Dietary cured meat and the risk of adult glioma: a meta-analysis of nine observational studies.

Abstract

N-nitroso compounds (NOCs) are recognized neural carcinogens in animal models and are suspected human carcinogens. A meta-analysis was performed examining the possible association of dietary N-nitroso intake from cured meats and the risk of gliomas in adults. A prospective protocol was developed outlining the intent, methods, and statistical analysis of the meta-analysis. Data from nine epidemiological studies were pooled using a general variance-based meta-analytic method employing confidence intervals as described by Greenland. The outcome of interest was a summary relative risk (SRR) reflecting the risk of brain tumor (glioma) development associated with high dietary intake of cured meats (as defined by individual study investigators). Sensitivity analyses were performed when necessary to explain any observed statistical heterogeneity. Nine observational studies met protocol-specified inclusion criteria. Analysis for heterogeneity demonstrated a lack of statistical heterogeneity (p = 0.58). Pooling the data on dietary cured meat intake of all types yielded an RR of 1.48 (1.20-1.83), suggesting a 48% increased risk of glioma development among adults ingesting high levels of cured meat. Analyzing brain tumor risk by meat type yielded an RR of 0.90 (0.63-1.25) for hotdogs (a nonstatistically significant result), 1.31 (1.00-1.71) for bacon, and 1.64 (1.27-2.14) for ham. Sensitivity analyses showed that the failure of most studies to adjust for total energy intake might lead to a spurious positive association between cured meat intake and brain tumor risk. Insufficient data were available for analyzing dose-response relationships, although a few individual studies showed evidence of increasing risk with increasing cured meat intake. The available data do not provide clear support for the suspected causal association between ingestion of NOCs from cured meat in adults and subsequent brain tumor risk. Uncontrolled confounding may account for the previously noted positive association seen in some epidemiological studies.