This document presents the official recommendations of the American Gastroenterological Association (AGA) on the management of malnutrition and cachexia, chronic diarrhea, and hepatobiliary disease in patients with human immunodeficiency virus infection. It was approved by the Patient Care Committee on July 15, 1996, and by the AGA Governing Board on July 27, 1996.
Human immunodeficiency virus (HIV) infection is a devastating illness with major public health implications throughout the world. Gastrointestinal and hepatobiliary complications are common in patients with acquired immunodeficiency syndrome (AIDS) and affect most patients at some point during the course of the disease. Gastrointestinal complications are important not only because of their frequency but because they result in significant morbidity, mortality, and use of health care resources. Management of these patients provides a challenge to the practicing physician for several reasons, including the atypical presentations, the broad spectrum of potential causes, and the multisystem nature of these complications. In addition, because these complications often occur in patients with advanced disease, ethical considerations must be taken into account.
The following guidelines provide recommendations for the management of HIV-infected patients with malnutrition and cachexia, diarrhea, and hepatobiliary diseases. These guidelines are based on a systematic, comprehensive review of the medical literature, which is provided in an accompanying report.1
Although weight loss is frequently multifactorial in etiology, a reduction in oral intake seems to be most critical. Reduced intake usually occurs in the setting of either systemic illnesses or oropharyngeal and/or esophageal disease. Evaluation of weight loss should be directed towards the systemic symptoms and signs that could indicate underlying disease(s). Successful treatment of the underlying disease frequently results in weight gain.
· Although no randomized, double-blind placebo-controlled trials have established that nutritional supplementation provides a survival benefit in patients with AIDS, nutritional supplementation, when clinically indicated, represents sound clinical care. Guidelines for nutritional supplementation are no different than for non-HIV-infected patients.
· Therapy with appetite stimulants can improve caloric intake and result in weight gain. However, no survival benefit or reduction in the incidence of opportunistic infections has been documented with the use of these agents.
· When nutritional supplementation is clinically indicated, enteral alimentation is the preferred route of administration.
· For patients in whom enteral alimentation is contraindicated, total parenteral nutrition may be indicated under selected conditions. The use of long-term total parenteral nutrition should be reserved for those patients in whom enteral nutrition is contraindicated and a meaningful life expectancy would be anticipated if nutritional therapy was given.
Diagnostic workup to identify the cause of HIV-related chronic diarrhea rests on the assumption that identifying and treating the cause(s) improves patient outcomes.
· A noninvasive evaluation, including stool test results (bacterial culture, parasite examination), should be initially performed. The number of stool tests that should be obtained is unknown, although three sets is recommended. Specific stool tests for cryptosporidia and microsporidia should be performed in patients with severe immunodeficiency (CD4 lymphocyte count, <200/mm3). Testing for Clostridium difficile stool toxin should be performed in the appropriate clinical setting. Blood cultures should be obtained in febrile patients. In those with a CD4 of <100/mm3, this should include mycobacterial culture.
· Upper endoscopy and flexible sigmoidoscopy or colonoscopy increase the diagnostic yield of enteric pathogens in patients in whom stool test results show no pathogen. The choice of upper or lower endoscopic evaluation should be tailored to the patient's clinical presentation and CD4 lymphocyte count. In patients with nonbloody diarrhea and negative stool test results, flexible sigmoidoscopy is a reasonable next step.
· Patients are at risk for cytomegalovirus colitis when the CD4 count is <100/mm3. The diagnosis is established by mucosal biopsy, which necessitates endoscopic evaluation. Flexible sigmoidoscopy and biopsy will be diagnostic in most cases. Colonoscopy is indicated for those patients in whom flexible sigmoidoscopy is negative and/or proximal colonic disease is suspected.
· Patients are at risk for small bowel Mycobacterium avium complex infection when the CD4 count is <100/mm3. The diagnosis is established by upper endoscopy and duodenal biopsy. If blood cultures are positive for mycobacteria, endoscopy and biopsy is unnecessary, and antimicrobial therapy may be initiated.
The causes of hepatobiliary disease differ depending on the extent of immunocompromise. In earlierstage HIV infection (CD4 cell count, >500/mm3), hepatic complications usually represent liver-specific processes, such as drug-related hepatotoxicity, primary neoplasms, or infection with hepatotrophic viruses. With progression of immunodeficiency to AIDS (CD4 cell count, <200/mm3), the liver is generally involved as part of a systemic opportunistic infection due to M. avium complex, fungi, or cytomegalovirus.
Because the liver is often one of many sites involved with disseminated opportunistic infections in patients with AIDS, liver disease per se is rarely the primary cause of death. Nonetheless, the progressive jaundice, fever, and/or abdominal pain that may accompany hepatobiliary disease can significantly reduce quality of life and therefore merits evaluation in most patients.
· The evaluation of suspected hepatobiliary disease should be tailored to the history, physical examination, pattern of liver tests, and CD4 lymphocyte count.
· Opportunistic infections and neoplasms usually involve the liver secondarily through lymphohematogenous dissemination. Thus, initial evaluation of blood or bone marrow specimens may provide an inferential diagnosis of liver disease.
· Ultrasonography is the test of choice when biliary disease is suspected. Abdominal computed tomographic scan is best reserved for patients with jaundice, marked hepatomegaly, or suspected peritoneal disease.
· Percutaneous liver biopsy should be reserved for patients in whom a treatable cause of parenchymal liver disease is suspected and when blood cultures and other noninvasive tests do not reveal a cause. It is unclear whether percutaneous liver biopsy is associated with a higher rate of complications in these patients.
This literature review and the recommendations therein were prepared for the American Gastroenterological Association Patient Care Committee. Following external review, the paper was approved by the Committee on May 19, 1996.
Disorders of the gastrointestinal tract and hepatobiliary system are among the most common complications associated with human immunodeficiency virus (HIV) infection. These disorders not only result in major morbidity but mortality as well. With increasing use of prophylaxis against Pneumocystis carinii pneumonia, the incidence of opportunistic gastrointestinal disorders has increased.1-3 Significant progress has been made in the last decade in characterizing the spectrum of pathogens involving the gastrointestinal and hepatobiliary systems, determining the pathophysiological mechanisms of these diverse processes, and defining management options. Despite these encouraging advancements, many questions remain unanswered. The purpose of this review is to evaluate and synthesize the published clinical research pertaining to three important HIV-related complications: malnutrition and cachexia, chronic diarrhea, and hepatobiliary disease. A separate review addresses disorders of the esophagus.4 Attention here is focused on etiology and pathogenesis, clinical features, diagnostic strategies, and efficacy of current treatment options. Recommendations for managing patients with these complications are provided based on the weight of the clinical evidence.
Literature searches for the section on malnutrition and cachexia were performed for the period from 1980 to 1994 using MEDLINE and the medical subject heading terms "cachexia" and "malnutrition" and cross-referenced with "HIV" and "AIDS." Secondary searches were performed using the terms "nutrition" and "malabsorption." Relevant articles for the section on chronic diarrhea were identified by searching the MEDLINE database for the period from 1989 to 1994. The initial search strategy used the medical subject heading terms "chronic diarrhea" and "diarrhea" cross-referenced with "HIV" and "AIDS." Secondary searches were undertaken using "enteric bacteria," "enteric parasite," "enteric pathogen," "mycobacteria," "cryptosporidia," "microsporidia," and "cytomegalovirus," which were cross-referenced with "HIV," "AIDS," "intestine," "colitis," and "gut." Articles for review for the section on hepatobiliary disease were obtained by searching the MEDLINE database for the period from 1985 to 1994 using the search terms "AIDS and liver," "HIV and liver," and "hepatitis and HIV." We identified further articles from the references cited in the literature identified by the MEDLINE searches. In selecting articles to include in this review, we gave preference to primary rather than secondary sources such as review articles and book chapters. We also gave preferences to larger series (n > 15) when available. Abstracts, articles from developing countries, and articles not published in English were excluded. Letters were included when they contained the only available data on a specific issue that was considered necessary for completeness. Articles pertaining to pediatric HIV infection were not reviewed.
Since the initial descriptions of acquired immunodeficiency syndrome (AIDS), weight loss has been recognized as a cardinal feature. The HIV wasting syndrome, which is an AIDS-defining illness, is defined as involuntary weight loss of >10% of baseline body weight plus either chronic diarrhea (at least two loose stools per day for 
30 days) or chronic weakness and documented fever (for at least 30 days and either intermittent or constant) in the absence of a concurrent illness or condition other than HIV infection that explains these findings.5 Given that opportunistic infections and neoplasms frequently cause fever, weight loss, and diarrhea, only a small subset of patients may truly have wasting in the absence of these complications.
Studies evaluating the nutritional status of hospitalized patients with AIDS have documented that 60% or more are underweight (>10% weight loss) at the time of admission.6-10 Even asymptomatic patients in the early stages of disease may have a reduction in body cell mass.11 A retrospective cohort study of 258 HIV-infected patients found a 37% incidence of weight loss during a 36-month period; weight loss was associated with a CD4 count of <500/mm3 (approximate normal range, 500-1500 mm3).12 In a study of 186 patients undergoing either upper or lower endoscopy for clinical indications, weight loss was reported in 26% of patients in Centers for Disease Control and Prevention (CDC) category IV disease compared with 7% of patients in categories II or III.13 Of a total of 147,225 adults or adolescents with AIDS reported to the CDC from September 1, 1987, to August 31, 1991, 17.8% had the HIV wasting syndrome, in 7.1% of whom it was the AIDS-defining illness.14 Geographic variations in the prevalence of wasting have been reported; in Puerto Rico, 47% of patients with AIDS had wasting in contrast to 11% in the northeastern United States.14 In developing countries, the symptom complex of weight loss, fever, and/or chronic diarrhea is particularly common and has been termed "slim disease."15-17 The diagnosis of wasting syndrome may be frequent in developing countries and in earlier published studies from the United States because complete evaluation for enteric opportunistic infections was not performed.
A number of risk factors for weight loss have been identified. Wasting has been associated with injection drug use as the route of HIV exposure (subsequently referred to as transmission category).14 Specific disorders most frequently associated with wasting included isosporiasis, pulmonary and/or esophageal candidiasis, HIV encephalopathy, chronic mucocutaneous herpes simplex virus infection, and coccidioidomycosis.14 Gastrointestinal cytomegalovirus (CMV) disease is also commonly accompanied by weight loss.13 Disseminated Mycobacterium tuberculosis plays an important role in weight loss in developing countries.18
Two prospective studies have determined risk factors for weight loss. In a cohort of 104 HIV-infected patients,19 a single risk factor was predominant in most patients with anorexia being most frequent. Other important risk factors included diarrhea, fever, and acute infections.19 In 1809 participants followed up semiannually for up to 6 years in the Multicenter AIDS Cohort Study,20 fever, thrush, and a CD4 count of <100/mm3 were each independently associated with a decrease in body mass. Although diarrhea was important, it was not as strongly associated with weight loss as these other factors. No significant decrease in body mass was found for those with a CD4 count of >200/mm3. Elevations in 
2-microglobulin21 and urinary neopterin22 levels have also been linked to subsequent weight loss, suggesting the importance of immune system activation.
Malnutrition has long been recognized to have deleterious effects on immune function.23 In non-HIV-infected patients, findings caused by malnutrition include alterations in B cell and phagocyte function as well as decreased secretory immunoglobulin A and complement activity. Lymphopenia commonly accompanies malnutrition, and reductions in B and T lymphocytes are well recognized. Non-HIV-infected malnourished patients also have decreased CD4 cells as well as moderate reductions in CD8 cells, which may normalize with improvement of nutritional status. These immune system derangements may play a role in the increased risk of infection found in malnourished patients. Nevertheless, it is often difficult to separate out the effect of malnutrition from the underlying disease on immune dysfunction.
