Mayo Clinic, Inc. (EXAS) Earnings Call Transcript & Summary

Good day, and thank you for standing by. Welcome to the Cologuard 2.0 ASCO GI Data conference call. [Operator Instructions] Please be advised that today's conference is being recorded. [Operator Instructions] I would now like to hand the conference over to your speaker today, Megan Jones. Please go ahead.

Thank you, operator, and thank you for joining us to discuss results of the second-generation Cologuard study presented at ASCO GI this week. On the call today are Kevin Conroy, the company's Chairman and CEO; and Dr. John Kisiel, practicing gastroenterologist at Mayo Clinic and Exact Sciences collaborator; Scott Johnson, our Senior Vice President of Research and Development; and Jeff Elliott, our Chief Financial and Chief Operating Officer, will also be available for questions. We'll only be taking questions on the results of the study today. Exact Sciences issued a news release yesterday afternoon detailing the study results. This news release and today's presentation are available on our website at exactsciences.com. During today's call, we will be making forward-looking statements based on current expectations. Our actual results may be materially different from such statements. Descriptions of the risks and uncertainties associated with Exact Sciences are included in our SEC filings, which can be accessed through our website. It is now my pleasure to introduce the company's Chairman and CEO, Kevin Conroy.

Good morning, and thank you for joining us. We released promising data yesterday at ASCO GI, demonstrating the power of our second-generation Cologuard test, which we're calling Cologuard 2.0. The Exact Sciences team has made incredible progress towards providing an even better colorectal cancer screening option for patients and health care providers. Through continuous innovation with our partners at Mayo Clinic, we've discovered new biomarkers to improve Cologuard test performance and the patient experience. The results we'll discuss today highlight the progress we've made toward achieving our goals for Cologuard 2.0 which are to: one, improve specificity for the false positive rate; improve precancer sensitivity; and three, maintain cancer sensitivity. Today, we'll discuss how the trial was designed to better predict performance in a real-world prospective setting, how we interpret the results and the positive impact Cologuard 2.0 could have on patient outcomes. I'll now turn the call over to Dr. John Kisiel, a practicing gastroenterologist at Mayo Clinic and one of our key collaborators and an author on this study.

Thank you, Kevin. More than a decade of research collaboration between our team at Mayo and Exact Sciences led to the discovery of a new panel of highly discriminate methylation markers, which have been included in the second-generation Cologuard test. And we believe that using this multi-target approach, which includes these improved methylated DNA markers and fecal hemoglobin could meaningfully increase the accuracy of Cologuard. We carefully designed this study to simulate a real-world screening population to better anticipate the performance that we would see when moving into a large prospective trial. First, all the adenomas and other precancers were prospectively collected, and we deliberately included different histologic subtypes and sizes. The median advanced adenoma size and precancer size was identical to that seen in the DeeP-C study, that was the prospective pivotal trial for the first-generation Cologuard. Second, the cancer stages were weighted to resemble those seen in asymptomatic screening population, specifically like BT. And here, 78% of the cancers we included in the study were Stage 1 or 2, and the median cancer size was also set to be identical to that seen in DeeP-C. Third, all 777 samples were collected from patients enrolled at 134 sites in the United States and Canada. Again, these were all average risk persons, including 112 with cancer, 98 who had advanced precancerous lesions and 176 with nonadvanced or small precancerous lesions and 391 persons with negative colon findings confirmed by colonoscopy. The samples were tested using an assay process that's similar to the current Cologuard. But this time, there's an exclusion in the analysis of mutant KRAS. And we believe that by replacing the markers in the current Cologuard panel with this newly discovered methylation marker set is going to help us improve specificity without sacrificing cancer sensitivity. To really ensure a rigorous evaluation of this novel marker panel, our analysis models were cross-validated 1,000 times, and this was done to control for risk of overfitting. And we also built this panel out of as few markers as possible. And what we see is nearly identical performance between the cross-validated results and the best fit estimates. The median performance in the cross-validated analysis showed that the second-generation Cologuard test was 95% sensitive for all stages of cancer at 92% specificity among patients with no findings on colonoscopy. Specificity was 90% when we included patients with small polyps and other incidental findings as having negative colonoscopy. The cancer sensitivity was 94% for Stage 1, 95% for Stage 2 and 100% for Stages 3 and 4. The panel also showed high sensitivity for advanced precursors with 57% sensitivity, including 83% sensitivity for high-grade dysplasia, the most dangerous form of precancer. Catching precancer before it progresses into cancer and catching outright cancer at its earliest stages when it's most treatable are really the most important attributes of a colon cancer screening test. And I believe that these results show the ability of second-generation methylated DNA markers and fecal hemoglobin in the Cologuard test 2.0 to reliably detect advanced precancerous lesions and early-stage colorectal cancer at high specificity. I think this is incredibly promising and will positively impact patient outcomes in colon cancer. I'll turn the call back to Kevin.

