Incidence of Noninfectious Conditions in Perinatally HIV-Infected Children and Adolescents 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 HIV-infected individuals. Highly active antiretroviral therapy (HAART) has dramatically decreased the rates of AIDS-related conditions and deaths.^1- 8 With the advent of life-prolonging HAART, HIV infection has become a chronic disease and is no longer the certain death sentence in children that it was in the pre-HAART era.

While HAART has dramatically decreased morbidity and mortality in HIV-infected infants, children, and adolescents in areas of the world with ready access to HAART,^4,6- 8 to our knowledge, no studies that estimate the incidence of noninfectious conditions of HIV during the HAART era have been conducted. Clinical trials for antiretroviral, immune, and vaccine-based therapies often evaluate only the short-term efficacy, safety, and toxicity of the various therapeutic modalities. The potential long-term outcomes (toxic effects and conditions) from the use of these agents cannot be adequately assessed given the short time frames of most clinical trials. The long-term consequences in youth with perinatal HIV infection related to HIV itself, immunosuppression, or drug toxic effects will require ongoing surveillance as the individuals age into adulthood.^9- 16

Our objective for this analysis was to estimate HAART-era incident rates (IRs) for the first episode of noninfectious conditions in HIV-infected infants, children, and adolescents followed up at Pediatric AIDS Clinical Trials Group (PACTG) sites to identify HAART-era changes in the natural history of HIV infection in infants, children, and adolescents.

PACTG protocol 219C was an ongoing National Institutes of Health–funded, observational, multicenter prospective cohort study designed to examine the long-term consequences related to HIV disease progression, treatment effects (including prophylaxis and treatment of opportunistic infections), and/or interactions of HIV disease and therapy in US infants, children, and adolescents. PACTG 219C opened in January 2000 and closed to follow-up in May 2007. Study eligibility criteria included age younger than 24 years and documentation of HIV infection or perinatal HIV exposure (including both HIV-infected and HIV-exposed uninfected children). An earlier version of the study, PACTG 219, enrolled children between 1993 and 2000 who had participated in any perinatal or treatment trial conducted by the PACTG.

Prior to site registration, the institutional review board at each participating site approved the protocol. Informed consent (and assent, if appropriate) was obtained for each patient before enrollment, and clinical research was conducted in accordance with guidelines for human experimentation, as specified by the US Department of Health and Human Services and by each participating institution. More than 80 sites in the United States including Puerto Rico enrolled and followed up children and adolescents enrolled in this comprehensive long-term follow-up study.

For this analysis, subjects were required to be perinatally HIV infected and to have enrolled prior to December 31, 2006. Two separate positive peripheral blood specimens from different days were used to confirm diagnosis of HIV infection. HIV DNA or HIV RNA viral load samples were used in children younger than 18 months to confirm HIV status, and either HIV DNA or RNA or HIV enzyme-linked immunosorbent assay with a confirmatory Western blot were used in children older than 18 months. No treatment was administered as part of this study, and there was no randomization or stratification of subjects. Subjects were declared lost to follow-up if they missed several consecutive scheduled visits despite efforts by the site to contact them. Study participation ended for subjects who reached 24 years of age or died during the study.

HIV-infected children were followed up through quarterly visits. Visits included interim history and physical examinations as well as routine specified blood testing or medical record abstraction, which were used to obtain CD4 cell percentage and count and HIV RNA results. At study entry, a lifetime history of antiretroviral treatment was obtained, and changes in and adherence to antiretroviral treatment were collected at each study visit, as described previously.^17- 18 If a patient moved, all efforts were made to ensure follow-up at another participating PACTG site. Starting in 2002, patients who moved and were not followed up at a PACTG site were permitted to have off-site clinical data entered into the 219C database. In the event of death, the death report was collected and results of an autopsy were obtained if one was performed.

HIV-related clinical conditions, adverse events associated with antiretroviral therapy, and other therapies designed to treat or prevent HIV-associated opportunistic infections were collected at each scheduled visit using standardized criteria on structured forms. Specific presumed or proven diagnoses were reported (rather than signs or symptoms) whenever possible. A list of coded diagnoses was used as the primary source for reporting diagnoses (https://www.fstrf.org/).