Micronutrient deficiencies have been documented to result in immune system dysfunction. Deficiencies of zinc, iron, fat-soluble vitamins, and vitamins C and B have been linked to defects in cellular immunity in non-HIV-infected individuals.23-27 In HIV-infected patients, reduced serum concentrations of zinc, vitamin B12, and some fat-soluble vitamins have been reported,24-28 which may contribute to immunodeficiency.25,26,28 However, clinically significant improvements in immune function after repletion of micronutrient deficiencies have not been well documented in patients with AIDS.
Both retrospective and prospective studies suggest a strong link between nutritional status and survival in patients with AIDS.29-34 Kotler et al.29 observed a critical level of body cell mass and weight at which death occurred (≈66% of ideal body weight) independent of the cause of wasting. In a study of 71 patients, death was strongly associated with reduced serum albumin concentration and percent weight loss.30 Similar observations by Guenter et al.31 indicate that this relationship is maintained even after controlling for CD4 lymphocyte count. In a prospective case-control study,32 a CD4 count of <200/mm3, wasting, and vitamin A deficiency were found to be independent risk factors for death. In a study of 100 outpatients, body cell mass of >30% of body weight or a serum albumin level of >3 mg/dL were associated with a longer survival independent of CD4 count.34 Body cell mass may be a better predictor of survival than weight loss; however, weight is a more readily available clinical measure. Although evaluation of a larger number of patients would be required to evaluate the importance of other factors, such as specific AIDS-defining conditions, the evidence strongly suggests that nutritional status is an important predictor of survival. Whether poor nutritional status is an additional cause of immune deficiency is not known.
A number of pathophysiological mechanisms have been documented as potential causes of weight loss in HIV-infected patients (Table 1). More than one mechanism may be operative at one or more times during the course of HIV infection, particularly in the late stages of disease. These mechanisms will be briefly reviewed as a background for understanding treatment options.
| Inadequate oral intake | |
|---|---|
| Intestinal malabsorption | |
| Alterations in metabolism | |
| Endocrine dysfunction | |
| Cytokines |
Sharkey et al.35 evaluated nutritional status and food intake for a 1-week period in HIV-infected and HIV-seronegative homosexual men. The HIV-infected patients were clinically stable (no detectable opportunistic infection, afebrile), were in the later stages of immunodeficiency, had >90% of ideal body weight, and were free of enteric pathogens. Overall, HIV-infected patients were thinner. As a whole, no differences were detected in food intake between the groups. Intake seemed to be compromised only in those patients with very low CD4 counts and reduced intake correlated with weight loss, suggesting a cause and effect relationship. Another study36 compared nutritional intake in clinically stable patients with AIDS, patients with AIDS-related complex, and asymptomatic HIV-infected controls. Similar to the study by Sharkey et al.,35 no overall differences in oral intake were observed among the three groups during the 72-hour recording period. Kotler et al.37 found no difference in weight as percent of ideal between clinically stable patients with AIDS and homosexual or heterosexual HIV-seronegative patients, although body cell mass, as determined by total body potassium, was significantly reduced in the patients with AIDS. In the study by Kotler et al.,37 caloric intake during the 6-week monitoring period was no different among the three groups, although intestinal absorption of both xylose and lipids was significantly reduced in the patients with AIDS. These and other studies35-38 suggest that, during periods of up to 6 weeks, stable patients with AIDS have adequate oral intake; thus, it seems that other mechanisms are important in these patients.
Studies evaluating oral intake in HIV-infected patients with acute opportunistic infections have found significant reductions in caloric intake.39,40 In a longitudinal study of 27 HIV-infected men, assessments of energy metabolism during stable periods were compared with episodes of weight loss.40 Most of these episodes were associated with a documented secondary infection. Although resting energy expenditures were higher in the HIV-infected men than in controls (9.6%), no differences were found in total energy expenditures. During weight loss episodes, total energy expenditures were reduced, primarily related to reductions in activity. However, evaluation of energy intake showed that the reduced oral intake exceeded the reduction in total energy expenditures. These studies suggest that, in patients with AIDS, acute systemic infections cause significant reductions in oral intake and play the major role in short-term weight loss (see below).
HIV-related chronic diarrhea is frequently accompanied by weight loss, particularly in those with more severe reductions in CD4 count. Evaluation of gastrointestinal function in these patients has documented reductions in D-xylose and fat absorption.41-46 The degree of fat malabsorption in some patients with diarrhea may be striking (i.e., >20 g/day).47 In one study,48 serum carotene, a marker of fat malabsorption, was decreased in 77% of HIV-infected patients and was associated with a decreased CD4 count. Vitamin B12 deficiency, which may be detected in 10%-20% of patients with AIDS,45,46 has been considered the result of intestinal malabsorption. Protein exudation49 may contribute to hypoalbuminemia, which is common in patients with AIDS.10,50 Pancreatic insufficiency does not seem to be the cause of fat malabsorption in patients with AIDS with diarrhea.51
In patients with diarrhea, abnormalities of small bowel morphology include chronic inflammation and variable degrees of villous flattening, which may be accompanied by malabsorption52-61 (see section on AIDS enteropathy). In some patients, enteric pathogens may explain these findings.54-56 However, even in patients in whom a thorough evaluation excludes pathogens, reductions in D-xylose and fat absorption and brush border enzyme activity may be observed.43,47,52,57,58
The normal response to malnutrition caused by fasting is a reduction in metabolic rate (resting energy expenditures).39 In contrast, patients with severe acute infections or critically ill patients typically have an increase in metabolic rate.39 Most studies have documented an increase in resting energy expenditures depending on the stage of immunodeficiency (denoted by the CD4 count) and the presence of active infections.39,40,62-64 In these studies, metabolic rate is typically measured by indirect calorimetry. The resting metabolic rate in HIV-infected patients with a normal CD4 count was 8% greater than observed in a group of HIV seronegative patients.64 Compared with healthy controls, patients with AIDS and active infections had a 34% increase in metabolic rate; stable patients with AIDS were found to have a 21% increase.65 Grunfeld et al.39 measured resting energy expenditures and caloric intake in HIV-infected patients and normal subjects. As documented in other studies,62-64 resting energy expenditure was increased in HIV-infected patients compared with controls. These increases were as follows: HIV infected without immunodeficiency, 11%; AIDS without systemic infections, 25%; and AIDS with systemic infection, 29%. Caloric intake in HIV-infected patients was similar to the controls, and no change in weight was observed except for the patients with AIDS with systemic infections in whom a 36% reduction in caloric intake occurred, accompanied by a 5% decrease in weight. Although smaller studies have reported a decrease in metabolic rate in stable patients with AIDS,36,66 patients with AIDS with systemic infections maintain an increased metabolic rate despite decreased caloric intake, similar to patients with sepsis, burns, or major trauma.
The pattern of weight loss may vary, depending on the patient's clinical manifestations of HIV infection. Acute weight loss may occur before or during acute infections or may occur as a preterminal event.67,68 In patients who recover from an acute infection (e.g., P. carinii pneumonia), weight may return to baseline without specific nutritional intervention.67 In contrast, patients with chronic progressive weight loss were more likely to have a gastrointestinal complication, particularly a diarrheal illness.67
These data in aggregate suggest that, even in early HIV infection, resting energy expenditures may be mildly increased although weight remains stable. In the absence of opportunistic infections, patients with AIDS may maintain a stable weight despite an increase in resting energy expenditures by appropriately increasing caloric intake and/or decreasing activity. An acute systemic illness seems to cause weight loss primarily through significant reductions in oral intake, although increased energy expenditures play an important role. With effective therapy for the systemic illness, weight gain may occur without specific nutritional interventions,69 although full recovery of weight may not necessarily occur with repeated infections. Progressive weight loss seems to be more characteristic of patients with gastrointestinal complications in whom decreased oral intake and malabsorption may occur. In patients with multiple simultaneous infections (e.g., P. carinii pneumonia and intestinal cryptosporidiosis), weight loss may result from multiple pathophysiological mechanisms.
Disturbances in a variety of endocrine systems have been documented in patients with AIDS. A reduction in serum testosterone hormone concentration, adrenal dysfunction, and an alteration in thyroid hormone levels have been documented,70 especially in patients with severe wasting,71 although not consistently.63,64 However, it is unclear whether these hormonal aberrations are causally related to wasting. In contrast to immunocompetent patients with severe systemic illness in whom serum T3 concentrations are reduced and reverse T3 concentrations are increased (euthyroid sick syndrome), some patients with AIDS maintain normal T3 concentrations, suggesting that thyroid function may be "overstimulated" for the degree of wasting, thus potentially contributing to weight loss.72,73 However, thyroid stimulation testing is usually normal in these patients.74 Grunfeld et al.74 found that thyroid hormone concentrations were normal in asymptomatic HIV-infected patients but decreased appropriately in those with systemic illnesses. In addition, for those with acute infections, anorexia, and weight loss, reduction in T3 concentrations occurred, consistent with euthyroid sick syndrome. In these patients, the resting energy expenditures were greater than in asymptomatic patients or normal individuals. These studies taken together suggest that endocrine dysfunction plays a minor role, if any, in HIV-related wasting.
Tumor necrosis factor has been suggested as a potential etiologic factor in HIV wasting syndrome. Conflicting data exist regarding serum concentrations of tumor necrosis factor.72,75-78 Recent studies using improved bioassays show elevated tumor necrosis factor levels in HIV-infected patients, which progressively increase with increasing immunodeficiency regardless of concurrent illness.79 Immunocompetent HIV-negative patients with acute illness and intravenous drug users may have slightly increased concentrations of tumor necrosis factor.76 Although an increased serum concentration of tumor necrosis factor may induce anorexia, its contribution to weight loss is unknown. In one study, serum concentrations of soluble tumor necrosis factor were significantly increased in HIV-infected patients compared with controls; there was an associated decrease in serum albumin and transferrin concentrations and reduced body cell mass.80 Whether the interaction of tumor necrosis factor with other cytokines such as interleukin 1 plays a role in wasting remains undefined.81 As a determinant of wasting, the paracrine effects of tumor necrosis factor may be more important in vital tissues rather than the serum concentrations.82 Despite the association of increased serum tumor necrosis factor concentrations with disease progression, the precise role of this cytokine in weight loss remains undefined.
As with any malnourished patient, repletion of nutritional deficiency represents sound clinical care. The goals of nutritional therapy should be to minimize loss of body mass, prevent micronutrient deficiencies, and restore body weight. Patient education regarding diet alone has been shown to significantly increase caloric intake.83 Although an aggressive approach to nutritional therapy has been advocated,84 its effects on outcome measures such as the incidence of opportunistic infections or mortality have not been documented.
A prospective randomized nonblinded trial of 80 patients compared a routine diet supplemented with standard enteral formula (Ensure; Ross Products, Columbus, OH), with a new peptide-based formula85 that contained greater amounts of fiber and carbohydrate content, less fat, and a "patented" polypeptide. The 56 patients (70%) who completed the trial (25 receiving Ensure and 31 receiving the new formula) consumed equal quantities of the treatment formula (approximately 1½ cans per day), although less than the targeted amount. Both formulas were well tolerated. Patients in the new formula group had a net weight gain of 4 lb compared with a loss of 1.5 lb in the standard formula group (P = 0.04). Interestingly, more patients receiving the standard formula were hospitalized after 3 months of treatment. The reasons for hospitalization were not reported, and survival differences were not evaluated. Although statistically significant differences were detected in some nutritional parameters such as triceps skin-fold thickness, these were not meaningful clinically.
