Performance of two modified two-tier algorithms for the serologic diagnosis of Lyme disease

ABSTRACT

We compared the performance of a new modified two-tier testing (MTTT) platform, the Diasorin Liaison chemiluminescent immunoassay (CLIA), to the Zeus enzyme-linked immunoassay (ELISA) MTTT and to Zeus ELISA/Viramed immunoblot standard two-tier testing (STTT) algorithm. Of 537 samples included in this study, 91 (16.9%) were positive or equivocal by one or more screening tests. Among these 91 samples, only 57 samples were concordant positive by first-tier screening tests, and only 19 of 57 were concordant by the three second-tier methods. For IgM results, positive percent agreement (PPA) was 68.1% for Diasorin versus 89.4% for Zeus compared to immunoblot. By contrast, the PPA for IgG for both Diasorin and Zeus was 100%. Using a 2-out-of-3 consensus reference standard, the PPAs for IgM were 75.6%, 97.8%, and 95.6% for Diasorin, Zeus, and immunoblot, respectively. The difference between Zeus MTTT and Diasorin MTTT for IgM detection was significant (P = 0.0094). PPA for both Diasorin and Zeus MTTT IgG assays was 100% but only 65.9% for immunoblot STTT (P = 0.0005). In total, second-tier positive IgM and/or IgG results were reported for 57 samples by Diasorin MTTT, 63 by Zeus MTTT, and 54 by Viramed STTT. While Diasorin CLIA MTTT had a much more rapid, automated, and efficient workflow, Diasorin MTTT was less sensitive for the detection of IgM than Zeus MTTT and STTT including in 5 early Lyme cases that were IgM negative but IgG positive.

IMPORTANCE

The laboratory diagnosis of Lyme disease relies upon the detection of antibodies to Borrelia species. Standard two tier testing (STTT) methods rely upon immunoblots which have clinical and technical limitations. Modified two-tier testing (MTTT) methods have recently become available and are being widely adopted. There are limited independent data available assessing the performance of MTTT and STTT methods.

INTRODUCTION

Lyme disease, first recognized in Lyme, Connecticut, is the most common tick-borne disease in North America, Europe, and Asia and a significant and increasing public health concern (1). The causative agents of Lyme disease are spirochetes of the Borrelia burgdorferi sensu lato complex. The three species most commonly associated with human disease are B. burgdorferi (sensu stricto), Borrelia afzelii, and Borrelia garinii. Cases in North America are largely due to B burgdorferi sensu stricto, whereas B. afzelii and B. garinii cause most cases in Europe, and B. garinii predominates in Asia (2). In the United States, most cases are reported from the Northeast, mid-Atlantic, and upper Midwest regions (3).

A Lyme disease diagnosis can be made based on the presence of the characteristic erythema migrans (EM) rash in endemic areas. Unfortunately, in 20%–30% of patients without EM, the bacteria cannot be reliably detected by culture or PCR during acute infection and the diagnosis of Lyme disease relies on serologic testing (1). However, antibody responses can take time to develop, leading to potentially negative serologic testing results early in the disease. The diagnostic accuracy of Lyme disease testing is further confounded by the propensity for nonspecific positive results in IgM assays, especially in acutely ill patients.

Serologic testing for Lyme disease is traditionally performed using a standard two-tier testing algorithm (STTT) (4). In STTT algorithms, an initial first-tier screening immunoassay is performed, and, if positive or equivocal, immunoblots detecting B. burgdorferi-specific IgM and IgG antibodies are performed in the second tier (5). The inclusion of IgM- and IgG- immunoblots adds specificity to the algorithm but immunoblots are laborious and lack sensitivity especially in early localized Lyme disease (6). When interpreted by visual inspection, immunoblots are subject to variation in reader interpretation, especially at low antibody titers. Even when scanners are used to standardize immunoblot interpretation, there is variation between laboratories and instruments in the cutoff intensity set to identify a positive band (7). Furthermore, immunoblots are typically performed once or twice a week or are sent to a reference laboratory, prolonging the time to result.

