Abstract
Background
Breast cancer patients regularly undergo adjuvant chemotherapies following surgery. However, these treatments are largely associated with chemotherapeutic toxicities ranging from nausea to severe myelosuppression. In this investigation, we examined the effects of four SNPs in NR1I2, CYP3A4 and CYP3A5 genes on chemotherapy-induced severe neutropenia in 311 female Chinese breast cancer patients undergoing a standard adjuvant chemotherapy regimen.
Methods
Patients were monitored for adverse reactions throughout the treatment, then divided into “none or mild” (80 %) or “severe” (20 %) toxicity groups according to whether they suffered grade 4 neutropenia defined as having an absolute neutrophil counts (ANC) of less than 0.5 × 109/L anytime during the treatment. DNA was extracted from patients’ peripheral blood samples, then genotyped using allele-specific Tm-shift PCR and melting analysis.
Results
Logistic regression revealed that rs776746 or CYP3A5*3 strongly associated with grade 4 neutropenia (OR = 2.56, P = 0.023) after adjustment for covariates, one of which more significant factor was baseline ANC (OR = 0.68, P = 0.020). Although univariate analysis in all patients did not reveal any association at first, further analysis indicated that rs776746 is significantly associated with severe neutropenia in subgroup of breast cancer patients with normal baseline ANC (≥2.0 × 109/L). These carriers of A-allele have 3.14-fold increased risk of developing severe neutropenia (P = 0.004).
Conclusion
Our results suggested that polymorphisms in CYP3A5 might be useful pharmacogenetic markers for the prediction of severe neutropenia during chemotherapy, however, only after screening patients by their baseline ANC in the presence of gene–environmental interaction. We demonstrate an approach of pharmacogenetic analysis, in which the genetic data should be analyzed in the perspective of other clinical parameters.
Electronic supplementary material
The online version of this article (doi:10.1007/s00432-012-1345-5) contains supplementary material, which is available to authorized users.
Keywords: Pharmacogenetics, CYP3A5 gene, Adjuvant chemotherapy, Neutropenia
Introduction
Chemotherapy is one of the standard adjuvant therapies for breast cancer patients. However, such treatment is associated with a number of side effects ranging from vomiting to severe pancytopenia due to toxicity to the bone marrow. Among patients treated with similar empirical standard chemotherapy dosage, there is wide variability in the response of individuals in terms of toxicity, which is an important problem in clinical practice (Fleeman et al. 2011; Regan et al. 2012; Sohn et al. 2004). Some of this variability is due to genetic factors, and pharmacogenetic studies have been carried out to identify what these genetic polymorphisms are and how they could be used in clinic practice. There have been quite a number of prominent successes in this field in the modern era of medical genetics, such that FDA made recommendations to include genetic testing before administration of certain drugs. For example, there are pharmacogenetic tests for thiopurine methyltransferase (TPMT) in treatment with thiopurine (Sohn et al. 2004). On the other hand, in routine adjuvant chemotherapy of breast cancer, it is not possible to predict which patient will have more severe adverse reaction.
We previously reported that, at the same empirical dosage of cytotoxic drugs for breast cancer treatment, Chinese patients experienced significantly more severe myelosuppression when compared with Caucasian patients (Ling and Lee 2011; Ma et al. 2002). Our retrospective study (Ma et al. 2002) showed a cohort of 85 Chinese breast cancer patients who received standard AC adjuvant chemotherapy was found to have a significantly higher incidence of grade 3 (52 %) and grade 4 (25 %) neutropenia, as well as hepatotoxicity (9 %) when compared with a historic Western cohort. In a more recent prospective study, (Beith et al. 2002) a total of 173 patients were recruited, in an inter-ethnic comparison study including Asian patients from our center (Hong Kong, 31) and Singapore (72) as well as Caucasian patients from Sydney (70). Toxicity was principally hematological, and when analyzed according to ethnicity, Asian patients had a higher rate of grade 4 neutropenia than Caucasians (53.9 vs. 18.9 %, respectively, χ 2 = 0.015). In addition, febrile neutropenia was encountered in the first cycle in 3.8 % Caucasian and 9.3 % Asian patients. Similar findings were reported in other studies (Yau et al. 2009; Ling and Lee 2011).
Inter-ethnic difference in pharmacogenetics is a well-recognized phenomenon (Phan et al. 2011). However, there is less understanding in the area of pharmacogenetics in breast cancer. In addition, conflicting results are not uncommon in the field of pharmacogenetics due to several reasons: (1) inter-ethnic difference in prevalence of variants will make an association more readily shown in one ethnic group but not the other, and (2) false positive association in initial report due to publication bias and chance. Therefore, it is important to validate or replicate genetic association results in local population before it is applied in routine clinical setting.
Nuclear receptors have an important role in the regulation of gene expression and integrate cellular stimuli into cellular response (Tirona 2010). Many drug transporter and enzymes are regulated by nuclear receptor at transcription, thus they are likely to have a key role in the determination of inter-individual variation in drug response (Ihunnah et al. 2011). The ligand-activated nuclear receptor pregnane X receptor (PXR) encoded by the NR1I2 gene and androstane receptor (CAR) are key xenobiotic responsive receptors. They regulate expression of a group of P450 enzymes (CYP3A), including phase II enzymes and drug transporter.
