A Xeroderma Pigmentosum Group D Gene Polymorphism Predicts Clinical Outcome to Platinum-based Chemotherapy in Patients with Advanced Colorectal Cancer

Introduction

Resistance to platinum compounds takes place through several mechanisms: decreased drug accumulation, drug inactivation, enhanced tolerance to platinum-DNA adducts, or enhanced DNA repair (1) . The NER4 pathway is one of the major DNA repair systems in mammalian cells. In fact, NER is the only known mechanism in mammalian cells for the removal of bulky, helix-distorting DNA adducts produced by platinum agents such as oxaliplatin (2) . The XPD gene, also known as ERCC2, encodes a helicase that is a component of transcription factor TFIIH (3) and an essential member of the NER pathway (4) . Mutations of the XPD protein such as a nucleotide substitution in codon 541 causing an amino acid Ser 224 Arg change can result in defective repair phenotypes (5) . Several polymorphisms in the XPD gene have been identified (6) , yet their functional significance remains to be elucidated. Reports have shown that three common polymorphisms in the XPD gene (in codons 156, 312, and 751) may be associated with differential DNA repair capacity (7, 8, 9) . The G312A polymorphism causes the amino acid change Asp 224 Asn, whereas the A751C polymorphism causes the amino acid change Lys 224 Gln. A third single nucleotide polymorphism C156A is silent, but it transforms a high usage codon into a low usage codon. Studies regarding the polymorphisms in the XPD gene and their effect on DNA repair capacity have yielded diverging results. This is especially true for the Lys751Gln polymorphism where the Lys variant was shown to be associated with decreased DNA repair in one study (7) , whereas a much larger study reported the opposite (8) . On the basis of the aforementioned data we designed a retrospective study to determine the association between the XPD polymorphisms at codons 156, 312, and 751, and clinical response and survival in patients with advanced colorectal cancer treated with platinum-based chemotherapy.

Subjects and Methods

Eligible Subjects.

Patients with refractory advanced colorectal cancer were enrolled in the compassionate oxaliplatin protocol 3C-98–3 at the USC/Norris Comprehensive Cancer Center from 1998–2000. This study was conducted at the Norris Comprehensive Cancer Center and approved by the Institutional Review Board of the University of Southern California for Medical Sciences. All of the patients signed an informed consent before entering the study. Patients entered in the treatment protocol also signed an informed consent for blood sample collection to establish the clinical significance of genetic polymorphisms in gastrointestinal cancer (USC protocol OS-99-10). All of the specimens (blood samples) were obtained at the time of treatment protocol entry. The age, race, and follow-up information for each subject were obtained from the tumor registry at USC/Norris Comprehensive Cancer Center. The clinical data and specimens were evaluated retrospectively.

All of the patients had bidimensionally measurable disease at the time of protocol entry. Responders to therapy were classified as those patients of which their tumor burden (the sum, overall measurable lesions, the products of the largest diameter, and its perpendicular diameter) decreased by 50% or more for at least 6 weeks. Progressive disease was defined as 25% or more increase in tumor burden (compared with the smallest measurement) or the appearance of new lesions. Patients who did not experience a response and did not progress within the first 12 weeks following start of 5-FU/oxaliplatin, were classified as having stable disease.

Survival was determined from the start day of the 5-FU/oxaliplatin treatment to death. Patients who were alive at the last follow-up evaluation were censored at that time. Two patients discontinued from the study too early for evaluation of response, but they were included for the determination of survival.

Treatment Protocol.

All of the patients enrolled on the study had pretreated advanced colorectal tumors. All of the patients had failed an earlier treatment with 5-FU, and 75% failed an additional second line treatment with irinotecan (CPT-11). All of the participants received the following combination therapy regimen: 130 mg/m² oxaliplatin every 3 weeks and weekly continuous infusion 5-FU (200 mg/m²/day).

Genotyping.

Blood sample was collected from each individual, and genomic DNA was extracted from peripheral blood lymphocytes using the QiaAmp kit (Qiagen, Valencia, CA). XPD genotypes were performed using a PCR-RFLP technique. Fragments containing the polymorphisms at codons 156, 312, and 751 were amplified as described previously (7, 8, 9) . The hot-start method was used to amplify the codon 312 fragment. Amplification of the PCR fragment for polymorphisms was not possible for some samples, especially for the codon 312 fragment and were noted as “unknown.”(see Table 1⇓ ). The restriction enzyme analysis of the PCR fragments was done as described previously (7, 8, 9) , and the fragments were visualized in a 2.5–3% agarose gel. All of the assays were performed two times by two independent investigators (D. P., J. S.), who were blinded for the analysis of the clinical data (W. Z., D. T-W., S. G.).

