Prediction and identification of synergistic compound combinations against pancreatic cancer cells

Resistance to current therapies is common for pancreatic cancer and hence novel treatment options are urgently needed. In this work, we developed and validated a computational method to select synergistic compound combinations based on transcriptomic profiles from both the disease and compound side, combined with a pathway scoring system, which was then validated prospectively by testing 30 compounds (and their combinations) on PANC-1 cells. Some compounds selected as single agents showed lower GI50 values than the standard of care, gemcitabine. Compounds suggested as combination agents with standard therapy gemcitabine based on the best performing scoring system showed on average 2.82-5.18 times higher synergies compared to compounds that were predicted to be active as single agents. Examples of highly synergistic in vitro validated compound pairs include gemcitabine combined with Entinostat, thioridazine, loperamide, scriptaid and Saracatinib. Hence, the computational approach presented here was able to identify synergistic compound combinations against pancreatic cancer cells.

Combenefit: an interactive platform for the analysis and visualization of drug combinations

MOTIVATION

Many drug combinations are routinely assessed to identify synergistic interactions in the attempt to develop novel treatment strategies. Appropriate software is required to analyze the results of these studies.

RESULTS

We present Combenefit, new free software tool that enables the visualization, analysis and quantification of drug combination effects in terms of synergy and/or antagonism. Data from combinations assays can be processed using classical Synergy models (Loewe, Bliss, HSA), as single experiments or in batch for High Throughput Screens. This user-friendly tool provides laboratory scientists with an easy and systematic way to analyze their data. The companion package provides bioinformaticians with critical implementations of routines enabling the processing of combination data.

AVAILABILITY AND IMPLEMENTATION

Combenefit is provided as a Matlab package but also as standalone software for Windows (http://sourceforge.net/projects/combenefit/).

CONTACT

Giovanni.DiVeroli@cruk.cam.ac.uk

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

An automated fitting procedure and software for dose-response curves with multiphasic features

In cancer pharmacology (and many other areas), most dose-response curves are satisfactorily described by a classical Hill equation (i.e. 4 parameters logistical). Nevertheless, there are instances where the marked presence of more than one point of inflection, or the presence of combined agonist and antagonist effects, prevents straight-forward modelling of the data via a standard Hill equation. Here we propose a modified model and automated fitting procedure to describe dose-response curves with multiphasic features. The resulting general model enables interpreting each phase of the dose-response as an independent dose-dependent process. We developed an algorithm which automatically generates and ranks dose-response models with varying degrees of multiphasic features. The algorithm was implemented in new freely available Dr Fit software (sourceforge.net/projects/drfit/). We show how our approach is successful in describing dose-response curves with multiphasic features. Additionally, we analysed a large cancer cell viability screen involving 11650 dose-response curves. Based on our algorithm, we found that 28% of cases were better described by a multiphasic model than by the Hill model. We thus provide a robust approach to fit dose-response curves with various degrees of complexity, which, together with the provided software implementation, should enable a wide audience to easily process their own data.

The EGFR demonstrates linear signal transmission

Cells sense information encoded in extracellular ligand concentrations and process it using intracellular signalling cascades. Using mathematical modelling and high-throughput imaging of individual cells, we studied how a transient extracellular growth factor signal is sensed by the epidermal growth factor receptor system, processed by downstream signalling, and transmitted to the nucleus. We found that transient epidermal growth factor signals are linearly translated into an activated epidermal growth factor receptor integrated over time. This allows us to generate a simplified model of receptor signaling where the receptor acts as a perfect sensor of extracellular information, while the nonlinear input-output relationship of EGF-EGFR triggered signalling is a consequence of the downstream MAPK cascade alone.

Gemcitabine diphosphate choline is a major metabolite linked to the Kennedy pathway in pancreatic cancer models in vivo

BACKGROUND

The modest benefits of gemcitabine (dFdC) therapy in patients with pancreatic ductal adenocarcinoma (PDAC) are well documented, with drug delivery and metabolic lability cited as important contributing factors. We have used a mouse model of PDAC: KRAS(G12D); p53(R172H); pdx-Cre (KPC) that recapitulates the human disease to study dFdC intra-tumoural metabolism.

