It would be difficult to find a greater proponent of surgical treatment for neuroendocrine tumors than I am. I have devoted much of my career to expanding the indications and patient eligibility criteria for surgical treatments.1,2,3,4,5,6,7 Dr. James Howe is my valued kindred spirit and equal proponent in this field.

However, even I had a hard stop line drawn at poorly differentiated grade 3 tumors. They are essentially extrapulmonary small cell carcinomas with a dismal prognosis that are chiefly treated with cisplatin and etoposide systemic chemotherapy.8 Surgical treatment has had no impact.8 Still, if it could be shown that surgical treatment helped in any subgroups of patients, I certainly would be among the first to join in offering operations to eligible patients.

In an effort to find any such subgroups, Dr. Howe and his co-authors have performed an exhaustive systematic review and meta-analysis of the world’s literature comparing surgical and medical treatments of grade 3 gastroenteropancreatic neuroendocrine neoplasms, and they used the word “neoplasms” in the title very intentionally. Moreover, they may have found a subgroup for which surgical treatment actually might be beneficial. However, proper interpretation and application of the findings reported in this manuscript requires a thorough understanding of the pathology nomenclature for neuroendocrine neoplasms, which is evolving and, admittedly, very confusing.

Pathology reports for neuroendocrine neoplasms should always, but often do not, include two components: grade and differentiation. They are subtly but critically different. Grade estimates growth rate judged by mitotic or Ki67 indices. Grade 1 neoplasms have indices of 2% or lower, grade 2 of 3% to 20%, and grade 3 of more than 20%.

Differentiation evaluates cell morphology. Well-differentiated neoplasms have monotonous round cells with regular nuclei, well-formed nucleoli, intracellular organoid structures, and they stain positive for chromogranin A. Poorly differentiated neoplasms have pleomorphic cells with bizarre nuclei, loss of nucleoli and organoid structures, loss of chromogranin A expression, and cellular necrosis. Well-differentiated neoplasms have been termed “neuroendocrine tumors” (although they are malignant and otherwise warrant the term “carcinoma”), and poorly differentiated neoplasms have been termed “neuroendocrine carcinomas.”

Historically, this scheme contained a gross misconception. It was believed that neuroendocrine tumors were always grade 1 or 2, whereas anything classified as grade 3 was automatically a neuroendocrine carcinoma, and never the twain shall meet. However, since 2017, a new entity has been defined: the grade 3 well-differentiated neuroendocrine tumor. This neoplasm has mitotic or Ki67 indices of more than 20% but a well-differentiated cell morphology. The prognosis for the grade 3 well-differentiated neuroendocrine tumor is not as good as for the grade 1 or 2 neuroendocrine tumor, but certainly is not as dismal as for neuroendocrine carcinoma.

Unfortunately, not everyone knows this. I still see pathology reports describing the diagnosis of neoplasms as neuroendocrine carcinomas simply because they are grade 3 with absolutely no description of the differentiation. In such cases, review by our experienced neuroendocrine pathologists can be game-changing for prognosis and treatment options if it shows a well-differentiated neoplasm. The opposite also occurs. A diagnosis of grade 3 neuroendocrine tumor with no description of differentiation can be changed to poorly differentiated neuroendocrine carcinoma. These examples underscore the critical importance of proper, competent interpretation and reporting of neuroendocrine pathology.

This recent evolutionary change in the pathology classification is the crux of the matter in this manuscript. A confounding problem is that series reporting favorable outcomes for surgical treatment compared with medical treatment of “neuroendocrine carcinomas” originated before the definition of grade 3 well-differentiated tumors. Thus, we do not know how many of the patients in those reports actually may have had well-differentiated neuroendocrine tumors rather than carcinomas. This means that the favorable outcomes in those reports actually may have been due to the so-called Will Rogers effect: if you move a group with a poor prognosis (grade 3 neuroendocrine tumors) from a group with a good prognosis (grades 1 and 2 neuroendocrine tumors) to a group with an even worse prognosis (grade 3 neuroendocrine carcinomas), then the prognosis of both groups improves. As the authors point out in their discussion, the question of how many patients in those reports actually had grade 3 neuroendocrine tumors is unresolvable. But we can extrapolate from the large sample in the study by Basturk et al.9 which subjected the pathology of patients treated for “neuroendocrine carcinoma” to a subsequent review using the new distinction, that likely more than half of the patients in such reports actually had grade 3 neuroendocrine tumors and not neuroendocrine carcinomas. Hence, the seemingly favorable results.

Consequently, the most that the authors can reasonably conclude from their analyses is that surgical treatment might yield a survival benefit for some patients with grade 3 neuroendocrine tumors and that surgical treatment should be carefully considered for select patients. It should certainly be considered for patients with localized disease, and possibly extended to some patients with metastatic disease, but in the context of a multimodality approach that would likely also include chemotherapy.

Since 2017, I have operated on some admittedly highly selected patients with grade 3 well-differentiated neuroendocrine tumors, which agrees with the authors’ recommendations in their conclusions. These operations were performed after adequate periods of careful observation to establish the cadence of the disease. Although these are rare tumors among an already rare group of tumors, they also are heterogeneous within their class, and many exhibit rapid progression, whereas some do not. Given their rarity, to date, I do not have enough patients to publish my own meaningful series.

This manuscript does not provide sufficient new evidence for clinicians to begin recommending surgical treatment for patients with true grade 3 poorly differentiated neuroendocrine carcinomas. The preponderance of data show rapid development of distant metastatic disease even among patients presenting with localized disease and no improvement in survival compared with medical therapy alone. Moreover, complications from surgical resections, such as prolonged leaks after pancreaticoduodenectomy, which are much more frequent with neuroendocrine neoplasms due to soft gland and thin-walled, small caliber ducts, could delay delivery of beneficial systemic chemotherapy.

I applaud the authors for their Herculean statistical effort to unravel the Gordian knot of confounded data within the literature on this challenging topic. Meanwhile, I advise readers to scrutinize all pathology reports on neuroendocrine neoplasms carefully for reporting of both grade and differentiation and insist upon them if absent in order to make appropriate treatment decisions. Aside from the aforementioned examples, I still too often see pathology reports that provide grade or differentiation, but not both, and occasionally neither. We must all do our part to correct this. Only with accurate classification and reporting of pathology, which enables us to know what we are treating and with which treatments, and then by tracking the corresponding outcomes can we ever hope to resolve that currently unresolvable question in this challenging area of grade 3 neuroendocrine neoplasms.