The impact of germline alterations in appendiceal adenocarcinoma

Abstract

Purpose:

Over 10% of assessed patients with appendiceal adenocarcinoma (AC) have a pathogenic (P) or likely pathogenic (LP) germline variant, including genes implicated in heritable gastrointestinal cancer syndromes, such as Lynch Syndrome. We defined the clinical and molecular impact of heritable alterations in AC to evaluate the need for dedicated appendiceal screening and prevention strategies in patients with LP/P germline variants.

Experimental Design:

We performed an integrated germline and somatic molecular analysis for patients with confirmed AC. Patients underwent paired tumor-normal sequencing for up to 90 hereditary cancer-risk genes and 505 genes for somatic mutation profiling. We defined the co-occurrence of LP/P germline variants and second-hit pathogenic somatic alterations. The associations between germline variants and patient clinicopathologic features were also evaluated.

Results:

Twenty-five out of 237 patients (10.5%) carried pathogenic or likely pathogenic germline variants in cancer susceptibility genes. Clinicopathologic characteristics and AC-specific survival were similar in patients with or without germline variants. Most (92%, N=23/25) patients with germline variants demonstrated no second-hit somatic alterations, including loss of heterozygosity. Two patients with a germline APC I1307K low-penetrance founder variant exhibited secondary somatic pathogenic alterations in APC. However only one patient tumor exhibited APC-mediated WNT-signaling dysregulation; a plausible consequence of multiple somatic APC mutations with no germline variant contribution. Four patients had germline variants in PMS2 or MSH2 associated with Lynch Syndrome, yet their cancers were microsatellite-stable.

Conclusions:

Germline variants are likely incidental without a contributory driver role in AC. AC screening in patients with germline variants is not clearly merited.

INTRODUCTION

Appendiceal adenocarcinomas (AC) are rare malignancies with diverse histological variation and clinical behavior that encompass less than one percent of gastrointestinal cancers (1–3). The appendix is contiguous with the right colon, yet ACs exhibit a distinct molecular landscape with fewer somatic BRAF, APC, and TP53 alterations and uncommon mismatch-repair (MMR) deficiency (<2%) compared to CRC (5–15%) (4–7).

In patients with gastrointestinal malignancies, identification of a driver germline mutation changes clinical management. CRC is associated with clinically-relevant heritable cancer syndromes caused by bi-allelic tumor suppressor gene inactivation (8,9). Lynch Syndrome (LS) is an autosomal dominant syndrome defined by MMR germline variants that cause tumorigenesis alongside co-occurring, “second hit” somatic mutations or loss of heterozygosity (LOH) in the opposing allele (10–12). Patients with LS-related CRC undergo frequent colonoscopy screening for detection of new malignancies, can be treated with preventative medications including aspirin to impede tumor formation (13), and in contrast to non-LS-associated tumors, demonstrate highly successful outcomes to immune checkpoint blockade (14–17). Germline mutations in APC may lead to familial adenomatous polyposis (FAP) characterized by innumerable gastrointestinal polyps (18). Patients with FAP may undergo prophylactic colectomies at a young age to prevent nearly inevitable CRC (19). Other rare germline alterations and preventative strategies associated with CRC and/or polyposis susceptibility are also described (20).

There are no appendix cancer specific accepted guidelines; the risk factors, screening, prevention, and management strategies for this disease remain to be defined independently from CRC. Pathogenic (P) or likely pathogenic (LP) germline variants have been observed in up to 11.5% of ACs, yet crucially it is unknown if these alterations are consequential drivers to appendiceal adenocarcinoma, or merely incidental, non-contributory variants (21). Definition of the importance of LP/P germline variants for tumorigenesis is crucial to both define the heritable risk factors of AC as well as design meaningful cancer surveillance, and prophylactic strategies in this rare disease. We performed an integrated germline and somatic molecular analysis to determine if germline variants co-occurred with a somatic event sufficient to cause pathogenic gene inactivation in patients with AC.

