Are treatment outcomes in gastric cancer associated with either hospital volume or surgeon volume?

Abstract Surgical resection is the only curative treatment for gastric cancer. Postoperative outcomes may be affected by the average or total number of surgeries carried out at an institution (hospital volume) or by a surgeon (surgeon volume). Among seven large‐scale studies that each enrolled over 10 000 patients who underwent gastrectomy, six showed that higher hospital volume contributed to a lower mortality rate after gastrectomy. Surgeon volume was also reported by three of four studies that each included over 1000 patients to be a significant factor contributing to heterogeneity in mortality rates after gastrectomy. In contrast, most studies showed no relationship between hospital volume and postoperative morbidity. A significant long‐term relationship was demonstrated in four of nine studies that each included more than 1000 patients, but the other five studies showed negative results. A recent correlative study of randomized phase III trials for gastric cancer surgeries showed a significant relationship between hospital volume and postoperative morbidity in one trial but not in another trial. There was no correlation between overall survival and either hospital or surgeon volume. In addition, another correlative study of a phase III trial of randomized chemotherapy for unresectable or recurrent gastric cancer found that there was no correlation between hospital volume and overall survival, although there was a large degree of heterogeneity in median overall survival among participating institutions.

each included over 1000 patients to be a significant factor contributing to heterogeneity in mortality rates after gastrectomy. In contrast, most studies showed no relationship between hospital volume and postoperative morbidity. A significant long-term relationship was demonstrated in four of nine studies that each included more than 1000 patients, but the other five studies showed negative results. A recent correlative study of randomized phase III trials for gastric cancer surgeries showed a significant relationship between hospital volume and postoperative morbidity in one trial but not in another trial. There was no correlation between overall survival and either hospital or surgeon volume. In addition, another correlative study of a phase III trial of randomized chemotherapy for unresectable or recurrent gastric cancer found that there was no correlation between hospital volume and overall survival, although there was a large degree of heterogeneity in median overall survival among participating institutions.

K E Y W O R D S
gastric cancer, heterogeneity, hospital volume, mortality, surgeon volume

| INTRODUCTION
Gastric cancer is the second leading cause of cancer death worldwide, 1 and surgical resection is the only curative treatment. However, recurrence can be observed even after curative resection for gastric cancer. 2 As previous studies have shown that the recurrence rate could be significantly affected by the occurrence of postoperative morbidities, 3,4 we should examine the significant factors associated with short-term outcomes including postoperative mortality and morbidity. The average and total number of surgeries carried out at an institution (hospital volume) and by an individual surgeon (surgeon volume) are considered factors that affect outcomes after surgery. Several large population-based studies and systematic literature reviews have shown a close relationship between hospital volume and postoperative mortality in various types of cancer after surgical resection. [5][6][7][8][9] Patients who underwent surgery at high-volume hospitals had lower rates of postoperative mortality than those at lowvolume hospitals. For gastric cancer surgery, operative mortality rates were reported to be 13.0% at very-low-volume hospitals and 8.7% at very-high-volume hospitals in the USA, 5 compared to 1.1% at low-volume hospitals and 0.4% at high-volume or very-highvolume hospitals in Japan. 10 However, fewer studies have evaluated heterogeneity in outcomes after gastrectomy than after other types of surgery. Furthermore, no studies thus far have investigated the correlation between hospital volume and outcomes after chemotherapy for gastric cancer.
Here we review the relationship between postoperative outcomes and both hospital and surgeon volumes in patients with resectable gastric cancer. In addition, we introduce two correlative studies of Japanese randomized phase III trials investigating interinstitutional heterogeneity in outcomes after gastrectomy for resectable gastric cancer or after chemotherapy for unresectable or recur- We included studies that evaluated short-term mortality, morbidity, or long-term survival in gastric cancer patients who underwent surgical intervention. We excluded non-English publications, narrative reviews, editorials, letters, and case reports, as well as studies with fewer than 1000 gastric cancer patients. Of the 615 studies initially identified in the PubMed search, 30 met the above criteria and were analyzed ( Figure 1). Classification of low-and high-volume groups were defined using the cut-off numbers in the original articles.

