chemical proﬁling of primary mesothelioma cultures …malignant pleural mesothelioma is...

Biology of Human Tumors

Chemical Profiling of Primary MesotheliomaCultures Defines Subtypes with DifferentExpression Profiles and Clinical ResponsesLaurel M. Schunselaar1, Josine M.M.F. Quispel-Janssen2, Yongsoo Kim1,Constantine Alifrangis3,Wilbert Zwart1, Paul Baas2, and Jacques Neefjes4

Abstract

Purpose: Finding new treatment options for patients withmalignant pleural mesothelioma is challenging due to the rarityand heterogeneity of this cancer type. The absence of druggabletargets further complicates the development of new therapies.Current treatment options are therefore limited, and prognosisremains poor.

Experimental Design:We performed drug screening on primarymesothelioma cultures to guide treatment decisions of correspond-ingpatients thatwereprogressive afterfirst- or second-line treatment.

Results: We observed a high concordance between in vitroresults and clinical outcomes. We defined three subgroupsresponding differently to the anticancer drugs tested. In addition,

gene expression profiling yielded distinct signatures that segre-gated the differently responding subgroups. These genes signa-tures involved various pathways, most prominently the fibroblastgrowth factor pathway.

Conclusions: Our primary mesothelioma culture system hasproved to be suitable to test novel drugs. Chemical profiling ofprimary mesothelioma cultures allows personalizing treatmentfor a group of patients with a rare tumor type where clinicaltrials are notoriously difficult. This personalized treatmentstrategy is expected to improve the poor prospects of patientswith mesothelioma. Clin Cancer Res; 24(7); 1761–70. �2017 AACR.

See related commentary by John and Chia, p. 1513

IntroductionMalignant pleuralmesothelioma (MPM) is a rare but aggressive

tumor arising from mesothelial cells in the pleural cavity. Itusually presents with pain or dyspnea caused by pleural fluid orshrinkage of the hemithorax (1). Palliative chemotherapy con-sisting of a platin and antifolate combination is consideredstandard of care and gives a modest survival advantage of around3 months (2). Further systemic treatment can be offered to fitpatients, but thus far, studies in second line have failed to detect asurvival benefit. Response rates in different second-line therapiesrange between 0% and 20% (3), which urges the need for moreeffective treatments.

Using genetic profiling to define drivers in cancer amendableto targeting by small molecular drugs has been successful in othertypes of tumors. MPM, however, has only a few mutations, and

none of these present as a likely target for therapy. Most geneticmutations found inMPMare loss of tumor suppressor genes, suchas CDKN2A, NF2, and BAP1, rather than activation of oncogenes(4). The absence of druggable molecular targets in MPM hindersthe development ofmore dedicated and effective therapies (5–9).

Basedonhistology, three typesofmesotheliomaare recognized:an epithelioid, a sarcomatoid, and a biphasic or mixed type (10).Epithelioid mesothelioma comprises the largest group and has abetter outcome than the sarcomatoid and mixed type. Regardingresponse to treatment, epithelioid mesothelioma is a heteroge-neous disease. To increase the effectiveness of current therapies, itis vital to find ways to more accurately profile this group ofpatients for personalized treatment and new therapeutic options.

Long-established cell lines are commonly used for in vitro drugscreens to select compounds for further clinical development(11). However, their resemblance to primary tumors is question-able because cells change pheno- and genotypically during theiradaptation to tissue culture conditions (12–15). This can have aprofound influence on their responses to anticancer drugs(16, 17). The use of cell lines in drug development programs didnot yield any active drugs for patients with mesothelioma. Oneexample is the VANTAGE-014 trial, which was based on positiveresults from established cell lines (18). This study exemplifies thedifficulty of conducting clinical trials in a rare disease like meso-thelioma (19). In this placebo-controlled trial that evaluated thehistone deacetylase (HDAC)–inhibitor vorinostat in second orthird line, the time to accrue 661 patients with mesotheliomafrom 90 international centers was 6 years. Unfortunately, therewas no clinical benefit from treatment with vorinostat in this verylarge study (20). This trial stresses the need for in vitro drug testingconditions that reflect genuine mesothelioma tumors more accu-rately. Primary mesothelioma cultures may provide a valuable

1Division of Oncogenomics, Netherlands Cancer Institute, Amsterdam, theNetherlands. 2Department of Thoracic Oncology, Netherlands Cancer Institute,Amsterdam, the Netherlands. 3Wellcome Trust Sanger Institute,Wellcome TrustGenome Campus, Hinxton, United Kingdom. 4Department of Chemical Immu-nology, Leiden University Medical Center, Leiden, the Netherlands

Note: Supplementary data for this article are available at Clinical CancerResearch Online (http://clincancerres.aacrjournals.org/).

L.M. Schunselaar and J.M.M.F. Quispel-Janssen contributed equally to thisarticle.

Corresponding Author: Paul Baas, Netherlands Cancer Institute, Plesmanlaan121, 1066 CXAmsterdam, the Netherlands. Phone: 31-20-512-9111; Fax: 31-20-512-2572; E-mail: [email protected]

doi: 10.1158/1078-0432.CCR-17-1345

�2017 American Association for Cancer Research.

ClinicalCancerResearch

www.aacrjournals.org 1761

on June 13, 2020. © 2018 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst October 24, 2017; DOI: 10.1158/1078-0432.CCR-17-1345

http://crossmark.crossref.org/dialog/?doi=10.1158/1078-0432.CCR-17-1345&domain=pdf&date_stamp=2018-3-7

http://clincancerres.aacrjournals.org/

model for personalized drug selection for patients with mesothe-lioma because they recapitulate the original tumor far moreaccurately than long-established MPM cell lines (21, 22).

We established a method of profiling primary mesotheliomacultures with commonly used anticancer drugs and validated theresults in corresponding patients. We distinguished three groups,not by means of genetic parameters, but based on the drugresponse patterns, which are ultimately more relevant to thepatient. We found that the three "chemical" profiles were asso-ciated with three distinct gene expression profiles relating to theFGFR pathway. Indeed, FGFR inhibition blocked proliferation ofprimary mesothelioma cultures, providing proof of concept ofchemical profiling as a method to reveal novel sensitivities totargeted agents.

