Purpose: Paclitaxel, a widely used chemotherapeutic drug, can cause peripheral neuropathies leading to dose reductions and treatment suspensions and decreasing the quality of life of patients. It has been suggested that genetic variants altering paclitaxel pharmacokinetics increase neuropathy risk, but the major causes of interindividual differences in susceptibility to paclitaxel toxicity remain unexplained. We carried out a whole-exome sequencing (WES) study to identify genetic susceptibility variants associated with paclitaxel neuropathy.

Experimental Design: Blood samples from 8 patients with severe paclitaxel-induced peripheral neuropathy were selected for WES. An independent cohort of 228 cancer patients with complete paclitaxel neuropathy data was used for variant screening by DHPLC and association analysis. HEK293 cells were used for heterologous expression and characterization of two novel CYP3A4 enzymes.

Results: WES revealed 2 patients with rare CYP3A4 variants, a premature stop codon (CYP3A4*20 allele) and a novel missense variant (CYP3A4*25, p.P389S) causing reduced enzyme expression. Screening for CYP3A4 variants in the independent cohort revealed three additional CYP3A4*20 carriers, and two patients with missense variants exhibiting diminished enzyme activity (CYP3A4*8 and the novel CYP3A4*27 allele, p.L475V). Relative to CYP3A4 wild-type patients, those carrying CYP3A4 defective variants had more severe neuropathy (2- and 1.3-fold higher risk of neuropathy for loss-of-function and missense variants, respectively, P = 0.045) and higher probability of neuropathy-induced paclitaxel treatment modifications (7- and 3-fold higher risk for loss-of-function and missense variants, respectively, P = 5.9 × 10−5).

Conclusion: This is the first description of a genetic marker associated with paclitaxel treatment modifications caused by neuropathy. CYP3A4 defective variants may provide a basis for paclitaxel treatment individualization. Clin Cancer Res; 21(2); 322–8. ©2014 AACR.

Translational Relevance

Paclitaxel is a cytotoxic agent widely used for the treatment of many cancers. Treatment with this drug frequently results in peripheral sensory neuropathy that can seriously affect patients' quality of life. We and others have shown that variant alleles moderately decreasing the expression of genes involved in paclitaxel metabolism (i.e., CYP2C8*3 or CYP3A4*22) are associated with paclitaxel-induced neuropathy. The identification of predictive genetic markers for paclitaxel-induced dose-limiting neuropathy could lead to individualized risk assessment, facilitating treatment decision-making and therefore being of great clinical value. In this study, by whole-exome sequencing of severe paclitaxel-induced peripheral neuropathy patients, we confirm the earlier described implication of CYP3A4 in paclitaxel-induced neuropathy and find an association of CYP3A4 defective variants with paclitaxel treatment modifications. This study emphasizes the need to screen for rare genetic variants in selected cohorts of patients and may provide a basis for paclitaxel treatment individualization.

Paclitaxel is an antimicrotubular agent widely used for the treatment of many solid tumors. Peripheral neuropathy is the major toxicity limiting the clinical utility of this drug (1, 2). The degree of neuropathy is highly variable among patients, and while some remain asymptomatic throughout treatment, those with severe neuropathy can require paclitaxel dose reductions and treatment suspension, and therefore receive potentially suboptimal treatment. The most severe cases sustain long-term damage to the peripheral nerves, substantially reducing their quality of life (3).

Paclitaxel-induced neuropathy is dose dependent, and there are various clinical conditions that have been suggested as risk factors, such as diabetes mellitus, chronic liver disease, alcoholism, and previous neuropathies (4). Genetic variation has also been suggested as a factor influencing neuropathy risk, based on both genome-wide association studies (5, 6) and candidate gene approaches focused on paclitaxel pharmacokinetics (7). A correlation between the severity of the neuropathy and paclitaxel levels in plasma has been described (8, 9). In fact, it has been shown that common polymorphisms in the two genes encoding paclitaxel metabolizing enzymes in the liver, CYP2C8 and CYP3A4, are associated with a moderately increased risk of developing neuropathy during paclitaxel treatment (CYP2C8*3 and CYP3A4*22 alleles) (8, 10, 11). The identification of predictive genetic markers for paclitaxel-induced dose-limiting neuropathy could lead to individualized risk assessment, facilitating treatment decision-making and therefore being of great clinical value.

The newly developed whole-exome sequencing (WES) technology facilitates the identification of mutations and rare variants in exons and exon/intron boundaries that may potentially be implicated in disease and in extreme phenotypes (12–16). Thus, WES could be applied to unveil novel high-impact alleles of importance for interindividual variation in drug metabolism and adverse drug effects. In this study, we identified a loss-of-function allele and a novel missense variant in the CYP3A4 gene among 8 patients with severe paclitaxel-induced neuropathy. Further screening for CYP3A4 variants in an independent patient cohort revealed additional loss-of-function and missense allele carriers. Patients with CYP3A4 defective variants had higher risk of neuropathy and a large increased risk of paclitaxel dose reductions or treatment suspensions. These results highlight the fact that genetic variants that are rare in the general population might be more prevalent in patient groups developing adverse drug reactions and indeed constitute pharmacogenomic biomarkers of value for individualized therapy.

Materials used

Dibenzylfluorescein (DBF), fluorescein, paclitaxel, 3ρ-hydroxypaclitaxel (3OH-P), NADP+, glucose-6-phosphate, and yeast glucose-6-phosphate dehydrogenase were purchased from Sigma-Aldrich (Sigma-Aldrich Sweden AB). Lipofectamine LTX/PLUS and cell medium were purchased from Invitrogen (Life Technologies Europe BV). All solvents for high-performance liquid chromatography (HPLC) assay were obtained from Merck KGaA.

