The Lyme Disease Polymerase Chain Reaction Test Has Low Sensitivity

Introduction

Lyme disease, the most common vector-borne disease in the United States and Europe, is caused by infection with Borrelia bacterial species. Available Lyme disease diagnostics can be broadly divided into direct and indirect detection methods (Aguero-Rosenfeld et al. 2005, Schutzer et al. 2019). Direct methods, such as a nucleic acid amplification tests (i.e., Lyme PCR) or culture, detect the presence of Borrelia bacteria whereas indirect methods measure the host's response to infection. Although the clinical utility of Lyme PCR has previously been questioned, we evaluated the test performance in a multicenter cohort of children undergoing evaluation for acute Lyme disease.

Methods

We performed a prospective cohort study of children presenting to one of six emergency departments (EDs) located in Lyme disease endemic areas that participate in Pedi Lyme Net (Nigrovic et al. 2017). The study protocol was approved by the institutional review board of each participating institution with permission for data sharing.

We approached children between 1 and 21 years who presented to the ED for evaluation of possible Lyme disease between June 2015 and July 2019 and obtained written informed consent for study participation. We limited our analysis to those children undergoing evaluation for acute Lyme disease who had both a Lyme PCR test (from cerebrospinal fluid [CSF], synovial fluid, or blood) and Lyme disease serology obtained. Specific Lyme PCR assays varied between centers (Supplementary Table S1). Treating clinicians provided clinical histories and we abstracted the test results from the medical records.

We defined a Lyme disease case with either a physician-diagnosed erythema migrans (EM) lesion or positive two-tiered Lyme disease serology (Maulden et al. 2019). Children with a positive or equivocal first-tier enzyme-linked immunoassay (EIA) test had a supplemental IgG and IgM immunoblot performed, interpreted using standard criteria (Centers for Disease Control and Prevention [CDC] 1995). For patients with >30 days of symptoms, a positive IgM alone was considered a false positive (Centers for Disease Control and Prevention [CDC] 1995, Seriburi et al. 2012, Lantos et al. 2016).

We used a chi square test to compare proportions and a Mann–Whitney test to compare medians. Our primary goal was to calculate the test performance of Lyme PCR for the diagnosis of Lyme disease. We report the sensitivity and specificity for all sample types combined and then by sample type (synovial fluid, CSF vs. blood) separately and Lyme disease stage (early disseminated: multiple EM lesions, facial palsy, meningitis, or carditis vs. late: arthritis).

We used SPSS version 23.0.0 for all statistical analyses (IBM SPSS Software; Armonk, NY).

Results

During the study period, we enrolled 2447 children of whom 125 (5.1%) had a Lyme PCR obtained by the treating clinicians. Lyme disease was more common in children who had a Lyme PCR test performed than those who did not (55/125 [44%] vs. 443/2322 [19.1%]; p < 0.001).

Of the 125 study patients who had a Lyme PCR test performed, the median patient age was 10 years (interquartile range [IQR] 6–14 years) and 80 (64.0%) were male. The median duration of symptoms at the time of initial ED evaluation was 3 days (IQR 2–5 days). Overall, 55 children had Lyme disease (44.0% of study cohort) of whom 23 had a positive Lyme PCR test (41.8% of those with Lyme disease). All children with a positive Lyme PCR test also had a positive two-tiered serology (positive EIA and supplementary immunoblot): positive IgM immunoblot alone with ≤30 days of symptoms (n = 1), positive IgG immunoblot alone (n = 7), and both positive IgM and IgG immunoblot (n = 15).

Of the 203 children who had a lumbar puncture performed, 39 (19.2%) had a CSF Lyme test performed, and of the 189 who had an arthrocentesis performed, 72 (38.1%) had a synovial fluid Lyme PCR test performed. The median CSF WBC count (5 cells/mm³ IQR 1–102 cells/mm³ CSF Lyme PCR obtained vs. 2 cells/mm³ IQR 1–58 cells/mm³ no Lyme PCR obtained; p = 0.37) and the synovial fluid WBC count (41,935 cells/mm³ IQR 12,236–67,500 cells/mm³ synovial Lyme PCR obtained vs. 42,187 cells/mm³ IQR 10,930–66,985 cells/mm³ no Lyme PCR obtained; p = 0.88) were similar. All 14 children who had a Lyme blood PCR obtained had symptoms compatible with early disseminated Lyme disease.

The sensitivity of Lyme PCR was highest in synovial fluid in those children with suspected late Lyme disease (Table 1). None of the Lyme PCR tests obtained from CSF or blood was positive. No Lyme PCR tests was falsely positive (specificity 100%; 95% confidence interval: 94.2–100).

Discussion

In a multicenter prospective cohort of children undergoing evaluation for acute Lyme disease, Lyme PCR testing had low sensitivity for synovial fluid, CSF, and blood. The sensitivity of the synovial fluid Lyme PCR was the highest, although still <50%. Importantly, every patient with a positive Lyme PCR also had a positive two-tiered Lyme disease test result. As both Lyme two-tier serology and PCR tests take several days to return results and PCR did not improve sensitivity, Lyme PCR is not helpful in most clinical situations (i.e., low diagnostic utility).

Our findings were similar to prior investigations. Previous estimates of Lyme PCR sensitivity ranged from 5% to 50%, with higher sensitivity for EM skin biopsies than for CSF samples (Christen et al. 1995, Lebech et al. 2000, Avery et al. 2005, Picha et al. 2005, Barstad et al. 2018). Although previous studies were small, included mainly adults, and were conducted in patients infected in Europe, all available Lyme PCR assays target sequences conserved between species. However, Lyme disease case definitions varied slightly between studies across patients. Our study was the largest conducted in U.S. children presenting to the ED, to include multiple centers, and to examine Lyme PCR testing of multiple specimen types.

Direct Borrelia bacterium detection could assist clinical decision-making, especially for children with suspected early Lyme disease. The low sensitivity of the Lyme PCR test reflects the paucity of bacterial DNA in the tested sample. Although increasing sample volume or other DNA amplification methods might improve sensitivity, these approaches will still have limited clinical utility, especially in children wherein sample volume may be limited (Bil-Lula et al. 2015). However, advances in molecular diagnostics for infectious diseases by direct detection methods such as antigen capture or enhanced nucleic amplification techniques appear promising, (Schutzer et al. 2019) but need more rigorous evaluation in children before clinical application.

Our study has several limitations. First, the treating clinicians decided whether to obtain a Lyme PCR test and we did not determine the reason for testing. Although we found that children who had the test obtained were more likely to have Lyme disease, the test sensitivity and specificity, unlike predictive values, measure the performance of the diagnostic test independent of disease prevalence. Second, the small number of tested patients limited our ability to rigorously evaluate Lyme PCR across specimen types and disease stage. Importantly, in our ED cohort, we could not evaluate the role for Lyme PCR of the synovial fluid in the management of antibiotic-refractory Lyme arthritis. Third, we aggregated results from Lyme PCR tests performed at four different commercial as well as one hospital laboratory (Supplementary Table S1). Although we were unable to determine whether the same primer sets or thermocycling parameters were used for each of these PCR tests, the Lyme PCR test performed similarly across testing laboratories. Finally, Lyme disease serologic assays in clinical use varied between institutions (Maulden et al. 2019). However, we applied a uniform Lyme disease case definition and all children with Lyme disease had positive two-tiered Lyme disease serology.

Conclusions

In conclusion, Lyme PCR did not improve the acute diagnosis or management of children undergoing evaluation for Lyme disease. Given the additional health care cost of this low-utility test, clinicians should not order Lyme PCR testing in the acute care setting.