You are here

Third-line systemic treatment versus best supportive care for advanced/metastatic gastric cancer: A systematic review and meta-analysis

Critical Reviews in Oncology/Hematology, August 2017, Volume 116, Pages 68-81


  • This review looks exclusively into metastatic gastric patients who failed second-line treatment.

  • All participants used study medications in third-line setting.

  • Third-line systemic therapies showed superior efficacy in OS, PFS, ORR and DCR.

  • More toxicities (both all grades and high grade) occurred in patients on third-line therapies.

  • Future studies should focus more on the QOL-related outcomes in third-line setting.


This review evaluated the efficacy, toxicities and quality of life of third-line systemic treatment (TLT) versus best supportive care (BSC) in metastatic gastric cancer patients after failing two lines of systemic treatment.

Six studies were included, involving 890 participants (TLT: 587, BSC: 303, Asian: 679, 76.3%), median 53–61 years old, ECOG 0–1 with no major co-morbidities. Compared with BSC, TLT improved overall survival (HR 0.63; 95% CI 0.46–0.87, corresponding to an improvement in medial OS from 3.20 to 4.80 months), progression-free survival (HR 0.29; 95% CI 0.18–0.45), objective response rate (RR 5.28; 95% CI 1.00–27.83) and disease control rate (RR 4.51; 95% CI 2.64–7.71). The efficacy results favoring TLT should be interpreted with caution for the substantial heterogeneities, wide confidence intervals and selection bias. More toxicities occurred in the TLT arms.

This review highlighted the paucity of QOL data. Future studies should focus more on QOL-related outcomes. PROSPERO registration: 2015 CRD42015017873.

Keywords: Gastric cancer, Metastatic, Chemotherapy, Targeted therapy, Palliative care.

1 Introduction

Gastric cancer is the fifth most common cancer and remains the world's third leading cause of cancer mortality ( Torre et al., 2015 ). Despite the improvement in multi-modality management approach, the recurrence rate is still high. About 40–80% patients suffer from disease relapse ( Gallo and Cha, 2006; Gunderson, 2002 ). Moreover, more than half of the patients at diagnosis are already too advanced and not operable.

In the last decade, first-line then second-line palliative chemotherapy with or without targeted agent has become a standard treatment in patients with advanced/metastatic gastric cancer. A Cochrane review and meta-analysis performed by Wagner demonstrated a significant survival benefit in favor of chemotherapy compared with best supportive care (BSC) (Hazard ratio (HR) 0.37; 95% confidence interval (CI) 0.24–0.55, P < 0.0001) ( Wagner et al., 2010 ). This could be interpreted as an improvement in median survival from 4.3 months with BSC to 11 months with chemotherapy. Combination chemotherapy is superior to monotherapy (HR 0.83, 95% CI 0.74–0.93, P = 0.001). Large proportions of patients are either non-responders or have progression after first-line chemotherapy, and proceed for second-line treatment. Several systematic reviews and meta-analyses had confirmed that second-line chemotherapy improved survival compared with BSC ( Kim et al., 2013; Lacovelli et al., 2014; Liepa et al., 2015 ). According to Lacovelli's study, the risk of death was decreased by 18% (HR = 0.82; 95% CI 0.79–0.85; posterior probability HR ≥ 1: <0.00001) with active therapies and the effect was even greater in patients with good performance status (ECOG = 0) ( Lacovelli et al., 2014 ).

With the development of new chemotherapies or targeted agents which are seemingly more effective and have less toxicity, many patients can still maintain a good general condition after failing second-line therapies. According to previous studies, around 20–90% patients continued on active third-line or further lines treatment ( Kang et al., 2012; Hironaka et al., 2013; Wilke et al., 2014 ). This gives patients another chance to control the malignancy, maintain quality of life (QOL), relieve symptoms and even improve survival.

Taxanes and irinotecan are the possible options in third-line setting as most of the metastatic gastric cancer patients had used 5FU/platinum based chemotherapy. Previous studies on taxane-based chemotherapy or irinotecan-based chemotherapy (either monotherapy or combination with 5FU) showed an overall response rate of 10–25% with progression-free survival (PFS) 2.1–3.3 months and overall survival (OS) 5.6–10.9 months ( Kang et al., 2013; Kim et al., 2007; Lee et al., 2008, 2012, 2013; Moon et al., 2010; Park et al., 2005; Shimoyama et al., 2009; Tarazona et al., 2016; [8] [9] ). A phase III RCT from Korea also demonstrated a trend of better survival (HR 0.81, 95% CI 0.45–1.46) with third-line chemotherapy using single agent irinotecan or docetaxel ( Kang et al., 2012 ).

Recent phase I to III studies with agents targeting on different molecular pathways including Her-2/EGFR/VEGFR/cMET/ATM/FGFR/IGFR/mTOR/PD-L1 are on-going internationally and some were published. These targeted agents were studied as monotherapy or in combination with chemotherapy, as first-line, second-line or further lines of treatment. Their results are conflicting and worth a deeper look to avoid publication bias of the positive studies.

1.1 Why it is important to do this review

Metastatic gastric cancer patients usually have poor prognosis with short life expectancy. Various anti-cancer agents have been used for third-line and beyond, but the data are not conclusive. Since active treatment can cause toxicities, the benefits must be balanced with the side effects. There is a lack of international standard for patients after second-line treatment. Hence a systematic review to evaluate the efficacy and toxicity of third-line treatment including both chemotherapy and targeted agents is performed.

2 Materials and methods

The methodology of this systematic review has been registered in PROSPERO in March 2015 (registration: CRD42015017873).

2.1 Study criteria

Phase II and III prospective randomized controlled trials (RCTs) that compared third-line or further lines of systemic treatment with BSC/placebo.

Inclusion criteria:

  • -

    Studies involving only patients with pathologically confirmed adenocarcinoma of stomach.

  • -

    Studies with at least one third-line or further lines treatment group.

