ToxSci Advance Access originally published online on May 22, 2007
Toxicological Sciences 2007 99(1):3-19; doi:10.1093/toxsci/kfm098
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
A Review of Large Granular Lymphocytic Leukemia in Fischer 344 Rats as an Initial Step Toward Evaluating the Implication of the Endpoint to Human Cancer Risk Assessment
* Toxicology & Environmental Research and Consulting, The Dow Chemical Company, Midland, Michigan 48674
JK Haseman Consulting, 1054 Tacketts Pond Drive, Raleigh, North Carolina, 27614
Department of Pharmacology & Toxicology, Michigan State University, East Lansing, Michigan, 48824
10513 Wayridge Drive, Montgomery Village, Maryland, 20886
¶ Penn State Cancer Institute, Penn State College of Medicine, Hershey, Pennsylvania, 17033
1 To whom correspondence should be addressed at Veterinary Pathologist, The Dow Chemical Company, Midland, MI 48674. E-mail: jthomas4{at}dow.com.
Received February 9, 2007; accepted April 24, 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONPATHOBIOLOGY OF LGLL IN...CELL OF ORIGIN OF...LYMPHOPROLIFERATIVE DISORDERS...INFERENCES ON LGLL IN...UTILITY OF INCREASED STATISTICAL...DISCUSSIONFundingREFERENCES |
|---|
Large granular lymphocyte leukemia (LGLL) is a common fatal disease in aging F344 rats. The current understanding of rat LGLL and a search for mechanistic data/correlations to human leukemia were examined with the goal of improving evaluation of the LGLL endpoint in cancer bioassays as it relates to human cancer risk assessments. The exact cell of origin of the F344 rat LGLL is not fully resolved, although natural killer (NK) cell characteristics were demonstrated in most, if not all cases. Similarities between rat LGLL and a rare human NK-LGLL exist, invalidating claims of no human counterpart, although the underlying etiopathogenesis may be different. There is insufficient data to establish a mode of action of chemical-induced rat LGLL. Evaluation of the National Toxicology Program database revealed only 34 substances (out of over 500 studied) that were possibly associated with increased incidences of LGLL. Of these, only five produced definitive LGLL effects in both sexes; the remaining 29 produced single sex responses and/or only "equivocal" associations with LGLL. Trends of increasing background/variability in LGLL incidence and its modulation by extraneous factors (e.g., corn oil gavage) are key confounders in interpretation. Given that LGLL is a common tumor in control F344 rats, interpretations of bioassays can be improved by increasing the statistical stringency (e.g., p < 0.01 over traditional p < 0.05), as an indicator of possible carcinogenic effects, but that alone would be insufficient evidence for declaring treatment-related increases. Thus, it was concluded that the evaluation of possible chemically related increases in rat LGLL utilize a "weight-of-evidence" approach.
Key Words: F344 rat; Large granular lymphocyte; Leukemia; LGLL and Human cancer risk assessment.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONPATHOBIOLOGY OF LGLL IN...CELL OF ORIGIN OF...LYMPHOPROLIFERATIVE DISORDERS...INFERENCES ON LGLL IN...UTILITY OF INCREASED STATISTICAL...DISCUSSIONFundingREFERENCES |
|---|
The Fischer 344 (F344) rat is the most commonly used rat in carcinogenesis bioassays in the United States, particularly by the National Toxicology Program (NTP) and the chemical industry sector. It is also frequently used in Japan (Haseman et al., 1994
; Ward et al., 1990). The F344 rat was adopted as the strain of choice for carcinogenicity testing by the National Cancer Institute of United States (NCI) in 1970 and subsequently by the NTP chiefly because of its genetically in-bred status, relatively small size, ease of maintenance and sensitivity to various carcinogens, and its relatively low spontaneous tumor rate except for testicular Leydig cell tumors (Rao and Boorman, 1990). One of the main challenges in the sound scientific interpretation of cancer bioassays with the F344 rat has been the elevated, variable, increasing trend in the incidence of leukemia in the untreated aging control rats. Marginal, but statistically identified elevations in the incidence of leukemia associated with treatment with a test article often is a cause of concern to the scientific community and the regulatory agencies as to its interpretation of relevance to human cancer risk, mainly because of the paucity in mechanistic data and the inherent variability in the background occurrence. The purpose of this article is to briefly review the biology of the Fischer rat leukemia, highlight the issues concerning its toxicological significance, its relevance in human cancer risk assessment, and finally concluding with recommendations to adopt a "weight-of-evidence" approach for the evaluation of the bioassays and classification of the compounds in question. |
PATHOBIOLOGY OF LGLL IN FISCHER 344 RATS |
|---|
|
TOPABSTRACTINTRODUCTIONPATHOBIOLOGY OF LGLL IN...CELL OF ORIGIN OF...LYMPHOPROLIFERATIVE DISORDERS...INFERENCES ON LGLL IN...UTILITY OF INCREASED STATISTICAL...DISCUSSIONFundingREFERENCES |
|---|
A transplantable acute leukemia with characteristics of a monocytic leukemia was first described in a female F344 rat by Dunning and Curtis (1957)
. Subsequently, more detailed descriptions of leukemias of a unique mononuclear cell type were reported in Fischer 344 rats (Moloney et al., 1970). The authors described characteristic mononuclear cells with distinctive reddish cytoplasmic granules in the peripheral blood. These authors preferred the use of a nonspecific term "mononuclear cell leukemia" (MCL) because of the inability to classify as granulocytic, lymphocytic or monocytic based on morphological and histochemical grounds. The same authors had also reported earlier a similar leukemia in Wistar-Furth rats (Moloney et al., 1969). In the subsequent years, the spontaneous occurrence of leukemia in aging Fischer 344 rats was studied and characterized by a number of investigators (reviewed in Ishmael and Dugard, 2006; Stromberg, 1990; Ward et al., 1990). Ward and Reynolds (1983) based on morphologic similarities of the neoplastic cells to those of normal large granular lymphocytes (LGLs) described the leukemia as one of LGLs and named it large granular lymphocyte leukemia (LGLL). The NTP, however, currently describes it as MCL. To avoid confusion and to relate to a comparable human leukemia, the term LGLL will be used to describe this disease in the current review.
