Pharmacogenomic Study—A Pilot Study of the Effect of Pharmacogenomic Phenotypes on the Adequate Dosing of Verapamil for Migraine Prevention

OBJECTIVE Using individual pharmacogenomic phenotypes to investigate the factors affecting the e�cacy and tolerability of verapamil for migraine prevention. BACKGROUND Verapamil has a wide range of dosing in headache disorders without reliable tools to predict the optimal doses for an individual. METHODS This is a retrospective chart review examining adults with existing pharmacogenomic reports at Mayo Clinic who had used verapamil for migraine. Effects of six cytochrome P450 phenotypes on the minimum effective and maximum tolerable verapamil doses for migraine prevention were assessed. RESULTS The mean minimum effective and maximum tolerable doses for the 33 patients in �nal analysis were 178.2(20–320) mg and 227.9(20–480) mg. A variety of CYP2C9, CYP2D6, and CYP3A5 phenotypes were found, without signi�cant association with the verapamil doses after adjusting for age, sex, body mass index, and smoking status.


Introduction
Migraine affects 12 to 15.9% of adults in U.S. and is a major leading cause of years lived with disability globally (1)(2)(3).Despite being an important public health concern, the preventive treatment for migraine remains a trial-and-error process.Lack of e cacy and side effects each account for around 40% of the reasons leading to discontinuation of migraine preventive medications (4).
Verapamil has long been used for migraine and cluster headache, with a wide range of optimal total daily dose, from 120 to 1 200 mg (5-7), for individual patients.Adverse events including ankle edema, bradycardia and heart block may occur, especially at higher doses.It would be bene cial if the adequate individual dosing can be predicted before undesirable side effects occur.Genetic variants known to interact with migraine medications will result in different therapeutic responses (8), and a recent study from our group identi ed several single nucleotide polymorphisms that were associated with verapamil's therapeutic responses in migraine (9).Since 2012, pharmacogenomic consultations became available in our institution for clinical use, though it has not been widely implemented in the headache practice.Existed literature suggests cytochromes P450 (CYP)1A2, CYP2B6, CYP2C8, CYP2C9, CYP2C18, CYP2D6, CYP2E1, CYP3A4, CYP3A5, and Pglycoprotein (P-gp) are involved in the metabolism or transportation of verapamil (10)(11)(12)(13)(14)(15).In theory, less active metabolic enzymes and transporter would lead to a lower minimum effective dose and/or lower maximum tolerable dose.
In this pilot, exploratory study, we aim to investigate the association between the pharmacogenomic pro le and the minimum effective and maximum tolerable doses of verapamil for patients with migraine to tackle the two major reasons leading to discontinuation of migraine preventive medications-lack of e cacy and intolerable side effects.

Methods
This is a retrospective chart review study approved by Mayo Clinic Institutional Review Board.We identi ed all Mayo Clinic patients, including Rochester, Scottsdale, Jacksonville, and all Mayo Clinic Health System sites, that have a pharmacogenomic report in the electronic medical record (EMR) with a diagnosis of migraine and have used verapamil.The diagnosis of migraine was identi ed using International Classi cation of Diseases (ICD) codes (ICD-9 346, ICD-10 G43) and con rmed by detailed chart review.Patients were excluded if they were less than 18 years old when initiating verapamil, without documented therapeutic responses to verapamil and its responding doses for migraine prevention, or without Minnesota Research Authorization (only for patients in Minnesota).At the time of data collection, available pharmacogenomic data in the EMR that are relevant to verapamil include CYP1A2, CYP2B6, CYP2C9, CYP2D6, CYP3A4 and CYP3A5.Individual phenotypes of these six genes, as well as demographic information including age, sex, race, smoking status while using verapamil, and body mass index (BMI) were collected through chart review.In pharmacogenomic study, phenotype indicates how a patient's individual genotype changes the innate enzyme activities.For example, a patient with a genotype of CYP2C9*1/*3 has decreased CYP2C9 enzyme activity that is classi ed as an intermediate metabolizer, when compared to CYP2C9*1/*1, normal metabolizer.Our EMR pharmacogenomic report classi es the enzyme activity from low to high as poor-, intermediate-, normal-, rapid-, and ultrarapidmetabolizers.Individual's therapeutic response to verapamil was reviewed from the date of initiating verapamil through May 31st, 2021.The minimum effective dose was de ned as the smallest total daily dose that has ever been documented as subjectively helpful for migraine prevention, and the highest verapamil total daily dose ever used in an individual without causing any intolerable side effects was de ned as the maximum tolerable dose.We used the chi-square test to evaluate if the doses of verapamil were associated with different demographic factors (age, sex, smoking status).Kruskal Wallis analysis of variance was performed to detect if speci c phenotypes within each CYP enzyme group affected responses to verapamil.Demographic factors were adjusted by linear regression.Enzyme classes that have dominant phenotypes are presented in descriptive fashion.Shapiro-Wilk test was conducted to examine the normal distribution of the verapamil doses.