The challenges associated with traditional Lyme disease diagnosis, especially gaps in the identification of early localized Lyme disease, prompted the development of modified two-tier testing algorithms (MTTT) comprised of successive immunoassays. Several studies comparing STTT with MTTT have demonstrated higher sensitivity of MTTT algorithms in early localized Lyme disease without compromising specificity (8–11). As a result, the first MTTT algorithm was granted FDA approval in July 2019 (12). MTTT was also found to be more cost-effective (13).

The first FDA-approved enzyme-linked immunoassay (ELISA) for MTTT Lyme testing uses a VlsE1/pepC10 IgM/IgG first-tier screening ELISA and second-tier ELISA involving B. burgdorferi whole-cell antigen for detection and potential differentiation of IgM and IgG antibodies (Zeus Scientific, Zeus, Branchburg, NJ) (11). The first- and second-tier ELISA tests can be automated and are typically performed in batches.

More recently, a chemiluminescent immunoassay (CLIA) (LIAISON, Diasorin, Saluggia, Italy) has become available for use as an MTTT option. CLIAs have several advantages over ELISAs due to inherent differences in their underlying chemistry (14). In contrast to batched ELISA testing, the Diasorin LIAISON is a random-access instrument allowing individual samples to be analyzed when received in the laboratory, with second-tier Lyme testing proceeding automatically on all appropriate samples. This results in a more efficient workflow, less technical time, and a shorter time to results. In this study, we compared the sensitivity and specificity of both the Zeus MTTT and the Diasorin MTTT systems with our current STTT algorithm. Samples submitted over a 1-month period to our clinical laboratory were used for this study and discrepant results were further evaluated by chart review.

MATERIALS AND METHODS

Case finding

Samples submitted to Yale New Haven Hospital for Lyme disease testing in the summer of 2021 were frozen at –20°C after the completion of all testing. The laboratory information system was queried for all samples tested for Lyme disease in the 1-month period between 27 July 2021 and 26 August 2021. This time period was selected due to higher test positivity patterns. Patient demographic information, specimen-specific information, and test results including ELISA sample to cutoff (S/CO) or index values and immunoblot results were extracted and/or recorded. Samples were thawed and subject to MTTT assays in May and June 2022. Among the 771 samples tested during this time, 234 were excluded from the analysis based on the following criteria (Fig. 1): no specimen found (n = 204), research opt out (n = 3), no second-tier result (n = 4), or repeat specimen within 3 days (n = 8). The absence of a specimen was most likely due to specimens being exhausted through Lyme or other testing. There was no significant difference in STTT algorithm first-tier test positivity among available or not available specimens (P = 0.5750; Data Not Shown). An additional 15 specimens were excluded due to Zeus screening results becoming negative after freeze-thaw. For these 15 specimens, the median (interquartile range) for the initial Zeus OD ratio was 1.020 (0.94–1.08). This study was reviewed by the Yale Institutional Review Board and determined to be Exempt and was granted a waiver of HIPAA authorization.

Fig 1.

Specimen testing flow chart with exclusion criteria, screening assay results, and secondary assay results. For IgM and IgG, results for Diasorin MTTT, Zeus MTTT, and Viramed Immunoblots were considered.

Serologic testing

The serologic testing methods are summarized in Table 1.

TABLE 1.

