The ENDEAVOR Trial: A Case Study in the Interpretation of Modern Cancer Trials

Get Permission

It can be easy to miss the forest for the trees in the interpretation of clinical trials. In particular, trials for the treatment of cancer are exceedingly complex, with long lists of inclusion and exclusion criteria, designs with hidden biases, drugs with unpronounceable names (if not cumbersome alphanumeric codes such as XYZ 14653), various secondary endpoints, and numerous surrogate metrics to assess efficacy.

The ENDEAVOR trial is a great example of a trial that is bold (head-to-head comparison of two leading competitors), answers an important question (is carfilzomib [Kyprolis] more potent than bortezomib [Velcade]?), and has high-quality data rigorous enough to secure approval from the U.S. Food and Drug Administration for a new drug indication. Yet, there is so much more going on beneath the surface that totally changes how one should interpret the results in clinical practice. In fact, interpreting this trial to mean that carfilzomib should be preferred ahead of bortezomib may cause harm.

Study Overview

The ENDEAVOR trial randomized 929 patients with relapsed myeloma to bortezomib plus dexamethasone vs carfilzomib plus dexamethasone.1 Bortezomib and carfilzomib are structurally different but share a similar mechanism of action as inhibitors of the intracellular proteasome pathway. The trial found that the response rate was superior with the use of carfilzomib plus dexamethasone compared to bortezomib plus dexamethasone—77% vs 63%, respectively. The median progression-free survival doubled with carfilzomib plus dexamethasone compared to bortezomib plus dexamethasone—18.7 months vs 9.4 months, respectively.

A reading of the abstract, or even the full paper, might lead the busy oncologist to conclude that when faced with a choice between carfilzomib and bortezomib, choosing the former is a no-brainer. There will be relentless continuing medical education meetings, webinars, and articles about the superiority of carfilzomib based on the “striking” results of the trial. I will point out why such conclusions are potentially flawed and, even if correct, may not be the wise strategy to follow.

Intrinsic Bias

First, the eligibility criteria of the trial were biased. Enrolled patients were allowed to have been previously exposed (or even refractory) to either drug, giving an illusion that the comparisons were fair. But in reality, when the study was conducted, it was nearly impossible for patients to have been previously treated with carfilzomib since the drug was not commercially available in most countries.

The concerns I raise in this piece are not unique to the ENDEAVOR trial. They are pertinent to many recent cancer trials involving new drugs. The accurate interpretation of such trials is not easy, given their inherent complexity, but we must endeavor to do so.
— S. Vincent Rajkumar, MD

Thus, the trial essentially evaluated a large group of patients who had previously received bortezomib, comparing those being reexposed to the same drug vs those given a new drug not previously administered. The question answered for these patients is not whether carfilzomib is superior to bortezomib, but rather, whether in relapsed patients it is better to try a drug they had previously received or to try something new.

To be fair, subset analysis showed that the results were similar in patients who were naive to bortezomib, but these findings carry the limitations of subset analysis. More importantly, the study illustrates the fact that cancer trials today can have hardwired biases that a priori favor one arm over the other in a manner that is not readily apparent unless one pores over the actual clinical trial protocol.2

Dosing Issues

Second, the dosing schedules used in the trial were adjusted to provide the best possible advantage to carfilzomib, using the drug at twice its earlier approved dose (for indications in combination with lenalidomide [Revlimid] and as monotherapy), presumably in order to squeeze out every last ounce of efficacy possible. At the same time, under an erroneous interpretation that a “previously approved schedule” was a regulatory requirement in the control arm, bortezomib was administered in a fashion that is seldom used in regular clinical practice: twice weekly (in all patients) and intravenously (in 22.5% of patients).

When the trial was designed, it was well known that the twice-weekly intravenous schedule of bortezomib causes unacceptably high rates of severe neuropathy,3,4 which could lead to dose reductions and premature withdrawal from the trial. Studies have shown that once-weekly or subcutaneous administration reduces severe neuropathy rates by two- to threefold and allows the drug to be administered for longer periods of time.3,4

Given the added toxicity of dexamethasone, the suboptimal dosing schedule chosen for bortezomib is another example of how a trial can be subtly biased to favor one arm, under the pretext of ostensibly reasonable considerations.