CYP3A gene cluster in long arm of chromosome 7 spans a genomic distance of 145,000 bps. It includes CYP3A5-CYP3A7-CYP3A4 genes. The encoded proteins CYP3A4 and CYP3A5 are well-characterized key players in drug metabolisms. There are only 2 alleles that are prevalent in the Chinese population, CYP3A4*1G (rs2242480) and CYP3A5*3 (rs776746) (Du et al. 2006; Phan et al. 2011; Zhou et al. 2011). It is also known that CYP3A5 plays an important role in the activation of cyclophosphamide from its pro-drug to the active cytotoxic metabolite (Anglicheau et al. 2007; Bray et al. 2010). However, its association in chemotherapy-related toxicity has been controversial (Bray et al. 2010; Kim et al. 2012; Low et al. 2009; Tsai et al. 2009). While many studies showed a trend for association with increased toxicity with CYP3A5*3, most of them fell short of significance at statistical analysis.
It has been reported that a specific haplotype of NR1I2 was associated with low expression of its downstream P450 genes, particularly CYP3A4, which had a poor clearance of doxorubicin and accounted for both inter-individual and inter-ethnic variation in disposition of doxorubicin (Sandanaraj et al. 2008). Another study from the same country using similar design and sample size did not confirm the association between NR1I2 genotype and pharmacokinetics or pharmacodynamics of doxorubicin (Hor et al. 2008). However, few follow-up studies attempted to replicate this finding. Therefore, it is important to evaluate if the association between CYP3A4, CYP3A5 and NR1I2 genotypes extend to manifestation of myelotoxicity in breast cancer patients undergo chemotherapy, and in an Asian population which is most affected by the cytotoxicity.
Methods
Study populations
Case study cohort
Chinese female breast cancer patients who attended the Department of Clinical Oncology at the Prince of Wales Hospital, Hong Kong, which is a tertiary referral center for oncological treatment, were recruited for this study between January 2009 and October 2011. To be eligible for inclusion in the study, subject must be a Chinese female between the age of 18 and 75 who had completed primary treatment with surgery and had planned to receive adjuvant chemotherapy treatment with 4 cycles of standard AC regimen (doxorubicin 60 mg/m2 and cyclophosphamide 600 mg/m2 based on body surface area (BSA) on Day 1 of a 21-day cycle). In addition, all subjects must have adequate hematological, hepatic and renal functions at the start of chemotherapy treatment (WBC: 3 × 109/L; ANC: 1.5 × 109/L; platelet: 100 × 109/L; total bilirubin <1.5 UNL; ALT ≤2.5 UNL; serum creatinine ≤1.5 UNL, or creatinine clearance ≥40 ml/min). In total, 322 breast cancer patients were eligible and provided written consent to enter this study.
Prior to the start of chemotherapy, 3 ml of peripheral blood sample was taken from each patient for DNA extraction and baseline complete blood counts. Throughout the four cycles of treatment, the complete blood count of each patient was assessed prior to Day 1, Day 10 and Day 21 of each of the 21-day cycle of chemotherapy and monitored for adverse neutropenic reactions. Treatment-related toxicity was graded according to the criteria of NCI-CTCAE v3.0. Considering absolute neutrophil count (ANC) being the most important indicator and most readily quantifiable parameter of myelotoxicity, a clinical significantly severe chemotherapy-induced myelotoxicity was defined as having grade IV neutropenia (neutrophil counts of <500 per microliter) in any cycle, leading to a change in dose of chemotherapy regimen in the subsequent cycle.
Laboratory methods
Selection of tagging SNPs
PXR and CYP3A encoding genes, which play important roles in the metabolism of doxorubicin or docetaxel and cyclophosphamide (Sandanaraj et al. 2008), were studied. PXR (NR1I2 gene) among 100 Chinese had been reported by Sandanaraj et al. (2008). Only 4 major haplotypes were found, with a frequency greater than 5 % in the population. They were clustered into *1A (H1-C), *1B (H2-C) and *1C (H3-C and H4-C). Therefore, 2 tagSNPs were used to determine these 3 haplotype clusters, and they were IVS2 + 55A/G (rs1464603) and IVS6-17C/T (rs2276707). The corresponding genotypes of NR1I2*1A, *1B and *1C at these 2 positions are A–C, A–T and G–T, respectively. And the only two SNPs that are polymorphic in the Chinese population, rs2242480 and rs776746, were selected to represent the CYP3A4 and CYP3A5 gene, respectively.
Genotyping
Genomic DNA was extracted from peripheral blood samples using a commercial extraction kit according to the manufacturer’s instructions (Favorgen Biotech Corp, Taiwan). The extracted DNA samples were stored at −80 °C before genotyping.
Genotyping was performed in a single-tube PCR using melting temperature (Tm)-shift allele-specific method, as previously described (Jiang et al. 2010; Wang et al. 2005). In brief, two allele-specific primers and one common complementary primer with less than 20 bp downstream from the SNP were designed for each SNP. For genotype identification, the pair of allele-specific primers would carry different lengths of 5′GC sequence that generate a 3–4 °C difference in Tm between the allele-specific PCR products. Primers sequences for all tested SNPs are presented in Supplementary Table 1. Biomek® 2000 and NX robotic systems (Beackman Coulter, USA) were used to set up the PCR. The PCR conditions optimization procedure has been described in (Jiang et al. 2009, 2010), and genotype identification was conducted by analyzing the melting curve profiles on the LightCycler®480 Real-Time PCR system (Roche Applied Science, USA).
