ELISA TESTING and the coin flip analogy

Tom Grier is a microbiologist with a long time interest in Lyme
disease, he has written extensively about it, lectured and led a
support group and been on the board of directors of the Lyme Alliance
of Michigan (I believe the group is now defunct).

In the following article which I have provided full text for he says
what I excerpt below:

We are told by manufacturers, health departments and clinics that the
Lyme ELISA tests are good, useful tests, but in two blinded studies
that tested laboratories for accuracy, they failed miserably. Lorie
Bakken, MS/MPH, showed in her studies that there was not only
inaccuracy and inconsistency between competing laboratories, but also
between identical triple samples sent to the same lab. In other words,
identical samples often resulted in different results! In the first
study, forty-five labs correctly identified the samples only 55% of
the time.

In the latest study by the College of American Pathologists, 516 labs
were tested. The overall result was terrible! There were almost equal
numbers of false positives as false negatives. Overall, the labs were
55% inaccurate. The labs could only give a correct result 45% of the

Laboratory Tests
By Tom Grier M.S.
Three Main Categories of Lyme Disease Tests:

1. Indirect Tests (serum antibody tests):

ELISA; Western Blot; IFA; Borreliacidal Antibody Assay (Gunderson
test);T-cell Activation Test

2. Direct detection tests:

PCR (DNA amplification); Lyme Urine Antigen Test (LUAT); Antigen
Capture Test; culturing of skin, blood, CSF, urine, or tissue; immune
complex / antigen-antibody test

3. Tissue Biopsy and Staining:

Silver Stain; Gold Stain; Fluorescent Tagged Monoclonal Antibody
Stains; Acrodine Orange; Gram Stain; Muramidase; etc.

There is a great deal of confusion and controversy surrounding Lyme
disease testing. The first problem is that most of the manufacturers
of these tests want you to believe that their tests are the best. At
every medical convention, I listen to sales pitch after sales pitch
from sales people making their product sound infallible. Often the
terminology is confusing and the customer frequently misinterprets
what is really being said.

For example, a salesman may say the rate of false positive or false
negative is less than one percent. This sounds like the test is more
than 99% accurate. In reality, what it is saying is if you have 1000
test samples from the same known laboratory sample, then in less than
ten samples will there be a result that differs significantly from the
other 990.

In any of this, did you hear the words: "percent reliability" or
"percent accuracy" in diagnosing Lyme disease in humans? No! People
often mistake "false positive rate" for accuracy. The truth is that no
Lyme disease test to date is close to 100% accurate, because each test
has its own particular set of shortcomings. So, while the first
problem with Lyme disease tests is in the way they are promoted, the
second problem is the way the tests are primed to recognize laboratory
strains of Bb, rather than wild types. Third, the Lyme spirochete can
hide in the human body, and fool the immune system into thinking it
isn't there. So, no antibodies are produced, resulting in negative
tests. Stealth technology isn't new, it evolved millions of years ago
by the first bacteria that evaded its host's defenses.

Immune Responses

The first antibody our body makes in response to a foreign invader is
usually immunoglobulin type M, abbreviated as IgM. This large antibody
takes two to four weeks to be made in quantities large enough to be
consistently measured. It is at its peak of production four weeks
after exposure to an antigen. The IgM antibody will only stay in
circulation for about six months, and then levels are usually too low
to detect. If infection persists, this antibody may also persist. In
general, a Lyme patient who consistently has detectable IgM levels is
usually chronically ill, but its absence is not a reliable indicator
of cure.

The second antibody we make after the IgM is the IgG antibody. This
antibody takes four to eight weeks to form, and is gone in less than
twelve months. It peaks at about six weeks. This antibody crosses the
placenta, so an infected mother can pass this antibody to her child.
An IgG antibody titer in a newborn does not have to mean active
infection. It does mean the mother has had exposure, and the child
must be carefully monitored for signs of the disease.

Because of the difference in the two antibodies, two separate tests
are available to test for their presence. Therefore, a physician must
specify whether or not a patient should have an IgM or IgG Western
Blot, or an IgM or IgG ELISA test.


This is the earliest of the antibodies to appear in response to an
infection. It is produced in quantity. It is six times larger than the
IgG antibody. Because of its size, this immunoglobulin does not cross
the placenta. Since it cannot enter the fetus from the mother, any
newborn that starts to make IgM antibodies against Lyme disease must
be infected. However, a fetus exposed to Borrelia burgdorferi early in
the pregnancy may never make an antibody response to the Lyme bacteria
because the baby's immune system doesn't recognize it as foreign.


