Search

Clinical trials often last several months or even several years. As the trial progresses, it can be tempting to find out whether the data obtained already answers the question posed at the start of the trial in order to stop inclusions or monitoring earlier. However, knowing and taking into account interim results can sometimes compromise the integrity of the results, which is counterproductive. To minimise this risk and ensure that the treatments are assessed reliably, safety and/or efficacy criteria are monitored during the study by a Data Monitoring Committee. After receiving the results confidentially, the Data Monitoring Committee assesses the benefit/risk ratio of the study treatment and recommends that the trial be continued, modified or terminated. Data Monitoring Committee members issuing these recommendations have an important responsibility: a hasty decision to end the trial may lead to inconclusive results unable to answer the initial question and, inversely, delaying the decision to end the trial may expose the subjects to potentially ineffective or even harmful interventions. The Data Monitoring Committee's task is therefore particularly complex. With this in mind, the round table discussion at the Giens workshops was a chance to review the scientific justification for creating Data Monitoring Committees and to recall the need for their members to receive comprehensive training on the complexities of multiple analyses, confidentiality requirements applying to the results and the need for them to be aware that recommendations to end a trial must be based on data that is robust enough to assess the benefit/risk ratio of the treatment studied., (Copyright © 2023. Published by Elsevier Masson SAS.)

Evidence-based medicine is the cornerstone of shared-decision making in healthcare today. The public deserves clear, transparent and trust-worthy information on drug efficacy. Yet today, many drugs are prescribed and used without solid evidence of efficacy. Clinical trials and randomised clinical trials (RCTs) are the best method to evaluate drug efficacy and side effects. In a shared medical decision-making approach, general practitioners need drug assessment based on patient-important outcomes. The aim of project rebuild the evidence base (REB) is to bridge the gap between the data needed in clinical practice and the data available from clinical research. The drugs will be assessed on clinical patient important outcomes and for a population. Using the Cochrane tools, we propose to analyse for each population and outcome: 1) a meta-analysis based on RCTs with a low risk of bias overall; 2) an evaluation of results of confirmatory RCTs; 3) a statistical analysis of heterrogeneity between RCTs and 4) an analysis of publication bias. Depending on the results of these analyses, the evidence will be categorized in 4 different levels: firm evidence, evidence (to be confirmed), signal or absence of evidence. Project REB proposes a method for reading and interpreting RCTs and their meta-analysis to produce quality data for general practitioners to focus on risk-benefit assessment in the interest of patients. If this data does not exist, it could enable clinical research to better its aim., (Copyright © 2022 Société française de pharmacologie et de thérapeutique. Published by Elsevier Masson SAS. All rights reserved.)

Introduction: Data on adverse drug reactions (ADRs) of immune checkpoint inhibitors (ICIs) used in oncology are mainly derived from clinical trials or cancer-specific reviews. We aim to analyze ADRs that occurred in patients treated with ICIs in real life., Materials and Methods: We conducted an observational study on a historical cohort of the University Hospitals of Lyon. All patients who initiated an ICI treatment for any cancer in 2017 were included. Patients were followed from the first infusion until 90 days after the last one, death, date of last news or end of the study period (28 February 2019), whichever came first. Two pharmacovigilance specialists assessed the accountability and the severity of each ADR using Naranjo algorithm and common terminology criteria for adverse events (CTCAE) classification, respectively., Results: 248 patients were included. They were treated with anti-PD-(L)-1, mainly nivolumab (70.6%) and pembrolizumab (25.8%). Lung cancer (62.1%) and melanoma (20.2%) were the most represented cancers. 139 ADRs occurred in 93 patients (37.5%), on average at the 6 th cure (±6.8). ADRs mainly concerned skin (29.5%), endocrine (19.4%) and digestive (10.8%) systems. 17.3% of ADRs were grades III-V and two patients died because of ADRs. By comparing patients with (N=93) or without (N=155) ADRs, all characteristics appeared similar except for age, number of infusions received and death status. The spontaneous notification rate found in this study was 5.8% for all grade ADRs (N = 8) but raised to 23.8% when only grades higher than III were considered (N = 5)., Discussion/conclusion: Our results are consistent with literature data in frequency and type of serious ADRs. We found a lower frequency of ADRs of any grade, which could be explained by a fairer causality assessment in our study than in clinical trials., (Copyright © 2022 Société française de pharmacologie et de thérapeutique. Published by Elsevier Masson SAS. All rights reserved.)