LGLL is one of the most common causes of moribundity and death in aging Fischer 344 rats (Ward et al., 1990) and is generally observed in rats over 18 months of age (Stromberg, 1985) with the incidence dramatically increasing through 27 months (Stromberg, 1983c). The pathology of spontaneous LGLL in Fischer 344 rats has been well characterized and well described (Stromberg, 1990; Stromberg and Vogtsberger, 1983; Stromberg et al., 1983b,c; Stefanski et al., 1990; Ward and Reynolds, 1983; Ward et al., 1990). Clinically, affected rats appear less active, with body weight loss, pale eyes, and ear pinnae, sometimes with yellowish discoloration of ear pinnae, and tail, with or without perineal soiling with dark urine. A gentle palpation of the abdominal area usually reveals an enlarged spleen. Death usually ensues within about 2 weeks after the clinical onset.
Splenomegaly is consistently seen in most advanced cases upon gross pathological examination. Other gross lesions of variable severity and incidence include enlarged and friable livers with light brown-yellow discoloration, and with irregular and roughened surface. Depending on the severity of leukemic cell infiltration, discoloration and/or hemorrhages may be seen in other organs. Leukemic cell infiltration of the splenic red pulp with variable lymphoid cell depletion is consistently seen in most moderate to advanced cases. Several organs are variably infiltrated but liver and lungs are frequently involved. The LGLL cells seen in imprints of spleen or blood smears, or cytocentrifuge preparations of splenic cell suspensions appear to be composed of a pleomorphic cell population, about 10–15 µm in diameter. The morphology of these leukemic cells can be variable. Well-differentiated tumor cells resemble normal LGLs with reniform nuclei and cytoplasmic granules. Intracytoplasmic granules are variable in size and number. Poorly differentiated cases have small or large lymphocytes with few or no apparent granules, but the granules may be seen ultrastructurally (Ward et al., 1990). Plasmacytoid cases of LGLL were also described containing prominent Golgi zones, but with obvious granules by ultrastructural examination (Ward et al., 1990). Purified cytoplasmic granules from the neoplastic LGLs are cytolytic to erythrocytes, splenocytes, and a number of different lymphoid tumor cells (Henkart et al., 1984). Neoplastic LGLs phagocytose erythrocytes (Kusewitt et al., 1982; Stromberg, 1990; Stromberg and Vogtsberger, 1983; Ward and Reynolds, 1983; Ward et al., 1990) and thrombocytes (Stromberg et al., 1983a). Although numerous organs may be infiltrated with the neoplastic LGLs, the spleen, liver, and lung are mostly affected. Bone marrow infiltration is variable and appears to occur late relative to spleen involvement and was reported only in less than half of the rats with leukemia (Stromberg and Vogtsberger, 1983). Histological findings in general, include diffuse infiltration of the splenic red pulp sinusoids of variable density, infiltration of the hepatic sinusoids with variable centrilobular hepatocellular degeneration, and necrosis and atrophy of hepatic cords and infiltration of the alveolar septa of the lungs. In the early stages of the disease, diagnosis is difficult, however, splenic congestion and lymphocytic depletion of the splenic white pulp appear to be the most significant and consistent finding (Losco and Ward, 1984). A hemolytic anemia associated with spherocytosis, reticulocytosis, anisocytosis, and polychromasia, neutrophilia with left shift, thrombocytopenia with variable numbers of atypical mononuclear cells characterizes the hemogram, particularly in advanced cases. An immune-mediated pathogenesis for the anemia was indicated due to the finding of antierythrocyte immunoglobulin with the direct Coomb's test (Stromberg et al., 1983c). Increases in serum bilirubin, enzymes associated with hepatocellular injury, hemoglobinuria, and bilirubinuria, also are generally observed, particularly in late stages of the disease (Stromberg et al., 1983b).
LGLL is readily transplanted by inoculation of the tumor cells into syngeneic recipients (Dieter et al., 1989; Reynolds et al., 1984; Stromberg et al., 1985, 1990; Ward and Reynolds, 1983) yielding large numbers of neoplastic LGLs for further detailed studies. A publicly available source of a F344 LGL rat leukemia cell line, denoted as RNK-16, is available in the NCI tumor repository http://dtp.nci.nih.gov/branches/btb/services.html. The morphologic and clinicopathologic features of the transplanted tumors are generally similar to those observed in the spontaneous cases (Stromberg et al., 1985).
The cause of spontaneous LGLL in Fischer 344 rats is unknown. Transplantation of the tumor by inoculation of cell free lysates has been unsuccessful (Moloney et al., 1971; Stromberg et al., 1985). There is no definitive evidence of any viral etiology and no reverse transcriptase activity has been associated with this leukemia (Stromberg, 1990). The relatively high incidence of LGLL in the F344 rat strain is suggestive of an age related genetic basis as a probable cause for this disease.
|
CELL OF ORIGIN OF LGL LEUKEMIA |
|---|
|
TOPABSTRACTINTRODUCTIONPATHOBIOLOGY OF LGLL IN...CELL OF ORIGIN OF...LYMPHOPROLIFERATIVE DISORDERS...INFERENCES ON LGLL IN...UTILITY OF INCREASED STATISTICAL...DISCUSSIONFundingREFERENCES |
|---|
The universal involvement of spleen in early and late stages of the disease, with less and variable infiltration of the bone marrow (Stromberg and Vogtsberger, 1983
), and the experimental evidence that splenectomy at 1–2 months of age dramatically reduced LGLL incidence (Moloney and King, 1973; Moloney et al., 1971) supports the hypothesis that the cell of origin of LGLL may reside in the spleen and/or requires the splenic microenvironment for differentiation/neoplastic transformation. The observation that normal bone marrow contained few LGLs (Reynolds et al., 1981) is consistent with the view that LGL differentiation from precursor cells takes place outside the bone marrow in certain tissues, spleen being one of them. The central role of spleen is further supported by the observation that certain chemicals that caused splenic toxicity also reduced the incidence of LGLL (Elwell et al., 1996) probably by causing some unfavorable change in the splenic microenvironment for the development of LGLL. It is also known that the precursor cell is radiosensitive as X-irradiation at 1–2 months of age dramatically reduces the LGLL incidence later in life of the rat (Hellman et al., 1982; Moloney et al., 1971), presumably by eliminating the precursor cell, but whether this occurs at the level of the spleen, bone marrow, or elsewhere is not certain. While there is considerable information on the ultrastructural, immunologic, and biochemical characteristics of the LGL leukemic cell (Reynolds et al., 1984; Stefanski et al., 1990; Stromberg et al., 1983a; Ward and Reynolds, 1983) and its resemblance to normal LGLs, the exact cell of origin of the leukemic cell is still not fully confirmed. In classifying rodent hematopoietic neoplasms, Frith et al. (1993), categorized LGLL as a "non-T, non-B, or null cell" leukemia.