Summary of first- and second-tier test methods used in this study for each algorithm under comparison

	STTT	Diasorin MTTT	Zeus MTTT
Tier-one test	Zeus ELISA Borrelia VIsE1/pepC10 IgG/IgM	LIAISON Lyme Total Antibody Plus	Zeus ELISA Borrelia VIsE1/pepC10 IgG/IgM
Tier-one test target(s)	VIsE1 and pepC10 (B. burgdorferi)	VlsE (B. burgdorferi and B. garinii) and OspC (B. afzelii)	VIsE1 and pepC10 (B. burgdorferi)
Tier-one test methodology	Indirect ELISA	Indirect Chemiluminescence Immunoassay (CLIA)	Indirect ELISA
Tier-one test automation	Dynex DSX	Liaison XL Analyzer	Dynex DSX
Tier-one test workflow	Batched	Random Access	Batched
Tier-two test(s)	Viramed IgM and IgG Blots	LIAISON Lyme IgM and LIAISON Lyme IgG assays	B. burgdorferi IgM and B. burgdorferi IgG Test Systems
Tier-two test target(s)	Purified and recombinant Borrelia sp. antigens	VlsE (B. burgdorferi) and OspC (B. afzelii)	B. burgdorferi whole cell antigen
Tier-two test methodology	Immunoblot	CLIA	Indirect ELISA
Tier-two test automation	Viramed BeeBlot	Liaison XL Analyzer	Dynex DSX
Tier-two test workflow	Batched	Random Access	Batched

Initial STTT

Initial STTT testing was performed using the Zeus ELISA Borrelia VIsE1/pepC10 IgG/IgM Test System (Zeus Scientific, Branchburg, NJ) with automation on a Dynex DSX (Dynex Technologies, Chantilly, VA); all aspects of the testing protocol were performed according to the manufacturer’s instructions. Samples testing positive or equivocal with OD ratios ≥0.9 were reflexively tested using Borrelia B31 IgG and IgM ViraStripe test strips (Viramed Biotech AG, ViraStripe, Planegg, Germany). This testing was performed on the BeeBlot instrument (Bee Robotics, Gwynedd, Wales, UK), and the thresholds for positivity were set at 60 for both IgM and IgG. IgM immunoblots were considered positive if ≥2 out of the three key IgM bands were present, and IgG immunoblots were considered positive if ≥5 out of the 10 key IgG bands were present consistent with CDC criteria. All blot results were reviewed by resident physicians and an attending laboratory medicine physician before being released into the chart. Seven samples were positive by the Diasorin screening assay but negative by the original Zeus assay and not subjected to immunoblot testing as part of the original evaluation. These samples were subsequently sent to a reference laboratory for Borrelia B31 ViraChip IgM and IgG (Viramed Biotech AG) and interpretation was performed by the testing laboratory using the same band criteria described above (15). STTT testing was not re-performed after freeze- thaw.

Zeus ELISA MTTT

Available specimens were thawed and retested with the Zeus ELISA Borrelia VIsE1/pepC10 IgG/IgM Test System as described above according to the manufacturer’s instructions. Samples testing equivocal or positive with OD ratios ≥0.9 were tested by the Zeus B. burgdorferi IgM Test System and the Zeus B. burgdorferi IgG Test System on the Dynex DSX instrument according to the manufacturer’s instructions.

Diasorin CLIA MTTT

Available thawed specimens were screened with the Diasorin LIAISON Lyme Total Antibody Plus assay (LIAISON, Diasorin, Saluggia, Italy) according to the manufacturer’s instructions on a Diasorin LIAISON XL instrument. Samples testing equivocal or positive with Index values ≥0.9 were then tested with the LIAISON Lyme IgM and LIAISON Lyme IgG assays according to the manufacturer’s instructions.

Chart review

After all testing was completed and inclusion and exclusion criteria were applied, all samples with a positive or equivocal screen result underwent chart review. All charts were reviewed by three reviewers who summarized pertinent clinical and laboratory findings and classified cases as probable, possible, or unlikely active Lyme disease. Signs and symptoms were consolidated into major groups that were categorized as being consistent with early localized Lyme disease, early disseminated Lyme disease, or late Lyme disease (Table S1). Since the study was conducted in a highly endemic region of Connecticut during peak Lyme season, additional cases were classified as possible Lyme if systemic symptoms without fever, atypical rash with fever, or fever alone occurred in a patient with risk factors and no other explanation for their symptoms. In addition, patients who were diagnosed with acute babesiosis by blood smear and had a positive Lyme serology test without EM rash were classified as possible Lyme since in the year samples were collected (2021), 39.4% of babesia-infected Ixodes scapularis ticks also harbored B. burgdorferi and dual infection could not be excluded (https://portal.ct.gov/-/media/CAES/DOCUMENTS/Tick_Testing/Summary-of-Tick-Testing-Results-2021.pdf). Lyme antibody test results and treatment decisions were not incorporated into the initial categorization.

