College of American Pathologists
Lyme labyrinth—solving a diagnostic puzzle
February 2009 Feature Story
William Check, PhD
It has been more than 30 years since Lyme arthritis was recognized as a new clinical entity, with several key features of the condition having been established quickly. Most cases are signaled by the erythema migrans cutaneous lesion, which appears soon after the tick bite that transmits the infection. In addition to mono or pauciarticular arthritis, clinical manifestations can include neurologic or cardiac involvement. Antibiotic therapy shortens the duration of erythema migrans and prevents subsequent arthritis. Within five years of the new clinical entity having been recognized, Ixodes ticks were discovered to be the vector and the spirochete Borrelia burgdorferi the causative agent. All of these facts remain undisputed.
Soon after, diagnostic techniques for Lyme disease were evaluated, leading to two conclusions: Culture has low sensitivity in most situations and takes a long time to produce positive results, and laboratory diagnosis is best done with a two-tier approach—ELISA using reagents from a whole cell sonicate of B. burgdorferi as the first tier followed by immunoblotting for immu no globulins G and M (IgG, IgM) as a supplemental test on specimens found positive or equivocal by the ELISA. This protocol was rapidly incorporated into recommendations (Grodzicki RL, Steere AC. J Infect Dis. 1988;157:790–797; MMWR Morb Mortal Wkly Rep. 1995;44:590– 591). However, “Because of the low yield of cultures and the delay in the specific antibody response,” it was emphasized that recognition of the clinical picture, particularly the erythema migrans lesion, is important in diagnosis (Shrestha M, et al. Am J Med. 1985;78:235–240). Finally, in the early 1990s, overdiagnosis of Lyme disease was reported, primarily due to the misdiagnosis of pain and fatigue syndromes, such as fibromyalgia, as chronic Lyme disease (Sigal LH. Am J Med. 1990;88:577–581; Steere AC, et al. JAMA. 1993;269:1812–1816).
Unlike the first set of findings about Lyme disease, confusion and controversy continue to surround the second set of conclusions. A surprising lack of recognition of the typical clinical picture of Lyme disease, even with regard to the erythema migrans rash, still exists among clinicians, and even in endemic areas. Overdiagnosis continues because of the persistent belief in chronic Lyme disease in people with symptoms similar to those of chronic fatigue syndrome, either those whose typical Lyme disease symptoms have resolved with therapy or those who have no evidence of having had Lyme disease in the first place. Most recently, a new ELISA was devised using only one B. burgdorferi antigen, called C6. A multicenter clinical comparison of this assay against the standard two-tier algorithm has been done but is not yet published. It is not clear whether this new assay can replace the two-tier approach as a single test or whether it should remain limited to its current FDA-approved role as an alternative to the whole cell sonicate ELISA as the first step in two-tier testing.
What is as true today as in the early days of Lyme disease is that this infection in the early rash stage is primarily a clinical diagnosis. Later clinical manifestations require both a clinical diagnosis and supportive laboratory test ing by the recommen ded two-tier procedure. A positive serological test confirms the clinical picture. “As a laboratorian, I like to think that I diagnose infections,” says J. Stephen Dumler, MD, professor of pathology and associate director in the Division of Medical Microbiology, Johns Hopkins Medical Institutions. “But with Lyme disease I am helping clinicians confirm their clinical impression.”
A primary element in the clinical recognition of Lyme disease is the erythema migrans, or EM, rash. In fact, one way to satisfy the Centers for Disease Control and Prevention’s case definition is “physician-diagnosed EM along with solitary lesions with diameters of at least 5 cm” (MMWR Morb Mort Wkly Rep. 2001;50:181–185). Yet, although true EM-negative early Lyme disease is uncommon or rare, misdiagnosis is not. The skin lesion may be missed if it is located on a difficult-to-see area of the body, says Gary P. Wormser, MD, chief of the Division of Infectious Diseases, Westchester Medical Center, and vice chairman of the Department of Medicine, New York Medical College. “Because the lesion is not painful,” he says, “the patient may not notice it. Even without antibiotic treatment it goes away in about four weeks and later symptoms are reported as ‘EM-negative.’” Dr. Wormser has a walk-in Lyme disease clinic to which many patients come thinking they have Lyme disease, but report no rash. “However,” he says, “it’s there when we examine them closely.”
