Lyme Neuroborreliosis: Known Knowns, Known Unknowns

John J. Halperin; Randi Eikeland; John A. Branda; Rick Dersch


Brain. 2022;145(8):2635-2647. 

In This Article

Diagnostic Testing

Since serum anti-Bbsl antibodies are usually detectable at initial LNB presentation, serology is the diagnostic test of choice.[9] To improve specificity, standard two-tiered testing (STTT)—screening samples using an ELISA or similar test, with reflex to IgM and IgG immunoblots—is recommended in the USA and many European countries.[9,46] First-tier ELISAs[47,48] use key immunodominant epitopes or antigens from relevant strains—in the USA primarily Bbss,[49] in Europe, multiple genospecies and regional strains.[50] When first-tier tests are positive or equivocal, IgM and IgG immunoblots are performed and interpreted according to specified criteria.[9,50] With less strain heterogeneity, US immunoblot criteria are more stringent (Table 1). Substantially greater strain heterogeneity makes it impossible to use uniform, pan-European assays and interpretive criteria (Table 1), with assays in each region incorporating different antigens from different strains to maximize reactivity to the most relevant species. Even within one region (Table 1) different criteria may be used to accommodate the range of locally prevalent strains.[51,52] Notably, while US and European assays and interpretive criteria perform comparably with US-acquired infections, not surprisingly US assays have lower sensitivity in European-acquired infections.[53,54]

Recently the US Center for Disease Control and Prevention endorsed alternative, modified two-tiered testing (MTTT), which still retains the first-tier ELISA but replaces immunoblots with a second (sometimes third) orthogonal ELISA (or similar) (Table 1).[55,56] MTTT is at least as sensitive as STTT in early LNB, without sacrificing specificity.[57,58] Comparable efforts to develop modified criteria with European assays appear promising.[59]

As with most serologic tests, those for Lyme borreliosis have two inherent limitations when used to support the diagnosis of an active infection. Since it takes time for the host immune response to produce measurable serum antibody, infected patients may initially be seronegative—a window which in Lyme borreliosis may extend to 4–6 weeks, occasionally overlapping with the onset of LNB. While obtaining acute and convalescent-phase serological tests, as in other infections, would help address this, this cannot inform initial treatment decisions. Because of this delayed antibody detectability, both European Guidelines[60] and older US studies[28] described a small proportion of patients with early LNB having negative serum ELISAs and/or two-tier testing. Newer serological techniques, such as the first tier ELISAs described above, using key immunodominant antigens, make such initially negative serum ELISAs quite rare[61]—although occasional positive ELISAs with negative second tier tests might still occur. In such uncommon instances with non-diagnostic ELISAs or two-tier tests, diagnosis requires a high clinical index of suspicion, including careful search for an EM rash, and potentially CSF examination.

The other limitation of serological testing is that serum antibodies typically remain elevated long after resolution of infection. Consequently, a positive serological test result can only be considered evidence of past or present infection, not proof of current infection. A corollary is that serological tests following treatment provide little information about treatment efficacy—positive results do not indicate continued infection.

More unconventional methods—urine antigen capture assays, the 'LM' blood microscopy method,[62] the 'improved' serum culture method[63] or non-serological indirect detection methods (e.g. lymphocyte transformation tests, quantitative CD57 lymphocyte assays)—should be avoided as either invalidated or lacking adequate clinical validation.[64–66]

Diagnosis of Lyme Neuroborreliosis

Assessment of CSF plays an important role in diagnosing CNS LNB and is required to meet European Federation of Neurologic Societies (EFNS) criteria for definite LNB[60]—with one exception. Many patients with acrodermatitis chronica atrophicans (ACA), a rare, late cutaneous manifestation seen almost exclusively in European Lyme borreliosis, have an associated polyneuropathy,[34,67] typically with normal CSF. Such individuals are considered to have definite LNB if other aetiologies of the neuropathy have been excluded and if the patient has positive two-tier testing—becoming positive at or within 6 weeks of presentation,[60] regardless of CSF findings (Table 2).

