The article, "Betting on the Biobank", by Matthew Dublin, October 2011, makes some important points, but has some implications which could be misinterpreted.
When referring to the Cancer Genome Atlas project, the article makes the statement that "only a small fraction of the samples being submitted were of suitable quality for research." That inference is incorrect in that it implies that most biobank samples are of poor quality and cannot be used for any research. Actually, most such samples might very well meet the needs for which they were originally banked, and could be used successfully in many types of research. The problem with samples needed by the TCGA was that the requirements as to specific sample characteristics were demanding. For example, if an investigator requests 20 samples of triple-negative breast cancer, each more than 25 grams in weight and from Japanese-American males 15 to 20 years of age, our bank could not meet this request — not because of any issues of quality of tissues, but because the request is too restrictive. Breast cancers are infrequent and typically small in males, are extremely rare in young males, and are less common in Asian populations; other requirements of the request (e.g. molecular features) also limit available tissues. If an investigator adds one additional requirement as to the tissue characteristics of a request, the available tissues that meet the needs of this investigator are greatly reduced.
Similarly, requests for tissues made by the Cancer Genome Atlas project were difficult to meet, primarily because of requirements as to the specimen characteristics — not quality of the specimens. Specifically, TCGA wanted relatively large specimens. More problematic, the requests for tissues by the TCGA were for 60 percent to 80 percent tumor nuclei (cellularity) and less than 20 percent to 30 percent necrosis. Most tumors are intermixed with inflammatory and/or stromal cells and many tumors typically have necrosis. For example, the diagnosis of glioblastoma multiforme is based upon the presence of necrosis. Thus, requiring little necrosis could limit the availability of some specimens and the funds available from the TCGA do not cover all costs to meet TCGA requirements, especially not micro/macrodissection or laser capture microdissection, which would be necessary to enrich the proportion of tumor cells to greater than 70 percent in many types of tumors. Of note, the standards established for TCGA specimens of breast cancers were necessary to meet three methods of analysis on each specimen. It was recognized and accepted that the TCGA requirements would bias the final results because only larger mammary tumors would qualify for the TCGA, and only tumors without intermixed inflammation, extensive stroma or necrosis, and features that are characteristic of many mammary cancers, would meet TCGA standards.
Of note, the TCGA ultimately had to shift to prospective approaches to collection. In attempting to prospectively meet the current standards of the TCGA for lung cancers, our biorepository has found that less than 5 percent of lung cancers meet a diagnosis of either squamous cell carcinoma or adenocarcinoma (no other diagnoses are acceptable), as well as at least 60 percent tumor nuclei (tumor cellularity) and less than 30 percent necrosis. Because the same efforts have to be devoted to prospective collection of all lung tumors without initially knowing the diagnosis or histopathological characteristics of the lung mass, collection and processing costs far exceed potential payments from the TCGA for specimen collections because more than twenty lung tumors must be collected to find just one that meets TCGA standards. This is a severe problem for biorepositories unless the associated institution covers deficits resulting from the efforts in collecting tumors acceptable to the TCGA.
None of the above problems suggest "poor" quality and most relate only to very demanding requests. Nevertheless, this is not to say that some biorepositories did not contribute to the confusion as to the initial availability of specific tumors to support the research needs of the TCGA. Specifically, some biorepositories may not have performed quality control when specimens were initially collected. In such cases, these biorepositories would not know the individual histopathologic features of the tissues in their collection.
A biorepository's inability to meet a demanding request should not be confused with "suitable quality [of tissue] for research" of biospecimens in a biorepository. Perhaps "specimen characteristics to meet specific research needs" would be more appropriate terminology and cause less misinterpretation as to the "quality of biospecimens."
William Grizzle, Katherine Sexton
Tissue Collection and Banking Facility
University of Alabama at Birmingham
Bell WC, Sexton KC, Grizzle WE. (2009). How to Efficiently Obtain Human Tissues to Support Specific Biomedical Research Projects. Cancer Epidemiology, Biomarkers, and Prevention. 18(6):1676-9.
Grizzle, WE, Sexton KC, Bell WC. (2008). Quality Assurance in Tissue Resources Supporting Biomedical Research. Cell Preservation Technology. 6(2): 113-118.
Bell WC, Sexton KC, Grizzle WE. (2010). Organizational Issues in Providing High-Quality Human Tissues and Clinical Information for the Support of Biomedical Research. Methods in Molecular Biology, (Eds. Robert Grützmann and Christian Pilarsky). 576:1-30.