Docetaxel-induced neuropathy: a pharmacogenetic case-control study of 150 women with early-stage breast cancer.

Abstract Background. Docetaxel is a highly effective treatment of a wide range of malignancies but is often associated with peripheral neuropathy. The genetic variability of genes involved in the transportation or metabolism of docetaxel may be responsible for the variation in docetaxel-induced peripheral neuropathy (DIPN). The main purpose of this study was to investigate the impact of genetic variants in GSTP1 and ABCB1 on DIPN. Material and methods. DNA was extracted from whole blood from 150 patients with early-stage breast cancer who had received adjuvant docetaxel from February 2011 to May 2012. Two polymorphisms in GSTP1 and three in ABCB1 were selected for the primary analysis, and a host of other candidate genes was explored and compared between 75 patients with clinician-reported DIPN grade ≥ 2 and 75 patients without DIPN. Results. Patients with the genetic variants GSTP1 rs1138272 C/T or T/T (114Ala/114Val or 114Val/114Val) genotype had an adjusted odds ratio of 3.82; 95% confidence interval 1.34–11.09 of developing DIPN. This result was confirmed in both analysis of cumulated docetaxel dose and haplotype analysis. None of the explorative genes investigated were significantly correlated with DIPN. Patients with a BMI ≥ 30 were five-fold more likely to have DIPN than patients with BMI < 25. Conclusion. We found that GSTP1 Ala114Val polymorphism is associated with occurrence of DIPN. This supports the theory that oxidative stress is involved in DIPN pathophysiology. If confirmed, this may be helpful in the risk assessment of DIPN and perhaps help to achieve better management of neurotoxicity.

Docetaxel is a highly effective treatment of a wide range of malignancies. In early breast cancer, it signifi cantly improves recurrence-free survival and overall survival [1]. However, the use of docetaxel, as well as other taxanes, is often associated with debilitating sensory peripheral neuropathy. The frequency of docetaxel-induced peripheral neuropathy (DIPN), grades 3 -4, lies between 0% and 17% [2]. We recently showed that 35% of patients with early breast cancer reported grades 2 -4 (grades 3 -4, 11%) DIPN during treatment, which led to dose reduction in up to 25% of treatment cycles [3]. The necessity of dose reduction has been confi rmed by others [4].
The variability of DIPN has been attributed to differences in treatment schedules and cumulated dose, concurrent administration of other neurotoxic chemotherapy, preexisting neuropathy [2,3], large observed inter-individual difference in pharmacokinetics of docetaxel [5], and genetic variability [6 -8]. However, no clear pattern has emerged, and currently there is no method for predicting which patients are at high risk of DIPN before treatment.
The genetic variability of genes involved in the transportation or metabolism of docetaxel could explain some of the variation in the exposure of the nerve fi bers to docetaxel. Therefore, there is a need for identifi cation and validation of single nucleotide polymorphisms (SNPs) that are strongly associated with the risk of developing DIPN, allowing oncologists to predict the toxicity before starting chemotherapy. Two pharmacogenetic studies suggest that SNPs in GSTP1 and ABCB1 are associated with DIPN [6,7], but these studies are like other SNP studies of DIPN characterized by a small number of subjects, heterogeneous range of cancers, pretreated patients, or patients receiving concomitant neurotoxic chemotherapy [6 -12].
The primary aim of this study was to test whether known alleles of GSTP1 and ABCB1 are associated with DIPN in patients with early-stage breast cancer who were not pretreated or receiving concomitant neurotoxic chemotherapy. The secondary aim was to explore associations of a host of variants in several other candidate genes and DIPN.

Patients
In this case-control study of the infl uence of genetic factors on DIPN, we compared genetic variation among patients who developed DIPN grade Ն 2 with patients who received the same treatment but did not develop DIPN. We recruited patients who received chemotherapy with docetaxel as adjuvant treatment for early-stage breast cancer in the Department of Oncology, Odense University Hospital, Odense, Denmark, consecutively from February 2011 until May 2012. Eligibility criteria included Western European origin, chemotherapy na ï ve at breast cancer diagnosis, at least one cycle of docetaxel received, information available on presence or absence of DIPN, and no major neurological disease or symptoms prior to start of docetaxel. The patients received either three cycles of epirubicin (90 mg/m 2 ) and cyclophosphamide (600 mg/m 2 ) followed by three cycles of docetaxel (100 mg/m 2 ) every third week, or six cycles of cyclophosphamide (600 mg/m 2 ) and docetaxel (75 mg/m 2 ) every third week. Information on diabetes (yes/no) and alcohol consumption (0, 1 -7, or Ն 8 units per week) was recorded for each patient at time of inclusion. All patients provided written informed consent; the trial was performed in accordance with the Helsinki II Declaration and was approved by the Regional Scientifi c Ethical Committee for Southern Denmark (S-20100131) in Denmark.

