We identified trials by searching PubMed, Cochrane Library, Ebsco, Chinese National Knowledge Infrastructure, and Chinese Biomedical databases (up until April 2017). No language restrictions applied. PubMed searched using trastuzumab and the exploded MeSH term gastric neoplasms . We reviewed the trial registries of Clinical Trials.gov, National Cancer Institute Clinical Trials to ensure the included trials.

We selected phase II or III randomized controlled trials (RCTs) comparing the efficacy of trastuzumab-based chemotherapy in gastric cancer. We included trials of gastric cancer patients with HER2-positive status. We excluded trials if they compared only the outcomes of different HER2 treatment dosing scheme, only measurement toxic effects, quality of life, pharmacokinetic efficacy or without control and case-control experiment. Studies were excluded without sufficient published data for estimation of the odds ratio (OR), and 95% confidence interval (CI) as well.

We extracted data strictly abide by Cochrane guidelines. Two reviewers (Huahui Li and Bihua Lin) evaluated independently the titles and abstracts of eligible publications to determine trial inclusion. Including design of trial, eligibility of patient, baseline characteristics of patient, dosing scheme, treatment line, HER2 identification method, follow-up duration and therapy changes on disease progression. Any disagreements between reviewers were resolved by a third researcher (Keyuan Zhou). If the test results repeat in various publications, we got the most complete endpoints.

We abstracted hazard ratios (HRs) and 95% confidence intervals (CIs) for the endpoints OS, TTP and PFS. We also abstracted the endpoints of overall partial response (PR), complete response (CR), stable disease (SD), poresponse rate (ORR), progressive disease (PD), time to treatment failure (TTF), clinical benefit rates (CBR), and total of deaths. The data extract documented relevant items, involving the surname of author, year of publication, age of the participants, HER2 status, and sample number of trial.

We identified the titles and abstracts of 446 citations, 69 articles excluded. The full-text reports of 377 articles were took to further determine eligibility. Night manuscripts were included in the meta-analysis171819202122232425. The night trials represented 1101 patients, 562 obtaining the trastuzumab and 536 receiving chemotherapy alone (Figure 2). If papers included both HER2 positive and negative patients, only HER2 patients were included in our meta-analysis18. Chemotherapy partners included capecitabine + cisplatin or 5-fluorouracil + cisplatin(5 trials)1719202125, Docetaxel + cisplatin + 5-fluorouracil (2 trials)1823, Irinotecan + 5-fluorouracil/Leucovorin (1 trial)21, trastuzumab + capecitabine + oxaliplatin (1 trial). Overall, 4 studies were reported in English and 5 in Chinese. Trials were either first line or a second or subsequent line of therapy. The trials of HER2 positive disease was via identified immunohistochemistry (IHC) + or FISH+. We further analysis of these studies characteristic in Table 1.

Our meta-analysis exhibited a 20% decrease in the hazard of death with the trastuzumab in combination with chemotherapy to chemotherapy alone (5 trials, 933 patients, HR = 0.80; 95% CI = 0.72 - 0.89) (Figure 3). The benefit of trastuzumab over chemotherapy alone was also seen with the endpoints of PFS (HR = 0.70; 95% CI = 0.59 - 0.83) and TTP (HR = 0.69; 95% CI = 0.57 - 0.83) and pooled results for outcomes (Figures 3). The addition of trastuzumab therapy to chemotherapy treatment have an insignificance ORR (Relative risk (RR) = 1.22; 95% CI = 0.94 - 1.59) (Figure 4). The addition of trastuzumab also was beneficial for the outcomes of partial response (RR = 1.26, 95% CI = 0.99 - 1.59). Disease control rate (complete   remission + partial response + Stable disease) demonstrated a 19% increase in the relative risk (6 trials, 942 patients, RR = 1.19; 95% CI = 1.10 - 1.28) (Figure 5). The incidence of adverse events was similar in the control and trial arms of the addition of trastuzumab therapy. A difference was only observed with increased in grade  3~4  adverse reactions in patients with diarrhea in the trial arms (4 trials, 875 patients, RR = 2.40; 95% CI = 1.34–4.32) (Figure 6). Testing publication bias by endpoint parameters showed no evident asymmetry in the funnel plots by the Egger’s test (P > 0.05).

Oursensitivity analysis was executed by successively eliminating studies one by one. The elimination of one study each time revealed no significant difference in whole estimates, suggesting the stabilization of these results.

Our pooled results of 9 studies show that the addition of trastuzumab therapies to chemotherapy provides a 20% improvement in OS. PFS and TTP show a 30% and 31% improvement, respectively, and significance increase in ORR of 41%. The pooled results improve some of the uncertainties about the benefit of trastuzumab therapies. Some trials included only small patients don t report on the endpoint of OS, PFS or TTP. A few HER2 combination therapies have been evaluated in a randomized controlled trials design, specifically capecitabine, cisplatin, 5-fluorouracil, irinotecan, docetaxel, Leucovorin and oxaliplatin). It may be helpful to carry out further clinical trials to compare the efficacy of trastuzumab therapy over chemotherapy. The clinical evidence suggest that trastuzumab therapy could be continued on patients with HER2-positive gastric cancer, even in patients who need to discontinue chemotherapy because of adverse reactions. Additionally, more data are still needed to confirm whether trastuzumab should be continued after disease progression. The introduction of trastuzumab led to the establishment of a new disease entity, HER2-positive gastric cancer, similar to HER2-positive breast cancer. It is expected that more and more anti-HER2 drugs should be developed and introduced into clinical practice to treat patients with HER2-positive gastric cancer.

Due to the various therapeutic scheme and small sample of trials, we can t undertake a subgroup analysis of individual therapeutic scheme. This study also has a lot of limitations, some of which are approach in meta-analysis and others detailed to our analysis. We don t have personal patient data in order to confirm outcomes data showed in publications. Moreover, we are incapable to confirm blinding procedures or randomization26. In addition, the current included trials are lack of long-term follow-up data. These limitations in clinical practice call for building an unbiased and current evidence to support the best use of trastuzumab therapies. Reporting diverse greatly in the included trials such as the endpoints are used and the way are expressed. The endpoints HRs and CIs included in our study are not always reported in every trial. However, many trials report the period of advantage in days, weeks or months. We try to fill some of the gaps by contacting individual researchers, but we fail to obtain additional data.

The results of this study show that trastuzumab-based chemotherapy can significantly improve overall survival in patients with HER2-positive advanced gastric cancer, compared with chemotherapy alone, and this improvement is particularly significant in patients with high HER2 expression. It is also notable that trastuzumab don t increase the incidence of adverse events that associated with chemotherapy and that the rate of cardiac events is also low. Therefore, These studies should support an indication for trastuzumab as part of a perioperative chemotherapeutic regimen for treating HER2-positive gastric cancer.

In conclusion, this study provides obvious efficacy of trastuzumab-based chemotherapy versus chemotherapy alone in gastric cancer, in terms of OS, PFS, TTP, RR and DCR endpoints. Trastuzumab-based chemotherapy increased grade 3 or 4 adverse reactions in patients with diarrhea, but there is not enough evidence that will add other adverse reactions in patients. A well-designed RCTs should establish to determine the ideal therapeutic scheme for trastuzumab therapies.