Stream: implementers
Topic: TNM Classification System URI
Alexander Henket (May 27 2020 at 13:14):
https://www.cancer.gov/about-cancer/diagnosis-staging/staging
Did anyone ever try to use TNM Cancer Classification in FHIR and invented a system URI and/or OID for it? We are on the verge of doing so, but it feels wrong to invent a "Dutch" identifier for an international system.
Lee Surprenant (May 27 2020 at 13:48):
Did you check mCODE?
Lee Surprenant (May 27 2020 at 13:50):
looks like they use LOINC for this http://hl7.org/fhir/us/mcode/StructureDefinition-mcode-tnm-clinical-stage-group.html ?
Lee Surprenant (May 27 2020 at 13:51):
scratch that, LOINC is just the code. the value is in valuesets like http://hl7.org/fhir/us/mcode/ValueSet-mcode-tnm-primary-tumor-category-vs.html. (http://cancerstaging.org)
Alexander Henket (May 27 2020 at 14:06):
As a non-US person I would not find a value set like this too easily in us core. Seems that all this is missing is:
- A listing in the http://build.fhir.org/terminologies-systems.html
- An OID for V3 usage
@Ted Klein / @Grahame Grieve Would a ticket requesting adding this system to the list of known systems including an OID stand a chance? FHIR#27736
Alexander Zautke (May 27 2020 at 14:09):
For the German base profiles we are using urn:oid:2.16.840.1.113883.15.16. See https://simplifier.net/ccr/tnm-m
http://www.oid-info.com/get/2.16.840.1.113883.15.16
Alexander Zautke (May 27 2020 at 14:10):
I thought the OID was also documented somewhere in art-decor, but I can't find it at the moment
Alexander Zautke (May 27 2020 at 14:11):
Don't know if @dr Kai U. Heitmann or @Stefan Lang know more?
Ted Klein (May 27 2020 at 14:22):
The TNM is published by the AJCC and a whole collection of folks have struggled for years to get them to publish this unencumbered and in some machine readable form for usage by the HIT implementation community. All to no avail. It is a perennial topic of discussion at WGMs. Many of the concepts were added to the US extension to SNOMED partly for these reasons some years ago. I do not know the current state of this.
Stefan Lang (May 27 2020 at 14:37):
That OID is the one used already in the German Diagnosis IG (CDA) which is pretty much at the beginning of my engagement with HL7 beyond v2 ;)
So probably @dr Kai U. Heitmann knows more.
But I would expect to find that OID (but not any of the concepts) in HL7 International's OID registry. As for concepts: I assume everybody literally browses the books published by UICC (for US: AJCC)
Stefan Lang (May 27 2020 at 15:00):
Anyway, I would very much appreciate a global canonical uri (url) for tnm; rather there would be two, since UICC and AJCC are not completely identical.
Also another one for ICD-O.
Alexander Henket (May 28 2020 at 10:29):
Interesting: if you need 2 system URIs, would that not imply two OIDs as well? oid-info.com lists the OID as registered by Kai Heitmann in the HL7 OID Registry so apparently that single OID already lives where I would hope it would be. Now we only need to find out if this singular OID is correct for both UICC and AJCC?
Stefan Lang (May 28 2020 at 11:45):
Well, that OID's description says: "TNM Edition8 UICC" which would imply the need for a different OID "TNM AJCC".
To be more precise, this is the OID specifically for the 8th edition of UICC TNM (which is the current one), but I would expect to have a single canonical URL and enforce usage of the version element in instances, just like we do with ICD-10-GM in the German Condition profile.
Stefan Lang (May 28 2020 at 11:59):
E.g. TNM UICC 7th edition has a different OID: http://www.oid-info.com/get/2.16.840.1.113883.15.6 (which actually is the one from my early HL7 days ;-))).
But for practical reasons, especially maintenance of IGs and corresponding conformance/terminology resources, these would be equivalent to the combination of system+version, like:
2.16.840.1.113883.15.16 => system = http://..../uicc-tnm , version = 8
2.16.840.1.113883.15.6 => system = http://..../uicc-tnm , version = 7
...
Alexander Henket (May 28 2020 at 12:41):
Ok, so we need an OID and a URI per version, one for UICC and one for AJCC.
But that must mean that US Core is using a too simple system URI? They are using http://cancerstaging.org for AJCC "preferably version 8"
Alexander Henket (May 28 2020 at 12:43):
TNM really needs to make sure they don't cause this new system per version thing, because this is bound to go sour in implementation
Stefan Lang (May 28 2020 at 12:58):
I would consider http://cancerstaging.org fine for AJCC if version was mandatory in resource instances.
Not using version is not exact since the meaning of codes may change between the TNM editions.
Stefan Lang (May 28 2020 at 13:03):
But that "new version versus new code system"thing is a discussion probably going on for decades now.
