540
AIDS COMMENTARY
Impact of Human Immunodeficiency Virus Infection on the Epidemiology, Clinical Features, Management, and Control of Tuberculosis Philip C. Hopewell
From the University of California, San Francisco, and the Division of Pulmonary and Critical Care Medicine, San Francisco General Hospital. San Francisco. California
-Merle A. Sande
Although the association of tuberculosis with AIDS was described in early reports of the syndrome [1-3], this important disorder received little public or scientific attention until recently [4]. Now, because of both the increasing number of cases and the occurrence of outbreaks of tuberculosis caused by multiple-drug-resistant (MDR) organisms [5], funding agencies, investigators, public health officials, and others who heretofore had paid little attention to tuberculosis are hastily attempting to make up for the lack of attention and resources directed toward the disease [4]. The epidemic of human immunodeficiency virus (HIV) infection has exposed the fundamental weaknesses of local programs for control-
Received 15 June 1992. Reprints or correspondence: Dr. Philip C. Hopewell. Professor of Medicine. Box 0841. University of California. San Francisco. 1001 Potrero Avenue, San Francisco. California 94110.
Clinical Infectious Diseases 1992;15:540-7 © 1992 by The University of Chicago. All rights reserved.
1058-4838/92/1503-0021$02.00
ling tuberculosis. Consequently, many communities are not equipped to confront the increasing number of cases or the shifting epidemiological and clinical features of the disease [6, 7]. Moreover, because of the inattention of the scientific community to tuberculosis (a surprising circumstance in light of its causing 2.5-3 million deaths per year worldwide before the HIV epidemic), there are no new tools with which to meet the increasing challenge [4]. The importance of tuberculosis in the context of HIV infection relates to at least four factors. The first of these is the high prevalence of tuberculosis in certain HIV-infected groups. Second, tuberculosis is probably the only HIV-related infection that is transmitted from person to person regardless of whether the exposed person is infected with HIV. Third, if tuberculosis is diagnosed promptly and treated appropriately, it is highly likely that it will be cured [8]. Finally, there is early evidence that tuberculosis can be prevented in HIV-infected populations [9]. This review presents an overview of the impact that HIV infection is having on tuberculosis and on its control primar-
Downloaded from http://cid.oxfordjournals.org/ at Lancaster University on October 24, 2014
After years of decreasing prevalence and increasing hope that tuberculosis, like smallpox, could be eliminated, the disease has resurfaced as a major public health problem in the United States. Particularly ominous are the appearance of multiple-drug-resistant strains and their impact on patients and health care workers who are infected with the human immunodeficiency virus, among whom mortality rates reach 80% 2-3 months postdiagnosis. To respond effectively to this new threat, it is critical that we reorient our thinking about tuberculosis and redirect health care resources to programs for tuberculosis control. We need to reinstitute screening of high-risk populations and ensure proper isolation of patients with the disease. Diagnosing tuberculosis at the earliest possible stage is obviously of the utmost importance. High priority must be given to the development of rapid diagnostic tests and techniques that screen for drug resistance. We must implement and adequately fund drug-discovery programs to develop new therapeutic agents that are effective against multiple-drug-resistant strains of Mycobacterium tuberculosis. Effective programs for monitoring the treatment of patients with tuberculosis must also be implemented. Failure to adequately support such programs has probably led to the recent upswing in multiple-drug-resistant tuberculosis, especially in large cities along the eastern seaboard. Leadership for and funding of these programs must come from the federal government, specifically the U.S. Department of Health & Human Services and the Centers for Disease Control (Atlanta). The Infectious Diseases Society of America is actively supporting a variety of tuberculosis control-related initiatives and will keep its members updated on progress in this area. We plan to publish a series of commentaries focusing on tuberculosis during the coming year. We invited Dr. Philip C. Hopewell to initiate this series with an AIDS Commentary that provides an overview of the critical interaction between infection with human immunodeficiency virus and tuberculosis. Dr. Hopewell is a professor of medicine at the University of California, San Francisco, and head of the Division of Pulmonary and Critical Care Medicine at San Francisco General Hospital.
CIO 1992; 15 (September)
HIV Infection and Tuberculosis
ily in the United States. Subsequent reviews will describe other specific aspects of the interaction between tuberculosis and HIV infection.
Pathogenesis of Tuberculosis in HIV-Infected Persons
losis developed tuberculosis within 120 days [13]. By the use of RFLP analysis it was substantiated that, in all 11 of the culture-positive patients, tuberculosis was caused by the same strain of M. tuberculosis. This pathogenetic sequence is undoubtedly responsible for the explosive outbreaks of tuberculosis caused by MDR organisms that have recently occurred [5]. Presumably because of the pathogenicity of M. tuberculosis, tuberculosis tends to occur relatively early in the course of HIV infection. This is attested to by the finding ofTheuer and associates [15] that HIV-seropositive patients with tuberculosis had a median circulating CD4 lymphocyte count of 354/f.tL. Among this group of 17 patients, tuberculosis was the initial manifestation of HIV infection in all but two patients; one of these patients had limited Kaposi's sarcoma, and the other had oral candidiasis.
