Senator Howard M. Metzenbaum,                    (Dated 3 November, 1987)
United States Senate,
140 Russell Senate Office Building,
Washington, DC, 20510

Dear Senator Metzenbaum,

   The following represents a continuation of my letter to you of last
week, October the 30th, 1987.  In that letter I discussed the many serious
problems with the quality or reliability of the experimental studies with
aspartame carried out by or for G.D. Searle & Co.;  I noted there that in
1976, the FDA Commissioner at that time, Dr. Alexander Schmidt, speaking
for the FDA as an agency, publicly stated that he agreed with a set of
conclusions, the first of which was that the FDA had no basis for reliance
on the quality of studies generated by or for that firm.

   Once such a determination is made at the highest level of the FDA, it
seems bizarre, to say the least, that essentially the same set of studies
could provide a foundation for the subsequent decision that those studies in
fact had demonstrated the safety of aspartame with "reasonable certainty"
as required by the Food Additive Amendment of the Federal Food, Drugs, and
Cosmetics Act.  As the television commercials for Weyerheauser, the
"tree-growing company", keep telling us:- "once the eagles are gone, they
are gone."

   Much the same is true also for experimental or laboratory rats:- once
they are gone, no one can bring them back for an interview to ask them how
much, if any, aspartame or DKP they had ingested while the experimental
studies in which they had participated were in progress and, without such
essential information, examination of their preserved tissues by even the
most skillful and competent of pathologists becomes largely a meaningless
exercise which cannot in any way resurrect in Phoenix-like fashion the
value of those studies.

   However, having said all of this, let us assume that in fact those
studies were of an acceptable quality;  let us pretend that the test
animals were actually exposed qualitatively and quantitatively to what G. D.
Searle & Co. would have us believe that they were exposed;  that there was
no post-mortem autolysis of their carcasses rendering vast numbers of their
tissues to a state unsuitable for pathology examination;  that the
technicians involved in the conduct of those studies were fully trained,
competent, and adequately supervised to make observations on those animals
prior to their death;  that the same was true with respect to the
observations made after their death;  that in fact those technicians actu-
ally made proper such observations;  that the proper samples of tissues with
grossly observed lesions were in fact collected for additional microscopic
examination;  that the identity of such tissue specimens corresponded (as
they should) to the identity of each animal that was their source, etc.  In
short, let us make believe in a spirit of Halloween that nothing which was
uncovered for the aspartame studies by the FDA investigations of 1975 and
1977 was actually true, i.e., that in fact we are dealing here with studies
of an absolutely perfect quality or reliability.  Of course, such assump-
tions belong to the domain of Fantasyland, but, nevertheless, let us play
this little game for awhile.

   Under such highly speculative hypothetical conditions, let us now ask
again whether aspartame can be viewed as being safe with "reasonable cer-
tainty".

   To answer this question, let us focus for a moment on the pathology
examinations carried out not by the pathologist originally retained by
GD Searle & Co. (those of the Experimental Pathology Laboratories, or EPL)
who examined the tissues from the rats in the Two-Year Rat Study) but,
rather, on the examinations carried out by the expert pathologists in the
UAREP.  Although in my last letter addressed to you last week I referred to
the investigative efforts of the UAREP as being "amateurish" by comparison
with those of the professional investigators in the FDA, I have no reason
to question or criticize in any way the  competence of UAREP pathologists in
their own specialty where they had examined first-hand tissue specimens
said to gave been collected from the animals in that study.

   The UAREP report (Volume 2, Chapter IV, dealing with that particular
study, reveals in Appendix IV-21 on its page 393 et seq. the animals which
were found by the UAREP pathologist to have harbored brain tumors:-

  Group   Sex   Path.No   Animal No.   Type of brain tumor   Weeks to death
    1      M    64-603    83-651       Astrocytoma               104
    2      M    64-775    83-745       Astrocytoma               104
    3      M    64-764    83-837       Astrocytoma                76
    4      M    64-707    83-919       Astrocytoma               104
           M    64-712    83-888       Oligodendroglioma          59
           M    64-713    83-892       Astrocytoma                49
           M    64-715    83-895       Astrocytoma               100
    5      M    none
    1      F    none
    2      F    64-989    83-769       Astrocytoma               104
                65-011    83-766       Astrocytoma                69
    3      F
    4      F    64-925    83-934       Astrocytoma                85
    5      F    64-881    84-010       Medulloblastoma/
                                       meningeal sarcoma          13
                64-888    84-019       Astrocytoma                67

