Overall survival benefits for irinotecan-containing regimens as first-line treatment for advanced gastric cancer: An updated meta-analysis of ten randomized controlled trials



The standard treatment for patients with advanced gastric cancer (AGC) is still debated, and the available data on the benefit of irinotecan-containing regimen as first-line treatment for those patients are controversial. We performed a systematic review and meta-analysis of randomized controlled trials to determine the survival benefits of irinotecan-containing regimens in this setting. A total of 1,837 patients from ten trials were included in the analysis. Our results showed that irinotecan-containing regimens significantly improved overall survival [OS: hazard ratio (HR) 0.86, 95% CI = 0.78–0.94, p = 0.002] and progression-free survival [HR = 0.82, 95% CI = 0.69–0.97, p = 0.026); however, the improvement of time to failure (HR = 0.90; 95% CI = 0.77–1.04, p = 0.15), 1-year survival rate [1-year SR: relative risk (RR) 1.10, 95% CI = 0.97–1.24, p = 0.13] and overall response rate (RR = 1.16, 95% CI = 0.91–1.49, p = 0.24] were nonsignificant. Equivalent frequencies of toxicities were found between the two groups excluding more Grade 3 or 4 fatigue (p = 0.001) in irinotecan-containing regimens. This updated meta-analysis provided strong evidence for a survival benefit of irinotecan-containing regimen as first-line treatment for AGC. A clear advantage of irinotecan-containing over nonirinotecan-containing regimen had not been established. These results should help to inform decisions about patient management and design of future trials.

Gastric cancer is the fourth most frequent malignant disease and the second most common cause of cancer-related deaths worldwide. A total of 989,600 new gastric cancer cases and 738,000 deaths are estimated to have occurred in 2008, accounting for 8% of the total cases and 10% of total cancer-related deaths, respectively.1–3 Although the overall incidence of gastric cancer in the Western countries, Japan, Korea and China has decreased, the incidence of proximal gastric cancer has increased.4 Moreover, the decline in gastric cancer incidence in China has been slower than in the Western countries, Japan and Korea and will be offset by population growth and aging.5 As a result, gastric cancer remains an important public health burden in the world.

As for early and locally advanced gastric cancer (AGC), surgery remains the mainstay of curative treatment; however, relapse is common, and the majority of patients present with advanced disease. Although systemic chemotherapy has a proven palliative role in patients with AGC, significantly improving quality of life and prolonging survival compared to best supportive care alone6; the prognosis of patients in this setting remains dismal with overall 5-year survival rate ranging from 10 to 15% in the Western countries, Japan and Korea.7 During the past decades, a large number of chemotherapy regimens including 5-fluorouracil (5-FU) and cisplatin (CF), epirubicin, cisplatin and infused 5-FU (ECF) or docetaxel, cisplatin and 5-FU (DCF) regimens have been tested in clinical studies6, 8–13; however, the additional survival advantage yielded by these combination therapies appears to be marginal. As a result, there is still no internationally accepted standard of care, and uncertainty remains regarding the choice of the regimen. Obviously, it is necessary to develop new active agents and combination regimens to achieve greater survival benefits in AGC.

Irinotecan is a semisynthetic, water-soluble derivative of the plant alkaloid camptothecin. Following conversion to its active metabolite, SN-38, irinotecan acts by inhibiting DNA topoisomerase-I, thereby interfering with DNA replication and cell division.14, 15 In early phase trials in AGC, single-agent irinotecan administered every 2, 3 or 4 weeks demonstrates efficacy in both first- and second-line treatment settings with response rates of 18–23%16–18 and 43.9–48% when combined with cisplatin.19, 20 Moreover, two published meta-analyses comparing irinotecan-containing versus nonirinotecan-containing chemotherapy prior to this study also confirm a nonsignificant survival benefit of irinotecan-containing chemotherapy due to relatively a few trials included.6, 21

Recently, more randomized controlled trials (RCTs) investigating the efficacy of irinotecan-containing chemotherapies in AGC have been conducted; however, the significance of irinotecan-containing chemotherapies in AGC should be reassessed.22–28 As a result, we systematize the available information to perform an updated meta-analysis of all RCTs on irinotecan-containing versus nonirinotecan-containing chemotherapies in AGC and to analyze and examine the survival benefits and the adverse reactions with irinotecan-containing chemotherapy.