Prevalence of Tuberculosis Among Patients with HIV Infection Matching of tuberculosis and AIDS registries in 43 states has revealed that 3.8% of AIDS cases appeared on tuberculosis case registries [16]. However, this percentage is widely variable in different states and areas. In Florida 10% of patients with AIDS have or have had tuberculosis, and in Connecticut and New York 5% have had the disease. In a few states, no cases in patients with AIDS have been reported. In Africa, although there are no precise data defining the incidence of tuberculosis in HIV-infected populations, it is clear that tuberculosis is the most common HIV-associated lung disease. In Dar Es Salaam, Tanzania, 51% of HI V-infected patients who were evaluated because of lung disease were found to have pulmonary tuberculosis [17]. These data confirm numerous anecdotal reports of the frequency with which tuberculosis is diagnosed as an HIV-associated disease in Africa. The frequency with which tuberculosis develops in persons who are infected with both HIV and M. tuberculosis is not well established but appears to be substantially in excess of what would be expected in a non-HIV-infected group. Selwyn and co-workers [ 18] performed a prospective study of tuberculosis among HI V-seropositive and -seronegative iv drug users in New York City. Of 49 HIV-seropositive subjects, seven drug users (14%) who were known to have had a positive tuberculin test previously and one anergic patient developed tuberculosis in a 2-year period. This was the equivalent of 7.9 cases per 100 person-years ofobservation. These observations suggest that seven or perhaps all eight patients who developed tuberculosis had had preexisting tuberculous infection, thus indicating that endogenous reactivation was the dominant if not the only pathogenetic mechanism for development of tuberculosis. Of additional concern was the observation that I I% and 13% of the seropositive and seronegative subjects, respectively, developed positive reactions to tuberculin tests during the study. If this observation is
Downloaded from http://cid.oxfordjournals.org/ at Lancaster University on October 24, 2014
There are two pathogenetic sequences by which tuberculosis can develop: (1) direct progression from recently acquired infection to disease and (2) recrudescence of previously acquired, latent infection. It is generally thought that in areas in which the prevalence of tuberculosis is low, most cases arise from latent infections because few new infections are occurring [10]. However, because HIV impairs the mechanisms by which new tuberculous infection is contained, the risk of direct progression is much greater in the presence of HIV infection; thus, an increasing number of cases may be occurring via this sequence. In addition, as the number of new cases of tuberculosis providing more sources ofinfection increases, there will be more transmission of infection and, thus, more opportunity for direct progression. Because of this shift in the pathogenesis of tuberculosis, the epidemiology of tuberculosis in some parts of the United States resembles that in developing countries. The mechanisms by which a contained tuberculous infection is kept quiescent are not well understood but clearly involve cell-mediated immunity [11]. Because of the effect HIV infection has on cell-mediated immunity, the likelihood of reactivation oflatent tuberculous infection leading to clinical tuberculosis is increased. Thus, persons who had been infected with Mycobacterium tuberculosis before developing HIV infection are at considerably increased risk of developing tuberculosis after being infected with HIV. In the healthy host, once the cell-mediated immune response to infection with M. tuberculosis develops, there is little likelihood of new exogenous infection being acquired [10]. However, because of the HIV-induced immune defect, an HIV-infected person who has been previously infected with M. tuberculosis may still be vulnerable to new infection. Reinfection has been documented under other circumstances as well [12]. Analysis of preliminary data with use of restriction fragment length polymorphism (RFLP) analysis suggests that reinfection can occur and may account for apparent relapses after successful completion of antituberculosis therapy (P.M. Small, personal communication). It has been speculated that HIV -infected patients are more likely to acquire tuberculous infection when exposed to M. tuberculosis [13]. Although this concept is presently unsubstantiated, it is clear that once an HIV-infected person becomes infected with M. tuberculosis, the infection can progress very rapidly to clinical disease [13, 14]. In situations where groups of HI V-infected persons are exposed to a patient with infectious tuberculosis, explosive outbreaks of tuberculosis may occur. For example, in a residential care facility for HIV-infected persons in San Francisco, 11 of 31 residents (35%) exposed to a person with infectious tubercu-
541
542
Hopewell
CID 1992; 15 (September)
truly indicative of new infections, it suggests that there was a large number of infectious cases within the population. Allen and colleagues [19] prospectively followed a group of urban women of childbearing age in Kigali, Rwanda, and found that the incidence oftuberculosis was ....... 2.5% per year. In comparison with HIV-negative women, the risk ratio for tuberculosis among the HIV-positive women was 22.9. For the HIV-positive women, a positive tuberculin test was associated with a significantly increased risk of tuberculosis, but the risk ratio was only 3; 9 of 17 patients with tuberculosis had negative tuberculin tests, perhaps as a result of anergy.
profound effects have occurred in New York City. Between 1984 and 1991 the case rate of tuberculosis increased from 19.9 to 36 per 100,000 persons, an 81%rise. The increase in case rates was even more dramatic in the central Harlem area, where in 1989 the rate reached 169/100,000 persons, a case rate quite similar to those in eastern and central Africa [6]. In 1979 the rate in central Harlem was 50.9/1 00,000 persons.
