American Heart Association

diagnosis and imaging

Perfusion MRI in Perinatal Stroke

Russell Mitesh Cerejo, MD

Watson CG, Dehaes M, Gagoski BA, Grant PE, Rivkin MJ. Arterial Spin Labeling Perfusion Magnetic Resonance Imaging Performed in Acute Perinatal Stroke Reveals Hyperperfusion Associated With Ischemic Injury. Stroke. 2016

In their paper, the authors described a novel method to assess perfusion in perinatal strokes in the ischemic as well as penumbra tissue. 
They included subjects less than 28 days old at time of diagnosis, and included both arterial and venous infarcts. MRI was carried out on 3T scanners with arterial spin labeling (ASL) techniques used for non-contrast perfusion imaging. Out of 25 neonates that participated, 16 were males (64%). Median gestational age at birth was 38.7 weeks (range: 35.7-41.9), median (estimated) age at stroke was 1 day (i.e., second day of life) (range: 0-8), and median age at MRI was 3 days (range: 0-16). The median time from symptom onset to MRI acquisition was 2 days (range: 0-8).

Arterial ischemic stroke was present in 11 (44%), while venous infarction was found in 9 (36%). Five patients (20%) had both arterial and venous stroke. Hyperperfusion was seen in 73% of arterial ischemic strokes, 11% with venous stroke, and 80% with both. Hypoperfusion was observed in 33% with venous and none with arterial stroke. Perfusion was normal in 45% with venous and 20% with both. In nearly all patients presenting with clinical or electrographic seizures, EEG abnormalities were present in the same hemisphere as the stroke; this clinical feature did not differ by stroke type.

This study demonstrates that perfusion imaging can be obtained in neonates with acute stroke, and often reveals hyperperfusion in the infarct core. Penumbra in arterial infarcts is seldom found. Hyperperfusion may be due to post-stroke reperfusion or to neuronal hyperexcitability of stroke-associated seizure.

FLAIR Vascular Hyperintensities in Bordezone Strokes

Allison E. Arch, MD

Kim and colleagues investigated the clinical significance of FLAIR vascular hyperintensities in watershed strokes, and they tried to predict poor prognosis using these FLAIR changes as a marker of impaired hemodynamics.

Watershed, or borderzone, strokes represent 10% of all ischemic infarcts. The authors of this study defined 2 types of borderzone strokes: internal borderzone infarcts (IBZ), which are lesions between the deep and superficial perforating arterial territories of the MCA, and cortical borderzone infarcts (CBZ), which are between the MCA/ACA or the MCA/PCA territories. A patient was then considered FLAIR-positive he had 2 or more FLAIR vascular hyperintensities in his MCA territory on MRI, which were thought to have occurred prior to the stroke.

Eighty-seven consecutive patients with acute borderzone strokes were identified, 62 with CBZ and 55 with IBZ. Thirty of all included stroke patients were considered FLAIR-positive. The authors found that FLAIR vascular hyperintensities were associated with a more severe clinical presentation and a poorer clinical prognosis in patients with CBZ strokes, but not in patients with IBZ strokes. They concluded the presence of FLAIR vascular hyperintensities, “may help to identify CBZ-infarcted patients who require close observation and hemodynamic control.”
Their findings are interesting. The authors noted that the presumed pathogenesis of watershed strokes is microembolization in combination with hemodynamic disturbance. However, in patients with FLAIR vascular hyperintensities on MRI, there may be an additional hemodynamic-compromised insult during the stroke, which then leads to poorer outcomes. Kim and colleagues pointed out that in the CBZ group, those who had FLAIR vascular hyperintensities had similar sized DWI lesions to those patients who did not have FLAIR lesions. However, there were significant perfusion differences between the FLAIR-positive and FLAIR-negative groups, lending support to the concept that FLAIR vascular hyperintensities on MRI may signify that the patient is more influenced by hemodynamic instability than his FLAIR-negative counterpart.

It is unclear why this would be on the case in CBZ strokes but not in IBZ strokes. Further investigations are needed to help elucidate the importance of hemodynamics in borderzone strokes. 

