Prevention strategies rely heavily on the use of tuberculin purified protein derivative (PPD) to identify persons harboring M. tuberculosis [14]. Anergy, a consequence of HIV infection, undermines these strategies in persons at the highest risk for tuberculosis infection and subsequent active disease [15-18]. A negative PPD test result in this setting could be attributable to a true lack of exposure to tuberculosis or simply to the incapacity of the patient to manifest an appropriate cell-mediated immune response. To reduce the measured prevalence of anergy and thereby increase the proportion of tuberculin nonreactors who can be considered "truly" PPD negative, the Centers for Disease Control and Prevention (CDC) has recommended the additional use of at least two delayed-type hypersensitivity control antigens (mumps antigen plus Candida antigen or tetanus toxoid) when screening HIV-infected patients. Thus, persons from populations with a prevalence of tuberculous infection of 10% or more and who are tuberculin negative but not anergic may be spared preventive therapy with isoniazid [19]. However, the ability of control antigens to predict the likelihood that a negative PPD test result is "truly" negative in this highly anergic population is unknown.

To improve approaches to tuberculosis prophylaxis, more data are needed about the relations among delayed-type hypersensitivity responsiveness, the prevalence of tuberculosis, and the waning immunity associated with progressive HIV infection. In an ongoing multicenter study of the natural history of the pulmonary complications associated with HIV infection, we have been examining these factors prospectively in a cohort of 1353 persons in 6 U.S. geographic areas. Recently, we evaluated baseline delayed-type hypersensitivity responses in this cohort of HIV-seropositive and HIV-seronegative persons and identified variables associated with tuberculin reactivity and anergy.

Methods

Top Methods Results Discussion Author & Article Info References

Patients and Study Design

The Pulmonary Complications of HIV Infection Study is a multicenter study designed to prospectively describe the frequency, types, and effect of pulmonary complications in HIV-infected persons, both before and after the development of AIDS. All diagnoses, treatments, and outcomes are recorded and monitored in a common database.

Because our purpose was to evaluate longitudinally both the early and late pulmonary manifestations of HIV infection, each center attempted to recruit about 170 HIV-seropositive participants, half with CD4 lymphocyte counts of 400 cells/mm³ or more and no HIV-related symptoms and half with fewer than 400 CD4 cells/mm³ or symptomatic HIV-infection (defined by a temperature of 38 °C or more for at least 2 weeks, involuntary weight loss of 10% or more from baseline, diarrhea of at least a 1-month duration, oral candidiasis, or oral hairy leukoplakia). Within each group, participants were drawn from one of three HIV transmission categories (homosexual men, male and female intravenous drug users, and women with heterosexually acquired HIV infection) to reflect their approximate distribution at each clinical site. About 30 HIV-seronegative homosexual men and intravenous drug users were also recruited at each site to serve as controls. Participants had to be willing and able to comply with the protocol and were required to give informed consent. The study was reviewed and approved by the institutional review board at each site. Exclusion criteria included Centers for Disease Control and Prevention (CDC)-defined AIDS [20], severe non-HIV-related disease likely to affect survival, lung disorders likely to interfere with the required evaluations, acute pulmonary processes, immunosuppressive therapies within the previous 6 months, and treatment for active tuberculosis within the past 12 months. From November 1988 through February 1990, we enrolled 1353 persons in the study, of whom 1171 were HIV seropositive and 182 were HIV seronegative. Human immunodeficiency virus serologic status was confirmed at study entry using a licensed enzyme-linked immunosorbent assay and a Western blot assay. Further baseline evaluation included a complete medical history, a physical examination, hematologic and biochemical studies, T-lymphocyte subset analysis, delayed-type hypersensitivity testing, a chest roentgenogram, and pulmonary function measurements.

