 75% and specificity of 96% for levels of CRP more than 20 mg/L (area under the curve, 0.917), and a sensitivity of 67% and specificity of 91% for fibrinogen levels more than 600 mg/dL (area under the curve, 0.646).
Discussion The major findings of the present study are (1) on this subject, our contribution is unique in the extensive body Table 3. Baseline fibrinogen and CRP levels in undetermined stroke patients OM group (n 5 8) Fibrinogen mean (mg/dL) CRP mean (mg/L)
572 17.7
Non-OM group (n 5 148) 457 4.8
P .045 .001
Abbreviations: CRP, C-reactive protein; OM, occult malignancy.
4
of literature, (2) the 2.1% of acute stroke patients had an OM, and this percentage was higher in undetermined etiology (5.3%), and (3) baseline levels of fibrinogen and CRP could be good predictors of OM in stroke patients with undetermined etiology. Autopsies studies of cancer patients, found evidence of cerebral infarction in 15% of cases,1 and the most common mechanism of stroke was atherosclerosis, but most of these infarctions were asymptomatic. The pathophysiology of cerebrovascular ischemic events in cancer in the reported clinical series is controversial. Nonbacterial thrombotic endocarditis (27%), hypercoagulable states (24%), undetermined (39.8%), and atherosclerosis (14.5%)1,2,7,8,11-13 have been the more frequently detected mechanisms, and usually the stroke occurs at the terminal stage of malignancy,11,14 whereas in other one study8 the mechanism of stroke was not different in patients with cancer compared with stroke patients without cancer. In addition, stroke can also occur as the first sign of cancer, and the diagnosis of the cause of stroke is crucial for treatment and prevention.4 In previous studies with stroke patients, OM was newly diagnosed in .4%-3% of cases, and the most common etiologic mechanisms were atherosclerosis, cardioembolic, and hypercoagulable states, and nonpatients with nonbacterial endocarditis were detected.5 This differences might suggest that different mechanisms of stroke-cancer–related exist according to the advanced malignancy. The more frequently detected malignancies in stroke patients were gastrointestinal (19%-32%), lung (11%38%), gynecological (21%), brain (9%), and prostate (9%).2,5,11,15 However, most of these reports included patients with cancer treated before stroke onset, or fecal occult blood test was systematically performed,5 and therefore, this could explain the higher prevalence of gastrointestinal malignancy. In our study, we did not systematically perform any screening of malignancy, and a predominance for any type of cancer was not found. Also, we did not detect differences in the distribution of vascular risk factors between both groups (non-OM versus OM), which is a consistent finding in previous studies. In our study, the prevalence of OM in stroke patients was similar to previous studies (2.1%).5 However, if we analyze these results according to etiological subtypes, we detected that the patients with UND had more frequent OM (5.3%). Similarly, to previously published studies,5 we did not detect any patient with nonbacterial thrombotic endocarditis. Stroke’s etiology was found in only 5 of the 13 cases (38%) and were LA in 2, CE in 2 and SV in the other one. Different factors have been proposed in stroke patients as markers of OM: laboratory findings (D-dimer .1.3 ng/ dL and hemoglobin ,12.8 g/dL), embolic signal detected by transcranial Doppler, and diffusion-weighted mag-
