Health Through Knowledge

The Serano Group

Monday, May 30, 2011

Lyme Borreliosis Tests

How good are they?

Get sick and you are filled with uncertainties. Did I catch something? How long have I had it? How bad is it? Am I going to have more problems? How do I know when I am over it? The questions go on and on.

It gets harder because, real-world, nothing guarantees every disease has to have a clear progression following a neat, concise course. Usually, the easy diseases are acute. You suddenly get sick. You get better in a week or two. The illness matches the textbook description. Your symptoms match those in the textbook.

If you get an infection, both you and your doctor would like it to be one of the easy ones.

But, real-world, some diseases are easy and others not. Lyme disease, more correctly, Lyme borreliosis, is one of the worst. Symptoms can come and go and may vary greatly from one individual to another. Some symptoms can go into a period of quiescence and reappear after an accident or operation. Other microbes can affect the course of the disease. Different individuals have varying responses to treatment.

Unfortunately, nothing messes up a complex disease more than a bad case of politics.

With Lyme borreliosis, the politics are some of the medical world’s most extreme. To summarize a thirty year history in a few words, a small number ideologues—primarily rheumatologists, neurologists, infectious disease researchers and microbiologists—looked at the complexity of the disease in the 1980s and decided their solution was to transform the realities of the infection into a neat package of an ideal infectious disease: easy to diagnose, easy to treat. Reality became a political opponent to be defeated using all means possible, including the production of biased research and censorship of disagreeing experts. Add the fact that many of the researchers were intertwined with a disastrous commercial vaccine effort (GlaxoSmithKline introduced LYMErix in 1999 and removed it from the market in 2002). The result was a distortion of definitions and an unfortunate undercurrent of “blame the patient” that persists to the present.  Caught in the middle, between the medical ideologues and the public who experiences something quite different from what the ideologues say, are our everyday doctors, the ones most of us go to when we get sick.

Doctors are pretty much taught a standard way of dealing with medical problems and are required to support that type of medicine if they want to get paid and stay out of trouble. A big part of this model is testing.

In the United States, clinical laboratory testing is estimated to be a $50B a year industry. Testing hits that sweet spot in the medical industry—readily paid for by insurers without question, and not over-utilizing the expensive services of doctors. Tests move doctors away from their rightful subjective decision-making role, a function insurers want under the insurance company’s control. The insurers want everything, especially payout costs (which they refer to as “losses”), as predictable as possible. Predict next year’s payouts, tack on profits, set the premiums accordingly, and insurers have a nice business. A doctor who decides to treat based on knowledge and experience disrupts the whole plan. It is better for insurers to have doctors diagnose based strictly on tests.

Test Basics

Before getting into the specifics of Lyme disease testing, let’s be clear on what we would like from clinical tests for infections. We want a test to produce a positive result 100% of the time in samples taken from subjects infected, producing perfect sensitivity. We want the test to produce a negative result 100% of the time in those uninfected, producing perfect selectivity. Another way of saying this is that sensitivity is the ability to detect the pathogen; selectivity is the characteristic of returning a positive result only for infected subjects. Of course, for many reasons, but primarily because we are dealing with complex biological systems, we never get perfect selectivity and sensitivity.  Let’s review what is available in Lyme disease tests.


If a doctor orders a Lyme disease test it will usually be an enzyme-linked immunosorbent assay (ELISA). A blood sample is sent to one of the large clinical laboratory testing corporations such as LabCorps or Quest.  You might encounter similar tests such as an enzyme immunoassay (EIA) or Immunofluorescence antibody analysis (IFA), but functionally, these are quite similar to the ELISA.  An ELISA is a technique, not specific to Borrelia or any other pathogen until an ELISA test kit or custom assay is designed to detect a particular antibody, hopefully specific to a particular microbe. A key point is that an ELISA looks for antibodies in the blood sample, not the Lyme disease bacteria, Borrelia burgdorferi (Bb). An ELISA is called an indirect test since it does not attempt to detect the bacterium itself, only the antibodies that your body creates to defend against the pathogen.