Materials and MethodsPatients

All patients provided written informed consent for the use andstorage of pleural fluid, tumor biopsies, and germline DNA.Separate informed consent was obtained to use the informationfrom the drug screens for making treatment decisions. The studywas conducted in accordancewith theDeclaration ofHelsinki andapproved by Netherlands Cancer Institute review board. Diagno-sis was determined on available tumor biopsies and confirmed bythe Dutch Mesothelioma Panel, a national expertise panel ofcertified pathologists who evaluate all patient samples suspectedof mesothelioma.

Culture methodShort-term primary mesothelioma cultures were generated by

isolating tumor cells from pleural fluid. Within half an hour afterdrainage, the pleural fluid was centrifuged at 1,500 rpm for 5minutes at room temperature (RT). When the cell pellet washighly contaminated with erythrocytes, it was incubated witherythrocyte lysis buffer (containing 150 mmol/L NH4Cl, 10mmol/L potassium bicarbonate, and 0.2 mmol/L EDTA) for 10minutes at RT. Cells were resuspended in Dulbecco's ModifiedEagle Medium (DMEM; Gibco) supplemented with penicillin/streptomycin and8% fetal calf serum. The cells were seeded in T75flasks at a quantity of 10 � 106, 15 � 106, or 20 � 106 cells and

incubated at 37�C at 5% CO2. Medium was refreshed dependingon metabolic activity of the cells, usually twice a week. Cells werecultured for a maximum period of 4 weeks.

Comparative genome hybridizationTo ensure that our cultures consisted mainly of tumor cells, we

performed comparative genome hybridization (CGH) on a num-ber of cultures. CGHwas performed as described by Schouten andcolleagues (23). Tumor DNA was labeled with Cy3, and femalepooled reference DNA (G1521; Promega) was labeled with Cy5using the Enzo labeling kit for BAC arrays (ENZ-42670; Enzo LifeSciences). Unincorporated nucleotides were removed with theQIAGEN MinElute PCR Purification Kit (28004; QIAGEN). Sub-sequently, tumor and reference DNA were pooled and pelletedusing an Eppendorf Concentrator (5301; Eppendorf). The pelletswere resuspended in hybridization mix (NimbleGen Hybridiza-tion Kit; Roche NimbleGen) and the sample loaded on the array.Hybridization was at 42�C for 40 to 72 hours (Maui Hybridiza-tion System; BioMicro Systems). Slides were washed three times(Roche NimbleGenWash Buffer Kit) and scanned at 2 mmdoublepass using Agilent'sHigh-ResolutionMicroarray Scanner (scannermodel: G2505C; Agilent Technologies). The resulting imagefiles were further analyzed using NimbleScan software (RocheNimblegen). Grids were aligned on the picture manually andper-channel pair files generated. The NimbleScan DNACopyalgorithm was applied at default settings, and the unaveragedDNACopy text files were used for further analyses.

Drug screensDrug screens were performed in biological duplicate after 1 and

2 weeks of culture. Seven single agents (cisplatin, carboplatin,oxaliplatin, vinorelbine, gemcitabine, pemetrexed, and doxoru-bicin) and five combinations (cisplatinþ pemetrexed, cisplatinþgemcitabine, carboplatin þ pemetrexed, oxaliplatin þ gemcita-bine, and oxaliplatinþ vinorelbine) were used. Cells were seededin flat-bottom, 96-well plates at a density of 5,000 cells per well.After overnight incubation, chemotherapeutics in a concentrationrange of 50 mmol/L to 5 nmol/L were added in technical tripli-cates. After 72 hours of incubation with the drugs, the cytotoxicitywas measured with a metabolic activity assay (CellTiter-Blue;G8081; Promega). Fluorescent readout was performed with theEnVision multilabel reader (PerkinElmer).

Interpretation dose–response curvesClassification of cultures in three groups. The classification ofcultures in three groups was based on results from all drugs anddrug combinations screened. For three concentrations (10 nmol/L,1mmol/L, and50mmol/L), cell survival cutoffwasdetermined.Cellsurvival cutoff for a drug concentration of 10 nmol/L was set at�90% cell survival, 1 mmol/L at �70%, and 50 mmol/L at �50%.For each concentration, the number of drugs above the cutoff valuewas counted. A culture was defined as nonresponsive when for allthree concentrations, five or more drugs were above the cellsurvival cutoff value. A culture was defined as an intermediateresponder when for one or two concentrations, five or more drugswere above thecell survival cutoff value.A culturewas classifiedasaresponder when for all concentrations, less than five drugs wereabove the cell survival cutoff value.

In vitro response prediction. An in vitro response prediction wasmade for each drug or drug combination individually. The in vitro

Translational Relevance

Mesothelioma, or asbestos cancer, is a tumor with a poorprognosis. Three mesothelioma subtypes have been definedbased onmorphology, and no effective treatment is available.Here, we describe a system allowing the culture of primarymesothelioma cells for drug testing and genetic analyses. Onthe basis of drug sensitivities, we define three new mesothe-lioma subtypes with a concomitant different gene expressionprofile, including the fibroblast growth factor (FGF) pathway.Translating the results of the primary cultures for the treatmentof a small set of patients correctly predicted clinical responses.Chemical profiling of patients with mesothelioma allowsidentification of subgroups separated by the feature mostrelevant to patients—drug responses. The correspondinggenetic analysis identifies the FGF pathway for targeting indefined mesothelioma subgroups.

Schunselaar et al.

Clin Cancer Res; 24(7) April 1, 2018 Clinical Cancer Research1762




responsewas correlated to the clinical response definedbyRECISTmodified for mesothelioma, thereby identifying patients withprogressive disease, stable disease, and partial response. A testset of dose–response curves was used to determine cutoff pointsfor areaunder the curve (AUC) values topredict clinical responses.Very low or very high drug concentrations were not expected to beclinically relevant. Therefore, the AUC was determined in aconcentration range of 50 to 5,000 nmol/L (GraphPad Prism).An AUC level of less than 1,485 predicted a partial response. AnAUC level higher than 2,970 predicted progressive disease. AllAUC levels between these numbers predicted stable disease.