Emphasizing the importance of EM lesions in diagnosis, Robert P. Smith, MD, an infectious disease physician at Maine Medical Center and co-director of the Vector-borne Disease Laboratory in the MMC Research Institute, says the majority of cases are seen in physicians’ offices at the time of the rash. For these patients, serology does not play an important role in diagnosis. In fact, ordering serology in a patient with early Lyme disease can confuse the diagnosis. “I have seen situations where the patient had a classic EM rash, the test result came back negative, and the patient was told they did not have Lyme disease because the physician was relying on serology for the diagnosis,” Dr. Smith says.
Allen C. Steere, MD, professor of medicine at Harvard Medical School, says the critical idea is that most people with early B. burgdorferi infection have the EM rash but are frequently seronegative at that time. “If the skin lesion is thought to be Lyme disease, the clinician should treat,” says Dr. Steere, whose work was instrumental in the recognition of Lyme disease and the identification of its etiology. Treatment should be started even if the physician sends for serology.
“While most physicians are well versed in recognizing EM,” Dr. Smith says, “I have been surprised by calls over the years where a physician sees a case with multiple EM lesions and doesn’t realize that is a typical presentation of Lyme disease. One circle they recognize, but if they see multiple circles, they may think it is something else.”
At the opposite end of the spectrum from misdiagnosis of true Lyme disease is false diagnosis of chronic Lyme disease, a term applied to patients with nonspecific presentations who may or may not have evidence of past infection with B. burgdorferi. Symptoms can include fatigue, musculoskeletal aches, and neurocognitive dysfunction. “Post-Lyme disease syndrome” may be a preferable term in a patient with a prior documented infection. Dr. Wormser says there was no evidence of B. burgdorferi infection in the cerebrospinal fluid of more than 100 patients diagnosed with chronic Lyme disease based on either culture or PCR testing. Some physicians order long-term treatment with antibiotics for patients with these symptoms, despite a lack of evidence of sustained benefit in several prospective randomized trials (Marques A. Infect Dis Clin North Am. 2008;22:341–360; Auwaerter PG. Clin Infect Dis. 2007;45:143–148; Feder HM Jr, et al. N Engl J Med. 2007;357:1422–1430; Klempner MS, et al. N Engl J Med. 2001;345:85-92).
“There is a real polarization between the mainstream medical community and what I call the counterculture,” Dr. Steere says. “Many people misdiagnosed with chronic Lyme disease do not have evidence of past B. burgdorferi infection. Even if they have had Lyme disease that was treated with antibiotic therapy followed by symptoms, it has not been shown in any controlled trial that further or long-term antibiotic therapy is beneficial. That’s the crux of the controversy. I’ve seen people on IV antibiotics for a year or longer.” Prolonged antibiotic therapy itself has risks, Dr. Steere adds, and the risks are even greater with intravenously administered antibiotics.
Dr. Smith refers to the “confusing information and misinformation” about Lyme disease. “If somebody with chronic symptoms goes to the Internet, they may come away convinced they have Lyme disease,” he says. “Using a serologic test in that setting is more likely to cause confusion and misdiagnosis than to be helpful.”
Laboratory testing contributes most to Lyme disease diagnosis in those who present with extracutaneous manifestations, such as a swollen knee or meningitis. “These patients usually will have had a rash,” Dr. Smith says. “It was just not recognized. They come in a month or two after the acute phase with later manifestations of the infection. By this time the rash has disappeared.”
The American College of Physicians guideline does not recommend serologic testing in early disease because often the antibody response is not well developed at the time the EM lesion is apparent. “In fact,” Dr. Steere says, “these patients do get tested a lot and the results can often be negative in the first few days or weeks [after the onset of infection].” Once the spirochete is disseminated, people become seropositive and develop the characteristic manifestations of disseminated infection. After several weeks, patients may develop neurologic symptoms such as meningitis, cranial neuropathy, and radiculopathy, or cardiac symptoms, including AV nodal block. After months, arthritis may develop in one or a few joints—most commonly the knee. Arthritis is the most common manifestation of disseminated infection. Because of the time since exposure, the patient may not associate these symptoms with a tick bite. “Those patients should certainly have serologic testing,” Dr. Steere says.