Measurement of Delayed-Type Hypersensitivity Response

We tested delayed-type hypersensitivity with the following antigens: mumps antigen (Connaught Laboratories, Inc., Swiftwater, Pennsylvania); Dermatophytin "0" (Candida) at 1:100 dilution (Hollister-Stier, Spokane, Washington); Dermatophytin (trichophytin) at 1:100 dilution (Hollister-Stier); and tuberculin PPD at a strength of 5 tuberculin units per 0.1-mL dose (Connaught Laboratories, Ltd., Willowdale, Ontario, Canada). Tests were administered by intradermal injection of 0.1 mL of antigen (Mantoux method) and read by a trained observer 48 to 72 hours after application in most participants (the interval exceeded 4 days in 18 persons). Response was recorded as the greatest diameter of induration. We used the current standard operational criteria for a positive response: induration of at least 5 mm for all antigens except PPD, for which an induration of 10 mm was required among HIV-uninfected persons [19]. Anergy was defined as 0 mm of induration for all delayed-type hypersensitivity antigens administered in a given panel. Unless otherwise specified, a test battery of tuberculin PPD, mumps antigen, and Candida antigen was used to define anergy.

At one site, however, the investigators did not distinguish between induration and erythema for the mumps, Candida, and trichophytin tests, documenting reactions to these antigens in terms of millimeters of erythema. When examined by zone diameter, their measurements were generally consistent with those from the other centers. Furthermore, the results of multivariate analyses with and without the data from this site were similar. Other investigators have shown a high degree of correlation between induration and erythema with these antigens [21]. Thus, for the purposes of our analysis, responses were recorded in millimeters of induration.

Because trichophytin elicited a positive reaction in only 14.0% of those tested, it was dropped from the delayed-type hypersensitivity battery midway through the enrollment period. Although lot numbers varied, the skin tests used at the centers were supplied by the same manufacturers, with a single exception: At one site, investigators used a different Candida preparation. The Candida test results for this center (247 participants) were excluded from all analyses involving this antigen.

Determination of Lymphocyte Subsets

Lymphocyte subsets were determined for CD3, CD4, and CD8 receptor-bearing cells by the same laboratory at each site. All laboratories participated in the flow cytometry quality control program sponsored by the National Institute of Allergy and Infectious Diseases [22].

Statistical Analysis

All analyses are based on data collected during the baseline evaluation. Statistical significance for comparisons of proportions was determined by chi-square or Fisher exact test [23]. For comparisons among nonindependent groups, repeated-measures analysis for categorical outcomes was used to determine statistical significance [24, 25]. Logistic regression models were used to study the relation between PPD positivity or anergy and potential risk factors [26]. Risk factors considered were HIV status; CD4 count among HIV-seropositive persons; intravenous drug use; race or ethnicity; a history of a positive PPD test result, tuberculosis, or BCG vaccination; age; gender; and socioeconomic status. Seventy-seven participants, including women with heterosexually acquired infection and persons who were not white, black, or Hispanic, were excluded from all multivariate analyses because of small sample sizes. Initial models included HIV status (positive or negative), intravenous drug use (presence or absence), race or ethnicity (white, black, or Hispanic), a history of a positive PPD test result, and age, as well as interaction terms, to determine whether the effect of HIV positivity varied among these groups or whether the effect of drug use varied by race. No statistically significant interactions were observed. All odds ratios presented were derived from subsequent models containing main effects only. A dichotomous variable indicating the 12% of the cohort who did not have a high school diploma was used as an index of socioeconomic status and was included in all final models.

All tests were two sided. A P value of 0.05 was considered statistically significant. Ninety-five percent CIs are given when appropriate.

Results

Top Methods Results Discussion Author & Article Info References

Patient Characteristics

During the 16-month enrollment period, 1171 HIV-seropositive and 182 HIV-seronegative persons entered the study. The two groups were similar with regard to age, sex, race, transmission category, and tuberculosis-associated history (Table 1). Of the 1165 men, 966 (82.9%) were homosexual; of the 188 women, 132 (70.2%) were intravenous drug users. As expected, baseline CD4 lymphocyte counts were higher in persons without HIV infection than in patients with HIV infection (median count, 889 cells/mm³ compared with 410 cells/mm³). A tuberculin PPD test was done and read in 91.2% of participants, a mumps antigen test in 92.5%, and a Candida antigen test in 91.9% (1001 of 1100 participants from five sites). Both the tuberculin PPD test and the mumps antigen test were done and read in 90.5% of participants, and the tuberculin, mumps, and Candida tests were done and read in 88.5% of participants. Intravenous drug users were the least compliant with testing; 76.3% underwent testing with all three antigens and returned for reading compared with 93.1% of all other cohort members. Within each transmission category, compliance was similar among HIV-seronegative controls, HIV-seropositive patients with 400 CD4 cells/mm³ or more, and HIV-seropositive patients with fewer than 400 CD4 cells/mm³.

