Imaging Negative Stroke: Diagnoses and Outcomes in IV t-PA Treated Patients

Abstract

Background/Purpose

Intravenous Alteplase (t-PA) improves outcome in patients with acute ischemic stroke. Of those with full recovery, some may not have had ischemia. We analyzed the frequency and post-treatment outcomes of stroke code patients with no imaging evidence of stroke in order to establish the incidence of neuroimaging negative cerebral ischemia (NNCI) and stroke mimics treated with t-PA. In addition, we compared these patients to the group of stroke patients with imaging evidence of acute stroke to determine whether there was a difference in adverse events, and functional outcomes.

Methods

We included all adult stroke patients treated with IV t-PA within 3 hours of stroke onset from the UCSD SPOTRIAS database through January 2013. The IPS group (Imaging Positive Stroke codes) was comprised of patients with neuroimaging evidence of acute ischemic stroke, while the INS group (Imaging Negative Stroke codes) included those patients without neuroimaging evidence of acute cerebral ischemia. All final diagnoses were reviewed by an adjudicating body. We reviewed medical records and neuroimaging; compared discharge diagnosis, 90-day mRS, and incidence of intracranial hemorrhage; and adjusted for age, admission NIHSS, pre-stroke mRS and diabetes in multivariable models.

Results

We identified 106 patients; 74 IPS patients and 32 INS patients, who had similar baseline characteristics, except for baseline NIHSS (IPS 12.9±8.2, INS 8.0±5.6, p=0.002) and incidence of cardiac arrhythmias (IPS 32.4%, INS 12.5%, p=0.034). The diagnoses in the INS group were stroke (23, 72%) – representing NNCI, somatization (6, 19%), tumor (1, 3%), seizure (1, 3%), and migraine (1, 3%). All IPS patients were diagnosed with acute ischemic stroke. Adjusted for age, baseline NIHSS, pre-stroke mRS and diabetes, the INS patients had significantly higher rates (OR 3.04, p=0.036) of good functional outcome (90 day mRS 0–1). ICH was found in 24% of the IPS patients and was symptomatic in 6.8%. None of the INS patients had ICH.

Conclusions

Since the majority of INS patients were found to have neuroimaging negative cerebral ischemia (NNCI), which may represent either TIA or aborted stroke, and there were no intracerebral hemorrhages in the INS group, our data support the safety of administering IV tissue plasminogen activator to all patients in whom acute ischemic stroke is clinically suspected. We have demonstrated that NNCI patients and stroke mimics are common and future larger scale prospective studies are required to delineate the true frequencies of each, and to evaluate differences in outcomes.

Introduction

The diagnosis of ischemic stroke is based on clinical findings. Differentiation between stroke mimics and patients with full recovery after stroke who show no evidence of acute ischemia on neuroimaging remains challenging. Acute imaging criteria are controversial and the routine use of Magnetic Resonance Imaging (MRI) with Diffusion Weighted Imaging (DWI) to prove the existence of an acute stroke has not been widely implemented.

In the absence of widely available and reliable acute surrogate markers, the treatment decision in acute stroke is currently based on the exclusion of advanced ischemia and ICH by plain head Computed Tomography (CT) and clinical evaluation.[1]

Many have reported on functional outcome after ischemic stroke, with or without reperfusion therapies. Outcome data in patients who receive thrombolytics without suffering a stroke is limited. [2] Neuroimaging with brain MRI and head CT is used to support the diagnosis of stroke. It is possible for patients to present with stroke-like symptoms, which are later found to be due to another disease process such as migraine, seizure, or conversion disorder.[3] A recent study showed the safety of administering IV rt-PA within the first 4.5 hours of symptom onset in patients with suspected ischemic stroke even when post-treatment DWI did not demonstrate cerebral infarction.[4] A retrospective study of 512 patients estimates the incidence of stroke mimics at 14% and NNCI at 7%.[5] Another study of 254 patients estimates the rate of nonischemic process (stroke mimics) at 3.5%, and TIA at 9.1%.[2]

We investigated the final diagnosis and 90-day outcome of stroke code patients who were treated with t-PA and who showed no evidence of acute ischemia on neuroimaging. Our analysis reports the incidence of IV t-PA administration to neuroimaging negative stroke code patients as well as the frequency of NNCI patients and stroke mimics within this group. In addition, the outcomes of the INS patients were compared to those of the IPS patients in order to investigate the safety of tPA administration in cases where the final diagnosis is non-ischemic, or where there is not imaging support for the diagnosis of acute cerebral ischemia. This may assist in the planning of future research in acute stroke treatment, as well as resource allocation and risk estimation, especially with regard to t-PA use in patients without ischemia.

