The primary consideration for prophylaxis is efficacy. Because prophylaxis is usually prolonged, long-term safety and tolerability are essential. Ease and frequency of administration are also important considerations. Oral agents are preferable, and complex regimens should be avoided because of diminished patient compliance. The potential for drug interactions resulting in altered metabolism and synergistic or additive toxicity is also of concern. Finally, the cost of preventive therapies is a critical factor in determining how widely used these treatments will be.

Although many pathogens are amenable to preventive therapy, not all infections are appropriate targets for prophylaxis. Diseases being considered for prophylaxis should be associated with substantial morbidity or mortality, and the consequences of the disease should outweigh those of the prophylactic regimen. For example, P. carinii pneumonia is associated with high morbidity, mortality, and cost, but prophylaxis is inexpensive and relatively nontoxic. Conversely, although mucosal candidiasis is common, it is generally mild at the time of presentation, easily diagnosed, and readily treated. Further, the benefits of prophylaxis must also be weighed against the risk for emergence of resistant pathogens.

The incidence of opportunistic infections and their occurrence in the course of HIV disease are important considerations in developing prevention strategies. The probability of infection depends on exposure to a pathogen and relative host susceptibility. Geographic variation in the prevalence of exposure, host cellular and humoral immunocompetence, the virulence of the pathogen, and the likelihood of latent infection are all components in this equation. More virulent pathogens, such as Streptococcus pneumoniae or Mycobacterium tuberculosis, may cause disease when host defenses are only mildly or moderately impaired. Truly opportunistic pathogens, such as M. avium complex, rarely cause disease until host cellular immune defenses are nearly ablated. Prophylaxis for these infections becomes important as HIV disease becomes more advanced.

A prophylaxis strategy must take into account the incidence of infections. Figure 1 shows the incidence of infections in a cohort of 1050 patients with advanced HIV disease and CD4 counts less than 250/mm³ who were starting zidovudine therapy [4, 10-13]. Disease caused by three pathogens (P. carinii, cytomegalovirus, and M. avium complex) was common, with a 2-year incidence of 20% to 40% (Figure 1). Because of the severity and relative frequency of disease, prophylaxis against these pathogens is desirable. Three other pathogens (Toxoplasma gondii, Cryptococcus neoformans, and M. tuberculosis) caused disease in only 5% to 8% of patients at 2 years; thus, prophylaxis is reserved for persons at highest risk. In this review, we analyzed the efficacy of chemoprophylaxis for opportunistic infections in persons infected with HIV.

View larger version (11K): [in this window] [in a new window]   Figure 1. Probability of patients developing specific opportunistic infections from time of enrollment. CMV = cytomegalovirus; CRY = cryptococcosis; MAI = Mycobacterium avium intracellulare; MTB = Mycobacterium tuberculosis; PCP = Pneumocystis carinii pneumonia; and TOX = toxoplasmosis.

Methods

Top Methods Results Author & Article Info References

Using MEDLINE (1985 to 1993), we searched the English-language medical literature for studies on the prevention of HIV-related opportunistic infections. Relevant abstracts presented at the International Conferences on AIDS, the Interscience Conferences on Antimicrobial Agents and Chemotherapy, and the National Conference on Human Retroviruses and Related Diseases were also reviewed. Importance was assigned in descending order to controlled clinical trials, uncontrolled trials and retrospective studies, and prospective observational studies. In this review, primary prophylaxis is the prevention of the first episode of opportunistic disease, and secondary prophylaxis (sometimes referred to as maintenance therapy) is the prevention of recurrence or relapse of disease.

Results

Top Methods Results Author & Article Info References

Organisms Causing Opportunistic Infections Pneumocystis carinii

Pneumocystis carinii pneumonia is the most common serious HIV-related opportunistic infection in developed countries [14] (Figure 1). Before the widespread use of prophylaxis, patients with an initial episode of P. carinii pneumonia had a median survival of 10 months, and virtually all died within 2 years [1]. Although antiretroviral therapy prolongs survival for patients with an initial diagnosis of P. carinii pneumonia [15], 60% develop a second episode within 1 year of initiation of zidovudine in the absence of secondary prophylaxis [16].

Primary pneumocystis prophylaxis is recommended for HIV-infected adults with a total CD4 count of less than 200/mm³ [17, 18]. In the Multicenter AIDS Cohort Study (MACS), the risk for developing P. carinii pneumonia within 6 months was 8% for persons with CD4 counts 200/mm³ compared with 0.5% for patients with CD4 counts of 201 to 350/mm³ [19]. Prophylaxis should also be given (regardless of CD4 count) to patients with a history of P. carinii pneumonia [15, 18] and to patients with more than 2 weeks of unexplained fevers or other constitutional symptoms, oral candidiasis, or opportunistic infections that typically occur with CD4 counts less than 200/mm³ [18, 19].

