The Dubious Origins of the Linear No Threshold Model of Carcinogenesis

Most regulation of chemical exposures for causing cancer outcomes is based upon a linear no-threshold exposure-response (LNT) model.  This model informs risk assessments and policy judgments in the United States, Europe, and throughout the world.  If the LNT model were offered to support only precautionary principle judgments, then we might excuse the pretensions of LNT advocates.  The authors of LNT models, environmentalists, and a cadre of anti-industry scientists (“The Lobby”) are not willing, however, to be understood to be offering mere prudential guidance.  They claim that they are offering scientific conclusions, worthy of being taken seriously, as knowledge.

Passing off prudence, aesthetics, and hostility to industrialization as science is a deception that takes place virtually every day in the front pages of the lay media, in scientific journals, as well as in the Federal Register.  Implicit in the stridency of the Lobby is a desire to impose an often impossible burden upon industry to prove that there is no increased risk at miniscule exposures, and to challenge negative results with claims of inadequate power.

Professor Calabrese, in the current issue of the University of Chicago Law Review, has traced a source of the deception behind LNT to a series of events tied to the United States nuclear weapons program during World War II.  Edward J. Calabrese, “US Risk Assessment Policy: A History of Deception — A Response to Arden Rowell, Allocating Pollution,” 79 U. Chi. L. Rev. 985 (2012).   In 1927, Hermann J. Muller, a radiation geneticist, discovered that X-ray radiation caused mutation of fruit fly germ cells. The importance of health outcomes from radiation exposure led to Muller’s service as an advisor to the Manhattan Project during World War II.  In 1946, Muller received the Nobel Prize in medicine, for his work in genetics.

Professor Calabrese points out that Muller so thoroughly dismissed threshold models of radiation mutagenicity in his Nobel address, that the LNT became accepted dogma in the regulatory and scientific agencies of the United States government, and later, around the world.   Edward J. Calabrese, “Muller’s Nobel Prize Lecture: when ideology prevailed over science,” 126 Toxicol. Sci. 1 (2012).   What Professor Calabrese adds to the historical narrative is that Muller was aware of high quality research that undermined the LNT, when he announced in his Nobel lecture that there was no evidence for thresholds.  In the 1950’s, Muller continued to use his influence to advance the LNT model of radiation-induced carcinogenesis with the National Academies of Science Biological Effects of Atomic Radiation (“BEAR I”) Committee.

Professor Calabrese has provided an important cautionary tale about how scientific beliefs take hold and are propagated.  Muller’s influence and the rise of the LNT model is a recurring tale of the role of power, persuasion, and prestige in claiming scientific truths. It is not just about money.  Conflicts of interest involving professional honors, institutional recognition, friendships, intellectual commitments, and investigative “zeal,” which are often much greater threats to the integrity of science and medical research than receipt of payments.  See Richard S. Saver, “Is It Really All about the Money? Reconsidering Non-Financial Interests in Medical Research,”  40 J. Law, Med. & Ethics (2012).

If Professor Calabresse is right in his historical analysis, Muller achieved his goal of protecting his discovery from challenge by a selective presentation of the relevant data at a crucial moment in the formation of scientific thinking about mutagenesis. The rise and fall of the LNT model carries with it an important lesson to judicial gatekeepers.  Unlike the audience of Muller’s Nobel speech, lawyers and judges can, and must, insist upon a declaration of all materials considered so that they can determine whether an expert witness has proffered an opinion based upon a thorough, complete review of the relevant data.