In the initial studies aimed at characterizing normal LGL population in rats in the early 1980's, it was discovered that most, if not all, natural killer (NK)–cell activity was associated with the LGL population (Reynolds et al., 1981). They reacted strongly with the NK-associated asialo GM1 antigen, expressed receptors for the Fc portion of IgG (Fc
R), exhibited antibody-dependent cellular cytotoxicity, and certain T-cell associated antigens recognized by the BC-84, W3/13, and OX-8 monoclonal antibodies, but not the T-helper phenotype W3/25. They were also reported to be weakly Ia positive and surface immunoglobulin (sIg) negative and therefore unrelated to a B-cell lineage. Evidence that they were not mature 
ß T cells came from the findings that freshly isolated LGL (Young et al., 1986) and neoplastic LGL from four rat LGL leukemia lines (Reynolds et al., 1985) did not express functional beta chain of the T-cell receptor suggesting that LGLs were derived from a lineage distinct from T cells or developed before a functional rearrangement of the T-cell receptor beta chain. Although the LGLs bear a superficial resemblance to monocytes, their failure to stain for esterase, nonadherence to plastic or nylon wool, and their nonphagocytic nature distinguished them from typical macrophages (Reynolds et al., 1981). Collectively, these findings indicated that LGL belonged to a distinct and separate lineage from both normal T or B lymphocytes and from normal macrophages or granulocytes (Reynolds and Sayers, 1990).
Ward and Reynolds (1983) studied the morphology, histochemistry, cell surface antigens, and NK-cell activity of 10 primary and 10 transplanted LGLL of aging F344 rats. The morphological features of the neoplastic LGLs closely resembled those of normal LGLs (therefore, the name LGL leukemia, LGLL). Although there were some morphologic variations in some cases, the use of antibodies against OX-8 and LGL granules, and the ultrastructural demonstration of cytoplasmic granules facilitated the identity of the neoplastic LGL (Ward et al., 1990). The LGLL cells were shown to have NK-cell activity against YAC-1 cells although it was variable from case to case and it appeared that the cytotoxic activity of individual tumors was generally associated with the differentiation of the tumor, the most poorly differentiated one generally showing the lowest cytotoxicity (Ward and Reynolds, 1983). Fluorescence-activated cell sorter analysis of surface antigens revealed the LGL leukemias to be heterogeneous, and there was no correlation between cytotoxic activity and cell surface antigens. Neoplastic cells were variably positive when stained with OX-8, Thy 1.1, M1/70, BC-84, W3/13, OX-1, and OX-7 antibodies, had Fc receptors, but were largely negative for Ia, sIg, and the T-helper phenotype W3/25. Thus, although the morphologic features of cells in LGLL largely resembled those of normal rat LGLs, differences in cytotoxic activity and surface antigens suggested that LGL tumors represented a heterogeneous group of leukemias. However, these investigators distinguished the LGLL cells from monocytes because they did not contain peroxidase or lysozyme and nonspecific esterase and were nonadherent in cell culture (Reynolds et al., 1984; Ward and Reynolds, 1983). LGLL cells exhibited some properties distinct from their normal counterparts such as the expression of Thy 1.1 antigen on most tumor cells indicating a relatively immature bone marrow precursor phenotype of the neoplastic cells and decreased or absent expression of the leukocyte common (L-C) antigen (OX-1) present on most, if not all, rat leukocytes (Reynolds et al., 1984; Ward and Reynolds, 1983). Other investigators (Stromberg et al., 1983a), characterizing leukemic cells from F344 rats with spontaneous leukemia, reported that more than half of the leukemic cells adhered rapidly to glass, and more than 97% showed consistent, strongly positive reaction for esterase that was sensitive to NaF suggestive of having some characteristics in common with macrophages. In this report, the surface markers, adherence and phagocytosis assays, provided support for the identification of the leukemia cell the as a non-T, non-B, adherent, esterase-positive cells with moderate Fc receptor activity and came to a general conclusion that the tumor cells had characteristics of both monocytes and lymphocytes. A relationship to a macrophage lineage was also suggested by other investigators based on the observations of erythrophagocytosis (Kusewitt et al., 1982) and the expression of an oncogene homologous with the csf-1 receptor (found on macrophages and their precursors) by the leukemic cells (Stefanski et al., 1990). These reports of heterogeneity of the cell phenotype and function may reflect neoplastic transformation process of mature LGLs or may reflect the malignant transformation of variety of individual subpopulations within the LGL population (Reynolds, 1985). Alternatively, it may be that a pluripotent LGL precursor cell could be the target of neoplastic transformation. Whether it is the mature LGL or its precursor is the target cell for neoplastic transformation should be an area of investigation, which may be linked to the observation of the heterogeneity in LGLL cells (Ward and Reynolds, 1983) and the differences in esterase staining and adherence characteristics reported by Stromberg et al. (1983a) and Ward and Reynolds (1983). Whatever may be the case, the available evidence of NK-cell cytotoxicity in most spontaneous and transplanted cases of F344 leukemia (Reynolds et al., 1984; Ward and Reynolds, 1983) suggests that F344 rat LGLL is an NK-cell leukemia from a functional perspective, although no further characterization of spontaneous or chemical-related cases has been reported. The use of newer and specific monoclonal anti-rat NK-cell antibodies such as 3.2.3 (Chambers et al., 1989), 10/78 (Kraus et al., 1996), and WEN23 (Westgard et al., 2004) and other rat leukocyte specific markers along with demonstrating functional NK-cell activity of more numbers of spontaneous or chemical-related LGLLs should help in firmly establishing the cell of origin.
|
LYMPHOPROLIFERATIVE DISORDERS INVOLVING LGL IN HUMANS |
|---|
|
TOPABSTRACTINTRODUCTIONPATHOBIOLOGY OF LGLL IN...CELL OF ORIGIN OF...LYMPHOPROLIFERATIVE DISORDERS...INFERENCES ON LGLL IN...UTILITY OF INCREASED STATISTICAL...DISCUSSIONFundingREFERENCES |
|---|
In humans, LGLs are a morphologically distinct lymphoid subset comprising 10–15% of normal peripheral blood mononuclear cells. Proliferative disorders involving LGLs have been reviewed elsewhere (Cheung et al., 2003
; Greer et al., 2001; Lamy and Loughran, 1999; Loughran, 1993; Rose and Berliner, 2004; Sokol and Loughran, 2006). Briefly, human LGLs can be classified into two major groups based on their lineages as CD3+ LGL and CD3– LGL. CD3+ LGLs are T cells that express the CD3/TCR complex and rearrange the TCR genes and are thought to represent in vivo activated cytotoxic T-lymphocytes. CD3– LGLs, on the other hand, are NK cells, do not express CD3/TCR complex, do not rearrange the TCR genes, and exhibit a non-MHC (major histocompatibility complex) restricted cytotoxicity.
