Biospecimen and Data Collection and Distribution
This module describes the principles governing the collection, processing, storage, and distribution of materials and data for the National Biospecimen Network (NBN), after the patient
has provided informed consent. NBN system requirements with respect to its organization, users
and contributors, tissue sources, collection, annotation, inclusion of longitudinal data, possibility
for advanced analyses, storage, distribution, safety, and incentives to participate are further
detailed. This module also reviews existing tissue banks and explains how the NBN is expected
to differ from and interact with them. |
3.1 Introduction
The National Dialogue on Cancer (NDC) Tissue Access Working Group (TAWG) called for the
establishment of a nationally coordinated, standardized, inclusive, high-quality network of
pathological and normal tissue collection, storage, and distribution, with associated clinical data,
that is accessible through a user-friendly informatics system. It also envisioned the need to
develop and implement national standards for the proposed system in areas such as sample
collection, annotation, storage, and distribution. Such standards would also be developed for
training, site monitoring, sample tracking, and quality management. The implementation of
standard collection protocols is expected to minimize experimental variability and accelerate
scientific progress. Variability in acquisition, processing, and storage of tissue may contribute to
undesired experimental variability in whole-genome analysis. One of the uncompromising
propositions of the NBN must therefore be the collection of biospecimens and associated data
according to known and widely accepted standards, and in conformance with ethical and legal
requirements (see 2. Management of Ethical and Legal Issues).
Although the NDC TAWG thought a fresh start would be optimal to achieve the goals of the
NBN, the NBN Design Team recognized the complementary value of many existing resources
and their usefulness for a range of purposes. The RAND Corporation review of selected existing
resources observed that some existing collections might be useful for some research, but
collection methods varied with not all storing frozen tissues, making comparisons across
resources difficult. Most repositories designed for research purposes collect demographic and
diagnostic data from tumor registries and/or patient medical records. It will be necessary for the
NBN to critically evaluate existing resources for their specific research utility, and to work with
valuable extant resources to maximize benefits to research.
3.2 Background
3.2.1 Existing Resources
Thousands of tissue repositories currently exist within the military, private industry, hospitals
and hospital consortia, cooperative groups, research programs, and academic institutions.
However, there is great variance among these existing resources in their focus (e.g., treatment or
research), the resources available for tissue testing and annotation, and the ability to control the
use of tissue samples. Many of these specimens were not collected or stored for research, but are
essentially pathology archives. These existing repositories share no specific standards of patient
consent procedures, acquisition, storage, or quality control (QC); are not able to share easily
either data or tissue; and are not linked in terms of tissue distribution.
Many existing tissue banks have served research well and are valuable for advancing knowledge
through the first stages after discovery. For example, the Cooperative Breast Cancer Tissue
Resource has more than 9,000 annotated samples and is a major source of tissue from patients
who have participated in randomized trials. Specimens from clinical trials in many organ
systems, mainly routine formalin-fixed paraffin-embedded tissues, are available in the
repositories of the major clinical cooperative group banks. High-quality specimens are also
stored in Specialized Program of Research Excellence (SPORE) repositories and laboratories
participating in the National Cancer Institute (NCI) Director’s Challenge: Toward a Molecular
Classification of Cancer.
RAND is examining a number of existing tissue resources to help clarify the nature of available
resources.1 Preliminary results confirm that few common standards exist among resources in
terms of how specimens are collected, stored, processed, and distributed. This is partly because
the repository design is integrally linked to an institution’s objectives, and these resources were
designed to serve multiple purposes including: Prospective collection, storage, and distribution;
standardized tissue microarrays; diagnosis; and study-driven banking. The RAND team also
observed that anticipating a repository in the initial study design, rather than assembling one
after the fact, increases the likelihood of standardization across samples.
Most of the resources examined by the RAND team receive a substantial amount of public
funding. In general, the repositories are integrating new technologies, and tissues from most
repositories are already being used for genomics/proteomics research. All of the repositories are
actively addressing ethical, legal, and privacy issues. These resources also market their
availability primarily to the scientific community, and the majority of consumers are academics.
Other highlights from the RAND study include the following:
Biospecimen Collection and Annotation
- All the repositories store paraffin-embedded tissue; many store frozen as well.
- All resources record pathological data. Some also record clinical data (taken from chart
review).
Bioinformatics and Data Management
- Most repositories use bioinformatics systems as a repository management tool.
- None of the repositories allows public access to individual patient data.
- Every repository has its own network security system.
Privacy and Ethical Issues
- Types of consent vary: Full consent, general surgical consent, study-dependent consent,
and exempt.
- Institutional Review Board (IRB) review for tissue use was required for all resources.
- In most cases, donors (patients and well volunteers) did not receive compensation.
3.2.2 Best Practices
Collection of specimens must first meet the needs of patient diagnosis, and procedures should
emphasize quality and follow standardized protocols to the extent possible. New national
standards based on best practices would be implemented as the basis for the new network
processes and upgrading of existing resources. These standards would cover every aspect of the
systemcollection, freezing/fixing, storing, and shipping. Such standardization would improve
comparative research, encourage the use of common technologies and testing methods, and make
it easier to merge data and conduct multidisciplinary research. The development and
implementation of national standards would require systems to assure compliance and oversight,
and provide incentives to ensure success. Quality assurance (QA) policies and procedures for the
NBN could be enforced by periodic site visits by a QA committee. Discussion of this subject is
found in 6. Governance and Business Models.
A potential model for creating and enforcing widely accepted standards and protocols nationally
is that of the organ-transplant banks and their associated organizations. Although the purpose of
these collection methods is primarily to maintain viability for transplant rather than to serve
research, the cumulated wisdom of the organ-transplant banks may still provide insights that can
be applied to the NBN.
