By Edward J. Calabresea, & Robert J. Golden – Re-Blogged From Junk Science
1. IntroductionThe US Congress passed, and President Richard Nixon signed intolaw the Safe Drinking Water Act in 1974. A significant provision of theAct involved engaging the US NAS to advise the EPA on multiple sci-entific and technical areas such as chemical and radiation risk assess-ment, including cancer risk assessment. To achieve these goals the NAScreated the Safe Drinking Water Committee (SDWC) in 1975. In 1977the SDWC published the 700 pageDrinking Water and Health reportoffering EPA widespread guidance, including cancer risk assessmentand its underlying scientific foundations that supported the LNT.Within two years EPA would issue the first national drinking waterstandard for a chemical carcinogen using the LNT for total trihalo-methanes (THM) . This action would jump start an avalanche ofother LNT based cancer risk assessments by EPA, not just for drinkingwater but for other environmental media as well.
The decision to golinear by the SDWC for chemical carcinogens was therefore as highlysignificant as it was precedent setting, and led the way for future EPAcancer risk assessment actions.The actions of the SDWC to recommend LNT for chemical carcino-gens was more than two decades after a similar recommendation of the1956 NAS BEAR Genetics Panel to switch from a threshold to LNT forradiation induced mutation. This action of the BEAR Genetics Panel wassoon followed by a recommendation of the National Committee forRadiation Protection and Measurement (NCRPM) to generalize the LNTconcept to somatic cells for cancer risk assessment. This two decadetime gap in the decision to go linear for cancer risk assessment forionizing radiation and chemical carcinogens suggests the possibilitythat chemical toxicologists and radiation geneticists/cancer researchersmay have evolved considerably differently with respect to the conceptof cancer risk assessment, prompting the present paper.
2. How radiation geneticists came to embrace LNTWhile ionizing radiation and chemicals induce cancer, their his-torical research foundations concerning cancer risk assessment havesome important differences. In the case of ionizing radiation, it wasknown as a human carcinogen within a decade of its discovery in 1895as well as having a range of diagnostic and therapeutic applications .The applications of X-rays lead to the development of substantial sci-entific, occupational health, and clinical data with important implica-tions for cancer risk assessment. There was also considerable researchassessing ionizing radiation induced mutations and their dose responserelationships in multiple biological models.During the initial two decades following Muller’s report on X-rayinduced gene mutations , the radiation genetics community becameconcerned with protecting humans from the harmful effects of X-rayson germ cells, developing embryos and fetuses.
The protection ofworkers exposed to ionizing radiation also became a priority with someradiation geneticists becoming members of national (e.g. NCRPM) andinternational advisory committees (e.g. ICRP) starting in the 1930’sconcerned with health and safety [5–7].These activities quickly drew the radiation genetics community intothe domain of human risk assessment, led by Muller’s Proportionalityhttps://doi.org/10.1016/j.cbi.2019.108736Received 15 May 2019; Received in revised form 18 June 2019; Accepted 3 July 2019*Corresponding author.E-mail addresses:email@example.com(E.J. Calabrese),firstname.lastname@example.org(R.J. Golden).Chemico-Biological Interactions 310 (2019) 108736Available online 03 July 20190009-2797/ © 2019 Elsevier B.V. All rights reserved.T
Rule and its LNT-single hit  dose response model for germ cell mu-tation and later application to cancer risk estimation [5, 7, 9–11]. Thislinear dose response assumption lead to the conclusion that there wasno safe exposure to ionizing radiation, challenging a threshold modelinterpretation for some birth defects and most cancer endpoints.What is striking during this time period, and perhaps little appre-ciated, is the paucity of experimental animal model studies concerningionizing radiation induced cancer (See Stannard and Baalman  for adetailed history of experimental radiation cancer research in the USfrom the 1930s through the 1970s).
Large-scale animal studies wereinitiated by the US Atomic Energy Commission (AEC) with extensivelifespan Beagle dog studies in the early 1950s and continued for severaldecades. These studies were undertaken following preliminary researchprincipally with mice and rats during and after World War II.Thefindings from animal studies during the period leading up to theNCRPM cancer linearity recommendation in late 1958 were thereforelimited and those which were potentially relevant were generallyviewed as not supportive of an LNT recommendation. In a 1958 articledealing with radiation-induced cancer in experimental modelsGlucksmann  noted that“none of the animal experiments haveindicated a linear relationship between tumor incidence and dose”.OnSeptember 19, 1958 in the journalScienceFinkel  claimed to havepublished“the best current information on the shape and origin of thedose-response curve”in lifetime experiments assessing the effects ofStrontium 90 on tumor formation in mice.
This study employed 12treatment groups with up to 150 mice in the control and lowest ex-posure group. In general, there was no treatment related responses atthe lowest four doses. The lowest dose tested was 100 fold greater thanthe level established for the general population. This most definitivestudy for that time-period also did not provide support for the LNTmodel. In further agreement, Upton , following a detailed review ofthe animal experimental cancer data, stated that“In no instance re-corded to date, has a linear relationship between neoplasia and radia-tion dose been adequately demonstrated.”These consistent animalmodel study perspectives merged with thefindings of Russell ondose rate in mouse spermatogonia and oocytes which discredited theradiation geneticist LNT dogma that all ionizing radiation induced ge-netic damage was cumulative, not repairable and irreversible.
