A specific missense mutation in GTF2I occurs at high frequency in thymic epithelial tumors

Authors

Iacopo Petrini

Paul S Meltzer

In-Kyu Kim

Marco Lucchi

Kang-Seo Park

Gabriella Fontanini

James Gao

Paolo A Zucali

Fiorella Calabrese

Adolfo Favaretto

Federico Rea

Jaime Rodriguez-Canales

Robert L Walker

Marbin Pineda

Yuelin J Zhu

Christopher Lau

Keith J Killian

Sven Bilke

Donna Voeller

Sivanesan Dakshanamurthy

Yisong Wang

Giuseppe Giaccone

Doi

10.1038/ng.3016

PMID: 24974848 · DOI: 10.1038/ng.3016 · Journal: Nature Genetics (2014)

TL;DR

Petrini and colleagues performed whole-exome sequencing on 28 thymic epithelial tumors (TETs) and identified a highly recurrent single-nucleotide missense mutation in GTF2I (chromosome 7 c.74146970T>A, encoding p.Leu404His in the β isoform and p.Leu383His in the δ isoform of TFII-I). In an extended cohort of 274 TETs, the GTF2I mutation was present in 82% of WHO type A and 74% of type AB thymomas — the most indolent histotypes — but only 8% of thymic carcinomas. GTF2I mutation correlated with significantly better disease-related survival (96% vs 70% 10-year survival; log-rank P < 0.001). Aggressive histotypes were enriched for mutations in known cancer genes (TP53, CYLD, CDKN2A, BAP1, PBRM1) and a higher overall mutation burden (mean 43.5 vs 18.4 mutations per sample).

Cohort & data

  • Cancer types: thymic epithelial tumors (TET), including thymoma (THYM) WHO subtypes A, AB, B1, B2, B3 and thymic carcinoma (THYC).
  • Dataset: tet_nci_2014 — 286 patients enrolled from the National Cancer Institute (Bethesda), Pisa University Hospital, Padua University Hospital, and IRCCS Istituto Clinico Humanitas (Rozzano).
  • Sample sizes by assay:
    • Whole-exome sequencing: 28 TETs (tumor/normal pairs).
    • Custom 197-gene targeted panel (high-depth MiSeq): 52 TETs (26 overlapping with WES).
    • Array CGH: 65 cases.
    • Transcriptome RNA-seq: 25 tumors (Illumina Genome Analyzer II / HiSeq2000).
    • GTF2I deep sequencing: 250 samples.
    • Sanger sequencing of GTF2I: 199 TETs with >50% cancer cells.
    • Total GTF2I prevalence cohort: 274 TETs (270 tumors + 4 cell lines: Ty82, T1682, T1889, IU-TAB1).
  • Survival data available for 204 patients (median follow-up 39.4 months, 95% CI 30.3–48.5).
  • Assays / methods: whole-exome sequencing, RNA-seq, array-CGH (Agilent), Sanger sequencing, GTF2I targeted deep sequencing on MiSeq. Bioinformatic pipeline used Novoalign + GATK + VarScan2 + SnpEff + Annovar for exome calls, TopHat + Cufflinks for RNA-seq, FusionMap and DeFuse for fusion detection, and GISTIC for copy-number peak calling. Data were deposited in GEO under accessions GSE55852 (aCGH) and GSE57892 (NGS).
  • Reference genome: hg19 / NCBI build 37.1.

Key findings

  • A single recurrent T>A missense mutation at chromosome 7 c.74146970 in GTF2I was identified by exome sequencing in TETs; the mutation was absent from dbSNP137, ESP5400, and COSMIC at the time of publication (PMID:24974848).
  • Histotype-specific prevalence (n=270 tumors): GTF2I mutation in 82% of type A, 74% of type AB, 78% combined; frequency decreased progressively in more aggressive histotypes to 8% in thymic carcinomas (3/36). Overall prevalence was 43.4% (119/270).
  • Stage association: GTF2I mutations were more frequent in early-stage disease (I–II, 57%) than advanced disease (III–IV, 19%; χ² P < 0.001). In binomial logistic regression including WHO histotype, stage, and completeness of resection, only histotype significantly predicted GTF2I status (P < 0.001).
  • Survival: Patients bearing GTF2I-mutant tumors had 96% 10-year disease-related survival vs 70% for wild-type (log-rank P < 0.001). Among thymoma patients alone the difference was 96% vs 88% (P = 0.057). All three thymic carcinoma patients with GTF2I mutation were alive at median 27.6 months. In multivariate Cox models combining stage, WHO classification and GTF2I, only stage remained an independent prognostic factor.
  • Mutation burden: Thymic carcinomas carried significantly more somatic mutations than thymomas (mean 43.5 vs 18.4 per sample; Mann-Whitney U P = 0.001). Across 28 WES tumors, 722 coding SNVs and 68 indels were identified (mean 28/sample; range 3–94).
  • Recurrent cancer-gene mutations in thymic carcinomas: TP53, CYLD, CDKN2A, BAP1, and PBRM1 were recurrently mutated.
  • Copy-number aberrations were enriched in aggressive histotypes (thymic carcinoma, B3, B2) and rare in types A/AB. Frequent arm-level losses: 6p (26%), 6q (29%), 3p (22%), 13q (18%); frequent gains: 1q (55%), 7p (20%), 7q (15%), 20p (17%). Focal amplification of the BCL2 locus correlated with increased BCL2 transcript expression.
  • GTF2I isoform expression: RNA-seq showed both wild-type and mutant alleles expressed (mean mutant allele fraction 47%, range 44–49%). The β (NM_033000.2) and δ (NM_001518.3) isoforms were significantly more abundant than the other three known isoforms (Kruskal-Wallis P < 0.0001).
  • Functional validation: Stable expression of mutant β/δ GTF2I isoforms in NIH-3T3 cells accelerated proliferation more than wild-type counterparts, but did not produce soft-agar colony formation — consistent with a growth-promoting, non-transforming oncogene. Mutant TFII-I protein was more abundant than wild-type at equal mRNA levels and degraded more slowly under cycloheximide, supporting post-transcriptional stabilization via a noncanonical RXXL destruction box (RILLAKE>RILHAKE). FOS mRNA was elevated in GTF2I-mutant tumors (consistent with TFII-I binding the FOS promoter).
  • Fusion transcripts: Identified in 7 of 25 tumors evaluated (1–16 fusions per case; mean 1). One B2 thymoma carried 16 fusions. The TY82 thymic carcinoma cell line confirmed the known BRD4–NUTM1 (BRD4-NUT) fusion. No fusion transcript involved GTF2I.

