Cancer is one of the United States’ major public health issues, and investigators have been working hard to find a cure. Researchers are now beginning to focus on active clinical trials, the trial that allows some modifications after it starts. This type is more flexible and less expansive than the traditional one. However, “Time to Adapt,” a Nature article, argues that researchers need to handle this experiment with care; the author claims the adaptive trial requires more time to plan and is less accurate than the traditional trial. As with any new experiment, I agree we should proceed with care, however the potential pitfalls are less severe than the author claims. The Bayesian statistics’ increasing acceptance and the new technology are making the adaptive trial more accessible; in some of the cases, it is even considered less risky than the traditional trial; therefore, the adaptive trial could become the new trend in clinical study.
The advantage of the adaptive clinical trial is its flexibility, which even the author of “Time to adapt” would agree. It allows the researchers to avoid being locked into a single trial and offers them chance to try more alternatives. In the traditional trial, after a protocol is developed, researchers executed it without modification. On one hand, the traditional trial protects the study against sources of bias and inflated alpha (alpha in statistics means the probability of a false-positive error, the error of rejecting a null hypothesis when it is indeed true). On the other hand, if the experiment in the traditional trial proves unsuccessful, the investigators have to drop the entire study since the traditional trial normally does not allow modification after it starts. This is why the adaptive clinical trial would be beneficial. In the adaptive trial, investigators use the data they gathered during the trial to change some aspects of the trial. It saves them a lot of time and money because they could make modifications to the trial instead of stopping it.
Even though the adaptive trial may take longer time to plan than the traditional trial, it still saves the cost. The author of “Time to Adapt” argues that the adaptive trial requires more statistical sophistication and consequently more time to plan. Moreover, pharmaceutical companies may have to change their practice of rewarding the speed in research if they are going to adopt the adaptive clinical trial. However, I think the adaptive would still cost less even if it needs more time to plan because the researchers would not have to drop an entire study if they find if unsuccessful. The sophisticated statistics used to make the investigators stay away from the adaptive trial, but nowadays the development in computers make it much easier to handle these statistics than before. Most importantly, if more researchers start to adopt the adaptive trial, they will have more experience in this kind of trial. It would not take this much time to plan in the future.
One factor that clears the way for the adaptive trial is the growing acceptance of the Bayesian statistical framework, which allows more flexibility than the traditional frequentist framework. The Bayesian statistics are usually computational intense. The recent breakthrough in computational algorithms and computer speed make it possible to compute such complex data. In “Guidance for the Use of Bayesian Statistics in Medical Device Clinical Trials”, the FDA explains the differences between the Bayesian statistics and the traditional statistics. According to the Guidance, “Bayesian statistics is an approach for learning from evidence as it accumulates. In clinical trials, traditional (frequentist) statistical methods may use information from previous only at the design stage...In contrast, the Bayesian approach uses Bayes’ Theorem to formally combine prior information with current information on a quantity of interest. The Bayesian idea is to consider the prior information and the trial results as part of a continual data stream, in which inferences are being updated each time new data become available.” Therefore, the advantage of Bayesian approach is its process of updating knowledge; researchers could update at any time without penalty. This characteristic makes Bayesian statistics more useful for the adaptive trial than the frequentist one does.
Moreover, as the author in “Tools Supporting Adaptive Trials” says, the new technology has made it much easier to conduct an adaptive trial than before. The success of the adaptive trial relies greatly on the real-time data from different resources. In the early days, researchers use fax to transmit the paper-based data. They first manually transcribed the data to a paper case report form (CRF) and then put them together periodically. Such inefficient collection of data had hindered the development of adaptive trial. But now, they use an electronic data capture (EDC) system to assess the data and make it more effective to obtain the real-time data. The researchers can also incorporate the EDC system with the interactive voice response system (IVRS), in which the patients can report the information on the phone.
Though the FDA advises against using an adaptive trial when a traditional one could be used, there are some cases when adaptive trial is less risky than the traditional trial. As the author in “An Assessment of Adaptive” states, “ if the assumptions and available data underlying the design of the traditional study are tenuous, then the risks may be greater for the program than those associated with the study being adaptive. “ When the assumptions in a design is not strong enough, it would be more useful to use the adaptive trial because the investigators will have the chance to explore different solutions.
The FDA also offers advises to solve the problems associated with the adaptive trial. In “Time to Adapt”, the author argues that the adaptive trial might increase the chances of reaching a false-positive conclusion (or Type I error). The FDA Guidance states that, to control the Type I error rate, the researchers can prospectively specify and include in the statistical analytic plan (SAP) all possible adaptations that may be considered during the course of the trial.
The adaptive trial is a great breakthrough in the cancer research area. Like every new finding in science, it has some disadvantages at the beginning stage, but if we put more resources in improving the adaptive trial, it would make great contribution in fighting against cancer in the future.
1. “Time to Adapt”, Nature April 29 2010.
2. “Tools Supporting Adaptive Trials”, by Graham Nicholls, Bill Byrom (2008). Article: ClinPage June 19,2008.
2. “An Assessment of Adaptive”, by Imogene Grimes, Barbara Tardiff.
