The first 2 authors contributed equally to this article.
Article first published online: 9 NOV 2011
Copyright © 2011 American Cancer Society
Volume 118, Issue 14, pages 3587–3598, 15 July 2012
How to Cite
Zhao, X., Wang, X., Wu, W., Gao, Z., Wu, J., Garfield, D. H., Wang, H., Wang, J., Qian, J., Li, H., Jin, L., Li, Q., Han, B., Lu, D. and Bai, C. (2012), Matrix metalloproteinase-2 polymorphisms and clinical outcome of Chinese patients with nonsmall cell lung cancer treated with first-line, platinum-based chemotherapy. Cancer, 118: 3587–3598. doi: 10.1002/cncr.26669
We thank Jun Yang and Datong Zheng for their help with sample collection, Shaohua Gu and Qihan Wu for their helpful comments and discussion, our volunteers for donating their blood, and our collaborators for collection of blood samples and information.
- Issue published online: 2 JUL 2012
- Article first published online: 9 NOV 2011
- Manuscript Accepted: 3 OCT 2011
- Manuscript Revised: 1 OCT 2011
- Manuscript Received: 16 AUG 2011
- matrix metalloproteinase-2;
- platinum-based chemotherapy;
- clinical outcome;
- nonsmall cell lung cancer
Matrix metalloproteinase-2 (MMP-2) is well known for its critical role in cell survival and cancer development. It also plays an important role in hematopoietic recovery after chemotherapy-induced myelosuppression. In this study, the authors investigated the association of MMP-2 polymorphisms with treatment efficacy and the occurrence of severe toxicity in patients with nonsmall cell lung cancer (NSCLC) who were receiving first-line, platinum-based chemotherapy.
A pharmacogenetic association study was performed in 663 Chinese patients who had inoperable stage III/IV NSCLC and were receiving first-line, platinum-based regimens. Information about objective response, progression-free survival, overall survival, grade 3 or 4 gastrointestinal toxicity (nausea/vomiting), and hematologic toxicity (neutropenia, anemia, thrombocytopenia) was available. Sixteen tag single nucleotide polymorphisms (SNPs) of MMP-2 were assessed.
In 7 polymorphisms, significant associations were observed with the incidence of grade 3 or 4 neutropenia. The variant homozygotes of reference SNP rs12934241 exhibited the most significant effect on the risk of neutropenia, leading to an incidence rate that increased from 12.3% (for the C/C genotype) to 50% (for the T/T genotype; odds ratio, 8.33; P = 8.8 × 10−5). Stratified analyses indicated that rs12934241 exhibited a much stronger influence in the cisplatin-gemcitabine regimen subgroup than subgroups that received other regimens (Pinteraction = .003). Further haplotype analyses produced results that were consistent with results from single-SNP analyses. However, no significant association was observed between MMP-2 polymorphisms and treatment efficacy, including response rate, clinical benefit, progression-free survival, and overall survival.
To the authors' knowledge, this study provides the first evidence for a predictive role of MMP-2 polymorphisms in the variability of severe chemotherapy-related neutropenia among Chinese patients with platinum-treated, advanced NSCLC. Cancer 2012;3587–3598. © 2011 American Cancer Society.
Platinum (cisplatin or carboplatin), in combination with navelbine, gemcitabine, or paclitaxel, has been widely used as first-line chemotherapy for advanced nonsmall cell lung cancer (NSCLC).1 These platinum-based regimens bring modest benefits but also adverse effects, which may cause distressing symptoms and prevent further therapies. Studies have suggested that efficacy and toxicity of platinum-based chemotherapy vary greatly between individuals.2, 3 Thus, the identification of predictive markers for optimal individualized therapy with better efficacy and minimal toxicity remains a continuing challenge in NSCLC treatment.
