A systematic review and meta-analysis of randomized trials on the role of targeted therapy in the management of advanced gastric cancer: Evidence does not translate?

Summary

It is still uncertain if targeted therapy-based regimens in advanced gastric cancer actually produce survival benefit. To shed light on this important question, we performed a systematic review and meta-analyses on each relevant targeted-pathway. By searching literature databases and proceedings of major cancer meetings in the time-frame 2005–2014, 22 randomized clinical trials exploring targeted therapy for a total of 7022 advanced gastric cancer patients were selected and included in the final analysis. Benefit was demonstrated for antiangiogenic agents in terms of overall survival (HR 0.759; 95%CI 0.655–0.880; p < 0.001). Conversely no benefit was found for EGFR pathway (HR 1.077; 95%CI 0.847–1.370; p = 0.543). Meta-analysis of HER-2 pathway confirmed improvement in terms of survival outcome, already known for this class of drugs (HR 0.823; 95%CI 0.722–0.939; p = 0.004). Pooled analysis demonstrated a significant survival benefit (OS: HR 0.823; PFS: HR 0.762) with acceptable tolerability profile for targeted-based therapies as compared to conventional treatments. This finding conflicts with the outcome of most individual studies, probably due to poor trial design or patients selection. In conclusion, our findings demonstrate a significant survival benefit for targeted therapy in its whole, which can be ascribed to anti-angiogenic and anti-HER2 agents.

Keywords:

• angiogenesis
• gastric cancer
• meta-analysis
• randomized clinical trials
• systemic chemotherapy
• targeted pathways
• targeted therapy

Abbreviations

Ab	=	monoclonal antibody
ADME	=	absorption, distribution, metabolism, and excretion
aGC	=	advanced gastric cancer
BSC	=	best supportive care
CHT	=	chemotherapy
EGFR	=	epidermal growth factor receptor
GC	=	gastric cancer
HER2	=	human epidermal growth factor receptor 2
HER3	=	human epidermal growth factor receptor 3
MET	=	mesenchymal epithelial transition factor
mTOR	=	mammalian target of rapamycin
mTORC	=	mTOR complex
NGS	=	next generation sequencing
NSCLC	=	non-small cell lung cancer
OR	=	odds-ratio
OS	=	overall survival
PARP	=	poly ADP ribose polymerase
PFS	=	progression free survival
PI3K	=	phosphatidylinositide 3-kinases
PRISMA	=	preferred reporting items for systematic reviews and meta-analyses
RAF	=	rapidly accelerated fibrosarcoma
RAS	=	rat sarcoma viral oncogene homolog
RCTs	=	randomized clinical trials
RR	=	response rate
TKI	=	tyrosine kinase inhibitor
VEGF	=	vascular endothelial growth factor
VEGFR	=	VEGF receptor

Background

Description of epidemiology and clinical management

Gastric cancer (GC) is an aggressive malignancy and prognosis for advanced disease (aGC) is poor.^1,2 Due to its heterogeneity, GC usually include different subgroups based on histological, anatomical, epidemiological and more recently, genomic, or molecular classifications.^3,4 Adenocarcinoma is the most frequent histotype, representing 90% of GCs. Histologically, an intestinal type (well-differentiated) with better prognosis and response to treatment, and a diffuse type (or undifferentiated) more frequent in young patients and characterized by worse outcome are recognized.⁵ As a further difference, intestinal GC seems to result from a multistep carcinogenesis process, while diffuse GC is characterized by de novo lesions frequently associated to E-cadherin loss.^6-8

Conventional treatment

Regardless of pathological subtype, the treatment of GC is based on a multidisciplinary approach that includes surgery, radiotherapy and chemotherapy.⁹ Usually, aGC patients receive diagnosis when the tumor is not suitable for radical surgery.¹⁰ At this stage, the mainstay of treatment is palliative systemic chemotherapy.^10,11 So far, several small randomized clinical trials (RCTs) provided evidence that systemic chemotherapy improves survival in aGC patients, but at the cost of relevant toxicity.^12,13 Whether fluoropyrimidines are the most common agents in this disease as single agent or in combination schedules, several other drugs showed efficacy as single agent such as docetaxel, irinotecan, anthracyclines, and platinum compounds (oxaliplatin and cisplatin).^14,15 Currently, combination chemotherapy schedules including fluoropyrimidines and platinum salts, with possible addition of docetaxel in fit patients, represent the landmark of first-line treatment for aGC with potential benefits in terms of quality of life and overall survival (OS).^10,16-18 Concerning the second-line treatment, available data on the safety and efficacy of therapy are limited.¹⁹ In particular COUGAR-02, a phase III randomized clinical trial, reported a significant OS improvement for docetaxel versus best supportive care (BSC) with a median advantage of 1.5 months.²⁰ Regarding subsequent lines of treatment, BSC or recruitment in clinical trials (fit patients only) is considered as the best choice.^12,21

