Relationship between Hypotension and Cerebral Ischemia during Hemodialysis

Abstract

The relationship between BP and downstream ischemia during hemodialysis has not been characterized. We studied the dynamic relationship between BP, real-time symptoms, and cerebral oxygenation during hemodialysis, using continuous BP and cerebral oxygenation measurements prospectively gathered from 635 real-world hemodialysis sessions in 58 prevalent patients. We examined the relationship between BP and cerebral ischemia (relative drop in cerebral saturation >15%) and explored the lower limit of cerebral autoregulation at patient and population levels. Furthermore, we estimated intradialytic exposure to cerebral ischemia and hypotension for each patient, and entered these values into multivariate models predicting change in cognitive function. In all, 23.5% of hemodialysis sessions featured cerebral ischemia; 31.9% of these events were symptomatic. Episodes of hypotension were common, with mean arterial pressure falling by a median of 22 mmHg (interquartile range, 14.3-31.9 mmHg) and dropping below 60 mmHg in 24% of sessions. Every 10 mmHg drop from baseline in mean arterial pressure associated with a 3% increase in ischemic events (P<0.001), and the incidence of ischemic events rose rapidly below an absolute mean arterial pressure of 60 mmHg. Overall, however, BP poorly predicted downstream ischemia. The lower limit of cerebral autoregulation varied substantially (mean 74.1 mmHg, SD 17.6 mmHg). Intradialytic cerebral ischemia, but not hypotension, correlated with decreased executive cognitive function at 12 months (P=0.03). This pilot study demonstrates that intradialytic cerebral ischemia occurs frequently, is not easily predicted from BP, and may be clinically significant.

Keywords: cardiovascular physiology; cognitive decline; hemodialysis; intra-dialytic hypotension; ischemia.

Figure 1.

Sensitivity/true positive rate (solid line) and specificity/true negative rate (broken line) of change in MAP from baseline for predicting the onset of cerebral ischemia in patients receiving hemodialysis.

Figure 2.

Sensitivity/true positive rate (solid line) and specificity/true negative rate (broken line) of absolute MAP thresholds for predicting the onset of cerebral ischemia in patients receiving hemodialysis.

Figure 3.

Relationship between MAP and cerebral oxygen saturations at the population level, approximating to an “average” autoregulation curve. The data for all patients was pooled, and BP data were sorted into 5-mmHg-wide bins sliding every 1 mmHg, e.g., 60–65, 61–66, 62–67 mmHg. Each BP data point was associated with a cerebral oxygen saturation. The mean and 95% limits of cerebral oxygen saturation data for each 5-mmHg bin were calculated, and plotted against the midpoint for that bin, e.g., the 60–65 mmHg bin is represented by the point 62.5 mmHg on the x axis.