By Val Turner

March 2002

View the 83 slides, and listen to the audio stream (Real audio, 65 min.), or read the transcript below.

This presentation has been prepared by Eleni Papadopulos and the Perth Group and several other colleagues. The subject is an analysis of the data claimed to prove nevirapine an effective agent for the prevention of mother to child transmission of HIV. The presenter is Dr. Val Turner from the Department of Emergency Medicine, Royal Perth Hospital.

Slide number 2 please

Members of the audience may wish to note that this presentation is an amplification of one part of our recently published monograph. Unfortunately time does not permit an analysis of every possible factor and most unfortunately of all, the data said to prove heterosexual transmission. Such proof is obviously a prerequisite for any scientific evaluation of this topic. Members of the audience who wish to study this and other relevant aspects are cordially invited to visit the link displayed at the bottom right hand side of the slide.

This presentation will focus on the HIVNET 012 study published on the 4th of September 1999 in the Lancet. However, before we do that, there are several matters we need to address.

Slide number 3 please

Any doctor, institution or indeed any government contemplating the use of particular drug for a specified disease is placed under an ethical obligation to be satisfied, beyond all reasonable doubt, that patients are definitively diagnosed with the disease in question, that the drug under consideration is capable of producing the intended result, and that the anticipated benefit is far outweighed by any undesirable effects and toxicities the drug may possess.

Slide number 4 please

The diagnosis of HIV infection is performed differently in mothers and infants.

Mothers are said to be HIV infected if a blood sample contains antibodies which react with particular proteins deemed unique to a retrovirus HIV. These antibody reactions may be demonstrated using either an ELISA or Western blot technique. We will not be discussing the ELISA as most experts do not regard reactive ELISAs as proof positive of HIV infection. That is, the ELISA lacks specificity. Rather we will focus on the Western blot which is said to be virtually 100% specific and if positive, proof of HIV infection.

Slide number 5 please

The identity of antigens used in HIV antibody tests is based on research published in 1983 by Montagnier and in 1984 by Gallo. Both scientists and their respective colleagues claimed to have isolated HIV from co-cultures of tissues of AIDS patients and achieving purification by passing supernatants of these cultures through a sucrose density gradient. The material which banded at a density of 1.16 gm/ml was claimed to be the QUOTE �purified virus� UNQUOTE, that is, purified HIV.

Although leading retrovirolgists had agreed at least a decade earlier that electron micrographic confirmation of the existence and purity of retroviral-like particles was an essential component of retrovirus isolation, neither Montagnier nor Gallo published electron micrographs to justify their respective claims. Thus at the time HIV was discovered, and in fact for the next fourteen years, it was not possible to conclude that �purified HIV� contained any particles, be they retroviral, viral or of any other morphologies, pure or impure. The EMs that were published in the Montagnier and Gallo papers, not of �purified virus� but of unpurified cultures, revealed a few particles with the appearances of type-C particles. These were said to be HIV yet nowadays HIV is classified as different genus. That of a Lentivirus.

Regardless of the omission of electron micrographs, both groups designated certain proteins in the banded material as HIV in origin, that is, HIV specific. The proof for this claim is not extraction from retrovirus particles but that from the many proteins present in the 1.16 gm/ml band these were the few proteins that reacted with antisera from AIDS patients.

Slide number 6 please

In an interview conducted on July 18th 1997 at the Pasteur Institute, Montagnier stated that analysis of the HIV proteins demands mass production and purification of HIV, but�

Slide number 7 please

That he and his colleagues did not purify HIV.

Despite these obvious scientific difficulties, laboratory scientists used these proteins banding at 1.16 gm/ml to develop antibody tests for proving HIV infection. The veracity of this procedure remains virtually unchallenged. Not even the discovery of HIV proteins in non-AIDS-related tissues, including tissues from healthy, no-risk individuals, led any HIV expert to question the validity of these tests. Certainly, the majority of clinicians appear unaware of these findings.

Slide number 8 please

From this abundant evidence here is an example we might consider germane to the present problem. Three of the most significant HIV proteins are found in normal placental tissue from healthy, non-HIV-infected women.

Slide number 9 please

And to continue this theme, from the very beginning and at present, Montagnier regards p41 not as an HIV protein but the ubiquitous cellular protein, actin.

In 1989 Pinter and colleagues proved that the p120 and p160 proteins in the HIV Western blot are not distinct HIV proteins but oligomers of p41. In another paper researchers expressed concern about mistaken diagnoses based on the mistaken belief that p120 and p160 are individual HIV proteins.

One should note that on this basis an African who has Western blot bands corresponding to the location of any two of the p41, p120 and p160 proteins, in reality has antibodies reactive with his own actin. Yet these are the criteria for a positive Western blot in Africa.

Let us return briefly to the Montagnier interview. This was conducted by Djamel Tahi and the text later published in the English magazine Continuum. There are videotapes of this interview in circulation. Eleni Papadopulos and I brought a copy to the Presidential AIDS Advisory Panel which we gave to Professor Mhlongo.

During the interview Professor Montagnier was asked why he did not publish electron micrographs of his �purified virus�.

Slide number 10 please

Montagnier�s response to this question is staggering. He replied that despite what he called a �Roman effort� no one at the Pasteur Institute could find particles in the just discovered �purified virus� that had the appearances of retroviral particles.

Pressed further he did not accept that Gallo�s laboratory had purified HIV.

In the same year two groups of researchers, one a European collaboration and the other American, provided the first electron micrographs of what has long been assumed to be �purified HIV�, as well as additional evidence that the HIV proteins have a cellular origin.

