INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

HIV infection increases the risk for pneumococcal pneumonia, as well as for other respiratory pathogens (1). Community-based studies performed between 1983 and 1994 in San Francisco, New Haven, and Ohio have estimated the rate of pneumococcal bacteremia to be 9.4 to 10.5 cases/1,000 person-years in HIV-seropositive populations compared with a community rate of 0.07 to 0.5 cases/1,000 person-years (4). HIV-seropositive intravenous drug users (IDU) are at an approximate fourfold greater risk of pneumococcal pneumonia than are their HIV-seropositive non-drug-using counterparts (8). These data suggest that both IDU and HIV seropositivity are independent risk factors for pneumococcal infections. However, the mechanisms that explain this increased susceptibility in HIV-infected IDU are unknown.

Our study focused on differences in -PLY titers between HIV-infected IDU developing pneumococcal bacteremia and control subjects remaining free of pneumococcal disease as there is in vivo evidence that PLY, the major Streptococcus pneumoniae cytotoxin, contributes to the early pathogenesis of pneumococcal pneumonia by facilitating intrapulmonary bacterial growth and invasion into the bloodstream (9). Furthermore, studies in children and elderly populations have demonstrated that the humoral immune response to PLY is protective against invasive pneumococcal disease (13, 14). Lastly, vaccination of animals with PLY protects against invasive pneumococcal infection (15).

To date, the protective role of -PLY in HIV-seropositive IDU populations has been inadequately characterized. A single study on the role of -PLY in HIV-seropositive persons demonstrated that the baseline geometric mean titers of -PLY in HIV-seropositive U.S. gay males and Kenyan female sex-workers were significantly lower than those in their HIV- seronegative control counterparts (16). The investigators were unable to document a significant difference in baseline -PLY titers between HIV-seropositive persons developing bacteremia from those free of disease.

Our study addressed the association between humoral immunity and the acquisition of pneumococcal bacteremia. We hypothesized that after controlling for CD4 cell count and time postenrollment, those developing severe pneumococcal infections have decreased baseline antibody levels against PLY compared with those remaining free of pneumococcal disease. We further postulated that within any category of CD4 cell count, there is a dampened B-cell mediated response to PLY in those developing pneumococcal bacteremia.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Design

The study was a matched case-control study that used participants from the NIDA/NIH-sponsored IDU cohort in the Baltimore area (ALIVE I and ALIVE II) (17). The study utilized prospectively collected sera and clinical data from 28 HIV-infected IDU who had a microbiologically confirmed incident episode of severe pneumococcal infection and 56 HIV-seropositive and 28 HIV-seronegative IDU remaining free of pneumococcal disease. We quantified the humoral immune response to the pneumococcal virulence factor pneumolysin before and after pneumococcal infection by the ELISA technique. Antihemolysin activity assessed the functional activity of the antibody response to pneumolysin.

To address whether there is a dampened antibody response to pneumococcal PLY in cases who develop pneumococcal bacteremia, we measured the change in antibody titers of -PLY postinfection from preinfection titers in 23 cases with available sera 3 to 9 mo postpneumococcal bacteremia. We used time matched HIV-seronegative IDU control subjects and time- and CD4-matched HIV-seropositive IDU control subjects to define the natural background decline of -PLY titer.

Study Population

The 112 study participants were members of the ALIVE Study Cohort who were observed every 6 mo for at least 12 mo in the ALIVE Study Clinic between July 1989 and December 1996. The procedures and methods for monitoring this cohort have been previously described (17). The visits consist of completion of a verbally administered standardized questionnaire on complications of IDU and drug use patterns, and acquisition of serum for measures of immunologic function and disease progression. Unused serum is stored in a study repository at 70° C. We retrospectively utilized this sera to measure differences in antipneumolysin antibody titers and function in members of this cohort. The follow-up rate has been > 92% through the first 16 visits.

To confirm the microbiologic diagnosis and to characterize the risk factors for bacterial pneumonia, the ALIVE study has routinely obtained hospital records, death certificates, and clinic records for persons reporting an episode of pneumonia or bacteremia on the ALIVE initial and follow-up questionnaires. Charts were extracted by trained study nurses using a standard pneumonia/ bacteremia study form. The protocols were approved by the institutional review board for human studies, and written informed consent was obtained from participants.

