Age dependence of oval cell responses and bile duct carcinomas in male fischer 344 rats fed a cyclic choline-deficient, ethionine-supplemented diet


  • Ian Guest,

    1. Department of Translational Medicine, Wadsworth Center, New York State Department of Health, Albany, NY
    Search for more papers by this author
  • Zoran Ilic,

    1. Department of Translational Medicine, Wadsworth Center, New York State Department of Health, Albany, NY
    Search for more papers by this author
  • Stewart Sell

    Corresponding author
    1. Department of Translational Medicine, Wadsworth Center, New York State Department of Health, Albany, NY
    2. Ordway Research Institute, Albany, NY
    • Wadsworth Center, New York State Department of Health, P.O. Box 509, Room C-551, Empire State Plaza, Albany, NY 12201
    Search for more papers by this author
    • fax: 518-473-2900

  • Potential conflict of interest: Nothing to report.


The age dependence of the oval cell response and bile duct carcinomas of male F344 rats exposed to a cyclic choline deficiency-ethionine (CDE) diet (2 weeks on, 1 week off) supports the concept of loss of potential of liver stem cells to form cancers with aging. Livers of rats exposed at 3 weeks of age demonstrated a robust and widespread oval cell proliferation followed by cholangiofibrosis and bile duct metaplasia with extensive mucinous cysts throughout all lobes, and induction of cholangiocarcinomas (CCAs) in seven of eight rats. Livers of rats exposed beginning at 8 weeks of age had much less oval cell response and cholangiofibrosis with only 1 of 15 rats developing a CCA. Livers in old (10-12 months when started) rats remained virtually unaffected, with minimal oval cell proliferation, only occasional and small foci of ductular dysplasia, and none of 16 rats developed CCAs. In contrast to most published studies using uninterrupted choline deficiency plus a carcinogen, hepatocellular carcinoma (HCC) was not observed under the conditions of this study. Conclusion: With aging, male F344 rats exposed to cyclic CDE diet display a diminished oval cell response and fewer CCAs. The absence of HCC is possibly due to the fact that during cyclic CDE, the week off may allow putative liver stem cells to avoid death or differentiation and survive to give rise to CCAs, whereas with continuous CDE exposure, the stem cells are forced to differentiate and develop into HCCs with relatively few CCAs. HEPATOLOGY 2010

The stem cell theory of cancer posits that cancers arise from tissue-determined stem cells present in various organs.1, 2 One of the processes inherent in aging is a loss in the number or potential of these tissue-renewing stem cells,3-5 believed to be mediated by a decreasing ability to repair ongoing DNA damage,6, 7 such as that which occurs when organs are exposed to carcinogens. In this report, we compare cellular responses in the livers of rats exposed to hepatocarcinogenesis beginning at 3 weeks of age, beginning at 8 weeks of age, and in retired breeders (10-12 months of age), using a previously well-studied chemical regimen, choline-deficient, ethionine-supplemented (CDE) diet.8-10

Following the hierarchical model of Pierce et al.,11 combined with analysis of the cellular events during chemical hepatocarcinogenesis, we previously concluded that, in adults, liver cancers can arise from liver stem cells (oval cells), transit-amplifying cells (ducts or immature hepatocytes), or mature hepatocytes, depending on the stage of maturation arrest.12-15 Although that model fit the experimental observations on the cellular origin of hepatocellular carcinomas (HCCs) and cholangiocarcinomas (CCAs), it did not include the step between pluripotent stem cells (teratocarcinoma) and the liver lineage cells. The missing link is a cancer known as hepatoblastoma (HB).16, 17 HB completes the cellular lineage of liver cancer that extends from pluripotent stem cells to liver-determined stem cells to ductular stem cells to mature liver cells (Fig. 1A).

Figure 1.

