Comparison of variants in TPMT and NUDT15 between sequencing and genotyping methods in a multistate pediatric institution

Abstract

The risk of severe adverse events related to thiopurine therapy can be reduced by personalizing dosing based on TPMT and NUDT15 genetic polymorphisms. However, the optimal genetic testing platform has not yet been established. In this study, we report on the TPMT and NUDT15 genotypes and phenotypes generated from 320 patients from a multicenter pediatric healthcare system using both Sanger sequencing and polymerase chain reaction genotyping (hereafter: genotyping) methods to determine the appropriateness of genotyping in our patient population. Sanger sequencing identified variant TPMT alleles including *3A (8, 3.2% of alleles), *3C (4, 1.6%), and *2 (1, 0.4%), and NUDT15 alleles including *2 (5, 3.6%) and *3 (1, 0.7%). For genotyped patients, variants identified in TPMT included *3A (12, 3.1%), *3C (4, 1%), *2 (2, 0.5%), and *8 (1, 0.25%), whereas NUDT15 included *4 (2, 1.9%) and *2 or *3 (1, 1%). Between Sanger sequencing and genotyping, no significant difference in allele, genotype, or phenotype frequency was identified for either TPMT or NUDT15. All patients who were tested using Sanger sequencing would have been accurately phenotyped for either TPMT (124/124), NUDT15 (69/69), or both genes (68/68) if they were assayed using the genotyping method. Considering 193 total TPMT and NUDT15 Sanger Sequencing tests reviewed, all tests would have resulted in an appropriate clinical recommendation if the test had instead been conducted using the comparison genotyping platforms. These results suggest that, in this study population, genotyping would be sufficient to provide accurate phenotype calls and clinical recommendations.

Study Highlights.

• WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC?

Although the use of TPMT and NUDT15 testing is widely accepted, current guidance for the methods used to identify their genotype is unclear.

• WHAT QUESTION DID THIS STUDY ADDRESS?

In this study, we compared the prevalence of the genotypes and phenotypes generated from patients with TPMT and/or NUDT15 results measured via Sanger sequencing from a multicenter pediatric healthcare system and compared them with results measured by different genotyping platforms. We also assessed the theoretical clinical accuracy of recommendation that would have been made if patients tested by Sanger sequencing were instead tested by genotyping.

• WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE?

The results of this study demonstrate the appropriateness of genotyping to accurately identify clinically meaningful genomic variants in TPMT and NUDT15.

• HOW MIGHT THIS CHANGE CLINICAL PHARMACOLOGY OR TRANSLATIONAL SCIENCE?

The results of this study could be used to build consensus recommendations regarding the methods of identifying genomic variants in TPMT and NUDT15.

INTRODUCTION

Thiopurines, including mercaptopurine, azathioprine, and thioguanine, are cornerstone therapies for certain cancers, including acute lymphoblastic leukemia, as well as autoimmune disorders. ¹ , ² , ³ , ⁴ These therapeutic agents are associated with severe adverse events, including myelosuppression and gastrointestinal disorders. ⁵ Pharmacogenomic (PGx) variants in thiopurine methyltransferase (TPMT) and nudix (nucleoside diphosphate linked moiety X)‐type motif 15 (NUDT15) can be used to personalize thiopurine dosing and reduce the risk for toxicity without compromising therapeutic efficacy. ⁶ , ⁷ Current clinical guidance from the US Food and Drug Administration (FDA), Children's Oncology Group (COG) and Clinical Pharmacogenetics Implementation Consortium (CPIC) suggests that identification of a variant genotype for TPMT and NUDT15 is clinically valuable and may be used to direct therapy. ⁸ , ⁹ Recommendations for dose‐adjustment of thiopurines based on TPMT and NUDT15 genotype is widely available through CPIC. ⁸

Whereas the use of TPMT and NUDT15 testing is widely accepted, and now mandated in recent B‐cell leukemia COG clinical trials, current guidance for the methods used to identify TPMT and NUDT15 genotype is unclear. ⁸ Accurate assignment of metabolizer phenotypes is vital for personalization of thiopurine therapy through accurate and precise genotyping. TPMT and NUDT15 genotypes can be identified through polymerase chain reaction (PCR)‐based genotyping (hereafter: genotyping) or sequencing methods, including Sanger and next‐generation sequencing. ¹⁰ Although genotyping holds several advantages over either sequencing method, it is restricted to identifying specific genomic variants selected prior to testing. Therefore, depending on the number of variants tested, rare but meaningful genotypes may be missed using a genotyping method. As sequencing approaches provide significantly more genomic information compared to genotyping, it is possible that these methods might identify the presence of variants not identified by genotyping.

