Machine learning outperformed logistic regression classification even with limit sample size: A model to predict pediatric HIV mortality and clinical progression to AIDS.

Logistic regression (LR) is the most common prediction model in medicine. In recent years, supervised machine learning (ML) methods have gained popularity. However, there are many concerns about ML utility for small sample sizes. In this study, we aim to compare the performance of 7 algorithms in the prediction of 1-year mortality and clinical progression to AIDS in a small cohort of infants living with HIV from South Africa and Mozambique. The data set (n = 100) was randomly split into 70% training and 30% validation set. Seven algorithms (LR, Random Forest (RF), Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Naïve Bayes (NB), Artificial Neural Network (ANN), and Elastic Net) were compared. The variables included as predictors were the same across the models including sociodemographic, virologic, immunologic, and maternal status features. For each of the models, a parameter tuning was performed to select the best-performing hyperparameters using 5 times repeated 10-fold cross-validation. A confusion-matrix was built to assess their accuracy, sensitivity, and specificity. RF ranked as the best algorithm in terms of accuracy (82,8%), sensitivity (78%), and AUC (0,73). Regarding specificity and sensitivity, RF showed better performance than the other algorithms in the external validation and the highest AUC. LR showed lower performance compared with RF, SVM, or KNN. The outcome of children living with perinatally acquired HIV can be predicted with considerable accuracy using ML algorithms. Better models would benefit less specialized staff in limited resources countries to improve prompt referral in case of high-risk clinical progression.

Approach to the management of paediatric HIV spontaneous controllers.

Paediatric HIV spontaneous controllers (HSCs) are a unique and understudied population with potential to inform alternative treatment options for patients living with HIV. As HSCs are so rare and often not recognised prior to antiretroviral treatment (ART) initiation, it can be difficult for clinicians to optimally manage this group. We describe the diagnosis, history and management of three paediatric HSCs, two girls and a boy who were followed for 2, 1.25 and 10.4 years, respectively, before starting ART. All had low but detectable viral loads throughout follow-up but mostly marginally low CD4:CD8 ratios. The reason for starting ART in all was a gradual tendency to poorer virological control. This case series should assist in recognising paediatric HSCs. Clinical dilemmas arising in the management of paediatric HSCs include arriving at a correct HIV-positive diagnosis, correct diagnosis as an HSC, as well as whether to initiate ART. Decision-making for initiation of ART in paediatric HSCs should be individualised. Factors supporting ART initiation in these patients included increased frequency of viral load blips, increasing detectable viral load, CD4 percentage and CD4:CD8 ratio. Other factors included Hepatitis C serology and highly sensitive C-reactive protein. All three patients ultimately required ART, which supports universal initiation of ART in paediatric HSCs, but further research is required.

Dolutegravir twice-daily dosing in children with HIV-associated tuberculosis: a pharmacokinetic and safety study within the open-label, multicentre, randomised, non-inferiority ODYSSEY trial.