Case and Control Selection

Case selection. The cases (n = 28) were survivors of a microbiologically confirmed incident episode of pneumococcal bacteremia (n = 24) or severe pneumonia (n = 4) who had questionnaire data and sera available within 9 mo prior to bacteremia, and 3 to 9 mo postbacteremia. Although all cases are microbiologically confirmed S. pneumoniae infections, in cases of severe pneumococcal pneumonia with presumed sepsis, the diagnosis of severe pneumonia is defined by American Thoracic Society criteria (18). Cases were excluded if they had known bronchiectasis, severe emphysema, use of prednisone, chemotherapeutic agents, or pneumococcal vaccination within 1 yr prior to developing bacteremia. In our study, nine potential participants were excluded because of missing sera.

Control selection. Each case was matched with two HIV-seropositive and one HIV-seronegative control subjects who remained free of pneumococcal disease from inception into the cohort until December 1996. The HIV-seropositive and HIV-seronegative control subjects were matched to the cases on date of study entry ± 3 mo and length of follow-up. The HIV-seropositive control subjects were further matched by duration of follow-up post-HIV seroconversion and CD4 cell count within 50 cells/mm³ of case CD4 count at visit prior to bacteremia. The same exclusion criteria that were applied to cases were also applied to control subjects. From the pool of eligible HIV-seropositive and HIV-seronegative control subjects, a random sample was chosen.

Measurements

The determination of demographic and clinical risk factors for developing severe pneumococcal infection utilized the responses from the semiannual questionnaire at the visit preceding severe infection.

Serologic measurements. -PLY were measured within 9 mo prior to infection and 3 to 9 mo postinfection by ELISA using the method previously described (19). The only modification to this protocol was pooled human -PLY used as a positive control. A range of human sera was tested for the presence of antipneumolysin antibodies. Sera containing high titers of antibody were pooled and the immunoglobulin (Ig) purified by affinity chromatography on protein A sepharose. A 1 mg/ml stock of human IgG was prepared and stored for standardization of ELISA assays. A standard curve was generated using 10, 5, 2.5, 1, 0.5, and 0.25 µg of antibody per well. Unknowns were compared with this curve, and results were expressed as arbitrary units of -PLY.

Antihemolytic activity of serum was measured using the method described for monoclonal antibodies, except that the first well in serial dilutions contained a 1/2 dilution of the appropriate serum sample (20). Hemolysin levels are expressed as the inverse ratio of antihemolysin titers.

Statistical Analysis

The data were analyzed using the statistical package STATA (version 5.0; College Station, TX). For the comparison of differences in demographic risk factors between cases and control subjects, we utilized a conditional logistic regression model with the primary outcome incident pneumococcal sepsis, and the covariates of primary interest were sex, prophylactic antibiotic use, smoking status, and injection drug use. The data were analyzed as a matched analysis, matching for length of follow-up and CD4 cell count; -PLY titers were log-transformed to more closely approximate a normal distribution. Two-tailed t tests with equal variance were used to assess the differences in mean geometric mean titer (GMT) -PLY before and after disease in cases and control subjects. Wilcoxon's rank sum statistic was used to determine differences in median antihemolysin titer between cases and control subjects.

Sample size. The sample size allowed for a 93% power to detect a 30% difference in GMT -PLY prior to disease between cases and control subjects with an alpha of 0.05, assuming a baseline titer of 70 units in control subjects (21). The sample size provided 86% power to detect a 30% difference between initial and follow-up -PLY titer in cases using a one-sample test of mean.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

There were 37 incident cases of pneumococcal bacteremia (n = 29) or severe pneumococcal pneumonia (n = 8) documented in the members of the ALIVE cohort between 1989 and 1996. We performed a matched case-control study to determine whether clinical or demographic factors differentiated between members of the cohort developing severe pneumococcal disease. Each case was matched to three HIV-seropositive control subjects (matched on calendar time and CD4 cell count) and three HIV-seronegative control subjects (matched on calendar time) remaining free of microbiologically confirmed pneumococcal disease. This analysis demonstrated that cases were similar to control subjects in age, race and sex, intravenous drug use, and smoking practices (Table 1). Cases did not differ from control subjects in type or frequency of comorbid diseases, including diabetes mellitus, renal disease, hepatic disease or tuberculosis. HIV-infected control subjects did not differ from the cases in prophylactic antibiotic use, antiretroviral therapy, or neutrophil counts. The median CD4 cell count in cases was 289.5 (range, 13 to 1,403 cells/mm³). The 28 cases with available sera and the 84 serologic control subjects did not differ significantly from the larger case-control group in clinical or demographic variables analyzed. A lack of significant differences in clinical and demographic factors suggests that the study of differences in humoral immunity to known virulence factors of S. pneumoniae may offer possible explanations for the increased susceptibility to pneumococcal infection in this high-risk population.