Genetic Factors that Impact White Matter Hyperintensities Increase Risk of Lacunar Stroke

Jay Shah, MD

Traylor M, Rutten-Jacobs LCA,Thijs V, Holliday EG, Levi C, Bevan S, et al. Genetic Associations With White Matter Hyperintensities Confer Risk of Lacunar Stroke. Stroke. 2016

Small vessel disease (SVD) can lead to various pathologies including lacunar infarcts, hemorrhage and microbleeds but the underlying pathophysiological mechanism remains unknown. White matter hyperintensities (WMH) are increased in lacunar stroke suggesting a shared pathological mechanism. Furthermore, WMH and lacunar infarcts co-exist in patients with inherited forms of SVD such as cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL). Genome-wide association studies (GWAS) have identified multiple genetic variants associated with WMH. In this study, the authors evaluated the impact of common genetic variants associated with WMH on the risk of lacunar stroke in patients with lacunar strokes and controls.
The authors used a genetic risk score approach to determine if SNPs associated with WMH were associated with lacunar strokes along with cardioembolic and large vessel infarcts. Each subgroup included approximately 1300 patients and 9000 controls. Secondly, lacunar strokes were separated into WMH (n = 568) and without WMH (n=787) to test for association. WMH genetic risk score was associated with lacunar stroke in patients regardless of WMH status but not with cardioembolic or large vessel strokes. However, none of the WMH-associated SNPs met significance for association with lacunar stroke.

This study supports the known belief that features of cerebral SVD share pathophysiology. Interestingly, the risk of lacunar stroke remained in patients without significant WMH. This could potentially be to the effect of “time”, in that patients without WMH could possibly be younger and have not had accumulated SVD damage. Another possibility is that patients without WMH have an unknown protective mechanism that protects against WMH but not lacunar stroke. The latter would be interesting in that it would suggest differing pathological mechanism. The study had large number of patients and controls but a potential confounding variable is that control patients did not have MRI images raising the possibility that controls had lacunar strokes and/or WMH as such pathology can potentially be clinically “silent”. Nonetheless, this study highlights a shared pathophysiological process that underlies various manifestations of CVD. 

What Makes a Lacune?

Peggy Nguyen, MD

The lacune, often used interchangeably with the definition of a stroke of small vessel atherosclerotic etiology, is traditionally based on a size definition of no greater than 15 mm. It is a classic feature of cerebral small vessel disease. However, despite its prolific use in the stroke literature, the exact characteristics and morphological features of a lacune are not well defined. Here, the authors analyzed the shape of incident lacunes in CADASIL, a genetically inherited small vessel arteriopathy, to better define the lacune’s morphological features.

Fifty-seven CADASIL patients with incident lacunes were included in the study, encompassing 88 incident lacunes, only 18 of which were associated with symptoms. The most common locations for lacunes were in the centrum semiovale (n=30) and the basal ganglia (n=27). In spectral shape analysis, elongation and planarity were found to be the primary determinants of lacune shape and tended to align along perforating arteries. Although 15 mm is traditionally used as the upper size limit of a lacune, about 10% of lacunes, particularly when evaluated in planes other than axial, exceeded this size, whereas only 1 lacune was larger than 15 mm in the axial plane.

Not all lacunes are created the same, but there are certainly similarities, and this may have to do with the mechanisms by which they develop. The findings in this study confirm some generalizations of lacunes, such as the common locations, but also refutes some others, for instance, the size of lacunes, particularly when viewed in non-axial planes. These findings are also suggestive of a mechanism in which lacunes of chronic small vessel diseases develop secondary to factors related to vascular anatomy, rather than tract degeneration.

Collateral Circulation Status as Assessed by MR-Perfusion Modulates Relationship Between Time and Development of FLAIR Signal

The exact time of symptom onset cannot be determined for up to a fourth of acute stroke patients. In such cases, clinicians often use magnetic resonance imaging (MRI), in particular a difference of signal intensities between diffusion-weighted imaging (DWI) and fluid-attenuated inversion recovery (FLAIR) sequences to approximate the age of a lesion.  Cytotoxic edema appears minutes after stroke onset and can be visualized using DWI.  Comparing this to FLAIR, which can demonstrate the vasogenic edema that appears hours after stroke onset, the so-called DWI/FLAIR mismatch has been proposed as a predictor with regards to the 4.5h window for intravenous thrombolysis. However, the sensitivity of DWI/FLAIR mismatch between studies. These variations are in part explained by imaging techniques, but it is likely that pathophysiological variables such as collateral circulation significantly contribute as well. Wouters et al. sought to investigate the association between timing of DWI/FLAIR mismatch and collateral circulation using hypoperfusion intensity ratio (HIR), a measurement of perfusion weighted imaging (PWI) severity that has been shown to be a good predictor for poor collaterals.