Studies of the safety and efficacy of nutritional supplementation by percutaneous endoscopic gastrostomy tubes are limited. In a 2-month study of 6 patients,86 a 14% increase in body mass was observed, which was accompanied by an increase in body fat content. However, no change in weight was detected from baseline. Although no procedure-related complications were reported, a subsequent case-control study suggests that HIV-infected patients are at a higher risk for major complications such as infection.87
Although personal clinical experience suggests that total parenteral nutrition (TPN) is frequently administered to patients with AIDS, there are no studies documenting the frequency of its use. Few studies have evaluated the nutritional effects and safety of TPN in HIV-infected patients. Kotler et al. studied the effects of TPN in 12 malnourished patients with AIDS.88 Malnutrition was caused by reduced oral intake or malabsorption in 5 patients and by underlying systemic infections with or without a malabsorption syndrome in 7 patients. Nutritional assessments were performed monthly for 4-42 weeks of follow-up. For the full study group, no change in mean body cell mass was observed, although body fat content increased significantly. The 5 patients in whom malnutrition was caused by reduced oral intake or malabsorption experienced weight gain and repletion of body cell mass, whereas only 2 of 7 patients with systemic infections had an increase in body cell mass; these differences were not significant. In a retrospective study of 22 patients with AIDS, TPN administration for a period of 1-11 months was evaluated.89 In these patients, 10 had an underlying neoplasm (Kaposi's sarcoma or lymphoma) and 9 developed Pneumocystis carinii pneumonia during the study period. In contrast to the findings of Kotler et al.,88 15 of the 22 patients gained weight, 5 maintained their weight, and only 2 lost weight.89 Metabolic disturbances related to TPN were uncommon, and the overall complication rate was low (0.12/100 catheter days). However, 2 patients developed catheter infections and 9 had to have their catheters replaced for other complications such as venous thrombosis. No difference in catheter-related complications was found when compared with a historical control group of patients with cancer receiving TPN. These small studies88,89 suggest that TPN is relatively safe; larger studies of HIV-infected patients receiving TPN or intravenous therapy for opportunistic infections by a central venous catheter have documented variable complication rates of 0.13-0.47/100 catheter days.90-92
In summary, weight gain may be achieved in some patients using enteral nutrition or TPN. However, the safety of TPN has not been evaluated in a sufficiently large number of patients to draw firm conclusions. In addition, weight gain has been modest at best, and no significant reduction in the incidence of opportunistic infections or hospitalizations or improvement in survival has been documented. Trials of TPN in patients with cancer have similarly failed to show improved survival.93 Nutritional supplementation may benefit selected patients; however, the available data do not help identify these patients or indicate which therapy is more appropriate.
Given the prevalence of anorexia, particularly in patients with end-stage disease, appetite stimulants have received attention. Megesterol acetate, a synthetic progestational agent, has been associated with appetite stimulation, an improved sense of well-being, and weight gain in patients with breast cancer.94 Following a pilot study showing weight gain in patients with AIDS,95 randomized placebo controlled trials were undertaken.96,97 In these two trials, entry criteria included AIDS with a 20% or greater decrease in body weight and a minimum weight loss of 10% below ideal body weight. Patients with impaired food intake, overt gastrointestinal disorders resulting in malabsorption (specific tests for malabsorbtion were not performed), dementia, malignancies, or opportunistic infections were excluded. One trial97 compared placebo with 800 mg daily of megesterol acetate, whereas the other96 was a dose-ranging trial comparing placebo with 100, 400, and 800 mg. The results from the two trials were remarkably similar. In both, caloric intake and body weight increased and sense of well-being improved in the patients treated with megesterol acetate. No difference in Karnofsky scores, opportunistic infections, or survival were observed between the treatment groups in either trial. Megesterol acetate was well tolerated with no significant side effects. However, one may question the clinical importance of a weight gain of 9-17 lb, particularly because this resulted from increased body fat and not lean body mass, which may be a more important determinant of survival.29 Finally, although an improved sense of well-being is important to patients, this must be weighed against the cost of the agent as well as lack of any survival benefit or reduction in the occurrence of opportunistic infections.
Dronabinol, the major active ingredient in marijuana, has also been used as an appetite stimulant.98 A multicenter randomized double-blind placebo-controlled trial of 139 HIV-infected patients was performed to evaluate the efficacy of this agent.99 In the 88 evaluable patients (63%), dronabinol was associated with statistically significant improvements in appetite and mood. However, no significant differences in weight were observed during the 6-week study period, and those who received dronabinol were twice as likely to discontinue therapy because of side effects. Survival differences were not reported.
A variety of other pharmacological therapies are currently undergoing evaluation. Pentoxifylline, a drug that reduces tumor necrosis factor concentrations and HIV replication,100 has been evaluated with inconclusive results in a pilot study of 5 patients.101 Growth hormone is also currently under investigation. In a study of 10 patients, intravenous administration of insulin-like growth factor resulted in only transient nitrogen retention, suggesting partial resistance to the anabolic effects of this drug.102 These novel pharmacological treatments must be evaluated in carefully conducted randomized clinical trials to determine their effect on overall patient nutritional status and/or survival.
Malnutrition is an important complication of HIV infection, although weight loss is not universal even in the later stages of disease. Alterations in metabolism that can be documented in early HIV infection become more pronounced in the later stages of disease. Opportunistic infections not only cause altered body metabolism but are also associated with reduced oral intake, which seems to be the most important determinant of weight loss. In patients with weight loss, the identification of treatable opportunistic infections, including gastrointestinal tract complications, should be the first concern. In the absence of any identifiable cause of weight loss, nutritional assessment should be performed followed by dietary counseling to increase nutritional intake. Although enteral supplements may promote weight gain, their effects have not been well documented. The use of TPN may be inappropriately aggressive, particularly in patients with advanced disease. In addition, no randomized clinical trials of TPN in patients with AIDS have been performed to evaluate the effects on quality of life and survival. Nevertheless, given the relationship between nutritional status and survival as well as the use of TPN in other situations in which data are lacking, the use of TPN for patients in whom gut function is inadequate may be appropriate. Although treatment with appetite stimulants has physiological rationale, randomized clinical trials have not documented meaningful improvement in important end points such as the occurrence of opportunistic infections or mortality. What these studies have shown are improvements in the patient's sense of well-being and modest weight gain, primarily body fat. Further studies are clearly needed to identify which patients are likely to benefit from nutritional supplementation and to determine the best approach.
Accurate definition is fundamental to systematic investigation and precise communication. The CDC defines chronic diarrhea as two or more loose stools per day for 30 days.5 However, this definition fails to capture the spectrum and severity of diarrheal patterns observed in clinical practice. For example, using this definition, no distinction is made between an individual passing three mushy stools per day who is otherwise functionally unimpaired and a bedridden patient who is incontinent with profuse, watery diarrhea. Given the lack of a valid, reliable, and widely accepted method for classifying diarrhea pattern and severity, various definitions have been used, making comparison among the published studies difficult.
In addition, from the standpoint of patient management, other features of the diarrheal illness beyond stool frequency are important because they provide clues to the etiology. As in non-HIV-infected patients, the passage of large-volume stools associated with periumbilical pain usually indicates small bowel disease. On the other hand, the passage of frequent, small-volume stools (which may be bloody) associated with urgency, tenesmus, and lower abdominal cramps or perianal pain usually indicates colonic and/or anorectal disease. The presence of fever is also important, because it may indicate mucosal invasive disease or the presence of an associated bacteremia.
Although diarrhea is frequently stated to be very common in HIV-infected individuals, few estimates of the magnitude of the risk of developing diarrhea are available. Two factors that predict a higher risk of diarrhea are decreased CD4 cell count and male homosexuality. In the Multicenter AIDS Cohort Study, which evaluated 493 HIV-infected homosexual men at two 6-month intervals, 0.9% of men with CD4 cell counts of >700/mm3 had diarrhea compared with 6% of men with CD4 counts of <250/mm3.103 A retrospective cohort study of 169 predominantly homosexual HIV-infected men, one third of whom had CD4 cell counts of <200/mm3, was performed for a 4-year period. The prevalence of diarrhea at any 6-month outpatient visit was 3%-7%.104 Diarrhea was three times more common among homosexual men than among those in other transmission categories.104 In a retrospective survey of gastrointestinal symptoms in 258 predominantly homosexual HIV-infected outpatients with a mean CD4 cell count of 502/mm3, 14% had diarrhea at the initial outpatient clinic visit.12 Thus, in these outpatient studies, the prevalence of diarrhea at any single visit ranged from 0.9% to 14%. On the other hand, in studies focusing on patients with more advanced disease, diarrhea is more common. For example, in a small study (n = 22) of severely ill hospitalized patients with AIDS, 50% had diarrhea.105
Although HIV-related chronic diarrhea can be attributed to infection with one or more common or opportunistic enteric pathogens in many patients, no cause can be identified despite intensive investigations in a significant proportion. Several hypotheses have been proposed to account for these cases of unexplained chronic diarrhea, including direct mucosal HIV infection. Although a full appraisal of the laboratory evidence supporting these hypotheses is beyond the scope of this report, the issues are summarized in a subsequent section. Because the goal of this report is to evaluate the clinical research and develop clinical guidelines for patient management based on this evidence, the main focus here is on enteric pathogens because they play the major role in HIV-related chronic diarrhea.
The prevalence and spectrum of enteric pathogens vary in published reports for at least two reasons. First, patient characteristics are important. The prevalence of enteric pathogens is greater among homosexual men,106 those with weight loss,107,108 and probably those with prolonged diarrhea.109 Another important patient characteristic is the severity of the immunologic deficit. For example, pathogens such as Mycobacterium avium complex (MAC) will not be observed unless patients with more advanced disease, denoted by a CD4+ cell count of <100/mm3, are included.110 Second, the prevalence and spectrum of enteric pathogens depends on the intensity of diagnostic workup. If diagnostic workup does not include endoscopic mucosal biopsies, organisms such as CMV will not be detected.111 Table 2 summarizes seven relevant published studies and lists the two most common enteric pathogens reported in each study. Excluded from Table 2 are studies that focused on specific pathogens, mucosal structure, or patients with negative stool examinations. All of the studies listed in Table 2 are cross-sectional. In the retrospective study,106 only patients with diarrhea were evaluated. Chronic diarrhea was not defined in three studies,105,106,112 and the definitions varied in the remainder. The patients were referred for gastroenterological workup in all but one study.113 The patients were severely ill, and all had AIDS except for one study in which 75% had AIDS.114 Inpatients, who tend to be sicker, were included in the six studies in which this could be determined105,106,113-116 and were the sole focus in two studies.105,106 For the four studies in which patient characteristics could be determined, the mean CD4 cell count was 62/mm3 and mean weight loss was 8.2 kg in the first study,115 the mean CD4 cell count was 126/mm3 and mean weight loss was 8.6% in the second study,114 the mean weight loss was 15 kg in the third study,105 and >15% body weight loss was present in 64% of patients in the fourth study.116 Taken together, in the six studies that reported the transmission category,105,106,113-116 two thirds of the patients were homosexual men. Diagnostic workup was intense. In six studies, esophagogastroduodenoscopy and colonoscopy or flexible sigmoidoscopy with biopsies was performed in all115,116 or in selected patients.105,106,112-114 Examination of biopsy tissue included viral culture in two studies.115,116
| Prevalence of pathogens | |||||
|---|---|---|---|---|---|
| Reference | (%)a | Intensity of diagnostic workupb | Patients with diarrhea (%) | Patients without diarrhea (%) | Enteric pathogenc |
| Dworkin et al.105 | 22 (55) | Stools | 38e | MAC Cryptosporidia | |
| Laughon et al.113 | 77 (64) | Stools | 50 | 11 | Cryptosporidia Campylobacter species |
| Smith et al.115 | 30 (67) | Stools, EGD, colonoscopy | 85 | 10 | CMV Entamoeba histolytica |
| Antony et al.106 | 66 (100) | Stools | 55 | NA | MAC CMV |
| Rene et al.116 | 132 (52) | Stools, EGD, colonoscopy | 59 | 28 | Cryptosporidia CMV |
| Cotte et al.114 | 81 (73) | Stools | 64 | 15 | Cryptosporidia CMV |
| Kotler and Orenstein112 | 194 (73) | Stoolsd | 83 | 2 | Microsporidia Cryptosporidia |
aNumber of patients studied and proportion with diarrhea (%).
bThe two most common organisms detected in the patients with diarrhea are listed, except for Dworkin et al.105 and Rene et al.,116 in which the overall results are given.
cEndoscopic procedures are listed only if they were performed in all patients.
dStool examinations performed in patients with diarrhea.
ePrevalence of pathogens (38%) not reported separately for patients with and without diarrhea.