Prevalence of HIV Infection Among Patients with Tuberculosis
Tuberculin skin testing and anergy testing. As would be expected, persons with advanced HIV infection commonly have little or no reaction to the tuberculin test. However, in earlier stages of the infection, reactivity may be maintained. The ability to respond to tuberculin is an indicator of the status of cell-mediated immunity, which in turn is an indicator of the stage of HIV infection. Most studies that describe the prevalence of reactive (indurations of > 10 mm) tuberculin tests indicate that ....... 40% of HIV-infected patients with tuberculosis have positive reactions. However, in the study by Theuer and associates [15], 80% of the HIV-seropositive patients with tuberculosis had positive reactions to tuberculin. Presumably, this finding relates to the fact that these patients' HIV disease was in the early stages. Because of the frequency of blunted skin test responses or anergy, it has been recommended that an induration of ~5 mm in reaction to 5 tuberculin units ofpurified protein derivative be regarded as indicative of tuberculous infection in HI V-infected persons [16]. The implications of using 5 mm as the cutoff point for defining tuberculous infection have not been determined. In a recent study Graham and associates [29] found that a 5-mm cutoff for determining tuberculin positivity probably resulted in underestimation of the rates oftuberculous infection, so they recommended a 2-mm cutoff. The practicality of using such a small reaction size to define infection would appear to impose significant limitations on the use of tuberculin skin tests, however [30]. To determine if a negative tuberculin test is the result of immunosuppression or is truly negative, control antigens should be applied. These include antigens made from Candida organisms, mumps virus, and tetanus toxoid, to which most persons with intact cell-mediated immunity will respond. A negative tuberculin reaction in the presence of'positive reactions to one or more of the control antigens can be interpreted as a true negative. Ifthere is no reaction to any of the antigens, it cannot be determined if the negative tuberculin test is a true or false negative [31]. Clinical features of tuberculosis. The clinical manifestations oftuberculosis occurring in patients with HIV infection vary considerably, depending on the severity ofthe immunosuppression [1-3, 15, 32-36]. As noted previously, tuberculosis tends to occur early in the course of HIV infection, presumably because of the virulence of M. tuberculosis. In
Impact of HIV Infection on the Incidence of Tuberculosis The full impact of HIV infection on the epidemiology of tuberculosis has not been defined in the United States or abroad. Inferential information suggests, however, that the influence may be substantial. Data from the CDC indicate that there have been ....... 28,000 cases oftuberculosis in excess of the number that would have been expected in the United States between 1984 and 1990 [28]. Although the excess cases that have occurred since 1984 have not been proven to be attributable to HIV infection, the largest increases in incidence have occurred in the areas with the greatest number of AIDS cases and, therefore, the highest prevalence of HIV infection. Use of countrywide data tends to result in underestimation of the effects of HI V infection on the incidence oftuberculosis in certain parts ofthe United States. Perhaps the most
Downloaded from http://cid.oxfordjournals.org/ at Lancaster University on October 24, 2014
Several prospective studies in public tuberculosis clinics have described the prevalence of HIV infection among patients with tuberculosis. Pitchenik and co-workers [20] reported that 31% of 71 consecutive tuberculosis patients in Miami were infected with HIV. In Seattle, Nolan and associates [21] found that 23% of non-Asian adults with tuberculosis were infected with HIV. Theuer and colleagues [15] reported that among a group studied in San Francisco, 28% of non-Asian patients between 15 and 55 years of age who had tuberculosis were seropositive for HIV. Beginning in 1988, a systematic nationwide sampling to determine the prevalence of HIV infection was undertaken by the Centers for Disease Control (CDC; Atlanta) in 14 urban tuberculosis clinics [22]. The median rate of seropositivity among 4,30 I persons who had or were suspected of having tuberculosis was 3.4%. The rates varied widely among clinics, ranging from 0 to 46%. The highest rate was reported from New York City (46%), followed by Newark, New Jersey (34%), Boston (27%), Miami (24%), and Baltimore (13%). Data from developing countries are scarce but indicate a substantial rate of HIV infection among patients with tuberculosis [23-27]. This finding would be expected given the concomitant high prevalence of tuberculosis infection and infection with HIV in developing countries.
Diagnosis of Tuberculous Infection and Tuberculosis in HIV-Infected Patients
ClD 1992; 15 (September)
HIV Infection and Tuberculosis
atypical findings was noted in the prospective study by Pitchenik and Rubinson [48]. These findings included a higher frequency of adenopathy and diffuse or lower-lobe infiltration and a lower frequency ofcavitation for HIV-seropositive patients than for HIV-seronegative patients. Bacteriologic and histologic examinations. Most reported series indicate that the prevalence of positive sputum smears and cultures in patients with pulmonary tuberculosis is the same in HIV-infected and -noninfected persons [15.20,34]. In some instances, sputum induction or bronchoscopic procedures have been necessary for diagnosing pulmonary tuberculosis. Specimens from any site of abnormality in patients who have or are suspected of having HIV infection should be examined for mycobacteria by means ofsmear and culture. Potential high-yield sources include lymph nodes, bone marrow, urine, and blood. In general, any acid-fast organism identified in any specimen should be regarded as being M. tuberculosis until proven otherwise. Such a policy will result in prompt initiation of appropriate therapy for tuberculosis and evaluation of persons in contact with the infected patient. With standard techniques, detection of growth and identification of the species of mycobacteria require 6-10 weeks. Radiometric culture techniques and DNA probes for identification of M. tuberculosis and Mycobacterium avium complex can shorten this procedure to 7-10 days, thereby adding greatly to the speed and efficiency of control measures for tuberculosis [49]. The use of the polymerase chain reaction for identification of M. tuberculosis will result in much more rapid diagnoses [50]. This technique should be available soon for routine application and will represent a major advance in the control of tuberculosis. In patients with more advanced HIV infection, mycobacterial infection does not produce classic granulomas. However, because tuberculosis tends to occur when HIV disease is less advanced, the ability to form granulomas may be intact. Thus, the finding of granulomas in either tissue sections or cytological preparations from needle aspiration biopsy should be interpreted as being more consistent with tuberculosis than nontuberculous mycobacterial disease.
