Rheumatology Advance Access originally published online on April 9, 2009
Rheumatology 2009 48(6):597-598; doi:10.1093/rheumatology/kep060
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
EDITORIALS |
Methotrexate pharmacogenomics in rheumatoid arthritis: introducing false-positive report probability
1Cypress Bioscience, San Diego, CA, USA
Correspondence to: Thierry Dervieux, Cypress Bioscience, 9393 Towne Center Drive, San Diego, CA 92121, USA. E-mail: tdervieux{at}cypressbio.com
In this issue of the Journal, Lee and associates [1] investigated the contribution of candidate polymorphisms in folate and purine pathways to disease activity in RA patients treated with low-dose MTX therapy. This cross-sectional study analysis in patients enrolled in the BRASS (Brigham and Women's Hospital Rheumatoid Arthritis Sequential Study) cohort sought to replicate the earlier findings from our group [2–4] as well as those from Leiden University [5,6] among others. The results clearly illustrate the challenges and difficulties we face when validating associations between low-penetrance genetic polymorphisms and complex phenotypes such as drug response. In particular, the authors establish that the minor allele of rs4673993 in 5-aminoimidazole-4-carboxamide ribonucleotide (AICAR) transformylase (ATIC, in proxy with a C347G non-synonymous SNP encoding a threonine-to-serine substitution at position 116) is associated with improved clinical status, a finding consistent with the results observed in our US cohort [3, 4], but in complete contradiction with the data reported by Wessels and colleagues [5] in the European Best cohort.
Recognizing that the finding was potentially false positive, the authors introduced and applied for the first time the concept of false-positive report probability (FPRP) to the field of MTX pharmacogenomics. FPRP is a recent method devised at the National Cancer Institute by Wacholder and colleagues [7] as an alternative to other methods, such as Bonferroni's or false-positive discovery rate. In essence, the method determines the likelihood that a significant finding is attributable to chance (false positive). The method does this by integrating the P-value observed (type I error) together with the statistical power [1-type II error to detect a given odds ratio (OR)] and a prior probability set forth by the investigator on the basis of the knowledge or credibility of the association between the SNP and the outcome. For example, the authors established that the FPRP to detect an OR of 2.5 (increased efficacy conferred by the minor allele of rs4673993) with a prior probability of 0.1 at the established P-level of 0.01 was 0.33. As such, a 33% chance that the present finding is false positive is below a preset 0.5 cut-off and the finding can be considered noteworthy. Arguably, the likelihood of false-positive association may appear relatively high but FPRP results should be interpreted with respect of a recent estimate that 95% of studies reporting significant associations between gene polymorphisms and complex outcomes may be indeed false positives [8]. Hence, the determination of the FPRP can be extremely helpful to reviewers, editors and readers to weigh the ‘noteworthiness’ of a publication and differentiate false positive from true positive associations especially when findings are contradictory or equivocal.
However, some caution should be exercised while interpreting the FPRP as the estimate of the prior probability can dramatically affect the interpretation of the findings. In fact, sensitivity analysis revealed that the FPRP would inflate from 0.33 to 0.85 by decreasing the prior probability from a high (0.1) to a moderate (0.01) level. Nonetheless, a high likelihood that non-synonymous genetic polymorphisms in ATIC is associated with MTX effects could be justified. First, the data have established that MTX is a potent inhibitor of de novo purine biosynthesis at the level of ATIC, and each addition of one glutamic residue on MTX produces a 10-fold increase in the inhibitory potency of ATIC (using 10-formyltetrahydrofolate as the co-substrate) [9]. Secondly, a large body of evidence suggests that ATIC inhibition produces an elevation of AICAR ribonucleic acid levels [10–12] and the build-up of AMP (through competitive inhibition of adenosine deaminase) [13] resulting in anti-inflammatory effects through extracellular conversion to adenosine by an ectonucleotidase-mediated process [14]. Thirdly, serine-to-threonine substitutions have been associated with differential effects on phosphorylation of cyclin-dependant kinases and ATIC phosphorylation enhances resistance to MTX [15,16]. Altogether, these biochemical data largely support the notion that alteration in ATIC expression at the transcriptional or post-transcriptional level may be central to MTX effects, and this contention is also supported in vivo, in RA [3–5]. Thus, we could agree with the authors that the choice of the prior probability was adequate. On the other hand, others may disagree given the lack of in vitro or in vivo models substantiating the true relevance of the polymorphisms to MTX effects and may use a more conservative moderate prior probability (i.e. 0.01).
