Introduction

Oxaliplatin, or trans-l-dach (1R, 2R-diaminocyclohexane) oxalatoplatinum (l-OHP), is a third-generation platinum analog similar to carboplatin and cisplatin [1]. In 1976, this drug was discovered in the Nagoya City University by Yoshinori Kidani, and subsequently it was developed as drug for colorectal cancer treatment [2]. The therapeutic spectrum of oxaliplatin has been extended to other malignancies such as gastric [3], pancreatic [4] and non-small cell lung cancers [5].

Peripheral neuropathy is one of the most frequent and clinically relevant adverse events associated with the use of oxaliplatin [6]. Although there are several approaches to reduce oxaliplatin-induced neuropathy such as stop-and-go strategy, calcium and magnesium infusions, venlafaxine [7], using neuroprotective agents [6], no well-accepted therapy to prevent oxaliplatin-induced neuropathy has been identified [8, 9].

In recent years, the pharmacogenomic approach has been proposed for the identification of people at high risk of development of chemotherapy-induced peripheral neurotoxicity [10]. Genetic variations in drug-targeted genes, or polymorphisms of genes encoding DNA repair enzymes, metabolism, and detoxification pathways may influence the neurotoxicity of oxaliplatin. Some studies have discovered the connection between genetic polymorphisms and oxaliplatin-induced neuropathy, such as ERCC1 C118T [11], GSTP1 Ile105Val [12], AXGT haplotype [13], SCN2A R19K polymorphism [14], ITGB3 L33P [15], polymorphic CAG motif of the SK3 gene [16], ABCC1/2 [17], whereas other polymorphisms such as GSTM1 deletion, ABCB1 Ser893Ala/Thr were not found to be associated with oxaliplatin-induced neuropathy [10]. Up to now, the most studied polymorphism related with oxaliplatin-induced neuropathy is GSTP1 Ile105Val.

The glutathione-S-transferases (GSTs), a superfamily of dimeric phase II metabolic enzymes, play an important role in the cellular defense system. The GST superfamily are divided into six major classes (α, μ, π, θ, ζ and ω) [18]. These enzymes contribute to the inactivation of various toxic compounds (unsaturated aldehydes, quinines, epoxides and hydroperoxides) by forming secondary metabolites during oxidative stress. Cytosolic human GST exhibits genetic polymorphisms which can increase the susceptibility to carcinogenesis and inflammatory diseases [19, 20]. Glutathione-S-transferase pi 1(GSTP1) is one member of GST family, known to strongly affect human’s susceptibility to several cancers and metabolite detoxification [21].

The gene coding for GSTP1 is located on chromosome 11q13, and the most studied polymorphism of GSTP1 is an A>G substitution in position 313 in exon 5 that gives rise to missense substitution Ile105Val (GSTP1*B alleles, 313A>G, I105V, rs1695) [22]. It is speculated that GSTP1 Ile105Val polymorphism displays a significantly lower enzyme activity, leading to less effective capability of detoxification [23]. Over the past two decades, a number of studies were conducted to investigate the relationship between GSTP1 codon 105 polymorphism and oxaliplatin-induced neuropathy [11–13, 17, 24–33]. However, these studies have reported conflicting results. For example, significant associations between the GSTP1 genotype and oxaliplatin-induced neuropathy were observed in several studies [11, 12, 24, 25, 27, 28, 31, 33], whereas no statistically significant association was found in other studies [12, 13, 17, 26, 29, 30, 32]. This systematic review and meta-analysis was designed to assess existing evidences concerning the clinical effectiveness of GSTP1 Ile105Val and neurotoxicity induced by oxaliplatin in cancer chemotherapy.

Materials and methods

Search strategy and selection criteria

We performed electronic searches of the English-language literatures on GSTP1 polymorphisms and oxaliplatin-induced neuropathy in PubMed, EMBASE and The Cochrane Library using the combined text words “GSTP1” and neuropathy-related words (e.g. neuropathy or neurotoxicity). The latest search was undertaken in December 2012. We also manually screened the reference lists of the retrieved articles to identify other relevant publications.

The including criteria were the followings: (1) they must be clinical researches; (2) the exposure of interest included GSTP1 Ile105Val; (3) the outcome of interest included oxaliplatin-induced neurotoxicity; and (4) the studies provided the numbers of patients with and without neurotoxicity in different genotypes. Studies that were published as reviews or involved children (<18 years) were excluded from analysis.