When reporting a diagnosis for the first time, it was noted on the case report form that the diagnosis was new. At subsequent visits, diagnoses continued to be reported as ongoing until they were resolved. Study criteria required that source documentation be available to support each reported diagnosis.

The noninfectious conditions listed in Table 1 were evaluated for each of the calendar years from 2001 to 2006 and overall for this cohort of 2575 perinatally HIV-infected children. These included cardiac conditions (acute myocardial infarction [MI] [MI or electrocardiogram and laboratory findings consistent with MI], cardiomyopathy [right or left ventricular dimensions >2 SD from mean for body surface area], or cardiac abnormality); encephalopathy (HIV encephalopathy, non-HIV encephalopathy, static encephalopathy [monophasic insult to the central nervous system causing dysfunction, cerebral palsy, or toxic encephalopathy], or other encephalopathy); gynecological abnormalities (including vaginal, vulvar, cervical, or anogenital atypia, dysplasia, or intraepithelial neoplasia, all grades); pancreatitis conditions (related to infection or medication with elevated enzymes [eg, amylase, lipase] specific to the pancreas); pregnancy-associated conditions (including intrauterine fetal demise or spontaneous miscarriage, ectopic pregnancy, chorioamnionitis, preeclampsia, intrauterine growth retardation, and premature delivery); renal disorders (including nephropathy, intermittent or persistent proteinuria, acute and chronic renal failure); and other noninfectious conditions (including peripheral neuropathy, birth defects, idiopathic thrombocytic purpura, non-AIDS–defining cancers, epilepsy, microcephaly, and substance abuse).

We calculated IRs for the first episode of each of the earlier-mentioned conditions. Incidence rates per 100 person-years were calculated for each condition under a Poisson distribution along with exact 95% confidence intervals (CIs). Incidence rates were calculated overall and by calendar year and evaluated for linear trends over time using Poisson regression models. Except as noted later, children with no history of a specific condition prior to study entry or prior to 2001 and with at least 1 follow-up visit after entry were considered at risk for a first event of that condition. A child could experience multiple first events in different condition categories. Exposure time for the first event was censored as of the first diagnosis date or, for those without an event, the last clinic visit date on or prior to December 31, 2006. Patients continued to be followed up for the development of other conditions for which they remained at risk.

Defining the subset of participants and appropriate time at risk required special considerations for certain noninfectious conditions. The risk set for pregnancies included all girls who reached 13 years of age by the end of follow-up (no girls reported pregnancies prior to that age in our study). Only girls who had ever reported a pregnancy were considered in the risk set for pregnancy conditions, and for these girls, time at risk began with their 13th birthday. The risk set for gynecological dysplasias/neoplasias included only females with at least 1 documented Papanicolaou (Pap) test. For this condition, time in the study prior to the earlier of the first Pap test or 13th birthday was not considered in calculating IRs, but for girls with a Pap test prior to study entry, all such gynecological abnormalities reported on Pap test after entry were included in the IR analysis, even if during 2000. Thus, for gynecological dysplasias, unlike most other conditions, some events in 2000 may have been considered incident events. Proportions of subjects with historical events (as defined earlier) within each condition were calculated along with exact CIs based on the binomial distribution. Data were frozen on February 6, 2007, and follow-up was censored at December 31, 2006.

Our analysis of 2575 perinatally HIV-infected children followed up from 2001 through 2006 indicated that the enrolled subjects were relatively healthy, with two-thirds of the group having a CD4 cell percentage higher than 25% and more than two-thirds of the group having viral loads less than 10 000 copies/mL at the time of entry. Seventy-five percent were receiving HAART, 15% were taking antiretroviral drugs but not receiving HAART, and 10% were not taking any antiretroviral drugs. Over the time studied, there were only 65 deaths in the perinatally HIV-infected cohort, of which 28 were not associated with opportunistic infections. Fifteen types of noninfectious diagnoses were tracked over the 6 years of the study.

