Integrated Genomic Classification of Melanocytic Tumors of the Central Nervous System Using Mutation Analysis, Copy Number Alterations, and DNA Methylation Profiling

Published OnlineFirst June 11, 2018; DOI: 10.1158/1078-0432.CCR-18-0763

Precision Medicine and Imaging Clinical Cancer Research Integrated Genomic Classiﬁcation of Melanocytic Tumors of the Central Nervous System Using Mutation Analysis, Copy Number Alterations, and DNA Methylation Proﬁling Klaus G. Griewank1,2, Christian Koelsche3, Johannes A.P. van de Nes4, Daniel Schrimpf3, Marco Gessi5,6, Inga Moller€ 1, Antje Sucker1, Richard A. Scolyer7,8,9, Michael E. Buckland8,10, Rajmohan Murali11, Torsten Pietsch5, Andreas von Deimling3, and Dirk Schadendorf1

Abstract

Purpose: In the central nervous system, distinguishing Results: Combining mutation, copy-number, and DNA- primary leptomeningeal melanocytic tumors from melanoma methylation profiles clearly distinguished cutaneous melanoma metastases and predicting their biological behavior solely metastases from other melanocytic tumors. Primary leptome- using histopathologic criteria may be challenging. We aimed ningeal melanocytic tumors, uveal melanomas, and blue to assess the diagnostic and prognostic value of integrated nevus–like melanoma showed common DNA-methylation, molecular analysis. copy-number alteration, and gene mutation signatures. Notably, Experimental Design: Targeted next-generation sequenc- tumors demonstrating chromosome 3 monosomy and BAP1 ing, array-based genome-wide methylation analysis, and BAP1 alterations formed a homogeneous subset within this group. IHC were performed on the largest cohort of central nervous Conclusions: Integrated molecular profiling aids in distin- system melanocytic tumors analyzed to date, including 47 guishing primary from metastatic melanocytic tumors of the primary tumors of the central nervous system, 16 uveal mel- central nervous system. Primary leptomeningeal melanocytic anomas, 13 cutaneous melanoma metastases, and 2 blue tumors, uveal melanoma, and blue nevus–like melanoma nevus–like melanomas. Gene mutation, DNA-methylation, share molecular similarity with chromosome 3 and BAP1 and copy-number profiles were correlated with clinicopatho- alterations, markers of poor prognosis. Clin Cancer Res; 24(18); logic features. 4494–504. 2018 AACR.

Introduction cytes. They are termed primary leptomeningeal melanocytic tumors, (1) and are classified according to histopathologic criteria Melanocytic tumors involving the central nervous system can (2), which include mitotic activity, parenchymal infiltration, and be diagnostically challenging. Melanocytic tumors can arise in the Ki67/MIB1 staining. Applying these criteria, primary leptomenin- central nervous system, however, more common are central geal melanocytic tumors are classified as melanocytomas, inter- nervous system melanoma metastases. Correct classification of mediate-grade melanocytomas, and primary leptomeningeal melanocytic tumors of the central nervous system is important for melanomas, referring to benign, atypical/borderline, and malig- prediction of their likely clinical behavior and selecting appro- nant tumors, respectively. However, histopathologic classification priate treatment. of primary leptomeningeal melanocytic tumors is not always The majority of primary melanocytic tumors of the central straightforward. For example, in some primary leptomeningeal nervous system presumably arise from leptomeningeal melano- melanocytic tumors, histopathologic criteria overlap and do not

1Department of Dermatology, University Hospital Essen, West German Cancer Note: Supplementary data for this article are available at Clinical Cancer Center, University Duisburg-Essen and the German Cancer Consortium (DKTK), Research Online (http://clincancerres.aacrjournals.org/). Germany. 2Dermatopathologie bei Mainz, Nieder-Olm, Germany. 3Department of K.G. Griewank, C. Koelsche, A. von Deimling, and D. Schadendorf contributed Neuropathology, Ruprecht-Karls-University Heidelberg, and Clinical Cooperation equally to this article. Unit Neuropathology, and DKTK, DKFZ, Heidelberg, Germany. 4Institute of Pathology, Ruhr University Bochum, Bochum, Germany. 5Institute of Neuropa- Corresponding Authors: Klaus G. Griewank, University Hospital Essen, thology, University of Bonn Medical Center, Bonn, Germany. 6Division of Histo- Hufelandstrasse 55, Essen 45147, Germany. Phone: 49 201 72385947; pathology, Fondazione Policlinico Universitario "A.Gemelli", Universita Cattolica Fax: 49 201 7235416; Orchid: 0000-0003-3899-9449; E-mail: del Sacro Cuore, Roma, Italy. 7Tissue Pathology and Diagnostic Oncology, Royal [email protected]; and Christian Koelsche, Ruprecht-Karls- Prince Alfred Hospital, Camperdown, NSW, Australia. 8The University of Sydney, University Heidelberg, Im Neuenheimer Feld 224, 69120 Heidelberg, Camperdown, NSW, Australia. 9Melanoma Institute Australia, The University of Germany. E-mail: [email protected] Sydney, North Sydney, NSW, Australia. 10Department of Neuropathology, Royal doi: 10.1158/1078-0432.CCR-18-0763 Prince Alfred Hospital, Camperdown, NSW, Australia. 11Department of Pathology, Memorial Sloan Kettering Cancer Center, New York, New York. 2018 American Association for Cancer Research.

