Greater tau pathology is associated with altered predictive coding

Abstract

Altered predictive coding may underlie the reduced auditory mismatch negativity amplitude observed in patients with dementia. We hypothesized that accumulating dementia-associated pathologies, including amyloid and tau, lead to disturbed predictions of our sensory environment. This would manifest as increased reliance on “observed” sensory information with an associated increase in feedforward, and decrease in feedback, signaling. To test this hypothesis, we studied a cross-sectional cohort of participants who underwent positron emission tomography imaging and high-density EEG during an odd-ball paradigm, and used dynamic casual modeling and Bayesian statistics to make inferences about the neuronal architectures (generators) and mechanisms (effective connectivity) underlying the observed auditory evoked responses. Amyloid-β imaging with [C-11] Pittsburgh Compound-B PET were qualitatively rated using established criteria. Tau-positive PET scans, with [F-18]MK6240, were defined by a MK-6240 standardized uptake value ratio positivity threshold at two standard deviations above the mean of the Amyloid(−) group in the entorhinal cortex (entorhinal MK-6240 SUVR > 1.27). The cross-sectional cohort included a total of 56 participants (9 and 13 participants in the Tau(+) and Amyloid(+) subgroups, respectively: age interquartile range of [73.50 - 75.34] and [70.5–75.34] years, 56% and 69% females, respectively; 46 and 43 participants in the Tau(-) and Amyloid(-) subgroups, respectively: age interquartile range of [62.72–72.5] and [62.64–72.48] years, 67% and 65% females, respectively). Mismatch negativity amplitudes were significantly smaller in Tau+ subgroup than Tau- subgroup (cluster statistics corrected for multiple comparisons: p = 0.028). Dynamic causal modelling showed that tau pathology was associated with increased feedforward connectivity and decreased feedback connectivity, with increased excitability of superior temporal gyrus but not inferior frontal regions. This effect on superior temporal gyrus was consistent with the distribution of tau disease on PET in these participants, indicating that the observed differences in mismatch negativity reflect pathological changes evolving in preclinical dementia. Exclusion of participants with diagnosed mild cognitive impairment or dementia did not affect the results. These observational data provide proof of concept that abnormalities in predictive coding may be detected in the preclinical phase of Alzheimer’s disease. This framework also provides a construct to understand how progressive impairments lead to loss of orientation to the sensory world in dementia. Based on our modeling results, plus animal models indicating that Alzheimer’s disease pathologies produce hyperexcitability of higher cortical regions through local disinhibition, mismatch negativity might be a useful monitor to deploy as strategies that target interneuron dysfunction are developed.

Delirium is associated with loss of feedback cortical connectivity

“…Nonetheless the data are consistent with our prior published hypothesis, 5 and with analyses showing diminished functional connectivity in the scalp EEG 7,31 . Similar to our findings with DCM, functional connectivity changes were associated with slow‐wave activity in the resting EEG and plasma cytokines 7,14 . We have found similar changes in a mouse model of systemic inflammation with intraperitoneal lipopolysaccharide 32 .…”

“…7,31 Similar to our findings with DCM, functional connectivity changes were associated with slowwave activity in the resting EEG and plasma cytokines. 7,14 We have found similar changes in a mouse model of systemic inflammation with intraperitoneal lipopolysaccharide. 32 Our current working model is that inflammation-driven prostaglandin release leads to adenosine signaling, 14,32 and activation of GABAergic-interneuron signaling pathways.…”

“…We have found similar changes in a mouse model of systemic inflammation with intraperitoneal lipopolysaccharide 32 . Our current working model is that inflammation‐driven prostaglandin release leads to adenosine signaling, 14,32 and activation of GABAergic‐interneuron signaling pathways 33 . We recently showed that caffeine, an adenosine antagonist, could reverse the EEG effects of inflammation in mice 32 .…”

“…previously to neurocognitive disorders such as schizophrenia and psychotic features, 13 and to detecting abnormalities during the preclinical phase of Alzheimer's disease (AD). 14 Patients with delirium become disoriented in space and time, demonstrating significant impairments of cognitive processing similar to late-stage dementia. Application of a PC framework to delirium may help explain the profound disturbances of cognition, attention, and orientation that characterize this state.…”

“…The PC may explain information processing across a wide range of systems such as sensory, motor, and higher order cognitive networks 12 . The PC model and related analyses have been applied previously to neurocognitive disorders such as schizophrenia and psychotic features, 13 and to detecting abnormalities during the preclinical phase of Alzheimer's disease (AD) 14 …”