Precision oncology for advanced-stage adenocarcinoma of the appendix: comprehensive molecular characterisation identifies actionable lesions and potential predictive biomarkers

Abstract

Objective

Appendiceal adenocarcinoma is a rare cancer with very limited therapeutic options. We aimed to determine whether molecular profiling of advanced appendiceal adenocancer can identify actionable therapeutic alterations.

Methods

We retrospectively analysed cohorts from two large German precision oncology programmes. Patient records and pathology reports from 19 patients with advanced appendiceal adenocarcinoma who were enrolled between 2015 and 2021 were included in this study. We report the molecular features, the resulting molecular tumour board recommendations and their clinical implementation.

Results

In 95% of the tumours, at least one potentially actionable alteration was identified, including mutations in ATM, PIK3CA and AKT1. An elevated tumour mutational burden was identified in 26% of the tumours. A total of 74% of all patients received a molecularly driven treatment recommendation, of which 2 (11%) received the recommended therapy.

Conclusion

Molecular profiling of appendiceal adenocarcinomas revealed potentially actionable alterations in a number of cases.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Appendiceal adenocarcinomas are a heterogenous group of rare cancers without satisfactory treatment options. It is essential to understand whether molecular profiling of these cancers can identify additional therapeutic targets.

WHAT THIS STUDY ADDS

• Potentially actionable genetic alterations can be found in the majority of appendiceal adenocarcinomas.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Patients with appendiceal adenocarcinomas should be evaluated for inclusion in precision oncology programmes and molecular profiling.

Introduction

Adenocarcinomas of the appendix are rare cancers and are part of the spectrum of histologically heterogeneous appendiceal neoplasms. These are detected incidentally in <1% of appendectomy samples.¹ Appendiceal adenocarcinomas can be further subdivided into different histological subtypes, including mucinous adenocarcinomas, signet-ring cell adenocarcinomas, goblet cell adenocarcinomas and adenocarcinomas not otherwise specified.² Given the rarity of the disease and its heterogeneity, there is no standard therapeutic regimen. Surgical resection can cure early-stage disease. Patients with advanced-stage disease are often treated similarly to those with colorectal cancer.³ This decision arises from the similar anatomical location and overlapping genetic profiles (including mutations in KRAS, TP53 and APC). However, various studies have revealed clear molecular differences between appendiceal and colorectal adenocarcinomas.3,7 The lack of standard therapeutic regimens and the lack of druggable lesions for targeted therapies in patients with appendiceal adenocarcinoma are still poorly understood, emphasising the need for further investigations. To identify actionable targets in this rare entity, we analysed molecular profiling data of advanced-stage adenocarcinomas of the appendix enrolled in two large German precision oncology programmes.

Methods

We included data from all patients with histologically-confirmed, advanced-stage adenocarcinoma of the appendix who were enrolled in the Precision Oncology Program at the Comprehensive Cancer Center Munich (CCCM) or the National Center for Tumor Diseases/German Cancer Consortium (NCT/DKTK) Molecular Aided Stratification for Tumor Eradication Research (MASTER)^{8 9} programme between 2015 and 2021 (n=19). Genomic data from the NCT/DKTK MASTER (n=12) were generated via RNA sequencing and either whole-genome sequencing (WGS) or whole-exome sequencing (WES). Genomic data from CCCM were generated via panel-based sequencing (n=7; DNA-based, RNA-based for fusion detection) using the Thermo Fischer Oncomine Comprehensive Assay V.3 (n=2) and Illumina Trusight Oncology 500 assays (n=5), respectively. Clinical and histopathological data were obtained from patient records and pathology reports. Microsatellite instability analysis was performed via immunohistochemistry (IHC) as previously described by Jesinghaus et al⁶ or DNA-based for the CCCM cases and obtained from pathology reports or molecular tumour board reports for the NCT/DKTK MASTER cases. For programmed cell death ligand-1 testing, immunohistochemical analysis was performed via DAKO 22C3, and three scores were reported: the tumour proportion score, combined positivity score and immune cell score, as previously described in a study by Schwamborn et al.¹⁰

