In an interview with Oncology Learning Network, Ramesh A. Shivdasani, MD, PhD, Professor of Medicine, Dana-Farber Cancer Institute, Harvard Medical School, Boston, Massachusetts, discussed a recent study that unveiled molecular information predictive of the risk for recurrence of non-functional pancreatic neuroendocrine tumors (pNETs; Nat Med. 2019 Jul 1. Epub ahead of print).

What existing data led you and your co-investigators to conduct this research?

pNETs are rising in incidence. These tumors usually do not secrete excess hormones (non-functional), and from the clinical and pathological standpoints they are viewed as a single, homogeneous disease entity. After surgery to remove the primary tumor, however, some pNETs are cured (ie, no recurrent metastases) while about half the patients develop recurrent metastases within a few years.

There has been no simple and reliable way to predict these different outcomes, and considering that pNETs >2 centimeters are later more likely to show metastases than smaller tumors, physicians use tumor size as a rough guide. A more reliable predictor is necessary.

Please briefly describe your study and its findings. Were any of the outcomes particularly surprising?

To start, we examined epigenome profiles (the landscape of gene regulatory elements) in about a dozen pNETs. Our goal was not to answer this prognostic question, but rather to better understand how pNETs might differ from normal pancreatic cells.

Among these few cases of non-functional tumors, we were surprised to detect two different epigenome profiles, one that superficially resembled that of normal insulin-producing beta cells and the other resembling normal glucagon-producing alpha cells. Each of these normal and tumor cell types expresses different key regulatory proteins, ARX (alpha cells) and PDX1 (beta cells).

We therefore tested expression of ARX and PDX1 by standard immunohistochemistry on another 142 pNET specimens and found that tumors express one or the other protein, but rarely both. Among these 142 cases, we had detailed clinical follow-up on 103 patients, often over many years after their initial operation.

We were surprised to find that pNETs whose cells exclusively expressed the ARX protein had >35% risk for recurrent metastases, compared to <5% risk if the tumor expressed PDX1. In other words, distant metastatic relapses occurred almost exclusively in patients whose tumors expressed ARX but not PDX1. Among these patients, cancers recurred in the liver within one to four years after surgery.

What are the possible real-world applications of these findings in clinical practice?

Based on these findings, pathologists can now use immunohistochemistry (a routine clinical test) on surgical resection specimens to classify pNET tumors as the ARX type or the PDX1 type. Patients whose tumors express PDX1 can be told that their risk for recurrence after surgery is very small; they will find this information most comforting.

On the other hand, patients whose tumors are of the ARX type have a high risk for recurrence and close follow-up may detect new metastases sooner. Although there is currently no way to prevent metastasis or cure recurrent disease, patients with ARX+ tumors are the candidates for new treatments as they emerge.

Ideally, these findings should be replicated in independent studies. We expect that other investigators who read our paper will do that. Our scientific interest is in understanding in the laboratory why ARX+ PNETs are more aggressive than the PDX1+ variety.