Recent clinical studies suggest a large potential for genomic findings in predicting treatment outcomes.4-6 Matrix metalloproteinase-2 (MMP-2) (also known as gelatinase A), the main physiologic mediator of extracellular matrix (ECM) degradation,7 which is important in cell proliferation, apoptosis, angiogenesis, and hematopoiesis,8-10 is playing an increasingly critical role in pharmacogenetic research of NSCLC treatment.11 Accumulating evidence has demonstrated that MMP-2 exhibits a high level of expression in many human tumors and plays an important role in cancer initiation and development.12 In lung cancer, activated MMP-2 can confer apoptosis resistance by modulating Fas-mediated death signaling; down-regulating MMP-2; increasing caspase-3, caspase-8, and caspase-9 activities; degrading the BH3-interacting domain death agonist Bid; and releasing cytochrome c.8 MMP-2 also directly modulates the activity of several proliferation and apoptosis-related growth factors, such as insulin-like growth factor-1, epidermal growth factor, and fibroblast growth factors.13-16 In addition, functional and mechanistic studies have demonstrated that MMP-2 plays an important role in hematopoietic recovery after chemotherapy-induced myelosuppression not only by promoting migration of hematopoietic stem cells through the ECM but also by facilitating the release of pivotal regulatory factors (eg, tumor necrosis factor-alpha, insulin-like growth factor, and transforming growth factor-beta) into the bone marrow microenvironment.10, 17-20 This suggests a potential impact of MMP-2 on the incidence and severity of drug-related hematologic toxicity.
Several studies have suggested that MMP-2 polymorphisms are associated significantly with lung cancer risk.14, 21 These functional differences between polymorphic variants can strongly influence gene transcriptional activity and function.22 Other reports have demonstrated that MMP-2 expression is associated significantly with tumor metastasis and a poor prognosis in patients with NSCLC.23, 24 However, in epidemiologic studies of polymorphisms, the issue has remained controversial.25-30 It is noteworthy that there are no reports concerning the correlation between MMP-2 genetic variants and the toxicity of cancer chemotherapy. In the current study, we investigated 16 tag-single nucleotide polymorphisms (SNPs) in the MMP-2 gene and assessed the association of the polymorphisms with grade 3 or 4 toxicities and treatment efficacy in 663 Chinese patients with stage III/IV NSCLC who received first-line platinum-based chemotherapy. The objective of this work was to explore the potential impact of MMP-2 variants on the treatment of advanced NSCLC.
MATERIALS AND METHODS
Study Design and Patient Recruitment
All patients enrolled in the study were Chinese living in China and had histologic stage IIIA through IV NSCLC diagnosed at the Shanghai Chest Hospital, Shanghai Zhongshan Hospital, and Shanghai Changhai Hospital between March 2005 and January 2010. The following criteria were used to identify eligible patients3: 1) informed consent available and adherence to the treatment schedule; 2) age ≥18 years; 3) histologically confirmed, stage IIIA through IV NSCLC with the presence of a measurable and evaluable lesion; 4) receiving first-line, platinum-based chemotherapy (no prior surgery, radiotherapy, or concurrent chemoradiotherapy for this cancer); 5) no other prior malignancy in the past 5 years except treated nonmelanoma skin cancer, treated cervical carcinoma in situ, or any other “cured” malignant tumor; 6) an Eastern Cooperative Oncology Group performance status between 0 and 2; 7) absolute neutrophil count ≥1.5 × 109/L, platelet count ≥100 × 109/L, serum creatinine ≤1.5 times the upper limit of normal, creatinine clearance ≥60 mL/minute, and alanine and aspartate aminotransferase levels ≤1.5 times the upper limit of normal; 8) no active congestive heart failure or cardiac arrhythmias requiring medical regimen; 9) no uncontrolled clinical infections; and 10) no other critical medical or psychological factors that might influence the treatment schedule.
Clinical data for all patients were recorded systematically at entry, including sex, age at diagnosis, smoking status, clinical stage, and tumor histology. Medical history interview, physical examination, and laboratory investigations were done before any treatment course was started. The complete medical record (including progress notes of the treating oncologist and nurses) was available and reviewed. Survival statistics were collected from several sources, including follow-up calls, the Social Security Death Index, and inpatient and outpatient clinical medical records. The study protocol was approved by the Ethical Review Committee of Fudan University and the participating Hospitals. Investigators were blinded to the genotype status of the patients.
