As with any diagnostic technology, there are 3 steps in the technology assessment process: evaluation of technical feasibility, evaluation of ability to accurately diagnose a clinical condition in comparison with the gold standard, and determination of whether use of the test results in an improved patient outcome. These factors are discussed below, both for pharmacogenomics and metabolite markers.
The genotypic analysis of the thiopurine methyltransferase (TPMT) gene is based on well-established polymerase chain reaction technology (PCR) to detect 3 distinct mutations. Currently, 3 alleles, TPMT*2, TPMT*3A and TPMT*3C, account for about 95% of individuals with reduced TPMT activity. Individuals homozygous for these alleles are TPMT-deficient and those heterozygous for these alleles have variable TPMT (low or intermediate) activity. A 2011 study from Sweden addressed the concordance between TPMT genotyping and phenotyping. (1) The investigators evaluated data from 7,195 unselected and consecutive TPMT genotype and phenotype tests. The genotype tests examined the 3 most common TPMT variants, noted above. TPMT genotyping identified 89% as TPMT wild type, 704 (10%) as TPMT heterozygous, and 37 (0.5%) as TPMT homozygous. The overall agreement between genotyping and phenotyping was 95%. Genotyping alone would have misclassified 3 of 37 (8%) homozygous patients as heterozygous; these 3 individuals were found to have uncommon mutations. All 3 had low TPMT activity. The phenotype test would have misclassified 4 of 37 (11%) of homozygous patients as they had test results above the cut-off level for low TPMT activity (<2.5 U/mL red blood cells).
Metabolite markers have been assessed using high performance liquid chromatography technology. It would be optimal to assess metabolite markers in peripheral leukocytes, since they reflect the status of bone marrow precursors. However, it is technically easier to measure metabolites in red blood cells (RBCs) instead of leukocytes.
Several systematic reviews of studies on the diagnostic performance of TPMT genotyping have been published. Among the most recent studies was a 2011 review by Booth and colleagues that was sponsored by the Agency for Healthcare Research and Quality (AHRQ). (2) A total of 19 studies on test performance were identified; most were cross-sectional or prospective observational studies and approximately 70% included patients with inflammatory bowel disease. Among the 1,735 total patients, 184 were heterozygous and 16 were homozygous for variant alleles, a small total sample of individuals with variant alleles. A pooled analysis of data from 19 studies found a sensitivity of 79.9% (95% confidence interval [CI]: 74.8% to 84.6%) for correctly identifying individuals with subnormal (intermediate or low) enzymatic activity. The specificity of the wild-type genotype for correctly identifying individuals with normal or high enzymatic activity approached 100%. Seventeen studies addressed the association between TPMT status and thiopurine toxicity. The studies included a total of 2,211 patients, of which 357 had intermediate and 74 had low enzymatic activity. In a pooled analysis of 3 studies (92 patients, 10 events), there were greater odds of myelotoxicity with low TPMT enzymatic activity than intermediate activity (pooled odds ratio [OR]: 14.5, 95% CI: 2.78-76.0). Similarly, in a pooled analysis of 3 studies (403 patients, 29 events), there were greater odds of myelotoxicity with low TPMT enzymatic activity than normal levels (pooled OR: 19.1, 95% CI: 4.6-80.2). It is worth noting that confidence intervals were wide due to few events and small sample sizes.
Another systematic review published in 2011, by Donnan and colleagues, identified 17 studies that reported the performance characteristics of TPMT genotyping tests (12 studies) and phenotyping (6 studies) compared to a reference standard. (3) No true gold standard was available. The enzymatic test was used as the reference standard in 9 studies, and the remainder used a genotyping test; 3 studies compared 2 methods of genotyping. All of the studies used a method of genotyping as either the investigational test or the reference standard; the tests varied somewhat in the number and type of polymorphisms they were designed to detect. Sixteen of 17 studies either reported sensitivity and specificity, or reported sufficient data for these measures to be calculated. Only 3 studies considered confounding factors such as concurrent medications and blood transfusions in their exclusion criteria. The authors of the systematic review did not pool study findings. In the included studies, sensitivity of enzymatic tests ranged from 92% to 100% and specificity ranged from 86% to 98%. The sensitivity of the genotype tests ranged from 55% to 100% and the specificity ranged from 94% to 100%. In general, the enzymatic tests had a high sensitivity and low-positive predictive value when genotype tests were used as the reference standard. Genotype tests showed a lower sensitivity and high positive-predictive value when enzymatic tests were used as the gold standard. The inconsistent use of a reference standard complicated the interpretation of the findings.
