Snowball in a Blizzard

by Steven Hatch

  The Lyme ELISA, then, is designed to capture that specific antibody response to Borrelia burgdorferi. Since Lyme-infected patients have ready-made anti-Borrelia antibodies in their blood, they can be captured if a patient’s serum is dropped into a well coated with Lyme proteins. After the patient’s serum is mixed in this well allowing for the binding, the rest of the serum—with all its other antibodies that aren’t specific for Lyme—can be washed away. After that, special antibodies that bind only to human antibodies are added. These antibodies are attached to fluorochromes, which are effectively tiny light beacons that can be activated, detected, and quantified by lasers. From there it’s a simple matter of a second wash, adding the proper reagents to activate the beacons and placing the well in a specialized laser system designed to read ELISAs.

  Keep in mind that if a person doesn’t have Lyme, when the serum is added to the well, there won’t be any binding of that person’s antibodies to the Lyme antigens that coat the surface of the well. The antibodies meant for organisms other than Borrelia will just bounce around in the fluid, and so when the serum is washed away, all the patient’s antibodies will be washed away, too. Thus, when the next step of the ELISA takes place, and the antihuman antibodies with the molecular beacons are added, they will have nothing to bind to themselves and will also be washed away. The resultant well will be read by the laser in the ELISA reader as negative—that is, the quantity of light emitted by those molecular beacons will approach zero.

  That’s the ELISA in principle, although the reality is a little more complicated, and those complications limit the effectiveness of the ELISA as the sole test for Lyme. The first is that, because the ELISA test measures the quantity of light emitted by the fluorochromes attached to the antibodies that bind in turn to the patient’s antibodies, it is an indirect test twice removed, and its result isn’t yes or no but is instead a number. Needless to say, because of the complex, multistep nature of the procedure, the numbers do not fall into nice, neat binary categories of zero and some high number that would indicate infection, but rather fall along a continuum. If we return to our body/murder crime scene analogy, the ELISA provides an answer as to how much blood is in the room. As a consequence, based on the characteristics of the test, some arbitrary number is chosen as the cutoff for “positive” and “negative.” Anything below that number is considered reliably negative, and above that positive.

  Unfortunately, the Lyme ELISA test has a false-positive problem. We’ve encountered this before elsewhere: the physicist Leonard Mlodinow’s “positive HIV test” that wasn’t really positive was an ELISA. The reason for this is that some antibodies the body produces can bind to Lyme antigens (or HIV antigens or whatever is being tested) even though they weren’t produced in response to an infection. Some diseases, like lupus, are caused by the body producing an abundance of antibodies: the high level of antibodies in the serum of these patients can often cross-react with antigens in ELISA tests, producing false positives. Even among healthy people, some just happen to have these cross-reactive antibodies. The Lyme ELISA turns out to have somewhere around a 5 percent false-positive rate: that is, for every hundred people with a positive test, five were never infected.

  This is where the Western blot comes into play, and it is used in precisely the same way HIV testing requires a Western blot to confirm infection (this test either wasn’t commercially available or wasn’t used when Leonard Mlodinow was “diagnosed” with HIV). The Lyme Western blot relies on measuring antibody responses for specific proteins found on the surface of every Borrelia burgdorferi bacteria. The test looks at thirteen proteins in all: ten can be bound by one type of antibody known as IgG, which are typically produced several weeks after infection, while the remaining three are bound by IgM antibodies, which are generated during the acute phase of infection and eventually dissipate. If the antibodies bind to the proteins, they are “labeled” in a manner similar to the antibodies in the ELISA and show up as little black lines, or bands, on photographic paper. Unlike the ELISA test, which in its crude form is a number, the Western blot bands either are present or not, and thus are either positive or negative.

