The Air Force Health Study Assets Research Program (2015)

3


Management of the AFHS Assets Research Program

The Institute of Medicine’s (IOM’s) research program is a unique undertaking because rarely, if ever, has such a large transfer of both data and biospecimens (assets) and the re-creation from the ground up of such an extensive database and research program been accomplished—especially in such a short time. This chapter begins with a discussion of the process of transferring the custodianship of the Air Force Health Study (AFHS) assets to the IOM. The chapter then details the numbers of and characteristics of assets received including their condition, quality, and usability for new research. The second half of the chapter describes the research program’s infrastructure, including development and implementation of the proposal and review process to allow qualified researchers access to the data and the biospecimens. The last section discusses examples of approved research studies, preliminary findings, and a summary of researchers’ experience using the IOM’s AFHS Assets Research Program.

TRANSFER ACTIVITIES OF THE AFHS ASSETS

Following on the findings and recommendations from the Disposition of the Air Force Health Study report (IOM, 2006), Congress took steps to implement its recommendations. AFHS personnel briefed the House Veterans Affairs Committee regarding their plans for close-out of the project and recommended that the Medical Follow-Up Agency (MFUA), the unit within the IOM responsible for data-intensive epidemiologic studies, be designated as the new custodian. MFUA had indicated that it would accept custodianship provided that sufficient resources accompanied it. Subsequently, Congress took steps to implement the

recommendations in the report, and passed two laws regarding the AFHS data and biospecimens. The first, Public Law 109-364, section 714, directed the U.S. Air Force to transfer custodianship of the AFHS’s data and biospecimens to the IOM of the National Academy of Sciences at the study’s conclusion (September 30, 2007). The second law, Public Law 110-389, section 803, directed the Department of Veterans Affairs (VA) to provide funding to the IOM to maintain and manage the AFHS assets in an effort to facilitate research aimed at “understanding the determinants of health, and promoting wellness, in veterans.” Appendix B contains the excerpts of those laws. The legislative wording was not restrictive regarding the type of research that could be conducted using the AFHS assets.

Several actions were initiated in early 2007 to prepare to close the project and to transfer the AFHS assets to the IOM’s MFUA. Cohort members were contacted to inform them of the conclusion of the study and to request consent to transfer custody of their data and biospecimens to the IOM. The Air Force contracted with Science Applications International Corporation to build a comprehensive digital database of all their records—electronic, paper, or otherwise. A biospecimens and data transfer contract was drafted between the Air Force and the National Academy of Sciences, the IOM’s parent organization, describing the general principles for the change in custodianship; it was signed in April 2007. In September 2007, a high-capacity storage drive containing the AFHS’s accumulated electronic data and digitized records for cohort members who gave consent for their study assets to be transferred to the IOM was given to MFUA and placed in a secure area at the IOM until funding could be secured to begin work on the data.

Because the IOM does not have the physical means 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 house the biospecimens. Several factors contributed to the decision to house the biospecimens at WPAFB, including willingness to invest in capital improvements that would prepare the site to hold the freezers containing the biospecimens, experience with managing biospecimens, having a secure space and a monitored environment with existing personnel and equipment, and having strong internal support for this initiative. Twenty-three ultralow temperature freezers were transported to WPAFB from Brooks Air Force Base, San Antonio, Texas, using trucks that were equipped to keep them continuously operating at –70°C.

Following the passage of Public Law 110-389, the IOM entered into negotiations with AFRL to establish a cooperative research and development agreement (CRADA) that would define the terms of AFRL’s management of the biospecimens under MFUA custody. The IOM and VA entered into a separate CRADA. It took approximately 3 years for a final agreement to be reached and for VA to provide funding to conduct the IOM’s work on the AFHS data and to provide AFRL with funds for the maintenance of the biospecimens. Funding for maintenance of the biospecimens and associated actions, such as aliquoting samples for researchers, at WPAFB was provided by a separate CRADA with VA.

To supplement the cost of biospecimen aliquoting and shipment, researchers are charged on a cost-recovery basis.

While the CRADAs were being negotiated, researchers had begun to contact MFUA to express interest in accessing the AFHS assets. As soon as funding was in place, MFUA was able to begin designing study procedures to make the AFHS data and biospecimens available to researchers. MFUA spent the first year developing infrastructure for the program, which included conducting an in-depth analysis of the files received, loading files containing contact information for cohort participants and building the corresponding contact database, and contacting participants to request consent for the continued use of their data and biospecimens. In addition, MFUA began loading the data text files transferred from the AFHS, which contained the data required to respond to researcher requests. An expert advisory committee was formed in late 2011 and held its first meeting in March 2012. The AFHS advisory committee consisted of eight members, including the chair. The membership changed over the years as some members rotated off and new members were appointed to replace needed expertise or include additional expertise based on program needs.

