Science fails the "Aids test"

When Bio/Technology's Dr Harvey Bialy told me of his conclusion that there was "absolutely no believable, persuasive evidence that Africa is in the midst of a new epidemic of infectious immuno-deficiency", my confidence in his view was boosted by an astonishing statement he had made about the HIV test. Bio/Technology had a paper in press, he said, which did more than highlight a problem with false positives: it challenged the very basis of the test as indicating the presence of a specific virus, HIV, arguing that it had never been validated against the accepted "gold standard" for a diagnostic test, isolation of the virus itself.

The implications were so enormous that I was once again nervous of the hostility such a revelation could arouse, and I did not pursue the story immediately. Indeed I found this conclusion hard to digest, simply by virtue of what it meant for the entire edifice of understanding that had been erected around HIV and Aids over the past 10 years. But during subsequent weeks, I studied the paper concerned and corresponded with the main author, Eleni Papadopulos-Eleopulos, a biophysicist at the Royal Perth Hospital. To my growing amazement I found that there was indeed a mass of evidence, pulled together in Eleopulos's enormous review article, that what had come to be called "the Aids test" was scientifically invalid. The proteins detected by the test kits were not specific to a unique retrovirus. Positive results were produced in people whose immune systems had been activated by a wide variety of conditions, including tuberculosis, multiple sclerosis, malaria, malnutrition, and even a course of flu jabs. Patients with Aids, and promiscuous gay men leading lives likely to expose their immune systems to multiple challenges, were certainly much more likely to test positive than healthy Americans, but for reasons that need not have anything to do with a deadly new virus.

The possible relevance of this paper for an understanding of Aids in Africa was particularly clear. African countries were those where the tests might be at their most meaningless, because of the widespread ill-health caused by malnutrition and associated chronic diseases. Had an entire continent been panicked by western scientists into believing it was in the grip of a deadly epidemic, on the basis of a test that had never been shown to be valid for the retrovirus whose presence it was claimed to detect? The paper's claims were of course also vitally important to people who had tested HIV-positive anywhere, for whatever reason.

The review article, which was the first of its kind and carried hundreds of references, was co-authored by Val Turner, a surgeon at the Royal Perth, and John Papadimitriou, professor of pathology at the University of Western Australia. It was headed, "Is a Positive Western Blot Proof of HIV Infection?" To understand the significance of the challenge that was being mounted, one first needs to distinguish between the two main types of HIV test in use, called western blot and Elisa. Both purport to show whether or not a person has been infected by HIV on the basis of detecting the presence in their blood of antibodies to various components of the virus.

The western blot test employs a selection of proteins thought to belong to the virus, separated and immobilised by the test's manufacturer, usually on a strip of nitrocellulose paper. The strip is then incubated with a dilution of the blood or blood components to be tested. If antibodies to those particular virus proteins are present in the blood, they bind at the appropriate points on the surface of the nitrocellulose strip. Unbound materials are washed away, and a colouring agent stains the strip where antibody is present, producing characteristic bands indicating the presence of the viral proteins. Because this test is thought to identify individual proteins that are specific to the virus - that is, which only HIV produces - it is widely believed to be so accurate as to mean that a positive result is synonymous with HIV infection.

The other main type of antibody test used, called Elisa (enzyme-linked immunosorbent assay), again uses a variety of proteins thought to be representative of HIV, and bound to a solid support, but as a group rather than separated from one another as in the western blot. The original tests of this kind were made from "whole virus lysate" - that is, from materials obtained from cell cultures thought to be infected with and growing HIV. It is now accepted (although I knew nothing of this before reading the article) that these materials often contained normal cell proteins as well, which confounded the test results, producing large numbers of positive results even when the person was not actually infected with HIV (false positives).

Later generations of Elisa tests, made with proteins manufactured synthetically or by genetic engineering techniques, have been claimed to be more reliable. Nevertheless, the Elisa test is still generally considered more prone to produce false positives than the western blot. In the US, for example, the Centres for Disease Control designate the Elisa a "screening" test, to detect suspicious blood samples, rather than as a "confirmatory" test, the role usually given to western blot. The CDC state in guidance issued in 1993 that Elisa results "should never be used alone to report a final positive result". Instead, if two or more blood tests are "reactive" on Elisa, "then the results are 'confirmed' using a second more specific antibody test such as the western blot." This more specific second test can tell the difference, the CDC says, between HIV antibodies, and other, non-HIV antibodies that react to the Elisa. "Results from this two-part testing are greater than 99 per cent accurate." Today in the US, people are only told they are positive if they are repeatedly reactive by Elisa, and are confirmed by western blot (there are other test systems, but these two are the ones in widespread use).

The significance of the Bio/Technology review article, which was published in June 1993, was that it not only presented evidence confirming the Elisa test's unreliability, but demonstrated that the western blot was also incapable of determining whether people were really infected with HIV. Yet as the authors said, "A positive HIV status has such profound implications that no one should be required to bear this burden without solid guarantees of the verity of the test and its interpretation."

To show that an antibody test for HIV is scientifically valid and reliable, the paper said, requires four steps. The first of these is to identify a source of HIV-specific antigens - the protein components of the virus to which antibodies bind. Here, one of the first surprises is that because HIV is extremely difficult - and perhaps impossible - to isolate in a clear-cut way, there is no guarantee that the method used really does obtain the virus or its components. I shall be discussing these problems of isolation in a later chapter, but for the moment suffice it to say that the manufacturers of the tests do not have an unequivocal collection of HIV viruses, visible through electron microscopy, which can then be broken down into their various components. Instead, a multi-step procedure has to be followed involving a variety of assumptions, each of which is questionable. The final assumption is that some material which bands at a particular density (1.16 grams per millilitre) when spun in a centrifuge represents "pure" HIV proteins from which to make the antibody test, or which can serve as a template from which to manufacture the proteins.

These proteins are then exposed to blood from patients with Aids, or at risk of Aids, and those to which antibodies attach become the basis of the HIV test. If other patients show similar reactions, they are considered to be HIV-positive. The proteins (p) are numbered according to their molecular weight. On this basis, those considered to be most important in triggering production of antibodies to HIV have been identified as p120, p41/45, p55, p31/32, p24/25, and p17/18.

In the Bio/Technology paper, Eleopoulos went through each of these in turn and showed that in most instances there were alternative explanations for the presence of the proteins, which did not require HIV. The p41 protein, for example, one of those detected by both Gallo's and Montagnier's groups in the first "HIV" isolates, could be accounted for by actin, a protein found in all cells as well as bacteria and several viruses. Montagnier and his colleagues, in fact, had seen that serum from Aids patients reacted with a p41 protein both in HIV-infected and in non-infected cells, concluding that the p41 band "may be due to contamination of the virus by cellular actin which was present in immunoprecipitates of all the cell extracts". Actin is known to play a key role in activity at the surface of a stimulated cell, and for that reason the binding of actin antibody to such cells has been proposed "as a sensitive marker for activated lymphocytes". There is also evidence that two other bands, p120 and p160, are simply oligomers (multiples) of p41.

The p31/32 protein has been shown to be identical to one coded by the major histocompatibility complex (MHC) set of genes. It is described as an MHC class 2 protein, which plays a part in controlling the reactions of T4 (helper) lymphocytes, so again there is a logic as to why it should be found in people whose immune systems are under pressure much more than in those who are healthy.

The p24 protein has been considered to be extremely specific for HIV infection, so much so that it is considered to correlate with the progression of "HIV disease" and is used to quantify virus levels. Until 1987, according to CDC criteria, the presence of a p24 band alone in the western blot test was considered sufficient to confirm HIV infection. However, it then became apparent that healthy blood donors at no risk for HIV infection were often showing up positive on this band - 14 out of 100 donors (who were negative on the Elisa test) had antibody to p24 in one 1989 study. Among the recipients of blood from such donors, 36 per cent showed a similar reaction on the western blot test six months after their transfusion - but so did 42 per cent of individuals who received western blot-negative blood. Both donors and recipients remained healthy. The authors of this study concluded that such patterns are "exceedingly common" in randomly selected donors and recipients and do not correlate with the presence or transmission of HIV.

The P24 thought to represent HIV may in fact be a protein of the same weight that is present in the outer lining of platelets, particles that circulate in the blood and whose main job is to clump together to stop bleeding. Antibody to it may be especially high in regular blood donors, arising as a harmless consequence of their blood loss. Most homosexual men with Aids and pre-Aids have been shown to carry an antibody that reacts with this protein. The reason is unknown, but may relate to platelet activation. In any case, the reaction seems to be very variable, contrary to what one would expect if the cause were HIV infection. In one study quoted by Eleopulos, the authors found that "In half of the cases in which a subject had a positive p24 test, the subject later had a negative test without taking any medications that would be expected to affect p24 antigen levels...the test is clinically erratic and should be interpreted very cautiously."

