2.6.1. Transporter Substrate Models

2.6.1. Transporter Substrate Models

If the investigational drug’s ADME data suggest that renal secretion contributes ≥ 25% of the total clearance, regulators will require that it be evaluated in vitro to determine whether it is a substrate of renal transporters such as OAT1/3, OCT2. The substrate classification models for P-gp, BCRP, OATP1B1, OATP1B3, OAT1, OAT3, OCT1 and OCT2 provided as part of the Transporter Module will help you anticipate the need for such tests.

The substrate categorization data were collected from various academic and commercial databases. Though a few data compilations have been published in the open literature, the majority of the data were gathered from the FDA label database hosted at the FDA website and individual reports. It was relatively easy to find data indicating that particular compounds are transporter substrates. Finding data showing that a particular compound is not a substrate for a given transporter was much more difficult: the absence of evidence is not a reliable evidence of absence. A compound was categorized as a nonsubstrate only if we found an original report specifically indicating that this was the case.

Compounds were dropped from the data sets if conflicting reports were found as to their substrate/non-substrate categorization. All ANNE classification models were built with 20% of the compounds in the external test set. The performance of all the transporter-substrate models is shown in the table below, with concordance, sensitivity and specificity above 85% in many cases.

P-glycoprotein (P-gp). 

The literature contains a wide range of criteria for deciding whether a compound is a substrate for P-gp transport. These range from high-throughput assays - e.g., stimulation of ATPase activity in microsomal vesicles or de-suppressed uptake of the P-gp substrate calcein acetoxymethyl ester (calcein-AM), whose intracellular cleavage releases the fluorescent dye calcein (Polli 2011) - to much more resource-intensive but more direct Caco-2 and MDR1-transfected monolayer permeability assays.

The data for Pgp_Substr were collected from the literature (Erić 2013) and commercial bi-directional Caco-2 permeability assays. The reliability of literature data was evaluated using criteria similar to those proposed by Zhang et al. (2003) and Didziapetris et al. (2003). This initial data set was expanded to cover a broader chemical space by including data from the primary literature by applying a preliminary substrate classification model to the data from Caco-2 permeability assays. Structures with an BA/AB efflux ratio ≥ 2.0 (basolateral to apical/apical to basolateral) predicted to be P-gp substrates were designated as positives ("Yes") whereas structures with efflux ratios BA/AB \< 2.0 predicted not to be substrates were designated as negatives ("No").

Breast Cancer Resistance Protein (BCRP).

We collected data on a wide variety of compounds, including those compiled by Mao and Unadkat (2015), by Hazai et al. (2013) and by Shaikh et al. (2017). In addition, we collected data directly from the primary literature. The final dataset was composed of 660 compounds (359 substrates and 301 non-substrates).

Organic Anion Transporting Polypeptides (OATP1B1 & OATP1B3)

OATPs are probably best known for facilitating hepatic uptake of statins (Hirano, 2004) but antibiotics, anticancer agents, and cardiac glycosides are also substrates. Data for our OATP substrate classification models came from commercial and public databases, the primary literature, and the FDA database.

Organic Anion Transporters (OAT1 & OAT3)

The substrate specificities of OAT1 and OAT3 overlap but are not identical. OAT1 has a very broad substrate specificity, whereas OAT3 handles bulkier and more lipophilic organic anions. Data used to build the model were drawn from commercial and public databases, the primary literature, and the FDA database.

Organic Cation Transporters (OCT1 & OCT2)

About 40% of orally administered drugs are cationic at physiological pH, and OCTs are primarily responsible for their intestinal absorption, whether they bear fixed positive charges (e.g., furamidine and paraquat) or basic groups such as amines, amidines or guanidines. Human OCT1 and OCT2 have largely overlapping but distinctly different substrate specificities.

Hendrickx and coworkers (2013) compiled an extensive OCT data set covering 354 compounds. We curated it extensively and augmented it with data obtained from the primary literature and the FDA database.