Does gemcitabine-based combination therapy improve the prognosis of unresectable pancreatic cancer?

Abstract

AIM: To assess whether gemcitabine-based combination therapy improves the prognosis of unresectable pancreatic cancer compared with gemcitabine treatment alone.

METHODS: A quantitative up-to-date meta-analysis was undertaken to investigate the efficacy of gemcitabine-based combination treatment compared with gemcitabine monotherapy in locally advanced or metastatic pancreatic cancer. Inclusion was limited to high-quality randomized clinical trials.

RESULTS: Twenty-six studies were included in the present analysis, with a total of 8808 patients recruited. The studies were divided into four subgroups based on the different kinds of cytotoxic agents, including platinum, fluoropyrimidine, camptothecin and targeted agents. Patients treated with gemcitabine monotherapy had significantly lower objective response rate [risk ratio (RR), 0.72; 95% confidence interval (CI): 0.63-0.83; P < 0.001], and lower 1-year overall survival (RR, 0.90; 95%CI: 0.82-0.99; P = 0.04). Gemcitabine monotherapy caused fewer complications, including fewer grade 3-4 toxicities: including vomiting (RR, 0.75; 95%CI: 0.62-0.89; P = 0.001), diarrhea (RR, 0.66; 95%CI: 0.49-0.89; P = 0.006), neutropenia (RR, 0.88; 95%CI: 0.72-1.06; P = 0.18), anemia (RR, 0.96; 95%CI: 0.82-1.12; P = 0.60), and thrombocytopenia (RR, 0.76; 95%CI: 0.60-0.97; P = 0.03) compared with gemcitabine combination therapies.

CONCLUSION: Gemcitabine combination therapy provides a modest improvement of survival, but is associated with more toxicity compared with gemcitabine monotherapy.

Key Words: Pancreatic cancer; Gemcitabine; Combination therapy; Outcome; Meta-analysis

INTRODUCTION

Pancreatic cancer is a highly lethal disease, and represents the fourth cause of cancer-related death in the Western world. In the United States, the incidence of pancreatic cancer almost equals its mortality, with 43  140 estimated new cases and 36  800 deaths in 2010. Globally, the incidence of new cases is approximately 230  000 each year[1,2]. Despite advances in diagnosis, staging, and surgical management of the disease during the past decade, the 5-year survival rate in the United States, Europe, and Australia is less than 5%[3-5]. Pancreatic cancer is characterized by a rapid disease progression and highly invasive tumor phenotype, with most patients having unresectable disease at diagnosis, and chemotherapy being the only possible treatment option for these patients[6].

Gemcitabine is a pyrimidine antimetabolite and analog of deoxycytidine. It was approved by the Food and Drug Administration as a first-line treatment for patients with locally advanced (stage II or stage III disease when surgery is not an option) or metastatic (stage IV) cancer of the pancreas, and has been widely used during the last decade. However, because of high levels of intrinsic and acquired chemoresistance, a large number of patients do not respond to gemcitabine[7]. To improve clinical efficacy, systemically administered gemcitabine is often combined with a second cytotoxic agent, such as platinum analogs, fluoropyrimidine, or a targeted cytotoxic agent. Numerous clinical trials have aimed at proving the superiority of gemcitabine-based combination therapy over single-agent gemcitabine treatment. However, most of the results of clinical trials have important limitations, including lack of statistical power because of small study populations. Thus, there remain several areas of controversy and uncertainty concerning optimal treatment regimens.

In the present study, we undertook a quantitative up-to-date meta-analysis to investigate the efficacy of gemcitabine-based combination treatment in locally advanced or metastatic pancreatic cancer. The aim of this study was to assess whether gemcitabine-based combination therapy improves the prognosis compared with gemcitabine treatment alone and to discuss possible mechanisms.

MATERIALS AND METHODS

Literature search strategy

We carried out a comprehensive search of the literature, including PUBMED (http://www.ncbi.nlm.nih.gov/pubmed), EMBASE (http://www.embase.com/home), American Society of Clinical Oncology abstracts (http://www.asco.org), and the Cochrane Database (http://www.thecochranelibrary.com). The following keywords were used in the search: (“Gemcitabine” or “Gemzar”), and (“pancreatic cancer” or “pancreatic tumor” or “pancreatic carcinoma”), and (“clinical trial”). The deadline for a publication to be eligible for this study was November 5, 2011. In addition to the online search, references from reviews and original articles were also scanned manually to identify further trials that met the eligibility criteria. No language restrictions were applied.