Delayed-Type Hypersensitivity Response by HIV Serologic Status and Transmission Category

A positive delayed-type hypersensitivity response at baseline was most frequently observed with mumps antigen (38.9%), followed by Candida antigen (28.3%), trichophytin (14.0%), and tuberculin PPD (6.1%). Delayed-type hypersensitivity responses were more frequent among HIV-seronegative controls than among HIV-seropositive patients: 10% compared with 5.5%, (P = 0.03) for tuberculin PPD; 69.5% compared with 34.3% (P < 0.001) for mumps antigen; and 22.2% compared with 12.7% (P = 0.009) for trichophytin. The prevalence of the response to Candida antigen was similar in HIV-seronegative and HIV-seropositive persons (29.9% compared with 28.0%, P > 0.2). Delayed-type hypersensitivity responses are summarized by HIV status and transmission category in Table 2. The prevalence of PPD reactivity in HIV-seronegative homosexual men was almost three times that observed in HIV-seropositive homosexual men (6.8% compared with 2.5%, P = 0.02). Among intravenous drug users, PPD reactivity was similar for HIV-seronegative and HIV-seropositive persons (19.1% compared with 15.1%, P > 0.2). Tuberculin PPD reactivity was 11% among women with heterosexually acquired HIV infection. Intravenous drug users had a higher prevalence of tuberculin PPD positivity than homosexual men among both HIV-seropositive patients (15.1% compared with 2.5%, P < 0.001) and HIV-seronegative controls (19.1% compared with 6.8%, P = 0.03). In each risk group, reactivity to mumps antigen was greater among HIV-seronegative controls than among HIV-seropositive patients. The prevalence of reactivity was higher for mumps antigen than for Candida antigen in both HIV-seronegative controls (69.5% compared with 29.9%, P < 0.001 by chi-square McNemar test) and HIV-seropositive patients (34.3% compared with 28.0%, P = 0.04 by chi-square McNemar test). Homosexual men had a higher prevalence of response to mumps antigen than did intravenous drug users among participants without HIV infection (75.8% compared with 52.3%, P = 0.05) but not among participants with HIV infection (34.8% compared with 33.3%, P > 0.2). Anergy to the battery of tuberculin PPD, mumps antigen, and Candida antigen was similar across transmission groups among HIV-infected patients. Among HIV-seronegative persons, the prevalence of anergy was higher in intravenous drug users than in homosexual men (44.1% compared with 15.6%, P = 0.001).

The relation between delayed-type hypersensitivity reactivity and CD4 lymphocyte count is shown in Figures 1 and 2. Because delayed-type hypersensitivity response did not vary in relation to CD4 count among HIV-seronegative controls, these participants were considered a single "immunologic" group in all comparisons. Figure 1 shows the effects of varying the cutoff for a positive reaction to PPD and mumps antigen in the combined homosexual male and intravenous drug user groups. Regardless of which diameter of induration was used ( 10 mm, 5 mm, or >0 mm), the prevalence of reactivity to PPD (see Figure 1, top) declined as the CD4 level decreased (chi-square for trend, P < 0.001 for each induration size). The prevalence of tuberculin PPD reactivity increased as the cutoff for a positive reaction was reduced (P = 0.002).

View larger version (38K): [in this window] [in a new window]   Figure 1. Effect of varying the induration cutoff for test positivity on delayed-type hypersensitivity response rates. The top panel shows the prevalence of a positive tuberculin purified protein derivative test result based on an induration diameter of 0, 5, or 10 mm. The bottom panel shows the prevalence of a positive mumps antigen reaction based on an induration diameter of 0, 5, or 10 mm. Women with heterosexually acquired human immunodeficiency virus (HIV) were excluded from the analysis. Participants were stratified by CD4 cell count as follows: 0 to 199 cells/mm3 (200 participants); 200 to 399 cells/mm3 (298 participants); 400 to 599 cells/mm3 (282 participants); and 600 cells/mm3 or more (232 participants). The study included 1012 HIV-positive patients and 160 HIV-negative participants. Prevalences were directly adjusted for differences in the distribution of homosexual men and intravenous drug users within each group. PPD = purified protein derivative.