A chronic lymphoproliferative disorder with unusual clinical, morphologic, ultrastructural, and membrane surface marker characteristics was first reported in 1977 (McKenna et al., 1977) in four patients. Although the term LGL was not used then, the description of the proliferating lymphocytes was consistent with what we now know as LGL. Since then, many studies have been published on LGL proliferative disorders and several different terms were utilized (Loughran, 1993; Rose and Berliner, 2004). Also, there was confusion if the proliferation represented a reactive or truly a neoplastic disorder. The observations of clonality, and invasion of bone marrow, spleen, and liver, led Loughran et al. (1985) to propose the term "LGL leukemia." Subsequently, based on the lineage and clonality, LGL leukemias were proposed to be further classified into T-LGL leukemia and NK-LGL leukemia (Loughran, 1993). Presently, in the recent World Health Organization classification, the terms T-cell LGL leukemia and aggressive NK-cell leukemia are used (Jaffe et al., 2001).
LGL leukemia comprises 2–5% of all T-cell/NK-cell malignancies, with only 400 cases reported in the literature (Sokol and Loughran, 2006 and reference cited therein). The clinical manifestations of the human LGL leukemia have been well reviewed elsewhere (Lamy and Loughran, 2003; Rose and Berliner, 2004; Sokol and Loughran, 2006). Greater than 85% of human LGL leukemia cases are of the CD3+/T-cell type and less than 15% of LGL proliferative diseases are of the CD3–/NK lineage. In general, majority of the CD3+/T-cell LGL leukemia cases are indolent although, aggressive variants are also described in the literature (Sokol and Loughran, 2006). CD3+/T-cell LGL leukemia is generally a disease of the elderly with a median age of 60 years with no specific predilection for either men or women. Surface marker expression of these cells reveal in majority of the cases CD3+, TCRß+, CD4–, CD8+, CD16+, CD 27–, CD45RO–, CD57+, CD94+ phenotype. Less frequently a CD3+/TCR
+ and CD3+/CD56+ phenotypes have been reported indicating some variations within the T-cell lineage. Neutropenia with recurrent bacterial infections is common. About half the patients are found to have modest splenomegaly and about 25% have hepatomegaly. Lymphadenopathy is rare. Neuropathy and hemophagocytic syndrome may very occasionally occur in T-LGL leukemia. Most patients present with lymphocytosis with LGL counts between 2 x 109 and 10 x 109/l, anemia and some have moderate thrombocytopenia. Coomb's positive hemolytic anemia has been described in a few cases. Majority of the patients have bone marrow infiltration, sinusoidal, and portal infiltrates in the liver and a characteristic pattern of red pulp infiltration of the spleen often with reactive follicular hyperplasia of the germinal centers of the white pulp. A number of autoimmune diseases, particularly rheumatoid arthritis are associated with this type of LGL leukemia. Although the exact cause of T-LGL leukemia is unknown, and no prototypical Human T-cell leukemia virus type I or II (HTLV-I or II) infection was reported (Loughran et al., 1994), there is evidence that a good proportion of the patients have seroreactivity against certain peptides of HTLV-I/II, specifically to a decapeptide PP-10 of the p21 env protein (Sokol et al., 2005). These findings coupled with the fact the neoplastic LGLs show all the characteristics of antigen activated T cells (Lamy and Loughran, 1999), raise a possibility of a variant HTLV-I/II viral infection as a possible cause in the pathogenesis of T-LGL leukemia although cross-reactivity to some human endogenous peptide cannot be fully ruled out.
In contrast to T-LGL leukemia, which in general, is characterized by a chronic indolent course whose major morbidity is related to neutropenia and anemia, the NK-LGL leukemia, generally has an aggressive clinical course (Burks and Loughran, 2005; Lamy and Loughran, 2003; Sokol and Loughran, 2006) and the outcome is generally poor. Despite aggressive multiagent chemotherapy, Loughran (1993) reported deaths in 9 of 11 patients within 1–2 months of diagnosis. Multiorgan failure associated with coagulopathy and hemophagocytic syndrome was the main cause of death. Anemia and thrombocytopenia are more common and pronounced in NK-LGL leukemia. Absolute LGL counts are higher than those in T-LGL leukemia with many patients reaching more than 10 x 109/l. Massive hepatosplenomegaly, involvement of the gastrointestinal system with jaundice, and ascites are often present. The majority have massive bone marrow infiltration sometimes with bone marrow fibrosis. Lymph node involvement is observed more often in NK-LGL leukemia than in T-cell LGL leukemia. In general, patients are younger, with a median age of 39 years with an equal male/female distribution. The usual phenotype is CD3–, TCRß–, TCR–, CD4–, CD8+, CD16+, CD56+ with variable expression of CD57. Most cases are described from Asia and are associated with clonal cytogenetic abnormalities with Epstein Barr virus (EBV) infection implicated in many of these cases (Chou et al., 1998; Kawa-Ha et al., 1989; Takechi et al., 2002). EBV RNA or DNA could be detected in LGL in some of the patients with in situ hybridization or Southern blot analysis (Chou et al., 1998; Kawa-Ha et al., 1989). Similar cases implicating EBV involvement in NK-LGL leukemia have been reported from New Zealand (Hart et al., 1992; Ruskova et al., 2004) and from the United States (Gelb et al., 1994). The exact mechanism(s) by which EBV is involved in the potential transformation of NK-LGL remains to be elucidated, but may likely be similar to pathways associated with other types of lymphomas in humans.
