Prospective Validation of Machine Learning Algorithms for Absorption, Distribution, Metabolism, and Excretion Prediction: An Industrial Perspective.

Absorption, distribution, metabolism, and excretion (ADME), which collectively define the concentration profile of a drug at the site of action, are of critical importance to the success of a drug candidate. Recent advances in machine learning algorithms and the availability of larger proprietary as well as public ADME data sets have generated renewed interest within the academic and pharmaceutical science communities in predicting pharmacokinetic and physicochemical endpoints in early drug discovery. In this study, we collected 120 internal prospective data sets over 20 months across six ADME in vitro endpoints: human and rat liver microsomal stability, MDR1-MDCK efflux ratio, solubility, and human and rat plasma protein binding. A variety of machine learning algorithms in combination with different molecular representations were evaluated. Our results suggest that gradient boosting decision tree and deep learning models consistently outperformed random forest over time. We also observed better performance when models were retrained on a fixed schedule, and the more frequent retraining generally resulted in increased accuracy, while hyperparameters tuning only improved the prospective predictions marginally.

Discovery and Preclinical Characterization of BIIB091, a Reversible, Selective BTK Inhibitor for the Treatment of Multiple Sclerosis.

Multiple Sclerosis is a chronic autoimmune neurodegenerative disorder of the central nervous system (CNS) that is characterized by inflammation, demyelination, and axonal injury leading to permeant disability. In the early stage of MS, inflammation is the primary driver of the disease progression. There remains an unmet need to develop high efficacy therapies with superior safety profiles to prevent the inflammation processes leading to disability. Herein, we describe the discovery of BIIB091, a structurally distinct orthosteric ATP competitive, reversible inhibitor that binds the BTK protein in a DFG-in confirmation designed to sequester Tyr-551, an important phosphorylation site on BTK, into an inactive conformation with excellent affinity. Preclinical studies demonstrated BIB091 to be a high potency molecule with good drug-like properties and a safety/tolerability profile suitable for clinical development as a highly selective, reversible BTKi for treating autoimmune diseases such as MS.

Optimization of dose and route of administration of the P-glycoprotein inhibitor, valspodar (PSC-833) and the P-glycoprotein and breast cancer resistance protein dual-inhibitor, elacridar (GF120918) as dual infusion in rats.

Transporters can play a key role in the absorption, distribution, metabolism, and excretion of drugs. Understanding these contributions early in drug discovery allows for more accurate projection of the clinical pharmacokinetics. One method to assess the impact of transporters in vivo involves co-dosing specific inhibitors. The objective of the present study was to optimize the dose and route of administration of a P-glycoprotein (P-gp) inhibitor, valspodar (PSC833), and a dual P-gp/breast cancer resistance protein (BCRP) inhibitor, elacridar (GF120918), by assessing the transporters' impact on brain penetration and absorption. A dual-infusion strategy was implemented to allow for flexibility with dose formulation. The chemical inhibitor was dosed intravenously via the femoral artery, and a cassette of known substrates was infused via the jugular vein. Valspodar or elacridar was administered as 4.5-hour constant infusions over a range of doses. To assess the degree of inhibition, the resulting ratios of brain and plasma concentrations, Kp's, of the known substrates were compared to the vehicle control. These data demonstrated that doses greater than 0.9 mg/hr/kg valspodar and 8.9 mg/hr/kg elacridar were sufficient to inhibit P-gp- and BCRP-mediated efflux at the blood-brain barrier in rats without any tolerability issues. Confirmation of BBB restriction by efflux transporters in preclinical species allows for subsequent prediction in humans based upon the proteomic expression at rodent and human BBB. Overall, the approach can also be applied to inhibition of efflux at other tissues (gut absorption, liver clearance) or can be extended to other transporters of interest using alternate inhibitors.

Discovery of the Oxadiazine FRM-024: A Potent CNS-Penetrant Gamma Secretase Modulator.

