How Immunoassays Work: The Curious Case of AIDS Denialist Roberto Giraldo and his Ignorance of the Basics

by John P. Moore, PhD

Roberto Giraldo is employed as a laboratory technologist in the molecular diagnostics laboratory at the New York Presbyterian Hospital, New York. He is not licensed to practice medicine in New York. He has no medical appointment at the Hospital. Again, Giraldo is employed at the Hospital as a technician, not as a medical doctor or a professional academic. Giraldo has no appointment at the Weill Medical College of Cornell University, an institution affiliated to but separate from the Hospital.

On his website Giraldo writes:

The ELISA test is a test for antibodies against what is supposed to be the Human Immunodeficiency Virus or HIV. To run this test, an individual's serum has to be diluted to a ratio of 1:400 with a special specimen diluent. According to the test kit manufacturer this diluent contains "0.1% triton X-100, Bovine and Goat Sera (minimum concentration of 5%) and Human T-Lymphocyte Lysate (minimum titer 1:7500). Preservative: 0.1% Sodium Azide" (1).

I first took samples of blood that, at 1:400 dilution, tested negative for antibodies to HIV. I then ran the exact same serum samples through the test again, but this time without diluting them. Tested straight, they all came positive.

Since that time I have run about 100 specimens and have always gotten the same result. I even ran my own blood which at 1:400 reacts negative. At 1:1 [undiluted] it reacted positive. I should mention that with the exception of my own blood, the patient samples all came from doctors who requested HIV tests. It is therefore likely that most of the blood samples that I tested belonged to individuals at risk for AIDS.

In this article, often parroted on other denialist websites, Giraldo displays scientific ignorance of truly staggering proportions for someone who states: "For the last 6 years I have been working at a laboratory of clinical immunology in one of the most prestigious University Hospitals in the City of New York."

What is an Enzyme-Linked Immunosorbent Assay (ELISA)?

In an ELISA, the HIV antigen source (originally detergent solubilized virus, nowadays purified recombinant HIV proteins or peptides) is coated onto a plastic surface, usually the bottom of a small well (of which there are traditionally 96) on a "microtiter plate". The unbound viral proteins are then washed away and any non-specific binding sites for antibodies on the wells are blocked with an irrelevant protein. A solution containing human serum is then added, diluted in a buffer containing a high concentration of unrelated proteins to further reduce non-specific binding of the serum antibodies to sites the ELISA wells. Hence the viral antibodies bind specifically to the viral antigens present on the filter, and not elsewhere. Any HIV-specific antibodies that are present in the solution bind to the HIV antigens on the surface of the well. The excess, unbound antibodies are washed away, and a labeled detection antibody is added. This is an antibody specific for human antibodies (for example, an antibody raised in rabbits against human IgG, then chemically labeled with an enzyme). The excess detection antibody is then removed by washing, so the only detection antibodies remaining attached to the membrane are those specifically bound to the HIV proteins via the intermediary of the human serum antibodies. In other words, a tri-partite layer is built up on the ELISA plate: HIV antigen, then serum anti-HIV antibody, and finally the labeled detection antibody, which is why reference is often made to a "sandwich" ELISA. The labeled antibody is then detected and quantified. When an enzyme labeled antibody is used, a solution is applied that causes a color change in the well in the presence of the enzyme. In an early variant of the ELISA, the radioimmunoassay or RIA, the detection antibody was labeled with a radioactive tracer, and the amount of bound radioactivity was quantified using radiation detectors. The final signal is a measurement of the presence of HIV-specific antibodies in the test serum; performing titrations (serial dilutions) allows the titer (a measure of the amount) of the antibodies to be estimated, if and when this is desirable.

What is a Western blot assay?

Although Giraldo does not refer to the Western blot assay, allegations of its non-specificity are sometimes made on other denialist web-sites. In the Western blot assay, antigens derived from a detergent-solubilized preparation of HIV are fractionated by electrophoresis on a gel (usually made from polyacrylamide). Under the experimental conditions most commonly used, the distances the different proteins migrate on the gel are proportional to their molecular weight, so the HIV proteins are separated by size. The separated proteins are then transferred onto a membrane made of nitrocellulose (more recently, of nylon) by a "blotting" procedure using high salt concentrations and/or an electric current. The membrane is then mixed with a solution containing antibodies (e.g., diluted serum from an infected or at-risk individual), along with high concentrations of unrelated proteins that occupy non-specific binding sites on the membrane to which proteins bind adventitiously. Hence the viral antibodies bind specifically to the HIV antigens present on the filter, and not elsewhere. The membrane is then thoroughly washed to remove unbound antibodies, and treated with a solution that contains a labeled antibody specific for human antibodies (for example, an antibody raised in rabbits against human IgG, then labeled with a radioactive tracer or an enzyme). The excess detection antibody is then removed by washing, so the only detection antibodies remaining attached to the membrane are those specifically bound to the viral proteins via the intermediary of the human serum antibodies. In other words, a tri-partite layer is built up on the membrane: HIV antigen, then serum anti-HIV antibody, and finally the labeled detection antibody, which is why reference is often made to a "sandwich" technique. The labeled antibody is then detected and quantified. In the case of a radioactive tracer, the membrane is exposed to a photographic film; for an enzyme label, a solution is applied that causes a colored compound to be deposited on the membrane in the vicinity of the enzyme. Overall, then, a picture is created of the HIV antigens with which serum antibodies are reactive.

Giraldo's folly

I first took samples of blood that, at 1:400 dilution, tested negative for antibodies to HIV. I then ran the exact same serum samples through the test again, but this time without diluting them. Tested straight, they all came positive.

