Molecular Profiling for Proper Diagnosis of Gastroenteropancreatic Neuroendocrine Tumor

INTRODUCTION

Next-generation sequencing has revolutionized oncology, with determination of cancer genetic profiles leading to precision medicine for the management and treatment of oncology patients. Although anatomic and morphologic characterization continues to be the foundation of cancer classification, molecular classification is now widely used to complement traditional classification methods.¹

Neuroendocrine neoplasms are rare malignancies derived from neuroendocrine cells, with those from the digestive system being the most common.^2-4 Currently, the fifth edition of the WHO classification of tumors of the digestive system classifies gastroenteropancreatic neuroendocrine neoplasms by grade and morphologic differentiation.⁵ Well-differentiated neuroendocrine tumors (NETs) characteristically show small, monomorphic cells in islets or trabeculae with fine to coarsely granular chromatin, whereas poorly differentiated neuroendocrine carcinoma (NEC) shows diffuse sheets of pleomorphic cells.^2,3 Because neuroendocrine neoplasms are heterogeneous, morphologic diagnosis alone is difficult. Here, we show the utility of molecular profiling for accurate diagnosis and appropriate treatment.

CASE REPORT

A 50-year-old man with no significant medical history presented at an outside hospital with right upper quadrant pain in April 2019. The patient reported no bowel symptoms. An abdominal and pelvic computed tomography (CT) scan showed multiple large hepatic lesions suspicious for metastatic disease, pancreatic tail heterogeneity, and sigmoid and rectum narrowing. The radiologist interpreted the sigmoid wall thickening as likely corresponding to known malignancy, although endoscopic evaluation of the large bowel was not performed. The largest right hepatic lobe lesion measured up to 12.4 cm in greatest dimension. Ultrasound-guided liver core needle biopsy showed a trabecular and vaguely gland-forming tumor (Fig 1A). The cells had abundant eosinophilic cytoplasm and hyperchromatic nuclei (Fig 1B). Immunohistochemical (IHC) stains were strongly positive for AE1/AE3, CDX2, and CD10; weakly positive for CK20; and negative for CK7, vimentin, Pax-8, and prostate-specific antigen, favoring a colorectal primary tumor (Figs 1C-1F). MLH1, MSH2, MSH6, and PMS2 expression were intact by IHC.

FIG 1.

Ultrasound-guided liver core needle biopsy performed at an outside hospital showed a tumor with trabecular and vaguely gland-forming architecture. Cytologically, the cells displayed abundant eosinophilic cytoplasm and hyperchromatic nuclei: (A) ×100, hematoxylin and eosin (HE); (B) ×400, HE. Immunohistochemical stains were strongly positive for (C) AE1/AE3 (×200), (D) CDX2 (×200), and CD10 (not shown); weakly positive for (E) CK20 (×200); and negative for (F) CK7 (×200), vimentin, Pax-8, and prostate-specific antigen (not shown). Subsequent immunohistochemical stains after results of molecular testing were strongly positive for (G) synaptophysin (×400) and (H) CD56 (×400) and negative for (I) chromogranin (×400). (J) Ki67 proliferation index was approximately 15% (×400).

The patient’s care was transferred to Rutgers Cancer Institute of New Jersey (RCINJ), and leucovorin, fluorouracil, and oxaliplatin (FOLFOX) plus bevacizumab was initiated in May 2019 according to National Comprehensive Cancer Network guidelines for metastatic colorectal cancer treatment. The patient provided informed consent to participate in a prospective trial for tumor genomic profiling (ClinicalTrials.gov identifier: NCT02688517) approved by the Rutgers University New Brunswick Health Sciences Institutional Review Board (Pro2012002075). This consent includes patient approval for publication of these results. The outside liver biopsy was reviewed by a GI pathologist at our institution who favored a colorectal primary diagnosis based on the supplied IHC slides. Another liver biopsy for tumor-only comprehensive genomic profiling (Foundation Medicine, Cambridge, MA) was obtained and diagnosed as metastatic carcinoma consistent with colorectal primary by a different pathologist. In total, 1 outside pathologist and 2 pathologists at our institution favored the diagnosis to be colorectal carcinoma. Genomic profiling results were reviewed by the RCINJ Molecular Tumor Board and were significant for MEN1 R521fs*15, ATRX splice site 5787-9_5809del32, CDKN1A C34fs*1, CDKN2A/B loss, low tumor mutational burden (3 mutations/Mb), microsatellite stability, and wild-type APC, KRAS, NRAS, BRAF, and TP53. MEN1 and ATRX mutations are commonly found in pancreatic NET,^6-8 with additional CDKN2A loss frequently reported in metastases.⁹ These results and the absence of commonly observed colorectal cancer mutations suggested a NET and prompted pathology rereview. Additional IHC performed on the original liver core biopsy specimen was synaptophysin and CD56 positive and chromogranin negative (Figs 1G-1I). Ki67 proliferative index was approximately 15% in > 500 assessed cells (Fig 1J). This immunostaining pattern was suggestive of intermediate-grade NET. In addition, gallium-68 DOTATATE positron emission tomography/CT scan performed in August showed tumor uptake in the liver.