This antibody remains the longest and is the foot soldier of the
immune system. It attacks viruses, bacteria, yeast, toxins, and
transplants. The IgG antibody can kill bacteria indirectly by tagging
or marking the foreign invaders for destruction by the killer cells (T-
cells, macrophage). Or, it can kill the bacteria directly by evoking
compliment, a series of enzymes and proteins that will dissolve the

Note: It was once thought that plasma cells could produce antibodies
that could conform to any shape necessary to attack foreign intruders.
If this were true, we would have almost unlimited immunity. It is now
thought that each person has a finite collection of specialized
lymphocytes that are able to create a finite number of antibodies.
Each antibody shape is predetermined, and can be produced by only one
type of lymphocyte. When the body is invaded by a foreign antigen, it
will stimulate one of these cells, and only that cell will begin to
clone itself. This process takes several weeks. If we lack the right
cell type to do the job, we are left with a gap in our immunity. This
might account for why some Lyme patients with certain tissue types
have greater morbidity, while others have relatively mild symptoms.

Dr. Alan Steere, M.D., observed that Lyme arthritis patients with
tissue type HLA-DR2 and HLA-DR4 had more severe arthritis and chronic
disease. Other tissue types have been associated with an increased
incidence of multiple sclerosis and other neurological diseases. It
might be that different patient tissue types might account for a
difference in patient's symptoms to a greater degree than different
strains of the bacteria.

It is known that this bacteria has an affinity for specific tissues.
If you have a specific lack of immunity, this may cause the disease to
manifest differently in those tissues. For example, let's say
hypothetically that your heart is infected with Borrelia burgdorferi
bacteria. Perhaps most people make an antibody that suppresses
attachment of Bb to certain fibers in the heart. If you lack that
antibody, the infection may continue more aggressively and manifest
differently - for instance, causing an enlargement of the muscle
fibers or destruction of the conduction pathways.

Instead of lacking a specific antibody, perhaps some individuals make
a different kind of antibody, an antibody that not only attacks the
bacteria - but may attack the heart as well! It is well known and
documented that some patients produce auto-antibodies, which are
antibodies that our own body produces that attack our own tissues.
This is the basis of autoimmune disease. In some Lyme disease
patients, an auto-antibody against cardiolipin has been clearly
established in Lyme patients with Lyme carditis.

Perhaps, in addition to other Lyme tests, we should also be tissue
typing patients and searching for auto-antibodies? Tissue typing
requires a small blood sample, and costs about $200.

Western Blot

The Western Blot essentially makes a map of the different antibodies
the immune system produces to the bacteria. The map separates the
antibodies by the weight of their respective antigens and are reported
in units called kilo daltons or kDa. For example, a Western Blot may
report bands at 22, 23, 25, 31, 34, 39, and 41 kDa. Each of these
bands represents an antibody response to a specific protein found on
the spirochete. The 41 band indicates an antibody to the flagella 41
kDa protein and is nonspecific. The 31 kDa band represents the OSPA
protein and is specific for just a few species of Borrelia, as is the
34 band OSPB, and 23 kDa OSPC.

In 1994, the Association of State and Territorial Public Health
Laboratory Directors, under a CDC grant, decided that there should be
consistency between labs reporting Lyme disease Western Blots, and
that a specific reporting criteria should be established. The
consensus committe, chaired by Dr. Michael Osterholm, Ph.D., MN, set
nationwide standards for Western Blot reporting. This sounds good, but
one could argue they made a bad situation worse. Prior to the hearing,
virtually every lab had accepted bands 22, 23, 25, 31, and 34 kDa as
specific and significant, and reported them as positive for exposure
to Borrelia burgdorferi. Not only are these bands specific for
Borrelia species, but they represent all of the major outer surface
proteins being used to develop the Lyme vaccines. The committee,
without any clear reasoning, disqualified those bands as even being

After the consensus meeting, those bands were no longer acceptable.
The result was that what had been a fair-to-good test for detecting
Lyme disease had now become poor, arguably useless. Many scientists
have questioned these new reporting criteria, and several wrote
letters of protest to both the committee and to laboratory journals.
Many labs stopped reporting the actual bands and instead, simply
reported the test as positive or negative, thus preventing any further
interpretations. (90)

How badly did the Lab Directors bootstrap this test? The following is
an analysis of the new guidelines presented as an abstract and lecture
at the 1995 Rheumatology Conference in Texas, chaired by Dr. Alan
Steere, MD. (1995 Rheumatology Symposia Abstract #1254, Dr. Paul
Fawcett, et al.)