Data analysis

Positive percent agreement (PPA), negative percent agreement (NPA), and 95% confidence intervals were calculated in GraphPad Prism v9.4.1 (GraphPad Software, Boston MA) using either historic STTT results or a composite 2-out-of-3 reference standard as indicated. For composite 2-out-of-3 reference results, IgM or IgG was considered positive if any two methods, for example, Diasorin MTTT, Zeus MTTT, or immunoblot, were positive for that antibody isotype. A reference result was considered negative if only one or no isotype-specific method was positive. McNemar’s test with continuous correction and Kappa statistics were performed using the GraphPad online calculator.

RESULTS

Comparison of STTT and MTTT methods

Of 537 available samples included in this study, 91 were positive or equivocal by one or more tier-one screening tests. Of the 91, only 57 were concordant on initial tier-one screening, and of these 57, only 19 were concordant on tier-two testing. Of the 34 samples discordant on the initial screen, 21 were negative for all tier-two testing and likely represented either very early infections or false-positive screening tests (Fig. 1). Two samples were Zeus-negative on initial STTT tier-one testing retested as very low Zeus positive after freeze/thaw on Zeus MTTT testing. Both samples were negative on tier-two testing, but one case was deemed early Lyme by subsequent chart review.

We first compared the IgM and IgG results of the Diasorin and Zeus MTTTs to the historic STTT results as a reference (Table 2). Diasorin and Zeus MTTT were significantly different for the detection of IgM. PPA for Diasorin IgM was 68.1% compared to 89.4% for Zeus IgM, as Diasorin detected 32 and Zeus detected 42 of 47 immunoblot IgM positives (P = 0.0094). However, NPA and overall agreement for both MTTT algorithms and STTT were not significantly different. By contrast, PPA for IgG was 100% for both Diasorin and Zeus as both MTTT algorithms detected all 27 IgG immunoblot-positive samples. However, NPA and overall agreement were significantly different between the two MTTT IgG assays (P = 0.0036), with Diasorin reporting 29 and Zeus 16 additional IgG positives not detected by immunoblot.

TABLE 2.

Diasorin and Zeus IgM and IgG MTTT performance compared to STTT as a reference method^c

MTTT method	STTT reference method	Agreement with STTT reference			Kappa (range)
		PPA (95% CI) no. pos / no. ref pos	NPA (95% CI) no. neg / no. ref neg	Overall agreement (95% CI) no. agree /total
Diasorin CLIA IgM	Viramed Immunoblot IgM	68.1%a (53.8%–79.6%) 32/47	99.0% (97.6%–99.6%) 485/490	96.3% (94.3%–97.6%) 517/537	0.74 (0.63–0.85)
Zeus ELISA IgM	Viramed Immunoblot IgM	89.4%a (77.4%–95.4%) 42/47	98.0% (96.3%–98.9%) 480/490	97.2% (95.4%–98.3%) 522/537	0.83 (0.75–0.92)
P values		P = 0.0094a	P = 0.2278a	P = 0.4a
Diasorin CLIA IgG	Viramed Immunoblot IgG	100% (87.5%–100%) 27/27	94.3%b (92.0%–96.0%) 481/510	94.6%b (92.4%–96.2%) 508/537	0.63 (0.50–0.75)
Zeus ELISA IgG	Viramed Immunoblot IgG	100% (87.5%–100%) 27/27	96.9%b (95%–98.1%) 494/510	97%b (95.2%–98.2%) 521/537	0.76 (0.64–0.87)
P values		P = 1.0b	P = 0.0036b	P = 0.0036b

^a P values refer to differences between Diasorin and Zeus MTTT, compared to STTT reference method IgM.