DETERMINING THE CONTENT AND VALUE OF THE TRANSFERRED AFHS ASSETS FOR RESEARCH

The AFHS data and biospecimens have long been recognized as valuable because of the availability of voluminous and detailed longitudinal data linked to numerous and unique biospecimens collected from a large sample of diverse men who were followed for more than 20 years. In addition to the contributions the AFHS assets have made to the scientific literature, it is their potential for use in research beyond the scope of the original study through the use of new assays, technologies, and methods developed since the samples were collected that continues to make the AFHS a valuable and useful resource. After the IOM became the custodian of the AFHS data and biospecimens, there were still many questions and uncertainties about the state of the data and biospecimens that had to be resolved before a research program could be realized.

AFHS Cohort Consent

As part of Public Law 109-364, the Air Force was required to contact all surviving AFHS participants to obtain their written consent for both their data and biospecimens to be transferred to MFUA within the IOM. Of the 2,758 persons who participated in at least one exam cycle, 2,488—more than 90%—cohort members gave their consent (1,031 Ranch Hands and 1,457 comparisons), indicating a high level of support for this activity. A complete set of data and biospecimens from those individuals, including deceased persons, was transferred to the IOM. Participants whose information was not transferred included those who

refused (5%) and those who could not be contacted after multiple attempts (5%). Two copies of the records for individuals who could not be contacted or who did not consent to the transfer exist: one was transferred to and is held by WPAFB, and one is held at the National Archives.

Once the data and biospecimens were transferred to IOM custody, additional consent was required to enable the AFHS data and biospecimens to be used for research other than studies pertaining to herbicide exposure or military health. Participants were informed that future research would likely involve a broad range of subject matter, including studies of health and well-being not directly related to military health or herbicide exposure. The IOM consent form contained the language below, which has implications for future custodianship (see Chapter 4):

Participants were asked to indicate whether they did or did not consent to allow their health records data and biospecimens to be made available for future research. The participants were informed that an advisory committee would approve all research and that their data and biospecimens would be subject to strict standards of security. Additionally, participants were asked whether or not they would agree to be contacted for any research involving new data collection. The entire consent process took approximately 18 months to complete. For participants who died between 2007 and when the IOM completed its own consent process for the cohort, consent was not legally required, but obtained from surviving spouses if possible, and participants’ data and biospecimens are available for use in research. Of the 2,488 participants who consented to have their records and biospecimens transferred to IOM custody, 2,210 (89%) gave consent to have their study materials used in future research (922 Ranch Hands and 1,288 comparisons). Moreover, 1,509 participants (61%; 606 Ranch Hands and 903 comparisons) indicated that they were willing to be contacted for research requiring new data collection.

Data Quality and Usability

Questions regarding the actual quality and usability of the data and biospecimens needed to be addressed before research using the AFHS assets could be realized. Although the Committee on the Disposition of the Air Force Health Study (hereafter referred to as the “Disposition Committee”) had concluded that the medical records and other study data appeared to have been properly maintained, it was not known whether the data were in a format that would allow trans-

fer from the system that the Air Force had used to systems used and maintained by the IOM. It was also unclear whether the data could be processed, understood, and made into a usable format by a different custodian, such as the IOM.

The Disposition Committee recommended that the AFHS electronic data be organized and fully documented, including the creation of detailed and comprehensive file specifications for data files and for each variable for each examination cycle; a codebook that would constitute a comprehensive distillation of data file contents; and a document describing the contents, format, and location of those materials that have been scanned into image files. None of those activities had been completed before termination of the AFHS in 2006, nor during the time between termination and when the contract between VA and the IOM was finalized in 2011.¹

The documentation provided by the Air Force custodians was limited to electronic file layouts, hard copies and PDFs of the questionnaires and translation codes (not the detailed and fully documented data dictionaries and codebooks that were recommended). Because the data were transferred as an assemblage of more than 123,000 disconnected text and other related files not in database form, the first year of the program was dedicated to documenting and reading in the disparate text files. For the next 3 years, in addition to continuing to read in the text files, much time and effort was spent on file integration, data cleaning, analyses required to create researcher data requests, and quality checks. Although the AFHS data are in electronic form, the data are not standardized between cycles or in a database format. Translation of these data into such a format is challenging and, since receiving the data from the Air Force, the IOM has committed a substantial amount of effort into quality control and database creation. One large difference between the IOM’s research program and other longitudinal epidemiologic studies, such as the Framingham Heart Study, is that data from the latter were collected in more organized fashion and immediately stored in study databases rather than according to the less systematic approach used in collecting the AFHS data. Beginning in the second year of the program (2012), the advisory committee began reviewing research proposals and with the first proposal approval, the program marked another milestone: the ability to process the data, compile a unique dataset for each approved study, and extract the data in a usable format for investigators.