EGD, esophagogastroduodenoscopy.
In this section, the focus is on the epidemiology, clinical features, diagnosis, and treatment of the four most commonly reported opportunistic enteric pathogens listed in Table 2.
Organisms of the MAC comprise two closely related species: M. avium and M. intracellulare. The organism is ubiquitous in the environment, and disseminated infection results from recent infection rather than reactivation of previous infection.117 The gastrointestinal tract is an important portal of entry from which the organism may disseminate, especially to the blood, bone marrow, liver, and lymph nodes. The major risk factor for MAC infection is the severity of the immunologic deficit. Infection is rare in patients with CD4 cell counts of >100/mm3.110,118 No differences have been observed in the frequency of MAC infection across transmission categories.118,119 The widespread use of prophylaxis against P. carinii pneumonia has prevented or delayed death from this infection, shifting clinical manifestations to illnesses that occur when immune function is more severely suppressed, such as MAC. For example, among 844 homosexual men in the Multicenter AIDS Cohort Study diagnosed with AIDS by January 1992 and followed up until April 1993, MAC infection developed in 33% of men who received prophylaxis compared with 17% of men who did not receive prophylaxis.1
The largest case series is a retrospective study of 114 patients with AIDS in whom MAC was first diagnosed by blood culture.120 The most common findings at diagnosis were fever and severe anemia (denoted by a hematocrit of <26%). Night sweats and 10% weight loss were less common. Findings referable to the gastrointestinal or hepatobiliary tract, which were present in 76% of patients, included diarrhea, abdominal pain, hepatomegaly, and increased alkaline phosphatase levels.
The largest published case series of patients with gastrointestinal MAC infection is a retrospective study of 35 patients with AIDS.121 The patients had been referred for gastrointestinal evaluation because of weight loss (30 patients); fever, sweats, or chills (21 patients); diarrhea (17 patients); and abdominal pain (7 patients). Endoscopic biopsy specimens showed duodenal involvement in 30 of 34 patients (88%). Unusual, characteristic white nodules (1-3 mm diameter) were observed in the duodenum in 12 patients. Of 11 patients who underwent flexible sigmoidoscopy with biopsies, 7 had involvement of the rectum. Gastrointestinal tract involvement was associated with dissemination; 25 of 28 men (89%) had bone marrow infection and 28 of 33 (85%) had bacteremia.
Although the gastrointestinal tract is a major portal of entry, gut colonization is uncommon, and routine screening of stool specimens with acid-fast stains and mycobacterial cultures is not an effective method for detecting MAC before dissemination.122,123
Disseminated MAC infection is most readily diagnosed by mycobacterial culture of blood or bone marrow. Using commercially available rapid blood culture systems, the average time required to establish a diagnosis can be decreased to 5-12 days.124 A single mycobacterial blood culture is adequate.125 Disseminated MAC infection may be diagnosed from mycobacterial culture of lymph node or liver biopsy specimens. New methods based on DNA probes are becoming available to replace the conventional approach of mycobacterial identification from colony growth; polymerase chain reaction (PCR) is being evaluated for detecting the organism in peripheral blood.
In the diagnosis of gastrointestinal infection using stool samples, mycobacterial culture is more sensitive than acid-fast stain.126,127 For patients undergoing gastrointestinal endoscopy, MAC infection can be detected in mucosal biopsy specimens examined under light microscopy using H&E and Ziehl-Neelsen staining. However, because granuloma formation is poor128,129 and Ziehl-Neelsen stains may be negative, mycobacterial culture of mucosal biopsy specimens is needed to increase the diagnostic yield for patients in whom the diagnosis is suspected.
Two randomized placebo-controlled clinical trials showed that rifabutin prophylaxis (300 mg once daily by mouth) reduced the frequency of MAC bacteremia in patients with AIDS with CD4 cell counts of 
200/mm3, although survival was not affected.130 On the basis of these results, a U.S. Public Health Service Task Force recommends rifabutin prophylaxis for HIV-infected patients with CD4 cell counts of <100/mm3.131
MAC infection occurs in patients with advanced disease; in the large case series cited above, the median survival after the initial diagnosis was 110 days.120 No randomized placebo-controlled trial has evaluated the effect of antimicrobial therapy on survival. In uncontrolled studies, single-agent therapy with azithromycin132 or clarithromycin133 was associated with suppression of bacteremia. Kemper et al.134 reported a prospective case series of 31 patients treated for 12 weeks with rifampin, ethambutol, clofazimine, and ciprofloxacin. Suppression of bacteremia and reduction in fever at 4 weeks occurred, although many patients experienced subsequent recurrence of fever and most continued to experience weight loss. Although patients with gastrointestinal involvement may experience improvement in symptoms, it is unknown whether antimicrobial therapy is associated with sustained microbiological or clinical benefit. Nonetheless, the U.S. Public Health Service Task Force recommends treating patients with MAC infection with at least two agents, including azithromycin or clarithromycin.131 Ethambutol can be used as a second drug and one or more of the following can be added as a third or fourth drug: clofazimine, rifabutin, rifampin, ciprofloxacin, and, in some situations, amikacin. Isoniazid and pyrazinamide are not recommended.131 Adverse drug reactions are a common problem with such multiple drug regimens. In the study by Kemper et al., one or more agents had to be withdrawn in 46% of the patients.134
Organisms of the genus Cryptosporidium, which belong to the phylum Apicomplexa, are protozoans. Cryptosporidium parvum is the species that infects both humans and cattle. The modes of transmission are water contamination, which has been responsible for community outbreaks, and direct fecal-oral, which may occur during male sexual contact.
The clinical course of infection is variable and depends on at least two factors. The first is the immunologic competence of the host. In immunocompetent persons, cryptosporidiosis is a self-limited diarrheal illness with an incubation period of 7 days and a mean duration of 12 days.135 In immunodeficient HIV-infected individuals, however, infection may be prolonged. The diarrhea is nonbloody and may be associated with nausea, vomiting, and abdominal cramps.136 Patients with CD4 cell counts of >180/mm3 tend to clear the infection.137 The largest reported case series describes the clinical course of cryptosporidia infection in 128 HIV-infected patients, of whom 38 (37%) had an additional enteric infection, most commonly CMV.138 Four clinical patterns of infection were observed: (1) asymptomatic (4% of patients); (2) transient symptomatic (28%), which spontaneously resolved within 2 months; (3) chronic (60%), which persisted for at least 2 months; and (4) fulminant (8%), which was refractory to therapy and was characterized by the passage of 2 L or more of liquid stool daily. Transient disease was more common in less immunosuppressed patients, whereas fulminant disease occurred in patients with CD4 cell counts of <50/mm3.138
The second factor that may be an important determinant of the clinical course of cryptosporidiosis is the intensity of infection or number of organisms present in the intestine.139 Given the variability in the clinical course of cryptosporidia infection, the importance of randomized placebo-controlled trials in evaluating potential therapies is clear.
Although the organism can be detected using light-microscopic examination of duodenal or, less commonly, colonic mucosal biopsy specimens, the availability of stool tests has lessened the need for endoscopy in the diagnostic workup. In many clinical laboratories, routine stool examination for ova and parasites does not include testing for cryptosporidia, which must be specifically requested. Cryptosporidia oocysts can be shown in stool samples using either a modified acid-fast stain or a commercially available kit based on an immunofluorescence technique. Current evidence indicates that the commercial kit is more sensitive than acid-fast staining.140 Recently, infection with Cyclospora species has been recognized in HIV-infected patients.141 When the acid-fast stain is used to examine stool samples, it is important that the larger Cyclospora species oocysts (8-9 µm) be distinguished from the smaller Cryptosporidium species oocysts (5 µm)141 because treatment is different.
Paromomycin, an orally administered and poorly absorbed aminoglycoside, has been evaluated in uncontrolled studies, the largest of which included 24 patients.142 In this study, paromomycin therapy was associated with a reduction in stool frequency and clearance of stool oocysts in some patients.142 The only published randomized clinical trial included 10 patients who were assigned to treatment with placebo or 500 mg paromomycin by mouth three or four times daily for 2 weeks and then crossed over to the other treatment group.143 Paromomycin therapy was accompanied by a decrease in total oocyst and daily stool excretion. Other agents that have been used to treat patients with HIV-related cryptosporidiosis, reported in uncontrolled studies and case reports, include spiramycin, erythromycin, zidovudine, and bovine serum colostrum.144-147 Important but unresolved issues include the length of therapy, the use of maintenance therapy, and the efficacy of preventive measures such as drinking boiled or bottled water rather than tap water.