For the past few years, platform trials have experienced a significant increase, recently amplified by the COVID-19 pandemic. The implementation of a platform trial is particularly useful in certain pathologies, particularly when there is a significant number of drug candidates to be assessed, a rapid evolution of the standard of care or in situations of urgent need for evaluation, during which the pooling of protocols and infrastructure optimizes the number of patients to be enrolled, the costs, and the deadlines for carrying out the investigation. However, the specificity of platform trials raises methodological, ethical, and regulatory issues, which have been the subject of the round table and which are presented in this article. The round table was also an opportunity to discuss the complexity of sponsorship and data management related to the multiplicity of partners, funding, and governance of these trials, and the level of acceptability of their findings by the competent authorities., (Copyright © 2022. Published by Elsevier Masson SAS.)

Background: Rehabilitation aims to improve hand-arm function, upper-limb strength, and functional independence that has been impaired by tetraplegia. On the basis of evidence derived from stroke rehabilitation, interventions aiming to increase intensity (i.e., duration and/or number of movements practiced) or alter brain plasticity (including motor imagery, virtual reality, transcranial direct-current or magnetic stimulations; i.e., neuromodulation) are now used during tetraplegic rehabilitation. However, no meta-analysis has investigated the efficacy of these interventions., Objective: This systematic review and meta-analysis investigated, separately, the efficacy of these interventions to alter hand-arm function, upper-limb strength, and functional independence of individuals with tetraplegia., Methods: Two independent reviewers followed the PROSPERO protocol (CRD42018098506) for this systematic review. MEDLINE, PEDro CENTRAL, and SCOPUS databases were searched for reports of randomized controlled trials of individuals with tetraplegia that were published in English. We performed a meta-analysis of intensive versus less intensive interventions and neuromodulation versus sham interventions considering hand-arm function, strength, and functional independence., Results: From 168 records identified, we included 29 studies (all but 1 were single-centre) in the systematic review (647 participants with C2 to T1 tetraplegia [American Spinal Injury Association impairment scale A to D]). Interventions lasted from 66 to 40,320min. Five studies were retained in the intensity meta-analyses and 5 in the neuromodulation meta-analyses. Overall, 3/5 and 1/5 studies had adequate methodology (Cochrane Risk of Bias score ≥6/10). For each outcome, the p-values for the overall effect were>0.05. Heterogeneity was low, but when analyzing intensity, it was moderate for functional independence and high for hand-arm function. Quality of evidence was very low to low., Conclusions: We can provide no recommendations for using intensive versus less intensive interventions or neuromodulation versus sham during tetraplegia rehabilitation. Further multicentre studies of high methodological quality are required to reduce uncertainty about the efficacy of these interventions., (Copyright © 2019 Elsevier Masson SAS. All rights reserved.)