The American Association of Tissue Banks (AATB) has developed a system for procurement,
standardization, storage, and access to tissue, and has an existing structure and network design. It
publishes standards to help ensure that the conduct of tissue banking meets acceptable norms or
technical and ethical performance, and provides technical information that describes procedures
to foster reasonable and responsible approaches to recovery, processing, preservation, and
distribution of transplantable tissue.2 The National Committee for Clinical Laboratory Standards
is developing standards for tissue collections, and the International Society for Biological and
Environmental Repositories (ISBER) is producing a comprehensive set of best practices for
tissue banks that will cover collection, storage, retrieval, packing, and shipping, and treatment of
human subjects (see Appendix K for the ISBER proposed best practices).
Another example of common practices is the set of standardized cancer protocols or formatting
standards for 42 tumor types to be used for the evaluation of surgical pathology specimens,
developed and published by the College of American Pathologists. The American College of
Surgeons Commission on Cancer (CoC) will require that pathologists at CoC-approved cancer
programs use the checklists in their surgical pathology reports starting January 1, 2004.3 All
operations at sites in the NBN and at the central site would be covered by standard operation
procedures to be refined as the network is formed.
3.3 NBN System Requirements
3.3.1 Organization of a Tissue Collection Network
The NBN Design Team recommended that the national tissue resource be organized as (a) a
decentralized network, possibly of nonprofit, tissue-repository organizations located near
academic medical centers; and (b) a virtual repository with data networked across the nation. The
approach should involve standardized acquisition, storage, and distribution of tumor tissue and
other samples (including, but not limited to, healthy adjacent tissues, buffy coat cells, serum,
plasma, and urine) by trained personnel whose primary responsibility is to meet NBN objectives.
Tissue would be obtained from patients who had given consent to dedicated, appropriately
trained personnel using standard protocols. The use of complex technologies that form the basis
for molecular, genetics-based studies would dictate a priority on fresh-frozen tissue acquisition.
The samples would be clinically annotated and “deposited” within a repository of both tissue
samples (likely a distributed physical network) and derived data (likely a virtual network feeding
a centralized national database). In addition, the NBN could contribute to the standardization of
existing repositories, improve flow of specimens to researchers, broaden the specimen types that
are collected, and develop data sharing processes and platforms to support more effective
specimen distribution.
To help communication to broad audiences, the Design Team found helpful a vignette that
follows the life cycle of a biospecimen through the NBN system for highlighting the salient
considerations from tissue donation consent by the patient, through researcher access to
specimens and data, and finally to efficacious treatment products. A summary of the scenario is
included as Appendix L.
3.3.2 Users and Contributors of Biospecimens
The proposed NBN is designed to serve scientists and clinical researchers from academia,
industry, and government. Access to both tissue and data derived from tissue should be broadly
available. Extensive external specimen sharing would be required of NBN collection centers on a
national scale. Additionally, a single, centralized policy for biospecimen distribution that
guarantees timely peer review of research applications seeking to access the resource needs to be
developed. Specimen allocation is further discussed in 6. Governance and Business Models.
It is imperative that the NBN have the full confidence and support of tissue donors, who are
integral to the building of the NBN into a national asset. The fundamental principle of informed
consent governs the use of materials and information that patients provide. These principles and
associated requirements are discussed in 2. Management of Ethical and Legal Issues. Experience
has shown that patients generally are willing to donate their tissue for research.4 If new drugs are
to be developed, commercial entities (pharmaceutical and biotech companies) should be allowed
full access to the resources of the NBN, and should not be prohibited from making a profit on
their investments.
Given the importance of studying health disparities across socioeconomic groups, the NBN
should be cognizant of fairly well-documented sociocultural factors that underlie the willingness
of different racial, ethnic, and cultural groups to donate biospecimens and participate in health
studies. For example, differences in willingness to donate blood, organs, and tissues for
transplantation have had the unfortunate consequence of reducing the availability of appropriate
organs for certain groups. If such factors also affect the willingness of some groups to donate
tissues for research, the NBN may face a similar problem. There may be reluctance on the part of
racial and ethnic minorities to donate to national repositories because of lack of trust in the
system.5 The NBN must make every effort to recruit a population of donors diverse in racial,
cultural, and socioeconomic characteristics if it is to support research that will help to further
elucidate genetic and environmental factors contributing to health disparities.
3.3.3 Tissue Sources
Eiseman and Haga (1999) estimate that “more than 307 million tissue specimens from more than
178 million cases are stored in the United States, accumulating at a rate of more than 20 million
per year” (p. xvii). They place the estimate of available specimens collected specifically for
research purposes at around 2.3 million. Sources of tissue may include academic medical centers,
large community hospitals, and existing repositories. Community hospitals, the primary point of
diagnosis for roughly 80 percent of cancers, would provide the surgical volume to be a major
source of tissues for a national network. The community setting also reduces the potential
selectivity that can occur in academic medical centers. Greater use of community hospitals
would also allow researchers using the NBN to benefit by access to more demographically
diverse patient populations.
Despite these many benefits for a national resource, community hospitals would need incentives
and assistance to develop the experience, infrastructure, and understanding of research necessary
to establish viable collection centers, particularly since these organizations conduct little or no
research on tissue. It remains an empirical question whether deploying a group of experts to community hospitals to help raise their levels of pathology and surgical expertise will greatly
improve medical care in their communities, provide incentives for hospital and medical
personnel participation, and allow the repository to reach the 80 percent of cancer patients who
are served in community hospital settings. What is more clear is that the full engagement of
community hospitals will require that human resources, coordination, and oversight challenges
be overcome, and that appropriate incentives exist to encourage their participation. The NBN
might consider financial support for infrastructure expansion, training, and salary support for
new staff. These incentives would be tailored to the unique needs of community hospitals.