4494 Clin Cancer Res; 24(18) September 15, 2018

Genetic Analysis of CNS Melanocytic Tumors

Each tumor was obtained from a unique patient. In all primary Translational Relevance leptomeningeal melanocytic tumors, a primary melanoma of Melanocytic tumors involving the central nervous system other site, including a uveal melanoma, had been excluded can be diagnostically challenging; in particular, distinguishing (clinical history, imaging, and frequently fundoscopy). The sam- primary leptomeningeal melanocytic tumors and central ner- ples were retrieved from the Institute of Neuropathology and the vous system metastases from cutaneous melanomas may be Department of Dermatology, Essen, Germany, the Department of difficult. We demonstrate that these entities can be distin- Neuropathology, Heidelberg, Germany, the Institute of Neuro- guished based on gene mutation, copy number, and DNA pathology Bonn, Germany, as well as the Melanoma Institute methylation profiles. Additionally, primary leptomeningeal Australia, Sydney, Australia. Some of these tumors have been melanocytic tumors demonstrated copy number, methyla- previously reported (8, 9, 11, 12, 26), but these reports did not tion, and mutation profiles similar to those of uveal melano- include the detailed genetic analyses reported herein. Tumor mas and blue nevus–like melanoma. Monosomy of chromo- slides were reviewed by at least two experienced histopatho- some 3 and BAP1 alterations, markers of poor prognosis in logists (C. Koelsche, J.A.P. van de Nes, M. Gessi, T. Pietsch, uveal melanoma, were also identified in primary leptomenin- K.G. Griewank, R.A. Scolyer, A. von Deimling, or M.E. Buckland). geal melanocytic tumors. In summary, we find that applying The study was performed in accordance with the guidelines genetic analysis in addition to histopathologic examination set forth by the ethics committee of the University of Heidelberg has the potential to improve diagnostic and prognostic eval- and Duisburg-Essen under the IRB protocol numbers 180073 and uation of melanocytic tumors of the central nervous system. 16-6951-BO, respectively. We believe that genetic profiling should become a routine component of the analysis of melanocytic tumors of the Histopathology and IHC central nervous system. Histopathologic examination was performed on routinely stained hematoxylin and eosin slides. Primary leptomeningeal tumors were diagnosed based on criteria described by Brat and colleagues (2). Well-differentiated primary leptomeningeal tumors with no or very low mitotic activity (0–1mitosesper permit ready classification. Others show aggressive biological 10 high-power fields), devoid of central nervous system paren- behavior despite lacking obvious high-grade histopathologic chymal infiltration, and Ki67/MIB1 index 2% were diag- features. nosed as melanocytomas (n ¼ 19). Tumors characterized by Primary leptomeningeal melanocytic tumors and uveal mel- increased mitotic activity (Ki67/MIB1 index 1%–4%) and/or anomas share a similar gene mutation profile (3–9). They microscopic central nervous system parenchymal invasion, frequently harbor mutually exclusive activating hot-spot muta- but not sufficiently anaplastic to warrant the designation of tions in either GNAQ, GNA11, PLCB4,orCYSLTR2 (3–9). melanoma were diagnosed as intermediate-grade melanocy- Additional mutations can also be found in EIF1AX, SF3B1, tomas (n ¼ 22). Tumors demonstrating higher mitotic activity and BAP1, which are also generally detected in a mutually and anaplasia were diagnosed as primary leptomeningeal exclusive pattern (10–12). melanomas (n ¼ 4). Two tumors diagnosed as retrobulbar Cutaneous melanomas consistently demonstrate a very high melanomas were also included in the primary leptomeningeal mutational load and a characteristic UV-exposure mutation sig- melanoma group. nature (13–16). A genetic classification into four groups based on In most cases, IHC was performed on a Ventana Benchmark mutations activating the MAP kinase pathway has been proposed: XT Autostainer using the following markers: S-100 (1:5,000, BRAF-mutated, RAS-mutated, NF1-mutated, or triple wild-type Dako; Z0311); Melan-A (1:100, Dako; M7196); HMB-45 (15, 16). Furthermore, TERT promoter mutations are highly (1:200, Dako; M0634); BAP1 [as previously described (26), recurrent in cutaneous melanomas (17, 18). applying a BAP1 rabbit polyclonal antibody raised against a A rare subset of cutaneous melanomas arising in the dermis is synthetic peptide corresponding to amino acids 430-729 of the termed blue nevus–like melanoma. These tumors lack BRAF, RAS, BAP1 molecule (clone C-4; Santa Cruz Biotechnology Inc.)]; NF1,orTERT promoter mutations, instead demonstrating genetic and the proliferation marker Ki67/MIB1 (1:200, Zytomed; alterations similar to those in primary leptomeningeal tumors MSK0810). BAP1 expression was scored as positive or negative and uveal melanomas (19–25). according to the presence or absence of nuclear staining. The The aim of our study was to explore molecular signatures of percentage of tumor cells staining with Ki-67/MIB1 was desig- primary leptomeningeal melanocytic tumors that would allow nated as the Ki-67/MIB1 proliferation index. classification into prognostically distinct subgroups. DNA isolation Ten-mm-thick sections were cut from formalin-fixed, paraffin- Materials and Methods embedded tumor tissues. The sections were deparaffinized and Sample selection manually microdissected according to standard procedures. A cohort of 47 primary leptomeningeal tumors comprising 19 Genomic DNA was isolated using the QIAamp DNA Mini Kit melanocytomas, 22 intermediate-grade melanocytomas, and 6 (Qiagen) according to the manufacturer's instructions. primary leptomeningeal melanomas (including two retrobulbar melanomas) was analyzed. The control group consisted of 17 Targeted sequencing central nervous system metastases (from 2 uveal melanomas, 2 Two previously published (11, 19) custom amplicon-based melanomas of unknown primary, and 13 cutaneous melanomas); sequencing panels were used, one covering 10 genes recurrently 14 primary uveal melanomas; and 2 blue nevus–like melanomas. mutated in uveal melanoma (Supplementary Table S1), and the

www.aacrjournals.org Clin Cancer Res; 24(18) September 15, 2018 4495

Griewank et al.