Treatment of Tuberculosis in HIV-Infected Patients Most reported series of patients with tuberculosis and HIV infection demonstrate a good response to antituberculosis treatment when regimens including isoniazid and rifampin are used [2, 3. 8, 15. 20, 32-35]. However, in the series reported by Sunderam and co-workers [33], there were three patients who did not respond to treatment and had progressive disease. A case report from the same institution described a recurrence of tuberculosis at an extrapulmonary (scrotal) site after apparently successful treatment of pulmonary disease [44]. These reports have led to expressions of concern regarding the adequacy of currently standard 6month therapy and, subsequently, to recommendations that a total of 9 months or 6 months beyond the time of sputum
Downloaded from http://cid.oxfordjournals.org/ at Lancaster University on October 24, 2014
most series of cases the majority of tuberculosis diagnoses have preceded the identification of an AIDS-defining disease. For a substantial number of patients, tuberculosis and AIDS are diagnosed at the same time, and for a smaller number, tuberculosis appears after a diagnosis of AIDS. This is in part a semantic distinction because the surveillance definition of AIDS, revised in 1987, now includes extrapulmonary tuberculosis in an HIV-seropositive person [37]. It seems that the earlier tuberculosis develops, the more usual its clinical presentation, whereas the later it occurs, the more atypical its features. This conclusion is somewhat difficult to support with published information, owing largely to the fact that most series of cases in the literature describe the features of tuberculosis in patients who either have or subsequently develop AIDS. Patients identified by cross-matching AIDS and tuberculosis registries clearly are likely to have advanced HIV disease even if their tuberculosis develops before an AIDS-defining diagnosis is made. Clinical reports have emphasized that tuberculosis in cases of advanced HIV infection is frequently disseminated, has unusual radiographic manifestations. and is associated with nonreactive tuberculin tests. Involvement of the lymph nodes, including intrathoracic adenopathy. has been described frequently. A variety of unusual manifestations have been noted. These include CNS involvement with brain abscesses, tuberculomas, and meningitis [33, 38, 39]; bone disease, including vertebral disease [33, 40]; pericarditis [33, 41]; gastric tuberculosis [42]; tuberculous peritonitis [43]; and scrotal tuberculosis [44]. In addition, M. tuberculosis has been cultured from the blood as well as bone marrow [45-47]. Despite the increased frequency of unusual forms of tuberculosis in persons with HIV infection. several reports have described a predominance of "standard" pulmonary disease [15, 18, 21]. These reports presented series ofcases collected prospectively, in which either HIV antibody was measured in patients with tuberculosis or HIV-seropositive subjects were followed for evidence of the development of tuberculosis. Thus, these patients presumably were less immunocompromised. Radiographic findings. The atypical findings on chest radiographs of HIV-infected patients who have tuberculosis have received considerable emphasis. In retrospective studies, radiographical features that are not regarded as typical for pulmonary tuberculosis have been the norm [34, 48]. Lower-lung-zone or diffuse infiltrations rather than the usual upper-lobe involvement commonly have been observed. Cavitation has been unusual, and intrathoracic adenopathy, an unusual finding in immunocompetent adults with tuberculosis, has been relatively frequent. In the prospective study by Theuer and colleagues [15], the radiographic findings for patients with HIV infection were not distinguishable from those for patients who were HIV seronegative. The findings included typical upper-lobe infiltration, often with cavitation. A greater prevalence of
543
544
Hopewell
Tuberculosis Caused by MDR Organisms Recently, explosive outbreaks of tuberculosis caused by MDR organisms (usually resistant at least to both isoniazid
and rifampin) have been reported [5, 54]. Although these outbreaks have taken place in hospital clinics and correctional institutionsand have predominantly involved HIV-infected patients, health care workers, both HIV seropositive and seronegative, have also been infected. Substantial epidemiological and laboratory (RFLP analysis) data indicate that transmission of resistant M. tuberculosis has taken place in these facilities. These outbreaks have been characterized by high case fatality rates, ranging from 72% to 89%within median periods of 4-16 weeks. Also noteworthy are the relatively high rates of tuberculin test conversions among health care workers exposed to these patients. A combination of at least three major factors have produced these outbreaks. The first is a relatively high prevalence of multiple-drug resistance in the community at large and particularly in groups that in some areas are most likely to be infected with HIV. This prevalence of drug resistance is a predictable outcome of the lack of attention and resources devoted to tuberculosis during at least the last decade and provides a reservoir of MDR organisms that are capable of being transmitted. The second factor is the effect ofHI V on the host response to tuberculous infection. As described, recently acquired infection with M. tuberculosis in an HIV-infected person may progress rapidly to clinical disease that in turn is capable of being transmitted. If the transmitted organism is MDR, all of the secondary infections will be caused by MDR organisms. The third factor relates to the fact that tuberculosis in HIV-infected persons may not be as easily recognizable as that in persons with a normal immune status. For this reason the disease may go undiagnosed, perhaps even in a hospital or clinic, for a relatively prolonged period. During this time, unless adequate infection control measures are applied presumptively, infected patients will be capable of transmitting the infection. Moreover, even if the disease is diagnosed, the fact that the organisms are MDR may not be appreciated for several weeks because of the time required to identify drugresistant organisms with standard techniques. An important contributing factor is the lack of effective environmental controls for infections due to airborne pathogens in many health care facilities. Even if tuberculosis is recognized, the isolation measures applied may not be effective. Consequently, transmission may occur within the facility. The need for prompt recognition and treatment of tuberculosis caused by MDR organisms presents considerable difficulty. First the causative organism must be recognized as drug resistant. With use of'radiornetrictechniques and direct inoculation ofdrug-containing media for every specimen, resistance patterns could be known in 7-10 days. However, this is an inefficient approach and could quickly overload a laboratory. A more efficient approach would be to screen for resistance to one agent, probably rifampin, and infer multiple-drug resistance if there is resistance to that drug. If this technique were applied only to specimens in which acid-fast
Downloaded from http://cid.oxfordjournals.org/ at Lancaster University on October 24, 2014
seroconversion (whichever is longer) be the minimum duration of treatment for HIV-infected patients [I6, 51). Current recommendations state that for adult patients with HIV infection, the treatment regimen for tuberculosis should include administration of isoniazid (300 mg/d), rifampin (600 mg/d, or 450 mg/d for persons weighing
AIDS COMMENTARY
Impact of Human Immunodeficiency Virus Infection on the Epidemiology, Clinical Features, Management, and Control of Tuberculosis Philip C. Hopewell
From the University of California, San Francisco, and the Division of Pulmonary and Critical Care Medicine, San Francisco General Hospital. San Francisco. California
-Merle A. Sande
Although the association of tuberculosis with AIDS was described in early reports of the syndrome [1-3], this important disorder received little public or scientific attention until recently [4]. Now, because of both the increasing number of cases and the occurrence of outbreaks of tuberculosis caused by multiple-drug-resistant (MDR) organisms [5], funding agencies, investigators, public health officials, and others who heretofore had paid little attention to tuberculosis are hastily attempting to make up for the lack of attention and resources directed toward the disease [4]. The epidemic of human immunodeficiency virus (HIV) infection has exposed the fundamental weaknesses of local programs for control-
Received 15 June 1992. Reprints or correspondence: Dr. Philip C. Hopewell. Professor of Medicine. Box 0841. University of California. San Francisco. 1001 Potrero Avenue, San Francisco. California 94110.