Chemotherapy toxicity was assessed twice weekly according to the National Cancer Institute's Common Toxicity Criteria, version 3.0. The worst toxicity record during the initial 2 cycles of therapy was collected for analysis. Toxicities included neutropenia, anemia, thrombocytopenia, nausea, and vomiting. Severe hematologic toxicity included grade 3 or 4 neutropenia, anemia, or thrombocytopenia. Severe gastrointestinal toxicity included grade 3 or 4 nausea or vomiting. No grade 5 toxicity (death) was observed in this study.
Patient responses to the treatment were evaluated after the first 2 cycles of chemotherapy according to Response Evaluation Criteria in Solid Tumors (version 1.0) guidelines, which classify the responses into 4 categories: complete response (CR), partial response (PR), stable disease (SD), and progressive disease (PD).3 The response rate corresponds to the percentage of patients with CR or PR, whereas clinical benefit (CR + PR + SD) represents the absence of PD.31 Progression-free survival (PFS) was calculated from the date the patient started chemotherapy to the date of disease progression or death (whichever occurred first) or the last progression-free follow-up. Overall survival (OS) calculated from the date of first chemotherapy to the date of death or last follow-up.
All patients enrolled in this study were considered inoperable and, thus, received first-line platinum-based chemotherapy as follows: either cisplatin 75 mg/m2 or carboplatin at an area under the curve of 5, both administered on day 1 every 3 weeks, in combination with navelbine 25 mg/m2 on days 1 and 8 every 3 weeks, or gemcitabine 1250 mg/m2 on days 1 and 8 every 3 weeks, or paclitaxel, 175 mg/m2 on day 1 every 3 weeks, or docetaxel 75 mg/m2, on day 1 every 3 weeks. A few patients received other platinum-based treatment. All chemotherapeutic drugs were administered intravenously, and all included patients were treated for 2 to 6 cycles.
Matrix Metalloproteinase-2 Single-Nucleotide Polymorphism Selection and Genotyping
The human MMP-2 gene is approximately 27.52 kb in size and is located on chromosome 16q13-q21. Genotype data on the MMP-2 gene region (including 2 kb upstream) from the Han Chinese in Beijing (CHB) population were downloaded from the phase 2 HapMap SNP database (available at http://www.hapmap.org/ [access August, 2011]), and tag-SNPs were selected in Haploview software (available at http://www.broadinstitute.org/haploview; [access August, 2011]) using a minor allele frequency (MAF) cutoff of 0.05 and a correlation coefficient (r2) threshold of 0.8. In this study, a total of 16 tag-SNPs were selected to represent genetic variation of 37 SNPs in MMP-2.
Blood samples were obtained in ethylene diamine tetracetic acid tubes from patients at the time of recruitment. Genomic DNA was extracted using the QIAamp DNA Maxi Kit (Qiagen GmbH, Hilden, Germany). All SNPs selected were genotyped using iSelect HD BeadChip (Illumina, San Diego, Calif) with the following quality-control criteria: genotyping call rate of SNP, >0.95; MAF, >0.05; and GenCall score, >0.2. All SNPs were genotyped successfully, except 3 probes that failed in the bead chip synthesis process. Therefore, a polymerase chain reaction-SNaPshot method was used to supplement the genotyping data from the 3 failed polymorphisms (reference SNPs [rs] rs243844, rs243849, and rs9928731). All selected SNPs had call rates >0.95. Concordance between replicates was >99.9%. GenomeStudioV2010.1 (Illumina) and GeneMap-per v 4.0 (Applied Biosystems) software were used to analyze the data and prepare reports.
