|Home | About | Journals | Submit | Contact Us | Français|
In the literature of experimental hepatic homotransplantation, it has not been emphasized that liver sepsis posed an unusual problem or occurred with exceptional frequency although reports from our laboratory1 and by Stuart et al2 and Fonkalsrud et al3 mentioned the development of liver abscesses as one postoperative complication. In these dogs, heavy antibiotic therapy was given. Moreover, Alican and Hardy4 observed hepatic abscesses in dogs which had received autotransplantation of the liver and biliary drainage by cholecystoduodenostomy. Their findings were particularly significant since neither an immunologic barrier for immunosuppression were involved in their experiments.
Since then, interest has been directed to the specific problem of liver homograft sepsis by the development of hepatic abscesses in four of six children who were treated at our institutions by orthotopic liver homotransplantation. The features and treatment of this complication in patients has been discussed elsewhere.5 The present report is concerned with experimental studies designed to elucidate the pathogenesis of the posttransplantation liver sepsis and to establish guidelines for its prevention by evaluating the influence of ischemia, type of biliary drainage, rejection, immunosuppression, and antibiotics.
The animals used were mongrel dogs and pigs of mixed breeds (predominantly Yorkshire and Hampshire) which weighed 13 to 17 kg (29 to 38 lb), and 15 to 27 kg (33 to 60 lb), respectively. The dogs were anesthetized with sodium pentobarbital and phencyclidine hydrochloride (Sernalyn) and a combination of sodium thiamylal (Surital) and succinylcholine chloride (Anectine) was used for the pigs. Postoperatively, liver function was followed with analyses of serum bilirubin, alkaline phosphatase, serum glutamic oxaloacetic transaminase (SGOT), and serum glutamic pyruvic transaminase (SGPT). Antibiotics were not administered during the preoperative or postoperative period except in group 5. Fifteen animals surviving less than three postoperative days were eliminated from the study since these early fatalities were generally attributable either to technical or postanesthetic complications. Three animals that died later of intussusception were also excluded. Further mention will be made only of the 51 completed experiments.
Five dogs had the abdominal cavity left open for three to 3½ hours after liver biopsy, approximately the same time as that required for orthotopic transplantation. Cholecystotomy and duodenotomy were made initially and closed at the end of the waiting period. The vascular supply to the liver was not interrupted.
Four pigs received the same operation described in group 1A with the addition of splenectomy.
In five pigs the liver was isolated from the circulation for 40 to 46 minutes by crossclamping the suprahepatic and subhepatic vena cava, portal vein, and hepatic artery. The blocked splanchnic system was decompressed with a splenicjugular venous bypass during the anhepatic phase. After isolation, the liver was quickly cooled by infusion through the portal vein of 1,000 to 2,000 ml of chilled (4 C) lactated Ringer’s solution. The fluid was allowed egress from a venotomy in the suprahepatic vena cava. The common duct was temporarily crossclamped. Cholecystotomy and duodenotomy were made and closed separately, thereby maintaining a natural biliary drainage.
Five pigs underwent the simulated autotransplantation described in group 2A with liver isolation for 39 to 56 minutes. However, the common duct was ligated and divided and a cholecystoduodenostomy was performed.
Five dogs were provided with orthotopic homografts from nonrelated donors using previously described techniques including a donor hepatic artery to recipient hepatic artery anastomosis and a cholecystoduodenostomy.6,7 The donors were cooled to between 30 to 34 C and killed by removal of their livers at the same time as the organs were core-cooled by intraportal infusion of chilled lactated Ringer’s solution.
Five pigs received orthotopic homografts. Deviations from the techniques used for dogs included splenectomy and the employment of a splenojugular venous bypass during the anhepatic phase. A portacaval shunt or separate decompression of the inferior vena cava was not employed.
Five dogs were pretreated for two to three days with subcutaneously administered canine antilymphocyte globulin (ALG). Azathioprine therapy was started at the time of transplantation and continued in the maximum daily doses which did not cause leukopenia. A one-week course of prednisone was administered in progressively diminishing daily doses, which started at 3 mg/kg of body weight on the day of operation. As in the other animals of groups 1 to 3, antibiotics were not given.
Five pigs were treated with azathioprine and prednisone as in group 4A. Antilymphocyte globulin was not employed.
Twelve dogs were treated with ALG, azathioprine, and prednisone as in group 4A. They were evaluated for survival and bacteriological studies were not obtained. On the day prior to transplantation, benzathine pencillin G therapy (1.2 million units) was started. In addition, a two-week course of chloramphenicol (250 mg to 500 mg daily) or tetracyline (500 mg daily) was begun three to four days prior to operation. Chronic survivors had their spectrum antibiotic changed every two weeks.
At operation, both the donor and recipient were examined bacteriologically by culturing pieces of the livers and gallbladder walls, bile, duodenal contents, and portal venous blood. Daily blood cultures were obtained by femoral venipunctures, beginning one day before transplantation. With the development of two or three positive blood cultures, or after an arbitrary period of approximately 21 days, the animals were killed. Seven animals which died before they could be killed were autopsied within two hours. In either case, the cultures obtained at the original operation were repeated under sterile conditions.
