An optimized PCR leads to rapid and highly sensitive detection of Borrelia burgdorferi in patients with Lyme borreliosis.

The present study aimed at developing an optimized PCR protocol fro the sensitive and specific detection of all three Borrelia burgdorferi genospecies pathogenic to humans in Lyme borreliosis patients. A rapid DNA extraction method using alkaline lysis was introduced and was found to be superior to other DNA extraction methods. Nested PCR was performed with primer sets targeting the plasmid-located ospA gene and a chromosomal gene segment encoding a 66-kDa protein (p66). In spiked synovial fluid (SF) fewer than three borreliae/sample were detected. The specificities of the amplicons were confirmed by Southern blot analysis with PCR-derived probes. Urine, cerebrospinal fluid (CSF), and SF specimens from 57 patients with Lyme borreliosis and from 58 controls were examined. In clinical samples the diagnostic sensitivity of PCR was 85% with SF samples, 79% with urine samples, and 91% with paired SF-urine samples from patients with Lyme arthritis and was 79% with CSF samples, 45% with urine samples, and 87% with paired CSF-urine specimens from neuroborreliosis patients. One patient each with neuroborreliosis and with Lyme arthritis had PCR-positive urine samples only. In 17% of all cases both primer sets yielded positive results, while the other patients were positive with only one primer set. Among these, more positive results were obtained with the p66 gene primer than with the ospA primer. The specificity exceeded 99%. We conclude that DNA from B. burgdorferi sensu lato species can sensitively and specifically be detected with the optimized PCR method described. At least two different primer sets should be used, and whenever possible, urine and CSF or SF should be analyzed in parallel to achieve maximum sensitivity of the test. This protocol, therefore, considerably enhances the diagnostic power of PCR in patients with B. burgdorferi infection.     

Detection of Borrelia burgdorferi DNA in urine samples and cerebrospinal fluid samples from patients with early and late Lyme neuroborreliosis by polymerase chain reaction.

A polymerase chain reaction (PCR) was developed for use in the identification of a 248-bp fragment of the Borrelia burgdorferi flagellin gene in urine and cerebrospinal fluid (CSF) from patients with Lyme neuroborreliosis. The specificities of the PCR products were confirmed by DNA-DNA hybridization with an internal probe. The assay had a detection limit of 10 in vitro-cultivated B. burgdorferi. The PCR assay seemed to be species wide as well as species specific, since DNA from all 21 B. burgdorferi isolates from humans tested but not from Borrelia hermsii or Treponema pallidum could be amplified. We tested 10 consecutively diagnosed patients with untreated neuroborreliosis. There was lymphocytic pleocytosis and intrathecal B. burgdorferi-specific antibody synthesis in the CSF of all patients. Urine and CSF samples were investigated by PCR before, during, and up to 8.5 months after therapy. B. burgdorferi DNA was detected in urine samples from nine patients; five patients, including two patients with chronic neuroborreliosis, were PCR positive prior to treatment, whereas urine samples from the remaining four patients obtained 3 to 6 days after the onset of therapy became PCR positive. All urine samples obtained greater than 4 weeks after therapy were negative by PCR. PCR of CSF was less sensitive, and samples from only four patients, including one with chronic neuroborreliosis, were positive. We conclude that urine is a more suitable sample source than CSF for use in B. burgdorferi DNA detection by PCR. Normalization of inflammatory CSF changes and the negative PCR results during follow-up even in patients with chronic neuroborreliosis do not point to a persistent infection. The future role of PCR as a diagnostic tool for Lyme neuroborreliosis is still uncertain.     

Detection of Borrelia burgdorferi sensu lato in lesional skin of patients with erythema migrans and acrodermatitis chronica atrophicans by ospA-specific PCR.

