12-phase flow (12)

1. 1FRAME
   Hyperprolactinaemia is a FINDING, not a diagnosis: this engine is the systematic Bayesian sieve (confirm → physiologic → drug → hypothyroid → renal/hepatic → macroprolactin → hook → stalk-vs-tumour → idiopathic) that ROUTES into endo.prolactinoma.core.v1 only after everything else is excluded (Endo Soc 2011 Melmed PMID 21296991)
   inputs: serum_prolactin
   advance: Raised PRL framed as a finding against the fixed exclusion ladder, not pre-labelled a prolactinoma
2. 2ENTRY
   Incidental/screening high PRL; woman with galactorrhoea/oligo-amenorrhoea/infertility; man with low libido/ED/gynaecomastia; prolactin-raising drug started; incidental sellar mass (Endo Soc 2011 Melmed; Pituitary Soc 2023 Petersenn PMID 37670148)
   inputs: serum_prolactin, sex
   advance: Engine entered via a recognised trigger
3. 3CONTEXT
   Capture sex/reproductive intent, pregnancy/lactation/nipple-stimulation, FULL medication review (antipsychotic — risperidone/paliperidone/amisulpride potent vs aripiprazole sparing; metoclopramide/domperidone; SSRI/TCA; opioid; verapamil; oestrogen; H2), chest-wall lesions, CKD/cirrhosis, and any psychiatric diagnosis that mandates psychiatry-shared deprescribing (Endo Soc 2011 Melmed)
   inputs: sex, pregnancy_status, current_meds, chest_wall_or_renal_hepatic_history, psychiatric_diagnosis
   advance: Physiologic + pharmacologic + systemic + psychiatric context fully captured
4. 4RED_FLAGS
   A LARGE/GIANT sellar mass with only mildly elevated PRL is HOOK EFFECT until a 1:100 serial dilution proves otherwise — undiluted assay can misroute a giant macroprolactinoma to surgery (case literature PRL 31 → 280,000 ng/mL on dilution, Barkan PMID 9574657). Macroadenoma with acute headache + visual field loss / ophthalmoplegia → screen for apoplexy and route OUT to the pituitary engines (Pituitary Soc 2023 Petersenn PMID 37670148)
   inputs: serum_prolactin, serum_prolactin_diluted, pituitary_mri
   actions: workup.hyperprolactinemia
   advance: Hook effect excluded by dilution in any large mass; acute mass-effect / apoplexy screened and routed out if present
5. 5INITIAL_WORKUP
   Confirm with a single RESTING non-stressed serum PRL (repeat without venepuncture stress if borderline — reclassifies stress artifact; dynamic testing NOT recommended). Then β-hCG (pregnancy halts the sieve), TSH/FT4 (primary hypothyroidism — reversible), creatinine/eGFR (CKD; PRL ∝ creatinine r≈0.61), LFT (cirrhosis). Medication reconciliation drives the drug-induced rung (Endo Soc 2011 Melmed PMID 21296991; Yavuz CKD PMID 16268803)
   inputs: serum_prolactin, serum_prolactin_repeat_resting, beta_hcg, tsh_ft4, creatinine_egfr, lft
   actions: panel.hormone, panel.tsh, panel.thyroid, panel.cmp, workup.hyperprolactinemia
   advance: Elevation confirmed resting/non-stressed AND physiologic/hypothyroid/renal-hepatic causes excluded or treated
6. 6BRANCHING_WORKUP
   Asymptomatic high PRL → macroprolactin PEG screen FIRST (recovery <40% = inert macroprolactin, ~21.5% of PRL >100 cohorts — reassure, no imaging; Kalsi PMID 30269265, Yu PMID 32597541). Large/giant mass with mild PRL → 1:100 dilution for hook effect. True hyperprolactinaemia after ALL exclusions → dedicated pituitary MRI; non-lactotroph mass with mild PRL → stalk-effect assessment and route to hypopituitarism (Endo Soc 2011 Melmed; Pituitary Soc 2023 Petersenn PMID 37670148)
   inputs: macroprolactin_peg, serum_prolactin_diluted, pituitary_mri
   actions: panel.hormone, workup.hyperprolactinemia