  If we return to our body/murder analogy, the ELISA provides us with a general sense of how much blood is in the room, while the Western Blot provides an even more specific quantification of the amount and confirms that the blood is from the murder victim. Over the past two decades, the Western blot has served as the definitive test for Lyme disease, becoming the final stop on a two-test system that utilizes the less expensive and less cumbersome ELISA as a primary screen, eliminating the false positives from the first test because of its greater accuracy. There are yet other legitimate Lyme tests that are beginning to gain acceptance, but the two-step testing remains the mainstay of Lyme diagnosis at the time of this writing. However, the alternative community has devised other Lyme tests that aren’t nearly as reliable.

  As with the ELISA (whose value is some number but gets interpreted as positive or negative), the Western blot similarly has a cutoff number: two out of three IgM bands indicates acute infection, while five of ten IgG bands make the diagnosis of past infection. Researchers developed these criteria for the same reasons that they chose the cutoff value in the ELISA: above the threshold, one can be reasonably confident that there’s a real infection, but below that number the test isn’t reliable. We know this by performing the Lyme Western blot on perfectly healthy people who don’t live in Lyme-endemic areas. Consider going to a mountain community in the Himalayas of Nepal, where Lyme definitely does not exist and collecting blood specimens to test for Lyme. These hypothetical Nepalese test subjects who have so graciously provided their blood aren’t even sick, yet even here the Christmas tree can light up to a limited extent, as some subjects will have a few IgG or IgM bands.

  To be sure, it is an imperfect test. Thus, the current criteria for Lyme diagnosis, both in terms of the ELISA and the Western blot, are constructed so as to minimize the false positives.

  Although I have described the Western blot in terms of clear positive or negative bands, even here the story is more complicated: sometimes the bands are faintly positive. Does this mean there are true anti-Lyme antibodies present in the patient’s serum, or is this just a fluke related to the imperfect characteristics of the test? One Lyme researcher named Gary Wormser looked at this question in relation to IgM antibodies. Since IgM antibodies can be produced only during the early, acute phase of the infection (usually the first four weeks), we know that if IgM testing is positive in the cold and snowy New England winter, when ticks are much less likely to feed and people are far less likely to be exposed to these ticks, it is highly probable to be a false positive. Dr. Wormser reviewed the records of approximately 250 patients using criteria such as these to evaluate interpretation of Lyme Western blots. Of the 250, nearly one hundred had positive IgM Western blots, and of these, about half were deemed to be false positive based on commonsense criteria.

  The entire Lyme diagnosis apparatus, then, is somewhat clunky, and it is sufficiently complicated that even mainstream physicians can misinterpret the tests and their cutoff points. Plus the two-step testing, in its effort to eliminate false positives, creates in turn the possibility of false negatives. What do you do with a person who has symptoms of textbook Lyme, lives in Connecticut, and whose Western blot has only four positive IgG bands and one positive IgM band? What happens when a newly sick patient has a negative ELISA, but the testing is done early enough that one cannot be completely confident of a negative result because the immune response isn’t immediate? Given the ease by which Lyme can be treated—three weeks of doxycycline typically does the trick—many doctors accept that they will overtreat a percentage of their patients, either by treating patients who don’t have Lyme or ones that do but don’t have definitive evidence of it at the time they come to the clinic.

  That said, the uncertainty of Lyme diagnosis has limits and only applies to a relatively small number of patients suffering from
classic Lyme symptoms in areas where deer ticks live. For these patients, suspicion of negative tests is sometimes warranted. But the believers in chronic Lyme disease have seized upon the uncertainty inherent in Lyme testing, convincing patients that the mainstream approach is far too restrictive and thus not to be trusted under virtually any circumstance, particularly in patients with persistent, low-grade symptoms such as chronic fatigue, muscle pain, and difficulty concentrating. The alternative practitioners fashion themselves as “Lyme-literate physicians,” thus implying that the mainstream doctors are Lyme illiterate—with the goal of maximizing doubt about the reliability of ELISAs and Western blots, unless, of course, the testing aligns with their preconceived notions about why a given patient is ill. The distrust of the standard two-test approach to Lyme, which magnifies the issue of the uncertainty inherent in the testing, is the starting point for going down the rabbit hole. Without it, the rest of the theory doesn’t make much sense.

