Main

The incidence of extrahepatic biliary atresia (EHBA) is approximately 1 in 15 000 live births. EHBA is defined as complete occlusion of the extrahepatic biliary tract of various severity. Associated anomalies of other organ systems occur in about 20% of the affected newborns. Jaundice is the initial symptom, and its parenchymal origin is difficult to distinguish from an obstructive cause. Untreated patients die due to complications of hepatic cirrhosis with portal hypertension(1, 2). The only treatment is early surgical intervention, such as bilioenteric reconstruction by portoenterostomy of Kasai and its modifications. This is a symptomatic approach with a curative effect in some patients(3, 4). Otherwise, the operation prolongs the interval to liver transplantation. EHBA is the most frequent indication for liver transplants in childhood(5, 6).

The etiology of EHBA remains unknown, with two different etiopathologic theories existing. One hypothesis emphasizes a developmental disturbance of the biliary tract within the embryologic period(7); the other assumes an inflammatory and progressive process, possibly induced perinatally by a viral infection(8–10). The variability of the clinical course does not allow favoring either one of these suggestions(11, 12). Simulation of EHBA in animals has failed until now(13). However, recently developed murine models demonstrate that viral infection is capable of inducing obstruction of the extrahepatic bile duct.

Worldwide infection with different rotaviruses is one major cause of acute nonbacterial diarrhea in children under 2 y, with a high lethality, especially in developing countries. Children with rotavirus infection often show hepatic involvement with increased serum transaminase levels(14). The enteritis is probably induced by NSP4, a nonstructural glycoprotein that acts as a viral enterotoxin(15). Different capsid proteins, mainly the highly immunogenic VP6(16), mediate a shared reactivity between human and animal rotaviruses, suggesting the possibility of developing a rotavirus vaccine, following the Jennerian concept(17, 18). Extraintestinal effects of rotavirus infection are hepatitis and cholangitis, which occur in mice after heterologous infection with rotavirus(19). The most important finding in this murine model was the emergence of extrahepatic biliary obstruction in Balb/c mice infected with RRV(20). Most of the infected and subsequently icteric and dystrophic pups died within 3 wk; some recovered spontaneously. The histomorphologic changes often revealed a complete obstruction of the common bile duct and were similar to findings in children with EHBA(21). These data support Landing's(8) hypothesis that EHBA is one possible manifestation of an infection of the whole hepatobiliary tract. Because RRV can induce EHBA in only newborn mice, the immature immune system in these mice could play a role here. This suggestion encourages a new therapeutic approach.

The aim of the present study was to describe the influence of IFN-α treatment on the course of RRV infection in newborn Balb/c mice; attention was directed exclusively to the hepatotropic symptoms. The objectives of the study were 3-fold: first, to test the protective influence of IFN-α by prophylactic application in newborn Balb/c mice; second, to investigate whether the progressive obliteration of the extrahepatic bile duct could be influenced and stopped by treating symptomatic mice with IFN-α, and finally, to determine whether acquired immunity of the dams can protect their descendants in this model.

METHODS

Virus and cells. The serotype 3 rotavirus (strain MMU 18006) was originally isolated from a rhesus monkey with diarrhea and is regularly used in different studies(17, 19, 20). The virus was passaged in MA 104 cells, a continuous line of embryonic African green monkey kidney cells (Bio Whittaker, Walkersville), until a titer of ≈106 pfu/mL was reached. All procedures were identical to those described elsewhere(22).

Animals, infection, and treatment. Four- to five-week-old, virus-free Balb/c mice were purchased from Charles River (Breeding Laboratories, Sulzfeld), isolated in laminar flow cages, and kept for breeding. The first litter of each brood animal was used for the study, if more than three pups survived the first 24 h. Newborns were infected on their 2nd d of life with 10 μL of 30% glucose, containing ≈104 pfu of RRV, plus 20 μL of NaCl, 0.9%, inoculated i.p. Recombinant IFN-α was used for the treatment of the mice (r/IFN-α B/D hybrid; Ciba-Geigy, Basel). This human IFN-α hybrid has biologic activity in mice(23). IFN-α B/D, 10 000 IU in 10 μL of NaCl 0.9%, was applied i.p. The dose of IFN-α chosen for this series was orientated to Gresser et al.'s(24) former studies.

Infected and/or IFN-α-treated mice, which died within 2 d p.i. or were not fed by their mother after injection, were listed as early lethality and excluded from further analysis. The weight curves of normal developing newborn Balb/c mice are shown in a group of 18 pups, documented in a former study(21). A reference group of infected mice was observed for at least 4 wk for the purpose of documenting mortality, lethality, and recovery of the untreated population. Another control group of six mice was not infected and received 10 000 IU of IFN-α from their 6th to their 12th d of life, once a day i.p.

