| |
Effect of early antiretroviral therapy on the risk of AIDS/death in HIV-infected infants: ART before age 3 months dramatically reduces AIDS/death
|
| |
| |
"In HIV-1 vertically infected infants, starting antiretroviral therapy before the age of 3 months is associated with a significant reduction in progression to AIDS and death.....Deferring treatment was associated with increased risk of progression [crude hazard ratio 5.0; 95% confidence interval (CI) 2.0-12.6; P = 0.001] that persisted after adjusting for cohort in multivariate models (adjusted hazard ratio 3.0".....In conclusion, results from this first analysis of the European Infant Collaborative study strongly suggest that ART initiated before the age of 3 months has a dramatic effect in reducing progression to AIDS and death in high income countries, and concur with results from a randomized trial in South Africa [13]. Deferring treatment in infected infants is no longer an option and guidelines for industrialized and poor-income countries are being updated [25]."
AIDS:Volume 23(5)13 March 2009p 597-604
Goetghebuer, Tessaa; Haelterman, Edwigea; Le Chenadec, Jeromeb; Dollfus, Catherinec; Gibb, Dianad; Judd, Alid; Green, Hannahd; Galli, Luisae; Ramos, Jose Tomasf; Giaquinto, Carlog; Warszawski, Josianebhi; Levy, Jacka; for the European Infant Collaboration group
aPediatric Department, CHU St Pierre, Brussels, Belgium
bInserm, U822, Le Kremlin-Bicetre, France
cAP-HP, Hopital Trousseau, Service d'Hematologie et d'oncologie pediatrique, Paris, France
dCollaborative HIV Paediatric Study (CHIPS), MRC Clinical Trials Unit, London, UK
eItalian Register, Department of Paediatrics, University of Florence, Italy
fHospital Universitario de Getafe, Madrid, Spain
gDepartment of Pediatrics, University of Padova, Padova, Italy
hUniversite Paris-Sud, Faculte de Medecine Paris-Sud, Le Kremlin-Bicetre, France
iAP-HP, Hopital Bicetre, Epidemiology and Public Health Service, France
Abstract
Objective: In the absence of treatment, rapid progression to AIDS occurs in approximately 20% of HIV-1-infected infants over the first year of life.. The prognosis of these children has considerably improved with highly active antiretroviral therapy. As data from well resourced countries are lacking, the objective of this collaborative study was to evaluate the impact of early treatment in vertically infected infants.
Design: Children born to HIV-infected mothers between 1 September 1996 and 31 December 2004, who were diagnosed with HIV and free of AIDS before 3 months, were eligible. Demographics and pregnancy data, details of antiretroviral therapy, and clinical outcome were collected from 11 European countries.
Methods: The risk of AIDS or death, by whether or not an infant started treatment before 3 months of age, was estimated by Kaplan-Meier survival analysis and Cox proportional hazards models.
Results: Among 210 children, 21 developed AIDS and three died. Baseline characteristics of the 124 infants treated before 3 months were similar to those of the 86 infants treated later. The risk of developing AIDS/death at 1 year was 1..6 and 11.7% in the two groups, respectively (P < 0.001). Deferring treatment was associated with increased risk of progression [crude hazard ratio 5..0; 95% confidence interval (CI) 2.0-12.6; P = 0.001] that persisted after adjusting for cohort in multivariate models (adjusted hazard ratio 3.0; 95% CI 1.2-7.9; P = 0.021).
Conclusion: In HIV-1 vertically infected infants, starting antiretroviral therapy before the age of 3 months is associated with a significant reduction in progression to AIDS and death.