Finally, there is evidence that p18, which along with p24 is most often detected by western blot in healthy blood donors, is also associated with immune system activation. Blood from Aids patients reacts with a p18 protein in HIV-infected T-cells that have been stimulated into dividing, but not with uninfected, unstimulated T-cells. However, when the uninfected T-cells are stimulated, the same reaction appears, so it would appear that the antibodies are binding to a protein associated with cell division and activation rather than with HIV.

In view of these observations, Eleopulos wrote, "it is difficult to defend the view that the bands p41 (and thus p160 and p120), p32, p24 or p18 represent specific HIV proteins."

Even if it could be shown that they really did belong to HIV, it could not simply be assumed that someone with antibodies reacting with them was infected with HIV. From the first antigen-antibody reactions performed by Montagnier and Gallo, it was found that not all the "HIV" proteins reacted with all sera from Aids patients, or even sera from the same patients obtained at different times. Other factors were evidently involved, so a second criterion of test validity had to be met, that of standardisation. There needed to be some criteria as to what constituted a positive western blot, so that a given test result had the same meaning in all patients, in all laboratories, in all countries.

That was not the case, however. An FDA-licensed western blot kit, used by a minority of laboratories, required a positive result on three different bands, p24, p31, and either p41, p120 or p160. When these stringent criteria were used, less than 50 per cent of Aids patients tested HIV-positive. The Consortium for Retrovirus Serology Standardisation, however, defined a positive western blot as the presence of antibodies to at least p24 or p31/32, and p41 or p120/160. Using these criteria, the proportion of Aids patients testing positive increased to 79 per cent. Other laboratories and organisations used still different standards. Moreover, even with the most stringent criteria, 10 per cent of control samples, which included specimens from blood donation centres, showed positive on western blot.

In the scientific literature, no strips had been published of a standard positive western blot. Interpretation, even with a single kit, left much to the subjective experience of a particular clinician or laboratory. For instance, one early instruction manual which reproduced examples of strong, weak, and non-reactive patient sera, went on to warn:

"This example shows typical reactive patterns only, and is not to be used as a reference for comparison with results from unknown serum samples...Patient samples may show varying degrees of reactivity with different proteins, thus showing different band development patterns...Each laboratory performing western blot testing should develop its own criteria for band interpretation. Alternatively, band interpretation may be left to the clinician."

The difficulties went still further. Other workers had shown that the patterns obtained depended on many factors including temperature, and the concentrations of a chemical used in the preparation of the "viral" material for the test.

The third criterion for a valid test was that it should be reproducible, but perhaps partly because of the above, there was evidence that even first-class laboratories could produce widely differing conclusions concerning the same samples. In addition, according to a 1989 CDC report, many laboratories were still using unlicensed western blot kits because of cost and the "stringent criteria required for interpreting the licensed test."

Once an antibody test was thought to have met the criteria described above, Eleopulos said, its performance in clinical practice had to be examined. For that, an alternative, independent method of establishing the presence of the infection for which the test is employed was needed. "This method, often referred to as the gold standard, is a crucial sine qua non, and represents the tenet upon which rests the scientific proof of validity." The only possible gold standard for the HIV antibody test was HIV itself. In other words, to be sure that the test was valid, you would need to show that HIV was present in people who tested positive, and that it was not present in people who tested negative.

This has never been done! Instead, some studies purport to validate one test by seeing how well it performs against another, but since the tests are usually based on similar principles, if the principles are wrong it means all such tests are wrong too. Other studies have used the clinical syndrome of Aids as a "gold standard", so that a test is considered valid if it can distinguish between serum from Aids patients and serum from healthy people. Thus, back in 1987, the CDC's Aids definition accepted a positive Elisa , without confirmation by western blot, "because a repeatedly reactive screening test result, in combination with an indicator disease, is highly indicative of true HIV disease." However, since most Aids patients have many physiological abnormalities, especially affecting their immune system, it is not surprising that many test positive for proteins which, as described above, are associated with immune system activation. Assuming that these proteins represented HIV was an unwarranted step.

All "HIV-positive" tests may be false positives

Since no one could say with any certainty whether the antibodies detected by the "HIV" test related to HIV or to these other proteins, all HIV-positive test results might be false positives, Eleopulos argued.

Since individuals from the main Aids risk groups, that is, gay men, drug users and haemophiliacs are exposed to many foreign substances such as semen, drugs, Factor 8, blood and blood components; and individuals belonging to the above groups commonly develop infections unrelated to HIV; one would expect these individuals to have high levels of antibodies directed against antigens other than HIV. In fact individuals with Aids, Aids-related complex, and those at risk have circulating immune complexes, rheumatoid factor, anti-cardiolipin, anti-nuclear factor, anti-cellular, anti-platelet, anti-red cell, anti-actin, anti-DNA, anti-tubulin, anti-thyroglobulin, anti-albumin, anti-myosin, anti-trinitrophenyl and anti-thymosin antibodies. Anti-lymphocyte auto-antibodies have been found in 87 per cent of HIV-positive patients, and their levels correlate with clinical status.

There was evidence that apart from specific cross-reactions with the "HIV" proteins used in the test, the general burden of immune system reactivity, with consequent heavy load of circulating antibody-antigen complexes, increased the chances of testing HIV-positive. A 1985 study found that in healthy Africans, the probability of a positive test result increased significantly with increasing immune-complex levels. The authors concluded that "reactivity in both Elisa and western blot analysis may be non-specific in Africans...the cause of the non-specificity needs to be clarified". A similar possibility was raised in relation to intravenous drug use, when researchers tested stored serum samples from 1,129 users, and 89 controls, dating back to 1971-72. They found that 17 of the samples from the drug users, but none of the controls, tested positive. The study authors commented: "On the basis of our positive western blot data, it appears that parenteral drug users may have been exposed to HTLV-3 [HIV] or a related virus as early as 1971. An alternative but equally viable explanation is that the HTLV-3 seropositivity detected in these specimens represents false positive or non-specific reactions." A 1988 study demonstrated that in drug addicts there was a strong association between high serum globulin, which includes immunoglobulins (the antibodies responsible for the immune response), and a positive HIV antibody test. This was "the only variable which remained significant" in statistical analysis.

In Eleopulos's view, the link was a cause-and-effect one: the immune system activation in such cases was causing the "HIV" test to show up positive. The reaction was similar to one that can confound testing for syphilis. Known as a "biological false positive" (BFP), it has been described by syphilis researchers as often being "a marker for an unidentified disorder of the immune system that predisposes to autoimmune diseases".

Several other pieces of evidence were cited in the paper in support of the idea that a positive HIV antibody test may be the result of a challenge to the immune system other than HIV.

· HIV was thought to be transmitted by infected needles, yet a higher percentage of prostitutes who use oral drugs (84 per cent) than intravenous drugs (46 per cent) tested positive, according to a 1988 study published in The Lancet.

· Mice exposed to T-cells from another mouse strain were shown to make antibodies against two "HIV" proteins.

· Recipients of HIV-negative blood have been shown to become HIV-positive and develop Aids, while the donors stayed healthy and seronegative.

· In healthy individuals, partners of HIV-positive individuals and organ transplant recipients, a positive HIV-test may revert to negative when exposure to semen, immunosuppressive therapy or clinical improvement occurs. [Several similar cases have since been reported in haemophiliacs].

· Hospital patients in the US (26 hospitals studied), who were not at risk of developing Aids and who did not have any infectious diseases, tested HIV-positive at a high rate on western blot (1.3 per cent to 7.8 per cent). That might either mean HIV was spreading to the general population, or that the HIV antibody tests were non-specific. The latter seemed more likely, in view of the fact that by 1992 in New York only 11 men were reported to have acquired Aids through heterosexual infection.

· Amazonian Indians with no contact outside their tribes, and with no Aids, were found to test HIV-positive with western blot at rates varying from 3.3 per cent to 13.3 per cent depending on the tribe. In another study by the same researchers, 25-41 per cent of malaria patients in Venezuela had a positive western blot, but no Aids.

All such problems could be avoided by use of the only suitable gold standard, HIV isolation, the paper argued. So why had the HIV tests not been validated in that way? The answer, set out in four more pages of closely argued and extensively documented text, was even more astonishing. It was that Aids scientists, including Montagnier and Gallo, had been unable to isolate HIV in an unequivocal way.

