Molecular Classification of Appendiceal Adenocarcinoma

Abstract

Purpose: Appendiceal adenocarcinomas (ACs) are rare, histologically diverse malignancies treated as colorectal cancers despite having distinct biology and clinical behavior. To guide clinical decision making, we defined molecular subtypes of AC associated with patient survival, metastatic burden, and chemotherapy response.

Patients and methods: A comprehensive molecular analysis was performed in patients with AC to define molecular subtypes. Associations between molecular subtype and overall survival, intraoperative peritoneal cancer index, and first-line chemotherapy response were assessed adjusting for histopathologic and clinical variables using multivariable Cox proportional hazards, linear regression, and logistic regression models.

Results: We defined distinct molecular lineages of mucinous appendiceal adenocarcinoma (MAAP) from co-occurring mutations in GNAS, RAS, and TP53. Of 164 MAAP tumors, 24 were RAS-mutant (mut) predominant (RAS-mut/GNAS-wild-type [wt]/TP53-wt) with significantly decreased mutations and chromosomal alterations compared with tumors with GNAS mutations (GNAS-mut predominant) or TP53 mutations (TP53-mut predominant). No patient with RAS-mut predominant subtype metastatic MAAP died of cancer, and overall survival in this subgroup was significantly improved compared with patients with GNAS-mut (P = .05) and TP53-mut (P = .004) predominant subtypes. TP53-mut predominant subtypes were highly aneuploid; increased tumor aneuploidy was independently (P = .001) associated with poor prognosis. The findings retained significance in patients with any metastatic AC. RAS-mut predominant metastases exhibited reduced peritoneal tumor bulk (P = .04) and stromal invasion (P < .001) compared with GNAS-mut or TP53-mut predominant tumors, respectively. Patients with RAS-mut predominant MAAP responded more to first-line chemotherapy (50%) compared with patients with GNAS-mut predominant tumors (6%, P = .03).

Conclusion: AC molecular patterns identify distinct molecular subtypes: a clinically indolent RAS-mut/GNAS-wt/TP53-wt subtype; a chemotherapy-resistant GNAS-mut predominant subtype; and an aggressive, highly aneuploid TP53-mut predominant subtype. Each subtype exhibits conserved clinical behavior irrespective of histopathology.

FIG 1.

Mutational landscape of AC. (A) Overview of mutational type and prevalence in all study tumors (N = 273) assessed through MSK-IMPACT hybrid capture–based sequencing platform. Tumors with sample purity ≤ 10% with no called synonymous or nonsynonymous mutations were not included in the analysis. (B) Prevalence of OncoKB-predicted clinically actionable alterations in AC. Numbers and red shading intensity represent the frequency of the mutation alterations in the respective AC histology. AC, appendiceal adenocarcinoma; CTAAP, colonic-type adenocarcinoma; GCA, goblet cell adenocarcinoma; MAAP, mucinous appendiceal adenocarcinoma.

FIG 2.

Co-occurrence patterns of driver mutations define prognostically relevant molecular subtypes in MAAP associated with varying molecular complexity. (A) Associations between prevalent driver mutations in MAAP tumors (n = 164) were assessed using the Fisher's exact method with Benjamini-Hochberg multiplicity correction and a significant FDR indicated by asterixis (**FDR-corrected P < .01, ***FDR-corrected P < .001). (B) Proportion of molecular subtypes with well, moderately, or poorly differentiated tumors. (C) Number of nonsynonymous mutations, pathogenic mutations in OncoKB-defined oncogenes, and pathogenic mutations in OncoKB-defined tumor suppressors organized by molecular subtype with aneuploidy score and copy-number alteration plot (n = 164). Data are represented as median with interquartile ranges. Aneuploidy scores were calculated as per the Data Supplement. Percent gain (red) or loss (blue) of chromosomal segments are represented. Brackets indicate statistically significant differences (*P < .05, **P < .01, ***P < .001) between compared samples using Wilcoxon-Mann-Whitney testing. Triple-negative subtype statistical comparisons are not shown. (D) OS from time of documented metastasis to death or censoring in patients with metastatic MAAP (n = 149) stratified by molecular subtype. HRs are determined by Cox univariable proportional hazards regression. (E) Multivariable Cox regression model of OS in patients with metastatic MAAP. HRs for death are shown with adjusted P values. Patients who received other chemotherapy types are included in the analysis, but the other chemotherapy covariate is not visually represented (n = 2 patients). Chemotherapy types are summarized in the Data Supplement. Cape, capecitabine; CAPOX, capecitabine and oxaliplatin; CRS, cytoreductive surgery; FDR, false discovery rate; FOLFIRI, FU, leucovorin, and irinotecan; FOLFIRINOX, FU, leucovorin, oxaliplatin, and irinotecan; FOLFOX, FU, leucovorin, and oxaliplatin; FU, fluorouracil; HR, hazard ratio; MAAP, mucinous appendiceal adenocarcinoma; OR, odds ratio; OS, overall survival; ref, reference; WT, wild-type.