Clinical Infectious Diseases 1992;15:540-7 © 1992 by The University of Chicago. All rights reserved.
1058-4838/92/1503-0021$02.00
ling tuberculosis. Consequently, many communities are not equipped to confront the increasing number of cases or the shifting epidemiological and clinical features of the disease [6, 7]. Moreover, because of the inattention of the scientific community to tuberculosis (a surprising circumstance in light of its causing 2.5-3 million deaths per year worldwide before the HIV epidemic), there are no new tools with which to meet the increasing challenge [4]. The importance of tuberculosis in the context of HIV infection relates to at least four factors. The first of these is the high prevalence of tuberculosis in certain HIV-infected groups. Second, tuberculosis is probably the only HIV-related infection that is transmitted from person to person regardless of whether the exposed person is infected with HIV. Third, if tuberculosis is diagnosed promptly and treated appropriately, it is highly likely that it will be cured [8]. Finally, there is early evidence that tuberculosis can be prevented in HIV-infected populations [9]. This review presents an overview of the impact that HIV infection is having on tuberculosis and on its control primar-
Downloaded from http://cid.oxfordjournals.org/ at Lancaster University on October 24, 2014
After years of decreasing prevalence and increasing hope that tuberculosis, like smallpox, could be eliminated, the disease has resurfaced as a major public health problem in the United States. Particularly ominous are the appearance of multiple-drug-resistant strains and their impact on patients and health care workers who are infected with the human immunodeficiency virus, among whom mortality rates reach 80% 2-3 months postdiagnosis. To respond effectively to this new threat, it is critical that we reorient our thinking about tuberculosis and redirect health care resources to programs for tuberculosis control. We need to reinstitute screening of high-risk populations and ensure proper isolation of patients with the disease. Diagnosing tuberculosis at the earliest possible stage is obviously of the utmost importance. High priority must be given to the development of rapid diagnostic tests and techniques that screen for drug resistance. We must implement and adequately fund drug-discovery programs to develop new therapeutic agents that are effective against multiple-drug-resistant strains of Mycobacterium tuberculosis. Effective programs for monitoring the treatment of patients with tuberculosis must also be implemented. Failure to adequately support such programs has probably led to the recent upswing in multiple-drug-resistant tuberculosis, especially in large cities along the eastern seaboard. Leadership for and funding of these programs must come from the federal government, specifically the U.S. Department of Health & Human Services and the Centers for Disease Control (Atlanta). The Infectious Diseases Society of America is actively supporting a variety of tuberculosis control-related initiatives and will keep its members updated on progress in this area. We plan to publish a series of commentaries focusing on tuberculosis during the coming year. We invited Dr. Philip C. Hopewell to initiate this series with an AIDS Commentary that provides an overview of the critical interaction between infection with human immunodeficiency virus and tuberculosis. Dr. Hopewell is a professor of medicine at the University of California, San Francisco, and head of the Division of Pulmonary and Critical Care Medicine at San Francisco General Hospital.
CIO 1992; 15 (September)
HIV Infection and Tuberculosis
ily in the United States. Subsequent reviews will describe other specific aspects of the interaction between tuberculosis and HIV infection.
Pathogenesis of Tuberculosis in HIV-Infected Persons
losis developed tuberculosis within 120 days [13]. By the use of RFLP analysis it was substantiated that, in all 11 of the culture-positive patients, tuberculosis was caused by the same strain of M. tuberculosis. This pathogenetic sequence is undoubtedly responsible for the explosive outbreaks of tuberculosis caused by MDR organisms that have recently occurred [5]. Presumably because of the pathogenicity of M. tuberculosis, tuberculosis tends to occur relatively early in the course of HIV infection. This is attested to by the finding ofTheuer and associates [15] that HIV-seropositive patients with tuberculosis had a median circulating CD4 lymphocyte count of 354/f.tL. Among this group of 17 patients, tuberculosis was the initial manifestation of HIV infection in all but two patients; one of these patients had limited Kaposi's sarcoma, and the other had oral candidiasis.