other covering 29 genes recurrently mutated in cutaneous mela- Gene mutations detected by targeted next-generation noma (Supplementary Table S2). Both sequencing panels were sequencing applied using the GeneRead Library Prep Kit from QIAGEN All cases were screened by a targeted amplicon-based next- according to the manufacturer's instructions. Sequencing was generation sequencing (NGS) panel covering genes known to be performed on an Illumina MiSeq next-generation sequencer. recurrently mutated in uveal melanoma, primary leptomeningeal Adapter ligation and barcoding was performed applying the tumors, and blue nevus–like melanoma (Supplementary Table NEBNext Ultra DNA Library Prep Mastermix Set and NEBNext S2). In 47 primary leptomeningeal tumors, 94% (44/47) tumors Multiplex Oligos for Illumina from New England Biolabs. harbored activating hotspot mutations in GNAQ (n ¼ 26, 55%), Sequencing analysis was performed applying CLC Cancer GNA11 (n ¼ 13, 28%), CYSLTR2 (n ¼ 3, 6%), and PLCB4 (n ¼ 2, Research Workbench from QIAGEN. The analysis workflow 4%). All of these mutations were mutually exclusive (Fig. 1; Table described in brief included adapter trimming and read pair 1; Supplementary Table S3). Co-occurring mutations in EIF1AX, merging before mapping to the human reference genome (hg19). SF3B1, and BAP1 were detected in 38% (n ¼ 18), 21% (n ¼ 10), Detection of insertions and deletions as well as single nucleotide and 6% (n ¼ 3) of tumors, respectively. Concurrent EIF1AX and variant followed. Additional information regarding potential SF3B1 mutations were identified in three primary leptomeningeal mutation type, known single nucleotide polymorphisms and tumors. conservation scores was obtained by cross-referencing various In the 16 uveal melanoma samples, 75% (n ¼ 12) had GNAQ databases (COSMIC, ClinVar, dbSNP, 1000 Genomes Project, and 25% (n ¼ 4) had GNA11 mutations. Additional mutations in HAPMAP, and PhastCons-Conservation_scores_hg19). BAP1, SF3B1, and EIF1AX were identified in 31% (n ¼ 5), 25% The sequencing results were of high quality, with an average (n ¼ 4), and 13% (n ¼ 2) of samples, respectively. coverage of 13,503-fold for all samples analyzed and a minimum In the blue nevus–like melanomas available, a GNAQ,a of 30-fold coverage achieved in 98% of the targeted sequence. GNA11, and two SF3B1 mutations were identified, as previously Mutations were reported if the overall coverage of the mutation reported (20). site was 30 reads, 10 reads reported the mutated variant and Metastases from cutaneous melanomas and melanomas of the frequency of mutated reads was 7%. All genetic variants unknown primary (n ¼ 15) harbored activating mutations in identified are presented in Fig. 1. genes of the MAPK pathway in 93% (n ¼ 14) of samples, including BRAF V600 (47%, n ¼ 7), NRAS (33%, n ¼ 5), HRAS, and KIT DNA-methylation profiling and copy number analysis mutations (each 7%, n ¼ 1). In addition, 80% (n ¼ 12) of tumors Copy number and methylation analysis required 500 ng of harbored TERT promoter hot-spot mutations. isolated DNA and was performed on 91% (n ¼ 43 of 47) primary leptomeningeal melanocytic tumors and 100% of blue nevus– BAP1 IHC like melanoma, cutaneous melanoma, and uveal melanoma Loss of nuclear BAP1 expression is an adverse prognostic samples (n ¼ 2, 15, and 16, respectively). The primary leptome- marker in uveal melanoma. Based on the known mutational ningeal melanocytic tumor samples in which these analyses were overlap between primary leptomeningeal tumors and uveal mel- not performed (due to small sample sizes and insufficient DNA) anomas, we examined BAP1 expression by IHC in cases with included two melanocytomas, one intermediate-grade melano- sufficient material available. Two of 42 primary leptomeningeal cytoma and one retrobulbar melanoma (Fig. 1). The Illumina tumors (one melanoma and one intermediate-grade melanocy- Infinium 450 k or EPIC array was used to obtain genome-wide toma) showed loss of nuclear BAP1 expression. In comparison, methylation data (Illumina), according to the manufacturer's 53% (8/15) uveal melanomas demonstrated nuclear BAP1 loss, instructions. DNA-methylation data were normalized by per- whereas BAP1 expression was retained in all eight metastases from forming background correction and dye bias correction. Filtering cutaneous melanomas that were analyzed (Fig. 1). of probes, that is, removal of probes containing single nucleotide polymorphism and not uniquely matching, was performed as Copy number analysis described previously (8). For unsupervised hierarchical clustering, Copy number analysis was performed on 76 samples from 80 the 10,000 most variably methylated probes (SD > 0.25) across tumors: 43 primary leptomeningeal tumors, 14 primary uveal the dataset were selected. Euclidean distance and average linkage melanomas, 17 melanoma metastases (from two melanomas of was applied for ordering of the probes (y-axis) and 1–Pearson unknown primary, two uveal melanomas and 13 cutaneous correlation as distance measure and average linkage was used for melanomas), and two blue nevus–like melanomas (Fig. 1). clustering samples (x-axis).The copy number profile was gener- Primary leptomeningeal tumors, uveal melanomas, and blue ated from the array data using the "conumee" R package in nevus–like melanomas showed similar copy number aberrations. Bioconductor (http://www.bioconductor.org/packages/release/ These frequently involved gains of chromosome arm 8q and 6p, bioc/html/conumee.html). monosomy of chromosome 3, and loss of chromosome arm 1p and 6q (Figs. 1–3). Copy number profiles of cutaneous melanomas differed from those of primary leptomeningeal tumors, Results uveal melanomas, and blue nevus–like melanomas (Figs. 1–3). Tumors and patients Cutaneous melanomas demonstrated similar 6p gains and 6q Overall, 39 (45%) tumors occurred in females and 46 (52%) losses, however less frequent chromosome 3 and chromosome in males; in two cases, the sex was unknown. The median age arm 1p losses. In addition, cutaneous melanomas frequently was 56 years and ranged from 15 to 87 years. The primary carried copy number alterations encompassing a loss of CDKN2A leptomeningeal tumor cohort consisted of 47 cases (19 melano- on chromosome arm 9p, a loss of PTEN on chromosome 10 and cytomas, 22 intermediate-grade melanocytomas, and 6 melano- frequent gains of chromosome 7 involving the BRAF locus, similar mas). Clinical details are listed in Table 1. to previous reports on cutaneous melanomas (27).

4496 Clin Cancer Res; 24(18) September 15, 2018 Clinical Cancer Research

Genetic Analysis of CNS Melanocytic Tumors

Primary leptomeningeal melanocytic tumors Nr. 1 2 3 4 5 6 7 8 9 1011121314151617181920212223242526272829303132333435363738394041424344454647 Diagnosis

GNAQ GNA11 PLCB4 CYSLTR2 EIF1AX SF3B1 BAP1 BRAF NRAS HRAS KIT TERT pro Gene mutations Other IHC BAP1 IHC 1p- 3p- 3q- 6p+ 6q- 8q+ BNLM Uveal melanomas Cutaneous melanomas

Nr. 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 Diagnosis

GNAQ GNA11 PLCB4 CYSLTR2 EIF1AX SF3B1 BAP1 BRAF NRAS HRAS KIT

Gene mutationsTERT pro CNV Other IHC BAP1 IHC 1p- 3p- 3q- 6p+

CNV 6q- 8q+

Diagnosis: Melanocytoma IG Melanocytoma CNS Melanoma Retrobulbar melanoma UM Primary UM Metastasis MUP Cutaneous Gene mutations: Activating Loss of function Activating TERT promoter Missense (unknown significance) R625 EIF1AX Exon 1 or 2 BAP IHC: Nuclear expression Retained Lost CNV: Loss Partial loss Gain Partial gain Polysomy/amplification

Figure 1. Distribution of mutations and select copy number variations. Demonstrated are the mutations and a selection of CNV identiﬁed in the analyzed tumor cohort. Results of BAP1 IHC (IHC) are also displayed. The annotation is according to the legend in the ﬁgure. UM, uveal melanoma; BNLM, blue nevus–like melanoma.

www.aacrjournals.org Clin Cancer Res; 24(18) September 15, 2018 4497

Griewank et al.

