To determine the frequency and specificity of antibodies to B. burgdorferi (as measured by ELISA) in patients with endocarditis, we studied patients who resided in an area devoid of B. burgdorferi-infected ticks. We also assessed whether increased levels of antibodies to B. burgdorferi in patients with endocarditis were attributable to rheumatoid factor often present in patients with endocarditis.

Methods

Top Methods Results Discussion Author & Article Info References

All patients had nonspirochetal, subacute bacterial endocarditis (proven by blood culture) diagnosed at Cornell Medical Center between 1979 and 1981. All resided in areas not endemic for Lyme borreliosis. No patient had a history of deer tick bite or erythema migrans rash. These patients with endocarditis were diagnosed before the advent of Lyme serologic tests, and none had their illness attributed to Lyme disease. Sera obtained from these patients during their endocarditis illness were stored at –40°C. Anti-B. burgdorferi antibodies were detected by ELISA using sonicated, whole B31 strain as previously described [13]. A convenience sample of frozen sera collected between 1989 and 1990 from asymptomatic persons residing in Suffolk County, New York, were used as negative controls (n = 30); this baseline control value was 1.8 ± 0.6 units. Sera were rated positive if they were more than 4.0 units (>3 SD above the mean of the controls). In the figures, data are reported as the ratio of the sample optical density to the negative cut-off value of 4.0 units.

Immunoblotting was done with these sera using the laboratory-adapted B31 strain of B. burgdorferi as previously described [13]. Spirochetes were sonicated in phosphate-buffered saline containing 1 µmol/L p-methyl sulfinylfluoride and were then concentrated and boiled in 2.5% sodium dodecyl sulfate/2.5% ß-mercaptoethanol for 3 minutes. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transfer to nitrocellulose (0.2 pore size; Hoeffer, San Francisco, California) were done according to standard methods [14]. Samples for specific IgG were incubated at a 1:200 dilution of sera, and samples for specific IgM and IgA were incubated at a 1:50 dilution. Bound antibodies were detected with a 1:5000 dilution of goat anti-human -, µ-, or -specific IgG conjugated to alkaline phosphatase and were developed with BICIP-Substrate System (5-bromo-4-chloro-3-indole phosphate, nitroblue tetrazolium; Kirkegaard and Perry Laboratories, Gaithersburg, Maryland). Reactions were stopped by rinsing the blots in distilled water. Immunoblots were interpreted to be diagnostic of Lyme disease according to published criteria [15] (that is, positive immunoblots have distinct 41-kd bands and at least four other defined bands).

IgM rheumatoid factor was quantified by a solid-phase radioimmunoassay using iodine 125-labeled F (Ab)₂ anti-µ reagents to detect IgM rheumatoid factor bound to solid-phase, adsorbed, heat-aggregated human IgG, as previously described [16]. A positive result was defined as more than 2.2% binding, a value 2 SD above the mean (10 normal controls had a mean = 1.8% ± 0.2 SD). Values are reported in the figures as the ratio of the sample percentage binding to the negative control cut-off binding of 2.2%.

Rheumatoid factor was also measured in two additional groups. IgM rheumatoid factor was measured by latex agglutination (Behring) in a convenience sample of frozen sera collected between 1991 and 1992 at the Stony Brook Lyme Center from 15 patients with seropositive Lyme disease who met criteria for Lyme disease established by the Centers for Disease Control and Prevention [17]. In addition, consecutive sera were collected from 20 patients with rheumatoid arthritis who were positive for rheumatoid factor and were treated in a private rheumatology office practice in Suffolk County between August 1991 and January 1992. Sera were tested for antibodies to B. burgdorferi by ELISA. All patients with rheumatoid arthritis fulfilled at least four of seven disease criteria of the American College of Rheumatology [18] and were positive for IgM rheumatoid factor using a latex agglutination test. None had coincident bacterial infection. All resided in Suffolk County, New York, an area endemic for Lyme disease, but none had a history of an erythema migrans rash.

Statistical Analyses

Values are presented as mean ±SE. Comparisons between dichotomous values were done using chi-square analysis with the Yates correction. For continuous measures, t-tests and Pearson correlations were used.

Results

Top Methods Results Discussion Author & Article Info References

Thirteen of 30 patients with subacute bacterial endocarditis (43%) had a positive ELISA result (> 4 units) for B. burgdorferi, whereas 3 of the 30 controls (10%) had ELISA values greater than 4.0 units (P < 0.01). The mean of the B. burgdorferi antibody-positive sera was 7.52 units in the patients with endocarditis (range, 4.2 to 19.8 units). The mean of the normal controls was 1.8 units (range, 0.1 to 10 units). The causative organisms in the 30 patients with endocarditis included a broad spectrum of streptococci in addition to five Staphylococcus aureus isolates and one isolate each of Staphylococcus epidermidis and Lactobacillus casei. No preponderance of any one organism was noted in patients with endocarditis who were either positive or negative for B. burgdorferi antibody.