Questionable Surrogate Endpoint

Third, even if the above concerns did not affect the actual outcome of the trial, one has to be aware that progression-free survival in trials like this may not be a true surrogate for clinical benefit (ie, improved overall survival or quality of life). The unanswered question of interest is whether patients receiving bortezomib can receive carfilzomib at progression and obtain the same overall benefit as patients who start out with carfilzomib?

This trial proved the value of carfilzomib as an antimyeloma agent but does not prove that giving carfilzomib ahead of bortezomib will prolong life or result in better quality of life. In this trial, overall survival was identical in the two arms, and given the more cumbersome nature of the carfilzomib schedule (twice-weekly intravenous infusion) and the increased risk of cardiac and vascular side effects, it is debatable as to whether quality of life was improved.

Had the trial compared bortezomib followed by carfilzomib vs carfilzomib followed by bortezomib, one could determine which strategy was better. The design utilized answers a surrogate endpoint question but not a clinical benefit question.

Further Considerations

Finally, one has to talk about convenience, safety, and cost. If all things were equal, I would have no hesitation in recommending carfilzomib ahead of bortezomib, even though the trial was not designed to address clinical benefit. But they are not equal.

Based on how these drugs are administered in practice, bortezomib is far more convenient for patients. Carfilzomib also has safety concerns that have not been fully resolved—particularly with regard to cardiac and vascular events.

Importantly, there is the cost issue. At the dosing used in the ENDEAVOR trial, carfilzomib will cost nearly four times more than bortezomib administered according to the standard schedule used in the clinic (approximately $28,000 per month vs $7,700 per month). This is an important consideration to keep in mind, not only to help patients, but also in recognizing that the system we have in the United States rewards oncologists with a higher total reimbursement if carfilzomib is prescribed rather than bortezomib.

This disparity in profit (and temptation to use carfilzomib) will only grow when generic bortezomib becomes available in 2017. And yet, continuing medical education highlighting this trial will provide ample cover to using carfilzomib ahead of bortezomib.

Possible solutions to these problems would be for the manufacturer of carfilzomib to lower the price of carfilzomib (at 56 mg/m2) to match the price of other proteasome inhibitors; for the Centers for Medicare & Medicaid Services to reimburse physicians with a standard reference pricing rather than 6% above average sales price; for us to identify patients who are best served by the early use of carfilzomib; and for researchers to conduct well-designed randomized trials that answer the sequence question in myeloma.

Closing Thoughts

In my practice, I use bortezomib ahead of carfilzomib for most patients and reserve carfilzomib for patients with high-risk disease at presentation or those who experience disease progression on bortezomib. Carfilzomib is one of the most active agents in myeloma, and as one of the principal investigators in an earlier randomized trial of this drug, I am fully aware of its importance to myeloma therapy.5 In fact, I applaud Amgen and the investigators of this trial for conducting a risky head-to-head comparison trial.

The concerns I raise here are not unique to the ENDEAVOR trial. They are pertinent to many recent cancer trials involving new drugs. The accurate interpretation of such trials is not easy, given their inherent complexity, but we must endeavor to do so.

Disclosure: Dr. Rajkumar reported no potential conflicts of interest.


1. Dimopoulos MA, Moreau P, Palumbo A, et al: Carfilzomib and dexamethasone versus bortezomib and dexamethasone for patients with relapsed or refractory multiple myeloma (ENDEAVOR): A randomised, phase 3, open-label, multicentre study. Lancet Oncol 17:27-38, 2016.

2. Prasad V, Berger VW: Hard-wired bias: How even double-blind, randomized controlled trials can be skewed from the start. Mayo Clinic Proc 90:1171-1175, 2015.

3. Palumbo A, Bringhen S, Rossi D, et al: Bortezomib, melphalan, prednisone and thalidomide (VMPT) followed by maintenance with bortezomib and thalidomide for initial treatment of elderly multiple myeloma patients. Blood 114:Abstract 128, 2009.

4. Mateos MV, Oriol A, Martinez-Lopez J, et al: Bortezomib/melphalan/prednisone (VMP) versus bortezomib/thalidomide/prednisone (VTP) as induction therapy followed by maintenance treatment with bortezomib/thalidomide (VT) versus Bortezomib/Prednisone (VP): A randomised trial in elderly untreated patients with multiple myeloma older than 65 years. Lancet Oncol 11:934-941, 2010.

5. Stewart AK, Rajkumar SV, Dimopoulos MA, et al: Carfilzomib, lenalidomide, and dexamethasone for relapsed multiple myeloma. N Engl J Med 372:142-152, 2015.