D. COCHO ET AL.
netic resonance imaging abnormalities in multiple territories.5,6,15,16 In our study, we did not systematically determine D-dimer levels. However, we detected that baseline fibrinogen levels more than 600 mg/dL and CRP more than 20 mg/L might be good predictors for OM in acute stroke patients, with good values of sensitivity and specificity. Although CRP and fibrinogen are nonspecific acute phase reactants that may be elevated in patients with ischemic stroke, this study suggests that they may also be markers for an OM in these patients. An elevation in these 2 markers should prompt consideration for cancer screening particularly in patients in whom the etiology of the ischemic event is not known. Our study has several limitations. First, our study is retrospective, and we have not performed any cancer screening; and therefore, the frequency of OM might be greater than reported. Second, after hospital discharge, the patients without malignancy were not evaluated, and the frequency of malignancy might be greater than reported. The incidence of OM might not be different in the patients with undetermined etiology with regard to the rest of subgroups, and perhaps neither CRP nor fibrinogen levels. Although we do not have enough evidence to accept that the incidence of OM is greater in patients with acute ischemic stroke in whom the etiology is not known, the incidence of malignancy detected in our study was greater than that of the general population. The annual incidence rate of cancer in general population over 65 years is 215 cases/100,000 people; therefore, the incidence expected in our sample with 631 patients should be 1.35 cases, and we have detected 13, and most patients had a stroke of UND. Third, D-dimer is a good biomarker of hypercoagulability and could be a good predictor of OM in patients with stroke of UND, but in our study was not examined, and it is difficult to definitively attribute a stroke mechanism to hypercoagulability based on fibrinogen and CRP level alone. Fourth, the type of study design did not allow us to establish a causal relation between the OM and ischemic stroke; however, our main objective was to evaluate if any biological marker could help to detect the presence of OM. In conclusion, at least 2.1% of acute ischemic stroke patients in our series had an OM, and this percentage was higher in patients with stroke of UND (5.3%). Higher baseline levels of fibrinogen and CRP were detected in stroke patients with OM and could be good predictors of OM. We believe that stroke patients with undetermined etiology and baseline levels of fibrinogen greater than 600 mg/dL or CRP more than 20 mg/L, should be examined for OM. Further studies with more patients, regarding the precise molecular mechanism of cancerrelated coagulopathy can confirm our findings. Acknowledgments: We thank Aitziber Aleu for her assistance with the editing of the article.
PREDICTORS OF OCCULT CANCER IN STROKE PATIENTS
References 1. Graus F, Rogers LR, Posner JB. Cerebrovascular complications in patients with cancer. Medicine (Baltimore) 1985;64:16-35. 2. Chaturvedi S, Ansell J, Recht L. Should cerebral ischemic events in cancer patients be considered a manifestation of hypercoagulability? Stroke 1994;25:1215-1218. 3. Rogers LR. Cerebrovascular complications in cancer patients. Neurol Clin 2003;21:167-192. 4. Grisold W, Oberndorfer S, Struhal W. Stroke and cancer: a review. Acta Neurol Scand 2009;119:1-16. 5. Uemura J, Kimura K, Sibazaki K, et al. Acute stroke patients have occult malignancy more often than expected. Eur Neurol 2010;64:140-144. 6. Arboix A. [Cerebrovascular disease in the cancer patient]. Rev Neurol 2000;31:1250-1252. 7. Taccone FS, Jeangette SM, Blecic SA. First-ever stroke as initial presentation of systemic cancer. J Stroke Cerebrovasc Dis 2008;17:169-174. 8. Kim SG, Hong JM, Kim HY, et al. Ischemic stroke in cancer patients with and without conventional mechanisms: a multicenter study in Korea. Stroke 2010;41: 798-801.
5 9. Bick RL. Cancer-associated thrombosis. N Engl J Med 2003;349:109-111. 10. Adams HP Jr, Bendixen BH, Kappelle LJ, et al. Classification of subtype of acute ischemic stroke. Definitions for use in a multicenter clinical trial. TOAST. Trial of Org 10172 in Acute Stroke Treatment. Stroke 1993;24:35-41. 11. Cestari DM, Weine DM, Panageas KS, et al. Stroke in patients with cancer: incidence and etiology. Neurology 2004;62:2025-2030. 12. Lim SP, Hakim SZ, Schoenhoff DD. Nonbacterial thrombotic endocarditis. Two-dimensional echocardiographic features in an autopsy-proven case. Chest 1987;92: 176-177. 13. Markides V, Nihoyannopoulos P. Non-bacterial thrombotic endocarditis. Eur J Echocardiogr 2000;1:291-294. 14. Yamamoto Y, Georgiadis AL, Chang HM, et al. Posterior cerebral artery territory infarcts in the New England Medical Center Posterior Circulation Registry. Arch Neurol 1999;56:824-832. 15. Seok JM, Kim SG, Kim JW, et al. Coagulopathy and embolic signal in cancer patients with ischemic stroke. Ann Neurol 2010;68:213-219. 16. Kwon HM, Kang BS, Yoon BW. Stroke as the first manifestation of concealed cancer. J Neurol Sci 2007;258:80-83.