Comparison of Baseline Antipneumolysin Antibody Titers

Our study demonstrated that the pre-disease -PLY titers were not significantly different between cases and control subjects (Table 2). Furthermore, the -PLY titers were similar between HIV-seropositive cases and the HIV-seropositive control subjects. Lastly, there was no significant difference in the baseline GMT -PLY based on CD4 cell category. In the 44 HIV-seropositive participants with CD4 cell counts < 300 cells/mm³, GMT -PLY was 73 ± 28 compared with a GMT -PLY of 67 ± 28 units in the 40 participants with CD4 cell counts > 300 cells/mm³, p = 0.81. 

Functional Activity of Baseline Antipneumolysin Antibodies

The study found a wide distribution of titers of antihemolysin in cases (range = 0 to 10,240) and control subjects (range = 0 to 10,240) over a wide range of -PLY titers. There was a lack of correlation between the baseline antihemolysin activity and the baseline -PLY titer in both cases and control subjects (Pearson's correlation, r² = 0.25). The median neutralization titer of antihemolysin was greater in HIV-seropositive cases (1,280) than in HIV-seronegative control subjects (160) (Table 2). However, the median antihemolysin titer was similar in HIV- seropositive cases (1,280) to that in the HIV-seropositive control subjects (640) (Table 2).

Change in Antipneumolysin Antibody Titer in Cases after Pneumococcal Bacteremia

The case sera were collected a median of 99 d (range, 34 to 156 d) after the bacteremic episode. Forty-eight percent (11 of 23) of the cases had an increase in -PLY antibody titer postdisease compared with 9% (two of 23) of the seronegative control subjects and 46% (21 of 46) of the HIV-seropositive control subjects during the same interval. In those with an increase in -PLY titer, the magnitude of the antibody response was greater in the 11 cases (mean increase, 48 PLY units; median increase, 20 PLY units) than in the 21 HIV-seropositive control subjects (mean increase, 31 PLY units; median increase, 11 PLY units) using a two-tailed unpaired t test (p = 0.06). However, only four of the 23 cases had a doubling of -PLY titer postinfection. The changes in -PLY titer and function between baseline and follow-up measurements in cases and control subjects are summarized in Table 3. The cases had a significantly greater fold-rise in -PLY titer (1.18) than did the HIV-seronegative control subjects who remained free of pneumococcal disease (fold-rise, 0.66). However, the fold-rise in -PLY titer in the cases did not differ significantly from that in the HIV-seropositive control subjects (fold-rise, 1.05). Furthermore, the median antihemolysin titer did not differ between cases and the HIV-seropositive control subjects or the HIV-seronegative control subjects.

Relationship between CD4 Level Prior to Infection and Postdisease Antipneumolysin Antibody Levels

Univariate analysis demonstrated that the CD4 cell count of the cases prior to infection was associated with the immune response to pneumolysin postinfection. Of the 11 cases with an increase in -PLY titer after infection, nine of 11 had a CD4 cell count > 300 at the visit prior to infection (median CD4 cell count, 477) and two of 11 had a CD4 cell count less than 300 at the visit prior to infection (median CD4 cell count, 217; two-tailed unpaired t test, p = 0.01). However, the magnitude of the fold-rise of -PLY titer in those with infection was not significantly different between those with CD4 cell counts > 300 (fold-rise, 1.60) and those with < 300 (fold-rise, 1.09) (p = 0.18).

Persistence of Antipneumolysin Antibody in the Cases

We measured -PLY titers in 14 cases with sera available 14 to 18 mo postinfection. The sera for measure of persistence were collected between 6 and 12 mo after the postinfection sera. There was no difference in timing of sera collection after the postdisease titer between those with and those without a fold-rise in -PLY titers postinfection. In the seven members with an increase in antibody titer postdisease, the -PLY titers at 14 to 18 mo postinfection were 91 ± 8% of the postdisease titer. In contrast, in the seven subjects with a decrease in antibody titer postdisease, the mean -PLY titers at 14 to 18 mo postinfection were only 74 ± 10% of the postdisease titer (p = 0.41).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study has demonstrated that the predisease -PLY titer and function did not differ significantly between HIV-seropositive IDU developing pneumococcal bacteremia from either HIV-seropositive IDU or HIV-seronegative IDU who remained free of disease. Our data suggest that the predisease -PLY titers do not wane with progression of HIV-infection. Furthermore, only four of 23 cases experienced a doubling of baseline -PLY titer postinfection. However, participants with CD4 cell counts greater than 300 cells/mm³ were more likely to have a fold-rise in -PLY titer than were participants with CD4 cell counts less than 300 cells/mm³. Lastly, those with an increased titer of -PLY after pneumococcal infection showed a persistently elevated response at 14 to 18 mo postinfection.

