A related recommendation says that when a resection specimen becomes available, pathologists should repeat Ki67 grading on the resection.
In patients with metastatic GEP-NETs, pathologists must report tumor grade for both primary and metastatic tumors for clinical management. If the tumors are resected simultaneously, the final reported grade should be based on the higher grade of the primary or metastatic tumor. In multiple metastatic samples, the largest metastasis per site should be tested for Ki67.
“Several studies have suggested tumor grade in the metastatic tumor is higher than the primary tumor,” Dr. Shi says, with 30 to 60 percent of tumors showing a significant increase in the Ki67 labeling index in metastatic tumors. “The tumor grade in the metastatic tumor is also prognostic,” she says. The guideline says pathologists may use mitotic count in addition to Ki67 grading. It notes that studies have shown that Ki67 identifies a higher grade than mitotic count alone.
The guideline discourages the eyeball estimation method, a technique for calculating the Ki67 proliferative index that can lead to grading discrepancies and poor reproducibility. “There are two exceptions,” Dr. Shi says: if the case is very low—obviously less than three percent Ki67—or very high, with a Ki67 of obviously more than 20 percent. The guideline recommends a manual count, with a printed camera-captured image preferred, although if that’s not available, the count may be performed under a microscope. “If you do it under a microscope, you can lose track of what you’re counting,” she says. “Print is the gold standard.”
Pathologists may use digital automated counting of Ki67 if the system is validated properly. Some studies evaluating the performance of digital automated counting showed good concordance with manual counts; others did not. “The key message from the guideline is this: If you want to use the digital imaging analysis system, that system needs to be validated,” Dr. Shi says.
Digital imaging comes with its own difficulties, she says, largely because of analytical variables. She recollects a case in which a patient’s tumor was given a grade 3. “The clinician was very worried,” she recalls. “We reviewed the Ki67 again [and saw that] most of the positive cells were lymphocytes. Whoever was doing this was not trained properly and was counting all the background lymphocytes as tumor cells. So there’s a lot regarding digital imaging that people need to be aware of.”
Preanalytical variables such as overstaining, too, can falsely elevate the Ki67 index. In one case, “we stained the same tumor twice,” she says. “One was dark or heavy with a strong background. The Ki67 index was 60 percent for the image with heavier staining, and 30 percent for the second image, which had lighter staining. So preanalytic [variables] can greatly affect the index.”
As for artificial intelligence, one promising future use would be to provide a continuous Ki67 percentage along with the WHO tumor grade. A Ki67 percentage between three and 20 percent is considered a tumor grade 2. “It’s an arbitrary cutoff,” Dr. Shi says, with data behind it but some studies indicating otherwise. Reporting a continuous number could help refine it. Multiple studies already separate the pancreatic neuroendocrine tumors into grades 2a (three to less than 10 percent Ki67) and 2b (10 to 20 percent Ki67), she notes. “You can see a very nice survival difference between 2a and 2b. That means in the future we might fine-tune the cutoff.” But without continuous Ki67 data, she says, “we can’t get a good conclusion from the cases we have.”
It’s worth noting one of the guideline’s good practice statements, which says to report histologic findings of the background gastric mucosa in patients with gastric NETs. Originally, Dr. Shi says, gastric NETs were classified into three subtypes. More access to genomic testing and more patients on long-term proton pump inhibitor (PPI) use has led to the discovery of two new subtypes. “These five subtypes have different clinical settings and different management,” she says.
One new subtype involves germline mutations that can impair gastric acid production and lead to neuroendocrine tumor development. The other is likely related to long-term PPI use, common in the U.S. PPI-related neuroendocrine tumors are not especially aggressive, “but they have relatively normal background mucosa that can be confused with type 3,” she says.
The PPI-related subtype was identified relatively recently, she says. “We saw it 10 to 15 years ago, but only occasionally.” Cases have become more common, she says, because the patients taking PPIs at that time now have been on them for a longer period. The PPI-related subtype will be included in the new WHO classification to be published this year.
“This is something new that much of pathology may not know,” she says.
Charna Albert is CAP TODAY senior editor.