Genomic and Transcriptomic Features of Dermatofibrosarcoma Protuberans

Cancer Genetics 241 (2020) 34–41

Contents lists available at ScienceDirect

Cancer Genetics

journal homepage: www.elsevier.com/locate/cancergen

Genomic and transcriptomic features of dermatoﬁbrosarcoma protuberans: Unusual chromosomal origin of the COL1A1 - PDGFB fusion gene and synergistic effects of ampliﬁed regions in tumor development

∗ Jan Köster a,b, , Elsa Arbajian a, Björn Viklund c, Anders Isaksson c, Jakob Hofvander a,

Felix Haglund d, Henrik Bauer e, Linda Magnusson a, Nils Mandahl a, Fredrik Mertens a,b a Division of Clinical Genetics, Department of Laboratory Medicine, Lund University, Lund, Sweden b Department of Clinical Genetics and Pathology, Division of Laboratory Medicine, Lund, Sweden c Science for Life Laboratory, Department of Medical Sciences, Uppsala University, Sweden d Department of Pathology and Cytology, Karolinska University Hospital, Stockholm, Sweden e Department of Orthopedics, Karolinska University Hospital, Stockholm, Sweden

a r t i c l e i n f o a b s t r a c t

Article history: The dermatoﬁbrosarcoma protuberans family of tumors (DPFT) comprises cutaneous soft tissue neo- Received 20 June 2019 plasms associated with aberrant PDGFBR signaling, typically through a COL1A1-PDGFB fusion. The aim of

Revised 6 November 2019 the present study was to obtain a better understanding of the chromosomal origin of this fusion and to Accepted 6 December 2019 assess the spectrum of secondary mutations at the chromosome and nucleotide levels. We thus investi- gated 42 tumor samples from 35 patients using chromosome banding, fluorescence in situ hybridization, Key words: single nucleotide polymorphism arrays, and/or massively parallel sequencing (gene panel, whole exome Dermatofibrosarcoma and transcriptome sequencing) methods. We confirmed the age-associated differences in the origin of

Fusion the COL1A1 - PDGFB fusion and could show that it in most cases must arise after DNA synthesis, i.e., in

SNP-array the S or G2 phase of the cell cycle. Whereas there was a non-random pattern of secondary chromoso- RNA-seq mal rearrangements, single nucleotide variants seem to have little impact on tumor progression. No clear Translocation mechanisms genomic differences between low-grade and high-grade DPFT were found, but the number of chromosomes and chromosomal imbalances as well as the frequency of 9p deletions all tended to be greater among the latter. Gene expression profiling of tumors with COL1A1 - PDGFB fusions associated with unbalanced translocations or ring chromosomes identified several transcriptionally up-regulated genes in the amplified regions of chromosomes 17 and 22, including TBX2, PRKCA, MSI2, SOX9, SOX10 , and PRAME . ©2019 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license. ( http://creativecommons.org/licenses/by-nc-nd/4.0/ )

Introduction extremities of young or middle-aged adults. Due to both molecular and genetic overlap, a variety of other entities, such as Bednar tu- Dermatofibrosarcoma protuberans (DP) is a rare ( < 1/10 0,0 0 0 mor and giant cell fibroblastoma (GCF), are nowadays included in individuals per year), locally aggressive mesenchymal tumor with the DP family of tumors (DPFT). often recurrent local disease (up to 50%), but only low metastatic Genetically, DPFT is, in the vast majority of cases, character- risk [1] . However, DP has a potential for transformation into a fi- ized by a COL1A1 - PDGFB gene fusion; the promoter and variable brosarcoma of higher grade. Tumors with fibrosarcomatous trans- portions of the collagen 1A1 ( COL1A1 ) gene are fused with exon formation show the same tendency for local recurrence as regular 2 of the platelet-derived growth factor beta ( PDGFB ) gene, result- DP, but are associated with metastatic disease in about 13% of the ing in deregulated expression of the PDGFB protein [3, 4] . At the cases [2] . DP occurs most commonly on the trunk and proximal chromosome level, the gene fusion is caused by an exchange of material between chromosome bands 17q21 ( COL1A1 ) and 22q13 ( PDGFB ). This exchange may appear in the form of a balanced or ∗ Corresponding author at: Division of Clinical Genetics, Department of Labora- tory Medicine, Lund University, Lund, Sweden. unbalanced t(17;22) or as one or more supernumerary ring chro-

E-mail address: [email protected] (J. Köster). mosomes [1,5,6]. The latter, which may contain multiple copies of https://doi.org/10.1016/j.cancergen.2019.12.001 2210-7762/© 2019 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license. ( http://creativecommons.org/licenses/by-nc-nd/4.0/ ) J. Köster, E. Arbajian and B. Viklund et al. / Cancer Genetics 241 (2020) 34–41 35 the fusion gene as well as of other portions of chromosome arms cleotide polymorphism (SNP) arrays (Cytoscan HD, Affymetrix, 17q and 22q, occurs more frequently in older patients [7] whereas Santa Clara, CA, USA), as described [32] . The position of the the former has been described particularly in children [8, 9] . The probes was aligned according to the USCS hg19/NCBI Build 37 se- COL1A1 - PDGFB fusion is specific for DPFT. Hence, both fluorescence quences. Tumor Aberration Prediction Suite (TAPS) and Rawcopy in situ hybridization (FISH) and RT-PCR based methods are well es- were used for segmentation of copy number shifts, copy number tablished diagnostic tools [4, 7, 10–12] . Rare cases combining PDGFB evaluation, and visualization of the data in the 12 samples show- with other 5 -partners have been described, and recently it was ing copy number variation within the amplified regions of 17q shown that at least a subset of DPFT of the breast in women have and/or 22q [33, 34] . Constitutional copy number variations were ex- a COL6A3 - PDGFD fusion that, similar to the more common COL1A1- cluded through comparison with the Database of Genomic Variants PDGFB fusion, activates the PDGFB receptor [13, 14] . (http://projects.tcag.ca/variation/). Whereas the pathogenetic significance of the COL1A1 - PDGFB fusion has been thoroughly documented, the information on sec- MPS-based analyses ondary genomic changes and to what extent these may vary with age or outcome is still limited. Abnormal karyotypes have been re- From eight of the samples analyzed by SNP array, also tar- ported in 47 cases [15] and a few studies have used comparative geted DNA sequencing of 50 neoplasia-associated genes was per- genomic hybridization (CGH) or array-based approaches to eval- formed ( Table 1 ), using the Ion AmpliSeq Cancer Hot Spot Panel v2 uate the spectrum of genomic imbalances in DPFT [16–21] . Even (ThermoFisher Scientific, Waltham, MA, USA) as described [35] . Ten fewer cases have been analyzed with techniques based on mas- samples were analyzed by RNA-seq ( Table 1 ), as described [36] . sively parallel sequencing (MPS; 20–22 ). Gene expression levels in the DPFT were compared with those in Here, we present a series of 35 patients with DPFT that were a cohort of 64 soft tissue tumors (STT). Differences between tu- analyzed with regard to genomic changes by chromosome banding, mor types in log2-transformed expression data were calculated us- FISH, and/or high-resolution single nucleotide polymorphism (SNP) ing a t -test, and corrections for multiple testing were based on the array analyses. In addition, selected cases were analyzed with re- Benjamini–Hochberg method (Qlucore AB). The two tumor samples gard to additional mutations and gene expression profile using and peripheral blood from Case 28 were analyzed by WES (Illu- MPS-based gene panel, whole exome (WES), and transcriptome se- mina, San Diego, CA, USA), with targeted resequencing for seven of quencing (RNA-seq). the mutations detected at WES ( Table 1 ); these analyses were performed as described [37, 38] . WES and targeted resequencing gen-

