While the spirochetal infections Lyme disease and syphilis are intriguingly similar, important differences also exist. Both infections start with unusual, often asymptomatic, cutaneous changes as spirochetes invade locally and multiply. Organisms can then disseminate widely, causing systemic symptoms as well as a multifocal rash, cardiac seeding, and meningitis. Both infections can become relatively asymptomatic only to cause more severe end organ involvement later and affect the nervous system and heart (syphilis) or joints (Lyme disease). Both infections elicit an easily demonstrable host immune response, although this seems to have limited ability to clear the infections or confer subsequent immunity. Both organisms remain highly sensitive to simple antibiotics. Unlike syphilis, however, Lyme disease is rarely fatal and rarely causes parenchymal central nervous system (CNS) damage. This article elaborates on the clinical characteristics of these two infections, focuses on the similarities and differences of the infections, and emphasizes both a rational diagnostic approach and appropriate treatment.
Lyme disease and the closely related disorders occurring in Europe and Asia are zoonoses that are caused predominantly by the three Ixodes tick-borne spirochetes, Borrelia burgdorferi sensu stricto (the only one of the three found in the United States), Borrelia garinii, and Borrelia afzelii. These multisystem infections, prevalent in temperate regions of northern and central Europe and Asia and the northeast and north central United States, typically begin with the characteristic erythroderma, erythema migrans (FIGURE 9-1).
This rash was first described more than a century ago by the Swedish dermatologist, Arvid Afzelius, who associated it with bites of the hard-shelled Ixodes sheep tick, Ixodes ricinus. The rash begins at the site of a tick bite, gradually enlarging over days to weeks as the spirochetes migrate centrifugally. Unlike the immediate pruritic reaction following the tick bite, which, like that following a mosquito bite, represents an allergic reaction to arthropod saliva, this rash begins after a delay of several days to a few weeks as the spirochetes initially multiply locally.
Ticks hatch uninfected; infection may ensue if they feed on an infected host, typically a small mammal such as a field mouse. Small numbers of Borrelia may then remain in the tick gut. Subsequent ingestion of blood triggers Borrelia proliferation and migration within the tick, ultimately reaching its salivary glands, from which Borrelia can be injected into the host and cause infection, a process that typically requires at least a day or two.
When infection disseminates in humans, it exhibits a number of organotropisms. B. burgdorferi sensu stricto (referred to from now on in this article as B. burgdorferi) causes multifocal erythema migrans in about one-fourth of patients, with each new focus representing a new nidus of infection arising from spirochetes that spread hematogenously from the original site. At each of these new sites, spirochetes again migrate centrifugally, causing additional slowly enlarging erythrodermas. Multifocal erythema migrans occurs significantly less frequently with B. garinii and B. afzelii, the predominant strains found in Europe.
About a decade after Afzelius’s original description of erythema migrans, two French clinicians described a sheep farmer who, again following a bite by an Ixodes tick, developed a large erythroderma associated with intractable radicular pain and focal muscle weakness and atrophy. He was found to have a CSF pleocytosis and a weakly positive reaginic test for syphilis. After explaining in detail why they did not believe this retired French foreign legionnaire could have had syphilis, they postulated that he had a different tick-borne spirochetal infection and treated him with arsphenamine, the “best practice” treatment for syphilis at the time. The patient rapidly became pain free for the first time in months. Two decades later, Bannwarth described a large series of such patients, whereupon this disorder became known as Garin-Bujadoux-Bannwarth syndrome, or, more commonly, Bannwarth syndrome. In the 1950s, the disorder was shown to respond to penicillin, which became standard treatment for many years.
Other than anecdotal reports of Montauk knee (spontaneous, recurrent, painful knee swelling among residents of eastern Long Island) and a single published report of erythema migrans from Wisconsin, no awareness of this disorder existed in the United States until the mid-1970s. At that time, several mothers in the towns of Lyme and Old Lyme, Connecticut, became alarmed at the number of children in their neighborhoods being diagnosed with juvenile rheumatoid arthritis. They reached out to the Centers for Disease Control and Prevention (CDC) and Yale University, where Allen Steere, then a rheumatology fellow, was assigned the task of investigating this disorder. In a series of landmark studies that continue to this day, Steere’s group found that these children had experienced Ixodes tick bites, had developed erythema migrans, and then went on to develop a relapsing-remitting large joint arthritis, clinically resembling juvenile rheumatoid arthritis. This would typically cause spontaneous painful swelling and redness of an individual (typically large) joint, which would then subside spontaneously, only to recur later in a different joint, a disorder that came to be known as Lyme arthritis.
Within several years, Reik and colleagues and Pachner and Steere, working with the group at Yale, observed that this disorder also caused a clinical syndrome of one or more of the clinical triad of lymphocytic meningitis, painful radiculitis, and cranial neuritis, much like Bannwarth syndrome. With the identification of B. burgdorferi as the causative organism by the Stony Brook and Yale groups in 1983 and the identification of the very similar European strains the following year by Asbrink and colleagues, it became clear that the European and US infections were closely related.
Although the infections share many similarities, some differences exist among the clinical syndromes associated with the different strains. B. garinii is the most likely to cause neuroborreliosis, although the frequency of nervous system involvement is only slightly lower in the United States. B. afzelii most commonly causes cutaneous involvement, particularly a peculiar late manifestation known as acrodermatitis chronica atrophicans. With this, the skin of a limb becomes purplish, tissue paper thin, and wrinkled, typically after prolonged infection. Biopsies demonstrate large numbers of spirochetes, just as in erythema migrans. This disorder has never been described outside of Europe. B. burgdorferi is the most likely to cause arthritis, although it is worth emphasizing that the title of Bannwarth’s article included the term rheumatism.
Overall, 10% to 15% of patients, both in Europe and in the United States, develop nervous system involvement, which is now generally referred to as Lyme neuroborreliosis.
