Methods

Top Methods Results Discussion Author & Article Info References

Samples from patients and controls were obtained over a 10-month period from both the university hospital and from clinicians in Minnesota and Wisconsin, areas where the background seroprevalence of B. burgdorferi among healthy individuals is 1% to 2%. Lyme arthritis was suspected in patients with mono- or oligoarticular large joint involvement, seropositivity for B. burgdorferi, and no other known underlying disease. Controls were from the same areas. They presented with various arthritic processes (see Results) and were having arthrocentesis of involved joints. Synovial fluid was processed in a building where B. burgdorferi had never been cultivated. Synovial fluid (0.5 to 1.0 mL) was centrifuged at 16 000 g for 15 minutes, and the pellet was resuspended in 200 µL of supernatant fluid. One half was used for PCR, and the other was cultured as described previously [13] but with 0.1% agar. Nucleic acids were guanidium isothiocyanate-extracted and subjected to PCR [10] using primers 991 and 992 corresponding to nucleotides 16-43 and 222-247, respectively, of a chromosomal sequence that amplifies 231 bp of B. burgdorferi DNA [10]. The PCR products were subjected to electrophoresis in agarose and transferred to nylon membranes. A 167-bp digoxigenin-11-deoxyuridine triphosphate probe (nucleotides 81-247, reference [10]) was used for Southern hybridization as specified by the manufacturer (GENIUS kit, Boehringer-Mannheim, Indianapolis, Indiana). Confirmation studies used two different primer pairs (A2, A4 and A149, A319) and internal probes for the plasmid-encoded B. burgdorferi outer surface protein A gene [11]. Study personnel were blinded to the suspected disease for all but one patient sample. Immunoglobulin G antibody against B. burgdorferi was measured by enzyme immunoassay with seropositivity defined as optical density values 3 standard deviations above the mean for 200 blood donors.

Results

Top Methods Results Discussion Author & Article Info References

Synovial fluid from six of seven (86%; 95% CI, 42% to 100%) persons thought to have Lyme arthritis was positive by PCR for B. burgdorferi DNA. Results from three of these (samples 5, 8, and 12) are shown (Figure 1). Results of the following tests were negative for all seven patients: rheumatoid factor, antinuclear antibodies, examinations for crystals, and aerobic and anaerobic cultures. All cultures for B. burgdorferi were negative.

View larger version (106K): [in this window] [in a new window]   Figure 1. Southern blot of polymerase chain reaction (PCR)-amplified samples from patients and controls. Synovial fluid samples 1, 5, 8, and 12 are from four patients with presumed Lyme arthritis. Sample 11 is a follow-up from the same patient who provided sample 12 but was obtained while the patient received antibiotic treatment. Positive control samples included 20 to 2000 B. burgdorferi organisms (top left); patient samples 1 to 4 were initially negative but then each spiked with 20 organisms (bottom left). Negative control samples included those from patients with other rheumatologic disorders (samples 2, 3, 4, 6, 7, 9, 10, and 13), human serum (sample 14), and laboratory controls in which known negative synovial fluid was extracted (X), or water, instead of sample DNA, was added in the laboratory (L) or to the final PCR master mix itself (N).

Case Histories

Sample 5

A 42-year-old man with erythema migrans was treated with oral tetracycline for 2 weeks, and the rash resolved. Six years later, he developed knee pain and swelling, was found to be B. burgdorferi seropositive, and was treated with oral doxycycline for 3 months without improvement. After 2 months of treatment, synovial fluid showed a leukocyte count of 16.7 x 10⁹/L (79% granulocytes) and was positive for B. burgdorferi by PCR. He was treated with intravenous ceftriaxone for 8 weeks, and symptoms resolved.

Sample 8

A 30-year-old woman reported 4 months of knee pain and swelling and was seropositive for B. burgdorferi. A synovial fluid specimen showed a leukocyte count of 37.8 x 10⁹/L (95% granulocytes) and was PCR positive. She was treated with oral doxycycline for 3 months, and symptoms resolved.