Neuropathy
Sensory DIPN using National Cancer Institute, Common Toxicity Criteria (NCI CTCAE) version 2.0 is routinely recorded for all patients by the oncologists at the department. In cases with missing data, the investigator extracted the information from patient medical records. All DIPN assessments were blind to patient genotype. The primary endpoint of the study was grade Ն 2 DIPN.

DNA extraction and genotyping
DNA was extracted from an aliquot of venous blood using the Maxwell 16 Blood DNA Purifi cation Kit (Promega Corporation, Madison, WI, USA). SNPs in GSTP1 , ABCB1 , NAT2 , ERCC1 , ATP7A , CYP3A5, SLCO1B3 , SLC10A2, and CHST3 were genotyped using predesigned TaqMan SNP genotyping assays on a StepOne Plus real-time instrument (Applied Biosystems, Foster City, CA, USA) in accordance with the manufacturer ' s protocol. TUBB2A (rs909964, rs909965, rs9501929, rs3734492, and rs13219681) and CYP3A4 (rs2740574) SNPs were genotyped by Sanger Sequencing. For sequencing, TUBB2A and CYP3A4 promoter regions were amplifi ed by PCR using specifi c primers (Supplementary Table I, to be found online at http://informahealth care.com/doi/abs/ 10.3109/0284186X.2014.969846). PCR products were sequenced in both directions on an ABI 3730xl DNA Analyzer (Applied Biosystems) using the BigDye Terminator v 3.1 Cycle sequencing kit (Applied Biosystems). Nested PCR of the CYP3A4 promoter region is as described in Lepper et al. [13] except that Taq DNA polymerase (Sigma-Aldrich, St. Louis, MO, USA) was used and the fi nal volume of the second PCR was 10 μ l. The Verbatim High Fidelity PCR Kit (Thermo Fisher Scientifi c, Waltham, MA, USA) was used for nested PCR of the TUBB2A promoter region in accordance with the manufacturer ' s protocol. Assay numbers and sequences of primers used for genotyping are listed in Supplementary Table I to be found online at http://informahealthcare.com/doi/abs/ 10.3109/0284186X.2014.96 9846. Genotyping was performed by laboratory personnel blinded for the study endpoint.

Selection of candidate genes and SNPs
Selection of candidate genes was based on a review of the literature. Genes and SNPs with a previous association with DIPN were given priority. Nineteen SNPs in 10 genes were picked for analysis. To meet the issue of multiple testing, the candidate genes were split, a priori, into two groups: one for primary analysis and one for explorative analysis. The fi ve SNPs in GSTP1 (rs1695 and rs1138272) and ABCB1 (rs2032582, rs1128503, and rs1045642) selected for the primary analysis have previously been association with DIPN [6,7] and are all non-synonymous (i.e. amino-acid changing) with demonstrated effect on function, except rs1128503 [14]. SNPs for the explorative analysis were selected based on previously reported association with either peripheral neuropathy after paclitaxel or with other neurologically symp-toms (e.g. dizziness, syncope, or hallucinations) after docetaxel [15]. Two additional SNPs in two docetaxel metabolizing genes were included [5,12].

Statistics
Fisher ' s exact test was used to evaluate the association between genotypes and DIPN and p Ͻ 0.05 was used to indicate statistical signifi cance for the primary analysis. The p-values calculated in the explorative analyses were calculated without correction for multiple testing and should be interpreted accordingly. Differences in patient characteristics for the group with DIPN versus no DIPN were analyzed using Wilcoxon rank-sum test and χ 2 -test for continuous and categorical variables, respectively. The covariates for the multivariate logistic regression were determined on the basis of prior studies focusing on the risk factors associated with DIPN [3,16] and other suspected variables and were as follows: age ( Ͻ 55 vs. Ն 55), BMI ( Ͻ 25, 25 -29, Ն 30), type of breast surgery (mastectomy vs. breast conserving surgery), tumor size ( Ͻ 2 cm vs. Ն 2 cm), number of positive lymph nodes (0 vs. Ն 1), alcohol consumption (0 vs. Ն 1), diabetes mellitus (yes vs. no), regimen (D100 vs. D75), and cumulated dose of docetaxel ( Ͻ 300 vs. Ն 300). If the p-value of the unadjusted odds ratio (OR) was below 0.05, the variable was included in the multivariate logistic regression analyses. For completeness, time-to-event analysis was also employed, in which time was defi ned as cumulated dose of docetaxel and an event was defi ned as the fi rst incidence of DIPN Ն 2. For patients not experiencing any event, the total docetaxel drug exposure was used. Since approximately two thirds of the patients in the trial were included after treatment, these data were extracted from the charts. Time-to-neuropathy data was unavailable for three patients. Cox regression was used to test the association between each SNP and DIPN. BMI, lymph node status and regimen were included as covariates. GSTP1 (2 SNPs) and ABCB1 (3 SNPs) haplotypes were analyzed in the same way and haplotypes were estimated separately for GSTP1 and ABCB1 using PHASE software program, version 2.1 [17]. The program was run 10 times with default settings; all calls for both genes were consistent. Stata version 11.2 (StataCorp, College Station, TX, USA) was used to perform statistics calculation.

