LOINC Tranche D Closing Ledger: Enzyme Activity and Metabolite Assays

Date: 2026-10-16 Scope: projects/LOINC_DIAGNOSTICS.md Phase 3, Tranche D — lysosomal-storage and inborn-error-of-metabolism groupings, bounded by kb/disorders/ entries already carrying an enzyme-deficiency or metabolite-storage mechanism. PRs: #14481, #14491, #14494, #14561, #14617 (all open, none merged as of 2026-10-16 — every "already bound on the sibling entry" fact below is cross-branch, not cross-main, as of that date, and should be re-checked against origin/main once any of these PRs lands).

Census: entries examined and their outcome

Entry	Outcome	Code / reason
Fabry_Disease	Bound	LOINC:LP14012-6 Alpha galactosidase A
Gaucher_Disease	Bound	LOINC:LP15604-9 Glucosylceramidase
Tay-Sachs_Disease	Bound	LOINC:LP14296-5 Beta-N-acetylhexosaminidase.A
Infantile-Onset_Pompe_Disease	Bound	LOINC:LP14013-4 Acid alpha glucosidase; LOINC:55827-0 DBS activity
Metachromatic_Leukodystrophy	Bound	LOINC:LP14023-3 Arylsulfatase A
Niemann-Pick_Disease_Type_A	Bound	LOINC:LP14035-7 Acid sphingomyelinase
Niemann-Pick_Disease_Type_B	Bound	LOINC:LP14035-7 Acid sphingomyelinase
Sandhoff_Disease	Bound	LOINC:LP14022-5 Beta-N-acetylhexosaminidase (plain, total A+B)
Krabbe_Disease_Due_To_Saposin_A_Deficiency	Bound	specimen-specific galactosylceramidase fibroblast codes (not the Part code LP14027-4, per that entry's own notes)
Farber_Disease	Declined	no LOINC code exists for acid ceramidase / ASAH1 / C26-ceramide
Attenuated_Mucopolysaccharidosis_Type_I	Bound	LOINC:LP14016-7 Alpha-L-iduronidase
Sanfilippo_syndrome	Bound (4-way split)	LOINC:LP14028-2 (IIIA, sulfamidase), LOINC:LP21324-6 (IIIB, alpha-N-acetylglucosaminidase), LOINC:LP14461-5 (IIIC, acetyl-CoA transferase), LOINC:LP14033-2 (IIID, N-acetylglucosamine-6-sulfatase)
Alpha_Mannosidosis	Bound	LOINC:LP14014-2 Alpha-mannosidase
Beta_Mannosidosis	Bound	LOINC:LP14020-9 Beta-mannosidase
Mucopolysaccharidosis	Declined	grouping-level entry; no single enzyme to bind
Tay-Sachs_Disease_AB_Variant	Bound	LOINC:LP14296-5 (the HexA assay the row performs; GM2A activator codes exist but no cited activator assay is run)
Gaucher_Disease_Due_To_Saposin_C_Deficiency	Bound	LOINC:LP15604-9 (the glucocerebrosidase assay the row performs; saposin-C activator codes exist but no cited activator assay is run)
Kanzaki_Disease	Bound	LOINC:LP14015-9 Alpha-N-acetylgalactosaminidase
Late-Onset_Pompe_Disease	Bound	LOINC:LP14013-4 + LOINC:55827-0, reused from Pompe_Disease.yaml
Chronic_Neurovisceral_Acid_Sphingomyelinase_Deficiency	Bound	LOINC:LP14035-7 Acid sphingomyelinase
Carbonic_Anhydrase_VA_Deficiency	Declined	diagnosis is a multi-analyte metabolic panel plus CA5A sequencing, none of it CA-VA-specific; no LOINC code for CA-VA/CA5A exists, and the biotinidase mention is a differential (multiple carboxylase deficiency)
DTYMK-Related_Neurodegeneration	Declined	dTMPK fibroblast assay is explicitly "a research method, not a clinical service"; no LOINC code exists
Autosomal_Recessive_Non-Syndromic_Intellectual_Disability	Declined	biotinidase mention is inside a normal newborn-screening exclusion panel