RNA isolation. Total RNA was extracted using TRIzol Reagent(15596-018; Ambion life technologies) according to the manu-facturer's protocol. Typically, 1mLof TRIzol Reagent was used per1 � 106 cells. The total RNA pellet was air dried for 8 minutes,dissolved in an appropriate volume of nuclease-free water(AM9937; Ambion life technologies), and quantified usingNanoDrop UV-VIS Spectrophotometer. Total RNA was furtherpurified using the RNeasy MinElute Cleanup Kit (74204;QIAGEN) according to the manufacturer's instructions. Qualityand quantity of the total RNA were assessed by the 2100 Bio-analyzer using a Nano Chip (Agilent Technologies). Total RNAsamples having an RNA integrity number (RIN) >8 were subject-ed to library generation.

RNA sequencing. Strand-specific libraries were generated using theTruSeq Stranded mRNA sample preparation kit (RS-122-2101/2;Illumina) according to the manufacturer's instructions (Part #15031047, Revision E). The libraries were analyzed on a 2100Bioanalyzer using a 7500 chip (Agilent Technologies), diluted andpooled equimolar into a 10-nmol/Lmultiplexed sequencing pool,and stored at �20�C. The libraries were sequenced with 65-bppaired-end reads on a HiSeq2500 using V4 chemistry (Illumina).

Gene expression analysis. The raw sequencing data were aligned toa human reference genome (build hg38) using TopHat 2.0, fol-lowed by measuring gene expression using our own protocolbased on htseq count (Icount). Normalized count per million(CPM) was measured using library sizes corrected with trimmedmean of M-values (TMM) normalization with the edgeR package(24). For differential expressed gene (DEG) identification, weused voom transformation (25) followed by empirical Bayesmethod with the limma R package. Then, DEGs were identifiedas the genes with P values less than 0.005 and log2 fold changeslarger than 2. The voom-transformed log-CPM of DEGs was usedin principal component analysis (PCA). For heatmap generation,voom-transformed log-CPM of DEGs was standardized by meancentering and scaling with standard deviation. Genes wereordered based on hierarchical clustering with Pearson correlationas a similarity measure and Ward linkage. ID number and corre-sponding fold changes of DEGs were uploaded in ingenuitypathway analysis (IPA; QIAGEN Bioinformatics). Analysis wasperformed with 224 mapped IDs.

Stability assessment of differential gene expression analysis. Toassess the reliability of DEGs, we performed differential expres-sion analysis with leaving out each of the responders and non-responders. TheP values and rankings ofDEGs thatwere obtainedwith this analysis were used in the downstream analysis. Further,for each of the held-out experiments, we obtained DEGs using

same P values and fold-change cutoffs. For each of the DEG lists,hierarchical clustering analysis was performed, after which con-sensus of the clustering was obtained.

ResultsProfiling and characterization of primary mesotheliomacultures

Between February 2012 and July 2016, 155 pleural fluids from102 patients with a confirmed histologic diagnosis of mesothe-lioma were collected for early passage primary cultures. Eighty-nine patients (87%) were male, the mean age was 67 years, andmost patients had an epithelial subtype similar to the conven-tional distribution of mesothelioma subtypes. Forty-one patientswere chemotherapy na€�ve at the time of cell isolation, and 61patients had received one or more lines of treatment (Supple-mentary Table S1A). Figure 1A shows a flow chart of the pleuralfluid pipeline depicting in vitro drug testing and subsequentclinical testing in patients. Eighty-one of the 155 isolations weresuitable for further culture and drug screening, resulting in a takerate of 52%. These 81 isolationswere derived from57patients.Wefailed to perform a drug screen for 45 patients. Patients' char-acteristics for both groups are given in Supplementary Table S1Band S1C. There was no significant difference between the twogroups for age (P ¼ 0.05), prior lines of treatment (P ¼ 0.54), orhistology (P ¼ 0.42). There was a significant difference in gender(P¼0.03), however, the number of female patientswas too low tomake conclusions about any effect of gender on success rate.Failure was mainly due to too low tumor cell count isolated fromthe pleural fluid. The time between isolation of pleural fluid andthe start of the first drug screen was generally 1 week. A biologicalduplicate screen was performed in the following week (Fig. 1B).

Because culturesmay changeover time,we assessed the stabilityof our cultures using CGH. Although mesothelioma is generallycharacterized by very fewmutations, they frequently show loss ofthe gene CDKN2A, located at the p16 locus on chromosome 9(26–28). This can be detected by CGH. There was no deletion ofthe p16 locus detected in samples of 2 patients. In the pleuralfluid of 3 other patients, deletion of the p16 locus was detectedin the first culture passages. At later passages, this deletion couldnot be detected anymore in 2 of the 3 patients. Because deletionscannot be repaired spontaneously, this suggests overgrowth ofreactive mesothelial cells co-isolated with the mesotheliomacells (Supplementary Fig. S2). These experiments validated theisolation and culture of primary mesothelioma cells and showedthat drug screens should be performed during the first 3 weeksafter isolation from patients, before overgrowth of other cellscould be expected.

Chemical profiling identifies three mesothelioma subgroupsDrug screeningwas performedon81different primary cultures,

with compounds selected on the basis of their current or historicaluse for treatment of patients with mesothelioma (2, 29–33).Cisplatin, carboplatin, oxaliplatin, gemcitabine, vinorelbine,pemetrexed, and doxorubicin have been tested as single agentand/or in combination. The different cultures showed markeddifferences in the dose–response profiles. This allowed clusteringof the primary cultures in three different groups: "responders,""nonresponders," and "intermediate responders" (see "Materialsand Methods"). The clustering is based on all drugs anddrug combinations screened. We defined a "responder" as aculture responding to most of the chemotherapeutics screened

Chemical–Genetic Profiling of Primary Mesothelioma Cultures

www.aacrjournals.org Clin Cancer Res; 24(7) April 1, 2018 1763




(Fig. 2A and Supplementary Fig. S3A). We defined a "nonrespon-der" as a culture failing to respond to more than five of the drugsscreened (Fig. 2B and Supplementary Fig. S3B). An "intermediateresponder" responded to some of the drugs, but not all of them,and visually did not fit into one of the other two categories(Fig. 2C and Supplementary Fig. S3C). From the 81 cultures, sixcultures classified as "responder," 27 as "nonresponder," and 48as "intermediate responder." Thirty-one drug screens were per-formed on chemo-naive cells. Fifty drug screens were performedon cells from patients who received one or more lines of treat-ment. The clustering in the three groups was not significantlydifferent for cells isolated from patients who had or had notreceived prior treatment (P ¼ 0.72; Supplementary Table S4A).These data suggested that primary mesothelioma cultures allowsubdivision of tumors based on drug sensitivity without signif-icant effects of earlier treatments of the corresponding patients.

