Chemotherapy for patients with gastric cancer after complete resection: A network meta-analysis

Abstract

AIM: To conduct a network meta-analysis to evaluate the effectiveness of different chemotherapy regimens for patients with gastric cancer.

METHODS: PubMed (1966-2011.12), the Cochrane Library (2011 Issue 2) and EMBASE (1974-2011.12) were searched with the terms “gastric cancer” and “chemotherapy”, as well as the medical subject headings. References from relevant articles and conferences were also included. Patients who had previous gastric surgery, radiation before or after surgery or chemotherapy before surgery were excluded. In this study, only randomized controlled trials (RCTs) were considered, and the end-point was the overall mortality. Direct comparisons were performed using traditional meta-analysis whereas indirect comparisons were performed using network meta-analysis.

RESULTS: In total, 31 RCTs with 7120 patients were included. Five chemotherapy regimens, fluorouracil (FU) + BCNU, FU + methyl-CCNU (mCCNU), FU + cisplatin, FU + anthracyclines and FU + mitomycin c (MMC) + cytarabine (Ara-c), were found to be less beneficial in terms of overall mortality. In contrast, four chemotherapy regimens were effective for the patients after surgery, including FU + MMC + adriamycin (FMA), FU + MMC (FM), Tegafur and MMC, There was no significant difference in terms of overall mortality among these regimens. The evidence for the FM regimen and MMC regimen was poor. Additionally, the FMA regimen, which includes a variety of chemotherapy drugs and causes many side effects, was not better than the Tegafur regimen.

CONCLUSION: Although the four chemotherapy regimens were effective in patients with gastric cancer after surgery and the overall mortality revealed no significant difference among them in the network meta-analysis, thorough analysis of the results recommends Tegafur as the first-line adjuvant chemotherapy regimen for patients after complete resection.

Core tip: Although adjuvant chemotherapy after complete resection of gastric cancer is therapeutically useful, which of the many regimens is most effective? To date, no regimen has been clearly recommended as the standard procedure post-operation; therefore, we performed a network meta-analysis, which is a useful tool to summarize the different clinical trials and to evaluate the effectiveness of different chemotherapy regimens for patients after complete resection of gastric cancer. Based on our findings, the Tegafur regimen, especially S-1, is the first therapy that should be recommend to the patients to reduce overall mortality.

INTRODUCTION

Gastric cancer (GC) remains the second leading cause of cancer-related deaths in the world and is the most common malignancy in Asia, South America and Eastern Europe. The overall outcome for patients with GC has not significantly improved over recent decades[1-4]. GC remains a considerable threat to public health around the world. Currently, complete resection still has the highest potential for curatively treating GC[5]. However, approximately 20%-60% of GC patients who have already had curative surgery develop recurrent diseases[6] and will need to undergo adjuvant chemotherapy.

No network meta-analysis has been conducted to compare the efficacy of different chemotherapy protocols for patients with GC. Network meta-analysis is a useful tool for summarizing different clinical trials[7], especially when many different regimens are effective for the same clinical condition. In this type of analysis, all binary comparisons are shown with labels indicating superiority, inferiority or no difference in a summary graph[8-12]. Some recent meta-analyses have indicated that adjuvant chemotherapy after complete resection produces a small survival benefit[13-18]. Several additional trials have also been conducted in this setting. However, they did not indicate which chemotherapy protocol had the best efficacy for treating patients who have undergone complete resection. There is no clearly recommended protocol for the standard treatment of patients with GC after complete resection, and a 5-fluorouracil (5-FU) and platinum-based regimen is usually administered. Surgeons need empirical evidence to determine the best treatment for GC patients. Therefore, it was deemed important to assess the benefits of various adjuvant chemotherapy regimens through a network meta-analysis based on data from all relevant randomized controlled trials (RCTs).

The purpose of this network meta-analysis was to evaluate the effectiveness of different chemotherapy regimens for patients with GC who had undergone surgery.

MATERIALS AND METHODS

Study selection

PubMed (1966.01-2011.12), the Cochrane Library (2011 Issue 12) and EMBASE (1974.01-2011.12) were searched with the terms “gastric cancer” and “chemotherapy”, as well as the medical subject headings. The relevant articles referenced in these publications were downloaded from the databases. The related article function was also used to widen the search results. All abstracts, comparative studies, non-randomized trials, and citations scanned were searched comprehensively. Additional searches were conducted by reviewing abstract booklets and review articles. Trials were included irrespective of the language in which they were reported.

Data extraction

Each article was critically reviewed by two researchers for eligibility in our network meta-analysis (Table 1). Only RCTs on palliative or adjuvant chemotherapy for treating GC patients who had undergone surgery were analyzed in this network meta-analysis. The two researchers extracted the data separately, which were then confirmed by a third researcher.