The authors utilized clinical and neuroimaging data from the AXIS 2 trial, a multicenter Phase IIb placebo-controlled, randomized and double blinded trial investigating recombinant Granulocyte Colony Stimulating Factor in acute stroke. A total of 141 patients were included for analysis, excluding patients with incomplete imaging sequences, severe FLAIR lesions overlapping the acute lesion or in the contralateral hemisphere (as the contralateral hemisphere was used for FLAIR intensity measurement), or reperfused core. Quantitative relative FLAIR maps (rFLAIR) were calculated in a voxel-based manner using in house software. Collateral status assessed by HIR was dichotomized into “good” (n= 87, 61.7%) and “poor” (n= 54, 38.3%). Patients with poor collaterals had more severe stroke symptoms at baseline (NIHSS 14 vs NIHSS 11, p= 0.01), larger DWI lesion volumes (47.2 mL vs 14.6 mL, p= <0.01), and larger TMax > 6s perfusion volumes (91.5 mL vs 45.8 mL, p=0.01). 

The predictive value of time for rFLAIR intensity was moderate in patients with poor collateral circulation (R2 = 0.28), but poor in patients with good collateral circulation (R2 = 0.03). The relationship between time from onset to rFLAIR signal intensity was stronger in patients with poor collaterals compared to those with good collaterals (p for interaction = 0.04). In addition, a strong interaction between increased Tmax in the region of perfusion deficit (a measurement of hypoperfusion severity) on the association between time and rFLAIR intensity was identified (p=0.001).

This study reinforces the concept that the development of DWI/FLAIR mismatch as a marker for ischemia is dependent on the severity of hypoperfusion. The authors’ omission of digital subtraction angiography (DSA) in the study precludes a definitive association with collateral status, but as the authors note, the HIR technique that was utilized has shown to have good correlation with DSA-assessed collateral circulation. These findings also provide valuable insight into the relationship between collateral status and the timing of DWI/FLAIR mismatch which will prove useful for clinical trials using this assessment as a substitute for last known well in cases where this is unable to be determined from history. Based on these results, patients with good collateral circulation may be misidentified as having lesions <4.5h old. Whether these patients would receive the same benefit from intravenous thrombolysis as compared to patients with poor collaterals and younger lesions presents an interesting avenue for future research, as DWI/FLAIR mismatch may emerge as a viable alternative to time of last known well in determining progression of ischemia in the context of thrombolysis eligibility. 

The Leakage Sign: A New Predictor for Hematoma Expansion in ICH

Alexander E. Merkler, MD

Orito K, Hirohata M, Nakamura Y, Takeshige N, Aoki T, Hattori G, et al. Leakage Sign for Primary Intracerebral Hemorrhage: A Novel Predictor of Hematoma Growth. Stroke. 2016

Intracerebral hemorrhage (ICH) is a devastating disease with a one-month mortality of 40%. Although initial ICH volume is the strongest predictor of mortality, hematoma expansion is a potentially modifiable risk factor that correlates well with both functional outcome and death and occurs in up to 40% of patients with ICH. Research has therefore been focused on methods to 1) identify patients at risk of hematoma expansion and 2) reduce hematoma expansion.

The “spot sign” has been previously correlated with both hematoma expansion and poor functional outcome; however, the spot sign is not a perfect predictor of hematoma expansion. Although the specificity of the spot sign is high, the sensitivity is only around 50%. In this study, Drs. Orito and Morioka et al evaluate a new predictor for hematoma expansion in ICH: the “leakage sign.”

The authors evaluate 80 patients with a primary ICH who underwent a CTA and a second CT (delayed phase image) 5 minutes later. The leakage sign was determined by comparing the arterial and delayed phase CT images. Each neuroradiologist measured a region of interest (ROI) of 1-cm diameter on the delayed phase images. This region was considered the highest change in Hounsfield Units between the delayed and arterial phase images and represented the leakage of contrast medium into the hematoma. The same ROI circle was then drawn on the arterial phase CT image and the Hounsfield Units were measured. A change >10% in Hounsfield Units was considered a positive leakage sign, or hematoma expansion. A follow-up CT was performed 24 hours later where hematoma expansion was defined as >10% change in hematoma volume from the initial CT.

The authors found that the spot sign was positive in 18 (22%) of patients and that the leakage sign was positive in 35 (43%) of patients. 33 of the 35 patients with a positive leakage sign also had positive spot signs. Overall, leakage sign had a higher sensitivity and specificity than the spot sign: 93.3% and 88.9%, respectively, versus 77.8% and 73.8%. In addition, the leakage sign proved to be a better predictor of outcomes than the spot sign. Patients with a positive leakage sign had significantly poorer outcomes (20.0% versus 51.5%), but outcomes were unaffected in patients with a positive spot sign.

