EXPERIMENTAL

AND

MOLECULAR

PATHOLOGY

53, 34-51 (1990)

Effects of 5-Azacytidine in Syrian Golden Hamsters: Toxicity, Tumorigenicity, and Differential Modulation of Bronchial Carcinogenesis WILLIAM G. HAMMOND,~ ALON YELLIN,~*~ AHARON GABRIEL,~'~ RAO R. PALADUGU,~,~ NORIOAZUMI,~'~ L. ROBERT HILL,~ AND JOHN R. BENFIELD'-* Departments of ‘Thoracic Surgery and ‘Biostatistics, and the Division of 4Anatomic Pathology, City of Hope National Medical Center, Duarte, California 91010 Received November 14, 1989; and in revised form May 7, 1990 5Azacytidine (AZC) was studied in a lung cancer mode1 in outbred and syngeneic (F,D) hamsters wherein benzo[a]pyrene (BP) from sustained release implants (SRI) induces preneoplastic mucosal changes which progress to bronchogenic cancer. In pilot studies to evaluate AZC toxicity, a dose schedule of 5 mg/kg biweekly was found suitable and was then used for long-term administration in all subsequent studies. Three groups of outbred hamsters were studied: BP SRI alone (n = 60), BP SRI + AZC (n = 60), and AZC alone (n = 54). AZC treatment was begun 3-5 days after SRI placement. Sixty-one days after the start of the experiment, seven or eight hamsters were sacrificed from each group. Later sacrifices were at 3-week intervals in groups receiving BP SRI and at 6-week intervals in the AZC only group. Four groups of F,D syngeneic hamsters were studied: BP SRI alone (n = SO); BP + AZC starting 3-5 days after SRI placement and continuing until death (n = 52); BP + AZC from 3 to 5 days until 75 days after SRI placement (n = 49); BP + AZC starting 80 days after SRI placement and continuing until death (n = 52). Hamsters (n = 9-14) from each group were sacrificed at 120, 150, 180, and 220 days after SRI implantation. AZC alone was not carcinogenic under these conditions. Both outbred and F,D hamsters treated with early or continuous AZC had slower rates of neoplastic change from BP SRI than did animals receiving BP SRIs alone or BP + late AZC. The incidences of epidermoid cancer were the same for all regimens, but the tumors in those receiving AZC early in carcinogenesis were smaller than in those receiving late or no AZC. The incidences of nonepidermoid cancer were lower in those receiving AZC during early carcinogenesis, and larger tumors were noted in the absence of AZC. Thus, within the study period in this unique hamster lung cancer model, AZC given early in carcinogenesis inhibited only the later (promotional) phase of BP epidermoid carcinogenesis, but inhibited all phases of nonsquamous cancer development induced by BP. This differential modulation of bronchial carcinogenesis, which occurs from AZC given during preneoplastic stages, may prove useful for delineating molecular mechanisms underlying specific phenotypic types of bronchogenic cancers . 0 1990 Academic Press. Inc.

’ Present address: Department of Thoracic Surgery, Sheba Medical Center, Tel Hashomer, 3 Present address: Department of Surgery, Ben-Gurion University, Omer 8X%5, Israel. ’ Present address: Department of Pathology, Greater El Monte Community Hospital, El CA. 6 Present address: Department of Pathology, Georgetown University School of Medicine, ington, DC. ’ To whom reprint requests should be addressed at University of California, Davis, Medical 4301 X Street, Sacramento, CA 95817.

34 0014-4800&O $3.00 Copyright 0 1990 by Academic Press. Inc. AU rights of reproduction in any form reserved.

Israel. Monte, WashCenter,

S-AZACYTIDINE

AND

BRONCHIAL

CARCINOGENESIS

35

INTRODUCTION In mammals, a variety of experiments suggest that DNA methylation is a significant factor in gene control (Riggs and Jones, 1983). Alterations in methylation state are correlated with variation in gene expression (Michalowsky and Jones, 1989; Andrews ef al., 1982). Cancers in vivo and in established transformed lines often have abnormal levels of methylated cytosine; the usual finding was hypomethylation, but hypermethylation has also been noted (Smith et al., 1982; Lapeyre and Becker, 1979; Carr et al., 1984; Michalowsky and Jones, 1989), and chemically induced experimental cancers have exhibited abnormal methylation states (Riggs and Jones, 1983; Michalowsky and Jones, 1989). The cytotoxic cytidine analog, 5-azacytidine (AZC), is a potent inhibitor of DNA methylation. Therefore, in a multistage model of carcinogenesis, AZC might be expected to act as a potentiator. Evidence in support of this expectation was reported from in vivo rat and mouse studies (NC1 Technical Report, 1978; Car-r et al., 1984, 1988; Denda et al., 1985; Stoner et al., 1983; Vesely and Cihak, 1973; Cavaliere et al., 1987), including experiments in liver carcinogenesis; the findings were interpreted as suggesting that AZC may also be a complete carcinogen as well as a potentiator. Because most of the results interpreted as showing that AZC alone is carcinogenic were obtained from animals wherein the indicator tumors (murine lymphoreticular tumors, leukemias, and pulmonary adenomas; rat interstitial cell tumors) occur spontaneously with substantial incidences, it is possible that these reported effects of AZC in carcinogenesis are from acceleration of processes already under way, rather than from initiation of cancer de novo. Since AZC continues to be used in humans as a remission induction agent for treatment of leukemia, additional data concerning its actions in carcinogenesis are of more than just mechanistic interest. We studied the systemic and bronchial tumorigenicity of AZC in hamsters, selected because (1) they exhibit a low incidence of spontaneous lung cancers (Altman and Katz, 1979; Chesterman, 1972; Homburger, 1983), (2) they are highly susceptible to bronchial chemical carcinogenesis (Shors et al., 1980; Schreiber et al., 1974), and (3) repetitive administration of AZC has not been evaluated in this species. In our hamster model, cancers are regularly induced at the site of endobronchial sustained release implants (SRI) which contain benzo(u)pyrene (BP) (Shors et al., 1980; Bentield et al., 1984). The BP-induced epithelial alterations are assessed by complete histopathological examination; the rates of the sequential progression of carcinogenesis (SPC) are assessed by logistic regression analysis of the histopathology results. Previous studies have shown that this model may be used as a standard against which to compare variations in the overall carcinogenesis response to stimuli that either accelerate (Benfield et al., 1984) or delay (Hammond et al., 1987) the rate of occurrence of the steps in the neoplastic progression. The majority of our work with this model had employed outbred (genetically nonuniform) hamsters. Hence, after identifying an AZC dose suitable for longterm use, we studied continuous AZC administration to outbred hamsters to examine the hypothesis that AZC would potentiate BP carcinogenesis. We then studied genetically uniform (syngeneic) hamsters to be sure that the AZC effects seen would not be related to susceptibility variation among outbred animals. We sought also to determine when during carcinogenesis the AZC effect occurred and

36

HAMMOND

ET AL.

to evaluate whether or not the chemotherapeutic action of AZC could explain our observations. The results of both experiments are presented below. MATERIALS

