Age and dPCR can predict relapse in CML patients who discontinued imatinib: the ISAV study.

Imatinib is effective for the treatment of chronic myeloid leukemia (CML). However even undetectable BCR-ABL1 by Q-RT-PCR does not equate to eradication of the disease. Digital-PCR (dPCR), able to detect 1 BCR-ABL1 positive cell out of 10(7) , has been recently developed. The ISAV study is a multicentre trial aimed at validating dPCR to predict relapses after imatinib discontinuation in CML patients with undetectable Q-RT-PCR. CML patients under imatinib therapy since more than 2 years and with undetectable PCR for at least 18 months were eligible. Patients were monitored by standard Q-RT-PCR for 36 months. Patients losing molecular remission (two consecutive positive Q-RT-PCR with at least 1 BCR-ABL1/ABL1 value above 0.1%) resumed imatinib. The study enrolled 112 patients, with a median follow-up of 21.6 months. Fifty-two of the 108 evaluable patients (48.1%), relapsed; 73.1% relapsed in the first 9 months but 14 late relapses were observed between 10 and 22 months. Among the 56 not-relapsed patients, 40 (37.0% of total) regained Q-RT-PCR positivity but never lost MMR. dPCR results showed a significant negative predictive value ratio of 1.115 [95% CI: 1.013-1.227]. An inverse relationship between patients age and risk of relapse was evident: 95% of patients <45 years relapsed versus 42% in the class ≥45 to <65 years and 33% of patients ≥65 years [P(χ(2) ) < 0.0001]. Relapse rates ranged between 100% (<45 years, dPCR+) and 36% (>45 years, dPCR-). Imatinib can be safely discontinued in the setting of continued PCR negativity; age and dPCR results can predict relapse.

Predicting myelosuppression of drugs from in silico models.

Anticancer agents targeting proliferating cell populations in tumor as well as in normal tissues can lead to a number of side effects including hematotoxicity, a common dose-limiting toxicity associated with oncology drugs. Myelosuppression, regarded as unacceptable for other therapeutic indications, is considered a clinical risk also for new targeted anticancer drugs acting specifically on tumor cells. Thus, it becomes important not only to evaluate the potential toxicity of such new therapeutics to human hematopoietic tissue during preclinical development but also to anticipate this liability in early drug discovery. This could be achieved by using in silico models to guide the design of new lead compounds and the selection of analogs with reduced myelosuppressive potential. Hence, the purpose of this study was to develop computational models able to predict the potential myelotoxicity of drugs from their chemical structure. The data set analyzed included 38 drugs. The structural diversity and the drug-like space covered by these molecules were investigated using the ChemGPS methodology. Two sets of potentially relevant descriptors for modeling myelotoxicity (i.e., 3D Volsurf+ and 2D structural and electrotopological E-states descriptors) were selected and a Principal Component Analysis was carried out on the entire set of data. The first two PCs were able to discriminate the highest from the least myelotoxic compounds with a total accuracy of 95%. Then, a quantitative PLS model was developed by correlating a selected subset of in vitro hematotoxicity data with Volsurf+ descriptors. After variable selection, the PLS analysis resulted in a one-latent-variable model with r(2) of 0.79 and q(2) of 0.72. The inclusion of 2D descriptors in the PLS analysis improved only slightly the robustness and quality of the model that predicted the pIC(50) values of 21 drugs not included in the model with a RMSEP of 0.67 and a squared correlation coefficient (r(0)(2)) of 0.70. Furthermore, in order to investigate whether the highly myelotoxic compounds are characterized by common structural features, which should be taken into consideration in the design of new candidate drugs, the entire data set was analyzed using GRIND toxicophore-based descriptors. One toxicophore emerged from the interpretation of the model. The toxicophore elements, at least determined by the molecules used in this study, are a pattern of H-bond acceptor groups, presence of a H-bond donor and H-bond acceptor regions at ∼15 Å distance and a hydrophobic and H-bond acceptor interacting regions separated by a distance of ∼12.4 Å. Moreover, the dimensions of the molecule play a role in its recognition as a myelotoxic compound.

Development and validation of in silico models for estimating drug preformulation risk in PEG400/water and Tween80/water systems.

