Use of NQO1 status as a selective biomarker for oesophageal squamous cell carcinomas with greater sensitivity to 17-AAG

Pan-cancer and single-cell analysis reveal the prognostic value and immune response of NQO1

Background: Overexpression of the NAD(P)H: Quinone Oxidoreductase 1 (NQOI) gene has been linked with tumor progression, aggressiveness, drug resistance, and poor patient prognosis. Most research has described the biological function of the NQO1 in certain types and limited samples, but a comprehensive understanding of the NQO1's function and clinical importance at the pan-cancer level is scarce. More research is needed to understand the role of NQO1 in tumor infiltration, and immune checkpoint inhibitors in various cancers are needed. Methods: The NQO1 expression data for 33 types of pan-cancer and their association with the prognosis, pathologic stage, gender, immune cell infiltration, the tumor mutation burden, microsatellite instability, immune checkpoints, enrichment pathways, and the half-maximal inhibitory concentration (IC50) were downloaded from public databases. Results: Our findings indicate that the NQO1 gene was significantly upregulated in most cancer types. The Cox regression analysis showed that overexpression of the NQO1 gene was related to poor OS in Glioma, uveal melanoma, head and neck squamous cell carcinoma, kidney renal papillary cell carcinoma, and adrenocortical carcinoma. NQO1 mRNA expression positively correlated with infiltrating immune cells and checkpoint molecule levels. The single-cell analysis revealed a potential relationship between the NQO1 mRNA expression levels and the infiltration of immune cells and stromal cells in bladder urothelial carcinoma, invasive breast carcinoma, and colorectal cancer. Conversely, a negative association was noted between various drugs (17-AAG, Lapatinib, Trametinib, PD-0325901) and the NQO1 mRNA expression levels. Conclusion: NQO1 expression was significantly associated with prognosis, immune infiltrates, and drug resistance in multiple cancer types. The inhibition of the NQO1-dependent signaling pathways may provide a promising strategy for developing new cancer-targeted therapies.

Markov neighborhood regression for statistical inference of high-dimensional generalized linear models

High‐dimensional inference is one of fundamental problems in modern biomedical studies. However, the existing methods do not perform satisfactorily. Based on the Markov property of graphical models and the likelihood ratio test, this article provides a simple justification for the Markov neighborhood regression method such that it can be applied to statistical inference for high‐dimensional generalized linear models with mixed features. The Markov neighborhood regression method is highly attractive in that it breaks the high‐dimensional inference problems into a series of low‐dimensional inference problems. The proposed method is applied to the cancer cell line encyclopedia data for identification of the genes and mutations that are sensitive to the response of anti‐cancer drugs. The numerical results favor the Markov neighborhood regression method to the existing ones.

Machine learning based anti-cancer drug response prediction and search for predictor genes using cancer cell line gene expression

Although many models have been proposed to accurately predict the response of drugs in cell lines recent years, understanding the genome related to drug response is also the key for completing oncology precision medicine. In this paper, based on the cancer cell line gene expression and the drug response data, we established a reliable and accurate drug response prediction model and found predictor genes for some drugs of interest. To this end, we first performed pre-selection of genes based on the Pearson correlation coefficient and then used ElasticNet regression model for drug response prediction and fine gene selection. To find more reliable set of predictor genes, we performed regression twice for each drug, one with IC₅₀ and the other with area under the curve (AUC) (or activity area). For the 12 drugs we tested, the predictive performance in terms of Pearson correlation coefficient exceeded 0.6 and the highest one was 17-AAG for which Pearson correlation coefficient was 0.811 for IC₅₀ and 0.81 for AUC. We identify common predictor genes for IC₅₀ and AUC, with which the performance was similar to those with genes separately found for IC₅₀ and AUC, but with much smaller number of predictor genes. By using only common predictor genes, the highest performance was AZD6244 (0.8016 for IC₅₀, 0.7945 for AUC) with 321 predictor genes.

Overcoming acquired resistance to HSP90 inhibition by targeting JAK-STAT signalling in triple-negative breast cancer

Background

Due to the lack of effective therapies and poor prognosis in TNBC (triple-negative breast cancer) patients, there is a strong need to develop effective novel targeted therapies for this subtype of breast cancer. Inhibition of heat shock protein 90 (HSP90), a conserved molecular chaperone that is involved in the regulation of oncogenic client proteins, has shown to be a promising therapeutic approach for TNBC. However, both intrinsic and acquired resistance to HSP90 inhibitors (HSP90i) limits their effectiveness in cancer patients.

Methods

We developed models of acquired resistance to HSP90i by prolonged exposure of TNBC cells to HSP90i (ganetespib) in vitro. Whole transcriptome profiling and a 328-compound bioactive small molecule screen were performed on these cells to identify the molecular basis of acquired resistance to HSP90i and potential therapeutic approaches to overcome resistance.