The European literature has long focused on the nervous system as the predominant extracutaneous target of Borrelia infection. Like the patient described by Garin and Bujadoux, typical manifestations include lymphocytic meningitis and painful radiculitis; cranial neuritis was subsequently added to clinical descriptions of this entity. Although involvement of virtually all cranial nerves has been described, facial nerve palsy is by far the most common. Less frequent manifestations were added over the years. Around the same time that Steere and colleagues were investigating Lyme arthritis, Hopf in Germany observed that patients with acrodermatitis chronica atrophicans commonly developed a patchy axonal neuropathy. Although this is often said to be limited to the affected limb, both Hopf’s original article and subsequent articles by others actually described this as a multifocal neuropathy, most severe in the limb with acrodermatitis chronica atrophicans but clearly involving other limbs as well. Similarly, it has been commonplace over the years for European descriptions to emphasize that painful radiculitis often occurs in the dermatome that was the site of the tick bite. Based on this, it was suspected that spirochetes track along peripheral nerves to reach the CNS. More recent studies have found this pattern to be less consistent than was originally thought. Finally, the 1980s saw the description of patients with progressive CNS parenchymal inflammation (with progressive white matter signs of spasticity, ataxia, and sensory loss), considered a form of encephalomyelitis. Most commonly, this CNS parenchymal inflammation involved the spinal cord at the same segmental level as Bannwarth radiculopathy; however, cases with prominent brain inflammation were described as well.
Findings in US patients have largely paralleled European observations. Reik and colleagues and Pachner and Steere focused on the clinical triad consisting of one or more of the following conditions: lymphocytic meningitis, cranial neuritis, and radiculoneuritis. Each of these conditions has its own subtleties. Although a CSF lymphocytic pleocytosis often occurs with other components of the triad, symptoms of meningitis are highly variable. To confuse matters further, Lyme meningitis typically occurs in summer, with much the same seasonality as enteroviral meningitis. Algorithms have been developed to differentiate between the two, with symptoms in Lyme meningitis perhaps a little less abrupt in onset and somewhat less severe. However, most of the predictive power of these algorithms resides with the presence or absence of co-occurring cranial neuritis, something that is common in Lyme disease and exceedingly unusual in enteroviral meningitis. Although seventh nerve paresis (CASE 9-1) probably accounts for 75% of cranial neuropathies seen in Lyme disease, targets can include the nerves to the extraocular muscles (causing diplopia), the trigeminal nerve (facial pain, numbness), and the acousticovestibular nerve (hearing loss, tinnitus, vertigo). Bilateral seventh nerve palsies are not uncommon; because idiopathic facial nerve palsies occur much less commonly in children than in adults, unilateral facial nerve palsy in a child with potential exposure should certainly raise a suspicion of Lyme disease; bilateral involvement in a child is exceptionally unusual except with Lyme disease.
A 32-year-old man who lived in an area endemic for Lyme disease presented for evaluation of left facial weakness. He had developed left retroauricular pain the day before presentation, and by the next morning, he noticed that sounds were louder on the left, and his left eye felt scratchy. When he brushed his teeth, water dribbled out of the left side of his mouth. He had not experienced any systemic symptoms.
When he presented for evaluation, he had complete paralysis in the distribution of the left seventh nerve. His general and neurologic examinations were otherwise normal. In particular, no evidence of erythema migrans was present, and he did not recall a tick bite. Serologic testing for exposure to Borrelia burgdorferi was strongly positive. He was treated with ceftriaxone 2 g/d for 14 days and recovered fully over the next 5 weeks.
The facial nerve palsy of Lyme disease is clinically indistinguishable from Bell’s palsy. In a minority of patients, it occurs with or soon after erythema migrans. Notably, tick bites are typically asymptomatic, so many of these patients do not recall one. Like many other patients with Lyme disease, this patient did not recall the erythema migrans. It is often stated that between 20% and 50% of patients do not develop erythema migrans, although in children, under the watchful eye of parents, this number is only about 10%. Whether this is a real difference or a reflection of the fact that the rash is often asymptomatic and can be on a part of the body that is not easily visualized is a matter of conjecture. Most affected individuals are strongly seropositive.
Finally, radiculoneuritis is probably the diagnosis most often missed. Just as originally described almost a century ago, pain is radicular in distribution and character and can be severe and intractable (CASE 9-2A). While sensory deficits may be subtle, weakness can be striking; there is often significant weakness and atrophy in the same dermatome as the pain. Unlike mechanical radiculopathies, often a few adjacent dermatomes are involved. In the United States, the affected dermatome has not been associated with the site of the bite, but this has not been studied systematically.
An 82-year-old man who lived in an area highly endemic for Lyme disease presented for evaluation of severe left upper quadrant pain that he had been experiencing for 2 months. He was severely constipated with some urinary difficulty but was on high doses of opiates for pain control. He had been to his internist, two gastrointestinal specialists, a urologist, and a pain management specialist. He had undergone CT and MRI scans of the chest and abdomen and upper and lower gastrointestinal endoscopies, none of which had identified any abnormalities. He had lost 15.8 kg (35 lb).
In desperation, he requested a neurology consultation. Careful history revealed that his pain was superficial, burning in character, and was aggravated by light touch but not deep palpation. Examination demonstrated hyperpathia in the left T9 and T10 dermatomes. When he performed a sit-up, his left upper abdominal muscles bulged. Otherwise, his neurologic examination was normal. No vesicles were visible, and he had received “the shingles vaccine” the prior year.
Hemoglobin A1c was 5.1%. MRIs of the thoracic and lumbar spinal canal were unremarkable. Serologic testing for exposure to Borrelia burgdorferi was strongly positive (5.5 times the upper limit of normal, with 10/10 IgG bands). Given the particularly challenging nature of this patient’s disorder, a lumbar puncture was performed; CSF was abnormal with 38 white blood cells/mm3 (all lymphocytes), protein was 94 mg/dL, and glucose was normal. Intrathecal antibody production was evident with a CSF to serum anti–B. burgdorferi antibody index of 2.8 (ie, after normalizing for IgG concentration, CSF contained 2.8 times as much antibody specific to B. burgdorferi as serum). He received 3 weeks of IV ceftriaxone. Within 2 days of starting, the pain disappeared for the first time in months and he recovered fully over the next 5 weeks.