Introduction
In the absence of highly active antiretroviral therapy (HAART) about 20% of children born in developed countries will progress to AIDS or die in the first year of life [1-3]. Markers for high risk of rapid disease progression in children under 12 months of age are inaccurate, although infants with positive PCR at birth (presumed to be infected in utero) appear to be at higher risk [4,5]. When HAART became available in the mid-1990s, initiation of combination therapy in the first months of life was advocated as a possible approach to avoid rapid progression of disease [6,7].. However, this remained controversial: the limited knowledge on pharmacokinetics of antiretroviral drugs in early life and the lack of appropriate formulations for infants complicated their administration, and there were concerns about the long-term toxicities and the risk of development of resistance, thereby limiting future treatment options.
Several nonrandomized studies in industrialized countries have suggested that HAART initiation before 6 months of age reduces the occurrence of early-onset severe disease [8-10]. In 2004 the PACTG356 trial demonstrated that initiating treatment at an age of 3 months or younger was associated with an improved long-term viral suppression [11]. Furthermore, children initiating HAART before the age of 3 months have been shown to maintain higher CD4 cell counts despite variable rates of viral load suppression [9-12]. Most recently, the clinical impact of early initiation of HAART has been established by a South African randomized, controlled trial including 375 infants [Children with HIV Early antiRetroviral therapy (CHER)]. This trial demonstrated a highly significant 76% reduction in mortality in infants without symptoms and with CD4 >25% initiating HAART before 12 weeks of age compared with those deferring therapy [13].
The clinical benefit of treating HIV-infected infants early in life has never been quantified in a large population in industrialized countries, and a prospective randomized trial is not anymore feasible in this setting.. As recommendations about the timing of HAART initiation in infants diagnosed with HIV have varied across Europe and have changed over time, we designed this collaborative cohort study to evaluate the benefit of starting therapy before 3 months of age in asymptomatic HIV-infected infants born between 1996 and 2004 in eleven European countries.
Results
Table 1 describes and compares the characteristics of mothers and children in each cohort. Overall, the median follow-up time was 58 months [interquartile range (IQR): 35-89], and differed significantly between cohorts ranging from 43 months (IQR: 33-81) for the UK/Ireland to 76 months (IQR: 41-101) for local cohorts. In the UK/Ireland, France, and local cohorts, most mothers originated from sub-Saharan Africa (60-84%). ART was more often administered during pregnancy and delivery in the UK/Ireland and French cohorts (approximately 80%) than in other cohorts. Caesarean section was performed for around half of the mothers in most cohorts, and for all in the Italian cohort. Around 20% of deliveries occurred before 37 weeks of gestational age, with highest proportions in Italian and other cohorts. Breastfeeding was very rare in all cohorts. CD4 at baseline differed between cohorts, with lower CD4 cell count in Italy and other cohorts than in France, UK, and local cohorts. The median age at ART initiation varied significantly according to cohort (P = 0.002), ranging from 2 months in France and in local cohorts to 5 months in the UK/Ireland.. At 1 year, 79% of the infants in the UK/Ireland had been treated with ART compared to approximately 95% for each of the other cohorts (P = 0..001), and this trend was reflected in the proportion of infants treated early, lowest in the UK/Ireland cohort (37%) and highest in France (68%) and in local cohorts (75%). Overall during the period 1996-2004, the proportion of infants treated before 3 months remained stable (P for trend = 0.6). The proportion of infants who started ART as mono or dual therapy was higher in France (26%) and Italy (36%) than in the UK/Ireland (9%) and local cohorts (8%).
Overall, 124 infants initiated treatment before 3 months of age (early treatment group), and 86 did not (deferred treatment group). Baseline maternal, infant characteristics and duration of follow-up were very similar in these two groups (Table 1). A total of 24 infants developed AIDS (n = 21) or died (n = 3). The risk of AIDS/death strongly varied according to cohort (P < 0.001), the highest risk being observed in the UK/Ireland cohort (Table 2). Differences remained significant when restricting the analysis to the end of the study period (2000-2004).
None of the baseline maternal, obstetrical, and infant characteristics in Table 2 was significantly associated with the occurrence of AIDS/death.
