Post-surgical chemotherapy versus surgery alone for resectable gastric cancer

  • Review
  • Intervention

Authors


Abstract

Background

For gastric cancer surgery is the mainstay treatment. Chemotherapy seems to improve the survival results. But chemotherapy is not a complication-free therapy and its role has been questioned by some trials.

Objectives

To determine whether post-surgical chemotherapy should be used routinely in resectable gastric cancer.

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL) in The Cochrane Library, MEDLINE, EMBASE, and Science Citation Index Expanded (July 2013).

Selection criteria

Randomised controlled trials (RCT) comparing post-surgical chemotherapy versus surgery alone for resectable gastric cancer.

Data collection and analysis

Two authors independently assessed trials for inclusion and independently extracted the data. We analysed the data with both the fixed-effect and the random-effects models using the RevMan analysis software. We calculated the hazard ratio (HR) with 95% confidence interval (CI) based on intention-to-treat or available case analysis.

Main results

The authors identified 34 studies (7824 patients) reporting overall survival (OS) and only 15 reporting disease free survival (DFS) as well. Post-surgical chemotherapy showed an improvement in OS (HR 0.85; 95% CI 0.80 to 0.90) and an improvement in DFS (HR 0.79; 95% CI 0.72 to 0.87), although all the trials had a high risk of bias.

The planned analysis of quality of life, return to work, and number of hospital admissions was impossible to complete as the outcome data for the analysis were not available from any trials.

Authors' conclusions

Post-surgical chemotherapy should be used routinely for resectable gastric cancer where possible. Further RCTs are needed to determine the role at each stage of disease.

Résumé scientifique

La chimiothérapie post-opératoire par rapport à la chirurgie seule pour le cancergastrique résécable

Contexte

Pour le cancer gastrique la chirurgie est le pilier du traitement. La chimiothérapie semble améliorer les résultats de survie. Toutefois, la chimiothérapie n'est pas un traitement sans complication et son rôle a été remis en cause par certains essais.

Objectifs

Déterminer si la chimiothérapie post-opératoire doit être systématiquement utilisée dans le cancer gastrique résécable.

Stratégie de recherche documentaire

Nous avons effectué des recherches dans le registre Cochrane des essais contrôlés (CENTRAL) dans la Bibliothèque Cochrane , MEDLINE, EMBASE et Science Citation Index Expanded (juillet 2013).

Critères de sélection

Les essais contrôlés randomisés (ECR) comparant la chimiothérapie post-opératoire par rapport à la chirurgie seule pour le cancer gastrique résécable.

Recueil et analyse des données

Deux auteurs ont évalué les essais à inclure et extrait les données de manière indépendante. Nous avons analysé les données avec le modèle à effets fixes et le modèle à effets aléatoires en utilisant le logiciel d'analyse RevMan. Nous avons calculé le hazard ratio (HR) avec un intervalle de confiance (IC) à 95% sur la base d'intention de traiter ou d'une analyse des cas disponibles.

Résultats principaux

Les auteurs ont identifié 34 études (7824 patients) rendant compte de la survie globale (SG) et seulement 15 rendant compte aussi de la survie sans maladie (SSM). La chimiothérapie post-opératoire a montré une amélioration de la SG (HR 0,85; IC à 95% 0,80 à 0,90) et une amélioration de la SSM (HR 0,79; IC à 95% 0,72 à 0,87), bien que tous les essais présentaient un risque élevé de biais.

L'analyse prévue de la qualité de vie, le retour au travail, et le nombre d'admissions à l'hôpital a été impossible à mener à bien car les données des résultats nécessaires à l'analyse n'étaient disponibles pour aucun essai.

Conclusions des auteurs

La chimiothérapie post-opératoire devrait être systématiquement utilisée pour le cancer gastrique résécable lorsque c'est possible. D'autres ECR sont nécessaires afin de déterminer sa place à chaque stade de la maladie.

摘要

比較胃切除手術後的化療和單獨胃切除手術的治療方式

背景

對於胃癌,外科手術是主要的治療方式。化療似乎提高了生存率。但化療並非是無併發症的一種治療方式,其在治療中的角色在一些試驗中,有受到質疑。

目的

決定手術後的化學治療是否該成為治療的可切除胃癌的常規療法。

搜尋策略

我們搜尋了 Cochrane Central Register of Controlled Trials (CENTRAL) in The Cochrane Library, MEDLINE, EMBASE, 和Science Citation Index Expanded (2013,6月) 。

選擇標準

採隨機對照試驗(Randomised controlled trials)去比較胃切除手術後的化療和單獨胃切除手術的治療。

資料收集與分析

兩位作者獨立的評估試驗所收案的條件並獨立萃取數據。我們使用RevMan分析軟體去分析彈性因素和隨機因素模型的數據。對於有意願治療或可用的案例進行分析並計算95%信賴區間(confidence interval)的風險比(hazard ratio)。

主要結果

作者確定34個研究(7824位病人)報告總生存率(overall survival),只有15件報告無疾病的生存率(disease free survival)。手術後的化學治療顯示有改善總生存率(風險比0.85;95%信賴區間為0.80〜0.90)和無疾病的生存率(風險比0.79;95%信賴區間為0.72〜0.87),但所有的試驗仍有高偏差的風險。

計畫分析生活品質、返回工作崗位和住院人數得數據分析結果不可能完成的工作,因為沒有任何試驗有分析以上結果的數據。

作者結論

手術後化學治療或許可以常規用於可切除胃癌的治療方式。需要更進一步的隨機對照試驗,以確定在疾病每一個階段中的治療角色。

譯註

翻譯者:臺北醫學大學考科藍臺灣研究中心(Cochrane Taiwan)

本翻譯計畫由臺北醫學大學考科藍臺灣研究中心(Cochrane Taiwan)、台灣實證醫學學會及東亞考科藍聯盟(EACA)統籌執行
聯絡E-mail:cochranetaiwan@tmu.edu.tw

Plain language summary

Chemotherapy after surgery versus surgery alone for stomach cancer

Stomach cancer is the second leading cause of cancer-related death worldwide. Surgery is the only curative treatment offered to patients suffering from this cancer. Survival rates, however, are still poor. Chemotherapy given after surgery has been developed in order to improve these results. We identified 34 trials with 7824 patients which randomised patients to surgery with post-surgical chemotherapy versus surgery alone. The group who received chemotherapy had a survival benefit and improvement in disease free survival (15% and 21%, respectively), although all the trials had a high risk of bias. There was no significant difference due to the stage of disease or the chemotherapy agent used.

Résumé simplifié

La chimiothérapie après la chirurgie par rapport à la chirurgie seule pour un cancer de l'estomac

Le cancer de l'estomac est la deuxième cause de mortalité liée au cancer dans le monde. La chirurgie est le seul traitement curatif proposé aux patients souffrant de ce cancer. Cependant, les taux de survie, sont encore médiocres. La chimiothérapie administrée après chirurgie a été développée afin d'améliorer ces résultats. Nous avons identifié 34 essais totalisant 7824 patients qui randomisaient les patients à la chirurgie avec chimiothérapie post-opératoire par rapport à la chirurgie seule. Le groupe ayant reçu de la chimiothérapie avait un bénéfice en termes de survie et d’amélioration de la survie sans maladie (15% et 21%, respectivement), bien que tous les essais présentaient un risque élevé de biais. Il n'y avait aucune différence significative en raison du stade de la maladie ou de l'agent de chimiothérapie utilisé.

Notes de traduction

Traduit par: French Cochrane Centre 15th December, 2013
Traduction financée par: Financeurs pour le Canada : Instituts de Recherche en Sant� du Canada, Minist�re de la Sant� et des Services Sociaux du Qu�bec, Fonds de recherche du Qu�bec-Sant� et Institut National d'Excellence en Sant� et en Services Sociaux; pour la France : Minist�re en charge de la Sant�

淺顯易懂的口語結論

比較胃切除手術後的化療和單獨胃切除手術的治療方式

胃癌是全世界排名第二名致死的癌症。手術是此疾病的唯一有效治療方式,但對於存活率仍不高。發展手術後的化學治療,以期改善其存活率。我們確定了34個試驗有7824位患者,被隨機分到手術後接受化學或是單獨手術中。接受化療的組別有其生存率上的優勢和無疾病生存率的改善(分別15%和21%),雖然其中的所有試驗都存在高風險的偏差率。由疾病的階段或是使用的化療藥物導致,在此沒有顯著的差異。

譯註

翻譯者:臺北醫學大學考科藍臺灣研究中心(Cochrane Taiwan)

本翻譯計畫由臺北醫學大學考科藍臺灣研究中心(Cochrane Taiwan)、台灣實證醫學學會及東亞考科藍聯盟(EACA)統籌執行
聯絡E-mail:cochranetaiwan@tmu.edu.tw

Summary of findings(Explanation)

Summary of findings for the main comparison. Post-surgical chemotherapy compared to surgery alone for resectable gastric cancer
Post-surgical chemotherapy compared to surgery alone for resectable gastric cancer
Patient or population: patients with resectable gastric cancer
Settings:
Intervention: Post-surgical chemotherapy
Comparison: Surgery alone
OutcomesRelative effect
(95% CI)
No of Participants
(studies)
Quality of the evidence
(GRADE)
Overall Survival (OS) HR 0.85
(0.80 to 0.90)
7523
(34 studies)
High quality
Disease Free Survival HR 0.79
(0.72 to 0.87)
4133
(15 studies)
High quality
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; HR: Hazard ratio;
GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on confidence in the estimate of effect and is likely to change the estimate.
Very low quality: Very uncertain about the estimate.

Summary of findings 2 Subgroup analysis for resectable gastric cancer

Summary of findings 2. Subgroup analysis for resectable gastric cancer
Subgroup analysis for resectable gastric cancer
Patient or population: patients with resectable gastric cancer
Settings:
Intervention: Subgroup analysis
OutcomesRelative effect
(95% CI)
No of Participants
(studies)
Quality of the evidence
(GRADE)
Chemobased OS - 5-FU based chemotherapy OS HR 0.88
(0.83 to 0.94)
5694
(28 studies)
Moderate quality
Chemobased OS - Platinum-based chemotherapy OS HR 0.9
(0.81 to 1)
1504
(9 studies)
Moderate quality
Chemobased DFS - 5-FU based chemotherapy DFS HR 0.86
(0.78 to 0.95)
2944
(13 studies)
Moderate quality
Chemobased DFS - Platinum-based chemotherapy DFS HR 0.89
(0.75 to 1.06)
969
(4 studies)
Moderate quality
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; HR: Hazard ratio;
GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on confidence in the estimate of effect and is likely to change the estimate.
Very low quality: Very uncertain about the estimate.