There are many technical details with several variations, but the basic process is conceptually simple. For a Lyme disease ELISA, a lab strain of Bb is grown in quantities. An extract of the bacteria is then fixed to a plate. Serum from a patient blood sample is washed over the plate and if antibodies are in the sample, they should attach to the Bb antigens on the plate. The antibodies attached to Bb antigen can be detected, after processing, visually or through immunofluorescence. If the antibody-antigen complex is visible at a a predetermined level, the test result is reported as positive.  Again, a key point is that the test does not attempt to detect Bb. It looks for antibodies to Bb.

Later in this article, I talk about how well a Lyme disease ELISA works in the real world, but first let’s look at some of the ways an ELISA could produce incorrect results.

Foremost, the test is entirely dependent on the person tested having a well-functioning immune system. Complicating things, the lab strain of Bb used in the test kit could differ from wild strains so greatly that antibodies from patient samples do not attach. Antigen-antibody attachment is not molecularly precise, more of a loose fit, but at some point the mismatch become so great that antibodies do not attach. Attachment will depend upon the strain the test kit manufacturer uses and important differences between wild strains and lab strains may cause antibodies to not attach.

One of the best Lyme borreliosis researchers, Reinhard Straubinger, has become expert at infecting dogs and ponies in his animal studies. His most reliable method is to use wild ticks containing wild strains of Bb. (For those perhaps thinking Lyme disease is rare, note he has no problem reliably collecting ticks carrying Bb.)

There are other subtleties to ELISA technique. The plate should be washed so that only antibody proteins attached firmly to the antigen substrate remain. Wash too vigorously and you wash away the antibody you are trying to detect.  Another striking deficiency of a Lyme disease ELISA is that it depends on the blood sample having free antibody molecules available for detection. In a severely sick individual, especially in a person immuno-compromised, they may not be producing excess, free antibodies. Whatever antibodies in the serum are already attached to Bb, making them unavailable to attach to the Bb on the ELISA plate. This is a reasonable explanation for the often observed phenomena where those sickest with Lyme borreliosis and can produce a negative ELISA.

You might encounter some variations in ELISA-type test reporting. Sometimes a numerical titer value is reported. Usually, this means the test antigen, as a reference, is put on a plate in progressively greater dilutions and complexed with the dye molecule as a referene.  Rather than just report that a single plate well was positive because it looked, say, pink, the technician looks at the degree of pinkness and compares it to the reference. (More likely an optical density meter and automation is used, but the concept is the same.) If your sample really had a lot of antibodies, it might not turn clear until dilution by a factor of of 1 to 1000. Your titer would be reported as 1:1000. If your sample produces less dye-attached antibodies, the solution might turn clear when diluted 1 to 100. Your reported sample titer would be 1:100. Sometimes labs set a cut-off point. For example, anything above 1:250 is reported positive. Samples 1:250 or less would be reported as negative.

A variation of the whole-cell ELISA is the VlsE C6 peptide ELISA which attempts to detect a specific antibody protein. This test has not proved to offer real advantages and is not often used.

Western Blot (WB)

Like the ELISA, the Western Blot (WB) is a technique, not specific to a particular disease. Only when designed to detect the antibody proteins unique (hopefully) to a disease pathogen does it become a diagnostic test. Like the ELISA, a WB for Lyme disease is an indirect test—it does not detect Bb, only proteins of antibodies to Bb.  Still, a WB is much more informative than an ELISA, since it attempts to detect antibodies to various components of the Bb bacterium

The basic technique is as follows. Instead of just using a slurry of Bb, the bacterium is first sonicated (busted up with sound waves). The resulting protein fragments are placed on a gel sheet and a current applied. causing the proteins to migrate varying distances down the plate depending on their weight, separating into bands. Proteins are weighed in kDa (kiloDaltons) producing the numerical identification of the bands.

