The LANDMARK trial we are discussing rests on the premise that calcium is bad, and that calcium based phosphate binders are inferior to non-calcium based phosphate binders. The focus on calcium here suggests it is not the superiority of the non-calcium based PO4 binders, rather something bad about calcium itself. Why might this be? Indeed, the relation between excess calcium consumption and vascular calcification, as well as cardiovascular outcomes has been made before (eg Bolland et al BMJ 2010). Surely non-calcium PO4 binders would be superior then? The extra money on sevelamer, lanthanum and others might be hence justified. Let us dig deeper.
The Sevelamer Stories: DCOR
One of the largest and well done trials of sevelamer is the DCOR trial (Suki et al, Kidney Int 2007). The primary outcome of this trial, all-cause mortality, was not different between arms.
However, the focus shifted to an age-based prespecified (allegedly - this was before the era of mandatory clinical trials reporting) subgroup. This purports to show that there is miraculously a mortality benefit that emerges once a person crosses from being 64 years 11 months and 29 days compared to 65 years. Sevelamer is (non significantly) worse for younger patients, and beneficial for those > 65 years. Talk of dichotomania gone wild. Why not use the mean age (~ 60 years), or even more robust, fit a spline function? The DCOR trial was funded by Genzyme, the maker of sevelamer, and the analysis was performed by an agency contracted by the company.
A much more reasonable analysis followed a year later, using claims based analysis, and accruing many more events (St Peter et al, AJKD 2008). With ~ 850 events (vs 550 in the original trial) this longer follow up confirmed there was nothing to see for a mortality benefit. This group, after appropriately adjusting for multiple comparisons, did not report an age-effect interaction. Notably, this study was not funded by Genzyme, the sevelamer manufacturer.
For further reading, this is strongly recommended:
Editorial by Winkelmayer and Tonelli, AJKD 2008 “Phosphate binder choice in dialysis patients: a call for evidence-based rather than marketing-based clinical practice”
The Sevelamer Stories: INDEPENDENT
This is an even more interesting story. In 2012, CJASN was hoodwinked into publishing the INDEPENDENT study, from di Iorio et al. This was a pilot trial in 212 predialysis CKD patients, of sevelamer compared to calcium based PO4 binders. It originally reported a massive benefit: HR 0.45 for mortality (unadjusted - after adjustment becoming 0.35), and lower need for dialysis (HR 0.55). Stop the presses. Add sevelamer to the drinking water. These effects are better than flozins. Except that they aren’t.
This trial was never registered so it is hard to say what went on behind the scenes. Nevertheless, none of the numbers in the initial published version make sense. In the methods section, it is stated that ‘The follow-up continued until death of any cause, dialysis inception, or 36 months from study entry.’ This suggests that a subject’s follow up was stopped (i.e. censored) if the secondary outcome of dialysis initiation occurred and all the mortality events occurred prior to initiation of dialysis. Of the 107 patients on sevelamer, 31 started dialysis and 12 died, leaving 64 with no events. Similarly, of the 105 patients on calcium carbonate, 42 started dialysis and 22 died, leaving 41 with no events. Hence, from their figure 1, the 12 deaths in the sevelamer group and the 22 deaths in the calcium carbonate group should have occurred in the pre-dialysis period. It also means that deaths in patients who initiated dialysis did not contribute to the primary outcome of all cause mortality, which is not appropriate considering that all-cause mortality is the primary outcome. The numbers at risk in their figure 3, however, are inconsistent with this assumption. For the sevelamer group, the number of individuals at risk for dialysis inception decrease from 107 to 76 over 36 months, with 31 starting dialysis. This would mean that deceased patients are included in the group at risk for starting dialysis.
Sadly this is before the days of twitter peer review and NephJC. And cJASN did not publish letters - but when we pointed out these inconsistencies, a whopping correction was published. Read it and one wonders why the paper was not retracted and republished. Even in the corrected version, one can wonder what is behind the straight line in the first 12 months - no one died or started dialysis (very suspicious for immortal time bias if this hadn’t purportedly been an RCT).
