Professional Documents
Culture Documents
DOI 10.1007/s11060-012-1011-4
CLINICAL STUDY
Jae-Kyu Roh
Received: 26 April 2012 / Accepted: 20 November 2012 / Published online: 9 January 2013
Springer Science+Business Media New York 2013
Abstract Cancer related stroke may have different pheno- Keywords Cerebral infarction Cancer Recurrence
types from non-cancer stroke, especially in terms of stroke
progression and recurrence. We performed a casecontrol
study to identify their incidences and risk factors in cancer
related stroke. Between January 2001 and December 2009, we Introduction
conducted a retrospective review of acute ischemic stroke
patients with cancer who were admitted to Seoul National Stroke and cancer are devastating diseases with significant
University Hospital, Seoul, Korea. The stroke patients without mortality and morbidity in aged populations. Stroke can
cancer served as control. We collected demographic variables, manifest as direct or indirect cancer effects, namely can-
vascular risk factors, stroke phenotype, clinical course, and cer related stroke. As cancer treatment improves and
cancer information including diagnosis, stage, and treatment patients survive longer, it is inevitable that cancer patients
status. Among cancer stroke patients, the potential risk factor of suffering from stroke will increase in the future [1, 2].
stroke recurrence was evaluated. The mean age of the 102 Cancer related stroke has attracted research interest
cancer patients was 66.4 10.8 years, and 64.7 % were men. because of its distinctive pathomechanism and clinical/
The mean time interval from cancer diagnosis to stroke onset radiologic characteristics [2, 3]. Initial studies focused on
was 39.7 60.9 months. The principal lesion pattern of can- pathologic features of cancer stroke from postmortem
cer stroke was multiple dots extending single vascular territory autopsy findings [4], and recent studies described the
(39.2 %), and they were associated with low hemoglobin and relationship between brain magnetic resonance imaging
high fibrinogen levels. Stroke progression and recurrence were (MRI) findings and laboratory data [57]. However, few
noted in 9.8 and 27.5 % of cancer stroke patients, and in 9.3 and studies have focused on clinical course of cancer stroke,
12.7 % of control patients, respectively. The stroke subtype probably due to the small number of patients and cross-
was independently associated with recurrence of cancer stroke sectional study design. Only one group reported that
after multiple logistic regression (odds ratio = 3.165, 95 % stroke patients with cancer had higher mortality than
confidence interval = 1.0809.277, p = 0.036). Cancer rela- general stroke patients because of poor general health
ted stroke has a distinct phenotype in terms of infarction pattern status by malignancy [6].
and laboratory findings. Stroke recurrence is frequently Early progression and recurrence of stroke are major
observed among cancer stroke patients, and its risk is related contributing factors affecting functional outcome of stroke
with stroke subtype. survivors [8, 9]. Cancer related stroke may have different
phenotypes from general stroke in terms of stroke pro-
gression and recurrence, but few studies have focused on
J.-M. Kim K.-H. Jung K. H. Park S.-T. Lee K. Chu their incidences and risk factors. In this study we con-
J.-K. Roh (&) ducted a retrospective review of cancer patients diagnosed
Stroke & Stem Cell Laboratory, Department of Neurology,
with acute ischemic stroke to evaluate clinical and
Clinical Research Institute, Seoul National University Hospital,
28, Yongon-dong, Chongro-gu, Seoul 110-744, South Korea radiological characteristics, and to find risk factors of
e-mail: rohjk@snu.ac.kr stroke recurrence.