Various indirect molecular, biochemical and genetic findings had been interpreted as meaning virus isolation, but none of these had offered conclusive evidence. In other words, the entire HIV story, with all the problems it had brought in its wake, might be a monumental error - a monstrous self-deception. It was not just that HIV was harmless, as Duesberg insisted. If Eleopulos was right, the HIV emperor did not just have no clothes, he had no body, no identity. He was not even a little man making a big noise, as in the Wizard of Oz; according to this evidence, HIV had never been shown to exist!

When I first read the paper, I did not have enough knowledge of molecular biology to comprehend fully the argument on virus isolation. It is only after another two years of familiarising myself with the field, including a period spent in Perth with the paper's authors, that I have gradually appreciated the significance of what they are saying. There are also limits to the human being's ability to grasp ideas and arguments - no matter how logically presented - when those arguments flatly contradict not just what one has previously believed oneself, but what the rest of the world believes too.

I was even reluctant to write a news story about the challenge to the HIV test until I could find some further support for it, although after several readings I felt the review article had certainly made a very powerful case. I talked it through with Sue Douglas, a senior Sunday Times executive who has a scientific background, and she encouraged me to go ahead. But first, I faxed the article to four experts in case some glaring error invalidating its reasoning had been missed by Bio/Technology, although the article had been peer-reviewed before publication.

One did not send any response to the article. He told me when I first called him that if false positives had been a common finding, "I think we would have spotted them. One would have been erroneously diagnosing HIV; patients will sue, and make a few millions." Another preferred not to comment, but referred me to a third, Dr Philip Mortimer, of the Virus Reference Division at Britain's Central Public Health Laboratory. Mortimer wrote a courteous reply acknowledging that the article "does make some fair points about the weakness of the western blot test when it is used incautiously and without follow-up." He added however that "the situation it describes is not typical of this country where initial positive serological (antibody) screening tests are confirmed by (i) further investigations, usually a combination of different Elisa assays but sometimes including western blot and (ii) a test of a follow-up specimen. Only if the positive reactions on both specimens are confirmed, usually in a reference laboratory, is a positive report issued." Perhaps this more stringent procedure helped to explain why Britain had only some 23,000 sero-positive people, compared with an estimated 1m in the United States and multi-millions in Africa.

But Eleopulos et al had not just criticised the western blot test. They had cited evidence indicating huge numbers of false positives with the Elisa test. In Russia in 1990, for example, out of 20.2m such tests performed, there were 20,000 false positives and only 112 "confirmed" positive results. A similar study in 1991 confirmed only 66 out of approximately 30,000 positive test results. Reporting these findings in The Lancet on 20 June 1992, Alexander Voevodin, of the Institute of Pulmonology in Moscow, said the severe consequences could largely be avoided "if the HIV status of screening-test positives is clarified rapidly, but this does not happen in Russia. The time scale for completing confirmation tests ranges from weeks to months and field epidemiologists often embark on contact tracing without waiting for the confirmation. Also, personal information on results of HIV tests is often leaked, resulting in discrimination and stigmatisation both of HIV-infected individuals and the large number of false positives."

Clearly, by using multiple tests giving very different results, false positives would be greatly reduced. But this still did not answer Eleopulos's charge that there was nothing in the literature to indicate why any of the tests should be considered reliable as indicating the presence of a specific retrovirus. Even if the damage done by false positives was being reduced in the UK and elsewhere by repeated testing, that was no comfort to those still left with a false diagnosis. Nor was it any help with regard to the situation in Africa, where because of cost considerations, most HIV diagnoses were being made on the basis of a single test. As Eleopulos told me, "the paper really shows that we have to question all types of the antibody test, especially in Aids patients, who have all types of infectious agents in them. How do we know that a positive test in Aids risk group patients is due to antibodies directed against HIV, and not against the many infectious agents to which Aids patients have been exposed? If the test is no good, you can repeat it a thousand times and it still won't be any good. When the principle of the test, the basis of it, has not been established, it doesn't matter how many times you repeat it, you still won't prove anything.

"The only way to show an antibody test is good, that it is telling us something specific, is to have a gold standard. For that, they have to have HIV; and it doesn't matter what they say, nobody has done that. So instead, when any other test is introduced, they have had to use the western blot as the gold standard. If the western blot is no good, no other test is any good either."

Mortimer also commented that diagnostic capability had recently been advanced by the introduction of a commercial polymerase chain reaction (PCR) assay for detecting minute quantities of HIV genetic material. "Comparison of results using this procedure with those obtained by antibody tests show a very close correlation confirming the reliability of HIV antibody tests," he wrote. However, as the Bio/Technology paper pointed out, this correlation might be the result of some quite different cause common to both the PCR test and the antibody test. PCR signalled the presence of only a small stretch of genetic material; perhaps it was picking up the presence of a sequence made detectable by the same stimulus as that which caused a person to test antibody-positive, a stimulus which need not having anything to do with "HIV". The Bio/Technology paper cited evidence in support of this idea. For example, a positive PCR reverted to negative when exposure to risk factors was discontinued; and cells from HIV-positive patients in which no HIV DNA could be detected, even by PCR, became positive for HIV RNA after immune activation by co-cultivation with activated T-cells.

The fourth expert was Professor Robin Weiss, head of the Chester Beatty Laboratories at the Institute of Cancer Research, London, who with Dr Richard Tedder, a virologist at the Middlesex Hospital in London, developed and patented Britain's first HIV test in conjunction with the drug company Wellcome. Weiss had invited me to visit his "blinkeredly orthodox" HIV research laboratory - "if you dare to expose yourself to possible contamination by orthodox views" - after the original "Aids: Can We Be Positive?" article. He had spent two hours trying to explain why he thought Duesberg was wrong. Although my experience of the meeting was mainly of being berated for having had the temerity to write as we did (I remember him showing me a front page headline from the Daily Mirror that week, "Your Surgeon Has Aids - Hospital Sends Shock Letter to 200 Patients", and telling me it was a more responsible story than our own) his exasperation had been laced with humour and besides, he had been involved with Aids research from the start.

Weiss told me on the phone that "I have always bemoaned the fact that western blots became the gold standard, I have never liked them." He then took the trouble to write a two-page letter concerning the Bio/Technology paper. His tone was set in the first paragraph: "It is the sort of paper I would have stopped reading by paragraph 5 if you hadn't requested an opinion." Later, he commented: "Sorry, if the authors were my students, I'd mark this essay B-minus. Of the 1,000 or so papers on HIV/AIDS that must have been published in the last six months, I'd put this in the bottom 10% for being worth reporting...I wonder if you will report it all the same. That would confirm my opinion that the Sunday Times is not interested in genuine controversies about Aids, which are, of course, in plenty." He quarrelled with some of the technical points raised, although he acknowledged that the paper might have had some merit if it had been published around 1986/7, as "there were serious difficulties and much variation in assessing western blot data, and some of the Elisa tests were still giving false positives." But since then, he argued, the tests had been greatly improved because almost all the commercial companies had switched to using HIV antigens produced in bacteria by recombinant DNA technology, rather than grown from sera taken from Aids patients.

It seemed to me that he had not answered the central complaint, that no one had ever established that the proteins held to indicate the active presence of HIV really are related to the virus in people who test positive, as opposed to other possibilities raised by the Bio/Technology authors. I wrote back along those lines, adding: "This is extremely important, surely. If they are right, it provides a perfect explanation of why 'HIV'-positive haemophiliacs who are switched to purified Factor 8 cease declining towards immune system failure. I have not heard any other explanation for this phenomenon."

On August 1, 1993, the Editor ran our most challenging story to date across the top of the front page. The headline read: "New Doubts Over Aids Infections As HIV Test Declared Invalid". The story began:

The "Aids test" is scientifically invalid and incapable of determining whether people are really infected with HIV, according to a new report by a team of Australian scientists who have conducted the first extensive review of research surrounding the test.

Doctors should think again about its use, say the authors. "A positive HIV status has such profound implications that nobody should be required to bear this burden without solid guarantees of the verity of the test and its interpretation," they conclude.

The findings, likely to cause intense debate in the medical fraternity and anguish for many HIV-positive people, are contained in an article published by the respected science journal, Bio/Technology.

Many people who appear to be infected by HIV, say the researchers, can be suffering from other conditions such as malaria or malnutrition that produce a positive result in the test. Even flu jabs can produce the same effect. As a result, predictions by the World Health Organisation that millions are set to die because of being HIV-positive may be wildly inaccurate.

The paper also lends powerful support to the theory, held by growing numbers of scientists, that HIV is not the true cause of Aids. One of its authors, Eleni Eleopulos, a biophysicist at the Royal Perth Hospital, said this weekend: "There is no proof that people labelled as 'HIV-positive' are infected with such a retrovirus. We should really question the role of 'HIV' in the causation of Aids."