AND METHODS

The outbred Syrian golden hamsters used were females from Simonson Laboratories (Gilroy, CA). The syngeneic Syrian golden hamsters were first generation male and female BIO F,D Alexander hybrids from Bio Breeders, Inc., Fitchburg, MA (Homburger et al., 1983; Bernfeld et al., 1986). All animals were 10-16 weeks old at the beginning of the experiment. In our use of this model, we have noted no sex difference in response to the BP SRI, and no age difference effect over the 10-16 week range used. Hamsters were maintained according to NIH guidelines for the care of laboratory animals and were fed standard laboratory chow and water ad libitum. The methods of preparation and placement and the BP release characteristics of SRI have been previously reported in detail (Shors et al., 1980). Finely powdered BP (Sigma Chemical Co., St. Louis, MO) was thoroughly mixed into liquid silicone polymer at 10% by weight and the solidification catalyst (0.5% stannous octoate) was then added. While still viscous, the mixture was drawn into 1.5 mm-inside-diameter glass tubing by a vacuum pump. After the polymer solidified, the glass was broken away and the resultant carcinogen-containing silicone rod was cut into 3.5mm segments. A 4-O gauge stainless steel suture wire was passed axially through the cylinder and the protruding end was bent to form a hook. Impaled upon a length of fine steel piano wire, the implant was inserted into the right bronchus intermedius through a tracheostomy performed aseptically under pentobarbital general anesthesia. After withdrawing the inserting wire, the tracheal and skin incisions were closed individually with fine sutures. The carcinogen release rate from SRI so prepared and inserted is a first-order exponential with a half-time of 40 days. AZC (Sigma Chemical Co., St. Louis, MO) was freshly prepared just prior to injection by dissolving the powder in sterile 0.9% NaCl solution to a concentration of 1.3 mg/ml; this solution was injected ip with a tuberculin syringe and a 27-gauge needle. Hamsters receiving AZC were weighed (unanesthetized) twice a week just before AZC injection; each individual AZC dose was based upon these weights. Hamsters not receiving AZC were also weighed twice a week. Animals were sacrificed by exsanguination under pentobarbital anesthesia. Blood studies (hemoglobin, hematocrit, total white blood count, differential counts, and platelet count) were done by standard hospital methods. Blood urea nitrogen, creatinine, serum glutamic oxalacetic transaminase (AST), serum glutamic pyruvic transaminase (ALT), alkaline phosphatase, total protein, albumin, cholesterol, calcium, and glucose were done by the SMA 1260 method in a veterinary reference laboratory. Experimental

Protocols

AZC toxicity: Study I. Fifteen outbred hamsters received AZC 10 mg/kg ip twice a week (biw); they were followed for up to 2 months. Blood samples were obtained from three hamsters. AZC toxicity: Study ZZ. Fifteen outbred hamsters were randomly allocated to one of five groups: (1) BP SRI plus AZC 5 m&g ip biw, (2) BP SRI plus AZC 10 mg/kg ip biw, (3) AZC 10 mg/kg ip biw, (4) BP SRI controls, and (5) untreated

FAZACYTIDINE

AND

BRONCHIAL

37

CARCINOGENESIS

controls. All hamsters were sacrificed after 32 days, and cardiac blood was collected from all. AZC tumorigenicity. Fifty-four randomly selected outbred hamsters received AZC 5 mg/kg biw until they were sacrificed in groups of seven or eight at 7, 15, 21, 27, 52, 61, and 70 weeks after the start of AZC injections. Modufution of BP carcinogenesis by AZ. Outbred hamsters were randomly allocated to receive either BP SRI (OB-BP ONLY, IZ = 60) or BP SRI plus AZC 5 mg/kg ip biw (OB-AZC-CONT, n = 60) beginning 3-5 days after SRI placement. Seven or eight animals from each group were sacrificed at 3-week intervals beginning 61 days after SRI placement; the last animals were sacrificed at 189 days. After BP SRI implantation, syngeneic (F,D) hamsters were randomly allocated as follows: One group received no further treatment (F,D-BP-ONLY, n = 50); another group received AZC 5 mg/kg ip biw beginning 2-5 days after SRI placement and continuing until death (F,D-AZC-CONT, n = 52); a third group received AZC 5 mg/kg ip biw early during carcinogenesis until 75 days after SRI placement but none thereafter (F,D-AZC-EARLY, n = 49); and the fourth group received AZC 5 mg/kg ip biw late during carcinogenesis, beginning 80 days after SRI placement and continuing until death (F,D-AZC-LATE, 12 = 52). Nine to 14 hamsters from each group were sacrificed at 120, 150, 180, and 220 days after SRI implantation, times selected to bracket durations of BP exposure previously found necessary for frequent cancer onset in BP-ONLY animals. These sacrifice times also allowed at least 45 days for (1) presumed subsidence of AZC effect in the FiD-AZC-EARLY group and (2) manifestation of AZC effect in the F,DAZC-LATE group. These protocols are outlined schematically in Fig. 1. This study did not include control groups of either untreated hamsters or hamsters treated with AZC and given an SRI without carcinogen. The incidence and variety of spontaneous tumors arising in untreated hamsters has been welldocumented (Altman and Katz, 1979; Chesterman, 1972; Bernfeld et al., 1986). Omission of the latter group was based on extensive past experience in which neither cancers nor preneoplastic changes were induced by SRIs that did not contain carcinogens (Shors et al., 1980; Benfield et al., 1984). Hamster Group

Treatment

OB-BP-ONLY

Sacrifice

Times

No AZC given

OB-AX-CONT

+IAZC Days-

< 61

”

0 3-5

FIO-BP-ONLY

5mg/kg i,p,biu

- ----

- __-________-______

82

103

I 124

_-

145

168

189

No AZC given

-IAZC

FlD-AX-EARLY

5mglkg

bi+ +AZC5mg/kgi.p.biw-----------------

FlD-AX-LATE

-IAZC5mg,kgl,p,biw-----

FID-AZC-CONT Days)

3-5

‘I

7580

”

--_-__---------------120

150

180

220

FIG. 1. Scheme of experimental protocols in studies of modulation of BP bronchial carcinogenesis by AZC. All hamsters received BP SRI on Day 0. OB, outbred hamsters; FlD, F,D hamsters.

38 Pathologic

HAMMOND

ET AL.