Statistics
For each SNP, its genotype distribution was examined for deviations from Hardy–Weinberg Equilibrium (HWE) using the chi-squared test. Descriptive analyses of breast cancer patients’ demographic, disease, treatment and toxicity characteristics were performed by chi-squared test, Fisher’s exact test and ANOVA test. Any missing values were excluded from mean or percentage calculation. Cochran–Armitage test for trend was used to evaluate the potential association between risk of breast cancer or chemotherapy-induced neutropenia and the 4 tested SNPs. Haploview 4.1 (Barrett et al. 2005) was used to infer NR1I2 and CYP3A haplotypes using an accelerated expectation maximization algorithm, and the haplotypic frequencies of subjects in different cohort or subgroups were compared. If significant association was detected, another measure of significance corrected for multiple testing bias was obtained using the implemented permutation test in Haploview. Multivariate logistic regression analyses were used to evaluate potential predictive value of NR1I2 and CYP3A genes as well as other risk factors on chemotherapy-induced neutropenia in breast cancer patients. Statistical analyses were conducted using SPSS 15 (SPSS Inc., Chicago, IL) and Haploview 4.1, and results were considered significant at two-tailed P value of <0.05.
Results
Characteristics of study cohorts
Out of the total 322 recruited breast cancer patients, 11 dropped out of the study during the treatment, resulting in a final cohort of 311 cases. As they received the standard adjuvant chemotherapy, all patients had a Karnofsky performance status of 100, and all received dexamethasone as antiemetic regimen. The patients were further divided into two study groups based on their experiences of chemotherapy-induced toxicity. Grade IV neutropenia toxicity at any one cycle throughout the entire treatment was found in 20 % of all patients (N = 63) who were categorized as “very severely” toxic group, while all other patients who experienced grade 0 to grade 3 were labeled as “none or mildly” toxic (N = 248).
The univariate comparison of anthropometric and disease characteristics between two study groups were analyzed as risk factors for adverse reaction using chi-squared test (see Table 1). The distribution of age at diagnostics, height and tumor size was similar between the two toxicity groups. The percentage of patients who were smokers, post-menopausal, having mastectomy for surgery and having positive ER status of the two case groups was also similar. However, it is apparent that body weight (and hence BMI, P = 0.001), and more impressively, baseline ANC and platelet counts (P < 0.001) were independent predictors for subsequent chemotherapy-induced neutropenia in univariate analysis. In addition, the use of growth factor support (GCSF, granulocyte colony-stimulating factor) significantly correlated with severe neutropenia toxicity (P < 0.001).
Table 1.
Demographic and disease/treatment characteristics of breast cancer patients
| Variables | None or mild (N = 248) |
Very severe (N = 63) |
P value |
|---|---|---|---|
| Age at diagnosis | 49.9 ± 8.4 | 50.8 ± 8.8 | NS |
| Height (cm) | 157.2 ± 6.1 | 157.2 ± 6.7 | NS |
| Weight (kg) | 59.3 ± 10.5 | 54.9 ± 8.4 | 0.002 |
| Body mass index (BMI) | 24.1 ± 4.4 | 22.2 ± 3.0 | 0.001 |
| Body surface area (BSA) | 1.6 ± 0.2 | 1.5 ± 0.1 | 0.066 |
| Post-menopausal (yes) | 108 (43.4 %) | 34 (54.0 %) | NS |
| Smokers (yes) | 14 (5.6 %) | 4 (6.6 %) | NS |
| 1st degree with CA breast (yes) | 25 (10.2 %) | 2 (3.2 %) | NS |
| Tumor size | 2.6 ± 1.8 | 2.3 ± 1.2 | 0.032 |
| Surgery types | |||
| Mastectomy | 184 (74.2 %) | 48 (76.2 %) | NS |
| BCS | 64 (25.8 %) | 15 (23.8 %) | |
| Positive ER status (yes) | 172 (69.1 %) | 44 (68.8 %) | NS |
| Baseline ANC count on day 0 | 3.63 ± 1.63 | 2.38 ± 1.53 | <0.001 |
| Baseline platelet count on day 0 | 251.2 ± 66.4 | 200.0 ± 73.1 | <0.001 |
| Use of GCSF during any cycle | 7 (2.8 %) | 16 (25.0 %) | <0.001 |
| Dose reduction during any cycle | 5 (2.0 %) | 4 (6.3 %) | 0.089 |
NS, not significant
Bold p-values indicate the association test was significant at p < 0.05
Univariate analysis of severe myelotoxicity and pharmacogenetic polymorphisms
A total of four SNPs from NR1I2 gene (rs1464603 and rs2276707) on chromosome 3, as well as CYP3A4 gene (rs2242480) and CYP3A5 gene (rs776746) on chromosome 7 were examined in this study. Their genotypic frequency distributions in the two groups are shown in Table 2. All four SNPs followed HWE (P > 0.05) in the mild case group.
Table 2.