Patients

Blood samples from 8 patients with breast cancer with severe paclitaxel-induced neuropathy were collected at Spanish Hospitals for this study. All 8 patients had developed grade 3 sensory neuropathy (NCI-CTC v4; http://www.eortc.be/services/doc/ctc/) during weekly paclitaxel treatment at a cumulative dose ≤800 mg/m2. Furthermore, the neuropathy resulted in treatment suspensions or paclitaxel dose reductions and continued with at least grade 2 intensity for more than 18 months after termination of paclitaxel treatment (Table 1). Other causes of neuropathy, such as diabetes mellitus, alcoholism, hepatic diseases, AIDS, and previous neuropathies, were ruled out. DNA from an independent cohort of 228 patients with breast and ovarian cancer treated with paclitaxel and recruited in different Spanish hospitals from Madrid, starting on January 2011, was available for genetic analysis. The primary objective of this series was to study paclitaxel-induced peripheral neuropathy, for this reason each patient had a complete neuropathy assessment. For each patient, the following information was available: demographics, tumor characteristics, maximum sensory neuropathy grade during paclitaxel treatment, neuropathy evolution once paclitaxel treatment was ceased, cumulative dose of paclitaxel, and paclitaxel dose reductions and suspensions and their causes (Supplementary Table S1). All patients with cancer were older than 18 years of age, had documented histologic cancer neoplasia, a life expectancy of ≥12 weeks and ECOG performance status ≤2, adequate bone marrow, renal and hepatic function and no previous history of neuropathy, and had taken some form of contraception.

Table 1.

Characteristics of the 8 patients included in WES

CharacteristicNa
Age (y) 
 Median (min–max) 57 (39–79) 
Tumor stage 
 I 
 II 
 III 
 IV 
First-line chemotherapy treatment 
 FEC+Tb 
 FEC+T+Hc 
 FEC75+T80d 
 T+He 
Chemotherapy cycles 
 Min–max 5–8 
Paclitaxel dose at grade 3 sensory neuropathy (mg/m2
 Median (min–max) 750 (400–800) 
Duration of sensory neuropathy grade 2–3f (months) 
 Median (min–max) 38 (19–66) 
Paclitaxel treatment modifications due to neuropathyg 
 Treatment suspension 
 Dose reduction 
CharacteristicNa
Age (y) 
 Median (min–max) 57 (39–79) 
Tumor stage 
 I 
 II 
 III 
 IV 
First-line chemotherapy treatment 
 FEC+Tb 
 FEC+T+Hc 
 FEC75+T80d 
 T+He 
Chemotherapy cycles 
 Min–max 5–8 
Paclitaxel dose at grade 3 sensory neuropathy (mg/m2
 Median (min–max) 750 (400–800) 
Duration of sensory neuropathy grade 2–3f (months) 
 Median (min–max) 38 (19–66) 
Paclitaxel treatment modifications due to neuropathyg 
 Treatment suspension 
 Dose reduction 

aUnless otherwise indicated.

bFEC+T: 5-fluorouracil 600 mg/m2, epirubicin 90 mg/m2, and cyclophosphamide 600 mg/m2, every 21 days, followed by paclitaxel 100 mg/m2, every 7 days.

cFEC+T+H: 5-fluorouracil 600 mg/m2, epirubicin 90 mg/m2, and cyclophosphamide 600 mg/m2, every 21 days, followed by paclitaxel 100 mg/m2 plus herceptin (6 mg/kg loading dose; 2 mg/kg subsequent doses), every 7 days.

dFEC75+T80: 5-fluorouracil 600 mg/m2, epirubicin 75 mg/m2, and cyclophosphamide 600 mg/m2, every 21 days, followed by paclitaxel 80 mg/m2, every 7 days.

eT+H: paclitaxel 80 mg/m2 plus herceptin (4 mg/kg loading dose; 2 mg/kg subsequent doses), every 7 days.

fDuration of grade 2 or 3 sensorial neuropathy after finishing paclitaxel treatment.

gWhen in the same patient paclitaxel dose was reduced and later treatment was suspended, the patient is included in the table as “treatment suspension”.

To ensure homogeneity in neuropathy grading across the different collaborating centers, a qualified nurse (L. Sánchez) trained by a neurologist (G. Gutiérrez-Gutiérrez) interviewed by telephone all patients included in the study in a systematic manner to determine severity of symptoms and impairment in activities of daily living (e.g. extension and intensity of paresthesia, sensitivity, strength in hands and feet) and evaluated the neuropathy grade according to NCI-CTC v4. The recruitment of patients and collection of samples was approved by local internal ethical review committees and all patients gave written informed consent to participate in the study.

WES

WES of DNA samples from the 8 patients with extreme sensory neuropathy (Table 1) was carried out at the National Centre for Genomic Analysis (CNAG). DNA was isolated from peripheral blood using the FlexiGene DNA Kit (Qiagen) and quality control was performed according to electrophoresis and spectrophotometric measurements. The Covaris S2 System (Covaris) was used for DNA fragmentation and exome capture was performed using the SureSelect XT HumanAllExon 50Mb kit (Agilent Technologies). Library size and concentration was determined using Bioanalyzer 2100 (Agilent Technologies). Exome sequencing at a mean coverage >50× was performed using 75-bp paired-end technology in a HiSeq2000 (Illumina). Real-time image analysis and base calling was performed using Illumina's Real Time Analysis software version 1.6 using standard parameters. The GEM (http://algorithms.cnag.cat/wiki/The_GEM_library) and BFAST (17) programs were used to align the reads against the whole human genome (hg19 assembly). To identify single-nucleotide variants (SNV) and insertion-deletions (indels), the SAMtools program was used (http://samtools.sourceforge.net). Variants were filtered to rule out those in genome regions with low mappability, those with a strand bias P value < 0.001 in at least one sample and those with low depth read (<15×), the alternative allele present in <20% of reads, and/or the alternative allele present only in forward or reverse reads.