Statistical Analysis.

Contingency tables and the Fisher’s exact test were used to summarize the association of response (grouped as response, stable disease, and progression disease) to 5-FU/oxaliplatin with XPD polymorphisms. Because there was no difference between the response and stable disease groups, we combined these two groups as no progression and compared it to the group with progressive disease. The proportion with progressive disease, odds ratio, and the associated 95% CI were calculated. Kaplan-Meier plots and the log-rank test were used to compare survival of patients according to genotype. Median survival was calculated based on the Kaplan-Meier estimator. The relative risk, the probability of survival at 18 months, and the Greenwood SE were used to summarize the data. All Ps are two-sided (10) .

Results

A total of 73 patients were enrolled in this study. The overall response to 5-FU/oxaliplatin was 14% (10 of 71; 95% CI 7%, 25%); 69% (49 of 71; 95% CI 57%, 79%) had stable disease; and 17% (12 of 71; 95% CI 9%, 28%) had progressive disease. There were no significant associations between the three polymorphisms and age, sex, and ethnicity (data not shown). The overall median survival was 11.7 (95% CI 7.9, 17.4) months. The median time of follow-up and range was 11.7 (1.0–26.5) months (see Tables 1⇓ and 2⇓ ).

Codon 751.

Seventy-one patients were evaluated for the Lys751Gln polymorphism. Thirty-one percent (22 of 71) of the patients were homozygous for the Lys/Lys genotype. Fifty-five percent (39 of 71) were heterozygous Lys/Gln, and 14% (10 of 71) were homozygous for the glutamine variant. We found that 24% (5 of 21) of patients with the Lys/Lys genotype responded, compared with 10% (4 of 39) and 10% (1 of 10) of those with the Lys/Gln and Gln/Gln genotypes, respectively (P = 0.015). One patient with the Lys/Lys genotype was not evaluable for clinical response for discontinuing from the study too early (see Tables 1⇓ and 3⇓ ).

More significantly, of those with the Gln/Gln genotype, 50% (5 of 10) had progressive disease compared with 14% (3 of 21) and 8% (3 of 39) of patients with the Lys/Lys and Lys/Gln genotypes, respectively (P = 0.008). (see Table 3⇓ ) The relative risk of progression (with the Gln/Gln group being the reference) was 0.17 (95% CI 0.03, 0.91) for the Lys/Lys genotype and 0.08 (95% CI 0.02, 0.45) for the Lys/Gln genotype (data not shown).

We also evaluated the survival and Lys751Gln polymorphism in 71 patients. Those with Lys/Lys genotype had a median survival of 17.4 (95% CI 7.9, 26.5) months. Patients with the Lys/Gln and Gln/Gln genotypes had a median survival of 12.8 (95% CI 8.5, 25.9) and 3.3 (95% CI 1.4, 6.5) months, respectively (P = 0.002). The relative risk of dying (with the Lys/Lys group being the reference) was 1.31 for those with the Lys/Gln genotype and 4.01 for the Gln/Gln group (see Table 2⇓ and Fig. 1⇓ ).

Codon 156.

A total of 69 of 73 patients were evaluated for the C 224 A polymorphism at codon 156. Thirty-two percent (22 of 69) and 55% (38 of 69) of the patients were homozygous and heterozygous for the C allele respectively. Thirteen percent (9 of 69) were homozygous for the A allele. Clinical response was evaluable in 67 of 69 patients evaluated for this polymorphism. Of those with the C/C genotype, 24% (5 of 21) responded compared with 11% (4 of 38) and 13% (1 of 8) of those with the C/A and A/A genotype respectively (P = 0.69; see Tables 1⇓ and 3⇓ ).

Data for survival was available for all 69 of the patients evaluated for this polymorphism. Those with the C/C genotype had a median survival of 11.7 (95% CI 7.8, 24.9) months. Patients with the C/A and A/A genotype had median survivals of 13.2 (95% CI 6.5, 25.9) and 8.5 (95% CI 3.3, 17.7+) months, respectively (P = 0.50; see Table 2⇓ ).

Codon 312.

