Classification of Proliferative Hepatocellular Lesions in Harlan Sprague–Dawley Rats Chronically Exposed to Dioxin-Like Compounds

Abstract

Over the years, the most appropriate classification scheme for nodular proliferative lesions of the hepatocyte has been heavily debated. In the most recent guidelines there appears to be a consensus for classifying these lesions as hepatocellular adenoma, hepatocellular carcinoma, or regenerative hyperplasia. Also, large foci of cellular alteration may appear somewhat nodular. Some nodular hepatocellular lesions from a group of 7 studies of dioxin and dioxin-like compounds conducted by the National Toxicology Program did not readily fit into these categories. Some of these lesions had morphologic features consistent with hyperplasia. However, there was not sufficient morphological or biological evidence to conclude that the entire response was regenerative. In other instances, these lesions had some features resembling adenoma, but contained a prominent component of biliary epithelium and/or oval cells. This component does not appear to be well described in the literature, and while its presence suggested a nodule to be nonneoplastic, this is inconclusive. This paper describes the morphology of these lesions, as well as the diagnostic approach taken in this series of studies.

Introduction

Polyhalogenated aromatic hydrocarbons such as 2,3,7,8-tetrachlorodiobenzo- p-dioxin (TCDD, dioxin) have the ability to bind to and activate the ligand activated transcription factor, the aryl hydrocarbon receptor (AhR). Structurally related compounds that bind to the AhR and exhibit biological actions similar to TCDD are commonly referred to as “dioxin-like compounds.” Ambient human exposure to dioxin-like compounds occurs through the ingestion of foods containing residues of dioxin-like compounds that bio-concentrate through the food chain. Due to their lipophilicity and persistence, once internalized they accumulate in adipose tissue resulting in chronic lifetime human exposure.

Since human exposure to dioxin-like compounds always occurs as a complex mixture, the Toxic Equivalency Factor (TEF) methodology has been developed as a mathematical tool to assess the health risk posed by complex mixtures of these compounds. The TEF methodology is a relative potency scheme that ranks the dioxin-like activity of a compound relative to TCDD. TCDD is considered the most potent of the group, and has a ranking of 1 with other compounds given lesser potency ranking (e.g., the TEF of PCB 126 is 0.1). This allows for the estimation of the potential dioxin-like activity of a mixture of chemicals, based on a common mechanism of action involving an initial binding of dioxin-like compounds to the AhR.

Dioxin-like compounds and the TEF methodology were nominated by the U.S. EPA for evaluation by the NTP, because of the widespread human exposure to dioxin-like compounds, and the lack of data on the adequacy of the TEF methodology for predicting relative potency for cancer risk. To address this, the NTP conducted a series of 7 (National Toxicology Program, 2004a–2004g) 2-year rat bioassays to evaluate the chronic toxicity and carcinogenicity of dioxin-like compounds and structurally related PCBs and mixtures of these compounds. There were dose-related increases in liver pathology including toxic changes and proliferative lesions. Many of the proliferative lesions possessed morphological features intermediate between categories from currently accepted classification schemes, and posed significant diagnostic difficulty.

The liver is the most common site for chemically induced proliferative lesions in rodent carcinogenicity studies conducted by the NTP (Haseman et al., 1984; Huff et al., 1991). However, the relevance to humans of treatment-related hepatocellular neoplasms, particularly in the B6C3F1 mouse and especially when it is the only neoplastic effect, is controversial. Additionally, the proper categorization of proliferative lesions of the hepatocyte has been heavily debated over the years (Squire and Levitt, 1975; Stewart et al., 1980; Maronpot et al., 1986). Proliferative lesions of presumably similar morphology have been termed nodular hyperplasia, regenerative hyperplasia, hyperplasia, neoplastic nodules, hepatoma, and adenoma. In 1975, a workshop recommended changing the term “hyperplastic nodule” to “neoplastic nodule” (Squire and Levitt, 1975). Later Maronpot et al. (1986) recommended that “neoplastic nodule” be changed to “hepatocellular adenoma” and this remains the currently accepted terminology.