Materials and methods erated an average coverage of 99X and 145X, respectively.

Patients and tumors Results The study encompassed 42 tumor samples from 35 patients Chromosome banding and FISH analyses (27 men, 8 women; age 1–76 years) collected from 1985 to 2018 ( Table 1 ). The patients were treated at the sarcoma centers in Lund All 39 samples from the 32 patients that were analyzed and Stockholm, Sweden. The study was based on 28 samples from by chromosome banding analysis showed abnormal karyotypes primary lesions, 12 from local recurrences, and one from a metas- ( Table 1 ). A balanced t(17;22)(q21;q13), 1–2 copies of an unbal- tasis; information on the origin was missing in one patient (Case anced der(22)t(17;22), and 1–4 ring chromosomes were found in 33). All diagnoses were revised, based on the morphological pic- one, nine, and 25 patients, respectively; one tumor (Case 3) had ture and immune profile, according to established criteria [1] . Two one clone with a balanced and one with an unbalanced transloca- tumors were diagnosed as GCF. Six of the tumors were classified as tion, and in four patients both an unbalanced der(22)t(17;22) and high-grade DP with fibrosarcomatous transformation, morphologi- ring chromosomes co-existed in the same or different clones. The cally indicated by architectural changes, more pronounced atypia, single case with a balanced t(17;22) was a 7-year-old boy, whereas increased mitotic activity, and in exceptional cases areas of tumor the tumors with an unbalanced der(22)t(17;22) or ring chromo- necrosis. somes were found in patients aged 1–50 years and 16–76 years, Cytogenetic and FISH analyses respectively. One patient (Case 28) did not display any of these cytogenetic features; both samples from this high-grade DP had a Chromosome banding analysis was performed on 39 samples complex aneuploid karyotype. from 32 patients, as described [23] . The nomenclature of the kary- Secondary chromosomal aberrations, here defined as rearrange- otypes followed the guidelines of the International System for Hu- ments other than ring chromosomes or affecting other chromo- man Cytogenomic Nomenclature [24] . The karyotypes from seven somes than 17 and 22, were found at G-banding analysis of 21 samples were published before ( Table 1 ; 25–30 ). Metaphase FISH tumors ( Table 1 ). Most tumors had a pseudo- or hyperdiploid analysis for the detection of PDGFB rearrangement and the COL1A1 - chromosome count; hypodiploid or triploid-hypertetraploid clones PDGFB fusion gene and to identify ring chromosomes was per- were seen in two tumors each. Among the secondary aberrations, formed in eight cases; one case was analyzed with interphase FISH numerical changes predominated over structural rearrangements, for PDGFB status only ( Table 1 ). FISH analyses were performed the most common being trisomy 8 (seven cases), loss of one or as described [31] , using the ZytoLight SPEC COL1A1 - PDGFB Dual more chromosomes 22 (six cases), and trisomy 5 and 18 (four Color Dual Fusion Probe and the ZytoLight SPEC PDGFB Dual Color cases each). Structural rearrangements, none of which was re- Break Apart Probe (ZytoVision, Bremerhaven, Germany). Sequences current, other than ring chromosomes or t(17;22)/der(22)t(17;22) of chromosomes 17 and 22 were identified using whole chromo- were seen in 10 cases, including three high-grade DP. some painting (WCP) probes (Applied Spectral Imaging, Migdal In seven cases, more than one sample was available for chromo- Haemek, Israel). some banding analysis. Intralesional clonal heterogeneity, studied in the four cases with data on both a preoperative needle biopsy SNP array analysis and the excised primary tumor, was restricted to a variable number of ring chromosomes in one (Case 11) and different structural Extracted DNA from 16 frozen tumor samples from 15 pa- and numerical secondary aberrations in one (Case 34). In three patients ( Table 1 ) was analyzed using high-resolution single nu- tients, clonal evolution with time could be studied. Two local re- 36 J. Köster, E. Arbajian and B. Viklund et al. / Cancer Genetics 241 (2020) 34–41

Table 1 Clinicopathologic features of 35 cases of dermatoﬁbrosarcoma protuberans family of tumors subjected to genetic analyses.

Case No Sample a Age Sex b Site c Size d DX e Karyotype Additional analyses f 1 P 1 M Th wall 4 GCF 46,XY,der(22)t(17;22)(q21;q13) SNP 2 P 1 M L arm NA GCF 46,XY,der(22)t(17;22)(q21;q13) FISH 3 P 7 M Back NA LGDP 46,XY,t(17;22)(q21;q13)/46,idem,der(22)t(17;22) SNP, FISH, GP 4 P 16 M L leg 8 LGDP 47,XY, + 18,der(22)t(17;22)(q21;q13)/47,XY, + 18,der(22)r(17;22) SNP, GP (q21q25;p13q13) 5 LR 19 M Back 6 LGDP 46–50,XY, + 2r SNP, FISH, RNA-Seq, GP 6 LR 21 F Tr wall 3.5 LGDP Not done SNP 7 P 22 M Thigh 0.4 LGDP 47,XY, + ?ider(22)(q10)t(17;22)(q21;q13)/47–48,idem, + 2 8a P 22 M L leg 3.5 LGDP 49,XY, + 8, + 18, + 22,der(22)t(17;22)(q21;q13)x2 8b P LGDP 49,XY, + 8, + 18, + 22,der(22)t(17;22)(q21;q13)x2 SNP, RNA-Seq, GP 9 LR 30 M Shoulder 2 LGDP 48,XY, + 8, + der(22)r(17;22)(q12-21q24-