The aim of this study was to develop a sensitive and specific PCR for the detection of Borrelia burgdorferi DNA. The plasmid-located gene coding for the outer surface protein A (OspA [31-kDa protein]) was used as a target. Nucleotide sequence information from different B. burgdorferi ospA genotypes was used to design primers homologous to different genotypes. The sensitivity of the nested PCR differed from 1 fg to 1 pg of borrelial DNA, depending on the strain analyzed. No cross-reactions with DNA from spirochetes other than B. burgdorferi or with human DNA were observed. A total of 22 skin biopsy samples from patients with erythema migrans (EM [n = 10]) or acrodermatitis chronica atrophicans (ACA [n = 12]) were examined for the presence of B. burgdorferi by nested PCR. Of 22 biopsies, 80% from EM patients and 92% from ACA patients were positive by PCR amplification. By comparison, 50% of the EM patients had elevated B. burgdorferi-specific immunoglobulin M (IgM) and/or IgG antibody levels as tested by enzyme-linked immunosorbent assay (ELISA) using purified B. burgdorferi flagella as antigen. A total of 33% of ACA patients had elevated IgM titers, and all had high IgG titers in their sera. Only 30% of specimens from patients with EM and none from patients with ACA were positive by culture. All culture-positive specimens were also positive by PCR. Thus, the sensitivities of the PCR were 80 and 92%, respectively, for patients with EM and ACA on the basis of the clinical and histopathological diagnoses of Lyme disease. From these results, we conclude that PCR is a suitable method to detect B. burdorferi sensu lato DNA in skin biopsy samples and could be applied as an additional diagnostic tool.     

Early dissemination of Borrelia burgdorferi without generalized symptoms in patients with erythema migrans

The diagnosis of erythema migrans (EM) is not always easy, and reports of culture- or PCR-confirmed diagnosis as well as reports of EM with simultaneous disseminated disease are few. Characteristics and incidence of EM in addition to frequency of early dissemination of B. burgdorferi were studied in the archipelago of South-Western Finland prospectively using questionnaires, skin biopsies and blood samples. Clinical EM was recognized in 82 patients (incidence 148/100,000 inhabitants/year). Of skin biopsy samples, 35.5% were positive by PCR (the majority B. garinii), and 21.5% by cultivation (all B. garinii). Of blood samples, 3.8% were positive by PCR, and 7.7% by cultivation. Of the patients, 30.9% were seropositive at the first visit, and 52.9% 3 weeks later. Of the patients with laboratory confirmed diagnosis, the EM lesion was ring-like in 31.8% and homogeneous in 65.9%. Dissemination of B. burgdorferi, based on culture or PCR positivity of blood samples, was detected in 11.0% of the patients. The frequency of generalized symptoms was nearly the same in patients with as in those without dissemination (22.2% vs 27.4%). Only 21.4% of the patients with culture-positive EM recalled a previous tick bite at the site of the EM lesion. We conclude that EM lesions are more often homogeneous than ring-like. B. burgdorferi may disseminate early without generalized symptoms.     

Detection of Borrelia burgdorferi-specific DNA in urine specimens from patients with erythema migrans before and after antibiotic therapy.

A nested PCR was developed for the detection of Borrelia burgdorferi-specific DNA in the urine of patients with erythema migrans. The target for the nested PCR was a specific region of the flagellin gene; the detection limit was less than five organisms of B. burgdorferi including all three species B. burgdorferi sensu stricto, B. afzelii, and B. garinii. A prospective, controlled, blinded study was performed with 26 patients with erythema migrans to evaluate the nested PCR method with clinical samples. B. burgdorferi-specific DNA could be detected in urine specimens from 22 of 24 patients with erythema migrans (sensitivity, 91.61%). Immediately after therapy, 11 of 19 patients still yielded positive results (58%). Eight weeks after therapy, 2 of 16 patients (13%) were positive by PCR of urine, and 20 weeks after treatment none of seven investigated urine samples was reactive. Essential for the sensitivity that was obtained was the development of a simple DNA extraction procedure. The results of the study indicate that the described method is highly sensitive and allows for the effective control of the efficacy of antibiotic therapy in patients with early Lyme borreliosis.     

Detection of Borrelia burgdorferi using the polymerase chain reaction.

Segments of the ospA gene of Borrelia burgdorferi were amplified by the polymerase chain reaction (PCR). Oligonucleotide primers used in the reaction flank a 309-base-pair segment within the ospA gene. After 35 cycles of amplification, the product could be detected by agarose gel electrophoresis or dot hybridization with a 32P-labeled probe. This segment was amplified in all strains of B. burgdorferi tested, but it was not detected in other bacterial species. An additional primer pair which has a broad specificity for conserved 16S rRNA sequences that are present in eubacteria amplified a 215-base-pair fragment in all organisms tested. The sensitivity of PCR for the detection of B. burgdorferi in clinical samples was evaluated by seeding blood and urine specimens with B. burgdorferi and subjecting them to amplification. We were able to detect 10 organisms per ml of blood or urine, using PCR with dot hybridization detection. DNA extraction is not required for sample preparation. Blood and urine specimens were obtained from canines with clinical and serologic evidence of Lyme disease and subjected to PCR analysis. Of 17 clinical specimens from 15 animals, one blood specimen showed reactivity in the PCR.     

