The Air Force Health Study Assets Research Program (2015)

4


Continued Management and Use of the AFHS Assets

This chapter addresses several issues key to enhancing the future use of the Air Force Health Study (AFHS) assets (the data and biospecimens) by the scientific community. The value of the assets for continued scientific research is discussed in depth in Chapter 3, but it is important to recognize that the availability and viability of such a resource depends on a careful balance between access by the scientific community and protection of the study participants. To that end, the committee begins the chapter with a discussion of options for the continued management and use of the AFHS data and biospecimens in research. That section is followed by an overview of considerations that any custodian or steward, including the Institute of Medicine (IOM), must take into account when making the assets available for research to the scientific community. The chapter ends with a discussion of opportunities for future research and possible metrics that can be applied to evaluate the future success of the program.

OPTIONS FOR CONTINUED MANAGEMENT AND USE OF THE AFHS ASSETS

The committee strongly believes that it is feasible and appropriate to maintain the AFHS data and biospecimens as their value and usefulness for research have been confirmed (as discussed in Chapter 3). The committee identified two options regarding the continued maintenance and management of the assets:

• Retain custody and management of the assets within the Medical Follow-Up Agency (MFUA) of the IOM; either maintaining the current manage-

• ment structure—where MFUA curates the assets—or forming a partnership between the IOM and another organization to manage and distribute the data to researchers (as is presently done with the biospecimens); or

• Transfer custody of the AFHS data and biospecimens to another organization.

Retain Custody and Management Under the IOM

At the time when the disposition of the AFHS was first considered, seven options for an appropriate custodian were identified (IOM, 2006). These included both nongovernment entities—specifically the IOM’s MFUA—and government research entities such as a Department of Veterans Affairs (VA) or VA-affiliated epidemiologic data management and distribution mechanism, a Department of Defense epidemiologic data management and distribution system, other federal agencies (such as the Centers for Disease Control and Prevention or the National Institutes of Health [NIH]) with expertise in epidemiologic data management and distribution mechanisms. The National Archives was also considered as a potential custodian, as was using the infrastructure that was in existence for the AFHS within the Air Force at the time, and identifying a new custodian through a competitive process (IOM, 2006). Each of these potential custodians had strengths and weaknesses that were discussed in the report, which did not name a recommended alternative.

Ultimately, Congress delegated custodianship of the AFHS assets to MFUA within the IOM. This was consonant with MFUA’s long history (since 1946) of conducting epidemiologic studies of military and veterans’ health; its experience maintaining data collections, including managing, analyzing, and disseminating data, protecting participants’ privacy and confidentiality; and its independence, objectivity, and thorough review and oversight processes.

The Disposition of the Air Force Health Study report identified several characteristics of a good custodian for the AFHS assets. Those characteristics included demonstrated ability to properly house, secure, and manage both very large and complex data and the accompanying biospecimens (IOM, 2006).

Since the IOM became the custodian of the AFHS electronic data, the storage devices containing them have been housed in a secured area with limited access to only those individual IOM staff members that have been approved to work with them. The IOM has invested heavily in data cleaning and preparation to foster the use of the data in new research, and, although the process is ongoing, program staff have considerable expertise managing, analyzing, and disseminating the data and biospecimen samples. Chapter 3 provides details regarding the condition of the received data and the preparation, cleaning, quality checks, and other management activities required to make it usable for research.

Because the IOM does not have the physical means needed to maintain a biospecimens collection, MFUA reached an agreement and contracted with the Air

Force Research Laboratory (AFRL) at Wright-Patterson Air Force Base (WPAFB) in Dayton, Ohio, to store, maintain, and ship samples to approved researchers. The IOM provides specific instructions on which samples to process based on analysis performed by the IOM staff using researcher requirements. In addition, the IOM provides all documentation, including vial labels and chain of custody forms for sample dissemination, as well as updates the electronic database with biospecimen sample locations and amounts. The Disposition of the Air Force Health Study report recommended that any custodian meet four guidelines pertaining to the storage and maintenance of the biospecimens: have a secure facility with controlled access and redundant power supplies to house freezers that are maintained at least –70°C, establish protocols for continuous monitoring of freezer function and other quality control and assurance practices for long-term biospecimen preservation, have appropriate local and remote alarm systems in place to ensure the preservation, viability, and integrity of the biospecimens, and have staff trained to respond to freezer breakdowns, power outages, or other emergencies at any hour (IOM, 2006). The storage, monitoring, maintenance, and emergency response guidelines outlined in the Disposition of the Air Force Health Study report have been met by AFRL (AFRL, 2012, 2014b).

More than 40 operating procedures and documents have been developed in collaboration with AFRL that detail the policies and procedures that inform the continuous monitoring of the freezers, ensure alarms systems are working, and staff are able to respond quickly and appropriately to any potential emergencies concerning the freezers and biospecimens, regardless of the day or the time. Operating instructions have been drafted, reviewed, approved, applied, and are in alignment with national best practices for biological sample repositories as set forth by the International Society for Biological and Environmental Repositories and the National Cancer Institute (NCI) (ISBER, 2011; NCI, 2011) for personnel qualification and training, establishing a quality assurance system, preparing the biospecimens, packaging and shipping the biospecimens, safe handling of biospecimens and laboratory equipment, equipment maintenance, and security and monitoring. These processes and procedures for storing, monitoring, maintaining the biospecimens, and responding to emergencies require periodic review and updating to ensure they continue to conform to national best practices for biological sample repositories.