CYP3A4 variant detection

The full CYP3A4 coding region was amplified in the prospective cohort by PCR using specific primers (Supplementary Table S2). Screening for CYP3A4 variants was performed using denaturing high-performance liquid chromatography (DHPLC), in the DNA WAVE system 4500 HT (Transgenomic), equipped with a DNASep column (Transgenomic).

Sequencing of PCR products was performed on an ABI PRISM 3700 DNA Analyzer capillary sequencer (Applied Biosystems).

Genotyping was performed on 15 ng of genomic DNA using the KASPar SNP Genotyping System (Kbiosciences). All assays included DNA samples with known genotypes and negative controls. The Sequence Detection System ABI PRISM 7900HT (Applied Biosystems) was used to determine fluorescence and for allele assignment.

CYP3A4 expression vectors and heterologous expression

The coding region of CYP3A4.1 cDNA (NM_017460.5) was cloned into pCMV5 at the XbaI and KpnI restriction enzyme sites, to generate pCYP3A4.1 plasmid. To introduce c.1165C>T and c.1423C>G variants, we used the QuikChange II Site-Directed Mutagenesis Kit (Agilent Technologies) following the manufacturer's instructions. The correct sequence of CYP3A4 variant plasmids (pCYP3A4-P389S and pCYP3A4-L475V) was confirmed by Sanger sequencing.

HEK293 cells were cultured in DMEM with 10% fetal bovine serum and 100 U/mL penicillin–streptomycin. Cells were transfected with pCYP3A4.1, pCYP3A4-P389S, pCYP3A4-L475V, or empty pCMV5, together with human cytochrome b5a (pCL-cytb5) using Lipofectamine LTX/PLUS, following the manufacturer's guidelines.

Western blot analysis

Cells were solubilized in RIPA buffer containing Complete Protease Inhibitor Cocktail (Roche Diagnostics). Cell lysates or microsomes, prepared as previously described (18) and containing an equal amount of total protein, were separated using 15% SDS-polyacrylamide gel (18). Membranes were probed with primary antibodies against CYP3A4 (α-hCYP3A4, 1:1,000; ref. 19), human cytochrome b5a (α-cytb5a, 1:1,000; Santa Cruz Biotechnology) and, as loading control, the housekeeping escort chaperone ERp29 (α-ERp29, 1:1,000; ref. 20). The amount of expressed CYP3A4 apoproteins was calculated by densitometric analysis (Image Gauge, v.4.0; Fujifilm) of Western blot bands using a standard calibration curve based on CYP3A4 supersomes (BD Biosciences). To determine protein stability, 48 hours posttransfection HEK293 cells were exposed to 50 μmol/L cycloheximide for 8 hours. The CYP3A4 degradation rate was estimated by immunoblotting and further densitometric analysis of protein bands.

Determination of CYP3A4 enzyme activity

All incubations were conducted as previously described (21, 22) with minor modifications. Briefly, microsomal fractions corresponding to 160 μg of protein were mixed with different concentrations of dibenzylfluorescein in 50 mmol/L potassium phosphate buffer (pH 7.4) and the reaction was initiated by adding a prewarmed NADPH-regenerating system. Fluorescein formation was proportional to incubation time and concentration of microsomes. After 60 minutes, formation of metabolites was measured in a SPECTRAmax Gemini microplate spectrofluorometer (Molecular Devices) and results were analyzed using SoftMax Pro5 software.

Statistical analysis

Michaelis–Menten constants Km, Vmax, and intrinsic clearance (CLint = Vmax/Km) were calculated by nonlinear regression analysis using GraphPad and statistical significance was assessed using a paired t test. Association between neuropathy grade (ranked 0 to 3) and specific genetic groups was assessed using the Goodman and Kruskal Gamma test. Association between treatment modifications (binary variable) and specific genetic groups was assessed using the Fisher exact test. In these analyses, potential confounders were accounted for by stratification using the Mantel–Haenszel test. The Pearson correlation test was used to compare all genetic groups simultaneously (the genetic variable ranked in three groups: wild-type, 0; missense variants, 1; loss-of-function variants, 2) with neuropathy grade and treatment modifications. The analysis SPSS software package v.19 was used for all statistical analyses. P values less than 0.05 were considered statistically significant.

Detection of CYP3A4*20 and CYP3A4 c.1165C>T (p.P389S) variants by WES in 2 patients with extreme paclitaxel neuropathy

We first screened for loss-of-function and missense variants in critical genes involved in paclitaxel pharmacokinetics (CYP3A4, CYP2C8, ABCB1, and SLCO1B3) in 8 patients with extreme sensory neuropathy, and identified two high-impact variants in CYP3A4 (Supplementary Table S3). One was the CYP3A4*20 allele, a rare deleterious indel causing a frameshift and premature stop codon (c.1461_1462insA, p.P488Tfs*494). The other was a missense variant (c.1165C>T, p.P389S) located in the highly conserved CYP β-helix 4; it had not been reported previously, and was given the name CYP3A4*25 by the CYP allele nomenclature committee (www.cypalleles.ki.se). The patients with these variants were women treated with adjuvant FEC+T for breast cancer that upon paclitaxel treatment developed grade 3 neuropathy with loss of sensitivity in hands and feet, dysesthesia and clumsiness, and walking problems. The CYP3A4*20 carrier had two paclitaxel dose reductions due to the neuropathy and more than 40 months after paclitaxel treatment the symptoms persisted with improvement to grade 2. The CYP3A4*25 carrier had suspension of paclitaxel treatment after cycle 6 due to the neuropathy and 19 months after paclitaxel treatment symptoms had decreased to grade 2 neuropathy; 25 months after paclitaxel treatment the patient did not report neuropathy symptoms. Sanger sequencing confirmed the presence of both variants in heterozygosity (Fig. 1).