The primary purpose of a classification scheme is to create categories that represent significantly separate biological entities, and to define criteria that allow accurate and consistent placement into these categories. The varied morphology of proliferative lesions of the liver coupled with insufficient information as to the biological nature of these lesions has confounded proper classification. With little or no biological information, classification schemes rely more upon morphology, and are therefore more speculative. The more recent guidelines (Eustis et al., 1990; Goodman et al., 1994; Narama et al., 2003) suggest separation of proliferative lesions into foci of cellular alteration, regenerative hyperplasia, adenoma, and carcinoma. There is some overlap of the criteria of various categories; however, criteria unique to a specific lesion are also included. These categories appear adequate for spontaneous lesions observed in untreated or control animals, and for most chemically induced lesions. No matter how good a classification scheme for the liver, lesions may occur that possess features intermediate between 2 categories (e.g., focus vs. adenoma). While these can pose a diagnostic dilemma and result in disagreements between pathologists, the incidences are generally low and study conclusions are not usually affected.

This paper describes the lesions, and discusses the morphological features and the diagnostic criteria used in this series of studies of dioxin and dioxin-like compounds. While the categorization reflects the current state of knowledge, further insight into the biological behavior may indicate whether these proliferative lesions have been properly categorized.

Methods and Materials

The studies were conducted in female Harlan Sprague–Dawley rats based on the prior observations of the carcinogenicity of TCDD in this strain (Kociba et al., 1978). Female rats were chosen based on the high potency of TCDD for hepatocarcinogenicity in the females of this strain. Animals were dosed by oral gavage since the majority of human exposure is oral.

Dose selection for TCDD (3–100 ng/kg/day) was based on the prior range used in the Kociba et al. (1978) study and on the demonstrated induction of liver tumor promotion over this dose range. Dosage levels for other dioxin-like compounds were based on the TCDD dosage range after adjustment for the current TEF values. These studies were designed to examine dose additivity rather than response additivity and therefore dose spacing was weighted in the 10–100 ng/kg dose range to increase dose density in the region where the greatest increase in liver tumors was expected. Doses higher than 100 ng/kg were not used in order to limit effects on body weight and liver toxicity. However, because of contamination of the study of PCB-118 with PCB-126, it ultimately was considered a mixture study. In that study, levels of PCB-126 were determined to be significantly higher than in the individual study of PCB-126.

Generally, there was 1 control group and 6–7 treated groups, including 1 stop study group that received the highest dose for 31 weeks, and then received vehicle only until study termination at 2 years. Interim necropsies at 14, 31, and 53 weeks were incorporated into the studies for the examination of mechanistically based biomarkers of AhR or PCB mediated effects. These endpoints included alterations in cytochromes P450 1A1, 1A2, and 2B, thyroid hormone levels, and hepatocyte replication. For cell proliferation analysis, up to 10 female rats per group received drinking water containing 40 mg BrdU/100mL Milli-Q water for 5 days. Tissue sections were stained with anti-BrdU antibody complexed with avidin and biotin. Cell turnover rate in the liver of dosed rats was compared to that in the vehicle control by determining the incorporation of BrdU into hepatocytes. Then, 2000 labeled or unlabeled hepatocyte nuclei were counted using a X20 objective and ocular grid. The labeling index was calculated as a percentage of BrdU labeled nuclei. Tissue analyses of parent compound in the liver, lung, blood, and adipose tissue were included at each interim necropsy and at terminal necropsy for dose response analysis using both administered dose, total body burden, and target tissue dose as the dose metric.

As with all NTP studies there was a comprehensive histopathological evaluation of tissues. However, this manuscript focuses on liver tissue from this group of studies as well as livers from a few F344/N rats from studies of other compounds conducted by the NTP. All studies were conducted under Good Laboratory Practice Standards. The livers were sampled consistently among all animals from a study, and were prepared for histological examination in the usual manner involving hematoxylin and eosin staining of paraffin embedded tissue. Special stains such as Perl’s stain for iron were applied to selected liver specimens.

