Description

A tremendous amount of literature has now accumulated on methodology of clinical trials.

Phases of clinical trials

Phase 0. This is also called microdosing and involves administration of a subpharmacological dose (100th of the pharmacological dose) of an investigative drug to human volunteers. It requires minimal preclinical toxicology safety testing. There is some uncertainty if such a small dose can adequately predict the pharmacokinetics of the therapeutically active dose. However, microdosing can provide data that may help the decision to proceed to phase 1 clinical trials. A phase 0 microdosing study has been conducted with an investigational, carbon 11-labeled antiamyloid drug and PET in healthy volunteers and patients with Alzheimer disease (05). The data provided important information on the blood-brain barrier penetration and metabolism of the drug.

Phase 1. These are initial safety trials of a new medicine conducted on healthy volunteers. An attempt is made to find the dose range. Pharmacokinetic studies carried out at any stage of development of a drug are phase 1 studies and may also be conducted in patients.

Phase 2a. These are pilot clinical trials conducted in patients with diseases to be treated or prevented. The focus is on dose-response, frequency of dosing, and other characteristics of safety and efficacy. Pilot trials are important prior to large randomized trials to ensure that the latter are justifiable and well-designed. Results of pilot trials should be reported with caution to avoid undue enthusiasm or pessimism based on unstable estimates (02).

Phase 2b. These are well-controlled clinical trials used to evaluate efficacy in selected populations of patients with the disease. They usually represent a rigorous demonstration of the efficacy of a drug and are also known as “pivotal” trials.

Phase 3a. These are conducted after efficacy of the drug has been demonstrated in phase 2 trials and prior to regulatory submission of a new drug application. They are conducted to demonstrate the efficacy of the drug in a large number of patients for whom the medication is intended. These include both controlled and uncontrolled studies and may be performed in special groups of patients or under special conditions. Most of the package insert and labeling information of a drug is derived from these studies.

Phase 3b. These trials are conducted after the regulatory submission of a new drug application and prior to the approval of the drug. These trials may supplement earlier trials or may deal with other aspects such as quality of life.

Phase 4. These studies are conducted after a drug is marketed to obtain additional information about the safety and efficacy. Some previously unknown adverse effects may be detected during this period. Different formulations, dosages, duration of treatment, drug interactions, and use in special patient populations are studied during this phase. These studies may be in postmarketing surveillance when they are observational and nonexperimental, serving to distinguish them from controlled phase 4 or marketing studies.

Open-label extension. This is a study that usually follows phase 3 clinical trials. Such studies have a legitimate but limited place in the clinical development of new medicines.

Types of clinical trials

There are 3 general types of clinical trials: prospective, retrospective, and a combination of the two.

Prospective trials. These are trials for procedures or drugs that are intended for future use if proven safe and effective. They involve an intervention, eg, administration of a drug or a procedure, to a given number of participants and observation of the resultant variation between them and others that were not exposed. The main purpose of the study design is to investigate cause-effect relationships by comparing intervention groups in a controlled environment. Most of the studies with drugs are prospective randomized studies, and most of the remarks in this article apply to prospective studies.

Retrospective studies. These studies involve evaluation of the data from the past treatment of patients. Retrospective controlled studies are case-control studies in which historical cases are used and are compiled under nonstandardized conditions. The lack of comparability of diagnostic and evaluation criteria is 1 of the many serious disadvantages of retrospective observational studies.

Patient grouping for clinical trials

For grouping of patients, one may use cross-sectional trials, which are short-term trials (usually less than 3 months) in a cross section of patients. Longitudinal trials are long-term trials lasting several months or longer. These are usually conducted in phases 2 and 3. Some of the epidemiological studies conducted in phase 4 are longitudinal studies.

Adaptive clinical trials

An adaptive clinical trial is defined as a trial that makes use of information obtained from within an ongoing trial to decide on modifications in further conduct of the trial. In “drop-the-losers” adaptive clinical trial design, treatment arms are dropped if they fail to meet a prescribed threshold, whereas those that meet this threshold are advanced to the next stage.

Adaptive designs can be used in both exploratory and confirmatory clinical trials. One example of exploratory trial is a dose ranging study to find the effective dose for confirmatory clinical trial. An adaptive design enables predictive biomarkers to identify patients who are likely to benefit from targeted therapies and, thus, increase the success rate of confirmatory clinical trials. To prevent biomarker-negative group from being denied the therapy, all participants should be randomized regardless of biomarker status, but an interim analysis should be used to identify whether the biomarker-positive patients benefit differentially from the targeted agent as compared with the biomarker-negative patients (06). If the interim results show that only the biomarker-positive patients are benefiting, further enrollment in the biomarker-negative subgroup can be terminated.

However, regardless of the degree of flexibility, the types of designs are limited to those in which all potential changes to the conduct of the trial are prospectively defined before the first patient is enrolled. Adaptive clinical trials are useful for acute neurologic disorders such as spinal cord injury in which specific treatments are lacking and assumptions made during trial planning are based on areas of uncertainty, such as dose, timing, and duration of treatment (36).

Efficacy and effectiveness

Efficacy trials (explanatory trials) determine whether an intervention produces the expected result under ideal circumstances. Effectiveness trials (pragmatic trials) measure the degree of beneficial effect under “real world” clinical settings. Hence, hypotheses and study designs of an effectiveness trial are formulated based on conditions of routine clinical practice and on outcomes essential for clinical decisions.

Virtual clinical trials

These imply incorporation of digital health technologies such as mobile devices and remote monitoring devices into study design. Such trials enable collection of information at each stage of the clinical trial, improve trial participant recruitment and retention, online informed consent, measurement of real-time clinical endpoints, and continuous tracking of adverse events. Such trials are important for enrolling patients with mobility issues or those who live in rural areas, which can be far from the research centers at which clinical trials are conducted. The use of virtual clinical trials is currently limited due to ethical, legal, regulatory, and data protection issues that need to be resolved.

Clinical trial designs

If 2 groups of patients are used, then the 2 most common designs are crossover and parallel. In the parallel design, patients are randomized to 1 group of 2 treatment groups and usually receive the assigned treatment (either the active medication or the placebo) during the entire trial. In the crossover design, each patient receives 1 treatment and then the other (in random order). The effect of the trial medication is expressed as the difference between the trial medication and the placebo. One of the problems with crossover studies is carry-over effect of a treatment after discontinuation of the medication. This design can only be used in stable chronic conditions such as epilepsy and migraine.

Historical control design. This design for 2 groups of patients was used more frequently in the past and involves the use of the same clinical trial medication in each group. The control group is composed of patients with the same disease and characteristics who were previously treated with other methods. The problem with this design is that it is not possible to have an adequately controlled historical group. Some of the conclusions reached from historical control trials have been refuted by randomized control trials. Data obtained from a historically controlled group may be limited to information on the natural history of the disease. Such data can be derived from the literature or from the previous clinical trials.