Transcriptomic analyses reveal distinct genomic subclassesthrough chemical profiles

Between primary mesothelioma cultures, divergent responsesto chemotherapeutic intervention were observed. To test whetherthere was a genomic basis for these three groups identified bychemical profiling, we performed RNA sequencing (RNA-Seq) on20 primary mesothelioma samples, taken immediately after iso-lation and representing four responder samples, nine nonre-sponder samples, and seven samples from the intermediategroup. We first identified a set of DEGs between responders andnonresponders with P values less than 0.005 and log2 foldchanges larger than 2 (see "Material and Methods"). A total of133 genes were downregulated, and 152 genes were upregulatedin the responder group compared with the nonresponder group(Supplementary Table S5). Indifferential gene expression analysiswith leave-one-out cross-validation, we confirmed that the 285

Figure 1.

Flow chart and timeline of the chemical and genetic profiling of primarymesothelioma cultures.A, Flow chart of the pleural fluid pipeline. Pleural fluid was extractedfrom 102 patients diagnosed with mesothelioma. The cultures were diagnosed with pathology, and primary cultures were made. Twenty primary tumorcultureswere genetically profiled. Eighty-one cultureswere suitable for drug screening. The results from 11 drug screenswere used inpatient treatment.B,Timeline ofdrug screens using primary mesothelioma cultures. The first screen was started within 10 days after isolation (day 0), and the biological duplicate screen wasperformed within 1 week after the first screen. The drug screening assays took 5 days, and primary cultures were analyzed within 3 weeks after cell isolation fromthe pleural fluid.

Schunselaar et al.





DEGs were consistently highly ranked, and the cutoffs (P < 0.005and log2 fold changes >2) provided genes that stably separatedpatients by response (Supplementary Fig. S6). The intermediategroup shows a signature that differs from both responders andnonresponders, also genetically defining it as a separate group(Fig. 3A).We observed the same trend in PCA on expression levelsof DEGs (Fig. 3B; "Materials and Methods"). IPA on DEGsrevealed 10 networks containing at least seven DEGs. The topnetwork with 23 DEGs contained the fibroblast growth factor(FGF) pathway (Fig. 3C). FGF9 was significantly upregulated inthe nonresponder group (Fig. 3D). Because this pathway has beendescribed previously in MPM (34), we analyzed gene expressionof the preferred receptors for FGF9: FGFR3 and FGFR1. Geneexpression of these receptors was also upregulated in the nonre-sponder group (Fig. 3D). The paired-end RNA-Seq analysis didnot reveal mutated expressed genes.

To test the relevance of the various components of the FGFpathway, primary mesothelioma cultures were exposed to com-pound PD-173074, an FGFR inhibitor with a high affinity forFGFR3 and FGFR1. Two nonresponder primary mesotheliomacultures were sensitive to the FGFR inhibitor (Fig. 3E). In meso-thelioma cell lines, we also found a statistically significant corre-lation between elevated FGF9 mRNA expression and IC50 toPD-173074 (P ¼ 0.0117). These experiments show thatchemical profiling of primary mesothelioma cultures allowsidentification of subgroups that are characterized by different

expression profiles. In addition, new targets for treatment ofmesothelioma subgroups can be identified, as is illustrated herefor the FGF pathway.

Clinical implication of in vitro drug screensTo study the correlation between in vitro drug screens and

clinical outcome, we quantified drug sensitivity by calculatingthe AUC values of dose–response curves. The AUC was deter-mined in a concentration range between 50 and 5,000 nmol/L.Lower or higher concentrationswere not expected tobe clinicallyrelevant. In vitro response was determined for each drug or drugcombination and was classified as one of the following theclinical responses: partial response, stable disease, or progressivedisease. Figure 2D illustrates dose–response curves for the druggemcitabine in three different patients. The boxes indicate theAUC in which progressive disease, stable disease, and partialresponse were predicted. We treated 10 patients who wereprogressive after first- or second-line treatment, with the drugthat was most effective based on the in vitro drug screen that wasperformed on the patient's primary mesothelioma cells(Table 1). Patient 1 was a 61-year-old woman with an epithe-lial-typemesothelioma. Her frontline treatment consisted of thestandard first-line combination of cisplatin and pemetrexed,which was followed by a surgical procedure consisting of apleurectomy/decortication. Upon progression, the in vitro drugscreen demonstrated oxaliplatin and vinorelbine as the most

Figure 2.

Dose–response curves for various drugs depicted for the differently responding subgroups. A–C, Dose–response curves of a responder, a nonresponder, and anintermediate responder are shown, as indicated. Drug screens were performed on chemo-naive cells. Survival (mean � SD) is shown in relation to increasingconcentrations of single agents and combinations, as indicated. D, Dose–response curves for the drug gemcitabine screened in three different patients: aresponder (green), an intermediate responder (blue), and a nonresponder (red). Boxes indicate the AUC fromwhich progressive disease (red), stable disease (blue),and partial response (green) are predicted. The AUC surface is pictured in the trend of the gemcitabine curves.






Figure 3.

Gene expression profiling of the differently responding mesothelioma subgroups. A, Heatmap showing 285 genes that are differentially expressed betweenresponders and nonresponders. Greenbars depict genes that are downregulated, whereas red bars depict upregulated genes in nonresponders. The gene expressionprofile of the intermediate group is different from the expression profile of responders and nonresponders. The list of genes is shown in Supplementary Table S2.B, PCA separates responders (red) from nonresponders (green). The intermediate group (black) is located between these groups. C, IPA illustrating the mostsignificant network containing 23 DEGs between responders and nonresponders. Green: upregulated, red: downregulated DEGs in nonresponders. D, Boxplotdepicting gene expression of FGF9 and interaction partners FGFR1 and FGFR3 in responders (red), nonresponders (green), and intermediate responders (black). Thelevel of gene expression is indicated on the y-axis. Boxplot showsmean expression level with 75th (top) and 25th (bottom) percentile value.Whiskers indicate rangeof values. E, Dose–response curves of two nonresponder cultures and reference cell lines NCI-H28 and H2810, treated with increasing concentrations of the FGFRinhibitor PD-173074. Cell viability is measured.

