Introduction

In gastric cancer, lymph node metastasis has been known to occur at a relatively early stage, and D2 gastrectomy has been recommended for the radical treatment of locally advanced gastric cancer [1]. Although D2 dissection is a standard surgery, D2 dissection is associated with a high risk of morbidity, such as pancreatic fistula and intra-abdominal abscess because supra-pancreatic nodal dissection is necessary for D2. If the surgical risk exceeds the benefit of D2 dissection, the surgeons may avoid D2 dissection.

Evidence of D2 dissection has been established in the era of surgery alone. In Japan, postoperative adjuvant chemotherapy using S-1 (TAIHO PHARMACEUTICAL CO., LTD, Tokyo, Japan) has been the standard treatment for pStage II/III locally advanced gastric cancer since 2007 [2]. However, no consideration has been given to the efficacy of dissection in the era of effective adjuvant chemotherapy. It remained unclear whether the efficacy of D2 was unchanged in the era of effective adjuvant chemotherapy depending on each nodal station. As a good indicator of the benefit of lymph node dissection, Sasako et al. introduced the therapeutic value index [3]. The therapeutic value index estimated the efficacy of nodal dissection based on the proportion of patients with metastasis who survived for more than five years as a result of dissection.

From these backgrounds, we evaluated the therapeutic value index of each nodal station in patients with locally advanced gastric cancer who were treated after the ACTS-GC trial in comparison to those treated before the ACTS-GC trial. The main objective of this study was to evaluate how the therapeutic value index of each lymph node station was affected in the era of effective adjuvant chemotherapy for locally advanced gastric cancer. This is the first report to discuss how the therapeutic value index was affected in the era of effective adjuvant chemotherapy in locally advanced gastric cancer.

Methods

This was a retrospective, single center, observational study. We reviewed clinical data acquired from hospital records, including general information, histopathological data, and treatment and follow-up data. The ethics committee of the National Cancer Center Hospital approved this study in November 2017 (No. 2017–077).

Selection of the patients

We reviewed 1298 patients with advanced gastric cancer, including 725 who received gastrectomy between January 2000 and December 2006 (the Former group, reflecting the era of surgery alone), and 573 cases who received gastrectomy between January 2007 and July 2015 (the Latter group, reflecting the era of effective adjuvant chemotherapy using S-1). In 2007, the standard practice for locally advanced gastric cancer changed from surgery alone to surgery followed by adjuvant chemotherapy using S-1 [2].

All patients underwent gastrectomy with lymphadenectomy, based on the Japanese Gastric Cancer Treatment Guidelines, at our department from January 2000 to July 2015 [1]. All cases met the criteria as follows: (1) histologically proven to be common-type adenocarcinoma, (2) advanced gastric cancer (pT2-4), (3) received total or distal gastrectomy with D1 or more lymphadenectomy, (4) R0 was achieved, (5) M0, (6) histologically proven to be pStage II or III (except for pT3N0) in accordance with the Union for International Cancer Control tumor–node–metastasis classification, eighth edition [4], and (7) preoperative therapy was not added.

Evaluating tumor

We evaluated tumors in accordance with the Union for International Cancer Control tumor–node–metastasis classification, eighth edition [4]. We numbered the lymph node stations in accordance with the Japanese Gastric Cancer Association [5, 6]. Then, we defined the perigastric stations (No. 1–7) as the D1 station and stations apart from the stomach (No.8a-12a) as the D2 station. We also defined ND1 as positive for lymph node metastasis up to the D1 station and ND2 as positive for lymph node metastasis beyond the D1 station and up to the D2 station.

Therapeutic value index of lymphadenectomy

We adopted the therapeutic value index, introduced by Sasako et al. (1995), evaluating the impact of each dissected station [3]. The index was defined as the product of the incidence of lymph node metastasis to the station and the 5-year overall survival rate of the patients.

Statistical methods

All data are shown as the median and range. To compare proportions of patients, we used Student’s t test and Fisher’s exact test. Overall survival was determined from the date of the operation to the date of death or the last follow-up. Overall survival rates were determined by the Kaplan–Meier method and compared by the log-rank test. All statistical analyses were performed using EZR version 1.52 [7].

Results

Study subjects

We screened 3028 patients who underwent radical gastrectomy for advanced gastric cancer between January 2000 and July 2015. The 1298 patients who were determined to be eligible were divided into the Former (N = 725) and Latter (N = 573) groups (Fig. 1).

