Cost Effectiveness of Gastric Cancer Screening and Prevention – 2022 Gastric Cancer Summit

[00:00] Thanks to the organizers for inviting me. This is a fantastic, fantastic effort and I congratulate all of you. I'm going to introduce Shelja Shah, who is a gastroenterologist and researcher based in San Diego. Her focus is on prevention of gastric cancer.

[00:20] and screening and improving outcomes. But by way of introduction, I want to reminisce a little bit about Julie's paper that Chin mentioned. I was a resident here in the mid-90s, and Howard Hack, who was one of Julie's co-authors, was a GI fellow, and he spoke at noon conference. This is after they gave us her.

[00:40] sandwiches and we're all sitting there eating and he starts talking about oh there's this infection and if we screen people and he's talking about screening millions of people and it's going to cost millions and millions of dollars but we're going to prevent people from dying from stomach cancer. Then we all stop it. What's how we're talking about? It was really really

[01:00] quite thought-provoking at the time, and here we are many years later. I now look back. I had no idea how hard it was to get published in Lancet and to publish a model in Lancet and to a large extent, as Chin said, you had the answer back then and the calls for trials and for implementation. So I'm really looking forward to hearing Shailleh just said.

[01:20] to see what you update us on since 25 years ago. Welcome. Thank you so much for that introduction. It really is wonderful to be here and to really pick up where we left off in March of 2020. And so really want to thank the organizers of course for this.

[01:40] kind invitation and really excited for the next two days. So the title of my talk today will be Cost Effectiveness of Gas or Cancer Screening in the United States. So this is a slide that is familiar to most if not all of you here in the room and one that you'll probably see a couple of times.

[02:00] presented at this conference, but just by way of background, it's quite important to the talk today. And so gastric cancer is the fifth most common cancer worldwide and the fourth leading cause of gastric cancer death, responsible for over a million new cases per year and over 780,000

[02:20] deaths per year. And we know that there's wide global variation with across countries and regions, but also within countries and regions. And even though the United States is considered a low to intermediate risk country, we know that there's actually that this burden

[02:40] proportionately spread across populations. And so this is the most recent SEER data. And so as you can see, even though we consider gastric cancer, next.

[03:00] So even though we consider stomach cancer to be a rare cancer in the US, it's actually responsible for nearly 27,000 new cases and almost 11,000 deaths per year. Next. And the majority of these, which will be the focus of our talks today, the majority

[03:20] are non-cardiac gastric cancers with only a minority being cardiac cancers. Next. Back one more. Back, yep. And so again, even though we've heard that stomach cancer, we think of it as a rare disease.

[03:40] of the talk is about esophageal cancer screening, colorectal cancer screening. We see that stomach cancer is actually responsible for nearly 6,000 more cases per year. And this is interesting because we see here that stomach cancer is ranked 15th most common in the US. But when we think about certain populations,

[04:00] especially non-white populations and immigrant populations. It actually ranks as one of the top five cancers in these groups. Next. Next. And so this was a population-based analysis trying to quantify that disproportionate burden of gastric cancer among the 10 most-

[04:20] of populist ethnic groups, racial and ethnic groups in the United States. And so with the focus on non-cardiac cancer, you can see that in all groups compared to non-Hispanic whites, that there is a two-fold to up to over 13-fold higher risk of non-cardiac gastric cancer in the United States.

[04:40] these groups. And even in some of these groups, the rates actually approach colorectal cancer, which is again obviously another cancer that we screen for. And these data are representative of a screening age population. And we look both at the age 50 as well as at the new recommendations age 45 based on the

[05:00] more recent colorectal cancer screening guidelines. Next. And so we also see disparity in terms of mortality and really mortality between the United States and other countries where screening does occur. And so as we all know, the

[05:20] The prognosis with a gastric cancer diagnosis is dismal. And that really is because of the advanced stage of most of these diagnoses and the absence of screening. Since most of these cancers are picked up once symptoms present and prompt a diagnostic endoscopy. Next. And this is reflected again in that the minority of cases

[05:40] are actually diagnosed in that localized state. And so 28% are diagnosed as local gastric cancers with an even smaller percentage, so about 15% diagnosed as early gastric cancers, which means that it's prior to submucosal invasion and that's the stage at which —

[06:00] or surgical resection is actually curative. Next. And this is in stark comparison to Asian countries where screening does occur. And in these countries, we actually see that with implementation of screening, it used to be that 15 to 20% were diagnosed in that early stage, but with implementation,