^b P values refer to differences between Diasorin and Zeus MTTT, compared to STTT reference method IgG.

^c For comparison between IgM and IgG results, a negative screening test that did not reflex to isotype-specific testing was considered negative for both IgM and IgG.

MTTT algorithms perform differently than STTT algorithms at different stages of the disease, and we analyzed the data using a “2-out-of-3” consensus reference standard as described above. This also allowed us to analyze the performance of the IgM and IgG immunoblots in the STTT algorithm (Table 3). When a 2-out-of-3 consensus was used as the reference standard, Diasorin MTTT detected only 34 of 45 consensus IgM positives, whereas Zeus MTTT detected 44 of 45 IgM (P = 0.0094). For IgG, Diasorin and Zeus MTTT both detected 41 of 41 consensus IgG-positive specimens, versus 16 detected by immunoblot. Thus, both MTTT algorithms detected significantly more IgG results than immunoblot (P = 0.0005). Zeus MTTT had a higher NPA and overall agreement with the 2-out-of-3 consensus reference than did Diasorin MTTT (P = 0.0036) due to the greater number of IgG positives reported by Diasorin.

TABLE 3.

Diasorin and Zeus IgM and IgG MTTT and STTT performance using a 2-out-of-3 consensus reference

Tier-two IgM test method	Agreement with 2-out-of-3 consensus reference			Kappa (range)
	PPA (95% CI) no. pos / no. ref pos	NPA (95% CI) no. neg / no. ref neg	Overall agreement (95% CI) no. agree /total
Diasorin IgM	75.6% (61.3%–85.8%) 34/45	99.4% (98.2%–99.8%) 489/492	97.4% (95.7%–98.4%) 523/537	0.82 (0.72–0.91)
Zeus IgM	97.8% (88.4%–99.6%) 44/45	98.4% (96.8%–99.2%) 484/492	98.3% (96.8%–99.1%) 528/537	0.90 (0.83–0.96)
Immunolot IgM	95.6% (85.2%–98.8%) 43/45	99.2% (97.9% to 99.7%) 488/492	98.9% (97.6%–99.5%) 531/537	0.93 (0.87–0.99)
Diasorin v. Zeus Diasorin v. Blot Zeus v. Blot	P = 0.0094a P = 0.0265a P = 1.0	P = 0.23 P = 1.0 P = 0.39	P = 0.40 P = 0.12 P = 0.61
Tier-two IgG test method
Diasorin IgG	100% (91.4%–100%) 41/41	97.0% (95.1%–98.2%) 481/496	97.2% (95.4%–98.3%) 522/537	0.83 (0.75–0.91)
Zeus IgG	100% (91.4%–100%) 41/41	99.6% (98.5%–99.9%) 494/496	99.6% (98.7%–99.9%) 535/537	0.97 (0.94–1.00)
Immunoblot IgG	65.9% (50.5%–78.4%) 27/41	100% (99.2%–100%) 496/496	97.4% (95.7%–98.4%) 523/537	0.78 (0.67–0.89)
Diasorin v. Zeus Diasorin v. Blot Zeus v. Blot	P = 1.0 P = 0.0005a P = 0.0005a	P = 0.0036a P = 0.0003a P = 0.48	P = 0.0036a P = 1.0 P = 0.006a

^a Differences between methods were significant. McNemar’s test.