Inclusion and exclusion criteria

Studies included in the analysis had to meet all of the following criteria: (1) prospective, randomized, controlled open or blinded trial; (2) patients with histologically confirmed locally advanced or metastatic pancreatic ductal adenocarcinoma; and (3) assessment of the efficacy of gemcitabine combination therapy vs gemcitabine alone. Non-randomized trials and quasi-randomized trials, studies of curatively aimed resection, and studies where patients had multiple cancers, were excluded to avoid clinical heterogeneities between different studies.

Data extraction and quality assessment

Data extraction and quality assessment were performed independently by two reviewers (CS and DA). Any disagreements between the reviewers were discussed with a third reviewer (RA) to achieve a consensus. The data extracted from the eligible studies included: first author, year of publication, patient characteristics, intervention, and clinical outcome (toxicity, response rate, overall survival and progression-free survival). If the same trial appeared on sequential or multiple publications, the data from the most recent or comprehensive one was included.

The methodological quality of included studies was assessed using the “Jadad scale” or “Oxford quality scoring system”[8]. This tool is an evidence-based quality assessment tool. There are three items (randomization, double blinding, withdrawals and dropouts) directly related to bias reduction for assessment. Each item is given a score of 1 point for each “yes” or 0 points for each “no”, and 1 additional point for appropriate randomization and double blinding. Every eligible study were assessed and given a score from 0-5.

Data analysis

The outcome measures were objective response rate [objective response rate (ORR) = complete response (CR) + partial response (PR)], as previously defined[9], 1-year overall survival (OS), median progression-free survival (PFS), median OS and toxicity. The analysis of ORR (number of partial and complete responses, as defined by Response Evaluation Criteria in Solid Tumors[10]), PFS (time from randomization to progression or death) and OS (time from the date of random assignment until date of death or date last known to be alive) were based on the intent-to-treat population, consisting of all patients randomly assigned onto every study. Toxicity was graded according to the National Cancer Institute Common Toxicity Criteria, and based on the safety population. The median PFS and median OS were assessed using the paired t-test. P < 0.05 was considered statistically significant.

Meta-analysis was performed using the Review Manager (version 5.1, provided by The Cochrane Collaboration). The strength of the associations between treatment and outcomes were estimated by risk ratio (RR, a ratio of the probability of the event occurring in the exposed group versus a non-exposed group) and 95% confidence interval (CI)[11]. The heterogeneities between different studies or different subgroups were estimated using Cochran’s Q test[12,13]. In addition, the I² value was used for evaluating the extent of heterogeneity between studies or subgroups[14]. The I² index measures the extent of true heterogeneity dividing the difference between the result of the Q test and its degrees of freedom by the Q value itself, and multiplied by 100. I² indicated the possibility of the variability among effect sizes caused by true heterogeneity between studies, not by sampling error[14]. If a significant Q test (P < 0.1) or I² > 50% indicated that heterogeneity existed between studies, the random effects model was used for meta-analysis, otherwise the fixed effects model was used. Publication bias was assessed by visual inspection of funnel plots.

RESULTS

Literature search and trial flow

The results of the literature search are depicted in Figure 1. Some 2707 studies were initially retrieved. After reading the title and abstract, 2675 studies were excluded, and 32 studies were left for further review. Six of these were excluded because of poor quality or failure to meet the inclusion criteria.

Figure 1 Flow chart of study selection.

Characteristics of selected trials

The characteristics of the included studies are shown in Table 1. Twenty-six studies were included in the present analysis, with 8808 recruited patients. The studies were divided into four subgroups based on types of cytotoxic agents used, including platinum[15-20], fluoropyrimidine[21-26], camptothecin[21,27-29] and targeted agents[30-41]. All included studies enrolled patients that were ≥ 18 years; had a histologically confirmed locally advanced or metastatic pancreatic cancer not amenable to surgical resection; no previous chemotherapy; a life expectancy of at least 12 wk; a Karnofsky performance status ≥ 50, or World Health Organization performance status of 0-2; and adequate liver, renal, and hematopoietic functions. The details of toxicity assessment are shown in Table 2 (only grade 3 and 4 toxic effects are presented), including vomiting, diarrhea, neutropenia, anemia and thrombocytopenia.

Table 1 Characteristics of selected trials.