We were able to determine the Asp312Asn polymorphism for only 59 of 73 patients. The Asp/Asp variant occurred at a frequency of 47% (28 of 59). Forty-one percent (24 of 59) were heterozygous (Asp/Asn) and 12% (7 of 59) were homozygous for the Asp variant. Nineteen percent (5 of 27) of the patients with the Asp/Asp genotype responded, whereas 17% (4 of 24) and 14% (1 of 7) with the Asp/Asn and Asn/Asn genotype, respectively, responded (P = 1.00; see Tables 1⇓ and 3⇓ ).

Data for survival was obtained for all 59 of the patients. The median survival for patients with the Asp/Asp genotype was 26.5 (95% CI 8.3, 26.5) months. Those with the Asp/Asn and Asn/Asn genotypes had a median survival of 9.2 (95% CI 5.4, 13.3) months and 17.3+ (95% CI 1.1, 17.3+) months, respectively (see Table 2⇓ ).

Association among the Three XPD Polymorphisms.

We determined the association among the three XPD polymorphisms to assess evidence of linkage. In the analysis between codons 751 and 156, patients with the Lys751 allele also tended to carry the C156 allele (P = 0.028). Similarly, the polymorphisms in codons 751 and 312 also seemed to be linked (P < 0.001) with the majority if the patients with the Lys751Lys genotype were also carrying the Asp312Asp genotype. There was no significant association among the polymorphisms of codons 156 and 312 (data not shown).

Discussion

In our study we assessed three common polymorphisms of the XPD gene that may influence DNA repair capacity and determined their association to clinical response and overall survival to combination oxaliplatin and 5-FU chemotherapy in advanced colorectal cancer patients. The frequencies of the various genotypes of the XPD codon 751, 156, and 312 polymorphisms were in accord with reports published previously (8 , 9 , 11) .

The overall response rate in our cohort was 14%, and median survival was 11.7 months (95% CI 7.9, 17.4). These results are in agreement with previous studies that report objective response rates between 15% and 25% and median survival ranging between 7 and 17 months in heavily pretreated patients treated with combination oxaliplatin and 5-FU (12) .

We found that a significant relationship exists between clinical response to chemotherapy and the Lys751Gln polymorphism of the XPD gene. Patients with the Gln/Gln genotype were 6–12 times more likely to have progressive disease compared with either the Lys/Lys or Lys/Gln group.

The study also shows that patients with the Lys/Lys genotype had the longest median survival followed by those with the Lys/Gln genotype. Patients with the Gln/Gln genotype faired the worst, having a significantly shorter survival or increased relative risk of dying when compared with the Lys/Lys group.

It is unclear how the change of amino acid at codon 751 of the XPD gene polymorphism influences clinical outcome to oxaliplatin based chemotherapy. One possible explanation is that the polymorphism influences DNA repair capacity. However, this polymorphic variation does not occur in any known helicase/ATPase domains. It has been posited that this change from a basic to a polar amino acid takes place ∼50 bp upstream to the poly(A) signal and, thus, may affect XPD protein function (9) . Therefore, patients with the Lys/Lys genotype may have a lower DNA repair capacity because of less efficient XPD function than those with the Gln/Gln genotype possibly making them more sensitive to oxaliplatin-based chemotherapy.

The single nucleotide/amino acid change may even affect post-transcriptional and/or post-translational stability and, thus, alter XPD protein levels (13) . Although it is not known whether XPD is highly expressed in colorectal tumor tissue, our data indicate that possibility. Currently there are no published reports that link the XPD Lys751Gln polymorphism to either gene expression or protein levels. A previous report studying a small number of patients showed no significant relationship between XPD gene expression and response to platinum chemotherapy in ovarian cancer tumors (14) . Furthermore, a recent study examining differential XPD expression in ovarian cancer cell lines with varying degrees of resistance to cisplatin showed only minimal relationship between XPD gene expression and cisplatin resistance (15) . These findings might point to mechanisms other than gene expression or protein levels as an explanation for the association between the Lys751Gln polymorphism and clinical outcome to platinum-based chemotherapy.

Structural changes caused by the amino acid change may also impact the interaction of XPD with other members of the NER complex leading to differential DNA repair capability. Lastly, the XPD Lys751Gln polymorphism may cosegregate with other DNA repair enzymes such as ERCC1 and XRCC1 because of their close proximity in the genome, thus being a marker for DNA repair capacity without influencing XPD gene expression or protein function (9) .