These studies were conducted at laboratories under contract to the National Toxicology Program, and in addition to the rigorous evaluation and peer review conducted internally at those study laboratories, were subjected to the usual peer review at and by the NTP. The extent of NTP’s usual peer review depends upon the study complexity and findings. The data are carefully evaluated, and tissues (approximately 20%) selected for review by a second (pathology quality assessment) pathologist. These same tissues are further reviewed by a 3rd (pathology working group) pathologist. Unresolved or potentially problematic lesions are presented to a group of experienced Toxicologic Pathologists for review, diagnosis and comment in the form of a Pathology Working Group.

Accuracy of diagnoses and consistency of diagnostic approach, within a study is paramount to NTP studies, and every reasonable effort is always made to ensure both. Primarily because of the critical need to be consistent with the diagnostic approach among all studies in this group, but also because of the complexity of many of the lesions, the pathology evaluation, diagnostic approach, and pathology peer review were more rigorous for this group of studies. For the sake of consistency, 4 pathologists were involved as principals in the initial evaluation and peer review. One pathologist (study pathologist) completed the histological evaluation for all 7 studies and the second (QA pathologist) completed the Quality Assessment evaluation. The third (PWG pathologist) completed the PWG review and chaired each of the PWGs, while the fourth (NTP pathologist) provided oversight of the NTP peer review process. Because some of the lesions in these studies were identified early on as rather unusual, there was a pre-PWG to gain insight into the best diagnostic approach early in the review process. To ensure consistency of approach, in addition to the individual PWGs, hepatocellular lesions from the first 5 studies were reviewed in a special PWG.

After completion of the studies, in addition to NTP’s routine pathology peer review (including PWGs) for each of the 7 studies, many of the lesions were evaluated in a special workshop. A panel of experienced pathologists with specific expertise in the evaluation of liver and/or carcinogenicity bioassays was convened to examine selected proliferative hepatocellular lesions from this series of studies. The panel of pathologists examined 93 slides with sections of liver divided into 6 groups which were selected to demonstrate the full spectrum of change observed within these studies. A few slides from other studies were provided for comparison. Early lesions (those observed in interim sacrifices) from the low- and high-dose groups from the PCB 126 study were also included.

Following the examination and discussion of lesions present on individual slides, the panel provided guidance concerning the most appropriate diagnostic criteria and nomenclature to classify the lesions which were often present in livers with extensive toxic change. The panel was chaired by Dr. Jerry Hardisty, and participants were: Drs. Peter Bannasch, John Cullen, Michael Elwell, James (Rick) Hailey, Ernest McConnell, James Swenberg, Jerrold Ward, and Douglas Wolf.

Results

Only data from the liver with emphasis on proliferative lesions are discussed in this report. While there were some quantitative differences between the hepatocellular proliferative lesions observed within these studies, they were qualitatively similar. Incidence data from all the TEF studies are not included, rather data from TCDD are included as representative of the group. However, lesions shown may be from any of the studies, including the mixture of PCB-126 and PCB-118 in which the doses were inadvertently higher than for the other studies. Effects observed in the liver and at other sites are included in detail in the NTP technical reports (National Toxicology Program, 2004a–2004g).

Cell Proliferation

Hepatic cell proliferation data was collected at 14, 31-, and 53-week interim terminations. Data from the TCDD are shown in Table 1, and are representative of the other studies. The labeling indices increased significantly with higher doses and later time points.

Nonproliferative Lesions

To illustrate the occurrence and progression of liver lesions in these studies, the incidences and severities of lesions recorded in the TCDD study are included in Tables 2 and 3. At 14 weeks, the principal finding in the liver was minimal-to-mild hepatocyte hypertrophy in most or all animals in the top 2–3 dose groups; this change was also prominent at 31 weeks. Additionally, at 31 weeks, minimal-to-mild multinucleated hepatocytes occurred in most animals in the top dose group, and minimal-to-mild pigmentation was observed in most animals from most dose groups. There was also minimal fatty change in 6 of 10 animals in the high-dose group. While most animals, including controls, had minimal inflammation in the liver, the severity tended to increase in higher dose groups. As early as 31 weeks, the incidences of mixed cell foci were increased in the high-dose group. Mixed cell foci were observed in control animals, but they were fewer and much smaller than those in the treated animals. At 53 weeks, all the changes observed at 31 weeks were present and were generally more severe. In addition, there was an increased incidence of minimal to mild bile duct hyperplasia. At 2 years, these lesions tended to be even more severe, and there were increased incidences of necrosis, oval cell hyperplasia, bile duct cysts, portal fibrosis, and cholangiofibrosis.