Usefulness of blood and urine samples collected on filter paper in detecting cytomegalovirus by the polymerase chain reaction technique

A rapid test for the diagnosis of congenital CMV infection is still needed. This study evaluated the usefulness of dried blood and urine samples collected on filter paper for detecting cytomegalovirus (CMV) by the polymerase chain reaction (PCR) assay compared with the use of liquid urine. Samples were obtained from 332 infants aged 1-7 days. Liquid urine samples were collected into bags, cultured in human fibroblasts, and processed using a multiplex PCR technique. Dried urine samples were obtained by placing a piece of filter paper in contact with the infant's genitals. The heels of neonates were punctured and capillary blood was blotted onto filter paper and dried. Dried blood and urine specimens were analyzed by multiplex PCR and nested-PCR assays. A diagnosis of congenital CMV infection was established by isolating the virus, and by detecting viral DNA in the liquid urine. Of the 332 liquid urine samples collected from 332 neonates, seven (2.1%) were positive for CMV and 325 were negative, by both cell culture and PCR assay. In dried samples, CMV DNA was detectable only with a nested PCR assay. Compared with known CMV infection status, 5/7 (71.4%) neonates were positive for congenital CMV infection using dried blood samples. All 325 uninfected neonates were negative. In the dried urine samples, 4/4 CMV-infected infants gave positive tests, and all 262 uninfected infants were negative. Although further improvements in sample collection and/or processing are still needed, PCR testing on dried urine or blood collected on filter paper is a promising approach in the diagnosis of neonatal CMV infection.     

Diagnosis of early Lyme disease by polymerase chain reaction amplification and culture of skin biopsies from erythema migrans lesions.

Current laboratory diagnosis of Lyme disease relies on tests for the detection of antibodies to Borrelia burgdorferi, the etiologic agent of the disease. These tests are often unreliable because of a lack of sensitivity and specificity and test-to-test variability. The purpose of this study was to evaluate the sensitivity and specificity of polymerase chain reaction (PCR) amplification for detection of B. burgdorferi in skin biopsy specimens. Forty-six 2-mm skin biopsy samples were obtained from 44 patients with a clinical diagnosis of erythema migrans, 9 of whom were receiving antibiotic therapy at the time of biopsy. Specimens were ground in BSK medium with separate aliquots taken for culture and PCR. Of the specimens from the untreated group, 57% (21 of 37) were positive by culture and 22% (8 of 37) were culture negative; 22% (8 of 37) of the cultures were uninformative because of contamination. By comparison, 22 (59%) of 37 specimens were positive by PCR amplification. Of 21 culture-positive samples, 13 (62%) were also positive by PCR analysis. Thus, the sensitivity of the PCR was 59 to 62%, based on either a clinical or cultural diagnosis of untreated Lyme disease. None of the nine specimens from antibiotic-treated patients grew in culture, whereas two of the nine were positive by PCR analysis. Given the complexity and time required for culture, PCR is a promising technique for the diagnosis of early Lyme disease.     

Qualitative and semiquantitative polymerase chain reaction testing for cytomegalovirus DNA in serum allows prediction of CMV related disease in liver transplant recipients

AIM: To identify cytomegalovirus (CMV) infection in liver transplant recipients by polymerase chain reaction (PCR) techniques and to separate the cases in which CMV related disease will occur, for whom treatment is indicated, from those in whom infection will remain innocuous. METHODS: The combination of qualitative and semiquantitative PCR of serum and urine was assessed to determine whether these assays can identify those at risk of CMV related disease and compared their performance with conventional approaches to diagnosis. RESULTS: Qualitative PCR of serum had superior specificity, sensitivity, and positive and negative predictive values compared with urine DEAFF (detection of early antigen fluorescent foci) and PCR of urine. All episodes of CMV related disease were associated with the presence of CMV DNA by PCR in serum or urine; CMV was detected before clinical onset in 70% and 60% of cases, respectively. The period over which CMV DNA could be detected was not correlated with CMV related disease. Both peak viral load and cumulative viral load estimated using a semiquantitative PCR method on serum samples positive by the qualitative method could be used to distinguish asymptomatic infection from CMV related disease with 100% specificity and sensitivity. In contrast semiquantitative PCR of urine was of little value. CONCLUSIONS: An approach based on PCR testing with a combination of qualitative and subsequently semiquantitative serum samples would improve the diagnosis of CMV infection and aid identification of those patients at risk of CMV related disease, allowing treatment to be targeted specifically.     