Based on clinical relevance, the primary endpoint was defined as the incidence of grade ≥2 neurotoxicity in OSS or grade ≥3 neurotoxicity in NCI-CTC (in OSS, grade 2 indicates moderate motor symptoms and sensory symptoms extended to ankle and wrist, and grade 3 indicates motor symptoms requiring help/assistance and sensory symptoms extended to knee and elbow; in NCI-CTC, grade ≥3 indicates sensory alteration or paresthesia interfering with activity of daily living (ADL), disabling or death).

Data extraction

Two investigators (Peng and Wang) independently did the search and data extraction. Any discrepancies were resolved by discussion. The following data elements were extracted from each study: first author, year of publication, racial descent, design of the study, source of population, sample size, mean age and gender percentage of cases and controls, type of tumors, chemotherapy regiments, oxaliplatin dosages, neuropathy criteria, genotype frequencies and genotyping methods.

Statistical analysis

In view of unknowing exact genetic model for the polymorphism, we examined contrasts for dominant model (assuming heterozygotes have the same increased risk as minor homozygous genotypes), as well as recessive model (only minor homozygous genotype having the effect) and allele frequencies (the minor allele frequency vs. the major allele frequency). The effect of associations was estimated as odds ratio (OR) with the corresponding 95 % confidence intervals (CI). Firstly, fixed-effect model was used for calculating pooled ORs. If there was significant heterogeneity across studies, random-effect model was selected. The existence of heterogeneity between studies was evaluated using the Dersimonian and Laird’s Q test [34]. I 2 was used to quantify heterogeneity; this measure describes the percentage of the observed between-study variability attributable to heterogeneity rather than chance. I 2 takes values between 0 and 100 %. Important heterogeneity was defined while I 2 > 50 % [35].

We used sensitivity analysis in which each study was excluded once at a time to examine the influence of one study on the overall summary estimate. In order to investigate the source of heterogeneity, studies were subdivided by ethnicity (Caucasian, including Caucasian of European origin; East Asian, including Chinese, Japanese, Korean), dosage (≤85 mg/m2 or >85 mg/m2) and the criteria for evaluating the oxaliplatin-induced toxicity (NCI-CTC or OSS) for subgroup analysis. We also used meta-regression to examine the effect of study-level predictor variables such as study design and mean age of study population on the effect size.

Publication bias was evaluated using inverted funnel plot and Egger’s test [36]. All analyses were carried out using Stata software (version 9.0). All P values were two-sided and the significance level was 0.05.

Results

Study characteristics

Three hundred and one studies were identified from online databases and the reference lists, of which 14 studies involving 2,191 patients (1,354 men and 837 women) were in accordance with the inclusive criteria and included in meta-analysis (Fig. 1). The characteristics of the studies are listed in Table 1. Individuals involved in these 14 studies received oxaliplatin-based chemotherapy regimens in terms of FOLFOX4, FOLFOX6, mFOLFOX6, FLOFOX7, CAPOX, IROX, oxaliplatin plus S1, FLO or FLP. All of these patients had not received oxaliplatin or other platinum agents previously. Sample size ranged from 51 to 520. Five trials were conducted in Asians [11, 24, 25, 27, 29] and the other 9 studies in Caucasians [12, 13, 17, 26, 28, 30–33]. Twelve trials were designed to be prospective and the other two studies to be retrospective [11, 12]. The mean age of patients ranged from 55 to 68. The patients of ten trials were mCRC, and the others were patients with advanced gastric cancer [24, 31], gastrointestinal solid tumors [12] or advanced carcinomas [13]. Single chemotherapy regimen was applied in 12 trials while more than 3 regimens in the other 2 trials [12, 28]. Oxaliplatin dosage included 85 mg/m2/day biweekly, 100 mg/m2/day biweekly, 130 mg/m2/day biweekly and 130 mg/m2/day every 3 weeks. The patients of 12 trials received the same oxaliplatin dose, while the dose varied among patients in the other one trial [12]. Eight trials [11, 13, 25, 27, 28, 30–32] applied National Cancer Institute Common Toxicity Criteria (NCI-CTC) to evaluate the neurotoxicity, while other six studies [12, 17, 24, 26, 29, 33] depending on oxaliplatin-specific scale (OSS).

Fig. 1
figure 1

Study selection process for the meta-analysis of GSTP1 Ile105Val polymorphism and oxaliplatin-induced neuropathy

Full size image
Table 1 Main characteristics of clinical trials included in meta-analysis
Full size table

Two oxaliplatin-containing regimens in McLeod’s [28] and Boige’s [32] studies were separated into 2 independent trials for analysis. We also analyzed the patients experiencing acute and chronic neuropathy as two separated samples in Gamelin’s study [13].