The lack of well-documented data dictionaries led to data requests from investigators that were often vague, generic, or incomplete, referring to categories of desired information rather than specific variables. This resulted in delays in identifying the most appropriate variables for provision and subsequent requests for additional data. Some researchers provided no listing of any variables that they needed, but instead relied on the program staff to determine a listing based upon the information in the researcher’s proposal. The proposal process was up-

_______________

¹ The IOM did not have funds to support work on the assets until the contract with VA was completed.

dated after the first few instances of nonspecific variable requests to require that researchers familiarize themselves with the available information from exams and questionnaires and, at a minimum, list the types of information and for which cycles it is requested (for example, all cancer diagnoses for cycles 1, 4, and 6). Although some investigators specified the information they needed, strict inclusion criteria at times resulted in sample sizes that were too small to be used for the research proposed. In these cases, the IOM program staff worked with the researchers, under guidance of the advisory committee, to refine the selection criteria for the study. Furthermore, some researchers requested all information available with the intent of “mining” the data for associations. The committee acknowledges that there are benefits to exploratory analyses and that they are an important part of scientific inquiry to further develop and refine hypotheses, study design, and understand the underlying data structure, but it requires that such analyses be supported by specific aims and an analysis plan. As such, the advisory committee required prospective investigators to submit hypothesis-driven proposals.

Characteristics intrinsic to the data made its distribution to researchers difficult. For example, some variables regarding basic information, such as the marital status of a participant at any given time, were simply not available because of the way the data was collected. Participants were asked their current marital status in only cycles 1 and 6. For other cycles, current marital status can only be determined by reviewing marriages and other relationships that took place after the previous cycle. Many of these seemingly basic variables could not be made available until the raw AFHS data were cleaned and processed.

Before data could be provided to researchers, each variable had to be defined along with all codes used for it. As part of ensuring accuracy and completeness for data disseminated to researchers, each file was individually reviewed. As part of the quality assurance procedures, researchers were provided with data dictionaries specific to their dataset that indicated the exact location of the item in the data file, along with the size and type of the variable. A complete dataset for an individual researcher might require integrating data from more than 30 of the original text files. Each data file generated for a research study is unique. The process of generating the data file was time consuming because each step required ad-hoc custom programming, verification, and research to ensure accurate cohort selection and researcher-specific data retrieval and analysis. For each data request, a complete Data Use Agreement had to be crafted, detailing all of the data provided, time period in which the data could be used, and policies for returning derivative files and disposing of the original data file at the study’s conclusion. Data files had to be structured to respond to the varying needs and expertise of the researchers, and documentation related to the process of creating the data files was continuously updated, verified, and maintained. Although this method of distributing the data to approved investigators is one of several that could be used, the IOM program staff and advisory committee considered it to be the best approach available at present

as it allows for a customizable approach to each study. The studies generally have few requested data elements in common, and this distribution method is integral to processing the biospecimen requests.

For studies that requested biospecimens, the quantity and availability of the biospecimens determined the available study population that could be selected. Integrating the physical biospecimen requirements with restrictive cohort selection criteria resulted in many additional challenges. The IOM program staff worked with the researchers to run multiple iterations of cohort selection using different parameters to finalize and select the cohort of interest. Major revisions to the study design or the cohort selection criteria had to first be approved by the advisory committee, and re-reviewed by the IRBs if applicable.

Biospecimen Quality and Usability

The Disposition Committee had also recommended that the biospecimens be inventoried and fully documented prior to the AFHS’s scheduled termination, but this had not been completed when the IOM assumed custodianship of the assets. The Air Force had a record (a Microsoft Excel file) of each freezer’s contents by shelf, box, and participant. However, they did not verify this information before transferring the biospecimens to AFRL. The record did not detail the biospecimen history (when it was received, whether additional aliquots had been made, number of freeze–thaw cycles, if it had been given to other investigators, and the like). The IOM created a biospecimen tracking database from the Excel file received. The database maintains location information for each participant’s biospecimens, and location information is updated and maintained as requests are made and fulfilled. The database was created with a standardized volume per biospecimen type for each participant. The IOM program staff carefully review each selected participant’s biospecimen availability to ensure that no one participant’s biospecimens will be depleted before allowing processing by AFRL. As biospecimens continue to be selected, aliquoted, and distributed, the advisory committee has instituted preliminary cutoff volumes to ensure samples are not depleted. The IOM biospecimen database includes information for each consented individual participant; their biospecimens by cycle, type, and amount; and number of aliquots, freeze–thaw cycles, aliquot date, and to whom samples were provided. However, because the original biospecimen data file was not based upon a full inventory, the IOM biospecimen database includes errors that are not identified until a particular sample is requested. Although this strategy of updating the biospecimen database works as an interim solution, the committee concurs with previous committees that an updated biospecimen inventory is critical to maximize the efficiency of the repository and better inform future research with the knowledge of what is actually available.

The IOM program staff use the database to select the biospecimen samples that are to be processed and shipped to researchers requesting them. Multiple docu-

ments, including chain of custody, aliquot vial labels, sample request information, and protocol, are sent to AFRL that identify the participant, location of the participant’s biospecimen(s) of interest, and the requested aliquot size. Printed labels are provided for all newly aliquoted samples along with chain of custody forms. One occasional problem encountered regarding requests for biospecimen samples was when the information on their location in the transferred spreadsheet did not match the actual location. For example, some biospecimens that were selected for research were missing completely, some had less volume than recorded, and some were not in the location indicated by the information sent with the biospecimens. To resolve these inherent problems and to avoid unnecessary handling, program staff identify and select additional biospecimens in the database to be potential replacements for primary selected biospecimens that cannot be located. Part of the processing procedure includes a preliminary check of requested study biospecimens by AFRL staff to ensure all of the requested biospecimens are available in the volume required before beginning the aliquoting process. Notably, the majority of biospecimens have been located and were found to be properly maintained. Those additional practices instituted by the advisory committee contributed to greater accuracy in preparing a study’s biospecimens because the study cohort did not have to be reselected or changed if biospecimens were missing. It also ensured that the biospecimens were not wasted or subject to unnecessary handling. In all of the biospecimens selected for use in approved research studies—2,270—only 20 could not be found and had to be replaced.

