A guideline consists of treatment recommendations intended to represent evidence-based best practices. Guidelines often appear in synchrony with dissemination of new information or emergence of new technology, informing their introduction to practice amid better-established treatments that have proved insufficiently effective or tolerable. In the example of applying genotype-inferred phenotypes intended to minimize adverse effects of slow metabolism and/or drug-drug interactivity, extant guidelines may exceed the evidence upon which they are based. They withal set a precedent, in that prior guidelines pertaining to therapeutic monitoring have relied upon correlation of substrate bioavailability with clinical effects in the same patient. Contrariwise, recently introduced assays for therapeutic drug monitoring measure cytochrome or conjugating enzyme activity in individual patients and then correlate the latter to retrospectively determined substrate bioavailability and clinical effects. Once the blood is drawn, the patient (and real-time blood drug concentrations) are out of the loop. Dosing and interpretation of drug effects thereafter rely upon genotype, which derives importance by virtue of phenotype. Yet, phenotype is affected as much or more by coadministered drugs in real time as it is by inference pertaining to (typically reduced) enzyme activity of inherited variant alleles, from which predictions drawn may not apply in real time (when, for example, an inhibitor or inducer of a cytochrome enzyme or conjugating enzyme, is coadministered). Not every insurer pays for genotyping and not all labs measure the same diplotypes of the same enzymes. Obliging a patient to pay five hundred dollars for a non-standardized research procedure is neither good medicine nor good ethics.
If the rule is that genotyping cytochrome enzyme genes yields modest clinical return, distractions can be found to prove it. One comes to us as a review cum guideline (of when to assay a patient’s genotype of drug-metabolizing cytochrome isoenzymes that operate in Phase I drug metabolism). It was published in 2017 online and in print by a highly regarded Korean group (Sollip Kim and colleagues, Annals of Laboratory Medicine 37(2):180-193, 2017), whose exhaustively consensualized methodology is adapted from those of each English-language guideline published by academically affiliated groups over the five years preceding their own.
The authors begin with the anticoagulant warfarin (whose S-isomer, a substrate of CYP2C9, exists as three alleles–*1, the normally active “wild type,” and *2 and *3, variants with 10 to 40 percent of the activity of the latter); they briefly discuss warfarin’s target enzyme, VKORC1 a vitamin K epoxide reductase , encoded from a gene of the same name, that reduces vitamin K epoxide to vitamin K (the rate-limiting step of vitamin K synthesis,. The authors discuss the three erstwhile identified alleles that “code for” variant (and less active) enzymes that metabolize the active isomer of warfarin (and their differences in anticoagulant potency, relative to the “wild-type” warfarin allele *1.
Yet, no specific warfarin guidelines are offered with respect to warfarin dosing in patients with low-activity *2 and *3 alleles of CYP2C9, except to say that they are lower than those of the more potent *1 allele. Epidemiologic and other “host factor” sources of variation of potency are alluded to but not discussed.
Also not discussed are CYP2C9 substrates other than warfarin, their likelihood of competitively inhibiting warfarin metabolism, or mechanism-based, irreversible inhibitors of CYP2C9. No mention is made of extents to which they may boost warfarin bioavailability and affect clinical effectiveness or safety.
The authors concede, finally, that the value of the warfarin guidelines is “low” and that no recommendations can be made for children. INR must be monitored as usual, whether or not algorithms based on CYP2C9 genotype are used. That is to say, the guideline on use of warfarin in persons with diplotypes comprising low-activity, variant alleles lacks specific recommendations about dosing and offers nothing about warfarin dosing during coadministration with CYP2C9 inhibitors, CYP2C9 inducers (which increase enzyme activity, accelerate clearance, and lower warfarin blood concentrations), or other CYP2C9 substrates.
Similarly perfunctory are the guidelines pertaining to CYP2D6 (whose substrates constitute 25 percent of all FDA-approved drugs): a small fraction are cited, for which recommendations are not offered; nor are they for the prodrugs codeine (metabolized to morphine), tamoxifen (metabolized to its anti-estrogenic active principle, endoxifen (4-hydroxy tamoxifen)), or the pro-drug tramadol (inert, until converted to its active analgesic principle). CYP3A4, which contributes to the metabolism of fifty percent or more of FDA-approved drugs, is not even mentioned (despite documentation of variant alleles with low activity, a plenitude of inhibitors, and the highest number of inducers among all of the human drug-metabolizing cytochrome enzymes. Nor is CYP1A2, which contributes to demethylation of several atypical antipsychotics and is robustly induced by tobacco smoking (so that blood drug concentrations of substrate drugs have been reported lower in smokers, and to rise by clinically significant increments in persons who quit smoking). CYP2B6 contributes to hepatic metabolism of several second-generation antidepressants, and a range of CYP2B6 inhibition has been associated with SSRIs and with several drugs used in primary care. Instead are offered generic mechanistic features of the hepatic conjugating enzymes, uridinyl glucuronosyltransferases (UGTs, responsible for clearance of lamotrigine and the provenience of drug-drug interactivity between the latter and divalproex likewise are not mentioned in the authors’ “guidelines.”
(Obiter: Marketers of “genetic testing” for CYP genotypes promote a service that has not been documented to enhance efficacy and safety of treatment, and because pharmacotherapy is necessarily unstable because of dosage adjustments intended to achieve optimality, seems unlikely to be. Substrate activity today may not be inferable from a genotype determined two weeks ago. (Adding bupropion today, for example, may rapidly lower CYP2D6 activity in a very rapid metabolizer (someone with duplicate wild-type alleles) to that of a poor metabolizer (homozygous variant alleles). A better and less expensive way to discover whether the blood concentration of a drug is high or low with respect to dosage is contemporaneous measurement of drug blood concentration for comparison with modal drug concentrations at comparable dosage.) Overall point: Use of a valid, sensitive, and specific test or device does not imbue a guideline that uses it with validity, sensitivity, and specificity. What justifies the use of a guideline is evidence that its use is related to clinical outcome, and what justifies the use of a test or device is the same. No evidence suggests that pharmacogenetic tests that detect alleles of genes that code for metabolic enzymes of high or low activity bear any relationship to clinical outcome. High or low activity changes from day to day and coadministered drug to coadministered drug. Be informed or be misled.