Frozen tissue, blood samples, as well as other body fluids are commonly stored in biobanks. Because these samples are stored for research purposes, it is of the utmost importance that they are of excellent quality. In a publication from Clinical Biochemistry, Dr. Bih-Rong Wei and Mark Simpson from the Laboratory of Cancer Biology and Genetics at the National Cancer Institute have discussed biobank sample quality assurance (QA), and presented solutions to improve QA, including the use of digital technology.1
Improving sample QA really begins with harmonizing sample collection protocols and timely preservation to avoid degradation. While some standard operating protocols have been released from the National Cancer Institute Cancer Human Biobank, and there are still multiple variables and logistical demands within patient care that can complicate specimen handling.
To illustrate the need for improved quality assurance within biobanking, the authors report on a survey of tumor specimens stored in the Lilly tissue/fluid biobank.2 Surprisingly, the researchers conducting the survey found that an average of only 59% of specimens contained <65% tumor cellularity, while 17% had no evidence of tumors. This is especially problematic for research, as DNA/RNA/protein experiments require a sufficient quantity/percentage of target tissues.
There are no standardized procedures to rank the level of quality of a particular sample; however, it is possible to gain information about the specimen composition through pathologist review. The pathologist can section tissue samples for histopathology as a surrogate for tissue components. As the authors point out, the review process can be tedious and is subjective.
Digital pathology is now emerging as an efficient strategy to improve biobank QA. Using digital microscopy, whole-slide images (WSIs) are captured into digitized image data files. In Canada, high resolution digital WSIs have been incorporated for clinical use, but have not yet been approved in the US.
After further innovations, the authors report that WSIs have been utilized in image analyses by incorporating algorithms constructed for pattern recognition with color identification/quantification functions.3 Pattern recognition image analysis (PRIA) is unique computer assisted strategy to discriminate between different tissue types in individual WSIs. Within a sample, areas containing representative features of each type of tissue and cancer type present are selected from WSIs for training an algorithm. PRIA can also provide data depicting proportions of neoplastic cells and stromal elements present in biorepository specimens.
While the researchers say it is impossible to create single algorithm to analyze an entire biobank, the authors report that multiple algorithms used to focus on phenotypic ranges and few numbers of tissue features can achieve good diagnostic agreement. The algorithms were designed with mean training accuracies of >98% and with >95% sensitivity and specificity. They also determined the median coefficient of variations (CV) among replicate portions of individual patient tumors to be 2%, 12%, and 33% for lymphomas, melanomas, and osteosarcomas, respectively. This indicates that PRIA can reasonably predict tissue components for some cancer types.
Digital pathology can minimize variability and subjectivity associated with routine pathologic evaluations and ensure that the target cells are present and in sufficient amounts. It is important to note that PRIA was never intended to replace the role of pathologist. The authors maintain that pathologists should continue to be involved in evaluating images and overseeing other aspects of QA within biobanking.
References
1. Wei, B.R., & Simpson, R.M. (2014) “Digital pathology and image analysis augment biospecimen annotation and biobank quality assurance harmonization,” Clinical Biochemistry, 47(4-5) (pp. 274–9), doi: 10.1016/j.clinbiochem.2013.12.008.
2. Sandusky G., Dumaual C., & Cheng L. (2009) “Reviewpaper: human tissues for discovery biomarkerpharmaceutical research: the experience of the Indiana University Simon CancerCenter—Lilly Research Labs Tissue/Fluid BioBank,” Veterinary Pathology, 46(1) (pp. 2–9).
3. Webste, J.D., et al. (2011) “Quantifying histological features of cancer biospecimens for biobanking quality assurance using automated morphometric pattern recognition image analysis algorithms. Journal of Biomolecular Techniques, 22(3), (pp. 108–18).
Post Author: Emily Humphreys. Emily has previous research experience in eye development, infectious diseases, and aging. While she enjoyed the thrill of research, She has since traded bench work for science journalism. Emily has been a regular contributor to Accelerating Science since 2012.
Leave a Reply