After the test gel is made, the rest of the technique is similar to an ELISA. Serum from the sample is washed over the plate to see if there are antibodies that stick to the separated Bb protein fragments on the plate. An added feature of WBs is that usually the sample antibodies are first separated out by weight into IgM and IgG antibodies and separate results reported. (It is thought that IgM antibodies develop early in an infection and that IgG antibodies develop later, providing long-term immunity, but with Lyme borreliosis, often new IgM antibodies are detected late in the disease.)  After washing and staining the result is something that looks like this:


There is a lack of consistency in how the bands are labelled and often there is no clear separation of the bands. But, generally, results are reported like this:

BAND INTENSITY: Low +, Medium ++, High +++, Equivocal +/- 

18 kDa   +/-
22 kDa   -
23-25 kDa +/-
28 kDa   -
30 kDa   -
31 kDa   +/-
34 kDa   +/-
37 kDa   -
39 kDa   +
41 kDa   ++
45 kDa   —
58 kDa   +
66 kDa   +/-
73 kDa   -
83 kDa   -
93 kDa   +/-

Sometimes, all you get is a positive or negative result on a report instead of a listing of bands. Results are highly dependent on the technician reading the blot. What could be an “indeterminate” for one technician could be a “+” for another. As one treating doctor has asked, what does indeterminate mean anyway? Certainly, they see something there or they would not report it.

Generally, patients find WBs impressive. Finally, they think, in the course of this confusing disease, they have a lab report with some hard numbers and all this precise data must mean something.  We can only wish.

The same group of ideologues who worked years to convince the world that Lyme disease was easy to diagnose and easy to treat, specified to the world what those Western blot numbers mean. The Centers for Disease Control (CDC), rubber-stamped or directed (depending on one’s level of cynicism) which bands mean a positive result. The detailed list of bands, used to this day, was based on a single study of 74 people, endorsed by a 1994 committee sponsored by The Association of State and Territorial Public Health Laboratory Directors, CDC, the Food and Drug Administration, the National Institutes of Health, the Council of State and Territorial Epidemiologists, and the National Committee for Clinical Laboratory Standards: 2 of 8 IgM bands from this list: 18, 21, 28, 37, 41, 45, 58, and 93 kDa; 5 of 8 IgG bands from this list: 18, 21, 28, 30, 39, 41, 45, 58, 66 and 93 kDa.

The CDC is always quick to add that this list is only to be used for surveillance, not for clinical diagnosis, and reported positive only when coupled with specific symptoms. The official stance is that Lyme disease is a clinical diagnosis not dependent on tests. But, de facto,  most doctors insist on a surveillance quality positive test before they will diagnosis and treat. The word on the street, or back office, since the mid-1980s has been that “good” doctors only diagnose the most obvious Lyme borreliosis infections and then if they do find a positive case, get someone else to treat it.

An especially interesting detail of the WB band criteria is that the highly diagnostic 31 kDa band is omitted from the list. Many of the researchers influencing the list were involved with GlaxoSmithKline , developer of the disastrous LYMErix vaccine which was on the market from 1999 to 2002. Lymerix was based on injecting the 31 kDa protein. Leaving it off the list was an attempt to ensure Lymerix did not contribute to creating positive test results, an unwanted effect of a vaccine. GlaxoSmithKline surely did not want their vaccine to change a patient’s negative test into a positive test. In reality, LYMErix did much worse, crippling many of those vaccinated and sometimes causing neurological disease. Additionally, for reasons not adequately explained, LYMErix produced another unexplained Lyme borreliosis anomaly: the vaccine often turned all the bands in a WB into a continuous dark blob.

Both the ELISA and WB, tests that look for antibodies, are referred to as serologies.


A seldom performed test, primarily because of cost, is a Bb PCR. This test looks for Bb DNA, typically in a blood sample, but sometimes in a tissue sample. It is highly insensitive, most likely because Bb has a tendency to sequester in fibrous tissue, such as the skin and nerve cells, rapidly clearing the bloodstream. While a Bb PCR is rarely positive, a positive test is a nearly certain indication of infection. Still, ideologues often claim positive PCR tests are detecting only dead Bb. Plus, as an exception in the infectious disease world, the CDC does not count a person with a positive PCR for Bb as a surveillance case, nor do doctors feel compelled to treat PCR-positive patients. Remember, prevailing ideology says to deny as many cases as possible.


