Editor’s Note: This is a substantially updated version of a previously published story. It includes new commentary from David Hillis, Rick Lange, and myself.
Additional update, December 11, 2010: The PROTECT II press release appears to be a very clear violation of the Ingelfinger Rule. Although companies are obligated to release the top line results of a trial if it contains material information, the level of (cherry-picked) detail contained in the press release is highly unusual, and would appear to jeopardize publication in a major publication or presentation at a scientific meeting, especially in a late-breaking clinical trials session.
Here’s a textbook example of a bad press release (reprinted below). Or, depending on your perspective, perhaps a good example. In any case, it’s a perfect example of modern day alchemy, the ancient pseudoscience that sought to turn base substances into gold or silver.
The occasion for the press release was the early termination of the PROTECT II study. The trial had been comparing the Impella circulatory support device to the intra-aortic balloon (IAB) in high risk PCI patients. The trial was terminated after an interim analysis by the Data and Safety Monitoring Board (DSMB), which reached a “futility determination… regarding the primary end-point,” according to the press release from the trial’s sponsor, Abiomed.
In the press release Abiomed said the source of the trial’s problem was an unexpected increase in the use of atherectomy in the Impella arm of the trial. The press quoted the trial PI, William O’Neill:
“Atherectomy was an unanticipated variable which resulted from the operators’ decision to ‘do more with Impella.’ Our investigators had unblinded knowledge of the treatment arm after randomization. It is interesting that operators felt that they could do more complex interventions once randomized to Impella and this in and of itself is an important finding.”
Comments: The press release for this trial strikes me as an excellent case study in how to spin a negative result. It includes egregious examples of cherry picking the data to find favorable subgroups and endpoints to support a positive view while downplaying or burying any contrary view. Here are two good examples:
There was an overall positive trend in the majority (88%, n=267) of patients in the study at the interim analysis
Impella provided a 47% reduction in major adverse events over IAB in a subgroup that represents 70% of the protocol study population
In addition, it’s not surprising when a company official tries to turn a negative result into a product commercial. It is– or should be– surprising when the principal investigator of the trial, William O’Neill, sounds more like a cheerleader or a spokesperson than a scientist:
I look forward to the scientific presentation at the upcoming American College of Cardiology (ACC) in April and firmly believe that Impella will play a major role in treatment of many patients who need hemodynamic support during PCI.
If he does “firmly believe” that the device is beneficial, what is the point of doing a clinical trial? Even more: on an investor call with a Wall Street analyst, O’Neill said that PROTECT II was “a landmark trial that will impact practice.”
Here’s how David Hillis and Rick Lange wrote about the trial over on CardioExchange:
What’s the spin?
The results leading to the futility determination are attributed to “unanticipated confounding variables related to the use of rotational atherectomy in the study.” Since the investigators had unblinded knowledge of which device was in place in each subject, they allegedly decided to perform atherectomy more often in patients with the Impella device than in those with IABPs. Yet the company claims that the results of the study were encouraging, with a “positive trend in the majority of patients“ and a “47% reduction in major adverse events over IABP in a subgroup that represents 70% of the protocol study population.”
Huh? The trial was stopped early because of a futility determination (i.e., the groups were not treated similarly because of operator bias), yet trends in favor of the device are touted in an unidentified (and possibly cherry-picked) subgroup of patients.
Here is the Abiomed press release:
| Abiomed Ends Protect II Study and Announces Interim Results, Detailed Clinical Analysis to Be Presented at ACC 2011 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| For the Entire Study Population, Impella Significantly Reduced Out of Hospital Major Adverse Events (MAE) by 52% Compared to IAB for the Duration of the 90 Day Monitoring (p=0.02, N= 302) DANVERS, Mass., Dec 06, 2010 (BUSINESS WIRE) – Abiomed Inc. (NASDAQ: ABMD), a leading provider of breakthrough heart support technologies, today announced completion of the PROTECT II study based on a futility determination at the planned interim analysis regarding the primary end-point, which the company views as likely to be due to unanticipated confounding variables related to the use of rotational atherectomy² in the study. The decision to end the study followed the recommendation of the Data Safety Monitoring Board. The study was designed to measure major adverse events at 30 days in high risk percutaneous coronary intervention (PCI) patients randomized to receive hemodynamic support during the procedure with the Impella versus intra-aortic balloon (IAB). Summary and Conclusion of Protect II:
“Atherectomy was an unanticipated variable which resulted from the operators’ decision to ‘do more with Impella.’ Our investigators had unblinded knowledge of the treatment arm after randomization,” said William O’Neill, M.D., University of Miami and Principal Investigator of the PROTECT II study. “It is interesting that operators felt that they could do more complex interventions once randomized to Impella and this in and of itself is an important finding.”