Table 1. Distribution of tumors with clinicopathologic and mutation data

Nr. Diag. GNA11 Type GNAQ SF3B1 BRAF KIT Age Sex CYSLTR2 BAP1 NRAS TERT pro. Other PLCB4 HRAS BAP1 IHC Location EIF1AX 3 M 53 m Spinal P Q209L G15D 4 M 81 m Spinal P Q209L G6D 7 M 48 m Posterior fossa P Q209L G9D 9 M 54 m Infratentorial basal R Q209P P2L 15 M 73 m Spinal C2-3 P Q209L 16 M 48 f Spinal T P Q209L 17 M 65 m Spinal P Q209L 18 M 55 m Spinal P Q209L 21 M 86 f Spinal P Q209L 22 M 23 f Spinal R Q209L 26 M 41 f Frontal P Q209P 30 M 43 f Spinal P Q209L K3T 32 M 67 f Sphenoid bone P Q209L R625H 40 M 41 m Craniocervical junc. NK D630F G8V 41 M 73 f Temporal right P D630Y E31del 42 M 85 m Intraventricular P L129Q G9R 43 M 77 f Spinal T7-8 P L129Q R13C 46 M 54 m NA NK 47 M 46 f Cavernous sinus P 1 IGM 55 m Cerebellum P Q209P G9D R625H 5 IGM 65 m Spinal C3 P Q209L G8V 6 IGM 69 f Spinal T8 P Q209L G9D 10 IGM 46 m Spinal P Q209L R625H 12 IGM 23 m Skull base R Q209P R625H 13 IGM 79 m Spinal T4-5 R Q209L 14 IGM 64 m Spinal C5 P Q209L 19 IGM 44 m Spinal C1 P Q209L 20 IGM NK NK NK P Q209L 23 IGM 34 m Spinal T10-12 P Q209P 24 IGM 20 m Frontal P Q209P 25 IGM 35 f Spinal C5-6 P Q209P 27 IGM 43 m Spinal P Q209L G6V R625C 28 IGM 15 m Supra- and infratent. P Q209L G8E 29 IGM 33 f NK P Q209L G9D 31 IGM 48 f Spinal T8-9 R Q209L R13P 33 IGM 58 f Spinal P Q209L R625H 35 IGM 47 m Spinal T10 P Q209L R625C 37 IGM 78 f Tentorium P Q209L R60* 38 IGM 67 m Plexus brachialis P Q209L 44 IGM 67 m Spinal T9-10 P L129Q 45 IGM 46 f Temporal R P2S 2 Mel 66 m Brain P R183Q G15D R625H 11 Mel 54 m Brain P Q209L R625H 34 Mel 54 m Parasagittal P Q209L R625S 36 Mel 66 f Supratentorial P Q209L Q267fs 8 RMel 54 m Retrobulbar P Q209P N4Y 39 RMel 74 m Retrobulbar P R183C 48 BNLM 70 m Left back P Q209P R625C 49 BNLM 39 m Scrotum left P Q209L R625H 50 UM 45 f Choroid P Q209P G15D 51 UM 73 f Choroid P Q209V R625H 52 UM 68 m Choroid cil. inv. P Q209P R625S 53 UM 52 m Choroid P Q209P L100fs 54 UM 81 f Choroid P Q209P F22fs 55 UM 72 f Choroid P Q209P S395* 56 UM 70 f Ciliary-choroid P Q209L 57 UM 85 m Choroid cil. inv. P Q209P 58 UM 55 f Ciliary-choroid P Q209P 59 UM 45 f Choroid P Q209P 60 UM 69 m Choroid P Q209L G6D 61 UM 77 m Choroid P Q209L R625H 62 UM 72 m Ciliary-choroid P Q209L P88fs 63 UM 66 f Choroid P Q209L Y223fs 64 UM-Met 49 f N. opticus M Q209P R625H 65 UM-Met 53 f Temporal M Q209L 66 MUP-Met 44 m Intraventricular M V600E 228 67 MUP-Met 39 m Leptomeningeal M V600E 68 CM-Met 74 f Occipital M V600K 228 69 CM-Met 68 f Parietal M V600E 228 70 CM-Met 60 f Frontal M V600E 250 71 CM-Met 74 m Temporal M V600K 250 72 CM-Met 74 f Occipital M V600K 228 73 CM-Met 68 f Occipital M Q61K 250 74 CM-Met NK NK NK M Q61K 75 CM-Met 60 f Frontal M Q61L 228 76 CM-Met 46 m Parietal M Q61L 228 77 CM-Met 51 m Cerebellum M G6S Q61R 250 78 CM-Met 60 m Frontal M G13S 250 79 CM-Met 87 m Parietal M A636V G466E V559A 250 80 CM-Met 57 f Temporal M Nr., sample number; Diag., diagnosis; M, melanocytoma; IGM, intermediate-grade melanocytoma; UM, uveal melanoma; BNLM, blue nevus–like melanoma; CM, cutaneous melanoma; MUP, melanoma of unknown primary; met, metastasis; m, male; f, female; P, primary; R, recurrence; M, metastasis; pro., promoter; n., nervus; inv, involvement; cil., ciliary; tent., tentorial; BAP1 IHC, BAP1 immunohistochemistry (green, retained nuclear expression; red, loss of expression; gray, not analyzed).

4498 Clin Cancer Res; 24(18) September 15, 2018 Clinical Cancer Research

Genetic Analysis of CNS Melanocytic Tumors

Figure 2. Comparison of methylation proﬁles. Shown are the results of unsupervised clustering, demonstrating the methylation areas with the largest difference in methylation status. Chromosome status of Chr. 1p, 3p, 3q, 6p, 6q, and 8q; mutation status of EIF1AX, SF3B1, and BAP1; and IHC BAP1 status of the presented tumors are also demonstrated above for comparison.

