This release from the University of Oxford describes a new blood test that researchers there developed after zeroing in on the protein that collects in the joints of people with rheumatoid arthritis. The presence of the protein can predict the advance of rheumatoid arthritis many years before arthritis symptoms develop, according to the research findings published in the journal Annals of the Rheumatic Diseases. Among some of the 2,000 individuals who were part of the study, the protein, called tenascin-C, was found as many as 16 years before disease symptoms appeared.
The release does a good job describing the science behind the test but may leave readers wondering, “What’s next?” That’s because we aren’t told when and where the test will be available and there’s no mention of early treatment strategies — what the options are, what they entail, how long it lasts and what it might cost.
Most doctors screen for rheumatoid arthritis among patients presenting with joint symptoms. Tests to diagnose RA aren’t perfect but have gotten better over time. This news release highlights a new test that may be able to identify patients who have not yet been diagnosed. As such, it would be predicting future RA in patients who do not yet have symptoms or other findings such as arthritis changes on x-rays. Identifying patients with so-called “pre-RA” might appear to be a good thing, but it may not be as simple as implied in this news release. First, it is one thing to retrospectively examine whether blood samples performed years earlier can identify people we’ve already classified. The next logical step would be to prospectively see if we can take patients who don’t have any symptoms and use the test to say “positive or negative” and then wait to see if the test correctly predicts a future diagnosis. This is complicated by the need to say who should we use the test on. Everyone? Someone who may be at risk for other reasons? It isn’t clear and the news release doesn’t help us. Second, since no test is perfect and there inevitably will be false-positives, what does it mean to label someone as being likely to develop future RA if in fact they won’t? Third, even if the test is correct, it isn’t clear what to do with the results. Should we start treating these patients with medications to prevent future symptoms? If the treatments had no side effects, one could argue to do so. But we know that the treatments for RA all have side effects, some quite serious. For those with debilitating symptoms, the risks of such medications may be justified. But for someone who has no such symptoms, it isn’t as clear. One would like to see studies saying that use of this test result leads to better outcomes. No such data exists. Finally, there is no effort to state what would be the potential costs of using such a test in routine practice. One would not only need to consider the cost of the test itself, but also the downstream costs associated with treatment and the potential savings in terms of RA complications avoided.
The release doesn’t provide discussion of what the cost of the blood test might be. If the test is promising enough to issue a news release about it, readers should be given some idea of the cost. A related test called the CCP assay, which detects certain peptides that also predict future RA, is already in use and would make a good comparison. Moreover, there would not only be the cost for this test, but also the downstream costs of earlier treatment (and potential benefits). In general, the potential benefits in terms of cost savings are usually offset by higher costs associated with treatment for a longer duration of time (earlier diagnosis = earlier “longer” treatment).
The release asserts that the key benefit of the test for tenascin-C protein is early detection of RA that can be used to initiate early treatment. After looking at results from 2,000 patient volunteers, the lead researcher was quoted saying the test “could diagnose RA in about 50% of cases” and that the test was 98% accurate in ruling out RA. It isn’t clear from the release if all 2,000 patients started out healthy and half went on to develop RA, or if the test accurately detected the RA precursor protein in only half the patients that went on to develop RA.
Overall, the release provides insufficient information to quantify the benefits. Accuracy is a reflection of both sensitivity (diagnosing the condition when it really exists) and specificity (not diagnosing the condition when it doesn’t exist). The very high accuracy implies the test is both sensitive and specific. However, this may be misleading depending on the nature of the population studied. If the population includes patients with established rheumatoid arthritis it may make the test look better than studying a population of individuals who don’t have the condition or may have a different condition.
The release briefly notes that the test carries a very low rate of false positives. However, the harms from screening tests are well-known. A false positive can result in additional tests, unnecessary treatment, anxiety and financial strains. A major concern is using the test in asymptomatic individuals who may be at risk for future development of symptoms due to rheumatoid arthritis. If the test isn’t 100% specific, it means we’re telling some patients who are otherwise fine that they’re going to develop something bad in the future – when in fact they won’t. Oops! Second, it isn’t clear that treating such asymptomatic patients may benefit them long-term. Remember the drugs for rheumatoid arthritis all have side effects – some of which can be quite serious. These potential harms are not mentioned at all.
The research is a statistical analysis of data from four separately conducted studies of patients in four different geographic locations. The study cohorts varied which could affect accuracy of the findings. One group included 20 British RA patients and 20 healthy controls; one included 101 pre-RA cases and 326 matched controls compiled from four Southern European cohorts; a Swedish study involving 1985 RA patients and 160 health controls; and 287 RA patients and 330 control patients with osteoarthritis (OA) in the United States.
The key missing point is that these are all retrospective cohorts. What one would like to see is a prospective cohort where you apply the test and classify the patient as positive or negative and then see what happens. Even before that, one would like information on the cut-points used in this study to define what is positive or not. Understanding whether there is a clear point where the test is positive or not would be useful information. In practice, that cut point is usually never 100% obvious.
One could argue that this study and news release are pushing for the creation of a new disease – called “pre-rheumatoid arthritis.” But the value of classifying people with such a condition is not established, and the release doesn’t explore any of the downsides of such an approach. Expanding the boundaries of disease without sufficient evidence that this is justified constitutes disease-mongering. So we’ll flag the release here.
There’s no mention of the funders in the release.
According to the published study, funding was provided by the Arthritis Research UK, the Kennedy Trust for Rheumatology Research and BT CURE. In addition, it was disclosed that “data in this manuscript are the subject of a patent filing” and that the lead investigator sits on the board of and is a consultant to Nascient.
The release provides some information about the cyclic citrullinated peptide (CCP) test, the standard test used, and states that the new blood test appears to be more accurate and reliable in comparison. The true alternative is actually to wait until symptoms develop.
As noted above under “Cost,” if the test is promising enough to issue a news release about it, readers should be given some idea of when and where it might be available.
We aren’t told until the very last paragraph that the research “provides the basis of tests that could improve diagnosis and, importantly, detect disease at a very early stage, with the promise even that people at risk of developing rheumatoid arthritis can be followed before the disease begins.” In other words, it could be a very long way off. We think there should have been a caveat that more studies are needed or that this test may be years off before it will be ready for routine practice.
Research into different proteins to diagnose RA in patients prior to the appearance of symptoms isn’t new. In fact, HealthNewsReview.org reviewed a WebMD news story on early detection of RA in 2010.
Another blood test for RA is already on the market in the United States, Canada, Europe, Japan and Australia.The JOINTstat test looks for another protein called 14-3-3η.
The release suggests the test will “improve diagnosis” and help patients get the right treatment early in order to manage their disease. It’s too early to make such claims about a test still under development. For the reasons cited, this release does an inadequate job in placing this potential test into a larger context.