A 74-yo caucasian man presented in 2020, with a 40-year history of skin lesions (including carcinomas, adenomas, basaliomas and epitheliomas) and multiple visceral cancers (Fig. 1). A large germline pathogenic deletion involving exons 1–6 of the MSH2 gene had been detected in DNA extracted from peripheral blood mononuclear cells and a diagnosis of MTS had been made back in 2002. (14, 15) Since then, the patient underwent intensive surveillance with annual colonoscopy, urine cytology and dermatologic examination. Upper gastrointestinal endoscopy was also performed, every 2 years.
After a history of multiple malignancies undergoing appropriate surgical treatment (onset at age 46), the patient was diagnosed with adenocarcinoma of the duodenum at 73 years. He underwent pancreaticoduodenectomy; immunostaining confirmed the loss of MSH2 and MSH6 and additionally showed Programmed death-ligand 1 (PD-L1) expression (TPS 1%, CPS 3). No adjuvant chemotherapy was given because of slow surgical recovery. Eight months later a metabolically active and dimensionally growing relapse in paracaval and inter-portocaval lymph nodes was identified by CT and PET/CT scans, together with a suspected primary lesion of the right kidney. A right nephroureterectomy with paracaval lymphadenectomy was performed. Histological examination of the renal tissue was compatible with metastasis from adenocarcinoma of gastrointestinal origin. Inter-portocaval lymph nodes could not be resected because of vascular infiltration.
Due to persistent, metabolically active nodal disease, anti-PD1 treatment with pembrolizumab was started. CT and PET/CT scans showed progressive reduction in the volume and uptake of inter-portocaval lymph nodes, until their complete disappearance after 1 year of treatment (metabolic CR). (Fig. 2, intermediate panel)
Furthermore, complete regression of a 2-cm, ulcerated cutaneous epithelioma in the lumbar region was observed over the first two months of pembrolizumab treatment. (Fig. 2, upper panel)
During the first year of ICB treatment, a progressively growing, hypermetabolic gastric lesion appeared and repeated endoscopic sampling documented adenoma with varying degrees of dysplasia. (Fig. 2, lower panel) Upon further progression and endoscopic assessment, a diagnosis of gastric adenocarcinoma was finally made and total gastrectomy, with D2 lymphadenectomy was performed; histologic examination showed a double localization of mixed gastric adenocarcinoma (tubulo-papillary and signet ring cell 10–90%; pT1b and pT3) with 7/25 positive lymph nodes. Immunostaining was negative for HER2 and PD-L1 expression (CPS < 1) and showed loss of MSH2 and MSH6 expression. Loss of MSH2 in MMRd tumors predictably results in loss of MSH6 protein due to failure to form the MSH2-MSH6 complex, in which MSH2 is a dominant partner. (5) Pembrolizumab was then restarted after multidisciplinary discussion, considering the ongoing CR on retroperitoneal nodes and cutaneous lesion, good treatment tolerance and no recurrence of oncologic disease or new neoplasia detected at regular follow-up.