Pain from radiculoneuritis is neuropathic in character, severe, and often dermatomal and resembles a mechanical, diabetic, or postherpetic radiculopathy. In patients in areas endemic for Lyme disease, it is important to keep the diagnosis of Lyme radiculoneuritis in mind, particularly if the patient has a CSF pleocytosis.
The occurrence of other clinical phenomena led to an early emphasis on the protean nature of Lyme disease. Cases of brachial and lumbosacral plexopathies, mononeuropathies, and mononeuropathy multiplex have been well described. Just as in acrodermatitis chronica atrophicans, patients with Lyme arthritis, the analogous manifestation of chronic infection, were often found to have a patchy multifocal axonal neuropathy. In some patients with long-standing untreated infections, a confluent mononeuropathy multiplex may occur. Detailed neurophysiologic studies in large groups of patients with the full range of peripheral nerve manifestations suggest that all of these neuropathic manifestations, including radiculitis and cranial neuritis, are varying presentations of a mononeuropathy multiplex. Although limited pathologic data exist, the few clinical biopsies that have been performed have consistently shown patchy axon loss and a few perivascular inflammatory infiltrates, particularly in the epineurium, without evidence of a true vasculitis. Importantly, none of the biopsies have shown spirochetes, spirochete antigens, immune complexes, complement deposition, or anything else that hints at a mechanism.
Although molecular mimicry has been suggested, with cross-reactivity between Borrelia and peripheral nerve antigens, several observations weigh against this. First, facial nerve palsy or painful radiculitis will occasionally occur before the patient develops a measurable antibody response, making it unlikely that enough antibody would be present to cause peripheral nerve damage. Second, it is highly unusual for peripheral nerve disease to continue to progress following antibiotic therapy, something that might be expected if cross-reacting antibodies were involved, particularly since serologies can remain positive indefinitely following fully treated infection. Histopathologically, no evidence has been seen of a severe, ongoing inflammatory process in the nerve or elsewhere after antimicrobial treatment. Finally, in the only animal model of neuroborreliosis (the experimentally infected rhesus macaque monkey), virtually every infected animal developed a patchy mononeuropathy multiplex. Neurophysiologic and neuropathologic changes have been identical to findings in humans with this disorder. Also, as in humans, no evidence of spirochetes, their antigens, complement, immune complexes, or antibody deposition has been observed to suggest a mechanism.
As in European patients, US patients very rarely may develop parenchymal CNS inflammation (encephalomyelitis). Patients have focal findings on examination and on CNS MRI imaging with inflammatory CSF. In a few patients, positron emission tomography (PET) scanning has demonstrated hypermetabolic foci. Symptoms, signs, and imaging changes can improve with antimicrobial therapy, although, as with any inflammatory CNS disorder, residual deficits can remain if structural damage has occurred. Less rarely, patients, particularly children, develop pseudotumor cerebri–like symptoms and signs, with transient and occasionally permanent visual loss, papilledema, and headaches. Most patients have had a CSF pleocytosis, so these may actually be cases of meningitis with raised intracranial pressure. Regardless, it is essential to recognize the clinical picture and treat the raised intracranial pressure as well as the infection to prevent visual loss.
While other neurologic disorders have been reported to occur in patients with some evidence of Lyme disease, proving a causal relationship has been far more challenging. Reports have described patients with serologic evidence suggesting exposure to B. burgdorferi or other reasons to suspect Lyme disease, who had parkinsonism, dementia, motor neuron disease, psychiatric disease, or other disorders. However, in most cases, there has been little to suggest any form of CNS infection, or even inflammation, that would create a plausible mechanism by which this infection could have caused this nervous system dysfunction.
Laboratory studies should follow a logical sequence, starting with testing of peripheral blood for serologic evidence of exposure to B. burgdorferi. As with any infection, it takes time to develop a measurable antibody response; in the case of Lyme disease, this may take 3 to 6 weeks. As a result, serology is negative in many patients with erythema migrans. Although neuroborreliosis requires disseminated infection and, therefore, typically occurs a little later, some patients with facial nerve palsies or other manifestations may seroconvert up to a few weeks after presentation. Curiously, while serologic responses are assessed using acute and convalescent samples in most infections, in Lyme disease, judgments are usually based on a single sample. However, because exposures to novel antigens typically lead to increasing amounts of antibody over time, follow-up serologic testing in 2 to 4 weeks is entirely reasonable in situations with a strong index of suspicion of early acute neuroborreliosis (symptom duration of just a few weeks) but initially negative serology. By the same token, positive serology in a person living in an endemic area need not reflect an acute infection. Either a symptomatic or asymptomatic infection may have occurred months or years before and be irrelevant to the current symptoms, although it might be prudent to treat such a patient if he or she has not been treated previously.
Historically, early serologic testing was confounded by false positives and false negatives. In the United States, two-tier testing was ultimately instituted to address this. An enzyme-linked immunosorbent assay (ELISA), quantitating total antibody reacting to B. burgdorferi, is performed first. In patients in whom this test is borderline or positive (originally, respectively, 2 and 3 standard deviations from the mean, now usually related to the area under the curve of a receiver operator curve), a Western blot is performed to determine the specific antigens to which patient serum reacts. Statistical studies in large populations found that patients with acute disease will typically have at least two of three specific IgM immunoreactivities (TABLE 9-1). In patients in whom disease has been active for more than 3 to 6 weeks, IgM reactivity is largely irrelevant (and actually more likely to be a false positive than a true positive), but most patients will have at least five out of 10 specific IgG immunoreactivities. Notably, if the ELISA is negative, a Western blot should not be performed; by the same token, a Western blot should not be performed without knowing the results of the ELISA. A number of large commercial laboratories now perform ELISAs and Western blots using well-standardized methodology, interpreted based on the recommended criteria by the CDC (TABLE 9-1); these tests are now highly reliable. Unfortunately, numerous other laboratories, often “specializing in tick-borne disease diagnosis,” have developed their own methodologies and interpretive criteria that have not been well validated. Unless and until such validation studies are performed and published, these assays cannot be considered reliable.