Randomization. Randomization is done to ensure that statistically significant tests are used in a valid manner. This is an attempt to decrease the investigator's bias in assigning patients to 1 group or the other. The patients are randomized after they have fulfilled the criteria for entry into the trial. Several methods are used for randomization. Simple randomization is a process in which a code is assigned to all patients, at the start of the trial and prior to treatment, determining which treatment of the 2 (or more) the patient will receive. The code assigns treatments based on the order of the patients admitted to the trial. In block randomization, a block of patients is chosen and the number of patients assigned to each treatment is proportional (eg, 1:1, 2:1, etc.) within the block. The block size chosen is divisible by the number of patients. The advantage of this method is that if the clinical trial fails to enroll the planned number of patients, there will still be an equal number of patients in each treatment group.

Double blind. Although the definition of double blind varies, a trial is double blind when the patient, investigators, and outcome assessors are unaware of the patient's assigned treatment throughout the conduct of the trial.

Stratification of enrollment. This may be based on characteristics such as age and sex or pharmacogenetic profile. Stratification may be done prospectively or retrospectively.

Number of patients. The number of patients required to demonstrate the effect of the treatment is determined in consultation with a statistician and may range from a few hundred to a few thousand. It depends on the disease, the drug, the method of trial, and the target difference, ie, expected difference between the treatment and placebo groups. A larger target difference generally means that fewer patients are required relative to a smaller target difference. There are no guidelines for selection of an appropriate target difference, but several methods are available, eg, opinion-seeking and review of the evidence base, mostly in clinical trials on chronic diseases (25). A statistically significant 100% improvement in an outcome measure can be detected with a randomized trial involving as few as 42 participants.

N-of-1 trials. These are randomized, double-blinded, multiple crossover comparisons of an intervention and a control treatment in a single patient with a relatively stable condition. They are particularly useful when recruitment of large numbers of patients is difficult or not possible.

Rating scales. Numerous rating scales are available for degenerative neurologic disorders. The numbers generated by most rating scales do not fulfill the criteria for rigorous measurements. Improvements in the scientific rigor of rating scales are needed for better evaluation of the effectiveness of treatments.

Termination of clinical trials. Clinical trials may be terminated for a variety of reasons based on positive or negative findings. For example, a clinical trial should be terminated based on positive findings if clearly superior results in the treatment group would render the continuation of the trial and denial of treatment to the placebo group unethical. Negative findings that warrant termination of the clinical trial include adverse events that surface during the trial or interim results that show lack of efficacy. The mechanism of discontinuation of a trial is complex and may take anywhere from a day to several months.

Extension of clinical trials. To obtain additional information about the long-term effect of a treatment beyond the clinical trial, an extension study may be conducted, which can last several years. This is done by continued observation with a simplified protocol, collecting information and cause-specific mortality. A controlled trial on a large number of patients for this length of time is impractical. The extension study is closer to the real-life situation of medical practice and enables a more realistic appraisal of the long-term effect of the medication.

Institutional oversight bodies. Various examples are:

Ethical considerations. In entering patients in clinical trials, physicians face the ethical dilemma of offering them optimal standard medical care versus treatment that is selected by chance in the context of a randomized clinical trial. A solution to this problem has been for physicians to ethically randomly assign patients to treatments if there is equipoise, ie, a state of professional uncertainty about their relative therapeutic merits. This approach has been criticized with the argument that clinical trials that violate equipoise are ethical and necessary to collect evidence for drug approval and coverage (38).

The U.S. Department of Health and Human Services has indicated that the regulatory framework for protecting human subjects is inadequate in comparative effectiveness research trials, and risks to subjects must be reasonable in relation to anticipated benefits (14). It is recommended that consent forms must adequately consider the potential risks as well as benefits, and these should be explained to the trial participant.

The “Common Rule,” a set of federal regulations for ethical conduct of human subjects research, has been updated and will go into effect in 2018 (35). It did not adopt the proposal to cover researchers’ use of unidentified biospecimens (such as leftover portions of blood samples) and to require informed consent for such research. To combat the growing complexity of consent forms, informed consent will be a “concise and focused” presentation of the key information that will assist a subject in making the decision about participation in a study. More protections provided to research participants will result in greater trust in the research enterprise.

Ethics committees, institutional review boards, and government regulatory agencies are official arbiters of ethical issues, but the investigators should keep the ethical issues in mind while preparing trial protocols. Important issues include the safety of patients enrolled in clinical trials, inclusion and exclusion criteria, and informed consent. A structured survey of numerous participants from a diverse sample of clinical trials showed that only 8% of participants have strong concerns about the risks of data sharing and if adequate security safeguards are provided, most participants are willing to share their data for a wide range of uses (34).

Pharmacoeconomic trials. These trials are carried out separately from the clinical trials and may overlap with any phase from 2 to 4. There are several reasons for conducting pharmacoeconomic trials, including the provision of evidence to the authorities for medication approval. Cost-benefit and cost-effectiveness analyses may be done retrospectively on clinical trial data.

Application of pharmacogenomics in clinical trials. The design of clinical trials is changing with advances in pharmacogenomics. Some of the advantages of using this knowledge for the planning of clinical trials are (27):

Genotyping is important in the design and interpretation of clinical studies. DNA samples should be obtained from all patients enrolled in clinical drug trials, along with appropriate consent to permit pharmacogenetic/pharmacogenomic studies. Because of marked population heterogeneity, a specific genotype may be important in determining the effects of a medication for 1 population or disease but not for another; therefore, pharmacogenomic relations must be validated for each therapeutic indication and in different racial and ethnic groups. Advantages of molecular genetic profiling in clinical studies are the correlation of drug response with the genetic background of the patient, the prediction of dose-response, and the identification of nonresponders to drugs. Single nucleotide polymorphism mapping data can be used to pinpoint a common set of variant nucleotides shared by people who do not respond to a drug.

Clinical trials should be structured in such a way that each test group contains adequate numbers of different phenotypes within polymorphisms. In case of a genotype-specific drug, test groups should contain only the targeted phenotypes. Molecular genetic methods may be applied both for the genetic profiling (polymorphisms, mutations, etc.) of cohorts and for the monitoring and guidance of therapies. Stratification of the patients in clinical trials according to their genetic profile would help in demonstrating the linkage between the subtypes and therapeutic efficacy, thus, optimizing therapeutic relevance.

Genotyping in clinical trials is still in its infancy. This approach is important in some therapeutic areas. An example of this is apolipoprotein E genotyping in clinical trials in Alzheimer disease patients. The overall response of Alzheimer disease patients to cholinomimetic as well as noncholinomimetic drugs varies according to the allotype of ApoE gene, and this is why genotyping is now being employed in clinical trials of various drugs for Alzheimer disease. ApoE may play a central role in determining the most appropriate therapy not only for Alzheimer disease but also for other disorders including depression, Parkinson disease, stroke, multiple sclerosis, and amyotrophic lateral sclerosis.

Some of the key items to be taken into consideration during design of a clinical trial are shown in Table 1.

Table 1. Key Items for Planning a Clinical Trial

Role of biomarkers in clinical trials. A biomarker is a characteristic that can be objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes, or pharmacological responses to a therapeutic intervention. For example, it may be DNA, protein, a metabolite, or a typical finding on brain imaging.

