In This Issue
Joint FDA / CMS statement on LDTs
Early pancreatic cancer Dx: A step forward
COVID’s post-holiday bump
What the increase in under-50 cancer is telling us
New and Noteworthy
FDA / CMS joint statement on diagnostic tests (really on LDTs)
Today, the FDA and the Centers for Medicare and Medicaid Services (CMS) issued a joint statement calling for “accurate and reliable diagnostic tests, wherever they are made.” The statement underscores the critical role played by diagnostics. Its headline comprises the two agencies’ rationale for phasing in FDA oversight of laboratory-developed tests (LDTs), a plan which was officially announced this past fall.
To date, the FDA has exercised enforcement discretion over LDTs. In this statement, the authors emphasize that “It is time to reconsider this approach.” Their central point is that “LDTs play an important role in healthcare, but when they perform poorly or are not supported by science, they put patients at risk.” The authors then outline the agencies’ concerns with LDTs, including over- and under-diagnosis as well as incorrect diagnoses.
Commentary: We will restate our thoughts.
First: It is important that everyone in the healthcare arena acknowledges and recognizes the critically important role of diagnostics, especially LDTs, for patients today. It is the glue that holds the whole system together. Without accurate, effective, and correctly used tests, treatment intervention will be neither appropriate for patients nor cost-effective. This has always been true, but it’s even more true (if that is possible) with the wider use of genomic biomarkers for cancer and other diseases.
Second: We believe that the FDA should oversee LDTs, but not with a cookie-cutter approach. Diagnostics have unique needs, different from those of other devices regulated by the FDA’s Center for Devices and Radiological Health, and thus should have their own regulatory process.
If the proposed plan goes forward, we are also concerned that the FDA will not be able to move fast enough to evaluate new, innovative tests. And that is the easy part. The truly gigantic task is reviewing existing tests. Though the proposed plan gives the agency four years and potential outside agency help, will it be able work through the backlog in that time? The initial estimate was 80,000 tests from 1,200 labs, but that’s likely to be an underestimate. The devil is in the details here, because the potential for disruption of patient care is huge.
We wonder if a compromise might be found. Could the FDA exempt currently marketed tests that are already being broadly used effectively? Might there be some way to, at a minimum, exempt these well-established tests - tests that are already in medical-association guidelines? If the FDA has data about systematic problems with some tests, then they should focus on those tests and not re-review products that are being used effectively for patient care today. It is critical, however, to ensure that any review acknowledges the difference between screening and diagnostic tests.
Diagnosing pancreatic cancer - light at the end of the tunnel?
Pancreatic cancer is a bear. Despite progress on the surgical side (a surgery professor once summarized the conventional wisdom as “you touch the pancreas, you go straight to hell”), twin challenges remain: limited treatments and late diagnosis. Recently released ACS statistics (extract for pancreatic cancer shown above) quantify the problem.
One of the main issues that has blocked early diagnosis is the fact that traditional imaging has not been sensitive enough to spot small, localized tumors - they just look too much like normal pancreatic tissue. An AI-enhanced CT technique from the Mayo Clinic has demonstrated improved success with these earlier tumors - it was able to detect tumors a median of 475 days before clinical diagnosis, with 84% positive predictive value and 90% specificity.
Commentary: This is a retrospective study, but a confirmatory prospective study is underway. Since the average annual incidence of pancreatic cancer in the general population is only 0.02%, this level of specificity is far too low for this test to be used for broad screening purposes (false positives would be overwhelming). However, it does represent a good test for known high-risk individuals and/or those with unusual early symptoms.
A side note: The authors hypothesize that AI may be able to recognize texture differences on imaging that are beyond the limits of the human eye (although the black box nature of AI strikes again - the authors do not know for certain how this AI actually works).
AI in diagnostics: This week’s roundup
Does it seem to you like every other announcement about diagnostics these days involves AI? Yeah, us too. Here are a few recent AI-based innovations that caught our eye.