DNA-methylation proﬁling meningeal tumors and uveal melanomas and blue nevus–like Unsupervised hierarchical clustering using the 10,000 most melanomas, which are indistinguishable at the DNA methylation variably methylated CpG probes clearly separates the study level. However, tumors with chromosome 3 monosomy, loss of cohort into two main clusters. One cluster includes all cutaneous BAP1 protein expression, and/or BAP1 mutations formed a dis- melanomas. The second cluster is composed of primary lepto- tinct subgroup within this methylation cluster. In contrast, tumors

Figure 3. Comparison of CNVs. Summaries of the detected CNV in each tumor group are shown. The blots demonstrate the amount of tumors demonstrating gains or losses of each region. Gains are shown above the x-axis, in the area on the y-axis above 0; losses are below the x-axis, in the area on the y-axis below 0.

www.aacrjournals.org Clin Cancer Res; 24(18) September 15, 2018 4499

Griewank et al.

not demonstrating these alterations and frequently harboring follow-up data allowing a comparable validation of the impact EIF1AX and/or SF3B1 mutations segregated together. of similar gene alterations in these tumors appears unlikely, given the rarity of these tumors. Incomplete follow-up is also a limitation of our primary leptomeningeal tumor cohort. Discussion Despite these limitations, the practically identical genetic pro- To the best of our knowledge, this study represents the most files observed in these tumor entities (uveal melanomas, blue comprehensive genetic analysis of primary leptomeningeal nevus–like melanomas, and primary leptomeningeal tumors) tumors to date. It also represents the first detailed direct genetic raises the possibility that the prognostic genetic markers in comparison of primary leptomeningeal tumors with the geneti- uveal melanoma may also be of prognostic relevance in pri- cally similar uveal melanomas and blue nevus–like melanomas. mary leptomeningeal tumors. The presented mutation and methylation profiles allowed a clear EIF1AX, SF3B1, and BAP1 mutations in uveal melanoma are distinction of primary leptomeningeal tumors from cutaneous associated with favorable, intermediate, and poor prognosis, melanoma metastases. Our analysis identified tumors clustering respectively (28, 31–33). Although our study identifies EIF1AX, according to chromosome 3 and BAP1 status which may be of SF3B1, and BAP1 mutations in primary leptomeningeal tumors, prognostic relevance. the gene mutation profile does not appear to fit well with the In our cohort of 47 primary leptomeningeal tumors, 94% (n ¼ histologic diagnosis in many cases (Figs. 1, 4, 5; Supplementary 44) of tumors harbored mutually exclusive activating mutations Fig. S2). Tumors diagnosed as melanocytoma or intermediate- in GNAQ, GNA11, PLCB4,orCYSLTR2 as previously reported (9, grade melanocytoma assumed to have a favorable prognosis were 28). These gene mutations were not identified in the analyzed found to harbor SF3B1 or BAP1 mutations (Fig. 1). Genetically metastases from cutaneous melanoma, consistent with the liter- comparable tumors were diagnosed as uveal melanoma or blue ature (9, 20). In addition to an activating mutation in one of the nevus–like melanoma, which is demonstrated in individual cases four above-mentioned genes, primary leptomeningeal tumors in Figs. 4, 5, and Supplementary Fig. S2. Here one can observe that also relatively frequently harbored additional EIF1AX (38%) or primary leptomeningeal tumors, uveal melanomas, and in one SF3B1 (19%) mutations, and only rarely (6%) BAP1 mutations. In case a blue nevus–like melanoma harboring similar or identical 45 of 47 primary leptomeningeal tumors (96%), at least one gene mutations [GNA11 and SF3B1 (Fig. 4), GNAQ and EIF1AX mutation in GNAQ, GNA11, PLCB4, CYSLTR2; EIF1AX, SF3B1,or (Fig. 5), GNA11 and BAP1 (Supplementary Fig. S1)] also have BAP1 was detected, suggesting that screening for presence of comparable copy number profiles. In all the presented cases, the mutations in these genes can be diagnostically useful in distin- primary leptomeningeal tumor was diagnosed either as a mela- guishing these tumors from metastases from cutaneous nocytoma or intermediate-grade melanocytoma, both assumed melanoma. to have a favorable prognosis. In contrast, the genetically almost The mutation profile detected in metastases from cutaneous identical uveal melanoma and blue nevus–like melanoma cases melanoma allows a clear distinction of these tumors from other were diagnosed as malignant melanoma. pigmented entities, in particular primary leptomeningeal tumors Despite these discrepancies, there was a trend toward certain and melanotic schwannomas (8). The most frequent mutations mutations being associated with tumors predicted to show poorer identified in the former were BRAF, NRAS,orTERT promoter prognosis based on their histopathologic features. The percentage mutations. In 14 of 15 (93%) cases, one of these three mutations of tumors harboring SF3B1 or BAP1 mutations was 11% (2/19) in was present and could clearly differentiate a cutaneous melanoma melanocytoma, 32% (7/22) in intermediate-grade melanocy- metastasis from other entities. Analyzing these genes for the toma, and 66% (4/6) in primary leptomeningeal melanoma presence of mutations could suffice for diagnostic purposes in (P ¼ 0.02). most cases. Poor prognosis being associated with BAP1 inactivation in The similar gene mutation, copy number variation (CNV), and uveal melanoma is well documented (26, 28, 32, 34). The limited methylation profiles we identified in primary leptomeningeal published data for blue nevus–like melanoma and primary lep- melanocytic tumors, uveal melanomas and blue nevus–like mel- tomeningeal tumors suggest a similar finding (11, 20, 21, 24, 25). anomas were remarkable. Unsupervised hierarchical clustering of Detecting losses of BAP1 genetically can prove challenging, par- these tumors based on genome-wide methylation data separated ticularly in picking up smaller deletions. IHC is a reliable method them according to chromosome and mutation status, but not of detecting BAP1 inactivation (26), but some mutations can according to diagnosis (uveal melanoma, blue nevus–like mela- functionally inactivate BAP1 without causing protein loss (9, 31). noma, or primary leptomeningeal tumor). Comparing individual Although chromosome 3 monosomy is frequently interpreted as uveal melanomas, blue nevus–like melanoma and primary lep- equivalent to BAP1 protein inactivation, there will be exceptions. tomeningeal tumors harboring the same or very similar gene We believe that screening these tumors for mutations, CNV and mutations showed similar CNV and methylation profiles methylation profiles should be performed when possible. How- (Figs. 4 and 5; Supplementary Fig. S2). As previously postulated ever, such cost-intensive and sophisticated approaches will not (29, 30), our study suggests that these tumors represent a com- always be available. Determining BAP1 status by IHC is relatively mon genetic tumor group. reliable and in our opinion should become a routine procedure in This relation may be of critical importance in terms of the pathologic evaluation of primary leptomeningeal tumors, interpreting the clinical meaning of genetic alterations. As uveal uveal melanomas, and blue nevus–like melanomas. melanoma are the most frequent and best studied tumor type In uveal melanoma, it has become accepted that tumor behav- in this group, the relevance of different mutations in terms of ior and prognosis can be better predicted by including genetic tumor behavior and prognosis is well-documented (28, 31, testing than relying solely on histopathologic analysis. We believe 32). Obtaining similarly sized cohorts of blue nevus–like that a similar approach has value in primary leptomeningeal melanomas or primary leptomeningeal tumors with long-term tumors. In our study, a definitive separation based on genome-

