Kicking the Sacred Cow

by James P. Hogan

  First, the two group (1) cases. In 1969, Tarján and Kemény, in Hungary, feeding low dosages to successive generations of a strain of inbred mice, reported a higher tumor incidence among the experimental animals than the controls, becoming statistically significant in the third and fourth generations. 196 But there were puzzling features. For one thing, tumors were lower in the controls than in the breeding stock, indicating some factor in the experiment that was not accounted for. Further, while the strain used was supposed to be leukemia-free, leukemia in fact occurred in both the experimental and control groups. Finally, nothing comparable—nor any cancer of any type—had been found by other researchers working with similar dose levels. A subsequent WHO investigation showed that all of the oddities could be explained by suspected contamination of the feed with aflatoxins—among the most potent of naturally occurring carcinogens. So one can agree with Epstein's assessment of the study's being defective in method or analysis. But of "highly suggestive"? Hardly.

  The remaining paper in this category described an earlier series of several experiments with rats. 197 Evaluation of the results was complicated by the fact that more than two-thirds of the test animals were lost for one reason or another over the two-year period. Of the 75 from 228 that survived, 4 developed liver tumors compared to 1 percent of the controls. Whether this is statistically significant is debatable, since the four came from different test groups, and no relationship was indicated between dose and effect. On the contrary, some of the rats that showed no liver cell necrosis had reportedly existed for twelve weeks on doses that other workers had found to be 100 percent lethal. Subsequent attempts to duplicate the reported results failed. The British pathologist, Sir Gordon Roy Cameron, who conducted one of these endeavors, a fellow of the Royal College of Pathologists, later knighted for his contributions to the field, observed that the 1947 study had employed formalin as a fixative agent for the tissues, which is not suitable for cytological studies on account of its tendency to produce artifacts of precisely the kind that had been identified as hyperplasia nodules. The inference could therefore be made that the results were "highly suggestive" of wrongly fixed tissue. 198

  This leaves us with two studies qualifying as well-designed and well-executed. In this connection Epstein sprang something of a surprise on the defense by introducing data from an in-house FDA experiment performed between 1964 and 1969 that nobody else had heard of, describing it as an excellent piece of work and expressing puzzlement that its findings had never been published. 199 The experiment involved two compounds, DDT and methoxychlor, fed to two strains of inbred mice. Epstein disregarded all the results for the strain-A mice, which experienced high mortality from a bacterial infection. This was not mentioned in the actual report but came to light in a memorandum from one of the researchers later, which included a terse comment that its effect on the study as a whole had not been determined. A second possible contributory factor was found in the sixty-seventh week with the discovery that the strain-A had been fed three times the intended dose for an undetermined period of time. Whether this had also been the case with the strain-B mice was not known. Well, one can only speculate that a good reason for refraining from publication and saving further expenditure of funds and effort might be found right there.

  Anyway, delving deeper regardless we find that taking both strains together there were actually more tumors among the controls (males 66, females 82) than among the experimental animals (males 63, females 73). For strain-B alone, a slight increase occurs for DDT-fed (males 42, females 50) versus controls (males 39, females 49). However, this applied to benign tumors only. If it is permissible to select just one subgroup from the entire study to make a point, then we could by the same token select just the strain-B females that developed malignant tumors, where the numbers were controls 10, DDT-fed 3. Hence, by a far larger margin, the same study could be shown as equally "conclusively" demonstrating DDT to be an anticarcinogen.

  The final paper, however, was the one presented as the proof, the main source that media articles ever since have cited in branding DDT as a carcinogen: the 1969 Bionetics Report, sometimes referred to as the Innes paper, after the first-named of its thirteen authors. 200 The work was part of an ambitious study performed by Bionetics Laboratories, a subsidiary of Litton Industries, under contract to the National Cancer Institute, on the effects of 123 chemical compounds in bioassays on 20,000 mice covering periods of up to eighty-four weeks. Epstein's confidence notwithstanding, the methods and findings have been widely criticized in the professional literature.

