Central Institutional Review Board Review for an Academic Trial Network

Abstract

Problem

Translating discoveries into therapeutics is often delayed by lengthy start-up periods for multicenter clinical trials. One cause of delay can be multiple institutional review board (IRB) reviews of the same protocol.

Approach

When developing the Network for Excellence in Neuroscience Clinical Trials (NeuroNEXT; hereafter, NN), the National Institute of Neurological Disorders and Stroke (NINDS) established a central IRB (CIRB) based at Massachusetts General Hospital, the academic medical center that received the NN clinical coordinating center grant. The 25 NN sites, located at U.S. academic institutions, agreed to required CIRB use for NN trials.

Outcomes

To delineate roles and establish legal relationships between the NN sites and the CIRB, the CIRB executed reliance agreements with the sites and their affiliates that hold federalwide assurance for the protection of human subjects (FWA); this took, on average, 84 days. The first NN protocol reviewed by the CIRB achieved full approval to allow participant enrollment within 56 days and went from grant award to the first patient visit in less than four months. The authors describe anticipated challenges related to institutional oversight responsibilities versus regulatory CIRB review as well as unanticipated challenges related to working with complex organizations that include multiple FWA-holding affiliates.

Next Steps

The authors anticipate that CIRB use will decrease NN trial start-up time and thus promote efficient trial implementation. They plan to collect data on timelines and costs associated with CIRB use. The NINDS plans to promote CIRB use in future initiatives.

Problem

The translation of discoveries into therapeutics is often delayed by lengthy start-up periods in multicenter clinical research trials. One cause of delay can be multiple institutional review board (IRB) reviews of the same protocol.

Federal regulations require that every site engaged in human subjects research either (a) have a local IRB to approve that research or (b) delegate IRB review to an external IRB via a formal reliance agreement (RA). Local IRB reviews can usually be completed efficiently for research projects that are being conducted at only one or at a few sites. However, local IRB reviews for projects with more than a few sites have been associated with delay and administrative burden. Wagner et al¹ found central IRB (CIRB) affiliation to be associated with cost savings, decreased IRB and investigator effort, and faster reviews. Additionally, it is unclear whether multiple local reviews add value toward the protection of human subjects. Concerns have been raised that review by multiple IRBs may decrease the chance that needed protocol changes will be attempted, because no single IRB takes charge.²

When launching the Network for Excellence in Neuroscience Clinical Trials (NeuroNEXT; hereafter, NN), the National Institute of Neurological Disorders and Stroke (NINDS) took an innovative approach to streamline IRB review for the network’s multicenter studies: it established a CIRB based in an academic medical center. The NN is an NINDS-funded network to support high-quality biomarker studies and phase 2 clinical trials in neurology. It consists of 25 sites, a data coordinating center, and a clinical coordinating center that hosts the CIRB. In this report, we describe the use of the CIRB in this multisite research network.

Approach

In September 2011, the NINDS initiated the creation of a CIRB for the NN. The grant solicitation for the clinical coordinating center asked for a plan detailing how the CIRB would be operated. The solicitation for NN sites stated a preference for institutions that indicated a willingness to rely on review by a CIRB. Because it was unknown how institutions would react to this statement, three “tiers” of IRB review were proposed (local, combined, or CIRB review). All 25 sites selected for the NN agreed to work with a CIRB, but 2 did so only after the CIRB process was known. In response to this level of acceptance, in summer 2012 the NINDS made CIRB use a requirement for the NN.

As the recipient of the clinical coordinating center grant, Massachusetts General Hospital (MGH) became the CIRB site. The MGH collaborates with other Partners HealthCare hospitals in a single IRB system, the Partners Human Research Committee (PHRC). One of the seven existing PHRC IRB panels was designated to serve as the CIRB. The NN CIRB model was informed by reviews of and discussions with leaders of existing CIRBs, particularly the CIRB at the Department of Veterans Affairs.³ In November 2011, at the NN inaugural meeting in Bethesda, Maryland, the NN CIRB model was presented and discussed in a breakout session with participating sites’ IRB leaders.