From the preceding sections it is apparent that human NK-LGLL and the F344 rat LGLL have some characteristics in common. A comparison between the human NK-LGLL and the rat LGLL is summarized in Table 1.
|
|
INFERENCES ON LGLL IN F344 RATS FROM NTP/NCI RODENT CARCINOGENCITY STUDIES |
|---|
|
TOPABSTRACTINTRODUCTIONPATHOBIOLOGY OF LGLL IN...CELL OF ORIGIN OF...LYMPHOPROLIFERATIVE DISORDERS...INFERENCES ON LGLL IN...UTILITY OF INCREASED STATISTICAL...DISCUSSIONFundingREFERENCES |
|---|
Leukemia in Fischer 344 Rats and its Spontaneous Incidence
Table 2 documents how the background incidence of LGLL in F344 rats has steadily increased over time. In addition, Haseman et al. (1985)
reported a mean background LGLL incidence 26.5% (range 10–46%) in untreated male F344 rats and 17.3% (range 6–38%) in untreated females. Although, the reasons for the increased LGLL prevalence over time and the high study-to-study variability are unknown, Rao et al. (1990) suggested that changes in diagnostic criteria and a combination of genetic and experimental variables over at least 30 generations may have contributed to the time-related increase in LGLL. Haseman et al. (1998) summarized the spontaneous incidences of neoplasia from 27 feeding studies and 18 inhalation studies whose pathology evaluation had been finalized as of 1 January 1997. In this study, the range of LGLL in untreated male F344 rats varied from 32% to 74% with an average of 50.5% in the feeding studies and 34% to 70% with an average of 57.5% in the inhalation studies. In the females, it varied from 14% to 52% with an average of 28.1% in the feeding studies and 24% to 54% with an average of 37.3% in the inhalation studies.
|
The latest background incidence of LGLL as of writing this review was reported again by Haseman et al. (2003)
and is summarized in Table 2. In this study, for the feed studies, these authors reported control LGLL incidences of 59% and 32% for males and females, respectively, given one diet (NIH-07) and 52% and 24% in males and females, respectively, given another diet (NTP-2000). There was a slight reduction of LGLL incidence in the control males from the inhalation studies with an average of 46% (NIH-07) and 47% (NTP-2000), but the incidences in the females were 36% and 35% given NIH-07 and NTP-2000, respectively. These findings illustrate that the background incidence of LGLL in F344 rats has been highly variable and has more than doubled during the two decades since the report of Haseman et al. (1985), the reason(s) for which is (are) not well understood.
Factors Influencing the Expression of LGLL in F344 Rats
A number of factors are known to modulate the expression of LGLL in F344 rats of which the effect of oil vehicles for oral gavage is particularly significant. The effect of corn oil gavage on the incidence of LGLL in control rats was first reported by Haseman et al. (1985), while reviewing all NTP carcinogenicity studies completed between 1 January 1979 and 1 August 1984. Among other findings, it was discovered that there was a statistically significant (p < 0.001) decrease in LGLL in control male rats gavaged with corn oil compared to the untreated controls (13.8% vs. 26.5%, respectively). Surprisingly, this effect was not observed in the females. In a subsequent study, Haseman and Rao (1992) evaluated 31 untreated control groups, 43 corn oil gavage control groups, and 14 water gavage control groups from NTP studies. They confirmed that corn oil reduced the background incidence of LGLL in male F344 rats relative to the untreated controls (21.4% vs. 48.9%, respectively) and not the gavage technique per se and also confirmed that LGLL incidence in females was not affected by corn oil administration.
This modulation of LGLL expression in males was not specific to corn oil (an oil containing high levels of polyunsaturated and monounsaturated fats), but also to other oils such as safflower oil (an oil very high in polyunsaturated fat), or tricaprylin, an oil containing saturated medium chain fatty acids (NTP, 1994). The mechanism(s) by which corn oil modulates the expression of spontaneous LGLL in males is (are) not clear. Hursting et al. (1994), using a transplant leukemia model, reported that corn oil may decrease the development of leukemia by slowing leukemic cell proliferation mediated at least in part, by altered levels of diffusible factors such as growth hormone, and/or by enhancing the immune response. In this study, how exactly corn oil administration mediated these effects were not clear, but the authors, based on the observation of reduced feed consumption in the corn oil gavage group (with no changes in body weights or caloric intake compared to controls) hypothesized that reduced consumption of a particular nutrient or combination of nutrients may have played a role.
However, the effect of feed restriction on the outcome of spontaneous LGLL is quite variable. Some have reported that diet restriction over a 2-year period lowers the incidence of LGLL (Stefanski et al., 1990). Diet restriction was also reported to lower the incidence of transplanted leukemia with longer latency and decreased severity in male F344 rats compared to ad libitum fed rats 12 weeks after inoculation with leukemia cells (Hursting et al., 1993). In this transplant model, dietary restriction was shown to modulate LGLL expression through both its influence on leukemic cell proliferation via suppression of the growth hormone: insulin-like growth factor axis and its enhancement of host defenses against tumor cells. However, a recent NTP (1997) study did not show a uniformly decreased LGLL incidence with restricted feeding compared to ad libitum controls at the end of 2 years. Other reports indicate lifetime feed restriction appears only to delay the onset of leukemia but not progression (Shimokawa et al., 1993; Thurman et al., 1994). Whatever may be the case, the expression of LGLL appears amenable to manipulation of nutritional factors.
Other dietary factors, animal care, and housing protocols may also play a role in the expression of LGLL. In the Haseman et al. (2003) study, there was a modest decrement (but not statistically significant) in the incidence of LGLL in rats given NTP-2000 as compared to another diet NIH-07, the former containing less protein and more fat than the latter amongst other differences. Curiously, in this study, the incidence of LGLL in the control males from the inhalation study was reduced (statistically significant) relative to feed controls, the reasons of which, may be related to differences in animal care and housing protocols (Haseman et al., 2003). LGLL expression can also be modulated by irradiation, splenectomy, and by some chemicals. X-irradiation (Hellman et al., 1982; Moloney et al., 1971) and splenectomy (Moloney and King, 1973; Moloney et al., 1971) at 1–2 months of age also markedly reduced the incidence of LGLL, likely by eliminating the precursor cells.
Certain chemicals that induced spleen toxicity also decreased the LGLL incidence (Elwell et al., 1996) who reported that, of the 20 chemicals tested in the NTP's 2-year carcinogenicity bioassay that caused significant decreases in the LGLL incidence, 16 of them had previously caused spleen toxicity in 13-week studies. These decreased LGLL rates were often quite dramatic. For example, in the isobutyl nitrite study (NTP, 1996), LGLL rates in male F344 rats were 4%, 2%, and 2% in the low-, mid-, and high-dose groups, respectively, compared with 59% in controls. The corresponding rates in females were 2%, 0%, and 0% versus 30% in controls. For both sexes, survival in the dosed groups exceeded that seen in the controls, significantly so for mid- and high-dose males. This improved survival in the dosed groups reflected elimination of a leading cause of death in F344 rats, namely LGLL. How exactly spleen toxicity is related to reduction in LGLL expression is unknown, but an "altered microenvironment" in the spleen causing suppression of the eventual development of LGLL was proposed. However, it was also noted that not all chemicals causing spleen toxicity ended up with reductions in LGLL and thus additional factors are also believed to be involved.
