Could High Be Better Than Low? Targeting Phosphate in Trials
Phosphate is bad, right?
As kidney function declines, phosphate excretion goes down. Serum phosphate levels rise, causing parathyroid hormone (PTH) to increase and bring phosphate down. Phosphate homeostasis is achieved, albeit at a higher PTH tradeoff. Once the glomerular filtration rate (GFR) reaches a critically low level, things become even more complicated, as the higher PTH doesn’t really help the nonfunctioning kidney to get rid of phosphate. The bones start dissolving. The vessels calcify. Fixing the hyperphosphatemia cascade would surely be beneficial, right?
Observational literature suggests the answer is yes. As an example, an analysis reported higher risk of death with higher phosphate levels (Block et al, JASN 2004). Even a systematic review (Palmer et al, JAMA 2011) of all the observational literature reported an 18% higher risk of death with every 1 mg/dL increase in phosphate. Phosphate is a poison to humanity. Based on the figure below, phosphate appears to be the root of all evil.
So, it’s not surprising that all the major guidelines suggest that we should lower phosphate levels by any means possible.
Looking deeper – Do we really know phosphate is bad?
Note the grade of evidence. It tells you a simple truth; there are no trials to inform these decisions. Why does that matter? Because achieved serum phosphate does not tell you anything about the outcomes if you intend to target lower phosphate. Consider the many possible ways and mechanisms that phosphate is lower and the underlying confounding between the phosphate-adverse outcomes relationship below:
If one examines the trial literature, the actual evidence is very equivocal. In another systematic review of randomized trials (Palmer et al, AJKD 2016) when compared to placebo, the benefit was not significant for any phosphate binder.
How do we get Phosphate lower, anyways?
There are a multitude of ways, each more onerous than the other. We start with dietary restrictions—eating less of all the good things in life. No more ice cream or cheese. No more meat and no more nuts. No more soft drinks. No more baked goods and fast food that often have phosphate additives. Tack this onto sodium and potassium restriction, and you’re left with unsalted crackers.
The next step is to add phosphate binders. Add them to every meal you eat; chew them down as flavoring agents to the food you are eating. Phosphate binders add to the pill burden so much that they account for roughly half of all the pills a dialysis patient takes (Chiu et al, cJASN 2009). Add on quality of life and cost, and we begin to wonder if making this number lower is really worth it. The United States spends $1.5 billion dollars a year on phosphate binders alone (St Peter et al, AJKD 2018).
If you haven’t had enough, another option is to increase dialysis time and frequency. Being an intracellular ion with complex kinetics, phosphate is not as easily removed, so one needs longer dialysis treatments (~ 1.6 mg/dL lower), with the benefit from short daily being much less (~ 0.6 mg/dL) from the FHN trials (Copland et al, AJKD 2016). Needless to say, longer and more frequent dialysis has failed to demonstrate improvement in quality of life (Jardine et al, JASN 2017).
So, in summary, all the ways of lowering phosphate lower quality of life, add huge costs and pill burden, and really have no evidence of increasing quantity of life. Despite that, some eminent nephrologists have said, “From an ethical point of view, it appears impossible to leave dialysis patients with hyperphosphatemia on placebo treatment for prolonged time periods.” (Drueke and Massy, JASN 2012). But let’s ignore this for now and think about the data.
It is possible to do Phosphate target trials?
Two pilot trials have demonstrated that it is possible to enroll patients in a trial and randomize them to lower or higher phosphate targets, and that it is feasible to achieve separation, with no huge signal of harm. In the Two phosphAte taRGets in End-stage renal disease (TARGET) trial (Wald et al, CJASN 2017) 104 patients were randomized to two different targets, and achieved a difference of 1.1 mg/dL (0.35 mmol/L) for 6 months. The amazing part is that the median dose of phosphate binders in the liberalized arm was 0. Zero, nil, nada, zilch. In the Serum Phosphate Intervention in Renal Replacement Therapy (SPIRIT) trial (Bhargava et al, BMC Nephrology 2019) 104 patients were enrolled and again achieved a similar separation of phosphate levels. The pill burden was similarly lower with the higher range phosphate target (median 8 pills/day versus 1 pill/day) with no safety signal. On to the larger outcome trials now.
Phosphate and HiLo Trials
The rationale for the Pragmatic Randomised Trial of High Or Standard PHosphAte Targets in End-stage Kidney Disease (PHOSPHATE) has been explained above, as well as a recent #NephTrials discussion of pragmatic trials. The Pragmatic Trial of Higher vs Lower Serum Phosphate Targets in Patients Undergoing Hemodialysis (HiLo) is the other large trial, and it is different in some noticeable ways. A key aspect of both trials is that they are both pragmatic in nature and aim to enroll a broad swathe of dialysis patients, unlike the tight inclusion and exclusion one typically sees in pharma-funded phase 3 trials. Most RCTs in dialysis patients include patients that are younger and with fewer comorbid conditions (Smyth et al, JAMA Int Med 2019) and do not represent the dialysis population as a whole very well. These trials aim to be more practical and powered for hard clinical outcomes, while also measuring patient reported outcome measures.
The major difference between the two designs is the cluster randomization aspect. What is a potential advantage of cluster randomization in this area? Check out this #NephTrials blog post for the intricacies of cluster randomization trial design. Then let’s see how phosphate control works as an intervention. It is not a pill one prescribes (“PO4 control < 5.5 mg/dL”) and walks away. It requires education on the part of the nephrologist, dialysis nurses, and the dietitian. In addition, dialysis is a very social area, and patients talk to each other. It is the culture of an entire unit to have a uniform policy. An intervention diffuses from one conversation with a patient to others in the same unit. This is what HiLo is betting on. Hence one of the “eligibility” criteria is that the medical director, dietitian, and lead administrator are willing to participate. That is a prerequisite before individual patients are approached.
HiLo in addition has some other key aspects which are somewhat novel and worth understanding more deeply:
- The eligibility is broad; all maintenance hemodialysis patients on dialysis > 3months and > 18 years can be included
- Though it is a cluster RCT, given that it is an individual patient who will receive the intervention and it is more than a “minimal” risk trial, individual patients will be consented, but with tablet devices that connect to secure web-based electronic consent
- Instead of reams of case report forms, the data will be collected automatically based on labs and events that are normally measured and reported anyways
Hospitalization is important and expensive, and mortality is the most important outcome. Since these events compete with each other, they are being analyzed using a hierarchical model, which… Let’s just see the graphical representation below:
In summary, after wandering around in the wilderness of ignorance for the last few decades (ie, treating a number rather than the patient, in the absence of data), we finally have taken the first halting steps. It is not unethical to do a trial of phosphate targets, it is unethical not to do a trial of phosphate targets in CKD.
– Post prepared by Swapnil Hiremath, AJKDBlog Contributor. Follow him @hswapnil.
To view Edmonston et al [Open Access], please visit AJKD.org.
Title: Design and Rationale of HiLo: A Pragmatic, Randomized Trial of Phosphate Management for Patients Receiving Maintenance Hemodialysis
Authors: Daniel L. Edmonston, Tamara Isakova, Laura M. Dember, Steven Brunelli, Amy Young, Rebecca Brosch, Srinivasan Beddhu, Hrishikesh Chakraborty, Myles Wolf
Leave a Reply