Introduction

Today’s drug discovery pipelines suffer from high rates of drug candidate failure that impose an economic burden on health care. Owing to the complexity of drug action, identifying drugs that exhibit therapeutic effect for a specific disease is a highly involved process. It has been estimated that US$1.78 billion and 13.5 years are required to take a single therapeutic from its initial identification as a candidate to its availability in the clinic.1 Therefore, there is enormous opportunity to develop more efficient drug identification platforms that integrate in silico approaches with available genomic data. In the case of cancer, there have been several efforts to characterize drug action at the genomic level to understand its functional effect. Large consortia such as the Connectivity Map (CMap) and Genomics of Drug Sensitivity (GDSC) have generated cell line-derived gene expression and drug response data to identify genomic predictors of drug effectiveness.2, 3, 4 Here, we chose to utilize data from blood-derived cell lines to develop a novel integrated pre-clinical drug-screening framework to identify drug leads for the treatment of acute myeloid leukemia (AML).

AML is a relatively rare, liquid cancer that is induced upon development of genetic lesions in immature hematocytes that comprise the myeloid lineage. The current model of AML holds that hematopoietic cells contain sequential mutations that propagate through the myeloid lineage during differentiation and self-renewal.5 As the age of an individual increases, mutations accumulate until a key driver mutation occurs that drives uncontrolled cell proliferation.5 Unfortunately, few new therapeutics have been introduced into the market over past decades.6 To address this issue, we have developed an integrated in silico and in vivo drug-screening framework that utilizes readily available high-throughput gene expression data to expedite pre-clinical drug discovery efforts in AML.

Gene expression profiles (GEPs) have been used to computationally search for drug candidates that exhibit potential anticancer activity.7, 8, 9, 10, 11, 12 For example, Hassane et al.8 utilized the gene expression signature of parthenolide, a known AML therapeutic, to search for drugs that could induce a similar signature by applying a correlation-based procedure. Furthermore, Sirota et al.9 performed a systematic repositioning analysis to identify potential drug candidates for several diseases using a compendia of publicly available gene expression data. Similarly, Dudley et al.10 identified topiramate as a potential candidate to treat inflammatory bowel disease. However, these studies did not incorporate patient clinical information into their prediction analyses, which is a key indicator of drug effectiveness.

Therefore, in this study we utilize an integrated approach whereby we implement the Integrated Drug Expression Analysis (IDEA) framework in AML to derive a set of drug candidate predictions.13 IDEA was previously developed to predict drug candidates by identifying drugs that could induce a GEP associated with breast cancer patient survival.13 This approach differs from previous methods by taking into account time-to-event patient clinical information and by using a more sensitive GEP-matching algorithm. We show that by applying IDEA, drugs associated with AML patient survival correlate with in vitro pharmacological metrics of drug potency and drug-associated molecular features. Finally, we carried out validation experiments in an AML mouse model to show that our predicted survival-associated drugs indeed exhibit pharmacological activity in vivo by substantially slowing tumor growth. Overall, we present a novel integrated pre-clinical drug discovery platform that combines both data-driven and experimental methodologies.

Materials and methods

Overview of integrated framework

Our integrated pre-clinical drug-screening procedure begins by implementing IDEA to identify drug candidates for AML treatment. IDEA integrates drug treatment profiles (DTPs) from CMap, AML patient tumor GEPs and AML patient survival information to output drugs that may induce a pharmacological effect that impacts patient survival (Figure 1). Briefly, IDEA calculates a drug regulatory score (DRS) between a DTP and an AML tumor GEP, which quantitatively measures the level of similarity or dissimilarity between the two profiles (see Methods in Ung et al.13). DRS is calculated for all DTP and tumor GEP pairs; and the DRS profile of each drug (DRS across patient samples) was fitted with a Cox proportional hazards model to identify top candidates that presumably have an effect on patient survival (Figure 1). Several predicted candidates are then selected for experimental validation in an AML mouse model (Figure 1).

Figure 1
figure 1

Flowchart of integrated screening pipeline. Drug treatment profiles from HL60 cell lines were integrated with acute myeloid leukemia (AML) patient gene expression profiles (GEPs) to derive drug regulatory score (DRS) profiles for each drug. Survival analysis of DRS profiles was implemented to identify drugs associated with patient survival. Selected candidates were then validated in an AML mouse model.