METHODS

Patient Cohort

We evaluated 237 patients with pathologist-confirmed AC, including mucinous appendiceal adenocarcinoma, goblet-cell adenocarcinoma, and colonic-type adenocarcinoma, treated at Memorial Sloan Kettering Cancer Center from 04/2015 to 09/2022. The study was approved by the institutional review board and was conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained for all patients for internal review board protocol 12–245 for germline and somatic matched hybridization capture-based sequencing with MSK-IMPACT (22). Clinicopathologic characteristics for each patient, including demographics, tumor histological subtype, grade, stage and purity, were confirmed by at least two physicians and study pathologists. For patients with stage IV AC, disease specific survival was determined from time of metastatic diagnosis to appendix cancer-specific death with censoring at last follow-up or non-AC related death.

Germline and Somatic Variant Detection

Germline mutations were identified as previously described using blood-derived DNA and MSK-IMPACT assay confined to up to 90 genes inclusive of all genes associated with cancer predisposition in the American College of Medical Genetics and Genomics guidelines (Supplementary Table 1)(22,23). DNA was extracted using the Qiagen DNAeasy Tissue kit and EZ1 Advanced XL system (Qiagen, Valencia, CA), and sheared with the Covaris E200 instrument (Covaris, Woburn, MA). Illumina HiSeq 2500 sequenced pooled libraries of DNA fragments were optimized to result in uniformly high coverage >500X.

Pathogenic (P) and likely pathogenic (LP) variants were manually reviewed before reporting and were assessed according to variant classification guidelines by the American College of Medical Genetics and Genomics; variants of uncertain significance (VUS) were not included (24). Penetrance levels were determined as high (relative risk, RR>4), moderate (RR=2–4), low (RR<2), recessive, or uncertain based on reported disease risks and accepted criteria (25,26). Patients with LP/P variants were offered comprehensive genomic counseling. IMPACT sequencing mapping quality thresholds are lowered to zero to call variants in highly homologous regions of PMS2. Before variant reporting, all PMS2 variants are confirmed to occur on the PMS2 gene through the use of long range PCR and Sanger Sequencing.

Patient tumors were assessed for somatic mutations, copy-number alterations and fusions using MSK-IMPACT. Over the course of the study the MSK-IMPACT panel was expanded, therefore the DNA custom probes captured targeted sequencing of all exons and selected introns of 341, 410, 468, or 505 genes for 1 (0.4%), 17 (7.2%), 139 (59%), and 80 (34%) of cases, respectively. The FACETs (Fraction and Allele-Specific Copy Number Estimates from Tumor Sequencing) algorithm identified the allele-specificity of somatic alterations (27). Somatic genomic alterations were called using the patient’s matched normal sample in every case. Pathogenicity of somatic alterations was confirmed using the OncoKB dataset.(28)

Molecular Analysis

Mismatch repair (MMR) tumor status was determined using the MSIsensor score (less than 10 defined as proficient) with additional confirmation with immunohistochemistry on available tissue. Tumor mutational burden (TMB) was calculated as the total non-synonymous mutations divided by the respective sample coverage. The fraction of genome altered (FGA) was determined as the percentage of the genome affected by copy number gains and losses. Β-Catenin immunohistochemistry was performed to evaluate for APC functionality.

Statistical Analysis

We utilized Fisher’s exact or Wilcoxon-Mann-Whitney univariable testing with Benjamini Hochberg false-discovery-rate (FDR) adjustment to compare pertinent clinicopathologic characteristics between patients with or without LP/P germline variants. Median overall survival for patients with stage IV AC with and without germline variants was determined using the Kaplan Meier method with the log-rank statistical test. All statistical analyses were performed with R (v4.0.0).

Data availability

Germline variants of the patient cohort are specified in the manuscript in Supplemental Table 2. Data generated in this study are not publicly available because data sharing of identifiable primary germline sequencing files is restricted through the institutional internal review board agreement. However, data and code is available at time of publication upon reasonable request to the corresponding author.

RESULTS

Patient clinical and molecular characteristics

Out of 237 total patients, twenty-five (10.5%) exhibited LP/P germline variants in genes with high (3.8%, N=9/237), moderate (0.4%, N=1/237), low (2.5%, N=6/237), recessive (2.5%, N=6/237), and uncertain (1.7%, N=4/237) penetrance (Figure 1A, Supplementary Table 2). One patient had germline LP/P variants in MUTYH and SDHA.