Such patients with distinct peripheral nerve disease notwithstanding, CSF exam is often of critical importance in patients with inflammatory CNS disorders, where it usually provides the only way to differentiate between LNB and other infectious or demyelinating processes. In patients in whom the differential diagnosis includes potentially more threatening meningeal or parenchymal CNS disorders, CSF examination is essential to exclude these other disorders. In individuals treated for LNB, knowledge of pretreatment CSF abnormalities can be helpful if a follow-up CSF examination is needed because treatment response appears incomplete. Crucially, when CSF is obtained the correct tests must be performed.

Direct organism detection in CSF, using nucleic acid detection or culture, has quite low diagnostic sensitivity (<20%),[9] likely due to very low spirochete numbers in CSF. With such poor negative predictive value, this is not recommended for routine use.9 However, other CSF findings are typically informative, particularly in CNS LNB. As in all CNS bacterial infections, CNS LNB almost invariably elicits inflammatory CSF changes. Even if some aspects of LNB were to have an autoimmune aetiology (although no compelling evidence supports this), autoimmune encephalitis is also usually associated with inflammatory CSF. Consequently, individual outliers notwithstanding, if a patient's CNS disorder is to be attributed to LNB, CSF should provide evidence of an inflammatory response.

CNS LNB typically elicits a mononuclear cell CSF pleocytosis, with a modest elevation of protein and normal glucose. Particularly in European patients, overall CSF immunoglobulin synthesis is often elevated, including CSF-specific oligoclonal bands. Intrathecal synthesis of Bbsl-specific antibody (ITAb) is often present. Assessment of intrathecal synthesis of pathogen-specific antibodies (Figure 2 and Table 3 and Table 4)—a technique also used in other CNS infections[68]—is predicated on the assumption that the presence of a pathogen in the CNS will lead to selective intrathecal production of antibodies targeting it (Figure 2C). If only an unpaired CSF sample is available (without matched serum), the biggest challenge in interpreting the result lies in identifying a valid 'normal' comparator. Collection of a large number of 'normal' CSF samples is impractical—and, as discussed below, problematic for other reasons. When laboratories are asked to assess isolated CSF samples, they typically can only compare results to normal serum values. Such determinations rely on the observation that, as demonstrated in Figure 2A, a small amount of serum immunoglobulin normally enters CSF (up to 4–5 mg% IgG in CSF, 500–1500 mg% in serum) with CSF, on average, containing about 1/250th of serum IgG. Laboratories therefore typically adopt a standard CSF dilution to match this 'normal' expectation and compare the resulting CSF antibody measurement to serum normal values.

Figure 2.

Intrathecal antibody concentration. Cartoon comparing CSF to serum immunoglobulin concentrations. (Serum to the left of the BBB, CSF to the right). In the normal state (A), a small amount of serum immunoglobulin enters CSF. CSF antibody is increased (B) with increased BBB/BCB permeability, such as in stroke, demyelinating disease or (C) with increased synthesis of targeted antibody in response to a specific pathogen within the CNS. When testing CSF in isolation, without comparison to that patient's serum, laboratories typically use a standard CSF dilution (e.g. 1:1) establishing an IgG concentration approximating that in diluted serum (typically diluted 1:500). Values are then compared to serum normal controls. (A) In seropositive patients with normal BBB/BCB permeability, this will result in falsely 'positive' results, reflecting passive entry of excess serum anti-Bbsl antibodies. (B) If, for any reason, BBB/BCB permeability increases, the total amount of CSF IgG will be increased. Unless a correction is made for this, measurements of any specific antibody will similarly be increased, again giving false positive results. (C) CNS infection with Bbsl leads to in-migration of B cells that synthesize Bb-specific antibody. After appropriately accounting for non-specific entry of all IgG into the CNS, measurement allows demonstration of the ITAb.

