Interpretation of results
We have shown a clear association between the presence of the DBH -1021T allele and AD (Table 4): odds ratio for -1021TT+TC versus CC = 1.2 (1.06-1.4, 0.005), controlling for centre, age, sex and APOE ε4 genotype. This association was nearly restricted to men < 75 years old: 2.2 (1.4-3.3, 0.0004). The interactions with sex and age were both significant (p = 0.01 and 0.03, respectively, Table 2). Table 3 shows that the effect of age was consistent between men and women and the effect of gender was consistent between the two age groups. All these results were consistent between North Europe and North Spain (Tables 2 & 4). We therefore believe these associations to be real. However, large numbers will be needed to replicate these interactions (see the power estimates in Tables 2 & 5).
We also found a probable interaction between the presence of DBH -1021T and IL1A -889TT (Table 5), thus partially replicating Mateo et al 2006 , who reported an interaction between DBH -1021TT and IL1A -889T. The synergy factors were consistent between North Europe and North Spain (Table 5). Also, each risk factor, i.e. DBH -1021T and IL1A -889TT, was only associated with AD risk in the presence of the interacting factor (Table 6), thus indicating epistasis. However, although the results were consistent in the three largest sample-sets, Rotterdam, Santander and OPTIMA, models for the smaller sample-sets proved unreliable. Thus we can only describe this interaction as probable, not definite. The IL1A -889TT genotype has been found to increase transcriptional activity in assays of promoter function [41, 42]. Meta-analyses [43–45] have shown heterogeneity between studies, but a possible, weak association of the -889T allele with AD: odds ratio = 1.07 (0.99-1.16) (23 Sept 2010, 29 sample-sets: http://www.alzgene.org/).
We also found a possible interaction between DBH -1021T and IL6 -174GG, partially replicating that between DBH -1021TT and IL6 -174GG reported by Mateo et al . However, in this case the interaction was only seen in North Europe and the results were inconsistent between the two European regions (Table 5) and between the seven centres. Thus, this apparent interaction may not be real. The only apparently significant results for the other two DBH SNPs studied in our full dataset, exon 3 Ala197Thr (rs5320) and intron 10 A/G (rs1611131), were somewhat inconsistent, precluding any firm conclusions.
The -1021T allele has consistently been associated with strikingly reduced plasma DBH activity [21, 23–27]. The allele partially disrupts consensus transcriptional motifs for n-MYC and MEF-2 . When DBH promoter/reporters were cotransfected with n-MYC or MEF-2, the allele affected the response . The allele is thus functional and, although we cannot assume that it has the same effect in the brain as in the plasma, we may plausibly speculate that it does also have some influence on DBH activity in the brain. DBH catalyses the conversion of dopamine to NA. The -1021C/T SNP may therefore affect levels of both catecholamines. However, although reduced levels of NA are seen in AD brain [8–13], raised levels of dopamine have generally not been found [8, 12, 13]. We will therefore base the discussion below on the hypothesis that the association of the -1021T allele with AD risk is mainly due to an effect on NA levels in the brain.
The control of inflammation in the brain
One likely result of changed DBH activity is misregulation of inflammation in the brain. The mechanisms that control inflammation in the brain differ from those in the periphery; an important part of the former control system is the noradrenergic network (reviewed in ). The anti-inflammatory role of NA has been shown in cultured cells and rodent brains (reviewed in ). Raised levels of NA reduced activation of astrocytes  and microglia [47–49], and lowered expression of pro-inflammatory cytokines [47–50] and chemokines . Noradrenergic depletion increased production of pro-inflammatory cytokines  and chemokines , and activation of astrocytes  and microglia , and impaired microglial phagocytosis of β-amyloid . Astrocytes are considered major targets of noradrenaline in the brain (reviewed in [54, 55]), through their β2-adrenoceptors [46, 54], which activate the cyclic AMP pathway [54, 56], which may lead to the activation of peroxisome proliferator-activated receptors (PPARs) [56–58]. These receptors down-regulate expression of pro-inflammatory genes (PPARγ: ; PPARδ: ). The importance of the cyclic AMP pathway in AD was underlined by the recent identification of the cyclic AMP-response element-binding protein as the transcription factor of most relevance to networks of AD-related genes . The inhibition of the pro-inflammatory transcription factor, nuclear factor κB, by its endogenous inhibitor, IκB, may also mediate the anti-inflammatory effects of NA [62–64]. However, the anti-inflammatory role of NA remains controversial  and it may even have pro-inflammatory actions in certain conditions [65–67]. Nevertheless, the predominant evidence suggests a mainly anti-inflammatory, regulatory role of NA in the brain. This role is weakened in ageing [1–3] and seriously disrupted in AD . Thus, elderly non-demented carriers of the DBH -1021T allele with presumed low activity may be more vulnerable to low-grade inflammation in the brain. This effect has been reported to be stronger in elderly men < 80 years old , consistent with our results.
Other potential mechanisms
In cell cultures and rodent brains, brain-derived neurotrophic factor (BDNF) has been reported: to be induced by NA in astrocytes and neurones [68–71]; to exert certain neuroprotective actions (reviewed in ); and to promote synaptic plasticity and contribute to learning and memory (reviewed in ). BDNF levels have been found to be decreased in the postmortem hippocampus and neocortex [74–76] in AD. A large recent meta-analysis of the BDNF Val66Met polymorphism  found that the Met allele was associated with AD in women, but not men.
Noradrenergic neurones also produce and secrete other neuromodulators and neurotrophins (reviewed in ). These neurones also have roles in glial energy metabolism [54, 55] and the maintenance of the microvasculature [79, 80] and of the blood-brain barrier . NA has actions against oxidative stress [57, 82, 83] and against excitotoxicity [84, 85]. Downstream of NA, the cyclic AMP pathway has neuroprotective and antioxidant actions in neuronal cultures [86, 87]. NA protects against the neurotoxicity of β-amyloid (reviewed in ). However, potentially pathogenic contributions of NA to AD have also been reported [65, 67, 89].