Figure 1: Germline mutation prevalence in patients with appendiceal adenocarcinoma with co-occurring somatic mutation and copy number status.

A, Germline mutations in the 237 cohort patients organized by predicted mutation penetrance. Percentage in parentheses indicates the percent of patients with the variant out of the total assessed patients. *The specific detected variants are associated with low (APC p.I1307K), recessive (FH c.1431_1433dupAAA (p.K477dup)), and uncertain (CHEK2 p.I157T) penetrance. B, Germline mutation penetrance with somatic and copy number alterations in respective genes. *Indicates a patient with two different germline mutations.

There were no significant differences in most clinicopathologic characteristics in patients with or without germline variants after multiplicity correction (Table 1). Most (84%; N=199/237) of the assessed patients exhibited stage IV disease. Patients with germline mutations were more likely (44%, 11/25) to have a history of additional cancers (FDR-adjusted q=0.019). No AC histology type was disproportionately seen in patients with germline variants. An increased proportion of poorly-differentiated (PD) tumors was seen in patients without germline variants (41%, N=86/212) versus patients with germline variants (20%, N=5/25), although this did not reach statistical significance after multiplicity correction.

Table 1:

Clinicopathologic Characteristics of the Study Cohort

Characteristic	Overall, N = 237	Germline Negative, N = 212	Germline Positive, N = 25	p-value	BH-adjusted q-value
Sex				0.32	0.42
Female	136 (57%)	124 (58%)	12 (48%)
Male	101 (43%)	88 (42%)	13 (52%)
Ancestry				0.023	0.08
Ashkenazi Jewish Ancestry	22 (9.3%)	16 (7.5%)	6 (24%)
No Ashkenazi Jewish Ancestry	167 (70%)	154 (73%)	13 (52%)
Unknown	48 (20%)	42 (20%)	6 (24%)
Race				0.22	0.36
Asian	18 (7.6%)	16 (7.5%)	2 (8.0%)
Black	13 (5.5%)	12 (5.7%)	1 (4.0%)
Hispanic	9 (3.8%)	9 (4.2%)	0 (0%)
Other	4 (1.7%)	2 (0.9%)	2 (8.0%)
Unknown	8 (3.4%)	7 (3.3%)	1 (4.0%)
White and Non-Hispanic	185 (78%)	166 (78%)	19 (76%)
Age at Diagnosis-Median (IQR)	55.0 (46.1, 65.0)	54.8 (46.0, 64.3)	59.7 (50.6, 66.6)	0.1	0.19
Cancer History				0.002	0.019
Additional Cancers	49 (21%)	38 (18%)	11 (44%)
No Additional Cancer	188 (79%)	174 (82%)	14 (56%)
Tumor Differentiation				0.03	0.08
WD	62 (26%)	53 (25%)	9 (36%)
MD	83 (35%)	73 (34%)	10 (40%)
PD	91 (38%)	86 (41%)	5 (20%)
Unknown	1 (0.4%)	0 (0%)	1 (4.0%)
Tumor Histology				0.84	0.84
MAAP	150 (63%)	134 (63%)	16 (64%)
CTAAP	18 (7.6%)	17 (8.0%)	1 (4.0%)
GCA	69 (29%)	61 (29%)	8 (32%)
Tumor Stage				0.61	0.7
I	2 (0.8%)	2 (0.9%)	0 (0%)
II	26 (11%)	22 (10%)	4 (16%)
III	10 (4.2%)	10 (4.7%)	0 (0%)
IV	199 (84%)	178 (84%)	21 (84%)

Statistics were assessed using Fisher’s Exact Testing or Wilcoxon-Mann-Whitney testing for categorical and continuous variables, respectively. Multiplicity testing was performed with the Benjamini-Hochberg (BH) correction. Abbreviations include number (N), interquartile range (IQR), well differentiated (WD), moderately differentiated (MD), poorly differentiated (PD), mucinous appendiceal adenocarcinoma (MAAP), colonic-type appendiceal adenocarcinoma (CTAAP), and goblet cell adenocarcinoma (GCA).