Christianson_Syndrome	Declined	hexosaminidase mention is a MODEL_ORGANISM mouse-histochemistry finding, not a human test; diagnosis is SLC9A6 sequencing
Combined_Saposin_Deficiency	Declined	arylsulfatase A mention is in the MLD differential; own biochemical finding (IHC saposin panel) has no LOINC code; PSAP gene codes exist only at method-specific granularity
Schindler_Disease	Bound	LOINC:LP14015-9, same Part code as allelic Kanzaki disease
Sea-Blue_Histiocyte_Syndrome	Bound	LOINC:LP14035-7; verified the row is this entry's own secondary-cause workup, not a differential mention
Maroteaux-Lamy_syndrome	Declined	no diagnosis: section; only biochemical marker is the dermatan sulfate substrate; no Part-level LOINC code exists for arylsulfatase B (only specimen-specific codes)
Mucopolysaccharidosis_type_X	Declined	entry's own text: "There is no clinically available ARSK enzyme assay"; no LOINC code for ARSK exists
X-linked_Chondrodysplasia_Punctata_1	Declined	entry's own evidence: "Testing of ARSL enzymatic activity is not currently available on a clinical basis"
Kufor-Rakeb_syndrome	Declined	glucocerebrosidase mention is a mechanism claim on an iPSC/cell-line biochemical marker explicitly flagged "not established as a routine clinical diagnostic biomarker"; no diagnosis: section
EDEM3-Congenital_Disorder_of_Glycosylation	Declined	"mannosidase" names the gene's own enzymatic identity, not a measured analyte; the real diagnostic test (plasma N-glycan ratio profiling) has no LOINC code
MOGS-Congenital_Disorder_of_Glycosylation	Declined	"mannosidase" is mechanism prose for the compensatory bypass pathway; the entry's own notes: already correctly states no LOINC code exists for the diagnostic analyte (urinary Glc3Man), verified independently
Action_Myoclonus-Renal_Failure_Syndrome	Declined	the entry's own diagnosis row states a leukocyte glucocerebrosidase assay "must not be used to exclude the diagnosis" (tissue-specific mistrafficking can give a false-normal result); diagnosis is SCARB2 sequencing
Hereditary_Spastic_Paraplegia_46	Declined	a genuine GBA2 activity diagnosis row exists, but GBA2 is a distinct non-lysosomal enzyme from GBA1 (classic Gaucher glucocerebrosidase); no GBA2-specific LOINC code exists, and reusing the GBA1 code would misidentify the analyte
Autosomal_Recessive_Cerebellar_Ataxia_With_Late_Onset_Spasticity	Declined	same GBA2/GBA1 situation as SPG46 — this entry states it is the same biallelic GBA2 disease spectrum
Hypertrophic_Cardiomyopathy_21	Declined	alpha-galactosidase A mention is a Fabry phenocopy-exclusion test, explicitly stated to reclassify the patient out of this entry if positive; the entry itself has no identified gene or biochemical marker
SLC5A6-Related_Multivitamin_Transporter_Deficiency	Declined	biotinidase activity is explicitly an exclusion finding — normal/near-normal biotinidase is what distinguishes this disease from biotinidase deficiency, which it mimics
Spinal_Muscular_Atrophy_Progressive_Myoclonic_Epilepsy	Declined	genuine diagnostic marker (acid ceramidase activity / C26-ceramide, ASAH1), but no LOINC code exists — same absence as the allelic Farber_Disease entry