Association between genotype and Grade IV neutropenia using a breast cancer case-only analysis on the genotype and haplotype frequency distribution of NR1I2 and CYP3A SNPs
| SNPs | Genotypes | All cases (N = 311) |
None or mild (N = 248) |
Very severe (N = 63) |
P value* |
|---|---|---|---|---|---|
| NR1I2 gene | |||||
| rs1464603 | CC | 40 (12.9) | 34 (13.7) | 6 (9.5) | |
| Position on chr3: 121009039 | CT | 148 (47.6) | 118 (47.6) | 30 (47.6) | 0.380 |
| Gene: NR1I2 | TT | 123 (39.5) | 96 (38.7) | 27 (42.9) | |
| Ancestral allele: T | HWE= | 0.662 | 0.813 | 0.571 | |
| rs2276707 | CC | 84 (27.5) | 66 (27.0) | 18 (29.0) | |
| Position on chr7: 121016843 | CT | 143 (46.7) | 120 (49.2) | 23 (37.1) | 0.434 |
| Gene: NR1I2 | TT | 79 (25.8) | 58 (23.8) | 21 (33.9) | |
| Ancestral allele: C | HWE= | 0.255 | 0.811 | 0.044 | |
| Haplotype analysis | T–C | 212 (34.2) | 171 (34.6) | 41 (32.5) | 0.650 |
| rs1464603–rs2276707 | T–T | 181 (29.2) | 138 (27.9) | 43 (20.3) | 0.168 |
| C–T | 124 (20.0) | 101 (20.5) | 23 (11.1) | 0.554 | |
| C–C | 103 (16.6) | 84 (5.3) | 19 (5.4) | 0.637 | |
| CYP3A cluster genes | |||||
| rs776746 | GG | 151 (49.2) | 126 (51.4) | 25 (40.3) | |
| Position on chr7: 99108475 | AG | 140 (45.6) | 105 (42.9) | 35 (56.5) | 0.306 |
| Gene: CYP3A5*3 | AA | 16 (5.2) | 14 (5.7) | 2 (3.2) | |
| Ancestral allele: A | HWE= | 0.022 | 0.191 | 0.015 | |
| rs2242480 | CC | 161 (54.0) | 128 (53.8) | 33 (55.0) | |
| Position on chr7: 99199402 | CT | 117 (39.3) | 94 (39.5) | 23 (38.3) | 0.887 |
| Gene: CYP3A4*1G | TT | 20 (6.7) | 16 (6.7) | 4 (6.7) | |
| Ancestral allele: T | HWE= | 0.839 | 0.822 | 0.998 | |
| Haplotype analysis | G–C | 413 (66.8) | 335 (67.7) | 78 (63.1) | 0.330 |
| rs776746–rs2242480 | A–T | 128 (20.8) | 103 (20.9) | 25 (20.3) | 0.889 |
| A–C | 44 (7.1) | 30 (6.1) | 14 (11.1) | 0.052 | |
| G–T | 33 (5.3) | 26 (5.3) | 7 (5.4) | 0.942 | |
Values presented are frequencies (percentages). NS, not significant. Genotype frequencies may not add up to 249 or 64 due to failed genotyping or missing samples
* P value represents comparison between non-toxic and toxic groups
Haplotype analysis was also performed for both NR1I2 and CYP3A cluster genes (using the CYP3A4*1G SNP rs2242480 and the CYP3A5*3 SNP rs776746, see Table 2). The two SNPs in each gene composed four different haplotypes in each gene: with T–C being the most common in NR1I2 accounting for more than one-third of all haplotypes in both case and control cohorts; while C–C being the most common haplotype for CYP3A4/5 cluster, accounting for about two-thirds of all haplotypes in both samples sets. As in genotype analysis, no association was detected between the haplotypes and chemotherapy-induced myelotoxicity in cases either.
Multivariate analysis of severe myelotoxicity and pharmacogenetic polymorphisms
As several other risk predictors were associated with severe chemotherapy toxicity (Table 1), it would be more informative to carry out a multivariate analysis in order to better understand the underlying primary predictors. Logistic regression analysis revealed four independent risk factors significantly associated with developing grade 4 neutropenia in breast cancer patients (see Table 3). One of which is the anthropometric measure of BMI (P = 0.014), for which each increase of 1 kg/m2 unit reduced the odds of developing grade 4 neutropenia by 0.86-fold (95 % CI = 0.77–0.97). The baseline ANC of the patients also shows an association (P = 0.020), where in each increase of 1 × 109/L baseline ANC, there was a decreased risk by 0.68-fold (95 % CI = 0.49–0.94). In addition, the use of supportive growth factor therapy appeared to be the strongest associating factor with severe neutrophil toxicity (P < 0.001). Patients who received GCSF therapy were at 9.15-fold (95 % CI = 2.98–28.15) more risk of developing grade 4 neutropenia reactions during chemotherapy. Finally, after controlling for the covariates, the CYP3A5*3 SNP rs776746 was found to be a strong predictor of chemotherapy-induced neutropenia (P = 0.023), as each additional T allele increased the risk of developing severe myelotoxicity 2.56-fold (95 % CI = 1.14–5.75).
Table 3.