  One objection was that the researchers did not assure random distribution of the litters. Genetic disposition counts heavily in tumor occurrences among mice. Subjecting litter mates to the same compound runs the risk that genetic factors can mask the results, making any ensuing statistics worthless. Others were that the study failed to distinguish between malignant tumors and benign nodules, the significance of what was being measured being further obscured by relying on a single figure of "maximum tolerated dose" in lieu of providing more orthodox dose-effect relationships—at levels 100,000 times higher than those of residues typically encountered in food.

  But perhaps the greatest oddity, pointed out by Claus and Bolander (p. 351) was that whereas the authors of the paper lumped all five control groups together on the grounds that there was no significant variation between them, the actual data showed there to be large differences. For strain-X, as an example, the percentage of mice developing tumors varied from 0 to 41.2 percent for the males and 0 to 16.3 percent for the females. Applying regular statistical procedures reveals that for the group showing the highest tumor incidence—strain-X males at 41.4 percent—with the above degree of variation present in the controls, the maximum part of this that could be attributed to DDT is 5.5 percent, which no amount of manipulation could make significant. Following the same procedure with the strain-Y mice yields higher tumor percentages among the controls than among the DDT-fed groups.

  Their skepticism highly aroused by this time, the authors of Ecological Sanity then turned their attention to the report's listing where the tested substances were classified as "tumorigenic," "doubtfully tumorigenic," or "not tumor-producing," i.e., safe. And again, a raft of inconsistencies was found. Substances widely agreed upon as being carcinogenic received a clean score. Others considered to be innocuous—one, in fact, used for the treatment of alcoholics—did not. Compounds with similar molecular structures and chemical behavior received very different ratings. And piperonyl butoxide, which was tested twice, managed to end up on both the "doubtful" and the "safe" lists.

  Claus and Bolander point out that a program of this magnitude would involve the preparation of about 48 million tissue sections placed on 2,400,000 microscope slides, requiring 60,000 boxes. Making what appear to be reasonable estimates of the likely workforce, rate of working through such a phenomenal task, and the time available, they doubt that meaningful results would even be possible. They ask (p. 362): "What kind of an expert panel is it that can not only so warmly endorse this ambitious but dismally sloppy study and dignify it with the name of the NCI but also provide rationalizations for its obvious weaknesses?"

  A Plague of Birds

  One of the most serious ravages that can befall woodlands is infestation by the gypsy moth. The larvae devour all the foliage of trees, especially oaks, and in large numbers can strip bare entire areas, forcing other life left without food or habitat to either perish or migrate. This can be unpleasant and even dangerous, as inhabitants of northern New Jersey discovered in the seventies when large numbers of rattlesnakes and copperheads invaded suburban areas there. In 1961, the year before Silent Spring was published, large areas of Pennsylvania were sprayed with DDT to eradicate this pest. The Scranton Bird Club kept careful records but didn't report a single case of bird poisoning. Officials of the National Audubon Society were satisfied that no harm was done to bird life, including nesting birds.

  Yet Silent Spring was to stat
e (p. 118) that the robin was on the verge of extinction, as was the eagle; (p. 111) "Swallows have been hard hit. . . . Our sky overhead was full of them only four years ago. Now we seldom see any." Here we have another instance of assertion being flatly contradicted by the facts, for the very next year ornithologist Roger Tory Peterson described the robin as "probably North America's number one bird" in terms of numbers. 201 The Audubon Society's figures for annual bird counts bore him out, reporting 8.41 robins per observer in 1941 (pre-DDT) and 104.01 for 1960, representing a more-than twelve-fold increase during the years when DDT use was at its highest. The corresponding figures for the eagle and swallow were increases by factors of 2.57 (counts per observer 3.18, 8.17) and 1.25 (0.08, 0.10) respectively. 202 This pattern was general for most of the species listed, showing 21 times more cowbirds, 38 times more blackbirds, and no less than 131 times more grackles, the average total count per observer being 1,480 in 1941 and 5,860 in 1960. Gulls became so abundant on the East Coast that the Audubon Society itself obtained permission to poison 30,000 of them on Tern Island, Massachusetts, in 1971. Wild turkeys increased from their rare status of the pre-DDT years to such numbers that hunters were bagging 130,000 annually. Of the few species that did decrease, some, such as swans, geese, and ducks, are hunted, while bluebirds are known to be susceptible to cold winters.