Somatic tumor profiling determined no significant difference in tumor mutational burden (p=0.45), fraction of genome altered (p=0.20), or tumor purity (p=0.54) between tumors with or without germline mutations. Tumor mutational burden and the fraction genome altered was not able to be calculated for six (2.5%) and 3 (1.3%) of patient tumors, respectively. All tumors were MMR-proficient except one MMR-deficient tumor with no germline mutation.

In an exploratory survival analysis of all patients with stage IV AC and evaluable outcomes (Supplemental Table 3), after an overall median follow-up time of 45 months, patients with any LP/P germline variant (n=21) exhibited no significant difference in median AC disease-specific survival (78 months, 95% Confidence Interval 22–Not Reached [NR]) compared to patients without germline variants (n=177; median 108 months, 95% Confidence Interval 55-NR: p=0.87) (Figure S1). Patients with metastatic AC with LP/P heritable variants in CHEK2 (n=3) or BRCA1 (n=1) consensus genes associated with homologous recombination deficiency (HRD) showed similar disease-specific survival to patients with metastatic AC and no heritable germline variant (Figure S2). Patients were treated similarly between the cohorts (Supplemental Table 3).

Incidence and impact of bi-allelic molecular events

Twenty-three out of 25 patients with LP/P germline variants (92%, N=23/25) lacked a second-hit somatic LOH or pathogenic mutation co-occurring in the same gene (Figure 1B). Median gene sequencing coverage was high (564X).

Two patients in the cohort had a low-penetrance APC germline founder mutation (c.3920T>A [p.I1307K]) with co-occurring inactivating APC somatic alterations (Figure 2). Both patients were of Ashkenazi Jewish ancestry and neither exhibited polyposis.

Figure 2: APC germline and somatic alterations in two patients with appendiceal adenocarcinoma.

The APC gene is shown with annotated functional domains and the pathogenic inactivating germline (top) and somatic (bottom) alterations for each patient. For patient 1, two proposed variations are indicated where somatic alterations occur trans (left arrow) or one in syn (right arrow) with the germline APC alteration. Patient 2 exhibited pathogenic loss of heterozygosity (LOH) in the non-germline APC allele. Tumor immunohistochemical staining for β-Catenin reveals nuclear and cytoplasmic (APC loss) versus membranous (APC intact) β-Catenin localization for Patient 1 and 2, respectively. Abbreviations include “chromosome” (Chr.)

Patient 1 had a metastatic colonic-type appendiceal adenocarcinoma with two different pathogenic APC somatic mutations most likely in trans alleles (Figure 2). In patient 1, immunohistochemical (IHC) tumor staining for β-Catenin demonstrated nuclear and cytoplasmic translocation consistent with APC inactivation and aberrant WNT-signaling activation. Patient 2 exhibited a metastatic goblet cell AC with the same APC p.I1307K germline variant and trans-inactivating APC LOH. The patient’s tumor exhibited β-Catenin membranous staining with intact APC functionality.

Four patients had mucinous ACs with pathogenic germline mutations in PMS2 (N=3) and MSH2 (N=1) associated with Lynch Syndrome (LS). Two of these patients exhibited additional malignancies (glioma and prostate cancer; multiple myeloma). All four patients with LS exhibited MMR-proficient ACs on MSIsensor with low tumor mutational burden (median TMB 6.4, range 2.5–8.9). All assessable (3/4) tumors exhibited MMR-proficiency on IHC. No patient was treated with immune checkpoint inhibitor therapy.

DISCUSSION

The rate of germline variants in our cohort of 237 patients with AC was similar (10.5%) to an observational study that examined the prevalence, but not the consequence, of germline variants in 131 patients with AC (21). Germline variants in our cohort were in tumor suppressors that require bi-allelic inactivation to drive tumor progression; a single germline variant, alone, is insufficient for tumor suppressor inactivation. Therefore, in contrast to previous work, we assessed if germline variants co-enable pathogenic gene inactivation alongside second-hit gene copy number alterations or somatic mutations.

