A.  Clinical laboratory procedures.

Hematologic, clinical, chemical and urinalysis examinations are described on pages 5-7 of Volume I of the submission.  The same rats were employed for all clinical laboratory examinations throughout the study.  In cases where one of these rats died during the study, another rat chosen from a corresponding group was substituted.

The following hematology parameters were measured at treatment days 42,92,189,364,547 and 734:  hematocrit, hemoglobin, total RBC, total WBC, differential WBC, and prothrombin time.

The following clinical chemistry (serum) measurements were made: pyruvic transaminase (days 42,92,189,364,547,736), glutamic oxaloacetic transaminase (days 41,92,189,364,547 and 734), alkaline phosphatase (days 42,92,189,364,547.734), total bilirubin (days 42,,92,189,364,547,734) blood (serum) urea nitrogen (days 42,92,189,364), total cholesterol (days 42,92,189,364,734), L-phenylalanine (days 42,92,189,364,547,734) sodium (day 734), potassium (day 734), calcium (day 734), protein electrophoresis (day 734).

The following urinalysis (2 hour collection) measurements were made at days 42,92,190,364,547, & 734: specific gravity, pH, occult blood, protein, bilirubin, microscopic on sediment, and phenylketones; glucose and ketones were determined at days 42,92,190,364, & 734; urobilinogen was measured at day 42,190,364 and 547.

We noted that some of the data sheets for urinalysis had erroneously labeled the phenylketones test values as "phenylalanine" (see exhibit #84).

Some cholesterol and BUN determinations were carried out which were not described in the submission to FDA.  They were as follows:

1)  Serum cholesterol determinations were done at days 796 & 798 (terminal bleeding), but not included in the submission to FDA.

The protocol indicated that clinical chemistry determinations, including serum cholesterol, were to have been performed at termination.  The submission to FDA (Vol. 1 p. 286) reported a significant decrease in serum cholesterol that was more perceptible towards the end of the study, and may have been related to compound administration.  Therefore, the omitted data may have been important.  (Copies of these data were obtained and are attached as exhibit #77, Section V.

2.  BUN determinations were done at day 546 but not reported in the submission to FDA (see exhibit #77 Section V).

3)  Serum cholesterols were also done on day 546 and not reported in the submission (see exhibit #77).  These determinations were only done for females, and only for a few animals, reportedly due to insufficient quantity of sample.

4)  BUN's were also done on day 735 and not reported in the submission. This data was not complete for all animals at day 735.

5)  Additional animals (other than those designated) were bled at the regularly scheduled times and determinations were made.  These determinations were not reported and we could not determine why the animals were bled.  (See Exhibit #77)

B.  A list of persons involved with lab analysis along with their responsibilities and duties is as follows:

1)  Judith R. Hehmal? - Nov l971 to present, Supv., Clinical Chemistry, section of Bioanalytical laboratory.

2)  Judith A. Beauchamp - Supervisor, Hematology Laboratory, April 1971 to present.

3)  Denise Prikins? - Supervisor Hematology Laboratory until April l971 (no longer employed by Searle).

Bart Tangonan

Tony Martinez

David Kie

Robert Spaet

The above four persons in the toxicology department were involved with assembling data for clinical chemistry and hematology determinations for April 1973 to Feb. l974.

Joyce Schulmann - performed urinalysis and hematology determinations from April l973 to Feb. l974.

Philip Muellner - Technician in Path-Tox Dept. July l970 till end of study.

Janet Praal - Technician, prepared individual work sheets for urinalysis. No longer employed by Searle.

C.  The following employees were interviewed regarding clinical lab procedures, and methods for recording clinical lab. data.

1)  Bart Tangonan on 6/1/77 regarding the recording of data.

2)  Judith Beauchamp, on 6/2/77 regarding hematology and urinalysis.

3)  Judith Schmal, on 6/2/77, 6/7/77, and 7/29/77 regarding clinical chemistry.

4)  Tony Martinez, on 6/3/77 regarding urine and blood collection, and recording of data.

5)  Jane Drury, on 6/7/77 regarding electrophoresis.

Accounts of these interviews are attached as exhibits #47-54.

