Response to Letter by Powers Regarding Article, “Failure of Cerebral Hemodynamic Selection in General or of Specific Positron Emission Tomography Methodology? Carotid Occlusion Surgery Study (COSS)”

In response to the Letter to the Editor by Powers, “Failure of Cerebral Hemodynamic Selection in General or of Specific Positron Emission Tomography Methodology? Carotid Occlusion Surgery Study (COSS),” we note the following.

Although we appreciate the great effort Dr Powers expended to design and carry out the COSS trial, it is imperative that we look critically at the study and consider all possibilities for the futility end of the study to define a difference between the surgical and the medical groups.

The ratio of 1.12 we reported was based on COSS instructions to determine eligibility for enrollment. The cutoff was apparently changed to 1.13 in final analysis. We also note that our publication was based on Dr Powers' presentation 6 months earlier at the International Stroke Conference.

Dr Powers was concerned that the patients in our study were not relevant to the COSS population. The 14 patients studied included patients who were eligible for COSS. Denied by the COSS executive committee to include our quantitative studies as part of COSS, we were allowed to do our quantitative studies after the COSS studies were completed. This is how the studies we report were done. Because these studies were not done as part of COSS, they were outside the jurisdiction of COSS requirements for publication.

We appreciate the restatement of how the COSS trial was derived from an initially quantitative oxygen extraction fraction (OEF) study and converted to a qualitative study based on post hoc analysis of 36 patients. The difficulty in recruiting patients if an arterial line was needed justified the qualitative study. We also appreciate that the post hoc derivation of a 40% stroke rate projection used in the COSS study design was based on post hoc analysis of the same 36 patients without prospective validation. No other study has shown a subsequent stroke rate as high as 40% in patients who were not identified with hemodynamic compromise. A study by Yamauchi et al¹ in 40 patients with occlusive vascular disease reported a stroke rate of 57% in patients with high OEF >53.3%. No difference in stroke rate was observed based on OEF asymmetry.

We also would like to note that the threshold of 50% OEF is based on many studies with quantitative OEF including some from Dr Powers' group in St Louis. Other studies have evaluated the qualitative OEF ratio method versus quantitative OEF and found the latter to fail in identifying hemodynamic compromise. These considerations are included in our brief report.

Dr Powers suggested that the reduction of stroke in the control group from 40% to 24% was most likely due to an improvement of outcome from improved medical therapy. However, 24% was also the stroke rate reported in the literature for symptomatic carotid occlusion. Determination of the stroke rate among patients who did not qualify for inclusion in COSS based on qualitative OEF ratio would clarify the issue as to whether they had a stroke rate equal to or less than those who were selected for inclusion into COSS and treated medically. These data should still be obtainable through follow-up and we look forward to those data being reported.

An equally plausible explanation for the similar results obtained in both treatment and control groups in COSS is that the qualitative approach chosen based on post hoc analyses of a small study (36 patients from the St Louis stroke study) is because they failed to identify the high-risk stroke group as suggested by an independent analysis of their qualitative versus a classic quantitative approach as published by our group. Unfortunately, a more plausible explanation for the inability of the COSS study to define a benefit due to surgery was the same reason the initial bypass trial failed, that is, failure to identify and select a high-risk, hemodynamically compromised population for the study.

The conclusion of the JAMA publication² of the results of this aborted trial due to futility is that “bypass surgery did not make a significant difference in outcome.” Although this conclusion may be factually correct, the suggestion that bypass surgery does not still have the potential to show a clinical benefit must be challenged until a methodologically correct study is performed.

We still conclude that the COSS trial most likely failed not due to the inability of bypass surgery to prevent stroke, but more likely due to the inability of COSS to correctly identify a hemodynamically compromised high-risk subgroup.