Examination

At necropsy, the entire tracheobronchial tree was excised en bloc. When tumor was visible or palpable in the intact lobe, the lesion was bisected, its largest diameter was measured, and touch preparations for cytologic examination were made from the cut surface. After removal of the SRI, representative portions of tumor and adjacent tissues were fixed in 10% formalin; paraffin-embedded sections were stained with hematoxylin and eosin. In the absence of visible or palpable tumor, the trachea and bronchus were opened through the membranous portion, the SRI was removed, and touch preparations were made from the mucosal surface at the SRI site. The whole tracheobronchial tree was then placed in buffered 10% formalin. After fixation, the entire bronchial segment containing the SRI site was mounted with liquified agar in a cylinder so that the longitudinal bronchial axis was also the longitudinal axis of the cylinder. When the agar had hardened, the cylinder was sliced into l.O-mm-thick discs which were placed flat on a glass slide. Excess agar was removed; the tissue slices were clumped together; and the aggregate was reembedded in agar, processed, and embedded in paraffin. From blocks so prepared, 4-pm sections were cut and stained with hematoxylin and eosin. Bronchial segments from the contralateral lung were obtained and prepared in the same manner. In animals that had received only AZC without an SRI, both mainstem bronchi were prepared like the bronchi that had been exposed to SRI. When invasive cancer was not found in the initial bronchial sections examined from 245 hamsters that received BP SRIs, additional progressive sections were made. At least one section from each 50 pm of the block was examined until invasive cancer was found or until representative sections of the entire area of bronchial mucosal abnormality adjacent to the SRI had been studied. This method of histopathological study reduces sampling error to negligible levels and provides a definitive diagnosis for each hamster so studied. For subsequent data analysis, the most severe change noted among all sections examined was taken as the stage of the SPC achieved in that hamster. Although carcinogenesis in respiratory epithelium is a progressively continuous process in hamsters (Schreiber et al., 1974; Benfield et al., 1984), dogs (Cohen et al., 1978), and humans (Saccamanno et al., 1974), 11 histopathologic diagnoses (shown as the ordinate in Figs. 2 and 3) within four broad categories were utilized to identify stages of the neoplastic progression for facilitating comparisons among the treatment regimens. The broad categories included (1) normal or reactive changes, including early nonspecific alterations such as inflammation and basal or columnar hyperplasia; (2) squamous metaplasia, ranging in severity over five steps from focal regular squamous metaplasia to severe atypia; (3) cancer detectable only by microscopic examination, i.e., carcinoma in situ and microinvasive carcinoma; and (4) grossly apparent and histologically confirmed invasive cancer of three size ranges. The histologic criteria used for diagnostic subdivisions within the broad categories were those previously described (Saccamanno et al., 1974). Diffuse squamous metaplasia was defined as characteristic mucosal change that either involved at least ‘/4 of the bronchial circumference or extended at least 400 p,m along the bronchial axis. The categories of grossly apparent cancer were minimal, w 8OYZ 2

0

-

8

-

0

. .

I 61

2B

0

0

8:

z

: :

.

09

:

.

.

0

I 145

I 168

I 169

0

IN SITU -

I 82

I 103

DAYS

39

CARCINOGENESIS

I 124

SRI IMPLANTATION

AFTER

FIG. 2. Shown is the most severe histopathologic diagnosis noted in each outbred indicated time of sacrifice. Each point represents an individual hamster.

hamster at the

mm in diameter; medium, 3-10 mm in diameter; and extensive, >lO mm in diameter. Histopathologic diagnoses were made independently by two observers on slides identified only by accession number. Initial interobserver diagnostic agreement was ~96%; nonconcordance was settled by simultaneous mutual review, BENZPYRENE

13 P ONLY, NO AZC

LATE

z

EXTENSIVE

-

82

g(3

MEDIUM MINIMAL

-- 8o 8o 0[ 81

”

u

@ u I‘

INVASIVE

-

-22

IN SITU

-08

8 8

23

0 8

o o [ 0

IMPLANT

*-

PLUS

CONTINUOUS .

=::

-i

$2:

.

.

l

.

;

.

8

I*

180

120 220

180 150

1

DAYS AFTER

ta f

180 150

SRI

t

A

.

120 220

’

f

: fA

8

.

150

.

8

I.

l

A :

:a** :

120

AZC

EARLY

120 220

180 150

220

PLACEMENT

3. Shown is the most severe histopathologic diagnosis noted in each F, D syngeneic hamster at the indicated sacrifice times. Each point represents an individual hamster. (0) F,D-BP-ONLY; (X) FIG.

FID-AZC-LATE;

(0)

F,D-AZC-CONT;

(A) F,D-AZC-EARLY.

40

HAMMOND

ET AL.

still without knowledge of from which hamsters the specimen had been obtained. Residual inaccuracies in diagnosis became part of the random variation which is accounted for in statistical analysis. At this level of diagnostic concurrence, this component of random variation will be quite slight. Additionally, since any undetected diagnostic bias equally affected all hamsters, the differences noted between groups are considered to be independent of residual histopathologic diagnostic error. During complete autopsies, samples were taken from the lung periphery, thyroid, small bowel, liver, spleen, pancreas, kidney, adrenal, ovary, uterus, femur, and any grossly abnormal area. Sections from all samples were stained with hematoxylin and eosin. Analysis Hamsters that died at times other than at scheduled sacrifice dates were unsuitable for autopsy and were therefore excluded from analysis. Hamsters in which (1) the BP SRI was not found in the bronchus at autopsy or (2) no main bronchus was seen in any of the histopathological sections were excluded from analysis of bronchial carcinogenesis; however, these hamsters were included in evaluation of the systemic effects of AX. Statistical analysis for the studies of AZC toxicity was done by a two-way analysis of variance and the t test (Dunn and Clark, 1974). Differences between incidences of tumor sizes and histologic patterns were compared utilizing the Fisher exact test (Afifi and Azen, 1979). Since the various histopathologic changes seen in the respiratory epithelium from BP exposure occur as an ordered progression with a temporal component rather than as random events, the method of logistic regression (Cox et al., 1970) was chosen for analysis of the effects of AZC upon BP carcinogenesis. As previously utilized (Hammond et al., 1987), this technique permits (1) comparison between distributions of histopathologic diagnosis in different animal groups, (2) evaluation of these distributions over time, and (3) exploration of the degree of interaction between these factors. The method also permits comparison of different rates of progression over time through the ordered series of changes which occur as a consequence of different carcinogenic regimens and provides P values for such comparisons to assess statistical significance of differences observed between regimens. To utilize logistic regression analysis, it is necessary to apply numerical rating scales to the 11 histopathologic stages of the neoplastic progression. Three different scales were utilized to assess the effects of scale selection upon the analyses. All three scales resulted in similar results; therefore the data reported below are from use of the simplest scale which expressed the changes as a series of consecutive numbers ranging from 0 to 10. The 11 histopathologic diagnoses described above are shown as the ordinate in Figs. 2 and 3. RESULTS Toxicity Studies The first dose of AZC 10 mg/kg biw was selected because it had been tolerated for prolonged periods in rats (Cat-r et al., 1984). However, 12 hamsters died l-2 months after initiation of AZC injections and 3 were sacrificed for obvious poor health at 3545 days. Initially, all treated hamsters stopped gaining weight after