  Ironically, some of the areas where birds seemed to thrive best were those of heaviest DDT use, such as in marshes sprayed to control mosquitos. Part of the reason seems to be that DDT is also effective in reducing insects that transmit bird diseases and which compete with birds for seeds and fruit. But perhaps even more important, DDT triggers the induction of liver enzymes that detoxify potent carcinogens such as aflatoxins that abound in the natural diets of birds

  None of this prevented any real or imagined decline in bird population from being immediately attributed to pesticides. 203 A severe reduction in eastern ospreys turned out to be due to high levels of mercury in the fish upon which they fed and to pole traps set around fish hatcheries—blamed on DDT even though reported as early as 1942. Alaska ospreys continued to do well despite high DDT residues. California brown pelicans increased almost threefold during the heavy DDT years but experienced a sharp decline at the beginning of the seventies—two months after an oil spill at Santa Barbara surrounded their breeding island of Anacapa (not mentioned in the reports of the state and federal wildlife agencies). In 1969 the colony had been severely afflicted by an epidemic of Newcastle disease transmitted from pelican grounds along the Mexican coast of the Gulf of California (also not mentioned). It was later established that helicopter-borne government investigators collected 72 percent (!) of the intact eggs on Anacapa for analysis and shotgunned incubating pelicans in their nests. 204 DDT was also implied as being connected with the reduction of the Texas pelican, even though the decline had been noted in 1939 and attributed to fishermen and hunters.

  The great eastern decline in the peregrine falcon was to a large degree due to the zealousness of egg collectors who have been known to rob hundreds of nests year after year, and then attribute the ensuing population collapse to the encroachments of civilization. In 1969, "biologists" studying peregrines in Colville, Alaska, collected fully one-third of the eggs from the colony and then dutifully reported that only two-thirds of the expected number of falcons subsequently hatched. 205

  Cracking Open the Eggshell Claims

  But by the time of the 1971 hearings, the main allegation, still perpetuating the fiction that a catastrophic fall in bird populations was taking place, had become that DDT was the cause not as a result of immediate toxicity, but indirectly through disruption of the reproductive cycle by the thinning of eggshells. This again goes back to Silent Spring, which states (p. 120), "For example, quail into whose diet DDT was introduced throughout the breeding season survived and even produced normal numbers of fertile eggs. But few of the eggs hatched." This was a reference to experiments performed by James DeWitt of the U.S. Fish and Wildlife Service, published in the Journal of Agricultural Food and Chemistry in 1956 (4: 10 pp. 853–66). The quail were fed 3,000 times the concentration typically encountered in the wild. 80 percent of the eggs from the treated group hatched, compared to 83.9 percent of the untreated controls, so the claim of a "few" was clearly false, and the difference in results hardly significant. Moreover, 92.8 percent of the eggs from the DDT-fed birds were fertile, compared to 89 percent from the controls, which reverses the impression created by the quoted text. Also omitted was that DeWitt's study was actually conducted on quail and pheasant. Of the pheasants, 80 percent of the eggs from the treated birds hatched compared to 57.4 percent for the controls, and 100 percent of the DDT birds survived against 94.8 percent of the control group.

  A number of later studies were cited at the EPA hearings, purporting to account for an effect (major decline in bird populations) that wasn't happening. All of them are examined in Ecological Sanity—with more care and thoroughness, it would appear, than by the witnesses who built their cases on them.