Sample 12

A 20-year-old man developed bilateral facial nerve palsies and lymphocytic meningitis. One year later, he developed fatigue, knee pain and swelling, and B. burgdorferi seropositivity. He received oral doxycycline therapy for 3 months and showed partial improvement. One month later, a large effusion developed and he had difficulty in walking; a synovial fluid sample showed a leukocyte count of 12.8 x 10⁹/L (90% granulocytes) and was positive for B. burgdorferi by PCR. Another specimen (sample 11), obtained 2 months later while the patient was receiving another course of doxycycline, showed a leukocyte count of 2.7 x 10⁹/L (only 15% granulocytes) and was negative by PCR.

Other Samples (not shown)

Sample 16: A 36-year-old man had knee pain and swelling and was seropositive for B. burgdorferi. Synovial fluid analysis showed a leukocyte count of 8.0 x 10⁹/L (predominantly granulocytes) and a positive PCR test result. He received oral doxycycline for 2 months, and symptoms resolved.

Sample 22: A 54-year-old woman had knee pain and effusion and was seropositive for B. burgdorferi. A synovial fluid specimen showed a leukocyte count of 54.1 x 10⁹/L (87% granulocytes). She is improving with oral doxycycline.

Sample 25: A 28-year-old man had tick exposures, no history of rash, and unexplained fevers with myalgias 16 months before the sample was taken. For 1 year he experienced worsening monthly episodes of arthritis of the knee, and he became seropositive for B. burgdorferi. Physical examination showed elbow, knee and ankle effusions. A synovial fluid specimen from the knee showed a leukocyte count of 26.8 x 10⁹/L (82% granulocytes) and was positive by PCR. He is currently receiving doxycycline.

One PCR-negative sample (sample 1) was obtained from a seropositive patient who was treated for probable chronic Lyme arthritis. This patient had an unexplained febrile illness several months before arthritis of the knee developed. After treatment with oral penicillin for 3 months and intravenous penicillin for 1 month, symptoms resolved. Five and 7 years later, arthritis recurred and the patient responded to intravenous penicillin and oral doxycycline, respectively. Synovial fluid obtained after the last treatment course showed a leukocyte count of 5.9 x 10⁹/L (all lymphocytes).

Summary

Six of seven patients with Lyme arthritis were positive by PCR. In contrast, all 18 synovial fluid samples from patients with other disorders, including rheumatoid arthritis, spondyloarthropathy, gout, pseudogout, hemarthrosis, degenerative joint disease, lupus, papillary synovitis, and trauma, were negative by PCR (P < 0.001, Lyme arthritis compared with controls, Fisher exact test). All 38 laboratory controls were negative by PCR. The assay reproducibly detected 20 or fewer B. burgdorferi cells directly or when added to extracted synovial fluid that was previously negative by PCR. Polymerase chain reaction was done four times with identical results, including analyses with both outer surface protein A primer sets.

Discussion

Top Methods Results Discussion Author & Article Info References

Six of 7 prospectively studied patients with Lyme arthritis had spirochetal nucleic acids present in affected joints, despite negative cultures. The fastidious techniques, the many controls with negative PCR results, and the confirmation of results with both chromosomal and plasmid targets make us confident there were no false-positive test results. Detection of organisms when PCR-negative samples were spiked assures against false-negative test results.

The number of organisms or equivalents detected by PCR was high in several patients (10^3/mL to 10^4/mL). Thus, the failure to cultivate spirochetes was not always caused by insufficient numbers. The discrepancy between PCR results and culture results suggests that organisms present could be injured, dead, or otherwise inhibited from multiplication, a situation that may help explain why some patients have apparent clinical resistance to antimicrobial agents.

Our results show the intra-articular persistence of B. burgdorferi nucleic acids in Lyme arthritis and suggest that persistent organisms and their components are important in maintaining ongoing immune and inflammatory processes even among some antibiotic-treated patients. Further studies are needed to determine the microbiologic state of these organisms and their therapeutic and prognostic implications.

Grant support: By NIH grants AR40448 and AI29739.