Diagnostic Utility of CXCL13 in Lyme Neuroborreliosis

(See the Major Article by Eckman et al on pages 1719–26.)

Lyme borreliosis, which is caused by the infection with Borrelia burgdorferi sensu lato, is the most common vector-borne disease in the Northern Hemisphere. In most patients, the illness begins with an expanding erythema migrans skin lesion. If untreated, the infection can disseminate to involve the heart, joints, and the central nervous system (CNS) [1, 2].

Lyme neuroborreliosis (LNB) is the most common extracutaneous manifestation of Lyme borreliosis in Europe and the second most common such manifestation in the United States after Lyme arthritis. The pathogenesis of LNB is not well understood, and there are differences in the disease presentation between North America and Europe. In adult European patients the typical and most common manifestation of LNB is painful meningoradiculoneuritis (Bannwarth’s syndrome), which is rare in the United States. In contrast, the most common LNB presentations in US patients are meningitis and cranial neuritis, most often involving facial nerves. Presumably, these differences are due to different Borrelia species on the 2 continents.

The diagnosis of LNB is based on 3 primary criteria: presence of clinical signs and symptoms suggestive of LNB, cerebrospinal fluid (CSF) pleocytosis (>5 × 10⁶ leukocytes/L), and borrelia intrathecal antibody synthesis (ITAbs). However, the clinical signs and symptoms and CSF pleocytosis may overlap with other neurological conditions such as viral meningitis or Bell’s palsy and are thus not specifically indicative of borrelia infection. In addition, although highly specific, borrelia-ITAbs takes time to develop and thus offers limited sensitivity in early infection. Furthermore, once elevated, antibodies may remain high for prolonged periods of time after antibiotic therapy and thus offer limited ability to distinguish active from past infection. Inflammatory mediators such as CXCL13, a B-cell chemoattractant, offer attractive diagnostic targets since presumably CXCL13 levels rise prior to B-cell influx and borrelia-ITAbs, and decrease dramatically following antibiotics and resolution of the infection, thus addressing the major limitations of current diagnostic protocols.

Numerous studies in Europe have evaluated the utility of CXCL13 as a possible diagnostic marker for LNB. The study by Eckman et al [3] in this issue of Clinical Infectious Diseases is the second such study from this group to evaluate the diagnostic utility of CXCL13 in US patients [3]. In their study, the authors assessed the levels of CXCL13 in patients with “definite” LNB (those with appropriate clinical presentation, CSF pleocytosis, and borrelia-ITAbs), “probable” LNB (without borrelia-ITAbs), or other neuroinflammatory disorders to determine the diagnostic utility of CXCL13 for US patients with LNB. As in European studies, CSF CXCL13 levels were markedly elevated in US patients with definite LNB compared to patients with probable LNB or other neuroinflammatory conditions. The levels of CXCL13 were strongly associated with presence of borrelia-ITAbs and directly correlated with lymphocyte and monocyte CSF counts. Moreover, Eckman et al [3] showed that some patients with probable LNB have elevated CXCL13 levels, implying potential utility of CXCL13 in early infection, before sufficient borrelia-ITAbs. However, these studies also highlight that, in many such patients, CXCL13 levels remain low prior to demonstratable pleocytosis or borrelia-ITAbs and often overlap with other infections or neuroinflammatory conditions including viral meningitis or Bell’s palsy. Thus, although CXCL13 can discriminate between definite LNB and other neuroinflammatory conditions (Eckman et al [3] reported 75% sensitivity, 97% specificity), it has limited ability to distinguish patients with probable or suspected LNB from those with other neuroinflammatory/immune conditions. One could postulate that neurological symptoms in these patients are not due to borrelia infection, or that early in the infection the modestly elevated CXCL13 levels do not yet provide sufficient discriminating power.

This raises questions regarding the diagnostic utility of CXCL13 beyond patients with LNB who already meet the established diagnostic criteria [4]. As highlighted by Eckman et al [3] and by other studies, there is potential utility of CXCL13 in early LNB; however, this needs to be evaluated in greater detail in patients with highly probable LNB—for example, in those with typical erythema migrans who develop signs and symptoms of CNS involvement but who may not yet have pleocytosis or borrelia-ITAbs. Another potential diagnostic benefit of CXCL13 is to distinguish patients with past from current infection. Several studies have demonstrated a dramatic decrease in CXCL13 levels after antibiotic therapy [5–7], implying that CXCL13 could differentiate patients with active LNB from those with persistent sequela due to past infection but in whom CSF pleocytosis may remain elevated for weeks and ITAbs for months to years. For example, in a study of pediatric patients with LNB, at 4 months post–antibiotic treatment, CXLC13 levels decreased below the cutoff in 82% of patients fulfilling LNB criteria, while CSF cell counts decreased to normal values (≤5 × 10⁶/L) in 60% and borrelia-ITAbs in only 10% of patients [8]. Similarly, in adult patients with Bannwarth’s syndrome 3 months after the start of ceftriaxone treatment, CSF leukocyte counts decreased below the cutoff in 47% of patients, whereas borrelia-ITAbs decreased in only 9% of patients [9].