Thanks, Dr. Kisiel. We're thrilled by these results and how Cologuard 2.0 could benefit patients. Cologuard is the most accurate noninvasive colorectal cancer screening test and Cologuard 2.0 will raise the bar, improving patient outcomes. When we model these performance data and compare it to current Cologuard performance, Cologuard 2.0 leads to an additional 11,000 life years gained and 150,000 fewer follow-up colonoscopies per 1 million people screened. Per the USPSTF modeling assumptions, this translates to a 3 percentage point reduction in colorectal cancer incidents and a 2 percentage point reduction in mortality, getting us one step closer to helping to eradicate this disease. Replacing the markers in Cologuard in the current sample preservative will improve patient compliance and help get more people screened. The chart on Slide 7 shows the power of the Cologuard 2.0 enhancement and how they will help achieve those financial targets and the efficiency of Cologuard. Some of the most impactful being improving the health economic case for Cologuard, sending fewer patients to invasive follow-up colonoscopies unnecessarily, increasing lab throughput, allowing patients to collect their sample any day of the week and decreasing the rejection of expired samples by 1 to 2 percentage points. These will lead to a better patient experience with fewer barriers to adoption and compliance as well as operational efficiencies. We have a multi-cancer pivotal trial ongoing called BLUE-C to further validate Cologuard 2.0 markers. We expect to finish enrollment later this year and report top line results afterwards. We'll use results of BLUE-C to support an FDA submission and make Cologuard 2.0 available broadly through our strong commercial, operations and lab teams. On a separate note, there was recently good news on the cost of follow-up colonoscopies. The U.S. Department of Labor issued guidance last week, clarifying that all follow-up colonoscopies after a positive noninvasive stool test like Cologuard, are part of the screening process and must be paid for by commercial insurance providers at no cost to the patient. The guidance applies to insurance plans starting after May 31 of this year. This follows the U.S. Preventive Services Task Force recommendation and removes another barrier to getting more people screened. We're focused on eradicating cancer and the suffering it causes by making earlier cancer detection a routine part of medical care. Data shared here today give us confidence Cologuard 2.0 will get us one step closer to achieving our mission. We're now happy to answer your questions.

[Operator Instructions] Your first question comes from the line of Brian Weinstein with William Blair.

I thought we'd just start with Dr. Kisiel, if I can start with a question with you. Can you just talk about the importance of advanced adenoma or precancer detection here? What that means clinically in practice for you and for clinicians around the world? I mean the difference between a 42% versus a 57%, how meaningful is that? So what does it do to clinical practice?

Yes. Thank you for the question, Brian. The -- that difference is pretty significant and very important within clinical practice. I think, obviously, we focused on emphasizing the performance that we're seeing in cancer detection, but detecting advanced precancerous lesions, those are the polyps that are most likely to progress towards cancer. And screening tests that don't detect precancerous lesions don't prevent cancer from happening. Many of them can maybe make the cancers that they find more treatable, but it's really by detecting precancerous lesions, especially the advanced ones. And by removing those lesions that we actually prevent the cancer from happening in the first place. There are numerous modeling studies, some conducted by us, some conducted by other experts using rigorously validated models that show that actually removing and finding the precancers makes the biggest impact on colorectal cancer mortality and quality of life gained by participating in a screening program. So I think this is very meaningful.

Your next question comes from the line of Catherine Schulte with Baird.

Just after your last 2.0 data presentation at ACG in 2019, and you talked about evaluating whether new markers could increase specificity while maintaining sensitivity, this update kind of showed the reverse. So I guess, as you think about your upcoming algorithm cutoff study for BLUE-C, is the goal to optimize specificity further in case you have some falloff in the pivotal study? Or are you now thinking that sensitivity optimization is preferable from here?

Scott, would you take that?

Yes, sure. Thanks, Kevin. Thanks for the question, Catherine. So the goal for this program is exactly, as Kevin described, to get the specificity threshold above 90%, while not reducing any cancer sensitivity. We think these data clearly suggest that this is possible in a prospective setting and that was the objective -- the sole objective of this study was to demonstrate that result.

Your next question comes from the line of Dan Brennan with Cowen.

So Kevin, in terms of the study design, you prospectively collected the samples for the healthy and the precancer adenoma and then you enlist for cancer. And I'd be interested in terms of this thought about how the data we should expect to progress when we get into the final readout of the pivotal study. I know you've discussed that prospective collection gives you a lot of confidence. So could you give us a sense, though, if we should expect some degradation just given this is still not a truly prospective study and you're going to a much larger sample size? So maybe kind of put on your hat and kind of statistical analysis hat, maybe help us think through what the right way to think about what the study means for the final readout?