PowerPoint slide

Full size image

Processing of DTPs

In all, 1229 DTPs corresponding to 1078 different drugs were downloaded from the CMap database.2 Data used in this study were derived from HL60 (human promyelocytic leukemia cells) cells treated with 1078 different drugs; and gene expression was measured after treatment by using Affymetrix arrays. The raw.CEL files were processed using the Robust Microarray Analysis method implemented in the R package ‘affy’;14 all probe sets were represented as absolute expression values. For each drug, relative log2 expression profiles were generated by comparing profiles from treated samples with a common reference profile containing the basal expression values of untreated HL60 cell lines. Probe set expression values were collapsed into gene level expression by taking the average of the probe set log2 ratios, resulting in final DTPs. For some drugs, multiple treatment profiles were available, which corresponded to different biological replicates and treatment with different dose concentrations. Each drug’s treatment profile is represented as a vector of log2 ratios, indicating the expression changes of all genes in response to the drug treatment. Each DTP was then split into upregulated gene and downregulated gene groups and z-transformed to derive a P-value. The P-values were then −log10 transformed and used as an input into IDEA. Detailed explanation of this process can be found in Ung et al.13

AML gene expression data

Normalized AML microarray data containing 562 patient GEPs from Herold et al. were downloaded from GEO under accession number GSE37642.15, 16 In all, 170 AML GEPs published by Wilson et al.17 was downloaded from the NCI caArray database under the accession number willm-0019. The Wouters (n=526) and Valk (n=293) AML data sets were downloaded from the GEO database under the accession numbers GSE14468 and GSE1159, respectively.17, 18, 19

In silico identification of survival-associated drugs

The IDEA computational framework was used to screen for drugs associated with patient survival in AML. Briefly, the DTPs characterize the effect of drugs on the expression of genes. Genes with larger positive/negative log ratios are those that are regulated more intensively by a drug. We examined the baseline expression levels of the genes in AML patient samples and calculated DRSs for each DTP-tumor sample pair. DRS is a quantitative measure of similarity between a DTP and an AML sample’s GEP. If genes that are upregulated in an AML patient sample also tend to be upregulated in HL60 cell lines in response to drug exposure, then the sample will be assigned a large positive DRS corresponding to the drug. Similarly, a large positive DRS will be assigned to a patient if genes that are downregulated in the sample are also downregulated in HL60 in response to drug treatment. Conversely, if there is an inverse relationship where genes that are upregulated in the AML patient sample tend to be downregulated in HL60 cell lines in response to drug treatment, the patient will be assigned a large negative DRS. When drug-regulated genes are randomly distributed in a sorted AML expression profile, the sample will yield a DRS close to 0. The DRS is calculated using a random-walk based algorithm implemented by IDEA. A detailed description of the IDEA framework can be found in Ung et al.13

Machine learning analysis

Unsupervised clustering of patients from the Wouters data set was performed using DRS profiles that were most deviant between good and poor/intermediate cytogenetic risk patients as determined by a two-sided Wilcoxon rank-sum test.19 This resulted in 94 significant DRS profiles (P<1E−4) that were used as features in the clustering analysis using complete linkage and Euclidean distance.

Random forest and linear discriminant analysis (LDA) were implemented to classify patients into cytogenetic risk groups on the basis of their DRS profiles. The random forest machine learning model was trained using all DRS profiles as features for both the two-group and three-group classification tasks. For the two-group classification task, model performance was evaluated by calculating the AUROC (Area Under the Receiver Operating Characteristic Curve). For the three-group task, classification accuracy was calculated from the model prediction. In the LDA, the 94 features identified in the unsupervised clustering analysis were used for the two-group classification task. Calculation of AUROC and 10-fold cross-validation were used to evaluate the model accuracy. For the three-group task, an ANOVA was used to compare DRS profiles between good, poor and intermediate cytogenetic risk groups and an adjusted P<0.01 was used as a cutoff to yield a total of 144 DRS profiles. These profiles were used as features in the LDA model; and classification accuracy was calculated from the model prediction.

