Cell proliferation not associated with carcinogenesis in rodents and humans.

Cell proliferation has often been found to be associated with carcinogenesis in rodents and humans at different stages of the multistage carcinogenesis process. The multistage process includes initiation, promotion, and progression phases. At each phase, increasing the normal level of cell turnover of target cells may enhance carcinogenesis. However, we present evidence that normal levels of cell turnover, or increasing the rate of cell turnover at these different stages, do not necessarily lead to enhanced carcinogenesis. In normal tissues, the length of the cell cycle depends on the age of the host and varies from tissue to tissue. Tissues with normal short cell cycles, such as intestine and bone marrow, do not show a high rate of spontaneous tumors in most species. Cells with higher turnover should be more susceptible to carcinogens at the initiation stage of carcinogenesis if cell proliferation per se causes cancer and if these cells or their progeny survive. Cancer in humans is more often associated with specific etiological factors rather than with the natural proliferative rate of specific tissues. For many tissues of humans and rodents, age-related diseases develop in a progressive, irreversible manner. Often, naturally occurring chronic degenerative and inflammatory changes in a tissue (e.g., kidney, liver, heart, reproductive tract) lead to chronic regeneration of the damaged tissue. Yet, cancer is rarely found in these tissues. In rodent carcinogenesis experiments, chronic toxic lesions, accompanied by increases in normal levels of cell turnover, have sometimes been observed in target organs of nongenotoxic carcinogens. More often, however, organ-specific nongenotoxic toxins are not carcinogens. These toxins include compounds toxic for the liver, kidney, and nasal cavity. In 19 inhalation bioassays conducted by the National Toxicology Program, 5/5 nasal carcinogens and 12/14 nasal noncarcinogens caused nasal lesions usually associated with chronic cell proliferation. Although cell proliferation may contribute to multistage carcinogenesis, cell proliferation is not necessarily a tumor promoter or cocarcinogen.


Introduction
Much information accumulated during the past 10 years reveals that cell proliferation (increased cell turnover, hyperplasia) is often associated with carcinogenesis in rodents and humans (1)(2)(3)(4)(5)(6)(7). Cell proliferation may contribute to carcinogenesis at stages of initiation, promotion, or progression (8). It has been postulated that cell proliferation can increase mutation rates or help fix spontaneous or induced mutations (1,2). With special pathology techniques, the labeling index (LI), determined by tritiated thymidine autoradiography or bromodeoxyuridine (BrdU) immunohistochemistry (9), is used as a crude estimate of the numbers of cells in S phase in tissues (10). Chemicals that have been found not to possess genotoxic activities or that are weak genotoxins sometimes cause tumors in rodent tissues, with chronic histological lesions of toxicity often indicative of cell proliferation. Similar situations may exist for genotoxic chemicals. In some tissues, such as mouse liver, chronic toxic lesions frequently accompany the hepatocarcinogenic or tumor-promoting activities of a chemical (11,12). As additional rodent bioassays are completed, recent information has become available that many organ-specific toxins are not carcinogens (13)(14)(15)(16).
This paper reviews some of these studies and addresses important questions relevant to these issues. We do not review or discuss the many associa-tions of cancer and cell proliferation reported in humans and rodents (4,5). Our review is limited by the type of information available. Many studies have no cell proliferation data available. A state of chronic cell proliferation is suggested when chronic lesions, including inflammation, epithelial hyperplasia, metaplasia, and other lesions, are reported. When available, cell proliferation data never contain cell cycle information.
We assume that the LI shows true cell turnover rates.
Sometimes, the nontarget cells are measured, and often transient increases in cell proliferation are seen rather than a sustained (persistent) increase. Little information is reported on the association of the level of induced cell proliferation and the level of the carcinogenic response. Should a 2-fold sustained increase in LI produce a 2-fold increase in tumor incidence or multiplicity? Also, the stage of carcinogenesis involved (initiation, promotion, or progression) is not usually determined for these effects. Much of the available information is suggestive but not conclusive of the stage of carcinogenesis affected.

