Note: Unpublished letter to Lancet with professor Gordon Stewart.

REPLY TO NUNN ET AL.

SIR: In the study (27th September), Nunn, Mulder and their colleagues claim to have proven that HIV-1 infection in Ugandan subsistence farmers is associated with high death rates, and in a second paper in the same issue, Evans and Mortimer et al claim that a commercial test kit gave "false negative results" in an "HIV antibody assay". The minimum scientific requirement for such claims are proof that (a) the antibody tests are HIV specific; (b) the HIV seropositive individuals died from AIDS.

According to Griner and colleagues, "The sensitivity of the test is determined by identifying the proportion of patients with disease in whom the test is positive…the specificity of the test is determined by identifying the proportion of patients without disease in whom the test is negative". When the "disease" is HIV infection, in order to determine the sensitivity and specificity of antigen/antibody reactivity for the "disease", one must (a) test a large number of individuals both sick and with and without AIDS, as well as those who are healthy and/or at no risk of AIDS; (b) simultaneously verify the presence or absence of HIV infection by employing an independent gold standard, that is, HIV isolation. However, according to one of the best known of all HIV/AIDS experts, William Blattner, for HIV this has not and could not be done. For unknown reasons the authors of the above studies determined the specificity of the HIV antibody tests by repeating the same or a combination of different antibody tests an arbitrary number of times or by using one antibody test as a gold standard for the other(s). This process is no different from claiming that the nature of a particular chest X-ray or ECG finding can be resolved by repeating the test on several different machines. Concordance of a number of antibody tests could be due to nothing more than reactivity occuring between the same or different antigens and the same or different antibodies in the sera tested. In the algorithm used in the first study, the true serostatus depended upon repeating two different ELISAs until they were concordant, an outcome which could have resulted from equally erroneous tests. A positive ELISA followed by a positive Western blot (WB, no criteria given) was also regarded as proof of HIV infection. In the second study a series of three ELISAs was considered a gold standard for HIV infection and was used to determine the sensitivity of another assay and the authors advised that antibody tests "should be evaluated using many known [antibody] positive specimens both by the manufacturer and by an independent laboratory".

In addition to violating the fundamental scientific principles of antibody test evaluation, the algorithm used in both studies raises another problem. In the vast majority of countries/institutions/laboratories, no positive HIV ELISA tests, no matter how many times repeated and concordant, are considered proof of HIV infection. In the second study, in addition to using ELISAs, the sera were tested with the Western blot, a test which, according to one of the study authors, "began as and should have remained, a research tool". Furthermore, in the only patient for whom the WB band pattern is given, there are only two bands, p24 and p160. However, at present in the majority of countries/institutions/laboratories, including Africa, Germany, France, the US Red Cross, the US Food and Drug Administration and the Australian National HIV Reference Laboratory, this result is regarded as "indeterminate" and caused by cross-reacting antibodies.

Since:

(a) no lesser authorities than Myron Essex and his colleagues have shown that "leprosy patients and their contacts show an unexpectedly high rate of false-positive reactivity of HIV-1 proteins on both WB and ELISA". The cross-reactivity was found to be caused by antibodies directed against two major carbohydrate-containing M. leprae antigens--phenolic glycolipid I and especially lipoarabinomannan, an arabinose-containing lipopolysaccharide which is also present in M. tuberculosis and other mycobacteria leading these authors to warn that, "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";

(b) According to Mortimer, in all parts of the world, "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 (e.g. HIV-2) which, though also associated with AIDS, are different viruses"; © the causes of death reported in the first study are unknown;

(d) tuberculosis (TB) is a protracted epidemic in Africa and it is estimated that of the 661 million people in sub-Saharan Africa, 2-3 million have active TB with an annual mortality of 790,000 and that in adults, "HIV infection" usually follows TB infection, and that 30-50 per cent of African "AIDS" deaths are from TB;

(e) unlike the AIDS definition for Western countries, the World Health Organisation Bangui definition for AIDS in Africa does not require immunological (T4 lymphocyte cell or antibody) tests or a specific disease diagnosis but consists largely of symptoms where, for example, an African with diarrhoea, fever and persistent cough for longer than one month is by definition an AIDS case. However, these symptoms are common, non-specific manifestations of many diseases endemic in Africa and were so long before the AIDS era;

(f) two of the diseases listed in the Bangui definition, cryptococcal meningitis and candidiasis, are caused by fungi, organisms which may induce antibodies that react with the "HIV" proteins;

(g) mycobacterial and fungal diseases are known to be associated with immune deficiency; one would expect Africans to have appreciable rates of positive HIV antibody tests; the death in these individuals including death from AIDS to be higher than in seronegative Africans without ever being infected with HIV.

Unless the authors of these studies provide valid scientific reasons for their methodology, the credibility of their conclusions is gravely undermined. *