Browse the corpus

Second-Line Antiretroviral Therapy for Children Living with HIV in Africa. BACKGROUND: Children living with human immunodeficiency virus (HIV) have limited options for second-line antiretroviral therapy (ART). METHODS: In this open-label trial with a 2-by-4 factorial design, we randomly assigned children with HIV who had first-line treatment failure to receive second-line therapy with tenofovir alafenamide fumarate (TAF)-emtricitabine or standard care (abacavir or zidovudine, plus lamivudine) as the backbone and dolutegravir or ritonavir-boosted darunavir, atazanavir, or lopinavir as the anchor drug. The primary outcome was a viral load of less than 400 copies per milliliter at 96 weeks. We hypothesized that TAF-emtricitabine would be noninferior to standard care, that dolutegravir and ritonavir-boosted darunavir would each be superior to ritonavir-boosted lopinavir and atazanavir analyzed in combination, and that ritonavir-boosted atazanavir would be noninferior to ritonavir-boosted lopinavir. Safety was also assessed. RESULTS: A total of 919 children underwent randomization; 458 were assigned to receive TAF-emtricitabine, and 461 to receive standard care. Assigned anchor drugs were dolutegravir (229 participants), ritonavir-boosted darunavir (232), ritonavir-boosted atazanavir (231), and ritonavir-boosted lopinavir (227). The median age of participants was 10 years, and 497 (54.1%) were male. The median viral load at baseline was 17,573 copies per milliliter. At week 96, TAF-emtricitabine was superior to standard care: the adjusted difference in the percentage of participants with a viral load of less than 400 copies per milliliter was 6.3 percentage points (95% confidence interval [CI], 2.0 to 10.6; P = 0.004). Dolutegravir was superior to ritonavir-boosted lopinavir and atazanavir analyzed in combination (adjusted difference, 9.7 percentage points; 95% CI, 4.8 to 14.5; P<0.001), but ritonavir-boosted darunavir was not (adjusted difference, 5.6 percentage points; 95% CI, 0.3 to 11.0; P = 0.04 [prespecified threshold, P = 0.03]). Ritonavir-boosted atazanavir was noninferior to ritonavir-boosted lopinavir. One child died, and 29 (3.2%) had serious adverse events, with no significant between-group differences. CONCLUSIONS: Second-line ART regimens including TAF-emtricitabine and dolutegravir were effective for children, with no evidence of safety concerns. Ritonavir-boosted darunavir was also effective. (Funded by the European and Developing Countries Clinical Trials Partnership and others; CHAPAS-4 ISRCTN Registry number, ISRCTN22964075.).

CHAPAS-4(ISRCTN22964075) was a randomised, open-label trial with a 2x4 factorial design. The trial was approved by ethics committees in Uganda, Zambia, Zimbabwe, and UK (protocol: www.mrcctu.ucl.ac.uk/studies/all-studies/c/chapas-4). Participants were recruited at six centres in three African countries: Uganda (Joint Clinical Research Centre (JCRC), Kampala; JCRC, Mbarara), Zambia (University Teaching Hospital, Lusaka; Arthur Davison Children’s Hospital, Ndola) and Zimbabwe (University of Zimbabwe Clinical Research Centre, Harare; Mpilo Central Hospital, Bulawayo). Participants were CLHIV aged 3-15 years, weighing ≥14kg, receiving first-line NNRTI-based ART, with treatment failure according to WHO criteria (confirmed VL>1000 copies/ml (after adherence counselling) or immunological/clinical criteria for failure) and screening visit VL>400 copies/ml. Post-menarchal females required a negative pregnancy test. Guardians provided written informed consent, with additional assent from older children, according to national guidelines. Children were excluded if they had severe hepatic impairment (alanine aminotransferase (ALT) ≥5x upper limit of normal (ULN), or ALT ≥3xULN and bilirubin ≥2xULN, or clinical liver disease). Full study details can be found in the protocol at nejm.org.