Time Effect of Storage

The Air Force performed a reassay on selected AFHS biospecimens that had been frozen for 9–24 years to evaluate their stability, condition, and viability before the biospecimens were transferred to the IOM’s custodianship (Fox and Pavuk, 2006; Pavuk, 2006). Five participants (two Ranch Hands and three comparisons) were randomly selected based on the 988 participants who met the following inclusion criteria: participated in all six physical examinations, had multiple 10 mL frozen serum specimens collected at each of the first five physical examinations, no history of cancer, had at least one quantifiable dioxin measurement, and signed an informed consent form. One 10 mL serum sample per examination per participant was selected from the first five exam cycles for a total of 25 serum samples. Stored samples were analyzed using a microsphere-based multiplexed immunoassay method to identify biomarker and protein expression patterns. Each serum specimen was analyzed for 177 analytes (78 specific serum antigens, 43 autoimmune serologies, and 56 infectious disease serologies). Sixteen analytes were measured in replicate in at least two exam cycles and were included for comparison. Eight analytes (alpha–1 antitrypsin, C3 complement, creatine kinase, IgA, IgM, PSA, SGOT/AST, and TSH) had continuous measured levels with normal, laboratory set ranges, and eight other analytes (hepatic panel,

mitochondrial antibodies, and thyroidal microsomal antibodies) had a positive or negative finding in the AFHS data. Results from earlier analyses and the immunoassay panels were similar and the correlation coefficients on each of the 16 replicate analytes were statistically significant (p-value ranged from < 0.001–0.04). Of the 177 analytes examined 170 (96%) provided measurable results using the immunoassay panels. The seven analytes that did not yield measurable results are usually present only in pathologic conditions (IL-1beta, IL-2, IL-4, IL-6, MMP-9, glutathion-S-transferase, and calcitonin). Complete results were available for 147 (83%) of all analytes in analyzed samples across all five cycles. Pavuk stated that the results below standard curves appeared to be participant-related rather than related to the examination cycle or storage time, and were not related to dioxin level or status as a Ranch Hand or comparison. No difference in the number of results was observed for earlier compared with later examinations. The results suggest that the biochemical integrity of the serum samples appeared to be well preserved and remains viable given that sensitive immunoassays could be successfully performed. Similarities in results between the AFHS measurements taken at the time of collection and these later reanalyses of the same measurements were further evidence of the stability of the stored frozen serum samples.

Although the AFHS samples are relatively old (from 32 years for cycle 1 to 12 years for cycle 6), they have been stored at –70°C or below since their collection and have been found to have retained their usefulness for epidemiological studies. While many biospecimen repositories are now stored in liquid nitrogen or ultralow freezers because the colder temperatures slow the deterioration of biomarkers during long-term storage, older repositories were not traditionally stored in such low temperatures, which is another unique and valuable feature of the AFHS. Many biomarkers are expected to remain viable over time and not be adversely affected despite the age of the samples. For example, the AFHS samples from the earliest cycle have been used for genetic studies and showed that the samples were not adversely affected by long-term storage.

Quality of Biospecimens

Although the small studies of serum samples by Pavuk demonstrated that the biospecimens were viable, there was no guarantee that the rest of the serum samples or samples from the other types of biospecimens were viable. Because an updated inventory of the biospecimens, which should have included a visual inspection, was not completed prior to the transfer to the IOM, the quality of the biospecimens was unknown. The Disposition Committee visited the AFHS biospecimen repository at Brooks Air Force Base, where the biospecimens were maintained before they were moved to WPAFB under the aegis of the IOM. The committee randomly selected three samples from three unique participants in order to observe the retrieval process and inspect the preservation of the samples. The committee observed no signs of leakage or major thaws with any of the three

biospecimens, but did note that the position of the fluid in one urine sample was consistent with either minor thawing and refreezing or that the initial freezing had occurred in two phases (IOM, 2006). Because only three samples were inspected out of a possible 91,500, many unknowns remained regarding the quality of the biospecimens, which could have implications for their use in new research.

Fortunately, the AFHS biospecimens had been appropriately stored and preserved and the AFHS biorepository is considered to be of comparable quality to other large repositories that house biospecimens for epidemiologic studies (IOM, 2006). To ensure the quality of the biospecimens was maintained, the IOM staff and the AFRL staff created operating procedures and protocols for aliquoting or sectioning each type of biospecimen, developed material transfer and custody agreements, and created forms to document and catalog the number of vials, volume, and location of each biospecimen that was selected for use in a research study as the database inventory was developed.