Slide number 11 please

In the top two micrographs we see, and for the first time 14 years after its discovery, what purified HIV actually looks like. In other words, these are the first published electron micrographs of the 1.16 gm/ml sucrose density gradient, banded material from which proteins and nucleic acids are obtained for use as diagnostic reagents. One does not need more than a glance, and one certainly does not need to be a scientist, to know that regardless of how this material is constituted, it is not pure. The authors themselves admit this and in fact labelled the top two pictures as �purified vesicles� and not purified HIV. Despite this, they still claimed that �purified vesicles� contain a few particles which are HIV. However, these particles, indicated by the arrow, do not have the requisite morphology and two such particles are present towards the right in the lower picture, which is material similarly obtained from a non-HIV-infected culture.

Slide number 12 please

This slide is taken from the second, the US study, published by a group led by Dr. Julian Bess, under the auspices of the National Cancer Institute, which provides material for the US Vaccine program. It is a gel electrophoresis of the proteins in the 1.16 gm/ml band.

Lane A is the pattern obtained from a non-HIV-infected peripheral blood mononuclear cell preparation which, under the EM, reveals cellular material similar to the previous slide. Bess and his colleagues refer to this material as �cellular microvesicles�.

Lane B is an HIV-infected, malignant cell line, the H9 cell line, which under the EM, also consists of cellular microvesicles and other cellular debris including a number of particles with diameters twice the HIV particles in the previous slide and twice the diameter of any known retrovirus particle, as well as lacking other morphological features of retrovirus particles. These are Bess and his co-workers� �HIV�.

Lane C is an HIV-infected H9 clone with similar appearances.

As far as the protein patterns are concerned, there is only one pattern. There are no qualitative differences between the three strips. Although there are darker bands, that is, quantitative differences between the three, including differences between Lanes B and C, which are said to be infected with the same virus, the same protein bands exist across all three cultures. Significantly there is a prominent actin band present in all three cultures in the region where we would expect to find p41, which appears to be missing, and in the region molecular weight of 120,000, there is the same band in all three cultures.

The quantitative differences between Lanes A, B and C are readily explicable in terms of the nature of the immortalised cell lines used in Lanes B and C, and differences in the histories of preparation of the cultures.

Since all the proteins are present in all the strips we emailed Dr. Bess and asked the reason for the labels, p24, p17 and p6/p7, adjacent to Lane C. Dr Bess informed us that the labels were added at the suggestion of the editor in order to orientate the reader. Dr. Bess and his colleagues did not prove the origin of these proteins or that they were HIV proteins.

We presented this slide during our presentation at the July 2000 Presidential AIDS Advisory Panel meeting. Dr. William Makgoba took us to task claiming expertise at interpreting gel electrophoresis patterns. Since Dr. Makgoba accused us of misrepresenting Dr. Bess�s data, after the meeting we sought Dr. Bess�s comments on what Dr. Makgoba had said.

Slide number 13

Dr. Bess agreed with us

QUOTE ��you can come to the conclusion from gel electrophoresis patterns that there are only quantitative differences between HIV and [cellular] microvesicles�, that is, between cellular proteins and HIV proteins� END OF QUOTE.

In conclusion, there is ample evidence that the antigens used in the HIV antibody tests are not proteins belonging to a unique retrovirus HIV but are in fact cellular proteins.

Slide number 14

Here are the proteins as they appear in the HIV Western blot strip. Although not in electrophoretic order. What we now need to consider is why patients may have or develop antibodies which react with these proteins, and why is there a correlation between these reactions and certain antediluvian diseases which, for the past two decades, have constituted a new syndrome defined as AIDS.

We also need to bear in mind why these antibody reactions should be so particularly prevalent in Africans compared to say Americans, and why almost equally in African women and men, but in American women and men.

Slide 15

One possibility, for which there is abundant evidence, is that the antibodies are antibodies directed against cellular proteins. In other words, autoantibodies. This is a list of a few of the autoantibodies that have been found in the sera of AIDS patients. A MEDLINE search will reveal many more.

But there are other non-retroviral reasons to account for a positive Western blot test.

Slide 16

Despite the acronym acquired immune deficiency, AIDS patients typically have high levels of all antibodies. That is, they have hypergammaglobulinaemia. Given that all antibodies have propensities for cross-reactivity, indeed this is the explanation AIDS experts provide to account for reactive but not positive ELISAs, as well as non-diagnostic or indeterminate Western blots, there is every reason to hypothesise the plethora of antibodies present in AIDS patients, with their attendant cross-reactivities, could account for many if not all positive Western blot tests, including those caused by autoantibodies. There is certainly no evidence to counter this possibility. In fact hypergammaglobulinaemia predicts HIV seropositivity.

AIDS patients are also infected with many infectious agents. For example, fungi and mycobacteria account for nearly 90% of all AIDS diagnoses. There is also plenty of evidence that antibodies directed against these agents react with the HIV antigens, both the so called envelope and internal proteins.

No doubt most are aware of the data published by Kashala from Uganda in 1995 which led to caution using the HIV ELISA and Western blot in mycobacterial prevalent areas. Significantly, in their paper Kashala and his coworkers published a series of Western blots from non-HIV-infected leprosy patients which, if using the stringent Australian criteria, would be reported HIV-infected.

These observations lead to the prediction that amongst individuals who are sick, that is, amongst those who have reasons for producing antibodies following exposure to either infectious or non-infectious agents, we will find reactions in the HIV Western blot, although these individuals do not belong to an AIDS risk group. Such is the case.

Slide 17

These are the results of a never followed up, never repeated, study from the US published in the New England Journal of Medicine in 1990.

HIV antibody tests including the Western blot were performed on 89,547 blood samples collected from patients at 26 US hospitals over a six month period. The patients were meticulously selected to avoid testing any patient in an AIDS risk group or with AIDS. The HIV seropositive rate at some hospital was impressive. Up to 21.7% of men and 7.8% of women aged 25-44 years were found to be positive.

In terms of retrovirus which in the US in 1990 was, and still largely is, restricted to certain identifiable groups, and predominantly men, these data make no sense. Why are the seroprevalence rates so high in no risk individuals and why are a third of positive tests in women?

