What Pharma/Biotech Investors Need To Know About Adaptive Trials

| About: Novo Nordisk (NVO)


The FDA allows for adaptive drug trials during which key protocols may be altered during the course of the trial depending on interim results.

Adaptive trials have strong benefits compared to conventional trials, but some significant risks as well.

These characteristics have investment implications.

Investors concerned with pharmaceutical or biotech drug development, from early to late stages, need to be aware of the benefits and risks of adaptive trials because their use and interpretation can impact corporate valuations. This is especially true for companies with either just a few key drug prospects or with a future that is expected to be driven by a new drug or discovery platform.

Adaptive trials are becoming much more commonplace in drug development and are being used by a wide range of companies, from microcaps to large caps. While not supplanting traditional trials in most cases, they are nevertheless gaining significant traction. By 2013, the number of pharmaceutical companies using adaptive studies approached 20%.

My interest in adaptive trials was piqued recently with the announcement by Novo Nordisk (NYSE:NVO) that they were going to do an adaptive, flexible-dose Phase 3 sub-trial. Some other former and current examples that I'll also discuss include Pfizer's (NYSE:PFE) stroke drug study ASTIN, a couple of GlaxoSmithKline's (NYSE:GSK) recent rheumatoid arthritis drug-testing protocols, Gilead's (NASDAQ:GILD) Phase 3 study of their leukemia treatment Zydelig, and a Phase 2A trial by Anavex (NASDAQ:AVXL) with their Alzheimer's drug ANAVEX 2-73.

Each of these examples presents revealing information, including valuable lessons, to look for in adaptive-study trials and how stock price valuations can be affected. The example list is only meant to be representative - I'm sure there are many more that could be included.

Conventional Clinical Trials

Traditional drug development has three phases that involve statistical comparisons between fixed-regimen, fixed-design, fixed-analysis active-drug and placebo arms.

Phase 1, which has a primary endpoint of safety, is typically performed with 20-80 healthy volunteers. If there is no toxicity found in Phase 1, the development can move to Phase 2 to investigate the primary effectiveness endpoint. Phase 2 typically has 30-300 subjects and also provides a chance to assess secondary endpoints like safety and short-term side effects. Success leads to Phase 3. A typical, large Phase 3 trial is performed with around 500 to 3,000 subjects. Phase 3 studies include further safety and effectiveness studies, dosage assessments, and effects in different subject populations (regional, by age, etc.).

Note that dose escalation is often performed primarily in the early stages, but it is usually escalated according to a fixed time schedule and so is essentially a fixed-dose regimen for specific time periods. Within the conventional drug-development paradigm, the FDA has stated that other forms of individualized dose titration (gradual dose adjustments) are problematic and tend to favor higher-than-necessary dosages.

Adaptive Clinical Trials

Adaptive trials are an FDA supported method of running all phases of drug trials with the wrinkle that trial regimen, design, or analysis changes can be implemented during the course of the study so long as it's done in a manner that does not undermine the study's validity. The draft FDA guidance for this type of trial was issued in February 2010, though the methods for such trials date back another 30 years or so.

Adaptive trials are typically designed to use multiple forms of accumulated data to determine trial modifications and to assess outcomes. Controls are put in place to ensure the trial's integrity is maintained. To aid the pharma industry, the FDA's guidance is focused on studies that demonstrate "adequate and well-controlled effectiveness"; however, many of these methods also apply to exploratory studies.

As an example, adaptive trials allow for variable, non-fixed dosages for each individual, with adjustments made after responses to the medication are analyzed during the course of the study. If a low dose is not showing sufficient effect in a subject, the dosage can be incrementally increased until a desired effect is observed. In this way, the treatment regimen becomes individually optimized in an adaptive manner for each subject.

Aside from dosage, several other potential adaptations can be made midstream of the trial: dropping a treatment arm, changing the randomization ratio, changing elements of the hypothesis under test, etc.


Just as it sounds, this can become very complicated both in implementation and interpretation when testing hundreds of subjects and potentially using multiple-adaptive design variations with multiple information streams.