In immunodeficient individuals, CMV disease can result from reactivation of latent virus, reinfection with a new virus, or primary infection in a previously uninfected host. In patients with AIDS, the source of CMV disease is either reactivation or reinfection because most individuals have been previously exposed to the virus. For example, CMV is sexually transmitted among homosexual men, 90% of whom have serological evidence of infection.148 Distinguishing between latent infection and disease is important. CMV gastrointestinal disease is defined as submucosal hemorrhage, mucosal erosions, or ulcerations in which the virus is present in a person in whom other causes of the lesion(s) have been excluded.149
In a cohort study of patients with AIDS or symptomatic HIV infection followed up for 2 years, the cumulative probability of developing CMV disease was 21% for those with initial CD4 counts of <100/mm3.150 In this study, the major risk factors for CMV disease were a CD4 cell count of <100/mm3 and male homosexuality.150
CMV disease may involve any site in the gastrointestinal tract; however, small bowel and colonic disease are most relevant to this report. No large series of patients with small bowel disease exists, although case reports show that pain, diarrhea, bleeding, and perforation may occur.151-153 Patients with colonic CMV disease may present with diarrhea, passage of blood per rectum, and/or abdominal pain. In the largest case series of patients (n = 44) with CMV colitis, 31 patients (70%) had persistent diarrhea, 13 (30%) had intermittent diarrhea, 28 (64%) had crampy abdominal pain, 10 (23%) had hematochezia, and 1 had colonic perforation.154 Case reports or small series of patients with CMV colitis indicate that patchy or confluent submucosal hemorrhage, erosion, and ulceration may be observed.155,156
Because of the high prevalence of anti-CMV antibodies due to previous exposure to the virus,148 serology is not helpful in diagnosing CMV gastrointestinal disease. Viral cultures of blood and urine are also unhelpful.157 In tissues, CMV infection produces a characteristic cytopathic effect, which is a large cell (referred to as a cytomegalic cell) containing an intranuclear inclusion and, frequently, intracytoplasmic inclusions. The presence of cytomegalic cells on light-microscopic examination of mucosal biopsy specimens stained with H&E is considered the gold standard for establishing a diagnosis of CMV infection. In some patients, however, the inclusions are atypical in appearance and few in number, making a definitive diagnosis difficult.158 The detection of cytomegalic cells in mucosal biopsy specimens obtained from areas of submucosal hemorrhage, erosion, or ulceration in the absence of other causes of the lesion indicates CMV gastrointestinal disease. The frequency with which cytomegalic cells are detected depends in part on the intensity of the search (number of biopsy specimens examined, effort of the pathologist).159 Because the sensitivity of cytomegalic cells in detecting disease is <100%, other diagnostic methods have been proposed. The disadvantage of viral cultures is the length of time between specimen collection and viral growth, which may be up to 2 weeks. Also, tissue specimens may be contaminated by blood, which in the presence of viremia may give rise to positive cultures. Immunohistochemistry detects CMV antigens using monoclonal or polyclonal antibodies, and PCR amplifies and detects minimal quantities of the viral genome in tissue specimens.159 In a comparison of light-microscopic examination, PCR, and immunohistochemistry, PCR was the most sensitive method for detecting CMV infection.159 However, further evaluation of immunohistochemistry and PCR is needed before recommending that these techniques replace light-microscopic examination for routine clinical use.
Patients with gastrointestinal CMV disease may be treated with ganciclovir or foscarnet, both of which are administered intravenously. In a case series of 17 patients with gastrointestinal CMV disease who were administered 2.5 mg/kg ganciclovir every 8 hours for 10 days, 9 of 14 patients (64%) had resolution of pain and 8 of 11 (73%) had resolution of diarrhea.160 A subsequent randomized placebo-controlled trial evaluated ganciclovir (5 mg/kg) every 12 hours for 14 days in 62 patients with AIDS and CMV colitis.161 Colitis severity was graded from 0 to 3 according to gross and histological criteria. Although ganciclovir therapy was associated with improvement in colitis scores, no improvement in symptoms (diarrhea, abdominal pain) was observed. However, the duration of therapy may have been too short (14 days). Because ganciclovir suppresses but does not eradicate the virus, relapse of CMV disease may occur after therapy.161 However, no accurate estimates of the magnitude of the risk for relapse are available for colitis. A prospective study of patients with CMV esophagitis reported a relapse rate of 39% at a median of 4 months of follow-up (range, 2-18 months).162 The usual duration of full-dose treatment (referred to as induction therapy) is 3-4 weeks, at which point symptoms and endoscopic features are reassessed. If no symptomatic or endoscopic response has occurred, a repeat course of induction therapy may be administered. In patients with a symptomatic and endoscopic response, the decision whether to advise maintenance therapy with 5 mg/kg ganciclovir daily should be individualized. No studies are available regarding the use of oral ganciclovir for maintenance therapy. For patients in whom a symptomatic but not an endoscopic response occurs, management should also be individualized. The toxicities of ganciclovir include renal impairment, neutropenia, and thrombocytopenia. The risk of cytopenia is greater when the drug is administered to patients receiving zidovudine.163
The largest case series of HIV-infected persons treated with foscarnet is a retrospective study of 66 patients with gastrointestinal CMV disease.164 In this study, the most commonly used dose of foscarnet was 60 mg/kg every 8 hours for 2-3 weeks. Improvement in symptoms and endoscopic healing of lesions occurred in 16 of 28 patients (57%) with colonic disease.164 In another case series of 19 patients with AIDS and gastrointestinal CMV disease who had failed a course of ganciclovir therapy,165 60 mg/kg foscarnet was administered every 8 hours for 14 days followed by maintenance therapy in a dose of 90 or 120 mg/kg daily. After 3 weeks of therapy, improvement in symptoms and gross and histological parameters was observed.165 As with ganciclovir, many would advise maintenance foscarnet after induction therapy. A randomized open label trial that compared ganciclovir with foscarnet for gastrointestinal CMV disease showed similar efficacy.166 The toxicities of foscarnet include renal insufficiency, hypocalcemia, hypomagnesemia, hypophosphatemia, anemia, and nausea.167 Ganciclovir and foscarnet may be safely administed together in patients with a poor response to either agent alone.168
The term microsporidia is a nontaxonomic designation used to describe organisms belonging to the order Microsporidia of the phylum Microspora. Microsporidia are obligate intracellular protozoan parasites. In HIV-infected individuals, gastrointestinal infection with organisms of two genera, Enterocytozoon and Septata, have been reported. The most common microsporidian intestinal infection is attributable to Enterocytozoon bieneusi. In six studies from different institutions encompassing 494 patients with HIV-related chronic diarrhea, E. bieneusi was detected in 137 (28%; range, 14%-33%).54,112,169-172 Few studies have evaluated the frequency of microsporidiosis in HIV-infected patients without diarrhea, taking into account CD4 cell count and the presence of other enteric organisms. In two studies in which this comparison has been made,170,171 the prevalence of microsporidia in the patients without diarrhea was low (3%) in one study,170 while the prevalence was higher (25%) and not significantly different from that observed in the patients with diarrhea in the other study.171 These discrepant findings may be accounted for by differences in the methods used to detect E. bieneusi infection. Light microscopy was used in one study,170 whereas electron microscopy, which is probably a more sensitive technique, was used in the other study.171 Clearly, E. bieneusi infection can be detected in patients without diarrhea, indicating a need for further study to identify factors that determine the presence of illness.173
Recently, gastrointestinal infection with a second microsporidian species, Encephalitozoon intestinalis, has been reported in HIV-infected individuals. This organism can be distinguished from E. bieneusi using electron microscopy.174 In some patients, the diagnosis may be suggested by the larger size of S. intestinalis as well its presence in lamina propria macrophages.175
In clinical practice, the detection of microsporidia infection has been hampered by lack of a simple, noninvasive diagnostic test. Light-microscopic detection of microsporidia in paraffin sections of mucosal biopsy tissue is difficult because the organisms are small and stain poorly with routinely used histological stains, such as H&E.171 Other stains have been used in light-microscopic examination of mucosal biopsy tissue,176 including Warthin-Starry,172 Gram, Giemsa, and modified chromotrope stains.177 Giemsa-stained touch preparations178 and semithin (1 µm) plastic sections stained with toluidine blue have also been used. In the past, the gold standard for detecting the parasite has been electron microscopy.179 However, more recently, a modified trichrome stain180 and other techniques for identifying the spores in stool specimens have been reported.181 No large comparative study of the different techniques for detecting intestinal microsporidia infection has been performed.
No randomized placebo-controlled clinical trial of therapy for intestinal E. bieneusi infection has been published. The largest study is a case series of 29 patients with AIDS, chronic diarrhea, and E. bieneusi infection who were treated with 400 mg albendazole by mouth twice daily.182 Of 26 patients who completed 1 month of treatment, 13 had a 50% or greater decrease in the mean daily stool frequency. Further study of albendazole and other agents is clearly needed.
In addition to symptoms (diarrhea, weight loss), abnormalities of gastrointestinal mucosal structure and absorptive function may occur. The term "AIDS enteropathy" was first used in 1984 in the context of these abnormalities.41 However, since then, confusion has arisen regarding the term itself.183 A critical appraisal of the published clinical research54,57,58,60,61 has shown that no consensus exists regarding the term "AIDS enteropathy," to whom it applies, the elements on which it is based, and the criteria for its definition.183 To avoid misunderstanding, it may be preferable to avoid the term "AIDS enteropathy" and to simply describe the abnormalities of mucosal structure and function that are observed, such as villus flattening and decreased D-xylose absorption.
Although the pathogenesis of these abnormalities of mucosal structure and function remains unclear, several hypotheses have been proposed to account for them. First, they may be due to as yet unknown or known but undetected enteric pathogens. Second, they may be a result of direct mucosal HIV infection.41,57,60 Third, bacterial overgrowth, possibly related to gastric hypoacidity,184 may play a role. In addition, decreased local (mucosal) immune function185,186 and altered enteric nerve and enteroendocrine cell function187 may be important.
The relationships between the abnormalities of mucosal structure and function, symptoms (diarrhea, weight loss), and CD4+ cell count remain unclear. Further research in this area is clearly needed to clarify the nomenclature, to define the pathogenesis of the alterations of mucosal structure and function, and to identify their relationships to other clinical features.
The main focus of diagnostic workup of patients with HIV-related chronic diarrhea is detecting enteric pathogens. However, the intensity of diagnostic workup remains controversial.111 On the one hand, some advise an intensive workup including upper endoscopy and colonoscopy with mucosal biopsies to identify all detectable enteric pathogens.105,115 Alternatively, others favor a minimal workup, consisting of stool cultures and microscopy followed by nonspecific antidiarrheal therapy if no treatable pathogen is identified.188 What evidence exists to support each approach?
The strongest evidence for minimal workup was obtained from a decision analysis model that compared three management strategies: full evaluation (stool culture, ova and parasite examination, blood cultures, upper endoscopy, and colonoscopy with biopsies), limited evaluation (stool culture, ova and parasite examination, and blood cultures), and minimal evaluation (stool cultures).188 For those patients in whom an enteric pathogen was detected, specific therapy was given; the remainder received a nonspecific antidiarrheal agent (diphenoxylate). The same diarrhea remission rate (75%) was obtained for each strategy; after taking cost into account, minimal evaluation was recommended.
Evidence cited to support an intensive workup was obtained in several studies in which enteric pathogens were detected in a large proportion of patients. For example, in one study, 85% (17 of 20) of homosexual men with AIDS and diarrhea (defined as 7 or more stools/day for at least 7 days) had detectable enteric pathogens. Diagnostic workup included microbiological evaluation of stool specimens, light-microscopic examination, and viral culture of duodenal and colonic mucosal biopsies. Based on these findings, the investigators advised intensive workup.115
However, an intensive, costly diagnostic workup of all HIV-infected patients with diarrhea cannot be advocated solely on the basis of such findings for several reasons.111 First, these results may not be generalizable to other clinical practice settings. One should not expect to identify enteric pathogens in 85% of individuals with HIV-related diarrhea unless similar patients with advanced disease are seen and investigated with similar intensity. Second, even if an enteric organism is detected, its role as a cause of chronic diarrhea may be uncertain. Third, no specific effective treatment is available for many enteric pathogens, and nonspecific antidiarrheal agents may be as effective as specific treatment. Fourth, the presence of enteric pathogens is not the only outcome to consider. Other outcomes such as quality of life must be addressed.189 Fifth, the presence of coexisting disease must be taken into account. In a patient with HIV-related chronic diarrhea, a severe coexisting illness may be the dominant clinical problem and the major determinant of the patient's quality of life. In such patients, the inconvenience, discomfort, and cost associated with intensive diagnostic workup may not be offset by any meaningful improvement in the patient's quality of life, even if a treatable enteric pathogen is detected and eradicated.111
Strategies intended to identify patients in whom further investigations are likely to improve patient outcomes have been devised. For example, a system based on the extent of weight loss and the Schilling test result has been reported,107 although this method has not met with widespread acceptance in clinical practice.