Summary of findings 3 Subgroup analysis for resectable gastric cancer

Summary of findings 3. Subgroup analysis for resectable gastric cancer
Subgroup analysis for resectable gastric cancer
Patient or population: patients with resectable gastric cancer
Settings:
Intervention: Subgroup analysis
OutcomesRelative effect
(95% CI)
No of Participants
(studies)
Quality of the evidence
(GRADE)
Stage OS analysis - T1/2 OS HR 0.86
(0.71 to 1.05)
1146
(5 studies)
Moderate quality
Stage OS analysis - T3/4 OS HR 0.66
(0.41 to 1.08)
1008
(4 studies)
Moderate quality
Lymph node mets analysis - N- OS HR 0.74
(0.41 to 1.33)
402
(2 studies)
Moderate quality
Lymph node mets analysis - N+ OS HR 0.78
(0.67 to 0.91)
877
(4 studies)
Moderate quality
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; HR: Hazard ratio;
GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on confidence in the estimate of effect and is likely to change the estimate.
Very low quality: Very uncertain about the estimate.

Background

Description of the condition

Gastric cancer is an important disease worldwide and it represents the second leading cause of cancer-related death in the world (CRUK 2012). In 2008, 3052 deaths due to gastric cancer were reported in the United Kingdom and 6450 in the United States of America (GLOBOCAN 2008).

Adenocarcinoma represents 90% of all gastric cancers. It has been classified as well-differentiated or intestinal type, and undifferentiated or diffuse type (Crew 2006). Other histological types like epidermoid tumours or stromal tumours have a significantly lower incidence and a very different clinical course, management and prognosis than adenocarcinoma.

Surgical resection is, to date, the main management and the only curative treatment that can be offered to patients with gastric cancer (Alberts 2003). Partial gastric resection or total gastrectomy are the surgical options commonly performed and both try to maximize elimination of tumour, with no significant difference in terms of survival when they are compared (Bozzetti 1999). Extended lymphadenectomy procedures aim to eradicate any residual disease or tumour cell progression through lymph nodes but the role of these procedures is controversial and there is an important disparity of results in the demonstration of real improvement in survival and the morbidity of this additional therapy (McCulloch 2003; Memon 2011).

The role of chemotherapy remains unclear and the presence in the literature of at least seven meta-analyses regarding this topic support this notion (Earle 1999; Hermans 1993; Hu 2002; Jeung 2009; Mari 2000; Oba 2006; Paoletti 2010). Chemotherapy and surgery can be combined in different regimes as chemotherapy can be administrated previous to or after surgery. Commonly, it is accepted as 'adjuvant chemotherapy' if it is administrated after surgery, 'neoadjuvant chemotherapy' when it takes place previous to surgery, and 'perioperative chemotherapy' when both neoadjuvant and adjuvant chemotherapy are combined. However, sometimes this is unclear in some trials, hence we refer to our intervention as post-surgical chemotherapy in order to avoid confusion.

Surgery alone is described as adequate treatment for early stage tumours (Kim 1992b), and it is commonly accepted that chemotherapy should be given to those patients with an advanced stage tumour (Wagner 2005). This is supported by recent publications (Sun 2009) but other well-defined trials and older meta-analyses do not share these results (Di Costanzo 2008; Hermans 1993). Secondly, early recurrence and disease free survival period are not described in some of the current reviews. Finally, morbidity and mortality related to chemotherapy have not been properly reported and they both may modify the recommendation for chemotherapy.

Analysis of old and new trials comparing chemotherapy versus surgery alone, once curative resection has been performed, may clarify the role of adjuvant chemotherapy in overall survival and disease free survival in patients with gastric cancer.

Description of the intervention

After potentially curative surgical resection for gastric cancer, chemotherapy using a single drug or a combination of drugs may be given (post-surgical chemotherapy).

How the intervention might work

Post-surgical chemotherapy may remove residual cancer cells or cancer cells in the systemic circulation that surgery cannot remove and which could be the reason for recurrence and death (Eguchi 2008).

Why it is important to do this review

The treatment options for resectable gastric cancer include resection alone, resection with pre-surgical and post-surgical chemotherapy (perioperative chemotherapy), and post-surgical chemotherapy. The role of gastric resection with pre-surgical and post-surgical chemotherapy is being assessed in another Cochrane Review (Slanger 2009). The role of post-surgical chemotherapy for gastric cancer in improving patient survival remains unclear. In addition, the use of chemotherapeutic agents is not risk free and they should be used only if justified. Post-surgical chemotherapy has been found to be effective in the treatment of other cancers such as breast cancer (EBCTCG 2008; Ferguson 2007; Mauri 2008). However, while some trials seem to support the use of post-surgical chemotherapy for resectable gastric cancers (Nakajima 2007; Paoletti 2010) others do not (Di Costanzo 2008b; Hu 2002). The aim of this review was to determine whether post-surgical chemotherapy improves overall survival and disease free survival in patients with resectable gastric cancer.

Objectives

To determine whether post-surgical chemotherapy should be used routinely in patients with resectable gastric adenocarcinoma.

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled trials (RCT). We excluded quasi-randomised studies for survival outcomes but included them for evaluation of the adverse effects of chemotherapy, such as anaemia and gastrointestinal toxicity.

Types of participants

Patients undergoing partial or total gastrectomy for gastric cancer, irrespective of the location of the cancer (distal, proximal or oesophagogastric junction).

Types of interventions

Post-surgical chemotherapy (irrespective of whether a single drug or a combination of drugs were used and the duration of the treatment) versus surgery alone without post-surgical chemotherapy in patients undergoing partial or total gastrectomy, irrespective of nodal status, whether pancreatosplenectomy was performed, or the presence of microscopic involvement of the resection margins.

We excluded trials assessing the following interventions.

  1. Combination of pre-surgical and post-surgical chemotherapy in patients undergoing resection for gastric cancer.

  2. Combination of post-surgical chemotherapy and radiotherapy in patients undergoing resection for gastric cancer.

  3. Combination of post-surgical chemotherapy and immunotherapy using drugs which have generic immunomodulation potential in patients undergoing resection for gastric cancer. However, we included any trials which used monoclonal antibodies that were targeted against specific proteins in combination with post-surgical chemotherapy. We only excluded non-specific immunotherapy, such as interferon or bacillus Calmette-Guerin (BCG) as it would be difficult to assess the effectiveness of the chemotherapy.

Types of outcome measures

Primary outcomes
  1. Overall survival:

    1. hazard ratio.

  2. Disease free survival:

    1. hazard ratio.

Secondary outcomes
  1. Quality of life (however defined by authors).

  2. Adverse effects attributed to chemotherapy, such as bone marrow suppression (anaemia, neutropenia, thrombocytopenia grades three and four of the World Health Organization (WHO) classification); gastrointestinal side effects (nausea, vomiting grades three and four of the WHO classification); cardiac toxicity (heart failure); and death related to post-surgical chemotherapy (WHO 1979).

  3. Number of hospital admissions after discharge.

  4. Return to work.

Search methods for identification of studies

We searched the following resources.

Electronic searches

Searching other resources

Reference lists from trials selected by electronic searching were handsearched to identify further relevant trials. Published abstracts from conference proceedings from the United European Gastroenterology Week (published in Gut), Digestive Disease Week (published in Gastroenterology) and ENT news also were handsearched.

In addition, members of the Cochrane Upper Gastrointestinal and Pancreatic Diseases (UGPD) Group and experts in the field were contacted and asked to provide details of outstanding clinical trials and any relevant unpublished materials. We also contacted the Cochrane Ear, Nose and Throat Diseases Group to confirm that our search was comprehensive.

The review authors carried out the search independently from each other. Reference lists of identified publications were scanned for additional trials and authors contacted if necessary. In addition, the reference lists of any previous reviews on the subject and the review authors' own files were scanned for relevant studies. No language restrictions were applied. The full text articles of the retrieved trials were reviewed by at least two review authors and the inclusion criteria applied independently. Any differences in opinion about which studies to include in the review were resolved by discussion between the review authors. Possible additional search terms were discussed at one of the conferences where the subject was presented.

Data collection and analysis

Selection of studies

Two authors (RDN and ROR) identified the trials for inclusion, independent of each other. We also listed the excluded trials with the reasons for exclusion in the 'Characteristics of excluded studies' table.

Data extraction and management

RDN and ROR extracted the data for the review independently. In addition, we extracted the population characteristics (such as sex, age, type of gastrectomies) and the interventions used in each trial. We clarified any unclear or missing information by contacting the authors of the individual trials. If there was any doubt as to whether the trials shared the same patients, completely or partially (by identifying common authors and centres), we contacted the authors of the trials to establish whether the trial report had been duplicated. We resolved any differences in opinion through discussion.

For overall survival and disease free survival, we extracted the logarithm of hazard ratios [ln(HR)] and the standard error (SE) of ln(HR) using the methods described by Parmar et al (Parmar 1998) and the Excel sheet provided by Tierney et al (Tierney 2007).

We followed the instructions given in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2008) and the Cochrane Upper GastrointestinaI and Pancreatic Diseases Group Module (Forman 2009). We imputed the standard deviation from P values according to the Cochrane Handbook for Systematic Reviews of Interventions and used the median for the meta-analysis when the mean was not available. If it was not possible to calculate the standard deviation from P values or confidence intervals, we imputed the standard deviation as the highest standard deviation noted for that group under that outcome.

Assessment of risk of bias in included studies

According to empirical evidence (Kjaergard 2001; Moher 1998; Schulz 1995; Wood 2008), we assessed the risk of bias in the trials based on sequence generation, allocation concealment, blinding (of participants, personnel and outcome assessors), incomplete outcome data, selective outcome reporting and other sources of bias. We assessed the risk of bias in the trials independently, without masking of the trial names. Quality components were based on the Cochrane Handbook for Systematic Reviews of Interventions (Gurusamy 2009; Higgins 2008).