Unlike many bacteria which are routinely cultured to diagnose infection, the Bb spirochete is quite difficult to culture and cultures are only performed in a research setting. Note that the spirochete that causes syphilis, Treponema pallidum, has never been cultured outside of living mammals, so any claim that Bb does not exist in a patient or group of patients because it cannot be cultured in a lab is the logical equivalent of saying syphilis does not exist.

What do all these tests mean?

Everyone wants Lyme disease tests to be more than they actually are. Doctors want some objectivity and certainty and so does the public. But we don’t get it. Instead, we get the ideology, dominating and often censoring the discussion. Medical journals and popular media usually report and conform to the ideologues advocacy. Sick and disabled people become especially interested in the precise-sounding band numbers in the Western Blots. All those numbers, they think, finally some precision and confirmation to this confusing, mutable illness. People post their Western blot results on forums and there is continuing speculation as to what the bands actually mean.

Exacerbating the testing problem are everyday doctors who are more familiar with HIV testing, where an ELISA seems to work as an effective screening test, and confirmation is made with the more accurate WB. Doctors assume Lyme disease ELISAs and WBs work just as well as HIV ELISAs and WBs, when there has never been any science performed that remotely indicates this. In Lyme disease testing, the two-tier testing, using an ELISA for a screen and a WB for confirmation does not meet acceptible standards for detection. For one thing, often studies show a sample that tests negative on a Bb ELISA produces a positive WB in a significant number of cases, meaning the ELISA is an invalid screening test.  This is yet another Lyme borreliosis anomaly.

What is almost jaw-droppingly irresponsible is that ideologues like Gary Wormser, who wish to deny as many Lyme borreliosis infections as possible, state that doctors should throw out a positive WB result when coupled with a negative result on the cheaper, less precise ELISA. The responsible approach, since studies show the ELISA is less sensitive,  would be to skip the ELISA entirely. Money keeps most doctors from ordering WBs. An ELISA cost about $50 versus the WB’s $250.

Testing tests

What do the published studies tell us about the accuracy of Lyme borreliosis tests? It seems as though it would be straightforward to evaluate tests. A committee associated with the World Health Organization recently published recommendations for evaluating clinical tests. Indicative of the distortion of clinical Lyme disease research, the few available evaluations of Lyme borreliosis testing seldom, if ever, follow established procedure.

Good science involves details and data, so only by looking somewhat closely at a single study can we see the distortions of routinely published clinical Lyme disease research.  In 2004, I worked as a contract researcher for Dr. Joseph Jemsek, the doctor shown being persecuted by the North Carolina Medical Board in the documentary film Under Our Skin. Patients were telling him that Johns Hopkins had some new tests that were 100% sensitive for detecting Lyme disease. It was easy to see why the patients were saying this— there were readily available early copies of a study abstract quoting impressive sensitivity statistics like 100% and 92% that practically jumped off the page. 

The Johns Hopkins study
looked at 86 people who were classified primarily by EM rashes as being "probable", "possible", and "unlikely" cases of Bb infection. All subjects had symptoms of Lyme borreliosis. There was no comparison to healthy controls. As an aside, there has never been an study objectively determining the prevalence of the supposed hallmark symptom of Lyme disease, the EM rash, so its validity as a study determinant is highly suspect.

Yet another problem making the Johns Hopkins study difficult to apply to clinical practice was that the researchers devoted a good deal of attention to culture, a test not available to the general public.  The study’s prominent Figure 1 showed “NO EVIDENCE OF LYME DISEASE” for about half of the subjects tested and I had difficulty discerning how the data were created. I wrote to the journal asking for clarification and the response from the lead investigator gave me enough information to figure out how the data were reported. Consistent with the politics, the lead investigator used the opportunity to pontificate and get in an unrelated dig in at IGeneX, a specialty lab that had been a long-time target of the ideologues, primarily for performing more sensitive tests than those a typical commercial lab uses.  To cut to the chase (more details), what the Johns Hopkins researchers were doing , astoundingly, was using the same tests they were evaluating to determine what was a true positive test. In other words, they assumed the tests were always right and then began their analysis. Unsurprisingly, the tests were determined to be 100% sensitive.