However, our study had several limitations. First, there were varying intervals of time between predisease and postdisease sera collection and the episode of infection. This difference in interval prior to disease is of greatest concern as rapid declines in immune function after the sera collection may have resulted in lower -PLY titers at the time of infection than those measured. Second, our measure of late convalescent -PLY titers may not have been a good predictor of response as the humoral response to pneumolysin may be transient. Third, -PLY titers increase in response to subclinical disease or asymptomatic colonization. We did not measure nasopharyngeal colonization rates in the study subjects; however, prior studies have demonstrated no significant difference in nasopharyngeal carriage of pneumococcus between HIV-seropositive and HIV-seronegative patients (2, 22). Lastly, restriction of the study to cases with severe pneumococcal infection could have led to an underestimate of response to PLY post-disease as those with milder disease may have had greater ability to mount an effective immune response.

Unlike the previous study by Amdahl and colleagues (16) of -PLY titers in HIV-infected gay males and African case sex-workers, -PLY titers in HIV-seropositive IDU were not different in those developing pneumococcal disease from those in the HIV-seronegative control subjects remaining free of documented disease. However, the longer interval between sera collection and disease in the previous study status may have accounted for the differences in results. The prior study had a mean of 16 mo between sera collection and disease compared with a mean of 5 mo in our study. Alternatively, we speculate that the high frequency of active IDU in both cases and control subjects could have decreased pre-disease -PLY titers biasing our results towards the null. Heroin and cocaine decrease humoral immunity in in vitro models (23, 24). Our study was similar to the study by Amdahl and colleagues in its inability to document a significant difference between GMT -PLY prior to disease in HIV-seropositive cases when compared with HIV-seropositive control subjects.

Our data demonstrate that the pre-disease -PLY titer is not correlated with the CD4 cell count of the case at the visit prior to disease. There is a wide range of -PLY titers at each category of CD4 level in both cases and control subjects, which suggests that multiple abnormalities in immune function are responsible for the increased susceptibility to pneumococcal disease.

The lack of correlation between disease state and anti-hemolysin titer may lead to one of two conclusions. Either the antihemolytic activity of serum is not the only important biologic effect of pneumolysin or the spectrum of epitopes to which antibodies are made is different in cases with HIV. Also pneumolysin itself can inhibit antibody synthesis by lymphocytes, which may affect the type of antibody generated in the different case-control groups. Antibodies to pneumolysin are important in conjunction with antibodies to other antigens. As -PLY are not opsonic, they mediate their effect by an antitoxic mechanism and by allowing development of opsonic antibodies to other antigens. HIV-infection is associated with a dampened -CPS response, which may diminish the importance of -PLY levels in HIV-infected persons (25). Furthermore, animal models do not show that antihemolytic titer reflects the level of protection.

The humoral immune response to pneumolysin postinfection was predicated on both the predisease CD4 cell count and the rate of decline of CD4 cell count between study intervals. The cases with CD4 cell counts > 300 cells/mm³ had a significantly greater proportion of members with an increase in -PLY titer postinfection than did members with CD4 cell counts < 300 cells/mm³. However, the actual mean fold-rise postinfection was less than 50% greater than baseline in cases with an increase in -PLY titer postinfection. This reduced IgG1 antibody response to pneumolysin in HIV-seropositive IDU is consistent with pneumococcal vaccine studies in HIV-seropositive IDU populations that document low levels of IgG₂ antibody response to pneumococcal capsular polysaccharides in persons with CD4 cell counts below 500 cells/mm³ (22). Our results emphasize the need to immunize this at-risk population at an early stage even if employing conjugate vaccines containing multiple immunogens from S. pneumoniae to augment protection. This study underscores the need to further characterize the immune dysregulation in HIV-seropositive IDU, particularly in the era of highly active antiretroviral therapy.