The recent approval of aducanumab for Alzheimer's disease has heightened the interest in therapies targeting the amyloid hypothesis. Our research has focused on identification of novel compounds to improve amyloid processing by modulating gamma secretase activity, thereby addressing a significant biological deficit known to plague the familial form of the disease. Herein, we describe the design, synthesis, and optimization of new gamma secretase modulators (GSMs) based on previously reported oxadiazine 1. Potency improvements with a focus on predicted and measured properties afforded high-quality compounds further differentiated via robust Aß42 reductions in both rodents and nonhuman primates. Extensive preclinical profiling, efficacy studies, and safety studies resulted in the nomination of FRM-024, (+)-cis-5-(4-chlorophenyl)-6-cyclopropyl-3-(6-methoxy-5-(4-methyl-1H-imidazole-1-yl)pyridin-2-yl)-5,6-dihydro-4H-1,2,4-oxadiazine, as a GSM preclinical candidate for familial Alzheimer's disease.

Design, Synthesis, and Evaluation of a Novel Series of Oxadiazine Gamma Secretase Modulators for Familial Alzheimer's Disease.

Herein we describe the design, synthesis, and evaluation of a novel series of oxadiazine-based gamma secretase modulators obtained via isosteric amide replacement and critical consideration of conformational restriction. Oxadiazine lead 47 possesses good in vitro potency with excellent predicted CNS drug-like properties and desirable ADME/PK profile. This lead compound demonstrated robust Aß42 reductions and subsequent Aß37 increases in both rodent brain and CSF at 30 mg/kg dosed orally.

Intracellular Retention of Three Quinuclidine Derivatives in Caco-2 Permeation Experiments: Mechanisms and Impact on Estimating Permeability and Active Efflux Ratio.

BACKGROUND: Three quinuclidine derivatives (FRM-1, FRM-2 and FRM-3) were subject to significant mass loss to cellular retention in Caco-2 permeation experiments. The apparent permeability coefficient (Papp) calculated with either 'sink' (Papp,sink) or 'non-sink' (Papp,nonsink) method was significantly biased. As a result, a simplified 3-compartmental distribution model was applied in this study to derive the 'intrinsic' Papp (Papp,int) and to understand the impact of cellular retention on estimating Papp and active efflux ratio (ER) values. METHODS: Time-courses of the amount of test compounds in the donor, receiver and cells were determined in the presence and absence of bafilomycin A1 (BFA, 100 nM) and / or cyclosporine A (CsA, 10 .M). A mathematical model was constructed to describe the mass transfer of test compounds among three compartments. The temporal profiles of directional Papp,sink, Papp,nonsink and the corresponding of ER values were compared with the counterpart parameters derived from data-fitting to the mathematical model. Simulations were performed for a better understanding of experimental observations. RESULTS: The mass recovery of test compounds deteriorated with incubation time and was direction dependent. Based on the directional Papp,sink values, the resulting ER is close to unity for FRM-1, and approximately 2 and 3.5 for FRM-2 and FRM-3. Treatment with BFA considerably enhanced mass recovery for FRM-1 and FRM-3 (by 5- and 2-fold) but elicited no impact on FRM-2, while ER values largely unchanged. Expectedly, Papp,nonsink was higher than Papp,sink, but the resulting ER was lower in most cases. In contrast, the model-derived Papp,int was much greater than the values of Papp,sink and Papp,nonsink. The model also quantitatively unveiled the respective contributions of lysosomal sequestration and nonspecific binding to the cellular retention of the compounds. CONCLUSION: Our work reveals the different mechanisms involved in cellular retention of these quinuclidine derivatives, and more importantly, demonstrates the value of kinetic analyses with mathematical modeling in minimizing the bias in Papp estimation when assumptions for conventional calculations are violated.

Characterization of FRM-36143 as a new γ-secretase modulator for the potential treatment of familial Alzheimer's disease.

