Cocarcinogenic Effects of Islet Hormones and N-Nitrosomorpholine in Hepatocarcinogenesis after Intrahepatic Transplantation of Pancreatic Islets in Streptozotocin-Diabetic Rats Free

Administration of the N-nitrosamine N-nitrosomorpholine (NNM) in rats is a well-examined model of chemically induced hepatocarcinogenesis (1–14). Weber and Bannasch have used several dosage protocols of NNM administration (4–6). They have shown that the intensity and velocity of carcinogenesis depended on the level of dosage administration, whereas the sequence of cellular changes during the carcinogenic process did not differ between these groups. Hepatocarcinogenesis in this model typically begins with the formation of focal preneoplastic lesions, the foci of altered hepatocytes, which continue developing to hepatocellular adenomas and carcinomas after weeks and months. The first preneoplastic lesion in this model is the clear cell focus of altered hepatocytes. It is characterized by an increase in glycogen storage, an increase in proliferative and apoptotic activity, and distinct alterations in the carbohydrate metabolism, i.e., a decrease in the activities of glucose-6-phosphatase, glycogen-phosphorylase, and adenylate cyclase, an increase in glyceraldehyde-3-phosphate dehydrogenase, hexokinase, glucose-6-phospate dehydrogenase and pyruvate kinase activity. These alterations, together with the glycogenosis and the increased cell turnover, resemble typical insulin effects on hepatocytes (13, 14).

Our group developed an endocrinologically induced hepatocarcinogenesis model based on the transplantation of a low number of isologous islets of Langerhans into the livers of diabetic rats (15–17). After the transplantation of a low number (n = 350) of islets, a mild diabetes persists, and the β cells of the islets are maximally stimulated to permanently synthesize and secrete insulin. As a result of the local hyperinsulinemia, the liver acini draining the blood from the islet grafts show adaptive alterations that were very similar to those known from the clear cell foci in the chemical model, i.e., glycogenosis, increase in proliferation and apoptotic elimination of hepatocytes, and similarly altered enzymes of carbohydrate metabolism (15, 16). In the first 3 months after transplantation, they are strictly restricted to the downstream liver acini, but then they gradually expand into the neighboring liver tissue and finally progress to hepatocellular adenomas and carcinomas between 6 and 24 months (17).

Insulin effects in the endocrine model were also reflected in alterations of the insulin-like growth factor (IGF) system, i.e., in the up-regulation of IGF-I and IGF binding protein-4 as well as in the down-regulation of IGF binding protein-1 (18). Interestingly, preneoplastic hepatocytes in both models strongly overexpress the key enzyme of the de novo fatty acid synthesis, i.e., the fatty acid synthase (19), which not only resembles another insulin effect, but may also be connected to carcinogenesis, as this enzyme has been shown to be overexpressed in preneoplasias and malignant tumors of many human tissues (20). Most of these alterations in the endocrine model at this early time point can be explained by receptor-mediated insulin signaling, as we have recently shown that the hepatocytes, located downstream of the islet grafts, show a translocation of the insulin receptor from the cell membrane to the cytoplasm and an overexpression of insulin receptor substrate-1 and components of the Ras-Raf–mitogen-activated protein kinase pathway (21).

Considering the similarities of the hepatocarcinogenic process of both models, it was tempting to examine whether chemically induced carcinogenesis after NNM administration could be modulated by simultaneous transplantation of pancreatic islets. In addition, it was interesting to observe whether high-number islet transplantation, which is increasingly used as a therapy for type 1 diabetes mellitus in humans, and which alone did not suffice to induce hepatocarcinogenesis in our animal model, may have cocarcinogenic potential and enhances the NNM-induced carcinogenic process. Thus, in this study, we combined different ways of NNM administration with different transplantation procedures to investigate a potential cocarcinogenic effect of these two models.

