The multidisciplinary meeting had twenty minutes for the case. A consultant asked whether the mother had been tested. Nobody in the room knew. Whether copy number calling had been run on the exome. Nobody knew. Whether the candidate gene had been covered adequately. Nobody knew. Whether the absent corpus callosum in the differential had been looked for on the imaging, or the imaging had simply not reported on it. Nobody knew.
The twenty minutes were spent establishing what had not been established. The decision, a repeat exome, was made on the guess that the first one had missed something, with no idea what. Six months and a second negative report later, the same questions were asked again, by the same people, with the same result.
What the field knows
A negative genomic result is a statement about what was examined, and the laboratory standards say so. The ACMG standards for next-generation sequencing (Rehm et al., 2013) require that a report describe the regions covered, the variant classes detected and the limitations of the assay, because a negative result over a poorly covered gene is no evidence about that gene. In practice the limitations paragraph is boilerplate, and the specific gaps in this patient's data are not in it. The reader learns that exomes in general miss things, and nothing about what this one missed.
Diagnostic uncertainty has been studied as a thing to measure and communicate. Bhise and colleagues (2018) reviewed how uncertainty in diagnosis is defined and found that it is rarely recorded in a structured form, so that what a clinician did not know at a decision point is reconstructed afterwards, if at all, from narrative notes. Medicine's experience with checklists (Haynes et al., 2009) rests on the same observation: the items a team forgets are not the hard ones, they are the ones nobody was responsible for stating. A gap that belongs to nobody is restated at every meeting and closed at none.
In reanalysis, the gaps are where the diagnoses are. Ewans and colleagues (2018) and others found that a share of the diagnoses made at reanalysis came from variant classes or regions the original analysis had not assessed, which is to say from known limitations that nobody had listed against the case and so nobody had revisited. The unknown was knowable. It was unrecorded, and so it was never scheduled.
Where practice breaks
The case record holds findings and loses gaps. A phenotype list holds the terms present; whether each absent feature was examined and absent, or never examined, is in the clinic letter if anywhere. A variant list holds what was called; which regions fell below coverage, which classes the pipeline could not detect, and which relatives were never sampled are in a technical report nobody reads twice.
The gaps are reconstructed at every meeting. Each discussion begins by re-establishing what is unknown, from memory and from documents, and the reconstruction is incomplete and different each time. The time that should go to the decision goes to the inventory, and the inventory is thrown away at the end of the meeting and rebuilt at the next.
The next step is chosen on a guess. A repeat exome is ordered when a copy number analysis of the existing data would have answered the question; a trio is requested when the father is unavailable and the question could be asked of the mother and a sibling; a biopsy is scheduled when the finding it would confirm was never looked for on the imaging that already exists. Each is an expensive answer to a question that was never stated, and each takes months, during which the family waits for a result that could not have changed the picture.
The report says negative. It does not say negative over these regions, for these variant classes, with these relatives untested and these features unassessed. A clinician reading it in three years takes it as a stronger statement than it was, and may stop looking on the strength of it.
How Tulparus handles it
Every case carries a list of what it cannot see, built from what was done. The list is derived, not typed: the pipeline knows which regions were below coverage threshold, which variant classes it called and which it did not, and which analyses have run with which versions; the pedigree knows which relatives have samples; the phenotype knows which features were recorded present, which examined and absent, and which never assessed. The list is a first-class part of the investigation, beside the candidates, and it updates as the case does. When a maternal sample arrives, the gap closes itself; when a new analysis runs, the regions it could not see are added.
The next step is proposed against the list. For each open gap the workbench shows what it would take to close it and what it would change if closed: a copy number analysis of existing data, a maternal sample, a question to the radiologist about a feature the report did not mention. The cheapest step that closes the biggest gap sits at the top, and the expensive steps are shown with the gaps they would and would not close. A repeat exome that would close none of the open gaps is shown as closing none.
The report renders the gaps. A negative report states what was examined and what was not, specifically for this patient: the genes of interest with coverage below threshold, the classes not assessed, the relatives not tested, the phenotype features not evaluated. The statement is generated from the case record, so it is accurate on the day and it is the same list the meeting sees.
For the meeting above, the case would have opened on the list: mother not sampled, copy number not run, candidate gene with two exons below threshold, corpus callosum not reported on. The twenty minutes would have gone to choosing among four stated options, and the repeat exome would not have been one of them.
What you can do tomorrow
The list can be kept by hand for every open case, and keeping it changes what the meeting does.
- Add a section to every case summary headed what we have not looked at, with four lines: relatives not tested, analyses not run, regions or classes not covered, features not assessed. Fill it before the meeting.
- Read the coverage report for the candidate genes before ordering any repeat test. The answer is often that the gene was never adequately seen.
- Phrase every negative report with its specific limits for this patient, in a sentence a clinician will read, above the standard paragraph.
- Open the multidisciplinary discussion with the gap list and spend the time on which gap to close.
Sources
- Rehm HL, Bale SJ, Bayrak-Toydemir P, et al. ACMG clinical laboratory standards for next-generation sequencing. Genetics in Medicine, 2013.
- Bhise V, Rajan SS, Sittig DF, Morgan RO, Chaudhary P, Singh H. Defining and measuring diagnostic uncertainty in medicine: a systematic review. Journal of General Internal Medicine, 2018.
- Haynes AB, Weiser TG, Berry WR, et al. A surgical safety checklist to reduce morbidity and mortality in a global population. New England Journal of Medicine, 2009.
- Ewans LJ, Schofield D, Shrestha R, et al. Whole-exome sequencing reanalysis at 12 months boosts diagnosis and is cost-effective when applied early in Mendelian disorders. Genetics in Medicine, 2018.
- Richards S, Aziz N, Bale S, et al. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genetics in Medicine, 2015.