Ref.	Intervention	n	Male (%)	Age (range)	Metastatic (%)	Response rate (%)	Median PFS (mo)	1-yr OS	Median OS (mo)	Quality
Louvet et al[15]	Gem	163	53	60.1 (22-75)	70	17.3	3.7	27.8	7.1	4
	Gem + Oxaliplatin	163	60	61.3 (35-77)	68	26.8	5.8	34.7	9.0
Poplin et al[16]	Gem	275	56.4	64 (31-88)	90.2	6	2.6	16	4.9	4
	Gem + Oxaliplatin	272	45.6	63 (29-96)	89.3	9	2.7	21	5.7
Heinemann et al[17]	Gem	95	61.9	66 (43-85)	78.9	9.0	3.1	24.7	6.0	4
	Gem + Cisplatin	95	65.3	64 (37-82)	80	11.5	5.3	25.3	7.5
Colucci et al[18]	Gem	54	50	63 (43-75)	65	9.2	2	11	5	4
	Gem + Cisplatin	53	35	60 (33-71)	68	26.4	5	11.3	7.5
Colucci et al[19]	Gem	199	56.8	63 (37-75)	82.9	10.1	3.9	34.0	8.3	4
	Gem + Cisplatin	201	62.2	63 (35-75)	84.6	11.4	3.8	30.7	7.2
Kulke et al[20]	Gem	64	66	58.9 (31-81)		14	3.3	NR	6.4	4
	Gem + Cisplatin	66	56	58.9 (36-84)	100	13	4.5	NR	6.7
	Gem + Irinotecan	64	68	60.8 (32-77)		14	4.0	NR	7.1
Berlin et al[21]	Gem	162	53.7	64.3 (33-85)	90.1	5.6	2.2	NR	5.4	4
	Gem + 5-fluorouracil	160	51.8	65.8 (28-84)	89.4	6.9	3.4	NR	6.7
Herrmann et al[22,24]	Gem	159	53	62 (36-84)	79	8	3.9	30	7.2	4
	Gem + Capecitabine	160	54	62 (27-83)	80	10	4.3	32	8.4
Cunningham et al[23]	Gem	266	58	62 (26-83)	71	12.4	3.8	22.0	6.2	4
	Gem + Capeitabine	267	60	62 (37-82)	70	19.1	5.3	24.3	7.1
Scheithauer et al[25]	Gem	42	55	66 (39-75)	100	14	4.0	37.2	8.2	4
	Gem + Capeitabine	41	66	64 (40-75)	100	17	5.1	31.8	9.5
Costanzo et al[26]	Gem	49	48	64 (34-75)	73	8	3.5	18	7.75	4
	Gem + 5-fluorouracil	45	63	62 (44-75)	67	11	4.5	20	7.5
Abou-Alfa et al[27]	Gem	174	57	62.3 (30-84)	78	5.7	3.8	21	6.2	4
	Gem + Exatecan	175	53	63.0 (36-85)	79	7.1	3.7	23	6.7
Stathopoulos et al[28]	Gem	74	42	64 (44-83)	66	10	2.9	21.8	6.5	4
	Gem + Irinotecan	71	39	64 (31-84)	60	15	2.8	24.3	6.4
Lima et al[29]	Gem	180	53.3	60.2 (32-82)	80.6	4.4	3.0	22	6.6	4
	Gem + Irinotecan	180	57.2	63.2 (38-81)	82.2	16.1	3.5	21	6.3
Moore et al[38]	Gem	284	57	64 (36.1-92)	75.0	8.0	3.55	17	5.91	4
	Gem + Erlotinib	285	47.7	63.7 (37-84)	76.5	8.6	3.75	23	6.24
Cutsem et al[39]	Gem	347	58	62 (30-88)	77	8	3.6	24	6.06	4
	Gem + Tipifarnib	341	57	61 (29-89)	76	6	3.7	27	6.43
Eckhardt et al[40]	Gem	120	59	60 (35-86)	73	NR	3.03	NR	7.36	5
	Gem + Tipifarnib	124	64	63 (35-81)	71	NR	2.3	NR	6.73
Philip et al[37]	Gem	371	54	64.3	78	7	3.0	NR	5.9	4
	Gem + cetuximab	372	51	63.7	79	8	3.4	NR	6.3
Kindler et al[36]	Gem	300	51	65.0 (35-86)	85	10	2.9	NR	5.9	4
	Gem + Bevacizumab	302	58	63.7 (26-88)	84	13	3.8	NR	5.8
Kindler et al[35]	Gem	316	59	62 (35-89)	72	1.6	4.4	NR	8.3	5
	Gem + Axitinib	316	61	61 (34-84)	72	4.9	4.4	NR	8.5
Spano et al[41]	Gem	34	47	61.0 (36-78)	19	3	3.7	23.5	5.6	4
	Gem + Axitinib	69	51	65.0 (44-81)	40	7	4.2	36.8	6.9
Friess et al[34]	Gem	43	42	66 (56-80)	91	14	3.83	0.24	7.7	4
	Gem + Cilengitide	46	57	68 (40-80)	94	17	3.66	0.15	6.7
Richards et al[33]	Gem	44	72.7	64.1 (41-83)	86.4	5.3	3.0	17	5.1	4
	Gem + Enzastarin	86	53.5	68.3 (39-86)	90.7	8.6	3.4	19	5.6
Bramhall et al[31]	Gem	119	71	62 (37-85)	62	16	3.2	17	5.46	5
	Gem + Marimastat	120	69	62 (32-83)	59	11	3.08	18	5.51
Richards et al[32]	Gem	85	60.2	65 (36-83)	83.0	13.9	3.43	NR	7.13	5
	Gem + CA-994	85	59.3	62 (32-82)	82.6	11.8	3.06	NR	6.47
Oettle et al[30]	Gem	282	53.5	63 (28-82)	91.1	7.1	3.3	20.1	6.3	4
	Gem + PPemetrexed	283	60.4	63 (27-82)	90.1	14.8	3.9	21.4	6.2