The analyses were specified to investigate the relations between MMP-2 polymorphisms and grade 3 or 4 toxicity and the clinical efficacy of platinum-based chemotherapy (response rate [CR + PR], clinical benefit [CR + PR + SD], PFS, and OS). Demographic, clinical, and pathologic characteristics were tested univariately against outcomes and genotypes using chi-square tests or log-rank tests, depending on which variables were analyzed. Association of each genetic variant with dichotomous outcomes were estimated by odds ratios (ORs) and their 95% confidence intervals (CIs) using univariate logistic regression, assuming an additive model (zero, 1, or 2 copies of variant alleles). Polymorphisms that had P values < .05 in univariate analysis were selected for multivariate logistic regression analysis in different genetic models (codominant, additive, dominant, recessive), with adjustment of patient characteristics (sex, age, Eastern Cooperative Oncology Group performance status, TNM stage, histologic type, smoking status, and chemotherapy regimens) to yield P < .1 in univariate analysis. Further stratification analyses for the significant polymorphisms were then performed, and differences in outcomes between characteristic subgroups according to genetic variants were investigated by calculating P values for interaction. The association between genotype and survival was examined by hazard ratios (HRs) with 95% CIs using univariate a Cox proportional hazards regression model. Polymorphisms with P < .05 in univariate analysis were selected for inclusion in the multivariate analysis. For statistical analyses, the SPSS statistical software package was used (version 15.0; SPSS Inc., Chicago, Ill). Pairwise linkage disequilibrium (LD) among the SNPs was examined using D′ and r2. Haplotype blocks were defined according to Gabriel et al in Haploview software (Broad Institute of Harvard and Massachusetts Institute of Technology, Cambridge, Mass).32 Individual haplotype frequencies were estimated based on the Bayesian algorithm using the PHASE 2.0: http://www.bioinf.manchester.ac.uk/resources/phase/ program (version 2.0.2).33 Haplotype-outcome associations were evaluated for each block using HAPLO.STATS package in R-plus (available at http://cran.r-project.org/web/packages/haplo.stats/index.html; [access August, 2011]).34 All P values reported were 2-sided. A P value < .05 was considered statistically significant, whereas a level of P < .1 was considered marginal. To protect results against multiple comparisons, Bonferroni correction was made for the P value from the results of any SNP by multiplying the number of SNPs tested for the gene.2
All 663 patients in our study had stage III through IV NSCLC and received 2 to 6 cycles of first-line, platinum-based chemotherapy except for 5 patients who died of disease progression before treatment was completed. The main patient characteristics and clinical outcomes are summarized in Table 1, and the correlations between variables were tested univariately using chi-square tests or log-rank tests, depending on the tested variables (data not shown). Of the 663 patients enrolled, hematologic toxicity was evaluated in 646 patients, gastrointestinal toxicity was evaluated in 642 patients, and the objective response rate was assessed in 657 patients. A few patients were not included because of loss to follow-up during first-line chemotherapy. For the survival analysis, at the time of final data collection (March 2011), the median follow-up was 32.0 months. The median PFS was 6.5 months, and the median survival was 18.4 months, similar to values in the literature.35 All 16 tag-SNPs involved were in Hardy-Weinberg equilibrium (P > .05), and the observed allele frequency was consistent with values in the published literature (Table 2). No significant associations between polymorphisms and patient characteristics were observed (data not shown).
|Characteristic||Total No.||No. of Patients (%)|
|Median age [range], y||58 [26-80]|
|Never smokers||277 (41.8)|
|Ever smokers||386 (58.2)|
|Squamous cell||141 (21.3)|
|Other platinum combinations||39 (5.9)|
|Median time to outcome, mo||663|
|Grade 3 or 4 hematologic toxicity||646||166 (25.7)|
|Grade 3 or 4 gastrointestinal toxicity|
|Minor Allele Frequency|