[06:20] implementation of screening, now about 50 to 60% and potentially more of gastrocancers are actually diagnosed in that early stage. And this is translated into lower gastrocancer mortality. Next. Next. And so now I'll briefly very, very

[06:40] Very briefly, pivot to the pathogenesis because this is really important to our discussion of cost-effectiveness of gastric cancer screening as well as surveillance and discussion of the models. And so non-cardiogastric adenocarcinoma, there's two intrinsic main histologic

[07:00] types, intestinal type as well as diffuse type. And our focus is on the intestinal type, which is the more common histologic type. And this develops as a stepwise progression from gastritis to atrophic gastritis, intestinal metoplasia, and in a small proportion of patients progresses to dysplasia.

[07:20] and in an even smaller proportion to malignant transformation. And next. The primary trigger for this cascade is H. pylori infection, which propagates this forward, along with the interaction of other host microbial and environmental factors.

[07:40] One very important thing though to really appreciate and recognize is that even in the absence of H. pylori, even though H. pylori is a trigger, even in the absence of H. pylori infection, once intestinal metoplesia develops, this cascade can still progress because of those other factors that are

[08:00] unfortunately still poorly understood. But that actually underlies the limitation of simply doing H. pylori testing, screening, and eradication, because patients that already have developed gastric pre-cancerous changes can still have that progression. And so that really underlies the importance of surveillance.

[08:20] patients. Next. And so with that background, so this is unfortunately an animation build which is going to be a little bit challenging, but we'll work through it. So with that background, if we take a step back and look to the WHO for

[08:40] screening criteria, we see that gastric cancer actually meets all of these or most if not all of these. So next. So the condition is important. Next. You can just keep clicking. Next. Okay. Stop there. So we know we have an understanding of the natural course of the condition.

[09:00] There is a recognizable latent or early symptomatic stage. We would accept that endoscopy is a suitable test and acceptable to the population. There is accepted treatment for patients with recognized disease. We do have the facilities for diagnosis and treatment available in the United States.

[09:20] Then this last point is an important point in the topic of our discussion. So whether or not the costs of case finding, so including the diagnosis and treatment, are economically balanced in relation to the possible expenditures on medical care as a whole. So with this it really is the consideration of

[09:40] How can we optimize the population that we're screening as well as minimize the economic burden, and that will allow these interventions to be cost-effective. Next. And so this has been looked at. So Dr. Herr gave us a nice overview of studies that have been done in the United States, and I pulled some of the information

[10:00] ones that have focused on endoscopic screening and really for that purpose that I mentioned of the potential progression of gastric pre-cancer even in the absence of H. pylori infection. And so these are two studies from the US. So the first one, next.

[10:20] by Gupta et al. in 2011. So this was done in a general US population, age 50, and where they built their model looking at upper GI cancers, including gastric as well as esophageal. And they modeled, oh, and then, so the second one is the IEA et al. study from 2016.

[10:40] 16, which here they looked only at men and at different ages and also modeled three different interventions. So I'll spend a little bit of time just talking through these studies. So in the Gupta et al. study, they looked at two screening modalities, so EGD during colonoscopy.

[11:00] at the age of 50, as well as then surveillance, but only if Barrett's esophagus was actually diagnosed. And they commented that the limitation was that they didn't actually model gastric intestinal metoplagia surveillance. So in this study, neither one was cost-effective, but if they added in

[11:20] bare esophagus surveillance, then it did slightly dip the udilobat willingness to patient threshold of 100,000. So the next study, the YEA-at-all study, so they modeled three different interventions that we heard about. And so seronepsinogen with reflex endoscopy if it was positive, starting at the age of

[11:40] 50, ECD alone starting at the age of 50, and then H. pylori screening starting at the age of 20. And none of these interventions were cost-effective in and of themselves. But like was mentioned, they did look at stratification based on smoking status. And so here, serum

[12:00] was actually cost-effective but only in smokers. And so, pepcinogen in the United States, as we all know, is not available clinically. And unfortunately, studies have not actually demonstrated in US populations, including high-risk US populations, appropriate sensitivity and specificity.

[12:20] It has demonstrated good test characteristics in other high risk populations, so certainly still could be a possibility. But one other thing to note about this study is that at least in nonsmokers, pepsynogen had to basically achieve perfect test characteristics to actually

[12:40] make the bottle and even then wasn't cost effective. And in smokers, the test characteristics were a little bit better, but sensitivity of 60% and specificity of 94%. Next. And so one of the key things in both these studies though, and also hinging back on.

