Peripheral inflammation acutely impairs human spatial memory via actions on medial temporal lobe glucose metabolism

Abstract

Background: Inflammation impairs cognitive performance and is implicated in the progression of neurodegenerative disorders. Rodent studies demonstrated key roles for inflammatory mediators in many processes critical to memory, including long-term potentiation, synaptic plasticity, and neurogenesis. They also demonstrated functional impairment of medial temporal lobe (MTL) structures by systemic inflammation. However, human data to support this position are limited.

Methods: Sequential fluorodeoxyglucose positron emission tomography together with experimentally induced inflammation was used to investigate effects of a systemic inflammatory challenge on human MTL function. Fluorodeoxyglucose positron emission tomography scanning was performed in 20 healthy participants before and after typhoid vaccination and saline control injection. After each scanning session, participants performed a virtual reality spatial memory task analogous to the Morris water maze and a mirror-tracing procedural memory control task.

Results: Fluorodeoxyglucose positron emission tomography data demonstrated an acute reduction in human MTL glucose metabolism after inflammation. The inflammatory challenge also selectively compromised human spatial, but not procedural, memory; this effect that was independent of actions on motivation or psychomotor response. Effects of inflammation on parahippocampal and rhinal glucose metabolism directly mediated actions of inflammation on spatial memory.

Conclusions: These data demonstrate acute sensitivity of human MTL to mild peripheral inflammation, giving rise to associated functional impairment in the form of reduced spatial memory performance. Our findings suggest a mechanism for the observed epidemiologic link between inflammation and risk of age-related cognitive decline and progression of neurodegenerative disorders including Alzheimer's disease.

Keywords: Alzheimer’s disease; PET; imaging; inflammation; memory; parahippocampus.

Figure 1

Study timeline. All participants completed three fluorodeoxyglucose (FDG) positron emission tomography (PET) scans. Each scan was preceded by a blood draw and followed by a memory testing session. The “early” inflammation group received the typhoid vaccination after the first PET scan (and saline injection after the second PET scan), and the “late” inflammation group received the typhoid vaccination after the second PET scan (saline injection after the first scan). In the first two sessions (T1 and T2), participants encoded and then recalled the identity and spatial location of two sets of objects (object set 1 and set 2). In the third session (T3), participants recalled the identity and spatial location of the two sets of objects encoded at T1 and T2. The mirror-tracing task was performed at each of the three testing sessions. MRI, magnetic resonance imaging. (Photo credit, copyright WGBH Educational Foundation.)

Figure 2

Cytokine and memory performance in the early and late inflamed groups. (A) Plasma interleukin-6 levels demonstrating a significantly greater increase in interleukin-6 (from session T1 to T2) in the early compared with late inflammation group. (B) Object location accuracy (proximity score) in units of 1/virtual meters at each encoding session (T1 and T2). Location accuracy decreased 4 hours after typhoid vaccination (early inflammation group) but increased 4 hours after placebo (late inflammation group). (C) Number of objects (maximum 16) correctly recalled during the two encoding sessions (T1 and T2). (D) Mean time taken to complete mirror tracing of a five-pointed star demonstrating a significant improvement across time in both groups. (E) Object location accuracy (proximity score) at the final session (T3) for objects encoded at T1 (set 1) and T2 (set 2). (F) Number of objects (maximum 16) correctly recalled at the final session (T3) for objects encoded at T1 (set 1) and T2 (set 2). Asterisk indicates statistical significance at p < .05. IL-6, interleukin-6.

Figure 3

Brain regions sensitive to acute inflammation and effects on object-location encoding. (A) Regions showing a greater reduction in glucose metabolism after inflammation compared with placebo between sessions 1 and 2. Contrast shown in T1 - T2 early group minus T1 - T2 late group. (B) Regions showing a greater reduction in glucose metabolism after inflammatory challenge in the late compared with early inflamed group between sessions 2 and 3. Contrast shown in T2 and T3 early group compared with T2 and T3 late group. The y axis in (A) and (B) shows estimated glucose metabolism in mL/100 g/min. (C) Regions showing a positive correlation between change in object-location accuracy and change in glucose metabolism between the two encoding sessions (T1 and T2) across all participants. (D) Medial temporal lobe region showing a significant group × location accuracy interaction between the two encoding sessions (T1 and T2). The y axis shows change in glucose metabolism between T1 and T2 in mL/100 g/min. E, early inflammation group (received vaccine after first scan); L, late inflammation group (received vaccine after second scan 2).

Figure 4

Right medial temporal lobe regions sensitive to inflammation, change in object-location accuracy, and interactions with inflammation. Cyan indicates regions showing a reduction in glucose metabolism after inflammation (T1 - T2 early group minus T1 - T2 late group). Yellow indicates regions showing a positive correlation between change in object-location accuracy (T1 and T2) and change in glucose metabolism (T1 and T2) across all participants. Red indicates area showing group × accuracy interaction—that is, the region mediating the effects of group membership (inflammation status) on change in accuracy for encoding object location.

Figure 5

Mediation analysis showing that the changes in parahippocampal glucose metabolism mediate the effects of inflammation on memory for object location. Path coefficients (standard error of path coefficients) are shown for each path of the mediation model.