Optimal method of collection of first-void urine for diagnosis of Chlamydia trachomatis infection in men

First-void urine (FVU) is the preferred specimen for the diagnosis of urogenital Chlamydia trachomatis infection in men. We have developed FirstBurst, a urine collection device that collects the first 4 to 5 ml of FVU and yields a specimen with a sixfold higher C. trachomatis organism load than the regular urine cup by quantitative PCR (32,533 versus 5,271 plasmids/ml; P < 0.0001). Consequently, the use of FirstBurst to collect a urine sample improved the sensitivity of a rapid test for Chlamydia over testing of samples collected with a urine cup (82 versus 47% sensitivity using PCR as a reference; P < 0.0015).     

Diagnosis of Chlamydia trachomatis urethritis in men by polymerase chain reaction assay of first-catch urine.

To determine the accuracy of a recently developed polymerase chain reaction (PCR) urine assay to detect Chlamydia trachomatis urethral infection in men, we obtained urethral swabs and first-catch urine from 365 men attending a sexually transmitted diseases clinic. Thirty-three (9%) of the 365 men were infected with C. trachomatis as defined by urethral culture. Thirty-two of the 33 men with culture-positive urethral swabs also had PCR-positive urine assays. Of 332 patients with culture-negative urethral swabs, 325 had PCR-negative urine. Compared with chlamydia culture of urethral specimens, PCR assay of urine samples thus had a sensitivity of 97% and a specificity of 98%. The positive predictive value of the urine PCR assay was 82%, and the negative predictive value was 99%. Analysis of discrepant results indicated that six of seven PCR-positive, urethral culture-negative patients probably had chlamydial urethritis. All six patients had symptoms of urethritis and had either a positive urethral swab PCR or a positive urine PCR with a different amplification target. After resolution of discrepant results, (defining true positives as the 33 culture-positive patients and the 6 PCR-positive, culture-negative patients just described), the sensitivity and specificity of culture were 85% (33 of 39) and 100% (326 of 326), respectively. The revised sensitivity and specificity of PCR were 97% (38 of 39) and 99.7% (325 of 326), respectively. We conclude that this urine PCR assay provides a highly sensitive, noninvasive alternative method for the detection of C. trachomatis urethral infection in high-risk men attending a sexually transmitted diseases clinic. This assay could greatly facilitate the testing of larger numbers of male patients for chlamydial infection and should be studied in other settings.     

Demonstration of Borrelia burgdorferi DNA in urine samples from healthy humans whose sera contain B. burgdorferi-specific antibodies.

Since the possibility of asymptomatic infection with Borrelia burgdorferi has been suggested by a positive serology found in healthy subjects, we hypothesized that these subjects might excrete borrelial DNA sequences in urine as happens in patients with Lyme borreliosis. We found borrelial sequences by nested PCR in the urine samples from 3 of 13 healthy B. burgdorferi antibody-positive adults but not in urine samples from 79 antibody-negative healthy controls. After therapy with doxycycline, the urine samples were repeatedly negative for B. burgdorferi DNA. We conclude that urinary excretion of borrelial DNA sequences may occur in seropositive healthy subjects during asymptomatic infection. Demonstration of such sequences in urine must be interpreted cautiously and may not necessarily prove a borrelial cause of disease.     

Comparison of polymerase chain reaction and culture for detection of Borrelia burgdorferi in naturally infected Peromyscus leucopus and experimentally infected C.B-17 scid/scid mice.

Culture and the polymerase chain reaction (PCR) were compared for detection of Borrelia burgdorferi infection in wild-caught Peromyscus leucopus and experimentally inoculated C.B-17 scid/scid (severe combined immunodeficient) mice. PCR targeted highly conserved regions of the ospA gene and could detect one to five cultured organisms and 10 to 50 copies of molecularly cloned ospA DNA. Organs (kidney, spleen, and urinary bladder) and/or ear biopsy samples were obtained from 108 captured P. leucopus mice, and tissues were obtained from 7 experimentally inoculated mice. A simple sample-processing procedure with proteinase K and detergent treatment was used in the PCR analysis. Overall, B. burgdorferi was detected in 29 of 108 (27%) P. leucopus mice by culture and in 31 of 108 (29%) mice by PCR. As assessed by the kappa statistic, agreement between PCR and culture was high for ear and bladder (kappa = 0.80 and 0.65, respectively) and low for kidney and spleen (kappa = 0.37 and 0.03, respectively). While concordant results were obtained from 98 animals, PCR detected B. burgdorferi from 6 additional mice for which cultures were negative and culture detected B. burgdorferi from 4 animals which were PCR negative. Further phenol-chloroform extraction of DNA in a limited number of samples improved the sensitivity of PCR compared with that of culture. These results indicate that PCR may be as sensitive as culture for detecting B. burgdorferi in ear samples and that PCR analysis is suitable for establishing the infection status of animals in mark-release-recapture studies.     

