Genomic landscape of endometrial polyps

Authors

Siiri Reinikka

Miika Mehine

Pernilla von Nandelstadh

Terhi Ahvenainen

Sara Khamaiseh

Susanna Nousiainen

Vilja Jokinen

Annukka Pasanen

Ralf Bützow

Nanna Sarvilinna

Esa Pitkänen

Pia Vahteristo

Doi

10.1186/s13073-025-01556-z

PMID: 28445112 · DOI: 10.1186/s13073-025-01556-z · Journal: Genome Medicine (2025)

TL;DR

Reinikka et al. performed whole-exome and whole-genome sequencing plus 3′RNA-seq on 23 endometrial polyps (with 18 matched bloods), validated key findings in a 54-polyp FFPE replication set, and report that chromosomal rearrangements affecting HMGA1 and/or HMGA2 occur in 17/23 (74%) of polyps and are the dominant driver event. They also nominate UBE2A as a novel candidate driver gene through highly localized, recurrent codon 6–8 mutations carrying high allelic fractions, and confirm earlier reports of low-allelic-fraction hotspot mutations in canonical endometrial-cancer genes (KRAS, PIK3CA, PIK3R1, PTEN). Source: PMID:28445112.

Cohort & data

  • Discovery cohort: 23 fresh-frozen endometrial polyps (18 nonfunctional, 5 functional) from 23 patients (11 postmenopausal, 12 premenopausal; mean age 52, range 31–78), collected 2016–2019 at Jorvi/Helsinki University Hospital. Matched blood available for 18/23. Endometrial polyps are benign uterine lesions related conceptually but not identical to uterine corpus cancers (UCEC) and to uterine leiomyomas (ULM) discussed as comparators in the paper.
  • Validation cohort: 54 archival FFPE polyps, screened by Sanger sequencing for the UBE2A codon-6 hotspot.
  • Assays: whole-exome sequencing on 22 polyps (Roche NimbleGen SeqCap EZ MedExome, Illumina HiSeq 2500, 101 bp PE, ~170× mean coverage); whole-genome sequencing on 23 polyps + 18 bloods (BGISEQ-500, 150 bp PE, ~81× tumor / ~38× normal); 3′RNA-seq (Lexogen QuantSeq 3′mRNA-Seq Library Prep FWD, NovaSeq 6000) on 25 polyps + 6 normal endometrium; Sanger sequencing for hotspot validation; break-apart FISH for HMGA1 (attempted on 9 rearranged + 3 wild-type FFPE polyps but FFPE quality was suboptimal); immunohistochemistry for HMGA1/HMGA2 protein.
  • Bioinformatics pipeline: GRCh38 alignment with BWA-MEM; somatic SNV/indel calling with Mutect2; structural variants with Delly (filter -q 0 -u 0, then quality ≥300, ≥1 discordant read, SVs >1 kb); somatic CNAs with CNVkit against a pooled normal of 18 bloods; mutational signatures with SigProfiler and MutationalPatterns 3.4.1; driver clustering with OncodriveCLUST; functional annotation with OncoKB annotator, CADD, REVEL, AlphaMissense, Splice-AI, and DynaMut (PDB 6CYO).