METHODS

LC-MS/MS and NMR were used to measure drug and physiological analytes. Cytotoxicity was assessed by the Sulphorhodamine B assay.

RESULTS

In KPC tumour tissue, we identified a new, Kennedy pathway-linked dFdC metabolite (gemcitabine diphosphate choline (GdPC)) present at equimolar amounts to its precursor, the accepted active metabolite gemcitabine triphosphate (dFdCTP). Utilising additional subcutaneous PDAC tumour models, we demonstrated an inverse correlation between GdPC/dFdCTP ratios and cytidine triphosphate (CTP). In tumour homogenates in vitro, CTP inhibited GdPC formation from dFdCTP, indicating competition between CTP and dFdCTP for CTP:phosphocholine cytidylyltransferase (CCT). As the structure of GdPC precludes entry into cells, potential cytotoxicity was assessed by stimulating CCT activity using linoleate in KPC cells in vitro, leading to increased GdPC concentration and synergistic growth inhibition after dFdC addition.

CONCLUSIONS

GdPC is an important element of the intra-tumoural dFdC metabolic pathway in vivo.

Anti-tumour efficacy of capecitabine in a genetically engineered mouse model of pancreatic cancer

Capecitabine (CAP) is a 5-FU pro-drug approved for the treatment of several cancers and it is used in combination with gemcitabine (GEM) in the treatment of patients with pancreatic adenocarcinoma (PDAC). However, limited pre-clinical data of the effects of CAP in PDAC are available to support the use of the GEMCAP combination in clinic. Therefore, we investigated the pharmacokinetics and the efficacy of CAP as a single agent first and then in combination with GEM to assess the utility of the GEMCAP therapy in clinic. Using a model of spontaneous PDAC occurring in Kras(G12D); p53(R172H); Pdx1-Cre (KPC) mice and subcutaneous allografts of a KPC PDAC-derived cell line (K8484), we showed that CAP achieved tumour concentrations (∼25 µM) of 5-FU in both models, as a single agent, and induced survival similar to GEM in KPC mice, suggesting similar efficacy. In vitro studies performed in K8484 cells as well as in human pancreatic cell lines showed an additive effect of the GEMCAP combination however, it increased toxicity in vivo and no benefit of a tolerable GEMCAP combination was identified in the allograft model when compared to GEM alone. Our work provides pre-clinical evidence of 5-FU delivery to tumours and anti-tumour efficacy following oral CAP administration that was similar to effects of GEM. Nevertheless, the GEMCAP combination does not improve the therapeutic index compared to GEM alone. These data suggest that CAP could be considered as an alternative to GEM in future, rationally designed, combination treatment strategies for advanced pancreatic cancer.

Paclitaxel and CYC3, an aurora kinase A inhibitor, synergise in pancreatic cancer cells but not bone marrow precursor cells

BACKGROUND

Amplification of aurora kinase A (AK-A) overrides the mitotic spindle assembly checkpoint, inducing resistance to taxanes. RNA interference targeting AK-A in human pancreatic cancer cell lines enhanced taxane chemosensitivity. In this study, a novel AK-A inhibitor, CYC3, was investigated in pancreatic cancer cell lines, in combination with paclitaxel.

METHODS

Western blot, flow cytometry and immunostaining were used to investigate the specificity of CYC3. Sulforhodamine B staining, time-lapse microscopy and colony-formation assays were employed to evaluate the cytotoxic effect of CYC3 and paclitaxel. Human colony-forming unit of granulocyte and macrophage (CFU-GM) cells were used to compare the effect in tumour and normal tissue.

RESULTS

CYC3 was shown to be a specific AK-A inhibitor. Three nanomolar paclitaxel (growth inhibition 50% (GI(50)) 3 nM in PANC-1, 5.1 nM in MIA PaCa-2) in combination with 1 μM CYC3 (GI(50) 1.1 μM in MIA PaCa2 and 2 μM in PANC-1) was synergistic in inhibiting pancreatic cell growth and causing mitotic arrest, achieving similar effects to 10-fold higher concentrations of paclitaxel (30 nM). In CFU-GM cells, the effect of the combination was simply additive, displaying significantly less myelotoxicity compared with high concentrations of paclitaxel (30 nM; 60-70% vs 100% inhibition).