Although the current laboratory approach is accurate, it is both imperfect and labor intensive. More recent efforts have focused on using more restricted antigens in testing, with a particular focus on what has been termed C6, a restricted segment of the VlsE peptide that is highly conserved among both European and US strains, with some data suggesting it may be possible to substitute C6 for the standard ELISA or even the Western blot, so long as two different measures are used to enhance specificity.
The other important diagnostic consideration relates to examining CSF. Deciding whether to perform this test in a particular patient rests on two key considerations, the first of which relates to the likelihood that the CNS is involved. In patients with no evidence of any nervous system involvement of any sort, or in whom involvement is strictly limited to the peripheral nervous system, there is no reason to think CSF will be abnormal or informative. The second consideration relates to how the information will inform the patient’s subsequent care. Since most CNS involvement can be successfully treated with oral doxycycline, the need for CSF information is debatable, although a case can be made for obtaining definitive information where concerns exist about other diagnoses (CASE 9-2B) or about parenchymal CNS involvement or to obtain baseline information in challenging situations where objective measures of treatment response may be helpful.
In assessing CSF, both nonspecific and specific measurements are helpful. As with any CNS infection, there should be a modest pleocytosis (typically <150 cells/mm3 or so, lymphocyte predominant). Protein is usually modestly elevated (<200 mg/dL or so). Total IgG is often elevated, particularly in European infections; in such instances, oligoclonal bands are not uncommon. Because serologic cross-reactivity with syphilis is not uncommon, a reaginic test for syphilis (eg, rapid plasma reagin [RPR], Venereal Disease Research Laboratory [VDRL]) should be routinely performed. Unfortunately, neither culture nor polymerase chain reaction (PCR)-based organism detection in Lyme borreliosis is particularly sensitive, so the most robust testing relies on serologic methods, ie, measurement of intrathecal production of specific antibody. To perform this, it is necessary to measure both serum and CSF antibody simultaneously, correcting for the amount of IgG that crosses the blood-brain barrier in that patient. Technically, this is most simply accomplished by means of a capture assay, which inherently measures specific antibody as a proportion of the total. Alternatively, one can measure IgG in CSF and serum and either dilute both fluids to the same final IgG concentrations before doing the ELISAs or perform the ELISAs at standard dilutions and correct arithmetically (this is the standard method used in Europe). Whichever method is chosen, the essential element is to measure CSF and serum simultaneously, correcting for blood-brain barrier function. This has two shortcomings; first, no gold standard diagnostic test exists to determine its overall sensitivity and specificity, and second, this ratio may remain elevated indefinitely, just like serum antibody. Hence, a positive CSF index reflects a past or present CNS infection. On the other hand, just as in neurosyphilis, the activity of CNS infection is usually reflected in the CSF cell count and protein, both of which slowly normalize after successful treatment.
Recent work has focused on the CSF concentration of CXCL13, a B-cell attracting cytokine that is elevated in many CNS infections but may be particularly elevated in spirochetal infections of the CNS. Although the specificity of this measure remains to be determined, it appears to be a reasonable measure of active infection and therefore may provide a useful measure of treatment success.
The patient in CASE 9-2A recovered fully over the next 5 weeks after treatment and functioned normally for 5 years. Then, over the course of a year, his gait became unsteady, and he fell several times. His voice became softer, and his handwriting became smaller. An enzyme-linked immunosorbent assay (ELISA) for Lyme disease was 2.8 times the upper limit of controls, with 6 IgG bands (10 originally), and no IgM bands. Retreatment with ceftriaxone was recommended by his primary care physician. The patient had not been bitten again, had no systemic or rheumatologic symptoms suggesting active Lyme disease, and had not had headaches. The course of his current neurologic symptoms had been slowly progressive. His neurologic examination was notable for normal cognitive function and the presence of obvious parkinsonian features.
Brain MRI with and without contrast showed only atrophy, with no focal areas of inflammation. CSF showed 2 lymphocytes/mm3 and protein of 65 mg/dL. CSF to serum Lyme antibody index was 2.7. Treatment with carbidopa/levodopa was recommended but not antibiotics.
This patient’s current serologic results in blood and CSF are consistent with past infection. The absence of a pleocytosis or significant elevation of protein makes active central nervous system infection highly unlikely. The presence of fewer bands and a lower ELISA value weigh against ongoing immune stimulation by active infection. Finally, parkinsonism is not one of the neurologic disorders causally linked to Lyme disease. In theory, although active inflammation in the basal ganglia could lead to a picture mimicking parkinsonism, the absence of brain MRI abnormalities or active CSF inflammation makes this theoretical possibility exceedingly unlikely.
LYME ENCEPHALOPATHY AND POSTTREATMENT LYME DISEASE SYNDROME
Before the wide availability of reliable diagnostic testing and establishment of standard treatment regimens, it was not uncommon to see patients with Lyme arthritis that had gone untreated for months or even years. Many of these individuals described memory and cognitive slowing that, while not profound, was sufficient to interfere with their cognitively demanding jobs (CASE 9-3).
A 48-year-old man who spent the summers in an area endemic for Lyme disease presented to his primary physician with an 18-month history of episodic large joint arthritis. On one occasion, his right knee had become spontaneously red, hot, swollen, and painful for 2 weeks and improved with ibuprofen. On other occasions, his left elbow was similarly affected. On specific questioning, he recalled an “odd bug bite” two summers previously, with a rather large but asymptomatic erythroderma that lasted several weeks. Throughout the previous 18 months he had felt run down and cognitively slowed but had no headaches. He worked as an engineer, and he found that solving routine engineering problems took him about twice as long to complete as previously. He had to write himself notes to be sure he completed needed tasks, while previously his memory never failed him.
