IDSA Review Panel Hearing Testimony- Oct 2010

Dr. Baker: Conversations, please step outside the room. For any of you, who are here to come, please turn off all cellphones, all of those pagers and now I’m going to turn the podium over to Dr. Barbara Johnson.

1. Dr. Barbara Johnson: Thank you very much. I am Barbara Johnson. I am a biochemist by training and I am speaking today on behalf of the Centers for Disease Control and Prevention concerning serodiagnosis of Lyme disease. Specifically, my statement is in support of IDSA Recommendation 1 page 1120 and references elsewhere in the guidelines to use two-tiered serology to support the diagnosis of Lyme disease in patients who have manifestations other than acute erythema migrans. Please consult my written statement for references to the scientific studies that support what I will say here today and for more details than I can cover in the time allowed.

Serologic testing recommended by the CDC was promulgated in 1995 and it is endorsed by the IDSA. It uses a two-tiered procedure that consist of a first-tier test commonly an ELISA, enzyme immuno assay but rarely an immunofluorescent assay followed by conditional immunoblotting if the result of the EIA is positive or indeterminate. This recommendation was developed with the participation of the medical research community and the relevant public health agencies. CDC recommended two-tiered testing to improve both the diagnosis of the individual patients and to improved surveillance for Lyme disease.

Two-tiered testing established the standard by against which potentially improved methods of testing could be judged and it was adopted by the FDA as a standard against which new testing methodologies would be evaluated. The salient feature of the two-tiered algorithm is that the first-tier is designed to ask the question how much antibody is present in the serum sample.

The second is designed to ask a more qualitative question, what types are present, are the types of antibodies specificity is more indicative of infection by Borrelia burgdorferi than some of the other antibodies present in the sample and scored in the ELISA. There is a very extensive peer-viewed literature both prospective and retrospective studies of the performance of these two tests algorithm. Of course, I can’t review the literature extensively today and so I thought I might be most helpful to you as panel members if I address the series of misconceptions that have complicated our discussion of Lyme disease.

This misconceptions are that ELISAs are insensitive, the immunoblots should be first-tiered tests, than Osps A and B should be scored in blots, that IgM testing is appropriate at any stage of Lyme disease, that the validation studies used to access the performance of serology use circular reasoning, and that testing of patients without objective signs of Lyme disease or exposure to vector ticks is useful

So the first misconception that ELISAs are insensitive and therefore unsuitable as first-tier test. The scientific literature has established that the sensitivity of first-tier tests varies by stage of disease. We have excellent sensitivity after the erythema migrans stage. I illustrate this point with data from my laboratory but there are many other references that have reached many other studies that have reached similar conclusions. As you see in this illustration, there’s relatively low sensitivity and acute erythema migrans.

When these patients are treated at convalescence and recovery, the EIA becomes considerably more positive but the low sensitivity at this stage of illness accounts for the recommendation of both CDC and the IDSA that serology not be used in this circumstance. But after organisms have disseminated, the sensitivity of the ELISA is really excellent. So where did this misconception come from that EIAs are insensitive and that other means of evaluation are necessary? I believe from two sources.

First, literature of the late 1980s and early 1990s is commonly quoted but does not represent current laboratory practice. These studies were performed before two-tiered testing was adopted as the national standard. The second reason I believe is that often it is said that well the EIA misses a whole third of cultured confirmed patients. What may not be said though is that cultured confirmed patients, culture confirmation is really only consistently possible during the EM phase, the phase that we did not recommended testing for and so it is not reasonable to quote the performance of a test in a population in which it is not recommended and generalized it to later stages of disease where it can be clinically useful.

Another misconception, it is sometimes said that immunoblots are more sensitive than ELISAs and should be used instead of two-tiered testing. I think it’s really important to just reflect on fact that the ELISA and the immunoblot are not independent assessment of serologic status.

Commonly, the very same antigens, same prep is used in both the ELISA and in the western blot, that is a lysate of whole cells. There is no a priority reason for immunoblots to be more sensitive than ELISAs.

