Response to Letter by Steiner et al
Steiner et al wrote a thoughtful comment on our recent article describing our off-label experience with activated factor VII (FVIIa) for spontaneous intracerebral hemorrhage (ICH).1 The points they raise are valid and we would like to comment.
The mean volumes were smaller in the treated patients than those of the phase-II FVIIa trial and our own controls. We certainly acknowledge that this may contribute to the lower magnitude of growth observed in our patients. Because we did not directly compare hematoma growth to that of untreated patients, we do not suggest that this is a proof of efficacy of FVIIa. Because we did not attempt this comparison, the purpose of including a control group was to compare clinical outcomes. The fact that there was no difference in clinical outcome despite the volume difference between the treatment and control groups (biased in favor of the treated group) is a signal that the efficacy of FVIIa in unselected ICH patients may be limited. The fact that very similar findings were found in the phase-III randomized controlled trial of FVIIa appears to support this conclusion.
We measured ICH volumes using the ABC/2 method. As Steiner et al point out, this method lacks the accuracy of a computerized measurement such as was used in the FVIIa trials. We concur that in ICH every milliliter is important and this may be a source of variance in our data. We would contend, however, that as the same method was used to blindly assess both groups, the probability of a systematic bias favoring one group is small.
We concur with Steiner et al that edema and intraventricular hemorrhage volume can make important contributions to clinical outcome.2 In our study we chose to concentrate on the primary efficacy end points of the FVIIa trials (namely ICH growth and clinical outcome). Because our primary end point was to compare 0 to 3 hours versus 3 to 4 hours treatment in terms of ICH growth and outcome, we did not include that in the analysis. In addition, the number of patients with IVH was too small to provide meaningful information. Notwithstanding, we have included this additional information (Table). Despite the limitations imposed by the sample size, it can be appreciated that the IVH and edema volume were very similar in both groups and there is no trend toward reduction of IVH growth or edema in the 0 to 3 hour group.
Steiner et al argue that the removal of warfarin-related ICH and patients who underwent surgery from the efficacy analysis is a factor that limits the meaningfulness of our conclusions. We respectfully disagree. In our experience, these subgroups of patients differ substantially from other ICH patients both in the degree and time frame of hematoma growth (warfarin-related ICH) and outcome (surgery may adversely affect outcome in patients with deep hematomas and may improve outcome in patients with superficial hematomas). Also, patients who undergo surgery within 24 hours cannot be assessed for hematoma growth. We believe that including those patients in the analysis would have made the interpretation of our results more difficult.
The safety of FVIIa is indeed an ongoing issue. The purpose of the safety analysis in our article was to provide additional “real life” data on the adverse events in patients treated with FVIIa. Because these events are not collected as rigorously in routine nontreated patients, some adverse events may be missed. It is likely that we were able to capture the events in the treated patients through adherence to a strict follow-up protocol mandating serial cardiac enzymes, ECGs and lower extremity ultrasound in the treatment group. For this reason, comparing the adverse events rate to historical controls is likely to produce a falsely low rate in the control group because of ascertainment bias. A previous publication from our center suggested, however, a rate of 3% troponin elevation among ICH patients that were not treated with FVIIa.3 We do agree that the only way to reliably test the safety of FVIIa is in a randomized placebo-controlled trial. While the publication of the FAST trial is pending, the safety results have been presented and it does appear that FVIIa is associated with a moderate increase in thrombotic events. Our rate is slightly higher than reported,4 and may be accounted for by 2 factors: less stringent exclusion criteria than in a clinical trial (for example, excluding patients with any history of cardiac disease) and smaller sample. In the latter case, this rate may be well within the play of chance.
Steiner et al point out the limited power of our study to detect any difference in hematoma growth and clinical outcome. We are, of-course, aware of this issue and concur that the lack of statistical significance does not exclude the possibility of a meaningful difference. Notwithstanding, we would expect to see a signal of efficacy if indeed there was a difference. The slightly lower relative hematoma growth in the 3 to 4 hours group may signify (as stated in the article) the natural dynamics of ICH growth in this time frame. If anything, our results emphasize the importance of refining patient selection for future hemostatic trials. It is clear that data coming from a large randomized trial provide better evidence of efficacy. A post hoc analysis of the data, however, may not be as robust. As a group we are firm believers in the hemostatic hypothesis and maintain that FVIIa could be beneficial in ICH despite the recent setbacks. We believe this treatment should target a selected group of patients with acute ICH. The selection criteria may include timing, but great care should be exercised to avoid too-strict timing criteria because this will exclude many patients and will limit the generalizability of the study.
We thank Steiner et al for their comments. We would like to point out that our conclusion was not that timing is unimportant in ICH, but rather that in unselected patients there appears to be little difference between treating within 3 hours or 3 to 4 hours. We are in favor of maintaining a longer time window and finding alternative means of patient selection to identify the population most likely to respond. In this way, it may be possible to maximize the efficacy of FVIIa in the treatment of ICH while keeping patient eligibility high.
James Grotta reports having received research grant support from Novo Nordisk; Nicole Gonzales reports receiving a one-time consulting fee from Novo Nordisk; Miguel Escobar served as a consultant for Novo Nordisk and has served on the DSMB for Novo Nordisk clinical trials. There are no other conflicts to report.
Hallevi H, Gonzales NR, Barreto AD, Martin-Schild S, Albright KC, Noser EA, Illoh K, Khaja AM, Allison T, Escobar MA, Shaltoni HM, Grotta JC. The effect of activated factor VII for intracerebral hemorrhage beyond 3 hours versus within 3 hours. Stroke. 2008; 39: 473–475.
Steiner T, Diringer MN, Schneider D, Mayer SA, Begtrup K, Broderick J, Skolnick BE, Davis SM. Dynamics of intraventricular hemorrhage in patients with spontaneous intracerebral hemorrhage: risk factors, clinical impact, and effect of hemostatic therapy with recombinant activated factor VII. Neurosurgery. 2006; 59: 767–773; discussion 773–764.
Sugg RM, Gonzales NR, Matherne DE, Ribo M, Shaltoni HM, Baraniuk S, Noser EA, Grotta JC. Myocardial injury in patients with intracerebral hemorrhage treated with recombinant factor viia. Neurology. 2006; 67: 1053–1055.
Diringer MN, Skolnick BE, Mayer SA, Steiner T, Davis SM, Brun NC, Broderick JP. Risk of thromboembolic events in controlled trials of rfviia in spontaneous intracerebral hemorrhage. Stroke. 2008; 39: 850–856.