D.  Other Documents and Procedures Used to Authenticate Clinical Laboratory Data values in Submission were as follows:

1)  One loose leaf volume entitled "SC-19192: 104 Week Oral Toxicity Study In The Rat.  PT - 988573 Protocols, Organ Weights, Dosage, Hematology, Urinalysis, Blood Chemistry, Protein Electrophoresis."  The volume was subdivided into sections according to the above parameters.  The individual pages (See Exhibit #77, Section IX for example) are composed of forms containing the appropriate measurements and units printed on the left side of the page onto which data on "sticky back sheets" corresponding to each of these measurements were pasted in columns representing the various time periods.  These pages, in addition to other information, were headed by the identifying number of the rat for which the measurements were made.  The information on the sticky back sheets (see Exhibit #77, Sec. IX) was copied (hand written) from laboratory notebooks, sheets, Autoanalyzer Charts, teletype sheets (on line data generated by analytical instruments) or computer printouts (containing raw and calculated data resulting from on line or off line input data from instruments) by individuals of General Toxicology Section (see interview with Bart Tangonan). Many of the pages were initialed "BRT" (apparently by Bart R. Tangonan).  Most of the final values transcribed into the sticky back sheets resulted apparently from calculations made directly by the analytical instruments or by external computer using the appropriate stored equations and data for the reference standards.

(2) Since the values appearing in the volume referred to in the above section were copied from other sources, an attempt was made to verify these values by examining the information in these sources.  No attempt was made to recover the teletype sheets, or computer printouts which we were told were no longer available or could not be recovered (see interview with Judith R. Schmal, Exhibit #54).  All laboratory notebooks that might contain the original data were requested.  Notebooks dated prior to the dates of the DKP study were excluded.  The appropriate laboratory notebooks were then identified by BA numbers which were listed on the top sections of the sticky back sheets included in the volume referred to above. Examination of these few laboratory notebooks revealed only a very small amount of data would could be used for additional verification of the values in the submission.  It was necessary to obtain the consultation of Judith Schmal to clarify the system used to relate the values in these books with the corresponding rat and period of time of bleeding.  The following notebooks, as designated by information on the front covers, were examined:

1)  Lab. notebook #N-26375 (hematology), 25 June 71 to 1/21/72.

2)  Lab notebook #127133 (phenylalanine), 10/8/71 to 4/21/72.

3)  *Lab notebook #113239 (cholesterol), dated 5/1/72.

4)  *Lab notebook #17, BA #0007118926 (SGOT), 12/27/71 to 2/25/72.

5)  Lab notebook #126472 (phenylalanine), 4/21/72 to 6/8/72.

6)  Blue Book #1591, identified "JF VON - 70" (hematology).

7)  Columnar book #21, identified "JF VON 27" (Differential cell counts).

8)  Spiral notebook identified "JABEA-" (coagulation/prothrombin) dated 7/23/71.

9)  *Spiral notebook #16, (SGOT), 8/27/71 to 12/16/71.

*Those books (3,4, & 9 above) marked with an asterisk provided us with no useable data, because a formula or standard curve (no longer available) was necessary to convert the data.

Copies of the applicable pages from all of the above notebooks were obtained, and are included in exhibit #77.

The following data were cross checked against available data from original entries (in addition to being checked against transcribed data on "sticky back sheets" in bound volume):

1.  Hematology - Erythrocytes: Treatment days 42,91,364 & 546 Males and Females.

2.  Hematology - Leucocytes, WBC: Treatment Days 42, 91, 364, & 546 Males and Females.

3.  Hematology - Leucocytes, Differential: Treatment Days 42,91,189,364, & 546, Males and Females.

4.  Hematology - Coagulogram, Prothrombin Time: Treatment Days 42 & 91, Males and Females.

5.   Phenylalanine. Treatment Days 42,189 Males and Females, Day 91 Males.

E.  Discrepancies were found between the clinical laboratory methods described on pages 5-7 of submission Volume 1 (referenced on page 120) and those actually carried out. These discrepancies were documented by the interviews described in Section C and in a document (Exhibit #77, Section II) voluntarily submitted by Jutidy Schmal, June 7, 1977 in response to requests for clarification of the clinical chemistry procedures as they were actually conducted in regard to analytical methology instrumentation, and processing and recording of data.

