Day 038 - 19 Oct 94 - Page 30


     
     1        For that group of 18 compounds, a man called David
     2        Zaltsberg took those known human carcinogens, carcinogens
     3        known to cause cancer in humans, from occupational
     4        epidemiological studies.  He then looked at the
     5        corresponding rat feeding studies and asked the question:
     6        How often do the rat studies correctly pick up the
     7        carcinogenicity?  He claimed the rat studies then got it
     8        right.  I think his figure was something like 38 per cent
     9        at the time, which indicates that they are extremely
    10        unreliable.
    11
    12        It perhaps will not surprise you to know that this provoked
    13        a lot of concern and further work.  Some of the compounds
    14        were studied again in different species, at higher doses,
    15        scrutinised more thoroughly.  By the late 80s, the most
    16        recent figure I have seen suggested that for known human
    17        carcinogens the rat feeding studies picked up the
    18        carcinogenicity maybe 80 per cent of the time, which is an
    19        improvement in the science but still a relatively poor
    20        basis of policy making, because it implies that the rat
    21        feeding studies were missing, on average, 20 per cent of
    22        all carcinogens.
    23
    24        Since we know more about carcinogenicity than anything
    25        else, than other toxic end points, we might reasonably
    26        suppose that carcinogenic toxicology is a more developed
    27        science than other parts of toxicology.  We therefore might
    28        plausibly think that animal feeding studies in respect of
    29        other end points are even less reliable.
    30
    31        So, I think the answer is, the relatively brief answer to
    32        your question is there have been all too few studies
    33        estimating the relative reliability of different kinds of
    34        animals; even less has there been the kind of research,
    35        which I have long been advocating, which is research
    36        intended to estimate the reliability of different kinds of
    37        animal studies, and, better still, to develop new kinds of
    38        studies which are demonstrably more reliable.  So, we know
    39        very, very little about the reliability of these studies.
    40
    41   Q.   As regarding when setting ADI's, are they set with
    42        reference to results in the most sensitive animal species?
    43        A.  Not always.
    44
    45   Q.   Should they?
    46        A.  There is a marked disparity between the theoretical
    47        literature in which the ADI concept was defined and the way
    48        it is used in practice.  All the early definitions said --
    49        firstly, all the early definitions said the concept of the
    50        ADI should not be used in respect of carcinogens at all. 
    51        There was then a retreat from that and they decided they 
    52        would not use it in respect of genotoxic carcinogens and 
    53        might use it for non-genotoxic carcinogens.
    54
    55        But normally the rule is supposed to be that the ADI is set
    56        by reference to the most sensitive species of those tested,
    57        not of all known species, because which species the tests
    58        are conducted in is itself something determined at the
    59        discretion of those conducting or sponsoring the tests.
    60        But, even where we have data in the public domain

Prev Next Index