Schunselaar et al.





effective compounds, and we predicted a partial response(Fig. 4A; patient 1). She was treated accordingly, resulting ina partial response, as is shown in Fig. 4B. The second patient, a52-year-old male with epithelial mesothelioma, was treatedwith cisplatin and pemetrexed, followed by a pleurectomy/decortication. Progression occurred 7 months after completionof his first-line therapy. The combination of oxaliplatin andgemcitabine was the most effective one, and stable diseasewas predicted (Fig. 4A; patient 2), which was indeed observedafter clinical treatment with these drugs (Fig. 4B). Patient 3, a36-year-old female patient with a mixed type of mesothelioma,had disease progression 4 months after her initial treatmentwith cisplatin, pemetrexed, and a pleurectomy/decortication.The in vitro drug screen showed a nonresponder profile, andprogressive disease was to be expected from treatment (Fig. 4A;patient 3). She was treated with consecutive courses of the bestcombination observed (carboplatin/gemcitabine and oxalipla-tin/vinorelbine) but experienced disease progression after twocourses of each combination (Fig. 4B) and died shortly there-after. In vitro drug screen results and CT scans before and aftertreatment for patients 4 to 10 are depicted in SupplementaryFig. S7. For patients 8 to 10, in vitro response predictioncorrelated with the actual patient response. For patients 4, 6,and 7, the patient response was better than predicted. Patient 5,a 71-year-old man with epithelial mesothelioma, was treatedtwice based on his chemosensitivity screen. After front-linetreatment with carboplatin and pemetrexed, he was first treatedwith gemcitabine and later with vinorelbine. The clinicalresponse for both treatments was stable disease. For gemcita-bine, this was predicted based on the in vitro screen. Forvinorelbine, however, the observed response was not as pro-nounced as was expected based on the in vitro results (Supple-mentary Fig. S7). For patient 6, vinorelbine was selected as thebest option to which oxaliplatin was added. Patients 7, 9, and10 did not receive the most potent drug based on the in vitrodrug screen because of contraindications for treatment withdoxorubicin. Due to the patients' history, vinorelbine or acombination with vinorelbine could not be given. From 11drug screens, seven in vitro response predictions were correct.For the four that were not correctly predicted, the actual clinicalresponse was better in 3 patients. These results suggest that thein vitro drug screens had added value in predicting actualindividual patient responses to selected drugs.

DiscussionCancer treatment strategies are changing from general therapy

regimens to more personalized treatment, often based on thegeneticmake-up of the tumor. Unfortunately, no druggable drivermutations have been identified in mesothelioma (5, 6, 8, 9, 35).Therefore, we "chemically" profiled primary mesothelioma cul-tures with common chemotherapeutic drugs and subsequentlytreated 10 patients with the most effective drug or drug combi-nation. This strategy has previously been successfully applied inlung cancer (36–38), ovarian cancer (39, 40), and breast cancer(41) and showed that in vitro drug responsiveness bears clinicallyrelevant information for patient treatment efficacy.

For the patients treated in this study, we observed considerableoverlap between the predicted drug responses in vitro and thecorresponding clinical responses. Although the number ofpatients is too small to make definite conclusions, we present asystem that can personalize the treatment of patients with meso-thelioma, a heterogeneous disease, with a limited number ofpatients available for clinical trials and only one registered sys-temic therapy option.

In addition to predicting the best chemotherapeutic option foran individual patient, we identified "chemical profiles" corre-sponding to gene signatures that distinguished tumors resistant tomost tested therapeutics from tumors thatwere largely responsive.A third group with intermediate responses to drugs had anexpression profile that was different from the responding andnonresponding groups. We expected that drug screens performedon chemo-naive cells would give a different chemosensitivityprofile comparedwith drug screens performed on pretreated cells.However, no significant differences were detected in the threechemical profiles between these groups. This corresponds to theresults of Mujoomdar and colleagues, who described similarresults for chemo-naive and pretreated biopsies treated in vitrowith three single agents (42).

The different chemical profiles that we identified could not havebeen identified based on pathology without prior knowledge. Incancer types like prostate and breast cancer, gene expressionprofiles were successfully used to define subclasses. These wereusually retrospectively correlatedwithprognostic features (43, 44),although one such a profile—the 70-gene signature in breastcancer—has recently been validated on the basis of a prospectivestudy (45). Our prospectively determined chemical profiles

Table 1. Overview of patients treated based on their in vitro drug screen

Patient Gender Histology DrugIn vitro–predictedresponse

Patientresponse

1 F Epithelial Oxaliplatin þ vinorelbine PR PR2 M Epithelial Oxaliplatin þ gemcitabine SD SD3 F Mixed Oxaliplatin þ vinorelbine PD PD4 M Epithelial Oxaliplatin þ gemcitabine SD PR5–1 M Epithelial Gemcitabine SD SD5–2 Vinorelbine PR SD6 M Epithelial Oxaliplatin þ vinorelbine PD SD7 M Epithelial Oxaliplatin þ gemcitabine PD PR8 M Epithelial Doxorubicin SD SD9 M Epithelial Oxaliplatin þ gemcitabine PD PD10 M Epithelial Oxaliplatin þ gemcitabine SD SD

NOTE: Ten patients were treated based on their in vitro drug screen. Gender, histology, chemotherapeutic given, in vitro response prediction, and actual patientresponse are given. Patient 5 was treated twice based on his in vitro drug screen.Abbreviations: F, female; M, male; PD, progressive disease; PR, partial response; SD, stable disease.






have predictive value, which—from the patients' perspective—isthe most important factor and clinically more relevant thanprognostic values.

Of note, there are some limitations to our pipeline. The drugscreening system was unable to test pemetrexed. Pemetrexed isan antifolate that inhibits multiple enzymes involved in theformation of nucleotides (46–49). Pemetrexed activity isreduced by folate (46, 47, 50, 51). The culture medium used inthis system contained folate, probably at supraphysiologiclevels. Serum also contains a variety of folate, nucleosides, andnucleotides and is expected to circumvent growth inhibition bypemetrexed (46, 52). The presence of folate, nucleosides, andnucleotides in the culture system could explain why primarycultures were not sensitive to pemetrexed. Another limitation ofthe system is that the culture does not include pharmacokineticsand dynamics of the different drugs. Every cell-based model lacks

features of the original tumor-like vasculature and tumor micro-environment, which makes it impossible to simulate pharmaco-kinetics and pharmacodynamics. On logical grounds, our systemalso cannot be used for the testing of the recently introducedclasses of immuno-oncology drugs. Our in vitro responseprediction method is arbitrary and expanding with more patientswould provide data to further define cutoffs for better drug re-sponse prediction.