Logistic regression of CYP3A and NR1I2 SNPs with risk factors
| Variables | β (±SE) | Odds ratio (95 % CI) | P value |
|---|---|---|---|
| Age | 0.038 (±0.035) | 1.039 (0.969–1.114) | NS |
| BMI | −0.148 (±0.060) | 0.862 (0.767–0.970) | 0.014 |
| Menopausal status | 0.377 (±0.551) | 1.458 (0.495–4.295) | NS |
| Smoker | 0.007 (±0.918) | 1.007 (0.167–6.094) | NS |
| 1st degree with CA breast? | −1.614 (±0.930) | 0.199 (0.032–1.231) | NS |
| BCS as primary surgery | −0.059 (±0.469) | 0.943 (0.376–2.363) | NS |
| Tumor size | −0.095 (±0.163) | 0.910 (0.661–1.251) | 0.067 |
| Negative ER status | 0.257 (±0.405) | 1.292 (0.584–2.859) | NS |
| Baseline ANC | −0.388 (±0.167) | 0.678 (0.489–0.941) | 0.020 |
| Baseline platelet count | −0.005 (±0.004) | 0.995 (0.987–1.002) | NS |
| Dose reduction | 1.438 (±0.914) | 4.211 (0.702–25.265) | NS |
| Use of GCSF | 2.214 (±0.573) | 9.151 (2.978–28.151) | <0.001 |
| rs1464603 (NR1I2) | 0.437 (±0.280) | 1.548 (0.893–2.682) | NS |
| rs2276707 (NR1I2) | 0.354 (±0.258) | 1.425 (0.859–2.364) | NS |
| rs776746 (CYP3A5) | 0.941 (±0.413) | 2.563 (1.142–5.754) | 0.023 |
| rs2242480 (CYP3A4) | −0.530 (±0.390) | 0.589 (0.274–1.265) | NS |
NS, not significant
Bold p-values indicate the association test was significant at p < 0.05
To elucidate the reason why CYP3A5*3 SNP showed a significant association with severe neutropenia in logistic regression model but not in the univariate analyses, breast cancer patients were further sub-grouped into risk groups based on their baseline ANC. According to criteria of NCI-CTCAE v3.0, the lower limit of clinically normal ANC range is 2.0 × 109/L, a lower value would signify myelotoxicity. Therefore, we divided patients into two subgroups: patients with abnormal baseline ANC <2.0 × 109/L and patients with normal baseline ANC ≥ 2.0 × 109/L, and re-performed the univariate analysis.
Re-analysis of the genotype frequencies of CYP3A genes between the two toxicity groups in patients with normal baseline ANC is shown in Table 4. It is revealed that rs776746 had a significant association with myelotoxicity in the subgroup of patient with normal baseline ANC (P = 0.016). In addition, the dominant mode analysis found a significant 3.14-fold increase (95 % CI = 1.41–6.98, P = 0.004) in the risk of developing grade 4 neutropenia if such patient possess any A-allele at rs776746, which is consistent with the results from logistic regression. Although rs2242480 gene (CYP3A4*1G) did not correlate with severe myelotoxicity in itself, the A–C haplotype forms with rs776746 did show a strong association with the risk of developing severe neutropenia (P = 0.003) in normal baseline ANC subgroup and remained significant even after 10,000 permutation tests that corrected for multiple testing bias (P = 0.023). The haplotype analysis confirmed a role of the CYP cluster locus in predisposition to developing grade 4 neutropenia in this subgroup of patients.
Table 4.
Genotype frequency of CYP3A SNPs of low-risk group patients with clinically normal baseline ANC (>2.00 × 109/L)
| SNPs | Genotypes | None or mild (N = 248) |
Very severe (N = 63) |
Odds ration P value |
P value after permutation |
|---|---|---|---|---|---|
| rs776746 | GG | 104 | 9 | 0.016 a | – |
| (CYP3A5*3) | AG | 91 | 26 | – | |
| AA | 12 | 2 | |||
| Dominant mode analysis | |||||
| GG | 104 | 9 | 0.004 b | – | |
| AG + AA | 103 | 28 | – | ||
| Odds ratio | 3.141 (1.413–6.984) | ||||
| rs2242480 | CC | 108 | 20 | 0.968a | – |
| (CYP3A4*1G) | CT | 79 | 13 | – | |
| TT | 14 | 3 | – | ||
| Dominant mode analysis | |||||
| CC | 108 | 20 | 0.840b | – | |
| CT + TT | 93 | 16 | – | ||
| Haplotype analysis | G–C | 286 | 45 | 0.078 | 0.437 |
| rs776746–rs2242480 | A–T | 89 | 18 | 0.630 | 0.999 |
| A–C | 27 | 13 | 0.003 | 0.023 | |
| G–T | 22 | 2 | 0.454 | 0.988 | |
P value obtained from aCochran–Armitage test for trend; bPearson’s chi-squared test
Bold p -values indicate the association test was significant at p < 0.05
Discussion
The ethnic difference in drug metabolism and toxicity from chemotherapy is an important area of research (Grann et al. 2008). Due to the great degree of ethnic difference, some underlying genetic diversity related to pharmacogenetics between ethnic groups is a likely contributing factor. Such inter-ethnic difference is not confined to the treatment of breast cancer, but also represents a general phenomenon in other cancer treatments (Law et al. 2007; Phan et al. 2009). In the case of chemotherapy for breast cancer, there was a high incidence of myelotoxicity among Chinese, or generally Asian women (Han et al. 2011; Ling and Lee 2011; Ma et al. 2002, 2010; Yau et al. 2009).