Results

Among a panel of seven TNBC cell lines, the most sensitive cell line (Hs578T) to HSP90i was selected as an in vitro model to investigate acquired resistance to HSP90i. Two independent HSP90i-resistant clones were successfully isolated which both showed absence of client proteins degradation, apoptosis induction and G2/M cell cycle arrest after treatment with HSP90i. Gene expression profiling and pathway enrichment analysis demonstrate significant activation of the survival JAK-STAT signalling pathway in both HSP90i-resistant clones, possibly through IL6 autocrine signalling. A bioactive small molecule screen also demonstrated that the HSP90i-resistant clones showed selective sensitivity to JAK2 inhibition. Inhibition of JAK and HSP90 caused higher induction of apoptosis, despite prior acquired resistance to HSP90i.

Conclusions

Acquired resistance to HSP90i in TNBC cells is associated with an upregulated JAK-STAT signalling pathway. A combined inhibition of the JAK-STAT signalling pathway and HSP90 could overcome this resistance. The benefits of the combined therapy could be explored further for the development of effective targeted therapy in TNBC patients.

Electronic supplementary material

The online version of this article (10.1186/s12885-019-5295-z) contains supplementary material, which is available to authorized users.

A quantile regression forest based method to predict drug response and assess prediction reliability

Drug response prediction is a critical step for personalized treatment of cancer patients and ultimately leads to precision medicine. A lot of machine-learning based methods have been proposed to predict drug response from different types of genomic data. However, currently available methods could only give a "point" prediction of drug response value but fail to provide the reliability and distribution of the prediction, which are of equal interest in clinical practice. In this paper, we proposed a method based on quantile regression forest and applied it to the CCLE dataset. Through the out-of-bag validation, our method achieved much higher prediction accuracy of drug response than other available tools. The assessment of prediction reliability by prediction intervals and its significance in personalized medicine were illustrated by several examples. Functional analysis of selected drug response associated genes showed that the proposed method achieves more biologically plausible results.

The germline genetic component of drug sensitivity in cancer cell lines

Patients with seemingly the same tumour can respond very differently to treatment. There are strong, well-established effects of somatic mutations on drug efficacy, but there is at-most anecdotal evidence of a germline component to drug response. Here, we report a systematic survey of how inherited germline variants affect drug susceptibility in cancer cell lines. We develop a joint analysis approach that leverages both germline and somatic variants, before applying it to screening data from 993 cell lines and 265 drugs. Surprisingly, we find that the germline contribution to variation in drug susceptibility can be as large or larger than effects due to somatic mutations. Several of the associations identified have a direct relationship to the drug target. Finally, using 17-AAG response as an example, we show how germline effects in combination with transcriptomic data can be leveraged for improved patient stratification and to identify new markers for drug sensitivity.

Little is known about the contribution of germline genetic variants to cancer drug sensitivity. Here, the authors devise an approach for joint analysis of germline variants and somatic mutations, identifying substantial germline contributions to variation in drug sensitivity.

Bayesian Neural Networks for Selection of Drug Sensitive Genes

Recent advances in high-throughput biotechnologies have provided an unprecedented opportunity for biomarker discovery, which, from a statistical point of view, can be cast as a variable selection problem. This problem is challenging due to the high-dimensional and non-linear nature of omics data and, in general, it suffers three difficulties: (i) an unknown functional form of the nonlinear system, (ii) variable selection consistency, and (iii) high-demanding computation. To circumvent the first difficulty, we employ a feed-forward neural network to approximate the unknown nonlinear function motivated by its universal approximation ability. To circumvent the second difficulty, we conduct structure selection for the neural network, which induces variable selection, by choosing appropriate prior distributions that lead to the consistency of variable selection. To circumvent the third difficulty, we implement the population stochastic approximation Monte Carlo algorithm, a parallel adaptive Markov Chain Monte Carlo (MCMC) algorithm, on the OpenMP platform which provides a linear speedup for the simulation with the number of cores of the computer. The numerical results indicate that the proposed method can work very well for identification of relevant variables for high-dimensional nonlinear systems. The proposed method is successfully applied to identification of the genes that are associated with anticancerdrug sensitivities based on the data collected in the cancer cell line encyclopedia (CCLE) study.

Aurora A Functional Single Nucleotide Polymorphism (SNP) Correlates With Clinical Outcome in Patients With Advanced Solid Tumors Treated With Alisertib, an Investigational Aurora A Kinase Inhibitor

Background

Alisertib (MLN8237) is an investigational, oral, selective Aurora A kinase inhibitor. Aurora A contains two functional single nucleotide polymorphisms (SNPs; codon 31 [F/I] and codon 57 [V/I]) that lead to functional changes. This study investigated the prognostic and predictive significance of these SNPs.