Prevalence of Tuberculosis Among Patients with HIV Infection Matching of tuberculosis and AIDS registries in 43 states has revealed that 3.8% of AIDS cases appeared on tuberculosis case registries [16]. However, this percentage is widely variable in different states and areas. In Florida 10% of patients with AIDS have or have had tuberculosis, and in Connecticut and New York 5% have had the disease. In a few states, no cases in patients with AIDS have been reported. In Africa, although there are no precise data defining the incidence of tuberculosis in HIV-infected populations, it is clear that tuberculosis is the most common HIV-associated lung disease. In Dar Es Salaam, Tanzania, 51% of HI V-infected patients who were evaluated because of lung disease were found to have pulmonary tuberculosis [17]. These data confirm numerous anecdotal reports of the frequency with which tuberculosis is diagnosed as an HIV-associated disease in Africa. The frequency with which tuberculosis develops in persons who are infected with both HIV and M. tuberculosis is not well established but appears to be substantially in excess of what would be expected in a non-HIV-infected group. Selwyn and co-workers [ 18] performed a prospective study of tuberculosis among HI V-seropositive and -seronegative iv drug users in New York City. Of 49 HIV-seropositive subjects, seven drug users (14%) who were known to have had a positive tuberculin test previously and one anergic patient developed tuberculosis in a 2-year period. This was the equivalent of 7.9 cases per 100 person-years ofobservation. These observations suggest that seven or perhaps all eight patients who developed tuberculosis had had preexisting tuberculous infection, thus indicating that endogenous reactivation was the dominant if not the only pathogenetic mechanism for development of tuberculosis. Of additional concern was the observation that I I% and 13% of the seropositive and seronegative subjects, respectively, developed positive reactions to tuberculin tests during the study. If this observation is
Downloaded from http://cid.oxfordjournals.org/ at Lancaster University on October 24, 2014
There are two pathogenetic sequences by which tuberculosis can develop: (1) direct progression from recently acquired infection to disease and (2) recrudescence of previously acquired, latent infection. It is generally thought that in areas in which the prevalence of tuberculosis is low, most cases arise from latent infections because few new infections are occurring [10]. However, because HIV impairs the mechanisms by which new tuberculous infection is contained, the risk of direct progression is much greater in the presence of HIV infection; thus, an increasing number of cases may be occurring via this sequence. In addition, as the number of new cases of tuberculosis providing more sources ofinfection increases, there will be more transmission of infection and, thus, more opportunity for direct progression. Because of this shift in the pathogenesis of tuberculosis, the epidemiology of tuberculosis in some parts of the United States resembles that in developing countries. The mechanisms by which a contained tuberculous infection is kept quiescent are not well understood but clearly involve cell-mediated immunity [11]. Because of the effect HIV infection has on cell-mediated immunity, the likelihood of reactivation oflatent tuberculous infection leading to clinical tuberculosis is increased. Thus, persons who had been infected with Mycobacterium tuberculosis before developing HIV infection are at considerably increased risk of developing tuberculosis after being infected with HIV. In the healthy host, once the cell-mediated immune response to infection with M. tuberculosis develops, there is little likelihood of new exogenous infection being acquired [10]. However, because of the HIV-induced immune defect, an HIV-infected person who has been previously infected with M. tuberculosis may still be vulnerable to new infection. Reinfection has been documented under other circumstances as well [12]. Analysis of preliminary data with use of restriction fragment length polymorphism (RFLP) analysis suggests that reinfection can occur and may account for apparent relapses after successful completion of antituberculosis therapy (P.M. Small, personal communication). It has been speculated that HIV -infected patients are more likely to acquire tuberculous infection when exposed to M. tuberculosis [13]. Although this concept is presently unsubstantiated, it is clear that once an HIV-infected person becomes infected with M. tuberculosis, the infection can progress very rapidly to clinical disease [13, 14]. In situations where groups of HI V-infected persons are exposed to a patient with infectious tuberculosis, explosive outbreaks of tuberculosis may occur. For example, in a residential care facility for HIV-infected persons in San Francisco, 11 of 31 residents (35%) exposed to a person with infectious tubercu-
541
542
Hopewell
CID 1992; 15 (September)
truly indicative of new infections, it suggests that there was a large number of infectious cases within the population. Allen and colleagues [19] prospectively followed a group of urban women of childbearing age in Kigali, Rwanda, and found that the incidence oftuberculosis was ....... 2.5% per year. In comparison with HIV-negative women, the risk ratio for tuberculosis among the HIV-positive women was 22.9. For the HIV-positive women, a positive tuberculin test was associated with a significantly increased risk of tuberculosis, but the risk ratio was only 3; 9 of 17 patients with tuberculosis had negative tuberculin tests, perhaps as a result of anergy.
profound effects have occurred in New York City. Between 1984 and 1991 the case rate of tuberculosis increased from 19.9 to 36 per 100,000 persons, an 81%rise. The increase in case rates was even more dramatic in the central Harlem area, where in 1989 the rate reached 169/100,000 persons, a case rate quite similar to those in eastern and central Africa [6]. In 1979 the rate in central Harlem was 50.9/1 00,000 persons.