Hepatocyte hypertrophy was characterized by hepatocytes that were enlarged with increased amounts of eosinophilic cytoplasm. Minimal hypertrophy affected periportal hepatocytes, but as severity increased, hepatocytes in other areas of the hepatic lobule were also affected. Multinucleated hepatocytes were characterized by scattered hepatocytes that were enlarged and contained multiple (more than 2 and often up to 10) nuclei. Inflammation was generally minimal to mild, consisting of lymphocytes and plasma cells with fewer macrophages within portal areas or scattered randomly throughout the liver. Pigmentation consisted of light brown to golden pigment within macrophages and occasionally hepatocytes. Pigmented macrophages were often seen in portal areas but were also scattered within the liver. The pigment was positive for iron with Perl’s stain.

Diffuse fatty change was generally a minimal to mild change consisting of discrete clear vacuoles in the cytoplasm of hepatocytes scattered diffusely throughout the liver. Necrosis consisted of scattered necrotic areas of hepatic parenchyma that were often randomly distributed, but occasionally, in more severe cases, were more diffuse. Oval cell hyperplasia (Figure 1A) was characterized by small ovoid cells with basophilic cytoplasm, and round to ovoid nuclei that were arranged in single or double rows and located predominantly in the portal areas. Bile duct hyperplasia (Figure 1A) consisted of increased numbers of portal bile ducts. Bile duct cysts were either unilocular or multilocular and lined by attenuated epithelium. Portal fibrosis consisted of fibrous connective tissue accumulation that extended between adjacent portal areas. Cholangiofibrosis was characterized by atypical ducts surrounded by abundant connective tissue. The ducts were often irregular, dilated, and contained mucinous material and cellular debris.

Toxic hepatopathy was diagnosed in some animals from 31 weeks on. Toxic hepatopathy comprised all the toxic effects (included in Table 2) observed in the liver. Summarizing the toxic effects of the liver into this single diagnosis made comparison of the liver toxicity between the various studies easier. It was diagnosed in addition to, and not instead of the individual toxic changes. When only findings of hepatocyte hypertrophy, pigment, and slight fatty change were present (as occurred at 14 weeks) no diagnosis of toxic hepatopathy was made. Minimal (grade 1) toxic hepatopathy was diagnosed when additional changes indicative of a toxic effect, usually a slight degree of bile duct and/or oval cell hyperplasia or a few large prominent altered hepatocellular foci, and occasionally a small focus of cholangiofibrosis were present. Mild (grade 2) toxic hepatopathy was characterized by the presence of multiple toxic changes, all of which were of minimal to mild severity. In addition, multiple prominent altered hepatocellular foci (usually mixed cell foci) and an occasional focus of nodular hyperplasia were sometimes present. Moderate (grade 3) toxic hepatopathy was diagnosed when most or all the spectrum of toxic changes were present, with some degree of distortion of the normal liver structure caused by prominent altered hepatocellular foci, nodular hyperplasia, and cholangiofibrosis. Marked (grade 4) toxic hepatopathy was diagnosed when severe toxic changes were present with pronounced distortion of the liver architecture. Livers with marked toxic hepatopathy often had a multinodular appearance due to the presence of numerous large foci of nodular hyperplasia that replaced much of the liver parenchyma.