Highly inbred 3-month-old male Lewis rats (n = 385; 250-300 g body weight) were used in this study. Housing of the animals, including measurement of body weight and blood glucose every 4 weeks, was described in detail previously (15) and was in line with the guidelines of the Society for Laboratory Animal Service and the strict German Animal Protection Law. Animals were divided into five experiments. In addition to the islet transplantation, different regimens of NNM administration were used in the main experiments A to C, whereas partial hepatectomy was done in experiment D. The classic transplantation model, i.e., islet transplantation without any further carcinogenic stimulus, was additionally done as experiment E for comparison with the late stage (50 weeks) animals of experiments A and D. Each of the experiments (A-C) consisted of one main group and four control groups (I-IV), whereas there were only one and three control groups in experiments D and E, respectively. Table 1 summarizes the different procedures in the respective experimental arms.

Diabetes induction. Diabetes was induced in animals of the main, control group I, and control group II by giving a single s.c. dose of streptozotocin (80 mg/kg body weight) 6 weeks prior to NNM administration/partial hepatectomy, whereas control groups III and IV remained nondiabetic.

Transplantation procedure. Details of the transplantation procedure have been described previously (15). Three weeks after induction of the diabetes and 3 weeks prior to NNM administration, all groups, except control group IV of E, underwent a transplantation procedure. Rats of the main group and control group III received a low number of islets of Langerhans (n = 350), whereas animals of control group I were transplanted with a high number of islets (n = 1,000). Instead, control group II received 350 polystyrene (Latex) particles (Polysciences, Warrington, PA) of about 90 to 100 μm in size. Animals of control group IV in A to C were sham-transplanted and received only 1 mL of the carrier solution [Hank's solution (pH 7.2), Sigma, Heidelberg, Germany]. All transplants were injected into the portal vein after clamping the branches that supply the left part of the livers (i.e., left lobe and left part of the middle lobe), thus verifying that the transplants were embolized only into the right part (i.e., right lobe, right part of the middle lobe, caudate lobe, anterior and posterior papillary processus).

N-nitrosomorpholine administration. Three weeks after the transplantation, the animals of experiments A to C received NNM by oral application via a pharyngeal tube, according to different administration protocols (Table 1). NNM was kindly provided by Prof. Dr. Rudolf Preussmann, Deutsches Krebsforschungszentrum Heidelberg (German Cancer Research Center), Heidelberg, Germany. In experiment A, the animals received a single but high dose (200 mg NNM/kg body weight). In experiment B, the “stop” experiment, rats were given a much lower dose (16 mg/kg body weight), but treatment was given daily, continuing for 6 weeks and then stopped. In experiment C, the dose was even lower (5 mg/kg body weight), but given throughout the whole experimental period until the animals were sacrificed. The NNM dose was adjusted to the alterations in body weight every 4 weeks. At the end of the experiment, total NNM dose was thus highest in experiment C (∼1,300 mg), medium in B (∼600 mg), and lowest in A (200 mg). The administration protocols were transferred from former experiments, done by Weber and Bannasch (4–6) but had to be slightly modified, as the original NNM doses proved to be too toxic in preliminary experiments of this study because of the concurrent diabetes of the animals. Thus, we lowered the NNM doses given in their studies by 20%, 33%, and 16% in our experiments A, B, and C, respectively.

Partial hepatectomy. In experiment D, instead of NNM treatment, the animals were partially hepatectomized. In ether anesthesia, after cutting the falciform ligament, the left lobe and middle lobe of the liver were ligatured (catgut 2-0) and cut with scissors 3 to 4 mm adjacent to the ligature. The peritoneal cavity and the skin were closed with stitches and clips, respectively.

Animal sacrifices. Animals were killed under anesthesia as described previously (15). The different sacrifice time points of the respective experiments (Table 1) were described by Weber and Bannasch (4–6). Animals that died spontaneously were included in the nearest time group. These animals were not analyzed stereologically regarding the volume fraction of preneoplastic foci because of the lack of perfusion fixation and the resulting loss in quality of tissue preservation. However, tissue preservation sufficed to evaluate the development of hepatocellular neoplasms.

Details of tissue processing are given in ref. (15). Briefly, after perfusion fixation, the liver was removed and cut into slices of 1 mm thickness. In addition to all macroscopically visible lesions (>2 mm diameter), 10 liver slices were embedded in paraffin, and slides of 2 to 3 μm thickness were cut and stained by H&E and the periodic acid Schiff (PAS) reaction.

