Scientific basis

Diagnostic testing. Culture of B burgdorferi from clinical material is difficult. The organism requires special medium that is not generally available in clinical laboratories. It is slow growing, so cultures must be maintained for weeks before they can be considered negative. The number of microorganisms present in readily obtainable tissue or fluid samples is small (except for cutaneous lesions, but these are clinically virtually pathognomonic, rendering culture unnecessary), limiting sensitivity because any given sample may or may not contain spirochetes.

Other technologies that might be expected to improve sensitivity have been disappointing (65). Results with nucleic acid detection-based techniques depend on the primer selected, are prone to false positives and negatives (if not carefully performed) and vary substantially among laboratories (51). Like culture, diagnostic sensitivity is particularly low in CSF, presumably because very few bacteria are present (68). Work with “pre-enrichment”—inoculating CSF directly into culture medium at the time of lumbar puncture, then incubating for 2 to 6 weeks before testing—has not materially improved diagnostic sensitivity (45). Antigen detection techniques (particularly the Lyme urine antigen test) have been popular in some circles, but reproducibility has been highly problematic (44). Assays of cell-mediated immunity, such as the EliSpot, are similarly popular but problematic (44). Detection of immune complexes containing Borrelia antigens has been reported by a few laboratories, but the findings have not been widely reproduced. As a result, testing for the presence of antibody to B burgdorferi, primarily by enzyme linked immunosorbent assay (ELISA), remains the diagnostic method of choice.

Two-tier testing was introduced to address problems with specificity that are inherent in serologic diagnosis. When (and only when) an initial quantitative assay (ELISA or IFA) demonstrates elevated serum immunoreactivity to the causative organism, a second test is performed to determine specificity. For 25 years, the second test has been a Western blot, which provides a qualitative overview of the specific proteins to which the patient's antibodies bind. Results are interpreted based on the statistical analysis of findings in large numbers of patients and controls, identifying combinations of bands with collectively high positive and negative predictive values, an approach that greatly increased specificity of serodiagnosis in North American patients. Unfortunately, the substantially greater strain variability among European Borrelia species has, to date, precluded the development of comparable criteria for European patients (03). Measurement of antibodies to the C6 antigen appears better at diagnosing infections with both U.S. and European strains, but it may be slightly less sensitive in early disease, such as facial palsy where the IgM response predominates (57). Studies have validated testing using two orthogonal recombinant antigen-based ELISAs (such as a C6 ELISA followed by one to VlsE) (08), allowing the use of two standardized, automated tests instead of an ELISA followed by a qualitative Western blot.

Finally, specific diagnosis of central nervous system infection has been particularly controversial. Because the sensitivity of culture for B burgdorferi in the cerebrospinal fluid of patients with Lyme meningitis has generally been on the order of 10%, this method has not been helpful. Nucleic acid detection results have varied widely, but clinical sensitivity of this also seems poor (68). Therefore, analogous to the diagnosis of systemic infection, the most useful diagnostic tool has been to measure the intrathecal production of anti-B burgdorferi antibodies to demonstrate that there is proportionately more specific antibody in the CSF than in serum, which is reflected in an antibody index. Here, too, estimates of sensitivity vary substantially. The published European guidelines for the diagnosis of neuroborreliosis require the presence of both a CSF pleocytosis and intrathecal antibody production to diagnose definite neuroborreliosis (53). Some work suggests false negatives occur primarily in very early disease just as in peripheral blood serologic testing (07). Other negatives that have been interpreted as “false” occur in patients whose neuroborreliosis is limited to the peripheral nervous system, in whom abnormal CSF is not necessarily expected. Interestingly, although CSF IgM responses might be expected to be highly diagnostic (because IgM does not cross the blood brain barrier), this has not proved to be the case (69). Notably, a common strategy in the U.S. has been to consider lumbar puncture only in patients with a clinically consistent picture and a positive or borderline serology. European studies have emphasized the importance of always performing lumbar punctures in patients suspected of having Lyme neuroborreliosis based on the occurrence of a small number of seronegative patients with intrathecal antibody production, thereby making such a strategy potentially problematic. A 2023 study may help clarify the situation (83). In 30 seronegative patients with intrathecal antibody production, 28 had a positive or borderline ELISA but were considered seronegative because of a negative confirmatory test. Thus, just as with “false negative” antibody index in early Lyme neuroborreliosis, this mismatch may relate to the incompletely developed peripheral serologic response in early infection, resulting in measurable total antibody but not meeting criteria for the second-tier confirmatory test. Whether this will be less likely to occur with modified 2-tier testing remains to be determined.

