Laboratory diagnosis of Lyme borreliosis: Current state of the art and future perspectives

Transcription

1 Critical Reviews in Clinical Laboratory Sciences ISSN: (Print) X (Online) Journal homepage: Laboratory diagnosis of Lyme borreliosis: Current state of the art and future perspectives Benedikt Lohr, Volker Fingerle, Douglas E. Norris & Klaus-Peter Hunfeld To cite this article: Benedikt Lohr, Volker Fingerle, Douglas E. Norris & Klaus-Peter Hunfeld (2018): Laboratory diagnosis of Lyme borreliosis: Current state of the art and future perspectives, Critical Reviews in Clinical Laboratory Sciences, DOI: / To link to this article: Published online: 02 Apr Submit your article to this journal Article views: 186 View related articles View Crossmark data Full Terms & Conditions of access and use can be found at

2 CRITICAL REVIEWS IN CLINICAL LABORATORY SCIENCES, REVIEW ARTICLE Laboratory diagnosis of Lyme borreliosis: Current state of the art and future perspectives Benedikt Lohr a, Volker Fingerle b, Douglas E. Norris c and Klaus-Peter Hunfeld a a Institute for Laboratory Medicine, Microbiology & Infection Control, Northwest Medical Centre, Medical Faculty, Goethe University, Frankfurt/Main, Germany; b Bayerisches Landesamt f ur Gesundheit und Lebensmittelsicherheit (LGL), Oberschleissheim, Germany; c W. Harry Feinstone Department of Molecular Microbiology & Immunology, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD, USA ABSTRACT This review is directed at physicians and laboratory personnel in private practice and clinics who treat and diagnose Lyme borreliosis (LB) in patients as part of their daily work. A major objective of this paper is to bring together background information on Borrelia (B.) burgdorferi sensu lato (s.l.) and basic clinical knowledge of LB, which is one of the most frequently reported vectorborne diseases in the Northern Hemisphere. The goal is to provide practical guidance for clinicians and for laboratory physicians, and scientists for a better understanding of current achievements and ongoing obstacles in the laboratory diagnosis of LB, an infectious disease that still remains one of the diagnostic chameleons of modern clinical medicine. Moreover, in bringing together current scientific information from guidelines, reviews, and original papers, this review provides recommendations for selecting the appropriate tests in relation to the patient s stage of disease to achieve effective, stage-related application of current direct and indirect laboratory methods for the detection of B. burgdorferi s.l. Additionally, the review aims to discuss the current state of the art concerning the diagnostic potential and limitations of the assays and test methods currently in use to optimize LB patient management and provide insight into the possible future prospects of this rapidly changing area of laboratory medicine. ARTICLE HISTORY Received 8 November 2017 Revised 21 February 2018 Accepted 6 March 2018 Published online 2 April 2018 KEYWORDS Lyme borreliosis; B. burgdorferi; diagnostics; culture; PCR; antibody detection Abbreviations: Ab: antibodies; ACA: acrodermatitis chronica atrophicans; AUC: area under the curve; B: Borrelia; BC: blood culture; BSK-H: Barbour-Stoenner-Kelly-H medium; CDC: Centres for Disease Control and Prevention, USA; CLIA: chemoluminescence immunoassay; CSF: cerebrospinal fluid; EBV: Ebstein Barr virus; ECLIA: electrochemiluminescence immunoassay; ELFA: enzyme-linked fluorescence assay; ELISA: enzyme-linked immunosorbent assay; ELISPOT: enzyme-linked immunospot assay; EM: erythema migrans; EQA: external quality assessment; fla: flagellin; I.: Ixodes; IA: immunoassay; INSTAND: Society for Standardization and Promotion of Quality in Medical Laboratories; LA: Lyme arthritis; LB: Lyme borreliosis; LNB: Lyme neuroborreliosis; LRFP: large restriction fragment pattern; LSI: liquor-serum index; LTT: lymphocyte transformation test; M: Morbus; MFI: multiplex fluorescence immunoassay; MKP: modified Kelly-Pettenkofer medium; MLST: multi locus sequence typing; NGS: next generation sequencing; Osp: outer surface protein; PCR/ESI-MS: polymerase chain reaction followed by electrospray mass spectrometry; PCR: polymerase chain reaction; PFGE: pulsed-field gel electrophoresis; PTLDS: post-treatment Lyme disease syndrome; RFLP: restriction fragment length polymorphism; rrna: ribosomal ribonucleic acid; RT- PCR: reverse transcription polymerase chain reaction; s.s.: sensu stricto, in the strict sense; s.l.: sensu lato, in the broad sense; TPPA: Treponema pallidum particle agglutination; USA: United States of America; VCS: visual contrast sensitivity test; VDRL: Venereal Disease Research Laboratory; VlsE: variable major protein-like expression site Introduction Lyme borreliosis (LB), an infectious disease caused by tick-borne spirochetes of the Borrelia (B.) burgdorferi sensu lato (s.l.) complex, is regarded as the most commonly reported vector-borne infection in the Northern Hemisphere [1 4]. Annual incidence rates in Europe range from 0.001/100,000 in Italy ( ) to 111/ 100,000 in Germany and 188.7/100,000 in Slovenia (2014) [5,6]. According to the Centres for Disease Control and Prevention (CDC), the incidence in the USA in 2014 was 7.9/100,000, with the majority of cases reported in the Northeastern and upper Midwestern CONTACT Klaus-Peter Hunfeld K.hunfeld@em.uni-frankfurt.de Institute for Laboratory Medicine, Microbiology & Infection Control, Northwest Medical Centre, Medical Faculty, Goethe University, Frankfurt/Main, Steinbacher Hohl 2-26, D Frankfurt/Main, Germany ß 2018 Informa UK Limited, trading as Taylor & Francis Group

3 2 B. LOHR ET AL. states [7]. Recent modeling studies based on claims data, however, suggest significant under-reporting and predict much higher numbers of >300,000 (93/100,000) and >200,000 (260/100,000) incident LB cases annually in the USA and Germany, respectively [7,8]. The natural reservoir of LB agents is primarily small rodents, but a variety of other small- to medium-sized wild and domestic animals may also serve as reservoirs [2]. The bacteria are transmitted by members of the Ixodes (I.) ricinus complex, predominantly I. ricinus and I. persulcatus in Europe and Asia, and I. scapularis and I. pacificus in North America [4,9]. Once infected with borreliae, these three-host hard-bodied ticks (Ixodidae) retain the infection for the remainder of their lives, including through molts, thereby effectively transmitting spirochetes to the next feeding stage and potentially to their hosts [10]. The geographical presence of the disease follows a belt-like distribution and mirrors the distribution of ixodes ticks that transmit LB agents in the Northern Hemisphere (Figure 1). The infection may go without signs and symptoms, but in clinically apparent cases, typical symptoms associated with infection (Table 1) include erythema migrans (EM), neurological manifestations (e.g. polymeningo-radiculo-neuritis, also named M. Bannwarth), Lyme arthritis (LA), and acrodermatitis chronica atrophicans (ACA), which, together with some other rare manifestations, were well recognized in Europe [11 15] decades before the final discovery of the causative pathogen, B. burgdorferi s.l. The causative agent of LB, however, remained a mystery until the discovery of spirochetal bacteria in the midgut of ticks collected at Long Island, NY, in 1982 by the Swiss-borne entomologist, Willy Burgdorfer [16]. The subsequent epidemiological and laboratory investigations led to the establishment of LB as a new multi-systemic infectious disease entity, one of the most important biomedical discoveries of the twentieth century [17]. Over the last few decades, tremendous scientific progress has resulted in better understanding of the clinical syndromes of LB, provided deeper insights into the pathophysiology of the infection, continually improved laboratory diagnostics, and established well-recognized treatment options. Nevertheless, LB, like syphilis, remains a chameleon of clinical medicine [18] for inexperienced clinicians and results in numerous problems, especially in the direct and indirect laboratory diagnosis of the pathogen. This review aims to present the current state of the art for the laboratory diagnosis of LB, including a discussion on the current pitfalls and limitations, as well as the future prospects in this challenging area of modern laboratory medicine. The causative agents Spirochetes that cause LB belong to the B. burgdorferi s.l. complex (Table 2); they are spiral-shaped bacteria of 4 30 mm in length and mm in diameter (Figure 2) in the family Spirochaetaceae and belong to the genus Borrelia that comprises both the so-called relapsing fever borreliae, and the closely related LB agents. Relapsing fever borreliae are transmitted by lice Figure 1. Occurrence of the main vector ticks transmitting LB agents in the northern hemisphere (modified and adapted from Stanek et al. [192]).