Whereas genotyping is considered a gold standard in PGx clinical trials, ¹¹ it is currently unknown what proportion of patients may be incorrectly phenotyped using a genotyping approach to TPMT and NUDT15. In this study, we report on the genotypes and phenotypes generated from patients with TPMT and/or NUDT15 results measured via Sanger sequencing from a multicenter pediatric healthcare system. We also report additional genotypes and phenotypes generated from patients with TPMT and/or NUDT15 results measured by different genotyping platforms. Finally, we report the minimum genotyping design necessary to accurately phenotype our patient population and report the rate of rare TPMT and NUDT15 genotypes.

METHODS

Data collection

TPMT and NUDT15 laboratory results prior to February 1, 2022, were abstracted from the electronic health records. All laboratory results collected were performed as standard of clinical care. Data included patients from inpatient and outpatient clinics across multiple states across the Nemours Children's Health (NCH) system. All functional assay results were excluded and only patients with genetic results were included in the analysis. As our analyses were focused on the comparison of laboratory assays only, no clinical inclusion or exclusion criteria were used in this study. Data collection was conducted and reviewed by at least two authors to ensure accuracy. Clinical data collected included patient reported race, patient reported ethnicity, and sex. This study was approved by the NCH Institutional Review Board (protocol no. 1845767).

Sanger sequencing

Sanger sequencing for TPMT and/or NUDT15 results were generated from an internal, Clinical Laboratory Improvement Amendment (CLIA)‐approved laboratory at NCH (Figure 1a,b). The assay is validated using reference samples. Genomic DNA was extracted from blood or buccal swabs and used to amplify the coding regions for exons 4, 6, and 9 of TPMT, and exons 1 and 3 of NUDT15 using standard PCR protocols. Notably, the use of PCR in this method is distinct from the use of PCR to call a genotype. Here, PCR is used to amplify target regions of a DNA sample prior to Sanger sequencing. Bidirectional sequencing analysis was performed using Big Dye Terminator version 3.1 Cycle Sequencing Kit (Applied Biosystems). Nucleotide sequences were compared to the control reference sequence (NM_000367.2 and NM_018283.3) by MacVector and Assembler Software (Apex, NC). Alleles previously reported as TPMT*1S were considered as TPMT*1 to maintain consistency due to the removal of reporting TPMT*1S during the study period. ⁸ , ¹² , ¹³

FIGURE 1.

Allelic prevalence by testing method. Comparative allelic coverage for TPMT (a) and NUDT15 (b) for tier 1 and 2 star alleles. (c) Allelic prevalence for TPMT between Sanger sequencing (248 alleles), TPMT #1 (270 alleles), and TPMT #2 (120 alleles). Boxes marked with “X” represent alleles not tested by individual assay. (d) Allelic prevalence for NUDT15 between Sanger sequencing (138 alleles) and NUDT15 #1 (104 alleles).

Polymerase chain reaction genotyping

PCR‐based genotyping results for TPMT testing were identified from five commercial laboratories using CAP/CLIA certified, clinically validated testing platforms. Four laboratories tested for TPMT alleles *2, *3A, *3B, and *3C (hereafter TPMT #1). One laboratory tested for TPMT alleles *2, *3A, *3B, *3C, *4, *5, *8, and *12 (hereafter TPMT #2). Results for NUDT15 testing were identified from one commercial laboratory which tested for NUDT15 alleles *2 or *3, *4, and *5 (hereafter NUDT15 #1). Notably, this commercial NUDT15 PCR platform could not distinguish between *2 and *3 variants of NUDT15. Both the *2 and *3 variants result in the same functional phenotype and are therefore reported here as “*2 or *3,” ¹⁴