In 1988, Fischl and colleagues [3] reported the results of the first primary prophylaxis trial in HIV-infected patients. Sixty patients with Kaposi sarcoma were assigned to receive either trimethoprim-sulfamethoxazole (800/160 mg or 1 double-strength tablet twice daily) with folinic acid, or no prophylaxis. Four (13%) of 30 patients assigned to trimethoprim-sulfamethoxazole developed P. carinii pneumonia compared with 16 (53%) of 30 patients who were not receiving prophylaxis (P < 0.005). The 4 patients who developed P. carinii pneumonia in the trimethoprim-sulfamethoxazole arm discontinued prophylaxis before the diagnosis because of drug toxicity. Trimethoprim-sulfamethoxazole prophylaxis was associated with longer median survival (23 months compared with 13 months, P < 0.002). Adverse reactions occurred in half of the patients taking trimethoprim-sulfamethoxazole and led to discontinuation of therapy in 17%.

The high frequency of adverse reactions because of trimethoprim-sulfamethoxazole prompted a search for alternative agents. Aerosolized pentamidine was developed in an effort to target drug delivery to the alveoli, thereby minimizing systemic toxicity [20-24]. In the San Francisco Community Prophylaxis Trial, 28 (20%) of 139 patients assigned to receive pentamidine (300 mg monthly) developed P. carinii pneumonia compared with 41 (30%) of 135 assigned to receive 30 mg every 2 weeks (P = 0.02) [25]. This study showed the efficacy of aerosolized pentamidine for secondary prophylaxis and was the basis for the recommended dose of 300 mg delivered by Respirgard II nebulizer (Marquest, Englewood, Colorado) once monthly. A subsequent placebo-controlled European trial [26] confirmed its efficacy for primary prophylaxis. The annual incidence of P. carinii pneumonia was 9% in the pentamidine group compared with 27% in the placebo group (P = 0.002).

Trimethoprim-sulfamethoxazole is superior to aerosolized pentamidine, as shown in retrospective studies [27-30] and in two randomized trials. Schneider and colleagues [9] compared three regimens for primary prophylaxis in 215 patients with CD4 counts less than 200/mm³: aerosolized pentamidine (300 mg monthly) or trimethoprim-sulfamethoxazole (1 single-strength tablet or 1 double-strength tablet daily). After a mean follow-up of 264 days, no patient assigned to trimethoprim-sulfamethoxazole developed pneumocystis pneumonia compared with 6 (8.5%) of 71 patients assigned to aerosolized pentamidine. Adverse reactions were more common in the trimethoprim-sulfamethoxazole groups and occurred sooner in patients taking the higher dose. In a randomized, multicenter secondary prophylaxis trial [8], the probability of recurrence of P. carinii pneumonia at 1 year was 18% for patients assigned to receive aerosolized pentamidine compared with 3% for trimethoprim-sulfamethoxazole (1 double-strength tablet daily). Of the 14 recurrences in patients initially assigned to trimethoprim-sulfamethoxazole, 7 were in patients whose therapy was changed to aerosolized pentamidine.

Thus, trimethoprim-sulfamethoxazole is the preferred agent for pneumocystis prophylaxis. The recommended regimen is one double-strength tablet daily, although smaller doses (one double-strength tablet three times weekly or one single-strength tablet daily) may be as effective and better tolerated [9, 28, 31]. Trimethoprim-sulfamethoxazole prophylaxis also appears to protect against toxoplasmic encephalitis [8, 9, 32] and bacterial infections [8]. All studies showed a higher incidence of adverse reactions in patients taking trimethoprim-sulfamethoxazole than in patients taking aerosolized pentamidine. The most common reactions are rash, fever, leukopenia, and hepatitis. Adverse reactions are less frequent with prophylactic doses of trimethoprim-sulfamethoxazole than with the higher doses used for treatment of P. carinii pneumonia [33]. A history of a non-life-threatening reaction to sulfonamides is not an absolute contraindication to prophylaxis with trimethoprim-sulfamethoxazole. The incidence of severe adverse reactions to trimethoprim-sulfamethoxazole in patients with a history of mild-to-moderate adverse reactions is similar to that in patients with no previous reactions [8]. Oral desensitization has been suggested for patients with sulfonamide allergies [34, 35], although no clear rationale exists for this approach because the toxicity is not mediated by IgE. No controlled trials testing this strategy have been conducted.