Immunoblotting is appropriately a second-tier test and why is this? Immunoblotting has a degree of subjectivity. Scoring of faint bands specially IgM bands can be a problem. It is the task of the mark reader to compare the signal which I’m sure you can barely see in the lane all C for calibrator.

The color density with each of the bands at the scored positions that requires some judgments and it is complicated particularly in the case of IgM blots by the fact that IgM antibodies are bit more sticky going to the fact that they are pentavalent rather than bivalent. And also, because of the scoring criteria themselves, in the case of an IgM blot, only two of three bands are sufficient for the blot to the be called positive. You can see that difficulty scoring just one band could impact IgM serology more readily than IgG. So these specificity problems in blotting have important consequences for public health.

In studies published both by [indiscernible] [0:08:31] and by myself depending on the population studied that is whether we’re looking at healthy donors or patients with other illnesses, the decrease in specificity if you skip the first tier was 2% to 8%. We don’t know exactly how many test I got in each year.

There have been crude estimates in a range of 1 to 2 million test a year and if this estimate is – this range of estimate is correct, for each 1% decrease in specificity, it would have the consequence of generating 10,000 to 20,000 positive tests a year. And I remind you of the context in which this result must be viewed. We have about 20,000 cases of Lyme disease reported each year through the National Surveillance System.

Third misconception, you may hear that the recommended criteria for scoring immunoblots are insensitive because they do not include scoring the Osps A and B bands. Sometimes referred to as 31 and 34 kiloDalton bands reflecting their apparent molecular mass on [indiscernible] [0:09:58].

This misconception is a little bit of a mystery to me how it got generated. The earliest work in this field published as long ago as 1992 indicated that the Osps A and B bands were less significant, are less commonly detected frequently detected compared with other protein bands. Sooner after Dressler and all, and the Steere laboratory published that the antibody responses to Osps A and B were not as diagnostic useful as they responses to 10 other antigens. When they were detected, it was in late infection specially Lyme arthritis when there is robust IgG response to numerous scored antigens.

Although this has been established now nearly a couple of decades, it’s more recently that people came to understand why is this so. Osps A and B are expressed by Borrelia burgdorferi mainly ticks not in the mammalian host and it is the basis for the mechanism of action of the one vaccine that has been marketed. Many times our patients or doctors will ask, well, if a vaccine – if it was in the vaccine, why is it not a good diagnostic antigen?

And the explanation is that the OspA vaccine blocks transmission of Borrelia burgdorferi from ticks to mammals. Antibodies OspA enter a tick as it takes the blood meal and Borrelia burgdorferi are killed or disabled because of they have an OspA on their surface when they are in ticks. So to just briefly summarize this scheme, here you have a defection of the life cycle of Borrelia burgdorferi as it travels between ticks and mammals. The red coding is simply to illustrate that Osps A and B are expressed in the tick base of the burgdorferi life cycle.

On the right hand side of the figure, it illustrates two antigens that are expressed in mammals and are particularly important in human serology. And finally, the human figure indicates an incidental host that is by the time a man, a human manifest with Lyme arthritis, these patients are essentially universally seropositive by standard criteria.

Next, I’ll address the question of IgM serology and the claim that it is appropriate for evaluating patients at any stage of illness. IgM serology should be re-observed for suspected early disseminated disease. It has diagnostic utility in the first month of infection when IgG antibodies are developing. This process of switching antibody classes from M to G is illustrated by a study of early disseminated disease characterized by facial paralysis. In this study that I’m citing here, 87% of the patients were IgM positive, 66% were IgG positive, 100% were M or G positive.

So there was diagnostic utility for looking at IgM in this circumstance. But patients with late Lyme disease are almost universally IgG seropositive and although IgM responses to Osps A and B are sometimes detected, that is because inflammation is capable of all that inducing expression of these Osps, this is IgM antibodies are not diagnostically useful. And so CDC does not recommend the use of IgM serology to diagnose any man of gestation of Lyme disease after one month of illness.

