Abstract
Purpose
To demonstrate whether single nucleotide polymorphisms (SNPs) of drug metabolic genes were associated with toxicity of gemcitabine-based chemoradiotherapy and overall survival (OS) of patients with pancreatic cancer.
Experimental Design
We evaluated 17 SNPs of the CDA, dCK, DCTP, RRM1, hCNT1, hCNT2, hCNT3, and hENT1 genes in 154 patients with potentially resectable pancreatic adenocarcinoma who were enrolled in clinical trials at The University of Texas M. D. Anderson Cancer Center (Houston, TX) from February 1999 to January 2006, with follow-up until April 2009. Patients received neoadjuvant concurrent gemcitabine and radiation therapy with or without gemcitabine-cisplatin induction therapy. The association of genotypes with toxicity or OS was tested, respectively, by logistic regression and Cox regression analysis.
Results
None of the 17 SNPs, individually, had a significant association with OS. A combined genotype effect of CDA A-76C, dCK C-1205T, DCTD T-47C, hCNT3 C-69T, hENT1 T-549C and hENT1 C913T on OS was observed. Patients carrying 0–1 (n=43), 2–3 (n=77) or 4–6 (n=30) variant alleles had median survival time of 31.5, 21.4 and 17.5 months, respectively. The hazard ratio of dying (95% CI) was 1.71 (1.06–2.76) and 3.16 (1.77–5.63) for patients carrying 2–3 or 4–6 at-risk genotypes (P=0.028 and P<0.001), respectively, after adjusting for clinical predictors. CDA C111T, dCK C-1205T, dCK A9846G and hCNT3 A25G, individually and jointly, had a significant association with nuetropenia toxicity.
Conclusions
These observations suggest that polymorphic variations of drug metabolic genes were associated with toxicity of gemcitabine-based therapy and OS of patients with resectable pancreatic cancer.
Keywords: gemcitabine metabolic genes, nucleotide transporter genes, single nucleotide polymorphism, pancreatic cancer
Introduction
Gemcitabine (2′,2′-difluoro 2′-deoxycytidine) is the standard first-line agent for treatment of pancreatic cancer. However, 75% of patients do not benefit from this therapy (1), and other than stage, it is not clear what factors predict clinical response to gemcitabine. A major dose limiting side effect of gemcitabine is hematological toxicity such as neutropenia and thrombocytopenia, which often result in dose reduction or longer intervals between gemcitabine administrations. However, there is no available biomarker that predicts the toxicity of gemcitabine.
Gemcitabine is a nucleoside analogue and a prodrug that requires cellular uptake and intracellular phosphorylation (2) (Fig. 1). Five of the nucleotide transporters found in humans—human concentrative nucleotide transporter (hCNT) 1–3 (aka solute carrier family 28 A1–A3); and human equilibrative nucleotide transporter (hENT) 1 and 2 (solute carrier family 29) —appear to be responsible for cellular uptake of gemcitabine (2). Once inside the cell, gemcitabine is phosphorylated by deoxycytidine kinase (dCK) to its monophosphate form. This first stage of phosphorylation is the rate-limiting step for further phosphorylation to the active triphosphate form and, thus, is essential for the activation of gemcitabine (3). The active diphosphate metabolite of gemcitabine inhibits DNA synthesis indirectly through the inhibition of ribonucleotide reductase (RR) (4). Inhibition of RR by gemcitabine blocks the de novo DNA synthesis pathway and decreases the intracellular concentrations of normal deoxynucleotide triphosphate pools. Gemcitabine is inactivated primarily by deoxycytidine deaminase (CDA)-mediated conversion to difluorodeoxyuridine.
Fig. 1.
Schematic description of gemcitabine (dFdC) transportation and metabolism. The italic letters indicate genes that are examined in this study.
Previous studies in cell lines and in patients have associated gemcitabine resistance to decreased expression of the activation enzyme (5–8), increased degradation (9), decreased nucleoside transport of drug into cells (10–12), and increased expression of RRM1 (10). Over expression of hENT1 and RRM1/2 in tumors has been significantly correlated to survival in pancreatic adenocarcinoma treated with gemcitabine (11–14).
It is theorized that an association exists between the activity of these proteins and the polymorphic variation of genes coding for the proteins (14,15). Few clinical studies have shown a positive association between CDA SNPs and drug toxicity (16,17). The current study tested the hypothesis that genetic variations in gemcitabine transport and metabolism, as well as in the drug’s target, may affect the clinical response, hematological toxicity, and overall outcome of pancreatic cancer patients treated with gemcitabine. We tested this hypothesis in a relatively homogeneous population of 154 patients with potentially resectable pancreatic cancer who had undergone neoadjuvant gemcitabine -based chemotherapy plus radiation therapy.
Materials and Methods
Patient recruitment and data collection
The study involved 154 patients who, at the time of diagnosis, had potentially resectable adenocarcinoma of the head of the pancreas and were enrolled in one of two phase II clinical trials (ID98-020 or ID01–341) of preoperative (neoadjuvant) combined chemotherapy-radiation therapy at The University of Texas M. D. Anderson Cancer Center (Houston, TX) conducted sequentially from February 1999 to January 2006 and were observed through April 2009. These 154 patients represented the subset of patients enrolled in these clinical trials who had a DNA sample available. The study was approved by the institutional review board of M.D. Anderson Cancer Center. Patients in the ID98-020 trial (n = 70) received gemcitabine-based chemoradiotherapy consisting of weekly gemcitabine (400mg/m2) for 4 weeks and radiation (30 Gy in 10 fractions) for 2 weeks. Patients in the ID01–341 trial (n = 84) received induction therapy of gemcitabine (750 mg/m2/d) and cisplatin (30mg/m2/d) every 2 weeks for 4 weeks and radiation (30 Gy in 10 fractions) for 2 weeks with weekly gemcitabine. The same eligibility criteria for patient recruitment had been applied in both trials, and no significant difference in any clinical feature was observed between the two patient populations (18,19).
