(P 372) COPY NUMBER GAIN OF MDM2 IN ROD-SHAPED CHROMOSOMES PROVIDES CLUES TO EARLY STAGES IN CIRCULARIZED GENE AMPLIFICATION IN LIPOSARCOMA

Objective: Well-differentiated liposarcoma (WDLS) is the most common subgroup of liposarcoma. The main characteristics are ring and/or giant marker chromosomes (RGMCs) involving amplification of chromosome arm 12q, including MDM2 and, often, HMGA2 and CDK4. RGMCs constitute a form of gene amplification, a common phenomenon in cancer, resulting in overexpression of key oncogenes. The underlying mechanism of the formation and development of RGMCs is poorly understood. It has been suggested that the initial event in this process is chromothripsis, followed by circularization and subsequent rounds of circular recombination.
The objective of this study was to assess the formation and development of RGMCs by investigating lipomatous tumors with low-level gain of 12q as a potential route between lipomas and WDLS. Further, we evaluated the validity of the suggested model involving chromothripsis as the initial event.

Methods: The study included 28 tumor samples from 21 patients with lipomatous tumors. Seven lipomatous tumors with continuous gain (26-110Mb) of 12q in rod-shaped non-RGMC structures were compared to 12 tumor samples with gain/amplification of 12q in ring chromosomes showing at least one uninterrupted contig of copy number gain (>10 Mb). Short-read (shortWGS) and long-read whole genome sequencing (longWGS), genomic arrays (SNP array), and mRNA sequencing (RNA-seq) were performed on selected samples to compare the characteristics of the two groups.

Results: The analysis of DNA from lipomatous tumors with rod-shaped chromosomes showed that the mechanisms behind gain of extra copies of 12q material vary considerably. In the simplest scenario, copy number (CN) gain of 12q was achieved through intrachromsomal duplication, followed by a few additional changes. None of these cases displayed genetic instability, and there was no CN oscillation, strongly arguing against a chromothriptic origin. In contrast, two other rod-shaped cases displayed multiple CN levels and numerous CN transitions, similar to findings in tumors with ring chromosomes. However, CN gained segments in rod-shaped chromosomes were less complex than those in rings, and both rod-shaped and circularized amplicons shared several key features, including lack of interspersed deletions; thus, none of the cases fulfilled the criteria for chromothripsis. ShortWGS and longWGS data showed more frequent internal structural variants (SVs) in CN gained regions of chromosome 12 (Chr12) than in other co-amplified chromosomes, suggesting that gain of Chr12 preceded that of other involved chromosomes.
The data suggest a stepwise process, starting with gain of parts of Chr12 after the S-phase, followed by, possibly circularized, cycles of BFB events. Then, the supernumerary derivative Chr12 formed a dicentric chromosome with supernumerary material from another chromosome. Hence, the amplicons in ring chromosomes in WDLS are formed through a stepwise procedure that is more complex than previously recognized.
Cluster analysis of RNA-seq data revealed no clear separation of tumors with rod-shaped chromosomes and tumors with ring chromosomes. However, a gradual increase in expression from normal fat over conventional lipomas to lipomatous tumors with rod-shaped chromosomes and then to lipomatous tumors with ring chromosomes was observed for CDK4, HMGA2 and MDM2. The gradual overlapping expression levels between the tumor types support the existence of an intermediary tumor type.

Conclusion: The study provides insights into the early steps of RGMC formation and suggests a potential evolutionary link between lipomas and WDLS. The amplicons in WDLS appear to develop through a complex, multistep mechanism, involving shattering of supernumerary DNA sequences, followed by recombination with other chromosomes as rod-shaped structures, and further gene amplification as ring chromosomes.