In their analysis, if a test came back positive it was assumed it was 100% accurate: the subject had Lyme disease. If a test came back negative, it was assumed it was 100% accurate: the subject did not have Lyme disease, even though a lot of those samples came from CDC-positive surveillance cases (which usually don’t depend on tests). Look at the tests, the study asserted, assume they are 100% sensitive, and then our analysis concludes they are 100% sensitive.

Taken to the logical extreme, using their methodology, flipping coins, reading tea leaves, and throwing darts, would all in combination show 100% sensitivity for detecting Lyme disease, just as they had for serology, PCRs, and culture. Moving into the world of clinical realities, a CDC-positive surveillance case in a citizen of Lyme, Connecticut, who had six Bb-PCR-positive ticks attached for three days, resulting in a 30 cm EM rash observed by a physician, coupled with severe Bell’s palsy, fever, and arthritis in one knee, where symptoms resolved after antibiotics; this subject would be classified in the Johns Hopkins study as having “NO EVIDENCE OF LYME DISEASE”, just because the subject had a negative ELISA. 

I made numerous requests to the authors for disclosure of their basic data—counts of how many in their three study cohorts tested positive or negative—and received no information, only multiple promises the data were forthcoming. The editors of the journal and peer reviewers have never required these data either.

How did the the tests regular doctors use almost exclusively, ELISA followed by WB, perform? Without data, we have to guess at whether test positives came from the probable, possible or unlikely groups, but in total, only 36 of the 84 subjects tested (43%) were positive.  I suspect that the researchers found that many gold-standard cases, those with positive PCRs or cultures, came from the group they deemed “Unlikely”. This would have encouraged a discussion ideologues do not seem to want to engage in.  Once I knew how the data were presented I wrote to the editor of the journal pointing out that even in its obscure presentation the study data showed…

if a physician required a positive test for diagnosis and only used the two-tier ELISA followed by Western Blot testing procedure (currently these are the only tests routinely ordered in clinical settings), 11 cases of certain infection would be entirely missed in the 86 subjects. Those 11 (13%) were positive by our present day gold standards, culture and PCR results, yet had negative serologies. This statistic furthermore ignores all subjects not positive on any test, a number not stated in the study as presented, but quoted in the abstract as 49/86 (57%). If only 20% of these subjects (who were selected because they exhibited symptoms suggestive of Lyme disease) were truly infected, the physician relying on two-tier serology testing would miss 21/86 (24%) cases of a potentially disabling or fatal disease.

This letter was promptly and summarily rejected by the journal’s chief editor even though I had some support from hands-on editors. Ironically, at about the same time a study submitted by Dr. Jemsek to the same journal was rejected. In Jemsek’s study, rather than just make the absurd assumption all Lyme tests are always correct and equally valid, Jemsek selected from 550 patients the gold-standard cases, those with positive PCRs. Only 27% of these definite Lyme cases, when concurrently tested, produced a positive Western Blot by standard CDC-endorsed criteria. So, in response to the protest voiced repeatedly for the last thirty years, that if the treating doctors are seeing something that differs from the published ideology they should report it, Jemsek’s study was censored. 

Soon, after reading much more, equally dubious, clinical Lyme disease research, I stopped relying on what medical journals said about Lyme disease.  I am not the only one who has lost faith:

It is simply no longer possible to believe much of the clinical research that is published, or to rely on the judgment of trusted physicians or authoritative medical guidelines. I take no pleasure in this conclusion, which I reached slowly and reluctantly over my two decades as an editor of The New England Journal of Medicine.

