HIV-1 in 1908? Another Sad Comedy of Errors from Michael Worobey

Printer-friendly version

October 9th, 2008.

As forecast in my piece "Worobey's wobbly research", first posted on this site on March 19th, 2008, the Canadian molecular biologist Michael Worobey has just published new calculations about the age of the AIDS virus, HIV-1, which place its origins even further back in time.

His work appears in the form of a lengthy letter to the journal Nature, entitled "Direct evidence of extensive diversity of HIV-1 in Kinshasa by 1960", by M. Worobey, D.E. Teuwen, M. Bunce, S.M. Wolinsky et al.; [Nature; 2008 (October 2nd); 455; 661-664.]

On the basis of this one newly-discovered sample of HIV-1 dating from 1960, Worobey and his colleagues contend that the first human infection with the AIDS virus occurred in 1908, with outer confidence limits stretching from 1884 to 1924.

The previous "best guess" of molecular biologists such as Bette Korber was that the first HIV-1 virus existed in a human being by 1931. This in itself was a highly dubious finding. But the 1908 "guestimate" by Worobey and his team from the Department of Ecology and Evolutionary Biology at the University of Arizona is now being highlighted by extensive coverage and publicity in Nature, and then spoon-fed to a largely compliant press corps.

Worobey's calculations, according to his supporters Beatrice Hahn and Paul Sharp, who have been invited by Nature to write the accompanying commentary, employ "state-of-the-art statistical analyses". But they are actually based on just the one crucial new piece of data, this being a fragment of genetic sequence allegedly obtained from the stored lymph node of an African woman in 1960. Just like the the famous 1959 sample (the oldest known sample of HIV), this 1960 sample comes from a subject who was then living in the Belgian Congo capital of Leopoldville, (now Kinshasa, Democratic Republic of Congo). This, quite clearly, is a significant detail, and yet not one of the press articles covering Worobey's paper seems to have picked it up. I, and the small group of people on whose wisdom and scientific expertise I informally rely (three quarters of whom are professional scientists, some of them quite eminent scientists) are strongly persuaded that these exciting-sounding dates from Worobey are highly dubious. We believe that his analysis and interpretation of the 1960 viral sequence are in reality little more than wishful thinking based on poorly-supported science.

We also believe that the discovery of that 1960 sample of HIV-1, and the coincidence of place with the 1959 HIV-1 sample, is the real story here, not Worobey's highly speculative 1908 guestimate of when the first AIDS virus might have existed. A much simpler and better-supported explanation for the recently- discovered 1960 HIV-1 fragment is that both it and the 1959 HIV-1 fragment are the results of the administering in Leopoldville and elsewhere in the Belgian Congo from 1957 onwards of different batches of an experimental live vaccine. This vaccine, an oral polio vaccine (OPV) called CHAT, was (as I have previously demonstrated) prepared locally in the Congo in chimpanzee cells, which cells were themselves almost certainly contaminated with SIVcpz (chimpanzee SIV, the immediate primate ancestor of HIV-1).

Part 1 of this essay gives a brief background to Worobey's latest paper. In Part 2 I provide some analysis of the paper, and attempt to demonstrate where Dr Worobey has gone wrong. In Part 3 I shall provide a much simpler explanation for the existence of this new HIV-1 sequence from 1960. In Part 4 I shall provide some historical background to Dr Worobey's involvement in this debate. And in Part 5 I shall provide more information about the large organised cover-up that has taken, and is taking place on this issue.

(1) Background.

In the 23 years since 1985 (just over a year after scientists had first developed the ability to test for the presence of HIV-1) there has been just one single truly ancient sample of HIV-1 known to scientists. This ancient sample of virus came from a blood specimen initially taken in 1959 from an African male from Leopoldville, in what was then the Belgian Congo. Scientists have searched hard for nearly a quarter of a century, but they have still found no earlier sample of HIV-1 in the entire world. [Originally, in 1985, this sample was described as HIV-positive because it tested positive in a series of antibody studies. Later, in 1998, the viral sequence (ZR59) of HIV-1 fragments obtained from this 1959 sample was published.]

Now, in 2008, the team of Michael Worobey have announced the discovery of another ancient sample. It apparently comes from the stored lymph node of an adult African female, and was taken in 1960. And it was obtained from the very same city, Leopoldville - now called Kinshasa in the Democratic Republic of Congo (DRC).

For perspective, the next earliest known sample of HIV-1 dates from 1976, fully sixteen years later - and it also was taken from a subject in the DRC (one who had worked for some years in Kinshasa as a femme libre, a "free woman"who had had several different sexual partners. The often-used translation of "prostitute" is a little too simplistic.)

