A 69-year-old white male presented to medical attention with 2–3 weeks of dyspnea and melena and a syncopal event. His medical comorbidities included gastroesophageal reflux disease, dyslipidemia, peripheral neuropathy, hypertension, colovesicular fistula with sigmoid resection related to an abdominal abscess, and sepsis following a root canal. By exam, a firm upper midline mass could be palpated. Initial investigations demonstrated profound anemia (hemoglobin 65 g/L) and a 10.5x4 cm gastric submucosal mass on computerized tomography (CT) imaging. Esophagogastroduodenoscopy (EGD) demonstrated 2 gastric ulcers on the greater curvature with distorted anatomy suggestive of extrinsic compression. A gastrointestinal stromal tumour was suspected, and he underwent exploratory laparotomy which revealed a locally advanced gastric tumour not amenable to immediate resection by the rural general surgery team. He was referred to a tertiary care hospital for consideration of radical resection.
Two weeks later, he underwent endoscopic ultrasound for tissue biopsy by gastroenterology. Endoscopic ultrasound at the area of the gastric mucosal ulceration revealed a hypoechoic mass arising from the 4th echolayer with internal heterogeneity that was highly vascular. A core biopsy was obtained.
On microscopic examination of the biopsy, neoplastic cells were monotonous with round nuclei and clear cytoplasm (see Fig. 1). A few mitotic figures were seen. Immunohistochemically, the tumour cells stained positive for actin, calponin and synaptophysin in a patchy distribution. The sample was most consistent with a glomus tumour by morphology and immunohistochemistry (see immunohistochemistry results in Table 1).
Table 1
Immunohistochemistry on gastric mass biopsy consistent with a glomus tumour
Test | Result |
Smooth muscle markers SMA, MSA, calponin, caldesmon Desmin | Positive Negative |
Neuroendocrine Synaptophysin Chromogranin, CD56, CDX2, CKAE1/3, CK8/18 | Positive Negative |
Epithelial markers (CKAE1/3, CK5/6, CK8/18) | Negative |
Lymphoid marker (LCA) | Negative |
Gastrointestinal stromal tumour markers (CKIT, DOG1) | Negative |
Schwanian/Neuroectodermal/melanocytic marker (S100) | Negative |
Mesothelial marker (WT1) | Negative |
Melanoma markers (MART1, HMB45) | Negative |
Vascular markers (CD34, ERG) | Negative* |
Abbreviations: CD, cluster of differentiation; CK, cytokeratin; CKIT, kit oncogene; DOG-1, discovered on GIST-1; ERG, ETS-related gene; HMB-45, human melanoma black 45; LCA, leukocyte common antigen; MART-1, melanoma-associated antigen recognized by T-cells 1; MSA, muscle-specific actin; SMA, smooth muscle actin; WT1, wilms tumor. *Except in a few vessels visible inside the tumour |
Approximately 5 months after his initial presentation, the patient continued to have symptomatic anemia requiring transfusion. Radical surgical resection was performed. Intraoperatively the tumour was localized at the gastric antrum (10.5 x 9.0 x 6.5 cm), invading omentum, pericolic fat, and pancreatic peritoneum with vascular invasion and lymph node involvement. A laparotomy, antrectomy and en-bloc resection of mid transverse colon, right hemicolectomy, Biliroth II gastrojejunostomy, cholecystectomy and biopsy of peripancreatic lesion was performed. Pathologic examination was consistent with malignant glomus tumour (2 of 4 lymph nodes positive for metastasis). The resection specimen had a high mitotic rate (30 mitoses / 50 high power fields). His clinical symptoms resolved and serial imaging surveillance for recurrence was initiated.
Approximately 1 year after his initial presentation, CT imaging revealed new lesions in the liver (largest 4.2 x 4.1cm) and thickening of the pancreatic tail/body suspicious for local recurrence. A liver biopsy confirmed recurrent malignant glomus tumour.
Gene sequencing and genomic profiling was performed (see Table 2). A 15-gene next-generation sequencing panel (TruSight15, Illumina, San Diego, CA) did not identify any mutations. Subsequent comprehensive genomic profiling (FoundationOne CDx, FoundationOne Medicine, Cambridge, MA) revealed a NOTCH2 rearrangement at exon 26 (partner gene not reported) and frameshift mutation in ATRX (C605fs*16). Tumor mutation burden was low (2/Mb) and there was no evidence of microsatellite instability, both suggesting a low probability of sensitivity to immuno-oncology drugs. Given evidence that ATRX plays a role in homologous recombination repair, it was hypothesized that a tumour lacking ATRX function may respond to a PARP inhibitor and to conventional platinum-based chemotherapy. Although clinical trials of NOTCH inhibitors were enrolling patients at the time, it was not feasible for the patient to travel for eligibility assessment.
Table 2
Summary of tumor genomic profiling and predictive response to therapy
Gene/Biomarker | Mutation/Result | Predictive response to therapy |
NOTCH2 | NOTCH2 rearrangement at exon 26, partner gene not reported | Notch inhibitor (clinical trial) |
ATRX | C605fs*16 | Pt-based conventional chemotherapy, PARP inhibitor (clinical trial) |
Tumor mutation burden | Low (2/Mb) | Low sensitivity to immune-oncology drugs |
Microsatellite instability | Negative | Low sensitivity to immune-oncology drugs |
Guided by medical oncology, palliative chemotherapy was pursued. His treatment course included: (1) doxorubicin, 4 cycles, stopped due to minimal response and cardiotoxicity; (2) ifosfamide, 3 cycles, stopped due to disease progression; (3) docetaxel/gemcitabine, 3 cycles, stopped due to disease progression; (4) dacarbazine, 3 cycles, stopped due to disease progression; (5) pazopanib, 6 weeks, stopped due to disease progression. At that point systemic therapies were stopped, and palliative care involved. He soon died from complications of his metastatic cancer approximately 2 years and 10 months after initial presentation.