Dissertation committee:
Τσάλτα Ελευθερία, Ομότιμη Καθηγήτρια, Ιατρική Σχολή, ΕΚΠΑ
Σμυρνής Νικόλαος, Καθηγητής, Ιατρική Σχολή, ΕΚΠΑ
Ψάρρος Κωνσταντίνος, Αν. Καθηγητής, Ιατρική Σχολή, ΕΚΠΑ
Ζέρβας Ιωάννης, Καθηγητής, Ιατρική Σχολή, ΕΚΠΑ
Δικαίος Δημήτριος, Καθηγητής, Ιατρική Σχολή, ΕΚΠΑ
Τζαβέλλας Ηλίας, Αν. Καθηγητής, Ιατρική Σχολή, ΕΚΠΑ
Φερεντίνος Παναγιώτης, Αν. Καθηγητής, Ιατρική Σχολή, ΕΚΠΑ
Summary:
INTRODUCTION
The impact of Bipolar Disorder (BD) on cognition may be obscured by the potential cognitive effects of its most commonly used pharmacological treatments, mood stabilizers. This possibility is supported by extensive data on the neurotrophic and neuroprotective properties of lithium in particular, as well as more limited data in that direction on alternative mood stabilizers of the anticonvulsant class.
This dissertation examined the effects of BD and chronic lithium administration on general cognitive functioning, with particular emphasis on Cognitive Flexibility (CF). The aim was to disentangle the cognitive effects of BD from those of its commonest pharmacotherapies, specifically of chronic lithium or valproate administration. To that aim, we attempted to bypass the methodological limitations imposed by clinical research ethics by using translational methodology. Specifically, we tested BD patients and healthy laboratory animals with closely analogous neuropsychological procedures assessing CF.
In humans, we examined a wide range of basic cognitive functions by 6 tests of the neuroscientifically informed neuropsychological CANTAB Eclipse battery (Cambridge Automated Neuropsychological Test Automated Battery CANTAB, Cambridge Cognition Ltd.). This provided an inclusive neuropsychological profile of BD patients, which could be compared to pre-existing literature data.
The translational approach was used on the higher-order cognitive function of CF, which is the focus of this dissertation. We used the intra–extradimensional shift CANTAB test (IED) which is closely mirrored by a rat CF experimental analogue. This methodological approach allowed the comparison of a clinical population of BD under chronic mood-stabilizing treatment to normal rats that were treated accordingly. It also allowed comparison of acute and chronic lithium effects, with a distinction of chronic administration in the present and in the past. This methodological approach in the particular context is the major innovative aspect of this dissertation.
METHODOLOGY
Clinical part - Studies 1 and 2a
The clinical studies followed a quasi-experimental design. The original subject pool included 99 euthymic BD patients (mostly type I but also type II), chronically treated either with lithium or valproate and 49 nonpsychiatric controls. All participants met the study inclusion criteria.
The three groups were then matched for demographic and -to the extent that this was possible - psychometric characteristics. After matching, the final sample included 32 nonpsychiatric CONTROLS, 32 BD patients under current, chronic lithium administration (LI-CHRONIC-CURRENT) and 30 BD patients under current, chronic administration of the commonest alternative mood stabilizer, valproate (VPA- CHRONIC-CURRENT).
Cognitive functioning was examined using the following 7 CANTAB tests:
Study 1: (1) psychomotor speed and processing (MOT), (2) visuospatial working memory (SWM), (3) visuospatial learning and memory (PAL), (4) strategy and problem solving (SOC), (5) decision making involving risk-taking (CGT), (6) response inhibition (SST) and
Study 2a: (7) cognitive flexibility (IED).
Experimental part of the dissertation - Study 2b
The study design was fully experimental (as opposed to the quasi-experimental schema of the clinical studies). Sixty-one male Wistar rats were randomly assigned to five pharmacological groups (saline-treated CONTROLS, n=12; LI-CHRONIC-CURRENT, n=11; VPA- CHRONIC-CURRENT, n= 12); in addition to these three groups, which corresponded to the clinical study design, the experimental part of the dissertation allowed the inclusion of two additional groups: a group of chronic lithium administration in the past (LI-CHRONIC-PAST, n=13) and a group of acute lithium administration (LI-ACUTE, n=13).
Cognitive flexibility of the five pharmacological groups was assessed in a set-shifting analogue of CANTAB IED (Birrell and Brown, 2000), which also partially corresponds to the most frequently used human cognitive flexibility test of Wisconsin Card Sorting. Similarly to the CANTAB IED, this animal model demands choices based on two informational dimensions. Unlike the IED, which is based on two visual informational dimensions, the model used odor and texture, sensory cues much more salient to rats than visual ones. As in IED, the readiness of the subjects to shift responses according to changes in the informational value of these two sensory dimensions assesses the ability to shift attentional sets in a series of 7 stages. These stages provide measures of the four abilities essential for cognitive flexibility: (1) simple discrimination learning, (2) discrimination reversal learning, (3) intra-dimensional shift and (4) extra-dimensional shift.