In our integrated germline and somatic molecular analysis of the largest germline-assessed cohort of patients with AC, nearly all (92%) patients with germline variants in tumor suppressors exhibited no plausible secondary somatic mutation or LOH event in the same gene necessary for pathogenic inactivation. The clinical characteristics, patient prognosis, and tumor pathological features were not significantly different in patients with or without P/LP germline variants. Altogether, this suggests that germline variants, despite being prevalent in AC, may be incidental to AC tumorigenesis.

The only two patients with co-occurring LP/P germline and somatic alterations exhibited a common, low-penetrance APC p.I1307K founder mutation prevalent in 6–10% of individuals with Ashkenazi Jewish ancestry that is associated with a ~2-fold increased CRC risk, but not FAP risk, in Ashkenazi Jews (29,29–31). Patient 1 exhibited APC inactivation alongside two somatic APC mutations which are rare events in AC (<10% prevalence), suggesting that these uncommon somatic events likely occurred in trans alleles to enable oncogenic signaling independently from the germline variant (5). To support this, patient 2 exhibited the same APC p.I1307K germline variant with trans-inactivating APC LOH, yet their tumor exhibited β-Catenin membranous staining with intact APC functionality.

For the remaining patients, germline variants clearly represent monoallelic, noncontributory events in genes associated with autosomal recessive syndromes, common founder mutations, and/or low penetrance variants. BRCA variants in our study occur proportionally to historical rates in the general population (32,33). Similarly, the MUTYH and CHEK2 variants seen in our cohort are known recurrent northern European founder mutations that are likely incidental (34). The specific CHEK2 c470T>C (p.I157T) variant exhibits uncertain penetrance and may not impact tumorigenesis; the alteration reduces protein autophosphorylation, but overall kinase activity is preserved (35,36,36,37). The SDHA variant is highly-penetrant and associated with increased Krebs cycle metabolites in neuroendocrine and gastrointestinal tumors (38); further study is needed to evaluate this variant’s significance in AC.

Our study has several limitations. Testing referral bias may elevate our cohort’s germline variant prevalence rates. Epigenetic silencing, historically extremely uncommon, was not evaluated as a possible mechanism of biallelic inactivation. Given the low sample size of recurrent LP/P germline variants, disease-specific survival analyses lacked sufficient power to detect a small effect size. Laboratory investigation is necessary to definitively examine causality between all germline variants and tumorigenesis. Further genomic discovery efforts in patients with unique family histories and/or unusual clinical phenotypes may reveal novel candidate genes associated with AC risk.

Overall, our results suggest that testing of patients for heritable germline variants may uncover pathogenic variants, yet upon molecular interrogation these variants do not clearly contribute to AC carcinogenesis. The paucity of contributory germline variants in our cohort suggests that AC, unlike CRC, is not clearly caused by known heritable LP/P variants. On a broader scale, our analysis highlights the vulnerabilities of causal inference in germline variant analysis; the impact of molecular alterations on carcinogenesis, not merely the variant prevalence, must be considered. Dedicated appendiceal cancer screening or prevention strategies in patients with germline variants is not merited given evidence in this large cohort of profiled patients.

TRANSLATIONAL RELEVANCE.

The heritable risk factors for appendiceal adenocarcinoma (AC), a rare gastrointestinal malignancy, remain unknown, complicating screening and prevention strategies. We integrated germline and somatic data in 237 patients with AC to define the driver role of germline variants. Twenty-five (10.5%) patients had pathogenic germline variants. Patients with or without germline variants exhibited similar clinicopathologic features and prognosis. Nearly all (92%) tumors with germline variants exhibited no additional somatic mutation or loss of heterozygosity necessary for pathogenic gene inactivation. In two patients with low-penetrant APC p.I3107K germline variants and plausible co-occurring second-hit alterations, APC was functional in one patient and inactivated in the other; this inactivation was likely due solely to multiple somatic mutations. Patients with Lynch Syndrome-associated germline variants all had mismatch-repair proficient AC. Overall, our results suggest that germline variants do not drive AC, and AC screening and prevention strategies are not clearly merited in patients with heritable alterations.