Thus far, we have tested only chemotherapeutics that arecommonly used in clinical practice because these allowed vali-dation of the results in patients with mesothelioma. By furtherexpanding the number and classes of compounds in thedrug screen, we may be able not only to further characterizethe more heterogeneous intermediate group but also toidentify more suitable therapeutic options for the nonresponderpatient population.

Figure 4.

Dose–response curves and clinical responses of 3 patients. A, Dose–response curves of primary mesothelioma cells isolated from patients 1 to 3 and treated withseveral single agents and combinations of cytotoxic drugs, as indicated. Cell viability measured after 72 hours of drug exposure as a function of increasingconcentrations of several drugs and combinations is depicted. B, CT scans of patients 1 to 3 before and after treatment with the drugs selected based on the in vitrodrug screens. Response evaluation was done using modified RECIST for mesothelioma. Colored boxes around CT scans indicate in vitro response predictionbefore treatment and the actual response after treatment. Green: partial response, yellow: stable disease, red: progressive disease. Patient 1 was treated with acombination of oxaliplatin and vinorelbine. The tumor rind indicated by the red line is irregular on her pretreatment scan and is smaller and smootheron her scan following treatment, indicating a partial response. Patient 2 received a combination of oxaliplatin andgemcitabine. The tumor nodule indicated by the redarrow remains similar between the scans, indicating stable disease. Patient 3 received successively carboplatin/gemcitabine and oxaliplatin/vinorelbine.The gray tumor rind on the pretreatment scan—encircled by the red line—is larger on the posttreatment scan, which illustrates progressive disease.

Schunselaar et al.





Our model will enable us to select drugs or drug combinationsthat are more likely to give a response in subgroups of patients.Because mesothelioma is a rare tumor type, such subgroupswould probably not have been detected in clinical trials. Prese-lection of drugs and patients will help to optimize the design andsuccess of clinical trials in this patient group.

We already have one example of a new drug selected on thebasis of our method. Based on gene expression profiling, the FGFpathway appeared upregulated in the nonresponder patient pop-ulation, for whom at this stage no active therapeutic options areavailable. Deregulated FGF signaling has been linked to cancerpathogenesis (53), and several groups have reported involvementof the FGF signaling cascade in mesothelioma (34, 54). Becausethis pathway appeared selectively upregulated in the nonrespond-er patient population, preselected patients may derive specificbenefit from therapeutic intervention using FGFR inhibitors, aswe successfully illustrate in our primary cultures (Fig. 3E). Chem-ical profiling of primary mesothelioma cultures revealed threeresponse groups corresponding to distinct gene signatures involv-ing the FGF signaling cascade. We demonstrated considerableoverlap between in vitro and in vivo responses, suggesting that ourpipeline represents a feasible method to personalize treatmentthat could ultimately improve the prospects of patients withmesothelioma.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design: L.M. Schunselaar, J.M.M.F. Quispel-Janssen, J. NeefjesDevelopment of methodology: L.M. Schunselaar, J.M.M.F. Quispel-Janssen,J. NeefjesAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): L.M. Schunselaar, J.M.M.F. Quispel-Janssen,C. Alifrangis, P. BaasAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): L.M. Schunselaar, J.M.M.F. Quispel-Janssen, Y. Kim,J. NeefjesWriting, review, and/or revision of the manuscript: L.M. Schunselaar,J.M.M.F. Quispel-Janssen, Y. Kim, C. Alifrangis, W. Zwart, J. NeefjesAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): L.M. Schunselaar, J.M.M.F. Quispel-JanssenStudy supervision: W. Zwart, P. Baas, J. Neefjes

AcknowledgmentsThe authors would like to acknowledge the NKI-AVL Core Facility Molecular

Pathology and Biobanking (CFMPB) and the NKI-AVL Genomics Core Facilityfor supplyingmaterial and lab support. The authors thankUltanMcDermott forhis data input. Thisworkwas supportedby grants from theDutchCancer Society(KWF) assigned to P. Baas and J. Neefjes and a gravity program "Institute ofChemical Immunology" assigned to J. Neefjes.

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

Received May 10, 2017; revised August 25, 2017; accepted October 19, 2017;published OnlineFirst October 24, 2017.

References1. Baas P, Fennell D, Kerr KM, Van Schil PE, Haas RL, Peters S, et al. Malignant

pleural mesothelioma: ESMO Clinical Practice Guidelines for diagnosis,treatment and follow-up. Ann Oncol 2015;26:v31–9.

2. Vogelzang NJ, Rusthoven JJ, Symanowski J, Denham C, Kaukel E, Ruffie P,et al. Phase III study of pemetrexed in combination with cisplatin versuscisplatin alone in patients with malignant pleural mesothelioma. J ClinOncol 2003;21:2636–44.

3. Buikhuisen WA, Hiddinga BI, Baas P, van Meerbeeck JP. Second linetherapy in malignant pleural mesothelioma: a systematic review. LungCancer 2015;89:223–31.

4. Guo G, Chmielecki J, Goparaju C, Heguy A, Dolgalev I, Carbone M, et al.Whole-exome sequencing reveals frequent genetic alterations in BAP1,NF2, CDKN2A, and CUL1 inmalignant pleural mesothelioma. Cancer Res2015;75:264–9.

5. Bonelli MA, Fumarola C, La Monica S, Alfieri R. New therapeuticstrategies for malignant pleural mesothelioma. Biochem Pharmacol2017;123:8–18.

6. Cheng YY,Wright CM,KirschnerMB,WilliamsM, SarunKH, SytnykV, et al.KCa1.1, a calcium-activated potassium channel subunit alpha 1, is targetedby miR-17–5p and modulates cell migration in malignant pleural meso-thelioma. Mol Cancer 2016;15:44.

7. Pignochino Y, Dell'Aglio C, Inghilleri S, Zorzetto M, Basirico M, Capozzi F,et al. The combination of sorafenib and everolimus shows antitumoractivity in preclinical models of malignant pleural mesothelioma. BMCCancer 2015;15:374.