Among methods for laboratory diagnosis, Dr. Wormser showed that, though the sensitivity of culture is low, it can be increased by using plasma and culturing larger volumes. “We now cultivate 9 mL of plasma and achieve greater than 40 percent yield in patients with the erythema migrans lesion,” he says (Wormser GP, et al. J Clin Micro biol. 1998;36:296–298; Wormser GP, et al. J Clin Microbiol. 2000;38:1648–1650; Wormser GP, et al. J Inf Dis. 2001;184:1070–1072).
After reagents for PCR were generated, PCR was compared with culture. In general, PCR has low sensitivity, except for synovial fluid, where its sensitivity can reach 90 percent or higher in patients with untreated or partially treated Lyme arthritis. PCR, however, can also give false-positive results (Aguero-Rosenfeld ME, et al. Clin Microbiol Rev. 2005; 18:484–509).
“One thing that continually harasses me is requests from clinicians for PCR for Lyme disease,” Johns Hopkins’ Dr. Dumler says. “There is a general belief that PCR is a panacea, but our best data now say that PCR in plasma, blood, or CSF is a very insensitive test for Lyme disease.” Dr. Dumler performed a meta-analysis of the sensitivity of this method. “A single PCR in a true-positive patient with disseminated Lyme disease is only 20 to 25 percent sensitive,” he says. “I’m always going over these data with clinicians.” There are those who advocate PCR testing for CSF, but Dr. Dumler does not agree. “In some cases of Lyme meningitis in the first 30 days you might use it on CSF,” he says. “But it still probably won’t detect the infection.” For meningitis he suggests doing serology, then starting treatment if the result is positive. A second approach is to use a ratio of CSF/serum antibodies. “If the level of CSF antibodies is higher, that probably means local stimulation of the immune system,” Dr. Dumler says. There is not a single serological test approved now in the United States for testing CSF, he notes.
The CDC recommended in 1995 two-tier testing to increase specificity. A positive or equivocal ELISA is followed by separate IgM and IgG immunoblots. “Compared to ELISA alone, two-tier testing adds cost, delay, and subjectivity,” Dr. Wormser says. Immunoblot for IgM causes the most problems. A positive result requires the presence of two of the three possible bands, which leads to many false-positive interpretations because of some laboratories’ “overreading” of weak bands. Specific IgM antibodies alone cannot be used to diagnose recent B. burgdorferi infection because the IgM response may persist for months or years despite effective antimicrobial therapy.
“The two-tier scheme has problems,” Dr. Dumler says. “Usually the screening ELISA is rapid in a high-volume laboratory. But the second phase, the confirmatory immunoblot, is very cumbersome and very expensive and increases the time to get a result. It causes a high number of queries about Lyme disease.” Many physicians have not paid attention to the details of the two-tier approach, Dr. Dumler says, drawing on his own experience. “That causes us innumerable difficulties helping physicians with this diagnosis.” In addition to requiring two of three possible bands, a positive IgM immunoblot result is valid only if the patient has had clinical manifestations for less than 30 days. “After 30 days the specificity of the IgM immunoblot as a single assay alone is very poor,” Dr. Dumler says. “It is more likely to be false-positive than true positive.” And the laboratory rarely knows how long a patient has had the infection, he notes.
Doing an immunoblot is also much more work. “It is a challenge to laboratorians to do these,” Dr. Dumler says. “They require much manual labor. And it is an assay that you have to read with your eye, so there is some degree of subjectivity in reading the intensity of these bands. I would prefer a more automated assay like an ELISA with a cutoff, even if it has an indeterminate zone for repeating.”
Because of general dissatisfaction with the two-tier approach, there was considerable interest when a new ELISA was introduced a few years ago. Called the C6 assay, it was based on measuring antibodies to a single antigen—a synthetic 26-amino acid peptide with the sequence of invariant region six of the VlsE protein, a B. burgdorferi surface antigen. After initial promising results, the FDA in 2001 approved a C6 kit from Immunetics as a first-tier test in the two-tier scheme for the diagnosis of Lyme disease.