Members of Michael Worobey's team have obtained a genetic sequence from the tiny fragments of virus that they have managed to recover from this 1960 sample of HIV-1. (A genetic sequence is a list, in sequence, of the individual nucleotide bases - A, C, G and T - that make up an organism, in this case the HIV-1 virus.) There are questions that could legitimately asked about this sequence, but it is too early to go into details, given that the sequence details have only recently been released to Genbank. I think that for now it is better to leave such concerns aside, and to assume that the 1960 sequence is a genuine sequence.

(2) An analysis of Worobey's paper, including some observations of where Worobey has gone wrong.

Worobey's team has gone on to use this genetic sequence from 1960 in an attempt to bolster their ideas about the age of AIDS. Employing their molecular clock hypothesis, they compare their segment of the 1960 sequence with the fraction of genetic sequence that exists from the 1959 HIV-1 virus, and they compare these in turn with other more modern HIV-1 sequences from around the world. Then, assuming that the evolution of HIV takes place at a constant rate, they extrapolate backwards to predict the age of the AIDS pandemic.

They claim that the two HIV-1 viral sequences from 1959 and 1960 are so genetically different one from the other (the difference is actually just under 12%) that this forces back the date of the first HIV-1 virus even earlier in time. In year 2000, the team of Bette Korber at Los Alamos, New Mexico (which is closely allied to the Worobey group in Arizona) proposed that the first HIV-1 had existed back in 1931. But now Worobey and his team propose that this index HIV-1 infection existed in 1908.

The aforesaid molecular biologists and geneticists deal with huge tranches of molecular data, and conduct their work on super-computers. They presume that they can measure the age of HIV-1 through a molecular clock (or phylogenetic clock) that ticks regularly, like a metronome.

However (as several scientists and I myself have been pointing out for more than six years now), the basic assumption that underlies all of their work, that the evolution of HIV-1 occurs at a constant rate, through mutation, is erroneous.

The molecular clock does offer a fairly reliable method for deciphering the past history of most viruses, which are DNA-based, and which do evolve predominantly through mutation. However, HIV is a retrovirus, and is RNA-based - and nine-tenths of HIV's evolution occurs through recombination, a completely different form of viral evolution. Unfortunately, the molecular clock is unable to measure recombination. HIV-1 is in fact the most recombinogenic virus (the virus most prone to recombination) known to medical science. And this means that the molecular clock, which measures only mutation, is an inappropriate hypothetical model for measuring the evolution of HIV-1.

Yet Worobey and his colleagues (just like other proponents of the bushmeat hypothesis of HIV origin, such as Paul Sharp, Bette Korber and Beatrice Hahn) ignore this simple fact.

They acknowledge that recombination in HIV-1 presents a problem for the dating of HIV-1, but argue that they have taken steps to remedy this. For instance, they say that they have excluded recombinant sequences from the dataset of HIV-1 viruses which they use for their analysis. But in reality they are unable to do this - the main reason being that you cannot identify recombination if it occurs early in the evolutionary history of a virus. Because they cannot recognise all recombinant sequences, they are unable to remove all recombinant sequences from their dataset.

It is interesting to note that what Dr Worobey finds most useful for his analysis (just like doctor Sharp before him) is something he refers to, deliciously, as a "relaxed molecular clock". This means a clock that ticks at a regular rate when they want it to, but which begins to bend time when they need some extra elbow-room with their calculations! Thus it is that Worobey and his team arrive at that HIV-1 start date of 1908. It sounds impressive. But in reality, it is simply a casserole of ambition and wishful thinking. When I used to speak with him back in 2000 to 2002, Michael Worobey used repeatedly to say that because of the impact of recombination, any attempt to date HIV-1 through phylogenetic means ended up as "a dog's dinner".

Now he has changed. Nowadays he states that the effect of recombination on dating HIV-1 can be ignored. But the work that he cites to support this contention relies on an assumption that recombination only occurs on the "terminal branches" of the phylogenetic tree, which represent the most recent HIV-1 sequences.

But Worobey offers nothing to support this assumption, which (as I demonstrate below) is inherently unsafe. What Professor Worobey has done, in effect, is to brush the whole issue of recombination under the carpet.

Let me explain in rather more detail why I dispute the findings of Michael Worobey, and of these eminent geneticists and molecular biologists who support him.

Essentially, Worobey and his friends are wedded to two preconceptions. These are:

(a) that one can date the age of HIV-1 (a virus which evolves mainly through recombination) by means of a molecular clock (a theoretical device which cannot allow for recombination, and which only measures evolution through mutation); and

(b) that there was just one single crucial transfer of immunodeficiency virus from the ancestral animal host (now accepted as the common chimpanzee) to a human being, and that this one event started the AIDS pandemic.

I strongly believe that both preconceptions are erroneous. Let me address them in order.