Chart review

With the high number of discordant results on both tier-one and tier-two testing, we attempted to ascertain the clinical accuracy of the three algorithms by examining the medical records. All 91 screen-positive or equivocal samples were subjected to chart review to assess for signs and symptoms of active Lyme disease, evidence of other acute infections, and evidence of past Lyme disease as summarized in Table S1. A concise summary of MTTT and STTT results and Lyme disease clinical assessment based on chart review is shown in Table 4. Nine cases of “Possible” Lyme were included as cases of active Lyme. None had recognized EM rash. Six of nine had acute babesia infections as well as positive Lyme serology, and a dual infection with Lyme could not be excluded. One of these babesia cases had concordant positive Lyme testing by all three algorithms, as did one other “Possible” Lyme sample with systemic symptoms compatible with early Lyme. The remaining two “Possible” Lyme cases with positive serology had either fever or headache, fatigue, and acute arthralgias without other explanation during peak Lyme season in Connecticut, a highly endemic area (see Table S1). Additional details on clinical presentations and stage of disease are presented in Table S2.

TABLE 4.

Comparison of MTTT and STTT test results and accuracy in detection of active lyme^d

Test	Result	No. with test result (no. with Lyme disease)a
		Diasorin MTTT	Zeus MTTT	Zeus ELISA/Viramed immunoblot STTT
Tier 1 Screen	Negative	27 (10)	7 (1)	9 (2)
	Positive	64 (45)	84 (54)	82 (53)
Tier 2 IgM and IgG	Negative	7 (2)	21 (5)	28 (10)
	IgM positive only	1 (1)	20 (16)	27 (22)
	IgM and IgG positive	36 (31)	32 (31)	20 (19)
	IgG positive only	20 (11)b	11 (2)	7 (4)
No. active Lyme cases confirmed by tier 2		43	49	45
No. unlikely active Lyme detected by tier 2		14 [6]c	14 [6]c	11 [4]c

^a () indicates the number with probable or possible Lyme disease by chart review. See supplemental tables for detailed information.

^b Five of 11 samples positive for IgG only by Diasorin were from patients with early or early disseminated Lyme.

^c [ ] indicates the number with past history of Lyme recorded in the medical record.

^d Charts were reviewed for all positive screening tests as described in Materials and Methods. Nine “possible” Lyme cases without EM rash included six babesia cases.

For the tier-one screen, Diasorin was negative for a greater number of active Lyme cases (n = 10) than Zeus MTTT (n = 1) or STTT (n = 2). Diasorin MTTT had a lower number of positive tier-one screening test results (n = 64) than Zeus MTTT (n = 84) and STTT (n = 82), and, among these samples, Diasorin had a lower number of tier-one-positive samples with negative results on tier-two testing (n = 7) than Zeus MTTT and STTT (n = 21 and 28, respectively). Of note, these samples testing negative on tier two included cases deemed to be active Lyme by chart review, namely 2, 5, and 10 cases for Diasorin, Zeus, and immunoblot, respectively.

As shown in Tables 2 to 4, the number of IgM detections by Diasorin MTTT (n = 37) was significantly lower than for Zeus MTTT (n = 52) and immunoblot (n = 47). Diasorin was notable for only one sample positive for IgM only, which was deemed a case of early Lyme. The IgM-only-positive samples detected by Zeus (n = 20) and immunoblot (n = 27) were also mostly cases of early localized or early disseminated Lyme. Both Diasorin and Zeus MTTT algorithms had significantly greater numbers of samples positive for both IgM and IgG (n = 36 and 32, respectively) compared to immunoblot (n = 20). As expected, these were predominantly early localized and early disseminated Lyme cases. In addition, four of five cases of late Lyme detected by Zeus MTTT were both IgM and IgG positive.