itional assent from older children, according to national guidelines. Children were excluded if they had severe hepatic impairment (alanine aminotransferase (ALT) ≥5x upper limit of normal (ULN), or ALT ≥3xULN and bilirubin ≥2xULN, or clinical liver disease). Full study details can be found in the protocol at nejm.org. Participants were randomised to one of two backbones (TAF/FTC or standard-of-care (SOC) (abacavir (ABC)/3TC or ZDV/3TC, whichever not used first-line)) and simultaneously to one of four anchor drugs (DTG, DRV/r, ATV/r, LPV/r). Randomisation was stratified by centre and first-line NRTI (ABC or ZDV). A computer-generated sequential randomisation list with variably sized permuted blocks was prepared by the trial statistician and incorporated securely into an online database. Allocation was concealed until eligibility was confirmed by local centre staff, who then randomised. Participants were seen at screening, ART switch (week 0), 2, 6, 12 weeks and 12-weekly thereafter to at least week-96 (primary endpoint): extended follow-up continued until 02/02/2023. Children with tuberculosis at enrolment or during follow-up underwent regimen modification to account for rifampicin drug-drug interactions. Additional measures ensured participant follow-up during the COVID-19 pandemic (Supplementary appendix: Supplementary methods).

nt): extended follow-up continued until 02/02/2023. Children with tuberculosis at enrolment or during follow-up underwent regimen modification to account for rifampicin drug-drug interactions. Additional measures ensured participant follow-up during the COVID-19 pandemic (Supplementary appendix: Supplementary methods). Primary outcome was VL <400 copies/ml at week-96 (death counted as failure). Secondary efficacy outcomes were VL <60 copies/ml (the lower limit at one site) and <1000 copies/ml at week-96, death/World Health Organisation (WHO) 3/4 events, changes in CD4 count/percentage, and genotypic resistance. Safety outcomes were grade 3/4, serious, and ART-modifying adverse events (AEs); and changes in total cholesterol, low-density lipoprotein (LDL), high-density lipoprotein (HDL), triglycerides, bilirubin and creatinine clearance (CrCl). Other outcomes included changes in weight-, height- and body mass index (BMI)-for-age and bone mineral density Z-scores. An economic analysis considered costs which were estimated from the health-system perspective and included ART, clinic visits and hospital stays in 2022 US dollars, discounted at 3% per annum (Supplementary appendix).

Primary outcome was VL <400 copies/ml at week-96 (death counted as failure). Secondary efficacy outcomes were VL <60 copies/ml (the lower limit at one site) and <1000 copies/ml at week-96, death/World Health Organisation (WHO) 3/4 events, changes in CD4 count/percentage, and genotypic resistance. Safety outcomes were grade 3/4, serious, and ART-modifying adverse events (AEs); and changes in total cholesterol, low-density lipoprotein (LDL), high-density lipoprotein (HDL), triglycerides, bilirubin and creatinine clearance (CrCl). Other outcomes included changes in weight-, height- and body mass index (BMI)-for-age and bone mineral density Z-scores. An economic analysis considered costs which were estimated from the health-system perspective and included ART, clinic visits and hospital stays in 2022 US dollars, discounted at 3% per annum (Supplementary appendix). For the backbone randomisation, assuming 80.0%-87.5% SOC achieved VL<400 copies/ml at week-96, 920 children provided ≥95% power to demonstrate TAF was non-inferior (10% margin) (two-sided alpha=0.05), assuming 2.5% loss-to-follow-up (reduced from 10% in original protocol). For the anchor randomisation, 920 children provided 88% power to demonstrate ATV/r was non-inferior (12% margin) to LPV/r (two-sided alpha=0.05), assuming 80% <400 copies/ml at week-96, and 89% power to detect 10% higher suppression in each of DTG and DRV/r than LPV/r and ATV/r combined (two-sided alpha=0.03; as multiple comparisons) (including 2.5% loss to follow-up). Margins reflect the clinical consensus and are within the range used in previous second-line treatment trials in adults (Supplementary appendix: methods). An independent data monitoring committee reviewed interim data four times using the Haybittle–Peto criterion (99.9% confidence interval).