This was a study designed to test the recently proposed changes to
Western Blot interpretation by the Second National Conference on
Serological Testing for Lyme Disease, sponsored by the CDC. The
committee proposed limiting the bands that could be reported in a
Western Blot for diagnosis of Lyme disease. Out of a possible 25
bands, 10 specific bands were selected as being reportable. An lgG
Western Blot must have five or more of these bands: 18, 21,28, 30, 39,
41,,45, 58, 66 and 93 kDa. An lgM Western Blot must have two or more
of the following three bands: 23, 39, 41.

Conspicuously absent are the most important bands, 22, 23, 25, 31, and
34, which include OSPA, OSP-B and OSP-C antigens - the three most
widely accepted and recognized Bb antigens. These antigens were the
antigens chosen for human vaccine trials. This abstract showed that,
under the old criteria, all of 66 pediatric patients with a history of
a tick bite and bull's-eye rash who were symptomatic were accepted as
positive under the old Western Blot interpretation.

Under the newly proposed criteria, only 20 were now considered
positive. (The number of false positives under both criteria was zero
percent.) That means 46 children who were all symptomatic would
probably be denied treatment! That's a success rate of only 31%.

*Note: A misconception about Western Blots is that they have as many
false positives as false negatives. This is not true. False positives
based on species specific bands are rare.

The conclusion of the researchers was: "the proposed Western Blot
reporting criteria are grossly inadequate, because it excluded 69% of
the infected children."

Elisa Test

The Enzyme-Linked Immunosorbant Serum Assay is the simplest, least
expensive, easiest to perform, and most common Lyme test ordered. It
is a test based on detecting the antibodies that our bodies make in
response to being exposed to Borrelia burgdorferi (Bb). It is a
preferred test by laboratories, not because it is more accurate than
other Lyme tests, but because it is automated. Many different patient
samples can be performed by a single machine simultaneously. This
allows for a faster turnover, less costs, and theoretically,
standardized test results that are consistent from lab to lab.

We are told by manufacturers, health departments and clinics that the
Lyme ELISA tests are good, useful tests, but in two blinded studies
that tested laboratories for accuracy, they failed miserably. Lorie
Bakken, MS/MPH, showed in her studies that there was not only
inaccuracy and inconsistency between competing laboratories, but also
between identical triple samples sent to the same lab. In other words,
identical samples often resulted in different results! In the first
study, forty-five labs correctly identified the samples only 55% of
the time.

In the latest study by the College of American Pathologists, 516 labs
were tested. The overall result was terrible! There were almost equal
numbers of false positives as false negatives. Overall, the labs were
55% inaccurate. The labs could only give a correct result 45% of the
time. You are actually better off to flip a coin!

The basis of the ELISA test is that it can be primed to be very
specific for particular antibodies. This is done by taking a
laboratory sample of the Lyme bacteria and breaking the sample down
into fragments. These fragments, or antigens, are then embedded on the
side of a reagent vessel like a test tube. Then the patient's serum is
added, and any free (non-complexed) antibodies specific for the test
strain will then bind to the antigens, which are linked to special
enzymes that will change color when antibodies are present. The sample
is continually diluted until the reaction no longer occurs and no
color change can be detected. The sample is then reported as a
dilution ratio, such as one part serum to 256 parts water, or 1:256.

The ELISA test sounds simple and straight forward, but it has a couple
of major flaws. Borrelia species are some of the most polymorphic
bacteria known to exist. In other words, most Borrelia species can
significantly change its surface proteins enough during cell division
as to evade our immune system, and may differ from laboratory strains
enough to result in negative tests, even if antiBb antibodies are
present! In Europe, this problem is intensified because they have
recognized three species of Borrelia that cause Lyme disease, and so
they have available three separate ELISA tests. The questions in
America are: 1) Have we recognized all the strains and species of
Borrelia that cause Lyme disease symptoms, and 2) are we incorporating
them into our tests? The answer is no. Convenience and expedience has
chosen that we don't prime our ELISA tests withwild strains, but use a
laboratory strain.