CONCLUSION

The combination of lower doses of paclitaxel and CYC3 merits further investigation with the potential for an improved therapeutic index in vivo.

Abstract 3771: Capecitabine pharmacokinetics and efficacy in spontaneous tumors occurring in a genetically engineered mouse model (GEMM) of pancreatic cancer

Capecitabine (CAP) is an oral fluoropyrimidine, converted sequentially and selectively to 5-FU at the tumour site. It is used in the treatment of a number of cancers as a single agent and in patients with pancreatic cancer, in combination with gemcitabine. However, pre-clinical data in pancreatic cancer models are limited. In this study, we investigated the pharmacokinetics (PK) and efficacy of CAP in a GEMM of spontaneous pancreatic adenocarcinoma (PDA) occurring in KrasG12D; p53R172H; Pdx1-Cre (KPC) mice, compared to an allograft model of a cell line isolated from a PDA arising in the KPC mice. In the PK study, tumour was collected 2 hours after CAP treatment (755 mg/kg by oral gavage), homogenates were analysed using an LC-MS/MS assay developed to simultaneously detect capecitabine and its 3 metabolites DFCR, DFUR and 5-FU (modified from S.M.Guichard, et al., J. Chrom. B. 2005). Data were compared to our previously reported studies in an allograft model (Proc. AACR 2011 a 5446). After a QDx7 treatment, 5-FU concentrations of 27 ± 13 μM were achieved (compared to 23.0 ± 8.1 μM and 22.7 ± 7.7 μM in allograft tumours after 1 and 5 consecutive doses respectively), confirming adequate drug delivery to the in situ tumour following oral administration of CAP. Therefore we proceeded to efficacy studies in this model. In the allograft model we had identified a significant reduction of the tumour doubling time with 755 mg/kg CAP (5 days/week, 3 weeks), compared to control (7.5 ± 3.0 vs 3.5 ± 0.5 days; P<0.001). In in situ tumours, a short term study over 7 days showed a reduction in tumour growth in CAP-treated KPC PDA tumours compared to control (199% ± 22% vs 121% ± 10%; P<0.01). In a survival study in KPC mice, CAP (755 mg/kg, 5 days/week) was compared to the standard treatment for advanced pancreatic cancer, gemcitabine (GEM, 100 mg/kg, Q3D), and there was no difference in the median survival of mice with spontaneous PDA tumours (P=0.61) suggesting a similar efficacy of CAP to GEM. There is conflicting evidence regarding the utility of the combination of GEM and CAP in this disease, so we also investigated the combination in mice bearing allograft PDA tumours. Full doses of both drugs were not tolerated, but the combination of GEM (75 mg/kg Q3D, 2 weeks) plus CAP (539 mg/kg, 5 days/week, 2 weeks) was feasible. This regimen was associated with significant growth inhibition, but this was not superior to GEM alone (75 mg/kg) at the same dose (tumour doubling time: 8.6 ± 10.8 vs 7.2 ± 2.8 days respectively). In summary, orally administered CAP achieves active concentrations of 5-FU in PDA tumours. Similar effects on survival compared to GEM are seen in PDA tumours. Growth inhibition data in allograft tumours did not show any additional benefit for the GEMCAP combination, when compared to GEM alone. CAP could be considered as an alternative to GEM in future, rationally designed, combination treatment strategies.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 3771. doi:1538-7445.AM2012-3771

Abstract 5446: The activity and pharmacokinetics of capecitabine in a mouse model (K8484) of pancreatic adenocarcinoma

The pharmacokinetics (PK) of the oral fluoropyrimidine capecitabine were investigated in a mouse model of pancreatic cancer. Allografts of a cell line (K8484) derived from a pancreatic adenocarcinoma which occurred in a KrasG12D; p53R172H; Pdx1-Cre (KPC) mouse (S. Hingorani, et al., Cancer Cell 2005) were grown subcutaneously in left and right flanks of syngeneic immunocompetent recipient mice. Mice were dosed with either a single dose or 5 daily doses of capecitabine at 755 mg/kg (2.1 mmol/kg/day) with tumour volumes of approximately 250 mm3. Plasma and tissue homogenates were analysed using an LC-MS/MS assay developed to simultaneously detect capecitabine and its 3 metabolites DFCR, DFUR and 5-FU (S.M.Guichard, et al., J. Chrom. B. 2005).