Recent and distinct studies that used the biospecimens under the IOM’s research program have found that the storage conditions used over the more than 30 years have been adequate and that the biospecimens have retained a similar quality as “fresher” samples (Pavuk, 2006; Janice Chambers, Mississippi State University, personal communication, April 2014). Preliminary results from approved studies have demonstrated that adequate quantities and quality of genetic material can be successfully obtained from various biospecimens, including semen, adipose tissue, and whole blood (Kim Boekelheide, Brown University, personal communication, April 2014). These and other approved studies have confirmed that numerous biomarkers of interest are still viable, reliably measurable, and consistent in the AFHS samples. One such example is the enzyme paraoxonase 1 that was shown to have measurable, although lower, activity (18%) in serum collected in cycle 1 compared with activity measured in serum collected in cycle 6 (53%) (Chambers, 2014). Several of the assays used in approved research projects have been multianalyte assays, which provide test results for multiple analytes simultaneously on each individual sample, making the sample usage relatively efficient.

Approved research studies have used the AFHS biospecimens for epidemiologic studies of biochemical factors that were not possible at the time the biospecimens were originally collected. Biochemical measurement technologies have improved dramatically over the past 30 years, and the AFHS samples could potentially be used to assess a broad spectrum of outcomes beyond their original intent. Moreover, a broad array of other health information is available for each participant making it possible to explore new avenues of correlates to overall health.

SUMMARY OF THE APPLICATION PROCESS TO ACCESS THE AFHS ASSETS

The advisory committee was charged with fostering the use of the AFHS data and biospecimens in new research and reviewing research proposals. To fulfill its

charge, it developed criteria to be used to evaluate the scientific merit of research proposals to make decisions about which would be given access. Additionally the advisory committee created an application process. Its first step was to develop and disseminate a request for proposals (RFP). The RFP was first announced on April 30, 2012. Researchers from the United States and abroad, from academic institutions, industry, and government organizations, were eligible to apply for access to the AFHS assets. The RFP was publicized on the AFHS project page of the IOM website and initially sent to researchers and organizations that were familiar with the AFHS or had an interest in research related to herbicide exposure.

Researchers were asked to first complete a letter of intent limited to 500 words that was reviewed by the committee to ensure that the primary hypothesis, research aims, and general proposed methods were well justified and scientifically sound, and that information being sought was available. Each letter of intent was assigned to two members of the advisory committee to review and evaluate independently. The reviewers for each letter of intent were specifically chosen based on their expertise that most closely matched the proposed investigation. Reviews were compiled, discussed by the entire committee, and a consensus decision was made about whether to invite the investigators to submit a full proposal. Those researchers who were approved were then invited to complete a full proposal. After experimenting with submission deadlines, beginning with the May 2013 RFP announcement, the committee chose to use a rolling submission process for both letters of intent and invited proposals. That served to both increase the number of interested researchers because they were not restricted by set deadlines, and to prevent a backlog of datasets and biospecimens to be prepared as would occur when several proposals were approved concurrently.

The same two committee members who were responsible for the primary review of the letter of intent were assigned to be the primary reviewers of the full proposal. Reviewers used the evaluation criteria outlined in the RFP, which are reproduced in Box 3-1. If necessary, a third committee member with relevant expertise was also asked to review specific sections of the proposal. For example, the committee member with biological sample repository and laboratory expertise would evaluate the parts of the proposal related to biospecimen usage. All evaluations were compiled, and the scientific merit of the proposal discussed by the full advisory committee before a decision on access to the requested assets was made. Although all proposals had to meet a minimum threshold to be approved, those proposals that requested the use of biospecimens had to meet a higher standard than proposals that requested data only because the biospecimens are a limited resource. Applications requesting biospecimens that have been judged scientifically meritorious, undergo review for feasibility testing of the requested biospecimens for the planned laboratory assays or evaluations. This feasibility assessment is conducted by committee members who are knowledgeable about the requested biospecimen types and the assays that are intended to be used. Factors considered in the review include adequacy and appropriateness of the requested

biospecimen type and amount, anticipated stability of the analyte under the known storage conditions, and known technical performance of the planned assays on the requested samples. Whenever possible, the amount of residual sample is minimized by careful planning of biospecimen amounts provided.

Proposals that included requests for pilot funds were critically reviewed for both scientific merit and to ensure that awarded pilot funds were likely to facilitate generation of preliminary data or results that could then be used to apply for larger grants to further the research, instead of solely being a mechanism to support research projects without future planning. Committee members would

sometimes request additional information or detail from the investigators before making a final decision.

Researchers whose proposals were approved were granted access to the requested electronic data and, if applicable, biospecimens, contingent on IRB approval from both the researcher’s home institution and the National Academy of Sciences’ IRB. Those investigators were required to give annual updates to the advisory committee that included their accomplishments, any difficulties encountered and their resolution, and preliminary results. Communication between investigators and the committee was maintained by the IOM. Thus, the advisory

committee provided scientific and administrative oversight of the research and evaluated results from both feasibility/pilot studies and larger studies.