Molecular pathways and targeted therapy

Several pathways appear to act as “drivers” in different aGC subtypes. In particular non-diffuse cancers seem to depend on different alterations in epidermal growth factor or other peptide growth factor signaling (HER2, EGFR, MET) or in angiogenesis-related signaling, while in diffuse cancers beta-catenin, PI3K/Akt/mTOR pathway and HER3 activity play a predominant role.^22-24

Recently, RCTs investigated the efficacy of the targeted therapy alone or in combination with chemotherapy, but results were mostly unsatisfactory.^25-38 While several RCTs demonstrated an improvement in terms of response rate (RR), and progression free survival (PFS) only one study reported a significant increase in terms of OS in a selected subgroup of patients in front-line treatment.²⁵ In that trial, patients were selected according to HER2 status (resulted to be overexpressed in 16–34% of patients with intestinal type and 2–7% of diffuse aGC), and subsequently treated with trastuzumab plus standard chemotherapy with a significant 2.7 months advantage in OS. To date, the addition of trastuzumab to conventional chemotherapy represents the best treatment choice for aGC overexpressing HER2.²⁵ Serum VEGF concentration, EGFR overexpression and PI3K/Akt/mTOR pathway alterations have been shown to be related to vascular involvement, metastases and poor outcome, thus representing potential targets in this disease.²³ Indeed, different antiangiogenic agents showed interesting activity in terms of response rate. Furthermore, as in many other cancers, it has been demonstrated the reliance of GC on angiogenesis, with the arrest of tumor growth in the absence of neovascularization.³⁹ In particular, 3 phase II studies that investigated the effect of bevacizumab-based therapy showed an encouraging RR (65–68%), subsequently confirmed in a phase III trial in the absence, however, of significant benefit in OS.^39-41 Recently, a meta-analysis confirmed the benefit of anti-VEGF target therapy in aGC on all endpoints evaluated (OS, PFS, RR).⁴² Despite EGFR overexpression is observed in 27–44% of all GC, different trials evaluating the role of anti-EGFR agents failed to demonstrate any improvement in either PFS, OS, or RR.^30,43 The role of targeted therapy in aGC remains therefore mostly undefined.

On this basis, we performed a systematic review to analyze the weight of each targeted pathway in aGC management through one by one meta-analysis.

Results

Studies selection

In Figure 1, the PRISMA chart related to RCTs selection and search strategy is shown. In the time-frame covered by the present systematic review (2005–2014), 7831 studies were reported as full papers or meeting abstracts, while 6689 studies were initially excluded because reviews and 962 were excluded for trial design. Subsequently, we examined in detail the remaining 180 trials. Among them, 158 were excluded because selection criteria were not met. Further, one study was excluded because reported data about a major trial previously examined and already included.^28,44 One trial was excluded due to missing retrievable data, as already reported.⁴⁵ Six studies couldn't be evaluated because still ongoing.^46-50 Twenty-two trials for a total of 7022 patients were selected and included in the final analysis.^{25-37,44,51-59} The TYTAN trial missed data about PFS. One trial missed results about OS. Two trials were analyzed only for RR and toxicity for missing data on survival endpoints.^33,51 Moreover, 3 trials, both designed for multiple arms comparison, were analyzed for single comparison considering an aggregate arm of different drug concentrations.^33,34,59 One trial was evaluated only for survival endpoints, because missing RR and toxicity data, as reported, an abstract in the meeting ASCO 2011.²⁷ At least one data-comparison in terms of survival, RR, or toxicity was reported in all selected RCTs, which were therefore deemed eligible for the end-point analysis. Summarizing the 22 trials included in final analyses: 19 were eligible for OS analysis (among them, we underlined, that: 9 were included in anti-angiogenic analysis; 4 studies were included in anti-EGFR analysis; 3 studies were included in anti-HER2 analysis; single trials respectively investigating a MET inhibitor, mTOR inhibitor and a PARP inhibitor), 19 were eligible for PFS analysis (among them, we underlined, that: 10 were included in anti-angiogenic analysis; 4 studies were included in anti-EGFR analysis; 2 studies were included in anti-HER2 analysis; single trials respectively investigating a MET inhibitor, mTOR inhibitor and a PARP inhibitor), and 19 were evaluable for RR analysis (among them, we underlined, that: 9 were included in anti-angiogenic analysis; 4 studies were included in anti-EGFR analysis; 3 studies were included in anti-HER2 analysis; single trials respectively investigating a Hedgehog inhibitor, a mTOR inhibitor and a PARP inhibitor). Twenty studies were evaluable for toxicity.