Proliferative Lesions

Eosinophilic and/or mixed foci in controls were generally small (Figure 1B); they consisted of hepatocytes that were generally somewhat enlarged, but were arranged in a relatively normal lobular pattern. The hepatic cords at the periphery of these foci generally merged imperceptibly with the surrounding liver, resulting in an indistinct border and little or no compression of the adjacent liver parenchyma. As early as 53 weeks, foci in treated animals were multiple, prominent and larger than those observed in controls (Figure 1C). At later time points the foci in treated animals generally had a more definite border and caused more compression than spontaneous foci. Although a relatively normal lobular pattern was maintained, the cords within the foci often were not smoothly continuous with those in the surrounding parenchyma, and the foci consisted of cells that were enlarged with abundant eosinophilic or clear vacuolated cytoplasm. Within some livers there appeared to be a morphological continuum between large foci and lesions with some features of hepatocellular adenoma (Figure 1D).

In the most severely affected animals (particularly from the PCB 126/118 study), livers had the appearance of a cirrhotic liver (Figure 1E); however, as is typical in the rat, there was less fibrosis than is usually associated with cirrhosis in man or dog. In these rats the liver was almost entirely composed of nodules ranging from a few hundred microns to over a centimeter in diameter. At lower doses and even at higher doses in some of the studies, the nodules were not as prominent or numerous, and in some instances there was only a single nodule (Figure 2A). These nodules were diagnosed as nodular hyperplasia, and were generally composed of hepatocytes that were considerably larger than normal (hypertrophic, Figure 2B), often with variable degrees of cytoplasmic vacuolation and sometimes mixed with areas of increased numbers of small hepatocytes. The cells appeared to be arranged in normal cords, but the cells often were so large as to obscure the sinusoids between the cords giving the appearance of solid sheets of hepatocytes. Areas of nodular hyperplasia sometimes blended with the surrounding parenchyma, but often had a distinct border (Figure 2B). Larger nodular hyperplasias sometimes caused significant compression of surrounding tissue, and bulging of the capsular surface. Portal areas were usually identifiable within the nodules, and blood vessels or central veins were also sometimes seen.

Also, variably scattered throughout many of the nodules was a proliferation of biliary epithelium and/or oval cells. These occurred as either single cells or clusters of spindled to cuboidal cells that often appeared to form ductular structures cut in cross- or tangential section (Figures 2C and 2D). Similar proliferations also occurred in the adjacent liver. While this was a common feature, some nodules had areas without these structures. The presence of hypertrophic, eosinophilic to vacuolated hepatocytes and/or apparent proliferation of biliary epithelium and/or oval cells was considered characteristic of nodular hyperplasia.

Many hepatocellular adenomas readily conformed to the generally accepted criteria, and were nodular masses that were usually larger than a focus, had a distinct border, and caused compression of surrounding parenchyma. They were composed of mildly to moderately pleomorphic hepatocytes that were of normal size or slightly larger than normal and were arranged in abnormal lobular patterns. The hepatic cords within an adenoma usually intersected the surrounding normal hepatic cords at an oblique or right angle. If present, a few portal triads or bile ducts were found only at the periphery of the lesion; these likely represented entrapped structures. In some of the adenomas that were considered to have potentially progressed from a nodular hyperplasia, the oval cell and/or biliary epithelium described before was present; however these particular lesions also contained large areas devoid of these components.

Hepatocellular carcinoma occurred rarely and was a large, poorly demarcated, locally invasive mass composed of atypical hepatocytes that were arranged in trabeculae three or more cells thick, in glandular formations, and solid sheets.

Discussion

Lesions of the liver were similar between all the studies of dioxin and dioxin-like compounds, and the incidences and severities of effects increased with dose and duration of exposure. At higher doses most animals had significant toxicity and numerous proliferative lesions including proliferations of hepatocytes, biliary epithelium and oval cells. This manuscript focuses on the hepatocellular proliferative lesions, and the difficulty of categorizing many of them.