Tumor response to therapy was evaluated by computed tomography (CT) before and after completion of the preoperative chemoradiation, and defined according to the RECIST criteria (Response Evaluation Criteria in Solid Tumors) as partial response (PR), stable disease (SD) or progressive disease (PD). Among patients with resected tumor, tumor response to preoperative treatment was histologically evaluated for percentage of viable tumor cells on resected tumor as previously described (20). Toxicity was graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE), version 3.0. Curative resection was defined by achievement of margin-negative resection. Treatment received after tumor recurrence was not considered in this study. Overall survival was calculated from the date of diagnosis to the date of death or date of last follow-up.
DNA extraction and genotyping
We selected 17 SNPs of the CDA, dCK, RRM1, deoxycytidylate deaminase (DCTD), hCNT1, hCNT2, hCNT3, and hENT1 genes in this exploratory investigation according to the following criteria: 1) The minor allele frequency of the SNP is greater than 10% among Caucasians; 2) coding SNPs including nonsynonymous and synonymous SNPs; and 3) SNPs that have been associated with cancer risk or clinical outcome in previous investigations. The genes, nucleotide substitutions, function (such as encoding amino acid changes), reference SNP identification numbers, and reported allele frequencies of the 17 SNPs evaluated in this study are summarized in Table 1.
Table 1.
SNPs evaluated
| Gene | Chromosome | SNP | Reference SNP ID number |
Minor allele frequency* |
|---|---|---|---|---|
| CDA | 1p36.12b | Ex4 +111C>T, T145T | 1048977 | 0.28 |
| Ex2 −76A>C, K27Q | 2072671 | 0.44 | ||
| dCK | 4q13.3b | IVS6 −1205C>T | 4694362 | 0.45 |
| IVS2 +9846A>G | 12648166 | 0.43 | ||
| RRM1 | 11q15.4d | Ex19 +42G>A, A744A | 1042858 | 0.11 |
| Ex19 +33A>G, T741T | 3177016 | 0.47 | ||
| Ex9 −27C>A, R284R | 183484 | 0.48 | ||
| DCTD | 4q35.1b | Ex4 −47T>C, V116V | 7663494 | 0.33 |
| hCNT1 | 15q25.3a | Ex15 −16A>G, Q456Q | 2242048 | 0.15 |
| Ex9 −9C>A, Q237K | 8187758 | 0.19 | ||
| hCNT2 | 15q21.1a | Ex4 −38C>A, S75R | 1060896 | 0.29 |
| Ex2 −17C>T, P22L | 11854484 | 0.34 | ||
| hCNT3 | 9q21.32c | Ex14 −69C>T, L461L | 7853758 | 0.15 |
| Ex5 +25A>G, T89T | 7867504 | 0.39 | ||
| hENT1 | 6p21.1b | IVS12 −201A>G | 760370 | 0.35 |
| IVS2 −549T>C | 324148 | 0.30 | ||
| IVS2 +913C>T | 9394992 | 0.32 |
Allele frequencies (Caucasian) were from the National Cancer Institute SNP500 cancer.
DNA was extracted from peripheral blood lymphocytes of 127 patients and from paraffin sections of normal adjacent tissues of 27 patients with resected tumors (20 from the ID98-020 trial) using Qiagen DNA isolation kits (Valencia, CA). Normal and tumor tissues are expected to have the same genotype for these germline common polymorphic sequence variants. Taqman 5′ nuclease assay was performed to determine all genetic variants using the ABI Prism 7900HT Sequence Detection System, and SDS 2.3 software (Applied Biosystems).
Approximately 5% of the samples were analyzed in duplicate, and discrepancies were seen in less than 1% of the total samples. Samples with discordant results were genotyped repeatedly and consistent results from at least two analyses were included in the final data analysis. The clinical information on each patient was unknown to the individual who performed the genotyping assay.
Statistical methods
The genotype distribution was tested for Hardy-Weinberg equilibrium using the goodness-of-fit χ2 test. Haplotype was inferred from the genotype data using the SNPAlyze software (version 4.1, DYNACOM Co., Ltd. Japan). The median follow-up time was computed with censored observations only, whereas the median survival time (MST) was calculated using data from all patients. Risk of dying was estimated by hazard ratios and 95% confidence intervals (CIs) in Cox proportional hazard models. Factors associated with tumor response to treatment or severe (grade 3–4) neutropenia toxicity was analyzed by logistic regression models. All clinical factors were modeled independently without additional variables in the model. Factors with P<0.05 in the univariate model were put into the initial multivariate model, and backward selection was then applied until all variables were statistically significant (P<0.05). These significant clinical factors were adjusted in all the regression models for genotype analyses. All statistical testing was conductedwith SPSS software, version 17.0 (SPSS, Chicago, IL), and statistical significance was defined as P ≤ 0.05.
We estimated the false-positive report probability (FPRP) for the observed statistically significant associations using the methods described by Wacholder et al (21). FPRP is the probability of no true association between a genetic variant and a phenotype given a statistically significant finding. It depends not only on the observed P value but also on both the prior probability that the association between the genetic variant and the phenotype is real and the statistical power of the test. In the current study, we set the OR and HR values of 2.0 to 4.0 as a likely threshold value. The prior probability employed was 0.25 for all SNPs. The FPRP value for noteworthiness was set at 0.2, which indicates any finding with a FPRP P value <0.2 is noteworthy.
Results
Patient characteristics and clinical predictors
The patients’ characteristics and clinical features of their tumors have previously been described in details (Table 2) (22). The median age of the 154 patients in this study was 63 years (range, 38–84 years). There are 96 male and 58 female patients. Non-Hispanic whites consisted 86% of the patients. One hundred and sixteen patients had the primary tumor surgically resected after preoperative treatment and pathologic evaluation of the surgical specimens demonstrated a microscopically positive margin (R1 resection) in 9 of the 116. There were 117 deaths (76%) among 154 cases. The median follow-up time was 49.9 months for the patients who were still alive. The MST of the 154 patients was 21.7 months (95% CI, 17.7 to 25.6). Information on tumor grade, lymph node metastasis, and tumor response to treatment by histological evaluation was not available in patients with unresected tumors. The factors which were significantly associated with overall survival time in log-rank test included diabetes status, tumor size, serum CA19-9 level at diagnosis, tumor response by CT evaluation, curative resection, tumor grade, lymph node metastasis, and the two clinical protocols for the preoperative treatment (Table 2). In the multivariable Cox regression model, diabetes status, serum CA19-9 at diagnosis, and curative resection remained as significant predictor for OS (P=0.003, 0.020, and <0.001, respectively).