  • -

    Studies with best supportive care group/placebo.

Exclusion criteria:

  • Studies with only treatment groups but no BSC/placebo group.

  • Trials with radiotherapy or intraperitoneal chemotherapy were outside the scope of our research and were excluded.

2.2 Search methods

Studies were identified by searching electronic databases including the Cochrane Register of Controlled Trials (CENTRAL, Issue 7, 2016), MEDLINE Ovid (1996 to July week 2, 2016), EMBASE Ovid (1996 to 21 July, 2016) and CINAHL (1996 to July week 2, 2016). The search strings used are reported in Appendix 1. Handsearching for published abstracts from conference proceedings was performed: The American Society for Clinical Oncology 2000 to 2016, The European Council of Clinical Oncology 2000 to 2015 (published in the European Journal of Cancer), and The European Society for Medical Oncology 2000 to 2015 (published in the Annals of Oncology).

Ongoing trials (accessed 21 July 2016) were identified with the databases: National Cancer Institute,, ISRCTN registry, and World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) search portal.

To limit publication bias, all published or unpublished randomized controlled trials were included. No limit was applied for language.

2.3 Types of outcome measures

Outcomes were divided into three groups: efficacy, toxicity and QOL. Efficacy outcomes included overall survival (OS), progression-free survival (PFS), objective response rate (ORR) and disease control rate (DCR). OS is defined as the time from the date of random assignment to date of death as a result of any cause. ORR comprises both complete and partial response. DCR comprises complete response, partial response and stable disease. Toxicity data was classified according to World Health Organization (WHO) or National Cancer Institute Common Toxicity Criteria (NCI-CTC). QOL was recorded qualitatively.

2.4 Selection of studies

After completing the searches, the search results were merged by the reference management software package Endnote X7.5. Duplicate records of the same report were removed. Eligibility assessment including scanning the title and abstract of every record identified, reviewing the full text articles, conference abstracts and protocols was performed independently by two authors. Disagreements were resolved by discussion and by consultation with a third author. Reasons for exclusions were documented.

2.5 Data extraction and management

Two authors independently extracted information using a standardized data extraction form. Information extracted included: study general information (country, published/unpublished, language and year of publication), study design, follow-up, settings, participants, interventions, quality components, efficacy outcomes, side effects and quality of life.

When a trial was presented in abstract form, we sought further information as necessary (from the internet, contacting authors or checking for other available resources/publications). If data were missing in a published report, the primary author was contacted to obtain the missing numerical data. For studies with more than one publication, data were extracted from all the publications; however, the final or updated version of each trial was considered the primary reference.

2.6 Assessment of risk of bias in included studies

Risk of bias in included trials was assessed independently by two authors using the tool of The Cochrane Collaboration for assessing risk of bias ( Higgins and Green, 2011 ): selection, performance, detection, attrition and reporting bias. Reporting bias was assessed by comparing the lists of outcomes from study protocols/trial registries to those reported in the published paper were compared.

The results were summarized in both a risk of bias graph ( Fig. 1 A ) and a risk of bias summary ( Fig. 1 B). Results of meta-analyses were interpreted in light of the findings with respect to risk of bias.


Fig. 1
(A) Risk of bias graph: review authors’ judgments about each risk of bias item presented as percentages across all included studies. (B) Risk of bias summary: review authors’ judgments about each risk of bias item for each included study.

2.7 Data synthesis

Meta-analyses were performed on the pooled outcomes if adequate numbers of clinically similar studies were available. They were all calculated with Review Manager 5.3 using the random-effects model ( The Nordic Cochrane Centre, 2014 ). Time-to-event data (OS, PFS) were expressed as hazard ratios (HR) of TLT over BSC, with 95% CI. Dichotomous outcomes (ORR, DCR) were expressed as risk ratios (RR). HRs were combined with inverse-variance method and RRs with Mantel–Haenszel method. Outcomes that were not suitable for meta-analysis were described qualitatively.

Heterogeneity among the trials in each meta-analysis was measured by Chi-square test and I 2 statistic. Substantial heterogeneity was considered to exist when the I 2 value was greater than 50% or there was a low P value (<0.10) in the Chi-square test ( Higgins et al., 2003 ).

3 Results

The search of electronic databases provided a total 9402 citations. Additional 37 records were identified from conference proceedings and trial registries, making a total of 9439 search results. After adjusting for duplicates, excluding the irrelevant topics and designs, 26 records remained for examine in more detail. Two were excluded as there were no BSC arms and 8 were on-going studies. This left 5 RCTs (16 reports) to be included in the review. Fig. 2 shows the search process.


Fig. 2
Study flow diagram.

A total of five RCTs comparing TLT with BSC were included ( Kang et al., 2012; Li et al., 2013, 2016; Ohtsu et al., 2013; Pavlakis et al., 2015; [5] ). One of the included trials Li et al. (2013) contained three arms: two intervention arms and one control arm. For data analysis and clearer presentations below, it was split into Li (2013a) and Li (2013b) so making a total of six eligible studies. Both published full papers and study protocols were available in all six studies ( Kang et al., 2012; Li et al., 2013, 2016; Ohtsu et al., 2013; Pavlakis et al., 2015; [5] ).

Three studies were conducted in China ( Li et al. 2013 ), one in Korea ( Kang et al., 2012 ), one in Canada, New Zealand, Australia and Korea ( Pavlakis et al., 2015 ), and one in 23 countries around the world ( Ohtsu et al., 2013 ). All studies reported intention-to-treat (ITT) analysis, thus recomputation of outcomes was not required.

