| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Molecular Oncology, Markers, Clinical Correlates |
1 Inserm 490 Laboratoire de Toxicologie Moléculaire, Université René Descartes ParisV, Paris; 2 Inserm U525 Faculté de Médecine Hôpital Pitié-Salpétrière, Paris; 3 Service d’Oto-Rhino-Laryngologie et de Chirurgie Cervico-Faciale, Paris; and 4 Pôle d’oncologie et de spécialité Hôpital Européen Georges Pompidou Assistance Publique–Hôpitaux de Paris, Paris, France
ABSTRACT
Purpose: Treatment of head and neck cancer often associates different therapeutic modalities, including surgery, radiotherapy, and chemotherapy. In an attempt to optimize therapeutics, the identification of molecular markers linked to response to chemotherapy remains important. Recently, the involvement of metalloproteinases in resistance to chemotherapy was suggested through their interaction with the Fas/Fas ligand pathway. Indeed metalloproteinases enhance Fas ligand shedding modulating chemotherapy efficiency. On the basis of these findings, we tested the existence of a correlation between response to chemotherapy and four metalloproteinase polymorphisms in a prospective series of 148 head and neck cancer patients.
Experimental Design: Patients were genotyped using automated fragment analysis and 5'-nuclease allelic discrimination assay. Response to chemotherapy was clinically assessed without knowledge of the genotype status.
Results: A significant relation between the metalloproteinase type 3 (MMP3) –1612insA polymorphism and response to chemotherapy was identified. Indeed, patients with the 6A/6A genotype responded more frequently (86%) to treatment as compared with patients with the 5A/6A (65%) or 5A/5A (55%) genotypes (P = 0.04). A multivariate analysis, including tumor stage, gender, TP53 mutations, and MMP3 polymorphism, showed that the 6A/6A genotype was an independent factor of response to 5-fluorouracil-cisplatin chemotherapy in head and neck cancer patients with an odds ratio of 6.7 as compared with the 5A/5A genotype.
Conclusions: This work showed that genotyping the MMP3 gene enhancer polymorphism –1612insA could help predict chemosensitivity in head and neck cancer patients.
INTRODUCTION
Head and neck squamous cell carcinoma (HNSCC) is a disease associated with major morbidity and mortality and has a worldwide incidence of 500,000 cases/year (1) . Despite the frequent success of surgical and chemotherapeutic measures in controlling tumor growth, some primary tumors and most metastatic diseases are seldom curable and display resistance to chemotherapy. A different mechanism can lead to resistance to chemotherapy, including modification of drugs, pharmacokinetic properties, or expression of the multidrug resistance gene, ABCB1 (MDR1), but there is also evidence that the ability of cancer cells to evade apoptosis plays a major role in tumor survival (2) . Cisplatin and 5-fluorouracil (5FU) combination therapy showed objective response rates of 70% with complete responses ranging from 14 to 31% in different randomized trials (3) . This regimen considered as the standard in HNSCC treatment is of interest in terms of organ preservation strategies in responder patients (4) . Understanding the mechanisms linked to resistance to chemotherapy in HNSCC patients could help select patients that would benefit from neoadjuvant chemotherapy and organ preservation.