Table 2.
Patient characteristics (n=154)
| Variable | No. of patients |
No. of deaths (%) |
MST (months) |
P (Log-rank) |
|---|---|---|---|---|
| Age (years) | .098 | |||
| ≤50 | 17 | 16 (94.1) | 18.5 | |
| 51–60 | 44 | 27 (61.4) | 36.0 | |
| 61–70 | 60 | 41 (68.3) | 21.5 | |
| >70 | 33 | 27 (81.8) | 21.2 | |
| Sex | .369 | |||
| Male | 96 | 70 (72.9) | 20.9 | |
| Female | 58 | 41 (70.7) | 24.5 | |
| Race | .704 | |||
| White | 133 | 97 (72.9) | 23.9 | |
| Hispanic | 10 | 7 (70.0) | 18.2 | |
| African American | 7 | 4 (57.1) | 33.6 | |
| Other | 4 | 3 (75.0) | 10.7 | |
| Diabetes status | .017 | |||
| Negative | 109 | 74 (67.9) | 27.6 | |
| Positive | 45 | 37 (82.2) | 18.2 | |
| Tumor size (cm) | .010 | |||
| ≤2 | 65 | 40 (61.5) | 34.0 | |
| >2 | 89 | 71 (79.8) | 20.7 | |
| CA19-9 (units/mL) | .001 | |||
| ≤47 | 40 | 21 (52.5) | 52.8 | |
| 48–500 | 78 | 58 (74.4) | 22.5 | |
| 501–1,000 | 14 | 12 (85.7) | 18.4 | |
| >1,000 | 22 | 20 (90.9) | 15.3 | |
| Tumor response in CT | < .001 | |||
| PR/SD | 126 | 85 (67.5) | 27.8 | |
| PD | 27 | 25 (92.6) | 10.3 | |
| Curative resection | < .001 | |||
| Yes | 107 | 64 (59.8) | 34.4 | |
| No* | 47 | 47 (100) | 12.8 | |
| Tumor grade | .017 | |||
| Well-to-moderate | 88 | 57 (64.8) | 33.6 | |
| Poor | 32 | 25 (78.1) | 19.8 | |
| Lymph node metastasis | < .001 | |||
| Negative | 61 | 33 (54.1) | 51.3 | |
| Positive | 55 | 40 (72.7) | 26.4 | |
| Treatment Effect† | .259 | |||
| ≤50% | 70 | 43 (61.4) | 27.9 | |
| >50% | 43 | 30 (69.8) | 37.1 | |
| Clinical Protocol | .026 | |||
| GEM/XRT | 70 | 49 (70.0) | 28.1 | |
| GEM/Cisplatin/XRT | 84 | 62 (73.8) | 18.4 |
Abbreviations: MST, median survival time; PR, partial response; SD, stable disease; PD, progressive disease.
No resection (n=38) and margin positive resection (n=9).
Percentage of viable cells by histological evaluation of resected tumor.
Genotype frequency and association with OS
The 17 genotypes of interest were successfully amplified in 95–100% of the samples. Genotype frequencies of the 17 SNPs were found to be in Hardy-Weinberg equilibrium (χ2 = 0.009 3.684; Ps=0.055–0.924) except RRM1 A33G (χ2 = 14.294; P=0.0002), RRM1 C-27A (χ2 = 15.112; P=0.0001), and hCNT1 A-16G (χ2 = 9.070; P=0.0026). No significant racial differences in genotype frequency were observed (data not shown). The two or three SNPs each of the dCK, RRM1, hCNT2, hCNT3 and hENT1 (IVS12 -201A>G and IVS2 -549T>C) genes were in linkage disequilibrium (|D′|>0.5, P < 0.01).
The genotype frequencies and their associations with OS are shown in Table 3. None of the 17 SNPs showed significant association with OS by log rank test. Six SNPs (CDA A-76C, dCK C-1205T, DCTD T-47C, hCNT3 C-69T, hENT1 T-549C and hENT1 C913T) showed weak associations with OS (Cox regression P<0.20). When these 6 SNPs were analyzed in combination, a gene-dosage effect on OS was observed. As the number of at-risk alleles increased the OS decreased (Fig. 2). Patients carrying 0–1 (n=43), 2–3 (n=77) or 4–6 (n=30) variant alleles had MST of 31.5, 21.4 and 17.5 months, respectively. The HR of dying (95% CI) was 1.71 (1.06–2.76) and 3.16 (1.77–5.63) for patients carrying 2–3 or 4–5 at-risk genotypes (P=0.028 and <0.001), respectively, after adjusting for clinical predictors. The FPRP was 0.102 and 0.005 for patients carrying 2–3 or 4–6 at-risk genotypes, respectively, indicating noteworthiness.
Table 3.
