We conclude that the ESRα rs2234693 and rs9340799 polymorphisms do not substantially contribute to susceptibility of schizophrenia, although a modest association was detected in males only for rs9340799. The results are nearly consistent with most previous studies. This difference may be due to these reasons: on one hand, the subjects in the study of Weickert were African American  only, while others were East Asians [28, 34] or southern Europeans . Susceptibility to the disease and distribution of the genotype for rs2234693 were distinctly different in these subjects (data was from International HapMap Project). On the other hand, the sample size and test power between these studies were difference. Interestingly, the haplotypes in the gene may increase the risk of schizophrenia. C-A rarely emerges with the extremely high frequency of approximately 25%, and it may contribute to schizophrenia. It is noteworthy that C-G may play the role of protective haplotype, especially in females. It is also meaningful that T-G and C-G may act only on females, not males.
The age of onset has been considered the single most valuable characteristic of schizophrenia that may yield a clue to its origin . A genomewide linkage study by Cardno et al.  has confirmed a genetic contribution to the age at onset of schizophrenia. This was the first time that significant associations of ESRα polymorphisms with age at onset of schizophrenia have been demonstrated. But in the study by Ouyang et al. , this association was not clearly verified. This inconsistency might be due to differences in population and in the definition of age at onset. In the present study, it is interesting that the T-allele of rs2234693 may relate to earlier pathogenesis of schizophrenia in all and female patients, not in males independently. Evidence that female schizophrenics are more easily influenced by inheritance than males may contribute to the gender difference .
Although many genes do not alter susceptibility to schizophrenia, they may affect clinical features [37, 38]. A study by Alonso et al.  verified that rs34535804 and five SNP haplotypes in ESRα were associated only with the psychopathic symptoms contamination obsessions and cleaning compulsions. In our study, we hypothesize that all, male, and paranoid schizophrenic patients with the T-allele had generally poor impulse control, as well as both tension and poor impulse control symptoms, respectively, suggesting an association of these psychopathic symptoms with rs2234693 of ESRα in special populations. In the therapeutic effect analysis, female and paranoid patients with the CC homozygote had a superior therapeutic effect in general psychopathology symptom after 6 weeks. Notably, we found that aripriazole was more effective in treating tension symptoms in patients with the CC homozygote, while risperidone controlled poor impulse symptoms, again, in patients having the CC homozygote. These findings suggest that T-allele protects against the onset of these symptoms, especially in general psychopathology, tension, and poor-impulse control psychopathic symptoms. Furthermore, CC-homozygote carriers seem to be more sensitive to antipsychotic treatment. It is also worth noting that two polymorphisms are closely located on the ESRα gene, but this study has demonstrated that rs2234693 is not linked in disequilibrium with rs9340799 in patients with schizophrenia. This explains why only rs2234693 without rs9340799 is associated with clinical characteristics of schizophrenia.
ESRα is markedly different from other neurotransmitter receptors in that it may influence mood, cognition, and synaptogenesis and be involved in neuroprotective effects [9, 39]. ESRα, as a ligand-dependent transcription factor, is widely distributed in the amygdala-hippocampal area, periamygdaloid cortex, and posterior cortical nucleus of the brain , especially in several limbic structures, suggesting its relationship with the above-mentioned processes . Furthermore, in the CNS, ESRα binds its specific ligand estrogen to influence various neurotransmitter systems such as dopamine, serotonin (5-HT), and norepinephrine. These phenomena may be responsible for the relation between ESRα and many psychiatric disorders such as schizophrenia . A previous review by Herrington et al.  has summarized the associations of ESRα polymorphisms with mood and cognition dysfunction, suggesting a strong relationship between them. We conclude that the association mechanism of ESRα variants with psychiatric characteristics may involve the estrogen-estrogen receptor effect, above. Whether our results can explain this mechanism needs further study.
In this study, we found a significant association of ESRα polymorphism with age at onset, general psychopathology symptoms, and, for the first time, the therapeutic effects in schizophrenia, but limitations also existed. Firstly, as only two polymorphisms were selected for analysis, they do not represent the whole gamut of the human ESRα, further studies should be done to analyze more variants. Secondly, not all the patients were included in the association analysis of clinical variables due to the finite clinical data. Thirdly, because the structured clinical interviews were difficult and time-consuming, our sample size was inadequate and could easily have led to false positive or negative results. Fourthly, although stratified analyses were performed, more subtle and stratified analyses should be used in further studies. Finally, whether ESRα variants can be considered as genetic markers to determine the risk or prognosis of schizophrenia needs further exploration.