25;p11q13)/48,XY, + der(22)r(17;22), + der(?)t(?;22) g 10 P 33 M Shoulder 3 LGDP Not done SNP, RNA-Seq

11a P 36 M Back 9 LGDP 47,XY, + 5, −21,der(22)t(?17;22)(q21;q13), + r g

11b P LGDP 47–50,XY, + 5, −21,der(22)t(?17;22)(q21;q13), + 1–4r g

12a LR1 40 F Shoulder 4 LGDP 47,XX, + der(1)r(1;?)(p36q44;?) g

12b LR2 2.5 LGDP 47,XX, + der(1)r(1;?)(p36q44;?) g 13 P 41 M Th wall 6 LGDP 48–50,XY, + 5, + 8, + 11, + 21, + r, + mar SNP, FISH, RNA-Seq, GP 14a P 42 M U arm 6.5 HGDP 50,XY, + 8, + 18, + 21, + 22,der(22)t(17;22)(q21;q13)x2 14b P HGDP 50,XY, + 8, + 18, + 21, + 22,der(22)t(17;22)(q21;q13)x2 FISH 15 P 43 F Shoulder 1.6 LGDP 47,XX, −5,der(14)t(5;14)(q11;q31),?del(22)(q11), + r, + mar/47,idem, add(7)(p11)/47,idem,t(3;9)(q25;q13), + 5,-mar 16 P 47 M Shoulder 4 LGDP 47,XY, + 5, −22, + r/46–47,XY,add(8)(p11), + 12, −14, SNP, RNA-Seq der(16)t(14;16)(q11;q23),add(22)(q11), + r 17 P 48 M Groin 3 LGDP 47,XY, + r/47,XY, + mar

18 P 48 M Tr wall 5 LGDP 47,XY, + r g 19 P 50 M L arm 1 HGDP 47,XY, + 22,der(22)t(17;22)(q21;q13)x2/48,idem, + r FISH 20 P 50 F Shoulder 3.5 LGDP 46–49,XX, + 4, + 8, −19, −20, −22, + r FISH 21 P 55 M Back 4 HGDP 46–48,XY,der(3)t(3;?5)(q29;q21), −22, + 1–2r FISH 22 P 57 M Groin 2.5 LGDP 87,XX,-Y,-Y, −3, −6, + 8, −11, −14, −22, + r/86,idem, −9 23 LR 59 M Tr wall 6 HGDP Not done SNP

24 LR 59 M Groin NA LGDP 47,XY, + r g 25 P 60 M L leg 2 HGDP 51–53,XY, + 5, + 7, + 11, + 12, −22, + r, + 1–2mar SNP, RNA-Seq 26 P 60 F Thigh 3.5 LGDP 47,XX, + r SNP, RNA-Seq, GP 27 P 60 M Shoulder 3 LGDP 46–47,XY, + r SNP, RNA-Seq, GP 28a P 64 M Shoulder 4.5 HGDP 100– SNP, WES 101,XXYY, −3,del(6)(q23), + 7, −10, + 15,?add(16)(q?12)x2, + 18, −19, + 20, + 5mar 28b Met Lung 5 HGDP 79–83,XYY,- SNP, WES, RNA-Seq X, −2, −3, −4,add(5)(p15), −6,del(6)(q23), −8, −9, −9, −10,add(11)(p11), −13, −14, −15, −15, + del(17)(p11), −19, −21, + 2–3mar 29 LR 64 M Shoulder 4 LGDP 49–51,XY, + 2, + r, + mar 30 P 66 F Shoulder 1.5 LGDP 46,XX, −22, + r/44–45,idem,-X/46–47,idem, + 4/46–48,idem, + 4, + r 31a LR1 66 M Shoulder 5 LGDP 45,XY,add(5)(p13),del(7)(q22q32), −10,add(20)(q12), −22, −22, + r, SNP, FISH, RNA-Seq, GP + der(?)t(?;22)(?;q1?) 31b LR2 Shoulder 4.5 LGDP 44–45,XY,add(5)(p13),del(7)(q22q32), −10,add(20)(q12), −22, −22, + mar/44– 45,idem, + r 32 LR 67 F Groin NA LGDP 49,XX, + ?del(4)(q21),dic(5;22)(p12- 13;p11),ins(5;?)(q31;?), + del(8)(p21), + r, + mar 33 NA 69 F Back NA LGDP 47,XX, + r 34a P 69 M Tr wall 6 LGDP 47,XY, + r/48,idem,add(1)(p11), + mar FISH 34b P LGDP 47–50,XY,add(?4)(p11), + 8, + r, + mar,inc 35 LR 76 M NA 0.7 LGDP 47,XY, + r/46,X,-Y, + r/45,X,-Y

a P = primary tumor; LR = local recurrence; Met = metastasis; NA = not available.

b M = male; F = female.

c L = lower; Th = thoracic; Tr = trunk; U = upper; NA = not available.

d Largest diameter in cm; NA = not available.

e DX = diagnosis; GCF = giant cell fibroblastoma; LGDP = low-grade dermatofibrosarcoma protuberans; HGDP = high-grade dermatofibrosarcoma protuberans.

f SNP = Single nucleotide polymorphism array (Affymetrix Cytoscan HD); FISH = ﬂuorescence in situ hybridization; WES = whole exome sequencing; RNA-seq = mRNA sequencing; GP = Ion AmpliSeq Cancer Hot Spot Panel v2.

g Previously published karyotype. currences in Cases 12 and 31 displayed no or only subtle differ- with t(17;22)/der(22)t(17;22) showed 1–2 extra copies of 17q21- ences, whereas the primary tumor and a metastasis in Case 28 dif- qter with the copy number shift in or near the COL1A1 locus and fered more extensively. loss of 22q13-qter with the copy number shift in or near the Metaphase FISH analysis was performed on nine tumors, show- PDGFB locus. Gain of the centromeric part of 22q (there are no in- ing one fusion event on each der(22) in the samples with translo- formative probes until nt position 16,877,134 in subband 22q11.1) cations, and up to three fusions in the ring chromosomes. In was seen in only one of these tumors (Case 8). There was no the DP from Case 19, fusion signals were seen only on the copy number variation within the gained or deleted segments and der(22)t(17;22) chromosomes, but not in the ring chromosome the segment from 17pter-q21 always displayed retained heterodi- found in a subclone ( Table 1 ; Fig. 1 a). somy, while the segment distal to the PDGFB locus on 22q consistently was mono-allelic. Case 6, a local recurrence 20 years after surgery for a primary tumor and for which no G-band data SNP array analysis were available, showed a pattern similar to that of other cases with der (22) t (17;22). SNP array analysis revealed copy number changes in all analyzed tumors (Supplementary Table 1 ; Fig. 2 a). All four tumors J. Köster, E. Arbajian and B. Viklund et al. / Cancer Genetics 241 (2020) 34–41 37

