Five-year results from the AntiRetroviral Research for Watoto (ARROW) Trial

Polly Clayden, HIV i-Base

Results from the ARROW Trial comparing routine and clinically driven monitoring in children were presented at CROI 2013. [1]

This trial also investigated stopping vs continuing cotrimoxazole in children on ART; a four-drug induction strategy for starting treatment; as well as and once- versus twice-daily 3TC and abacavir. [2, 3, 4] Results from these evaluations were also presented at the conference.

ARROW was an open-label parallel-group trial conducted in Ugandan and Zimbabwean children, aged 3 months to 17 years and eligible for ART.

Small difference between routine and clinical monitoring

Adeodata Kekitiinwa presented findings from the monitoring comparison in an oral presentation. These results were also published in the Lancet on 7 March 2013. [5] Overall, long-term survival on ART was high in ARROW at 95%. Routine CD4 monitoring provided clinical benefit after the first year of treatment but event rates were very low, with small absolute differences between monitoring strategies and no difference in adverse events.

The trial enrolled 1206 children who were randomised to either clinically driven monitoring or routine laboratory and clinical monitoring for toxicity (haematology and biochemistry).

Clinicians were given12-week results for children assigned to routine laboratory monitoring (n=600) but in the clinically driven monitoring arm (n=606), only had toxicity results returned if requested for clinical reasons or grade 4 toxicities. Children switched to second-line ART if they had new WHO stage 3 or 4 events or (routine laboratory monitoring only) age-dependent CD4 criteria. Primary endpoints were new WHO 4 events or death and grade 3/4 adverse events, with the intention to show non-inferiority for efficacy margin +1.6/100 patient years.

At baseline, children were a median age of 6 years (range 0.4-17). Approximately one third overall were 3 years or younger and half were girls. Their median CD4 percent was 12% and weight for age -2.2; the majority (56%) was WHO stage 3. Sixty-two percent of children received nevirapine as their NNRTI and the remainder received efavirenz.

Only 33 (0.3%) children were lost to follow up over median of 4 years (IQR 3.7-4.4) from 2007 to 2012 (total 4685 child years). Few clinical participants had external CD4 tests done privately (4/571 at trial exit) and clinicians remained blinded.

At the end of the trial 94% and 95% of children in the laboratory and clinical arms respectively were still on their first line regimens with low rates of substitutions (7.5% overall) – where this occurred it was usually to accommodate TB treatment or for toxicity.

Thirty-five (6%) and 28 (5%) children in the laboratory and clinical arms switched to second line treatment. This was due to CD4 in 28 children in the laboratory monitoring arm, and for clinical reasons in 7 vs 28 children in the laboratory and clinical arms respectively. A further, 3 and 16 children in the laboratory and clinical arms switched due to unexplained failure to thrive.

The rates of switching to second line was low overall: 1.2 with laboratory and with 1.5 clinical monitoring per 100 child years, p=0.22. At switch the median CD4 percent was similar, 7% and 8% in the laboratory and clinical arms respectively. Dr Kekitiinwa noted that monitoring weight gain appeared to avoid an excess of very low CD4 switches.

There were 39 (1.7 per 100 child years) vs 47 (2 per 100 child years) new WHO stage 4 events or deaths in the laboratory and clinical monitoring arms, HR (clinical: laboratory) 1.13 (95% CI 0.73-1.53), p=0.59. This gave an absolute difference of +0.3 per 100 child years (95% CI -0.5 to +1.1), which fell within the specified non-inferiority margin.

Most of these events occurred in the first year of ART: 33/39 and 24/47 in the laboratory and clinical monitoring arms respectively. In years 2-5 of ART the difference was significant, with respectively 6 (0.4 per 100 child years) vs 23 (1.3 per 100 child years) events. This gave an absolute difference of +1.0 per 100 child years (95% CI +0.4 to +1.6), p=0.002, which remained within the non-inferiority margin.