4 CRITICAL REVIEWS IN CLINICAL LABORATORY SCIENCES 3 Table 1. Clinical case definitions and indications for medical laboratory investigation (modified from Stanek et al. [20]). Term Erythema migrans (early localized infection) Borrelia lymphocytoma (localized infection) Lyme neuroborreliosis (early disseminated infection) Lyme carditis (rare) (early disseminated infection) Ocular manifestation (rare) (early disseminated infection) Lyme arthritis (late manifestation) Acrodermatitis chronica atrophicans (late manifestation) Clinical case definition (indications for testing) Expanding red or bluish-red patch (>5 cm in diameter) a, with or without central clearing. Advancing edge typically distinct, often intensely colored not markedly elevated a Painless bluish-red nodule or plaque, usually on ear lobe, ear helix, nipple or scrotum. More frequent in children (especially on the ear) than in adults In adults mainly meningo-radiculitis (Bannwarth syndrome), meningitis; rarely encephalitis or myelitis; very rarely cerebral vasculitis; in children mainly symptom-poor meningitis and facial palsy Acute onset of atrio-ventricular (I-III) conduction disturbances; Rhythm disturbances, sometimes myocarditis or pancarditis. Alternative explanations must be excluded Conjunctivitis, uveitis, papillitis, episcleritis, keratitis Recurrent attacks or persisting objective joint swelling in one or a few large joints. Alternative explanations must be excluded Long-standing red or bluish-red lesions, usually on the extensor surfaces of extremities. Initial doughy swelling. Lesions eventually become atrophic. Possible skin indurations and fibroid nodules over bony prominences Medical laboratory evidence necessary b None in typical erythema Seroconversion or positive serology CSF-pleocytosis and demonstration of intrathecal specific antibody synthesis c Specific serum antibodies Specific serum antibodies Specific serum IgG antibodies, usually in high concentrations High level of specific serum IgG antibodies Supporting medical laboratory/ clinical evidence Detection of B. burgdorferi s.l. by culture and/or PCR from skin biopsy material Skin biopsy in unclear cases; histology, detection of B. burgdorferi by culture and/or PCR; recent or concomitant EM Detection of B. burgdorferi s.l. by culture and/or PCR from CSF. Intrathecal synthesis of total IgM, and/or IgG and/or IgA. Detection of B. burgdorferi s.l.- specific serum antibodies. Recent or concomitant EM Detection of B. burgdorferi s.l. by culture and/or PCR from endomyocardial biopsy material. Recent or concomitant erythema migrans and/or typical neurologic disorders Recent or concomitant Lyme borreliosis manifestations. Detection of B. burgdorferi s.l. by culture and/or PCR from ocular fluid/biopsy Synovial fluid analysis and detection of B. burgdorferi s.l. by PCR (and rarely culture) from synovial fluid and/or biopsy material Histology. Detection of B. burgdorferi s.l. by culture and/or PCR from skin biopsy a If <5 cm in diameter, a history of tick bite, a delay in appearance (after the tick bite) of at least 2 days and an expanding rash at the site of the tick bite is required. Can occur as single or more rarely as multiple erythema. b As a rule, initial and follow-up samples have to be tested in parallel in order to avoid changes by inter-assay variation. c In early cases intrathecally produced specific antibodies may still be absent. Table 2. Members of the B. burgdorferi sensu lato (s.l.) complex of confirmed or possible human pathogenic significance (modified and extended from Hunfeld et al. [21]). Genospecies Typical vectors Main reservoir Human pathogenicity Epidemiological distribution B. burgdorferi sensu stricto I. scapularis, Mammals, birds þþþ North America, Europe I. pacificus, I. ricinus, I. persulcatus (?) B. garinii I. ricinus, Birds þþþ Europe, Asia I. persulcatus B. bavariensis I. ricinus, Small mammals, birds þþþ Europe, Asia I. persulcatus B. afzelii I. ricinus, Small mammals þþþ Europe, Asia I. persulcatus B. spielmanii I. ricinus, Garden dormouse þþþ Europe I. persulcatus B. bissettiae I. pacificus, Neotoma fuscipes (þ) Europe, North America I. spinipalpis, I. ricinus (wood rat) B. mayonii I. scapularis, Mammals þþ North America I. pacificus B. lusitaniae I. ricinus Lizards (þ) Europe B. valaisiana I. ricinus, I. granulatus, I. columnae Birds? Europe, Japan, Taiwan, Korea

5 4 B. LOHR ET AL. Figure 2. Borrelia burgdorferi spirochetes. Immune fluorescent staining, oil, (epidemic relapsing fever, B. recurrentis) and ticks (endemic tick-borne relapsing fever, B. caucasica, B. hispanica, B. hermsii, etc.). Tick-borne relapsing fever is endemic to certain regions around the world and is transmitted in Europe, particularly in Eastern and Southern Europe, by soft ticks belonging to Argasidae [19]. In contrast, epidemic relapsing fever caused by B. recurrentis is louse-borne and occurs worldwide [19 21]. Not long ago, another relapsing fever borrelia, B. miyamotoi, was described in Eurasia and North America. This relapsing fever borrelia is transmitted by the same ixodid ticks as LB agents and causes a mostly mild, flulike illness, but it may also result in a more severe fever and in neurological disease, especially in immunosuppressed patients [22 25]. For the diagnosis of relapsing fever, detection of borreliae by microscopy, specific serological assays (using GlpQ protein as an antigen, especially for B. miyamotoi), and, when possible, molecular methods, is crucial [26]. The diagnosis of the relapsing fever borreliae is not reviewed here but is discussed in several recent publications on this topic [19,21,24]. A phylogenetic relationship of B. burgdorferi s.l. also exists to treponemes and leptospires that, due to possibly resulting cross-reactivity, can be relevant both for molecular and serological aspects of the laboratory diagnosis of LB [21,27 29]. The protoplasmic cylinder of the cell of LB agents is surrounded by 7 11 flagellae [26] and external surfaces are covered by an outer membrane with several immunodominant proteins. To date, six outer surface proteins (OspA to OspF) and various other diagnostically-relevant immunodominant protein components have been identified. The description and characterization of the variable major protein-like sequence expression site (VlsE), which shows no recombination and a less intense expression in culture or ticks but considerable recombination and strong expression in the mammalian host [30], is of special serodiagnostic relevance [31]. The VlsE protein itself, and biotechnologically derived components and peptides (i.e. C6), are part of many modern immunological assays designed to increase the sensitivity and specificity of diagnostic tests for early phase infection [21,32]. Sequence differences of the 5S-23S intergenic spacer region of the ribosomal ribonucleic acid (rrna) operon have been used to establish a classification of closely related genospecies in the B. burgdorferi s.l. complex (Table 2) [33,34]. The human pathogenic species belonging to the B. burgdorferi s.l. complex show marked genetic and phenotypic heterogeneity, which can complicate the microbiological diagnosis of LB [35]. Borrelia isolates can be typed by molecular methods (DNA DNA hybridization, plasmid analysis, restriction fragment length polymorphism analysis, whole genome sequencing) and by immunological methods using phenotypic OspA-specific monoclonal antibodies [10,34,36,37]. Currently, the favored method to characterize new species and interspecies relationships in the LB agents is multi locus sequence typing (MLST) that typically includes eight house-keeping gene loci [34,37,38]. At present, the B. burgdorferi s.l. complex includes 21 different genospecies. However, relevant pathogenicity for humans has been established only for B. burgdorferi s.s., B. afzelii, B. spielmanii, B. garinii, and B. bavariensis (Table 2) [34]. Human pathogenicity is very probable, but remains a matter of debate, for B. valaisiana [39], B. lusitaniae, B. mayonii, and B. bissettiae [40],

6 CRITICAL REVIEWS IN CLINICAL LABORATORY SCIENCES 5 although very recently B. bissettiae and B. mayonii were described as a possible cause of LB in clinically ill patients from Germany, the USA, and Canada [41 43]. All of the above-mentioned species that have been assumed to be pathogenic to humans, except B. mayonii, are found in Europe. B. burgdorferi s.s. and the probable pathogens B. bissettiae and B. mayonii are present in the USA, and all of the mentioned species are present in Asia except for B. burgdorferi s.s. and B. mayonii. The various genospecies of the B. burgdorferi s.l. complex are genetically very heterogeneous [34] and, despite the fact that EM is triggered by all seven human pathogenic genospecies, some exhibit specific symptoms in human infections. For example, B. afzelii is almost exclusively detected in ACA, B. garinii and B. bavariensis are often present in neurological manifestations, and B. burgdorferi s.s. mainly affects the joints [43]. B. spielmanii has so far been isolated only from EM [34,36], and the organotropisms of B. mayonii and B. bissettiae are not yet finally delineated. Direct detection methods Direct detection of B. burgdorferi s.l. According to the classical Henle Koch postulates, direct detection of pathogens remains the gold standard for establishing the diagnosis in infectious diseases [44 46], but this approach suffers from many limitations when it comes to fastidious agents and paucibacillary disease manifestations of infections such as tuberculosis, syphilis, and LB. Nevertheless, a variety of direct diagnostic laboratory approaches have been established for the detection of B. burgdorferi [2,3,10]. Such techniques can be used to demonstrate the presence of intact spirochetes or spirochete components such as DNA or protein in tick vectors, reservoir hosts, or patients. Currently, five different direct diagnostic tactics are used in the microbiological laboratory: (i) culture, (ii) nucleic acid-based methods, (iii) microscopic detection methods, (iv) direct detection of B. burgdorferi s.l.-specific proteins, and (v) xenodiagnosis. Of these, only culture and xenodiagnoses of B. burgdorferi s.l. detect viable organisms, and undoubtedly these offer the best confirmation of active infection in patients [2,3]. Bacterial culture has been increasingly used by researchers on both sides of the Atlantic, and the availability of cultured organisms has made investigations on the structural, molecular, antigenic, and pathogenetic properties of the different species of the B. burgdorferi s.l. complex possible [2,3]. Direct detection of Borreliae from ticks Various methods can be applied to detect the presence of LB agents in tick vectors [10,24]. Widely used approaches with only partly known or highly variable sensitivity, specificity, and reliability include culture, multiple formats of polymerase chain reaction (PCR) (mostly nested PCR that targets different genomic loci, reverse-line blotting based on hybridization of amplified B. burgdorferi s.l. genes with specific probes, and multilocus sequence analysis of amplified genetic fragments of B. burgdorferi s.l.), and microscopy of stained spirochetes in the tick midgut or salivary glands [2,3,38,47,48]. The most recently applied techniques also include next-generation sequencing (NGS), broadrange PCR combined with electrospray mass spectrometry (PCR/ESI-MS), and proteomic approaches [10,49]. These methods are invaluable research tools and facilitate epidemiological studies [10,24], but they have not gained traction in diagnostic laboratories to date. Alarmingly, some home-use point-of-care-test kits are available to allow individuals to test collected ticks for the presence of borreliae [24]. The reliability of all these tests for the direct detection of borreliae in ticks, however, is very concerning because, even in certified microbiological laboratories, both false negative and false positive results do occur. Moreover, such tests have not been standardized or evaluated by external quality assessment (EQA) schemes [24,50] and sound clinical application studies are largely absent. Most importantly, the detection of a pathogen in the vector does not necessarily imply that it has been successfully transmitted to the host upon which the tick has fed [24]. Vector competence and transmission dynamics are complex, multifactorial and beyond the scope of this review, but even successful transmission of a pathogen to the host will not necessarily result in active infection or clinical disease [24,51]. In addition, tick bites frequently go unnoticed and may demonstrate only that the individual has been in a risky environment; they do not necessarily correlate with infection. Recommendation: The use of the above-mentioned tests is of modest value for diagnosis and should be limited to epidemiological studies [24]. Xenodiagnostic approaches A recent investigation examined the safety of using I. scapularis larvae for xenodiagnosis for the detection of B. burgdorferi s.l. infections in humans [52,53] including individuals with post-treatment Lyme disease syndrome (PTLDS). Briefly, laboratory-reared larval I. scapularis ticks were placed on 36 subjects and allowed to feed