Sequence generation
  • Low risk of bias (the method used is either adequate (e.g. computer-generated random numbers, table of random numbers) or unlikely to introduce confounding).

  • Uncertain risk of bias (there is insufficient information to assess whether the method used is likely to introduce confounding).

  • High risk of bias (the method used (e.g. quasi-randomised trials) is improper and likely to introduce confounding).

Allocation concealment
  • Low risk of bias (the method used (e.g. central allocation) is unlikely to induce bias in the final observed effect).

  • Uncertain risk of bias (there is insufficient information to assess whether the method used is likely to induce bias in the estimate of effect).

  • High risk of bias (the method used (e.g. open random allocation schedule) is likely to induce bias in the final observed effect).

Blinding of participants, personnel and outcome assessors
  • Low risk of bias (blinding was performed adequately, or the outcome measurement is not likely to be influenced by lack of blinding).

  • Uncertain risk of bias (there is insufficient information to assess whether the type of blinding used is likely to induce bias in the estimate of effect).

  • High risk of bias (no blinding or incomplete blinding, and the outcome or the outcome measurement is likely to be influenced by lack of blinding).

Incomplete outcome data
  • Low risk of bias (the underlying reasons for missing data are unlikely to make treatment effects depart from plausible values, or proper methods have been employed to handle missing data).

  • Uncertain risk of bias (there is insufficient information to assess whether the missing data mechanism in combination with the method used to handle missing data is likely to induce bias in the estimate of effect).

  • High risk of bias (the crude estimate of effects (e.g. complete case estimate) will clearly be biased due to the underlying reasons for missing data, and the methods used to handle missing data are unsatisfactory).

Selective outcome reporting
  • Low risk of bias (the trial protocol is available and all of the trial's prespecified outcomes that are of interest in the review have been reported, or similar).

  • Uncertain risk of bias (there is insufficient information to assess whether the magnitude and direction of the observed effect is related to selective outcome reporting).

  • High risk of bias (not all of the trial's prespecified primary outcomes have been reported, or similar).

Other bias
Baseline imbalance
  • Low risk of bias (there was no baseline imbalance in important characteristics).

  • Uncertain risk of bias (the baseline characteristics were not reported).

  • High risk of bias (there was a baseline imbalance due to chance or due to imbalanced exclusion after randomisation).

Early stopping
  • Low risk of bias (sample size calculation was reported and the trial was not stopped or the trial was stopped early by a formal stopping rule at a point where the likelihood of observing an extreme intervention effect due to chance was low).

  • Uncertain risk of bias (sample size calculations were not reported and it is not clear whether the trial was stopped early or not).

  • High risk of bias (the trial was stopped early due to an informal stopping rule or the trial was stopped early by a formal stopping rule at a point where the likelihood of observing an extreme intervention effect due to chance was high).

Source of funding bias
  • Low risk of bias (the trial's source(s) of funding was from no parties that might have conflicting interest (e.g. pharmaceutical company)).

  • Uncertain risk of bias (the source of funding was not clear).

  • High risk of bias (the trial was funded by a pharmaceutical company).

We considered trials which were classified as low risk of bias in sequence generation, allocation concealment, blinding, incomplete data and selective outcome reporting as low bias-risk trials.

Measures of treatment effect

The hazard ratio (HR) was measured with 95% confidence interval (CI) for overall survival and disease free survival. For dichotomous outcomes, the risk ratio (RR) was calculated with 95% CI. For continuous outcomes, we calculated the mean difference (MD) or standardised mean difference (SMD) with 95% CI. The risk difference (RD) was also estimated with 95% CI for dichotomous outcomes and results were only reported if they were different from the RR.

Unit of analysis issues

The unit of analysis was each patient recruited into the trials.

Dealing with missing data

Analysis was performed on an 'intention-to-treat' basis (Newell 1992) whenever possible. Otherwise, the 'available case' analysis was adopted.

Assessment of heterogeneity

Heterogeneity was explored by the Chi² test with significance set at a P value of 0.10, and we measured the quantity of heterogeneity with the I² statistic (Higgins 2002). An I² statistic of > 50% was considered statistically significant heterogeneity. We calculated the 95% CI for I² using StatsDirect statistical software (StatsDirect 2008). Meta-regression would have been performed in the presence of at least 10 trials.

Assessment of reporting biases

Asymmetry in a funnel plot of trial size against treatment effect was used to assess bias (Egger 1997; Macaskill 2001). The linear regression approach described by Egger et al was performed to determine the funnel plot asymmetry (Egger 1997).

Data synthesis

The meta-analyses was performed according to the recommendations of The Cochrane Collaboration (Higgins 2008). The software package RevMan 5 provided by the Cochrane Collaboration was used (RevMan 2008). A random-effects model (DerSimonian 1986) and a fixed-effect model were used (DeMets 1987). In the case of discrepancy between the two models, both results were reported; otherwise only the results from the fixed-effect model were mentioned. HRs of overall survival and disease free survival obtained from the different trials were combined using the generic inverse variance method.

Subgroup analysis and investigation of heterogeneity

The following subgroup analyses were planned.

  • Trials with low risk of bias versus those with high risk of bias.

  • Different types of chemotherapy as well as number of regimens and rate of cycles completed.

  • Different surgical procedures (partial versus total gastrectomy).

  • Different lymphadenectomy procedures (D1 versus D2 resection).

  • Different approaches (open versus laparoscopic approach).

  • Microscopically clear resection margin versus microscopically involved tumour margin.

  • Different locations of tumour.

  • Different histological types of gastric cancer.

  • Different stages of gastric cancer (T1, T2 versus T3, T4).

The Chi² test for subgroup differences was set at a P value of 0.05.

Sensitivity analysis

Sensitivity analysis was performed by excluding the trials which included gastric cancers at the gastro-oesophageal junction.

Results

Description of studies

Results of the search

A total of 3391 references were identified through the electronic searches of the Cochrane Central Register of Controlled Trials (CENTRAL) in The Cochrane Library, MEDLINE, EMBASE, and Science Citation Index Expanded. The flow of references are shown in Figure 1. A total of 467 duplicates and 2879 clearly irrelevant references were excluded through reading abstracts. Forty-five references were retrieved for further assessment. Of the 45 references, 11 references were excluded because of the reasons listed in the table 'Characteristics of excluded studies'. Three references corresponded to the same study (Alcobendas 1983; Estape 1991; Grau 1993) and hence the inclusion of two studies under the excluded studies. Of the remaining references, 34 randomised trials fulfilled the inclusion criteria. Once the full analysis of data was achieved the search strategy was rerun up to July 2013, in order to assure the presence of the most recent trials. This new search resulted in 189 new citations (MEDLINE= 129, EMBASE= 27, CENTRAL= 21, Science Citation Index= 68) of which only three were considered relevant. One of these remaining was excluded and listed in the table 'Characteristics of excluded studies' (Chen 2011) and the other two trials were added to the section “studies awaiting classification” (Bang 2012; Sasako 2011).

Figure 1.

Study flow diagram.

Included studies

All 34 trials were completed trials and could provide data for the analyses. Details of the trials are shown in the table 'Characteristics of included studies'. Overall the 34 trials included 7824 patients. The mean age of the individuals in the trials varied between 54 (Kim 1992) and 65 years (Tentes 2006). The mean proportion of females varied between 19% (Popiela 2004) to 47% (Hallissey 1994). There was no difference in the characteristics of patients in the intervention group and control group in any of the trials that reported these baseline characteristics. However, a wide variation was found between trials as the interventions were not exactly the same in all the studies. Differences were in terms of the drugs, doses and duration of the treatments as well as the design of the studies; but almost every trial based their intervention on 5-fluorouracil (5-FU) or platinum-based chemotherapy (Platino), considered for separate analysis by subgroup.

Excluded studies

The reasons for exclusion of the studies are listed in the table 'Characteristics of excluded studies'.

Risk of bias in included studies

The risk of bias is summarised in the risk of bias graph (Figure 2) and risk of bias summary (Figure 3).

Figure 2.

Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies.

Figure 3.

Methodological quality summary: review authors' judgements about each methodological quality item for each included study.

Allocation

The generation of allocation sequence was adequate in six trials (Bajetta 2002; Chipponi 2004Nakajima 1999; Nakajima 2007; Nitti 2006 EORTC; Sakuramoto 2007) and allocation concealment was adequate in 11 trials (Bonfanti 1988; Cirera 1999; Coombes 1990Hallissey 1994; Higgins 1983; Huguier 1980; Nakajima 2007; Nashimoto 2003; Ochiai 1983; Popiela 2004; Sakuramoto 2007).

Blinding

Blinding was adequate in one trial (Allum 1989) where the authors reported the use of placebo for the surgery alone arm in the same regimen as the chemotherapy was administrated to the intervention arm.

Incomplete outcome data

Five trials were free from bias due to incomplete outcome data (Allum 1989; Bonfanti 1988; Douglas 1982; Engstrom 1985; Higgins 1983). In 28 trials, there were post-randomisation dropouts resulting in high risk of bias from incomplete outcome data.

Selective reporting

Twenty-eight trials were free from bias due to selective reporting (Allum 1989; Bajetta 2002; Bouché 2005; Chipponi 2004; Chou 1994; Cirera 1999; Coombes 1990De Vitta 2007; Di Costanzo 2008; Douglas 1982; Engstrom 1985; Fielding 1983; Fujimoto 1977; Hallissey 1994; Higgins 1983; Huguier 1980Krook 1991; Lise 1995; Macdonald 1995; Nakajima 1999; Nakajima 2007; Neri 2001; Nitti 2006 EORTC; Nitti 2006 ICCG; Ochiai 1983; Popiela 1982; Popiela 2004; Sakuramoto 2007).

Other potential sources of bias

Fifteen trials were free from other potential sources of bias (Allum 1989; Bajetta 2002; Bonfanti 1988; Bouché 2005; Cirera 1999; Coombes 1990; Di Costanzo 2008; Douglas 1982; Engstrom 1985; Fielding 1983; Fujimoto 1977; Hallissey 1994; Huguier 1980; Macdonald 1995; Nakajima 1999).

