ANNALS OF SURGERY Vol. 227, No. 2, 205-212 © 1998 Lippincott-Raven Publishers Increased Intestinal Permeability and Altered Mucosal Immunity in Cholestatic Jaundice Fenella K. S. Welsh, F.R.C.S,* Carol W. Ramsden, F.I.M.B.L.S.,* Kenneth MacLennan, F.R.C.Path.,* Maria B. Sheridan, F.R.C.R.,* G. Robin Barclay, Ph.D.,t Pierre J. Guillou, F.R.C.S.,* and John V. Reynolds, F.R.C.S.I. * From the Professorial Surgical Unit, Department of Pathology, and Department of Radiology, St. James's University Hospital, Leeds, England,* and the Regional Blood Donor Center, t Edinburgh, Scotland Objective To examine the effects of cholestatic jaundice on gut barrier function. Summary Background Data Gut barrier failure occurs in animal models of jaundice. In humans, the presence of endotoxemia indirectly implicates failure of this host defense, but this has not previously been investigated in jaundiced patients. Methods Twenty-seven patients with extrahepatic obstructive jaundice and 27 nonicteric subjects were studied. Intestinal permeability was measured using the lactulose-mannitol test. Small intestinal morphology and the presence of mucosal immunologic activation were examined in endoscopic biopsies of the second part of the duodenum. Systemic antiendotoxin core IgG antibodies and serum interleukin-6 and C-reactive protein were also quantified. Intestinal permeability was remeasured in 9 patients 5 weeks after internal biliary drainage. Results The median lactulose-mannitol ratio was significantly increased in the jaundiced patients. This was accompanied by upregulation of HLA-DR expression on enterocytes and gut- associated lymphoid tissue, suggesting immune activation. A significant increase in the acute phase response and circulating antiendotoxin core antibodies was also observed in the jaundiced patients. After internal biliary drainage, intestinal permeability returned toward normal levels. Conclusions A reversible impairment in gut barrier function occurs in patients with cholestatic jaundice. Increased intestinal permeability is associated with local immune cell and enterocyte activation. In view of the role of gut defenses in the modern paradigm of sepsis, these data may directly identify an important underlying mechanism contributing to the high risk of sepsis in jaundiced patients. 205 206 Welsh and Others Ann. Surg. * February 1998 Sepsis and its sequelae are the leading cause of morbid- Twenty-seven patients with extrahepatic obstructive jaun- ity and mortality in jaundiced patients,' and this risk cor- dice and 27 nonicteric controls were investigated. The relates with the severity of the jaundice. A potential rela- groups were matched for age, sex, and the presence of tion with gut barrier function was suggested more than neoplastic disease. Postoperative patients, patients with 30 years ago, when Fine et al.2 postulated that the intestine evidence of infection or sepsis, and patients taking ste- was a source of systemic endotoxemia. Many subsequent roids or nonsteroidal anti-inflammatory drugs were ex- investigators have reported a high incidence of portal and cluded from the study. Each patient was defined as being systemic endotoxemia in jaundice, and endotoxin has well-nourished or mildly or moderately malnourished, us- been incriminated in the pathophysiology of postoperative ing the method of subjective global assessment, as pre- complications in such patients.3 Bile salts and lactulose viously described.12 Severely malnourished patients were may neutralize gut endotoxin and alter enteric microflora, excluded from the study because of the impact nutritional and their administration has been reported to result in depletion may have on mucosal permeability. clinical benefit in jaundiced patients.45 Although a num- ber of related abnormalities result from jaundice, includ- ing impaired endotoxin clearance by Kupffer's cells6 and Intestinal Permeability altered systemic immunity,7 these observations indirectly Intestinal permeability was quantified using the lactu- suggest that impairment of gut barrier function may be lose-mannitol test.'3 The principle of this test is that relevant in jaundiced patients. lactulose and mannitol are absorbed from the intestine by There is an emerging consensus that a failure of the different pathways. Mannitol, the smaller molecule, is gut barrier to exclude endogenous bacteria and toxins absorbed via the transcellular route, lactulose by the para- from the portal and systemic circulation contributes to the cellular route. Both probes are rapidly excreted in the pathophysiology of the systemic inflammatory response urine. In conditions associated with increased permeabil- syndrome, sepsis, and multiple organ failure.89 In hu- ity, the lactulose-mannitol excretion ratio is