Role of Serology in the Diagnosis of Lyme Disease

Lyme disease was first described after a cluster of cases initially diagnosed as juvenile rheumatoid arthritis was reported in Lyme, Conn.¹ The newly identified tickborne spirochete, Borrelia burgdorferi, was isolated from the blood, skin, and cerebrospinal fluid of patients in the early 1980s.^{2 - 3} An IgM and IgG response to this spirochete was observed by immunofluorescence antibody (IFA) technique,² forming the basis for serologic tests for antibody to B burgdorferi.

Since 1987, the US Food and Drug Administration (FDA) has cleared 58 serologic assays for B burgdorferi for commercial marketing. The regulatory route for clearance is through the 510(k) premarket notification process that requires the manufacturer to provide information and data to demonstrate that the new test is substantially equivalent and as safe and effective as a test already cleared by the FDA.⁴ Forty-five assays from 15 manufacturers are first-step assays and use enzyme-linked immunosorbent assay (ELISA), IFA, or immunodot techniques for detection of IgG, IgM, or total anti-B burgdorferi antibodies. Eight second-step supplemental⁵ Western blot IgG and IgM assays from 4 manufacturers have been cleared. The FDA has cleared tests for detection of antibodies to B burgdorferi in blood, serum, or plasma. No test using other body fluids, such as urine or cerebrospinal fluid, have been cleared by the FDA.

The FDA discussed concerns about the accuracy of these assays with the FDA Microbiology Advisory Panel and industry in an open public meeting in 1996.⁶ The FDA then issued a public health advisory alerting clinicians and clinical laboratories to the potential for misdiagnosis of Lyme disease using commercial assays for detecting antibody to B burgdorferi.⁷ The advisory stressed the use, limitations, and interpretation of these assays for supporting a clinical diagnosis of Lyme disease. The agency also promoted the 2-step laboratory testing algorithm recommended by the Second National Conference on Serologic Diagnosis of Lyme Disease.⁵ Two-step testing requires the use of a second test (Western blot) for positive or equivocal first-step test results (ELISA, IFA, immunodot). In October 1997, the FDA sent a letter to manufacturers of ELISA, IFA, and immunodot testing assays requesting that they modify the labeling to reflect current information on appropriate use and interpretation of antibody assays for B burgdorferi.

ELISA, IFA, or immunodot IgG, IgM, or total antibody assays are qualitative first-step assays. All positive or equivocal results therefore are presumptive and should not be reported until a more specific second-step (Western blot) is completed. A negative result, whether first- or second-step, indicates that there was no reliable serologic evidence of B burgdorferi infection present at the time the specimen was collected. A positive second-step result indicates that there is either serologic evidence of past or current infection with B burgdorferi or presence of cross-reacting antibodies.

The FDA is concerned about the inability of B burgdorferi commercial antibody test kits to serve as the primary basis for diagnosing early or late Lyme disease. The variable performance of these test kits is also a problem (Table 1). The purpose of this article is to clarify the clinical value of serologic testing for Lyme disease by providing examples of problems with serologic diagnosis of Lyme borreliosis and the consequences that may result.

Three genospecies of Borrelia are known to cause Lyme disease and are collectively referred to as B burgdorferi sensu lato.⁸ The primary North American vector for Borrelia is the black-legged tick from the genus Ixodes. The Centers for Disease Control and Prevention (CDC) surveillance definition of Lyme disease is the presence of an erythema migrans rash 5 cm or greater in diameter or laboratory confirmation of the infection by culture with evidence of at least 1 manifestation of musculoskeletal, neurologic, or cardiovascular disease.⁹ The clinical manifestations of Lyme borreliosis may occur in early localized disease (days to a month after tick bite), early disseminated disease (days to 10 months after tick bite), or late disease (months to years after tick bite). Early localized disease includes erythema migrans and viral flulike symptoms (fever, malaise, headache, stiff neck, and fatigue). Early disseminated disease may result in severe systemic symptoms, including progression to cardiac abnormalities such as atrioventricular block or myopericarditis, neurologic manifestations including acute meningitis or encephalitis, and development of arthritis, which is characteristically episodic and occurs in large joints.^{8 ,10} In 25% to 50% of individuals, early disseminated disease is not preceded by recognized erythema migrans.⁸ Late disease occurs months to years after tick bite and may present as migratory polyarthritis or chronic monoarthritis (usually in the knee), as well as central nervous system manifestations or cutaneous manifestations including sclerodermalike lesions.

Disease manifestations may be mild or severe and may occur within days, weeks, or months after the initial infection. The characteristic skin lesion (erythema migrans) may occur within 2 days to 2 weeks after exposure⁸ and is considered diagnostic for infection with B burgdorferi.^{9 ,11} Because of the protean manifestations of this disease, in the absence of erythema migrans, clinicians rely on serologic testing for IgM or IgG antibodies for evidence of infection with B burgdorferi. Limitations of commercially available kits used in clinical laboratories make them inappropriate as the primary diagnostic criteria for Lyme disease. Serologic testing for antibodies to Borrelia should only be used to support a clinical diagnosis of Lyme disease.