A quality assurance program has been instituted for the biospecimens to prevent inconsistencies in handling, provide standardization for conducting repository activities, and ensure best practices are used (AFRL, 2012). The program ensures that all repository systems function properly and according to guidelines. The continued monitoring and evaluation of processes and equipment allow for early identification of deviations from proper operations that may compromise the quality and integrity of the AFHS biospecimens. Biosafety cabinets are certified annually and are used to process and aliquot all of the AFHS biospecimens.

Laboratory staff notify the IOM staff about concerns regarding the freezers or biospecimen integrity and document any incidents in annual progress reports.

Standard operating procedures for thawing, freezing, and aliquoting of frozen biospecimen samples were developed and followed during the fulfillment of investigator requests (AFRL, 2013, 2014a). Investigators review the procedures that are proposed to be used on their samples in advance and sign their acknowledgement, or they propose a revised protocol that is agreed upon and then used to process the request. The first time a serum or urine biospecimen is needed from a previously unaccessed vial, it is aliquoted from its (5 mL or more) container into smaller 0.5 mL aliquots; only the amount needed is sent to the investigator, and the remainder is returned to storage in smaller aliquots. This process serves to minimize freeze–thaw cycles for subsequent requests, and thus, helps to maintain biospecimen integrity as well as facilitating biospecimen selection for future requests. The aliquoted samples are placed into storage containers that are stable for long-term storage. To date, the most commonly requested amount of serum is 0.5 mL, so larger vials are realiquoted to 0.5 mL samples. The serum sample volumes requested and used to date have left most of the biospecimen remaining for future studies. When samples of adipose tissue and semen are requested, the amount requested is removed and the rest of it refrozen without separation into additional aliquots. The committee finds the storage, monitoring, and emergency response guidelines recommended by the Disposition of the Air Force Health Study report have been met by AFRL and has no concerns about the maintenance of the biospecimens.

Although MFUA has much experience in storing, managing, and disseminating epidemiologic data, the committee observes that partnering with an agency or organization that administers databases or funds research as its primary mission may be in a better position to actively promote and leverage use of the AFHS data and biospecimens for new and original research than the IOM alone. This is not meant to cast doubt on the ability of the IOM to manage this program, but instead to suggest a possible alternative approach that uses the strengths of the IOM in scientific review while enhancing the breadth of the research program. In such a partnership, the IOM would retain custodianship of the AFHS assets and approve the type and amount of information given to researchers through an advisory committee. The partner organization would physically hold the electronic data; have responsibility for cleaning and applying quality control, managing, and maintaining them; preparing and distributing data (and selecting any biospecimens associated with research studies) to researchers as directed by the IOM; and apply its expertise in promoting the use of large datasets for research. Such a partnership would be similar to the one currently in place for the storage and maintenance of the biospecimens between the IOM and AFRL.

There are various examples of organizations partnering, including public and private alliances, to perform data and biorepository management. One such example is the Framingham Heart Study, a large ongoing prospective epidemio-

logic study that collected and stored biospecimens that can be used by outside researchers, which is maintained under the joint stewardship of Boston University and the National Heart, Lung, and Blood Institute (NHLBI) at NIH (Framingham Heart Study, 2014). The molecular genetic data, including genomic sequencing results, and gene expression and microRNA profiles collected from the Framingham Heart Study are available through NIH’s database of Genotypes and Phenotypes, and exam data are available for use in research through the NHLBI. Therefore, although much of the Framingham Heart Study data and biospecimens are maintained and made available by other entities, the stewards continue to maintain overall control of the assets, and such an arrangement may be possible for the AFHS data and biospecimens. Certain organizations have experience in the management of broader data repositories, such as the Inter-university Consortium for Political and Social Research (ICPSR), housed at the University of Michigan, which holds, curates, and provides access to hundreds of diverse datasets from more than 740 universities, government agencies, and other institutions that are used in new research studies (ICPSR, 2014).

Shifting storage, curation, and maintenance of the data to another party would require the completion of the data dictionaries and other documentation activities that MFUA has been pursuing and could not take place without funding for the staff time needed to prepare the assets.

Transfer Custody

Congress may decide to designate an alternate custodian to manage the AFHS data and biospecimens and to administer the research program. Such a decision should be carefully considered and based on thorough comparison of the attributes and characteristics between the IOM and proposed alternatives. If a new agency or entity were chosen to be the custodian of the AFHS assets, there are many considerations that would have to be resolved before such a transfer could take place. First, the cohort would have to consent to have their materials transferred to a new custodian. Approximately 10% of the cohort did not consent to have their AFHS materials transferred from the Air Force to the IOM in 2007, and another 10% did not consent (or could not be located) to have their materials used in new research. These losses do not appear to have significantly affected the research designs of the types of studies proposed. However, if cohort members were asked to consent a third time, the likely result would be a further diminution of the cohort, the extent of which would be dependent on the new selected custodian. Loss of additional cohort members will reduce the value of these assets for use in additional research because the sample size may not be adequate for many types of studies. Loss of additional participants may introduce further selection bias as well as weaken the scientific validity of the analyses. Second, if the AFHS assets were transferred to a different custodian, their availability for continued use in research would likely suffer a setback because the new custodian would have to set up a new research