Among the 124 infants treated early, four developed AIDS (three encephalopathies diagnosed at 7, 8, and 13 months of age; one wasting syndrome identified at 82 months) and two died (at 17 and 45 months, respectively). Among the 86 infants from the deferred treatment group, 58 started ART within the first year of life. This included six children who started therapy because of the occurrence of an AIDS-defining event [three encephalopathies, one Pneumocystis jiroveci pneumonia (PCP), one cytomegalovirus (CMV) infection, one serious recurrent bacterial infection], and 52 children who started therapy in the absence of AIDS. Among the latter, six developed AIDS several weeks or months after treatment start (three encephalopathies, two CMV infections, and one serious recurrent bacterial infection) and one died without treatment. Among the 28 children who were untreated at 1 year of age, six AIDS events were reported.
The Kaplan-Meier cumulative probability risk of AIDS or death was considerably higher in infants with deferred treatment compared with those initiating ART early: 11.7 vs.1.6% at 1 year, and 21.5 vs. 4.6% at 5 years (P < 0.001) (Fig.. 1). The crude hazard ratio of AIDS/death associated with deferred compared with early ART was 5.0 [95% confidence interval (CI) 2.0-12.6; P = 0.001] (Table 2). This association remained highly significant and of the same magnitude after adjusting separately for each possible confounder (year of birth, sex, maternal origin, ART during pregnancy, ART during delivery, mode of delivery, gestational age <37 weeks, birth weight <2500 g, CD4 cell percentage <25%, number of drugs in postnatal prophylaxis, PCP prophylaxis, and number of drugs in the first ART). Only adjustment for cohort had an impact on the estimated value of hazard ratio, which decreased to 3.0 (95% CI 1.2-7.9; P = 0.021). In a sensitivity analysis restricted to children who started HAART (including triple NRTI or two NRTI + nevirapine) as first antiretroviral therapy, the association was even stronger (hazard ratio 7.9; 95% CI 2.3-27.6; P < 0.001).
Discussion
The prognosis of HIV-infected children has improved considerably since the introduction of HAART, with a rate of progression to AIDS/death in infancy and early childhood decreasing significantly after 1996 [14,16,17]. However, criteria for initiation of HAART in HIV-infected infants have varied over time and between countries and clinical centers. In previous US and European guidelines [18-20], indications to start treatment have been based on level of CD4% or viral load, according to analysis of the risk of AIDS in untreated infected children from the large longitudinal HIV Paediatric Prognostic Markers Collaborative Study (HPPMCS) [21]. However, these biological markers are poor predictors of rapid disease progression in infancy. Although there was a consensus on the need to treat infants who present clinical symptoms of disease progression or with very low CD4, at the time when the study patients were born early initiation of HAART in asymptomatic infants was still controversial [18-20,22].
Several single-arm studies (some of which contributed data to the European Infant Collaborative study) from Europe [9,10,12] and the USA [11] have previously reported variable virological responses, but good clinical and immunological responses to the early initiation of HAART in infancy. More importantly, the CHER randomized trial in South Africa, which included 375 infants from birth, recently reported a 76% reduction in mortality in those who started HAART before the age of 12 weeks, compared with the deferred treatment arm, after only 32 weeks of follow-up [13]. A randomized trial is no longer feasible in industrialized countries. As vertically infected children are prospectively followed in most European cohorts, this collaborative study provided the opportunity to evaluate in a large number of infants followed up prospectively from birth, whether early initiation of HAART is associated with a clinical benefit in high-income countries. The results of the analysis indicate that the risk of AIDS/death was three times lower in infants starting antiretroviral therapy before 3 months of age as compared with those deferring treatment. The reduction in the risk of progression to AIDS or death with early vs.. deferred ART observed in this study is consistent with the reduction in mortality risk reported in the randomized controlled trial, although in a very different setting.