Myelotoxicity is an important complication of chemotherapy. It increases risk of infection, septicemia and even results in mortality. In the past, only clinical data or cancer parameters were used as predictor of adverse events after chemotherapy. Many previous studies investigated patients’ samples on different adjuvant regimen. This approach effectively reduce the sample size, and thus in logistic regression, it is virtually equivalent to the power of the sample size of each individual treatment regimen (Chew et al. 2009; Low et al. 2009; Yao et al. 2010a, b). On the other hand, small sample size (around or less than 100 subjects) are common in studies, which confined to a single regimen (Hor et al. 2008; Nakajima et al. 2007; Rizzo et al. 2010; Tsai et al. 2009; Wong et al. 2011). Here, we advance by a step further to predict which patient is going to experience myelotoxicity by combining 2 sets of predictors, namely clinical and pharmacogenetics. We studied one of the largest sample sizes, so far, of 311 patients on the same regimen based on AC. We only studied the most potentially important pharmacogenetic genes, NR1I2 and CYP3A cluster, both of which play important role in metabolism of doxorubicin, docetaxel and cyclophosphamide (Sandanaraj et al. 2008). In fact, NR1I2 also acts as an inducer (transcription factor) for upregulation of expression of CYP3A4 and thus forming a gene interaction network. These information forms the basis of the analysis carried out in this study, single gene analysis and gene–gene interaction analysis.
Although no associations were detected in univariate analysis of single maker (SNP genotype), logistic regression and subsequent subgroup examination, after further dividing the patients by their baseline ANC, revealed that CYP3A genetic polymorphisms can in fact predict severe neutropenia due to chemotherapy. The most significant association with severe neutropenia was CYP3A5*3, also known as rs776746. It is a common A to G polymorphism at position 6,986 in intron 3, which leads to a splicing variant resulting in protein truncation (King et al. 2003; Kuehl et al. 2001; Shih and Huang 2002). This defective allele was prevalent among Chinese, and the allelic frequency of G variant was as high as 0.7 in Chinese (Hu et al. 2005). Yet, it seems having a GG genotype protected patients from developing severe neutropenia, possibly due to its faulty enzymatic activity and cannot transform the chemotherapy pro-drug into the active cytotoxic metabolites. While carriers of A-allele as this locus, with a normal functional CYP3A5 enzyme, have a 3.14-fold increased risk of experiencing grade 4 neutropenia due to chemotherapy toxicity, even if they originally had a normal ANC level at the start of the treatment. These results showed a scenario of gene–environmental interaction in chemotherapy-induced toxicities.
The use of growth factor support is an interesting independent risk factor, as it strongly associated the development severe neutropenia in our univariate analysis and logistic regression. However, all of the total 23 patients (from both toxicity group) who ended up receiving GCSF therapy only started using growth factor support after the cycle with their worst ANC. Therefore, it is most likely a consequential factor rather than a predictive factor for the development of severe myelotoxicity.
The results have major implication in clinical practice and future studies. In the clinical situation, it would be not informative to perform pharmacogenetic test in all (unselective) patients receiving chemotherapy, as it will predict adverse events in the whole group. For example, in this study patients with abnormally low baseline ANC were very likely to experience severe neutropenia irrespective of the genotypes. In another word, the non-genetic risk factor (low baseline ANC) is sufficient to lead to severe neutropenia by itself, and there is little information gained by pharmacogenetics. However, our study showed that, while in a selected subgroup of patients with clinically normal baseline ANC, it will be very informative to perform pharmacogenetic test to predict subsequent adverse event. This is also likely the reason underlying the conflicting results obtained in previous pharmacogenetic studies, some of which supported an association while other did not.
Conclusion
Examining the genetic polymorphisms in these genes or other drug-metabolizing enzyme genes may contribute significantly to the development of personalized medicine, by which the effects of chemotherapeutic toxicity may be minimized for cancer patients. In our study, a large cohort of 311 patients receiving a single adjuvant chemotherapy regimen, independent risk factors for grade IV neutropenia included weight in kg, BMI, baseline ANC, baseline platelet count and the use of growth factor support (univariate analysis). In multivariate analysis, BMI, baseline ANC, the use of growth factor support and rs776746 of the CYP3A5 gene were independent risk factors. In addition, rs776746 associated with severe neutropenia in patient with normal baseline ANC (ANC ≥2.0 × 109/L). Our results showed that there was no gene–gene interaction between NR1I2 and CYP3A4/5. However, there is significant interaction between gene and non-genetic risk factors in predisposition to severe neutropenia. Therefore, future studies should be more concentrated on this interaction, and such interaction may account for conflicting results in previous reports.
Electronic supplementary material
Below is the link to the electronic supplementary material.
Acknowledgments
This investigation was supported by GlaxoSmithKline Oncology International Ethnic Research Initiative (GSK-ERI) funding (Co-PI: Nelson LS Tang and Winnie Yeo).
Ethical Standards
The study was approved by the Clinical Research Ethics Committee of the Chinese University of Hong Kong.
Conflict of interest
None.