Methods

This study evaluated associations between Aurora A SNPs and overall survival (OS) in The Cancer Genome Atlas (TCGA) database. The Aurora A SNPs were also evaluated as predictive biomarkers for clinical outcomes to alisertib in two phase 2 studies (NCT01045421 and NCT01091428). Aurora A SNP genotyping was obtained from 85 patients with advanced solid tumors receiving single-agent alisertib and 122 patients with advanced recurrent ovarian cancer treated with alisertib plus weekly paclitaxel (n = 62) or paclitaxel alone (n = 60). Whole blood was collected prior to treatment and genotypes were analyzed by PCR.

Findings

TCGA data suggested prognostic significance for codon 57 SNP; solid tumor patients with VV and VI alleles had significantly reduced OS versus those with II alleles (HR 1.9 [VI] and 1.8 [VV]; p < 0.0001). In NCT01045421, patients carrying the VV alleles at codon 57 (n = 53, 62%) had significantly longer progression-free survival (PFS) than patients carrying IV or II alleles (n = 32, 38%; HR 0.5; p = 0.0195). In NCT01091428, patients with the VV alleles at codon 57 who received alisertib plus paclitaxel (n = 47, 39%) had a trend towards improved PFS (7.5 months) vs paclitaxel alone (n = 32, 26%; 3.8 months; HR 0.618; p = 0.0593). In the paclitaxel alone arm, patients with the VV alleles had reduced PFS vs modified intent-to-treat (mITT) patients (3.8 vs 5.1 months), consistent with the TCGA study identifying the VV alleles as a poor prognostic biomarker. No significant associations were identified for codon 31 SNP from the same data set.

Interpretation

These findings suggest that Aurora A SNP at codon 57 may predict disease outcome and response to alisertib in patients with solid tumors. Further investigation is warranted.

•

Aurora A contains two single nucleotide polymorphisms (SNPs) at codons 31 and 57 that lead to functional amino acid changes

•

We evaluated the potential prognostic and predictive value of these SNPs and revealed the SNP at codon 57 may predict disease outcome and response to Alisertib in patients with solid tumors

Alisertib, an investigational Aurora A kinase inhibitor, was evaluated in clinical trials and showed clinically meaningful benefit in patients with solid tumors. Two coding region single nucleotide polymorphisms (SNPs) in the Aurora A gene have been reported to be associated with functional changes of the protien. Here we assessed the prognostic and predictive value of Aurora A SNPs in a range of solid tumors. The results suggest that codon 57 SNP may predict disease outcome and response to alisertib in patients. These findings warrant further investigation and may ultimately provide a patient selection strategy for alisertib in certain cancers.

A Robust Model-Free Feature Screening Method for Ultrahigh-Dimensional Data

Feature screening plays an important role in dimension reduction for ultrahigh-dimensional data. In this paper, we introduce a new feature screening method and establish its sure independence screening property under the ultrahigh-dimensional setting. The proposed method works based on the nonparanormal transformation and Henze-Zirkler's test; that is, it first transforms the response variable and features to Gaussian random variables using the nonparanormal transformation and then tests the dependence between the response variable and features using the Henze-Zirkler's test. The proposed method enjoys at least two merits. First, it is model-free, which avoids the specification of a particular model structure. Second, it is condition-free, which does not require any extra conditions except for some regularity conditions for high-dimensional feature screening. The numerical results indicate that, compared to the existing methods, the proposed method is more robust to the data generated from heavy-tailed distributions and/or complex models with interaction variables. The proposed method is applied to screening of anticancer drug response genes.

Azoreductases in drug metabolism

Azoreductases are flavoenzymes that have been characterized in a range of prokaryotes and eukaryotes. Bacterial azoreductases are associated with the activation of two classes of drug, azo drugs for the treatment of inflammatory bowel disease and nitrofuran antibiotics. The mechanism of reduction of azo compounds is presented; it requires tautomerisation of the azo compound to a quinoneimine and provides a unifying mechanism for the reduction of azo and quinone substrates by azoreductases. The importance of further work in the characterization of azoreductases from enteric bacteria is highlighted to aid in the development of novel drugs for the treatment of colon related disorders. Human azoreductases are known to play a crucial role in the metabolism of a number of quinone-containing cancer chemotherapeutic drugs. The mechanism of hydride transfer to quinones, which is shared not only between eukaryotic and prokaryotic azoreductases but also the wider family of NAD(P)H quinone oxidoreductases, is outlined. The importance of common single nucleotide polymorphisms (SNPs) in human azoreductases is described not only in cancer prognosis but also with regard to their effects on the efficacy of quinone drug-based cancer chemotherapeutic regimens. This highlights the need to screen patients for azoreductase SNPs ahead of treatment with these regimens.

