Multiplex High-Definition Polymerase Chain Reaction Assay for the Diagnosis of Tick-borne Infections in Children

Abstract

Over the past 2 decades, Lyme disease incidence has increased to nearly half a million new cases each year in the United States [1–3]. Borrelia burgdorferi, the primary causative agent of Lyme disease, is the most recognized and reported cause of tick-borne illness. The Ixodes scapularis or black-legged tick that transmits B burgdorferi can also carry 1 or more other pathogens [4], including other Borrelia species [5] as well as additional bacteria, viruses, or parasites [5–9]. Rapid and accurate detection of tick-borne pathogens in addition to B burgdorferi could help guide clinical decision making and allow for timely and targeted treatment.

The approach to evaluation for tick-borne infections that complicate or mimic Lyme disease varies considerably. Currently available diagnostics frequently measure the host antibody response, which can be less specific than directly detecting the infecting pathogen [10], potentially falsely negative in early infection, and will not distinguish active from past infection [11]. While a blood polymerase chain reaction (PCR) test is insensitive for detection of B burgdorferi due to low bacterial counts during infection [12–14], other tick-borne pathogens are typically present in sufficient quantity to be detected by a PCR assay with sensitivity superior to microscopic examination of the blood smear or serology [15]. A multiplex PCR panel could improve recognition of other tick-borne infections and coinfections by simultaneously detecting multiple pathogens [13, 16]. However, the unknown prevalence of tick-borne coinfections in children creates clinical uncertainty about the optimal approach to diagnostic testing.

To this end, we assembled a multicenter prospective cohort of children undergoing evaluation for Lyme disease in endemic areas [17]. Our primary goal was to determine the prevalence of tick-borne infections other than B burgdorferi including coinfections in children serologically or clinically diagnosed with acute Lyme disease, and compare to matched controls using a multiplex high-definition PCR (HDPCR) tick-borne pathogen panel. Our secondary goal was to explore the potential impact of identification of non-Borrelia tick-borne coinfections on patient management.

METHODS

Pedi Lyme Net Cohort

We assembled a prospective cohort of children presenting for evaluation of potential Lyme disease between June 2015 and December 2021 to 1 of 8 Pedi Lyme Net emergency departments located in 3 geographic regions: Northeast (Boston Children's Hospital [Boston, Massachusetts] and Hasbro Children's Hospital [Providence, Rhode Island]); Mid-Atlantic (Nemours Children's Hospital [Wilmington, Delaware], Children's Hospital of Philadelphia [Philadelphia, Pennsylvania], and Children's Hospital of Pittsburgh [Pittsburgh, Pennsylvania]); and Midwest (Children's Wisconsin [Milwaukee, Wisconsin] and Minnesota Children's [Minneapolis and St Paul, Minnesota]).

Patient Consent Statement

When available, study staff approached potentially eligible children year-round to obtain written informed consent for participation with assent when appropriate. The institutional review board at each of the study sites approved the study protocol with permission for data sharing and material transfer.

Data Collection

At the time of enrollment, we obtained clinical history from medical providers as well as subjects and families, including demographics, symptoms, and duration. We defined peak Lyme season between June and October. After collection of clinical blood tests, we obtained research biosamples (serum and whole blood), which were initially processed at the participating study site, stored at −80°C, and batch shipped to the Pediatric Lyme Disease biobank at Boston Children's Hospital [17]. Approximately 1 month after enrollment, study staff reviewed the medical record and performed telephone follow-up to determine test results, treatment, and complications as well as clinical outcome. All data were collected prospectively using Research Electronic Data Capture (REDCap) [18] housed at Harvard University.

Lyme Disease Case Definition

We defined a case of Lyme disease with a clinician-diagnosed erythema migrans (EM) lesion or positive 2-tier Lyme disease serology with clinical symptoms associated with acute Lyme disease [19]. We defined positive 2-tiered testing as a positive or equivocal first-tier enzyme immunoassay (EIA) followed by a positive immunoblot interpreted using standard criteria [20]. For this study, we performed first-tier EIAs at a single research laboratory (2015–2019: Immunetics C6; 2020–2021: Diasorin VlsE; Branda Laboratory, Massachusetts General Hospital, Boston). For children with Lyme disease, we subclassified according to stage: early (a single EM lesion), early disseminated (multiple EM lesions, headache, and facial nerve palsy), or late (arthritis). Children without EM lesions who had negative 2-tier Lyme disease serology were classified as controls.

