Diagnostic Performance of C6 Enzyme Immunoassay for Lyme Arthritis

We conducted a prospective cohort study of children presenting to 1 of 6 EDs in Pedi Lyme Net. Participating EDs were each located in Lyme disease endemic areas: Alfred I. Dupont Hospital for Children (Wilmington, DE), Boston Children’s Hospital (Boston, MA), Hasbro Children’s Hospital (Providence, RI), Children’s Hospital of Philadelphia (Philadelphia, PA), Children’s Hospital of Pittsburgh (Pittsburgh, PA), and Children’s Hospital of Wisconsin (Milwaukee, WI).⁹  We obtained informed consent and assent from eligible children and their caregivers, as appropriate. The study protocol was approved by the institutional review boards of each participating center, with permission for data and biosample sharing.

For the parent study, we enrolled children ≥1 and ≤21 years of age presenting to a participating ED undergoing clinical evaluation for Lyme disease, defined by the treating clinician ordering Lyme disease serology between June 2015 and February 2019. If the treating clinician later cancelled clinical Lyme disease serology, the patient was excluded after enrollment. For this planned substudy, we selected enrolled children with either a painful or swollen joint on examination. A patient could be enrolled more than once if the child had more than 1 eligible encounter.

Treating clinicians completed data forms that were used to capture the following historical and examination factors: antibiotic treatment within 72 hours, tick bite history, fever, duration of symptoms, and affected joint involvement. Previous Lyme disease history was obtained from the patient or their caregivers. One month after enrollment, study staff reviewed medical records to determine laboratory results as well as treatment received. We defined a normal CRP as ≤0.5 mg/dL and a normal ESR as ≤20 mm/hour, using established laboratory cut points.¹⁰  Phone follow-up was performed as needed to determine clinical outcome 1 month after enrollment. All data were entered in REDCap data collection tools hosted by Harvard University.¹¹ 

Serological diagnosis of Lyme disease requires a positive first-tier test (eg, C6 EIA) followed by a supplemental immunoblot. Because the specific commercial Lyme disease test in clinical use differed between participating centers, which can impact 2-tier results in some cases,^12–14  we performed C6 2-tier Lyme disease testing for all enrolled children. Study participation included collection of a research serum sample after obtaining clinical samples. Initial sample processing consisted of centrifuging at 3600 rpm for 10 to 15 minutes with collection of the supernatant. The resulting research serum samples were then stored at the local institution at −80°C before batch shipment to the Pediatric Lyme Disease Biobank located at Boston Children’s Hospital (Boston, MA). We then performed a C6 EIA test using a single commercial assay¹⁵  at the Branda Laboratory at Massachusetts General Hospital (Boston, MA). We interpreted C6 EIA index values using cut points recommended by the assay manufacturer to classify C6 EIA index values as negative (optical density index values <0.90), equivocal (optical density index values 0.90–1.09), or positive (optical density index values ≥1.10).¹⁵  Children with a positive or equivocal C6 EIA result had immunoglobulin G (IgG) and immunoglobulin M (IgM) Borrelia burgdorferi immunoblots interpreted using standard criteria.¹⁶  If immunoblots were not performed as a part of clinical care, we obtained research immunoblots at a single commercial laboratory (ARUP, Salt Lake, UT) using standard interpretative criteria. For patients with more than 30 days of symptoms, a positive IgM alone was considered a false-positive immunoblot result.^16–18  A child with a positive Lyme disease 2-tier serology result had a positive or equivocal C6 EIA result, followed by a positive supplemental immunoblot result.

We defined a case of septic arthritis as a child with arthritis and growth of a pathogenic bacteria from synovial fluid culture or with synovial fluid pleocytosis (synovial fluid white blood cell [WBC] count ≥50 000 cells/μL) and growth of pathogenic bacteria from blood culture. A priori, we defined bacillus species, Staphylococcus nonaureus, and Streptococcus viridans as contaminants. For children without septic arthritis, we defined a case of Lyme arthritis as a child with arthritis and a positive 2-tiered Lyme disease serology result. All other children were classified as having inflammatory arthritis.

We first compared the clinical characteristics of patients by arthritis type: septic, Lyme, or inflammatory arthritis. We compared medians using a Kruskal-Wallace test and proportions using a χ² test. Next, we calculated the performance of a positive or equivocal C6 EIA result for the diagnosis of Lyme arthritis by reporting sensitivity, specificity, negative predictive value, and positive predictive value. We then examined the proportion of children with a positive or equivocal C6 EIA result who had a diagnostic arthrocentesis, an operative joint wash performed, or a final diagnosis of septic arthritis. Last, we calculated the test accuracy in the subgroup of patients who had diagnostic arthrocentesis performed.