Figure 1. PRISMA chart showing the trial exclusion and inclusion process in the meta-analysis. SEER, Surveillance, Epidemiology, and End Results.

Study characteristics

Our analysis included 13 front-line (involving 4268 patients), 4 second-line (1140 patients), and 5 beyond second-line (1614 patients) trials in which targeted therapy-based treatment was compared to other conventional treatment. In particular, we evaluated 10 trials involving anti-angiogenic agents in experimental arm, 3 studies involving HER2 targeting agents, 5 trials involving anti-EGFR agents in experimental arm, single trials respectively investigating a MET inhibitor, mTOR inhibitor, a PARP inhibitor, and a Hedgehog inhibitor.

OS analyses

Three trials were excluded from OS analysis because they did not report data for this end-point.^33,51 In our analysis targeted therapy plus conventional therapy showed a benefit in terms of OS in aGC patients compared to conventional therapy alone (pooled HR 0.823; 95%CI 0.743–0.912; p ≤ 0.001; Fig. 2A). We reported meta-analysis results on each single target-therapy pathway. In particular, in our analyses, the most significant benefit occurred for antiangiogenic agents in terms of OS (HR 0.759; 95%CI 0.655–0.880; p < 0.001; Fig. 2B). Regarding angiogenesis pathway, we reported the advantage in subgroup analysis for second-line treatment (pooled HR 0.808; 95%CI 0.685–0.953; p = 0.011) and subsequent lines of treatment (pooled HR 0.633; 95%CI 0.474–0.846; p = 0.002), but not for first-line treatment. No statistically significance difference was found about subgroup analysis for class of drug (data not shown). Conversely by meta-analysis conducted for EGFR pathway, no benefit was found (HR 1.077; 95%CI 0.847–1.370; p = 0.543; Fig. 2C). No advantage for line of treatment was demonstrated by subgroup analysis for this pathway.

Figure 2. Comparison of OS according to involved pathway and treatment line ((A) all pathways), ((B) anti-angiogenic drugs), ((C) anti-EGFR drugs) and ((D) anti-HER2 drugs), between patients treated with a targeted therapy-containing regimen versus conventional schedule. Abbreviation: overall survival, OS; hazard ratio, HR; chemotherapy, CHT; best supportive care, BSC.

Our meta-analysis performed on HER2 pathway confirmed the improvement in terms of OS outcome known for this class of drugs (HR 0.823; 95%CI 0.722–0.939; p = 0.004; Fig. 2D). No advantage for line of treatment was demonstrated by subgroup analysis for this pathway.

PFS analyses

PFS data were not provided in one trial that was therefore excluded from PFS analysis.⁵¹ Targeted therapy-based treatment was overall found to be significantly associated with improved PFS (pooled HR 0.762; 95%CI 0.664–0.875; p < 0.001; Fig. 3A). The analysis for single pathway demonstrated a significant PFS benefit in the anti-angiogenic pathway (pooled HR 0.695; 95%CI 0.567–0.853; p < 0.001; Fig. 3B). For this subgroup, we demonstrated an advantage both in first-line of treatment and in subsequent lines of treatment (pooled HR 0.873; 95%CI 0.770–0.990; p = 0.034; HR 0.463; 95%CI 0.351–0.610; p < 0.001, respectively). Conversely, there is no benefit in second-line of treatment. No differences about TKI/Ab subgroup analysis for this pathway. Interestingly, the analysis of anti-EGFR subgroup did not reach a statistically significant PFS advantage (pooled HR 1.117; 95%CI 0.985–1.268; p = 0.085; Fig. 3C). No advantage for line of treatment was demonstrated by subgroup analysis for this pathway. Regarding anti-HER2 analyses, we reported a significant benefit in term of PFS (pooled HR 0.780; 95%CI 0.646–0.941; p = 0.009; Fig. 3D). No evaluable subgroup analysis for line of treatment and for class of drug was performed for limited available studies.