In conclusion, the leakage sign appears to be both an easy and reliable method to predict hematoma expansion in patients with ICH.

Missed Strokes in the ED: An Opportunity for Systems Change

Ilana Spokoyny, MD
Arch AE, Weisman DC, Coca S, Nystrom KV, Wira III CR, Schindler JL. Missed Ischemic Stroke Diagnosis in the Emergency Department by Emergency Medicine and Neurology Services. Stroke. 2016

Ischemic stroke presentations can vary significantly and some presentations are more likely to be overlooked or thought to be stroke mimics. Additionally, some patient populations (such as young patients, women, and minorities) are more likely to be attributed nonstroke etiologies. The danger is, of course, missed treatment opportunities. However, patients with missed strokes are also less likely to receive appropriate monitoring for neurological progression or stroke-related complications.

The authors of this study performed a retrospective chart review to determine the rates of diagnosis and misdiagnosis of stroke patients at one academic hospital and one community hospital. Patients were identified by discharge billing code. TIAs were excluded, as were imaging-negative strokes. A stroke was “missed” if practitioners in the Emergency Department (ED) did not initially consider stroke in the differential or the diagnosis was delayed, causing the patient to miss the therapeutic window for thrombolytic therapy. A stroke was also considered “missed” if ED physicians consulted neurology for a possible stroke diagnosis and the neurology consultant felt that the patient did not have a stroke and admitted the patient to a medicine service. 

A third of missed cases presented within a 3-hour time window, and an additional 11% presented within 3 to 6 hours. Of all missed cases of ischemic stroke at the academic hospital, 20/55 (35%) were seen by neurology in the ED but early diagnosis was still missed. Nine (45%) of these cases missed by neurology presented within the time window for rt-PA, and an additional 3 (15%) presented within 6 hours. Only 8% of missed stroke patients were triaged in the ED as stroke codes. Comparing these patients to those with accurate diagnoses, of whom 46% presented as stroke codes (p<0.001), the 46% seems low. However, stroke codes are not routinely called at this institution for patients who are out of the time window for intervention or clinical trial, and patients where acute cerebrovascular event is not part of the initial differential diagnosis. Forty percent of missed strokes patients did not have neurological examinations with elements of the NIHSS, compared with 8% of the accurately diagnosed stroke patients (p<0.001). Patients with nausea/vomiting, dizziness, and prior strokes were more likely to be misdiagnosed. Patients with focal weakness, vision changes, gaze preference, and dysarthria were more likely to be correctly diagnosed. More posterior strokes were initially misdiagnosed than anterior strokes. There did not appear to be a difference in the rates of misdiagnosis between an academic and community hospitals. For readmission rates, 33% of misdiagnosed patients were readmitted at 60 days post-discharge, compared with 17% of accurately diagnosed patients (p=0.012). 

The study population consisted of predominantly white, older adults who presented to an ED at a primary stroke center. A larger scale analysis would be important to perform, to determine if the rates of misdiagnosis vary among race/ethnicity. This study did not include truly misdiagnosed patients (those presenting with symptoms, who had a stroke that was never diagnosed, and who did not have brain imaging). It would be interesting to evaluate patients diagnosed with stroke who have had ED visits within the week prior to admission (for similar symptoms/presentation), to identify characteristics of truly misdiagnosed strokes. 

There are several factors identified in this study which can lead to systemic changes in the stroke triage process. First, there is a large discrepancy in the frequency of stroke codes called when comparing those with correct diagnoses to those who were missed. A balance must be found between high sensitivity and high specificity. Limiting the stroke code activation to those within the treatment window may be problematic. We know from other studies that the last known well time is often incorrectly identified in the ED, and that patients are only later refined into or out of the treatment window and their treatment options may change. Additionally, as the endovascular window extends, more patients may benefit from having acute neurologic evaluation in the ED, in the form of a stroke code. So, calling stroke codes on anyone with neurologic symptoms within a longer period of time, such as 12 hours, would greatly improve sensitivity. However, this comes at a resource costspecifically, neurology consultant time, CT scanner immediate availability, and pharmacy time spent at bedside. An analysis could be performed to determine how many additional stroke codes would have been activated with these broader criteria, and the resource cost of this, and compare this to how many additional patients would have been correctly identified and treated promptly. Overall, this was an eye-opening study on how we need to do better in the correct identification of our stroke patients.