Another model is to have one or two larger community hospitals paired with a regional academic
medical institution with experience in tissue resources. Institutions more experienced in research
could assist in establishing the infrastructure and training to develop quality tissue collection
programs. Criteria for selecting collection sites should include the following: Adequate specimen
flow, good followup, willingness to comply with collection protocols, and diversity in
geographic and other demographic characteristics of patients. It should be noted that diversity in
the resource does not necessarily require that every site be diverse.
The NBN might consider recruiting other specimen collection sites through merit-based
competitions based on NBN specifications (see discussion in 6. Governance and Business
Models). Any qualified organization would be encouraged to participate. Existing specimen
resources will be eligible to join the NBN if they meet the NBN certification criteria, including
rates of longitudinal data collection and diversity of tumor characteristics. Thus, existing
resources such as the SPOREs, Clinical Cooperative Groups, and Cancer Centers could play an
important role in the NBN. It will be important to develop incentives for these sites to participate
in the NBN, as they provide high levels of expertise and they together can contribute to the
diversity of the patient or donor pool in terms of cancer types, geography, and demography. The
NBN could supplement current funding to encourage participation in broader tissue research
efforts and provide a ready source of high-quality specimens, including frozen tissue, with
clinical annotation, correlative/translational research results, and followup data. The NBN might
also develop special access provisions for clinical trial-based samples, since these samples will
likely have a unique importance to a particular study group. Restricted access may be beneficial
and an incentive for clinical trial groups to associate with the NBN.
3.3.4 Specimen and Data Collection
The Design Team recommended that specimens from a broad range of cancer types be collected
initially and that the NBN should be viewed primarily as a biospecimen bank with general
guidelines on types and quantities of biospecimens available, rather than a prospective targeted
collection source where specific types of specimens are collected only after a researcher has
made a request for them. The goal should be to collect and bank the number and types of
specimens needed by the research community, not just the ones easiest to obtain. Some cancers
would be more commonly represented, at least initially, because they produce more abundant
tissue, occur more frequently, or are less “damaged” by adjuvant therapies. It is possible to apply
statistical models to determine the optimum amount of each tissue type, as collection for all types
of tissue should not need to expand indefinitely. As it becomes very expensive to continue to
collect specimens once a sufficient number has been banked to meet research needs, the NBN should be cognizant of the number and amount of banked specimens needed to meet researcher
needs. The size of tissue fragments available to the repository would vary with the type of tumor.
The availability of metastatic and uninvolved tissue, expected specimen size, and most likely
institutional source is shown for 15 cancer types in Table 3-1. This information, along with the
incidence of specific cancers and annual deaths, can provide parameters for evaluating the
adequacy of biospecimen inventories.
The suitability of the tissue samples for genomic and proteomic research varies with the organ
system. Organs and their neoplastic derivatives that have high contents of proteolytic and other
degrading enzymes such as pancreas, kidney, and liver, tend to degrade rapidly following ischemia
and vary in their usefulness for research, especially for studies at the mRNA level. The usefulness
for research of any tissue obtained from surgery varies with the time the organ is isolated from its
vascular supply at surgery and how fast the tissue for research is cooled from body temperature.
Nevertheless, many tissues remain useful for extended times and, even after death, remain
potentially viable. For example, pituitary fibroblasts and hematopoietic cells have been grown in
culture from tissue obtained from autopsies performed up to 48 hours after the deaths of patients.
Specimens not used by the NBN could be available for other collection efforts or would be
discarded. Some experience suggests that limiting collection of tumors in the operating room
(OR) to specific types has adverse effects on the overall collection process. On the other hand,
the Cooperative Human Tissue Network (CHTN) has demonstrated that specimens can be
effectively targeted to meet needs without compromising the collection process.6 Tissue would
be divided into sections and would be preserved in several ways (e.g., paraffin block: formalin
fixation; paraffin block: ethanol fixation; frozen: optimal cutting temperature compound; frozen:
no preservative) that would allow it to be used in a variety of techniques. A small portion of each
section would be taken before preservation for QC purposes.
Most samples are (and likely will continue to be) gathered in the OR, or in the surgical pathology
laboratory associated with the OR; consequently, surgeons and pathologists currently have initial
control over the samples. The NDC TAWG suggested that the preferred scenario would have
highly qualified, trained personnel (employed by the organization or consortium that would
govern the new national network, working under the supervision of a pathologist) present in the
OR/surgical pathology laboratory. Such dedicated staff would have the responsibility for
monitoring the surgical schedule, arriving at the OR/surgical pathology laboratory at a preset
time, taking possession of the tissue in an appropriate container, bringing it to the pathologist for
evaluation, and overseeing initial processing in a standardized fashion. (Patient prescreening and
consent would have occurred before the patient entered the OR.) A training and oversight
program would be required for these individuals. This scenario would maximize the potential for
following NBN data collection and associated annotation standards.
A less costly scenario that would rely upon independent surgeons and pathologists to comply
with NBN collection standards was considered less preferable because of reduced NBN control
over the handling of the samples and data. The TAWG also observed that surgeons and pathologists have roles
in patient care that might preclude their taking responsibility for the
initial tissue processing. Surgeons are legally obligated to provide specimens to pathology for
diagnostic evaluation, so pathologists at the submitting institution would be responsible for the
initial evaluation of tissue and should determine which samples are available for research.
