Encephalopathy is like fine art: Most people know it when they see it, but there is very little agreement on how to define it. At the 14th International Lyme Disease Conference, Brian A. Fallon, MD,[1] of Columbia University and the New York State Psychiatric Institute, New York, NY, tried to do just that. More importantly, he described the different ways one can define encephalopathy, the strengths and limitations of each approach, and significantly, what other aspects of life can give the impression of encephalopathy where none exists. First, one must evaluate patients with persistent Lyme encephalopathy by asking the following questions:

Is the diagnosis correct?
Are there comorbid psychiatric disorders that could be treated better? Does the patient have a psychogenic medical illness? What was the patient’s response to prior antibiotics?
Was previous treatment adequate? How long was the course, and what was the route of administration? Was there a subsequent relapse

Defining the Problem
The first question can pose a problem for the clinician. There is currently no agreed-upon definition of Lyme encephalopathy, and this has caused a great deal of confusion in the field. Encephalopathy was not included in the CDC’s case definition of Lyme disease, so NIH-funded studies of this condition can be hard to defend without a government- or society-sanctioned definition.
For investigators in this field, there have been numerous and differing definitions of Lyme encephalopathy. One of the earlier attempts at defining the problem was made by Logigian and colleagues in 1990.[2] This group listed the chronic neurologic abnormalities of Lyme encephalopathy as memory loss, depression, sleep disturbance, irritability, and difficulty finding words. However, there is much overlap between these symptoms and those of depression unrelated to Lyme disease.

Some further possibilities for defining the condition include self reporting of cognitive deficits, self reporting plus laboratory signs of CNS involvement, objective evidence of deficits on cognitive testing, or objective deficits plus laboratory signs of central nervous system (CNS) involvement. Signs of CNS involvement have included elevated cerebrospinal fluid (CSF) protein or pleocytosis, abnormal brain scans or tests (single photon emission computed tomography [SPECT], magnetic resonance imaging [MRI], or electroencephalogram [EEG]), intrathecal antibody production, or a positive polymerase chain reaction (PCR) for Borrelia burgdorferi DNA or a positive culture. However, objective tests often do not agree with patients’ perceptions. This is especially true for memory in depression

One Deficit, or Many?
One challenge is determining which and how many cognitive deficits to include in the definition and evaluation. For instance, some investigators have looked at a single, representative deficit (ie, single-domain methods) such as memory.[3] Others have looked at numerous deficits in each patient (ie, multiple-domain methods), such as memory, verbal fluency, and attention.[4] The advantage of the single domain method is that it focuses on one main problem and makes for a more homogeneous study sample. However, such a study may exclude patients who are impaired in other cognitive areas. The multiple domain method is particularly well suited for a disease that affects multiple cognitive domains, as one would expect for a global term such as encephalopathy. However, if patients have deficits primarily in one domain, this method becomes less sensitive (by dilution with less affected cognitive areas).
A further complication is how one defines and measures these deficits. One approach is to compare to age-matched norms. A second approach is to compare to actual or estimated premorbid or general ability levels. Comparison to age-matched norms provides simple, clear criteria for measurement and comparison, a cut-off score. But if the deficit being compared is correlated with a general ability (such as memory and general intelligence), then people with higher intelligence but with memory impairment may not be detected by this method. For example, if a subject has a general intelligence IQ of 130 and a memory score of 100, his full-scale IQ is 2 standard deviations above the age-norm, while his memory score is exactly at the age-norm. Compared to age-norms, this subject would not have memory impairment. Compared to his full-scale IQ, his memory score would be 2 standard deviations below expected — which clearly would suggest impairment. One method identifies this subject as normal, the next identifies him as impaired.

In contrast to the age-norm method, comparison to general levels of ability can allow for a more customized approach to assessing cognitive impairment, thus enhancing sensitivity. However, general abilities can be decreased by illness, regardless of a specific effect on that ability. In addition, this approach assumes that the domain of interest is strongly correlated with general ability.

The ideal screening tool for Lyme encephalopathy should have maximal sensitivity and specificity. Premorbid ability should be taken into account, by using norms adjusted for age, sex, and education level, or with balanced premorbid assessment of ability. It is still not clear whether a single- or multiple-domain definition of neurocognitive impairment should be used.

There are many screening tools for assessing premorbid ability, including verbal IQ, verbal comprehension index, vocabulary subtest score, reading subtest score (WRAT-R [Wide range achievement test-Revised]), demographic composite (Barona demographic equation), and national adult reading test (NART). Of course, different investigators have used different methods of assessing premorbid ability, thus further complicating comparisons between studies.

One of the few studies that actively evaluated patients with Lyme encephalopathy was conducted by Logigian and coworkers in 1997.[5] This study reveals some of these diagnostic problems discussed above. In this study, the investigators screened patients for Lyme encephalopathy and then evaluated the change in SPECT scan perfusion after treatment. Starting with clearly defined criteria for “definite Lyme encephalopathy” — subjective complaints of cognitive deficits, along with either a past or present CSF abnormality (intrathecal antibodies or PCR positivity) or objective cognitive deficits (as measured by 2 standard deviations below normal on verbal or visual memory tests, or 1 SD below normal on both tests) — they examined whether brain perfusion improved after treatment. While the brain imaging results were of interest in that all of the patients with definite Lyme encephalopathy showed improved perfusion after treatment, this study also demonstrated that the definition of Lyme encephalopathy, if restricted to cognitive testing, would have been too restrictive using their criteria. In other words, 5 of the 13 patients with “definite Lyme encephalopathy” did not have cognitive impairment using their criteria for impairment. Yet, these very same patients had abnormal SPECT scans that improved after treatment. Their cognitive criteria, which used age-norms for comparison, failed to identify 38% of patients with CNS abnormalities.