Results
The initial search resulted in 74 patients.After excluded for reasons listed in Fig. 1, 33 patients were included in the nal analysis.The mean age was 51 years old with a mean BMI of 28 kg/m 2 .79% of patients were female and 97% were white.Except two people with unknown smoking status, 90% were non-smokers (Table 1).We were able to collect complete data for CYP1A2, CYP2C9, CYP2D6, CYP3A4 and CYP3A5, though only 8 patients had CYP2B6 phenotype information.Figure 2 shows the overall distribution of verapamil dose used in this dataset.Overall, the mean minimum effective and maximum tolerable doses were 178.2 ± 73.4 mg (range 20 to 320 mg) and 227.9 ± 97.7 mg (range 20 to 480 mg), and both were normally distributed.There were no signi cant association between demographic factors (age, sex, smoking, and BMI) and the minimum effective and maximum tolerable doses.
All, except one, were CYP1A2 rapid metabolizers.66.7% of patients carried CYP1A2*1F/*1F.For CYP2B6, six of eight patients were intermediate metabolizers and they carried CYP2B6*1/*6 genotype (75%).For CYP 3A4, 31 patients were normal metabolizers with CYP3A4*1/*1 genotype (94.0%).For CYP2C9, CYP2D6, and CYP3A5, there was a variety of phenotypes ranging from poor-to-intermediate metabolizers to normal metabolizers.There were ve genotypes of CYP2C9 in our cohort, including 57% of patients with CYP2C9*1/*1, and 21.2% with CYP2C9*1/*2.Genotypes of CYP2D6 were rather complex, which included potentially sixteen different genotypes and 21.2% carried CYP2D6*1/*2A.There were two genotypes of CYP3A5, 78.8% were CYP3A5*3/*3 and 21.2% were CYP3A5*1/*3.Phenotypes within each CYP enzyme groups and their minimum effective and maximum tolerable doses were listed in Table 1.The allele frequencies of the six CYP enzymes in our cohort were listed in Table 2. Overall, we did not nd any signi cant association between the different phenotypes of CYP2C9, CYP2D6, and CYP3A5 and the minimum effective and maximum tolerable doses.The results remain non-signi cant for these three groups after adjusting for age, sex, and smoking status.