Surrogate endpoint. This is a biomarker that is intended to serve as a substitute for a clinically meaningful endpoint and is expected to predict the effect of a therapeutic intervention. A clinical endpoint is a clinically meaningful measure of how a patient feels, functions, or survives. Clinical endpoints may be further classified as intermediate endpoints, which are clinical endpoints that are not the ultimate outcome, but are nonetheless of real clinical usefulness (eg, exacerbation rate), and ultimate clinical outcomes, which are clinical endpoints reflective of the accumulation of irreversible morbidity and survival. These definitions indicate a clear hierarchical distinction between biomarkers and surrogate endpoints. Whereas numerous laboratory biomarkers may be associated with a certain disease state, the term “surrogate” indicates the ability of a biomarker to provide information about the clinical prognosis or efficacy of a therapy. The word ‘surrogate’ implies a strong correlation with a clinical endpoint, but to be clinically useful a surrogate must provide information about prognosis or therapeutic efficacy in a significantly shorter time than would be needed by following the clinical endpoint.

Assessment of clinical trial results

Data collection. Data collected by clinical trial monitors are recorded in computers by data processors.

Data analysis. The data are analyzed in collaboration with statisticians, and a statistical report is issued.

Data interpretation. The data are interpreted by various persons connected with the clinical trial including the investigator and the sponsor. Statistical as well as clinical significance is determined. The aims of this interpretation are to evaluate the overall significance of the trial results, to relate the results to the original objectives, and to compare the data with similar data obtained from other trials.

Results. The final medical report is usually written by the sponsor after the other procedures have been completed. An important component is analysis of adverse events. The blind code is occasionally broken to evaluate a serious adverse reaction. Finally, the results of the clinical trials are published in a journal. The results of the trial influence the decision for further clinical trials and the translation into clinical practice.

A widespread but unreasonable and oversimplistic practice is the labeling of all randomized trials as either positive or negative on the basis of whether the P value for the primary outcome is less than 0.05. A thorough examination of the whole evidence, including secondary end points, safety issues, and the size as well as quality of the trial are required for better assessment of results. If the overall results of a trial show little or no evidence of efficacy of treatment, and if there are safety concerns as well, discontinuation of further investigation is recommended. If the primary outcome of a trial fails to achieve statistical significance, the investigators should critically evaluate the scientific basis and the results to ascertain if the new treatment may still have value and if further trials with modified design would be worthwhile (45).

The minimum requirement if a trial is to be declared “positive” is a significance level of 5% for the primary efficacy outcome (46). Eventually, physicians at the point of care are responsible for accurate interpretation of clinical trial results and for integrating them with regulatory and guideline recommendations to make the best decisions for treatment of each patient under their care.

Nonresponders. This may result from a high degree of refractoriness to treatment that may have been present prior to start of the trial. In such patients no medicine with characteristics of the trial drug would have elicited any results. There may be other reasons for nonresponse, and these should be analyzed before classifying the patient as a nonresponder.

Interim analysis. Occasionally an interim analysis of the clinical trial is conducted prior to completion of the trial. The results of the interim analysis may affect the decision of whether to continue the trial or not. Important considerations are safety and efficacy. Interim analyses may be used to modify sample size, randomization, dosage of a drug, or regimen of a treatment in adaptive clinical trials that allow modifications in the design of ongoing trials according to a prespecified plan. In most of the clinical trials, no changes in design are permitted during the trial.

Quality of life. There is an increasing emphasis on measuring the effect of treatments in clinical trials on quality of life. Several instruments are available for this purpose, but the results are difficult to assess, particularly when the outcome is a continuous variable, and the difference between a treatment and a control is less than what is significant. However, the treatment may still have an important impact on many patients.

Subgroup analysis. The purpose of subgroup analyses is to investigate the consistency of the trial conclusions among different subpopulations defined by each of multiple baseline characteristics of the patients. Subgroup analyses can provide valuable information if they are properly planned, reported, and interpreted. However, there is a possibility of misuse.

Trial resources and limitations

Clinical trials and physicians in private practice. Most of the clinical trials are conducted in large university medical centers and are sponsored by the pharmaceutical industry. A physician in private practice can also participate in a sponsored clinical trial, particularly in phases 3 and 4. Important requirements for participation are motivation, staff support to fulfill the sponsor's data collection requirements, and liability insurance to protect against any claims arising out of damage from a new medicine or treatment. Such a project should be undertaken only after thoughtful review of the Clinical Investigator's Brochure and protocol. A private practitioner may also conduct a nonsponsored trial if the objective is straightforward and only a small number of patients are required. However, if the physician will be using an investigational drug it may be necessary to obtain permission from regulatory authorities. The manufacturer may supply the medications free of charge to the investigator. Another area of clinical trials in which practicing physicians may help is in the reporting of adverse drug reactions in the postmarketing phase of the drug; these reactions are generally underreported.

One criticism of clinical trials conducted at large medical centers is that the selection of patients does not reflect the conditions in private practice of neurology in communities. The setting of clinical practice, which more closely reflects standard patient care, may be more appropriate for some clinical trials.

Academic and industrial collaborations in clinical trials. Collaborations among various players, including funding agencies, academic centers, and pharmaceutical companies, have helped in financial support and meeting the higher bars for regulatory compliance of clinical trials (49). Patient advocacy groups and practicing neurologists also play an important role.

Role of the internet in clinical trials. The widespread use of the internet in the general population facilitates rapid recruitment of participants as well as instant collection of data in a secure and confidential manner. Internet-based trial method is most suitable when the intervention is safe, the medical disorder can be confirmed by remote means, and the outcome measures can be applied by using electronically transmissible technologies. Within these parameters, the internet trial method offers opportunities for studying treatments quickly and efficiently.

United States Government database of clinical trials. Registered clinical trials are openly accessible at www.ClinicalTrials.gov. This is the largest registry of clinical trials with plans, design, and contact information as well as current status. As of March 2021, it listed 370,218 trials with locations in all 50 states and in 219 countries. The number of trials retrieved with the search term “neurological disorders” is 43,049. Data are self-reported by trial sponsors or investigators worldwide through a web-based system. The registry was expanded in 2008 to include a database for reporting summary results. A publication analyzed data that were publicly available during a period of 1 year ending September 2010 and pointed out some of the limitations of the available information (62). Although the database enables access to study results not otherwise available and allows examination of clinical trials, its ultimate usefulness depends on the submission of accurate and informative data by the investigators.

Drawbacks and limitations of clinical trials. Although randomized controlled trials have become the standard technique for changing diagnostic or therapeutic methods, the use of this method poses several problems. The results of well-conducted clinical trials are not being translated into practice for several reasons that include ignorance of patients as well as physicians, uncertainty as to the applicability of trials to individual patients, and financial considerations. Some of the issues are as follows:

Disparity between the results of animal experiments and human clinical trials. This is well recognized because of failure of animal models to adequately mimic human diseases. An example is that of a corticosteroid, which showed benefit in animal models that could not be proven in clinical trials of human patients with head injury.

Significance of results for clinical practice. Randomized trials may lack relevance to clinical practice. The enrolled subjects are highly selected and unrepresentative of the general population affected by the condition under consideration. Randomized clinical trials provide high-quality evidence about the benefits and harm of medical interventions but may not answer clinicians' questions about a specific treatment in the real-world practice of medicine. To balance this limitation, practical or pragmatic trials have been suggested that compare 2 or more medical interventions that are directly relevant to clinical care, even though they may lack the statistical power (56).