123
296 J Neurooncol (2013) 111:295301
123
J Neurooncol (2013) 111:295301 297
Total of 116 stroke patients was combined with cancer at Patient number 102 171
stroke onset. Among those patients, 11 patients with hemor- Demographic variable
rhagic stroke and 3 patients with hematologic malignancy Age (years) 66.4 10.8 64.4 11.8 0.175
were excluded in our study, and finally 102 patients were Gender (male) 66 (64.7) 107 (62.6) 0.795
reviewed in this study. The mean age of the CSP was Cardiovascular risk factors
66.4 10.8 years and 64.7 % of the patients were male Hypertension 53 (52.0) 110 (64.2) 0.100
(Table 1). Brain MRI of CSP showed distinctive findings Diabetes mellitus 24 (23.5) 56 (32.7) 0.179
(Fig. 1). The most common lesion pattern among CSP was Dyslipidemia 17 (16.7) 37 (21.6) 0.396
multiple dots that were not confined within a single vascular Smoking 40 (39.2) 45 (26.3) 0.061
territory (39.2 %), and the unconventional subtype (25.5 %) Heart disease 23 (22.5) 45 (26.3) 0.482
was more prevalent in CSP than in control group. On the other Stroke type
hand, control stroke patients more frequently had large artery Conventional 76 (75.5) 164 (96.1)
atherosclerosis, and small vascular pathologies including Unconventional 26 (25.5) 7 (3.9) \0.001*
periventricular white matter change and old lacunes than did Stroke lesion pattern
CSP. Hemoglobin level was significantly lower in CSP than No lesion 8 (7.8) 24 (14.0)
in control patients (p = 0.005) and fibrinogen level tended to Anterior territory 29 (28.4) 79 (46.2)
be higher in CSP (p = 0.05). The prevalence of vascular risk Posterior territory 25 (24.5) 54 (31.6)
factors and stroke severity in terms of NIHSS were not dif- Multiple territory 40 (39.2) 14 (8.2) \0.001*
ferent between the two groups. Although stroke progression Functional outcome
was not different between the CSP and control patients (9.8 NIH stroke scale 5.9 5.7 4.8 4.7 0.102
vs. 9.4 %), stroke recurred more commonly in CSP than in Symptom progression 10 (9.8) 16 (9.4) 0.903
control group (27.5 vs. 12.9 %; p = 0.003). Six patients died Stroke recurrence 28 (27.5) 22 (12.9) 0.003*
during hospitalization in each group. More CSP did not Laboratory finding
receive antithrombotic treatment than control stroke patients Hemoglobin (g/dL) 12.2 2.4 13.1 1.8 0.005*
because of terminal cancer state (12.8 vs. 1.2 %; p \ 0.001). C-reactive protein (mg/dl) 0.98 1.49 1.09 1.76 0.646
Fibrinogen (mg/dL) 387 123 359 97 0.050
Combined cancer status Brain imaging finding
Intracranial stenosis 34 (33.3) 90 (52.6) 0.016*
The most common malignancy among CSP was stomach Extracranial stenosis 25 (24.5) 45 (26.3) 0.976
cancer (17.7 %), followed by genitourinary cancer (14.7 %), White matter change 0.6 2.0 7.6 28.7 0.017*
lung cancer (12.7 %), pancreatico-biliary cancer (12.7 %), and volume (mm3)
brain neoplasm (11.8 %) (Table 2). When compared with the Microbleeds on T2* 1.2 3.6 1.8 5.1 0.360
total number of cancer patients, the ratio of cancer stroke was gradient echo
high in patients with brain tumor (0.24 %) and pancretico- Old lacunes 0.1 0.4 0.7 1.1 \0.001*
biliary (0.14 %), and low in breast cancer patients (0.01 %). Stroke treatment
Cancer stage was confined within the primary site in 49 patients Antiplatelet 69 (67.6) 129 (75.4)
(48.0 %), and the other patients had local invasion or systemic Anticoagulant 17 (16.7) 32 (18.7)
metastasis. Three patients had limited staging work-up due to Combination of both 3 (2.9) 8 (4.7)
terminal stage and they were included in systemic group when No treatment 13 (12.8) 2 (1.2) \0.001*
performing statistical analysis. The patients combined with
* Statistically significant difference (p \ 0.05)
extended or systemic cancer had lower hemoglobin level than
the patients with localized malignancy (p = 0.001). Various (39.4 %) experienced stroke within 1 year, and nine out of ten
treatment options including surgery, chemotherapy, radio- patients (90 %) who did not receive any anticancer treatment
therapy or combination of them were performed in CSP, but 10 experienced stroke within 1 year.
