Quality control assessment of Canadian laboratories testing for Lyme disease

H ARTSOB, M GARVIE. Quality control assessment of Canadian laboratories testing for Lyme disease. Can J Infect Dis 1991 ;2(1):4 1-45. In June 1990 a quality control assessment was undertaken of Canadian public health laboratories testing for antibodies to Borrelia burgdorjeri, the etiological agent of Lyme disease. Twenty sera were distributed to nine laboratories, including 12 obtained from patients in Lyme endemic areas and presumed to be serological positives, and eight prescreened negative controls. Seventeen serological reports were submitted, comprising nine enzyme-linked immunosorbent assays (ELISA). six immunofluorescent assays and two Western blot assessments. Antibodies were detected in 11 of the 12 sera which had been presumed to be positive. Assuming 11 positive sera had been submitted, the test sensitivities varied from 88.9 to 100% by ELISA, and 54.5 to 90.1% by immunofluorescent assay. Specificities were 100% for all but one ELISA and one immunofluorescent assay assessment. The results indicate a satisfactory performance by ELISA but a need for upgrading or replacement of some immunofluorescent assay tests.

L YME DISEASE WAS INITlALLY RECOGNIZED AS A DIS- tinct clinical entity in 1975 ( 1).although the etiological agent of Lyme disease, a spirochete named Borrelia burgdorjeri, was not isolated until 1982 (2).Laboratory testing for Lyme disease usually involves serological assays, since routine isolation and/or culture of B burgdorjeri from patients is presently not a viable diagnostic option (3).
Problems in the serodiagnosis of Lyme disease include low test sensitivity during the early stages of Lyme disease (4).as well as false positive serology often due to cross reactivity with other spirochetal organisms (5).Serological tests performed by most laboratories include enzymelinked immunosorbent assay (ELISA) and/or indirect immunofluorescent assay.Procedures for these assays have not been well standardized, and laboratories vary in their antigen preparations as well as their 'cutoff values for positive tests (6).
Hence considerable inter-laboratory as well as intra-laboratory variation in Lyme disease serological results have been documented between several laboratories in the United States (7)(8)(9).In June 1990 the Laboratory Centre for Disease Control (LCDC) initiated a quality control assessment of Canadian public health laboratories that currently test for Lyme disease.The results of the assessment are reported here.

MATERIALS AND METHODS
Nine laboratories, designated 1 to 9, participated in this exercise.including eight provincial health laboratories and the zoonotic diseases lab- oratory at the LCDC, which also tested sera under code.Twenty sera were included in the assessment and coded 1 to 20 by random selection of numbers.Twelve sera which were presumed to be serological positives were kindly supplied by Dr MG Golightly of the State University of New York in Stony Brook, New York, and Dr JE Craft of the Yale University School of Medicine in New Haven, Connecticut.These included two sera categorized as high ELISA reactors (sera 3 and 14), four as medium ELISA reactors (sera 4, 5, 12 and 13).two as low ELISA reactors (sera 7 and 16) and four for which no information was provided (sera 2, 10, 11 and 18).
Eight sera (1, 6, 8, 9, 15, 17, 19 and 20) were obtained from areas of Canada considered to be nonendernic for Lyme disease, prescreened by ELISA and immunofluorescent assay and included as negative controls.No conscious effort was made to include false positives in this initial assessment of Canadian public health laboratories.
All sera were diluted 1:8 in phosphate-buffered saline, and 0.4 mL aliquots of diluted sera were provided to participating laboratories.Sera were sent by courier on dry ice.All sera were apparently received promptly by the laboratories, with the exception of laboratory 5, which did not locate the serum panel until several days after shipment.