A 2010 meta-analysis by Dong and colleagues evaluated the relationship between TPMT polymorphisms and adverse-drug reactions in patients with inflammatory bowel disease (IBD) taking thiopurine drugs. (4) The review included cross-sectional studies, prospective cohort studies, and case-control studies conducted in patients 18 years and older with IBD. In addition, studies needed to compare TPMT polymorphism frequencies in thiopurine-tolerant and thiopurine-intolerant patients and report at least one of several outcome measures related to adverse drug reactions. The investigators identified a total of 181 publications, 17 of which were considered potentially appropriate after applying initial eligibility criteria e.g., publication type and study population. Another 8 studies were excluded because relevant outcome variables were not reported or could not be calculated. The final sample consisted of 9 studies with 1,309 participants. In a pooled analysis of data from 6 studies, 39 of 273 (14%) patients with an adverse drug reaction were TPMT heterozygous or homozygous compared to 39 of 708 (55%) patients without an adverse drug reaction. This difference was statistically significant (OR: 2.93, 95% CI: 1.68-5.09). In analyses of specific adverse reactions, there was a statistically significant association between the presence of TPMT alleles and bone marrow toxicity, but not hepatotoxicity or pancreatitis. In the non-statistically significant analyses, the number of events was small and the analyses have been underpowered. For example, 3 of 37 (8%) IBD patients with hepatotoxicity were TPMT heterozygous/ homozygous compared to 82 of 1,017 (81%) patients without hepatotoxicity (OR: 1.51, 95% CI: 0.54 to 4.19).
Studies have focused on assessing the correlation between metabolite levels and disease remission or adverse drug effects; study designs have generally been cross-sectional or retrospective. For example, Cuffari and colleagues measured the metabolites 6-thioguanine nucleotides (6-TGN) and 6-methyl-mercaptopurine ribonucleotides (6-MMRP) in 25 pediatric patients with Crohn’s disease. (5) Achievement of clinical remission was correlated with 6-thioguanine nucleotides (6-TGN), levels, but not 6-methyl-mercaptopurine ribonucleotides (6-MMRP). In 1 patient, low 6-thioguanine nucleotides (6-TGN) levels suggested noncompliance, which was subsequently confirmed on further questioning. Dubinsky and colleagues measured 6-thioguanine nucleotides (6-TGN), and 6-methyl-mercaptopurine ribonucleotides (6-MMRP) levels in 92 pediatric patients. (6) Higher median levels of 6-thioguanine nucleotides (6-TGN) were observed at points of clinical response versus non-response. Quartile analysis on all samples revealed that the best probability of treatment occurred when 6-thioguanine nucleotides (6-TGN) levels were greater than 235 (measured in pmol/8 x 10 to the eighth). The 6-methyl-mercaptopurine ribonucleotides (6-MMRP) levels did not correlate with disease activity, but elevated levels did correlate with hepatotoxicity, observed in 16 patients. Elevated 6-thioguanine nucleotides (6-TGN) levels were also associated with hematologic toxicity. In contrast, Gupta and colleagues reported discordant results in a case series of 54 patients with IBD being treated with azathioprine. (7) A total of 36% of patients in relapse had 6-thioguanine nucleotides (6-TGN) levels greater than 230, compared with 30% of those in remission. Conversely, 57% of patients with 6-thioguanine nucleotides (6-TGN) levels less than 230 were in remission, versus 50% of patients with 6-thioguanine nucleotides (6-TGN) levels greater than 230. These results question the 235 cut-off point suggested by Dubinsky et al.
A 2010 cross-sectional study by Waljee et al. compared the ability of metabolite tests and a clinical algorithm using routine laboratory values (CBC and chemistry panel) and patient age to predict the clinical response of patients with IBD to thiopurines. (8) The study included 346 patients taking thiopurines who underwent thiopurine metabolite analysis, a CBC, and a comprehensive chemistry analysis in one 24-hour period. Clinical response was determined through medical record review. Data on clinical response were available for 240 patients; 119 (49.6%) were classified as responders. Using area under the curve (AUC) analysis, an algorithm using patient age and laboratory values differentiated clinical responders from nonresponders with an AUC of 0.86. In comparison, the metabolite marker 6-thioguanine nucleotides (6-TGN) had an AUC of 0.59 for differentiating between clinical response and nonresponse. The difference between these two areas was statistically significant, p<0.001. The variables with the strongest independent associations with clinical response were neutrophil count, alkaline phosphatase, red cell distribution width, age, and white blood cell count. When 6-TGN levels were added as an independent variable to a model containing the above variables, it did not significantly improve the AUC (which increased from 0.856 to 0.862). The authors noted that a limitation of the study was that data were obtained from a tertiary care center, and it is possible that patients in whom metabolites performed poorly were likely to be referred to that center. The authors concluded that the algorithm they developed and tested using routine laboratory tests for differentiating thiopurine clinical responders and nonresponders performed significantly better than metabolite monitoring in predicting clinical response.