Fig. 1. A, FISH analysis of dermatoﬁbrosarcoma protuberans (DP; Case 19) using a break-apart probe for PDGFB (5 in green, 3 in red) and a whole chromosome painting probe for chromosome 22 (blue). The two der(22)t(17;22) showed a split PDBGF signal (red, arrowhead), while the normal chromosome 22 showed an intact PDGFB locus (yellow; thick arrow). The ring chromosome was negative for PDGFB and chromosome 22 material (thin arrow). B, Circos plot showing copy number changes and non- synonymous exonic variants detected with single nucleotide polymorphism (SNP) array and whole exome sequencing (WES), respectively, in the primary tumor (PT) and a metastasis (Met) from Case 28. The circles are ordered chronologically, starting from the center with the PT. The pink/green inner circles represent the location and amplitude of the allelic imbalances in relation to the ploidy level (4n); blue is gain, gray is loss and yellow is copy neutral loss of heterozygosity. Differences between the PT and the Met are indicated by red changes in the most peripheral circle, which otherwise appears green. The light blue circles represent the location of the variants reported by the whole exome sequencing. C, Possible outcomes of a t(17;22) in different phases of the cell cycle. The two parental copies of chromosomes 17 and 22 are shown in different shades of blue and gray, respectively. (1) Trisomy 17 is followed by t(17;22) in G0-G1 and loss of der(17), resulting in UPID for the disomic part of chr 17 in 1/3 of the cases. (2) t(17;22) in G0-G1 with subsequent loss of der(17) and duplication of the normal homologue, always leading to UPID of the disomic parts of chr 17. (3) Recombinations in G2 would consistently result in retained heterodisomy for the disomic parts of chromosomes 17 and 22 in DP cases with (I) der(22)t(17;22) or (II) ring chromosomes. (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of this article.) 38 J. Köster, E. Arbajian and B. Viklund et al. / Cancer Genetics 241 (2020) 34–41

Fig. 2. Single nucleotide polymorphism (SNP) array and RNA-sequencing data in cases of dermatofibrosarcoma protuberans (DPFT). A, SNP array with genome-wide overview on copy number changes in the subset of the DPFT showing copy number variation within the amplified regions of 17q and/or 22q. Green signals indicate gain, blue signals loss, and gray signals copy neutral loss of heterozygosity. B and C, Copy number changes are shown in detail for chromosomes 17 and 22. D and E, Heat maps of expression levels for genes on chromosomes 17 and 22 in DPFT and other soft tissue tumors; red indicates increased and green decreased gene expression. Light blue = benign fibrous histiocytoma; dark blue = DPFT; orange = inflammatory leiomyosarcoma; green = myxoinflammatory fibroblastic sarcoma; white = myxofibrosarcoma; purple = ossifying fibromyxoid tumor; red = PRDM10 -positive tumors; black = sclerosing epithelioid fibrosarcoma; yellow = undifferentiated pleomorphic sarcoma. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) J. Köster, E. Arbajian and B. Viklund et al. / Cancer Genetics 241 (2020) 34–41 39

