Alterations in ERBB2 and BRCA and microsatellite instability as new personalized treatment options in small bowel carcinoma

Small intestinal carcinomas account for about 3% of all gastro-intestinal tract tumors [1]. Recent epidemiological data were reported from a Netherland registry showing a 0.7 / 100,000 incidence [2]. Small bowel carcinoma has an annual incidence of about 6 cases per million and is therefore much rarer than colo-rectal carcinomas, which have an incidence of about 420 cases per million inhabitants in the same period (1995–2002) [3‐5].

The most common tumor location is the duodenum, followed by jejunum and ileum. According to a French study men and women are equally effected. The mean age at occurrence is in their middle to late sixties [6].

The main cause of small bowel carcinoma is unknown. Predisposing factors can be chronic inflammatory bowel syndromes, celiac disease or Lynch-syndrome [3]. In general, the prognosis is worse than for colon carcinoma [7].

According to one publication distal tumor location (ileum) is one predictor for poor survival [8]. On the other hand a recent large study reported that duodenum localisation is a negative predictor of survival after resection of SBA [9].

Thus, there is a strong need for a more effective and personalized systemic treatment option in small bowel carcinoma due to limited effects of standard chemotherapy-based treatments in a metastatic setting.

Seven studies in the past (in years: 1997–2014 analyzed between 15 and 89 tumors each) found 5–35% MSI-small bowel carcinomas. More recently Schrock, A et al. [10] described molecular alterations in 317 small bowel carcinomas as well as Hänninen et al. [1] in additional 160 SBACs. Microsatellite-instability (MSI) was found in 7.6% [10] and 14.2% [1], respectively.

The main molecular mutations considering both publications include (up to): TP53 (48.0%), KRAS (53.6%), APC (26.8%), CDKN2A (14.5%), SMAD4 (17.4%), SOX9 (12.0%) and BRAF (9.1, 10% of these BRAF mutated cases in the study of Schrock showed the common p.V600E mutation whereas Hänninen did not find any BRAF V600E mutations) as well as ERBB2 mutations in 8.2%. [1, 10].

Furthermore, Hänninen et al. described novel candidate driver genes like ACVR1B, BRCA2, and SMARCA4. Copy number gains were observed mainly in KRAS (18.9%), BRAF (17.9%) and PIK3CA (15.3%). Nearly 10% of the Finnish patient cohort suffered from celiac disease. This group revealed a higher amount of MSI tumors [1].

Eleven small bowel carcinomas were selected from the registry of the Institute of Pathology of the University Hospital Cologne, Germany. We identified these cases over a time-frame of six years. We considered primary small bowel tumors (no metastasis) with existing paraffin material for further molecular analyses. Ten of these patients had adenocarcinomas, one a mixed neuroendocrine-non neuroendocrine neoplasm (MiNEN) of the small bowel. All samples were routinely formalin-fixed and paraffin embedded (FFPE) according to local practice.

We analyzed formalin-fixed and paraffin-embedded tumor material of eleven patients in total (for patients’ characteristics see Table 1). In about half of our cohort we found potentially treatable genomic alterations in the genes BRCA, ERBB2 and PIK3CA as well as microsatellite instability (see Table 2).

In this study we were able to confirm the results of Hänninen et al., who described for the first time pathogenic and therapeutically relevant BRCA2 mutations in their analyses of 106 SBAC. Additionally, we found two patients with pathogenic BRCA1 mutations, and one of them turned out to be germline related (patient 3). In patient 4 the BRCA1 mutation was a point mutation with a low allele frequency of 5.5% leading to a truncated protein (p.E1540*) and described as pathogenic in the ARUP and ClinVar databases. In a third patient a somatic truncating BRCA2 mutation in exon 11 with an allele frequency of 38.0% was detected (patient 1, p.N986Ifs*5). According the ENIGMA criteria this truncating mutation is likely pathogenic (class 4). The tumor of a fourth patient with known Lynch-syndrome revealed a BRCA1 as well as a BRCA2 mutation both classified as class 3 mutations and therefore probably not therapeutically important. Both mutations are presumably due to the microsatellite-instability-related higher mutational burden.

DNA repair is essential to maintain DNA integrity – BRCA1 as well as BRCA2 deficient cells show a high degree of chromosomal instability, increasing the risk of malignant transformation [15‐18]. Ovarian carcinomas with a somatic BRCA mutation are likely to respond equally well to therapies that include PARP inhibitors as those with germline related BRCA-mutations [9, 16, 19‐23].

In the recent study by Hänninen et al. BRCA mutation were detected for the first time in 7% of 106 patients [1] In our study BRCA mutations were detected with an even higher percentage of 36% (Table 3). However, results could be hampered by the small sample size.

