Andras Lakos M.D., PhD
Laboratory diagnostic methods in Lyme borreliosis
Lyme borreliosis (Lb) is easily recognisable on the basis of some clinical manifestations. Most cases of erythema migrans (EM), Bannwarth-syndrome with facial palsy as well as acrodermatitis chronica atrophicans do not cause too much diagnostic difficulties for an experienced medical team. However some cases of EM do not show its typical face, moreover many cases of neurological involvement, especially the chronic and subtle encephalopathy accompanied by polyneuropathy as well as the rheumatological form of LB all needs laboratory confirmation. The available laboratory tests can be divided into two groups:
- demonstration of Borrelia burgdorferi sensu lato (Bbsl),
- immonological reaction against Bbsl.
Direct visualisation of Bbsl is rarely successful from blood or CSF in comparison to the tropical disease, febris recurrens. Some experiments have shown that visualisation by dark field microscopy or electronmicroscopy is possible, especially from CSF. Visualisation of Bbsl by dark field microscopy from CSF is most successful during the early phase of neuroborreliosis when antibody reaction has not developed yet. In a consecutive series of 29 cases of Bannwarth's syndrome, we found moving borrelia cells only in 2 CSF samples. The advantage of direct visualisation is the rapidity and the low cost of the examination, but the low sensitivity of this test resulted in a restricted application. Looking for borrelia in blood has a risk in finding filaments seemingly moving and resembling to spirochetes.
Cultivation is the gold standard of every microbiological diagnosis. EM is the easiest target of borrelia cultivation. In an unpublished study, we were able to isolate Bbsl from 14 skin biopsy sample of 15 consecutive, untreated EM patients. Disinfecting of the skin is the most critical point of borrelia isolation. Because of borrelia grow very slowly, contaminating microbes overgrow borrelia and prevent cultivation of the target bacteria. When we used alcohol washing and Arugeen (AVR) or Egisept (EGIS) spray for disinfecting, we had 30% contamination rate. Than we turned to Sagrosept washing and Kodan spray (Schülke & Mayr) we had no contamination in a series of 96 consecutive isolations. It is extremely important to wait until complete drying up of the disinfecting solution. Applying antibiotics in culture media may decrease the rate of contamination but unable to completely prevent it. We have a little experience of using antibiotics in the culture media, this shown that this may decrease the success rate of cultivation. Therefore we suggest avoiding antibiotics in the media. There are some differences between disposable punch biopsy knives made by different manufacturer, but this has no influence on the cultivation success rate. The size of the punch biopsy sample is also critical. We had significantly less success rate of isolation when we used 2 mm size punch biopsy knife instead of our "regular" 4 mm in diameter. When we applied 2 mm punches and we could not find one single borrelia by dark field microscopy after 2 months of incubation, sub-cultures were successful within two weeks. The site of the sampling is also important. The best result can be achieved when biopsy is made inside and close to the advancing border of EM. Culture medium is also a critical point of cultivation success. Several recipes of BSK medium and its modifications have been described. In our experience, selection of manufacturers of the components is more important than the components themselves. Filtration of the media for sterilising is also a determinative step. The best result can be achieved with PES membranes (e.g. Millipore express). Using this membrane we do not need "pre-filtering" the media. Pre-filtration with a draft filter may decrease the amount of important components in the medium. After inoculation of the biopsy sample, the culture tube is incubated on 32-35 C. Incubation time mainly depends on the size of the inoculum, and the isolated strain. From CSF, as we experienced the same in direct visualisation - the success rate of isolation is the best when antibody reaction has not developed yet. If the isolation is successful, before changing the colour of the medium, a faint cloud appears in the liquid. A few days later, the red colour turns to yellow. In some case, borrelia can be detected on the 10th day of incubation, but the duration of incubation time can be as long as 3 months. We usually look for borrelia by dark field microscopy on every 2 weeks. If the 6-week-check shows no borrelia, 1 ml should be inoculated into a fresh media.