R packages ‘gplots’, ‘randomForest’ and ‘MASS’ were used to implement clustering, random forest and LDA analyses, respectively.

Correlation of trichostatin A DRS with IC50 and mRNA expression

GEPs from 102 blood-derived cell lines along with the IC50 information for 125 drugs across these 102 blood-derived cell lines were downloaded from the GDSC database.4 A Trichostatin A (TSA) DRS was computed for each cell line using the TSA DTP derived from CMap and the cell line GEP. Each cell line will have a DRS and a drug IC50 value associated with it. Correlation between the TSA DRS profile and IC50 profile of each GDSC drug across the 102 blood-derived cell lines was calculated using Spearman correlation. The correlation of the TSA DRS with HDAC2, MEK and Bcl2 mRNA expression across the 102 cell lines was also performed using Spearman correlation.

In vivo validation of survival-associated drugs

The following animal study was approved by the IACUC of National Chung-Hsing University. Athymic BALB/c nu/nu nude mice (4–6 weeks of age) were purchased from the National Laboratory Animal Center (Taipei, Taiwan); and mice were maintained in pathogen-free conditions with irradiated chow. HL60 (ATCC®, CCL-240™) cells were re-suspended in serum-free RPMI-1640 medium mixed with Matrigel (BD Biosciences, San Jose, CA, USA) at a 1:1 ratio. Mice were injected s.c with 5 × 106 cells in 0.5 ml matrigel into the ventral flank; and tumors were allowed to grow for 10 days or until palpable tumors formed (approximately 50 mm3). Tumor-bearing mice were randomly assigned to the following treatment groups: control (10% DMSO+90% glyceryl trioctanoate), sulfasalazine (250 mg kg−1), fluoxetine (30 mg kg−1), betulinic acid (20 mg kg−1), clozapine (1 mg kg−1) so that each treatment group contained five mice. All four drugs were purchased from Sigma‐Aldrich (St Louis, MO, USA); and the drug dosages were used by referring to previous reports describing anti-tumor activity of these compounds.20, 21, 22, 23 Mice were treated every other day by intraperitoneal injection using 100 mL total volumes. Mean tumor volumes were measured according to the formula: length × width × thickness × 0.5, and expressed as mm3 values before each treatment. Mice were killed when tumors reached a size of 200 mm3 or became ulcerated. If individual mice within a group were killed, then the final measurement was carried over to subsequent time points. For each drug, a permutation-based procedure (R package ‘statmod’) was used to calculate the significance of drug-treated tumor growth curve compared with vehicle.24, 25 Center bars on growth curves indicate mean and error bars represent s.d.

Identification of candidate drugs for AML treatment

We calculated the DRS for 1078 drugs in all samples of an AML expression data set. To identify the candidate drugs that might be effective for treating AML, we examined the correlation between DRS of these drugs and patient survival. Our rationale is that if the expression of drug-regulated genes is correlated with patient survival, then the drug might be used to modulate these genes in AML to induce a pharmacological effect. For each drug, we fitted a univariate Cox proportional hazards model using the DRS as the independent variable and patient survival as the dependent variable.26 We also fitted multivariate Cox regression models to adjust for potential confounding clinical factors such as age, tumor stage, tumor grade and estrogen receptor status. Analysis of Schoenfeld residuals was used to evaluate the proportional hazards assumption for all models. The Wald test was used to assess the significance of the model parameters; and P-values were adjusted for multiple hypotheses testing using the Benjamini-Hochberg procedure.27 The ‘survival’ R package was used to implement the survival analysis.

GO enrichment analysis

Genes that were upregulated or downregulated twofold upon treatment with TSA were used to calculate GO enrichment of BP terms via the DAVID bioinformatics tool (http://david.abcc.ncifcrf.gov/).28, 29

Results

Systematic screening of drugs associated with AML patient survival

At an adjusted P-value cutoff of 0.01, we identified 66 drugs using the Herold data set that are significantly associated with AML patient survival to validate in vivo (Supplementary Table S1).30 We re-implemented this analysis in the Verhaak, Valk and Wilson AML data sets and achieved similar results.17, 18, 19 Figure 2 shows an example where TSA DRS profiles effectively stratify AML patients into favorable and poor prognosis groups in four independent data sets.17, 18, 19 This indicates that IDEA’s output is reproducible and that TSA exhibits pharmacological activity in several cancer types, as shown by previous studies.21, 26, 31, 32, 33 This in silico screening approach filters out biologically inactive drugs and efficiently identifies the most probable candidates based on gene expression and patient survival.