Further research is clearly indicated in this area. A full evaluation of the outcomes associated with alternative diagnostic and therapeutic interventions is needed so that evidence-based clinical guidelines can be developed. Such guidelines would assist clinical decision-making by identifying patients in whom intensive workup is unlikely to improve outcome so they can be spared this intervention.
A stepwise approach is suggested here to identify enteric pathogens using few diagnostic tests, tailoring the workup to the patient. This approach relies on the patient's clinical features and CD4 cell count to guide investigations. The hypothesis underlying this approach is that identifying and treating the cause(s) of HIV-related chronic diarrhea improves patient outcomes, such as quality of life. This approach is based on the following principles: (1) the CD4 cell count is helpful because the presence of severe immunodeficiency (defined as a CD4 cell count of <100/mm3) denotes a subgroup of patients at significant risk for CMV disease and MAC infection110,150; (2) compared with colonoscopy, the yield of flexible sigmoidoscopy is adequate for detecting enteric pathogens108,190; and (3) endoscopic mucosal biopsies are preferable to barium radiographs, because the goal is to detect enteric pathogens. As Table 3 indicates, the first step begins with stool tests to identify common enteric bacteria (e.g., Salmonella species) and parasites (e.g., G. lamblia) as well as specific opportunistic agents (e.g., C. parvum species). A stool sample should be sent for Clostridium difficile toxin assay if the patient is at risk for antibiotic-associated diarrhea.191 The number of stool samples that should be submitted is unknown, although many would advise at least three specimens. In febrile patients, blood should be drawn for bacterial culture, and in those with severe immunodeficiency (a CD4 cell count of <100/mm3), blood should also be drawn for mycobacterial culture.
In patients with rectal symptoms (small volume stools, tenesmus, passage of blood per rectum) or those whose symptoms suggest a colonic process (bloody diarrhea, lower abdominal cramps or pain), flexible sigmoidoscopy is the next step. In patients with a CD4 cell count of <100 cells/mm3, CMV disease should be kept in mind; a careful search for submucosal hemorrhage, mucosal erosions, or ulcerations should be undertaken. Rectal and colonic mucosal biopsy specimens should be obtained for light-microscopic examination. Mycobacterial culture of the biopsy specimens may be considered, depending on the clinical presentation. An additional mucosal biopsy specimen may be placed in glutaraldehyde and kept aside for subsequent electron-microscopic examination (if available) if other studies are unrevealing and microsporidiosis is suspected.
The third step is to perform esophagogastroduodenoscopy. Duodenal mucosal biopsy specimens should be obtained for light-microscopic examination. In patients with a CD4 cell count of <100/mm3 and clinical features suggestive of MAC infection, biopsy specimens may also be obtained for mycobacterial culture. An additional mucosal biopsy specimen may be put aside for subsequent electron-microscopic examination, as noted above.
This stepwise approach is intended as a guideline. Clearly, clinical judgment plays an important role in determining the sequence and content of the investigations. In selected patients whose symptoms suggest a small bowel process (large volume diarrhea, periumbilical cramps), one may omit flexible sigmoidoscopy and proceed directly from stool tests (if negative) to esophagogastroduodenoscopy with biopsies. In selected patients, colonoscopy may be performed rather than flexible sigmoidoscopy.
As noted above, opinion varies regarding the role of empiric antidiarrheal therapy in patient management. A case for such therapy can be made in patients with nonbloody diarrhea and negative stool test results before deciding whether to proceed with endoscopic evaluation. A variety of agents may be used, including loperamide and diphenoxylate and atropine (Lomotil). Tincture of opium may be useful in some patients. Although octreotide, a synthetic analogue of somatostatin, initially showed promise, recent results indicate that this agent is not more effective than placebo in patients with HIV-related chronic diarrhea.192
An important problem in HIV-infected individuals is chronic diarrhea, which occurs with greater frequency in male homosexuals and those with severe immunodeficiency. The major cause of HIV-related chronic diarrhea is infection with enteric pathogens. In published reports, the prevalence and spectrum of enteric pathogens varies depending on the characteristics of the study population and the intensity of the diagnostic workup. In patients with advanced immunodeficiency who undergo intensive workup, the opportunistic agents most commonly identified are MAC, C. parvum, CMV, and E. bieneusi.
In clinical practice, the main goal of diagnostic workup in patients with HIV-related chronic diarrhea is to detect enteric pathogens. However, the intensity of diagnostic evaluation remains controversial. On the one hand, some advise an intensive evaluation that includes esophagogastroduodenoscopy and colonoscopy with mucosal biopsies to identify all detectable enteric pathogens. Alternatively, others favor a minimal workup, consisting of stool cultures followed by nonspecific antidiarrheal therapy if no treatable pathogen is identified. No prospective studies have compared the full effects of alternative management strategies on broad patient outcomes, including quality of life. In the absence of such evidence, a stepwise workup is reasonable based on the assumption that identifying and treating the cause of HIV-related chronic diarrhea improves patient outcomes. According to this approach, the first step is to obtain stool tests to identify common enteric bacteria and parasites as well as specific opportunistic agents. If stool tests fail to detect an enteric pathogen, the second step is to perform flexible sigmoidoscopy with mucosal biopsies. If these studies are unrevealing, the third step is to perform esophagogastroduodenoscopy with duodenal mucosal biopsies. Appropriate antimicrobial therapy is prescribed if specific enteric pathogens are identified. In patients in whom either enteric pathogens are not identified or antimicrobial therapy is ineffective, nonspecific antidiarrheal agents are prescribed.
The causes of hepatobiliary disease differ depending on the extent of immunocompromise. In earlier-stage HIV infection (a CD4 cell count of >500/mm3), hepatic complications usually represent liver-specific processes, such as drug-related hepatotoxicity, primary neoplasms, or infection with hepatotrophic viruses. With progression of immunodeficiency to AIDS (a CD4 cell count of <200/mm3), the liver is generally involved as part of a systemic opportunistic infection due to MAC, fungi, or CMV.
Because the liver is often one of many sites involved with disseminated opportunistic infections in patients with AIDS, liver disease per se is rarely the primary cause of death.193-195 Nonetheless, the progressive jaundice, fever, and/or abdominal pain that may accompany hepatobiliary disease can significantly reduce quality of life and therefore merits evaluation in most patients.
Most published studies of the manifestations of hepatobiliary disease in HIV infection have been descriptive. Initial observations were made in autopsy studies, but more recently case series have clarified the prevalence, clinical features, and clinical course of several HIV-related hepatobiliary complications. For some diseases, such as bacillary peliosis hepatis, findings previously considered to be nonspecific are now recognized as manifestations of a specific infection. Continued clinical experience with HIV disease will likely lead to further refinements in our understanding of how the liver responds to opportunistic infection. Studies examining the optimal mode of diagnosis and therapy have only begun to appear, and greater progress in these areas is anticipated in the next 5-10 years.
Clinical manifestations of hepatobiliary disease in HIV infection range from asymptomatic liver biochemical abnormalities to hepatic failure. Few studies have attempted to correlate clinical and biochemical abnormalities with specific diagnoses, although patterns have been recognized that suggest the most likely possibilities (see below). Specific causes of hepatobiliary disease are listed in Table 4 and are reviewed individually below.
| Hepatic disease | |
|---|---|
| MAC | |
| Drug induced, especially sulfa, zidovudine | |
| M. tuberculosis | |
| Bacillary peliosis hepatis | |
| Lymphoma | |
| CMV | |
| Cryptococcus, histoplasmosis | |
| Kaposi's sarcoma | |
| Hepatitis C | |
| Hepatitis B, D | |
| P. carinii | |
| Microsporidia | |
| Biliary disease | |
| AIDS cholangiopathy | |
| Lymphoma | |
| Kaposi's sarcoma | |
| Acalculous cholecystitis | |
| Non-AIDS disorders, including gallstone disease |
NOTE. In decreasing order of frequency.
MAC is the most common pathogen identified in the liver in autopsy series.193,196-198 It is typically a late-stage complication, associated with disseminated infection and characterized by fevers, night sweats, weight loss, and diarrhea. Although liver involvement can be documented in up to 46% of patients,199 the infection is usually clinically silent. Liver biopsy appears to be more sensitive than bone marrow biopsy in diagnosing infection (75% positive vs. 25% in patients with documented MAC)200 and usually shows large numbers of acid-fast organisms; granulomas, if present, are usually poorly formed.197,201,202
Mycobacterium tuberculosis, unlike MAC, can occur in earlier stages of HIV infection and is more prevalent in injection drug users than other transmission categories.196 Tuberculosis is usually a pulmonary infection in patients with a CD4 count of >200/mm3, whereas atypical presentations are more likely, including hepatic involvement, in patients with more severe immunodeficiency.203 Clinical disease in most cases arises from reactivation rather than primary infection.204 As with hepatic MAC infection, clinical expression of liver disease is unusual.
CMV, like MAC, involves the liver late in the course of HIV disease as part of a systemic infection. Although a frequent finding in autopsy series,198 for unclear reasons, hepatic CMV is rarely diagnosed antemortem. The infection is associated with fever and weight loss, giant cell formation, and mononuclear cell infiltration in infected tissue.
Bacillary peliosis hepatis is a lesion characterized by dilated vascular lakes in the hepatic parenchyma and is associated with one of two bacillary organisms, either Bartonella quintana or Bartonella henselae. Autopsy series initially described peliosis hepatis in patients with AIDS as a nonspecific finding205,206 until the reports by Perkocha et al.207 and Kemper et al.208 in 1990 identified its cause. The infections are systemic and associated with fever, lymphadenopathy, hepatosplenomegaly, and cutaneous and bony lesions.209,210 The syndrome is virtually identical to cat-scratch fever.190 The bacteria can be identified in hepatic sinusoidal endothelial cells211 and are most easily visualized using silver-based stains such as Warthin-Starry.209 Peliotic spaces, which may be macroscopic or microscopic, resolve with appropriate antibiotic therapy.212
Fungal infections of the liver are almost exclusively seen in late-stage HIV disease as part of a systemic infection. Histoplasmosis occurs predominantly, but not exclusively, in endemic areas, with the infected liver showing fungal organisms within granulomas. In a series of 48 patients, 5 had documented hepatic involvement.213 Clinically, disseminated histoplasmosis presents as a severe systemic illness; hepatomegaly is a common finding.213 Typically the organism is first identified in other sites such as bone marrow, lung, or on blood smear, so liver biopsy is rarely required for diagnosis. Indeed, pulmonary disease is observed in one half of patients with disseminated histoplasmosis,214,215 often with cutaneous involvement. Relapse may be detectable by capsular antigen assay.216 Cryptococcus neoformans, like Histoplasma capsulatum, may be detected in the liver during systemic involvement and typically incites only a weak inflammatory response.217 Similarly, hepatic involvement from coccidiomycosis218 or blastomycosis219 has been reported.
Disseminated P. carinii has been reported to involve the liver.220,221 This appears to occur primarily, but not exclusively, in patients receiving inhaled pentamidine therapy for prophylaxis. Microsporidia have also been reported to cause hepatitis.222
Evidence of infection with hepatitis A virus (HAV), hepatitis B virus (HBV), hepatitis C virus (HCV), and/or hepatitis delta virus (HDV) is frequently recognized in patients with HIV disease, in part because risk factors for acquiring these infections are similar to those for HIV (e.g., unprotected sex with multiple partners, needle use, or transfusions received before the introduction of screening methods for HCV and HIV). Behavior of hepatotrophic viruses is altered in the setting of HIV infection, especially as immunocompromise becomes severe. Only the more recent studies have stratified patients based on the extent of immunodeficiency (i.e., CD4 cell count).
