Thiopurine S -methyltransferase genotype-phenotype concordance: used as a quality assurance tool to help control the phenotype assay

Abstract

Background

As part of the quality control system for our TPMT phenotyping service we monitor the genotype-phenotype concordance for patient samples with deficient and low TPMT activity. We have studied the genotype-phenotype concordance over the last year to demonstrate its effectiveness as a quality assurance tool.

Methods

From July 2007 to July 2008 TPMT genotyping was performed on all routine samples analysed using our phenotypic assay with an activity of ≤40 nmol 6-MTG/gHb/h. The monthly genotype-phenotype concordance was calculated between: all deficient TPMT activity results and a homozygous mutant or compound heterozygote genotype, low TPMT activity and a heterozygote genotype, normal TPMT activity and a wild-type genotype.

Results

A total of 14,832 samples were analysed by TPMT phenotyping and 1769 of these by genotyping. The monthly mean concordance between low TPMT activity and a mutant heterozygote genotype was 83%, ranging from 67–90%. The number of individuals with deficient TPMT activity identified by phenotyping was 44. For two of these individuals only one mutant allele was detected, and for one no common mutations were identified.

Conclusions

Monitoring the genotype-phenotype concordance is an effective quality assurance tool for the TPMT phenotyping assay. As demonstrated in this study current genotyping assays risk missing some deficient patients.

Introduction

We provide a service for measuring whole blood TPMT activity used for screening patients prior to starting treatment with thiopurine drugs. This can identify the 0.3% of individuals with deficient TPMT activity and the 10% with low TPMT activity who are at risk of developing serious adverse reactions. ^1–3 TPMT phenotypic assays are difficult to control and typically all reagents, calibrators and controls are manufactured in-house. Assay drift is a particular problem, and currently there is no active external quality control programme for TPMT phenotyping. This has led us to develop a robust quality control system for checking all reagents prior to use, and with continual performance monitoring using the genotype-phenotype concordance. For genotyping, we use a strategy of screening for the common TPMT mutations: TPMT*2, TPMT*3A and TPMT*3C. ^3,4 Here, we have examined the genotype-phenotype concordance over the last year for our routine TPMT service to investigate its effectiveness as a quality assurance tool.

Materials and methods

From July 2007 to July 2008 TPMT genotyping was performed on all routine samples analysed using our whole blood TPMT phenotypic assay with an activity of ≤40 nmol 6-MTG/gHb/h. The total number of samples received was 14,832 and the number of samples for which genotyping was also performed was 1769 (11.9%).

TPMT phenotyping

Determination of whole blood TPMT activity was carried out as described previously using 6-thioguanine as substrate using high-performance liquid chromatographic (HPLC) analysis and fluorimetric detection. ³ Reference intervals for our whole blood phenotypic assay have been established ³ and are deficient, ≤5; low, 6–34; normal, 35–79; high, ≥80 (units, nmol6-MTG/gHb/h).

TPMT genotyping

Total DNA (100–500 μg) was extracted from buffy coats using a standard phenol/chloroform extraction method. TPMT genotyping was performed using a previously described multiplex amplification refractory mutation system (ARMS) strategy to simultaneously screen for the common TPMT mutations TPMT*2, TPMT*3A and TPMT*3C. ² Each ARMS run was controlled using a water and environmental blank together with control samples for the genetic variations being examined.

Genotype-phenotype concordance

The genotype-phenotype concordance was calculated for each month as follows:

Samples with deficient TPMT activity (≤5 nmol 6-MTG/gHb/h) and a homozygote TPMT*3/*3 or compound heterozygote TPMT*3/*2 genotype;

Samples with low TPMT activity (6–34 nmol 6-MTG/gHb/h) and a heterozygote TPMT*1/*2 or TPMT*1/*3 genotype;

Samples with a TPMT activity just above the cut-off between the low and normal reference interval (35–40 nmol 6-MTG/gHb/h) and a wild-type (normal) TPMT*1/*1 genotype.

Results

The monthly TPMT genotype-phenotype concordance between June 2007 and June 2008 is presented in Table 1. The monthly mean concordance between low TPMT activity and a mutant heterozygote genotype was 83%. This is comparable with the genotype-phenotype concordance of 79% we previously reported from a study of 402 whole blood routine patient samples. ³ It is also similar to that reported by others of 50–80%. ⁴ For TPMT activity results just above the low reference interval cut-off of 34 nmol 6-MTG/gHb/h, the mean concordance for the year was lower at 68%.