Figure 2
figure 2

Trichostatin A (TSA) was identified as a drug candidate in four data sets. TSA drug regulatory score (DRS) profile is prognostic in the (a) Herold data set, (b) Verhaak data set, (c) Valk data set (d) and William data set with P<0.01. Patients were stratified at DRS=0.

PowerPoint slide

Full size image

Drug DRS profiles are predictive of cytogenetic risk in AML

Since drug DRS is prognostic, we hypothesized that DRS can predict cellular phenotypes that have been traditionally known to correlate with patient survival. As such, we attempted to predict cytogenetic risk of patients based on drug DRS profiles.34, 35 Using the Wouters data set, we selected DRS profiles that differed significantly between patients classified as having ‘good’ cytogenetic risk or ‘poor/intermediate’ cytogenetic risk using an adjusted P-value cutoff of 1E−4 (Wilcoxon rank-sum test).19 This yielded 94 DRS profiles that were used to cluster patients into subgroups. We found that these DRS profiles were informative such that they were able to cluster patients into cytogenetic risk groups (Figure 3a). In particular, we identified three apparent clusters corresponding to patient cytogenetic risk. The first cluster (top) had 50 samples with good cytogenetic risk (49.4%), the second cluster had 40 samples with good cytogenetic risk (25.3%) and the third cluster had 7 samples with good cytogenetic risk (3.8%) (Figure 3a). This indicates that DRS profiles effectively identified differences in cytogenetic risk between AML samples, showing that they reflect prognostic molecular features of tumors.

Figure 3
figure 3

Drug regulatory score (DRS) profiles are predictive of cytogenetic risk in acute myeloid leukemia (AML). (a) Unsupervised clustering of DRS profiles. Magenta sample labels denote good cytogenetic risk and aqua sample labels indicate poor/intermediate cytogenetic risk. (b) Receiver operating characteristic curve for random forest classification of patients into good or poor/intermediate cytogenetic risk categories.

PowerPoint slide

Full size image

To further evaluate our unsupervised results, we trained a random forest model using DRS profiles from all drugs to verify their ability predict patient cytogenetic risk. We performed two-group (good vs poor/intermediate) and three-group (good, poor and intermediate) classification tasks using DRS profiles as covariates and cytogenetic risk group as the response variable. We were able to achieve an AUROC of 0.97 for the two-group task, and an accuracy rate of 71% for the three-group task (Figure 3b). Additionally, we implemented LDA and were able to achieve an AUC of 0.92 for the two-group task and an accuracy rate of 76% for the three-group task. Similarly, Zhou et al.36 reported comparable accuracy when predicting cytogenetic risk using patient GEPs. Ultimately, these results indicate that DRS profiles contain information that reflects phenotypic differences between patient groups.

Correlation of DRS with in vitro pharmacological metrics

To explore potential mechanisms of action underlying our predicted drugs, we adopted a novel integrated validation procedure whereby we implemented IDEA in 102 GDSC blood-derived cell lines and correlated the outputted DRS with treatment response metrics. As the drugs included in the GDSC data set have known targets, this correlation analysis allowed us to gain insight into the biological mechanisms underlying drug treatment with CMap drugs. We calculated DRS between CMap DTPs and GDSC blood-derived cell line GEPs and correlated the DRS of each CMap drug with the IC50 of each GDSC drug across blood-derived GDSC cell lines (Figure 4a). We repeated this analysis using other pharmacological metrics (area under the curve and EC50) derived from dose-response curves and achieved consistent results. To note, CMap provides before and after drug treatment GEPs for three cell lines (drug-centric), whereas GDSC provides profiles that measure basal gene expression of several cell lines (cell line-centric).