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

Of the 3002 eligible children approached, we enrolled 2062 children (68.7% enrollment rate). Of these, 20 children (0.1% of enrolled) were later excluded because the clinical team later cancelled Lyme disease testing. Of the 2064 included children, we collected a research serum sample from 1898 patients (92.0% of enrolled; Fig 1). For this study, we evaluated the 911 who presented for evaluation of acute arthritis (48.0% of those with research sample available). The number of children enrolled per participating ED ranged from 21 to 265. The median age was 7 years (interquartile range [IQR]: 4–11 years), and 572 (62.9% of study patients) were male patients.

Of the 911 study patients, 789 (86.6%) had pain of a single joint, 93 (10.2%) had pain of 2 joints, and 29 (3.2%) had pain of 3 or more joints. In order of descending frequency, the following joints were involved: knee (n = 539; 59.2%), hip (n = 339; 37.2%), ankle (n = 103; 11.3%), elbow (n = 51; 5.6%), wrist (n = 35; 3.8%), and other joints (n = 44; 4.8%; 22 shoulder, 11 spine, 6 finger, and 5 toe). Of the entire study population, 587 (64.4%) had joint swelling on physical examination. Overall, 11 (1.2% of study patients) had septic arthritis, 211 (23.2%) had Lyme arthritis, and 689 (75.6%) had inflammatory arthritis.

Of the 11 children who had septic arthritis, all were diagnosed during the index hospital encounter and had a single joint involved. All but 1 had an elevated CRP (9 out of 10 tested; 90%), and most had an elevated ESR (7 out of 10 tested; 70%) (Table 1). Staphylococcus aureus was the most common bacterial pathogen. Less than half of the children with septic arthritis had a positive synovial fluid Gram-stain result (5 out of 11 tested; 45%).

Of the 211 children with Lyme arthritis, 87 (41.3%) had a positive IgG result alone, 10 (4.7%) had a positive IgM result alone, and 114 (54.0%) had both a positive IgG and IgM result. The proportion of children with Lyme disease varied from 9.4% to 33.8% by participating EDs. The vast majority of children with Lyme arthritis had a single joint involved (209 out of 211; 99.1%). Overall, 209 (99.1% of children with Lyme arthritis) had a positive C6 EIA result and 2 (0.9%) had an equivocal C6 EIA result.

Next, we compared the clinical characteristics of children by arthritis type (Table 2). More children with Lyme arthritis were boys. One in 5 children had been pretreated with antibiotics within 72 hours of study enrollment. Few children recalled a history of a tick bite, including those children with Lyme arthritis. Lyme arthritis most commonly affected the knee. Children with Lyme and septic arthritis both had elevated WBC counts, ESR, and CRP amounts, although there is overlap in the range for children with all 3 types of arthritis. A substantial minority of children with Lyme and inflammatory arthritis had arthrocentesis and/or operative joint washout performed.

A positive or equivocal C6 EIA result had high specificity and positive predictive value for Lyme arthritis (Table 3). None of the 250 children with a positive or equivocal C6 EIA result had septic arthritis (0%; 95% confidence interval [CI]: 0%–1.5%). However, of those with a positive first-tier Lyme disease test result, 75 children underwent arthrocentesis (30.0%; 95% CI: 24.7%–36.0%) and 27 had an operative joint washout performed (10.8%; 95% CI: 7.5%–15.3%).

Of the 142 children who had a diagnostic arthrocentesis performed, 11 children (7.7% of those who had arthrocentesis) had septic arthritis, 73 (51.0%) had Lyme arthritis, and 59 (41.3%) had inflammatory arthritis. Of these, 75 children had either a positive or equivocal C6 EIA test result. Of the 11 children with septic arthritis, 5 had a positive synovial fluid Gram-stain result.

We assembled a large prospective cohort of children with acute arthritis undergoing evaluation for Lyme disease at 1 of 6 participating EDs located in Lyme disease endemic areas. Whereas septic arthritis was uncommon, almost one-third of children with acute arthritis had Lyme arthritis. Of the children with positive or equivocal C6 EIA test results, no child was diagnosed with septic arthritis, although a substantial minority underwent arthrocentesis. In the appropriate clinical scenario, clinicians could use the C6 EIA to identify children at low risk for septic arthritis who could potentially safely avoid invasive procedures and parenteral antibiotics while awaiting confirmatory test results.

In Lyme disease endemic areas, distinguishing between Lyme and septic arthritis can be particularly challenging because Lyme and septic arthritis have considerable clinical overlap.^1,3  A child with septic arthritis requires prompt initiation of parenteral antibiotics and, if the hip joint is affected, operative joint washout to avoid permanent joint damage. The Kocher criteria are used to identify children with hip arthritis who are at low risk of septic arthritis and have undergone broad validation.^19,20  Low-risk children have none of the following high-risk predictors: inability to bear weight on the affected limb, fever (temperature ≥38.5°C), peripheral WBC count ≥12 000 cells/mm³, and ESR ≥40 mm/hour. First, the Kocher criteria were not developed for children with arthritis of other joints, and only 37% of our study patients had hip arthritis. Also, in a recent 3-center study of children with hip arthritis who underwent arthrocentesis conducted in a Lyme disease endemic area, 12% of those with septic arthritis were classified as low risk by the Kocher criteria.⁶  Clearly, the Kocher criteria should not be used in isolation to guide the initial management of children with acute inflammatory arthritis in Lyme disease endemic areas.