ng 2.5% loss to follow-up). Margins reflect the clinical consensus and are within the range used in previous second-line treatment trials in adults (Supplementary appendix: methods). An independent data monitoring committee reviewed interim data four times using the Haybittle–Peto criterion (99.9% confidence interval). Analyses were intention-to-treat using Stata (version 17.0). Primary endpoint analyses used logistic regression (adjusting for stratification factors), then marginal estimation of risk differences. For non-inferiority comparisons, secondary per-protocol analyses included children who received randomised backbone/anchor drug for >90% follow-up. For death/WHO 3/4 events, and grade 3/4 serious and ART-modifying AEs, groups were compared via Cox regression (unadjusted). Changes in continuous outcomes were analysed using Normal generalised estimating equations adjusting for visit, stratification factors and baseline (and interactions between these factors and visit), for an overall analysis of difference between groups over all visits (independent correlation; mean difference reported). 95% confidence intervals were not adjusted for multiple testing (Supplementary appendix: methods). European Developing Country Clinical Trial Partnership (funder), and pharmaceutical companies donating additional funding (Gilead Sciences, Johnson and Johnson) and drugs (ViiV Healthcare, Gilead Sciences, Johnson and Johnson, CIPLA), did not participate in trial design, conduct or analysis.

919 children were randomised between 17/12/2018 and 01/04/2021 (Figure 1). Baseline characteristics were similar between arms (Table 1; Table S3). 497(54.1%) children were male; median age 10 years (IQR 8,13); 777(84.5%) were WHO stage 1/2. Median weight-, height- and BMI-for-age Z-scores were between -1 and -1.6. Median VL was 17,573 copies/mL (IQR 5549,55700); CD4 count 669 cells/mm3 (413,971), CD4% 28%(19%,36%). Median time on first-line ART was 5.6 years (44% nevirapine, 56% efavirenz). Over 96 weeks, 98.9% of visits were attended. Eleven children (1.2%) were lost to follow-up. 674(73.3%) entered extended follow-up (median 60 (IQR 30,75) additional weeks). In SOC, 217/461(47.1%) initiated ABC/3TC, 244(52.9%) ZDV/3TC. Prior to week-96, children spent 99.1% of time on allocated backbone (99.5% TAF/FTC vs. 98.8% SOC) and five (0.5%) initiated third-line ART (2(0.4%) TAF/FTC vs. 3(0.7%) SOC). In extended follow-up, children spent 93.5% of time on allocated backbone (95.6% TAF/FTC, 91.4% SOC) (Figure S1).

ed ABC/3TC, 244(52.9%) ZDV/3TC. Prior to week-96, children spent 99.1% of time on allocated backbone (99.5% TAF/FTC vs. 98.8% SOC) and five (0.5%) initiated third-line ART (2(0.4%) TAF/FTC vs. 3(0.7%) SOC). In extended follow-up, children spent 93.5% of time on allocated backbone (95.6% TAF/FTC, 91.4% SOC) (Figure S1). At week-96, 406/454(89.4%) TAF/FTC vs. 378/454(83.3%) SOC had VL <400copies/mL (adjusted difference +6.3% [95% confidence interval (CI) +2.0%,+10.6%]; p=0.004) (Figure 2). Therefore, TAF/FTC was non-inferior (and superior) to SOC according to the pre-specified 10% margin. There was no evidence of heterogeneity in the effect of TAF/FTC vs. SOC in any of 11 prespecified sub-groups (Figure S2), including first-line NRTI, anchor randomisation, country and baseline VL. Results of per-protocol analyses were similar: 403/449(89.8%) TAF/FTC vs. 370/445(83.1%) SOC had VL <400copies/mL (adjusted difference +6.8%[+2.4%,+11.1%]; p=0.002). Differences between arms in suppression <60 and <1000 copies/mL were similar between arms, as were results at weeks 48 and 144 (Table S4).

nd baseline VL. Results of per-protocol analyses were similar: 403/449(89.8%) TAF/FTC vs. 370/445(83.1%) SOC had VL <400copies/mL (adjusted difference +6.8%[+2.4%,+11.1%]; p=0.002). Differences between arms in suppression <60 and <1000 copies/mL were similar between arms, as were results at weeks 48 and 144 (Table S4). Over 96 weeks, 127/919(13.8%) children experienced 176 grade 3/4 AEs (63(13.8%) TAF/FTC vs. 64(13.9%) SOC) (Cox p=0.93) (Table 2; Table S6), including eight infections, all in SOC (4 malaria, 3 tuberculosis, 1 herpes zoster). Twenty-nine(3.2%) children experienced a total of 31 serious AEs (15(3.3%) TAF/FTC vs. 14(3.0%) SOC) (p=0.84) (Table 2; Table S7); most were hospitalisations with infection. One child died (TAF/FTC+DTG, from hypotension/toxic shock/severe malnutrition, judged by the investigators as ART-unrelated). Twenty-four (2.6%) children experienced a total of 41 ART-modifying AEs (any grade) (11(2.4%) TAF/FTC vs. 13(2.8%) SOC) (p=0.68), of which 33 were tuberculosis-related protocol-specified modifications (Table 2).