Of the six included studies, three studies ( Kang et al., 2012; Ohtsu et al., 2013; Pavlakis et al., 2015 ) included not only patients who failed second-line systemic treatment, but also those just failed first-line. For these three studies, only data from the patients who used study medications/BSC in third-line setting were extracted. The six studies included total 890 metastatic gastric cancer patients who failed two lines of chemotherapy. Of these, 587 were assigned to TLT and 303 to BSC. Median age was 53–61 years. 679 participants (76.2%) were Asian. Performance status was well balanced in all studies. All patients were either ECOG 0 or 1. All patients had previously failed at least two lines of chemotherapy with agents including pyrimidine derivatives, platinum and taxanes. Five studies compared targeted agents (apatinib, regorafenib or everolimus) with BSC, and one study compared third-line chemotherapy (irinotecan or docetaxel) with BSC. 461 participants received targeted agent and 176 received chemotherapy.

A summary of the included studies is presented in Table 1 . A detail description of each study is presented in the “Supplementary data: Appendix 1”. A “Summary of findings” table using the recommendations in the Cochrane Handbook for Systematic Reviews of Interventions is in Table 2 ( Higgins and Green, 2011 ). The pooled HRs/RRs and their 95% CI were in details in “Supplementary data: Appendix 2”.

Study Design (Phase) Carry out country Primary tumor site No. intervention No. control Total no. ECOG Mean/median age (yr) Treatment (intervention vs. control) Outcomes (only third-line) Outcomes (second + third line) Full article available
Kang et al. (2012) III Korea GC 33 21 54 0,1 I: 58
C: 61
Intervention: Single agent docetaxel 60 mg/m 2 on D1 every 3 weeks or irinotecan 150 mg/m 2 every 2 weeks
Control: Best supportive care
Overall survival Toxicities Yes
Li (2013a) II China GC/GEJ 46 24 70 0,1 I: 53
C: 54
Intervention: Oral apatinib 425 mg twice daily
Control: Oral placebo
Overall survival
Progression-free survival
Objective response rate
Disease control rate
Li (2013b) II China GC/GEJ 47 24 71 0,1 I: 55
C: 54
Intervention: Oral apatinib 850 mg once daily
Control: Oral placebo
Overall survival
Progression-free survival
Objective response rate
Disease control rate
Li et al. (2016) III China GC/GEJ 176 91 267 0,1 I: 55
C: 56
Intervention: Oral apatinib 850 mg once daily
Control: Oral placebo
Overall survival
Progression-free survival
Objective response rate
Disease control rate
Ohtsu et al. (2013) III 23 countries worldwide GC 229 114 343 0,1 I: 62
C: 62
Intervention: Oral everolimus 10 mg/day + best supportive care
Control: Oral placebo + best supportive care
Overall survival Progression-free survival
Objective response rate
Disease control rate
Pavlakis et al. (2015) II Australia
New Zealand
GC/GEJ 56 29 85 0,1 I: 60
C: 60
Intervention: Oral regorafenib 160 mg daily on days 1–21 each 28-day cycle + best supportive care
Control: Oral placebo + best supportive care
Progression free survival Overall survival
Objective response rate
Disease control rate
I: Intervention, C: Control, GC: Gastric cancer, GEJ: Gastroesophageal junction.

Table 1Summary of the included studies.


Table 2Summary of findings table.

3.1 Efficacy of third-line treatment

3.1.1 Overall survival (OS)

Five studies ( Kang et al., 2012; Li et al., 2013, 2016; Ohtsu et al., 2013; [4] ) provided overall survival data on patients receiving TLT vs. BSC. In Kang et al. (2012) , 33 patients received salvage chemotherapy and 21 patients received BSC. The HR was 0.821, 95% CI 0.450–1.464, suggesting no significant difference between salvage chemotherapy and BSC. In Li et al. (2013) , the mOS was significantly longer in patients given apatinib. The mOS of the placebo group was 2.50 months while those of apatinib 425 mg twice daily (46 patients) and 850 mg (47 patients) daily were 4.83 months (HR 0.41, 95% CI 0.24–0.72, p = 0.0017) and 4.83 months (HR 0.37; 95% CI 0.22–0.62, p < 0.001), respectively. Li et al. (2016) , a confirmatory phase III RCT on apatinib in patients failed two or further lines of chemotherapy, also demonstrated significant improvement in mOS in the apatinib group (176 patients) compared with placebo group (91 patients) (6.5 months vs. 4.7 months; HR 0.709, 95% CI 0.537–0.937; P = 0.0156). In Ohtsu et al. (2013) , everolimus (229 patients) did not improve the OS for advanced gastric cancer that progressed after two lines of previous systemic chemotherapy compared with BSC (114 patients) (HR 0.9, 95% CI 0.7–1.15).

Meta-analysis was performed for OS which involved total 805 participants. 531 of 805 participants received TLT and the remaining 274 received BSC. The resulting HR for OS was 0.63 (95% CI 0.46–0.87, p = 0.006) demonstrating a statistical significant benefit in OS in favor of TLT ( Fig. 3 ). This can be interpreted as an improvement in median OS from 3.20 months (weighted average in BSC) to 4.80 months (with TLT). There was substantial heterogeneity present in the data ( I 2 = 70%) and it was taken into account with the use of a random-effects model. There were 679 Asians in these 805 participants. The HR for medial survival in Asian populations was 0.63 (95% CI 0.45–0.90, P = 0.01) in favor of TLT ( Fig. 3 ). This can be interpreted as an improvement in median survival from 3.20 months (weighted average in BSC) to 4.83 months (with TLT).


Fig. 3
Efficacy outcome: A = overall survival, B = overall survival (Asian), C = overall survival (targeted agents), D = progression-free survival, E = objective response rate and F = disease control rate.

Subgroup analysis was performed on the targeted agents only trials. The main aim was to check if the heterogenity was due to mixing of studies using chemotherapy. The resulting HR for OS with targeted agents was 0.60 (95% CI 0.41–0.88, P = 0.004) ( Fig. 3 ). There was again substantial heterogeneity present in the data ( I 2 > 50%).