Table 1.

Characteristics of randomized trials included in the network meta-analysis

Trial	Year	Postoperative chemotherapy regimens	Sample size		Overall mortality		Follow-up (mo)	Jadad score
			Chemotherapy group	Control group	Chemotherapy group	Control group
Lawton et al[20]	1981	FU + BCNU	13	12	11/13	10/12	60	2
Stablein et al[21]	1982	FU + MCCNU	71	71	29/71	40/71	48	3
Higgins et al[22]	1983	FU + MCCNU	156	156	121/156	117/156	36	3
Nakajima et al[23]	1984	FM + Ara-c	128	124	11/128	17/124	60	3
Engstrom et al[24]	1985	FU + MCCNU	91	89	57/91	51/89	24	3
Schlag et al[25]	1987	FU + BCNU	42	53	21/42	28/53	72	2
Bonfanti et al[26]	1988	FU + MCCNU	75	69	63/75	56/69	84	4
Coombes et al[27]	1990	FMA	131	148	101/133	123/148	68	3
Estape et al[28]	1991	MMC	33	37	16/33	31/37	120	2
Krook et al[29]	1991	FA	61	64	41/61	43/64	60	3
Kim et al[30]	1992	MMC + FU	77	94	54/77	71/94	60	2
Grau et al[31]	1993	MMC	68	66	40/68	49/66	105	2
Hallissey et al[32]	1994	FMA	138	145	101/138	110/145	60	3
Macdonald et al[33]	1995	FMA	93	100	59/93	68/100	114	2
Lise et al[34]	1995	FMA	155	159	88/155	99/159	78	3
Tsavaris et al[35]	1996	FMA	42	42	27/42	34/42	60	3
Cirera et al[36]	1999	MMC + Tegafur	76	76	33/ 76	44/72	37	3
Nakajima et al[37]	1999	MMC + FU + UFT	288	285	41/288	49/285	72	3
Neri et al[38]	2001	Epirubicin + FU	69	68	48/69	59/68	60	2
Bajetta et al[39]	2002	FU + Adriamycin etoposide + cisplatin	137	137	66/137	71/137	66	2
Nashimoto et al[40]	2003	MMC + FU + Ara C	128	124	11/128	23/124	69	2
Popiela et al[41]	2004	FAM	53	52	42/53	47/52	120	2
Chipponi et al[42]	2004	Cisplatin + FU	101	104	62/101	63/104	60	2
Bouché et al[43]	2005	Cisplatin + FU	127	133	68/127	77/133	97.8	3
Nitti et al[44]	2006	FU + Adriamycin + methotrexate + LV	103	103	54/103	49/103	60	3
Nitti et al[44]	2006	FU + Epirubicin + methotrexate + LV	91	100	63/91	64/100	60	3
De Vita et al[45]	2007	FU + Epirubicin + LV + etoposide	112	113	58/112	64/113	60	2
Nakajima et al[46]	2007	Uracil-Tegafur	95	95	18/95	30/95	60	4
Di Costanzo et al[47]	2008	FU + Epirubicin + cisplatin + LV	130	128	69/130	70/128	60	3
Miyashiro et al[48]	2011	Cisplatin + FU	132	132	50/132	52/132	60	4
Sasako et al[49]	2011	S-1	529	530	149/529	206/530	60	4

FU: Fluorouracil; MCCNU: Methyl-CCNU; MMC: Mitomycin c; LV: Leucovorin; Ara-c: Cytarabine; CDHP: 5-Chloro-2,4-dihydropyrimidine; Oxo: Potassium oxonate; FM: FU + MMC; FMA: FU + MMC + adriamycin; S-1: Tegafur + CDHP + Oxo.

Inclusion criterion: Patients with GC after complete resection and age < 71 years.

Exclusion criteria: Patients who had previous gastric surgery, radiation before or after surgery, chemotherapy before surgery, a history of deep venous thrombosis or pulmonary embolism and severe cardiovascular, respiratory, hepatic or renal disease.

End point: Overall mortality was defined as the time from randomization to death from any cause, or to the last follow-up, which was used as the date of censoring.

Quality evaluation

The quality of the studies included was assessed using the Jadad score[19].

Statistical analysis

The traditional meta-analysis method was used for extracting the crude rates of our pre-specified clinical end-point for each treatment group when the trials reported suitable information. We summarized the available data on overall survival from the reported results in all trials, computing pooled odd ratios and their respective 95% confidence intervals (95%CI) by means of a fixed-effects model. All statistical analyses were performed using Review Manager (RevMan version 5.0), the Cochrane Collaboration’s software for preparing and maintaining Cochrane systematic reviews. We used the chi-square statistic to assess the heterogeneity between trials and the I² statistic to assess the extent of inconsistency. Subgroup analysis was used to explore important clinical differences among trials that might be expected to affect the magnitude of the treatment effect.