NR: No record; PFS: Progression-free survival; OS: Overall survival.

Table 2 Grade 3 or 4 toxicity of selected clinical trials.

Ref.	Intervention	Vomiting	Diarrhea	Neutropenia	Anemia	Thrombocytopenia
Louvet et al[15]	Gem	3.2	1.3	27.6	10.3	3.2
	Gem + Oxaliplatin	8.9	5.7	20.4	6.4	14.0
Poplin et al[16]	Gem	7	4	33	10	13
	Gem + Oxaliplatin	16	6	22	6	11
Heinemann et al[17]	Gem	5.9	4.7	NR	10.6	10.6
	Gem + Cisplatin	22.2	3.3	NR	13.3	4.4
Colucci et al[18]	Gem	2	0	9	4	2
	Gem + Cisplatin	2	4	18	6	2
Colucci et al[19]	Gem	< 1	1.5	13.7	1	18.9
	Gem + Cisplatin	2.6	< 1	24.7	4.8	15.3
Kulke et al[20]	Gem	24.1	NR	82.8	20	43.1
	Gem + Cisplatin	29	NR	74.2	25.8	79
	Gem + Irinotecan	16.7	NR	41.7	8.3	23.3
Berlin et al[21]	Gem	8	NR	5	10	11
	Gem + 5-fluorouracil	7	NR	7	10	19
Herrmann et al[22,24]	Gem	4	2	19	6	8
	Gem + Capeitabine	4	5	23	8	4
Cunningham et al[23]	Gem	6	4	22	6	6
	Gem + Capeitabine	6	5	35	4	11
Scheithauer et al[25]	Gem	0	0	7.6	0	2.5
	Gem + Capeitabine	0	5	10	5	0
Costanzo et al[26]	Gem	0	0	2	6	0
	Gem + 5-fluorouracil	2	0	2	7	2
Abou-Alfa et al[27]	Gem	3.1	< 1	14.6	7.6	4.4
	Gem + Exatecan	5.3	1.1	30.3	5.9	15.4
Stathopoulos et al[28]	Gem	1.4	2.8	15.7	4.2	0
	Gem + Irinotecan	1.6	3.2	26.6	5	5
Lima et al[29]	Gem	8.2	1.8	31.9	13	14.2
	Gem + Irinotecan	13.9	18.5	37.6	16.2	19.6
Moore et al[38]	Gem	NR	< 1	27%	NR	11%
	Gem + Erlotinib	NR	2	24%	NR	10%
Cutsem et al[39]	Gem	9	3	30	16	12
	Gem + Tipifarnib	7	4	40	20	15
Eckhardt et al[40]	Gem	9.3	0	33.9	11	15.3
	Gem + Tipifarnib	4	0	35.5	12.1	16.1
Philip et al[37]	Gem	2.2	2.5	23.9	6.2	8.5
	Gem + cetuximab	6.6	2.8	23.3	9.7	6.6
Kindler et al[36]	Gem	NR	NR	29	8	12
	Gem + Bevacizumab	NR	NR	33	5	12
Kindler et al[35]	Gem	3.2	1.6	< 1	< 1	< 1
	Gem + Axitinib	3.9	1.3	0	0	0
Spano et al[41]	Gem	10	0	33	17	13
	Gem + Axitinib	6	6	28	8	20
Richards et al[33]	Gem	0	2.6	28.2	5.1	25.6
	Gem + Enzastarin	2.4	3.7	18.3	3.7	14.6
Bramhall et al[31]	Gem	14%	NR	NR	7%	NR
	Gem + Marimastat	6%	NR	NR	3%	NR
Richards et al[32]	Gem	9	3	NR	5	11
	Gem + CA-994	9	5	NR	13	2
Oettle et al[30]	Gem	3.7	0.7	12.8	2.9	6.2
	Gem + Pemetrexed	3.3	2.9	45.1	13.9	17.9