SAZACYTIDINE

AND

BRONCHIAL

CARCINOGENESIS

41

several injections; later there was scoliosis, hair loss, and severe weight loss. Blood tests in 2 animals revealed severe leukopenia (1200 and 1400/mm3), granulocytopenia (350 and 400/mm3), and hypoalbuminemia (2.5 g/dl). The second 32-day toxicity study was performed to see if the chronic administration of AZC 5 mg/kg ip biw would be tolerated in the presence of a BP SRI. Controls included animals getting no AZC and animals getting the toxic dose of 10 mg/kg. Since cessation of weight gain had been the earliest sign of toxicity noted previously, serial weights were assessed by percentage change in weight. To determine the effects of BP and AZC on weight change, we compared hamster groups by the AZC dose given: 0,5, or 10 mg/kg biw. After 32 days, hamsters that had received no AZC (groups 4 and 5) gained 21.7 + 4.5% of their initial body weight and those that had received 5 mg/kg (group 1) gained 15.1 + 3.0% (difference not significant; P > 0.05). However, those that had received 10 mg/kg AZC (groups 2 and 3) lost 9.7 + 14.5% (difference from 0 or 5 mg/kg significant; P < 0.05). As compared to untreated controls (which usually increase body weight by 2&25% from l&12 weeks to 20-22 weeks of age), the presence of BP SRI, with or without AZC 5 mg/kg, was not associated with significant weight change (P > 0.5). Mean values for hemoglobin, hematocrit, platelets, albumin, AST, ALT, cholesterol, bilirubin, calcium, phosphorus, glucose and creatinine were similar and within normal limits among all treatment groups. Mean values for alkaline phosphatase (P < 0.01) and blood urea nitrogen (P < 0.05) were decreased in hamsters treated with AZC 10 mg/kg, although all values were within normal limits. Hamsters that received 0, 5, or 10 mg/kg of AZC differed significantly (P < 0.01) with regard to the total leukocyte counts; 3383 + 804/mm3, 2400 2 361/mm3, and 1667 & 58/mm3, respectively. These respective differences were even more significant (P < 0.001) when the number of polymorphonuclear leukocytes was analyzed: 1393 ? 166/mm3, 1023 ? 216/mm3 and 470 ? 212/mm3, respectively. Long-term systemic toxic effects of AZC were assessed from comparisons of serial weighings. For each sacrifice time group in each treatment regimen, the group mean percentage body weight variation in the twice weekly weighings was calculated. These values were each compared to the -t4.7% random variation (mean -+ 1.67 standard deviations) we have found from serial (twice weekly) weighings of unanesthetized, developmentally mature, and apparently healthy untreated hamsters over 4- to ‘I-month periods. Interregimen comparisons were also made between those regimens including AZC and those without AZC. The only significant weight losses noted included the 189-day sacrifice groups of regimens. OB-BP-ONLY and OB-BP-AZC and the 220-day sacrifice group of F,DAZC-LATE; in these groups the mean weight losses of -9.7%, -7.6%, and -6.8%, respectively, were due entirely to animals that had grossly apparent cancers at autopsy. In all other groups, the hamsters either (1) maintained their initial weight throughout or (2) initially gained weight and later maintained a stable weight if developmentally immature at the start of the study. Extrapulmonary pathologic changes were uncommon. They consisted of splenic or renal congestion in two hamsters and agonal congestion of both organs in five AZC animals. Other incidental findings were one liver granuloma, one adrenal and one renal benign hyperplasia, and three cases of chronic pyelonephritis. Fifteen hamsters that received only AZC had benign liver cysts, noted only after 360 days of study when the hamsters were 14-19 months old. Because

42

HAMMOND

ET

AL.

most liver cysts occur in hamsters older than 14 months (Chesterman, consider them a consequence of age, unrelated to AZC administration. Tumorigenicity

1972), we

of AZC

Fifty-four hamsters receiving only AZC (5 mg/kg ip biw) were sacrificed over 7-70 weeks according to the previously stated schedule. None of them had any grossly apparent tumor, and specimens from the various organs studied histologically revealed no microscopic neoplasms. From the OB-AZC-CONT, F,DAZC-CONT, and F,D-AZC-LATE groups combined, there were 140 more hamsters subjected to the systemic effects of AZC for prolonged periods. Among the 194 hamsters, only one cancer was detected in the lungs without SRI and in all other organs: a malignant histiocytoma of the thigh. Thus, the incidence of cancer remote from the endobronchial SRI was 0.51%. There were 248 lungs without BP-SRI; in none of these was gross or microscopic cancer, atypia or squamous metaplasia detected. Differential

Modulation

of Carcinogenesis

by AZC

The outbred hamster experiments were performed prior to the syngeneic hamster work and used different sacrifice schedules; therefore results will first be described separately with regard to rates of the SPC. We will then present analyses in which SPC rates in both outbred and syngeneic animals are compared according to treatment regimen. Finally, we will consider the incidences and types of cancer that resulted in both varieties of hamster. In outbred hamsters, there were 11 perioperative deaths; 11 additional hamsters had SRIs inserted to restore the initial numbers. Thereafter, 5 OB-BP-ONLY hamsters and 4 OB-AZC-CONT hamsters died 14-50 days after the start of the experiment. Of the 55 and 56 hamsters sacrificed in these two groups, the SRI was not found in the bronchus in 3 OB-BP-ONLY animals and 1 OB-AZC-CONT hamster. Four OB-BP-ONLY hamsters and 2 OB-AZC-CONT hamsters were excluded from analysis of carcinogenesis because no major bronchus was identified in the histologic sections. Available for final analysis of bronchial carcinogenesis were 48 OB-BP-ONLY and 53 OB-AZC-CONT animals. Three additional OB-AZC-CONT hamsters were available for analysis of systemic effects of AZC (total = 56). The entire spectrum of the bronchial neoplastic progression at the SRI site in outbred hamsters is shown in Fig. 2. Logistic regression analysis revealed that, within each of the two treatment groups (OB-BP-ONLY and OB-AZC-CONT), the increasing severity of epithelial change seen with increasing duration of carcinogen exposure was continuously progressive and statistically significant (P < 0.05). The histopathological changes seen in OB-BP-ONLY animals were proportionately more severe by logistic regression analysis than those noted in OBAZC-CONT hamsters (P < 0.05) at any single sacrifice time. This difference between OB-BP-ONLY and OB-AZC-CONT was consistent all through the study, since logistic regression analysis showed that the interaction between passage of time and degree of change was not significantly different between the two treatment groups (P > 0.05). Had this difference developed during the early periods of observation, an interaction difference would have been expected; as interaction difference was not seen, it presumably developed earlier. Thus, the

5-AZACYTIDINE

AND BRONCHIAL

43

CARCINOGENESIS

differential effect of AZC was already present at the time of first sacrifice after 60 days of SRI placement and persisted throughout the study. In syngeneic hamsters, there were 12 perioperative deaths due equally to SRI dislodgement with asphyxia and to unknown causes. Fifteen hamsters died at times other than at scheduled sacrifice dates and their carcasses were unsuitable for autopsy. In both instances, deaths were about equally distributed among the four experimental groups. One F,D-AZC-CONT animal and 3 F,D-AZC-LATE hamsters were excluded from analysis of bronchial carcinogenesis because no main bronchus was seen in the sections, and 1 F,D-AZC-LATE hamster was excluded because no SRI was found at autopsy. Available for analysis of the effects of AZC upon BP pulmonary carcinogenesis in syngeneic hamsters were 44 FiD-BP-ONLY hamsters, 40 F,D-AZC-CONT hamsters, 43 F,D-AZC-EARLY hamsters, and 44 F,D-AZC-LATE hamsters. The histologic abnormalities noted at the SRI site are shown in Fig. 3. By logistic regression analysis, the SPC rate in syngeneic hamsters given AZC starting after 80 days of SRI exposure (F,D-AZC-LATE) was like that in hamsters that received no AZC (F,D-BP-ONLY). In syngeneic hamsters receiving AZC for only the first 75 days after SRI implantation (FiD-AZC-EARLY), the SPC rate was similar to that seen in hamsters that received AZC throughout (F,D-AZC-CONT). The SPC rates noted in hamsters that received no AZC or that received AZC only after 80 days of SRI exposure (FiD-BP-ONLY and FiD-AZC-LATE) were both more rapid than the SPC rates noted in hamsters that received AZC during the early stages of BP carcinogenesis (F,D-AZC-EARLY and F,D-AZC-CONT). Comparisons between outbred and syngeneic hamsters receiving the same regimens were made by logistic regression analysis. The SPC rate in BP SRI alone groups was not different between outbred (OB-BP-ONLY) and F,D (syngeneic) hamsters (F,D-BP-ONLY). In both varieties of hamsters, the effects of continuous AZC were the same (OB-AZC-CONT and F,D-AZC-CONT); the SPC occurred more slowly in these than in hamsters that did not receive AZC (OBBP-ONLY and F,D-BP-ONLY). As analyzed by logistic regression, comparisons of SPC rates from the carcinogenesis modulation studies in both varieties of hamster are summarized in Table I. Although there were clear differences in the SPC rates among the various regimens, the final incidences of epidermoid cancers were not different (Table II). The distribution of epidermoid cancer sizes was the same between the BP-ONLY