Figure 3. Comparison of PFS according to involved pathway and treatment line ((A) all pathways), ((B) anti-angiogenic drugs), ((C) anti-EGFR drugs) and ((D) anti-HER2 drugs), between patients treated with a targeted therapy-containing regimen vs. conventional schedule. Abbreviation: overall survival, OS; hazard ratio, HR; chemotherapy, CHT; best supportive care, BSC.

RR analyses

One trial did not report data in term of RR, and was excluded from this analysis.²⁷ Regarding RR analyses for anti-angiogenic and for anti-EGFR agents, no advantage was reported (OR for RR 1.096; 95%CI 0.724–1.660; p = 0.665; OR for RR 0.913; 95%CI 0.449–1.858; p = 0.802, respectively; Fig. S1A–C). In the anti-HER2 drugs analysis we reported a significant improvement in term of RR (OR for RR 2.013; 95%CI 1.283–3.160; p = 0.002; Fig. S1D).

Toxicity analyses

In Table 1, we reported in detail common adverse events for each pathway. We analyzed specifically haematological toxicities, nausea and vomiting finding no difference for each pathway. In anti-HER2 agents group we observed an increase for diarrhea, while in anti-EGFR group, in first-line of treatment, we reported an increase of rash. No relevant toxicity for anti-angiogenic agents was demonstrated.

Table 1. Most common grade 3–4 adverse events analyzed in the meta-analysis

	3–4 grade (%)	Overall risk ratio (95%CI)
Adverse events	TT	CT	p Value
Anaemia	12.8	11.9	1.040 (0.842–1.286)	0.714
Thrombocytopenia	5.8	4.0	1.147 (0.568–2.314)	0.703
Neutropenia	31.6	27.1	1.145 (0.935–1.404)	0.191
Nausea	6.0	5.2	1.142 (0.830–1.571)	0.416
Astenia	4.7	5	0.965 (0.601–1.548)	0.882
Fatigue	9.2	7.9	1.224 (0.966–1.551)	0.095
Vomiting	4.9	5.6	0.859 (0.629–1.174)	0.340
Diarrhea	8.9	5.8	1.622 (1.062–2.477)	0.025
EGFR			1.451 (0.955–2.206)	0.081
HER2			4.042 (1.802–9.065)	0.001
Angiogenesis			1.029 (0.453–2.337)	0.946
mTOR			3.000 (0.317–28.353)	0.338
Rash	17.4	6.4	3.455 (1.449–8.234)	0.005
EGFR			4.784 (1.075–21.295)	0.040
mTOR			3.000 (0.124–72.771)	0.500
Hand-foot syndrome	4.1	6.4	0.725 (0.351–1.497)	0.384

Abbreviations: conventional therapy, CT; targeted therapy, TT.

Risk of bias in individual studies

Begg's funnel plot showed no significant evidence for publication bias (Fig. 4).

Figure 4. Funnel plot (Begg's test) assessing publication bias.

Discussion

With our meta-analysis we investigated the role of targeted therapy in aGC by comparison of targeted therapy-based regimens to any other conventional treatment. Moreover, we performed single meta-analysis stratified for each involved pathway. The major finding from our study is a significant survival benefit (OS: HR 0.823; PFS: HR 0.762) for targeted therapy in its whole, which can be ascribed to the reported advantage of anti-angiogenic and anti-HER2 agents as compared to conventional treatments in aGC, including all endpoints and subgroups. In fact we demonstrated OS improvement for studies on agents targeting anti-angiogenic and anti-HER2 pathways (HR 0.759; HR 0.823, respectively), paralleled by a significant PFS benefit (HR 0.695; HR 0.780, respectively). In term of survival outcomes no benefit was found for anti-EGFR agents. Only meta-analysis investigating anti-HER2 pathway showed a significant benefit in RR (OR 2.013). By evaluation of studies investigating each pathway, our work confirmed a good tolerability profile of targeted therapy, thus underscoring its potential role in management of aGC particularly in lines of treatment subsequent to front-line.