After setting out key aspects of the case against the test, the article noted that the World Health Organisation - which was seeking an extra £2bn a year for its Aids prevention programme - estimated that about 14m people had been infected with HIV worldwide. WHO claimed the total would reach 30-40m by the year 2,000, and that most would eventually contract Aids. Developing countries were said to face the biggest threat, with Africa alone already having an estimated 8m HIV-infected people.

However, according to the Bio/Technology report, those were the countries where the tests might be at their most unreliable because of widespread ill-health caused by other diseases. Severe malnutrition and multiple infections were especially likely to produce a misleading result in the test. Claims that current Aids tests were virtually 100 per cent accurate were based on studies of healthy subjects.

We also quoted Peter Duesberg as welcoming the report, on the grounds that it helped to explain how "a false correlation" had been found between "HIV" antibodies and Aids. "The whole virus hypothesis of Aids is based on this correlation," he said. "Its proponents have nothing else: no mechanism whereby HIV could do the damage attributed to it, no animal tests, no cure, no vaccine, no virus activity. They have nothing conventional in terms of virus-disease argument, except this correlation with antibodies. If this study is correct, and I have no reason to doubt it, it means that even that is now falling apart."

The article added that the findings had led to a call by the New York Native for legal action against the American government by relatives of people who had killed themselves, or suffered toxic effects from taking AZT, as a result of positive HIV tests.

Circular reasoning means test remains unvalidated

Having already had to cope with so much protest over previous articles reporting on the calls for a reappraisal of HIV's role in Aids, I expected this fresh challenge to provoke a new wave of criticism and comment. After all, the claims were completely at odds with conventional thinking on this enormously important subject. But this time, there was hardly a word of protest, let alone any arguments of rebuttal. No scientific papers to validate the tests. And no comment elsewhere in the media. I felt sure we were being privately criticised by the Aids experts to whom specialist writers turn for advice on such issues, but perhaps they now felt it better that there should be no public debate. .

Robin Weiss wrote again on 2 August, as follows:

Yes I did answer their central complaint - you chose to ignore it:

(a) Since ~ 1988, most tests including western blots, slot blots and other variants of antibody blotting have used HIV antigens recombinant DNA methods and the bands are therefore nothing other than HIV.

(b) Ref (3) by Pinter et al, the major complaint was misunderstood. The gp120 was genuine gp120 and the gp160 a mixture of gp160 and gp41 tetramer - in both cases, real HIV antigens.

(c) Western blots are only ever used as part of a panoply of confirmatory tests, not as screening tests.

It is difficult to believe you are as slow a learner as you make out. I cannot respect the integrity of yesterday's article, including again a gratuitous, petty and wrong smear on the Wellcome Trust (which does not fund my lab's Aids research).

Weiss was wrong in his challenge to Eleopulos's claim that p120 and p160 had been found to be multiples of the p41 protein, and therefore did not qualify as separate indications for HIV infection in the western blot test. In a letter to the New England Journal of Medicine (11 May 1989) Dr Abraham Pinter and co-authors stated that they had "unexpectedly" found that these two viral antigens "are primarily multimers of the HIV transmembrane glycoprotein gp41 that react with antibodies to gp41. Confusion over the identification of these bands has resulted in incorrect conclusions in experimental studies. Similarly, some clinical specimens may have been identified erroneously as seropositive, on the assumption that these bands reflected specific reactivity against two distinct viral components." Tests indicated that some true gp120 was also present, but no gp160.

Weiss's last sentence was referring to an AZT story written by another reporter the same weekend, and apart from allowing Weiss to let off some steam, had nothing to do with the question of the HIV test's validity. I replied:

I am getting used to your style and now rather like it. However, please have patience with this slow learner and explain how the bands which you say are "nothing other than HIV" are the equivalent of virus isolation as a gold standard for testing the specificity of the diagnostic kits? The purity of the bands overcomes one of the complaints in the Bio/Technology paper, that the proteins previously came in a mixed soup; but there was a lot of evidence in the paper that pure or not, the proteins are not specific to HIV and that therefore the test should be proved effective by comparing positive results with virus isolation.

Weiss responded: "As I wrote, that might have been a valid argument six years ago, but not today as the proteins have been specific for some years."

We were getting nowhere, except that I now felt more confident than ever that Weiss had no effective response to the Perth team's criticisms of the test. Like so much in HIV/Aids thinking, his argument was circular. Philip Mortimer, despite being more generous towards the Bio/Technology paper, seemed to have fallen into the same trap. In a 1991 Lancet paper he had favoured new Elisa tests "based on recombinant and peptide antigens", which, as Weiss said, overcame the problem of not knowing precisely what antigens were present in the test kits. But it was not much use knowing what had gone into the tests if you still did not know whether or not those antigens were specific to HIV.

This, surely, was exactly what had happened. As Mortimer said, the western blot was used as a "gold standard" for the other HIV antibody tests. Yet his paper, entitled "The fallibility of HIV western blot", concluded that "western blot detection of HIV antibodies began as, and should have remained, a research tool". Earlier, in a 1989 article, he had written:

Diagnosis of HIV infection is based almost entirely on detection of antibodies to HIV, but there can be misleading cross-reactions between HIV-1 antigens and antibodies formed against other antigens, and these may lead to false-positive reactions. Thus it may be impossible to relate an antibody response specifically to HIV-1 infection. In the presence of clinical and/or epidemiological features of HIV-1 infection there is often little doubt, but anti-HIV-1 may still be due to infection with related retroviruses (eg HIV-2) which though associated with Aids, are different viruses."

Mortimer revealed here another important indication of why HIV-positivity has appeared to be so closely correlated with Aids risk groups, apart from the fact that for several years a positive HIV test has formed part of the definition of Aids: clinical and epidemiological features are sometimes, perhaps commonly, invoked as part of the diagnosis. Thus, a gay man with a given test result may be judged positive, while a heterosexual with the same test result may not be burdened with an HIV "diagnosis", unless some risk factor can be identified. This policy, reflecting the difficulties Aids scientists have encountered as the non-specificity of the test became more evident, was spelled out by the Consortium for Retrovirus Serology Standardisation in 1988: "For symptomatic individuals clinical evidence of infection with a positive western blot test provides criteria for diagnosis. However, for asymptomatic individuals, other criteria including history, high-risk behaviours or results of other virological or serological studies may be necessary for the diagnosis of HIV infection." The consortium added that "the western blot test result is highly dependent on the technical skill and subjective interpretation of the experienced laboratorian...The establishment of even the most rigorous criteria will not completely eliminate the subjectivity and variability inherent in western blot testing."

In another big critique of the HIV test, Eleopulos and her co-authors have pointed to evidence which could also help to explain why in Africa as well as elsewhere, HIV-positivity in healthy people tends to concentrate in certain age groups. The scientific literature "abounds" with data, they say, demonstrating cases of mistaken identity when researchers thought they had found retrovirus causes for diseases such as cancer, only to discover that the antigens they had detected did not belong to retroviruses at all, but were of a naturally occurring "heterophil" (general) type. This type of antibody, which may be related to growth spurts as well as generalised stress, peaks at about three years of age and at about 35 years. Levels are also influenced by tumour growth and chemotherapy, and have been seen to be raised in older people suffering from several types of cancer. One group of authors said that their findings "emphasise the need for extensive controls in determining the specificity of human antibody recognition of viral protein determinants." The need was especially great when radioimmunoprecipitation (RIP), a third, ultra-sensitive type of antibody test was used.

The idea that the HIV antibody tests are non-specific - that is, that positive results can be triggered by a wide variety of conditions - may answer a number of puzzles in Aids research, Eleopulos says. For example, researchers have found that TB responds just as well to treatment in African patients who are HIV-positive as in those who are not. The reason may be that the HIV-positivity is a false positive reaction. TB itself can produce symptoms that mimic Aids, including a decrease in T4 cells and prolonged, depressed immunity.

Scientists in Nairobi have been surprised to find that many prostitutes stay HIV-negative despite continuing to have unprotected sex with HIV-positive men. The idea that they have some kind of special resistance to the virus has been discussed, attracting world headlines. The real reason may be that the HIV-positivity in their clients is a non-specific reaction that therefore does not put the prostitutes at risk of becoming infected with a deadly virus.

