Developing Novel Genomic Risk Stratification Models in Soft Tissue and Uterine Leiomyosarcoma

Authors

Dermawan JK

Chiang S

Singer S

Jadeja B

Hensley ML

Tap WD

Movva S

Maki RG

Antonescu CR

Doi

10.1158/1078-0432.CCR-24-0148

PMID: 38488807 · DOI: 10.1158/1078-0432.CCR-24-0148 · Journal: Clinical Cancer Research (2024)

TL;DR

This study developed three-tier genomic risk stratification models for soft tissue leiomyosarcoma (STLMS) and uterine leiomyosarcoma (ULMS) using MSK-IMPACT targeted sequencing data from 195 STLMS and 238 ULMS cases at MSKCC. In STLMS, co-occurrence of RB1 mutation with chr12q deletion or ATRX mutation defined a high-risk group with significantly worse disease-specific survival (DSS), outperforming traditional FNCLCC grading. In ULMS, concurrent TP53 mutation with chr20q amplification or ATRX mutation defined high risk. The STLMS model was externally validated in the AACR GENIE cohort (317 STLMS cases). Longitudinal sequencing confirmed that most driver alterations are early clonal events persisting through disease progression.

Cohort & data

  • STLMS cohort: 195 cases from MSKCC (151 primary at presentation, 44 metastatic), profiled by MSK-IMPACT (341-505 genes), with at least one-year follow-up. Median age 58 years; 64% female; 74% retroperitoneal/intraabdominal/pelvic (PMID:38488807).
  • ULMS cohort: 238 cases from MSKCC (177 primary at presentation, 61 metastatic), profiled by MSK-IMPACT, with at least one-year follow-up. Median age 54 years (PMID:38488807).
  • External validation: 317 STLMS cases from AACR GENIE consortium (128 metastatic at presentation, 40%), using multiple tumor-only targeted panels from different institutions (PMID:38488807).
  • Longitudinal subset: 18 STLMS and 15 ULMS patients with sequential tumors sequenced (PMID:38488807).
  • Dataset: lms_msk_2024 (cBioPortal public)

Key findings

  • STLMS genomic landscape: TP53 mutations (56%), TP53 deletions (12%), RB1 mutations (26%), RB1 deletions (24%), ATRX mutations (10%), PTEN deletions (12%), CDKN2A deletions (2.6%) (PMID:38488807).
  • ULMS genomic landscape: TP53 mutations (55%), TP53 deletions (16%), RB1 mutations (13%), RB1 deletions (39%), ATRX mutations (32%), PTEN deletions (18%), CDKN2A deletions (10%), MED12 mutations (15%) (PMID:38488807).
  • STLMS 3-tier genomic risk for DSS (MSKCC, primary, n=151): high risk (co-occurrence of RB1 mutation + chr12q deletion or ATRX mutation, n=10, median DSS 75 months); intermediate risk (RB1 mutation, ATRX mutation, or del12q, n=60, median DSS 101 months); low risk (none, n=81, median DSS 158 months; global log-rank P=0.04) (PMID:38488807).
  • STLMS entire cohort (n=195): genomic risk stratification remained significant for DSS (global log-rank P=0.00057); high risk: 5-year DSS 43%; intermediate: 64%; low: 79% (PMID:38488807).
  • STLMS external validation (AACR GENIE, n=317): simplified model using RB1 alterations + ATRX mutation was significant for OS (global log-rank P=0.00013); high risk (n=36, median OS 23 months, 5-year OS 10%); intermediate (n=114, median OS 29 months, 5-year OS 21%); low (n=167, median OS 44 months, 5-year OS 43%) (PMID:38488807).
  • ULMS 3-tier genomic risk for DSS (primary, n=177): high risk (concurrent TP53 mutation + ATRX mutation or chr20q amplification, n=50, median DSS 52 months); intermediate (TP53 mutation, ATRX mutation, or amp20q, n=66, median DSS 92 months); low (none, n=61, median DSS 157 months; global log-rank P=0.00012) (PMID:38488807).
  • ULMS entire cohort (n=238): genomic risk for DSS remained significant (global log-rank P=0.00011) (PMID:38488807).
  • Traditional vs genomic risk in STLMS: FNCLCC grading weakly predicted DSS (P=0.047) but tumor size did not (P=0.47). On multivariate Cox regression, only genomic high-risk group remained significant for DSS (PMID:38488807).
  • Traditional vs genomic risk in ULMS: tumor size, mitotic rate, and necrosis were all associated with worse DSS and PFS. On multivariate analysis, only tumor size and genomic high-risk group remained significant for DSS (PMID:38488807).
  • Mutual exclusivity: RB1 mutations and RB1 deletions were mutually exclusive (P<0.01) in both STLMS and ULMS, as were TP53 mutations and TP53 deletions (P<0.01) (PMID:38488807).
  • Co-occurrence patterns: In STLMS, RB1 and TP53 mutations co-occurred (P<0.01). In ULMS, TP53 and ATRX mutations co-occurred (P<0.05), and RB1 deletions with PTEN deletions co-occurred (P<0.01) (PMID:38488807).
  • TMB: Median TMB was 2 mt/Mb (STLMS) and 3 mt/Mb (ULMS), with 99% and 98% of cases <10 mt/Mb respectively (PMID:38488807).
  • Longitudinal sequencing: Most molecular alterations were early clonal events that persisted during progression; a subset of cases acquired RB1 loss or ATRX mutations later (PMID:38488807).