BACKGROUND: Children with HIV-associated tuberculosis (TB) have few antiretroviral therapy (ART) options. We aimed to evaluate the safety and pharmacokinetics of dolutegravir twice-daily dosing in children receiving rifampicin for HIV-associated TB. METHODS: We nested a two-period, fixed-order pharmacokinetic substudy within the open-label, multicentre, randomised, controlled, non-inferiority ODYSSEY trial at research centres in South Africa, Uganda, and Zimbabwe. Children (aged 4 weeks to <18 years) with HIV-associated TB who were receiving rifampicin and twice-daily dolutegravir were eligible for inclusion. We did a 12-h pharmacokinetic profile on rifampicin and twice-daily dolutegravir and a 24-h profile on once-daily dolutegravir. Geometric mean ratios for trough plasma concentration (Ctrough), area under the plasma concentration time curve from 0 h to 24 h after dosing (AUC0-24 h), and maximum plasma concentration (Cmax) were used to compare dolutegravir concentrations between substudy days. We assessed rifampicin Cmax on the first substudy day. All children within ODYSSEY with HIV-associated TB who received rifampicin and twice-daily dolutegravir were included in the safety analysis. We described adverse events reported from starting twice-daily dolutegravir to 30 days after returning to once-daily dolutegravir. This trial is registered with ClinicalTrials.gov (NCT02259127), EudraCT (2014-002632-14), and the ISRCTN registry (ISRCTN91737921). FINDINGS: Between Sept 20, 2016, and June 28, 2021, 37 children with HIV-associated TB (median age 11·9 years [range 0·4-17·6], 19 [51%] were female and 18 [49%] were male, 36 [97%] in Africa and one [3%] in Thailand) received rifampicin with twice-daily dolutegravir and were included in the safety analysis. 20 (54%) of 37 children enrolled in the pharmacokinetic substudy, 14 of whom contributed at least one evaluable pharmacokinetic curve for dolutegravir, including 12 who had within-participant comparisons. Geometric mean ratios for rifampicin and twice-daily dolutegravir versus once-daily dolutegravir were 1·51 (90% CI 1·08-2·11) for Ctrough, 1·23 (0·99-1·53) for AUC0-24 h, and 0·94 (0·76-1·16) for Cmax. Individual dolutegravir Ctrough concentrations were higher than the 90% effective concentration (ie, 0·32 mg/L) in all children receiving rifampicin and twice-daily dolutegravir. Of 18 children with evaluable rifampicin concentrations, 15 (83%) had a Cmax of less than the optimal target concentration of 8 mg/L. Rifampicin geometric mean Cmax was 5·1 mg/L (coefficient of variation 71%). During a median follow-up of 31 weeks (IQR 30-40), 15 grade 3 or higher adverse events occurred among 11 (30%) of 37 children, ten serious adverse events occurred among eight (22%) children, including two deaths (one tuberculosis-related death, one death due to traumatic injury); no adverse events, including deaths, were considered related to dolutegravir. INTERPRETATION: Twice-daily dolutegravir was shown to be safe and sufficient to overcome the rifampicin enzyme-inducing effect in children, and could provide a practical ART option for children with HIV-associated TB. FUNDING: Penta Foundation, ViiV Healthcare, UK Medical Research Council.

Viral suppression is associated with HIV-antibody level and HIV-1 DNA detectability in early treated children at 2 years of age.

OBJECTIVES: Early infant HIV diagnosis and antiretroviral therapy (ART) initiation are now implemented shortly after birth. Maintaining and monitoring ART adherence is difficult and requires frequent visits. We, therefore, investigated whether HIV antibodies and HIV-1 DNA levels are markers of cumulative viremia. DESIGN: We conducted a cross sectional investigation at 2 years of age of HIV antibodies and HIV-1 DNA levels in a well characterized cohort of 31 children who started ART shortly after birth. METHODS: HIV antibodies were measured by a combination of the Abbott ARCHITECT HIV Ag/Ab Combo and Geenius HIV 1/2 supplemental assays; and total HIV-1 DNA quantified using a sensitive quantitative PCR (qPCR) assay targeting the HIV-1 integrase gene. RESULTS: Infant post-exposure prophylaxis consisted of zidovudine (AZT) and nevirapine (NPV) (or NVP only, in one child) within 1 day of birth, transitioning, after positive diagnosis, to three-drug ART, at a median [interquartile range (IQR)] of 7 (4-9.5) days. Twelve of 31 children had well suppressed HIV plasma viral loads (HIVVL) and the remainder periods of viremia (HIVVL > 100 copies/ml after 3 months of ART), classified as non-suppressed. At 24 months of age: 11 of 12 (92%) of well suppressed children had undetectable HIV-1 antibodies versus 3 of 19 (16%) non-suppressed children (P < 0.001) and 7 of 12 (58%) well suppressed children had undetectable HIV-1 DNA versus 3 of 19 (16%) non-suppressed children (P = 0.02). CONCLUSION: Considering low assay costs and the high proportion of well suppressed children with undetected antibody levels at 2 years, HIV antibody levels may be a valuable marker of cumulative adherence in children who start treatment shortly after birth and could prompt adherence and viral load investigation.

Pharmacokinetics of adjusted-dose 8-hourly lopinavir/ritonavir in HIV-infected children co-treated with rifampicin.