Figure 1.

Coding variants identified by WES in CYP3A4. Sanger sequencing confirming the presence of (A) CYP3A4*20 (c.1461_1462insA) and (B) CYP3A4*25 (c.1165C>T) variant alleles. The blue-highlighted regions indicate the position of the nucleotide change. C, segment of CYP3A4 protein sequence alignment and predicted secondary structure. Proline 389, marked in red, is highly conserved across species. *, positions with a single, fully conserved residue; “:,” amino acids with very similar properties; “.,” amino acids with mildly similar properties.

Figure 1.

Coding variants identified by WES in CYP3A4. Sanger sequencing confirming the presence of (A) CYP3A4*20 (c.1461_1462insA) and (B) CYP3A4*25 (c.1165C>T) variant alleles. The blue-highlighted regions indicate the position of the nucleotide change. C, segment of CYP3A4 protein sequence alignment and predicted secondary structure. Proline 389, marked in red, is highly conserved across species. *, positions with a single, fully conserved residue; “:,” amino acids with very similar properties; “.,” amino acids with mildly similar properties.

Close modal

In the CYP2C8 gene, encoding the other CYP enzyme metabolizing paclitaxel, and in SLCO1B3 and ABCB1, encoding the uptake and efflux paclitaxel transporters, respectively, we only detected previously described missense variants, none of which were predicted to affect protein function. In addition, with the exception of CYP3A4, the frequency of most variants in the patients with extreme neuropathy was similar to that reported in the general population (Supplementary Table S3 shows coding polymorphisms and the regulatory intronic CYP3A4*22). Thus, we selected the CYP3A4 gene for further study.

Screening for CYP3A4 variants in a cohort of paclitaxel-treated patients

To determine whether additional CYP3A4 coding variants are carried by patients with paclitaxel-induced neuropathy, we examined by DHPLC an independent cohort of 228 patients with cancer treated with the drug (Supplementary Table S1). We detected three additional patients carrying the CYP3A4*20 allele, another patient with CYP3A4*8 allele (c.389G>A, p.R130Q), and another one with a novel CYP3A4 missense variant (c.1423C>G, p.L475V, named CYP3A4*27). The characteristics of the patients carrying CYP3A4 coding variants are shown in Table 2.

Table 2.

Characteristics of patients with CYP3A4 coding variants

PatientCYP3A4 variant allele effectCYP3A4 genotypeDetection techniqueaNeuropathy gradebTotal cycles (n)Treatment modificationscTime with neuropathy (mo)d
11S1213  *1/ *20 WES Red (at cycles 4 & 8) (40) 
11S919 Loss-of-function *1/ *20 DHPLC Red (at cycle 4) (9) 
13S812  *1/ *20 DHPLC 12 Red (at cycle 9) (10) 
13S120  *1/ *20 DHPLC 11 No 16 
11S872  *1/ *25 WES Susp (after cycle 6) 19 
12S513 Decr. activity *1/ *27 DHPLC No (19) 
11S918  *1/ *8 DHPLC 12 No 
PatientCYP3A4 variant allele effectCYP3A4 genotypeDetection techniqueaNeuropathy gradebTotal cycles (n)Treatment modificationscTime with neuropathy (mo)d
11S1213  *1/ *20 WES Red (at cycles 4 & 8) (40) 
11S919 Loss-of-function *1/ *20 DHPLC Red (at cycle 4) (9) 
13S812  *1/ *20 DHPLC 12 Red (at cycle 9) (10) 
13S120  *1/ *20 DHPLC 11 No 16 
11S872  *1/ *25 WES Susp (after cycle 6) 19 
12S513 Decr. activity *1/ *27 DHPLC No (19) 
11S918  *1/ *8 DHPLC 12 No 

aConfirmation of variants was performed by Sanger sequencing.

bMaximum sensory neuropathy grade during paclitaxel treatment (NCI-CTC v4).

cModifications due to neuropathy. Red, reduction of the paclitaxel dose; Susp, suspension of paclitaxel treatment.

dDuration of neuropathy after finishing paclitaxel treatment. For cases for which the neuropathy was ongoing or further evaluation could not be performed because the patient was lost, the duration is written in parentheses.

Stability and enzymatic activity of CYP3A4.25 and CYP3A4.27

The CYP3A4*20 allele has been shown to encode a nonfunctional enzyme (23) and the CYP3A4*8 allele was shown to cause decreased CYP3A4 activity (24), but no functional data exist on the novel missense variants CYP3A4*25 and CYP3A4*27. HEK293 cells transiently expressing CYP3A4-P389S or CYP3A4-L475V (CYP3A4.25 and CYP3A4.27, respectively) showed substantially lower amounts of protein compared with CYP3A4-wild-type (CYP3A4.1; Fig. 2). The level of the CYP3A4.27 protein in the expression system was estimated to be 10% of the corresponding expression of CYP3A4.1. The level of CYP3A4.25 was also relatively low, about 40% of the CYP3A4.1 levels and treatment of the cells with the protein synthesis inhibitor cycloheximide confirmed that the P389S substitution in CYP3A4.25 caused an increased rate of degradation (Supplementary Fig. S1). By analyzing the mRNA levels, it was found that the levels were slightly lower (about 80%–70% of the control) for the mutant variants as compared with CYP3A4.1 in the expression system (data not shown). Analyses of catalytic activities of the variant enzymes using dibenzylfluorescein as a CYP3A4 substrate revealed a similar Km value for CYP3A4.27 (Km, 7.4 ± 1.8 μmol/L) as compared with CYP3A4.1, whereas the Km value for CYP3A4.25 was somewhat higher (Km, 31.7 ± 2.8 μmol/L). The true Vmax was difficult to determine because of low expression of the variant proteins (data not shown). We conclude that both CYP3A4.25 and CYP3A4.27 have decreased stability in the expression system used.