Fig. 1
figure 1

Flow diagram of the 3028 patients who underwent gastrectomy for advanced gastric cancer between January 2000 and July 2015

Full size image

Patient backgrounds and histopathological findings

Table 1 shows the patient background factors and histopathological findings. In this cohort (N = 1298), the Latter group was older population (p < 0.001), their tumors were more frequently located in the upper stomach (p = 0.008), they showed less intraoperative blood loss (p < 0.001), and tended to have smaller tumors (p = 0.081) in comparison to the Former group. Adjuvant chemotherapy was administered to 75.2% of the Latter group and 9.4% of the Former group. In almost all cases, the adjuvant chemotherapy regimen was S-1 or a regimen containing S-1.

Table 1 Clinicopathological backgrounds (n = 1298)
Full size table

Patient survival

In both groups, the median follow-up period of the survivors exceeded 5 years (Former group, 104.5 months; Latter group, 77.1 months). Figure 2 shows the 5-year overall survival rates. The overall survival curve of the Latter group was superior to that of the Former group (hazard ratio = 0.80, p = 0.025). In ND1 and ND2, the overall survival curves of the Latter group were also improved in comparison to the Former group (p = 0.027 and p = 0.201, respectively). The hazard ratios of ND1 and ND2 were 0.73 in ND1 and 0.81 in ND2.

Fig. 2
figure 2

Overall survival. a Overall survival of the whole population, b Overall survival of ND1 cases, c Overall survival of ND2 cases. ND1, positive for lymph node metastasis up to D1 station; ND2, positive for lymph node metastasis beyond D1 station and up to D2 station; 5y-OS, 5-year overall survival

Full size image

Therapeutic value index for each station

Table 2 shows the rate of nodal metastasis, the 5-year overall survival rates of patients with metastasis, and the measured therapeutic value index for each group. The metastasis rates of the D1 and D2 stations were similar between the Former and Latter groups (85.1% vs 84.5% in D1 station and 25.5% vs 26.7% in D2 station). On the other hand, the 5-year survival rates of ND1 and ND2 both improved from the Former to the Latter groups: 67.8% vs 74.5% for ND1 and 53.5% vs 60.9% for ND2. The improvement in the 5-year survival rate was similar between ND1 and ND2. The therapeutic value index was also improved from the Former to the Latter group at almost all lymph node stations. The improvement ratio of the therapeutic value index was 1.09 (57.7 to 62.9) for the D1 station and 1.19 (13.7 to 16.3) for the D2 station.

Table 2 The nodal metastasis rate, 5-year overall survival rate, and measured therapeutic value index
Full size table

Discussion

In this study, we compared the therapeutic efficacy of lymphadenectomy in locally advanced gastric cancer in the era of surgery alone with that in the era of effective adjuvant chemotherapy. Survival in the era of surgery alone represents the cure rate achieved by local control alone and would indicate the proportion of tumors limited to the local site, while the survival increment in the era of effective adjuvant chemotherapy suggests the efficacy of adjuvant chemotherapy for micro-metastatic tumor cells that are not eradicated by surgery alone. In comparison to the survival of the Former group, the hazard ratio of ND1 was 0.73 and that of ND2 was 0.81 in the Latter group, which was similar to the results of the ACTS-GC trial [8]. On the other hand, the present study demonstrated that the therapeutic value index improved at almost all stations in the era of effective adjuvant chemotherapy regardless of whether the D1 or D2 nodes were involved. Taken together, adjuvant chemotherapy would be effective for both the D1 and D2 nodes. Accordingly, D2 gastrectomy would be still important for locally advanced gastric cancer in the era of effective adjuvant chemotherapy.

Then, in comparison to the previous report, are the index values calculated in our study reasonable? Many reports examined the therapeutic value index to evaluate the value of dissection for gastric cancer arising in a specific area, such as proximal lesions, lesions on upper gastric curvature, and lesions with duodenal invasion [9,10,11]. However, few reports have used the index to evaluate the value of dissecting whole gastric cancer. Furthermore, there are few reports on the components of the index, that is, the frequency of metastasis and survival rate for each nodal station. Regarding the frequency of metastasis of each nodal station, our search only identified a report related to T2 gastric cancer from China. Although the frequency of metastasis was rather low, the report showed a similar tendency to our study [12]. The survival rate reported for each nodal station with metastasis was similar to that of the original report by Sasako (although it was an old report from the same institute) [3]. Accordingly, it was considered that the index values of the Former group were appropriate for values obtained by surgery alone.