   Altogether the table just above lists a total of 12 animals with brain
tumors, 7 males and 5 females;  for both sexes there are 1 in Group 1, 3, in
Group 2, 1 in Group 3, 5 in Group 4, and 2 in Group 5 for a total of 12.
Note that the GAO report which refers to those animals at the bottom of its
page 45 is in error in that it lists 4 (rather than 3) animals with brain
tumors in Group 2 (the low dosage group).  Because of this error, the GAO's
Figure 4.1 on page 46 of its report is somewhat misleading.

    The GAO report also indicates under item (2) on its page 34:-
         "According to UAREP's president at the time of its review"

         "...the thing that impressed (UAREP) throughout the
         study,... which is reflected in our final statements
         and conclusions, was that the interpretations of the
         experimental results by previous observers did not
         really differ very significantly from ours following
         our review of the material."

   Yet, Appendix IV-25 beginning on page 446 of the UAREP report repre-
sents a 6-page table entitled "Significant Discrepancies Between Histo-
pathologic Diagnoses By UAREP and EPL", the last mentioned having been, as
stated above, the "previous observers", i.e., the pathologists originally
retained by GD Searle & Co. to examine those tissues and whose report was
submitted by that firm to the FDA in support of their petition to have
aspartame approved for marketing.  In that table I have counted some 207
such "significant discrepancies" between the diagnoses of the UAREP and EPL
and these involve some 162 animals or 37% of all the 440 animals in that
study.  This  was not reported by the GAO representatives who apparently
were content with merely chatting with the UAREP president about his
reminiscences of some 10 years ago.

   Moreover, that same UAREP report reveals in that very same Appendix
IV-25 as cited above for the 12 animals with brain tumors the characteri-
zations or diagnoses reached by the pathologists from the EPL:-

for animal No. 83-651 with an astrocytoma of the brain
                      EPL lists the brain as unremarkable;
for animal No. 83-745 with an astrocytoma of the brain
                      EPL lists no comparable diagnosis;
for animal No. 83-837 with an astrocytoma of the brain
                      EPL lists no comparable diagnosis;
for animal No. 83-919 with an astrocytoma of the brain
                      EPL lists no comparable diagnosis;
for animal No. 83-888 with an Oligodendroglioma of the brain
                      EPL lists no comparable diagnosis;
for animal No. 83-892 with an astrocytoma of the brain
                      EPL lists no comparable diagnosis;
for animal No. 83-895 with an astrocytoma of the brain
                      EPL lists no comparable diagnosis;
for animal No. 83-769 with an astrocytoma of the brain
                      EPL lists no comparable diagnosis;
for animal No. 83-766 with an astrocytoma of the brain
                      EPL lists no comparable diagnosis;
for animal No. 83-934 with an astrocytoma of the brain
                      EPL lists an ependymomo\a i.e., a different
                      kind of brain tumor;
for animal No. 84-010 with a medulloblastoma/meningeal sarcoma of
                      the brain, EPL lists a meningioma i.e., a
                      tumor of the membranes covering the brain;
for animal No. 84-019 with an astrocytoma of the brain
                      there was no discrepancy in the EPL diagnosis.

   In other words, for the 12 animals identified as having brain tumors
in this study by the UAREP pathologists, EPL pathologists (i.e., the "pre-
vious observers" as the president of the UAREP has it) had completely
missed no less than 9 or 75% of these.  Such difference between the
diagnoses of those two groups cannot by any stretch of the imagination be
interpreted by any reasonable person as being "not very significant" as
that same president of the UAREP is quoted by the GAO to have stated.
Incidentally, the GAO representatives themselves also  failed in their
report to highlight this tremendous difference between the diagnoses of the
UAREP and EPL.