The clinical assessment of the infected mice included daily weight checks and the appearance of cholestatic signs, such as color of stools, icterus of the non-fur-covered skin, and bilirubinuria, tested by Bilugen-Test™ (Boehringer Mannheim). The observation period for the treated animals was orientated to the survival time of the infected mice in the control group. All treated mice were prepared under a dissecting microscope at the end of the procedure. Specimens from all mice were taken and fixed in formalin. Sections of the paraffin-imbedded organs were stained with H×E. The comparative interpretation of the histomorphology was orientated to former studies(21). Liver and ligamentum hepatoduodenale were investigated in all mice and, additionally, kidney, spleen, and thymus in the IFN-α-treated control group.

All procedures in this animal study were made in compliance with the national regulations for protection of animals and under supervision of the veterinarian of the responsible institute.

Statistics. The results are presented as percentages. Fisher's exact test was used to test categorical variables. Statistical significance was accepted at p < 0.05. Statistical software (SPSS for Windows™ V. 6.1.3.© (SPSS Inc. 1989-95) was used for all tests, including the cumulative survival tests.

RESULTS

Control groups. Twenty-nine mice from four litters were taken for the first control group. After infection with RRV, the early lethality was two mice, and 23 of 27 pups (85%) developed cholestasis after a short episode of diarrhea. This control group, indicated as “group A,” shows the spontaneous course of the disease (Table 1). Twenty-one mice (91%) died between the 9th and 27th d p.i. (mean 15th d) with persisting signs of cholestasis and stationary weight. Two mice recovered spontaneously on the 20th and 24th d, respectively, p.i. and subsequently showed good weight increase (Table 2). (The survival curve of this group is used as reference in Figs. 2 and 3.)

Table 1 Morbidity of RRV-infected Balb/c mice with and without treatment with IFN-α
Full size table
Table 2 Lethality and recovery of RRV-infected and cholestatic Balb/c mice after treatment with IFN
Full size table
Figure 2
figure 2

Survival curves of RRV-infected and cholestatic Balb/c mice without treatment (control group A) and after treatment with 10 000 IU of IFN-α for 1 wk (group E); see also Table 2.

Full size image
Figure 3
figure 3

Normal increase in weight of healthy Balb/c mice between the lines min and max. Average increase in weight of 22 RRV-infected and cholestatic mice, treated with 10 000 IU of IFN-α individually for 1 wk which recovered between d 23 and 43 (group E in Table 2 and Fig. 2).

Full size image

A second group of 20 mice was infected in the same way. One animal died the 2nd d, and 14 animals became icteric. These mice were used to document gross pathologic and histomorphologic changes and were killed 1 wk after onset of symptoms. At this time they presented with cholestasis and reduced weight gain. At dissection, the involvement of the liver was visible with edema and red spots on its surface. The extrahepatic bile ducts were also edematous, and the beginning of obstruction appeared in varying degrees and locations. The gallbladder was distended and filled with bile in six animals, atretic in one, and normal sized in the other mice. The histologic sections revealed intra- and extrahepatic cholangitis and some foci of reactive necrosis in the liver. These findings corresponded to results of former studies, when pancholangitis led mainly to EHBA-like occlusion of the bile ducts(20, 21).

Another control group was formed from one litter with six pups. These noninfected mice were treated with IFN-α for 1 wk and observed for 1 mo. IFN-α, 10 000 IU, was injected i.p., starting d 5 after birth. All animals remained asymptomatic and increased weight as did the pups in the healthy control group. The animals were killed at 4 wk p.i.; under the dissecting microscope, no pathologic changes could be found in tissue samples. Histomorphologic sections of liver, ligament hepatoduodenal, and kidney were normal. The only remarkable finding was a nodular hyperplasia in the spleen.

Single-dose treatment with IFN- α. Within the 2nd d of life, 29 mice from four litters received 10 000 IU of IFN-α i.p. as a single dose, prophylactically 7 h before infection (group B). Twenty-five pups of this group presented no clinical signs, and their weight curves were normal. Four mice developed cholestasis between the 6th and 10th day p.i., with two of them recovering spontaneously (Table 1). All animals were dissected on the 15th d p.i. The asymptomatic pups showed no pathologic changes, neither gross nor histomorphologically. The two mice that recovered spontaneously also presented no evident residues of the infection. However, the still icteric animals had the same intensity of hepatotropic involvement as the corresponding control group, with beginning occlusion of the bile ducts. Therefore, a single dose of IFN-α significantly altered (p < 0.0005) the course of the disease.

