Sirs: Most studies of genetic association in sporadic Alzheimer's disease (AD) are actually concerned simply with single candidate gene polymorphisms, rather than in estimating the size of the effect associated with gene-gene interaction. AD brains reveal signs of a robust inflammation [1]. Genes encoding proteins in inflammatory pathways might modulate risk for sporadic AD by increasing production of the protein and provoking a chronic inflammatory response in the brain with consequent neurodegeneration. A polymorphism in the promoter region [-889] of the interleukin-1A (IL-1A) has been shown to confer increased risk for AD [3,7,12], but other studies failed to replicate such an association [5,6,11], perhaps owing to the interaction of genetic modifiers in individual patients. The postulated common pathway of IL-1A and IL-8 in the cerebral inflammatory response [4,8] prompted us to examine the combined contribution of both genes to the susceptibility for AD.The study included 276 AD patients (69 % women; mean age at the time of study 75.6 years; SD 8.9; range 50-98; mean age at onset 71.8 years; SD 8.7; range 48-95) who met NINCDS/ADRDA criteria for probable AD, and 237 control individuals (71 % women; mean age 79.9 years; SD 7.9; range 63-98) who had complete neurological and medical examinations that showed that they were free of significant illness and had Mini Mental State Examination scores of 28 or more. Control subjects were randomly selected from a nursing home. It is often considered that individuals living in nursing homes do not represent the general population of that particular age group but might represent a subset of people; however, the fact that control genotypes were in HardyWeinberg equilibrium does argue against a significant bias being introduced by this method of selection. The AD and control samples were Caucasians originating from a homogeneous population in a limited geographical area in Northern Spain. The IL-1A [-889] and IL-8 [-251] polymorphisms were determined as described previously [9,12]. Association between dichotomous variables was analysed with odds ratio, and 95 % confidence intervals were estimated by the Cornfield method or the Fisher exact method. P-values were estimated by chi-square or Fisher exact tests. Interrelations were analysed by stratification.The distributions of the IL-1A (P = 0.078) and Il-8 (P = 0.997) genotypes were in Hardy-Weinberg equilibrium. The presence of the IL-1A [-889] allele 2 (1/2 and 2/2 genotypes) conferred a marginally significant increase in the risk for the disease (OR 1.42, 95 % CI 1.00-2.02, P = 0.053), and the presence of the IL-8 T/T genotype was not associated with AD (OR 1.33, 95 % CI 0.91-1.94, P = 0.135). However, evaluating the interactive effects of both polymorphisms by stratification (Table 1), the subjects carrying both the IL-1A [-889] allele 2 and the IL-8 [-251] T/T genotype had about twice the risk of developing AD than had subjects without these risk genotypes (OR 2.12, 95 % CI 1.22-3.69, P = 0.007), suggesting a gene-gene ...