Immunoblot analysis of 22 of the 30 sera are shown in Figure 1. Positive Lyme sera are positioned in lanes 1 and 24 (lane 24 is serum from a previously reported patient with both endocarditis and Lyme disease [1]). The three normal control sera from patients with increased antibodies for B. burgdorferi (using immunoblot tests) showed only isolated 41-kd reactivity (data not shown). Of the 13 patients with endocarditis who had positive ELISA results for B. burgdorferi antibodies, 1 (lane 19) had high antibody levels (19.8 units), and the immunoblot of this serum showed multiple bands, including a strong 41-kd band consistent with, but not diagnostic of, previous exposure to B. burgdorferi [15]. Five of the remaining 12 patients (42%) had 60-to 67-kd reactivity (Figure 1; lanes 7, 18, 19, 22, and 23). An additional dark 41-kd reactivity line, seen in almost all seropositive Lyme patients, was not seen. The weak anti-41-kd reactivity seen in many of the sera from patients with endocarditis is indistinguishable from the faint anti-41-kd bands seen in sera from normal patients [13]. Seven of 12 patients had positive ELISA results (mean, 7.86 units), yet 3 (43%) of these had no strong bands on immunoblots. These three immunoblots, despite positive ELISA results, were indistinguishable from normal negative controls. Overall, 12 of 30 (40%) patients with endocarditis who had increased antibodies to B. burgdorferi, as measured by ELISA, were falsely positive. This represents a specificity of only 60% (95% CI, 42% to 78%). Of our 30 patients with endocarditis, sufficient sera were available from 13 patients to measure IgM rheumatoid factor. Eleven of the 13 patients (85%) tested positive for quantitative IgM rheumatoid factor, but the titers varied widely. Figure 2 shows the comparison between normalized levels of IgM rheumatoid factor in the endocarditis groups positive (n = 5) and negative (n = 8) for B. burgdorferi antibody. Although the mean levels of IgM rheumatoid factor were slightly higher in the group seropositive for antibody to B. burgdorferi (n = 5; mean, 5.6 ± 1.15 units) than in the seronegative group (n = 8; mean, 3.96 ± 0.90 units), the differences were not statistically significant (P > 0.2). No correlation was found between the IgM rheumatoid factor level and the B. burgdorferi antibody level for both groups combined (R = 0.2; P > 0.2).

View larger version (97K): [in this window] [in a new window]   Figure 1. Immunoblot analysis of sera from 23 of 30 patients with nonspirochetal bacterial endocarditis who reside in nonendemic areas for Lyme disease. Sera was blotted against solubilized Borrelia antigen (B. burgdorferi strain 31) and was probed with anti-human IgG, -chain-specific goat antibody conjugated to alkaline phosphatase. A positive control serum sample in lane 1 shows an expanded humoral response in a patient with newly diagnosed chronic Lyme disease. Thirteen of the sera from patients with carditis yielded positive ELISA results for B. burgdorferi (lanes 3, 4, 6, 7, 8, 12, 13, 16, 17, 18, 19, 22, and 23). Immunoblots from the remaining 17 patients with endocarditis, who had nonreactive ELISA results for B. burgdorferi, were indistinguishable from negative control results (data shown for nine patients).

View larger version (15K): [in this window] [in a new window]   Figure 2. Relation between IgM rheumatoid factor and B. burgdorferi antibody levels in 13 patients with endocarditis. The values are shown for quantitative IgM rheumatoid factor and anti-Borrelia ELISA indices that represent the ratio of the sample values to the negative cut-off value. No significant correlation was noted between IgM rheumatoid factor and B. burgdorferi antibody levels (Lyme antibody index) using linear regression analysis (R = 0.2; P > 0.2).

View larger version (12K): [in this window] [in a new window]   Figure 3. Relation between IgM rheumatoid factor and B. burgdorferi antibody levels. Left. In 15 patients from Suffolk County with seropositive Lyme disease. Right. In 20 patients with documented rheumatoid arthritis. Rheumatoid factor and Borrelia antibody levels are shown as in Figure 2.

Discussion

Top Methods Results Discussion Author & Article Info References

The diagnosis of noncutaneous Lyme borreliosis is based on the serologic confirmation of previous exposure to B. burgdorferi plus clinical manifestations suggestive of Lyme borreliosis [1, 2, 19]. Such manifestations are defined by the Centers for Disease Control and Prevention for the purpose of detecting patients with Lyme borreliosis (National Surveillance of Lyme borreliosis) [17]. Lyme borreliosis shares with other illnesses nonspecific symptoms such as fatigue, malaise, arthralgia, and myalgia [19-22]. In such patients who do not meet strict criteria for Lyme borreliosis, the use of serologic assays for antibodies to B. burgdorferi has been advocated [23]. However, our data suggest that positive B. burgdorferi serologic test results must be interpreted with caution.

