Distinguishing Between Companion and Complementary Diagnostic Tests
Benjamin P. Levy, MD: Companion diagnostic means that you have to have the test in order to give the drug. And this is a term that we’ve known very well. There are companion diagnostics for targeted therapies, and there is a companion PD-L1 test that you have to have. You have to be PD-L1 positive to give pembrolizumab, or Keytruda.
So, that’s a companion diagnostic. That term has been in our lexicon for quite some time. Complementary is very new, and many of us are trying to understand what this means. Nivolumab does not need a PD-L1 test, based on the data that showed a survival advantage in all patient populations in squamous and in adenocarcinoma. The PD-L1 testing is not required, but it’s recommended. My interpretation of what a complementary test is, is that it’s not required, but it’s recommended. A companion diagnostic is one you’ve got to do if you want to give the drug, and we’re still trying to sort out these definitions. There’s a lot of complexity with these two drugs given that one requires a companion diagnostic, one is complementary. Those are my interpretations of what those two are. But, I imagine if you asked other medical oncologists, they may tell you something different.
The complementary test gives a little wiggle room to the physician, because it’s not required. And if a patient’s PD-L1 is positive, there’s no doubt that that patient is going to respond. But if they’re negative, that doesn’t mean that they’re not going to derive a benefit. So, the question is, if you have a test in which it’s negative, and it still may mean that the patient’s going to respond, why would you test in the first place? And there really have been two camps that have come out within medical oncology.
Some medical oncologists say, “Look, I want to make sure up front that I know what the chances of response with this drug are going to be, so I’m going to order the PD-L1 testing.” If the PD-L1 testing is greater than 50%, I know with Keytruda, based on the data, that the response rate is going to be in the 40%-to-45% range. However, if you look at the response rates that are in some of the data in PD-L1–negative patients, it’s somewhere between 20% to 30% depending on what platform is used. In my mind, in my practice, it’s not ready for prime time. I do not routinely test for PD-L1 for my patients, given the fact that even if they’re negative, there’s a chance that they will derive a benefit. If they’re positive, yes, that will define for me patients more likely to respond. But I would hate to exclude a patient from getting these drugs based on a test where, if negative, it means that they’re not going to derive a benefit.
There are other people, however, who think that it’s important to do the PD-L1 testing on all patients because it helps define those patients who are more likely to respond. An issue that’s not talked about much is how much tissue we have left. So, we have to do molecular analysis on the tissue. Sometimes, the tissue is a fine needle aspirate, so if there’s not a lot of tumor content, I’ve got to do that first. Then, if a patient is part of a clinical trial, I have to make sure there’s tissue available for that. And then I have to make sure that there’s tissue available for PD-L1 testing, which generally is only one or two slides. But sometimes our tissue procurement yields very low tumor content.
And so we have to be judicious and sort out, in terms of relevance, where our tissue goes. Molecular testing has to be done; that’s first. If the patient is going on a clinical trial, at least in my center, that’s next. If there’s room, or there’s tissue left over, maybe consider PD-L1 testing. And I think that’s a lot of what people are thinking about: how we have to be careful with tissue allotment based on the competitive needs from different diagnostic platforms.