Patients
One hundred and sixty-fi ve patients were invited to participate. Forty-six patients were included during treatment and 104 patients were included after completion of treatment, three of whom declined. Twelve patients were excluded before genotyping because they did not fulfi ll the inclusion criteria (nine patients because of prior chemotherapy, one received paclitaxel, one with symptoms mimicking neuropathy, and one patient of Asian origin). We included 75 patients with sensory DIPN and 75 patients without sensory DIPN. Time since last treatment with docetaxel until inclusion in this study was median 7.5 months (range 0 -56.5). The clinical characteristics of patients with and without DIPN were similar except for more patients in the DIPN group having a BMI Ն 30 than in the group without DIPN (25% vs. 8%, p ϭ 0.008), and more patients in the DIPN group being lymph node positive (75% vs. 51%, p ϭ 0.002) ( Table I). In all, 75 (50%) patients had grade Ն 2 DIPN, including 46 patients with grade 2, 19 patients with grade 3, and 10 patients with grade 4.

Genotyping
All SNPs were in Hardy-Weinberg equilibrium. Table II gives an overview of the frequency of genotypes in the primary analyses, and Table III gives an overview of the frequency of the genotypes in the exploratory analysis. SNP genotyping of TUBB2A was unavailable in three patients (see Table III).

Haplotypes
Analyses of estimated GSTP1 and ABCB1 haplotypes and association with DIPN were carried out as a supplement. Generally the results confi rmed the associations already described. Table V shows the frequency of haplotypes depending on DIPN status. Data was analyzed with time-to-neuropathy. More patients had haplotype GSTP1 * A /GSTP1 * C in the DIPN group compared to the no DIPN group, HR ϭ 2.10 (95% CI 1.08 -4.07). None of the  * Wilcoxon rank sum test; † χ 2 -test; ‡ Breast conserving surgery, * * Only ductal and lobular carcinomas are graded -the rest were recorded as unknown, † † D100 ϭ 3 cycles of epirubicin (90 mg/m 2 ) and cyclophosphamide (600 mg/m 2 ) followed by 3 cycles of docetaxel (100 mg/m 2 ), D75 ϭ six cycles of cyclophosphamide (600 mg/m 2 ) and docetaxel (75 mg/m 2 ). haplotypes of ABCB1 were associated with DIPN in this study (Table V).

Exploratory analysis
No associations were found for the SNPs in the explorative analysis and DIPN (Table III).