View larger version (39K): [in this window] [in a new window]   Figure 2. Relation between delayed-type hypersensitivity response rates and immune function in homosexual men, intravenous drug users, and the total cohort. The top panel shows the prevalence of a positive tuberculin PPD test result ( 10 mm of induration for HIV-negative controls and 5 mm of induration for HIV-positive patients). The bottom panel shows the prevalence of anergy (0 mm of induration for PPD, mumps antigen, and Candida antigen tests). Sample size is indicated above each bar. HIV = human immunodeficiency virus; HIV = homosexual men; IDU = intravenous drug users.

The reduction of the cutoff point for PPD response from 10 mm to 5 mm in HIV-seropositive patients increased overall reactivity by 1%. Further reduction of the zone diameter to the presence of any induration (>0 mm) accounted for another 0.9% increase in prevalence. The prevalence of PPD positivity among HIV-infected patients with 400 CD4 cells/mm³ or more based on the 0-mm cutoff approximated the prevalence of 9.9% in participants without HIV infection based on the conventional 10-mm criterion. Although the number of reactors with CD4 counts of fewer than 200 cells/mm³ was too small to draw any firm conclusions, the prevalence of PPD positivity was the same, regardless of the cutoff used for a positive test. These relations are shown for mumps antigen testing in Figure 1, bottom. A decreased prevalence of response to mumps antigen was found as the CD4 level decreased (chi-square for trend, P < 0.001 for each induration size), whereas an increased response was found as the zone-size criterion was decreased (P < 0.001). Additionally, even among HIV-seropositive participants with 600 CD4 cells/mm³ or more, a positive response to mumps antigen occurred less frequently than among HIV-seronegative participants (51.6% compared with 69.8% for an induration cutoff of 5 mm, P < 0.001).

The prevalence of delayed-type hypersensitivity response by CD4 level is shown for homosexual men and intravenous drug users in Figure 2. Standard definitions for a positive tuberculin test were used ( 10 mm of induration for HIV-seronegative controls and 5 mm of induration for HIV-seropositive patients). For the total study population, tuberculin PPD reactivity was lower among HIV-seropositive persons with CD4 levels of fewer than 400 cells/mm³ than among HIV-seronegative persons (P < 0.001). However, the prevalence of tuberculin reactivity in HIV-infected participants with 400 or more CD4 cells/mm³ was similar to that in persons without HIV infection (P > 0.2), among both homosexual men (3.7% compared with 6.8%, P = 0.2) and intravenous drug users (22.0% compared with 19.1%, P > 0.2). Among HIV-infected homosexual men, the prevalence of a positive tuberculin PPD test result decreased as CD4 level decreased, from a high of 3.9% in patients with a CD4 count of at least 600 cells/mm³ to a low of 0.6% in those with fewer than 200 CD4 cells/mm³ (chi-square for trend, P = 0.03). Changes among HIV-infected intravenous drug users were similar, declining by a factor of 10, from 24.0% in patients with 600 CD4 cells/mm³ or more to 2.4% in those with fewer than 200 CD4 cells/mm³ (chi-square for trend, P < 0.001).

Reduction in the prevalence of PPD reactivity was accompanied by a corresponding increase in anergy. Anergy was more prevalent in HIV-seropositive than in HIV-seronegative participants overall (P < 0.001) (see Figure 2, bottom). In HIV-seropositive patients, the prevalence of anergy increased as the CD4 count decreased; the prevalence was 25.5% in persons with at least 600 CD4 lymphocytes/mm³ and 72.0% in those with fewer than 200 CD4 lymphocytes/mm³ (chi-square for trend, P < 0.001). Sequential trends were similar for both seropositive homosexual men and intravenous drug users. At CD4 counts below 600 cells/mm³, the prevalence of anergy among HIV-infected homosexual men was 50.1% compared with 15.6% among homosexual men without HIV infection (P < 0.001). However, the prevalence of anergy among HIV-seropositive intravenous drug users with more than 200 CD4 cells/mm³ was similar to that among HIV-seronegative intravenous drug users (40.0% compared with 44.1%, P > 0.2).