NR: No record.

The methodological quality of selected studies was high. Most trials had a Jadad score of 3, while three studies had a score of 4. Methods of double blinding were infrequently reported.

Gemcitabine combination therapy improves objective response rate compared with gemcitabine treatment alone

This analysis evaluated 26 trials (8146 patients) comparing gemcitabine monotherapy with the combination of gemcitabine and some other cytotoxic agent. Based on all studies, patients treated with gemcitabine monotherapy had a significantly lower ORR than gemcitabine combination therapy (RR, 0.72; 95%CI: 0.63-0.83; P < 0.001). The RRs in the different subgroups were 0.67, 0.71, 0.53 and 0.82, for platinum, fluoropyrimidine, camptothecin and targeted agents, respectively. All RRs in the subgroups were also significant. Data are shown in Figure 2.

Figure 2 Fixed effect model of risk ratio of objective response rate. CI: Confidence interval.

Gemcitabine combination therapy improves 1-year overall survival rate compared with gemcitabine alone

5717 patients from 18 trials were included in this meta-analysis comparing gemcitabine monotherapy with combination therapies for 1-year overall survival. The RRs of 1-year overall survival were analyzed both totally and in the different subgroups. Subgroup analysis showed that the RRs of the monotherapy-based 1-year overall survival were lower than for the combination groups (RR: 0.91, 0.95, 0.97 and 0.85, for platinum, fluoropyrimidine, camptothecin and targeted agents, respectively), but no significant differences were found. When analyzed in terms of total events, the RR shown in Figure 3 was 0.90 [(95%CI: 0.82-0.99); P = 0.04]. The result showed that the 1-year overall survival rate of gemcitabine monotherapy was almost 90% of that of the combination therapy.

Figure 3 Fixed effect model of risk ratio of 1-year overall survival rate. CI: Confidence interval.

Gemcitabine combination therapy increases the toxicity effect compared with gemcitabine alone

Outcomes of the meta-analysis of the main toxicities are presented in Figures 4, 5, 6, 7 and 8. All RRs with grade 3-4 toxicities analyzed in this study were lower in the gemcitabine monotherapy group than in the combination group. The incidence of vomiting (RR, 0.75; 95%CI: 0.62-0.89; P = 0.001), diarrhea (RR, 0.66; 95%CI: 0.49-0.89; P = 0.006) and thrombocytopenia (RR, 0.76; 95%CI: 0.60-0.97; P = 0.03) were all significantly different between the treatment groups, while no significant difference was noted regarding neutropenia (RR, 0.88; 95%CI: 0.72-1.06; P = 0.18) and anemia (RR, 0.96; 95%CI: 0.82-1.12; P = 0.60).

Figure 4 Fixed effect model of risk ratio of grade 3 or 4 vomiting. CI: Confidence interval.

Figure 5 Fixed effect model of risk ratio of grade 3 or 4 diarrhea. CI: Confidence interval.

Figure 6 Random effect model of risk ratio of grade 3 or 4 neutropenia. CI: Confidence interval.

Figure 7 Fixed effect model of risk ratio of grade 3 or 4 anemia. CI: Confidence interval.

Figure 8 Random effect model of risk ratio of grade 3 or 4 thrombocytopenia. CI: Confidence interval.