So if US core users are fine with not knowing the version (= TNM edition), so be it.
For our German IG I'm a strong defender of enforcing version, still ;-)
(and hoping for a globally defined code system URL expressing "UICC TNM" instead of OIDs)
Alexander Henket (May 28 2020 at 13:16):
In our mixed V3/FHIR world, we really need OIDs but are fine to convert to/from system URIs when necessary
Alexander Henket (May 28 2020 at 13:18):
Does this one make sense? http://oid-info.com/get/1.2.840.10008.6.1.397 (Tumor Stages From AJCC (CID 6068))
May Terry (May 28 2020 at 17:26):
Ted Klein said:
The TNM is published by the AJCC and a whole collection of folks have struggled for years to get them to publish this unencumbered and in some machine readable form for usage by the HIT implementation community. All to no avail. It is a perennial topic of discussion at WGMs. Many of the concepts were added to the US extension to SNOMED partly for these reasons some years ago. I do not know the current state of this.
More like AJCC has a prognostic staging system which leverages TNM but also includes prognostic factors which will vary based on cancer type, and are not included in other international staging systems (e.g.: UICC). In some respects, it deserves its own URL.
Robert McClure (May 29 2020 at 17:45):
@Carol Macumber @Julie James Weigh in here. Folks, when HL7 needs a code system identifier for an externally defined code system, you need to route that through the HTA first looking at the list of systems already addressed or under consideration here and possibly follow the approach noted here. That said, TNM is on the existing list but unfortunately is not completed.
Carol Macumber (Jun 01 2020 at 14:08):
@Julie James , looks like you created the page for TNM in April. Anything I can do to help this one along?
Julie James (Jun 01 2020 at 14:22):
Looking at our Minutes from 2020-04-29, it appears that nobody is sure whether TNM is a code system per se. The Europeans tend to use it as such by extracting codes from the book. SNOMED has the content up to V6. I created the page but got no further than that....I almost wonder if on that page it should say "this is not a code system in our terms" because there is no machine processable set of content as such. If I understand correctly - and I may well not - it's not a code system per se but a means of building an individual's cancer "score" based on observations- so more like your particular Glasgow Coma Score, which gets measured at various points... or your 2minute APGAR.
Julie James (Jun 01 2020 at 14:32):
BTW - why can folk not use the codes in the NCI-T and take a value set from that?
Michael van der Zel (Jun 04 2020 at 13:34):
I was hoping on a machine processable set with OIDs and not some textual list of strings as it is now. Or a description of how to build the "score" from its components.
Michael van der Zel (Jun 10 2020 at 09:33):
What about this computable (through an API) version of AJCC https://ajcc.3scale.net ?
Linda Mook (Aug 21 2020 at 12:31):
by now we have proceeded and set up most valuesets for UICC and temporarily 'forget' about the OID discussion. I have checked the FHIR valuesets but it looks like most complicated valuesets that are different per disease are not completed and one is referred to 'the TNM book', or am I wrong? In the Netherlands mostly UICC is used so AJCC is not our most urgent need and implementation use-case for consuming from the AJCC API by healtcare systems is limited over here. Saying that, if there is any new information or insights anybody can share it is very welcome.
Patrick Werner (Nov 01 2021 at 13:53):
Looking at the HTA confluence i found: https://confluence.hl7.org/display/TA/TNM and https://confluence.hl7.org/pages/viewpage.action?pageId=113679187
Patrick Werner (Nov 01 2021 at 13:54):
`
Patrick Werner (Nov 01 2021 at 13:54):
So HTA is thinking about using https://cancerstaging.org/
Patrick Werner (Nov 01 2021 at 13:59):
i think the trailing slash should be removed here to allign with the rest of external CS uris
Patrick Werner (Nov 01 2021 at 13:59):
(mcode doesn't use the trailing slash)
Patrick Werner (Nov 01 2021 at 14:09):
I also was wondering what http://cancerstaging.org includes? T-N-M Stage codes, UICC stages and AJCC Prognostic Stage Groups?
Patrick Werner (Nov 01 2021 at 14:09):
Could be a little bit messy.
Patrick Werner (Nov 01 2021 at 14:10):
Maybe a CS for TNM (universal, international), a CS for UICC stages and a CS for AJCC Prognostic Stage Groups would be easier to understand?
Patrick Werner (Nov 01 2021 at 14:10):
fyi @May Terry
Grahame Grieve (Nov 01 2021 at 20:01):
@Patrick Werner Agree but you need to make sure that feedback goes back into the HTA process
Patrick Werner (Nov 03 2021 at 11:09):
will try.
Added a comment here: https://confluence.hl7.org/pages/viewpage.action?pageId=113679187 hope this is a proper way to comment.
fyi @Reuben Daniels @May Terry @Sylvia Thun
Last updated: Apr 12 2022 at 19:14 UTC