In addition to the factors known to modulate LGLL incidence, we know that there are as yet undiscovered factors that affect LGLL, because the high variability in background incidence from study-to-study is not due to chance, nor is it due to the modulating factors identified thus far and discussed above.
NTP Studies Showing Evidence of Chemically Related Leukemia Effects in F344 Rats
The NTP has evaluated more than 500 chemicals for potential carcinogenicity in rats and/or mice. Of these, 34 showed evidence of chemically related increases in the incidence of leukemia. Overall NTP conclusions for each chemical are not presented or discussed.
The terminology used by the NTP to describe Fischer rat leukemia has evolved over the years. The current terminology is "mononuclear cell leukemia." In the earlier days of the testing program, a variety of other terms were used to describe this neoplasm, including monocytic leukemia, granulocytic leukemia, lymphocytic leukemia, leukemia (not otherwise specified), undifferentiated leukemia, and even malignant lymphoma. The interpretation of leukemia findings in these early studies was based on the combination of all these neoplasms, and these are the tumor rates reported in the Table 3 for these early studies (and simply referred to as "leukemia").
|
Just as the NTP's diagnostic nomenclatures for LGLL have changed over time, there has also been an evolution in the statistical methodology used to assess possible carcinogenic effects. The early NTP studies used Cochran–Armitage trend tests and Fisher's exact tests for pairwise comparisons. These procedures do not adjust for survival differences among groups. More recent NTP studies use survival-adjusted methods: life table tests (appropriate for fatal tumors) and incidental tumor test or logistic regression analyses (appropriate for tumors observed "incidentally" in an animal dying of an unrelated cause). Since the NTP does not require cause of death information, it is not possible to determine for an individual tumor its "context of observation" (i.e., fatal or incidental).
Consequently, the most recent NTP studies use a poly-3 test (Portier and Bailer, 1989) a survival-adjusted method that does not require information on the context of observation of individual tumors. The p-values reported in Table 3 reflect the statistical methodology that was in use at that time and are taken directly from the technical reports. In those studies that used survival-adjusted methods, p-values from the Cochran-Armitage and Fisher exact tests were often reported to maintain continuity with the earlier technical reports, even though interpretation of experimental results in such studies was based on the p-values from the survival-adjusted tests. For completeness, Table 3 includes p-values from all statistical tests whose results appear in the technical report.
The pattern of carcinogenicity observed in the 34 NTP studies showing leukemia effects in F344 rats is summarized in Table 4. Note that there were a total of 44 sex-species groups showing evidence of chemical-related leukemia effects. The "positive" studies are those judged by the NTP to provide definitive evidence of a carcinogenic effect. "Equivocal" outcomes are those in which it is judged to be uncertain if the increased incidences of leukemia are chemical-related.
|
For eight of the 34 NTP studies showing evidence of leukemia effects in F344 rats, equivocal leukemia effects (acetaminophen, chlorinated water, chloraminated water, diallyl phthalate, and phenol) or positive leukemia effects (2-amino-5-nitrothiazole, butyl benzyl phthalate, and dimethylmorpholinophosphoramidate) in male and/or female F344 rats provided the only evidence of carcinogenicity in the entire study (rats or mice). For an additional six chemicals (allyl isovalerate, bisphenol A, hydroquinone, pyridine, tetrachlorethylene, and 2,4,6-trichlorophenol), leukemia was the only neoplastic change for either male or female rats, but other sex-species groups showed evidence of carcinogenic effects other than leukemia. For the remaining 20 chemicals, leukemia was one of multiple neoplastic changes observed for male and/or female rats.
As can be seen in Table 4, there are only five chemicals that produced definitive leukemia effects in both sexes. More often, in contrast to most other target sites in NTP studies, the leukemia effect is specific to one sex or the other. Note also that there is a relatively high frequency of "equivocal" outcomes compared to the number of "positive" outcomes.
|
UTILITY OF INCREASED STATISTICAL STRINGENCY |
|---|
|
TOPABSTRACTINTRODUCTIONPATHOBIOLOGY OF LGLL IN...CELL OF ORIGIN OF...LYMPHOPROLIFERATIVE DISORDERS...INFERENCES ON LGLL IN...UTILITY OF INCREASED STATISTICAL...DISCUSSIONFundingREFERENCES |
|---|
The NTP uses no rigid statistical decision rule in its interpretation of experimental results, although the observed p-values for the trend test and pairwise comparisons are certainly important in the decision-making process. Other factors, consistent with a weight of evidence approach are used, such as the historical control tumor rate, whether or not the leukemia effect was seen in both sexes, whether or not it was dose-related, whether or not tumor latency was reduced, and in some cases even the stage of the leukemia (e.g., was it more advanced in the dosed groups?) (Dunnick et al., 1989
). A detailed review of the weight of evidence concept is beyond the scope of this review but this concept is discussed in greater detail in the U.S. Environmental Protection Agency Guidelines for Carcinogen Risk Assessment (USEPA, 2005). Briefly, unlike a rigid statistical decision rule, the weight of evidence approach takes into account all of the available evidence to arrive at an interpretation and "emphasizes the importance of weighing all of the evidence in reaching conclusions about the human carcinogenic potential of agents" (USEPA, 2005). This includes but is not limited to, the tumor findings in bioassays, the physical/chemical properties of the agent, any structure activity relationships to other carcinogenic agents, genotoxicity studies, and studies addressing possible carcinogenic modes of action. Together these data provide evidence or lack of evidence, for the biological plausibility for tumor induction. In other words, are the collective data consistent with the current understanding of biology?