For HBV, the prevalence of seropositivity indicating past or present infection is unchanged in HIV-infected individuals; however, there can be marked alterations of antibody and/or antigen levels with late-stage infection and AIDS.223 Reactivation of HBV is not unusual with the onset or progression of HIV infection. For example, several reports have described the reappearance of hepatitis B surface antigen in patients previously believed to be immune to HBV based on the presence of circulating hepatitis B surface antibody.223,224 Additionally, increased expression of e and core antigens and increased DNA polymerase levels may result from coinfection with HIV.225-229 These serological findings have not been confirmed in all studies, even when stratifying based on HIV stage.230,231 Most but not all studies have noted that while an increased risk of reactivation or chronic carriage of HBV exists in homosexual men with HIV infection, the liver disease is not histologically more severe226,231,232 and survival is not affected by HBV.205 In contrast, histological progression and worsened survival have been reported by Housset et al.232 in those chronic HBV carriers who were injection drug users. The difference between these two transmission categories may be the higher prevalence of HCV among drug users, which was not assessed in the study by Housset et al.232 Indeed, in a study of 122 HIV-infected injection drug users, severity of liver disease was significantly greater in patients with both HBV and HCV infections compared with those with either HBV or HCV alone.233 The presence of HIV infection does not increase the prevalence of HDV within specific transmission categories. However, as in HIV-negative patients, HDV in those with HIV infection is associated with much more severe liver disease than in those with HBV alone.234 Unfortunately, in patients with HIV who have no HBV markers, for unclear reasons, vaccination is not effective in generating immunity.235,236 Lack of vaccine efficacy seems to be unrelated to the stage of HIV disease.
The reported prevalence of HCV in patients with HIV infection primarily depends on the transmission category of the patients studied and the specificity of the assay used. Studies using first-generation anti-HCV assays may have overestimated HCV prevalence.237 However, antibody assays alone may underestimate the true HCV prevalence, because loss of antibody has been associated with progressive HIV infection.238,239 HCV RNA does not correlate closely with positive HCV immunoassay in HIV-infected patients, because it is only detectable in 53% of patients with a positive second-generation immunoassay.240 Prevalence is vastly higher in injection drug users than all other transmission categories, ranging from 52% to 89%.238,241 The prevalence is 1%-3.5% in HIV-infected military personnel,242,243 7% in a university clinic setting,237 and 4%-11% in non-drug users (including homosexual men).240,244 Many HIV-infected hemophiliacs are coinfected with HCV.245 These prevalence data are consistent with the notion that HCV is almost always transmitted parenterally, even in patients with HIV infection. However, HIV infection seems to increase the risk of sexual or perinatal transmission of HCV to HIV-seronegative female sex partners246,247 and possibly to newborns of HIV-infected mothers, respectively.248
Clear evidence for acceleration of HCV replication and related liver injury by HIV has emerged from studies of hemophiliacs. Eyster et al. have shown a 58-fold increase in HCV RNA levels during a several-year period after HIV seroconversion (by branched chain complementary DNA assay or PCR) compared with a 3-fold increase during the same interval in those who were HCV seropositive but HIV seronegative.249 Moreover, increases in HCV RNA accompanied progressive immunodeficiency during a 10-20-year period and were correlated with aspartate aminotransferase levels and evidence of hepatomegaly, liver injury, and failure.248-251 Despite this overall increased incidence of HCV-related liver injury and mortality in HIV-infected hemophiliacs,245 HCV infection is not an independent determinant of mortality in nonhemophiliacs with AIDS,240 probably because these patients have high mortality rates from other AIDS-related opportunistic illnesses.
In summary, there does not seem to be an increase in prevalence or severity of HAV infection associated with HIV infection. HBV infection in HIV disease is associated with increased levels of HBV DNA but no apparent increase in severity of liver disease or independent effect on mortality. HCV infection may be associated with increased viral replication and more severe liver disease, especially in HIV-infected hemophiliacs.
Neoplasms may be first diagnosed in the liver. Non-Hodgkin's lymphoma, also referred to as large cell or Burkitt's lymphoma depending on the histology, presents in extranodal sites in the majority of cases in those with HIV infection. It may appear at any stage of HIV disease, although it usually occurs in the later stages of immunodeficiency. In a large series of patients with HIV-associated non-Hodgkin's lymphoma, 9%-16% had a primary presentation in the liver.252,253 The most common presentation is abnormal liver test results (elevated alkaline phosphatase levels, often with jaundice) and hepatomegaly. Fever is inconsistent. Abdominal computed tomography almost always shows one or multiple mass lesions. Prognosis depends primarily on the stage of immunocompromise rather than the lymphoma itself.252
Kaposi's sarcoma, which occurs most frequently in homosexual men, is a multicentric neoplasm derived from lymphatic endothelial cells.254 Hepatic Kaposi's sarcoma has been detected on liver biopsy or autopsy in 14%-49% of patients,194,201,255 although it is rarely evident clinically. Interestingly, the prevalence of Kaposi's sarcoma seems to be decreasing.256 Its occasional presence in HIV-negative homosexual men has suggested that it may arise from a transmissible agent, and recent studies identifying herpes virus-like DNA in Kaposi's sarcoma tissue supports this possibility.257,258
In any patient with signs or symptoms of hepatobiliary disease, drug-induced liver abnormalities are important to consider because they are so readily treatable by drug withdrawal in most cases. Sulfonamides in particular have been associated with adverse reactions requiring alternative agents in >50% of patients in one series.259 Typical manifestations include fever, rash, leukocytosis, eosinophilia (occasional), and transferase level elevations.259 Antimycobacterial drugs have also been linked to severe liver disease.260 In some patients, antiretrovirals, particularly zidovudine and dideoxyinosine, have been associated with severe, sometimes fatal hepatic failure261-264 in conjunction with marked steatosis. The basis for this lesion seems to be mitochondrial toxicity with impairment of mitochondrial DNA synthesis.265 Survival is possible if the syndrome is recognized early and the offending agent promptly withdrawn.260 Other agents associated with hepatotoxicity include pentamidine258 and acetaminophen.262 An additional form of iatrogenic liver injury attributable to iron overload has been described in patients who have received multiple transfusions.198,266 In patients with HIV infection, the heightened sensitivity to drugs combined with the frequent use of concomitant medications suggest that any agent should be considered hepatotoxic, including alcohol, nonprescription medications, and herbal remedies.
In a small study, amyloidosis was observed in 5 of 12 patients with AIDS and a history of injection drug use.267 All had hepatomegaly, and none had underlying myeloma or plasma cell dyscrasia. In these patients, it is unclear whether the lesion represents a complication of HIV disease or of injection drug use alone.
HIV itself may infect hepatocytes and Kupffer cells. Several studies have identified the virus within Kupffer cells (hepatic macrophages) and occasionally within hepatocytes using immunocytochemistry or in situ hybridization.268-271 The findings have been complemented by evidence of infection of human Kupffer cells and sinusoidal endothelial cells in culture,272,273 probably via CD4 cell receptors.274 A study quantitating HIV-1 DNA in human liver showed a broad range of DNA concentrations in different patients, with some samples from patients with AIDS lacking HIV-1 DNA by PCR.271 No correlation was evident between HIV-1 RNA levels and histological liver injury.271 The demonstration of HIV in liver has two major implications. First, as the largest macrophage population in the body, Kupffer cells represent a major potential reservoir of virus; this must be considered in designing antiretroviral strategies. Second, active HIV infection of hepatic macrophages might lead to impaired cellular function, which could explain the increased rate of systemic enteric bacteremia in some patients with HIV infection.275
Nonspecific histopathologic findings have been frequently described in autopsy and liver biopsy series. These include granulomas (without organisms), Kupffer cell hyperplasia (which may reflect direct HIV infection), steatosis, portal inflammation, focal necrosis, bile stasis, or congestion.194,276 Such findings rarely have clinical significance.
Disorders of the biliary tract in patients with HIV infection consist predominantly of the syndrome known as AIDS cholangiopathy.277-280 This abnormality resembles ductular changes of sclerosing cholangitis and is characterized by intrahepatic and/or extrahepatic duct strictures. The most common cholangiographic pattern is diffuse (intrahepatic and extrahepatic) sclerosing cholangitis in combination with papillary stenosis.278,281-283 Less commonly, there may be isolated intrahepatic sclerosing cholangitis, papillary stenosis, or long extrahepatic strictures.281 Clinically, the presentation is variable, although right upper quadrant pain accompanied by an elevated alkaline phosphatase level is the most common manifestation. Liver biochemical test results may be normal.278,282 Interestingly, jaundice is very unusual,278,280,282,283 suggesting that complete ductal obstruction is rare. AIDS cholangiopathy has been more commonly reported in homosexual men than in other transmission categories. A significant proportion of patients may harbor ductal changes that are clinically silent, as shown in a study of patients who had CMV viremia but no abdominal symptoms.284 When using endoscopic retrograde cholangiopancreatography (ERCP) as the gold standard, ultrasonography or computed tomography have an approximately 75%-87% sensitivity in detecting ductal disease,278,280 implying that the syndrome cannot be excluded definitely using these latter two imaging studies. The etiology is usually related to a specific infection (cryptosporidiosis, microsporidiosis, CMV), neoplasm, pancreatic disease, or possibly some other process or immune mechanism. One study from France has reported an extremely high frequency of biliary microsporidiosis in affected patients,285 whereas other studies from the United States and France have more commonly identified cryptosporidium or CMV,279,281,286 particularly in patients with involvement of large intrahepatic ducts.283 The presence of AIDS cholangiopathy does not seem to affect survival.287
Acalculous cholecystitis occurs as a late-stage complication when opportunistic infections are more likely.288 Common findings include thickening of the gallbladder wall; radioisotope imaging with 2,-6 dimethylphenylcarbamoylmethyliminodiacetic acid usually shows cystic duct obstruction.288 Cholecystectomy is curative, although associated sclerosing cholangitis of the bile ducts may subsequently become clinically evident. Histopathologic examination of the gallbladder will frequently identify an opportunistic pathogen such as CMV, cryptosporidia, microsporidia, or Isospora belli. Gallstones may also be detected.288-290 Rare cases of lymphomatous or nodal biliary obstruction have been reported.
As in all patients, the magnitude and pattern of liver test abnormalities in those with HIV infection can help distinguish cholestatic from hepatocellular lesions but taken alone do not indicate a specific diagnosis. Marked elevations of serum alkaline phosphatase levels are particularly common, having been reported in 17% of 90 consecutive patients with AIDS in one series.291 The majority of these patients had no identifiable biliary or hepatocellular disease, although liver biopsy and ERCP were not performed in all patients. In one large series,194 the mean serum alkaline phosphatase level among patients with AIDS cholangiopathy was 600 (normal value < 115). As noted above, the serum bilirubin levels are usually normal so that jaundice from cholangiopathy alone is extremely unusual.280,282,283,285 Alternatively, marked elevation of alkaline phosphatase levels in the absence of ductal pathology may indicate the presence of MAC194 or some other disseminated infection or malignancy.