Table 3-1. Characteristics of Tissue Sample Collection by Type of Cancer
(1) Type of Cancer |
(2) Incidence 1999 (new cases) |
(3) Annual Deaths 2001 |
(4) Tissue Amt Available from biopsy or surgical resection |
(5) Likely Institutional Sources |
(6) Adjacent Normal Tissue |
(7) Metastatic Tissue |
Breast |
201,500 |
41,844 |
0.2 gm Small |
Community Hospitals |
Challenging with fewer mastectomies |
Yesfrom lymph node dissection |
Lung and Bronchus |
188,900 |
156,005 |
1-2 gm Medium |
Large Cancer Centers |
Alveolaryes Bronchus limited |
Unlikely |
Prostate |
195,400 |
30,714 |
0.2 gm Small |
Widely available |
Yes – but may have BPH |
Unlikely |
Colorectal |
155,300 |
56,799 |
1-3 gm Med/Large |
Community hospitals |
Yes |
Yes |
Non-Hodgkin’s Lymphoma |
50,900 |
22,340 |
0.5 gm Small |
Limited sources |
Usually Not |
Unlikely |
Kidney |
32,300 |
12,084 |
5 gm Large |
Large Cancer Centers |
Yes |
Unlikely |
Liver and Intrahepatic bile duct |
13,000 |
13,263 |
1-2 g Med/Large |
Large Cancer Centers |
Yes |
Unlikely |
Ovarian |
24,000 |
14,361 |
5 gm Large |
Widely available |
Omentum Only |
Yes |
Cervical |
13,500 |
4,064 |
0.5 gm Small |
Widely available |
Yes |
Limited |
Pancreas |
29,900 |
29,723 |
In resections this will be large |
Large Cancer Centers, ~3,600 resection/yr U.S. |
Yesfrom resections |
Very limited |
Bladder |
60,600 |
12,115 |
0.5 gm Small |
Large Cancer Centers |
Limited from cystectomy |
Yes |
Esophagus |
14,000 |
12,509 |
0.2 gm Small |
Large Cancer Centers |
Yesin small quantities with limited quality |
Limitedneeded for staging |
Uterus |
36,800 |
14,361 |
0.5 gm Small |
Widely available |
YesWidely |
Yes |
Stomach |
21,200 |
12,340 |
1 gm Medium |
Widely available |
YesWidely |
Yes |
Oral cavity and Pharynx |
29,100 |
7,638 |
0.5 gm Small |
Widely available |
Yes |
Yes |
Sources: [Col 2] based on incidence rates from Surveillance, Epidemiology, and End Results (SEER) registries and population figures in U.S. Cancer Statistics Working Group, 2002. [Col 3] Arias E. and Smith B., 2003, p. 15. [Cols. 4-7] Personal communications, W. Grizzle and S. Hewitt, July 2003. |
The anticipated utilization of the specimens will determine how they need to be processed and
what associated data should be collected. For each donor with a specific cancer, the NBN would
collect high-quality specimens, sometimes with adjacent non-neoplastic tissue, and qualitycontrolled
tumor RNA and DNA. These specimens would ideally be paired with serum, plasma,
buffy coat, DNA, and urine samples. Quality-controlled clinical and pathology annotation,
treatment, response-to-treatment, and outcomes data would be linked to the specimens.
It was suggested that a standardized kit be developed and utilized, with associated protocols for
tissue collection procedures, equipment, and supplies. Similar kits, with established protocols
and a procedure manual, are being used by the Biopathology Center, Children’s Research
Institute, in Columbus, Ohio, and at other institutions. This center provides centralized
histopathology and tissue bank services for the pediatric division of the CHTN program, the
Childhood Cancer Survivor Study, the Children’s Oncology Group (COG), and the Gynecologic
Oncology Group. Although the NBN can learn from best practices of child cancer groups, the
Design Team suggested that NBN focus on tumors from adult patients. Pediatric tumors are
already well covered by programs operated by the COG.
Pathologists are responsible for the diagnostic evaluation of patient surgical specimens and must
first take what is needed for clinical (diagnostic) purposes. If enough tissue remains for donation
to the NBN, the pathologist will evaluate representative sections of the samples designated for
the repository for QC purposes. Excess tissue would be embargoed in the event that it is needed
for additional diagnosis and patients have the right to revoke consent for unused samples.
Potential challenges in sample collection include the increase in preoperative neoadjuvant
therapy, the increasing use of biopsy techniques that yield small specimens, and the impact of
surgical clamping and the resulting anoxia on a specimen’s suitability for advanced analytical
techniques, particularly RNA analysis.
Tissue specimens would include snap frozen tissue and other necessary samples (including but
not limited to matching adjacent tissues of grossly normal appearance, serum, plasma, buffy
coat, and urine), with appropriate tumor representation and accurate depiction of patient
background (e.g., age, ethnicity, race, gender, socioeconomic status, and geographical location).
A portion of the specimens would be fixed in an appropriate reagent to allow RNA expression
analysis. Other specimens would be snap frozen, thus preserving the ability to analyze specimens
at the proteomic level. Tissue microarrays could be prepared using selected specimens. A model
for this could be the Tissue Array Research Program within the intramural program of the NCI,
which provides high-density tissue arrays for researchers throughout the country. Other forms of
processing might be available. The goal should be that tissue is processed (annotated and frozen)
within 15 to 30 minutes of removal from the patient. Tissue collection techniques and protocols
would be validated in the demonstration project phase (see 8. Demonstration Project).