Cardiologists (and other interested folks) now have access to a free, online app that predicts risk of complications or death after balloon angioplasty +/- stenting (minimally invasive treatments for blocked arteries). It was developed using machine learning and feedback from patients, and it outperforms other current risk-prediction models.
A deep learning model was able to diagnose obstructive coronary artery disease based on retina photos from people suspected to have the condition. While its success was only moderate (86% sensitivity, 54% specificity, and 81% accuracy in the test group), it still had “better predictive power for events than the traditional pre-test probability models.”
The FDA has granted de novo approval to an AI-based program that predicts whether a patient will progress from amnestic mild cognitive impairment (aMCI) to full-blown Alzheimer’s disease within five years. The tool uses a standard 3D MRI of the brain to make the assessment and grades patients on a scale of zero to 100.
Another de novo approval came through for a machine-learning-based tool that uses CT scans to help diagnose idiopathic pulmonary fibrosis. According to the company that developed it, this is the first diagnostic tool for lung fibrosis that the FDA has authorized; diagnosis typically requires a biopsy.
A third FDA approval came through for an AI- and spectroscopy-based device that evaluates potentially cancerous skin lesions in folks over 40. In the study that led to the approval, it had 97% overall sensitivity for skin cancer and 90% negative predictive value. The tool is intended to help general practitioners know which patients need to be referred to a dermatologist for further workup.
COVID’s 2024 post-holiday party: Big but not like 2022
As has been widely reported, COVID is back with a punch this winter. But just how many cases are really happening is unclear, as so much testing now happens at home and is not reported. (Reminder: You can report at-home test results at makemytestcount.org - and you should report negative results, too.)
Fortunately, wastewater analysis allows us to get a sense of national viral loads - or at least compare where we are to where we’ve been. As shown below, COVID levels are up in all regions of the country - but are still below those of January 2022.
Of note: While mortality is significantly lower than in the past, between 1300 and 1500 people died each week from COVID in Q4 2023. Hospitalizations are up for the ninth straight week but at a slower rate of late. There are some signs that numbers are leveling out now that the holidays are a couple of weeks behind us. Fingers crossed.
Food for Thought
Cancer diagnosis significantly up in under-50s
This week the American Cancer Society released its 2024 “state of the union,” updated with 2020 incidence and 2021 mortality data.
Despite the overall good news that both the incidence and mortality rates of the two big killers (lung and breast cancer) have been in steady decline for 50 years, the press take has been quite alarmist: e.g., WSJ: “Cancer is striking more young people, and doctors are alarmed and baffled”; CNN: “A global epidemic…could be emerging.” (WSJ chart indexed to 1975 reproduced here.)
Commentary: There is no question that this increase is a real trend that needs to be analyzed and addressed, but let’s not forget that cancer remains rare in young adults. Whereas 1.7% of over-65s are diagnosed with cancer each year, incidence in young adults (20-49 years) is less than a tenth of that, at an average of 0.16%.
Are “doctors baffled”? Yes and no. On the yes side: Just how much are new unhealthy behavioral and social patterns responsible (e.g., sedentary lifestyles, poor nutrition)? If so, as this group ages, we can expect increased incidence and mortality among older adults too.
On the no side: The two biggest mortality drivers by far are colorectal cancer (which has been increasing for 25 years, primarily in younger men), and breast cancer in women. We already have highly effective screening and therapeutic tools for both these diseases. What this data may be telling is that we need to start using them in younger people.
Quick Hits
Recognizing GI dysfunction in dogs is easy (we can show you our carpets to prove that), but figuring out why it’s happening can be tough. After analyzing the microbiomes of almost 300 dogs, Texas A&M researchers have validated a Diagnostic Dysbiosis Index to distinguish acute and chronic GI issues and detect non-GI disorders.
The largest genome-wide association study of gestational diabetes to date demonstrated that there are two different classes of genetic variants related to the disease: One that is shared with type 2 diabetes and one that’s mostly associated only with the gestational form. Gestational diabetes affects more than 16 million pregnancies worldwide each year.