patients (9.8 %) did not receive anticancer treatment because
of terminal disease status. The mean time period from cancer Stroke recurrence among CSP
diagnosis to stroke onset was 39.7 60.9 months, and vari-
able according to cancer treatment status (Table 3). Among the Stroke recurrence was reported in 28 patients and was related
38 patients who had received surgical treatment, 15 patients with the number of microbleeds, extracranial stenosis,
123
298 J Neurooncol (2013) 111:295301
Fig. 1 Brain MRI disclosed various lesion patterns of cancer related stroke with unconventional etiology
unconventional stroke etiology, and cancer treatment status Table 2 Cancer characteristics of the 102 cancer stroke patients
(Table 4). After multiple logistic regression, the unconven- Number (%) Total cancer patients
tional etiology (p = 0.036, OR = 3.165, 95 % CI = (CSP %)
1.0809.277) was the only independent variable predicting
Cancer type
recurrence (Table 5), and no treatment of cancer showed a
Stomach 18 (17.7) 27,035 (0.0665 %)
tendency to increase the OR of stroke recurrence (p = 0.086,
Genitourinary 15 (14.7) 29,742 (0.0504 %)
OR = 3.742, 95 % CI = 0.82916.897).
Lung 13 (12.7) 19,542 (0.0665 %)
Pancreatico-biliary 13 (12.7) 9,296 (0.1398 %)
Brain 12 (11.8) 4,955 (0.2422 %)
Discussion
Colon 10 (9.8) 29,358 (0.0341 %)
Head & Neck 9 (8.8) 29,122 (0.0309 %)
This study illustrates the distinct radiological phenotype of
Breast 5 (4.9) 53,006 (0.0094 %)
CSP, which is widespread lesion regardless of single vas-
cular territory. Cancer related stroke was combined with Liver 5 (4.9) 22,984 (0.0218 %)
low hemoglobin and high fibrinogen level; whereas both Musculoskeletal 2 (2.0) 7,534 (0.0265 %)
large artery and small vessel pathologies were less frequent Total 102 (100) 232,574 (0.0439 %)
in CSP than in control patients. Our data also suggest Cancer stage
different stroke prevalence in terms of primary cancer Localized 49 (48.0)
origin. Stroke progression and recurrence among CSP are Extended 23 (22.6)
frequent, and stroke subtype of unconventional etiology is Systemic 27 (26.5)
an independent factor predicting stroke recurrence. Unknown 3 (2.9)
The majority of CSP had widespread multiple lesions Cancer treatment
that were not confined in one vascular territory, and brain Surgery only 38 (37.3)
MRI with diffusion weighted images disclosed that Chemotherapy only 16 (15.7)
infarctions were principally located in multiple end artery Radiotherapy only 3 (2.9)
zones. Intracranial large artery atherosclerosis and small Combination of above 35 (34.3)
vessel disease in terms of old lacune numbers and white No treatment 10 (9.8)
matter hyperintensity volume were less extensive in CSP, CSP cancer stroke patients
implying that conventional vascular pathologies are less
appreciated in cancer associated stroke than distinct cancer
related conditions such as coagulation disorder. The studies [6, 12, 13]. Recently, one group detected multiple
prevalence of vascular risk factors was not different embolic signals from transcranial Doppler and elevated
between the two groups, in agreement with previous serum D-dimer level in CSP, which was attenuated by
123
J Neurooncol (2013) 111:295301 299
Within 1 years 11 15 1 13 9
Within 5 years 4 17 0 12 1
More than 5 years 1 6 2 10 0
Subtotal 16 38 3 35 10
Mean time interval (months) 16.1 22.6 36.1 66.4 115.3 106.1 51.8 63.1 2.8 8.6
Table 4 Factors related to stroke recurrence in patients with cancer- Table 5 Logistic regression analysis of stroke recurrence among
related stroke cancer stroke patients
No recurrence Recurrence p value Odds 95 % Confidence p value
(n = 74) (n = 28) ratio interval
Age (years) 65.3 11.5 69.3 8.2 0.090 Age 1.032 0.9711.098 0.307
Sex (Male) 46 (62.2) 20 (71.4) 0.488 Stroke etiology, 3.165 1.0809.277 0.036
Hypertension 38 (51.4) 15 (53.6) 0.708 unconventional
Diabetes mellitus 17 (23.0) 7 (25.0) 0.794 Extracranial stenosis 2.301 0.7666.910 0.138
Dyslipidemia 12 (16.2) 5 (53.5) 1.000 Cancer treatment, no 3.742 0.82916.897 0.086
therapy
Smoking 28 (37.8) 12 (42.9) 0.656
Heart disease 15 (20.3) 8 (28.5) 0.421
Stroke type 0.181
Conventional 57 (77.0) 16 (57.1) prostate cancer [12]. The difference between studies may
Unconventional 17 (23.0) 12 (42.9) reflect distinct regional cancer prevalence and different
Lesion pattern 0.739 study inclusion criteria. Stomach cancer was the most
Single lesion 47 (63.5) 15 (53.6) common malignancy with stroke in this study, probably
Multiple territory 27 (36.5) 13 (46.4)
because it is the most common cancer in Koreans [15].