First-step immunoassay test kits cleared by the FDA use either ELISA, IFA, or immunodot technology. The antigen preparation is typically a lysate prepared from a tissue culture strain of B burgdorferi.^{12 - 13} The lysate is a complex mixture containing several immunogenic proteins, lipoproteins (eg, outer surface proteins [Osp] A-G), and carbohydrates.^{13 - 14} In IFA, the antigen is whole B burgdorferi organisms fixed to slides.¹³ Unlike human immunodeficiency virus tests, the commercially available serologic tests for antibody to B burgdorferi are not standardized against a reference serum specimen panel.¹¹

Performance deficiencies of ELISA (or other first-step tests), coupled with their use in a population with a low pretest probability of Lyme disease, is a serious concern. The positive predictive value (the ability to diagnose disease) of a test is dependent on the validity of the assay (sensitivity and specificity) and the prevalence of the disease in the population to which the test is being applied.¹⁵ Table 2 provides an example of the influence of sensitivity of the test and disease prevalence on the positive predictive value of a test.

Current CDC recommendations call for confirmation of positive or equivocal ELISA results by Western blot. The antigen preparation is lysate that contains a range of B burgdorferi antigens separated into distinct molecular weight proteins by electrophoresis. The ability of Western blot to detect antibodies to individual proteins, including some specific for B burgdorferi, reduces the number of false-positive test results obtained with ELISA.¹⁶

Table 1 is a compilation of results obtained from data submitted to the FDA by 6 first-step test manufacturers and all 4 Western blot manufacturers after testing the CDC masked panel of serum specimens from patients with Lyme disease. Only those manufacturers responding to the FDA letter are included in this table. The FDA is requiring that this information be included in product labeling. This information can be used by laboratories or physicians to compare performance of tests from different manufacturers. The results of the annual Lyme Disease Survey from the College of American Pathologists can also be used for this purpose.

False-negative test results may occur during the first few weeks after infection before the production of a humoral response or when the level of antibody is below the analytical sensitivity of the test. In first-step IgM assays, false-positive test results may be due to cross-reacting antibodies to B burgdorferi in patients with other tickborne diseases such as relapsing fever (Borrelia spp) and an ehrlichiosis infection, autoimmune disease, or viral (Epstein-Barr virus and others), musculoskeletal, and neurological diseases (eg, chronic encephalomyelitis). False-positive test results may occur when cross-reacting IgG antibodies to B burgdorferi are present in patients with syphilis, an ehrlichiosis infection, babesiosis, Helicobacter pylori, or musculoskeletal disease (eg, systemic sclerosis or systemic lupus erythematosus).^{13 ,16 - 18}

Several studies have described comparisons of commercial test kits. One such study tested diagnostic kits from 5 manufacturers on 53 serum specimens, 25 of which were from patients with symptoms of Lyme disease.¹² The sensitivity of ELISA kits for serum specimens collected early ranged from 58% to 92%, and for late serum specimens, all had sensitivities of 77%. In this high-prevalence panel, the specificity of the ELISA kits varied from 83% to 100%. A similar study indicated that concordance between test kits was very low.¹⁹ Given the results seen with different manufacturers' test kits, interlaboratory results predictably show poor agreement. A test in 4 laboratories, which used either IFA or ELISA, indicated that neither interlaboratory nor intralaboratory testing were reliable.²⁰

In interlaboratory testing with a panel of 18 serum specimens from diagnosed cases sent to 9 laboratories, positive serologic test results were reported for between 44% to 100% of the cases.²¹ In this study, discordant results were obtained in half of the laboratories on paired aliquots, indicating that intralaboratory testing of the same sample may yield different results. Recent studies have indicated that neither sensitivity nor specificity have improved and that there is a continued inability in some laboratories to reproduce results on split aliquots.^{22 - 23} These findings are troubling and reinforce the maxim that serologic testing is of little value in the absence of clinical findings suggestive of Lyme disease.

As described in detail by Tugwell et al,¹⁵ when the pretest probability of Lyme disease is low, testing will result in more false-positive than true-positive test results. They recommended no serologic testing if the pretest probability is below 0.20 or above 0.80 and described the following examples of each case. If a patient has nonspecific signs and symptoms of illness such as headache, fatigue, myalgia, or arthralgia, then their pretest probability of Lyme disease will be low (<0.20) even in highly endemic areas. In this case, Tugwell et al recommend treating the symptoms. If the patient presents with a rash resembling erythema migrans, or with arthritis, history of rash resembling erythema migrans, and a previous tick bite, then their pretest probability is high (>0.80). Serologic testing is not required and antibiotic therapy is recommended.¹⁵ Patients presenting without erythema migrans or a history of erythema migrans but with other objective clinical signs such as recurrent attacks of marked painful swelling affecting 1 or a few large joints followed by periods of remission, serologic testing may be useful. In this case, the pretest probability of Lyme disease is likely between 0.20 and 0.80 and can be further refined by considering prevalence of Lyme disease. It is these intermediate cases in which serologic testing for Lyme disease is most helpful.¹⁵