Streamlined Hyperacute MRI Identifies tPA Eligible Stroke Patients Among Stroke Mimics

Peggy Nguyen, MD

Goyal MS, Hoff BG, Williams J, Khoury N, Wiesehan R, Heitsch L, et al. Streamlined Hyperacute Magnetic Resonance Imaging Protocol Identifies Tissue-Type Plasminogen Activator–Eligible Stroke Patients When Clinical Impression Is Stroke Mimic. Stroke. 2016

Despite advances in imaging, the radiologic component of the tPA decision-making is predicated on a non-contrast CT head, guided by the clinical history and exam. Sometimes, however, the clinical exam or history can be confusing and the CT scan does not provide much additional diagnostic data; stroke mimics make up anywhere between 1-16% of the patients presenting with stroke-like symptoms at large institutions. The use of a hyperacute MRI (hMRI) can help differentiate strokes from stroke mimics, and potentially minimize tPA given to mimics and, perhaps more importantly, ensure that tPA is not withheld from patients who are suspected to be mimics, but are actually strokes.

Here, the authors report an institution-specific streamlined hMRI protocol in the setting of acute stroke. The hMRI protocol described here provides DWI/ADC, FLAIR, and T2*GRE sequences in just under 6 minutes. In order to avoid overutilization, physicians were instructed to order the hMRI only when the initial diagnostic impression was likely stroke mimic, but ischemic stroke could not be entirely ruled out and, if MRI was not available at their institution, the physician would not give the patient tPA. 57 patients, identified as stroke mimics, underwent the hMRI protocol, with 11 having the final diagnosis of stroke, 4 with the final diagnosis of TIA, and the remaining diagnoses being conversion disorder, seizure, complicated migraine, and other. Seven of the 11 stroke patients received IV tPA. There were no differences in door-to-needle, onset-to-needle, or door-to-arrival times for all IV tPA treated patients pre- and post-hMRI; however, the door-to-needle time for tPA treated patients screened with CT alone were significantly shorter than the 7 tPA patients screened with hMRI (37 minutes vs 112 minutes).

Although the overall metrics (door to needle, onset to needle, etc) did not change much with the institution of hMRI protocol, likely due to the minority of patients who went on to receive tPA under the protocol, using the hMRI protocol did lead to substantially longer door-to-needle times for patients who received tPA. However, longer door-to-needle times are preferable than withholding tPA, and it is probable that these patients, having been initially identified as stroke mimics, would not have received tPA otherwise. The use of a hMRI does have its limitations, given it is not widely available and many institutions may not have the resources to staff it emergently, but in institutions where the resources are available, it could potentially increase tPA usage to patients with strokes and decrease tPA usage to patients without strokes.

Small Vessel Disease is Associated With a Worse Outcome After Intracranial Hemorrhage

Neal S. Parikh, MD

Question & Rationale
Small vessel disease (SVD), as reflected by white matter lesions (WML), brain atrophy and lacunar strokes, is associated with hypertension and possibly abnormal autoregulation. The INTERACT 2 investigators therefore hypothesized that SVD is associated with a poor outcome after intracranial hemorrhage (ICH) and investigated whether this association is mediated by intensive blood pressure lowering in ICH treatment.
The INTERACT2 cohort was used for this study. In this international, multicenter, open, blinded endpoint, randomized trial, patients with spontaneous ICH were randomized within 6 hours to SBP<140 versus SBP<180. The cohort did not include patients with pre-existing advanced dementia or disability.

Exposures, Outcomes, and Covariates

There were three measures of SVD on baseline CT: WML, atrophy, and lacunar strokes. On baseline CT, WML were graded by the van Swieten scale.  Atrophy was measured by linear measurements (frontal ratio, third ventricle Sylvian fissure distance) and visual inspection. Lacunar stroke was defined as a round/ovoid cavity of 3-15 millimeters in diameter. The outcome was defined as 90 days death or major disability. Interaction term analysis was used to determine effect of intensive BP lowering on outcome. 

Covariates were age, gender, location, history of ischemic stroke, hypertension, diabetes, use of anti-thrombotic and lipid lowering agents, onset to randomization time, systolic BP, NIHSS, volume/location of hematoma, and intraventricular extension.

In crude and adjusted regression analyses, measures of WML and atrophy were associated with death or major disability. Lacunar stroke was not associated with the outcome in crude or multivariate models. High intensity BP lowering did not result in excess poor outcomes in patients with evidence of SVD.