Summary of Comparisons between

TABLE I SPC Rates in Carcinogenesis Modulation

Studies

Groups compared”

Significance

OB-BP-ONLY (48) = F,D-BP-ONLY (44) OB-BP-ONLY (48) > OB-AZC-CONT (53) F,D-BP-ONLY (44) > F,D-AZC-CONT (40) F,D-BP-ONLY (44) = F,D-AZC-LATE (44) F,D-AZC-EARLY (43) = F,D-AZC-CONT (40) F,D-BP-ONLY (44) > F,D-AZC-EARLY (43) F,DBP-ONLY (44) > F,D-AZC-CONT (40) F,D-AZC-LATE (44) > F,D-AZC-EARLY (43) F.D-AZC-LATE (44) > F,D-AZC-CONT (40)

N.S. P -c 0.05 P < 0.001 N.S. N.S. P < 0.001 P < 0.001 P -c 0.05 P < 0.01

D Parenthetic numbers = number of hamsters in group.

44

HAMMOND

ET AL.

TABLE II Incidences of Epidermoid Cancer among Four Regimens BP + AZC given BP” only No. evaluable hamsters All epidermoid cancers (%) Microscopic”

Late

Early

Continuousb

15 25 (33.3)

44

42

16

&)

&)

(i.0)

(ii. 1) 13** (31.0)

(%I) 24** (31.6)

epidermoid cancers (%)

Grossly apparenr’ epidermoid cancers (%) (:L)

(K2)

L2OB-BP-ONLY and F,D-BP-ONLY combined; no difference between them for these parameters. b OB-AZC-CONT and F,D-AZC-CONT combined; no difference between them for these parameters. c Includes carcinoma in situ plus microinvasive cancers. d Includes all sizes of grossly apparent cancers. * P < 0.05, **P < 0.01, as compared to BP-ONLY; others not signikant.

and the AZC-LATE groups and between the AZC-EARLY and the AZC-CONT groups, but the latter two groups had significantly smaller (earlier) epidermoid cancers. In contrast, the incidences of nonepidermoid cancer (Table III) were significantly lower in the AZC-EARLY and AZC-CONT groups as compared to the BP-ONLY and AZC-LATE groups. The distribution of nonepidermoid cancer sizes was also different from that seen for epidermoid neoplasms; the incidences of microscopic cancer were the same for BP-ONLY as for the AZC-treated groups, while the incidence of grossly apparent cancer was significantly higher in the BP-ONLY group. There were no significant interregimen differences in the proportion of the three nonepidermoid cancer varieties-adenocarcinoma, adenosquamous carcinoma, and spindle cell carcinoma.

Incidences of Nonepidermoid

TABLE III Cell Cancers among Four Regimens BP + AZC given BP” only

No. evaluable hamsters All nonepidermoid” cancers (%) Microscopic nonepidermoid

cancers (%)

Grossly apparentd nonepidermoid cancers (%)

Late

Early

Continuousb 76 4** (5.3) (k3)

75

44

(:l.O, 4 (5.3)

(E3)

42 7** (l&7)$

$3)

$9)

(&)

(IL),

2** (4.8)

3** (3.9)

a OB-BP-ONLY and F,D-BP-ONLY combined; no difference between them for these parameters. b OB-AZC-CONT and FiD-AZC-CONT combined; no difference between them for these parameters. ’ Adenocarcinomata, adenosquamous, and spindle-cell carcinomata. d Includes all sizes of grossly apparent cancers. * P < 0.05, **P < 0.01, as compared to BP-ONLY; others not significant. t Horizontally adjacent values significantly different with P < 0.05;SP< 0.01.

S-AZACYTIDINE

AND

BRONCHIAL

CARCINOGENESIS

45

DISCUSSION AZC

Toxicity

To our knowledge, long-term systemic administration of AZC has been studied previously only in Rhesus monkeys, baboons, mice, rats, and humans. The maximum tolerated dose varied widely. Baboons succumbed to infections after 4-6 weeks of 4-6 mg/kg iv daily (Heller and DeSimone, 1984); Rhesus monkeys tolerated 14 daily iv doses of up to 1.1 mg/kg (Palm et al., 1972a, b). Mice tolerated iv daily doses of 10 mg/kg for 5-day courses (Reno et al., 1983). Rats tolerated biweekly doses of 10 mg/kg ip for as long as 9 months or triweekly doses of 7.5 mg/kg ip for 12 months (Cat-r et al., 1984, 1988). Doses of 5 to 17 mg/kg were tolerated in humans in different regimens; the tolerance limits were inversely proportional to frequency of treatment (Von Hoff et al., 1976). In hamsters, AZC has only been used topically or as a single jugular vein injection (Palm et al., 1972a, b). Our toxicity studies indicated that AZC at 10 mg/kg biw would be regularly lethal to hamsters when administered for periods much longer than 1 month. We observed significant bone marrow depression and some hepatic and renal damage with this dose, findings similar to those reported by others in laboratory animals (Palm et al., 1972a, b; Reno et al., 1983) and humans (Von Hoff et al., 1976). After we determined that AZC at a dose of 5 mg/kg ip twice a week was not associated with significant weight change or biochemical abnormality during 5 weeks of treatment, this dose was then given for the longer term studies. For periods as long as 16 months, it was tolerated without weight loss or persistent hematological, renal, or hepatic toxicity of significance. However, the observed mild decrease in leukocyte counts and the differences in types of tumors seen both indicate that this dose of AZC was biologically effective. Tumorigenicity

of AZC

Although the effects of AZC upon neoplasia have been studied extensively in studies evaluating the influence of AZC upon carcinogenesis in vivo are sparse and limited to observations of the end product--cancer. The Golden Syrian hamster is susceptible to chemical carcinogenesis in the lung as well as elsewhere (Shors et al., 1980; Schreiber et al., 1974; Benfield et al., 1984; Hammond et al., 1987; Bernfeld et al., 1986), but neither tumorigenicity nor promoter-like actions of AZC have previously been studied in this species. Under the conditions of our studies, it seems clear that AZC is not a complete carcinogen at any site in hamsters. We examined all the viscera, especially those organs in which spontaneous neoplasms in hamsters are found (Chesterman, 1972; Hornburger et al., 1983; Bernfeld ef al., 1986); only a single malignant fibrous histiocytoma was found. Cancers were found only in the lungs that contained an implant of carcinogen; not in the lungs of 54 hamster that received only AZC, nor in the contralateral lung of 140 hamsters given systemic AZC plus locally applied ipsilateral carcinogen. Neither in our studies using the SRI method (Shors et al., 1980; Benfield et al., 1984) nor in any reports from other investigators using intratracheal instillation of larger carcinogen doses were there increases in tumors outside the airways, indicating that systemic absorption of intraairwayadministered carcinogen is negligible at the doses used. In outbred hamsters that received only AZC, treatment with this agent began at vitro,

46

HAMMOND

ET AL.