   advance: Macroprolactin/hook resolved; MRI obtained only when indicated; mass characterised stalk-effect vs lactotroph
7. 7DIFFERENTIAL
   MECE terminal split with the named pivot + discriminating LR per pair: physiologic (β-hCG) vs drug-induced (med review/washout; aripiprazole-sparing vs risperidone-raising) vs primary hypothyroidism (TSH/FT4, reversible — route endo.hypothyroidism.core.v1) vs hyperthyroidism/TSH-secreting tumour (thyrotoxicosis + galactorrhoea, raised PRL reverses with the thyroid lesion — Kamoi PMID 3934894 → endo.hyperthyroidism.core.v1) vs PCOS-with-hyperprolactinaemia (oligomenorrhoea overlap; 11.4–11.6% of PCOS hyperprolactinaemic, PRL band lower 52.9–85.2 ng/mL — Kim PMID 37057215, Kyritsi PMID 29845629 → endo.pcos.core.v1) vs CKD/cirrhosis (creatinine∝PRL r=0.609) vs macroprolactinaemia (PEG recovery <40%) vs hook effect (1:100 dilution; 5.8–14.2% of macroadenomas) vs stalk effect (MRI + DA trial: PRL falls but mass does NOT shrink — Kawaguchi border-zone false-positive 18.9% PMID 25142896) vs true prolactinoma (PRL ≥85.2 ng/mL LR+ ≈28 / PRL 62.45 LR+ 16.6; >250 + macroadenoma; PRL falls AND mass shrinks) vs idiopathic (exclusion + surveillance) (Endo Soc 2011 Melmed; Pituitary Soc 2023 Petersenn)
   inputs: serum_prolactin, macroprolactin_peg, pituitary_mri
   advance: Terminal cause assigned via the discriminating test for that rung
8. 8RISK_STRATIFICATION
   WIRED PRL-magnitude likelihood ratios for prolactinoma (post-confirm, secondary causes excluded): PRL ≥85.2 ng/mL — sens 0.77, spec 1.00, AUC 0.91 → LR+ ≈ 28 (Haldane-corrected from 0/18 false-positives), the STRONGEST wired band ≥20 (Kyritsi PMID 29845629); PRL 62.45 ng/mL prolactinoma-vs-NFA — sens 0.857, spec 0.948 → LR+ 16.6, LR− 0.15 (Wright PMID 33966173); PRL/tumour-volume ratio 21.62 (ng/mL)/cm³ — sens 1.00, spec 0.83 → LR− ≈ 0 (rules OUT prolactinoma when low); PRL 204 µg/L micro-vs-macro — sens 0.891, spec 0.932 → LR+ 13.1, LR− 0.12, PRL∝size r=0.750 (Leca PMID 33963239); border-zone 90–200 ng/mL false-positive 18.9% — magnitude alone UNSAFE here (Kawaguchi PMID 25142896). CONDITIONAL DEPENDENCE (modelled, not naive-independent): (i) the magnitude LR is conditional on ASSAY state — a hook-saturated assay INVERTS it (giant adenoma can read 11.3 → 5795 µg/L on 1:100 dilution; hook in 5.8–14.2% of macroadenomas, 100% male/mean 51 mm in the hook group — Petakov PMID 9591215, St-Jean PMID 8729527); (ii) macroprolactin lowers the symptomatic-prolactinoma posterior even when total PRL is high (PEG BEFORE magnitude is trusted — Kalsi PMID 30269265, Yu PMID 32597541); (iii) the PRL-band LR is conditional on the PCOS/oligomenorrhoea population (11.4–11.6% of PCOS are hyperprolactinaemic; PCOS cohort cut-points 52.9–85.2 ng/mL — Kim PMID 37057215, Kyritsi PMID 29845629 → endo.pcos.core.v1); (iv) drug-history LR and magnitude LR are linked — on a potent antipsychotic a 25–100 ng/mL value is explained by the drug and does NOT independently raise the tumour posterior until persistence after washout re-enables magnitude (Ma PMID 39411853). T_test (when to MRI): confirmed PERSISTENT non-physiologic non-drug elevation after thyroid/renal/macroprolactin exclusion. Routing threshold to endo.prolactinoma.core.v1: confirmed true hyperprolactinaemia + concordant sellar lesion + secondary causes excluded — carry magnitude band + imaging state (PRL ≥290 µg/L predicts long-term DA-dependence HR 23.9 — Andereggen PMID 39904877) (Endo Soc 2011 Melmed; Pituitary Soc 2023 Petersenn PMID 37670148)
   inputs: serum_prolactin, macroprolactin_peg, pituitary_mri