RESULTS
Reports received from participating laboratories included nine ELISAs, six immunofluorescent assays and two Westem blot assessments.The ELISAs were performed with commercial kits from  1. Two commercial and four in-house antigen preparations were employed for immunofluorescent assays.The fluorescein-labelled antihuman conjugates varied between laboratories , as did the cutoff standards for interpreting reactivity.In addition, laboratory 6 adsorbed all reactive and borderline sera with treponema-sorbent, and only considered sera positive if reactivity was not absorbed out.Western blot assays included one complete assessment by laboratory 4 with an in-house Western blot (11).and a partial assessment (sera 6, 7, 8, 16, 18 and 20) by laboratory 6 with the Whittaker Bioproducts accublot test system.ELISA assessments: The results included four Cambridge , three Whittaker, one Hillcrest and one in-house ELISA assessments (Table 2).There was 100% agreement for 15 of the serum samples, including eight positives (sera 2, 3, 4, 10, 11, 12, 14 and 18) and seven negatives (sera 1, 8, 15, 16, 17, 19 and 20) .Three of the remaining five sera (sera 6, 7 and 9) were reported as positive only by laboratory 9 using the Cambridge kit although laboratory 2 with its in-house ELISA reported sera 6 and 7 as equivocal.There was strong agreement on a fourth specimen, ie, serum 13 was reactive in eight ELISAS, while in the ninth, the in-house ELISA of laboratory 2 , the results were equivocal.
The only notable discordance in ELISA results was for serum 5 , which was reported as reactive by all four Cambridge assessments, positive in one and equivocal in two Whittaker reports, and negative in both the Hillcrest and in-house ELISA assessments.
There was close agreement between the commercial kits.Three of four Cambridge ELISAs detected 10 positives with 100% concordance of results, whereas the fourth Cambridge assessment from laboratory 9 reported all of these 10 sera positive, as well as three additional reactors.Further investigation revealed that laboratory 9 had problems adapting the Cambridge kit to their ELISA washer, and it is possible that a washing problem was responsible for the three extra positives.
One of three Whittaker assessments identified the same 10 positives as the Cambridge kits, whereas the other two Whittaker assessments recorded nine of these sera as reactive and the 1Oth as equivocal.The Hillcrest ELISA detected nine positives, which were all in agreement with the two other commercial kits.The in-house ELISA was slightly less sensitive, with only eight positive and three equivocal results.These equivocal results encompassed presumed positive and negative sera.Immunofluorescent assay assessments: The immunofluorescent assay results were more variable, with the number of reported reactors ranging from six to 11 (Table 2).There was 100% agreement by immunofluorescent assay for 12 sera, including five positives (sera 2, 3, 4 , 10 and 18) and seven negatives (sera 1, 6, 9 , 15, 16, 19 and 20).In addition, serum 8 was reported as negative in five and borderline in the sixth assessment.
Two sera, 13 and 17, were reported as reactive only by laboratories 6 and 4, respectively, whereas serum 14 was found positive by all except laboratory 5.The remaining four members of the serum panel yielded more divergent results.Sera 7 and 11 were positive in four and sera 5 and 12 in three of the six assessments.Western blot assessments: Nine Western blot positive sera were reported by laboratory 4, as well as two indeterminates and nine negatives (Table 2).Laboratory 6 performed Western blot on a known reactor (serum 18) and on five specimens (sera 6, 7, 8, 16 and 20) which it initially found to be borderline or reactive by immunofluorescent assay, but for which the immunofluorescent assay reactivity was removed by use of treponema-sorbent.The only specimen positive in laboratory 6 was serum 18.Thus, the only Western blot discrepancy between laboratories was serum 7, which was indeterminate in laboratory 4 and negative in laboratory 6.

