Article under review: Blanc F, Jualhac B, Fleury M, et al. Relevance of the antibody index to diagnose Lyme neuroborreliosis among seropositive patients. Neurology 2007;69:953-8.
Background: Lyme neuroborreliosis has a wide clinical variability of presentation, including craniospinal meningoradiculitis, isolated peripheral facial palsy, meningitis, encephalomyelitis, encephalopathy, mononeuropathy multiplex, and other less common syndromes.
Previous studies that assessed the diagnostic value of an intrathecal anti-Borrelia antibody index for diagnosis of Lyme neuroborreliosis found a sensitivity of 55% to 80%, based generally on typical cases of meningitis or meningoradiculitis (Stiernstedt et al 1985; Halperin et al 1989; 1991; 1996; Hansen et al 1990; Steere et al 1990; Hansen and Lebech 1991; Tumani et al 1995; Picha et al 2000).
Purpose: To assess the diagnostic value of the intrathecal anti-Borrelia antibody index for diagnosis of Lyme neuroborreliosis.
Methods: The intrathecal anti-Borrelia antibody index was determined as the ratio of anti-Borrelia IgG in CSF/anti-Borrelia IgG in serum to total IgG in CSF/total IgG in serum. The antibody index was considered positive when greater than or equal to 2, intermediate when between 1.5 and 1.9, and negative when less than 1.5 [the basis for this determination was not given].
This was a retrospective study spanning the years 1998 to 2005 at a university hospital in France. The authors studied a "sample of 123 consecutive patients with clinical signs of neurologic involvement and CSF anti-Borrelia antibodies" in whom the antibody index was also determined. The statement that these were "consecutive" seems to be somewhat of an overstatement. It is not clear how many total patients were suspected to have Lyme neuroborreliosis, how many of these actually had CSF antibodies, and how many of these had a positive antibody index as so defined.
The cases were categorized into 3 groups: Group I, definite neuroborreliosis (40 patients); Group II, possible neuroborreliosis (9 patients); and Group III, other non-neuroborreliosis diagnosis (74 patients). Group I, with definite neuroborreliosis, included those with typical meningoradiculitis, and/or the disappearance of neurologic signs after ceftriaxone treatment (2 g/day for "almost 2 weeks"), and the presence of B. burgdorferi antibodies in the CSF. Group II, with possible neuroborreliosis, had no other (non-neuroborreliosis) neurologic disease that could account for the clinical signs, no clinical response to ceftriaxone treatment, and the presence of B. burgdorferi antibodies in the CSF. The categorization into the 3 groups was made after determination of the antibody index, and without blinding of the antibody index results.
Results: In Group I, with clinically definite neuroborreliosis, the antibody index was positive in 30 (75%), intermediate in 6 (15%), and negative in 4 (10%). In Group II, with possible neuroborreliosis, the antibody index was positive in 5 (56%), intermediate in 1 (11%), and negative in 3 (33%). In Group III, with another etiologic diagnosis (apparently presumed to be the sole active neurologic diagnosis causing the concerning symptoms), the antibody index was positive in 2, intermediate in 10, and negative in 6.
The authors calculated sensitivity as the proportion of test positives (30) in Group I, giving an estimate of 0.75. The authors calculated specificity as the proportion of negative and intermediate test results in Group III, giving an estimate of 0.97. This, however, is problematic and leads to biased estimates as discussed below.
The individuals in Group III had a wide range of reported neurologic diagnoses, including intervertebral disk protrusion; lumbar or cervical spinal stenosis; carpal tunnel syndrome; hereditary polyneuropathy; acquired neuropathy due to Sjögren's syndrome, diabetes, or alcoholism; acute polyradiculoneuritis; chronic inflammatory demyelinating polyneuropathy; various dementias (vascular dementia, Alzheimer disease, Lewy body dementia, Creutzfeldt-Jakob disease, frontotemporal dementia, normal pressure hydrocephalus); familial spastic paraplegia; Parkinson disease; multiple sclerosis; acute stroke or stroke sequelae; arachonid cysts; herpes encephalitis; tick-borne encephalitis; Alport syndrome; cerebral hemorrhage; CNS tumor; malignant hypertension; migraine; "worsening paraparesis sequelae of myelitis secondary to acute prostatitis"; CNS involvement secondary to alcoholism; and presumed viral labyrinthitis.
In examining the list of Group III diagnoses (ie, with another non-neuroborreliosis diagnosis), one is hard-pressed to consider some of these as likely Lyme neuroborreliosis mimics, and it appears that CSF was obtained in some for questionable reasons (eg, intervertebral disk protrusion, spinal stenosis, carpal tunnel syndrome, hereditary polyneuropathy, diabetic neuropathy, alcoholic neuropathy, most of the chronic dementia cases, familial spastic paraplegia, Parkinson disease, acute stroke or stroke sequelae, arachnoid cysts, CNS tumors, malignant hypertension, and migraine).
Conclusions: The authors concluded that the anti-Borrelia antibody index had excellent specificity (97% as tested, but see concerns that this is biased as discussed in commentary below), but was insufficiently sensitive to diagnose all neuroborreliosis cases. In fact the measured sensitivity was 75%, indicating that many neuroborreliosis cases with even characteristic clinical features will not be diagnosed as such based on this test alone.