Network meta-analysis was used after traditional meta-analysis. When efficient chemotherapy regimens were compared through network meta-analysis, the head-to-head comparisons (in this case, indirect comparisons) were handled and consequently assigned a statistical result in terms of superiority/inferiority or no difference along with the level of statistical significance. Statistical calculations and graph generation were carried out. The HR, with a 95%CI, for each indirect comparison was estimated according to the ITC software (Canadian Agency for Drugs and Technologies in Health, Indirect Treatment Comparison software, Ottawa, Ontario, Canada). This approach allows an indirect HR, with a 95%CI, to be estimated on the condition that both treatments included in the indirect comparison had been compared in actual trials against a common comparator.

Role of funding source

No sponsors were involved in the study design; during the collection, analysis, and interpretation of the data; in the writing of the report; or in the decision to submit the report for publication. All authors had access to the raw data. The corresponding author had full access to all of the data and the final responsibility to submit the report for publication.

RESULTS

Flow diagram of trial selection

In total, 31 RCTs, with a total of 7120 patients, were included (Figure 1) from the electronic databases. Figure 1 shows a flow chart of studies from the initial results of the publication searches to the final inclusion or exclusion. The RCTs that met the criteria for our analysis are described in Table 1. There were 12 RCTs that had a Jadad score of 2, 15 RCTs that had a Jadad score of 3 and 4 RCTs that had a Jadad score of 4.

Figure 1.

Flow diagram of trial selection.

Analysis of regimen groups

In terms of direct comparisons, this analysis divided the chemotherapy regimens into 9 subgroups, and 8 subgroups were assessed by the fixed effects models, while only 1 was assessed by the random effects models. In terms of overall mortality, at least 5 chemotherapy regimens were found to be of equal efficacy when compared to a blank control. The values of HR were as follows: 0.92 (95%CI: 0.43-1.96) for FU + BCNU regimen, 1.00 (95%CI: 0.76-1.32) for FU + methyl-CCNU (mCCNU) regimen, 0.93 (95%CI: 0.69-1.24) for FU + cisplatin regimen, 0.92 (95%CI: 0.74-1.14) for FU + anthracyclines regimen, and 0.67 (95%CI: 0.41-1.10) for FU + mitomycin c (MMC) + AraC regimen. In contrast, in terms of overall mortality, 4 chemotherapy regimens were found to be more effective than the blank control. The values of HR were as follows: 0.74 (95%CI: 0.58-0.94) for FAM regimen, 0.68 (95%CI: 0.49-0.94) for FM regimen, 0.60 (95%CI: 0.47-0.76) for Tegafur regimen, and 0.33 (95%CI: 0.13-0.86) for MMC regimen. These outcomes are described in Figures 2 and 3.

Figure 2.

Eight subgroups in the fixed effects models.

Figure 3.

One subgroup in the random effects model. FU: Fluorouracil; mCCNU: Methyl-CCNU; MMC: Mitomycin c.

In terms of indirect comparisons, 4 chemotherapy regimens were found to be equal in terms of overall mortality. The values of HR were as follows: 1.09 (95%CI: 0.73-1.63) for 5-FU + adriamycin + MCC (FAM) regimen vs FM regimen; 1.23 (95%CI: 0.88-1.73) for 5-FU + MMC + adriamycin (FMA) regimen vs Tegafur regimen; 2.24 (95%CI: 0.85-5.95) for FMA regimen vs MMC regimen; 1.13 (95%CI: 0.76-1.70) for FM regimen vs Tegafur regimen; 2.06 (95%CI: 0.76-5.60) for FM regimen vs MMC regimen; and 1.82 (95%CI: 0.67-4.80) for Tegafur regimen vs MMC regimen. These outcomes are described in Figure 4.

Figure 4.

Network meta-analysis in terms of mortality. MMC: Mitomycin c; FAM: 5-fluorouracil, adriamycin, and mitomycin c.

DISCUSSION

In total, 31 RCTs, with a total of 7120 patients, were included in this analysis, and 12 RCTs had a Jadad score of 2, 15 RCTs had a Jadad score of 3, and 4 RCTs had a Jadad score of 4. This study divided these chemotherapy regimens into 9 subgroups. The result of this analysis indicated that 5 chemotherapy regimens had little benefit to the patients, including the FU + BCNU, FU + mCCNU, FU + cisplatin, FU + anthracyclines, and FU + MMC + AraC regimens. In contrast, 4 chemotherapy regimens were effective for patients after surgery, including the FMA, FM, Tegafur, and MMC regimens. In this study, Tegafur and the S-1 regimen were assigned to one regimen because S-1 was composed of Tegafur, CDHP and Oxo, as CDHP and Oxo reduced the side effects of Tegafur. As Tegafur is a fluorouracil derivative, the FM regimen was included in 3 RCTs. Additionally, anthracyclines, including adriamycin, epirubicin and doxorubicin, were part of the FMA regimen, which was included in 6 RCTs. Indirect comparisons were estimated according to the ITC software, and the results indicated that there was no difference among these four chemotherapy regimens in the terms of overall mortality.