Therapeutic targets were identified in two ways: we used the OncoKB¹¹ (www.oncokb.org) database to identify established therapeutic targets among all single nucleotide variants (SNVs) and small insertions/deletions (indels), which we termed listed target lesions (amino acid changes, as listed in OncoKB). Lesions in listed genes with differing amino acid changes were termed potential target lesions. These variants were classified as likely pathogenic or variants of uncertain significance and not submitted via ClinVar (ncbi.nlm.nih.gov/clinvar/) or Varsome (varsome.com). Furthermore, we summarised all therapeutic targets, including SNVs, indels, copy number variants and messenger RNA (mRNA) expression levels listed in the NCT/DKTK MASTER and CCCM molecular tumour board recommendations (potential target lesions).

Tumour mutational burden (TMB) was defined differently between the centres. For WES and WGS (NCT/DKTK MASTER), the absolute number of SNVs per exome/genome was used. The cut-off for high TMB was 100 SNVs per Exome/Genome, which was the cut-off for recommending immune checkpoint inhibitor (ICI) therapy on the NCT/DKTK MASTER molecular tumour board (summarised as elevated TMB). For panel sequencing data (CCCM), mutations per megabase (mut/Mb) with a cut-off of 10 mut/Mb (high TMB) were used.

The study complies with the RECORD (REporting of studies Conducted using Observational Routinely-collected health Data) statement (online supplemental file 1).

Results

Clinicopathological features

19 patients with advanced-stage adenocarcinoma of the appendix who were not amenable to curative treatment (peritoneal metastasis in 76% of patients) were enrolled in the CCCM and the NCT/DKTK MASTER precision oncology programmes between 2015 and 2021 (table 1). The median age at the time of inclusion was 48 years, and the male-to-female ratio was 10:9. Histologically, the cohort consisted of nine mucinous adenocarcinomas, including one with partial neuroendocrine differentiation, three signet-ring cell adenocarcinomas, two goblet cell adenocarcinomas and five adenocarcinomas not otherwise specified. Most patients (13/19, 68%) had received chemotherapy, 8/19 (42%) had undergone surgical resection, 4/19 (21%) had received hyperthermic intraperitoneal chemotherapy and 3/19 (16%) had received radiotherapy.

Table 1. Summary of patient and tumour characteristics and sequencing methods.

	n=19	%
Sex
Male	10	53
Female	9	47
Median age at time of inclusion (range)	48 (30–76)
Histological subtype
Mucinous adenocarcinoma	9	47
Signet-ring cell adenocarcinoma	3	16
Goblet cell adenocarcinoma	2	10
Adenocarcinoma NOS	5	26
Precision oncology programme
NCT/DKTK MASTER	12	63
CCCM	7	37
Sequencing method
Panel	7	37
WGS	7	37
WES	5	21
RNA sequencing	7	37
Received therapy
Chemotherapy	13	68
Surgery	8	42
HIPEC	4	21
Radiotherapy	3	16
Targeted therapy	2	11

CCCM, Comprehensive Cancer Center Munich; HIPEC, hyperthermic intraperitoneal chemotherapy; MASTER, Molecular Aided Stratification for Tumor Eradication Research; NCT/DKTK, National Center for Tumor Diseases/German Cancer Consortium; NOS, not otherwise specified; WES, whole-exome sequencing; WGS, whole-genome sequencing.

Molecular characterisation

WES and WGS identified 31–243 non-synonymous somatic SNVs and 0–11 indels per case. Panel sequencing identified 2–9 potentially pathogenic SNVs and 0–2 potentially pathogenic indels per case. The most frequently mutated genes were KRAS (12/19, 63%), TP53 (11/19, 58%) and GNAS (5/19, 26%) (figure 1). The majority of the KRAS mutations were p.G12D/V/C mutations (10/12, 83%), whereas p.G13D and p.Q61H mutations were detected in one patient each. All GNAS mutations were p.R201H/C mutations (all with concomitant KRAS mutations), of which the majority (4/5, 80%) occurred in mucinous adenocarcinomas. 18 samples were tested for microsatellite instability, and no microsatellite-unstable tumours were found. PD-L1 status was only available for the CCCM cohort. The details are shown in figure 2.

