Day 039 - 20 Oct 94 - Page 76
1 conducted".
2
3 Q. "No new studies"?
4 A. "No new studies", only the ones that were already
5 available. Now, the data show that, yes, indeed, only at
6 two per cent over that period do you generate the full
7 tumours, the carcinomas, but at lower doses you get the
8 induction of hyperplasia which is the precancerous lesion.
9
10 Q. It gives you the percentages of diets at which that
11 happens, does it not?
12 A. It does deeds in those species, but you have to be
13 careful with interpreting this. It looks as though it is
14 saying there were no changes at all at the .1 per cent dose
15 level. But that is not strictly correct. Typically in
16 these documents what they fail to do is emphasise the
17 limited sensitivity of these experiments, and what they
18 should -- I mean, the way in which these results, I think,
19 should more properly and more scientifically be reported is
20 that, well, first something about the level of statistical
21 confidence that one could have in the conclusion, given the
22 sample that were used on the conclusion that .1 per cent
23 did not produce lesions. I suspect if they had done that
24 it would be clear that while no effect, no statistically
25 significant effect, occurred, the groups were so small that
26 one could not have substantial confidence that that was a
27 genuine negative rather than merely an artefact of the
28 small, sample size.
29
30 Now, they say: "After re-evaluating the data in pigs, it
31 was concluded that the evidence is ... questionable". What
32 I find when I read these documents is that every parts of
33 the evidence is questionable, but the committees
34 selectively choose to question in detail some evidence but
35 not others. Just to say "it is questionable" seems to me
36 to be a rhetorical device and not a systematic, scientific
37 observation.
38
39 Then they say again a couple of sentences down:
40 "Considering the absence of any significant effects in the
41 two dog studies, it was concluded that further
42 investigations in animals without forestomachs are not
43 required".
44
45 Now, that is a judgment about what they deem sufficient,
46 but there is no reason to think that the dog is a better
47 model of the human, it provides a better model of the human
48 digestive tract than a rat does. What happens is compounds
49 get tested -- here is a typical example -- in a wide
50 variety of species. There are no a priori grounds for
51 thinking in advance of the conduct of any study that any
52 particular species is a better or worse model than any
53 other.
54
55 Q. May I suggest a explanation which might appeal to a less
56 suspicious mind, Dr. Millstone? It is this, that it had
57 been observed for some considerable time that the
58 forestomachs of rats were susceptible or vulnerable to the
59 formation of hyperplasia and carcinomas at certain
60 dosages. By comparing that phenomenon with the
