Long-term Profiles of CD4 Reconstitution in HIV-infected Children Initiating Antiretroviral Therapy in Uganda and Zimbabwe

Abstract

Background: Studies of long-term immune reconstitution after antiretroviral therapy (ART) initiation are scarce in HIV-infected children in resource-limited settings. Methods: HIV-infected ART-naïve children meeting WHO criteria for ART in Uganda and Zimbabwe were enrolled in the ARROW trial between 2007 and 2008. CD4 counts were measured by flow cytometry at ART initiation and every 12 weeks subsequently. Based on our previous work in European children, we assumed an asymptotic CD4 reconstitution over time on ART with CD4 rising to a constant age-corrected level after a period of time. Using non-linear mixed-effects models for log (CD4 per expected CD4 in uninfected child of the same age), we investigated the effects of pre-ART age, sex, WHO stage and weight-forage z-score on pre-ART and long-term CD4-for-age. Results: 1206 children were enrolled (51% girls; ages 0.4-17.6 years) and followed for median (IQR) 2.8 (2.5, 3.2) years. Our model identified two groups of children. Group 1: 930 (77%) children had asymptotic CD4 reconstitution (median (IQR) age 5.8 (2.4, 9.0) years; pre-ART CD4 344 (158,634) cells per μl, estimated CD4 after 2.8 years of follow-up 693 (645, 866) cells per μl). The two key findings were: 1- a strong correlation between their pre-ART and long-term CD4 (r=0.37; p less than 0.0001), i.e. children who started ART with lower CD4 achieved lower CD4 in the long-term and 2- an association of older age at ART initiation with lower long-term CD4 reconstitution (p less than 0.0001). Lower pre-ART CD4-for-age was also associated with older age and lower weight-for-age z-score at ART initiation (both p less than 0.0001) and lower long-term CD4-for-age with male sex (p=0.002). Group 2: The 276 children for whom asymptotic CD4 reconstitution did not fit were older (7.2 (2.4,10.7) years; p=0.005), with higher pre-ART CD4 (469 (191,908) cells per μl) and CD4-for-age (-1.1(-1.7,-0.6); both p less than 0.0001). We used linear regression to identify four types of CD4-for-age trajectory within this sub-population. Conclusions: Our data suggest that CD4 and age at ART initiation may both drive different long-term profiles of immune reconstitution. These drivers appear to act in both a continuous and a stepwise manner to separate qualitatively different responses to ART in the long term. In the context of increasing availability of ART, our results highlight the importance of considering long-term immunological health, as well as short-term disease progression, in formulating future guidelines for ART initiation