Misconception 5, it is sometimes said that the studies of the performance of serologic test use circular reasoning and are therefore invalid. Here is a brief review of how gold standard serum samples have been gathered for test evaluation. The problem of course is that the clinical signs and symptoms of Lyme disease after the first weeks of infection are not unique to this illness. Diagnosis cannot be confidently made without laboratory testing, where test evaluations are previous positive result should not be the basis or inclusion of the serum in the study.

So how do you deal with that problem? It’s quite readily addressed with proper technical skill in the early phase of Lyme disease. Erythema migrans lesions consistently yield, cultures of Borrelia burgdorferi when cultured in -- in vitro. This option unfortunately is not available in later disease and the extracutaneous manifestations of Lyme disease are seldom culture positive. Therefore, we look to the antecedent clinical science of Lyme disease to defy – clinically define populations of illness.

This is because – this is possible because of the well known natural history of Lyme disease. Facial paralysis for example is associated with erythema migrans in 72% to 87% of patients. Lyme carditis is seen with previous or concurrent EM in 80% of patients and sometimes, they also have neurologic disease and late neurologic disease, a rare condition may have antecedent EM, early neurologic disease and/or Lyme arthritis. In our 2003 report of late neurologic disease, 11 of 11 patients were seropositive. They were included in the study on the basis of their antecedent clinical findings.

And the last misconception that I have time to address today, laboratory testing in patients without objective signs of Lyme disease or history of exposure of infected vector ticks is medically useful. So far, I’ve focused on sensitivity and specificity but as everyone knows, it is the predictive value of a test that is clinically relevant. I’ll illustrate this with first, a situation where there is appropriate pretest probability. [indiscernible] [0:17:07] you in the audience may not be able to read all the terms of this illustration but I hope you’ll be able to look at it on the printed material.

In this situation, we presume an objective intermittent swelling of a knee, an individual who lives in an endemic area and the pretest likelihood of Lyme disease in this circumstance is about 50%. People trained in the interpretation of diagnostic data will immediately recognize this way of calculating predicted values. I won’t go through in any detail but the punch line if you will is that the test has an excellent predictive value in this circumstance, 90% of the test positives are true positives.

If you use the same excellent test, 99% sensitivity and specificity assumed in this illustration but apply it in a population with very low pretest probability, you get a situation where the test result is not clinically useful. It is necessary to analyze a larger population, a million people, in this illustration when a pretest probability is very low. So at your leisure, please look through the Math which illustrates that when testing is done inappropriately, it’s not clinically useful.

I’ll wrapped by trying to pull together some of these ideas. Misdiagnoses the patients with the low pretest probability of Lyme disease can occur if a healthcare provider over estimates the pretest likelihood, orders laboratory testing, and then may test with invalidated or sub-optical methods. Upon receiving a positive result, may mistakenly have high confidence that the individual has Borrelia burgdorferi infection and then adapted you with the clinical features and course of Lyme disease that are not evidence-based and so we have ourselves reinforcing circle here.

I think it’s sometimes difficult to appreciate this cycle of misdiagnosis because of some of the widely circulated misconceptions about lab testing and people may – except on critically that a test that is positive is therefore better because it’s so called more sensitive. So in conclusion, CDC recommends the appropriate use of two-tiered serology to support the diagnosis of Lyme disease, a physician that is endorsed by the IDSA and its practice guidelines.

Dr. Baker: Thank you very much. While the panels are thinking other questions for you, Dr. Johnson, I have a question based on an analogy. Many years ago, in the Houston Public School, they did routine PPD testing for tuberculosis screening.

Dr. Barbara Johnson: Yes.

Dr. Baker: And in this particular private school, the likelihood of having tuberculosis in positive 50 was very low, so the pretest probability analogy was very low. And we ended up with -- by the criteria at that time false positive test. So my question with regard to testing is in the non-endemic Lyme areas and if physician would order 2-tiered testing for a patient who hadn’t left that particular region so the exposure was only in a non-endemic area. What are State Health Departments do testing or send testing onto the CDC? You know what the recommendation is?