The patient was not known to have any signs or symptoms related to multiple endocrine neoplasia type 1 (MEN1) syndrome or NET such as flushing or diarrhea. Chromogranin A (Fig 2) and neuron-specific enolase (NSE) were elevated (chromogranin A, 141 ng/mL [normal < 93 ng/mL]; NSE, 47 ng/mL [normal ≤ 15 ng/mL]). Gastrin was within normal limits (21 pg/mL [normal < 100 pg/mL]). With the NET diagnosis and after already receiving 4 cycles of FOLFOX plus bevacizumab, in July 2019, the patient’s treatment regimen was adjusted to fluorouracil, leucovorin, and bevacizumab every 2 weeks and lanreotide monthly because the regimen of fluorouracil, leucovorin, and bevacizumab has been shown to have activity in advanced NET.^10,11 Chromogranin A decreased to 23 ng/mL after 2 months of therapy with the adjusted regimen and continued to stay within normal limits (Fig 2). CT scan in September 2019 of the largest right hepatic lobe lesion demonstrated a decrease from 12.4 cm to 8.9 cm in greatest dimension. In December 2019, fluorouracil and bevacizumab were discontinued while continuing lanreotide as a result of splenic vein thrombosis resulting in GI bleeding and ultimately splenectomy. In February 2020, everolimus was initiated before disease progression, and the patient continues to be treated with everolimus and lanreotide. Chromogranin A levels have continued to remain within normal limits (Fig 2).

FIG 2.

Chromogranin A levels in the presented patient (blue line) from June 2019 to February 2020. Upper limit of normal is < 93 ng/mL (dashed black line).

DISCUSSION

Gastroenteropancreatic neuroendocrine neoplasms are rare and increasing in incidence.⁴ Neuroendocrine neoplasms histologically range from well-differentiated NETs to poorly differentiated high-grade NECs.^2,3,5 Although well-differentiated NETs are classically composed of small, monomorphic cells with fine to coarsely granular chromatin forming islets or trabeculae, morphologic heterogeneity exists and may overlap with other entities.^2,3,5 The discrimination between well-differentiated NET with a high-grade component and poorly differentiated NEC is particularly important.¹² In this case, glandular architecture without classic nuclear features led to the consideration of adenocarcinoma. Functional somatostatin-producing duodenal NETs often display glandular architecture that may be mistaken for adenocarcinoma.³ Diagnosis of high-grade pancreatic neuroendocrine neoplasms is particularly challenging with ambiguous morphology.¹³ Ancillary IHC is routinely used for diagnosis, with chromogranin A, synaptophysin, and CD56 indicating neuroendocrine differentiation and high-grade tumors showing more limited chromogranin A staining.²

Molecular studies are not traditionally used to aid neuroendocrine neoplasm diagnosis; morphology and IHC in combination are often enough. In this case, neuroendocrine IHC was not performed initially, and the tumor molecular profile with MEN1 and ATRX alterations combined with wild-type APC, KRAS, and TP53 prompted suspicion of a neuroendocrine neoplasm rather than a colorectal adenocarcinoma. MEN1, a tumor suppressor gene encoding MENIN; DAXX/ATRX, a transcription/chromatin remodeling complex; and mammalian target of rapamycin pathway genes are most frequently mutated in pancreatic NET.⁶ With the patient’s tumor reclassified as an intermediate-grade NET and the continued response to the adjusted chemotherapy regimen, this patient’s case exemplifies molecular studies as a useful tool for the proper classification of cancer.