Aerosolized pentamidine is an alternative for patients who cannot take trimethoprim-sulfamethoxazole. Aerosolized pentamidine is well tolerated, although some patients may have bronchospasm during administration [36]. Bronchospasm can usually be prevented by pretreatment with inhaled ß-adrenergic agonists. The use of aerosolized pentamidine may be associated with a lower diagnostic yield of induced sputum or bronchoalveolar lavage fluid and with unusual radiographic presentations of P. carinii pneumonia, spontaneous pneumothoraces, and a higher incidence of extrapulmonary pneumocystosis [37-41]. The annual cost of aerosolized pentamidine is substantially higher than that of trimethoprim-sulfamethoxazole.

Oral dapsone appears to be an effective and inexpensive alternative to aerosolized pentamidine in patients who are intolerant of trimethoprim-sulfamethoxazole [29, 42-44]. In a randomized trial comparing dapsone (100 mg daily) and trimethoprim-sulfamethoxazole (1 double-strength tablet daily) for primary prophylaxis in 86 patients, only 1 episode of P. carinii pneumonia developed in each group during 1638 patient-months of observation [45]. Dapsone (100 mg twice weekly) and aerosolized pentamidine were equally effective as primary or secondary prophylaxis in a randomized, trial of 96 patients [46]. In a multicenter, randomized trial comparing trimethoprim-sulfamethoxazole, dapsone, and aerosolized pentamidine for primary prophylaxis, dapsone (100 mg daily) was less effective than trimethoprim-sulfamethoxazole but was more effective than aerosolized pentamidine in preventing P. carinii pneumonia (Bozzette SA. Unpublished observations).

The absorption of dapsone may be impaired by achlorhydria or by coadministration of didanosine [47-49]. Toxic reactions to dapsone include rash, nausea, and methemoglobinemia, as well as hemolytic anemia in patients with glucose-6-phosphate dehydrogenase (G6PD) deficiency. The optimal prophylactic dose of dapsone has not been determined. Dapsone has a long plasma half-life (28 hours) [50], and weekly or twice weekly dosing may be justified [42].

Regimens containing dapsone and pyrimethamine have been studied for their efficacy in preventing P. carinii pneumonia and toxoplasmosis [51-55]. In a randomized trial [56] of primary prophylaxis, aerosolized pentamidine, trimethoprim-sulfamethoxazole (1 double-strength tablet three times weekly), and dapsone-pyrimethamine (100 mg and 25 mg weekly) were equally effective. In a randomized trial of 349 patients, dapsone-pyrimethamine (50 mg daily and 50 mg weekly) was as effective as aerosolized pentamidine [57]. Folinic acid (25 mg weekly) is added to pyrimethamine-containing regimens to minimize bone marrow toxicity.

Other regimens that may be effective in preventing P. carinii pneumonia include pyrimethamine-sulfadoxine (Fansidar; Roche Laboratories, Nutley, New Jersey) [55, 58, 59], intravenous or intramuscular pentamidine [60], atovaquone (Mepron; Burroughs Wellcome, Research Triangle Park, North Carolina) [61-64], clindamycin with primaquine [65-67], and trimethoprim-dapsone. However, further studies are needed before these agents can be recommended.

Toxoplasma gondii

It is estimated that 20% to 47% of patients infected with HIV who have latent T. gondii infection will develop cerebral toxoplasmosis during their HIV disease [68-70] and that 10% to 20% of U.S. patients with AIDS will eventually develop toxoplasmic encephalitis [69]. Evidence supports the use of primary prophylaxis against toxoplasmosis in anti-toxoplasma IgG seropositive patients with advanced HIV disease.

Ideally, a regimen effective in preventing toxoplasmosis would also prevent P. carinii infection. Trimethoprim-sulfamethoxazole given to prevent P. carinii pneumonia also appears to be effective in preventing cerebral toxoplasmosis [8, 9, 30, 71-74]. In a retrospective study [32], none of the 60 patients receiving trimethoprim-sulfamethoxazole for Pneumocystis prophylaxis developed toxoplasmosis compared with 12 (13%) of 95 receiving pentamidine (P < 0.05). In two large prospective studies comparing aerosolized pentamidine with trimethoprim-sulfamethoxazole for pneumocystis prophylaxis, 9 of 10 and 3 of 3 patients who developed toxoplasmosis were taking aerosolized pentamidine at the time of diagnosis [8, 9]. In a randomized trial [75] comparing two doses of didanosine, only 1 (1%) of the 80 patients taking trimethoprim-sulfamethoxazole developed toxoplasmosis compared with 30 (13%) of 228 patients taking aerosolized pentamidine (P = 0.002). Toxoplasma seroprevalence was 73% and did not vary between groups.

