Examining the Promise of Multicancer Early Detection Tests: Lessons Learned From PATHFINDER

In this fifth episode of OncChats: Examining the Promise of Multicancer Early Detection Tests, Toufic A. Kachaamy, MD, Madappa Kundranda, MD, PhD, and Niloy Jewel J. Samadder, MD, walk through the key lessons learned from the prospective, multicenter PATHINDER study (NCT04241796), which evaluated a multicancer early detection (MCED) test developed by GRAIL.

Kundranda: Dr Samadder, when you have a test, you have a randomized study [which in this case is] the Vanguard study, which is upcoming, like you mentioned, and it’s at 10 centers within the United States. There are a lot more centers that will not have the study, but you have a lot of patients, either due to a family history or due to personal history, that will be ordering these tests. There will be that buzz about the trial, and hence, a lot of our physicians will be dealing with the results of this test, even before we have any results, or even before enrollment is complete on [the study]. That can overwhelm the system; it can lead to a lot of unnecessary invasive and noninvasive testing, and then can potentially have a lot of negative effects. Any comments about that?

Samadder: Yeah, that’s a great question. I don’t think I have the perfect answer, but I will start by saying that we need to consider that there are 3 characteristics to an optimal MCED test. They include high sensitivity or aggregate sensitivity if it’s covering multiple cancer types; the ability to screen for a wide range of tumors anywhere from 3 to 20; and probably most importantly, to your point, [have] a low false-positive rate, which should be less than 1% because that’s what’s going to lead to less need for diagnostic procedures which will easily overwhelm the system if you start screening thousands of primary care patients, for example, every year.

The other way to potentially look at this is [to question], and you made that point, [whether] this test is better utilized in higher-risk populations, where your positive predictive rate may be better [and] your false-positive rate may be more reasonable to accept because they are already under intensive surveillance. [This can include], for example, patients with Lynch syndrome or hereditary breast and ovarian cancers, so [those with] genomic or genetic predisposition. [It can also include] patients with a strong family history of breast, colon, or ovarian cancer, where we know they’re also at higher risk and undergo more intensive screening. Finally, [it could also include] patients with maybe environmental risk factors, for example, heavy smokers. All that needs to still be [evaluated] in a trial format. We don’t know whether such products will work equally well in these populations; that is a guess at best.

I will say, that looking at the PATHFINDER data that were presented at ASCO and ESMO, there is a suggestion that this test will work better in those who are at increased risk. As such, of the roughly 7000 patients who were enrolled to the PATHFINDER study, 3681 were at increased risk either due to smoking, a prior history of cancer, or hereditary cancer risk factor. The majority of the positive cancers they found were in this population with increased risk. To me, there’s a signal there that this product may ultimately find its greatest benefit in those with some level of increased cancer risk—environmental, genetic, or familial or personal history.

Kachaamy: Since you mentioned the PATHFINDER study, I wanted to talk a little bit in detail about it. We know in [this] study, the positive signal was around 1.5%, and 65% of these reached the diagnostic resolution. However, the median time for diagnosis was 78 days, so 2 to 3 months. Ninety-three percent of participants had imaging and 72% had invasive procedures. In your opinion, and in your experience, is this what we should anticipate? One to 2% positivity and then up to 3 months to get an answer for patients? Because it’s important for people who decide to undergo these tests to know what to expect. Is this consistent with your experience?

Samadder: Absolutely, yes. It's about a 2% positivity rate at the upper end, and about half of them work out diagnostic resolution. [For] PATHFINDER, like you said, it was about a 60% diagnostic resolution to cancer diagnosis. [Also,] the split in cancers seemed to be equal between solid cancers and hematologic malignancies identified. Hematologic malignancies are quite well screened, in theory, by a circulating DNA product because it’s a bloodborne cancer and you’re taking a blood sample. There are certain cancers that are not well screened for, and patients need to also know about that; these include brain cancers. Because of a blood-brain barrier, they are not screened for at all. There [also] seems to be some black holes around cancers that seem to be missed by these MCED products, including estrogen receptor–positive breast cancers, prostate cancers, renal cancers, and uterine cancers. It’s unclear why they’re being missed. Is it due to a lack of tumor DNA being shed? Is it due to a problem in the AI algorithms or machine learning algorithms that are being identified? There are a lot of areas that [still] need to be [understood by] patients and providers, as well as new research [efforts that need to be made to understand] what can be done to eliminate some of these black holes in cancers.

In terms of diagnostic workup, it has been our typical experience that the workup can take 2 months between multiple modalities being offered, each one being waited for in terms of results before the next test is ordered, [and] getting insurance authorization for these tests. Then, at least for the experience in a trial, as well as at our own institution, much of this occurred during the COVID-19 pandemic so there were a lot of limitations on entry into the hospital system, [as well as] bandwidth of diagnostic- and screening-type procedures. That will hopefully be compressed in terms of the window of time it takes to come to diagnostic [resolution]. We’re hopefully out of the pandemic [and] we will have clear algorithms of care of what tests to order sequentially for any specific MCED or tissue of origin positive signal, and we’ll learn from our experiences of what tests are most likely to lead to diagnostic resolution. But do I think it’s going to be a few days to get diagnostic resolution? That’s unlikely. It seems like there is a bit of a needle in the haystack and you’re going after it over probably several weeks, at the very least, to get diagnostic resolution.

Check back on Wednesday for the next episode in this series.