There are increasing number of different kind of PCRs. With this examination, we are able to find the genom of a very little number of living or presently dead borrelia cells. This technique allows detecting borrelia on the first 3 days of antibiotic treatment, when the culture is usually ineffective. Theoretically, PCR would be the best diagnostic tool in Lb. As a matter of fact, we really need a good PCR in Lb. Most papers dealing with this topic agreed on that neither specificity nor sensitivity is better than culturing borrelia from skin, CSF or blood. The only and important exception is Lyme arthritis. Borrelia cultivation is rarely successful from synovial fluid. In contrast with culture, PCR is quite sensitive from synovia. This is especially important, because Lyme arthritis is the less effectively diagnosed on the basis of clinical signs and the diagnosis is the most complicating when we want to rely on serology. Moreover, we may experience treatment failure more frequently in Lyme arthritis. With a good PCR, we are able to show borrelia genom weeks or months or even years after the treatment. As it has been proven, a positive PCR some days after finishing the treatment, is equivalent with treatment failure. (Borrelia particles are swept off quite rapidly after a successful treatment.) Some papers have been published on PCR results on urine. Basic idea of this test is that borrelia particle can be isolated from urine. The test - as stated - is parallel with the number living borrelia cells in the whole body, and it is useful tool for control the efficiency of antibiotic treatment. Unfortunately, subsequent studies had not been able to confirm these results.
Serology is the easiest and widely used method for supporting the clinical diagnosis. Unfortunately, serology has serious limitations. Firstly, the test is negative during the first days or weeks of infection. Antibiotic treatment even if it is only partly effective can abort the antibody production. Secondly, the test can be positive long after the successful antibiotic treatment. Because many parts of Europe are highly endemic for Lyme disease, subclinical infection is quite frequent. Examining a high-risk population (e.g. forestry workers) we have to consider spontaneous recovery from Lyme disease. In every-day practice, we have to face with the possibility, that we see the serological scar of an earlier, spontaneously recovered infection as a possible cause of the present symptom. Excluding or proving the causative connection between these actual symptoms and serological reaction (the "positivity") is sometimes easy, but usually not. Thirdly, because of cross-reacting antigens of Bbls, false positive reaction can occur. Moreover, especially in Europe, Bbls is antigenically heterogeneous, and that is only one but probably the most important part of the reasons why the serological tests can not be standardised. We have tested many commercially available kits, and the results were terribly poor. False positivity was as frequent as 20-80 per cent. (This means that 20-80 of one hundred samples of healthy blood donors were positive with an actual test. False positivity in IgM ELISA tests was dominant.)
IgM is less relevant in the diagnosis of Lb than we would expect. In most cases of EM we do not see IgM reaction, even the patient is tested in the very early phase. Interestingly, older women or young children after a longer untreated period of EM (6 to 12 weeks) have a higher chance to produce strong IgM reaction. IgM can reappear or persist years after the first sign of infection. In the suspicion of neuroborreliosis, it is mandatory to demonstrate intrathecal borrelia antibody production. It is easiest with immunoblot (see there). The other tests need several measurements (e.g. total immunglobulin in serum and CSF, total albumin in serum and CSF, or specific immunglobulin against - for example tetanus in serum and CSF). The more measurement you do that gives the more chance for doing mistake.
Immunofluorescens assay (IFA) is the cheapest method. Home-made IFA is quite cost-effective. Unfortunately, IFA is the most difficult test to standardise. Day to day variation is high. Evaluation of the results needs experience. The results can change seriously if the person who reads the test or any of the equipment changes. IFA can not be automated, and evaluation of a big serious of tests is extremely tiring.
Enzyme-linked immunosorbent assay (ELISA) is a rapid, automated, less subjective test. Day-to-day variation however can not be avoided. Most ELISA kits apply whole cell sonicated borrelia. Several efforts have been made to improve Lyme ELISA performance. Absorption of cross-reacting antibodies as well as application of purified or recombinant antigens (instead of whole cell sonicate) may result in a decrease of sensitivity.
Immunoblot (Western blot). The proteins of borrelia are separated by polyacrylamid gel electrophoresis (PAGE). The separated proteins (antigens) are transmitted to a nylon membrane and the immune reaction is performed on this membrane. The basic idea of this test is that we can separately examine the antibody production against the different antigens. In this way, we can omit the cross reactive antibodies during the test evaluation, and we only calculate with those antibody response which has a high specificity ratio. Specificity ratio is calculated as a frequency of an antibody reaction against a particular borrelia protein in a group of clinically determined Lb divided by the frequency of the same antibody reaction amongst the control group.
