Cumulative Exposure to Frequent Intradialytic Hypotension Associates With New-Onset Dementia Among Elderly Hemodialysis Patients

To the Editor:

Cognitive impairment, including dementia, affects up to 70% of individuals receiving maintenance hemodialysis,¹ a prevalence that exceeds that of the elderly general population by 3- to 7-fold.² Cardiometabolic risk factors for dementia, such as diabetes, hypercholesterolemia, hypertension, and vascular disease, are highly prevalent in end-stage kidney disease.³ However, the rate of new-onset dementia is higher among hemodialysis patients compared with peritoneal dialysis patients,⁴ suggesting that aspects of the hemodialysis procedure may contribute to the development of dementia.

Imaging studies show that hemodialysis can cause significant circulatory stress via ultrafiltration-induced hypoperfusion of vital vascular beds, including the brain.⁵ Maintenance of adequate cerebral perfusion during dialysis is dependent on intradialytic blood pressure and the brain’s intrinsic ability to preserve relatively constant blood flow despite changes in perfusion pressure. Hemodialysis patients may be particularly susceptible to the neurologic consequences of dialysis-induced blood pressure declines due, in part, to diminished autoregulatory capacity from autonomic and endothelial dysfunction.6, 7 It is thus plausible that repeated exposure to intradialytic hypotension (IDH) and associated cerebral hypoperfusion may increase dementia risk. In fact, general population data indicate that orthostatic hypotension⁸ as well as cerebral hypoperfusion⁹ are associated with accelerated cognitive decline and a higher risk of new-onset dementia. Although a recent study found an association between IDH and reversible cognitive decline in the immediate post-dialysis period,¹⁰ the relationship between long-term, cumulative IDH exposure and the development of dementia in the hemodialysis population has not been established.

The objective of our study was to examine the association between the cumulative exposure to frequent IDH and the 5-year risk of new-onset dementia among elderly individuals initiating maintenance hemodialysis at a large U.S. dialysis organization. After a 30-day lag period following hemodialysis initiation, we determined if patients experienced frequent IDH in successive 90-day exposure intervals (Supplementary Figure S1). Within a given exposure interval, we classified an individual as having frequent IDH if he or she experienced a nadir intradialytic systolic blood pressure <90 mm Hg in at least 30% of hemodialysis treatments. This IDH definition has been associated with all-cause mortality, an important competing event in our analysis.¹¹ We evaluated the association between the time-updated, cumulative number of exposure intervals with frequent IDH after dialysis initiation and new-onset dementia using marginal structural Fine and Gray proportional subdistribution hazards models. All-cause death was treated as a competing event.

A total of 31,055 individuals initiated maintenance hemodialysis from June 1, 2005, to October 1, 2013, and met study selection criteria (Supplementary Table S1). The study cohort had an average age of 76.0 ± 6.6 years, 46.2% were women, 19.9% were black, 8.7% were Hispanic, and the most common cause of end-stage kidney disease was diabetes (43.7%). Baseline cardiovascular comorbid conditions were common: 39.9% of the cohort had an arrhythmia or conduction disorder, 59.2% had heart failure, 56.0% had ischemic heart disease, 29.1% had peripheral arterial disease and 18.8% had a history of stroke (Table 1 and Supplementary Table S2).

Table 1.

Select characteristics of the study cohort at baseline

Characteristic	Total study population (N = 31,055)
Age at dialysis initiation, yr	76.0 ± 6.6
Female	14,359 (46.2)
Race
Black	6179 (19.9)
White	23,356 (75.2)
Other	1520 (4.9)
Hispanic	2705 (8.7)
Cause of end-stage kidney disease
Diabetes	13,565 (43.7)
Hypertension	11,636 (37.5)
Glomerular disease	1742 (5.6)
Other	4112 (13.2)
Current smoker at dialysis initiation	1090 (3.5)
Inpatient dialysis initiation	13,054 (42.0)
Vascular access
Catheter	22,398 (72.1)
Fistula	6580 (21.2)
Graft	2077 (6.7)
Arrhythmia or conduction disorder	12,393 (39.9)
Diabetes	20,379 (65.6)
Dyslipidemia	9972 (32.1)
Heart failure	18,370 (59.2)
Ischemic heart disease	17,398 (56.0)
Peripheral arterial disease	9046 (29.1)
Stroke	5834 (18.8)
Valvular disease	2464 (7.9)
Pre-dialysis systolic blood pressure, mm Hg
≤130	8475 (27.3)
131–150	12,240 (39.4)
151–170	7736 (24.9)
≥171	2604 (8.4)

Values are given as number (percent) for categorical variables and as mean ± SD for continuous variables. Baseline covariates, including demographics, comorbid conditions, and health care utilization metrics were obtained in the 365 days preceding dialysis initiation. Baseline laboratory and dialysis treatment-related covariates were obtained in the 30 days immediately after dialysis initiation. The complete list of study cohort baseline characteristics is displayed in Supplementary Table S2.