Figure 1. Molecular profiling of appendiceal adenocarcinomas. All recurrently altered genes that were mutated and listed in more than one case are shown. Patients are sorted by histological subtype. Only single nucleotide variants and small insertions/deletions are shown. F, female; M, male; NOS, not otherwise specified.

Figure 2. Biomarkers for treatment with immune checkpoint inhibitors. CCCM, Comprehensive Cancer Center Munich; CPS, combined positivity score; F, female; IC, immune cell infiltrate; IHC, immunohistochemistry; indels, insertions/deletions; M, male; MASTER, Molecular Aided Stratification for Tumor Eradication Research; mut/Mb, mutations per megabase; NCT/DKTK, National Center for Tumor Diseases/German Cancer Consortium; NOS, not otherwise specified; PD-L1, programmed cell death ligand-1; SNVs, single nucleotide variants; TMB, tumour mutational burden; TPS, tumour proportion score; WES, whole-exome sequencing; WGS, whole-genome sequencing.

Actionable alterations and potential predictive biomarkers

Compared with OncoKB, OncoKB identified 14/19 tumours (74%) that harboured at least one listed target lesion. The identified genes were KRAS (12/19, 63%), PIK3CA (p.H1047R in Patient 14 and p.E545K in Patient 23), AKT1 (p.E17K in Patient 1) and ATM (p.R3008H in Patient 4). Additionally, 5/19 tumours (26%) harboured an elevated TMB, of which three cases had a high TMB (243 SNVs in Patient 11, 107 SNVs in Patient 8 and 100 SNVs in Patient 9) and two cases had a TMB just below the specified cut-off (95 SNVs in Patient 15 and 9.4 mut/Mb in Patient 23). The therapeutic options according to OncoKB are shown in table 2. Furthermore, we identified several potential target lesions, including three additional variants in ATM, two of which were classified as likely pathogenic (c.4437–2A>C and c.5006–1G>A in Patient 16) and one further alteration of AKT1 (p.Q79K in Patient 11), which was also classified as likely pathogenic.

Table 2. Summary of definitive potential therapeutic targets and drugs identified in the OncoKB database.

Gene	Alteration		Pathway	OncoKB
				Drugs	Level of evidence	Level-assoc. cancer types
KRAS	WT	(6/19)	MAPK signalling	Regorafenib, panitumumab, cetuximab	1	Colorectal cancer
	G12C	(1/19)	MAPK signalling	AMG-510	3A	Non-small cell lung cancer
	G12V/G12D/G13D/Q61H	(11/19)	MAPK signalling	Cobimetinib, binimetinib, trametinib	4	All solid tumours
	Oncogenic mutations	(12/19)	MAPK signalling	Panitumumab, cetuximab	R1	Colorectal cancer
PIK3CA	H1074R / E545K	(2/19)	PIK3-AKT-mTOR signalling	Fulvestrant+alpelisib	1	Breast cancer
AKT1	E12K	(1/19)	PIK3-AKT-mTOR signalling	AZD5363	3A	Breast cancer, endometrial cancer, ovarian cancer
ATM	R3008H	(1/19)	Homologous recombination	Olaparib	1	Prostate cancer
Elevated tumour mutational burden		(5/19)	Immune checkpoint	Ipilimumab+nivolumab	1	Colorectal cancer
				Pembrolizumab	1	All solid tumours

OncoKB levels of evidence are defined as; level 1: FDA-approved, level 2: standard care, level 3A: clinical evidence, level 4: biological evidence and R1: resistance.

AKT, protein kinase B; FDA, Food and Drug Administration; MAPK, mitogen-activated protein kinase; mTOR, mammalian target of rapamycin; PIK3, phosphoinositide 3-kinases; SNV, single nucleotide variant; WT, wild type.