4500 Clin Cancer Res; 24(18) September 15, 2018 Clinical Cancer Research

Genetic Analysis of CNS Melanocytic Tumors

Figure 4. Mutation and copy number proﬁle of GNA11 and SF3B1 mutated tumors. Shown are examples of a primary leptomeningeal melanocytic tumor (PLMT) diagnosed as a melanocytoma and a uveal melanoma and blue nevus–like melanoma. All these samples harbored the same GNA11 c.626A>T Q209L and SF3B1 R625H c.1874G>A mutations. The copy number proﬁles of these three tumors are depicted on the right, demonstrating similar alterations, with common gains of Chr. 6p, 8q, and 11p as well as losses of Chr. 6q and 11q.

wide methylation, CNV, or mutation signatures was not possible, assigned to the intermediate-risk group and 3/47 (6%) cases to however tumors were found to group together based on chro- the potential high-risk group. The majority of tumors 28/47 mosome 3 and BAP1 status (Fig. 2). Primary leptomeningeal (60%) demonstrated a genetic profile suggesting a favorable tumors harboring chromosome 3 loss and BAP1 alterations prognosis. (mutations and/or BAP1 IHC loss) should be considered high- A comparison of the genetic tumor classification with the risk, that is, potentially malignant. Tumors lacking these altera- histopathologic diagnosis [according to Brat and colleagues tions being EIF1AX, SF3B1, and BAP1-wild-type or harboring an (2)] demonstrated relatively poor correlation (Supplementary EIF1AX mutation (SF3B1 and BAP1-wild-type) could be assumed Table S4). This is not entirely surprising, given that the distribu- to have a favorable prognosis. SF3B1-mutant tumors should be tion of EIF1AX, SF3B1, and BAP1 mutations did not fit well with considered to be intermediate-risk tumors, as both blue nevus– the histopathologic diagnoses (Fig. 2; Table 1). The availability of like melanomas and uveal melanomas with these mutations can only limited follow-up data for the patients in our cohort (Sup- metastasize (20, 35). In addition, we would suggest including plementary Table S3) precludes comparisons of the genetic and tumors with chromosome 3 loss but no detectable BAP1 alter- the Brat and colleagues diagnostic schemes with respect to clinical ation (mutation or BAP1 IHC loss) to the intermediate-risk group. outcomes. Histopathologic evaluation remains a key component Applying this risk stratification scheme to our cohort of primary of the diagnostic process, and tumors with overt cell pleomor- leptomeningeal tumors, 16/47 (34%) cases would have been phism, high mitotic index, and necrosis should be considered

www.aacrjournals.org Clin Cancer Res; 24(18) September 15, 2018 4501

Griewank et al.

Figure 5. Mutation and copy number proﬁle of GNAQ and EIF1AX mutated tumors. Shown is an example of a primary leptomeningeal melanocytic tumor (PLMT) diagnosed as a melanocytoma, as well as a uveal melanoma. These samples harbored similar mutations with a GNAQ c.626A>T Q209L or c.626A>CQ209P mutation, respectively, as well as EIF1AX G15D c.44G>A mutations. The copy number proﬁles of these tumors, depicted on the right, demonstrate a common gain of Chr. 6p.

malignant (or high-risk) tumors independently of genetic find- leptomeningeal tumors may profit from these therapies. How- ings. However, similar to the case in uveal melanoma, we believe ever, as long as other promising therapies are still lacking, that genetic testing may become a useful tool to help predict immune checkpoint blockade should still be considered a viable tumor behavior and patient prognosis in primary leptomeningeal therapeutic option in these patients. melanocytic tumors in which unequivocal histopathologic fea- Our study has some limitations. The cohort of primary lepto- tures of malignancy are not observed. meningeal tumors is large, but in many cases the follow-up data The two retrobulbar melanomas included in the study were were missing or incomplete. The comparison with uveal mela- classified as primary leptomeningeal melanomas rather than noma and blue nevus–like melanoma is informative, but the uveal melanomas, because in both cases, imaging and fundo- available sample numbers of the latter tumors was limited scopy showed no evidence of a primary uveal tumor. Ultimately, due to their rarity. Studies involving larger cohorts with complete given the genetic similarities between primary leptomeningeal follow-up data would be valuable to be able to validate associa- melanocytic tumors and uveal melanomas, this distinction may tions of gene alterations with prognosis in primary leptomenin- not necessarily be clinically important for management of affected geal tumors. patients. In summary, our study is to date the most comprehensive The observed genetic similarity of primary leptomeningeal genetic analysis of primary leptomeningeal tumors, demonstrat- tumors with uveal melanomas and blue nevus–like melanomas ing they can be clearly distinguished from cutaneous melanoma may have therapeutic consequences. Uveal melanomas and blue metastasis and have a strong genetic relationship with uveal nevus–like melanomas have shown very poor response rates to melanoma and blue nevus–like melanoma. Our data suggest that immunotherapy with anti CTLA-4 and PD-1 antibody therapies determining copy number and mutation status in primary lepto- (30, 36–38). These suggest few patients with advanced primary meningeal tumors may assist in prognostic assessment.

