In some cases original data could be recorded in several areas, making it difficult, and sometimes impossible to determine which was actually the original.  This was a particular problem in dealing with dates of deaths, as some conflicted on the "source" documents.  Many of the responsible individuals involved with the study, including stability testing of DKP, are no longer employed by Searle.  Dr. K.S. Rao, Study Monitor, the only individual who could have possible answered some questtions, had left Searle.  He was contacted, but permission for an interview was refused by his attorney.  Due to the absence of various individuals it was not always possible to a accurately determine methods used in some analyses and operations carried out in conducting this study.  In a number of areas, including chemistry, statistics, diet preparation and feeding, it was necessary to use assumptions, or information supplied by current employees who were not involved with the study.

At the beginning of this investigation , Mr. James R. Phelps, Vice-President and General Counsel for G.D. Searle & Co., advised us that an attorney and scientific coordinator would have to be present at all times to protect their interest in the data. This did not present any insurmountable problems, but on several occasions an attorney would question our request for data, stating that it was not relevant for authentication.  At no time did we make any statement to the effect that our goal was to authenticate the study.  Two memos were discovered dealing with reaction of animals to the diet.  This was a significant factor in the study.  Permission to copy them was initially refused, but finally granted after Searle was contacted by FDA General Counsel.  We were not allowed to make xerox copies of any documents for about two and one-half weeks, due to Searle's concern over confidentiality.  This was eventually reconciled between Searle and FDA General Counsel.

The major discrepancies concerning Study PD 988S73, SC-19192:

1)  Control and treated animals were randomly distributed on the same rack. (See diagram of housing group attached as exhibit 7.)

2)  No ear clips or other methods of uniquely identifying each animal were used.  Identification consisted of two types of cards attached to the front of each cage.

3)  Compound inventory cards were deficient in that only one of 18 such cards stated the purpose (study 988S73) for withdrawing the compound from inventory.  Three of the cards did not include the date withdrawn, amount withdrawn, or signature of requestor.  Therefore it was impossible to reconcile the amount withdrawn and the amount used. (See exhibit #28.)

4)  Food jars were not individually identified, yet all the filled jars for a given housing group (control, low, mid, and high dose) were placed on a mobile cart, which was wheeled to the housing rack.  The position of the jar (in rows) on the cart was the only means of identifying the proper dose level.  The arrangement of the food cups on the cart is shown in exhibit #8.

5)  A total of 79 "observations for drug effects" records were not signed or initialed.

6)  Observation records indicated that animal A23LM was alive at week 88, dead from week 92 through week 104, alive at week 108, and dead at week 112.

7)  Records indicated that at the scheduled 104 week bleeding,  animal E2CM was substituted for animal A11CM.  Records also indicated that animal A11CM was alive on this date and therefore should have been bled as scheduled.

8)  Records indicated that penicillin was administered to four rats beginning on May 16, 1973, and continuing daily through May 28, 1973.  This third occurrence of infections disease and penicillin administration was not reported in the submission to FDA.

9)  In many cases the actual number of tissues embedded was less than the 24 (control and high dose) or 19 (low and mid dose) specified in the final histology lab protocol dated 1/21/74.

10) Ophthalmoscopic examination records were present for animals H26MF and J29CM, yet the findings were not reported in the submission to FDA.  Two other discrepancies of this type were noted.

11) Records indicate that a tissue mass measuring 1.5 x 1.0 cm was excised from animal B3HF on 2/12/72, and that a "skin incision over mass" was performed on animals C22LM and G25LM on Feb. 10, 1972.

B.  Stability and Homogeneity of DKP in Diet Mixtures

1)  There were no batch records to show the quantities of DKP and basal diet weighted, type of mixer used, mixing time, dates, or names of individuals performing the weighing and blending operations.

2)  There was no evidence that any tests had been done to determine the blending characteristics of the mixer, or to validate the mixing time.

3)  No homogeneity tests were performed on any batches of diet used in the study, and to stability study assay reports (A7738 and A7739) indicated that samples were not homogeneous.  (See exhibit #29.)

4)  A stability study was conducted with DKP in 1972.  However, the 115 week rat study employed Basal Diet from week 2 to its conclusion, and to stability studies had been conducted with Basal Diet.

5)  Methods of assay for DKP in the diet were deficient in that: The titration method was discontinued after 1 week of the stability study.  Some of the TLC photographs showed no DKP reference standards and photographs also showed that there was something in the basal diet itself producing a spot on the TLC plate which had an Rf value corresponding to DKP.  Only one solvent system was  used for development of the TLC Plates.  Some of the chromatograms showed poor separation.

6)  No reserve samples of any of the lots of DKP used in this study were retained by Searle.

7)  Three different sets of specifications for DKP were found, and Searle could not determine with any degree of certainty which of the three were applicable to the 7 lots of DKP used in the study.

8)  The analytical records for DKP lots IR through 5R refer to reference standard IR #3701.  None of the three sets of DKP specifications lists reference #3701.  No data was made available as to dates, methods of preparation and authentication of DKP reference standards.

9)  Analytical records A-9129 for DKP lot 5R showed an assay of 1000%.  Examination of laboratory notebooks showed that eleven (11) samples had been analyzed from this lot, and the analytical record only reflected an average of the last three of theses.  The other assays (not reported) ranged from 87.93% to 114.83%.