(a) Although the impact of recombination on phylogenetic dating is hugely controversial, Worobey spends just two sentences addressing it in his latest paper. The key passage of these two sentences reads: "Despite initial indications that recombination might seriously confound phylogenetic dating estimates, subsequent work has suggested that recombination is not likely to systematically bias HIV-1 dates in one direction or the other."

To support this claim, he cites another paper from 2004, on which he is a co-author. It is by Philippe Lemey et al., and it is entitled: "The Molecular Population Genetics of HIV-1 Group O" [Genetics; 2004; 167; 1059-1068, available on-line through PubMed]. In this paper, Lemey and colleagues do briefly address the key published paper which proposes that recombination in HIV-1 does not allow effective phylogenetic dating estimates of that virus to be made. This latter paper is by Mikkel Schierup and Roald Forsberg, and is called "Recombination and Phylogenetic Analysis of HIV-1"; it was presented in 2001 at the Lincei conference on "Origin of HIV", discussed elsewhere in this essay. The paper was published in 2003, and appears on pages 231-245 of the conference proceedings.

Paul Sharp, who also spoke at the Lincei conference [see below] was so angry about Schierup's paper that he declined to allow his own paper to be published in the Lincei conference proceedings. This was probably a wise move, for Schierup's paper fairly blows apart the arguments of the HIV-1 molecular dating school. It concludes with the telling words that if recombination occurs early in the history of a virus like HIV-1, then "it is not valid to use a phylogenetic method to obtain the time estimate" of when HIV first appeared.

In his 2004 paper, Lemey acknowledges (page 1061) that "the results probably indicate significant levels of recombination" in their dataset. Having noted that Schierup and Forsberg's 2003 paper arrives at different conclusions from their own, he then proceeds, however, to argue that Schierup's arguments can be ignored. The basis for doing so actually boils down to the following claim: "recombination events in a very rapidly growing population will mostly occur on the terminal branches of the 'star-like' sample geneaology." [Here Lemey is asserting that most of the recombination in HIV-1 occurs in the more recent sequences ("the terminal branches") on the phylogenetic tree of HIV-1 viruses. This may appear a reasonable claim, but further examination reveals its shortcomings; see below.] Lemey goes on to state that although recombination on the terminal branches will result in "rejection of the molecular clock and an increase in the variance of TMRCA estimates, importantly, it will not systematically bias estimates of the TMRCA in either direction." [TMRCA means "Time to the Most Recent Common Ancestor" - ie the time from the present (2008) back to that start date of 1908, or whatever.]

Later, Lemey effectively repeats this. He once again grants the possibility that recombination might "bias the TMRCA upward" [ie cause scientists to over-estimate the time back to the HIV-1 start date] and also that he and his team might be employing "a circular argument". However, he promptly rejects the latter possibility, again by assertion rather than supported argument.

Lemey and colleagues conclude by claiming that their analysis "provides some assurance that recombination is not strongly biasing the estimates of TMRCA", and that "the methods we have used here present a framework that goes some way toward a more realistic description of viral evolution". But these claims are far from convincing. In fact, the authors do not even appear to be convinced themselves!

I find it astonishing that in his latest (2008) paper, and purely on the basis of Lemey's 2004 paper on which he was a co-author, Worobey now baldly asserts that "recombination is not likely to systematically bias HIV-1 dates in one direction or the other". All that he has done is driven us a few times round the houses, and then asserted that since recom only occurs in recent sequences, it stands to reason that it cannot affect estimates of the age of HIV-1. If this is not a circular argument, I really don't know what is. In reality, the Lemey paper entirely fails to address the sort of SIV recombination which I believe took place in the chimpanzee tissue cultures that we now know were being made in the Laboratoire Medical de Stanleyville (LMS).

According to the report of one of the vaccine-makers, Stanley Plotkin, different pools of CHAT vaccine were prepared in the 1950s by adding new tissue culture material in sequence. (He writes: "No seed system was used. Rather, each pool served as the seed virus for a subsequent pool.") [S.A. Plotkin, "Unthuths and Consequences"; Phil. Trans. Roy. Soc. Lond. B; 2001; 356; 815-823; see page 816.] This means, amazingly, that any viral contaminants in one vaccine pool would have been passed on and added to the next vaccine pool - and to the next one after that. It is not a technique that would be acceptable today. It is highly likely that if the vaccine-makers adopted such an approach during the important task of preparing brand new pools of vaccine, then the same technique (of adding materials sequentially) would have been employed for the far less crucial task of preparing new vaccine batches.

[Definitions. A pool of vaccine is the term for vaccine that has been prepared to a certain level of attenuation; however, different batches of that pool can be made in different labs, at different times, and in different tissue cultures. By contrast, a batch of vaccine is produced in a single production run. Unlike a vaccine pool, a vaccine batch is homogeneous.]

