Complement Deficiency IEIs (IUIS Table 8)

name: Complement Deficiency IEIs
display_name: Complement Deficiency IEIs (IUIS Table 8)
creation_date: "2026-08-20T00:00:00Z"
description: >-
  IUIS Table 8 — complement deficiencies. These inborn errors of immunity
  disable a component or regulator of the complement cascade, an innate humoral
  effector system that operates without cells. The infection phenotype is
  strikingly pathway-specific: terminal-component defects impair membrane
  attack complex assembly and admit Neisseria almost exclusively, while early
  classical-pathway defects impair immune-complex clearance and present as
  lupus-like autoimmunity.
grouping_basis:
- SHARED_MECHANISM
- SHARED_PHENOTYPE
grouping_rationale: >-
  Grouped on the IUIS Table 8 lesion in the complement system. Complement is
  kept as its own table, and its own grouping, rather than being folded into
  innate immunity (Table 6): it is a soluble cascade rather than a cellular
  program, its deficiencies are inherited as clean autosomal recessive
  enzymopathies of individual components, and the position of the defect within
  the cascade — early classical, alternative, lectin, terminal, or regulatory —
  predicts the clinical phenotype at the level of the arm. That prediction is
  strong for the terminal, early classical and regulatory arms and weakens where
  an arm has redundant inputs: the lectin arm is the worked exception, and the
  `complement_table8_lectin_arm_qualifies_the_claim` discussion carries the
  argument. Members are kept as separate Disease entries because that cascade
  position differs.
mappings:
  mondo_mappings:
  - term:
      id: MONDO:0003832
      label: complement deficiency
    mapping_predicate: skos:broadMatch
    mapping_source: MONDO
    mapping_justification: >-
      IUIS Table 8 aligns with MONDO complement deficiency (MONDO:0003832).
      broadMatch rather than exactMatch because MONDO's class is not restricted
      to germline monogenic inborn errors, whereas IUIS Table 8 is; the MONDO
      term is therefore wider than this grouping.
    consistency:
    - reference: MONDO
      consistent: CONSISTENT
      notes: >-
        Re-verified 2026-09-13 with `runoak -i sqlite:obo:mondo descendants -p i
        MONDO:0003832`. Two of the three members are is-a descendants:
        Late Complement Component Deficiency (MONDO:0015700, immunodeficiency due
        to a late component of complement deficiency) and Complement Component 2
        Deficiency (MONDO:0009006, complement component 2 deficiency).

        Hereditary Angioedema is not, and the mismatch is a MONDO modelling
        choice rather than a membership error. MONDO splits the molecular lesion
        from the clinical syndrome: MONDO:0007361 "C1 inhibitor deficiency" is a
        descendant of MONDO:0003832, but the dismech entry is anchored on
        MONDO:0019623 "hereditary angioedema", which MONDO places under
        angioedema and skin vascular disease (`runoak ancestors -p i
        MONDO:0019623` returns no complement ancestor). IUIS Table 8 keys its row
        on the C1-inhibitor deficiency, so the member belongs here under this
        grouping's own criterion even though the anchored MONDO class sits
        elsewhere.
membership_criteria:
- description: >-
    A disorder belongs to IUIS Table 8 if it is an inborn error of immunity
    caused by deficiency of a complement component, a complement pathway
    recognition molecule, or a complement regulatory protein.