   Furthermore, Appendix IV-20 on page 391 of that same UAREP report
reveals in the first row of the table on that specific page that GD Searle
& Co. or their agents had provided to the subcontracting EPL pathologists,
i.e., to those whose report that firm had originally submitted to the FDA:-

a)  only 8 (or only 10%) of the brain sections for the 80 animals
    in group 2;

b)  only 7 (or only less then 9%) of the brain sections for the 80
    animals in Group 3;

c)  only 5 (or only less than 7%) of the brain sections for the 80
    animals in Group 4;

and the UAREP were proved with the brain sections of 2 fewer animals than
were provided to the EPL.  Again, this is another little wrinkle not high-
lighted in the GAO report.

   This, quite by itself, is sufficiently eloquent on just how G.D.
Searle & Co. saw fit to discharge their responsibilities in reporting fully
and completely their results of the Two Year Rat Study with aspartame to
the FDA;  it is just as eloquent on precisely how thoroughly the Bureau of
Foods of the FDA (the predecessor of the CFSAN) had reviewed the data
emanating from that study prior to its initial approval in 1974 for the
marketing of that food additive.

   I note at the bottom of page 54 in the GAO report that CFSAN had
objected to the medulloblastoma that was noted in a female rat at the top
exposure level on the grounds that "it was unlikely aspartame caused this
tumor".  Such statement would imply that aspartame had caused all the other
tumors (the nine astrocytomas and the solitary oligodendroglioma noted in
animals exposed to it) which is vastly more than enough to lead to a
conclusion that, because of this, it cannot be regarded as being a safe
food additive.  The reason for such conclusion by the CFSAN appear in the
first paragraph of page 46 of the GAO report.  As is also true for many of
the other arguments advanced by the CFSAN and by G.D. Searle & Co., those
reasons are largely speculative and without much merit.  Still, to accom-
modate the CFSAN's views regardless of their validity, I am willing to
ignore the occurrence of that particular tumor in a female animal at the top
exposure level.

   If we are to analyze the distribution of the rest of those brain
tumors, we ought ignore also the response of any animals at the top level
of exposure (Group 5) on the grounds that completing toxicity may well have
inhibited the expression of brain tumors in the animals of that  group.

   Accordingly we have for the male animals with brain tumors:-

in Group 1 i.e., at      0 mcm/kgm body-weight 1/59 =  1.69% positive rats;

in Group 2 i.e., at  1,000 mcm/kgm body-weight 1/36 =  2.78% positive rats;

in Group 3 i.e., at  2,000 mcm/kgm body-weight 1/40 =  2.50% positive rats;

in Group 4 i.e., at  4,000 mcm/kgm body-weight 4/40 = 10.00% positive rats;

   This particular distribution yields a dose-response slope as high as
0.000,019,865 with standard error of only 0.000,009,729,2 leading to a chi
square with one degree of freedom for slope as high as 4.118, whose
one-sided probability is as low as p = 0.021,217;  in other words, the

dose-dependent increase in frequency of brain tumors for the male rats
in that study was highly significant and, therefore, attributable to
aspartame, the agent on test.

   That particular slope of the dose-response function yields the  follo-
wing expected incidences of brain tumors amongst male animals:-

at     0 mcm/kgm body-weight - 0.867%
at 1,000 mcm/kgm body-weight - 2.854%
at 2,000 mcm/kgm body-weight - 4.840%
at 4,000 mcm/kgm body-weight - 8.813%

   Note that the four expected values given just above are fairly close
to their respective observed values listed near the bottom of the preceding
page, which indicates a close fit of the observations to the dose-response
or regression function.

   If we have reference to the animals of both sexes with brain tumors,
we have:-

in Group 1 i.e., at      0 mcm/kgm body-weight 1/118 = 0.847% positive rats;
in Group 2 i.e., at  1,000 mcm/kgm body-weight 3/ 76 = 3.948% positive rats;
in Group 3 i.e., at  2,000 mcm/kgm body-weight 1/ 80 = 1.250% positive rats;
in Group 4 i.e., at  4,000 mcm/kgm body-weight 5/ 80 = 6.250% positive rats;

   This particular distribution yields a dose-response slope as high as
0.000,011,578 with a standard error of only 0.000,005,831,8 leading to a chi
square with one degree of freedom for slope almost as high as the one for
merely the male animals, 3.920, with on-sided probability almost as low as
that for merely the male animals, p = 0.023,860.  The conclusion that
follows is identical with that reached above for merely the male animals.