NIH stroke scale 5.6 6.6 7.0 6.2 0.285
Brain tumor (0.2422 %) and pancreatico-biliary tumor
Hemoglobin (g/dL) 12.1 2.2 12.3 2.8 0.826
(0.1398 %) patients had relatively higher CSP proportion
than breast cancer patients (0.0094 %), suggesting the
Fibrinogen (mg/dL) 352 100 401 129 0.071
different susceptibility of cancer related stroke. The dif-
Intracranial stenosis 25 (33.7) 11 (39.2) 0.704
ference seems to stem from different cancer cell histology,
Extracranial stenosis 14 (18.9) 12 (42.8) 0.012*
local tumor effect, or treatment modality. Primary brain
White matter volume 0.5 1.7 0.8 2.5 0.475
(mm3) tumor has been associated with an increased risk for stroke,
Microbleeds on T2* 0.5 1.7 2.6 5.7 0.011* mainly from treatment complications such as surgery and
gradient echo radiotherapy [16]. However, it is also probable that cancer
Old lacunes 0.1 0.4 0.1 0.3 0.840 categorization results in a selection bias. For example,
Cancer stage 0.534 many breast cancer patients are diagnosed at an early stage
Localized 38 (51.4) 11 (39.3) by routine healthcare check-up when the systemic effect of
Extended or systemic 36 (48.6) 17 (60.7) cancer is minimal, whereas pancreatico-biliary tumors are
Cancer treatment 0.022* relatively hard to diagnose at an early stage. Patient follow-
Treatment 70 (94.6) 22 (78.6) up status such as death at home or referral to other hospital
No treatment 4 (5.4) 6 (21.4) is another important factor that may distort cancer stroke
prevalence. Studies including multiple centers or with
* Statistically significant difference (p \ 0.05)
prospective design may be necessary to confirm stroke
susceptibility of different cancer origin.
anticoagulation [14]. Based on these findings, it is sug- Cancer stroke patients had lower level of hemoglobin
gested that CSP management should focus on the evalua- than control patients, possibly because cancer patients were
tion of embolic source or hypercoagulable condition. anemic as the result of their chronic disease, and many
Stroke susceptibility seems to be different in terms of patients had gastrointestinal malignancies that were com-
primary cancer location based on our data. A previous monly related with blood loss. The relationship between
study reported that lung cancer is the most common neo- anemia and stroke outcome is controversial in the general
plasm combined with stroke, followed by brain and population, and both low and high hemoglobin levels are
123
300 J Neurooncol (2013) 111:295301
detrimental to stroke outcome [1719]. In some diseases, and mortality. Different inclusion time between cancer and
such as sickle cell anemia, a low level of hemoglobin is control stroke groups is another weak point of our study,
associated with increased stroke risk, and transfusion can although stroke diagnosis and treatment strategy were similar
attenuate stroke incidence by increasing oxygen carrying in the two time periods. The relationship between cancer
capacity [20, 21]. Further studies are required to confirm treatment option and stroke risk is an important issue, but this
the pathological contribution of anemia in CSP and the study was not able to suggest anticancer treatment effect on
effect of its correction to stroke outcome. We also found stroke risk because of not enough patients to consider
that fibrinogen, which is a coagulation cascade protein, underlying cancer status, patient condition and combined
tended to be increased in CSP than in control patients. vascular risk factors at the same time [31]. Future studies with
Fibrinogen increase has been reported in stroke patients larger number of patient inclusion or with homogeneous
and one study reported its association with atherosclerotic cancer patient cohort will disclose possible association
progression and stroke recurrence [22, 23]. Underlying between anticancer treatment and stroke risk.