|NCBI SNP ID||Location in Gene Region||Base Change||Genotyping Rate, %||P (HWE)||No. of SNPs Tagged||In Current Data||CHBa||JPTa||YRIa||CEUa|
Single-Nucleotide Polymorphism Analysis
A complete set of associations between each polymorphism and each severe toxicity was assessed using univariate logistic regression, assuming an additive model (data not shown). Seven of the 16 SNPs with P values < .05 in univariate analysis were examined further in multivariate logistic regression analysis, and all 7 retained significance (Table 3). We observed that those 7 SNPs had significant associations with grade 3 or 4 neutropenia (the variant homozygotes of rs1477017, rs17301608, rs12934241, rs1992116, and rs11639960 displayed a risk effect for neutropenia; whereas homozygous rs243847 and rs243844 exhibited a protective influence on neutropenia). Among these, the most significant correlation was observed between the variant homozygotes of rs12934241 and the risk of grade 3 or 4 neutropenia (incidence rate, 50% vs 12.3%; OR, 8.33; 95% CI, 2.89-24.01; P = 8.8 × 10−5 × 16 = .0014 < .05; remaining significant after Bonferroni correction). Thus, further stratification analyses between rs12934241 and severe neutropenia were performed for patient characteristic subgroups (Table 4). Because 100 patients were sufficient to investigate the associations between rs12934241 (MAF, 0.16) and grade 3 or 4 neutropenia with 90% power to detect a minimum OR of 1.83 at the 2-sided 5% level of significance, several subsets were selected for stratified examination (total, >100 patients): aged ≤58 years, aged >58 years, men, women, never-smokers, ever-smokers, stage IIIB or IV disease, histologic type of adenocarcinoma or squamous cell carcinoma, patients receiving cisplatin-navelbine, cisplatin-gemcitabine, or carboplatin-paclitaxel. The rs12934241 variant had a similar effect in all the other patient characteristic subgroups, whereas it exhibited a much stronger influence on severe neutropenia occurrence in the cisplatin-gemcitabine subset than the other 2 regimens (cisplatin-gemcitabine subgroup, 80% vs 3.3%; OR, 8.39; 95% CI, 2.78-25.29; P = 1.6 × 10−4), and with a statistically significant interaction was observed (compared between platinum plus gemcitabine or platinum plus other agents; P = .003).
|Outcome||Reference SNP No.||Genotype and Best Fitting Model||No./Overalla||%||ORb||95% CIb||Pb|
|T/T||8/16||50||8.33||2.89-24.01||8.8 × 10−5c|
|Recessive||8.08||2.83-23.06||9.5 × 10−5c|
|Cisplatin-gemcitabine||3/90||3.3||3/27||11.1||4/5||80||8.39||2.78-25.29||1.6 × 10−4c|
|Objective Response||Total No.||No. of Respondersa||ORb||95% CIb||Pb|
|Survival||Total No.||Median, mo||HRc||95% CIc||Pc|
We also observed significant associations of rs1992116 with severe hematologic toxicity (P = .017), neutropenia (P = .033), and anemia (P = .002). Further stratified analyses were not performed because of insufficient subgroup sample size. No significant association was observed between MMP-2 polymorphisms and grade 3 or 4 gastrointestinal toxicity or thrombocytopenia.
Associations between each MMP-2 polymorphism and treatment efficacy (response rate, clinical benefit, PFS, OS), were assessed using univariate logistic regression or Cox proportional hazards regression, assuming an additive model (data not shown). Although 2 polymorphisms had marginally significant correlations with PFS (rs865094, P = .091; rs243843, P = .071), there was no statistically significant association between MMP-2 polymorphisms and treatment efficacy. For rs12934241, the most significant variant in toxicity analysis, the variant homozygote had a longer PFS and OS than the wild-type homozygote, although neither PFS nor OS was statistically significant (PFS: HR, 0.92; P = .330; OS: HR, 0.94; P = .452) (Table 5).
Linkage Disequilibrium and Haplotype Analysis
Pairwise LD relations between MMP-2 polymorphisms are illustrated in Figure 1. The SNP rs12934241 was observed in high LD with rs1477017, rs17301608, and rs1992116 (0.993 < D′ < 1.000; 0.304 < r2 < 0.494), all of which were associated significantly with severe chemotherapy-related neutropenia. High LD also was observed for rs11639960, rs17301608, and rs1477017 (0.765 < D′ < 0.986; 0.404 < r2 < 0.687). In addition, rs243847 and rs243844, both of which demonstrated a protective effect on severe neutropenia, were correlated with each other (r2 = 0.416 with each other).