Because apparent intrathecal antibody production can persist for years after the infection has been successfully treated, and because the CSF pleocytosis may be slow to resolve, much effort has focused on possible alternative markers more closely linked to active infection, particularly the concentration in CSF of CXCL13. The CSF level of this B cell attracting cytokine does appear to increase very early in infection and decrease rapidly with successful treatment (64; 59). Its major limitation is that it is elevated in many CNS inflammatory states, leading to difficulty in defining both a useful diagnostic cutoff value and its overall specificity (58; 15). Work measuring plasma neurofilament light chain concentration suggests that, as in other CNS insults, levels may be elevated in acute Lyme neuroborreliosis, normalizing rapidly with successful treatment (52). Notably, no relationship was observed between this plasma marker and persistent post-treatment symptoms.

Clinical phenomena. There has been particular confusion and controversy surrounding the clinical phenomenology of nervous system Lyme disease. Some clinicians (primarily non-neurologists) and many patients seem to forget that there are three prerequisites for the diagnosis of nervous system Lyme disease: (1) the existence of nervous system disease, (2) the presence of Lyme disease, and (3) a causal relation between the two.

Most agree that Lyme disease causes lymphocytic meningitis, radiculoneuritis (painful inflammation of nerve roots, typically with objectively demonstrable sensory, motor, and reflex changes), and cranial neuropathies, particularly the facial nerve. In European patients, the most common manifestation in adults is painful radiculitis; in children it is cranial neuropathies, almost always facial palsy (74). Less agreement exists on the characteristics of other types of neuropathy and other CNS changes, particularly encephalomyelitis and encephalopathy (54).

A variety of forms of peripheral nervous system involvement has been described. A mononeuritis multiplex often occurs both in the affected limb and more diffusely in patients with acrodermatitis chronica atrophicans (34), a chronic cutaneous manifestation of Borrelia infection seen in Europe but rarely, if ever, in North America. North American patients also develop a mononeuropathy multiplex, with neurophysiologic and neuropathologic characteristics similar to those seen in European patients (31). Clinical presentation can be as a plexopathy, a severe patchy or diffuse polyneuropathy (similar clinically to that first described by Garin and Bujadoux), or a mild confluent mononeuropathy multiplex, presenting clinically as a subacute stocking-glove neuropathy. Interestingly, the identical peripheral nervous system changes have been consistently demonstrated in rhesus macaque monkeys experimentally infected with B burgdorferi (19).

Central nervous system parenchymal involvement, though very rare, similarly can take a variety of forms. Best described is the focal spinal cord inflammation sometimes occurring at the same level as nerve root inflammation in the syndrome of Garin-Bujadoux-Bannwarth (37). Rarely, patients may develop primarily white matter inflammation, a leukoencephalitis, with spasticity and other "white matter" signs, abnormal brain MRIs, and inflammatory changes in CSF (39). The spirochete may have a specific tropism for oligodendroglia, leading to predominant involvement of the white matter (63). Patients with this prolonged CNS infection often have oligoclonal bands and increased total IgG synthesis in the CSF, raising the possibility of misdiagnosis as multiple sclerosis. However, in these patients, because the chronic immune stimulation causing these abnormalities is directed against a specific organism, virtually all should have demonstrable intrathecal production of Borrelia-specific antibody. Treatment should arrest disease progression in most of these patients; if the infection is eradicated, improvement may occur over time as with any self-limited encephalitis.