Those prostitutes who themselves test HIV-positive may similarly not be infected with any such virus. This would explain the puzzlement of researchers who reported on "Aids virus infection in Nairobi prostitutes" in 1986 in the New England Journal of Medicine. They found that two thirds of female prostitutes from an economically depressed neighbourhood tested positive, just under one third of prostitutes from a better-off area, eight per cent of a group of men treated at a clinic for sexually transmitted diseases and two per cent of medical personnel. But they added: "The high prevalence of [HIV] antibody among the prostitutes was a surprising finding in view of the paucity of cases of overt Aids diagnosed in Kenya...The prostitutes whom we interviewed did not describe any deaths within the prostitute community during the past five years that were suggestive of Aids."

In 1985, Robert Gallo and colleagues reported how they tested sera dating back to 1972/3 from 75 healthy children from the West Nile district of Uganda. Both Elisa and western blot tests were performed. Two thirds of the samples were positive. Since the children, whose mean age was 6.4 years, were presumed infected by their mothers, the authors expected at least an equal percentage of adults to be infected, raising the question of "why the incidence of Aids in the Ugandan population (and neighbouring Zaire) has gone unnoticed for so long...it is possible that Aids existed in African populations without being recognised as a separate disease entity." That point of view, Eleopulos said, was inconsistent with the conventional idea of HIV as being an extremely infectious agent with an incubation period for Aids of about four years in African victims. "The Ugandan population should by now...be drastically reduced - perhaps decimated. This is not the case and it would seem prudent to accept that the majority, if not all of the positive tests, were false positives induced by other diseases endemic in the area of study." Tragically, however, after whole nations had been panicked by these false test results into believing they were about to be decimated by HIV, many of the traditional causes of death became redesignated "Aids" in Africa.

Also in 1985, US and French researchers reported that the wife of a haemophiliac Aids patient who was found to have low T4 cells and to be HIV-positive reverted to normal after exposure to her husband's semen (vaginal, oral and anal) was stopped. According to the researchers, "her discontinuation of sexual exposure [to HIV], or her lack of antigenic stimulation or both, may have played a role in her conversion to seronegativity and increase in the number of T-helper lymphocytes." According to Eleopulos, since semen is known to be capable of provoking antibody production as well as causing immune suppression, especially in anal intercourse, it may have been the original cause both of the decrease in T4 cells and of the HIV-positivity, "irrespective of HIV infection".

All of this evidence, she and her co-authors conclude, makes it likely that in healthy people as well as Aids patients a positive test does not indicate HIV infection but represents a non-specific marker for a variety of unrelated conditions. "Consequently, the general belief that almost all individuals, healthy or otherwise, who are HIV antibody positive are infected with a lethal retrovirus, has not been scientifically substantiated."

Just as when Joe Sonnabend received the cold shoulder - and worse - for his multi-factorial theory of Aids; and just as Peter Duesberg had met months of total silence, and ultimately utter rejection by the Aids establishment, for demonstrating his belief that HIV was harmless; so Eleopulos et al have been unable to engage the scientific mainstream in any open debate on the dramatic possibilities they have raised - though I suspect that behind the scenes, discomfort over what to do about the HIV test is growing acute.

Was it possible that, as Robin Weiss tried to persuade me, "the authors could have written a similar paper about any virus save, perhaps, measles, and found similar discrepancies; polio, herpes simplex, CMV, influenza"? "There is nothing special here about HIV," Weiss wrote. "Their whole concept of a gold standard is bizarre, as technology moves forward and tests are better now than five years ago."

That was not how Harvey Bialy saw it. Bio/Technology is a journal whose specialities include the validation of diagnostic tests, which was the issue at the heart of the Eleopulos paper. According to him, Eleopulos was absolutely right in her analysis of the circular reasoning surrounding the attempts to validate the HIV test by reference to other, similar tests.

Besides, I knew that during the second half of the 1980s, as the world was being driven into a panic by virologists over HIV's allegedly rapid heterosexual spread in Africa, there had been no public acknowledgement of inadequacies in the HIV test. Weiss had been among those responsible for this panic: in a February 1985 Lancet paper that he handed to me when we first met in 1992 - without any reservations about the quality of the data - he and his co-authors stated: "In Africa HTLV-3 [HIV] appears to be transmitted through heterosexual contact or exposure to blood through insect bites or scarification." The paper included the devastating information that a fifth of "control" samples taken in Uganda from healthy people as well as hospital patients tested HIV-positive, and declared that "the serological tests are highly specific."

WHO meeting hears of test unreliability

Although it had never been made plain to the public, experts knew from an early point that there were exceptional problems with the HIV test. Some of these doubts and uncertainties came up at a meeting at WHO's headquarters in Geneva on 14-16 April 1986, called to discuss the safety of blood supplies and issues related to antibody screening. There were more than 100 participants, from 34 countries. The proceedings were published by WHO as a book, Aids: The Safety of Blood and Blood Products, the following year, priced at £39.95. It was an academic publication and I never heard about it at the time, but came across a copy several years later.

Dr James Allen, assistant director for medical science in the CDC's Aids programme, told the meeting that the importance of eliminating technical errors was emphasised "by the high proportion (66.7 per cent) of initially reactive specimens in blood donors that give negative results when the tests are repeated". Another source of error derived from the inability of some manufacturers to provide uniformly reliable test kits and reagents. The result was variations in the rates at which screened specimens were repeatedly reactive by Elisa. Additional reasons for false positive test results included non-specific reactions, particularly antibody in specimens that cross-reacted with non-HIV antigens in the test system. "This latter problem has proved to be both real and annoying for blood collection agencies which find that less than 20 per cent of donors who are repeatedly reactive by EIA [Elisa]also have a positive western blot." Studies suggested some people were reacting to components of the cell line used to grow HIV for many of the test kits licensed in the US. Other cross-reactivity was occurring because of antibodies to naturally occurring cell surface proteins, the so-called HLA antigens.

Allen gave a warning that the problems could be magnified in areas of the world that did not have the sophisticated facilities of the United States. Blood centre laboratories that were not routinely processing large numbers of specimens might find the tests uneconomical and technically difficult. "In addition, laboratories that are not equipped to handle the dilutions, critical incubation times and temperatures, and washing procedures required, or that do not have a constant and dependable electrical supply for automated readers, may have trouble obtaining accurate and reproducible results. Blood centres that do not have refrigerated storage facilities for donated blood may find the delay of several hours required to perform the test unacceptably long." Other problems that needed to be assessed included the impact of variable shipping and storage conditions on the test kits and reagents, and potential problems of nonspecific cross-reactivity of sera collected from donors in different areas.

Allen added that although the tests licensed for screening blood supplies were sensitive enough to detect most infected donors, a highly specific confirmatory test was needed to distinguish which people with a repeatedly reactive result had genuinely been infected with HIV, and which had a nonspecific reaction. "To date, no entirely satisfactory confirmatory test has been developed, standardised and licensed to help resolve this dilemma, although it is probable that such tests will be available in the future." Later, challenged about the CDC's criteria for determining test results, he said: "A major part of the problem in reading western blots is that methods are not standardised and interpretation is subjective. It varies from one laboratory to another, depending on how the gels and the blots are set up, on the source and amount of antigen that is used, and on multiple other technical factors."

The CDC had spent many hours of discussion with groups and different laboratories trying to establish the definition of a positive western blot. It had ended up deciding that most infected people had antibody against the p41 and p24 proteins, and that therefore the presence of one or both of those bands could be interpreted as a positive result, although "if the only band present is p24 we are reluctant to call the blot positive, and we try to get an additional specimen to repeat the blot." That led Professor F. Deinhardt, director of a microbiology institute in Munich, Germany, to say that in his unit, non-infected cells had been shown to produce a false-positive band in the p41 region, so even that could not be trusted on its own. "We ourselves would never report a sample as confirmed positive if we do not have at least two virus-specific bands which we can be sure of. We would also further investigate before we let the axe fall on the poor individual and tell him he is anti LAV/HTLV-3 [HIV] positive." He felt it would be "highly dangerous" to license a western blot test so that any laboratory could start running it, because of the technical difficulties and the pitfalls in interpretation.

Dr Thomas F. Zuck, from the FDA's office of biologics research and review, commented: "Dr Deinhardt's concerns are the same concerns we have; interpretation is as much a part of the test as are the physical reagents one uses for it. So we have difficulties at this point deciding what is going to be required to validate a claim a manufacturer may make that a test is confirmatory." Later he added: ""One of the difficulties we have in looking at 'claims' for confirmatory tests or designing systems to validate what in fact is going to be 'confirmatory', is determining how you define and validate it...What will be confirmatory? What claims are we permitted to make? It's going to take some time for us to sort through this and to do the appropriate clinical trials."

