Cinacalcet is a drug that lowers levels of parathyroid hormone, a factor that is increased in almost all people with kidney failure and has been linked to an increased risk of fracture, cardiovascular events and death. It has been hoped that by reducing parathyroid hormone with cinacalcet, the risk of these clinical events will also be reduced. Cinacalcet was approved in the US in 2004 and has been widely used since to lower levels of parathyroid hormone. However, there was no proof that it actually prevents fractures or other important clinical events that impacts patients.

The EVOLVE trial was developed to define whether cinacalcet truly prevents important clinical events in people with kidney failure. 3883 hemodialysis patients with moderate to severe hyperparathyroidism were randomized into a prospective, randomized controlled double-blind trial of cinacalcet vs placebo. The primary endpoint was a composite of time until death, MI, hospitalization for unstable angina, heart failure, or a peripheral vascular event. In an unadjusted intention-to-treat analysis, no statistical difference was found in these primary endpoints, which many found surprising.

This result was disappointing to clinicians and patients alike, and a huge waste of the hundreds of millions of dollars that was likely spent to fund the trial. This is particularly so when a number of prespecified secondary analyses hint that the trial may have missed the detection of a real benefit. As cinacalcet continues to be on the market, clinicians will continue to prescribe the drug and patients will continue to take it. Yet neither group will have a clear, objective understanding of the balance between risks and benefits.

So what went wrong with EVOLVE?

Timing an outcome trial

EVOLVE started in late 2006 and marketing approval for cinacalcet was given in early 2004. During the life of the trial, which ended in 2012, cinacalcet was widely available to the patient population randomized into this trial. About a third of the patients in the placebo group ended up coming off the study and being started on commercial cinacalcet, despite being participants in a clinical trial looking at the effects of this agent. In addition, two thirds of patients in the cinacalcet group discontinued active therapy due to the significant side-effect profile. Thus, the treatment crossover for this trial was high and substantially reduced the power of the trial to test its hypotheses.

This highlights the important of planning an outcome trial before marketing approval is obtained, so that such a trial can be implemented in parallel to, or immediately after, marketing approval. It also highlights the critical nature of clear strategies to maximize adherence to randomized treatment in randomized trials.

Role of regulators and clinicians

The treatment crossover seen in EVOLVE further underlines that clinicians do not yet understand what the role of cinacalnet is. Although higher levels of the parathyroid hormone may be associated with an increased risk of death and cardiovascular events, no randomized trials investigated this until EVOLVE. Yet clinicians were willing to prescribe this drug, itself having a high adverse event profile, without demanding for outcome data that demonstrated real benefit.

The FDA has defined hard outcome-data requirements for diabetes therapies before and immediately after drug registration. Many elements of these, such as the need to demonstrate safety for cardiovascular and other endpoints, can be translated across therapeutic areas. However, regulators have not defined clear requirements for other therapeutic areas even though most of them would share the same treatment goals of reducing morbidity and mortality, measured only by clinical events. Cinacalcet’s marketing approval was based on its efficacy at changing blood tests.

Design and analysis

Another key factor that impeded the ability of EVOLVE to demonstrate a difference was that there was an imbalance in patient characteristics, especially age, that meant that the control group was at lower risk of cardiovascular events due to chance. This made it more difficult to demonstrate a treatment benefit due to cinacalcet, and likely contributed to the lack of a statistically significant treatment benefit. The impact of this imbalance was highlighted by analyses that adjusted for the baseline differences, where a statistically significant benefit was found. As this was not the pre-specified analysis, however, the results carried much less weight than the unadjusted analyses.

This highlights the need to stratify for key prognostic variables and to consider using analyses adjusted for baseline variables. Any differences in distribution of baseline variables will generally be known long before the results of an outcome trial such as EVOLVE are known, offering an opportunity to pre-define an adjusted primary analysis in a reliable and unbiased fashion.

Naturally, demand for additional data puts more pressure on industry and researchers alike, against a background of growing and potentially unsustainable clinical development costs. This piece addresses in part, how we can ensure that time and resources spent on trials such as EVOLVE are optimally spent. However, it is also just as critical to formulate better design and encourage transformation of the clinical trial space.