Also all of the eight DP with ring chromosomes showed gain of both a balanced and an unbalanced translocation in different sub- 17q21-qter, usually between 1–4 extra copies, with the shift from clones or different samples, and six (2–13 yrs) had only an un- a normal, heterodisomic state in or near COL1A1 ; in contrast to the balanced der(22)t(17;22) or variants thereof. Of the 38 DP from tumors with der(22)t(17;22), six of the cases showed copy number adults, 32 displayed one or more ring chromosomes, the youngest variation within the gained segment. The pattern for 22q was even patient being 30 years at diagnosis. None of the adults had a more complex, with varying portions of 22q11-q13 being gained balanced t(17;22), but two, aged 38 and 47 yrs, had the same at different copy number levels; a copy number shift, correspond- der(22)t(17;22) as seen in the pediatric setting [15] . ing to 1–4 extra copies, was always seen at the PDGFB locus. In In our cohort of 32 cytogenetically analyzed DPFT, the age at di- four tumors, part of 22q11-q13 was disomic. In 5/7 tumors, the agnosis ranged from 1–76 years. All four patients aged 1–16 years segment distal to the PDGFB locus was disomic with retained het- had at least one clone with a balanced or unbalanced t(17;22). The erodisomy (Supplementary Fig. 1); in the remaining two tumors only tumor with a balanced t(17;22), from a 7-year-old boy, also this segment was either monosomic (corresponding to the mono- had a subclone with an unbalanced der(22)t(17;22). The youngest somy 22 observed at G-banding) or showed complex copy number patient displaying only ring chromosomes (Case 5) was 19 years at variation (possibly explained by the combination of monosomy 22 diagnosis. In agreement with literature data, also some adult pa- and marker chromosomes at G-banding), respectively. Also Case 10, tients, the oldest being 50 years at diagnosis (Case 19), displayed a for which karyotypic data were missing, showed a profile similar to der(22)t(17;22). most cases with known ring chromosomes. In summary, a balanced t(17;22) is a rare finding in DPFT and SNP array data were also available for the primary tumor and has so far not been observed in patients older than 13 years. The the metastasis from Case 28. The primary tumor was more similar unbalanced der(22)t(17;22), or variants thereof, predominates in to the tumors with der(22)t(17;22) than to those with ring chro- the pediatric/adolescent setting, but is occasionally seen also in mosomes: uniform low-level gain of 17q21-qter and loss of 22q13- adults up to the age of 50 years. Ring chromosomes, on the other qter, respectively. However, there was also relative gain of proximal hand, are found in the vast majority of DPFT from adults, and have 22q. The metastasis showed further structural rearrangements of so far not been observed in patients below the age of 16 years. 17p (deletion of one copy) and 22q (deletion of one copy of 22q11- The age-dependent association between the COL1A1 - PDGFB fu- qter). sion and three different chromosomal variants suggests that the Imbalances affecting other chromosomes than 17 and 22 were fusion arises through different mechanisms in different age groups. mostly in good agreement with aberrations detected at G-banding On the other hand, there are several examples of one cytogenetic analysis. For a detailed view of all observed copy number shifts in variant coexisting with another, suggesting that there could be a the analyzed tumors see Supplementary Table 1. The distribution stepwise evolution from a balanced t(17;22) over der(22)t(17;22) of gained and lost segments on chromosomes 17 and 22 in a subset to one or more ring chromosomes containing parts of chromo- of the tumors is shown in Fig. 2 b and c. somes 17 and 22 [10, 15] . For instance, in the present cohort one DPFT had one clone with a t(17;22) and one with a t(17;22) and a MPS-based analyses der(22)t(17;22), and one case had one clone with a der(22)t(17;22) and one clone with a ring chromosome ( Table 1 ); FISH analy- RNA-seq of ten cases identified in-frame COL1A1 - PDGFB fusion sis verified the presence of the COL1A1 - PDGFB fusion in all these transcripts in all cases (Supplementary Table 2). At the gene ex- clones. pression level, genes mapping within the commonly amplified re- While we, in line with previously reported data [15] , thus in gions on 17q (i.e., from COL1A1 to qter) and 22q (i.e., from 22q11 a few cases could provide direct evidence for an evolutionary re- to PDGFB ) were expressed at higher levels than genes from the lationship between the t(17;22) and the der(22)t(17;22) and be- non-amplified parts of these chromosomes ( Fig. 2 d and e). Of the tween the der(22)t(17;22) and ring chromosome formation, respec- 410 and 499 genes within the amplified regions on chromosomes tively, this does not imply that the der(22)t(17;22) or ring chro- 17 and 22, 267 (65%) and 260 (52%) genes, respectively, showed mosomes in DPFT generally derive from a balanced t(17;22). In- higher expression in the DP compared to other soft tissue tumors. deed, the SNP array results do not support such a clonal evolution. Among the amplified genes showing at least 5-fold increased ex- The most common karyotype in DPFT, i.e., one with two normal pression in DP compared to the cohort of 64 STT were PRKCA, copies of chromosomes 17 and 22, and one or more supernumer- TBX2, MSI2 , and SOX9 on chromosome 17 and PDGFB, PRAME , and ary der(22)t(17;22) or ring chromosomes, could arise through three SOX10 on chromosome 22 (Supplementary Tables 3 and 4). different mechanisms: 1) An exchange between chromosomes 17 The gene panel analysis in eight cases revealed only one poten- and 22 in the G0 or G1 phase of the cell cycle that was preceded tially pathogenic mutation (scored as deleterious by the SIFT algo- by trisomy 17 followed by loss of the derivative chromosome 17; 2) rithm): a c.524T > A (p.Val175Glu) at an allele frequency of 14% in An exchange in G0-G1 between chromosomes 17 and 22 that was the PTEN gene in a local recurrence (Case 31a). WES of the primary followed by loss of the derivative chromosome 17 and duplication tumor and a metastasis from Case 28 revealed 23 non-synonymous of the normal chromosome 17 homologue; or 3) An exchange be- exonic structural variants (ESV) in the primary tumor and 24 in the tween chromatids after DNA replication in S-G2, with the der(17) metastasis; 19 of these were shared ( Fig. 1 b and Supplementary and the der(22) ending up in different daughter cells. Mechanism Table 5). 1 would result in uniparental isodisomy (UPID) for the disomic part of chromosome 17 in one-third of the cases and mechanism Discussion 2 would always give rise to UPID for the disomic part of chromosome 17. Only scenario 3 would consistently result in retained A unique aspect of DPFT is the association between patient age heterodisomy for the disomic parts of chromosomes 17 and 22, and karyotype. As noted before, the COL1A1 - PDGFB fusion is in an which is the allelic pattern that was found in all four DPFT with age-dependent manner associated with either a classical translo- der(22)t(17;22) and all seven DPFT with ring chromosome(s) ana- cation or with one or more supernumerary ring chromosomes [9] . lyzed here ( Fig. 1 c and Supplementary Fig. 1 ). More specifically, none of the nine previously published karyotypes Whereas it thus seems clear that the COL1A1 - PDGFB fusion from children with DPFT included a ring chromosome; instead, one in most DPFT with a der(22)t(17;22) or ring chromosome(s) is (from a child aged 15 months) showed only a balanced transloca- due to an exchange between chromosomes 17 and 22 after DNA tion t(17;22), two (from children aged 8 months and 13 yrs) had synthesis, the origin of the balanced t(17;22) cannot be deduced 40 J. Köster, E. Arbajian and B. Viklund et al. / Cancer Genetics 241 (2020) 34–41 from the currently available data. Although this translocation, five tumors did not differ in any distinct way from the cases diag- like most other balanced neoplasia-associated translocations, most nosed as low-grade DP. Similarly, SNP array analysis did not find likely arises in G0-G1, an origin in S-G2 could also result in the any distinct biomarker among the few high-grade DP cases. segregation of a der(17) and a der(22) to the same daughter cell; It has also been assessed whether the number of fusion copies hence, more cases than the only tumor (Case 3) with a balanced or the presence of mutations in selected cancer-associated genes t(17;22) analyzed so far need to be studied before an S-G2 origin could correlate with sarcomatous transformation, but no strong can be excluded. evidence for such an association has emerged, in line with the The question, then, is why the COL1A1 - PDGFB fusion arises findings in our study [16, 21, 42–44] . One possible exception could through S-G2 errors in the vast majority of DPFT. The timing and be the inactivation of the tumor suppressor CDKN2A , which has origin of translocations and other types of erroneous DNA double- been reported to occur more often, but not exclusively, in high- strand repair is still incompletely understood, but based on cy- grade DP [21, 45] . Among the 15 cases analyzed by SNP array in the togenetic data most neoplasia-associated translocations resulting present study, two of three high-grade DP and one of 12 low-grade in pathogenetic fusions are due to G0-G1 errors [39, 40] . Possibly, DP showed deletion of sequences in 9p21, where CDKN2A resides. the COL1A1 - PDGFB fusion is sufficient for tumorigenesis only in Thus, the significance of 9p deletions in DPFT remains to be eluci- a pediatric setting, whereas most children and all adults require dated. not only the gene fusion but also the extra copies of distal 17q We also had the opportunity to study two separate lesions, a and/or a distorted ratio between genes centromeric and telomeric primary tumor and its metastasis 6 years later, with WES. In line to the PDGFB locus on chromosome 22. Hence, the different types with previously published results on other soft tissue sarcomas of translocation may occur at equal frequencies in all age groups, [38] , neither the mutational load nor the extent of the amplified but only the unbalanced ones have a selective advantage in older regions showed much variation with time or disease progression. patients. Thus, it may well be that tumor progression in DPFT is largely The pathogenetic impact of the copy number changes affecting driven by epigenetic changes. chromosomes 17 and 22 is, however, not obvious. As shown before [18] , genes included in amplified regions are, in general, ex- Declaration of Competing Interest pressed at higher levels than are non-amplified genes on chromosomes 17 and 22. However, the copy number increase is relatively The authors declare that they have no conflicts of interest. moderate (on average 1–3 extra copies in DPFT with ring chromosomes) with usually only minor fluctuations along the amplified segment ( Fig. 2 b and c). Our RNA-seq analysis showed that of 499 Acknowledgments transcripts (long non-coding and micro RNA molecules excluded) located on chromosome 22, 344 were from the amplified region This study was supported by grants from the Swedish Cancer and 155 mapped to the non-amplified, sometimes deleted, region Society, the Swedish Childhood Cancer Foundation, and Govern- telomeric of PDGFB . While 71 (21%) of the genes in the amplified mental Funding of Clinical Research within the National Health region showed increased expression (here defined as a fold change Service. ≥2), the corresponding figure for genes in the non-amplified region was only 7 (5%). For the 410 genes located in the amplified Supplementary material region of chromosome 17, 125 (30%) showed increased expression ( Fig. 2 d and e; Supplementary Tables 3 and 4). Supplementary material associated with this article can be RNA-seq identified a few genes (26 on chromosome 17 and 18 found, in the online version, at doi: 10.1016/j.cancergen.2019.12.001 . on chromosome 22) in the amplified regions with at least 5-fold higher expression compared to the other STT. Many of the am- References plified genes on chromosomes 17 and 22 that were found to be highly expressed in DPFT by Linn and co-workers [18] showed [1] Mentzel T , Pedeutour F , Lazar A , Coindre J-M . In: Fletcher CDM, Bridge JA, high expression also in the present study. For instance, PDGFB Hogendoorn PCW, Mertens F, editors. Dermatofibrosarcoma protuberans. Lyon: was, as expected, one of the top up-regulated genes, showing a WHO Classification of Tumours of Soft Tissue and Bone; 2013. p. 77–9 . [2] Voth H, Landsberg J, Hinz T, Wenzel J, Bieber T, Reinhard G, Höller T, Wendt-