3.1.2 Progression-free survival

Four studies ( Li et al., 2013, 2016; Pavlakis et al., 2015; [3] ) reported the PFS data in gastric cancer patients after two lines of chemotherapy. Li et al. (2013) showed apatinib improved PFS compared with placebo. The PFS of placebo group was 1.5 months while that of apatinib 850 mg daily was 3.67 months (HR 0.18, 95% CI 0.10–0.34, p < 0.001) and that of apatinib 425 mg twice daily was 3.20 months (HR 0.21, 95% CI 0.11–0.38, p < 0.001). In Li et al. (2016) , similarly, apatinib significantly prolonged median PFS compared with placebo (2.6 months vs. 1.8 months, HR 0.444, 95% CI 0.331–0.595, p < 0.001). Pavlakis et al. (2015) also reported significantly longer PFS in the regorafenib group than the placebo group (HR 0.32, 95% CI 0.19–0.55, P < 0.001).

Data from these four studies including 493 participants were summarized in the PFS meta-analysis ( Fig. 3 ). The overall HR for PFS was 0.29; 95% CI 0.18–0.45, P < 0.00001, in favor of TLT. There was substantial heterogeneity present in the data ( I 2 = 69%).

3.1.3 Objective response rate (ORR) and disease control rate (DCR)

Data were available in total 408 participants in three eligible studies ( Li et al. 2013 ), with 269 participants in the TLT arms and 139 in the BSC arms. Both Li (2013a,b) and Li et al. (2016) compared apatinib to BSC. ORRs were between 2.84% ( Li et al., 2016 ) and 13.04% ( Li et al., 2013 ). The pooled RR was 5.28 (95% CI 1.00–27.83; P = 0.05), indicating a higher ORR with TLT when compared with BSC ( Fig. 3 ). DCRs were 34.78–51.06% with TLT and 8.79–10.42% with BSC. The pooled RR was 4.51 (95% CI 2.64–7.71, P < 0.00001) in favor of TLT ( Fig. 3 ). Heterogeneity was not detected in both analyses.

3.2 Toxicities

All six studies reported on the toxicity profile of both intervention arms and placebo arms. Three studies ( Kang et al., 2012; Ohtsu et al., 2013; Pavlakis et al., 2015 ) did not separately present the toxicity profile of patients in second-line or third-line settings, so it is not possible to combine all safety data into meta-analyses.

3.2.1 Death due to drug toxicity

Ohtsu et al. (2013) reported three patients in the everolimus arm died and their deaths were suspected to be a result of study treatment ( n = 1 each for sudden death, grade 3 pneumonitis, and grade 4 gastrointestinal hemorrhage). In the placebo arm, two patients died and their deaths were suspected to be a result of study treatment ( n = 1 each for multi-organ failure and cerebrovascular accident). The other five studies ( Kang et al., 2012; Li et al., 2013, 2016; Pavlakis et al., 2015; [4] ) did not report any data on death due to toxicities.

3.2.2 Specific side effects reported in each study

Reported side effects from each study were summarized in Table 3A “Table of side effects of all grades ≥10%” and Table 3B “Table of severe side effects Grade 3/4 ≥5%”. According to data of each trial, generally more participants in the intervention groups experienced treatment related toxicities (both all grades and high grade G3/4) than those in the BSC groups. Myelosuppression (neutropenia and anemia) and nausea were more common in chemotherapy recipients but less in targeted agents recipients. On the contrary, other unique side effects like hypertension, proteinuria, hand–foot syndrome, liver function derangement and electrolyte disturbances occurred more frequently in targeted agents.


Table 3ATable of side effects of all grades ≥10%.


Table 3BTable of severe side effects Grade 3/4 ≥ 5%.

Meta-analyses were performed on side effects commonly seen in three studies ( Li et al. 2013 ). The toxicities (all grades) with significant higher incidence in the treatment arms include neutropenia, thrombocytopenia, leukopenia, diarrhea, hypertension, proteinuria and hand–foot syndrome. For G3/4 toxicities, significant differences were seen in hypertension (RR: 6.25, 95% CI 1.20–32.57, p = 0.03) and proteinuria (RR: 4.65, 95% CI 1.09–19.87, p = 0.04) ( Fig. 4 ).


Fig. 4
Toxicities: G3/4: A = hypertension and B = hand–foot syndrome.

3.3 Quality of life

QOL was reported as secondary endpoint in five studies ( Li et al., 2013; Li et al., 2016; Ohtsu et al., 2013; Pavlakis et al., 2015 ). The QOL data could not be combined in a meta-analysis because only brief descriptions were reported in their final publications and the QOL data of Ohtsu et al. (2013) and Pavlakis et al. (2015) involved both second-line and third-line participants.

3.3.1 Methods of measurement of QOL

All five studies ( Li et al., 2013, 2016; Ohtsu et al., 2013; Pavlakis et al., 2015; [4] ) measured QOL using European Organization for Research and Treatment of Cancer quality of Life Questionnaire C30 (QLQ-C30) ( Scott et al., 2008 ). The QLQ-C30 questionnaire reports five functional scales (physical, role, emotional, cognitive and social functioning), three symptom scales (fatigue, nausea/vomiting, pain), a global health and overall QOL scale and five single item scales (dyspnea, appetite loss, sleep disturbance, constipation and diarrhea, and the financial impact of disease and treatment).

Both Li et al. (2013 2016) did not show any significant differences between TLT and BCS with regard to the scores of the QLQ-C30. Li et al. (2013) reported insomnia was the only significant change in QOL score over the course of treatment. Ohtsu et al. (2013) reported everolimus recipients had higher mean scores for the global health status/QOL scale of the QLQ-C30 questionnaire over time. However, significant difference in mean score of QOL was only seen at week 36, which was at a time much later than the median survival (mOS of everolimus: 23.5 weeks and placebo: 18.4 weeks) and only total 30 participants left. In Pavlakis et al. (2015) , QLQ-C30 Global Health Subscale mean scores for regorafenib vs. placebo were 53 (95% CI 48–58) vs. 58 (95% CI 51–65) at week 4 and 54 (95% CI 48–60) vs. 56 (95% CI 45–67) at week 8, respectively. The full report of QOL of Pavlakis et al. (2015) will be published later.