Most controversial has been the entity referred to as Lyme encephalopathy (31). This disorder was originally described in individuals with unequivocal and long-standing systemic Lyme disease, such as Lyme arthritis or other chronic manifestations. Described before there was widespread lay and medical recognition of Lyme disease, and consequently, at a time when a significant number of patients were untreated for an extended period of time, this consisted of difficulty with cognitive functioning clearly demonstrable both on mini-mental status testing and on formal neuropsychologic testing. In most patients, brain MRI scans and CSF exams failed to suggest CNS infection. It is likely that this disorder represents the same type of cognitive difficulty seen in patients with most other systemic (non-neurologic) inflammatory illnesses, perhaps mediated by circulating cytokines produced outside the CNS in response to infection, which then enter the CNS where they have neuroactive effects (14). Because the nervous system is neither infected nor damaged in these individuals, neurologic recovery following treatment is typically excellent (55).

A small proportion of these individuals have indeed had evidence of CNS infection, with abnormal brain MRIs and CSF. Like other patients with Lyme encephalitis, this disorder does respond to antimicrobial therapy (49). Because the parenchymal damage tends to be rather limited, recovery is generally excellent.

Emphasis on these encephalopathic symptoms in active Lyme disease led to the notion that this state could occur either as the sole manifestation of Lyme disease, absent other evidence of inflammation, or after otherwise successful treatment of the infection–the latter becoming known as “post-treatment Lyme disease symptoms/syndrome” (PTLDS). PTLDS is thought to consist of persistent difficulties with cognition, memory, and fatigue despite appropriate treatment for Lyme disease (13). Although one study suggested a slight excess in this symptom complex among patients who had previously been treated for Lyme borreliosis (3.9% higher than in population controls) (72), most studies describe the prevalence of 10% to 20% in individuals 1 year after treatment as no different than among contemporaneously-followed healthy controls (67; 66; 11; 70; 82; 78; 01). Several non-Lyme disease-specific factors have been identified as predictors of persistent symptoms: presence of multiple comorbidities (76); patient initial negative expectation of a treatment response and assumption that treatment was not with active drug (73); and the presence initially of either paresthesia, elevated stress score, or a combination of elevated Beck Depression Inventory score plus number of pain sites (79). Importantly, multiple studies have found that few if any patients with persistent symptoms have had prior neuroborreliosis (13; 24; 02). Even more importantly, multiple studies have found that only a minority of patients seen in referral centers for these symptoms have any objective evidence of prior Lyme borreliosis. In fact, many have had other diagnoses for which appropriate treatment was delayed by this misdiagnosis (24; 40). Notably, additional antimicrobial therapy clearly does not provide meaningful benefit to these patients and carries significant risk (43; 20; 05; 22; 06; 16).

Given the 2% prevalence of these nonspecific symptoms in the general population, post-treatment Lyme disease symptoms/syndrome may well represent the chance co-occurrence of unrelated disorders, with symptoms attributed to Lyme disease because of anchoring bias (50; 31). Finally, although some work has suggested possible persistence of small numbers of B burgdorferi in experimentally infected monkeys (18), there is no evidence that this elicits any type of host response, causes any symptoms, requires treatment, or in any way contributes to continued disease.

