News:Acute Myocardial Infarction with Hyperoxemic Therapy II (AMIHOT II) in ST Elevation MI with Rebuttal by Dr. Stone

C. Michael Gibson, M.S., M.D. October 23rd, 2007

WASHINGTON DC:

Please see rebuttal regarding the interpretation of this trial from Drs. Stone and Martin at the bottom of this page

It has been hypothesized that intracoronary administration of supersaturated oxygen (hyperoxemic reperfusion) would reduce infarct size in the setting of ST elevation MI. The AMIHOT II investigators set out to explore this hypothesis by administering supersaturated oxygen for 90 minutes following primary PCI using the TherOx® AO System (TherOx Inc., Irvine, California) in patients with an anterior MI. This therapy had been previously evaluated in the AMIHOT 1 trial, which showed no benefit overall. However, in the subgroup of patients with an anterior MI who were reperfused early (i.e. within the first 6 hours of symptom onset), there was a reduction in infarct size with hyperoxemic reperfusion. Because neither the AMIHOT I trial or the present AMIHOT II trial were adequately powered, the design of the present study was to pool data from the AMIHOT 1 trial with the AMIHOT II trial using Bayesian statistics.

Patients were randomized following successful PCI for an anterior MI to either hyperoxemic reperfusion (n = 222) or control (n = 79). The primary endpoint was SPECT infarct size evaluated at 14 days. The two groups were well matched with respect to duration of symptoms and door to balloon times were brisk at 77 only minutes. There was a trend for fewer patients to have normal TIMI grade 3 flow at the completion of the PCI in the hyperoxemic group (88.4% vs 93.0% of control patients). SPECT imaging could only be performed in 78.8% of hyperoxemic patients at 14 days and in only 87.3% of the control group, and infarct size was imputed in those patients with missing data.

SPECT infarct size (the primary endpoint) did not differ between the hyperoxemic and control groups in AMIHOT II (median 26.5% for hyperoxemic group vs. 20%, p = 0.10). When data from AMIHOT II was pooled with the positive subgroup data from AMIHOT I, SPECT infarct size was lower in the hyperoxemic group (median 18.5% vs. 25%, p = 0.023). The reader should be aware that the inclusion of a positive subgroup along with the results from the overall trial in the present study should be viewed with caution.

With respect to clinical outcomes, there was no benefit for hyperoxmic therapy. 30 day MACE occurred in 5.4% of patients treated with supersaturated oxygen compared with 3.8% of the control group (p = 0.77). Although no patients in the control group died, there were 4 deaths in the hyperoxemic group (p = 0.58). There was significantly greater bleeding in the hyperoxemic group (24.3% vs. 12.7%, p = 0.04).

Following the initial report of this article, I received this rebuttal from Drs. Stone and Martin:

"1) The primary pre-specified efficacy endpoint as per the FDA approved statistical analysis plan was infarct size reduction based on a Bayesian model incorporating data from AMIHOT I while maintaining a type one error of <5%. Thus it does not seem accurate to describe AMIHOT II as a negative trial given that it met its pre-specified primary efficacy endpoint. 2) It is not completely accurate to state there was pooling of a positive subgroup from an overall negative AMIHOT I trial. The Bayesian model did NOT allow for simple pooling of data and did NOT pre-specify any degree of pooling. The model was designed to assign weight to data from AMIHOT I based on the consistency of data between the two trials. 3) The suggestion that "extreme caution" is called for with regard to the statistical methodology in this trial calls into question the use of Bayesian methodology in clinical investigation in general. The rationale for and the importance of Bayesian models in device trials is documented in the FDA draft guidance cited in the presentation. Although it may be appropriate to raise issues related to the size of the trial and the level of evidence which it provides, it does not seem accurate to suggest that inappropriate models were applied."

Gregg W. Stone, M.D. Professor of Medicine Director of Cardiovascular Research and Education Center for Interventional Vascular Therapy Columbia University Medical Center Chairman The Cardiovascular Research Foundation New York City, NY

Jack L. Martin, M.D.