Quality assessment

The use of quality scoring system in meta-analyses of observational studies is controversial [37]. We did not assign a single grade or score to represent the quality of a study. Instead, we focused on certain items that were reflective of methodological and reporting quality of the studies. Study year, country, race, study design, source of population, number of patients, age, type of tumors, chemotherapy regimens, oxaliplatin dose/schedule and grade criteria were considered in our evaluation of the quality of an included study (Table 1).

The role of GSTP1 polymorphism on oxaliplatin-induced neuropathy

GSTP1 AA or AG versus GG

All of 14 included trials reported the relevant data for the comparison of the risk of severe oxaliplatin-induced neuropathy between patients with GSTP1 AA or AG genotype and those with GSTP1 GG genotype. We did not find significant associations between GSTP1 G allele and the risk of oxaliplatin-induced neuropathy in dominant model (OR = 1.08, 95 %CI 0.67–1.74, P = 0.754). Statistical significance of heterogeneity was detected across all studies (I 2 = 67.9 %, P < 0.05). We carried out influence analyses to explore the source of heterogeneity. The results were not significantly affected by the removal of any study (supplement Fig. 1), and we failed to find out the source of heterogeneity after subgroup analysis where the studies were stratified by study characteristics, such as ethnicity, dosage of oxaliplatin, the grade criteria (Table 2) and meta-regression analysis (data not shown) (Fig. 2).

Table 2 Effect of GSTP1 Ile105Val polymorphism on oxaliplatin-induced neuropathy by prespecified study characteristics in different genetic model
Full size table
Fig. 2
figure 2

Meta-analysis for the relationship between GSTP1 Ile105Val polymorphism and oxaliplatin-induced neuropathy risk in dominant model. Dominant model (AA or AG vs. GG). Year represents publish year. The solid squares represent odds ratios (ORs) from individual studies; the diamonds are shown as overall effect

Full size image

GSTP1 AA versus AG or GG

There were seven clinical trials for these analyses. The risk of severe neuropathy was not significantly different between patients with GSTP1 AA genotype and those with AG or GG genotype (OR = 1.67, 95 %CI 0.56–4.93, P = 0.357). There was significant heterogeneity between studies (I 2 = 59.7 %, P < 0.05) (Fig. 3). However, no statistical heterogeneity was detected in the patients received >85 mg/m2 dosage of oxaliplatin, although the risk was similar to the original result (OR = 0.58, 95 %CI 0.12–2.72, P = 0.951) (Table 2).

Fig. 3
figure 3

Meta-analysis for the relationship between GSTP1 Ile105Val polymorphism and oxaliplatin-induced neuropathy risk in recessive model. Recessive model (AA vs. AG or GG). Year represents publish year. The solid squares represent odds ratios (ORs) from individual studies; the diamonds are shown as overall effect

Full size image

GSTP1 A allele versus G allele

We compared the risk of oxaliplatin-induced neuropathy between GSTP1 A allele and G allele in seven trials including 1,043 patients. There were also no significant differences between the two groups (OR = 1.22, 95 %CI 0.67–2.24, P = 0.513). Heterogeneity was statistically significant across studies (I 2 = 73.6 %, P < 0.05). No statistical heterogeneity was detected in the subgroup of studies using NCI-CTC criteria, and the result did not change materially (OR = 1.12, 95 %CI 0.71–1.76, P = 0.636) (Table 2; Fig. 4).

Fig. 4
figure 4

Meta-analysis for the relationship between GSTP1 Ile105Val polymorphism and oxaliplatin-induced neuropathy risk in allelic analysis. Allele model (A allele vs. G allele). Year represents publish year. The solid squares represent odds ratios (ORs) from individual studies; the diamonds are displayed as overall effect

Full size image

Publication bias

For publication bias estimating, we did not observe visually or statistically significant asymmetry according to the inverted funnel plot (supplement Fig. 2) and Egger’s test in all analyses (data not shown).

Discussion

Present study is the first meta-analysis examining the effect of GSTP1 Ile105Val polymorphism on the risk of oxaliplatin-induced neuropathy. Fourteen articles including 2,191 patients were used for final analysis. There was no evidence supporting the hypothesis that the GSTP1 Ile105Val polymorphism is significantly associated with oxaliplatin-induced neuropathy.

