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

Pathophysiology of Lyme Neuroborreliosis

Two factors underlie the importance of differentiating between LNB and non-neurological Lyme borreliosis. First, nervous system infections are generally more difficult to treat, and second, progressive, difficult-to-treat brain disorders are among the most feared of all illnesses.[21] Since extra-neurological infections may alter consciousness, cognition or other neurobehavioural functions, differentiating these from LNB is essential. Two mechanisms can mediate neurobehavioral effects of extra-neurological infections—nervous system entry either of bacterial exotoxins, which Bbsl appears to lack, or of peripherally produced cytokines. By definition, neither constitutes nervous system infection, so concerns about treatment-responsiveness are not relevant. This makes consideration of how pathogens invade the nervous system pivotal.

The nervous system is well protected by the blood–nerve (BNB), blood–CSF (BCB) and blood–brain barriers (BBB). The comparative simplicity of the peripheral nervous system (PNS) provides a useful paradigm—particularly as it is a frequent target in LNB—in both patients and experimentally infected rhesus macaques,[22] the only animal model of LNB. Peripheral nerve disease (non-LNB) generally falls into just one of three categories: demyelinating, length dependent axonal (typically reflecting neuronal failure), or multifocal with patchy axon damage. The last, mononeuropathy multiplex, generally attributable to vasculopathic processes, is virtually the only pattern seen in patients and experimental animals with PNS LNB.

Two potential mechanisms have been proposed for Bbsl PNS involvement—haematogenous spread[23] versus tracking along nerves from the site of inoculation.[24] In European Garin-Bujadoux-Bannwarth syndrome (GBBS) radicular symptoms often occur in the same dermatome as EM, suggesting spirochete migration along nerves.[25] Evidence against this model includes that pain often is not limited to a single dermatome[24] and the frequent involvement of anatomically dispersed nerves—e.g. patients with radiculopathy and cranial neuropathy, bilateral facial nerve palsies, or neurophysiological demonstration of multiple unrelated sites.[26] In one study no patients with EM and facial nerve palsy had EM on the face;[27] in about 30% radicular symptoms did not match the EM site. Notably, most experimentally tick-bite-infected rhesus macaques develop inflammatory changes in multiple anatomically unrelated nerves.[22]

If infection spreads haematogenously, pathogens must access nerves either as they cross the already haematogenously infected subarachnoid space, or within peripheral nerves themselves. Notably, although meningitis often co-occurs with PNS involvement, CSF has been normal in a third of patients with early LNB radiculopathy[15] (personal communication, F. Strle). CSF was similarly acellular in half[28] with early LNB facial nerve palsy (FNP). Moreover, in LNB FNP differing involvement of dysgeusia, hyperacusis or muscle function implicates varying sites peripheral to the subarachnoid space, even when CSF is abnormal.[28] Similarly, neuroimaging can demonstrate prominent peripheral involvement[29–31] (Figure 1). Finally, neurophysiological testing, and more recently ultrasound imaging of involved peripheral nerve,[32] often indicates more disseminated peripheral nerve damage than is apparent clinically—typically indicative of damage peripheral to the nerve roots,[26,33,34] in aggregate, suggesting a mononeuropathy multiplex. If LNB meningitis, radiculopathy and cranial neuropathy can all occur independently, it appears meningeal involvement, while often co-occurring with PNS LNB, is not pathophysiologically essential.

Figure 1.

MRI of involved facial nerve in LNB. 3D T1-weighted and fat saturated post-contrast image (reformatted in oblique sagittal plane parallel to the course of the nerve with 10 mm maximum intensity projection) demonstrating pathological enhancement of the right facial nerve in a patient with LNB-associated right facial palsy. Superiorly is the typical tuft of enhancement at the distal intracanalicular segment, typically evident in standard axial views in customary diagnostic imaging in early disease. The abnormal, intense enhancement of the nerve is evident as it continues first horizontally, then vertically, in the geniculate ganglion, tympanic and mastoid segments. In this case, but not in all LNB-associated facial palsies, enhancement extends extracranially into the parotid segment. Image courtesy of Dr Elisabeth S. Lindland, BorrSci study group. Sorlandet Hospital, Norway.