Figure 2.

Expression of CYP3A4.25 and CYP3A4.27. Lysates from transfected HEK293 cells were immunoblotted with antibodies for CYP3A4 and ERp29, the latter as a loading control. A representative image is shown for (A) CYP3A4.25 and (B) CYP3A4.27, with samples loaded in triplicates (duplicates for mock transfected cells). Densitometric analysis estimated that CYP3A4 protein expression was 40% and 10% of the corresponding level of CYP3A4.1, for CYP3A4.25 and CYP3A4.27, respectively.

Figure 2.

Expression of CYP3A4.25 and CYP3A4.27. Lysates from transfected HEK293 cells were immunoblotted with antibodies for CYP3A4 and ERp29, the latter as a loading control. A representative image is shown for (A) CYP3A4.25 and (B) CYP3A4.27, with samples loaded in triplicates (duplicates for mock transfected cells). Densitometric analysis estimated that CYP3A4 protein expression was 40% and 10% of the corresponding level of CYP3A4.1, for CYP3A4.25 and CYP3A4.27, respectively.

Close modal

CYP3A4 variants are associated with an increased risk of neuropathy and paclitaxel treatment modifications

Thus, in a total of 236 patients, composed of 8 WES-studied patients and 228 patients used for CYP3A4 screening, 4 carried a loss-of-function variant (CYP3A4*20) and three carried rare missense variants giving rise to decreased enzymatic activity (CYP3A4*25, CYP3A4*27 and CYP3A4*8; Table 2).

The patients with CYP3A4 loss-of-function variants and patients with missense variants showed a 2- and 1.3-fold increased risk of grade 3 neuropathy, respectively, when compared with wild-type CYP3A4 patients (Fig. 3A). The neuropathy grade was significantly different between patients with loss-of-function variants and wild-type homozygotes (P = 0.042), and including missense variants in the analysis only minimally changed the P value (P = 0.045). Furthermore, 14% of patients with paclitaxel dose reductions or treatment suspensions due to neuropathy carried CYP3A4 variants, and a 7- and 3-fold increased risk of treatment changes was observed in patients with loss-of-function and missense variants, respectively, when compared with wild-type CYP3A4 patients (Fig. 3B). This increased risk of treatment modifications for patients with genetically decreased CYP3A4 activity was statistically significant (P = 5.8 × 10−3 for loss-of-function variants, P = 5.9 × 10−5 when missense variants were included in the analysis). When all paclitaxel treatment modifications (i.e., including those due to reasons other than neuropathy) were considered, CYP3A4 variants were still associated with an increased risk of dose changes (P = 7.7 × 10−3).

Figure 3.

CYP3A4 defective variants confer an increased risk of paclitaxel-induced neuropathy and treatment modifications. A, neuropathy grade was compared among patients with different CYP3A4 activity. All 4 patients with loss-of-function variants (100%), 2 of 3 patients (67%) with missense variants, and 116 of 229 patients (51%) with wild-type CYP3A4 had grade 3 sensory neuropathy. B, treatment modifications due to neuropathy were compared among patients with different CYP3A4 activity. Three of 4 patients (75%) with loss-of-function variants, 1 of 3 patients (33%) with missense variants, and 25 of 229 patients (11%) with wild-type CYP3A4 had treatment modifications. As described in Patients and Methods, the Goodman and Kruskal gamma test and Fisher exact test were used to assess association with neuropathy grade (γ = 1) and treatment modifications. To perform an analysis including simultaneously all variants categorized according to CYP3A4 activity (loss-of-function, missense, and wild-type), Pearson correlation test was used. Treatment modif PN, treatment modifications due to peripheral neuropathy.

Figure 3.

CYP3A4 defective variants confer an increased risk of paclitaxel-induced neuropathy and treatment modifications. A, neuropathy grade was compared among patients with different CYP3A4 activity. All 4 patients with loss-of-function variants (100%), 2 of 3 patients (67%) with missense variants, and 116 of 229 patients (51%) with wild-type CYP3A4 had grade 3 sensory neuropathy. B, treatment modifications due to neuropathy were compared among patients with different CYP3A4 activity. Three of 4 patients (75%) with loss-of-function variants, 1 of 3 patients (33%) with missense variants, and 25 of 229 patients (11%) with wild-type CYP3A4 had treatment modifications. As described in Patients and Methods, the Goodman and Kruskal gamma test and Fisher exact test were used to assess association with neuropathy grade (γ = 1) and treatment modifications. To perform an analysis including simultaneously all variants categorized according to CYP3A4 activity (loss-of-function, missense, and wild-type), Pearson correlation test was used. Treatment modif PN, treatment modifications due to peripheral neuropathy.

Close modal

Tumor type was not associated with the neuropathy, but conditions considered to be neuropathy risk factors (diabetes, high alcohol intake, restless-legs-syndrome) were significantly associated with neuropathy grade (Supplementary Fig. S2A). Cumulative paclitaxel dose was associated with treatment modifications, as expected, as these result in lower cumulative doses (Supplementary Fig. S2D). Concerning CYP3A4*22 allele, we found a trend toward higher treatment modifications in carriers of this variant (P = 0.066); however, no statistically significant differences were obtained for neuropathy grade and treatment modifications due to neuropathy (Supplementary Fig. S3). Accounting for neuropathy risk factors, cumulative paclitaxel dose or CYP3A4*22 allele did not substantially change the association observed for CYP3A4 defective variants.

