A mechanism-based population K-PD model for long-term testosterone inhibition in prostate cancer patients under intermittent androgen deprivation therapy
Joost DeJongh (1) , Maurice Ahsman (1) and Nelleke Snelder (1)
(1) LAP&P consultants, The Netherlands
Objectives: Intermittent androgen deprivation therapy (iADT) is a treatment option for selected subpopulations of prostate cancer (PCa) patients that can prevent or delay disease progression and development of castration resistant prostate cancer (CRPCa). It can also reduce risk and severity of side effects associated with chemical castration in PCa patients. The aim of this investigation was to derive a mechanism-based population model for testosterone in PCa patients undergoing long term iADT during active treatment and recovery phases, to predict PSA response in stable and relapsing patients.
Methods: One of the earliest comprehensively documented clinical trials on iADT was reported more than a decade ago [1]. In this study, PCa patients showing PSA relapse after previous radiation therapy started iADT treatment under a pre-defined titration scheme, with a 32 week treatment cycle followed by an off-drug period. Repeated PSA recurrence above a threshold re-initiated active treatment, until PSA returned to below the target level. 72 patients underwent one to five active treatment cycles during a period of up to six years, in which testosterone and PSA levels were monitored continuously every three months. A K-PD model for long-term testosterone inhibition and recovery via GNRH-receptor downregulation was developed, based on a recently reported testosterone-PSA model for Leuproline [2], under Monte-Carlo importance sampling (IMPMAP) in NONMEM.
Results: The model could successfully describe testosterone inhibition during and after each active treatment cycle on population and individual patient level. Precision of model predicted testosterone response decreased approximately three-fold after the first active cycle, but remained constant during subsequent cycles. Model accuracy was equally adequate during each cycle. Some disease progression-related aspects of long-term iADT in the population, originally reported as clinical observations [1], were quantified from the data by model inference: Pre-treatment testosterone concentration for the population at baseline was 11.5 nmol/L and a first-order constant of 0.097 h-1 was derived for long-term decrease of endogenous testosterone production, visible after treatment recovery at the end of each cycle. However, the median testosterone nadir during active treatment remained constant (0.13 nmol/L) for patients remaining in the trial. A non-normal distribution of individual estimates (EBE’s) for testosterone production decrease was observed and a negative Box-Cox parameter for this could be identified. In addition, the return rate of testosterone concentrations after the end of active treatment cycles was approximately 60% lower after cycle 2-5, compared to the wash-out phase after cycle 1.
For a small sub-population of 12 patients designated as PSA reIapsers, slightly higher testosterone levels at the nadir of cycle 1 and 2 were observed compared to stable patients, which indicates that the individual nadir estimates may be predictive for development of long-term PCa resistance to castration, but a bi-modal distribution for this could not be derived using the $MIX option in NONMEM. From model-derived parameters, obtained by fitting to the data reported from cycle 1 and 2, the response during subsequent active cycles 3-5 could be adequately predicted.
Conclusions: A mechanism based K-PD population was developed that can describe and predict the long-term testosterone response under iADT in this patient population, including some typical aspects of receptor down regulation and post-recovery decay in testosterone production.
In future, this model may be used to explain and predict the long-term disease progression as reflected by PSA response in the same PCa patient population.
References:
[1] Bruchovski N et al. Cancer, 109 (5): 858-67, 2007
[2] Snelder et al. Br J Clin Pharmacol., https://doi.org/10.1111/bcp.13891 , Feb 7, 2019