Dapsone is synergistic with pyrimethamine and has been evaluated in combination with pyrimethamine for toxoplasmosis prophylaxis [76, 77]. In a study [51] comparing trimethoprim-sulfamethoxazole (1 double-strength tablet three times weekly) with dapsone-pyrimethamine (100 mg and 25 mg weekly), 2 (2.4%) of 81 patients in the trimethoprim-sulfamethoxazole group and 3 (3.5%) of 85 patients in the dapsone-pyrimethamine group developed toxoplasmosis. In another study of 109 patients (50% of whom were seropositive for toxoplasma) who were given dapsone-pyrimethamine (100 mg and 25 mg twice weekly), no patient developed toxoplasmosis during a mean follow-up of 15 months [54]. In a trial [57] comparing dapsone-pyrimethamine (50 mg daily and 50 mg weekly) with aerosolized pentamidine, toxoplasmosis developed in 32 (18%) of 176 patients in the pentamidine group and 19 (11%) of 173 patients in the dapsone-pyrimethamine group (P = 0.02). In the subset of patients who were seropositive for T. gondii, the adjusted relative risk for toxoplasmosis was 2.37 in the pentamidine group (P = 0.006).

Pyrimethamine at a dose of 25 mg weekly appears to be ineffective as single-agent prophylaxis [78] and was associated with increased mortality compared with no prophylaxis in one study [79]. In a trial of pyrimethamine (50 mg three times weekly) in 554 patients seropositive for toxoplasma, no difference was noted in the incidence of toxoplasmosis between patients randomly assigned to pyrimethamine compared with those assigned to placebo [80]. Several other agents have activity against T. gondii and merit further investigation as therapeutic and prophylactic agents. These include pyrimethamine with sulfadoxine (Fansidar, Roche Laboratories, Nutley, New Jersey) [55, 59], clindamycin [79, 81], atovaquone [82], azithromycin [83, 84], and clarithromycin [85-87]. Atovaquone and azithromycin are active against the cyst and trophozoite forms of T. gondii in cell culture [88].

On the basis of these studies, we recommend the use of trimethoprim-sulfamethoxazole to prevent toxoplasmosis in persons with CD4 counts less than 100/mm³ who have a positive test result for anti-toxoplasma IgG. Patients who cannot tolerate trimethoprim-sulfamethoxazole should receive dapsone and pyrimethamine with folinic acid. Both regimens are also effective in preventing P. carinii pneumonia. Neither the Centers for Disease Control and Prevention nor the U.S. Public Health Service has recommended routine primary prophylaxis against toxoplasmosis.

Life-long maintenance therapy is required to suppress toxoplasmic encephalitis after initial therapy because drugs currently used in the management of toxoplasmosis are inactive against the cyst form of T. gondii [89]. Suppressive regimens usually consist of lower doses of the same drugs used in acute therapy.

Fungal Infections

Persons infected with HIV often have fungal infections. However, decisions about prophylaxis against specific fungal pathogens should be based not only on efficacy data but also on the prevalence of infection, the risk for disease, the consequences of prophylaxis, and the relative cost and difficulty of the treatment regimen compared with the prophylactic agent [90]. Routine antifungal prophylaxis in adults infected with HIV has several drawbacks. Drug-resistant oral candidiasis caused by Candida krusei, Torulopsis glabrata, or Candida albicans has been observed in patients infected with HIV who are receiving fluconazole [91-93]. Fluconazole-resistant Cryptococcus neoformans has also been reported and appears to be associated with a poor response to oral therapy for cryptococcal meningitis [94].

Esophageal candidiasis accounts for approximately 15% of initial AIDS diagnoses, and oropharyngeal and vaginal candidiasis are exceedingly common [95, 96]. Nevertheless, candidal infections may not be appropriate targets for primary prophylaxis because they respond rapidly to treatment with topical or oral antifungal agents. However, patients who develop recurrences may require long-term suppressive therapy. Suppression of oropharyngeal candidiasis can be accomplished with topical agents (such as clotrimazole or nystatin) or with systemic agents (such as ketoconazole or fluconazole). Fluconazole (at doses of 50 to 100 mg daily or 150 mg weekly) has been shown to prevent recurrent oral candidiasis [97-99]. Patients with recurrent esophageal candidiasis often require suppressive therapy with systemic antifungal agents. Fluconazole is more effective than ketoconazole in the treatment of candida esophagitis [100], and it also appears to be effective in secondary prophylaxis [101]. Although few comparative studies exist of secondary prophylaxis for candida esophagitis, fluconazole is more consistently absorbed, has a longer half-life, and has greater in vivo activity against Candida species than does ketoconazole [102, 103]. It is more expensive, however, and many patients may achieve satisfactory suppression of candidiasis with ketoconazole.