Single-arm studies are sometimes used as pivotal studies but they have methodological limitations which prevent them from obtaining the high level of reliability as for a randomised controlled study which remains the gold standard in the evaluation of new treatments. The objective of this roundtable was to discuss the limitations of these single-arm studies, to analyse available and acceptable solutions in order to propose guidelines for their conduct and assessment. Single-arm studies themselves are intrinsically inappropriate for demonstrating the benefit of a new treatment because it is impossible to infer the benefit from a value obtained under treatment without knowing what it would have been in the absence of the new treatment. The implication is that comparison with other data is necessary. However this comparison has limitations due to (1) the post hoc choice of the reference used for comparison, (2) the confusion bias for which an adjustment approach is imperative and, (3) the other biases, measure and attrition among others. When these limitations are taken into account this should, first and foremost, lead to the conduct of externally controlled trials instead of single-arm trials as is proposed by the latest version of ICH E10. Moreover, the external control must be formalised in the study protocol with a priori selection of both the reference control and the formal method of comparison: test in relation to a standard, adjustment on individual data, a synthetic control group or matching-adjusted indirect comparisons (MAIC). Lastly, externally controlled studies must be restricted to situations where randomisation is infeasible. To be acceptable, these studies must be able to guarantee freedom from residual confusion bias, which is only truly acceptable if the observed effect is dramatic and the usual course of the disease is highly predicable., (Copyright © 2019. Published by Elsevier Masson SAS.)

After looking at some case studies, we will remind in this paper how observational studies regarding drug exposure or delivery are prone to various biases and structural limitations. These biases lead us to be extremely cautious regarding the implementation of results from such studies in clinical practice, and to question the reliability of such studies to determine the position of a given treatment into the therapeutic strategy. We will conclude on the respective place of randomized controlled trials (RCTs) and observational approaches, which are obviously complementary, but not interchangeable., (Copyright © 2018 Société française de pharmacologie et de thérapeutique. Published by Elsevier Masson SAS. All rights reserved.)

The use of statistical test is central in the clinical trial. At the statistical level, obtaining a P<0.05 allows to claim the effectiveness of the new studied treatment. However, given its underlying mathematical logic the concept of "P value" is often misinterpreted. It is often assimilated, mistakenly, to the likelihood that treatment is ineffective. Actually the "P value" gives an indirect information about the plausibility of the existence of treatment effect. With "P<0.05", the probability that the treatment is effective may vary depending on other statistical parameters which are the alpha level of risk, the power of the study and especially the a priori probability of the existence of treatment effect. A "P<0.05" does not always produce the same degree of certainty. Thus there exist situations where the risk of a result "P<0.05" is in reality a false positive is very high. This is the case if the power is low, if there is an inflation of the alpha risk or if the result is exploratory or chance discoveries. This possibility is important to take into consideration when interpreting the results of clinical trials in order to avoid pushing ahead significant results in appearance, but which are likely to be actually false positive results., (Copyright © 2017 Société française de pharmacologie et de thérapeutique. Published by Elsevier Masson SAS. All rights reserved.)

Medical devices are characterized notably by a wide heterogeneity (from tongue depressors to hip prostheses, and from non-implantable to invasive devices), a short life cycle with recurrent incremental innovations (from 18 months to 5 years), and an operator-dependent nature. The objective of the current round table was to develop proposals and recommendations concerning the prerequisites needed in order to meet the French health authorities expectations concerning requests for post-approval studies for medical devices, required in cases where short and long-term consequences are unknown. These studies, which are the responsibility of the manufacturer or the distributor of the medical device, are designed to confirm the role of the medical device in the therapeutic management strategy in a real-life setting. There are currently approximately 150 post-approval studies underway, mainly concerning class III devices, and the majority face difficulties implementing the study or meeting the study objectives. In light of this, the round table endeavored to clearly identify the conditions for implementation of post-approval studies specific to the characteristics of medical devices. Various areas of progress have been envisaged to improve the performance of these studies, and by consequence, the efficiency of reimbursement of medical devices by the national health insurance. These include providing manufacturers with the opportunity to better anticipate post-approval requirements, defining a study-specific primary objective, integrating a phase allowing dialogue between the manufacturer, the health authorities and the scientific committee, and increasing awareness and training of health professionals on the impact of post-approval clinical studies in terms of the reimbursement of medical devices by the national insurance., (© 2014 Société Française de Pharmacologie et de Thérapeutique.)

