Incidence of Opportunistic and Other Infections in HIV-Infected Children in the HAART Era FREE

The human immunodeficiency virus (HIV) epidemic has spurred the development of new antiretroviral, immune, and vaccine-based therapies geared to block vertical transmission, prevent disease progression, and prolong the survival of individuals who are HIV positive. HAART has dramatically decreased rates of AIDS-related opportunistic complications and deaths in adults.^1- 5 The Swiss HIV Cohort Study, involving 2410 HIV-infected persons 16 years or older, reported that incidence rates (IRs) for all opportunistic infections combined had rapidly decreased from 15.1 per 100 person-years before therapy to 7.7 in the first 3 months after starting treatment, 2.6 in the following 6 months, and 2.2 between 9 and 15 months later.⁶ Due to advances in the treatment of HIV infection including HAART, earlier diagnosis of infection status in exposed infants and quantitative virology and immunologic monitoring, HIV-infection has become a chronic illness with substantially reduced mortality among children.^7- 10 Immune reconstitution engendered by HAART in HIV-positive children has significantly lowered mortality and frequency of bacterial infections and opportunistic infections, such as Pneumocystis jeroveci and disseminated Mycobacteriumavium complex, compared with the pre-HAART era.¹¹

The Pediatric AIDS Clinical Trials Group (PACTG) has conducted multicenter studies of HIV-positive children since 1988. In 2000, the PACTG reported a meta-analysis of 3331 HIV-positive infants and children who were enrolled in 13 PACTG studies conducted before treatment with HAART had became available to determine the rates of various HIV-associated infectious complications.¹² The meta-analysis sample included 49% girls, 52% blacks, and 27% Hispanics. The children were followed up for a median of 24 months (range, 0.23-104.8 months) in 1 or more studies. Antiretroviral regimens studied in the 13 pre–HAART era studies included single, dual, or triple combinations of nevirapine, zidovudine, didanosine, stavudine, or lamivudine. In that study by Dankner et al,¹¹ 5 infections occurred at event rates of greater than 1.0 per 100 patient-years: serious bacterial infections, herpes zoster, disseminated Mavium complex, Pjeroveci, and tracheobronchial or esophageal candidiasis. Pneumonia and bacteremia were the most common bacterial infections.

Although HAART has dramatically decreased morbidity and mortality in HIV-infected infants, children, and adolescents in the United States, no studies comparing the incidence of opportunistic and other related infections before and during the HAART era have been conducted. The objectives of this study were to estimate the IRs for the first occurrence of 29 targeted opportunistic and other related infections between January 1, 2001, and December 31, 2004, in HIV-infected infants, children, and adolescents followed up in PACTG 219C; to compare IRs in the HAART era to those of the pre-HAART era; and to test for linear trends in IRs to evaluate whether the rates continued to decline or remained stable over time in the HAART era. The 29 opportunistic infection categories were selected based on clinical importance and high frequency in the pre-HAART era or HAART era.

Pediatric AIDS Clinical Trials 219C is an ongoing National Institutes of Health–funded, multicenter, prospective cohort study designed to examine the long-term consequences of HIV infection, treatment effects (including prophylaxis and treatment of opportunistic diseases), and interactions of HIV disease and therapy in US infants, children, and adolescents. Prior to participation, the institutional review board at each site approved the protocol. Written informed consent (and assent, if appropriate) was obtained before enrollment from the parent, legal guardian, or participant if legally able to provide his/her consent. Clinical research was conducted in accordance with guidelines for human experimentation, as specified by the US Department of Health and Human Services and by the participating institutions.

Pre-HAART Versions. Version 1.0 of PACTG 219 was implemented in 1993. The major aim was to longitudinally follow up infants, children, and adolescents who had participated in PACTG clinical trials in order to monitor late consequences of therapy including effects on long-term survival and quality of life.⁶ Version 2.0 was implemented in 1995 to expand long-term follow-up of infected infants who were surviving to older ages and to include HIV-exposed children born to HIV-infected mothers who participated in PACTG perinatal transmission studies. A total of 1993 HIV-positive children 0 to 21 years old were enrolled into Versions 1.0 and 2.0 of the protocol implemented in the period corresponding to the pre-HAART era.

HAART Era Versions. To redefine, further update, and expand the objectives and hypotheses, including evaluating the frequency of opportunistic infections in the HAART era over time, a revision of the protocol, version 3.0 (renamed PACTG 219C) was implemented on September 15, 2000, to expand inclusion criteria to allow enrollment, regardless of coenrollment to other PACTG studies, of all HIV-infected children and adolescents at the participating sites.