Discussion and commentary: Blanc and colleagues note that in Europe an anti-Borrelia CSF-serum antibody index is used as an indicator of intrathecal synthesis for diagnosis of Lyme neuroborreliosis. However, they incorrectly indicate that such is not the case in the United States, citing the surveillance criteria of the US Centers for Disease Control (CDC) (Centers for Disease Control 1997). An accompanying editorial by Roos and Berger concurs with these authors: "In the United States, the demonstration of anti-Borrelia burgdorferi antibodies in CSF is regarded as definitive evidence of neurologic Lyme disease based on the assumption that the presence of antibodies in CSF is evidence of intrathecal antibody production" (Roos and Berger 2007). Roos and Berger correctly note that "determination of the intrathecal production of antibodies required more than the detection of antibodies in CSF...as antibodies can be passively transferred from serum to CSF…"
CDC surveillance criteria in fact state explicitly that they are "not intended to be used in clinical diagnosis" (emphasis added) and these criteria do not in any case indicate that the presence of CSF anti-Borrelia antibodies are sufficient to diagnose Lyme neuroborreliosis (Centers for Disease Control 1997). According to CDC surveillance criteria, encephalomyelitis must be confirmed for surveillance purposes by demonstrated production of B. burgdorferi antibodies in CSF (ie, intrathecal synthesis), as evidenced "by a higher titer of antibody in CSF than in serum." Also, several American research groups have published methods of determining intrathecal antibody synthesis, and none relied on the presence of anti-Borrelia burgdorferi antibodies in CSF alone. CSF-serum antibody indices for assessment of intrathecal synthesis of anti-Borrelia burgdorferi antibodies are commercially available in the United States.
For surveillance purposes the CDC has defined a reportable case of Lyme disease as (1) physician-diagnosed erythema migrans at least 5 cm (2 inches) in diameter; or (2) at lease 1 objective late manifestation of Lyme disease (ie, neurologic, musculoskeletal, or cardiac) with laboratory evidence of infection with Borrelia burgdorferi in a person with possible exposure to infected ticks (Centers for Disease Control 1997).
Erythema migrans is defined for the US surveillance definition as an expanding skin lesion that starts as a red macule or papule and then enlarges over days or weeks to form a round rash with central clearing. Note that annual erythematous lesions developing within several hours of a tick bite are hypersensitivity reactions and not erythema migrans. Most patients with erythema migrans have other acute symptoms such as fatigue, fever, headache, mild neck stiffness, arthralgias, or myalgias. In the absence of known exposure, laboratory confirmation is recommended.
Late nervous system manifestations according to the US surveillance definition include any of the following (alone or in combination): lymphocytic meningitis, cranial neuritis (particularly facial palsy, which may be unilateral or bilateral), radiculoneuropathy, or rarely encephalomyelitis. Note that isolated nonspecific symptoms such as headache, fatigue, paresthesias, or mild neck stiffness are not criteria for neurologic involvement.
Practice parameters for diagnosis of Lyme neuroborreliosis authored by Halperin and colleagues and supported by the American Academy of Neurology concerning assessment of intrathecal antibody production state: "Properly performed (i.e., correcting for immunoglobulin leakage into the CSF because of blood-brain barrier disruption and eliminating the possibility of other spirochetal diseases, such as neurosyphilis), this method is highly specific (specificity of approximately 95%). Sensitivity estimates are difficult in the absence of other absolute criteria for CNS infection. Use of clinical criteria in patients with B. burgdorferi-associated meningoradiculitis leads to a sensitivity estimate of approximately 90%. Estimates in more chronic CNS disease are controversial and range from approximately 50% to 100%....When CNS involvement is suspected, demonstration of intrathecal antibody production is highly specific and quite sensitive, particularly in patients with acute disease; it may even rarely be diagnostic despite negative peripheral blood serologic tests....Intrathecal antibody production may persist for years following successful treatment, so this does not provide a useful marker of disease activity" (Halperin et al 1996). Even the brief summary statement of these practice parameters by the American Academy of Neurology Quality Standards Subcommittee, when I was a member, included the statement that "CSF analysis for cells, protein, and intrathecal production of specific antibody is indicated if CNS infection is suspected" (Quality Standards Subcommittee 1996).
An additional concern with the study by Blanc and colleagues is how they assessed sensitivity and specificity. They grouped test results into 3 categories (ie, positive, intermediate, and negative), and similarly grouped disease states into 3 categories (ie, definite neuroborreliosis, possible neuroborreliosis, and other conditions) - a 3x3 or 9-cell categorization. However, proper calculation of sensitivity and specificity requires a 2x2 or 4-cell categorization (Feinstein 1985). They performed the calculations by combining the intermediate and negative test results for only cases with either definite neuroborreliosis or other conditions, while inappropriately ignoring the possible neuroborreliosis cases in these calculations (Blanc et al 2007). This produces biased results (spectrum bias) as can be readily seen by considering such cases as either diseased or not diseased (Feinstein 1985; Nierenberg and Feinstein 1988; Lachs et al 1992). Moreover, the possible neuroborreliosis cases are precisely the cases that pose diagnostic challenges and in whom the utility of such a diagnostic test should be evaluated. The test was given the much easier task of discriminating between Lyme neuroborreliosis and many conditions for which the test was done without much suspicion of Lyme neuroborreliosis (ie, very low pre-test probability), which tends to inflate the calculated test indices.