Genotype and Overall Survival
| Genotype | No. of Cases | No. of Deaths | MST ± SE (months) | P (log-rank) | * HR (95% CI) | P |
|---|---|---|---|---|---|---|
| CDA C111T (T145T) | .567 | |||||
| TT | 20 | 15 | 23.8 ± 2.1 | |||
| CT | 68 | 46 | 27.8 ± 6.2 | |||
| CC | 65 | 49 | 20.9 ± 1.2 | |||
| CT versus TT/CC | 1.13 (0.76–1.69) | .544 | ||||
| CDA A-76C (K27Q) | .183 | |||||
| CC | 20 | 13 | 27.8 ± 12.6 | |||
| AC | 60 | 39 | 27.6 ± 5.0 | |||
| AA | 69 | 54 | 20.4 ± 1.5 | |||
| CC/AC versus AA | 1.30 (0.88–1.93) | .193 | ||||
| dCK C-1205T | .449 | |||||
| CC | 22 | 15 | 28.7 ± 4.0 | |||
| CT | 72 | 48 | 24.3 ± 5.7 | |||
| TT | 57 | 45 | 20.7 ± 1.5 | |||
| CC/CT versus TT | 1.44 (0.96–2.17) | .079 | ||||
| dCK A9846G | .583 | |||||
| AA | 24 | 16 | 31.5 ± 4.1 | |||
| AG | 71 | 48 | 21.5 ± 3.6 | |||
| GG | 55 | 43 | 20.9 ± 1.5 | |||
| AA versus AG/GG | 1.33 (0.76–2.32) | .319 | ||||
| RRM1 G42A (A744A) | .069 | |||||
| AA | 126 | 88 | 23.9 ± 2.6 | |||
| AG | 20 | 15 | 16.1 ± 3.1 | |||
| GG | 3 | 3 | 14.1 ± 2.8 | |||
| AA versus GG/AG | 1.32 (0.79–2.21) | .292 | ||||
| RRM1 A33G (T741T) | .908 | |||||
| GG | 49 | 34 | 26.4 ± 4.8 | |||
| AG | 51 | 36 | 21.2 ± 3.1 | |||
| AA | 48 | 35 | 22.5 ± 2.4 | |||
| GG versus AG/AA | 1.18 (0.78–1.79) | .442 | ||||
| RRM1 C-27A (R284R) | .911 | |||||
| CC | 52 | 38 | 22.8 ± 3.8 | |||
| AC | 52 | 37 | 20.7 ± 1.7 | |||
| AA | 48 | 34 | 24.3 ± 4.1 | |||
| AA versus AC/CC | 1.16 (0.77–1.76) | .471 | ||||
| DCTD T-47C (V116V) | .395 | |||||
| CC | 12 | 8 | 28.1 ± 10.8 | |||
| CT | 52 | 39 | 17.5 ± 2.2 | |||
| TT | 87 | 61 | 26.4 ± 2.0 | |||
| CC/TT versus CT | 1.35 (0.90–2.01) | .148 | ||||
| hCNT1 A-16G (Q456Q) | .918 | |||||
| GG | 125 | 89 | 23.9 ± 2.8 | |||
| AG | 21 | 16 | 21.5 ± 3.8 | |||
| AA | 5 | 3 | 21.7 ± 18.2 | |||
| GG versus AA/AG | 1.24 (0.74–2.08) | .425 | ||||
| hCNT1 C-9A (Q237K) | .942 | |||||
| CC | 88 | 65 | 23.9 ± 2.8 | |||
| AC | 53 | 37 | 20.9 ± 3.2 | |||
| AA | 10 | 6 | 13.1 ± 17.8 | |||
| CC versus AA/AC | 1.16 (0.77–1.76) | .480 | ||||
| hCNT2 C-38A (S75R) | .636 | |||||
| AA | 56 | 39 | 24.5 ± 2.5 | |||
| AC | 66 | 48 | 18.4 ± 1.7 | |||
| CC | 29 | 21 | 23.8 ± 12.7 | |||
| AA/CC versus AC | 1.04 (0.71–1.53) | .852 | ||||
| hCNT2 C-17T (P22L) | .282 | |||||
| TT | 46 | 31 | 28.1 ± 4.3 | |||
| CT | 67 | 47 | 20.7 ± 4.3 | |||
| CC | 39 | 31 | 18.7 ± 2.9 | |||
| TT/CT versus CC | 1.23 (0.80–1.90) | .353 | ||||
| hCNT3 C-69T (L461L) | .565 | |||||
| CC | 111 | 77 | 24.3 ± 2.7 | |||
| CT | 36 | 29 | 20.4 ± 3.7 | |||
| TT | 6 | 4 | 16.9 ± 7.4 | |||
| CC versus CT/TT | 1.39 (0.91–2.11) | .125 | ||||
| hCNT3 A25G (T89T) | .492 | |||||
| AA | 78 | 53 | 24.5 ± 2.8 | |||
| AG | 53 | 42 | 19.8 ± 2.6 | |||
| GG | 17 | 11 | 18.4 ± 6.8 | |||
| AA versus GG/AG | 1.03 (0.70–1.52) | .883 | ||||
| hENT1 A-201G | .966 | |||||
| AA | 56 | 40 | 23.8 ± 5.1 | |||
| AG | 65 | 44 | 20.9 ± 5.1 | |||
| GG | 28 | 22 | 23.9 ± 2.8 | |||
| AA/GG versus AG | 1.27 (0.84–1.91) | .258 | ||||
| hENT1 T-549C | .338 | |||||
| CC | 86 | 59 | 24.5 ± 3.2 | |||
| CT | 57 | 44 | 18.9 ± 2.2 | |||
| TT | 11 | 8 | 17.5 ± 11.0 | |||
| CC versus CT/TT | 1.32 (0.91–1.92) | .142 | ||||
| hENT1 C913T | .175 | |||||
| TT | 8 | 5 | 26.4 ± 19.0 | |||
| CT | 68 | 43 | 27.9 ± 7.0 | |||
| CC | 74 | 59 | 21.2 ± 1.4 | |||
| CT/TT versus CC | 1.44 (0.98–2.10) | .064 | ||||
| No. of at-risk genotypes† | .019 | |||||
| 0–1 | 43 | 25 | 31.5 ± 5.4 | 1.0 | ||
| 2–3 | 77 | 56 | 21.4 ± 2.6 | 1.71 (1.06–2.76) | .028 | |
| 4–6 | 30 | 26 | 17.5 ± 1.6 | 3.16 (1.77–5.63) | <.001 | |
Abbreviations: MST, median survival time; SE, standard error.
HR was adjusted for history of diabetes, serum level of CA19-9, and curative resection.
At risk genotypes: CDA -76AA, dCK -1205TT, DCTD -47CT, hCNT3 -69CT/TT, hENT1 -549CT/TT, and hENT1 913CC.
Fig. 2.
Combined genotype effect of CDA -76AA, dCK -1205TT, DCTD -47CT, hCNT3 -69CT/TT, hENT1 -549CT/TT, and hENT1 913CC on overall survival. The number of 0 to 6 indicates the number of deleterious genotypes associated with reduced survival.