   The expected incidences for both sexes are:-

at     0 mcm/kgm body-weight - 1.006%
at 1,000 mcm/kgm body-weight - 2.164%
at 2,000 mcm/kgm body-weight - 3.322%
at 4,000 mcm/kgm body-weight - 5.638%

or, again, fairly close agreement tot he observed values given just above.

   Note that in the analyses outlined above I have not combined the
response noted at two or more experimental groups, as was done by the PBOI
and as objected to by the CFSAN.

   If we now analyze the data in the same "uncombined" fashion, while
still excluding from consideration the medulloblastoma manifested by a
female in the top exposure level group, and even if we do consider the poor
response of the animals in the top exposure level group (which, as noted,
may have been due to competing toxicity interfering with the expression of
brain  tumors), but consider the so-called "historical control" incidence of
brain tumors (49/59,812 cited by Dr. Olney in his table 2 on page 2 of Part
III of his written statement presented to the PBOI as well as the rate of
4/115 positive control animals noted by both the UAREP and EPL for the
Lifetime Toxicity study of aspartame in the rat - see UAREP report, Chapter
V, page 559) along with the contemporaneous (local) control rate of 1.118
positive animals of both sexes noted in the Two-Year aspartame study in the
rat, we end up with a total of 54/60,045 = 0.090% for the control inci-
dence for both sexes.  The weighted averages of the exposure level in Group
5 animals was 7,420 mgm/kgm body-wieght.  Accordingly we would have:-

at     0 mcm/kgm body-weight - 54/60,045 = 0.090% rats with brain tumors;
at 1,000 mcm/kgm body-weight -  3/    76 = 3.947% rats with brain tumors;
at 2,000 mcm/kgm body-weight -  1/    80 = 1.250% rats with brain tumors;
at 4,000 mcm/kgm body-weight -  5/    80 = 6.250% rats with brain tumors;
at 7,420 mcm/kgm body-weight -  1/    77 = 1.299% rats with brain tumors;

   This distribution yields a slope of dose-response function as high
as 0.000,005,297 with a standard error of only 0.000,000,423,4, leading to a
chi square with one degree of freedom for slope as high as 156 whose one-
sided probability is as low as 4.031E-36, i.e., 4 with 35  zeros ahead of it
and to the right of the decimal point.  This is nothing short of astronomi-
cally high significance.

   Alternatively, if one considers merely the contemporaneous or local
control value in the two-year rat study with aspartame, 1/118 = 0.85% ani-
mals positive for brain tumors, the response at the lowest level of expo-
sure, 1000 mgm/kgm body-weight, 3/76 = 3.95% animals similarly positive
for brain tumors, is elevated by comparison with that control rate more
than 4.5 times which is borderline significance at p = 0.058,674.  The
response at the next to the highest level of exposure of 4,000 mgm/kgm
body-weight, 5/80 = 6.25% animals with brain tumors, is elevated more than
7.3 times over that same control rate of 0.85%, and this is highly signifi-
cant at the p = 0.009,975 probability level.

   Finally in this entire consideration of significance for the brain
tumors, one could set up yet another sort of contrast by making believe
that all animals exposed to aspartame were in fact exposed to the highest
level tried, 7,420 mgm/kgm body-weight.  This would extend a great deal of
the benefit of doubt to aspartame.  that particular contrast of 0.090% the
control rate versus 10/313 = 3.195% for all exposed animals (still
excluding the medulloblastoma objected to by the CFSAN), leads to a chi
square adjusted for continuity and with one degree of freedom as high as
254 which is, again, of almost astronomical significance.

   In other words, even if one is willing to give aspartame a very
generous benefit of doubt on the quality or reliability of the two-year
study in rats as well as several other considerable benefits of the doubt
involved in the test of significance, it still emerges that the rate of
brain tumors amongst the animals exposed to it vastly exceeds that for
animals not exposed to it and such excess is very highly significant.  What
this says is that there cannot be any reasonable, or even shadow of a doubt
that aspartame had caused such an increase in the incidence of brain tumors.

   It follows, therefore, that the conclusion of the PBOI and of several
members of the FDA Commissioner's panel of experts is the right conclusion,
and that reached by the CFSAN and by the FDA Commissioner who overturned
the PBOI view in this respect is the wrong conclusion.