  To aid those efforts, the alternate Lyme community has a panel of yet more tests for Lyme diagnosis, most of which are not approved by standard scientific labs. Among others, they include “urine antigen capture assays,” which look for proteins produced by Borrelia in a urine sample; CD57 assays, which look for the presence of mature immune cells known as “natural killer” cells that once looked promising in diagnosing Lyme but turned out to be elevated in all kinds of conditions; and the Lyme lymphocyte transformation test, about which the less said the better. The CDC has included a page in its Lyme disease section referencing all of these tests, noting with bureaucratic dryness that “some laboratories offer [these tests] using assays whose accuracy and clinical usefulness have not been adequately established.”

  The website for the alternative chronic Lyme advocacy group ILADS provides a fair illustration of its suspicion of mainstream test interpretation. In a section titled “Basic Information about Lyme Disease,” it argues that the CDC intended the two-step testing process to be used for only “disease surveillance,” not diagnosis; that the CDC Western blot criteria are defined too narrowly and that “the Western blot should be performed by a laboratory that reads and reports all of the bands related to Borrelia burgdorferi”; and, moreover, that the ELISA is the wrong test to use in the first place because, according to ILADS, the ELISA misses about a third of Lyme cases to begin with.

  Taken together, these objections to Lyme testing sets the stage for a two-pronged assault on the validity of the mainstream community’s Lyme diagnosis strategy. The first prong asserts that alternative laboratories, which use a variety of tests that either are not approved by the Food and Drug Administration or are not interpreted according to the CDC guidelines, are the only ones that can be trusted in diagnosing patients. The second prong asserts that Lyme is a “clinical diagnosis”—that is, labs aren’t really relevant anyway. “Familiarity with [Lyme’s] varied presentations is key to recognizing disseminated disease,” the website says, implying that a mainstream physician’s reliance on a negative Lyme test is not only misplaced faith in technology but speaks of that physician’s ineptitude in clinical diagnosis.

  In short, it’s a heads-I-win-tails-you-lose response to mainstream medicine: first they use a different set of tests with a different set of standards, ensuring that there are far more positive than negative tests; second, in the unlikely event that a test from one of their approved labs is negative, these practitioners will often argue that the Lyme bacteria is still present and has evaded laboratory detection, or that a patient’s disease is due to a different tick-borne illness such as babesiosis.*

  I have seen many patients referred to me with tests from these alternative companies endorsed by chronic Lyme advocacy groups. Personal experience can be trusted only so far, but with that caveat in mind, I have never once seen a patient who has had completely negative testing from them.

  This entire elaborate alternate hypothetical structure is founded to a large extent upon the moderate degree of uncertainty inherent in Lyme diagnosis, especially in the first few weeks of the infection when even mainstream doctors can disagree about cases. The mainstream approach during that window is, by and large, to overtreat some patients: give appropriate antibiotics to a plurality, or even a majority, of patients who present with these vague symptoms during the warm-weather months in Lyme-endemic areas, knowing that some people will actually have some nonspecific viral or other illness. Given the limitations of the testing, it is a balancing act between the clear benefits for those infected with Lyme and risk for those with other problems. Here, overtreatment in the right population (those with classic Lyme symptoms in the early period of disease) carries real benefits at the cost of small risks (as I’ll explain shortly, the harms of antibiotics in patients who don’t have Lyme but whose presentation mimics it).

  Contrast this with the downsides of overdiagnosis and overtreatment of breast cancer through the false-positive problems of screening mammography, which introduces the possibility of the horrifying spectacle of unnecessary surgery. Again, we are in a different place on the spectrum of certainty: the harms of overtreating Lyme in the face of legitimate uncertainty are smaller, and the benefits real.