Since the IOM became the custodian of the AFHS assets, more than 80 inquiries from researchers have been received. Many of these were related to the specific diseases and conditions that developed in the cohort over the 20-year follow-up of the AFHS, several of which had been attributed to Agent Orange exposure as outlined in IOM reports (see Veterans and Agent Orange series reports from 1994–2012). As of February 1, 2015, 26 letters of intent were submitted by investigators, 24 of whom were invited to submit full proposals. The Web-based system used to submit the letters of intent and proposals also tracks the number of applications started but not completed or submitted; 16 additional letters of intent were started but not submitted. Of the 24 proposal invitations, 19 complete proposals were submitted to the committee for review. Among those investigators who chose not to submit a full proposal, staff learned through communications that for at least 4 of them the lack of funding support influenced their decision, especially after pilot funds were no longer available. Of the 19 proposals reviewed by the committee, 13 were approved; 6 studies required access to the data only, and 7 studies used both data and biospecimens. Table 3-1 provides a summary of approved studies. The abstracts of approved studies are found in Appendix C. The proposals that were not approved did not meet the standards for scientific merit outlined in Box 3-1. In its decision letters to principal investigators of proposals that were not approved when first reviewed, the committee detailed its major concerns and allowed the investigators to respond and clarify any of the points and refine issues of their proposal before the committee made a final decision on access to the AFHS assets. The committee requested additional information on 11 proposals, and ultimately approved 5 of them for access to the AFHS assets.

Generating Interest

Following the announcement of the first RFP, the committee designed a two-page flyer to advertise the availability of the AFHS assets. The flyer was intended to increase awareness of the availability of the AFHS assets for researchers, particularly in fields outside herbicide exposure and toxicology. Both committee members and the IOM staff brought copies of the flyer to national meetings including the American Public Health Association, the American Geriatric Society, the American Society of Clinical Oncologists, and the American Association for Cancer Research. Project staff made presentations about the AFHS research program at national meetings to encourage use in new research directions. The committee announced the availability of the data and biospecimens for use in research on several listservs, including the VA listserv of all chiefs of staff for research, listservs targeting aging and geriatrics research, and announcements to the toxicology-research community, all of which resulted in increased interest as demonstrated by many new queries and submitted letters of intent. An invited

TABLE 3-1 Studies Approved for Use of the AFHS Data or Biospecimens Since 2012^a,b

^a This list represents studies approved as of February 1, 2015.

^b The date that a proposal was approved and the date the study was able to start varied depending on the type and extensiveness of the requested assets. For example, studies that requested more biospecimens (for example, more than 100) took longer to process than those requesting fewer samples.

^c Denotes studies that were awarded pilot funding.

paper summarizing the accomplishments of the AFHS Assets Research Program is under review by Military Medicine.

Limited pilot funding was available in the first 3 years of the program, as was required by Public Law 110-389, section 803, although no funds were awarded until the second year. As noted above, the first year was dedicated to transferring the assets and setting up the program infrastructure. All investigators were eligible to apply for pilot funding; however, only the most meritorious proposals were awarded funds. The goal of awarding funding was to provide assistance for feasibility studies (such as for obtaining biospecimens and reagents), rather than to entirely support the proposed work. The committee could award up to $250,000 per year. Three proposals were awarded pilot funds in the second year and four proposals were awarded pilot funds in the third year. Funding ranged from approximately $12,000 to $84,000 per proposal. The availability of pilot funding served to generate additional interest in the assets from the research community.

RESEARCH APPROVED FOR THE USE OF THE AFHS ASSETS

Although the Disposition Committee recommended that a 5-year commitment should be given to creating and operationalizing the new AFHS management program, and an evaluation of its success should not occur less than 2 years after the final pilot funding awards, the VA contract supported only a 4-year commitment with a report on the feasibility and advisability of maintaining the AFHS assets for future research required in the fourth year. The first proposals were approved in 2012. Given the amount of work required to prepare the data to be used in research, this did not provide sufficient time for most researchers to complete their work. While it is premature to expect publications so soon into the program, the committee was able to review preliminary results from approved studies in the feasibility stage, and one investigator was able to publish his results (Mazur et al., 2013, 2014).

Summary of Approved Studies

The research proposals reviewed and approved for use of the AFHS biospecimens and data span a wide range of scientific and biomedical research topics (see Table 3-1 and Appendix C). The extensive data and biospecimens collected over the extended period of the AFHS makes it valuable for examining potential etiological factors relating to rare diseases or diseases and conditions that have a long latency periods, or for examining biomarkers associated with the natural history of slowly progressive disorders. For example, diagnostic data have been combined with DNA analysis from whole blood, semen, and adipose tissue samples to investigate genetic markers of disease predisposition or prognosis. Use of technology, such as next-generation sequencing, enables DNA data from an entire genome to be obtained from a whole blood sample of less than 0.5 mL

(Koboldt et al., 2013). Investigators generating whole genome data from the AFHS participants can use the subset of the data that is relevant to their study. Ideally, whole genome data generated by a single investigator will also be made available to other investigators (see Chapter 4), so that the samples do not need to be used to generate whole genome data more than once.