Although this analysis indicated that MMC was effective for patients after surgery, the evidence for this result was poor because of the low quality of the 2 RCTs included. Specifically, one trial had a small sample size, and only 204 patients were contained in the subgroup analysis. Additionally, because there was also significant heterogeneity among the trials (P = 0.14, I² = 54%), the analysis was carried out using the random effects models. The curative effect of MMC needs to be further validated. The evidence for the Tegafur regimen included 1249 patients, the RCTs were of high quality, and there was no significant heterogeneity among the trials (P = 0.59, I² = 0%). Accordingly, the analysis was carried out using the fixed effects model, and we found strong evidence to confirm the efficacy of the Tegafur regimen. The joint application with 5-chloro-2,4-dihydropyrimidine (CDHP) and potassium oxonate (Oxo) reduced the side effects of Tegafur; therefore, the S-1 regimen (Tegafur + CDHP + Oxo) is recommended.

The combination of Tegafur and MMC in the FM regimen was similar to treatment with each component individually, as determined by indirect comparison, and further studies are needed to confirm which treatment is the primary effector. Additionally, if the side effects of Tegafur and MMC will reduce the overall efficacy, further studies are needed to identify an adjuvant that can reduce these side effects, as in the case of S-1. If the treatments have a mutual antagonist effect on each other, they should be used separately. As the evidence for the FM regimen is not very strong, larger sample sizes and multicenter RCTs are still needed. While the FMA regimen is available, surprisingly, it is not better than Tegafur or MMC. Traditional analysis indicated that the FU + anthracyclines regimen is not available, and thus, MMC may contribute to the efficacy of the FMA regimen to a great extent. Accordingly, based on these results, FMA is not recommended.

In summary, chemotherapy regimens, especially Tegafur, are available for GC. However, the efficacy of the FM regimen and MMC regimen needs to be further validated. The evidence for the Tegafur regimen is more credible, and S-1 may be the best current choice. Future studies should focus on identifying better adjuvants that can reduce the side effects of MMC as much as possible. Their combination could be a better regimen than S-1, and perhaps, the combination of MMC, Tegafur and adjuvant can achieve better outcomes than mono-chemotherapy alone. However, based on recent evidence, the Tegafur regimen, especially S-1, is most commonly recommended to patients after complete resection.

In conclusion, this analysis indicated that four chemotherapy regimens are effective for patients with GC after surgery, including the FMA regimen, FM regimen, Tegafur regimen and MMC regimen. However, the evidence for the FM regimen and MMC regimen was poor in terms of overall mortality. The FMA regimen, which includes many chemotherapy drugs and thus has many side effects, is not better than the Tegafur regimen. Based on this study, the Tegafur regimen is recommended as a better choice for doctors when dealing with GC patients after complete resection.

COMMENTS

Background

Gastric cancer is very common worldwide and, in most cases, will lead to serious health problems, even after complete resection. Currently, treatment with adjuvant and palliative chemotherapies are essential to prevent and treat recurrence disease. A standard chemotherapy regimen has not been established; therefore, the evaluation of which regimens may be better for gastric cancer patients is needed.

Research frontiers

This network meta-analysis was performed to evaluate the effectiveness of different chemotherapy regimens for patients with gastric cancer. The end point was overall mortality, which was defined as the time from randomization to death from any cause, or to the last follow-up.

Innovations and breakthroughs

The meta-analysis shows the following: four chemotherapy regimens [fluorouracil (FU) + mitomycin c + adriamycin, fluorouracil + mitomycin c (FM), tegafur and mitomycin c (MMC)] are effective for patients after surgery, whereas the other five regimens [fluorouracil + BCNU, FU + methyl-CCNU (mCCNU), FU + cisplatin, FU + anthracyclines and FU + mitomycin c + cytarabine] were found to be less beneficial.

Applications

From the analysis, Tegafur is recommended as the first-line adjuvant chemotherapy regimen for patients after complete resection. This recommendation is due to the high quality of the randomized controlled trials (RCTs), homogeneity among trials and fewer side effects.

Peer review

The current network meta-analysis evaluated the effectiveness of different chemotherapy regimens for gastric cancer patients after curative surgery, and we found that the outcomes and analysis were good. However, further RCTs are needed to study the FM regimen, MMC regimen and combination chemotherapy.