Specific detection of human BK polyomavirus in urine samples of immunocompromised patients

A semiquantitative PCR assay for the detection of BK virus in urine was developed using primers for BK virus that specifically amplified BK but not JC virus. DNA was extracted from urine through treatment with proteinase K followed by DNA precipitation with sodium acetate. Semiquantitation was achieved by amplifying serial dilutions (1:1, 1:10, 1:100, and 1:1,000) of the urine specimens. Each assay included both positive (stock BK virus and previously positive patient urine) and negative (no template) controls. A urine sample was interpreted as positive if any of the serial dilutions showed amplification of the DNA fragment of the expected size. For some patient-derived samples, amplification of the expected-size fragment was achieved with a dilute template whereas no amplification was achieved with a concentrated template. This was attributed to interfering substances in the urine. PCR results were compared with urine cytology and shown to be more sensitive. Validation studies were performed at the University of Nebraska Medical Center, utilizing a separate qualitative PCR assay that detects both BK and JC virus and distinguishes between them by restriction enzyme digestion patterns. Of 46 urine samples analyzed using both methods, 22 were positive by both assays, 18 were negative by both assays, 5 were positive only by the Nebraska method, and 1 was positive only by our method. In comparison with the Nebraska PCR, our PCR assay had a sensitivity of 81% and specificity of 95%. For twenty-one (43%) of 49 immunocompromised patients, tests were positive when specimens were submitted because of clinical suspicion of BK virus infection.     

Specific Detection of Human BK Polyomavirus in Urine Samples of Immunocompromised Patients

A semiquantitative PCR assay for the detection of BK virus in urine was developed using primers for BK virus that specifically amplified BK but not JC virus. DNA was extracted from urine through treatment with proteinase K followed by DNA precipitation with sodium acetate. Semiquantitation was achieved by amplifying serial dilutions (1:1, 1:10, 1:100, and 1:1,000) of the urine specimens. Each assay included both positive (stock BK virus and previously positive patient urine) and negative (no template) controls. A urine sample was interpreted as positive if any of the serial dilutions showed amplification of the DNA fragment of the expected size. For some patient-derived samples, amplification of the expected-size fragment was achieved with a dilute template whereas no amplification was achieved with a concentrated template. This was attributed to interfering substances in the urine. PCR results were compared with urine cytology and shown to be more sensitive. Validation studies were performed at the University of Nebraska Medical Center, utilizing a separate qualitative PCR assay that detects both BK and JC virus and distinguishes between them by restriction enzyme digestion patterns. Of 46 urine samples analyzed using both methods, 22 were positive by both assays, 18 were negative by both assays, 5 were positive only by the Nebraska method, and 1 was positive only by our method. In comparison with the Nebraska PCR, our PCR assay had a sensitivity of 81% and specificity of 95%. For twenty-one (43%) of 49 immunocompromised patients, tests were postive when specimens were submitted because of clinical suspicion of BK virus infection.     

Detection ofLegionella DNA in human and guinea pig urine samples by the polymerase chain reaction