   advance: Magnitude band + symptom burden documented; MRI/routing thresholds applied
9. 9TREATMENT
   CAUSE-DIRECTED (largely deprescribing/workup, not a drug engine): (1) physiologic → reassure/observe; (2) drug-induced → stop/swap the culprit WITH psychiatry (NEVER unilaterally) — switch/adjunct aripiprazole (prolactin-sparing: adjunctive aripiprazole lowered PRL 58% vs +22% placebo, normalised 46%, NNT 2, Raghuthaman PMID 27703744); (3) primary hypothyroidism → levothyroxine (reverses PRL and thyrotroph hyperplasia, Fernández-Real PMID 9534344); (4) macroprolactinaemia → REASSURE, do NOT treat (inert); (5) CKD/cirrhosis → treat underlying; (6) true symptomatic prolactinoma after exclusions → ROUTE to endo.prolactinoma.core.v1 for the dopamine-agonist ladder (cabergoline preferred there: normalisation 87.4% vs bromocriptine 41.4%, Arduc PMID 25421155) — this engine does NOT initiate DA (Endo Soc 2011 Melmed PMID 21296991; Pituitary Soc 2023 Petersenn PMID 37670148)
   inputs: current_meds, tsh_ft4, macroprolactin_peg, serum_prolactin
   actions: workup.hyperprolactinemia
   advance: Reversible cause treated/reassured, OR true symptomatic prolactinoma routed to endo.prolactinoma.core.v1 with magnitude + imaging carryover
10. 10DISPOSITION
    Outpatient endocrinology for almost all. Cross-engine routing by engine_id with carryover: true symptomatic prolactinoma → endo.prolactinoma.core.v1 (DA ladder owned there; carry PRL magnitude band + post-dilution value + MRI state — PRL ≥290 µg/L flags likely long-term DA-dependence, Andereggen PMID 39904877); stalk-effect / non-lactotroph mass / hypopituitarism → endo.hypopituitarism.core.v1 (carry mass size + which axes screened); primary hypothyroidism as the cause → endo.hypothyroidism.core.v1 (carry TSH/FT4 + that PRL is expected to reverse on levothyroxine); thyrotoxicosis / TSH-secreting tumour with raised PRL → endo.hyperthyroidism.core.v1 (Kamoi PMID 3934894 — carry thyroid panel + galactorrhoea); PCOS-with-hyperprolactinaemia / oligomenorrhoea overlap → endo.pcos.core.v1 (carry PRL band + that the PCOS-cohort cut-point is lower, Kyritsi PMID 29845629). Bidirectional intent: each sibling can route a raised-PRL finding BACK into this sieve. Co-manage drug-induced disease with psychiatry (Endo Soc 2011 Melmed; Pituitary Soc 2023 Petersenn)
    advance: Setting set and cross-engine routing executed with carryover
11. 11MONITORING
    After a reversible cause is removed, RECHECK serum PRL (drug washout / after levothyroxine to target / after delivery-lactation ends) before any imaging — re-imaging is only for persistent unexplained elevation. Macroprolactin: no follow-up imaging needed. Idiopathic: periodic PRL ± symptom review (34% normalise; 1/41 develop a tumour — Martin PMID 3980670) (Endo Soc 2011 Melmed PMID 21296991)
    inputs: serum_prolactin, macroprolactin_peg
    actions: panel.hormone
    advance: Post-intervention PRL trajectory documented; imaging reserved for persistence
12. 12FOLLOWUP
    Idiopathic hyperprolactinaemia surveillance (most stable/resolve — Martin PMID 3980670); preconception counselling (prolactinoma pregnancy risk owned by endo.prolactinoma.core.v1 — micro ~1% vs macro 23% growth if DA stopped, Molitch PMID 10649822); reinforce that the antipsychotic is managed with psychiatry; return precautions (visual change, severe headache, new galactorrhoea) (Endo Soc 2011 Melmed; Pituitary Soc 2023 Petersenn)
    advance: Long-term surveillance / preconception / psychiatry-shared plan booked