The interpretations of Lee and associates might be taken a step further by retrospectively determining the FPRPs in the two other studies reported by Dervieux et al. [3], Wessels et al. [5] and compare them to those reported here (Table 1). In our cohort of patients [3], the ATIC 347G allele was also significantly associated with higher efficacy [disease activity score (DAS) < 3.2; OR = 1.81; 95% CI 1.02, 3.22; P = 0.046] and the FPRP to detect a conservative OR of 2.0 with a prior probability of 0.1 was 0.38 vs 0.34 in the present study (using unadjusted OR). Conversely, in the Best cohort, the ATIC 347G allele was associated with lower efficacy (OR = 0.45; 95% CI 0.26, 0.79; P = 0.007) and the FPRP to detect an OR of 0.5 with a prior probability of 0.1 was 0.12 (Table 1). Thus, taken independently all three studies are noteworthy in their own right (FPRP < 0.5), although it remains unclear why contradictory findings were observed between the two US cohorts and the European cohort. In fact, applying a moderate to low prior probability to FPRP calculation turns cautious optimism to scepticism and reveals that all associations are indeed false.
|
As stated by the authors, the difference between the results could be explained by false positives in any or all of the studies. The authors suggest that demographic dissimilarities between cohorts, including positivity for RFs, ethnicities and duration of disease might have explained the discrepancies observed. However, the consistency of the findings between the BRASS cohort and ours may not be explained by the percentage of RF-positive patients, which was 69.0% in our population, a value actually closer to that reported in the Best cohort (67.3%) vs the BRASS cohort (79.2%). Also, the notion that racial status contributed is also unlikely because >95% patients enrolled in our study were of Caucasian descent vs 93% in the Best cohort (the analysis was restricted to Caucasians in BRASS).
It remains that disease duration was much longer in patients enrolled in the US cohorts (mean of 18 and 11 years, respectively) compared with those enrolled in the European cohort (mean 2 weeks), and patients had been treated with MTX for an extended period of time in both US cohorts (mean 74 and 54 months) vs only 6 months in the Best cohort. Assuming that the reports have value in their own right, it is tempting to suggest that the contribution of ATIC C347G to MTX effects may differ as a function of treatment duration. ATIC is a member of the one carbon folate metabolism, a highly conserved intricate network of over 20 genes tightly regulated by folate homeostasis and data indicate a differing contribution of genetic variants in folate enzymes as a function of folate administration [17]. For example, both clinical data and mathematical models have established that the effect of MTHFR C677T on homocysteine concentrations is dependent on folate pools, and higher folate concentrations overcome the elevation in homocysteine levels observed in patients carrying the MTHFR 677TT genotype [18]. In addition, folate pools mediate substantial alteration in ATIC velocities [18], and the changes in the concentration of folate polyglutamates occurring at the initiation of MTX therapy [2] or during treatment (as a function of supplements to prevent side effects) may mask the effect of the polymorphism, or alternatively reveal its dominant effect. Moreover, the effect of dietary folate supplementation which is more pronounced in the USA, where fortification of bread and cereals is now routine, compared with European countries, where fortification does not routinely occur, may also contribute to the discrepancies observed [19]. It should also be emphasized that ATIC deficiency is associated with AICAR ribosiduria, a highly penetrant devastating neurological disorder [20], and it is extremely unlikely that any common ATIC C347G polymorphism has dramatic effects on the protein functions. More likely, are minor subtle differences that can together with other as yet unidentified non linear gene–gene interactions produce a phenotype involving the interplay of multiple other variants ‘superimposed’ to other complex metabolic interactions. As a matter of fact, MTX exposure and the concentration of MTXPG achieved together with folate polyglutamate levels contribute to the efficacy of the drug [2,3], and thus confound the qualitative and quantitative role of constitutive genetic background to the multi-factorial phenotype of drug response.
In conclusion, our growing field of MTX pharmacogenetics may benefit from the concept of FPRP introduced here. Whether the contribution of the ATIC C347G to MTX effects is real or spurious is not known, but cautious optimism should stem our enthusiasm to conduct additional studies to bring the field of MTX pharmacogenetics to clinical reality.
Disclosure statement: T.D. is employed by Cypress Bioscience and holds stock options in the company. He is the inventor of patent related to ‘Methods for optimizing clinical responsiveness to methotrexate therapy using metabolite profiling and pharmacogenetics’.
References
- Lee YC, Cui J, Costenbader KH, Shadick NA, Weinblatt ME, Karlson EW. Investigation of candidate polymorphisms and disease activity in rheumatoid arthritis patients on methotrexate. Rheumatology (2009) 48:613–7.
[Abstract/Free Full Text] - Dervieux T, Greenstein N, Kremer J. Pharmacogenomic and metabolic biomarkers in the folate pathway and their association with methotrexate effects during dosage escalation in rheumatoid arthritis. Arthritis Rheum (2006) 54:3095–103.[CrossRef][Web of Science][Medline]
- Dervieux T, Furst D, Lein DO, et al. Pharmacogenetic and metabolite measurements are associated with clinical status in patients with rheumatoid arthritis treated with methotrexate: results of a multicentred cross sectional observational study. Ann Rheum Dis (2005) 64:1180–5.