program or integrate the AFHS into an existing one. It is unclear in what format the data would be transferred. For example, the IOM received more than 100,000 text files that had to be read in, integrated, and cleaned. Whether the custodian would choose to create a new database using the original text files or would try to incorporate the work the IOM has done (if the systems are compatible), each option has large implications for continuing a program of research using the AFHS data and biospecimens. Third, the custodian would need to find an appropriate and secure location to house the freezers in which the biospecimens are stored, form a relationship with AFRL, or to transfer them to a different ultralow temperature storage facility. As was noted above, funding would need to be provided to support the staff needed to prepare assets for the transfer and assure that it was successfully completed.

Regardless of which custodial option is ultimately selected by Congress, a number of aspects need to be considered and accounted for to ensure continued successful use of the AFHS data and biospecimens for new and innovative research. The next section highlights some of these.

CONSIDERATIONS FOR CUSTODIANS OR STEWARDS OF THE AFHS ASSETS

The IOM’s AFHS Assets Research Program has continued to develop and generate interest. There are several issues that need to be considered for managing the data and biospecimens and making them available for research. They fall into three major categories and are discussed below:

• storage (including data security), data curation, and privacy protection;
• laws, rules, and policies related to biospecimen management and dissemination; and
• sustained funding to support these activities.

Storage, Data Curation, and Privacy Protection

Data Storage

Practices and technologies to enable data security have advanced substantially since the AFHS began in the early 1980s. Data must be stored in a manner that maintains its physical and logical integrity and allows it to be easily retrievable, such as in a database file format, while protecting it from external and internal threats, such as accidental or deliberate deletions. Physical measures to achieve this include buildings with restricted access, locked offices, restricting personnel who have access, password-protected computers, and firewalls. Technical measures include restricting access to files to individuals with appropriate privileges (designated by the custodian), separating personally identifiable information from

the rest of the data, and ensuring that devices storing AFHS data are not connected to the Internet. Database content must also be reviewed periodically for accuracy.

Data Curation

Data curation refers to the active management of data required to maintain it for long-term preservation and reuse. There is not a single set of nationally agreed upon criteria or standards for curation of collected data, such as that of the AFHS. However, a number of organizations including leading universities, research-driven institutions, and government agencies have developed and proposed standards and guidelines for best practices that broadly aim to maximize internal data management of a study and address making it available to the scientific community. Although there are differences across the guidelines, Box 4-1 summarizes the main elements that should be considered and applied to most datasets that are to be used for broader research. Considerations related to allowing broader access to existing scientific data include adequate preparation of data and its associated documentation, protection of human subjects, formation and use of data management plans, and appropriate attribution of the data source. An important component of any strategic plan concerning the maintenance of the AFHS data and its wide scientific use is long-term curation plans that align with best practices

and archival standards such as those proposed by ICPSR or the Data Documentation Initiative consortium (Data Documentation Initiative, 2015; ICPSR, 2014).

Proper data documentation, quality, and accessibility are key to enhancing and enabling use internally as well as by the broader scientific community. Because of the longitudinal nature of the AFHS and the diverse manners by which data were often collected, recorded, and documented, as well as the different statistical methods used to analyze and account for missing data and loss to follow-up, the structure of the data is complex. For example, even basic demographic variables that were collected at each visit were not necessarily formatted in the same way or asked as a question in the same way. While not an insurmountable issue, it has and continues to require substantial efforts to integrate and process data over time.

Based on the IOM’s experience of working with the AFHS data and biospecimens for the past 3 years, additional issues have been identified that have affected the information available for research and timeliness of fulfilling data requests, such as the lack of standardization between examination cycles and the inability to abstract information from some types of files. To achieve a more efficient operation of the AFHS data and biospecimen resource, these issues need to be addressed. First, the data are not standardized between cycles or in a database format. As a consequence, they have not been completely subject to the types of quality control procedures necessary to ensure standardization and interpret-ability for future investigators. Translating the data into such a format is a large undertaking, and the IOM has already committed a substantial amount of effort into quality control and database creation.

Second, medical chart information, such as electrocardiograms and radiographs, are stored in digital media files that have not been processed owing to the costs involved in abstracting and de-identifying this information, which can only be accomplished using proprietary software. The approximately 20,000 files hold potentially important variables and information, but they are not currently in an accessible form and will require further substantial investment of money and time to become usable. Translation of the files into a usable format is the first step to making this information available; however, each file has personally identifiable information (for example name, social security numbers, medical record numbers, date of birth, etc.) that must then be redacted. Redacting personally identifiable information from these images and files will be a labor intensive process because, as the IOM staff were told when these files were transferred to the IOM’s custodianship, the information can be found in multiple places within the file and is not in a standard format or place in each file. Therefore, redacting the identifiable information will require custom programming by an individual staff member to ensure that all such information has been removed and that the removal does not interfere with the image. Custom programming and the need for dedicated individual staff time for translating these thousands of files is an expensive endeavor, but until interest from investigators necessitates the need

for translating some of these files, the time or expense cannot be estimated. The committee believes that these records should continue to be preserved and that dedicated funding for purchasing the software and staff time to complete the abstraction should be provided by either the funding agency or the investigators who require the information as part of their work if it is concluded that there is a demand for the information contained in them. To date, no studies have been approved that require those data.

