We proceeded to conduct a complete review of the literature with the support of the Federate Library of Medicine of Turin. The body of evidence for this review was primarily comprised of mature RCT data and meta-analyses of RCTs.

We searched PubMed, Embase, American Society of Clinical Oncology (ASCO) and European society of Medical Oncology (ESMO) meeting abstracts as well as the Cochrane Library by computer for non-randomized and randomized works published from 1993 to March 2017 (Figure 1).

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Literature selection flowchart.

Keywords were matched and included “NAC or neoadjuvant chemotherapy or neo-adjuvant chemotherapy or neo-adjuvant chemo-therapy or neoadjuvant chemo-therapy or preoperative chemotherapy or pre-operative chemotherapy or preoperative chemo-therapy or pre-operative chemo-therapy or perioperative chemotherapy or peri-operative chemotherapy or peri-operative chemo-therapy or perioperative chemo-therapy ”, “GC or gastric cancer or stomach cancer or stomach neoplasm or gastric neoplasm or gastric adenocarcinoma or stomach adenocarcinoma”, and “RCT or randomized controlled trial or randomised controlled trial”.

All published trials comparing NAC-containing procedures vs no treatment before surgery in patients with gastric cancer (or gastric cancer + cardia cancer + lower-esophagus cancer) were included earlier in our analysis. Blinding the trial conditions was not necessary. There was no language restriction.

We screened titles and abstracts of all identified papers (5957) and included only trials that satisfied the following criteria: patient populations with gastric cancer (diagnosed and classified according to 6^th or 7^th TNM and/or Japanese Gastric Cancer Association) without age, gender and racial limitations[19-21]; intervention and a comparative intervention that was clearly documented (NAC + surgery -NAC+S versus surgery alone -SA) for GC regardless of the detailed NAC regimens that were administered, surgical techniques performed (type of gastrectomy and LN dissection), pathological classification, location and stage of the disease. All patients needed to have a history of potentially curative surgery.

NAC could be performed through oral or intravenous (IV), intraperitoneal (IP) and intra-arterial (IA) infusion.

Esophagogastric junction and lower esophagus cancer patients were considered only if they were enrolled together with proper gastric cancer patients in the same study[15,17,18,22]. In these cases, the results of tests for heterogeneity effects according to the site of the primary tumor were documented when reported by the authors.

When results were reported or updated in more than one publication, only the most recent paper was used.

At the end of the process, only RCTs were included in this review (Figure 1).

We included only GC patients according to the theory of site-dependent differences in tumor genomics and biology[23]. Therefore, studies recruiting lower-esophagus and cardia cancers without gastric cancer were excluded. Studies including GC and diseases other than GC (lower esophageal cancer, esophagogastric junction tumor) were also excluded unless analyses were conducted separately.

Studies of preoperative radiotherapy or immunotherapy were excluded if they were not associated with chemotherapy.

We excluded patients with history of prior treatment before entering the trial as well.

Any chemotherapy regimen performed before resective surgery with or without postoperative adjuvant treatment was included and located in the NAC-arm.

For the surgery alone-arm (SA-arm), we enrolled all types of gastrectomy and lymphadenectomy that were performed (D0, D1, D1plus, D2, D3).

Studies in which GC was associated with another synchronous malignancy or studies containing multivisceral resections were excluded.

The primary outcomes were the overall survival rates (time from random assignment to the last follow-up or death) and/or disease-free survival rates (DFS) (time from random assignment to tumor recurrence or death) and/or disease-specific survival rates (DSS) (time from random assignment to death from disease). Secondary outcomes included the perioperative response rates, the R0 (margin negative) resection rate after perioperative treatment, and OS or DFS or DSS according to the type of lymph node dissection performed (D1, D2, others) following Japanese Gastric Cancer Treatments Guidelines[21]. Preoperative tumor stages were all recorded following TNM classification of malignant tumors (mostly from the 6^th edition[19]).

The objective response to NAC was evaluated as either complete response (CR), partial response (PR), stable disease (SD) or progressive disease (PD) according to the indications of the Japanese Gastric Cancer Association[21].

Some trials assessed the down-staging effect of NAC by separating out patients with negative nodes from patients with positive nodes on pathological examinations after surgery.

