To Cool the Dialysate or Not? Question Answered?

The prevalence of kidney failure continues to increase, with more than 480,000 patients treated with hemodialysis as their KRT modality in the United States in 2020.1 Since its inception, patients receiving maintenance hemodialysis have been plagued by concomitant cardiovascular risk, which remains the major cause of morbidity and mortality, accounting for approximately 40% of deaths.1 Therapies that have proven beneficial to reduce cardiovascular risk among patients without CKD have often not extrapolated to (or been tested in) those receiving maintenance hemodialysis,2 suggesting the need to target alternative or nontraditional risk factors unique to the hemodialysis population. BP typically declines during the course of a hemodialysis session for most patients,3 with a mean decline of approximately 30 mm Hg.4 For many patients, this is thought to be appropriate and correlates with the removal of excess volume that has accumulated since the prior session. However, some patients experience an excessive BP decline, a phenomenon called intradialytic hypotension. This entity has been challenging to define, which is reflected by the plethora of definitions in the literature. Notwithstanding these issues, several studies have described associations of intradialytic hypotension with adverse outcomes (hypothesized to be primarily mediated through end-organ hypoperfusion), including access thrombosis, myocardial stunning, cerebral atrophy, deterioration in residual kidney function, and mortality.5 Therefore, greater emphasis has been placed on exploring interventions with the potential to prevent intradialytic hypotension. Although several approaches have been tried, many are accompanied by significant downsides,5 and few have been tested in randomized trials that are powered to detect differences in important clinical outcomes. Cooling of the dialysate has been suggested as a method to promote hemodynamic stability for decades, with hypothesized mechanisms including activation of the sympathetic nervous system and peripheral vasoconstriction.6 Dialysate cooling has enjoyed a resurgence of interest, with smaller studies suggesting benefits in terms of optimizing myocardial perfusion7 and a modest-sized trial suggesting benefits in minimizing cerebral white matter changes, compared with a dialysate temperature of 37°C.8 Dialysate cooling also has favorable characteristics such as its low cost, ease of implementation, and general acceptance as a mostly safe form of intervention—all advantageous for development of large pragmatic trials. However, a deeper dive uncovers multiple nuances that should be considered. These include various cooling options such as fixed, programmed, isothermic dialysis, thermoneutral dialysis, and negative energy dialysis.6 Owing to practical limitations and ease of use, perhaps the most widely advocated approach is cooling to a fixed difference below the measured body temperature, with many studies choosing 0.5°C below that of the predialysis body temperature. Enter the MyTEMP trial, which was a pragmatic, two-arm, parallel-group, registry-based, open-label, cluster-randomized superiority trial performed from 2017 to 2021 in Ontario, Canada.9 This study enrolled 84 centers and 15,413 patients, representing the largest trial of patients receiving hemodialysis to date. For this, the investigators, the entire study team, and the patients deserve tremendous credit. The intervention group was prescribed a dialysate temperature 0.5°C below body temperature, with a minimum of 35.5°C and a maximum of 36.5°C; the standard group was prescribed a dialysate temperature of 36.5°C. At the end of follow-up, there was no difference in the arms for the primary outcome (composite of cardiovascular death, or hospitalization for myocardial infarction, ischemic stroke, or heart failure; adjusted hazard ratio, 1.00; 96% confidence interval, 0.89 to 1.11). Similar findings were reported for several sensitivity analyses, as well as for the examination of the decline in systolic BP from pre to nadir during the hemodialysis session, while patients in the lower temperature arm were more likely to report feeling uncomfortably cold during dialysis. Given the promising results of physiological studies and preliminary trials that appeared to support cooled dialysate as an intervention to prevent intradialytic hypotension, the nephrology community has been left pondering the reasons underlying the null findings from this trial. In this respect, there are a number of features of the study design and execution that warrant further discussion. First, one should consider the generalizability of the results to other populations. This trial was performed in Canada, a country with a publicly funded health care system, and included patients with a mean age of 66 years, 8% were Black, and around 59% had diabetes. Interestingly, around 70% dialyzed with a catheter, despite a median dialysis vintage of 3.3 years. There were some minor differences between randomized groups in racial subgroups, some cardiovascular comorbidities, and use of various BP medications, with a slightly larger proportion taking four or more antihypertensives in the standard arm compared with the cooled arm. The unit of clustering was individual dialysis units, and individual patients were given the option to opt out of the intervention (although all patients were included in the intention-to-treat approaches). Adherence to the dialysate temperature intervention was assessed by random