[Abstract/Free Full Text] - Dervieux T, Furst D, Lein DO, et al. Polyglutamation of methotrexate with common polymorphisms in reduced folate carrier, aminoimidazole carboxamide ribonucleotide transformylase, and thymidylate synthase are associated with methotrexate effects in rheumatoid arthritis. Arthritis Rheum (2004) 50:2766–74.[CrossRef][Web of Science][Medline]
- Wessels JA, Kooloos WM, de JR, et al. Relationship between genetic variants in the adenosine pathway and outcome of methotrexate treatment in patients with recent-onset rheumatoid arthritis. Arthritis Rheum (2006) 54:2830–9.[CrossRef][Web of Science][Medline]
- Wessels JA, de Vries-Bouwstra JK, Heijmans BT, et al. Efficacy and toxicity of methotrexate in early rheumatoid arthritis are associated with single-nucleotide polymorphisms in genes coding for folate pathway enzymes. Arthritis Rheum (2006) 54:1087–95.[CrossRef][Web of Science][Medline]
- Wacholder S, Chanock S, Garcia-Closas M, El Ghormli L, Rothman N. Assessing the probability that a positive report is false: an approach for molecular epidemiology studies. J Natl Cancer Inst (2004) 96:434–42.
[Abstract/Free Full Text] - Colhoun HM, McKeigue PM, Smith GD. Problems of reporting genetic associations with complex outcomes. Lancet (2003) 361:865–72.[CrossRef][Web of Science][Medline]
- Allegra CJ, Drake JC, Jolivet J, Chabner BA. Inhibition of phosphoribosyl aminoimidazolecarboxamide transformylase by methotrexate and dihydrofolic acid polyglutamates. Proc Natl Acad Sci USA (1985) 82:4881–5.
[Abstract/Free Full Text] - Allegra CJ, Hoang K, Yeh GC, Drake JC, Baram J. Evidence for direct inhibition of de novo purine synthesis in human MCF-7 breast cells as a principal mode of metabolic inhibition by methotrexate. J Biol Chem (1987) 262:13520–6.
[Abstract/Free Full Text] - Cronstein BN, Naime D, Ostad E. The antiinflammatory mechanism of methotrexate. Increased adenosine release at inflamed sites diminishes leukocyte accumulation in an in vivo model of inflammation. J Clin Invest (1993) 92:2675–82.[Web of Science][Medline]
- Morgan SL, Oster RA, Lee JY, Alarcon GS, Baggott JE. The effect of folic acid and folinic acid supplements on purine metabolism in methotrexate-treated rheumatoid arthritis. Arthritis Rheum (2004) 50:3104–11.[CrossRef][Web of Science][Medline]
- Baggott JE, Vaughn WH, Hudson BB. Inhibition of 5-aminoimidazole-4-carboxamide ribotide transformylase, adenosine deaminase and 5’-adenylate deaminase by polyglutamates of methotrexate and oxidized folates and by 5-aminoimidazole-4- carboxamide riboside and ribotide. Biochem J (1986) 236:193–200.[Web of Science][Medline]
- Montesinos MC, Takedachi M, Thompson LF, Wilder TF, Fernandez P, Cronstein BN. The antiinflammatory mechanism of methotrexate depends on extracellular conversion of adenine nucleotides to adenosine by ecto-5’-nucleotidase: findings in a study of ecto-5’-nucleotidase gene-deficient mice. Arthritis Rheum (2007) 56:1440–5.[CrossRef][Web of Science][Medline]
- Kaldis P, Cheng A, Solomon MJ. The effects of changing the site of activating phosphorylation in CDK2 from threonine to serine. J Biol Chem (2000) 275:32578–84.
[Abstract/Free Full Text] - Boccalatte FE, Voena C, Riganti C, et al. The enzymatic activity of 5-aminoimidazole-4-carboxamide ribonucleotide formyltransferase/IMP cyclohydrolase (ATIC) is enhanced by NPM-ALK: new insights in ALK-mediated pathogenesis and the treatment of ALCL. Blood. Advance Access published Oct 9, 2008, doi:113: 2776–2790.
- van Ede AE, Laan RF, Blom HJ, et al. The C677T mutation in the methylenetetrahydrofolate reductase gene: a genetic risk factor for methotrexate-related elevation of liver enzymes in rheumatoid arthritis patients. Arthritis Rheum (2001) 44:2525–30.[CrossRef][Web of Science][Medline]
- Reed MC, Nijhout HF, Neuhouser ML, et al. A mathematical model gives insights into nutritional and genetic aspects of folate-mediated one-carbon metabolism. J Nutr (2006) 136:2653–61.
[Abstract/Free Full Text] - Quinlivan EP, Gregory JF III. Effect of food fortification on folic acid intake in the United States. Am J Clin Nutr (2003) 77:221–5.
[Abstract/Free Full Text] - Marie S, Heron B, Bitoun P, Timmerman T, Van Den Berghe G, Vincent MF. AICA-ribosiduria: a novel, neurologically devastating inborn error of purine biosynthesis caused by mutation of ATIC. Am J Hum Genet (2004) 74:1276–81.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  T. Dervieux Comment on: Methotrexate pharmacogenomics in rheumatoid arthritis: introducing false positive report probability: reply Rheumatology, December 1, 2009; 48(12): 1620 - 1620. [Full Text] [PDF]
|
|
|
|
|
|
L. Beretta Comment on: Methotrexate pharmacogenomics in rheumatoid arthritis: introducing false positive report probability Rheumatology, December 1, 2009; 48(12): 1619 - 1620. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||