In his own presentation, Zuck emphasised that an initial positive reaction using the Elisa test meant very little, "unless the test is highly reactive or the subject is known to be at high risk" for HIV infection. Only 3.8 per cent of initially reactive sera from more than 2.5m blood donors screened by American Red Cross centres could be confirmed by western blot. Surveys conducted jointly by the College of American Pathologists and the American Association of Blood Banks, covering more than 5.5m donations, produced a similar result, with a confirmation rate of four per cent. He went on: "Because knowledge of the degree of risk [for HIV infection] of the person being tested is of such importance when interpreting test results, an additional more specific test, such as western blotting, is needed to interpret the meaning of repeatedly reactive sera from people of unknown risk." If tests were interpreted that way, few diagnostic errors would be made. "Further, because a positive test has such enormous social, medical, and psychological implications, persons with sera reactive by additional more specific tests need medical evaluation and counselling."

Similar difficulties with false positive reactions were raised by Dr John Barbara, head of microbiology at the North London Blood Transfusion Centre in England. Because the virus replicated by budding through the wall of an infected cell, it carried cell antigens with it. "In addition to this, viral purification for preparation of reagent antigen can never be perfect." As a result, non-specific but repeatable "cross-reactions" were a considerable problem. There was a big need for a definitive confirmatory test. "It is important to remember that the western blot (commonly used as a confirmatory method) is itself an antiglobulin assay. It is liable, therefore, to the same kind of false-positive reactions as the screening test it might be confirming." Barbara also pointed out that when (as with HIV in blood donors) the prevalence of an antibody is very low, even a test that sounds highly specific may still produce a high rate of false positive results. For example, if the true prevalence were one in 1,000, a test with 99.8 per cent specificity would find one true-positive and two false-positives in every 1,000 samples - "a 67 per cent false positive rate!"

Commenting on the difficulties of separating the virus from normal cell components, Dr M.V. O'Shaughnessy, head of viral surveillance at the Laboratory Centre for Disease Control at Ottawa, Canada, said that when the proteins from an HIV preparation were electrophoretically separated and stained using ultra-sensitive techniques, "more than 30 individual proteins can be recognised". Several large Canadian studies had shown extensive non-specific cross-reactivity, especially in haemophiliacs and in North American native populations. He also reported problems with cross-contamination. The western blot procedure was so sensitive that some positive sera could be diluted up to a million times and still be clearly positive. Accidental transmission of one microlitre (one millionth of a litre) of diluted sample from one well to another in the incubation tray could result in a false-positive being reported. O'Shaughnessy added that new tests based on genetically engineered or synthetic antigens were now being evaluated, which had the theoretical advantage of not being contaminated with proteins from the host cells or from the cell line in which HIV was grown. "However, these second-generation procedures will almost assuredly result in a second generation of problems for the diagnostic laboratory."

Today, with the advantage of hindsight, distanced from the madness that surrounded discussion of Aids in the mid-1980s, one can feel sympathy for the experts who struggled to the best of their ability with the tangled web of issues surrounding blood testing. The more I read of their difficulties, however, the more it seemed that Eleopulos might be right, and that the root of their problems could be that the deadly new virus they had assumed they were dealing with was itself a "false positive" - a mistaken concept. Questioning that root assumption was not a part of their agenda at that time.

Nevertheless, sympathy comes still more readily for all those people who have been falsely given to understand that they are infected with HIV. Because of its ability to detect a generalised immune system activation, the Elisa test helped clean up blood supplies, but it was never suitable for more than that. Regardless of whether or not the virus is a real entity, it is clear that in the early years at least, the uncertainties surrounding the HIV test were so great as to render it unfit for diagnostic use.

The FDA's Thomas Zuck admitted as much at the WHO meeting. Indications for the use of the Elisa kits in the US had expanded beyond those for which they were designed and initially licensed, he said. "Consistent with the data obtained in support of test licensure, use of the test was intended to be limited by phrases in the package inserts stating, 'the primary use of the HTLV-3 antibody test is to screen blood and plasma donations...it is inappropriate to use this test as a screen for Aids or as a screen for members of groups at increased risk for Aids in the general population'." However, "enforcing the intent of this language would be analogous to enforcing the Volstead Act which prohibited alcoholic beverage sales in the United States in the 1920s - simply not practical."

Broad application of the test was evident, he said. The nonspecificity of the test would cause difficulties when indications for its use were expanded, particularly when diagnostic and policy decisions might be based on the results. But he took some comfort from the fact that Elisa testing "to slow the spread of ...infection may serve the public health well," backed up by additional more specific testing, such as western blot, of all repeatably reactive sera. "Public policies about confidentiality of test results, economic and social decisions based on the results of these tests, and uses dictated by public health needs must be determined by judicial bodies rather than control authorities."

In fact, as Gordon Stewart has constantly emphasised, the public health has been ill-served by an infectious disease paradigm, with a non-specific diagnostic test at its heart, that told the world it was in the grip of a deadly new disease putting everyone at risk. As well as inflicting great psychological and physical damage on "HIV-positive" people, the test led to the misappropriation of billions in public funds. Science disgraced itself by adopting the test on the basis of inadequate evidence, allowing its indiscriminate use, and by failing to make the non-specificity of the test absolutely clear to policy-makers, physicians and individuals as soon as this flaw became apparent, as it clearly had done to the experts who gathered in Geneva as far back as April 1986.

Are today's tests more reliable?

According to the Bio/Technology analysis, the lack of validation of HIV antibody tests was an intrinsic defect. No matter what permutations manufacturers developed with the proteins used in the tests, they would still be in difficulties as long as they did not check how well the tests performed against the gold standard of virus isolation - which they were unable to do, because they could not isolate the virus. Robin Weiss, on the other hand, had told me that "the proteins have been specific for some years". Who was right?

Evidence that the HIV test is still beset with major difficulties is contained in the latest comprehensive guide to the subject, the 1994 edition of Aids Testing, a 400-page textbook edited by two CDC experts, Dr Gerald Schochetman and Dr J. Richard George. There are bland assertions that the tests are "highly sensitive and specific", just as used to be said of the earlier tests. But the science tells a different story.

In a chapter on FDA regulation of the tests, Dr Jay Epstein, of the agency's centre for biologics evaluation and research, comments on "a continuing controversy" surrounding the inconsistency of some screening test kits "because of the familiar problem of false-positive reactions" with non-specifically "sticky" antibodies. "Such a problem has been observed for several blood donor screening Elisa tests following immunisation for influenza," he says. Another concern with screening tests is the accuracy of rapid tests introduced in recent years, which require subjective interpretation by the operator. "Poorly trained operators can easily misinterpret test results."

The western blot remains by far the most widely used procedure for "confirmatory" testing, Epstein says. With this, "the greatest concern has been the high prevalence of nonspecific banding patterns, resulting in indeterminate test results. Unfortunately, this phenomenon is intrinsic to the technology due to a variety of causes that include antibodies in the patient sample that bind to non-viral proteins in the viral antigen preparation and antibodies of unknown specificity that cross-react with viral proteins."

Furthermore, the FDA has actually relaxed its criteria for a positive western blot, to get over the embarrassing fact - spotted by Eleopulos - that with the first kit licensed for confirmatory testing, which required a positive result on three different bands, fewer than half of Aids patients tested HIV-positive. "Currently used criteria are less stringent and have resulted in tests that are more sensitive but have a higher false-positive rate," Epstein admits. His justification for this move sounds legalistic rather than scientific. "The FDA has accepted product amendments for revised western blot interpretative criteria that are consistent with the prevailing scientific view and the recommendations of the Public Health Service."

It has also "attempted to reduce the occurrence of falsely reactive bands by close attention to quality control issues in manufacturing and through lot release criteria." This seems an inadequate recompense for reducing the stringency of a test which, as Eleopulos has demonstrated and as the UK's Philip Mortimer and other experts have agreed, was already severely flawed.

Epstein adds that the FDA has become "increasingly concerned" about the sale of unlicensed confirmatory test kits. Initially, it did not act against them because of a lack of availability of licensed kits, and because of "the pressing public health need for a means to validate HIV screening test results". Now that licensed tests were widely available, action would be taken against unlicensed products. Again, with hindsight, it seems extraordinary that such action, to reduce the margin of error in a procedure with such devastating potential impact on a person's life, should have been delayed for so long.

In a review of the various testing methods used, George and Schochetman reiterate the need not to tell people they are HIV-positive, or take medical decisions about them, based on Elisa testing alone. This is because of the continuing risk of false positive results. "Testing by the sensitive EIA [Elisa] is done to identify those persons who need additional, more specific supplemental testing. Counseling and medical decisions are made based on the results of the supplemental assay, not on those of the screening test alone." Apart from Elisa, they say, probably the most widely used screening test is one called SERODIA-HIV, made in Japan. It is not licensed by the FDA but is widely available in most other countries, particularly those in Asia, Australia, Latin America and Africa. It works on a principle known as agglutination. Particles such as gelatin or latex, coated with viral proteins, settle into distinctive patterns depending on the presence or absence of antibodies attributed to HIV in the patient's sera. False positive reactions are slightly more of a problem than with Elisa, the authors say, though the factors that trigger them seem to be different.