We also confirmed the importance of pathogenic BRCA mutations. In the case of the germline-related BRCA1 mutated MiNEN we could successfully proof the efficacy of a combination of platin-based chemotherapy and the PARP inhibitor olaparib; more than two years after his initial diagnosis of a diffuse metastasized MiNEN (cerebral and different liver metastases) the patient is in a general good condition still without metastases [24].

In addition, first indications are reported that BRCA-mutated ovarian cancers respond well to immuno-checkpoint inhibitors. This is probably due to the higher mutation burden of these tumors compared to BRCA non-mutated ovarian carcinoma [25]. It remains to be shown whether BRCA mutated small bowel adenocarcinoma also benefit from immune-checkpoint inhibition as a second option after PARP inhibition.

ERBB2 is a well-known tyrosine kinase and belongs to the ERBB-family (ERBB1–4). ERBB2 amplification is especially important in breast- and gastric carcinomas and is therapeutically targetable using e.g. the tyrosine-kinase inhibitor trastuzumab. Few publications describe the importance of ERBB2 mutations in small bowel carcinomas [1, 10, 26]. Recent studies by Schrock et al. and Hänninen et al. found comparable results to our study with 8.2 and14% of activating ERBB2 mutations in their patient population [1, 10] In our study we detected in 9% of patient samples an ERBB2 mutation (Table 3). The majority of the ERBB2 mutations clustered into four known hotspots (L755S, was found exclusively in MSI tumors), S310F/Y, R678Q, and V842I). Concurrent hotspot mutations were reported. In concordance with the results mentioned above we could also detect potentially treatment sensitive ERBB2 mutations in our cohort (9%). Our ERBB2 mutated tumor revealed a activating exon 21 mutation (c.2584 A > G p.T862A) which is sensitive to inhibition by neratinib and lapatinib [26].

According to previous studies microsatellite instability (MSI) occurs in a significant number of cases (5–35%). MSI can be germline-related (Lynch-syndrome like in one patient in our cohort) or more frequently somatically induced by an epigenetic silencing of the MLH1 promotor.

Currently, the largest studies by Schrock et al. and Hänninen et al. found high-levels of MSI in SBCAs in 7.6 and 14.1%. We demonstrated a high-level MSI in two out of ten analyzable patients (20%) in our cohort (Table 3). There is growing evidence that microsatellite-instable tumors as well as tumors with a high tumor mutational burden respond well to checkpoint inhibitors and that microsatellite as well as tumor mutational burden status can predict therapy outcome [27].

Results from the keynote studies 158 and 164 confirmed the anti-tumor efficacy of the PD-1-inhibitor pembrolizumab in patients with microsatellite instable colon cancers. Pembrolizumab was approved by the FDA in 2017 for MSI high or mismatch-repair-deficient solid tumors irrespective of tumor origin [28].

The most often altered cancer pathway in SBAC is the PI3K/AKT-pathway. In line with this activating PIK3CA mutations were described in 16% of SBCA by Schrock eal. and were confirmed by us (27%). Tumors driven by an activated PI3K/AKT-pathway might benefit from treatment with a PIK3CA or MEK inhibitor. Currently, clinical trials are ongoing with targeted therapies for PIK3CA mutated tumors of different entities (https://​clinicaltrials.​gov, for example: NCT02389842, NCT02644122). These therapies are based on convincing preclinical and clinical studies [29, 30].

Beyond the mutations described above (ERBB2, BRCA and PIK3CA) the tumors we investigated revealed additional mutations including in the genes KEAP1, KRAS and TP53 (compare Table 3). There is a growing evidence that TP53 mutated tumors harbor a higher mutational burden and a higher chromosomal instability in comparison to TP53 wild type tumors. Until now nothing is known about different treatment responses to e.g. checkpoint inhibition considering the TP53 mutational status of the tumors. TP53 mutations were detected in our cohort in 55% of samples. This is also in concordance with Schrock et al. and Hänninen et al. Here, PIK3CA mutations were detected with 51 and 48% [1, 10] (Table 3).

Limitations of our study include the small number of cases analyzed. Nevertheless, we were able to demonstrate important and rarely described molecular alterations in SBAC (e.g. BRCA mutations) and could confirm findings of much larger studies (e.g. Schrock et al. and Hänninen et al.) in our small collective. We and other detected potentially targetable molecular alterations with similar percentages (e.g. microsatellite-instability or activating ERBB2-mutations) (Table 3) in SBAC. However, the percentage of BRCA mutations in our smaller sample cohort was higher than previously published. In the future, further clinical validations are needed regarding treatment response. Currently, treatment response against specific genomic alterations in SBACs was extrapolited from other tumor entities like colon or gastric carcinoma.

Parallel sequencing on RNA using the Archer FusionPlex CTL panel did not reveal any gene fusions.