Noninvasive imaging is an important tool in evaluating hepatobiliary pathology. In one large series of HIV-infected patients undergoing abdominal ultrasonography, abnormalities were detected in 264 of 399 ultrasonography studies.292 Typical findings included gallbladder and biliary duct abnormalities (n = 80), hepatosplenomegaly (n = 337), and hepatomegaly (n = 77). Associated abnormal liver biochemistries were present in 270 patients. The frequency of abnormalities detected in this series increased substantially during a 10-year period (from 1984 to 1994); it is uncertain whether this represents increased prevalence of abnormalities, increased experience of the ultrasonographer, or technological improvements in imaging. Computed tomographic scanning is also useful in evaluating hepatobiliary abnormalities.293 In particular, lesions attributable to lymphoma294 and large peliotic spaces212 are usually well visualized. Focal lesions attributable to MAC may occasionally be detected by either computed tomography or ultrasonography,295 although this infection is usually diffuse. No studies have directly compared the sensitivity and specificity of noninvasive techniques in detecting hepatobiliary disease. Liver-spleen scintigraphy rarely adds additional information.296
In patients with AIDS cholangiopathy, noninvasive imaging (ultrasonography, computed tomography) plays an important role in diagnosis, although this approach is less sensitive than ERCP (see above). In the patient with biliary dilation shown by noninvasive imaging studies, ERCP may be appropriate to diagnose AIDS cholangiopathy, exclude other disorders, and provide the opportunity for therapeutic intervention such as sphincterotomy, stenting, or biopsy, depending on the lesion.
In assessing the use of liver biopsy, at least three issues must be considered: the probability of making a specific diagnosis, the probability of identifying a lesion not identified by other diagnostic methods, and the probability of making a diagnosis for which therapy will improve the patient's quality of life or survival. Retrospective data are emerging that address these issues, but no study has prospectively assessed the impact of liver biopsy on patient outcomes.
According to most reports, percutaneous liver biopsy will uncover one or more specific abnormalities in the majority of patients when undertaken to evaluate fever, hepatomegaly, and/or abnormal liver biochemistries.194,196,298 In a study of HIV-infected patients who were injection drug users,196 liver biopsy was helpful in the evaluation of fever of unknown origin. Several difficulties arise, however, in evaluating liver biopsy data from retrospective studies. First, the indications and frequency of performing liver biopsy have evolved considerably during the 15 years since AIDS was first described. Whereas biopsy was commonly performed when the AIDS epidemic began, it is now performed less frequently. Second, other diagnostic methods have been refined, and we have a greater appreciation of the spectrum of hepatobiliary disease. A clear example is the recognition of bacillary peliosis hepatis as a specific infection, which might obviate the need for biopsy if diagnosed by skin or blood culture. Third, it is possible that liver biopsies are now performed later in the course of HIV infection, when it is more likely that a specific diagnosis will be uncovered.297,298 Fourth, the spectrum of abnormalities will differ among different transmission categories. For example, the prevalence of HCV in injection drug users is much greater than in homosexual men.
Liver biopsy rarely uncovers a lesion not previously identified in other tissues. In one study,194 only 2 of 26 biopsy specimens revealed a new diagnosis (1 of CMV and 1 of lymphoma). For diagnosing fever of unknown origin, liver biopsy may have special value if other methods such as bone marrow biopsy and blood cultures are nondiagnostic. In a prospective study of 24 patients undergoing liver biopsy that had unexplained fever and increased alkaline phosphatase or 
-glutamyl transferase levels, a microbiological diagnosis was made in 13 (54%), in 9 of whom MAC was found.299 Liver biopsy seems to be more sensitive than bone marrow biopsy in diagnosing MAC infection.200 Thus, when MAC is suspected and blood and bone marrow cultures are negative, liver biopsy may enable a prompt diagnosis.
No reliable data exist that specifically address how often findings from liver biopsy lead to improved quality of life or survival. However, this is not likely to be a frequent occurrence given the low probability of uncovering new diagnoses. Moreover, enthusiasm for uncovering unrecognized pathogens or neoplasms is tempered by the realization that many are untreatable, especially in late-stage HIV infection. Similarly, biopsy is not likely to be useful in HCV infection if HIV disease is advanced, because patients with low CD4 counts are unlikely to respond to interferon alfa (see below).
Despite reports to the contrary,300,301 no comprehensive evidence suggests that complications from liver biopsy are more likely in patients with HIV infection, provided that coagulation parameters are normal. Concern persists, however, about the possibility of hemorrhage from vascular lesions such as Kaposi's sarcoma302 or peliosis hepatis.
A valuable alternative to Klatskin or Menghini needle biopsy for focal lesions is directed fine-needle aspiration. This technique greatly reduces the risk of hemorrhage and can be directed to specific abnormalities using computed tomography or ultrasonography guidance. It is particularly useful to diagnose primary hepatic lymphoma, where it was successful in 22 of 24 patients in one series.303 Laparoscopy may also be helpful to perform a biopsy on focal hepatic lesions as well as to identify associated peritoneal disease.304
The therapy of hepatobiliary disease in HIV-infected patients depends on at least three factors: (1) specific findings of culture, biopsy, or imaging studies; (2) extent of the patient's immunocompromise; and (3) presence and severity of other infections or neoplasms. The treatment of opportunistic infections involving the liver is similar to that for other sites. The focus here is on the use of interferon in viral hepatitis.
The value of interferon alfa for treating chronic HBV or HCV infections in patients with HIV is debatable, primarily because these infections are rarely a cause of morbidity or mortality in this setting. Before considering therapy, the goal must be defined. For patients with HBV who are e antigen positive, a reasonable goal might be seroconversion to e antigen negative status, but only if this clearly translates into prolonged survival. However, the data needed to address this issue are lacking. Moreover, responsiveness to interferon in patients with chronic HBV carriage is lower in HIV infection. Wong et al.305 reported that a response (defined as loss of HBV DNA and e antigen and appearance of anti-e antibody) to interferon in patients with HBV infection was one fifth as likely in those coinfected with HIV compared with those who were HIV seronegative. These results are similar to those reported by Marcellin et al.306 For chronic HCV infection, the likelihood of response seems to be greatest in those who are not severely immunocompromised.307,308 However, treatment is most clearly indicated when HCV replication increases, which occurs in the late stages of HIV infection when other opportunistic life-threatening infections develop. Thus, treatment with interferon alfa is least likely to be helpful when it is needed most. Optimal dosing and duration of therapy in this setting are not established. Use of interferon alfa in patients coinfected with HBV and HIV is not indicated, except within the context of clinical trials. As with HBV, no study has shown that treating HCV with interferon improves the patient's quality of life or prolongs survival. Given that HCV does not seem to adversely influence survival in HIV-infected patients without hemophilia,240 the need for treatment is less clear.
Papillary stenosis associated with AIDS cholangiopathy can be effectively treated by endoscopic sphincterotomy, and this approach should be considered in the symptomatic patient because pain is improved in most patients.278,283 Nevertheless, controlled trials of sphincterotomy or stenting for papillary stenosis are lacking. In two series totaling 22 patients with abdominal pain and papillary stenosis,278,283 sphincterotomy resulted in complete relief of pain. However, in another series of 21 patients, pain relief occurred in only 9.282 After sphincterotomy, progressive elevation of alkaline phosphatase levels continues in most of these patients, reflecting ongoing intrahepatic biliary disease.309 In some patients, stenosis may reoccur, necessitating repeat ERCP. There is no evidence to support the use of sphincterotomy in the patient with intrahepatic sclerosing cholangitis who has a normal common bile duct.
Acalculous cholecystitis can be a surgical emergency, or it may result in chronic progressive symptoms. In otherwise stable patients, cholecystectomy is indicated and is remarkably well tolerated.288 In patients too ill to tolerate open surgery, laparoscopic cholecystotomy, cholecystostomy, or minilaparotomy may be feasible.310 If the clinical evidence supports CMV infection as an underlying etiology, empiric administration of ganciclovir or foscarnet may be considered before definitve therapy.
In general, the diagnostic workup should be similar to the immunocompetent host. A suggested approach is summarized in Table 5. Mild elevations in liver chemistry test results (<2 times the upper limit of normal) in an asymptomatic patient may warrant no specific evaluation. Elevation of the transferase levels may indicate hepatitis and should be evaluated with hepatitis serologies and consideration of drug-induced disease. All prescription and nonprescription drugs should be considered potentially hepatotoxic and stopped, if possible, before or concurrent with additional evaluation. Marked elevation of serum alkaline phosphatase levels usually indicates an infiltrative disorder (neoplasm, infections) or cholangiopathy. In this setting, evaluation should depend on the clinical features. In a patient with systemic symptoms and hepatomegaly, computed tomography is probably the preferred test because adenopathy can be appreciated and mass lesions are better delineated. The presence of mild hepatomegaly in the absence of significant liver test abnormalities, right upper quadrant pain, or systemic symptoms demands no specific evaluation. Computed tomography and ultrasonography are especially useful in guiding fine-needle aspiration or biopsy of focal lesions. Abdominal ultrasonography should be the test of choice when cholangiopathy is suspected, such as in a patient with right upper quadrant pain and elevation of the serum alkaline phosphatase level. If biliary duct dilation is detected, particularly of the common bile duct, ERCP is appropriate because sphincterotomy or stenting may be helpful. In the jaundiced patient without clinical evidence of hepatitis, CT may be the preferred test to evaluate for mass lesions or adenopathy.311 Liver biopsy rarely uncovers new unrecognized disease. Biopsy may be most useful in those patients with persistent fever in whom other less invasive methods have not uncovered a treatable lesion.
CT, computed tomography; US, ultrasonography.
In HIV-infected individuals, abnormal liver test results, jaundice, or hepatomegaly can develop as a result of hepatic parenchymal disease, biliary lesions, or both. More than one abnormality is not unusual, particularly in late-stage HIV infection. The spectrum of potential causes for liver or biliary disease are dependent on the stage of immunocompromise. However, the liver is rarely the only site of involvement due to infection or neoplasm, so that evidence of infection in other sites may obviate the need for invasive hepatic evaluation. Liver or biliary disease is rarely the cause of death in patients with HIV infection. As such, reasonable end points apart from survival must be defined before considering treatment. These may include relief of symptoms (pain, fever, jaundice) or improved quality of life.
C. MEL WILCOX, M.D.
Department of Medicine
University of Alabama at Birmingham
Birmingham, Alabama
LINDA RABENECK, M.D., M.P.H.
Department of Medicine
Baylor College of Medicine
VA Medical Center
Houston, Texas
SCOTT FRIEDMAN, M.D.
Department of Medicine
University of California San Francisco
San Francisco, California
in advanced HIV infection in the absence of AIDS-related secondary infections. J Acquir Immune Defic Syndr 1992;5:1266-1271.
(TNF-) and soluble TNF receptors in human immunodeficiency virus type 1 infection-correlations to clinical, immunologic, and virologic parameters. J Infect Dis 1994;169:420-424.
Abbreviations used in this paper: AIDS, acquired immunodeficiency syndrome; CDC, Centers for Disease Control and Prevention; CMV, cytomegalovirus; ERCP, endoscopic retrograde cholangiopancreatography; HIV, human immunodeficiency virus; MAC, Mycobacterium avium complex; PCR, polymerase chain reaction; TPN, total parenteral nutrition.
© 1996 by the American Gastroenterological Association 0016-5085/96/$3.00
Address requests for reprints to: Chair, Patient Care Committee, AGA National Office, 7910 Woodmont Avenue, 7th Floor, Bethesda, Maryland 20814. Fax: (301) 654-5920.
The Patient Care Committee acknowledges the following individuals whose critiques of this review paper provided valuable guidance to the authors: Robert M. Craig, Richard W. Goodgame, Donald P. Kotler, and Christina M. Surawicz.