In all, study staff at PACTG sites throughout the United States and Puerto Rico directly recruited and enrolled 2767 HIV-infected children and adolescents aged 0 to 21 years from September 15, 2000, until December 31, 2004. Participants included in this current analysis were the aforementioned 2767 entered into the database by August 1, 2005, at which time the database was frozen and data analysis conducted.

Human immunodeficiency virus infection was determined at PACTG-certified laboratories using specific and standard guidelines, techniques, and methods. Participants were considered lost to follow-up if they missed 3 consecutive scheduled visits despite efforts by the site staff to contact them. Study participation ended at 24 years of age. Self-reported race/ethnicity information, provided by the parent, guardian, or participant, were collected to assess generalizability and comparability with other studies, using the following list of categories: White non-Hispanic, black non-Hispanic, Hispanic (regardless of race), Pacific Islander, American Indian or Alaskan Native, and Other.

Study visits were conducted every 3 months. For participants who relocated and were not able to transfer to another PACTG site, had enrolled in hospice care, or had become severely handicapped to the extent that follow-up visits were a hardship, off-site evaluations were allowed if they had completed at least 2 regularly scheduled visits and were older than 12 months. Human immunodeficiency virus–related clinical complications and adverse events associated with antiretroviral therapy and other therapies designed to treat or prevent HIV-associated infections were collected at each scheduled visit using standardized criteria and structured forms. Criteria used for the diagnoses of specific infections were those developed by the Pediatric ACTG for clinical trials (“Pediatric Diagnoses for Determining Study Endpoints,” also known as Appendix 40 through the Frontier Science Foundation Web site http://www.fstrf.org). Infections with an etiology documented by culture or histology were designated as proven; those without documentation were designated as presumptive. Health status was assessed at each visit based on CD4 T-lymphocyte cell percentage (CD4%) and HIV-1 viral load. We measured CD4% because the CD4 cell count is an unreliable measure of functional immunity in infants and children infected with HIV. CD4% is relatively stable in children, whereas CD4 cell counts are high in young infants but slowly decrease with age.¹³

For each of 29 targeted infections, we calculated IRs and exact 95% confidence intervals (CIs) per 100 person-years under a Poisson distribution, both overall and by calendar year, for the 2001-2004 period. Participant-specific exposure time for a particular diagnosis was calculated as the time from study entry until the diagnosis date for those with the event and censored at the date of latest clinic visit or death for those who discontinued or died without experiencing an event, or December 31, 2004, for those still participating in the study at the end of the follow-up period. Incidence rate calculations included only children with at least 1 follow-up visit after study entry. Poisson regression models were used to evaluate trends in IRs over the 2001-2004 period and were summarized via a χ² test for linear trend.

The pre-HAART era IRs used for comparison were obtained from a meta-analysis of opportunistic infections occurring in 3331 HIV-infected children participating in 13 PACTG clinical studies conducted by Dankner et al¹¹ in 2000. The median entry age in the pre-HAART era study was 3.3 years (range, 0.1-20.9 years). The median CD4% was 20.5% (range, 0% to 69%). Forty-nine percent of the study sample were girls; 52%, black; 19%, white; and 27%, Hispanic. The pre-HAART era study provides IRs against which our HAART era IRs can be compared. The pre-HAART era analysis did not include information about HIV viral load because too few of the patients' HIV viral load had been measured. Furthermore, the 29 opportunistic infection categories that we considered were not all evaluated in the pre-HAART era analysis. For the 5 opportunistic infection categories that were evaluated in both eras (Herpes Zoster, disseminated M avium complex, P jeroveci pneumonia, fungal infections, and candidiasis), we conducted the overall analysis described above along with supporting analyses that adjusted for CD4% category (<15%, 15%-24%, ≥25%) and were stratified by CD4% level using a Poisson regression model.

For each opportunistic infection, age was calculated on the date of the initial occurrence of the opportunistic infection. We used available viral load and CD4% values measured at the most recent study visit prior to the initial opportunistic infection. All analyses were conducted using the SAS Statistical Software version 8.0 (SAS Institute Inc, Cary, NC); a 2-sided P<.05 was used to determine statistical significance.

