Blood Pressure Variability and Dialysis: Variability May Not Always Be the Spice of Life

In 1785, the great English poet William Cowper composed the famous lines, “Variety’s the very spice of life, That gives it all its flavour.” While Cowper was referring to the physical world, his sentiment also captures the importance of variety and variability in numerous biologic processes. Genetic variability facilitates evolution through the differential survival of the fittest. Physiologic process variability takes on many beneficial forms, including cyclical neurohormonal release, respiratory sinus variation, and nocturnal BP dipping. Cyclic variability in these evolutionary and biologic processes facilitates health. Not all variability, however, confers advantage. A growing body of evidence suggests that BP variability (BPV) portends worse outcomes in both the general and kidney disease populations.

Numerous studies in the nondialysis population have identified BPV as a risk factor for the progression of CKD,¹ stroke,² and death.^2,3 The mechanistic underpinnings of these associations are not fully understood but probably hinge on BPV-induced end-organ damage that may be driven, in part, by arterial system changes.⁴ The finding that calcium-channel blockers (CCBs) ameliorate BPV in nondialysis populations lends some credence to the arterial system’s central role in BPV because CCBs induce vasodilation and are associated with arterial structural remodeling.⁴

BPV is also a compelling risk factor among patients undergoing hemodialysis (HD). Given the high burden of vascular disease in HD patients and the nonphysiologic fluid and osmolar shifts inherent to the prevailing pattern of intermittent thrice-weekly HD, it is not surprising that previous studies have shown associations between BP phenomena and adverse outcomes among HD patients. For example, the absence of nocturnal dipping,⁵ BP rise over the course of dialysis,⁶ and visit-to-visit pre-HD BPV^7–9 have all been linked to increased death risk. In a recent analysis, Chang et al. demonstrated an association between visit-to-visit pre-HD BPV and death across all pre-HD BP strata, but surprisingly did not find a significant association with cardiovascular death.⁹ Flythe et al. examined BP fluctuations during dialysis (i.e., intradialytic BPV) and reported a link between intradialytic BPV and all-cause and cardiovascular death.¹⁰ In a prior study, the same authors identified older age, shorter dialysis vintage, and greater ultrafiltration (rate and volume) as associated with greater intradialytic BPV.¹¹ Contrary to findings in the general population, neither Chang et al. nor Flythe et al. found antihypertensive agents to play a significant role in visit-to-visit or intradialytic BPV–outcome associations among HD patients.^9,11

In this issue of JASN, Shafi et al. present further observational data linking visit-to-visit BPV and all-cause and cardiovascular morbidity and mortality.¹² In a contemporary cohort of 11,291 incident HD patients from a single dialysis organization, the authors demonstrate an association between greater predialysis systolic BPV and higher body mass index, higher calcium-phosphate product levels, and lower hemoglobin concentrations. Nonmodifiable factors, including female sex, black race, diabetes, and comorbid cardiovascular disease, were also associated with greater BPV. Most important, lower visit-to-visit BPV was associated with greater ultrafiltration volume, dry weight attainment, and antihypertensive regimens without β-blockers or renin-angiotensin system (RAS)–blocking agents. All associations persisted across strata of pre-HD BP. The authors’ findings extend the existing BPV knowledge base by shedding light on the potentially protective role of non–blocker-based and non–RAS-blocking agent–based antihypertensive regimens and by elucidating yet another reason to focus on fluid removal and the attainment of dry weight.

The strengths of Shafi and colleagues’ study include its large sample; inclusion of multiple important clinical, pharmacologic, and dialysis-related patient characteristics that were absent in previous analyses; and the selection of a statistically robust BPV metric. As the BPV literature has expanded, investigators have used a range of metrics to describe BP fluctuations, including SD, SD independent of the mean, average real variability, coefficient of variation, and residuals derived from generalized linear models. The BPV metric must adequately reflect BP fluctuations without overinfluence from ambient BP levels, such that the independent effects of each can be distinguished. Independence of the BPV metric from the pre-HD BP measurement is of critical importance given the strong association of pre-HD BP and outcomes.¹³ To accomplish this, Shafi et al. fit a mixed linear effects model to the natural log of a patient’s pre-HD systolic BP measurements over time and defined BPV as the SD of the model’s residuals. Differences in metric definitions between Shafi and Chang and their colleagues’ investigations of visit-to-visit BPV may have led to the discrepant cardiovascular mortality findings across the two studies. Chang et al. defined BPV using the coefficient of variation, the ratio of the SD to the mean.⁹ This method also measures BPV independent of the pre-HD SBP, but it is derived from a mean BP value and thus does not account well for individual visit-to-visit fluctuations and outlying data points. Such detail is better reflected in Shafi and coworkers’ non–mean-based, mixed linear model approach. Shafi and associates’ rigorous statistical approach lends strong credibility to the reported outcome associations, and the authors should be commended for their efforts.