Data and statistical analysis

All data were analyzed in R statistical software version 3.4.3 (R Project for Statistical Computing) or Prism analysis and graphing software version 8.0.0 (GraphPad). Neither testing approach is able to phase results, therefore TPMT*1/*3A (vs. TPMT*3B/*3C) and NUDT15*1/*2 (vs. NUDT15*3/*6) were reported based on population frequency data. ⁸ As this study was a retrospective analysis of all available genomic data for the TPMT and NUDT15 genes across our institution, no power calculation was conducted and all data meeting our criteria were included. Genotype to phenotype translations and dosing recommendations were determined by the CPIC clinical guidelines. ⁸ For categorical variables, χ ² tests were used to generate p values. Hardy–Weinberg equilibrium was assessed using the genetics_v1.3.8.1.3 package in R. Two‐tailed p values ≤ 0.05 were considered statistically significant.

RESULTS

Cohort characteristics

We identified 124 patients with TPMT Sanger sequencing data and 69 patients with NUDT15 Sanger sequencing data (Tables S1 and S2). A total of 68 patients had both TPMT and NUDT15 Sanger sequencing data available. A separate set of 195 patients with TPMT genotyping data and 52 patients with NUDT15 genotyping data were also identified. Of patients with genotyping data available, 52 had both TPMT and NUDT15 genotyping data available. No patients had a combination of Sanger sequencing and genotyping results for TPMT and NUDT15. Demographic features were not significantly different between patients with Sanger sequencing or genotyping results. Full demographic characteristics are available in Table 1.

TABLE 1.

Patient demographics.

Demographic	Sanger sequencing (n = 125)	PCR genotyping (n = 195)
Sex
Female	54 (43.2)	100 (51.3)
Male	71 (56.8)	95 (48.7)
Race
Asian Indian	3 (2.4)	0 (0)
Black or African American	21 (16.8)	25 (12.8)
Chinese	1 (0.8)	0 (0)
Filipino	0 (0)	1 (0.5)
Other	21 (16.8)	36 (18.5)
Other Asian	4 (3.2)	1 (0.5)
Unknown	0 (0)	3 (1.5)
White or Caucasian	75 (60)	129 (66.2)
Ethnicity
Another Hispanic, Latino, or Spanish Origin	20 (16)	33 (16.9)
Mexican, Mexican American, Chicano/a	1 (0.8)	0 (0)
Non‐Hispanic or Latino	100 (80)	154 (79)
Puerto Rican	1 (0.8)	0 (0)
Unknown	3 (2.4)	8 (4.1)
Pharmacogenetic results
TPMT and NUDT15	68 (54.4)	52 (26.7)
TPMT only	56 (44.8)	143 (73.3)
NUDT15 only	1 (0.8)	0 (0)

Note: All data presented as number (percent). Percentages may not add to 100 due to rounding.

Genomic variants by Sanger sequencing

Of 124 patients with TPMT Sanger sequencing results available, 112 patients (90.3%) were called as a genotype of TPMT*1/*1. Variant alleles identified included TPMT*3A (8, 3.2% of alleles), *3C (4, 1.6%), and *2 (1, 0.4%). TPMT metabolizer phenotypes predicted for these patients included 112 (90.3%) normal metabolizers, 11 (8.9%) intermediate metabolizers, and one (0.8%) poor metabolizer. Of the 69 patients with NUDT15 Sanger sequencing results available, 63 patients (91.3%) were called as a genotype of NUDT15*1/*1. Variant alleles identified included NUDT15*2 (5, 3.6% of alleles) and *3 (1, 0.7%). NUDT15 metabolizer phenotypes predicted for these patients included 63 (91.3%) normal metabolizers and six (8.7%) intermediate metabolizers.