In our view, these data represent the authentic explanation for a positive HIV test in Africa. That is, illness from a large variety of causes which are not retroviral. The only differences between the US and Africa is that in America diseases are not so prevalent while hospitals are.

Slide 18

Even if we discount everything said so far, as physicians, how can we accept the assertion of the Centers for Disease Control, that the HIV Western blot possesses �extraordinary� specificity. How can we reconcile this statement when the criteria for a positive test vary so enormously between countries and institutions, and even between laboratories in the same city?

How can a man with two antibody bands be HIV infected in New York City but not in Sydney, Australia?

How can an African man be positive with a p41, p120, p32, p24 band in Kinshasha while his brother or sister in Australia would not be positive with the identical bands.

This slide illustrates 11 different sets of criteria for diagnosing a positive Western blot. Can any of us imagine 11 sets of criteria for the diagnosis of myocardial infarction on an electrocardiogram? Or tuberculosis on a CXR? Is it possible to have a heart attack or TB in the UK but to have this negated merely by crossing the English channel?

How can doctors practice medicine under such circumstances? How can public health officials compare data? And most importantly, how can we subject mothers and babies to these tests and claim proof of heterosexual and mother to child transmission?

Slide 19

This dilemma could have been solved many years ago if scientists had validated the antibody tests against a gold standard.

The only scientific gold standard for proving the specificity of an antibody test for HIV infection, is HIV itself. This means performing an experiment to compare the presence of absence of antibody reactivity with the presence of absence of what we wish to measure. That is, the virus, HIV, as determined by isolating it.

But there is no such evidence and at present could not be because no one has presented evidence for HIV isolation. Rather, when we analyse what HIV experts such as Montagnier and Gallo actually present as isolation, the data consist of a collection of non-specific findings, including antibody/antigen reactions, all of which have non-retroviral, non-viral and other non-microbial causes, and which have been reported from material which does not even contain retroviral-like particles.

Slide 20

What may a scientist conclude?

We agree with the HIV/AIDS experts that a positive HIV antibody test is a risk factor for the development of illness, at least in the AIDS risk groups. Our disagreement is the underlying cause of a positive test. We have argued the tests are non-specific and should be regarded in the same manner as an elevated erythrocyte sedimentation rate or C-reactive protein. Physicians find these to be extremely useful investigations. The ESR for example predicts many diseases and, like serial antibody titres, response to treatment. But no one imagines that diseases such as tuberculosis or osteomyelits are caused by red cells clumping together.

In our view in the mid 1980s laboratory scientists serendipitously discovered an ESR-like test but in their haste to find the cause of a new syndrome, and in disregard for scientific principles, recommended its general use as a diagnostic agent for a retrovirus which was never isolated. And which in reality exists only because of this antibody test.

Slide 21

Regardless of how HIV antibody tests are interpreted in mothers, there is an additional problem in children.

Everyone accepts that maternal IgG immunoglobulin is transferred via the placenta from mother to infant and increasingly in the latter weeks of pregnancy.

Slide 22

In this slide the tent shaped, dashed line illustrates that at delivery, infant IgG antibody levels approximate maternal IgG levels and thereafter decline to zero over the next several months as maternal IgG is catatabolised. The disappearance of the mother�s IgG antibodies has an an exponential decay with a half life of approximately 30 days. By 9 months of age the mother�s antibodies have totally disappeared from her child�s circulation.

Slide 23

This slide is taken from a study published in the Lancet in 1988 and is the only study to report the disappearance of infant HIV seropositivity over time. There were 271 children in this study, all born to mothers from 8 centres in Europe.

We see that because of the presence of maternal IgG, 100% of offspring of all HIV seropositive mothers are seropositive at birth.

By nine months of age approximately 25% of the children have seroreverted. This is explained by the disappearance of maternal HIV antibodies from the infant�s circulation. By 21-22 months 15% have not seroreverted which HIV experts explain as infant HIV antibodies representing the proportion of infants infected by their mothers. However, 60%, over half the children, serorevert between 9 months and 21-22 months. This cannot be explained by the loss of maternal HIV antibodies in the absence of biochemical evidence that the metabolism of maternal HIV IgG is selectively prolonged and takes over twice as long as all other antibodies of this class. This leaves two possibilities. The first is that the infant synthesises and then loses HIV antibodies which that untreated infants are able by some means to cure themselves of HIV infection. The second is that the are antibodies are synthesised by the infants, they react with the HIV antigens but they are not HIV antibodies. In other words, whatever their origin, they react non-specifically in the ELISA and Western blot. If this is the case then the same can be argued for all the infants and also all their mothers and fathers.

These data cannot be explained by persistence of IgA HIV antibodies acquired by the infant from breast milk. For no other reason than in Europe only 20% of women breastfeed beyond six months yet 60% of seropositive infants serorevert after this time.

In our view this argument alone is sufficient to call into question the whole HIV theory of AIDS.

Slide 24

HIV/AIDS experts have circumvented what they see as the maternal antibody problem by recommending infants are diagnosed by detecting or measuring HIV RNA or DNA using the PCR.

A few of the many problems using this approach are listed on this and the next few slides.

We have already seen that as with the HIV proteins, nucleic acid primers and probes, said to be those of a unique retrovirus, are not obtained from purified retroviral particles.

In fact, the material from which nucleic acids are obtained is anything but pure but more importantly, it does not contain particles bearing the morphology typical of retroviruses.

Even if these particles were proven to be a retrovirus, there are no data demonstrating that the nucleic acids in question originate in these particles.

And regardless of their origin, there is no proof that positive tests which depend on these primers and probes, are specific for HIV infection.