When too much serum is used in any immunoassay (ELISA, RIA Western blot), there is a false positive reaction. The point with having a standard protocol for operating a diagnostic test is that it should be followed; if it is not, the test is unreliable. This is not a particularly difficult concept to grasp: If a recipe for soup says that one should add a pinch of salt, and one adds a handful for the hell of it, one ends up with an inedible meal. If the refill instructions for a Zippo lighter say to "use lighter fuel" and one instead adds gasoline, one's eyebrows will be singed.

Why does adding too much serum matter? Well, as outlined above, serum antibodies can stick non-specifically to the surface onto which the test antigens (HIV proteins) are coated, the surface being either the plastic wells of the ELISA plate or the nylon or nitrocellulose membranes used in Western blots. To prevent this from happening, the ELISA plate surfaces or the membranes are routinely treated with a solution containing a high concentration of irrelevant carrier proteins (e.g., bovine serum albumin, or non-fat milk proteins) to try to block as many of the non-specific binding sites for proteins as possible. For the same reason, the buffer used to dilute the test serum contains a high protein concentration to reduce the chances of the human serum antibodies being themselves able to get access to the non-specific protein binding sites on the plastic surface or membrane. Often sheep, goat or bovine serum is used for this purpose, which is why the dilution buffer for the early generation ELISA that Giraldo refers to contains "Bovine and Goat Sera (minimum concentration of 5%)". The bovine or goat antibodies bind to the same non-specific sites as the human antibodies would, so they compete out the human antibodies. However, the detection antibodies are specific for human IgG, and do not efficiently react with the bovine or goat counterparts.

The extent of the non-specific absorption of the human antibodies to the plastic surface or membrane is dependent on the antibody concentration in the input sample; the more antibody is present, the more of it will bind non-specifically even in the presence of competing proteins. Optimization procedures are therefore performed when creating the assay, intended to identify the lowest serum dilution at which known negative control samples fail to give a signal in the assay. The 1:400 standard dilution in the test referred to by Giraldo is reasonable and typical, although the precise dilution will vary between assays of slightly different design. Human serum contains typically contains about 20 mg/ml of IgG, so a 1:400 dilution equates to about 50 ug/ml of IgG antibody in the test reaction. Of note is that if one takes any purified monoclonal antibody to an irrelevant, control antigen, and puts it into an HIV ELISA at around 50 ug/ml, it is likely to give a false positive signal for exactly the same reason: non-specific binding to the plastic surfaces of the assay wells.

Having established the optimum serum dilution that avoids false positive signals, test samples are therefore diluted to the same extent, in the knowledge that any resulting assay signal is not likely to be due to non-specific absorption of irrelevant antibodies to the plate surface. Of note is that truly HIV-positive serum samples can often be diluted by as much as 1:100,000, sometimes even more, and still give a positive reaction. This is because the serum antibodies to HIV react SPECIFICALLY with the HIV antigens that are coated onto the ELISA plate surface, or present on the western blot membrane.

None of this is rocket science; it is absolutely standard operating procedure in designing and using immunoassays, the kind of basic knowledge any professional immunologist or diagnostic specialist would learn in the first few days at work. It is simply shocking that a technician with the experience claimed by Giraldo would not know this.

John P. Moore, PhD, is Professor of Microbiology and Immunology, Weill Medical College of Cornell University, New York

For additional information on ELISA including an animation of the test, see the The University of Arizona's Biology Project website.

Definitions

antibody: an infection-fighting protein molecule in blood or secretory fluids that tags, neutralizes, and helps destroy pathogenic microorganisms (e.g., bacteria, viruses) or toxins. Antibodies, known generally as immunoglobulins, are made and secreted by B lymphocytes in response to stimulation by antigens. Each specific antibody binds only to the specific antigen that stimulated its production.

antigen: any substance that stimulates the immune system to produce antibodies. Antigens are often foreign substances such as invading bacteria or viruses.

binding antibody: an antibody that attaches to some part of HIV. Binding antibodies may or may not lead to the killing of the virus.

ELISA (enzyme-linked immunoabsorbent assay): a blood test that detects antibodies based on a reaction that leads to a detectable color change in the test tube. The HIV ELISA is commonly used as the initial screening test because it is relatively easy and inexpensive to perform. Because the HIV ELISA is designed for optimal sensitivity -- that is, it detects all persons with HIV antibodies as well as some who don't have them (false positives) -- a positive HIV ELISA test must be confirmed by a second, more specific test such as an HIV Western Blot.

IgG: a class of immunoglobulins that include the most common antibodies circulating in the blood, that facilitate the phagocytic destruction of microorganisms foreign to the body.

immunoassay: a technique or test (as the enzyme-linked immunosorbent assay) used to detect the presence or quantity of a substance (as a protein) based on its capacity to act as an antigen or antibody.

immunoglobulin: a general term for antibodies, which bind to invading organisms, leading to their destruction.There are five classes of immunoglobulins: IgA, IgG, IgM, IgD and IgE.

microtiter plate: a flat plate with multiple "wells" used as small test tubes. The microplate has become a standard tool in analytical research and clinical diagnostic testing laboratories. It typically has 6, 24, 96, 384 or even 1536 sample wells arranged in a 2:3 rectangular matrix.

peptide: any of various amides that are derived from two or more amino acids by combination of the amino group of one acid with the carboxyl group of another and are usually obtained by partial hydrolysis of proteins.

serum: the watery portion of an animal fluid remaining after coagulation: the clear yellowish fluid that remains from blood plasma after fibrinogen, prothrombin, and other clotting factors have been removed by clot formation -- called also blood serum.

ug/ml: micrograms per milliliter. A measure of an amount (in weight) per volume.

western blot: a blood test to detect antibodies to several specific components of a virus such as HIV. This test is most often used to confirm a positive ELISA.