Figure 4
figure 4

Correlation of trichostatin A (TSA) with drug IC50 and mRNA expression across 102 blood-derived Genomics of Drug Sensitivity (GDSC) cell lines. (a) Correlation of TSA drug regulatory score (DRS) with IC50 of 125 drugs from the GDSC database. (b) TSA DRS is anticorrelated with HDAC2 mRNA expression. (c) TSA DRS is correlated with ABT-263 (Bcl2 inhibitor) IC50. (d) TSA DRS is correlated with AZD6244 (MEK inhibitor) IC50. (e) TSA DRS is correlated with MEK mRNA expression.

PowerPoint slide

Full size image

Interestingly, we found that TSA DRS was most correlated with the IC50 of ABT-263 (navitoclax), a Bcl2 inhibitor, with a Pearson correlation coefficient (PCC) of 0.57 and P=1E−5 (Figures 4a and c). Furthermore, we identified MEK inhibitors that were also highly correlated including AZD6244 (PCC=0.53), RDEA119 (PCC=0.47), PD0325901 (PCC=0.4) and CI-1040 (PCC=0.4) (Figure 4a). These results suggest that Bcl2 and MEK pathways are involved in response to TSA treatment. To further evaluate their involvement, we correlated the TSA DRS with HDAC2, Bcl2 and MEK mRNA expression across the 102 GDSC blood-derived cell lines. As expected, the TSA DRS was anticorrelated with HDAC2 expression (PCC=−0.44, P=4.5E−6) (Figure 4b). However, we found that there was a significant, albeit weak, anti-correlation between TSA DRS and Bcl2 expression (PCC=−0.2, P=0.04) even though there was a strong correlation between TSA DRS and ABT-263 IC50 (Supplementary Fig S1, Figure 4c). In spite of this, the directionality of the correlation is in accordance with previous studies reporting that increased Bcl2 expression confers sensitivity to Bcl2 inhibitors.37, 38 However, the small effect size suggests a more complicated relationship between HDAC inhibition and BCL-2 expression. In the case of MEK, we observed that TSA DRS was correlated with both MEK expression (PCC=0.47, P=6.2E−7) and AZD6244 IC50 (PCC=0.53, P=4.1E−5) (Figures 4d and e). This suggests that high TSA DRS, which indicates decreased HDAC expression, results in an upregulation of MEK that explains why increased dosage concentration of AZD6244, RDEA119, PD0325901 and CI-1040 is required to achieve 50% cellular inhibition in vitro. Indeed, several studies have shown that HDAC inhibitors and MEK inhibitors exhibit synergistic effects in leukemia, indicating that our analysis was able to identify molecular mechanisms underlying drug effect.39, 40, 41, 42

To compare, we carried out Gene Ontology enrichment analysis of the top upregulated and downregulated genes in the TSA DTP (Supplementary Table S2).43 Surprisingly, we found no significant cancer-related biological processes enriched in the differentially expressed genes. We speculate that as TSA is a histone deacetylase inhibitor, it will have widespread, yet small downstream effects on genes that may or may not have a functional effect. As a result, standard enrichment analysis may not be sensitive enough to detect key genes involved in apoptosis or cell proliferation. These results suggest that identifying enriched pathways in drug-regulated genes may not be able reveal drug mechanisms, at least for some drugs such as TSA, and that correlating DRS with phenotypic response to targeted inhibitors provides an alternative strategy.

In vivo validation of novel predicted drugs for AML treatment: sulfasalazine, fluoxetine, betulinic acid, clozapine

