Inherited NUDT15 variant is a genetic determinant of mercaptopurine intolerance in children with acute lymphoblastic leukemia

Abstract

Purpose: Mercaptopurine (MP) is the mainstay of curative therapy for acute lymphoblastic leukemia (ALL). We performed a genome-wide association study (GWAS) to identify comprehensively the genetic basis of MP intolerance in children with ALL.

Patients and methods: The discovery GWAS and replication cohorts included 657 and 371 children from two prospective clinical trials. MP dose intensity was a marker for drug tolerance and toxicities and was defined as prescribed dose divided by the planned protocol dose during maintenance therapy; its association with genotype was evaluated using a linear mixed-effects model.

Results: MP dose intensity varied by race and ethnicity and was negatively correlated with East Asian genetic ancestry (P < .001). The GWAS revealed two genome-wide significant loci associated with dose intensity: rs1142345 in TPMT (Tyr240Cys, present in *3A and *3C variants; P = 8.6 × 10(-9)) and rs116855232 in NUDT15 (P = 8.8 × 10(-9)), with independent replication. Patients with TT genotype at rs116855232 were exquisitely sensitive to MP, with an average dose intensity of 8.3%, compared with those with TC and CC genotypes, who tolerated 63% and 83.5% of the planned dose, respectively. The NUDT15 variant was most common in East Asians and Hispanics, rare in Europeans, and not observed in Africans, contributing to ancestry-related differences in MP tolerance. Of children homozygous for either TPMT or NUDT15 variants or heterozygous for both, 100% required ≥ 50% MP dose reduction, compared with only 7.7% of others.

Conclusion: We describe a germline variant in NUDT15 strongly associated with MP intolerance in childhood ALL, which may have implications for treatment individualization in this disease.

Trial registration: ClinicalTrials.gov NCT00137111 NCT00268528.

Conflict of interest statement

Authors' disclosures of potential conflicts of interest are found in the article online at www.jco.org. Author contributions are found at the end of this article.

© 2015 by American Society of Clinical Oncology.

Figures

Fig 1.

Mercaptopurine (MP) dose intensity and genetic ancestry. For patients enrolled onto AALL03N1 protocol, MP dose was adjusted during maintenance therapy on the basis of host toxicities (myelosuppression and infections), and dose intensity was defined as ratio of prescribed dose over protocol planned dose (75 mg/mg2 per day). Dose intensity was measured longitudinally over 6 months and is shown as single cumulative value for study period. Patients were grouped into five racial/ethnic categories on the basis of genetic ancestry. Genetically defined East Asians had lowest MP dose intensity (ie, most likely to be MP intolerant). P value was estimated by using Kruskal-Wallis test. Each box includes data between 25th and 75th percentiles, with horizontal line indicating median. Whiskers indicate maximal and minimal observations within 1.5× length of box.

Fig 2.

Genome-wide association study identified TPMT and NUDT15 variants associated with mercaptopurine (MP) dose intensity. (A) Total of 40,889 single-nucleotide polymorphisms (SNPs) were tested for associations with MP dose intensity in 657 children with acute lymphoblastic leukemia in AALL03N1 cohort. Association P value (y-axis) is plotted against respective chromosomal position of each SNP (x-axis). Dashed horizontal line indicates genome-wide significance threshold (P < 5 × 10−8). TPMT and NUDT15 loci are indicated by arrows. MP dose intensity was significantly lower for patients heterozygous or homozygous for risk alleles at (B) TPMT SNP rs1142345 (contributing to *3A and *3C) and (C) NUDT15 SNP rs116855232, compared with those with wild-type genotypes. P value was estimated using linear mixed-effects model, with genetic ancestry as covariate. Each box includes data between 25th and 75th percentiles, with horizontal line indicating median. Whiskers indicate maximal and minimal observations within 1.5× length of box.

Fig 3.