7 6 B. LOHR ET AL. to repletion. Ticks were then tested for LB agents by a highly sensitive diagnostic approach combining PCR, culture, and/or isothermal amplification followed by PCR and ESI-MS. In addition, attempts were made to infect immune deficient mice by tick bite or molted nymphal ticks or inoculation of tick contents. In principle, xenodiagnosis was well tolerated, with the most common adverse event being mild itching at the tick attachment site. Xenodiagnosis was negative in 16 patients with PTLDS and/or high C6 antibody levels and in five patients who completed antibiotic therapy for erythema migrans. Xenodiagnosis resulted in ticks positive for B. burgdorferi s.s. DNA from one patient at an early stage of antibiotic therapy for erythema migrans and from another patient with PTLDS [52]. However, these methods present insufficient evidence to conclude that viable spirochetes were present in either patient [54]. Recommendation: Currently xenodiagnostic methods are not recommended outside the experimental setting. Such methods need further clinical evaluation to address the question of borrelial persistence post treatment, and further clinical studies are needed to determine the sensitivity of xenodiagnosis in patients with LB and the possible significance of a positive result [53 55]. Direct microscopy and borrelia antigen detection from clinical samples Direct microscopic detection of B. burgdorferi s.l. has limited clinical utility in the laboratory confirmation of LB due to the sparseness of organisms in clinical samples [2,3,56 58]. This remains especially true for methods such as the modified microscopy protocol (LMmethod) that has recently garnered significant publicity. The LM-method has been promoted for direct detection of tick-borne pathogens from patients blood after a tick bite. However, in a recent clinical evaluation of the LM-method, structures interpreted as borreliae and babesiae could not be verified by PCR and the method was determined to be invalid [50]. Such clinical diagnostic studies underline the importance of doing proper test validation before new or modified assays are introduced in the field of routine diagnostics [50]. Antigen detection assays (aside from PCR) also suffer from the same limitations as microscopic detection. Although antigen capture tests have been used to detect B. burgdorferi s.l. antigens in cerebrospinal fluid (CSF) of patients with neuroborreliosis [2,59,60] and in urine samples from patients with suspected LB [61 63], such assays are not regarded as useful because of their lack of diagnostic validity under routine laboratory conditions, and many guidelines advise against their use [2,3,32,64,65]. Recently, the application of a nanotrap technique in combination with specific immunoassay technology for highly sensitive measurement of urinary excretion of the OspA carboxyl-terminus domain in early LB was proposed [66]. This method may be an interesting and novel approach but cannot be regarded as an established diagnostic method without further evaluation and validation [67]. Recommendation: The reliability of the methods discussed here for clinical practice is poor or at best, questionable [3,32,63]. These methods are not recommended by the currently established national and international guidelines for laboratory procedures for LB diagnosis [3,32,64,65,68 71]. Culture of B. burgdorferi s.l. directly from clinical samples Direct detection by culture with modified Kelly Pettenkofer (MKP) and Barbour Stoenner Kelly-H (BSK-H) medium is considered to be the gold standard and to provide clear proof of infection with B. burgdorferi s.l. [2,3,24,68]. Usually, the best borrelial growth is observed at 33 C under microaerophilic conditions after two to three weeks, while no growth usually occurs at 4 C. In a recent comparative study, the growth rate of B. afzelii was found to be significantly slower in comparison to B. garinii and B. burgdorferi s.s. at 22, 28, 33, and 37 C[72]. In contrast, there was no statistically significant difference in the growth rates of B. garinii and B. burgdorferi s.s. at 22, 28, 33, and 37 C. However, the inoculum had a statistically significant influence on growth of all three Borrelia spp. at all tested temperatures except at 4 C. In another study on the performance of MKP and BSK-H for borrelial culture, initially the numbers of spirochetes were higher in BSK- H than in MKP; however, in comparison with MKP, the bacteria subcultured in BSK-H medium were more frequently irregular, thin and non-motile, and rapidly died. In addition, the successful isolation rate from EM in skin samples was higher in MKP than in BSK-H medium (108/171, 63.2% versus 70/171, 40.9%; p <.0001). Although BSK-H medium supports better initial growth of borreliae, MKP is superior with regard to the isolation rate, morphology and motility of cultured spirochetes [73]. Direct detection of spirochetes from LB skin manifestations by culture is frequently successful. The sensitivity of culture in European studies is between 40% and 90% for EM and between 20% and 60% for ACA [10,32,64,65,68]. To a limited degree, detection by culture is also possible from CSF (10 26%), in very rare cases from synovial fluid and synovial biopsies

8 CRITICAL REVIEWS IN CLINICAL LABORATORY SCIENCES 7 (anecdotal, <1%), and from blood culture (BC) (9% [Europe] to >40% [using larger volumes of blood from EM patients in the USA]) [2,3,68]. Approximately 45% of untreated American patients with early LB associated with EM have a positive BC based on microscopic detection of B. burgdorferi s.l. in BSK-H medium after 2 12 weeks of incubation [74]. The use of culture combined with real-time PCR can boost the detection of B. burgdorferi s.l. from BC to 70.8%. This approach can also reduce the detection time to just 7 d of incubation of the BC and result in positive results for more than 90% of patients with EM and near 100% for patients with multiple EM [74]. This approach, however, is costly and impractical because of the necessity of taking larger amounts of blood from patients with EM, a condition that can usually be diagnosed and treated clinically. In addition, the positivity rate of BC is clearly much lower in late infections and in those without concomitant cutaneous disease manifestations [75]. In individual cases, the detection of B. burgdorferi s.l. has also been achieved from other tissue samples, e.g. heart muscle and iris biopsies [76,77]. In terms of laboratory diagnostics, cultivation of borreliae from patient samples generally remains laborious and costly, and it usually takes more than 2 weeks to detect growth of spirochetes [2,3,32,65,68]. Regarding the specificity of the complex and error-prone cultivation procedure, it is important to note that contamination can occur. In a recent study, the authors claimed that they could cultivate B. burgdorferi s.l. from serum in 94% of 72 patients [78]. In an assessment study, however, it was shown that almost all of these isolates corresponded to one of the control strains [79]. Recommendation: Because of the invasiveness of sample acquisition, direct detection by culture should be reserved for clear indications following cases definitions such as those proposed by Stanek et al. [20], and the procedures should be performed only in specialized reference laboratories and reference centers [32,65,69 71]. In addition, molecular confirmation of positive culture results and the characterization of the isolate to the genospecies level is required to guarantee specificity of the results [32,65]. Methods for molecular diagnosis and typing Methods for molecular biological diagnosis Important requirements for high sensitivity and specificity in molecular diagnostic tests is to choose the correct analytical technique for a given clinical manifestation, to improve the acquisition and pre-analytical processing of samples, to enhance extraction of target DNA, and to optimize the performance of molecular detection methods [32,65]. The clinical presentation and stage of disease together with a clear diagnostic indication should dictate the selection of the appropriate specimens, for example, skin biopsies for EM and ACA, CSF in suspected cases of neuroborreliosis, and synovial fluid or biopsies in LA, as recently outlined by Stanek et al. [20]. Most European and American guidelines advise against blood and urine as specimens for molecular diagnosis because the low and mostly transient presence of spirochetes results in highly variable performance of such materials for molecular testing, at least under routine diagnostic laboratory conditions [10,32,64,65,69 71,80]. Although it is difficult to obtain clinically, the largest possible amount of sample material (e.g. 2 3mm biopsies, >1 ml of CSF or synovial fluid) should be collected because of the extremely variable borrelial load in most clinical material (e.g ,000 spirochetes in 2 mm EM skin biopsies and 20 41,000/mL spirochetes in synovial fluid as determined by quantitative real-time PCR) [2,3,81,82]. Additionally, these samples must be sent directly to the diagnostic laboratory as quickly as possible under optimized transport conditions (at 4 8 C for less than 2 h) [10]. Detection of specific DNA works better from fresh or fresh-frozen material compared with stored or fixed specimens [80,83]. To circumvent the problems associated with a high ratio of human to bacterial DNA in most samples, modified manual and automated extraction methods that include specific lysis and degradation of human material (cells and DNA), selective binding, separation, and removal of human DNA, CpG motive-based selection, with final enrichment of bacterial DNA, have been developed. These reagents are in part commercially available, and extraction methods for PCR diagnostics have been extensively discussed in a recent review by Ruzic- Sabljic and Cerar [10]. Knowledge of the genetic architecture of the B. burgdorferi complex is an important prerequisite for the development of well-designed tools for the molecular diagnosis of LB [10]. The B. burgdorferi s.l. genome consists of a linear chromosome (app. 910 kb) and linear and circular plasmids that can differ in number and size (comprising up to 40% of the genomic DNA) [84]. Whereas housekeeping genes are mainly located on the chromosome, genes involved in virulence and hostpathogen interaction are predominantly situated on the plasmids. The organizational structure of ribosomal genes is unique and comprises a single 16S rrna gene and a duplication of the 23S-5S region interspaced by a non-coding region, the 5S-23S spacer [10,85,86].

