Genome-wide study of methotrexate clearance replicates SLCO1B1

Abstract

Methotrexate clearance can influence the cure of and toxicity in children with acute lymphoblastic leukemia (ALL). We estimated methotrexate plasma clearance for 1279 patients with ALL treated with methotrexate (24-hour infusion of a 1 g/m2 dose or 4-hour infusion of a 2 g/m2 dose) on the Children’s Oncology Group P9904 and P9905 protocols. Methotrexate clearance was lower in older children (P = 7 x 10(-7)), girls (P = 2.7 x 10(-4)), and those who received a delayed-intensification phase (P = .0022). A genome-wide analysis showed that methotrexate clearance was associated with polymorphisms in the organic anion transporter gene SLCO1B1 (P = 2.1 x 10(-11)). This replicates findings using different schedules of high-dose methotrexate in St Jude ALL treatment protocols; a combined meta-analysis yields a P value of 5.7 x 10(-19) for the association of methotrexate clearance with SLCO1B1 SNP rs4149056. Validation of this variant with 5 different treatment regimens of methotrexate solidifies the robustness of this pharmacogenomic determinant of methotrexate clearance. This study is registered at http://www.clinicaltrials.gov as NCT00005585 and NCT00005596.

Figure 1

Flow chart of patients included in this analysis. A total of 1883 patients were treated with HDMTX on P9904 and P9905, and 1588 were eligible and had germline DNA evaluable. Patients who had congenital abnormalities (usually Down syndrome) were excluded because they may not have received the full methotrexate dose. Patients with implausible or no evaluable methotrexate plasma concentrations were excluded (see “Methods” for details).

Figure 2

Concentration versus time plots. (A) Twenty-four-hour methotrexate infusion of 1 g/m². Concentrations shown were scheduled to be drawn at 24 hours (end of infusion) and at 48 hours (data are jittered [horizontal offset added] for better visualization). The pink lines represent the cut-offs (based on the range of the St Jude pharmacokinetic simulation) outside of which samples were excluded. There is no lower threshold for exclusion (pink line) for postinfusion samples. (B) Plasma concentrations were estimated for St Jude Total XV patients on the basis of a simulated dose of 1 g/m² methotrexate over 24 hours (solid black line) and COG patients (dotted black line) receiving the same dose. The 1-99th percentile range is shown with shading for the St Jude Total XV patients (blue) and COG patients (red). (C) 4-hour methotrexate infusion of 2 g/m². Concentrations shown were scheduled to be drawn at 4 hours (end of infusion) and at 28 hours (data are jittered for better visualization). (D) Plasma concentrations were estimated for St Jude Total XV patients based on a simulated dose of 2 g/m² methotrexate over 4 hours (solid black line) and for COG patients (dotted black line) receiving the same dose. The 1-99th percentile range is shown with shading for the St Jude Total XV patients (blue) and COG patients (red).

Figure 3

Histogram of each patient's average methotrexate clearance. (A) Unadjusted methotrexate clearance for the 655 COG patients receiving the 4-hour infusion (blue), and the 624 receiving the 24-hour infusion (red). (B) Methotrexate clearance in all COG patients (n = 1279), adjusted for age, sex, race, and treatment arm (residual from a linear regression model).

Figure 4

Manhattan plot of P values for a genome-wide association with methotrexate clearance. Each chromosome is plotted along the x-axis, whereas the y-axis plots the negative log₁₀ of the P value for the association of each SNP with methotrexate clearance. Higher points on the graph indicate a stronger association with methotrexate clearance. Typed and imputed SNPs are shown in all panels (5.2 million SNPs). Methotrexate clearance is adjusted for age, sex, race, and treatment arm. (A) Association of each SNP with methotrexate clearance in COG patients (n = 1279). (B) Meta-analysis of St Jude (n = 699) and COG patients (n = 1279). (C) Meta-analysis of St Jude and COG patients, after we adjusted for rs4149056.

Figure 5

LD of SNPs with P < 10⁻⁵ in meta-analysis spanning SLCO1B3, LST-3TM12, SLCO1B1, and SLCO1A2 generated with the use of Haploview software in European-ancestry patients only (COG n = 806 and SJ n = 459). The darkness of each box represents the association between 2 SNPs (R²), with white being R² = 0, black being R² = 1, and 0 < R² < 1 being shades of gray. LD blocks, highlighted with the yellow lines, were defined by confidence intervals. Assembly hg18 from http://genome.ucsc.edu used for display.