Search strategy

We searched PubMed (up to January 2012), Embase (1980 to January 2012) and the Cochrane Register of Controlled Trials using various combinations of different terms “advanced,” “metastatic,” “gastric cancer,” “first-line treatment,” “irinotecan,” “randomized” and “CPT-11.” We also looked at posters from the annual meetings of the European Society of Medical Oncology and the American Society of Medical Oncology in the past 10 years. The search was limited to clinical studies in English language, and reference lists from relevant primary studies and review articles were also examined to find additional publications.

Study selection

The relevant clinical trials were manually selected carefully based on the following criteria: (i) trails comparing irinotecan-containing with nonirinotecan-containing chemotherapy; (ii) patients were pathologically confirmed of AGC; (iii) prospective phase II and III RCTs and (iv) the included study has sufficient data for extraction. Trials investigating immunotherapy or neoadjuvant or perioperative chemotherapy were excluded. Likewise, trials with radiotherapy or intraperitoneal chemotherapy were not in the scope of our research. If multiple publications of the same trial were retrieved or if there was a case mix between publications, only the most recent publication (and the most informative) was included.

Data extraction and quality assessment

Two independent investigators reviewed the publications and extracted the data. The following information was extracted from each article: (i) basic information from papers such as year of publication, country and author name; (ii) characteristics of patients such as age, percent of female patients and disease burden; (iii) information of study designation such as sample size per group, study design, randomization scheme, inclusion criteria and type of end point used and (iv) information of treatment such as treatment regimens, median overall survival (OS), progression-free survival (PFS), time to failure (TTF), 1-year survival rate (1-year SR), overall response rate (ORR) and adverse events (AEs). Available information was extracted and recorded to a data collection form and entered into electronic database. The quantitative 5-point Jadad scale was used to assess the quality of included trials based on the report of the methods and results of the studies.29

Data analysis

The analysis was undertaken on an intention-to-treat basis: patients were analyzed according to treatment allocated, irrespective of whether they received that treatment. Statistical analysis of the overall hazard ratio (HR) for OS, PFS and TTF, the risk ratio (RR) for 1-year SR, ORR and Grade 3 or 4 AEs was calculated using Stata version 12.0 software (Stata Corporation, College Station, TX). When OS, PFS and TTF could not be extracted from the original reports directly in several RCTs, we deciphered them from the survival curve as reported by Parmar et al.30 Between-study heterogeneity was estimated using the χ2-based Q statistic.31 Heterogeneity was considered statistically significant when pheterogeneity < 0.05 or I2 > 50%. If heterogeneity existed, data were analyzed using a random-effects model. In the absence of heterogeneity, a fixed-effects model was used. Sources of heterogeneity were appraised by subgroup stratification analysis based on several study characteristics such as ethnicity, country and chemotherapy regimen. A statistical test with a p-value less than 0.05 was considered significant. HR > 1 reflected more deaths or progression in irinotecan-containing regimen, and RR > 1 indicated more toxicities, ORR in irinotecan-containing regimen and vice versa. The presence of publication bias was evaluated by using the Begg and Egger tests.32, 33 All p-values were two sided. All CIs had a two-sided probability coverage of 95%. Power calculation was carried out with the power and sample size calculation software34 (PS version 3.0).


