Conflict of Interest Notices

Transcription

1 BIAS IN DRUG DEVELOPMENT AND USE: HOW DO WE GET THINGS RIGHT? Ross McKinney, Jr, MD Duke University School of Medicine Conflict of Interest Notices I am a member of Data Monitoring Committees for Gilead Sciences I have funding from the NIH through several mechanisms I am chair of the Duke School of Medicine Conflict of Interest Committee Cautionary Warning This talk may appear to be anti-industry Because of the pharmaceutical industry, the HIV infected patients for whom I provide care are doing very well, for which I am extremely grateful View the talk as cautionary the pharmaceutical industry has lost much of the good will it generated for decades, and the concerns I will raise are real 1

2 Overview The nature of conflict of interest and bias The ubiquitous nature of COI Why do we care? Examples of bias and the effects of COI How do we manage COI appropriately? Regulations Concepts The Nature of Conflict of Interest What is conflict of interest? A conflict of interest exists when a primary interest or responsibility is (perhaps unduly) affected by a secondary interest or responsibility 2

3 Illustrations of COI - #1 Picture yourself in dentist s chair Illustration of COI - #2 You walk in to Best Buy to purchase a new interchangeable lens camera. You know a lot about your needs, but not about the available resources You turn to the store s sales clerk for domain knowledge how do you treat the source? Illustration of COI - #3 What Blue are You???? 3

4 Key to the Example Our perception is flavored by pre-set expectations and rationalizations of which we may not be aware The whole crowd saw the same data, but drew very different conclusions COI is part of every day life Human beings are very aware of COIs Every sales encounter All fee-for-service medical encounters Validating Science Some degree of bias is almost inevitable We have hypotheses We are rewarded for establishing new ideas Publications Grants Higher pay Personal satisfaction In short, in science we have strong incentives to prefer positive outcomes 4

5 Validating Science In basic science, the means to control the effects of bias: Controls often blinded Consistency with a logical hypothesis Reproduction Often done by others Requires publication of methods, provision of reagents Note may be a challenge in proprietary research Peer review Concerns: In Clinical Research In science, reproducibility is the key test for validation In clinical research, trials are often too expensive to reproduce Don t want to put people at risk unnecessarily Clinical Equipoise If one therapy is already established as better, how do we randomize? Use the IRB and FDA as judges Concerns in Clinical Research Primary means of validation is audit (specifically, monitoring) Audit is not generally effective as a means to identify bias Problems occur in study design Subjectivity in endpoint and AE assessments Statistical criteria may be biased Articles as written may not reflect the initial study design (rarely checked against the protocol) 5

6 Problems with Corporate Research Jobs depend on positive outcomes When a project gets killed, layoffs often ensue It s hard to kill a new drug if the company has already made a significant investment Preclinical work and early phase studies are very expensive Some drugs live on beyond their logical demise On the flip side, sometimes the science looks great, but the market looks too small In short Bias can enter into clinical research at many levels Nearly all corporately sponsored research is designed to produce a positive outcome Bright people doing careful study design Approval process and marketing needs may force designs to meet FDA or EMA expectations & requirements. Why do we fear bias? 6

7 The Nature of Science Science builds on what s happened before Previous experiments found through publications, presentations, personal communication Experimenters expected to provide key results and methods to others Proprietary research. What if Shoddy research (let s say irreproducible) makes it in to print People will waste time trying to reproduce the results People may try to extend results that weren t valid in the first place In clinical research and future practice, people may be treated with ineffective regimens Wasting Time & Resources Neale et al 2010 Science and Engineering Ethics 102 Articles noted in official findings of ORI to be affected by misconduct, some retracted Analyzed 503 of 5,393 citations of these articles <5% of citations showed any awareness the article was invalidated by misconduct 7

8 Falsified Clinical Research The Kyoto Heart Study: University-run clinical trial of Valsartan (angiotensin II receptor antagonist) versus best practice antihypertensives led by Hiroaki Matsubara, a cardiologist Published clinical trial of 3000 patients (European Heart Journal, 2009) demonstrated improvement in angina, 50% reduced risk of stroke Led to $1B/year sales of Valsartan in Japan Kyoto Heart Study This July, Kyoto Prefectural University of Medicine announced the data had been manipulated in order to produce a positive effect Reductions in angina and stroke were not supported by the real data 8 articles retracted Damage Studies like the Kyoto Heart Study are damaging at many levels Loss of trust Waste of resources Inappropriate treatments May harm Lose the opportunity for more effective options 8

9 Can problems happen in the US? Is the FDA sufficient protection? Merck Again NY Times Friday February 15 Merck Settles Suits Over Cholesterol Drug Merck has agreed to pay $688 million to settle lawsuits claiming it had harmed investors by delaying the release of unfavorable study results for its cholesterol drug Vytorin, the company announced Thursday The back story The ENHANCE study compared a combination drug Vytorin (ezetimibe plus simvastatin) against Zocor (simvastatin) alone Vytorin was a Schering-Plough g drug, Zocor a Merck drug that was going off patent 9