Genes & alterations

  • TP53: Mutations (56% STLMS, 55% ULMS) and deletions (12% STLMS, 16% ULMS); point mutations (not deletions) were prognostically significant in ULMS; truncal clonal event (PMID:38488807).
  • RB1: Mutations (26% STLMS, 13% ULMS) and deletions (24% STLMS, 39% ULMS); inactivating mutations (not deletions) predicted worse DSS in STLMS; key component of STLMS genomic risk model (PMID:38488807).
  • ATRX: Mutations (10% STLMS, 32% ULMS); prognostically significant in both STLMS and ULMS risk models; contributes to high-risk classification when co-occurring with RB1 mutations (STLMS) or TP53 mutations (ULMS) (PMID:38488807).
  • PTEN: Deletions/mutations (12% STLMS, 18% ULMS); PTEN deletions co-occurred with RB1 deletions in ULMS; associated with worse PFS in STLMS (PMID:38488807).
  • CDKN2A: Deletions more common in ULMS (10%) than STLMS (2.6%) (PMID:38488807).
  • MED12: Recurrent mutations in ULMS (15%), infrequent in STLMS (1%) (PMID:38488807).

Clinical implications

  • Genomic risk stratification using clinically available targeted NGS data (RB1, ATRX, chr12q status for STLMS; TP53, ATRX, chr20q status for ULMS) provides superior or comparable prognostic information relative to traditional FNCLCC grading and tumor size (PMID:38488807).
  • IHC for TP53, RB1, and ATRX may serve as surrogate biomarkers for genomic alterations, potentially enabling risk stratification in settings without NGS access (PMID:38488807).
  • The simplified model (RB1 + ATRX alterations only) validated in the AACR GENIE cohort suggests feasibility across different NGS platforms and institutions (PMID:38488807).
  • Risk stratification could inform decisions on neoadjuvant or adjuvant therapy for LMS patients (PMID:38488807).

Limitations & open questions

  • The MSKCC STLMS cohort was biased toward higher-grade tumors referred for NGS, which may explain why tumor size was not prognostic in this cohort (PMID:38488807).
  • The AACR GENIE validation cohort used tumor-only panels from multiple institutions with variable probe coverage, and lacked chromosomal arm-level CNV data, limiting direct comparison (PMID:38488807).
  • OS in the AACR GENIE cohort was calculated from time of sequencing rather than diagnosis, introducing potential lead-time bias (PMID:38488807).
  • Insufficient ULMS cases with follow-up in AACR GENIE precluded external validation of the ULMS risk model (PMID:38488807).
  • Why RB1 point mutations but not deletions are prognostically significant requires further functional investigation; gain-of-function/dominant-negative effects of mutant TP53 may similarly explain why mutations but not deletions predict outcome in ULMS (PMID:38488807).
  • No therapeutic interventions were evaluated; future work needed to determine if risk-stratified patients benefit from different treatment approaches (PMID:38488807).

Citations from this paper used in the wiki

  • “In STLMS, FNCLCC grade but not tumor size predicted progression-free survival (PFS) or disease-specific survival (DSS).” (PMID:38488807)
  • “high risk – co-occurrence of RB1 inactivating mutations (exclusive of RB1 deletion) and chr12q deletion (del12q) or co-occurrence of RB1 mutation and ATRX mutation (n = 10; median DSS: 75 months)” (PMID:38488807)
  • “high risk – co-occurrence of TP53 mutation (exclusive of TP53 deletion) and ATRX mutation or co-occurrence of TP53 mutation and chr20q amplification (amp20q) (n = 50; median DSS: 52 months; 5-year DSS: 41%)” (PMID:38488807)
  • “Longitudinal sequencing showed that most of the molecular alterations appeared to be an early clonal event that persisted during disease progression” (PMID:38488807)
  • “Compared to traditional clinicopathologic models, genomic risk stratification demonstrates superior prediction of clinical outcome in STLMS and is comparable in ULMS.” (PMID:38488807)

This page was processed by crosslinker on 2026-05-04.