Drug-Metabolizing Enzyme Polymorphisms Predict Clinical Outcome in a Node-Positive Breast Cancer Cohort
http://www.100md.com
《临床肿瘤学》
the Department of Biostatistics and Epidemiology
Division of Hematology/Oncology, Department of Medicine
Abramson Cancer Center
Abramson Family Research Institute
Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania School of Medicine
Division of Hematology/Oncology, Children's Hospital of Philadelphia, Philadelphia, PA
ABSTRACT
PURPOSE: Adjuvant chemotherapy cures only a subset of women with nonmetastatic breast cancer. Genotypes in drug-metabolizing enzymes, including functional polymorphisms in cytochrome P450 (CYP) and glutathione S-transferases (GST), may predict treatment-related outcomes.
PATIENTS AND METHODS: We examined CYP3A4*1B, CYP3A5*3, and deletions in GST μ (GSTM1) and (GSTT1), as well as a priori–defined combinations of polymorphisms in these genes. Using a cohort of 90 node-positive breast cancer patients who received anthracycline-based adjuvant chemotherapy followed by high-dose multiagent chemotherapy with stem-cell rescue, we estimated the effect of genotype and other known prognostic factors on disease-free survival (DFS) and overall survival (OS).
RESULTS: Patients who carried homozygous CYP3A4*1B and CYP3A5*3 variants and did not carry homozygous deletions in both GSTM1 and GSTT1 (denoted low-drug genotype group) had a 4.9-fold poorer DFS (P = .021) and a four-fold poorer OS (P = .031) compared with individuals who did not carry any CYP3A4*1B or CYP3A5*3 variants but had deletions in both GSTT1 and GSTM1 (denoted high-drug genotype group). After adjustment for other significant prognostic factors, the low-drug genotype group retained a significantly poorer DFS (hazard ratio [HR] = 4.9; 95% CI, 1.7 to 14.6; P = .004) and OS (HR = 4.8; 95% CI, 1.8 to 12.9; P = .002) compared with the high- and intermediate-drug combined genotype group. In the multivariate model, having low-drug genotype group status had a greater impact on clinical outcome than estrogen receptor status.
CONCLUSION: Combined genotypes at CYP3A4, CYP3A5, GSTM1, and GSTT1 influence the probability of treatment failure after high-dose adjuvant chemotherapy for node-positive breast cancer.
INTRODUCTION
Combination chemotherapy using cyclophosphamide-containing regimens is integral to standard adjuvant therapy of node-positive breast cancer, resulting in significant improvements in disease-free survival (DFS) and overall survival (OS).1 Doses of cyclophosphamide in the 600 mg/m2 range remain the standard of care, although autologous transplantation approaches have typically used higher doses.2,3 Although these high-dose studies have not resulted in improved outcomes compared with standard-dose regimens,4,5 such studies provide an opportunity to study the effects of individual variation in drug-metabolizing enzyme (DME) genes with clinical outcome.
Although many factors determine the efficacy of chemotherapy, interindividual genetic differences in DME activity are likely to contribute to systemic drug levels and tissue exposure to active drug metabolites.6 Cyclophosphamide metabolism serves as a paradigm for the role of DMEs to predict treatment response. The inactive prodrug undergoes a sequence of activating steps in the liver,7 as shown in Figure 1. The inactive prodrug undergoes a sequence of activating steps in the liver,7 which is initiated by a phase I 4-hydroxylation reaction that is catalyzed by multiple cytochrome P450 (CYP) enzymes,8 including CYPs 3A4 and 3A5, among others.9,10 The 4-hydroxy-cyclophosphamide intermediate, in equilibrium with its tautomer aldophosphamide, is either further activated to the alkylating compound phosphoramide mustard (through B-elimination) or inactivated by a variety of mechanisms, including phase II conjugation with thiols or sulfates through the glutathione S-transferase (GST) multigene family11,12 and generation of chloro- and dechlorocyclophosphamide via CYP3A4 or carboxycyclophosphamide via aldehyde dehydrogenase (ALDH1).
Several functional single nucleotide polymorphisms (SNPs) have been described in the CYP3A genes that may impact cyclophosphamide metabolism. CYP3A4*1B contains an A to G transition mutation that alters the 10–base pair nifedipine-specific element, which is located –287 to –296 base pairs from the transcription start site of the CYP3A4 gene13 and may be associated with increased transcriptional activity.14 SNPs in CYP3A5*3 and CYP3A5*6 cause alternative splicing and protein truncation that results in the absence of CYP3A5 activity.15 GST μ (GSTM1) and (GSTT1) variants have been described in which homozygous deletions result in a null phenotype, leading to complete absence of enzyme activity.16,17
Thiotepa is also hepatically metabolized via the CYP system to the active metabolite N,N',N''triethylene phosphoramide (TEPA) and a less active metabolite monochloroTEPA. Fifty percent to 80% of thiotepa administered is converted to these active compounds. A small additional amount of thiotepa is converted to thiotepa-mercapturate via GSTs A1 and P1, but this metabolite has only weak alkylating activity. CYP3A4 plays the major role in the metabolism of TEPA, as evidenced in studies of liver microsomes18 in which thiotepa incubated with cloned, expressed CYP3A4 showed turnover to TEPA, whereas incubations with cloned, expressed CYP2A6 showed no TEPA production.18
On the basis of these preclinical and clinical data, we hypothesized that a priori–defined combinations of genetic variants in the major DMEs involved in cyclophosphamide and thiotepa predict interindividual variability in drug metabolism and subsequent treatment response. To test this hypothesis, we used clinical data and stem-cell DNA from a phase II trial of high-dose cyclophosphamide and thiotepa for node-positive breast cancer to examine the impact of genetic variability in CYP3A4, CYP3A5, GSTM1, and GSTT1 on clinical outcome.19
PATIENTS AND METHODS
After institutional review board approval, we evaluated 124 patients with high-risk, node-positive breast cancer ( four positive nodes) who had been previously enrolled onto one of two clinical trials at the University of Pennsylvania Cancer Center from June 1992 through December 1997.4,19 The majority of patients identified (n = 119) were enrolled onto a phase II, single-institution protocol, the results of which have been published previously.19 Five additional patients were treated on the high-dose therapy arm of an Eastern Cooperative Oncology Group protocol (E2190) through the University of Pennsylvania Cancer Center during the same time period. Eligibility criteria and treatment regimens were identical in both of these studies. Patients who provided consent for these protocols had concurrently consented to the use of their stem-cell materials, including DNA, for clinical research.
Stored stem-cell aliquots from 92 (74%) of the 124 eligible patients were available for the current study. The 32 patients without stem-cell specimens did not differ significantly from patients who did provide specimens with respect to age, tumor size, or number of positive lymph nodes. After excluding two patients who did not have clinical follow-up, a final study cohort of 90 patients was available for the current study.
All patients received conventional adjuvant chemotherapy with an anthracycline-containing regimen followed by bone marrow and/or peripheral-blood stem-cell harvest. Patients then received concurrent cyclophosphamide 1,500 mg/m2 (continuous infusion daily for 4 days) and thiotepa 200 mg/m2 (continuous infusion daily for 4 days). Stem-cell reinfusion, radiotherapy, and hormonal therapy were administered as previously reported.19,20 Patients were observed from the time of treatment until disease recurrence and death.