Nested Case-Control Study

For this study, we conducted a case-control study nested within the larger Pedi Lyme Net prospective cohort using patients with an available whole blood research sample. We selected cases from each participating center with early Lyme disease (a single EM lesion), Lyme meningitis, or Lyme facial palsy. For each Lyme disease case, we selected a control with similar symptoms (meningitis of facial palsy), matched for age group (1–7 years, 8–12 years, 13–18 years, or 19–21 years), gender, and participating center. Children with a single EM lesion were matched to a control with a recognized recent tick bite without signs or symptoms of infection.

Multiplex HDPCR Test

Research biosamples were frozen at −80°C after collection and shipped in batches from the enrolling sites to the Pediatric Lyme Disease biobank at Boston Children's Hospital (Boston, Massachusetts). We batch shipped whole blood samples from the selected case and the control patients to our laboratory collaborator (Buchan Laboratory, Children's Wisconsin, Milwaukee) who performed the research use–only multiplex HDPCR Tick-borne Panel (ChromaCode, Carlsbad, California) to test for 8 bacterial and 1 protozoan distinct tick-borne pathogens or pathogen groups (Table 1). A 200-µL portion of each whole blood specimen was subjected to total nucleic acid extraction using the EMAG assay (bioMérieux), eluted into 50 µL of buffer, and analyzed using the research use–only multiplex HDPCR Tick-borne Panel (Chroma Code). Real-time amplification and detection of each assay target was conducted using the ABI 7500 FastDx thermocycler (Applied BioSystems) with 96-well PCR plate format in accordance with manufacturer's instructions for use. Each plate contained both positive and negative controls, as well as 5 calibrators to ensure quality and accuracy of results.

For the children found to have a non-Borrelia tick-borne infection with our research HDPCR panel, we re-reviewed the existing medical record to understand clinical presentation, diagnostic test results (blood smear, hemoglobin, liver function tests, and chest radiograph), and treatment received.

Statistical Analysis

We describe categorical variables with proportions including 95% confidence intervals (CIs) and medians with interquartile range (IQR). We determined the frequency of tick-borne coinfections in children with Lyme disease and matched clinical mimics. We then calculated the difference in proportions with 95% CI.

We used IBM SPSS software (version 27.0; IBM Corporation) for all analyses.

RESULTS

Over the study period, we approached 3983 children (68% of eligible), of whom 3523 (88% of approached) agreed to participate in the Pedi Lyme Net cohort study. For this nested case-control study, we selected 306 early Lyme disease cases matched to 306 control subjects without Lyme disease. Of the 306 with Lyme disease, 55 (18.0%) had a single EM lesion and 251 (82.0%) had 1 or more symptoms consistent with neurologic Lyme disease. The median patient age was 10 years (IQR, 6–14 years) and 247 (40.4%) were female. A minority of both cases and controls had a recognized tick bite in the previous year (55/306 [18.0%] Lyme disease cases vs 38/306 [12.4%] symptomatic controls; difference, 5.6% [95% CI, −.2% to 1.3%]). We included children from the following geographic regions of the United States: Northeast (237 [38.7%]), Mid-Atlantic (326 [53.3%]), and Midwest (49 [8.0%]). At the time of patient enrollment and biosample collection, a minority had started antibiotics active against Lyme disease (55/612 [9.0%]): 15 amoxicillin, 15 cephalexin, 12 doxycycline, 6 azithromycin, 4 trimethoprim-sulfamethoxazole, and 3 clindamycin. The median duration of pretreatment was 4 days (IQR, 1–7 days).

Of the 612 selected samples, 594 (97.1%) had a successful multiplex HDPCR panel, 12 (2.0%) failed sample extraction, and 14 (2.3%) had an indeterminant result. Samples that failed extraction were unable to be repeated due to limited sample volume. Among the 14 specimens initially reported as initially indeterminate, 8 had a definitive result after repeat PCR testing and 6 (0.9%) remained indeterminate (Table 2).

Of the 297 HDPCR panels performed on samples from children with Lyme disease, 12 had Borrelia species identified, of which 10 were B burgdorferi/Borrelia mayonii species (incidence, 3.4% [95% CI, 1.8%–6.1%]) and 2 were relapsing fever borreliae (Borrelia hermsii, Borrelia parkeri, or Borrelia turicatae, undifferentiated). The single control subject with a positive B burgdorferi PCR was 11 years of age with 3 days of headache and negative 2-tier Lyme disease serology. None of the 9 children with a recent tick bite but no other symptoms of infection had any tick-borne pathogen detected by the HDPCR panel. All of the non-Borrelia tick-borne infections (n = 15) were identified in children with serologic or clinically confirmed Lyme disease (15/297 [5.1%] Lyme cases vs 0/297 [0%] controls; difference, 5.1% [95% CI, 2.2%–8.7%]).