Outdated results. Prolonged periods of planning, conduct, and analysis of clinical trials limits their ability to keep pace with advances in products and treatments that develop after initiation of trials.

Surrogate biomarkers. High costs and time limitations of clinical trials may lead to reliance on surrogate biomarkers that may not adequately represent outcome of interest.

Multiplicity in clinical trials. This problem usually occurs in clinical trials with several objectives based on the evaluation of multiple end points or multiple dose-control comparisons, and many patient populations. Multiple comparisons in a clinical trial increases the likelihood that a chance association would appear causal. Statistical strategies such as multiplicity adjustments based on clinical trial simulations are available to address various sources of multiplicity and the probability of an erroneous conclusion that the experimental treatment is effective (10).

Geographical variations of results of clinical trials. Although most variations in results among countries are likely due to chance, they may also be due to differences among populations in the benefits of an intervention or in adverse reactions. If a randomized clinical trial shows marked variations in results among countries, additional supporting evidence should be sought to determine whether the observed results are real, a variation in methodology and analysis of trial, or simply due to chance (59).

Data from the Matrix and Platinum Science Trial show that endovascular aneurysm retreatment occurs with different frequency and at different times in different regions of the world, but clinical outcomes are similar (55). When retreatment is a primary endpoint, it may or may not take place within the defined study follow-up period. Differences in practice patterns between various countries must be accounted for when comparing outcomes between different trials.

Ethical issues. Physicians participating in these trials are required to modify their commitment to their individual patients. The physician-patient relation of mutual confidence may be disturbed as the patient becomes the research subject for the interest of humanity in general.

Protection of subjects in clinical trials. Several mechanisms have been instituted in the United States to protect the subjects in clinical trials, eg, informed consent and institutional review boards. However, there has been no new regulation since 1991, although the number of clinical trials and innovative therapies has increased considerably. Several changes and suggested improvements in oversight of clinical trials and informed consent are currently under discussion.

Legislation designed to protect clinical trial study participants, close dangerous loopholes, and bring certainty and public transparency to life-saving research studies has been introduced in the United States House of Representatives. The Trial and Experimental Studies Transparency (TEST) Act of 2013 updates and expands the United States Government clinical trial registry data bank with stronger reporting requirements and requires that all foreign clinical studies meet the same registration and reporting requirements as domestic trials if they are used to support an application to market a product in the United States. Eighty percent of the drugs entering the market in 2008 were clinically tested overseas, and a growing number of device trials are also being outsourced to other countries. Many of these trials are not required to be registered with the clinical trials database.

Institutional review board (IRB) delays. Start-up periods for multicenter clinical trials can be delayed because of multiple IRB reviews of the same protocol. During the development of the Network for Excellence in Neuroscience Clinical Trials, the National Institute of Neurological Disorders and Stroke established a central IRB based at Massachusetts General Hospital, which received a grant as the clinical coordinating center for 25 sites located at U.S. academic institutions. It is expected that use of this central IRB will decrease clinical trial start-up time (28).

Significance of information. Licensing trials may provide insufficient information on which to base clinical decisions particularly if the sample size is small, difference in effect is not marked, or adverse effects are common.

Limitations of intention-to-treat analysis. Intention-to-treat maintains treatment groups that are similar except for random variation. The primary goal of efficacy trials is to determine if a treatment works under ideal circumstances, which requires minimization of factors that can alter a treatment effect. Therefore, statistical analyses in efficacy trials frequently exclude patients that deviate from the protocol. In clinical practice, however, factors such as compliance, adverse events, drug regimens, comorbidities, concomitant treatments, or costs all can alter efficacy. An analysis limited to subjects that complete the trial would not take these factors adequately into account.

Bias. The object of the randomized trial is to remove bias. This is only applicable in a situation where the physician has an opinion about the treatment. Randomization is not applicable in the trial of a new, untested treatment for which a physician has not yet formed an opinion.

Blinding. This is difficult to maintain in certain conditions such as cancer and AIDS, in which the side effects of the treatment are evident.

Fraud. This has most commonly been reported in the evaluation of results and, if undetected, may cause promotion of an ineffective and potentially harmful treatment.

Adverse drug reactions. The collection of adverse event data is an important component of clinical trials, but it is controversial whether solicited or unsolicited collection methods are better at distinguishing drug effects from the effects of placebo. Although solicitation produces higher reporting rates, spontaneous reporting provides larger drug-placebo differences.

Rare adverse drug reactions (with a frequency of occurrence less than 1:1000) may not be detected during clinical trials and may become a cause for concern some years after the introduction of the drug. Identification of rare but serious adverse effects of treatment often manifest only during postmarketing surveillance and long-term follow-up (18).

Seeding trials. These are like clinical trials conducted by pharmaceutical companies. The purpose of seeding trials is not to advance research, but to make physicians familiar with a new drug for facilitating marketing. An example of this is the STEPS (Study of Titrate to Effect, Profile of Safety) study of Neurontin (gabapentin), which was a seeding trial posing as a legitimate scientific study (29).

Statistical manipulation. In some types of data analysis, such as performance several subgroup analyses, data can be manipulated to give the result needed to support a hypothesis. Using the definition of a significant p value as 0.05 or less, every statistical test has a 1 in 20 chance of giving a false positive result. If numerous statistical tests are used in situations where there is no real difference between 2 groups, some tests will give statistically significant results by chance. Furthermore, the number of tests can be increased to a point that a false positive result is obtained.

Difficulties in demonstrating the slowing of progression of a disease. Some interventions for chronic neurodegenerative diseases may not produce any improvement but merely delay the progression of the disease to clinically important end points. These require a long-term follow-up and the results are difficult to evaluate. A surrogate outcome measure such as cessation of formation of new lesions on MRI in multiple sclerosis with campath-1 is a sensitive measure of the biological effect of the treatment but a poor index of the clinical effect.

Publication of clinical trial data. In an industry-sponsored trial, the investigators may not have full access to the data or may not be allowed to reveal it in writing or presentations at conferences. In 2016, the International Committee of Medical Journal Editors proposed that as a condition of consideration for publication of a clinical trial report in its member journals, authors are required to share with others the de-identified individual patient data underlying the results presented in the article (including tables, figures, and appendices or supplementary material) no later than 6 months after publication. Sharing of data generated by interventional clinical trials is considered to be an ethical obligation as participants have put themselves at risk (53). Apart from improving the understanding of clinical trial results, this information will help avoidance of mistakes and repetitions in further clinical trials.

Missing data. This is a serious problem that limits the drawing of any conclusions from clinical trials. At the request of the FDA, the Panel on the Handling of Missing Data in Clinical Trials was created by the National Research Council's Committee on National Statistics (44). A special report by members of the panel has analyzed the causes of missing data in clinical trials and offered suggestions to remedy it (30). One of the causes is patient dropout from clinical trials. It is important to design clinical trials carefully as there is no satisfactory way to fix this problem at the data analysis stage. Some analyses apply naive methods to adjust for missing data and make unjustified assumptions. The report concluded that the best approach is to prevent missing data by designing and carrying out the trial in a way that limits the problem.