Genomic DNA was extracted from residual stem-cell or bone marrow samples available from the University of Pennsylvania Department of Pathology. Samples were identified only by study number, and thus, clinical outcome data was not available to laboratory personnel. No patient identifiers were linked to clinical data. DNA isolation was performed using the Purgene DNA Isolation Kit (Gentra, Inc, Minneapolis, MN), as described previously.20 A multiplex pyrosequencing assay was used for CYP3A4*1B and CYP3A5*6; a simplex pyrosequencing assay was used for CYP3A5*3; and homozygous deletions at GSTM1 and GSTT1 were identified by multiplex polymerase chain reaction, followed by visualization on a 4% metaphor gel, as previously reported.21,22
We generated genotype combinations that would be predicted to result in varying tissue concentrations of active cyclophosphamide metabolite. Table 1 lists the genotype combinations that were used in this analysis, based on known functional significance of the genetic variants studied.14-17 First, because of known linkage disequilibrium between CYP3A4*1B and CYP3A5*3,23 these genotypes were combined into a single variable before combining them with GST genotype for the prospectively defined drug concentration group. Presence of the variant allele in either CYP3A4 or CYP3A5*3 (n = 13) defined one group, whereas absence of the variant allele in both CYP3A4 and CYP3A5*3 (n = 77) constituted another group. Forty-eight patients had null genotypes for GSTM1, and 15 patients had null genotypes for GSTT1. First, we defined a high-drug genotype group, which was composed of patients who carried nonvariant CYP3A4 (*1A/*1A) and CYP3A5 (*1/*1) and carried homozygous deletions in both GSTM1 and GSTT1, resulting in the combined GST genotype null/null. These patients should be capable of normal 4-hydroxylation, resulting in biologically active cyclophosphamide metabolites, but may have reduced rates of 4-OH-cyclophosphamide conjugation because of absence of GSTT1/GSTM1 activity. Second, a low-drug genotype group was defined, which was composed of patients who carried any variant in CYP3A5*3 and CYP3A4*1B genotypes and nondeleted genotypes for both GSTM1 and GSTT1 (GST non-null/non-null). This group was predicted to have low rates of 4-hydroxylation (because of decreased CYP3A5 activity) but enhanced CYP3A4-inactivating activity and normal GST-inactivating activity. Third, an intermediate-drug genotype group was composed of patients who carried only one CYP variant (either 3A4*1B or 3A5*3) and discordance in GSTM1 and GSTT1. These patients would be expected to have active cyclophosphamide concentrations intermediate to the extreme groups and, thus, would be predicted to have intermediate clinical outcomes.
Descriptive statistics, including medians and ranges for continuous measures and frequencies and percentages for categoric measures, were computed. The Kaplan-Meier method was used to estimate DFS and OS. DFS was defined as days from stem-cell reinfusion to breast cancer recurrence, death from any cause, or last patient contact. OS was defined as days from stem-cell reinfusion to death from any cause or last patient contact. Patients were censored for recurrence if a new primary breast cancer (invasive or noninvasive) was diagnosed, although this occurred in only one patient. Exact log-rank tests were used to test for differences in DFS and OS functions among analytic groups. Exact methods were used because the accuracy of P values from asymptotic tests are questionable with small group sizes.24 Cox proportional hazards regression analysis25 was used to obtain estimates of the hazard ratios (HRs) for individual genotypes and combination groups, adjusting for other prognostic variables, and to evaluate interactions. A reduced multivariate Cox proportional hazards model was obtained by eliminating prognostic variables from the full multivariate model that did not contribute significant predictive information for DFS and OS. Although we defined a priori hypotheses specifying the anticipated direction of genotype effects, all tests of significance were two sided. Analyses were performed with StatXact (Version 6.0; Cytel Software Corporation, Cambridge, MA) and STATA (Release 8; Stata Corporation, College Station, TX).
RESULTS
Characteristics of the 90 patients in the study cohort are listed in Table 2 . As of May 2002, 36 patients (40%) had a breast cancer recurrence, and 52 patients (58%) remained disease free; recurrence status was unknown in two patients. Thirty-five patients (39%) were confirmed as dead, 54 patients (60%) remained alive at the end of follow-up, and the survival status of one patient (1%) was unknown. At the time of this analysis, the median DFS of the cohort as a whole was 37 months (range, 0.76 to 107 months), and the median OS for the group was 44.5 months. All patients in the current study were white.
Table 3 lists allele frequencies and associations between both individual and a priori–defined genotype combinations on both DFS and OS in the study cohort. At each locus, no more than 3% of specimens could not be genotyped. Although we included CYP3A5*6 in our genotyping, no CYP3A5*6 variants were identified, and thus, this SNP was excluded from further analysis.
Patients who fell into the low-drug genotype group had a significantly decreased DFS and OS distribution function compared with both the high-drug and intermediate-drug genotype groups in univariate analysis (P = .077 for DFS, and P = .032 for OS). The Kaplan-Meier curves reflect these differences in both DFS and OS (Fig 2). Because the high-drug and intermediate-drug groups did not differ significantly in either DFS or OS (P = .964 for DFS, and P = .755 for OS), we combined these groups into a single variable and compared them with the low-drug group (Table 3). When compared with the combined high- and intermediate-drug group, the low-drug group clearly defines a population with a significantly decreased DFS and OS (P = .018 for DFS, and P = .006 for OS). The CYP3A4 and CYP3A5 genotypes were each marginally associated with both DFS and OS, whereas the GSTT1 and GSTM1 genotypes were not. In further exploration of various genotype combinations, only the combination of CYP3A4, CYP3A5, and GSTM1 genotypes remained significantly associated with outcome (data not shown).
Table 4 lists both the unadjusted and adjusted HRs from the univariate and reduced multivariate Cox proportional hazards models for DFS and OS using the a priori–defined DME analysis groups. The full multivariate models for DFS and OS included indicator variables for the low- and intermediate-drug genotypes, estrogen receptor status, number of positive lymph nodes, tumor size, and age. After reduction of the full multivariate model, only the DME genotype variables and estrogen receptor status remained in both the reduced DFS and OS multivariate Cox models as independent prognostic factors. In the reduced models, patients with the low-drug genotype retained a significantly higher risk of recurrence (HR = 5.4; 95% CI, 1.4 to 20.9) and death (HR = 4.5; 95% CI, 1.3 to 16.3) compared with patients with the high-drug genotype. The risk of recurrence and death in the intermediate-drug genotype group was not significantly greater than the risk in the high-drug genotype group (DFS: HR = 1.1, P = .872; OS: HR = 0.9, P = .866). Estrogen receptor status remained significant in the reduced multivariate model as well. Number of lymph nodes, tumor size, and age were not significant risk factors for recurrence or death in either univariate or multivariate analyses.
Table 5 lists the univariate and reduced multivariate HRs for DFS and OS Cox proportional hazards models, with the high-drug and intermediate-drug genotypes combined into a single DME group. In the reduced multivariate model, the low-drug genotype group had a significantly higher risk of recurrence (HR = 4.9; 95% CI, 1.7 to 14.6) and death (HR = 4.8; 95% CI, 1.8 to 12.9) compared with the combined high- and intermediate-drug genotype group. Similar to the model in Table 4, estrogen receptor status was the only other prognostic factor that remained significant in the reduced multivariate DFS and OS Cox proportional hazards models.