3.3.4.1 Quality Control Considerations
Human tissues used in research must be of as high a quality as possible, with sufficient
annotation to be able to determine appropriate uses. Initial steps should confirm the tissue type,
check to see that the tumor is present and the percent cellularity, and look for signs of degradation. QC should
also, at a minimum, consist of preparation of a stained slide obtained
from tissue adjacent to a section. This would be analyzed at the central site even if storage were
maintained locally.
It is suggested that the NBN utilize an industry-accepted bioanalyzer to evaluate all specimens
coming from the OR to determine the quality of RNA and protein. Specimens should be
prioritizied based on the results of this testing. It should be possible to collect a fair amount of
chip data using this technique, and provide a printout of RNA characteristics with each
specimen.
An independent, offsite pathology review should also be a part of any QC procedures. QC
procedures should confirm that involved tissue has the correct diagnosis, and that uninvolved
tissue is actually free of the tumor or disease process. Samples should be validated histologically,
as well as any molecular derivatives produced, prior to distribution. The successful production of
high-quality tissue microarrays would provide one indication of the physical quality of the
specimens. The ISBER Best Practices document provides an informative description of QC
requirements (see Appendix K).
Ultimate responsibility for QA and QC activities would rest with the appropriate NBN entity, in
coordination with the collection site (see also 6. Governance and Business Models).
3.3.5 Annotation and Clinical Data
To accelerate drug and cancer therapy development, more complete datasets are needed.
Requirements for data must be defined with input from researchers, as researcher needs should
dictate the types of data that are most valuable to collect. Also, the magnitude of both the
currently available and projected data, and researchers’ ease of access, should drive the design of
the NBN database; the needs of users for data should be anticipated and well served by the
selected bioinformatics platform. Ideally, information to be tied to tissue samples via identifiers
should include the following:
- Demographic data and social history (e.g., smoking, alcohol use), including familial
cancer history
- Diagnostic and clinical information
- Pathology reports
- Initial staging procedures
- Tissue collection procedures
- Treatment data
- Information that could track the patient in the future for clinical outcomes.
Investigators need to know the pathologic diagnosis of the tumor and characteristics of
associated tissues. Clinical annotation of specimens should ideally include site of origin (e.g.,
lung), primary diagnosis (e.g., adenocarcinoma), secondary descriptors (e.g., poorly
differentiated), tumor size and stage, a copy of the blinded surgical pathology report, and any
additional diagnostic studies. For every case (multiple specimens), a base set of patient
demographic and clinical data should be collected. The demographic data would ideally include
age, race, ethnicity, sex, state of birth, state of current domicile, and urban/rural classification.
Social history should ideally include detailed tobacco usage, detailed alcohol usage, and potential
toxic (occupational, previous chemotherapy or radiation therapy) exposures. Clinical history also
should include (if possible) familial histories of cancer, with detailed, more specific histories
focused on each tumor type. Data collection should be standardized and QC enforced across
biospecimen collection sites. It is recognized that such an extensive data collection will be costly.
Except for outcome data available from tumor registries and use of questionnaires, treatment and
outcome data are extremely difficult to obtain and might not be available, even though these
would be very valuable for research. New policies and protocols would be needed to facilitate the
submission of these data, ensure uniformity, ensure patient privacy, and track treatment and
outcomes. Tracking recurrence is difficult because patients often do not return to the same
hospital for treatment of the second tumor incidence. Standardization, QC approaches, and
monitoring would be required at all stages of the process. Additionally, the database system
should be designed to grow and evolve as technologies advance and the information base grows.
The system should also be capable of collecting longitudinal data, and the methods to be used to
collect longitudinal data must be carefully planned. This is further discussed in 4.
Bioinformatics and Data Management.
Various groups that might offer models for best practices in annotation include the following:
- American Joint Committee on Cancer (AJCC) or American Cancer Society Tumor
Registries and State registries
- Clinical Cooperative Group banks
- Children’s Oncology Group (COG)
- College of American Pathologists
- Familial cancer groups/Cancer Family Registries program
- Shared Pathology Information Network
- Military or Veterans Administration (VA) systems
- Health Maintenance Organizations
- Single Nucleotide Polymorphism (SNP) Consortium
- Early Detection Research Network (EDRN)
Coding standards, databases, and communications protocolskey to making these data useful
are discussed in 4. Bioinformatics and Data Management.
3.3.6 Advanced Analysis
It may be reasonable and desirable for the NBN to provide advanced analysis services to
researchers. To respond to researchers’ demands for the collection and cataloging of baseline
genomic data, it will be necessary to define the data that are in highest demand, how they will be
accessed, and at what price. Possible advanced analysis services might include proteomic and
genomic-based services (SNP profiling, mRNA profiling, proteomic profiling) and advanced
tissue approaches (laser capture microdissection, tumor cell lines, and advance biospecimen
production). A discussion of baseline genomic and proteomic analysis techniques likely to be
employed by researchers is found in Appendix M.
Collection of genomic data, while highly desirable to researchers, would dramatically increase
cost and management issues. The informal questionnaire administered by the NDC Research
Team and the NCI at the AACR in July of 2003 indicated that approximately 70 percent of
respondents expressed an interest in obtaining gene expression by DNA microarray (mRNA
transcript expression profiles of about 30,000 genes) along with specimens; however, the fees
respondents were willing to pay for these studies were insufficient to cover even 50 percent of
the costs for such data (see Appendix E for complete results from the survey). The Design Team
recommended that a professional survey of researchers be commissioned to further expand the
findings of the AACR survey, with particular emphasis on the priority of researcher needs for
advanced analyses and cost sensitivities.
3.3.7 Longitudinal Data
The usefulness of the biospecimen resource would be fully realized only after high-quality
genomic and proteomic data are linked with clinical outcomes, events that may occur months to
years after collection of the samples.
