Response to Lindor et al.

To the Editor: We greatly appreciate the comments of Dr Lindor and colleagues in the letter titled “Preserving Personal Autonomy in a Genomic Testing Era”¹ regarding our approach to managing incidental findings from genome-scale sequencing.² We wholeheartedly agree that a consensus-based approach for determining how to handle the broad spectrum of incidental findings is unlikely to satisfy all constituents in the long term. Likewise, we are in complete agreement that patient preferences should play a central (although not exclusive) role in determining the return of results. Indeed, our “binning” approach attempts to balance the ethical responsibilities of the clinician (such as the duty to warn) with the autonomy of the patients to determine what information they want to know and what information they prefer not to know.

There are certainly many valid approaches to dividing the genome into categories that can be used to manage the return of incidental findings, but we strongly believe that some measure of clinical actionability will be a critical parameter in any successful strategy. In our approach, Bin 1 can be considered the category of incidental information in which the degree of clinical actionability invokes a duty to warn that supersedes patient preferences. Bin 2 contains the bulk of incidental information with limited clinical actionability that some patients may desire to know, whereas others may not, which is the very definition of individual informed decision making. Of course, there will be differing opinions about where to draw the line between Bin 1 and Bin 2, which is essentially the crux of the problem with consensus-based approaches to “binning” the genome. Instead, as pointed out by Lindor and colleagues,¹ there is a continuum of actionability.

We are therefore intrigued to hear about the efforts at the Mayo Center for Individualized Medicine to develop the Tailored Result Selection Tool with a scoring system for “actionability,” and we were gratified to see that our provisional bin assignments correlated reasonably well with the Mayo group’s actionability scores. Our group has come to the very same conclusion that a semiquantitative measure is required to score the clinical actionability of gene–phenotype pairs in order to categorize them in a transparent and evidence-based fashion. We have focused on four key components of clinical actionability: (i) the severity of the threat to health for an undiagnosed individual carrying an incidentally identified deleterious allele; (ii) the likelihood that a serious threat will materialize, akin to penetrance; (iii) the effectiveness of interventions at preventing harm from occurring; and (iv) the acceptability in terms of the burdens or risks placed on the individual. These components of actionability have also been adopted as part of an evidence-based framework being developed by the Evaluation of Genomic Applications in Practice and Prevention working group³.

Our local “binning” committee is now systematically scoring gene–phenotype pairs much in the same way as described by Lindor and colleagues¹. In the process, we have revised the bin assignments of some genes. We ultimately plan to use the semiquantitative actionability scores to set a threshold for Bin 1, and we anticipate exploring different weighting systems for the key components of clinical actionability and/or thresholds to define Bin 1 in different clinical contexts. Thus, one could set a very high threshold such that Bin 1 contains very few genes, leaving more genes in Bin 2 for individualized decision making. One could also imagine using the continuum of actionability scores to facilitate individual decision making regarding return of results. It will be fascinating to hear more about the Tailored Result Selection Tool system, and we very much look forward to the results and lessons learned.