The hazard ratios of the present study seemed slightly inferior to those of the ACTS-GC trial. One explanation is that this was a retrospective study in a current real-world setting and a certain number of cases who had not received adjuvant chemotherapy were included in the era of adjuvant chemotherapy.

Moreover, in this study, reflecting the situation of Japanese society, the cohort in the era of adjuvant chemotherapy was older and had more proximal cancer than the cohort in the era of surgery alone. In general, these populations are considered to have a poorer prognosis [13, 14]. These factors may have affected the present results. Nevertheless, the prognosis of the cohort in the era of adjuvant chemotherapy was improved. Regarding the optimal extent of lymph node dissection and postoperative adjuvant chemotherapy for the very elderly, further research is necessary.

Some factors, such as perioperative care, perioperative complications and the development of surgical devices, may have differed between the Former and Latter periods. Changes in perioperative care or complications would strongly affect postoperative stay. As shown in this study, the frequency of hospital stay exceeding 3 weeks, reflecting major surgical morbidities, was high in both periods and did not differ to a statistically significant extent. Also, the median hospital stay in the latter period was only 2 days shorter than that in the former period, and the difference did not reach statistical significance. Although these differences could be a factor for improved survival in the latter period, the effect would be not so large. On the other hand, intraoperative blood loss was significantly reduced in the Latter period. This result may have been influenced by the progress of surgical devices. The amount of intraoperative blood loss is known to be correlated with the patient’s prognosis [15], and the decrease in blood loss may have improved the prognosis of the Latter period and overestimated the value of adjuvant S-1 therapy.

In the ACTS-GC trial, the hazard ratio of the whole cohort, N1, and N2 were 0.68, 0.61, and 0.84, respectively [2, 8]. On the other hand, the present study showed that the hazard ratio of the whole cohort, ND1, and ND2 were 0.80, 0.73 and 0.81, respectively. Considering that the N factor of the JCGC thirteenth edition was associated with the location of nodal metastasis [16], our results showed a similar tendency to the ACTS-GC trial. These results may show that S1 was less effective for the D2 station. However, the improvement in the therapeutic value at the D2 station was almost equal to that at the D1 station.

The present study was associated with several limitations. First, this study just compared the index based on the metastatic rate and the survival between the Former and Latter period. In the Latter period, most (but not all) patients received adjuvant chemotherapy including S-1. Ideally, surgery alone should be compared to S-1 adjuvant chemotherapy, which would be only possible by randomized comparison of the ACTS-GC trial; however, this was not been performed. Moreover, a prospective study is impossible in the era of effective adjuvant chemotherapy. The present study would be inferior in scientific reliability but would be superior in terms of reflecting real-world data in the new era in which several effective regimens are available in which some patients do not receive adjuvant chemotherapy due to surgical morbidities. Before initiating adjuvant chemotherapy, surgeons must determine extent of nodal dissection. Thus, our study would be helpful for optimizing nodal dissection with consideration of the benefit of dissection in the new era as well as the surgical risk. Second, the comparison between the two periods includes factors other than the rate of adjuvant chemotherapy administration. The effects of different background factors cannot be ignored. Moreover, the anticancer drugs administered after recurrence would have differed between the Former and Latter periods. This factor would have improved survival without S-1 adjuvant chemotherapy. Thus, our results would overestimate the index obtained by effective adjuvant chemotherapy in the Latter period. Third, this is a single-center study. Validation of the present results in a multicenter study may be required. Finally, this method cannot be applied to cases treated with neoadjuvant chemotherapy. At present, we cannot calculate the therapeutic value index for patients who receive perioperative chemotherapy because the accurate evaluation of preoperative lymph node metastasis is difficult. Therefore, the generalizability of the present results would be limited in areas where perioperative chemotherapy is a standard of care.

Conclusion

We showed that the therapeutic value index of each station improved in the era of effective S-1 adjuvant chemotherapy, regardless of whether the D1 or D2 nodes were involved. The results suggest that D2 gastrectomy is still important for locally advanced gastric cancer in the era of effective adjuvant chemotherapy.