Glaucoma Checks in Primary Care; Wastewater to Screen for Colorectal Cancer

TTHealthWatch is a weekly podcast from Texas Tech. In it, Elizabeth Tracey, director of electronic media for Johns Hopkins Medicine in Baltimore, and Rick Lange, MD, president of Texas Tech Health El Paso, look at the top medical stories of the week.

This week’s topics include brain architecture and diet, use of wastewater to screen for colorectal cancer, new cholesterol management guidelines, and artificial intelligence (AI)-based screening for glaucoma in a primary care setting.

Program notes:

0:35 New cholesterol-lowering guidelines

1:35 Risk scores new

2:35 Mainstay is behavioral change

3:35 For those with a longer view

4:00 AI-based glaucoma screening in primary care settings

5:00 Either AI or pressure

6:00 Fundus photograph and intraocular pressure

7:00 Sensitivity and specificity

7:50 Diet and brain structure

8:50 Brain imaging over 10 or 15 years

9:50 Diet that’s been advocated for a long time

10:55 Surveying wastewater for colorectal cancer markers

11:55 CDH1 marker measurement

12:55 Shows promise for intervention

13:57 End

Transcript:

Elizabeth: Can we use wastewater to survey for colorectal cancer?

Rick: New guidelines on the management of cholesterol.

Elizabeth: Using AI to screen for glaucoma in a primary care setting.

Rick: And can your diet affect your brain function?

Elizabeth: That’s what we’re talking about this week on TTHealthWatch, your weekly look at the medical headlines from Texas Tech University Health Sciences Center in El Paso. I’m Elizabeth Tracey, a Baltimore-based medical journalist.

Rick: And I’m Rick Lange, president of Texas Tech Health El Paso.

Elizabeth: Rick, let’s turn right to Circulation. This has been getting a lot of press — modification of cholesterol-lowering guidelines.

Rick: And Elizabeth, this is a very extensive evaluation of what we need to be doing with regard to cholesterol and lipid management. Gosh, it’s about 150 pages long, so I’m just going to summarize some of the major changes that have occurred since the last recommendation.

The first is that they recommend that we treat dyslipidemia earlier to reduce lifelong risk of prolonged exposure. Why wait until you’re 60 or 70? Let’s begin to do health counseling very early on. And then if people have high cholesterol at a young age, let’s begin to treat that earlier.

Secondly, we’re going to use a different equation. It’s called the PREVENT, Predicting Risk of Cardiovascular Disease EVENTs, equation instead of the older one called the Pooled Cohort equation. They assessed both 10- and 30-year risk. We’ll do that in adults age 30 to about age 80.

There are some risks that aren’t included in that, and we can add another, something called the CPR model. We’re going to talk about a 10-year risk of between 3% and 5% of having cardiovascular disease, as being borderline risk. Those with a risk of 5% to 10% are considered intermediate — we certainly want to begin to treat those — and those that are a higher risk.

We’ve changed the goals a bit. We used to try to lower the LDL cholesterol below 100, and now we want to get it below 70 in anybody that has any risk, and in those that have high risk below 55.

We’re going to talk a lot more in the future about not just measuring LDL cholesterol, but something called ApoB, because it identifies adults that have elevated risk, even though their LDL appears to be normal, and it underestimates the risk. We’re going to talk about ApoB [apolipoprotein B], which includes both LDL and some other high-triglyceride lipoproteins.

Elizabeth: Okay. So this is going to put a whole lot more people into the category of taking a daily med to lower their cholesterol. There must be estimates of that.

Rick: It will affect literally millions of individuals. The mainstay of therapy still remains, first of all, behavioral changes. If pharmacotherapy is required, [we use] statins, and those medications cost pennies per day. They’re effective. Now, in many places around the world where the diet is better, these LDL cholesterols are already low, and so they have a lower risk of cardiovascular disease than we do in the United States. So we’re addressing partly our behavior and partly our diet.

Elizabeth: Talk to me about your own practice. When you have somebody who comes in there with that 10-year risk of 5%, what are you going to do?

Rick: I sit down and I have a conversation with them. First of all, lowering LDL cholesterol by addressing lifestyle changes, and if it doesn’t get down far enough, using pharmacotherapy. And I let the patient be involved in that decision-making process very early on.