Paclitaxel peripheral neuropathy affects a large number of patients and can lead to treatment modifications (25). Most patients recover from the neuropathy, but long-term nerve damage can also occur, compromising the quality of life of these patients. The extent of paclitaxel exposure is associated with the severity of the neuropathy (8, 9), and paclitaxel elimination is mediated by CYP2C8, CYP3A4, OATP1B3, and P-glycoprotein (7, 26, 27). Thus, alterations in the activity of these proteins could decrease drug elimination and consequently increase toxicity risk (e.g., CYP2C8*3, refs. 10, 11; CYP3A4*22, refs. 8, 28–30; rs1045642 in ABCB1, ref. 31).

The selection of phenotypic outliers has been shown to be an effective strategy to identify genetic variants associated with diseases and drug outcomes (13, 14, 23, 32), and when combined with massive parallel DNA sequencing (33) it is a powerful method to detect low-frequency susceptibility variants (16, 34, 35). Consequently, we designed a study to identify genetic variants in patients with severe paclitaxel-induced neuropathy. Chemotherapy-induced neurotoxicity studies are challenging due to subjectivity of the common toxicity scales and the difficult application of more accurate neuropathy scales across multiple centers (36). In this study, neuropathy was assessed in a systematic manner in all collaborating centers, and for the identification of extreme-phenotype patients, not only the severity of symptoms during treatment, but also modifications of treatment regimen and long-lasting disabilities were taken into account.

WES revealed two rare high-impact variants in CYP3A4 (CYP3A4*20 and CYP3A4*25). CYP3A4*20 is an indel leading to a premature stop codon previously described in one individual with impaired elimination of CYP3A4 substrates (23) and the novel CYP3A4.25 (P389S) protein had decreased stability and reduced amounts of apoprotein in the HEK293 expression system (Fig. 2A and Supplementary Fig. S1). In an independent cohort we found three more carriers of CYP3A4*20, and two carriers of missense variants, CYP3A4.8, described to have diminished activity (24), and CYP3A4.27, here found to be less expressed in comparison with CYP3A4.1 (Fig. 2B). In total, 3% of the Spanish patients carried CYP3A4 defective variants, suggesting that these could explain part of CYP3A4 variability. Although CYP3A4*20 allele is present in Spain it has a low frequency in most European, Asian, and African populations (37) however, alternative CYP3A4 defective variants might be relevant in other populations. In this respect, 3% of European Americans and 2% of African Americans seem to carry potentially defective CYP3A4 allele variants (loss-of-function or missense variants likely damaging according to Polyphen; see Exome Variant Server database).

Patients carrying loss-of-function CYP3A4 variants had a significantly higher risk of neuropathy and paclitaxel treatment modifications, when compared with wild-type CYP3A4 patients (P = 0.042 and P = 5.8 × 10−3, respectively; Fig. 3), and carriers of missense variants showed an intermediate phenotype, concordant with a decreased but not abolished CYP3A4 activity. For CYP3A4*22 allele, we only detected a trend toward increased treatment modifications, which might reflect a lower effect of this variant on paclitaxel metabolism and/ or a low statistical power due to the small number of CYP3A4*22 carriers (Supplementary Fig. S3). Four of the 29 patients with paclitaxel treatment modifications due to neuropathy carried CYP3A4 defective variants (missense or loss-of-function), indicating that genetic testing of CYP3A4 before treatment, would have a very high specificity (99%) but poor sensitivity (14%). If pathologic risk factors were also taken into consideration, the sensitivity would increase, and an estimated 27% of patients carrying CYP3A4 defective variants or with preexisting conditions associated with neuropathy, would require treatment modification due to severe neuropathy upon paclitaxel chemotherapy. Paclitaxel-induced neuropathy is a multifactorial and polygenic trait, and additional genetic variants, some yet to be identified, will further improve the predictive power of genetic testing. In this regard, Supplementary Table S4 shows genetic variants that have been described as moderate risk factors for paclitaxel-induced neuropathy (CYP3A4*22, CYP2C8*3, EPHA5-rs7349683, and XKR4-rs4737264, the latter two identified in a meta-analysis of genome-wide association studies; refs. 5, 6). Thus, the identification of genetic variants and physiopathologic risk factors predictive of paclitaxel-induced neuropathy may provide a basis on which to individualize this treatment. It is important to highlight that previous reports have only identified markers associated to neuropathy grade, but not with neuropathies resulting in treatment modifications. In fact, this is the first study describing a marker associated with paclitaxel dose-limiting neuropathy.

In summary, in this study we found an overrepresentation of defective CYP3A4 variants in patients with paclitaxel treatment modifications and in those with high-grade paclitaxel-induced neuropathy. This supports and confirms the earlier described implication of CYP3A4 in paclitaxel-induced neuropathy. These results emphasize the need to screen for rare genetic variants in selected cohorts of patients.

No potential conflicts of interest were disclosed.

Conception and design: M. Apellániz-Ruiz, M.-Y. Lee, G. Gutiérrez-Gutiérrez, I. Calvo, J. García-Donás, M. Ingelman-Sundberg, C. Rodríguez-Antona

Development of methodology: M. Apellániz-Ruiz, M.-Y. Lee, G. Gutiérrez-Gutiérrez, L.J. Leandro-García

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.-Y. Lee, L. Sánchez-Barroso, G. Gutiérrez-Gutiérrez, L. García-Estévez, M. Sereno, J. García-Donás, B. Castelo, E. Guerra, M. Robledo

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Apellániz-Ruiz, M.-Y. Lee, J. García-Donás, L.J. Leandro-García, I. Johansson, M. Ingelman-Sundberg, C. Rodríguez-Antona

Writing, review, and/or revision of the manuscript: M. Apellániz-Ruiz, M.-Y. Lee, L. Sánchez-Barroso, G. Gutiérrez-Gutiérrez, I. Calvo, L. García-Estévez, M. Sereno, J. García-Donás, B. Castelo, E. Guerra, L.J. Leandro-García, A. Cascón, I. Johansson, M. Robledo, M. Ingelman-Sundberg, C. Rodríguez-Antona

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.-Y. Lee, L. Sánchez-Barroso

Study supervision: M. Apellániz-Ruiz, C. Rodríguez-Antona

The authors thank CNAG personnel and especially Sergi Beltran and Raul Tonda for their support in whole-exome sequencing. The authors also thank Lucia Inglada-Pérez for her contribution to the statistical analyses.

