Alternative splicing with CYP3A4*22: Implications for the steroid-tacrolimus drug interaction

Abstract

We read with interest the recent study by Saqr et al. (2024) examining steroid-tacrolimus drug interactions in 2462 kidney transplant recipients.¹ Their findings provide important insights into the complex interplay between CYP3A4 and CYP3A5 genotypes and drug–drug interactions involving enzyme induction in the setting of immunosuppressive therapy. An unexpected finding was that patients with three or four loss of function (LOF) alleles in these genes showed a similar proportional increase in tacrolimus clearance with steroids relative to no LOF alleles (10.3% vs. 11.8%), whereas those with 1- and 2-LOF alleles showed a minor increase (2.6% and 5%, respectively). Consideration of the genetic groupings, and the functional consequences of the alleles, may provide mechanistic insight into this observation.

Due to the prevalence of LOF alleles in CYP3A5, the 2-LOF group in Saqr et al. represents 67.5% of participants, 99% of whom have two LOF alleles in this gene (assigned as 2-LOFa, Table 1). The 3/4-LOF group represents people that additionally carry CYP3A4*22, predominantly as a single copy (3-LOF group, Table 1; 98%). The subgroups with less frequent CYP3A5 and CYP3A4 genotypes (2-LOFb and 4-LOF in Table 1) account for a small proportion of participants that are unlikely to impact the median clearance estimates for the larger groups.

The unexpected clearance patterns with steroid induction may be explained by examining how CYP3A4*22 differs mechanistically from other LOF alleles. CYP3A5*6 and *7 prevent production of functional protein through coding region changes that eliminate enzyme activity, while *3 occurs in an intronic splice site, which causes premature termination.² In contrast, CYP3A4*22 occurs in intron 6 and influences the proportion of a nonfunctional alternatively spliced transcript, which retains the intron; however, this is tissue specific, occurring in the liver but not the small intestine.³ Wang and Sadee quantified both this transcript and functional CYP3A4 mRNA in tissue samples.³ Only 5% of the alternatively spliced transcript was present in liver samples from people with the CYP3A4*1/*1 genotype, while those with at least one *22 allele showed approximately 19%.³ Notably, in small intestine samples, there was no difference in the spliced transcript proportions between genotype groups. This suggests *22 might exclusively impact the clearance of drugs metabolized by CYP3A4, possibly without impacting intestinal absorption.

The functional consequences of CYP3A4*22 may involve additional mechanisms. Recent work identified rs62471956, a variant in complete linkage disequilibrium with *22, which affects CYP3A4 expression through altered enhancer interactions.⁴ Despite a complex picture, data from human liver samples consistently show hepatic CYP3A4 mRNA, protein quantity and enzyme activity are reduced in CYP3A4*22/rs62471956 carriers.^5-7 Unlike CYP3A5*3, *6 and *7, the data do not support a complete LOF scenario, aligning with the observed reduction in median clearance without steroids in the 3/4-LOF group relative to 2-LOF (~32%).

We propose that in CYP3A4*22 carriers, steroid induction impacts the proportion of the spliced transcript in the liver. This is in line with the observed reduction in median clearance with steroids between the 2-LOF and 3/4-LOF groups (~27%). The above reasoning may provide a mechanistic explanation for the finding of greater relative steroid inducibility in the 3/4-LOF group. Future studies should investigate how steroids modulate alternative splicing affected by CYP3A4*22, with consideration for possible tissue-specific differences.

These findings have direct implications for clinical practice. While most patients (67.5%) fall into category 2-LOFa and show minimal steroid-related changes in tacrolimus clearance, patients with other genotype combinations require careful monitoring. The more pronounced impact of steroid induction for people in these categories has implications for adjusting tacrolimus doses, especially when steroids are stopped or started, highlighting the value of pharmacogenetic information for managing drug interactions involving induction.

All authors contributed to conceptualisation and review/approval of the final manuscript. Jana Stojanova prepared the original draft.

The authors declare no conflicts of interest.