Lymphadenectomy for gastric cancer.

Abstract

Perhaps the greatest criticism of reports from Japan and other Asian countries demonstrating significant benefits and modest morbidity from extended lymph node dissection for gastric cancer has been the retrospective nature of the data. Despite enormous numbers of patients afflicted with gastric cancer in Japan, these retrospective reports have often been considered suspect by many Western surgeons because of a lack of prospective randomized studies. These suspicions will be challenged with the first report of the Japan Clinical Oncology Group (JCOG) study of lymphadenectomy in this issue of the Journal of Clinical Oncology. Although some questions may yet persist as to the applicability and rationale for extended lymphadenectomy for gastric cancer, Sano et al are to be commended for completing this ambitious trial of D2 lymph node dissection compared with a more extensive lymphadenectomy. A D1 lymphadenectomy includes only those nodes adjacent to the stomach; a D2 dissection also removes the nodes around the branches of the celiac axis. The JCOG trial randomly assigned patients at surgery, after confirming the ability to perform a curative resection (R0) to either a D2 dissection or to a more radical node dissection that included the nodal tissue along the aorta. This study built on lessons learned by Professor Sasako (one of the major collaborators of this study) from his significant involvement in the Dutch randomized trial of D1 dissection compared to D2 dissection. The JCOG study was conducted at 24 centers, each of which entered an average of more than 20 patients; this compares with the Dutch trial, in which 80 centers averaged fewer than 10 entries each. In addition, surgeons in this study had to have personally performed more than 100 gastric resections or be at an institution with a specialized unit where more than 80 gastric resections were performed annually. There are relatively few surgeons in the United States who would have been able to participate in this study. To put this level of experience into perspective, we conducted a recent review of all patients who underwent a gastric resection for cancer at a hospital in the state of Texas during a 3-year period. We found that the median annual number of resections per institution was only two, and only two hospitals in the entire state performed more than 15 gastric resections per year. Neither would have met the criteria for participation in the JCOG study. Not surprisingly, our review found that operative mortality was significantly lower at the higher volume centers. When the operative survival rates of the JCOG study are compared with large population-based studies of US patients, where the operative mortality for gastrectomy approximates 9%, the mortality rate presented here of less than 1% seems remarkable. There are, however, likely to be significant differences in the patient populations. Virtually all of the patients in this study were considered to be potentially curative after a rigorous intraoperative evaluation that included negative peritoneal cytology. Patients who undergo a palliative resection are much more likely to suffer postoperative morbidity and mortality than those undergoing a potentially curative resection. However, one must also consider the experience at major US centers such as Memorial Sloan-Kettering, where D2 dissection is routinely used, with a reported operative mortality rate of 3.6% in 865 patients undergoing an R0 resection. At The University of Texas M.D. Anderson Cancer Center, where spleenpreserving D2 dissection is performed, the mortality rate was 1.7% in 175 patients treated on eight prospective trials of neoadjuvant therapy. These rates, although not quite as low as the rates reported in the current JCOG study, are well below those seen in both the Dutch and Medical Research Council trials, as well as the large US national database surveys. Another important factor seen in this report is the pathologic evaluation of the specimen. Patients who had the “limited” procedure (D2) had an average of 54 nodes evaluated. This is compared with the experience in the United States, where, in the retrospective study by Wanebo et al, more than one third of patients had no nodes reported in the pathologic specimen. Though surgery was not part of the protocol in the Intergroup 0116 trial, which demonJOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L VOLUME 22 NUMBER 14 JULY 15 2004