Treatment. Numerous studies have demonstrated that both oral and parenteral antimicrobial regimens are highly effective in treating patients with clear-cut Lyme disease (44). Several analyses have found that most commonly used antimicrobial regimens are equally efficacious (70; 81; 04). The usual approach is to treat all but the most serious infections with oral regimens first, followed by parenteral regimens in the small number of individuals in whom this is not curative. Oral regimens have been shown to be effective in European patients with Lyme meningitis and cranial neuritis (42). Published European neuroborreliosis treatment guidelines explicitly recommend oral treatment for Lyme meningitis or peripheral nervous system involvement, with parenteral treatment solely for individuals with parenchymal CNS involvement (53; 36; 62). Corresponding studies have not been performed in patients infected in the United States. However, given that the antimicrobial sensitivities of U.S. and European B burgdorferi strains are comparable, similar treatment recommendations are reasonable in U.S. patients. Moreover, one European study has demonstrated that oral doxycycline can effectively treat many patients with parenchymal CNS infection (09). Controlled, randomized trials show no clear benefit of extending parenteral treatment beyond 2 weeks (56); however, many centers dealing with large numbers of patients have seen enough relapses after 2 weeks of treatment that it has become customary to treat for 3 to 4 weeks.

Discussion

Diagnostic testing. Although ELISAs are widely used in the serodiagnosis of many infections, many of the conventions usually employed in other disorders have, unfortunately, been ignored in the diagnosis of Lyme disease. In most other diseases, changes in antibody titer (acute vs. convalescent) are used to indicate a change in the immune response and, hence, to infer an immune challenge (ie, a current or recent infection). In Lyme disease, a single serologic test result is typically used to infer active infection. In most other diseases, the presence of detectable antibody is taken as evidence of present or past exposure; in Lyme disease, it is often used to suggest current, active infection in need of treatment. In most infections, it is readily accepted that it takes time for the immune system to develop a detectable antibody response; the absence of detectable antibody in very early Lyme disease sometimes leads to delayed diagnosis and treatment. Finally, although in most other disorders it is widely accepted that no diagnostic technology has 100% sensitivity and specificity, in Lyme disease this has led to popular misconceptions that laboratory testing is completely unreliable and without any positive or negative predictive value. This unfortunate list of misconceptions and misuse of testing has resulted in tremendous uncertainty on the part of physicians and patients on the appropriate diagnostic and therapeutic strategy.

Similarly, although the use of Western blots substantially improved the specificity of serodiagnosis, it must be recognized that as with any technique that improves specificity, there is some corresponding loss of sensitivity. In the original work that led to these criteria, the sensitivity of the IgM criteria was only 32% in patients with early disease, whereas that of the IgG criteria was 83% in individuals with disease of longer standing, a value that has now been improved to over 95% but may still be somewhat limiting in early Lyme neuroborreliosis (80; 83). Even so, common sense would still suggest that in a strongly seropositive patient with a clinical disorder likely to be caused by Lyme disease, a negative Western blot would not absolutely exclude the diagnosis (85). Perhaps more troubling has been the widespread use of Western blots without ELISAs. Because the criteria for Western blot interpretation were developed almost exclusively in patients with positive ELISAs, interpretation in an entirely different group (those with total borrelia binding antibody no different than controls) is problematic and must be done extremely cautiously. Hopefully, the alternative strategy of sequentially using two independent ELISAs will at least remove difficulties with Western blot misinterpretation (42).

In the diagnosis of CNS infection, the role of measurement of intrathecal production of specific antibody has been controversial, a controversy made even more problematic by the observation that apparent excess of intrathecal antibody can persist for many years after successful treatment. Most European experts in the field originally considered this measure an absolute requirement to diagnose CNS infection and only rarely make the diagnosis of nervous system Lyme disease in its absence (53). Studies in early North American Lyme neuroborreliosis also suggest that at least 90% of such patients have specific intrathecal antibody production (31). Some studies of patients with more long-standing infection and CNS symptomatology suggest that only 50% of patients will meet this criterion, although most others have a CSF pleocytosis or increased protein (48). Still other practitioners, viewing these differences, consider CSF examinations to be completely worthless and misleading. It seems clear that the demonstration of intrathecal antibody, particularly if accompanied by other CSF abnormalities, is a sufficient basis for a diagnosis of CNS Lyme disease. Whether it is necessary remains to be established.