Longitudinal data useful for research should include specific treatment, time to progression, and
survival outcome. Relevant biomarkers should be used, if available. Suggested data elements
include primary chemotherapy (including chemotherapeutic agents); primary radiation therapy;
primary surgical therapy; secondary therapy; concomitant drugs; narcotic pain medication usage;
and date of death. Chemotherapy and radiation therapy would ideally specify doses, cycles
delivered, dose modification, response, toxicity, reason for termination of therapy, and time to
progression of disease. The NDC-NCI questionnaire results (Appendix E) suggest that
researchers want all of these data elements.
Longitudinal data collection holds enormous promise for scientific discovery but is very resource
intensive. The substantial costs involved in realistically collecting the most potentially useful
longitudinal data must be anticipated. Statistical expertise for incorporating such diverse
followup data into interpretation of laboratory results also will be needed.
There are several ongoing potential sources of longitudinal data that can offer opportunities and
insights for the NBN, including the VA medical centers, clinical cooperative group banks (of
which the COG is one), the National Cancer Database (NCDB), selected academic centers, and
the Woman’s Health Initiative. The COG is considered an excellent example of a successful
model system, featuring a uniform treatment protocol, collection of biospecimens, and tracking
for outcomes. Although the NCDB, run by the American College of Surgeons CoC, collects
longitudinal data, it does not have a legal mandate to enforce cooperation, and its ability to share
the data once it is collected is not clear. A partnership between the NCDB and NBN could
provide the NBN with outcomes data linked to tissue sample data, particularly if a legal mandate
could be established to facilitate sharing.
Collecting longitudinal data presents a variety of challenges. Heterogeneity of current therapy,
quality and nature of followup data, and changing therapies with availability of new agents have
an impact on the utility of followup data. A periodic link must be sustained from the patient to
the tissue sample, but patient privacy must be maintained. Experience has shown that collecting
longitudinal data requires dedicated onsite data managers and a flexible information technology
(IT) structure. Using existing resources within hospitals and clinicsfor example, clinical trial
data managers or state tumor registry personnelmay be possible. The Design Team
recommends that an onsite data manager for each collection site be designated to maximize the
collection of longitudinal and outcome data, as well as ensure the accuracy and validity of
clinical annotation.
Epidemiologists have shown that it is possible to collect longitudinal data over a period of years
with minimal loss to followup.7 Individuals who
are part of a study and know that they are part
of a study do not mind being recontacted. The response rate tends to be high, and the loss to
followup is low. The combination of active contact, change-of-address systems (operated by the
Postal Service), checks on cancer registries, and the National Death Index means that only a
small percentage is lost to followup. Sometimes individuals fail to respond at one contact attempt
but resurface on the next contact. Additionally, acceptable loss to followup rates can be defined a
priori. The NBN should build on current epidemiological techniques and expertise to maximize
its success in collecting this valuable information. Additionally, a well-considered marketing
campaign, coupled with a privacy-protected Web site, could encourage patients to provide the
NBN with updated address information.
Some specific treatment, response, and time-to-progression information (or time between
treatments, as a surrogate) might be obtained from chart reviews, although additional
investigation may be necessary to determine more effective ways to obtain this type of
information. Survival data can be obtained from SEER or high-quality hospital or state tumor
registries. Linking the NBN database to existing state tumor registry data has been considered,
but may require solving additional privacy and IT challenges. Hospital and some state tumor
registries can provide accurate data for patients who remain in the local area, and also provide
information on who and what percentage of patients is lost to followup at that location.
Because state and hospital registries vary in their followup success, and mobility rates of
patients differ in various locales, the average stay in a medical system can be determined and
the failure rate in longitudinal collection can be built into the NBN system expectations.
Why Are Longitudinal Data so Important? Example: Colorectal Cancer
The current standard-of-care for patients with Dukes B stage colorectal cancer is surgery
(alone). However, 20 percent of patients so classified go on to have recurrences, and may
have benefited from adjuvant therapy; unfortunately, there is no currently accepted
method for identifying that 20 percent. Longitudinal clinical outcome data associated with
biospecimens from patients with colorectal cancer would be invaluable for studying the
correlation between gene expression patterns and long-term outcomes. These data could
lead to development of a test to more accurately stage these cancers, and thus to make
more informed recommendations for treatment (surgery alone or surgery plus adjuvant
therapy).
|
3.3.8 Storage
A decentralized network of tissue repositories would require solutions to logistical issues
such as inventory control and ease of access. The design of the physical plants of the network
would require highly standardized procedures and practices. These would include power
outage/electrical backup, operation of liquid nitrogen freezers (including procedures to buy,
store, and use liquid nitrogen; ample backup; and safety equipment), and shelving and
materials for room-temperature storage. Issues such as heating/air conditioning and security
systems, physical layouts, and geographical locations (e.g., seismic areas and local
transportation centers) also should be considered.
Storing all frozen specimens in the vapor phase of liquid nitrogen freezers will ensure that physical
and chemical reactions within tissue are slowed so that the specimens remain stable for use over
many years, and biomarkers of interest to researchers preserved. If cryopreservation for viable cells
is required, biospecimens should undergo controlled rate freezing to prevent damage by ice
crystallization.
A high-quality inventory system using bar coding and an inventory database should be employed
so that sample location can be tracked. Tracking and life cycle management of biospecimens is
critical. Written procedures for all facets of facility management and security will be required.
The use of secondary storage sites is complex. A collection site would remove specimens from
temporary storage, package the specimens, and ship them to a central site. Central sites would
unpack the specimens, check identification of each specimen, store them according to
identification, and wait for usage. Some specimens should remain at the collection site as a
redundant measure to avoid failures in storage. All participating tissue collection sites that
possess the expertise to collect specimens according to NBN requirements would need to be
trained in NBN storage methods. Storage methods would be validated in the demonstration
project phase (see 8. Demonstration Project).