The cohorts participating to the study were quite heterogeneous. Some were incident cohorts including children born to HIV-infected mothers and using prospective standardized follow-up from birth; others were hospital-based cohorts including HIV-infected children at time of diagnosis which may occur later in infancy or childhood; others were a mix of both types of populations. In order to overcome potential selection biases due to the heterogeneity of the cohorts' populations, we restricted inclusion in the study to infants who received ART in the postnatal period to ensure that they were identified as at risk of HIV infection since birth and prospectively followed, and who were diagnosed with HIV within 3 months of age. Furthermore we excluded infants with AIDS before the age of 3 months. Reporting of type and age of onset of AIDS clinical events may vary according to cohort, especially for complex events such as encephalopathies or wasting syndrome. However, undiagnosed stage C events are likely to be rare for these children followed in specialized pediatric clinical centers. The association between deferred therapy and the occurrence of AIDS/death decreased but remained strong after adjustment for cohorts (hazard ratio = 3.0). Other potential confounders such as year of birth, maternal origin, and type of ART had little effect on the relationship between treatment group and risk of AIDS/death so that crude and adjusted estimates of hazard ratio were very similar.
Prior to the availability of ART in European countries, approximately 20% of infants experienced early disease progression in the first year of life and 10% died [1,3]. The benefit or systematic antiretroviral therapy before 3 months of age to avoid this rapid progression has been recently demonstrated in the South African controlled randomized trial [13]. Our analysis confirmed that this is also the case in routine clinical practice in industrialized countries. As shown in the figure, the AIDS events in the early treatment group occurred uniformly over time during the follow-up, whereas in the deferred treatment group, most events occurred before the age of 2 years. The rate of occurrence of AIDS or death of infant from this latter group is similar to historical reports of natural history in infected infants without treatment in similar populations [2,3].. HIV encephalopathy is a particular concern for children exposed to HIV during brain development, and is difficult to detect early enough to prevent irreversible damage. A study from the French Perinatal Cohort showed that early initiation of ART was associated with a significantly reduced risk of HIV encephalopathy [8]. Immunological progression is also limited by early treatment in infants [23]. The recent prospective study from South Africa showed that 85% of untreated vertically infected infants presented CD4 percentage under 25% over the first 6 months of life [24].
However, initiating HAART soon after the diagnosis of vertical transmission remains difficult in daily practice. Parents or carers need support to help them to cope with the diagnosis and optimize adherence to medication. Palatable pediatric syrups, appropriate low-dose tablet formulations, and pharmacokinetic data to inform correct dosing of antiretroviral drugs in infancy are still needed. The best ART regimen for use in early life remains debated, and has not been addressed here. Other important issues are the long-term toxicities of antiretroviral agents, the problems of therapy adherence relying on caregiver administration of medication, and the development of resistance, limiting future therapy options. The safety of interrupting therapy in older children is currently being addressed in the PENTA 11 trial and in the CHER trial [13].
In conclusion, results from this first analysis of the European Infant Collaborative study strongly suggest that ART initiated before the age of 3 months has a dramatic effect in reducing progression to AIDS and death in high income countries, and concur with results from a randomized trial in South Africa [13]. Deferring treatment in infected infants is no longer an option and guidelines for industrialized and poor-income countries are being updated [25].