Discussion
This study evaluated the infl uence of polymorphisms in ABCB1 and GSTP1 on DIPN in a homogenous population of chemotherapy na ï ve, Western European females treated for early breast cancer. We found that patients with DIPN more often carried GSTP1 rs1138272 C/T or T/T (114Ala/114Val or 114Val/114Val). In contrast none of the SNPs tested in ABCB1, TUBB2A, NAT2, ERCC1, ATP7A, CYP3A5, CYP3A4, CHST3, SLCO1B3, or SLC10A2 were found to be linked to DIPN. The role of polymorphisms in GSTP1 for DIPN has been studied previously. Mir et al. found that patients with rs1695 (105Ile/105Ile) had a higher risk of DIPN [7]. In the very comprehensive retrospective SCOTROC1 trial with 539 patients treated with docetaxel, Marsh et al. found the same association with rs1695 (105Ile/105Ile) with a p-value of 0.018; however, after correcting for multiple testing, the association was not signifi cant [8]. No association was found for the 454 patients treated with paclitaxel. Others have reported an association between rs1695 (105Ile/105Ile) and oxaliplatininduced neuropathy, but a recent meta-analysis could not confi rm this association [18].
In contrast to our results Marsh et al. and Mir et al. did not fi nd an association of GSTP1 rs1138272/ Ala114Val with DIPN [7,8]. This discrepancy may be due to differences in trial setups. Hence, Mir et al. included 58 patients (including 10 patients with DIPN) in their study, and therefore their study may not have had suffi cient statistical power. Marsh et al. included 539 patients (including 57 patients with DIPN), but these patients were treated concomitantly with carboplatin, which could account for some of the difference between our results and theirs because carboplatin is also known to be neurotoxic. Furthermore, a debate has been raised concerning the power and interpretation of p-values in the SCOTROC study [19].
Exactly how GSTP1 is linked to DIPN is unknown. However, the fact that GSTP1 encodes an enzyme that plays an important role in the inactivation of various toxic compounds, e.g. metabolites after oxidative stress, offers a plausible link [20]. Polymorphisms in the gene encoding GSTP1 have been associated with alterations in enzyme activity [14], and studies have shown that that expression and catalytic activity of GSTP1 were related to resistance to docetaxel treatment in vitro and in vivo [21]. A recent study of paclitaxel-induced neuropathy in rats showed that paclitaxel was associated with an accumulation of atypical mitochondria in sensory neuron cell bodies and peripheral nerves and a twofold increase in the production of mitochondrial reactive oxygen species (ROS) [22]. Hence, it could be speculated that if the inactivation of oxidative stress is hampered because of reduced enzyme activ- Interestingly, we found that obesity was significantly more common in patients with DIPN [25% of DIPN patients were obese (BMI Ն 30), whereas only 8% of patients without DIPN were obese (p ϭ 0.008)]. No other study has reported this association, and we did not fi nd the same association in a prospective randomized study of 1725 patients [3]. One possible explanation for this observation is that the presence of obesity alters the metabolism of docetaxel [25]. It has been shown that paclitaxel clearance is greater in obese patients (BMI Ն 30) ity, the risk of peripheral neuropathy is greater, but the exact mechanism remains unclear. Sissung et al. have studied the role of polymorphisms in ABCB1 for DIPN [6]. They found that patients with the ABCB1 2677GG genotype developed neuropathy signifi cantly slower than carriers of at least one variant allele [6], however this was only observed in patients who also received thalidomide, which also has neurotoxic properties [23]. These fi ndings are therefore consistent with our results. The SCOTROC1 study [8] also tested ABCB1 in 539 patients with ovarian cancer and found no association between SNPs in ABCB1 (including 3435C Ͼ T) and DIPN. The recently published study by Abraham et al. found an association between ABCB1 (rs3213619) and taxane-induced PN in a meta-type analysis of 1303 patients with  [25], but the opposite is true for docetaxel, where the half-life is longer in the obese [25], resulting in greater exposure to the drug. The association with no positive lymph nodes and a higher risk of DIPN was also observed in a recent prospective randomized study by our group [3]. There is no obvious explanation for the association, but at least it indicates that axillary dissection in node positive patients does not increase the risk of DIPN. One explanation for the association of DIPN with low-dose docetaxel is that the dose of the drug is reduced in patients who develop DIPN during treatment, and hence the lower odds of a patient receiving a higher dose of docetaxel.
The strengths of the present study are that it is a case-control study that has included the largest number of patients with DIPN published until now. Due to the nature of a case-control study, our study can only be hypothesis generating, and the results should be confi rmed either in a prospectively trial or in several other, well-conducted retrospective trials with the same methodology. However, we included 75 patients with DIPN who had a very homogeneous background, supporting the fact that the fi ndings of this study are genetic.
In conclusion, we demonstrated that the occurrence of grade Ն 2 DIPN was signifi cantly more frequent in patients carrying GSTP1 rs1138272 C/T or T/T (one allele of 114Val). This fi nding supports the theory of a role of oxidative stress in DIPN pathophysiology, and, if confi rmed, can play a role in risk assessment of DIPN and perhaps also lead to better management of neurotoxicity.

Acknowledgments
We wish to acknowledge the work and invaluable assistance of laboratory technicians Susanne Hill-  3.30 (1.29 -8.40) 3.85 (1.34 -11.09) T/T * * Two patients with T/T genotype are calculated as C/T; † Adjusted for BMI, number of positive lymph nodes, cumulated dose of docetaxel and rs1138272. brandt, the Department of Clinical Biochemistry and Pharmacology, University of Southern Denmark, and the clinical staff at the Department of Oncology, Odense University Hospital, Denmark. This study was supported fi nancially by " Kr ae ftFonden " and " Aase og Ejnar Danielsens Fond " .

Declaration of interest:
The authors report no confl icts of interest. The authors alone are responsible for the content and writing of the paper.