(T0930-10-65) Human Renal Clearance Prediction Using Monkey

Purpose: Prediction of human renal clearance prior to the first human dose is important for human PK and DDI predictions. Multiple approaches such as in silico, in vitro and PBPK models have been evaluated over the years. The published method⁽¹⁾ most widely applied is simple allometry using data collected in dogs corrected for protein binding (PB) and kidney blood flow (KBF) differences between human and dog. This method, on average, results in human renal clearance predictions within 2-fold of the actual values. However, dog data may not be available in some discovery programs due to issues such as emesis or the selection of monkey as the large animal non-rodent species. The purpose of this investigation was to assess if monkey renal clearance (mCLr) predicts human renal clearance (hCLr) and to test whether the concordance of mCLr to hCLr is better than dog due to potentially better transporter homology between monkeys and humans.

Methods: Twelve drugs with various physicochemical properties comprising of acids, bases, neutrals, or zwitterions that ranged from low to high hCLr were evaluated in cynomolgus monkey excretion experiments using non-radiolabeled material. The drug set also included molecules known to be cleared renally via different excretion mechanisms i.e. filtration, secretion, and/or reabsorption. Based on reported hCLr, five drugs possess high hCLr ( >1 ml/min/kg), four have moderate hCLr (0.1-1 ml/min/kg) and three are low hCLr ( < 0.1 ml/min/kg). Monkey excretion studies were carried out in a cross-over design using four male monkeys given 1-mg/kg IV infusion over five minutes using a pH 8 Captisol solution formulation. Plasma, urine, and feces were collected up to 72 hours. Creatinine levels were measured in plasma and urine. Drug concentrations in plasma and excreta were determined using protein precipitation and LC-MS/MS methods. PK parameters and CLr values were calculated for each monkey and summarized. The results obtained in monkey were compared to the previously reported⁽¹⁾ evaluations in dogs, using the same three animal scaling methods for the prediction of the hCLr. The first method assumed mCLr equals to hCLr. The second method applied simple allometry correcting for PB differences between monkey and human and the third method used simple allometry correcting both PB and KBF differences between monkey and human.

Results: Creatinine clearance was generally above glomerular flow rate (GFR) in monkeys, indicating potential active secretion in monkeys. Iohexol was used as a positive control for urine recovery, which is completely renally cleared and therefore should be fully recovered in urine; the %parent recovered from urine was highly variable and ranged from 12.1 to76.5%. Low drug recovery is proven to be a challenge in discovery excretion studies, particularly in monkey and when performed using non-radiolabeled material. Despite the variability amongst animals, the predicted mean hCLr was within 3-fold of published data using the three scaling methods. All twelve drugs showed a reasonable trend between predicted hCLr when categorized as high, mid and low CLr from monkey data using the three scaling methods. Simple allometry correcting for both PB and KBF differences between human and monkey provided the best result. In the high hCLr category ( >1ml/min/kg), 4 out of 5 drugs (metformin, sematilide, iohexol and furosemide) had predicted hCLr within 3-fold of the literature values. Ranitidine had predicted mean hCLr that was 5-fold different from the reported values. Among the four monkeys dosed with ranitidine, one animal had 12.6% drug recovered in urine and predicted hCLr was within 3-fold. The remaining three monkeys provided poor drug recovery (2.6-5.9%) in urine and the predicted hCLr was outside the 3-fold range. In this case, data from one animal that provided the highest drug recovery from urine predicted better than the mean data from all animals. In the mid hCLr category (0.1–1ml/min/kg), two out of four drugs (moxifloxacin and fluconazole) had predicted hCLr within 3-fold of the reported values. Losartan and carbamazepine had predicted hCLr greater than 3-fold of the reported values. When hCLr was low ( < 0.1 ml/min/kg), variability was high, and only 1 out of 3 drugs (sulfinpyrazone) had predicted hCLr within 3-fold of the reported value. The other two drugs (naproxen and dolasetron) provided greater than 3-fold of reported values. All drugs in this category were predicted to have low renal clearance.

Conclusion: We have demonstrated that predictions of hCLr using monkey data may be useful for categorical renal prediction, but not superior to quantitative predictions using dog data⁽¹⁾, likely due to high variability observed in monkey studies. Human renal clearance prediction using dog data and simple allometry correcting for PB and KBF remains the best prediction method when dog data is available.

References: (1) Paine, S.W., Menochet K., Denton R., McGinnity D.F., and Riley J., Drug Metab Dispos 39: 1008-1013, 2011