Diasorin MTTT had the highest number of total IgG detections (n = 56) compared to Zeus MTTT (n = 43) and immunoblot (n = 27), as well as the greatest number of samples positive for IgG only (n = 20). Thus, both MTTT were notable for the earlier and more sensitive detection of IgG seroconversion compared to immunoblot in early localized and early disseminated Lyme disease, providing greater diagnostic confidence. There were more cases of IgG-only-positive samples in the two MTTT algorithms that did not have symptoms compatible with active Lyme or a prior history of Lyme documented in the chart. These may represent past Lyme infection or false-positive results. Diasorin MTTT, Zeus MTTT, and immunoblot detected 43, 49, and 45 total cases of active Lyme confirmed by positive tier-two testing and chart review. For cases determined to be unlikely active Lyme, 14 were positive by both MTTT algorithms; however, 6 were from patients with evidence of previous Lyme disease; for the 8 remaining cases prior Lyme diagnosis was not documented in the chart. For STTT, 11 tier-two-positive samples were considered unlikely active Lyme, and 4 of these had histories of prior Lyme. Thus, the number of potential false-positive detections was similar between the two MTTT assays (n = 8) and STTT (n = 7), but these results could also reflect true past Lyme infections in a highly endemic area.

DISCUSSION

The laboratory diagnosis of B. burgdorferi infection is suboptimal. Due to the inability to reliably detect B. burgdorferi in clinical specimens in the early Lyme disease, laboratory diagnosis relies on detecting the antibody response, which takes time to develop and for IgM especially may be nonspecific. Thus, both false negatives and false positives are possible, and more accurate test strategies are needed.

Recent publications have found that the MTTT algorithm using ELISA for second-tier testing detects more early Lyme infections than STTT using immunoblots (8, 10, 16–18). Thus, we sought to compare our STTT with two MTTT assays, the ELISA-based Zeus MTTT, which has been the subject of a previous study (19), and the more recently available CLIA-based Diasorin MTTT.

Of 537 samples included in the final analysis, 91 (16.9%) were positive by one or more screening tests, but only 57 samples had concordant positive tier one results. Of 34 samples discordant on the tier-one screen, 27 were Zeus positive and 7 were Diasorin positive. By chart review, 7 Zeus-positive, but no Diasorin-positive, discordant samples were deemed early localized Lyme.

When tier-two test results using the two MTTT platforms were compared to STTT results as the reference standard for IgM, Zeus MTTT detected significantly more IgM positives than Diasorin MTTT, and both MTTT detected more IgG positives than STTT. When a 2-out-of-3 consensus reference standard was used for tier-two IgM tests, these differences were confirmed. These results suggested a lower sensitivity for IgM detection by Diasorin MTTT compared to both Zeus MTTT and STTT second-tier testing and a greater sensitivity for both MTTT compared to STTT for the detection of IgG.

The reasons for the lower sensitivity of the Diasorin IgM assay are unclear. In general, CLIA methods claim to have greater sensitivity and specificity than ELISA methods. However, manufacturers have latitude in defining positive and equivocal cutoffs for their assays. Diasorin may have opted to prioritize specificity over sensitivity for their Lyme IgM assay. While the Zeus ELISA detected more true Lyme IgM-positive results, it also generated more false-positive results. In addition, a subset of Zeus IgM positive samples were negative by first-stage Diasorin testing. Finally, the interaction of Borrelia spp. antigen selection and assay thresholds cannot be determined.

Based on a chart review of all screen-positive or -equivocal samples, more cases of active Lyme were missed by tier-one screening or not confirmed by tier-two testing by Diasorin MTTT compared to Zeus MTTT. Whether falsely negative tier-one or tier-two testing impacts the decision to treat is variable. However, providers may decide not to prescribe antibiotic therapy if screening or tier-two tests are negative and the patient has nonspecific symptoms. Thus, identifying these additional cases would be expected to improve patient outcomes.

There were three cases of early localized Lyme disease and two cases of early disseminated Lyme, all with EM lesions, that were positive for IgG-only on the Diasorin tier-two test. One “possible” Lyme case with systemic symptoms but without EM rash was also positive for IgG only. Detection of IgG alone in early Lyme was both surprising and concerning since, if EM rash is not seen, detection of only IgG could be interpreted as past infection rather than early active infection and treatment may not be given. Among these six cases, three were positive for both IgM and IgG, and three were IgM-only positive by Zeus MTTT. No cases of early localized or early disseminated Lyme were positive for IgG only by Zeus MTTT or STTT.