Quantity and quality of evidence

The flow chart of our study is shown in Figure 1. A total of 313 studies were retrieved electronically, 82 duplicated papers were excluded and 174 articles were removed on title and abstract. The articles did not meet the inclusion criteria for the following multiple reasons: five retrospective studies, two Phase I studies, 18 case reports and 149 irrelevant contents. Full-text copies of the remaining 57 citations were obtained and evaluated in more detail. Of these citations, 51 were excluded for the following reasons: four citations were meta-analysis of RCTs; nine citations were review articles; 22 trials were single-arm phase II trials; two trials were RCTs, but irinotecan was included in both treatment arms; seven trials were RCTs, but irinotecan was used as second-line treatments; two RCTs were preliminary reports of ongoing RCTs; three RCTs reported quality of life and toxic effects only and two RCTs included both AGC and esophagogastric cancer.35, 36 The remaining six trials were included in the review. Four additional conference abstracts were located as a result of hand searching. Finally, a total of ten publications were included in the review; these publications are related to six clinical trials reported in the full-text publications22–24, 37–39 and four conference abstracts.25–28 The total number of patients randomly assigned in these trials was 1,837, with 786 in the irinotecan-containing chemotherapy arm and 1,051 in the nonirinotecan-containing chemotherapy arm. Two of ten included trials were large, Phase III, multicentered, and randomized clinical studies22, 24; there was no placebo-controlled double-blinded trial. Characteristics of these eligible trials are given in Table 1. The quality of each included study was roughly assessed according to Jadad scale, and five trials had Jadad scores of 3 and five trials had Jadad scores of 2 (Table 1).

Figure 1.

Flow chart of trial selection process.

Table 1. Overview of studies in the pooled analysis (N = 1,837)
inline image

Overall survival

Five of the ten trials reported OS data.22, 24, 37–39 As two trials were three-arm studies,22, 37 the number of comparison was seven. The pooled HR for OS showed significant improvement in OS in irinotecan-containing chemotherapy yielding HR of 0.86 (95% CI = 0.78–94, p = 0.002; Fig. 2) using a fixed-effects model; there was no significant heterogeneity between studies (I2 = 0%, p = 0.91).

Figure 2.

Fixed-effects model of hazard ratio (95% CI) of OS associated with irinotecan-containing versus nonirinotecan-containing regimens.

Progression-free survival

Four trials reported PFS data22, 37–39 and two trials were three-arm studies,22, 37 and therefore, the number of comparison was six. The pooled HR for PFS confirmed that irinotecan-containing chemotherapy was superior to nonirinotecan-containing chemotherapy (HR = 0.82, 95% CI = 0.69–0.98, p = 0.026; Fig. 3). There was significant heterogeneity (I2 = 64.1%, p = 0.016), and the pooled HR for PFS was performed by using random-effects model.

Figure 3.

Random-effects model of hazard ratio (95% CI) of PFS associated with irinotecan-containing versus nonirinotecan-containing regimens.

Time to failure

Three trials reported TTF data.22, 38, 39 The pooled HR for TTF showed that there was a nonsignificant improvement for irinotecan-containing chemotherapy giving HR = 0.90 (95% CI = 0.77–1.04, p = 0.15). There was no significant heterogeneity between trials (I2 = 0%, p = 0.37), and the pooled HR for TTF was performed by using fixed-effects model.

One-year survival rate

Three trials reported 1-year survival data.22, 24, 37 As two trials were three-arm studies,22, 37 the number of comparison was five. The pooled RR for 1-year survival rate showed a nonsignificant improvement for irinotecan-containing chemotherapy (RR = 1.10, 95% CI = 0.97–1.24, p = 0.13). There was no significant heterogeneity (I2 = 0%, p = 0.97), and the pooled RR for 1-year survival rate was performed using fixed-effects model.

Overall response rate

All ten trials reported ORR data, and the pooled RR for ORR showed that there was a nonsignificant improvement for irinotecan-containing chemotherapy with RR = 1.16 (95% CI = 0.91–1.49, p = 0.24; Fig. 4). There was significant heterogeneity between the trials (I2 = 50.9%, p = 0.031), and the pooled RR for overall response was performed using random-effects model.

Figure 4.

Random-effects model of risk ratio (95% CI) of ORR associated with irinotecan-containing versus nonirinotecan-containing regimens.


