Since the publication of NINDS, a large number of registry based observational studies have been used for the weaponization of tPA in the clinical setting. Initially utilized to justify the use of tPA outside the controlled setting of a clinical trial, and later used to support the profitable indication creep we have observed in the intervening 28-years. Each of the publications share an important commonality, the lack of control group. Without a comparator the results can be made to appear however one wishes. The recent publication of the Potential of rtPA for Ischemic Strokes With Mild Symptoms (PRISMS) trial in JAMA by Khatri et al serves as a reminder of the dangers of utilizing such observational data to assess the efficacy and safety of therapeutic interventions.
Prior to the PRISMS trial, much of the previous data justifying the practice of administering tPA to the subset of stroke patients with mild symptoms has been observational in nature. Take for example a trial published in Stroke in 2016 by Romano et al cited as evidence in favor of administering tPA to patients with mild symptoms at presentation1. The authors examined patients from the Get With The Guidelines–Stroke registry (GWTG), a US based stroke registry sponsored by the AHA. The authors examined 42,394 patients who arrived ≤4.5 hours from symptom onset not treated with thrombolytics because of mild or rapidly improving symptoms. The authors noted 27% were unable to ambulate independently at discharge, and 27.2% could not be discharged home. The authors argued that since a subset of patients were not discharged directly home, the entire cohort would benefit from tPA. Of course such a statement is absurd. Overall this cohort did quite well. Mortality was low (0.8), and the majority of patients (73%) were discharged home directly from the hospital. Of course this study is unable to determine whether the administration of tPA will improve these outcomes.
Exposing their biases, the very same authors published a similar analysis in JAMA Neurology in 20152. Like their prior paper, the authors examined the GWTG registry, focusing on the subset of patients with mild strokes who received tPA. Mortality was 1.3%, slightly higher than what was observed in the patients who did not receive tPA. Patient discharged directly home was less than the previous paper at 70.6%. Despite these two cohorts demonstrating almost identical results, the authors concluded:
Given the significant proportion of treated mild stroke and real-world safety and outcomes observed, our results provide reassurance about the safety of IV rtPA in patients with low NIHSS scores.
Simply put, even in these hopelessly flawed observational cohorts one can see that tPA offers no benefit. But since each cohort was presented independent of the other, each with no control group, and with a very clear agenda, the results can be manipulated to suggest the beneficial properties of tPA.
Such manipulations are far more difficult when a control group is present, especially when patients are randomly assigned to receive the treatment in question. In the Khatri et al publication3 (3), the authors enrolled patients 18 years or older presenting with symptoms consistent with a mild stroke, defined as NIHSS score of 0 to 5, and deficits judged to not be “clearly disabling” within 3 hours of symptom onset. Patients were randomized to receive either intravenous alteplase at 0.9mg/kg and placebo oral aspirin or oral aspirin, 325 mg and placebo intravenous alteplase.
Like the majority of the stroke literature the authors selected the proportion of patients alive and independent at 90-days (mRS or 0 or 1) as their primary outcome. The also examined the effect of tPA on the ordinal shift of the mRS score at 90-days, and a global favorable recovery measure, which was the conglomeration of the mRS score, NIHSS, Barthel Index, and Glasgow Outcome Scale. Additional exploratory analyses were conducted examining ambulatory speed, stroke-related quality of life, and general health-related quality of life.
From May 2014 to December 2016, the authors enrolled 313 patients before the trial was stopped well short of the planned 948 patient sample size due to slow enrollment. The authors found no difference in their primary outcome, 78.2% of patients randomized to alteplase and 81.5% of patients randomized to the aspirin group had a favorable neurological outcome at 90-days. Nor did they observe a difference in the ordinal analysis of mRS scores (OR, 0.81; 95% CI, 0.5-1.2) or the rate of global favorable recovery (OR, 0.86; 95% CI, 0.5-1.4). In fact, the only statistically significant difference noted was the rate of symptomatic intracranial hemorrhage (ICH), experienced by 3.2% of the patients in the alteplase group, with no ICH reported in the aspirin group.
Although the results of PRISMS were technically negative, the authors argue that given the early stoppage, they could not definitively exclude a benefit of tPA. While this statement is true in principle, it strikes me as fairly disingenuous. The confidence intervals surrounding the 90-day outcomes are wide (−9.4% to 7.3%), but there is a complete lack of signal to suggest benefit.
Since the earliest clinical endeavors into thrombolytic therapy for acute ischemic stroke, our view of the true efficacy of tPA has been obscured. There has been a systematic effort to accept only data that supports its use and ignore or discredit the data that calls its efficacy into question. The observational trials used to expand tPA’s indications are even more dangerous because they lack a comparator with which to ensure scientific honesty. Without a control group, observational data can be used to inflate efficacy and minimize harms simply through patient selection. The PRISMS trial should remind us of this fundamental concept. But PRISMS skirts around the larger issue, its intent not to identify the subset of stroke patients with the potential to benefit from thrombolytic therapy, but simply to broaden the number of patients eligible to receive tPA. It is yet another study in a systematic effort that has been ongoing for 28 years since the publication of NINDS. The primary goal of the research examining the efficacy of thrombolytics for CVA has not been to determine the underlying truth, but rather to portray the most profitable reality possible.
University of Georgetown
Resuscitation and Critical Care Fellowship Graduate