Cryptococcal meningitis causes substantial morbidity and mortality, and often requires prolonged hospitalization and treatment with relatively toxic drugs. Although the 2-year incidence of cryptococcosis in patients with advanced HIV disease is less than 10%, the morbidity, mortality, and cost of therapy for cryptococcal meningitis make it an attractive target for primary prophylaxis. Nightingale and colleagues [104] treated 329 patients infected with HIV who had CD4 counts less than 68/mm³ with fluconazole (100 mg daily) during a 1-year period [104]. The risk for invasive fungal infection at 1 year was estimated to be 1.8% when compared with 7.5% in a group of 337 untreated historical control patients with similar CD4 cell counts. In a randomized trial of high-dose fluconazole (200 mg daily) compared with clotrimazole troches (10 mg 5 times daily) for primary prophylaxis of fungal infections in patients with CD4 counts less than 200/mm³, patients receiving fluconazole had a statistically significant lower incidence of cryptococcosis and invasive fungal infections. The differences were most pronounced among patients with CD4 counts less than 50/mm³. No difference was noted in mortality, and data on resistant fungal infections were not collected (Powderly WG. Unpublished observations).

Recurrence of cryptococcal meningitis occurs in 40% to 60% of successfully treated patients who do not receive suppressive therapy [105, 106]. Patients who have completed therapy should receive life-long fluconazole suppression [107]. Itraconazole, although it penetrates poorly into cerebrospinal fluid, may be an alternative for patients unable to take fluconazole. Relapses of cryptococcal disease may result from dissemination of organisms in extraneural sites, such as the prostate [108]. Thus, azoles that do not penetrate into the cerebrospinal fluid may still be effective in preventing recurrences.

Disseminated histoplasmosis occurs in 2% to 25% of patients with AIDS who live in areas endemic for Histoplasma capsulatum [109, 110]. In areas of high endemicity, patients with advanced HIV disease may benefit from primary prophylaxis for histoplasmosis, but no data are available to support this practice. Prevention of relapse is essential because more than 60% of successfully treated patients will have a recurrence without secondary prophylaxis [109]. Neither amphotericin B nor ketoconazole is sufficiently effective for secondary prophylaxis of histoplasmosis [109, 111]. Itraconazole (200 mg twice daily) was effective in preventing recurrences of histoplasmosis in 42 patients who had completed primary therapy [112]. Fluconazole may also be effective for secondary prophylaxis, although adequate trials have not been reported.

Patients infected with HIV who have lived in areas endemic for Coccidioides immitis are at risk for severe coccidioidomycosis [113]; patients with CD4 counts less than 250/mm³ or AIDS are at highest risk [114]. Primary prophylaxis for coccidioidomycosis has not been studied. Coccidioidomycosis has been reported in patients taking ketoconazole for other reasons [113], and the limited data available support the use of itraconazole (200 mg twice daily) or fluconazole (400 mg daily) as maintenance therapy [113, 115, 116].

Mycobacterium avium complex

Disseminated infection with M. avium complex is the most common opportunistic bacterial infection in patients with AIDS. A late manifestation of HIV disease, it rarely occurs in patients with more than 100 CD4 cells/mm³, and the median CD4 count is often less than 20/mm³ at the time of diagnosis [10, 117-119].

Rifabutin, a semi-synthetic rifamycin, was efficacious in preventing M. avium complex bacteremia in two randomized controlled trials comparing rifabutin (300 mg daily) with placebo in patients who had CD4 counts less than 200/mm³ [120]. The mean duration of prophylaxis was approximately 270 days. Rifabutin was found to decrease the incidence of M. avium complex bacteremia by 50% and to prolong the time to fever, fatigue, anemia, increases in alkaline phosphatase levels, and decreases in Karnofsky performance score. The decrease in M. avium complex bacteremia was 70% when the analysis was restricted to those patients who were taking assigned treatment. The efficacy was most pronounced in patients whose initial CD4 count was less than 75/mm³. Fewer patients in the rifabutin arm required hospitalization, but no significant decrease in mortality was noted. Rifabutin was well tolerated with no serious toxic reactions. A U.S. Public Health Service Task Force recommended that rifabutin prophylaxis be given to patients with CD4 counts less than 100/mm³ [121]. Clinical evaluation to rule out active mycobacterial disease (including tuberculosis) is important in patients who are candidates for rifabutin therapy. Like rifampin, rifabutin induces levels of hepatic microsomal enzymes, which causes altered metabolism of oral contraceptives, warfarin, dilantin, methadone, and zidovudine. Patients taking rifabutin should be monitored closely for evidence of drug interactions.