Genomic variants by genotyping

Of patients genotyped, the majority were called as a genotype of *1/*1 for TPMT (TPMT #1: 125, 92.6% and TPMT #2: 53, 88.3%). The variant alleles identified by TPMT #1 included TPMT*3A (7, 2.6%), *2 (2, 0.7%), and *3C (2, 0.7%) resulting in TPMT metabolizer phenotypes of 125 (92.6%) normal metabolizers, nine (6.7%) intermediate metabolizers, and one (0.7%) poor metabolizer. Of patients genotyped by TPMT #2, variant alleles identified included TPMT*3A (5, 4.2%), *3C (2, 1.7%), and *8 (1, 0.8%) resulting in TPMT metabolizer phenotypes of 53 (88.3%) normal metabolizers, five (8.3%) intermediate metabolizers, one (1.7%) poor metabolizer, and one (1.7%) indeterminate metabolizer. Of patients genotyped with NUDT15 #1, the majority were called as a genotype of NUDT15*1/*1 (49, 94.2%). The variant alleles identified included NUDT15*4 (2, 1.9%) and *2 or *3 (1, 1%) resulting in NUDT15 metabolizer phenotypes of 49 (94.2%) normal metabolizers, one (1.9%) intermediate metabolizer, and two (3.9%) indeterminate metabolizers.

Comparison of genotype, phenotype, and clinical action determined by testing method

Between Sanger sequencing and genotyping, no significant difference in allele frequency for TPMT (p = 0.53; Figure 1c) or NUDT15 (p = 0.09; Figure 1d), genotype frequency (TPMT p = 0.44; NUDT15 p = 0.09), or phenotype frequency (TPMT p = 0.47 and NUDT15 p = 0.24) was identified. Considering TPMT, if patients who were tested using Sanger sequencing were tested by either genotyping method, all patients would have been accurately genotyped and phenotyped. If patients who received genotyping of TPMT with TPMT #2 were tested by the less broad TPMT #1, 99.3% (59/60) would have been accurately genotyped and phenotyped. In this case, the one TPMT*1/*8 result identified by TPMT #2 would be misclassified as a normal metabolizer. All patients tested using Sanger sequencing of NUDT15 would have been accurately phenotyped if tested by the available genotyping method. However, six patient genotypes would be reported as NUDT15*1/*2 or *3 due to the inability to design probes to accurately detect the six base pair insertion which would differentiate the *2 from *3 alleles. Considering the 193 combined TPMT and NUDT15 Sanger sequencing tests reviewed, all tests would have resulted in an appropriate clinical recommendation if the test had instead been conducted using the comparison genotyping platforms.

DISCUSSION

Genomic testing for both TPMT and NUDT15 are highly utilized in clinical care for patients receiving thiopurine therapy. Optimal testing for these genes is therefore a critical component of delivering high‐quality care to patients. ¹⁵ Although Sanger sequencing can identify a broader number of variants compared to genotyping, there are key benefits associated with genotyping which could influence an institution's choice in testing method. A genotyping test is more amenable to multi‐gene panel‐based testing, is less costly, and requires less informatics infrastructure to implement in a Clinical Decision Support System. ¹⁶

From our evaluation across a multicenter pediatric healthcare system, all patients in our population would be accurately phenotyped with genotyping in comparison to Sanger sequencing. One patient would have been classified as a TPMT*1/*1 normal metabolizer with TPMT #1 when TPMT #2 genotyping detected a *1/*8 genotype. The TPMT*8 allele function is currently unknown, and the TPMT*1/*8 genotype would translate to an indeterminate metabolizer phenotype. ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ Current recommendations for patients with a TPMT*1/*8 genotype are to evaluate TPMT erythrocyte activity or monitor closely for toxicity if thiopurines are required. ⁸ No NUDT15 Sanger sequencing results would have changed clinical recommendations if tested using genotyping. The six results that would have been reported as NUDT15*1/*2 or *3 would ultimately translate to a NUDT15 intermediate metabolizer status. Importantly, the clinical recommendations for dose reduction would be the same regardless of the testing method meaning the ultimate clinical care with either methods would be appropriate.