Mario T. Philipp, PhD, and colleagues first identified and characterized C6 for use in the serodiagnosis of Lyme disease. “We were hoping to try VlsE as a vaccine candidate,” says Dr. Philipp, pro fessor of microbiology and immu nology and chair of the Division of Bacteriology and Parasitology, Tulane National Primate Research Center. However, other researchers discovered that V1sE was a protein that changed its antigenic properties as infection progressed, making it possibly unsuitable as a vaccine. Attention then turned to invariant regions of VlsE, segments whose antigenic properties did not change over time. “What struck us was that invariant region six reacted with serum specimens taken from nonhuman primates early in infection,” Dr. Philipp says. “We thought we might have a candidate for early diagnosis.” When they tested C6 with a battery of human specimens, sensitivity ranged from 74 to 100 percent, depending on stage of infection. More important, specificity was close to 100 percent, with only two false-positives out of 176 samples (Liang FT, et al. J Clin Microbiol. 1999;37:3990–3996).
Barbara J.B. Johnson, PhD, of the CDC’s Bacterial Diseases Branch in the Division of Vector-Borne Infectious Diseases, says that between 1999 and 2008, 34 publications in the scientific literature in the U.S. and Europe established the value of C6 peptide as an antigen in immunoassays for evidence of exposure to Borrelia burgdorferi. “CDC scientists have authored some of these reports and recognize the importance of this immunodominant antigen,” Dr. Johnson says. “Currently, there is considerable interest in whether the performance of the C6 assay is sufficiently good to merit its use as a stand-alone assay as an alternative to standard two-tier testing.”
To compare the C6 assay with the two-tier scheme, a multicenter study was done using more than 500 well-characterized Lyme disease patient sera—“the largest serological study ever attempted with a diagnostic,” Dr. Wormser says. It included 356 sera from patients with EM. Sensitivity of the C6 assay among patients with extracutaneous manifestations was comparable to the two-tier scheme, but these data are still preliminary and unpublished. For those with EM, the sensitivity of C6 was approximately double that of two-tier testing (Wormser GP, et al. Clin Infect Dis. 2008;47:910–914; and unpublished).
For the C6 assay as well as for the conventional first-tier assays and for two-tier testing, sensitivity was significantly higher in patients with multiple EM lesions relative to those with a single EM lesion. In untreated patients, sensitivity rose with time after onset, reaching 100 percent for single EM cases by 28 to 30 days (Wormser GP, et al. Clin Vaccine Immunol. 2008;15:1519–1522).
Specificity was evaluated in blood donors from endemic and nonendemic areas and in patients with a range of other disease conditions. Preliminary results show that the overall specificity was about 99 percent for both C6 and two-tier assays, similar to the figure Dr. Philipp and his colleagues reported in their 1999 paper. The two-tier assay showed slightly higher specificity in both the endemic and nonendemic blood donor groups, while C6 showed slightly higher specificity in patients with other disease conditions. These individual differences were not significant. Combining all controls, preliminary results show there was a roughly 0.6 percent advantage for the two-tier algorithm, which was statistically significant.
Dr. Wormser emphasizes the theoretical impact of even a small increase in false-positive results. If 2 million to 3 million Lyme disease assays are done each year in the U.S., a one percent increase in the false-positive rate means 20,000 to 30,000 more people will be treated, doubling the current number of reported cases. [Dr. Wormser and colleagues have found that the true number of cases is probably “several-fold higher” than the number of reported cases (Campbell GL, et al. Am J Epidemiol. 1998;148:1018–1026).]
However, the inappropriate ordering of Lyme tests on patients with no risk factors or other clinical evidence of infection also contributes significantly to the false-positive rate.
Dr. Philipp raises two criteria aside from sensitivity and specificity that, in his view, should be considered in the evaluation of the C6 assay. “I admit that I am biased toward C6,” he begins. “But there is another aspect that is very important, and that is ease of use. In this study the numbers were generated by very adept labs that know how to run immunoblots and interpret them by the two-tier criteria. But when immunoblot is done by standard clinical labs, the possibility of misinterpretation is very high. So simplicity of use is an advantage of the C6 assay.” He notes, too, that the C6 assay is much cheaper because it is one test as opposed to three tests in the two-tier approach an ELISA followed by separate IgG and IgM determinations for immunoblotting.
Dr. Steere’s laboratory recently published a prospective study of serologic tests for Lyme disease (Steere AC, et al. Clin Infect Dis. 2008;47:188 195). “What we are reporting is that all Lyme disease patients with [extracutaneous] involvement were seropositive,” he says. “Patients with arthritis had the highest antibody titers.” He found slightly better specificity with two-tier testing.
