Prachi Mehndiratta, MD

Navi BB, Singer S, Merkler AE, Cheng NT, Stone JB, Kamel HK, et al. Cryptogenic Subtype Predicts Reduced Survival Among Cancer Patients With Ischemic Stroke. Stroke. 2014


It is traumatic to experience a stroke and probably even more to not know what caused it. In about one-third of patients with acute ischemic stroke no clear etiology is found. It is hard to predict the recurrence risk of stroke in such patients particularly as the inciting factor or cause is not known. About 7% of patients with cancer experience a clinically significant stroke. These strokes can break their treatment cycles and often lead to significant disability. The authors in this article attempt to elucidate the frequency of cryptogenic stroke in patients with underlying malignancy as well as their outcomes after a cryptogenic stroke.



The authors identified 263 patients with acute ischemic stroke in a retrospective cohort of cancer patients admitted to their center between 2005 and 2009, by utilizing ICD 9 codes for stroke or transient ischemic attack. Patients that underwent only a CT scan and those seen exclusively in the outpatient setting were excluded due to concern for insufficient work up. Patients were followed till the primary outcome of death was reached or till July 31st 2012. Demographic, outcome and imaging data was rigorously collected and strokes were classified by using the TOAST classification. Stroke type was classified as cryptogenic if no clear cause was identified or if there were possibly two competing mechanisms. Multivariate cox proportional hazards analysis was used to determine an independent relationship between etiology of stroke and death and Kaplan-Meier survival curves were constructed.

Their results indicate that adenocarcinomas of solid organs such as lung, GI or GU tract were most often associated with ischemic stroke and stroke typically occurred about 10 months after cancer diagnosis. 69% of patients had disseminated disease at the time of stroke occurrence and stroke work up was equally comprehensive for the cryptogenic and non-cryptogenic stroke types. 133 of 263 patients were identified to have cryptogenic stroke and 76 of these had a cardioembolic appearing imaging pattern. Median survival among those with cryptogenic stroke was significantly lower (47 days vs. 141 days) as compared to those with known stroke mechanism and even lower in those that appeared cardioembolic (31 days). After adjusting for age, functional status, adenocarcinoma histology and known metastases, cryptogenic stroke type remained and independent predictor of death. (HR 1.64 95% CI (1.25-2.14)).

This study raised several important questions – does a cryptogenic stroke in a cancer patient spell doom? And if it does then what can we do to ascertain the cause better? One possible suggestion is that all patients undergo transesophageal echocardiogram and that the search for marantic endocarditis is more rigorous and a more comprehensive classification system such as the Causative Classification of Stroke (CCS) is employed to determine stroke etiology. I would have also liked to know a little more about differences in stroke severity between the two groups and if the patients underwent any repeat evaluations during follow up.  Practically speaking, should the patients that have a cryptogenic stroke and a cardioembolic appearing stroke be empirically treated with anticoagulation?