In the present study, for evaluating the implications for efficacy of the different dosage regimens of UST for IBD, we constructed a PK/PD model, and simulated the sequential changes of therapeutic effect following various dosage regimens.
The predicted sequential changes of serum UST concentration, CDAI, and partial Mayo score after repeated administration of UST using the PK/PD model were in good agreement with the observed values (Figs. 2–5). Furthermore, the results of cross-validation revealed the validity of the model generalization (Fig. 6). Therefore, it is suggested that the PK/PD model is capable of predicting sequential changes in CDAI and partial Mayo score following administration of UST.
The data on the sequential changes of therapeutic effect after administration of UST was also reported in the induction treatment trials conducted prior to the maintenance treatment trials [7, 11]. However, in the induction treatment trials, there was less data that measured therapeutic effect at 0, 3, 6, and 8 weeks after a single intravenous administration than at 13 time points from 0 to 52 weeks in the maintenance treatment trials. Therefore, in the present study, the sequential changes of therapeutic effect were extracted from the maintenance treatment trials.
The estimated KUST and Kelse values for CD were approximately 160 and 93.6 times greater, respectively, than those for UC. It was considered that PD parameters are affected by the scale of therapeutic effect, as the scale of CDAI was greater than that of the partial Mayo score. To evaluate the contributions of the activities of UST to the partial Mayo score and CDAI, we calculated the KUST/Kelse ratios, which were 14.7 for CD and 8.64 for UC patients, thus the inflammation suppressed by the action of UST was greater than that caused by factors not affected by the action of UST in the patients in this analysis. However, the patients participating in the maintenance treatment trials were those who obtained the therapeutic effect of UST in the induction treatment trials, thus if the patients who did not obtain the therapeutic effect after the induction treatment trials were included in this analysis, KUST/Kelse ratios might have been smaller. Therefore, in the present study, there is a limitation that the estimated PD parameters are values for patients who obtained the therapeutic effect after the induction therapy.
The validity of the initial intravenous administration of UST for IBD was evaluated (Fig. 7). The Emin of CDAI and partial Mayo score was the smallest in Regimen 1, and clinical remission was obtained only in Regimen 1 for both diseases. Therefore, it was shown by the model-informed approach that the initial intravenous administration is important for UST instead of the initial subcutaneous administration, such as in Regimen 2 or Regimen 3, in order to induce remission. Our results are considered reasonable, as the clinical trials found that the clinical response rate and the remission rate of inflammation markers after initial intravenous administration of UST at 4.5mg/kg were superior to those after initial subcutaneous administration at 90mg [18, 19].
The different dose intervals and administration routes (Regimens 1, 4, 5, 6, 7, 8) were evaluated (Fig. 8). The results showed that Regimen 4 remained at a slightly lower value than Regimen 1, though there was almost no difference in the fluctuation range of CDAI (Regimen 1: 17, Regimen 4: 11) and partial Mayo score (Regimen 1: 0.17, Regimen 4: 0.08), and a stable therapeutic effect was obtained. The dose interval for subcutaneous administration in the maintenance therapy has been recommended at every 12 weeks (Regimen 1) in PMDA and EMA [8, 9], while the FDA recommends every 8 weeks (Regimen 4) [10]. In CD patients, previous studies showed that the remission rate at Week 44 tended to be slightly lower in Regimen 4 than in Regimen 1 (48.8% and 53.1%, respectively), though a clear difference was not found [6, 18]. In addition, in UC patients of the maintenance treatment trial, it was reported that the clinical remission rate at Week 44 was similar between Regimen 1 and Regimen 4 [7, 20]. On the other hand, the predicted value of Regimen 5 was lower than of Regimen 1 and 4. Some real-word studies have reported the experience with ustekinumab intensification from 90mg q8W to q4W and even to q3W [21–25]. Recently, the University of Chicago group reported the effectiveness of ustekinumab dose interval shortening from 90mg q8W to q4W in 51 patients with a Harvey–Bradshaw score > 4 [26]. They showed that dose escalation resulted in improvement in clinical indices of disease activity. There are important points to consider with regard to the present findings that the safety of q4W was not examined in this study, thus further study is needed.
The sequential changes of CDAI and partial Mayo score were lower in intravenous administration than in subcutaneous administration (Fig. 8). It has been reported that the re-induction with intravenous administration after secondary loss of response is considered an important rescue treatment option in patients with refractory CD [27, 28]. Therefore, it was suggested that intravenous administration will present a higher therapeutic effect than subcutaneous administration during the maintenance therapy. However, as far as we know, there is no study reporting the actual efficacy and safety of repeated intravenous administration of UST for IBD during the maintenance therapy, thus it is necessary to examine the efficacy and safety of repeated intravenous administration.
The amount of the initial intravenous administration of UST for IBD was also evaluated (Supplementary Table 1, Supplementary Fig. 1). The Emin of CDAI and partial Mayo score in each dosage regimen are shown in Supplementary Table 2. The Emin was the smallest in 6mg/kg, and clinical remission was obtained only in 6mg/kg and 3mg/kg for CD and 6mg/kg for UC. Therefore, it was shown that the dose of 6mg/kg is important for the induction of remission after the initial intravenous administration. Our results are considered reasonable because of the results of the following clinical trials. In CD patients, the clinical response rate at Week 6 was compared among the 1mg/kg group, 3mg/kg group, and 6mg/kg group of initial intravenous administration [18, 29]. The results were 23.5% (31/132 patients) in the placebo group, 36.6% (48/131 patients) in the 1mg/kg group, 34.1% (45/132 patients) in the 3mg/kg group, and 39.7% (52/131 cases) in the 6mg/kg group [18, 29]. Furthermore, in a phase III clinical trial comparing the secondary endpoint of therapeutic effect rate at Week 8 between the 6mg/kg group and the 130mg (2mg/kg at 65kg) group for UC patients, the results were 61.8% (199/322 patients) in the 6mg/kg group and 51.3% (164/320 patients) in the 130mg group [7].
In conclusion, his study showed that the presented PK/PD model is capable of predicting sequential changes of therapeutic effect of UST for IBD, and the implications for efficacy of the different dosage regimens could be evaluated by simulation with various dosage regimens. These findings are useful for model-informed precision dosing of individual IBD patients by predicting the serum UST concentration and therapeutic effect in individual patients [30].