WSO - April 2024 - 407

Naftali et al.
407
Introduction
Cerebral microinfarcts (CMIs) are microscopic lesions of
cellular death or tissue necrosis and are probably the most
common type of brain infracts.1 However, due to their
microscopic size, they are undetectable on conventional
structural magnetic resonance imaging (MRI) sequencing
and even on gross pathological examination.1 Special MRI
sequence, the diffusion-weighted imaging (DWI), can
detect CMI, but only within a limited time period of
5-14 days following their formation.2
CMI can be detected in different clinical settings such as
cerebral small vessel disease (CSVD),3,4 ipsilateral carotid
occlusion5 or stenosis,6 acute ischemic stroke,7 and cardioembolism.8
A randomly detected CMI on MRI implies the
annual incidence of hundreds of new CMI9 and may therefore
serve as a strong predictor for future stroke and cognitive
decline.10-13
Cancer-related stroke (CRS) is an emerging subtype of
ischemic stroke recently gaining special clinical and scientific
attention.14 CRS has unique pathomechanisms and
high occurrence risk following cancer diagnosis, especially
lung cancer (LC).15-17 Stoke incidence in LC patients was
found to be 1.5 times higher (25.9 vs 17.4 per 1000 person/
years) compared with non-cancer patients.16 In this study,
we aimed to detect incidental CMI as a possible marker for
ongoing cerebral ischemia among patients with active LC
relative to other cancer types.
Methods
Setting and data source
This is a retrospective cohort study, part of the PREACHER
(Prediction Analysis for Intra-Cerebral Hemorrhage) collaboration
group. Our group investigated clinical outcomes,
risk factors, and comorbidities in patients with CSVD
imaging markers using data from Clalit Health Services
(CHS) database. CHS is the largest of the four health maintenance
organizations in Israel with 14 hospitals in Israel
(inpatients' data) as well as widespread community care.
CHS has approximately 5 million members representing
over half of the Israeli population, making it a good representative
of the Israeli population. The electronic data
include data of continuous real-time inputs from physicians
and health service providers, including medical diagnoses,
laboratory results, and imaging acquisitions. The study was
approved by the Carmel Medical Center Ethics Committee
(study no. 0041-17-CMC). All procedures were carried out
in accordance with relevant guidelines and regulations.
Due to the retrospective observational nature of this study,
informed consent was waived.
Sample
We included any patient aged ⩾18 years with a diagnosis of
LC (The International Classification of Diseases, Ninth
Revision (ICD-9) 162), pancreas cancer (ICD-9 157), colorectal
cancer (ICD-9 153), or breast cancer (ICD-9 174175)
between January 2014 and April 2020, who underwent
brain MRI including DWI sequence for any indication
within 5 years after cancer diagnosis (index event).
Resource constraints did not allow us to review MRI
scans from all cancer patients in our database, and we
had to focus on specific subgroups. LC is associated
with pro-thrombotic complications including increased
CRS incidence and cardiovascular risk factors,16,18,19
while the risk of CRS is high for pancreatic cancer and
intermediate for colon cancer. Finally, breast cancer is
associated with the lowest cancer-associated stroke risk,
although it is still more than the risk in the general
population.20
Variables and groups
Demographic and cardiovascular data were recorded at
the time of the latest MRI scan date. Baseline variables
were defined using the CHS chronic disease registry based
upon ICD-9 codes. Patients were divided into two groups:
LC patients, who undergo brain MRI scans as part of their
routine follow-up21-23 as well as for other reasons, and
non-lung cancer (NLC) control group (pancreas, colorectal,
and breast cancer) who underwent brain MRI scans
for various reasons. For positive lesions, we also looked at
indications for performing imaging. We hypothesized that
LC patients would have more CMI compared with other
cancers types.
Outcomes
The main study outcome was CMI on brain MRI performed
following index event. CMI were defined as small hyperintense
lesions on DWI, with a correlated hypointense lesion
in apparent diffusion coefficient (ADC) sequence.1,24 We
also looked for location (e.g. cortical vs sub-cortical) and
CMI size. In all suspected lesions, we also looked at the
susceptibility-weighted imaging (SWI) sequence to exclude
cerebral microbleeds (CMBs), and T1 with contrast
sequence and fluid-attenuated inversion recovery (FLAIR)
to exclude potential metastases.24,25 In case of uncertainty,
follow-up scan was also reviewed to exclude small metastasis.
If CMI mimics were subsequently found, the lesion
was not classified as CMI.
All MRI scans were reviewed by a group of neurologists
and neuroradiologists who underwent teaching and practical
training. All suspected positive CMI scans underwent
further confirmation by an experienced neuroradiologist
(R.E.) and an experienced vascular neurologist (E.A.). Our
group had excellent inter-rater reliability.26 Intraclass correlation
coefficients (ICCs)27 from two-way analysis of
variance (ANOVA) analysis were 0.89 (95% confidence
interval (CI) = 0.82-0.94) for CMI.
International Journal of Stroke, 19(4)
WSO - April 2024

Table of Contents for the Digital Edition of WSO - April 2024