Background: Infusion of saline attenuates the decrease in renal function induced by radiographic contrast agents among patients with chronic renal insufficiency., Aim: The Preventing Renal alteration in Coronary Disease (PRECORD) trial was a randomized trial to assess the effect on renal function of saline infusion during and after coronary angiography in 201 patients without severe chronic renal insufficiency (serum creatinine<140micromol/L)., Methods: All patients received standard oral hydration: 2000mL of tap water within the 24 hours after coronary angiography. Patients were randomized before the procedure to intravenous hydration (1000mL of 0.9% saline infusion) or no additional hydration. The infusion was started in the catheterization laboratory and continued for 24 hours. The primary endpoint was the change in calculated creatinine clearance between baseline and 24 hours after coronary angiography. The same ionic low osmolar radiographic contrast agent (ioxaglate) was used in all patients., Results: Both groups had similar baseline characteristics, including age, serum creatinine, volume of contrast and proportion of patients undergoing ad hoc coronary angioplasty. The overall decrease in serum creatinine clearance 24 hours after the procedure was -3.44 (0.68)mL/min. The change in serum creatinine clearance 24 hours after the procedure was -2.81 (1.07)mL/min in the infusion group vs -4.09 (0.91)mL/min in the control group (p=0.38)., Conclusion: Renal function is altered only slightly 24 hours after coronary angiography with standard oral hydration alone and is not affected by saline infusion started at the beginning of coronary angiography, even in patients with mild-to-moderate renal dysfunction.

Frequently, in data packages submitted for Marketing Approval to the CHMP, there is a lack of relevant head-to-head comparisons of medicinal products that could enable national authorities responsible for the approval of reimbursement to assess the Added Therapeutic Value (ASMR) of new clinical entities or line extensions of existing therapies.Indirect or mixed treatment comparisons (MTC) are methods stemming from the field of meta-analysis that have been designed to tackle this problem. Adjusted indirect comparisons, meta-regressions, mixed models, Bayesian network analyses pool results of randomised controlled trials (RCTs), enabling a quantitative synthesis.The REAL procedure, recently developed by the HAS (French National Authority for Health), is a mixture of an MTC and effect model based on expert opinions. It is intended to translate the efficacy observed in the trials into effectiveness expected in day-to-day clinical practice in France.

Noninferiority trials make it possible to show that the efficacy of a new treatment is not worse than that of the reference treatment, but not to conclude that efficacy is strictly equivalent. Even though a noninferiority trial is conclusive, there remains the possibility that the new treatment is inferior to the reference treatment (within the margin of the noninferiority margin set a priori, that is, before the results were known). A treatment shown not to be inferior does not represent therapeutic progress unless it has other advantages, in terms, for example, of tolerance or ease of use. Setting the noninferiority margin is a clinical and not a statistical problem. It requires determining the greatest acceptable loss of efficacy relative to the reference treatment. This loss of efficacy set by the noninferiority margin cannot exceed the efficacy of the reference treatment, since that could amount to loss of the entire treatment benefit.

Results of meta-analyses may differ from those of megatrials only because of the bias that could be introduced by a unadequate methodology of some meta-analyses. A meta-analysis must be based on an exhaustive review of the literature - of all studies, those with negative as well as positive results. The quality of a meta-analysis depends on the methodological quality of the studies it analyzes; accordingly they must be selected according to strict methodological inclusion and exclusion criteria, defined a priori. The value of a meta-analysis is that allows a treatment strategy to be assessed in populations more heterogeneous than those in clinical trials. Nonetheless the heterogeneity of protocols, of dosing, and of outcome measures can lead to bias of the treatment effect estimation. Different analytic techniques must be used to assess possible publication or selection bias (the so-called "funnel plot" method) and potential sources of heterogeneity (heterogeneity test, meta-regression, etc.). Meta-analyses of individual data make it possible to assess the treatment effect according to patient characteristics. Prospective meta-analyses or those planned before clinical trials can help to limit data heterogeneity by making study protocols less varied (treatment, follow-up, evaluation, etc.).