A variety of rapid screening tests are also on offer, many performing comparably to the Elisa, but some of which "have been developed and offered for sale at low prices in developing countries without being validated through clinical trials or reviewed by any recognised institution".

Problems also clearly remain with the western blot. Production of the strips for this test is not a precise process, George and Schochetman say. "Differences in protein concentrations, identities, and positions are observed between manufacturers and even between lots from the same manufacturer." Since the first recommendation, in 1988, by the US Public Health Service that Elisa-positive specimens should be further tested by western blot, "there has been extensive debate over the appropriate interpretive criteria for this assay". As Eleopulos pointed out, even within the US the criteria recommended by various organisations still differ considerably. Outside the US, the expense of confirmatory tests - they can cost US$60-100 in some African and Latin American countries - means that they are often not used at all.

A chapter on quality control makes plain that there are concerns in that area, too. "Each year the number of tests performed and the number of laboratories performing tests increase in the United States. Yet programmes for quality control of laboratory testing have not kept pace." Standard panels of serum for quality control and evaluation of Elisa and western blot tests are not generally available, the authors say. Quality control is more difficult for western blot than for Elisa, the major reason for the difficulty being the need to compromise "between what is scientifically necessary and what is commercially feasible". Measures to guard against deterioration of the kit, proficiency testing and performance evaluation, and daily monitoring and routine maintenance of the refrigerators, freezers, waterbaths, incubators, microplate readers and washers and pipettes are all needed to prevent inaccurate results being given to doctors and patients.

The polymerase chain reaction (PCR), which can amplify DNA a million times or more, is increasingly used for direct detection of genetic sequences attributed to HIV. However, because of its exquisite sensitivity, and the huge amplification it employs - it can multiply even a single molecule up to detectable levels - it is also very vulnerable to error, both in terms of procedure and interpretation. In a chapter concerning such techniques, Schochetman and Dr John Sninsky, of Roche Molecular Systems, say that PCR "does not have an intrinsic analytic or diagnostic sensitivity and specificity. The diagnostic sensitivity and specificity are inextricably linked to the laboratories performing the procedure. As a result, the confidence in the reported results is directly proportional to the experience and critical interpretive criteria used by the laboratories performing the assay." Factors that make a dramatic difference to results include the genetic sequence "primers" chosen to start the reaction off, the probes used to analyse the results, the concentration of the various reagents used, and the temperature and time over which the reaction is run. False positives result from cross-contamination of samples and reagents, and "carryover" of PCR products from previous reactions. The latter is the usual reason for false positives, because of the number of copies of a particular genetic sequence that it generates - anywhere between a million and a million million, for example.

Manoeuvres such as PCR, which are replete with opportunities for confusion, have had to be invoked in HIV research because, as Aids Testing says, the virus cannot be detected directly by conventional molecular biology techniques. The reason usually given for those difficulties in detection is that the virus is very inactive. The book cites estimates that generally only 1 in 10,000 to 1 in 100,000 lymphocytes in the peripheral blood of infected people are actively replicating virus. That in itself puts HIV - should it exist - in a very different category from most other viruses for which diagnostic testing has been developed.

"HIV-positive" is also a very different diagnosis from most in that HIV and Aids have been invested with so much fear and prejudice. "No other disease, except perhaps leprosy from the beginning of the millenium to the 1800s, has brought so much societal pressure on its victims and caused such cataclysmic social consequences over such a long period," says Dr Ann James, another of the contributors to Aids Testing.

It is a rare individual who has the strength and good fortune - and good humour - enabling them to fly over the cuckoo's nest of an HIV diagnosis. One such is Christine Maggiore, from Los Angeles, who told her story in the March/April 1994 issue of Rethinking Aids, the newsletter of the group seeking a reappraisal of the HIV/Aids hypothesis. She wrote:

I am writing to thank you for your commitment to reason, science and truth amidst all the compromised interest and hysteria that surround Aids, and to let you know of my own adventures in the wacky world of HIV.

I was diagnosed as positive in March of 1992 after having been annoyed into an antibody test by a gynaecologist who relentlessly recommended testing to all her new patients. I finally consented even though I had been in a mostly monogamous relationship for over seven years, did not fit into any of the groups described as high risk and had even tested negative in 1990 to qualify for life insurance. The news that I was suddenly positive devastated me. I knew nothing about HIV except that everyone that had it was supposed to die and end up on a quilt.

I immediately saw an Aids specialist who had me retake the test since only two of the seven bands on the western blot were reactive. A week later, for reasons he never explored or explained, the test was completely reactive. At that point I was considered positive beyond a doubt and began to live as if I were going to die within the five to eight years officially allotted me. Fortunately I had (and have) a high T-cell count, so none of the many doctors I went through pushed me to start AZT or any of the other alleged antivirals.

Since all the experts I met with told me there was nothing I or anyone could do to halt the inevitable and unattractive demise of my immune system, I started looking for solutions on my own. This led me out of mainstream medicine and away from the typical Aids-think supported by those depressingly helpful groups and foundations.

Somehow I stumbled upon Dr Duesberg's telephone number (I had no idea who he was or that the entire world was mad at him) and the conversation we had made a tremendous and irrevocable impact on my life. Mostly because he told me I could expect to have one [a life] if I stayed away from AZT and other such drugs. He generously sent me copies of all his writings and news articles which I reviewed with the same sceptical interest I now applied to everything about this subject. What I understood shocked and amazed me: HIV was a runaway hypothesis that grant money had turned into a belief system and that science had never bothered to substantiate! No wonder all those doctors made no sense to me - there was no sense to be made of any of this!

She read books and contacted alternative organisations that were challenging HIV and Aids, and started to think differently about what Duesberg referred to as a "boring" retrovirus "that seemed to be doing nothing but make finding dates and insurance coverage difficult".

I began to wonder what, if anything, was going on in my body one year into "HIV disease"...

I took all my tests all over again and found I still had the same T cell count but was now considered "undetermined" for HIV: certain bands previously positive were now negative. I had a PCR done to determine what undetermined might mean and that came up "detected". Curious as to how an apparently diminishing virus was detectable, I followed the PCR with another antibody test. The result was completely negative. My amazing seroconversion then mysteriously reseroconverted to positive on the next visit to the lab.

I wish I could say I find antibody testing an entertaining pastime or that the varying results mean nothing to me. It is, instead, an obnoxious and confusing nightmare no doctor I know can explain. Depending on the week and according to the accepted wisdom, I am or am not dying and may or may not be considered an object of pity and/or fear. The whole thing sucks.

Does your group have any clue as to how tests can vary to such degrees and is there anyone else out there with a similar experience? Thank you again for your continued existence and for allowing me to be cranky about this in a public forum.

What the "HIV" test means

Despite cases like Christine's, the antibody test does tell us something in people who are clearly and repeatedly positive. "When you are ill, especially when you have an infectious disease, you have a lot of antibodies to a variety of things," Eleopulos says. "When a foreign substance comes into your body, and you make antibodies to it, these do not react only with that substance, but with other substances too. So the more antibodies you have in your body, the higher the probability that you will react to a substance to which you have never been exposed. The Aids risk groups - the gay men, the illicit drug users, the haemophiliacs - have millions of antibodies, because they are exposed to so many antigens. So whether or not they have HIV in them, they are more likely to test positive for the antigens in the HIV test. The same is true for Africans. Black people, even the healthy ones, have much higher antibody levels than whites, and on top of that they are often poor, they suffer TB, malaria, and many other infectious diseases. But if you test a young American military recruit who has never been exposed to foreign antigens, it is very unlikely that his blood will react. The HIV antibody test can be used as a marker of risk for the development of Aids, but not as proof of HIV infection."

A year after Bio/Technology published Eleopulos's review article, powerful confirmation of its conclusions was provided by a study headed by Dr Max Essex, of the Harvard School of Public Health, one of the originators of the hypothesis linking HIV with Aids and a leading exponent of the theory that the virus originated in Africa. Essex had been working with scientists from the University of Kinshasa and the health ministry in Zaire to see whether leprosy patients, and those in close contact with them, were at increased risk of being infected with HIV.