cancer may have contributed elevated fibrinogen level This study describes distinct clinical characteristics of
because fibrinogen is a reactive protein that can be elevated stroke patients combined with malignancy, and suggests
in any form of inflammation. possible risk factors of stroke recurrence. Stroke subtypes
Stroke recurrence is related with a patients functional and malignancy related conditions may deserve more
outcome and mortality, and attempts to prevent its recur- attention in predicting stroke outcome than conventional
rence are essential to minimize neurological deficit after vascular pathology. Further studies are warranted to dis-
stroke [24, 25]. Major vessel occlusion, diabetes mellitus, close any modifiable risk factor of recurrence and to sug-
and low blood pressure have been suggested as a risk factor gest better secondary preventive strategy in cancer stroke.
of early recurrence among general stroke population [25
27]. In terms of stroke subtypes, large artery disease or Acknowledgments This study was supported by a grant of the
Korean Health Technology R&D Project, Ministry of Health &
cardioembolism confer a higher risk of recurrence than Welfare, Republic of Korea (A110045).
other factors in general stroke patients [2830]. However,
those parameters were not evident in cancer related stroke.
The stroke recurrence among CSP was associated with References
unconventional stroke etiology, no cancer treatment,
extracranial stenosis, and microbleed number from uni- 1. Wen PY, Glantz MJ (2003) Neurologic complications of cancer.
variate analysis. Unconventional stroke subtype indepen- Preface. Neurol Clin 21:xixiii
dently predicted stroke recurrence among CSP, and the 2. Grisold W, Oberndorfer S, Struhal W (2009) Stroke and cancer: a
review. Acta Neurol Scand 119:116
difference of vascular pathology was not significant after 3. Rogers LR (2004) Cerebrovascular complications in patients with
adjustment. Hidden embolic source or coagulopathy might cancer. Semin Neurol 24:453460
exist among CSP with unconventional etiology, which had 4. Nguyen T, DeAngelis LM (2006) Stroke in cancer patients. Curr
not been disclosed during hospitalization. Since stroke Neurol Neurosci Rep 6:187192
5. Kwon HM, Kang BS, Yoon BW (2007) Stroke as the first man-
recurred more commonly among CSP than general stroke ifestation of concealed cancer. J Neurol Sci 258:8083
patients, and the portion of unconventional etiology were 6. Zhang YY, Cordato D, Shen Q et al (2007) Risk factor, pattern,
considerable (25.5 %), more proactive treatment may be etiology and outcome in ischemic stroke patients with cancer: a
required for secondary prevention. nested case-control study. Cerebrovasc Dis 23:181187
7. Taccone FS, Jeangette SM, Blecic SA (2008) First-ever stroke as
Several limitations exist in this study. First of all there is initial presentation of systemic cancer. J Stroke Cerebrovasc Dis
debate on how to define cancer related stroke, because it 17:169174
should be differentiated from stroke accidentally combined 8. Petty GW, Brown RD Jr, Whisnant JP et al (2000) Ischemic
with cancer. The causal relationship between the two dis- stroke subtypes: a population-based study of functional outcome,
survival, and recurrence. Stroke 31:10621068
eases is sometimes evident in such cases of multiple infarc- 9. Sumer M, Ozdemir I, Erturk O (2003) Progression in acute
tions in cancer patients without any vascular risk factor, but ischemic stroke: frequency, risk factors and prognosis. J Clin
strokes combined with clear conventional mechanisms by Neurosci 10:177180
TOAST criteria may have little pathologic relationship with 10. Aho K, Harmsen P, Hatano S et al (1980) Cerebrovascular dis-
ease in the community: results of a WHO collaborative study.