20-fold higher expression in DPFT (Supplementary Table 4). For ner C-M, Schmid-Wendtner M-H. Management of dermatofibrosarcoma pro- genes mapping to the amplified part of chromosome 17, we could tuberans with fibrosarcomatous transformation: an evidence-based review of confirm that the transcription factor-encoding gene TBX2 and the the literature. J Eur Acad Dermatol Venereol 2011;25:1385–91. doi: 10.1111/j. 1468-3083.2011.04141.x . kinase-encoding gene PRKCA were relatively highly expressed (6.4- [3] Simon M-P, Pedeutour F, Sirvent N, Grosgeorge J, Minoletti F, Coindre J-M, and 6.8-fold increase, respectively). Two transcriptionally upregu- Terrier-Lacombe M-J, Mandahl N, Craver R, Blin N, Sozzi G, Turc-Carel C, lated genes on 17q that have not previously been implicated in O’Brien KP, Kedra D, Fransson I, Guilbaud C, Dumanski JP. Deregulation of β the platelet-derived growth factor -chain gene via fusion with collagen gene DPFT were MSI2 (16.1-fold higher expression) and SOX9 (10.5-fold COL1A1 in dermatofibrosarcoma protuberans and giant-cell fibroblastoma. Nat higher expression). From a clinical point of view, a potentially Genet 1997;15:95–8. doi: 10.1038/ng0197-95 . interesting target for the amplified segment on 22q was PRAME [4] O’Brien KP , Seroussi E , Dal Cin P , Sciot R , Mandahl N , Fletcher JA , et al. Var-

(12.2-fold higher expression in DPFT). This gene codes for a cancer- ious regions within the alpha-helical domain of the COL1A1 gene are fused to the second exon of the PDGFB gene in dermatofibrosarcomas and giant-cell testis antigen, and it was recently suggested that overexpressed fibroblastomas. Genes Chromosomes Cancer 1998;23:187–93 . PRAME could constitute a good antigen for immunotherapy in sar- [5] Pedeutour F , Coindre J-M , Nicolo G , Bouchot C , Ayraud N , Turc Carel C . comas [41] . In summary, the gene expression profiling of DPFT Ring chromosomes in dermatofibrosarcoma protuberans contain chromosome 17 sequences: fluorescence in situ hybridization. Cancer Genet Cytogenet shows that the characteristic gain of 17q and 22q material that 1993;67:149 . in most cases accompanies the COL1A1 - PDGFB fusion is associated [6] Pedeutour F , Simon MP , Minoletti F , Sozzi G , Pierotti MA , Hecht F , Turc-Carel C . with a reproducible differential expression of many genes, some of Ring 22 chromosomes in dermatofibrosarcoma protuberans are low-level am- plifiers of chromosome 17 and 22 sequences. Cancer Res 1995;55:2400–3 . which are potential treatment targets. [7] Terrier-Lacombe MJ, Guillou L, Maire G, Terrier P, Vince DR, Somerhausen N, Little is known about the mechanisms behind transformation of de SA, Collin F, de Saint Aubain Somerhausen N, et al. Dermatofibrosarcoma a low-grade into a high-grade DP. None of the previously reported protuberans, giant cell fibroblastoma, and hybrid lesions in children: clinicopathologic comparative analysis of 28 cases with molecular data - A study karyotypes was from a tumor diagnosed as high-grade DP. In the from the french federation of cancer centers sarcoma group. Am J Surg Pathol present study there were cytogenetic data on five such cases. These 2003;27:27–39. doi: 10.1097/00000478-200301000-00004 . J. Köster, E. Arbajian and B. Viklund et al. / Cancer Genetics 241 (2020) 34–41 41

