Increasing the success rate for Alzheimer's disease drug discovery and development

Abstract

This paper responds to the fact that over 200 Alzheimer's disease (AD) drug candidates have failed to date and draws on searches of the literature for studies of error effects in drug developments and the authors' published works. In the same period, basic knowledge of AD pathology has greatly expanded providing both potential therapeutic targets and rationales for modifications in strategies for testing AD drug candidates. Current opinion generally holds that AD drug candidates have failed because they address pathology that is already too advanced. Less attention is paid to numerous reported methodological weaknesses capable of biasing AD clinical trials and drug developments and thus invalidating conclusions to be reached about the drugs being tested. The costs of quality controls possibly needed to better insure validity in AD drug developments raises concerns that progress toward success in AD drug development may be hindered by the costs of intervening against current methodological barriers to the successful completions of AD drug developments.

Keywords::

• Alzheimer's disease
• clinical trials
• drug development
• methodological errors

1. Introduction

After successful introduction of drugs able to provide some degrees of symptomatic relief to Alzheimer's disease (AD) patients, over 200 AD drug candidates failed to date to modify the course of the neurodegenerative (ND) pathology or the clinical deterioration in AD patients [1]. In the same period, basic knowledge of AD pathology greatly expanded providing both potential therapeutic targets and rationales for modifications in strategies for testing AD drug candidates. Regardless, success in showing relationships between these drug targets and clinical benefits has eluded AD investigators. Unfortunately, this background appears to have discouraged commercial pharmaceutical firms. Closings of firms' in-house neuroscience development divisions have led to widespread concerns that advances in therapies for AD and other NDs will be delayed. This paper provides an overview of recently suggested salient and in-hand strategies for potentially improving AD drug development successes, reasons for giving these strategic priorities, views of why these much-needed advances may be out of reach for the field at present, and predictions for what this means to AD clinical neuroscience research.

2. Why are AD drug candidates failing?

Current opinion generally holds that AD drug candidates have failed because they address pathology that is already too advanced [2]. Consequently, recent expert advice has centered on detecting and treating persons who are at risk of AD rather than persons diagnosable with AD [2]. This strategy and the methodological advances it will spawn have not yet been fully tested for effectiveness because of the difficulties associated with implementation: the problems of distinguishing non-AD, yet destined to develop AD, subjects from simply cognitively impaired but clinically stable subjects; the availability of outcome variables appropriate to document progressions into disease modifications or failures of at-risk subjects to proceed toward AD; the slow development of this ND disease requiring double-blinded, placebo or otherwise controlled clinical trials of 18 months or longer duration and similar factors [3]. These same problems complicate the evaluation of design and method proposals meant to overcome the difficulties associated with current approaches to AD drug developments [4,5].

Less attention is paid to a more fundamental issue potentially able to explain some of the 200 failed AD trials and perhaps critical to the success of future AD clinical research and advances in design and methods. In analyses of various AD and other neuropsychiatric drug developments, various authors found numerous methodological weaknesses capable of biasing these clinical trials and drug developments and thus invalidating conclusions to be reached about the drugs being tested [6-9]. Given the number of failures of AD drug developments it may be the case that adherence to current AD drug development norms or standards of practice does not adequately protect validity [1]. All too prevalent risks shown to compromise the validity of AD and other clinical trials include dosing, biases, ineptitude and deception, protocol violations, measurement errors and so forth [6-9] and can be found to interfere with validity in other's and our analyses [5-16]. A recent analysis of documentation for one clinical trial revealed that some of the trial conditions were statistically significantly predictors for outcomes, grounds to declare that trial invalid as a test of the drug [10]. Yet, this trial had contributed to the abandonment of the drug and to general acceptance that the drug had failed in development for AD. Based on these and other findings and evidence from cognitive science methodological deficiencies in AD drug developments could too easily go undetected and invalidate an unknown number of AD drug research efforts and future efforts to assess the effectiveness of new strategies, such as the current turn toward prevention of AD [11].

3. Possible strategies for improving AD drug development successes

New designs for clinical trials and drug developments and new target populations are essential elements to the advance of medical sciences and therapies. Yet, advances from these innovations can be realized only if the test conditions that will determine the fates of innovations are fair and unbiased. All aspects of AD drug development, method advances, subject selections, drug design, preclinical and clinical phases of development, are most usefully regarded as interdependent. Investigators reach the ultimate verdict on the effectiveness of new methods, drug efficacy, safety and other factors only after successfully negotiating the hazards characteristic of having to progress through stages to reach an outcome. Early errors, if James Reason [11] is correct about error behaviors in complex undertakings, often silently abrogate further investments of time and money. Experiences in other industries have shown that these properties of error intrusions into complex systems can be effectively controlled only by preemptions or effective early interventions sufficient to prevent error and mistake intrusions [11]. If one accepts that error or bias risks in AD drug development can remain undetected and make convincing appearing outcomes invalid [10], then a preemptive strategy seems an important element in any approach taken to reversing the current epidemic of AD drug development failures.

4. Giving methodological advances strategic priority

The recently reported plans for redevelopment of an abandoned AD drug candidate may reveal one source for the relatively limited responses by investigators to the widely published bias risks found in AD and NP drug studies [10]. The redevelopment proposes to equip a Phase III clinical trial with needed countermeasures by adding state-of-the-art scientifically indicated quality controls over methods ranging from controls over inaccuracy to standardizations of biomarker testing. Current costs for adding the indicated technology would increase estimated per subject $40,000 costs threefold. These same cost differentials may not arise for other AD drug developments. On the other hand, AD drug developments currently face a wide range of risks to validity. The costs of quality controls needed to better insure validity in AD drug developments suggest that progress toward success in AD drug development may be hindered by the costs of intervening against barriers to success.