Neurologic examination was unrevealing, although formal neuropsychological testing confirmed mild memory and cognitive deficits. He was found to have a strongly positive Lyme enzyme-linked immunosorbent assay (ELISA) (7 times negative cutoff) with 8 bands on an IgG Western blot. Brain MRI with and without contrast was normal, as was CSF examination. After receiving 2 weeks of IV ceftriaxone, he had some intermittent arthralgia over the next 2 to 3 months but no further joint swelling. His malaise gradually improved, and his problem-solving ability returned to baseline. By 6 months following treatment, he was asymptomatic.
Termed Lyme encephalopathy, it was initially unclear if this picture in patients was due to brain infection or an indirect toxic metabolic encephalopathy. Although a few early patients with Lyme disease were found to have areas of increased T2 signal on brain MRI, in the majority of patients, MRIs were normal. No patients had focally abnormal neurologic examinations; only rare individuals had any CSF abnormalities. In virtually all patients, the cognitive and memory difficulties resolved with successful treatment of their Lyme arthritis. At this point, it seems clear that these cognitive difficulties, occurring in the context of systemic B. burgdorferi infection such as in this patient, were a nonspecific likely cytokine-mediated remote neurobehavioral effect of infection and should not be considered evidence of neuroborreliosis, ie, nervous system infection with B. burgdorferi.
Interestingly, some patients described these symptoms after otherwise successful antimicrobial treatment. In some, this was a continuation of symptoms occurring during acute infection; in others, onset apparently was after treatment. The true nature of this disorder, which has been termed posttreatment Lyme disease syndrome, remains unclear. It is clear that the symptoms are not improved by additional antibiotics; none of the patients have clinical phenomena, such as recurring arthritis, that would suggest ongoing infection. While often stated to occur frequently after Lyme disease, prospective studies of patients with erythema migrans suggest it occurs in a very small proportion (perhaps 3%) of patients, a finding consistent with the great difficulty identifying qualifying patients for clinical trials, in which about 1% of screened patients had sufficient evidence of actual Lyme disease infection, with symptoms persisting after appropriate treatment. European studies of the same phenomena have found it least likely to occur in patients with clinically definite neuroborreliosis, making an association with nervous system infection implausible. At the same time, population and cohort studies suggest the prevalence of persistent symptoms or objective abnormalities is comparable in patients who have had Lyme disease and in control populations, suggesting that this reflects anchoring bias in many individuals. At this point, it remains unclear if posttreatment Lyme disease syndrome actually exists or if it largely reflects misattribution of a very common symptom complex to sequelae of this infection.
Importantly, Lyme neuroborreliosis is a highly treatable disease. Approximately 95% of patients with Lyme meningitis, radiculitis, cranial neuritis, and other forms of peripheral nerve involvement will respond well to 2- to 4-week courses of appropriate antibiotics (now typically doxycycline 100 mg orally, 2 times a day for 2 to 4 weeks). While early studies focused on meningitis-level doses of parenteral penicillin, with the subsequent introduction of IV ceftriaxone (2 g/d IV for 2 to 3 weeks) specifically because of its excellent CNS penetration, numerous studies (all from Europe) have shown that oral doxycycline is as effective as parenteral β-lactams. Because of its extreme rarity, no systematic studies of treatment of parenchymal CNS infection have been conducted. Although anecdotal reports exist of such patients responding to oral doxycycline, parenteral ceftriaxone or similar antibiotics are recommended because of the lack of systematic data.
Several important questions remain. Doxycycline, the usually recommended antibiotic, is not currently recommended for children 8 years of age or younger because tetracycline irreversibly stains the teeth of younger children. More recent data suggest that this is not a common problem with relatively short courses of doxycycline, and the American Academy of Pediatrics will likely relax this recommendation in the near future. However, it is worth noting that no cases of breakthrough neuroborreliosis have occurred in children treated for erythema migrans with cefuroxime axetil or amoxicillin, something that should have occurred if these antibiotics were insufficient.
A second remaining issue relates to the concurrent use of steroids, particularly in patients with Lyme disease–associated facial nerve palsy. Steroids have been shown to improve survival in patients with Streptococcus pneumoniae meningitis and lessen neurologic sequelae in meningitis due to Haemophilus influenzae, so a rationale exists for use of steroids in bacterial meningitis. A small series found that steroids accelerated the resolution of radicular pain in patients with Bannwarth syndrome. Within that study cohort, some patients also had Lyme disease–associated facial nerve palsy. Long-term outcomes were equally excellent with or without steroids; those receiving steroids recovered more rapidly, although because of the very small sample size this was not statistically significant.
Importantly, steroids are clearly beneficial in idiopathic Bell’s palsy if given in the first 72 hours, a time period in which it is not always possible to obtain Lyme disease laboratory results. In light of this, several observations must be considered. In spring through autumn in areas endemic for Lyme disease, about one-fourth of facial nerve palsy cases is attributable to Lyme disease. In patients with idiopathic facial nerve palsy, the number needed to treat with steroids is seven to benefit one patient. Hence, in endemic areas, for every 10 patients with facial nerve palsy, withholding steroids from all until laboratory results return would harm one of the seven patients without Lyme disease. Giving steroids to all these patients might theoretically cause harm to a subset of the two or three with Lyme disease, although, to date, no data indicate this to be the case. In areas where the risk of Lyme disease is lower, the risk to benefit ratio of omitting steroids is skewed even more strongly toward harming patients without Lyme disease in the hopes of minimizing a theoretical risk in rare patients with this infection. Consequently, unless the patient has clear evidence of Lyme disease at presentation, idiopathic facial nerve palsies should be treated with steroids, absent any other contraindications. Although no systematic data exist in Lyme disease–associated facial nerve palsy, in this author’s opinion, the pathophysiology and rationale for steroids is the same in facial nerve palsy due to Lyme disease as it is in Bell’s palsy. Hence, starting steroids either just before antibiotics or simultaneously with them makes sense.
Several other important aspects are currently the subject of active investigation. Improvements in diagnostic tools, particularly those that differentiate between active infection and past exposure, would be very helpful. Our understanding of the pathophysiology of neuroborreliosis remains woefully inadequate. Finally, whether subsets of patients exist in whom oral antibiotics will be predictably ineffective remains an extremely important subject.