70-112 days of age. The last three groups were sacrificed at 4351175,500-540, and 560-600 days of life, respectively. Since the median survival for female outbred hamsters has been reported as varying from 345 to 481 days, it seems clear that these last three groups of hamsters received AZC throughout their adult life and were followed well into the extremes of hamster longevity. Nevertheless, no cancers were observed from AZC in these hamster lifetime studies. To our knowledge, there are only six reports in which repetitive administration of AZC was evaluated for carcinogenicity. At three dose levels, Stoner et al. (1973) treated strain A mice with three injections of AZC/week for 8 weeks. Twenty-four weeks after starting AZC injections, there were no significant differences in the proportions of tumor-bearing mice between the control groups and groups receiving any dose of AZC. In their initial study, Cat-r et al. (1984) treated male Fischer 344 rats with biweekly injections of AZC at two dose levels (2.5 and 10.0 mg/kg) for 9 months and sacrificed the animals 9 months after ending the injections. In their subsequent study, Car-r et al. (1988) administered AZC at three dose levels (2.5,0.25, and 0.025 mg/kg) to the same rat strain thrice weekly for 12 months and then sacrificed the animals. Skin tumors at the injection site in the two highest AZC dose groups (7.5 and 2.0 mg/kg/wk) and four renal tumors noted in the group receiving 7.5 mg/kg/wk were the only tumors that (1) were significantly different (P < 0.05) from controls or (2) did not occur spontaneously in untreated rats with an incidence of 2% or more (Goodman et al., 1979). In Sprague-Dawley rats treated with AZC at doses of 2.6 and 5.2 mg/kg given thrice weekly, no carcinogenic effects were seen; in female B6C3Fr mice given AZC at 2.2 or 4.4 mg/kg thrice weekly, an equivocal increase in lymphomas was noted with the higher dose (NC1 Technical Report, 1978). Female AKR mice, given six doses of 1.5 mg/kg AZC twice weekly and then six doses of 0.8 mg/kg AZC spread over the next 4 weeks, developed the typical AKR leukemia earlier in life and at a higher incidence than did untreated female AKR mice (Vesely and Cihak, 1973). Eightweek-old BALB/c mice of both sexes were treated with 2.0 mg/kg AZC weekly for 50 weeks. Lymphoreticular system neoplasms were noted in both treated and control mice of both sexes, with an increased incidence in treated animals. Eulmonary adenomas were similarly present in both sexes, but the incidence in treated mice was higher only in males. A low but significant incidence of mammary and skin neoplasms was noted in treated females (Cavaliere et al., 1987). Since the majority of the results interpreted as showing that AZC is carcinogenic were obtained in animals that have substantial incidences of spontaneous occurrence of the indicator tumors (mouse lymphoreticular tumors, leukemias, and pulmonary adenomas; rat interstitial cell tumors), the observations described above suggest that these reported effects of AZC may be from acceleration of a process already under way, rather than from initiation of carcinogenesis de nova. Carr et al. (1984, 1988) found no evidence for de lzovo tumor initiation in their rat hepatic carcinogenesis model. Moreover, Denda and associates (1985) administered AZC at 10 mg/kg to the same rat strain used by Carr et al. and observed no evidence of liver cancer initiation as manifested by foci of y-glutamyltransferase-positive hepatocytes. Our findings also fail to implicate AZC alone as an initiator of carcinogenesis. Although permanent cell lines maintained in vitro have eventually shown neoplastic transformation when serially passaged following exposure to AZC (Hsiao et al., 1985), such cell lines have undergone many alterations of biological char-

SAZACYTIDINE

AND

BRONCHIAL

CARCINOGENESIS

47

acteristics during successful adaptation to the in vitro environment prior to AZC exposure. Under such circumstances, it is not clear whether AZC truly initiates transformation de nova or whether it only potentiates a process already initiated during adaptation. Considering the reported in vitro and in vivo results described above together with our findings, we believe that the case for AZC as a de nova initiator of carcinogenesis remains unproven, especially as regards any extrapolation to the human situation. In this regard, we agree with the conclusions of the IARC as expressed in 1983 (International Agency for Research on Cancer, 1983). Differential

Modulation

of BP Carcinogenesis

by AZC

The effects of AZC in modulating BP carcinogenesis were more clearly observable as a direct result of the use of logistic regression. Typically, this type of experiment is analyzed using contingency table analysis. That method, however, does not allow exploitation of the progressive nature of the histopathological changes or the temporal component. As a result, patterns such as those found here are more difftcult to detect in the data; ie., the consistent progression in atypia over time and the consistently greater atypia in the BP-ONLY groups. There is both a theoretical and an experimental basis for expecting potentiation of tumorigenicity by methylation inhibition. AZC has caused transformation in many in vitro experiments (Riggs and Jones, 1983). In vivo studies are sparse, but support for the hypothesis of tumor potentiation by AZC was specifically addressed and demonstrated in two studies utilizing a hepatic cancer model in the Fischer rat (Cat-r et al., 1984; Denda et al., 1985). However, in hamsters that received systemic AZC 5 mg/kg biw in addition to being exposed to a focal carcinogen, we did not find enhancement of the focal carcinogen-induced neoplastic progression. Instead, as compared to controls, there was inhibition of the growth of epidermoid cancers in both groups that received AZC early during carcinogenesis and no effect on epidermoid cancers in the group that received AZC late in carcinogenesis (Table II). Since the final incidences of epidermoid cancers were similar for all regimens, it appears that the effect of AZC on epidermoid carcinogenesis is not upon the early (initiation) phase but on the late (promotional) phase of the actions of the complete (self-promoting) carcinogen, BP. Further, since only those groups that received AZC early in carcinogenesis manifested this effect, the action of AZC appears to occur in the earliest part of the late (promotional) phase, the time when already initiated cells must proliferate actively to express the neoplastic potential inherent in the initiated state. These effects of AZC upon the neoplastic progression and upon the incidence and appearance of squamous cell cancers were unanticipated, but the results may be related to other known actions of AZC. For instance, AZC has chemotherapeutic (cytotoxic) properties that are useful for the treatment of acute leukemias in humans. However, the administration of AZC beginning 80 days after BP SRI insertion had no significant effect upon the SPC rate nor upon the incidence of epidermoid cancers induced by the BP SRI. Carcinomas in situ and microinvasive carcinomas (those expected to be most susceptible to the chemotherapeutic action of AZC) do not appear until after 100 days of BP exposure in control animals. That treatment with AZC during this period had no effect indicates that the chemotherapeutic property of AZC was not responsible for the differences noted be-

