TB in the ARROW trial of children on ART

45th lung health logoPolly Clayden, HIV i-Base

Incidence of TB disease was highest in the first three months in children with HIV receiving antiretroviral treatment (ART) in the ARROW trial. [1]

TB was associated with lower weight-for-age z-score and lower CD4 percent. Fewer children receiving ART and continuing cotrimoxazole prophylaxis beyond 96 weeks were diagnosed with TB than those that stopped. These findings were shown at the 45th Union World Conference on Lung Health.

In ARROW 1206 Ugandan and Zimbabwean children and adolescents, aged 3 months to 17 years and eligible for ART, were randomised to clinical or laboratory monitoring and to receive standard 3-drug or 4-drug ART. There was a secondary randomisation to continue or stop taking cotrimoxazole after 96 weeks of ART.

Retention and outcomes were excellent in ARROW: only 3% of children were lost to follow-up, 95% remained on first line ART and 96% were alive at the end of the trial.

Angela Cook presented findings from the TB sub-study on behalf of the ARROW trial team.

TB data were collected using a standardised form at enrolment and throughout a median of 4 years follow up. There were 120 TB diagnoses in 111 participants (106 pulmonary, 9 extra pulmonary disseminated and 5 lymphnode). An Endpoint Review Committee adjudicated the diagnoses using all available data but blind to randomised arm; 71% of diagnoses were presumptive.

Of 1206 children, 237 with a previous history of TB were excluded. Of the remaining 969, 900 (93%) had no TB and 69 (7%) confirmed TB. TB incidence decreased over time from initiation of ART: 11.5/100 patient years (py) at 0 – 3 months; 3.1/100 py at 3 – 12 months and 1.4/100 py at >12 months.

In adjusted analysis, lower weight z-score, and lower CD4 percent were associated with TB diagnoses: respectively -3.3 (IQR -4.5 to -2.0) vs -2.1 (-3.2 to -1.2) and 8% (IQR 5 to 13%) vs 13% (7 to 19%) in the TB vs no TB groups. Both comparisons p<0.001.

Dr Cook noted that these results underline the importance of TB screening before starting ART.

In the second part of the sub-study the investigators looked at TB and cotrimoxazole. There were 760 participants randomised to continue contrimoxaxole beyond 96 weeks on ART, of whom 138 with a previous history of TB were excluded. A total of 622 of had a median follow up of 2 years, of this group, 20 (3%) had confirmed TB.

Participants with TB had a median CD4 percent of 22% (IQR 13 – 34%) vs 33% (IQR 27 – 39) in the group with no TB, p<0.001. Of the 20 participants with TB, 15 (75%) were in the group that stopped cotrimoxazole and 5 (25%) in the group that continued, HR (stop: continue) 3.1 (95% CI 1.1 – 8.4) p=0.03. These data add to the evidence of the benefits of prolonged use of cotrimoxazole.


Crook A et al. Incidence of TB and risk factors in HIV co-infected children in Africa: an analysis of the ARROW trial. 45th Union World Conference on Lung Health. Barcelona, Spain. 28 October – 1 November, 2014. OAP-200-30. Abstract presentations, Thursday, 30 October. S202.


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