Clinical phenomena. Although there is little debate about the classic triad of neurologic disorders (which tend to occur early in infection and are unambiguous in their clinical appearance), considerable controversy surrounds the more indolent and typically later-occurring phenomena. Unfortunately, the early descriptions of the milder, late polyneuropathy (30) emphasized the subjective symptoms more than the equally prominent objective abnormalities on exam and on neurophysiologic testing. Subsequently, some non-neurologists have taken to diagnosing Lyme neuropathy in patients with fluctuating paresthesias, but with no objective evidence of a peripheral neuropathy by any criterion; this is a questionable practice at best. More problematic, these purely subjective symptoms are now considered typical of post Lyme disease syndrome (75; 76).

Similarly, several issues concerning CNS disease remain confusing. Although intrathecal antibody measurement should differentiate patients with active Lyme encephalomyelitis from those with multiple sclerosis, some clinical situations may not be straightforward. In some patients who appear to have typical multiple sclerosis and also have evidence of Lyme disease, an indirect relationship might exist. CNS infection with B burgdorferi (either meningitis or parenchymal involvement) can stimulate local production of gamma-interferon (17) and tumor necrosis factor-alpha. Because the former is known to trigger multiple sclerosis attacks, it should not be surprising that in some patients, multiple sclerosis attacks might be triggered by this infection. In patients in whom this constitutes the first episode of multiple sclerosis, it might be concluded that their relapsing remitting illness is attributable to this infection. However, it seems more likely that Lyme disease is merely one of the many potential triggers of an acute exacerbation, and not the "cause" of relapsing remitting multiple sclerosis or of a relapsing inflammatory CNS disorder.

Similarly, some patients with apparent amyotrophic lateral sclerosis have had serologic evidence of B burgdorferi infection (28; 77). In some, this probably is Lyme neuroborreliosis presenting as a combination of a Lyme-related myelopathy and a predominantly motor polyradiculopathy. In others, the reason for the association is unknown, but may be a chance co-occurrence (61). Importantly, a large study of all Danish patients found to have intrathecal antibody production to B burgdorferi SL over 3 decades found no excess incidence of dementia, Alzheimer disease, Parkinson disease, motor neuron disease, epilepsy, or Guillain-Barre syndrome (25).

Perhaps most confusing has been the concept of Lyme encephalopathy. Initially described in patients with objectively demonstrable cognitive difficulties, this construct is now sometimes used in patients with either psychiatric diagnoses or purely subjective disorders but who are neurologically normal. Studies spanning decades have shown that these patients rarely have CNS infection or any evidence of neurologic damage of any sort (29; 55; 14). As for psychiatric disorders, the few systematic studies addressing this subject have indicated that psychiatric disease is no more common in patients with Lyme disease than in those with other chronic illnesses (84; 71), nor is any specific psychiatric disorder associated with this infection (26). Although numerous anecdotes describe individuals with psychiatric disease with acute Lyme disease (including some in whom both disorders improved concurrently), there is little to prove that a consistent or causal relationship exists. On the other hand, some, but by no means all, patients who have been diagnosed as having “chronic Lyme disease” without objective evidence of this infection have been shown to have significant psychiatric comorbidities – although these may more often be Axis II than Axis I disorders (33; 46; 41).

As it has become more difficult to support the argument that some of these syndromes are attributable to Lyme disease, the possibility of coinfections has received increasing attention. The same ticks that transmit B burgdorferi can also carry babesia, ehrlichia, anaplasma, and tick-borne encephalitis complex viruses. To date, however, the evidence in support of these as coinfections causing chronic symptoms has remained limited (23), and there is evidence that patients with both Lyme disease and anaplasma may be less ill than those with either infection alone (35). The possibility of co-infections with the agent of cat-scratch fever has been suggested; this too, though, seems improbable (32). Interestingly, in the current ILADS Guideline recommendations, early treatment with amoxicillin or cefuroxime is recommended to prevent the occurrence of late sequelae of Lyme disease, even though these agents have no efficacy against these co-infections (10).