Another group of 17 pups from two litters received the same dose of IFN-α 6 h p.i. (group C). Three mice with a low initial weight of 0.5 g died the same day after infection and application of IFN-α. The remaining 14 animals remained healthy and developed normally (Table 1 and Fig. 1). These mice were not dissected, but were kept for breeding and further studies. The incidence of cholestatic symptoms in group B and C differs significantly from that of control group A.

Figure 1
figure 1

Survival curves of RRV-infected Balb/c mice without treatment (control group A), with single dose treatment of 10 000 IU of IFN-α 6 h p.i. (group C) and with single dose treatment of 10 000 IU of IFN-α 5 d p.i. (group D); see also Table 1.

Full size image

A third group of 17 pups, also from two litters (group D), was infected on the 2nd d with early lethality of one animal. The remaining 16 mice received 10 000 IU of IFN-α i.p. on the 5th d p.i., independent of clinical signs. Thirteen became icteric between the 5th and 10th day p.i., nine died between the 8th and 23th d p.i., whereas four mice recovered (Tables 1 and 2). The survival curves of this and the former group are shown in Figure 1. The incidence of cholestasis, as well as the lethality and the rate of recovering mice in this group, is statistically not different from the control group.

Symptomatic treatment with IFN- α. Thirty-eight pups from five litters were combined as one test group (group E) with no early lethality occurring after infection. Twenty-nine mice presented signs of cholestasis between the 5th and the 10th d p.i. The morbidity in this group is statistically not different to control group A (Table 1). With onset of the first symptoms of cholestasis, each animal received 10 000 IU of IFN-α i.p. daily for 1 wk. During the treatment with IFN-α, the animals showed side effects such as remarkable transpiration and a reversible alopecia on the head and tail. All mice completed the treatment cycle of 7 d with none perishing during this time. Seven animals died between the 19th and 27th d p.i. (mean 20th d.) and 22 recovered (Table 2 and Fig. 2). Recuperation times were different, and the symptoms persisted on average to the 35th d p.i. The weight curves of these mice were below the range of healthy and spontaneously recovering mice (Fig. 3)(21). Eighteen animals were killed at the 44th d p.i., whereas the rest were kept for further observation. Their weight curves became normal at 3 mo of age. Dissection under the microscope revealed no evident residues of the disease. Mild edema of the ligamentum hepatoduodenale occurred in 10 mice. However, obstruction or atresia of the DC was not visible, and no dilatation of the DC was seen. The surface of the liver was normal in all mice, and the gallbladder showed hydrops in only one case. The histomorphologic sections of the liver were mostly normal, apart from mild cellular infiltration in some sections. No proliferation of small bile ducts could be observed, particularly in the porta hepatis. The extrahepatic bile ducts in all animals were patent, and residues of cholangitis were visible in some slide preparations. Fibrosis could not be detected, neither intra- nor extrahepatically. Nodular hyperplasia in the spleen occurred, similar to the findings in the control group (Fig. 4). The rate of lethality and recovering mice is statistically different from the untreated control group A (p < 0.0005).

Figure 4
figure 4

Forty-five-day-old RRV-infected Balb/c mouse treated with 10 000 IU of IFN-α daily for 1 wk and recovery from cholestatic symptoms 37 d p.i. Normal findings in the hepatobiliary tract and poorly differentiated germinal follicles in the spleen as the only abnormality on histomorphologic examination (H×E, 100×).

Full size image

Acquired immunity for RRV. All infected mice, which were prophylactically treated with IFN-α 6 h p.i., survived without any symptoms and were reared for breeding (group C). Seventeen pups from three litters were infected with RRV in the same way as their mothers before them. One mouse died 2 d p.i., and three animals developed cholestasis at the 6th and 7th d, respectively, p.i (19%). One of these recovered at the 23th p.i., whereas the other two continued to show clinical signs of cholestasis. However, they grew slowly and were killed at d 30. The findings were different in both. One mouse presented with an atretic gallbladder and obstructed DC, with foci of necrosis in the liver, but no real atresia. The other animal showed a small prestenotic dilatation of the DC, which was not completely blocked, and normal findings in the liver parenchyma. Morbidity and lethality in this group is statistically different from control group A (p < 0.0005).