Despite these additional factors, it is of interest to investigate the degree to which the statistical decision rule suggested in the Food and Drug Administration (FDA) Guidelines for a commonly occurring tumor (p < 0.005 for a trend test; p < 0.01 for pairwise comparisons; Lin and Rahman, 1998) would have reproduced the NTP outcomes (Table 5).
|
NTP does not routinely provide cause of death information, so it is impossible to determine which specific leukemias were fatal and which were incidental. However, leukemia is generally regarded as a fatal neoplasm, and is one of the leading causes of death in F344 rats (Haseman et al., 1994
), so when conflicting test results are obtained by the life table and incidental tumor/logistic regression analyses, the life table test is probably closer to reflecting the true statistical significance of the carcinogenic effect. Thus, when conflicting results were obtained by the two approaches, the p-values from the life table test were the ones compared to the statistical criteria from the FDA Guidelines. The "Yes*" outcomes in Table 5 are those that satisfy the statistical criteria of significance by the life table test, but not by the incidental tumor/logistic regression analyses. These conflicting results occur in studies showing chemically related increases in mortality, which the life table test attributes in part to increased incidences of (and shortened latency for) leukemia, while the incidental tumor/logistic regression analysis assumes that leukemia did not contribute to the shortened survival seen in the dosed animals.
Comparison of the NTP calls with application of the FDA Guidelines reveals approximately the same overall number of chemical-related leukemia effects in F344 rats, although some of the individual chemicals would have been redistributed among the "positive" and "equivocal" categories. For example, if the FDA trend test criteria were rigidly followed, then eight NTP positives would be negative or equivocal and seven NTP equivocals would be positive. For the FDA pairwise comparison criteria, the corresponding changes are seven from positive to equivocal/negative and eight from equivocal to positive. As noted above, the NTP evaluation appropriately considers a number of other factors in addition to the observed p-values in its decision-making process.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONPATHOBIOLOGY OF LGLL IN...CELL OF ORIGIN OF...LYMPHOPROLIFERATIVE DISORDERS...INFERENCES ON LGLL IN...UTILITY OF INCREASED STATISTICAL...DISCUSSIONFundingREFERENCES |
|---|
Among the laboratory strains of rat, spontaneous LGLL occurs at the highest incidence in the aging F344 rat. Aging Wistar-Furth rats have also been reported to develop spontaneous leukemia involving an unusual type of mononuclear cell containing distinct reddish granules (similar to the descriptions of LGLL in F344 rat) with an incidence of 15–22% (Moloney et al., 1969
; Ward et al., 1990). It is relatively uncommon in other strains and stocks of rats (Ward et al., 1990). The incidence of LGLL in the Sprague–Dawley rat was reported to be 0.6% (Frith, 1988) and more recently reported to be 0.5% in female Harlan Sprague–Dawley rats from 2-year studies performed by the NTP (Brix et al., 2005). Naturally occurring LGLLs have not yet been reported in mice to the best of our understanding, although mice transgenic for the tax gene of HTLV-I targeted to the mature T-lymphocyte compartment were reported to develop LGLL (Grossman et al., 1995). LGLL has also been reported in dogs (Ghernati et al., 2000; McDonough and Moore, 2000), in cats (Darbes et al., 1998; Roccabianca et al., 2006), and in a horse (Kramer et al., 1993).
Comparisons of the F344 rat LGLL with lymphoproliferative diseases involving LGL in humans have been reviewed in the past (Reynolds, 1985; Reynolds and Foon, 1984) and F344 rat leukemia was proposed to be a useful animal model for human T-cell leukemias (Stromberg, 1985). Although some heterogeneity in phenotype and function (NK-cell activity) was reported in F344 rat LGLL (Ward and Reynolds, 1983) and the exact cell of origin of F344 rat LGLL is still not fully resolved, the available evidence of NK-cell activity (Reynolds et al., 1984; Ward and Reynolds, 1983) of the neoplastic cells indicates that most of F344 rat LGLL, if not all, is of an NK-cell type. Thus, contrary to the reports that F344 LGLL does not have a human counterpart (Caldwell, 1999; Moloney et al., 1970), it is apparent from Table 1, that the F344 LGLL is quite comparable to the aggressive human NK-LGL leukemia on a morphological, functional, and clinical basis and is consistent with the view expressed in a recently published similar review article (Ishmael and Dugard, 2006).
It must be emphasized, however, that although there are multiple similarities between the F344 rat leukemia and human NK-LGL leukemia, the mechanisms of leukemogenesis may be very different. The facts that human NK-LGL leukemias are rare, seen in younger patients, reported mainly from the far-east with strong implications to EBV as the probable causative agent, contrast sharply with the high background incidence in aging F344 rats with no known etiology, viral, or otherwise. While research into molecular mechanisms of human LGLL has advanced over the years (Epling-Burnette et al., 2001, 2004a,b; Schade et al., 2006) very little is known about the cellular/molecular pathways of leukemogenesis in the F344 rat. Clearly, more research is needed to explain the predisposition of this strain of rat to LGLL, fully characterize the leukemic cell, and to define candidate molecular targets to understand the process of leukemogenesis which should hopefully explain how it is modulated by administration of exogenous agents. Needless to say, more mechanistic information is needed for arriving at scientifically sound conclusions as to its relevance in human cancer risk assessments.
Evaluation of the NTP database of rodent carcinogenicity studies confirms a time-related increase in the incidence of F344 rat leukemia in untreated animals. These increases are quite striking, and the reasons for these changes in tumor incidence are unknown. Possible contributing factors included changes in the diagnostic criteria and/or increased LGLL incidences related to associated increases in the body weights of F344 rats over time. However, it is unlikely that these factors alone could account for the marked increases in LGLL incidence seen.
It was also noted that F344 leukemia was more likely to be regarded as an equivocal effect by the NTP than other site-specific tumors. For example, of the 44 individual sex-species groups identified by the NTP as possibly showing chemically related leukemia effects, 43.2% (19/44) were considered equivocal. The relative percentage of equivocals versus positives is much higher than seen for other tumors evaluated by the NTP (which showed a range of 12–23% equivocals relative to positives; Table 6). There are likely two possible reasons for this disparity: (1) the high and variable incidence of LGLL makes a definitive interpretation more difficult; and (2) the increased incidences of LGLL are generally relatively modest, as can be seen from the data reported in Table 3. In fact, considered collectively, the chemically related increased incidences of LGLL in NTP studies are much less impressive than the chemically related decreased incidences of LGLL in NTP studies as reported by Elwell et al. (1996).
|
It is also noted that LGLL effects were more often than not confined to one sex, whereas for most tumors evaluated in NTP studies, similar effects are frequently seen in males and females (with of course, the exception of reproductive system neoplasms). Since there are modulating factors known to affect leukemia in one sex only (corn oil), and the majority of carcinogenic effects appear to be sex-specific for this neoplasm, it seems plausible to speculate that there may also be as yet unidentified factors/modes of action that are unique to one sex or the other for inducing LGLL in the F344 rat.