48

HAMMOND

ET

AL.

tween those groups given late AZC and which received AZC throughout the experiment. For the development of bronchial cancer other than epidermoid (adenosquamous and spindle cell carcinomas, adenocarcinomas), the effects of AZC on BPinduced carcinogenesis were more extensive than those noted for epidermoid cancers. BP initiation of nonepidermoid cancers was decreased by early AZC, as shown by the significantly reduced incidences of those cancers in the groups that received AZC during the early period. However, the effect of AZC upon nonepidermoid cancer development was different from that seen for epidermoid cancers; continuous administration of AZC reduced the final incidence more than did either early or late AZC alone, while the incidence of grossly apparent nonepidermoid cancers from late AZC was intermediate between those of early and continuous AZC and that of BP alone. Hence, for nonepidermoid cancer, AZC not only inhibits initiation, it also inhibits both the late (promotional) phase and the subsequent growth of an established tumor. The specific mechanisms for the inhibitory effects of AZC during carcinogenesis are unclear. One of the possibilities considered was that AZC modified carcinogenesis through an effect upon nutrition, since nausea and vomiting have been the commonest dose limiting toxicity in AZC use in humans (Von Hoff ef al., 1976). However, we found no evidence for significant inhibition of food intake in AZC-treated hamsters; on the contrary, animals receiving AZC, either alone or in combination with BP SRI, gained weight in a normal developmental fashion or maintained adult weight. At the dosage used in this study, selective cytotoxic effects of AZC might have outweighed whatever carcinogen potentiating effect that may have coexisted. Evidence to support this possibility includes the observations that (1) in studies of the diaplacental effect of AZC on tumorigenesis in mice, the lower dose of AZC used resulted in more tumors than with higher doses (Schmahl et al., 1985), and (2) there was an inverse relationship between AZC dose and its effect upon concavalin A stimulation of rat thymic lymphocyte proliferation in vitro (Schauenstein et al., 1988). Further, random alteration of DNA methylation patterns by AZC may lead to variably selective expression of certain histologic characteristics and repression of others, suggested by the finding that AZC-induced hypomethylation of L1210 cells passaged in vivo was stable when low doses of AZC were used, but was clearly less stable when doses greater than 4 mg/kg were employed (Lu and Randermath, 1984). Finally, although AZC is known to change the expression of immunogenic cell surface markers (Frost et al., 1984; Olsson and Forchhammer, 1984), a recent report (Carlow et al., 1989) indicates that AZCinduced increases in tumor cell expression of Class I major histocompatibility antigens are not associated with increased immunogenicity of the affected tumor cells. Which, if any, of these effects of AZC may have contributed to the observed modulation of carcinogenesis remains to be determined. As noted in the Introduction, DNA hypomethylation by AZC in some instances eventuates in a predominantly carcinogenic response. Insofar as we have been able to determine, all the biological effects of AZC appear to result from its action on DNA methylation, even the cytotoxicity useful for treatment of patients with myeloid and lymphoid leukemia (Momparler et al., 1984). Thus, that the systemic administration of AZC at the dose and schedule used in this study did not cause cancer in hamsters suggests that selectively specific alterations of DNA methyl-

5-AZACYTIDINE

AND BRONCHIAL

CARCINOGENESIS

49

ation are required for carcinogenesis alone; other additional AZC-induced DNA hypomethylations may occur concomitantly, with the net effect being inhibition of carcinogenesis. Alternatively, the demonstration that BP alone may inhibit DNA methylation in a manner than interferes with the hypomethylating action of AZC in some cell lines but not in others (Wilson and Jones, 1983) provides a potential explanation for the absence of the expected increase in cancers from both agents given together, as well as for the tumor type dissociation of AZC effects. In this investigation, we identified a dose and schedule of AZC that produce definite biological effects but which are tolerated well by hamsters for long periods, even when coupled with BP SRI. In a species that has an infinitesimal incidence of spontaneous lung cancer, we have shown that this dose and schedule of AZC are not carcinogenic, thereby maintaining the previous uncertain status of AZC as a bona tide potential carcinogen, especially in humans. Finally, we have shown that, while early concomitant administration of AZC does not affect BPinduced initiation of epidermoid cancer, it does inhibit the later (promotional) phases of BP-induced epidermoid carcinogenesis. Differentially, during BPinduction of other types of bronchial cancer, AZC inhibits not only the early (initiation) phase, but also the later (promotional) phases and subsequent tumor growth. This tumor type relationship of AZC effects upon bronchial carcinogenesis suggests the possibility that the molecular mechanisms underlying the genesis of the various types of BP-induced bronchial cancer in hamsters may be tumortype specific. ACKNOWLEDGMENTS This research was supported by NIH Grants CA29372 and CA26529.

REFERENCES AFIFI, A. A., and AZEN, S. P. (1979). Statistical Analysis: A Computer-Oriented Approach. Academic Press, New York. ALTMAN, P. L., and KATZ, D. D. (Eds.) (1979). “Inbred and Genetically Defined Strain of Laboratory Animals: Part 2. Hamster, Pig, Rabbit, and Chicken,” pp. 493-495, FASEB, Bethesda, MD. ANDREWS, A. K., DZIADEK, M., and TAMAOKI, T. (1982). Expression and methylation of the mouse a-fetoprotein gene in embryonic, adult and neoplastic tissues. J. Biol. Chem. 257, 5148-5153. BENFIELD, J. R., HAMMOND, W. G., SHORS, E., PALADUGU, R. R., PAK, H. Y., and TEPLITZ, R. L. (1984). A family of epidermoid lung cancer models. Ann. Thorac. Surg. 38, 627-632. BERNFELD, P., HOMBURGER, F., ADAMS, R. A., VAN DONGEN, C. G., and SOTO, E. (1986). Baseline data in a carcinogen-susceptible first-generation hybrid strain of Syrian golden hamster: F,D Alexander. J. Natl. Cancer Inst. 77, 165-171. CARLOW, D. A., KERBEL, R. S., and ELLIOT, B. E. (1989). Failure of Class I major histocompatibility antigens to alter tumor immunogenicity of a spontaneous murine carcinoma. J. Natl. Cancer Inst. 81, 759-767. CARR, B. I., RAHBON, S., and ASMERON Y. (1988). Carcinogenicity and haemoglobin synthesis by cytidine analogues. Brit. J. Cancer 57, 395402. CARR, B. I., REILLY, J. G., and SMITH, S. S. (1984). The tumorigenicity of 5azacytidine in the male Fischer rat. Carcinogenesis 5, 1583-1590. CAVALIERE, A., BUFALARI, A., and VITALI, R. (1987). 5-Azacytidine carcinogenesis in BALBlc mice. Cancer Lett. 37, 51-58. CHESTERMAN, F. C. (1972). Background pathology in a colony of golden hamsters. In “Pathology of the Syrian Hamster” (F. Homburger, Ed.), pp. 50-68. Karger, London. COHEN, A. H., SHORS,E. C., OKITA, M., MATSUMURA, K., JENSEN, T., and BENFIELD, J. R. (1978). Morphologic aspects of experimental canine bronchial carcinogen exposure. Eur. J. Cancer 14, 401409. Cox, D. R. (1970). “The Analysis of Binary Data.” Methuen, London.

