Old drug, new lessons

In this issue of Blood, French and colleagues use several genome-wide approaches and report that acquired genetic variation has a stronger impact on methotrexate polyglutamate accumulation in acute lymphoblastic leukemia cells than inherited genetic variation.

This paper by French et al is among the first to combine high-throughput analyses of malignant cells for acquired genetic variation (mRNA expression and DNA copy number variation) and of normal cells to detect inherited genetic variation (DNA single nucleotide polymorphisms and DNA copy number variation) in the same patients.¹  A relatively large cohort of children with acute lymphoblastic leukemia (ALL) was tested, and methotrexate polyglutamate (MTXPG) accumulation was determined ex vivo in samples obtained 42 to 44 hours after the patients had been treated with single-agent MTX at 1 g/m² given intravenously over 4 or 24 hours. Acquired genetic variation was assessed in leukemic cells obtained at diagnosis while inherited genetic variation was studied in DNA extracted from whole blood sampled after the patients achieved complete remission.

This study is important for several reasons. First, it demonstrates the importance of characterizing malignant cells themselves in order to predict or explain their sensitivity to a particular drug. Apparently, the biology of malignant cells is more important in this respect than inherited factors, which, for instance, influence pharmacokinetics. Second, it identifies novel genes, especially on chromosomes 10 and 18, which seem important in explaining variation in MTXPG accumulation. Third, the paper contains a wealth of information on the relevance of individual chromosomes and genes regarding MTX accumulation. Finally, it shows that combining information on acquired and inherited genetic variation can be useful, since this analysis identified 7 genes that had the strongest impact on MTXPG accumulation.

Inevitably, the study also has some weaknesses, and several questions remain to be answered. While 248 patients were eligible for the study, actual characterization was limited to 145 patients for mRNA expression, 82 patients for leukemia cell DNA copy number variation, and 144 patients for inherited DNA genotyping. The authors demonstrate that this did not result in a statistically significant selection bias, but some impact of the subset of patients available for each assay type cannot be excluded. Moreover, patient numbers limited the power of the study to detect smaller but still relevant correlations between genetic variation and MTXPG levels. An open question is whether the genes that explained variation in MTXPG accumulation have a causal role in determining the clinical response to MTX treatment. After all, this was a correlative study and moreover, MTXPG levels are a surrogate for sensitivity or resistance to MTX, although other studies demonstrated a correlation between MTXPG levels and both in vitro²  and in vivo³  efficacy of MTX. It also would be interesting to know what the study results would have been in case of a higher dose of MTX (eg, 5 g/m²), a dose now being used in many protocols. French et al indeed report that MTXPG levels differed between the 4- and 24-hour infusion schedules, and they cite literature that reported that gain-of-chromosome 21 was associated with increased MTXPG accumulation, but only in the case of treatment with MTX at 180 mg/m² given orally over 36 hours and not in the case of 1 g/m² given as a 24-hour infusion.⁴  Similarly, the findings explain up to two-thirds of the variation in MTXPG levels, but what about the remaining one-third? Finally, the authors do not provide data on toxicity of MTX. The current analysis might suggest that characterization of inherited genetic variation is of limited value. However, such variation is more likely to explain side effects of specific treatment elements and is important as well. It was beyond the scope of this study and will be extremely challenging, but one wonders if similar findings will be made if minimal residual leukemic cells are studied.

Meanwhile, this study encourages similar efforts focusing on other drugs. Given the large interindividual variation in drug sensitivity, the complicated biological processes that determine their antileukemic activity and toxicity, and their importance in the treatment of childhood acute leukemias, obvious candidates are cytarabine, l-asparaginase, and glucocorticoids.^5-7  The understanding of the efficacy of other drugs and in a wide range of cancers will also benefit from this type of integrated research. In the longer term, this approach is not feasible for routine use in multicenter clinical studies. Therefore, customized chips should be developed, which likely need to be disease- and treatment-specific. Ultimately, this type of research will enable more individualized and tailored chemotherapy of cancer, aiming at more effective and less toxic therapy.