Slide 25

These problems of the HIV PCR can be summarised by a study published by Owens and his colleagues in 1996. This in-depth meta-analysis concluded that when the HIV PCR is compared against serology, not HIV isolation which should be the case, the false positive rates are QUOTE �too high to to warrant a broader role for PCR in either routine screening or in the confirmation of diagnosis of HIV infection. This conclusion is true even for the results reported from more recent, high-quality studies that used commercially available, standardized PCR assays...We did not find evidence that the performance of PCR improved over time� END OF QUOTE

Slide 26

Virologists, such as Ian Christie from the UK Public Health Laboratories, lend further weight and strongly advise that PCR should NOT be used to confirm HIV infection.

Slide 27

The Centers for Disease Control in the United States have what can only be described as a doubly shizophrenic attitude to HIV diagnosis in infants. Firstly they assert that their original and four times revised AIDS definition does not apply to the diagnosis of HIV infection, as if there are two separate processes involved, one for counting the number of HIV/AIDS individuals for public health purposes and another for individual diagnosis. One would assume these statistics are identical.

Slide 28

Second, the CDC forbids the use of RNA PCR to diagnose HIV infection in adults and adolescents and children infected by all other means except perinatally.

In other words, RNA cannot be used to prove infection of an infant by blood transfusion, but it is the method of choice to prove infection transmitted via the mother.

Slide 29

As recently as December 14th 2001, 74 experts in Pediatric HIV/AIDS, representing The Working Group on Antiretroviral Therapy, The National Pediatric and Family HIV Resource Center, The Health Resources and Services Administration and The National Institutes of Health,advised that the specificity of the HIV RNA PCR is unknown. Yet this is the test used in many studies of mother to child transmission including the HIVNET 012 study.

Slide 30

If we dismiss every problem discussed so far, how much confidence can a physician or a patients place in a test when three different PCR techniques, the three columns on the right, carried out on the same quantity of virus, the two columns of the left, yield results varying almost a millionfold? If the RNA PCR had anything the recommend its use at least the minimum we would expect the numbers in the rightmost three columns all be of the same order.

There is no scientific proof of a retroviral origin for the PCR RNA reagents, we have scientists such as Owens and his colleagues warning that the PCR test parameters are extremely dubious, we have practising virologists recommending that the PCR is NOT to be used to confirm HIV infection, we have the CDC claiming a test not to be used in adults is perfectly acceptable in perinatally but not otherwise infected children to diagnose the same virus and finally,

Slide 31

we have Roche, the manufacturer of the only licensed PCR in the US, including a caveat against the use of their test in their packet insert.

Yet such tests are the raison d�etre of mother to child transmission and underlie for example the recommendation that the South African Government provide nevirapine to all pregnant HIV positive women as well as their babies.

Slide 32

Let us now consider the data on nevirapine,

Slide 33

We begin with drug efficacy.

The most reliable evidence regarding the effects of a drug on a disease are obtained by conducting randomised, double blind, placebo controlled clinical trials.

Slide 34

The HIVNET 012 Ugandan study was preceded by a phase I, phase II trial called the HIVNET 006 study. The authors of this study are substantially the same authors of the HIVNET 012 study.

HIVNET 006 studied the safety and pharmacokinetics of nevirapine in 21 Ugandan women.

Cohort one consisted of 8 women who received 200 mg of nevirapine �when in active labour�.

Cohort two were 13 women similarly treated whose infants were given nevirapine, 2mg/Kg �at 72 h of age�.

Slide 35

Musoke P et al. (1999). A phase I/II study of the safety and pharmacokinetics of nevirapine in HIV-1-infected pregnant Ugandan women and their neonates (HIVNET 006). AIDS 13:479-86. OBJECTIVE: To determine the safety, pharmacokinetics, tolerance, antiretroviral activity, and infant HIV infection status after giving a single dose of nevirapine to HIV-1-infected pregnant women during labor and their newborns during the first week of life. DESIGN: An open label phase I/II study. SETTING: Tertiary care hospital, Kampala, Uganda. PATIENTS AND INTERVENTIONS: Nevirapine, 200 mg, was given as a single dose during labor to 21 HIV-1-infected pregnant Ugandan women. In cohort 1, eight infants did not receive nevirapine whereas in cohort 2, 13 infants received a single dose of nevirapine, 2 mg/kg, at 72 h of age. OUTCOMES: The number and type of adverse events; nevirapine concentrations in the plasma and breast milk; maternal plasma HIV-1 RNA copy number before and up to 6 weeks after delivery; and HIV-1 infection status of the infants were monitored. RESULTS: Nevirapine was well tolerated by women and infants; no serious adverse events that were related to nevirapine were observed. Median nevirapine concentration in the women at delivery was 1623 ng/ml (range 238-2356 ng/ml); median cord/maternal blood ratio of 0.75 (0.37-0.93). The median half-life in women was 61.3 h (27-90 h) and the transplacental nevirapine half-life in infants who did not receive a neonatal dose was 54 h. The median half-life after a single dose at 72 h in infants was 46.5 h. During the first week of life, the median colostrum/breast milk to maternal plasma nevirapine concentration was 60.5% (25-122%). The median nevirapine concentration in breast milk 1 week after delivery was 103 ng/ml (25-309 ng/ml). Plasma nevirapine concentrations were above 100 ng/ml in all infants from both cohorts tested at age 7 days. Maternal HIV-1 RNA levels decreased by a median of 1.3 logs at 1 week postpartum, and returned to baseline by 6 weeks postpartum. Detectable plasma HIV-1 RNA was observed in one out of 22 (4.5%) infants at birth; three out of 21 (14%) at 6 weeks; and four out of 21 (19%) at 6 months of age. CONCLUSION: The administration of a single dose of nevirapine to women during labor and to their newborns at 72 h was well tolerated and showed potent antiretroviral activity in the women at 1 week after dosing without rebound above baseline 6 weeks after a single dose. The nevirapine concentration was maintained above the target of 100 ng/ml in infants at age 7 days, even in those infants not receiving a neonatal dose. This regimen has promise as prophylaxis against intrapartum and early breast milk transmission in a breastfeeding population.