    "Recognition molecule" was added 2026-09-22 when Ficolin 3 Deficiency was
    listed. Ficolin-3 is not a cascade component: it binds acetylated ligands on
    microbial and altered-self surfaces and recruits the MASPs, so it sits
    upstream of the C4/C2 cleavage step rather than being one of the proteins
    cleaved. The earlier two-part wording therefore did not literally admit it,
    and IUIS Table 8 does carry it as a row, as this grouping's own notes have
    recorded since it was created. Widening the description is the resolution
    rather than listing the member as an exception, because the same gap applies
    to the uncurated MASP2 row and to the C1q recognition subunit.
  criteria_semantics: NECESSARY
  logic:
    criterion_predicate: HAS_CLASSIFICATION
    classification: "iuis_category:complement deficiency"
    description: >-
      Assigned to IUIS Table 8 (complement deficiencies) via
      classifications.iuis_category on the member Disease entry. Stated in the
      keyed `<slot>:<value>` form so the audit reads the structured
      classifications block.
members:
- member: Complement Component 2 Deficiency
  member_type: DISEASE
  differentiating_mechanisms:
  - description: >-
      Biallelic loss of C2, most often the 28-bp exon 6 splice-boundary deletion,
      leaves C4b with no partner and so prevents assembly of the classical-pathway
      C3 convertase C4b2a. The lectin pathway converges on that same convertase,
      so it fails too; only the alternative pathway is intact, and the
      lesion is not one of lytic failure but of impaired opsonisation and of poor
      clearance of immune complexes and apoptotic debris. That gives the
      early-classical arm its characteristic pairing, absent in the terminal arm:
      pyogenic infection with encapsulated organisms, Streptococcus pneumoniae
      rather than Neisseria, alongside lupus-like autoimmunity in roughly a
      quarter of homozygotes.
    gene:
      preferred_term: C2
      term:
        id: hgnc:1248
        label: C2
- member: Complement Component 3 Deficiency
  member_type: DISEASE
  differentiating_mechanisms:
  - description: >-
      The convergence point of the cascade, and the only member whose lesion
      disables all three activation pathways at once. Biallelic C3 loss leaves
      no C3b to opsonise, no alternative-pathway amplification loop, and no
      C3b-containing C5 convertase. It therefore combines the phenotypes of the
      arms on either side: severe pyogenic infection with pneumococci,
      Haemophilus and meningococci, as in the early classical arm but without
      an intact alternative pathway to compensate, together with
      immune-complex membranoproliferative glomerulonephritis and SLE-like
      disease, each reported in roughly a quarter of patients.
    gene:
      preferred_term: C3
      term:
        id: hgnc:1318
        label: C3
- member: Ficolin 3 Deficiency
  member_type: DISEASE
  differentiating_mechanisms:
  - description: >-
      The lectin-pathway arm, and the only member whose lesion is in a
      recognition molecule rather than in the cascade it triggers. Homozygosity
      for the FCN3 +1637delC frameshift leaves no detectable serum ficolin-3 and
      no ficolin-3-dependent complement activation. Because mannose-binding
      lectin and the ficolin-2 arm still initiate the same pathway, the block is
      partial in a way the other members' are not: C2 deficiency removes a
      convertase subunit that both the classical and lectin pathways converge
      on, and terminal deficiency removes the lytic machinery outright, whereas
      losing one of several redundant recognition molecules leaves the pathway
      reachable by other routes.

      That redundancy is why this member's phenotype is contested rather than
      characteristic, and so why it bounds this grouping's organising claim
      instead of illustrating it. The argument is recorded in the
      `complement_table8_lectin_arm_qualifies_the_claim` discussion rather than
      duplicated here.
    gene:
      preferred_term: FCN3
      term:
        id: hgnc:3625
        label: FCN3
- member: Hereditary Angioedema
  member_type: DISEASE
  differentiating_mechanisms:
  - description: >-
      The regulatory arm, and the one member whose lesion is disinhibition rather
      than deficiency of effector function. Heterozygous SERPING1 variants reduce
      functional C1 inhibitor well below the 50% haploinsufficiency predicts —
      increased catabolism, and trans-inhibition of the wild-type allele by some
      mutants, typically leave 5-30% of normal activity, which is why a
      heterozygous serpin defect causes disease at all. C1 inhibitor is the
      serpin that restrains C1r/C1s and MASP1/MASP2
      in the complement cascade and factor XIIa and plasma kallikrein in the
      contact system. Complement consumption shows up as the low C4 used
      diagnostically, but the attacks themselves are bradykinin-driven, so the
      phenotype is recurrent non-urticarial subcutaneous, abdominal and laryngeal
      edema rather than infection or autoimmunity. Note the scope caveat recorded
      on that entry's `iuis_category`: only its SERPING1 subtypes are IUIS Table 8
      disease, not its normal-C1-INH arm.
    gene:
      preferred_term: SERPING1
      term:
        id: hgnc:1228
        label: SERPING1
- member: Immunodeficiency Due To MASP-2 Deficiency
  member_type: DISEASE
  differentiating_mechanisms:
  - description: >-
      The second lectin-pathway member, and the one whose lesion is in the
      enzyme rather than in a recognition molecule. MASP-2 is the only protease
      that cleaves C4, so homozygosity for MASP2 p.D120G - which abolishes
      binding to mannose-binding lectin - removes lectin-pathway activity by
      every route, where ficolin-3 deficiency removes one of several routes
      into it.