Implementation difficulties can arise from the significant additional upfront planning efforts and also from logistics in getting test drugs to the right patients at the right time.

Interpretation issues can arise from the complexity of the study and their required advanced statistical analysis. Adaptive trials typically make use of multiple sources of information, including prior data and potentially different types of data collected as part of the study. This data is obviously much more complex and informative than traditionally obtained data and, therefore, requires more sophisticated, unified statistical analyses (e.g., Bayesian statistics).

The FDA noted some major concerns with adaptive trials. Adaptive designs may contain flaws that lead to false positive effectiveness conclusions (a.k.a. Type I errors). Their complexity can also lead to positive results that are difficult to interpret.

In particular, there is significant increase in the risk of operational bias if adaptive studies are not properly blinded, especially with respect to the unblinding protocols necessary to do the analyses and make the adaptive changes. Such biases can occur when investigators have an opportunity to choose between trial options or if the investigators having knowledge of the adaptive decisions "treat, manage, or evaluate patients differently."

The introduction of such biases cannot be corrected by statistical methods. Consequently, there is potential for additional risk that could be exposed by a correctly performed Phase 3 study. Essentially, partial risk of failure can be transferred from Phase 2 to Phase 3.


On the other hand, the advantages of adaptive trials can far outweigh the disadvantages, which can be mitigated with proper planning and study protocols. Advantages include reduced cost owing to fewer subjects or shorter trial duration, more informative data gathering abilities, improved chances of trial success, and better understanding of both the treatment effects and regimens such as titration. Importantly, adaptive trials can increase the chance that an individual patient will have a favorable outcome.

Adaptive studies are most advantageous for use in studies that have: substantial uncertainty around key trial parameters (population variance, effect size, treatment, appropriate dose, etc.); well understood predictive relationships between biomarkers and clinical effects; a means to quickly measure clinical efficacy; an ability to perform the proposed adaptations in a time frame supportive of the trial; reasonable and well understood logistics costs; and candidates for which specific ethical issues can benefit from the adaptive trial structure.

Adaptive Trial Examples

Prior, Large-Cap Pharma: Pfizer's ASTIN Trial

Pfizer, the world's largest biopharmaceutical company with a nearly $200B market cap, is a pharmaceutical innovator that performed one of the early high-profile adaptive trials. This was done with their ASTIN (Acute Stroke Therapy by Inhibition of Neutrophils) Phase 2 study. ASTIN, whose results were published in 2003, examined dose-response characteristics and proof-of-concept of a neuroprotectant stroke drug.

The purpose of the trial was to determine dose and efficacy relative to placebo. To implement the trial, a predictive model was utilized and continuously updated throughout the study period. An adaptive treatment allocation, i.e., variable dosage based on data, was used with 15 dose options. This permitted more precise dose determinations with fewer subjects than would a conventional study.

A generalized adaptive trial dose-response finding flow chart is shown below. Patients are randomized to active dose or placebo. Predictive models, with inputs obtained from patient data, are run to analyze the cases and estimate dose-response curves. Decisions about efficacy and futility are made according to pre-set rules. Either the trial is stopped for futility or doses are altered in an attempt to optimize them. This cycle continues until either the conditions of efficacy or futility are hit.

In Pfizer's implementation, their ASTIN adaptive trial provided for multiple interim unblinding points that would permit comparison of results to the model, adjustment of doses, feedback to the analysis and, most importantly, the assessment of the trial's progress.

These evaluations indicated insufficient efficacy and signaled a decision for early termination due to futility. Ending the trial early saved both time and money for Pfizer in this case. A detailed adaptive trial lessons learned was published in 2005.

The lesson for investors here is that adaptive trials provide an avenue for early decision making that can save money by minimizing the costs of drug trials, in this case through reduced trial length.

Recent, Large-Cap Biotech: Gilead's Zydelig Trials

Gilead, the world's largest biotech company at $150B market cap, provides examples of studies that were ended early because efficacy was demonstrated.

The first example is the 220 subject study on Zydelig with rituximab for relapsed chronic lymphocytic leukemia. This trial was adaptive owing to periodic unblinding and assessment with an option to stop the trial and possibly other undisclosed options.