Storage and Maintenance of Biospecimens

A complete inventory of the biospecimens has not been completed before or since the biospecimens were transferred to AFRL, and would likely be a necessary condition if the biospecimens were to be moved to another facility. Commercial biological repositories typically require an up-to-date inventory of all materials along with labeling or barcoding to align with their systems in order to accept biospecimens for storage. Thus, the approximately 91,500 biospecimens in the AFHS would likely have to be inventoried and relabeled before they could be transferred to another facility. Such an inventory is important to good management practices and an exercise that will be valuable for continued use of the biospecimens in research.

The Disposition of the Air Force Health Study report also called for a plan detailing biospecimen access and residual return policies, including clear documentation of the entire biospecimen access application process, the review process, decision-making criteria, biospecimen processing and shipping costs, and final disposition of biospecimens (IOM, 2006). The MFUA staff with guidance from the advisory committee has operationalized each of these items.

As recommended by NCI’s Best Practices for Biorepositories and Biospecimen Resources, a legacy or contingency plan for the AFHS biospecimens is part of their successful management. This is especially important because there is currently no long-term sustained funding source for the biospecimens (or data) in place (see Sustained Funding). The conflict between currently limited and unknown additional funding and stakeholders’ desire that the AFHS biorepository be permanently maintained and widely used results in an ethical dilemma that further highlights the need for a long-term strategy and legacy plan. The biospecimens have been shown to retain their biochemical properties and are valuable for many types of research studies. Should funding not be continued or a new custodian chosen, the IOM will need to consult with AFRL as they develop and implement a legacy plan for the future disposition of the biospecimen repository. The advisory committee has discussed needed components of a legacy plan as part of its responsibilities, but ultimately any transfer and legacy plan will have to be enacted by the IOM and be consistent with any prior agreements, such as the need to reconsent the cohort, laws, and institutional policies that may apply.

Privacy Protection

The extensive data collected on the AFHS participants and the linkage of that data to the biospecimens provides an extremely rich resource for research. However, while those data are being made available for a broad array of research, it is critical to maintain appropriate safeguards to protect participants’ privacy and confidentiality. The safeguards in place encompass not only data already available but should extend to additional data that may be generated in the future from the biospecimens or from studies that seek to collect new information on cohort participants, such as their cause of death. Thus, balancing access to the AFHS data and biospecimens while safeguarding the privacy of the AFHS participants is essential to maintaining a successful research program. Human subjects’ protection is an important consideration for all studies of human health conducted under routine research conditions, and the IOM practices regarding the AFHS assets fall within normal boundaries for protection of human subjects.

Data must be de-identified as necessary to protect participant privacy and confidentiality. Variables that could lead to the re-identification of the research participants such as date of birth, geographical location, and even diagnoses of rare diseases must be removed or binned into larger categories (e.g., the first three rather than all five digits of zip code). However, overly aggressive stripping of information that does not pose a threat to participant privacy and confidentiality risks diminishing the value and usefulness of the data. A further consideration is that the AFHS is not a randomly selected cohort but rather a specific and well-defined group, especially the Ranch Hands. De-identification processes for these data must take into account that publicly available information, such as tours of duty, military occupation, and military honors, can potentially be linked to information in the database that would allow identification of a participant. For example, in 1999 the Government Accountability Office (GAO) criticized the AFHS for its slow progress of issuing results and recommended that the data be made available to the broader community of scientists (GAO, 1999). Although the then data custodians removed all personally identifying variables (e.g., names, identification numbers, addresses, and birthdates) before posting the dataset to their website, within a few months the dataset had to be removed due to concerns about insecure data sharing practices and violations of participant privacy because the small size and special nature of the cohort led to the identification of individual participants. Therefore, any custodian must consider what other publicly available data might exist that would allow individuals with knowledge of both datasets to identify particular AFHS study participants.

The policy of the IOM advisory committee has been to limit the information released to approved researchers to only that information required for the specific research question of interest. Variables that may lead to the identification of a participant are suppressed. For example, dates of birth and death are not included with any dataset, instead only the year of birth or death is included. For studies

requiring matching based on closest age, the IOM staff perform the match and include the difference of the number of days between birthdates in the dataset. Furthermore, each research study is limited to include only those participants necessary for the study. For example, the study population is limited to the specific inclusion criteria approved in the proposal. A unique dataset is generated for each approved study—there is no premade dataset that is available for all approved researchers—and for each, individual participants are assigned unique case numbers so datasets cannot be combined across studies. Only the IOM data managers have the key to the case numbers. Similarly, aliquots of requested biospecimen samples are given unique identifiers so they could be tracked independently of the parent biospecimen and adhere to confidentiality and privacy provisions. These case numbers are the same as the ones used in the corresponding data provided to the researchers. Aliquots for subsequent researchers were labeled with a unique identifier for each study. The committee finds the practices in place for maintaining privacy protection of the cohort are appropriate and should be continued.