Mothers were diagnosed HIV infected on the basis of the ELISA and Western blot.

Infants were diagnosed HIV infected by RNA PCR on 2 separate specimens, or if they had a reactive ELISA and positive Western blot at 18 months, OR if the infant had a single positive PCR but died.

For reasons not explained by the authors, infant infection was �confirmed� by culture but no data were reported.

We should note that although the HIVNET 006 authors refer to a single RNA PCR in a child who died as �probable infection�, the CDC classify this as �not definitely diagnosed�.

This study reported a transmission rate of 19%

Slide 36

The HIVNET 012 study was published a year later.

The study compared AZT with nevirapine. The trial also commenced with a placebo but this was dropped after 49 women had enrolled and given birth. The actual placebo, which was not identified, was used in only 19 of the 645 patients. The study did not have a non-treatment arm, it was not double blind but did claim to be randomised.

Slide 37

The drug dosage regime was the same as the HIVNET 006. The timing of of administration in the mothers was described as �at the onset of labour� and the children were given the drug �at 72 h after birth or at discharge from hospital, whichever occurred first�.

Slide 38

There were 645 mothers assigned to the HIVNET 012 study of whom 313 were assigned to the AZT arm, 313 to the nevirapine arm and 19 placebo group.

One should note that this number fell well short of the 1500 mother child pairs that the authors stated was �to investigate the safety and efficacy of oral zidovudine and oral nevirapine for the prevention of vertical transmission of HIV-1 from pregnant women to neonates in Uganda�.

Slide 39

The study reported a transmission rate at 14-16 weeks for AZT of 25.1% versus 13.1% for nevirapine with a p value of 0.0006. This efficacy of nevirapine over AZT was calculated at 48%.

Slide 40

As a result of this study, twelve international HIV experts including Barre-Sinoussi recommended this nevirapine regime as the most practical, effective and safe available for the prevention of mother to child transmission.

Its use was not only advised for HIV positive pregnant women and their babies but also for women who were not proven to be HIV positive but who lived in high seroprevalence areas, and in those women who did not have access to testing or those who refused to be tested.

Slide 41

It is our view that there are many scientific reasons to question the validity of this conclusion and these recommendations. Even more so when there are recommendations to give this drug to women whose HIV status is unknown.

Slide 42

In HIVNET 012 the diagnosis of HIV infection in infants was made by one qualitative PCR confirmed by a second quantitative PCR or culture. As in HIVNET 006, no data on culture were published and in contrast to HIVNET 006, a single RNA PCR followed by death, was elevated from �probable� to definite infection.

One must question why a test must be repeated to �confirm� infection in a live individual but not in an individual who dies. Especially when not all causes of death are given and not all deaths even in HIV infected children are caused by HIV.

One should also note that the PCR RNA test used by the HIVNET 012 authors is the Roche AMPLICOR MONITOR assay, which,

Slide 43

Manufacturer Roche asserts �is not intended to be used as a screening test for HIV-1 or as a diagnostic test to confirm the presence of HIV-1 infection� in anyone.

Slide 44

There are also problems in the manner in which the HIVNET 012 study was randomised.

Slide 45

The mothers eligible for randomisation in this study were selected from 2144 women amongst 13, 839 women attending antenatal clinics at Mulago Hospital. The authors reported these 2144 women as having a QUOTE �positive HIV-1 test� END OF QUOTE. In another section of their paper they defined their HIV testing protocol for mothers as two ELISAs followed by a Western blot. A large proportion of the women with a positive test for HIV-1 were excluded from the study. In fact 70% of the women with a positive test were excluded.

Slide 46

The reasons given for excluding 1499 women included �an indeterminate or negative western blot�. The number of such women was not specified. This exclusion criterion is extremely puzzling.

Since the HIVNET 012 study was investigating the effect of drugs on mother to child transmission of HIV one assumes that all 2144 women would not be eligible for randomisation unless were all HIV infected. Which leads us to ask what did the authors mean by a �positive HIV-1 test�?

Did they mean 2144 women with reactive ELISAs who had not had a Western blot? If so, then why were they included in the group eligible for randomisation? And why were the results of the ELISAs and Western blots not separately reported? The proportion of the 2144 mothers selected from all mothers HIV tested at the antenatal clinics was 15%, the same seroprevalence observed in Kampala during the study period. This confirms that the 2144 women had passed all steps of the author�s protocol, including the Western blot.

If this is the case then why did an undefined number of women have a second Western blot, if not further Western blots? And why did the authors not include this step in the description of their protocol and publish the results?

If an undefined number of 2144 women with a positive test for HIV-1 had an indeterminate or negative Western blot, how many of the 645 mothers randomised had a second or further Western blot and how many also had an indeterminate or negative Western blot? And how were these distributed amongst the AZT, nevirapine and placebo groups?

Even the remotest possibility that HIV indeterminate or negative mothers could have been included in the study groups negates any claim this study can make in regard to transmission.

Slide 47

There were also statistically significant differences in two of the characteristics making up the AZT and nevirapine groups.

Duration of labour was shorter in the AZT group and children born to mothers given nevirapine had lower birthweights.

Thus we must question the effective ness of the randomisation procedure.

Slide 48

The HIVNET 012 study also had several numerical inconsistencies.

Slide 49

The authors stated they �measured directly HIV-1 free survival at age 14-16 weeks in 496 of the 616 assessable babies�. But in their Figure 1 there are a total of 609 infants �assessable for HIV-1 infection�. Thus there are 7 missing infants but it is not possible to tell from which group or groups.

In Table 3, which lists the results of Cox�s regression models, 618 infants were assessed, which is two more than the 616 the authors assessed for HIV free survival at 14-16 weeks. Or, if we count 609 assessable infants, there were 9 missing infants.