The 15 cases with a non-Borrelia coinfection occurred in children with Lyme disease who presented with clinical features compatible with Lyme disease alone (Table 3). The 2 children with Anaplasma phagocytophilum and the 1 child with Erlichia chaffeensis infection presented with headache and fever and received 14 days of doxycycline.

All 12 children with Babesia microti identified were previously healthy without a immunodeficiency including previous splenectomy. Two had an EM lesion (single or multiple) and 11 had neurologic Lyme disease (eg, headache or facial palsy). None of the 12 children with B microti identified had anemia (58% tested), hepatitis (33% tested), or pneumonia (0% had a chest radiograph performed). Although none of these children received antibiotics active against B microti, each made a complete recovery based on medical record review and telephone follow-up. None of the children with a non-Borrelia coinfection were tested for these infections by their treating clinicians.

Children with non-Borrelia coinfections had a similar seasonality (14/15 [93.3%] coinfection during peak Lyme season vs 228/270 [84.4%] Lyme disease alone; difference, 8.9% [95% CI, −14.6% to 16.2%]) and rates of hospitalization (5/15 [33.3%] coinfection vs 72/270 [26.7%] Lyme disease alone; difference, 6.7% [95% CI, −12.3% to 32.1%]). The median duration of antibiotics prescribed was similar (14 days [IQR, 14–21 days] coinfection vs 14 days [IQR, 14–21 days] Lyme disease alone; difference, 0 days [IQR, −2 to 3 days]).

DISCUSSION

A significant minority of children with serologically or clinically diagnosed Lyme disease had non-Borrelia tick-borne infections identified by the multiplex HDPCR panel, whereas none of the matched controls had these infections identified. Reliance on traditional diagnostic methods may lead to underdiagnosis of tick-borne coinfections. However, currently recommended first-line antibiotic treatments active for Lyme disease will also treat Anaplasma and Erlichia infections [21], even when these tick-borne coinfections are not recognized. Additionally, Babesia infections are typically subclinical in immunocompetent children.

Ixodes scapularis ticks are endemic in the Northeast, Mid-Atlantic, and Midwest, with their geographic range expanding over the past few decades [2], and are capable of simultaneously carrying and transmitting multiple human pathogens [9]. However, rigorous studies to determine how frequently I scapularis ticks transmit non-Lyme infections or coinfections to humans are limited [6, 22, 23], especially for children. In 230 adults from Maine with a recognized tick bite within the previous 5 years, 6 (2.6%) had a positive serologic test to B miyamoti and 1 (0.4%) to Powassan virus, consistent with recovery of previous infection or exposure. In 2 small prospective studies of adults with a biopsy proven EM lesion, 2% had a positive B microti PCR result (active infection) and 10% had positive B microti serology (active or previous infection) at the time of enrollment [22, 24]. A prior study evaluated the same HDPCR panel we utilized in this study using residual whole blood samples from 530 adults undergoing evaluation for Lyme disease presenting to 2 centers located in the Upper Midwest [13]. Interestingly, the 6% coinfection rate in these adults was similar to that observed in our multicenter pediatric cohort.

Babesiosis, caused by infection with the parasite B microti, shares a similar geographic distribution to Lyme disease. The pathogen can be identified either by examination of peripheral blood whole blood smear or PCR assay [21]. The preferred antibiotic treatment for nonsevere babesiosis includes atovaquone plus azithromycin, while first-line Lyme disease antibiotics (eg, amoxicillin, doxycycline, or ceftriaxone) are not active against Babesia. While babesiosis can be a serious infection in older adults or those with a compromised immune system, most immunocompetent patients, especially children, have subclinical infections that do not require treatment [21]. In a retrospective cohort of 336 adults with Lyme disease, serologic evidence of B microti infection was not associated with more severe symptoms than Lyme disease alone [25]. Our study patients with B microti identified did not demonstrate clinical evidence of babesiosis and all recovered without targeted antibiotic therapy.