We therefore performed next generation sequencing (NGS) analysis on tumor tissues obtained from pancreaticoduodenectomy, right nephrectomy (metastasis from the previous duodenal adenocarcinoma) and total gastrectomy, using the SureSelectXT HS CD Cancer Core assay (model 5191–6730, Agilent, Wokingham, UK) targeted panel was performed. Specimens from duodenal and renal lesions demonstrated a substantial overlap in terms of pathogenic/likely pathogenic variants (18 shared variants out of 22 and 18 total variants respectively), while the gastric adenocarcinoma specimen displayed much less overlap. Indeed, only 2 of 18 total pathogenic/likely pathogenic variants (TGFBR2 p.P129fs*3 and RNF43 p.G659fs*41) were shared with the duodenal primary and its renal metastasis; all 3 specimens featured MSI and high tumor mutational burden (TMB) (> 30 mut/Mb), and the previously described germline deletion of exons 1–6 of MSH2 was present in all three samples. The second MSH2 allele was affected by somatic mutation causing a nonsense variant (p.R680*) in the duodenal and renal lesions, but not in the gastric one where a different alteration MSH2 was detected (p.G673fs*12). The remaining alterations of the gastric neoplasm were entirely different from the duodenal and renal lesions, witnessing a completely distinct origin of the former from the latter. The two neoplasms (duodenal and gastric) featured distinct gain-of function alterations in proliferative genes: the duodenal lesion and its metastasis displayed ERBB2 (p.L755S) and PIK3CA (p.V105E) mutations; dysregulation of the PI3K pathway in these lesions was further sustained by a double frameshift mutation affecting both alleles of PIK3R1. Conversely, the gastric lesion displayed KRAS (p.G12D) and GNAS (p.R210H) mutations. Homozygous B2M inactivation by a double somatic event was detected in the duodenal and renal lesions, but not in the gastric one. The gastric lesion featured homozygous inactivation of TGFBR2 by two frameshift variants (p.K128fs*3 and p.P129fs*3), each affecting a different individual allele of the gene (Fig. 4); the duodenal and renal lesions also featured the TGFBR2 p.P129fs*3 variant, but lacked the second somatic hit. Several genes involved with epigenetic/chromatin remodeling regulation were affected in both lesions, albeit by different variants. ARID1A was the only gene showing multiple alterations both in the duodenal/renal (2 frameshift variants) and gastric (3 frameshift variants lesions, although it was impossible to determine whether homozygous inactivation occurred in either lesion, due to the long mapping distance between the different variants. (Fig. 3)
The tumor immune microenvironment (TiME) of the duodenal and gastric lesions was evaluated immunohistochemically A prominent CD8 + infiltrate was detected in the duodenal specimen (score 4), but not in the gastric one (score 1); CD4 + proportion was similar between the two samples (score 2 for both samples), while a marked difference was observed in the CD11c + dendritic cells’ infiltrate which was abundant in the duodenal primitive (score 5) and lower in the gastric counterpart (score 2). Staining with an anti-interleukin-8 (IL-8, also known as CXCL-8) antibody confirmed the presence of a significant IL-8 + tumor-associated macrophages (TAM) population in the duodenal sample (score 3), which was completely absent in the gastric specimen (score 0).
During a 20-yr follow-up period since the diagnosis of MTS, the patient produced 66 neoplasms, predominantly premalignant (average of 3.4 neoplasms/year, with an increasing trend over the years), whereas no other new neoplasms (benign, premalignant, or malignant), besides the breakthrough gastric cancer, were detected during almost 4 years of ICB treatment.
The patient's tissue samples underwent targeted multigene sequencing using an NGS panel of 174 genes. The anatomic site of each lesion is indicated on the left. Abbreviations: MSI microsatellite instability, TMB tumor mutation burden, InDels genetic variants with nucleotide insertion or deletion, LOH loss of heterozygosity.
For investigating cell composition of tumor microenvironment and distribution of inflammatory cells, specific immunohistochemical staining were performed following standardized procedures of immunohistochemistry according to manufacturer instructions. The following antibodies were used: CD3 (clone: LN10, source: Leica / Germany, prediluted), CD4 (4B12, Novocastra / UK, prediluted), CD8 (C8/144B, Dako / Germany, 1:200), CD11c (5D11, Novocastra, 1:100), CD20 (L26, Novocastra, prediluted), Granzyme B (GrB-7, Monosan DBA / Italy, 1:200), PD-L1 (22C3, Dako, protocol-dilution), and IL-8 (CXCL8/IL-8, Novus Biologicals / USA, 1:50). Cells were considered “positive” when the cell membrane was stained. The expression of these biomarkers was evaluated using a semi-quantitative (0–5) scoring system: 0 = negative (no positive cells), 1 = rare (1–10 positive cells per high-power field, 400X), 2 = low (11–20 positive cells per HPF), 3 = moderate (21–30 positive cells per HPF), 4 = high (31–50 positive cells per HPF), 5 = very high (>50 positive cells per HPF).