Methods

We analyzed all patients with a Stroke Code alert at the UCSD Hospitals who were prospectively entered into the UCSD SPOTRIAS database from June 2004 to January 2013. Inclusion criteria were age >18 years, ED admitting diagnosis of Acute Ischemic Stroke, IV t-PA administration within 3 hours of symptom onset, neuroimaging from hospital admission and 90-day modified Rankin score (mRS) availability. This study was approved by the local IRB. All patients’ medical records and neuroimaging studies were reviewed. An adjudicating group of at least two board certified vascular neurologists reviewed all Stroke Code alerts and determined the final diagnosis.

Good outcome was defined as mRS 0 or 1 at 90 days. We captured the final diagnosis, discharge disposition, frequency of intracranial hemorrhages and 90-day mortality.

The IPS group (Imaging Positive Stroke codes) was comprised of patients with neuroimaging evidence of acute ischemic stroke, while the INS group (Imaging Negative Stroke codes) included those patients without neuroimaging evidence of acute cerebral ischemia. A patient was classified as imaging positive (IPS) if acute stroke findings were documented on pre- or post-treatment head CT or brain MRI. Patients were classified as imaging negative (INS) if no acute stroke was identified and post treatment (within 72 hours) neuroimaging was available and negative for acute ischemia. A board certified vascular neurologist, who was blinded to the clinical outcome, reviewed the neuroimaging studies.

Statistical Analysis

Group comparisons were made with the Wilcoxon Rank-sum for continuous variables and Fisher’s Exact test for categorical variables. Multivariable logistic regression models were used to adjust for age, baseline NIHSS, pre-stroke mRS and history of diabetes as covariates to test the association of the outcomes between groups.

Results

We identified 74 IPS and 32 INS patients. The baseline characteristics were similar between the two groups, except for incidence of cardiac arrhythmias (IPS 32.4%, INS 12.5%, p=0.034) and baseline NIHSS (IPS 12.9±8.2, INS 8.0±5.6, p=0.002) (Table 1).

Table 1.

Baseline Characteristics

Baseline Variable	IPS (n=74)	INS (n=32)	p-value
Age (years) (mean ± SD)	67.0 (±17.8)	62.7 (±18.2)	0.262
Female (%)	41.9	46.9	0.673
Baseline NIHSS	12.9 (±8.2)	8.0 (±7.8)	0.002*
Pre-Stroke mRS (0 or 1) (%)	77.0	84.4	0.447
History of Coronary artery disease (%)	17.6	25.0	0.43
History of Myocardial infarction (%)	12.2	15.6	0.756
Prior Stroke or TIA (%)	18.9	31.3	0.207
Atrial fibrillation (%)	32.4	18.8	0.168
Diabetes mellitus (%)	13.5	18.8	0.558
Hypertension (%)	64.9	46.9	0.091
Cardiac Arrhythmias (%)	32.4	12.5	0.034*
Time from symptom onset to IV t-PA (minutes), mean (SD)	127.5 (44.6)	136.6 (40.5)	0.348
Time from symptom onset to CT (minutes), mean (SD)	87.5 (51.1)	83.8 (54.3)	0.983

^*p-value <0.05

In the unadjusted analysis, 90-day mortality was less frequent in the INS group (9.38% vs. 27.03%, OR 0.28, p=0.07), 90-day mRS of 0–1 was more frequent in the INS group (68.75% vs. 32.43%, OR 4.51, p=0.001), and good 90-day outcome (measured by discharge disposition) was more frequent in the INS group (84.38% vs. 64.86%, OR 2.90, p=0.062). Adjusted for age, admission NIHSS, pre-stroke mRS and diabetes, the differences in mortality (OR 0.33, p=0.156) and discharge disposition (OR 1.87, p=0.331) were no longer statistically significant. However, the 90 day mRS (0–1) was significantly more frequent in the INS group (OR 3.04, p=0.036).

The analyses of final diagnosis, cerebral hemorrhage, and symptomatic cerebral hemorrhage were not performed with multivariate modeling because of the small number of events. In the IPS group, symptomatic intracranial hemorrhage (sICH) was found in 6.8%, any ICH in 24.3%. None of the INS patients had ICH.

All IPS patients had the final diagnosis of ischemic stroke. In the INS group, a final diagnosis of stroke was given to 23 of 32 patients (72%), representing NNCI, or neuroimaging negative cerebral ischemia. This may represent either TIA or stroke which resolved completely with tPA.

The remainder of the INS group had diagnoses of somatization (6, 19%), tumor (1, 3%), seizure (1, 3%), and migraine (1, 3%). Overall, of the 104 t-PA treated patients, 8.5% were stroke mimics and 21.7% had NNCI. Table 2 demonstrates the average NIHSS and onset-to-treatment (OTT) times for each final diagnosis. The average OTT time for NNCI was 143 minutes.

Table 2.