Clinical practice for the diagnosis of non-Lyme tick-borne infections varies considerably. Serologic tests rely on detection of the human host's immune response to infection and may take weeks to develop. Therefore, early in infection, serologic tests may be falsely negative, and later may be falsely positive due to prior infection or cross-reactivity [26]. Direct pathogen detection tests such as PCR identify the infecting pathogen rather than relying on host response to infection. Consistent with previous evaluations [14, 27], the B burgdorferi blood PCR assay had inadequate sensitivity to be utilized clinically for evaluation of patients with potential Lyme disease. Similarly, Rickettsiae species infections are frequently limited to the vascular endothelium with few organisms circulating free in the bloodstream, which limits the sensitivity utility of PCR for these pathogens. However, blood PCR is both a sensitive and specific diagnostic for other non-Borrelia tick-borne pathogens and is recommended for the diagnosis of acute infections [21, 28]. Previous evaluations of the multiplex HDPCR panel utilized in our study demonstrated that this panel performed similarly to individual PCR assays with a 97.4%–100% positive agreement and a 98.9%–100% negative agreement, compared with validated single PCR assays [13, 16]. The ability to evaluate for multiple pathogens with a single assay provides a widely adaptable and easy-to-implement testing strategy for the majority of hospital laboratories without the need for additional instrumentation or specialized training. However, this panel has some important limitations, including the inability to distinguish between Borrelia species associated with Lyme disease (B burgdorferi and B mayonii), and does not detect tick-borne viral pathogens (ie, Powassan virus). In addition, specimens with very low concentration of B microti may be misclassified as positive for relapsing fever–causing Borrelia species (Borrelia group 2) [16]. This characteristic of the HDPCR test utilized could explain the 2 children who tested positive for relapsing fever “Borrelia group 2” targets who did not have documented travel to regions endemic for ticks that transmit these bacterial species. Further sequence analysis of the resulting amplicon would be necessary to definitively resolve these 2 unexpected results.

Treating clinicians evaluating a patient for Lyme disease must decide whether also to evaluate for other tick-borne infections. Two of the tick-borne coinfections identified, Erlichia and Anaplasma, are effectively treated with doxycycline, a first-line therapy for Lyme disease [21]. Babesia requires targeted antibiotic therapy but typically causes subclinical infection in previously healthy children. Importantly, none of the symptomatic control patients who were undergoing clinical evaluation for Lyme disease had any other tick-borne infection identified. Routine use of molecular panels in the initial assessment of children with a potential tick-borne infection would add costs without clear benefit given the relatively low positivity rate. Available clinical algorithms to limit utilization of molecular diagnostics (eg, tick-borne panels) to the most appropriate patients [29] need further evaluation before widespread implementation, especially for children. Our findings suggest that multiplex HDPCR or other molecular tick-borne panels should be reserved for children with Lyme disease with an atypical clinical presentation or immunodeficiency (eg, splenectomy) or who have inadequate response to initial therapy.

Our study has several limitations. First, we selected cases and controls from the larger Pedi Lyme Net cohort. Although we matched by symptoms, demographics, and participating center, we were unable to match by time of presentation. Therefore, controls may have differed from cases in ways that could have influenced the risk of tick-borne coinfection. Second, we were limited to the pathogens included in the multiplex panel and did not have an independent diagnostic gold standard for the non–Lyme disease tick-borne infections. Therefore, we may have failed to identify some pathogens including viruses with our molecular panel. Third, antibiotics could render PCR assays falsely negative. However, only a minority of children received antibiotics active against tick-borne pathogens prior to research sample collection. Fourth, all study enrollment sites were located in urban areas. However, many families bypass local care sites to seek care for their children at regional pediatric centers [30]. Fifth, we only performed short-term clinical follow-up (1 month after enrollment) and may not have identified subtle symptoms in children with an identified coinfection. Last, although coinfection risk may vary between regions in the United States, the small number of coinfections identified limits our ability to identify geographic predictors of infection.

CONCLUSIONS

A substantial minority of children with acute Lyme disease have an unrecognized tick-borne infection, although current therapies directed toward the primary agent of Lyme disease may be adequate. Clinical use of multiplex molecular tick-borne panels should be reserved for atypical Lyme disease cases or when initial response to therapy is inadequate. Larger studies are needed to identify clinical and geographic risks of tick-borne coinfection to further guide clinical decision making and ensure appropriate utilization of tick-borne illness testing.

Notes

Acknowledgments. We gratefully acknowledge the participation of our beloved colleague, Aris C. Garro, MD, MPH (deceased), as a co-investigator in this study.

Disclaimer. The funder had no role in the research conduct or results interpretation.

Financial support. This work was supported by a grant from the Emergency Medicine Foundation (to L. E. N.) and the Global Lyme Alliance (to L. E. N.).

References

Author notes