Dr. Barbara Johnson: Well, the recommendation of CDC, which is based on this study of an export review panel of The American College of Pathologist is that the pretest probability should fall between .2 and .8. Above point 8 which only occurs in the case of erythema migrans, diagnostic testing doesn’t add value. Below .2, the probability of false positives exceeds true positives.

The illustrations that I presented represented some extremes of this discussion, but even if you work to presume the pretest probability, there will be 100-fold higher than in the case that I illustrated, there would still be more -- there would not be more true positives than false positives. So we can entertain some various views about what the pretest probability is in a given circumstance, but it definitely does not serve the patient’s interest to test people where the pretest likelihood is very low.

Dr. Baker: Thank you very much. The reason I gave you the tuberculosis screening example is that resulted in some unnecessary treatments, some unnecessary adverse effects and based on the pretest probability criteria we recommended in school screening not be done in that situations. So parallel --

Dr. Barbara Johnson: I have a lot of empathy for patients who say I’ve been sick for years and besides it’s been proven that I have Lyme disease because of such and such a test, and I think this the knob of what we’re talking about here today and we have to break through the cycle.

Dr. Baker: Well, I completely agree with you and I would speak for members of the panel, is that we very much are interested. Our key objectives in doing our review and making a report ultimately after getting a diverse testimony of opinions and written material is to find the best diagnostic capability and best treatment for these patients who are suffering as well as the family sort of they are living with. Some questions from the panel. Yes, David.

Dr. Mushatt: Thank you, Dr. Johnson, for an excellent overview. Given the mirky and really podium manifestations of later forms of Lyme disease, I think we’d acknowledge that that is probably the case. I’m looking at your excellent summary side of the ELISA. The ELISA sensitivities can be very important in those later forms. I’m a little bit disappointed to see the small numbers of individuals of early and late neurologic disease, there are about 26 individuals that constituted and they gave rise to the finding of this sensitivity 100%. That’s not a lot today’s inclusions.

Dr. Barbara Johnson: Sure and I can care with you.

Dr. Mushatt: These are much more?

Dr. Barbara Johnson: There are two reasons for that. The first is that Lyme disease is most commonly seeing in its early stages and most of the samples are from that stage that numerically well-represented in the population. The second is that we try to evaluate goals standards samples which I find and they’re not so easily obtained. And so yes, the sample sizes are small and there are other studies with, you know, similar sizes that might surprise you that this study [indiscernible] [0:24:32] all in 2003, that altogether had more than 800 samples was the most robust study of these questions in the literature. So I acknowledge the limitations and if you’re going to study something that is thought to be rare as the idea say, a clinical review – clinical guidelines panel reported then having robust numbers is pretty much impossible.

Dr. Baker: Other questions? Dr. Sanders?

Dr. Sanders: I know it wasn’t specific in your talk but can you make any comments on other testing modalities that are frequently used?

Dr. Barbara Johnson: I focused on two-tier testing using whole-cell bacteria because that was the most commonly used at the time the Guidelines were developed. There is a good evidence that recombinant antigen known as VlsE and particularly a fragment of it expressed as a synthetic peptide, is an excellent first-tier test and if both the whole molecule and a synthetic peptide had been approved by the FDA as a first-tier test and to try and get more indirectly to your question study is in the manuscript of the study that has been completed about the use of C6 as a Stand-Alone test is in preparation.

So, that is one aspect and then there are efforts to understand if there are other important immuno-dominant antigens that could add value to the VlsE and Dr. Alan Barber’s group has been doing a comprehensive proteomic analysis of all the open reading frames and Borrelia burgdorferi and it maybe that that work will bear fruit to add value, but that has not been established so far.

Dr. Baker: Thank you. Dr. Parsonnet.

Dr. Parsonnet: Thank you very much. I have a question. In that written text that you distributed, you referred an American College physician paper from 1997 about prior probability which is really all important in doing any diagnostic test. And you say that the ACP panel members pointed out that the patient have only nonspecific science and symptoms of illness such as headache, fatigue, muscular joint pains, so not including a rash. Even when they reside in a geographic area endemic for Lyme disease, have a free pretest probability less than .2.