Haplotype association with OS
The CDA C111T and A-76C TA haplotype was significantly associated with increased risk and the hENT1 A-201G, T-549C and C913T ACT haplotype was significantly associated with reduced risk of death compared to the most common haplotype of each gene respectively (Ps<0.05, Table 4). No other haplotype showed significant association with OS.
Table 4.
Haplotype and Overall Survival
| Haplotype | Frequency | * HR (95% CI) | P |
|---|---|---|---|
| CDA C111T/A-76C | |||
| CA | 0.45 | Reference | |
| TA | 0.21 | 1.49 (1.03–2.15) | .036 |
| CC | 0.19 | 1.03 (0.68–1.54) | .898 |
| TC | 0.15 | 0.71 (0.48–1.05) | .089 |
| dCK C-1205T/A9846G | |||
| TG | 0.60 | Reference | |
| CA | 0.38 | 0.77 (0.58–1.03) | .081 |
| Others | 0.02 | 0.98 (0.45–2.12) | .959 |
| RRM1 G42A/A33G/C-27A | |||
| AGA | 0.48 | Reference | |
| AAC | 0.41 | 1.18 (0.88–1.58) | .270 |
| Others | 0.11 | 1.33 (0.85–2.08) | .210 |
| hCNT1 A-16G/C-9A | |||
| GC | 0.67 | Reference | |
| GA | 0.23 | 0.90 (0.64–1.25) | .518 |
| Others | 0.10 | 1.22 (0.77–1.93) | .394 |
| hCNT2 C-38A/C-17T | |||
| AT | 0.50 | Reference | |
| CC | 0.39 | 0.96 (0.72–1.29) | .802 |
| Others | 0.11 | 1.04 (0.65–1.68) | .863 |
| hCNT3 C-69T/A25G | |||
| CA | 0.67 | Reference | |
| CG | 0.17 | 0.87 (0.60–1.27) | .474 |
| Others | 0.12 | 1.17 (0.80–1.71) | .416 |
| hENT1 A-201G/T-549C/C913T | |||
| GCC | 0.30 | Reference | |
| ACC | 0.22 | 0.81 (0.55–1.22) | .314 |
| ATC | 0.16 | 1.26 (0.85–1.88) | .243 |
| ACT | 0.15 | 0.65 (0.43–0.98) | .037 |
| Others | 0.17 | 1.04 (0.65–1.67) | .875 |
| dCK C-1205/A9846G & DCTD T-47C | |||
| TGT | 0.45 | Reference | |
| CAT | 0.28 | 0.70 (0.49–1.01) | .054 |
| TGC | 0.15 | 1.34 (0.86–2.10) | .196 |
| CAC | 0.10 | 1.03 (0.68–1.54) | .895 |
| Others | 0.02 | 1.02 (0.47–2.22) | .954 |
HR was adjusted for history of diabetes, serum level of CA19-9, and curative resection.
Genotype association with tumor response to treatment
Tumor size was the only clinical factor that was significantly associated with tumor response to therapy by histological evaluation. None of the 17 SNPs showed significant association with tumor response to therapy by radiological evaluation (data not shown). Four SNPs, i.e. dCK C-1205T, dCK A9846G, hCNT3 C-69T, and hCNT3 A25G had significant associations with tumor response by histological evaluation (Table 5). For example, 26.2% of the dCK 1205 TT carriers versus 46.4% of the CT/CC carriers had a poor response to preoperative chemoradiotherapy, i.e. >50% tumor cells were viable in resected tumor. Patients carrying 3 or more of the 4 at risk genotypes had a 5.77-fold higher risk of poor response to therapy after adjusting for tumor size (95% CI, 2.23–14.9, P<0.001). The FPRP was 0.058 for patients carrying 3–4 at-risk genotypes, indicating noteworthiness.
Table 5.
Genotype and Tumor Response to Preoperative Treatment
| Genotype | ≤50%* |
>50%* |
||
|---|---|---|---|---|
| N (%) | N (%) | † OR (95% CI) | P value | |
| dCK C-1205T | ||||
| TT | 31 (73.8) | 11 (26.2) | 1.0 | |
| CT/CC | 37 (53.6) | 32 (46.4) | 2.73 (1.15–6.45) | .022 |
| dCK A9846G | ||||
| GG | 31 (75.6) | 10 (24.4) | 1.0 | |
| AG/AA | 37 (53.6) | 32 (46.4) | 2.96 (1.23–7.13) | .015 |
| hCNT3 A25G | ||||
| AA | 42 (70.0) | 18 (30.0) | 1.0 | |
| AG/GG | 24 (49.0) | 25 (51.0) | 2.733 (1.21–6.17) | .016 |
| hCNT3 C-69T | ||||
| CC | 55 (68.8) | 25 (31.2) | 1.0 | |
| CT/TT | 14 (43.8) | 18 (56.3) | 3.08 (1.30–7.31) | .011 |
| No. of at-risk genotypes | ||||
| 0–2 | 52 (72.2) | 20 (27.8) | 1.0 | |
| 3–4 | 14 (38.9) | 22 (61.1) | 5.77 (2.23–14.9) | <.001 |
Percentage of viable cells by histological evaluation of resected tumor.
OR was adjusted for tumor size.
Genotype association with toxicity
None of the clinical factors was predictive for severe nuetropenia. The CDA C111T, dCK C-1205T, dCK A9846G and hCNT3 A25G genotype individually and in combination were significantly associated with toxicity (Table 6). For example, 40.9% of the CDA 111 CT/TT carriers versus 24.6% of the CC carriers had grade 3–4 neutropenia (P=0.037). Patients carrying 3 or 4 variant alleles compared to those carrying 0–2 variant allele had a significantly higher risk for grade 3–4 neutropenia (OR: 3.57, 95% CI: 1.38–9.22, P=0.009; OR: 5.88, 95% CI: 2.10–16.5, P=0.001, respectively). The FPRP was 0.182 and 0.102 for patients carrying 3 or 4 at-risk genotypes, respectively, indicating noteworthiness. No significant association of toxicity with the remaining SNPs was noticed (data not shown).
Table 6.