3.3.9 Distribution and Access
A clear research need is timely and equitable access to biospecimens and associated data without
undue administrative burden, as well as a prescribed mechanism for rapid turnaround of requests.
Research would have to be supported through a uniform candidate biospecimen identification
and distribution system, rather than one that requires the NBN to negotiate with a series of
institutions. Distribution of scarce tissue resources should be prioritized through an evaluation of
needs at the national level, and a peer review process should determine distribution.
To facilitate this, the NBN will provide access to tissue specimens through a Biospecimen
Utilization Review Committee. Members of this committee would be recruited from the research
community, advocacy groups, industry, and possibly government. The NBN Biospecimen
Utilization Review Committee would utilize a peer review system to set priorities as to how the
collected tissue should be allocated, and to guarantee fair and equitable access. Evaluation of
researchers’ needs and capabilities against competing demands for specimens would be required
to assure maximal benefit of the NBN. The importance and quality of the laboratory studies
should be commensurate with the value of the specimen, its associated annotated database, and
followup data. As part of the peer review process, the Research Administration and Support
Business Unit might consider factors such as the applicant’s willingness to redeposit data and
past collaboration with the NBN.8
Most investigators would probably use fewer than 100 specimens per year of one particular type
of tissue. Some entities might use up to 1,000 samples per year (for example, a large
pharmaceutical company or large academic medical center conducting many concurrent studies).
All studies should justify the number of specimens required for the specific proposed study.
Also, the Biospecimen Utilization Review Committee should determine if the tissue needed
might be obtained from other sources (e.g., a prospective collection for which outcome data are
not available, if the designated study does not require outcome data).
It is important to clarify that the roles of the NBN Tissue Utilization Committee and
IRBs/industrial review boards are distinct. The IRBs provide assurances that human subject
regulations and policies are being followed. The NBN Tissue Utilization Committee would look
at the quality of the proposed research and determine its priority for access to specimens.
Deidentified data associated with the biospecimens would be made available to industry,
academia, and government users through a transparent IT platform that is Web based, searchable,
secure, and workable across virtually all systems (see 4. Bioinformatics and Data Management.)
3.3.9.1 Shipping
Designated carriers who are reliable and who routinely ship human tissue samples would be used
for all shipments from the NBN repository to the research users. Packages would be bar coded
and could be tracked via software provided by the carriers, and the carriers would have
acceptable procedures for resolution of shipment problems. The carriers would ensure that
paperwork accompanying shipments would be standardized and compliant with regulations.
Standard operating procedures would be developed for shipping and would be provided to
members of the network.
Facilities applying to be part of the NBN would have to comply with regulations for storage and
shipment of human tissue to domestic and overseas destinations. The International Air Transport
Association (IATA) regulates overseas shipments, and the U.S. Department of Transportation,
which regulates domestic shipments, has now adopted the IATA standards.9
3.3.10 Safety
Biohazards and other safety issues must be considered by personnel who collect tissues, by
investigators who receive tissues, and by laboratorians who work with tissues. The NBN must
ensure that persons who come into contact with specimens procured by its collection facilities
are trained properly in how to handle potentially hazardous materials. In particular, all specimens
must be handled as if infectious. It is an Occupational Safety and Health Administration (OSHA)
requirement that organizations that employ persons handling human tissues establish a written
safety program and a comprehensive training program to protect personnel from blood-borne
pathogens.
To be an approved site within the NBN, a facility should demonstrate its compliance with the
OSHA regulationse.g., regulations for handling blood-borne pathogens (29 CFR Part 1910)
and state and local biosafety requirements. As part of this compliance, safety and emergency
procedures, including a facility safety plan with standard operating procedures and standardized
personnel training, would be developed.10 This
responsibility would probably lie at the Business Unit level, with a corresponding QA matrixed responsibility at the
NBN Operations Center level. Fulfillment would include site visits, the distribution of written and online
materials, and training.
Organizations such as CHTN require investigators to sign an agreement that they will educate
their staff about the proper handling of biohazardous materials, and sign an agreement that
indemnifies the tissue procurement and distribution facility from any claims, costs, or damages
resulting from the use of tissues provided by that facility. It is recommended that the NBN
consider similar agreements with persons who will receive specimens.
3.3.11 Submission of Research Data
Creating a framework for sharing and comparing research results in a timely fashion would add
substantial value to the NBN. Therefore, it is recommended that validated, investigator-derived
data be returned to the NBN and linked back to original NBN tissue samples. Encouraging the
entry of validated data in a standardized format (e.g., outcomes, possibly research results) back
into the NBN system would create a rich, valuable, and unique resource available to all
investigators, providing new possibilities for the corroboration of research findings across
methodologies; the emergence of successful new therapeutic/preventive targets; and the
acceleration of important scientific breakthroughs. The research database will also be useful for
selection of related specimens remaining in the bank. The data would be able to be used
repeatedly, unlike the actual tissues, and would serve as a stimulus for participation in the NBN,
further enhancing the benefits for the scientific community.
