Your report on tuberculosis (TB) vaccines perpetuates a flawed but widely held view. In fact, the lack of efficacy of the MVA85A vaccine in a recent human clinical trial was no surprise (Nature 502, S8–S9; 2013).

The trial followed an immunization regime previously used in four animal models in the past ten years. Careful examination of those results reveals that MVA85A offered no statistically significant increase in protection over the BCG (Bacillus Calmette–Guérin) vaccine alone in mice, guinea pigs, cows and non-human primates (see, for example, F. A. W. Verreck et al. PLoS ONE 4, e5264; 2009, and S. A. Sharpe et al. Clin. Vaccine Immunol. 17, 1170–1182; 2010).

The only exception is a mouse study (N. P. Goonetilleke et al. J. Immunol. 171, 1602–1609; 2003), which is not comparable because a different immunization route was used (see C. N. Horvath and Z. Xing Adv. Exp. Med. Biol. 783, 267–278; 2013).

In aggregate, therefore, the preclinical animal data for MVA85A predicted the outcome of the reported clinical trial. It remains to be seen how successfully animal models will predict the efficacy of other TB vaccine candidates.