      That makes this member the sharper test of the grouping's organising
      claim, and it fails it in the same direction. A complete, non-redundant
      block of the arm produces no consistent phenotype: the published
      homozygotes split roughly evenly into affected individuals with no shared
      presentation and healthy controls, and a targeted study found none among
      1495 adults with community-acquired pneumonia or 186 patients with lupus.
      So the lectin arm's position in the cascade predicts no phenotype even
      when the block is total, which is a stronger form of the qualification
      recorded in the `complement_table8_lectin_arm_qualifies_the_claim`
      discussion than ficolin-3 deficiency alone supplies.
    gene:
      preferred_term: MASP2
      term:
        id: hgnc:6902
        label: MASP2
- member: Late Complement Component Deficiency
  member_type: DISEASE
  differentiating_mechanisms:
  - description: >-
      Biallelic loss of any terminal component (C5, C6, C7, C8, C9) prevents
      assembly of the membrane attack complex C5b-9 and so abolishes direct
      serum bactericidal activity. Because Neisseria are killed principally by
      MAC lysis rather than by opsonophagocytosis, the deficiency produces an
      almost pathognomonic susceptibility to recurrent invasive meningococcal
      and disseminated gonococcal disease with otherwise unremarkable immunity —
      the terminal-pathway arm of Table 8.
    gene:
      preferred_term: C5
      term:
        id: hgnc:1331
        label: C5
notes: >-
  Created 2026-08-20 to give the IUIS Table 8 disorders a place in the Inborn
  Errors of Immunity tree, initially with a single terminal-pathway member.

  Resynchronised 2026-09-13. The earlier note claiming the early classical-pathway
  and regulatory arms were uncurated was already stale: Complement Component 2
  Deficiency merged in #9667 carrying `iuis_category: complement deficiency`, and
  Hereditary Angioedema was backfilled with the same classification in this pass,
  so both had become classified-but-unlisted. That drift is invisible to tooling.
  `just check-groupings` evaluates the membership criteria only against members
  already named here; it never discovers a classified entry that the list omits.
  The sync is therefore a manual diff, and the command that defines the expected
  member set is:

  rg -l "classification_value: complement deficiency" kb/disorders/

  Run it before trusting this list, and make the two agree. As of 2026-09-22 it
  returns exactly the four members below.

  Ficolin 3 Deficiency added 2026-09-22, supplying the lectin-pathway arm. The
  entry was curated in #11885 and deliberately left unlisted there to keep that
  diff to a single disease; this is the follow-through (#11920). Adding it
  required widening the membership criterion's description, which spoke only of
  components and regulators and so did not literally admit a recognition
  molecule — the reasoning is recorded on the criterion itself rather than
  here, so it sits next to the text it explains.