AF/FTC+DTG, from hypotension/toxic shock/severe malnutrition, judged by the investigators as ART-unrelated). Twenty-four (2.6%) children experienced a total of 41 ART-modifying AEs (any grade) (11(2.4%) TAF/FTC vs. 13(2.8%) SOC) (p=0.68), of which 33 were tuberculosis-related protocol-specified modifications (Table 2). Over 96 weeks, weight-, height- and BMI-for-age Z-scores increased more with TAF/FTC vs. SOC (mean Z-score difference (averaged over all visits to week 96) +0.09[95% CI +0.04,+0.13], +0.04 [+0.01,+0.07] and +0.10 [+0.04,+0.16], respectively). In extended follow-up, increases were maintained and similar (Figure 3; Figure S6). Comparing TAF/FTC vs. SOC at week-96, corresponding mean weight increase was 7.0 vs. 6.2kg; height increase was 10.2 vs. 9.8cm. There was a small reduction in mean CrCl in both arms at week 96, greater in TAF vs. SOC (mean -16 vs. -11ml/min), which persisted in extended follow-up (Figure S4). Phosphate excretion was similar between arms and no child discontinued TAF for renal dysfunction (Figure S5). At randomisation, 910/919(99.0%) initiated their randomised anchor drug (eight with tuberculosis coinfection randomised to ATV/r or DRV/r initiated LPV/r or DTG (protocol-specified modification), one error). Through week-96, children spent 98.6% follow-up on allocated anchor drug (99.1% DTG, 98.5% DRV/r, 98.6% ATV/r, 98.4% LPV/r) and five (0.5%) initiated third-line ART (1 DRV/r, 2 ATV/r, 2 LPV/r). In extended follow-up, children spent 86.2% of time on allocated anchor drug (99.1% DTG, 95.6% DRV/r, 93.7% ATV/r, 54.9% LPV/r) (Figure S1).

6, children spent 98.6% follow-up on allocated anchor drug (99.1% DTG, 98.5% DRV/r, 98.6% ATV/r, 98.4% LPV/r) and five (0.5%) initiated third-line ART (1 DRV/r, 2 ATV/r, 2 LPV/r). In extended follow-up, children spent 86.2% of time on allocated anchor drug (99.1% DTG, 95.6% DRV/r, 93.7% ATV/r, 54.9% LPV/r) (Figure S1). At week-96, 92.0% DTG, 88.3% DRV/r, 84.3% ATV/r and 80.7% LPV/r had VL <400 copies/mL (Figure 2). Considering the pre-specified comparisons (Table S5), DTG was superior to LPV/r and ATV/r arms combined (adjusted difference +9.7% [95%CI +4.8%,+14.5%]; p<0.001). DRV/r was not superior to LPV/r and ATV/r combined as the comparison did not meet pre-specified significance (adjusted difference +5.6% [+0.3%,+11.0%]; p=0.04 vs. threshold p=0.03 from multiple comparisons). ATV/r was non-inferior to LPV/r (adjusted difference +3.4% [-3.4%,+10.2%]; p=0.33). Per-protocol analysis was similar (Supplementary appendix: results). For each comparison, there was no evidence of heterogeneity among 11 prespecified sub-groups, including first-line NRTI, randomised backbone, country and baseline VL, apart from marginally greater VL response for DTG vs. LPV/r and ATV/r combined following nevirapine vs efavirenz first-line(Figures S2). In a post-hoc analysis, VL suppression was +4.0% [-1.3%,+9.4%] higher with DTG vs. DRV/r (Table S5). For each comparison, results using <60 and <1000 copies/ml VL thresholds were similar, as was suppression at weeks 48 and 144 (Figure 2; Table S5).