These developments just prior to the NCRPM’s late December 1958generalization of the 1956 BEAR Genetics Panel germ cell linearityrecommendation to cancer risk assessment raised the question of howthis NCRPM committee made its LNT cancer risk recommendation inlight of the mounting scientific questions and uncertainties, if not openchallenges.In NCRPM Committee discussions in 1958, E.B. Lewis indicated thatthe LNT model predicted a 1 × 10−6leukemia risk with 1 rad/year,which was translated into a 10% increase in leukemia incidence peryear when exposure was assumed to be twice the background dose (page 613). This risk at low doses was hard to reconcile with theemergingfindings of Russell  and others that mutation thresholds occurred at many thousands of times greater than background dosesand the non-supportive animal model cancer studies.
The battle lineswere therefore drawn in the radiation community between those sidingwith the extrapolative predictions of the LNT model and those sup-porting empirical animal studies that were consistent with a thresholdconclusion. With both sides articulating their concerns with opposingviews, a compromised position was adopted based on a PrecautionaryPrinciple driven LNT policy acknowledging it was not based on“sound”science (see Ref. -page 5, left column, for a summary) but uponboth a fear of ionizing radiation and possible limitations in availabledata sets such as sample size, the capacity of animal models to predicthuman responses, the capacity of epidemiology to detect very low risks,amongst other factors.Such Committee decisions often are affected by the backgrounds,beliefs, and potential biases of those comprising the committee. In thiscase the NCRPM was comprised of highly prestigious individuals butonly a few with relevant education, training and experience such asJames Crow , Edwin B Lewis , and Clinton C. Powell [17, 20]each of whom had a published record of support for the LNT, despite itslimitations. Their LNT position was challenged by Austin Brues ,director of research at Argonne National Labs, probably leading to thecompromised position, yet one still favoring the adoption of LNT.3.
Why toxicologists were skeptical of LNTWith respect to chemical carcinogenesis risk assessment, tumorswerefirst induced in 1918 by chemicals in tars rubbed on rabbit ears and Ichikawa 1918]. By the early 1930s extensive experimentalstudies had established that numerous polycyclic aromatic hydro-carbons (PAHs) were carcinogenic in animal models [23–34]. Suchfindings were subsequently submitted to quantitative analyses via lowdose biostatistical modeling which set the stage for similar regulatorycancer risk assessment estimates some four decades later by EPA[35, 36].During these early decades of the 20th century, thefield of chemicalcarcinogenesis profited greatly from the collaboration of pathologists(e.g., Kennaway) [25–27] and synthetic organic chemists, such as JohnW. Cook who synthesized and tested many hundreds of compounds fortheir carcinogenic effects [28, 29].
Cook was widely credited with usinghis organic chemistry synthesis skills to assess tumorigenicity withsingle compounds, thereby permitting the reproduction of tumorsunder experimental conditions. This research was also integrated withadvances influorescent spectroscopy that lead to the identification ofsimilar spectra of carcinogens ,facilitating the development ofstructure activity frameworks for predicting carcinogenic hydro-carbons. By 1947 the number of papers published on the carcinogeni-city of individual compounds had exceeded an astonishing 5000 .While these developments revealed a striking relationship between chemical structure and biological effects and had a profound influenceon the development of thefield, it failed to provide a sound mechanisticfoundation for tumor induction.Despite these limitations,
Cook explored further mechanistic un-derstandings when it was revealed that sterol hormones and relatedagents (e.g., vitamin D, ovarian hormones, bile acids) had the samechemical backbone as did the carcinogenic polycyclic hydrocarbons.Such structural similarities suggested functional relationships implyingthat sterols might be precursors of carcinogenic hydrocarbons. This leadto the hypothesis that endogenous sterols may affect the occurrence ofspontaneous tumors via abnormal metabolic mechanisms . Theconverging role of sterols and hydrocarbon carcinogens in cancerbiology then lead to attempts to implicate hormones in cancer causa-tion. These developments suggested that PAH carcinogens may mod-ulate sterol metabolism, closing the gap between chemically inducedcancer and normal hormonal effects. From these developments emergedthe idea that chemically induced cancer was governed by pharmaco-logical principles that there were well known to be mediated viathreshold dose responses.
As suggested above, the research of Kennaway, Cook and others hada dominating impact on thefield of cancer risk assessment. For ex-ample, by 1936 the British Medical Association annual meeting had asession entitled“Substances Promoting Normal and Abnormal Growth”,with Cook presenting hisfindings that linked sterols and hydrocarboncarcinogens. By 1939, Kennaway and Cook would receive thefirst AnnaFuller Memorial Prize for their research on carcinogenic aromatic hy-drocarbons. Further reflecting the impact of his chemical carcinogenresearch, Kennaway would receive the King’s Medal in 1941 and beknighted in 1947, an honor that would also be given to Cook.Kennaway was also nominated for the Nobel Prize in 1951 and 1953.Despite extensive research on radiation induced mutation followingthe dramatic discovery of radiation-induced gene mutation by Muller(1927), chemical induced mutation evaded discovery, not being re-ported until 1946, some 20 years later, achieving this with mustard gasE.J. Calabrese and R.J. GoldenChemico-Biological Interactions 310 (2019) 1087362