A more stringent statistical criterion, i.e., p < 0.01 rather than 0.05 for a pairwise comparison and p < 0.005 rather than 0.01 for a trend test is appropriate for assessing potential treatment-related increases in common tumors (spontaneous incidence of 1% or more) such as rat LGLL (Lin and Rahman, 1998). Use of the stringent p-value criteria for a commonly occurring tumor produces essentially the same overall number of Fisher rat leukemogens as the more rigorous NTP decision-making process, and can reduce the likelihood of a false positive outcome.
Given the potential relevance of the F344 rat LGLL to the rare human NK-LGLL and in light of the factors that complicate definitive interpretation of chemical-induced increases in LGLL (i.e., that spontaneous LGLL in F344 rat occurs at a high and variable incidence, is capable of being modulated by dietary factors such as corn oil, and has little evidence to support a mode of action [MOA]), it is proposed, like other reported recommendations (MacDonald, 2004) to adopt a "weight-of-evidence" approach when statistically identified increases in LGLL occur with exposure to a given compound. The "weight-of-evidence" approach, similar to the NTP's rigorous evaluation approach, should include assessment of the nature of dose–response curve in terms of incidence and/or severity, appropriate historical control data, reduction in latency time, reproducibility, or lack thereof when exposed through different routes, reproducibility, or lack thereof when tested in another strain or species, involvement of both sexes or only one, comparative species metabolism of the administered compound, genotoxicity, cytotoxicity, and any other relevant information. Most importantly, is there a biological plausible reason for tumor induction, or increased incidence? Does the chemical have toxic or carcinogenic effects on LGLs or their precursors? In addition, increasing the stringency of statistical analysis to further reduce the identification of false positives is also recommended. Moreover, detailed analyses of LGLL "associated" chemicals in NTP bioassays along with their genotoxicity and subchronic toxicity data may reveal a "model" LGLL-inducing chemical which could be used for future studies aimed at determining a MOA for LGLL in the F344 rat.
|
Funding |
|---|
|
TOPABSTRACTINTRODUCTIONPATHOBIOLOGY OF LGLL IN...CELL OF ORIGIN OF...LYMPHOPROLIFERATIVE DISORDERS...INFERENCES ON LGLL IN...UTILITY OF INCREASED STATISTICAL...DISCUSSIONFundingREFERENCES |
|---|
The Dow Chemical Company, Midland, MI, and BASF AG, Ludwigshafen, Germany.
|
ACKNOWLEDGMENTS |
|---|
The Panel gratefully acknowledges Gene McConnell for his expert review and insightful comments to the manuscript, Bhaskar Gollapudi for his active discussions and continued support during this project, and Carrie Houtman for her assistance in the manuscript preparation. This is a report of the Fischer Rat LGLL Expert Panel Review.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONPATHOBIOLOGY OF LGLL IN...CELL OF ORIGIN OF...LYMPHOPROLIFERATIVE DISORDERS...INFERENCES ON LGLL IN...UTILITY OF INCREASED STATISTICAL...DISCUSSIONFundingREFERENCES |
|---|
Brix AE, Nyska A, Haseman JK, Sells DM, Jokinen MP, Walker NJ. Incidences of selected lesions in control female Harlan Sprague-Dawley rats from two-year studies performed by the National Toxicology Program. Toxicol. Pathol. (2005) 33:477–483.[CrossRef][Web of Science][Medline]
Burks EJ, Loughran TP Jr. Perspectives in the treatment of LGL leukemia. Leukoc. Res. (2005) 29:123–125.[CrossRef]
Caldwell DJ. Review of mononuclear cell leukemia in F–344 rat bioassays and its significance to human cancer risk: A case study using alkyl phthalates. Regul. Toxicol. Pharmacol. (1999) 30:45–53.[CrossRef][Web of Science][Medline]
Chambers WH, Vujuanovic NL, DeLeo AB, Olszowy MW, Herberman RB, Hiserodt JC. Monoclonal antibody to a triggering structure expressed on rat natural killer cells and adherent lymphokine-activated killer cells. J. Exp. Med. (1989) 169:1373–1389.
Cheung MMC, Chan JKC, Wong K. Natural killer cell neoplasms: A distinctive group of highly aggressive lymphomas/leukemias. Semin. Hematol. (2003) 40:221–232.[CrossRef][Web of Science][Medline]
Chou WC, Chaiang IP, Tang JL, Su IJ, Huang SY, Chen YC, Liu MC, Lee FY, Wang CH, Shen MC, et al. Clonal disease of natural killer large granular lymphocytes in Taiwan. Br. J. Haematol. (1998) 103:1124–1128.[CrossRef][Web of Science][Medline]
Darbes J, Majzoub M, Breuer W, Hermanns W. Large granular lymphocyte leukemia/lymphoma in six cats. Vet. Pathol. (1998) 35:370–379.[Abstract]
Dieter MP, Jameson CW, French JE, Ganjee S, Stefanski SA, Chhabra RS, Chan PC. Development and validation of a cellular transplant model for leukemia in Fischer rats: A short-tem assay for potential anti-leukemic chemicals. Leukoc. Res. (1989) 13:841–849.[CrossRef]
Dunnick JK, Eustis SL, Huff JE, Haseman JK. Two-year toxicity and carcinogenicity studies of ampicillin trihydrate and penicillin VK in rodents. Fundam. Appl. Toxicol. (1989) 12(2):252–257.[CrossRef][Web of Science][Medline]
Dunning WF, Curtis MR. A transplantable acute leukemia in an inbred line of rats. J. Natl. Cancer Inst. (1957) 19:845–852.[Web of Science][Medline]
Elwell MR, Dunnick JK, Hailey JR, Haseman JK. Chemicals associated with decreases in the incidence of mononuclear cell leukemia in the Fischer rat. Toxicol. Pathol. (1996) 24:238–245.
Epling-Burnette PK, Liu JH, Catlett-Falcone R, Turkson J, Oshiro M, Kothapalli R, Li Y, Wang JM, Yang-Yen HF, Karras J, Jove R, et al. Inhibition of STAT3 signaling leads to apoptosis of leukemic large granular lymphocytes and decreased Mcl-1 expression. J. Clin. Invest. (2001) 107:351–362.[Web of Science][Medline]