Ohtsu et al. (2013) also measured QOL by gastric-specific STO22 module ( Blazeby et al., 2004 ). Over time, everolimus recipients had lower mean scores (i.e. lower level of symptoms) for the pain and reflux subscales of the STO22 module.

3.3.2 Time to deterioration in performance status/QOL

Ohtsu et al. (2013) reported on the time of deterioration in performance status/QOL. The median time to definitive deterioration in ECOG PS by >1 category did not differ between treatment groups (2.30 months with everolimus vs. 2.23 months with placebo; HR 0.96; 95% CI 0.76–1.20; P = 0.6925). There was no significant difference in the time to definitive 5% deterioration in global QOL from baseline (1.51 months with everolimus vs. 1.45 months with placebo; HR 0.84; 95% CI 0.69–1.03; P = 0.0936). The times to definitive 5% deterioration in the social (HR 0.87; 95% CI 0.70–1.09; P = 0.2108), emotional functioning (HR 0.83; 95% CI 0.67–1.02; P = 0.0735) and physical functioning (HR 0.95; 95% CI 0.78–1.15; P = 0.5711) were not statistically different.

3.3.3 Questionnaire completion rate

Li et al. (2016) reported on the questionnaire completion rate. The rates of compliance for responding to the QOL questionnaire at baseline and end of cycles two and three were 100%, 60.8%, and 34.7% for the apatinib group and 100%, 47.3% and 7.7% for the placebo group, respectively.

4 Discussion

The median OS was statistically significant in favor of systemic therapy (HR 0.63; 95% CI 0.46–0.87, P = 0.006), corresponding to 4.80 vs. 3.20 months. Asian population also had similar finding (HR: 0.63; 95% CI 0.45–0.90, P = 0.01). HR for PFS was superior with use of TLT (HR 0.29; 95% CI 0.18–0.45, P < 0.00001). The results should be interpreted with cautions for their substantial heterogeneities. Possible sources of the heterogeneity among studies included geographical participations (single country [ n = 4], multinational [ n = 2]), ethnicity of participants (Asian [ n = 679], non-Asian [ n = 126]), primary tumor site (GC only [ n = 2], GC/GEJ [ n = 4]), and use of anti-cancer agents (chemotherapy [ n = 1], targeted agent [ n = 5]).

Participants in the included trials were only in part representative of all gastric cancer patients. They were generally younger compared with the overall population of gastric cancer patients (peak incidence age: 69). They all had good performance status, ECOG 0–1. Patients with co-morbidities, such as uncontrolled hypertension (blood pressure >140/90 mmHg), history of venous thromboembolic events, impaired hepatic or renal function were excluded from the studies. Also it is unknown if the underrepresented group, such as elderlies or patients with marginal performance status ECOG ≥2, can gain equal benefits as reported in the trials. As more than 75% of the participants in these trials were Asian, both efficacy and tolerability of TLT in non-Asian populations need further evaluation. Therefore it is considered as unclear whether our analyses results can be generalized to all gastric cancer patients who failed second-line systemic treatment.

In our review, the agents that showed statistical significant benefits in third-line settings were all targeted therapies. They can be easily administered at home orally and avoid hospital visits. Targeted agents are often perceived to have fewer side effects and more tolerable than chemotherapies. Undoubtedly, the frequencies of myelosuppression, nausea and vomiting are less often in targeted agents. They are nevertheless associated with other unique adverse effects, such as hypertension, proteinuria, hand–foot syndrome, and lethal complications like pneumonitis and gastrointestinal hemorrhage.

Previous lines of treatment may affect the efficacy of TLT. All the included participants had previously exposed to chemotherapy alone with agents including fluoropyrimidine, platinum and taxanes. With the promising results of targeted agents like trastuzumab and ramucirumab, the combination of targeted agents and chemotherapy are used in first and second-line settings. The integration of these anti-cancer agents into a treatment continuum for initial treatment, and treatment after initial progression may help patients better tolerate and even attain more benefits when going on TLT.

The decision on TLT in advanced gastric cancer patients is complex in the light of the above reasons. The use of TLT can be considered in patients who have good performance status, minimal co-morbidities, longer expected survival (≥3 months) and are willing to risk additional side effects or toxicities. It is essential for the physicians to explore whether this small gain in short term survival (from 3.20 to 4.80 months, absolute gain 1.60 months) and response at expense of higher incidence of treatment toxicities meets patients’ goals and expectations before starting treatment. An open and honest communication on the prognosis, more specifically on the life expectancy, should be discussed with the patients and carers. Patients should be informed about the goal of treatment, efficacy, response, side effects and potential life-threatening complications of TLT, as well as the alternative with BSC which involves symptoms control, daily care, social support, financial planning and spiritual needs.