Discussion
In our cohort, a wide range of the minimum effective (range 20 to 320 mg) and maximum tolerable (range 20 to 480 mg) doses were reported in patients with migraine.The one patient that used verapamil 20 mg reported marked reduction in headache frequency and severity and was unable to tolerate a higher dose due to lightheadedness, nausea, heart racing, and insomnia.For the six pharmacogenomic phenotypes currently available in EMR reports, most patients that used verapamil for migraine prevention were CYP1A2 rapid metabolizers, CYP2B6 intermediate metabolizers, and CYP3A4 normal metabolizers.There was a wide variety of CYP2C9, CYP2D6, and CYP3A5 phenotypes.
Overall, the distribution of different alleles of the CYP enzymes in our cohort is consistent with the American population data reported in the literature.In a meta-analysis (16), Zhou et al reported the haplotype frequencies of 176 alleles distributed over the 12 major CYP genes known to be the most impactful on drug metabolism.The authors gathered and analyzed whole genome and exome sequencing data from 56 945 unrelated individuals from ve major populations-Africans, admixed Americans, East Asians, Europeans, and South Asians.The allele frequencies of the leading alleles in our cohort, CYP1A2*1F, CYP2B6*1, CYP2C9*1, CYP2D6*1, CYP3A4*1, and CYP3A5*3 are similar to the frequencies reported from the analysis of the admixed American population as listed in Table 2 (ref. 16).
Although this pilot study did not identify any statistical signi cance, we cannot exclude the effect of pharmacogenomic phenotypes on the effective and tolerable doses of verapamil.We aimed to use existing pharmacogenomic reports in the EMR.However, as pharmacogenomic testing is not routinely used in headache care, this study was limited by the small sample size.Given the wide range of verapamil used for cluster headache, initially we also investigated the cluster headache population.However, within identi ed patients that had a pharmacogenomic report, we found only one patient that was really using verapamil for the prevention of cluster headache yet without documented therapeutic response.Therefore, our current report focused on migraine.Predicting the adequate doses of verapamil in patients with cluster headache would still be a future research direction, given verapamil is a rst-line medication for the prevention of cluster headache.Additionally, as pharmacogenomic testing is more frequently implemented for depression treatment, exploring the association between pharmacogenomic phenotypes and the therapeutic doses of antidepressants commonly used for migraine prevention, including amitriptyline, nortriptyline, and venlafaxine, might potentially yield to larger study sample sizes, though determining the therapeutic or tolerated doses used for migraine as opposed to depression could potentially be complex in some cases.Of note, CYP2D6 is listed in U.S. Food & Drug Administration label as an actionable pharmacogenomic biomarker for these three medications as the systemic concentrations can be altered by certain phenotypes of CYP2D6 and may require dose adjustment (17)(18)(19).For example, using tricyclic antidepressants such as amitriptyline or nortriptyline in CYP2D6 ultrarapid metabolizers may lead to suboptimal therapeutic response in treating depression since there might be increased metabolism of the tricyclic antidepressants to less-active compounds.If such use cannot be avoided, individuals may need to target a higher dose to reach an adequate therapeutic response.On the contrary, for CYP2D6 poor metabolizers, a 50% dose reduction should be considered to avoid side effects caused by higher-than-normal plasma concentration of active drugs if the use of tricyclic antidepressants is necessary (20).For venlafaxine, the prediction of therapeutic response and the occurrence of side effects by individual CYP2D6 metabolism activities is rather complex and no precise dose adjustment recommendation can be reached per current literature.CYP2D6 facilitates the metabolism of venlafaxine to active metabolite O-desmethylvenlafaxine, both are active antidepressants.It has been found that a high O-desmethylvenlafaxine/venlafaxine ratio is associated with better e cacy and less moderate to severe side effects (21).CYP2D6 ultrarapid metabolizer may increase the conversion from venlafaxine to O-desmethylvenlafaxine yet reduce the total sum of both venlafaxine and O-desmethylvenlafaxine.On the other hand, CYP2D6 intermediate or poor metabolizer may reduce the conversion from venlafaxine to O-desmethylvenlafaxine.It would be hard to predict the ratio change while adjust dose in altered metabolism between venlafaxine and O-desmethylvenlafaxine.A reduced e cacy and increased risk of side effects had been identi ed in CYP2D6 intermediate or poor metabolizers (22).Our cohort presents a variety of CYP2D6 phenotypes, 45.5% were normal, 27.3% were intermediate, 15.2% were intermediate-to-normal, and 12.1% were poor-to-intermediate metabolizers.
CYP3A4 is the most recognizable metabolic enzyme known to affect verapamil concentrations.CYP3A4 inhibitors such as erythromycin and ritonavir can increase verapamil concentration whereas CYP3A4 inducers such as rifampin could potentially cause an opposite effect (23).In our cohort, most subjects are CYP3A4 normal metabolizers, making it di cult to further examine the impact of CYP3A4 on verapamil.Smoking has been shown to decrease verapamil concentration, potentially by increasing CYP1A2 activity (24).In theory, smokers would require a higher verapamil dose to achieve therapeutic response.Our study did not nd such association, likely limited by the small sample size or the existence of other confounding enzymes that could be affected by smoking but not listed in the current report, such as CYP2E1 (25).Besides those enzymes currently in the EMR pharmacogenomic report, CYP2C8, CYP2C18 and P-gp can also affect the metabolism or transportation of verapamil and the association between the phenotypes of those enzymes and verapamil dosages could be further investigated.P-gp plays an important role in transporting verapamil out of the brain.This leads to the low verapamil concentration in the central nervous system despite that it can easily cross the blood-brain barrier as a lipophilic medication.It is possible that less active P-gp may increase the presence of verapamil in central nervous system and enhance its neurological effect as demonstrated by a study via P-gp inhibitor (26).Our previous study did not identify an association between P-gp variants in the e cacy of verapamil at a daily dose of 240 mg for migraine prevention (9), though whether P-gp phenotypes affect the therapeutic window of verapamil in responders could be further investigated.
In addition to the small sample size, other limitations of our study include recall bias of patients and the inability to account for all confounding factors that were not documented.Examining patients that already had an existing pharmacogenomic report could also create selection bias toward patients with certain pre-existing conditions.Initially we attempted to gather information on concurrent medications for this study; however, changes of the concurrent medications and the compliance of using those medications were not consistently provided or updated in the existed EMR records with each documented response to verapamil and so we were unable to include this information.As the use of certain medications could potentially affect the concentration of verapamil through affecting the CYP enzymes, future studies could consider including the effect of concurrent medications into the analysis.Regardless, our pilot study demonstrates a wide range of the minimum effective and maximum tolerable dosages of verapamil for migraine prevention in patients with a variety of pharmacogenomic phenotypes.Our study also highlights the importance of future prospective studies to investigate the effect of pharmacogenomic variants on the e cacy and safety of headache preventive medications, ultimately to facilitate individualized medicine in headache care.

Conclusion
Our pilot study demonstrated a wide range of minimum effective (range 20 to 320 mg) and maximum tolerable (range 20 to 480 mg) doses for migraine prevention in patients with various pharmacogenomic phenotypes.Most patients in our cohort were CYP1A2 rapid metabolizers, CYP2B6 intermediate metabolizers, and CYP3A4 normal metabolizers.There was a wide variety of CYP2C9, CYP2D6, and CYP3A5 phenotypes.The overall distribution of CYP enzyme alleles in our cohort is similar to the allele frequencies of the American population reported in the literature.As pharmacogenomic testing is not routinely used in headache care, retrospective studies could be limited by small sample sizes.Therefore, although this pilot study did not identify any statistical signi cance, we cannot exclude the effect of pharmacogenomic phenotypes on the effective and tolerable doses of verapamil.Further prospective studies investigating how pharmacogenomic factors affect the appropriate dosing of preventative medications in headache care are warranted to facilitate individualized medicine.

Figure 1 Study
Figure 1

Table 1 and
2 are available in the Supplementary Files section.