Cell Turnover in Normal Tissues and Natural Cancer
The length of the cell cycle varies with tissue, cell type, age, and other factors (17). Adult tissues and cells with the shortest cell cycle times include the small intestine (10 hr in mice and rats), lymphoid germinative cells (germinal centers), and bone marrow. A comparison of cell cycle times and tumor incidence in humans (18) and rodents shows that there is no general correlation between cell cycle time and tumor incidence ( Fig. 1). In humans, etiological factors (viruses, genetic, or environmental factors) appear to be more important. Although colon epithelial cell-cycle times are short (25 hr in humans), low incidences of colon carcinoma are found in some populations (Japan), whereas high incidences are seen in others (United States). Mouse leukemia of lymphocyte origin, from cells with low cell-cycle times, occur at high incidences in some mouse strains but at very low incidences in many other strains. Evidently, the length of the cell cycle may be associated with carcinogenesis in some organs only or else other factors are involved (8).
It has been suggested that small intestinal (duodenal) proliferative cells migrate toward the tip of the villus and are sloughed into the gut lumen (1,2). This phenomenon may explain why the duodenum has low cancer rates despite low cell-cycle times. However, duodenal stem cells also have low cell-cycle times (at 25 hr) (17), do not migrate, and remain in the epithelium (Fig. 2), where they may be exposed to carcinogens and potential DNA damage.
Humans and animals that are larger (taller or heavier) than normal may develop cancer at higher rates than smaller humans and animals (19)(20)(21). The size of a specific cell type in an organ of each species should be identical in males and females, although there may be species differences. Thus, organs of larger humans or animals should have more cells to account for increased body and organ size. If a man weighs twice as much as another man (assuming similar body-fat mass), he may have twice as many cells in his body as the smaller man. If normal cell cycles and spontaneous mutation rates are similar for each cell in a specific tissue, then more spontaneous mutations should occur in a larger organ. If these rates were associated with spontaneous cancer (1,2), more cancer should be found in larger animals and people. Yet, the latter situation does not occur for most tissues. Some differences in humans of larger stature were reported for incidences of male colorectal cancer and a few other sites but not for most tissues, especially in females (20,21).

Cell Proliferation in Natural Diseases
Many tissues of animals and humans develop agerelated degenerative lesions. These changes occur from exposure to endogenous and exogenous substances and from genetically programmed processes. Damaged cells and tissues attempt to repair the degenerative processes through regeneration of the affected cells (Fig. 3). This process of damage and repair is continuous throughout life and is common in the kidney (22), liver, heart, skin, and reproductive tissues of many species (23). Repair is evidenced by the presence of hyperplastic basophilic cells in epithelial tissues, increased LI (10,12,22), and increased frequency of mitotic cells. Tumors may develop in these tissues in older individuals but often occur at low incidence (24).

Rats
Aging nephropathy (chronic progressive nephrosis) is often associated with regenerative renal cortical tubules in rats (22,24). Increased levels of cell proliferation are often seen in these regenerative tubules (22,25), yet spontaneous renal tubular cell tumors are uncommon [less than 1% incidence in rats (24)]. Interestingly, the levels of cell proliferation found in cortical tubules of aging nephropathy are similar to those induced by nongenotoxic renal carcinogens, such as unleaded gasoline (25). Renal tumors induced by these nongenotoxic renal toxins and carcinogens usually occur at rates greater than 20% (26). Bile duct hyperplasia is seen in most aged F344 rats, but tumors of these cells are extremely rare (22). Focal hyperplasia of hepatocytes (basophilic foci) is often observed in aging F344 rats, with little evidence of spontaneous tumorigenesis (24,27).