The results, published in the Journal of Infectious Diseases, were remarkable. Out of serum samples from 57 leprosy patients, 41 tested positive using one Elisa kit, 39 using another, and 37 (65 per cent) by both. Among the sera from 39 contacts, the figures were 12, 10, and 9 (23 per cent) respectively. When the sera were tested with western blot (WB) and radioimmunoprecipitation analysis, however - tests which, because of their expense, are never normally performed in Africa ­- only two of the leprosy patients and none of the contacts were confirmed as positive. Even those two could be considered false positives, because the diagnosis was made on the basis of reactivity with two of the three bands gp160/120 and gp41, and as we have seen, the gp160/120 proteins are normally just multiples of the gp41 band.

Most of the blood samples reacted with the proteins on the WB strips, but not strongly enough to provide a positive result. These "indeterminate" reactions were even seen in 85 per cent of patients, and 57 per cent of contacts, who were negative with the Elisa tests. In normal individuals used as controls, however, only 2½ per cent gave a positive WB reaction.

Laboratory investigations indicated that proteins from Mycobacterium leprae, the microbe responsible for leprosy, were causing many of these cross-reactions and false positives, both with western blot and Elisa. P24 was the most frequently recognised "HIV" protein in the WB test, but cross-reactivity also occurred with all the others. M. leprae might have this potential "since the disease it causes is associated with an immunodeficiency that resembles HIV-1 in several respects," the researchers said. "In addition, the immune dysregulation induced by M. leprae is often accompanied by the production of autoantibodies to numerous cellular proteins."

They concluded that leprosy patients and their contacts "show an unexpectedly high rate of false-positive reactivity of HIV-1 proteins on both WB and Elisa." Since M. leprae shared several antigens with other members of the mycobacterial family, including M. tuberculosis, the agent responsible for TB, "our observations of cross-reactivity...suggest that HIV-1 Elisa and WB results should be interpreted with caution when screening individuals infected with M. tuberculosis or other mycobacterial species. Elisa and WB may not be sufficient for HIV diagnosis in Aids-endemic areas of Central Africa where the prevalence of mycobacterial diseases is quite high."

"Quite high" is an understatement. According to a WorldAIDS briefing paper published by the Panos Institute in September 1992, about one third of the world's population are latently infected with TB, and at any one time between nine and 11 million people are suffering from the active infection - 95 per cent of them in Asia, Africa, and Latin America. "In Africa, TB has already become the prime cause of death in adults with HIV," the paper said. "A recent study in the Ivory Coast showed that 35 per cent of adults with HIV died of TB. 'TB is clearly the most important HIV-associated disease in Africa', Dr Sebastian Lucas, author of the Ivory Coast study, told the 8th International Conference on Aids in Amsterdam in July. 'Since Africa has 65 per cent of all cases of HIV, that makes TB the most important Aids-associated infection in the world." According to Panos, "the established epidemic of TB and the new epidemic of HIV have shown a disturbing tendency to coalesce andto co-infect individuals. It is a dangerous liaison both for those who are co-infected and for those communities in the developing world at risk of TB."

It seems clear from the Zaire study that this "epidemic of TB/HIV co-infection", as WHO has taken to calling it, is another of the tragic errors created by the non-specificity of the "HIV" test. People with active TB infection are at greatly increased risk of testing positive because of M. tuberculosis, not HIV.

When I wrote a news report on the study in May 1994, I tried several times to speak to Max Essex to explore the implications of the research further with him, but he was unwilling to come to the phone. I did however quote Professor John Papadimitriou, one of the co-authors of the original Bio/Technology paper, as saying that the Zaire research demonstrated the inadequacy of the HIV tests. "Why condemn a continent to death because of HIV," he said, "when you have other explanations for why people are falling sick?"

An HIV diagnosis is potentially lethal in the West as well as in Africa, as the following letter to The Sunday Times, written in response to our article on the Eleopulos paper, made clear:

I read Neville Hodgkinson on the recent report questioning the reliability of the HIV test as a predictor of Aids with the anguish which he accurately predicted. The night before I had learned that yet another of my circle of friends had by virtue of being unwell and gay been subject to and had received an HIV+ result from this iniquitous procedure.

Would he read this article and would it sufficiently impress him that he need not conclude - as so many have during the last 10 years - that HIV spells untimely death?

Your newspaper has been almost alone in this country in reporting the opposition to the orthodox view that HIV causes Aids. There must eventually come a time when this lethal paradigm which has caused untold misery and untimely destruction of lives is demolished.

When I get the opportunity to meet with this friend I shall urge him to resist this dismal view of his future as do a growing number of healthy alleged HIV positives. I hope that during a period when he will be extremely vulnerable his doctor will not be pressuring him to accept medication. Especially not the highly dubious reverse transcriptase inhibitors AZT and DDI/DDC.

I shall urge him with the conviction of personal experience, having been "diagnosed" two years ago and warned that without drugs, I could expect to count life in weeks rather than years. I have since chosen to disregard that bleak counsel and the drugs and continue with my life. The intellectual justification for such obstinacy is expanding thanks to brave souls such as the authors of the report described and Prof. Duesberg, Dr Sonnabend and others who resist the pressure to subscribe to the prevailing views of the Aids industry. The empirical support can be found in the lives I am glad to say of many of my friends and acquaintances.

Much much more questioning, challenging and overturning is needed. But thank you for playing your part.

Despite the absence of scientific debate or discussion over the Bio/Technology review, it was clear that the Sunday Times reports were beginning to have an impact at the grass roots. A freelance illustrator in Yorkshire told us of how he felt life was only just beginning to return to him, four years after an HIV diagnosis had robbed him of his confidence, self-esteem and drive. On the day his HIV-positive partner died of pneumonia, he was told he was positive too and had probably two years to live if he took AZT - 18 months if he did not. He went home thinking about how conventional treatments had failed to save his partner's life, and of side-effects friends were suffering.

I had lots of friends who were HIV-positive but absolutely fine until they started taking the drugs. Everyone thought I was barmy for refusing medication but time has confirmed my suspicions.

I knew the time would come when the validity of the tests themselves would be questioned. Now that has happened, I am so furious I feel like suing the doctors and the drug companies for stealing four years of my life.

When I had to tell people I had tested positive, I was thrown out of the house I had been sharing with my partner; I lost my self-esteem, my sense of direction and my drive and subsequently suffered from depression.

Things got so bad at one time, I thought about suicide. I am angry when I think about the people who have killed themselves because of a so-called positive test.

I realise the doctors have a difficult job to do, but someone has to question the whole empire that has been built up around the industry before even more people are killed through ignorance.

There is a growing element of defiance, with more and more people flushing their drugs down the toilets and saying "Look, I'm still here and I've been so-called HIV-positive for years, so what is going on?"

It would be nice though to hear something like regret or an apology from the medical profession instead of the deafening silence out there now.

We also learned of the difficulties faced by care workers and others in knowing how best to advise those diagnosed HIV-positive. One health adviser specialising in HIV/Aids said:

I would hate to be newly diagnosed as HIV-positive because the whole thing has become horrendously confusing. Five years ago, I would have told clients with a low T-cell count they should take the drugs, now I tell them they have to decide for themselves but that they should question the doctors so they are making an informed choice.

Having said that, there is no independent body to give them the advice and no forum for the wider debate of all the issues raised so how they are supposed to make an informed choice, I don't know.

I am amazed at the increase in the number of people diagnosed as HIV-positive who are questioning mainstream thinking from the start but if I want to keep my job, I cannot be seen to be rocking the boat. I feel I am walking on eggshells all the time because there are so many things it is politically not acceptable for me to be telling clients.

There is this terrible fear throughout the field that it will be discovered we have been walking down the wrong route all these years and advising people to do things which will turn out to be the wrong things. I think about quitting all the time.

That was a good reminder, amidst the anger that was beginning to be aroused among patients as the inadequacies of the HIV theory became more widely known, that people in the "Aids industry" were also sometimes suffering greatly from the pathological science affecting their field of employment.

It was and is wrong to tell people they are carrying a deadly new virus on the basis of an unvalidated test, beset with technical problems and pitfalls in interpretation, vulnerable to shipping, climatic and storage conditions, and subject to unmeasured and probably immeasurable cross-reactivities and hence false positive results. It is very hard for doctors, scientists, politicians, the World Health Organisation, gay leaders, Aids charities and even journalists to admit to this today, since they have all been instrumental in bringing about the climate of opinion in which this unvalidated test was inflicted on millions. But those are the facts. Regardless of whether or not the test has any relevance to a retrovirus, there are so many other possible causes of a positive result that on present knowledge, no one should be diagnosed as suffering from "HIV" infection or disease. No one cognisant of these facts will ever wish to allow themselves to be tested. The sooner the error is acknowledged and the test relegated to history, the quicker we may see a return to sanity in Aids science.