This approach has two important limitations—as would comparison to accumulated 'normal' CSF. Since antibodies of a given class and subclass should enter CSF equally, regardless of their antigenic target, seropositive patients, in whom elevated peripheral antibodies are reflected in CSF, will presumably all have 'positive' CSF results—reflecting passive antibody entry, not local synthesis (Figure 2A). On the other hand, in seronegative patients with increased BBB/BCB permeability (Figure 2B), there will be disproportionate entry of all peripheral IgG. In this setting, measurement of pathogen-specific CSF antibody, without correction for the overall increase in immunoglobulin, will reflect the excess of all antibodies, with the increase in non-specific background reactivity causing apparent specific antibody excess—even when no Bbsl-specific antibody at all is present. To avoid such errors, comparing the 'proportion' of CSF and serum antibodies specific for Bbsl is essential—with demonstration that the proportion of specific antibodies is greater in CSF providing compelling evidence that that pathogen has been present in the CNS.

Several methods are used for this (Table 4), each with strengths and weaknesses. Widely used in Europe[69] and in some major US labs[70] is the Reiber formula. For this, CSF and serum albumin and immunoglobulin concentrations are quantitated, pathogen-specific antibody is measured in both at standard dilutions, then the albumin and immunoglobulin concentrations are used to mathematically correct for multiple variables that affect their CSF concentrations, producing a calculated specific antibody index (AI). Perhaps the most technically elegant approach, and requiring the least CSF, is a capture assay[71] which directly measures the proportion of total antibody in CSF and serum that is specific for the target antigens. A third approach, like the Reiber method, starts by measuring CSF and serum immunoglobulin concentrations, but then dilutes both fluids so post-dilution immunoglobulin concentrations are identical, and then performs pathogen-specific ELISAs on both[72]—an approach developed to circumvent the non-linear relationship between antibody concentration and the measured pathogen-specific ELISA result. Although one very small study suggests the capture method may be more sensitive,[73] no large systematic study has compared methods—and the capture assay is no longer widely commercially available. Moreover, procedures have not been standardized; none is US FDA cleared.[70]

Regardless of technique, ITAb measurement provides a useful diagnostic tool—although precise estimates of its sensitivity and specificity are limited by the absence of alternative reference tests for comparison. However, in any patient with elevated overall intra-CNS immunoglobulin synthesis (oligoclonal bands, increased IgG synthesis rate) if this increase is due to LNB there should be evidence that much of that overall immunoglobulin directly targets Bbsl.

Notably, presumably because of lower background CSF IgG concentration, ITAb can sometimes be detectable before a patient is seropositive, providing a potential diagnostic tool in very early LNB. However, like peripheral blood serologies, ITAb can remain elevated years after successful treatment, similarly limiting its utility in differentiating active from past infection.[74] However, if CNS LNB is active other CNS inflammatory changes should be evident (Table 3).

Finally, immunoblots can be performed on CSF—but are not recommended.[9,70] As with measurements of ITAb, these would need to be performed on CSF and serum concurrently, with appropriate adjustment for overall antibody concentration in both. Since interpretation of individual bands is typically not quantitative, arithmetically correcting for differences in total antibody concentration is not possible. If a standard compensatory CSF dilution is used, this will be susceptible to the same limitations as in ITAb assessment. If total antibody concentrations are measured and dilutions adjusted accordingly (rarely done in clinical laboratories) the required multiple non-standardized steps will potentially introduce technical errors in a procedure that is quantitatively imprecise to begin with.

Measurement of CSF CXCL13—although not yet widely available—may also prove informative. Assay techniques and normal value ranges have not yet been standardized, limiting clinical utility at this point. However, this cytokine is often elevated in CNS inflammatory disorders,[75,76] particularly so in CNS spirochaetal infections.[77,78] While it lacks specificity, it is almost always quite elevated early in CNS-involving LNB, providing excellent negative predictive value. Moreover, it decreases quickly with successful antimicrobial therapy, making it one of the few objective markers of treatment efficacy (although its concentration parallels CSF leucocyte count and IgG synthesis[76]).