The quality of the included RCTs was assessed using modified Jadad’s scoring system[24] and Cochrane reviewers’ handbook 5.0.1 RCT criteria[25]. The assessment was based on the randomization methods, the report of dropout rates, allocation concealment, the use of intention-to-treat (ITT) analysis, and losses to follow-up, the extent to which valid results were depicted. Based on these criteria, the studies were divided into high quality group (score ≥ 4) and low quality group (score < 4) (Table 1).

Figure 1 shows the literature trial selection flowchart. In brief, 5957 studies were identified from the literature search. Out of all these studies, a total of 202 relevant papers documenting the results of any comparison of NAC + S vs SA for gastric cancer were selected. Then, we proceeded to a further selection of 19 papers reporting the results of controlled trials that were described as randomized in the title, (6 abstracts and 13 full-text papers). Nevertheless, a more detailed examination showed that in 3 of these studies, patient allocation had in fact not been conducted randomly[26-28]. Therefore, only 16 papers were proper randomized controlled trials that fulfilled our research criteria and were included in this review.

These RCTs were published between 1993 and 2011 with a 24 to 83 mo follow-up period. A total of 2062 patients was included in the analysis with 1089 receiving NAC (52.81%) and 973 (47.18%) undergoing SA.

In Yonemura’s trial[29] (1993), fifty-five patients with far advanced gastric cancer (TNM stage IV) were enrolled with 29 who received neoadjuvant treatment and 26 who underwent surgery alone (Table 2). NAC patients were treated with a PMUE regimen, a combination of cisplatin (CDDP) 75 mg/m², mitomycin C (MMC) 10 mg/body, etoposide 150 mg/body, and 400 mg/d UFT. The author reported response rates, resectability rates with curative intent and a survival rate advantage in the NAC group, but, overall, this was not statistically significant. Furthermore, the type of lymph node dissection performed in both arms was not reported.

In 1995, Shchepotin et al[30] included 146 GC patients in a RCT in which the tumor stage was not specified (Table 2). Fifty patients were treated with surgery alone, and 96 patients were submitted to NAC. Out of these patients, 49 received intravenous (systemic) chemotherapy (IVCH) and 47 received super selective intra-arterial chemotherapy (IACH). The chemotherapy regimen was not mentioned. Only IACH + S showed a response rate and a survival rate advantage over surgery alone, and this survival benefit was statistically significant (P < 0.01). Patient selection and the type of lymph node dissection performed were not reported.

The Kang et al[31] trial enrolled 107 patients with operable gastric cancer at TNM stages III and IV. Fifty-four patients received surgery alone and 53 received NAC with a PEF regimen followed by surgery. The response rate was not specified. The curative resection rate was higher in the NAC + S group than in the SA group. However, there was no significant difference in overall survival between the two arms (P = 0.114). The type of lymph node dissection that was performed was not reported (Table 2).

Lygidakis et al[32] recruited 58 patients with resectable gastric cancer at all stages. Patients were randomly assigned to 3 groups. Group A (No. 20) patients received preoperative hypoxic upper abdominal chemotherapy using mitomycin-C, 5-fluorouracil, leucovorin, and farmorubicin combined with adjuvant systemic chemotherapy; Group B (No 19) patients received preoperative hypoxic upper abdominal chemotherapy without systemic chemotherapy; and Group C (No 19) patients received surgery alone. Improved survival was reported by combining hypoxic upper abdominal perfusion NAC with surgery compared to surgery alone. Response rate, resectability rate and type of LN dissection were not documented (Table 2).

Two-hundred-sixty-two patients with resectable GC mostly with serosal invasion were treated by Takiguchi et al[33] with either SA (239 patients) or NAC (123) between May 1993 and May 1998. NAC patients received 5-FU (F group, n = 39) until the day before surgery or 5-FU combined with CDDP (FP Group, n = 84). NAC with 5-FU combined with low-dose CDDP was reported to reduce the rate of intraperitoneal positive cytology, increase the rate of successful resections and improve OS without statistical significance. The extension of the lymph node dissection was not described (Table 2).