DISCUSSION
This study provides evidence that genetic variability in CYP3A and GST DMEs is significantly and independently associated with DFS and OS in a cohort of patients with node-positive breast cancer who received high-dose cyclophosphamide as part of an adjuvant therapy regimen. Patients who carried a combination of any variant in CYP3A4 or CYP3A5 and both nondeletion genotypes in GSTM1 and GSTT1 (the low-drug genotype group) had a particularly poor DFS and OS compared with all other patients. Patients in this group were more than four times as likely to experience recurrence or death during the study period. The contribution of genotypes on outcome was of higher statistical significance than tumor estrogen receptor status in multivariate analysis. These findings suggest that genetic variation in drug metabolism may play an important role in chemotherapy efficacy in high-risk breast cancer.
We initially hypothesized that combinations of functional polymorphisms in CYPs 3A4 and 3A5 and GSTs M1 and P1 would lead to distinct phenotypes of drug metabolism that would predict outcome to high-dose therapy in this cohort of node-positive breast cancer patients. We based this hypothesis on data suggesting a dose-response relationship between cyclophosphamide dose and outcome in breast cancer as observed in some previous studies but not others. Preclinical studies suggested a dose-response relationship between both cyclophosphamide26 and thiotepa27 and fractional cell kill,28 which led to the development of high-dose cyclophosphamide regimens, many of which required stem-cell support. However, subsequent clinical trials of escalating doses of adjuvant cyclophosphamide demonstrated that regimens containing greater than 600 mg/m2 did not yield significant improvements in either DFS or OS,2-5,29,30 whereas regimens containing less than 600 mg/m2 resulted in inferior outcomes,31 suggesting a threshold necessary for efficacy. It is this threshold effect that may explain why we saw a substantial difference in outcome between the low-drug group compared with the other groups but no significant difference in outcomes between the intermediate- and high-drug groups.
CYP genotypes as predictors of breast cancer treatment efficacy have not been previously reported. Only three prior studies have examined GSTM1 and GSTT1 genotypes and outcome. Ambrosone et al32 genotyped 251 women with incident breast cancer identified through the Arkansas Cancer Research Center Tumor Registry. Patients in this study represented all stages of disease, and no detailed information was available regarding receipt of chemotherapy. After adjusting for age, race, and stage, GSTM1- and GSTT1-null genotypes predicted significantly better DFS and OS, both individually and in combination. Nedelcheva Kristensen et al33 found similar results in a study of 239 breast cancer patients; in the study, patients who were homozygous for the deleted GSTM1 allele were also found to have a significantly shorter OS (P = .036). Conversely, Lizard-Nacol et al34 found no effect of GSTM1-null genotype on DFS or OS among 92 women with advanced breast cancer who had received cyclophosphamide, doxorubicin, and fluorouracil. In contrast to these previous studies, our study is the only one to examine adjuvant therapy in a population of patients with a relatively uniform recurrence risk, all of whom had the identical chemotherapy regimen, providing a homogeneous patient population in which to study treatment-related genotypes and outcomes.
Other CYP enzymes have been found to have significant oxazaphosphorine 4-hydroxylase activity in vitro, including 2A6, 2B6, 2C9, 2C18, and 2C19. Of these, CYP2B6 has been shown in in vitro studies to have the highest level of activity9,35; however, CYP2B6 represents only a small proportion of total in vivo hepatic CYP (< 1% to 5% of total P450 in human liver)36,37 and is, therefore, likely to make only a minor contribution to hepatic cyclophosphamide metabolism. Thus, genotyping for CYP2B6 was not included in this model, in which we sought to determine the effects of major DME polymorphisms in vivo. Although CYP2B6 may contribute a small degree of additional precision to this model, determination of the contribution of CYP2B6 requires a much greater sample size than what was available in this study. Additional studies incorporating larger sample sizes will be necessary to determine the additional prognostic value of CYP2B6 and other CYPs in this setting.
There are several important limitations to this study that suggest appropriate caution be taken in its interpretation. In our study population, the low-drug group was quite small (n = 5 in the DFS analysis; n = 6 in the OS analysis) because of the limited sample size and the a priori–defined analysis groups. We attempted to limit the type I error of our study by using an exact log-rank test, and our results remain significant for group comparisons. However, confirmation in a larger cohort of patients is necessary and is currently underway. Our study population also had a high risk of recurrence, limiting generalizability to patients with a more favorable prognosis. In addition, the patients in this study cohort all received high-dose therapy, which is a regimen that has been largely abandoned after the failure of clinical trials to support improved efficacy with a high-dose therapy approach. However, patients receiving high-dose therapy with autologous transplantation do have DFS and OS comparable to patients receiving standard-dose adjuvant therapy,4 and these end points in the current study were in line with those findings.19 Only further studies in a standard-dose adjuvant population will confirm that these results are applicable in a lower dose setting, and these studies are currently underway.
Toxicity is an important component of treatment outcome when it leads to dose modification, reduction, or early cessation of treatment. We did not have adequate statistical power to address the relationship of genotype with treatment toxicities, but we did examine the genotypes of the patients who died during the time frame between administration of high-dose therapy and engraftment, presumably from acute complications of therapy. Of the three patients in this group, two had genotype combinations that would have placed them in the low-drug genotype group, and one would have been placed in the intermediate-drug genotype group (data not shown).
Other DME polymorphisms are likely to be important in the metabolism of the drugs received by patients in this study. Additional CYP (2B6, 2C9, and 2C19) and GST (P1 and A1) polymorphisms and others may play a role in the metabolism of cyclophosphamide, doxorubicin, and the taxanes, which are the drugs most commonly used in the adjuvant breast cancer setting. Our analysis of these and other polymorphisms is currently underway. Serum or tumor samples were not available in these patients, precluding our ability to conduct direct measurement of drug levels in these compartments and precluding investigation into biologic mechanisms by which DME polymorphisms may act to influence clinical outcome after high-dose therapy. Future prospective studies are underway to address these issues.
This study provides important new information about the potential role of DME polymorphisms in outcome after adjuvant therapy for breast cancer. Confirmation of these findings and supportive mechanistic data will ultimately allow the potential for DME genotyping to be realized in the clinic to individualize and optimize breast cancer therapy.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
NOTES
Supported by grant No. K23-CA81009 (A.D.), University of Pennsylvania Cancer Center Pilot Project (A.D.), and National Cancer Institute Core Grant (J.G.).