DISCUSSION

The development and interpretation of animal models for EHBA will remain difficult as long as the etiology of the disease in human newborns is not clear. Although the findings of irreversible occlusion and atresia of the extrahepatic bile ducts could result from different pathologic mechanisms, an infective origin seems to be plausible. However, in humans the causative agent could not be identified in spite of encouraging observations(25). Possibly there is no specific, EHBA-producing virus, but rather a still unidentified viral infection with considerable hepatovirulence, which can induce EHBA, especially under a specific immunologic constellation. Many factors might play a role when a likely ubiquitous virus leads to severe damage of the bile ducts. The immaturity of the local or the whole immunologic system might be responsible for the extent of the infection, as well as the absence of maternal antibodies(26–29).

The results of the presented study demonstrate that immunity plays an important role in rotavirus infection and the development of extrahepatic biliary atresia in mice. It is useful to look at this observation from three different aspects.

First, the prophylactic application of IFN-α in a single dose (groups B and C) gave a significant protection from hepatobiliary complications. IFN-α had to be given either some hours before or after infection. The later administration of the same single dose at the time of expected onset of symptoms did not show a protective effect in group D.

Second, IFN-α treatment of cholestatic mice for 1 wk is successful when the therapy starts in mice that are already ill. Similar to those of the untreated population in the control group, the weight curves of these mice, which died within 3 wk, differ from those of the development of spontaneously recovering mice(21). However, not all of these mice were cured. The design of the present investigation could not uncover the intraindividual effect of IFN-α on intra- and extrahepatic cholangitis. The symptomatic mice were too small and vulnerable for additional and invasive diagnostic procedures. Therefore, at present, it is not possible to clarify at which phase of the disease the fatal outcome can be prevented by IFN-α. However, the our results indicate that in this model the immature immune system can be successfully supported by external IFN-α.

Finally, infected and prophylactically IFN-α-treated mice in group C remained asymptomatic and were probably able to produce antibodies against the heterologous RRV infection. These dams acquired immunity to protect their descendants from severe hepatobiliary symptoms of i.p. RRV infection. How this congenital immunity occurs is not clear. Maternal immunoglobulin is a likely candidate for the immunity transmitted via the placenta or by mother's milk in suckling mice. A similar protective mechanism could be expected by vaccination of the mother animals.

Severe side effects of the IFN-α treatment, i.e. liver cell necrosis, glomerulonephritis, and inhibition of growth, as reported by others to occur after higher doses and different administration of IFN-α, were not seen in our study. The only corresponding findings were poorly differentiated germinal follicles in the spleen(30) of RRV-infected (group E) and not infected mice (control group) receiving IFN-α for 1 wk in therapeutic quantity.

How does the infection of Balb/c mice with RRV fulfill the conditions of an animal model for EHBA? Members of the reoviridae family are favored for murine EHBA models, due to their hepatotropism(13, 28). Particularly infections with different rotaviruses, which are responsible for severe gastroenteritis in all mammals, present with hepatic involvement in both mice and children(14, 25). The overall view of animal studies show that hepatovirulence of rotavirus infection can be influenced by different factors. It depends on the serotype of the virus; hepatobiliary reaction is higher in heterologously infected mice. The extraintestinal involvement intensifies by increasing the dose of virus and by i.p. application. Other determining factors are infection at an early age and the state of nutrition(19, 20, 31). Basically, the immunocompetence of the mouse line is essential for the course of the hepatobiliary disease(19). Balb/c mice, which are regarded as immunocompetent, lack the IFN-α-induced Mx proteins which take part in developing resistance to viral infection(32, 33). It is remarkable that this mouse line develops occlusion of extrahepatic bile ducts as a consequence of RRV infection. However, further studies are mandatory to investigate specific immunologic parameters, using other mouse lines and different virus strains.

From the observations in the presented model, we can conclude that IFN-α is effective in preventing hepatobiliary symptoms after infection with RRV. Most important in the study are the results of successful IFN-α treatment in those mice, which presented already showing signs of cholestasis. The findings of acquired immunity in this animal model have to be interpreted phenomenologically only. However, they provide a basis for further investigation.

The presented animal model consists of empirical observations, and its specific constellation is artificial. However, in this particular case, a high incidence of EHBA-like disease can be reproduced, although its course is not identical, but similar, to the human disease. The main question is whether any parameter in this model is congruent to the situation in human newborns developing EHBA. If we assume that EHBA can be induced by infection, we have to continue the search for the infective agent. Indirect signs of viral infection in the hepatobiliary tract, such as local IFN-α activity, may be helpful for proceeding in this direction. However, the EHBA-inducing virus has to be identified before any immunologic treatment or vaccination can be started as in the present animal study.