Solubility is one of the most important properties of drug candidates for achieving the targeted plasma concentrations following oral dosing. Furthermore, the formulations adopted in the in vivo preclinical studies, for both oral and intravenous administrations, are usually solutions. To formulate compounds sparingly soluble in water, pharmaceutically acceptable cosolvents or surfactants are typically employed to increase solubility. Compounds poorly soluble also in these systems will likely show severe formulation issues. In such cases, relatively high amount of compounds, rarely available in the early preclinical phases, are needed to identify the most appropriate dosing vehicles. Hence, the purpose of this study was to build two computational models which, on the basis of the molecular structure, are able to predict the compound solubility in two vehicle systems (40% PEG400/water and 10% Tween80/water) used in our company as screening tools for anticipating potential formulation issues. The two models were developed using the solubility data obtained from the analysis of approximately 2000 chemically diverse compounds. The structural diversity and the drug-like space covered by these molecules were investigated using the ChemGPS methodology. The compounds were classified (high/low preformulation risk) based on the experimental solubility value range. A combination of descriptors (i.e. logD at two different pH, E-state indices and other 2D structural descriptors) was correlated to these classes using partial least squares discriminant (PLSD) analysis. The overall accuracy of each PLSD model applied to independent sets of compounds was approximately 78%. The accuracy reached when the models were used in combination to identify molecules with low preformulation risk in both systems was 83%. The models appeared a valuable tool for predicting the preformulation risk of drug candidates and consequently for identifying the most appropriate dosing vehicles to be further investigated before the first in vivo preclinical studies. Since only a small number of 2D descriptors is need to evaluate the preformulation risk classes, the models resulted easy to use and characterized by high throughput.

Evaluation of a basic physiologically based pharmacokinetic model for simulating the first-time-in-animal study.

The objective of this study was to evaluate a physiologically based pharmacokinetic (PBPK) approach for predicting the plasma concentration-time curves expected after intravenous administration of candidate drugs to rodents. The predictions were based on a small number of properties that were either calculated based on the structure of the candidate drug (octanol:water partition coefficient, ionization constant(s)) or obtained from the typical high-throughput screens implemented in the early drug discovery phases (fraction unbound in plasma and hepatic intrinsic clearance). The model was tested comparing the predicted and the observed pharmacokinetics of 45 molecules. This dataset included six known drugs and 39 drug candidates from different discovery programs, so that the performance of the model could be evaluated in a real discovery case scenario. The plasma concentration-time curves were predicted with good accuracy, the pharmacokinetic parameters being on average two- to three-fold of actual values. Multivariate analysis was used for identifying the candidate properties which were likely associated to biased predictions. The application of this approach was found useful for the prioritization of the in vivo pharmacokinetics screens and the design of the first-time-in-animal studies.

Computational models for identifying potential P-glycoprotein substrates and inhibitors.

Multidrug resistance mediated by ATP binding cassette (ABC) transporters such as P-glycoprotein (P-gp) represents a serious problem for the development of effective anticancer drugs. In addition, P-gp has been shown to reduce oral absorption, modulate hepatic, renal, or intestinal elimination, and restrict blood-brain barrier penetration of several drugs. Consequently, there is a great interest in anticipating whether drug candidates are P-gp substrates or inhibitors. In this respect, two different computational models have been developed. A method for discriminating P-gp substrates and nonsubstrates has been set up based on calculated molecular descriptors and multivariate analysis using a training set of 53 diverse drugs. These compounds were previously classified as P-gp substrates or nonsubstrates on the basis of the efflux ratio from Caco-2 permeability measurements. The program Volsurf was used to compute the compounds' molecular descriptors. The descriptors were correlated to the experimental classes using partial least squares discriminant analysis (PLSD). The model was able to predict correctly the behavior of 72% of an external set of 272 proprietary compounds. Thirty of the 53 previously mentioned drugs were also evaluated for P-gp inhibition using a calcein-AM (CAM) assay. On the basis of these additional P-gp functional data, a PLSD analysis using GRIND-pharmacophore-based descriptors was performed to model P-gp substrates having poor or no inhibitory activity versus inhibitors having no evidence of significant transport. The model was able to discriminate between 69 substrates and 56 inhibitors taken from the literature with an average accuracy of 82%. The model allowed also the identification of some key molecular features that differentiate a substrate from an inhibitor, which should be taken into consideration in the design of new candidate drugs. These two models can be implemented in a virtual screening funnel.