In 2000, Wang et al[22] published the results of an RCT that included sixty patients with gastric cardia cancer that were randomly assigned to surgery alone (No. 30) or to NAC with 5-FU (No. 30). As Siewert III cancers are nowdays considered as proper gastric cancers, this study was included in the study despite our inclusion criteria[34]. The 5-year OS rate was improved in the NAC group but this survival benefit was not statistically significant. The resectability rate and type of node dissection performed in both groups were not reported (Table 2).

Kobayashi et al[35] performed a multicenter randomized clinical trial by recruiting 171 patients with advanced gastric cancer (a more detailed TNM tumor stage was not specified). These patients were randomized either to NAC with oral 5’-DFUR (No 91) followed by surgery and adjuvant therapy with iv MMC and oral 5’-FUDR or for surgery alone (No. 80) followed by the same adjuvant therapy. This clinical trial failed to demonstrate any survival benefit of preoperative chemotherapy over surgery alone. Response rate, resectability rate and the type of node dissection performed in both arms were not documented (Table 2).

In 2004, Hartgrink et al[4] reported the long-term results of the Dutch randomized FAMTX trial. In this trial, 59 AGC patients were recruited with no further details about the stage of the disease. Of these patients, 29 were allocated to the FAMTX regimen prior to surgery, while 30 patients received surgery alone. The resectability rates were equal for both groups. A complete or partial response was registered in 32% of the FAMTX group. The median survival since randomization was 18 mo in the FAMTX group vs 30 mo in the SA group (P = 0.17). Therefore, the trial could not show a beneficial effect of preoperative FAMTX. The standard surgical procedure included a limited (D1) lymphadenectomy. The authors concluded that surgery alone is the best treatment for operable gastric cancer (Table 2).

In 2004, Nio et al[36] published the results of a RCT in which 295 patients with GC of all stages were included (more than 50% were stage I). Overall, 193 patients were randomly allocated to the SA-arm and 102 were allocated to the NAC-arm with the UFT regimen (Tegafur + 5-FU). The response rate to NAC was 33.3%. The resectability rate was not reported. There were no significant differences in the OS rates between the NAC and SA groups. A survival benefit of NAC over SA was documented only in stages 2 and 3 and was higher in stage 2 and 3 patients with a complete or partial response to NAC compared to non-responders. The type of LN dissection performed was D2 only in 48% in the SA arm and 56% in the NAC-arm.

Zhao et al[37] studied the apoptosis induced by preoperative oral 5’-FUDR administration in GC and its mechanism of action. Sixty patients with gastric cancer of all TNM stages were randomly assigned to 3 groups (20 each group) before surgery. Group one patients received a NAC treatment with 5’-DFUR oral administration; group two patients received 5-FU + CF by venous drip for 3-5 d; and group three patients received surgery alone. The authors documented a significant increase in the apoptosis of gastric carcinoma cells and a decrease in the tumor cell proliferation index in group one patients after oral 5’-FUDR administration compared to groups 2 and 3. Nevertheless, preoperative NAC did not improve patient prognosis. The resectability rate and type of lymph node dissection were not reported.

In the most known RCT on NAC (MAGIC Trial) published in 2006 in the New England Journal of Medicine (NEJM), Cunningham et al[17] studied the effects of preoperative treatment on 503 recruited patients with all stages of resectable adenocarcinoma of the stomach, esophagogastric junction, or the lower esophagus (Table 2). Patients were randomly assigned to either a perioperative treatment (250 patients) followed by surgery or surgery alone (253 patients). Perioperative chemotherapy consisted of three preoperative and three postoperative cycles of i.v. epirubicin (50 mg/m²) and cisplatin (60 mg/m²) on day 1 as well as a continuous i.v. infusion of fluorouracil (200 mg/m² per day) for 21 d. The resected tumors were significantly smaller and less advanced in the perioperative-chemotherapy group (P < 0.001), and among patients undergoing resection, there was a greater proportion of stage T1 and T2 tumors in the perioperative-chemotherapy arm compared to the surgery alone arm (P = 0.002). Compared with the surgery alone arm, the perioperative chemotherapy arm had a higher likelihood of overall survival (hazard ratio for death, 0.75; 95 percent confidence interval, 0.60 to 0.93; P = 0.009; five-year survival rate, 36% vs 23%) and progression-free survival (hazard ratio for progression, 0.66; 95 percent confidence interval, 0.53 to 0.81; P < 0.001). Nevertheless, the tumor was located in the stomach in 74% of cases in the NAC- arm and 73.9% in the surgery alone-arm; the tumor originated from the lower esophagus or from the esophago-gastric junction in the ramaining cases. D2 lymphadenectomy was performed only in 40% of patients in the SA arm and in 42% of patients in the NAC arm. Furthermore, according to tests for heterogeneity of the treatment effects related to the baseline characteristics of patients, the survival benefit of perioperative chemotherapy was stronger for the esophagogastric junction and the lower esophagus site of the primary tumor compared to the proper stomach site. Resectability rates were not significantly different among the two groups.