With a test turnaround time of only a few hours, the currently available C6 EIA could be used along with other clinical factors to risk stratify children with acute arthritis. Children with a positive or equivocal C6 EIA result could be started on empirical Lyme disease antibiotics while safely avoiding invasive procedures such as arthrocentesis or operative joint washout because the risk of septic arthritis was low. However, given the C6 EIA specificity of only 94%,⁸  confirmatory immunoblots are still needed to confirm the diagnosis of Lyme arthritis before making final treatment decisions. On the other hand, children with a negative C6 EIA result need careful clinical assessment, including consideration of arthrocentesis to obtain a synovial fluid culture when septic arthritis is clinically suspected.

Currently used 2-tier Lyme disease testing combines a sensitive first-tier test with a more-specific second-tier test. The higher specificity of newer first-tier tests (eg, C6 EIA) when compared with older tests (eg, whole-cell sonicate EIA) suggests a greater use to guide initial decision-making.^8,21  The current commercially available C6 EIA assay is a moderate-complexity test that must be performed by trained laboratory personnel, although lower-complexity and more-rapid assays are under development (J.A.B., personal communication, 2019). However, the results of C6 EIA tests performed at the hospital laboratory could be available to clinicians within a few hours, before decisions about the need for invasive procedures are made, often while the child is still in the ED. Although we dichotomized the C6 EIA results as directed by the test manufacturer, higher C6 EIA values have been associated with increased risk of Lyme disease and could further assist clinical decision-making.²²  Point-of-care Lyme tests are under development and will further help to guide rapid decision-making for children with arthritis in Lyme disease endemic areas.

With our study, we underscore the clinical importance of rapid and accurate Lyme disease testing. Two EIA testing strategies, referred to as modified 2-tier testing strategies,^7,23  may replace conventional 2-tier testing, shortening the turnaround time of current Lyme disease tests.^24,25  Newer multiplex serological panels may be more accurate than the conventional 2-tiered strategies.²⁶  Beyond serology, novel host-based diagnostic approaches such as metabolomics²⁷  or RNAexpression²⁸  profiles have been piloted and may provide an alternate approach. Older direct pathogen detection methods, such as a polymerase chain reaction test of synovial fluid, lack sensitivity compared with serological testing.^29,30  Newer direct detection methods appear promising,³¹  although more-robust evaluation in well-curated patient cohorts will be required before widespread implementation.

Our study has the following limitations. First, each of the participating centers is located in Lyme disease endemic areas. Our findings should not be applied to regions where Lyme disease is seen less frequently or to children without potential Lyme disease exposure. Second, we did not enroll all eligible children because of availability of study staff and refusal of patients and families to participate in the study. However, we did not detect substantial differences between enrolled and not-enrolled children (data not shown). Third, we may have misclassified children with septic arthritis because many children did not have a diagnostic arthrocentesis, and we included those who had received antibiotics before enrollment, which could render culture results falsely negative. However, after clinical and medical record follow-up 1 month after enrollment, we did not identify any children with a delayed diagnosis of septic arthritis. In addition, we know that most children treated in the ED of a pediatric referral center will return to the same center for subsequent care.³²  Fourth, Lyme disease serology results may be falsely positive due to past rather than current infection, highlighting a limitation of the current approach to Lyme disease diagnosis. However, the background seroprevalence rate in children without signs of infection presenting to the Pedi Lyme Net centers was low.³³  Fifth, we relied on the physical examination of ED providers, and some joint effusions, especially when small, may not have been recognized clinically. Last, septic arthritis was uncommon, limiting the certainty of our risk estimates.

The C6 EIA could assist initial clinical decision-making for providers caring for children presenting to the ED for evaluation of arthritis, although additional validation is required. In the appropriate clinical scenario, children with a positive or equivocal C6 EIA result could safely avoid arthrocentesis and operative joint washout while starting empirical Lyme disease antibiotics. Further work is needed to demonstrate the impact of C6 EIA test availability at the time of initial clinical decision-making.

We thank the providers, research staff, and children and families who agreed to participate in the study.

CI

confidence interval

CRP

C-reactive protein

ED

emergency department

EIA

enzyme immunoassay

ESR

erythrocyte sedimentation rate

IgG

immunoglobulin G

IgM

immunoglobulin M

WBC

white blood cell