Key findings

  • Chromosomal rearrangements involving HMGA1 and/or HMGA2 in 17/23 (74%) polyps, the highest reported frequency to date and substantially higher than prior cytogenetic and targeted-sequencing estimates. One sample (1092_1_S1) carried both. Source: PMID:28445112.
  • Rearrangement architecture: 9/23 (39%) carried a balanced two-way translocation involving HMGA1 or HMGA2; 5/23 (22%) carried a 3-way translocation; 6/23 (26%) showed additional independent chromosomal alterations on top of the HMGA event.
  • Recurrent rearrangement partners (all linked to HMGA1 and/or HMGA2): a ~12 kb region of 7p15.2 containing the candidate enhancer ENSR00001272277 (3 polyps, no nearby genes — argues for enhancer hijacking rather than fusion); LRMDA at 10q22.3 (2 polyps); RAD51B at 14q24.1 (2 polyps); TRAF3IP2 at 6q21 (2 polyps).
  • No intragenic breakpoints in HMGA1 were observed; intragenic breakpoints in HMGA2 were seen in 2/4 HMGA2-rearranged polyps. Authors interpret this as evidence that fusion-gene formation is not the primary mechanism — instead, rearrangement disrupts cis-regulation, consistent with enhancer hijacking.
  • Expression effects of HMGA1/2 rearrangements (3′RNA-seq vs normal endometrium): HMGA1 upregulated in HMGA1-rearranged polyps (p = 0.02, log2FC = 1.06, Student’s t-test); HMGA2 upregulation trended higher but did not reach significance (p = 0.37, log2FC = 1.90); the downstream target PLAG1 was significantly upregulated in both HMGA1-rearranged (p = 3.91 × 10⁻⁶, log2FC = 3.22) and HMGA2-rearranged (p = 0.02, log2FC = 3.38) polyps.
  • Most upregulated gene in HMGA1/2-aberrant polyps: ZMAT3 (q = 6.98 × 10⁻¹³, log2FC = 1.65) — a p53-induced RNA-binding protein previously reported by the same group as the most upregulated gene in uterine leiomyomas regardless of driver. C19ORF38 (q = 3.49 × 10⁻⁴, log2FC = 2.03) was the 13th most upregulated, also flagged by the authors as a leiomyoma HMGA1/2/PLAG1 biomarker.
  • Concordance of HMGA IHC with WGS-called rearrangements: strong stromal HMGA1 staining in 9/12 HMGA1-altered polyps; strong stromal HMGA2 staining in 3/3 HMGA2-only–rearranged polyps. Fisher’s exact test for IHC vs WGS: HMGA1 p = 0.008, HMGA2 p = 0.012. Glandular epithelial cells stained positive in both polyps and normal endometrium, indicating the rearrangement-driven overexpression is anchored in the stromal compartment.
  • UBE2A as novel candidate driver: 3/23 (13%) WGS-cohort polyps carried recurrent codon-6 mutations (mean VAF 0.28, range 0.10–0.38 — much higher than other recurrently mutated genes); 2 additional codon-6 mutations identified in the 54-polyp FFPE validation set (5 total across 77 polyps). Variants observed: c.16C > T p.(Arg6Trp) (4 polyps in this study; also reported in a TCGA endometrial cancer C3N-00386 and in NSCLC, colorectal, and pancreatic samples in cBioPortal) and c.16_18del p.(Arg6del) (1 polyp). All mutations confirmed somatic by Sanger; absent from gnomAD. OncodriveCLUST score 0.83, q = 1.51 × 10⁻⁵.
  • Functional predictions for UBE2A p.(Arg6Trp): CADD 24.4, REVEL 0.56, AlphaMissense 0.96 (likely pathogenic); DynaMut ΔΔG = +1.650 kcal/mol on PDB 6CYO (predicted stabilizing). The c.16_18del p.(Arg6del) sits at a position with phyloP100 = 7.9 (highly conserved). The same arginine stretch (codons 6–8) hosts a germline UBE2A p.(Arg7Trp) variant linked to X-linked intellectual disability (Leiden Open Variation Database #0000408054).
  • Low-VAF cancer-gene mutations recapitulating endometrial cancer drivers: OncoKB-annotated oncogenic / likely-oncogenic mutations in 14 driver genes including hotspots in KRAS, PIK3CA, ERBB2, PIK3R1, PPP2R1A, PTEN, and FBXW7. KRAS was the most frequently mutated (6/23 polyps, 26%); one sample carried two distinct hotspots (G12V and G13V) on different reads at low VAF, consistent with subclonal independent events. Possible biallelic missense + loss-of-function combinations were seen in ARHGAP35 and PIEZO2. KEGG pathway enrichment of recurrently mutated genes flagged “endometrial cancer” as the top hit (FDR-adjusted p = 2.824 × 10⁻⁵).
  • Tumor mutation burden: mean 2,149 mutations per sample (range 290–5,463; ~25 nonsynonymous coding mutations, range 2–64) — comparable to many cancers despite the benign nature of the lesion. No significant TMB difference between functional and nonfunctional polyps (p = 0.06 / p = 0.08).
  • Mutational signatures: SBS1, SBS5, and SBS40 dominant (SigProfiler); MutationalPatterns added SBS8 and SBS89. Signatures were similar in functional vs nonfunctional polyps. SBS1/5/40 typically correlate with age and have been seen in normal endometrium; SBS8 is unusual in benign tissue and may reflect late replication errors.
  • Copy-number landscape: very few CNAs — no high-level amplifications, no homozygous deletions, no aneuploidy — implying milder genomic instability than most cancers. Conversely, 420 chromosomal breakpoints distributed across simple-to-complex SVs were called.
  • Other structural events: retrocopy insertions of MATR3 into chromosomes 1p21.1, 3q26.33, and 12q24.12 in one polyp (germline vs somatic indeterminate due to lack of matched normal).

