Koch's Postulates and the Proof of Causation

Epidemiologists and biomedical researchers rely on a range of principles to establish causation in disease. However, since the acceptance of different versions and interpretations of these principles of causation has itself become one of the stakes in the controversy over the causation of AIDS, these principles cannot be independently invoked as neutral measures. The most well known causation criteria are called "Koch's postulates," named after Robert Koch, the nineteenth-century German microbiologist. The postulates consist of four steps that are easily stated. First, the causal agent must be found in all cases of the disease. Second, the agent must be isolated from a carrier and grown in pure culture. Third, when the culture is injected into a susceptible laboratory animal, the animal must contract the disease. Finally, the causal agent must then be recovered from the diseased animal.

The precise relevance of Koch's postulates to contemporary biomedical research (particularly with regard to viruses, which were unknown at the time of Koch's own work) is in dispute, and this dispute has been magnified as a result of recent debates about the causation of AIDS. Many have proposed that researchers nowadays must work with "modern" or revised versions of the postulates, and have argued that Koch himself did not intend them to be followed rigidly.^[122] Nevertheless, Koch's postulates remain a well-known reference point for considering questions of etiology in scientific medicine.

For instance, Richard Krause, the director of the National Institute of Allergy and Infectious Diseases of the NIH, gave a conference talk in the summer of 1983 on "Koch's Postulates and the Search for the AIDS Agent," noting that "technical difficulties" often "impede the fulfillment" of all of Koch's postulates, but concluding: "If we abide by the scientific guidance of Koch's postulates, we are sure to discover the cause of AIDS."^[123] Similarly, Lawrence Altman focused squarely on Koch's postulates in an article, published later in 1984, on "How AIDS Researchers Strive for Virus Proof."^[124] Altman presented Koch's postulates as an important medical tradition that researchers have looked to for a century, but he noted that doctors are sometimes forced to rely on immunological or other experimental evidence when Koch's postulates cannot fully be satisfied. With less equivocation, James D'Eramo wrote in the New York Native soon after the Heckler press conference: "The definitive classical proof that a virus or bacterium causes disease rests on causing the disease in animals by injecting them with the putative agent. AIDS has yet to occur in a laboratory animal."^[125] Dr. Nathan Fain, the medical writer for the national, West Coast-based gay newsmagazine the Advocate, made roughly the same claim in May when he explained why "work must continue to prove beyond all doubt that the candidate virus does cause AIDS."^[126]

Clearly, if Koch's postulates are the benchmark, then Gallo's May 1984 articles in Science by no means established HTLV-III as the cause of AIDS. But since the criteria for proving causation have been contested, it may be useful to assess the credibility of Gallo's claims-making by looking at a relatively weak version of the causation criteria presented in a recent epidemiology textbook. According to Mausner and Kramer, the likelihood that an association is causal can be evaluated by examining several criteria.^[127] First, there is the strength of the association, which they describe as the "ratio of disease rates for those with and without the hypothesized causal factor": here Gallo's evidence is compelling but far from perfect, since he was able to isolate the virus only in fewer than half of the samples from people actually diagnosed with AIDS. Second, the "dose-response relationship": does a higher dose of the causal factor result in higher rates of disease expression? Gallo had no data on this point. Third, the consistency of the association across different studies: clearly this was yet to be determined. Finally, is the association a "temporally correct" one, meaning that the cause precedes the expression with a sufficient "induction period" or "latency period"? With the exception of the

one virus-positive, clinically healthy gay man who developed AIDS within six months, Gallo had no relevant data to report.

Given the state of the evidence in early 1984, perhaps a more plausible claim was that of Jay Levy, whose results in isolating what he called "ARV," or AIDS-associated retrovirus, were published in August.^[128] Levy found signs of ARV in about half his AIDS patients and in about 20 percent of clinically healthy homosexual men, but in only 4 percent of clinically healthy heterosexuals. Levy recalled agonizing over how to phrase his conclusion: "I called a good friend of mine … who's an editor, and I said, 'How do I do this? I don't want to say it isn't, but I don't want to say it is .'"^[129] In the end, Levy's wording was cautious: "Although no conclusion can yet be made concerning their etiologic role in AIDS, their biologic properties and prevalence in AIDS patients certainly suggest that these retroviruses could cause this disease."^[130]

But for Gallo, the notion that he had proven the virus to be the cause became something crucial to defend, particularly as his credibility on other claims was challenged. In 1985, the Pasteur Institute sued the U.S. government in a patent dispute over the discovery of the virus, and in 1987 the heads of the two governments, Jacques Chirac and Ronald Reagan, signed an agreement splitting the royalties for the commercial antibody test. Especially after it became apparent that Montagnier's LAV had found its way into Gallo's viral cultures—a point that Gallo would formally concede in 1991—Gallo gradually backed off from claiming any primacy. And although Gallo continued to present the discovery of HTLV-III as a natural outgrowth of HTLV research,^[131] he was eventually forced to accept the prevailing view that, from a genetic standpoint, the new virus was not reasonably classifiable as an HTLV virus. In response to the confusing array of acronyms then in use—HTLV-III, LAV, ARV, HTLV-III/LAV, and others—the Human Retrovirus Subcommittee of the International Committee on the Taxonomy of Viruses rebuffed Gallo and agreed on a new, compromise, name in 1986: HIV, human immunodeficiency virus. (Levy and Montagnier signed the agreement; Gallo and his close associate, Max Essex, dissented.)^[132]