Fixed-panel tissue RNA prioritization in extraskeletal myxoid chondrosarcoma: CSPG4 evidence across cohorts with comparator and sequencing-year limits

Abstract

Background: Tissue RNA enrichment can prioritize experimental assessment of candidate therapeutic addresses in extraskeletal myxoid chondrosarcoma (EMC), but the answer depends on comparator histology, specimen provenance and assay. We evaluated a fixed 11-gene panel and a separate CHRNA6 context control without treating bulk RNA as evidence of accessible protein or normal-tissue sparing. Methods: We analyzed publicly released gene TPM from a 704-patient soft-tissue-tumor study. Nine primary EMC remained after excluding three explicitly previously reported EMC cases and one recurrence. Primary comparators were myxoid liposarcoma (n=14), low-grade fibromyxoid sarcoma (LGFMS; n=13) and synovial sarcoma (n=18). We calculated per-gene probability of superiority, A=P(EMC>comparator)+0.5P(tie), with equal histology weights, separately for marginal and sequencing-year-matched contrasts. Each matched comparison contained three EMC; their union comprised four patients. A prioritization rule and sensitivity analyses were frozen before target values were inspected. Original GSE24369 arrays provided a separate LGFMS replication anchor (6 EMC biopsies; 17 LGFMS) and three secondary shared histologies. Results: CSPG4 alone met the 11-gene prioritization rule: marginal A=0.895 and matched A=0.811. CSPG4's LGFMS-specific A was 1.000 in GSE24369 and 0.966 marginal/0.933 matched in the RNA-sequencing cohort. Single-EMC and single-histology deletions preserved the positive composite direction. However, removing sequencing-year 2019 reversed the matched composite to 0.433; the marginal DFSP context comparison was also nonpositive (A=0.467). CHRNA6, the separate control, had A=1 in all primary comparisons. Other candidates displayed weak, reversed or comparator-dependent effects. Normal-expression sources and earlier 3SEQ observations did not support normal-tissue restriction. Conclusions: The fixed-panel analysis supports a qualified CSPG4 tissue-validation rationale and identifies consequential negative directions for other candidates. The year-sensitive matched result, normal-expression overlap and unresolved malignant-cell localization preclude a broadly robust EMC-selectivity or therapeutic-validation claim. Code and data: https://raw.githubusercontent.com/trimcrae/Rare-cancers/785cda386aa401c19439fdecd1bfdd5b03f56c0f/research/release-candidates/PUB-SURFACE-TARGETS/2026-09-06/emc-tissue-rna-code-data-supplement.zip Archive files retain their source-specific attribution and reuse terms.

Keywords