Because the above-mentioned problem, we strongly suggest before application any commercial or home made Lyme diagnostic test, the first step should be the test of the test. We need 100 serum samples of healthy blood donors (HBD). HBD persons should represent the locality of the population where from most of the patient samples is expected. If the region has high and low endemic populations, HBDs should be tested separately for each region. The cut-off level should be put at level of 99 percentile. It means, that only one per cent is allowed to be above the cut-off level. Applying different cut-off levels is also a good idea. We can use the term of strong positive, week positive or borderline. In this way we can give information for the clinicians of the reliability of the test results. If this method is applied, the description of the test results should contain the information of specificity of each cut off level (see later). The second step of testing the test is calculating the sensitivity. We need 100 clinically diagnosed (by an experienced clinician) EM patients with the exact date of first recognition of EM and if possible, the date of tick bite. The most important factor is the time elapsed from the first sign of EM to the antibiotic treatment/sampling. In this way we can describe the sensitivity of the test by duration of the infection (see later).
Immunoblot has some special features, so we describe it in details.
The critical point in every serological test and especially in Lyme immunoblot is the definition of a "positive" test result. The protein components of Bbsl especially in Europe are highly variable. Even using the same strain, the protein pattern may change during cultivation, sonication or other steps of purification and antigen preparation. Because of these sensitive points, application of a definition of "positivity" drawn from other Lyme immunoblots (e.g. CDC - Centers for Disease Control criteria) can generate tremendously different test results. Calculation of the cut-off should be done for each borrelia immunoblot individually. Because of the lack of appropriate number of serum and CSF samples drawn from clinically well-defined Lyme patients, most of the borrelia immunoblot kits are marketed without any reliable definition of "positivity".
We have developed a Western blot kit. This kit was tested on 100 samples of each of the following 7 clinically defined groups: EM, Bannwarth's syndrome (BS), acrodermatitis chronica atrophicans (ACA), infants, healthy blood donors (HBD), forestry workers and uveitis patients. This latter represents a group of cross-reacting and problem sera what were "positive" in other labs but the clinical symptoms were incongruent with Lyme disease. IgG and IgM were tested separately. The intensity of bands was estimated and scored from 0 to 3. Using discriminate analysis we defined the most specific bands with Borrelia afzelii (ACA1) as antigen.
IgM antibodies against both the 22 kD and 41 kD proteins were 100 % specific. However, IgM testing has a low diagnostic value because only five EM and nine BS patients had IgM but no IgG antibodies. In IgG testing, anti-flagellar (41 kD) and anti-39 kD antibodies were not specific in contrast with many previous studies. We found the 22; 29; 35; 44; 47; 49; and 93 kD bands to be most specific.
The best specificity/sensitivity ratio was found when the cut-off was set up at score 1.5 (one strong plus one faint reaction at the above-mentioned bands). According to these criteria, only three HBDs and no infants were positive. The sensitivity of IgG tests among consecutive EM and BS patients was 40 and 63%, respectively. In BS, when CSF and serum, IgG and IgM were evaluated, the sensitivity was 88%. All 42 BS and 24 of 25 EM patients whose symptoms had been started more than six weeks before sampling were positive. Based on the testing of serum and liquor samples of 700 persons - Lyme patients and controls - the IgG results are qualified according to the 5 following classes: "negative", "borderline", "weakly positive", "positive" or "strongly positive". The specificity of the strongly positive finding is 100 %, that of the positive finding 99 %, that of the weakly positive finding is 95 %.
In comparison to an ELISA test where the cut-off was defined on similar base, our immunoblot kit is more sensitive and more specific at the same time. The sensitivity and specificity were calculated on 1600 tests.
Since 1996, we routinely use as a first and only serological test the immunoblot assay. During this period we have tested eight thousand patients. We found immunoblot superior to ELISA in the following reasons:
||Immunoblot is more sensitive and specific than ELISA is (Figure 1).
||In neuroborreliosis, intrathecal borrelia antibody synthesis can be demonstrated in single step by parallel testing of a serum and a CSF sample (Figure 1, 2 and 3).
||Immunoblot is more sensitive in detecting seroprogression (Figure 4 and 5).
||It is (the only) appropriate method for controlling microbiological cure (Figure 6).
2.5 Many papers have been published on examination of cellular immune response. It has the same limitation as serological tests, moreover it need special equipment accompanied by higher price. Standardisation seems to be even more difficult in comparison to serological tests. These problems explain that these methods remained inside the territories of research laboratories.