The additional costs for countermeasures would allow a test of the effectiveness of a new state of the art for adding certainty to ratings, biomarkers, monitoring, design and other practices vulnerable to errors in AD drug developments. We had previously demonstrated variance control for ratings by using means of observations in place of single patient assessments [12] and, drawing on Cogger [7], proposed accuracy improvements through screening of raters for quality during earlier phases of a drug's development [13]. With these interventions we are able to achieve power with only 100 subjects; however, from our earlier identifications of undetected error effects common to AD clinical trials [14-16], we would have confidence in this or even a larger study only if the more extensive error controls were in place and tested for effectiveness [10].

Given the usual size of 400 or more subjects in current Phase II and III clinical trials, countermeasures may become economical if they reduce the overall size of these studies [7]. Clearly whatever the outcomes from any one redesigned clinical trial, one test will not justify any conclusions about how AD drug developers might best proceed to insure progress toward success finding efficacy. On the one hand, aware of the costs of methodological innovations that other's and our earlier research has supported, we have become concerned that higher per-subject costs may well make potentially sounder and thus more likely successful AD drug developments impractical to implement without access to less costly quality controls. On the other hand, the scientific progress in biomarker, monitoring, subject status rating and other clinical trial methods may offer routes to overcome problems of error compromises that are leading AD drug candidates to be abandoned. The lack of adequate financial resources or know-how to integrate these innovations may hold back evaluation and adoption of methodological innovations needed to support progress toward successful AD drug developments. AD clinical neuroscience may find appropriate increased emphases on funding for methodological research and innovation with priority given to economical preemptive interventions able to guard against invalidation of clinical trials. Our earlier studies of error and bias effects in AD drug development, more focused on the risks themselves, appear to have underestimated the costs of solutions to these risks to validity.

5. What might the status quo mean for AD clinical neuroscience research?

Hypotheses about biases putting at risk the validity of AD drug developments have not been widely accepted as significant risks to developments and require more research to fully understand the interactions among clinical neuroscience practices, drug development applications of these practices and risks faced by investigators that conclusions drawn from studies will reflect error and not independent variable or drug effects [12]. AD drug development may be in a crisis brought on by failures to support the relevance to disease of any one or more drug targets and lack of progress toward linking drug-target engagement to clinical benefits. While many accept as to be expected the Phase II AD successes that were followed by Phase III failures, such as occurred with Dimebon and was claimed for tarenflurbil [14], in cases for which data are available we have found on analysis potential methodological explanations for the failures [14]. One might say that AD drug development may have an infrastructure problem that no one is positioned or resourced to fix. In this period of fiscal austerity and retrenchment, government, industry, philanthropy or others may not have the resources to afford the work required to better insure the validity of AD drug developments. Pharmaceutical firms that take on the methodological challenges will burden themselves in relations to competitors with added expenses and time and, if successful, will gain no proprietary advantage since competitors could use any new methods and practices. Government policies worldwide leave clinical drug development to industry. The problems of invalidity have not risen to a level of acceptance such that government funded academic AD drug development programs have considered coordinated attacks on potential risks to validity.

What does the future hold for AD drug development? Perhaps as Lon Schneider has observed, success in AD drug development awaits a dramatically effective therapy able to provide a convincing demonstration of its worth despite problems with the methods available to support increased successes in drug developments (personal communication). This ultimately compelling effect size for outcomes occurred with sulfa drugs and early antibiotics. It seems unfortunate if AD drug development must wait on similar circumstances to recur. One alternative is the further documentation and quantification of risks from biases, recruiting of lower cost but effective interventions against bias effects, justifications for advances short of disease modifications and their implicit need for definite supports in addition to clinical status. Each of these has the potential to provide a step toward the identification of a critical path through the slough of biases and lead investigators toward needed certainty for validity realized at reasonable costs.

6. Expert opinion

For current AD drug discovery and development recent experience suggests that the Expert can, at present, only be better informed about what she or he does not know rather than what is known. Academic and commercial experts have been sponsors, investigators or consultants to the consistent failures of AD drug candidates. Experts have taken AD drug candidates through phases of development that failed to predict the outcome of the next phase. Experienced scientists have noted that current critical preclinical and early clinical drug development practices should have less confidence given to them since their failures to predict efficacy lend them no validity [15,16]. The cognitive scientist Daniel Kahneman noted from his own experiences and confirmed in his research that even a failed practice will be continued if it offers the only known approach to the problem [17]. The situation of the AD drug developer may be as a member of the positivist Vienna Circle, Otto Neurath, described for the uncertainty of methods comprising science: ‘We are like sailors who have to rebuild their ship on the open sea' [18]. Fortunately AD drug development has available many of the tools needed for a reconstruction. However, the costs to systematically implement and test these innovations may exceed currently available funding levels. This hurdle may define the transformation needed to achieve a fully functional clinical neuroscience translational medicine [1].

In our view, the current situation diminishes the authority of Expertise and elevates the need for those about to embark on a voyage of drug discovery, expert and novice alike, to challenge themselves to assure that at each step sound science prevails and enthusiasm, commercial priorities, a rush to publish or proceed through a development, do not compromise an orderly and systematic effort to allow the molecule to reveal itself as a potential therapeutic.

Declaration of interest

Nigel H. Greig is supported by the Intramural Research Program of the National Institute on Aging, NIH, and declares no conflicts of interest.