Effects of interventions

See: Summary of findings for the main comparison Post-surgical chemotherapy compared to surgery alone for resectable gastric cancer; Summary of findings 2 Subgroup analysis for resectable gastric cancer; Summary of findings 3 Subgroup analysis for resectable gastric cancer

Primary outcomes

Overall survival

A total of 34 trials (7523 participants) provided data on overall survival with homogeneous results (I² = 31%). Post-surgical chemotherapy showed a significant beneficial effect on overall survival (HR 0.85; 95% CI 0.80 to 0.90) (Figure 4; Figure 5).

Figure 4.

Forest plot of comparison: 1 Post-surgical chemotherapy versus surgery alone, outcome: 1.1 Overal survival.

Figure 5.

Funnel plot of comparison: 1 Post-surgical chemotherapy versus surgery alone, outcome: 1.1 Overal survival.

Disease free survival

Of the 34 trials, 15 trials (4133 participants) also reported results regarding disease free survival, where post-surgical chemotherapy showed a beneficial effect on disease free survival (HR 0.79; CI 0.72 to 0.87) (Figure 6).

Figure 6.

Forest plot of comparison: 1 Post-surgical chemotherapy versus surgery alone, outcome: 1.2 Disesase free survival.

Secondary outcomes

The planned analysis of quality of life, return to work, and number of hospital admissions was impossible to complete as the outcome data for the analysis were not available from any trials.

Subgroup analysis

Planned subgroup analysis of different gastrectomy procedures, surgical approach, location of the tumour, microscopically affected margins, tumour location, and histological type were not performed as no outcome data for the different subgroups were available from the trials.

Subgroup analysis based on the risk of bias in the trials was not performed as no trials were at low risk of bias, accepting as low risk of bias those where blinding, sequence generation and allocation concealment were properly applied and clearly reported. Despite the impossibility of achieving this analysis, allocation concealment was considered an important element and a subgroup analysis was performed according to this bias. Eleven trials with adequate allocation concealment were compared with 23 trials where allocation concealment was unclear or inadequate. The first subgroup showed an HR of 0.82 (95% CI 0.74 to 0.91) and the second subgroup showed an HR of 0.86 (95% CI 0.79 to 0.93) with no statistical difference when compared (P > 0.05).

For subgroup analysis by stage, only five trials provided data related to the tumour size (T1-2 versus T3-4). Analysis of the patients with T1-2 tumour size revealed an HR of 0.86 (95% CI 0.71 to 1.05) while those with T3-4 tumours showed an HR of 0.66 (95% CI 0.41 to 1.08). However, no significant difference was shown when compared (Chi² = 0.93, df = 1, P = 0.34; I² = 0%).

For subgroup analysis by lymph node involvement, four included studies described independent analyses related to lymph node status (N). For N - (absence of lymph node metastasis) the HR was 0.74 (95% CI 0.41 to 1.33) and for N + (lymph node metastasis) the HR was 0.78 (95% CI 0.67 to 0.91). Again, no differences were shown between groups (Chi² = 0.03, df = 1, P = 0.86; I² = 0%).

In the subgroup analysis by chemotherapy agent, from all studies 29 trials based the post-surgical chemotherapy on the use of 5-FU, alone or combined with other drugs. The analysis revealed an overall survival benefit with an HR of 0.88 (95% CI 0.83 to 0.94); while those trials where platinum-based chemotherapy was used (eight included trials) the HR for overall survival was 0.90 (95% CI 0.81 to 1.00). The analysis of disease free survival for both groups showed similar results. Fourteen 5-FU based trials reported data on disease free survival and showed an HR of 0.86 (95% CI 0.78 to 0.95). Only four platinum-based studies reported data on disease free survival where the HR was 0.89 (95% CI 0.75 to 1.06). Nevertheless, overall survival and disease free survival did not show statistically significant differences between groups when compared. Test for subgroup differences for overall survival: Chi² = 0.16, df = 1, P = 0.69; I² = 0%; and for disease free survival: Chi² = 0.15, df = 1, P = 0.70; I² = 0%.

Discussion

Gastric cancer is one of the most important neoplasms of the digestive tract due to its high incidence and mortality. Over 21,000 new cases were expected in USA in 2010 as well as more than 10,000 expected deaths for both sexes (Jemal 2010). Such a rate is caused by the fact that despite the advances achieved in the management of this pathology, overall survival rates including all stages do not exceed 20% (Crew 2006). A particular consideration of gastric cancer is the historical difference between oriental and occidental series. Countries like Japan and China have described a higher incidence of this pathology and their screening programs have reached historically higher early stage diagnosis rates. Also, the reported overall survival and disease free survival rates were commonly better than those reported in occidental trials.

Surgery

Irrespective of the series origin, treatment of this malignancy is based on surgical resection, which represents the only curative option today. Complete tumour resection is currently the target of treatment, and provides the basis on which the therapeutic strategy is supported.

The described interventions are partial or total gastrectomy and, in relation to the technique, available data do not allow us to assess differences between the procedures regarding survival. Main series of patients where total gastrectomy is compared with subtotal gastrectomy do not indicate any improvement in terms of survival (Shi 2010). Factors determining survival are margin invasion and lymph node status (Gouzi 1989; Shi 2010). In addition, for total gastrectomy the postoperative quality of life is significantly worse when compared with quality of life after subtotal gastrectomy (Davies 1998). In the present review, the surgical procedure was only mentioned as part of the initial description of baseline characteristics of patients in the trials and no further comparison was reported related to it.

Lymphadenectomy

Every surgical resection commonly includes some degree of lymphadenectomy. Gastric cancer may be one of the neoplasms, similar to breast cancer, where lymphadenectomy procedures are very well described and also have standardised grades. Worldwide, four grades of lymphadenectomy are accepted for gastric cancer and they are meticulously described, in ascending order of extent, as D1, D2, D3 and D4.

Differences between oriental and occidental series arise again in relation to this topic. D2 was initially described as a safe procedure that produced incremental overall survival and disease free survival rates in oriental series (Nakajima 2002); while prospective European trials like the Dutch Gastric Cancer Trial (DGCT) and UK Medical Research Council (MRC) did not show any survival benefit for D2 lymphadenectomy, and described a significantly higher morbidity rate associated with this procedure (Cuschieri 1999; Hartgrink 2004). Recently, similar studies have reported that pancreas resection seems to be the main cause of morbidity related to lymphadenectomy, hence pancreas preserving D2 lymphadenectomy may have a survival benefit without an incremental increase in morbidity (Degiuli 2004; Wu 2006). The standardisation and implementation of this type of surgery appears to be, at present, the main recommendation (Songun 2010).

Chemotherapy

Once an adequate surgical procedure and D2 lymphadenectomy have been performed, post-surgical chemotherapy would be the next step in the management of gastric cancer. Results reported in this meta-analysis conclude that chemotherapy should be the general recommendation, with an overall survival benefit of 15% (HR 0.85) and disease free survival benefit of 21% (HR 0.79), however no standard regimen is available due to the contradictory results. Platinum-based regimens of chemotherapy have clearly failed to show survival benefit in recent trials (Bajetta 2002; Kulig 2010) and our subgroup analysis also support these results. Considering the estimated HRs for overall survival (OS) and disease free survival (DFS) in the platinum-based chemotherapy group (HR 0.90; 95% CI 0.81 to 1.00 and HR 0.89; 95% CI 0.75 to 1.06, respectively) no survival benefit can be assumed with the use of these regimens. The estimated HR for the 5-FU based chemotherapy group does reveal a survival benefit (HR 0.88 for OS and HR 0.86 for DFS; P < 0.05 in both cases); despite no statistically significant difference when both groups were compared (P > 0.05). These findings are consistent with regard to recent recommendations and current guidelines (Okines 2010).

As a result, new therapeutical regimens are being developed and the combination of perioperative chemotherapy (pre and post-surgical chemotherapy) with radiotherapy is accepted in current treatment protocols based on two studies that reported survival benefit (Cunninghan 2006; Macdonald 2001), both based on a fluouracil-type agent regime.

Toxicity

Chemotherapy administration has important related morbidity. The US Food and Drug Administration (FDA) reports an adverse effect rate close to 100% for the use of 5-FU (FDA), including grades 1 to 4 of the WHO adverse effects classification, and this is confirmed in some trials that report a toxicity rate of 94% when considering grades 1 and 2 (Macdonald 1995). Commonly these initial toxicity grades are not included in studies and this rate varies from 15% to 55% (Bouché 2005; Di Costanzo 2008; Kulig 2010; Tentes 2006). Special mention is required concerning the current use of S-1 (contains tegafur, a substance that converts to 5-FU in the body) therapy, with reports of an adverse effect rate of 6% (Sakuramoto 2007).

Adverse effects related to chemotherapy do not represent an outcome that can be used to compare groups as they are only relevant to one arm of the trials, however they have a huge impact on patients receiving these kind of treatments. Our data reveal that only 21 of the included studies reported data related to adverse effects (Table 1). Nausea and vomiting is the most common toxicity related to chemotherapy, considering our data, with 565 events in 2345 patients (24% incidence) followed by leukopenia with 500 events in 2506 patients (20% incidence).

Table 1. Adverse effects
  1. Interv.: Intervention group (postsurgical chemotherapy).

    Cont.: Control group (surgery alone).

    NR: not reported.

Study ID Participants Anaemia events Leukopenia events Thrombopenia events Nausea/Vomiting
Bajetta 2002

Interv.: 135

Cont.: 136

NR

NR

28

NR

2

NR

18

NR

Bouché 2005

Interv.: 127

Cont.: 133

NR

NR

NR

NR

NR

NR

57

NR

Chipponi 2004

Interv.: 93

Cont.: 103

10

NR

24

NR

13

NR

29

NR

Coombes 1990

Interv.: 133

Cont.: 148

53

21

70

11

28

2

71

10

De Vitta 2007

Interv.: 112

Cont.: 113

11

NR

30

NR

14

NR

20

NR

Di Costanzo 2008

Interv.: 130

Cont.: 128

4

NR

24

NR

5

NR

25

NR

Fielding 1983

Interv.: 141

Cont.: 130

113

64

47

6

30

6

93

73

Fujimoto 1977

Interv.: 97

Cont.: 120

NR

NR

17

NR

NR

NR

12

NR

Grau 1993

Interv.: 68

Cont.: 66

NR

NR

15

NR

17

NR

12

NR

Higgins 1983

Interv.: 66

Cont.: 68

37

NR

?

NR

36

NR

?

