Abstract
The recent results of the DIAS II trial have thrown the current strategies for translational stroke research into sharp relief (1). The general approach being used is to extend the time window for therapy beyond the relatively short 3–h time limit designated by the original NINDS trial (2). This is being done using both the large mega-trial strategy (IST 3) (3) and the smaller penumbral selection phase II trials (1, 4) (DIAS II, EPITHET). The former is only in its early phases but the latter are well advanced with the results in the process of being analyzed and released. Given that the results of the DIAS II trial were surprisingly negative, it is worth examining why and re-evaluating the penumbral approach generally (1).
The desmoteplase in acute stroke program (DIAS) was based on the development of a novel plasminogen activator, which had the favorable features of high fibrin specificity, long half life and lack of neurotoxicity or beta amyloid activation in animal models of stroke (5). The general principle of the series of phase II trials conducted with desmoteplase was based on the understanding that the ischemic penumbra may exist for periods beyond 3 h in individual patients and that these could be selected based on MR PWI/DWI mismatch. The initial DIAS trial included a total of 104 patients (27 placebo) in a dose-escalation study between 3 and 9h post stroke and showed that reperfusion rates were significantly higher (71·4%) with desmoteplase 125 μg/kg compared with 19·2% with placebo (5). Early reperfusion correlated favorably with clinical outcome. The results of DIAS were mimicked by the DEDAS trial with much the same protocol and included 37 patients (8 placebo). With 125 μg/kg desmoteplase reperfusion was seen in 53·3% of cases compared with 37·5% in placebo (NS). In both DIAS and DEDAS symptomatic hemorrhage rates were extremely low when doses of 125 μg/kg or less were used (5).
Although using a different thrombolytic agent (tPA), the nonrandomized DEFUSE study was informative because the results were analyzed to determine whether the presence of significant MR PWI/DWI mismatch within 3·6 h after stroke onset predicted favorable clinical outcomes (6). The investigators confirmed a strong relationship between the early reperfusion and good clinical outcome among 80 patients randomized. This built on previous work where it had been clearly established that infarct growth on DWI was strongly correlated with clinical outcome as was recanalization and/or reperfusion (7, 8). In other words, the penumbral hypothesis has been substantiated in a number of different studies using MR DWI/PWI mismatch as a penumbral marker.
In DIAS II, a prospective double-blind single-bolus design was used with desmoteplase 90 and 125 μg/kg, again with a 3–9–h time window. However, on this occasion clinical endpoints were to be the primary outcome measures (1). On intention-to-treat analysis, no significant difference in clinical outcomes was seen among 186 patients (63 placebo). Furthermore, all-cause mortality at 90 days was increased in the 125 μ/kg group (21%) compared with the 90 μg/kg group (5·3%) and placebo group (6·3%). Again the symptomatic intracerebral hemorrhage rate was low at 4·5% in the 125 μg/kg group.
The reasons for these negative results are, as yet, unclear. Obviously, one possibility is that desmoteplase is unsuitable for use as a thrombolytic agent in humans. However, this would fly in the face of the promising early data from the DIAS and DEDAS trials and our understanding of the action of thrombolytic agents generally (5, 9). There are a number of possible explanations for the DIAS II results which deserve consideration. First, the initial entry criteria may have allowed selection of patients with only modest neurological deficits (median NIHSS nine). This may have allowed the observed excellent clinical response rate of about 50% among placebo patients and created a difficult baseline to improve upon with therapy. Second, the excess death rate in the 125 μg desmoteplase group occurred late and was not stroke related. Hence, this was most likely due to the play of chance. Third, patient selection was based on both MR and CT perfusion criteria; while there has been much experience using the former in clinical trials, the latter is a new technique that may require further validation for this type of study. In particular, CT perfusion allows only a portion of the brain to be imaged (usually four slices), which may introduce penumbral selection errors. Finally, DIAS 2 was a small trial, with just over 60 patients in each of the placebo and two different dosage groups, consistent with a Phase IIb design. There was a highly optimistic prediction of treatment effect and the trial was probably underpowered to detect differences in clinical outcomes.
Where do we go from here? Obviously, we must await the full publication of the DIAS II results to better understand the data. Then, an individual patient meta-analysis including DIAS, DEDAS and DIAS II needs to be undertaken. Outcome measures should include both clinical response and reperfusion rates. Analysis of the effect of those patients included, based on CT perfusion criteria, also needs to be assessed. Further studies using desmoteplase may well be worthwhile. The results of other studies in which patients with significant penumbral mismatch are the target for therapy need to be taken into consideration. For example, in EPITHET, patients have been randomized 1:1 placebo vs. 0·9 mg/kg tPA after imaging with MR DWI/PWI to identify a target population of mismatch patients (4).
In planning future trials, particularly in those with tPA, data from DEFUSE (6) and EPITHET should be used to conservatively estimate sample size requirements for phase III trials in which the primary endpoints are clinical outcome measures. In doing so, entry criteria must be carefully planned to ensure a range of patients with sufficient neurological deficit, perhaps in the order of a median NIHSS value of around 14 as in the original NINDS trial (2). Hence, while we need to learn from the results of DIAS II, the general strategy of enriching trial populations with patients in whom a significant penumbral target is present is a logical one. Momentum must not be lost in our efforts to extend the time window for thrombolytic therapy.