LINKED ARTICLES

This article is part of a themed section on Drug Metabolism and Antibiotic Resistance in Micro-organisms. To view the other articles in this section visit http://onlinelibrary.wiley.com/doi/10.1111/bph.v174.14/issuetoc.

P-glycoprotein confers acquired resistance to 17-DMAG in lung cancers with an ALK rearrangement

BACKGROUND

Because anaplastic lymphoma kinase (ALK) is dependent on Hsp90 for protein stability, Hsp90 inhibitors are effective in controlling growth of lung cancer cells with ALK rearrangement. We investigated the mechanism of acquired resistance to 17-(Dimethylaminoethylamino)-17-demethoxygeldanamycin (17-DMAG), a geldanamycin analogue Hsp90 inhibitor, in H3122 and H2228 non-small cell lung cancer cell lines with ALK rearrangement.

METHODS

Resistant cell lines (H3122/DR-1, H3122/DR-2 and H2228/DR) were established by repeated exposure to increasing concentrations of 17-DMAG. Mechanisms for resistance by either NAD(P)H/quinone oxidoreductase 1 (NQO1), previously known as a factor related to 17-DMAG resistance, or P-glycoprotein (P-gp; ABCB1/MDR1) were queried using RT-PCR, western blot analysis, chemical inhibitors, the MTT cell proliferation/survival assay, and cellular efflux of rhodamine 123.

RESULTS

The resistant cells showed no cross-resistance to AUY922 or ALK inhibitors, suggesting that ALK dependency persists in cells with acquired resistance to 17-DMAG. Although expression of NQO1 was decreased in H3122/DR-1 and H3122/DR-2, NQO1 inhibition by dicumarol did not affect the response of parental cells (H2228 and H3122) to 17-DMAG. Interestingly, all resistant cells showed the induction of P-gp at the protein and RNA levels, which was associated with an increased efflux of the P-gp substrate rhodamine 123 (Rho123). Transfection with siRNA directed against P-gp or treatment with verapamil, an inhibitor of P-gp, restored the sensitivity to the drug in all cells with acquired resistance to 17-DMAG. Furthermore, we also observed that the growth-inhibitory effect of 17-DMAG was decreased in A549/PR and H460/PR cells generated to over-express P-gp by long-term exposure to paclitaxel, and these cells recovered their sensitivity to 17-DMAG through the inhibition of P-gp.

CONCLUSION

P-gp over-expression is a possible mechanism of acquired resistance to 17-DMAG in cells with ALK rearrangement.

High expression of NQO1 is associated with poor prognosis in serous ovarian carcinoma

Background

NAD(P)H:quinone oxidoreductase (NQO1) is a flavoprotein that catalyzes two-electron reduction and detoxification of quinones and its derivatives. NQO1 catalyzes reactions that have a protective effect against redox cycling, oxidative stress and neoplasia. High expression of NQO1 is associated with many solid tumors including those affecting the colon, breast and pancreas; however, its role in the progression of ovarian carcinoma is largely undefined. This study aimed to investigate the clinicopathological significance of high NQO1 expression in serous ovarian carcinoma.

Methods

NQO1 protein expression was assessed using immunohistochemical (IHC) staining in 160 patients with serous ovarian carcinoma, 62 patients with ovarian borderline tumors and 53 patients with benign ovarian tumors. Quantitative real-time polymerase chain reaction (qRT-PCR) was performed to detect NQO1 mRNA expression levels. The correlation between high NQO1 expression and clinicopathological features of ovarian carcinoma was evaluated by Chi-square and Fisher’s exact test. Overall survival (OS) rates of all of ovarian carcinoma patients were calculated using the Kaplan-Meier method, and univariate and multivariate analyses were performed using the Cox proportional hazards regression model.

Results

NQO1 protein expression in ovarian carcinoma cells was predominantly cytoplasmic. Strong, positive expression of NQO1 protein was observed in 63.8% (102/160) of ovarian carcinomas, which was significantly higher than in borderline serous tumors (32.3%, 20/62) or benign serous tumors (11.3%, 6/53). Importantly, the rate of strong, positive NQO1 expression in borderline serous tumors was also higher than in benign serous tumors. High expression of NQO1 protein was closely associated with higher histological grade, advanced clinical stage and lower OS rates in ovarian carcinomas. Moreover, multivariate analysis indicated that NQO1 was a significant independent prognostic factor, in addition to clinical stage, in patients with ovarian carcinoma.

Conclusions

NQO1 is frequently upregulated in ovarian carcinoma. High expressin of NQO1 protein may be an effective biomarker for poor prognostic evaluation of patients with serous ovarian carcinomas.