Improvement in Health Outcomes
The use of pharmacogenomics and thioprine metabolite testing creates the possibility of tailoring a drug regimen for each individual patient, with the ultimate goal of attaining disease remission and elimination of steroid therapy. The preferred study design would compare patient management (e.g., drug choice) and health outcomes in patients managed with and without testing.
A randomized controlled study (RCT), known as the TARGET study, randomized 333 patients to receive TPMT genotyping or usual care (no genotyping) prior to azathioprine therapy. (9) Study eligibility included age 16 years or older with a diagnosis of inflammatory bowel disease (IBD). In the testing arm, test results were generated within 1 week, and the study clinician was informed of the results. Clinicians were advised to recommend the following: maintenance dose of azathioprine (i.e., 1.5 to 3 mg/kg/day) for individuals with wild-type TPMT, low-dose azathioprine (i.e., 25-50 mg/day) titrated to a maintenance dose for individuals with heterozygous TPMT variant alleles and an alternative therapy (no azathioprine) for patients homozygous for TPMT variant alleles. All final treatment decisions were at the discretion of the individual provider (i.e., this was a pragmatic RCT). Genotyping was also done on samples from patients in the control group, but results were not made available until the end of the study.
Data were available for 322/333 (97%) patients at 4 months. The primary study endpoint was stopping azathioprine due to any adverse drug reaction in the first 4 months of treatment. At 4 months, a total of 91 of 322 (28%) patients had stopped taking azathioprine due to an adverse drug reaction, 47 of 163 (29%) in the genotyping group and 44 of 159 (28%) in the non-genotyping group. The difference between groups was not statistically significant, p=0.74. In the genotyping arm, the average starting dose of azathioprine was significantly lower in TPMT heterozygotes than wild-type individuals (p=0.008), suggesting that clinicians were following dosing recommendations. However, at 4 months, the mean dose was similar across both arms (1.68 mg/kg/day, p=0.25), and there was no difference in dose between individuals heterozygous or wild type for TPMT variant alleles (p=0.99). Moreover, at 4 months, there was not a significant difference between groups in the level of clinical symptoms between groups. The mean Harvey-Bradshaw symptom index score was 4.5 in each group, p=0.80 (54 patients in the genotyping group and 56 patients in the non-genotyping group were included in this analysis). It is important to note that, in this study, few individuals had non-wild type gene variants. In the genotyping group, there were 7 heterozygous patients and in the non-genotyping group, there were 2 heterozygous patients and 1 homozygous patient. Thus, the study was underpowered to evaluate the impact of TPMT genotyping on patients with variant alleles.
Several uncontrolled prospective studies have also been published. For example, in a study that involved 131 patients with IBD, investigators from Europe did not find that the choice of azathioprine/ mercaptopurine (6-MP) dose based on RBC TPMT activity prevented myelotoxicity; no patients in this study exhibited low activity. (10) In a 2008 study from New Zealand, Gardiner et al. noted that initial target doses to attain therapeutic levels in patients with IBD might be 1 mg/kg/d and 3 mg/kg/d in intermediate (heterozygous) and normal (wild-type) metabolizers. (11) This conclusion was based on a study of 52 patients with IBD who were started on azathioprine or mercaptopurine and who were followed up for 9 months while 6-thioguanine nucleotides (6-TGN) levels and clinical status were assessed. This study suggested that knowledge of TPMT activity can assist with initial dosing. In a study from Europe including 394 patients with inflammatory bowel disease, Gisbert et al. noted that the probability of myelotoxicity was 14.3% in the TPMT intermediate group compared to 3.5% in those with high (wild-type) activity. (12) These authors concluded that determining TPMT activity prior to initiating treatment with azathioprine could help to minimize the risk of myelotoxicity.
A 2007 review article by Teml et al. noted that patients with TPMT deficiency treated with standard doses of azathioprine/ mercaptopurine (6-MP) have double the risk of developing severe myelosuppression within a few weeks of starting therapy. (13) As noted here, this risk is explained by a marked increase of 6-thioguanine nucleotides (6-TGN) levels. The review also noted that the data are more limited in those who are heterozygous for TPMT activity However, for patients who developed myelotoxicity, the reviewers noted that it tended to occur earlier (1 to 1.5 months into treatment) for those who were heterozygous, compared to about 3 months in wild-type patients.