Maturation arrest models of hepatocarcinogenesis. (A) Experimental hepatocarcinogenesis in rats is explained by maturation arrest at the adult liver stem cell or hepatocyte level. The missing link to pluripotent stem cells is the postulated cell giving rise to hepatoblastoma. From Sell.16 (B) Revised liver-cell lineage and maturation arrest model of hepatocarcinogenesis is shown. We now postulate that in rats, between birth and 3 weeks of age, liver stem cells differentiate to a stage of bile duct determination, such that exposure to the cyclic CDE regimen results in induction of CCAs. Our model predicts that at a younger age, there still exists a more primitive liver-determined stem cell that under appropriate conditions could give rise to hepatoblastomas.

We reasoned that if more liver stem cells are present, or if liver stem cells have greater potential in young rats as compared to old rats, then treatment of very young rats with a potent hepatocarcinogenic regimen should induce a more intense oval cell response and result in production of less well-differentiated HCCs, possibly such as HBs, than does treatment of older rats.18 We report here that cyclic CDE treatment of rats beginning at 3 weeks of age indeed caused a much higher level proliferation of oval cells than did treatment of rats beginning at 8 weeks of age and older. However, unexpectedly, exposure of the younger rats resulted in a high incidence of cholangiofibrosis and bile duct cancers, rather than HBs, suggesting that by 3 weeks of age the reactive liver-specific stem cells of the rat that give rise to cancer are already determined beyond the hepatoblast stage to ductal cell precursors.


CCA, cholangiocarcinoma; CD, choline-deficient; CDE, choline-deficient, ethionine-supplemented; HCC, hepatocellular carcinoma

Materials and Methods


Male Fischer 344 rats were obtained from Taconic Farms, Germantown, NY. Rats were housed in the Wadsworth Center's animal facility and had free access to standard laboratory chow and water when not in the cyclic CDE regimen. All animal experiments were approved by and performed under the guidelines of the Wadsworth Center's Institutional Animal Care and Use Committee (IACUC).


A choline-deficient diet was obtained from Dyets, Inc., Bethlehem, PA (catalogue #518753). D,L-Ethionine was purchased from Sigma Chemical Co., St. Louis, MO (catalogue #E5139).

Dietary Manipulation.

A cyclic feeding regimen of the CDE diet was followed. One cycle consisted of 2 weeks on the choline-deficient (CD) diet, followed by 1 week off. During the time when the rats were fed the CD diet, the drinking water was supplemented with D,L-ethionine at 1 g/L.

Three groups of rats were fed the cyclic CDE diet. Group 1 consisted of weanling rats that were 3 weeks old at the start of feeding; some of these rats were left untreated to serve as controls for both the 3 and 8 week old rats that were exposed to CDE feeding. Group 2 were 8 weeks old at the start of feeding and the group 3 were retired breeders (age 10-12 months). Some retired breeders were also left untreated to serve as controls for this third group. The 3-week cycle was repeated five times. At the end of cycles 1, 3, and 5, one to six rats from each of the experimental groups were euthanized for pathological analysis (Table 1). All surviving rats were left for long-term observation of tumor development and were only sacrificed when found moribund or at the termination of the study. Depending on the group, final groups of rats were 17.5 to 30 months old at study termination. See text below for details.

Table 1. Summary of Changes in Liver and Pancreas After Cyclic CDE Feeding for Three Age Groups of Male F344 Rats
Number of RatsOval CellsDuct MetaplasiaMitoses*Small HepatocytesLoss of GlandsOval Cells
  • For grading criteria see Fig. 2.

  • *

    If any mitotic cell seen on liver section, graded as 1. If several mitotic cells seen, graded as 2.

  • If a few clusters of small hepatocytes seen, graded as 1.

  • NS, tissue not on slide.

  • (Individual grading for each rat listed in Supporting Table 1).

3-Week-old rats
Cycle 141.801.5001.2
Cycle 364210.32.52.8
Cycle 554.2320.733.5
3-Week-old rats
Cycle 130.502.3100
Cycle 362.20.5112.02.0
Cycle 5421.
12- to 18-Month-old rats
Cycle 111010NSNS
Cycle 330.300.300.30.3
Cycle 5510.50.40.211

Tissue Collection and Analysis.