The complexity of pathophysiological mechanisms is beyond the capabilities of traditional approaches. Many of the decision-making problems in public health, such as initiating mass screening, are complex. Progress in genomics and proteomics, and the resulting extraordinary increase in knowledge with regard to interactions between gene expression, the environment and behaviour, the customisation of risk factors and the need to combine therapies that individually have minimal though well documented efficacy, has led doctors to raise new questions: how to optimise choice and the application of therapeutic strategies at the individual rather than the group level, while taking into account all the available evidence? This is essentially a problem of complexity with dimensions similar to the previous ones: multiple parameters with nonlinear relationships between them, varying time scales that cannot be ignored etc. Numerical modelling and simulation (in silico investigations) have the potential to meet these challenges. Such approaches are considered in drug innovation and development. They require a multidisciplinary approach, and this will involve modification of the way research in pharmacology is conducted.

The simulation of therapeutic models and clinical trial simulation have recently attracted attention as emerging techniques for developing new active molecules and the exploration of possible clinical trial results. Such approaches have benefited from fundamental progress in the development of 'in silico' models, as well as progress in nonlinear mixed-effect pharmacokinetic-pharmacodynamic models. Mixing the two approaches allows simulation of 'virtual' patients, who receive virtual treatments or placebo. These have various uses, such as proof of concept, decision analysis or experimental design optimisation. Also, the effect of departures from protocol on clinical trial results can easily be evaluated by the use of simulation. This technique is now implemented by the pharmaceutical industry for optimising phase II and III experimental designs when a good biomarker or a clinical outcome model is available, but the use of an in silico therapeutic model as a proof of concept is only just beginning. In order to see such methodologies used more widely in drug development, multidisciplinary efforts need to be initiated, new modelling and simulation tools developed, and sound modelling and simulation practice documents need to be adopted. A reduction in the number of failed clinical development projects, the number of negative phase II and III clinical trials, or in just their cost and duration, are among the expected benefits of modelling and simulation in clinical drug development.

Primary Prevention: A meta-analysis of the 7 available randomized cholesterol lowering trials (2 on statins, 2 on fibrates, 2 on resins and 1 on diet) demonstrates a significant 24% relative reduction in the frequency of fatal and non-fatal coronary events, leading to a significant 14% relative reduction in coronary disease related mortality. For total mortality, the statin trials were not comparable with other treatments where there was a trend to overmortality. With the 2 statin trials, there was a nonsignificant 13% reduction in total mortality., Secondary Prevention: A meta-analysis of the 13 available randomized trials (3 with clofribrate, 1 with gemfibrozil, 2 with nicotinic acid, and 4 with diet) showed a significant 21% reduction in the frequency of fatal and non-fatal coronary events, coronary mortality and total mortality. For total mortality, only the result of the 2 statin trials was significant (-20%). BENEFICIAL EFFECT OF STATINS, ABSOLUTE VALUES: Measured as the number of patients to treat for 5 years in order to avoid 1 event (NNT), there is a clinically pertinent benefit of secondary prevention (NNT = 15 for events, NNT = 30 for coronary mortality). The absolute value is weaker when patients with minimally elevated cholesterol levels are included. For primary prevention, the absolute value is low with NNT = 44 for events and NNT = 300 for coronary mortality. In addition, as these results were obtained in trials including populations with a much higher risk than the general French population, the absolute beneficial effect may not be pertinent in France except in high-risk patients who present, in addition to a high cholesterol level, other cardiovascular risk factors.

In chronic illness, when death or a non-fatal event can occur at any time, the current efficacy indices are no longer appropriate to express the effect of the treatment on the potential therapeutic objectives. The inappropriateness is not dependent on the effect model. Clues for solutions are proposed.