Presented at the 94th Annual Meeting of the American Association of Cancer Research, Washington, DC, July 11-14, 2003 and the 26th Annual San Antonio Breast Cancer Symposium, San Antonio, TX, December 3-6, 2003.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Early Breast Cancer Trialists' Collaborative Group: Polychemotherapy for early breast cancer: An overview of the randomised trials—Early Breast Cancer Trialists' Collaborative Group. Lancet 352:930-942, 1998
Aapro MS: Adjuvant therapy of primary breast cancer: A review of key findings from the 7th International Conference, St. Gallen, February 2001. Oncologist 6:376-385, 2001
Antman KH: Randomized trials of high dose chemotherapy for breast cancer. Biochim Biophys Acta 1471:M89-M98, 2001
Tallman MS, Gray R, Robert NJ, et al: Conventional adjuvant chemotherapy with or without high-dose chemotherapy and autologous stem-cell transplantation in high-risk breast cancer. N Engl J Med 349:17-26, 2003
Zander A, Kroger N, Schmoor C, et al: High-dose chemotherapy with autologous hematopoietic stem-cell support compared with standard-dose chemotherapy in breast cancer patients with 10 or more positive lymph nodes: First results of a randomized trial. J Clin Oncol 22:2273-2283, 2004
West WL, Knight EM, Pradhan S, et al: Interpatient variability: Genetic predisposition and other genetic factors. J Clin Pharmacol 37:635-648, 1997
Boddy AV, Furtun Y, Sardas S, et al: Individual variation in the activation and inactivation of metabolic pathways of cyclophosphamide. J Natl Cancer Inst 84:1744-1748, 1992
Chang TK, Yu L, Maurel P, et al: Enhanced cyclophosphamide and ifosfamide activation in primary human hepatocyte cultures: Response to cytochrome P-450 inducers and autoinduction by oxazaphosphorines. Cancer Res 57:1946-1954, 1997
Roy P, Yu LJ, Crespi CL, et al: Development of a substrate-activity based approach to identify the major human liver P-450 catalysts of cyclophosphamide and ifosfamide activation based on cDNA-expressed activities and liver microsomal P-450 profiles. Drug Metab Dispos 27:655-666, 1999
Ren S, Yang JS, Kalhorn TF, et al: Oxidation of cyclophosphamide to 4-hydroxycyclophosphamide and deschloroethylcyclophosphamide in human liver microsomes. Cancer Res 57:4229-4235, 1997
Ludeman SM: The chemistry of the metabolites of cyclophosphamide. Curr Pharm Des 5:627-643, 1999
Dirven HA, van Ommen B, van Bladeren PJ: Involvement of human glutathione S-transferase isoenzymes in the conjugation of cyclophosphamide metabolites with glutathione. Cancer Res 54:6215-6220, 1994
Walker AH, Jaffe JM, Gunasegaram S, et al: Characterization of an allelic variant in the nifedipine-specific element of CYP3A4: Ethnic distribution and implications for prostate cancer risk—Mutations in brief no. 191. Online. Hum Mutat 12:289, 1998
Amirimani B, Ning B, Deitz AC, et al: Increased transcriptional activity of the CYP3A4*1B promoter variant. Environ Mol Mutagen 42:299-305, 2003
Kuehl P, Zhang J, Lin Y, et al: Sequence diversity in CYP3A promoters and characterization of the genetic basis of polymorphic CYP3A5 expression. Nat Genet 27:383-391, 2001
Seidegard J, Pero RW: The genetic variation and the expression of human glutathione transferase mu. Klin Wochenschr 66: 125-126, 1988 (suppl 11)
Seidegard J, Vorachek WR, Pero RW, et al: Hereditary differences in the expression of the human glutathione transferase active on trans-stilbene oxide are due to a gene deletion. Proc Natl Acad Sci U S A 85:7293-7297, 1988
Jacobson PA, Green K, Birnbaum A, et al: Cytochrome P450 isozymes 3A4 and 2B6 are involved in the in vitro human metabolism of thiotepa to TEPA. Cancer Chemother Pharmacol 49:461-467, 2002
Moore HC, Mick R, Solin LJ, et al: Autologous stem-cell transplant after conventional dose adjuvant chemotherapy for high-risk breast cancer: Impact on the delivery of local-regional radiation therapy. Ann Oncol 10:929-936, 1999
DeMichele A, Martin AM, Mick R, et al: Interleukin-6-174GC polymorphism is associated with improved outcome in high-risk breast cancer. Cancer Res 63:8051-8056, 2003
Aplenc R, Glatfelter W, Han P, et al: CYP3A genotypes and treatment response in paediatric acute lymphoblastic leukaemia. Br J Haematol 122:240-244, 2003
Davies SM, Robison LL, Buckley JD, et al: Glutathione S-transferase polymorphisms in children with myeloid leukemia: A Children's Cancer Group study. Cancer Epidemiol Biomarkers Prev 9:563-566, 2000
Dally H, Bartsch H, Jager B, et al: Genotype relationships in the CYP3A locus in Caucasians. Cancer Lett 207:95-99, 2004
Wilcoxan F: Individual comparisons by ranking methods. Biometrics 1: XXX, 1945
Cox DR: Regression models and life tables. J R Stat Soc B 34:187-220, 1972
Frei E III, Teicher BA, Holden SA, et al: Preclinical studies and clinical correlation of the effect of alkylating dose. Cancer Res 48:6417-6423, 1988
Teicher BA, Holden SA, Cucchi CA, et al: Combination of N,N',N''-triethylenethiophosphoramide and cyclophosphamide in vitro and in vivo. Cancer Res 48:94-100, 1988
Jusko WJ: Pharmacodynamics of chemotherapeutic effects: Dose-time-response relationships for phase-nonspecific agents. J Pharm Sci 60:892-895, 1971
Fisher B, Anderson S, Wickerham DL, et al: Increased intensification and total dose of cyclophosphamide in a doxorubicin-cyclophosphamide regimen for the treatment of primary breast cancer: Findings from National Surgical Adjuvant Breast and Bowel Project B-22. J Clin Oncol 15:1858-1869, 1997
Fisher B, Anderson S, DeCillis A, et al: Further evaluation of intensified and increased total dose of cyclophosphamide for the treatment of primary breast cancer: Findings from National Surgical Adjuvant Breast and Bowel Project B-25. J Clin Oncol 17:3374-3388, 1999
Wood WC, Budman DR, Korzun AH, et al: Dose and dose intensity of adjuvant chemotherapy for stage II, node-positive breast carcinoma. N Engl J Med 330:1253-1259, 1994
Ambrosone CB, Sweeney C, Coles BF, et al: Polymorphisms in glutathione S-transferases (GSTM1 and GSTT1) and survival after treatment for breast cancer. Cancer Res 61:7130-7135, 2001
Nedelcheva Kristensen V, Andersen TI, Erikstein B, et al: Single tube multiplex polymerase chain reaction genotype analysis of GSTM1, GSTT1 and GSTP1: Relation of genotypes to TP53 tumor status and clinicopathological variables in breast cancer patients. Pharmacogenetics 8:441-447, 1998
Lizard-Nacol S, Coudert B, Colosetti P, et al: Glutathione S-transferase M1 null genotype: Lack of association with tumour characteristics and survival in advanced breast cancer. Breast Cancer Res 1:81-87, 1999
Chang TK, Weber GF, Crespi CL, et al: Differential activation of cyclophosphamide and ifosphamide by cytochromes P-450 2B and 3A in human liver microsomes. Cancer Res 53:5629-5637, 1993
Ekins S, Vandenbranden M, Ring BJ, et al: Further characterization of the expression in liver and catalytic activity of CYP2B6. J Pharmacol Exp Ther 286:1253-1259, 1998
Hanna IH, Dawling S, Roodi N, et al: Cytochrome P450 1B1 (CYP1B1) pharmacogenetics: Association of polymorphisms with functional differences in estrogen hydroxylation activity. Cancer Res 60:3440-3444, 2000(Angela DeMichele, Richard)
Division of Hematology/Oncology, Department of Medicine
Abramson Cancer Center
Abramson Family Research Institute
Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania School of Medicine
Division of Hematology/Oncology, Children's Hospital of Philadelphia, Philadelphia, PA
ABSTRACT
PURPOSE: Adjuvant chemotherapy cures only a subset of women with nonmetastatic breast cancer. Genotypes in drug-metabolizing enzymes, including functional polymorphisms in cytochrome P450 (CYP) and glutathione S-transferases (GST), may predict treatment-related outcomes.