Although Shafi and colleagues’ findings add to the growing body of evidence that pre-HD BP fluctuations are poor prognostic indicators, several important limitations must be considered in interpreting the results. BPV is a dynamic phenomenon that is influenced by interactions among environmental, humoral, and neural factors. Confounders of these interrelated processes must be carefully considered in interpreting outcome associations. Although the authors included many dialysis-related confounders in the adjusted models, they did not include measures of sodium balance or objective measures of volume status. A positive dialysate-to-serum sodium gradient has been associated with greater interdialytic weight gain and predialysis BP;¹⁴ its effects on BPV per se have not been reported. Additionally, sodium loading may directly induce higher BP via nitric oxide inhibition and associated endothelin-1 rise.¹⁵ From Shafi and colleagues’ analysis, we cannot assess associations between BPV and outcomes independent of sodium balance. Improved understanding of the role of sodium balance in BPV could potentially shed light on modifiable procedural modifications, such as dialysate sodium concentration manipulation for pre-HD BPV reduction. Finally, the development of objective volume status measures to facilitate dry weight determination is critical to mitigating BPV and other harms of chronic volume expansion.

Similarly, the study does not take into account intradialytic hemodynamic measures. Episodes of intradialytic hypotension or hypertension and/or intradialytic BPV could influence interdialytic BP patterns and fluctuations. The issue of intradialytic BPV is of particular interest because Flythe et al. demonstrated an association between greater intradialytic BPV and greater ultrafiltration rate and volume.¹¹ On the other hand, Shafi et al. found an association between lesser pre-HD BPV and greater ultrafiltration rate and volume. Shafi and colleagues’ finding might reflect the practice of stopping or reducing ultrafiltration in hemodynamically unstable patients; this hypothesis is supported by the fact that patients in the highest BPV quartile were less likely to achieve target weight than patients in lower BPV quartiles. Therefore, the directionality of the ultrafiltration volume–visit-to-visit BPV association remains in question. Given other compelling evidence that greater peridialytic weight gains/losses are associated with adverse clinical outcomes, increasing interdialytic weight gain/ultrafiltration volume should not be pursued as a means to reduce visit-to-visit BPV unless further data become available.

Most noteworthy, Shafi et al. found an association between reduced BPV and non–β-blocker–based and non–RAS-based antihypertensive regimens. Unfortunately, the authors did not examine the effects of specific antihypertensive classes within these regimen categories. Of particular interest are CCBs, a class shown to reduce BPV in non-HD populations.⁴ Prospective studies assessing the effect of CCBs on pre-HD BPV are needed. Such investigations should account for patient adherence to antihypertensive agents and the administration timing of antihypertensive agents with respect to treatments, two aspects acknowledged as limitations in the current study by Shafi et al.

In conclusion, the study by Shafi et al. corroborates and expands the existing evidence that greater visit-to-visit pre-HD BPV has significant prognostic value and points to some potentially modifiable practice patterns, including antihypertensive agent selection and consistent achievement of target weight. Identification of optimal therapeutic strategies to stabilize interdialytic BPV requires prospective investigations. Altering practice without the benefit of evidence accounting for important patient differences, such as sodium balance, volume status, and timing and adherence to antihypertensive agents, should be approached with caution. Although Shafi et al. provide valuable observational data regarding the harms of and risk factors for visit-to-visit BPV, it is now time to set our sights on interventional studies to test these associations in real time and in real patients.

Variety often is the spice of life. The rich diversity and variety that exist among HD patients and HD treatments should compel us to test putative observational associations, such as those between BPV and outcomes in the real, variable world through prospective assessments as we seek therapeutic strategies to improve HD patient outcomes.

Disclosures

S.M.B. is an employee of DaVita Clinical Research, has received speaking honoraria from Fresenius Medical Care North America, and has served on advisory boards for Amgen and C.B. Fleet. His spouse is employed by Astra Zeneca. J.E.F. has received speaking honoraria from Dialysis Clinic Inc.