No prospective comparative trials were identified in which use of metabolite markers was compared to current approaches to care. Studies continue to report on aspects of this testing. For example, recent studies have reported that there is only weak correlation between metabolite levels and dose of drug. (14) In addition, studies have reported that levels obtained with testing are often outside of the therapeutic range. For example, Gearry and colleagues study reported that 41% of values were within the therapeutic range. (15) and Armstrong and colleagues found that 32% of values were within therapeutic levels. (16) The review by Teml et al. concluded that at present therapeutic drug monitoring of 6-thioguanine nucleotides (6-TGN) can be recommended only to estimate patients’ compliance. (13)
There are a large number of studies on the diagnostic performance of TMPT genotyping and phenotyping tests. A recent meta-analysis found a pooled sensitivity of about 80% and specificity near 100% for identifying individuals with subnormal enzymatic activity. In addition, studies have found a greater likelihood of adverse drug reactions with low TPMT activity. One RCT on health outcomes was identified; this study did not find a significant difference in outcomes in patients managed with and without TPMT genotyping testing, but the study may have been underpowered. One-time genotype or phenotype testing is considered medically necessary in select patients.
There is insufficient evidence from prospective studies on whether metabolite markers will lead to improved outcomes (primarily improved disease control and/or less adverse drug effects). Thus, analysis of metabolite markers is considered experimental, investigational and unproven.
Practice Guidelines and Position Statements
A 2011 guideline from the British Association of Dermatologists addressed the safe and effective prescribing of azathioprine. (17) The guideline includes the following recommendations on analysis of TMPT activity:
- There is strong evidence that baseline testing before starting azathioprine predicts severe neutropenia in patients with absent TMPT activity
- There is good evidence that intermediate TMPT activity is associated with myelotoxicity in patients using conventional doses of azathioprine
- There is a continued need for regular monitoring of blood counts in addition to TMPT testing.
A 2010 guideline from the National Academy of Clinical Biochemistry (NACB) states, “thiopurine methyltransferase (TPMT) genotyping is recommended as a useful adjunct to a regimen for prescribing azathioprine” (18) This is an “A-I” recommendation, indicating that the NACB strongly recommends adoption and the recommendation is based on evidence with consistent results from well-designed, well-conducted studies in representative populations.
A 2006 position statement from the American Gastroenterological Association includes the following recommendations (19):
- “Current U.S. Food and Drug Administration (FDA) recommendations suggest that individuals should have thiopurine methyltransferase (TPMT) genotype or phenotype assessed before initiation of therapy with AZA or 6-MP in an effort to detect individuals who have low enzyme activity (or who are homozygous deficient in TPMT) in an effort to avert AZA or 6-MP therapy and thus avoid potential adverse events. Individuals who have intermediate or normal TPMT activity (wild type or heterozygotes) need measurement of frequent complete blood counts (as above) in addition to TPMT assessment because these individuals may still develop myelosuppression subsequent to use of AZA or 6-MP (Grade B)”.
- “Thiopurine metabolite monitoring in the treatment of patients with 6-MP or AZA is useful when attempting to determine medical noncompliance and may be helpful for optimizing dose and monitoring for toxicity (Grade C).”
Evidence grades used in this guideline are:
“Grade A: Homogeneous evidence from multiple well-designed, randomized (therapeutic) or cohort (descriptive) controlled trials, each involving a number of participants to be of sufficient statistical power.
Grade B: Evidence from at least 1 large well-designed, clinical trial with or without randomization from cohort or case-control analytic studies or well-designed meta-analysis.
Grade C: Evidence based on clinical experience, descriptive studies, or reports of expert committees.”
Disclaimer for coding information on Medical Policies
Procedure and diagnosis codes on Medical Policy documents are included only as a general reference tool for each policy. They may not be all-inclusive.
The presence or absence of procedure, service, supply, device or diagnosis codes in a Medical Policy document has no relevance for determination of benefit coverage for members or reimbursement for providers. Only the written coverage position in a medical policy should be used for such determinations.
Benefit coverage determinations based on written Medical Policy coverage positions must include review of the member’s benefit contract or Summary Plan Description (SPD) for defined coverage vs. non-coverage, benefit exclusions, and benefit limitations such as dollar or duration caps.