To translate our in silico drug-screening procedure into the pre-clinical testing phase, we identified five novel survival-associated drugs that were predicted by IDEA and experimentally evaluated their effectiveness in an AML mouse model. In particular, we tested sulfasalazine, fluoxetine, clozapine, betulinic acid and ceforanide, which were originally intended to treat arthritis, depression, schizophrenia, viral infections and bacterial infections, respectively.44, 45, 46, 47, 48 All five of these drugs were predicted to impact patient survival via pharmacological activity as shown in Figures 5a and d in the Herold data set.15 To generate our AML mouse models, we engrafted athymic BALB/c nu/nu mice with HL60 (Human promyelocytic leukemia cells) via subcutaneous xenografts. Each drug was then tested for therapeutic activity by treating mice with either vehicle (10% DMSO+90% glyceryl trioctanoate) alone or standard doses of the drug candidate. We found that four out of five (ceforanide showed no substantial effect) of our predicted drugs exhibited significant therapeutic activity. First, we found that treatment with 250 mg kg−1 of sulfasalazine substantially decelerated tumor growth compared with vehicle over a 21-day period (P<0.01) (Figures 6a and b). Second, in the case of fluoxetine, daily treatment with 30 mg kg−1 also decreased the rate of tumor growth compared with vehicle (P<0.01) (Figures 6c and d). Third, we observed that daily treatment with 20 mg kg−1 of betulinic acid also decelerated tumor growth (P<0.01) (Figure 6e). Finally, we observed similar antineoplastic activity of clozapine at a dose of 1 mg kg−1 (P<0.01) (Figure 6f). These results suggest that our initial in silico screen was able to output several survival-associated drugs, the majority of which could be verified in vivo.

Figure 5
figure 5

Survival analysis of sulfasalazine, fluoxetine, clozapine and betulinic acid drug regulatory score (DRS) profiles. (a) Sulfasalazine, (b) fluoxetine, (c) clozapine (d) and betulinic acid were identified as drugs associated with acute myeloid leukemia (AML) patient survival and chosen for further experimental validation. DRS profiles of all drugs were associated with AML patient survival with P<0.05 (log-rank test).

PowerPoint slide

Full size image
Figure 6
figure 6

In vivo validation of sulfasalazine, fluoxetine, clozapine and betulinic acid in acute myeloid leukemia (AML) mouse models. (a) 250 mg kg−1 Sulfasalazine decreased the rate of tumor growth compared with vehicle-treated control over a course of 21 days. (b) Image of tumor from mouse treated with 250 mg kg−1 sulfasalazine compared with control. (c) 30 mg kg−1 Fluoxetine decreased the rate of tumor growth compared with vehicle-treated control over a course of 21 days. (d) Image of tumor from mouse treated with 250 mg kg−1 fluoxetine compared with control. (e) 20 mg kg−1 Betulinic acid decreased the rate of tumor growth compared with vehicle-treated control over a course of 21 days. (f) 1 mg kg−1 Clozapine decreases the rate of tumor growth compared with vehicle-treated control over a course of 21 days.

PowerPoint slide

Full size image

Discussion

Drug discovery has long been a focus of intense research due to the constant need for therapeutics that can treat disease and ameliorate symptoms. In the case of cancer, rapid development of acquired resistance to commonly prescribed chemotherapeutics and targeted therapies necessitates the formulation of faster and more efficient drug-screening pipelines. Here, we implemented a computational drug prediction framework in AML and were able to experimentally validate several of our drug predictions to identify candidates. We explored the association between IDEA and pharmacological metrics to gain mechanistic insight into drug action; and our experimental results show a substantial reduction in tumor growth when mice were treated with four different survival-associated drugs over a 21-day period.

We note that there are limitations to our approach. First, the reliability of DTPs, which were derived by averaging gene expression over several replicate experiments, may exhibit variability. Second, these DTPs were generated over different concentrations and may not reflect optimal drug activity. Third, we note that the limited time interval over which mice were treated provides a short-term evaluation of drug effectiveness and that anti-tumor activity may not be sustained and/or side effects may present itself after prolonged exposure. Finally, it is difficult to interpret the hazard ratios of the top drugs outputted by IDEA. Since patient tumors from our data sets were collected before treatment, the survival results may have been influenced by subsequent therapy. This may be why some known anticancer drugs were associated with a hazard ratio of >1. Thus, we claim that any significant association that exists between the drug and patient survival indicates pharmacological activity, which merits further experimental investigation.

Despite these obstacles, we maintain that our integrated pipeline is robust and sensitive enough to detect potential drug candidates. In particular, we have shown that we could computationally identify known therapeutics (for example, TSA) and novel candidates. We further support our predictions by conducting in vivo experiments that show our drug leads exhibit therapeutic activity in an AML mouse model. Our results strongly support the effectiveness of using an integrated in silico/in vivo approach to drug screening in the context of patient survival.