Pharmacogenetics encompasses almost 100 diverse monogenically transmitted conditions, including polymorphisms of N-acetyltransferase, serum butyrylcholinesterase (pseudocholinesterase), serum arylesterase (paraoxonase), glutathione S-transferase, debrisoquine, and mephenytoin [2, 3]. Drug accumulation generally arises from an abnormal enzyme that blocks the drug's usual pathway of inactivation. The parent drug is sometimes inactive, as in debrisoquine polymorphism, in which affected patients do not obtain adequate analgesia with codeine because they cannot convert it to morphine.

Many drugs participate in some pharmacogenetic conditions. For example, 30 different drugs act, to some extent, as substrates in the debrisoquine polymorphism caused by mutations of the cytochrome P450 isozyme designated CYP2D6. Generally, more than half of the total metabolism of a drug must pass through this aberrant isozyme for an adverse effect to arise. It should be stressed that several different P450 isozymes often participate in the metabolism of a single drug. Thiopurine S-methyltransferase biotransforms at least three drugs: mercaptopurine, azathioprine, and thioguanine.

Most pharmacogenetic conditions go undetected in affected persons until a drug substrate for the aberrant isozyme is administered. Deficiency in TPM differs in two respects from most other pharmacogenetic conditions: 1) TPM genetic polymorphism was discovered in normal, unmedicated persons (1 in 300 [0.3%] was homozygous for the mutant allele and had low activity, 10% to 11% were heterozygous and had intermediate activity, and 89% to 90% were homozygous for the normal allele and had high activity [4, 5]); and 2) toxicity develops not from the parent drug or its initial metabolite but from a distant metabolite, thioguanine nucleotide. Blockage of one main pathway for metabolic inactivation of mercaptopurine or azathioprine through TPM shunts more mercaptopurine and azathioprine through another pathway to form toxic thioguanine nucleotide concentrations. In 1987, Lennard and colleagues [6] first reported the clinical significance of TPM polymorphism, an observation that has been amply confirmed [7].

Thioguanine nucleotide concentrations and leukocyte counts are reciprocally related [8]. Thus, mercaptopurine and azathioprine toxicity can be estimated indirectly by monitoring leukocyte counts. Determining phenotype for TPM polymorphism by measuring enzymatic activity in erythrocytes has become a standard aspect of care for patients with leukemia before thiopurine therapy. This is done to avoid the toxicity that can occur if patients with inherited low levels of TPM activity are exposed to standard doses of thiopurine drugs. One approach is to measure TPM activity before therapy to avoid potentially life-threatening myelosuppression in homozygous recessive patients with low TPM activity. Because of the possibility of TPM induction, enzyme activity is then measured again during therapy in persons without low activity. Erythrocyte thioguanine nucleotide concentrations can also be monitored throughout therapy. This approach can be invalidated by recent blood transfusions, which can falsely elevate TPM activity. Moreover, thiopurines are often given in combination with other myelosuppressive drugs, making it difficult to ascertain which medication is being given in too high a dose. Thus, Yates and colleagues' approach has the advantage of providing a molecular indication of low TPM activity as a basis for mercaptopurine and azathioprine toxicity.

Yates and colleagues report exemplifies current widespread applications of molecular biology to pharmacogenetics. The authors used new, automated DNA technology that could make genotyping of polymorphisms routine in clinical chemistry laboratories. In affected persons, these methods allow detection of mutant alleles at the genomic level, thereby providing a way to prevent drug toxicity before drug administration. Therapeutic safety and reduced risk for future toxicity can be achieved by 1) selecting an entirely different agent that is metabolized by other routes or 2) reducing the drug dose. In TPM deficiency, the mercaptopurine dose is reduced by 10- to 15-fold. Dose adjustments of this magnitude would be difficult to make empirically without previous knowledge of the biochemical or molecular mechanisms. Isolation and characterization of mutant alleles causing TPM deficiency in humans and development of polymerase chain reaction and restriction fragment-length polymorphism assays to detect these mutations in genomic DNA now enable identification of more than 80% of all TPM mutant alleles in white persons [1, 7]. However, both phenotyping and genotyping of TPM have advantages and disadvantages. Numerous variant alleles create technical problems with unequivocal genotyping of samples. Although the genotyping methods reported by Yates and colleagues had excellent phenotypic concordance in the white persons studied by these authors, additional mutant alleles will be discovered, especially in other ethnic groups.

Previous observations that led to Yates and colleagues' research must be acknowledged. Development of these genotyping methods was possible only after Honchel and colleagues [9] cloned the complementary DNA and active gene for TPM, an accomplishment that was followed by isolation and characterization of the structural gene and mutant alleles associated with TPM deficiency [10-12].

A fundamental pharmacogenetic principle is that genetic factors rarely act alone; rather, they act in concert with many environmental and developmental factors [13]. This principle applies to TPM polymorphism. Mercaptopurine, azathioprine, and thioguanine are prodrugs that require metabolic activation to produce therapeutic effects. However, the biotransformation of mercaptopurine is complex, involving not only TPM but also xanthine oxidase, hypoxanthine phosphoribosyltransferase, inosine monophosphate dehydrogenase, and guanosine monophosphate synthetase. Thus, many host factors are able to influence mercaptopurine disposition; this explains why little relation is seen between the dose of mercaptopurine administered and the therapeutic effects of the drug. Instead, therapeutic effects correlate with thioguanine nucleotide concentrations. Apparently, interindividual differences between mercaptopurine dose and thioguanine nucleotide concentrations cannot be explained entirely by TPM polymorphism. This conclusion warns against relying solely on TPM assays to identify patients who need individualization of mercaptopurine doses. All patients require such tailoring, guided by leukocyte and platelet counts and thioguanine nucleotide concentrations. For example, patients with cancer who are receiving allopurinol require reduction of initial oral doses of mercaptopurine because allopurine inhibits the metabolic inactivation of mercaptopurine by xanthine oxidase.

Other factors can influence TPM activity. Sulfasalazine and its metabolite 5-aminosalicylic acid inhibit TPM activity [14], whereas mercaptopurine [15] or diuretics [16] can induce this activity. Elevated TPM activity resulting from induction increases mercaptopurine dose requirements because patients with high TPM activity are undertreated with standard thiopurine doses. In addition, TPM activity is affected by patient age [17], sex [16], and renal function [18]. Many interacting host factors that influence mercaptopurine disposition require close observation of all patients receiving mercaptopurine for potential toxicity and careful titration of dose against leukocyte and platelet counts.

A final therapeutic principle derived from pharmacogenetics and illustrated by TPM deficiency is that patients homozygous or even heterozygous for the defect are more susceptible to potential toxicity from drug interactions, especially such inhibitors as sulfasalazine [12] that further reduce the activity of an abnormal enzyme.

The significance of the TPM polymorphism transcends its relation to any drug or drug class. Thiopurine S-methyltransferase probably exists in humans not to metabolize any single drug but rather to catalyze S-methylation of currently unknown endogenous substrates. Further research is needed to identify these substrates and provide reasons for the genetic polymorphism, ethnic differences in its gene frequency [7, 16, 19, 20], and its true role in health and disease.