Foci of altered hepatocytes were easily identified because of the clear-cell (H&E) morphology of the hepatocytes that stained intense purple with the PAS reaction owing to the high glycogen content. Later stages of preneoplastic foci showed increased basophilia in some (mixed-cell foci) or in virtually all of the hepatocytes (basophilic cell foci) in H&E stain and a corresponding partial or complete loss of PAS reactivity.

The volume fraction of foci of altered hepatocytes was estimated from their area fraction in the PAS reaction at ×100 magnification, determined by the point-counting stereological technique with an ocular grid described by Weibel (22). At least 600 points were counted per section.

Hepatocellular adenomas were diagnosed if (a) the acinar morphology was lost, (b) the lesion showed a trabecular pattern, (c) displayed no or only mild atypia, and (d) compressed the adjacent parenchyma. Hepatocellular carcinomas were diagnosed if the lesion showed a pseudoglandular or a trabecular pattern with more than three cell layers in at least two different areas.

We only tested differences reflecting possible cocarcinogenic effects, for example by comparing the right part with the left part of the liver within one experimental group or the right liver part of the main group with the right liver part of the respective control groups. Differences in volume fraction of preneoplastic foci and total number of hepatocellular neoplasms were tested with the Wilcoxon-Mann-Whitney test, whereas the frequency of animals bearing at least one neoplasm was tested using Fisher's exact test. P < 0.05 was considered statistically significant.

The highest level of hyperglycemia was noted in the control group II animals, which were diabetic and received polysterene particles instead of islet transplants (blood glucose permanently ∼400 mg/dL). In all the experiments, the main group animals also showed hyperglycemia throughout the entire experimental period, albeit less than the control group II animals (260-400 mg/dL); thus, the low-number transplantation of pancreatic islets could not fully compensate the streptozotocin-induced diabetes, as planned in the experimental design. As expected, the high-number islet transplantation in control group I animals of experiments A to C were able to establish normoglycemia (60-100 mg/dL). Animals of control groups III and IV were nondiabetic and thus also stayed normoglycemic for the whole experiment.

Hyperglycemia was inversely correlated with body weight gain; therefore, weight gain was lowest in control group II animals. They showed an increase of at most 10% after 50 weeks (when compared with body weight immediately before streptozotocin administration). The animals in the main group, suffering from a milder diabetes, increased in body weight by 30%. In the normoglycemic control groups I, III, and IV, body weight gain was highest, varying between 40% and 60% (after 30 weeks).

In general, the modus and stepwise development of hepatocarcinogenesis was not different from former experiments either in the endocrine or in the chemical model (Figs. 1–3). The clear cell focus was the exclusive type of preneoplastic lesion at the beginning of the experiments (Figs. 2A,, D and Figs. 3A), whereas at later stages, a considerable number progressed to mixed-cell and basophilic foci (Fig. 3B). Hepatocellular adenomas (Fig. 3C and D) were of mixed cellular composition but usually still contained a considerable number of glycogenotic hepatocytes; hepatocellular carcinomas, however, were predominantly composed of basophilic cells (Fig. 3E and F). There were considerable differences in the intensity of the carcinogenic process between the experiments because of the differential cocarcinogenic effects of NNM and the islet hormones.

Experiment A. Seven and 11 weeks after NNM treatment, the animals of the main group showed a moderate number of foci of altered hepatocytes (Table 2). They were located only in the right part of the liver and were strictly confined to the liver acini, draining the blood from the islet grafts; thus, they were evidently a consequence of the transplantation and cannot be attributed to NNM treatment. The first chemically induced and randomly distributed foci of altered hepatocytes (Fig. 2D), occurred in both liver lobes of the main group and in all control groups not before week 20.