Finally, several groups have attempted to obtain corroboration of this encephalopathy using brain SPECT scanning. Unfortunately, conventional, qualitative SPECT is highly susceptible to numerous artifacts. Although one highly sophisticated quantitative analysis in a group of patients with other compelling evidence of neuroborreliosis demonstrated significant abnormalities (47), most other work in this field is at best difficult to interpret. Interestingly, in contrast to the regional hypometabolism described in these SPECT studies, at least two PET studies in well characterized patients have demonstrated hypermetabolism in involved areas (38; 60) – more fitting to the local inflammatory state known to be present. Unfortunately, the diagnosis of Lyme encephalopathy is now being made in many individuals in whom the basis for the diagnosis of Lyme disease is tenuous, albeit more compelling than the evidence of an encephalopathy.

Treatment. The observation that some patients who have been diagnosed as having Lyme disease have symptoms after receiving recommended treatment regimens has led to the use of ever more prolonged and varied courses of treatment. Unfortunately, there is little biological rationale for treatment for longer periods or for many of the unconventional regimens being employed (44). Some clinicians have tried to use resolution of the serologic results to gauge treatment response. Because the "purpose" of the antibody response in any infection is, in part, to persist and protect in the event of future exposures, this strategy is illogical. In other patients in whom laboratory support for the diagnosis of Lyme disease is initially tenuous and who suffer from incapacitating subjective symptoms but have no objective evidence of neurologic (or other medical) disease, response to conventional courses of treatment is often disappointing (41). This occurrence can be attributed to one of two possibilities: either the problem was not caused by this infection in the first place, or the treatment was ineffective. Unfortunately, by drawing the second conclusion, many patients are being exposed to ever longer and more toxic regimens, with no clear goal or endpoint in mind, but with considerable potential risk (21).

Resolution

Diagnostic testing. Even though the preponderance of scientific data suggests there should be little remaining controversy, some continue to advocate strongly that important issues remain unresolved. That notwithstanding, several concepts are helpful when evaluating a patient for possible CNS Lyme disease. Because the principal issue is whether or not the CNS is infected, usual common sense principles should apply. CNS infections almost invariably elicit a local inflammatory response, manifest as a reactive CSF pleocytosis, increased CSF protein, or both. To have a CNS infection with completely normal CSF seems implausible. Moreover, as in the vast majority of other chronic CNS infections in which this has been evaluated, there is local stimulation of the immune system with demonstrable intrathecal production of organism-specific antibodies. Hence, although not necessarily addressing whether or not a patient has infection outside the CNS, it is probably fair to say that completely normal CSF makes CNS infection with B burgdorferi highly unlikely.

Similarly, common sense must be applied to serologic diagnosis. This requires recognition that no test has 100% accuracy, and all testing must be interpreted in its clinical context. If applied appropriately, 2-tier testing provides extremely useful information.

As for the clinical spectrum of this disease, as in any disorder, there will be a continuing evolution in our concept of the full spectrum of clinical manifestations, a fact made surprisingly evident with the identification of genetic markers in other disorders originally described syndromically. Although our concepts of the clinical spectrum of nervous system Lyme disease will undoubtedly continue to evolve, the guiding principle should remain that there must be objective evidence both of nervous system disease and of Lyme disease, and of a plausible link between the two.

Treatment. Unfortunately, this issue often leads to the most dogmatic and heated debates. Despite this, the microbiological and clinical data are clear. B burgdorferi is highly susceptible to the commonly recommended regimens, which typically result in cure rates of 90% to 95%, comparable to many other infections. Exposing patients to the risks of extremely prolonged regimens or of other more toxic agents seems ill advised.

Table 1. Treatment Regimens