At selected times or when found ill as defined by IACUC criteria, rats were sacrificed by CO2 exposure/cervical dislocation. Samples of organs were fixed in formalin, processed, and embedded in paraffin; 5 μM sections were then cut and stained with hematoxylin and eosin. Selected sections were immunostained for epithelial cell adhesion molecule (EpCAM), hepatocyte nuclear factor 6 (HNF6), and C-Met (hepatocyte growth factor receptor) (see Supporting Fig. 5 for methods). Images were captured on an Olympus BX 51 microscope equipped with fluorescence detection and Optronics PictureFrame Version 1.2 software. More than 600 individual microscopic slides were examined.


Early Oval Cell Response.

The histologic grading of early changes for each experimental animal fed the CDE diet is presented in Supporting Table 1, and the results are summarized in Table 1 and Fig. 2A,B. The early changes associated with feeding of CDE are mainly seen in the liver and pancreas, and take the form of replacement of the normal cells with various numbers of oval cells. The extent of the oval cell response was graded as shown in Fig. 1A, and is clearly related to the age at the time of initiation of the CDE diet. Thus, up to 80% of the liver was replaced by oval cells after 5 cycles of CDE feeding to 3-week-old rats. A quarter or less of the liver was involved in rats fed CDE at 8 weeks of age, and very little response was seen in the retired breeders. A similar correlation with age was seen in the oval cell response in the pancreas (Fig. 2B). The oval cell response increased from CDE cycle 1 to 3 to 5 in the rats started at 3 weeks of age, but it decreased after three cycles in the rats started at 8 weeks and in the retired breeders (approximately 1 year of age). Early bile duct hyperplasia and metaplasia were also more frequent in the younger rats, but much more extensive bile duct changes were seen in the rats surviving until spontaneous death or euthanasia. Consistent lesions were not identified in other organs in the rats from the early-euthanized groups.

Figure 2.

(A) Histologic grading of oval cell reaction in liver (magnification, ×20). Panel 0, normal 4-month-old liver; panel 1, Grade 1; panel 2, Grade 2; panel 3, Grade 3; panel 4, Grade 4; panel 5, Grade 5. In (A), small round cells are limited to the portal plate. Grades 1 and 2 show short extension of cords of small oval cells into the hepatic parenchyma. Grades 3 through 5 show increasing replacement of the normal hepatocytes with small oval cells. In livers with grade 5 lesions, up to 80% of the liver may be replaced by oval cells. The bar graph shows the average grade for the rats from each group; the numbers at the top of the bars indicate the number of animals examined. (B) Oval cell response in rat pancreas. Panel 0, normal, ×20 (left) and ×40 (right); panel 1, 8-weeks-old, 5 cycles, ×40; panel 2, 3-weeks-old, 5 cycles; panel 3, 3-weeks-old, 5 cycles, ×20; panel 4, 3-weeks-old, 5 cycles. Islets of Langerhans may remain after acinar structures are lost. The grading indicates that from 20%-90% of the pancreatic acinar cells are lost. The bar graph shows the average grade for the rats from each group; the number of rats is the same as in (A). (C) Grading of bile duct hyperplasia and metaplasia. Panel A, Grade 1 (slide 09-216; magnification, ×10); panel B, Grade 2 (slide 09-696; magnification, ×10); panel C, Grade 3 (slide 09-202; magnification, ×20); panel D, Grade 4 (slide 09-208; magnification, ×4); panel E, focus of proliferation of small ducts (09-207; magnification, ×10); panel F, CCA. Grading is based on extent of the proliferation. For example, in grade 4, larger zones of the liver are involved than was the case for grade 3. There is extensive gastrointestinal metaplasia, with production of mucus (panels B,C,D). The difference between panel E (small duct proliferation) and panel F (CCA) is the entrapment of hepatocytes between the ducts as shown in panel F (arrows).