In our study, 43% of patients (CI, 25% to 61%) with nonspirochetal endocarditis produced antibodies reactive to B. burgdorferi that were detectable by ELISA using whole-sonicated B. burgdorferi. Positive serologic test results for B. burgdorferi appeared to be related to antibodies that were cross-reactive to shared epitopes between B. burgdorferi and the endocarditis organisms. Importantly, none of our 30 documented patients with endocarditis satisfied the immunoblot criteria for definitive B. burgdorferi exposure [15]. Only one patient with endocarditis had an immunoblot compatible but not diagnostic of B. burgdorferi exposure.

The spectrum of antibodies reactive against B. burgdorferi as detected on immunoblot is diverse and includes antibodies to the 60-kd protein. Fifty-eight percent (8 of 13) of our patients with endocarditis who were positive for B. burgdorferi had antibodies reactive to the 60-kd protein on immunoblot. The 60-kd groEL antigen of B. burgdorferi is homologous to the 65-kd heat-shock protein of mycobacteria [24]. Epitopes within this antigen are widely shared by many bacteria and eukaryotic cells [25]. Therefore, antibodies to this antigen are nonspecific and do not confirm previous exposure to B. burgdorferi. Bacteria-specific regions of the B. burgdorferi 60-kd heat-shock protein have recently been reported and may form the basis of a more specific B. burgdorferi assay [26].

Only 1 of 13 patients with endocarditis who were seropositive for B. burgdorferi had antibodies strongly reactive to the 41-kd protein. This serum also had reactivity to other Borrelia antigens, perhaps suggesting true exposure to B. burgdorferi [19, 27]. Epitopes within the 41-kd endoflagellar protein are conserved among other spirochetes including commensal mouth treponemes [28]. Low-level reactivity to this protein is frequently seen in sera of normal, non-Lyme persons [13]. This reactivity may represent polyclonal B-cell activation or anamnestic response or both to either the endoflagellar protein of B. burgdorferi, or related intraoral spirochetes [28].

Three of seven patients with positive ELISA test results to B. burgdorferi had negative results using immunoblot analysis (that is, they had patterns indistinguishable from normal controls). This phenomenon may be due to antibodies reactive to capsular lipopolysaccharides or peptidoglycans that are not detected using immunoblot analysis. Such reactivity has been shown for antibodies to streptococcal organisms [29]. Attempts to detect lipopolysaccharides in B. burgdorferi have been unsuccessful [30]. However, peptidoglycans are present in B. burgdorferi [31] and may account for positive ELISA and negative immunoblot test results.

Increased rheumatoid factor is often present in patients with endocarditis [32]. Rheumatoid factor has been associated with false-positive serologic test results for B. burgdorferi measured by immunofluorescent assay [33, 34]. However, one study of patients with rheumatoid arthritis did not find an association between rheumatoid factor and antibodies to B. burgdorferi [35]. Similarly, we found no correlation between IgM rheumatoid factor levels and antibodies to B. burgdorferi in that only 1 of our 20 patients with rheumatoid arthritis who were positive for rheumatoid factor also had a positive ELISA test result for B. burgdorferi (Figure 3, right). This patient had an immunoblot compatible with, but not diagnostic of, previous B. burgdorferi infection [26]. Thus, a positive test result for IgM rheumatoid factor does not appear to yield a positive ELISA result for B. burgdorferi in the absence of true infection. Conversely, our patients with Lyme disease and high-titer antibodies to B. burgdorferi tested negative for rheumatoid factor (Figure 3, left). Although transient positive test results for rheumatoid factor have been reported in a small number of patients with Lyme arthritis [36, 37], our data suggest that this positive result does not persist. This is consistent with the absence of rheumatoid factor reported in 80 patients with frank Lyme arthritis [38].

The rate of false-positive test results for B. burgdorferi antibodies found in our patients with endocarditis highlights the hazards associated with assuming that the specificity of a test established in a healthy control population is applicable to persons with clinical illness that can be confused with Lyme disease [1]. Although a patient with endocarditis who has a positive ELISA test result for B. burgdorferi may be a "true" positive if immunoblot analysis also confirms previous exposure to B. burgdorferi, a positive serologic result alone does not ensure that the current clinical problem is due to Lyme borreliosis [1, 39]. The advent of immunoblot analysis [25], standardized ELISA procedures using selected pathogen-specific antigens of B. burgdorferi [40-42], and the application of polymerase chain reaction technology to detect B. burgdorferi DNA [43] do not circumvent the difficulty in interpreting the clinical significance of positive test results.