Dysregulation of CO2-Driven Heart-Rate Chemoreflex Is Related Closely to Impaired CO2 Dynamic Vasomotor Reactivity in Mild Cognitive Impairment Patients

Figures

• 

  Fig. 1. Illustrative time-series data over 5 min for one control subject, representing beat-to-beat spontaneous variations of CFV (top panel), TOI (2nd panel), ABP (3rd panel), HR (4th panel), and CO2 (bottom panel), before (left column) and after (right column) pre-processing. The units are: cm/s for CFV, % for TOI, beats/min for HR, and mmHg for ABP and CO2.

• 

  Fig. 2. Average kernel estimates for the CO2 input (±1 SD bounds marked with dotted lines) over 18 or 17 CS (blue line) and 45 or 43 MP (red line), for CFV (left) or TOI (right) output, respectively.

• 

  Fig. 3. Average model-predicted response to a unit-step CO2 input over 18 CS (blue line) and 45 MP (red line) for the CFV output (left panel), and over 17 CS (blue line) and 43 MP (red line) for the TOI output (right panel). The SD bounds are also shown with dashed lines.

• 

  Fig. 4. Average kernel estimates (±1 SD bounds marked with dotted lines) over 18 CS (blue line) and 45 MP (red line) of the predictive dynamic model for the HR output and the ABP input (left panel) or the CO2 input (right panel). These two sets of kernels describe the baroreflex and chemoreflex dynamics, respectively. The baroreflex kernels are similar for CS versus MP, but the chemoreflex kernels are distinct.

• 

  Fig. 5. The average model-predicted HR responses (in beats/min) to a unit-step ABP input (left panel) and CO2 input (right panel) over 18 CS (blue line) and 45 MP (red line) that illustrate the average time-course of the baroreflex and CO2-driven heart-rate chemoreflex effects, respectively. The CRG index for each CS or MP is defined as the time-average of the shaded area of the HR unit-response (indicated on the right panel). The BRG index for each CS or MP is defined as the difference between the peak and the trough of the HR unit-response (shown on the left panel for the average kernels).

• 

  Fig. 6. The gain of the Transfer Functions (Fourier Transform magnitudes) of the kernels for CO2 input and CFV output (left), TOI output (middle) and HR output (right). Notable differences are seen in the low-frequencies between MP and CS, as well as in the location of resonant peaks.

• 

  Fig. 7. Regression lines of the scatter-plots of DVR-CFV versus CRG (left panel) and DVR-TOI versus CRG (right panel), for all MP (red stars) and CS (blue circles). The indices DVR-CFV and CRG have significant correlation (p = 0.0146), but the indices DVR-TOI and CRG do not (p = 0.1066). The Fisher Discriminants are also plotted as green lines (see text).

• 

  Fig. 8. Scatter-plots and regression lines for Average-TOI values versus CRG indices (right) or DVR-TOI indices (left) for 17 CS (blue circles) and 43 MP (red stars) taken together. The Average-TOI values correlate strongly with the CRG indices but not with the DVR-TOI indices.

• 

  Fig. 9. Scatter-plots for Delayed Logical Memory Recall (DLMR) scores versus DVR-CFV index (left) and DVR-TOI index (right) with the regression lines (dotted black) indicating significant correlation between DLMR and both DVR indices (p = 0.0011 for DVR-CFV and p = 0.0088 for DVR-TOI). The Fisher Discriminants are also shown as dashed green lines and suggest the potential use of the Composite Indices: [DLMR + 0.51×DVR-CFV] and [DLMR + 8.33×DVR-TOI] to achieve improved delineation between MP and CS (p∼10–10).