The anticancer action of drugs such as cisplatin and 5FU is largely attributed to their ability to induce programmed cell death (5)
. The integrity of signaling pathways that drive cell death is crucial for cells to commit apoptosis. Head and neck cancer cells demonstrate alterations in apoptosis pathways that allow them to override control cell death. For example, mutations in the TP53 gene, which has a pivotal role in apoptosis, have been reported to correlate with resistance to treatment in HNSCC (6
, 7)
. More recently, the involvement of the TNFRSF6/TNFSF6 [Fas/Fas ligand (FasL)] pathway in genotoxic agent-induced apoptosis has been reported previously (8)
. Indeed, FasL is up-regulated in cells treated with genotoxic agents, including cisplatin and 5FU, and could act as an autocrine/paracrine mediator of apoptosis. FasL is normally expressed in T lymphocytes, macrophages, and splenocytes but also in tumor cells, including HNSCC, in which its role in cytotoxic agent-induced cell death can therefore be hypothesized (9)
. In terms of resistance to treatment, it would be expected that the down-regulation of FasL, at the surface of HNSCC tumor cells, decreases the activity of anticancer drugs. Matrix metalloproteinases (MMPs) are a family of proteins with many biological functions, including extracellular matrix degradation, cell proliferation, and angiogenesis, that have been implicated at many levels in carcinogenesis (10)
. Recent evidence suggested that MMP7 (11)
and MMP3 (12)
could be implicated in the shedding of FasL from the cell surface. This results in the generation of soluble FasL with less potent activity in terms of triggering apoptosis by cross-linking with Fas (13)
. Moreover, MMP7 cleavage of FasL was related to resistance to doxorubicin in Ewing sarcoma and colon carcinoma cells, and blockade of MMP7 activity by MMP inhibitor resulted in sensitization to treatment, providing evidence for the role of MMPs and FasL in chemotherapy resistance (14)
. MMP3 (15)
and MMP7 (16)
are expressed in HNSCC and have been correlated to invasiveness. However, the importance of MMP expression in surrounding tumor cells could be of major influence on tumor growth. Indeed, MMP3 is essentially expressed by fibroblasts and macrophages. Different MMP expression patterns could be related to the existence of functional genetic polymorphisms in MMP promoter regions. Concerning MMP7, two polymorphisms were described, one at position –181A
G and the other at position –153CT. The two rare alleles conferred an increased promoter activity (17)
. For MMP3, the deletion or the insertion of an adenine at position –1612 (–1612insA) defined two alleles 5A or 6A. The 6A allele is associated with low transcription levels (18)
. Concerning the other MMP3 polymorphisms, six are in linkage disequilibrium with –1612insA and one (–A709G) is not of proven functional importance (19)
. In our study, only the –1612insA polymorphism was retained. Seven other MMPs map a cluster region of 400 kb at 11q22. In particular, the MMP1 gene, for which a functional polymorphism in the promoter region at position –1607 (–1607insG) gene, was found in linkage disequilibrium with the MMP3 –1612insA allele (20)
. The MMP1 polymorphism defines two alleles 1G or 2G. The presence of 2G creates an Ets site and results in higher transcription level (21)
. It is noteworthy that the haplotypes 2G-6A and 1G-5A are more frequently found than expected leading to a preferred association of the higher and the lower transcriptional activity alleles for MMP1 and MMP3 genes, respectively.
In this work, we focused on the relations between MMPs and chemotherapy and analyzed the correlation between response to 5FU-cisplatin neoadjuvant chemotherapy and MMPs (MMP1, MMP3, and MMP7) genotypes in a series of 148 head and neck cancer patients. To test whether MMP polymorphisms are independent risk factors of resistance to chemotherapy, a multivariate analysis was performed, including TP53 mutations.
PATIENTS AND METHODS
Patients.
Patients attending the Laennec Hospital in Paris with histologically proven head and neck squamous cell carcinoma, without previous history of cancer, multiple tumor locations, nor contraindication for cisplatin-based chemotherapy were eligible for entry onto this study. This study was in agreement with French laws and received the authorization of the local ethic committee (CCPPRB no. 96,017). A total of 148 patients (129 males, 19 females; mean age, 59 ± 5.5 years) was included in this study. Tumors were located in the oral cavity (n = 16), the oropharynx (n = 61), the hypopharynx (n = 40), and the endolarynx (n = 31). They were grouped according to the Tumor-Node-Metastasis classification and staged as recommended by the American Joint Committee on Cancer. Six tumors were T1, 53 were T2, 42 were T3, and 47 were T4 and 69 were N0 and 79 were n +. Three tumors were stage I, 31 were stage II, 38 were stage III, and 76 were stage IV. Tobacco consumption was available for 142 patients. Among them, 79 smoked >35 pack-years; 47, 15–35 pack-years, and 16 < 15 pack-years. All patients had neoadjuvant chemotherapy before surgery or radiotherapy that consisted of cisplatin (25 mg/m2/day) and 5FU (g/m2/day) delivered as a daily continuous i.v. dosage in 4-day courses. Three courses were repeated at 16–21-day intervals. Clinical response was assessed as defined by the Eastern Cooperative Oncology Group. Responder patients (R), n = 99, showed at least a 50% decrease in tumor size, and nonresponder patients (NR), n = 49, showed <50% decrease in tumor size. In this series, 26% of responder patients were complete responders (CR). All clinical data were reviewed by three of the authors (O. Laccourreye, H. Blons, and P.Laurent-Puig) without knowledge of patient genotypes.
Sample Collection.
Tumor samples and 10 ml of blood were collected at initial diagnosis during endoscopy under general anesthesia. Lymphocyte and DNA tumor samples were stored and extracted as described previously (22)
.