Methods
Participating cohorts
Thirteen cohorts from 11 European countries participated to this study, providing their dataset for 327 HIV-infected children born between 1 September 1996 and 31 December 2004, and followed since birth. Of these 327 infants, 201 (61%) originated from three national multicenter cohorts: the French Pediatric Cohort - ANRS/EPF CO10 (France, EPF) (n = 96), the Collaborative HIV Paediatric Study (UK and Ireland, CHIPS) (n = 52), and the Italian Register for HIV Infection in Children (Italy, ITLR) (n = 53). All HIV-infected mothers delivering at participating clinical sites and their infants are eligible for these cohorts and data from the infants are prospectively collected. The vast majority of clinical sites participate in the national cohorts' data collection. For the EPF cohort, it has been estimated that 70% of the detected HIV-positive pregnant mothers delivering in France are included (Josiane Warszawski, personal communication). These three cohorts are described in details elsewhere [8,9,14]. Various other cohorts contributed a total of 65 infants, principally from individual large tertiary clinical centers [Amsterdam (n = 6), Rotterdam (n = 3), Munich (n = 5)], four centers in Belgium (n = 20), and six centers in Spain (n = 31) and are referred as 'local cohorts' in the study. 'Other cohorts' included the following: study patients from the Paediatric European Network for the Treatment of AIDS (PENTA) 7 study [12] and from the European Collaborative Study (ECS) [3] (n = 20) that did not belong to the above-mentioned cohorts, children from the Swiss Mother and Child HIV Cohort Study (MoCHiV) (n = 15) and from single centers in Eastern Europe [Warsaw (n = 17) and Bucharest (n = 9)].
Study population
To minimize selection bias linked to the heterogeneity of inclusion criteria in the different cohorts, the following inclusion criteria for this study were used: infants born between 1 September 1996 and 31 December 2004 to mothers known to be HIV-infected at delivery, who received antiretroviral prophylaxis within 3 days after birth (n = 259), and who were diagnosed for HIV-1 infection by at least two positive tests (PCR or culture or Ag p24) during the first 91 days of life (n = 216). Infants who were lost to follow-up (n = 2) or who developed AIDS before 3 months of age (n = 4) were excluded. Taking into account these inclusion criteria, 210 infants constitute the population of the present analysis.
Variables and data collection
Demographics and pregnancy data, details of prophylaxis and ART in early life, Centers for Disease Control (CDC) events and death, and immunological and virological measurements since birth were collected. Data on resistance and on compliance were available from very few cohorts and were thus not considered for pooling. Crude data were abstracted from structured routine databases (National Cohorts, and some of the local and other cohorts) or from medical records (most local and other cohorts) and transferred to the coordinating center at Saint Pierre Hospital, Brussels, which sent queries to the cohort coordinators. Data were checked for inconsistencies, cleaned, and pooled with SPSS software (version 15.0; SPSS Inc., Chicago, Illinois, USA).
Definitions
Postnatal antiretroviral prophylaxis was defined as any antiretroviral regimen started within 3 days following birth for a maximum of 12 weeks. Treatment was defined as any antiretroviral regimen started after 3 days of life or, if started within 3 days of life, continued for longer than 12 weeks. Infants were classified into the 'early treatment group' if they started ART before 3 months of age (0-91 days) and into the 'deferred treatment group' if they did not start ART before 3 months of age. HAART was defined as any regimen including at least three antiretroviral drugs.. Outcome was AIDS or death from any cause combined. AIDS was defined as the first occurrence of a CDC category C event [15]. CD4 at baseline was defined as CD4 measure closest to the age of 1 month, and not after the age of 3 months.
Statistical analysis
Baseline characteristics of the two treatment groups were compared using Student's t-test for normally distributed continuous data or Mann-Whitney U and Kruskal-Wallis tests for nonnormally distributed continuous data. Chi-squared tests, Fisher exact tests, or Monte Carlo methods were used for categorical data.. The risk of AIDS/death in the two treatment groups was estimated by Kaplan-Meier survival analysis censored at last follow-up, using the log-rank to test for significance. We performed univariate and multivariate analyses using Cox proportional hazards models to estimate crude and adjusted hazard ratios for the associations between baseline risk factors, treatment group, and the occurrence of AIDS/death combined. Adjusted models were undertaken introducing potential confounders one by one because of the limited number of events. We performed an intention-to-treat analysis and did not take into account interruption or changes in treatment after the initial regimen.
The proportional hazard assumption was tested by the Schoenfeld's residuals method. We used SPSS version 15.0 and STATA version 10 (STATA Corp., College Station, Texas, USA) for the statistical analyses, and considered differences to be significant at the 5% probability level.
|
|
| |
| |
|
|
|