9 8 B. LOHR ET AL. Table 3. Sensitivity and specificity of molecular diagnostic detection methods for LB (modified from Ruzic-Sabljic and Cerar [10]). Clinical specimen No. of studies No. of patients Targets (genes) Median sensitivity (range) (%) Specificity (%) Skin biopsy EM p66, 23S rdna, flagellin, rrf-rrl, ospa, reca, 68 (30 81) Europe S rdna, OspC 70 (30 80) USA (33 81) Skin biopsy ACA p66, ospa, chromosomal DNA, 23S rdna, rrf- rrl, flagellin 75 (20 100) 100 CSF chromosomal DNA, ospa, flagellin, rrf-rrl, SrRNK, p (5 100) Europe (9 100) USA (5 93) Synovial fluid rrs-rrl, ospc, ospa, p66, flagellin 77.5 (23 100) 100 Europe (23 100) USA (60 100) Blood, serum or plasma polc, ospa, 16S rdna, rrf-rrl, rpoc 18 (0 100) Europe ( ) USA (0 62) Conventional, isothermal, qualitative, and quantitative real-time PCR assays targeting the unique rrna gene (16S rdna, 23S rdna, 23S-5S rdna intergenic spacer), a variety of other single copy chromosomal targets (flagellin (fla), hbb, rrf-rrl, polc, SrRNK, p66, reca, bmpa, rpob, rpoc,and gyr), plasmid bound targets such as dbpa, vlse, and outer surface protein genes (ospa, ospb, ospc), have been developed and are, in part, commercially available [2,3,10,85]. An important drawback to a more standardized application of such assays is the fact that some molecular targets, especially the plasmid encoded target genes (e.g. ospa, ospb, ospc), have high inter- and intra-strain variability that can lead to amplification drop outs (false-negative results). Others (e.g. fla) lack discriminatory power between the different Borrelia genospecies because of their more conserved genetic nature. An overview of these targets and their potential usefulness for the detection of borreliae in different clinical specimens is provided in Tables 1 and 3 [2,3,10,21,29]. Assays targeting plasmid-encoded genes (e.g. osps, vlse) are believed to be more sensitive than those targeting chromosomal ones (e.g. 16S rrna-, flagellin (fla)-genes). This is likely because borreliae often shed plasmid containing blebs, which leads to a potentially higher copy number of plasmid-encoded genes in comparison to those encoded in chromosomal DNA [80]. Interestingly, such blebs may also persist in body fluids and be sequestered into tissues of infected individuals [87]. This can lead to false positive PCR results that do not necessarily reflect ongoing active disease. In contrast, chromosomal targets such as the 16S rdna that occur primarily as single copy genes may lead to lower sensitivity but may be better predictors of viable borreliae [80,87,88]. Such analytical limitations and technical variabilities have led to expert recommendations that molecular tests should employ at least two different DNA target sequences to increase diagnostic reliability [32,65]. PCR followed by electrospray ionization-mass spectrometry (ESI-MS) A relatively recently introduced molecular approach is broad-range PCR followed by electrospray ionizationmass spectrometry (ESI-MS). Briefly, this approach uses a conventional real-time PCR assay utilizing broad range primers that target conserved sequences of eubacterial DNA. Resulting amplicons are subjected to ESI-MS, which measures the mass of the amplicons with sufficient accuracy to determine the nucleic acid base composition [49,89]. Unlike hybridization-based detection, ESI-MS does not require prior knowledge of the target sequence but identifies the composition of amplicons based on mass with reference to databases without prejudice. This PCR/ESI-MS approach does not produce as detailed information as DNA sequencing, but it provides identification of the detected pathogen including additional information about the genotype and the possible presence of genetic markers and virulence factors, and it can be applied to a wide range of vector-borne pathogens [49,90]. In a recent study using this method on 1.5 ml whole blood samples from patients with early LB, higher sensitivities (67%; 95% CI: 40 79%) of detection were revealed than with methods previously discussed for BC from EM patients [91]. PCR followed by ESI-MS may be more useful for rapid borrelia detection, but it is currently not recommended because it needs more extensive clinical evaluation to demonstrate that it is not limited by the same general drawbacks in sensitivity in patients with late stage disease as the other PCR-based detection methods. As well it is costly and technically demanding. Sensitivity of molecular diagnostics In general, the sensitivity of molecular test methods (mainly PCR-based) for the detection of borreliae under

10 CRITICAL REVIEWS IN CLINICAL LABORATORY SCIENCES 9 routine laboratory conditions correlates with the known detection limits for culture [2,3,10,21,29,32,65]. In principle, the detection of borreliae from a skin biopsy from EM and ACA using nucleic acid amplification techniques (usually PCR) is very reliable (72%) (Table 3), and in the case of early manifestations, may be more sensitive (68%) than serological antibody detection (40 60%) [32,65]. However, the question remains whether molecular testing, which is costly, is necessary for most skin manifestations. For the typical EM, the molecular test is clearly not warranted [20,49,80]. The sensitivities of molecular diagnostic tests for clinical specimens other than skin biopsies, based on a recent meta-analysis [10], are in the range of 20% for CSF, approximately 77% for synovial fluid, and at best 18% for blood (summarized in Table 3) [10,29,92 96]. Patients with LA remain a diagnostic exception (Table 1); with this manifestation, molecular tests can achieve much higher sensitivities (77.5%; Table 3) than culture (<1%) [10,29,32,94]. This is why molecular investigation of synovial fluid or synovial biopsies is of the highest importance in diagnosing LA [10,32,64,65], whereas molecular diagnostics for urine samples is generally disregarded due to its inadequate analytical specificity [97 99]. Most importantly, positive results should be confirmed by amplicon sequencing and identification of the genospecies [32,65]. After treatment, B. burgdorferi s.l. DNA can still be detected for weeks or even months in samples taken from formerly affected areas of skin [100] and in treated LA [101,102]. Since PCR does not discriminate between residual DNA and viable organisms, no conclusions should be drawn as to whether the therapy has failed, especially in patients without typical symptoms [100,101]. Indeed, molecular detection of pathogens borreliae without the simultaneous presence of typical disease manifestations is not clinically relevant [101,103,104]. Problems with molecular test standardization A significant drawback to molecular diagnosis of LB is the absence of standardization of methods [10,29,105]. This applies to DNA isolation and selection of the clinical sample as well as elution volumes, analytical conditions, and the selection of the molecular targets as outlined above. Results from EQA programs for these diagnostic techniques are currently too heterogeneous to demonstrate that one method is better than another [21,29,106]. This is further supported by evidence from the EQA scheme run biannually by INSTAND e.v. (Society for Standardization and Promotion of Quality in Medical Laboratories) for molecular detection of B. burgdorferi s.l., which cover several major commercial manufacturers and many in-house assays. Between 2013 and 2015, the average pass rates for 241 participating laboratories enrolled in this program were 86 97% for positive samples and % for negative samples [107], reflecting in part the situation in molecular diagnostic laboratories for LB. Due to the invasiveness of sample acquisition, direct detection by PCR should be based on a clear indication (e.g. unexplained skin or joint manifestations after excluding other differential diagnoses) and explicitly remain limited to specialized reference laboratories and specific tissues. Many guidelines for the management of LB, therefore, advise against the broad use of molecular detection methods for the diagnosis of LB from skin biopsies or blood [32,64,65,69 71], and the CDC has cautioned against their use on specimens such as blood and urine that it considers inappropriate [108]. Recommendation: Direct molecular (mostly PCRbased) detection methods should not be used as primary screening tools if there is suspicion of LB. Importantly, a negative PCR test result does not exclude LB. Positive results need to be confirmed with regard to specificity (e.g. probe hybridization, sequencing of the amplicon), and identification of genospecies must be indicated in the laboratory report. A positive PCR test result after adequate therapy and without typical clinical manifestations has no clinical relevance. Direct molecular detection should be limited to patients with defined skin and joint manifestations and should be performed exclusively by specialized reference laboratories [20,21,32,64,65,68 71]. Genetic typing methods of amplicons and isolates Genetic characterization of B. burgdorferi s.l. isolates is mainly relevant for epidemiological, clinical, and evolutionary studies [10,38]. However, it also has value in the diagnostic laboratory to determine and identify positive test results from molecular and culture-based diagnostics for LB [48]. A variety of molecular methods that have been developed for these purposes include large restriction fragment pattern (LRFP) analysis plus plasmid profiling, PCR-based typing techniques targeting single genes, PCR-based restriction fragment length polymorphism (RFLP) analysis of rrs-rrla (16S-23S rrna) and rrfa-rrlb (5S-23S rrna) intergenic spacer, molecular ospc-analysis [48], flagellin typing, real-time PCR-based typing using melting temperature analysis, MLST of mostly eight house-keeping B. burgdorferi s.l. genes [38], and whole genome sequencing (WGS). Whereas studies using WGS and NGS from culture and field samples in LB diagnostics are rare and the implementation of such techniques in the routine laboratory is still

