AI/ML-driven simulation of the skin retention of opioid analgesics following topical administration in normal skin and cutaneous lesions.

Topical drug administration in the treatment of skin disorders is known to have high efficacy with a favorable safety profile in clinical settings. Disposition of the active pharmaceutical ingredient (API) in the skin layers plays a critical role for the dose regimen calculator required for desired therapy efficiency. Experimental measurements of the API skin concentration are complicated and inaccurate and can be replaced with physiologically based pharmacokinetic modeling predicting local and systemic drug exposure. Opioid analgesics are commonly used in treating severe pain and it is frequently associated with putative adverse effects. Topical applications of opioids were shown to have high efficacy with a favorable safety profile in clinical settings. Clinical investigations of transdermal drug formulations, especially in patients with skin lesions, are often hindered due to the patient intolerance, high cost of the studies, and ethical aspects. BIOiSIM, a hybrid artificial intelligence computational platform integrating machine learning optimization with mechanistic modeling was implemented to accurately simulate in vivo plasma concentration profile and predict partition coefficients and API skin retention for two opioid analgesics in healthy skin and simulated skin lesions. Publicly available drug parameters and physiological data were used for the model building. The platform successfully predicted API systemic exposure and skin concentration following the epicutaneous application of central opioid agonist buprenorphine and peripheral opioid agonist oxycodone in healthy human skin and simulated subjects impaired skin. Experimentally observed plasma concentration for both drugs demonstrated validity of the in-silico methods. The results indicated that buprenorphine permeates rapidly through the stratum corneum and partitions into the dermis, where its concentration slowly declines as the reservoir in the compound vehicle is depleted. Thus, the impact of lesions is not as significant. In contrast, the hydrophilic oxycodone, where the rate-limiting stratum corneum layer prevents a significant portion of the compound from permeating deeper in healthy patients. Hence, concentration of oxycodone sharply declines over the first hours of exposure in deeper indicating poorer accumulation capacity in the skin in the presence of lesions. An AI/ML-driven PBPK model was applied for accurate simulation skin concentrations of two opioid analgesics - buprenorphine and oxycodone. The results have confirmed that both these drugs have good skin retention and can stimulate opioid receptors located in skin. The results suggest that buprenorphine showing substantial skin retention capacity can be applied topically for the treatment of skin disorders in in patients.