Just over half of the studies that the advisory committee approved were for use of the AFHS biospecimens and data, while the others requested and received data only. Several of the studies that were approved for use of the AFHS biospecimens were pilot studies for which limited preliminary data were available due to prior nonavailability of suitable biospecimens. These studies were successful in establishing the feasibility of applying novel assays to the stored biospecimens to generate interesting scientific leads. Results from pilot studies, which may be insufficient on their own to produce a research publication, provide critical information to inform next steps in these challenging areas of research.

The study of monoclonal gammopathy of undetermined significance (MGUS) is one good example. This study made use of the longitudinally collected serum samples to examine expression levels for a panel of microRNAs and other novel biomarkers for their association with development of MGUS and its progression to multiple myeloma, a disease with incidence that increases with age. Serum samples for individuals were examined at three time points to determine biomarker changes. Promising results from this pilot study could provide justification for study of these biomarkers in expanded collections of serum samples or in prospective clinical trials of MGUS.

A second approved study investigated potential etiologic factors for the rare blood disorder amyloid light chain amyloidosis, which has a reported incidence rate of 1,200 to 3,200 new cases per year in the United States (Amyloidosis Foundation, 2014). Matched sets of serum and adipose tissue specimens were required to conduct the biomarker evaluations necessary for the study. The first step was to provide 228 serum samples that had matching adipose tissue samples for the same year of collection. All serum samples were initially screened for abnormal levels of free immunoglobulin light chains, a biomarker of a plasma cell dyscrasia in the bone marrow due to amyloid light chain amyloidosis, multiple myeloma, or MGUS. Additional diagnostic tests of serum protein electrophoresis and serum immunofixation electrophoresis were performed on the six samples (3%) that had screened positive for abnormal levels of free immunoglobulin light chains, which is suggestive of the possible presence of plasma cell disease. Supplementary tests to identify biomarkers of organ dysfunction were also performed. Matched adipose tissue samples for the six serum samples that screened positive were then tested for tissue confirmation of light chain amyloidosis (David Seldin, Boston University, personal communication, April 2, 2014). The rarity of biorepositories with clinically annotated adipose tissue and matched serum samples would have rendered this research project infeasible had it not been for the AFHS resources. It would be prohibitively expensive to conduct new prospective studies to obtain

the rich amount of data and unique biospecimen collections that were already available in the AFHS for these projects.

Other approved proposals examined long-term health outcomes and changes in biomarkers, DNA methylation, or other measures with regard to response or susceptibility to health endpoints (Boekelheide, Chambers, Mitchell, Ramos, Roth, Seldin, Shim); using biospecimens for early detection, identification of therapeutic targets, and/or disease response markers (Chambers, Mitchell, Seldin); and new statistical analyses, methods, and models of exposures on health effects (Haws, Mandel). Research that consisted of using new analytical techniques to examine questions that were not addressed as part of the AFHS was approved (Batty, Mazur), as was using publicly available information to expand the period of follow-up for the cohort (Haws).

The research performed using the AFHS assets in the course of the original study has benefited veterans through findings from studies of dioxin and herbicide exposure that were unique risks in this population. Several of the earliest approved studies under the IOM’s program focused on the health effects and potential biomarkers of dioxin and other herbicide contaminants that may provide further evidence of association and other occupational exposures or environment effects specific to individuals’ service in the Air Force during Vietnam. The fact that these were among the first proposals is not surprising given that these investigators were already familiar with the “Ranch Hand Study” and its rich data and biospecimen resources and they were specifically made aware of the RFP. For example, three research investigations proposed using different assumptions and approaches than have been applied to date to reanalyze outcomes examined by the AFHS (Haws, Knafl, Mandel). Another research study linked administrative records from VA to gather additional follow-up health and outcome information from a subset of AFHS participants (Ross). The AFHS resource remains relevant for dioxin-related research through the application of novel technologies (e.g., genome-wide epigenetic assays) to more deeply interrogate the molecular consequences of dioxin exposure, for the application of novel statistical approaches, or for further research focusing on new or longer-term disease or quality-of-life endpoints (such as reproductive outcomes).

Awareness of the AFHS resource and its potential value beyond dioxin research has been disseminated throughout the research community, as evidenced by a steady increase in the diversity and number of proposals received. The AFHS cohort is a subset of the U.S. population, and although not representative since the cohort consists of mostly white men whose average age was 64 years at the end of the original study period, it has a diversity of socioeconomic status and educational background and the members have many of the health outcomes that align with those observed in the broader similar demographic subset of the U.S. population (e.g., prevalence of heart disease and diabetes). Findings from new epidemiologic or biomarker studies continue to be relevant and applicable to veterans but are also relevant to all U.S. men of similar demographic makeup.

Additional research topics that have emerged among the submitted and approved proposals include aging, neurologic diseases, and cancer research, showing that AFHS data and biospecimens remain relevant for important biomedical and public health research projects.