[13:00] disparity data that I presented earlier is that neither study accounted for the differential risk of gastroprenia plastic changes according to race and ethnicity and neither took into account gastrointestinal metaplegia surveillance. Next. And so this led us to hypothesize that

[13:20] race and ethnicity might actually be one way to identify high-risk populations. And so we developed a Markov model where the base case was a 50 year old individual undergoing colonoscopy for routine colorectal cancer screening and looked at this for non-Hispanic women.

[13:40] non-Hispanic black, Hispanics and Asians. And I will mention that we also did look at 45-year-olds as well and same outcomes. And so then this base case modeled the base case and could undergo one of three interventions. So EGD and mapping biopsies with non-Hispanic white and Hispanic black and Asian white.

[14:00] continued surveillance of intestinal metoplasia every three years if it was diagnosed based on biopsies. EGD plus mapping biopsies every two years even if it was normal and that was actually based on current screening guidelines in Japan and Korea and then current standard of care in the US which is no endoscopic

[14:20] screening. Then patients would fall into Markov model based on their probabilities according to their race and ethnicity and then transition probabilities again based on race and ethnicity. Next. So these transition probabilities cause

[14:40] as well as complications, were all taken from published literature. And we did also model the other probabilities, such as the probability of a missed neoplastic lesion, probability of the availability of ESD versus gastrectomy, along with some other things.

[15:00] see here with these transition probabilities that somebody, if they underwent upper endoscopy and were found to have normal gastropulae mucosa, including on biopsies, that then there would be no follow-up screening unless they were in that second category. But anybody diagnosed with intestinal metoplasia

[15:20] or more advanced lesions would fall into a surveillance or treatment categorization. So patients with intestinal metoplasia would undergo endoscopy every three years. If there was resectable dysplasia or neoplasia, then they would undergo resection with appropriate follow-up. And so this was then modeled over a 30-year time.

[15:40] horizon. Next. Next. And so based on this model, so we found that gastric cancer screening with upper endoscopy and mapping biopsies with continued surveillance of gastric intestinal metoplasia was cost-effective.

[16:00] But for Asians, Hispanics, and non-Hispanic blacks, but not for non-Hispanic white populations. And we were also interested, again, based on the disparity data that I presented about the differential risk in Asian American groups, we were also interested to see if it would be more cost effective in.

[16:20] other populations. And so we were also interested in Hispanic Americans as well, but unfortunately there weren't enough granular data to develop the model. But we did look at Asian Americans specifically. Next. And so this was the same exact model, but looking

[16:40] at the six most populous Asian American groups in the United States and took the same approach, but instead now the data inputs relied on the estimates for each of these populations. Next. Next.

[17:00] next. And so you can see that it was actually cost-effective for all groups, all Asian American groups, next. But it was particularly more cost-effective for the highest risk Asian American groups in the United States. So Japanese American, Chinese American, Korean American, as well as Vietnamese American.

[17:20] populations. Next. And so this, as I mentioned, we, you know, we were limited to Asian American populations, but really plausibly this same model could apply to high risk immigrant groups who are immigrating to a low risk country, which the United States would be considered low intermediate.

[17:40] risk. And so plausibly the same model would apply to these high risk groups. And we hypothesized this based on a recent meta-analysis that was published that did in fact demonstrate that immigrants do retain that elevated incidence as well as mortality. Next.

[18:00] But to actually quantify this and to support this, we did do several sensitivity analyses with the Markov model. And so we did this to identify factors with the greatest impact on cost effectiveness, as well as to determine the influence of various assumptions on the estimates for all variables that were used.

[18:20] model. And so you can see here that the model was particularly sensitive to the baseline prevalence of intestinal metoplasia, as well as those transition probabilities, so from intestinal metoplasia to dysplasia and further, as well as the cost of the interventions. But the greatest cost-saving in the model was actually

[18:40] the detection of early gastric cancer so that opportunity for cure and to actually prevent gastric cancer death, that that was the most important cost benefit. So didn't do anything for incidence, but really dramatically reduced the mortality. Next.

[19:00] And so when we think of baseline prevalence of gastric and teslometoplasia in the United States, if we look at the largest study that has been published, which was based on a pathology database of nearly 900,000 people in the US undergoing EGD with biopsy, we could see that the prevalence of gastric and teslometoplasis

[19:20] and tesla metoplasia overall is about 5% of the US population, which is decently high, especially when we consider the economic burden that can potentially be incurred by endoscopic surveillance for everybody. But what's important to note next is that the prevalence of gastrocodilators

[19:40] and tessel metoplasia varies substantially according to the population. So there's higher risk populations where the prevalence is on the order of about 20% and potentially even higher. And so this of course includes non-white racial and ethnic groups as well as immigrant groups and those with a family history of gastric cancer.