References
- Anglicheau D, Legendre C, Beaune P, Thervet E (2007) Cytochrome P450 3A polymorphisms and immunosuppressive drugs: an update. Pharmacogenomics 8(7):835–849 [DOI] [PubMed] [Google Scholar]
- Barrett JC, Fry B, Maller J, Daly MJ (2005) Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 21(2):263–265. doi:10.1093/bioinformatics/bth457 [DOI] [PubMed] [Google Scholar]
- Beith J, Goh B, Yeo W, Sullivan A, Lim S, Zhong S, Rivory L (2002) Interethnic differences in the myelotoxicity of adriamycin/cyclophosphamide (AC) for adjuvant breast cancer. In: Proceedings of American Society of Clinical Oncology Annual Meeting: J Clin Oncol 21:64a, #252
- Bray J, Sludden J, Griffin MJ, Cole M, Verrill M, Jamieson D, Boddy AV (2010) Influence of pharmacogenetics on response and toxicity in breast cancer patients treated with doxorubicin and cyclophosphamide. Br J Cancer 102(6):1003–1009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Chew HK, Doroshow JH, Frankel P et al (2009) Phase II studies of gemcitabine and cisplatin in heavily and minimally pretreated metastatic breast cancer. J Clin Oncol 27(13):2163–2169 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Du J, Xing Q, Xu L et al (2006) Systematic screening for polymorphisms in the CYP3A4 gene in the Chinese population. Pharmacogenomics 7(6):831–841 [DOI] [PubMed] [Google Scholar]
- Fleeman N, Martin Saborido C, Payne K et al (2011) The clinical effectiveness and cost-effectiveness of genotyping for CYP2D6 for the management of women with breast cancer treated with tamoxifen: a systematic review. Health Technol Assess 15(33):1–102 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Grann VR, Bowman N, Joseph C et al (2008) Neutropenia in 6 ethnic groups from the Caribbean and the U.S. Cancer 113(4):854–860 [DOI] [PubMed] [Google Scholar]
- Han HS, Reis IM, Zhao W et al (2011) Racial differences in acute toxicities of neoadjuvant or adjuvant chemotherapy in patients with early-stage breast cancer. Eur J Cancer 47(17):2537–2545 [DOI] [PubMed] [Google Scholar]
- Hor SY, Lee SC, Wong CI et al (2008) PXR, CAR and HNF4alpha genotypes and their association with pharmacokinetics and pharmacodynamics of docetaxel and doxorubicin in Asian patients. Pharmacogenomics J 8(2):139–146 [DOI] [PubMed] [Google Scholar]
- Hu YF, He J, Chen GL et al (2005) CYP3A5*3 and CYP3A4*18 single nucleotide polymorphisms in a Chinese population. Clin Chim Acta 353(1–2):187–192 [DOI] [PubMed] [Google Scholar]
- Ihunnah CA, Jiang M, Xie W (2011) Nuclear receptor PXR, transcriptional circuits and metabolic relevance. Biochim Biophys Acta 1812(8):956–963 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Jiang J, Tang NL, Ohlsson C et al (2009) Association of genetic variations in aromatase gene with serum estrogen and estrogen/testosterone ratio in Chinese elderly men. Clin Chim Acta 411(1–2):53–58 [DOI] [PubMed] [Google Scholar]
- Jiang J, Tang NL, Ohlsson C et al (2010) Association of SRD5A2 variants and serum androstane-3alpha,17beta-diol glucuronide concentration in Chinese elderly men. Clin Chem 56(11):1742–1749 [DOI] [PubMed] [Google Scholar]
- Kim K-P, Ahn J-H, Kim S-B, Jung KH, Yoon DH, Lee JS, Ahn S-H (2012) Prospective evaluation of the drug-metabolizing enzyme polymorphisms and toxicity profile of docetaxel in Korean patients with operable lymph node-positive breast cancer receiving adjuvant chemotherapy. Cancer Chemother Pharmacol 69(5):1221–1227. doi:10.1007/s00280-011-1816-4 [DOI] [PubMed] [Google Scholar]
- King BP, Leathart JB, Mutch E, Williams FM, Daly AK (2003) CYP3A5 phenotype-genotype correlations in a British population. Br J Clin Pharmacol 55(6):625–629 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kuehl P, Zhang J, Lin Y et al (2001) Sequence diversity in CYP3A promoters and characterization of the genetic basis of polymorphic CYP3A5 expression. Nat Genet 27(4):383–391 [DOI] [PubMed] [Google Scholar]
- Law C-C, Fu Y-T, Chau K-K, Choy T-S, So P-F, Wong K-H (2007) Toxicity profile and efficacy of oral capecitabine as adjuvant chemotherapy for Chinese patients with Stage III colon cancer. Dis Colon Rectum 50(12):2180–2187 [DOI] [PubMed] [Google Scholar]
- Ling WHY, Lee SC (2011) Inter-ethnic differences—how important is it in cancer treatment? Ann Acad Med Singap 40(8):356–361 [PubMed] [Google Scholar]