Survival was high overall with 96% and 95% in the laboratory and clinical monitoring arms alive at 4 years. After the first year there was an absolute difference of +0.6 per 100 child years (95% CI =0.2 to +1.0), p=0.009. Of the 14 children that died during this period 12 were > 8 years old at time of death.

There was no difference in change of absolute CD4 counts or percent between the two arms. Viral load suppression rates were tested retrospectively and were also similar across both arms. Suppression rates were higher among children receiving NNRTI containing regimens and did not differ by strategy.

Dr Kekitiinwa concluded that it is possible to deliver ART safely to children with good quality clinical care without the need for routine monitoring. She noted that monitoring weight gain could be an important indicator of first line failure. However she added that there might be a role for targeted CD4 monitoring from the second year of ART.

“Resources should be focused on getting as many children onto treatment as possible rather than providing routine laboratory monitoring to fewer on ART”, she said.

She also noted that ARROW analyses did not find routine laboratory monitoring cost effective. These data were shown in a separate poster authored by Paul Revill and colleagues from the trial group.[6]

Analyses restricted to 12-228 weeks from starting ART revealed mean total costs per child of $2068 (laboratory) and $1532 (clinical), driven by higher costs with laboratory monitoring ($679 vs $25), although these were slightly offset by lower hospitalisation costs ($105 vs $145). There was an incremental cost-effectiveness ratio per life-year gained of $595,870. In weeks 52-228, mean total costs and life-years were $1536 and 3.37 (laboratory) compared to $1131 and 3.34 (clinical), leading to an incremental cost-effectiveness ratio per life-year gained of $14,560.

Removing the toxicity monitoring tests from the evaluation gave incremental cost-effectiveness ratio per life-year gained of $356,500 and $3,121 in weeks 12-228 and 52-228 respectively. In weeks 52-228, in <3-, 3- to 6-, 7- to 11-, and >11-year-olds these values were $56,784, $19,242, $11,235, and $3,144 respectively.

Continuing cotrimoxazole with ART beneficial

An important sub-study of ARROW looked at continuing daily cotrimoxazole prophylaxis in children on ART. This study found benefit to continuing cotrimoxazole in children on ART for 96 weeks or more, with persisting reductions in hospitalisations for malaria and other infections across all ages and CD4 levels. [2]

Mutsa Bwakura-Dangarembizi showed these data in an oral presentation following the main trial results. She explained that DART showed the benefit of continuing with cotrimoxazole in adults in the first 18 months of ART but there are no data on discontinuation in children. Based on expert opinion the WHO recommends that children >5 years who are stable on ART for at least 6 months with CD4 >350 cells/mm3, may stop.

For this analysis, 758 children were randomised to stop (n=382) or continue (n=376) daily cotrimoxazole after median 2.1 years (IQR 1.8 to 2.2) years on ART. Eligible children were aged >3 years, on ART >96 weeks, currently on receiving cotrimoxazole, using insecticide-treated bed-nets if living in malaria endemic areas and had no previous pneumocystis pneumonia (PCP). Primary endpoints were hospitalisation or death and grade 3 or 4 adverse events. Secondary endpoints were malaria, pneumonia, and gains in weight, height, BMI and CD4.

At baseline children were a median age of approximately 8 years old with a median CD4 percent of 33%. Dr Bwakura-Dangarembizi noted that this was a substantial improvement compared to 12% when the children entered the trial pre-ART.

The study found children stopping contrimoxazole had higher rates of hospitalisation or death, HR 1.57 (95% CI 1.09 to 2.26), p=0.007. This effect did not vary by age, sex, centre, country or monitoring strategy. Benefits in continuing were greatest in children with the highest CD4 (> 30%), HR 2.15 (95% CI 1.30 to 3.54).

Mortality was low and similar in the two groups, 2/382 and 3/376 in the group that stopped and continued respectively. Increased hospitalisations in the stop group were for malaria (49 vs 2I), HR (stop:continue) 2.10 (95% CI 1.43 to 3.09), as well as non-malarial infections (53 vs 25), particularly pneumonia, sepsis, and meningitis.