Clofazimine (50 mg daily) did not prevent disseminated M. avium complex disease in a small, open-label trial in which 110 patients with advanced HIV disease were randomly assigned to receive either clofazimine or no treatment [122]. Clarithromycin and azithromycin are effective in treating M. avium complex bacteremia [123-125]. Studies evaluating these new macrolides for the prevention of M. avium complex infection are now in progress. Use of a macrolide as prophylaxis may be appropriate for patients unable to tolerate rifabutin. Dapsone has in vitro activity against mycobacteria [126, 127], and dapsone-pyrimethamine (which is currently used to prevent P. carinii pneumonia and toxoplasmosis) may also help prevent M. avium complex disease [128, 129].

Mycobacterium tuberculosis

Persons infected with HIV are uniquely susceptible to developing active tuberculosis. The risk for reactivation of latent infection is estimated to range from 2% to 10% per year compared with a lifetime risk for 10% in immunocompetent persons [130]. A person with HIV infection who acquires new M. tuberculosis infection has a high risk for developing primary disease [131].

Candidates for chemoprophylaxis are persons with a tuberculin skin test (5 tuberculin units using purified protein derivative [PPD]) reaction 5 mm of induration, a history of a positive skin test reaction, or a recent history of close contact with a person with infectious tuberculosis. Because HIV-induced anergy decreases the sensitivity of a tuberculin skin test, the interpretation of PPD tests in persons infected with HIV is problematic. The Centers for Disease Control and Prevention (CDC) recommend that control skin test antigens (such as candida, mumps, or tetanus toxoid) be used to help interpret the tuberculin test and that prophylaxis be considered for anergic persons from populations at risk for tuberculosis [132]. Two small, retrospective studies [133, 134] suggested that HIV-infected, anergic intravenous drug users have a risk for tuberculosis similar to that of patients infected with HIV who are positive for PPD. In one large cohort study [135], however, the prevalence of tuberculin positivity was 7% in patients who were anergic to control antigens and was 6% in nonanergic persons.

The efficacy of isoniazid prophylaxis in HIV-infected persons is thought to be similar to that in nonimmunosuppressed patients [130]. In a controlled clinical trial [136] in Haitian adults infected with HIV, 6 months of daily isoniazid and pyridoxine decreased the risk for tuberculosis by 80% compared with pyridoxine alone and appeared to be associated with improved survival [136]. The benefit occurred in persons with a positive tuberculin skin test result at entry. Zambian patients with HIV infection who were randomly assigned to receive 6 months of daily isoniazid had a 60% decrease in the risk for tuberculosis compared with patients who received a B-complex vitamin [137]. The optimal duration of prophylaxis in persons infected with HIV is unknown. The CDC recommends that prophylaxis be administered for 12 months. Studies of shorter duration prophylaxis are under way. Twice-weekly isoniazid, 15 mg/kg, may be substituted for daily therapy in patients whose therapy can be supervised. Pyridoxine, 50 mg daily, is often added to isoniazid therapy to decrease the incidence of peripheral neuropathy.

For patients unable to take isoniazid, rifampin may be given for 6 to 12 months [138]. An alternative regimen of rifampin and pyrazinamide given for 2 months is under investigation in several trials, based on superior activity in the animal model of chronic tuberculous infection [139]. Rifabutin has activity against M. tuberculosis (similar to rifampin) and may be an effective alternative [140], although clinical trials have not yet been conducted.

Persons infected with HIV who are exposed to patients with multidrug-resistant tuberculosis pose a special problem [141]. The CDC recommends that such patients receive high-dose ethambutol and pyrazinamide, with or without a fluoroquinolone. Because many of the outbreak strains of multidrug-resistant tuberculosis are resistant to ethambutol and pyrazinamide, however, the use of a fluoroquinolone (such as ofloxacin or ciprofloxacin) may be essential. Susceptible and multidrug-resistant strains are inhibited by concentrations of these drugs of 1 to 2 µg/mL, levels that can be achieved with oral doses of 400 to 600 mg of ofloxacin or 750 mg of ciprofloxacin. The duration of prophylaxis against multidrug-resistant tuberculosis in a person infected with HIV should be at least 1 year.