Slide 50

During the study 12 sets of twins and 1 set of triplets were born. These babies were tested for HIV but their status was not reported. Second and third born infants, 14 in all, were also excluded from the analysis.

Why was their individual HIV status not reported and why were they ? Are these infants not considered important in such a study, especially since 9 of these infants were in the nevirapine group. If all nine were either concordantly negative or positive this would produce considerable bearing on the outcome of the study.

Slide 51

The HIVNET 012 study was not double or even single blind.

Slide 52

The authors admitted that after randomisation everyone knew to which group mothers belonged and thus which drugs would be administered.

It is impossible to accept this could not influence the outcome of the study. Given the fear of infecting newborn children, and the hope and possible hype that surrounds any new treatment, mothers or others could have inadvertently adopted behaviours that amongst other things, altered the timing and dosing of the actual drugs themselves. One only has to recall the first trial of AZT in the USA where gay men were able to distinguish AZT from placebo which led to drug sharing between the two groups.

Slide 53

Despite the fact that the HIVNET 012 was designed to be a randomised, placebo-controlled, double-blind, phase three trial of 1500 mother-infant pairs, the placebo group was dropped after results of another trial in Thailand were announced in February 1998.

Slide 54

By this time HIVNET 012 had enrolled only 49 women of which 19 women were assigned placebo.

This is a critical omission for any scientist wishing to prove a drug effect. Since no treatment or placebo may be associated with the benefit being sought, there is no possible means by which a scientist can claim a benefit over no or inactive treatment for the drug under investigation.

In the case of mother to child transmission of HIV this is not a trivial problem. Transmission rates vary widely between countries. For example, 15-20% in Europe; 16-30% in USA; 25-40% in Africa and 13-48% Asia and SE Asia. The authors themselves cite estimated transmission rates between 21-43%.

Slide 55

Experts themselves admit that methodological differences account for the large variations between studies.

One can only ask why has the large, double blind, properly randomised, placebo controlled drug trial become such a rarity during the AIDS era? Why design a trial containing all these elements and then abandon most? Are HIV positive mothers and their babies not deserving of the expected scientific rigour?

Slide 56

This slide documents three examples taken from studies with placebos to illustrate the unexpected, unpredictable and inexplicable variations in transmission which would otherwise remain hidden and lead to erroneous conclusions in the calculation of transmission rates. The examples are from two CDC studies conducted in Thailand.

The first is differences even between placebo transmission rates at two different hospitals within the same study. 14.3% versus 23.7%.

The second demonstrates how transmission rates may vary across the time span of a study, in this case according to the study midpoint.

The third that placebo can improve transmission in relation to no treatment. 18.6% versus 24.2%.

In relation to these data the authors commented that: �The lower than expected background transmission rate highlights the importance of having included a randomised, concurrently enrolled, untreated control group. Had the test regimen been inactive, a transmission rate of 18.6% may have suggested some efficacy when compared with historical data�.

Slide 57

Besides lacking a placebo group, the HIVNET 012 study did not study a non-treatment group. But in a prospective study reported in 1998, the transmission rate in 561 African women given no antiretroviral treatment was 12%, 1.1% less than the 13.1% reported for nevirapine in the HIVNET 012 study.

Slide 58

An additional problem is to do with the manner of reporting transmission rates.

Slide 59

The authors reported that they obtained �Blood samples were collected at 24 h, 6 weeks, and 14 weeks after birth for all babies�. The samples were frozen within 24 hours of collection and tested typically within a week.

This means that the authors had data as to the actual numbers of infants infected at these times.

But the results were presented as cumulative infection rates calculated from the Kaplan Meir method at times other than when the authors said they obtained the blood samples. In fact at day 3, and 8 weeks 16 weeks. These estimates were used to calculate the efficacy of nevirapine.

Nowhere in the paper are the infection rates reported at the times the infant blood samples were collected.

Why did the authors need to estimate infection rates? Why did they not report the actual data free from statistical manipulation?

Slide 60

Inevitably we must put the question �Is it possible for nevirapine to decrease the rate of mother to child transmission of HIV�.

Slide 61

All HIV experts agree that the level of maternal viral load predicts transmission to the infant.

The HIVNET 012 authors themselves reported a two fold increase in risk of transmission for every unit increment in log10 maternal viral load before entry into the study.

Slide 62

In fact in their discussion, the authors set out two necessary conditions which must be fulfilled in order to reduce mother to child transmission. QUOTE ��maternal viral load must be substantially decreased by the time of labour or the baby must have systemic concentrations of active drug present at the time of HIV-1 exposure to successfully lower risk of transmission� END OF QUOTE

Let us consider their first condition

Slide 63

In mothers the HIVNET 012 authors measured plasma HIV-1 viral load before entry, at delivery, and at 7 days and 6 weeks post partum.

But they reported only the baseline viral load.

Amongst the authors� pharmacological rationale for the choice of nevirapine was their claim that the drug reduces the plasma HIV-1 RNA concentration by at least 1.3 log after a single dose and cited a reference number 13.

Reference 13 is the authors� HIVNET 006 study, The phase I/II study of the safety and pharmacokinetics of nevirapine in HIV-infected pregnant Ugandan women and their infants, published in AIDS in the same year as the 012 study.

Slide 64

In this study the authors measured RNA in 19 women and found a 1.3 log reduction 7 days after a single dose.

Two of the mothers, both who transmitted HIV to their infants, had viral loads of 556 and 672 copies per ml at delivery.

Amongst the non-transmitting mothers an unspecified number had a viral load less than 400 copies per ml.

We should note that a plasma RNA less than 400 copies per ml is considered a viral load of zero.

At six weeks viral load was the same as baseline.

These data invite two comments:

First, if a viral load < 400 copies per ml is considered zero, then viral loads of 556 and 672 are very close to zero. Especially on a log scale. If we consider these two mothers, as well as the unspecified number of mothers who had viral loads which were zero, then amongst a cohort of only 19 women we may ask how many mothers had a substantial viral load to reduce?