PATIENTS AND METHODS: We examined CYP3A4*1B, CYP3A5*3, and deletions in GST μ (GSTM1) and (GSTT1), as well as a priori–defined combinations of polymorphisms in these genes. Using a cohort of 90 node-positive breast cancer patients who received anthracycline-based adjuvant chemotherapy followed by high-dose multiagent chemotherapy with stem-cell rescue, we estimated the effect of genotype and other known prognostic factors on disease-free survival (DFS) and overall survival (OS).
RESULTS: Patients who carried homozygous CYP3A4*1B and CYP3A5*3 variants and did not carry homozygous deletions in both GSTM1 and GSTT1 (denoted low-drug genotype group) had a 4.9-fold poorer DFS (P = .021) and a four-fold poorer OS (P = .031) compared with individuals who did not carry any CYP3A4*1B or CYP3A5*3 variants but had deletions in both GSTT1 and GSTM1 (denoted high-drug genotype group). After adjustment for other significant prognostic factors, the low-drug genotype group retained a significantly poorer DFS (hazard ratio [HR] = 4.9; 95% CI, 1.7 to 14.6; P = .004) and OS (HR = 4.8; 95% CI, 1.8 to 12.9; P = .002) compared with the high- and intermediate-drug combined genotype group. In the multivariate model, having low-drug genotype group status had a greater impact on clinical outcome than estrogen receptor status.
CONCLUSION: Combined genotypes at CYP3A4, CYP3A5, GSTM1, and GSTT1 influence the probability of treatment failure after high-dose adjuvant chemotherapy for node-positive breast cancer.
INTRODUCTION
Combination chemotherapy using cyclophosphamide-containing regimens is integral to standard adjuvant therapy of node-positive breast cancer, resulting in significant improvements in disease-free survival (DFS) and overall survival (OS).1 Doses of cyclophosphamide in the 600 mg/m2 range remain the standard of care, although autologous transplantation approaches have typically used higher doses.2,3 Although these high-dose studies have not resulted in improved outcomes compared with standard-dose regimens,4,5 such studies provide an opportunity to study the effects of individual variation in drug-metabolizing enzyme (DME) genes with clinical outcome.
Although many factors determine the efficacy of chemotherapy, interindividual genetic differences in DME activity are likely to contribute to systemic drug levels and tissue exposure to active drug metabolites.6 Cyclophosphamide metabolism serves as a paradigm for the role of DMEs to predict treatment response. The inactive prodrug undergoes a sequence of activating steps in the liver,7 as shown in Figure 1. The inactive prodrug undergoes a sequence of activating steps in the liver,7 which is initiated by a phase I 4-hydroxylation reaction that is catalyzed by multiple cytochrome P450 (CYP) enzymes,8 including CYPs 3A4 and 3A5, among others.9,10 The 4-hydroxy-cyclophosphamide intermediate, in equilibrium with its tautomer aldophosphamide, is either further activated to the alkylating compound phosphoramide mustard (through B-elimination) or inactivated by a variety of mechanisms, including phase II conjugation with thiols or sulfates through the glutathione S-transferase (GST) multigene family11,12 and generation of chloro- and dechlorocyclophosphamide via CYP3A4 or carboxycyclophosphamide via aldehyde dehydrogenase (ALDH1).
Several functional single nucleotide polymorphisms (SNPs) have been described in the CYP3A genes that may impact cyclophosphamide metabolism. CYP3A4*1B contains an A to G transition mutation that alters the 10–base pair nifedipine-specific element, which is located –287 to –296 base pairs from the transcription start site of the CYP3A4 gene13 and may be associated with increased transcriptional activity.14 SNPs in CYP3A5*3 and CYP3A5*6 cause alternative splicing and protein truncation that results in the absence of CYP3A5 activity.15 GST μ (GSTM1) and (GSTT1) variants have been described in which homozygous deletions result in a null phenotype, leading to complete absence of enzyme activity.16,17
Thiotepa is also hepatically metabolized via the CYP system to the active metabolite N,N',N''triethylene phosphoramide (TEPA) and a less active metabolite monochloroTEPA. Fifty percent to 80% of thiotepa administered is converted to these active compounds. A small additional amount of thiotepa is converted to thiotepa-mercapturate via GSTs A1 and P1, but this metabolite has only weak alkylating activity. CYP3A4 plays the major role in the metabolism of TEPA, as evidenced in studies of liver microsomes18 in which thiotepa incubated with cloned, expressed CYP3A4 showed turnover to TEPA, whereas incubations with cloned, expressed CYP2A6 showed no TEPA production.18
On the basis of these preclinical and clinical data, we hypothesized that a priori–defined combinations of genetic variants in the major DMEs involved in cyclophosphamide and thiotepa predict interindividual variability in drug metabolism and subsequent treatment response. To test this hypothesis, we used clinical data and stem-cell DNA from a phase II trial of high-dose cyclophosphamide and thiotepa for node-positive breast cancer to examine the impact of genetic variability in CYP3A4, CYP3A5, GSTM1, and GSTT1 on clinical outcome.19
PATIENTS AND METHODS
After institutional review board approval, we evaluated 124 patients with high-risk, node-positive breast cancer ( four positive nodes) who had been previously enrolled onto one of two clinical trials at the University of Pennsylvania Cancer Center from June 1992 through December 1997.4,19 The majority of patients identified (n = 119) were enrolled onto a phase II, single-institution protocol, the results of which have been published previously.19 Five additional patients were treated on the high-dose therapy arm of an Eastern Cooperative Oncology Group protocol (E2190) through the University of Pennsylvania Cancer Center during the same time period. Eligibility criteria and treatment regimens were identical in both of these studies. Patients who provided consent for these protocols had concurrently consented to the use of their stem-cell materials, including DNA, for clinical research.
Stored stem-cell aliquots from 92 (74%) of the 124 eligible patients were available for the current study. The 32 patients without stem-cell specimens did not differ significantly from patients who did provide specimens with respect to age, tumor size, or number of positive lymph nodes. After excluding two patients who did not have clinical follow-up, a final study cohort of 90 patients was available for the current study.
All patients received conventional adjuvant chemotherapy with an anthracycline-containing regimen followed by bone marrow and/or peripheral-blood stem-cell harvest. Patients then received concurrent cyclophosphamide 1,500 mg/m2 (continuous infusion daily for 4 days) and thiotepa 200 mg/m2 (continuous infusion daily for 4 days). Stem-cell reinfusion, radiotherapy, and hormonal therapy were administered as previously reported.19,20 Patients were observed from the time of treatment until disease recurrence and death.