| Short-term outcomes after gastrectomy
Since the 2000s, a total of 23 studies that each included more than 1000 patients have evaluated the relationship between hospital volume and mortality after gastrectomy ( Table 1). As a whole, a higher hospital volume was significantly associated with a lower postoperative mortality. Among seven large-scale studies with over 10 000 patients each, six showed that a higher hospital volume contributed to a significantly lower mortality rate after gastrectomy. 5,[11][12][13][14][15] Birkmeyer et al. 5  This number included patients who underwent esophagectomy or pancreatectomy. CI, confidence interval; HR, hazard ratio; NS, not significant; OR, odds ratio. one could not show a significant relationship between higher hospital volume and lower hospital mortality, but the absolute difference in mortality between low-and high-volume hospitals was greater than 1% (8.7% vs 6.9%). 25 In contrast, there is less evidence regarding the relationship between hospital volume and postoperative morbidity. To the best of our knowledge, most studies showed no relationship between hospital volume and postoperative morbidity. 18,20,21,[26][27][28] Morbidity rate is generally affected by patient characteristics such as comorbidities, acuity of presentation, and age. 29 High-volume hospitals have a great deal of experience treating postoperative morbidities without failure to rescue. 18 Surgeon volume may contribute to heterogeneity in postoperative outcomes. Birkmeyer et al. 31 reported that the majority of hospital volume effect was largely as a result of surgeon volume, and surgeon volume was positively related to postoperative mortality in the USA for all eight procedures they investigated. Unfortunately, their study did not include patients with gastric cancer. In terms of gastrectomy, three of four studies that each included more than 1000 patients showed a relationship between surgeon volume and mortality after gastrectomy ( Table 2). Xirasagar et al. 32 reported that the adjusted hazard ratio (HR) for death within 6 months was 1.3 (P<.01) for low-volume surgeons (6-month mortality, 23.0%) relative to very high-volume surgeons (6-month mortality, 16.7%).
Their study also showed that increasing surgeon age, which is correlated with the total number of surgeries carried out, was a significant prognostic factor. Yu et al. 33 reported that there was no significant association between surgeon volume and short-term mortality, but a significant difference between specialized and general surgeons was found in the 5-year survival rate (63.9% vs 59.7%, P=.038).

| Long-term outcomes after gastrectomy
As described above, most studies have focused on the relationship between hospital volume and short-term outcomes in patients undergoing gastrectomy. Since the 2000s, nine studies that each included at least 1000 patients have focused on long-term survival after gastrectomy (Table 3), and four of these found a significant relationship between survival and hospital volume. 23,[34][35][36] In a study from the USA showing a positive relationship between hospital volume and longterm outcomes (HR, 0.82; 95% CI, 0.73-0.92; P<.001), differences in outcomes could not be explained by discrepancies in patient characteristics or operative mortality, suggesting that hospital volume was an independent predictor of long-term survival after gastrectomy. 35 In contrast, a Dutch study showed that high hospital volume was associated with long-term survival after esophagectomy, but not after gastrectomy. 37 A Japanese study showed that in patients with localized or regionally metastasized gastric cancer, 5-year survival between 1975 and 1979 was worse when resection was carried out at very low-volume hospitals compared to high-volume hospitals. 34 This correlation, however, decreased in later time periods and disappeared in the period between 1990 and 1994, except for very low-volume hospitals. The authors commented that improvements in medical technology and cancer care were widespread among various types of hospitals in Japan, so the influence of hospital volume on long-term survival might have diminished for common malignancies such as gastric cancer. However, patients who were treated at very lowvolume hospitals remained at significantly higher risk of death than those treated at high-volume hospitals.
However, several studies could not show a positive relationship between hospital volume and long-term outcomes. 22,24,32,37,38 One of these negative studies suggested that this is because longterm outcomes are affected by two different dimensions of expertise: surgical technical skill and diagnostic ability. 32 T A B L E 2 Studies with more than 1000 patients evaluating the relationship between surgeon volume and mortality after gastrectomy Describes mortality rates when specialized surgeons had at least four consecutive years of surgical practice. c Describes 6-month mortality and hazard ratio. CI, confidence interval; HR, hazard ratio; NS, not significant; OR, odds ratio.
Unfortunately, several of these studies lacked data on such important factors, which might have led to the lack of an observed relationship between hospital volume and long-term outcomes.
Therefore, we now review the results of important studies using data in the landmark phase III trials that made an effort to assure the qualities of surgical or oncological skills and diagnostic ability.

| Correlative study of randomized phase III trials
Recently, a correlative study of two randomized phase III trials evaluated the inter-institutional heterogeneity in short-and long-term outcomes after gastrectomy for resectable gastric cancer. 39   These studies included esophageal cancer patients. CI, confidence interval; HR, hazard ratio; N(À), node negative; N(+), node positive; NS, not significant. turn, may have been a result of variations in medical oncology clinical experience.

| CONCLUSION
As with other types of surgery, heterogeneity in post-gastrectomy mortality according to hospital volume was observed in multiple studies. Hospital and surgeon volumes have also been reported to be significant factors contributing to heterogeneity in postoperative mortality. However, most studies showed no relationship between hospital volume and postoperative morbidity. The relationship between hospital volume and long-term outcomes after gastrectomy is also controversial. Recent correlative studies of randomized phase III trials for gastric cancer demonstrated that there was no correlation between postoperative OS and either hospital or surgeon volume, or between OS after chemotherapy and hospital volume.