[20:00] Next. But as I mentioned, with 5% of the population potentially affected, that's where those transition probabilities come into place and so really identifying who is at highest risk of progression. And so based on a large systematic review and meta-analysis, the cumulative risk of

[20:20] gastric cancer in people with intestinal metoplesia is, as you can see here, about 0.4% at 3 years, about 1.6% at 10 years. Next. And so these rates are actually very similar to what we see for non-displastic barrets, as well as low-risk colorectal

[20:40] adenomas, again, conditions that we routinely survey for without question. But these transition probabilities that you can see in that green box are really the baseline risks of progression. And so there's factors that are associated with higher versus lower risk of progression.

[21:00] And so you can see in the other box that there's certain other modifying factors, including whether or not there's persistent H. pylori infection, family history, the severity of those preneoplastic changes, incomplete subtype or extensive intestinal metoplesia.

[21:20] And the presence of any of those risk factors can increase the risk of progression even up to four and higher fold. Next. Next. And so when we think about how we can really refine our cost effectiveness models, certainly we can add in some of those factors.

[21:40] but they're imprecise. And unfortunately, we did just finish another model looking at those factors, and unfortunately, didn't have them in time to present here today. But still, we see that people with limited intestinal metoplasia still do progress, and those with extensive intestinal metoplasia might not progress.

[22:00] Ideally, we can move more to more precise means to identify people who are at high risk and ideally really move away from race and ethnicity, which we know are surrogate risk factors. Things like molecular biomarkers as well as genetic factors. Next.

[22:20] Next. And so we've talked about, you know, targeted screening as well as surveillance and how we can optimize some of these cost-effectiveness models. But really the elephant in the room is that the success of systemic gastrocancer screening really does hinge on. Next.

[22:40] Both the uptake by the intended population as well as not missing lesions. And so uptake of the intended intervention, we've learned quite a bit from colorectal cancer screening and uptake and difficulties there because there certainly are relevant cultural barriers and other barriers there.

[23:00] there. But the other thing is not missing lesions. Next. And so as I mentioned in our model we did actually vary the probability of missing neoplastic lesions. And that's important because the purpose of these exams, sure, we can, you know, stage intestinal metoplesia and get a

[23:20] idea of gastric pre-neoplasia and risk stratification that way. But really the purpose is to identify early neoplasia so that you do have that opportunity for curative resection. And so unfortunately, even in expert hands, the endoscopic misrate for these lesions, which are quite subtle, is on average about 10 to 11

[23:40] but reports are even demonstrated higher, potentially up to 25%. And so in the US, that's particularly problematic since as gastroenterologists, it's not something that compared to colorectal polyps and lesions that we're really trained for. So certainly a lot of opportunity there as well.

[24:00] Next. Next. Next. So in the US, so some take-home points. So in the US, there are marked differences in gastric cancer incidents based on race, ethnicity, and immigrant status. And this pool of at-risk individuals is only growing.

[24:20] So about as of 2018, which again, still growing, about 30% of the US population consists of foreign born immigrants and the US born offspring. And most are from regions of high gastric cancer incidence and mortality. And as I showed, these do these populations do retain that elevated risk.

[24:40] Next. Next. Next. And so, endoscopy for gastric cancer screening at the time of colonoscopy for average risk colorectal cancer screening is potentially a cost-effective intervention for certain populations, including non-Hispanic black.

[25:00] Hispanics, Asian Americans, and likely other immigrant groups from those high-risk countries. And this really does support a model of targeted opportunistic screening for these high-risk populations. Next. But unfortunately, we lack—oh, next. Unfortunately, we lack robust estimates.

[25:20] for progression. And as I mentioned, it really is that transition probability that drives some of these models. And so this is a major knowledge gap. Next. And so ideally, we'll have further risk stratification models that include validated biomarkers to enrich the population that we're screening for.

[25:40] Next. Next. But one of the biggest things that we have to do is really understand that certainly we can model this, but then also that human element of making sure that we're doing adequate endoscopic exams and really appreciating the subtleness of these neoplastic lesions, because really it's

[26:00] that early cancer detection and the opportunity for curative resection that's a driver of these cost savings. Next. And that's it. Thank you.