Pooled analysis of reported Grades 3 and 4 AEs of interest was also performed. There were more incidences of Grade 3 or 4 fatigue (RR = 1.97, 95% CI = 1.31–2.97, p = 0.001) in irinotecan-containing regimens. With regard to the risk of Grade 3 or 4 anemia (RR = 1.25; 95% CI = 0.77–2.04, p = 0.37), neutropenia (RR = 1.41; 95% CI = 0.56–3.44, p = 0.46), febrile neutropenia (RR = 1.30, 95% CI = 0.56–3.02, p = 0.55), thrombopenia (RR = 0.71; 95% CI = 0.22–2.31, p = 0.57), diarrhea (RR = 2.05, 95% CI = 0.91–4.66, p = 0.085), stomatitis (RR = 0.47, 95% CI = 0.21–1.06, p = 0.07) and nausea and vomiting (RR = 1.22, 95% CI = 0.58–2.57, p = 0.60), equivalent frequencies were found between the two groups.

Subgroup analysis

Subgroup analysis could help us discover potential information of what the clinicians were interested in. Therefore, we studied some factors that might be related with survival between the two groups. The studies from the Eastern or the Western countries, phase of trials and different chemotherapy regimens were considered as the subgroup analysis factors. Finally, all subgroup analyses favored irinotecan-containing chemotherapy in terms of OS; however, the improvement was not significant in Phase II trials and non-Asian patients (Table 2).

Table 2. Subgroup analysis for OS
inline image

Publication bias

Begg's funnel plot and Egger's test were performed to assess the publication bias of literature. The shapes of the funnel plots did not reveal any evidence of obvious asymmetry (p = 0.881 for OS; p = 0.851 for PFS; p = 0.602 for TTF; p = 0.221 for 1-year SR; p = 0.929 for ORR). Then, Egger's test was used to provide statistical evidence of funnel plot symmetry. The results still did not suggest any evidence of publication bias (p = 0.798 for OS; p = 0.785 for PFS; p = 0.52 for TTF; p = 0.054 for 1-year SR; p = 0.167 for ORR).


There were two meta-analyses published prior to our study. In 2006, Wagner et al.6 did the first meta-analysis to assess the efficacy and tolerability of chemotherapy in patients with AGC and demonstrated a nonsignificant survival benefit in favor of irinotecan-containing regimens by subgroup analysis. Moreover, the rates of treatment-related deaths in the irinotecan-containing arm were lower than that in the nonirinotecan-containing arm (0.7% vs. 2.6%, respectively). After that the second study reported by Wang et al.21 in 2010 also favored irinotecan-containing regimens without significantly improving OS rate; however, a significant benefit of TTF was observed in irinotecan-containing group. However, several limitations of these two meta-analyses needed to be considered. First, all RCTs included in both meta-analyses were conducted before 2008, and the numbers of RCTs included in both studies were small (three in Wagner's study and four in Wang's study). Recently, at least six new trials have been conducted to evaluate the effectiveness and toxicities of irinotecan-containing regimens as first-line treatment for AGC. Second, although most clinical trial data available emerged by including esophagogastric cancer either in gastric or esophageal cancer studies, more and more evidence demonstrated that esophagogastric cancer might represent a specific histopathological and biologic entity. For example, Wang et al.40 demonstrated that Barrett's metaplasia, a precancerous condition for esophagogastric tumorigenesis, was formed by residual embryonic cells, whereas intestinal metaplasia, a precancerous condition for gastric tumors, had been repeatedly reported to be induced by transdifferentiation of gastric type cells. This indicated that gastric and esophagogastric cancers were formed by different mechanisms. In addition, Reim et al.41 did a retrospective study of 551 patients and found that esophagogastric cancer was more likely to respond to preoperative chemotherapy than gastric cancer. However, authors of the previous meta-analyses did not exclude patients with esophagogastric caner in their analysis. Additionally, they also did not do subgroup analysis according to patient characteristic factors. Therefore, this led us to re-evaluate the role of irinotecan-based regimens in this setting.

This updated meta-analysis combined 1,837 patients from ten RCTs so that treatment effect could be evaluated with greater statistical power. With the present sample size, we had a power of 84.8% to reject the null hypothesis or a possibility of β-error of 15.2%. The final results demonstrated that irinotecan-containing regimens significantly improved OS compared to nonirinotecan-containing regimens, which were in accordance with previous two meta-analyses6, 21; however, the improvement in OS was not significant in previous studies. On the contrary to the study of Wang et al., we did not find a significant improvement of TTF in irinotecan-containing chemotherapy. Moreover, we also did pooled analysis for ORR, PFS and 1-year SR and found that irinotecan-containing regimens were superior to nonirinotecan-containing regimens; however, the improvement was not significant. Overall, our study demonstrated a survival benefit of irinotecan-containing over nonirinotecan-containing chemotherapy as first-line treatment for AGC.