The ultimate objective of toxicity/carcinogenicity findings in rodent studies is identification of potential hazards to human health, but the certainty of the response in the rodent is a crucial first step. Classification of lesions is most precise when morphology can be correlated with biological behavior. However, biological information is often lacking and pathologists are left with only the morphology. In the livers of treated animals from this set of studies, there was a plethora of nodular lesions with a spectrum of morphologies and a paucity of biological information. Contemporary classification schemes generally include regenerative hyperplasia, hepatocellular adenoma, and hepatocellular carcinoma as possible diagnoses for nodular proliferative hepatocellular lesions.

Generally accepted criteria for the spectrum of proliferative lesions of the liver are well described (Greenblatt, 1982; Newberne, 1982; Eustis, 1990; Goodman et al., 1994; Narama et al., 2003). While there are several morphological features that define adenoma, key criteria that allow distinction from foci of hepatocellular alteration include larger size, loss of normal hepatic architecture including an absence of, or few, portal triads, distinct compression of adjacent liver, and angulation of hepatic cords where they intersect with the adjacent parenchyma. For regenerative hyperplasia, the morphological features tend to be intermediate between focus and adenoma. There may be more compression of the surrounding hepatic parenchyma than for a focus, but less than for adenoma. Typically, there is retention of hepatic architecture (including vasculature and bile ducts), but hepatic cords may be irregular and/or distorted by hypertrophic cells. Component cells are often eosinophilic and hypertrophic (Figure 2B). The criterion that may be most helpful in discerning regenerative hyperplasia from focus or adenoma is not a feature of the lesion itself, but rather alteration in the surrounding liver suggesting prior or ongoing liver damage, and therefore the likelihood of a regenerative response. This is the basis for use of the qualifier “regenerative” as opposed to the use of hyperplasia alone.

Metastasis is the most definitive criterion for identification of hepatocellular carcinoma, but there are other morphological features that appear to be harbingers for metastatic potential, and are routinely used to identify hepatocellular carcinomas. One such feature is thickened hepatocellular trabeculae (Eustis, 1990; Goodman et al., 1994). However, identification of hepatocellular carcinoma was not an issue in this group of studies, as lesions with morphological features of malignancy were rarely encountered despite the presence of multiple nodular lesions in many animals. Also, there were no incidences of distant metastasis of hepatocellular neoplasms.

In these studies, the earliest evaluation was at 14-weeks, and the most notable change involving the hepatocyte was hypertrophy. Treatment-related foci of cellular alteration were observed at the 31-week interim sacrifice and at later time points; they were larger and more numerous compared to foci in control groups. At 2 years, the most severely affected livers contained many nodular lesions and had the appearance of a cirrhotic liver (Figure 1E). This was most apparent in the study of PCB-118/126 (NTP TR 531) in which the highest relative doses were given.

Many lesions in these studies had classic morphological features that allowed straightforward classification as hepatocellular adenoma or focus of hepatocellular alteration. Many other lesions conformed to a category intermediate between focus and adenoma. There was significant variation in the size of these lesions, with variable compression of adjacent parenchyma. They were comprised of enlarged eosinophilic to vacuolated cells and the architecture was disorganized and/or altered. While these features are most consistent with regenerative hyperplasia, it was the overall appearance of the liver that gave confidence that most nodular lesions in the higher dose groups were regenerative hyperplasia and not adenoma (Figure 1E).

Some lesions within these moderately to severe toxic livers appeared intermediate between adenoma and regenerative hyperplasia, and were therefore difficult to categorize. Similar lesions occurred in animals with less toxic change, generally from lower dose groups; in these animals, the role of regeneration is not clear. A feature of many of these lesions that does not appear to be well described in the literature was the biliary and/or oval cell component (Figures 2C and 2D). Proliferations of these cell types were also present in the adjacent liver. Their presence throughout suggests that the proliferative lesions were not neoplastic. However, the distribution and amount of biliary and oval cells was not uniform in a lesion. In some, there were areas devoid of these cells, and instead composed of a monotonous population of hepatocytes; if these areas were large, the lesion was diagnosed as adenoma.

