CIOMS VI - Management Of Safety Information From Clinical Trials

CIOMS VI - MANAGEMENT OF SAFETY INFORMATION FROM CLINICAL TRIALS

RATIONALE

Pharmacovigilance has traditionally focused on detection and evaluation of signals in the post-approval environment in order to secure early detection of new adverse reactions or patient subgroups of exceptional sensitivity, and to introduce measures to manage those risks. It was and remains the vision of CIOMS VI even though there are some important differences between pre-marketing and post-marketing safety monitoring and management that there should be a much stronger and closer relationship between them. In providing practical, and sometimes completely new approaches for managing safety information in the clinical trial setting, CIOMS VI enables a more seamless tran-sition in conducting high quality pharmacovigilance from the development stage to the post-approval period. The sixth CIOMS working group addressed the collection, monitoring, analysis, evaluation and overall management of safety information from clini-cal trials. The output of the CIOMS VI working group is dedicated to the many thousands of patients and other volunteers who generously participate in clini-cal research programmes so vital for the development and advancement of medicines.

PROCESS

In 2000–2003, drug regulatory authorities, pharma-ceutical companies and clinical investigators were challenged by several new national, regional and inter-national guidelines and regulations, including those dealing with ethical aspects of biomedical research. Implementation of ICH Guideline E6 on GCP was completed, the World Medical Association’s Declara-tion of Helsinki was revised in 2000 (and subsequently clarified in 2002 and 2004), the European Commis-sion published the Clinical Trials Directive in 2001 and its guidances in 2003, and CIOMS published the revised International Ethical Guidelines for Biomed-ical Research Involving Human Subjects in 2002. Moreover, the working group reviewed new develop-ments in drug safety regulations and concepts and in risk-management put forth by the US FDA and the EU EMEA. Similarly, it was also kept up to date on new initiatives in Japan, Australia and South Amer-ica. All these aspects are reflected or referred to in the final report of the working group.

A survey of pharmaceutical companies on their safety practices during clinical trials was conducted in early 2003; the results of that survey helped inform the working group’s deliberations.The topics covered in the survey included broad organisation and policy issues (regarding, e.g., risk management, Investigator’s Brochure management) as well as case processing and data management issues (e.g. causality assessment, study/case blinding, use of AE terms and coding dictionaries, and much more).

OVERVIEW

The report is divided into the following six main subject areas:

•   Ethical Considerations for Clinical Trial Safety Management.

•   Good Pharmacovigilance and Risk Management Practices: Systematic Approach to managing safety during clinical development.

•   Collection and management of safety data during clinical trials.

•   Identification and evaluation of risk from clinical trial data.

•   Statistical analysis of safety data in clinical trials.

•   Regulatory reporting and other communication of safety information from clinical trials.

Ethical Considerations for Clinical Trial Safety Management

The key messages of this chapter are that for anyone designing and conducting a clinical trial, the funda-mental principle should be that any study that is not scientifically sound can be considered unethical.

It also endorses the concept of transparency of results and outcomes for all clinical research, espe-cially safety data.

Systematic Approach to Managing Safety During Drug Development

This chapter is destined to become one of the most influential pharmacovigilance texts in recent memory. It recommends that sponsors of clinical trials ensure that a well-defined and well-structured process is in place that will allow them to readily identify, evaluate and minimize potential safety risks relative to poten-tial benefits for study subjects in pre-approval trials. Such a process should start before initiating the first Phase I study and continue through post-approval use of the drug or biologic in the general population.

A dedicated Safety Management Team (SMT) should be formed for each development programme, to review all the available safety information on a regular basis so that decisions on safety can be made in a timely manner. It also recommends that these reviews generally take place at least quarterly pre-approval and be co-ordinated with pre-approval and, if applicable, post-approval periodic reporting. Quar-terly and ad hoc safety reviews should consider the overall evolving safety profile of the investigational product, make necessary changes to the Investiga-tor’s Brochure (IB), Development Core Safety Infor-mation (DCSI) and informed consent, determine if any changes to the conduct of the trials need to be considered, and initiate prompt communications to investigators, ethics committees and regulators when appropriate. The team should be empowered to make decisions that will accomplish the goal of minimizing risk while maximizing benefits to subjects in clinical trials, as well as anticipating the use of the product once marketed.

A formal Development Risk Management Plan (DRMP) should be created and modified as needed during a clinical programme. In the initial planning stages of a new clinical development programme, one goal is to gather the necessary knowledge and infor-mation to adequately plan the optimum programme from the standpoint of safety. The plan should include early documentation of known, anticipated and poten-tial risks along with plans for addressing them during development and, where appropriate, the DRMP would eventually evolve into a post-marketing risk management plan that will accompany the registration application.

All pertinent data must be readily available to the safety team from the clinical trial and safety databases as well as from other relevant sources, such as the pre-clinical toxicology department (e.g. carcinogenic-ity and development and reproductive toxicology), in vitro mutagenicity studies, and pharmacokinetic and drug-interaction studies.