NR

Krook 1991

Interv.: 61

Cont.: 64

NR

NR

50

NR

1

NR

28

NR

Lise 1995

Interv.: 155

Cont.: 159

65

NR

94

NR

44

NR

107

NR

Macdonald 1995

Interv.: 93

Cont.: 100

3

NR

1

NR

20

NR

13

NR

Nakajima 1999

Interv.: 288

Cont.: 285

NR

NR

27

1

2

1

NR

NR

Nakajima 2007

Interv.: 93

Cont.: 95

1

0

11

0

NR

NR

1

0

Nashimoto 2003

Interv.: 127

Cont.: 123

2

3

2

0

0

1

0

0

Neri 2001

Interv.: 69

Cont.: 68

3

NR

6

NR

2

NR

44

NR

Nitti 2006 EORTC

Interv.: 103

Cont.: 103

NR

NR

28

NR

4

NR

11

NR

Nitti 2006 ICCG

Interv.: 91

Cont.: 100

NR

NR

20

NR

1

NR

11

NR

Popiela 2004

Interv.: 53

Cont.: 52

NR

NR

?

NR

4

NR

NR

NR

Sakuramoto 2007

Interv.: 517

Cont.: 526

6

4

6

2

1

2

25

16

Staging

Stage-related treatment is the most controversial topic. Results of our review show general OS and DFS benefit including all stages, nevertheless the prognosis is different in every stage and the beneficial effect may be related to the more important benefit on higher grading. Subgroup analysis was completed by tumour size but few articles reported their data and no differences between groups were found (P > 0.05). Despite no statistical significance a greater beneficial effect was shown in the T3-4 group (HR 0.66 versus HR 0.86 for T1-2 group). Against this, Nakajima 2007 described survival benefit with the use of post-surgical chemotherapy for patients with serosa-negative gastric cancer, but all included patients also had lymph node metastasis.

Regarding post-surgical chemotherapy administration and N status, some studies affirm that lymph node metastases (N +) represents an independent poor prognosis factor (Bouché 2005; Popiela 2004), however few studies reported separate analyses. Subgroup analysis again shows a more important survival benefit for the N + group (HR 0.78; 95% CI 0.67 to 0.91 versus HR 0.74; 95% CI 0.41 to 1.33 for N -) despite the fact that there was no statistical significance when both groups were compared (P > 0.05). It seems easy to assume that post-surgical chemotherapy should be considered as part of the treatment of gastric cancer with lymph node involvement in relation to trials that failed to demonstrate general survival benefit including all the stages but that report survival benefit for the N + group (De Vitta 2007). However, doubts arise in the situation where D2 lymphadenectomy has been performed correctly and there is no evidence of lymph node metastasis. We understand that potential survival improvement with chemotherapy remains unclear in these stages. In summary, from the current analysis we agree that patients without nodal or serosa involvement are unlike to benefit from post-surgical chemotherapy (Nashimoto 2003).

Quality of life

Quality of life is not described in any of the included studies, but it represents a fundamental fact in oncological patients. The closest related reference is the study of Al-Batran (Al-Batran 2010) where they analysed the quality of life related to chemotherapy of patients with metastatic oesophagogastric cancer. Generally, studies that are available do not describe any improvements in the quality of life due to the administration of chemotherapy. Only fluorouracil-based and platinum-based regimens seem to improve, in some measure, the quality of life or at least they preserve it. Direct effects of adjuvant treatment on quality of life in those patients operated on with curative intention should be evaluated in the future and properly designed trials with low risk of bias should include quality of life as a main outcome.

Cost-effectiveness

None of the included or excluded studies mentioned the cost-effectiveness of implementation of post-surgical chemotherapy. Recent trials that evaluate the cost-effectiveness of new drugs in advanced gastric cancer have concluded that these therapies might not be cost-effective (Wu 2012). In this sense, well-developed trials should investigate the cost-effectiveness of post-surgical chemotherapy.

Summary of main results

Post-surgical chemotherapy

Post-surgical chemotherapy shows a survival benefit for patients with gastric cancer, for OS (HR 0.85; 95% CI 0.80 to 0.90) and for DFS (HR 0.79; 95% CI 0.72 to 0.87).

Subgroup analysis

Post-surgical chemotherapy shows overall survival benefit for patients with gastric cancer and N + status (HR 0.78; 95% CI 0.67 to 0.91) despite the fact no significant differences were found when compared with the N - group (P > 0.05).

Overall completeness and applicability of evidence

This review included patients undergoing gastric surgery for resectable gastric adenocarcinoma, including total or partial gastrectomies, irrespective of the lymphadenectomy procedure. Hence this review is applicable to most patients with resectable gastric cancer.

Quality of the evidence

Most of the trials were at high risk of bias. Allocation was unclear in most of the trials. Blinding was not performed in most of the trials. Some of the trials did not report all the prespecified outcomes.

Potential biases in the review process

The Cochrane Handbook for Systematic Reviews of Interventions was followed for this review. There were no language, publication status, or sample size restrictions. Thus, bias due to selection of trials was minimized. There was no reporting bias as suggested by the funnel plot and Egger's linear regression approach (Egger 1997).

Per protocol analysis was used when intention-to-treat analysis or available case analysis was not possible. This could lead to bias favouring the intervention.

Agreements and disagreements with other studies or reviews

Old reviews regarding the use of post-surgical chemotherapy failed to demonstrate any survival benefit (Hermans 1993). More recent reviews show higher rates of overall survival with statistical significance in patients treated with chemotherapy after surgical resection (Paoletti 2010; Sun 2009). Results of this review are similar to those more recent publications and they support that there is not a standard regimen of chemotherapy accepted.

Authors' conclusions

Implications for practice

Post-surgical chemotherapy should be use for patients with resectable gastric cancer following the results of this review, however it is important to note that some of the included trials present limitations and definitive assessments on this topic may be delayed until future trials are properly developed.

Implications for research

Further low bias-risk trials are necessary to determine wether post-surgical chemotherapy should be used routinely in every stage of gastric cancer. The chemotherapy agent and the standard regimen and duration are still topics that require clarification. Quality of life and cost-effectiveness represent two important topics that should be considered in future studies.

Acknowledgements

We thank the Cochrane Upper Gastrointestinal and Pancreatic Diseases Group with special mention of Ms Karin Dearnes for all her help and professionalism.

We thank Mr Kurinchi Gurusamy, Department of Surgery, UCL Medical School, London for his guidance. We also thank Mr Massimo Varcada for his work in the elaboration of the protocol.

We thank Ms Marta Roqué, Iberoamerican Cochrane Group, for her support.

Data and analyses

Download statistical data

Comparison 1. Post-surgical chemotherapy versus surgery alone
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Overal Survival347523Hazard Ratio (Fixed, 95% CI)0.85 [0.80, 0.90]
2 Disesase free survival154133Hazard Ratio (Random, 95% CI)0.79 [0.72, 0.87]
Analysis 1.1.

Comparison 1 Post-surgical chemotherapy versus surgery alone, Outcome 1 Overal Survival.

Analysis 1.2.

Comparison 1 Post-surgical chemotherapy versus surgery alone, Outcome 2 Disesase free survival.

Comparison 2. Subgroup analysis
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Chemobased OS297198Hazard Ratio (Random, 95% CI)0.89 [0.84, 0.94]
1.1 5-FU based chemotherapy OS285694Hazard Ratio (Random, 95% CI)0.88 [0.83, 0.94]
1.2 Platinum-based chemotherapy OS91504Hazard Ratio (Random, 95% CI)0.90 [0.81, 1.00]
2 Stage OS analysis5 Hazard Ratio (Random, 95% CI)0.80 [0.66, 0.98]
2.1 T1/2 OS5 Hazard Ratio (Random, 95% CI)0.86 [0.71, 1.05]
2.2 T3/4 OS4 Hazard Ratio (Random, 95% CI)0.66 [0.41, 1.08]
3 Lymph node mets analysis4 Hazard Ratio (Random, 95% CI)0.78 [0.68, 0.90]
3.1 N- OS2 Hazard Ratio (Random, 95% CI)0.74 [0.41, 1.33]
3.2 N+ OS4 Hazard Ratio (Random, 95% CI)0.78 [0.67, 0.91]
4 Chemobased DFS133913Hazard Ratio (Random, 95% CI)0.86 [0.79, 0.94]
4.1 5-FU based chemotherapy DFS132944Hazard Ratio (Random, 95% CI)0.86 [0.78, 0.95]
4.2 Platinum-based chemotherapy DFS4969Hazard Ratio (Random, 95% CI)0.89 [0.75, 1.06]
5 Low Risk of Bias347523Hazard Ratio (Random, 95% CI)0.84 [0.79, 0.90]
5.1 Allocation concealment112835Hazard Ratio (Random, 95% CI)0.82 [0.74, 0.91]
5.2 No allocation concealment234688Hazard Ratio (Random, 95% CI)0.86 [0.79, 0.93]
Analysis 2.1.

Comparison 2 Subgroup analysis, Outcome 1 Chemobased OS.

Analysis 2.2.

Comparison 2 Subgroup analysis, Outcome 2 Stage OS analysis.

Analysis 2.3.

Comparison 2 Subgroup analysis, Outcome 3 Lymph node mets analysis.

Analysis 2.4.

Comparison 2 Subgroup analysis, Outcome 4 Chemobased DFS.

Analysis 2.5.

Comparison 2 Subgroup analysis, Outcome 5 Low Risk of Bias.