Neutropenia toxicity and genotype
| Genotype | Grade1–2 |
Grade 3–4 |
||
|---|---|---|---|---|
| N (%) | N (%) | * OR (95% CI) | P value | |
| CDA C111T | ||||
| CC | 49 (75.4) | 16 (24.6) | 1.0 | |
| CT/TT | 52 (59.1) | 36 (40.9) | 2.12 (1.05–4.30) | .037 |
| dCK C-1205T | ||||
| CC | 19 (86.4) | 3 (13.6) | 1.0 | |
| CT/TT | 80 (62.0) | 49 (38.0) | 3.88 (1.09–13.8) | .036 |
| dCK A9846G | ||||
| AA | 20 (83.3) | 4 (16.7) | 1.0 | |
| AG/GG | 78 (61.9) | 48 (38.1) | 3.08 (0.99–9.55) | .052 |
| hCNT3 A25G | ||||
| GG/AG | 52 (74.3) | 18 (25.7) | 1.0 | |
| AA | 45 (57.7) | 33 (42.3) | 2.12 (1.05–4.26) | .035 |
| No. of at-risk genotypes | ||||
| 0–2 | 39 (84.8) | 7 (15.2) | 1.0 | |
| 3 | 39 (60.9) | 25 (39.1) | 3.57 (1.38–9.22) | .009 |
| 4 | 18 (48.6) | 19 (51.4) | 5.88 (2.10–16.5) | .001 |
Crude odds ratio.
Discussion
In this study, we observed a significant association of combined genotype of gemcitabine metabolic genes CDA, dCK, and DCTD as well as transporter genes hCNT3 and hENT1 with overall survival of patients with resectable pancreatic cancer and treated with preoperative gemcitabine-based chemoradiotherapy. We also observed a significant association of dCK and hCNT3 gene variants with tumor response to therapy and drug toxicity. These data support the hypothesis that genetic variations in gemcitabine metabolism affect the clinical outcome of pancreatic cancer patients receiving gemcitabine-based chemotherapy.
CDA, an enzyme involved in the pyrimidine salvage pathway, is the major gemcitabine inactivation enzyme. Three main SNPs were identified in the CDA gene: C111T (T145T), A-76C (K27Q), and G208A (14,23,24). CDA A-76C (K27Q) has previously been reported to result in a moderate decrease in activity with gemcitabine (25). The CDA 208AA homozygote allele and related haplotype have been associated with severe drug toxicity in Japanese cancer patients treated with gemcitabine plus cisplatin (16,17). However, CDA G208A was not detected in Caucasians (26). Although the CDA A-76C (K27Q) and G208A SNPs have been associated with drug toxicity, no previous study has shown an association with patient survival. In the current study, the CDA A-76C (K27Q) variant C allele was associated with a better overall survival suggesting reduced enzyme activity conferred by this allele resulted in a higher level of drug availability. In consistency, the CDA C111T and A-76C TA haplotype was significantly associated with increased risk of death. The CDA C111T (T145T) T allele showed a significant association with gemcitabine toxicity, although the functional significance of this SNP is not clear. DCTD is another gemcitabine degradation enzyme. The DCTD gene polymorphic variants, including the nonsynonymous A172G, have not been associated with clinical response to gemcitabine in previous studies (15). We observed a weak association of the DCTD T-47C (V116V) SNP with OS in our patient population. These observations were consistent with previous reports, which support a role of CDA genotype in affecting gemcitabine toxicity and clinical outcome of patients receiving gemcitabine.
dCK plays a key role in the activation of gemcitabine, and its activity has been correlated with gemcitabine sensitivity and clinical outcomes in several studies (27–29). One study reported that the haplotype containing dCK C-360G and C-201T was correlated with the clinical outcomes of cancer patients treated with Ara-C (30). In our study, we found a significant difference in toxicity and tumor response to therapy but not in OS associated with the dCK C-1205T and dCK A9846G SNPs, both are located in the intron region. It is not clear whether these SNPs are directly responsible for gemcitabine sensitivity or whether they are in linkage disequilibrium with other SNPs or other genes. Alternatively, it is conceivable that their functional difference may be mediated by affecting RNA splicing. Further haplotype analysis of the dCK gene is required to answer these questions.
The activity of nucleotide transporters is expected to play a role in gemcitabine cytotoxicity and efficacy (31). In the current study, we observed a marginally significant association of the hCNT3 C-69T (L461L), hENT1 C-549T and hENT1 C913T genotype and a significant association of hENT1 A-201G, T-549C and C913T ACT haplotype with overall survival. Further more, hCNT3 genotypes were associated with drug toxicity and tumor response to therapy. The major route for transporting gemcitabine is hENT1 and, to a lesser extent, hCNT1 and hCNT3. A previous study reported that the hENT1 promoter region haplotype containing the C1345G, G1050A, and G706C SNPs might influence gene expression (32). Two studies have explored the hENT1 haplotype and no functional significance was reported (33, 34). Considering that hENT1 expression has been associated with pancreatic cancer survival (35), it would be important to demonstrate the genotype and phenotype association to determine whether the genotype may be used as a surrogate for tumors.
A recent study in pancreatic adenocarcinoma cell lines showed that the ratio of the expression level of hENT1, dCK, RRM1, and RRM2 genes was correlated to acquired gemcitabine chemoresistance (36). However, tissue samples are not available for most pancreatic cancer patients and it is possible that some SNPs might alter substrate specificity, resulting in altered function without increased levels of mRNA and protein (37). Thus, if confirmed in other patient populations, the genotype information might be useful in stratifying patients on protocol and in predicting response and toxicity.
In summary, we observed significant associations of gemcitabine metabolic genes on the toxicity and tumor response to gemcitabine-based preoperative therapy and overall survival of patients with resectable pancreatic cancer. Even though the effect of individual common SNPs may be trivial, the combined genotype effects are remarkable. Genetic profiling of patients may provide the fundamental information required for future “individualized” therapy.
Acknowledgments
Grant Support: Supported by National Institutes of Health (NIH) RO1 grant CA098380 (D.L.), a SPORE P20 grant CA101936 (J.L.A.), an NIH Cancer Center Core grant CA16672, and a research grant from the Lockton Research Funds (D.L.).