Prevalence of HIV Infection Among Patients with Tuberculosis
Tuberculin skin testing and anergy testing. As would be expected, persons with advanced HIV infection commonly have little or no reaction to the tuberculin test. However, in earlier stages of the infection, reactivity may be maintained. The ability to respond to tuberculin is an indicator of the status of cell-mediated immunity, which in turn is an indicator of the stage of HIV infection. Most studies that describe the prevalence of reactive (indurations of > 10 mm) tuberculin tests indicate that ....... 40% of HIV-infected patients with tuberculosis have positive reactions. However, in the study by Theuer and associates [15], 80% of the HIV-seropositive patients with tuberculosis had positive reactions to tuberculin. Presumably, this finding relates to the fact that these patients' HIV disease was in the early stages. Because of the frequency of blunted skin test responses or anergy, it has been recommended that an induration of ~5 mm in reaction to 5 tuberculin units ofpurified protein derivative be regarded as indicative of tuberculous infection in HI V-infected persons [16]. The implications of using 5 mm as the cutoff point for defining tuberculous infection have not been determined. In a recent study Graham and associates [29] found that a 5-mm cutoff for determining tuberculin positivity probably resulted in underestimation of the rates oftuberculous infection, so they recommended a 2-mm cutoff. The practicality of using such a small reaction size to define infection would appear to impose significant limitations on the use of tuberculin skin tests, however [30]. To determine if a negative tuberculin test is the result of immunosuppression or is truly negative, control antigens should be applied. These include antigens made from Candida organisms, mumps virus, and tetanus toxoid, to which most persons with intact cell-mediated immunity will respond. A negative tuberculin reaction in the presence of'positive reactions to one or more of the control antigens can be interpreted as a true negative. Ifthere is no reaction to any of the antigens, it cannot be determined if the negative tuberculin test is a true or false negative [31]. Clinical features of tuberculosis. The clinical manifestations oftuberculosis occurring in patients with HIV infection vary considerably, depending on the severity ofthe immunosuppression [1-3, 15, 32-36]. As noted previously, tuberculosis tends to occur early in the course of HIV infection, presumably because of the virulence of M. tuberculosis. In
Impact of HIV Infection on the Incidence of Tuberculosis The full impact of HIV infection on the epidemiology of tuberculosis has not been defined in the United States or abroad. Inferential information suggests, however, that the influence may be substantial. Data from the CDC indicate that there have been ....... 28,000 cases oftuberculosis in excess of the number that would have been expected in the United States between 1984 and 1990 [28]. Although the excess cases that have occurred since 1984 have not been proven to be attributable to HIV infection, the largest increases in incidence have occurred in the areas with the greatest number of AIDS cases and, therefore, the highest prevalence of HIV infection. Use of countrywide data tends to result in underestimation of the effects of HI V infection on the incidence oftuberculosis in certain parts ofthe United States. Perhaps the most
Downloaded from http://cid.oxfordjournals.org/ at Lancaster University on October 24, 2014
Several prospective studies in public tuberculosis clinics have described the prevalence of HIV infection among patients with tuberculosis. Pitchenik and co-workers [20] reported that 31% of 71 consecutive tuberculosis patients in Miami were infected with HIV. In Seattle, Nolan and associates [21] found that 23% of non-Asian adults with tuberculosis were infected with HIV. Theuer and colleagues [15] reported that among a group studied in San Francisco, 28% of non-Asian patients between 15 and 55 years of age who had tuberculosis were seropositive for HIV. Beginning in 1988, a systematic nationwide sampling to determine the prevalence of HIV infection was undertaken by the Centers for Disease Control (CDC; Atlanta) in 14 urban tuberculosis clinics [22]. The median rate of seropositivity among 4,30 I persons who had or were suspected of having tuberculosis was 3.4%. The rates varied widely among clinics, ranging from 0 to 46%. The highest rate was reported from New York City (46%), followed by Newark, New Jersey (34%), Boston (27%), Miami (24%), and Baltimore (13%). Data from developing countries are scarce but indicate a substantial rate of HIV infection among patients with tuberculosis [23-27]. This finding would be expected given the concomitant high prevalence of tuberculosis infection and infection with HIV in developing countries.
Diagnosis of Tuberculous Infection and Tuberculosis in HIV-Infected Patients
ClD 1992; 15 (September)
HIV Infection and Tuberculosis
atypical findings was noted in the prospective study by Pitchenik and Rubinson [48]. These findings included a higher frequency of adenopathy and diffuse or lower-lobe infiltration and a lower frequency ofcavitation for HIV-seropositive patients than for HIV-seronegative patients. Bacteriologic and histologic examinations. Most reported series indicate that the prevalence of positive sputum smears and cultures in patients with pulmonary tuberculosis is the same in HIV-infected and -noninfected persons [15.20,34]. In some instances, sputum induction or bronchoscopic procedures have been necessary for diagnosing pulmonary tuberculosis. Specimens from any site of abnormality in patients who have or are suspected of having HIV infection should be examined for mycobacteria by means ofsmear and culture. Potential high-yield sources include lymph nodes, bone marrow, urine, and blood. In general, any acid-fast organism identified in any specimen should be regarded as being M. tuberculosis until proven otherwise. Such a policy will result in prompt initiation of appropriate therapy for tuberculosis and evaluation of persons in contact with the infected patient. With standard techniques, detection of growth and identification of the species of mycobacteria require 6-10 weeks. Radiometric culture techniques and DNA probes for identification of M. tuberculosis and Mycobacterium avium complex can shorten this procedure to 7-10 days, thereby adding greatly to the speed and efficiency of control measures for tuberculosis [49]. The use of the polymerase chain reaction for identification of M. tuberculosis will result in much more rapid diagnoses [50]. This technique should be available soon for routine application and will represent a major advance in the control of tuberculosis. In patients with more advanced HIV infection, mycobacterial infection does not produce classic granulomas. However, because tuberculosis tends to occur when HIV disease is less advanced, the ability to form granulomas may be intact. Thus, the finding of granulomas in either tissue sections or cytological preparations from needle aspiration biopsy should be interpreted as being more consistent with tuberculosis than nontuberculous mycobacterial disease.