Between September 15, 2000, and December 31, 2004, 2767 HIV-infected children enrolled into PACTG 219C. Fifty-two percent of the participants were girls, 59% were black, and 47% were aged 6 to 12 years at enrollment or rollover from a prior version of the protocol; 90% acquired HIV perinatally (Table 1). Sixty-five percent had a CD4% of 25% or more and 20% had a CD4% between 15% and 24% (median [interquartile quartile range {IQR}], 30%, [22%-37%]). As of December 31, 2004, the median duration of follow-up was 3.4 years (interquartile range [IQR], 1.4-4.0 years), with 2200 participants (79%) still being followed up. Five hundred twelve participants (19%) were lost to follow-up: 263 of them because their sites were closed. In addition, 48 died and 7 had completed the study. All these children were censored as of their last clinic visit or when an opportunistic infection was reported. A similar approach was used for the pre-HAART analysis.

Nearly half of the participants who had been part of the pre-HAART protocol continued to this HAART era. Of those 53% were black; 13%, white; and 32%, Hispanic.

Table 2 shows the proportion of children with a history of each targeted infection prior to entry to PACTG 219C. The most common infections reported by at least 5% of children were oral candidiasis, bacterial pneumonia, dermatophyte infections, varicella, lymphoid interstitial pneumonitis, herpes zoster, bacteremia, and molluscum contagiosum.

Table 3 shows IRs for each of the targeted first-time infections diagnosed in 4 or more participants between January 2001 and December 2004. The most common first-time infections were bacterial pneumonia (2.15 per 100 person-years), herpes zoster (1.11 per 100 person-years), oral candidiasis (0.93 per 100 person-years), and dermatophyte infections (0.88 per 100 person-years). The IRs for the first occurrence of varicella, urinary tract infection, bacteremia, molluscum contagiosum, viral hepatitis, disseminated M avium complex and other nontuberculosis mycobacteria, systemic herpes simplex virus, infectious diarrhea, esophageal or pulmonary candidiasis, Pjeroveci, lymphoid interstitial pneumonitis, systemic fungal infection, cytomegalovirus retinitis, and tuberculosis were all less than 0.50 events per 100 person-years. There was no significant linear trend in incidence over the 4 calendar years for any of these infections.

Overall, 553 first episodes of a specific infection occurred among 395 (14%) of the participants, with each participant contributing between 1 and 8 first-time events; 286 participants contributed only 1 event. The demographics of this group of participants were similar to that of the study population as a whole: 89% were perinatally infected; 63%, black; 27%, Hispanic; and 54%, girls. Their age, CD4%, and HIV viral load at the time of occurrence of each infection are presented in Table 4. The median age at the time of the occurrence of any infection was 12 years (IQR, 8-16 years). Disseminated M avium complex, Pjeroveci, and systemic fungal infections occurred only in perinatally infected children. The most common first-time infections among children younger than 12 years were bacterial pneumonia, dermatophyte infections, and mucocutaneous candidiasis, whereas in older children, bacterial pneumonia, herpes zoster, urinary tract infection, dermatophyte infections, and molluscum contagiosum were most common. At the time of the first infection, 49% of the children had a CD4% of at least 25% (IQR, 12%-34%), 59% had a CD4 cell count of more than 400 cells per mL (IQR, 214-913 cells per mL), and 43% had an HIV viral load lower than 5000 RNA copies/mL. Most episodes of dermatophyte infection (54%), varicella (79%), urinary tract infection (52%), molluscum contagiosum (59%), and viral hepatitis (64%) occurred at a time when the child had a normal percentage of CD4 cells (≥25%). On the other hand, all first episodes of lymphoid interstitial pneumonitis and most episodes of disseminated M avium complex (82%), esophageal or pulmonary candidiasis (84%), systemic herpes simplex (63%), herpes zoster (60%), bacteremia (56%), and candida thrush (73%) occurred in children with reduced CD4% (<25%). Six of the 7 first events of Pjerovici occurred in children with CD4 cell count of less than 15%.

The causes of death for the 48 patients who died during follow-up are shown in Table 5. Mean age at death was 15.0 years, (IQR, 12.5-19.1 years), 22 were girls; 31, black non-Hispanic; 12, Hispanic; and 5, white non-Hispanic.

Figure 1 compares HAART-era IRs with pre–HAART era IRs for the 5 most common serious infections of children infected with HIV.¹¹ The IRs (per 100 person-years) of bacterial pneumonia decreased from 11.1 (95% CI, 10.3-12.0) to 2.15 (95% CI, 1.79-2.56; P<.001), herpes zoster decreased from 2.9 (95% CI, 2.6-3.3) to 1.11 (95% CI, 0.88-1.39; P<.001). Bacteremia decreased from 3.3 (95% CI, 2.9-3.8) to 0.35 (95% CI, 0.22-0.51; P<.001), disseminated M avium complex decreased from 1.8 (95% CI, 1.5-2.1) to 0.14 (95% CI, 0.07-0.25; P<.001), and P jeroveci decreased from 1.3 (95% CI, 1.1-1.6) to 0.09 (95% CI, 0.04-0.19; P<.001).