Researches on gastric carcinogenesis and molecular subtypes may help in better understanding this heterogeneous cancer, shedding some light on the development of more successful targeted agents. Several cellular signaling pathways and their targeted agents are now currently under investigations, e.g. HGF/c-MET pathway, PI3K/AKT/mTOR pathway, immune checkpoints. Rilotumumab ( Cunningham et al., 2015; Doi et al., 2015 ), an anti-HGF monoclonal antibody, and onartuzumab ( Shah et al., 2015 ), an anti-c-MET monoclonal antibody, have been studied as first-line agent for advanced gastric cancer in phase III studies. Regorafenib, after demonstrated its superior efficacy in PFS in Phase II study, has been tested in Phase III INTERGRATE-II study on gastric cancer patients in third-line setting ( NCT, 2016a ). Studies on immune checkpoint inhibitors are now growing rapidly in recent few years. In the KEYNOTE-012 study, pembrolizumab, a PD-1 inhibitor, showed durable efficacy and manageable safety profile for the heavily pre-treated, PD-L1 positive gastric cancer patients (OS 11.4 months, PFS 1.9 months, ORR 22.2%) ( Bang et al., 2015 ). Studies on pembrolizumab are now in progress, including KEYNOTE-062 comparing pembrolizumab as first-line monotherapy and in combination with cisplatin/5FU versus placebo, and KEYNOTE-061 comparing pembrolizumab with paclitaxel as second-line agent ( NCT 2015a ). Avelumab, a human IgG1 anti-PD-L1 antibody, has also been tested in phase III RCTs as first-line and third-line settings for gastric cancers (JAVELIN Gastric 100 and JAVELIN Gastric 300) ( NCT 2015b ). At the time when revising this manuscript, nivolumab has reported its superior efficacy compared with placebo in advanced gastric cancer patients as salvage treatment (OS: 5.32 months vs. 4.14 months, p < 0.0001; 12-month OS: 26.6% vs. 10.9%) in 2017 ASCO GI Cancer Symposium ( Kang et al., 2017 ). Results of these studies are encouraging. Probably in the future, the combination of cytotoxic chemotherapy, targeted agents and immunotherapy may be the management approach to maximize benefits in both survival and quality of life.

This systematic review highlighted the scarcity of QOL data. QOL assessments were included in the clinical studies only as one of the secondary endpoints to complement the efficacy results of the primary endpoint. Hence, the included trials were not powered to detect significant differences in QOL. The response rates of the QOL questionnaires were low in the included studies. Data from non-responders may provide important information for adequate assessment of QOL. Overcoming the challenge of non-response is essential in adequate assessment of QOL in coming studies.

With poor long term survival, future researches should compare QOL in patients going on TLT verses BSC. QOL-related outcomes, like “quality adjusted survival” or “time of preservation of functional capacity or independence”, might be used as primary outcome measures in future studies.

Several limitations of this meta-analysis should be noted. First, all the included studies were industry funded and prone to reporting bias. Second, this systematic review was conducted using aggregated data from studies instead of individual patient data (IPD). IPD was not sought from the study authors. Although ideally it should be conducted using IPD, this is rarely done and considered not practical as IPD are usually not easily obtainable. Third, all studies risked the bias in the control group for the lack of standardization of the delivery of BSC. The method of BSC was not described in details in the reports or protocols. Fourth, the cost effectiveness on the use of TLT compared with BSC in advanced gastric cancer is worth for evaluation. However, this analysis is out of our scope in this review.

5 Conclusions

The results suggested a statistically significant but modest clinical benefit in favor of TLT, in the cost of more toxicities. QOL data after failure of second-line therapy is scarce and incomplete. Truthful communication with patient and carers is of utmost importance before starting TLT. Future researches should focus on the combination and sequencing of different anti-cancer agents for advanced gastric cancer to maximize the efficacy and minimize the toxicities.

Conflicts of interest

The authors have declared no conflicts of interest.


This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.


The authors would like to acknowledge Prof. Jung Hung Kang, Prof. Atsushi Ohtsu, Prof. Nick Pavlakis for providing further information on their researches; Ms. Delyth Morris from the University of Cardiff in the searching strategies of this review.