Schuhmacher et al[15] in 2010 reported the preliminary results of the European Organisation for Research and Treatment of Cancer ( EORTC) randomized trial Nr. 40954 that compared NAC to SA for locally advanced cancer of the GC and EGJ (Table 2). The original trial design required that 360 patients with locally advanced (UICC stages III and IV cM0) adenocarcinoma of the stomach or esophagogastric junction (Sievert II and III) were randomly assigned to preoperative chemotherapy with cisplatin, d-L-folinic acid and fluorouracil after surgery or surgery alone. This trial was stopped for poor accrual after 144 patients were enrolled, including 72 patients in the surgery alone arm and 72 in the perioperative-chemotherapy treatment arm. More than half of patients had tumors located in the proximal third of the stomach, including EGJ Siewert type II and III. The trial showed a significantly increased R0 resection and pN0 rates in the NAC arm (P = 0.036) compared to the SA arm, but failed to demonstrate a survival benefit of perioperative treatment. In this trial, adequate surgery with D2 lymph node dissection was performed in 92.6% of SA arm patients and in 95.7% of NAC + S arm patients. The survival rates at 2 years were 72.7% (95%CI: 60.7%-81.7%) and 69.9% (95%CI: 57.7%-79.2%) in the neoadjuvant and surgery-only arms, respectively. The better outcome than expected after surgery alone was ascribed to the high quality of surgery and nodal dissection that was performed.

Imano et al[38] in 2010 reported the results of a four-arm randomized trial designed to evaluate the effects of short-term NAC (chemotherapy performed for 72 h before surgery) on the proliferative ability of cancer cells in gastric cancer. Sixty-three patients with gastric cancer were randomly assigned to 4 groups as follows: Group F, which had a single administration of 5-FU (16 patients); Group C, which had a single administration of CDDP (15 patients); Group FC, which had administration of both 5-FU and CDDP; and the Control Group, which had surgery alone (16 patients). Gastric cancer TNM or UICC stages were not reported. The D2 lymph node dissection was performed in all patients.The authors found that the combination of CDDP and 5-FU reduced proliferative ability and increased cellular apoptosis in gastric cancer cells. Nevertheless, there were no differences in overall survival rates between each group. Resectability and R0 resection rates were not reported (Table 2).

A group from the European Institute of Oncology in Milan[39] published the data of a multicenter RCT in which the objective was the non-inferiority of preoperative NAC with TCF regimen followed by surgery (Arm A) compared to surgery followed by the same chemotherapy regimen (Arm B) on clinical outcomes (morbidity and mortality of surgery) in GC patients (Table 2). Based on the initial design, a target sample size of 240 patients with locally advanced GC was required. Nevertheless, this trial was prematurely stopped at 69 randomized patients due to insufficient accrual with 34 in Arm A and 35 in Arm B, D2 lymphadenectomy was performed in almost all cases in both arms. The resectability rate was greater in the Arm A compared to the Arm B. In conclusion, preoperative TCF was safe and had similar morbidity compared to surgery followed by a TCF regimen. Furthermore, due to its early discontinuation for slow accrual, this study did not provide information about the efficacy of preoperative TCF, and no survival results were reported.

Qu et al[40] enrolled 78 patients with cTNM stage III or IV (M0) gastric cancer (Table 2). Thirty-nine of these patients were randomized to the NAC arm (paclitaxel combined with FOLFOX4 regimen) and 39 to the surgical arm. The clinical response rate was 66% in the NAC arm. The R0 resection rate was significantly higher, and the number of lymph node metastases was significantly lower in the NAC arm compared to the SA arm. In addition, the 2-year survival rate was higher in NAC- than in the surgery alone-arm. The location of the disease and type of lymph node dissection performed were not described. The sample size was very small.