Outcomes

The NN CIRB model

The NN CIRB model gives the CIRB sole regulatory review authority for NN trials, including initial review, continuing review, amendments, adverse events, protocol violations, and required regulatory IRB reporting. The model reflects Association for the Accreditation of Human Research Protection Programs (AAHRPP) criteria, and the CIRB and 15 of the 25 NN sites are AAHRPP accredited. Below, we describe the CIRB model by following a protocol through the system.

The principal investigator (PI) of a given trial develops the “parent” protocol and the informed consent form (ICF) containing a “locked” section (e.g., study description and procedures) and a customizable section that allows for local issues. For the CIRB review of the parent protocol, the ICF is customized to the trial PI’s site. Initial review is conducted by a CIRB chair who (1) determines if the submission is complete and ready for full CIRB review (i.e., deemed “IRB ready”) and (2) identifies concerns related to the protocol (e.g., recruitment issues, inclusion of minors) that should be considered by each site. The trial PI and NN Executive Committee decide whether a subset or all of the NN sites will be invited to participate in the trial on the basis of availability of patients, equipment, expertise, and other factors.

The IRB-ready parent protocol, with identified issues of potential concern, is sent to the participating sites for review by the site PI, the site study coordinator, and a site IRB representative. This is an institutional review only: The local IRB does not have regulatory authority for the protocol. Sites have 14 days from receipt of the protocol to provide substantive comments and local concerns for consideration when the protocol goes for full CIRB review. To facilitate the site reviews, the CIRB, in collaboration with the trial PI and the clinical coordinating center, hosts conference calls to review the CIRB process, provide an overview of the protocol, and address any immediate questions. Upon review of the comments received, the CIRB, in conjunction with the clinical coordinating center, contacts the trial PI for protocol changes, if needed, and sends the protocol to the full CIRB for review.

After approval by the CIRB, the protocol and PI-site ICF are sent to the sites selected for the trial. At that time, changes to the approved protocol are not considered, but sites have the option to decline participation. Sites that agree to the protocol as approved are added to the trial by submitting an amendment and a customized ICF; they are considered a “child” to the parent protocol. The protocol PI’s “parent” site, as well as all “child” sites, are responsible for obtaining site-specific ancillary committee reviews (e.g., radiation safety, biomedical engineering, nursing, conflict of interest).

All sites are on one timeline so that continuing review is synchronized. All adverse events, amendments, deviations, and other incidents or regulatory concerns are reported directly to the CIRB. The local IRBs do not have regulatory oversight. Issues that are site-limited (e.g., local study staff changes) are handled by the CIRB in coordination with the site PI. Events that have broader implications are discussed with the trial PI and the clinical coordinating center and, as appropriate, all participating site PIs.

Setting up the CIRB infrastructure

The CIRB had to exercise due diligence regarding each site as well as any affiliates that would be performance sites for NN research. (A performance site is any location where clinical studies take place within the network.) The 56 NN performance sites include the 25 main sites as well as 31 of their affiliates. The CIRB decided that an RA would be required for each performance site that held its own federalwide assurance for the protection of human subjects (FWA). For efficiency, one RA template limited to NN research was used for all FWA-holding performance sites to delineate roles and responsibilities and to establish the legal relationships between these sites and the CIRB. The NN also allows nonmember sites (i.e., sites that are not part of the network infrastructure) to participate in NN research when they are needed in a specific study for recruitment purposes or other reasons. Nonmember sites are also obligated to use the NN CIRB and to sign an NN RA limited to the specific protocol.

A draft IRB RA template was distributed in December 2011 to member sites and their affiliates. Two webinars with the sites were held (January 2012) before the final RA template was completed (February 2012). As soon as site-specific information was received, a site-specific RA was prepared and circulated for signatures. It took on average 84 days (median, 59 days; range, 48–350 days) to fully execute RAs at required performance sites.

Protocol approval

At the time of writing in March 2013, the NN CIRB had reviewed and approved its first protocol. The parent protocol was approved 56 days after submission to the CIRB; this period included the 14 days allowed for local parent site review and a 17-day delay related to experimental device ancillary review by the parent site’s biomedical engineering committee. Delayed availability of the experimental device for each site’s required ancillary review led to several sites delaying the submission of their site amendments to the CIRB. However, once site amendments were received by the CIRB, they were approved in an average of 12 days (median, 2 days; range, 1–44 days); when a delay occurred, it was the result of negotiating language regarding research-related injury and the customizable portions of the ICF. Overall, the protocol went from grant funding release to the first patient visit in less than four months.