Syphilis and neurosyphilis became well-known in the years following Columbus’s return from the New World. Viewed by some as just retribution for measles and smallpox, bestowed upon the native people of North America by European sailors, infection with Treponema pallidum was initially highly lethal. Over time, presumably as a result of selection of organisms that could persist without killing their only host and vector and death of the most susceptible humans, syphilis evolved into a more indolent, persistent infection. Initial descriptions of the manifestations of this infection were quite varied, probably in substantial part because of limitations of medical knowledge. At a time when specific diagnostic tools were nonexistent, it was easy to attribute a broad range of otherwise poorly understood and unrelated disorders to this chronic infection. While reminiscent of the misattribution of numerous illnesses and symptoms to Lyme disease today, it is important to appreciate that ignorance about the biology of infections was universal in the days before Van Leeuwenhoek created the first practical microscope and first viewed bacteria and long before the development of accurate diagnostic testing for syphilis. Currently, ignorance about the biology of Lyme disease is more difficult to explain, given the vast amount we have learned during the 4 decades since it was first described.
The similarities between these two spirochetal infections are intriguing and informative. Both infections are spread almost exclusively by very close contact with a single vector species. Typically, both infections initially present as quite striking cutaneous abnormalities, which, despite being quite angry in appearance, are typically painless, presumably because of the peculiarities of the local immune reaction to these organisms. Untreated, both can cause indolent chronic infections that can be asymptomatic for a period of time but ultimately result in end organ damage, probably at least in part because spirochetes expose only a limited range of antigens to the host immune system and are capable of changing these antigens over time. Both have a propensity for crossing the blood-brain barrier early in infection, probably aided by their intrinsic motility. Because of the practical limitations of more direct microbiologic diagnostic tools, diagnosis for both relies on indirect diagnostic tools, primarily serologic, with their inherent limitations. Neither has developed significant antimicrobial resistance, despite which questions remain about whether a subset of patients exists in whom conventional antimicrobial therapy may cure the infection but not halt the disease. Interestingly, the numbers of cases meeting CDC surveillance criteria annually in the United States are now comparable, with about 28,000 cases of primary and secondary syphilis and 30,000 meeting the case definition for Lyme disease. Not surprisingly, while a sizable constituency has focused on the likely underestimation of the actual number of cases of Lyme disease, no such advocacy group makes comparable claims about syphilis.
Like B. burgdorferi, T. pallidum is a motile, corkscrew-shaped bacterium. It tends to be slightly smaller than B. burgdorferi, about 6 μm to 20 μm versus 10 μm to 30 μm in length, but about the same diameter (0.2 μm to 0.3 μm). Both infections are transmitted by intimate contact with a single vector species, and both have similar antimicrobial sensitivities. Antibiotic resistance has not been a substantial problem with either infection. Both are fastidious in vitro; T. pallidum remains impossible to culture, and culturing B. burgdorferi is impractical for clinical purposes.
The earliest descriptions of syphilis focused on cutaneous changes, the painless ulcerated chancre occurring at the site of inoculation (primary syphilis). Just like the painless erythema migrans at the site of B. burgdorferi inoculation, the chancre contains innumerable spirochetes. The standard method of diagnosing primary syphilis remains dark field microscopy of scrapings from the ulcerated surface. As in Lyme borreliosis, early spirochetal dissemination can result in multifocal cutaneous abnormalities (secondary syphilis) usually within a month of initial infection. The maculopapular rash of secondary syphilis is unusual in that it frequently involves the palms of the hands and soles of the feet. Unlike disseminated Lyme borreliosis, secondary syphilis often includes widespread lymphadenopathy, with frequent involvement of the epitrochlear nodes. In both Lyme borreliosis and syphilis, cutaneous abnormalities subside spontaneously, but resolution is more rapid with simple antibiotic regimens. In both, in the absence of early antimicrobial therapy, infection can persist and cause later manifestations.
Just as in Lyme borreliosis, early meningeal seeding is quite common in syphilis, sometimes in the absence of any apparent immune reaction. In the absence of antimicrobial therapy, secondary syphilis can become asymptomatic and latent. This is often divided into early (first year) and late latent disease, a categorization that is useful as patients with early latent infection are more likely to transmit infection.
About one-third of untreated patients with latent infection will progress to symptomatic late (tertiary) syphilis and develop either aortitis or late nervous system involvement. Although the clinical manifestations of tertiary syphilis are commonly referred to as protean, from a pathophysiologic perspective, they are due to one of three processes: an endarteritis obliterans, responsible for syphilitic aortitis and cerebral arteritis and stroke; creation of granulomata (referred to as gummas) that act as masses in the brain or elsewhere; and chronic meningitis.
Given the lack of direct microbiologic diagnostic tools, such as culture or PCR-based organism detection, the diagnosis of syphilis depends upon serologic techniques. Fortuitously, relatively high-titer reaginic antibodies (antibodies to cardiolipin) were identified in the sera of patients with syphilis long before the availability of more sensitive immunoassays for specific antitreponemal antibodies. Although the reason these antibodies occur remains unclear, the ability to measure them allows simple, inexpensive screening tests (eg, Wasserman test, Hinton test), some of which (eg, VDRL, RPR) remain in widespread use today. As confirmed in one recent study, while specific antitreponemal antibodies, as with IgG antibodies in most other infections, remain elevated long after successful treatment, these high-titer nonspecific antibodies are present primarily in active infection and fall with successful treatment. As such, they provide a useful measure of treatment efficacy.
Just as in early Lyme disease, serologic tests (both reaginic and treponemal assays) are often negative in primary syphilis, although the sensitivity varies; in one study, 38 of 90 patients with primary syphilis were VDRL negative but only six patients were negative for T. pallidum particle agglutination assay (TPPA) (a direct treponemal test). Just as in Lyme disease, this variability is due to the normal evolution of the host immune response during infection and in no way invalidates testing in later disease, in which virtually all patients are seropositive by either assay. Again, as in Lyme borreliosis, the diagnosis in early disease is quite straightforward independent of serologic testing, and the presence of motile spirochetes on dark field microscopy of scrapings from a chancre is diagnostic.