The results of the FNCLCC and FFCD multicenter Phase III trial[18] were published in the Journal of Clinical Oncology in 2011 (Table 2). In this trial, 224 patients with resectable adenocarcinoma of the lower esophagus, gastroesophageal junction (EGJ), or stomach were randomly assigned to either perioperative chemotherapy with cisplatin and a fluorouracil regimen (NAC arm, n = 113) followed by surgery or surgery alone (SA arm, n =111). The curative resection rate was significantly higher in the NAC arm compared to that in the SA arm (84% vs 73%, P = 0.04). A non-significant decrease of lymph node metastases in the NAC arm compared to that of the SA arm was reported (67% vs 80%; P = 0.54). Compared to the SA group, the NAC group showed a better OS (5-year rate 38% vs 24%; HR for death 0.69; P = 0.02) and a better disease-free survival (5-year rate: 34% vs 19%; HR for death 0.65; P = 0.003). Nevertheless, in this trial, only 25% of patients had a tumor located in the stomach, while in 64% of cases, the tumor site was the esophagogastric junction, and another 11% of patients had cancer of the lower esophagus. While an extended lymphadenectomy (D2) was recommended in the “Study Treatment” protocol, the lymph node dissections performed in both arms were not described in the results. The authors concluded that compared to surgery alone, perioperative chemotherapy using a combination of cisplatin and fluorouracil significantly improved the curative resection rate, OS and DFS of patients with adenocarcinoma of the lower esophagus, EGJ and stomach. Nevertheless, this strong conclusion is an evident contrast to their previous comment reported in the “RESULTS” section of this paper where they stated that although the multivariable analysis showed the two significant prognostic factors for OS were the administration of a preoperative chemotherapy (P = 0.01) and tumor site (P < 0.01), the chemotherapy effect was only significant in the esophagogastric junction subgroup, which included around two-thirds of the patients. The two other subgroups (lower esophagus and stomach) were too small to distinguish between no effect and a small effect. Should these comments have been emphasized in the conclusion of the paper as well?

As already reported above, several studies from either eastern or western countries have reported survival rates after adequate surgery with extended nodal dissection that were significantly higher than rates reported both in the MAGIC[17] and FNCLCC-FFCD[18] trials after neoadjuvant treatment followed by incomplete surgery. Furthermore, these studies consider only cancer arising from the stomach, while the two RCTs included the lower esophagus and EGJ cancers as well (Table 4).

Recently, Maruyama et al[48] reported an overall 5-year survival rate of 70.1% among all UICC stages of the disease (data from Japanese Nationwide Registry of 11.261 gastric patients over the period 1991-2009) after surgery alone with accurate D2 lymph node dissection. Removing data for Stage I patients, which usually do not have indications for preoperative treatment, the rates referring to UICC stage II and III, which are commonly submitted to NAC according to the current guidelines, were 73% and 44.5, respectively and were still significantly higher with respect to figures reported in the neoadjuvant settings discussed above (Table 4).

In a recent RCT from Taiwan, Wu et al[53] observed an overall 5-year survival of 59.5% after extended surgery, and the rate of T1 patients in this study population was 26% (Table 4).

In 2017, Kim et al[49] published survival data as observed in high volume centers in Korea. Overall, the 5-year survival was surprisingly very high at approximately 80% in all patients, but AJCC stage Irepresented 63% of the study population. Anyway, the 5-year survival rates for stage II and III were still 86.5% and 63.7%, respectively (Table 4).

Survival rates after SA higher than those published in MAGIC and FNCLCC-FFCD neoadjuvant settings have been reported also in the western world, both in non-randomized and randomized studies on D2 gastrectomy. In historical non-randomized trials published in the 80s and 90s in Germany[50], England[51] and Norway[54], the 5-year OS rates coming from reference centers after extended procedures were reported as 46.5, 55% and 47%, respectively. In all these study populations, T1 cancers represented less than 10% of all patients (Table 4).