Table 1

Monthly concordance between whole blood thiopurine S-methyltransferase (TPMT) phenotypic activity and TPMT genotype

	TPMT activity (nmol 6-MTG/gHb/h)
	Deficient (≤5)		Low (6–34)		Normal cut-off (35–40)
Month	n	% Concordance	n	% Concordance	n	% Concordance
July 2007	1	100	153	67	20	100
August 2007	5	100	128	78	47	81
September 2007	1	100	108	90	32	69
October 2007	1	100	128	84	40	75
November 2007	6	83	98	83	7	43
December 2007	1	100	85	74	6	33
January 2008	4	75	80	86	2	100
February 2008	5	80	92	85	4	50
March 2008	5	100	101	88	3	33
April 2008	3	100	110	88	2	50
May 2008	7	100	92	82	1	100
June 2008	5	100	119	85	6	83
Mean	4	95	108	83	14	68

% Genotype-phenotype concordance were determined for each month as follows: patients samples with deficient TPMT activity (≤5 nmol 6-MTG/gHb/h) and either a homozygote TPMT*3/*3 or compound heterozygote TPMT*3/*2 genotype, samples with low TPMT activity (6–34 nmol6-MTG/gHb/h) and a heterozygote TPMT*1/*2 or TPMT*1/*3 genotype and samples with TPMT activity just above the cut-off between the low and normal reference interval (35–40 nmol 6-MTG/gHb/h) and a wild-type TPMT*1/*1 genotype

On average we identified four deficient TPMT status patients each month. The mean monthly genotype-phenotype concordance (presence of homozygote mutant or compound heterozygote) was 95% for deficient patients.

Discussion

There is increasing evidence to support treating patients with low TPMT activity with less than the normal standard dose of thiopurine, for example, 50–33% of the recommended dose. ^1,2 It is important to identify individuals with reduced enzyme activity to ensure that they receive the right treatment. This is a particular challenge for TPMT phenotypic assays that are difficult to control. Examination of the monthly mean concordance between low TPMT activity and a heterozygote genotype for our phenotypic assay reveals 3 months in particular, where the concordance was low compared with the overall mean of 83%; July 2007 (67%), August 2007 (78%) and December 2007 (74%). For July and August 2007, this decrease in concordance coincided with the use of a new lot number of S-adenosylmethionine (SAM). SAM is present as the methyl donor in the TPMT enzyme reaction. ⁵ The low genotype-phenotype concordance prompted us to perform enzyme kinetic studies, which identified a problem with a particular lot of SAM that was not revealed by batch comparison of the TPMT reaction reagent, or internal control results. A different lot of SAM gave improved assay performance with increased concordance. The decrease in concordance in December 2007 was seen at a time when the samples were delayed in the Christmas post and this may be a preanalytical factor being picked up by our genotype QC regimen.

The genotype-phenotype concordance was lower for TPMT activity results just above the cut-off for low TPMT activity (35–40 nmol 6-MTG/gHb/h) and a wild-type TPMT*1/*1 genotype, with an overall mean of 68%. This lower concordance around the cut-off is of course to be expected. Although it is partly weighted by the small sample number obtained for some months, it remains useful as it can further help identify changes in assay performance and the cause of such changes.

Patients with deficient TPMT activity are exposed to utmost risk of life-threatening reactions to thiopurine drugs, such as myelosuppression, and normally treatment is contraindicated. ^1,2 A primary function of any method used for determining patient TPMT activity status is to identify individuals correctly. Out of the 44 patients, we identified with deficient TPMT activity between July 2007 and June 2008, only two had one mutant allele (TPMT*1/*3 carrier genotype, consistent with low TPMT activity) and one no mutant alleles (consistent with normal TPMT activity). Results were confirmed by repeat analysis with a second sample. This demonstrates the utility of the phenotypic approach to TPMT screening. So far, approximately 23 different mutations of the TPMT gene have been identified. Our TPMT genotyping strategy, along with most others, ^2–4 screens for the common mutations TPMT*2, TPMT*3A and TPMT*3C. ³ These common mutations are reported to be responsible for between 60% and 95% of mutant alleles for deficient TPMT activity in Caucasian, Asian and African-American populations, which is consistent with our findings. ^3,4

Apart from using the genotype as a QC tool, we also routinely report TPMT genotypes for deficient patients, those who have recently had an adverse reaction, and where patients have received recent blood products, as this can give a misleading phenotypic result. ⁶

We have demonstrated that monitoring of the TPMT genotype-phenotype concordance is able to identify poor assay performance not revealed by reagent checks or internal quality control. This study also demonstrates that genotyping can sometimes miss TPMT-deficient patients due to the occurrence of genotypes that are not being screened for, whereas a well-controlled phenotypic assay will always identify individuals with deficient TPMT activity.