The primary questions were whether many of the proliferative liver lesions in this group of studies were neoplastic or nonneoplastic, and if they were nonneoplastic, what was the most appropriate categorization? The rare occurrence of hepatocellular carcinomas in the face of a marked proliferative response shows a lack of progression to malignancy. This supports the hypothesis that most of the nodular lesions were nonneoplastic since it is known that hepatocellular adenoma may progress to carcinoma in rodents. Nodular lesions were generally not observed at the end of 2 years in the stop groups, which might indicate a regression of the lesions. However, they were not present in animals evaluated at 31 weeks either (the point at which chemical administration was stopped), so it may be that the stop exposure group was not exposed long enough for nodular lesions to develop. Only in the PCB-126/118 study, where higher relative doses of chemical were administered, were nodular lesions observed in animals at 31 weeks. None were considered neoplastic at 31 weeks and hepatocellular adenomas were not increased in this group at 2 years. Interpretation of stop study data is potentially confounded since these chemicals accumulate in tissues and are cleared slowly, resulting in continued exposure, albeit exposure would certainly be less than with sustained administration.

It is clear that in this group of studies hepatocellular damage and regeneration were present. Regeneration may be associated with cirrhosis which is defined by the WHO as “a diffuse process characterized by fibrosis and the conversion of normal liver architecture into structurally abnormal nodules” (Anthony et al., 1978). Although rodent livers may develop a nodular appearance, fibrosis is not as pronounced as in other species such as dogs and primates, and was not a prominent feature in these studies. Administration of a number of xenobiotics to laboratory animals can result in a nodular liver, and some of these have been used to develop models of the human disease (Nuber et al., 1980). It has been reported (Rogers and Newberne, 1973) that dietary manipulation can produce nodular livers in rodents. Regenerative nodules in cirrhosis may occasionally grow quite large and appear tumor-like, and have been called macroregenerative nodules, adenomatous hyperplasia, and dysplastic nodules. Some have been reported to contain nodules within nodules and foci of malignant change and to develop into hepatocellular carcinoma over time (Wada et al., 1988; Eguchi et al., 1992; Kaji et al., 1994; Nakanuma et al., 1993, 1998; Hytiroglou et al., 1995; Hytiroglou and Theise, 1998).

When associated with partial hepatectomy, regeneration is a regulated process that results in re-establishment of the appropriate ratio between liver mass and body size (Fausto, 1999). It appears that restoration of the removed lobes does not occur, but rather the remaining liver is enlarged (Fausto and Mead, 1989). Partial hepatectomy in rodents closely reflects the capacity of the human liver to regenerate after partial hepatectomy (Nagasue et al., 1987). In humans a transplanted liver will increase or decrease in size until the organ reaches the optimal mass required for that person (Kam et al., 1987; Van Thiel et al., 1987).

An important criterion for regenerative hyperplasia is evidence of significant previous and/or ongoing hepatic degeneration/necrosis. However, in this group of studies, particularly in low-dose groups, there were some livers with nodular lesions, but with minimal-to-mild liver damage. In these instances, it may be inappropriate to imply a pathogenesis by diagnosing regenerative hyperplasia. There was no evidence of necrosis or severe toxicity at 14, 31, and 53 weeks in dose groups in which proliferative lesions were later observed. In other studies of TCDD using comparable doses, minimal-to-mild hypertrophy and minimal fatty change and pigmentation were the primary toxic changes observed at 5 and/or 15 weeks (Walker et al., 1998; Wyde et al., 2001). Although unlikely in these studies, there are potential mitigating factors. For example, some significant changes may occur in the liver, yet not be readily identifiable via routine histopathological evaluation. Apoptosis can develop as a response to a wide range of changes (Alison and Sarraf, 1995; Wyllie et al., 1999). It can be completed very quickly so that there is a relatively low probability of identifying apoptotic cells in static tissue sections. It has been calculated that a tissue may lose half its cell complement in 2–3 days but never show more than 5% of its cells in apoptosis at the same time (Wyllie et al., 1999). It is also possible that significant damage to the livers occurred very early and resolved, and therefore was not detected at the times of evaluation in these studies. However, significant hepatotoxicity observed later in the studies suggests that an early insult with adaptation likely did not occur.