Appendix A Supplementary data

The following are the supplementary data to this article: Attached file


  • Bang et al., 2015 Y.J. Bang ,H.C. Chung ,V. Shankaran ,R. Geva ,D.V.T. Catenacci ,S. Gupta ,J.P. Eder ,R. Berger ,E.J. Gonzalez ,A. Ray ,et al. Relationship between PD-L1 expression and clinical outcomes in patients with advanced gastric cancer treated with the anti-PD-1 monoclonal antibody pembrolizumab (MK-3475) in KEYNOTE-012. J. Clin. Oncol.. 2015;33(Suppl. 15):4001
  • Blazeby et al., 2004 J.M. Blazeby ,T. Conroy ,A. Bottomley ,C. Vickery ,J. Arraras ,O. Sezer ,et al. European Organisation for Research and Treatment of Cancer Gastrointestinal and Quality of Life Groups. Clinical and psychometric validation of a questionnaire module, the EORTC QLQ-STO 22, to assess quality of life in patients with gastric cancer. Eur. J. Cancer. 2004;40(October (15)):2260-2268 Crossref
  • Cunningham et al., 2015 D. Cunningham ,N.C. Tebbutt ,I. Davidenko ,A.M. Murad ,S.-E. Al-Batran ,D.H. Ilson ,et al. Phase III, randomized, double-blind, multicenter, placebo (P)-controlled trial of rilotumumab (R) plus epirubicin, cisplatin and capecitabine (ECX) as first-line therapy in patients (pts) with advanced MET-positive (pos) gastric or gastroesophageal junction (G/GEJ) cancer: RILOMET-1 study. J. Clin. Oncol.. 2015;33(Suppl. 15):4000
  • Doi et al., 2015 T. Doi ,Y.K. Kang ,K. Muro ,Y. Jiang ,R.K. Jain ,R. Lizambri. A phase 3, multicenter, randomized, double-blind, placebo-controlled study of rilotumumab in combination with cisplatin and capecitabine (CX) as first-line therapy for Asian patients (pts) with advanced MET-positive gastric or gastroesophageal junction (G/GEJ) adenocarcinoma: the RILOMET-2 trial. J. Clin. Oncol.. 2015;33(Suppl. 15):TPS226
  • Gallo and Cha, 2006 A. Gallo ,C. Cha. Updates on esophageal and gastric cancers. World J. Gastroenterol.. 2006;(12):3237-3242 Crossref
  • Gunderson, 2002 L.L. Gunderson. Gastric cancer – patterns of relapse after surgical resection. Semin. Radiat. Oncol.. 2002;12 :150-161 Crossref
  • Higgins and Green, 2011 Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 [Updated March 2011]. J.P.T. Higgins, S. Green. (The Cochrane Collaboration, 2011) Available from
  • Higgins et al., 2003 J.P. Higgins ,S.G. Thompson ,J.J. Deeks ,D.G. Altman. Measuring inconsistency in meta-analyses. BMJ. 2003;327(7414):557-560 Crossref
  • Hironaka et al., 2013 S. Hironaka ,S. Ueda ,H. Yasui ,T. Nishina ,M. Tsuda ,T. Tsumura ,et al. Randomized, open-label, phase III study comparing irinotecan with paclitaxel in patients with advanced gastric cancer without severe peritoneal metastasis after failure of prior combination chemotherapy using fluoropyrimidine plus platinum: WJOG trial. J. Clin. Oncol.. 2013;31(December (35)):4438-4444 10.1200/JCO.2012.48.5805 Epub 2013 November 4 Crossref
  • Kang et al., 2012 J.H. Kang ,S.I. Lee ,H. Lim do ,K.W. Park ,S.Y. Oh ,H.C. Kwon ,et al. Salvage chemotherapy for pretreated gastric cancer: a randomized phase III trial comparing chemotherapy plus best supportive care with best supportive care alone. Clin. Oncol.. 2012;30(May (13)):1513-1518 10.1200/JCO.2011.39.4585 Epub 2012 March 12 Crossref
  • Kang et al., 2013 E.J. Kang ,S.A. Im ,D.Y. Oh ,S.W. Han ,J.S. Kim ,I.S. Choi ,et al. Irinotecan combined with 5-fluorouracil and leucovorin third-line chemotherapy after failure of fluoropyrimidine, platinum, and taxane in gastric cancer: treatment outcomes and a prognostic model to predict survival. Gastric Cancer. 2013;16(October (4)):581-589 10.1007/s10120-012-0227-5 Epub 2012 December 25 Crossref
  • Kang et al., 2017 Y.K. Kang ,T. Satoh ,M.H. Ryu ,et al. Nivolumab (ONO-4538/BMS-936558) as salvage treatment after second or later-line chemotherapy for advanced gastric or gastro-esophageal junction cancer (AGC): a double-blinded, randomized, phase III trial. J. Clin. Oncol.. 2017;35(Suppl):4S Abstract 2
  • Kim et al., 2007 S.G. Kim ,S.Y. Oh ,H.C. Kwon ,S. Lee ,J.H. Kim ,S.H. Kim ,et al. A phase II study of irinotecan with bi-weekly, low-dose leucovorin and bolus and continuous infusion 5-fluorouracil (modified FOLFIRI) as salvage therapy for patients with advanced or metastatic gastric cancer. Jpn. J. Clin. Oncol.. 2007;37(October (10)):744-749 Epub 2007 October 8 Crossref
  • Kim et al., 2013 H.S. Kim ,H.J. Kim ,S.Y. Kim ,T.Y. Kim ,K.W. Lee ,S.K. Baek ,et al. Second-line chemotherapy versus supportive cancer treatment in advanced gastric cancer: a meta-analysis. Ann. Oncol.. 2013;24(November (11)):2850-2854 10.1093/annonc/mdt351 Epub 2013 August 13 Crossref
  • Lacovelli et al., 2014 R. Lacovelli ,F. Pietrantonio ,A. Farcomeni ,C. Maggi ,A. Palazzo ,F. Ricchini ,et al. Chemotherapy or targeted therapy as second-line treatment of advanced gastric cancer. A systematic review and meta-analysis of published studies. PLOS ONE. 2014;9(September (9)):e108940 10.1371/journal.pone.0108940 eCollection 2014
  • Lee et al., 2008 J.L. Lee ,M.H. Ryu ,H.M. Chang ,T.W. Kim ,J.H. Yook ,S.T. Oh ,et al. A phase II study of docetaxel as salvage chemotherapy in advanced gastric cancer after failure of fluoropyrimidine and platinum combination chemotherapy. Cancer Chemother. Pharmacol.. 2008;61(April (4)):631-637 Epub 2007 May 23 Crossref
  • Lee et al., 2012 M.J. Lee ,I.G. Hwang ,J.S. Jang ,J.H. Choi ,B.B. Park ,M.H. Chang ,et al. Outcomes of third-line docetaxel-based chemotherapy in advanced gastric cancer who failed previous oxaliplatin-based and irinotecan-based chemotherapies. Cancer Res. Treat.. 2012;44(December (4)):235-241