Genomic DNA was extracted from residual stem-cell or bone marrow samples available from the University of Pennsylvania Department of Pathology. Samples were identified only by study number, and thus, clinical outcome data was not available to laboratory personnel. No patient identifiers were linked to clinical data. DNA isolation was performed using the Purgene DNA Isolation Kit (Gentra, Inc, Minneapolis, MN), as described previously.20 A multiplex pyrosequencing assay was used for CYP3A4*1B and CYP3A5*6; a simplex pyrosequencing assay was used for CYP3A5*3; and homozygous deletions at GSTM1 and GSTT1 were identified by multiplex polymerase chain reaction, followed by visualization on a 4% metaphor gel, as previously reported.21,22
We generated genotype combinations that would be predicted to result in varying tissue concentrations of active cyclophosphamide metabolite. Table 1 lists the genotype combinations that were used in this analysis, based on known functional significance of the genetic variants studied.14-17 First, because of known linkage disequilibrium between CYP3A4*1B and CYP3A5*3,23 these genotypes were combined into a single variable before combining them with GST genotype for the prospectively defined drug concentration group. Presence of the variant allele in either CYP3A4 or CYP3A5*3 (n = 13) defined one group, whereas absence of the variant allele in both CYP3A4 and CYP3A5*3 (n = 77) constituted another group. Forty-eight patients had null genotypes for GSTM1, and 15 patients had null genotypes for GSTT1. First, we defined a high-drug genotype group, which was composed of patients who carried nonvariant CYP3A4 (*1A/*1A) and CYP3A5 (*1/*1) and carried homozygous deletions in both GSTM1 and GSTT1, resulting in the combined GST genotype null/null. These patients should be capable of normal 4-hydroxylation, resulting in biologically active cyclophosphamide metabolites, but may have reduced rates of 4-OH-cyclophosphamide conjugation because of absence of GSTT1/GSTM1 activity. Second, a low-drug genotype group was defined, which was composed of patients who carried any variant in CYP3A5*3 and CYP3A4*1B genotypes and nondeleted genotypes for both GSTM1 and GSTT1 (GST non-null/non-null). This group was predicted to have low rates of 4-hydroxylation (because of decreased CYP3A5 activity) but enhanced CYP3A4-inactivating activity and normal GST-inactivating activity. Third, an intermediate-drug genotype group was composed of patients who carried only one CYP variant (either 3A4*1B or 3A5*3) and discordance in GSTM1 and GSTT1. These patients would be expected to have active cyclophosphamide concentrations intermediate to the extreme groups and, thus, would be predicted to have intermediate clinical outcomes.
Descriptive statistics, including medians and ranges for continuous measures and frequencies and percentages for categoric measures, were computed. The Kaplan-Meier method was used to estimate DFS and OS. DFS was defined as days from stem-cell reinfusion to breast cancer recurrence, death from any cause, or last patient contact. OS was defined as days from stem-cell reinfusion to death from any cause or last patient contact. Patients were censored for recurrence if a new primary breast cancer (invasive or noninvasive) was diagnosed, although this occurred in only one patient. Exact log-rank tests were used to test for differences in DFS and OS functions among analytic groups. Exact methods were used because the accuracy of P values from asymptotic tests are questionable with small group sizes.24 Cox proportional hazards regression analysis25 was used to obtain estimates of the hazard ratios (HRs) for individual genotypes and combination groups, adjusting for other prognostic variables, and to evaluate interactions. A reduced multivariate Cox proportional hazards model was obtained by eliminating prognostic variables from the full multivariate model that did not contribute significant predictive information for DFS and OS. Although we defined a priori hypotheses specifying the anticipated direction of genotype effects, all tests of significance were two sided. Analyses were performed with StatXact (Version 6.0; Cytel Software Corporation, Cambridge, MA) and STATA (Release 8; Stata Corporation, College Station, TX).
RESULTS
Characteristics of the 90 patients in the study cohort are listed in Table 2 . As of May 2002, 36 patients (40%) had a breast cancer recurrence, and 52 patients (58%) remained disease free; recurrence status was unknown in two patients. Thirty-five patients (39%) were confirmed as dead, 54 patients (60%) remained alive at the end of follow-up, and the survival status of one patient (1%) was unknown. At the time of this analysis, the median DFS of the cohort as a whole was 37 months (range, 0.76 to 107 months), and the median OS for the group was 44.5 months. All patients in the current study were white.
Table 3 lists allele frequencies and associations between both individual and a priori–defined genotype combinations on both DFS and OS in the study cohort. At each locus, no more than 3% of specimens could not be genotyped. Although we included CYP3A5*6 in our genotyping, no CYP3A5*6 variants were identified, and thus, this SNP was excluded from further analysis.
Patients who fell into the low-drug genotype group had a significantly decreased DFS and OS distribution function compared with both the high-drug and intermediate-drug genotype groups in univariate analysis (P = .077 for DFS, and P = .032 for OS). The Kaplan-Meier curves reflect these differences in both DFS and OS (Fig 2). Because the high-drug and intermediate-drug groups did not differ significantly in either DFS or OS (P = .964 for DFS, and P = .755 for OS), we combined these groups into a single variable and compared them with the low-drug group (Table 3). When compared with the combined high- and intermediate-drug group, the low-drug group clearly defines a population with a significantly decreased DFS and OS (P = .018 for DFS, and P = .006 for OS). The CYP3A4 and CYP3A5 genotypes were each marginally associated with both DFS and OS, whereas the GSTT1 and GSTM1 genotypes were not. In further exploration of various genotype combinations, only the combination of CYP3A4, CYP3A5, and GSTM1 genotypes remained significantly associated with outcome (data not shown).
Table 4 lists both the unadjusted and adjusted HRs from the univariate and reduced multivariate Cox proportional hazards models for DFS and OS using the a priori–defined DME analysis groups. The full multivariate models for DFS and OS included indicator variables for the low- and intermediate-drug genotypes, estrogen receptor status, number of positive lymph nodes, tumor size, and age. After reduction of the full multivariate model, only the DME genotype variables and estrogen receptor status remained in both the reduced DFS and OS multivariate Cox models as independent prognostic factors. In the reduced models, patients with the low-drug genotype retained a significantly higher risk of recurrence (HR = 5.4; 95% CI, 1.4 to 20.9) and death (HR = 4.5; 95% CI, 1.3 to 16.3) compared with patients with the high-drug genotype. The risk of recurrence and death in the intermediate-drug genotype group was not significantly greater than the risk in the high-drug genotype group (DFS: HR = 1.1, P = .872; OS: HR = 0.9, P = .866). Estrogen receptor status remained significant in the reduced multivariate model as well. Number of lymph nodes, tumor size, and age were not significant risk factors for recurrence or death in either univariate or multivariate analyses.
Table 5 lists the univariate and reduced multivariate HRs for DFS and OS Cox proportional hazards models, with the high-drug and intermediate-drug genotypes combined into a single DME group. In the reduced multivariate model, the low-drug genotype group had a significantly higher risk of recurrence (HR = 4.9; 95% CI, 1.7 to 14.6) and death (HR = 4.8; 95% CI, 1.8 to 12.9) compared with the combined high- and intermediate-drug genotype group. Similar to the model in Table 4, estrogen receptor status was the only other prognostic factor that remained significant in the reduced multivariate DFS and OS Cox proportional hazards models.