Efficacy indices do not contain the same information although they are all combinations of the same two quantities. Therefore, one should choose the proper index. Actually, none is entirely appropriate. Each more or less meets the specifications, depending on the underlying effect model for the therapy considered. However, one can say that the absolute benefit is more appropriate from the patient's point of view, the relative from the scientific point of view and the number of patients to treat from the policy maker's point of view. Nevertheless, this classification needs to be considered with caution. Finally, it emerges from the review that none is fully relevant to express the efficacy of a therapy, even in the most suitable condition, the acute illness.

The four indices for a binary outcome or therapeutic objective are: the odds ratio, the relative risk, the absolute benefit and the number of patients to treat. For a continuous outcome, the effect size is the best choice. The odds ratio approximates the relative risk. The difference may be large in some instances. The number of patients to treat is the reciprocal of the absolute benefit. Although they are built on the same two quantities, they are not interchangeable and should not be considered in the same way. Moreover, their meaning is not straightforward and they can be misused.

Efficacy indices measure the efficacy of therapies. They derive, by definition, from two quantities, the basal or control risk of event, Rc, observed in the control group, and the on-treatment risk, Rt, observed in the treated group. In clinical trials and meta-analyses, each is an unbiased measure of efficacy. Although they are a combination of frequencies, these indices are used in clinical practice to predict the benefit in treated patients. Their relevance to express efficacy depends on the type of clinical condition, and is better for acute diseases than for chronic diseases. In order to be useful for prescribers, they should meet certain specifications. In addition, they should be considered in the more general framework of effect models.

Meta-analyses of Furberg's original data (after correction of two minor errors) were performed using six different methods. Only three of them gave significant results at p < 0.05. The sensitivity analysis showed that taking into account some of the criticisms applied to the original meta-analysis did not change the results. When all the criticisms were considered together, the 95 per cent confidence interval of the odds ratio for mortality was [0.96; 1.31] instead of [1.06; 1.37] originally (p = 0.14 and p = 0.03 respectively), and the dose excess mortality relationship stressed by Furberg disappeared. When the selection of the studies to be entered in a meta-analysis is not straightforward, a sensitivity analysis should be performed.

The efficacy of a drug is a quantitative concept rather than a qualitative one. This quantity is expressed by several efficacy indices. None of them meet all the requirements. However, that of absolute benefit is especially suitable for the patients because it tells them the exact gain they can expect from taking the treatment. The absolute benefit varies according to patients' profiles because it interacts with some components of these profiles. In theory, such interactions can be used to predict the size of the absolute benefit for each patient, as well to describe better than with the current tools the therapy target population. We explain why meta-analysis and effect models are means of improving the prediction of the size of the effect and the definition of the therapy target population.

The effect models are defined as the simple or complex relation that the risk in the treatment group follows when the risk in the control group varies. The standard statistical methods of meta-analysis are based on simple effect models. The use of these methods could induce inaccurate or erroneous results in more complex situations. In this case, it is necessary to adopt a more appropriate effect model, such as the linear effect model. The properties of this kind of model allow the possibility that a treatment can be beneficial and harmful at the same time, in function of the risk without treatment. From this observation, it is advisable to be careful with the use of simple effect models in meta-analysis which can submerge interesting information in the synthesis.

The aim of therapeutic information is that people who need it have access to data of the highest level of evidence. In this context, a message is defined as the medium of information. The conditions for an ideal message to the prescriber are described: its content, the qualities it must meet, and the functional issues it will tackle. An example illustrates the importance of the form of a message. The difference between recommendation and message is highlighted.

In this paper we shall present the general principles of meta-analysis and will then discuss the various factors needed to evaluate a meta-analysis: description of the problem; definition of the outcome(s) (primary and secondary); methods for identifying and selecting trials for inclusion; statistical methods used; and the presentation and discussion of the results. We shall then examine other problems such as the detection of bias, the validity of the information provided by the meta-analysis, the problem of heterogeneity, the sensitivity and robustness of the meta-analysis, quality criteria for a meta-analysis, and how to locate published meta-analyses. Finally we present a decision algorithm which should help answer the question: should and can the results from the meta-analysis be integrated into clinical practice?