Concordance between Sanger sequencing and genotyping methods has been previously studied and high concordance has been reported across multiple pharmacogenes. ¹⁰ , ²¹ In a study using genomewide genotyping, the concordance between Sanger sequencing and imputed genotypes for TPMT was 84%. ³ In our study population, the concordance of our clinical TPMT phenotype results with those generated by Sanger sequencing was 100%, suggesting that Sanger sequencing can be used to determine the minimum TPMT genotyping allele design for our patient population. It is not surprising that our concordance is higher than the previously published data, as the sequencing method was targeted to a select set of exons of each gene as opposed to full sequencing of each gene. It is, however, encouraging that our phenotype concordance for both TPMT and NUDT15 is perfect in this dataset. Selection of the alleles identified in this study is further supported by the recently published TPMT and NUDT15 Genotyping Recommendation joint consensus. ¹⁴ This consensus recommends that minimum genotyping should include *2, *3A, *3B, and *3C for TPMT and *3 for NUDT15. Using these minimum genotyping recommendations, 100% of the patients who received Sanger sequencing in this study would be accurately phenotyped.

Our findings are limited in their internal and external validity and care should be taken in extrapolating these results. First, this study is limited by its retrospective design as well as lack of randomization between patients who received genotyping or Sanger sequencing. It is not possible with our data to prevent bias from a lack of controlling which patients were sequenced or genotyped. Furthermore, a lack of paired sequencing and genotyping data for the same patients limits the true accuracy of our findings. Additionally, the sequencing method used in this study detects variants in exons targeted, and may not account for variants outside these regions. Our external validity is also limited by our sample size and specific patient population. Whereas the alleles tested for in this study are likely appropriate for a largely European/White populations such as this, key variants prevalent in other ancestral populations (i.e., TPMT *8 in Sub‐Saharan African, *24 in African American/Afro‐Caribbean, and NUDT15 *5 and *6 in East Asian) were not included across all platforms. ⁸ , ¹⁴ Design of testing platforms should consider the addition of these variants despite the low allele frequency. Institutions should review their internal data and laboratory guidance before selecting their minimum allele testing criteria. There are also limitations on the methods included in this study, despite their current clinical use. Neither testing approach is well‐equipped to identify phasing of results (i.e., TPMT*1/*3A vs. TPMT*3B/*3C and NUDT15*1/*2 vs. NUDT15*3/*6). Few clinically available tests are able to accurately phase these data; therefore, most clinical laboratories reported phasing based on population frequency data. ⁸ , ¹⁴ Novel methods including long‐read sequencing and single molecule sequencing may be able to overcome this limitation in the future.

CONCLUSION

In this study, we aimed to test the concordance between Sanger sequencing and genotyping methods as well as identify the minimum genotyping design necessary to accurately phenotype our patient population for TPMT and NUDT15. In this cohort, all therapeutic recommendations would have been unchanged if patients receiving Sanger sequencing for TPMT and NUDT15 instead were genotyped. This demonstrates that most patients will receive appropriate dose adjustments based on the phenotype determination from a limited genotyping panel. Institutions who are considering switching from Sanger sequencing to genotyping should evaluate their institution's specific population allele frequency to ensure relevant variants are included in the assay in addition to following published recommendations for minimum genotyping design. The results of this study may be used by healthcare systems to determine the optimal methods of identifying TPMT and NUDT15 genotypes in their patient population.

AUTHOR CONTRIBUTIONS

K.J.C., V.G., L.G., B.Q.D., R.C.G., E.A.K., M.B., A.S.B., R.N., S.M.K., K.V.B., and N.D.S. wrote the manuscript. K.J.C. and N.D.S. designed the research. K.J.C., V.G., L.G., and N.D.S. performed the research. K.J.C., B.Q.D., S.M.K., and N.D.S. analyzed the data.

FUNDING INFORMATION

Publication made possible in part by support from the Nemours Grants for Open Access from Library Services and the Nemours Precision Medicine Program.

CONFLICT OF INTEREST STATEMENT

The authors declared no competing interests for this work.

Supporting information

Appendix S1

Cook KJ, Grusauskas V, Gloe L, et al. Comparison of variants in TPMT and NUDT15 between sequencing and genotyping methods in a multistate pediatric institution. Clin Transl Sci. 2023;16:1352‐1358. doi: 10.1111/cts.13539