DISCUSSION
This study provides evidence that genetic variability in CYP3A and GST DMEs is significantly and independently associated with DFS and OS in a cohort of patients with node-positive breast cancer who received high-dose cyclophosphamide as part of an adjuvant therapy regimen. Patients who carried a combination of any variant in CYP3A4 or CYP3A5 and both nondeletion genotypes in GSTM1 and GSTT1 (the low-drug genotype group) had a particularly poor DFS and OS compared with all other patients. Patients in this group were more than four times as likely to experience recurrence or death during the study period. The contribution of genotypes on outcome was of higher statistical significance than tumor estrogen receptor status in multivariate analysis. These findings suggest that genetic variation in drug metabolism may play an important role in chemotherapy efficacy in high-risk breast cancer.
We initially hypothesized that combinations of functional polymorphisms in CYPs 3A4 and 3A5 and GSTs M1 and P1 would lead to distinct phenotypes of drug metabolism that would predict outcome to high-dose therapy in this cohort of node-positive breast cancer patients. We based this hypothesis on data suggesting a dose-response relationship between cyclophosphamide dose and outcome in breast cancer as observed in some previous studies but not others. Preclinical studies suggested a dose-response relationship between both cyclophosphamide26 and thiotepa27 and fractional cell kill,28 which led to the development of high-dose cyclophosphamide regimens, many of which required stem-cell support. However, subsequent clinical trials of escalating doses of adjuvant cyclophosphamide demonstrated that regimens containing greater than 600 mg/m2 did not yield significant improvements in either DFS or OS,2-5,29,30 whereas regimens containing less than 600 mg/m2 resulted in inferior outcomes,31 suggesting a threshold necessary for efficacy. It is this threshold effect that may explain why we saw a substantial difference in outcome between the low-drug group compared with the other groups but no significant difference in outcomes between the intermediate- and high-drug groups.
CYP genotypes as predictors of breast cancer treatment efficacy have not been previously reported. Only three prior studies have examined GSTM1 and GSTT1 genotypes and outcome. Ambrosone et al32 genotyped 251 women with incident breast cancer identified through the Arkansas Cancer Research Center Tumor Registry. Patients in this study represented all stages of disease, and no detailed information was available regarding receipt of chemotherapy. After adjusting for age, race, and stage, GSTM1- and GSTT1-null genotypes predicted significantly better DFS and OS, both individually and in combination. Nedelcheva Kristensen et al33 found similar results in a study of 239 breast cancer patients; in the study, patients who were homozygous for the deleted GSTM1 allele were also found to have a significantly shorter OS (P = .036). Conversely, Lizard-Nacol et al34 found no effect of GSTM1-null genotype on DFS or OS among 92 women with advanced breast cancer who had received cyclophosphamide, doxorubicin, and fluorouracil. In contrast to these previous studies, our study is the only one to examine adjuvant therapy in a population of patients with a relatively uniform recurrence risk, all of whom had the identical chemotherapy regimen, providing a homogeneous patient population in which to study treatment-related genotypes and outcomes.
Other CYP enzymes have been found to have significant oxazaphosphorine 4-hydroxylase activity in vitro, including 2A6, 2B6, 2C9, 2C18, and 2C19. Of these, CYP2B6 has been shown in in vitro studies to have the highest level of activity9,35; however, CYP2B6 represents only a small proportion of total in vivo hepatic CYP (< 1% to 5% of total P450 in human liver)36,37 and is, therefore, likely to make only a minor contribution to hepatic cyclophosphamide metabolism. Thus, genotyping for CYP2B6 was not included in this model, in which we sought to determine the effects of major DME polymorphisms in vivo. Although CYP2B6 may contribute a small degree of additional precision to this model, determination of the contribution of CYP2B6 requires a much greater sample size than what was available in this study. Additional studies incorporating larger sample sizes will be necessary to determine the additional prognostic value of CYP2B6 and other CYPs in this setting.
There are several important limitations to this study that suggest appropriate caution be taken in its interpretation. In our study population, the low-drug group was quite small (n = 5 in the DFS analysis; n = 6 in the OS analysis) because of the limited sample size and the a priori–defined analysis groups. We attempted to limit the type I error of our study by using an exact log-rank test, and our results remain significant for group comparisons. However, confirmation in a larger cohort of patients is necessary and is currently underway. Our study population also had a high risk of recurrence, limiting generalizability to patients with a more favorable prognosis. In addition, the patients in this study cohort all received high-dose therapy, which is a regimen that has been largely abandoned after the failure of clinical trials to support improved efficacy with a high-dose therapy approach. However, patients receiving high-dose therapy with autologous transplantation do have DFS and OS comparable to patients receiving standard-dose adjuvant therapy,4 and these end points in the current study were in line with those findings.19 Only further studies in a standard-dose adjuvant population will confirm that these results are applicable in a lower dose setting, and these studies are currently underway.
Toxicity is an important component of treatment outcome when it leads to dose modification, reduction, or early cessation of treatment. We did not have adequate statistical power to address the relationship of genotype with treatment toxicities, but we did examine the genotypes of the patients who died during the time frame between administration of high-dose therapy and engraftment, presumably from acute complications of therapy. Of the three patients in this group, two had genotype combinations that would have placed them in the low-drug genotype group, and one would have been placed in the intermediate-drug genotype group (data not shown).
Other DME polymorphisms are likely to be important in the metabolism of the drugs received by patients in this study. Additional CYP (2B6, 2C9, and 2C19) and GST (P1 and A1) polymorphisms and others may play a role in the metabolism of cyclophosphamide, doxorubicin, and the taxanes, which are the drugs most commonly used in the adjuvant breast cancer setting. Our analysis of these and other polymorphisms is currently underway. Serum or tumor samples were not available in these patients, precluding our ability to conduct direct measurement of drug levels in these compartments and precluding investigation into biologic mechanisms by which DME polymorphisms may act to influence clinical outcome after high-dose therapy. Future prospective studies are underway to address these issues.
This study provides important new information about the potential role of DME polymorphisms in outcome after adjuvant therapy for breast cancer. Confirmation of these findings and supportive mechanistic data will ultimately allow the potential for DME genotyping to be realized in the clinic to individualize and optimize breast cancer therapy.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
NOTES
Supported by grant No. K23-CA81009 (A.D.), University of Pennsylvania Cancer Center Pilot Project (A.D.), and National Cancer Institute Core Grant (J.G.).