Herpesvirus Infections

Cytomegalovirus is the most common cause of serious opportunistic viral disease in patients with HIV infection [142]. Autopsy studies have shown that as many as 90% of AIDS patients develop active cytomegalovirus infection and that up to 40% may develop life- or sight-threatening disease [143-145]. Patients with a baseline CD4 count less than 100/mm³ have a 2-year probability of 21% of developing end-organ disease, which is an independent predictor of mortality [11]. Although the association with mortality has not been shown to be causal, the high incidence, substantial morbidity, and high cost and toxicity of therapy make cytomegalovirus disease an excellent target for prophylaxis.

Conflicting data are available from studies in solid-organ transplant recipients about the efficacy of high-dose acyclovir in preventing cytomegalovirus disease [146-148]. In a double-blind, randomized, placebo-controlled trial of combination therapy with zidovudine (250 mg every 6 hours) and acyclovir (800 mg every 6 hours) compared with zidovudine alone in patients with advanced HIV disease, acyclovir did not decrease the frequency of cytomegalovirus disease [149].

Studies of valacyclovir, an acyclovir congener that is rapidly metabolized to acyclovir and that results in serum acyclovir levels three- to fourfold higher than oral acyclovir [150] are in progress. Oral preparations of ganciclovir and foscarnet are also being studied in early trials. If oral prophylaxis becomes available, the most likely candidates would be persons with CD4 counts less than 100/mm³ who are seropositive for anti-cytomegalovirus antibodies [11]. The presence of cytomegalovirus viremia or viruria may be useful in identifying patients at high risk for end-organ disease [151], although in the absence of effective prophylaxis, it is not useful to obtain viral cultures of blood or urine [152].

Patients with cytomegalovirus retinitis require life-long treatment to prevent recurrence because 86% of patients relapse within 120 days without suppressive therapy [153]. The agent used for initial therapy, foscarnet or ganciclovir, is typically given daily for life, or until relapse (Appendix Table 1). Combination ganciclovir and foscarnet may be more effective than either agent alone.

Infections with herpes simplex virus are common among patients infected with HIV. Frequent recurrence is the rule, and oral acyclovir (600 mg to 800 mg daily) is effective as secondary prophylaxis [154-156]. In one trial, the combination of zidovudine and acyclovir decreased initial and recurrent herpesvirus infections compared with zidovudine alone [149]. The use of suppressive acyclovir in patients with HIV infection has been associated with the emergence of thymidine-kinase deficient, acyclovir-resistant herpes simplex infections [157]. The incidence of disease caused by these resistant isolates appears to be low.

Varicella zoster virus infection occurs with great frequency in persons infected with HIV and often presents before the onset of AIDS [158, 159]. Because most adults have had primary varicella (chickenpox) during childhood, patients infected with HIV typically develop dermatomal zoster (shingles). Patients at risk for primary varicella should be given varicella zoster immune globulin after exposure to persons with active varicella infection. Because immune globulin is only protective if given within 4 days of exposure, it may be useful to obtain baseline results of anti-varicella IgG levels in patients who cannot recall having chicken- pox during childhood. Recurrences of herpes zoster are less frequent than with herpes simplex virus infection, and secondary prophylaxis is not usually necessary [160].

Bacterial Infections

The incidence of pyogenic bacterial infections in persons infected with HIV is high, and bacterial infections are independent predictors of progression to AIDS [161]. Recent reports [162, 163] indicate that the incidence of serious bacterial infections increases as the CD4 cell count decreases and that smoking tobacco or other drugs is an independent risk factor for bacterial pneumonia.

The most common pathogens causing pneumonia or bacteremia in adults infected with HIV are Streptococcus pneumoniae and Haemophilus influenzae. Other types of bacterial infections that occur with increased frequency include sinusitis [164]; central venous catheter infections [165, 166]; and infections caused by Salmonella species [167], Nocardia species [168], and Rhodococcus equi [169, 170].

The use of trimethoprim-sulfamethoxazole for pneumocystis prophylaxis decreases the risk for bacterial infections [8, 17, 171]. The incidence of serious infections (including pneumonia, bronchitis, and sinusitis) was 50% lower than in patients receiving aerosol pentamidine in a study of secondary prophylaxis [8]. The distribution of pneumococcal serotypes that cause pneumonia and bacteremia is similar in patients with and without HIV infection; 86% of serotypes isolated from patients infected with HIV in one study are included in the 23-valent pneumococcal polysaccharide vaccine [172]. For this reason, pneumococcal vaccination is recommended for all adults as early as possible in the course of HIV infection [173-175]. Efficacy studies are lacking, however, and vaccine failures have been reported [176-178]. Zidovudine therapy may improve antibody responses to the vaccine, and immunization can be delayed for 4 weeks in unvaccinated patients who are beginning zidovudine therapy [179].