An advantage seen in both MTTT platforms was the enhanced detection of IgG in early Lyme compared to immunoblots, leading to a reduction in IgM-only results compared to STTT. The detection of IgM-only on tier-two tests can represent IgM false positivity, a recognized limitation of all IgM assays (19). Our laboratory test interpretation for IgM-only results alerts providers to the possibility of a false-positive IgM due to other infections and autoimmune diseases, and we recommend follow-up testing to document seroconversion of IgG if indicated. However, if the patient is treated promptly, a detectable IgG response may not develop leading to a failure to seroconvert. Thus, the earlier detection of IgG on the initial sample by the two MTTT platforms not only provides greater confidence in result accuracy but also reduces the need for follow-up testing.

Among the nine cases classified as “possible” Lyme disease included in Table 4, six patients were diagnosed with babesiosis (see Table S1). While these may represent false Lyme reactivity, dual infections with Babesia sp. and B. burgdorferi cannot be excluded since close to 40% of Babesia sp.-infected I. scapularis ticks in Connecticut also carry B. burgdorferi (https://portal.ct.gov/-/media/CAES/DOCUMENTS/Tick_Testing/Summary-of-Tick-Testing-Results-2021.pdf).

Our study was limited by the relatively small number of samples that were tested in a single laboratory. Nevertheless, we found significant differences between the two MTTT assays and the incumbent STTT algorithm. Chart reviews were used to help resolve discrepant results to the best of our ability, but documentation in the medical record is often incomplete for outpatients, patients may be unaware of tick bites or rashes, and nonspecific presentations or dual infections with other tickborne pathogens contribute to uncertainty. Few samples have follow-up testing that might assist in establishing test accuracy and for those that do, results may be impacted by therapy. In addition, we did not perform a chart review on specimens testing negative by all screening methods, and some cases of acute Lyme disease could have been missed, especially in patients with impaired antibody responses. Nevertheless, we strongly felt that, since we had access to medical records, clinical correlation by chart review should be attempted to better assess discordant results and compare test accuracy.

Our study was conducted during the peak of Lyme disease season in a highly endemic area, and we did not assess test performance at times or in regions of low prevalence. While this could be specifically investigated, we would expect a lower proportion of IgM positives to be true positives in the off-season by both STTT and MTTT, but the earlier detection of IgG seroconversion by MTTT would continue to provide greater confidence in result accuracy as was seen in our study.

Although not the focus of this study, a major reason for evaluating the Diasorin MTTT assay was the great advantage of random access instrumentation. Testing can be promptly initiated for the initial screen upon receipt in the laboratory, and positives can reflex to tier-two testing while the samples are still on the instrument. This allows for more efficient workflow, less hands-on technical time, and shorter time to result than batched ELISAs. By contrast, Zeus testing is batched for the initial screen, and positives are then batched in a separate run for tier-two testing. In smaller laboratories, batch size may restrict testing to several times a week especially for tier-two testing, thus greatly increasing the time to final result compared to Diasorin MTTT.

In sum, both MTTT options showed better sensitivity for IgG detection than immunoblots, leading to fewer IgM-only results and greater diagnostic confidence for early Lyme. Zeus MTTT detected more IgM positives than Diasorin MTTT and more active Lyme cases than Diasorin or STTT. Our findings also suggest that the sensitivity of Diasorin IgM detection should be improved. Clinicians should be aware that IgG-only detection can occur in early Lyme infections with Diasorin MTTT and that this result does not exclude active Lyme in patients with compatible symptoms and risk factors.

The views expressed do not necessarily represent the views of Yale New Haven Hospital, the Yale University School of Medicine, the State of Connecticut, or the U.S. Federal Government.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/jcm.00139-24.

Tables S1 and S2. jcm.00139-24-s0001.docx.

Classification scheme and detailed testing data.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.