Elizabeth: It’s going to be interesting because, as we’re both aware, it’s been a challenge to implement, for example, hypertension guidelines in people. They tend to be pretty resistant. I’m wondering how resistant they’re going to be to this.

Rick: There will always be some people that are resistant. But for those that want to take a longer look, and especially those that have some risk of cardiovascular disease already, [like] a family history. Anybody with diabetes is already at high risk. People with chronic kidney disease are already at high risk, and those are individuals we should target as well. It means we need to arm them with the information they can use to make an informed decision.

Elizabeth: Let’s turn now to something that also has the potential to impact millions of people. That’s the implementation of an AI-based glaucoma screening in primary care settings. That’s in The Lancet.

It turns out glaucoma is a very prominent cause of blindness and often goes undiagnosed. This study decided they were going to evaluate whether an AI-based glaucoma screening is clinically and economically feasible within a publicly funded primary care setting, and this was in Portugal.

They used individual-level participant data from a single primary care screening facility in Lisbon, Portugal. And these were participants 55 to 65 years of age with and without diabetes who were randomly selected from this population. And they were already taking a look at diabetic retinopathy, so the infrastructure was in place. They looked at fundus photographs using an AI algorithm to generate a glaucoma risk score, and they referred for specialist evaluation by either a positive AI result or an intraocular pressure of 25 mmHg or higher.

What they found when they drilled down — out of over 1,000 invited individuals, they had 671 who attended screening. They diagnosed glaucoma in 40 of these 629 participants who were included in this analysis. Of that number, the AI algorithm referred 66 participants, compared with 118 referrals through adjudicated expert assessment. The AI algorithm sensitivity was 78%, its specificity 95%. What they showed was that there is a huge cost-effectiveness relative to employing this method, even when they only have a 1% prevalence of glaucoma.

Rick: There are some things about this I thought were really interesting. It’s tacked on to diabetic retinopathy screening. As you mentioned, they did two things. One is they took a picture of the eye and they used these AI algorithms to assess whether someone had glaucoma. They also measured intraocular pressure. That’s not something that most primary care physicians do. A sensitivity of 78% means that you’re missing about a fourth of the individuals that actually have glaucoma. That only works out well if people are going to have repeat measurements, because glaucoma causes blindness over a long period of time. So this only works if you’re going to repetitively screen these individuals, so hopefully the sensitivity becomes better. The specificity of 90%+ means there are few false positives.

This is a proof of concept. The authors say, “Well, it’s feasible.” There are additional studies that need to be performed to say, “Is it scalable? Is it reproducible? How are we going to deal with the individuals in whom the initial screening doesn’t catch it?” You don’t want someone saying, “Well, I don’t have glaucoma. I don’t need to be screened again.” They need to make sure that they have annual screenings.

Elizabeth: When you compare that, however, with the human evaluations, it, I think, performs really pretty well. And also, the AI system refers people who are more suitable for referral than the human graders did.

Rick: They missed a lot of cases. It needs to be reproduced, but it does say it’s at least feasible. And really, whether it’s scalable or not, that’s a whole different issue.

Elizabeth: We’ll see, I guess, about that because people are already being evaluated for diabetic retinopathy, so using that infrastructure seems pretty persuasive to me. I’ll just cite this one statistic they provide, which is that human graders referred nearly twice as many participants as the AI system, with substantially more false-positive referrals.

Rick: Right. You have to ask yourself, “Do you want to have more false positives — it needs additional testing, i.e. intraocular pressure — or do you want to have more false negatives, where you actually miss the diagnosis?”

Elizabeth: We’re going to see more about this, I suspect. Let us turn to The BMJ.

Rick: Does diet affect brain function? I should have said brain structure, actually, because we know that there’s an association between diet and neurodegenerative diseases. Individuals that are on a Mediterranean-type diet, especially if it’s combined with other things that stop hypertension — this is called the MIND diet, the Mediterranean-Dietary Approaches to Stop Hypertension Intervention for Neurodegenerative Delay. Wow, that’s the MIND diet.

They combined some Mediterranean dietary approaches with the DASH diet. And we know that it’s associated with better brain aging, less neurodegenerative disease, such as Alzheimer’s and Parkinson’s disease. But what nobody has done before is on a longitudinal basis associate that with brain structural changes.