  The sync command above also returns Complement_Component_4A_Deficiency,
  which carries the classification but is not yet listed; it was left for a
  separate change rather than added without review of its own entry.
  Complement Component 3 Deficiency is the convergence point of all three
  pathways rather than a member of one arm, so it adds a cascade position
  without filling the alternative-pathway gap recorded in
  `complement_table8_pathway_coverage`.

  Still uncurated in kb/disorders/, and so still absent here: the remaining
  early classical rows (C1q, C1r, C1s, C4), the alternative pathway (factor B,
  factor D, properdin, factor H, factor I), and the membrane regulators (CD46,
  CD55/CHAPLE, CD59, thrombomodulin). See the
  `complement_table8_pathway_coverage` discussion. The lectin arm is now
  complete: MASP2 deficiency was curated and listed 2026-10-15, alongside the
  ficolin-3 row.

  Mannose-binding lectin deficiency is deliberately not listed as an outstanding
  Table 8 row. IUIS 2022 Table 8 carries a MASP2 deficiency row and a Ficolin 3
  deficiency row but no MBL row at all, so counting one would overstate the
  table.

  Three Table 8 rows are deliberately not claimed as members, and the reasoning
  should not be rediscovered. Periodontal Ehlers-Danlos syndrome is literally
  Table 8, as the C1R and C1S autosomal dominant gain-of-function rows, but
  dismech curates it as a `has_subtypes` entry inside Ehlers-Danlos_Syndrome
  rather than as its own Disease; `iuis_category` is entry-level and
  single-valued, so classifying it would assert Table 8 membership for every
  other EDS subtype in that file. Atypical Hemolytic Uremic Syndrome overlaps six
  Table 8 gene rows, but IUIS anchors those rows on the gene deficiency and lists
  aHUS as an associated feature rather than as the Table 8 disease, and the
  dismech entry additionally lumps an anti-factor-H antibody arm that IUIS places
  in Table 10 (phenocopies) and a DGKE arm that is not complement at all.
  Paroxysmal nocturnal hemoglobinuria is somatic PIGA and appears in neither
  Table 8 nor Table 10.
discussions:
- discussion_id: complement_table8_pathway_coverage
  kind: KNOWLEDGE_GAP
  prompt: >-
    Do four of thirty Table 8 rows support the phenotype claim this grouping
    makes about complement deficiency as a class?
  rationale: >-
    The clinically important property of Table 8 is that the position of the
    defect within the cascade predicts the phenotype — terminal defects give
    neisserial infection, early classical defects give lupus-like autoimmunity
    and immune-complex disease, regulatory defects give hereditary angioedema
    or atypical haemolytic uraemic syndrome.

    Updated 2026-09-13. The original form of this gap, a single terminal-pathway
    member, is closed: three cascade positions are now represented, so the
    contrast is demonstrable across members rather than only asserted in the
    rationale. Complement Component 2 Deficiency supplies the early classical
    arm with pneumococcal infection and lupus, Hereditary Angioedema supplies the
    regulatory arm with bradykinin-mediated attacks and no infection phenotype,
    and Late Complement Component Deficiency supplies the terminal arm with
    isolated neisserial susceptibility.

    Updated again 2026-09-22. Ficolin 3 Deficiency supplies the lectin arm, so
    four cascade positions are represented. It does not extend the contrast in
    the way the other three do, because its phenotype is contested rather than
    characteristic; that is a finding about the claim rather than a gap in the
    member, and it is recorded separately in
    `complement_table8_lectin_arm_qualifies_the_claim`.