TV/r combined following nevirapine vs efavirenz first-line(Figures S2). In a post-hoc analysis, VL suppression was +4.0% [-1.3%,+9.4%] higher with DTG vs. DRV/r (Table S5). For each comparison, results using <60 and <1000 copies/ml VL thresholds were similar, as was suppression at weeks 48 and 144 (Figure 2; Table S5). Over 96 weeks, 127/919(13.8%) children experienced grade 3/4 AEs (Table 2; Table S6), most commonly hyperbilirubinemia, predictably almost exclusively ATV/r-associated (Figure S7). Fewer children experienced grade 3/4 AEs with DTG(5.2%) vs. LPV/r(11.5%) (p=0.02); there was no evidence of differences between DRV/r(8.6%) vs. LPV/r(11.5%) (p=0.31). Twenty-nine(3.2%) children experienced serious AEs (6 DTG, 8 DRV/r, 5 ATV/r, 10 LPV/r) (p>0.1) (Table S7). Twenty-four(2.6%) experienced ART-modifying AEs of any grade, with no differences across arms (7 DTG, 5 DRV/r, 5 ATV/r, 7 LPV/r) (p>0.5). Weight- and BMI-for-age Z-scores increased more with ATV/r, DRV/r and DTG vs. LPV/r (Figure 3; Table S8). There was no evidence that anchor drugs’ effects on weight-for-age Z-scores differed by backbone (Figure S6). Additional secondary outcome analyses (including lipid (Figure S9) and bone health (Figure S10) comparisons) are reported in Supplementary appendix. TAF/FTC had lower cost than SOC (by $37.68), resulting in a high probability of being cost-saving. DTG was the least costly anchor drug, saving $190.77 compared to ATZ/r; DRV/r was the most expensive (Supplementary appendix).

In SOC, 217/461(47.1%) initiated ABC/3TC, 244(52.9%) ZDV/3TC. Prior to week-96, children spent 99.1% of time on allocated backbone (99.5% TAF/FTC vs. 98.8% SOC) and five (0.5%) initiated third-line ART (2(0.4%) TAF/FTC vs. 3(0.7%) SOC). In extended follow-up, children spent 93.5% of time on allocated backbone (95.6% TAF/FTC, 91.4% SOC) (Figure S1). At week-96, 406/454(89.4%) TAF/FTC vs. 378/454(83.3%) SOC had VL <400copies/mL (adjusted difference +6.3% [95% confidence interval (CI) +2.0%,+10.6%]; p=0.004) (Figure 2). Therefore, TAF/FTC was non-inferior (and superior) to SOC according to the pre-specified 10% margin. There was no evidence of heterogeneity in the effect of TAF/FTC vs. SOC in any of 11 prespecified sub-groups (Figure S2), including first-line NRTI, anchor randomisation, country and baseline VL. Results of per-protocol analyses were similar: 403/449(89.8%) TAF/FTC vs. 370/445(83.1%) SOC had VL <400copies/mL (adjusted difference +6.8%[+2.4%,+11.1%]; p=0.002). Differences between arms in suppression <60 and <1000 copies/mL were similar between arms, as were results at weeks 48 and 144 (Table S4).

AF/FTC+DTG, from hypotension/toxic shock/severe malnutrition, judged by the investigators as ART-unrelated). Twenty-four (2.6%) children experienced a total of 41 ART-modifying AEs (any grade) (11(2.4%) TAF/FTC vs. 13(2.8%) SOC) (p=0.68), of which 33 were tuberculosis-related protocol-specified modifications (Table 2). Over 96 weeks, weight-, height- and BMI-for-age Z-scores increased more with TAF/FTC vs. SOC (mean Z-score difference (averaged over all visits to week 96) +0.09[95% CI +0.04,+0.13], +0.04 [+0.01,+0.07] and +0.10 [+0.04,+0.16], respectively). In extended follow-up, increases were maintained and similar (Figure 3; Figure S6). Comparing TAF/FTC vs. SOC at week-96, corresponding mean weight increase was 7.0 vs. 6.2kg; height increase was 10.2 vs. 9.8cm. There was a small reduction in mean CrCl in both arms at week 96, greater in TAF vs. SOC (mean -16 vs. -11ml/min), which persisted in extended follow-up (Figure S4). Phosphate excretion was similar between arms and no child discontinued TAF for renal dysfunction (Figure S5).