Blood samples were placed in tryptic soy broth bottles, and if no growth was evident by 14 days, subcultures were made to blood agar. The other specimens of tissues or body fluids were inoculated on duplicate plates of 5% sheep blood trypticase soy agar and lactobacillus selection agar; one of each kind of plate was incubated aerobically at 35 C and the other anaerobically in BBL gas pack jars. MacConkeys agar plate was also used for some of the samples. In addition, fungus cultures were made on bottles of plain and antibiotic-containing mycology agar and incubated at room temperature for 21 days before being discarded as negative.
Prior to the foregoing inoculations, the liver specimen was made into a brei by fragmenting it in a tissue grinder (Ten Broeck) which contained tryptic soy broth. By weighing the tissue introduced, adding a given volume of broth to the grinder, and inoculating the agar plates with a known quantity of the resulting brei supernatant, it was possible to estimate the bacterial count of different organisms for each gram of tissue.
Positive cultures from the blood, fluid, or tissue specimens were further studied by conventional staining and differential testing procedures. Anaerobic organisms were retested for inability to grow aerobically. Facultative organisms were counted as aerobes. Partial identification schemes based on methods used by Zubrzycki and Spaulding8 were used on some organisms. An identification of Bacteroides was made if the colonies were strictly anaerobic and if they were sporeless, gram-negative pleomorphic bacilli, and had a pungent, fetid odor. Strictly anaerobic gram-positive bacilli, (with or without spores) were identified as clostridia. Clostridium perfringens was identified by gram-stain morphology, colony appearance on blood agar, and stormy fermentation of milk. Other clostridia were simply listed as Clostridium species.
The results of approximately 1,500 cultures are summarized in Tables 1 to to5.5. Before and during operation, the bacterial flora were similar in both the 25 normal dogs and the 34 normal pigs (Table 1). In both species, organisms were occasionally found in the peripheral and portal venous blood, the liver, gallbladder wall, and bile. Bacteria, most commonly of the gram-negative variety, were found in more than half the duodenal cultures.
Postoperatively, this bacteriologic profile was altered more or less drastically according to the procedure carried out. The least profound changes followed the sham operations (group 1). However, even in these animals there was a high incidence of postoperative bacteremia as well as an increased number of positive cultures from the gallbladder, bile, and liver (Table 2).
After either simulated autotransplantation (group 2), or homotransplantation to untreated (group 3), or immunosuppressed (group 4) recipients, bacterial growth became ubiquitous inasmuch as sterile cultures were rare from any of the areas of sampling (Table 2). The bacteria were of all varieties with a strong representation of kinds normally found in the gastrointestinal tract (Table 3).
In the pigs with simulated autotransplantation, the enteric organisms appeared in the peripheral venous blood at a later time when the common duct was left intact than when cholecystoduodenostomy was performed. This resulted in the killing of the former animals (group 2A) after only eight days as compared to 16 days in group 2B.
In all instances in groups 2 to 4 in which the liver tissue contained bacteria, at least one of the other specimens had the same microorganisms. Nevertheless, there was wide variability in given animals in the bacterial strains isolated from the individual samples. Thus, the microorganisms ultimately isolated from the liver tissue often had not been previously found in the peripheral venous samples (Table 4). Similarly, bacteria grown from the portal vein, the gallbladder wall, bile, or duodenum were not necessarily represented in the infected liver tissue; this was particularly true after autotransplantation with an intact common duct (Table 4).
Although there were not significant qualitative differences in the livers of the animals with simulated autotransplantation (group 2) as opposed to those with true homotransplantation (group 3 band 4), the magnitude of sepsis seemed greater in the latter two series as quantitated by rough bacteriologic counting (Table 5). It was also in groups 3 and 4 that evidence of major hepatic necrosis was demonstrated as reflected primarily by increases in serum enzyme values and by other measures of liver function (Table 6).
The virulence of the septic complication was also highly variable in the four test groups as judged by the clinical courses. The animals subjected to sham operation or simulated autotransplantation usually seemed quite normal in spite of the well-documented postoperative bacteremia. All survived until they were killed. In contrast, both the untreated and immunosuppressed homograft recipients tended to follow a malignant course. Frequently, they passed from apparent well-being to a moribund state within a few hours. For example, the deterioration was so rapid in the immunosuppressed dogs (group 4A) that only one of five experiments could be electively terminated. The survival in the other four was three, seven, eight, and nine days, respectively.
The addition of intensive antibiotic therapy after canine homotransplantation to immunosuppressed recipients (group 5) radically changed the outlook. Of the latter 12 dogs, two died of hepatic artery occlusion after four and seven days, respectively, and two more of rejection at seven and eight days. The eight remaining animals lived for three weeks or more, and six survived for at least six weeks. Using a maximum credit of 70 days for any individual dog, the mean survival for the group was 38.7 ± 29.8 days (standard deviation).