From Dr. Marcia Angell, The New York Review of Books, Volume 56, Number 1. January 15, 2009 Drug Companies & Doctors: A Story of Corruption.  (Dr. Angell resigned as editor-in-chief of The New England Journal of Medicine in June, 2000.)

Bottom Line

If you have read even a few parts of this article, you may suspect Lyme disease tests are not acceptably accurate. We do not even have a good estimate of test accuracy because there have not been any credible studies evaluating test accuracy of which I am aware. But that is not to say the tests are useless.

Lyme borreliosis is complex and the tests are obviously not definitive. Possibly, they miss 90% or more of those infected. Patients want the tests to mean more. Doctors often insist that they mean more. That does not mean they do.

I’ve studied Lyme borreliosis for over ten years. My guess is that if you have a Western Blot that is positive for any band other than 41, this is highly indicative of Bb infection. Another guess is that if you have a ELISA for Lyme disease that comes back negative, there is still significant probability you have a Bb infection. I won’t even hazard a guess at a percentages—everything depends on your history and entire clinical picture. You should be especially concerned if you have a set of symptoms that are not typically expected in combination. Be particularly concerned if the symptoms developed about the same time without other plausible explanation.

Most importantly, if you are negative on every Lyme tests you have ever had and you have symptoms, consider that there is still a good chance you have a Bb infection.

I’ve come to put more confidence in another test of sorts: how someone responds to antibiotics. A significant reaction, either good or bad, strongly suggests a bacterial infection. An anonymous treating doctor writing in an excellent blog agrees with me.

Tests are only part of a responsible diagnosis. Get one or more tests, talk to a doctor not constrained by the prevailing ideology, and take someone with you whose judgement you trust. Decide what is important to you and do it.

Selected References

"Case Definitions for Infectious Conditions Under Public Health Surveillance. Centers for Disease Control and Prevention." MMWR. Recommendations and reports : Morbidity and mortality weekly report. Recommendations and reports / Centers for Disease Control , no. 46 (1997): 1-55.

Coulter, Peggy, Clara Lema, Diane Flayhart, Amy S Linhardt, John N Aucott, Paul G Auwaerter, and J Stephen Dumler. "Two-Year Evaluation of Borrelia Burgdorferi Culture and Supplemental Tests for Definitive Diagnosis of Lyme Disease." Journal of clinical microbiology 43, no. 10 (2005): doi:10.1128/JCM.43.10.5080-5084.2005.

Dressler, F, J A Whalen, B N Reinhardt, and A C Steere. "Western Blotting in the Serodiagnosis of Lyme Disease." The Journal of infectious diseases 167, no. 2 (1993): 392-400.

Latov, Norman, Anita T Wu, Russell L Chin, Howard W Sander, Armin Alaedini, and Thomas H Brannagan. "Neuropathy and Cognitive Impairment Following Vaccination with the Ospa Protein of Borrelia Burgdorferi." Journal of the peripheral nervous system : JPNS 9, no. 3 (2004): doi:10.1111/j.1085-9489.2004.09306.x.

Straubinger, R K, B A Summers, Y F Chang, and M J Appel. "Persistence of Borrelia Burgdorferi in Experimentally Infected Dogs After Antibiotic Treatment." Journal of clinical microbiology 35, no. 1 (1997): 111-6.

Wormser, G P, C Carbonaro, S Miller, J Nowakowski, R B Nadelman, S Sivak, and M E Aguero-Rosenfeld. "A Limitation of 2-Stage Serological Testing for Lyme Disease: Enzyme Immunoassay and Immunoblot Assay Are Not Independent Tests." Clinical infectious diseases : an official publication of the Infectious Diseases Society of America 30, no. 3 (2000): doi:10.1086/313688.

Minor edits added August 6, 2011

Posted by Joel Spinhirne
(19) Comments | Permalink
Next entry: Paul Krugman and the IDSA Previous entry: Maureen Dowd Is Going To Talk To Her Doctor

Post a comment


Email (required, not posted):


Remember my personal information

Notify me of follow-up comments?