Table 4.

Multivariate regression model of the 3 strongest determinants. Variables that after correction for confounders had a significant correlation in both cohorts were included in 1 multivariate regression model. FPGS rs4451422 was dichotomized into wild type versus heterozygous/variant. Correlations with p < 0.05 are considered significant. Confounders are not shown and were serum folate (for erythrocyte folate) and sex (for FPGS rs4451422).

MTX-PG1MTX-PG2MTX-PG3MTX-PG4+5MTX-PG2–5
stβpstβpstβpstβpstβp
Derivation cohort
  Age0.210.060.20.090.240.030.20.080.280.01
  Erythrocyte folate–0.190.130.040.750.210.080.180.150.190.1
  FPGS rs4451422 WT vs het/var–0.080.47–0.170.13–0.36< 0.001–0.330.003–0.38< 0.001
  R20.140.040.210.110.21
Validation cohort
  Age0.130.080.220.0020.230.0010.220.0020.28< 0.001
Erythrocyte folate0.050.530.030.660.120.090.10.160.110.099
  FPGS rs4451422 WT vs het/var0.010.84–0.070.3–0.080.25–0.090.18–0.100.12
  R200.030.070.070.1
Combined cohort
  Dose–0.070.26–0.22< 0.0010.23< 0.0010.3< 0.0010.2< 0.001
  Age0.150.0130.20.0010.23< 0.0010.21< 0.0010.27< 0.001
  Erythrocyte folate0.030.650.030.660.150.010.110.060.130.024
  FPGS rs4451422 WT vs het/var–0.010.8–0.090.106–0.160.003–0.150.005–0.18< 0.001
  R20.010.080.150.150.16
  • FPGS: folylpolyglutamate synthetase; MTX: methotrexate; PG: polyglutamate; WT: wild type; het: heterozygous; var: homozygous variant.