At randomisation, 910/919(99.0%) initiated their randomised anchor drug (eight with tuberculosis coinfection randomised to ATV/r or DRV/r initiated LPV/r or DTG (protocol-specified modification), one error). Through week-96, children spent 98.6% follow-up on allocated anchor drug (99.1% DTG, 98.5% DRV/r, 98.6% ATV/r, 98.4% LPV/r) and five (0.5%) initiated third-line ART (1 DRV/r, 2 ATV/r, 2 LPV/r). In extended follow-up, children spent 86.2% of time on allocated anchor drug (99.1% DTG, 95.6% DRV/r, 93.7% ATV/r, 54.9% LPV/r) (Figure S1).

TV/r combined following nevirapine vs efavirenz first-line(Figures S2). In a post-hoc analysis, VL suppression was +4.0% [-1.3%,+9.4%] higher with DTG vs. DRV/r (Table S5). For each comparison, results using <60 and <1000 copies/ml VL thresholds were similar, as was suppression at weeks 48 and 144 (Figure 2; Table S5). Over 96 weeks, 127/919(13.8%) children experienced grade 3/4 AEs (Table 2; Table S6), most commonly hyperbilirubinemia, predictably almost exclusively ATV/r-associated (Figure S7). Fewer children experienced grade 3/4 AEs with DTG(5.2%) vs. LPV/r(11.5%) (p=0.02); there was no evidence of differences between DRV/r(8.6%) vs. LPV/r(11.5%) (p=0.31). Twenty-nine(3.2%) children experienced serious AEs (6 DTG, 8 DRV/r, 5 ATV/r, 10 LPV/r) (p>0.1) (Table S7). Twenty-four(2.6%) experienced ART-modifying AEs of any grade, with no differences across arms (7 DTG, 5 DRV/r, 5 ATV/r, 7 LPV/r) (p>0.5). Weight- and BMI-for-age Z-scores increased more with ATV/r, DRV/r and DTG vs. LPV/r (Figure 3; Table S8). There was no evidence that anchor drugs’ effects on weight-for-age Z-scores differed by backbone (Figure S6). Additional secondary outcome analyses (including lipid (Figure S9) and bone health (Figure S10) comparisons) are reported in Supplementary appendix.

TAF/FTC had lower cost than SOC (by $37.68), resulting in a high probability of being cost-saving. DTG was the least costly anchor drug, saving $190.77 compared to ATZ/r; DRV/r was the most expensive (Supplementary appendix).

TAF/FTC provided superior virological suppression vs. ABC/3TC or ZDV/3TC. DTG-based regimens were virologically superior vs. LPV/r and ATV/r arms combined; DRV/r-based regimens achieved higher virological suppression than LPV/r and ATV/r arms combined but could not be declared superior (although significance was close to the multiple-comparison adjusted threshold). LPV/r was associated with the poorest virological outcomes, growth, lipid profiles (Figure S9) and bone health (Figure S10). These comparisons between TAF/FTC (including a new 120/15mg paediatric formulation) and SOC, and the four main currently available second-line anchor drugs for children provide much-needed robust evidence to guide future drug formulation development and paediatric guidelines. Children did well clinically with infrequent hospitalisation or disease progression and only one death over 96 weeks (due to advanced disease). This is in part attributable to relatively high baseline CD4 counts, supporting the principle of switching to second-line before evidence of significant immune-compromise.