sampling of deidentified patient data (predialysis patient and prescribed dialysate temperature) from 15 treatments from each center on a monthly basis, with different patients sampled each month. These were all taken from Friday or Saturday sessions at the end of each month and also served as the sampling mechanism for the ascertainment of the intradialytic BP measurements (a key secondary outcome). Of note, some reports have suggested that the frequency of intradialytic hypotension is higher as the dialysis week progresses, so the lack of sampling from earlier weekdays seems less likely to have missed hypotensive events.10 However, the proportion of patients who opted out at each center was not reported, and all patients at a given center were eligible for data sampling (even those who opted out of the intervention). Considered in addition to the nonadherence rate of 20% and crossover of some patients, the sampling of approximately 1% of the dialysis sessions introduces the potential for bias in the data collected and the possibility that an effect of the intervention on BP was missed. The standard group was prescribed a dialysate temperature of 36.5°C, which is lower than that used in many preceding studies.11 If we assume for a moment that cooling the dialysate below 37°C is beneficial, setting the standard temperature of the comparator group at 36.5°C may have already conferred benefits and thus limited the ability to detect additional risk reduction from even further temperature lowering in the intervention group. Indeed, the mean prescribed dialysate temperature of the intervention group was only 0.6°C lower than the standard group, while the lack of data on the delivered dialysate temperature or the effect of the intervention on body temperature also leaves us uncertain if the cooled group was cool enough. Furthermore, as the authors alluded to, the lack of patient-level data precluded the ability to examine the effects of the intervention in hypotensive-prone patients. The trial was 80% powered to detect a 20% reduction in the primary composite outcome. In some respects, this is a relatively optimistic effect size to hope for. Examination of the lower confidence interval around the results of the primary composite outcome (adjusted hazard ratio, 1.00; 96% confidence interval, 0.89 to 1.11) allows for the possibility of a risk reduction of around 11%. To be adequately powered to detect such an effect would have required a sample size around four times as large as the current trial (or much longer follow-up), which seems beyond reach in the current paradigm. Again, one is left wondering if a smaller effect size than the anticipated 20% reduction truly exists and whether a comparator of 37°C may have helped demonstrate this. Of course, opposing arguments can also be made, including that there is truly no treatment effect or even the possibility that the cooled intervention could result in a 11% higher risk of the composite outcome. This was an open-label study, meaning that neither patients nor providers were masked to the intervention. This is perhaps the most important limitation of this study, as a lack of blinding has the potential to introduce bias in clinical treatment. For example, providers (and patients) in the standard arm could have paid extra attention to the review of BP medications and dry weight in an effort to minimize the risk of hypotension, thereby diluting any potential changes in BP and adverse sequelae. Furthermore, owing to the pragmatic nature of the trial, the frequency of interventions in response to low BP events or patient symptoms was not recorded. For example, it is possible that providers in the standard arm could have had a lower threshold for intervening, limiting the ability to detect differences in BP changes between groups. Again, assuming that cooled dialysate does lower the risk of the primary outcome, whether the existence of the described bias would be enough to dilute a true treatment effect is likely to remain a matter of much debate. So, where do we go from here? This well-executed pragmatic trial has answered the question that facility-wide lowering of the dialysate temperature 0.5°C below body temperature versus 36.5°C does not result in a 20% risk reduction in the cardiovascular composite outcome. However, it does not exclude the possibility that a smaller and still clinically meaningful risk reduction may be present (and does not answer the comparison with a standard dialysate temperature of 37°C or with a comparator group with greater separation [e.g., 36.5°C versus 35.5°C]). Future studies may wish to focus on higher-risk subgroups, such as those prone to intradialytic hypotension, and consider innovative approaches to blinding for these interventions. The massive effort for this trial has unquestionably advanced our knowledge in this area but, as always, more needs to (and must) be done. Disclosures D.M. Charytan reports consultancy for Allena Pharmaceuticals (DSMB), Amgen, AstraZeneca, Boehringer Ingelheim, CSL Behring, Eli Lilly, Fresenius, Gilead, GSK, Medtronic, Merck, Novo Nordisk, Renalytix, and Zogenix; research funding from Amgen, Gilead, Medtronic-clinical trial support, and Novo Nordisk; royalties from UpToDate for authorship/edits on reviews; role as an Associate Editor of CJASN; and expert witness fees related to proton pump inhibitors. F.R. Mc Causland reports consultancy for GlaxoSmithKline and Zydus Therapeutics Inc., and research funding paid to institution from Advanced Medical, Fifth Eye, NIH, and Satellite Healthcare.