During the whole observation period, the volume fraction occupied by foci of altered hepatocytes in the main group animals was significantly higher in the right part of the liver, bearing the islet transplants, than either the left part or in control group II-IV rats, mainly reflecting the carcinogenic effect of islet transplantation (Fig. 1A-C; Table 2). Interestingly, the single NNM dose did not lead to a significant gradual increase in volume fraction of foci of altered hepatocytes, but resulted in a pronounced progression from pure clear cell foci to morphologically more advanced phenotypes, such as mixed-cell or basophilic phenotypes. At early time points, main group animals also showed a significantly higher volume fraction occupied by foci of altered hepatocytes in the right part of the liver than did control group I animals; however, after 50 weeks, this difference diminished. Instead, control group I animals showed a significant increase in volume fraction of preneoplastic foci in the right part of the liver when compared with the left part and control group II to IV animals (Table 2).

Hepatocellular neoplasms were observed only in the late time-group, i.e., at week 50 after NNM administration (Table 3). With the exception of a single hepatocellular adenoma in 1 animal each from control groups I and II, all hepatocellular neoplasms, i.e., 13 hepatocellular adenomas and 1 hepatocellular carcinoma, occurred in the right part of the liver in the main group animals. This difference was statistically significant not only when compared with the left part of the liver of the same animal group and to control groups I, II, and IV animals, but also when compared with the late stage main group animals of control experiment E, in which the long-term effects of sole islet transplantation without additional NNM treatment was tested.

Experiment B. Although the effects of the islet transplantation dominated the NNM influence in experiment A, the cocarcinogenic effect of the intensified NNM treatment was clearly stronger in this experiment. At the earliest time point, i.e., after only 10 weeks, several foci of altered hepatocytes occurred in all experimental and control groups (Table 2). However, the additional low-number islet transplantation was still effective as a cocarcinogenic stimulus. This was reflected in the significantly higher volume fraction of foci of altered hepatocytes in the right part of the liver of main group animals when compared with the graft-free left part and to the control groups at all time points with one exception: at week 16, the carcinogenic effect of NNM was so strong in control group I animals that the volume fraction of foci of altered hepatocytes was not significantly lower than that in main group animals, although control group I animals received a higher number of islet grafts and the influence of insulin on the adjacent hepatocytes must have been lower in this group. Instead, in these animals, the combined carcinogenic stimulus of the chemical and endocrine model leads to a significant increase in volume fraction of foci of altered hepatocytes in the right part of the liver, when compared with the left part at 16 and 32 weeks (Fig. 1D-F; Table 2). It should also be noted that the combination of streptozotocin-induced diabetes and NNM treatment also lead to an increase in volume fraction of foci of altered hepatocytes in the severely diabetic control group II (after 16 weeks) and the left liver lobe of the diabetic main group (after 10 and 16 weeks) when compared with the nondiabetic and not streptozotocin-treated control group III and/or control group IV (Table 2).

The intensified carcinogenic process in this experiment was also reflected in the accelerated formation of neoplasms. After only 10 weeks, main group animals showed a significantly higher number of hepatocellular adenomas when compared with the control groups (Table 3). The enhanced tumorigenesis in the main group was maintained after 16 and 32 weeks, although the strong NNM effect also led to the development of several hepatocellular neoplasms in the control groups.

Experiment C. At 11 weeks, the volume fraction occupied by foci of altered hepatocytes in the right part of the liver of main group animals was not as high as in experiment B because the total NNM dosage was lower at this time point. Main group animals showed a higher volume fraction of foci of altered hepatocytes in the right part of the liver when compared with the left part and to the control groups (Table 2). This situation was maintained after 20 weeks. After 27 weeks, and in particular after 37 weeks, the continuous NNM treatment reached such high total levels that carcinogenesis was also very strong in the control groups, although the volume fraction of foci of altered hepatocytes was still highest in the right part of the liver of the main group, indicating that the transplant grafts were still of cocarcinogenic effect (Table 2).

Regarding the neoplasms, cocarcinogenic effects became gradually more visible (Fig. 1G-L; Table 3). After 11 weeks, only one single hepatocellular adenoma occurred in the right part of one main group animal. After 20 and 27 weeks, however, main group animals showed 6 and 12 hepatocellular adenomas in the right parts of their livers, respectively, a significantly higher number than in the left part or in any control group animal. After 37 weeks, 44 hepatocellular neoplasms, i.e., 18 hepatocellular adenoma and 26 hepatocellular carcinoma, occurred in the right liver parts of 10 main group animals, which significantly outnumbered the tumors noted in the left part of the liver and in all control groups, although these animals also developed many tumors because of the strong NNM effect.