Genotyping.
MMP1 and MMP3 polymorphisms have been studied by genotyping lymphocyte DNAs using a PCR based approach. Amplifications were performed using 5'-CCCTCTTGAACTCACATGTTATG-6FAM-3' and 5'-ACTTTCCTCCCCTTATGGATTCC-3' for MMP1; 5'-TCCTCATATCAATGTGGCCAAA-3' and 5'-CGGCACCTGGCCTAAAGAC-6FAM-3' for MMP3. Briefly, 40 ng of genomic DNA were amplified by 0.5 units of HotStarTaq polymerase (Qiagen, Les Ulis, France), 200 nM deoxynucleoside triphosphates, 2 mM MgCl2, and 300 nM of each primer on a GeneAmp PCR system 9700 (Applied Biosystems, Courtaboeuf, France). Fragments were separated after dilution on an ABI 310 genetic analyzer (Applied Biosystems).
The two MMP7 polymorphisms were studied by 5'-nuclease allelic discrimination assay on an ABI 7900HT Sequence Detection System (Applied Biosystems). For both polymorphisms, amplifications were performed with forward 5'-AGTCAATTTATGCAGCAGACAGAAA-3' and reverse 5'-GTGTTATTTTTCATTAACTAAAACGAGGAA-3' primers. Specific probes for each allele were respectively labeled with the fluorescence reporter dyes VIC and FAM at their 5'-end: for MMP7 A/G, 5'-ACAATGTATTTGTCTTTC-3' and 5'-CAATGTATTCGTCTTTC-3'; and for MMP7 C/T, 5'-CTGCCAATAACGAT-3' and 5'-CTGCCAATAATGAT-3'. Briefly, reactions were performed in 10 µl comprising 40 ng of DNA, 1x mix of specific primers and probes, and 1X TaqMan Universal PCR Master Mix (Applied Biosystems). Data were analyzed with SDS2.0 software (Applied Biosystems).
Statistical Analysis.
The 
2 test was used to determine differences in the repartition of the genetic polymorphisms analyzed among groups of patients according to tumor stage, tumor location, smoking habits, and response to chemotherapy. A logistic regression was used to analyze the contribution of patient tumor parameters to a classification in two groups: responders and nonresponders. A t test was used to compare quantitative variables. These statistical tests were performed using the STATA software (STATA 7.0; College Station). The pairwise D’ values between MMP1 and MMP3 polymorphisms and between both MMP7 polymorphisms were estimated using the expectation maximization algorithm. Furthermore, a statistical analysis was performed using haplotype data with the software developed by David Tregouet (23)
. The power of the study was calculated to answer the following question: is the response rate different between subjects with MMP1 or MMP3 variant alleles? The computation shows that 
70 subjects/group should allow seeing a relative risk of 1.3 with 80% of power. Because we hypothesized that the frequency of variant alleles for MMP1 and MMP3 is 50%, we should have considered 140 patients.
RESULTS
A total of 148 patients was included in this study; among them, 66% (99 of 148) had an objective response rate of >50%, and 34% (49 of 148) were nonresponders as defined in "Patients and Methods." Response to chemotherapy was assessed with different clinicopathological parameters: tumor location; gender; age; tobacco consumption; node involvement; tumor size; and tumor stage. No clinical or histological parameters were significantly linked to response to chemotherapy; however, 75% of the patients with T1 or T2 tumors were responders versus only 62% of the patients with T3 or T4 tumors (P = 0.1), and 84% of females were responders as compared with 64% of males (P = 0.08; Table 1
). Allele frequencies and different genotypes for the three genes analyzed are shown in Table 2
. The distribution of the genotypes was in agreement with Hardy and Weinberg equilibrium. As expected, a linkage disequilibrium was found between the MMP1 2G and the MMP3 6A alleles, as well as between the MMP1 1G and the MMP3 5A alleles. The D’ was estimated to be 0.39 (P < 10-9). Concerning the MMP7 polymorphisms, the D’ was estimated to be –1.00, assuming that these two close polymorphisms were in complete disequilibrium. A significant correlation was found between MMP3 genotypes and response to chemotherapy. Indeed, patients with the 6A/6A genotype responded more frequently (86%) to treatment as compared with patients with the 5A/6A (65%) or 5A/5A (55%) genotypes (P = 0.04; Table 3
). Although nonsignificant, a similar trend was observed for MMP1 genotypes. Indeed, 56% of 1G/1G versus 70% of 1G/2G and 72% of 2G/2G responded to treatment. No significant correlation was observed between MMP7 genotype and response to chemotherapy.