Between September 2000 and December 31, 2006, 2869 HIV-infected children enrolled in PACTG 219C. Of these, 2575 were perinatally HIV infected, 74% of whom enrolled in 2000 and 2001. Table 2 shows the characteristics of the subjects at study entry. More than half (51.3%) were female, 59% were black, and more than 57% were 6 to 12.9 years of age at enrollment. Nearly two-thirds had a CD4 cell percentage of 25% or higher and 18.6% had a CD4 cell percentage between 15% and 24.9%, while 9.9% had less than 15% CD4 cells; 38.6% had undetectable HIV RNA (≤400 copies/mL), and only 6.7% had more than 100 000 copies/mL. Median follow-up was 59 months as of December 31, 2006. By the end of study follow-up, 710 girls had reached 13 years of age and constituted the risk set for normal pregnancies. Fifty-four girls who had a pregnancy prior to the end of follow-up were counted as at risk for pregnancy conditions and similarly for HELLP syndrome. Finally, 123 girls with Pap tests were considered at risk for gynecological dysplasias and carcinomas. These risk sets were reduced by any prevalent events for the incidence analysis.

Table 1 shows the numbers, percentages, and 95% CIs of children with a history of each condition prior to study entry or prior to January 1, 2001. For pregnancy-related conditions, there were no historical cases (events prior to 13th birthday). The most common historical conditions were encephalopathy (8.93%), cardiac disorders (MI, cardiomyopathy, and coronary heart disease) (6.10%), idiopathic thrombocytic purpura (4.54%), and renal disorders, including nephropathy and nephrolithiasis (1.94%). Fewer than 9 events at entry were reported in each of the following categories: epilepsy, erythema multiforme, gynecological dysplasia/neoplasia, hairy leukoplakia, substance use disorder, presumed peripheral neuropathy, and noncytomegalovirus retinitis. Two new diagnoses were added at different times after PACTG 219C was established: lipodystrophy and mitochondrial disorder. Few data are available on the presence of these conditions at study entry and over the study follow-up; they are therefore not included in our analyses.

Figure 1 describes the overall incidence rate per 100 person-years of noninfectious condition occurring from 2001 through 2006 for those conditions with at least 4 events during this period. The 10 noninfectious conditions with the highest IRs per 100 person-years were pregnancy conditions (22 events in 357 person-years; IR = 6.16; 95% CI, 3.9-9.3), gynecological dysplasias (28 events, 473 person-years; IR = 5.92; 95% CI, 3.9-8.6), encephalopathy (36 events, 9392 person-years; IR = 0.38; 95% CI, 0.3-0.5), pancreatitis (31 events, 10 304 person-years; IR = 0.30; 95% CI, 0.2-0.4), cardiac disorders (27 events, 9697 person-years; IR = 0.28; 95% CI, 0.2-0.4), renal disorders (27 events, 10 220 person-years; IR = 0.26; 95% CI, 0.2-0.4), proven peripheral neuropathy (24 events, 10 284 person-years; IR = 0.23; 95% CI, 0.2-0.4), birth defects (19 events, 10 218 person-years; IR = 0.19; 95% CI, 0.1-0.3), idiopathic thrombocytic purpura (15 events, 9917 person-years; IR = 0.15; 95% CI, 0.1-0.3), and condyloma (15 events, 10 322 person-years; IR = 0.15; 95% CI, 0.1-0.2). Among those conditions with 4 or more new events reported overall, the lowest rates per 100 person-years were for nephrolithiasis, epilepsy, microcephaly, and substance use with 7, 6, 5, and 4 events, respectively.

Table 3 and Figure 2 present the incidence rates for noninfectious conditions occurring from 2001 through 2006 by calendar year for those conditions with temporal trends (P < .10). Incidence rates decreased over time (Figure 2A) for encephalopathy (P < .001), pancreatitis (P = .002), and cardiac disorders (P = .007). In contrast, as shown in Figure 2B, the rates of gynecological dysplasia/neoplasia increased (P = .02). There was a slight but nonsignificant increase in women with pregnancies, while among women who reported at least 1 pregnancy, there was an increasing rate of conditions (including therapeutic abortions; P < .001). The median age of those considered at risk for normal pregnancies was 16.2 years (interquartile range [IQR], 14.7-18.2 years), while the median age at the time of reported pregnancies was 17.9 years (IQR, 16.7-19.1 years). In contrast, the median age of those at risk for abnormal pregnancies was 19.5 years (IQR, 18.2-21.3 years) and the median age for abnormal pregnancy was 18.9 years (IQR, 17.3-20.7 years).

The rest of the noninfectious condition categories had minimal event rates and did not demonstrate significant linear trends. Only 3 events of AIDS-defining cancers, 2 events of noncytomegalovirus retinitis, and 1 of hairy leukoplakia were observed, and no events of HELLP syndrome, erythema multiforme, or Stevens-Johnson syndrome were reported. There were 2 events of mental retardation reported.