Responsibility for maintaining privacy and confidentiality extends to investigators receiving biospecimens and data. The Disposition of the Air Force Health Study report identified several ways in which these responsibilities could be addressed. The review process for access to the AFHS assets and the subsequent data use and asset transfer agreement stipulates that researchers must meet and comply with several criteria to ensure the privacy and confidentiality of the individuals who have provided data and biospecimens for their use. These criteria include researchers must provide a detailed description of how they intend to use the data and biospecimens and be specific about the data variables needed; receive only de-identified data and biospecimens; agree to not attempt to identify any individuals; store the biospecimens in a secure location with limited access only by authorized personnel; store data on secure computer systems and not share it (as provided or generated from the biospecimens); and have institutional review board (IRB) approval by both the researcher’s institution and the National Academy of Sciences for each approved study.

Additional restrictions on access to the AFHS data and biospecimens, such as a multistep review process, can further promote data security. Other safeguards might include varying levels of access to specific types of data and using a secure data enclave (ICPSR, 2014). However, the committee cautions that data and biospecimens are only valuable and useful if they are actually used, and investigators are far less likely to pursue research on assets that present high barriers to access.

Federal Policies Related to Management and Dissemination of the AFHS Assets

Practices governing research on biospecimens and associated data in the United States are generally related to the protection of study participants and their health information and regulations for protection of human subjects in research

that have been adopted by federal departments and agencies that conduct and sponsor such research (such as VA). Laws and policies related to management and dissemination of the AFHS data and biospecimens must be taken into account by the custodian or steward if the assets are to continue to be made available for research. The IOM and AFRL staff working with the AFHS assets are trained in and adhere to regulations and protocols for ensuring the privacy of the AFHS cohort participants and are in compliance with all relevant legal requirements and ethical and social norms and expectations.

Legal requirements include federal ones, such as the Policy for the Protection of Human Subjects (Common Rule), as well as state- or jurisdiction-specific laws and regulations where the data and/or biospecimens are held. Privacy standards, such as the Freedom of Information Act and the Public Health Service Act, provide additional guidance regarding the conduct of research using the AFHS assets. Although the AFHS data and biospecimens were collected by the Air Force and at that time subject to military rules and regulations related to biomedical and behavioral research, because custodianship and responsibility for these assets was transferred to the IOM, these regulations no longer apply.

The Common Rule governs most federally funded research conducted on human beings and aims to ensure that the rights of human subjects are protected during the course of a research project (IOM, 2012). Simply, it seeks to protect human subjects through requiring informed consent of the research participant and requiring review of proposed research by an IRB. Data are considered personally identifiable if the identity of a participant is or may be readily ascertained by the investigator or associated with the information accessed by the researcher (45 CFR § 46.102(f); OHRP, 2008). However, the Common Rule exempts from its requirements research that involves solely the collection or study of existing data, documents, records, pathological biospecimens, or diagnostic biospecimens, if these sources are publicly available or if the information is recorded by the investigator in such a manner that participants cannot be identified, directly or through identifiers linked to the participants (45 CFR § 46.101(b)(4)).

The Common Rule and related regulations may be applicable to future research conducted with the AFHS data or biospecimens, but it does not apply to the assets themselves. The Office for Human Research Protections of the Department of Health and Human Services has provided guidance stating that the use of biospecimens and data in research that were originally collected for different research—such as the case with the AFHS assets—is not considered research on human subjects under the Common Rule if the investigators cannot readily identify the source individuals (OHRP, 2008). Therefore, as long as personally identifiable information is removed from the biospecimen samples and associated data before they are provided to researchers, such that the individuals cannot be readily identified by the researchers, research conducted using these biospecimens or data would not constitute human subjects research under the Common Rule. To the best of the IOM’s and the committee’s knowledge, there have been no

breaches of confidentiality concerning the data or biospecimens while the assets have been under the IOM stewardship, and the IOM has taken active steps to ensure this remains the case.

In February 2013 the U.S. Office of Science and Technology Policy (OSTP) published a memo outlining federal policy with regard to data access (Holdren, 2013). Specifically, OSTP directed federal agencies with more than $100 million in research and development expenditures to develop plans to make the published results of federally funded research freely available to the public within 1 year of publication. At the same time OSTP has asked such agencies to require researchers to account for and manage the digital data resulting from federally funded scientific research. NIH has also promulgated guidelines and policy statements that are similar to OSTP and is increasingly focused on the broad sharing of data from NIH investigations. Investigators were encouraged to use the AFHS resource, and the associated seed funding, to generate pilot data and findings for future investigations with the expectation that such investigations would be funded by various not-for-profit and governmental agencies, such as NIH.

Data-sharing expectations with respect to NIH date back to February 2003 when NIH published the final data sharing policy, which stated that studies provided with at least $500,000 in any year of a study would need to provide a data-sharing plan (NIH, 2003). In this policy, NIH stated that shared data should be devoid of identifiers that would permit linkages to individual research participants, as well as variables that could lead to “deductive disclosure” of an individual participant’s identity. If data sharing could not be performed then investigators would need to provide an explanation. The IOM’s AFHS Assets Research Program has already implemented measures that align with the NIH policy to prevent identification of any individual (see Privacy Protection section) including generating a unique dataset for each approved study and assigning randomly generated case numbers for each individual included in the dataset so datasets cannot be combined across studies. While grant panel reviewers have been instructed not to factor such plans into the determination of scientific merit or a priority score for a proposal, NIH program staff is responsible for overseeing the data-sharing policy and for assessing the appropriateness and adequacy of the proposed data-sharing plan.

