Background
Disseminated TB is a common cause of hospitalisation for people with HIV and has high early mortality, which is associated with innate immune activation and signalling. We hypothesized that adjunctive corticosteroids could improve outcomes.
Methods
The NewStrat-TB trial, conducted at 3 hospitals in Cape Town, South Africa, had a 2×2 factorial randomization to standard versus intensified TB treatment (additional rifampicin dosed to 35mg/kg/d plus levofloxacin), and to prednisone (1.5mg/kg/d) versus matching placebo, for the first 14 days, followed by standard TB treatment. Adults admitted to hospital with HIV-associated disseminated TB (positive urine LAM, urine Xpert Ultra and/or blood Xpert Ultra) were eligible. Primary endpoint was 12-week mortality. Absolute difference between arms was estimated by the Kaplan-Meier method with bootstrapping. Intensified TB treatment randomization was stopped early by the DSMB due to higher 2-week mortality in that arm, which we previously reported. We now report the prednisone versus placebo and interaction analyses.
Results
732 participants were enrolled with 712 in the modified intention to treat population: 389 (55%) female, median age 36 years (interquartile range, IQR=30-43), CD4 count 43 cells/μL (IQR=17-96), Hb 8.3 g/dL (IQR=7.0-9.7), venous lactate 2.2 mmol/L (IQR=1.6-3.0), and creatinine 70 μmol/L (IQR=52-99). 83%, 68%, and 31% were positive on urine Xpert Ultra, urine LAM, and blood Xpert Ultra, respectively. Only 75/696 participants (11%) had HIV RNA <50 copies/mL.
At 12 weeks, mortality was 17.6% (95%CI 14.0-22.0) in the prednisone arm and 19.2% (95%CI 15.5-23.7) in the placebo arm (difference -1.6; 95%CI -7.5-4.2; p=0.588) (Figure). 14-day mortality was 6.8% (95%CI 4.6-9.9) versus 9.5% (95%CI 6.9-13.1) in the prednisone and placebo arms, respectively (difference -2.8; 95%CI -6.9-1.3; p=0.185). There was no evidence of interaction between the two concurrent factorial randomizations at 14 days or 12 weeks (interaction p=0.447 and p=0.314, respectively; n=564). Grade 3 (50% vs 52%), Grade 4 (25% vs 30%), and non-fatal serious (20% vs 22%) adverse events occurred at similar frequency in the prednisone and placebo arms, respectively.
Conclusions
Although well tolerated, adjunctive prednisone did not reduce mortality in patients hospitalized with disseminated HIV-associated tuberculosis. Case fatality rates among these patients are high and research aimed at reducing mortality remains a priority.
Table or Figure