Demographic characteristics and age at each first noninfectious condition diagnosis were evaluated for each condition category with at least 4 events in the time frame studied. Overall, 248 children (10%) experienced at least 1 of the incident noninfectious conditions studied. Of these 248 unique subjects, 61% were black and 27% were Hispanic. Slightly more than half (58%) were female. The median age at the occurrence of any first noninfectious condition was 13.3 years (IQR, 8.9-16.7 years). Certain disorders were more prevalent among males (cardiac disorders) while others were more common among females (peripheral neuropathy, birth defects, and microcephaly). Cardiac disorders were more common among 6- to 13-year-olds while renal disorders were most common in the older age group (>13 years). Twenty eight of the 248 participants (11%) with at least 1 of the noninfectious events subsequently died.

The immunologic and virological status at each first noninfectious condition diagnosis was also investigated. Only the first condition event regardless of category was considered. Of the 248 children with at least 1 event, 43% had a CD4 cell percentage lower than 25% at the time of the event (IQR, 17%-34%). Eleven percent of patients with first events had HIV RNA of more than 100 000 copies/mL; 41% were undetectable. Fifty-four percent of the patients with non-AIDS–defining cancers had viral loads less than 5000 copies/mL, 45% of patients diagnosed with cardiac disorder had viral loads less than 5000 copies/mL, and 85% who were diagnosed with peripheral neuropathy had viral loads less than 5000 copies/mL. More children with renal disorders had low CD4 cell percentage values and higher viral loads compared with the distribution of values for all children with any condition, while children with encephalopathy and pancreatitis also had high viral loads at diagnosis.

There were 65 deaths seen in the cohort of children followed up in 219C. These included 37 opportunistic infection–proven or–presumed deaths and 28 deaths related to other causes. The other categories included wasting syndrome, advanced or end-stage AIDS with no pathogens identified, cancer-related deaths, trauma-related deaths, and miscellaneous other deaths, unrelated to infections. There was no 1 category of deaths that had more than 4 identified causes.

The data that were collected as part of PACTG 219 and 219C have greatly contributed to our understanding of immune dysfunction in HIV-infected children and have helped address hypotheses that cannot be addressed in randomized clinical trials. The incidence of first episodes of these noninfectious conditions shows encouraging decreasing trends and continues to remain low. We observed statistically significant decreases in the incidence of encephalopathy, pancreatitis, and cardiac-related conditions. Myocardial infarction, cardiomyopathy, and coronary heart disease rates may be overreported because of the need to use the nonspecific diagnosis of “cardiac abnormality, unspecified.” Ten percent or 248 subjects experienced at least 1 new noninfectious event during the years studied, with only 20% of these presenting with a CD4 cell percentage less than 15% and only 11%, with a viral load more than 100 000 copies/mL at the time of diagnosis. There was a short lag between identification of an efficacious antiretroviral treatment for adults and its use in pediatric patients. By April 1998, guidelines were in place that recommended combination therapies for HIV-infected children and adolescents, and by 2000, there was excellent uptake of recommended combination therapy.¹⁷ Several disorders were seen more often in patients whose viral loads were more than 100 000 copies/mL, such as renal disorders, pancreatitis, and non-AIDS–defining cancers, while others, such as peripheral neuropathy, were almost exclusively seen in patients with viral loads less than 5000 copies/mL (85%). Perhaps these patients had good control of virus and their conditions were related to their long-term antiretroviral drug use (and excellent adherence) and not to failure of their immune system.

One specific set of conditions, those associated with gynecological/dysplasias, neoplasias, and carcinomas, showed a worrisome increase (IR increased from 2.3 per 100 person-years in 2001 to 32.6 per 100 person-years in 2006). A recent publication reviewing reproductive health in perinatally HIV-infected female adolescents documents differences between HIV-infected and noninfected adolescents, with rates of 29.7% for abnormal cervical cytologic results in HIV-infected adolescents¹⁹ as compared with rates of 11.5% in women attending a university health clinic in Montreal, Quebec, Canada.²⁰ We did not assess age at first intercourse, number of partners, or other high-risk behaviors that could have contributed to an increased incidence of gynecological abnormality–associated conditions. Human papillomavirus–associated conditions were also not serotyped; thus, we were unable to assess if specific serotypes (as in other populations) were more often associated with disease.