Of the C6 assay, Dr. Steere says, “It’s a good test. I don’t think of it as a stand-alone test. Its specificity is not quite as good as two-tier testing. A number of labs are using C6 as the first test; it works well in that regard. We are not suggesting in our paper that a change should be made, just that it works well.”
Dr. Smith has been impressed by the sensitivity and specificity of the C6 test compared with standard two-tier testing. “If the specificity is confirmed to be comparable, there are some potential advantages to the routine use of the C6 test. We see a lot of misinterpretation of two-tier testing in the community. It remains confusing to many clinicians, particularly interpretation of immunoblot results.” Dr. Smith is concerned about immunoblots being interpreted as positive when in terms of the clinical picture they should not be. In his experience, the prime example is interpreting an IgM immunoblot as serologic evidence of Lyme disease in a patient who has had symptoms for months or years and in the absence of an IgG response.
“If you had a simpler, well-standardized test that was of equal sensitivity and specificity, it might lead to less overdiagnosis and unnecessary treatment,” Dr. Smith says. He realizes there is concern about the specificity of C6 alone and substituting it for two-tier testing. “I think that is an important issue given the high use of Lyme disease testing,” he says, “particularly in low-prevalence settings. But I am encouraged by what I have seen in the literature so far.”
To the CDC’s Dr. Johnson, the main result in the findings published to date is that “the commercial C6 test was more sensitive than two-tier testing in patients with early Lyme disease presenting as erythema migrans in New York, confirming earlier work with noncommercialized C6 assays.” A comparison of the commercial C6 test and two-tier testing in a large population of Lyme disease patients with diverse manifestations of the disease is underway but not yet published, she says. Recommended standards for evaluating serologic tests for Lyme disease were adopted at a 1994 meeting in Dearborn, Mich. (MMWR Morb Mortal Weekly Rep. 1995;44:590–591). “CDC remains committed to this evidence-based process for determining whether any assay should be recommended as an alternative to two-tier testing,” Dr. Johnson says. “For the C6 EIA, not all of these criteria have yet been met by publication in the peer-reviewed scientific literature.”
Dr. Dumler, who took part in the multicenter study, also has a positive impression of the C6 test from the data, saying C6 is a good test that might replace completely the two-tier strategy. “One of the things we are looking at is equivalent sensitivity and specificity that is better in some regards,” he says. The C6 assay “would represent a less expensive, more rapid method than the existing two-tier approach for serological confirmation of Lyme disease.”
In the multicenter study, among samples taken during the first seven days after EM was identified, the sensitivity of the two-tier approach was only about 30 percent. For the C6 peptide ELISA alone, Dr. Dumler notes, sensitivity was “markedly better”—greater than 50 percent. “Those numbers increase over the next 14 days, as they should,” Dr. Dumler says. “People who have been infected 14 days ago are more likely to have antibodies.” Higher sensitivity soon after infection is important because physicians often order the test in that interval. With the two-tier approach, results typically come back negative even though the patient almost always has Lyme disease, perplexing the clinician. A test that would be positive more often early after infection would prevent some of this confusion.
Dr. Dumler has evaluated another assay that is also a candidate to replace two-tier testing: DiaSorin’s Liaison chemiluminescence assay for antibodies to the VlsE protein. “From the predominant data it looks like both the C6 and the VlsE assays have overall sensitivity equivalent to the two-tier approach,” he says. “That’s very promising. In early Lyme disease they may be superior [in sensitivity].
“From the statistical standpoint, in thousands of individuals, we can show that these two assays are almost equivalent for specificity,” Dr. Dumler says. Although statistically speaking C6 or VlsE ELISAs are equivalent in specificity to two-tier testing in population samples, he says, the specificity of C6 and VlsE ELISAs seems to be consistently lower than with the two-tier approach. “This concerns me because if this ‘nonstatistical’ difference is extrapolated over tens of thousands of tests, the difference could become significant and result in an increase in false-positive test results compared to the two-tier approach.” As a result, in Dr. Dumler’s view, a conclusion cannot yet be reached about the ability of either assay to replace the two-tier approach.
“My personal opinion is that, because of reservations due to the specificity issue, more testing will be required,” he says. “The last thing we need is a lot more people out there with a misdiagnosis of Lyme disease.”
William Check is a medical writer in Wilmette, Ill.