Second, the reduction in viral load reported by these authors is not confirmed by other researchers.

Slide 65

For example, de Jong and 22 colleagues, using a higher dose of nevirapine, observed an average reduction of 0.46 ? 0.47 log RNA copy numbers after 4 weeks of treatment, which returned to baseline 8 weeks later.

Slide 66

Most importantly, and as one might anticipate from a median 1.3 log reduction measured 7 days after dosing, it is highly unlikely, within a few hours of administration, that is, at delivery, the HIV RNA viral load could have decreased at all.

Indeed, this is exactly what the authors observed.

QUOTE �Maternal plasma HIV-1 RNA levels were also not significantly different at delivery from baseline� END OF QUOTE

Slide 67

We conclude that these data contravene the author�s first condition. That maternal viral load must be substantially reduced by the time of labour.

Thus nevirapine cannot successfully lower the risk of transmission during labour and delivery by lowering the concentration of virus presented to the infant.

Where does this leave us in relation to the authors� second condition? That the infant must have a systemic concentration of active drug at the time of HIV exposure?

Slide 68

There are two sources of exposure to consider. The immediate source is the mother�s blood and birth canal. And following this, the mother�s colostrum and breastmilk.

There are two mechanisms by which nevirapine could act. The first is prophylactically in the infant following exposure to both sources mentioned, or by reducing the viral load in breastmilk and colostrum.

These lead to the question, what is a therapeutically effective plasma concentration of nevirapine?

Slide 69

The HIVNET authors selected a target plasma concentration of nevirapine of 100 ng/ml which, according to the authors, is ten times the 50% inhibitory concentration.

The authors did not determine the IC50 themselves.

The data cited in support of the authors IC50 figure of 10 ng/ml were published by 13 authors from Boehringer Ingelheim Pharmaceuticals and 2 from the Department of Pediatics and Medicine at the University of Massachusetts Medical School in December 1990.

Other authors, for example, Grob and his colleagues, also from Boehringer Ingelheim, report an IC50 twice this concentration.

These and other studies reporting these data are obtained from in vitro experiments and measure inhibition of reverse transcription using not the HIV RNA but synthetic template-primers.

Slide 70

This slide shows infant nevirapine pharmacokinetics from the PACTG 250 and HIVNET 006 studies.

PACTG 250 was and independent study conducted at seven hospitals in the USA and published by a year earlier than HIVNET 006.

In mothers both 250 and 006 used the same nevirapine regime as the HIVNET 012 study. In the infants the drug was administered at �between 48 and 72 h after birth� in 250 and �at 72 hours� in 006. In 012 nevirapine was given �at 72 h after birth or at discharge from hospital, whichever occurred first�. Also in 012 were an unspecified number of children who were born at home or an outside and who were given the drug �as soon as they arrived at the clinic if they presented within the first 7 days of life�.

One should note the pre-nevirapine dose differences, that is, the concentration which resulted from the mother�s nevirapine, the time of dosings, the range of C maximums, the 12.6 hour difference in the median T maximums, the range of T maximums and the twofold differences in half lives of the administered drug.

At one week there are differences between the median and maximum concentrations of over 70%.

In other words, similar to the viral load data, the HIVNET pharmacokinetic data are substantially different from those reported in other studies.

However, they do demonstrate concentrations of nevirapine exceeding the HIVNET 012 authors� target of 100 ng/ml.

The only question of any practical significance in these data is not whether the drug levels achieve a concentration arbitrarily set to exceed an in vitro IC 50 level, but do they result in systemic concentrations of active drug present at the time of HIV-1 exposure sufficient to prevent the replication of HIV and thus lower the risk of transmission in infants?

Slide 71

Not according to data published by Havlir and colleagues in the Journal of Infectious Diseases in 1995.

Here it was demonstrated that plasma trough levels of nevirapine required for an in vivo virological response range from 3.4 to 8 ug/ml with a median value of 4.7 ug/ml.

Expressed as nanograms per ml it can be seen that the C maximum in the infants does not reach the minimum concentration required for a virological response.

Thus on this basis we conclude that the authors second condition, that is, achieving therapeutically active concentrations of nevirapine in the infant, is not fulfilled.

Slide 72

This brings us to breast feeding. For the sake or argument, let us give credence to the authors� own suggestion, expressed in the HIVNET 006 study, that QUOTE ��if nevirapine turns out to be efficacious in preventing vertical transmission at the time of delivery, it is unlikely to be caused by a reduction in maternal viral load. The decrease in viral load during colostrum feeding might, however, impact on postnatal transmission�. END OF QUOTE

Indeed, this is how Hudson and Moodley, from the University of Natal, interpreted the HIVNET 006 study data.

��findings from HIVNET 006 suggest maternal dose may primarily act by reducing early breastmilk transmission�*

Could nevirapine reduce transmission by lowering maternal viral load during the time infants are breast fed?

Slide 73

Even if the authors� target of 100 ng/ml is an effective in vivo concentration necessary reduce maternal breast milk viral load, we can see that because the half life is 72 hours at the most, the target concentration will be sustained for only a few weeks at most. Well short of the time mothers at least in this study, breastfed their babies. So any benefit from lowering breastmilk viral load will be transient. In fact it may be even worse, as James McIntrye expressed in the British Medical Journal on January 26th, 2002. That even after single dose, nevirapine may cause the the development of resistance to the drug. If, as Brooks Jackson and associates suggest, rebound after stopping antiretroviral treatment leads to higher viral loads, the HIVNET 012 regime may increase transmission via breastfeeding.

Regardless of what exposure the infant receives via breastfeeding, the previous argument applies. The concentration provided via the mother and the single dose given to the infant does not result in levels effective in vivo.