  However, as people’s symptoms become more prolonged, the likelihood that a negative Western blot or a negative ELISA really does mean that a patient isn’t suffering from Lyme becomes more and more reliable. But ILADS and its supporters, who have overestimated the uncertainty involved in Lyme diagnosis to the point that they don’t trust standard testing at all, can’t be convinced of this. To me, they appear to refuse to be convinced, despite an avalanche of scientific evidence that has accumulated since the research on Borrelia began in earnest.

  If this description of an advocacy group resisting mainstream medical opinion by evaluating evidence in a highly selective manner sounds familiar, it should: this is more or less the intellectual approach used by antivaccine advocates. (I will talk about the HPV vaccine briefly in the chapter on media.) But it is noteworthy that the strategies of these two movements bear a striking resemblance to one another. To a great extent, they could be described as flipping mainstream uncertainty into alternative-medicine capital-C Certainty. This child with autism? It must have been due to vaccinations because he was vaccinated and now he has autism. (One cannot, after all, prove the negative.) The Lyme ELISA eliminates negative tests from consideration for the disease, and this patient with six months of fatigue has a negative ELISA? No, it’s the wrong test, he needs a Western blot, and we know this because he has Lyme. The Western blot has only two positive bands? You’re interpreting the test wrong, and we know this. Why do we know? Because he has Lyme!*

  This also is the same type of circular reasoning used by the psychiatrist who concluded his patient lacked “affective stability” in the Rosenhan experiment.

  The schism between mainstream doctors and their alternative counterparts was years in the making, with much of the contentiousness focused on what did or did not constitute unimpeachable evidence that someone didn’t have Lyme. In general, among mainstream doctors evaluating patients with several months of symptoms, a negative Lyme test meant that Lyme was not to blame. The alternative practitioners, meanwhile, frequently attributed these symptoms to Lyme. Over the years, companies were founded that allowed the alternative practitioners to have their diagnosis and eat it, too, so to speak, using the testing that I’ve described above (and which led to real-world profits for those companies that fed this diagnostic hunger). Thus a good deal of the disagreement was based on diagnosis.

  Throughout the 1980s and 1990s, there was, however, one area in which these two very distinct groups shared a somewhat similar approach: the use of long-term antibiotics in patients with lingering symptoms after Lyme infection. What was clear was that a small minority of patients who were clearly suffering from Lyme disease were failing to improve, even after completing several weeks of antibiotics. David, the patient whose story began this chapter, would have been an apt example but for the fact t
hat his Lyme screens were never positive. Many others who did have Lyme disease had stories matching David’s, however.

  Unsure of what to do and without clear-cut scientific evidence to guide them, some mainstream physicians gave ever-longer courses of antibiotics under the assumption that there was some hidden reservoir where Borrelia burgdorferi was riding out the antibiotic storm, only to crop up after the medication was stopped. Given that Lyme had a predilection for joints, and that antibiotic penetration into joints is fairly low compared to that of other parts of the body, the reasoning was that Lyme managed to survive there and could recur if subjected to only a few weeks of antibiotics. Thus, some physicians opted for courses as long as three or six months, and a smaller number treated patients even longer.

  To give a sense of context, most bacterial infections are treated with antibiotic courses lasting from days to weeks. Osteomyelitis—infections in the bone—are often treated for up to three months. Tuberculosis is the leader in terms of length of treatment for a bacterial infection: depending on where the tuberculosis is found, courses can range from as short as four months to as long as twelve. But, barring very unusual circumstances, bacterial infections are almost never treated for more than a year; such prolonged treatments are rare, reserved for uncommon organisms in addition to TB.†

  Patients with implanted devices (like artificial hips) that become infected but cannot be removed because the procedure is too risky for the patient, are typical candidates for “antibiotics for life.” There are also a couple of unusual bacteria that require more than a year of treatment. But, again, these are exceptional situations.