Next, we summarised the recommendations given in the CCCM and NCT/DKTK MASTER molecular tumour boards that were not included in OncoKB. Frequently altered genes with potential therapeutic options, in addition to those previously mentioned, were GNAS (p.R201H/C, 5/19, 26%), FBXW7 (3/19, 16%) and NBN (3/19, 16%). One patient (Patient 14) with high expression levels of genes related to immune cell infiltration (CTLA4, LAG3, PDCD1) was subjected to RNA sequencing. In case 7, the tumour cell content of the sample was too low to identify therapeutic targets on the basis of WES; only transcriptome sequencing was analysed for targetable alterations. Overall, we identified 18/19 tumours (95%) that harboured at least one potential therapeutic target (figure 3).

Figure 3. Potentially targetable genetic alterations in a cohort of advanced appendiceal adenocarcinomas. Summary of all identified target lesions, including OncoKB searches of single nucleotide variants and small insertions/deletions, as well as tumour board recommendations. Tumour board recommendations include druggable copy number variants (CNVs) and gene expression levels. CA, carcinoma; F, female; HR, homologous recombination; M, male; NOS, not otherwise specified.

Pathway-based analysis of identified therapeutic targets

Next, we assigned the identified therapeutic targets to signalling pathways. Only pathways that were affected in more than one patient were included. Most frequently, tumours harbour target lesions in genes of the mitogen-activated protein kinase pathway (13/19, 68%), the majority of which have KRAS mutations. The PI3K-AKT-mTOR pathway and genes involved in homologous recombination were altered in 7 and 19 patients (37%), respectively. Wingless/Integrated signalling was affected in 4/19 patients (21%) (online supplemental table 1).

Molecular therapy recommendations from the NCT/DKTK MASTER and CCCM molecular tumour boards and the clinical course

Overall, 14/19 patients (74%) received recommendations for targeted therapies. Frequently recommended therapeutic options are MEK inhibitors (MEKis) (9/19, 47%), RAF inhibitors (6/19, 32%), PARP inhibitors (5/19, 26%), mTOR inhibitors/AKT inhibitors (4/19, 21%), EGFR inhibitors/HER inhibitors (2/19, 10%) and WNT inhibitors(2/19, 10%). Hydroxychloroquine was recommended in combination with MEKi in 4/19 patients (21%). 4/19 patients (21%) with mutations in PIK3CA, AKT1 or CTNNB1 or amplification of RAP1B received a recommendation for additional therapy with acetylsalicylic acid on the basis of previously published data by Liao et al.¹² BET inhibitors, CDK4/6 inhibitors, ERK inhibitors, EZH2 inhibitors, FGFR inhibitors, MDM2 inhibitors, PIK3CA inhibitors, SHP2 inhibitors, SOS1 inhibitors and anti-oestrogen therapy were each recommended in one case (5% each).

ICI therapy was recommended for 5/19 patients (26%). Three of these patients had an elevated TMB (patients 8, 9, 11), one had high mRNA expression levels of genes related to immune cell infiltration (Patient 14) and one patient was marginally below the cut-off of 100 SNVs used in the NCT/DKTK MASTER programme (Patient 15; 95 SNVs). In four patients, ICI therapy was recommended in combination with MEK inhibition (based on KRAS mutations or amplification of RAP1B).

Therapy recommendations were implemented in 2/14 patients (14%) who received recommendations. These patients (Patients 8 and 11) were both treated with ICIs but had either progressive disease or deceased disease before follow-up. Both rapid clinical deterioration and denial by the respective healthcare insurances precluded implementation of recommended therapies in most other patients.