There is nothing in the publication about how this prospective trial was funded.
The strange survival curve of the corrected INDEPENDENT trial . Note the lack of events in the first 12 months.
The Sevelamer stories: INDEPENDENT part 2
But di Iorio and colleagues were not done. The first trial was just a pilot, remember? The larger trial of 466 patients was now done in incident dialysis patients. The target journal was AJKD (see di Iorio et al, AJKD 2013). It had a strange primary outcome not of all cause mortality, nor CV mortality, nor MACE. It was that of CV mortality due to arrhythmias. Quite pioneering for a trial in dialysis patients of a PO4 binder, right? A funny line in the methods goes ‘All patients were followed up for 36 months or censored earlier in the case of a lethal event’ - quite unique to consider death as a censoring event.
Again with dramatic results. Primary arrhythmia outcome HR 0.06 ; all cause CV mortality (yes, you read that right - all cause CV mortality) HR 0.09 and all cause mortality HR 0.23. Sevelamer is closest to the elixir that defies Thanatos. Ironman would be jealous.
Beyond the gullibility required to swallow these findings, note this wonderful letter that followed (Obi et al, AJKD 2014). A key excerpt:
According to the history of the NCT00710788 record, the authors made substantial modifications to the study protocol on February 8, 2013, a full 7 months after submitting their manuscript to AJKD: the primary outcome was changed from “death” to “death due to cardiac arrhythmias or as sudden cardiac death.”
Meta-analytic muddling
This might be the end of the stories, if the issue was not complicated by systematic reviews, the pinnacle of the evidence pyramid. An SR in the Lancet did the fatal (pun intended) mistake of including these trials and reporting calcium based binders of being bad (Jamal et al, Lancet 2013). Classic GIGO. Not KDIGO, though that might be close. We wrote a letter pointing this out (Hiremath and Akbari, Lancet 2014). For more illustrious lineage, check out this erudite commentary (Uhlig, AJKD 2014) titled “Evidence of Comparative Effectiveness Without Evidence of Effectiveness: The Case of Phosphate Binders in CKD”. Despite this, AJKD went on to publish another meta-analysis (this time with network methods) which had the same limitations (Palmer et al, AJKD 2016). These limitations are clearly are mentioned in the text - but who reads the text when a nice and misleading evidence table like this is provided?
Guidelines or Misguidance?
Lastly, the KDIGO CKD-MBD guidelines must be one of the most misleading product ever produced by the KDIGO workgroup (The KDOQI 2006 anemia and the recent 2021 KDOQI nutrition would be fierce competitors of course). They rely on these flawed trials and these fancy meta-analysis to give recommendations favoring non-calcium based PO4 binders. What a shame. As an aside, check out the KDIGO supporters here (sadly missing the manufacturer of cheap old tums).
What happens next?
Fortunately the right trials are happening now. We discuss LANDMARK this week. Meanwhile pragmatic trials like PHOSPHATE (Nephtrials summary from Manasi Bapat) and HiLO (AJKD Blog from Swapnil Hiremath) are ongoing. Another old question also lingers - is it calcium that matters or the anion? The negative DCOR trial mostly had patients on calcium acetate and not carbonate. Perhaps once we get over these industry funded trials, we can answer some real questions that matter.
Update
In typical fashion, Eric Weinhandl (who also was on the DCOR re-analysis team above) tweeted a few studies during the chat - these are not trials, but they demonstrate the lack of difference between calcium acetate as a binder and sevelamer, and calcium vs calcium free PO4 binders:
Yusuf et al, AJKD 2014
Spoendlin et al, JAMA IM 2019
Unsaid in the rant above is any mention of PTH - but Anitha Vijayan covered that very well in this tweetorial some time ago
Another dysphoric editorial is a delightful read as well:
Soomro and Goldfarb, cJASN 2015: “Dysphoria Induced in Dialysis Providers by Secondary Hyperparathyroidism”
Commentary by Swapnil Hiremath