[8] Pedeutour F, Simon MP, Minoletti F, Barcelo G, Terrier-Lacombe MJ, Combe- [28] Örndal C, Mandahl N, Rydholm A, Willén H, Brosjö O, Heim S, Mitelman F. male P, Sozzi G, Ayraud N, Turc-Carel C. Translocation, t(17;22)(q22;q13), in Supernumerary ring chromosomes in five bone and soft tissue tumors of low dermatofibrosarcoma protuberans: a new tumor-associated chromosome rear- or borderline malignancy. Cancer Genet Cytogenet 1992;60:170–5. doi: 10.1016/ rangement. Cytogenet Genome Res 1996;72:171–4. doi: 10.1159/0 0 0134178 . 0165-4608(92)90011-V . [9] Sirvent N, Maire G, Pedeutour F. Genetics of dermatofibrosarcoma protuberans [29] Gisselsson D , Höglund M , O’Brien KP , Dumanski JP , Mertens F , Mandahl N . A family of tumors: from ring chromosomes to tyrosine kinase inhibitor treat- case of dermatofibrosarcoma protuberans with a ring chromosome 5 and a re- ment. Genes Chromosomes Cancer 2003;37:1–19. doi: 10.1002/gcc.10202 . arranged chromosome 22 containing amplified COL1A1 and PDGFB sequences. [10] Macarenco RS, Zamolyi R, Franco MF, Nascimento AG, Abott JJ, Wang X, et al. Cancer Lett 1998;133:129–34 . Genomic gains of COL1A1-PDFGB occur in the histologic evolution of giant [30] Walther C, Domanski HA, Vult von Steyern F, Mandahl N, Mertens F. Chro- cell fibroblastoma into dermatofibrosarcoma protuberans. Genes Chromosomes mosome banding analysis of cells from fine-needle aspiration biopsy samples Cancer 2008;47:260–5. doi: 10.1002/gcc.20530 . from soft tissue and bone tumors: is it clinically meaningful? Cancer Genet [11] Walluks K, Chen Y, Woelfel C, Yang L, Cui T, Seliger C, Geier C, Knösel T, 2011;204:203–6. doi: 10.1016/j.cancergen.2011.01.001 . Hauke S, Petersen I. Molecular and clinicopathological analysis of dermatofi- [31] Jin Y, Möller E, Nord KH, Mandahl N, Vult Von Steyern F, Domanski HA, brosarcoma protuberans. Pathol Res Pract 2013;209:30–5. doi: 10.1016/j.prp. et al. Fusion of the AHRR and NCOA2 genes through a recurrent translocation 2012.10.005 . t(5;8)(p15;q13) in soft tissue angiofibroma results in upregulation of aryl hy- [12] Italiano A, Di Mauro I, Rapp J, Pierron G, Auger N, Alberti L, Chibon F, Es- drocarbon receptor target genes. Genes Chromosomes Cancer 2012;51:510–20. cande F, Voegeli A-C, Ghnassia J-P, Keslair F, Laé M, Ranchère-Vince D, Ter- doi: 10.1002/gcc.21939 . rier P, Baffert S, Coindre J-M, Pedeutour F. Clinical effect of molecular meth- [32] Walther C, Mayrhofer M, Nilsson J, Hofvander J, Jonson T, Mandahl N, Øra I, ods in sarcoma diagnosis (GENSARC): a prospective, multicentre, observational Gisselsson D, Mertens F. Genetic heterogeneity in rhabdomyosarcoma revealed study. Lancet Oncol 2016;17:532–8. doi: 10.1016/S1470-2045(15)00583-5 . by SNP array analysis. Genes Chromosomes Cancer 2016;55:3–15. doi: 10.1002/ [13] Dadone-Montaudié B, Alberti L, Duc A, Delespaul L, Lesluyes T, Pérot G, gcc.22285 . Lançon A, Paindavoine S, Di Mauro I, Blay J-Y, de la Fouchardière A, Chi- [33] Rasmussen M, Sundström M, Göransson Kultima H, Botling J, Micke P, Bir- bon F, Karanian M, MacGrogan G, Kubiniek V, Keslair F, Cardot-Leccia N, Mi- gisson H, et al. Allele-specific copy number analysis of tumor samples with chot A, Perrin V, Zekri Y, Coindre J-M, Tirode F, Pedeutour F, Ranchère-Vince D, aneuploidy and tumor heterogeneity. Genome Biol 2011;12:R108. doi: 10.1186/ Le Loarer F, Pissaloux D. Alternative PDGFD rearrangements in dermatofi- gb- 2011- 12- 10- r108 . brosarcomas protuberans without PDGFB fusions. Mod Pathol 2018;31:1683– [34] Mayrhofer M, Viklund B, Isaksson A. Rawcopy: improved copy number analysis 93. doi: 10.1038/s41379- 018- 0089- 4 . with Affymetrix arrays. Sci Rep 2016;6:36158. doi: 10.1038/srep36158 . [14] Dickson BC, Hornick JL, Fletcher CDM, Demicco EG, Howarth DJ, Swanson D, [35] Hofvander J, Jo VY, Ghanei I, Gisselsson D, Mårtensson E, Mertens F. Com- Zhang L, Sung Y-S, Antonescu CR. Dermatofibrosarcoma protuberans with a prehensive genetic analysis of a paediatric pleomorphic myxoid liposar- novel COL6A3-PDGFD fusion gene and apparent predilection for breast. Genes coma reveals near-haploidization and loss of the RB1 gene. Histopathology Chromosomes Cancer 2018;57:437–45. doi: 10.1002/gcc.22663 . 2016;69:141–7. doi: 10.1111/his.12913 . [15] Mitelman F, Johansson B, Mertens F, editors. Mitelman database of chromo- [36] Arbajian E, Puls F, Antonescu CR, Amary F, Sciot R, Debiec-Rychter M, some aberrations and gene fusions in cancer; 2018 . (Queried August 13, 2018). Sumathi VP, Järås M, Magnusson L, Nilsson J, Hofvander J, Mertens F. In- https://mitelmandatabase.isb-cgc.org/ . depth genetic analysis of sclerosing epithelioid fibrosarcoma reveals recur- [16] Kiuru-Kuhlefelt S, El-Rifai W, Fanburg-Smith J, Kere J, Miettinen M, Knuu- rent genomic alterations and potential treatment targets. Clin Cancer Res tila S. Concomitant DNA copy number amplification at 17q and 22q in der- 2017;23:7426–34. doi: 10.1158/1078- 0432.CCR- 17- 1856 . matofibrosarcoma protuberans. Cytogenet Cell Genet 2001;92:192–5. doi: 10. [37] Hofvander J, Arbajian E, Stenkula KG, Lindkvist-Petersson