increased mans, this hypothesis is supported by the frequent absence through relatively greater absorbence of lactulose com- of an identifiable focus'0 in septic patients with microbio- pared to mannitol. After a 6-hour fast, a pretest urine logic evidence of infection and by unexplained sepsis and specimen was collected from each patient, who then drank multiple organ failure in patients with no microbiologic the test solution containing 5 g lactulose, 2 g mannitol, evidence of infection." This theory of gut-origin sepsis and 22.4 g glucose in 100 mL of water. Urine was then is supported by data from animal models.9 Although the collected for the following 5 hours into 0.2 g chlorhexi- integrity of intestinal defenses and barrier function is cen- dine. Urinary mannitol was measured by spectrophotome- tral to the thesis, there have been few studies of gut barrier try, using a color change reaction with potassium perio- function in relevant clinical situations in humans. date; urinary lactulose was measured by paper chromatog- The aims of this study were to determine whether gut raphy.'4 The results are expressed as a ratio of the barrier function was impaired in jaundiced patients and recovery of the two molecules. The normal range for the to investigate the effects of any altered barrier function urinary lactulose-mannitol ratio is 0 to 0.07. on intestinal morphology or gastrointestinal immune function. In addition, we attempted to determine whether increased intestinal permeability was associated with an Intestinal Morphology and altered cytokine profile and acute phase response, or with Immunohistochemistry evidence of endotoxin exposure. Intestinal permeability All patients underwent endoscopy, at which time biop- was also examined before and after internal biliary drain- sies were taken from the second part of the duodenum, age. distal to the ampulla of Vater. One biopsy was fixed in 10% buffered formalin and embedded in paraffin, and 5- /,m sections were cut and stained with hematoxylin and METHODS eosin. Intestinal villus height and crypt depth were mea- Patients sured using a microscope eyepiece graticule. At least 10 Ethical approval was obtained from the Ethics Commit- villi or crypts were measured for each section, and the tee of St. James's University Hospital, according to the mean value was determined. A second biopsy was snap- Helsinki agreement. Informed consent was mandatory. frozen in liquid nitrogen; 5-,um sections were cut in a cryostat and the sections were stained using a standard three-stage immunoperoxidase reaction as follows. After Address reprint requests to John V. Reynolds, Academic Surgical Unit, blocking nonspecific binding sites with 5% serum in phos- St. James's University Hospital, Leeds LS9 7TF, United Kingdom. phate buffered saline (PBS) and avidin and biotin Accepted for publication January 1997. blocking solutions (Vector AB Blocking kit; Vector, UK), Vol. 227 * No. 2 Jaundice Impairs Gut Barrier Function 207 Table 1. PRIMARY ANTIBODIES USED Antibody Cellular Distribution Source Description Dilution CD3 T cells DAKO* Rabbit polyclonal 1:100 CD4 Helper T cells PGt Mouse monoclonal CD8 Cytotoxic T cells PG Mouse monoclonal CD30 Late activated T cells DAKO Mouse monoclonal IgGl 1:40 CD68 Macrophages DAKO Mouse monoclonal IgGl 1:50 HLA-DR (MHC class 11) B cells, activated T cells, DAKO Mouse monoclonal IgGl 1:100 macrophages IgA Secretory IgA DAKO Rabbit polyclonal 1:500 * DAKO, High Wycombe, Buckinghamshire, United Kingdom. t PG = personal gift, Dr. Ros Banks, Institute of Cancer Studies, University of Leeds, United Kingdom. tissue sections were incubated at room temperature for signed ranks test. Analysis of categorical variables was 30 minutes with a primary antibody (Table 1) or serum performed using the chi square test with Yates' correc- (negative control). After washing in PBS, sections were tion. incubated with a biotinylated second-layer antibody for 30 minutes at room temperature (swine antirabbit RESULTS [DAKO, High Wycombe, Buckinghamshire, UK]; poly- clonal primary antibodies) or rabbit antimouse (DAKO; Patient Demographics monoclonal primary antibodies). After a further washing The characteristics of the control subjects and jaun- step, samples were incubated with StreptABComplex/ diced patients are shown in Table 2. The groups were HRP (DAKO) and visualized with a diaminobenzidine matched for age, sex, and the presence of neoplastic dis- chromogen (Sigma, Poole, Dorset, UK). Sections were ease. Twenty of the 27 jaundiced patients had pancreatic counterstained with hematoxylin. The numbers of positive cancer. By subjective global assessment, 5 subjects (19%) cells and the intensity of staining within cells were scored from the control group and 12 patients (37%) from