Although additional studies will clarify the utility of CXCL13 in LNB, perhaps another approach to improve sensitivity and specificity is to incorporate other markers in conjunction with CXCL13. A recent European study by Markowicz et al [10], using 2 separate CXCL13 assays, demonstrated that inclusion of CSF cell counts may improve sensitivity above that of CXCL13 alone. In both assays, the inclusion of CSF pleocytosis increased the sensitivity from approximately 84–88% (without inclusion of CSF cell counts) to 94–98% (with inclusion of CSF cell counts) with better or equivocal specificity (98–100%). Similarly, a study in children found that inclusion of total immunoglobulin M (IgM) index in addition to CXCL13 levels improved the sensitivity from 74% to 91%, particularly in those with probable LNB; however, the specificity was reduced from 97% to 91% [11].

The moderate success of CXCL13 as an adjunctive diagnostic marker in LNB lends support for approaches using immune mediators as biomarkers of disease course and outcome, particularly in conditions in which immune response is important in pathogenesis. Indeed, a recent study by Skogman et al [12] demonstrates that other inflammatory mediators, such as the ratios of interleukin 10 (IL-10):CXCL1, may improve sensitivity in early LNB. Of the 41 cytokines and chemokines tested, including CXCL13, the ratio of IL-10:CXCL1 had the greatest discriminatory power in patients with definite LNB as well as those with probable LNB who lacked borrelia-ITAbs and CSF pleocytosis (receiver operating characteristic curves: area under the curve was 0.997 for IL-10:CXLC1 and 0.868 for CXCL13). Of interest, both CXCL1 and IL-10 may have neuroprotective properties, presumably by regulating the immune responses in the brain.

A contributing factor to limited discriminatory power of diagnostic approaches in LNB are gaps in knowledge of disease pathogenesis. Although the immune response is thought to be an important driver of clinical signs and symptoms in LNB, the types of inflammatory and cellular responses involved have not been well defined. Studies of inflammatory mediators demonstrate that in LNB the immune responses are primarily concentrated in CSF and are thus presumed important in disease pathogenesis. Along these lines, during borrelia CNS infection, CXCL13 is likely produced locally in germinal centers in the brain and its production is dependent on specific cell types including monocytes/macrophages and dendritic cells [13, 14]. Additionally, borrelia/spirochetes may possess greater capacity to induce CXCL13 than other microorganisms. For example, Rupprecht et al [14] found that stimulation of cultured monocytes with Borrelia gariniii or Treponema pallidum (spirochetal bacterium that causes syphilis) resulted in higher CXCL13 levels than stimulation with pneumococci (causative agent of pneumococcal meningitis). A greater capacity of borrelia lipids to stimulate CXCL13 may explain the exceptionally high levels of CXCL13 in LNB and suggests at least some specificity of this marker.

Collectively, these studies imply that, in patients with LNB, host immune responses, particularly those occurring locally in the brain, are likely important for the control of spirochetal infection, but the consequence of this robust immune response is more severe disease. However, the mechanisms responsible for neurological symptoms in patients with suspected LNB, who often lack CNS measures of immune activation (inflammation, immune cell infiltrates, borrelia-ITAbs), are less clear. Patients with mild cognitive symptoms during acute Lyme borreliosis suspected of having LNB, and those after appropriate treatment, often have normal levels of inflammatory mediators in CSF, but elevated serum levels of Th17 cytokines and T-cell growth factors. These mediators are also elevated in patients who have similar symptoms without evidence of Lyme borreliosis. In the absence of CSF abnormalities, neurobehavioral symptoms appear to be associated with systemic inflammation, rather than CNS inflammation, and are not specific for borrelia infection [3]. A better understanding of disease pathogenesis will help elucidate the underlying mechanisms and identify new biomarkers for disease course and outcome.

In summary, current findings, including those by Eckman et al [3], demonstrate that the diagnostic utility of CXCL13 in both the United States and Europe is primarily limited to patients with definite LNB who meet other established diagnostic criteria. Although inflammatory immune mediators remain attractive targets for diagnostic assays, particularly in conditions where immune response is a key component in disease pathogenesis, the general lack of specificity limits their discriminating power. Optimization of such assays is critically important to improve sensitivity in early LNB and to help discriminate between active and past borrelia infection, likely the greatest diagnostic value of CXCL13. Finally, integration of other immune markers into the diagnostic algorithm should further improve the specificity and sensitivity. As with many human diseases, thorough clinical evaluation, history, and exposure risk remain important components in the diagnosis of LNB.

Note

Potential conflicts of interest. F. S. reports grants from the Slovenian Research Agency and personal fees from Roche and is an unpaid member of the steering committee of the European Society of Clinical Microbiology and Infectious Diseases Study Group on Lyme Borreliosis/ESCMID. K. S. reports grants from the National Institutes of Health and consulting fees from Roche and T2. A. J. H. reports collaboration on the European Union project on Borrelia serology within the Eurostars program with Reagena Oy Ltd, personal fees from Abbott Laboratories for collaboration on Borrelia serology, and collaboration on diagnostics for tick-borne diseases with Mabtech AB. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

References