Expanding Access to Biospecimens for Lyme Disease Test Development

TEXT

Lyme disease (LD), caused by Borrelia burgdorferi sensu stricto, is the most frequently reported tick-borne illness in the United States. Recent analyses estimate over 300,000 cases annually in the United States (1). Most cases occur in the Northeast and Upper Midwest; however, the range of B. burgdorferi-infected ticks may be expanding, resulting in transmission in areas where they were previously considered nonendemic (2). LD occurs in stages, including the early localized (erythema migrans [EM]), early disseminated, and late disseminated stages. EM is a pathognomonic sign that is, however, not universally present and that may be atypical (3, 4). As such, laboratory-based diagnostics are often used to confirm clinical suspicion.

Direct detection of B. burgdorferi, including culture and nucleic acid amplification tests (NAT), are possible but are not practical or sufficiently effective (reviewed in reference 5). Culture is technically challenging due to the slow replicative rate of the B. burgdorferi and thus are neither useful nor practical. NAT can be used for the detection of B. burgdorferi DNA in cerebrospinal fluid (CSF), synovial fluid (SF), and EM lesions. However, the sensitivity of CSF and SF testing by NAT is poor. NAT of EM lesions is more sensitive than NAT of CSF and SF but is usually unnecessary due to the characteristic nature of the EM lesion. Given these limitations, serologic tests are the mainstay of LD diagnostics.

People infected with B. burgdorferi mount robust antibody responses. Patients with disseminated stages typically have levels of antibody readily detectable with current antibody tests and algorithms (6). However, antibody testing in patients with early, localized disease is challenging due to the relatively slow developing antibody response (7). The majority of cases presenting with EM are clinically diagnosed and treated without laboratory confirmation of infection. In addition to sensitivity, specificity concerns with LD antibody tests are well-known. There are cross-reactions in LD assays due to other bacterial infections. In addition, viral infections, such as infections with Epstein-Barr virus, have been associated with false reactivity. A number of approaches have been adopted and are being investigated to improve upon the suboptimal sensitivity and specificity of antibody detection in LD.

The current standard approach for the serologic diagnosis of early disseminated and late disseminated infection with B. burgdorferi is the standard two-tier testing algorithm (STTTA), adopted in 1994 (8). The algorithm relies on the use of a relatively sensitive enzyme immunoassay (EIA) as a screening test. A negative result is reported as such. If the result is equivocal or positive, the sample is then reflexively tested by Western immunoblotting (WB). WB enhances the specificity of testing by requiring IgM and/or IgG reactivity with multiple spirochetal proteins. Unfortunately, the enhanced specificity achieved with this reflexive testing approach is at the cost of sensitivity. The WB assay has lower sensitivity than the screening tests (9). This is particularly problematic for the detection of antibodies during early localized infection, when antibody levels are low. Specificity has been improved with the adoption of the two-tiered testing algorithm (8) as well as with tests using conserved proteins from B. burgdorferi (10, 11).

Among the improvements to testing for LD is the recent adoption of a modified two-tiered testing algorithm (MTTTA) (12). This algorithm replaces WB with a second EIA that uses spirochetal antigens different from those used in the screening EIA. This change addresses, in part, the sensitivity concerns with the WB. Studies have documented this improvement (13–15) and have led the U.S. Centers for Disease Control and Prevention (CDC) to endorse this algorithm as an alternative to the standard two-tiered testing algorithm. A second approach to increase the sensitivity of antibody detection is the use of novel technologies. Arumugam et al. (16) described the use of a multiplexed assay on a microfluidic platform as a more sensitive alternative. In their study, a sensitivity of 80 to 85% was achieved in early LD patients. In comparison, the two-tiered testing algorithm achieved sensitivities of 48.5 to 75%. As expected, the sensitivity was 100% in Lyme arthritis patients, as was the case for the two-tiered testing algorithm.

Given the effectiveness of early treatment (17) and the diagnostic challenges, a significant research effort is under way to develop more sensitive methods for diagnosing early LD. Advances in understanding the biology of B. burgdorferi and the host response to infection and technological advances are paving the way for improvements in laboratory diagnostics (5, 9). As additional novel methods for the detection of B. burgdorferi antibody or other direct detection methods are developed and evaluated, there is a major need for reliable sources of well-documented, quality-assured biospecimens. In particular, biospecimens from early LD patients are particularly desirable.

To this point, Horn et al. (18) describe the development of the Lyme Disease Biobank (LDB). The purpose of LDB is to be a resource for biospecimens from early LD patients and healthy controls from areas of endemicity for use in test development and evaluation. In their publication, the authors describe the identification and assessment of LDB samples. Case patients included subjects from an area of endemicity with EM (lesion size, >5 cm) or annular skin lesions (lesion size, ≤5 cm) or subjects with one or more signs/symptoms of early LD (headache, fatigue, fever, chills, joint or muscular pain) but lacking skin lesions. Cases were enrolled within 30 days of onset and could have been treated with antibiotics for less than 48 h. Controls included healthy subjects from areas of endemicity with no history of LD or other tick-borne infection. LD test results were not used in the selection process for case patients or controls.

Biospecimens were collected from consenting case patients and controls (greater than 10 years of age) identified in 3 areas of endemicity: Martha’s Vineyard, MA; East Hampton, NY, and north central Wisconsin (5 clinic sites). In this initial report, 550 subjects (298 case patients and 252 controls) were enrolled. Whole blood and urine were collected from the case patients and controls and shipped to a central biorepository (Precision for Medicine, Frederick, MD). At the repository, samples were processed and stored as serum, plasma, whole-blood, and whole-urine aliquots. The biorepository is College of American Pathologists (CAP) accredited and Clinical Laboratory Improvement Amendments (CLIA) certified. In addition to biospecimens, a case report form was completed to gather demographic information, information on the signs and symptoms of LD, treatment history, other medical history, relevant epidemiological information, tick exposure, and a photograph of any skin lesions.