According to the method described by Gabriel et al, 3 blocks were defined for further analyses between haplotypes and severe neutropenia. In multivariate logistic regression analysis adjusted for type of treatment regimen, significant correlations of haplotypes with severe neutropenia were observed in all 3 blocks (Table 6). In the cisplatin-gemcitabine subset, a global score test revealed much stronger associations between haplotypes and the incidence of severe neutropenia in all the 3 blocks, especially Block 2 (global statistics, 24.4522, degrees of freedom, 4; P = 6.0 × 10−5; simulation P value [Psim] = 2.7 × 10−4), consistent with results from the single-locus analysis.
|Platinum Compounds||Cisplatin-Gemcitabine Combinations|
|Block||No.a||%||ORb||95%CIb||Pb||Global Score Test||No.a||%||ORc||95%CIc||Pc||Global Score Test|
|AA||54/531||10.2||1||Chi-square, 9.2442; df, 2; P = .0098; Psim = .0105d||6/96||6.3||1||Chi-square, 12.3995;df, 2; P = .0020; Psim = .0019d|
|CCCG||74/609||12.2||1||Chi-square, 8.9457; df, 4; P = .0625; Psim = .0667d||7/119||5.9||1||Chi-square, 24.4522; df, 4; P = 6.0 × 10−5; Psim = 2.7×10-4d|
|ACG||39/390||10||1||Chi-square, 7.4135; df, 3; P = .0598; Psim = .0615d||6/74||8.1||1||Chi-square, 13.0307; df, 3; P = .0046; Psim = .0041d|
Epidemiologic studies have indicated that sequence variations within MMP-2 may strongly influence the risk of several diseases, including lung cancer.14, 21 It was reported recently that rs1477017, rs17301608, and rs11639960 displayed associations with the risk of prostate cancer.36 More recently, it also was demonstrated that rs1992116, rs1477017, and rs17301608 strongly influence functional outcome after stroke.37 Several other reports demonstrated that 2 variants (rs243866 and rs243865) in the gene promoter region were associated significantly with the risk of both lung cancer and gastric cancer.14, 28 In addition, there is experimental evidence concerning the biologic functions of these 2 latter variants.22 However, pharmacogenetic studies related to MMP-2 polymorphisms are few; and, to our knowledge, no reports have focused on drug-related adverse events. In the current study, the tag-SNP selection strategy was chosen to assess a broader spectrum of MMP-2 polymorphisms and to characterize the effects of whole gene variants on chemotherapy-related outcomes, including both toxicity and treatment efficacy. The reportedly significant variants, rs243866 and rs243865, were not noted because of their low MAF in the Chinese population (< 0.05; National Center for Biotechnology Information [NCBI] dbSNP database; NCBI, Bethesda, Md).
In our study, significant associations were observed between MMP-2 polymorphisms and the occurrence of grade 3 or 4 toxicity. Single-SNP analysis identified 7 SNPs that were associated significantly with severe neutropenia, and rs12934241 had the strongest influence, leading to an incidence rate that increased from 12.3% (C/C genotype) to 50% (T/T genotype; (P = 8.8 × 10−5). Moreover, LD analysis displayed high levels of LD between rs12934241 and other significant SNPs, suggesting the important role of rs12934241. Consistent findings were exhibited in stratified analyses of both single SNPs and haplotypes. It is noteworthy that we observed a stronger influence of that polymorphism on severe neutropenia in the cisplatin-gemcitabine subgroup, and the interaction was statistically significant (P = .003). Our results suggest a more important role for MMP-2 in the cytotoxic mechanism of DNA-damaging agents (eg, gemcitabine) than other kinds of drugs, such as microtubule-targeted agents (eg, navelbine, paclitaxel).
To our knowledge, rs12934241, located in intron 5, has not been reported in previous association studies. Accumulating reports previously demonstrated that intronic polymorphisms located in splice donor, acceptor, or cis-regulatory element sites may perform an important role through modulating messenger RNA (mRNA)splicing and gene expression.38-40 However, we know of no experimental evidence supporting the same hypothesis for MMP-2 variants. Conversely, it is likely that rs12934241 is not a functional polymorphism but, rather, is a substitute for the underling causative variant in the same region. For example, we observed that the rs12934241 variant was in perfect LD (D′ = 1.00; r2 = 1.00) with a synonymous polymorphism (glycine [Gly]226Gly; rs1132896; phase 2 HapMap SNP database). Synonymous variants do not lead to altered coding sequences but could affect mRNA stability and protein translation efficiency by converting a high-usage codon to a low-usage codon.2, 41 It also has been confirmed that silent SNPs influence protein conformation and substrate affinity.42, 43 It is noteworthy that 1 study demonstrated the ability of chemotherapy to disturb hematopoietic cell proliferation and survival through disrupted translation of MMP-2,44 suggesting the importance of the MMP-2 translation process in chemotherapy-induced hematologic toxicity. Further studies will be required to identify the actual causative genetic variation. The findings that MMP-2 plays an important role in hematopoietic progenitor cell recruitment and bone marrow microenvironment modulation suggest that the genetic variant tagged by rs12934241 leads to inhibition of MMP-2 and the disruption of hematopoietic recovery after chemotherapy-related myelosuppression.