10 The back story Trial found the combination was no better at preventing atherosclerosis than simvastatin alone, although it was better at lowering cholesterol In short, the biomarker was not adequately linked to the desired outcome Companies delayed releasing the information (and publications) more than 2 years (2006 to 2008) During that time sold billions of dollars of Vytorin Merck ultimately bought S-P in 2009 The point should be clear When financial incentives are sufficient, researchers and pharmaceutical companies may hide information that leads to patient harm or death When corporate profits are considered more important than patient lives, know that COI is a serious issue Most cases of COI are, however, far more subtle shades of gray, not black and white To this point COI is ubiquitous we all deal with it every day without batting an eye Scientific COI is expected, but in some cases it can lead to significant distortions in the Science and can put human beings at risk We all need to be aware 10

11 What s the evidence? What is the evidence? This being science, is there evidence that COI actually disturbs science, that bias is introduced? Are the cases I ve described so far the exceptions? JAMA 2003 Als-Nielsen, et al 370 randomized trials Considered: 1) Outcome of trial; 2) Sponsor Non-profit Sponsor - 16% recommended experimental drug as treatment of choice Funding not reported: 30% Mixed funding: 35% For-profit organization: 51% Difference significant (p<0.001)or: 5.3; 95% CI ) 11

12 Turner NEJM Study Studies of 12 anti-depressants; 12,564 patients 38 studies with positive FDA results, of which 37 were published, 1 not published 36 with negative FDA results 3 published, 22 not published 11 published with data selection to appear positive In literature, 94% of publications were positive Turner: NEJM 2008;258: Completeness of Reporting Wieseler, et al, PLOS Medicine October 2013 Compared unpublished Clinical Study Reports submitted to the EMA with publicly available information (journal articles and registry entries) CSRs were submitted voluntarily to the investigators by the companies, not obtained from the EMA 101 trials with full CSRs used in 16 technology assessments by German Institute for Quality and Efficiency in Health Care (similar to NICE) Completeness of Reporting Of the 101 CSRs, 86 had a public document (65 at least one journal article, 50 a registry report) 822 key findings in those 86 cases Postive finding: (54%) in CSR only Negative finding: (45%) in CSR only In short lots of missing information, although it wasn t obviously tilted toward missing harms 12

13 Merck and VIGOR Compared rofecoxib (Vioxx) to naproxen for rheumatoid arthritis 8076 patients Similar symptomatic efficacy Confirmed GI events: rofecoxib 2.1/100 pt-years Naproxen 4.5/100 pt-years (i.e. ~2x worse) Noted MI s were less common in naproxen group (0.1% vs 0.4%) Vioxx and VIGOR A study was published in the NEJM in 2000 Data was provided supporting Merck s claim that rofecoxib (Vioxx) only increased cardiovascular risk in high risk individuals NEJM found a deleted figure on the submitted floppy disk that included data on heart attacks for three low risk patients (the deletion 2 days before submission to NEJM), making Merck s claim suspect. Vioxx removed from the market in September 2004 Supplemental information Data not included in the VIGOR report Serious CV events: Confirmed thromboembolic cardiovascular events Cumulative Incidence (%) Rofecoxib 47 25% 2.5% 2.0% Naproxen % Net: prevented 65 0% Upper GI events at Months Cost of 27 additional thromboembolic events 1.5% 1.0% 13

14 Bias It is again clear that decisions were made about how the research was to be presented that was biased toward a goal One might say this is to be expected of pharmaceutical company research, but if it is, why do it? Why allow people to volunteer, take risks, and not produce credible information How do we manage? Conflict of Interest management strategies and the federal regulations Role of FDA and EMA The FDA and EMA act as COI checks Their role is to represent the public s interest to be sure appropriate studies are done to achieve safe and effective drugs They review study design, performance, and interpretation 14

15 Skepticism as a Protection Kesselheim, et al, NEJM internists shown composed abstracts describing a fictional Randomized Controlled Trial Three levels of rigor to the methodologies Funding attributed to a pharmaceutical company, NIH, or not described 269 physicians responded (53.5%) Kesselheim results Within each funding type, physicians could accurately judge the study rigor Physicians said they would be less willing to prescribe a new medicine on the basis of low rigor studies (OR: 0.64, p=0.008) than medium rigor More willing to prescribe on the basis of high rigor study than medium (OR: 3.07, p<0.001) Kesselheim Results Physicians were half as likely to prescribe if the study was pharma funded than NIH funded (OR: 0.52, p<0.001), regardless of the rigor level In short,,physicians have come to distrust research results produced in industry trials 15

16 What s a company to do? Consider ways to make the study teams more verifiably independent Make granular clinical trial results available for independent review Accurately update results in ClinicalTrials.gov Continue to design high rigor studies (realizing rigor increases costs) How to interact with academics? Be aware that with the Physician Sunshine Act, all physician payments will be posted on a publicly accessible web site (which will affect many physicians willingness to interact for hire) To protect proprietary information, hire faculty directly as consultants If the information can be shared, hire the faculty through research grants/contracts with academic institutions Summary 16

17 Problems Bias is a significant problem in pharmaceutical research The public, including prescribers, has become skeptical about the results presented as a result of corporately sponsored research Greater data transparency will go a long way to improve the credibility of corporate research A Shared Problem Medicine, including physicians, hospitals, device manufacturers, and pharmaceutical companies, has been portrayed as greedy and self serving (unfortunately often for good reason) Changing the public image of medicine in a positive direction is going to be very difficult If we don t make an effort to act in patients best interest, it s highly likely the public will have little sympathy as cost constraints increase In Short Trust in medicine, and the pharmaceutical & medical device industries, requires that we keep the patients interest first in mind, not our own If we lose the people s p trust, regulations and decreasing funding will be sure to follow 17