Footnotes
Statement of Translational Relevance
This study demonstrated associations of polymorphic variants of gemcitabine metabolic genes and nucleotide transporter genes with toxicity and overall survival of 154 patients with resectable pancreatic cancer treated with preoperative gemcitabine-based chemoradiotherapy. This information might be helpful for treatment selection and dose management in future “individualized” cancer therapy.
References
- 1.Jimeno A, Hidalgo M. Molecular biomarkers: their increasing role in the diagnosis, characterization, and therapy guidance in pancreatic cancer. Mol Cancer Ther. 2006;5:787–96. doi: 10.1158/1535-7163.MCT-06-0005. [DOI] [PubMed] [Google Scholar]
- 2.Mini E, Nobili S, Caciagli B, Landini I, Mazzei T. Cellular pharmacology of gemcitabine. Ann Oncol. 2006;17 (suppl 5):v7–v12. doi: 10.1093/annonc/mdj941. [DOI] [PubMed] [Google Scholar]
- 3.Heinemann V, Schulz L, Issels RD, Plunkett W. Gemcitabine: a modulator of intracellular nucleotide and deoxynucleotide metabolism. Semin Oncol. 1995;22 (Suppl 11):11–18. [PubMed] [Google Scholar]
- 4.Baker CH, Banzon J, Bollinger JM, et al. 2′-Deoxy-2′-methylenecytidine and 2′-deoxy-2′,2′-difluorocytidine 5′-diphosphates: potent mechanism-based inhibitors of ribonucleotide reductase. J Med Chem. 1991;34:1879–84. doi: 10.1021/jm00110a019. [DOI] [PubMed] [Google Scholar]
- 5.Kroep JR, Loves WJP, van der Wilt CL, et al. Pretreatment deoxycytidine kinase levels predict in vivo gemcitabine sensitivity. Mol Cancer Ther. 2002;1:371–6. [PubMed] [Google Scholar]
- 6.Jordheim LP, Galmarini CM, Dumontet C. Gemcitabine resistance due to deoxycytidine kinase deficiency can be reverted by fruitfly deoxynucleoside kinase, DmdNK, in human uterine sarcoma cells. Cancer Chemother Pharmacol. 2006;58:547–54. doi: 10.1007/s00280-006-0195-8. [DOI] [PubMed] [Google Scholar]
- 7.Galmarini CM, Clarke ML, Jordheim L, et al. Resistance to gemcitabine in a human follicular lymphoma cell line is due to partial deletion of the deoxycytidine kinase gene. BMC Pharmacol. 2004;4:8. doi: 10.1186/1471-2210-4-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Shi JY, Shi ZZ, Zhang SJ, et al. Association between single nucleotide polymorphisms in deoxycytidine kinase and treatment response among acute myeloid leukaemia patients. Pharmacogenetics. 2004;14:759–68. doi: 10.1097/00008571-200411000-00007. [DOI] [PubMed] [Google Scholar]
- 9.Bergman AM, Pinedo HM, Peters GJ. Determinants of resistance to 2′,2′-difluorodeoxycytidine (gemcitabine) Drug Resist Update. 2002;5:19–33. doi: 10.1016/s1368-7646(02)00002-x. [DOI] [PubMed] [Google Scholar]
- 10.Achiwa H, Oguri T, Sato S, Maeda H, Niimi T, Ueda R. Determinations of sensitivity and resistance to gemcitabine: The roles of human equilibrative nucleoside transporter 1 and deoxycytidine kinase in non-small cell lung cancer. Cancer Sci. 2004;95:753–7. doi: 10.1111/j.1349-7006.2004.tb03257.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Giovannetti E, Tacca MD, Mey V, et al. Transcription analysis of human equilibrative nucleoside transporter-1 predicts survival in pancreas cancer patients treated with gemcitabine. Cancer Res. 2006;66:3928–35. doi: 10.1158/0008-5472.CAN-05-4203. [DOI] [PubMed] [Google Scholar]
- 12.Spratlin J, Sangha R, Glubrecht D, et al. The absence of human equilibrative nucleoside transporter 1 is associated with reduced survival in patients with gemcitabine-treated pancreas adenocarcinoma. Clin Cancer Res. 2004;10:6956–61. doi: 10.1158/1078-0432.CCR-04-0224. [DOI] [PubMed] [Google Scholar]
- 13.Nakano Y, Tanno S, Koizumi K, et al. Gemcitabine chemoresistance and molecular markers associated with gemcitabine transport and metabolism in human pancreatic cancer cells. Br J Cancer. 2007;96:457–63. doi: 10.1038/sj.bjc.6603559. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Fukunaga AK, Marsh S, Murry DJ, Hurley TD, McLeod HL. Identification and analysis of single-nucleotide polymorphisms in the gemcitabine pharmacologic pathway. Pharmacogenomics. 2004;4:307–14. doi: 10.1038/sj.tpj.6500259. [DOI] [PubMed] [Google Scholar]
- 15.Ueno H, Kiyosawa K, Kaniwa N. Pharmacogenomics of gemcitabine: can genetic studies lead to tailor-made therapy? Br J Cancer. 2007;97:145–51. doi: 10.1038/sj.bjc.6603860. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Yonemori K, Ueno H, Okusaka T, et al. Severe drug toxicity associated with a single-nucleotide polymorphism of the cytidine deaminase gene in a Japanese cancer patient treated with gemcitabine plus cisplatin. Clin Cancer Res. 2005;11:2620–4. doi: 10.1158/1078-0432.CCR-04-1497. [DOI] [PubMed] [Google Scholar]
- 17.Sugiyama E, Kaniwa N, Kim SR, et al. Pharmacokinetics of gemcitabine in Japanese cancer patients: the impact of a cytidine deaminase polymorphism. J Clin Oncol. 2007;25:32–42. doi: 10.1200/JCO.2006.06.7405. [DOI] [PubMed] [Google Scholar]
- 18.Varadhachary GR, Wolff RA, Crane CH, et al. Preoperative Gemcitabine and Cisplatin Followed by Gemcitabine-Based Chemoradiation for Resectable Adenocarcinoma of the Pancreatic Head. J Clin Oncol. 2008;26:3487–95. doi: 10.1200/JCO.2007.15.8642. [DOI] [PubMed] [Google Scholar]