After accounting for CD4% (ordinal or categories), HAART-era IRs were consistently significantly lower than pre–HAART era rates (P<.001) for all infections shown in Table 6 except for systemic fungal infections, which showed no significant decline, possibly due to the small number of events observed. HAART-era rates also showed statistically significant decreasing linear trends with increasing CD4% level for all 5 infections shown in Table 6 and, in addition, for bacteremia, candida (thrush), and bacterial pneumonia (data not shown). Pre-HAART era rates for these latter 3 infections were not evaluated by CD4% and could not be compared with HAART era rates accounting for CD4%.

The percentage of HIV-infected individuals receiving HAART as of January 1 of each year remained very stable over the period of this study, ranging from 69% in 2001 to 71% in 2004; the majority of HAART use included a protease inhibitor (86% of HAART use, within each year). An additional 3% to 4% reported use of 3 or more nucleoside reverse transcriptase inhibitors, and 13% to 15% reported not taking antiretroviral therapy. The median cumulative duration of prior HAART therapy increased over the period from 2.6 years in 2001 to 4.5 years in 2004. For participants who discontinued the study, 68% were taking HAART, 5% were taking 3 or more nucleoside reverse transcriptase inhibitors, and 13% were not currently taking antiretroviral therapy. Patients who discontinued the study had a median cumulative duration of 3.7 years of HAART therapy at the time of discontinuation.

Data on use of primary or secondary opportunistic infection prophylaxis is shown in Figure 2. Over the period 2001-2004, a statistically significant decreasing linear trend existed for those receiving either prophylaxis against P jeroveci pneumonia (P <.001 for trend), disseminated M avium complex (P <.001 for trend), or fungal infections (P <.05 for trend). The proportion of those receiving prophylaxis against P jeroveci pneumonia decreased from 31% to 11% and from 8% to 3% for those receiving prophylaxis against disseminated Mavium complex, while the proportion of those receiving antifungal prophylaxis decreased from 1.9% to 1.0%.

For those receiving prophylaxis against P jeroveci pneumonia, the most common drugs used included trimethoprim-sulfamethoxazole (79.8%), dapsone (11.6%), atovaquone (5.4%), and pentamidine delivered either via aerosol or intravenously (3.2%). For those receiving prophylaxis against disseminated M avium complex, the most common drugs used included azithromycin (78.8%) followed by clarithromycin (16.6%) and rifabutin (4.7%). Fluconazole was the only drug used for those receiving antifungal chemoprophylaxis.

The results of this study demonstrate a substantial reduction in the incidence of several opportunistic infections in HIV-infected children since the introduction of HAART therapy. As previously reported 80% of the PACTG 219C cohort has been treated with an antiretroviral therapy regimen considered to be highly active, but the proportion is possibly an underestimate because the study did not include children who were receiving antiretroviral therapy through a clinical trial.¹⁴ Although HIV-associated opportunistic infections and other related infections continue to occur in HIV-positive children since the introduction of HAART, most infection occur at rates that are substantially lower than those seen in the pre-HAART era.¹⁵ Increased understanding of the pathogenesis of HIV and associated opportunistic pathogens has led to effective prophylaxis for many of these infections. Although we are unable to directly estimate the individual effect of antiretroviral therapy and prophylaxis, the combined effect of the 2 are associated with a substantial reduction in the incidence of previously common opportunistic infections.

The risk of opportunistic infections decreases in concert with an increase in the CD4 lymphocyte count following initiation of HAART with rates of mycobacterial diseases and cytomegalovirus disease remaining high for the first 3 months after initiating HAART before declining.¹⁶ Likewise, the Swiss cohort analysis reported that the risk of developing an opportunistic infection for a person receiving potent antiretroviral therapy was highest during the initial months of therapy.⁶ In adults, achieving a CD4 cell count of at least 200 × 10⁶/L following initiation of HAART is a valuable marker for a reduction in the risk of subsequent opportunistic infections. In another analysis of 219C, lack of a sustained response to HAART was predictive of opportunistic infection in children.¹⁷ Primary prophylaxis against Pjeroveci, toxoplasmosis, and disseminated Mavium complex infection can be safely discontinued in adults and children following immune reconstitution. PACTG study 1008 demonstrated the safety of discontinuing maintenance therapy following HAART–induced immune reconstitution in HIV-positive children.¹⁸