11 10 B. LOHR ET AL. under evaluation [10, ], PCR-based RFLP analysis, restriction enzyme LFRP analysis in combination with pulsed-field gel electrophoresis (PFGE), and PCR-based flagellin typing have made their way into research and diagnostic laboratories with variable success [10]. Flagellin typing is technically demanding and not widely used [115], whereas 16S-23S rrna typing has been applied primarily for epidemiological characterization of pathogenicity in B. burgdorferi s.l. strains of different geographic origins [10,116,117]. In contrast, MluI- LRFP in combination with PFGE from culture material, 5S-23S PCR-based RFLP, and real-time PCR melting temperature analysis from positive borrelia culture or directly from clinical samples are widely used in both research and diagnostic laboratories. LRFP together with plasmid profiling is laborious but very suitable for Borrelia spp. identification and subgroup delineation because of its high discriminatory power [ ]. OspC-PCR typing requires DNA-sequencing because of the high variability of the ospc gene and, like LRFP, is more or less restricted to reference laboratories [10,122,123]. In contrast, application of reverse transcription polymerase chain reaction (RT-PCR) and melting temperature analysis can be automated and is much easier to apply, so that many laboratories have established this method for diagnostic and research purposes. Such assays, however, remain problematic if they are applied directly to clinical samples [10]. One of the most widely applied methods for confirmation and characterization of B. burgdorferi s.l. isolates in many research and diagnostic laboratories is PCR-based 5S- 23S DNA RFLP analysis using MseI and DraI enzymes [124], because of its relative ease of use, high discriminatory power, and excellent reproducibility. Serological testing methods Indirect detection of the pathogen (serological testing) Because of its ease of performance and potentially substantial diagnostic value to the diagnosis of LB, indirect pathogen detection by serology is indicated in all cases of clinically suspected LB, except in typical EM, to achieve possible laboratory support for the disease [20,32,65,69 71,125]. Case definitions for Europe which contain descriptions of the various clinical manifestations of LB (Table 1) and which should also work for most other parts of the Northern Hemisphere have been published so as to establish guidelines for clinical diagnosis and to provide advice on appropriate indications for ordering diagnostic tests in cases of suspected LB [20]. Despite these definitions, laboratory testing for B. burgdorferi s.l.-specific antibodies continues to be used frequently in many clinical situations where testing is not recommended by current guidelines [8,32,65, 69,70, ]. For example, in the Netherlands, only 9% of the patients tested had clinical signs indicated in the guidelines [127]. In Denmark, 43% of samples from general practice originated from patients with suspected EM [126]. Unnecessary testing may delay proper diagnosis and treatment and will increases healthcare costs. In Germany alone, the annual cost of laboratory testing for LB in the outpatient sector was estimated to be e51 million, a substantial portion of which resulted from over-testing [8]. Epidemiological considerations for adequate serodiagnostics In this context, it remains important to consider that the less specific the symptoms, the weaker the a priori probability of LB, and the lower the predictive value of serological methods [2,3,20,68,125,128]. Confirmed indications for serological testing in suspected LB, that is, that the patient displays the signs indicated under the clinical case definitions, are summarized in Table 1. The probability that a patient with a positive serological test has LB (positive predictive value) and the probability that a patient with a negative test does not have the disease (negative predictive value) depend on the performance characteristics of a given assay (sensitivity and specificity) and also on the prevalence of the disease in a given population and geographic region [20,125,129,130]. The pretest probability of a patient having or not having LB therefore determines the predictive value of a given test result, and the clinical significance of a given test result for antibodies to B. burgdorferi must, therefore, be interpreted with caution, especially outside endemic areas [129,130]. These considerations have been discussed in detail in a recent review by the European Study Group for Lyme Borreliosis (ESCBOR) [125]. Recommendation: Relevant clinical signs must be present before LB can be suspected [20,125]. Serology is indicated only when such appropriate clinical symptoms and signs are present, and possible serological follow up is helpful only 2 3( 6) weeks after a possible infection [21,32]. Advantages and limitations of current serological test strategies Currently, many guidelines and the CDC in the USA recommend a two-tiered laboratory testing strategy that utilizes a screening test of high analytical sensitivity (e.g. enzyme-linked immunosorbent assay [ELISA]) with

12 CRITICAL REVIEWS IN CLINICAL LABORATORY SCIENCES 11 Figure 3. Diagnostic algorithm for EIA and immunoblot based two-tier testing (modified from Hunfeld et al. [152]). a confirmatory test of high specificity (immunoblot) as the most reliable diagnostic procedure (Figure 3) [21,32,64,65,131]. Recent attempts have been made to improve the currently used diagnostic test strategies by integrating new antigen components such as VlsE or closely related peptides (C6) into a new generation of immunoassays (IA). These could possibly replace the widely used two step or two-tiered approach (i.e. confirmation of positive screening results by subsequent immunoblots) and provide improved diagnostic sensitivity and specificity over those tests currently used in the USA [20]. In Europe, however, several investigations have raised doubt about whether a single serological test for LB diagnosis has adequate sensitivity and specificity to serve as a stand-alone test [20, ]. The primary concern is that the much higher biodiversity of Borrelia genospecies and strains in Europe pose a more complicated challenge in designing an appropriate IA with a simple one size fits all approach [21,132,134]. This single-test strategy may be more applicable to the USA due to the presence of fewer genospecies [135]. This assessment is further supported by a recent study that confirmed that the standard two-tiered serologic testing using assays developed for use in the USA had inferior performance compared to the European assays for the evaluation for LB that is acquired in Europe [136]. As outlined above, this result is not surprising, because LB in North America is caused mainly by B. burgdorferi s.s., whereas in Europe, there is greater species heterogeneity and infection can be caused by B. afzelii, B. garinii, B. bavariensis, B. spielmannii, and less commonly, B. burgdorferi s.s. [2,3,21,32,65]. Importantly, the specificity of stand-alone tests has been questioned by studies comparing the specificity of two-tiered (IA & C6 peptide ELISA) algorithms (99.5%) to the lower specificity of 98.4% of the stand-alone C6 ELISA [137]. Although such a difference seems negligible, in the United States alone, where at least 3.4 million LB tests are performed annually [137], such a difference would lead to an additional 37,000 false-positive test results per year, more than the reported incidence of LB in the United States (up to 35,000 cases annually) [138]. The high specificity of the two-tiered approach (Figure 3) is thus a critical advantage [135,137] and therefore it has prevailed as the gold standard for LB serology in most diagnostic laboratories in Europe and North America [21,32,65,131]. However, it is worth noting that the question of whether the second tier must necessarily be an immunoblot is debatable in the view of the known limitations of immunoblot testing, which include little standardization across laboratories and problems with specific IgM antibody detection [135]. Recommendation: A two-tiered strategy applying a screening test of high analytic sensitivity (e.g. ELISA) with a confirmatory test of high specificity (e.g. IA, immunoblot) is the most reliable diagnostic procedure currently available (Figure 3) [21,32,64,65,131].

13 12 B. LOHR ET AL. Figure 4. Heterogeneous sensitivities of different antigens of diverse European B. burgdorferi s.l. strains for the detection of the specific antibody response of patients with different manifestations of Lyme borreliosis (EM: erythema migrans, NB: neuroborreliosis, late LB: late Lyme borreliosis) (modified from Goettner et al. [134]). Relevant immunodominant antigens of B. burgdorferi Clinical serodiagnostic studies have revealed the antigenic repertoire of LB agents and identified the following immunodominant B. burgdorferi s.l. antigens as being most suited for diagnostic purposes: p83/100, which also serves as a marker of the late phase of the specific immune response; p58, p41 (flagellin), the recombinant internal fragment of p41 (p41i), which is less cross reactive than the native protein; and the outer surface proteins OspA, OspC, p39 (BmpA), and DbpA (Osp17/p18) [2,3,21,32,65,134, ]. All these antigens are available as recombinant proteins and are included in many diagnostic assays as single proteins or as mixtures of various combinations and concentrations. The known genospecies, and in part also the strain-dependent variabilities of most B. burgdorferi s.l. antigens, can result in variable serodiagnostic sensitivities, especially within Europe (Figure 4) but also between Europe and North America [21,32,35,65,134,136,142]. Such variability also significantly affects the VlsE, OspA, OspC, and Osp17 proteins and has direct diagnostic impact because the resulting borrelial surface antigenic diversity impairs serodiagnostic performance, especially in techniques such as the immunoblot [21,135,143]. For example, intraspecies differences in B. garinii detected by using a panel of monoclonal antibodies has led to the recognition of at least 13 different OspC serotypes [122]. To a lesser extent, heterogeneity in the immunoreactivity to VlsE has been described in Europe [134]. As a result, European immunoassays are often prepared from mixtures of specific spirochetal lysates and/or purified antigens including fusion proteins, and then the diagnostic criteria are adjusted to the diagnostic needs best suited for the local epidemiological situation [21,32,65,144]. Due to strong cross-reactivity, the flagellin protein, a very sensitive but not very specific marker of early infection, should be used only as a recombinant truncated internal fragment (p41 int.) [140,145]. An important finding in the sense of improved serodiagnostics was the discovery and production of recombinant VlsE protein and its conserved peptide region C6 [134, ]. Interestingly, VlsE shows no recombination and a less intense expression in culture or ticks, but considerable recombination and strong expression in the mammalian host [30]. Therefore, using recombinant VlsE as an important immunodominant antigen or the VlsE-derived C6 peptide for diagnostic purposes has dramatically improved the sensitivity and the specificity of detecting IgG antibodies against B. burgdorferi s.l. by ELISA and immunoblot during the very early stages of