Erythema migrans is diagnostic when there is a history of tick bite (or exposure in an endemic region) and systemic signs and symptoms are consistent with Lyme disease.^{9 ,11 ,15} Serologic tests do not become positive until weeks after a tick bite.²⁴ Therefore, when a patient presents within a few days of a tick bite, serologic tests for B burgdorferi are unnecessary and unreliable (Table 1). The positive predictive value of the test is higher in patients known to have the disease who are tested at more than 12 months after onset of the disease (Table 1). Later serologic testing may be negative if patients were treated with antibiotics.^{16 ,25} Without erythema migrans, positive 2-step serology indicates only exposure and does not indicate whether viable spirochetes are present.

Patient anxiety may contribute to the inappropriate use of testing and treatment.¹⁶ A study of the use of serologic testing in an endemic area indicated that 67% of patients presenting with tick bite underwent serologic testing; of these, 98% had negative or equivocal results.²⁶ None of the patients with tick bite developed Lyme disease. However, 55% of those presenting with a history of tick bite were treated with antibiotics, regardless of whether serologic testing was performed.

Recent concerns about the overuse of antibiotics and subsequent emerging resistant strains of a wide variety of microorganisms should be considered when prescribing antibiotics that may be of questionable or limited value to the patient. In addition to public health concerns, the overdiagnosis of Lyme disease results in increased antibiotic-related adverse events, distress in patients when their symptoms are not relieved, and can be expensive,^{27 - 28} particularly when intrathecal administration is used. Especially alarming is that incorrect diagnosis and treatment may delay a correct diagnosis and appropriate treatment. For instance, incorrect diagnosis of Lyme disease in a patient with babesiosis may delay the use of the appropriate antibiotic regimen.¹⁶

The incorrect diagnosis of Lyme disease was reported to have occurred in 452 (57%) of 788 patients referred to a Lyme disease clinic.²⁹ Of the 452 patients who did not have Lyme disease, 45% (302) had a positive serologic test result prior to referral. In nearly 80% of patients treated with an antibiotic prior to referral, the reason for failure to respond to treatment was disease misdiagnosis. This is a graphic example in which false-positive, first-step test results likely contributed to large-scale misdiagnosis and treatment.

Recent progress in developing Lyme disease vaccines in which the recombinant B burgdorferi protein, OspA, is the immunogen^{30 - 31} has raised additional concerns. For example, ELISA-based assays that use lysate with the full range of antigens will not distinguish vaccinated from infected individuals. Zhang et al³² reported results with a lysate from a natural variant of B burgdorferi that does not produce OspA or OspB. Individuals immunized with OspA were negative in this test, but had false-positive test results with a conventional ELISA using lysate from a strain of B burgdorferi with a full complement of antigens, including OspA. Serum specimens from infected individuals were positive in both assays. The mean titers were the same in both assays, indicating that absence of OspA did not reduce the analytical sensitivity of the ELISA using lysate without OspA.

Use of tests constructed with lysate from appropriately selected variants of B burgdorferi may mitigate or eliminate the problem of distinguishing vaccinated from infected individuals by ELISA. However, to date, such tests have not been cleared by the FDA, are not commercially available, and other false-positive reactions owing to antibody cross-reactivity would be likely to persist.

Physicians are encouraged to report adverse events related to clinical laboratory assays to MEDWATCH, the voluntary reporting program of the FDA. Pertinent adverse events could include anti-B burgdorferi test results that led to inappropriate patient management or a problem with quality control of an assay.

Adverse events may be reported to the FDA via the MEDWATCH program by telephone at 1-800-FDA-1088, by fax at 1-800-FDA-0178, via the MEDWATCH Web site at http://www.fda.gov/medwatch, or by mail at MEDWATCH, HF-2, FDA, 5600 Fishers Ln, Rockville, MD 20852-9787.

Results from commercially available ELISA, IFA, or immunodot test kits used in clinical laboratories for the detection of antibody to B burgdorferi cannot reliably be used as the primary diagnostic criterion for Lyme disease. The correct use for these tests is to support the clinical diagnosis of Lyme disease. A positive or equivocal first-step result on an ELISA (or IFA or immunodot) assay requires supplemental testing with a Western blot. A negative result on the Western blot or ELISA indicates that there is no serologic evidence of infection by B burgdorferi at the time the sample was drawn.

Erythema migrans is diagnostic with a history of tick bite (or exposure in endemic regions) and clinical signs and symptoms consistent with Lyme disease. Early serologic testing based only on tick bite or potential exposure to ticks is an especially dubious practice. This practice will lead to false-positive test results since the specificity of the test kits is low, or to false-negative test results since the patient has insufficient time to develop an immune response to B burgdorferi or because antibody is below the detection limit of the test. Inappropriate use of tests for B burgdorferi may lead to inappropriate treatment of patients and the occurrence of associated adverse events.