Multiple Antigens and the Prevalence of Reactivity

A subset of 716 participants (demographically and clinically representative of the entire study sample) received all four skin tests and returned for the measurement of zone diameter. This provided the opportunity to examine the effect of the sequential addition of antigens on the prevalence of reactivity (Figure 3). Multiple antigens elicited more responses than any antigen alone. Among HIV-seropositive patients, 40.3% reacted (>0 mm) either to tuberculin PPD or mumps antigen. When Candida antigen and then Trichophyton antigen were successively added to the delayed-type hypersensitivity battery, the number of responders increased to 55.7% and 59.4%, respectively. Among HIV-seronegative controls, 73.1% responded to tuberculin PPD or mumps antigen; 75.3% reacted to tuberculin PPD, mumps antigen, or Candida antigen; and 77.4% responded to one or more antigens when all four were considered in the interpretation of anergy. With the addition of Candida antigen, the relative increase in reactivity was greater among HIV-seropositive than among HIV-seronegative persons (25.8% compared with 8.2%, P = 0.05); with the addition of Trichophyton antigen, the relative increase was similar in HIV-seropositive and HIV-seronegative participants (8.4% compared with 8.5%, P > 0.2).

View larger version (67K): [in this window] [in a new window]   Figure 3. Multiple antigens and the prevalence of reactivity. A subset of 716 participants had all four skin tests. Each set of three bars shows the distribution of tuberculin purified protein derivative (PPD) responses in relation to those for successive batteries of delayed-type hypersensitivity control antigens in human immunodeficiency virus (HIV)-seropositive patients (n = 623) and HIV-seronegative controls (n = 93). The height of each defined segment within a bar indicates the prevalence of each response (expressed as a cumulative percentage). The number associated with each segment indicates the number of persons who had that response. A PPd-positive response and a PPd-negative response were defined as induration 5 mm and <5 mm, respectively, for HIV-seropositive patients and induration 10 mm and <10 mm for HIV-seronegative controls. Induration greater than 0 mm defined a positive reaction to a control antigen. In the battery including all four antigens, PPD positivity was the only response in 21 participants (2.9%); mumps antigen positivity was the only response in 115 (16.1%); Candida antigen positivity was the only response in 78 (10.9%); and trichophytin positivity was the only response in 25 (3.5%). MP = mumps; Cd = Candida; and Tr = trichophytin.

Tuberculin PPD Positivity and Response to Control Antigens

The association between tuberculin PPD positivity ( 5 mm for HIV-seropositive participants and 10 mm for HIV-seronegative participants) and reactivity (>0 mm) to control antigens was also studied in this same subset of 716 subjects. Regardless of which delayed-type hypersensitivity control panel was used, no association was observed between the tuberculin PPD response and reactivity to control antigens (see Figure 3). For example, the prevalence of a positive tuberculin PPD response was 6.2% (20 of 324) for HIV-seropositive participants who reacted to either mumps or Candida antigen and 7.0% (21 of 299) for those who showed no response (0 mm) to either antigen (difference, 0.8%; 95% CI, –3.0% to 5.0%; P > 0.2). Similarly, among HIV-seronegative participants, the prevalence of PPD positivity did not differ between responders (5 of 67) and nonresponders (3 of 26) to mumps or Candida antigen (7.5% compared with 11.5%; difference, 4% [95% CI, –12.0% to 20.0%]; P > 0.2), although the number of tuberculin PPD responders was small. These relations were similar in all HIV transmission groups.

Predictors of Tuberculin PPD Reactivity

Multivariate analysis was used to assess the ability of various factors to predict tuberculin PPD reactivity and anergy. Unadjusted odds ratios are presented with those adjusted for the effects of the other variables shown. Age was treated as a continuous variable in both analyses. Although unadjusted prevalences are discussed below, all odds ratios shown are adjusted. Because of the somewhat high prevalences in most groups, the odds ratios exceed the prevalence ratios.