GSTP1 is directly involved in the detoxification of cisplatin by the formation of cisplatin–glutathione adducts, indicating that GSTP1 plays a role in the acquisition of resistance to this platinum compound [38, 39]. Amino acid 105 of GSTP1 lies in close proximity to the active center and may directly influence GSTP1’s catalytic activity. The Val allele causes a variant GSTP1 protein with a lower enzymatic capacity for the conjugation of various cytotoxic drugs as compared to the wild-type Ile allele [22, 40]. However, the precise mechanism has not been well clarified yet. Kweekel et al. [30] summarized two theories postulated to explain the association between the GSTP1 Ile105Val polymorphism and oxaliplatin-induced neuropathy. Firstly, previous studies indicated that the Ile105Val polymorphism decreased the substrate affinity of the GSTP1 enzyme [22]. The protective effect of GSTP1*G allele on platinum may result in a lower incidence of neurotoxicity in Val/Val carriers [23]. Secondly, GSTP1*A reduces cellular proliferation and protects against apoptosis through a JNK-independent mechanism. In contrast, GSTP1*G did not influence cellular proliferation but protected cells from apoptosis through JNK-mediated mechanisms, which are involved in protecting the cells from platinum-induced toxicity [41]. In addition, Mir et al. [42] discovered a significant correlation between GSTP1 105Ile/105Ile genotype and the occurrence of grade ≥2 docetaxel-induced peripheral neuropathy (DIPN). These findings strongly suggested 105Ile/105Ile variants playing a role of oxidative stress in the pathophysiology of DIPN. However, which signaling pathway or additional pathway is more important for the pathophysiologic effects is still not well known. And the results of our study also did not contribute to explaining the pathophysiologic effects of GSTP1 Ile105Val polymorphism and oxaliplatin-induced neuropathy. It might be explained by the fact that the effect of a gene polymorphism depends on its complicated genetic background, and expression of compensatory functions of other genes or proteins could mask the effect of one GSTP1 mutation [21].

The difference of criteria used to evaluate oxaliplatin-induced neuropathy may partly account for the result. There is a wide discrepancy in the literature on how to measure and grade oxaliplatin-induced neuropathy, such as NCI-CTC, Total Neuropathy Score (TNS), Eastern Cooperative Oncology Group (ECOG) toxicity criteria, OSS and criteria from individual studies or World Health Organization [6]. Most clinical trials evaluated oxaliplatin-induced neuropathy based on NCI-CTC or OSS criteria. However, after subgroup analysis by different criteria, no significant association between GSTP1 Ile105Val polymorphism and oxaliplatin-induced neuropathy was observed in both NCI-CTC and OSS groups (Fig. 4).

Some limitations of this meta-analysis must be considered. Firstly, extreme heterogeneity was still observed between studies although we performed subgroup analysis stratified by study characteristics, such as ethnicity, dosage of oxaliplatin, the grade criteria and meta-regression analyses. Since the limited detailed presented data in the included studies, other potential confounding effects such as selection bias, population stratification, population-specific gene–gene or gene–environment interaction, genotyping errors or chance that might be contributable to the source of heterogeneity across studies were not well examined [43]. Secondly, oxaliplatin-induced neuropathy is usually cumulative. More cycles of chemotherapy and more doses of oxaliplatin might account for a higher incidence of severe neuropathy after treatment with oxaliplatin. Chen et al. [25] found that the statistically significant only shows after 12 cycles but not after 4 or 8 cycles. Whereas due to limited data, we were unable to further examine the effect of cycles of chemotherapy on the association between GSTP1 and neuropathy. Thirdly, oxaliplatin dosage was an important factor to be considered. Although study-level dosage of oxaliplatin was extracted from each study and was used for subgroup and meta-regression analyses, individual-level was more influential on the relationship, which was not available. Fourthly, the neurotoxicity induced by oxaliplatin can manifest as two distinct syndromes: a transient, acute syndrome that can appear during or shortly after infusion, and a dose-limiting, cumulative sensory neuropathy [44]. In this meta-analysis, there was only one study that separated into two samples based on acute or chronic neuropathy. However, the result did not change materially after we included the patients with acute neuropathy or not (data not shown).

In conclusion, this is the first meta-analysis to examine the effect of GSTP1 Ile105Val polymorphism on the risk of oxaliplatin-induced neuropathy. No significant association between the GSTP1 Ile105Val polymorphism and the risk of oxaliplatin-induced neuropathy was identified in the present study. Given the limited number of studies and potential bias, large-scale and well-designed, controlled clinical trials will be required to confirm these hypotheses.