In historical randomized trials from Honk-Kong[55] and South Africa[56], despite evident limitations due to the low accrual, OS after D2 gastrectomy were reported as 35% and 67%, respectively (Table 4).

Additionally, in more recent RCTs coming from Europe, the Dutch[57] and the IGCSG[13] trials, the rates of overall survival observed after proper D2 gastrectomy alone were at least comparable to the results reported by Cunningham and Ychou in the MAGIC and FNCLCC-FFCD RCTs after chemotherapy followed by incomplete surgery. The results of the Dutch trials were first published in the NEJM in 1999. The 5-year OS rate after extended nodal dissection was 47% in all patients. It was 77%, 44% and 22% in T1, T2 and T3 patients, respectively, and 69% and 30% in lymph node-negative or lymph node-positive patients, respectively[57] (Table 4).

The survival results of the IGCSG RCT were published in 2014 in the British Journal of Surgery[13]. The overall 5-year survival and the 5-year disease-specific survival rates were 64.2% and 72.6%, respectively, for the whole cohort of patients submitted to D2 lymphadenectomy. Tumors were stratified by the depth of invasion and by lymph node involvement, and both the OS and DSS were calculated for pT2-pT4 patients and for pT2-pT4 with node-positive patients. In these two subsites of patients (which represent a common indication for neoadjuvant treatments), the 5-year OS and DSS were 59% and 69%, respectively. The observed morbidity and mortality were extremely low (24% and 2.2%, respectively), and therefore, they did not negatively impact the effects of extended surgery on survival, in contrast to what was described previously by Cuschieri et al[58] and Bonenkamp et al[57] in their studies. This trial is a clear demonstration that adequate surgery with proper extended nodal dissection with low complications and mortality can provide impressive survival results that are even significantly higher than those reported in trials adopting neoadjuvant treatments in a similar patient population (Table 4).

On the other hand, due to the high rates of morbidity and mortality, the MAGIC trial designed by Cuschieri to demonstrate the superiority of the extended gastrectomy over the standard D1 procedure in Great Britain failed to document a survival advantage[58,59].

In fact, the OS reported by Cuschieri 5 years after surgery was low with 33% in the D2 arm and 35% in the D1 arm, which was only comparable to the results observed in the MAGIC and FNCLCC-FFCD trials when taking all patients into consideration. It was even lower if we removed AJCC Stage I patients. In fact, in the remaining AJCC stage II and stage IIIpatients, it was 31% and 11%, respectively. These poor survival rates have been related to a combination of the observed severe morbidity and mortality with the inadequacy of the procedures performed, including many protocol violations, especially due to high rates of noncompliance (e.g., non-removal of several Ln stations that were expected to be removed during lymph node dissections). A few years later, the positive results of the MAGIC trial on NAC do seem a direct consequence of this lack of quality and adequacy of surgical treatment provided to gastric cancer patients in many upper GI centers from Great Britain until the revolutionary centralization of treatments adopted since recent years. The chemotherapy effect was more evident when the quality of surgery was lower.

In conclusion, neoadjuvant (preoperative or perioperative) chemotherapy for GC has been rapidly introduced in international western guidelines without evidence-based medicine related demonstrations of its efficacy for proper stomach cancer in patients who receive a complete extended (D2) lymph node dissection. The currently available data support the adoption of NAC procedures in several national guidelines based on low-volume study populations with mixed sites of the primary disease (lower esophagus, esophagogastric junction and stomach), and statistical tests have often shown evidence of the heterogeneity of chemotherapy effects according to the site of the primary tumor. In both MAGIC and FNCLCC-FFCD trials, tests for heterogeneity showed that perioperative chemotherapy effects were much more evident for esophagogastric junction cancer and much less for stomach cancer. Furthermore, in both these RCTs supporting the adoption of NAC, surgery provided to patients was incomplete because extended nodal dissection was performed in less than 50% of patients or not described, and no results on the LN yield were documented to warrant its adequacy. The only well-designed RCT with extended surgery performed in almost all cases was prematurely ended for poor accrual, but at the moment of its closure, it failed to demonstrate a survival benefit for NAC.

