Reading Between the Codes: How Seemingly Unrelated Conditions May Signal Earlier Multiple Myeloma

New real-world data suggest that the path to a multiple myeloma (MM) diagnosis is often marked by a constellation of nonspecific encounters, some of which may appear unrelated at first glance.

In this interview with Pharmacy Times, Faith Davies, MD, discusses why increased gastroesophageal reflux disease (GERD) and cardiovascular-related codes emerged in the pre-diagnosis period, how pharmacists can interpret these signals within the broader diagnostic journey, and what opportunities exist to shorten diagnostic delays. She also explores how pattern recognition, population-level data, and future electronic health record (EHR)-based tools could empower pharmacists to play a more proactive role in identifying patients at risk for MM earlier in the disease course.

Q: The study also highlighted increased gastroesophageal reflux disease (GERD) and cardiovascular-related codes in the pre-MM population. How could pharmacists interpret or contextualize these seemingly unrelated conditions within the broader diagnostic journey of MM?

Faith Davies, MD: So in the study, some unusual codes came up, particularly GERD and endoscopy. And I think it's difficult to say how they're related. I guess that's one of the joys of big data, is that you often actually can't delve into it. It could well be that those patients presented with anemia, and therefore were going through a workup for anemia, and so that's how they ended up having those codes being identified. Or it could just be some other complex kind of symptom related to multiple myeloma, we're not really sure. But again, I think it's just one of those flags that we need to think about.

Q: Considering that many patients with multiple myeloma experience multiple health care encounters before reaching a specialist, what opportunities exist for pharmacists to help shorten the diagnostic delay, particularly in ambulatory, primary, or community settings?

Davies: I think it's putting the pieces of the puzzle together, so having that patient that may have a collection of nonspecific symptoms with some nonspecific slight changes in their tests, and thinking, “Oh, I wonder if this might be myeloma.” And one of the common ones, I guess, is just that slightly high total protein that you get these nonspecific kind of symptoms, and then on their renal profile, they have the slightly higher protein. One of the things we're hoping to do with the data moving forward, though, is that you could imagine that, using AI or whatever, because we have this selection of codes, is there something that we may be able to proactively program into an electronic health care record to say, okay, although these kind of diagnostic codes appear unrelated, they may actually signify that something's going on. And so, there's plenty of work that needs to be done around this, but you could imagine that there might be able to develop a tool that we could have in an EHR that raised a flag and said, “Oh, you've got this very specific pattern a year earlier than you've presented. Let's think about myeloma.”

Q: Given the matched control design and the use of large administrative claims data, how might these findings influence pharmacists’ approach to population health initiatives or risk identification programs for older adults?

Davies: Yeah, so at the moment, we don't have a national screening program for myeloma, and interestingly, there's some work coming out of Iceland. They do have one, and [we are] just discussing whether that is an approach we may or may not want to go down. But I think the way that this data is going to be very useful is if we can use it, rather than the individual factors, but as a pattern to say, “Okay, if patients have this kind of pattern in their health care records, then we should be thinking about at least screening them for multiple myeloma.” And so, what we kind of need to do as the next step in this puzzle is obviously test out our pattern in a second data set. That's clearly going to be important. But then try and figure out how we might be able to kind of utilize electronic health care records to do this prospectively rather than retrospectively.