These encouraging results were observed despite other significant treatment variables that were identified in the interim report. These included: two times the treatment of saphenous vein graft (SVG) (p=0.04) in Impella arm, higher contrast media (p=0.01) in Impella arm, 32% of IAB but only 6% of Impella patients left the catheterization lab with device implanted (p<0.001) and IAB patients had four times longer mean support (p<0.001, 2 hours of Impella support, 9 hours of IAB support). “This groundbreaking study data will be utilized along with the presented USpella registry to help physicians identify the patients that receive maximum benefit from Impella as well as the small subset of patients where it may not be appropriate or for those that require longer support outside of the cath lab,” said Dr. O’Neill. “I look forward to the scientific presentation at the upcoming American College of Cardiology (ACC) in April and firmly believe that Impella will play a major role in treatment of many patients who need hemodynamic support during PCI.” “Impella significantly reduced major adverse events as compared to IAB in the majority of patients in PROTECT II and the Impella arm had 52% lower out of hospital major adverse events as well. This study applies to the 26,000 of the 124,000 IABs utilized in the U.S. annually and we are excited at the prospect of converting the majority of the prophylactic IAB PCI patients,” said Michael R. Minogue, Chairman, President and Chief Executive Officer, Abiomed. “With PROTECT II interim results, we now have both the clinical and cost effectiveness data to pave the way to change the standard of care for PCI requiring prophylactic hemodynamic support. These data will augment our increasing commercial momentum as we now focus on execution to capture this market opportunity.” The PROTECT II is a landmark study because it is a first in many categories such as: the complexity and advanced disease of the patient population receiving PCI with hemodynamic support, number of endpoints measured prospectively at 30 days post hemodynamic supported PCI, first use of atherectomy in low ejection fraction (EF) PCI patients and the first FDA Premarket Approval (PMA) study measuring IAB major adverse events. These data reported have been collected, validated, and monitored by a third party academic research organization and the MAE’s were assessed at 30 and 90 days. An additional 125 patients have been enrolled since the midpoint of the study and are not yet included in this interim analysis. Additionally, all hospital charges were tracked by an independent health economic organization in coordination with the study. —————————————————————————————————————————– Conference Call Details: The Company will host a conference call to discuss the results on Tuesday, December 7, 2010, at 8:00 a.m. ET. Michael R. Minogue, Abiomed Chairman, President and Chief Executive Officer, Karim Benali, Abiomed Chief Medical Officer and William O’Neill, M.D., University of Miami and Principal Investigator of the PROTECT II study, will host the conference call. To listen to the call live, please tune into the webcast via http://investor.abiomed.com or dial 800.510.9691; the international number is 617.614.3453. The access participant code is 47378706. A replay of this conference call will be available beginning at 12 p.m. ET on December 7, 2010 through 11:59 p.m. ET on December 21, 2010. The replay phone number is 888.286.8010; the international number is 617.801.6888. The replay access code is 62614381. ——————————————————————————————————————————————– References: 1. The PROTECT II study is a prospective, multi-center, randomized controlled study of the Impella 2.5 versus the intra-aortic balloon (IAB) in patients undergoing non emergent high risk PCI requiring hemodynamic support. The purpose of the PROTECT II study is to determine the safety and effectiveness of the Impella 2.5 as compared to optimal medical management with an IAB, during “high-risk” angioplasty procedures. The study protocol was for low ejection fraction (EF) patients with unprotected left main (EF less-than or equal to 35%) or with triple vessel disease (EF less-than or equal to 30%), measuring a primary endpoint of 10 major adverse event (MAE) components including MACCE at 30 days for a total of 654 patients. Primary MAE endpoint: Prospective 30-day Post PCI includes: death, myocardial infarction, stroke, repeat revascularization (PCI/CABG), need for any cardiovascular operation, acute renal dysfunction, increase in aortic insufficiency, severe hypotension, CPR or arrhythmia requiring Trt, failure to adequately reopen the vessel. The thesis assumed MAE rates of : greater-than or equal to 20 % for Impella and greater-than or equal to 30% for IAB. 2. Coronary atherectomy or commonly referred to as rotablation, is a catheter-based procedure that includes a high-speed rotating metallic burr that abrades calcified (hardened) plaque that is blocking an artery and blood supply to the heart, rotating at speeds of up to 200,000 RPM. Complications listedper FDA labeling include: ventricular perforation, contrast media reaction, stroke, slow flow, no flow, myocardial infarction (Q-wave and non Q-wave), arrhythmia requiring treatment, cardiac tamponade, and death. 3. Statistically Significant p value means pless-than or equal to 0.05, this means that the likelihood that the phenomena tested occurred by chance alone is less than or equal to 5%. ABOUT ABIOMED Based in Danvers, Massachusetts, Abiomed, Inc., is a leading provider of medical devices that provide circulatory support to acute heart failure patients across the continuum of care in heart recovery. Our products are designed to enable the heart to rest, heal and recover by improving blood flow and/or performing the pumping of the heart. For additional information please visit: www.abiomed.com. FORWARD-LOOKING STATEMENTS This Release contains forward-looking statements, including statements regarding the interim results of the Protect II study. Future patient outcomes, including the results from the additional 125 patients enrolled in the PROTECT II study but not yet analyzed, may differ materially in the future and the Company’s actual results may differ materially from those anticipated in these forward-looking statements based upon a number of factors, including uncertainties associated with development, testing and related regulatory approvals, including anticipated future losses, physician usage, complex manufacturing, high quality requirements, dependence on limited sources of supply, competition, technological change, government regulation, future capital needs and uncertainty of additional financing, and other risks and challenges detailed in the Company’s filings with the Securities and Exchange Commission, including the Annual Report filed on Form 10-K and most recently filed Quarterly Report on Form 10-Q. Readers are cautioned not to place undue reliance on any forward-looking statements, which speak only as of the date of this Release. The Company undertakes no obligation to publicly release the results of any revisions to these forward-looking statements that may be made to reflect events or circumstances that occur after the date of this Release or to reflect the occurrence of unanticipated events. SOURCE: Abiomed Inc. Abiomed Inc. Aimee Maillett, 978-646-1553 Public Relations Specialist ir@abiomed.com |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
[…] I wrote then, the press release struck me as a perfect example of how to spin a negative result, relying on […]