14 CRITICAL REVIEWS IN CLINICAL LABORATORY SCIENCES 13 the infection and immune response [2,3,21,32,65, 134,144]. Screening tests Various modifications of IA such as ELISA, enzyme-linked fluorescence assays (ELFA), chemiluminescence immunoassays (CLIA), and electrochemiluminescence immunoassay (ECLIA) are used as screening tests because they are well-suited for the polyvalent, selective, and quantitative determination of specific IgG and IgM antibodies [150]. Other serological tests such as indirect hemagglutination assays, complement fixation or indirect immunofluorescence are no longer considered appropriate [21,32,65]. In conventional whole cell lysate assays, the patient s serum is absorbed to T. phagadenis to increase the specificity of the screening test. In addition, treatment of the serum with a rheumatoid factor (RF) absorbent is necessary for selective detection of IgM antibodies to avoid false positive results due to possible presence of RF. In most recombinant test formats, however, such pretreatment is normally not required [21,65]. The antigen fractions for specific antibody detection in the ELISAs and CLIAs consist of ultrasonicated borreliae, borrelial whole cell extract or purified recombinant proteins capable of stimulating a specific immune response in vivo; these include Osp17/p18 (DbpA), flagellin (p41), p39, p58, OspC, and VlsE (selective ELISA). Hybrid tests combining both cell-derived and recombinant antigens are also available, for example, enrichment of conventional antigen extracts with recombinant OspC or VlsE. CLIA and ELISA methods differ not only in the type of detection they utilize but also in the antigen preparations they use. The specificity of advanced IAs is 80 90% [2,3,21,32,65]. Such assays are routine in serological laboratory testing for LB agents because automated reading, accuracy of measurement, and ease of handling are assured, and these tools are readily adapted for modern high-throughput laboratories using automated analytical instrumentation [21]. The results of the available immunoassays can vary significantly depending on the specific antigen composition and the manufacturer of the test system [150]. Hence, the results of different tests and/or different laboratories are comparable only to a limited extent [8,21,150,151]. Standardization would require parallel testing across laboratories with archived control sera, but this is rarely carried out [21,32,65]. Confirmatory immunoblots In addition to conventional whole cell lysate immunoblots (Western blots), immunoblots and line immunoblots, or modified variations of similar test formats using recombinant antigens (recombinant blot), are increasingly employed (Figure 5) as confirmatory tests [152]. Hybrid tests combining both types of antigen preparation are also applied to an increasing extent. Whole cell lysate or whole cell antigen immunoblots Classical Western blots apply ultrasonicated borreliae as an antigen source (Figure 5). Thus, all B. burgdorferi s.l. antigens, i.e. specific immunodominant and non-specific proteins, are used for antibody detection. Optimal expression of the specific immunodominant antigens of the borrelial strain targeted is essential for the quality of such a blot [2,3,139]. Studies have shown that different strains of B. burgdorferi s.l. display pronounced variability in their immunodominant antigens. The B. afzelii strain PKo has proved to be particularly suitable for Europe [32,35,142, 153,154]. Criteria for the interpretation of whole cell antigen immunoblots employing the PKo strain have been compiled [68,142,152] under standardized conditions (Table 4). It is a disadvantage that immunodominant antigens such as VlsE are strongly expressed in vivo but show less intense expression in conventional culture [155]. Hence, many manufacturers of conventional lysate blots selectively add certain recombinantly produced proteins (VlsE, OspC) to form hybrid tests that close these diagnostic gaps [152]. The use of lot-specific evaluation templates and antigen localization verification with monoclonal antibodies in whole cell lysate immunoblots by the manufacturers are essential for diagnostic quality [32,68,142,152]. Recombinant immunoblots Highly specific recombinant immunoblots (selective blots) that employ antigen preparations from recombinant proteins for the detection of B. burgdorferi s.l. antibodies are being used to an increasing extent [68,152]. Some of the most common recombinant antigens for these assays are: Osp17/p18 (DbpA), VlsE, OspC, OspA, p39/bmpa, p41/i (flagellin, internal fragment), p41 (flagellin), and p83/100 (Figure 5). In addition, specific antigens of different genospecies can be used in the same test mixture to account for the immunological variability of the different Borrelia species [142]. The diagnostic sensitivity of the recombinant immunoblots depends on the type and number of antigens used. The colored bands of the antigen pattern are more easily assigned to defined antigens in the recombinant immunoblot than the whole cell lysate blot (Figure 5). The recombinant immunoblot is, therefore, recommended, especially

15 14 B. LOHR ET AL. Figure 5. Different formats of diagnostic immunoblots for the antigen- and antibody class-specific detection of the anti-b. burgdorferi s.l. immune response (modified from Hunfeld et al. [152]). for laboratories with little experience in LB serodiagnostics [152]. Recommendation: Studies examining quality control have revealed large numbers of inconsistent results between laboratories because of variability of test results of the commercially available test systems and the low recovery rates of specific immunoblot bands [8,151]. Therefore, the criteria presented in Table 4 are intended only for orientation and are not appropriate for general application and evaluation of the wide variety of differently designed commercial test kits that are available [8,21,32,65,151]. There are no criteria for the standard handling of borderline findings. Line blots In principle, line blots represent a modification of conventional recombinant immunoblots with respect to the production and processing of the antigens used. However, no electrophoretic separation step is necessary prior to the blotting procedure [152]. For such tests, the individual antigens, without previous denaturation, are gently and selectively sprayed directly onto the carrier membrane (Figure 5). The line immunoblot achieves high levels of diagnostic sensitivity and specificity for the detection of B. burgdorferi s.l.-specific antibodies [152]. Moreover, the line immunoblot technique has the advantage of resolving differences in the diagnostic reactivity of sera by combination or supplementation of tests with highly specific immunodominant antigens (e.g. VlsE, OspC) from different genospecies and strains from different geographic origins [32,134,152]. Considering the number and kind of B. burgdorferi s.l.- specific bands, these test systems also provide information as to the quality and duration of the immune response, thus allowing the result to be classified more accurately in the clinical context [20,21,125]. The identification of blot bands via lot-specific evaluation templates and their adequate weighting, as well as the parallel testing of positive controls and cutoff controls, are essential to avoid false positive findings and confusion of antigens [32,152]. These technically demanding issues and the resulting difficulties in standardization of tests and test results are confirmed by interlaboratory comparison studies and round robin trials that demonstrate the limitations of commercially manufactured conventional and recombinant immunoblots [8,150,151]. Recommendation: Based on a clear diagnostic indication, the immunoblot continues to be one of the most specific diagnostic tests for the detection of

16 CRITICAL REVIEWS IN CLINICAL LABORATORY SCIENCES 15 Table 4. Examples of established interpretative criteria for immunoblots (modified from Hunfeld and Kraiczy [152]). B. afzelii (strain PKo) whole cell antigen immunoblot evaluation criteria for Europe IgG positive: 2 bands IgM positive: > 1 band p100, p58, p43, p39, p30, OspC, p21, p17, p14 p41 (strongly positive), p39, OspC, p17 B. burgdorferi s.s. (strain G39/40) whole cell antigen immunoblot evaluation criteria (CDC recommendations for the USA only) IgG positive: 5 bands IgM positive: 2 bands p83/100, p66, p58 (not GroEL), p45, p41, p39, p30, p28, OspC, or p18 p39, OspC, p41 Recombinant immunoblot evaluation criteria IgG positive: 2 bands IgM positive: 2 bands p100, p58, p39, VlsE, OspC, p41 internal fragment, p18/p17 p39, OspC, p41 internal fragment, p18/p17 or strong reaction against OspC only B. burgdorferi s.l.-specific antibodies. Immunoblots that use recombinant antigens are especially suited for most diagnostic use and provide reliable results [21,32,65,152]. It is important to note that given reactivity with strain-specific immunoblot bands does not allow specific serological identification of the infecting genospecies in the tested patient. Multiplex fluorescence immunoassays (MFI) The introduction of novel multiple parameter test systems based on single-antigen Luminex (Luminex Multiplex Assays, Thermo Fisher Scientific, Darmstadt, Germany) technology allows simultaneous analysis of many analytes such as antigens in the same microtiter well or by flow cytometry in a single process [21]. This test method is based on tiny, antigen-coated polystyrene beads that serve as the solid phase for a variety of detection reactions that are similar to Western blots and ELISAs [156]. The resulting multi-analyte profiles combine the advantages of immunoblots with the analytical principles of quantifiable immunoassays. A study recently evaluated a multiplex-bead-based assay for the detection of serum antibodies to B. burgdorferi s.l. [157]. The assay tested IgG and IgM responses against 13 different B. burgdorferi s.l. antigens in 49 Danish and 61 Swedish patients with Lyme neuroborreliosis (LNB), 139 Swedish non-lnb patients, and 218 Danish blood donor controls. The VlsE IgG- and OspC IgM-testing showed an area under the curve (AUC) of 96% and 80% on the receiver operating characteristic curve. All the other antigens were much less discriminatory in LNB compared with controls. Recommendation: Given the complexity and cost of the assay, the results of the current study did not show much advantage over conventional diagnostic approaches, because the VlsE IgG component alone was the largest contributor to the diagnostic outcome [157]. The method, however, may be helpful in the future to serve as an assay platform for simultaneous serological testing against a variety of tick-borne pathogens, other than B. burgdorferi s.l., that are of diagnostic importance. Interpretation of serological test results The results of interlaboratory proficiency testing and scientific evaluation studies have shown that the individual sensitivity and specificity of different test methods may diverge considerably depending on the test used and test system manufacturer [8,150,151]. Prior to the introduction of a new test procedure into a laboratory, internal evaluation is necessary. For this purpose, serum samples from patients with clinically confirmed LB and serum samples from healthy blood donors (control group) should be assessed [158]. In borreliosis serology, test results are reported as negative, positive or borderline depending on the manufacturer s specification (cutoff index or U/mL). The definition of the borderline value for ELISA and the assessment of specific bands in the immunoblots from different manufacturers do not yet follow general, standardized guidelines [8,150,152]. Moreover, international, regional, or national reference preparations and materials with standardized borderline values are not available. Hence, the qualitative and quantitative test results (e.g. U/mL in ELISA, band pattern in immunoblot) are not directly comparable but depend on the manufacturer [151]. This must be considered during monitoring when one is interpreting test results from different laboratories. Recommendation: The significance of changes in test results must always be verified by parallel testing with a previously collected sample from the patient [32,65,139,159,160]. Moreover, borderline titers and immunoblot interpretations as well as the diagnostic sensitivity and specificity are dependent on the test and the manufacturer. Specified borderline titers and cutoffs are to be understood as guidance values and should be critically verified for each test used in a given laboratory through in-house performance reviews on positive