The development of immunofluorescent assays for specific antibody (fluorescent treponemal antibody-absorption [FTA-ABS] test absorbed against other cross-reacting treponemes) and, more recently, ELISAs led to a standard two-tier approach, which is important to eliminate false-positive screening tests because of the presence of anticardiolipin antibodies for other reasons. Sera screening positive for reaginic antibodies should be confirmed with a specific antitreponemal antibody test. Although used for many years, this approach has been confounded by the lack of a gold standard diagnostic test for validation purposes, leading to consideration of either changing the order of testing or using different tests. As in Lyme disease, considerations are driven by the positive and negative predictive values of different approaches. Some recent studies support continuing the current approach, finding that changing the sequence results in too many false positives. Others have led to the opposite conclusion. Importantly, in one large study, both reaginic (VDRL) and specific tests (TPPA) were positive in all patients with secondary or tertiary syphilis. Notably, that study confirmed that the VDRL fell with successful treatment while the treponemal test did not. Equally important, that study confirmed that both tests are equally valid in individuals who are positive for human immunodeficiency virus (HIV).
While the CDC recommends testing with a reaginic test followed by a treponemal one, the European Centre for Disease Prevention and Control recently recommended a different approach: to start with a treponemal assay as a primary screening test, then confirming with a different treponemal assay rather than a reaginic one. These conflicting findings and recommendations may, in part, have a Bayesian explanation. It may well be that different approaches have been used in populations with differing a priori risks of prior or current infection, where the relative importance of false positives and false negatives will differ. Notably, recent efforts to develop a simple automated dual approach (combining recombinant treponemal and nontreponemal antigens in a single assay) shows promise. Insofar as these use well-defined antigens and are more standardized and technically simple to implement, such an approach may render much of this debate moot.
Testing of CSF has long played an essential role in the diagnosis of neurosyphilis. Given that treatment of patients with neurosyphilis differs from that in patients without nervous system infection, this will continue to be important. Diagnosis of definite neurosyphilis requires the following: “(1) positive treponemal and nontreponemal serum test results; (2) positive CSF VDRL or positive CSF FTA-ABS test result and one CSF laboratory test abnormality, such as pleocytosis (cell count >20/μL) or high protein level (>0.5 g/L); and (3) clinical symptoms.” As in neuroborreliosis, even in the presence of clear inflammation in the CSF, PCR is positive in a minority of samples. In at least one systematic study of patients meeting the preceding diagnostic criteria, 48% had a CSF pleocytosis, and 88% had elevated protein. Only 35% had a positive CSF VDRL, while all others had a positive CSF FTA-ABS. Other common abnormalities in neurosyphilis, as in other disorders with prominent B-cell stimulation, include increased overall IgG synthesis in the CSF and even oligoclonal bands.
The relative diagnostic utility of measurement of different antibodies in the CSF remains unclear, and studies vary widely depending on the test used. Reaginic tests are in many ways preferable, as these must be present in high titer to be measurable and therefore are unlikely to be due to nonspecific cross-reactivity. A positive CSF VDRL is considered diagnostic; the CSF RPR has more false positives and is less reliable. Specific antitreponemal antibodies are subject to the same limitations as Lyme serologies. These very sensitive assays are prone to false positives from related infections (Lyme neuroborreliosis can cause strong cross-reactivity in T. pallidum ELISAs or FTA-ABS, but not VDRL). As in serum, CSF treponemal antibodies often remain positive long after successful treatment, unlike the CSF VDRL, which, along with the cell count and protein, reliably falls after successful treatment. Perhaps the biggest challenge with specific antibodies relates to differentiating between antibody produced within the CNS and that crossing the blood-brain barrier. In uninflamed meninges, about 0.2% of peripheral blood IgG enters the CNS, accounting for the 2 mg/dL to 3 mg/dL normally found in CSF. If a patient has a high-titer treponemal antibody in his or her blood, this will be detectable in CSF, but would only be evidence of CNS infection if the CSF elevation were disproportionate to that in serum. Similarly, if the meninges are inflamed for other reasons, with leakage of more IgG into the CSF, measurement of specific antibodies could be artefactually elevated. Thus, measurement of CSF antibody compared to serum, normalized for total CSF and serum IgG concentrations, would appear to be the most appropriate approach.
Finally, just as in neuroborreliosis, the prominent role of B cells in response to this infection is mirrored by elevation of CSF concentrations of CXCL13, although this elevation appears to be less dramatic than in neuroborreliosis.
In the acute disseminated phase (stage two), patients develop a substantial spirochetemia with seeding of multiple organs, with the skin, nervous system, and kidney being particularly susceptible targets. As many as 40% of untreated patients will develop CNS seeding, typically meningitis. About half of patients with CNS seeding develop typical meningitis symptoms (headache, nausea, vomiting, photophobia, with or without cranial nerve abnormalities). Ocular inflammation, including uveitis and retinitis, occurs in up to 80% in some series.
Late or tertiary neurosyphilis occurs in a smaller proportion of untreated patients and takes several different forms. Meningovascular syphilis, in which brain arteries develop an obliterative endarteritis, presumably as a result of the chronic meningitis surrounding them, presents as strokes, occurs several years into the illness, and represents about 10% to 15% of neurosyphilis cases. Chronic meningitis can also affect the underlying brain parenchyma, causing parenchymatous neurosyphilis. Involvement of the upper brainstem and surrounding cisterns can lead to Argyll Robertson pupils (small, irregular pupils unreactive to light but reactive to accommodation). Tabes dorsalis results from involvement of the dorsal root ganglia and their roots, with secondary degeneration of the posterior columns. Patients develop both positive and negative sensory symptoms with lightning-like tabetic pains and marked sensory loss. Loss of both proprioceptive and pain perception can leave patients unaware of repeated and severe joint damage, leading to joint destruction (ie, Charcot joints). Finally, about 10% to 15% of patients with neurosyphilis may develop late parenchymatous brain involvement. This may consist of focal gummas, presenting as mass lesions, or as late dementia, often referred to as general paresis of the insane. Thought to be due to chronic meningitis, this presents with an insidious frontotemporal dementia with personality changes, often with hallucinations or delusions. Pathologic changes are typically subtle, and the exact pathophysiology is unclear.
