The results of the present clinical study show that plasma BDNF levels at 6 months after first hospitalisation for first episode psychosis (when patients have recovered from the acute episode) are positively associated with several cognitive domains. That is, cognitive performance is better in patients with higher BDNF levels. Specifically, we found positive associations between plasma BDNF levels and learning capacity (Logical Memory and Verbal Paired Associates I Learning Curve), verbal delayed memory (Verbal Paired Associates II), abstract verbal reasoning (Similarities) and processing speed (Digit-Symbol Coding) in patients. In contrast, an association between plasma BDNF levels and cognition was not seen in healthy control subjects. Similarly, some studies have found an association between immediate memory and BDNF levels in schizophrenic patients but not in healthy controls
. A possible explanation for our findings is that healthy volunteers and patients with higher BDNF levels may have sufficient cognitive reserve to compensate for other possible deficits present in patients with low BDNF levels; i.e., the brains of individuals with higher BDNF have a greater resistance to damage. Higher cognitive reserve has been shown to have a protective effect against dementia, schizophrenia and depression
[35, 36]. In the New Zealand Dunedin cohort, lower IQ was found to be a risk factor for the development of schizophrenia spectrum disorders
. In our sample, the healthy volunteer group had a higher IQ than the patient group and performed better in almost all of the cognitive tests. In both groups, IQ was positively associated with abstract thinking and processing speed.
Our results that patients with higher BDNF levels have better cognitive performance suggest that plasma BDNF levels could be used as a biological marker of cognition in patients with a first psychotic episode. Cognitive performance is an important clinical variable associated with prognosis in severe mental illness
. The nature of the relationship is unclear but, based on previous studies
, we can hypothesize that lower BDNF functioning in the brain (e.g. during an acute episode of psychosis) can lead to cognitive impairment and could contribute towards the differences in cognition observed between different patients and between patients and healthy subjects. Other investigations have found positive relationships between BDNF levels and cognition in patients with diseases that result in cognitive impairment. One study found an association between maintained increases in serum BDNF levels and improved cognition in patients with schizophrenia, and suggested that serum BDNF levels could be used as a biological marker of cognitive improvement
. Another study found a positive correlation between serum BDNF levels and hippocampal volumes, which could explain the relationship between the BDNF and memory
. Moreover, BDNF levels could change during treatment with some medications. In patients with early Alzheimer’s disease, up-regulation of BDNF with lithium treatment has been associated with increases in cognitive performance
. In addition, some studies in rats have shown that BDNF mediates exercise-induced enhancements in learning and memory
Plasma BDNF levels vary over time in psychosis
 and schizophrenia
, as demonstrated in the present study, where mean BDNF levels were lower during the acute first episode of psychosis and increased to normal levels (comparable to those in healthy subjects) after 6 months of naturalistic treatment when patients were in remission. Plasma BDNF levels have a significant negative correlation with positive symptoms at psychosis onset
. A recent study found that after 6 weeks of antipsychotic treatment, BDNF levels did not increase in patients treated with risperidone, haloperidol or olanzapine, but increased significantly in the subgroup treated with olanzapine
. Another study found a significant increase plasma BDNF levels during the first 6 months of follow-up after olanzapine treatment
. Recently, low serum levels of BDNF were found to be a state marker of depression that normalise during remission
. Our finding of a positive association between BDNF levels and cognition in FEP patients who have recovered from an acute episode of psychosis may be partly due to the effects of pharmacotherapy on BDNF levels in these patients.
In contrast to the variability of BDNF levels, cognitive impairments are more stable in patients with psychosis
. The observed associations between BDNF levels and cognition were significant when patients had recovered from their positive symptoms. These results raise the question of whether plasma BDNF levels may be used a possible marker beyond the acute clinical state.
The results of the multivariate regression models show that plasma BDNF levels are independently associated with several cognitive domains (learning capacity, verbal delayed memory, abstract verbal reasoning and processing speed); that is, patients with lower BDNF levels have a worse cognitive performance independent of premorbid IQ. However, our study shows that IQ is also an important independent factor associated with some cognitive domains (processing speed and delayed memory), as has been reported previously
[44, 45]. Cognitive performance has also been associated with negative symptoms
. We found that negative symptoms were inversely related to the Similarities subtest of the WAIS. However, the association was no longer significant when PANSS negative symptoms were introduced in the multivariate regression model together with BDNF and IQ.
This study has some limitations. The patients received pharmacological treatment in a naturalistic setting but we cannot analyse the possible differential effects of antipsychotics on BDNF because the sample is not large enough. In addition, other biological factors, such as oxidative stress
 and environmental factors not measured in this study, could be associated with cognition. The small sample size can limit the statistical power. Nevertheless, the sample was large enough to suggest that BDNF and cognitive performance are associated, and that plasma BDNF levels after clinical stabilisation could be used as a biological marker of cognition in psychotic disorders.