   As a result of all the considerations above, I would add my full
endorsement to the conclusion of the unidentified statistician mentioned in
paragraph 3 on page 56 of the GAO report who apparently reached the same
conclusion as I did in an independent manner.

   I would also support the views of the similarly unidentified
carcinogenicity specialist mentioned in paragraph 2. of that same page in
the GAO report who felt that the relatively high exposure rates in the two-
year rat study with aspartame were a necessary compensation for the relati-
vely low power of this study to detect as significant increases as high as
5% in the brain tumor rate for humans exposed to aspartame, which would
constitute a downright catastrophe.

           The Acceptable daily Intake (ADI) of aspartame.

   Still under the hypothetical assumption that these experimental stu-
dies were of an  impeccable quality, let us now turn to a different aspect of
the interpretation of results arising from them.

   Near the bottom of page 60 of the GAO report it is disclosed that the
Acceptable Daily Intake (ADI) of aspartame was raised from 20 mgm/kgm body-
weight to 50 mgm/kgm body-weight after aspartame was approved for use in
carbonated beverages and after it became evident to the FDA that very young
children could potentially consume almost 50 mgm/kgm body-weight of it per
day.

   It appears that the justification for such sudden and considerable in-
crease of 150% in the ADI for aspartame was provided by the results of five
clinical studies as well as five other studies published in the literature;
however, it is unclear from the GAO report whether any of those studies
were of a long duration (such as a major part of the life-span) -
clearly, such  studies conducted with humans could not possibly have been of
this nature.

   To examine whether an ADI of 50 mgm/kgm body-weight can be justifiably
regarded as "safe", let us return to the issue of the brain tumors and
conduct for these a formal Risk Assessment.  Although it has been esta-
blished here that the incidence of brain tumors in rats was highly signifi-
cantly related to the dosage of aspartame in the two-year rat study (and,
therefore, that aspartame had cause that increase in incidence of brain
tumors amongst exposed animals by reference to the rate noted in comparable
unexposed ones) such determination of high significance is in fact not a
necessary requirement for a formal risk assessment.

   I have carried out such risk assessment by utilizing two separate pro-
cedures which are widely accepted for this purpose:- the Mantel-Bryan ap-
proach (also known as the log-probit method) and the Hone Hit method of
extrapolation.

    To extend again the benefit of doubt to aspartame, I have had refe-
rence in such assessment to the control rate of brain tumors noted merely
in the local or contemporaneous control animals (1/118 = 0.847%) rather
then to the almost ten times lower rate of the "expanded" control group
discussed in the previous section here (54/60.045 = 0.090%); also, I have
assumed all non-control rats to have been exposed at the top levels
of exposure (7,420 mgm/kgm body-weight) rather then to a series of levels
starting at merely 1,000 mgm/kgm body-weight;  I have also excluded from
consideration the medulloblastoma observed for Animal No. 84-010, but have
not excluded the response of any other animal in that study.  Each of these
features, as mentioned, provides the benefit of doubt to aspartame, i.e.,
to its "producers" as distinct from its "consumers".

   With such additional assumptions, we may tabulate the estimated
"virtually safe" levels of aspartame in the mgm/kgm body-weight/day for a
variety of upper limits on the risk indicated in the column at the extreme
left of the table that follows here .

   Note that for each of the two methods of extrapolation, two estimates
are given in the table opposite each upper limit on the risk:-  on for rats
and one for humans.  The estimate for the humans is related to the
corresponding one for rats by being 5.23 time smaller then it.  This
is the factor necessary for "translation" from rats to humans by correcting
for the body-area of the two species:-  due to its larger size, the human
has a body-area per unit mass smaller than does the rat:-

   An average male rat in the study considered here weighed 506 Gms., and
an average female rat 331 Gms., for a mean weight of 418.5 for the two
sexes.  This a human of average weight of 60 Kgms., say, is "worth" on a
 mass or weight basis 60,000/418.5 = 143.37 rats of average weight.  But
that same human weighing 60 Kgms is worth on a body-area basis only the
two-thirds power of 143.37 i.e., only 27 .39 such rats.  Thus, to have
equivalence for doses expressed in mgm/kgms body-weight rats and
humans, the dosage for the rats must be divided by the one-third power of
143.37, i.e., by 5.23.  Hence the factor used in the table that follows.