Tuberculin PPD reactivity Table 3 was more common in participants with a history of a positive PPD test result (38.0%) than among those without such a history (3.5%; odds ratio, 12.5). Similarly, among 36 participants with a history of BCG vaccination, the prevalence of tuberculin PPD positivity was 17.7% compared with 5.4% among those who reported no history of vaccination (odds ratio, 5.9). The odds of a positive reaction were higher in intravenous drug users than in homosexual men (prevalence, 15.8% compared with 2.9%; odds ratio, 3.3); the odds of a positive reaction were also higher in blacks than in whites (prevalence, 15.2% compared with 3.1%; odds ratio, 2.7). The other variables in this model partially accounted for the strong univariate associations between black race and intravenous drug use (for example, 84% of the homosexual men were white, whereas 69% of the intravenous drug users were black). Hispanics showed similar levels of tuberculin PPD positivity when compared with whites (odds ratio, 1.4; CI, 0.4 to 4.8). Although the crude analysis revealed a strong inverse association between socioeconomic status and PPD positivity, the adjusted odds ratio for persons not completing high school relative to high school graduates showed no such association (odds ratio, 0.92; CI, 0.4 to 2.0). The prevalence of response to tuberculin PPD increased with increasing age.

Seropositivity for HIV and low CD4 counts were negatively associated with PPD reactivity. Compared with HIV-seronegative controls, HIV-seropositive persons with CD4 cell counts below 400/mm³ had reduced odds of a positive PPD reaction. Although the prevalence of tuberculin PPD positivity was reduced among HIV-seropositive persons with 400 CD4 lymphocytes/mm³ or more compared with HIV-seronegative controls, this reduction was not statistically significant. This relation, however, became significant at all CD4 levels when induration of 5 mm was used to define a positive reaction in both HIV-seronegative and HIV-seropositive persons and at all levels below 600 CD4 cells/mm³ when induration of 0 mm or 10 mm was used. No gender differences in tuberculin reactivity were detected.

Predictors of Anergy

Factors associated with anergy were evaluated in 910 participants. Anergy was more common among HIV-seropositive patients than among HIV-seronegative controls for all CD4 levels below 600 cells/mm³. Above this level, HIV-seropositive patients were no more likely to be anergic than were controls. Patients with HIV infection who had fewer than 200 CD4 cells/mm³ were most likely to be anergic (odds ratio, 9.0; CI, 5.2 to 15.4). The odds of being anergic were reduced in persons who had a history of a positive tuberculin PPD test result relative to those who had no such history (prevalence, 28.6% compared with 42.5%; odds ratio, 0.4 [CI, 0.2 to 0.8]). Blacks and Hispanics had similar odds of being anergic when compared with whites (odds ratios, 1.2 [CI, 0.8 to 1.8] and 1.1 [CI, 0.6 to 1.9], respectively). Overall, intravenous drug users were no more likely to be anergic than were homosexual men (odds ratio, 1.3; CI, 0.9 to 2.1). Neither gender, history of BCG vaccination, age, nor socioeconomic status was predictive of anergy.

Discussion

Top Methods Results Discussion Author & Article Info References

We have provided a cross-sectional view of delayed-type hypersensitivity responsiveness in HIV-infected patients and HIV-seronegative controls. Because of the study sample's diversity, our analysis was able to include several predictors of delayed-type hypersensitivity responsiveness that cannot be evaluated in samples limited to a single HIV transmission category, gender, or racial group. The baseline analysis of our relatively large cohort revealed independent associations of HIV serologic status, CD4 level (among HIV-seropositive patients), transmission category, race, previous tuberculin PPD positivity, a history of BCG vaccination, and age with the prevalence of a positive tuberculin PPD reaction. The strongest predictor of anergy was HIV serologic status. Previous tuberculin PPD positivity and CD4 level (among HIV-seropositive patients) had lesser effects, whereas transmission category, race, and age were not significant predictors of anergy. Finally, neither gender nor socioeconomic status was associated with the prevalence of either tuberculin positivity or anergy.

The increasing prevalence of tuberculin positivity with advancing age seen in our study has been reported by others [18, 27]. This finding may be due to a birth cohort effect. Other surveys have also indicated that the risks for tuberculin PPD positivity and active tuberculosis depend on racial, ethnic, and socioeconomic factors [8, 27]. Although we failed to detect an effect of socioeconomic status on tuberculin response, our index, a complete high school education (absent in only 12% of the cohort), may have lacked the power to detect statistically significant differences.