- Low SK, Kiyotani K, Mushiroda T, Daigo Y, Nakamura Y, Zembutsu H (2009) Association study of genetic polymorphism in ABCC4 with cyclophosphamide-induced adverse drug reactions in breast cancer patients. J Hum Genet 54(10):564–571 [DOI] [PubMed] [Google Scholar]
- Ma B, Yeo W, Hui P, Ho WM, Johnson PJ (2002) Acute toxicity of adjuvant doxorubicin and cyclophosphamide for early breast cancer—a retrospective review of Chinese patients and comparison with an historic Western series. Radiother Oncol 62(2):185–189 [DOI] [PubMed] [Google Scholar]
- Ma BBY, Hui EP, Mok TSK (2010) Population-based differences in treatment outcome following anticancer drug therapies. Lancet Oncol 11(1):75–84 [DOI] [PubMed] [Google Scholar]
- Nakajima M, Komagata S, Fujiki Y et al (2007) Genetic polymorphisms of CYP2B6 affect the pharmacokinetics/pharmacodynamics of cyclophosphamide in Japanese cancer patients. Pharmacogenet Genomics 17(6):431–445 [DOI] [PubMed] [Google Scholar]
- Phan VH, Moore MM, McLachlan AJ, Piquette-Miller M, Xu H, Clarke SJ (2009) Ethnic differences in drug metabolism and toxicity from chemotherapy. Expert Opin Drug Metab Toxicol 5(3):243–257 [DOI] [PubMed] [Google Scholar]
- Phan VH, Tan C, Rittau A, Xu H, McLachlan AJ, Clarke SJ (2011) An update on ethnic differences in drug metabolism and toxicity from anti-cancer drugs. Expert Opin Drug Metab Toxicol 7(11):1395–1410 [DOI] [PubMed] [Google Scholar]
- Regan MM, Leyland-Jones B, Bouzyk M et al (2012) CYP2D6 genotype and tamoxifen response in postmenopausal women with endocrine-responsive breast cancer: the breast international group 1–98 trial. J Natl Cancer Inst 104(6):441–451 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Rizzo R, Spaggiari F, Indelli M, Lelli G, Baricordi OR, Rimessi P, Ferlini A (2010) Association of CYP1B1 with hypersensitivity induced by taxane therapy in breast cancer patients. Breast Cancer Res Treat 124(2):593–598 [DOI] [PubMed] [Google Scholar]
- Sandanaraj E, Lal S, Selvarajan V et al (2008) PXR pharmacogenetics: association of haplotypes with hepatic CYP3A4 and ABCB1 messenger RNA expression and doxorubicin clearance in Asian breast cancer patients. Clin Cancer Res 14(21):7116–7126 [DOI] [PubMed] [Google Scholar]
- Shih PS, Huang JD (2002) Pharmacokinetics of midazolam and 1′-hydroxymidazolam in Chinese with different CYP3A5 genotypes. Drug Metab Dispos 30(12):1491–1496 [DOI] [PubMed] [Google Scholar]
- Sohn KJ, Croxford R, Yates Z, Lucock M, Kim YI (2004) Effect of the methylenetetrahydrofolate reductase C677T polymorphism on chemosensitivity of colon and breast cancer cells to 5-fluorouracil and methotrexate. J Natl Cancer Inst 96(2):134–144 [DOI] [PubMed] [Google Scholar]
- Tirona RG (2010) Impact of nuclear receptors CAR, PXR, FXR, and VDR, and their ligands on enzymes and transporters. In: Pang KS, Rodrigues AD, Peter RM (eds) Enzyme- and transporter-based drug–drug interactions, progress and future challenges, 1st edn. Springer, New York, pp 75–105. doi:10.1007/978-1-4419-0840-7_4
- Tsai SM, Lin CY, Wu SH, Hou LA, Ma H, Tsai LY, Hou MF (2009) Side effects after docetaxel treatment in Taiwanese breast cancer patients with CYP3A4, CYP3A5, and ABCB1 gene polymorphisms. Clin Chim Acta 404(2):160–165 [DOI] [PubMed] [Google Scholar]
- Wang J, Chuang K, Ahluwalia M et al (2005) High-throughput SNP genotyping by single-tube PCR with Tm-shift primers. Biotechniques 39(6):885–893 [DOI] [PubMed] [Google Scholar]
- Wong ALA, Yap HL, Yeo WL et al (2011) Gemcitabine and platinum pathway pharmacogenetics in Asian breast cancer patients. Cancer Genomics Proteomics 8(5):255–259 [PubMed] [Google Scholar]
- Yao S, Barlow WE, Albain KS et al (2010a) Gene polymorphisms in cyclophosphamide metabolism pathway, treatment-related toxicity, and disease-free survival in SWOG 8897 clinical trial for breast cancer. Clin Cancer Res 16(24):6169–6176 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yao S, Barlow WE, Albain KS et al (2010b) Manganese superoxide dismutase polymorphism, treatment-related toxicity and disease-free survival in SWOG 8897 clinical trial for breast cancer. Breast Cancer Res Treat 124(2):433–439 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yau T, Ng T, Soong I et al (2009) Toxicity of docetaxel plus cyclophosphamide as adjuvant therapy for breast cancer in Chinese patients—the Hong Kong experience. Asia Pac J Clin Oncol 5(2):123–128 [Google Scholar]
- Zhou Q, Yu X, Shu C et al (2011) Analysis of CYP3A4 genetic polymorphisms in Han Chinese. J Hum Genet 56(6):415–422 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.