This work was supported by projects from the Spanish Ministry of Economy and Competiveness (grant number SAF2012–35779) and by grants from The Swedish Cancer Foundation, The Swedish Research Council, and Karolinska Institutet in Sweden. María Apellániz-Ruiz is a predoctoral fellow of “la Caixa”/CNIO international PhD programme.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Mielke
S
,
Sparreboom
A
,
Mross
K
. 
Peripheral neuropathy: a persisting challenge in paclitaxel-based regimes.
Eur J Cancer
2006
;
42
:
24
30
.
2.
Argyriou
AA
,
Koltzenburg
M
,
Polychronopoulos
P
,
Papapetropoulos
S
,
Kalofonos
HP
. 
Peripheral nerve damage associated with administration of taxanes in patients with cancer.
Crit Rev Oncol Hematol
2008
;
66
:
218
28
.
3.
Hershman
DL
,
Weimer
LH
,
Wang
A
,
Kranwinkel
G
,
Brafman
L
,
Fuentes
D
, et al
Association between patient reported outcomes and quantitative sensory tests for measuring long-term neurotoxicity in breast cancer survivors treated with adjuvant paclitaxel chemotherapy.
Breast Cancer Res Treat
2011
;
125
:
767
74
.
4.
Lee
JJ
,
Swain
SM
. 
Peripheral neuropathy induced by microtubule-stabilizing agents.
J Clin Oncol
2006
;
24
:
1633
42
.
5.
Baldwin
RM
,
Owzar
K
,
Zembutsu
H
,
Chhibber
A
,
Kubo
M
,
Jiang
C
, et al
A genome-wide association study identifies novel loci for paclitaxel-induced sensory peripheral neuropathy in CALGB 40101.
Clin Cancer Res
2012
;
18
:
5099
109
.
6.
Leandro-Garcia
LJ
,
Inglada-Perez
L
,
Pita
G
,
Hjerpe
E
,
Leskela
S
,
Jara
C
, et al
Genome-wide association study identifies ephrin type A receptors implicated in paclitaxel induced peripheral sensory neuropathy.
J Med Genet
2013
;
50
:
599
605
.
7.
Rodriguez-Antona
C
. 
Pharmacogenomics of paclitaxel.
Pharmacogenomics
2010
;
11
:
621
3
.
8.
de Graan
AJ
,
Elens
L
,
Sprowl
JA
,
Sparreboom
A
,
Friberg
LE
,
van der Holt
B
, et al
CYP3A4*22 genotype and systemic exposure affect paclitaxel-induced neurotoxicity.
Clin Cancer Res
2013
;
19
:
3316
24
.
9.
Mielke
S
,
Sparreboom
A
,
Steinberg
SM
,
Gelderblom
H
,
Unger
C
,
Behringer
D
, et al
Association of Paclitaxel pharmacokinetics with the development of peripheral neuropathy in patients with advanced cancer.
Clin Cancer Res
2005
;
11
:
4843
50
.
10.
Leskela
S
,
Jara
C
,
Leandro-Garcia
LJ
,
Martinez
A
,
Garcia-Donas
J
,
Hernando
S
, et al
Polymorphisms in cytochromes P450 2C8 and 3A5 are associated with paclitaxel neurotoxicity.
Pharmacogenomics J
2011
;
11
:
121
9
.
11.
Hertz
DL
,
Roy
S
,
Motsinger-Reif
AA
,
Drobish
A
,
Clark
LS
,
McLeod
HL
, et al
CYP2C8*3 increases risk of neuropathy in breast cancer patients treated with paclitaxel.
Ann Oncol
2013
;
24
:
1472
8
.
12.
Biesecker
LG
. 
Exome sequencing makes medical genomics a reality.
Nat Genet
2010
;
42
:
13
4
.
13.
Cirulli
ET
,
Goldstein
DB
. 
Uncovering the roles of rare variants in common disease through whole-genome sequencing.
Nat Rev Genet
2010
;
11
:
415
25
.
14.
Nebert
DW
. 
Extreme discordant phenotype methodology: an intuitive approach to clinical pharmacogenetics.
Eur J Pharmacol
2000
;
410
:
107
20
.
15.
Manolio
TA
,
Collins
FS
,
Cox
NJ
,
Goldstein
DB
,
Hindorff
LA
,
Hunter
DJ
, et al
Finding the missing heritability of complex diseases.
Nature
2009
;
461
:
747
53
.
16.
Gurwitz
D
,
McLeod
HL
. 
Genome-wide studies in pharmacogenomics: harnessing the power of extreme phenotypes.
Pharmacogenomics
2013
;
14
:
337
9
.
17.
Homer
N
,
Merriman
B
,
Nelson
SF
. 
BFAST: an alignment tool for large scale genome resequencing.
PLoS ONE
2009
;
4
:
e7767
.
18.
Lee
MY
,
Borgiani
P
,
Johansson
I
,
Oteri
F
,
Mkrtchian
S
,
Falconi
M
, et al
High warfarin sensitivity in carriers of CYP2C9*35 is determined by the impaired interaction with P450 oxidoreductase.
Pharmacogenomics J
2014
;
14
:
343
9
.
19.
Edwards
RJ
,
Adams
DA
,
Watts
PS
,
Davies
DS
,
Boobis
AR
. 
Development of a comprehensive panel of antibodies against the major xenobiotic metabolising forms of cytochrome P450 in humans.