1)   Glutamic Pyruvic Transaminase.

Reference:  Russell, C.D. and Cotlove, E. (1971)/ Clin. Chem. 17; 1114

The reference describes a coupled reaction U.V. assay for serum glutamic oxaloacetic transaminase in which malic dehydrogenase is used.

As described by Judith R. Schmal (June 7,1977) glutamic pyruvic transaminase was assayed by a method adopted from Sigma Kit Technical bulletin #410 - U. V. using Lactic acid dehydrogenase.

2)   Glutamic Oxaloacetic Transaminase

Reference: Same as in (1) above.

As described by Judith R. Schmal (June 7, 1977), from November 1971 to March 15, 1972 a manual colorimetric method (Fermco Kit) was used (employing dinitrophenylhydrazine). From March 15, 1972 the method used was adapted from Sigma Kit, Technical bulletin #410 - U. V. using Lactic acid dehydrogenase.

3)   Blood (serum) urea nitrogen.

Reference: Marsh, (Marsh in Submission) W.H., Fingerhut, B. and Miller, H. (1965). Clin Chem 11, 624.

The referenced method calls for reaction of urea with diacetyl monoxime in the presence of thiosemicarbazide and ferricions in a relatively weak acid solution.

As described by Judith R. Schmal (June 7, 1977) the method used from November 1971 to February 1, 1974 was adapted from Fermco Kit Bulletins #20 and 20-1.  Urea is hydrolyzed to ammonia and carbonic acid in the presence of urease. Ammonia is detected by the Berthelot reaction to produce indophenol.  From February 1, 1974 the "direct serum methond" modified from the method of Marsh et al was used.

4)  Phenylalanine

Reference: Hill, J. B., Summer, G. K. Pencer, M. W. Resz N.O. (1965) Chin Chem 11, 541

From Nov. 1971 to about Septembe 1972 there is no documentation in file as to method used.  From about September 1972 the method used was a flurometric determination in the presence of ninhydrin and l-leucyl-l-alanine as adapted from McCaman and Rubins.  (This is a manual method modified for automation by Hill et al - reverenced above) (Judith R. Schmal, June 7, 1977)

5)  Calcium:

Reference Pybus J., (Pylrus in submission), Feldman, F. J., and Browers (Borrers in Submission) Jr., G. N. (1970) Clin.Chem. 16 (11 in submission), 998.

The referenced method involves the measurement of total calcium in serum by atomic absorption spectrophotometry.  As described by Judith R. Schmal (June 7, 1977) from November, 1961 to February 1974  the procedure used was a colorimetric procedure using Corinth dye as adapted from Kingsley and Robnet.  From May 21, 1973 the method used was atomic absorption spectrophotometry, as adapted from Pybus et al (reference above)

6)  Total Cholesterol

Reference:  Levine J. S., Morgenstern, S., and Vlastelica, D. (1968).  Automation Anal Chem. pp 25-28, Technicon Symposia 1968.

We were unable to check the above reference because of difficulty up to now in obtaining a copy of the publication, but as shown below two different procedures were employed to measure total serum cholesterol at different times during the study.

(Judith R. Schmal, June 7, 1977).  From November 1971 to July 2, 1973 the method involved reacting an isopropanol extract of serum with ferric chloride (modified from Block, Tirret and Levine). From July 2, 1973 the method used was a direct serum method using a modified Lieberman - Burchard reaction.

7)  Glucose

Reference:  Frings, C.S., Ratliff, C.R. and Dunn, R.T. (1969) Advances in Automated Analysis 1, 73

This reference was not checked (because of difficulty up to now in obtaining a copy of the publication) but as shown below two different procedures were employed to measure serum glucose at different times during the study (Judith R. Schmal, June 7, 1977).

From November 1971 to October 16, 1972 the method was a glucose oxidase determination (modified Gertrud Acrow) using protein free filtrates.  From October 16, 1972 the method was a direct serum O-Toluidine reaction as modified from Frings, Ratlif and Dunn.