4. “The Opportunities and Advantages of Using Bayesian Statistics in Clinical Trials”, http://www.tessella.com/wp-content/uploads/2009/11/Capability-statement-Bayesian-Statistics-in-Clinical-Trials.pdf
5. “Guidance for the Use of Bayesian Statistics in Medical Device Clinical Trials”,
http://www.fda.gov/downloads/MedicalDevices/DeviceRegulationandGuidance/GuidanceDocuments/ucm071121.pdf
6. “Guidance for Industry: Adaptive Design Clinical trials for Drugs and Biologics”,
http://www.fda.gov/downloads/Drugs/guidancecomplianceregulatoryinformation/guidances/ucm201790.pdf
The advantage of the adaptive clinical trial is its flexibility, which even the author of “Time to adapt” would agree. It allows the researchers to avoid being locked into a single trial and offers them chance to try more alternatives. In the traditional trial, after a protocol is developed, researchers executed it without modification. On one hand, the traditional trial protects the study against sources of bias and inflated alpha (alpha in statistics means the probability of a false-positive error, the error of rejecting a null hypothesis when it is indeed true). On the other hand, if the experiment in the traditional trial proves unsuccessful, the investigators have to drop the entire study since the traditional trial normally does not allow modification after it starts. This is why the adaptive clinical trial would be beneficial. In the adaptive trial, investigators use the data they gathered during the trial to change some aspects of the trial. It saves them a lot of time and money because they could make modifications to the trial instead of stopping it.
Even though the adaptive trial may take longer time to plan than the traditional trial, it still saves the cost. The author of “Time to Adapt” argues that the adaptive trial requires more statistical sophistication and consequently more time to plan. Moreover, pharmaceutical companies may have to change their practice of rewarding the speed in research if they are going to adopt the adaptive clinical trial. However, I think the adaptive would still cost less even if it needs more time to plan because the researchers would not have to drop an entire study if they find if unsuccessful. The sophisticated statistics used to make the investigators stay away from the adaptive trial, but nowadays the development in computers make it much easier to handle these statistics than before. Most importantly, if more researchers start to adopt the adaptive trial, they will have more experience in this kind of trial. It would not take this much time to plan in the future.
One factor that clears the way for the adaptive trial is the growing acceptance of the Bayesian statistical framework, which allows more flexibility than the traditional frequentist framework. The Bayesian statistics are usually computational intense. The recent breakthrough in computational algorithms and computer speed make it possible to compute such complex data. In “Guidance for the Use of Bayesian Statistics in Medical Device Clinical Trials”, the FDA explains the differences between the Bayesian statistics and the traditional statistics. According to the Guidance, “Bayesian statistics is an approach for learning from evidence as it accumulates. In clinical trials, traditional (frequentist) statistical methods may use information from previous only at the design stage...In contrast, the Bayesian approach uses Bayes’ Theorem to formally combine prior information with current information on a quantity of interest. The Bayesian idea is to consider the prior information and the trial results as part of a continual data stream, in which inferences are being updated each time new data become available.” Therefore, the advantage of Bayesian approach is its process of updating knowledge; researchers could update at any time without penalty. This characteristic makes Bayesian statistics more useful for the adaptive trial than the frequentist one does.
Moreover, as the author in “Tools Supporting Adaptive Trials” says, the new technology has made it much easier to conduct an adaptive trial than before. The success of the adaptive trial relies greatly on the real-time data from different resources. In the early days, researchers use fax to transmit the paper-based data. They first manually transcribed the data to a paper case report form (CRF) and then put them together periodically. Such inefficient collection of data had hindered the development of adaptive trial. But now, they use an electronic data capture (EDC) system to assess the data and make it more effective to obtain the real-time data. The researchers can also incorporate the EDC system with the interactive voice response system (IVRS), in which the patients can report the information on the phone.
Though the FDA advises against using an adaptive trial when a traditional one could be used, there are some cases when adaptive trial is less risky than the traditional trial. As the author in “An Assessment of Adaptive” states, “ if the assumptions and available data underlying the design of the traditional study are tenuous, then the risks may be greater for the program than those associated with the study being adaptive. “ When the assumptions in a design is not strong enough, it would be more useful to use the adaptive trial because the investigators will have the chance to explore different solutions.
The FDA also offers advises to solve the problems associated with the adaptive trial. In “Time to Adapt”, the author argues that the adaptive trial might increase the chances of reaching a false-positive conclusion (or Type I error). The FDA Guidance states that, to control the Type I error rate, the researchers can prospectively specify and include in the statistical analytic plan (SAP) all possible adaptations that may be considered during the course of the trial.
The adaptive trial is a great breakthrough in the cancer research area. Like every new finding in science, it has some disadvantages at the beginning stage, but if we put more resources in improving the adaptive trial, it would make great contribution in fighting against cancer in the future.
1. “Time to Adapt”, Nature April 29 2010.
2. “Tools Supporting Adaptive Trials”, by Graham Nicholls, Bill Byrom (2008). Article: ClinPage June 19,2008.
2. “An Assessment of Adaptive”, by Imogene Grimes, Barbara Tardiff.
4. “The Opportunities and Advantages of Using Bayesian Statistics in Clinical Trials”, http://www.tessella.com/wp-content/uploads/2009/11/Capability-statement-Bayesian-Statistics-in-Clinical-Trials.pdf
5. “Guidance for the Use of Bayesian Statistics in Medical Device Clinical Trials”,
http://www.fda.gov/downloads/MedicalDevices/DeviceRegulationandGuidance/GuidanceDocuments/ucm071121.pdf
6. “Guidance for Industry: Adaptive Design Clinical trials for Drugs and Biologics”,
http://www.fda.gov/downloads/Drugs/guidancecomplianceregulatoryinformation/guidances/ucm201790.pdf
No comments:
Post a Comment