Investigator Experience Using the AFHS Assets

In an effort to determine whether researchers found the AFHS to be a valuable and useful resource, the committee designed a short questionnaire and solicited feedback from the principal investigators of approved research studies after they had received all of their requested data or biospecimen samples. The committee asked these investigators to give specific examples of how the AFHS assets were unique for their research and any suggestions they could provide to make the AFHS Assets Research Program even more useful or valuable. Questions were primarily subjective and related to the overall application process, quality and completeness of the data received, quality and usefulness of the biospecimens, and future plans regarding the results and additional uses of the AFHS assets.

The overall feedback was supportive of both the AFHS assets as well as the application process. A majority of the investigators indicated that they learned about the AFHS through the IOM’s project website, but a number of them were already familiar with the AFHS and learned about the opportunity for using the assets for new research through colleagues or direct inquiry. Every one of the investigators used the information found on the project website to plan their research and found it to be helpful for planning their study. All of the investigators had contacted the IOM staff before submitting a letter of intent, and none reported any issues or problems with the application, pilot funding (if applicable), or award process. Principal investigators reported that the IOM program staff was able to provide helpful guidance, and that the committee’s feedback on letters of intent when making invitations to complete a full proposal was also helpful in refining project goals or improving study design or approach.

The investigators who received and used the AFHS data generally rated the completeness of it as 4 or 5 on a scale of 1–5, with 5 being very complete. While at least half of the investigators’ studies were still in progress when they provided feedback to the committee, they were confident that the quality of the data, both provided and generated, would allow them to draw reliable conclusions. Even though work was continuously ongoing to prepare the data for use in research, preparing researcher data requests could take weeks to months depending on the variables of interest; however, all of the investigators who had received all of their requested materials reported that it had been provided to them in a “timely manner.”

More than half of the approved research studies required the use of biospecimens in addition to data. Numbers of biospecimens requested for studies ranged from 12 to 1,163. All investigators reported that the samples were clearly labeled

and in good condition and they were able to perform the proposed assays successfully on the samples. Both DNA and microRNA were isolated successfully from the samples.

The AFHS Assets Research Program began accepting applications in mid-2012, and several of the research studies are still in process (especially for those studies approved in 2013 and 2014). However, one principal investigator was able to publish two peer-reviewed articles (Mazur et al., 2013, 2014) in this relatively short time period. The other investigators have indicated that they are either in the process of drafting or submitting publication(s) or expecting to do so once their study is complete. Although no investigators have used the preliminary data generated by analyses of the AFHS assets in a subsequent grant application, it is too soon to know whether the pilot projects will produce results that lead to such applications. All but one of the investigators stated that their research could not have been completed without the AFHS research assets. The investigators, except one who was interested in pursuing a different line of research, stated that they found the AFHS data and biospecimens to be unique and valuable, and would consider using the AFHS assets for future research if possible. When asked about their overall experience with the AFHS Assets Research Program, all of the respondents characterized their experience as very good or excellent. And all of the investigators stated they would recommend use of the AFHS to others. As an example, Dr. Chambers stated, “Such cohorts and data collection are not easy to find…. This could be a tremendously valuable resource for answering questions … to the toxicology community.” Dr. Chambers further stated,

This chapter described the AFHS management program in terms of its ability to provide valuable data and biospecimens for research despite the many difficulties, unknowns, uncertainties, and complications that had to be addressed by the IOM. The following chapter provides an overview of possible options for the continued management of—and access to—the AFHS data and biospecimens. It also outlines several aspects of data management that need to be considered as the research program continues to grow and move forward and offers some of the many possible opportunities for future research using the AFHS data and biospecimens.

REFERENCES

Amyloidosis Foundation. 2014. Description and background. http://www.amyloidosis.org/treatmentinformation/primaryal.html (accessed August 5, 2014).

Chambers, J. 2014. A longitudinal study of paraoxonase 1 (PON1) in relationship to type 2 diabetes and aging. PowerPoint presented at the Advisory Committee meeting, Washington, DC.

Fox, K. A., and M. Pavuk. 2006. Testing the viability of stored frozen serum samples from the Air Force Health Study using human multi-analyte profiles (MAP). Organohalogen Compounds 68:1390-1393.

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

Koboldt, D. C., K. M. Steinberg, D. E. Larson, R. K. Wilson, and E. R. Mardis. 2013. The next-generation sequencing revolution and its impact on genomics. Cell 155(1):27-38.

Mazur, A., R. Westerman, and U. Mueller. 2013. Is rising obesity causing a secular (age-independent) decline in testosterone among American men? PLoS ONE 8(10):e76178.

Mazur, A., R. Westerman, A. Werdecker, and U. Mueller. 2014. Testosterone and type 2 diabetes in men. Aging Male 17(1):18-24.

NIH (National Institutes of Health). 2014. How reviewers score applications. http://www.niaid.nih.gov/researchfunding/grant/strategy/pages/5scoring.aspx (accessed December 10, 2014).

Pavuk, M. 2006. Testing the viability of stored frozen serum samples from the Air Force Health Study using human multi-analyte profiles (MAP). AFRL-HE-BR-TE-2007-0015. Brooks-City-Base, TX: Air Force Research Laboratory Human Effectiveness Directorate, Information Operations and Special Program Division.

This page intentionally left blank.