4502 Clin Cancer Res; 24(18) September 15, 2018 Clinical Cancer Research

Genetic Analysis of CNS Melanocytic Tumors

Disclosure of Potential Conﬂicts of Interest J.A.P.vandeNes,I.Moller,€ A. Sucker, T. Pietsch, A. von Deimling, No potential conﬂicts of interest were disclosed. D. Schadendorf Study supervision: K.G. Griewank, A. von Deimling

Authors' Contributions Acknowledgments The authors thank Nadine Stadler and Julia Kretz for excellent technical Conception and design: K.G. Griewank, C. Koelsche, J.A.P. van de Nes, A. von assistance. Assistance from colleagues at Melanoma Institute Australia and the Deimling Royal Prince Alfred Hospital is also gratefully acknowledged. The research was Development of methodology: K.G. Griewank, A. von Deimling supported by a grant from the Dr. Werner-Jackst€adt-Stiftung (www.jackstaedt- Acquisition of data (provided animals, acquired and managed patients, stiftung.de) and Hiege-Stiftung gegen Hautkrebs. This research was also funded provided facilities, etc.): K.G. Griewank, C. Koelsche, J.A.P. van de Nes, in part through the NIH/NCI Cancer Center Support Grant P30 CA008748. The M. Gessi, A. Sucker, R.A. Scolyer, M.E. Buckland, R. Murali, T. Pietsch, funders had no role in study design, data collection and analysis, decision to A. von Deimling, D. Schadendorf publish, or preparation of the manuscript. Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K.G. Griewank, C. Koelsche, D. Schrimpf, R. Murali, A. von Deimling, D. Schadendorf The costs of publication of this article were defrayed in part by the payment of advertisement Writing, review, and/or revision of the manuscript: K.G. Griewank, C. page charges. This article must therefore be hereby marked in Koelsche, J.A.P. van de Nes, M. Gessi, R.A. Scolyer, M.E. Buckland, R. Murali, accordance with 18 U.S.C. Section 1734 solely to indicate this fact. T. Pietsch, A. von Deimling, D. Schadendorf Administrative, technical, or material support (i.e., reporting or Received March 7, 2018; revised April 28, 2018; accepted June 5, 2018; organizing data, constructing databases): K.G. Griewank, C. Koelsche, published ﬁrst June 11, 2018.

References 1. Brat D, Perry A, Wesseling P, Bastian B. Melanocytic tumours. In:Louis D, 14. Krauthammer M, Kong Y, Ha BH, Evans P, Bacchiocchi A, McCusker JP, Ohgaki H, Wiestler O, Cavanee W, editors. WHO classification of tumours et al. Exome sequencing identifies recurrent somatic RAC1 mutations in of the central nervous system. Lyon: IARC Press; 2016. p266–70. melanoma. Nat Genet 2012;44:1006–14. 2. Brat DJ, Giannini C, Scheithauer BW, Burger PC. Primary melanocytic 15. Cancer Genome Atlas N. Genomic classification of cutaneous melanoma. neoplasms of the central nervous systems. Am J Surg Pathol 1999;23: Cell 2015;161:1681–96. 745–54. 16. Hayward NK, Wilmott JS, Waddell N, Johansson PA, Field MA, Nones K, 3. GessiM,vandeNesJ,GriewankK,BarresiV,BucklandME,KirfelJ,etal. et al. Whole-genome landscapes of major melanoma subtypes. Nature Absence of TERT promoter mutations in primary melanocytic tumours 2017;545:175–80. of the central nervous system. Neuropathol Appl Neurobiol 2014;40: 17. Horn S, Figl A, Rachakonda PS, Fischer C, Sucker A, Gast A, et al. TERT 794–7. promoter mutations in familial and sporadic melanoma. Science 2013; 4. Kusters-Vandevelde HV, van Engen-van Grunsven IA, Kusters B, van Dijk 339:959–61. MR, Groenen PJ, Wesseling P, et al. Improved discrimination of mel- 18. Griewank KG, Murali R, Puig-Butille JA, Schilling B, Livingstone E, Potrony anotic schwannoma from melanocytic lesions by combined morpho- M, et al. TERT promoter mutation status as an independent prognostic logical and GNAQ mutational analysis. Acta Neuropathol (Berl) factor in cutaneous melanoma. J Natl Cancer Inst 2014;106:dju246. 2010;120:755–64. 19. Moller I, Murali R, Muller H, Wiesner T, Jackett LA, Scholz SL, et al. 5. Kusters-Vandevelde HV, Klaasen A, Kusters B, Groenen PJ, van Engen-van Activating cysteinyl leukotriene receptor 2 (CYSLTR2) mutations in blue Grunsven IA, van Dijk MR, et al. Activating mutations of the GNAQ gene: a nevi. Mod Pathol 2017;30:350–6. frequent event in primary melanocytic neoplasms of the central nervous 20. Griewank KG, Muller H, Jackett LA, Emberger M, Moller I, van de Nes JA, system. Acta Neuropathol (Berl) 2010;119:317–23. et al. SF3B1 and BAP1 mutations in blue nevus–like melanoma. Mod 6. Murali R, Wiesner T, Rosenblum MK, Bastian BC. GNAQ and GNA11 Pathol 2017;30:928–39. mutations in melanocytomas of the central nervous system. Acta Neuro- 21. Held L, Eigentler TK, Metzler G, Leiter U, Messina JL, Glass LF, et al. pathol (Berl) 2012;123:457–9. Proliferative activity, chromosomal aberrations, and tumor-specific muta- 7. Kusters-Vandevelde HV, van Engen-van Grunsven IA, Coupland SE, Lake tions in the differential diagnosis between blue nevi and melanoma. Am J SL, Rijntjes J, Pfundt R, et al. Mutations in g protein encoding genes and Pathol 2013;182:640–5. chromosomal alterations in primary leptomeningeal melanocytic neo- 22. Loghavi S, Curry JL, Torres-Cabala CA, Ivan D, Patel KP, Mehrotra M, et al. plasms. Pathol Oncol Res 2015;21:439–47. Melanoma arising in association with blue nevus: a clinical and pathologic 8. Koelsche C, Hovestadt V, Jones DT, Capper D, Sturm D, Sahm F, et al. study of 24 cases and comprehensive review of the literature. Mod Pathol Melanotic tumors of the nervous system are characterized by distinct 2014;27:1468–78. mutational, chromosomal and epigenomic profiles. Brain Pathol 23. Yeh I, Mully TW, Wiesner T, Vemula SS, Mirza SA, Sparatta AJ, et al. 2015;25:202–8. Ambiguous melanocytic tumors with loss of 3p21. Am J Surg Pathol 9. van de Nes JAP, Koelsche C, Gessi M, Moller I, Sucker A, Scolyer RA, et al. 2014;38:1088–95. Activating CYSLTR2 and PLCB4 mutations in primary leptomeningeal 24. Costa S, Byrne M, Pissaloux D, Haddad V, Paindavoine S, Thomas L, et al. melanocytic tumors. J Invest Dermatol 2017;137:2033–5. Melanomas associated with blue nevi or mimicking cellular blue nevi: 10. Kusters-Vandevelde HV, Creytens D, van Engen-van Grunsven AC, Jeunink clinical, pathologic, and molecular study of 11 cases displaying a high M, Winnepenninckx V, Groenen PJ, et al. SF3B1 and EIF1AX mutations frequency of GNA11 mutations, BAP1 expression loss, and a predilection occur in primary leptomeningeal melanocytic neoplasms; yet another for the scalp. Am J Surg Pathol 2016;40:368–77. similarity to uveal melanomas. Acta Neuropathol Commun 2016;4:5. 25. Dai J, Tetzlaff MT, Schuchter LM, Elder DE, Elenitsas R. Histopathologic 11. van de Nes J, Gessi M, Sucker A, Moller I, Stiller M, Horn S, et al. Targeted and mutational analysis of a case of blue nevus–like melanoma. J Cutan next generation sequencing reveals unique mutation profile of primary Pathol 2016;43:776–80. melanocytic tumors of the central nervous system. J Neurooncol 2016; 26. van de Nes JA, Nelles J, Kreis S, Metz CH, Hager T, Lohmann DR, et al. 127:435–44. Comparing the prognostic value of BAP1 mutation pattern, chromosome 3 12. van de Nes J, Wrede K, Ringelstein A, Stiller M, Horn S, Sucker A, et al. status, and BAP1 immunohistochemistry in uveal melanoma. Am J Surg Diagnosing a primary leptomeningeal melanoma by gene mutation sig- Pathol 2016;40:796–805. nature. J Invest Dermatol 2016;136:1526–8. 27. Curtin JA, Fridlyand J, Kageshita T, Patel HN, Busam KJ, Kutzner H, et al. 13. Hodis E, Watson IR, Kryukov GV, Arold ST, Imielinski M, Theurillat JP, et al. Distinct sets of genetic alterations in melanoma. N Engl J Med 2005;353: A landscape of driver mutations in melanoma. Cell 2012;150:251–63. 2135–47.