NUDT15 variant was more frequent in East Asians but associated with mercaptopurine (MP) dose intensity regardless of ancestry. (A) T (risk) allele at rs116855232 was most common in individuals of East Asian descent followed by Hispanics, rare in Europeans, and not observed in Africans. T allele was consistently linked to lower MP dose intensity in (B) East Asians and (C) Hispanics; (D) similar trend was observed in Europeans. P values were estimated using linear mixed-effects model. Each box includes data between 25th and 75th percentiles, with horizontal line indicating median. Whiskers indicate maximal and minimal observations within 1.5× length of box.

Fig 4.

Combined effects of TPMT and NUDT15 variants on mercaptopurine (MP) intolerance. Each patient in (A) AALL03N1 and (B) St Jude cohorts was first assigned genetic risk score based on genotype at NUDT15 variant (rs116855232) and TPMT variants (rs1142345, rs1800462, and rs1800460; contributing to *2, *3A, *3B, and *3C). MP dose intensity decreased as genetic risk score increased in both cohorts. P value was estimated using Spearman correlation. Each box includes data between 25th and 75th percentiles, with horizontal line indicating median. Whiskers indicate maximal and minimal observations within 1.5× length of box. HET, heterozygous; HOM, homozygous; WT, wild type.

Fig A1.

Genetic ancestry and self-reported race/ethnicity. Genetic ancestry was estimated based on genotypes at 30,000 randomly selected single-nucleotide polymorphisms using STRUCTURE software (version 2.2.3; http://pritchardlab.stanford.edu/structure.html). (A) European, (B) East Asian, (C) Native American, and (D) African ancestries are plotted for four self-reported race/ethnic groups in AALL03N1 cohort.

Fig A2.

East Asian genetic ancestry was related to lower mercaptopurine (MP) dose intensity within self-reported Asians. In self-identified Asian patients in AALL03N1 cohort, those with greater East Asian genetic ancestry had lower MP dose intensity. P value was estimated using Wilcoxon test. Each box includes data between 25th and 75th percentiles, with horizontal line indicating median. Whiskers indicate maximal and minimal observations within 1.5× length of box.

Fig A3.

Association of TPMT genotype (rs1142345) and thiopurine methyltransferase (TPMT) enzymatic activity. TPMT activity was measured from erythrocyte extracts for (A) 643 children on AALL03N1 protocol and (B) 311 patients in St Jude cohort. P value was estimated using linear regression with ancestry as covariate. Each box includes data between 25th and 75th percentiles, with horizontal line indicating median. Whiskers indicate maximal and minimal observations within 1.5× length of box.

Fig A4.

Frequency of TPMT variants in different racial/ethnic groups in AALL03N1 cohort. Composite TPMT genotype was defined by rs1142345, rs1800462, and rs1800460 (contributing to *2, *3A, *3B, and *3C variants). Genetically defined racial/ethnic groups in AALL03N1 cohort were based on genetic ancestry, as described in Patients and Methods.

Fig A5.

Longitudinal mercaptopurine (MP) dose intensity and genotype at NUDT15 SNP rs116855232. MP dose intensity was determined monthly during 6-month study period of AALL03N1 for patients with (A) TT, (B) TC, and (C) CC genotypes at rs116855232. Numbers on x-axis denote month during which MP dose intensity was measured. Each box includes data between 25th and 75th percentiles, with horizontal line indicating median. Whiskers indicate maximal and minimal observations within 1.5× length of box.

Fig A6.

Mercaptopurine (MP) dose intensity and genotype at NUDT15 SNP rs116855232 and TPMT SNP rs1142345 in St Jude cohort. Of 371 children with acute lymphoblastic leukemia treated at St Jude Children's Research Hospital, cumulative MP dose intensity during entire continuation phase (maintenance) was lower in those with CT genotype at (A) rs116855232 (NUDT15) or (B) rs1142345 (TPMT) than in wild-type individuals. P value was estimated using linear mixed-effects regression model with ancestry and number of longitudinal MP dose-intensity measurements as covariates. Each box includes data between 25th and 75th percentiles, with horizontal line indicating median. Whiskers indicate maximal and minimal observations within 1.5× length of box.