Structure-based approaches to improve selectivity: CDK2-GSK3beta binding site analysis.

An evaluation and comparison of two different approaches, GRID/CPCA and GRIND/CPCA (CPCA = consensus principal component analysis; GRIND = GRid-INdependent Descriptors), suitable for visualizing the structural differences between related proteins is presented. Ten crystal structures of CDK2/cyclin A and GSK3beta solved in-house with different inhibitors were compared with the aim of highlighting regions that could be potential sites for gaining selectivity for CDK2 versus GSK3beta. The analyses pointed out remarkable differences in the backs of the CDK2-GSK3beta ATP binding pockets that guided the optimization toward a selective benzodipyrazole CDK2 inhibitor. The gain in selectivity can be associated with the two main differences in the ATP pocket between the enzymes. Phe80 of CDK2, the so-called gatekeeper residue often exploited for the design of kinase selective ligands, is replaced by a leucine in GSK3beta, and Ala144 is replaced by a cysteine. As a consequence of these mutations, CDK2 has a less elongated and less flat buried region at the back of the ATP pocket.

Virtual screening to enrich a compound collection with CDK2 inhibitors using docking, scoring, and composite scoring models.

Docking programs can generate subsets of a compound collection with an increased percentage of actives against a target (enrichment) by predicting their binding mode (pose) and affinity (score), and retrieving those with the highest scores. Using the QXP and GOLD programs, we compared the ability of six single scoring functions (PLP, Ligscore, Ludi, Jain, ChemScore, PMF) and four composite scoring models (Mean Rank: MR, Rank-by-Vote: Vt, Bayesian Statistics: BS and PLS Discriminant Analysis: DA) to separate compounds that are active against CDK2 from inactives. We determined the enrichment for the entire set of actives (IC50 < 10 microM) and for three activity subsets. In all cases, the enrichment for each subset was lower than for the entire set of actives. QXP outperformed GOLD at pose prediction, but yielded only moderately better enrichments. Five to six scoring functions yielded good enrichments with GOLD poses, while typically only two worked well with QXP poses. For each program, two scoring functions generally performed better than the others (Ligscore2 and Ludi for GOLD; QXP and Jain for QXP). Composite scoring functions yielded better results than single scoring functions. The consensus approaches MR and Vt worked best when separating micromolar inhibitors from inactives. The statistical approaches BS and DA, which require training data, performed best when distinguishing between low and high nanomolar inhibitors. The key observation that all hit rate profiles for all four activity intervals for all scoring schemes for both programs are significantly better than random, is evidence that docking can be successfully applied to enrich compound collections.

Influence of molecular flexibility and polar surface area metrics on oral bioavailability in the rat.

The relationship of rotatable bond count (N(rot)) and polar surface area (PSA) with oral bioavailability in rats was examined for 434 Pharmacia compounds and compared with an earlier report from Veber et al. (J. Med. Chem. 2002, 45, 2615). N(rot) and PSA were calculated with QikProp or Cerius2. The resulting correlations depended on the calculation method and the therapeutic class within the data superset. These results underscore that such generalizations must be used with caution.

Model based on GRID-derived descriptors for estimating CYP3A4 enzyme stability of potential drug candidates.

A number of computational approaches are being proposed for an early optimization of ADME (absorption, distribution, metabolism and excretion) properties to increase the success rate in drug discovery. The present study describes the development of an in silico model able to estimate, from the three-dimensional structure of a molecule, the stability of a compound with respect to the human cytochrome P450 (CYP) 3A4 enzyme activity. Stability data were obtained by measuring the amount of unchanged compound remaining after a standardized incubation with human cDNA-expressed CYP3A4. The computational method transforms the three-dimensional molecular interaction fields (MIFs) generated from the molecular structure into descriptors (VolSurf and Almond procedures). The descriptors were correlated to the experimental metabolic stability classes by a partial least squares discriminant procedure. The model was trained using a set of 1800 compounds from the Pharmacia collection and was validated using two test sets: the first one including 825 compounds from the Pharmacia collection and the second one consisting of 20 known drugs. This model correctly predicted 75% of the first and 85% of the second test set and showed a precision above 86% to correctly select metabolically stable compounds. The model appears a valuable tool in the design of virtual libraries to bias the selection toward more stable compounds.