Genes & alterations

  • HMGA1 — chromosomal rearrangements upstream (12/14), in the 3′UTR (2/14), or downstream (1/14) of the gene in 14/23 polyps; mean rearrangement VAF 0.24 (range 0.06–0.43). Recurrent partners: 7p15.2 enhancer ENSR00001272277, LRMDA (10q22.3), RAD51B (14q24.1), TRAF3IP2 (6q21). One sample also carried a high-VAF (0.34) somatic missense c.268C > G p.(Leu90Val) in the penultimate exon of HMGA1 (Splice-AI weakly predicts donor-site loss, delta 0.25). Result: HMGA1 mRNA and protein overexpression in the stromal compartment.
  • HMGA2 — rearrangements in 4/23 polyps (intragenic in 2); average VAF 0.24 (range 0.14–0.45). Same partner repertoire as HMGA1. mRNA upregulation trended but did not reach significance (p = 0.37); protein overexpression by IHC in 3/3 HMGA2-only–rearranged polyps.
  • UBE2A — recurrent codon-6 hotspot mutations: 4 × p.(Arg6Trp) and 1 × p.(Arg6del) across 77 polyps (3 from WGS discovery, 2 from Sanger validation). Mean VAF 0.28 in discovery cohort — clonally dominant relative to other low-VAF mutations. Predicted likely pathogenic by CADD/REVEL/AlphaMissense; predicted stabilizing by DynaMut. Authors propose gain-of-function. Same codon stretch hosts a germline X-linked intellectual disability allele p.(Arg7Trp).
  • PLAG1 — no DNA-level alterations reported, but transcriptional upregulation (downstream of HMGA1/2) in HMGA-rearranged polyps (log2FC 3.22 in HMGA1-rearranged, 3.38 in HMGA2-rearranged).
  • KRAS — 6/23 (26%) low-VAF hotspot mutations including G12V and G13V; one sample bore both on independent reads.
  • PIK3CA, PIK3R1, PTEN, ERBB2, PPP2R1A, FBXW7 — recurrent low-VAF hotspot or oncogenic mutations annotated by OncoKB annotator. The pattern phenocopies the canonical endometrial-cancer driver landscape.
  • ARHGAP35, PIEZO2 — possible biallelic events (missense + LoF), recurrently mutated.
  • ZMAT3 — top differentially upregulated gene in HMGA-aberrant polyps; not directly mutated.
  • C19ORF38 — 13th most upregulated in HMGA-aberrant polyps; previously reported as a HMGA1/HMGA2/PLAG1-leiomyoma biomarker.
  • LRMDA, RAD51B, TRAF3IP2 — recurrent translocation partners of HMGA1/HMGA2 (intragenic and downstream breakpoints); not classical drivers themselves in this context.
  • JAZF1 — discussed but explicitly excluded as a likely target: 7p15.2 breakpoints in this cohort sit ~2 Mb upstream of JAZF1, arguing against a JAZF1 fusion (the classic endometrial stromal tumor mechanism via t(7;17)(p15;q21)) in endometrial polyps.
  • MED12 — discussed only as a comparator; the dominant driver of uterine leiomyomas (70–80%), but not reported as recurrently mutated in these polyps.
  • MATR3 — retrocopy insertion at three genomic positions in one polyp (somatic vs germline status indeterminate).

Clinical implications

  • Mechanistic insight, not yet a clinical biomarker: the paper reframes endometrial polyps as stromal-driven neoplasms in which non-fusion HMGA1/2 enhancer hijacking is the dominant initiating event. Authors argue this could open avenues for “targeted, non-invasive” treatment in symptomatic polyps (currently treated by hysteroscopic polypectomy), but no specific drug or biomarker is proposed for current clinical use.
  • Risk stratification for malignant progression: endometrial polyps can progress to endometrial cancer (up to ~3%). The presence of OncoKB-annotated low-VAF driver mutations in canonical UCEC genes (KRAS, PIK3CA, PIK3R1, PTEN) supports treating a subset of polyps as precursor lesions. Authors flag — but do not test — the question of whether VAF or specific co-mutations predict transformation risk.
  • Cross-disease shared biology with ULM: HMGA1/2 alterations and ZMAT3 upregulation parallel uterine leiomyoma biology, suggesting any HMGA-targeted therapy (e.g., gene silencing of HMGA2 explored in ovarian carcinoma) could in principle be repurposed across both lesion types.
  • No association with tamoxifen in this cohort: although tamoxifen is a known risk factor for endometrial polyps and KRAS-mutated polyps in prior literature, none of the 23 patients in this study had been on tamoxifen, so the link could not be tested here.