8. Remon J, Reguart N, Corral J, Lianes P. Malignant pleural mesothelioma:newhope in the horizonwith novel therapeutic strategies. Cancer Treat Rev2015;41:27–34.

9. Sekido Y.Molecular pathogenesis of malignant mesothelioma. Carcino-genesis 2013;34:1413–9.

10. Churg A, Roggli V, Galateau-Salle F, Cagle PT, Gibbs AR, Hasleton PS,et al. Tumours of the pleura: mesothelial tumours. In: Travis WD BE,Muller-Hermelink HK, Harris CC, editor. Pathology and genetics oftumours of the lung, pleura, thymus and heart. vol. 10. Lyon (France):International Agency for Research on Cancer; 2004. p. 128–36.

11. Garnett MJ, Edelman EJ, Heidorn SJ, Greenman CD, Dastur A, Lau KW,et al. Systematic identification of genomic markers of drug sensitivity incancer cells. Nature 2012;483:570–5.

12. Daniel VC,Marchionni L,Hierman JS, Rhodes JT,DevereuxWL, RudinCM,et al. A primary xenograft model of small-cell lung cancer reveals irrevers-ible changes in gene expression imposed by culture in vitro. Cancer Res2009;69:3364–73.

13. Gillet JP, Calcagno AM, Varma S, Marino M, Green LJ, Vora MI, et al.Redefining the relevance of established cancer cell lines to the study ofmechanisms of clinical anti-cancer drug resistance. Proc Nat Acad Sci U S A2011;108:18708–13.

14. Roschke AV, Tonon G, Gehlhaus KS, McTyre N, Bussey KJ, Lababidi S, et al.Karyotypic complexity of the NCI-60 drug-screening panel. Cancer Res2003;63:8634–47.

15. Tveit KM, Pihl A. Do cell lines in vitro reflect the properties of the tumoursof origin? A study of lines derived from human melanoma xenografts. Br JCancer 1981;44:775–86.

16. Das V, Bruzzese F, Konecny P, Iannelli F, Budillon A, Hajduch M. Patho-physiologically relevant in vitro tumor models for drug screening. DrugDiscov Today 2015;20:848–55.

17. Eglen RM,Gilchrist A, Reisine T. The use of immortalized cell lines inGPCRscreening: the good, bad and ugly. Comb Chem High Throughput Screen2008;11:560–5.

18. Paik PK, Krug LM. Histone deacetylase inhibitors in malignant pleuralmesothelioma: preclinical rationale and clinical trials. J Thorac Oncol2010;5:275–9.

19. Garassino MC, Marsoni S. A lesson from vorinostat in pleural mesothe-lioma. Lancet Oncol 2015;16:359–60.

20. Krug LM, Kindler HL, Calvert H, Manegold C, Tsao AS, Fennell D, et al.Vorinostat in patients with advanced malignant pleural mesotheliomawho have progressed on previous chemotherapy (VANTAGE-014): a phase3, double-blind, randomised, placebo-controlled trial. Lancet Oncol2015;16:447–56.

21. Szulkin A, Otvos R, Hillerdal CO, Celep A, Yousef-Fadhel E, Skribek H,et al. Characterization and drug sensitivity profiling of primary






malignant mesothelioma cells from pleural effusions. BMC Cancer2014;14:709.

22. Chernova T, Sun XM, Powley IR, Galavotti S, Grosso S, Murphy FA, et al.Molecular profiling reveals primary mesothelioma cell lines recapitulatehuman disease. Cell Death Different 2016;23:1152–64.

23. Schouten PC, van Dyk E, Braaf LM, Mulder L, Lips EH, de Ronde JJ, et al.Platform comparisons for identification of breast cancers with a BRCA-likecopy number profile. Breast Cancer Res Treat 2013;139:317–27.

24. RobinsonMD, Oshlack A. A scaling normalizationmethod for differentialexpression analysis of RNA-seq data. Genome Biol 2010;11:R25.

25. Law CW, Chen Y, Shi W, Smyth GK. voom: Precision weights unlocklinear model analysis tools for RNA-seq read counts. Genome Biol2014;15:R29.

26. Musti M, Kettunen E, Dragonieri S, Lindholm P, Cavone D, Serio G, et al.Cytogenetic and molecular genetic changes in malignant mesothelioma.Cancer Genet Cytogenet 2006;170:9–15.

27. Illei PB, Rusch VW, Zakowski MF, Ladanyi M. Homozygous deletion ofCDKN2A and codeletion of the methylthioadenosine phosphorylasegene in the majority of pleural mesotheliomas. Clin Cancer Res2003;9:2108–13.

28. Lopez-Rios F, Chuai S, Flores R, Shimizu S, Ohno T, Wakahara K, et al.Global gene expressionprofiling of pleuralmesotheliomas: overexpressionof aurora kinases and P16/CDKN2A deletion as prognostic factors andcritical evaluation of microarray-based prognostic prediction. Cancer Res2006;66:2970–9.

29. Ceresoli GL, Zucali PA, Favaretto AG, Grossi F, Bidoli P, Del Conte G, et al.Phase II study of pemetrexed plus carboplatin in malignant pleuralmesothelioma. J Clin Oncol 2006;24:1443–8.

30. Ceresoli GL, Zucali PA. Vinca alkaloids in the therapeutic management ofmalignant pleural mesothelioma. Cancer Treat Rev 2015;41:853–8.

31. Kindler HL, van Meerbeeck JP. The role of gemcitabine in the treatment ofmalignant mesothelioma. Semin Oncol 2002;29:70–6.

32. Schutte W, Blankenburg T, Lauerwald K, Schreiber J, Bork I, Wollscg-kaeger B, et al. A multicenter phase II study of gemcitabine andoxaliplatin for malignant pleural mesothelioma. Clin Lung Cancer2003;4:294–7.

33. Arrieta O, Medina LA, Estrada-Lobato E, Hernandez-Pedro N, Villanueva-Rodriguez G, Martinez-Barrera L, et al. First-line chemotherapy withliposomal doxorubicin plus cisplatin for patients with advanced malig-nant pleural mesothelioma: phase II trial. Br J Cancer 2012;106:1027–32.

34. Schelch K, Hoda MA, Klikovits T, Munzker J, Ghanim B, Wagner C, et al.Fibroblast growth factor receptor inhibition is active againstmesotheliomaand synergizes with radio- and chemotherapy. Am J Resp Crit Care Med2014;190:763–72.