Final Diagnoses, NIHSS, Onset-to-Treatment times

Group	Diagnosis	Number (%)	NIHSS (SD)	Onst-to-Treatment time (SD)
IPS	Stroke	74 (100)	12.85 (8.18)	127.5 (44.58)
INS	Neuroimaging Negative Cerebral Ischemia	23 (71.88)	6.91 (5.44)	143.36 (43.47)
INS	Somatization	6 (18.75)	10.83 (4.67)	124.17 (33.15)
INS	Migraine	1 (3.12)	6	120
INS	Tumor	1 (3.12)	19	105
INS	Seizure	1 (3.12)	6	110

Discussion

This study investigated the rate of imaging-negative stroke code patients within the t-PA treated population as well as these patients’ discharge diagnoses and long term outcomes. The rate of stroke mimics among t-PA treated patients was 8.5%, and the rate of NNCI was 21.7%. This is higher than anticipated, and higher than reported in other studies [2, 5] which may be explained by the tendency of our team to treat aggressively with t-PA, base treatment largely on clinical symptoms, and not use a low NIHSS to exclude from therapy.

The only significant differences in patient characteristics were the incidence of cardiac arrhythmias and the baseline NIHSS, which were higher in the IPS group. Cardiac arrhythmia is a risk factor for stroke and patients with higher NIHSS are more likely to show early ischemic signs on neuroimaging.[6, 7] Furthermore, patients with somatization, the most common non-stroke diagnosis in our study, are less likely to have facial weakness, aphasia, neglect, or visual field deficits,[8] corresponding to a lower NIHSS than patients with true ischemic stroke.

A recent study suggests the safety of administering IV tissue plasminogen activator to patients with suspected acute cerebral ischemia within 3 hours of symptom onset, even when the diagnosis ultimately is found not to be stroke or imaging does not show an infarct.[5] One analysis suggests that based on availability and sophistication of current diagnostic tools for acute ischemic stroke, the upper limit for an acceptable rate of treatment of neurovascular mimics should be <3% at centers using non-contrast CT alone and <1% at centers using multimodal imaging.[3] A prior study at our institution identified the incidence of non-ischemic stroke mimics among stroke codes called;[9] however, data on rates of t-PA treatment of stroke mimics is less complete. In addition, studies which do analyze the subset of treated patients do not often include post-discharge outcome data.

Analyses which take into account discharge diagnoses solely based on discharge summaries may miss important distinctions, as coded diagnoses may be chosen for billing purposes. Our patients were prospectively entered in the SPOTRIAS database and the diagnosis was adjudicated by at least 2 board certified vascular neurologists. Utilizing this review system, we identified 72% of INS patients with the clinical diagnosis of ischemic stroke (NNCI). The rest of the INS patients were stroke mimics, with somatization being the next most common diagnosis in this patient group. These findings underscore that while imaging plays an important role as a surrogate marker in determining the patient’s diagnosis, a detailed clinical evaluation is essential in the correct treatment of acute ischemic stroke.

There was no ICH in the INS group. We considered that this may have been in part due to a limitation in our diagnostic criteria, in that a patient with ICH on post t-PA neuroimaging would have been classified as imaging positive due to the hemorrhage. To address this, we reevaluated the records of all IPS patients with ICH and found early ischemic changes on initial CT or clinical findings making acute ischemia highly probable in those patients. The lack of ICH in the INS group supports the safety of tPA administration to patients with clinical suspicion of stroke, regardless of final discharge diagnosis or lack of imaging evidence of ischemia.

Limitations of our study include the relatively small sample size from a single center. This may have limited the variability of stroke mimic etiologies, such as hypoglycemia, vertigo, and head trauma. The adjudicating body was comprised of a limited number of vascular neurologists, who may have had a bias towards making the diagnosis of acute ischemic stroke as they were aware that the patients had received t-PA.

We included only patients who had a final outcome (death or 90-day mRS) available. This may have biased the study to include younger patients, who were more likely to follow-up in 90 days, and exclude patients who could not consent to follow-up evaluations. This may have induced bias against patients with language impairment and those for who surrogate consent could not be obtained.

Most TIA patients clinically improve within the first hour of onset of symptoms. This study is limited in that we analyzed only the initial NIHSS score, without information about symptom evolution. While it may be postulated that some patients with somatization were misclassified as NNCI, the central adjudication reviewed all aspects of the case including the evolution of symptoms and can be considered a gold standard of clinical diagnosis. In addition, if patients were rapidly improving, they would have been less likely to be treated based on the relative tPA exclusion criteria. Overall, since the majority of INS patients were found to have NNCI, which may represent either TIA or aborted stroke, and there were no intracerebral hemorrhages in the INS group, our data support the safety of administering IV tissue plasminogen activator to patients in whom acute ischemic stroke is suspected, despite the possibility that the patients may ultimately be found to be stroke mimics or have NNCI. In fact, these tPA-treated INS patients had better adjusted mRS scores than their IPS counterparts. We have demonstrated that NNCI patients and stroke mimics are common and future larger scale prospective studies are required to delineate the true frequencies of each, and to evaluate differences in outcomes.