In the international release there a small number of ‘Product containing only’ medicinal product (MP only) concepts that are children of another MP only concept.
For example:
775397002 |Product containing only crisantaspase (medicinal product)|
Is_A
774654007 |Product containing only asparaginase (medicinal product)|
It seems odd that a concept stated as only containing something can be a child of a concept that only contains something different.
I will add I don’t think the modelling on the MP only concepts is wrong, but instead I think the issue lies in the modelling for the substances.
For example:
370893003 |Crisantaspase (substance)|
Is_A
371014004 |Asparaginase (substance)|
Medicinal Products are generally authored as closed world, and it seems wrong to me that the MP only can be a child of a ‘different’ ingredient.
It makes some sense for the modifications, such as Daunorubicin and Daunorubicin (liposomal) and Theophylline and Aminophylline, but less sense for Crisantaspase as a type of Asparaginase.
Crisantaspase is a kind of asparaginase, and is modeled as a subtype rather than a modification. MPOnly classes are usually modeled using an active ingredient, which has a property chain attached to infer that modifications of a substance could be inferred as supertypes when used as targets of has active ingredient. In this case, the domain experts might need to discuss whether they consider the inference valid. As a physician, I think that the modeling is correct, because the Erwinia asparaginase (and the E Coli asparaginase) are not necessarily modifications and the subtype relationship might be better, but open to be corrected by domain experts). If the inference is invalid, the modeling of the substance would need to consider is modification of, AND the MPOnly asparaginase would need to be an MPOnly-precise if cisantaspace MPOnly should not classify under asparaginase MPOnly.. Just elaborating for completeness, not because I would change it without domain expert input.
I agree with Guillermo I think the modelling of the substances is correct asparaginase is a “class” - Colaspase and Crisantaspase are specific types (from different organisms).
I don’t think the MP is wrong either. But is it necessary? SNOMED generally doesn’t have “Product containing only (medicinal product)” concepts.
For example:
39487003 |Insulin-containing product|
890458001 |Penicillin-containing product|
Both exist but no “ONLY” subtype with the same ingredients (only the specific subtypes).
That seems to be an editorial choice though.. “ONLY class” concepts aren’t inherently wrong.
774654007 |Asparaginase only product| could subsume a product like “Product containing ONLY both Colaspase and Crisantaspase” - modelled as precise ingredients.
The obvious example for me is a concept like Vitamin C products.
This product contains two ingreidents but could be described as a “Product containing only Vitamin C”
(noting that modifications and Is A - function the same with the hasIngredient role chains)
The question is really “Is 785935005 |Asparaginase 10000 unit powder for solution for injection vial| a valid concept” ?
AMT has an equivalent concept 38005011000036100 |Asparaginase 10 000 Kyowa units injection, vial|
For the specific brand- 37642011000036101 |Leunase 10 000 Kyowa units powder for injection, 1 vial|
So it could be modelled with that? Perhaps as a precise ingredient?
But it looks like there was an editorial decision in AMT or TGA that the specific Asparaginase was not relevant (I’m only speculating).
So it comes back to how precise do the products need to be modelled?
From previous discussions, it seems the requirements vary between countries.
I agree that both Crisantaspase (substance) and Escherichia coli asparaginase (substance) are subtypes (and not modifications) of Asparaginase (substance) (manufactured). So I think the substance hierarchy is correct.
However having done some research, regulations of some countries seem to permit labelling using L-asparaginase to represent Escherichia coli asparaginase. I was unable to find detailed product information for these products. I agree Matt that 785935005 |Asparaginase 10000 unit powder for solution for injection vial is ambiguous. There are also recombinant versions now.
We could look to inactivate ‘Product containing only asparaginase (medicinal product)’ as Product containing only x (medicinal product)| concepts are abstract representations of the active ingredient(s) for a medicinal product. For asparaginase products, given that the dosing and adverse reaction profiles differ, it is important to specify the precise active ingredient (which includes the source).
We could also look at all concepts that do not specify Crisantaspase (substance) or Escherichia coli asparaginase (substance) with a view to inactivate as ambiguous and replace with the two options?
I’m not sure we need to retire the Medicinal product concept.
And I’m not sure it’s ambiguous.
It’s currently behaving like the other MP groups that group specific salts/modification.
It comes back to are the different forms interchangeable, clinically significant - and what the actual use case for the MP class is.
As for the Clinical Drug - we have pretty much an identical concept in AMT. So it’s possibly down to local regulations - and “precise ingredient” rules.
I think it makes sense that asparaginase is a class of substances, and as such the MP only concept really means “products that only contain some type of asparaginase but no other ingredients”.
So I think the MP is OK. I think there’s a broader discussion about where MP only concepts are and aren’t useful compared to their “at least” cousins, and where they should and shouldn’t exist as a result.
I think part of the problem here is that these medications have evolved over time. Initially, in the late 70s, there was only Asparaginase. So that is what was prescribed and recorded in patient notes. A couple of years later a new variant Erwinia Asparaginase was identified. Only then it became important to specify E coli originated or Erwinia. With the benefit of hindsight all of the original products should probably have been re-named as E. coli. asparaginase. However if any of the products that were marketed as “Asparaginase” had been discontinued the likelyhood of being able to confirm the type of asparaginase in that product is slim. So what was once considered to be a drug concept is now perceived as a grouper concept. The same has occurred with the interferons and insulins.
From a UK use case point of view (other NRCs may have different use cases) having a MP only with a matching ingredient / substance to a ‘Product containing precisely’ (Clinical Drug).
You could argue that a MP only that does not infer any clinical drugs that have the same ingredient, have become a grouper by definition. In this case it isn’t true as 785935005 |Product containing precisely asparaginase 10000 unit/1 vial powder for conventional release solution for injection (clinical drug)| exists.
@mcordell Great visual. (We’d love the same for RxNorm-in-OWL clinical drugs!)
I think some of this modeling could be based on the Substance hierarchy. The substance is probably a child of penicillin, signaling a decision?
FWIW, In RxNorm ingredients (INs) and precise ingredients (PINs) are separated but do have relationships to one another (albeit not hierarchical, but something SNOMED gives us for free). In most cases, these terms assert to both SNOMED substance and medicinal product IDs from editorial curation.