K, Larsson M, Nils- 1159/0 0 0 056901 . son J, Magnusson L, Vult von Steyern F, Rissler P, Hornick JL, Mertens F. [17] Nishio J , Iwasaki H , Ohjimi Y , Ishiguro M , Isayama T , Naito M , Iwashita A , Frequent low-level mutations of protein kinase D2 in angiolipoma. J Pathol Kikuchi M . Overrepresentation of 17q22-qter and 22q13 in dermatofibrosar- 2017;241:578–82. doi: 10.1002/path.4865 . coma protuberans but not in dermatofibroma: a comparative genomic hy- [38] Hofvander J, Viklund B, Isaksson A, Brosjö O, Vult von Steyern F, Rissler P, bridization study. Cancer Genet Cytogenet 2002;132:102–8 . Mandahl N, Mertens F. Different patterns of clonal evolution among differ- [18] Linn SC, West RB, Pollack JR, Zhu S, Hernandez-Boussard T, Nielsen TO, Ru- ent sarcoma subtypes followed for up to 25 years. Nat Commun 2018;9:3662. bin BP, Patel R, Goldblum JR, Siegmund D, Botstein D, Brown PO, Gilks CB, van doi: 10.1038/s41467- 018- 06098- 0 . de Rijn M. Gene expression patterns and gene copy number changes in der- [39] Weinstock DM, Richardson CA, Elliott B, Jasin M. Modeling oncogenic translo- matofibrosarcoma protuberans. Am J Pathol 2003;163:2383–95. doi: 10.1016/ cations: distinct roles for double-strand break repair pathways in translo- S0 0 02- 9440(10)63593- 6 . cation formation in mammalian cells. DNA Repair (Amst) 2006;5:1065–74. [19] Kaur S, Vauhkonen H, Böhling T, Mertens F, Mandahl N, Knuutila S. Gene doi: 10.1016/j.dnarep.2006.05.028 . copy number changes in dermatofibrosarcoma protuberans –a fine-resolution [40] Biehs R, Steinlage M, Barton O, Juhász S, Künzel J, Spies J, Shibata A, Jeggo PA, study using array comparative genomic hybridization. Cytogenet Genome Res Löbrich M. DNA double-strand break resection occurs during non-homologous 2006;115:283–8. doi: 10.1159/0 0 0 095925 . end joining in G1 but is distinct from resection during homologous recombi- [20] Hong JY, Liu X, Mao M, Li M, Choi DI, Kang SW, et al. Genetic aber- nation. Mol Cell 2017;65:671–84 e5. doi: 10.1016/j.molcel.2016.12.016 . rations in imatinib-resistant dermatofibrosarcoma protuberans revealed by [41] Roszik J, Wang W-L, Livingston JA, Roland CL, Ravi V, Yee C, Hwu P, Futreal A, whole genome sequencing. PLoS ONE 2013;8:e69752. doi: 10.1371/journal.pone. Lazar AJ, Patel SR, Conley AP. Overexpressed Prame is a potential immunother- 0069752 . apy target in sarcoma subtypes. Clin Sarcoma Res 2017;7:11. doi: 10.1186/ [21] Stacchiotti S, Astolfi A, Gronchi A, Fontana A, Pantaleo MA, Negri T, et al. Evo- s13569- 017- 0077- 3 . lution of dermatofibrosarcoma protuberans to DFSP-derived fibrosarcoma: an [42] Wang J, Morimitsu Y, Okamoto S, Hisaoka M, Ishida T, Sheng W, Hashimoto H. event marked by epithelial-mesenchymal transition-like process and 22q loss. COL1A1-PDGFB fusion transcripts in fibrosarcomatous areas of six der- Mol Cancer Res 2016;14:820–9. doi: 10.1158/1541- 7786.MCR- 16- 0068 . matofibrosarcomas protuberans. J Mol Diagn 20 0 0;2:47–52. doi: 10.1016/ [22] Oh E, Jeong HM, Kwon MJ, Ha SY, Park HK, Song J-Y, et al. Unforeseen clonal S1525-1578(10)60614-9 . evolution of tumor cell population in recurrent and metastatic dermatofi- [43] Takahira T , Oda Y , Tamiya S , Yamamoto H , Kawaguchi K , Kobayashi C , Oda S , brosarcoma protuberans. PLoS ONE 2017;12:e0185826. doi: 10.1371/journal. Iwamoto Y , Tsuneyoshi M . Microsatellite instability and p53 mutation associ- pone.0185826 . ated with tumor progression in dermatofibrosarcoma protuberans. Hum Pathol [23] Mandahl N , Heim S , Arheden K , Rydholm A , Willén H , Mitelman F . Three ma- 2004;35:240–5 . jor cytogenetic subgroups can be identified among chromosomally abnormal [44] Abbott JJ, Erickson-Johnson M, Wang X, Nascimento AG, Oliveira AM. Gains solitary lipomas. Hum Genet 1988;79:203–8 . of COL1A1-PDGFB genomic copies occur in fibrosarcomatous transformation of [24] McGowan-Jordan J , Simons A , Schmid M ISCN 2016. ISCN 2016. An international dermatofibrosarcoma protuberans. Mod Pathol 2006;19:1512–18. doi: 10.1038/ system for human cytogenomic nomenclature. Basel: Karger; 2016 . modpathol.3800695 . [25] Mandahl N, Heim S, Arheden K, Rydholm A, Willén H, Mitelman F. Rings, di- [45] Eilers G, Czaplinski JT, Mayeda M, Bahri N, Tao D, Zhu M, Hornick JL, Lin- centrics, and telomeric association in histiocytomas. Cancer Genet Cytogenet deman NI, Sicinska E, Wagner AJ, Fletcher JA, Mariño-Enríquez A. CDKN2A 1988;30:23–33. doi: 10.1016/0165- 4608(88)90089- 1 . /p16 loss implicates CDK4 as a therapeutic target in imatinib-resistant der- [26] Mandahl N, Heim S, Willén H, Rydholm A, Mitelman F. Supernumer- matofibrosarcoma protuberans. Mol Cancer Ther 2015;14:1346–53. doi: 10. ary ring chromosome as the sole cytogenetic abnormality in a dermatofi- 1158/1535- 7163.MCT- 14- 0793 . brosarcoma protuberans. Cancer Genet Cytogenet 1990;49:273–5. doi: 10.1016/ 0165-4608(90)90153-2 . [27] Mandahl N, Limon J, Mertens F, Arheden K, Mitelman F. Ring marker containing 17q and chromosome 22 in a case of dermatofibrosarcoma protuberans. Cancer Genet Cytogenet 1996;89:88–91. doi: 10.1016/0165- 4608(95)00307- X .