Experiment D. The additional proliferative stimulus of partial hepatectomy had only slight effects on carcinogenesis in the islet transplantation model. Throughout the experiment, lesions in islet-transplanted animals clearly outnumbered the very few (sporadic) clear cell foci that occurred in the latex particle–transplanted rats of the control group. Interestingly, the volume fraction of foci of altered hepatocytes in the main group more than doubled within 50 weeks observation, a feature that we have never observed in the classic transplantation model.

In addition, at that late time point, main group animals of experiment D had a higher volume fraction of foci of altered hepatocytes and significantly more hepatocellular adenoma than the solely islet-transplanted main group animals of the control experiment E (Table 3). Thus, a mild stimulatory effect of partial hepatectomy on tumor development could be noted. Hepatocellular carcinoma did not develop in this experiment within the observation period.

This investigation has shown that hepatocarcinogenesis chemically induced by administration of NNM and a simultaneous hormonally induced hepatocarcinogenesis after transplantation of islets of Langerhans into the liver of streptozotocin-diabetic rats exert cocarcinogenic effects. The different experiments have shown that the carcinogenic effects are dose-dependent. In experiment A, NNM was given only once and was by far the lowest total dose. Here, the hepatocarcinogenic process was very similar to that of the classic islet transplantation model with the major difference that control group I animals also developed significantly more preneoplasias in their right, i.e., islet-bearing, liver part in late stages of the observation period. The transplantation model has been the subject of investigation for several years (15–19, 21, 23, 24). In contrast to the main group, control group I animals, i.e., rats receiving a high number of islets, never showed the development of foci of altered hepatocytes or even hepatocellular neoplasms. The high number of grafts (1,000 islets) resulted in normoglycemia. Thus, the B cells of the islet grafts are not stimulated to reach maximal insulin production, and the downstream hepatocytes are not exposed to such a strong insulin influence as observed in the animals in the main group. This difference is also reflected in the islet graft morphology and ultrastructure, as islet grafts in this group showed neither a tendency towards growth nor an increased transformation into insulinomas as is the case in the main group (15–17, 24). Thus, in the classic endocrine model, the insulin concentration in the downstream liver acini of the high number transplantation group is too low to induce foci of altered hepatocytes. However, as observed in experiment A (after 50 weeks) and in experiment B (after 16 and 32 weeks), it is possible to induce pronounced right-sided carcinogenesis in these animals by simultaneously administrating a certain dose of NNM. In addition to their preferential location in the right liver half, many of these foci of altered hepatocytes typically developed in the downstream area of the transplants and not haphazardly in the parenchyma, as shown in Fig. 2A; thus, their existence must be interpreted as a combined effect of insulin and NNM.

Regarding these control group I animals in particular, it is worth reflecting on the questions of why NNM and insulin have cocarcinogenic effects and how they may act together. The carcinogenic mechanism of NNM is insulinomimetic and thus similar to the islet model. These similarities include alterations in the carbohydrate and lipid metabolism as well as a translocation and/or overexpression of the insulin receptor, the IGF-I receptor, and components of the insulin signal transduction pathway, such as insulin receptor substrate-1, RAF-1, and mitogen-activated protein/extracellular signal-regulated kinase-1 (13, 25–27). The main difference is probably that an overexpression of the insulin receptor is responsible for the initiation of the insulinomimetic effect in the NNM model with normal levels of insulin and glucose in the blood, whereas the insulin signaling cascade in the transplantation model is stimulated by hyperinsulinemia owing to high blood glucose levels. The cocarcinogenic effect of the high-number islet transplantation on hepatocarcinogenesis induced by NNM could thus result from somewhat increased intrahepatic levels of insulin exerting an additional stimulus on the insulin signaling cascade in that foci which already overexpress the insulin receptor.

