This study identified distinct neuropsychological profiles that distinguished those young people with ‘attenuated syndromes’ from those with a discrete or persistent disorder, independent of other diagnostic considerations. As expected, those in the later stages showed the most impaired neuropsychological profile with the attenuated syndrome patients showing an intermediate profile compared to controls (as well as the standardised ‘norm’). These neuropsychological findings are especially important given the lack of differences between the patient groups in terms of their overall current symptoms and levels of distress. These findings provide further important validation of our clinical staging model, particularly with respect to the notion that the change from stage 1b to stage 2 and 3 represents a ‘key point of differentiation’ (Hickie et al. 2013a).
The findings presented here are consistent with our other studies showing a similar demarcation in both neuroimaging (Lagopoulos et al. 2012) and circadian (Naismith et al. 2012) measures. In the former study, there were frontal grey matter volume differences between the stage 1b and stage 2/3 groups, suggesting a major transition point (Lagopoulos et al. 2012). In the latter study, stage 2/3 patients, but not stage 1b patients, showed a disruption in a circadian system marker (reduced melatonin secretion) which was associated with less subjective sleepiness and poorer performance in a memory task (Naismith et al. 2012). In relation to neuropsychological profiles, there is very little other literature to compare our results to. While numerous studies have described the neuropsychological profiles of prodromal or ‘ultra-high risk’ states for psychosis there are, to our knowledge, no studies that have included young patients with unipolar and/or bipolar illnesses. This may be a critical oversight, given evidence that affective and psychotic disorders probably represent different combinations of the same continuously distributed dimensions of symptoms, particularly at early stages (Hafner et al. 2008). Importantly, our clinical staging model (Hickie et al. 2013a) maintains that there is inherent heterogeneity of cases within each clinical stage and that more detailed profiling (using syndromal, psychological and neurobiological measures) is required to better understand the key underlying factors that cause patients to express a discrete disorder or not (that is, despite being similar in age and current symptomatology).
Our samples are representative of young help-seeking outpatients with admixtures of depressive, (hypo)manic and psychotic symptoms. However, there is some evidence to suggest that those with psychotic spectrum illness show the most marked neuropsychological impairments at various ages (Quraishi and Frangou 2002). Therefore we also examined the neuropsychological profiles of only those with a mood syndrome or disorder and our results confirm that such patients showed a similar overall pattern as the larger sample (with psychosis included), albeit to a lesser degree. Critically, the two key neuropsychological variables (i.e. verbal memory and set-shifting, an aspect of executive functioning) remained significantly different across the clinical-stage groups (and markedly reduced in the stage 2/3 patients). Separate lines of research have shown that cognitive decline in the form of verbal memory and executive function deficits is characteristic of (and often precedes) the early stages of both affective (Burt et al. 1995) and psychotic (Brewer et al. 2005; Seidman et al. 2010) disorders. Similarly, there are several studies showing that impaired neuropsychological function (particularly with regards to memory and executive function) in early stage young patients with mental disorders predicts longer-term poor (typically functional) outcomes (Bodnar et al. 2008; Seidman et al. 2010). Thus, it is becoming increasingly important to identify the best neuropsychological markers for early intervention. This is particularly warranted given that pharmacological (e.g. antidepressant) (Sheline et al. 2003) and non-pharmacological (e.g. cognitive training) (Naismith et al. 2010) strategies may offer neuroprotection against further cognitive damage (Simon et al. 2007).
This study has limitations. Firstly, the cross-sectional design impacts any conclusions about which neuropsychological variables reflect trait versus state aspects of these stages of illness. The presence of some significant associations between the sustained attention or cognitive flexibility measures and current depressive or general psychiatric symptoms (in the stage 1b group) suggests that at least some aspects of executive functioning may be modulated by an individuals state. Clearly longitudinal studies would provide very important information about such trait versus state aspects. Secondly, we did not control for any potential effects of psychotropic medication. Although we opted to assess these young patients under ‘treatment as usual’ conditions, the real impact that such medications have on neuropsychological function is unknown. Given the early stage of illness it is unlikely that the current medications afforded any neuroprotection, but rather offered some amelioration of affective and/or psychotic symptoms. While the clinical-stage groups did not differ in the prevalence of current antidepressant treatment there were differences in the frequency of antipsychotics and mood stabilisers. While there is good evidence to show that the former have very little direct effects on cognition, particularly at early stages in the course of treatment, there is less known about these effects from the latter. Thirdly, the control group in this study were more educated than the patient groups. Despite this, all three groups were matched in their predicted IQ and the standardised scores for each neuropsychological variable were adjusted for age and years of education. Fourthly, while the lack of a significant difference in age among groups was helpful in evaluating the differences in neuropsychological function it may also limit the generalizability of our findings. In our previous study (Scott et al. 2012), utilising a much larger (N = 1260), albeit younger (i.e. 12 to 25 years of age) sample of patients (accessing the same services as those in the current study), we reported age differences among the three stage groups (stage 1b = 17.4 ± 3.4 years; stage 2 = 18.7 ± 3.2 years; stage 3 = 20.3 ± 3.4 years). Given the different age range in the current study (in particular the minimum age of 18 years) these findings may only represent young adults at various stages of illness; future studies should include younger patients (despite the limitations in normative data and valid neuropsychological subtests for younger subjects). Another limitation may be the significant differences among groups in terms of the proportions of females-to-males. Just over two-thirds (62%) of those in the stage 1b group were female, compared to the lower proportion of females (43%) in the stage 2/3 group. These ratios are quite different to those in our larger, younger cohort (Scott et al. 2012) with 47% and 54% females in stage 1b versus stage 2/3, respectively. Although our statistical analyses attempted to control for the effects gender, our findings should be treated with some caution until future studies with larger sample sizes (and presumably more equal proportions of the genders) are conducted. Finally, as highlighted in a recent systematic review (Cosci and Fava 2013) there are numerous variations of staging models for mental disorders. In their distillation of this literature, Cosci and Fava (2013) propose separate models for a range of disorders (including schizophrenia, unipolar depression, bipolar and alcohol use disorders). Thus, it is important to recognise the distinctions between the model investigated in this current study and others in the literature. Comprehensive longitudinal research will help to determine the utility of staging within single disorders (see (Cosci and Fava 2013)) versus staging across a range of syndromes (Hickie et al. 2013; Hickie et al. 2013a).