It is important to incorporate epidemiology into the development planning process, not only for defining the natural history of the disease being treated, but for anticipating important confounding factors and back-ground rates of occurrence of concurrent illnesses. Understanding these will help to put the evolving safety profile into proper perspective.

When planning for the development of virtually any new medicinal product, there are certain cate-gories of potential toxicities that should always be considered. These include abnormalities in cardiac conduction, hepatotoxicity, drug–drug interactions, immunogenicity, bone marrow toxicity and reactive metabolite formation.

Collection and Management of Safety Data During Clinical Trials

In early phases of drug development, it is generally necessary to collect more comprehensive safety data than in post-marketing studies. In addition, certain drug types may require longer routine follow-up as in the case of vaccines, immunotherapies and some biotechnology products. As a general rule, it is recommended that safety data event-collection should continue after the last dose of the drug for at least a further five half-lives of the experimental product. Also, investigators should be instructed to always be diligent in looking for possible latent safety effects that may not appear until after a medication is discontinued.

There are no definitive methods for distinguish-ing most adverse drug reactions (i.e. events that are causally attributable to study therapy) from clinical adverse events that occur as background findings in the population and have only a temporal association with study therapy. The CIOMS VI Working Group thus recommends that:

All adverse events, both serious and non-serious, should be collected for any clinical trial during development, regardless of presumed relationship to the study agent by the investigator or sponsor, in order to allow for subsequent assessment of causal-ity using standardized methods for individual cases and aggregate data. This applies not only to the experimental product but to placebo, no treatment or active comparator.

•   Causality judgments based on analysis of multi-ple cases/aggregate data, rather than on individual cases, are almost always more meaningful and typi-cally have a greater impact on the conduct of clin-ical trials, including changes to informed consent documents, study design and core safety informa-tion. However, causality assessment of individual adverse events by the investigator may play a role in the early detection of significant safety prob-lems, and these are the only source of information on rare events.

•   The investigator should be asked to use a simple binary decision for drug causality (related or not related) for serious adverse events. One possi-ble approach that has been suggested is to ask simply whether there is a ‘reasonable possibility’ or ‘no reasonable possibility’ that the study treat-ment caused the event. Alternatively – ‘Was there a reasonable possibility?’ Yes or No.

In order to assure standardized signal detection and evaluation processes, data quality and completeness are paramount. The CIOMS VI Working Group recommends the following principles for this impor-tant objective:

•   Individual case safety reports from studies should be as fully documented as possible.

•   There should be diligent follow-up of each case, as needed.

•   The reporter’s verbatim AE terms must be retained within all relevant databases.

Sponsors should avoid ‘excessive coding’ of events reported in serious adverse event cases. Each such report should contain only the minimum number of dictionary terms needed to ensure retrieval in the rele-vant clinical context(s). Conversely, sponsors should take great care not to ‘undercode’ events, namely assign codes that might downgrade the severity or importance of an event term or terms. Some compa-nies and health authorities maintain a list of event terms that are always regarded as medically seri-ous and important even if the specific case might not satisfy the criteria for serious in a regulatory sense (require expedited reporting, for example). Such ‘always serious’ events are used routinely to trigger special attention and evaluation. Although such lists were originally created for post-marketing purposes, especially for spontaneous reports, they might be useful for pre-approval clinical research purposes.

Identification and Evaluation of Risk from Clinical Trial Data

The purpose of ongoing safety evaluation during drug development is to ensure that important safety signals are detected early and to gain a better understanding of the benefit–risk profile of the drug. Safety monitoring, evaluation and analysis should be performed in such a manner as not to compromise the integrity of the indi-vidual studies or the overall development programme. Study sponsor should be fully aware at every stage of development of the potential risks of the investiga-tional product and the morbidities characteristic of the study population. They must also ensure that activi-ties involved in the management of clinical trial safety data (e.g. data entry, edit checks, data queries, coding of adverse events using a standard dictionary) are undertaken with care and precision in order to ensure that the safety database is accurate and complete.

The working group recommended that frequent review of serious and special interest adverse events, as well as overall assessment of all AEs, regardless of seriousness, causality or expectedness, should be performed periodically:

•   ad hoc, for serious and special interest AEs;

•   routine, periodic, general review of all data, whose frequency will vary from trial to trial and from development programme to development programme and depend on many factors; and

•   reviews triggered by specific milestones estab-lished for a trial or a programme (e.g. numbers of completed patients, end-of-trial, end-of-program, preparation of integrated summary of safety, and a marketing application).

Aggregate safety data should be monitored and eval-uated periodically during the course of the overall developmental programme, during each study, and at the end of every study to provide an ongoing appraisal of benefit–risk balance.