So in this particular study, they took over 1,600 middle-aged and older individuals. These are from the original Framingham study. They calculated their MIND diet score — how well did they adhere to this? — and then they did brain imaging over the course of about 10 or 15 years.

Over the follow-up of almost 12½ years, greater adherence to the MIND diet was associated with a slower decline in gray matter volume and a slower increase in ventricular volumes. Ventricular volumes represent brain shrinkage. And they can actually quantitate that. Each 3-unit increase in the MIND diet corresponded to a 20% attenuation in age-related change and that was equivalent to about 2½ years of reduced brain aging during the 12½-year follow-up. That’s pretty impressive.

Elizabeth: Let’s talk some more about the diet.

Rick: It is primarily a Mediterranean diet that also includes features to reduce hypertension. And so it’s low sodium, high potassium, high in fresh fruits, vegetables, whole grains, olive oils, reduced saturated fats, reduced processed foods, reduced red meat and fatty meats as well. So, a diet that we’ve been advocating for a long period of time because we know it reduces the risk of cardiovascular disease, reduces the risk of hypertension, diabetes, chronic kidney disease, and now it’s been shown to actually directly affect brain structure.

Elizabeth: Yeah. I’m really interested in this intersection between brain structure and cognitive decline. I guess I’d like to see that specifically evaluated also.

Rick: It’s been clear that this diet is associated with a reduced incidence of neurodegenerative disease, including Alzheimer’s and Parkinson’s. They’ve just never had the structural longitudinal follow-up and that’s what this does.

There’s biologic plausibility because this diet is rich in antioxidants. It’s got high-quality protein sources. It’s thought that this reduces the oxidative stress and it mitigates the neuronal damage that occurs with aging. If you compare that to fast and fried foods, they’re unhealthy. They have advanced glycation end products and they contribute to inflammation and vascular damage. There’s a definite improvement in neural function and now the associated anatomic improvements.

Elizabeth: Remaining in The BMJ then, let’s take a look at this very novel notion, to me, of surveying wastewater for population-level colorectal cancer surveillance. We, of course, have been surveying wastewater now that’s come into great prominence with COVID and has been used subsequently for lots of other infectious diseases. As far as I know, this is the first time I’ve seen it looking for cancer markers. As we’re both aware, colorectal cancer [is] increasing in incidence quite a lot and concerningly in those younger than age 50. Is there a way that we could sort of figure this out? And maybe this is going to be one tool that will help to do that.

These authors looked at feasibility data demonstrating the detection of colorectal cancer-associated RNA biomarkers in community wastewater. They looked at RNA expression values for one housekeeping marker and for a colorectal cancer-associated marker called CDH1. They used droplet digital PCR to do this and they looked at 4 neighborhood clusters. It basically showed that they are able to identify the shedding of these markers into wastewater and they can drill that down into neighborhoods and say, “Wow, this neighborhood actually has more folks who could potentially have colorectal cancer.”

Rick: This is fascinating. As you mentioned, it does represent the first use of specific human RNA biomarkers to detect colorectal cancer in wastewater.

There are two possible outcomes. One is to examine why, in that particular area, would colorectal cancer be more prevalent. But secondly, because it’s appearing earlier and earlier, the age at which screening is proposed has gone down from 65 to 50 to now age 45. If we can target specific neighborhoods where we really need to emphasize early screening, that would be particularly helpful.

This shows some promise. It’s not definitive, but preliminary data suggests that once we identify certain areas, we can target those residents to make sure that they’re involved with early screening.

Elizabeth: Yep, and a sobering statistic in here. Right now in the United States, colorectal cancers are the third most common cancer and the second leading cause of cancer-related deaths. So if these shed biomarkers in the wastewater system help us to pinpoint some of the drivers behind that or to identify folks early, that would be a really good thing.

Rick: I would never have thought about using wastewater for doing geolocation of people at high risk or that have colorectal cancer.

Elizabeth: On that note then, that’s a look at this week’s medical headlines from Texas Tech.

I’m Elizabeth Tracey.

Rick: And I’m Rick Lange. Y’all listen up and make healthy choices.

Please enable JavaScript to view the comments powered by Disqus.