Some points need however to be addressed. The first point is that, while the survival improvement in this meta-analysis is remarkable, our results do not reflect the common clinical practice because only few trials reached their primary endpoint of improved aGC patient survival and therefore led to regulatory agency drug approval and treatment changes. One possible explanation for this finding is the underpowering of single trials both for the limited number of enrolled patients and for the overestimated expected benefit in the trial design. In particular, we observed that the major limit of the trials included in our meta-analysis is the lack of a priori selection or patient stratification. We are now aware that GC is an heterogeneous disease in which differences in anatomical site, race, and histotype characterize distinct diseases in terms of biology, genotype, and prognosis. Indeed, none of the included studies specifically reported the outcome of the adenocarcinoma subtype (intestinal vs. diffuse tumors). Moreover, with the exception of 3 trials in which a stratification for biomarker analysis (HER2 status) was reported, all the other trials were performed on unselected patient populations.^25,36,53 This could be an important information on patients survival outcome, because it is known that the expression of HER2 is different in the different histotypes (30% in the intestinal type and 7 % in the diffuse type). Apart from this specific condition there is no evidence about different prognostic and predictive effect of other pathways. For anti-angiogenic agents, this limitation is often due to the lack of validated predictive biomarkers.^{26,31,33-35,37,39,52,60-64} In other cases, in particular for the anti-EGFR agents, even if the prominent role of RAS/RAF mutational status in colon cancer has been widely recognized so far, it is conceivable that this information was unknown at the time of studies design.^{27,30,51,65-67} Therefore it is highlighted the need of new trials designed to investigate the relevance of RAS mutation in predicting response to anti-EGFR mAbs, in aGC. In particular, preliminary findings suggest that a worse outcome could be demonstrated in chemotherapy-treated patients in the presence of N-RAS mutations.⁶⁸ Furthermore, our results showed that, regardless a prior selection of patients, the use of anti-angiogenic drugs is overally statistical significant on all survival endpoints, specifically in subsequent lines of treatment and with a good tolerability profile.^30,43 Concerning RR results, we demonstrated a significant advantage for targeting anti-HER2 pathway alone, thus mimicking the results obtained in breast cancer where this pathway is determinant in driving tumor progression and represents a valuable therapeutic target in neoadjuvant treatment.^69-72

A second important point is that the most significant results on survival endpoints were achieved on subsequent lines of treatment as compared to first-line treatment. A possible explanation is that the most effective agents evaluated in subsequent lines of treatment, such as ramucirumab, have not yet been investigated in first line setting. Futhermore, combination regimens used in first line chemotherapy are investigated in trials heterogeneous for sample size which is generally small. Taking into account the choice of chemotherapy, trial design didn't consider drug interactions. Indeed, the combination with recognized anti-angiogenetic effect such as paclitaxel intensifies and supports bevacizumab activity regardless of cancer type.⁷³ Similarly, the combination with agents which has a recognized detrimental effect could underestimate the effectiveness of targeted therapy (e.g. capecitabine and cetuximab combination in colorectal cancer).⁷⁴

In subsequent lines of treatment conventional therapy is often represented by BSC, mainly because there is not, at present, a recognized standard/control arm in third line treatment.^29,34,35 Indeed, we think that these not expected findings open avenues for use of these new drugs in the maintenance settings, as already widely demonstrated in other cancers. New research paths are strongly suggested.^75-78

A third point regards the interpretation of our findings about subgroup analysis for TKI/Ab. We reported no differences for class of drugs in each pathway analyzed.

