Transmission of HIV cannot be proved

by Henry Bauer 

“Genetic analysis has shown . . .” makes the claim that something has been proven scientifically, which is commonly taken to mean beyond any doubt.
I’ve long wondered how such claims could be made with respect to HIV, since HIV is also said to mutate at such an astounding rate that
“no two virus isolates are identical….  Within a single … host, HIV-1 population represents a complex mixture, or swarm, of mutant virus variants … [whose] prevalence … is changing … on almost a daily basis (intrahost evolution)”

— Lukashov, Goudsmit, & Paxton, “The genetic diversity of HIV-1 and its implications for vaccine development”, ch. 3, pp. 93-120 in AIDS Vaccine Research, ed. Wong-Staal & Gallo, Marcel Dekker  2002.
Furthermore, one should not imagine that “genetic analysis” means that genomes are being compared, that whole DNA or RNA chains are being compared. Rather, the researchers decide to focus on some short segment of the presumed HIV genome and compare that short segment in the various samples.

Reliance on HIV genetic analysis has become increasingly common in criminal cases when an individual is charged with transmitting HIV. Earlier this year, Lancet Infectious Diseases (Abecasis et al., “Science in court: the myth of HIV fingerprinting”, 11: 78-9) published a clear exposition of the reasons why HIV genetic analysis cannot provide proof of transmission:
“the risk  of incorrect conviction is substantial because of a flawed  view  of  the  science  behind  forensic  phylogenetics.  A  scientific discussion of  HIV  fingerprinting  in  cases of  HIV  transmission is therefore urgent. . . . By  calling  such  investigations  HIV  fingerprinting,  scientists  raise  unrealistic  expectations . . . . Unlike   for   DNA   fingerprinting,  where  a  likelihood  can  be  calculated for  a  full  match between the evidential DNA and the suspect DNA, there  is never a full match between the RNA or the DNA of HIV  in two samples, even within an individual. HIV is constantly evolving *”
* Shankarappa et al., “Consistent viral evolutionary  changes associated with the progression of human immunodeficiency  virus type 1 infection”, Journal of Virology 73 (1999) 10489-502

This means that one can calculate only probabilities, in the first instance the probability that two samples come from the same hypothesized chain of transmission. Even when that probability is high, however, the data cannot distinguish between direct and indirect transmission in that chain: in other words, the supposed victim of transmission (V) might have been infected by someone other than the suspect (S) because S might have infected A, B, C . . ., any one of whom might have infected V.
Beyond that, a high probability that two samples come from the same chain of transmission says nothing about the direction of transmission: it might even have been V who infected S rather than the opposite.
Above all, it should be recognized that all evidence from genetic analysis is no more than circumstantial. It should not be confused with proof.
The point made by Abecasis et al. was seconded by Leitner et al. (Nature 473 [2011] 284; the correction published 6/7 July does not pertain to the following point):
“Scientists must explain to  courts that phylogenetic analysis  cannot ‘prove’ any particular  hypothesis . . . . results may be compatible with  several hypotheses, or support  one over another. . . . Although the direction  of viral transmission can  sometimes be supported, it does  not prove direct transmission.”
The AIDS Beacon website (“Independent, up-to-date news and information about HIV and AIDS”) interviewed authors of theLancet article who emphasized, “Phylogenetics can prove that people cannot have infected each other, but it can never prove that people infected each other.”
Guidelines are being drafted for how researchers should give testimony about HIV phylogenetic analysis, in view of the universal agreement that it should never be described as “fingerprinting” and that courts should understand that

phylogenetics alone can never prove transmission:
 it can never prove any alleged transmission
occurred between two specific individuals.

I’ve not yet seen a discussion of how this point applies to research that attempts to trace where HIV first entered human beings; or more important, how this point applies to research on measuring supposed rates of transmission, and on studies of attempts to prevent transmission.
As it happens, Cohen et al. recently published “Prevention of HIV-1 infection with early  antiretroviral therapy” (New England Journal of Medicine, 10.1056/NEJMoa1105243, 18 July 2011) in which they claim that “The early initiation of antiretroviral therapy reduced rates of sexual transmission of  HIV-1”.
The studied population comprised 1763 HIV-1 discordant couples in 5 African countries and in Brazil, India, Thailand, and the USA. Early antiretroviral treatment appeared to result in transmission at a rate of 0.3 per 100 person-years, whereas the untreated population showed a rate of 1.2 per 100 person-years. The total number of actual apparent transmissions was 39:
“Through viral genetic analysis, 28 transmissions  were linked to the HIV-1–infected participant”.
But genetic analysis cannot prove such linkage!
Note too that 11 of the 39 apparent transmissions were not even claimed to be direct transmission. How did they occur?
Perhaps the same sort of process that produced these unlinked seroconversions might also have caused the 28 apparently linked transmissions as well?
Furthermore, 23 of the 28 purportedly linked transmissions were at African sites. It is mainstream dogma that the prevalence of HIV in Africa bespeaks a very high level of multiple concurrent sexual relationships. Since genetic analysis cannot prove any given case of transmission, it seems unwarranted to assume that those 23 apparent transmissions in particular can be taken as proven direct transmission within the discordant couple.
There are a number of other reasons as well for questioning the validity of conclusions from this study, for example the wide variety of antiretroviral drugs used: “Study drugs included a combination of lamivudine  and zidovudine (Combivir), efavirenz, atazanavir,  nevirapine, tenofovir, lamivudine, zidovudine, didanosine, stavudine, a combination of lopinavir  and ritonavir (Kaletra and Aluvia), ritonavir, and a  combination of emtricitabine and tenofovir (Truvada). A prespecified combination of these drugs  was provided to participants at monthly or quarterly visits. Sites could also use locally supplied,  FDA-approved drugs if they could be purchased  with nonstudy funds. For participants with virologic failure, specified second-line treatment regimens were provided”.
Even on the study’s own terms, surely one would need to consider the possibility that only one or some of these were effective, instead of lumping them all together as “early versus delayed therapy”?
Beyond that, it seems rather unlikely that precisely the same protocol and level of supervision could have been in effect in all the different study sites in the different countries, and with investigators that number at least the named 35 co-authors.

The chief point, however, is this. Experts in phylogenetic analysis are unanimous that this procedure cannot prove any given case of transmission. Cohen et al. rely on it precisely for such proof, and therefore their conclusions cannot be accepted as valid. Less politely, one might say that this peer-reviewed paper is BS.

*                    *                    *                    *                    *                    *

Most studies of claimed transmission have relied  only on “HIV” tests. Those are even less to be believed, of course, than the genetic analyses, given the high rate of seroconversion for reasons having nothing to do with “HIV infection”. Just as with genetic analysis, though, an absence of matching tests can demonstrate a lack of transmission, as in the notorious Padian study.