Overall, grade 3/4 adverse events were similar HR 1.17 (95% CI 0.82 to 1.68), p = 0.39, but there were more grade 4 events in children stopping contrimoxazole, HR 2.03 (95% CI 0.98 to 4.18), p=0.05. This was mostly driven by differences in anaemia (12 stop vs 2 continue).

Changes in height-for-age and CD4 were similar between groups; there was a trend towards greater weight gain in the children that continued contrimoxazole.

Dr Bwakura-Dangarembizi concluded that continuing cotrimoxazole in children on ART for >96 weeks is beneficial, with persisting reductions in hospitalisations for malaria and other infections across all ages and CD4 levels. Based on these results the ARROW investigators recommend that the guidelines for cotrimoxazole should be updated. She stressed that healthcare systems and supply-chains will have to be strengthened to avoid stockouts.

Short-term benefits of induction strategy do not persist

A late breaker poster authored by Patricia Nahirya-Ntege and colleagues showed results from the induction strategy, which found short-term benefits of four-drug first-line ART do not persist with three-drug maintenance. [3]

This randomisation was simultaneous with that by monitoring strategy and was to a standard three-drug or four-drug induction first-line ART. All children received an NNRTI, abacavir and 3TC (Arm A received this regimen alone). The children receiving four-drug induction regimens also received AZT. At 36 weeks those on the four-drug regimen reduced treatment to either NNRTI, abacavir and 3TC (Arm B) or abacavir, 3TC and AZT (Arm C).

At the trial end, 371 (93%) Arm A vs 387 (96%) Arm B vs 385 (95%) were still on first-line treatment with the majority on their original regimen. Only 5% switched to second-line ART and this was in similar proportions across the treatment groups.

Thirty children in Arm A, 20 in Arm B and 9 in Arm C stopped first-line nevirapine for TB treatment. In Arm A, nevirapine was mainly substituted with AZT in children less than 3 years old or efavirenz in older children. In the four-drug arms about a third just dropped the nevirapine to continue with three NRTIs.

There was no difference in CD4 percent change to 72 weeks across arms when all children were on three-drugs, p=0.33 or 144 weeks, p=0.69. At week 36 the four-drug (B and C) arms were superior to Arm A; +14.3% versus 12.4%, p <0.001.

At week 24, viral load suppression was significantly greater in induction four-drug arms (B and C) with 88% <400 copies/mL vs 77% arm A, p=0.002, but was similar across all arms by week 48, p= 0.76. At latest retrospective test at a median 3.7 years on ART, 83% A vs 84% B vs 65% C had viral load <400 copies/mL, p< 0.001.

One hundred and fifty seven (40%) A vs 190 (47%) B vs 218 (54%) C children had grade 3/4 events, p<0.001. Increases in arms B and C were driven by increased asymptomatic neutropenia in AZT-containing arms there were only 6 associated drug substitutions. Grade 3/4 anaemia occurred in similar proportions across arms.

The investigators noted that it is unknown whether continuing with four drugs would provide ongoing benefit and they suggested that triple NRTI ART might be useful during relatively short-term TB co-treatment.

Once- vs twice-daily abacavir and 3TC similar

The final presentation from ARROW was a poster showing findings from a further sub-study, a randomised comparison of once- vs twice-daily abacavir and 3TC in 669 children, authored by Victor Musiime and colleagues

The ARROW investigators previously demonstrated bioequivalence of plasma concentrations between abacavir and 3TC taken once- or twice-daily in 41 Ugandan children aged 3- 12 years.

In this comparison, children in ARROW receiving abacavir and 3TC-containing regimens twice-daily for at least 36 weeks, were randomised to remain on twice-daily or change to once-daily dosing (open-label). The children received single or co-formulated (Kivexa) antiretrovirals.

Primary outcomes were viral load at 48 weeks – measured retrospectively with a 12% non-inferiority margin for suppression – and study drug related grade 3/4 adverse events.