         RESULTS EMANATING FROM THE FORMAL RISK ASSESSMENT
                        INVOLVING BRAIN TUMORS

              "virtually safe" level of aspartame in mgm/kgm bw/day
                            Log robit method       One Hit method
   Upper limit on risk   for rats  for humans   for rats    for humans

   1/100,000,000         0.700       0.134      0.001,278    0.000,244
   5/100,000,000         1.349       0.258      0.006,392    0.001,22
   1/ 10,000,000         1.809       0.346      0.012,78     0.002,44
   5/ 10,000,000         3.674       0.702      0.063,92     0.012,2
   1/  1,000,000         5.050       0.966      0.127,8      0.024,,4
   5/  1,000,000        10.95        2.09       0.639,2      0.122,0
   1/    100,000        15.55        2.97       1.278        0.244
   5/    100,000        36.81        7.04       6.392        1.22
   1/     10,000        54.63       10.45      12.78         2.44
   5/     10,000       146.5        28.01      63.93        12.2
   1/      1,000       232.3        44.42     127.9         24.5
   5/      1,000       759.1       145.2      640.8        123.0

   It turns out from the entries in the table just above that an ADI of
50 mgm/kgm body-weight for  humans is associated by both methods of extrapo-
lation with an upper limit on the risk as high as between 1/1,000 and
5/1,000 population exposed to aspartame to develop brain tumors as a result
of exposure to that food additive.  For this to actually become evident, it
would take many years since such tumors have a very long latent period,
i.e., it takes a long time for them to become manifest.  Thus, it seems to
me that we are dealing here with a huge time bomb.

    There is hardly any need for me to emphasize here that this represents
an unacceptably high risk or hazard posed by aspartame.

                           SUMMARY AND CONCLUSIONS.

   From what has been discussed in my letter addressed to you last week
as well as from what has been presented in the previous pages of this com-
munication, I can conclude the following:-

   1.  It is impossible for anyone to appreciate just how a determination
by the FDA that the G.D. Searle & Co. experimental studies with aspartame
were of an unacceptable quality in 1976 can be metamorphosed several years
later into a view by that same Agency that essentially the same studies
were sufficiently reliable for anyone to assess that this food additive is
"reasonably certain" to be safe for consumption by humans.

    2.  Even if, contrary to the FDA's view in 1976, the quality of the
conduct of those studies could be relied upon by the same agency to even
begin making such a determination, at least one of those studies had
reveled a highly significantly dose-related increase in the incidence of
brain tumors as a result of exposure to aspartame.

   The full incidence of those brain tumors was not disclosed by G.D.
Searle & Co. to the FDA prior to the initial approval for the marketing of
aspartame in 1974;  moreover, the  review of that study in the FDA was so
flawed that the Agency apparently did not even realize tat that time that
only a portion of the observations on brain tumors had in fact been submit-
ted by G.D. Searle & Co. in their petition for that approval.

   3.  Quite aside from the remarkable significance of the increased
incidence with dose of those brain tumors, the ADI of 50 mgm/kgm body-
weight recently set by the FDA for the human consumption of aspartame is
alarmingly dangerous in that it involves an extremely high and, therefore,
a totally unacceptable upper limit on the risk for those consuming aspar-
tame:  between 1/1,000 and 5/1,000 population to develop brain tumors as a
result of such exposure.

   4.  Although in their report the GAO express the view that the FDA
"followed its required process in approving aspartame (for marketing)" I
would sharply disagree with such evaluation.  Although the FDA may have
gone through the motions or it may have given the appearance of such a
process being in place here, the people of this country expect and require
a great deal more from that agency charged with protecting their public
health:- in addition to mere facade or window-dressing on the part of the
FDA, they require a thorough and scientifically based evaluation by the
Agency on the safety of the products it regulates.

   Unfortunately this has clearly not been the case here.  And
without this kind of assurance, any such "process: or dance represents no
more than a farce and a mockery of what is truly required.

Sincerely yours

M. Adrian Gross,
Senior Science Advisor,
Benefits and Use Division,
Office of Pesticide Programs
Sworn  to be a true copy on 3 November, 1987.