We previously demonstrated that CYP2C19-guided voriconazole dosing significantly reduces the proportion of patients with subtherapeutic trough concentrations in those undergoing allogeneic HCT – this is particularly true for CYP2C19 RMs and UMs (12). However, a substantial proportion of patients, especially CYP2C19 NMs, are still subtherapeutic at the initial steady state level (< 1 mg/L) and may require higher up-front doses. To our knowledge, this is the first study to investigate genetic and clinical predictors of voriconazole trough concentrations in patients receiving CYP2C19-guided dosing. Apart from CYP2C19 phenotype, we identified that letermovir use consistently resulted in lower voriconazole concentrations. We also identified potential associations with ABCG2 and the CYP2C TG haplotype that warrant further study.
Substantial interpatient pharmacokinetic variability exists in voriconazole trough concentrations when using flat or weight-based dosing. Numerous prior reports demonstrate that voriconazole trough concentrations vary by CYP2C19 phenotypes (7, 8). Original reports by Patel et al.(12) and Hicks et al.(13) demonstrated that CYP2C19-guided dosing in adults improves the attainment of target trough concentrations. Prior to completion of these studies, CPIC guidelines were also published summarizing the evidence supporting the association between CYP2C19 phenotype and voriconazole pharmacokinetics and response, and recommending that alternative antifungals be considered in CYP2C19 PMs and RM/UMs (8). While increasing the voriconazole dose in CYP2C19 RM/UMs significantly increases the overall proportion of patients achieving target concentrations, a substantial number of non-RM/UM patients remain subtherapeutic; however, there are no prior studies investigating contributing factors in this subgroup.
Prior candidate gene studies have evaluated other potential SNPs associated with voriconazole concentrations in patients receiving standard dosing. A study of 177 Thai patients with IFIs receiving voriconazole treatment demonstrated no significant association between SNPs in CYP3A4, ABCB1, and FMO3 with voriconazole concentrations (17). Another study of 68 pediatric Chinese patients receiving voriconazole treatment similarly failed to demonstrate an association between CYP3A4 and voriconazole concentrations (18). In a larger study of 233 pediatric patients receiving voriconazole treatment, SNPs in SLCO1B3, ABCG2, and ABCB1 were significantly associated with trough concentrations (19). A study of 36 pediatric patients receiving voriconazole treatment demonstrated that CYP2C19, CYP3A4, ABCC2, and ABCG2 were associated with voriconazole concentrations (20). However, the minor allele frequency for CYP3A4 was 5%, and the study sample size was small. A systematic review and meta-analysis including 203 patients and 754 voriconazole trough concentrations from six studies demonstrated that voriconazole trough concentrations were independently influenced by age, dose, C-reactive protein level, CYP2C19 genotype, and CYP3A4 genotype (21). Prior studies suggesting that CYP3A4 genotype is independently associated with voriconazole concentrations are small (low minor allele frequency of the CYP3A4*22 allele) and include potential confounders such as proton pump inhibitor use (22–24). Like some prior studies, we identified lack of an association between certain genes, including CYP2C9, CYP3A4/5, and ABCB1, with voriconazole concentrations but did identify a potential signal with ABCG2, suggesting that NF patients have lower trough concentrations compared to PF or DF patients. However, this finding may have little clinical significance given ABCG2 was not associated with odds of having subtherapeutic concentrations.
To our knowledge, this is the first report to investigate the association between the novel CYP2C haplotype and voriconazole concentrations. Braten et al. identified three new haplotypes of the CYP2C locus (TG, TA, and CG). In a study of 875 previously genotyped escitalopram-treated patients, the CYP2C haplotype was significantly associated with ultrarapid metabolism of escitalopram, whereby the serum concentrations of escitalopram in homozygous CYP2C:TG and CYP2C19*17 carriers were 25 and 17% lower compared with CG and TA carriers (14). Subsequently, Braten et al. also demonstrated the same haplotype was significantly associated with sertraline exposure (15). Given the similarities in hepatic metabolism between these selective serotonin reuptake inhibitors and voriconazole, we investigated the association of the same haplotypes on voriconazole trough concentrations. In the univariate analysis of patients receiving voriconazole 200 mg twice daily (non-RM/UM patients), TG carriers had significantly lower mean voriconazole trough concentrations compared to non-TG carriers (1.1 vs 1.5; p = 0.019); however, this effect was not retained in the multivariable model when accounting for other variables. Further, there was no association between the CYP2C haplotype and odds of having subtherapeutic concentrations. We also conducted an additional analysis limited to CYP2C19 NMs, but there was no association between the novel haplotype and mean voriconazole concentration or odds of subtherapeutic concentration (data not shown), which may be limited by the small sample size (n = 45 CYP2C19 NMs with haplotype information). Larger studies are warranted to determine the effect of the novel CYP2C haplotype on voriconazole concentrations and potential clinical utility when combined with CYP2C19 genotype information.
In addition to genetic factors, we also evaluated the impact of clinical factors on voriconazole concentrations. Race, sex, disease, and BMI were not significantly associated with voriconazole concentrations; however, presence of letermovir resulted in about 4-fold higher probability of having subtherapeutic voriconazole concentrations when analyzed across all patients and in those limited to receiving 200 mg twice daily. Letermovir is a CYP2C19 inducer. Two small studies in allogeneic hematopoietic cell transplant recipients also showed that letermovir significantly reduces voriconazole concentrations (25, 26). A study of healthy subjects who received concomitant letermovir and voriconazole showed that voriconazole AUC and maximum serum concentration were reduced by 44% and 39%, respectively (27). Based on these collective findings and our report, voriconazole trough levels should be checked when initiating letermovir in transplant patients and adjusted accordingly. Further studies are needed to determine whether these patients should have upfront voriconazole dose escalations when initiating letermovir.
It is important to recognize the limitations of this study, including its retrospective nature, small sample sizes for subgroup analyses, and unknown clinical relevance of pharmacokinetic findings. Although a substantial number of patients underwent clinical CYP2C19 genotyping, fewer patients had voriconazole trough levels and banked DNA for retrospective genotyping of candidate genes. Further, subgroup analyses were performed on those receiving 200 mg twice daily since one of the objectives was to identify candidate genes associated with mean voriconazole concentration and/or subtherapeutic levels in non-RM/UM patients. Other patient-related factors were not assessed such as compliance, however, most patients were still inpatient at the time of initial voriconazole steady state trough collection. Lastly, we did not include data on incidence of IFIs or voriconazole-related side effects. In our prior study, we reported fewer IFIs with CYP2C19-guided dosing compared to historical control data, but no such evaluation was performed in this study as the objective was to identify other clinical and genetic factors associated with voriconazole pharmacokinetics only.
In conclusion, while CYP2C19 genotype-guided dosing improves the ability to achieve voriconazole target trough concentrations, many patients are still subtherapeutic. In the first study to evaluate clinical and genetic predictors of voriconazole concentration in patients already receiving CYP2C19-guided dosing, we identified that concomitant letermovir, ABCG2, and possibly the novel CYP2C haplotype may further modulate mean voriconazole trough concentrations. If validated in larger independent cohorts, these clinical and genetic variables can be used to identify the most appropriate up-front prophylactic dose (e.g., 200 mg vs 300 mg twice daily), followed by therapeutic drug monitoring to further refine dosing. Additional studies are needed in those receiving treatment dosing. Given the relationship between voriconazole trough concentrations and clinical efficacy, it is imperative that personalized approaches to dosing are used to improve drug efficacy, especially in high-risk immunocompromised patients.