Although germline MEN1 mutations may occur in the context of MEN1 familial syndrome, up to 44% of MEN1 mutations in pancreatic NETs occur sporadically.^14,15 DAXX (∼20%) and ATRX (∼10%) mutations also occur frequently in sporadic pancreatic NETs.⁸ Studies suggest that MEN1-, DAXX-, and ATRX-mutated pancreatic NETs form a subgroup with distinct genetic expression, DNA methylation profiles, and prognostication.⁸ Although Jiao et al⁶ initially found prolonged survival in patients with MEN1-, DAXX-, and ATRX-mutated tumors compared with those with wild-type pancreatic NETs, subsequent studies showed worse prognosis with MEN1-, DAXX-, and ATRX-mutated pancreatic NETs.^8,9 With further elucidation of this distinct molecular subgroup, it is likely that neuroendocrine neoplasm molecular classification may be incorporated in the future.

Neuroendocrine neoplasms commonly metastasize to liver, lung, and bone, with nonspecific features making primary site identification difficult.^3,15 Here, CDX2 positivity and liver metastases strongly suggest gastroenteropancreatic origin.^3,15 Although the primary site of this patient’s NET is not known definitively, the molecular profiling results in combination with CT findings of heterogeneity within the tail of the pancreas and splenic vein thrombosis suggest a pancreatic origin. Although studies are scarce, data suggest site-specific differences between NET molecular pathogenesis.¹⁶ For example, β-catenin mutations have been reported in 38% of GI NETs.¹⁷ In addition, small intestine NETs have been reported to have recurrent mutations in CDKN1B (∼10%), APC (∼8%), and CDKN2C (∼8%).¹⁸

In conclusion, this patient case illustrates the recognition of specific genomic alterations to appropriately diagnose and treat a patient with metastatic cancer in the setting of unusual tumor morphology. This is another example of genomic profiling providing more information than just finding an actionable alteration.¹⁹ Although none of individual mutations are clearly targetable, the pattern of mutations led to the diagnosis of NET, which in turn led to a change in management and treatment. In addition, this change in diagnosis has implications for prognosis and future therapies for this patient, such as possible treatment with sunitinib, peptide receptor radionuclide therapy, or cytoreductive procedures. Although it is possible the patient may have been having some response to FOLFOX plus bevacizumab, which was initiated in May 2019, chromogranin A levels did not decline below the upper limit of normal until August 2019, after lanreotide was added to the regimen in July 2019. As routine molecular profiling continues to grow, this will emerge as another important tool to aid in proper diagnosis. Emerging data on gastroenteropancreatic NETs show that these tumors have distinct molecular profiles dependent on their morphology, grade, prognosis, and site of origin that will be important for patient management.

AUTHOR CONTRIBUTIONS

Conception and design: Virian D. Serei, Shridar Ganesan, Gregory Riedlinger

Provision of study materials or patients: Elizabeth Poplin

Collection and assembly of data: Virian D. Serei, Elizabeth Poplin, Gregory Riedlinger

Data analysis and interpretation: Virian D. Serei, Shridar Ganesan, Gregory Riedlinger

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/po/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).

Elizabeth Poplin

Research Funding: Neogenix Oncology (Inst)

Travel, Accommodations, Expenses: Clovis Oncology

Shridar Ganesan

Employment: Merck (I)

Stock and Other Ownership Interests: Ibris, Inspirata, Merck (I)

Consulting or Advisory Role: Inspirata, Novartis, Roche, Foghorn Therapeutics, Foundation Medicine, Merck Sharp & Dohme

Patents, Royalties, Other Intellectual Property: I hold 2 patents for digital imaging that may be licensed to Ibris, and Inspirata

Other Relationship: National Cancer Institute/National Institutes of Health

Gregory Riedlinger

Honoraria: MJH Healthcare Holdings, Gerson Lehrman Group

Consulting or Advisory Role: Personal Genome Diagnostics

No other potential conflicts of interest were reported.