Adam Roth

Roth 2004 Roth, Adam. 2004. Specific and non-specific effects of BCG - implications for routine immunisation. Bandim Health Project, Division of Epidemiology, Statens Serum Institut. University of Copenhagen.

The aim of the present thesis was to further describe the Bacille Calmette-Guerin (BCG) response and the recently suggested non-specific beneficial effect of BCG on overall mortality, and to investigate the possible implications of this effect for routine immunisation in low-income countries.

In the capital of Guinea-Bissau, a population currently of currently around 57.000 individuals has been followed through a demographic census system for 25 years at Projecto de Saúde Bandim (PSB). Detailed data on socio-economic and health status has been routinely collected and there has been a registry of births and vaccinations at the health centres and the maternal ward of the only hospital in the area. The studies forming the basis of the present thesis were carried out in four different cohorts of children: in a measles vaccine trial in 1996-1998, children were screened for the presence of a BCG scar at 6 months of age (cohort A) (Garly et al, 2003)); children between 3 months and 5 years of age were visited in the home during 1998 and 1999 for BCG scar assessment (cohort B); for children born during 2000 and 2002, a BCG scar and a tuberculin skin test (TST) were assessed at 2, 6 and 12 months of age (cohort C); BCG vaccination status and birth weight were determined for children born at the central hospital during 1989-1999 (cohort D)

One-year overall mortality from scar reading and onwards was described according to scar status for BCG-vaccinated children in cohort B. In cohort C, children with a TST and a BCG scar assessment at 2 and 6 months of age were followed for survival till 18 months of age.

A primary cause of death was determined by verbal autopsies (VA) for the children in cohort A and B who died within the year following scar assessment. The risk of dying from a specific disease was calculated through a cause-specific hazard function.

Determinants of the TST reaction and BCG scar formation at 6 months of age were evaluated in cohort C, and in a subgroup of cohort C the impact of vaccination technique on BCG response was studied in detail.

In cohort D the impact of the current policy of not BCG vaccinating low birth weight (LBW, <2500g) children at birth was evaluated by comparing vaccination coverage, TST reaction and BCG scarring for LBW and normal-birth-weight (NBW, >=2500g) children. The importance of early vaccination of LBW children for one-year overall mortality was also assessed.

A BCG scar and a positive TST reaction were related to better overall survival, and the beneficial effect was stronger for girls. The beneficial effect of having a BCG scar seemed to persist through childhood and was not confounded by exposure to tuberculosis (TB) infection.

The risk of dying from malaria was significantly lower for children with a BCG scar and the effect seemed to be similar, though not significant, for other major causes of death.

The rate of TST reactors after BCG vaccination was low; 26% at 2 months and 25% at 6 months of age, and the TST response to BCG had almost completely waned by 12 months of age. BCG vaccination resulted in a BCG scar for 89% of the children by 6 months of age. The strain of BCG vaccination, the actual injected dose and the monitoring of vaccination technique were of major importance for the BCG response.

BCG coverage for LBW compared to NBW children was lower till 18 months of age due to the policy of not vaccinating LBW children at birth, even though the difference was probably only clinically relevant till 4 months of age. Early BCG vaccination of LBW children was associated with lower mortality the first year of life despite trends of fewer BCG scars and positive TST reactions among LBW children vaccinated early.

Conclusion and future perspectives
The observations that BCG vaccination, BCG scar and a positive TST reaction are related to lower overall infant mortality in low-income countries and that the beneficial effect is sex-specific are confirmed in the present thesis. BCG reduces the risk of dying from malaria, but the effect seems to be non-specific, affecting several major causes of death in childhood in low-income countries. The limitations of the method of verbal autopsy should though be remembered when interpreting these results. The non-specific beneficial effect of BCG on survival is not explained by confounding of TB, nor is it possible to explain the sex-differential effect of BCG by selection bias or confounding from different socio-economic status for boys and girls. The observation that vaccination technique is of major importance for BCG scar and TST reaction makes it unlikely that a good BCG response merely marks immunologically strong children and therefore further supports the non-specific beneficial effect of BCG on survival. The low rate of reactors to TST after BCG is comparable to the rates observed in other African studies using a similar tuberculin. The policy of not BCG-vaccinating LBW children at birth lowers the BCG coverage for LBW children for longer than intended. The slight trend of fewer BCG scars and TST reactions among LBW children with early BCG vaccination probably has little clinical importance considering the poor correlation of BCG scar and TST reaction to BCG efficacy against TB and the lower mortality for LBW children vaccinated early that is observed in this study.

There are several practical implications of these findings; the strain of BCG is of importance for the BCG response; the vaccination technique should be monitored; there are no good reasons for delaying BCG vaccination for LBW children; a more potent tuberculin than 2 T.U. PPD RT23 may be better for monitoring BCG response in low-income countries; children with no BCG scar or a negative TST reaction after BCG vaccination may benefit from an early revaccination. This warrants further studies: a cohort study on whether specific morbidity is affected by having a BCG scar or a positive TST reaction; a randomised trial of different BCG strains and doses with BCG scar, TST reaction, morbidity and overall mortality as outcomes; a randomised trial of early BCG revaccination of non-responders with overall morbidity and mortality as outcomes; a randomised trial of early versus later BCG vaccination for LBW children.

Until these suggested studies have been carried out it would be hasty to suggest any change in immunisation policy, other than emphasizing the importance of a high coverage of BCG, early BCG vaccination and the monitoring of vaccination technique in low-income countries.

The non-specific beneficial effect of BCG on infant survival in low-income countries is well supported in recent studies and in the present thesis. An introduction of a new TB vaccine should therefore be evaluated and compared to BCG with overall mortality as the end point.