T. pallidum remains highly sensitive to penicillin (TABLE 9-2), with treatment for late latent or tertiary syphilis requiring longer courses. Neurosyphilis requires substantially more aggressive treatment. Treatment with ceftriaxone has been suggested as an alternative, particularly in individuals allergic to penicillin. At least one recent study showed ceftriaxone 1 g/d for 10 days to be as effective as benzathine penicillin G in early syphilis. Individuals coinfected with HIV require special consideration. While a single dose of benzathine penicillin G is as effective as three doses in such individuals with early infection, overall, treatment may be less effective, with up to one-third of patients with HIV failing to serorevert after treatment, although symptomatic treatment failures have been rare. At this point, while treatment recommendations are the same regardless of HIV status, very close follow-up is prudent in patients with HIV.
Neurosyphilis has been well-known for many years. Although diagnostic and therapeutic approaches have been developed, recent work suggests opportunities to improve diagnostic sensitivity using updated tests and algorithms. Treatment is generally highly successful, although the completeness of cure in some individuals coinfected with HIV remains to be established. Whether additional treatment regimens will be needed in such patients remains to be determined.
As with neurosyphilis, much is already known about neuroborreliosis. Both infections are highly responsive to readily available antibiotics. Diagnostic tools in both are robust, but in both, recent efforts have focused both on simplifying the techniques used and improving test sensitivity and specificity. Although in both we understand fundamental aspects of pathophysiology, perhaps the greatest difference is how much more we need to learn about the detailed mechanisms by which Borrelia species affect the nervous system.
- Overall, 10% to 15% of patients, both in Europe and in the United States, develop nervous system involvement from Borrelia infection, which is generally referred to as Lyme neuroborreliosis.
- Although a CSF lymphocytic pleocytosis often occurs with other components of the triad of lymphocytic meningitis, cranial neuritis, and radiculoneuritis, symptoms of meningitis are highly variable.
- Bilateral seventh nerve palsies are not uncommon in Lyme neuroborreliosis; because facial nerve palsies in children occur much less commonly than in adults, unilateral facial nerve palsy in a child with potential exposure should certainly raise a suspicion of Lyme disease; bilateral involvement in a child is exceptionally unusual except with Lyme disease.
- Radiculoneuritis from Lyme neuroborreliosis is probably the diagnosis most often missed.
- All neuropathic manifestations of Lyme neuroborreliosis, including radiculitis and cranial neuritis, are varying presentations of a mononeuropathy multiplex.
- As in European patients, US patients very rarely may develop parenchymal central nervous system inflammation (encephalomyelitis) from Lyme neuroborreliosis.
- Because exposures to novel antigens typically lead to increasing amounts of antibodies over time, follow-up serologic testing in 2 to 4 weeks is entirely reasonable in situations with a strong index of suspicion of early acute neuroborreliosis (symptom duration of just a few weeks) but negative initial serology.
- In patients in whom Lyme disease has been active for more than 3 to 6 weeks, IgM reactivity is largely irrelevant (and actually more likely to be a false positive than a true positive).
- If the enzyme-linked immunosorbent assay is negative, a Western blot should not be performed; by the same token, a Western blot should not be performed without knowing the results of the enzyme-linked immunosorbent assay.
- In patients with no evidence of any nervous system involvement of any sort or in whom involvement is strictly limited to the peripheral nervous system, there is no reason to think CSF will be abnormal or informative.
- Cognitive difficulties occurring in the context of systemic Borrelia burgdorferi infection are a nonspecific, likely cytokine-mediated remote neurobehavioral effect of infection and should not be considered evidence of neuroborreliosis, ie, nervous system infection with B. burgdorferi.
- It is unclear if posttreatment Lyme disease syndrome exists or if it largely reflects misattribution of a very common symptom complex to sequelae of the infection.
- Approximately 95% of patients with Lyme meningitis, radiculitis, cranial neuritis, and other forms of peripheral nerve involvement will respond well to 2- to 4-week courses of appropriate antibiotics (now typically doxycycline 100 mg orally, 2 times a day for 2 to 4 weeks).
- Unless the patient has clear evidence of Lyme disease at presentation, idiopathic facial nerve palsies should be treated with steroids, absent other contraindications.
- The similarities between Lyme disease and syphilis are intriguing and informative.
- Antibiotic resistance has not been a substantial problem with either Lyme disease or syphilis. Both are fastidious in vitro; Treponema pallidum remains impossible to culture, and culturing Borrelia burgdorferi is impractical for clinical purposes.
- Just as in Lyme borreliosis, early meningeal seeding is quite common in syphilis, sometimes in the absence of any apparent immune reaction.
- About one-third of untreated patients with latent infection will progress to symptomatic late (tertiary) syphilis.
- While specific antitreponemal antibodies, as with IgG antibodies in most other infections, remain elevated long after successful treatment, high-titer nonspecific reaginic antibodies are present primarily in active infection and fall with successful treatment of syphilis.
- Both reaginic and treponemal assays are often negative in primary syphilis.
- Sera screening positive for reaginic antibodies should be confirmed with a specific antitreponemal antibody test.
- Testing of CSF has long played an essential role in the diagnosis of neurosyphilis. Given that treatment of patients with neurosyphilis differs from that in patients without nervous system infection, this will continue to be important.
- Meningovascular syphilis presents as strokes, occurs several years into the illness, and represents about 10% to 15% of neurosyphilis cases.
- Treponema pallidum remains highly sensitive to penicillin.
- In patients with human immunodeficiency virus and syphilis, very close follow-up is prudent.