6-mercaptopurine/methotrexate (6MP/MTX) maintenance therapy is essential for high cure rates of childhood acute lymphoblastic leukemia (ALL). 6MP exerts its cytotoxicity through conversion into thioguanine nucleotides (TGN) that are substrates for DNA polymerase and compete with normal guanine for DNA incorporation. DNA-TGN occasionally mismatches to thymidine causing cell death due to futile attempts of repetitive mismatch repair. We recently showed that higher levels of DNA-incorporated TGN were associated with a reduced relapse risk in non-high risk ALL (Nielsen, Lancet Oncol 2017). Common germline variants in TPMT that codes for thiopurine methyltransferase (TPMT) lead to reduced TPMT activity and higher levels of cytosol TGN when measured in erythrocytes. Thus, low TPMT activity could theoretically lead to higher DNA-TGN and a lower relapse rate. Low TPMT activity (phenotype) has been associated with a lower relapse risk in the NOPHO ALL92 protocol (Schmiegelow, Leukemia 2009). We explored if TPMT status evaluated by either geno- or phenotype was related to DNA-TGN levels and relapse rate in the NOPHO ALL2008 protocol.


TPMTgenotype and repeated phenotyping (2141 measurements in total), and levels of DNA-TGN (10830 measurements in total) during maintenance therapy were determined for 918 children treated for non-high risk ALL (89% of all eligible patients). On a subset (N=204) TPMT phenotype was available both at diagnosis and during maintenance therapy. 6MP was started at 50 and 75 mg/m2 for TPMT heterozygous and wildtype patients, respectively, and subsequently adjusted to a target WBC of 1.5-3.0 x109/L. Cox analyses were adjusted by age, sex and WBC at diagnosis.


78 of the 918 non-high risk patients were TPMT heterozygous (none homozygous deficient). TPMT activities were significantly higher during maintenance therapy than at diagnosis for both TPMT wildtype and heterozygous children (N=204; wildtype: 16.6 vs. 11.3 U/mL ery., p <0.001; heterozygous: 9.9 vs. 6.8 U/mL ery., p=0.047). Mean DNA-TGN levels were higher for TPMT heterozygous than wildtype children (N=523, children in remission at end of therapy with Maintenance II measurements, median 760.9 fmol/µg, range 125.7−1255.0 vs. 494.9, range 44.0−1559.0, p <0.001; Figure). The 5-year cumulative incidence of relapse was 4.2% (95% CI: 0-8.7%) for TPMT heterozygous and 5.7% (95% CI: 3.8-7.5%) forwildtype children (hazard ratio =0.87, 95% CI: 0.27−2.83, p=0.81). No association between mean TPMT activity during maintenance therapy (as a time-dependent covariate) and relapse-specific hazard was seen (N=812; hazard ratio =0.98 per one unit increase in TPMT activity (U/mL ery.), 95% CI: 0.89−1.08, p=0.63). In patients with positive MRD at end of induction (d29) results were similar (N=476; hazard ratio =1.04 per one unit increase in TPMT activity (U/mL ery.), 95% CI: 0.93−1.16, p=0.50).


Although mean DNA-TGN levels were higher in TPMT heterozygous than wildtype children, variability was high in both groups and DNA-TGN levels were largely overlapping, and no association with relapse hazard was seen for neither TPMT geno- nor phenotype. In accordance with previously published data TPMT phenotype determinations done in erythrocytes present at ALL diagnosis can lead to TPMT misclassification.


a) Mean DNA-TGN levels (N=523) during Maintenance II of children in clinical remission with completed therapy and at least one measurement.

b) Cumulative incidence curves of relapse of children carrying TPMT wildtype vs. heterozygous.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.