Better yet, Dan, we'll let Scott Johnson put on his statistical analysis hat. Thanks, Scott.

Thanks, Kevin. Yes. So great question. And as is common, and I think known to all, anytime you do a case control study, there are limitations relative to what you expect to see at a prospective setting. However, this study was specifically designed to mimic the distribution of stages and sizes of lesions that we saw in DeeP-C, and so that's one way of mitigating against any delta that you would see from a case control to a prospective setting. The other point that I think is really important to highlight here is the extensive cross validation that was done in this analysis. Briefly, cross-validation involves taking all of the data in the study, partitioning it randomly multiple times and revealing the average performance out of that repeated cross-validation analysis. So you have multiple test and training sets. And what you see is that the point estimate or the best fit analysis in this study was very close to the mean of the cross-validation, suggesting that the risk of overfitting largely due to the few number of biomarker inputs in this analysis is likely to hold up in a prospective setting.

And just to be clear, this study was the precancerous polyp for collected prospectively also, which mimics what is occurring in the BLUE-C study and occurred in the DeeP-C study.

Your next question comes from the line of Brandon Couillard with Jefferies.

This is Matt on for Brandon. One for you, Jeff. In the past, I think you've talked about kind of 5% to 10% reduction in cost of goods due to Cologuard 2.0, one of the reasons being fewer markers and lower spoilage rate. And based on kind of what you've seen so far, is that kind of still in the ballpark of the improvement you expect to see on the cost side of the equation?

Yes, Matt. We're still very optimistic about generating not only kind of lab efficiencies, improved customer experience. You touched on some big ones there. The fact that we can increase the -- improve the false positive rate by 3 points means that every time you run Cologuard, and the guidelines recommend every 3 years, every 3 years, we'll keep 3 percentage points more of patients, which not only is the big win for patients, but also helps lift our revenue. On top of that, reducing the expired sample rate, today, 1% to 2% of the samples that come in are expired, reducing that not only improves the patient experience and also helps generate additional revenue and lower our cost of goods. And then Kevin touched on, in his remarks, removing the KRAS mutation from -- that's in Cologuard 1 reduces the -- some of the lab costs and also improves the throughput in our lab pretty significantly. Perhaps the biggest improvement that we expect to see in Cologuard 2.0 is improved adoption. We're already on this rapid rate of adoption but often and even better, more accurate test should help us appeal to more patients than more providers and lift the overall revenue.

Your next question comes from the line of Jack Meehan with Nephron.

My questions are on the design for Cologuard 2.0. So the abstract includes details on the new methylation markers, which are different than the initial Cologuard test. Just talk about the process you went through to hone in on these markers? And scientifically, the rationale for why these would show better detection or specificity than the initial Cologuard?

Yes, sure. I'll start, and John, feel free to jump in. Obviously, this is an extension of a collaboration -- a long-standing collaboration with the Mayo Clinic. High level, there was some sequencing-based discovery of biomarkers that resulted in dozens of candidate biomarkers that could be candidates for this next-generation test. And then subsequently multiple case-control studies to eliminate biomarkers that had lack of specificity for other benign diseases as well as those that were universal across a variety of different lesion types in colorectal cancer. And so essentially, this study that we're presenting today is the culmination of the down selection, as I said, dozens of markers into the 4 methylation markers and hemoglobin that they're in the current configuration.

Yes. Scott, I agree with everything you've said. I think the markers that are in Cologuard 1.0 have come from candidates that were developed several decades ago. But the scientific community, really since the proliferation of next-gen sequencing about a decade to 1.5 decades ago, has really generated far more representative and far more accurate markers of colon cancer. It should also be noted that these markers were also sequenced from samples that included advanced adenomas. So we feel that they are really present throughout the adenoma-carcinoma sequence and probably have a significantly lower background in normal tissue than the current methylated DNA markers and KRAS do. There's also a novel normalizing marker, a normal reference marker in this, which is ZDHHC1, and this is a marker that is methylated in both cancer and normals to give a much more even and consistent background than we currently achieve from beta actin. So that's another new technology in this panel.

Your next question comes from the line of Kyle Mikson with Canaccord Genuity.

Good to see that sensitivity for APL high-grade dysplasia of 83% was higher than DeeP-C, which was 69%, I believe. But that's lower than the same sensitivity in the 2012 validation study, where the metric was 86%, whereas overall APL detection of 57% was identical to this. So are there any reason, I guess, why the metric would be lower in this study? And I guess, do you expect the BLUE-C high-grade dysplasia sensitivity will be lower than DeeP-C? And overall, I mean, how important is the high-grade dysplasia sensitivity from like a clinical utility perspective? We know we have some individuals with medical backgrounds on the line.