Challenges

The development of this model highlighted a number of challenges, only some of which we anticipated. Whereas the CIRB has sole authority for all regulatory reviews, the NN sites remain responsible for many aspects of oversight, such as ancillary committee reviews, conflict-of-interest reviews, and analysis of local context issues, including relevant state and local laws. At many institutions, the IRB office manages these nonregulatory, institutional tasks and, at some sites, voiced discomfort about retaining these responsibilities without controlling the IRB review itself. The detailed RA used in the NN CIRB model includes delegation of each of these responsibilities to the site or the CIRB, respectively, and clarifies lines of communication, similar to what has been suggested by Flynn et al.⁴

Among the unexpected challenges was the variability in complexity among the NN sites, which range from single sites to sites with affiliates that are often geographically distant and may hold their own FWA. This heterogeneity created challenges in identifying which affiliate sites needed a separate RA.

Next Steps

The NN CIRB experience is in the pilot stage. A formal evaluation is planned after more protocols have been processed; the critical issues we plan to address include timelines, costs, and stakeholder satisfaction. Also, there has not yet been an opportunity to test how the model functions in case of incidents that require quality correction, such as major protocol violations.

We anticipate that CIRB use will decrease NN trial start-up time and thus promote efficient trial implementation. Although it is not reasonable to generalize from a single completed protocol review process, it is our impression that the time from submission of the parent protocol to CIRB approval of the parent protocol will not be shortened by this process; in fact, the need to solicit and consider participating sites’ input may add to the time required for initial review. However, efficiencies should be realized in shortened time for the approval of the multiple “child” sites, hence achieving approval for all participating sites more expeditiously. We also anticipate that continuing review and protocol amendments will be handled more efficiently using this CIRB model.

Our next steps include comparing NN CIRB review timelines with those of recent multicenter trials relying on local IRB review. Another important evaluation will focus on costs of this system, for both the CIRB and the sites. CIRB costs include establishing the CIRB model (e.g., due diligence of sites, completion of RAs, support for the CIRB IT systems, NN-specific forms) as well as ongoing implementation. Local sites will also incur costs; for example, some will need to develop new processes for completing the institutional tasks for protocols that have not been reviewed by the local IRB. However, these costs should be considered in the context of potential savings realized from avoiding review by multiple IRBs, from shortening start-up times, and from streamlining continuing review, amendments, and adverse event reporting.

The agreement of 25 U.S. academic institutions to participate in this innovative CIRB model suggests that there is a willingness to consider alternate IRB models. We note that the proposed mandatory use of a CIRB for domestic multisite research found in the July 2011 advance notice of proposed rule making for revisions of the Common Rule⁵ has probably added some momentum.

It is our hope that our experience with developing and implementing the NN CIRB model will inform ongoing and future discussions about CIRBs and suggest potential solutions to perceived barriers.⁴ At the present time, there is no standard CIRB model; hence, lessons learned from any one model must be carefully assessed for generalizability. The NN CIRB model has three distinct characteristics that differentiate it from other models. First, use of the NN CIRB is mandatory for all NN protocols; this contrasts with other models in which use of the CIRB is voluntary and can be considered on a protocol-by-protocol basis. Second, the NN CIRB model is a rigid central model. All regulatory authority is held by the CIRB, and there is no regulatory role for local IRBs. Other models currently in use maintain a role for local IRBs in review of site-specific amendments and events. Third, the NN CIRB model is supported by a robust research infrastructure with defined governance, a clinical coordinating center, a data coordinating center, a research pharmacy, a Web site, and a process for NN review and revision of proposed research protocols—all of which are reflected in detailed standard operating procedures. This infrastructure provides a level of standardization and quality control that likely cannot be realized in all multisite research efforts.

As next steps, the NINDS is planning to use a CIRB in another network it has recently launched, the National Institutes of Health (NIH) Stroke Trials Network (StrokeNET). Also, the NINDS is engaged in the NIH’s internal and broader initiatives to promote the use of CIRBs.