Humans
In humans, many chronic conditions are associated with regenerative hyperplasia or chronic hyperplasia without evidence of increased risk from cancer. These conditions can involve almost any tissue. Enlargement of the male breast (gynecomastia) may occur in response to excess estrogen (23). Gynecomastia is one manifestation of Klinefelter's syndrome and may occur in patients with functioning testicular neoplasms such as Leydig cell and, rarely, Sertoli cell tumors. Gynecomastia may occur any time during adult life when there is cause for hyperestrinism. The most important cause of hyperestrinism in the male is cirrhosis of the liver, since the liver is responsible for metabolizing estrogen. Gynecomastia also may be seen in chronic marijuana smokers and heroin addicts. Microscopically, there is marked hyperplasia of the ductal linings with piling up of multilayered epithelium ( Fig. 4). Gynecomastia does not appear to increase the risk of male breast cancer (28,29). Non-neoplastic nodular enlargement of the prostate (nodular hyperplasia, benign prostatic hypertrophy or hyperplasia) is the most common symptomatic tumorlike condition in humans. Studies have shown that nodular hyperplasia originates almost exclusively in the inner periurethral portion of the classically defined middle and lateral lobes (30). This distribution is in striking contrast to that of the prostatic carcinoma, which usually involves the posterior lobe. Microscopically, the nodularity may be due mainly to glandular proliferation or dilatation or to fibrous or muscular proliferation of the fibromuscular stroma. The epithelium is characteristically organized into many papillary buds and infoldings, which are more prominent than in the normal prostate (Fig. 5). Nodular hyperplasia of the prostate has been suggested as a cancer precursor, but most experts do not believe that this benign lesion has any relationship to the development of cancer (31).
Peptic ulcers are chronic, most often solitary, lesions that occur at any level of the gastrointestinal tract exposed to the aggressive action of acid-peptic juices. (23). Adjacent to the ulcer there is chronic inflamma-REPAIRED CELL tion and epithelial hyperplasia (Fig 6). Approximately 98-99% of peptic ulcers occur in either the duodenum or the stomach in a ratio of about 4:1. About 10-20% of patients with a gastric ulcer have concurrent duodenal lesions. In contrast to the stomach and colon, carcinomas are rare in the small intestine (18), despite the enormous surface area of the epithelium at risk and its constant replicative activity. Malignant transformation is unknown with duodenal ulcers but has been reported with gastric lesions (23). More likely, the "ulcer" was malignant from the outset. Cancers ulcerate, but ulcers rarely become cancerous (23). It is extremely difficult to establish the occurrence of a sequence of events in any particular case. Gastric cancer may arise in association with other types of proliferative lesions (4,23). Psoriasis is an inflammatory and epidermal hyperproliferative disease of the skin (32). Ki-67 immunoreactivity, BrdU exposure in vitro, and immunohistochemistry were used to evaluate rates of cell proliferation in epidermis of normal and psoriatic subjects. Patients with psoriasis had greatly increased numbers of labeled cells in the basal layer (some of which are probably stem cells) and other layers of the epidermis (32). Yet, epidemiological studies have not found an increased risk for skin tumors in these patients (33). However, after therapy with psoralens and ultraviolet A irradiation (PUVA) and ultraviolet B radiation, an increased risk of genital skin tumors was seen (34). Although the hyperproliferative state of psoriasis alone does not lead to cancer, it may serve as an important cofactor with PUVA therapy.

Induced Cell Proliferation and Tumors in Rodents
There is an increasing volume of information that shows chemically induced chronic toxicity, with histological evidence of epithelial regeneration (cell proliferation, hyperplasia) and with or without cell proliferation data (tritiated thymidine autoradiography and BrdU or proliferating cell nuclear antigen [PCNA] immunohistochemistry) is not associated with carcinogenesis in rodents (9)(10)(11)(14)(15)(16)35,36). This phenomenon may be more common than are experiments with associations of chronic toxicity, cell proliferation, and carcinogenesis.

Nasal Carcinogenesis
Nasal carcinogenesis is often induced by potent genotoxic chemicals, especially specific genotoxic nitrosamines (36,38). Some nongenotoxic or weakly genotoxic nasal carcinogens have induced such tumors accompanied by inflammatory and hyperplastic epithelial lesions. Formaldehyde-induced nasal carcinomas and hyperplastic lesions are seen with increased levels of cell proliferation in rats (39,40). More recently, bioassays sponsored by the National Toxicology Program (NTP) (41)(42)(43)(44) have provided information on inhalation studies with 10 chemicals in rats and mice (Tables 1-3). Five chemicals were found to be nasal carcinogens, and 14 chemicals were not carcinogenic for the nasal cavity (Tables 1-4). Of the 5 carcinogens, all induced inflammatory and proliferative (regenerative) nasal lesions. Of the 14 noncarcinogens, 12 induced similar inflammatory and proliferative nasal lesions. Of the 5 carcinogens, 3 also caused tumors at other sites, and 7 of the 14 nasal noncarcinogens caused other tumors. By   15 15 13 11 Nasal noncarcinogens 57h 36 26 23 aLesions include inflammation of various types, epithelial hyperplasia, squamous metaplasia, and metaplasia. bIncludes five sex/species groups given chemicals inducing nasal tumors in other sex/species groups.  (Table 2). Often, a whole spectrum of lesions was seen, similar to those lesions described for formaldehyde (39).
Recently, in a detailed 6-week inhalation experiment with formaldehyde, the L3 segment of the rat nasal cavity was found to have the highest level of persistent epithelial cell proliferation, although tumors rarely developed in this region (45). Inhalation of 1,3-dichloropropene produced nasal erosions and fibrosis with hyperplasialhypertrophy in mice but no hyperplasia in rats, although these types of lesions are often associated with hyperplasia (46).

