Culling Decision Tools for Biobankers

Culling unused biospecimens and data from biobank inventories is a necessary and integral part of responding to the evolving needs of research. Culling is necessary in many circumstances. Culling can help with the sustainability of a biobank by curtailing the operating costs of expanding inventories or creating space within an existing storage footprint for new collections. Culling is also part of our responsibility to accomplish “green biobanking.”^1,2

Among the important barriers to undertaking culling has been acceptance of the need to consider this activity as an essential part of biobanking. This reticence is reflected in a relatively low profile given to the topic of culling within past and current Best Practices documents. ³ This has also meant that the topic has received little discussion up till now in international biobanking forums, and determining how to value a collection has also received only limited attention. However, this changed at the International Society for Biological and Environmental Repositories annual meeting in Montreal in 2025 with several presentations on the topic from groups based in Australia and Canada, France, and the USA. ⁴ These groups have also now recently published the first models for determining relative value of collections in “Biopreservation and Biobanking.”^5–7 A forth group from the Netherlands has also posted a national guideline and valuation model on the topic. ⁸ The development of these guidelines and models is an essential prerequisite for “keep” or “cull” decisions. ⁵

From the perspective of biobankers faced with culling decisions, these models provide for the first time a possible approach to adopt to undertake valuation of collections. This also provides a choice of model to adopt or improve upon. For those choosing, there are several reassuring similarities and also some useful differences to consider.

On inspection of the four valuation models,^5–8 all incorporate multiple factors for a full assessment of a collection that can be categorized as relating to aspects such as “Ethics and Legal” status, “Quality” of the biospecimens and data, “Prospects” for future use, “Scarcity/Rarity,” and “Age/Scale” of the collection. “Ethics and Legal” status relates to a range of features such as specifications of the informed consent, ability to share, custodianship status, and participant status. “Quality” of the biospecimens and data relates to features such as fitness for purpose, biospecimen storage features, extent of the annotating data and QC data availability, nature of the database, and indicators of good management. Some features of quality are more easily assessed because they are directly visible without any detailed analysis, such as the state of labeling and condition of data records, and are separated out into “Integrity” factors in one model. ⁵ “Prospects” for future use relate to past usage data, including numbers of users, publications based on the collection, importance for the institution, and relevance to anticipated areas of research. “Scarcity/Rarity” of the collection relates to feasibility and future cost to replace. In one model the “Age” and “Size” of the collection are also considered as distinct factors. ⁶

In terms of similarities in the approaches used, all four models^5–8 propose assessment across many aspects of a collection, and most incorporate a scoring system to summarize multiple factors. The common factors assessed within all models are those relating to the areas of “Ethics and Legal” status, “Quality” of the biospecimens and data, and “Prospects” for future use. “Quality” factors are also the main drivers of the overall scores used in assessments. Furthermore, a common approach used within all models is to initially consider those factors that might “eliminate” collections of very low value early in the assessment process.

However, there are also several differences between the models. In one model factors are not scored. ⁸ In the other models the approach and relative weight given to different factors varies in each model (but could easily be adjusted). For example, while “Quality” factors dominate in all models, these factors contribute to between 40% and 69% of the overall maximum scores. In one model, negative scores are possible for some Quality subfactors to more strongly discriminate against some collections. ⁶ As noted above, Quality factors are also assigned to two separate factors in two models and considered using different features in all the models.

“Ethics and Legal” factors are also handled differently across the models. These factors contribute to either 7% or 30% of the overall maximum scores in two of the models.^6,7 In the other models,^5,8 these factors are not scored, partly because in most jurisdictions, an Ethics Review Board will ultimately decide on the impact of these factors. Therefore, yes/no/maybe responses to multiple ELSI considerations are anticipated that can then be summarized. One model also considers the “Size” and “Age” of the collection as distinct factors that together contribute up to 10% of the overall maximum score. ⁶

Another difference in the overall approach in one model is in the identification of very high-quality collections, in addition to very low-value collections, early in the assessment process to avoid unnecessary effort involved in the full valuation of all collections. ⁵ This is achieved by an initial-valuation step that precedes moving onto a full extended valuation.

Perhaps most importantly, these models are based on distinct expertise and different perspectives. Therefore, each model may be a better fit for different circumstances and categories of biobank. ⁹ One was developed for academic institutional classic type human health research biobanks, ⁵ another was initially developed for valuation of microbial collections but then adapted for human collections, ⁶ the third for valuation of collections from multiple types of biobanks, ⁷ and the forth for medical centers and biomedical research institutions to evaluate their biobank collections. ⁸

In summary, the notion that culling is an important activity for a biobank is now becoming accepted. These valuation models may appeal to different biobanks and in different circumstances. Furthermore, all models can all be adapted as appropriate but can provide the basis for a relatively standardized approach to support “keep or cull” decisions.