Recent studies proposed a possible alternative mechanism of action for mAbs apart intracellular pathway perturbation.⁷⁹ Indeed, in several tumors such as breast cancer, non-small cell lung cancer (NSCLC), and colorectal cancer, it was suggested an immune-modulating role for Abs. Based on some experimental findings, it appears that an immune modulating effect should be considered even for mAbs used in aGC treatment.^80-86 Interestingly, in aGC, while bevacizumab did not demonstrate a significant benefit, indeed, in our analysis, we observed that trials evaluating other agents showed a highly significant efficacy in term of survival benefit in the anti-angiogenesis subgroup in aGC.^17,35,37 Several studies suggested that targeting VEGFR2 may elicit and improve the immune response directed toward the tumor and its microenvironment trough different mechanisms including direct activation of CD8⁺ cells and targeting of the FOXP3^high regulatory T cells.^87-95 These findings could explain, at least in part, the efficacy of ramucirumab, a mAb IgG1 direct to VEGFR2 and appear of interest taking also into account that GC is a disease in which VEGFR2 expression has not been reported to retain a prognostic weight.⁹⁶ Moreover, recent evidence underscored the prognostic roles of both VEGFA and VEGFC expression in GC.⁹⁷ VEGFC, in particular, is reported to induce VEGFR2/VEGFR3 heterodimer formation thus leading to cell proliferation (via RAS signaling) and angiogenic sprouting.⁹⁸ Of interest, VEGFC has been associated to both innate and acquired resistance to bevacizumab, and should be noted again that in our analysis VEGFR2-targeting ramucirumab appears to perform better than bevacizumab.⁹⁹ Consistently with these results, TKIs that mainly target VEGFR2 signaling, such as apatinib, showed the better results in term of OS. Considering the role played by the VEGFR2/VEGFR3 axis in inflammation, immunology, and angiogenesis these agents seems to be promising and warrant further investigation in aGC. Another interesting pathway involved in the process of progression and invasion of cancer is PI3K/AKT/mTOR. This pathway seems to play a central role in aGC, for these reasons specific and taking into account the currently available mTORC1 inhibitors, such as everolimus that proved to be effective in other cancer type, these compounds should be investigated in aGC.^100,101

Finally it has to be pointed out that our meta-analysis has some limitations: it was performed on literature data and it was not possible to retrieve data about all end-points from all studies; another potential bias is the heterogeneity of agents used in the included trials. An important limitation, as suggested by pre-planned subgroup analyses in AVAGAST on regional differences in the efficacy of anti-angiogenic therapy, is that we could not analyze data for single ethnic population.²⁶ Additionally, the weight assessment for each drug is difficult and it has to be considered that specific polymorphisms in ADME genes among different ethnic groups could modulate response to treatment.¹⁰² Taking into account all these points, it is clear that our findings are not intended to identify the new standard treatment for aGC. Indeed, the goal of a meta-analysis is not to change clinical practice but to suggest rational new trials design, and underscore novel areas of investigation. At this aim our data strongly support additional investigation on anti- angiogenic pathways-targeted agents in aGC. On these basis future trials design could foresee enrollment of patients selected on one or more specific targets predefined by new technologies such as NGS that describe a peculiar molecular profile for related disease but not correlated to the site of disease.

In conclusion, at our knowledge, our work provides the first proof-of-concept that targeted therapy is an effective therapeutic approach against aGC. However, at this time, if we exclude trastuzumab in HER2-selected patients and ramucirumab, this important evidence cannot be transferred into clinical practice for other targeted agents. New trials designed on predefined biomarkers and aimed to identify the right drug for the right patient are eagerly awaited. In the absence of good designed biomarker-driven clinical trials, the use of targeted therapy is presently an unmet goal for clinical practice in a prognostically unfavourable disease as aGC.

Methods

Study design

We performed a systematic review and meta-analysis of all published randomized studies focused on each previously reported targeted pathway, to investigate the weight of targeted therapy-based schedules compared to conventional therapies in the management of aGC. The study includes therefore prospective and RCTs on aGC in all treatment lines with OS, PFS, RR, and toxicity, as predefined endpoints.

Comparisons for each one meta-analysis were performed as follows:

• Targeted therapy plus conventional therapy versus conventional therapy alone:

  • Line of treatment subgroup;

  • Class of drugs subgroup: monoclonal antibody (mAb)/tyrosine kinase inhibitor (TKI);

• Endpoints were OS, PFS, RR and toxicity.