All blood samples were tested in a blinded manner by real-time PCR (RT-PCR) for B. burgdorferi, Anaplasma phagocytophilum, Babesia microti, and Borrelia miyamotoi. Samples obtained from Wisconsin were also tested by RT-PCR for Borrelia mayonii and Ehrlichia muris subsp. eauclairensis. Some samples were cultured for B. burgdorferi. A subset of these had the culture fluid tested by RT-PCR for B. burgdorferi. All samples were tested by both a screening EIA and WB. Laboratory testing was performed at multiple sites and included 3 different screening EIAs, depending on the laboratory and time period. Both in-house-developed and commercial WB assays were used, with interpretations being performed according to CDC criteria (8). Many samples were also tested by 2 EIAs, allowing classification according to the MTTTA. The testing laboratories were CAP accredited and CLIA certified.

Based on signs/symptoms and serologic test results, case patients were classified as having laboratory-confirmed LD (positive results by use of the STTTA, RT-PCR, or culture or positive results by two EIAs and an EM lesion size of >5 cm; 35% of cases), probable LD (an EM lesion size of >5 cm and a negative result by use of the STTTA), suspected LD (am EM/annular lesion size of <5 cm and a negative result by use of the STTT), or symptomatic with no lesions. A negative STTTA result could have a positive individual test (EIA or WB) result in the algorithm.

Laboratory testing with direct methods confirmed the poor sensitivity of these approaches: two EM cases were culture positive, nine case patients and one control were positive by RT-PCR-augmented culture, and four cases were RT-PCR positive. B. microti and A. phagocytophilum were also detected in 9% and <1% of samples, respectively.

The STTTA was positive for 29% of cases with EM lesion sizes of >5 cm and for 10 to 12% of cases with EM lesion sizes of <5 cm and cases without lesions. Two percent of controls from regions of endemicity had a positive STTTA result. Just over half of the subjects provided a convalescent-phase blood sample, which was collected from 2 to 3 months after the acute-phase sample was obtained. Interestingly, only 4% of convalescent-phase samples from EM-positive patients demonstrated seroconversion (seronegative in the acute-phase sample and seropositive in the convalescent-phase sample) by STTTA. In total, 35% of patients in the group classified as type 1 (those with EM) had laboratory-confirmed LD, 44% had probable LD, and 21% had suspected LD. In the type 2 category (patients who did not have EM but who were symptomatic), 14% had laboratory-confirmed LD and 86% were seronegative. Eighty-three percent of controls from areas of endemicity were negative by all serologic tests.

This LDB is a source for biospecimens from well-characterized donors that is available to test developers and evaluators. Potential users can access application forms online (info{at}lymebiobank.org). This biorepository is focused on early LD samples. This includes samples from patients with potentially nonclassical or absent EM. While this presents some degree of uncertainty in their use as standards to judge the performance of novel assays, they do represent the patient group for which novel tests are most needed. This study did not enroll subjects with early or late disseminated disease, as these are not priority areas for test development since robust IgG antibody responses will be detected in most patients. However, the bank has more recently enrolled samples from subjects with disseminated disease.

Of concern is the relatively low seropositivity rates of EM-positive cases. This may present a challenge for test development, as the classification of cases with atypical lesions or cases with symptoms of early LD but no lesions could be questioned. While a significant number of donors provided convalescent-phase samples, the STTTA seroconversion rate was <5%, which is concerning (7, 19) and which is lower than that reported for the CDC Lyme Serum Repository (LSR) (20). A close look at the data shows a seroconversion rate of only 7% for IgG blots but a rate of 33% for IgM blots. Applying the STTTA criteria of not using the IgM WB result after 30 days postonset reduces the seroconversion rate significantly. The lack of IgG seroconversion might be due to the use of antibiotics and the 2- to 3-month time frame for collection of the convalescent-phase samples. However, 82% of cases were antibiotic naive. For now, the low seropositivity rate remains unexplained.

Overall, the Lyme Disease Biobank provides another useful biospecimen resource for individuals involved in LD test development and evaluation. It is unique in the variety of biospecimens available (whole blood, serum, urine). It can also be of use for other tick-borne diseases as well, given the laboratory testing conducted. An often-cited resource for serum specimens from Lyme disease patients is the CDC LSR (20), which contains serum samples from EM-positive patients (acute- and convalescent-phase samples) as well as patients with early disseminated infection (carditis, neuroborreliosis) and late disseminated infection (Lyme arthritis). In contrast to the LDB, the LSR samples from patients with early Lyme disease excluded samples from subjects that had received antibiotic therapy (patients who provided convalescent-phase samples had not received antibiotic therapy). In addition, more samples in LSR than in LDB were culture and/or PCR positive. Forty percent of acute-phase samples were STTTA positive, while 61% of convalescent-phase samples were positive, as defined by strict STTTA application (i.e., disregarding the IgM WB result for samples collected >30 days after symptom onset). These results for samples from patients with early disease are noticeably different from those for samples from LDB.

There is a growing appreciation for the value of carefully created, curated, and managed biorepositories to support basic, translational, and clinical research (21). Preanalytic variables can significantly affect testing results; as such, standardization of practices is highly desirable. In fact, there are organizations that accredit biorepositories. Accreditation, such as that provided by the College of American Pathologists, offers a mechanism to ensure best practices in the acquisition, storage, and distribution of biospecimens for research (22). It is worthwhile noting the parent organization of the LDB is accredited by CAP and has CLIA certification. This should provide additional confidence in the quality of the associated biospecimens provided.