Statistical Analysis of Safety Data in Clinical Trials

Use of the most appropriate statistical techniques for analysis and display of the data are essential for plac-ing the absolute and relative safety of a medicinal product in proper perspective. Early in drug devel-opment (Phase I and early Phase II trials), much of the assessment of safety depends on individual case assessment. However, as the database increases, aggregate analysis tends to become more important, and that is where statistics play a crucial role. The techniques and approaches to use of statistics for analysing safety data have not been developed as fully as they have for efficacy and it is not uncommon to find inappropriate or incomplete displays and analysis of adverse event data, even in refereed publications.

This chapter is not intended to be a manual for statistical analysis of safety data as the subject is much too broad and complex. However, it does highlight key points that need attention when considering anal-ysis, and areas which the working group believed may not be adequately understood or appreciated.

Statistical approaches have application at several stages of clinical trials: protocol design, during a trial, for final analysis and writing of the trial report or publication, and when combining data across differ-ent trials. Professional statistical help is required and should be obtained at each of those stages. Statis-tical association (P-values or other measures) alone may or may not be of clinical value. In random-ized trials they have great strength in testing causality but they inevitably have uncertainty. Examination of both statistical and clinical significance must involve a partnership.

The ability of a study to detect causal effects in the face of variation within and between individu-als is dependent on sample size; the smaller or rarer an effect, the larger the sample size required, if any degree of certainty is to be given to the study conclu-sions. It is necessary to acknowledge when the data are insufficient to draw conclusions on safety, i.e. ‘absence of evidence is not evidence of absence’. In such situations, the use of descriptive methods and well-designed graphics will be helpful in this process. Finally, although this chapter concentrates on ways of graphically representing safety data, the recom-mendation is that the unwanted effects must always be considered in the context of the benefits of the medicine.

Regulatory Reporting and Other Communication of Safety Information from Clinical Trials

This chapter makes a number of detailed recommenda-tions on the expedited reporting process for clinical trial reports. More importantly and contrary to established regulations, CIOMS VI proposes that routine expedited case reporting by sponsors to investigators and Indepen-dent Ethics Committees (IECs)/Institutional Review Boards (IRBs) be eliminated. Instead, sponsors should provide regular updates of the evolving benefit–risk profile and highlight important new safety information. Only significant new information, occasionally a single case report, that has implications for the conduct of the trial or warrants an immediate revision to the informed consent would be communicated on an expedited basis. More commonly, important new safety information would be communicated periodically, based on the assessment of accumulating, aggregate information.

For unapproved products, instead of sending indi-vidual expedited clinical trial case reports to investi-gators and IECs/IRBs, the CIOMS VI Working Group recommended periodic reporting. It was suggested that such reports include a line listing of unblinded clinical trial cases that were expedited to regula-tory authorities since the last periodic report, a copy of the current DCSI along with an explanation of any changes, a statement if there are no changes, and a brief summary of the emerging safety profile. Although it is recommended that the default would be quarterly updates, there may be circumstances when either a more immediate or less frequent communica-tion would be appropriate.

For approved products, the time frame for peri-odic reports to investigators and IECs/IRBs would depend on the extent to which new indications are being developed. For a product undergoing Phase III trials, continuation of the quarterly reports would be advisable. For well-established products, less frequent updates would be appropriate and, at some point, there should only be a need to update investigators and IECs/IRBs when there is significant new infor-mation to report. For Phase IV investigators and their associated IECs/IRBs, communications of changes to the CCSI should be sufficient.

The working group proposed that there be a single Development Safety Update Report (DSUR) for submission to regulators on an annual basis, with a consistent format and content which were yet to be defined. Additional recommendations for the DSUR were taken up by the CIOMS VII working group (see CIOMS VII).

If a significant safety issue is identified, either from an individual case report or review of aggre-gate data, then the sponsor should issue a prompt notification to all parties, namely regulatory author-ities, investigators, IECs/IRBs and, if relevant, Data and Safety Monitoring Boards (DSMBs). A signifi-cant safety issue could be defined as one that has a significant impact on the course of the clinical trial or programme (including the potential for suspension of the trial programme or amendments to protocols) or warrants immediate update of informed consent. DSMBs are most commonly employed for a single large clinical trial and are not usually charged with providing oversight of an entire clinical program. It would therefore be important to ensure that important new safety information is communicated to a DSMB even if the information did not originate from the DSMB-monitored study.

There was much discussion on whether the previ-ously recommended concept and level of threshold for changes to the CCSI (CIOMS III/V report) should be applied to the DCSI and informed consent informa-tion. Although there was agreement on the concept, there was not agreement on the threshold. As reflected in Appendix 7 of the report, there was a body of opin-ion that fewer and less stringent criteria for including new ADR information in the DCSI be applied for events that might have a significant adverse outcome for the trial population. This opinion was not reflected in the main chapter.

The CIOMS VI report was published in 2005 (CIOMS, 2005).