Use of a conjugated H. influenzae type b vaccine has been suggested, although efficacy data are lacking [174, 175]. A recent study suggests that the risk for invasive H. influenzae type b infection is increased by HIV infection [180], although the actual number of patients is probably small. The vaccine for H. influenzae (a polysaccharide-mutant diphtheria-toxoid conjugate vaccine [PRP-CRM]) was more immunogenic than the polysaccharide (PRP) vaccine in persons infected with HIV before the onset of AIDS [181]. The antibody response to the PRP-CRM vaccine was highly correlated with CD4 count.

Monthly intravenous IgG was shown to be superior to placebo in preventing bacterial infections in children infected with HIV [182]. Incidence rates per 100 patient-years in the IgG and placebo groups were 115 compared with 160 for minor bacterial infections (P = 0.01) and 26 compared with 48 for serious bacterial infections (P = 0.002). However, a comparison of IgG with trimethoprim-sulfamethoxazole has not been done.

Conclusions and Future Directions

The widespread use of pneumocystis prophylaxis has had a dramatic impact on the natural history of HIV disease, resulting in decreased morbidity and prolonged survival but in increased risk for developing other opportunistic infections. As efforts to extend the success of prophylaxis to other pathogens increase, several new issues have arisen. Drug interactions between prophylactic and antiretroviral agents are an increasing problem. For example, rifabutin and clarithromycin have been shown to decrease the serum concentration of zidovudine [183, 184]. Didanosine may impair the absorption of dapsone [47]. Fluconazole results in increased serum levels of rifabutin, and recent reports indicate that rifabutin may decrease serum levels of clarithromycin. As the number and types of drug therapies increase, these problems will assume greater importance.

An additional problem is the cost of preventing opportunistic infections. Although some prophylactic regimens are relatively inexpensive, others (such as aerosolized pentamidine and rifabutin) cost thousands of dollars per patient per year. The emphasis on cost containment in health care will result in increased scrutiny of these therapies; however, even expensive regimens may be cost-effective because of the relative expense of hospitalization and of the treatment of acute disease. For example, in a cost–benefit analysis of secondary pneumocystis prophylaxis, the strategy of using trimethoprim-sulfamethoxazole as a first-line agent and reserving aerosolized pentamidine for patients who were intolerant of trimethoprim-sulfamethoxazole resulted in prolonged survival and a savings of $16 503 per patient compared with not using prophylaxis [185]. Although the use of aerosolized pentamidine as a first-line agent cost $2904 more per patient, both strategies were more cost-effective than not using prophylaxis. In a review of all 79 patients with P. carinii pneumonia at Johns Hopkins Hospital in 1991, 61 (77%) had not been receiving pneumocystis prophylaxis either because of previously undiagnosed HIV infection, failure to return for outpatient appointments, or noncompliance with therapy [186]. Death occurred only in patients not receiving prophylaxis; they also accounted for 85% of the hospital days, 100% of the intensive care unit days, and 89% of the total inpatient charges (unpublished observations).

Drug interactions and costs may be minimized by limiting the number of agents used. So-called "multiple opportunistic pathogen prophylaxis" attempts to identify drugs whose broad spectrum will prevent infections caused by different classes of infectious agents. An example is the use of trimethoprim-sulfamethoxazole to prevent P. carinii pneumonia, toxoplasmosis, and serious bacterial infections. Other agents under study include the macrolides (to prevent mycobacteria, P. carinii, Cryptosporidium species, and T. gondii) and rifabutin (to prevent mycobacteria, bacteria, and T. gondii). Nevertheless, a combination of several agents will probably be required to prevent the serious infections to which patients infected with HIV are susceptible.

Finally, with the spread of HIV to new populations and with advances in microbial detection and identification, new opportunistic pathogens will be discovered. In Thailand, for example, the fungus Penicillium marneffei has been shown to cause disseminated disease in a large proportion of patients infected with HIV [187]. The recent elucidation of the cause of bacillary angiomatosis and peliosis using molecular techniques has led to the recognition of more prevalent Rochalimaea quintana and Rochalimaea henselae infections [188-190].

Continued progress in antiretroviral therapy and in prophylaxis for opportunistic infections will undoubtedly prolong survival and improve the quality of life in patients with advanced HIV disease. Changes in patient populations and the microbial causes of opportunistic disease will continue to present considerable challenges to physicians caring for persons with HIV infection.