Scott?

Yes, sure. Thanks, Kevin. So the 2019 study was a similar case control study. However, the stage distribution was more heavily biased towards a later stage than the early stage. Likewise, the lesion sizes were larger in that study than in this study. This study, and again, at the risk of being redundant, was intended to mimic the distributions by many measures, including demographics of the study population of DeeP-C. And so this is a more representative outcome that we expect to see in the BLUE-C population.

Your next question comes from the line of Puneet Souda with SVB Leerink.

The first question is really on the level of deterioration or degradation, if I may say, on the BLUE-C that you expect. And I know that's been asked a little bit before, but if I could ask about I appreciate the inclusion of average-risk samples and extensive cross-validation you're doing here for the study, and that's really helpful. But we did see a reduction in advanced adenoma data from the ACG 2019 data of 65% to now 57%. And I appreciate this has more prospective data set this time around. So I just wanted to get a sense of why that could be the case. And also, a quick clarification, if I may. Kevin, the 92% no neoplasia, would that be the one on the FDA label eventually as a specificity? Or 89.8% no CRC APL would be on the FDA label and marketed as the specificity of Cologuard?

Yes. So Puneet, I'll try to describe this in layperson's terms. Size does matter. It does -- in the 2019 study, as Scott mentioned, those are -- were larger findings. So they were larger precancers leading to a higher level of sensitivity. Scott, you can clarify that, if I'm incorrect.

Correct. Just to add some details there, Puneet. If you look back at the 2019 precancer data, there was kind of -- if you look at back at that slide, we readjusted the precancer by size. If you adjust the size similar to what we saw on DeeP-C, the precancer sensitivity there was 46%. What we're showing here now is 57% in a very -- the same prospective samples that were reflected in the DeeP-C. So the way I see it, we actually have improved precancer performance today versus what we showed back in 2019.

Yes. And to your question about FDA's specificity in the label, when you include small findings, so nonadvanced adenomas, as a false positive as the FDA required in the label, the comparison is to the 90% specificity. So 87% specificity in DeeP-C would equate to 90% specificity in this study. And the 90% specificity that are in the data tables from the DeeP-C study and the FDA submission would equate to the 92% specificity here. That is excluding those adenomas that are not yet advanced as -- so they would be excluded from the false positives. Does that clarify things?

The next question comes from the line of Alex Nowak with Craig-Hallum Capital.

This is Trenton McCarthy on for Alex. I just had one question here. Have you ran Cologuard 2.0 on your DeeP-C sample database? And if not, do you have a plan to? And what are your thoughts on the potential there to maybe provide a more apples-to-apples comparison to Cologuard 1.0?

Yes. We have not run the DeeP-C specimens as a cohort against the prototype Cologuard 2.0, but a number of these specimens, as highlighted, did -- in this study did come from that collection. And so we -- this is essentially a sampling of those specimens again to represent the total population.

Your next question comes from the line of Mark Massaro with BTIG.

I guess first one is for Kevin. You decided to enroll participants aged 40 and older. The guidelines right now are aged 45 and older. I guess, can you speak to any activity around talking to different groups about trying to lower the screening guidelines to age 40? And then I also had a question for Dr. Kisiel about a number of companies are increasing the size of their enrollments, including DeeP-C. And I think one of the reasons for that is that the prevalence for cancer in the average risk setting has declined. I'd be curious to hear why you think the prevalence of cancer in the average risk community is declining?

So I'll take the first part, and John, you take the second question. The first question, with the -- so age 40, the guideline group, USPSTF lowered the screening age in the summer of 2021 down to 45. And that was based on the growing incidence of cancer in the sub-50 population. Will someday that be lower to age 40? We don't know. Certainly, if it increase in prevalence of the disease continues to increase, there may be a case for that. And in which case, we want to be prepared with Cologuard 2.0 to have been studied in that population so that we don't have to go back and enroll a new study. So we don't know what's going to happen in the future, but we will be prepared for it. John, if you want to take the second question?

Yes, absolutely. So I'll push back a little bit. The number of estimated new cases of colorectal cancer in the United States has not changed appreciably, at least not by more than a few percentage points per year. So the Rebecca Siegel publication, Cancer Statistics, as of their -- the estimates for 2021, the 2022 estimates will be published probably any week now, are showing that about 150,000 new cases are diagnosed each year. And I also want to emphasize that the coronavirus pandemic has been anticipated by NCI to be leading to at least 10,000 additional deaths from reduced screening for colon and breast cancers alone. What I think we're finding is that for studies that require a criterion standard colonoscopy, there may be fewer patients who are now willing to get one. And that's very different than a reduction in the prevalence of cancer in the population.

Ladies and gentlemen, this concludes today's conference call. Thank you for participating. You may now disconnect.