- 19.Evans DB, Varadhachary GR, Crane CH, et al. Preoperative Gemcitabine-Based Chemoradiation for Patients With Resectable Adenocarcinoma of the Pancreatic Head. J Clin Oncol. 2008;26:3496–502. doi: 10.1200/JCO.2007.15.8634. [DOI] [PubMed] [Google Scholar]
- 20.Evans DB, Rich TA, Byrd DR, et al. Preoperative chemoradiation and pancreaticoduodenectomy for adenocarcinoma of the pancreas. Arch Surg. 1992;127:1335–9. doi: 10.1001/archsurg.1992.01420110083017. [DOI] [PubMed] [Google Scholar]
- 21.Wacholder S, Chanock S, Garcia-Closas M, El ghormli L, Rothman N. Assessing the probability that a positive report is false: an approach for molecular epidemiology studies. J Natl Cancer Inst. 2004;96:434–42. doi: 10.1093/jnci/djh075. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Dong X, Jiao L, Li Y, et al. Significant Associations of Mismatch Repair Gene Polymorphisms with Clinical Outcome of Pancreatic Cancer. J Clin Oncol. 2009;27:1592–9. doi: 10.1200/JCO.2008.20.1111. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Yue L, Saikawa Y, Ota K, et al. A functional single-nucleotide polymorphism in the human cyticine deaminase gene contributing to ara-C sensitivity. Parmacogenetics. 2003;13:29–38. doi: 10.1097/00008571-200301000-00005. [DOI] [PubMed] [Google Scholar]
- 24.Schröder JK, Kirch C, Seeber S, Schütte J. Structural and functional analysis of the cytidine deaminase gene in patients with acute myeloid leukaemia. Br J Haematol. 1998;103:1096–103. doi: 10.1046/j.1365-2141.1998.01084.x. [DOI] [PubMed] [Google Scholar]
- 25.Gilbert JA, Salavaggione OE, Ji Y, et al. Gemcitabine pharmacogenomics: cytidine deaminase and deoxycytidylate deaminase gene resequencing and functional genomics. Clin Cancer Res. 2006;12:1794–803. doi: 10.1158/1078-0432.CCR-05-1969. [DOI] [PubMed] [Google Scholar]
- 26.Fitzgerald SM, Goyal RK, Osborne WRA, Roy JD, Wilson JW, Ferrell RE. Identification of functional single nucleotide polymorphism haplotypes in the cytidine deaminase promoter. Hum Genet. 2006;119:276–83. doi: 10.1007/s00439-006-0142-0. [DOI] [PubMed] [Google Scholar]
- 27.Kroep JR, Loves WJP, Wilt CLVD, et al. Mol Cancer Ther. 2002;1:371–6. [PubMed] [Google Scholar]
- 28.Achiwa H, Oguri T, Sato S, et al. Determinations of sensitivity and resistance to gemcitabine: The roles of human equilibrative nucleoside transporter 1 and deoxycytidine kinase in non-small cell lung cancer. Cancer Sci. 2004;95:753–7. doi: 10.1111/j.1349-7006.2004.tb03257.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Jordheim LP, Galmarini CM, Dumontet C. Gemcitabine resistance due to deoxycytidine kinase deficiency can be reverted by fruitfly deoxynucleoside kinase, DmdNK, in human uterine sarcoma cells. Cancer Chemother Pharmacol. 2006;58:547–54. doi: 10.1007/s00280-006-0195-8. [DOI] [PubMed] [Google Scholar]
- 30.Shi JY, Shi ZZ, Zhang SJ, et al. Association between single nucleotide polymorphisms in deoxycytidine kinase and treatment response among acute myeloid leukaemia patients. Pharmacogenetics. 2004;14:759–68. doi: 10.1097/00008571-200411000-00007. [DOI] [PubMed] [Google Scholar]
- 31.Spratlin J, Sangha R, Glubrecht D, Dabbagh L, Young JD, Dumontet C, et al. The absence of human equilibrative nucleotide transporter 1 is associated with reduced survival in patients with gemcitabine-treated pancreas adenocarcinoma. Clin Cancer Res. 2004;10:6956–61. doi: 10.1158/1078-0432.CCR-04-0224. [DOI] [PubMed] [Google Scholar]
- 32.Myers SN, Goyal RK, Roy JD, Fairfull LD, Wilson JW, Ferrell RE. Functional single nucleotide polymorphism haplotypes in the human equilibrative nucleotide transporter1. Pharmacogenet Genomics. 2006;16:315–20. doi: 10.1097/01.fpc.0000189804.41962.15. [DOI] [PubMed] [Google Scholar]
- 33.Kim SR, Saito Y, Maekawa K, et al. Thirty novel genetic variations in the SLC29A1 gene encoding human equilibrative nucleotide transporter 1(hENT1) Drug Metab Pharmacokinet. 2006;21:248–56. doi: 10.2133/dmpk.21.248. [DOI] [PubMed] [Google Scholar]
- 34.Osato DH, Huang CC, Kawamoto M, et al. Functional characterization in yeast of genetic variants in the human equilibrative nucleoside transporter, ENT1. Pharmacogenetics. 2003;13:297–301. doi: 10.1097/00008571-200305000-00010. [DOI] [PubMed] [Google Scholar]
- 35.Farrel JJ, Elsaleh H, Garcia M, et al. Human equilibrium nucleoside transporter 1 levels predict response to gemcitabine in patients with pancreatic cancer. Gastroenterology. 2008;136:187–95. doi: 10.1053/j.gastro.2008.09.067. [DOI] [PubMed] [Google Scholar]
- 36.Nakano Y, Tanno S, Koizumi K, et al. Gemcitabine chemoresistance and molecular markers associated with gemcitabine transport and metabolism in human pancreatic cancer cells. Br J Cancer. 2007;96:457–63. doi: 10.1038/sj.bjc.6603559. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Kimchi-Sarfaty C, Oh JM, Kim IW, et al. A “Silent” polymorphism in the MDR1 gene changes substrate specificity. Science. 2007;315:525–8. doi: 10.1126/science.1135308. [DOI] [PubMed] [Google Scholar]