Aditya Bardia, MD, MP: That’s a good segue to the different tests, Ki-67 and Oncotype DX. Virginia, can you comment on the role of Ki-67, particularly related to the use of abemaciclib in the adjuvant setting?

Virginia Kaklamani, MD: Absolutely. Thanks, Aditya. This was something we started doing in the past few years because of the monarchE clinical trial and the inclusion criteria, as well as the way the FDA approved abemaciclib in the adjuvant setting. Now Ki-67 is our standard-of-care test for all cancers. We primarily use it for decisions on giving adjuvant abemaciclib to our patients.

Aditya Bardia, MD, MPH: Have you received pushback from the pathology department in terms of Ki-67 and the cutoff? We’d also love to hear from you, Michelle. What’s your opinion about Ki-67 from an analytical perspective? But first, Virginia.

Virginia Kaklamani, MD: We used to do Ki-67 before I moved here, and I asked them to stop—that was in 2015—because we didn’t have any data at the time to use it. Some people were arguing that we should be against Oncotyping our patients if we have a low or high Ki-67. I stopped that practice. With monarchE data, we started that again. Pathologists were happy to see that test back in their practice.

Aditya Bardia, MD, MPH: Michelle?

S. Michelle Shiller, DO, AP/CP, MGP: The utility of Ki-67—certainly you want testing to help inform your clinical decision-making—also helps us look at tumor behavior. We’ve consistently tested Ki-67 for a number of years, and it’s good to see that there are clinical data that correlate with it and help drive therapeutic decisions that will help the patients have a different and better response.

One issue we run into, just to put it out there in case anyone else has come across this, is that insurance companies and payers aren’t necessarily paying us when we run that test. As a group of providers, oncologists and pathologists need to unite and help inform payers as to why we’re running the assay and how it informs the clinical decision-making. Payers tend to lag behind a bit, but I wanted to mention that in the interest of anyone who’s dealing with that locally. That’s a real thing. We’re seeing pushback and lack of reimbursement for that biomarker, but we do it anyway.

Virginia Kaklamani, MD: Michelle, I was puzzled when I saw some data several years ago from world experts on Ki-67—I’m talking about pathologists. The results they got were very discordant from one another. Now we have this 20% [standard]. If it’s 19%, a patient may not be eligible for a pretty effective adjuvant therapy. But if it’s 20% or 21%, they will. How standardized is Ki-67?

S. Michelle Shiller, DO, AP/CP, MGP: The better question is how standardized is trying to quantify an expression with human interpretation. Artificial intelligence is a hot topic in general. In pathology it’s more of a gestalt or an impression, unless you specifically count and enumerate the cells. Therefore, a lot of studies have shown that pathologists aren’t consistent at calling a specific numeric value—5%, 10%, 25%. The greatest consistency is in tertiles: 1% to 33%, 34% to 66%, and 67% to 100%. Beyond that, that’s when you start seeing that concordance fall apart. That’s also been reflected in PD-L1 IHC [immunohistochemistry] interpretation, and anything in which we’re being asked to quantitate. Unless you’re numerically counting them, it’s going to be difficult to reproduce it. My solution is to consider artificial intelligence [AI]–based platforms, where you have a machine counting it, for those kinds of numeric evaluations.

Virginia Kaklamani, MD: Is that how you guys do it?

S. Michelle Shiller, DO, AP/CP, MGP: No, that’s not how we do it.

Virginia Kaklamani, MD: How common is AI?

S. Michelle Shiller, DO, AP/CP, MGP: Not common. It’s a hot topic in the world in general. In pathology, artificial intelligence is under consideration for a lot of things. For screening high-volume things such as Paps [Papanicolaou tests] for abnormal cytology and sending more complex or difficult borderline cases to a pathologist for more specific and detailed review. But it has its role when we’re trying to quantify. They’re looking at it with PD-L1 expression as well, and they found it with counting NIK, for example, and other tumors. It’s not common, but it’s an area that we should all be aware of for future growth.

Aditya Bardia, MD, MPH: That’s great. That was really helpful. Thanks so much.

Transcript edited for clarity.