Slide 74

Again, for the sake of argument, let us assume that nevirapine as administered in the HIVNET 012 study is 100% effective in preventing HIV infection via breastfeeding and that this effect lasts not for a few weeks but 11 months. We choose 11 months because the HIVNET authors cite a paper by Miotti and colleagues from Malawi which reported a 7% cumulative transmission risk due to breastfeeding at 11 months of age.

In HIVNET 012 the authors claimed nevirapine lowered the risk of HIV-1 infection by nearly 50% in a breastfeeding population. Therefore, let us assume that without treatment, or with a placebo, the transmission rate would have been twice their reported rate of 13.1%, that is, 26.2% This figure is in line with what the authors quoted in the introduction to their paper.

If the placebo rate is 26.2%, and we subtract the 7% breastfeeding risk, we are left with 19.2%. THIS REPRESENTS THE MINIMUM TRANSMISSION RATE WITH NEVIRAPINE

If we now compare the efficacy of nevirapine with AZT we arrive at a figure of 24%, approximately half that reported by the authors.


Slide 75

If the placebo transmission rate is 26.2% and the AZT transmission rate 25.1% then the AZT transmission rate is not significantly different from the placebo rate.

Yet the authors state their short course AZT may have had some benefit.

Slide 76

Let us recount the conditions the authors stated were necessary for nevirapine to prevent mother to child transmisson.

�Maternal viral load must be substantially decreased by the time of labour or the baby must have systemic concentrations of active drug present at the time of HIV-1 exposure to successfully lower risk of transmission�

But the authors demonstrated that nevirapine does not lower maternal viral load during labour and delivery.

We have demonstrated that the concentrations achieved in vivo cannot lead to a virological response.

And it goes without saying that nevirapine, as administered and recommended in this trial cannot effect transmission before the onset of labour, that is, during pregnancy.

Thus nevirapine does not pass the authors� own set of criteria for the purpose proposed.

Slide 77

In conclusion,

There are many questions regarding the design, execution, analysis and interpretation of the HIVNET 012 study. And as researchers from South Africa point out, just one factor, the wide confidence intervals reported for the efficacy of nevirapine over AZT, is sufficient to argue that the HIVNET data should be confirmed before this treatment becomes standard clinical practice.

��CIs for their estimate of efficacy are wide, with a lower value of 20%. Further studies are needed, and are in progress, to confirm their findings�.

According to one of the best known European experts on MTCT, Marie Louise Newell, QUOTE �A randomised, double blind placebo controlled trial of Nevirapine versus placebo in addition to routine anti-retroviral prophylaxis (ACTG316) is currently underway in the USA and Europe. The regimen under evaluation is as the HIVNET [012] trial�but is being studied in a very different population� Furthermore, women participating in ACTG316 are asked not to breastfeed their infants, It is expected that enrolment in this trial will be completed by mid 2000�. END OF QUOTE

The HIVNET 012 trial was published in 1999, five months after the completion of enrollment. On this time scale, dating from mid 2000, the 012 study could have been published four times over. We may well ask, why has the ACTG316 study and others, not been published?

Thus even if nevirapine were totally devoid of toxicities, which is by no means the case, at present there is no basis whatsoever for recommending its use in preventing mother to child transmission of HIV.

Slide 78

We conclude this presentation with a very brief look at deaths and adverse events in the two HIVNET trials as well as toxicity data.

Slide 79

In the HIVNET 006 study 4/22 infants died. The causes of death were not reported in three of these children.

There were also 12 serious adverse events of which one was thought possibly but not likely to be drug related.

Slide 80

In HIVNET 012 adverse events in infants were uniformly recorded up to age 6 weeks, but after than only serious adverse events continued to be recorded at each visit up to age 18 months.

38 babies died. 22 in the AZT group and 16 in the nevirapine group. Pneumonia, gastroenteritis, diarrhoea, dehydration and sepsis were given as the causes of death. There were 59 serious adverse events in the first 8 weeks of life. Those specified were sepsis, pneumonia, fever, congenital anomaly, asphyxia and dyspnoea.

Four adverse events in the AZT group and 2 in the nevirapine group were thought �possibly, but unlikely to be, related to the study drug�.

Overall there were 18 babies with maculopapular rash, 22 had anaemia.

In light of these data there are two claims in the HIVNET 012 study worthy of comment.

The first is that adverse events were �similar up to the 18-month visit�. Since the 012 study did not have a placebo all that can be claimed is that nevirapine and AZT have equal probabilities of adverse events. This may be because AZT and nevirapine are both toxic and equally so. These data do not prove that nevirapine or AZT is non-toxic.

The second point relates to the authors� claim that �the nevirapine regimen lowered the risk of HIV-1 infection or death up to age 14-16 weeks by 48%�. In the context of HIV infection this could only be valid if the causes of death are all given and they are all caused by HIV infection. If the causes of death are not all caused by HIV infection then the authors are claiming that nevirapine reduces deaths from independent causes, such as dehydration.

Slide 81

As far as drug toxicities are concerned, the recently published guidelines for the use of antiretroviral agents in children mention several. Some serious and some fatal.

Given these toxicities and the unproven benefit of nevirapine it would seem most unwise to recommend its use in children.

Although generations of medical students are taught the child is not a little man, there is still a sufficient nexus between the two to heed toxicities that occur in adults.

Slide 82

The CDC warn that the toxicities of nevirapine are such that it should not be given prophylatically to healthy adults. Its use for this purpose, for example, after needle stick injuries, has resulted in severe and life threatening effects with at least one patient requiring a liver transplant for fulminant hepatitis. Many pregnant women are healthy adults.

The European agency for the evaluation of medicinal products warns of the same toxicities and recommends the use of nevirapine only for advanced or progressive immunodeficiency.

Given these and all the preceding data, it is problematic to speak of nevirapine in terms of a benefit to risk ratio.

We conclude it is absolutely vital that further evidence be presented before nevirapine can be considered a proven modality for the prevention of mother to child transmission of HIV.

Thank you for your attention.