It is noteworthy that, although MMP-2 polymorphisms had significant associations with the incidence of adverse events, they were not correlated significantly with the efficacy of platinum-based treatment, suggesting a different mechanism of MMP-2 function between tumor cells and normal cells. A recent meta-analysis called into question the prognostic significance of MMP-2 expression in tumor cells and stromal cells in patients with NSCLC: MMP-2 expression in tumor cells was a poor prognostic factor, whereas this was not the case in stromal cells.23 In analyses of polymorphisms, several reports have suggested that a functional variant of MMP-2, although it demonstrated an association with cancer risk and had an impact on gene expression, failed to demonstrate a significant correlation with tumor metastasis and prognosis in patients with gastric cancer.14, 21, 22, 27-29 Considering the abnormally high level of MMP-2 expression in NSCLC tumor cells,12, 45 it has been suggested that MMP-2 expression and function may be modulated by some other factors in the background of carcinoma,46 and genetic variants may not play a major role in this situation.
Although the new finding of an association between MMP-2 polymorphisms and severe neutropenia is interesting, this study is limited by the lack of validation in an independent population. Another limitation is the insufficient sample size in the subgroup analyses. This may lead to a failure to detect a small effect from SNPs of low penetrance on a low toxicity incidence (ie, grade 3 or 4 anemia or thrombocytopenia). Several other important regimens (eg, a platinum with docetaxel) were not involved in subgroup analysis, also because of limited numbers. However, in terms of the stratified analysis of rs12934241 and severe neutropenia involved in this study, the sample size was large enough (for rs12934241, the MAF was 0.16; assuming n = 100 and α = .05, the analysis had 90% power to detect a minimum OR of 1.83 for grade 3 or 4 neutropenia).
In the current pharmacogenetic study, the sample size is large, and the recruitment criteria are strict. All 663 patients enrolled are Han Chinese and were treated in 3 medical centers that shared identical criteria for patient recruitment and clinical data collection, thereby limiting the potential confounding effects of genetic and clinical heterogeneity. The results focused mainly on neutropenia, which is an objective and comparable outcome. Therefore, these findings do not appear have been obtained by chance. Because the frequencies of MMP-2 polymorphisms are quite different among ethnic populations (eg, the frequency of the rs12934241 T allele is much lower in the Han Chinese population [CHB, 0.11] than in the Western Europe population [Utah residents with ancestry from northern and western Europe from the Center for the Study of Human Polymorphisms, 0.40] according to the HapMap database); and, considering the retrospective nature of this study, independent prospective validations in different populations will be required.
It is well known that severe chemotherapy-related neutropenia, frequently leading to serious infections, greatly influences patients' quality of life, further therapy, and survival. Thus, biomarkers for predicting neutropenia and risk, if validated, would be clinically significant. In the current study, we investigated the relation between MMP-2 polymorphisms and clinical outcomes in Han Chinese with NSCLC who received first-line, platinum-based chemotherapy and identified several significant genetic variants (especially rs12934241). However, there was no significant association with treatment efficacy, and the only significance was for the incidence of grade 3 or 4 chemotherapy-related toxicity, especially neutropenia. This finding may provide insight not only into the plausible role of MMP-2 in the pathogenesis and treatment of neutropenia but also into the predictive effect of the genetic variant on adverse events; thus, these findings potentially may have significant benefit for the individualized treatment of NSCLC in the future.
This work was partially supported by Shanghai Science and Technology Research Program (09JC1402200 and 10410709100), Natural Science Foundation of China (30800622 and 81001114) and Shanghai Leading Academic Discipline Project B111.
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.