This study was conducted in a subgroup of 669 children who were a median age of 5.5 years (range 1.8-16.9), had a median CD4 percent of 33% and about half were girls. They were randomised to twice- daily (n = 333) vs once-daily (n=336) strategies after a median 1.8 years (range 0.9-3.0) on twice-daily abacavir and 3TC-containing first-line ART.

At the time of the twice-/once-daily randomisation, 51% vs 44%, 15% vs 22% and 34% vs 34% of children were receiving nevirapine, efavirenz and AZT respectively as their third drug. The majority (approximately 80%) took tablet formulations and the remainder syrups.

Over median 2.2 years follow-up, 98% vs 97% child-time was spent on twice- vs once-daily abacavir and 3TC in the two groups respectively. Self reported adherence was similar in the two groups.

There was no difference between groups in grade 3/4 or serious adverse events, WHO 3/4 events or weight- or height-for-age.

Proportions with viral load <80 copies/mL were similar in the twice- vs once-daily groups: at 48-weeks (n=661) there was a difference of -1.6% (95% CI, –8.4% to +5.2%), p = 0.65; at 96 weeks (n=539, assays ongoing) this was +0.7% (95%CI +6.9% to +8.3%), p=0.86.

Increases in CD4 percent in twice- vs once-daily were respectively +1.3% vs +0.8% at 48 weeks, p=0.25, and +2.5% vs +1.5% at 96 weeks, p = 0.06.

The investigators concluded that once daily dosing of abacavir and 3TC can be used to simplify treatment in children. Using efavirenz, this has the potential for a once-daily regimen.


ARROW gives us a very rich data set to further inform children’s treatment, particularly in resource-limited settings.

The most recent US guidelines include discussion on switching from twice-daily to once-daily dosing of 3TC at 8 to 10 mg/kg, based on review of data from the ARROW and PENTA 13 and 15 trials. [7, 8]

Triple NRTIs are expected to be the preferred regimen for children receiving ART and TB treatment in the next WHO guidelines, according to the discussion at the WHO hosted satellite just before the opening of CROI 2013. These revised guidelines will be released at the International AIDS Society conference in Malaysia – June 30 to July 3 2013.


Unless stated otherwise, all references are from the 20th Conference on Retroviruses and Opportunistic Infections (CROI), 3-6 March 2013, Atlanta.

Webcasts of the oral abstracts are fifth and sixth presentations at:

  1. Kekitiinwa A et al. Impact of routine laboratory monitoring after ART initiation in 1206 HIV+ African children: the 5-year AntiRetroviral Research for Watoto trial. Oral abstract 85.
  2. Bwakura-Dangarembizi M et al. Randomised comparison of stopping vs continuing cotrimoxazole prophylaxis among 758 HIV+ children on long-term ART: the Anti-Retroviral Research for Watoto trial. Oral abstract 86.
  3. Nahirya-Ntege P et al. Short-term benefits of 4-drug 1st-Line ART do not persist with 3-drug maintenance in 1206 HIV+ African Children: The 5-Year ARROW Trial. Poster abstract 184LB.
  4. Musiime V et al. Randomised comparison of once- vs twice-daily abacavir and lamivudine among 669 HIV+ children in the Anti-Retroviral Research for Watoto Trial. Poster abstract 977.
  5. ARROW Trial Team. Routine versus clinically driven laboratory monitoring and first-line antiretroviral therapy strategies in African children with HIV (ARROW): a 5-year open-label randomised factorial trial. The Lancet, Early Online Publication, 7 March 2013. doi:10.1016/S0140-6736(12)62198-9
  6. Revill P et al. Within-trial cost-effectiveness analysis of clinically-driven vs laboratory and clinical monitoring for 1206 African children on ART: The AntiRetroviral Research for Watoto Trial. Poster abstract 949.
  7. US paediatric guidelines updated. HTB. 1 February 2013.
  8. Panel on antiretroviral therapy and medical management of HIV-infected children. Guidelines for the use of antiretroviral agents in pediatric HIV infection.

Links to other websites are current at date of posting but not maintained.