Late Cholangiofibrosis and Bile Duct Cancer.

The histologic grading of the lesions for each of the rats followed to death is listed in Supporting Table 2, and the results are summarized in Table 2. As expected, major lesions were present in the liver in the form of bile duct hyperplasia, metaplasia, and fibrosis, also known as cholangiofibrosis and termed “tubuloform degeneration” in the older literature.14 Intestinal metaplasia is a common feature of cholangiofibromas seen after oval cell proliferation in response to a chemical hepatocarcinogen.19 In the furan model of cholangiocarcinoma (CAA),20 intestinal metaplasia preceding CAA is associated with expression of CDX1, a caudal-type homeobox intestine-specific transcription factor,21 as well as overexpression of the tyrosine kinase growth factor receptors, C-NEU (epidermal growth factor) and C-Met,22 and hepatocyte growth factor/scatter factor.23 Although not tested in this article, it is likely that these factors play a critical role in the ductal differentiation of oval cells. The degree of cholangiofibrosis correlated with the age of initiation of the CDE feeding. Up to 30% of the liver was replaced by cholangiofibrosis in four of eight rats of the 3-week age group, whereas this occurred in only 2 of 15 of the 8-week age group, and in none of the retired breeder group. A striking finding is that seven of eight rats in the 3-week age group had bile duct cancers, whereas only 1 of 15 of the 8-week age group, and none in the retired breeder group demonstrated this cancer. Bile duct cancer (CCA) was identified on the basis of infiltration of small bile ductules into the liver, such that mature hepatocytes became entrapped between the expanding bile ducts (Fig. 2C, panel F).24 By contrast, in cholangiofibrosis involving small ducts, the ducts were surrounded by fibrous tissue (Fig. 2C, panel E). In some of the rats, large zones of the liver were occupied by CCA (see the 3-week age group in Supporting Table 2). Unexpectedly, no HCCs were seen.

Table 2. Summary of Lesions of Long-Term Survivors at Euthanasia or Death
No. of RatsAge Rangeat DeathLesion
(CIP)ScarringBDH&MBD CancerAtrophy(CIN)AtrophyCA
  • For grading of bile duct hypertrophy and metaplasia (BDH&M), lung, pancreas, kidney, and testes lesions, see Supporting Fig. 2. BD, bile duct; CA, interstitial cell carcinoma; CIP, chronic interstitial pneumonitis; CIN, chronic interstitial nephritis.

  • *

    Higher grades in older rats at time of death.

  • All rats over the age of 18.5 months had an interstitial cell cancer of the testes.

  • There was no testes tissue on slides from two rats; two rats over the age of 20 months did not have interstitial cell carcinomas of the testes.

  • (Individual grading for each rat listed in Supporting Table 2.).

Control rats for 3 and 8 weeks old groups. No CDE treatment
3-Week-old rats, completed 5 cycles of CDE
8-Week-old rats, completed 5 cycles of CDE
Retired breeder controls, No CDE treatment
1- to 1.5-Year-old rats, completed 5 cycles of CDE

Other Late Lesions.

Lesions of the lung (chronic interstitial pneumonitis), pancreas (atrophy and fibrosis), kidney (chronic interstitial nephritis), and testes (atrophy and interstitial cell carcinomas) were commonly seen (Table 2). In addition, there were cancers of various tissue origin in single rats (Supporting Information Table 2). Severe chronic interstitial pneumonia (Supporting Information Fig. 1) and nephritis (Supporting Information Fig. 2), as well as testicular atrophy (Supporting Information Fig. 3), were seen in both the control and experimental rats, but were marginally more severe in the experimental rats. These lesions have been previously reported in normal aged Fischer 344 rats.25 In fact, a major cause of death and decision to euthanize is renal failure in the aged Fischer rat. Interstitial cell cancers of the testes (Supporting Information Fig. 4A) are also commonly noted in aging Fischer 344 male rats.25 Indeed, those tumors were also found in high incidence in our control rats (15 of 26 rats, Table 2), but were absent in the 3-week and 8-week groups that underwent five cycles of CDE. On the other hand, 11 of 14 rats in the retired breeder group fed CDE had these tumors. The reason for the low number of such tumors in the younger rats is not clear, but could be due to sampling error or the fact that rats receiving CDE die at an earlier age than the controls or retired breeders fed CDE and thus could be dying before the interstitial cell tumors develop.

