Our results suggest that ACRT with 45Gy IMRT and concurrent oral capecitabine at a dose of 800 mg/m2 twice daily had an acceptable efficacy and toxicity profile in patients with local advanced gastric cancer after radical gastrectomy and D1/2 LND. The 3-year DFS was 66.2%, which did not reach the primary hypothesis endpoint of our phase II study.
The role of ACRT for local advanced gastric cancer remains debatable. The benefits or drawbacks of this scenario mainly depend on whether a D1 or D2 lymphadenectomy has been performed. The INT-0116 study was the first trial to prove the benefit of ACRT in patients after radical gastrectomy and D0/1 LND: it showed that the 3-year OS and DFS increased from 41% to 50% and 31% to 48%, respectively . Even after 10 years of follow-up, ACRT was associated with superior DFS and OS . Dikken et al suggested that addition of ACRT after D1 LND had a major impact on local recurrence in gastric cancer . Zhang, N.et al suggested that patients with D1 or D1 plus LND benefit from adjuvant RT, and adjuvant RT may be beneficial for some patients with D2 LND . Yu, J. I. et al. re-analyse the ARTIST study and conclude that adjuvant RT after D2 resection in gastric cancer reduced locoregional recurrence risk, especially in group 3 LNs, and improved locoregional recurrence-free survival. Patients with positive LN benefited more from the adjuvant RT than the other subgroup . The National Comprehensive Cancer Network (NCCN) guideline recommends ACRT as an adjuvant treatment in patients with less than D2 LND.
In China, D2 LND is considered a routine surgical procedure for locally advanced gastric cancer because it is the most widely accepted surgical procedure in Asian and European countries . However, given the many differences between centers or institutions in terms of hospital volume, patient populations, surgical practices and training, postoperative nursing experience, and pathological identification and examination of LNs, it is difficult to standardize and generalize D2 LND, even in our specialized cancer hospital. In our previous retrospective study of 297 local advance gastric cancer patients who received radical surgery alone, the median number of LNs resected was 18 (range, 4–68 nodes) with a 27.6% 5-year locoregional recurrence rate . In our phase I trial (performed between 2007 and 2009), D2 LND was performed in only 16.7% (3/18) of patients with gastric cancer, with a median of 19 LNs (range, 5-35) examined in our hospital. The median number of LNs resected in the present study (performed between 2009 and 2012) reached up to 24 nodes (range, 5–56 nodes). In a large observational study conducted in patients who underwent radical resection for gastric cancer, the survival benefits significantly associated with an increase in the number of LNs resected, even when as many as 40 LNs were examined . According to published reports [18, 19], the greater number of LNs resected in our study may have provided better locoregional control and possibly a survival advantage, although this number is still much smaller than those reported in studies conducted in Japan and Korea (D2 LND with a median of more than 40 LNs examined) [2, 3]. Furthermore, although patients underwent a very high-quality D2 LND, the ARTIST trial (ACRT Trial of Capecitabine Plus Cisplatin for Gastric Cancer) demonstrated that DFS could be further improved by ACRT in positive LN gastric cancer. A randomized trail published in 2013 showed recurrence free survival benefit (median time 36 months vs. 50 months, P = 0.029) in ACRT group after D2 LND, which did not provide the resected LN number . Thus, it is reasonable that ACRT could be considered as an adjuvant treatment for patients with LN metastasis or those who did not undergo a high-quality D2 LND.
In the past decade, capecitabine has been widely used in gastrointestinal cancer, as it has a much safer side effect profile and does not require invasive delivery [21, 22]. Oral capecitabine was not inferior to in fusional 5-FU in randomized control trials for patients with advanced gastric cancer . Therefore, capecitabine has been considered as a standard chemotherapy regimen for the treatment of advanced gastric cancer worldwide. The NCCN guidelines also suggest that infusional 5-FU can be replaced by oral capecitabine in gastric cancer. Our previous study determined that the maximum tolerated dose and recommended dose of capecitabine was 800 mg/m2 twice daily when administered concurrently with IMRT for gastric cancer as adjuvant therapy , which is similar to the dose used in the ACRT group concurrent with RT in the ARTIST trial . Lee et al. evaluated the efficacy and toxicity of ACRT using FP (5-FU+cisplatin) chemotherapy and capecitabine combined with RT for advanced gastric cancer; in their study, capecitabine was administered at a dose of 825 mg/m2 twice daily throughout the duration of RT . Jansen et al. evaluated the dose escalation of capecitabine monotherapy concurrently with postoperative RT in gastric cancer and recommended a capecitabine dose of 1000 mg/m2 twice daily on each treatment day during the RT period .