In subgroup analysis, the RRs were 0.46-0.98 for vomiting, 0.65-0.70 for diarrhea, 0.70-1.03 for neutropenia, 0.87-1.09 for anemia and 0.65-0.85 for thrombocytopenia, respectively. No subgroup differences in overall toxic outcomes, expect for vomiting, were found between the gemcitabine monotherapy group and the combination groups. Gemcitabine monotherapy decreased the incidence rate of vomiting by more than 50% compared with the gemcitabine plus platinum subgroup (RR, 0.46; 95%CI: 0.33-0.64; P < 0.001), while no significant differences for other subgroups was seen. In addition, the results showed that gemcitabine plus fluoropyrimidine therapy significantly increased the incidence of neutropenia (RR, 0.70; 95%CI: 0.55-0.88; P = 0.002).

The median PFS and median OS between gemcitabine monotherapy and combination therapies

The data of median PFS and OS in every study were extracted and assessed by a paired t-test. The results showed that only gemcitabine plus fluoropyrimidine significantly increased the median PFS (3.480 vs 4.520; P = 0.045) and median OS (6.950 vs 7.840; P = 0.038). Data are shown in Figure 9.

Figure 9 Median progression-free survival and overall survival between gem-monotherapy and combination therapy.

Publication bias

The funnel plots are shown in Figure 10. There was no obvious publication bias found in the analysis.

Figure 10 Funnel plots. A: Objective response rate; B: 1-year overall survival rate; C: Vomiting; D: Diarrhea; E: Neutropenia; F: Anemia; G: Thrombocytopenia. RR: Risk ratio.

DISCUSSION

Pancreatic cancer is a devastating disease with a dismal prognosis. Most of patients present with locally advanced or metastatic disease and are thus not candidates for surgical resection, thereby having to rely on palliative chemotherapy as the only treatment option. In this situation, the goals of chemotherapy should be to control tumor progression, decrease toxicity, and improve the survival rate. At present, gemcitabine monotherapy remains the standard treatment for patients with locally advanced or metastatic pancreatic cancer, but its efficacy is unsatisfactory. During the last decade, several randomized controlled clinical trials have evaluated gemcitabine in combination with various agents in an attempt to improve the prognosis of pancreatic cancer. The aim of this meta-analysis was to compare the therapeutic efficacy of gemcitabine-based combination treatments with gemcitabine alone in patients with locally advanced or metastatic pancreatic cancer. The results of this study showed that various combination therapies overall did not provide any major benefit compared with gemcitabine monotherapy.

Moreover, gemcitabine monotherapy had an almost 30% lower ORR than combination therapy regimens, but only had a 10% lower 1-year OS than combination therapy. According to the present data, 1-year OS in advanced or metastatic pancreatic cancer was low (less than 20%) after gemcitabine monotherapy; therefore, the combination therapy options studied did not improve outcome in a substantial way, the prognosis still being poor[21,31,42]. In addition, there were no significant improvements found in this analysis in median PFS and median OS after combination therapy compared with monotherapy, except for gemcitabine plus fluoropyrimidine treatment. Our results are similar to previously published analyses[23,43,44]. We also assessed the five most common toxicities related to chemotherapeutic treatment of locally advanced or metastatic pancreatic cancer, including hematological and gastrointestinal side effects. Grade 3-4 toxicities tended to be higher following combination therapy compared with monotherapy, and three of the five investigated toxicities reached significant differences. The RRs of vomiting, diarrhea, and thrombocytopenia in the monotherapy group were about 70% of that noted in the combination therapy group. Furthermore, subgroup analysis showed that the RR of vomiting after treatment with gemcitabine plus platinum was twice that of monotherapy. The results presented here showed that combination therapy induced more toxicity than gemcitabine alone. Similar results were found in a previous study[45]. This finding might explain why combination therapy received higher ORR, but without affecting the overall prognosis, as toxicities counterbalanced the positive effects.

In the present analysis, we included several kinds of combination therapies with gemcitabine, which could provide different anti-tumor effects and different uptake and toxicity profiles, including platinum based agents, topoisomerase inhibitors, taxanes, bevacizumab, cetuximab and other biologically targeted agents. However, no significant differences were found between the subgroups, except for side-effects (vomiting). In addition, the analysis of heterogeneity was not significant in most subgroups. This is interesting, and indicates that effectiveness of the different combinations regarding outcomes could not clearly be shown. There are two possible reasons that could explain these results. One is related to gemcitabine-uptake receptors. The other one could be the inherent drug-resistant character of pancreatic carcinoma cells and the effect of the surrounding stroma.