Various Other Cancers.

Other cancers found in single rats (Supporting Information Table 2, Supporting Fig. 2B-F) included leukemia, lung adenocarcinomas, renal cell carcinoma, mucinous carcinomatosis of the peritoneum, transitional cell carcinoma of the bladder, and an osteogenic sarcoma. These tumors are also found characteristically in aged Fischer 344 rats25 and did not show any clear association with CDE feeding. For example, we found two leukemias in the control groups and one in the experimental. The incidence of leukemia here was lower than what has been reported in the literature, but we found that a number of both control and experimental rats showed extramedullary hematopoiesis in the liver; that condition could have been reported as leukemia in other studies.


Localization of EpCAM, HNF6, and C-Met were done to determine expression in oval cells, cholangiofibrosis (CF), and CAA (Supporting Information Table 3, Supporting Fig. 5). EpCAM was present in normal ducts, oval cells, epithelial cells in CF, and CAA, but not in normal or reactive hepatocytes, thus showing consistency of expression in biliary cell types. HNF6 was present in oval cells and CF, but unexpectedly not in CAA. C-Met was expressed weakly in normal hepatocytes, but was highly expressed in focal hepatocytes (notably in mitotic hepatocytes) in CDE-fed rats, as well as in oval cells, CF, and CAA.


The cellular response of rats to a hepatocarcinogenic regimen is age-dependent. When fed five cycles of CDE diet beginning at age 3 weeks, seven of eight rats developed CCA, preceded by florid oval cell proliferation. In rats fed the CDE diet beginning at 8 weeks of age, the oval cell response was much lower and a CCA developed in only 1 of 15 rats. When rats were fed the diet beginning at 1 year of age, there was minimal oval cell proliferation and no bile duct carcinomas were seen. This result supports the concept that cancers arise from tissue-determined stem cells, and that the number or potential, or both characteristics, of tissue stem cells to respond to carcinogens declines with aging.

Unexpectedly, our cyclic CDE-fed rats did not develop either HBs, as predicted by our working hypothesis, or HCCs. In the literature, oval cell proliferation has generally been considered to be a precursor for development of HCC (for an extensive review, see Sell16). Early studies clearly showed that a combined deficiency of lipotropic factors enhanced the chemical induction of tumors of the liver.8, 9, 26 When Lombardi et al. combined CD with azaserine27 (not known to be a hepatocarcinogen) or with either of the liver carcinogens ethionine10 and acetylaminofluorene28 and fed to Sprague-Dawley rats at approximately 8 weeks of age, they reported rapid development of HCC, with an incidence of up to 80% after 6 months, and only a 38% incidence of CCAs.28 When initiation by diethylnitrosamine was followed by CDE, Takahashi et al. obtained similar results.29 Mikol et al.30 found a 90%-100% incidence of HCC in Fischer 344 rats initiated with diethylnitrosamine and fed a diet devoid of methionine and choline. We have no clear explanation for why our present finding of a high incidence of CCA in rats fed cyclic CDE beginning at 3 weeks of age but no HCC in either the rats started on CDE at 3 weeks or 8 weeks, is different from other results previously reported. Clearly, most investigators have reported HCC in the various hepatocarcinogenic regimens that induce oval cell proliferation. However, to the best of our knowledge, no other study used cyclic CDE exposure. Perhaps the week off during each cycle allows putative liver stem cells to evade death or differentiation and thus be able to give rise to CCAs; in contrast, with continuous CDE exposure, the stem cells would be forced to differentiate, such that they give rise to relatively few CCAs and more HCCs. An alternative explanation is that the default differentiation pathway of oval cells is to form ducts. If this is true, then when hepatocarcinogenic regimens induce large numbers of oval cells, CCAs would be expected rather than HCCs.