There is other evidence that dioxin-like compounds may result in a proliferative stimulus of the liver that is separate and apart from degeneration and subsequent regeneration. Increases in cell proliferation were recorded at the interim sacrifices particularly at 31 and 53 weeks. Hepatocellular hypertrophy was significant from 14 weeks onward. Treatment related increases in foci of hepatocellular alteration were present at 31 weeks and pronounced by 53 weeks. Multiple, large foci clearly exceeding that observed in controls were often present in livers with minimal evidence of toxicity (Figures 1B–D). TCDD is also reported to be a liver tumor promoter (Pitot et al., 1980; Maronpot et al., 1993), and an inhibitor of apoptosis (Stinchcombe et al., 1995).

The use of the term “regenerative” implies a pathogenesis. In the absence of information clearly indicating a past or ongoing toxic/necrotizing event, one cannot know for certain if the hyperplasia is regenerative. In light of this, as well as the stimulatory role of Dioxin-like compounds in cell proliferation it seemed inappropriate to conclude that all nonneoplastic nodular lesions in these studies were secondary to degeneration/necrosis. Therefore, the more encompassing term, “hyperplasia,” was considered preferable for this group of lesions regardless of the potential pathogenesis. “Nodular” was used as a modifier because it described the gross morphology of the lesions. Historically, terms such as neoplastic nodule, nodular hyperplasia, hyperplastic nodule, and hepatoma have been used as all encompassing terms for proliferative nodular lesions of the liver, including neoplasms. In retrospect, because of this historical connotation, it may have been appropriate to use focal as a modifier rather than nodular, or as well as.

As mentioned, the presence of the biliary/oval cell component (Figures 2C and 2D) identified in this group of lesions is not well described in recent classification schemes, and its significance is uncertain. In humans, “multifocal nodular hyperplasia” is a term sometimes employed for the regenerative hyperplasia that occurs in cirrhosis. Nodules are variable in size but show an intimate relationship with bile ducts and pseudo bile ducts that can be found deep within their parenchyma. This feature serves to aid in the distinction of these nodules from adenomas, which are devoid of biliary structures and show more striking cellular pleomorphism (Edmundson, 1958). While the presence of a second cell type in these studies may suggest a lesion to be nonneoplastic, there are clearly instances of neoplasms with mixed cell types. There were proliferative lesions within these studies composed of both hepatocytes and biliary epithelium, and diagnosed as hepatocholangiomas. It is possible that the proliferative nodules resulted from a progenitor cell that has the ability to differentiate into biliary epithelium and/or hepatocytes.

Proliferative hepatocellular lesions from other NTP studies were evaluated to determine if biliary/oval cell proliferations occurred and if they were specific for hyperplasia. While the vast majority of proliferative lesions that were examined fit the classical criteria, and did not include a prominent biliary/oval cell component, there were instances where this component was present. For example, there were lesions diagnosed as focus of hepatocellular alteration or adenoma in which the biliary/oval cell component was present, yet it was not present in the adjacent liver, and instances where it was present in the lesion as well as within the adjacent liver. There were also adenomas without biliary/oval cell proliferations where such proliferations were present in the adjacent liver. There were some very large lesions diagnosed as adenoma in which some areas contained the proliferations while other areas did not. A few unequivocal carcinomas contained a component of spindled cells reminiscent of biliary/oval cells that were present within the neoplasms. Biliary epithelium/oval cell proliferations appear to occur under a variety of circumstances, and their usefulness in differentiating nonneoplastic and neoplastic nodular lesions is not completely clear.

In conclusion, it may be inappropriate to use regenerative as a qualifier for hyperplasia in the liver if evidence for such a pathogenesis is lacking. Intuitively, the presence of a second cell type such as biliary and/or oval cells suggests that a proliferative, nodular, hepatocellular lesion is nonneoplastic; however, this is unclear. Studies designed to characterize the pathogenesis are needed to fully understand and most accurately categorize the spectrum of lesions observed in this group of studies with dioxin and dioxin-like compounds.