LAUSR

In this issue, Keung et al present the results of their study using the National Cancer Data Base (NCDB) to compare the outcomes for patients undergoing surgery for primary retroperitoneal sarcoma (RPS) at low-volume hospitals (LVHs) with those of patients treated at high-volume hospitals (HVHs). This provocative article raises several issues that we will address in this editorial: 1) the challenge of using large national data sets for rare malignancies such as softtissue sarcoma (STS); 2) the difficulty in determining what constitutes a high-volume center for a rare disease; 3) the importance of centralizing surgery for patients with RPS at experienced centers; and 4) the lessons learned from this study and other experiences in planning and implementing STS surgical trials in the United States. First is the caveat of using data sets such as the NCDB for performing sarcoma research. Because sarcoma is not an organ-based malignancy (in contrast to breast cancer or colon cancer, for example), there is some potential for having inaccurate data submitted to these national databases. We recently identified significant underreporting of sarcoma surgical volumes due to coding errors at one of our institutions. Such volume data are fed into various national data sets and, if inaccurate at the onset, potentially will underrepresent volume. If such a problem is more pervasive nationally, then databases such as the NCDB may underestimate both the number of cases performed in total and the number performed at specific centers (and thus the number of centers that are HVHs for RPS). Nevertheless, data included in the data sets may be accurate even if not fully representative. Thus, a carefully crafted analysis tailored to the strength of the data, as the article by Keung et al is, still can be valid. However, using such data sets to track volume trends overall may not be. Assuming the question asked is appropriate for this data set, we then must consider a second, perhaps central, issue raised by the study by Keung et al: determining what constitutes an HVH. How do you define HVH versus LVH for any disease or malignancy in general and sarcoma in particular? The authors defined an HVH RPS center as any institution that treats >10 patients with primary RPS per year. This appears to be a very low bar and a somewhat arbitrary selection. However, defining a cutoff value based on any other rationale is challenging. The median value is only 1.1 cases per year per center and thus using that as a cutoff value (as has been done in other analyses) would be meaningless. Thus, by comparison, 10 appears to be a reasonable if somewhat random number. The authors did demonstrate the progressive improvement in outcomes noted with progressively higher volumes (from 0-5 cases/year to 6-10 cases/year to >10 cases/year). But what does this cutoff value of >10 cases per year truly represent? Does it matter that it is >10 primary RPS cases in particular or high volumes of STS cases in general? Case volume in part includes the experience of the surgeon. However, it also likely reflects the fact that the HVH have comprehensive cancer centers, and thus volume is a proxy for the experience of the multidisciplinary clinic in which these patients are evaluated. It appears logical that centers that treat more patients with RPS would have more experience with regard to the subtleties of care and potentially offer more treatment options. Therefore, the next obvious question is whether treatment at a proposed HVH actually makes a difference. In a landmark article, Birkmeyer et al raised alarms regarding the correlation between volume and surgical mortality. Experienced centers simply have better outcomes. In the current study by Keung et al, of the 1131 hospitals reporting data concerning 6950 patients treated over a 13-year period and included in the study, only 4 hospitals (0.4%) met