    What remains is breadth. Four of the thirty Table 8 disease rows are
    curated, and one whole arm still has no member at all: the alternative
    pathway (factor B, factor D, properdin, factor H, factor I), plus the
    membrane regulators CD46, CD55, CD59 and thrombomodulin, and the remaining
    lectin row MASP2. A SUFFICIENT criterion still cannot be written,
    because it would have to enumerate phenotypes the KB does not yet carry for
    those arms. This is a coverage gap in kb/disorders/ rather than an open
    scientific question, but it bounds what this grouping can audit.
  proposed_experiments:
  - experiment_id: curate_early_and_regulatory_complement_defects
    name: Curate the alternative-pathway and lectin-pathway complement defects
    description: >-
      Superseded in part. The early classical-pathway arm (C2 deficiency), the
      regulatory arm (C1-inhibitor deficiency) and, as of 2026-09-22, the
      lectin-pathway arm (ficolin-3 deficiency) called for here are curated and
      listed as members. The outstanding work is the one arm with no member —
      curate at least one alternative-pathway deficiency (factor B, factor D,
      properdin, factor H or factor I, presenting with neisserial or encapsulated
      organism infection and, for factor H/I, aHUS), assign it
      `classifications.iuis_category: complement deficiency`, and add it here.
      With five cascade positions represented,
      the pathway-position-predicts-phenotype claim becomes auditable as a
      structured contrast between members rather than an assertion in the
      grouping rationale.

      One correction to the expectation this experiment was written with. It
      assumed each new arm would sharpen the contrast. The lectin arm did not:
      ficolin-3 deficiency has no settled phenotype to contrast with, so it
      qualifies the claim instead of extending it. An alternative-pathway member
      may or may not behave like C2 and the terminal components; the experiment
      is worth running either way, but "five positions represented" should not be
      read as five clean contrasts in advance of curating the fifth.
- discussion_id: complement_table8_lectin_arm_qualifies_the_claim
  kind: KNOWLEDGE_GAP
  prompt: >-
    Does cascade position predict phenotype for a deficiency of a redundant
    recognition molecule, or only for a deficiency of a shared cascade
    component?
  rationale: >-
    This grouping's organising claim is that the position of the defect within
    the complement cascade predicts the clinical phenotype. Three of the four
    members bear it out cleanly. Ficolin 3 Deficiency, added 2026-09-22 as the
    lectin-pathway member, does not, and the reason is structural rather than a
    shortfall in that entry's curation.

    Ficolin-3 is one of several redundant recognition molecules — mannose-binding
    lectin and ficolin-2 initiate the same pathway — so its loss does not remove
    the lectin pathway, only one route into it. Its phenotype is correspondingly
    unsettled. The index report read the deficiency as an immunodeficiency; a
    later series of deficient individuals concluded that reading is "more
    complicated and less clear-cut"; a systematic review found a third of the
    published deficient patients carried an SLE diagnosis. The member entry
    curates both readings as competing pathophysiology nodes and carries a
    REFUTE evidence item beside a SUPPORT one on the infection claim.

    The part that bears on this grouping: both candidate readings converge on
    cascade positions this grouping already has a member for — the pyogenic
    infection of impaired opsonisation, and the immune-complex and lupus
    phenotype of defective apoptotic-cell clearance, which is what Complement
    Component 2 Deficiency shows. Cascade position cannot adjudicate between
    them, because a lesion upstream of the shared C4b2a convertase is compatible
    with either. So the claim is not falsified; it is bounded. It holds at the
    granularity of the cascade arm rather than the individual protein, and it
    weakens where an arm has redundant inputs.

    What would settle it is a direct measurement rather than more members. No
    study has measured opsonophagocytosis or apoptotic-cell clearance in
    ficolin-3-deficient cells; both nodes rest on phenotype distribution across
    fewer than a dozen people, and that entry's own
    `ficolin3_infection_versus_autoimmunity` gap records the same absence from
    the disease side. Until then, a SUFFICIENT criterion for this grouping
    cannot key on phenotype for the lectin arm, whatever the coverage of the
    remaining Table 8 rows.

    The prediction this makes, and the thing to check when the alternative
    pathway is curated: factor B and factor D are non-redundant cascade
    components and should behave like C2 and the terminal components, whereas
    properdin is a stabiliser rather than a component and may qualify the claim
    the way ficolin-3 does. If it does, redundancy and position-in-cascade are
    two separate axes and this grouping's rationale should say so.