Liver
Chronic toxicity has also been implicated as the driving force behind the carcinogenic potential of some chemicals (especially the nongenotoxic carcinogens) in the liver, via the indirect stimulation of chronic compensatory cell proliferation with "higher dosages" (1,2,7,10). However, several inconsistencies have been reported in the literature (47). Recently, a retrospective evaluation of the NTP's carcinogenesis database revealed that only 25% of the F344 rat bioassays evaluated exhibited a positive correlation between chemically induced toxicity and increased liver tumor occurrence (14). In the B6C3Fj mouse studies, a positive correlation occurred in only 51% of the studies assessed. Studies conducted in our laboratory revealed similar noncorrelative results. Chronic administration of acetaminophen to B6C3Fj mice resulted in severe liver toxicity; however, no increases in tumor incidence were noted (9,48). It must be noted that in those studies, only limited cell proliferation data were sometimes available, and therefore we can only deduce that there is, at most, a small correlation between chemically induced toxicity and hepatic carcinogenesis.
Inconsistencies in correlating the induction of cell proliferation by a chemical with its known or potential tumorigenic capabilities have also been identified in the liver (47). An example of a positive correlation between the two was reported by Cunningham et al. (12) using two isomers of diaminotoluene (DAT). The compounds 2,4-DAT and 2,6-DAT are equally absorbed, metabolized, and excreted and are genotoxically active. However, only 2,4-DAT produced liver tumors in chronic rodent bioassays. When similar doses of each compound were evaluated over a 9-day period for their effects on hepatic cell proliferation, 2,4-DAT produced a dose-dependent increase in DNA synthesis, whereas 2,6-DAT had no effect. It is not known if this proliferative response is sustained. In contrast to these reports, other investigators have reported studies in which no such correlations can be made. Chronic exposure of B6C3Fj mice to unleaded gasoline vapors increased liver tumorigenesis in female B6C3Fj mice but not in male mice. In contrast, when DNA synthesis was evaluated, a 6to 10-fold increase in the LI was found in the livers of both female and male mice (49). Sex specificity for tumorigenicity and cell proliferation induced by 1,4-dichlorobenzene was not seen (50).
Di(2-ethylhexyl)phthalate) (DEHP) is a plasticizer, rodent hepatocarcinogen, and hepatotoxin (51). As a peroxisomal proliferator, DEHP causes severe liver enlargement (more than twice normal) and persistent (sustained), dose-related increase in the hepatocyte LI [more than 10 times normal at the highest dose (9)]. In the original NTP bioassay, hepatocellular adenomas or carcinomas were found in 14/50 (28%) of the control males, 25/48 (52%) of the low-dose males (3000 ppm), and 29/50 (58%) of the high-dose (6000 ppm) males (52). Thus, tumors were found in twice as many low-dose and high-dose males as in controls. In a 40-week toxicity study, we found doses of 6000 ppm caused a 1-to 3fold increase in hepatocyte LI (9). These findings suggest a persistent LI at the level of the tumor increase. In a recent 91-week study, we found that although the persistent hepatocyte LI of mice given 12,000 ppm was 10-to 15-fold over normal, the tumor incidence was only 3-to 5-fold over normal (J. M. Ward and H. Uno, unpublished data). The numbers of preneoplastic foci per liver, however, were more closely correlated with LI. Also, the tumor incidence in mice receiving 6,000 ppm was similar to that in mice at 12,000 ppm despite a 2-to 10-fold difference in LI. It was recently reported that cell proliferation by the carcinogenic peroxisomal proliferators, clofibric acid and nafenopin, was not sustained beyond 4 weeks (53). Additional studies should help clarify the quantitative relationships, if any, between dose, LI, and tumor incidence.

Skin
Tumor promotion in mouse skin is usually associated with epidermal hyperplasia (10,54). Most skin irritants are tumor promoters in mouse epidermis, except turpentine (54). In skin tumor-promotion assays and in a short-term assay for hyperplasia, turpentine did not possess promoting activity but did cause epidermal hyperplasia (Fig. 7). the level of cell proliferation was similar to that of ethylphenylpropiolate, a tumor promoter (54).