Appendices

Appendix 1. CENTRAL search strategy

#1 Digest* or Gastr* or gut or epigastr* or stomach*
#2 oesophagogastric junction or esophagogastric junction
#3 Lower Esophageal Sphincter
#4 carcin* or cancer* or neoplas* or tumour* or tumor* or cyst* or adenocarcin* or malig*
#5 (#1 OR #2 OR #3)
#6 (#4 AND #5)
#7 gastrectomy
#8 Postoperative Care
#9 chemotherapy
#10 (#8 AND #9)
#11 Antineoplastic Combined Chemotherapy Protocols
#12 (#8 AND #11)
#13 adjuvant NEAR/3 Chemotherap*
#14 (#9 OR #13)
#15 (#10 OR #12 OR #14)
#16 (#6 AND #7 AND #15)

Appendix 2. MEDLINE search strategy

1. randomised controlled trial.pt.
2. controlled clinical trial.pt.
3. randomized.ab.
4. placebo.ab.
5. drug therapy.fs.
6. randomly.ab.
7. trial.ab.
8. groups.ab.
9. or/1-8
10. (animals not (humans and animals)).sh.
11. 9 not 10
12. (Digest$ or Gastr$ or gut or epigastr$ or stomach$).mp.
13. (oesophagogastric junction or esophagogastric junction).mp. [mp=title, original title, abstract, name of substance word, subject heading word, unique identifier]
14. exp Esophagogastric Junction/
15. exp Esophageal Sphincter, Lower/
16. (carcin$ or cancer$ or neoplas$ or tumour$ or tumor$ or cyst$ or adenocarcin$ or malig$).mp.
17. exp Adenocarcinoma/
18. or/12-15
19. exp Stomach Neoplasms/
20. 18 and (16 or 17)
21. 19 or 20
22. exp gastrectomy/
23. gastrectomy.tw.
24. or/22-23
25. *Postoperative Care/
26. chemotherapy.mp.
27. 25 and 26
28. Postoperative Care/
29. Antineoplastic Combined Chemotherapy Protocols/
30. 28 and 29
31. exp Chemotherapy, Adjuvant/
32. (adjuvant adj2 Chemotherap$).tw.
33. 31 or 32
34. 27 or 30 or 33
35. 21 and 24 and 34
36. 11 and 35

Appendix 3. EMBASE search strategy

1. Clinical trial/
2. Randomized controlled trial/
3. Randomization/
4. Single-Blind Method/
5. Double-Blind Method/
6. Cross-Over Studies/
7. Random Allocation/
8. Placebo/
9. Randomi?ed controlled trial$.tw.
10. Rct.tw.
11. Random allocation.tw.
12. Randomly allocated.tw.
13. Allocated randomly.tw.
14. (allocated adj2 random).tw.
15. Single blind$.tw.
16. Double blind$.tw.
17. ((treble or triple) adj blind$).tw.
18. Placebo$.tw.
19. Prospective study/
20. or/1-19
21. Case study/
22. Case report.tw.
23. Abstract report/ or letter/
24. or/21-23
25. 20 not 24
26. (Digest$ or Gastr$ or gut or epigastr$ or stomach$).mp.
27. (oesophagogastric junction or esophagogastric junction).mp. [mp=title, abstract, subject headings, heading word, drug trade name, original title, device manufacturer, drug manufacturer name]
28. exp Lower Esophageal Sphincter/
29. (carcin$ or cancer$ or neoplas$ or tumour$ or tumor$ or cyst$ or adenocarcin$ or malig$).mp.
30. exp Adenocarcinoma/
31. or/26-28
32. exp stomach tumor/ or exp stomach cancer/ or exp stomach adenocarcinoma/
33. 31 and (29 or 30)
34. 32 or 33
35. exp gastrectomy/
36. gastrectomy.tw.
37. or/35-36
38. *Postoperative Care/
39. chemotherapy.mp.
40. 38 and 39
41. Postoperative Care/
42. exp antineoplastic agent/
43. 41 and 42
44. exp Chemotherapy, Adjuvant/
45. (adjuvant adj2 Chemotherap$).tw.
46. 44 or 45
47. 40 or 43 or 46
48. 34 and 37 and 47
49. 25 and 48

Appendix 4. Science Citation Index search strategy

# 1Topic=(Digest* or Gastr* or gut or epigastr* or stomach*) OR Topic=(oesophagogastric junction or esophagogastric junction) OR Topic=(Lower Esophageal Sphincter)

Databases=SCI-EXPANDED, SSCI, A&HCI Timespan=1976-2011

# 2 Topic=(carcin* or cancer* or neoplas* or tumour* or tumor* or cyst* or adenocarcin* or malig*)

Databases=SCI-EXPANDED, SSCI, A&HCI Timespan=1976-2011

# 3 #2 AND #1

Databases=SCI-EXPANDED, SSCI, A&HCI Timespan=1976-2011

# 4 Topic=(gastrectomy)

Databases=SCI-EXPANDED, SSCI, A&HCI Timespan=1976-2011

# 5 Topic=(postoperative) AND Topic=(chemotherapy)

Databases=SCI-EXPANDED, SSCI, A&HCI Timespan=1976-2011

# 6 #5 AND #4 AND #3

Databases=SCI-EXPANDED, SSCI, A&HCI Timespan=1976-2011

Contributions of authors

RDN wrote the review, and assessed the trials for inclusion and extracted data on included trials. ROR independently assessed the trials for inclusion and extracted data on included trials. MW provided advice for improving the review; reviewed the data and the manuscript.

Declarations of interest

None

Sources of support

Internal sources

  • Virgen de la Victoria University Hospital, Malaga, Spain.

  • Reina Sofía University Hospital, Cordoba, Spain.

  • Royal Free Hospital and Medical School, London, UK.

External sources

  • No sources of support supplied

Differences between protocol and review

One-year, two-year, three-year, four-year, five-year and 10-year actual survival were originally planned for OS and DFS but were omitted because these analyses were made redundant by the use of hazard ratios.

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Allum 1989

MethodsRandomised Clinical Trial
Participants

Country: United Kingdom.

Sample size: 455.

Post-randomisation dropouts: 44.

Revised sample size: 411.

Females: 130.

Mean age: 63.

Intervention/Control: 141/130. *140 patients received induction chemotherapy (3rd group).

Median follow-up (months): 100.

Inclusion Criteria:

-Gastric adenocarcinoma.

Exclusion Criteria:

-Previous malignancy.

-Previous chemo-radiotherapy.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Futher details: 5-FU 15mg/kg and mitomycin C 150g/kg very 3 weeks.

Outcomes

Primary outcome: OS

Secondary outcomes: Not reported.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
Low riskPlacebo
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasLow riskAdequate early stop quote and no sponsor bias.

Bajetta 2002

MethodsRandomised Clinical Trial
Participants

Country: Italy.

Sample size: 274.

Post-randomisation dropouts: 14.

Revised sample size: 260.

Females: 97.

Mean age: 57.

Intervention/Control: 135/136.

Median follow-up (months): 66.

Inclusion Criteria:

-Resected gastric adenocarcinoma.

Exclusion Criteria:

-WHO performance status > 2.

-Previous malignancy.

-Previous chemo-radiotherapy.

-pT4.

-N +.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: Etoposide 120mg/m² + Adriamycin 20mg/m² +Cisplatin 40mg/m² + 5-FU + Leucovorin.

Outcomes

-Primary outcome: OS and DFS.

-Secondary outcomes: toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskComputer generated
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasLow riskAdequate baseline imbalance, early stop and sponsor biases.

Bonfanti 1988

MethodsRandomised Clinical Trial
Participants

Country: Italy.

Sample size: 235.

Post-randomisation dropouts: 22.

Revised sample size: 213.

Females: 75.

Mean age:

Intervention/Control: 75/69. *Other 69 patiens received chemo and levamisole.

Median follow-up (months): NR.

Inclusion Criteria:

-Complete removal of gastric adenocarcinoma.

-Availability for frequent follow-up.

Exclusion Criteria:

-T1 stage.

->70 years old.

-Previous malignancy.

-Previous chemo-radiotherapy.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Postsurgical chemotherapy.

Further details: Me-CCNU 130mg/m² + 5-FU 325mg/m² + levamisole.

Outcomes

Primary outcome: OS.

Secondary outcomes: NR.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Low riskSealed envelopes
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)High riskSome important outcomes were not reported.
Other biasLow riskAdequate baseline imbalance.

Bouché 2005

MethodsRandomised Clinical Trial
Participants

Country: France.

Sample size: 278.

Post-randomisation dropouts: 18.

Revised sample size: 260.

Females: 74.

Mean age: 61.

Intervention/Control: 127/133.

Median follow-up (months): 84.

Inclusion Criteria:

-Gastric adenocarcinoma.

-R0.

Exclusion Criteria:

-WHO performance status >2.

-Linitis plastica.

-Previous-concurrent malignancy.

-Previous chemo-radiotherapy.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: 5-FU 500mg/m² + Cisplatin 100mg/m².

Outcomes

Primary outcomes: OS and DFS.

Secondary outcomes: toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasLow riskAdequate base imbalance.

Chipponi 2004

MethodsRandomised Clinical Trial
Participants

Country: France.

Sample size: 205.

Post-randomisation dropouts: 9

Revised sample size: 196.

Females: 67.

Mean age: 61.

Intervention/Control: 93/103.

Median follow-up (months): NR.

Inclusion Criteria:

-Resected Gastric adenocarcinoma with no macroscopic margin involvement.

Exclusion Criteria:

-> 75 years old.

-Previous malignancy.

-Previous chemo-radiotherapy.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: Leucovorin 200mg/m² + 5-FU 375mg/m² + Cisplatin.

Outcomes

Primary outcome: OS.

Secondary outcome: toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskCentralized random permuted block technique
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported.
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasHigh riskEarly stopping bias

Chou 1994

MethodsRandomised Clinical Trial
Participants

Country: Taiwan.

Sample size: 126.

Post-randomisation dropouts: 11.

Revised sample size: 115.

Females: 48.

Mean age: 60.

Intervention/Control: 59/56.

Median follow-up (months): NR.

Inclusion Criteria:

-Stage II and III Gastric adenocarcinoma with clear margin resection.

Exclusion Criteria:

-Stage I and IV.

Interventions

Post-surgical chemotherapy.

Further details: Ftorafur 10mg/kg. Oral vitamins for control group.

OutcomesPrimary outcome: OS.
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)High riskn
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
Unclear riskNot reported
Incomplete outcome data (attrition bias)
All outcomes
High riskn
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasHigh riskMissing outcome bias.

Cirera 1999

MethodsRandomised Clinical Trial
Participants

Country: Spain.

Sample size: 156.

Post-randomisation dropouts: 8.

Revised sample size: 148.

Females: 50.

Mean age: 60.9.

Intervention/Control: 76/72.

Median follow-up (months): 37.

Inclusion Criteria:

-Stage III Gastric adenocarcinoma.

Exclusion Criteria:

->75 years old.

-Karnofsky index < 70.

-R2 surgery.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: Mitomycin 20mg/m² + Tegafur 400mg.

Outcomes

Primary outcomes: OS and DFS.