The trial was stopped early at its first interim analysis point on the recommendation of their data and safety monitoring board owing to overwhelming efficacy. This was at month 14 of up to 17 months of study. Zydelig was subsequently approved in 2014 and continues to show success in other related clinical trials.

The second example is the recently announced (December 2015) 416 subject study on Zydelig with rituxmab and bendamustine for relapsed chronic lymphocytic leukemia. This trial was similarly ended early on recommendation by their independent data monitoring committee.

These trials show excellent statistical data and routes to approval with the former, and likely the latter as well, with sub-500 subject sizes and relatively short durations.

The lesson here is again that adaptive trials provide an avenue for early decision making that can save money by minimizing the costs of drug trials, but in this case through reduced trial length owing to the positive outcome of demonstrated efficacy.

Current, Micro/Small-Cap Pharma: ANAVEX 2-73 Trial

Anavex is a $200M micro/small-cap development-stage biopharmaceutical company focused on Alzheimer's disease, other central nervous system diseases, and pain treatments. Anavex is studying their drug ANAVEX 2-73 for the treatment of Alzheimer's disease.

Their Phase 2a trial involves 32 mild-to-moderate AD patients. They have an adaptive trial with population pharmacokinetics. Part A of the study investigates bioavailability, dose finding, and exploratory efficacy over 36 days. Part B is a 6-12 month open-label extension with analyses at 3, 6, 9, and 12 months. This trial followed a single-ascending dose Phase 1 trial that investigated safety in healthy subjects.

Anavex cites the following benefits to their adaptive trial: fewer subjects, shorter duration, more likely to demonstrate effect if one exists, more informative on treatment's effects, and broader and better dose-response information and subgroup effects. They believe this is a more efficient study that will help optimize trial parameters for Phase 3 which they envision to be a 6-12 month efficacy trial. Anavex has stated they hope to reduce their Phase 3 trial size to 100-300 subjects, that is, by an order of magnitude relative to conventional studies and similar in size to Gilead's Zydelig studies.

Preliminary Phase 2a, Part A data was released in July 2015 followed by full data released in November 2015. Surrounding these data releases, there were wild stock price fluctuations.

Being a small company with small float seemed to allow for excessive price impacts by players with media influence. The use of an adaptive trial with interim data releases opened the door to speculation and promotional claims by investors through media such as message boards and news outlets.

Strong misleading statements of efficacy by investors/traders followed Anavex's first interim news release and led to a price run-up. While the company press release stated that their interim biomarker "…increase is approximately 4 times higher than donepezil (Aricept), the current standard of care, in the same time frame," this was restated and hyped in message boards as indicating ANAVEX 2-73 is 4 times more effective than donepezil, which is clearly not equivalent in meaning.

Hence, this interim release provided an opportunity for market players to drive prices up from around $4/share to nearly $15/share. An uplisting to NASADAQ followed and gave additional opportunity for price influence by large traders. The price was crashed after the second data release couldn't live up to the hype of the first.

The lesson here is that small companies with multiple data releases during early stages of adaptive trials open up opportunities for both unscrupulous and astute traders. Money can be made by trading into and out of momentum.

On the other hand, long-term investors should understand that there could be significantly more price variation when such companies are running their adaptive trials, so they must really understand what they're getting into and why. They must be confident in their investment as these variations will test an investor's resolve.

Recent/Current, Large-Cap Pharma: GlaxoSmithKline's Multiple Trials

To get a look at the scope and variety of adaptive trials, take a look at GSK's online research and development listings. GSK is a $100B large-cap pharmaceutical company with a broad scope of treatment areas. GSK is currently performing, or recently performed, multiple adaptive studies, primarily in Phase 2.

GSK makes it easy to find what studies are being performed and their key attributes on their clinical study site. For example, I took a look at GSK's rheumatoid arthritis clinical study page where I found adaptive studies 104972 and 201755. The first is a completed adaptive dose-finding study to investigate safety, tolerability, pharmacokinetics and pharmacodynamics of single and repeat doses. The second is an actively recruiting dose-adaptive study.