When a lab reports that their ELISA test has had high specificity and
high sensitivity, it is usually interpreted by doctors as being a more
accurate test, but the doctors don't know what the lab is actually
measuring. One of the hidden problems of serologic Lyme tests is the
fact that the tests must be primed with a source of bacteria to create
the reactions with the patient's antibodies. To do this, virtually all
labs rely on a laboratory strain of Bb known as strain B-31.Taking
purified antigens from strain B-31 and injecting them into mice, they
then can extract a monoclonal antibody to each antigen, or a
polyvalent antibody soup. This antibody is concentrated and purified,
and then added to the ELISA test to test the efficacy and performance
of the test. Unlike the wild strains, B-31 grows well in culture, and
this makes it a perfect choice as a consistent and inexpensive source
of Bb. But the affinity the mouse monoclonal antibody has to B-31
antigen is quite different from the affinity the patients' antibodies
have to the same antigen. This means the test may register as negative
because the test cannot detect the slightly different antibody profile
that a wild strain of Bb can produce. In other words, the labs are
really comparing apples to oranges! This is why, when the American
College of Pathologists used human sera to test the accuracy of 516
different laboratories ELISA tests nation wide, the overall accuracy
was only 45%.

In the quest for specificity, most ELISA tests have become so specific
that the test may fail to detect antibodies from related strains of
Borrelia. This would include different genospecies that cause Lyme
disease, as well as different Borrelia species that cause Tickborne
Relapsing Fever. Would a cross reaction to the Borrelia species that
cause Tick-borne Relapsing Fever be so bad?

The real Achilles' Heal of an ELISA Test is that it can only detect
free antibody. It cannot detect any antibody that has become complexed
with antigen.

The ELISA test depends on the active, free antibodies to attach to the
free antigens that have been embedded on the walls of the test tube.
If the antibodies in the serum being tested are already attached to
antigens, then the enzyme reaction cannot take place. If we think of
antibodies as sort of keys that fit into locks, and that on the
surface of the bacteria are specific locks we now call antigens, you
can see that once a key is inserted into a lock, the key is no longer
available to open any other locks.

What makes this test so misleading is that many doctors accept high
readings as an indication that the patient must really be sick. This
logic is exactly backwards. If a patient is really infected with lots
of bacteria, that means they have a lot of bacterial antigens floating
around in the blood that are complexing free antibodies. So, as free
antigen increases, free antibody decreases. Since the ELISA test
detects only free antibody, a negative test might actually indicate a
more serious infection. Many times, I have seen totally asymptotic
patients with ELISA titers over 1000 be treated as though they were on
death's doorstep simply because they had a high titer, while patients
with borderline titers who are practically disabled are ignored,
because a low titer is perceived as meaning less infected! These
conclusions are erroneous and actually opposite to the truth, which is
that a high titer means greater natural immunity.

This phenomena can actually be observed by using vaccines. If a
patient has been vaccinated for a disease like tetanus, they will
carry a high titer of free antibodies. If you try to measure those
antibodies an hour after a booster shot is given, they will test
negative. This is because the injected tetanus antigen complexes all
available free antibody before the body can make more, so the
measurable free antibody level drops.

The nature of all antibody is to seek out the proper antigen. The
level of free antibody available is variable and often inadequate for
the amount of antigen available. As antigen increases (i.e. The
bacteria are dividing faster than the immune system can handle), free
antibody drops.

What a high ELISA test may be a better indicator of is what level of
immunity is the patient capable of mounting against this infection? A
high titer is the same thing as saying the patient has a high natural
immunity, and a low can mean that the patient may be overwhelmed by

In one year-long study by Dr. Sam Donta, MD, done on chronic Lyme
patients, the initial ELISA tests proved to be more than 66+%
inaccurate (1996 LDF Conference lecture). Other researchers have also
found the ELISA tests to be inaccurate. Using a 45-panel diagnostic
testing protocol from the NIH for testing the efficacy of the ELISA
and Western Blot, researchers found the accuracy of the Lyme ELISA
varied from about 5075%, and were routinely inconsistent. The CDC's
ELISA test did no better on average than any other ELISA. It is the
CDC ELISA test which is used for surveillance of emerging Lyme disease
in the United States, yet the test was correct only about two out
every three tests. Too often, a single negative ELISA test can prevent
a sick patient from getting treatment, even despite having serious

In my opinion, the ELISA test is worthless as a diagnostic tool in
Lyme disease. It is inconsistent and inaccurate, and should be
discontinued as a tool to diagnose Lyme. If the NIH and CDC truly
believe, as they've stated, that the diagnosis of Lyme disease is to
be made on the basis of symptoms, then these tests should be
temporarily banned until each manufacturer can prove efficacy using
human serum.