BACKGROUND: Familial Alzheimer's disease (FAD) is caused by mutations in the amyloid precursor protein (APP) or presenilin (PS). Most PS mutations, which account for the majority of FAD cases, lead to an increased ratio of longer to shorter forms of the amyloid beta (Aß) peptide. The therapeutic rationale of γ-secretase modulators (GSMs) for Alzheimer's disease is based on this genetic evidence as well as on enzyme kinetics measurements showing changes in the processivity of the γ-secretase complex. This analysis suggests that GSMs could potentially offset some of the effects of PS mutations on APP processing, thereby addressing the root cause of early onset FAD. Unfortunately, the field has generated few, if any, molecules with good central nervous system (CNS) drug-like properties to enable proof-of-mechanism studies. METHOD: We characterized the novel GSM FRM-36143 using multiple cellular assays to determine its in vitro potency and off-target activity as well as its potential to reverse the effect of PS mutations. We also tested its efficacy in vivo in wild-type mice and rats. RESULTS: FRM-36143 has much improved CNS drug-like properties compared to published GSMs. It has an in vitro EC50 for Aß42 of 35 nM in H4 cells, can reduce Aß42 to 58 % of the baseline in rat cerebrospinal fluid, and also increases the non-amyloidogenic peptides Aß37 and Aß38. It does not inhibit Notch processing, nor does it inhibit 24-dehydrocholesterol reductase (DHCR24) activity. Most interestingly, it can reverse the effects of presenilin mutations on APP processing in vitro. CONCLUSIONS: FRM-36143 possesses all the characteristics of a GSM in terms of Aß modulation Because FRM-36143 was able to reverse the effect of PS mutations, we suggest that targeting patients with this genetic defect would be the best approach at testing the efficacy of a GSM in the clinic. While the amyloid hypothesis is still being tested with ß-site APP-cleaving enzyme inhibitors and monoclonal antibodies in sporadic AD, we believe it is not a hypothesis for FAD. Since GSMs can correct the molecular defect caused by PS mutations, they have the promise to provide benefits to the patients when treated early enough in the course of the disease.

Neuropharmacokinetics of two investigational compounds in rats: divergent temporal profiles in the brain and cerebrospinal fluid.

Two investigational compounds (FRM-1, (R)-7-fluoro-N-(quinuclidin-3-yl)benzo[b]thiophene-2-carboxamide and FRM-2, (R)-7-cyano-N-(quinuclidin-3-yl)benzo[b]thiophene-2-carboxamide) resided in rat brain longer than in systemic circulation. In Caco-2 directional transport studies, they both showed good intrinsic passive permeability but differed significantly in efflux susceptibility (efflux ratio of <2 and ∼7, respectively), largely attributed to P-glycoprotein (P-gp). Capitalizing on these interesting properties, we investigated how cerebrospinal fluid (CSF) concentration (CCSF) would be shaped by unbound plasma concentration (Cu,p) and unbound brain concentration (Cu,b) in disequilibrium conditions and at steady state. Following subcutaneous administration, FRM-1CCSF largely followed Cu,p initially and leveled between Cu,p and Cu,b. However, it gradually approached Cu,b and became lower than, but parallel to Cu,b at the terminal phase. In contrast, FRM-2CCSF temporal profile mostly paralleled the Cu,p but was at a much lower level. Upon intravenous infusion to steady state, FRM-1CCSF and Cu,b were similar, accounting for 61% and 69% of the Cu,p, indicating a case of largely passive diffusion-governed brain penetration where CCSF served as a good surrogate for Cu,b. On the contrary, FRM-2CCSF and Cu,b were remarkably lower than Cu,p (17% and 8% of Cu,p, respectively), suggesting that FRM-2 brain penetration was severely impaired by P-gp-mediated efflux and CCSF underestimated this impact. A semi-physiologically based pharmacokinetic (PBPK) model was constructed that adequately described the temporal profiles of the compounds in the plasma, brain and CSF. Our work provided some insight into the relative importance of blood-brain barrier (BBB) and blood-CSF barrier (BCSFB) in modulating CCSF.