DISCUSSION
Twelve sera, believed to be positive for antibodies to B burgdorferi, were included in the serum panel.Nine of these sera (2, 3, 4, 10, 11-14 and 18) were confirmed as reactors in most ELISAs, immunofluorescent assay and Western blot tests.A lOth serum, 16, categorized as a low ELISA reactor by the contributing source, was not reported as reactive in any of the 17 serological tests undertaken and thus, by consensus, must be considered nonreactive.
The remaining two sera were more problematic.Serum 5 yielded divergent results between laboratories in ELISA and immunofluorescent assays, and was categorized as an indeterminate reactor by Western blot.Serum 7 showed a pattern of reactivity in most immunofluorescent assay tests but was negative in most ELISAs (Table 2).Furthermore, laboratory 6 demonstrated that the immunofluorescent assay reactivity of serum 7 was removed by treponema-sorbent, and Western blot assays on this serum were reported as negative and indeterminate.Thus, the possibility was entertained that serum 7 was a syphilis reactor.However, this serum was negative for antibodies to Treponema pallidum by microhemagglutination assay and fluorescent treponemal antibody-absorption tests.
Currently it is difficult to discern the true extent of Lyme disease occurring in Canada.As of June 1990 provisional data have led to estimates of approximately 100 cases from seven provinces (12).Many of these cases are presumed to have been acquired outside of the country.To date, however, Long Point in southern Ontario is the only Canadian location where there is definite proof for the occurrence of B burgdorjeri ( 13).It is unclear whether cases of Lyme disease are being missed and/ or misdiagnosed in Canada at present.
A proper diagnosis of Lyme disease relies heavily on clinical presentation supported by epidemiological and laboratory input.This assessment was undertaken in an attempt to determine the current effectiveness of Lyme disease laboratory testing in Canada.The results indicate a satisfactory performance by laboratories undertaking ELISA, but a need for upgrading or replacing some immunofluorescent assay tests.

ACKNOWLEDGEMENTS:
The authors thank Dr MG Golightly of the State University of New York.Stony Brook, New York and Dr JE Craft of the Yale University School of Medicine, New Haven.Connecticut, for supplying sera for this assessment, and the following individuals and laboratories for participating in this study: Dr SHS Lee of the Victoria General Hospital, Halifax; M Fauvel of the Laboratoire de Sante Publique du Quebec, Sainte-Anne-de-Bellevue, Quebec; Dr RH Notenboom of the Laboratory Services Branch, Toronto.Ontario; Dr L Sekla of the Cadham Provincial Laboratory, Winnipeg, Manitoba; Dr E Chan of the Saskatchewan Health Laboratory and Disease Control Services, Regina, Saskatchewan; K Fonseca of the Provincial Laboratory of Public Health for Southern Alberta, Calgary, Alberta; E Prasad of the Provincial Laboratory of Public Health for Northern Alberta, Ed-

Source of reagents and cutoff values for each laboratory performing the immunofluorescent antibody and Western blot tests for antibodies to Borrelia burgdorferi
Sera were graded as reactive if ELISA ratios were greater than or equal to 0 .4,equivocalwith ratios between 0.25 and 0 .4,andnonreactive if ratios were less than 0.25.The source of reagents and cutoff values for the immunofluorescent assays and Western blot assessments are summarized in Table 6 Western blot Whittaker Antihuman lgG (as per test kit) <41gG bands 17.25.31/34.41kD bandsTABLE 2 Summary of serological test results from Canadian laboratories testing for antibodies to Borrelia burgdorferi ' There was 100% concordance by ollloborotories in all tests for the remaining 10 sera including five positive (2, 3, 4, 10, 18) and five negative (1 , 15, 16, 19, 20) sera. 1 These sera (as well as sera 16 and 20) were reactive or borderline by the cutoff established by laboratory 6 but were reported as negative because reactivity was removed with treponemo-sorbent.ELISA Enzyme -linked immunosorbent assay; /FA Immunofluorescent assay; WB Western blot; + Reactive, -Nonreactive; ± Borderline, equivocal or indeterminate three sources (Cambridge Bioscience, Hillcrest Biologicals, Whittaker Bioproducts) and with an in-house ELISA used by laboratory 2. The cutoffs for the commercial kits were as specified by the manufacturers.The in-house ELISA antigen was prepared as described by Russell et al ( 1 0), and the cutoff determined by a ratio, ie , the mean absorbance of the test divided by the mean absorbance of the positive control.