Moreover, in national reference centers for gastric cancer care and research, OS and DSS rates of subsites of patients theoretically fit for NAC submitted to surgery alone with complete D2 dissection and performed for cancers located in the stomach only, are even higher than those reported after NAC followed by incomplete surgery provided to a patient population with mixed sites of the primary tumor.

We think additional high volume sample-size multicenter (and eventually multinational) RCTs comparing newer treatment regimens of neoadjuvant settings (capecitabine, oxaliplatin, docetaxel) combined with molecular therapies (epidermal growth factor receptor inhibitors or antiangiogenic agents) followed by controlled D2 surgery with controlled D2 surgery alone are necessary to investigate the survival effects of modern preoperative chemotherapy in patients with only stomach cancer who receive proper extended surgery.

Actually, curative surgery with complete D2 lymph node dissection is the treatment of choice for gastric cancer (GC). In high-volume reference centers surgery alone gives excellent survival rates. Anyway, in western countries, prognosis remains poor outside reference centers even after curative resection, especially for the high rate of recurrence mainly due to incomplete nodal dissection. Many trials have investigated the impact of adjuvant treatment by comparing surgery alone with surgery followed by adjuvant chemotherapy, but there is no definitive evidence of any survival advantage.

Therefore, in the last decade, a multimodal approach of GC has been suggested with the adoption of preoperative or perioperative treatment (NAC). The problem is that NAC has been introduced in the national guidelines of many countries, especially after publication of MRC and French RCTs, without a clear and evidence-based medicine (EBM)-related demonstration of a survival benefit compared to controlled surgery with proper enlarged LN dissection in patients with only stomach tumors.

In fact, MAGIC and FNCLCC-FFCD RCTs have shown a survival advantage of NAC + surgery vs surgery alone, but surgery performed in these patients did not include a proper extended lymphadenectomy in the majority of cases, and many patients had lower esophageal or cardia cancer instead of only GC. Therefore, it is possible that a proper radical surgery (gastrectomy with D2 LN dissection) may nullify the survival benefit of NAC described after incomplete surgery and also that the location of the tumor outside the stomach (cardia and lower esophagus), may amplify the response to preoperative treatment.

To document a possible benefit of NAC after adequate D2 gastrectomy for only stomach cancer, we proceeded to review the literature on neoadjuvant treatment for resectable gastric cancer by mainly investigating these variables.

Neoadjuvant (pre- or peri-operative) chemotherapy has been recently introduced in the national guidelines of many countries mainly after publication of MRC and French RCTs, without a clear and evidence-based medicine (EBM)-related demonstration of a survival benefit compared to controlled surgery with proper enlarged LN dissection in patients with only stomach tumors. The advantage reported after NAC seems mostly related to incomplete nodal dissection performed in both arms of these trials and to heterogeneous recruitment of lower esophagus and esophago-gastric junction cancer together with proper stomach cancer. Are there any trials among RCTs available in Literature investigating the effect of NAC over an homogeneous population of patients with only-stomach cancer treated with complete D2 lymph node dissection in both arms? Otherwise we could not exclude that: 1) a proper radical surgery (gastrectomy with D2 LN dissection) may nullify the survival benefit of NAC described after incomplete surgery and 2) the location of the tumor outside the stomach (esophago-gastric junction and lower esophagus) may amplify the response to preoperative treatment. In other words, NAC could be unuseful in proper gastric cancer submitted to curative complete D2 dissection.

The main objective was to investigate whether a possible survival benefit (OS and/or DSS and/or DFS) of NAC after adequate D2 gastrectomy for only stomach cancer was documented in all RCTS available in Literature till now. Secondary outcomes included the perioperative response rates, the R0 (margin negative) resection rate, and OS or DFS or DSS according to the type of lymph node dissection performed (D1, D2, others) after neoadjuvant treatment. We realized that no RCTs available till now could document a survival benefit of NAC over surgery alone for homogeneous population of patients with tumor of the only stomach treated with complete D2 gastrectomy and that a further large volume randomized trial is needed.