Nonreporting of negative results. The published results in the medical journals deal mostly with positive outcomes. The negative results are published rarely. An attempt was made to remedy this by public registration of all clinical trials to ensure that information about all clinical trials undertaken would be available regardless of whether their results were positive or negative. The FDA Amendments Act of 2007 required that new clinical trials conducted in the United States post summaries of their results at www.ClinicalTrials.gov within a year of completion, or face a fine of $10,000 a day or loss of NIH funding. The first open audit of the process, covering all clinical trials registered in 2011, found that 4 out of 5 trials covered by the legislation had ignored these reporting requirements (48). No fine was levied. A systematic assessment of the availability of published research reports for publicly registered trials showed that publication proportion for registered trials of licensed drugs is 75%, whereas most of the information collected in unsuccessful drug trials is inaccessible to the broader research and practice communities (21).

Another study found that despite ethical and legal obligations to disclose findings promptly, most clinical trials that were highly likely to be subject to the FDA Amendments Act did not report results to ClinicalTrials.gov in a timely fashion from 2008 through 2013 (01). Compliance was higher in industry-funded trials than in trials funded by the NIH or other government or academic institutions. These findings provide a rationale for policy reforms aimed at promoting transparency, ethics, and accountability in clinical research.

A proposed legislation in the United States House of Representatives called the Trial and Experimental Studies Transparency (TEST) Act of 2012 would require that results from all clinical trials be posted to the database within a year of the trial's end (13). Also, the trials would need to be registered prior to enrolling any participants, and the NIH as well as the FDA would have to report to Congress on new database standards.

Systematic posting of full protocols and statistical analysis plans, which is now required at www.ClinicalTrials.gov under the FDA Amendment Act 2007 final rule and the NIH policy, will enable investigators to discuss and eventually develop consistent standards needed to ensure the valid interpretation of reported results (64). Nevertheless, several opportunities for analyzing the data more systematically as a basis for key decisions by investigators, funders, institutional review boards, and others remain unexploited. In a follow-up to this statement, the authors endeavor to support researchers and institutions in maximizing the value of their efforts and those of the research participants as well as the overall value of the ClinicalTrials.gov results database to the scientific enterprise (61).

Sharing of data from clinical trials. It is generally recognized that sharing of data from clinical trials is useful, but some of the issues under discussion currently are the amount and format of data to be stored and shared as well as how to protect the privacy of the trial participants.

Merely depositing the data somewhere is not enough, but it should be possible for others to find and reanalyze the data. One concern is that it would be difficult to get someone to enter the clinical trial field and spend years collecting data if others could simply obtain the finished data effortlessly. A more important concern is that secondary analysis without involvement of the primary investigators might yield erroneous results. To address these concerns, it may be better to define a core set of data that can be freely accessed from all trials and to allow sharing of the more complex data for specific diseases only through collaborations, which are already an accepted practice in global health care (22). A workshop on sharing data from clinical trials made the following observations and suggestions (42):

Reporting of randomized controlled trials. For accurate assessment of a trial, the published report should contain complete, clear, and transparent information on its methods and findings, but this is not always available. Unfortunately, fewer than half of the trials funded by NIH are published in a peer reviewed biomedical journal indexed by MEDLINE within 30 months of trial completion (50). Because of the high rate of unpublished clinical trials, one needs to search www.ClinicalTrials.gov for both unpublished and published trials, particularly for serious adverse events, as they are better reported in the clinical trial database than in published articles.

According to the FDA Amendments Act of 2007 and the NIH policy, all parties responsible for clinical trials need to ensure that their systems as well as procedures promote complete and timely clinical trial reporting; summary results information must be posted publicly on the following website: ClinicalTrials.gov (63).

Publication of clinical trial results in nonindexed journals. Some clinical trials are published in the national languages of small countries and are not retrievable by MEDLINE search. Such sources need to be searched manually to obtain a comprehensive controlled trials register.

Placebo-controlled studies. It is a general belief that placebo interventions induce powerful effects. Several randomized trials that compared placebo-treated to untreated patients found no evidence that effect of placebo exceeds that of active therapeutic intervention. The ethics of performing placebo-controlled studies of new medications in neurologic disorders where treatment is already available is questionable. Nevertheless, there are ethical reasons that justify study designs that use placebo, provided that the rights and safety of participants are adequately safeguarded.

If placebo effect interferes with evaluation of results of clinical trials, particularly those of antidepressants, sequential parallel comparison design can enrich the study population to be less responsive to the placebo by removing placebo responders between the first and second phases of a trial (37). The final summary statistic in this type of trial is then based on a weighted combination of effects from the 2 stages. Furthermore, adaptive designs in clinical trials were found to recover unnecessary resources spent in the traditional sequential parallel comparison design with overestimated initial sample sizes and provide moderate gains in power.

Nocebo effect. This term is used when an otherwise harmless drug produces effects that are perceived to be harmful by a subject and lead to treatment discontinuation. Nocebo effects are driven by expectancy and contribute to variations of outcome in both placebo and active treatment arms of clinical trials. A study that analyzed pooled clinical trial data in the placebo arms of controlled trials of antidepressant medications to investigate treatment-emergent adverse events in placebo arms found that they were commonly due to nocebo effect with no evidence to associate them with adverse clinical outcomes (11).

Evaluation of nonpharmacological therapies. The randomized placebo-controlled trial was developed to test new drugs. It may lead to negative results in nonpharmacological therapies such as acupuncture.

Limitations of clinical trials for personalized medicine. Average treatment effects shown by clinical trials cannot be reliably applied to an individual patient or even patient subgroups, leading to treatment of patients for whom the treatment is not effective and may be harmful. Better stratification of patients by genotyping, disease stage, or baseline risk of relevant outcomes is more likely to identify those who will benefit and those who will be harmed by an intervention, leading to the development of appropriate diagnostic and treatment thresholds, which will reduce overdiagnosis as well as overtreatment (39). This approach is in line with personalized medicine.

Tips on data analysis

P values. Physicians are not always taught how to interpret clinical trial results and detect flaws. Many would accept the results of a randomized, double-blind, placebo-controlled study in which p values are quoted as proof that 2 treatments led to different results. However, p values convey meaningful information only if they are put into a clinical context. A p value of 0.05 has historically been used to support treatment differences. A confidence interval usually consists of 2 values, an upper and a lower limit, attached to a level of probability. For example, in a study comparing the abilities of treatments A and B to reduce a symptom, a mean difference of 20% is reported and the 95% confidence interval is 13% to 27%. This means that if we treat the whole study population with treatments A and B, we are 95% confident that the difference between the efficacy of the 2 treatments would lie somewhere between 13% and 27%. Confidence intervals can be used to indicate the clinical significance of a p value. This is important because a small clinical difference may be statistically significant because of large sample size, whereas a clinically important effect may appear statistically insignificant if the number of subjects studied is too small.

The use of confidence levels as well as the plotting of clinical significance curves and risk-benefit contours has been proposed to provide degrees of probability of both the potential benefit of treatment and the risk due to toxicity (52). These should be incorporated in reports of clinical trials to provide clinicians with a mechanism of translating the results of studies into treatment for individual patients and to facilitate the clinical decision-making process.