Forty minutes after a single dose the mean plasma concentrations (n=3) were 28 +/- 20 mg/ml capecitabine, 78 +/- 48 mg/ml DFCR, 41 +/- 17 mg/ml DFUR and 0.19 +/- 0.07 mg/ml (1.5 µM) 5-FU, with concentrations of all analytes falling at 2 and 4 hours. In the tumour tissue, mean concentrations at 40 mins were capecitabine: 37 +/- 19 ng/mg tissue, DFCR: 56 +/- 12 ng/mg; DFUR: 20 +/- 8 ng/mg and 5-FU: 3.3 +/- 0.7 ng/mg. At 2 and 4 hours the tumour 5-FU concentrations were 3.0 +/- 1.1 and 1.4 +/- 0.6 ng/mg respectively. In the liver, DFCR concentrations were higher than in the tumour from the same mice but DFUR concentrations were lower and 5-FU was below the limit of quantification (<0.4 ng/mg) in most, consistent with the reported distribution of carboxylesterases, cytidine deaminase and thymidine phosphorylase in tissues (M. Miwa, et al., Eur. J. Cancer 1998). In animals dosed for 5 days, there was no evidence of accumulation of capecitabine or its metabolites in tumour tissue when compared to the single dose. The concentration of tumour 5-FU, ranging from 1.4 to 3.3 ng/mg, is estimated to be equivalent to 11 to 25 µM, more than 10-fold higher than the plasma concentration. The IC50 for 5-FU in K8484 cells, grown in vitro, was 1.4 +/- 0.8 µM suggesting that oral dosing with capecitabine delivers a therapeutically effective dose to the allograft. This was then confirmed in an efficacy study, treating tumour-bearing mice with oral capecitabine for 5 days per week for 3 weeks, which resulted in significant inhibition of tumour growth rate compared to the vehicle treated group, with mean tumour doubling time of 7.5 +/- 3.0 days for capecitabine compared to 3.5 +/- 0.5 days for the vehicle treated group (P < 0.001).

Despite identical germline Kras and p53 genotypes, the KPC PDA and KPC allograft tumour types have previously shown differences in gemcitabine sensitivity in vivo, predicted to be due to differences in drug delivery (K. Olive, et al., Science 2009). Therefore, capecitabine PK and efficacy are now being investigated in the autochthonous tumours arising in the KPC model.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 5446. doi:10.1158/1538-7445.AM2011-5446

Cytochrome P450 gene induction in rats ex vivo assessed by quantitative real-time reverse transcriptase-polymerase chain reaction (TaqMan)

Drug-induced changes in expression of cytochrome P450 (P450) genes are a significant issue in the preclinical development of pharmaceuticals. For example, preclinically, P450 induction can affect safety studies by reducing the systemic exposure of a compound undergoing toxicological evaluation, thus limiting the exposure that can be safely investigated in patients. Therefore, the induction potential of candidate drugs has been studied as part of the drug development process, typically using protein and/or catalytic end points. However, measuring changes in the levels of mRNA using TaqMan technology offers the opportunity to investigate this issue with the advantages of better dynamic range and specific enzyme identification. Here, we describe the TaqMan application to study ex vivo the P450 gene induction in the rat. Initially, livers from rats dosed with the prototypic P450 inducers beta-napthoflavone (BNF), phenobarbital (PB), dexamethasone (DEX), and clofibric acid (CLO) were analyzed for mRNA levels of CYP1A1, 1A2, 2B1, 2B2, 2E1, 3A2, 3A23, and 4A1 and compared with control animals. The maximum fold induction of mRNA varied: 2500-fold for CYP1A1 with BNF, 680-fold for CYP2B1 with PB, 59-fold for CYP3A23 with DEX, and 16-fold for CYP4A1 with CLO. This method was then applied to estimate the inductive potential of putative drug candidates undergoing rodent toxicological evaluation. We present a summary of these data that demonstrates the sensitivity and specificity of the TaqMan assay to distinguish between inducers and noninducers and that offers a highly specific alternative to the quantification of drug effects on P450 expression using immunodetection and substrate metabolism.