An eighty four-year old male visited the outpatient clinic complaining of a painful mass that had been present on his left thigh for 3 years. He was medically healthy and did not have a previous medical or familial history of malignancy. There was no clinical sign of neurofibromatosis. He had twice undergone excisional biopsies (18 and 3 years ago) of the mass at the same location under the presumed diagnosis of lipoma. Preoperative magnetic resonance imaging revealed an approximately 17.5 × 16.4 × 30.2 cm sized, extensive, lobulated heterogeneous mass with T2 high, T1 high signal intensity involving the left thigh. A wide local excision was performed. Gross pathology of the tumor showed a well circumscribed tumor mass with internal multi-lobulated areas (Fig. 1a). Multiple lung and pleural metastases were diagnosed via chest computed tomography (Fig. 1b). Histological findings showed dedifferentiated areas with spindle cells without a lipomatous portion and less dedifferentiated areas consisting of round cells with lipomatous portions, along with infiltrated polymorphonuclear cells (Fig. 1c). Based on these clinical and pathological findings, the mass was diagnosed as DDLPS. The patient died 2 months following surgery.
A DDLPS frozen tissue was obtained from the biobank of Seoul St. Mary Hospital (Seoul, Republic of Korea). Genomic DNA was extracted by microdissection of tumor cell rich area (> 70% of tumor cell purity) and whole blood of the patient using the DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany). WES was performed using the Agilent SureSelect Human All Exome 50 Mb Kit (Agilent Technologies, Santa Clara, CA) and Illumina HiSeq 2500 platform (Illumina, San Diego, CA). Data pre-processing was done using the best practices workflows of The Genome Analysis toolkit (GATK, v4.1.1) (https://software.broadinstitute.org/gatk/) to align the sequence reads with the human reference genome (UCSC hg19) and local realignments with base recalibration, and to identify somatic mutations. The web ANNOVAR package was used to select somatic mutations located in the exonic sequences and to predict their functional consequences [8]. In order to obtain reliable and robust mutation calling, the following variants were eliminated: (i) read depth fewer than 20 in either the tumor or matched constitutional tissues; (ii) polymorphisms listed in the population databases of East Asians with a minor allele frequency 0.1% or more; and (iii) variant allele frequencies less than 5%. Catalogue of Somatic Mutations in Cancer (COSMIC) mutation signatures were obtained via a Mutalisk package [9] using known mutation signatures of soft-tissue sarcoma [2] (signature 1, 2, 5, and 13). To define CNAs, we used the ngCGH module and SNPRank Segmentation statistical algorithm in NEXUS software 9.0 (Biodiscovery, El Segundo, CA). Segments were classified as gains or losses when the log2 ratio was greater than 0.25 or less than − 0.25, respectively. Amplification was defined as a log2 ratio greater than 1.0.
The average sequencing depths for tumor and constitutional DNA were 218X and 223X, respectively (Table S1). A total of 36 non-silent mutations were identified in the exonic area (Table S2), which corresponded to a mutation rate of 0.73 per Mb. The NF1 stop gain mutation (c.7486C > T, p.Arg2496*, variant allele frequency 15.7%) was identified among the cancer-related genes listed in the Cancer Gene Census of COSMIC database (Fig. 2a). This variant was classified as ‘pathogenic’ in ClinVar DB (RCV000218957.1), and not reported in population level variant databases. There is no germline variant classified as ‘pathogenic’ or ‘likely pathogenic’ in ClinVar DB. Thirty-five regions affected by CNAs were found, which correspond to 2% of the genome, including amplifications on 12q13–15, which included known genes associated with DDLPS pathogenesis, such as DDIT3, CDK4, MDM2, YEATS4, and FRS [2, 4,5,6] (Figs. 2b & c; Table S3). In addition to 12q13–15 amplifications, copy gains or amplifications on 1q32, 5p13, 7p22, 8q11–12, 9q31, 12q15–21, 14q12, 19p12–13, and 20q12–13 as well as copy losses on 2p11, 4p11-4q11, 7p21-p14, and 20q13.2 were observed. Several potential cancer-related genes, such as MDM4, ELK4, SLC45A3, RAC1, KLF4, MAFB, TOP1, PLCG1, and PTPRT, were located in the copy gain/amplification regions (Fig. 2c; Table S3). Chromothripsis events were also observed in 7p, 9q, and 20q (Fig. 2c). In a mutational context, C > T mutations were predominant and signature decomposition analysis indicated predominant signatures 1 and 5 (clock-like mutational process) followed by a minor proportion of signatures 2 and 13 (APOBEC-related), which was consistent with the results of a TCGA study on DDLPS [2] (Fig. S2).