Searching

We retrieved the most widely recognized bibliographic sources (PubMed, Embase, and the Central Registry of Controlled Trials of the Cochrane Library), major meeting proceeding databases (ASCO and ESMO), and selected studies presented between January 2005, at the time of consolidation of chemotherapy-based treatment in the management of aGC, and December 2014. Prospective studies only, were allowed in this analysis in order to reduce or minimize the risk of selection or information bias.^103-105 The search was performed by the following key-words: “gastric”, “stomach”, “tumor”, “cancer”, “advanced”, “metastatic”, “therapy”, “targeted”, “prospective”, and “randomized” in different combinations: i.e. “advanced gastric cancer, targeted therapy”. The ‘related articles’ function and references retrieved from articles were used to perform the search of all related studies, abstracts and citations. For this search only papers written in English language were considered.

Selection

In the studies included in the present review, patients had to be enrolled according to the characteristics described in Table 2.

Table 2. Main characteristics of the randomized trials included in the meta-analysis. Abbreviations: overall survival, OS; progression free survival, PFS; hazard ratio, HR; TT: target therapy; ST: standard therapy; chemotherapy, CHT; best supportive care, BSC

Author	Year	n° pts	Arms	Phase	PFS (HR)	OS (HR)	Response rates% (TT arm)	Response rates% (ST arm)
I line
Rao^32	2010	72	CTH ± Matuzumab	II	1.13 (0.63–2.01)	1.02 (0.61–1.71)	31	58
Eatock^33	2013	112	CTH ± Trebananib 10 mg/Kg	II	0.99 (0.63–1.55)	−	27	35
Eatock^33	2013	115	CTH ± Trebananib 3 mg/Kg	II	0.98 (0.64–1.51)	−	43	35
Koizumi^52	2013	93	CHT ± TSU-68	II	1.23 (0.74–2.05)	0.74 (0.46–1.19)	62	56
Richards^51	2013	150	CHT ± Cetuximab	II	−	−	38	26
Waddell^44 REAL-3	2013	553	CHT ± Panitumumab	II/III	1.22 (0.98–1.52)	1·37 (1.07–1.76)	46	42
Bang^25 ToGA	2010	594	CTH ± Trastuzumab	III	0.71 (0.59–0.85)	0.74 (0.6–0.91)	47	35
Ohtsu^26 AVAGAST	2011	774	CHT ± Bevacizumab	III	0.80 (0.68–0.93)	0.87 (0.73–1.03)	46	37
Hecht^36 LOGIC	2013	545	CHT + Lapatinib vs CHT + Placebo	III	0.86 (0.71–1.04)	0.91 (0.73–1.12)	53	40
Lordick^30 EXPAND	2013	904	CHT ± Cetuximab	III	1.09 (0.92–1.29)	1 (0.87–1.17)	30	29
Cohen^54	2013	123	CHT ± Vismodegib	II	−	−	35	35
Iveson^59	2014	121	CHT + Rilotumumab	II	0.60 (0.45–0.79)	0.73 (0.53–1.01)	−	−
Yoon^57	2014	168	CHT + Ramucirumab vs CHT + Placebo	II	0.98 (0.69–1.37)	1.08 (0.73–1.58)	45	46
II line
Wilke^37 RAINBOW	2014	665	CHT + Ramucirumab vs CHT + Placebo	III	0.635 (0.536–0.752)	0.807 (0.678–0.962)	28	16
Bang^53 TYTAN	2013	261	CHT ± Lapatinib	III	−	0.84 (0.64–1.11)	27	9
Bang^55	2013	124	CHT ± Olaparib	II	0.80 (0.54–1.18)	0.56 (0.35–0.87)	26	19
Moehler^56	2013	90	CHT + Sunitinib vs CHT + Placebo	II	1.107 (0.703–1.742)	0.816 (0.495–1.343)	20	29
>II line
Kim^27	2011	82	CHT ± Nimotuzumab	II	0.86 (0.51–1.43)	0.584 (0.24–1.4)	−	−
Yi^31	2012	107	CHT ± Sunitinib	II	0.77 (0.52–1.16)	0.94 (0.6–1.49)	41	14
Li^34	2013	96	Apatinib 850 mg vs Placebo	II	0.29 (0.16–0.52)	0.37 (0.22–0.62)	7	0
Li^34	2013	96	Apatinib 425 mg vs Placebo	II	0.34 (0.18–0.64)	0.41 (0.24–0.72)	15	0
Fuchs^35 REGARD	2013	355	Ramucirumab vs Placebo	III	0.48 (0.37–0.62)	0.77 (0.6–0.99)	3	3
Ohtsu^29 GRANITE-1	2013	656	BSC + Everolimus vs BSC + Placebo	III	0.66 (0.56–0.78)	0.90 (0.75–1.08)	4	2
Qin^58	2014	270	Apatinib vs Placebo	III	0.44 (0.33–0.61)	0.71 (0.54–0.94)	3	−