35. Bueno R, Stawiski EW, Goldstein LD,Durinck S, De Rienzo A,Modrusan Z,et al. Comprehensive genomic analysis ofmalignant pleuralmesotheliomaidentifies recurrent mutations, gene fusions and splicing alterations. NatGenet 2016;48:407–16.

36. Cortazar P, Gazdar AF, Woods E, Russell E, Steinberg SM, Williams J, et al.Survival of patients with limited-stage small cell lung cancer treated withindividualized chemotherapy selected by in vitro drug sensitivity testing.Clin Cancer Res 1997;3:741–7.

37. Wilbur DW, Camacho ES, Hilliard DA, Dill PL, Weisenthal LM. Chemo-therapy of non-small cell lung carcinoma guided by an in vitro drug

resistance assay measuring total tumour cell kill. Br J Cancer 1992;65:27–32.

38. Miyazaki R, Anayama T, Hirohashi K, Okada H, Kume M, Orihashi K. Invitro drug sensitivity tests to predict molecular target drug responses insurgically resected lung cancer. PloS One 2016;11:e0152665.

39. Holloway RW, Mehta RS, Finkler NJ, Li KT, McLaren CE, Parker RJ, et al.Association between in vitro platinum resistance in the EDR assay andclinical outcomes for ovarian cancer patients. Gynecol Oncol 2002;87:8–16.

40. Matsuo K, Bond VK, Im DD, Rosenshein NB. Prediction of chemotherapyresponse with platinum and taxane in the advanced stage of ovarian anduterine carcinosarcoma: a clinical implication of in vitro drug resistanceassay. Am J Clin Oncol 2010;33:358–63.

41. Takamura Y, Kobayashi H, Taguchi T, Motomura K, Inaji H, Noguchi S.Prediction of chemotherapeutic response by collagen gel droplet embed-ded culture-drug sensitivity test in human breast cancers. Int J Cancer2002;98:450–5.

42. Mujoomdar AA, Tilleman TR, Richards WG, Bueno R, Sugarbaker DJ.Prevalence of in vitro chemotherapeutic drug resistance in primary malig-nant pleural mesothelioma: result in a cohort of 203 resection specimens.J Thoracic Cardiovasc Surg 2010;140:352–5.

43. Glinsky GV, Glinskii AB, Stephenson AJ, Hoffman RM, Gerald WL. Geneexpression profiling predicts clinical outcome of prostate cancer. J ClinInvest 2004;113:913–23.

44. van 't Veer LJ, Dai H, van de Vijver MJ, He YD, Hart AA, Mao M, et al. Geneexpression profiling predicts clinical outcome of breast cancer. Nature2002;415:530–6.

45. Cardoso F, van't Veer LJ, Bogaerts J, Slaets L, Viale G, Delaloge S, et al. 70-gene signature as an aid to treatment decisions in early-stage breast cancer.N Eng J Med 2016;375:717–29.

46. Zhao R, Zhang S, Hanscom M, Chattopadhyay S, Goldman ID. Loss ofreduced folate carrier function and folate depletion result in enhancedpemetrexed inhibition of purine synthesis. Clin Cancer Res 2005;11:1294–301.

47. Suchy SL, Landreneau RJ, Schuchert MJ, Wang D, Ervin PR Jr., Brower SL.Adaptation of a chemosensitivity assay to accurately assess pemetrexed inex vivo cultures of lung cancer. Cancer Biol Ther 2013;14:39–44.

48. Calvert AH.Biochemical pharmacology of pemetrexed. Oncology (Will-iston Park) 2004;18 (13 Suppl 8):13–7.

49. Assaraf YG.Molecular basis of antifolate resistance. Cancer Metast Rev2007;26:153–81.

50. Zhao R, Gao F, Goldman ID. Marked suppression of the activity of some,but not all, antifolate compounds by augmentation of folate cofactor poolswithin tumor cells. Biochem Pharmacol 2001;61:857–65.

51. Backus HH, Pinedo HM, Wouters D, Padron JM, Molders N, van Der WiltCL, et al. Folate depletion increases sensitivity of solid tumor cell lines to 5-fluorouracil and antifolates. Int J Cancer 2000;87:771–8.

52. Lorenzi M, Porcelli B, Vannoni D, Leoncini R, Pizzichini M, Di Stefano A,et al. Plasma oxypurines in gastric and colorectal cancer. Biomed Pharmac-other 1990;44:403–7.

53. Touat M, Ileana E, Postel-Vinay S, Andre F, Soria JC. Targeting FGFRsignaling in cancer. Clin Cancer Res 2015;21:2684–94.

54. Marek LA, Hinz TK, von Massenhausen A, Olszewski KA, Kleczko EK,Boehm D, et al. Nonamplified FGFR1 is a growth driver in malignantpleural mesothelioma. Mol Cancer Res 2014;12:1460–9.


Schunselaar et al.




2018;24:1761-1770. Published OnlineFirst October 24, 2017.Clin Cancer Res Laurel M. Schunselaar, Josine M.M.F. Quispel-Janssen, Yongsoo Kim, et al. ResponsesSubtypes with Different Expression Profiles and Clinical Chemical Profiling of Primary Mesothelioma Cultures Defines

Updated version

10.1158/1078-0432.CCR-17-1345doi:

Access the most recent version of this article at:

Material

Supplementary

http://clincancerres.aacrjournals.org/content/suppl/2017/10/24/1078-0432.CCR-17-1345.DC1

Access the most recent supplemental material at:

Cited articles

http://clincancerres.aacrjournals.org/content/24/7/1761.full#ref-list-1

This article cites 53 articles, 12 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/24/7/1761.full#related-urls

This article has been cited by 3 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://clincancerres.aacrjournals.org/content/24/7/1761To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/lookup/doi/10.1158/1078-0432.CCR-17-1345

http://clincancerres.aacrjournals.org/content/suppl/2017/10/24/1078-0432.CCR-17-1345.DC1

http://clincancerres.aacrjournals.org/content/24/7/1761.full#ref-list-1

http://clincancerres.aacrjournals.org/content/24/7/1761.full#related-urls

http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://clincancerres.aacrjournals.org/content/24/7/1761


chemical proﬁling of primary mesothelioma cultures …malignant pleural mesothelioma is...

Documents