Camera pills: They’re not just for GI anymore
Volume 8, Issue 22 | November 30, 2023
In This Issue
Could these be the markers for Long COVID diagnosis?
AI-based Dx system achieves 100% accuracy
Getting people to screen is one thing. Getting them to follow up is quite another
New and Noteworthy
Long COVID diagnosis: Light at the end of the tunnel?
Our September 28 newsletter highlighted how long-lasting immune-system dysfunction generates the persistent symptoms of Long COVID. A recent preprint supports this conclusion and narrows the target to persistent over-activation of the very earliest stages of the innate immune response – complement-cascade-driven inflammation.
The authors found that an immunoassay test for just four of the >50 complement-associated proteins would have a predictive power of 78.5% in differentiating Long COVID from other diseases or syndromes. Those four proteins may be the best place to focus treatments (drugs that suppress elements of this cascade are already available).
COMMENTARY: The demand for a simple and effective Long-COVID diagnostic is urgent. On average, 45% of COVID survivors report at least one persistent and debilitating symptom at three months post infection.
Beyond fatigue and shortness of breath, symptoms are quite varied, suggesting that Long COVID may not be a single condition. However, attempts to subdivide cases based on clusters of symptoms have been unconvincing so far (e.g., one AI-based attempt).
The immune response reaches into every organ system and is a likely place to look for the root causes of this syndrome. On the side of caution, however, we have to note:
This study is relatively small (166 had Long COVID, 79 didn’t).
Complement activity is not specific to Long COVID.
The complement levels in cases had substantial overlap with those of controls (see Figure 4B above).
Trials of existing complement-inhibiting drugs have been disappointing (although several of the best candidates have not yet been evaluated).
But don’t despair. The good and bad news is that we are still relatively early in the exploration of Long COVID. New insights are being generated weekly. We believe that this nut will be cracked.
Camera Pills: Literally and Figuratively Easy to Swallow
Cameras in the form of swallowable pills have been used for the last 20 years to diagnose and monitor GI-related diseases. Called capsule endoscopy, the use of these tools helps to identify bleeding, polyps, varices, tumors, and other abnormalities in the GI tract. We can debate whether this sort of procedure counts as non-invasive - but it is certainly less invasive than an endoscopy, colonoscopy, or full-on surgery.
New research shows that a camera pill the size of a multivitamin can detect not only what is happening in the digestive tract, but monitor heart and lung function, as well. In this small study, a digital pill was used to monitor cardiac and respiratory function in sleep-apnea patients. (Sleep apnea affects 17% of women and 33% of men in the US.) This technology could also be used to identify respiratory distress in patients at high risk for drug overdose.
COMMENTARY: We are bullish on this technology and look forward to larger studies. We are also intrigued about the possibilities. If one section of a diagnostic pill detects pathology, could another chamber in the pill then open up and provide treatment?
On differential Dx for MIS-C vs. typhus, AI bats 1.000
One of the central challenges in medicine is differential diagnosis: differentiating between diseases / conditions that have similar symptoms but very different treatments. AI is increasingly being tested as a way to work through this issue.
One particularly vexing challenge that has arisen in recent years is differentiating multisystem inflammatory syndrome in children (MIS-C), a rare result of a COVID infection, from other inflammatory or acute diseases. A new study looked at the use of AI to distinguish the syndrome from typhus, a flea-borne tropical disease that’s common in Texas and other areas in the US.
Researchers used an AI system to identify 30 demographic, clinical, and laboratory features that together can be used to differentiate typhus from MIS-C. It’s a two-part process: In the first part, clinicians use 17 of those features to manually calculate a score. If the score is high enough, the patient has typhus. If it’s low enough, the kid has MIS-C. If it’s somewhere in the middle, the clinician moves to phase 2, which uses AI and all 30 features to make the diagnosis.
Historically, typhus takes days to diagnose. Using this system, clinicians can get their answer within six hours - and half the time, they don’t need to consult the software at all, they can just use an old-fashioned questionnaire.