The foregoing findings establish that highly characteristic infectious complications occur with orthotopic hepatic transplantation, and that sepsis selectively afflicts the liver homograft probably by virtue of its strategic location in relation to the rest of the bacteriarich gastrointestinal tract. The multiple factors which could contribute to invasion of hepatic parenchyma by pathogenic microorganisms have been analyzed in the present study in which antibiotic therapy was for the most part avoided.
In both dogs and pigs, bacteria were cultured from both the liver and portal vein at the time of initial laparotomy with a much lower frequency than that noted by previous authors.9–13 When the wounds were closed after a relatively nontraumatic sham operation, consisting only of duodenotomy and cholecystotomy, the incidence was only slightly increased with later sampling from the same location. However, postoperative systemic bacteremia was now demonstrable in eight of nine animals. The peripheral microorganisms were usually gram-positive cocci which in only one case were subsequently cultured from the liver.
When an ischemic injury was superimposed in pigs by performing simulated autotransplantation with or without division of the common duct, the bacteriologic pattern was markedly altered. Postoperative bacteremia again occurred, but there was a much higher representation of organisms which are indigenous to the gastrointestinal tract (Table 3). These same organisms were now also commonly although not invariably found in the liver, portal vein, gallbladder, bile, and duodenum; the presence in multiple sites of the same bacterial species was more pronounced after cholecystoduodenostomy than when the common duct had been left intact. Mixed flora were common in all the samples.
The foregoing sham procedures or simulated autotransplantations appeared to have provoked a potentially serious infectious complication to a degree which was roughly proportional to the magnitude of the procedure. However, the majority of the animals did not appear ill and all 19 survived until the experiments were electively ended in order to obtain specimens; presumably many could have lived indefinitely. The introduction of two additional factors into this dangerous environment resulted in explosive infections when antibiotic therapy was not used.
The first was the replacement of the animals’ own livers with homografts. The transplanted organs seemed to have little capacity to resist the bacterial invasion when subjected to the added duress of rejection. All the nonimmunosuppressed dogs and pigs developed overwhelming sepsis before or at the same time as there was biochemical evidence of liver necrosis. Often, there was an early bacteremia with gram-positive organisms which was succeeded within a few days by gram-negative septicemia. When the latter finding appeared, death was usually imminent. At sacrifice or autopsy, the same gram-negative organisms were very often found as part of a mixed and variable flora in the liver, gallbladder, upper gastrointestinal tract, and portal vein.
In animals which did not receive antibiotic therapy, attempts with host immunosuppression to prevent the liver necrosis caused by rejection did not improve the situation in terms of the promptness and the virulence of the sepsis; the postoperative course in these dogs and pigs was not significantly improved over that in animals in which rejection was allowed to run its natural course. Only when antibiotics were used in conjunction with immunosuppression was there significant prolongation of life.
Appreciation that septic complications after homotransplantation of the liver often center around the homograft should make it possible to avoid them. That this can be done is demonstrated by the fact that survival of as long as four years has been achieved in our laboratory after orthotopic transplantation of the dog liver. Two general approaches seem feasible.
First, incisive antibiotic therapy is mandatory as has been shown after other kinds of severe liver injury.14,15 Since bacterial seeding is probably through the biliary tract and portal vein, prophylactic treatment should be primarily based upon culture data on the flora in the recipient gastrointestinal tract. Mechanically, it may be possible to reduce the extent of contamination by providing biliary tract drainage with choledochocholedochostomy.
Of far greater importance, however, is the prevention of either immediate or delayed ischemic injury since, once extensive liver necrosis has started, the chance of preventing or controlling growth in this ideal culture medium has probably been lost. This was pointed out in a recent more complete analysis of four orthotopic liver recipients who died from two to six months after transplantation. Death was the direct or delayed consequence of septic hepatic infarction of the right liver lobe. In all four children, there was selective thrombosis of the right hepatic artery.16 Evidence was presented that distortion of the hepatic arterial anatomy could have been an underlying mechanical factor, to which low-grade rejection could conceivably have contributed.
A bacteriologic study was performed in dogs and pigs after sham operation, simulated hepatic autotransplantation, and orthotopic hepatic homotransplantation to unaltered and immunosuppressed recipients. When antibiotics were not given, each of the procedures was followed by bacteremia. With either autotransplantation or homotransplantation, bacterial invasion of the liver apparently occupied a central role in the complex septic course that followed. Ischemic injury, rejection, and immunosuppression without antibiotic coverage all aggravated the infections. After simulated autotransplantation, sepsis developed earlier after cholecystoduodenostomy than when the common duct was left intact. Long-term survival could be obtained in the recipients of homografts only if intensive antibiotic therapy was added to effective immunosuppression. These findings have direct implications for clinical liver transplantation.
This study was supported by Public Health Service grants AM-06344, HE-07735, AM-07772, AI-04152, FR-00051, FR-00069, and FO5-TW-1154.
Read before the 25th annual meeting of the Central Surgical Association, Cleveland, Feb 23, 1968.
Generic and Trade Names of Drugs
Prednisone—Deltasone, Deltra, Meticorten, Paracort.
Benzathine penicillin G—Bicillin.
Tetracycline—Achromycin, Bristocycline, Panmycin, Steclin.