cancer itself [13]. Therefore, unconventional subtype may Bull World Health Organ 58:113130
represent genuine cancer related stroke. Secondly, selec- 11. Adams HP Jr, Bendixen BH, Kappelle LJ et al (1993) Classifi-
tion bias may exist in stroke detection when cancer patients cation of subtype of acute ischemic stroke: definitions for use in a
refuse brain imaging due to terminal disease status. Since this multicenter clinical trial TOAST Trial of Org 10172 in Acute
Stroke Treatment. Stroke 24:3541
study was performed retrospectively, several necessary 12. Cestari DM, Weine DM, Panageas KS et al (2004) Stroke in
information including pathologic and laboratory data such as patients with cancer: incidence and etiology. Neurology 62:
D-dimer were not obtained, as well as overall quality of life 20252030
123
J Neurooncol (2013) 111:295301 301
13. Kim SG, Hong JM, Kim HY et al (2010) Ischemic stroke in 23. Audebert HJ, Pellkofer TS, Wimmer ML et al (2004) Progression
cancer patients with and without conventional mechanisms: a in lacunar stroke is related to elevated acute phase parameters.
multicenter study in Korea. Stroke 41:798801 Eur Neurol 51:125131
14. Seok JM, Kim SG, Kim JW et al (2010) Coagulopathy and 24. Thanvi B, Treadwell S, Robinson T (2008) Early neurological
embolic signal in cancer patients with ischemic stroke. Ann deterioration in acute ischaemic stroke: predictors, mechanisms
Neurol 68:213219 and management. Postgrad Med J 84:412417
15. Jung KW, Park S, Kong HJ et al (2011) Cancer statistics in 25. Hillen T, Coshall C, Tilling K et al (2003) Cause of stroke
Korea: incidence, mortality, survival, and prevalence in 2008. recurrence is multifactorial: patterns, risk factors, and outcomes
Cancer Res Treat 43:111 of stroke recurrence in the South London stroke register. Stroke
16. Kreisl TN, Toothaker T, Karimi S et al (2008) Ischemic stroke in 34:14571463
patients with primary brain tumors. Neurology 70:23142320 26. Davalos A, Toni D, Lweins F et al (1999) Neurological deteri-
17. Huang WY, Chen IC, Meng L et al (2009) The influence of oration in acute ischemic stroke: potential predictors and asso-
anemia on clinical presentation and outcome of patients with ciated factors in the European cooperative acute stroke study
first-ever atherosclerosis-related ischemic stroke. J Clin Neurosci (ECASS) I. Stroke 30:26312636
16:645649 27. Ay H, Gungor L, Arsava EM et al (2010) A score to predict early
18. Tanne D, Molshatzki N, Merzeliak O et al (2010) Anemia status, risk of recurrence after ischemic stroke. Neurology 74:128135
hemoglobin concentration and outcome after acute stroke: a 28. Jackson CA, Hutchison A, Dennis MS et al (2009) Differences
cohort study. BMC Neurol 10:22 between ischemic stroke subtypes in vascular outcomes support a
19. Irace C, Ciamei M, Crivaro A et al (2003) Hematocrit is asso- distinct lacunar ischemic stroke arteriopathy: a prospective,
ciated with carotid atherosclerosis in men but not in women. hospital-based study. Stroke 40:36793684
Coron Artery Dis 14:279284 29. Soda T, Nakayasu H, Maeda M et al (2004) Stroke recurrence
20. Adams RJ, McKie VC, Hsu L et al (1998) Prevention of a first within the first year following cerebral infarctionTottori Uni-
stroke by transfusions in children with sickle cell anemia and versity Lacunar Infarction Prognosis Study (TULIPS). Acta
abnormal results of transcranial Doppler ultrasonography. N Engl Neurol Scand 110:343349
J Med 339:511 30. Lovett JK, Coull AJ, Rothwell PM (2004) Early risk of recur-
21. Lee MT, Piomelli S, Granger S, STOP Study Investigators et al rence by subtype of ischemic stroke in population-based inci-
(2006) Stroke Prevention Trial in Sickle Cell Anemia (STOP): dence studies. Neurology 62:569573
extended follow-up and final results. Blood 108:847852 31. Falanga A, Marchetti M (2012) Anticancer treatment and
22. Tsuda Y, Satoh K, Kitadai M et al (1997) Hemorheologic profiles thrombosis. Thromb Res 129:353359
of plasma fibrinogen and blood viscosity from silent to acute and
chronic cerebral infarctions. J Neurol Sci 147:4954
123
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.