Integrated mutational landscape analysis of poorly differentiated high-grade neuroendocrine carcinoma of the uterine cervix

Abstract

High-grade neuroendocrine cervical cancers (NETc) are exceedingly rare, highly aggressive tumors. We analyzed 64 NETc tumor samples by whole-exome sequencing (WES). Human papillomavirus DNA was detected in 65.6% (42/64) of the tumors. Recurrent mutations were identified in PIK3CA, KMT2D/MLL2, K-RAS, ARID1A, NOTCH2, and RPL10. The top mutated genes included RB1, ARID1A, PTEN, KMT2D/MLL2, and WDFY3, a gene not yet implicated in NETc. Somatic CNV analysis identified two copy number gains (3q27.1 and 19q13.12) and five copy number losses (1p36.21/5q31.3/6p22.2/9q21.11/11p15.5). Also, gene fusions affecting the ACLY-CRHR1 and PVT1-MYC genes were identified in one of the eight samples subjected to RNA sequencing. To resolve evolutionary history, multiregion WES in NETc admixed with adenocarcinoma cells was performed (i.e., mixed-NETc). Phylogenetic analysis of mixed-NETc demonstrated that adenocarcinoma and neuroendocrine elements derive from a common precursor with mutations typical of adenocarcinomas. Over one-third (22/64) of NETc demonstrated a mutator phenotype of C > T at CpG consistent with deficiencies in MBD4, a member of the base excision repair (BER) pathway. Mutations in the PI3K/AMPK pathways were identified in 49/64 samples. We used two patient-derived-xenografts (PDX) (i.e., NET19 and NET21) to evaluate the activity of pan-HER (afatinib), PIK3CA (copanlisib), and ATR (elimusertib) inhibitors, alone and in combination. PDXs harboring alterations in the ERBB2/PI3K/AKT/mTOR/ATR pathway were sensitive to afatinib, copanlisib, and elimusertib (P < 0.001 vs. controls). However, combinations of copanlisib/afatinib and copanlisib/elimusertib were significantly more effective in controlling NETc tumor growth. These findings define the genetic landscape of NETc and suggest that a large subset of these highly lethal malignancies might benefit from existing targeted therapies.

Keywords: cancer; genetic analysis; mutations; neuroendocrine; oncogene.

Fig. 1.

Somatic mutation landscape underlying NETc. (A) Distribution of somatic mutation in 64 NETc samples. (B) Frequency and type of somatic mutations. Rows represent genes affected by at least two somatic SNVs or small INDELs, and columns represent individual NETc.

Fig. 2.

Mutational gene burden in NETc. (A) Q-Q plot of significantly mutated genes according to Mutpanning including RB1, ARID1A, KMT2D, PTEN, and WDFY3. (B) Representation of SNVs and INDELs detected in the WDFY3 gene. (C) Frequency and type of PI3K and AMPK somatic mutations in samples harboring WDFY3 gene alterations. Pathway: PI3K/AMPK (deep navy), PI3K (light pink), and AMPK (light navy).

Fig. 3.

Recurrent CNV pattern. Significant amplifications (Left) and deletions (Right). The green line indicates the cutoff for significance (q = 0.25). In the deletion result, the black peak indicates the false positive by visual inspection.

Fig. 4.

(A) Genomic comparison of neuroendocrine and adenocarcinoma cells and evolution paths of two mixed NETc samples. Venn diagrams present the total number of somatic exonic mutations unique to neuroendocrine (NET) vs. epithelial (ADK) cells or shared between the ROM15 and ROM16 matched mixed tumors. (B) Clustering of subclones and clonal evolution inference in the mixed ROM15 and ROM16 NETc. Left: Clusters of variants based on the variant allele fraction (VAF) in neuroendocrine and epithelial cells. Points represent the VAFs of each variant in each cluster. Middle: Node-based clonal evolution trees. Right: Branch-based clonal evolution trees. Branch lengths represent the number of somatic variants in the clusters.

Fig. 5.

Copanlisib, afatinib, and elimusertib alone or in combination inhibited cell proliferation in NET19 and NET21 PDXs. (A) Mice harboring NET19 PDX were treated with vehicle, copanlisib and afatinib or the combination of the two up to 30 d. Tumor volumes are reported as mean ± SEM. Tumor growth was significantly different between control and mice treated with copanlisib, afatinib, or the combination of copanlisib and afatinib (Fig. 5A, P < 0.0001, P < 0.0001, and P < 0.0001, respectively). The combinations of the two inhibitors induced a more durable tumor growth inhibition in the NET19 PDX model when compared to either copanlisib or afatinib alone (Fig. 5A, P = 0.0004 and P = 0.0056, respectively) (B) Mice harboring NET21 PDX were treated with vehicle, copanlisib and elimusertib or the combination of the two up to 30 d. Tumor volumes are reported as mean ± SEM. Tumor growth was significantly different between control and mice treated with copanlisib or elimusertib or the combination of copanlisib and elimusertib (Fig. 5B, P = 0.007, P = 0.0008, and P < 0.0001, respectively). While the combination of the two inhibitors was not able to induce a durable treatment response in this tumor model, it significantly slowed down tumor growth when compared to copanlisib and elimusertib alone (Fig. 5B, P = 0.0004 and P = 0.009, respectively).