In August 2014, NIH published a data-sharing policy that focused on genomic sequence data, which extended and revised a 2007 policy that focused mainly on genomewide scans of single nucleotide polymorphic regions (NIH, 2007, 2014). This policy states that NIH expects sharing of large-scale human genome sequencing data for broad reuse regardless of the quantity of funding provided by the institute. These policies reinforced the statement that the data should be devoid of identifiers, but went on to state “expect(s) investigators generating genomic data to seek consent from participants for future research uses and the broadest possible sharing” and that “the breadth of the sharing permitted by the consent may be taken into consideration during program priority review.”

The language of the consent form used by the IOM complies with NIH and other federal policy. Consent was obtained from participants for their biospecimens and medical information to be transferred from the Air Force to the IOM. Following the transfer of the AFHS materials, the IOM embarked on an 18-month effort to obtain consent from all living participants to allow their data and biospecimens to be used for research (see Chapter 3). The language of the consent was broad as to not limit the types of research that could use the AFHS data and biospecimens, and meets the suggested guidance of the Department of Health and Human Services that has been codified in federal regulations (HHS, 2011) and the criteria suggested by NIH and other institutions to allow the distribution of properly protected data to qualified scientists. Participants were assured that all research would have to be approved by an expert advisory committee appointed by the National Academy of Sciences and subject to the approval of its IRB. If a different custodian were to be selected in the future, a new consent of the current AFHS cohort would be required. Experience has shown that with each new consent process, approximately 10% of participants do not give consent, thus, any new consent requirements would likely result in a diminution of the current cohort thereby reducing its value for continued research.

This most recent NIH policy does not require consent for data generated from de-identified clinical biospecimens and cell lines collected before the effective date of the policy, which is January 25, 2015, “because of the practical and ethical limitations in recontacting participants to obtain new consent for existing collections and the fact that such data may have already been widely used in research.” From this language, it appears that NIH funding may be used to generate or study existing sequences of DNA from the AFHS biospecimens. One possibility for researchers wishing to conduct genomic studies using the biospecimens is to create a data-sharing plan that includes returning generated data to the IOM (as is currently required in the data use agreement). This would allow the generated data to be linked to the other available data which would then be made available to other researchers and prevent exhausting the biospecimens by using them to generate the same data multiple times. The data use agreement specifies that investigator-derived data does not need to be returned to MFUA until 3 years from the end of the project to allow adequate time for research results to be published. Because the AFHS Assets Research Program has only been distributing data to investigators for just over 2 years, no derivative data from researchers has been returned to MFUA or incorporated with the AFHS data files. Data use agreements and assets transfer agreements (for those studies using biospecimens), such as those currently in place between the IOM and researchers, will require periodic review and revision, to include requirements for data sharing.

Ethical and social norms and expectations include respect for the individuals who have voluntarily provided biospecimen samples, meeting public expectations of the treatment of participants and the use of their materials in research, and supporting efficient functioning of the biorepository and transparent practices.

Obtaining informed consent from the AFHS cohort members to use their data and biospecimens in research studies is one example of the ethical principle of respect for individuals and promoting their autonomy. Social norms include concepts such as engendering public trust, by being aware of and responsive to interested stakeholder groups. For example, the IOM staff have presented updates of the program and approved research to the Ranch Hand Vietnam Association (a veteran organization composed of former participants in Operation Ranch Hand) at their annual meetings to update them on the types of research questions that have been approved for use of the AFHS data and biospecimens. Other considerations include retention of participants and how biospecimens and associated data are maintained and distributed (Caulfield et al., 2014).

Sustained Funding

To properly plan for the future management of the AFHS assets, it is important to understand the limitations and conditions that were extant at the time the IOM began its stewardship of the AFHS data and biospecimens and that are likely to persist with any steward. While funding was considered an issue to the disposition of the AFHS, and rough estimates of funding levels were provided in the 2006 report, experience of the current research program has shown that the extent of work required to prepare the dataset for access was underestimated. The IOM requested approximately $4 million in total costs over 4 years to support the development of the research management program, but was awarded approximately $3 million to support the work over 4 years. Any unobligated funds remaining at the end of each contract year were deobligated or not carried forward. Because of the condition of the data received, the majority of staff effort went into cleaning and organizing the data at a level of time and effort that was not originally anticipated. Of the total award of $3.1 million, $750,000 was dedicated to funding pilot studies, which left just over $2 million over 4 years in direct costs. (This does not include costs for maintaining the biological samples, which were a separate contract between VA and AFRL.) Despite the limited resources to support it, the committee finds that the IOM has done a commendable job overcoming the problems identified before and after transfer of the AFHS assets to continue to maintain them in accordance to best practices and make them available to qualified researchers. However, sustainable funding at the level necessary to continue to manage the assets as well as make them available for researchers is necessary to continue the program.