  • Lee et al., 2013 J.H. Lee ,S.H. Kim ,S.Y. Oh ,S. Lee ,H. Lee ,H.J. Lee ,et al. Third-line docetaxel chemotherapy for recurrent and metastatic gastric cancer. Korean J. Intern. Med.. 2013;28(May (3)):314-321 10.3904/kjim.2013.28.3.314 Epub 2013 May 1 Crossref
  • Li et al., 2013 J. Li ,S. Qin ,J. Xu ,W. Guo ,J. Xiong ,Y. Bai ,et al. Apatinib for chemotherapy-refractory advanced metastatic gastric cancer: results from a randomized, placebo-controlled, parallel-arm, phase II trial. J. Clin. Oncol.. 2013;31(26):3219-3225 Crossref
  • Li et al., 2016 J. Li ,S. Qin ,J. Xu ,J. Xiong ,C. Wu ,Y. Bai ,et al. Randomized, double-blind, placebo-controlled phase III trial of apatinib in patients with chemotherapy-refractory advanced or metastatic adenocarcinoma of the stomach or gastroesophageal junction. J. Clin. Oncol.. 2016;34(May (13)):1448-1454 10.1200/JCO.2015.63.5995 Epub 2016 February 16 Crossref
  • Liepa et al., 2015 A. Liepa ,S. Mitchell ,S. Batson ,M.H. Jen ,A. Davie ,K. Taipale ,L. Hess. Systematic review and meta-analysis of recommended second-line therapies for advanced gastric cancer (GC). Eur. J. Cancer. 2015; Conference: European Cancer Congress 2015, ECC 2015 Vienna Austria. Conference Start: 20150925 Conference End: 20150929. Conference Publication: (var.pagings). 51 (pp S437), 2015. Date of Publication: September 2015
  • Moon et al., 2010 Y.W. Moon ,S.Y. Rha ,H.C. Jeung ,C. Kim ,M.H. Hong ,H. Chang ,et al. Outcomes of multiple salvage chemotherapy for advanced gastric cancer: implications for clinical practice and trial design. Cancer Chemother. Pharmacol.. 2010;66(September (4)):797-805 10.1007/s00280-010-1295-z Epub 2010 March 11 Crossref
  • NCT, 2015a NCT02494583: Study of. (, 2015) Available from:
  • NCT, 2015b NCT02625610: Avelumab in. (, 2015) Available from:
  • NCT, 2015c NCT02625623: Avelumab in. (, 2015) Available from:
  • NCT, 2016a NCT02773524: A. (, 2016) Available from:
  • NCT, 2016b NCT02370498: A. (, 2016) Available from:
  • Ohtsu et al., 2013 A. Ohtsu ,J.A. Ajani ,Y.X. Bai ,Y.J. Bang ,H.C. Chung ,H.M. Pan ,et al. Everolimus for previously treated advanced gastric cancer: results of the randomized double-blind phase III GRANITE-1 study. J. Clin. Oncol.. 2013;31(November (31)):3935-3943 10.1200/JCO.2012.48.3552 Epub 2013 September 16 Crossref
  • Park et al., 2005 S.H. Park ,E.Y. Choi ,S.M. Bang ,E.K. Cho ,J.H. Lee ,D.B. Shin ,et al. Salvage chemotherapy with irinotecan and cisplatin in patients with metastatic gastric cancer failing both 5-fluorouracil and taxanes. Anticancer Drugs. 2005;16(July (6)):621-625 Crossref
  • Pavlakis et al., 2015 N. Pavlakis ,K.M. Sjoquist ,E. Tsobanis ,A.J. Martin ,Y.-K. Kang ,J. Bang ,et al. INTEGRATE: a randomized, phase II, double-blind, placebo-controlled study of regorafenib in refractory advanced oesophagogastric cancer (AOGC): a study by the Australasian Gastrointestinal Trials Group (AGITG)-final overall and subgroup results. J. Clin. Oncol.. 2015; Conference: 2015 Annual Meeting of the American Society of Clinical Oncology, ASCO Chicago, IL United States. Conference Start: 20150529 Conference End: 20150602. Conference Publication: (var.pagings). 33 (15 SUPPL. 1) (no pagination), 2015. Date of Publication: 20 May 2015
  • Scott et al., 2008 N.W. Scott ,P.M. Fayers ,N.K. Aaronson ,A. Bottomley ,A. de Graeff ,M. Groenvold ,et al. EORTC Quality of Life Group. EORTC QLQ-C30 Reference Values. (EORTC Quality of Life Group Publications, Brussels, 2008) Available from
  • Shah et al., 2015 M.A. Shah ,Y.J. Bang ,F. Lordick ,J. Tabernero ,M. Chen ,S.P. Hack ,S.C. Phan ,D.S. Shames ,D. Cunningham. METGastric: a phase III study of onartuzumab plus mFOLFOX6 in patients with metastatic HER2-negative (HER2−) and MET-positive (MET+) adenocarcinoma of the stomach or gastroesophageal junction (GEC). J. Clin. Oncol.. 2015;33(Suppl. 15):4012
  • Shimoyama et al., 2009 R. Shimoyama ,H. Yasui ,N. Boku ,Y. Onozawa ,S. Hironaka ,A. Fukutomi ,et al. Weekly paclitaxel for heavily treated advanced or recurrent gastric cancer refractory to fluorouracil, irinotecan, and cisplatin. Gastric Cancer. 2009;12(4):206-211 10.1007/s10120-009-0524-9 Epub 2010 January 5 Crossref
  • Tarazona et al., 2016 N. Tarazona ,E.C. Smyth ,C. Peckit ,I. Chau ,D. Watkins ,S. Rao ,et al. Efficacy and toxicity of salvage weekly paclitaxel chemotherapy in non-Asian patients with advanced oesophagogastric adenocarcinoma. Ther. Adv. Med. Oncol.. 2016;8(March (2)):104-112 10.1177/1758834015621669 Crossref
  • The Nordic Cochrane Centre, 2014 The Nordic Cochrane Centre, The Cochrane Collaboration. Review Manager (RevMan). 5.3. (The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, 2014)
  • Torre et al., 2015 L.A. Torre ,F. Bray ,R.L. Siegel ,J. Ferlay ,J. Lortet-Tieulent ,A. Jemal. Global cancer statistics, 2012. CA Cancer J. Clin.. 2015;65(March (2)):87-108 10.3322/caac.21262 Epub 2015 February 4 Crossref
  • Wagner et al., 2010 A.D. Wagner ,S. Unverzagt ,W. Grothe ,G. Kleber ,A. Grothey ,J. Haerting ,et al. Chemotherapy for advanced gastric cancer. Cochrane Database Syst. Rev.. 2010;17(March (3)):CD004064 10.1002/14651858.CD004064.pub3
  • Wilke et al., 2014 H. Wilke ,K. Muro ,E. Van Cutsem ,S.C. Oh ,G. Bodoky ,Y. Shimada ,et al. Ramucirumab plus paclitaxel versus placebo plus paclitaxel in patients with previously treated advanced gastric or gastro-oesophageal junction adenocarcinoma (RAINBOW): a double-blind, randomised phase 3 trial. Lancet Oncol.. 2014;15(October (11)):1224-1235 10.1016/S1470-2045(14)70420-6 Epub 2014 September 17 Crossref