In general, the prevalence of a delayed-type hypersensitivity response was lower in HIV-seropositive patients, even those with 600 CD4 cells/mm³ or more, than in HIV-seronegative controls, although these differences were statistically significant only at CD4 counts below 400 cells/mm³. When the presence of any induration (>0 mm) was used to define test positivity, the prevalence of tuberculin PPD reactivity among HIV-seropositive patients with at least 400 CD4 cells/mm³ more closely approximated that observed in HIV-seronegative controls evaluated by standard criteria ( 10 mm). It is unknown whether reduction of the tuberculin cutoff point among HIV-seropositive patients actually succeeds in correctly identifying those harboring M. tuberculosis [18]; much of the specificity of the test may, in fact, be lost [27-29]. Using the criterion of "any induration" for a positive tuberculin test in HIV-seropositive patients resulted in the identification of only nine additional positive reactors in our cohort. Our results resemble those from Johnson and colleagues' study [30] of 2042 Haitians; induration in response to PPD testing was 10 mm or more in most patients and less than 5 mm in relatively few patients, regardless of HIV serologic status. Comparison of the results of our study with those of other surveys is confounded by the use of different delayed-type hypersensitivity tests, criteria for interpretation, and population bases. However, investigators generally agree that HIV infection is associated with a 2- to 10-fold decrease in the prevalence of a positive tuberculin PPD test result, depending on such factors as the clinical status of the patients [16, 18, 19, 31-33]. Several studies have shown that among patients with active tuberculosis, the prevalence of PPD positivity in patients with HIV infection was half that observed in seronegative controls [4, 16, 18, 34]. Among various antigens evaluated in delayed-type hypersensitivity panels, mumps antigen gave the highest frequency of positive responses [15, 18, 19]. In 1120 homosexual and bisexual men, Sears and colleagues [15] found that the best single predictor of anergy was HIV serologic status. These investigators also showed that the prevalence of anergy was higher in patients with CD4 counts of 400 cells/mm³ or less and that the prevalence was reduced by the addition of more antigens to the delayed-type hypersensitivity test battery. Data also suggest that some patients with advanced disease, even those with CD4 counts below 50 cells/mm³, may respond to delayed-type hypersensitivity testing, including the tuberculin PPD test [1, 18, 20]. Similarly, our study shows that 17.3% (18 of 104) of HIV-infected patients with fewer than 100 CD4 cells/mm³ had a positive reaction to at least one antigen, although none had a positive tuberculin PPD test result. The 44.1% prevalence of anergy among HIV-seronegative intravenous drug users (see Table 2) observed in our study is about six times higher than that recently reported by Graham and colleagues [18], despite the intradermal administration of the same control antigens supplied by the same manufacturers and our use of a more restrictive definition of anergy. Other than active drug use, our study participants had no additional known risk for anergy. We therefore cannot explain this large difference in anergy prevalence. Because these 34 anergic persons represented only a small proportion of our total study sample, they had little effect on the overall analysis.