Human equilibrative nucleoside transporter-1 (hENT-1) is the main membrane channel responsible for intracellular bioavailability of gemcitabine. Preclinical and clinical data have revealed that only a minority of patients have high hENT1 expression and reduced levels of hENT-1 expression correlate with increased resistance to gemcitabine[7]. To increase gemcitabine efficiency and avoid overtreatment, determination of the hENT1 status at the time of diagnosis and modifying gemcitabine to allow it to bypass the receptor may represent a future mode of overcoming gemcitabine resistance[7]. Multidrug resistance proteins, including ABC transporters, have also been implied in drug resistance in pancreatic cancer and limit the efficacy of gemcitabine[46]. The pancreatic stellate cell has a key role in stroma formation and secretes factors that promote pancreatic tumor growth and resistance to radiation and gemcitabine chemotherapy[47]. The hypoxic stroma could be a physical barrier preventing chemotherapeutic drugs from reaching the pancreatic cancer cells per se, and depletion of the stroma could enhance cancer drug delivery[48]. Recently, cancer stem cells and epithelial to mesenchymal transition (EMT)-type cells have been implicated in metastasis formation and drug resistance[49]. Selective targeting of cancer stem cells and EMT-type cells might represent a future strategy to enhance chemosensitivity. The plasma circulating time of gemcitabine is short and its rapid metabolism thus limits tumor uptake. Using novel formulas such as PEGylated gemcitabine may result in a prolonged circulation time, and possibly inducing cytotoxicity at lower concentrations than those required with treatment with the native drug[50]. Pancreatic cancer, including its concurrent stroma, is complex; therefore, treatment with a single anti-tumor agent may not be effective enough. Determining the underlying mechanisms of drug resistance in pancreatic cancer will provide ways of overcoming chemoresistance and provide a foundation for novel therapies.

During the last decade, several meta-analyses have focused on the efficacy of combination therapy in locally advanced or metastatic pancreatic cancer[43,44,51-54]. Most of these studies have reported positive trends or have concluded that combination treatments may improve outcome, but only a limited number of the studies have reported toxicity data, or have only focused on one anti-tumor agent. In the present study, we collected and analyzed the outcomes in different chemotherapeutic subgroups, as well as toxicity, to make the result more comprehensive and convincing. However, several limitations in this analysis should still be considered. Firstly, some endpoints were not evaluated in all of the included randomised controlled trials. Secondly, heterogeneity of certain endpoints in some subgroups might limit the comparability between studies and affect the validity, despite the use of randomized models.

In conclusion, we compared the outcomes of different combination therapies with gemcitabine monotherapy, including ORR, OS, PFS and major toxicity. The results showed that gemcitabine combination therapy leads to a modest improvement in survival, though with more toxicity reported compared with gemcitabine monotherapy.

COMMENTS

Background

Pancreatic cancer is a devastating disease with a dismal prognosis. The majority of patients (80%-90%) have unresectable disease at diagnosis and have to rely on palliative chemotherapy alone. In this situation, the goals of chemotherapy should be to control tumor progression and improve the survival rate with minimal toxicity.

Research frontiers

Gemcitabine has long been the first-line chemotherapeutic agent in pancreatic cancer. However, because of high levels of intrinsic and acquired chemoresistance, a large number of patients do not respond to the drug. In an attempt to improve clinical efficacy, systemically administered gemcitabine has been combined with a second cytotoxic agent, such as platinum analogs, fluoropyrimidine, or a targeted cytotoxic agent. However, the added value of these additive agents is unknown.

Innovations and breakthroughs

The present meta-analysis included 26 high-quality randomized trials comparing gemcitabine monotherapy with gemcitabine combination therapy. The primary outcome variables evaluated were objective response rate, overall survival, progression-free survival and major toxicity. The results showed that gemcitabine combination therapy was associated with only a modest improvement in survival, though with more toxicity reported, compared with gemcitabine monotherapy.

Applications

The results of this study imply that current treatments for unresectable pancreatic cancer are not sufficiently effective. Further research concerning the mechanisms involved in drug resistance in pancreatic cancer could discover ways of overcoming chemoresistance and lead to novel types of therapy.

Peer review

This is a meta-analysis concerning combination chemotherapy with Gemcitabine unresectable pancreatic cancer. The manuscript is well written and represents elaborative work.