Stream: Cancer Interoperability
Topic: CS URI UICC/AJCC
Patrick Werner (Mar 12 2020 at 10:54):
We are currently searching for sources of the TNM CodeSystem. In mcode "http://cancerstaging.org" is used for this. To have this aligned internationally i want to propose to use a UICC based CodeSystem.
Patrick Werner (Mar 12 2020 at 10:54):
URI could be: http://www.uicc.org/tnm which forwards to: https://www.uicc.org/resources/tnm
Grahame Grieve (Mar 12 2020 at 11:18):
there is not one tnm code system. it's a framework but each cancer has it's own codes
Stefan Lang (Mar 12 2020 at 11:49):
@Grahame Grieve The viability of actually having separate code systems for each kind of cancer has been tested e.g. in the context of IHE APSR 1.0. Since this apparently was hard to maintain, APSR 2.0 stepped back to having a generic TNM model and code system across all cancer types for practical reasons.
There already is a code system OID for the whole of TNM (2.16.840.1.113883.15.6), for which we are seeking an internationally aligned canonical URL.
Grahame Grieve (Mar 12 2020 at 12:01):
well, awesome. Glad some practicality has broken out
Stefan Lang (Mar 12 2020 at 12:02):
Well, this was some 5 years+ ago ;)
Halina Labikova (Mar 23 2020 at 16:30):
@Patrick Werner What is the workflow to establish the uicc uri as the official one? Does it go through the vocab group first?
Patrick Werner (Mar 23 2020 at 16:32):
Yes, create an JiraTracker item, and talk to vocab. They then possibly refer this to UTG.
Patrick Werner (Mar 23 2020 at 16:32):
Which usually isn't a fast process at all.
Patrick Werner (Mar 23 2020 at 16:33):
In the meantime it would be nice to be "unofficially" alligned on using the same CS URI.
Patrick Werner (Mar 23 2020 at 16:34):
i personally like: http://www.uicc.org/tnm
@May Terry what do you think?
Halina Labikova (Mar 23 2020 at 16:55):
Great! On that note, shouldn't then this uri be used for the AJCC non-TNM criteria as well, for the sake of consistency? (https://ajcc.3scale.net/docs/staging-elements#/ includes criteria like PSA value or Gleason score as well)
Patrick Werner (Mar 23 2020 at 17:00):
hmm maybe i'm misunderstanding. But PSA is a (LOINC or SCT coded) Observation to me
Patrick Werner (Mar 23 2020 at 17:01):
So Gleason or PSA aren't CodeSystems, but Observations.
Patrick Werner (Mar 23 2020 at 17:02):
For TNM UICC and AJCC are equivalent (at least to my knowledge), i'm pushing UICC because AJCC will cause irritations outside the US i guess. UICC is international, shouldn't be a problem.
Halina Labikova (Mar 23 2020 at 17:15):
This type of modelling is sort of problematic for non-us EHRs - LOINC is too granular to use (the data is entered at point when the information about what test was used is already lost), SNOMED has licencing restrictions. Also, AJCC uses the non-tnm elements (hist grade, gleason score, hpv status, ER/PR, oncotype dx) to calculate the stage group, so there should be at least a hasMember relationship to those observations, no?
Halina Labikova (Mar 23 2020 at 17:16):
So Gleason or PSA aren't CodeSystems, but Observations.
Yes, I meant that their codeable concepts, when used as part of Cancer Stage, could use uicc uri in the system
May Terry (Mar 23 2020 at 17:20):
Patrick Werner said:
i personally like: http://www.uicc.org/tnm
May Terry what do you think?
I'm not completely familiar with how UICC differs from AJCC, but I wasn't able to find UICC distinguishes between clinical and pathological staging. Also, we'd have to pass it by ASCO and the mCODE executive committee but I suspect they won't be receptive to having UICC be the mCODE default staging system. Regardless, we should be able to represent UICC. mCODE PrimaryCancerCondition.stage.assessment points to CancerStageParent http://hl7.org/fhir/us/mcode/StructureDefinition-mcode-cancer-stage-parent.html which should allow you to specify UICC as a staging system and also still include a Clinical TNM profiles as well as Pathologic TNM profiles.
Patrick Werner (Mar 23 2020 at 17:39):
Halina Labikova said:
This type of modelling is sort of problematic for non-us EHRs - LOINC is too granular to use (the data is entered at point when the information about what test was used is already lost), SNOMED has licencing restrictions. Also, AJCC uses the non-tnm elements (hist grade, gleason score, hpv status, ER/PR, oncotype dx) to calculate the stage group, so there should be at least a hasMember relationship to those observations, no?
To me the UICC can be derived from PSA, pearson score,.... observations.
Patrick Werner (Mar 23 2020 at 17:40):
these Observations are then accessed and put into: https://en.wikipedia.org/wiki/TNM_staging_system#Other_parameters
Patrick Werner (Mar 23 2020 at 17:42):
When they are part of the UICC the code.system of these components should start with an uicc uri as well.
Patrick Werner (Mar 23 2020 at 17:43):
fyi, this is the german TMN profile (currently reviewed and converted to R4): https://simplifier.net/ccr/observation-cancer-base-tnm-0.1-duplicate-2
Patrick Werner (Mar 23 2020 at 17:44):
@Stefan Lang we don't have the G parameter of UICC modeled as a component. There sure was a good reason, which i don't know. Also want to include you into the discussion - asking for your opinion
Stefan Lang (Mar 23 2020 at 17:50):
G is part of the histology, but not of TNM, though it is sometimes included in the human readable representation of TNM, just as R is.
But in our discussions a couple of years ago, we came to the conclusion to keep G and R seperate from TNM/UICC.
Patrick Werner (Mar 23 2020 at 19:10):
just had a quick lookup. PSA is used in the AJCC Prognostic Stage Group, so discrete from TMN itself.
Halina Labikova (Mar 24 2020 at 20:03):
Hmmm. So when the AJCC API for staging is called (input "MAG-PRO"), both AJCCfactor-PSA and AJCCfactor-GG are listed alongside the Tvalue, Nvalue and Mvalue, to calculate stagegroup. To me, it would make sense to use the same system uri for all five when used in that context.
Patrick Werner (Mar 25 2020 at 13:30):
Ok, no we are getting there. Stage groups are coming from AJCC -> system uri should be AJCC based.
TNM is defined internationally by the UICC -> system uri should be UICC based.
Last updated: Apr 12 2022 at 19:14 UTC