Secondary outcomes: Toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Low riskSealed envelopes
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasLow riskAdequate baseline imbalance.

Coombes 1990

MethodsRandomised Clinical Trial
Participants

Country: United Kingdom.

Sample size: 330.

Post-randomisation dropouts: 49.

Revised sample size: 281.

Females: 90.

Mean age: 56.6.

Intervention/Control: 133/148.

Median follow-up (months): NR.

Inclusion Criteria:

-Gastric adenocarcinoma.

Exclusion Criteria:

-Karnofsky index <70

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: 5-FU 600mg/m² + Adriamycin 30mg/m² + Mitomycin 10 mg/m².

Outcomes

Primary outcomes: OS and DFS.

Secondary outcomes: Toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Low riskSealed envelopes
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasLow riskAdequate baseline imbalance.

De Vitta 2007

MethodsRandomised Clinical Trial
Participants

Country: Italy.

Sample size: 228.

Post-randomisation dropouts: 3.

Revised sample size: 225.

Females: 94.

Mean age: 63.5.

Intervention/Control: 112/113.

Median follow-up (months): 60.

Inclusion Criteria:

-R0 resected Gastric adenocarcinoma and D1 lymphadenectomy.

Exclusion Criteria:

->70 years old.

-Performance Status >2.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: Epirubicin 60mg/m² + Leucovorin 100mg/m² + 5-FU 375mg/m² + etoposide 80mg/m².

Outcomes

Primary outcomes: OS and DFS.

Secondary outcomes: Toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasUnclear riskNot reported

Di Costanzo 2008

MethodsRandomised Clinical Trial
Participants

Country: Italy.

Sample size: 258.

Post-randomisation dropouts: 13.

Revised sample size: 245.

Females: 101.

Mean age: 59.0.

Intervention/Control: 130/128.

Median follow-up (months): 73.

Inclusion Criteria:

-Resected Gastric adenocarcinoma.

Exclusion Criteria:

->75 years old.

-Performance Status >2.

-Previous malignancy.

-Previous chemo-radiotherapy.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: Cisplatin 40mg/m² + leucovorin 100mg/m² + 5-FU 300mg/m².

Outcomes

Primary outcomes: OS and DFS.

Secondary outcomes: toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
High riskIntention to treat.
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasLow riskAdequate baseline imbalance and sponsor quote.

Douglas 1982

MethodsRandomised Clinical Trial
Participants

Country: United States.

Sample size: 165.

Post-randomisation dropouts: 23.

Revised sample size: 142.

Females: 42.

Mean age: NR.

Intervention/Control: 71/71.

Median follow-up (months): NR.

Inclusion Criteria:

-Gastric adenocarcinoma.

Exclusion Criteria:

-Metastatic disease, including peritoneal spread or malignant ascitics.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: Me-CCNU 150mg/m² + 5-FU 325mg%m².

Outcomes

Primary outcome: OS.

Secondary outcome: toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Low riskOf the 165 randomised patients, 71 patients in each treatment arm were included in the analysis.
Selective reporting (reporting bias)Unclear riskNot reported
Other biasLow riskAdequate baseline imbalance.

Engstrom 1985

MethodsRandomised Clinical Trial
Participants

Country: United States.

Sample size: 213.

Post-randomisation dropouts: 33.

Revised sample size: 180.

Females: 60.

Mean age: NR.

Intervention/Control: 91/89.

Median follow-up (months): NR.

Inclusion Criteria:

-Gastric adenocarcinoma.

Exclusion Criteria:

-Previous malignancy.

-Incomplete resection.

-Metastatic disease including malignant ascitics.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Futher details: 5-FU 350mg/m² + Me-CCNU 150mg/m².

OutcomesPrimary outcomes: OS and DFS.
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Low riskIntention to treat
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasLow riskAdequate baseline imbalance and sponsor quote.

Fielding 1983

MethodsRandomised Clinical Trial
Participants

Country: United Kingdom.

Sample size: 455.

Post-randomisation dropouts: 44.

Revised sample size: 411.

Females: NR.

Mean age: NR.

Intervention/Control: 141/130. *Other 140 patients received induction chemotherapy (3rd group).

Median follow-up (months): NR.

Inclusion Criteria:

-Gastric adenocarcinoma.

Exclusion Criteria:

-Previous malignancy.

-Previous chemo-radiotherapy.

-Stage I or IV.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: 5-FU 15mg/m² + Mitomycin 150 mg/m².

Outcomes

Primary outcome: OS.

Secondary outcome: toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)High riskSome important outcomes were not reported.
Other biasLow riskAdequate baseline imbalance and sponsor quote.

Fujimoto 1977

MethodsRandomised Clinical Trial
Participants

Country: Japan.

Sample size: 249.

Post-randomisation dropouts: 32.

Revised sample size: 217.

Females: NR.

Mean age: NR.

Intervention/Control: 97/120.

Median follow-up (months): NR.

Inclusion Criteria:

-Gastric adenocarcinoma.

Exclusion Criteria:

->80 years old.

-Previous malignancy.

-Previous chemo-radiotherapy.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: 5-FU 3.5mg/kg + FT-207 8mg/kg.

Outcomes

Primary outcome: OS.

Secondary outcome: Toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)Unclear riskNot reported
Other biasLow riskAdequate baseline imbalance

Grau 1993

MethodsRandomised Clinical Trial
Participants

Country: Spain.

Sample size: 134.

Post-randomisation dropouts: 0.

Revised sample size: 134.

Females: 46.

Mean age: 56.5.

Intervention/Control: 68/66.

Median follow-up (months): 105.

Inclusion Criteria:

-Gastric adenocarcinoma.

Exclusion Criteria:

-T1 or N3

-Previous malignancy.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: Mitomycin 20mg/m².

Outcomes

Primary outcome: OS.

Secondary outcome: Toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)Unclear riskNot reported
Other biasUnclear riskNot reported

Hallissey 1994

MethodsRandomised Clinical Trial
Participants

Country: United Kingdom.

Sample size: 283.

Post-randomisation dropouts: 0.

Revised sample size: 283.

Females: 133.

Mean age: 64.0.

Intervention/Control: 138/145.

Median follow-up (months): 84.

Inclusion Criteria:

-Gastric adenocarcinoma.

Exclusion Criteria:

-Previous malignancy.

-Stage I.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: Mitomycin 4mg/m² + Doxorubicin 30mg/m² + 5-FU 600mg/m².

OutcomesPrimary outcome: OS.
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Low riskAllocation by a central office
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
High risk 
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasLow riskAdequate baseline imbalance and sponsor quote.

Higgins 1983

MethodsRandomised Clinical Trial
Participants

Country: United States.

Sample size: 135.

Post-randomisation dropouts: 1.

Revised sample size: 134.

Females: NR.

Mean age: NR.

Intervention/Control: 66/68.

Median follow-up (months): NR.

Inclusion Criteria:

-Gastric adenocarcinoma.

Exclusion Criteria:

-Previous malignancy.

-Previous chemo-radiotherapy.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: 5-FU 9mg/kg + Me-CCNU 4mg/kg.

Outcomes

Primary outcome: OS.

Secondary outcome: toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Low riskSealed envelopes
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
High risk 
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasUnclear riskNot reported

Huguier 1980

MethodsRandomised Clinical Trial
Participants

Country: France.

Sample size: 53.

Post-randomisation dropouts: 3.

Revised sample size: 50.

Females: 15.

Mean age:

Intervention/Control: 27/26.

Median follow-up (months): NR.

Inclusion Criteria:

-Radical resection of Gastric adenocarcinoma.

Exclusion Criteria:

-Previous chemo-radiotherapy.

Interventions

Post-surgical chemotherapy.

Further details: 5-FU 500mg + Cyclophosphamide.

Outcomes

Primary outcomes: OS.

Secondary outcomes: Toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Low riskSealed envelopes
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)Unclear riskNot reported
Other biasLow riskAdequate baseline imbalance and early stopping.

Kim 1992

MethodsRandomised Clinical Trial
Participants

Country: Korea.

Sample size: 200.

Post-randomisation dropouts: 29.

Revised sample size: 171.

Females:

Mean age: 54.

Intervention/Control: 77/94.

Median follow-up (months): NR.

Inclusion Criteria:

-Stage III Gastric adenocarcinoma.

Exclusion Criteria:

->70 years old.

-Performance Status >2.

-Previous malignancy.

-Previous chemo-radiotherapy.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: Mitomycin 4mg/50kg + 5-FU 500mg/50kg + Arabinoside 40mg/50kg.

OutcomesPrimary outcome: OS.
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
High risk 
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasUnclear riskNot reported

Krook 1991

MethodsRandomised Clinical Trial
Participants

Country: United States.

Sample size: 127.

Post-randomisation dropouts: 2.

Revised sample size: 125.

Females: 27.

Mean age: 62.5.

Intervention/Control: 61/64.

Median follow-up (months): NR.

Inclusion Criteria:

-Gastric adenocarcinoma.

Exclusion Criteria:

-Evidence of margin infiltration.

-Previous malignancy.

-Previous chemo-radiotherapy.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further Details: 5-FU 350mg/m² + Doxorubicin 40mg/m².

Outcomes

Primary outcomes: OS and DFS.

Secondary outcomes: Toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasHigh riskInadequate baseline imbalance by age.

Kulig 2010

MethodsRandomised Clinical Trial
Participants

Country: Poland.

Sample size: 309.

Post-randomisation dropouts: 54.

Revised sample size: 155.

Females: 84.

Mean age: 62.5.

Intervention/Control: 154/101.

Median follow-up (months): 37.

Inclusion Criteria:

-Histologically verified nonmetastatic Gastric adenocarcinoma.

Exclusion Criteria:

-<18 years old.

-Karnofsky <70.

-T1-N0.

-Other malignancies in the last 5 years.

Interventions

Post-surgical chemotherapy.

Further details: Doxorubicin 20mg/m²; Cisplatin 40mg/m²; Etoposide 120mg/m².

OutcomesOS and DFS.
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskComputer-generated randomisation list
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)Unclear riskNot reported
Other biasUnclear riskNot reported

Lise 1995

MethodsRandomised Clinical Trial
Participants

Country: Italy.

Sample size: 326.

Post-randomisation dropouts: 12.

Revised sample size: 314.

Females: 112.

Mean age: NR.

Intervention/Control: 155/159.