Discussion

Advanced appendiceal adenocarcinoma is a rare and poorly studied entity with limited treatment options and a relatively poor prognosis that typically affects relatively young patients. The cohort described in this study was representative of patients with appendiceal adenocarcinoma. The genes most frequently mutated in our cohort were KRAS, TP53 and GNAS. This is in accordance with previously published large-scale studies of appendiceal adenocarcinoma.^{3 13 14} The findings that GNAS, a gene typically associated with mucinous differentiation, was exclusively mutated at codon 201 and comutated with KRAS also correspond to what Ang et al reported in their genomic profiling study of 703 appendiceal cancer specimens.³ Recently, Weitz et al have shown that GNAS mutations in peritoneal mucinous carcinomatosis can be actionable through CDK4/6 blockade.¹⁵ Our pathway-based analysis of identified targets revealed three main affected pathways: the MAPK pathway, the PI3K-AKT-mTOR pathway and homologous recombination genes. We did not investigate predictive biomarkers based on histological subtypes or mutational profiles owing to the limited size of our cohort. Our study, as one of the very few, focused on actionable targets in appendiceal adenocarcinoma. The primary limitations of our study are the retrospective design, small cohort size of a rare cancer, heterogeneous sequencing platforms, incomplete PD-L1/IHC data and short clinical follow-up, limiting generalisability.

Our comprehensive molecular analyses of advanced-stage appendiceal adenocarcinoma revealed diverse druggable targets and potential predictive biomarkers. Most patients in our cohort presented potential molecular target lesions and received treatment recommendations for targeted therapies on the molecular tumour boards of two large German precision oncology programmes. There is a relationship between the identified targets in our study and the recommendations of the tumour board. Our analysis revealed more listed and potential targets than patients received actual therapy recommendations, while very few recommendations were clinically implemented.

With respect to the former, most of the differences between identified targets and tumour board recommendations can be attributed to the retrospective design of our study and, therefore, changes in publicly available data, such as study results that support or contradict therapy recommendations based on molecular targets.

With respect to the implementation of therapy recommendations, in landmark studies of precision oncology programmes, the implementation rate typically ranges between 10% and 30%, which is determined mainly by the year of recruitment.^{8 16 17}

Overall, this cohort of young (ranging from 30 to 76) and fit patients (Eastern Cooperative Oncology Group performance status scale ≤1), who have been extensively pretreated, is typical for inclusion in a precision oncology programme. Whether these findings can be extended to all patients with appendiceal carcinomas cannot currently be answered. Notably, previous studies of the NCT/DKTK MASTER programme identified previously unknown targets in rare pancreatic cancer subtypes, resulting in successful therapy implementation in two patients.¹⁸

In addition to targeted therapies, the relevance of ICIs has not been systematically addressed for appendiceal adenocarcinoma thus far. In recent years, different rationales for the application of ICIs have been identified, such as high TMB, elevated mRNA expression of genes signalling immune cell infiltration, microsatellite instability or high PD-L1 expression.¹⁹ Several studies have shown increased response rates to programmed cell death protein 1 (PD-1) inhibitors, to combined PD-1/cytotoxic T-lymphocyte-associated protein 4 (CTLA4) inhibitors in patients with non-small cell lung cancer and to CTLA4 inhibitors in patients with melanoma when their TMB is high.20,22 In 2020, the results from KEYNOTE-158²³ led to the approval of pembrolizumab for TMB-high solid tumours. Bendell et al reported that a combination of the MEK inhibitor cobimetinib with the anti-PD-L1 antibody atezolizumab showed efficacy in microsatellite-stable colorectal cancer with activating KRAS mutations,²⁴ which is a molecular profile that three of our patients with a rationale for ICI therapy shared; however, a later randomised phase 3 study showed no benefit over regorafenib.²⁵ In two studies of patients with melanoma receiving ipilimumab or tremelimumab, a mutational load of more than 100 non-synonymous somatic mutations identified by WES was associated with clinical benefit.^{22 26} However, there is no universal cut-off for the definition of high TMB. Samstein et al analysed a cohort of 1662 patients with advanced tumours who had received ICI therapy and reported that the TMB cut-off points for predicting clinical benefit vary across cancer types.²⁷ Although ICIs seem to be a promising approach for treating advanced-stage appendiceal adenocarcinoma, no systematic data are available. Our very limited data from two patients revealed an unfavourable therapeutic response. However, further studies are needed to investigate this approach in patients with appendiceal carcinoma.

In this study, most patients with advanced appendiceal adenocarcinoma harboured potentially druggable lesions. However, more data are needed to investigate whether this should inform treatment decisions.

Data availability statement

Data may be obtained from a third party and are not publicly available.