Intention-to-treat analysis. Such an analysis is the most rigorous that can be performed on clinical trial data and includes all the patients who were randomized to a study. If the number of patients analyzed is less than the number randomized (such as in subgroup analyses), there can be a serious bias as the randomization ensures that the groups are similar in terms of patient characteristics.

Errors in clinical trials

Errors occur when one incorrectly evaluates the difference in outcomes between the placebo and the treatment groups. Type 1 error (false-positive signal) is the erroneous conclusion of difference when in fact no difference exists. The probability of a type 1 error (alpha) is usually set at 0.05. Type 2 error (false-negative result) occurs when a false conclusion is made that the 2 outcomes are not significantly different when they are. A type 2 error can result from erroneously failing to reject the null hypothesis. The probability of a type 2 error (beta) decreases as the sample gets larger or the statistical power (1-beta) increases.

Clinical applications

Influence of clinical trials on patterns of medical practice

The results of controlled clinical trials are a potentially important supplement to a clinician’s uncontrolled clinical experience and an extension of knowledge about pathomechanisms of a disease. With knowledge of clinical trial results, a physician can make a more informed decision about appropriate treatment. Some well-executed clinical trials, the results of which have been published in high profile medical journals, have resulted in significant changes in medical practice. This impact, though, is not consistent because the factors that influence the patterns of practice are complex. Physicians do not necessarily change their prescribing habits following the introduction of new drugs, despite the reported success of clinical trials, especially in cases for which several drugs are available, as in epilepsy. More cautious physicians tend to observe the drugs for longer periods in the postmarketing phase and review the results of meta-analyses before forming opinions about adopting a new drug in their practice.

Explanatory versus pragmatic clinical trials. Although explanatory trials confirm a physiological or clinical hypothesis, pragmatic trials form the basis of a decision by providing evidence for adoption of the intervention into the real world clinical practice. Very few trials can be fully pragmatic, and some novel interventions can have a significant impact on clinical practice without being pragmatic. A practical approach to pragmatism is to adopt the features of pragmatic trials whenever feasible without compromising trial quality and the ability to answer the clinical question of interest (15).

Consensus. Randomized clinical trials are the best evidence considered by consensus conferences that also determine the practical importance of the trials. Common features of consensus conferences and clinical trials are an impersonal nature and objectivity. Consensus is the opinion and recommendations of a group, not an individual.

Metaanalysis of clinical trials. Some problems with the translation of clinical trials into real life practice are revealed by metaanalysis of randomized, placebo-controlled trials to determine the efficacy and adverse effects of atypical antipsychotics for the management of agitation, aggression, delusions, and hallucinations in patients with Alzheimer disease. Although small statistical effect on symptom rating scales support the evidence for the efficacy of atypical antipsychotics, it is limited by incomplete reporting, dropouts, and adverse effects. It leaves the decision up to the physician to prescribe atypical antipsychotics within the context of medical need and the efficacy and safety of alternatives.

Standard systematic reviews and meta-analyses rely on published aggregate data to provide robust clinical conclusions. A study found that hazard ratios from published aggregate data were most likely to agree with those from individual participant data when the information size was large (54). Based on these findings, Tierney and colleagues provide guidance for determining systematically when standard meta-analysis of published aggregate data will likely generate robust clinical conclusions and when the individual participant data approach will add considerable value.

Clinical trials and evidence-based medicine. The practice of evidence-based medicine is the most judicious use of currently available medical evidence. It involves the integration of individual clinical expertise with the best external evidence from clinical trials and systematic reviews such as the Cochrane Collaboration. The Cochrane Collaboration is a systematic review (also called overview or metaanalysis) of randomized trials and has 2 main components: 1) a systematic search for all relevant randomized clinical trials (published or unpublished), and 2) the use of appropriate statistical methods to derive the best estimate of treatment effect. Industry sponsored reviews of drugs are less transparent, have fewer reservations about methodological limitations of the included trials, and have more favorable conclusions than the corresponding Cochrane reviews. Some neurologic disorders are included in the Cochrane Database of Systematic Reviews.

In relation to evidence-based medicine, emphasis is placed on the design of double-blinded randomized controlled trials, statistical power, and the level of significance. Other significant aspects of trial quality—biological plausibility, reproducibility, and generalizability—that affect the validity of the findings should also be considered.

In some instances, recommendations for treatment need to be made in the absence of definitive clinical trial data. Limitations of recommendations of this type should be taken into consideration and treatment decisions can be based on a combination of careful clinical assessment of the patient and an objective evaluation of whatever clinical trial data are available.

Clinical trials in neurology

Instruments for measurement of outcome in management of neurologic disease with disability should be evaluated for clinical usefulness as well as scientific soundness.

Few instruments for measuring neurologic outcomes have been evaluated comprehensively. Many of the commonly used instruments, such as the Expanded Disability Status Scale, provide limited data about their scientific properties. The Barthel Index, which is a benchmark for measurement of disability, has incomplete psychometric data.

Brain imaging provides objective documentation of several neurologic disorders. Several brain mapping techniques are now available to enable the clinician to monitor disease progression and therapeutic effects of a drug in either the routine clinical setting or experimental clinical trials.

Recruiting patients for clinical trials in neurology. Recruiting patients for clinical research studies in neurology is challenging. Distance, disability, and the need for frequent in-person visits are major barriers to participation in clinical trials. Home visits may enable greater participation in clinical trials, but this is not practical in geographically diverse populations or for persons residing in nursing homes or not located near research centers. The Michael J Fox Foundation has a Fox Trial Finder online registry that identifies individuals with and without Parkinson disease who are willing to participate in future research studies. In 1 study of virtual research visits with videoconference assessment, neurologists validated self diagnosis of Parkinson disease in 97% of cases with overall satisfaction rate of 79% (12). Despite some limitations, this approach may facilitate future clinical trial participation.

Applications of clinical trials in neurology

Stroke. Because stroke is a complex condition with several clinical variables that compound care and influence outcome, few therapies will produce such a dramatic response that the medical community will accept the benefits of a treatment based solely on anecdotal, uncontrolled reports. Trial methodology in stroke can be problematic. Obtaining agreement on trial design, eligibility criteria, primary efficacy parameters, neurologic scales, and study sample size is not always easy to accomplish. Application of proposed treatments in appropriate animal models is a prerequisite for interventional clinical trials in humans.

Controlled clinical trials are expensive, and such projects should not be launched unless strong evidence from experimental and pilot clinical studies shows that there is reasonable chance for success. Data from experimental studies should be clearly positive. The pharmacological properties of the drug should be known with a rational basis for the desired effect. Issues of safety and efficacy are best addressed by pilot clinical studies. In stroke clinical trials, the data safety and monitoring committee provides recommendations to the steering committee of the trial on whether to continue or to stop patient recruitment and on protocol amendments that may be necessary to protect the safety of the patients.

Because the stroke population is heterogeneous, treatment effect may be biased even in balanced randomized trials. Risk adjustment statistically addresses some of the heterogeneity and can reduce bias in the treatment effect estimate.