Limitations & open questions

  • Only 23 polyps in the WGS discovery cohort and 54 in the Sanger validation cohort — UBE2A frequency estimates carry wide confidence intervals.
  • FFPE quality precluded formal break-apart FISH cell counting / scoring for HMGA1 rearrangements (Supplementary Fig. S11).
  • Some samples showed strong HMGA1 or HMGA2 staining without a corresponding WGS-detected rearrangement and vice versa — likely a mix of (a) cellular composition (stromal signal masked by glandular HMGA expression), (b) FFPE cautery artifact, and (c) sensitivity limits of short-read SV detection.
  • The functional impact of UBE2A codon-6 mutations is inferred from in silico tools and orthologous yeast/germline evidence — direct functional validation in polyp-relevant models is not done here.
  • One sample’s MATR3 retrocopy events could not be classified as somatic vs germline (no matched normal).
  • 3′RNA-seq clustering did not separate polyps by HMGA1/HMGA2/UBE2A status, possibly because epithelial HMGA expression in normal glands swamps the stromal signal — limits the use of expression as a stand-alone classifier.
  • Open question for cross-paper synthesis: how do the low-VAF KRAS/PIK3CA/PIK3R1/PTEN hotspot patterns in benign endometrial polyps compare quantitatively with the same hotspots seen in normal endometrial glands and in UCEC? Authors note all three contexts share these mutations but do not provide a head-to-head VAF / clonality comparison.
  • The paper’s conclusions about JAZF1 hinge on the precise location of 7p15.2 breakpoints; the small enhancer ENSR00001272277 has no known target gene in this tissue, so the proposed enhancer-hijacking mechanism is plausible but not yet functionally proven.

Citations from this paper used in the wiki

  • “The most common alterations were chromosomal rearrangements affecting HMGA1 and HMGA2, identified in 74% (17/23) of the polyps. These rearrangements involved LRMDA, RAD51B, TRAF3IP2, and 7p15.2 as recurrent rearrangement partners.” (Abstract)
  • “We also identified recurrent low-allelic fraction mutations in well-established cancer genes KRAS, PIK3CA, PIK3R1, and PTEN.” (Abstract)
  • “The only exception was ubiquitin-conjugating enzyme E2 A (UBE2A), which harbored three mutations with an average allelic fraction of 0.28 (range 0.1–0.38).” (Results, p. 5)
  • “Two samples harbored the same missense mutation c.16C > T, p.(Arg6Trp), and the third an in-frame deletion c.16_18del, p.(Arg6del).” (Results, p. 5)
  • “Targeted screening of an independent validation set of 54 FFPE polyp samples revealed two additional polyps with the c.16C > T, p.(Arg6Trp) mutation.” (Results, p. 6)
  • “Analysis of recurrent breakpoint regions revealed simple and complex chromosomal rearrangements involving 6p21.31 (HMGA1) and/or 12q14.3 (HMGA2) in 17 of 23 (74%) polyps.” (Results, p. 6)
  • “We confirmed upregulation of HMGA1 in polyps harboring an HMGA1 rearrangement when compared to normal endometrium (p = 0.02, Student’s t-test, log2FC = 1.06).” (Results, p. 9)
  • “we identified significant upregulation of PLAG1 in endometrial polyps harboring either an HMGA1 (p = 3.91 × 10⁻⁶, Student’s t-test, log2FC = 3.22) or an HMGA2 (p = 0.02, Student’s t-test, log2FC = 3.38) rearrangement when compared to endometrium.” (Results, p. 9)
  • “ZMAT3 was the most significantly upregulated gene in these polyps.” (Results, p. 9–10)
  • “Fisher’s exact test revealed a significant association between the IHC and WGS results (HMGA1, p = 0.008, HMGA2, p = 0.012).” (Results, p. 12)
  • “We have characterized the genomic landscape of endometrial polyps. We show that chromosomal alterations affecting HMGA1 and HMGA2 are a major underlying cause for polyp development. In addition, we present UBE2A as a novel candidate gene for human tumorigenesis.” (Conclusions)

This page was processed by crosslinker on 2026-05-14.