The current AFHS research program should be viewed as a pilot effort. To truly realize the potential of the AFHS assets as a research asset like the Framingham Heart Study, the Atherosclerosis Risk in Communities Study, and other longitudinal epidemiologic studies, will require a more sustained and possibly greater investment of money and personnel. A well-defined process to access the assets is in place, but it could be improved by requiring investigators to provide

data management and sharing plans, which would align with the new practices of NIH (see the Federal Policies section above). Continued resources are needed to transform the data from a disparate system of tables into a high-quality database that would facilitate rapid response to investigator requests.

FUTURE DIRECTIONS

A number of steps have been taken to prepare the data and biospecimens for use in research since the IOM began its AFHS Assets Research Program. As the research program moves forward, short- and long-term metrics are needed to evaluate its reach, value, and success. Interest in the assets is evidenced by the numbers of contacts and inquiries about them, the numbers of submitted letters of intent and ultimately approved proposals, and the amount of data disseminated. However, any evaluations of the program should consider not only the number of requests but the quality or importance of the research published from use of the assets. Since the program infrastructure has been developed and the application process is operational, the value of the AFHS research program should continue to be evaluated through the use of short- and long-term metrics. Examples of these metrics may include sustained interest in the AFHS assets quantified as described above, numbers of peer-reviewed publications, the journals that publish the results, and numbers and types (for example, R01 applications versus internal institution funding) of successful grant applications that used results from the AFHS.

Different metrics may be applied to the data and biospecimens. For example, NCI, which has funded the development and operation of many biospecimen resources, proposed guidelines for their evaluation (NCI, 2014). The questions asked of NCI-funded biospecimen resources may be adapted and used to evaluate the AFHS biospecimens:

• How effectively has the resource performed?
• What impact has the resource had on research?
• Is there a continuing need for the resource?

Effectiveness can be measured by number and variety of biospecimens or datasets provided, number of researchers served, the number of difficult to obtain biospecimens or data provided, affordability, rate of repeat requests, number of research publications generated, and usability of the provided data and biospecimens. Metrics for impact of the scientific studies conducted using the resource data and biospecimens include importance of published studies, prestige of journals, and rate of citation; role of the study in furthering useful research techniques or technologies; and whether the study led to a regulatory approval or commercialization of a product (e.g., an in vitro diagnostic or a drug), or to a new public health recommendation. Evaluation of continued need should consider whether the resource continues to see a steady flow of requests for biospecimens

and data, whether the resource continues to contribute to the scientific value of studies, whether the data and biospecimen types remain relevant for contemporary scientific needs, and whether alternative sources for similar data or biospecimens are now available without prohibitive restrictions or costs. The numbers of requests and submitted applications are one measure of distribution or popularity, but are not indicative of the broader picture of value and use in the scientific community.

Opportunities for Continued Research

The longitudinal nature of the AFHS—with its extended follow-up, high rates of retention, and repeat biological samples—provides a valuable opportunity for research beyond the original aims of the study. As the AFHS research program moves forward, it can draw upon the experiences of other data resources and biorepositories to ensure success, such as the Armed Forces Institute of Pathology’s tissue repository, the Atherosclerosis Risk in Communities Study, the Cardiovascular Heart Study, and the Framingham Heart Study.

There is widespread recognition that clinical and survey data collected for one study can be repurposed to study a wide array of phenotypes (Pathak et al., 2013). Moreover, there is evidence that such data can be used to support biomolecular (e.g., genomic and proteomic) association studies (Jensen et al., 2012; Kohane, 2011). Phenotypes have been shown to have merit across a wide range of topics, ranging from variations of arthritis to metabolic disorders and psychiatric issues. A certain portion of the phenotypes for which investigators provide the specifications focus on longitudinal aspects of an individual’s life and there is potential that the long arc of following many of the participants in the AFHS could be studied using such methods, and has been the focus of a few of the approved proposals. Additionally, recent evidence has shown that the reuse of existing biological sample repositories and medical records can enable research studies at a significantly faster and more cost-effective pace than development of a new biological sample repository designed to target a specific disease or population (Bowton et al., 2014).

For example, the AFHS assets could be used to study the determinants of incidence or natural history of common conditions (e.g., hypertension, diabetes, and obesity) as well as correlates of health in aging. Proposals addressing these kinds of topics have been few and have only been offered recently (e.g., Chambers’ on serum paraoxanase 1 in relation to type 2 diabetes and aging, and Roth’s on caveolin in relation to aging) and results from those studies (other than feasibility data) are not yet available. However, the time period for evaluating the use of the data and biospecimens for other purposes has been short, and owing to the level and length of funding for this effort, this program may be more appropriately considered as a pilot research program, and could not have the impact of more established research programs.

At the same time, there are certain challenges that the AFHS research program will need to surmount in order to achieve its goals of facilitating meaningful

biomedical research across a range of topics that extend far beyond investigations associated with the original purpose of data collection. An IOM report on the future uses of the Armed Forces Institute of Pathology’s tissue repository covered issues that pertain to the AFHS biospecimen repository including use of the biospecimens in research while continuing to protect and respect the individuals who provided the biospecimens, meeting public expectations, and supporting the efficient functioning of the repository (IOM, 2012). Additionally, as was established in Chapter 2, many of the biospecimens in the AFHS repository are unique in type and quality. Those cases that have longitudinal data and biospecimen collection or which have concurrent matched samples of different types (e.g., blood and adipose tissue) are relatively rare and use of such valuable cases may warrant additional considerations. A highly visible marketing strategy will likely serve to increase interest in and use of the AFHS assets for new research. The next chapter summarizes the committee’s findings, conclusions, and recommendations regarding the value of the AFHS data and biospecimens and their continued maintenance and use for research.

