Discussion
The present study showed that CYP2D6 genotype has a great impact on atomoxetine exposure, where IM and PM patients have 1.9-fold and 9.6-fold higher serum concentrations of atomoxetine per mg dose compared to NMs/UMs. When adding CYP2C19 genotype as a factor of relevance for personalised atomoxetine dosing, CYP2C19*2 carriers had 1.5-fold higher serum concentrations of atomoxetine per mg dose than non-carriers regardless of CYP2D6 genotype. CYP2D6 PMs and IMs carrying theCYP2C19*2 allele should have doses equivalent to ≤10% and 35% of the normal starting dose, respectively, to target the same exposure as CYP2D6 NMs not carrying the CYP2C19*2 allele. These findings suggest that pre-emptive genotyping of CYP2D6/CYP2C19 should be performed to individualise atomoxetine dosing and minimise the risk for adverse effects.
According to CPIC dosing recommendations for atomoxetine12, dosing for CYP2D6 PMs should be reduced to approximately half of the approved dose in the product label information. Our study, however, suggests that an even lower dose (i.e., ≤10% of the starting dose) should be used in CYP2D6 PMs to avoid potential dose-dependent adverse effects. In practice, starting dose selection will be limited by available formulation and lowest available dose, with only oral solution forms suitable for the lowest doses. Further, our study suggests that for CYP2D6 IMs, a dose reduction equivalent to 50% of the recommended dose should be considered to target the same drug exposure as NMs, consistent with results from a previous study18. An even larger dose-reduction is warranted if the CYP2D6 PMs and IMs are carrying the CYP2C19*2allele (dose reduction of ≥90% and 65%, respectively), compared to CYP2D6 NMs not carrying the CYP2C19*2 allele. There was no significant interaction effect between the various CYP2D6 genotype subgroups and CYP2C19*2 allele carriers (p≥0.13), thus, indicating an additive effect on atomoxetine exposure ofCYP2C19*2 allele variant carriers regardless of CYP2D6 genotype. In addition, our study shows that the quantitative differences in atomoxetine exposure between the various CYP2C19 andCYP2D6 genotypes were similar regardless of age. In patients who are CYP2D6 UM, very limited data exists, but it is unlikely these patients would achieve adequate serum concentrations with standard atomoxetine dosing12,19. Indeed, we showed that CYP2D6 UMs had a higher probability of having at least one undetectable atomoxetine concentration (three out of four patients) compared to all other CYP2D6 genotypes, which indicates higher risk of undetectable levels and therapeutic failure. However, due to the low numbers of CYP2D6 UMs in the present study, no statistically significant differences were observed between NMs and UMs for C D-1 ratio, absolute concentrations and daily dose of atomoxetine, as well as risk of undetectable samples in the mixed model analyses. The CPIC’s recommendations are based on studies showing superior treatment response and compliance in CYP2D6 PMs versus NMs. It is therefore likely that the CYP2D6 PMs in the present study may have experienced improved therapeutic response but also increased dose-dependent side effects compared to the other CYP2D6 genotype subgroups. It is possible that this superior treatment response also resulted in better treatment adherence and fewer undetectable atomoxetine samples. Unfortunately, clinical data on therapeutic response were unavailable. A study based on pooled data from clinical trials conducted by Michelson et al.20 found that PMs showed a greater clinical improvement, measured by ADHDRS-score, and lower treatment drop-out than patients with normal CYP2D6 activity. Interestingly, but unsurprisingly, these patients also had higher serum levels of atomoxetine (Cmax sampled 1-hour post-dose, PMs vs NMs: approx. 3300 vs 650 nM on the same dose) and risk of adverse effects, including raised pulse and diastolic blood pressure, tremor, decreased appetite and insomnia. Trzepacz, PT et al. found a significant difference in weight, BMI and pulse in two open-label phase 3 studies but, conversely, no differences in clinical effect were observed between the CYP2D6 genotype subgroups9. Fijal, BJ et al. studied the effect of CYP2D6 activity and safety associated with atomoxetine treatment in adult patients over a period of 25 weeks at multiple international centers21. In addition to comparing NM and PM phenotypes, as in earlier studies, they distinguished between patients with UM and IM phenotypes. The study found an increased risk of adverse effects in patients with poor CYP2D6 activity, including reduced appetite, dry mouth, erectile dysfunction, hyperhidrosis, insomnia and urinary retention. Diastolic blood pressure and pulse were also significantly higher, while BMI was significantly lower. Patients with IM also had a higher risk of dry mouth and insomnia. Therefore, it is possible that the higher burden of adverse effects observed in CYP2D6 PMs may outweigh the potential improvement in therapeutic response compared to NMs.
A strength of the present study was the relatively large dataset, of not only adult users, but also children and adolescents that are previously less studied. We also had the possibility to correct the results for blood sampling time (trough concentrations, blood sampling 4-8 hrs post-dose). One obvious weakness of the present study was the lack of clinical information about each patient, including therapeutic effectiveness and treatment outcome, information on renal and hepatic status (e.g., Child-Pugh B and C category), as well as existing somatic comorbidities that would be of value. Concentrations are not corrected for patient weight as this data was not available. Patient weight may play a role in atomoxetine pharmacokinetics, especially in younger patients where weight-corrected dosing is recommended (i.e., patients under 70 kg in weight). Therefore, the patients’ age was included in the statistical models as age may account for weight indirectly. In addition, dose administration intervals were not known and may generate some noise in the estimates but are unlikely skewed between the variousCYP2D6 genotype subgroups and CYP2C19*2 allele carriers versus noncarriers. Information about steady-state could not be confirmed but steady-state is a prerequisite for TDM, and this is clearly communicated to doctors requesting TDM. It is therefore likely that the great majority of atomoxetine serum samples are at steady-state. A higher proportion of males (59%) was observed in the present study population which reflects the higher proportion of males diagnosed with ADHD as reported elsewhere1,2. Thus, the study population is likely representative and comparable to other studies.
In conclusion, both CYP2D6 and CYP2C19 genotypes have an impact on atomoxetine exposure regardless of age, where our real-world data suggest atomoxetine dose requirements to be equivalent to ≤10% and 35% of the starting dose in CYP2D6 IM and PM patients carrying theCYP2C19*2 allele variant versus CYP2D6 NM patients not carrying the CYP2C19*2 allele variant, respectively, to target the same atomoxetine exposure. These findings suggest that pre-emptive genotyping of CYP2D6 /CYP2C19 should be performed to individualize atomoxetine dosing and prevent adverse effects.