Genes & alterations

  • GTF2I — recurrent missense mutation chr7:74146970 T>A → p.Leu404His (β isoform) / p.Leu383His (δ isoform). Highly enriched in indolent types A/AB thymomas (78% combined) and a favorable prognostic biomarker. Predicted deleterious by SIFT/PolyPhen-2; alters a noncanonical RXXL destruction box leading to post-transcriptional TFII-I stabilization.
  • TP53 — recurrent somatic mutations in thymic carcinomas.
  • CYLD — recurrent mutations in ~19% of thymic carcinomas (cited from prior literature; observed in this cohort).
  • CDKN2A — recurrent mutations in thymic carcinomas.
  • BAP1 — recurrent mutations in thymic carcinomas.
  • PBRM1 — recurrent mutations in thymic carcinomas.
  • BCL2 — focal amplification (GISTIC peak); correlated with increased mRNA expression by RNA-seq.
  • KIT — known to be mutated in ~10% of thymic carcinomas (literature context; the three GTF2I-mutant thymic carcinomas in this study were all positive for KIT by immunohistochemistry, including two squamous cell and one undifferentiated carcinoma).
  • BRD4–NUTM1 — BRD4-NUT fusion present in the TY82 thymic carcinoma cell line.
  • FOS — elevated mRNA expression in GTF2I-mutant tumors, consistent with TFII-I-driven transcriptional regulation.

Clinical implications

  • Prognostic biomarker: GTF2I mutation status stratifies TET patients into a favorable-prognosis subgroup. Only 2/83 GTF2I-mutant patients died of tumor progression vs 26/121 (21%) wild-type (P < 0.0001). The authors propose GTF2I mutation testing as a complement to WHO histological classification, which suffers from interobserver variability.
  • Diagnostic adjunct: Because GTF2I mutation occurs at the same nucleotide (chr7:74146970 T>A) in all mutant cases, a single targeted assay can identify the favorable subgroup.
  • Molecular dichotomy: The data support treating thymomas and thymic carcinomas as molecularly distinct diseases — thymic carcinomas resemble aggressive epithelial cancers with TP53/CYLD/CDKN2A/BAP1/PBRM1 mutations and high copy-number burden, while indolent A/AB thymomas are dominated by the GTF2I hotspot.
  • Targeted therapy implications: The authors note KIT mutations in ~10% of thymic carcinomas (literature) as a potential targetable lesion; this study does not directly evaluate targeted therapy response. The GTF2I mutation itself acts as a cell-growth–promoting (not strongly transforming) oncogene and is not directly druggable as characterized here.

Limitations & open questions

  • The functional characterization relied on NIH-3T3 fibroblasts, an immortalized line that may not recapitulate thymic epithelial biology. The authors explicitly call out this limitation.
  • Only three thymic carcinomas in the cohort carried GTF2I mutation — too few to robustly characterize the mutation’s behavior in aggressive disease.
  • The authors propose two non-mutually-exclusive models for the inverse correlation of GTF2I mutation with aggressiveness — (1) GTF2I mutation marks a distinct indolent lineage, or (2) GTF2I mutation is a founder event lost via clonal evolution in aggressive tumors — but cannot distinguish between them without multiregion sampling.
  • Findings need confirmation in an independent, well-annotated TET cohort.
  • The molecular mechanism by which p.Leu404His stabilizes TFII-I (beyond the proposed RXXL destruction-box disruption) remains incompletely understood; alternative mechanisms warrant further study.
  • The single-cohort study does not address treatment-response correlates for GTF2I status.

Citations from this paper used in the wiki

  • “We observed a GTF2I mutation (chromosome 7 c.74146970T>A) in 119 of the 270 TETs evaluated (43.4%), which was present most commonly in type A (82%) and AB (74%) thymomas (78% overall).”
  • “Patients with tumors bearing GTF2I mutations had a better prognosis than those bearing wild-type GTF2I (96% compared to 70% 10-year survival, respectively; log-rank P < 0.001).”
  • “Thymic carcinomas had a significantly higher number of mutations than thymomas (Mann-Whitney U P = 0.001)” (mean 43.5 vs 18.4 per sample).
  • “We identified recurrent mutations of known cancer genes, including TP53, CYLD, CDKN2A, BAP1 and PBRM1, in thymic carcinomas.”
  • “Focal amplification of the BCL2 locus, which correlated with increased expression of BCL2 transcripts according to RNA-seq data.”
  • “We identified fusion transcripts in 7 of the 25 tumors evaluated, including the TY82 thymic carcinoma cell lines known to carry the BRD4-NUT fusion.”
  • “Both the β and δ mutant isoforms increased cell proliferation more than their wild-type counterparts” in NIH-3T3 assays.

This page was processed by crosslinker on 2026-05-11.