Dr. Fallon described an ongoing study of Lyme encephalopathy he is conducting at the NY State Psychiatric Institute. Because previous studies disagree about which is more accurate, single- or multiple-domain evaluation, they will collect data in 6 domains: motor skills, psychomotor skills, attention, memory, working memory, and verbal fluency. Data are preliminary, but so far the most sensitive method for the detection of impairment appears to be one that incorporates testing from multiple cognitive domains. Memory and working memory seem to be the most affected neurocognitive areas. Even so, about one quarter of the control subjects appeared to have neurocognitive deficits, compared to two thirds of the patients with complaints of memory impairment secondary to Lyme disease. Dr. Fallon suggested that the best method for detecting memory impairment in particular might be one that makes use of ethnicity- and education-adjusted norms. The Psychological Corporation is expected to publish such norms within the next year.

Complicating Factors
When determining the presence of encephalopathy in patients with Lyme disease, one must take into account other causes of cognitive complaints (Table 1), including the use of medications that can impair neurocognitive function (Table 2).

[Hysterical dementia is often over diagnosed in patients with Lyme encephalopathy, but it is rare for this to occur alone as a conversion symptom. The label of “hysterical” is often applied when phenomena are outside the clinician’s experience. For instance, females and male homosexuals more often receive this label. Based on studies over the past 50 years, many patients who are initially given the diagnosis of hysterical dementia go on to develop an organic CNS disorder.

Dr. Fallon is conducting a randomized, placebo-controlled study of brain imaging and treatment of persistent Lyme encephalopathy (Columbia University – National Institute of Neurological Disorders and Stroke [NINDS]). Treatment will involve IV ceftriaxone for 10 weeks, with a 14-week antibiotic-free follow-up period. At the end of the 24-week study, patients who had been randomized to receive placebo will be given 6 weeks of IV ceftriaxone. Evaluations will be conducted at baseline and 12 and 24 weeks. PET and MRI imaging as well as neuropsychiatric tests are being used to evaluate response to treatment. Patients will be recruited for this study over the course of the next 3 years.

An In-depth Study of Neurocognitive and Behavioral Lyme Disease

Patricia K. Coyle, MD,[6] and colleagues from the State University of New York at Stony Brook School of Medicine have conducted a prospective, controlled study to characterize the neurologic and neurobehavioral manifestations of Lyme disease in North America. They examined 3 groups: adults with acute disease, adults with chronic disease, and children with disseminated disease (ie, more than 1 erythema migrans [EM] lesion). They attempted to characterize changes to the CSF and identify pathogenetic mechanisms and predictors of outcome. This work builds on a previous study by Coyle and colleagues.[7]
The clinical syndromes studied included cranial (facial) neuropathy, radiculoneuritis, meningitis, and arthritis. The major symptoms (ie, seen in more than two thirds of patients) in adults with acute disease included fatigue, headache, sleep problems, stiff neck, and myalgia. Those with chronic disease had a different constellation of major symptoms, such as concentration difficulties, fatigue, arthralgias, myalgias, mood disturbance, memory loss, sleep problems, word-finding difficulties, knee pain, confusion, and stiff neck. Children with disseminated Lyme disease experienced major symptoms of headache and fatigue. By symptom score, the major initial defining syndrome for the acute adult group was meningitis, followed by multifocal EM, cranial nerve palsy, radiculoneuritis, and single EM; for the chronic adult group, single EM was the most common presenting symptom, followed by arthritis, cranial nerve palsy, and multifocal EM; and for children with disseminated disease, extraneural symptoms.

These investigators concluded that the children with disseminated disease are less symptomatic than adults, but they have more inflammatory CSF changes. Among the adult patients, those with chronic disease were more symptomatic than those with acute disease. They had more cognitive, mood, and joint disturbances; more severe symptoms; and more current depression, anxiety, and adjustment problems. However, it was the acute disease patients who were more likely to show objective cognitive deficits.

Surprisingly, CSF changes were not marked in either group of adult patients. In adults with acute disease, 45% had reactive or borderline CSF serology, 7% had intrathecal antibody production (a common test for CNS infection), 30% had an elevated white blood cell count (WBC), 23% had elevated protein, only 14% had evidence of oligoclonal bands, and only 7% had an elevated IgG index. Similarly, in the adult chronic disease group, 35% had reactive or borderline CSF serology, 6% had intrathecal antibody production, 15% had elevated WBC, 15% had elevated protein, 6% had evidence of oligoclonal band testing, and 3% had an elevated IgG index. Dr. Coyle plans to follow up with each group of patients to measure long-term (18-month) sequelae of the disease