In the case of four of the above parameters (glutamic pyruvic transaminase, glutamic oxaloacetic transaminase, blood urea nitrogen and calcium) different methodology was used during part of the study then was indicated in the submission.  For one parameter (phenylalanine), there was no documentation as to the method used for one period of the study and for two other parameters (total cholesterol and glucose), two different methods were used for each of the parameters while only one was referenced in the submission.

Alkaline phosphatase was measured generally as referenced in the submission (McComb, R.B. and Borrers, G.N. (1972). Clin. CHem., 18, 97 in that the method involved measuring the production of p-nitrophenol from p-nitrophenylphosphate However starting July, 1973 there was a "re-optimization of reagent concentrations" (Judith R. Small, June 7, 1977).

The above changes in procedure could conceivably result in differences in the apparent absolute values for the concentration of the substances measured.  Changes in the method of conversion of raw data to calculated values as was don in the determination of sodium and potassium by atomic absorption spectrophotometry during different periods of the study, (Judth R.Schmal, June 7, 1977) could also possibly produce differences in final values.

In an interview with Judith Schmal on June 2, 1977, she did state in response to a question that two levels of "Serum Controls" were used in each run to check the method and instruments and that the data was not reported if the values were more than two standard deviations greater than that for the expected values.

NO evidence was obtained that any attempts were made to determine whether or not DKP cold interfere with any of the clinical laboratory tests conducted.  For that matter no information was made available to us as to whether DKP itself or related compounds did appear in the blood or the urine of rats fed diet containing this compound.

Neither, as a result of interviews held or reference to available laboratory notebooks were we able to obtain information helpful in explaining the unusually low values for BUN for the control males at treatment days 189 and 364 and for all the treated male groups at treatment day 364.  No raw laboratory data in reference to this could be found and may have been recorded on discarded teletype sheets referred to previously.  In reference to the low BUN values, Page 29 of the submission contains the following statement: "BUN" values for the control males at treatment day 189 were unusually low and may possibly be related to a technical artefact; as a result, the group mean values for all treated males at this interval were significantly higher but, in fact, these values were in the normal range.  BUN values both in control and al treated male groups at treatment day 364 were unusually low;  this again reflects a possible technical artefact."

F.  A total of 21 disparities between individual clinical laboratory analysis values appearing in the submission Volume I and those values appearing in data sheets and/or laboratory notebooks were found (Table 4).  Of these, 17 were in hematology, one in clinical chemistry, and three in urinalysis.  As a result of the discussion with Robert Bost, it was apparent that some of the hematology discrepancies may have resulted for Searle personnel mistaking recorded instrument readings for calculated values.  In two cases no value or crossed out values appeared in the laboratory notebooks while values were found entered onto the appropriate places in the data sheets.  For animal number A01HM and treatment day 546 four discrepancies (hematocrit, hemoglobin, RBC and WBC) were noted.

G. Discrepancies Found In Statistical Analysis:

The mean and standard errors for the three dose levels and the controls for the various measurements using the values in the submission Volume I or values noted in the data sheets (where there values differed from those found in the submission) were calculated by the Division of Mathematics, FDA. Also supplied were the results of the T-TEsts comparing the controls to the treated groups.  See memo to Leonard Friedman from Dennis Wilson, dated July 20, 1977 with attached Tables 1 and 2 (Exhibit #87).

A total of 49 disparities were found, which were comprised of 6 means, 23 standard errors and 20 significant differences. As stated in the memo, in all cases where there is a disparity, it appears to be due to differences in the data.

Calculations were also carried out for cholesterol data found in the data sheets but not reported in the submission.  As shown in Table 5 the mean values for the median and high level treated females and the high level treated males were significantly lower than the mean values for the respective controls.  To illustrate the possible significance of these changes and disparities between the values calculated by Searle and FDA for cholesterol data at the other time periods of treatment, table 5 was constructed.  Very few disparities are seen between the calculated values obtained by FDA and those in the submission but a fairly consistent trend is seen for treatment related lowering of serum cholesterol, particularly at the two highest dose levels and for the female rats.