Estimation of number needed to treat in parallel design clinical trials is a useful measure of a treatment's clinical benefit or harm. Several randomized clinical trials have evaluated the long-term use of antiplatelet drugs in reducing the risk of new vascular events in patients with ischemic stroke and the results form the basis of various guidelines for the management of such patients in clinical practice. A review of data from the Netherlands Stroke Survey showed that patients with ischemic stroke enrolled in randomized clinical trials are only partially representative of patients in clinical practice and suggested that less restrictive enrollment criteria could improve the selection of patients (31).

The information overlap that is misclassified in healthcare data from electronic medical records can be used to calibrate stroke risk factor management (41). This enables more accurate reweighting of the trial findings for application to the target population. For example, if in the overlapped part of the target population, 1 of 3 individuals have false-positive results and 2 of 3 have false-negative results, the same proportion of corrections is expected in the whole target population when measurements are calibrated, ie, 8 of 24 are false-positive, meaning no transient ischemic attack, and 14 of 21 are false negative-meaning transient ischemic attack.

The traditional approach to acute stroke clinical trial design has been problematic on several levels. There is a need to validate surrogate biomarkers, particularly brain imaging, and designs need to be developed for combination therapies of stroke.

Evaluation of restorative and regenerative therapies for stroke requires functional imaging such as fMRI and PET for assessment. Functional imaging for measuring tissue function introduces some complexities not encountered in structural imaging as the effectiveness of these interventions is influenced by factors such as activity (09). However, imaging measures may also serve as a biomarker of treatment effect, which might secondarily guide restorative trial design.

The DAWN trial (DWI or CTP Assessment with Clinical Mismatch in the Triage of Wake-Up and Late Presenting Strokes Undergoing Neurointervention with Trevo) investigated the efficacy and safety of endovascular thrombectomy that is performed 6 to 24 hours after the onset of stroke (43). The trial was halted because of the results of a prespecified interim analysis that suggested high probability of success. A discussion of the results assured that the positive outcome of the trial was not due to complex statistical maneuvering or the use of an unconventional end point that involved a utility-weighted modification of the usual Rankin scale (20). Even with the use of basic parametric statistics and an end point that is typically used in stroke trials, the DAWN trial had strikingly positive results. The trial required evidence of a small infarct core on diffusion-weighted imaging and no evidence of an infarct on fluid-attenuation inversion recovery imaging; this combination of findings indicates that the onset of stroke symptoms (which were observed at awakening) most likely occurred less than 4 hours earlier.

Multiple sclerosis. Signs and symptoms of multiple sclerosis are subjective and difficult to measure. Fatigue and numbness can only be assessed by self-reports because objective neurologic examination cannot measure this experience. Current clinical research and outcomes rely heavily on outcome assessment that is influenced by ambulation. These parameters may be insensitive to some therapeutic measures. Possible adverse reactions to drugs produce another complicating factor.

The unpredictable natural history of multiple sclerosis is a problem concerning evaluating the results of clinical trials. Patients are selected for clinical trials of multiple sclerosis based on criteria designed to identify patients with high event or treatment failure rates. The power of the clinical trial is determined by the event rate in the control group on a short-term basis. The aim of the treatment, however, is modification of the long-term outcome such as time to reach the Expanded Disability Status Scale. Baseline Expanded Disability Status Scale and the duration of the disease should be taken into consideration in designing clinical trials of progressive multiple sclerosis.

A phase 2b placebo-controlled study of imilecleucel-T, an autologous T-cell immunotherapy for multiple sclerosis, showed no statistically significant clinical or radiological benefit, but a prospective analysis of subjects with more active disease favored immunotherapy in terms of annualized relapse rate (16). An analysis also found a possible beneficial effect of prior disease-modifying treatment, which might explain the lower relapse rate in the placebo group and suggests limitation of future trials to treatment-naive patients.

The placebo group has other problems as well. In some studies, placebo groups have fared better than expected. Determination of the magnitude of placebo effect in multiple sclerosis patients would require randomization of patients, with 1 group receiving no therapy and another group receiving an inactive placebo. The placebo effect is usually short-lived and does not produce a significant change in disability scales.

Clinical trials in multiple sclerosis are complex because there are many variables in the disease, and there is a lack of universal definition for certain aspects of relapsing-progressing forms. Duration of disease and number of relapses prior to entry into clinical trials are the best predictors for the relapse rate during the study, whereas gadolinium enhancement status does not have a significant influence.

A study on patients experiencing multiple sclerosis relapse compared the outcome of being on- or off- to a randomized controlled trial for the same intervention, ie, a 3-day regimen of intravenous methylprednisolone (33). Results showed that those receiving interpersonal care as outpatients in the off-trial group were significantly worse than those in the clinical trial. The beneficial effect observed is likely secondary to clinical trial participation as both groups had similar baseline features and were received the same way. The findings support the incorporation of the structured clinical trial–style approach into the routine management of multiple sclerosis. Because multiple sclerosis is a chronic disease, long-term follow-up of clinical trials is necessary to evaluate the disease-modifying effect of a therapy (17).

In a review of the 100 most cited clinical trials of multiple sclerosis carried out between 2010 and 2015, most reported limitations were sample size, outcomes, trial duration as well as design, and population characteristics due to variability in disease status (40). Given variations in the natural history of multiple sclerosis and various forms of presentations, severities, and progression outcomes, it is difficult for subjects enrolled in clinical trials to represent the entire population of multiple sclerosis.

Epilepsy. Although there is a long list of available drugs for the treatment of epilepsy, refractory cases have prompted the development of new drugs as add-on therapy and, in some cases, as monotherapy. Controlled clinical trials provide evidence for the efficacy and safety of new medications but do not define the best use of these agents. The trials have ignored some of the factors, such as etiology and epilepsy syndromes, that determine prognosis. To resolve some of these issues, large multicenter studies on patients with well-characterized seizures have been suggested. Restriction of clinical trials to major teaching hospitals creates special problems in the design of monotherapy protocols because referral institutions usually get treatment-refractory patients.

The past standard of clinical trials in epilepsy has been add-on therapy. To obtain approval of a new antiepileptic drug as monotherapy, the regulatory authorities require a placebo-controlled trial, and the pharmaceutical companies usually comply with this. The patients entered in such trials are withdrawn from their usual antiepileptic medication and may suffer from seizures prior to the onset of effect of the new medication, which may or may not be able to control the seizures effectively.

An ongoing multicenter, randomized, double-blind, Bayesian adaptive, phase 3 comparative effectiveness trial, the “established status epilepticus trial (ESETT),” is designed to identify the best treatment for 35% to 45% of status epilepticus patients who are resistant to benzodiazepine, the first-line treatment (NCT01960075). Three arms of the trial will compare fosphenytoin, levetiracetam, and valproic acid. ESETT is a Bayesian adaptive design. Proposed total sample size is 795, which provides 90% power to identify the most effective and/or the least effective treatment when 1 treatment arm has a true response rate of 65% and the true response rate is 50% in the other 2 arms (07).