Treatment of Tuberculosis in HIV-Infected Patients Most reported series of patients with tuberculosis and HIV infection demonstrate a good response to antituberculosis treatment when regimens including isoniazid and rifampin are used [2, 3. 8, 15. 20, 32-35]. However, in the series reported by Sunderam and co-workers [33], there were three patients who did not respond to treatment and had progressive disease. A case report from the same institution described a recurrence of tuberculosis at an extrapulmonary (scrotal) site after apparently successful treatment of pulmonary disease [44]. These reports have led to expressions of concern regarding the adequacy of currently standard 6month therapy and, subsequently, to recommendations that a total of 9 months or 6 months beyond the time of sputum
Downloaded from http://cid.oxfordjournals.org/ at Lancaster University on October 24, 2014
most series of cases the majority of tuberculosis diagnoses have preceded the identification of an AIDS-defining disease. For a substantial number of patients, tuberculosis and AIDS are diagnosed at the same time, and for a smaller number, tuberculosis appears after a diagnosis of AIDS. This is in part a semantic distinction because the surveillance definition of AIDS, revised in 1987, now includes extrapulmonary tuberculosis in an HIV-seropositive person [37]. It seems that the earlier tuberculosis develops, the more usual its clinical presentation, whereas the later it occurs, the more atypical its features. This conclusion is somewhat difficult to support with published information, owing largely to the fact that most series of cases in the literature describe the features of tuberculosis in patients who either have or subsequently develop AIDS. Patients identified by cross-matching AIDS and tuberculosis registries clearly are likely to have advanced HIV disease even if their tuberculosis develops before an AIDS-defining diagnosis is made. Clinical reports have emphasized that tuberculosis in cases of advanced HIV infection is frequently disseminated, has unusual radiographic manifestations. and is associated with nonreactive tuberculin tests. Involvement of the lymph nodes, including intrathoracic adenopathy. has been described frequently. A variety of unusual manifestations have been noted. These include CNS involvement with brain abscesses, tuberculomas, and meningitis [33, 38, 39]; bone disease, including vertebral disease [33, 40]; pericarditis [33, 41]; gastric tuberculosis [42]; tuberculous peritonitis [43]; and scrotal tuberculosis [44]. In addition, M. tuberculosis has been cultured from the blood as well as bone marrow [45-47]. Despite the increased frequency of unusual forms of tuberculosis in persons with HIV infection. several reports have described a predominance of "standard" pulmonary disease [15, 18, 21]. These reports presented series ofcases collected prospectively, in which either HIV antibody was measured in patients with tuberculosis or HIV-seropositive subjects were followed for evidence of the development of tuberculosis. Thus, these patients presumably were less immunocompromised. Radiographic findings. The atypical findings on chest radiographs of HIV-infected patients who have tuberculosis have received considerable emphasis. In retrospective studies, radiographical features that are not regarded as typical for pulmonary tuberculosis have been the norm [34, 48]. Lower-lung-zone or diffuse infiltrations rather than the usual upper-lobe involvement commonly have been observed. Cavitation has been unusual, and intrathoracic adenopathy, an unusual finding in immunocompetent adults with tuberculosis, has been relatively frequent. In the prospective study by Theuer and colleagues [15], the radiographic findings for patients with HIV infection were not distinguishable from those for patients who were HIV seronegative. The findings included typical upper-lobe infiltration, often with cavitation. A greater prevalence of
543
544
Hopewell
Tuberculosis Caused by MDR Organisms Recently, explosive outbreaks of tuberculosis caused by MDR organisms (usually resistant at least to both isoniazid
and rifampin) have been reported [5, 54]. Although these outbreaks have taken place in hospital clinics and correctional institutionsand have predominantly involved HIV-infected patients, health care workers, both HIV seropositive and seronegative, have also been infected. Substantial epidemiological and laboratory (RFLP analysis) data indicate that transmission of resistant M. tuberculosis has taken place in these facilities. These outbreaks have been characterized by high case fatality rates, ranging from 72% to 89%within median periods of 4-16 weeks. Also noteworthy are the relatively high rates of tuberculin test conversions among health care workers exposed to these patients. A combination of at least three major factors have produced these outbreaks. The first is a relatively high prevalence of multiple-drug resistance in the community at large and particularly in groups that in some areas are most likely to be infected with HIV. This prevalence of drug resistance is a predictable outcome of the lack of attention and resources devoted to tuberculosis during at least the last decade and provides a reservoir of MDR organisms that are capable of being transmitted. The second factor is the effect ofHI V on the host response to tuberculous infection. As described, recently acquired infection with M. tuberculosis in an HIV-infected person may progress rapidly to clinical disease that in turn is capable of being transmitted. If the transmitted organism is MDR, all of the secondary infections will be caused by MDR organisms. The third factor relates to the fact that tuberculosis in HIV-infected persons may not be as easily recognizable as that in persons with a normal immune status. For this reason the disease may go undiagnosed, perhaps even in a hospital or clinic, for a relatively prolonged period. During this time, unless adequate infection control measures are applied presumptively, infected patients will be capable of transmitting the infection. Moreover, even if the disease is diagnosed, the fact that the organisms are MDR may not be appreciated for several weeks because of the time required to identify drugresistant organisms with standard techniques. An important contributing factor is the lack of effective environmental controls for infections due to airborne pathogens in many health care facilities. Even if tuberculosis is recognized, the isolation measures applied may not be effective. Consequently, transmission may occur within the facility. The need for prompt recognition and treatment of tuberculosis caused by MDR organisms presents considerable difficulty. First the causative organism must be recognized as drug resistant. With use of'radiornetrictechniques and direct inoculation ofdrug-containing media for every specimen, resistance patterns could be known in 7-10 days. However, this is an inefficient approach and could quickly overload a laboratory. A more efficient approach would be to screen for resistance to one agent, probably rifampin, and infer multiple-drug resistance if there is resistance to that drug. If this technique were applied only to specimens in which acid-fast
Downloaded from http://cid.oxfordjournals.org/ at Lancaster University on October 24, 2014
seroconversion (whichever is longer) be the minimum duration of treatment for HIV-infected patients [I6, 51). Current recommendations state that for adult patients with HIV infection, the treatment regimen for tuberculosis should include administration of isoniazid (300 mg/d), rifampin (600 mg/d, or 450 mg/d for persons weighing
