Discussion

We report a significant positive relationship between phosphene threshold and BOLD activity in early visual areas (V1, V2/V3) in our controls, which suggests that the greater an individual's inherent visual cortical excitability, the smaller the visual cortical activation in response to a simple checkerboard stimulus. This observation implies that individuals with sensitive visual systems need less ‘activation’ to an external stimulus. In contrast, the opposite relationship in DLB infers a breakdown in this dynamic. Assuming similar visual cortical neuronal populations are activated by the checkerboard and TMS, then in the controls a positive relationship between BOLD activity and phosphene threshold implies inhibition must outweigh excitation, whereas in DLB this is reversed. If the BOLD signal represents spiking excitatory neuronal activity, ^{Reference Lee, Durand, Gradinaru, Zhang, Goshen and Kim4} then the lack of difference between groups in mean BOLD activity suggests a loss of inhibition.

There needs to be critical consideration to our interpretations; our knowledge of the underlying neuronal activity of the BOLD signal is incomplete ^{Reference Logothetis3,Reference Logothetis7} and the BOLD effect reflects not just neuronal activity but also haemodynamic coupling, which may be altered with neurodegeneration; ideally cerebrovascular reactivity would be assessed via a CO₂ challenge ^{Reference Cantin, Villien, Moreaud, Tropres, Keignart and Chipon8} or similar paradigm. However, we did not observe any differences in occipital perfusion between the control and DLB group, and similarly, there was no relationship between perfusion and either BOLD response or phosphene threshold. Furthermore, cognitive function in DLB was not associated with phosphene threshold or any of the neuroimaging measures. Thus, although not providing a direct measure of cerebrovascular reactivity, these findings suggest that the early visual system in these participants with DLB was similar to controls.

From the TMS perspective, the origin of phosphene perception may not be co-localised to the visual cortex and thus the relative state of the visual cortex may be less crucial to phosphene perception, although reciprocal connections between early visual and association areas are likely to be important. ^{Reference Taylor, Walsh and Eimer9} In addition, cholinergic function has been implicated in modulating visual attention and associated BOLD fMRI responses; ^{Reference Ricciardi, Handjaras, Bernardi, Pietrini and Furey10} our observation of differences in BOLD activity between those patients with DLB taking cholinesterase inhibitors v. those not, supports this and fits into the broader framework of reported cholinergic receptor changes in DLB ^{Reference O'Brien, Colloby, Pakrasi, Perry, Pimlott and Wyper11} as well as changes in occipital metabolism in individuals with DLB with hallucination amelioration treated with donepezil. ^{Reference Satoh, Ishikawa, Meguro, Kasuya, Ishii and Yamaguchi12}

Finally, a loss of inhibitory activity in patients might be expected to decrease phosphene threshold relative to the control group, something which was not observed. However, it is notable that for a range of phosphene thresholds, the overall BOLD response in the DLB group was larger than in the control group, significantly so for V1 after excluding TMS non-responders (whose relative level of visual cortical excitability is unclear). Hence, one could speculate that compensatory mechanisms may be operating in patients with DLB that attempt to ‘normalise’ the balance of excitation/ inhibition in an underactive early visual system, in the face of ‘upstream’ pathology in the visual system thus leading to a maintenance of the phosphene threshold within a normative range and possibly improved bottom-up visual processing, but at the cost of increasing the risk of visual hallucination occurrence, as evidenced by our previous work linking phosphene threshold with visual hallucination frequency and severity. ^{Reference Taylor, Firbank, Barnett, Pearce, Livingstone and Mosimann1}

In summary, our findings of a double dissociation suggest a marked deviance in visual neuronal processing between patients with DLB who hallucinate v. healthy controls. Whether this represents a specific loss of inhibitory drive in the visual cortex of people with DLB and what role cholinergic function plays remains to be clarified; neurochemical post-mortem studies or novel spectroscopic imaging assessing for example gamma-aminobutyric acid-ergic tone may be helpful in this regard.