There was only fair-moderate intra-class correlation for WML, brain atrophy, and especially lacunes. Additionally, the study was not pre-specified and therefore subject to type I error. Last, pre-ICH cognitive and physical disability were not rigorously assessed; however, this is compatible with clinical practice – it is often not possible to thoroughly assess a patient’s pre-ICH function at the time of ICH.

Pre-existing SVD burden may be associated with poor outcome after ICH, and this relationship is not mediated by intensity of blood pressure control. Therefore, these data allay fears regarding intensive BP management after ICH in patients with sequela of chronic hypertension such as SVD. 

Meta-Analysis Concludes Utility of CTA Spot Sign is Dependent Upon Timing and is Not Sufficient to Predict Hematoma Expansion in Acute ICH

Danny R. Rose, Jr., MD

Dowlatshahi D, Brouwers HB, Demchuk AM, Hill MD, Aviv RI, Ufholz L-A, et al. Predicting Intracerebral Hemorrhage Growth With the Spot Sign: The Effect of Onset-to-Scan Time. Stroke. 2016

Intracerebral hemorrhage (ICH) causes a significant amount of stroke-related morbidity and mortality. Of the various prognostic factors in ICH, hematoma expansion is one of the few potentially modifiable ones and as such has been a topic of increasing research. Unfortunately, large-scale randomized controlled trials aimed at preventing hematoma expansion have not shown robust results, possibly owing to the limited ability of clinicians to predict which patients are at greatest risk. One of the more promising diagnostic features in identifying such patients is the “spot sign” of contrast extravasation in the hematoma bed of ICH patients undergoing CT angiography (CTA). However, the predictive value of the spot sign has differed widely across studies, which may reflect variability in delay between ictus and CTA acquisition. Dowlatshahi et al. sought to examine the predictive value of the spot sign in relationship to onset-to-CTA times in patients with acute ICH by conducting a systematic review and patient-level meta-analysis.

The authors found eight studies in which patient-level data was able to be obtained regarding spot sign status, baseline and follow-up hematoma volumes and time from onset to presentation and CTA. Onset-to-CTA time was categorized a priori into five strata: <120 minutes, 120-239 minutes, 240-359 minutes, 360-479 minutes, and >480 minutes.  Hematoma size was measured by computer-assisted planimetry in three studies and the ABC/2 in five studies. A total of 1343 patients were evaluated across the eight studies, with 1176 having baseline hematoma volumes, spot sign status and time-to-CTA, 1039 of whom had follow-up hematoma volume measurements. 
There was significant heterogeneity of spot sign frequency among the studies (I2 = 20.67, p=0.004), due to heterogeneity in the 0-2h (I2 = 15.5, p=0.016) and the 2-4h stratas (I2 = 16.5, p=0.011). There was no heterogeneity for the later time strata. The frequency of the spot sign was 26% for the group as a whole with a significant relationship with onset-to-CTA time strata (p<0.001), decreasing from 39% within two hours of onset to 13% after eight hours. There was no heterogeneity in hematoma expansion between studies (F=0.45, p=0.87) or time strata (F=0.75, p=0.56). Across all time intervals, the median volume of hematoma expansion was greater for spot positive as compared to spot negative patients, and there was no change in median hematoma expansion by time-strata in spot positive patients (F=1.28, p=0.14). Spot positivity and significant hematoma expansion (defined as 6mL or ≥33%) as onset-to-CTA time strata increased. Sensitivity and PPV of the spot sign for hematoma expansion was greatest in the earlier time strata, whereas the specificity and negative predictive value (NPV) of spot sign increased with time. 

This study suggests that the utility of spot sign assessment may be dichotomized depending on onset-to-CTA time. The earlier time strata had the highest sensitivity and PPV which both decreased with time, and the later time strata suggested a high specificity and NPV. Identifying patients at risk for hematoma expansion as well as patients who likely have stable hematomas are both of value to clinicians, and understanding this information in the context of CTA timing suggests clinically relevant information can be gleaned regardless of the time of onset-to-CTA. Even in the acute setting, however, the best sensitivity to detect hematoma expansion was only 60%. This suggests that the spot sign alone is insufficient to identify at-risk patients and should be incorporated into future studies designed to identify other potential predictive factors. As the authors were unable to include any studies published after 2013 due to the time needed to acquire regulatory approvals for patient-level data, analysis of more recent studies would be helpful to validate these results.