REFERENCES

AFRL (Air Force Research Laboratory). 2012. Air Force Health Study Quality Assurance Program. Wright-Patterson AFB, OH: AFRL.

AFRL. 2013. Processing of AFHS specimen requests. Wright-Patterson AFB, OH: AFRL.

AFRL. 2014a. Processing of AFHS adipose tissue specimen requests. Wright-Patterson AFB, OH: AFRL.

AFRL. 2014b. AFHS Annual progress report. Wright-Patterson AFB, OH: AFRL.

Bowton, E., J. R. Field, S. Wang, J. S. Schildcrout, S. L. Van Driest, J. T. Delaney, J. Cowan, P. Weeke, J. D. Mosley, Q. S. Wells, J. H. Karnes, C. Shaffer, J. F. Peterson, J. C. Denny, D. M. Roden, and J. M. Pulley. 2014. Biobanks and electronic medical records: Enabling cost-effective research. Science Translational Medicine 6(234):234.

Caulfield, T., S. Burningham, Y. Joly, Z. Master, M. Shabani, P. Borry, A. Becker, M. Burgess, K. Calder, C. Critchley, K. Edwards, S. M. Fullerton, H. Gottweis, R. Hyde-Lay, J. Illes, R. Isasi, K. Kato, J. Kaye, B. Knoppers, J. Lynch, A. McGuire, E. Meslin, D. Nicol, K. O’Doherty, U. Ogbogu, M. Otlowski, D. Pullman, N. Ries, C. Scott, M. Sears, H. Wallace, and M. H. Zawati. 2014. A review of the key issues associated with the commercialization of biobanks. Journal of Law and Biosciences 1(1):94-110.

Data Documentation Initiative. 2014. Best practices. http://www.ddialliance.org/getting-started#best (accessed January 16, 2015).

Framingham Heart Study. 2014. Participating investigators. https://www.framinghamheartstudy.org/about-fhs/participating-institutions.php (accessed November 4, 2014).

GAO (General Accounting Office). 1999. Agent Orange: Actions needed to improve communications of Air Force Ranch Hand Study data and results. GAO/NSIAD-00-31. Washington, DC: GAO.

HHS (Department of Health and Human Services). 2011. Human subjects research protections: Enhancing protections for research subjects and reducing burden, delay, and ambiguity for investigators. Federal Register 76(143):44512-44531.

Holdren, J. P. 2013. Memorandum: Increasing access to the results of federally funded scientific research February 22, 2013. Office of Science Technology and Policy. http://www.whitehouse.gov/sites/default/files/microsites/ostp/ostp_public_access_memo_2013.pdf (accessed December 1, 2014).

ICPSR (Inter-university Consortium for Political and Social Research). 2014. Elements of a public access plan for scientific data. ICSPR Data Management and Curation. http://www.icpsr.umich.edu/icpsrweb/content/datamanagement/ostp.html (accessed September 22, 2014).

IOM (Institute of Medicine). 2006. Disposition of the Air Force Health Study. Washington, DC: The National Academies Press.

IOM. 2012. Future uses of the Department of Defense Joint Pathology Center Biorepository. Washington, DC: The National Academies Press.

ISBER (International Society for Biological and Environmental Repositories). 2011. Best practices for repositories: Collection, storage, retrieval and distribution of biological materials for research. http://c.ymcdn.com/sites/www.isber.org/resource/resmgr/Files/ISBER_Best_Practices_3rd_Edi.pdf (accessed December 1, 2014).

Jensen, P. B., L. J. Jensen, and S. Brunak. 2012. Mining electronic health records: Towards better research applications and clinical care. Nature Review Genetics 13:395-405.

Kohane, I. S. 2011. Using electronic health records to drive discovery in disease genomics. Nature Review Genetics 12:417-428.

NCI (National Cancer Institute). 2011. NCI best practices for biospecimen resources. http://biospecimens.cancer.gov/bestpractices/2011-NCIbestpractices.pdf (accessed December 1, 2014).

NCI. 2014. How the NCI evaluates NCI-supported resources. http://www.cancerdiagnosis.nci.nih.gov/humanSpecimens/establish/evaluates.htm (accessed July 31, 2014).

NIH (National Institutes of Health). 2003. Final NIH statement on sharing research data. NOT-OD-03-032. Bethesda, MD: NIH.

NIH. 2007. Policy for sharing data obtained in NIH supported or conducted genome-wide association studies (GWAS). NOT-OD-07-088. Bethesda, MD: NIH.

NIH. 2014. NIH genomic data sharing policy. NOT-OD-14-124. Bethesda, MD: NIH.

OHRP (Office for Human Research Protections). 2008. Guidance on engagement of institutions in human subjects research. http://www.hhs.gov/ohrp/policy/engage08.html (accessed November 20, 2014).

Pathak, J., A. N. Kho, and J. C. Denny. 2013. Electronic health records-driven phenotyping: Challenges, recent advances, and perspectives. Journal of the American Medical Informatics Association 20:e206-e211.

This page intentionally left blank.