Investigation of issues related to the sequence of ACT or ACRT is important since poor compliance to adjuvant treatment after gastrectomy is the main problem that may affect patient prognosis. Theoretically, for patients with high-risk pathological features (such as poorly differentiated cancer, lymphovascular invasion, or multiple lymph nodes metastasis) leading to a higher probability of distant failure, more cycles of ACT may be administered soon after surgery to avoid more cancer cell micrometastasis. However, excessive chemotherapy before RT would reduce patient tolerance to ACRT. Soyfer et al. reported an association between total RT treatment time and, to some extent, the time of the initiation of RT on local control and distant metastases . McMillan et al. reported that prolonged intervals between surgery and RT initiation were not associated with inferior OS in gastric cancer, while prolonged RT treatment duration was . In the studies reported by Janson et al., RT started one 21-day cycle of ACT after surgery, which means that patients might tolerate ACRT well compared to the tolerance observed in those who received several cycles of combination chemotherapy before RT [24, 27, 28]. With recent randomized evidence reinforcing the beneﬁt of ACRT in node-positive gastric cancer [11, 29], it is desirable to explore issues related to the proper sequence of ACT and ACRT, since poor compliance to adjuvant treatment after gastrectomy is the main problem that may impact patient prognosis. Based on our clinical experience, compliance to ACRT would be better if it started after no more than four cycles (21 days/cycle) of ACT for patients with many adverse prognostic factors. Furthermore, monotherapy administered as concurrent chemotherapy during RT, rather than as part of a combination chemotherapy regimen, would also improve patient compliance to ACRT. Thus, in our opinion, the adjuvant treatment design of the ACRT arm in the ARTIST trial seems reasonable (two cycles of capecitabine plus cisplatin followed by capecitabine-based ACRT and then two additional cycles of capecitabine plus cisplatin) .
The most commonly observed grade 3/4 hematologic and gastrointestinal toxicities in this study were leukopenia (12.5%) and vomiting (10%), which were much less frequent than those in INT 0116 (54% and 32% of the patients developed grade 3/4 hematologic and gastrointestinal toxicity) and CALGB 80801 study (about 50% and 16% of the patients developed grade 3/4 hematologic and gastrointestinal toxicity) [1, 30]. The exclusion of the remnant stomach from the target volume and the use of IMRT technology and capecitabine monotherapy (noninferior efficacy and lower gastrointestinal toxicity than 5FU) may account for the relatively lower rate of severe toxicities. Nam et al demonstrated that the exclusion of the remnant stomach from the radiation field could significantly reduce acute side effects without compromising long-term survival rates .And after the long-term follow-up of ARTIST study which CTV did not include remnant stomach, local recurrence in the remnant stomach was seen in only 2% of all patients, and this result was similar to Nam et al . RT fields including or excluding the remnant stomach. Several studies have found that IMRT was superior to two-dimensional or three-dimensional RT, providing a more conformal and homogeneous dose to the PTV and accordingly minimizing the probability of toxicity [32-34].We had previously determined that tomotherapy is a better option for adjuvant treatment of gastric cancer due to its superior bowel and bone marrow dose sparing, dose conformity, and dose homogeneity [6, 7, 35, 36]. Given these evidences, the present study showed acceptable safety and comparable compliance with the treatment course. Our study showed that 95% (38/40) and 72.5% (29/40) of patients complete RT and concurrent capecitabine monotherapy, respectively.
The 3-year DFS of the ACRT arm in INT 0116 was used in the power calculation for the present phase II study, as this is the only randomized trial evaluating the effect of ACRT in gastric cancer patients with an LND level less than D2. However, the final 3-year DFS in our study was 66.2%, which did not meet the primary endpoint (3-year DFS = 70%). This could be attributable to the max number of PLNs found (as high as 7) and the fact that only 55% of our patients had D2 LND. Despite the previous findings, our results are still better than those obtained with ACRT treatment by Janson et al [27, 28].The 2-year OS of their phase II trials evaluating capecitabine/cisplatin chemotherapy with concurrent RT after D0/1/2 LND (18%-22% of patients had D2 LND) was 45%-61% .According to A National Cancer Data Base Analysis, they find that patients with adjuvant RT 5-year OS rate was 45%. While our study showed a 3-year OS of 75% and 5-year OS of 58.9%. The higher incidence of D2 LND performed during radical gastrectomy in the present study may have contributed to our better prognosis. Subgroup analysis of the ARTIST trial showed that the significant 3-year DFS effect of ACRT in node-positive disease improved from 72% to 78%, which may be due to the very high-quality D2 LND (median number of lymph nodes dissected was 40) and relatively lower rate of metastatic lymph nodes (median number was 3) .
This study has several limitations that warrant emphasis. Due to the poor patient recruitment for this study, we did not limit the regimens or cycles of adjuvant chemotherapy administered before or after ACRT. Accordingly, this may have influenced the results for the toxicity profile of ACRT and led to different intervals between surgery and initiation of ACRT. However, patients were presumably recruited postoperatively, yielding a subgroup of patients who had undergone surgery. This is relevant if comparisons are to be made with other treatment strategies where patients are recruited preoperatively.