We proceeded to conduct a complete review of the literature with the support of the Federate Library of Medicine of Turin. The body of evidence for this review was primarily comprised of mature randomized controlled trial (RCT) data. All published randomized trials comparing NAC-containing procedures vs no treatment before surgery in patients with gastric cancer (or gastric cancer + cardia cancer + lower-esophagus cancer) were included in our analysis. Blinding the trial conditions was not necessary. There was no language restriction. Esophagogastric junction and lower esophageal cancer patients were considered only if they were enrolled together with proper gastric cancer patients in the same study. The reason for considering also esophago-gastric junction and lower esophageal cancers in the recruitment process was that these locations represented a consistent part of the study population of the main RCTs actually claimed to be the basis for NAC treatment in gastric cancer.

At the end of the process, only RCTs were included in this review. As we wanted to include all RCTS available in Literature at the moment, any chemotherapy regimen performed before resective surgery with or without postoperative adjuvant treatment was included and located in the NAC-arm as well as in the surgery alone-arm (SA-arm) we enrolled all types of gastrectomy and lymphadenectomy that were performed (D0, D1, D1plus, D2, D3). The quality of the included RCTs was assessed usin nodified JADAD’s scoring system and Cochrane reviewers’ andbook 5.0.1 RCT criteria.

The currently available data support the adoption of NAC procedures in several national guidelines based on low-volume study populations with mixed sites of the primary disease (lower esophagus, esophago-gastric junction and stomach). Statistical tests have often shown evidence of the heterogeneity of chemotherapy effects according to the site of the primary tumor. In fact, in both Magic and FNCLCC-FFCD trials, tests for heterogeneity showed that perioperative chemotherapy effects were much more evident for esophago-gastric junction cancer and much less for stomach cancer.

Furthermore, in both these RCTs supporting the adoption of NAC, surgery provided to patients was incomplete because extended nodal dissection was performed in less than 50% of patients or not described, and no results on the LN yield were documented to warrant its adequacy. The only well-designed RCT with extended surgery performed in almost all cases was prematurely ended for poor accrual, but at the moment of its closure, it failed to demonstrate a survival benefit for NAC.

Moreover, in national reference centers for gastric cancer care and research, OS and DSS rates of subsites of patients theoretically fit for NAC submitted to surgery alone with complete D2 dissection and performed for cancers located in the stomach only, are even higher than those reported after NAC followed by incomplete surgery provided to a patient population with mixed sites of the primary tumor.

This study documented that the adoption of NAC procedures in several national guidelines is actually based on low-volume study populations with mixed sites of the primary disease (lower esophagus, esophago-gastric junction and stomach) and with inadequate surgery as concern the extent of nodal dissection (incomplete D2 dissection). In fact, statistical tests reported in these trials have shown evidence of the heterogeneity of chemotherapy effects according to the site of the primary tumor, documented a major effect mainly over esophagogastric junction cancer. Moreover, survival rates reported in Surgery-Alone arm by main RCTs claimed to be fundamental for supporting the adoption of NAC ( MRC and French trials) are really too low as compared to rates documented in reference centers study population after proper D2 gastrectomy without preoperative chemotherapy. Effectively, in these surgical series coming from high volume referral centres, survival rates after extended surgery without NAC are even higher than those reported after NAC followed by incomplete surgery.

Our hypothesis is that, with current data available it ‘s not possible to exclude that: (1) The location of the tumor outside the stomach (esophago-gastric junction and lower esophagus) may amplify the response to preoperative treatment. (2) A proper radical surgery (gastrectomy with D2 LN dissection) may nullify the survival benefit of NAC described after incomplete surgery. (3) On the contrary, NAC effect on survival is more likely to be evident after incomplete surgery. In other words, NAC could be unuseful in cancer of the only stomach submitted to curative complete D2 dissection.

We think additional high volume sample-size multicenter (and eventually multinational) RCTs comparing newer treatment regimens of neoadjuvant settings (capecitabine, oxaliplatin, docetaxel) combined with molecular therapies (epidermal growth factor receptor inhibitors or antiangiogenic agents) followed by controlled D2 surgery with controlled D2 surgery alone are necessary to investigate the survival effects of modern preoperative chemotherapy in patients with gastric cancer only who receive proper extended surgery. This is the ony way to investigate if neoadjuvant therapy can positively impact survival also in homogeneous population of gastric cancer patients without any biases related to location of the tumor and adequacy of surgery administered.