Amyotrophic lateral sclerosis. The World Federation of Neurology has developed guidelines for the optimal clinical evaluation of potential therapies for amyotrophic lateral sclerosis. These guidelines capture the viewpoints of the FDA and of academic neurologists from North America and Europe:

The first endpoint is unrealistic because a cure is unlikely at the present state of knowledge. The patients usually have lost 80% of the neurons before the diagnosis is established. Prevention of further deterioration is a measure of efficacy (even though no improvement can be demonstrated), and prolongation of life may be used as a criterion. World Federation of Neurology guidelines address the issue of survival as an important measurement technique in the evaluation of potential therapies for amyotrophic lateral sclerosis. However, mere prolongation without improvement in the quality of life may not make a strong case for the new drug. Improvement of quality of life and objective measures of neurologic assessment should be incorporated to demonstrate advantages over the existing treatments.

From the point of view of both the practicing neurologists and the patients with amyotrophic lateral sclerosis, the following effects would be desired from a new treatment:

Consideration should be given to diagnosis by MRI. Any changes seen in amyotrophic lateral sclerosis are diagnostic and correlate with the rate of disease progression. MRI provides documentation of neuronal loss both into the brain and the spinal cord. However, inclusion of MRI in a clinical trial for amyotrophic lateral sclerosis patients would increase the cost considerably.

There are many variations in the clinical trial design for amyotrophic lateral sclerosis, which has made the comparison of treatments difficult. There is a trend towards shorter duration trials with smaller numbers so that more drugs can be tested. New outcome measures have been developed that have reduced the sample size requirement as compared with survival studies.

Alzheimer disease. In Alzheimer disease pivotal phase 2 and phase 2a clinical trials usually involve those diagnosed as probable Alzheimer disease patients. In patients with suspected Alzheimer disease, virtual clinical trials using wearable devices and unobtrusive passive sensors enable the collection of frequent or continuous, objective, and multidimensional real-time data during daily activities of life and may detect subtle changes in cognition and functional capacity much earlier than the onset of dementia (19).

Guidelines for the evaluation of antidementia drugs have been prepared by the FDA. Efficacy of most of the cholinesterase inhibitors currently in use was established by parallel group studies. Time to reach clinical milestones is an additional feature in useful trial designs for new studies that aim to delay progression of the disease. Several clinical trials of Alzheimer disease have either failed or have been discontinued. Lessons can be learned from failed trials that should influence clinical trial design, selection of participants, and choice of study outcome measures.

Several treatments under development aim to slow the rate of progression or delay the time of onset of Alzheimer disease. Clinical trials designed to prevent the development of Alzheimer disease in patients with mild cognitive impairment have been initiated. Neuroimaging-based strategies for measuring the rate of progression of Alzheimer disease may be useful in addition to the behavioral instruments typically used in these studies.

Parkinson disease. In order to rapidly identify promising candidate drugs for definitive assessment of efficacy, a series of relatively small phase 2 trials involving less than 100 patients should be supported to establish the safety, tolerability, and dosage of new drugs. Trials of Parkinson disease involve glial cell line-derived neurotrophic factor, the monoamine oxidase type B inhibitor rasagiline, glutamate antagonists, inhibitors of the dopamine transporter, and neuroimmunophilins. Results of such studies would be used to determine the feasibility of definitive long-term phase 3 trials involving 500 to 2000 patients. Because Parkinson disease is heterogeneous, subgroups of these populations may need to be studied for nonmotor symptoms such as cognitive impairment and depression to identify which subpopulations respond best to a specific intervention. Prevention trials should be done in large families with identified genetic markers for Parkinson disease.

Positive effects of placebo-treatment are well-known, and differentiation of a disease-modifying therapy from symptomatic relief may be difficult to determine in a neurodegenerative disorder such as Parkinson disease. Imaging of dopaminergic pathways as a biomarker is extremely expensive and not always practical in clinical trials. A rapid and inexpensive approach was used in a proof-of-concept clinical trial of exenatide, an approved diabetes treatment that has been shown to have neuroprotective properties in preclinical studies (03). A single-blind trial design evaluated the progress of patients with Parkinson disease randomly assigned to receive exenatide injections for 12 months or to act as controls. Blinded video assessment and use of clinical rating scales during the study and in the following 2 months wash-out period showed better results in treated patients as compared to controls.

Traumatic brain injury. Clinical trials in severe traumatic brain injury are problematic for many reasons, including a short treatment window, uncertain passage of therapeutics across the blood-brain barrier, and inability of unconscious patients to give consent. A review of the 100 most cited clinical trials of traumatic brain injury revealed that they did not produce scientifically significant results possibly due to high variability of patients, which may compromise the reliability or generalizability of the study findings (60). The authors recommended that future investigators should be encouraged to pay attention to the principles of clinical research to improve the quality of traumatic brain injury studies and achieve a higher level of evidence.

Backache. There have been several clinical trials of therapies for chronic low back pain but there is less success in the development of treatments for neuropathic low back pain. Failure to identify effective treatments is likely due to poor design of clinical trials for neuropathic low back pain because of the lack of specific clinical signs and imaging findings that can be addressed by outcome measures based on symptoms that matter most to patients (32).

Use of clinical trial registries for further research in neurologic disorders

Apart from www.ClinicalTrials.gov, several other registries are maintained in other countries (eg, European Union, Japan, and Australia). In addition, there are registries for specific diseases maintained by patient organizations that serve as source of referrals for clinical trials. All these sources provide data that can be used for further research.

Future of clinical trials in neurology

The drug development cycle will be shortened to 6 to 8 years in the postgenomic era with full application of genomic and proteomic technologies. With application of pharmacogenomics and pharmacogenetics, the clinical trials will be stratified so that the drugs deemed most effective and safe for a certain group of patients will be allocated to such patients as personalized medicine. A lesser number of patients need to be recruited in such trials. It is conceivable that phase 3 trials may not be necessary for approval in such an environment. The safety of the drugs will be tested at the preclinical stage and potentially toxic drugs will not be administered to patients who are liable to suffer ill effects from them. There would be less concern for adverse reactions in the postmarketing surveillance.

Failure rate of clinical trials in neurology is still high, and some of the suggestions for improving neuroscience trials of the future in a workshop include the following (47):

Clinical trials in neurosurgery

In contrast with nonsurgical treatments, clinical trials of neurosurgical procedures have been limited. A systematic analysis of study design, quality of reporting, and trial results of neurosurgical randomized controlled trials against another procedure, nonsurgical treatment, or no treatment has shown that design and reporting have improved over time, but better powered and accurately reported trials are needed to obtain evidence for achieving optimal outcomes (04). Confounders, which are variables associated with the risk factors and the outcome, should not be mistaken for effect modification, particularly in neurosurgical clinical trials. A confounder affects the association between a potential predictor and an outcome. Modifications in the design of neurosurgical research studies can control for confounders, but if it is not possible, data analysis techniques can provide an effective control (57).

Because of the problems with CNS drug delivery, neurosurgical trials offer an opportunity to test promising agents with demonstrable brain tumor penetration and molecular effects. In brain tumor drug development, however, phase 0 study design poses significant limitations due to the absence of predictive animal models. Adaptation of the phase 0 trial design by the following specific modifications for neurooncology patients is an effective avenue to obtain direct evidence (51).