The tissue resource and associated data would be made maximally useful if investigators were
required to submit their experimental data to the database after a set period. The Design Team
recommended that a multitier policy on submission of research data be considered, as wholescale
submission of data may not be acceptable for some researchers. However, multiple levels
of acceptable validation exist among industry and academia, which may limit the usefulness of
the database to some groups of customers. Therefore, researcher-submitted data, unless validated
to the standards of all constituencies, may suffer from lack of utilization. Standards to ensure as
much data validity as possible will be necessary. When such data should be submitted, and the
appropriate level of data to be submitted (e.g., broad profiling genomic data versus focused
pathway mapping studies) by various types of researchers (academic or industry) with sufficient
validity standards should be determined by the Research Support Business Unit. The Design
Team offered the following guiding principles:
- As close to practicable after publication, researchers would be encouraged and invited to
submit data (especially DNA, RNA, and proteomic data) to the NBN.
- Researchers would be required to report all publications resulting from the use of NBN
samples, as well as reference the source of the samples in their report (as is required by
CHTN and other resource sharing facilities).
- If DNA, RNA, and proteomic data are published, the raw data must be available for
public use, similar to publication standards set by major scientific journals.
Creation of such a data resource, while of great potential value, certainly presents many technical
and policy implementation challenges. It will be necessary to create data standards to allow
resubmission in a usable format into a centralized database. Fortunately, data and nomenclature
standards are currently being created for many types of research results (e.g., Systemized
Nomenclature of Medicine and AJCC for cancer staging; Minimum Information About a
Microarray Experiment for gene expression data; and the common data elements created by the
EDRN), and these standards should also be used by an information resource such as that
suggested by the NBN. It will be difficult for the NBN to revalidate all submitted results, so
steps must be taken to ensure submission of data that other researchers will find valuable and
want to use. Researchers are often reluctant to submit data for various reasons, so incentives to
encourage this submission may be necessary. Finally, it will take time for the NBN to address
these challenges; consequently, the creation of a usable database with high value for researchers
likely will occur later, rather than sooner. Specific recommendations about how this might be
accomplished may be found in 6. Governance and Business Models.
3.3.12 Incentives for Participation in the NBN
To implement the NBN, potential sites may need incentives to overcome possible institutional
barriers. For example, they will need to adhere to strict policies, procedures, and standards;
invest in new infrastructure and support (staff, equipment, training, etc.); and revisit the role of
IRBs in the process. Incentives could be developed to encourage surgeons and pathologists to
cooperate in increasing samples within the repository for research. For example, increased
deposits of tissues could be rewarded with increased or priority access to tissue. Participating in
the NBN, and thus having met certain quality standards, might make an institution’s tissues more
valuable for other research purposes. NBN participationand the associated increased
resourcesmight help existing repositories (e.g., SPOREs, clinical cooperative group banks, and
cancer centers) to improve their capacity for their entire range of activities.
Another incentive would be provided by access to the NBN database that would be developed as
part of this resource. Academic medical centers would want access for their researchers to use
this new valuable resource, which would not otherwise be available. The data could be used
repeatedly, unlike the actual tissue; thus, the data could serve as a stimulus for long-term
participation.
Different incentives would need to be developed for community hospitals because their needs
differ from those of academic centers. Absent resource constraints, community hospital
pathologists are generally willing to participate for the satisfaction of contributing to the research
enterprise. Providing a community hospital with funds to hire employees, such as technologists,
would be likely to encourage their participation. Community hospitals and academic centers
could be paired to share various resources. Alternatively, these hospitals could be tied in with
clinical trial groups to involve them in clinical research.11 Cost and reimbursements as incentives are discussed in 6. Governance and Business Models.
3.4 Summary of Key Findings and Recommendations
The NBN would be distinguished from existing resources for tumor tissue and other specimens
by highly standardized procedures for collection, processing, storage, annotation, and
distribution. The NBN would be developed to provide biospecimens and clinical information in
compliance with Federal, state, and local regulations. In particular, the following should be
noted:
- The NBN should be organized as (a) a decentralized network of collection facilities with
regional storage, possibly of nonprofit, tissue-repository organizations located near
academic medical centers and community-based hospitals that serve large and diverse
patient populations; and (b) a virtual data repository networked across the nation.
- Best practices should be incorporated and/or developed for every aspect of biospecimen
and data collection, processing, storage, and distribution in the operation of the tissue
repository; and should be consistently applied through the use of standard operating
procedures that would be monitored.
- Biospecimens and data should be collected from sources meeting NBN criteria, while
applying standardized clinical annotation.
- Incentives tailored to each kind of source would be developed to encourage many entities
to participate. Although community hospitals are the primary point of diagnosis for
roughly 80 percent of cancers and could thus provide the needed volume, community
hospitals would need specific incentives and assistance to develop the experience,
infrastructure, and understanding of research necessary to establish viable collection
centers.
- Well-trained personnel would collect tissue using NBN-provided protocols, applying
standardized clinical annotation, with selected biospecimens undergoing advanced
analysis.
- The NBN must ensure that persons who come into contact with specimens procured by
its collection facilities are trained properly in how to handle potentially hazardous
materials.
- Specimens from all cancer types should be collected (with matched normal specimens,
whenever possible), but the NBN should be structured to provide the quantity and
diversity of biospecimens required to meet researcher needs.
- Both fresh frozen and formalin fixed/paraffin embedded preparations should be used.
- A minimal dataset would be established for each specimen, with collection of additional
longitudinal data for a high percentage of specimens, and provision of genomic and
proteomic-based data.
- A professional survey of researchers should be commissioned to further explore
researcher needs for advanced analyses and cost sensitivities.
- Distribution of specimens would be guided by a Biospecimen Utilization Review
Committee peer review process that would evaluate researchers’ needs against competing
demands for specimens.
- It should be expected that validated, investigator-derived data derived from NBN
resources be submitted to the NBN and linked back to original NBN tissue samples. An
expanded dataset, created by the return of this experimental data to the NBN, could then
be made available to all investigators.
|