Every code above was read directly from the branch that carries it (origin/loinc-d-batch1 through loinc-d-batch4, and loinc-d-batch6) in the same step this table was written, not from an earlier summary.

Across 39 entries examined, 20 bindings landed and 19 declines were recorded with a stated reason. That ratio is itself informative: this tranche's entries disproportionately describe tests by their absence (a research-only assay, a method that exists but carries no LOINC identity code, an enzyme that must not be used diagnostically) rather than by a clean clinical test with an obvious code.

Finding 1: the candidate filter's false-positive rate

The candidate pool for this tranche was built by matching lysosomal/IEM enzyme names against kb/ text. Two successive filters were tried, and both over-matched in a tranche-specific way:

• Whole-file matching (tranche D batch 5): every one of five candidates was a false positive, and each failed for a different reason. Biotinidase sat in a differential or a normal-exclusion panel on three entries — Carbonic_Anhydrase_VA_Deficiency, DTYMK-Related_Neurodegeneration, and Autosomal_Recessive_Non-Syndromic_Intellectual_Disability — plus an enzyme name inside a MODEL_ORGANISM mouse-histochemistry finding (Christianson_Syndrome) and an enzyme named only in another disease's differential-diagnosis block (Combined_Saposin_Deficiency). None was the entry's own test.
• Section-scoped matching (diagnosis:/biochemical: only, tranche D batch 6 onward): better, but still not clean. Two of the batch-5 entries already declined above — Autosomal_Recessive_Non-Syndromic_Intellectual_Disability and DTYMK-Related_Neurodegeneration — kept resurfacing as candidates even under the improved filter, because their biotinidase mention genuinely sits inside a diagnosis: section — as a normal exclusion panel, not as the entry's own positive finding. Section membership does not distinguish "this section names the test that diagnoses this disease" from "this section names a test that was run and came back normal, ruling out something else." (Both batch-7 declines — Kufor-Rakeb_syndrome and the eight other entries in that batch — are a separate set; neither Autosomal_Recessive_Non-Syndromic_Intellectual_Disability nor DTYMK-Related_Neurodegeneration is among them.)

The recommendation for whoever extends this tranche: filter the candidate pool on the enzyme appearing as a measured analyte — a biomarker_term, a measurements row, or a named assay with its own evidence — not on the word appearing anywhere in a section. That distinction cannot be made by string matching alone; every one of the 19 declines above needed the row read in context (whose test is this, and does the row's own evidence describe a result for this disease or for the thing it is being distinguished from).

Finding 2: enzyme-family name collisions that validate perfectly while being wrong

Two gene/enzyme families in this tranche share a name stem across genuinely different molecular entities, and binding the wrong family member's code validates perfectly while being wrong.

• ARSA / ARSB / ARSE (ARSL) / ARSK. Four distinct arylsulfatases, and the four entries that depend on them took four different paths. ARSA has a Part-level identity code (LOINC:LP14023-3, bound on Metachromatic_Leukodystrophy). ARSB (Maroteaux-Lamy syndrome) has no Part-level code at all — only specimen-specific result codes exist, and Maroteaux-Lamy_syndrome was declined on that basis, with no diagnosis: section to attach even a specimen-specific code to. Searched and confirmed no code exists at any granularity for ARSE/ARSL (X-linked_Chondrodysplasia_Punctata_1) or ARSK (Mucopolysaccharidosis_type_X). A search for "arylsulfatase" without the letter suffix returns only ARSA and ARSB/ARSC hits — it will not surface the absence of ARSE/ARSK coverage, so a future search needs to check each letter explicitly rather than trusting a generic "arylsulfatase" query.

Krabbe_Disease_Due_To_Saposin_A_Deficiency (galactosylceramidase, not an arylsulfatase, but the same specimen-granularity question) is the instructive contrast to Maroteaux-Lamy's decline: it also declined the Part-level code, but for the opposite reason — not because no code exists, but because the entry's own diagnostic claim is specifically the leukocyte-versus-fibroblast activity discordance that distinguishes this activator-deficiency variant from classic GALC-deficient Krabbe disease. A single Part-level code would have erased exactly the fact the row exists to state, so two specimen-specific codes were bound instead. Maroteaux-Lamy has no code of any granularity to choose between; Krabbe has one and deliberately did not use it. Both are specimen-level decisions, reached for opposite reasons. - GBA1 / GBA2. Two distinct glucosylceramidases (lysosomal vs. non-lysosomal), sharing enough vocabulary ("glucocerebrosidase", "glucosylceramidase", "beta-glucosidase") that a GBA2 diagnostic row reads, at a skim, like a GBA1/Gaucher mention. Hereditary_Spastic_Paraplegia_46 and Autosomal_Recessive_Cerebellar_Ataxia_With_Late_Onset_Spasticity both carry a genuine, cited GBA2 enzyme-activity diagnosis row — not a mechanism mention, not a differential — and both were still correctly declined, because no GBA2-specific LOINC code exists and the GBA1 code (LOINC:LP15604-9) names the wrong gene product. This is the one methodological case in the tranche where the candidate was right (a real test exists) and the binding would still have been wrong.

Closing summary

39 stratum entries were examined in this tranche: 20 carry a new LOINC binding, and 19 were declined with a stated reason, across PRs #14481,

14491, #14494, #14561 and #14617 (all open as of 2026-10-16). Both counts

are denominators worth stating explicitly rather than quoting a bare total later — 20 bound, 19 declined, 39 examined, and all three numbers should still add up for anyone re-deriving them from the table above. The declines are recorded here with their reason so the next pass over this tranche — or the next tranche that touches a lysosomal-storage or IEM entry — does not re-derive the same conclusions from scratch.