17 16 B. LOHR ET AL. reference samples and negative blood donor sera that reflect the local epidemiological situation [139,158,160]. Stage-dependent antibody kinetics in Lyme borreliosis Antibodies to B. burgdorferi s.l. antigens usually form approximately 2 6 weeks after the onset of the B. burgdorferi s.l. infection. In most cases, the IgM antibody response precedes IgG antibody production [32,65,159,161,162]. The absence of an IgM response has been reported in some cases [32]. An IgM response may also be absent during a reinfection event because reinfections are usually associated with significant IgG response without major IgM production [162]. In the early phase of infection, the immune response of both immunoglobulin classes is at first directed against a narrow range of B. burgdorferi s.l. antigens, especially flagellin (p41), VlsE, and OspC. Antibodies to VlsE and OspC are of special diagnostic significance because of their relatively high specificity. The introduction of the VlsE antigen has provided better sensitivity in serodiagnostic testing for B. burgdorferi s.l. [31,134,144]. If it is used in the early stages of disease manifestation (e.g. EM, LNB), this antigen achieves high detection rates for B. burgdorferi s.l.-specific IgG antibodies. However, many patients (50 70%) still remain seronegative in the early stage of infection [32,65]. The positivity rates investigated for various disease manifestations within the scope of studies of seroprevalence are summarized in Table 5. The number of seropositive patients increases as the spirochete infection progresses and reaches almost 100% in late disease [20,32,65,125]. In these cases, the immune response is directed against a wide range of B. burgdorferi s.l.-specific antigens. Antibodies against specific antigens such as the p83/100 protein, p39 (BmpA) and Osp17/p18 (DbpA) are of particularly high diagnostic significance during the late stage of the immune response. In contrast, antibodies against OspA, which are also specific, are rare and are most often observed in patients with LA [21,32,65,152]. As with other infectious diseases, immunocompromised patients may show delayed or a complete absence of an immune response. As a rule, seronegative LB is extremely rare in immunocompetent patients except in the very early stage of the disease. In patients with a short duration of disease, direct detection of the pathogen should always be considered [20,21,32,65,125]. Specific B. burgdorferi s.l. IgM antibody testing Detection of IgM antibodies to LB agents is useful for early infection, but their detection does not contribute to serodiagnosis of late LB [20]. As outlined above, stand-alone IgG assays clearly suffer from limitations, at least in the non-american setting, but such assays may suffice if highly sensitive screening tests that include the VlsE protein or C6 peptide are utilized [21,32,65]. In this instance, additional IgM detection appears to have little significant advantage over IgG testing in the recognition of early LB, and it may actually reduce the specificity of diagnostic testing in ambiguous clinical situations, although this may depend on the antigen mix applied in the assay [163,164]. In patients with extended disease manifestations such as chronic LNB, ACA, or LA, only the detection of IgG antibodies to B. burgdorferi s.l. should be considered diagnostic [20]. This works well because, in some individuals, a lowgrade IgM antibody response can persist for months or even years after treatment or a past infection, although this phenomenon is not associated with a (persistent) infection with B. burgdorferi s.l. [ ]. Recommendation: In late LB manifestations, the presence of an IgG response is required for the diagnosis of LB, according to current case definitions [20]. In contrast, an isolated and persistent IgM response clearly speaks against a long-lasting infection or late manifestation of LB [20,21,32,65]. Table 5. Type of disease and prevalence, and kind of B. burgdorferi s.l.-specific antibody response (modified from Rauer et al. [170]). Type of disease Type of antibody response/comments Prevalence of antibodies (Early) localized disease > 3 weeks post infection occurrence of IgM antibodies (may be missing in re-infection) 20 to >50 % Early predominance of IgM 3 6 weeks post infection occurrence of IgG Antibody response can be absent in early infection Early disseminated disease Antibody response similar to localized disease 70 to >90 % After 2 weeks start of intrathecal antibody response Normally presence of IgM und IgG antibodies After >6 weeks of symptoms >99% of LNB cases show a positive intrathecal antibody response Late manifestations High IgG antibody concentration IgM antibody response is variable Only IgG is of diagnostic relevance Intrathecal IgG antibody response is a must in late LNB Almost 100 %

18 CRITICAL REVIEWS IN CLINICAL LABORATORY SCIENCES 17 Interpretation in the case of a negative screening test In the case of a negative screening test result, no further investigation is needed. However, a negative serological finding does not exclude LB. Patients during early disease stages often present with negative serological results. If infection continues to be suspected, the serological diagnosis can be repeated after 2 6 weeks. In individual cases, a measurable immune response may have been suppressed by early initiation of successful antibiotic treatment, which would prevent seroconversion for IgG and IgM from occurring (the socalled abrogative antibody response with absence of IgG and/or or IgM seroconversion) [20,21,32,65,125]. Interpretation in the case of a positive screening test A borderline or positive test result must be interpreted with caution because false positive findings occur, e.g. in cases of syphilis, other bacterial infections, and Ebstein Barr virus (EBV) infection, despite absorption of cross-reacting antibodies by T. phagadenis [169]. In the case of a borderline or positive result of the LB screening test, a TPPA (Treponema pallidum particle agglutination) test should be performed (see above) to exclude present or resolved syphilis as a possible cause of false positive borreliosis serology. Recommendation: A positive test result gives strong evidence of active or possibly past B. burgdorferi s.l. infection. If this serological finding coincides unambiguously with the suspected clinical diagnosis (Table 1), further laboratory investigations are not required. In all other cases, further confirmatory diagnosis (Table 1) must be performed [20,21,32,65,125]. Interpretation in the case of a negative confirmatory test A negative confirmatory test (IA, immunoblot, or line blot) suggests that the screening test provided a false positive result and B. burgdorferi s.l. serology is to be reported as negative. Recommendation: In such cases, further investigations are usually not necessary. However, a comment should be added to the report that in early infection stages (EM, LNB), the confirmatory test may produce false negative findings [65,152]. Interpretation in the case of a borderline test Borderline values in the immunoblot are particularly difficult to interpret because of the absence of criteria for the detection of B. burgdorferi s.l. infection and the diagnosis of the disease. There is no available standard on how to define or interpret indeterminate results [125]. There is no reason to repeat the test with the same technique, because the reproducibility of modern automated immunoassays is high and there are only small random variations in measurement results [125]. In proficiency testing surveys, the reproducibility of immunoblot testing has been reported to be 82% in 239 participating laboratories [8]. In these cases, performance of a back-up test (e.g. verification of a whole cell antigen blot result by recombinant immunoblot) can be helpful. If screening and back-up tests provide divergent results in the presence of a borderline immunoblot result, this provides strong evidence for a false positive serodiagnosis [21,65]. In the case of a positive or borderline back-up test, the serological finding is consistent with LB during an early stage of infection. However, such results almost always speak against an existing, persistent infection, or late manifestation of LB [20,21,32,65,125,159]. Recommendation: Depending on the clinical signs and indications for testing (Table 1), only in the case of a suspected recent infection is it recommended that a borderline or negative serological test be repeated after 2 6 weeks. If the finding does not change, active LB is unlikely [20,21,32,65,125,159]. Interpretation in the case of a positive confirmatory immunoblot As stated earlier, many guidelines and the CDC in the USA recommend a two-tiered strategy that applies a screening test of high analytical sensitivity (e.g. ELISA) with a confirmatory test of high specificity (immunoblot) as the most reliable diagnostic procedure (Figure 3) [21,32,64,65,131]. In the case of a positive immunoblot, analysis of the banding pattern is an important part of the serological diagnosis. The data required for test interpretation [21,32] are the class of reactive antibodies (IgM and/or IgG), the intensity of the bands, and the number of bands (antigens) recognized, including their molecular weights. In suspected early or late stage LB, it is essential that the diagnostic findings also fit the clinical presentation and current case definitions (Table 1). Isolated detection of IgM antibodies against VlsE, p41, or OspC is an important marker of early LB. Although the finding of an isolated p41 band is not proof of LB because the antigen may cross-react to the flagellin proteins of other bacteria, it is compatible with the clinical diagnosis of early LB when corresponding clinical information is present [21,32,65]. By detecting antibodies against p41 or p41/i (the B. burgdorferi