We have evidence from a military paper from 1958 that refers to the making of chimpanzee tissue cultures (in the US) from the cells of the Lindi chimps, and this paper shows that both kidney cells and sera from the chimpanzees were part of the process. In practice, both kidney cells and primitive preparations of sera will contain SIVs, provided the source animal (in this case the chimp) is SIV-infected. Surveys reveal that 13% of wild central African chimps are naturally SIV-infected, even before any co-caging and gang-caging, which is what routinely happened at Lindi. This means that SIV-infected kidney cells and sera would almost certainly have been added sequentially to the tissue cultures as each new batch of polio vaccine was made at the LMS. It only takes two SIVs in a cell, or in a tissue culture, to spark recombination - and in this case we have the picture of new SIVs constantly being added to the "pot", producing a melange of new SIV sequences and recombinant SIV sequences that would have become more and more complex with time. Given such a background, a 12% genetic difference between two different vaccine batches (one of which infected the 1959 Leopoldville male and the other the 1960 Leopoldville female) would be well within the bounds of possibility.

Crucially, the chimpanzee SIVs so produced (or HIV-1 viruses, as they are known once they appear in humans) would not appear on the more recent "terminal branches" of Worobey's phylogenetic tree, but would appear right in the centre, in the main trunk. In this core position on the phylogenetic tree, such ancient recombination is, quite simply, undetectable from the perspective of today.

And, as Schierup and Forsberg state in the conclusion of their 2003 paper: "if recombination has occurred in the viral population originating from the MRCA, it is not valid to use a phylogenetic method to obtain the time estimate, and our results suggest that doing so would give a certain overconfidence to the previous estimate of 1931 +/- 10 years." Their phrase "would give a certain overconfidence" is believed to be polite under-statement, or else a slight Scandinavian misunderstanding of English usage. What is really meant, it would seem, is: "would definitely indicate overconfidence"!

(b) Another shortcoming of phylogenetic analysis is that it can only embrace the idea of a single index virus sparking an epidemic. But the concept we have with the OPV theory is entirely different. It is of several different SIV-contaminated batches of CHAT vaccine being administered, within a brief period of time, to "volunteers" in some 30 different vaccination trials across central Africa. In other words, it involves several separate (and near-simultaneous) introductions into humans of variants and recombinants of chimp SIV. From an OPV perspective, this is exactly what appears to have happened around 1959-60, right at the start of the HIV-1 epidemic. Yet the molecular clock supporters are unable to compute two different forms of a virus that crossed over from chimps to humans in the same town at almost the same time. Because of their preconceptions, they are forced to invent an imaginary index virus from decades earlier which (they say) gave birth to both the 1959 and 1960 viruses. The 1959 HIV-1 virus from Leopoldville is almost 12% different, Worobey explains, from the 1960 virus from Leopoldville. He concludes that this is such a big genetic difference that this "indicates that the HIV-1 M group founder virus began to diversify in the human population......decades before 1960." However, this interpretation is unsupported.

The following explanation is largely based on an e-mailed commentary from one of my "advisors". I reproduce it here with only slight alterations, because it summarises the arguments especially simply and clearly.

"The key fact that Worobey presents is that in 1959 and 1960 there were in existence in Leopoldville two strains of HIV-1 which had notable differences in their genetic structure. Their interpretation is that the two viruses must have evolved apart from a common ancestor over a considerable number of years. An alternative view is that the origins of HIV-1 lie in chimpanzee cultures that contained a variety of SIVs with considerable genetic heterogeneity. Thus, at the moment of formation, HIV-1 would have possessed considerable genetic diversity inherited from these chimp SIV progenitors. There would have been several separate SIV transfers from chimps to humans via the vaccinations, and what we see in the 1959 and 1960 Leopoldville samples is merely some of this genetic variability of chimp SIV."

(3) A much simpler explanation for the 1959 and 1960 HIV samples.

Using their molecular clock, Worobey's team use the 1959 and 1960 HIV-1 samples (and compare them with the 1976 sample, and dozens of other samples obtained from nearer the present, and from different countries around the world) in order to predict that HIV-1 has been in existence since 1908.

If they were not so wedded to these two faulty principles, they might examine a much simpler and more logical explanation. This is that the emergence of the two earliest examples of HIV-1 from the same city (Leopoldville/Kinshasa) and within a year of each other suggests that there might have been a causative event in that city in the years immediately preceding 1959 and 1960.

Was there such a candidate event? There was indeed.

Starting in August 1958, a mass vaccination was staged of all the young children (up to the age of five years) in Leopoldville with CHAT, an experimental oral polio vaccine (OPV). CHAT had been developed by a Polish-American scientist, Hilary Koprowski, who was then director of the Wistar Institute, an independent biomedical research institut