Earlier studies suggested that many HIV-infected patients who develop tuberculosis have a positive PPD test result months to years before the onset of tuberculous disease [4, 34, 35]. Furthermore, most cases of tuberculosis develop at a time when tuberculin PPD testing is more sensitive, before rather than after the development of AIDS. Although the potential to develop tuberculosis rapidly after exposure has been documented in outbreak settings, most cases of HIV-associated tuberculosis are believed to be caused by reactivation of latent infection [3, 6, 9, 11-13]. By the time AIDS-defining opportunistic infection develops, the pool of potential tuberculosis reactivators may be largely depleted. Most of these patients will be anergic. The risk for tuberculosis among anergic patients may be considerable. In a sample of HIV-infected intravenous drug users in New York City, the prevalence of active tuberculosis among anergic patients was similar to that among PPd-positive patients, although the total number of cases was small and the contribution of primary tuberculosis to this total was unknown [36]. Nevertheless, strategies targeting anergic persons for prophylactic therapy may result in the treatment of disproportionately large numbers of people to prevent few cases. Because isoniazid poses a considerable management problem in patients with advanced HIV infection, strategies for tuberculosis prevention should seek to minimize unnecessary use of the drug. One such approach is to better assess the patient's overall capacity to manifest a delayed-type hypersensitivity response through the application of multiple control antigens [19]. For example, based on our data, the addition of Candida antigen to a delayed-type hypersensitivity test battery of PPD and mumps antigen increased the number of reactors from 40.3% to 55.7% among HIV-seropositive patients. Thus, if PPd-negative patients who react to control antigens can be considered to have "true-negative" results, more HIV-positive patients may be spared isoniazid therapy if immune response is assessed using both Candida and mumps antigen rather than mumps antigen alone. This strategy is rendered less effective by the phenomenon of selective anergy to tuberculin PPD, which may be seen in many cases of active tuberculosis [37-39]. Moreover, our data showed no association between the response to control antigens and tuberculin reactivity, unlike those reported by Palmer and Reed [21] in a sample with considerably higher prevalences of antigen reactivity. Regardless of HIV serologic status, our participants were just as likely to be PPD positive, regardless of their response to control antigens. The importance of this observation is unclear, and further studies are needed to define the rate of active tuberculosis among the various categories of delayed-type hypersensitivity responders. Some investigators have, in fact, questioned the need for tuberculin PPD and anergy testing in persons with HIV infection and have proposed extending prophylaxis to members of groups in which the prevalence of tuberculosis is 10% or more [40].

Baseline skin testing in our cohort detected the largest prevalence of PPD positivity among intravenous drug users as well as a diminished delayed-type hypersensitivity response to all antigens in all HIV-infected groups. Among persons with HIV infection, both tuberculin PPD positivity and anergy depended highly on CD4 cell count, with significantly lower reactivity being observed in patients with CD4 counts of fewer than 400 cells/mm³. In addition, our findings also support the combined use of at least the mumps and Candida antigens to reduce the proportion of HIV-seropositive persons considered to be anergic, although this approach needs further elucidation [41]. Therefore, until predictors of tuberculosis are more accurately defined, better methods to detect latent infection with M. tuberculosis are developed, or less toxic preventive therapies are identified, it is essential that tuberculin screening be done as early as possible in the course of HIV infection and repeated periodically, even in patients with low CD4 counts.

Appendix: The Pulmonary Complications of HIV Infection Study Group

Participating clinical centers included the University of California, San Francisco, California (Philip C. Hopewell, John Stansell, Joan Turner, and Dennis Osmond); Northwestern University, Chicago, Illinois (Jeffrey Glassroth, Melinda Mossar, and Robert Hirschtick); Beth Israel Medical Center, New York, New York (Mark J. Rosen, Lori Meiselman, Kim K. Manghisi, Christopher Cardozo, and Thomas H. Kalb); University of Medicine and Dentistry of New Jersey, New Brunswick, New Jersey (Lee B. Reichman, Bonita T. Mangura, Claudia Hanson, Fran Dowell, and Margaret O'Toole); University of California, Los Angeles, Los Angeles, California (Jeanne M. Wallace, Bert Shapiro, Barbara LeMaire, Barbara Richer, Janet Au, and Anne Coulson); Henry Ford Hospital, Detroit, Michigan (Paul A. Kvale, Norman Markowitz, Louis Saravolatz, Christine Johnson, Joanne Huitsing, and Annmarie Krystoforski).

The data coordinating center was Research Triangle Institute, Research Triangle Park, North Carolina (W. Kenneth Poole, A. Vijaya Rao, Kim Clayton, Nellie I. Hansen, Matthew C. Jordan, Jim Thompson, David Myers, Lisa LaVange, Judith Katzin, William Fulkerson, Timothy Wilcosky, Yu Lou, and Steven Game).

National Heart, Lung, and Blood Institute, Bethesda, Maryland (Anthony R. Kalica, Janet Wittes, Dean A. Follmann, and Robert Wise [Consultant to NHLBI from Johns Hopkins University, Baltimore, Maryland]).

The policy and safety monitoring board included Reuben Cherniack (Chair), John G. Bartlett, John E. Connett, Ronald P. Daniele, John F. French, Dixie E. Snider, Jr., and Gerard M. Turino.