Recently, Spitz et al. (8) studied the functional significance of the XPD Lys751Gln and Asp312Asn polymorphisms among lung cancer patients and healthy controls. The report comprised the largest cohort studied to date in relation to the XPD polymorphisms. They reported that the variant Gln/Gln was associated with suboptimal DNA repair capacity. This was determined by assessing the ability of cells to remove DNA-adducts induced by benzo(a)pyrene, a major constituent of tobacco smoke. This association was statistically significant among the lung cancer cases but not among the healthy controls, indicating a role for XPD in tobacco-related cancers.

A small case-control study also reported increased risk for squamous cell carcinoma of the head and neck in patients with the Gln/Gln genotype when compared with the Lys/Lys group. This risk was additionally increased in a subgroup of older subjects and current drinkers (16) .

In contrast, Lunn et al. (7) reported a significant association between the Lys/Lys genotype and decreased DNA repair proficiency (odds ratio 7.2; 95% CI 1.01–87.7). In this study, DNA repair proficiency was determined by quantifying X-ray-induced chromatid aberrations in lymphocytes from a small number of women. Furthermore, Dybdahl et al. (9) and Tomescu et al. (17) reported that individuals with the Lys variant were associated with increased risk for basal cell carcinoma and melanoma. This presumably would be attributable to suboptimal DNA repair capacity, thus decreasing their ability to repair oncogenic insults to the genome. Others such as Duell et al. (18) and Moeller et al. (19) reported no significant relationship between the Lys751Gln polymorphism and DNA repair proficiency.

It is difficult to explain the seemingly conflicting data regarding the effect of the Lys751Gln polymorphism on DNA repair capacity. A hypothesis presented first by Lunn et al. (7) and later by Spitz et al. (8) suggests that the Lys allele may have different effects on different DNA repair pathways. Therefore, the use of varying assays to assess DNA repair may lead to disparate results. Also, it is not known how well the assays used in the studies of Duell et al. (18) , Moeller et al. (19) , Lunn et al. (7) , and Spitz et al. (8) mimic repair of DNA damage caused by platinum-adducts. To our knowledge, there are no published reports studying the association of the XPD Lys751Gln polymorphism and repair of DNA damage caused by platinum agents. Additional studies are needed to clarify the role of XPD in platinum resistance.

Previous reports have shown potential effects on DNA repair capacity associated with the Asp312Asn and C156A polymorphisms (8 , 9 , 11 , 17 , 20) . However, we were not able to show a significant relationship between the polymorphisms, and response and survival to 5-FU/oxaliplatin chemotherapy.

Studies have also reported evidence of linkage between the XPD 751and 312 (20) , and XPD 751 and 156 (9) polymorphisms, which was confirmed in our study. However, we did not find a significant association between the XPD 156 and 312 polymorphisms. The clinical significance of the linkage disequilibrium is unclear, because neither the XPD 156 nor 312 polymorphisms were independent predictors of response or survival to chemotherapy. Whether synergistic or opposing effects exist between the three XPD polymorphisms that may influence clinical outcome to chemotherapy remains to be determined. We hope that future studies with larger sample sizes will aid in the clarification of this possibility.

To conclude, we have assessed the clinical significance of three common XPD polymorphisms in the outcome of patients with advanced colorectal cancer treated with combination 5-FU/oxaliplatin. Our results show that a significant association exists between response to 5-FU/oxaliplatin chemotherapy and survival, and the Lys751Gln polymorphism. To our knowledge this is the first study that shows a relationship between an XPD gene polymorphism and clinical outcome to chemotherapy. The mechanisms for this relationship remain unknown, and the studies addressing the functional significance of this polymorphism are conflicting.

There are a few issues that must be considered when evaluating the findings of the current study. Although these patients were consistently treated with a common protocol and the clinical data were prospectively collected, this was still a retrospective study. Furthermore, the number of patients in this study is relatively small. Thus, there is some possibility that the reported association between the XPD Lys751Gln polymorphism and clinical outcome to 5-FU/oxapliplatin may be a spurious finding. Conversely, there may be moderate (differences of ≤25%) or small associations between the polymorphisms at the 156 and 312 codons and clinical outcome that would have a low probability of being detected in this study. For these reasons, a larger prospective study will be needed to confirm the findings of this study.

As the role of oxaliplatin becomes additionally established in the treatment of advanced colorectal cancer both as first- or second-line agents, it will be important to determine the various factors that may predict response or resistance to this agent. Our study indicates that the XPD Lys751Gln polymorphism may be a powerful predictor of clinical outcome to oxaliplatin-based chemotherapy.