TAF/FTC provided superior virological suppression vs. ABC/3TC or ZDV/3TC. DTG-based regimens were virologically superior vs. LPV/r and ATV/r arms combined; DRV/r-based regimens achieved higher virological suppression than LPV/r and ATV/r arms combined but could not be declared superior (although significance was close to the multiple-comparison adjusted threshold). LPV/r was associated with the poorest virological outcomes, growth, lipid profiles (Figure S9) and bone health (Figure S10). These comparisons between TAF/FTC (including a new 120/15mg paediatric formulation) and SOC, and the four main currently available second-line anchor drugs for children provide much-needed robust evidence to guide future drug formulation development and paediatric guidelines. Children did well clinically with infrequent hospitalisation or disease progression and only one death over 96 weeks (due to advanced disease). This is in part attributable to relatively high baseline CD4 counts, supporting the principle of switching to second-line before evidence of significant immune-compromise. The superior virological suppression of 89.4% at 96 weeks observed with TAF/FTC is comparable to the 93-100% reported in four small single-arm paediatric trials of TAF.13 Of note, >85% were virologically suppressed at baseline in these studies, whereas all children in CHAPAS-4 had baseline VL >400 copies/ml. Our results are also similar to the 86-92% virological suppression on TDF or TAF in the adult African NADIA and VISEND second-line trials,5–7 and the 84-86% VL suppression at 96 weeks in a pooled analysis of TDF/TAF in 14 adult initial treatment trials.14

whereas all children in CHAPAS-4 had baseline VL >400 copies/ml. Our results are also similar to the 86-92% virological suppression on TDF or TAF in the adult African NADIA and VISEND second-line trials,5–7 and the 84-86% VL suppression at 96 weeks in a pooled analysis of TDF/TAF in 14 adult initial treatment trials.14 Weight-, height- and BMI-for-age z-scores all increased more with TAF/FTC, suggesting overall better growth which is potentially a consequence of improved virological suppression. There was no evidence of bone toxicity with TAF, and if anything, greater increases in bone mineral density vs. SOC as assessed by total-body-less-head dual-energy X-ray absorptiometry (irrespective of anchor drug) (Figure S10). These findings, alongside the additional benefits of smaller pill size, once-daily administration, lower cost and lower risk of hypersensitivity, make TAF a valuable second-line option. Although mean CrCl decreased slightly more over 96 weeks with TAF/FTC, values remained within normal limits, with no clinician-assessed associated grade 3/4 adverse events; no child discontinued medication for renal impairment, and there was no evidence of tubulopathy.

nsitivity, make TAF a valuable second-line option. Although mean CrCl decreased slightly more over 96 weeks with TAF/FTC, values remained within normal limits, with no clinician-assessed associated grade 3/4 adverse events; no child discontinued medication for renal impairment, and there was no evidence of tubulopathy. The superior virologic suppression with DTG vs. ATV/r and LPV/r combined extends findings from the ODYSSEY trial which showed superiority of DTG vs. SOC for both first- and second-line ART (ODYSSEY second-line SOC being 72% LPV/r, 24% ATV/r, 1% DRV/r).15 CHAPAS-4 provides additional evidence through direct randomised comparisons of DTG and DRV/r vs. ATV/r or LPV/r. Given DTG’s cost-effectiveness, small milligram dosing and authorisation for use below 3 years, these results further support DTG as second-line anchor drug of choice in WHO guidelines (when not used first-line).8 WHO also recommends DTG combined with optimised NRTI backbone for adults failing NNRTI-based ART,8 based in part on superiority of DTG vs. LVP/r in the DAWNING trial,16 and non-inferiority of DTG vs. DRV/r (with TDF or ZDV) in the NADIA trial.6,7

line anchor drug of choice in WHO guidelines (when not used first-line).8 WHO also recommends DTG combined with optimised NRTI backbone for adults failing NNRTI-based ART,8 based in part on superiority of DTG vs. LVP/r in the DAWNING trial,16 and non-inferiority of DTG vs. DRV/r (with TDF or ZDV) in the NADIA trial.6,7 CHAPAS-4 demonstrated immune reconstitution for all drugs, particularly during 24 weeks after second-line ART initiation (Figure S8). Age-appropriate weight-gain was observed with all anchor drugs except LPV/r, which showed minimal increases in weight-for-age Z-scores in a population with already low baseline scores (Figure 3). A systematic review and meta-analysis evaluating weight-gain among adults reported greater weight-gain among those receiving DTG with TAF compared to other NRTIs,17 but we observed no excessive weight-gain with any anchor/backbone combination, including DTG+TAF/FTC. Excess weight-gain in adults has been associated with advanced immunosuppression at ART initiation, high VL, female sex and black race, mostly occurring in the first 2 years of therapy.18 This phenomenon has been described as “return to health” where resting energy expenditure returns to normal as HIV viremia and inflammation are controlled.19 CHAPAS-4 participants were either normal or underweight at baseline (Table 1), and none had evidence of obesity. Results may therefore not be generalisable to more overweight paediatric populations. As expected, lipid profiles were less favourable for children on LPV/r (Figure S9) and hyperbilirubinemia was predictably seen with ATV/r (Figure S7).