Median follow-up (months): NR.

Inclusion Criteria:

-Curative resection of Gastric adenocarcinoma.

Exclusion Criteria:

->71 years old.

-Performance Status >2.

-Stage I.

-Previous malignancy.

-Previous chemo-radiotherapy.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: 5-FU 400mg/m² +Doxorubicin 40mg/m² + Mitomycin 10mg/m².

Outcomes

Primary outcomes: OS and DFS.

Secondary outcomes: Toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)Unclear riskNot reported
Other biasUnclear riskNot reported

Macdonald 1995

MethodsRandomised Clinical Trial
Participants

Country: United States.

Sample size: 221.

Post-randomisation dropouts: 28.

Revised sample size: 193.

Females: 70.

Mean age: 59.5.

Intervention/Control: 93/100.

Median follow-up (months): 114.

Inclusion Criteria:

-Gastric adenocarcinoma.

Exclusion Criteria:

-Resection margins infiltration.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further Details: 5-FU 600mg/m² + Doxorubicin 30mg/m² + Mitomycin C 10mg/m².

Outcomes

Primary outcomes: OS and DFS.

Secondary outcomes: Toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)Unclear riskNot reported
Other biasUnclear riskNot reported

Nakajima 1999

MethodsRandomised Clinical Trial
Participants

Country: Japan.

Sample size: 579.

Post-randomisation dropouts: 6.

Revised sample size: 573.

Females: 210.

Mean age: NR.

Intervention/Control: 288/285.

Median follow-up (months): 72.

Inclusion Criteria:

-Gastric adenocarcinoma.

Exclusion Criteria:

-T1.

->75 years old.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: Mitomycin 1-4mg/m² +5-FU 166mg/m² + uracil + tegafur 300mg.

Outcomes

Primary outcome: OS.

Secondary outcomes: Toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskComputer assisted
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Low riskIntention to treat
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasLow riskAdequate early stopping bias and sponsor bias.

Nakajima 2007

MethodsRandomised Clinical Trial
Participants

Country: Japan.

Sample size: 190.

Post-randomisation dropouts: 2.

Revised sample size: 188.

Females: 45.

Mean age: 63.5.

Intervention/Control: 93/95.

Median follow-up (months): 75.

Inclusion Criteria:

-Gastric adenocarcinoma T2, N1-2 stage.

Exclusion Criteria:

-<20 and >75 years old.

-Performance Status >2.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: Uracil-Tegafur 360mg/m².

Outcomes

Primary outcomes: OS and DFS.

Secondary outcomes: Toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskMinimization
Allocation concealment (selection bias)Low riskMinimization
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
High riskDrops out
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasHigh riskSponsor bias

Nashimoto 2003

MethodsRandomised Clinical Trial
Participants

Country: Japan.

Sample size: 252.

Post-randomisation dropouts: 2.

Revised sample size: 250.

Females: 81.

Mean age: 58.5.

Intervention/Control: 127/123.

Median follow-up (months): 69.

Inclusion Criteria:

-Gastric adenocarcinoma.

Exclusion Criteria:

->75 years old.

-T1 or N0.

-Serosal involvement.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: Mitomycin 1.33mg/m² + FU 166.7 mg/m² + Ara-C13.3mg/m².

Outcomes

Primary outcomes: OS and DFS.

Secondary outcomes: Toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Low riskMinimization
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)Unclear riskNot reported
Other biasUnclear riskNot reported

Neri 2001

MethodsRandomised Clinical Trial
Participants

Country: Italy.

Sample size: 137.

Post-randomisation dropouts: 0.

Revised sample size: 137.

Females: 39.

Mean age: 63.0.

Intervention/Control: 69/68.

Median follow-up (months): NR.

Inclusion Criteria:

-Gastric adenocarcinoma.

Exclusion Criteria:

-Karnofsky index < 60.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: Epidoxirubicin 75mg/m² + Leucovorin 200mg/m² + 5-FU 450mg/m².

Outcomes

Primary outcome: OS.

Secondary outcome: Toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasUnclear riskNot reported

Nitti 2006 EORTC

MethodsRandomised Clinical Trial
Participants

Country: Italy.

Sample size: 206.

Post-randomisation dropouts: 0.

Revised sample size: 206.

Females: 79.

Mean age: 56.0.

Intervention/Control: 103/103.

Median follow-up (months): 79.

Inclusion Criteria:

-Gastric adenocarcinoma.

Exclusion Criteria:

-Macroscopically affected margins.

->71years old.

-Karnosfky index < 60 or Permormance Status >2

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details:Methotrexate 1.5mg/m² + 5-FU 1.5g/mg + leucovorin 15mg/m² + adriamycin 30mg/m².

Outcomes

Primary outcomes: OS and DFS.

Secondary outcomes: Toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskMinimization and stratification
Allocation concealment (selection bias)Low riskMinimization
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Low riskIntention to treat
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasHigh riskEarly stopping bias and sponsor bias.

Nitti 2006 ICCG

MethodsRandomised Clinical Trial
Participants

Country: Italy.

Sample size: 191.

Post-randomisation dropouts: 8.

Revised sample size: 183.

Females: 66.

Mean age: 56.0.

Intervention/Control: 91/100.

Median follow-up (months): 77.

Inclusion Criteria:

-Gastric adenocarcinoma.

Exclusion Criteria:

-Macroscopically affected margins.

->71years old.

-Karnosfky index < 60 or Permormance Status >2.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: 5-FU + leucovorin + Methotrexate + epirubicin 70mg/m².

Outcomes

Primary outcomes: OS and DFS.

Secondary outcomes: Toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
High risk 
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasHigh riskSponsor bias

Ochiai 1983

MethodsRandomised Clinical Trial
Participants

Country: Japan.

Sample size: 140.

Post-randomisation dropouts: 2.

Revised sample size: 138.

Females: 32.

Mean age: NR.

Intervention/Control: 49/40. *Other 51 patients received immunochemotherapy.

Median follow-up (months): NR.

Interventions

Post-surgical chemotherapy.

Further details: Mitomycin 2mg + 5 FU 250mg + Ara-C 20mg.

OutcomesPrimaru outcome: OS.
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Low riskSealed envelopes
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)Unclear riskNot reported
Other biasUnclear riskNot reported

Popiela 1982

MethodsRandomised Clinical Trial
Participants

Country: Poland.

Sample size: 99.

Post-randomisation dropouts: 0.

Revised sample size: 99.

Females:

Mean age: 57.

Intervention/Control: 16/44. *Other 39 patients received immunochemotherapy.

Median follow-up (months): 24.

Inclusion Criteria:

-Gastric adenocarcinoma.

Interventions

Post-surgical chemotherapy.

Further details: 5-FU 15mg/kg.

OutcomesPrimary outcome: OS.
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasUnclear riskNot reported

Popiela 2004

MethodsRandomised Clinical Trial
Participants

Country: Poland.

Sample size: 164.

Post-randomisation dropouts: 8.

Revised sample size: 156.

Females: 31.

Mean age: 58.3.

Intervention/Control: 53/52. *Other 51 patients received immunochemotherapy.

Median follow-up (months): NR.

Inclusion Criteria:

-R0-R1 resection of gastric adenocarcinoma and D2 lymphadenectomy.

Exclusion Criteria:

->70 years old.

-Performance Status>2.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: 5-FU 600mg/m² + Adriamycin 30mg/m² + Mitomycin-C 10mg/m².

Outcomes

Primary outcome: OS.

Secondary outcome: Toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Low riskSealed envelopes
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasUnclear riskNot reported

Sakuramoto 2007

MethodsRandomised Clinical Trial
Participants

Country: Japan.

Sample size: 1059.

Post-randomisation dropouts: 25.

Revised sample size: 1034.

Females: 323.

Mean age: 63.

Intervention/Control: 517/526.

Median follow-up (months): NR.

Inclusion Criteria:

-R0 surgery of gastric adenocarcinoma.

Exclusion Criteria:

-<20 and >80 years old.

-T1.

-Previous malignancy.

-Previous chemo-radiotherapy.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: Tegafur + gimeracil + oteracil.

Outcomes

Primary outcomes: OS and DFS.

Secondary outcomes: Toxicity.

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskMinimization
Allocation concealment (selection bias)Low riskMinimization
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
High riskDropouts
Selective reporting (reporting bias)Low riskAll important outcomes were reported.
Other biasHigh riskSponsor bias

Tentes 2006

MethodsRandomised Clinical Trial
Participants

Country: Greece.

Sample size: 40.

Post-randomisation dropouts: 0.

Revised sample size: 40.

Females: 15.

Mean age: 65.1.

Intervention/Control: 20/20.

Median follow-up (months): 52.

Inclusion Criteria:

-Gastric adenocarcinoma.

Exclusion Criteria:

-T1-2.

-Karnofsky index <50.

-Previous malignancy.

-Previous chemo-radiotherapy.

-Metastatic disease.

-Contraindication for surgery or chemotherapy.

Interventions

Post-surgical chemotherapy.

Further details: 5-FU 600mg/m² + doxorubicin 15mg/m² + Mitomycin 3.7mg/m².

OutcomesPrimary outcome: OS.
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Unclear riskNot reported
Blinding (performance bias and detection bias)
All outcomes
High riskNo blinding described
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)Unclear riskNot reported
Other biasUnclear riskNot reported

Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion
Alcobendas 1983Same study as Grau 1993.
Bresciani 2000Non-randomized.
Chen 2005Hormonal therapy.
Chen 2011Non-randomized.
Cunninghan 2006Pre-surgical chemotherapy.
Estape 1991Same study as Grau 1993.
Macdonald 2001Chemoradiotherapy.
Moertel 1984Radiotherapy.
Nio 1992Inmunochemotherapy.
Strong 2010Non-randomized.
Yamamura 1986Inmunochemotherapy.
Yu 1995Radiotherapy.

Characteristics of studies awaiting assessment [ordered by study ID]

Bang 2012

MethodsRandomized clinical trial.
Participants1035
InterventionsPost-surgical capecitabine and oxaliplatin.
OutcomesImprovement of 3-years DFS rate in the intervention group.
Notes 

Sasako 2011

MethodsRandomized clinical trial.
Participants1059
InterventionsPost-surgical chemotherapy with S-1.
OutcomesImprovement of OS and DFS in the intervention group.
Notes 

Ancillary