Abbreviations: number of patients, n; pts; overall survival, OS; hazard ratio, HR; progression free survival, PFS; TT: target therapy; ST: standard therapy; chemotherapy, CHT; best supportive care, BSC.

Inclusion criteria

The studies had to report patients with diagnosis of locally aGC or metastatic disease. Randomized controlled trials, with or without blinding were included. We considered abstracts or unpublished data if sufficient information on study design, characteristics of participants, interventions, and outcomes were available. In the experimental arm patients received a targeted therapy-based schedule. In the control arm patients received a conventional schedule for disease stage.

Exclusion criteria

Non-comparative studies; non-prospective studies; languages other than English; non-comparable end-points; way of chemotherapy or targeted agents administration different from systemic, or oral (e.g., Intra-arterial or intra-peritoneal infusion); no radiotherapy.

Data extraction

The studies were examined, independently, by 2 investigators (DC and NS) in order to select homogeneous studies.¹⁰⁶ Different variables from selected trials such as number of patients enrolled, year of publication, treatment schedule, and efficacy results were extracted and evaluated. Data regarding the occurrence of toxicities were obtained from the safety profile of each study. Any discrepancy was resolved by an arbiter (P.T.). Primarily, we analyzed OS and PFS, subsequently, RR and Toxicity Rates (TRs), regarding grade 3–4 of adverse events, for all patients. Data extraction was conducted according to the PRISMA statement.¹⁰⁷

Validity assessment

The quality assessment of selected studies was evaluated according to the Cochrane reviewers' handbook for 5 requirements: method of randomization, allocation concealment, blindness, withdrawal/dropout, and adequacy of follow-up.^108,109 10 trials were scored A (low risk of bias), 6 trials were scored B (intermediate risk of bias), and 6 trials were scored C (high risk of bias) (Table S1).

Quantitative data synthesis

For each targeted-pathway the meta-analyses were carried out in order to evaluate the effects of the targeted-based treatments [biological +/− chemotherapeutic agents] on the pre-specified end-points.¹¹⁰ Regarding the outcome of patients, survival data have been extracted as HRs of OS, and PFS with relative confidence intervals (95%CI). The interaction between survival and experimental treatment was obtained by each study from the HRs logarithm. The overall effect of combined treatments on RR and toxicity rate (TR) was calculated using method for dichotomous data (odds ratio and risk ratio assessment; 95%CI). In each meta-analysis subgroup analyses, based on the class of agents and line of treatment, were performed for all end-points. Cochrane's Q-test and I² statistics were used to assess heterogeneity between studies and the random-effects model was used for the analysis taking into account the intent of comparing trials based on drugs with different mechanisms of action.¹¹¹ Pooled data analysis was performed according to the DerSimonian and Laird test.^112,113 The presence of publication bias was investigated through Begg's test for single pathway and by visual inspection of funnel plots.¹¹⁴ A 2-tailed p value equal or lower than 0.05 was considered statistically significant. All the statistical analyses were performed by using STATA SE v. 13.1 (STATA_ Corporation, Texas, USA).¹¹⁵

Disclosure of Potential Conflicts of Interest

All authors declare that they have no conflicts of interest.

Contributors

DC, NS, PT, and PT designed the study. DC, NS, FC, and FM did the literature search, and extracted data. DC realized the figures. FC, LF, SC, AMDA and SG performed the tables. All authors collected data. DC, NS, CB, PT, and, PT interpreted and analyzed the data, and wrote the manuscript. All authors read, and approved final version of the manuscript.

Acknowledgments

This work is supported by PhD program of Magna Graecia University: PhD in molecular oncology and translational and innovative medical and surgical techniques.

Supplemental Material

Supplemental data for this article can be accessed on the publisher's website.