In addition, the high rate of pregnancy-associated conditions may be related to both pregnancies in a young population as well as advanced immune suppression related to both resistant virus and adherence in adolescents perinatally infected with HIV. We are concerned that these immunologically challenged perinatally infected females will have higher rates of pregnancy complications, dysplasia, and more rapid risk of progression to gynecological neoplasias. Studies that include risk behavior education as well as early routine vaccination with human papillomavirus vaccines will need to be conducted in this very high-risk and immunologically unstable population.

Despite treatment advances, however, HIV-infected children will continue to develop noninfectious HIV-related complications. Some children fail to respond to antiretroviral therapies, whereas others are unable to tolerate the complex medication regimens. Drug interactions, complex dosing schedules, adverse effects, and high costs of medication further limit the efficacy of these therapies. New medication and HIV-related toxicities, such as lipodystrophy and lipid abnormalities, may in turn increase event rates of new conditions (MI, coronary heart disease) as our perinatally infected youth transition into adult care. In addition, with the aging of our population, many of our patients are entering or are in adolescence, a time of life classically associated with poor adherence, perhaps potentiating current and future treatment failures.

The strength of the present analysis lies in the demographics of patients enrolled under the 4 versions of the protocol, well representing the population of pediatric HIV-infected patients followed up at PACTG sites in the United States. By encouraging coenrollment of all patients enrolled in primary therapy protocols, the PACTG 219C cohort hoped to eliminate potential bias in determining incidence of late outcomes. However, even though the PACTG 219C sample size is large and the demographics are similar to the overall group of patients followed up at PACTG sites, the cohort is not a random sample. The median age at entry was 9 years (IQR, 6-12 years), suggesting that enrollment bias may be possible since only the children who survived the pre-HAART era could be enrolled. The nonsurvivors could have been immunologically weaker and survivors had access to more potent protease inhibitors at younger ages therefore underestimating incidence rates.

In the future it will be important to continue to study this population of perinatally HIV-infected youth. It will also be critical to assess these noninfectious conditions as well as new conditions that arise that may be related to either HIV or its therapies.

Correspondence: Sharon A. Nachman, MD, Department of Pediatrics, Stony Brook University Medical Center, Health Sciences Center T-11, Room 031, Stony Brook, NY 11794-8111 (sharon.nachman@stonybrook.edu).

Accepted for Publication: May 14, 2008.

Author Contributions:Study concept and design: Nachman, Gona, Van Dyke, Dankner, Seage, and Williams. Acquisition of data: Van Dyke, Dankner, and Oleske. Analysis and interpretation of data: Nachman, Chernoff, Gona, Van Dyke, Dankner, Seage, and Williams. Drafting of the manuscript: Nachman, Chernoff, Gona, Seage, and Williams. Critical revision of the manuscript for important intellectual content: Nachman, Chernoff, Van Dyke, Dankner, Seage, Oleske, and Williams. Statistical analysis: Chernoff, Gona, Seage, and Williams. Obtained funding: Van Dyke. Administrative, technical, and material support: Gona and Van Dyke. Study supervision: Nachman, Van Dyke, Oleske, and Williams.

Financial Disclosure: None reported.

Funding/Support: This study was supported by the US National Institute of Allergy and Infectious Diseases and National Institute of Child Health and Human Development grant U01AI068632. This work was supported by the Statistical and Data Analysis Center at Harvard School of Public Health under the National Institute of Allergy and Infectious Diseases cooperative agreement 5 U01 AI41110 with PACTG and cooperative agreement 1 U01 AI068616 with the International Maternal Pediatric Adolescent AIDS Clinical Trials Group.

Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Allergy and Infectious Diseases or the National Institutes of Health. The institutions involved in funding and support were involved in the design, data collection, and conduct of protocol 219C but were not involved in the present analysis, the interpretation of the data, the writing of the manuscript, or the decision to submit for publication.

Previous Presentations: This study was presented in part at the Infectious Diseases Society of America Annual Meeting; October 3, 2004; Boston, Massachusetts.

Additional Contributions: We thank the children and families for their participation in PACTG 219C and the individuals and institutions involved in the conduct of 219C.