We chose a cyclic CDE dietary schedule for two principal reasons. The main reason was to reduce morbidity and mortality during a chronic feeding schedule, based on previous studies using a cyclic feeding of acetylaminofluorene.31, 32 Rats continuously fed CDE for more than 2 weeks in our previous studies died with massive oval cell proliferation.33 A secondary reason was as an attempt to augment oval cell proliferation and tumor development, by repetitive exposure to a CD diet.10, 28 Choline deficiency induces a state of fat deposition, apoptosis, and compensatory regeneration; in this abnormal situation, hepatocytes are forced to divide repeatedly, such that the deficiency has been termed a nutritional partial hepatectomy.34 The aberrant hepatocyte proliferation within the liver is believed to be the fundamental alteration that is ultimately responsible for the development of HCC from a methyl-deficient diet.35

As discussed above, our results are in contrast to those found previously. In a study entailing a CD diet, only 51% of male F344 rats fed that CD diet for 13-24 months had developed HCC by the end of the 24-month period.36 Three months duration on a continuous CD diet has been considered to be the minimum period required to induce HCC.34 It is thus possible that we did not expose the rats to total CDE for a sufficiently long enough period. In addition, ethionine was administered in the drinking water in the present study; consequently, dosing may not be comparable to a regimen in which ethionine is incorporated into the diet, given that rats drink less water when it contains ethionine.

Many others have used various carcinogenic regimens to study the origin of oval cell proliferation in rats, assuming that such proliferation is a precursor to development of HCC, but without actually following the treated rats to determine whether any cancers subsequently develop.37-40 However, after rapid proliferation, most oval cells, including those involved in bile-duct proliferation, are either lost by apoptosis or differentiate into mature liver cells41-43 or duct-like cells.44, 45 Thus, oval cells are part of a normal response to liver injury, producing progeny to replace hepatocytes and ducts. It is not known where the critical step occurs in this process that results in induction of cancers. However, shared marker phenotypes between oval cells and HCCs identified by monoclonal antibodies to cells in the liver lineage support the concept that oval cells could give rise to both HCC and CCA.14, 46, 47 Our marker studies using EpCAM, HNF6, and C-Met (Supporting Information Fig. 5) are not definitive but are consistent with an oval cell–to–duct cell lineage in development of CAA. Bile ducts, oval cells, CF, and CAA are all EpCAM-positive, whereas hepatocytes are not. In human liver cancer, EpCAM expression defines HCCs with stem cell features.48 C-Met is known to be positive in mucin-producing cholangiocarcinomas.22 The unexpected finding is that oval cells and bile duct cells in CF are positive for HNF6, but CAAs are negative. HNF6 is a transcription factor proposed to drive cholangiocyte differentiation.49 Thus, there appears to be a loss of this ductal phenotype with malignant transformation.

Although a direct comparison of our results to the human liver is not possible, it is likely that there is a decrease of functional liver stem cells in humans with aging. The major confounding factor is that there is no generally accepted marker for putative stem cells in the adult human liver. In fact, the adult human liver stem cell is functionally defined as “facultative”. That is, such cells are not identifiable in normal liver, but there are cells in the liver that respond to injury.50 OV 6 has been proposed as a marker for “transitional hepatocytes” that may serve this function.51 It has also been proposed that stem cells are located in the canals of Hering.52 There is a decrease in the number of biliary cells expressing the putative liver stem cell marker CD133 from 96.3% at 3 years of age to 59% in the adult.53