That paper that you referred to is now 12 years old and it it’s just an anecdotal experience but I’ll tell you I have seen many patients this summer who have not had erythema migran but had clearly have Lyme disease and have seroconverted. I wonder what would happen if that range instead of being .2 to .8 was .1 to .8 or whether there should be a reconsideration of this prior probability range for appropriate testing.

Dr. Barbara Johnson: I think you raised a very useful point. I think estimates of prior probability should be informed by Ticicology where infected vectors can be found. It can be informed by probably new research to actually interrogate the general population and do a study of heavy reprehended diagnosis of Lyme disease so that we can be get more maybe count the specific data that to my knowledge is that kind of data is not available right now.

Dr. Baker: Dr. Lantos.

Dr. Lantos: Are you aware of any data that compares the either tests rather serological criteria, the seroprevalence against Lyme and asymptomatic unfitted patients and how that compares the patients who have chronic symptoms that have been attributed to Lyme disease?

Dr. Barbara Johnson: Okay, I needed a little help here. Asymptomatic patients –

Dr. Lantos: In other words, in a particular geographic area to asymptomatic patients who will have a certain degree of seropositivity, what is the rate of that and does that -- is that comparable to the rate of seropositivity among the patients who have chronic symptoms attributed to Lyme disease?

Dr. Barbara Johnson: Okay, the rate of asymptomatic seroconversion was estimated from the vaccine study with the Smith -- GlaxoSmithKline preparation and there was an asymptomatic seroconversion rate of about 11% if I can recall correctly. And those people were followed, [indiscernible] [0:29:34] say exactly but my memory is maybe 6 months and they did not develop other manifestations of Lyme disease that they’d seen not to have clinical science and symptoms, and then the estimation of the second point was –

Dr. Lantos: Is to have rate of asymptomatic seropositivity comparable to the rate of seropositivity among people who have chronic symptoms?

Dr. Barbara Johnson: Seropositivity among people who are chronic symptoms, okay. So to answer your question, one must define the clinical stage of illness that they achieved prior to treatment. So, for example, on the Klempner study, there was a seronegative group. Who were they? They were people who had erythema migrans that were treated early. They did not become seropositive.

There are some evidence that there’s aggregation of class switch and failure to develop an immune response and treated the young people patients so nevertheless they developed a chronic syndrome. So, there’s that situation and then there is a situation of people who are -- some people contend have chronic encephalopathy, encephalitis, clinical manifestations that in terminology of the idea say are late neurologic disease, and these people the data suggest had been infected for a long time and have had time to develop antibodies, so essentially all of those people would be seropositive.

Dr. Baker: One final question. There is one? Yes, Dr. Charini.

Dr. Charini: Yeah, getting back to that pretest probability issue, in the where I live, there’s– one of the first questions that I asked the patients is what county they come from because that all, you know, I mean what their answers is that will affect my estimation of, you now, with their pretest probability of having the diseases, but given that Lyme disease spreading, spreading in land from the coast down the Eastern sea border and other parts of the United States, they will be very useful I think to a lot of physician in other parts of the country, other than Massachusetts, to have some sort of means of, you know, monitoring the problems of the disease in their areas in a more up-to-date basis and have that easily accessible any kind – what does the CDC do besides publishing case -- number of reported cases today?

Dr. Barbara Johnson: So there are two things CDC does. The first, as you well know, surveillance, and surveillance is useful for looking through, it trends their time to look at where something it occurs. There has been a lot of controversy about does it reflect in the true disease burden?

Well, we generally do other studies to look at the degree of under or over reporting to get an estimate of the true disease burden. The other things CDC does besides surveillance is support research in the field to identify areas that have infected vector ticks and Dr. Fish has been a recipient in one of those awards and has extensive data on the distribution of infected ixodes ticks in the United States. And so if I’m understanding your suggestion, maybe both an up-to-date tick map of infected vectors overlaid with the latest data on human disease could be useful to clinicians.

Dr. Baker: Thank you very much.