19 18 B. LOHR ET AL. Table 6. Comparison of CSF findings in early and late LNB (modified from Djukic et al. [174] and Kaiser [173]). Comparison of CSF findings in early and late LNB [173] CSF parameter Early LNB [174] Early LNB, N ¼ 37 Late LNB, N ¼ 10 Cell count (/ml) (57.0; 369) a 218 (6 757) b 95 (23 312) b Total protein (g/l) (697; 1926) a n.n. n.n. Albumin quotient 17.2 (9.7; 28.4) a 19.6 (8 58.4) b 45 (8 140) b (10 3 ) IgG-synthesis-rated Pathologic at: n.n % 20 % (17) c 81 % 50 % (20) c 100 % IgM-synthesis-rate d Pathologic at: n.n % 54 % (32) c 84 % 9 % (13) c 40 % IgA-synthesis-rate d Pathologic at: n.n. 9% 7 % (17 ) c 19 % 39 % (28) c 80 % Lactate 2.0 (1.6; 2.6) a n.n. n.n. n.n.: not noted. a 50% percentile (25%; 75% percentiles). b Mean (range). c Mean (standard deviation). d According to Reiber et al. (2013) [172]. s.l.-specific internal fragment of flagellin) and against VlsE or OspC, the diagnosis of early LB is much more probable, especially if at least one of the bands is present strongly in the blot [21,32,65]. The clinical diagnosis of late stage disease is supported by laboratory testing only if several bands of intensive signal are recognizable across a wide range of antigens. The p83/ p100 and Osp17/p18 bands are of particular diagnostic relevance because of their specificity. In combination with a wide band pattern, they point to the late phase of the immune response [21,32,65]. Recommendation: For the diagnostic assessment of an immunoblot result as outlined above, detailed clinical information (Table 1) is essential. Importantly, any conclusions as to the need for treatment cannot be made based merely on a positive immunoblot or ELISA result because antibodies (including IgM) do not per se point to active disease, and they may persist for long time periods (months to even years) after resolution of the infection and even after treatment [21,32,65]. Therefore, the detection of antibodies does not automatically confirm a clinical infection. Confirmation is provided by merging the result of the serodiagnostic test with the clinical symptoms (Table 1), especially because a marker of disease activity (such as the Venereal Disease Research Laboratory (VDRL) test for syphilis) is currently not available for LB [21,32,65]. Correspondingly, any reinfections can be unambiguously diagnosed only by verifying significant changes of the serology by parallel testing with a previously collected sample from the patient. It must be emphasized again that, in almost all cases, the isolated positive detection of IgM antibodies in this context clearly speaks against manifestation of late LB [21,32,65]. Lyme neuroborreliosis LNB can manifest primarily or following EM as a disorder of the central nervous system. Overall about 10% of LB patients show LNB with typical clinical symptoms and syndromes (Table 1). Laboratory diagnosis, in principle, follows the usual diagnostic approach. Except for very early LNB and in cases with polyneuropathy of the peripheral nerves, conventional laboratory investigations of the CSF show signs of inflammation including lymphocytic pleocytosis, activated plasma cells, and a disturbance of the blood-brain barrier (i.e. elevated protein and albumin) [170]. Cell counts in the CSF can vary between 6/mL and 1100/mL, with a mean of 170/mL [170,171]. In addition, general intrathecal immunoglobulin production of IgM and of IgG is present in early LNB in % and in 60% of patients, respectively [170,172]. In late LNB, high general intrathecal IgG and also IgA production is frequently seen [170,173,174]. Lactate determination does not play a significant role in the diagnosis of LNB (Table 6) [170]. Depending on the duration of the disease and the antigen preparation used for diagnostic testing (which should contain VlsE or C6 peptide), specific intrathecal antibodies are detected in 60 90% of cases with LNB depending on the duration of the infection. An isolated intrathecal antibody response (IgM) without a specific antibody response from peripheral blood has been noted rarely, mainly in very early LNB of children. Antibody production in the central nervous system is detectable only by parallel quantitative testing of CSF and serum for B. burgdorferi s.l.-specific antibodies. For this purpose, the serum and CSF IgG concentration or albumin concentration is interrelated with the concentration of the pathogen-specific antibodies (IgG, IgM) determined in serum and CSF taken at the same time [20,125,170,175]. These data are used to determine the liquor serum index (LSI) according to the following formula [170]: LSI ¼ IgG or albumin concentration ðserumþ specific antibody concentration ½ELISA ðu=mlþš in CSF IgG or albumin concentration ðcsfþ specific antibody concentration ½ELISA ðu=mlþš in serum Unless other tests have been specifically evaluated and determined to have a different cutoff, LSI values >2 in the ELISA corroborate intrathecal antibody production due to LNB. LSI values of are considered to be borderline results [20,125,170,175]. Advanced CSF analysis should employ IT-based evaluation programs coupled to medical laboratory protein analysis for

20 CRITICAL REVIEWS IN CLINICAL LABORATORY SCIENCES 19 assessment of the CSF flow (function of the blood brain barrier) to be able to take into account the special analytical constellations of today s diagnostic algorithms [21,32]. This remains true especially in cases with local general intrathecal IgG, IgA, or IgM production when the calculation of the LSI for B. burgdorferi s.l.-specific local antibody production must be based on Q-Lim (the empiric cutoff for the IgG- (or IgA- and IgM-) fraction originating from peripheral blood in relation to albumin) to avoid a false negative LSI determination [21,32,170]. In this case, the following formula must be applied: LSI ¼ ½Specific Ab ðigg; IgM; IgAÞ in CSF ðu=mlþš= ½Specific Ab ðigg; IgM; IgAÞ in serum ðu=mlþš Q-Lim High diagnostic accuracy is also achieved by immunoblot analysis of concurrently collected CSF and serum samples that have been adjusted to equal concentrations of class-specific immunoglobulin (cross-match immunoblot). Additional bands in CSF or higher band intensity compared with the serum are evidence of immunoglobulin class-specific and antigen-specific autochthonous antibody production in CNS [21,152]. The diagnostic conclusiveness of positive findings (Table 5) must always be assessed in the context of other protein analysis and CSF serology data (presence of blood brain barrier disorders, presence of lymphocytic pleocytosis) (Table 6). Recommendation: The absence of inflammation and lack of antibody response suggests that the patient does not have LNB if the duration of disease is longer than 4 6 weeks [21]. However, in very early LNB, a systemic and/or intrathecal antibody response can be missing even though signs of inflammation are present, except in peripheral neuropathy. Most importantly, due to past infections and even after adequately treated LNB, specific autochthonous antibody formation in the CSF can persist for months or even years [170,175]. It must also be kept in mind that not all serological test systems achieve the same detection sensitivity for B. garinii, the primary causative agent of LNB. This especially applies to tests that do not include VlsE or C6 peptide and that use only B. afzelii or B. burgdorferi s.s. as an antigen source [170,175]. LNB can be excluded, however, by adequate testing if, in the absence of pleocytosis and the presence of normal CSF protein concentrations, neurological symptoms persist for more than 2 months [170,175]. CXCL13 as a marker for Lyme neuroborreliosis The chemokine (cytokine) CXCL13 may be a useful parameter in early diagnosis of LNB [176]. Among other effects, this chemotactic cytokine attracts B-lymphocytes to the central nervous system [175,177]. The presence of B-lymphocytes in CSF in the case of LB (as well as in neurosyphilis) is an already established phenomenon. Recent studies have suggested that the concentration of CXCL13 increases reliably in the CSF of patients with well-defined early LNB and can precede specific antibody formation in the CSF [175,177]. Some studies have reported that CXCL13 shows high sensitivity in early LNB when the B. burgdorferi s.l.-specific Antibody Synthesis Index (ASI) in CSF is still negative [178]. The diagnostic sensitivity and specificity of CXCL13 are currently specified as % and 63 96% by contemporary clinical evaluation studies [177,178]. Moreover, the CSF CXCL13 concentration in treated patients decreases relatively rapidly, and it is, therefore, suggested as a potential biomarker for treatment response [178]. Information on the diagnostic specificity and discriminatory power with respect to other infectious and inflammatory CNS disorders is still insufficient. It has been reported that elevated CXCL13 concentrations are also detectable in infections with closely related pathogens (e.g. T. pallidum), and in tuberculous meningitis and CNS-lymphoma [176, ]. Most importantly, there is no consensus on a standardized assay and the clinical cutoff value. Recommendation: The CXCL13 test may be a useful parameter, especially in the early diagnosis of LNB and for monitoring treatment success; however, it is not yet established as a routine diagnostic tool, and it needs standardization and further clinical evaluation [20,125,170,175]. Quality assurance Following the guidelines of many national medical associations (e.g. German Medical Association: Bundes arztekammer ), diagnostic laboratories must participate in infection-related serological round robin tests several times a year [182]. This also applies to serological antibody detection and to direct molecular-biological detection of borreliae. The molecular detection of LB agents is also an option offered as part of interlaboratory proficiency testing scheme on bacterial genome detection. The results of the EQA tests, which INSTAND e.v. (D usseldorf, Germany) has been carrying out for years, reveal extensive heterogeneity in the testing systems currently on the market (see above). The analytical pass rates for the conventional serological and molecular test systems, which have been collected from meta-analytical data, are relatively good for immunoassays and molecular tests. However, the

21 20 B. LOHR ET AL. Figure 6. Average number of participants and mean pass rates (%) with standard deviations (bars) for different assay systems as observed between 2006 and 2008 in the German Lyme borreliosis proficiency testing program (modified from M uller et al. [8]). interpretation of the results often proves problematic and can hamper medical treatment in daily clinical practice [8,107,152]. Thus, when LB is suspected, diagnosis of infection should be conducted in laboratories that meet standards in accordance with the diagnostic guidelines of the expert medical societies [65]. Physicians treating patients with LB should verify that these prerequisites are met in the laboratories charged with carrying out their diagnostic testing [65]. If questionable or implausible test results are produced, expert laboratories should be consulted. Figure 6 provides a summary of the pass rates for common test systems based on meta-analytical data from in Germany [8]. Recommendation: Attending physicians should be aware of whether their diagnostic laboratories comply with the respective diagnostic standards and qualifications, and the extent to which the diagnostic assays that are used conform to actual guidelines [21,65]. Diagnostic tests that are not recommended In addition to the traditional diagnostic methods listed above, which are used when LB is suspected, the literature describes a series of diagnostic techniques that, in part, have been inconclusively evaluated. This includes the immuno-histochemical detection of B. burgdorferi s.l. in biopsies and of antigens from blood and urine as well as functional tests that assess cellular immunity (lymphocyte transformation tests [LTT] and cytokine detection) [ ]. Currently there is a paucity of scientific investigations that support a diagnostic benefit of such methods. In addition, available LTT methods lack specificity and should not be used [184]. Immunohistochemical detection of borreliae from tissue is another method that is not recommended for use in the diagnosis of cutaneous manifestations of LB [65]. Similarly, the enzyme-linked immunospot assay (ELISPOT) [187] is not recommended. Detection of LB agents in engorged ticks by xenodiagnostic (see above) or molecular methods is similarly not recommended [52,54]. The detection of cystic forms, L-forms or spheroplasts [188], CD57 þ /CD3-lymphocyte subpopulation tests [189], the detection of circulating immunocomplexes, and the visual contrast sensitivity test (VCS) [190] are neither helpful nor recommended. Also pointof care testing [191] is not recommended in the diagnosis of LB. Recommendation: Most expert guidelines explicitly recommend against the diagnostic application of the tests listed above for the diagnosis of LB due to poor diagnostic sensitivity and specificity [2,3,20,64,65,69, 70,108,125,127,170,175]. Conclusion Over the last few decades, tremendous scientific progress has been achieved in improving the laboratory diagnosis for LB. However, problems with LB laboratory diagnosis persist because of its variable clinical