FIG 3.

Molecular subtypes are prognostic for overall survival in all patients with metastatic appendiceal adenocarcinoma. (A) Composition of genomic subgroups for each metastatic appendiceal adenocarcinoma histology type. (B) OS of patients with metastatic appendiceal adenocarcinoma stratified by molecular subtype. (C) Multivariable Cox regression analysis of patients with metastatic appendiceal adenocarcinoma. HRs, 95% CIs, and P values were determined from univariable (B) and multivariable (C) Cox proportional hazards regression models. Chemotherapy types are summarized in the Data Supplement. Cape, capecitabine; CAPOX, capecitabine and oxaliplatin; CRS, cytoreductive surgery; CTAAP, colonic-type appendiceal adenocarcinoma; FOLFIRI, FU, leucovorin, and irinotecan; FOLFIRINOX, FU, leucovorin, oxaliplatin, and irinotecan; FOLFOX, FU, leucovorin, and oxaliplatin; FU, fluorouracil; GCA, goblet cell adenocarcinoma; HR, hazard ratio; MAAP, mucinous appendiceal adenocarcinoma; mut, mutant; OS, overall survival; ref, reference.

FIG 4.

RAS-mut predominant mMAAP exhibits decreased stromal destruction and increased response to chemotherapy compared with other subtypes. (A) Tumor PCI was assessed intraoperatively in patients with metastatic MAAP who underwent a cytoreductive surgery (n = 81). Scores are represented as median with interquartile ranges for RAS-mut predominant subtype tumors versus the other molecular subtypes. P value determined by Wilcoxon-Mann-Whitney statistical testing. (B) Multivariable linear regression to determine odds ratios for increased PCI on the basis of various study covariates. (C) Proportion of assessed intraoperative mMAAP samples exhibiting destructive stromal invasion organized by molecular subgroup with a P value representing statistical significance with Fisher's exact testing. (D-G) Representative tumor sections from intraoperative samples of different patients with mMAAP. RAS-mut predominant MAAP cells were typically observed surrounded by peritoneal-adjacent mucin pools (D) or adherent to the surface of tissues without destructive stromal invasion (splenic tissue, E). By contrast, TP53-mut predominant cells exhibited destructive stromal invasion of peritoneal tissue and organs such as the visualized omentum (F and G). Arrows indicate tumor tissue and scaling bars provide magnification orientation. (H) Radiographic best overall disease response to first-line therapy in patients with mMAAP (n = 49). (I) Percent change in CEA or CA19-9 biochemical markers from the beginning of therapy to the time of radiographic best response scan in patients with mMAAP (n = 14). Bars are colored by molecular subgroup. Only patients with at least one historically-elevated tumor marker and markers evaluated pretreatment and periradiographic scan were included. CA19-9, carbohydrate antigen 19-9; CEA, carcinoembryonic antigen; MAAP, mucinous appendiceal adenocarcinoma; mMAAP, metastatic MAAP; mut, mutant; PCI, peritoneal cancer index.

FIG 5.

Clinical outcomes summary for patients with metastatic MAAP organized by molecular subtype. Three-year overall survival was calculated using Kaplan-Meier analysis for all patients with metastatic mucinous appendiceal adenocarcinoma (n = 149). Molecular subtypes are represented with corresponding average tumor aneuploidy score (Data Supplement). The average peritoneal cancer index (scaled dot size) and the percent of tumors per subtype with radiologic tumor regression after chemotherapy (dot color) are represented from each respective study analysis. MAAP, mucinous appendiceal adenocarcinoma.