www.aacrjournals.org Clin Cancer Res; 24(18) September 15, 2018 4503

Griewank et al.

28. Robertson AG, Shih J, Yau C, Gibb EA, Oba J, Mungall KL, et al. Integrative 34. Matatall KA, Agapova OA, Onken MD, Worley LA, Bowcock AM, Harbour analysis identifies four molecular and clinical subsets in uveal melanoma. JW. BAP1 deficiency causes loss of melanocytic cell identity in uveal Cancer Cell 2017;32:204–20.e15. melanoma. BMC cancer 2013;13:371. 29. Bastian BC. The molecular pathology of melanoma: an integrated taxon- 35. Yavuzyigitoglu S, Koopmans AE, Verdijk RM, Vaarwater J, Eussen B, van omy of melanocytic neoplasia. Annu Rev Pathol 2014;9:239–71. Bodegom A, et al. Uveal melanomas with SF3B1 mutations: a distinct 30. Johnson DB, Roszik J, Shoushtari AN, Eroglu Z, Balko JM, Higham C, et al. subclass associated with late-onset metastases. Ophthalmology 2016; Comparative analysis of the GNAQ, GNA11, SF3B1, and EIF1AX driver 123:1118–28. mutations in melanoma and across the cancer spectrum. Pigment Cell 36. Algazi AP, Tsai KK, Shoushtari AN, Munhoz RR, Eroglu Z, Piulats JM, et al. Melanoma Res 2016;29:470–3. Clinical outcomes in metastatic uveal melanoma treated with PD-1 and 31. Harbour JW, Onken MD, Roberson ED, Duan S, Cao L, Worley LA, et al. PD-L1 antibodies. Cancer 2016;122:3344–53. Frequent mutation of BAP1 in metastasizing uveal melanomas. Science 37. Zimmer L, Vaubel J, Mohr P, Hauschild A, Utikal J, Simon J, et al. Phase II 2010;330:1410–3. DeCOG-study of ipilimumab in pretreated and treatment-naive patients 32. Martin M, Masshofer L, Temming P, Rahmann S, Metz C, Bornfeld N, et al. with metastatic uveal melanoma. PLoS One 2015;10:e0118564. Exome sequencing identifies recurrent somatic mutations in EIF1AX and 38. Heppt MV, Heinzerling L, Kahler KC, Forschner A, Kirchberger MC, SF3B1 in uveal melanoma with disomy 3. Nat Genet 2013;45:933–6. Loquai C, et al. Prognostic factors and outcomes in metastatic uveal 33. Harbour JW, Roberson ED, Anbunathan H, Onken MD, Worley LA, Bow- melanoma treated with programmed cell death-1 or combined PD-1/ cock AM. Recurrent mutations at codon 625 of the splicing factor SF3B1 in cytotoxic T-lymphocyte antigen-4 inhibition. Eur J Cancer 2017;82: uveal melanoma. Nat Genet 2013;45:133–5. 56–65.

4504 Clin Cancer Res; 24(18) September 15, 2018 Clinical Cancer Research

Integrated Genomic Classification of Melanocytic Tumors of the Central Nervous System Using Mutation Analysis, Copy Number Alterations, and DNA Methylation Profiling

Klaus G. Griewank, Christian Koelsche, Johannes A.P. van de Nes, et al.

Clin Cancer Res 2018;24:4494-4504. Published OnlineFirst June 11, 2018.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-18-0763

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2018/06/09/1078-0432.CCR-18-0763.DC1

Cited articles This article cites 37 articles, 2 of which you can access for free at: http://clincancerres.aacrjournals.org/content/24/18/4494.full#ref-list-1

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

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/24/18/4494. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.