OBJECTIVES: To evaluate the proportion of children with lopinavir Cmin ≥1 mg/L when receiving a novel 8-hourly lopinavir/ritonavir dosing strategy during rifampicin co-treatment. METHODS: HIV-infected children on lopinavir/ritonavir and rifampicin were enrolled in a prospective pharmacokinetic study. Children were switched from standard-of-care lopinavir/ritonavir-4:1 with additional ritonavir (1:1 ratio) twice daily to 8-hourly lopinavir/ritonavir-4:1 using weight-banded dosing. Rifampicin was dosed at 10-20 mg/kg/day. After 2 weeks, plasma samples were collected ∼2, 4, 6, 8 and 10 h after the morning lopinavir/ritonavir-4:1 dose, ALT was obtained to assess safety and treatment was switched back to standard of care. ClinicalTrials.gov registration number: NCT01637558. RESULTS: We recruited 11 children in two weight bands: 5 (45%) were 10-13.9 kg and received 20-24 mg/kg/dose of lopinavir and 6 (55%) children weighed 6-9.9 kg and received 20-23 mg/kg/dose of lopinavir. The median age was 15 months (IQR = 12.6-28.8 months). The median (IQR) lopinavir Cmin was 3.0 (0.1-5.5) mg/L. Seven (63.6%) of the 11 children had Cmin values ≥1 mg/L. Children with a lopinavir mg/kg dose below the median 21.5 were more likely to have Cmin <1 mg/L (P = 0.02). There was a strong positive correlation between lopinavir and ritonavir concentrations. No associations were found between lopinavir AUC2-10 and age, sex, weight, nutritional status or mg/kg/dose of lopinavir. CONCLUSIONS: These data do not support the use of 8-hourly lopinavir/ritonavir at studied doses. Evaluation of higher doses is needed to optimize treatment outcomes of TB and HIV in young children.

Strong sex bias in elite control of paediatric HIV infection

BACKGROUND: Reports of posttreatment control following antiretroviral therapy (ART) have prompted the question of how common immune control of HIV infection is in the absence of ART. In contrast to adult infection, where elite controllers have been very well characterized and constitute approximately 0.5% of infections, very few data exist to address this question in paediatric infection. METHODS: We describe 11 ART-naive elite controllers from 10 cohorts of HIV-infected children being followed in South Africa, Brazil, Thailand, and Europe. RESULT: All but one of the elite controllers (91%) are females. The median age at which control of viraemia was achieved was 6.5 years. Five of these 11 (46%) children lost control of viraemia at a median age of 12.9 years. Children who maintained control of viraemia had significantly higher absolute CD4þ cell counts in the period of elite control than those who lost viraemic control. On the basis of data available from these cohorts, the prevalence of elite controllers in paediatric infection is estimated to be 5­10-fold lower than in adults. CONCLUSION: Although conclusions are limited by the study design, these data suggest that, whilst paediatric elite control can be achieved, compared with adult elite controllers, this occurs rarely, and takes some years after infection to achieve. Also, loss of immune control arises in a high proportion of children and often relatively rapidly. These findings are consistent with the more potent antiviral immune responses observed in adults and in females

Strong sex bias in elite control of paediatric HIV infection.

BACKGROUND: Reports of posttreatment control following antiretroviral therapy (ART) have prompted the question of how common immune control of HIV infection is in the absence of ART. In contrast to adult infection, where elite controllers have been very well characterized and constitute approximately 0.5% of infections, very few data exist to address this question in paediatric infection. METHODS: We describe 11 ART-naive elite controllers from 10 cohorts of HIV-infected children being followed in South Africa, Brazil, Thailand, and Europe. RESULTS: All but one of the elite controllers (91%) are females. The median age at which control of viraemia was achieved was 6.5 years. Five of these 11 (46%) children lost control of viraemia at a median age of 12.9 years. Children who maintained control of viraemia had significantly higher absolute CD4 cell counts in the period of elite control than those who lost viraemic control. On the basis of data available from these cohorts, the prevalence of elite controllers in paediatric infection is estimated to be 5-10-fold lower than in adults. CONCLUSION: Although conclusions are limited by the study design, these data suggest that, whilst paediatric elite control can be achieved, compared with adult elite controllers, this occurs rarely, and takes some years after infection to achieve. Also, loss of immune control arises in a high proportion of children and often relatively rapidly. These findings are consistent with the more potent antiviral immune responses observed in adults and in females.