  A 1969 experiment (Heath et al.) on mallard ducks, performed at the Patuxent Wildlife Center at Laurel, Maryland, reported that birds fed DDT and DDE (the major metabolic residue from DDT breakdown) suffered a mortality among embryos and hatchlings of from 30 to 50 percent. The first thing that struck Claus and Bolander upon reviewing the paper was an enormous range of variation with the control groups that nobody else had apparently objected to. For instance, the number of eggs laid per hen in one control group was 39.2, whereas in another it was 16.8. This difference alone was far greater than any of the differences said to be "significant" among the experimental birds. The difference in live fourteen-day-old hatchlings in the same groups was 16.1 versus 6.0, which again was greater (69 percent) than the 50 percent deficit in ducklings per hen reported for the birds fed the highest DDT diet. When the variations among the controls are greater than the variations between the control and experimental animals, it should be obvious that some factor other than the test variable is operating (a bacterial infection, for example), which affects both groups. Claus and Bolander conclude (p. 406):

  On the basis of these absurd differences . . . the entire study becomes meaningless, and all of the conclusions presented by the authors have to be discarded." And (p.408) "How this paper could have been passed for publication in Nature is unfathomable, for even rapid scanning of the tables presented in the article should have made it immediately evident to the referees that the data for the two series of control birds invalidated the whole experiment.

  But published in Nature it was (227: 47–48), and it remains virtually unchallenged as one of the most frequently cited references in support of the contention that sublethal concentrations of DDT can be held responsible for declines in wildlife populations.

  The same issue of Nature (was there some editorial decision to convey a message here?) carried another paper cited at the hearings, this time by Dr. Joel Bitman and coworkers, describing experiments on Japanese quail. It concluded from direct measurements that DDT and related compounds induce a decrease in eggshell calcium and produce thinner eggshells. How were these conclusions arrived at?

  The quantities of test compound fed to the experimental birds were of the order of 100 times that found in the natural environment. As if this were not enough, the experimenters also introduced "calcium stress" in the form of a diet (given both to the control and two experimental groups) reduced from the normal level of around 3 percent calcium content to 0.56 percent. The question that the results needed to answer was, "Did the feeding of DDT add to any calcium depletion caused by the calcium stress conditions? The authors of the experiment claimed that it did.

  Their results, however, showed no significant differences in the calcium content of the blood or the bones that were analyzed from the three groups. It seems odd that if the calcium reserves of the parent birds showed no reduction, there should be a significant difference in the eggshells they produce. The significance reported was
0.07 percent, arising from 2.03 percent calcium content measured in the shells from the controls, versus 1.95 and 1.96 for the test groups. While mathematically the claim of significance is correct, it turns out that the method followed was to analyze the shell for the weight of calcium, which was then expressed as a percentage of the fresh weight of the entire egg. The weights of the shells themselves were not given, leaving wide open the possibility that eggs smaller and presumably lighter in total weight could nevertheless have possessed shells that were heavier. Hence, it's not possible to tell from the presented data whether the percentage of eggshell calcium was reduced or not, which was the whole point of the exercise.

  It gets even more interesting when we look at the measuring of eggshell thickness. This was done with a mechanical screw-type micrometer after the shell membranes were removed. An average reduction is reported of 69.5 x 10-4 inches for the controls to 66.7 x 10-4 in. and 65.6 x 10-4 in. for the eggs of two groups of test birds, and is described as "highly significant." Well, let's look at it.

  Converting these figures to metric at the same accuracy as that given—namely of three significant figures—yields figures for reduction in thickness of 0.00711 and 0.00991 millimeters. The last two digits of each are below the resolving power of a light microscope, and eliminating them leaves reported thinnings of 7 and 10 microns. (1 micron = 0.001 mm—about half the size of a well-developed bacterium.) Screw micrometers available at the time were not considered to resolve differences below 50 microns reliably. More recent devices graduated to 10 microns are useful for gauging thicknesses of materials like metal foil, but for compressible samples such as paper—or eggshells—the determination of "end point" is a subjective quantity based on feel, typically resulting in variations of 10 to 30 microns. To borrow the phrase so beloved of textbook writers, it is left as an exercise for the reader to judge if such methods could have given results justifiably acclaimed as being highly significant.