Direct nerve infection can be best understood in terms of peripheral nerve anatomy, particularly its neurovascular unit. Nerves consist of fascicles, each with myriad axons enveloped by endoneurium, surrounded by perineurium, with fascicles bound together by a tough connective tissue sheath, the epineurium. Nerve blood supply, the vasa nervorum, originates in the epineurium, where it has fenestrated endothelium, then penetrates the perineurium, where endothelial cells become joined by tight junctions, with surrounding pericytes. Perineurial cells similarly are joined by tight junctions, together with the endothelium forming the difficult to penetrate BNB—a barrier with relative weaknesses at nerve terminals and dorsal root ganglia (DRG). BNB notwithstanding, most experimentally infected rhesus macaques develop a mild mononeuropathy multiplex, with nerve biopsies demonstrating the expected patchy axon loss and perivascular inflammatory infiltrates—at a minimum involving epineurial vessels. Nerve biopsies in PNS LNB,[31,35–37] while infrequently obtained, show similar changes—with epineurial perivascular inflammation; less frequently plasma cells and lymphocytic infiltrates surround endoneurial vessels.[22,33,37] Importantly, pathologic changes in sural (sensory) nerves would not be expected if damage were limited to the subarachnoid space, proximal to DRG, so, like neurophysiological findings, biopsies indicate LNB causes a mononeuropathy multiplex. Notably in neither patients nor experimental animals is there evidence of blood vessel wall necrosis—required to diagnose vasculitis. Paradoxically, although the antibiotic responsiveness of this nerve damage implicates active infection in its pathogenesis, in only one instance[31] has it been possible to demonstrate intact spirochetes in nerve. No study has demonstrated spirochete antigens or immune complexes in nerve that might directly link infection to nerve damage. A reasonable argument could be made that the absence of demonstrable spirochetes may result from sampling limitations, with very patchy nerve involvement making it unlikely that a biopsy would capture the key site. However, the demonstrated extent of nerve involvement—as seen in the facial nerve in Figure 1, and in published ultrasound images of involved nerves[32]—suggest peripheral nerve pathology is sufficiently extensive that biopsies should be informative. A plausible inference is that infection, perhaps starting with haematogenous spread to the epineurium, triggers a local inflammatory response, occasionally extending into the endoneurium damaging the nerve itself. In sum, considerable evidence suggests LNB peripheral nerve involvement begins with haematogenous dissemination of a small number of spirochetes to the vasa nervorum, where immune amplification results in vasocentric damage to fascicles with multifocal—potentially reversible—axonal damage.

Analogously, to infect the CNS, pathogens must cross the BBB or BCB. The three-layered BBB is more formidable, with endothelial cells joined by tight junctions, surrounded by pericytes, (which can serve as microglia, 'backstopping' the endothelium), in turn surrounded by astrocytic foot processes, the glia limitans. CSF is formed primarily in the choroid plexus, principal site of the simpler BCB. Here, fenestrated vessel endothelium is immediately apposed to epithelium (CSF-facing ependyma) with these epithelial cells joined by tight junctions—a monolayer barrier. The remainder of the CSF space is lined by ependymal cells (ventricles) or pia (brain's surface), both presenting mechanical barriers but neither with tight junctions—the BBB-BCB difference perhaps accounting for Bbsl (like West Nile virus and other pathogens) more commonly causing meningitis than encephalitis. However, analogous to PNS epineurial perivascular inflammation, this suggests a mechanism by which pathogens, once in the subarachnoid and contiguous perivascular CSF spaces, could occasionally reach the brain from the BBB's abluminal side. In extreme instances this might explain what has been attributed to a CNS vasculitis in LNB.[38,39] Extremely rare LNB patients develop strokes, beaded changes on cerebral angiography and inflammatory CSF. No histopathological evidence has ever confirmed a true vasculitis. However, inflammation in the subarachnoid space could either directly, or through involvement of the vasa vasorum, cause large artery strokes, while a vasocentric process in the Virchow Robin spaces, analogous to that around epineurial vessels in peripheral nerve, could provide a plausible pathophysiological mechanism for small vessel strokes.

A paradox in CNS LNB is how, as in peripheral nerve, so few organisms elicit such a marked inflammatory response. Bbsl seeds the CNS early in acute dissemination.[40,41] In response, brain resident monocytes and microglia, likely in the CSF-contiguous perivascular spaces, produce cytokines, particularly CXCL13.[42–44] CXCL13, present in CSF in particularly high concentration in spirochetal nervous system infections,[45] then attracts B cells into the CNS. Once in the CNS, these B cells proliferate to produce Bbsl-specific antibodies intrathecally—providing a quite specific tool to diagnose CNS LNB. Notably, the underlying mechanism of CNS LNB is always inflammatory; considering a CNS disorder to be linked to LNB without CNS inflammation is not consistent with the known pathophysiology. Importantly, an analogous immune amplification mechanism in peripheral nerve could underlie a process by which a very small number of spirochetes could trigger significant multifocal peripheral nerve injury, similarly explaining this multifocal infectious neuropathy without demonstrable spirochetes.