The incidence of initial opportunistic infection was substantially lower than previously reported by Dankner et al¹¹ from the same sites during the pre-HAART era. The demographics of the pre-HAART participants were similar to that of the HAART-era participants in the current analysis. The pre-HAART analysis study sample comprised 49% girls and 52% black and 27% Hispanic children followed up for a median of 24 months (range, 0.23-104.8 months) in one or more studies. The median age was 3.3 years (range, 0.1-20.9 years), although the median age for the current HAART-era analysis cohort was 10.4 years. The median CD4% in the pre-HAART analysis was 20.5 (range, 0%-69%). Antiretroviral regimens studied in the 13 pre-HAART era analyses included single, dual, or triple combinations of nevirapine, zidovudine, didanosine, stavudine (ddC or d4T), or lamivudine. For the pre-HAART analyses, the authors censored patients in trials at the time of initiating HAART therapy. However, the pre-HAART analysis did not specifically analyze the use of opportunistic infection prophylaxis, which prevents direct comparison of the effect of prophylaxis on the IRs of opportunistic infections.

In the previous report by Dankner et al,¹¹ 5 infections occurred at event rates of greater than 1.0 per 100 patient-years: serious bacterial infections, 15.1; herpes zoster, 2.9; disseminated M avium complex, 1.8; P jeroveci, 1.3; and tracheobronchial or esophageal candidiasis, 1.2 events per 100 person-years. Pneumonia (IR, 11.1) and bacteremia (IR, 3.3) were the most common bacterial infections. Notably, the present analysis included participants from the same source population as those in the analysis conducted by Dankner et al, thus providing substantial overlap and relative comparability of participants between the 2 periods. Children enrolled in the current study were healthier, as measured by their baseline CD4%, than those analyzed in the study by Dankner et al. In addition, the current study population represents approximately 22% of the 9885 children and adolescents younger than 19 years living with HIV/AIDS in the United States during 2003,¹⁹ which increases the generalizability of our findings.

In a meta-analysis of some very early ACTG studies conducted before the advent of HAART, Finkelstein et al¹⁶ reported that the baseline CD4 cell count and the immunologic and virologic response to treatment were strong predictors of disease progression and that the occurrence of one opportunistic infection increased the risk of subsequent opportunistic infections, even after adjusting for the CD4 count. For example, the occurrence of Pjeroveci significantly increased the risk of subsequent disseminated Mavium complex and cytomegalovirus, disseminated Mavium complex increased the risk of subsequent cytomegalovirus, and cytomegalovirus increased the risk of disseminated Mavium complex. Individuals with CD4 cell counts less than 50 × 10⁶/L remain at increased risk of infection and require close clinical observation. However, there were too few recurrent events in our database to evaluate the risk of subsequent occurrence of opportunistic infection among participants who experienced the first event, but studies are underway to evaluate the incidence of opportunistic infections in children who reported a history of an infection at entry to PACTG 219C.

In our current study, the actual power for detecting a significant linear trend in yearly IRs for any given infection was low due to the overall rarity of targeted opportunistic infections. Because we did not observe significant linear trends in the annual incidence of any of the 29 infections during 2001, 2002, 2003, and 2004, we assessed our statistical power to detect IR decreases. We estimated that these data provided 70.4% power at a significance level of .05 to detect a decrease in IR from 2.0 to 1.0 per 100 person-years assuming person-years at risk of 1500, 1750, 1800, 1850 during 2001, 2002, 2003, and 2004, respectively. Another limitation is that we did not examine risk factors for occurrence of opportunistic infections in relation to HAART use. A detailed risk-factor analysis among patients in this cohort with perinatally acquired HIV-infection is reported elsewhere.¹⁷ Furthermore, we did not analyze the potential impact of the use of the more immunogenic pneumococcal conjugate vaccine on rates of bacteremia and pneumonia despite its use in the general population and availability for children infected with HIV during the study observation period.

Despite these current advances due to HAART, some HIV-infected children continue to develop opportunistic infections.²⁰ Some children fail to respond to antiretroviral therapy as a result of viral resistance, poor adherence, or inability to tolerate complex treatment regimens. Furthermore, prophylactic therapies are not fully effective and poor adherence can further reduce their efficacy. Drug interactions, complex dosing schedules, adverse effects, and high costs can further limit the efficacy of these therapies. Although these issues do present challenges, our findings demonstrate a substantial reduction in the incidence of several opportunistic infections in HIV-infected children since the introduction of HAART therapy.