A detection system forLegionella DNA in urine samples based on the polymerase chain reaction (PCR) was developed and tested on infected guinea pigs and patients suffering from pneumonia. Results were compared with standard methods for diagnosis of Legionnaires' disease. A primer system was selected which amplifies a 108 bp DNA fragment of the 5S rRNA gene. The sensitivity of the PCR system was one femtogram of extractedLegionella DNA. Three methods were tested for pretreatment of urine samples. Of these, the Geneclean II kit (Bio 101, USA) gave the best results for artificially contaminated urine samples as well as those from infected guinea pigs or patients. Thirty-seven urine samples from 15 guinea pigs intraperitoneally infected with eitherLegionella pneumophila serogroup 1, 3 and 6 orLegionella micdadei, 26 urine samples of 21 patients suffering from pneumonia, and 30 control samples of patients with urinary tract infection (UTI) were tested.Legionella DNA was detected in 29 of the guinea pig urine samples; whereas, urinary antigen detection using EIA was positive in only 20 of the samples. PCR was also positive in the samples of 11 patients with pneumonia, 9 of which were confirmed by other microbiological methods, such as culture, direct fluorescent antibody test, urinary antigen detection and antibody testing. However, of the 30 control samples from patients with UTI, three samples yielded positive results. The results demonstrate thatLegionella DNA is excreted in the urine of infected individuals and that the PCR shows a higher degree of sensitivity than EIA to the detection of solubleLegionella antigen in urine. Although the positive results obtained from control samples are a potential limitation of the test, these preliminary data suggest that PCR can be a sensitive method for the diagnosis of legionellosis from urine samples.     

Storage and preservation of whole blood samples for use in detection of human immunodeficiency virus type-1 by the polymerase chain reaction.

Methods used in the diagnosis of human immunodeficiency virus type-1 (HIV-1) infection by the polymerase chain reaction (PCR) usually require the separation of lymphocytes from a whole-blood sample within 24 hours of patient sampling. A method is described in which blood samples are mixed with a cryopreservative ('Glycigel'), stored frozen, and DNA suitable for use in an HIV PCR recovered. Samples can be stored at -20 degrees C for up to 3 months and still give positive results with all samples from infected patients; storage at -80 degrees C for at least 3 months shows no loss of titre. The method shows no loss of sensitivity compared to previously described sample preparation methods. Deglycerolised Glycigel supernatants were found to be suitable for conventional anti-HIV-1 serological studies and loss of sensitivity only represented the dilution effect due to sample preparation. Application of the method as a means of storing samples frozen at the point of sampling and transporting them to a central laboratory for processing is demonstrated using samples taken from HIV-1-infected mothers and their babies.     

Detection of Borrelia burgdorferi in human blood and urine using the polymerase chain reaction.

We investigated the use of the polymerase chain reaction (PCR) to detect Borrelia burgdorferi strain B-31 in human blood and urine experimentally inoculated with 5 and 1 borreliae/cm3, respectively, and to biotinylate a DNA probe specific for B. burgdorferi in the dot blot and Southern blot assays. When the blood and urine samples were subjected to PCR, a 370-bp amplified product was consistently visible on agarose gel electrophoresis after 30 and 45 cycles, respectively. The total human genomic DNA extracted from a 1-cm3 sample of inoculated blood was approximately 6.25 micrograms, and the total amount of B. burgdorferi DNA was estimated to be 0.01 pg/6.25 micrograms of the human DNA. For PCR, 2.5 micrograms of human DNA which contained the equivalent of 0.004 pg of borrelia DNA (approximately two borreliae) were used for enzymatic amplification. When 1/20 or 1/10 of the PCR-amplified products were used either for dot blot or Southern blot hybridization, the accessible copies of amplified B. burgdorferi DNA were sufficient for detectable hybridization to occur. PCR amplification of B. burgdorferi DNA in clinical specimens followed by dot blot hybridization may be a valuable adjunct or alternative to current but inadequate laboratory methods for the diagnosis of Lyme disease.     

Detection of leptospires in urine by PCR for early diagnosis of leptospirosis.

We tested urine samples from patients at different stages of current leptospirosis and thereafter to determine whether use of the PCR for detection of leptospires in urine can be a valuable alternative to culturing. The procedure of DNA extraction and subsequent PCR applied to 15 freshly voided urine samples proved to be twice as sensitive as culturing. Overall, we were able to detect leptospires in approximately 90% (26 of 29) of the urine samples. Urine and serum samples were obtained from seven patients, before the eighth day of illness. Although it is generally assumed that leptospiruria starts approximately in the second week of illness, we were able to detect leptospires in all of these early urine samples. In contrast, only two of seven corresponding serum samples gave positive PCR results, which suggests that PCR analysis of urine can be more successful for early diagnosis of leptospirosis than PCR analysis of serum. Urine samples from six patients who had been treated with antibiotics at the time of illness were positive by PCR, implying that the patients were still shedding leptospires in their urine despite treatment. Some of these samples were even taken years after the infection, indicating that shedding of leptospires in urine may last much longer than is generally assumed. We conclude that detection of leptospires in urine with PCR is a promising approach for early diagnosis of leptospirosis and may also be useful in studying long-term shedding.