Interestingly, compared with the sole transplantation model, we also observed an enhancement of tumor development when combining islet transplantation with partial hepatectomy. Owing to its proliferation-promoting effect, partial hepatectomy is regarded as a classic promotor in accordance with the initiation-promotion model. Thus, this phenomenon might be explained as the promotion of foci of altered hepatocytes by partial hepatectomy after being previously initiated by primarily islet-derived effects, i.e., insulin. Based on this model, however, it is difficult to explain the combined effects of NNM and islet transplantation on the hepatocytes. NNM is considered a complete carcinogen, and hepatocytes initiated by NNM are preferentially located in zones 1 and 2 of the liver acinus (13). Now, one might believe that NNM-initiated cells which by chance lie within the downstream area of an islet graft may have growth advantages owing to the tumorigenic stimuli of the local hyperinsulinemia, thus, in this context, allotting insulin only promoting effects. This might also be an explanation for the development of foci of altered hepatocytes in control group I animals, in which the simultaneous NNM administration is indeed crucial for the induction of hepatocarcinogenesis. On the other hand, low-number islet transplantation without administration of NNM completely suffices to induce hepatocarcinogenesis in diabetic rats in the classic endocrine model (15–17), which would classify low-number islet transplantation, i.e., strong hyperinsulinism, as a complete carcinogen.

The diabetes in these animals was caused by streptozotocin. Therefore, one might argue that the tumors might be a result of initiation by streptozotocin administration and promotion by islet-derived effects, as streptozotocin has indeed been shown to produce renal and pancreatic neoplasms (28). However, in that study, no carcinogenic effect of streptozotocin on the liver was noted. The fact that the volume fraction of foci of altered hepatocytes at some time points was higher than in some non-diabetic control groups in the left liver part of the main group and control group II in experiment B might be the result of an additional directly hepatocarcinogenic effect of streptozotocin. It is, however, more likely that this is caused indirectly by the diabetes of these animals, as normoglycemic control group I animals that have also been treated by streptozotocin do not show an increase in foci of altered hepatocytes in the left liver part. Finally, we have recently finished experiments using the islet transplantation model (without NNM treatment) in spontaneously diabetic BB/Pfd rats that did not receive streptozotocin, and found the carcinogenic process to be identical.⁴

We believe that our results are important regarding therapeutic islet transplantation in human diabetes mellitus. This promising therapy is an increasingly used approach, and in the meantime, >750 patients have received islet transplantation into the liver (29). Owing to new immunosuppression protocols, patients may become independent of exogenous insulin therapy, which significantly improves their metabolic situation and life quality, although lifelong immunosuppression is required (29–32). Naturally, experiences regarding long-term effects of islet-derived hormones on the liver are still very limited and, to the best of our knowledge, detailed histopathologic studies of human donor livers do not exist. However, one experimental study in non–human primates has been done, using an experimental setting similar to the situation in humans (33). Although mainly concentrating on islet morphology, the authors describe focal hepatocellular glycogenosis in the hepatocytes surrounding the islet grafts in one of five animals examined. Only this animal periodically required exogenous insulin, which could be explained histomorphologically by finding only sparse and small but functioning islets. This description of the liver tissue and the metabolic situation in this primate exactly corresponds with our main group animals after low-number islet transplantation, demonstrating that foci of altered hepatocytes may also develop in primates. Moreover, there is growing evidence that hepatocellular carcinoma might also originate from glycogenotic foci in humans (26, 34). Thus, we conclusively suggest that lower levels of insulin might also promote hepatocarcinogenesis against the background of concurrently active noxae, which should also be considered when treating human diabetic patients with therapeutic intrahepatic islet transplantation.

Grant support: Deutsche Forschungsgemeinschaft (Do 622/1-3, 1-4, and 1-5).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Gabriele Becker, Jörg Bedorf, Danuta Chrobok, Mariana Dombrowski, Mathilde Hau-Liersch, Kirsten Herrmanns, and Barbara von Netzer for technical assistance; Yvonne Fischer and Kurt Rüdel for animal care; as well as Gerrit Klemm and Thomas Jonczyk-Weber for photographic work. We thank Prof. Dr. Rudolf Preussmann, Deutsches Krebsforschungszentrum Heidelberg, for providing us with N-nitrosmorpholine, and Prof. Dr. Peter Bannasch, Deutsches Krebsforschungszentrum Heidelberg, for many helpful discussions.