|
|
|
. The absence of TP53 mutations in tumor DNA and the MMP3 6A/6A genotype are independent factors that influence positively the response to 5FU-cisplatin neoadjuvant chemotherapy in head and neck cancer patients. Moreover, the hypothesis that MMP3 polymorphism could be related to TP53 mutations was tested. No correlation was shown between the different MMP3 genotypes and TP53 mutations in tumor DNA. To better characterize the contribution of the MMP3 and MMP1 genes to a specific phenotype, a haplotype analysis was performed using a maximum likelihood method recently proposed for haplotype-phenotype association (23)
. This method allows the estimation of haplotype frequencies, as well as covariate-adjusted haplotype effects, expressed as odds ratios for a binary phenotype (case-control status) by comparison to a reference haplotype. The frequencies of the different haplotypes in the group of responders and nonresponders were given in Table 5
. A significant haplotypic effect was observed (P = 0.03). The odds ratio for each haplotype was estimated using the 1G-5A haplotype as the reference after adjustment on TP53 mutation status, gender, and age. A similar significant haplotypic effect was observed for 1G-6A and for 2G-6A on response to chemotherapy (Table 6)
.
|
|
|
In this series of 148 head and neck cancer patients, the hypothesis that MMPs could influence response to chemotherapy by promoting FasL shedding and interacting with the Fas/FasL pathway was investigated. The approach used in this work consisted of the analysis of four MMP promoter functional polymorphisms known to modulate MMP expression levels. All together, the results of this screening demonstrated that the different MMP1, MMP3, and MMP7 allele frequencies were in accordance with previous reports on Caucasian population (21 , 24 , 25) . A correlation was identified between MMP3 promoter polymorphism and response to chemotherapy in HNSCC. It was shown for the first time that patients carrying the low transcription level allele 6A had a better response to 5FU-cisplatin combination therapy than did patients carrying the 5A allele. Moreover, multivariate analysis, including TP53 mutational status, showed that MMP3 contribution to response to treatment was an independent factor. This is in accordance with the demonstration that the induction of FasL after DNA damage is p53 independent (26) . It would be expected that a decrease of MMP3 could result in a lower FasL cleavage and a better autocrine and paracrine death signaling between cancer cells. The difference in allele transcription caused by polymorphisms in promoters is certainly subtle as compared with overexpression due to gene amplifications. Nevertheless, in various cancer types, MMP polymorphisms were related to different cancer risks, suggesting that genotypes could influence by themselves cancer development or cancer cell response to DNA damage (24) .
The finding that MMP3 could influence response to chemotherapy in HNSCC suggested its implication in Fas-FasL-mediated apoptosis as it was shown for MMP7 in Ewing sarcoma. Indeed, the direct implication of MMP7 in FasL cleavage and its influence on response to chemotherapy was strongly suggested by its effect on the reduction of Fas-mediated apoptosis (14) . In our study, MMP7 promoter polymorphisms do not influence response to treatment. The importance of MMP3 seemed at first an unexpected result, but the fact that other MMPs could play a role in cell types in which the Fas/FasL pathway mediates drugs cytotoxicity has not yet been studied. It suggests that, depending on cell type, different MMPs could mediate similar activities. It is known that MMPs are highly regulated proteinases and are expressed at very low levels in normal adult tissues. MMP3 and MMP7 expression levels are documented in HNSCC as in other epithelial tumor types. Moreover, increased expression in normal surrounding tissue has been documented for MMP3, underlying the importance of matrix components on tumor progression (15 , 27) . MMP3 is expressed at very low levels in normal adult tissues, and peritumoral expression could be because the host tissue reacted as though the tumor was a wound. This extratumoral MMP expression could influence tumor growth and behavior toward cytotoxic stress. In our study, polymorphisms have been analyzed in lymphocyte DNA and reflected stroma cell patterns. It could be that tumor and lymphocyte haplotypes do not always match if loss of heterozygosity in tumor cells is taken into account. We determined allelic imbalance among MMP3 heterozygous patients by comparing genotypes between lymphocyte and tumor DNA. Allelic imbalance was found in 56% of samples, with 52% showing loss of the 6A allele. Response rate to chemotherapy was similar in patients with or without allelic imbalance at this locus (data not shown).
