Attack on Radiation Geneticists Triggers Furor

Calabrese is best known for his work on hormesis, the hypothesis that most if not all chemicals at low doses are actually good for you. He also says a bit more radiation than the natural background would be healthy for humans: "I believe we live in a radiation-deficit environment."

But current regulations on radiation exposure are based on a so-called linear no-threshold model, which says the effect is proportional with the dose—even at extremely low exposures. Calabrese says he got interested in Muller and Stern when he wanted to find out exactly how that model became so dominant after World War II. His research, published in September issue Archives of Toxicology and the October issue of Environmental and Molecular Mutagenesis, led him back to the BEAR I report, and ultimately to the studies that underpinned it. He spent years digging through the archives of the American Philosophical Society and read Muller's and Stern's correspondence to retrace how the scientific consensus arose.

The field has its origins in 1927, when Muller discovered that x-rays caused mutations in male fruit fly germ cells, which led him to voice concerns about the use of x-rays in medicine. He also started investigating the dose-response relationship of radiation and found that the evidence pointed in the direction of a linear relationship. Conventional wisdom in those days held that most toxic chemicals are harmless at low doses and that everything depends on the dose; as Paracelsus had put it, "Poison is in everything, and no thing is without poison."

Studies in the 1930s and '40s confirmed the linear relationship at intermediate and high doses of radiation. But questions about lower doses remained, in part because the studies were extremely cumbersome; to rack up enough mutations to reach statistical significance, scientists had to expose tens of thousands of flies to ionizing radiation.

In 1946, when Muller won the Nobel for his work, he was a consultant on two major studies funded by the Atomic Energy Commission that sought to answer the low-dose question. Both were led by Stern, then a professor of genetics at the University of Rochester. Calabrese says Muller "guided the group on breeding practices, data interpretation, and manuscript refinement."

The first of the studies, conducted by Stern and Warren Spencer, supported a linear relationship at low doses. But in August 1946, a month before Muller was awarded the Nobel, the results of the second study, which Stern had done with Ernst Caspari, became available. That work showed that flies exposed to similarly low doses of radiation didn't have significantly more mutations than control flies, supporting a "threshold model," in which radiation is innocuous below a certain level.

Muller received the manuscript of the Stern-Caspari study before he went to Stockholm for his Nobel lecture, and on 12 November 1946 he sent Stern a letter in which he wrote:

However, I see that it [the Caspari paper] is very important and shall do all I can to go through it in a reasonable time, surely before I leave again early in December. I hope that Caspari can wait that long if necessary. In the meantime I wonder whether you are having any steps taken to have the question tested again, with variations in technique. It is of such paramount importance, and the results seem so diametrically opposed to those which you and the others have obtained, that I should think funds would be forthcoming for a test of the matter.

Apparently he had not read the entire manuscript but was aware that it contradicted the no-threshold model, says Calabrese—and yet in his speech in Stockholm, Muller felt confident enough to say that there is "no escape from the conclusion that there is no threshold." "It was a misleading and dishonest statement that Muller made," Calabrese says. (A press release from his university about the papers bluntly says that "Muller knowingly lied.")

Schwartz says that conclusion is totally out of proportion. "Muller ... had 10 years of extensive experience in radiation genetics. To expect him to make an acknowledgement of this unpublished manuscript in his Nobel Prize address is the height of hypocrisy, it is totally absurd."

Kenneth Muller, a grandson of the Nobelist who's a professor of physiology and biophysics at the University of Miami, defended his grandfather as well in a recent story in The Chronicle of Higher Education. But in the same story, John Beatty, a professor of history and philosophy of science at the University of British Columbia in Canada who has studied Muller's career, agrees that given the paucity of data, Muller may have been too confident about the linear model. "How could Muller be so sure?" Beatty asks.

In the second paper, Calabrese aims to show that the Stockholm remarks were part of a broader pattern. He contends that Muller and Stern were biased by their post-war anxieties about radiation and tried to downplay the evidence for a threshold model in their handling of the Spencer-Caspari manuscript and later in the influential BEAR I report.

The Stern-Spencer and Stern-Caspari studies were both published in 1948 in the same issue of the journal Genetics, of which Stern was the editor at the time. Calabrese says that for a variety of methodological reasons, the Stern-Caspari study, which yielded evidence that radiation isn't harmful below a low threshold, was the superior one. Yet in the paper, Stern and Caspari seem to stress that the results could be just a statistical fluke, Calabrese says. "They say: 'don't accept our results unless you can explain why they differ from the Spencer paper.' "

Calabrese also discovered that a sentence mentioning the "possibility of a tolerance dose for radiation" appeared in the manuscript, but not in the published paper. He sees that as evidence that Stern, with the consent of Muller, suppressed scientific evidence. "With removal of this conclusion and the cautions about the differences, the paper was acceptable for Muller," Calabrese says.

To Schwartz, the accusations are way over the top. "They openly published their results and in the published paper they freely acknowledge the results," he says. Despite the deletion, the paper still contains a mention that the results point in the direction of a threshold. And it's not strange for Stern to be cautious about Caspari's results. "The Spencer results were entirely consistent with all the previous results. ... The Caspari results were unexpected and therefore Stern and Muller decided they had to be replicated."

Elof Axel Carlson, a former student of Muller who published a biography about him in 1981, is furious about the assault on his mentor. "Muller was enormously conscientious about the accuracy of the work he read," he wrote in an e-mail to Science.

After the two 1948 papers, Stern hired a young new researcher, Delta Uphoff, to do three new studies. The key outcomes were presented and compared with the previous two in a one-page Technical paper in Science in 1949. All three of the new studies appear to support the no-threshold model, although the authors acknowledged that "a final interpretation of the results cannot be offered," in part because in two of the new studies, control flies that received no radiation at all had unusually low mutation numbers, thereby inflating the apparent effects of exposure. The authors promised to publish full data on the three new studies later, but that never happened.

Despite its shortcomings, the Science paper had a major impact on the further development of the field, says Calabrese. It was cited in an influential 1957 review on leukemia and radiation by future Nobelist Edward Lewis in Science that promoted the linear model, as well as the BEAR I report, which concluded that "there is no minimum amount of radiation dose which might be exceeded before any harmful mutations occur." Calabrese says the report was a defining moment when the threshold model was replaced by the linear model as the standard in risk assessment. Within a few years, the new paradigm would be adopted all over the world.

Critics say that Calabrese's view may be colored by conflicts of interest. On his Web site, Carlson calls it "bothersome" that Calabrese's research is funded by chemical companies and the nuclear industry. Calabrese counters that around 80% of his funding comes from the government, that he is transparent about his sponsors, and that he does not consult for them.

Schwartz also notes that Calabrese has an intellectual bias, in that he's pushing the hormesis theory, which Schwartz calls "way out there." He points out that Calabrese tried in vain to convince a National Academy of Sciences panel of the benefits of low-level radiation in 2005. "The experts in the committee listened to his evidence and rejected it," says Schwartz.