2022-05-2 - Cancer Synoptic Reporting Project Group Meeting

Dear Scott,

thinking about yesterdays discussion, it might be wise to think of a genetic abnormality hierarchy:

Many of the acute leukemias are now defined by genetic change and these categories have a range of morphology according to the FAB classification.

In the bone marrow diseases you will find hairy cell leukemia wich has a subtype of BRAF V600E mutated hairy cell leukemia.

However BRAF V600E (and variants) mutation is the driver for many other neoplasms of different lineages (melanoma, papillary thyroid carcinoma, ameloblastoma).

The same ist true for may other mutations (KRAS, EGFR, BRACA) and in times of precision oncology these drive the therapeutic possibilities

The morphologic differences of mutated vs. unmutated tumors are not yet understood in most cases.

Therefore, it might be feasible to have an independent genetic hierarchy that can also be used to model disorders. (In addition in records it might be needed to use a string for so far unclassified mutations.)

Hairy cell leukemia is also a good example for a disorder now defined genetically. Cases that lack the BRAV V600E mutation are now separated out as splenic B-cell lymphoma/ leukemia, unclassifiable, a subgroup of which is variant hairy cell leukemia. Still the WHO-"FSN" remains hairy cell leukemia.

HI Thomas,

I think this is an interesting idea.  I will say that this may be possible in morphology abnormalities as it is a primitive hierarchy.  Unfortunately, it is a primitive hierarchy and subject to more complexity.   For example, leukemia with XYZ genetic signature would have one parent concept pointing to leukemia and another pointing to abnormal genetic morphology.  However, the abnormal genetic morphology hierarchy would be all encompassing.  Would it actually add value?  Not sure at this point in SNOMED model maturity.  With regards to representation of the specific genetic change, there is not a mechanism in SNOMED to model that, currently.  It is something we have been contemplating and have some modeling direction.

I do have wrestle with including the genomic variation/cytogenetic abnormality in the histology name.   This relegates the genomic data to a simple word string which is not computable nor really suitable for decision support at scale.