C.  Dosage, Body Weight and Food Consumption

1)  Examination of the raw data sheets revealed the following discrepancies:

1. Empty feed cup weights were missing for the D housing group at the 12th week, in the raw data sheets.  (See exhibit #75.)
2. In several instances, the dietary concentration shown on the weight sheets did not agree with the concentration listed for the same level in the other housing groups.  (For example; C group Males, mid & high levels for week 13,; A group Males, high levels for week 99.)

2)  Comparison of the Searle submission and the independent FDA analysis of the raw body weight and food consumption data revealed the following discrepancies:

1. We found a total of 15 differences of 1 gram or more in the average body weight and of 0.1 percentage points or more in weight gain.  (See table 1.)
2. We found approximately 82 discrepancies of one gram or more in the food intake when expressed in grams/day. (See  table 2.)
3. We found approximately 40 errors of 5 or more grams in food intake when expressed in grams/kg./day.  (See table 2.)
4. Most of our dosage calculations differed from Searle's dosage calculations by 10 or more mg., when the dosage is expressed as mg/kg/day.  (See table 2.)

D.   Gross and Microscopic Pathology

1)  98 of the 196 animals that died during the study were fixed in toto and autopsied at some later date, in some cases more than one year later.

2)  A total of 20 animals were excluded from the study due to excessive autolysis.  Of these, 17 had been fixed in toto and autopsied at a later date.

3)  Records indicated that animal F6HF, a high dose female, was found dead at 787 days of treatment and the gross pathology sheet reported a tissue mass measuring 5.0 X 4.5 X 2.5 cm. The submission to FDA reported no tissue mass and the animal was excluded from the study due to marked autolysis.

4)  Records for approximately 30 animals showed substantial differences between gross observations on pathology sheets, when compared with the gross observations on pathology sheets submitted to FDA.  A detailed description of 10 of these is included in the report.  Copies of all the gross pathology sheets, and the pathology summaries submitted to FDA are attached as exhibits.

5)  Dr. Charles H. Frith, D.V.M., Ph.D., Directory, Pathology Services, NCTR, examined slides for a total of 150 animals, or about 42 percent of the animals on study.  He noted the following discrepancies:

1. The reporting of a mass (by Searle) as missing which was actually present (animal M1LF.)
2. The finding of a polyp of the uterus which was not diagnosed by Searle (animal K9MF).  The finding of this additional uterine polyp by Dr. Frith increases the incidence in the midi dose to 5 of 34. (15 percent.)
3. The finding of ovarian neoplasms in animals H19CF, H19C, and H7HF, and the finding of diffuse hyperplasia in animal D29CF, which were not diagnosed by Searle.
4. The finding of additional inconsistencies in 21 animals.

6)  No microscopic worksheets or other "raw data" relating to microscopic pathology could be found for this study.

7)  A mammary tumor found in animal F27CF was described as a papillary cystadenoma on the pathology summary sheet, (page 105, Vol. II of the submission) and as an adenocarcinoma on summary table 12 (p. 95, Vol. I of the submission).

8)  In several instances the histopathology technician made notes at the bottom of the gross pathology sheet to indicate that certain organs were not present in the bottle of fixative (and were therefore not available for sectioning).  Yet, in three of these instances (animals A4CM, K23CF, and J3CM) a diagnosis appears in the submission to FDA.

E.   Organ Weights

1)  Organ weights were entered on the gross pathology sheets at the time of autopsy.  We compared all of the individual organ weights on appendix table 5 in the submission to FDA (Vol. 1, pgs. 222-226) with the original data on the gross pathology sheets.  A total of eleven (11) errors were noted in transcribing the raw data from the pathology sheets to the tables in the submission to FDA.

F.  Survival

1.  We were  unable to determine the exact method used by Searle in constructing the survival table in the submission to FDA. We constructed a survival table using the body/feeder weight Teletype sheets.  A Life Table Analysis was constructed from our survival table by Dennis Wilson, FDA Department of Mathematics.  The female control population differed from the high level population (p 0.05) and the mail control population differed from the mid and high level population (p 0.05).  In all cases the differences are due to higher mortality in the controls.

G.  Clinical Laboratory Procedures

1. Laboratory records of one sort or another for all assays reported in the submission were obtained.  In some cases data sheets were noted with results of assays carried out at treatment days not indicated in the protocol or protocol amendment.  For example, serum cholesterol determinations were done at days 796 and 798 (terminal bleeding) but not included in the submission to FDA. Because 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, the omitted data is of some importance.

2.  No data was seen for two assays (serum insulin and serum ornithine carbamyl transferase) which were called for in an amendmend to the protocol.

3.  Original data was not always available for authentication of results or examination of procedures for conversion of raw data into the calculated values submitted to FDA.

4.  Data pages for clinical chemistry and urinalysis were initialled by a technician who transcribed data but apparently was not directly involved in the assays indicated.  He stated in an interview that Dr. Rao told him to initial the data sheets.

5.  The methodology as referenced in the submission to FDA is incomplete and not always an accurate reflection of the methodology actually used in the study.  The fact that more than one method was sometimes used for a particular assay during different times of the study was not indicated in the submission to FDA.

6.  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.

7.  A  total of 49 disparities were noted  between statistical computations reported by Searle in the submission and those calculated by FDA.  The disparities are constituted by the values for 6 means, 23 standard errors, and 20 significant differences (as measured by T tests).