Finally, the functionality of the MMP3 promoter polymorphism has been demonstrated in fibroblasts and results in the binding of a repressor. Nothing is known of the presence of this repressor in cancer cells. All together, these findings reinforce the idea that surrounding normal stroma cells play a important role in cancer evolution.
The implication of the Fas-FasL pathway in drug-induced apoptosis is still discussed, and the relation between MMP3 and FasL shedding has to be demonstrated in HNSCC, but arguments exist that link MMPs, Fas-FasL pathway, and drug cytotoxicity (28, 29, 30) . First, both drugs used in this study have been shown to drive cell death through an up-regulation of this pathway (29 , 31) . Second, MMP inhibitors can block shedding of FasL, causing its accumulation on cell surface and the induction of apoptosis (8) . Third, clinical trials have shown that MMP inhibitors could increase response rate in association with conventional chemotherapy, although Phase III trials failed to demonstrate clear benefits (32) .
Finally, other putative mechanisms may explain the link between MMP3 and resistance to chemotherapy. Indeed, the composition of extracellular matrix proteins could be dependent of MMP3 expression: higher expression could facilitate growth factor delivery from the matrix, leading to lower response to chemotherapy in 5A/5A patients. In transgenic mice model, enhanced MMP3 expression facilitates mammary tumor promotion, suggesting that MMP3 is involved in the early steps of carcinogenesis (33) . Whether MMP3 level modulates different carcinogenesis pathways remains to be studied.
The usefulness of cancer biomarkers is based on their informativity (i.e., sensitivity and specificity) but also on the analytical accessibility of the assay. Indeed, the quantification of proteins or RNAs can be difficult to standardize because of the host/tumor heterogeneity of samples and the necessity of high-quality material. At the opposite, the determination of gene polymorphisms linked to altered protein expression is suitable for automated analysis and can easily be used in large-scale studies or in hospital laboratories. For the first time, the role of MMP3 in response to 5FU-cisplatin chemotherapy was demonstrated in a large series of 148 HNSCC undergoing same neoadjuvant treatment. Indeed, the MMP3 low expression allele 6A correlates with a better response to treatment. In this work, we suggest that MMP3 could modulate cancer cell behavior toward chemotherapy possibly by interfering with the Fas-FasL pathway and that targeting and inactivating MMP3 may be an option to enhance the efficacy of conventional cancer chemotherapy. The development of specific MMP inhibitors may have important therapeutic implications based on a better understanding of MMP biology.
FOOTNOTES
Grant footnote: La région Ile de France, La ligue Nationale de Lutte Contre le Cancer, and La Fondation de France.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Requests for reprints: Pierre Laurent-Puig, Inserm 490 Laboratoire de Toxicologie Moléculaire, Université René Descartes Paris V, 45 rue des Saints-Pères, 75006 Paris, France. Phone: 33142862081; Fax: 33142862072; E-mail: pierre.laurent-puig{at}biomedicale.univ-paris5.fr
Received 7/30/03; revised 12/15/03; accepted 1/21/04.
REFERENCES
This article has been cited by other articles:
|
|
 |
|
  J. Hopkins, D. W. Cescon, D. Tse, P. Bradbury, W. Xu, C. Ma, P. Wheatley-Price, J. Waldron, D. Goldstein, F. Meyer, et al. Genetic Polymorphisms and Head and Neck Cancer Outcomes: A Review Cancer Epidemiol. Biomarkers Prev., March 1, 2008; 17(3): 490 - 499. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. Zinzindohoue, T. Lecomte, J.-M. Ferraz, A.-M. Houllier, P.-H. Cugnenc, A. Berger, H. Blons, and P. Laurent-Puig Prognostic Significance of MMP-1 and MMP-3 Functional Promoter Polymorphisms in Colorectal Cancer Clin. Cancer Res., January 15, 2005; 11(2): 594 - 599. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Zhang, X. Jin, S. Fang, Y. Li, R. Wang, W. Guo, N. Wang, Y. Wang, D. Wen, L. Wei, et al. The functional SNP in the matrix metalloproteinase-3 promoter modifies susceptibility and lymphatic metastasis in esophageal squamous cell carcinoma but not in gastric cardiac adenocarcinoma Carcinogenesis, December 1, 2004; 25(12): 2519 - 2524. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cancer Prevention Research |
| Cancer Prevention Journals Portal | Cancer Reviews Online |
| Annual Meeting Education Book | Cell Growth & Differentiation |