ther normal or underweight at baseline (Table 1), and none had evidence of obesity. Results may therefore not be generalisable to more overweight paediatric populations. As expected, lipid profiles were less favourable for children on LPV/r (Figure S9) and hyperbilirubinemia was predictably seen with ATV/r (Figure S7). Our findings also show that DRV/r and ATV/r are effective once-daily treatment options which could be considered if DTG cannot be used second-line. Previous small studies have shown ATV/r to be effective in children and potentially a preferred and better tolerated second-line option compared to LPV/r,20 as long as hyperbilirubinemia is not associated with discontinuation. LPV/r use in children has considerable challenges of unpalatability and twice-daily dosing. The additional data on poorer growth, abnormal lipid profiles and lower virological suppression in CHAPAS-4 emphasize that LPV/r may be suboptimal.

o LPV/r,20 as long as hyperbilirubinemia is not associated with discontinuation. LPV/r use in children has considerable challenges of unpalatability and twice-daily dosing. The additional data on poorer growth, abnormal lipid profiles and lower virological suppression in CHAPAS-4 emphasize that LPV/r may be suboptimal. Our trial strengths include its power to compare both DTG and DRV/r with ATV/r and LPV/r while employing a factorial design to compare TAF-based with SOC backbones. The trial was conducted in three African countries, including three centres outside capital cities, increasing generalisability of results across sub-Saharan Africa where the majority of CLHIV live. Whilst the findings can inform guidelines on second-line regimen after NNRTI-based first-line ART, children currently initiating first-line DTG will also require robust second-line options. A limitation is that CHAPAS-4 does not provide direct evidence to inform anchor/backbone choice in this situation; however, safety and efficacy could be inferred (given lack of evidence of interaction) and they will undoubtedly remain important future options. The relatively high CD4 counts at enrolment may also limit generalisability to severely immunocompromised children. One factor that may have impacted ATV/r and DRV/r efficacy was the lack of co-formulated tablets, resulting in a relatively high pill burden (although a small 25mg ritonavir generic pill was used). Overcoming this barrier through FDC manufacture may further enhance the effectiveness of ritonavir-boosted PIs for children in future. The open-label design of the trial could have potentially introduced bias; however the primary endpoint (VL) was objective. See Table S1 for further review of representativeness/generalisability.

this barrier through FDC manufacture may further enhance the effectiveness of ritonavir-boosted PIs for children in future. The open-label design of the trial could have potentially introduced bias; however the primary endpoint (VL) was objective. See Table S1 for further review of representativeness/generalisability. The impact of baseline genotypic NRTI resistance on risk of virological failure, as well as development of acquired resistance mutations during second-line ART, are important considerations for product/formulation prioritisation. Retrospective analyses of resistance results from all children at baseline and those with VL >400 copies/ml at weeks 48 and/or 96 are ongoing. Overall, CHAPAS-4 results provide efficacy and safety data for TAF/FTC and DTG for paediatric second-line ART. If scaled up, TAF/FTC could also result in cost savings (Supplementary appendix). DRV/r offers several benefits over ATV/r (e.g. higher resistance barrier, ongoing FDC development) but cannot be used under 3 years and is relatively costly so alternative ritonavir-boosted PI/non-INSTI anchor options for young children remain important.21 CHAPAS-4 results support further development of child friendly FDCs of TAF/FTC, with or without anchor drugs, and their inclusion on the priority list of the WHO Paediatric Drug Optimization (PADO) program,22 which in turn should inform future guidelines and prioritisation of the most effective paediatric drugs and formulations for roll-out in Africa and globally.

f child friendly FDCs of TAF/FTC, with or without anchor drugs, and their inclusion on the priority list of the WHO Paediatric Drug Optimization (PADO) program,22 which in turn should inform future guidelines and prioritisation of the most effective paediatric drugs and formulations for roll-out in Africa and globally. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org.