50

HAMMOND

ET AL.

DENDA, A., RAO, P.M., and RAJALAKSHMI, S. (1985). S-Azacytidine potentiates initiation induced by carcinogens in rat liver. Carcinogenesis 6, 145-146. DUNN, 0. J., and CLARK, V. A. (1974). “Applied Statistics: Analysis of Variance and Regression.” Wiley, New York. FROST, P., LITEPOLO, R. G., and DONAGHUE, T. P. (1984). Selection of strongly immunogenic “turn’‘-variants from tumors at high frequency using 5-azacyridine. J. l&p. Med. 159, 1491-1501. GOODMAN, D. G., WARD, J. M., and SQUIRE, R. A. (1979). Neoplastic and non-neoplastic lesions in aging F344 rats. Toxic01 Appl. Pharmacol. 48, 236-248. HAMMOND, W. G., GABRIEL, A., PALADUGU, R. R., AZUMI, N., HILL, L. R., and BENFIELD, J. R. (1987). Differential susceptibility to bronchial carcinogenesis in syngeneic hamsters. Cancer Res. 47, 5202-5206. HELLER, P., and DESIMONE J. (1984). 5-Azacytidine and fetal hemoglobin. Amer. J. Hematol. 17, 43w7. HOMBURGER, F., ADAMS, R. A., BERNFIELD, P., VAN DONGEN, C. G., and SOTO, E. (1983). A new first-generation hybrid Syrian hamster, BlO F,D Alexander, for in vivo carcinogenesis bioassay, as a third species or to replace the mouse. Surv. Synch. Pathol. Res. 1, 125-133. HSIAO, W. L. W., GATTONI-CELLI, S., and WEINSTEIN, I. B. (1985). Effects of 5-azacytidine on the progressive nature of cell transformation. Mol. Cell. Biol. 5, 1800-1803. International Agency for Research on Cancer: Some Antineoplastic and Immunosuppressive Agents. (1983). “IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans. Vol. 26, pp. 3746. LAPEYRE, J. N., and BECKER, F. F. (1979). 5-Methylcytosine content of nuclear DNA during chemical hepatocarcinogenesis and in carcinomas which result. Biochem. Biophys. Res. Commun. 87, 698705. Lu, I. J. W., and RANDERRNATH, K. (1984). Long term instability and molecular mechanism of 5-azacytidine-induced DNA hypomethylation in normal and neoplastic tissues in vivo. Mol. Pharmacol. 26, 594403. MICHALOWSKY, L. A., and JONES, P. A. (1989). DNA methylation and differentiation. Environ. Health Perspect. 80, 189-197. MOMPARLER, R. L., BOUCHARD, J., and ONETTO, N. (1984). Effects of 5-AZA-2’-deoxycytidine on DNA methylation in patients with acute leukemia. Proc. Amer. Assoc. Cancer Res. 25, 188. NC1 Technical Report: Bioassay of 5-Azacytidine for Possible Carcinogenicity. (1978). Series No. 42, CAS No. 320-67-2. OLSSON, L., and FORCHHAMMER, J. (1984). Induction of the metastatic phenotype in a mouse tumor model by 5-azacytidine, and characterization of an antigen associated with metastatic activity. Proc. Natl. Acad. Sci. USA 81, 3389-3393. PALM, P. E., ARNOLD, E. P., and RACHWALL, P. C. (1972a). “Repeated Dose Toxicity of 5Triazine-2(1H)-l,4-amino-l-2-D-ribofuranozyl-5-azacytidine (NSC 102816) in the Rhesus Monkey.” Report to the Laboratory of Toxicology, National Cancer Institute, Arthur D. Little Co. PALM, P. E., ARNOLD, E. P., and RACHWALL, P. C. (1972b). “Preclinical Toxicology of 102816 5-triazine-2(1H)-l,4-amino-l-2-D-ribofuranozyl-5-azacytidine Clinical Formulation Containing PVP in Hamsters and Dogs.” Report to the Laboratory of Toxicology, National Cancer Institute, Arthur D. Little Co. RENO, R., BOYSEN, B., and GLAZA, S. (1983). Task II Preclinical Single and Five Daily Dose RangeFinding and Lethality Studies of 5-Azacytidine (NSC 102816) in DC2f, Mice. Hazleton Laboratories American, Inc., Report to Toxicology Branch, Developmental Therapeutics Program, National Cancer Institute. RIGGS, A. D., and JONES, P. A. (1983). 5-Methylcytosine, gene expression, and cancer. Adv. Cancer Res. 40, l-30. SACCAMANNO, G., ARCHER, V. E., AUERBACH, O., SAUNDERS, R. P., and BRENNAN, L. M. (1974). Development of carcinoma of the lung as reflected in exfoliated cells. Cancer 33, 256270. SCHAUENSTEIN, K., ROSSI, K., and CSORDAS, A. (1988). Differential inhibition of nitrogen induced T-cell proliferation by 5-azacytidine and cytosine-arabinoside. Biochem. Biophys. Res. Commun. 151, 548 and 553.

SCHMAHL, W., GERBER, E., and LEHMACHER, W. (1985). Diaplacental effects of 5-azacytidine in NMRI mice. Cancer Let?.27, 81-90. SCHREIBER,H., SACCAMANNO, G., MARTIN, D. H., and BRENNAN, L. (1974). Sequential cytological

S-AZACYTIDINE

AND

BRONCHIAL

CARCINOGENESIS

51

changes during development of respiratory tract tumor induced in hamsters by benzo[n]pyreneferric oxide. Cancer Res. 34, 689-698. SHORS, E. C., Fu, P. C., and MATSUMURA, K. (1980). Sustained release of benzo[a]pyrene from silicone polymer into the tracheobronchial tree of hamsters and dogs. Cancer Res. 40, 2288-2294. SMITH, S. S., Yu, J. C., and CHEN, C. W. (1982). Diierent levels of DNA modification at S’CCGG in murine erythroleukemia cells and the tissues of normal mouse spleen. Nucleic Acids Res. 10, 4305-4320. STONER, G. D., SHIMKIN, M. B., and KNIAZEFF, A. J. (1973). Test for carcinogenicity of food additives and chemotherapeutic agents by the pulmonary tumor response in strain A mice. Cancer Res. 33, 3069-3085. VESELY, J., and CIHAK, A. (1973). High frequency induction in vivo of mouse leukemia in AKB strain by 5azacytidine and S-iodo-2’-deoxymidine. Experientia 29, 1132-l 133. VON HOFF, D. D., SLAVIK, M., and MUGGIA, F. M. (1976). S-Azacytidine, a new anticancer drug with effectiveness in acute myelogenous leukemia. Ann Intern. Med. 85, 237-245. WILSON, V. L., and JONES, P. A. (1983). Distribution of DNA methylation by chemical carcinogens in vitro. Cell 32, 239-246.