RESULTS
Clinical Study 1: In visuospatial learning (PAL), groups LI-CHRONIC-CURRENT and VPA-CHRONIC-CURRENT performed worse than CONTROLS in the dependent variables of “total errors” and “1st trial memory score”. In working memory (SWM) both the LI-CHRONIC-CURRENT and VPA-CHRONIC-CURRENT groups performed worse than CONTROLS only on the variable of “total errors-between”, and only on performance in the first level of increased difficulty. Both pharmacological groups were also slower than CONTROLS in the “reaction time” variable. It is noteworthy that, despite these deficits of the two clinical groups in these more basic cognitive variables, they actually performed as well as CONTROLS in the “strategy” variable of the test.
In strategy and problem-solving (SOC), the LI-CHRONIC-CURRENT group performed worse than the VPA-CHRONIC-CURRENT group and CONTROLS on the variable of “problems solved in minimum moves”. Further analysis suggested that the deficit of the LI-CHRONIC-CURRENT group was restricted to the most difficult stage of the test.
In a direct test of response inhibition (SST), the LI-CHRONIC-CURRENT group performed worse than the VPA-CHRONIC-CURRENT and CONTROL groups on the variable “direction errors – stop & go”, while further analysis showed that this difference appeared only on “direction errors-go” trials. This indicates that both clinical groups had intact inhibitory ability (directly assessed in “stop-trials”), though the LI-CHRONIC-CURRENT group possibly presented some psychomotor retardation in the “go-trials”.
Finally, in decision-making involving risk-taking (CGT, which also measures impulsivity and inhibitory ability), the LI-CHRONIC-CURRENT group was actually better, though not statistically so, than CONTROLS in the variable of “decision-making quality”. The LI-CHRONIC-CURRENT group was also statistically better than the VPA-CHRONIC-CURRENT group, which did not differ significantly from CONTROLS. On the variable of “risk adjustment” the LI-CHRONIC-CURRENT group did not differ from CONTROLS, while the VPA-CHRONIC-CURRENT group performed significantly worse than CONTROLS. It should be noted that the two clinical groups, LI-CHRONIC-CURRENT and VPA-CHRONIC-CURRENT did not differ in any other variables except the CGT “decision-making quality” variable, in favor of lithium.
Study 2a: None of the five variables of the IED test showed statistical differences between the two clinical pharmacological groups and CONTROLS. This was confirmed in a more detailed, stage-by-stage analysis.
Study 2b: The same picture appeared in the experimental part of the dissertation with the animal IED analogue. No significant differences emerged between the five pharmacological groups in any response count variable. Statistically significant differences emerged only for the response latency variable between the LI-ACUTE group and CONTROLS. This is an indication of psychomotor retardation after acute lithium administration, an effect often reported in the clinic. However, this effect was transient, no longer evident after chronic administration of lithium. It is unfortunate that it was not possible to include a LI-ACUTE group in the clinical part of the studies. The fact that the CANTAB Eclipse-IED does not provide reaction time measurements also made it impossible to explore this effect in the clinical studies.
CONCLUSIONS
Euthymic BD patients with similar demographic and psychometric profiles, under chronic, current administration of either lithium or valproate, showed more frequent and systematic deficits compared to CONTROLS in the more basic cognitive functions (psychomotor speed and processing, visuospatial working memory, visuospatial learning and memory). Given these basic deficits, it would be expected that performance in the more complex, integrative executive functions requiring response inhibition, planning and risk-taking, such as SST and CGT should suffer, but this was not the case.
Even more striking was the finding that CF, which is hierarchically the most integrative executive function, was also found intact in both clinical (BD patients) and normal (healthy animals) groups treated with mood stabilizers. Overall, neither BD patients nor normal laboratory animals exposed to chronic administration of the two basic mood stabilizers differed from CONTROLS. In conclusion, cognitive flexibility was not compromised by either bipolar disorder or by its commonest pharmacological treatments.
Taking into account the limitations of translational comparisons, the compatibility of clinical and experimental findings of this dissertation on cognitive flexibility is remarkable. The combined results lead to the conclusion that cognitive flexibility is preserved despite deficits in basic cognitive functions in the clinical groups. The experimental study confirms that normal cognitive flexibility in euthymic BD patients is not due to a masking of disease-related cognitive deficits by a beneficial cognitive impact of the neurotrophic - neuroprotective actions of lithium, a cancelling-out of disease and drug effects.
The finding stands against the expectations of cognitive decline raised by the obstreperous clinical phenotype of BD. It also fails to support our initial hypothesis that chronic lithium administration may improve cognitive function by means of a prophylactic neuroprotective - neurotrophic action.
Finally, always keeping in mind the limitations in homology and analogy at the neuroanatomical and cognitive/behavioral level, the translational approach can be very useful in disentangling psychiatric illness effects on cognition from the cognitive effects of their pharmacological treatments.