The kicker? Using the system, clinicians diagnosed all 220 kids in the study with 100% accuracy. Damn.
Food for Thought
Home testing might get people to screen for disease.
Getting them to follow up on the results is a lot tougher.
We’ve talked a lot about the cost/benefit analysis that comes with screening testing. Is the test sensitive and specific enough, with a high enough positive predictive value, to be worth the cost of doing it? Are there risks involved with the test itself? Are there costs - both monetary and other - associated with false positives?
Today we’re going to look at the other side of the coin: Let’s say that the screening test IS worth it. Can we get people to take it? And once they’ve taken it, can we get them to follow up appropriately if they get an abnormal result?
Three recent studies provide potential answers to these questions- and those answers aren’t always the ones you might expect.
Let’s start with Step 1: Getting folks to take the test. (Spoiler alert: You probably know the answer to this one.) A study published in JAMA this week looked at screening for cervical cancer. It found that sending a self-sample-collection kit for a human papillomavirus (HPV) test to folks who were due or overdue for screening made them more likely to get tested than just sending them educational materials. Giving them the option to take samples at home (but not actually sending the kit) also worked better than just education, but not as well as sending the kit. So far, so good: The easier you make it for people to get tested, the more likely they are to test.
But screening itself is only the first step of the process. Once people have test results, getting folks to act is necessary. And that’s where screening testing, whether at-home or in-clinic, runs into a snag.
The HPV study above didn’t follow patients past the screening process, but studies on HIV have done so. And according to one recent meta-analysis of eight studies (all randomized controlled trials or close to it), self-testing for HIV generally didn’t change the likelihood that people would go to a health-care provider or get treated if they had HIV.
So what DOES get people to follow up on abnormal screening test results? If the results of a third recent study are any indication, not much. This one looked at follow-up to screening testing for several of the most common cancers (colorectal, cervical, breast, lung). It found that after someone received an abnormal screening result, the more reminders and help they got about doing follow-up testing, the more likely they were to do those follow-up tests (but most of them still didn't follow up).
Electronic health record (EHR) reminders only:
22.7% did follow-up testingEHR reminders + a phone call reminder: 31%
EHR reminders + phone call reminder + help getting to the test if needed: 31.4%
Those numbers suck, huh? A caveat: the HPV and multi-cancer studies were done during the pandemic, so that could skew results. But still - even with practitioners practically begging them to follow up on abnormal test results, more than ⅔ of people didn’t do so.
COMMENTARY: What’s the take-home here? As exciting as the prospect of increased self-testing may be, we have to remember that for screening testing, that first test is only the first step. It’s HARD to get people to follow up on abnormal test results. So test manufacturers who are working on at-home tests need to think not just about how to develop tests that work. They need to think about how they can make follow-up so automatic that testers can’t help but do it.
(Disclosure: Mara is on the board of an HIV self-test manufacturer.)
Quick Hits
The US Department of Education and HHS’s Administration for Strategic Preparedness and Response (ASPR) announced yesterday that they are making millions of free OTC COVID antigen tests available for schools to order. Any school system, traditional or charter, can order tests and distribute them to students, staff, parents, or their local communities.
A report issued today showed that data on respiratory diseases from the National Wastewater Surveillance System (NWSS) Centers of Excellence correlated well with clinical surveillance and in some cases was able to provide an early heads-up about disease surges. The system has sites in Denver, Houston, California, and Wisconsin - the latter’s wastewater analysis showed a spike in flu and RSV weeks before emergency departments saw the cases.
EUA Update
The FDA issued no new 510(k) premarket notifications, one new EUA, two amendments to existing EUAs, and no new revocations in November. Data is available at TestingCommons.com
510(k) Premarket Notifications: 0
New EUAs (1):
COVID + Flu Panel: (1): Roche cobas SARS-CoV-2 & Influenza A/B v2
Amendments to Existing EUA’s (2):
COVID Antigen: 2
Revocations: 0