Geographical origin was an important factor that affected the incidence and prognosis of AGC. In Eastern Asia, the incidence of gastric cancer was still high despite advancement in treatment and subsequent improvement in prognosis. On the contrary, in the West, where the incidence continuously decreased, the overall and stage-specific survival was worse than that in Eastern Asia. For example, Strong et al.42 reported that the disease-specific survival of gastric cancer was significantly better in Korea than in the United States (HR = 1.3, 95% CI = 1.0–1.6, p = 0.008). The reasons for the geographical differences in terms of incidence and prognosis were unclear; they were probably attributable to various factors in gastric carcinogenesis as well as in diagnostic and therapeutic strategies. As a result, we did subgroup analysis according to geographical origin and found that the estimated survival benefit for irinotecan-containing regimens was similar for those in Asia (HR = 0.86; 95% CI = 0.78–0.96) and out of Asia (HR = 0.82; 95% CI = 0.64–1.05).

With regard to the chemotherapy used, the combined regimens of irinotecan differed between included studies; however, all studies used irinotecan-based doublet therapy, and combined drugs included docetaxel,25 cisplatin,22, 39 oxaliplatin,28 S-123, 24, 27 or 5-FU.26, 37, 38 In addition, the controlled groups in trials were also different; three trials22–24 used monotherapy as controlled group, whereas the others25–28, 37–39 used combination therapy as controlled group. Therefore, we did subgroup analysis based on the controlled therapy and found that for each of these patient subsets, irinotecan-containing chemotherapy had slightly better efficacy than monotherapy (HR = 0.88, 95% CI = 0.77–1.00) or nonirinotecan-containing combination therapy (HR = 0.83, 95% CI = 0.71–0.96). The consistency of these findings suggested that benefits gained from irinotecan-containing regimens were robust. Based on the results of our meta-analysis, irinotecan-containing regimens seemed to be an appropriate alternative to FU and cisplatin, with significant survival benefits and a more favorable toxicity profile. However, irinotecan-containing regimens had never been compared against to a three-drug combination. For this reason, there was a reasonable doubt whether they truly represented therapeutic advances over an established three-drug regimen epirubicin and cisplatin/FU (ECF) or docetaxel and cisplatin/FU (DCF). As a result, more high-quality RCTs were warranted to confirm the efficacy and toxicities of irinotecan-containing regimens versus established three-drug regimens in AGC.

In this study, seven of these ten trials were newly included. Compared to the studies of Wagner et al.6 and Wang et al.,21 the number of eligible trials in this study was the most. However, this meta-analysis was still far from perfect. First, it was not an individual patient data analysis, and therefore, it precluded a more comprehensive analysis such as adjusting for baseline factors and other differences that existed between the trials from which the data were pooled. Furthermore, we could not discover the possible survival benefits of irinotecan-based regimen in different groups of patients with gastric cancer with different histologic types, detailed stages, ages, general conditions and so forth, because of inadequateness of corresponding data in these eligible trials. Although all these eligible trials used irinotecan-based chemotherapy as first-line treatment for AGC, the exact regimens among these trials were multitudinous. Thus, our study could not answer that which regimens would be the best choice. Finally, the difference in treatment schedules contributed to increase the clinical heterogeneity of the meta-analysis, which made the interpretation of a meta-analysis more problematic; however, clinical heterogeneity might improve the generalizability of the observed heterogeneity.

In summary, this updated meta-analysis of ten eligible RCTs provided strong evidence for a survival benefit of irinotecan-containing over nonirinotecan-containing chemotherapy as first-line treatment for AGC. A clear advantage of irinotecan-containing over nonirinotecan-containing chemotherapy had not been established. These results should help to inform decisions about patient management and design of future trials.