Presented at the 94th Annual Meeting of the American Association of Cancer Research, Washington, DC, July 11-14, 2003 and the 26th Annual San Antonio Breast Cancer Symposium, San Antonio, TX, December 3-6, 2003.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Early Breast Cancer Trialists' Collaborative Group: Polychemotherapy for early breast cancer: An overview of the randomised trials—Early Breast Cancer Trialists' Collaborative Group. Lancet 352:930-942, 1998
Aapro MS: Adjuvant therapy of primary breast cancer: A review of key findings from the 7th International Conference, St. Gallen, February 2001. Oncologist 6:376-385, 2001
Antman KH: Randomized trials of high dose chemotherapy for breast cancer. Biochim Biophys Acta 1471:M89-M98, 2001
Tallman MS, Gray R, Robert NJ, et al: Conventional adjuvant chemotherapy with or without high-dose chemotherapy and autologous stem-cell transplantation in high-risk breast cancer. N Engl J Med 349:17-26, 2003
Zander A, Kroger N, Schmoor C, et al: High-dose chemotherapy with autologous hematopoietic stem-cell support compared with standard-dose chemotherapy in breast cancer patients with 10 or more positive lymph nodes: First results of a randomized trial. J Clin Oncol 22:2273-2283, 2004
West WL, Knight EM, Pradhan S, et al: Interpatient variability: Genetic predisposition and other genetic factors. J Clin Pharmacol 37:635-648, 1997
Boddy AV, Furtun Y, Sardas S, et al: Individual variation in the activation and inactivation of metabolic pathways of cyclophosphamide. J Natl Cancer Inst 84:1744-1748, 1992
Chang TK, Yu L, Maurel P, et al: Enhanced cyclophosphamide and ifosfamide activation in primary human hepatocyte cultures: Response to cytochrome P-450 inducers and autoinduction by oxazaphosphorines. Cancer Res 57:1946-1954, 1997
Roy P, Yu LJ, Crespi CL, et al: Development of a substrate-activity based approach to identify the major human liver P-450 catalysts of cyclophosphamide and ifosfamide activation based on cDNA-expressed activities and liver microsomal P-450 profiles. Drug Metab Dispos 27:655-666, 1999
Ren S, Yang JS, Kalhorn TF, et al: Oxidation of cyclophosphamide to 4-hydroxycyclophosphamide and deschloroethylcyclophosphamide in human liver microsomes. Cancer Res 57:4229-4235, 1997
Ludeman SM: The chemistry of the metabolites of cyclophosphamide. Curr Pharm Des 5:627-643, 1999
Dirven HA, van Ommen B, van Bladeren PJ: Involvement of human glutathione S-transferase isoenzymes in the conjugation of cyclophosphamide metabolites with glutathione. Cancer Res 54:6215-6220, 1994
Walker AH, Jaffe JM, Gunasegaram S, et al: Characterization of an allelic variant in the nifedipine-specific element of CYP3A4: Ethnic distribution and implications for prostate cancer risk—Mutations in brief no. 191. Online. Hum Mutat 12:289, 1998
Amirimani B, Ning B, Deitz AC, et al: Increased transcriptional activity of the CYP3A4*1B promoter variant. Environ Mol Mutagen 42:299-305, 2003
Kuehl P, Zhang J, Lin Y, et al: Sequence diversity in CYP3A promoters and characterization of the genetic basis of polymorphic CYP3A5 expression. Nat Genet 27:383-391, 2001
Seidegard J, Pero RW: The genetic variation and the expression of human glutathione transferase mu. Klin Wochenschr 66: 125-126, 1988 (suppl 11)
Seidegard J, Vorachek WR, Pero RW, et al: Hereditary differences in the expression of the human glutathione transferase active on trans-stilbene oxide are due to a gene deletion. Proc Natl Acad Sci U S A 85:7293-7297, 1988
Jacobson PA, Green K, Birnbaum A, et al: Cytochrome P450 isozymes 3A4 and 2B6 are involved in the in vitro human metabolism of thiotepa to TEPA. Cancer Chemother Pharmacol 49:461-467, 2002
Moore HC, Mick R, Solin LJ, et al: Autologous stem-cell transplant after conventional dose adjuvant chemotherapy for high-risk breast cancer: Impact on the delivery of local-regional radiation therapy. Ann Oncol 10:929-936, 1999
DeMichele A, Martin AM, Mick R, et al: Interleukin-6-174GC polymorphism is associated with improved outcome in high-risk breast cancer. Cancer Res 63:8051-8056, 2003
Aplenc R, Glatfelter W, Han P, et al: CYP3A genotypes and treatment response in paediatric acute lymphoblastic leukaemia. Br J Haematol 122:240-244, 2003
Davies SM, Robison LL, Buckley JD, et al: Glutathione S-transferase polymorphisms in children with myeloid leukemia: A Children's Cancer Group study. Cancer Epidemiol Biomarkers Prev 9:563-566, 2000
Dally H, Bartsch H, Jager B, et al: Genotype relationships in the CYP3A locus in Caucasians. Cancer Lett 207:95-99, 2004
Wilcoxan F: Individual comparisons by ranking methods. Biometrics 1: XXX, 1945
Cox DR: Regression models and life tables. J R Stat Soc B 34:187-220, 1972
Frei E III, Teicher BA, Holden SA, et al: Preclinical studies and clinical correlation of the effect of alkylating dose. Cancer Res 48:6417-6423, 1988
Teicher BA, Holden SA, Cucchi CA, et al: Combination of N,N',N''-triethylenethiophosphoramide and cyclophosphamide in vitro and in vivo. Cancer Res 48:94-100, 1988
Jusko WJ: Pharmacodynamics of chemotherapeutic effects: Dose-time-response relationships for phase-nonspecific agents. J Pharm Sci 60:892-895, 1971
Fisher B, Anderson S, Wickerham DL, et al: Increased intensification and total dose of cyclophosphamide in a doxorubicin-cyclophosphamide regimen for the treatment of primary breast cancer: Findings from National Surgical Adjuvant Breast and Bowel Project B-22. J Clin Oncol 15:1858-1869, 1997
Fisher B, Anderson S, DeCillis A, et al: Further evaluation of intensified and increased total dose of cyclophosphamide for the treatment of primary breast cancer: Findings from National Surgical Adjuvant Breast and Bowel Project B-25. J Clin Oncol 17:3374-3388, 1999
Wood WC, Budman DR, Korzun AH, et al: Dose and dose intensity of adjuvant chemotherapy for stage II, node-positive breast carcinoma. N Engl J Med 330:1253-1259, 1994
Ambrosone CB, Sweeney C, Coles BF, et al: Polymorphisms in glutathione S-transferases (GSTM1 and GSTT1) and survival after treatment for breast cancer. Cancer Res 61:7130-7135, 2001
Nedelcheva Kristensen V, Andersen TI, Erikstein B, et al: Single tube multiplex polymerase chain reaction genotype analysis of GSTM1, GSTT1 and GSTP1: Relation of genotypes to TP53 tumor status and clinicopathological variables in breast cancer patients. Pharmacogenetics 8:441-447, 1998
Lizard-Nacol S, Coudert B, Colosetti P, et al: Glutathione S-transferase M1 null genotype: Lack of association with tumour characteristics and survival in advanced breast cancer. Breast Cancer Res 1:81-87, 1999
Chang TK, Weber GF, Crespi CL, et al: Differential activation of cyclophosphamide and ifosphamide by cytochromes P-450 2B and 3A in human liver microsomes. Cancer Res 53:5629-5637, 1993
Ekins S, Vandenbranden M, Ring BJ, et al: Further characterization of the expression in liver and catalytic activity of CYP2B6. J Pharmacol Exp Ther 286:1253-1259, 1998
Hanna IH, Dawling S, Roodi N, et al: Cytochrome P450 1B1 (CYP1B1) pharmacogenetics: Association of polymorphisms with functional differences in estrogen hydroxylation activity. Cancer Res 60:3440-3444, 2000(Angela DeMichele, Richard)