Results
In total, 315 patients (671 serum samples) aged from 6 to 73 years were included (table 1). The average age was 24.9 years (CI: 23.3, 26.4). Information on blood sampling time adhering to the recommended time interval (4-8 hrs) was available for 47% of the patients. For these patients, the average blood sampling time post-dose was 5.4 hrs (CI: 5.2, 5.6 hrs). Information about dose was available for 87% of the TDM samples. As many as 21.5% of the patients had recorded at least one undetectable serum sample. The proportions of CYP2D6 UMs, NMs, IMs and PMs were 1.2%, 52%, 37% and 10%, respectively, while the proportion of CYP2C19*2 allele carriers was 35% (table 1).
Figure 1 shows the absolute serum concentration and daily dose of atomoxetine for patients subgrouped to the various CYP2D6 andCYP2C19 genotypes. The serum concentration of atomoxetine was systematically higher for PMs (p<0.001) and IMs (p<0.048) regardless of carrying the CYP2C19*2 allele variant but were not significantly altered in UMs (p>0.9), when compared to NMs not carrying the CYP2C19*2 allele variant (figure 1A). Despite this, insignificant differences in dosing were registered for PMs and IMs when compared to NMs or UMs (figure 1B; p>0.05).
In the studied population, a high variability of atomoxetine C D-1 ratio was observed (>85-fold difference; 10-90 percentile: 0.88-75 nM mg-1). As CYP2D6 NMs and UMs had no statistical differences in absolute concentrations (p=0.90) and C D-1 ratio of atomoxetine (p=0.77), these groups were merged for the remaining analyses. Compared to CYP2D6 NM/UMs, a significantly increased atomoxetine C D‑1 ratio was observed for CYP2D6 IMs and PMs (IM: +95%, p<0.001; PM: +863%, p<0.001; table 2), respectively. A similar effect was observed when CYP2D6 UMs were excluded (model 5, supplementary table 1). Moderately higher C D-1 ratio of atomoxetine was observed inCYP2C19*2 allele carriers versus noncarriers (+54%, p=0.011; table 2). A statistically significant decrease in C D-1 ratio and increase in daily dose were observed with increasing age, respectively (C D-1 ratio: -0.020 nM/mg per year, p<0.001; Dose: +0.83 mg per year, p<0.001; table 2). Furthermore, significantly increased absolute atomoxetine concentrations were observed for CYP2D6 IMs and PMs when comparing to NM/UMs (IMs: +65%, p=0.006; PMs: +570%, p<0.001; table 2), respectively. Statistically higher absolute concentration of atomoxetine was observed for CYP2C19*2 allele carriers versus non-carriers (+49%, p=0.022; table 2). Moreover, daily dose was statistically lower for CYP2D6 IM and PMs compared to NM/UMs (IMs: -16%, p=0.022; PMs: -22%, p=0.029), but not betweenCYP2C19*2 allele carriers vs non-carriers (p=0.78; table 2). In addition, inclusion of interaction terms to the mixed model such as age x CYP2D6 genotype, age x CYP2C19*2 allele carriers or CYP2D6 genotype x CYP2C19*2 allele carriers turned out to be insignificant (model 1-4 supplementary table 1). Sensitivity analysis, only including blood samples taken 4-8 hrs post-dose, revealed similar quantitative effects on atomoxetine C D-1 ratio of the various CYP2D6 genotype subgroups, CYP2C19*2 allele carriers and age (model 6, supplementary table 1).
For CYP2D6 UMs (three out of four; 75%) and PMs (12.5%), the proportions of patients with at least one undetectable serum sample were higher and lower when compared to the other genotype subgroups (NMs, 26.4%; IMs, 15.5%; p=0.0071), respectively. In the mixed logistic regression model, including all serum samples, a lower odds ratio (OR) of having an undetectable serum sample was observed for CYP2D6 IMs (OR: 0.50, p=0.037) and PMs (OR: 0.34, p=0.068), compared to CYP2D6 UM/NMs (table 2). No effect on the risk of having an undetectable serum sample was observed between CYP2C19*2 carriers vs non-carriers (p=0.12; table 2). Interestingly, increasing age was associated with an increased OR of undetectable concentrations by +3% per year (p=0.002; table 2).