The cohort was followed for a total of 64,982 person-years and had an average follow-up duration of 2.1 ± 1.6 years. During the 5-year follow-up period, 4991 individuals developed all-cause dementia (incidence rate = 7.7 cases of new-onset dementia/100 person-years) and 11,037 individuals died (incidence rate = 17.0 deaths/100 person-years). Greater cumulative exposure to frequent IDH after hemodialysis initiation was incrementally associated with a higher 5-year risk of new-onset dementia (Figure 1). Individuals who experienced frequent IDH in ≥ 7 (vs. zero) 90-day exposure intervals across time had the highest 5-year risk of new-onset dementia (hazard ratio [95% confidence interval] = 1.36 [1.20–1.48]).

Figure 1.

Association between time-updated, cumulative exposure to frequent intradialytic hypotension and the 5-year risk of new-onset dementia. We used marginal structural Fine and Gray proportional subdistribution hazard models, treating death as a competing event, to estimate association between the time-updated, cumulative exposure to frequent intradialytic hypotension and the 5-year risk of new-onset dementia. Within a given 90-day exposure interval, we classified an individual as having frequent intradialytic hypotension if he or she experienced a nadir intradialytic systolic blood pressure <90 mm Hg in at least 30% of hemodialysis treatments. Supplementary Table S3 contains outcome and competing event definitions. We used inverse probability of exposure weighting to adjust for baseline and time-updated covariates listed in Supplementary Table S4. CI, confidence interval; HR, hazard ratio.

We provide initial evidence linking cumulative IDH exposure to new-onset dementia among individuals receiving maintenance hemodialysis. Prior mechanistic studies have shown that hemodialysis-induced circulatory stress may contribute to the development of cerebral ischemic injury, but these studies were not powered to consider clinical outcomes. For example, in a cross-sectional study of 12 elderly hemodialysis patients from the Netherlands, Polinder-Bos et al.⁵ demonstrated that hemodialysis induces a significant reduction in global and regional cerebral blood flow. In a pilot study of 58 prevalent hemodialysis patients from the United Kingdom, MacEwen et al.¹² found that more pronounced declines in intradialytic mean arterial pressure associated with a higher incidence of intradialytic cerebral ischemic episodes. Furthermore, in a prospective cohort study of 32 nondiabetic, Japanese hemodialysis patients, Mizumasa et al.¹³ noted that the number of IDH episodes (defined as a fall in systolic blood pressure >50 mm Hg within 30 minutes of starting hemodialysis plus associated symptoms of hypoperfusion) across time was weakly correlated with greater cerebral frontal lobe atrophy.

Our findings extend this mechanistic evidence by linking frequent IDH to the clinical outcome of new-onset dementia. Dementia is associated with a range of adverse outcomes, including lower quality of life¹⁴ and treatment nonadherence,¹⁵ as well as higher hospitalization and mortality rates.16, 17 Therefore, identification of effective interventions that reduce dementia risk is needed to improve patient outcomes. One promising hemodialysis-based intervention may be the use of cooled dialysate. Cooled dialysate reduces intradialytic hemodynamic instability,¹⁸ likely via cold-induced vasoconstriction and associated improvements in systemic vascular resistance during ultrafiltration. In a randomized controlled trial of 73 incident hemodialysis patients from the United Kingdom, Eldehni et al.¹⁹ found that hemodialysis with cooled dialysate (0.5°C below core body temperature) versus standard temperature dialysate (37°C) led to preservation of brain white matter microstructure at 1 year. This relatively low-risk intervention may be a viable, low-cost neuroprotective treatment strategy for individuals initiating hemodialysis. Other potential IDH reduction strategies include treatment time extension, more frequent dialysis, and ultrafiltration profiling, but supporting data, particularly with regard to the latter, are limited.

In conclusion, we found that increased cumulative exposure to frequent IDH after dialysis initiation was incrementally associated with a higher 5-year risk of new-onset dementia in a cohort of more than 30,000 elderly hemodialysis patients. Interventional studies are needed to determine if IDH mitigation reduces dementia risk among individuals receiving maintenance hemodialysis.

Disclosure

MMA and JEF received investigator-initiated research funding from the Renal Research Institute, a subsidiary of Fresenius Medical Care, North America. JEF received speaking honoraria from American Renal Associates, American Society of Nephrology, Dialysis Clinic, Incorporated, National Kidney Foundation, and multiple universities; consulting fees from Fresenius Medical Care, North America; and serves on the medical advisory board to NxStage Medical. LW declared no competing interests.

Supplementary Material

Figure S1

Study design.

Table S1

Study inclusion and exclusion criteria.

Table S2

Complete list of study cohort baseline characteristics.

Table S3

Outcome and competing event definitions.

Table S4

Baseline and time-updated covariates.