The point here is that adaptive studies have become mainstream among big pharma and numerous examples abound. Especially pertinent in this example is that GSK's open clinical trial descriptions make it clear that GSK is well-trenched in state-of-the-art adaptive drug trial methods.

Upcoming, Large-Cap Pharma: Novo Nordisk's Oral GLP-1 Trial

Novo Nordisk is a $120B large-cap pharmaceutical company focused on diabetes treatments and biopharmaceuticals. They are developing an oral GLP-1 diabetes pill, with their partner Emisphere (OTCQB:EMIS), for the treatment of type-2 diabetes. They plan to begin their oral GLP-1 Phase 3 trial, called PIONEER and containing 10 subtrials, in early 2016.

The PIONEER 7 trial is a flexible dose escalation study with 500 subjects. This subtrial allows for outcome-based adjustment of dosages during the study.

While details have not yet been released, this study is likely to be designed similar to Pfizer's ASTIN dose-optimization study. It will differ in that an escalating dose regimen optimization to obtain the best outcome with minimized side-effects will be assessed rather than straightforward dose optimization. Far fewer dose choices will be used as well, with potentially 3 versus 15.

For example, given an initial lowest dose, if there is insufficient fasting glucose or HbA1c response, they will move up to a higher dose. Side effects will also be monitored, which can affect the timing between increased doses.

This study is intended to mimic real-world applications. My expectation is that this trial will provide doctor's with treatment guidelines that will produce improved patient compliance and outcomes. According to Novo, several worldwide regulatory agencies have commended them for this approach.

Investment Implications

Adaptive trials have attributes that produce some pertinent investment implications.

  1. Adaptive trials have the potential to save significant study costs, especially in Phases 2 and 3, where the most subjects are traditionally required. They can also make it more likely that the effectiveness of a drug is observed thereby reducing costs in terms of additional pre-clinical work and trials, and the additional time as well as cost to perform them. They also provide points to determine early efficacy or futility. These are likely the drivers behind both GSK's and Anavex's rationale for using adaptive Phase 2 trials. The ability to spend less on trials helps make small companies more competitive and allows big pharmas to broaden their scope.
  2. Adaptive trials may have occasional or frequent unblinding requirements to support data needs at decision points. If such data is released by companies, it may be open to interpretation and manipulation by outside interests. Such ambiguity can open the door to large speculative price swings of small-cap companies leading to both long and short opportunities. I believe that this is in large part the reason for Anavex's recent price volatility.
  3. If a flawed adaptive Phase 2 study is performed and a positive outcome is achieved, additional risk is transferred to Phase 3. Investors should generally be aware of this possibility.
  4. Adaptive trials can provide more drug performance information than traditional trials, potentially suggesting the most effective treatment regimens for certain types of patients. This has the potential for reducing treatment trial and error regimens and improving patient outcomes. As a consequence, more successful treatments could provide a competitive advantage and help maximize pharmaceutical sales. These are likely the primary reasons for Novo's flexible-dose Phase 3 subtrial.
  5. Cost savings and sales increase effects can play into expense calculations and peak sales, respectively, by investors and analysts. Also, time to peak sales could be reduced with more effective treatment regimens leading to more rapid physician acceptance.
  6. Adaptive trials may also increase the overall probability of regulatory approval for a drug's development; however, there is not yet enough data available to generally quantify this benefit. On the other hand, an analyst might want to adjust the weight to the probability of success relative to a traditional estimate at the end of a phase if it's done with an adaptive trial.

Finally, I'll note that adaptive trials are usually tailored to the specific drug development under consideration. There is no one-size-fits-all adaptive trial, only general guidelines are available. Hence, when it comes to investment impacts relative to traditional drug development, a close look at what is being done, what information will be released and when, and how the adaptive trial will impact the outcome must be considered in concert for an effective evaluation.

Disclosure: I am/we are long EMIS.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

Editor's Note: This article covers one or more stocks trading at less than $1 per share and/or with less than a $100 million market cap. Please be aware of the risks associated with these stocks.