Biochem Pharmacol
1998
;
56
:
377
87
.
20.
Mkrtchian
S
,
Sandalova
T
. 
ERp29, an unusual redox-inactive member of the thioredoxin family.
Antioxid Redox Signal
2006
;
8
:
325
37
.
21.
Cheng
Q
,
Guengerich
FG
. 
High-throughput fluorescence assay of cytochrome P450 3A4.
In:
Phillips
IR
,
Shephard
EA
,
Ortiz de Montellano
PR
,
editors
. 
Methods in Molecular Biology: Cytochrome P450 Protocols.
New Jersey
:
Humana Press Inc
; 
2013
. p.
157
62
.
22.
Stresser
DM
,
Blanchard
AP
,
Turner
SD
,
Erve
JC
,
Dandeneau
AA
,
Miller
VP
, et al
Substrate-dependent modulation of CYP3A4 catalytic activity: analysis of 27 test compounds with four fluorometric substrates.
Drug Metab Dispos
2000
;
28
:
1440
8
.
23.
Westlind-Johnsson
A
,
Hermann
R
,
Huennemeyer
A
,
Hauns
B
,
Lahu
G
,
Nassr
N
, et al
Identification and characterization of CYP3A4*20, a novel rare CYP3A4 allele without functional activity.
Clin Pharmacol Ther
2006
;
79
:
339
49
.
24.
Eiselt
R
,
Domanski
TL
,
Zibat
A
,
Mueller
R
,
Presecan-Siedel
E
,
Hustert
E
, et al
Identification and functional characterization of eight CYP3A4 protein variants.
Pharmacogenetics
2001
;
11
:
447
58
.
25.
Postma
TJ
,
Vermorken
JB
,
Liefting
AJ
,
Pinedo
HM
,
Heimans
JJ
. 
Paclitaxel-induced neuropathy.
Ann Oncol
1995
;
6
:
489
94
.
26.
van de Steeg
E
,
van Esch
A
,
Wagenaar
E
,
Kenworthy
KE
,
Schinkel
AH
. 
Influence of human OATP1B1, OATP1B3, and OATP1A2 on the pharmacokinetics of methotrexate and paclitaxel in humanized transgenic mice.
Clin Cancer Res
2013
;
19
:
821
32
.
27.
Sparreboom
A
,
van Asperen
J
,
Mayer
U
,
Schinkel
AH
,
Smit
JW
,
Meijer
DK
, et al
Limited oral bioavailability and active epithelial excretion of paclitaxel (Taxol) caused by P-glycoprotein in the intestine.
Proc Natl Acad Sci U S A
1997
;
94
:
2031
5
.
28.
Klein
K
,
Thomas
M
,
Winter
S
,
Nussler
AK
,
Niemi
M
,
Schwab
M
, et al
PPARA: a novel genetic determinant of CYP3A4 in vitro and in vivo.
Clin Pharmacol Ther
2012
;
91
:
1044
52
.
29.
Wang
D
,
Guo
Y
,
Wrighton
SA
,
Cooke
GE
,
Sadee
W
. 
Intronic polymorphism in CYP3A4 affects hepatic expression and response to statin drugs.
Pharmacogenomics J
2011
;
11
:
274
86
.
30.
Elens
L
,
Nieuweboer
A
,
Clarke
SJ
,
Charles
KA
,
de Graan
AJ
,
Haufroid
V
, et al
CYP3A4 intron 6 C>T SNP (CYP3A4*22) encodes lower CYP3A4 activity in cancer patients, as measured with probes midazolam and erythromycin.
Pharmacogenomics
2013
;
14
:
137
49
.
31.
Sissung
TM
,
Mross
K
,
Steinberg
SM
,
Behringer
D
,
Figg
WD
,
Sparreboom
A
, et al
Association of ABCB1 genotypes with paclitaxel-mediated peripheral neuropathy and neutropenia.
Eur J Cancer
2006
;
42
:
2893
6
.
32.
Zhang
G
,
Nebert
DW
,
Chakraborty
R
,
Jin
L
. 
Statistical power of association using the extreme discordant phenotype design.
Pharmacogenet Genomics
2006
;
16
:
401
13
.
33.
Shendure
J
,
Ji
H
. 
Next-generation DNA sequencing.
Nat Biotechnol
2008
;
26
:
1135
45
.
34.
Ng
SB
,
Buckingham
KJ
,
Lee
C
,
Bigham
AW
,
Tabor
HK
,
Dent
KM
, et al
Exome sequencing identifies the cause of a mendelian disorder.
Nat Genet
2010
;
42
:
30
5
.
35.
Panoutsopoulou
K
,
Tachmazidou
I
,
Zeggini
E
. 
In search of low-frequency and rare variants affecting complex traits.
Hum Mol Genet
2013
;
22
:
R16
21
.
36.
Cavaletti
G
,
Frigeni
B
,
Lanzani
F
,
Piatti
M
,
Rota
S
,
Briani
C
, et al
The Total Neuropathy Score as an assessment tool for grading the course of chemotherapy-induced peripheral neurotoxicity: comparison with the National Cancer Institute-Common Toxicity Scale.
J Peripher Nerv Syst
2007
;
12
:
210
5
.
37.
Apellániz-Ruiz
M
,
Inglada-Pérez
L
,
G Naranjo
M
,
Sánchez
L
,
Mancikova
V
,
Currás-Freixes
M
, et al
High frequency and founder effect of the CYP3A4*20 loss-of-function allele in the Spanish population classifies CYP3A4 as a polymorphic enzyme.
Pharmacogenomics J
2014 Nov 4
.
[Epub ahead of print]
.