Kidney
Unleaded gasoline and 2,2,4-trimethylpentane caused a2u-globulin accumulation in the P2 segment of the proximal renal tubules, persistent increases in the LI of the P2 segment, and renal tumors in this segment (25,26). The P3 segment, however, also had high persistent levels of cell proliferation without evidence of renal carcinogenesis (25). Although several compounds classified as chemicals inducing a2u-globulin accumulation (CIGAs) have induced renal tumors and apparent renal tubular hyperplasia, rats administered gaba- pentin did not develop renal tumors despite the induction of renal hyaline droplets and renal tubular hyperplasia (55). DEHP, a renal toxin in B6C3F1 mice, produced sustained increases in LI of renal tubules but was not a renal carcinogen or tumor promoter (35). Streptozotocin initiation inhibited the nephropathy and cell proliferation induced by barbital sodium but failed to inhibit the tumor-promoting effects of the barbital sodium (36). Mercuric chloride produced a chemi-cally related increase in nephropathy and parathyroid hyperplasia (Fig. 8) in rats but no renal tumors in a recent NTP bioassay (56). Four other renal toxins were not carcinogenic in 2-year bioassays (15). Diphenyl induced renal pelvic stones, obstructive pyelonephritis, renal tubular atrophy, cystic tubules, and fibrosis (57), lesions usually associated with increases of cell proliferation in renal tubules. When diphenyl was studied for renal tumor-promoting activi-   (57). Renal pelvic hyperplasia was induced in male rats by d-limonene but no pelvic tumors were found (58).

Urinary Bladder
Hyperplasia of the urothelium has been suggested to play an important role in bladder carcinogenesis and tumor promotion (3)(4)(5). Yet, examples can be found that suggest hyperplasia is not associated with carcinogenesis in these instances. Hyperplasia of the urinary bladder induced in mice by 1,3-dichloropropene was not significantly associated with any bladder tumors (46). Melamine caused bladder stones and epithelial hyperplasia in rats and mice, yet bladder tumors were only found in rats (59). Papillary-nodular bladder hyperplasia, type A, did not develop into papilloma, but type-B hyperplasia progressed to papilloma (60). Butylated hydroxytoluene (BHA) induced a high level of persistent increases in bladder epithelial DNA synthesis (Fig. 9), and sodium L-ascorbate produced a 5-fold lower level increase of DNA synthesis (61). Yet, sodium L-ascorbate promoted bladder hyperplasias, papillomas, and carcinomas to a similar degree as did BHA (62)(63)(64). BHA was not carcinogenic for the rat urinary bladder in long-term studies (65,66).    were not available (14,15). These reviews involved many tissues. Besides the tissues noted above, we occasionally found published reports that showed histological evidence of chronic toxicity and/or hyperplasia without carcinogenesis. Although forestomach hyperplasia was associated with carcinogenesis in male and female rats gavaged with 3-chloro-2-methylpropene, low dose female rats had basal cell hyperplasia at an incidence of 84% without evidence of carcinogenesis (67). Four other chemicals caused forestomach hyperplasia without evidence of carcinogenesis (15). Gastric and duodenal ulcers induced by freeze ulceration increased cancer incidence induced by a nitrosamine in the gastric fundus and proximal duodenum, but not in the other parts of the duodenum, pylorus, or forestomach (68). An interesting phenomenon has been described in amphibia exposed to carcinogens. Forelimb blastemas, generated after limb amputation, have marked levels of regeneration, yet are resistent to carcinogenesis by potent carcinogens (69).

Summary and Conclusions
Cell proliferation has often been associated with carcinogenesis in specific tissues of humans and rodents (4)(5)(6). However, an increasing number of examples can be found that suggest cell proliferation is often not associated with carcinogenesis (9,14,16,35,45,47,49,50). In this review, we presented some examples of cell proliferation without carcinogenesis. Many other cases can be found. If chronic cell proliferation occurs but is not associated with an increased risk for carcinogenesis, several possible hypotheses may be considered. If one assumes that cell proliferation should increase cancer risk in most cases (4,5), the lack of carcinogenesis in some situations mnay occur for the following reasons.
Cell proliferation may occur without involvement of stem cells, which are important targets of carcinogens (4,5). Often, increases in the compartment of proliferating cells are responsible for cell proliferation in a tissue. Stem cells should be involved in the proliferative process (5) but may not be the only targets of carcinogens (70). Preneoplastic cells and foci occurring naturally or induced, may also be targets of carcinogens or tumor promoters. If transient or sustained cell proliferation occur before the appearance of these lesions, the maximal effect of cell proliferation will be minimized. The effect of cell proliferation should be dose related, but limited evidence suggests otherwise (9,10,15). One might expect a dose-related increase of the level of cell proliferation and carcinogenesis if the two phenomena were entirely related.
Finally, cell proliferation may not play a role in carcinogenesis in specific cases only or in all cases in which it is found. Cell proliferation may be a red herring in the multistage carcinogenesis process. There is evidence that some mutations may be time dependent instead of replication dependent (71). Research on the role of spontaneous mutations in carcinogenesis by nongenotoxic and genotoxic carcinogens should provide some important information on the role of cell proliferation in carcinogenesis.