12-phase flow (12)

1. 1FRAME
   A thyroid nodule is a FINDING, not a diagnosis: ≈65% population prevalence, only ≈7–15% clinically significant cancer. The job is a sequential Bayesian funnel (TSH → US/TI-RADS → FNA/Bethesda → molecular → management) that spares the benign majority biopsy/surgery (Durante JAMA 2018; ATA 2015 Haugen)
   inputs: serum_tsh, diagnostic_neck_ultrasound
   advance: Scope set to the subacute/chronic outpatient nodule funnel with the TSH-first / autonomy-gate frame explicit
2. 2ENTRY
   Incidental nodule on neck imaging; palpable nodule/neck mass; suppressed TSH with a nodule; compressive/invasive neck symptoms; elevated-prior patient (family MEN2/familial PTC, childhood neck radiation) (Durante JAMA 2018; ATA 2015; ATA MTC 2015 Wells)
   inputs: serum_tsh
   advance: Engine entered via a recognised trigger
3. 3CONTEXT
   Capture age (<20 / >60 raise prior; AS progression age-stratified), sex (male ↑ prior), pregnancy (defers scan/surgery; FNA safe), childhood head/neck radiation (↑ prior), family MEN2/MTC or germline RET (medullary re-frame), prior thyroid disease and growth history (ATA 2015 Haugen; ATA MTC 2015 Wells; ATA pediatric 2015 Francis)
   inputs: age, sex, pregnancy_status, childhood_head_neck_radiation, family_men2_mtc_or_ret
   advance: Demographic + radiation + family + pregnancy prior-modifiers fully captured
4. 4RED_FLAGS
   Fixed/hard rapidly enlarging mass + hoarseness/vocal-cord palsy + fixed cervical adenopathy ± compressive airway symptoms = anaplastic / aggressive thyroid malignancy or invasive disease — this bypasses the routine funnel: urgent ENT + oncology, expedited imaging/biopsy, airway assessment (ATA 2015 Haugen)
   inputs: invasive_red_flag_features
   actions: workup.thyroid_nodule
   advance: Anaplastic / invasive presentation screened and escalated to urgent ENT/oncology if present
5. 5INITIAL_WORKUP
   TSH FIRST: suppressed TSH → radionuclide (I-123/Tc-99m) scan for an autonomously functioning ("hot") nodule — hot nodules are almost never malignant, FNA generally NOT indicated, route OUT to endo.hyperthyroidism.core.v1 (toxic adenoma / toxic MNG). Normal/high TSH → dedicated diagnostic neck US is the core risk tool (ATA 2015 Haugen; ATA 2016 hyperthyroidism Ross)
   inputs: serum_tsh, diagnostic_neck_ultrasound
   actions: panel.tsh, panel.thyroid, calc.tsh_reflex_interpretation, workup.thyroid_nodule
   advance: Functional status established (autonomy routed out) and diagnostic US obtained
6. 6BRANCHING_WORKUP
   ACR TI-RADS point score (composition/echogenicity/shape/margin/echogenic-foci) → TR1–TR5 → size-based FNA threshold (TR5 ≥1.0 cm, TR4 ≥1.5 cm, TR3 ≥2.5 cm; TR1/TR2 none; sub-threshold → interval US). CONDITIONAL DEPENDENCE #1 (FNA size cutoff ⟂ TI-RADS category): the size threshold is NOT a single number — it is conditional on TR category. CONDITIONAL DEPENDENCE #4 (sonographic-feature LR NON-INDEPENDENCE): individual feature LR+ are NOT multiplied naïvely — microcalcification LR+ ≈3.26, taller-than-wide LR+ ≈8.07, irregular margin LR+ ≈2.99 (Remonti meta-analysis 12,786 nodules, PMID 25747526; per-feature OR range 1.78–35.7); the features co-occur in true papillary cancers so the composite high-suspicion pattern (solid + marked hypoechoic + microcalcification + taller-than-wide + irregular/infiltrative margin) is de-rated for conditional dependence to a wired composite LR+ ≈20 (NOT the naïve product ≈78.6; bounded by Remonti per-feature OR ceiling 35.7 and Woliński taller-than-wide OR 13.7, PMID 24473342; Moon nondiagnostic-nodule TIRADS feature-count gradient, PMID 25133852). FNA → Bethesda I–VI. Indeterminate Bethesda III/IV → molecular rule-out (Afirma GSC / ThyroSeq v3). Suspicious + family MEN2/diarrhoea/indeterminate → serum calcitonin ± CEA + germline RET (medullary pivot) (Tessler JACR 2017; Cibas/Ali 2017; Remonti 2015; Woliński 2014; Steward/Patel/Livhits; Wells ATA MTC 2015)
   inputs: acr_ti_rads_features, nodule_size_mm, radionuclide_thyroid_scan, fna_bethesda_category, molecular_test_result, serum_calcitonin
   actions: calc.acr_ti_rads, calc.bethesda_thyroid, panel.tumor, workup.thyroid_nodule
   advance: TR level assigned, FNA decision resolved, Bethesda category (and molecular/calcitonin where indicated) obtained
7. 7DIFFERENTIAL
   MECE terminal split: benign colloid/hyperplastic nodule vs indeterminate follicular-patterned lesion vs papillary thyroid carcinoma vs MEDULLARY carcinoma (calcitonin/RET pivot — does NOT follow the papillary FNA pathway) vs anaplastic carcinoma vs primary thyroid lymphoma vs metastasis. The TSH→TI-RADS→Bethesda→molecular chain (conditional on age/radiation/family) is the decisive pivot (ATA 2015 Haugen; Cibas/Ali 2017; Wells ATA MTC 2015)
   inputs: fna_bethesda_category, molecular_test_result, serum_calcitonin
   advance: Terminal diagnosis (or indeterminate-needs-molecular state) assigned
8. 8RISK_STRATIFICATION
   Integrate TR level + Bethesda implied-risk + molecular result + the composite sonographic LR+ ≈20 (microcalcification + taller-than-wide + irregular-margin + solid-marked-hypoechoic high-suspicion pattern, conditional-dependence-corrected, Remonti PMID 25747526 / Woliński PMID 24473342). CONDITIONAL DEPENDENCE #2 (molecular PPV ⟂ Bethesda category × institutional prevalence): sens/NPV are stable but PPV moves materially with the prior — Steward ThyroSeq v3 PPV 66% @28% prevalence vs Patel Afirma GSC PPV 47% @24% prevalence; a positive molecular call is read against the institution-specific Bethesda-category prevalence, never as a fixed probability. CONDITIONAL DEPENDENCE #3 (malignancy prior ⟂ host context): the SAME US pattern carries a HIGHER posterior in a child (Francis), after childhood neck irradiation (Haugen), with a coexisting Hashimoto background (papillary OR 2.12, Lai PMID 28977955; primary thyroid lymphoma OR 12.92 in HT, Abbasgholizadeh PMID 34861884), and a LOWER posterior if the nodule is hot (Ross) or spongiform/pure-cyst (Durante). Add ATA recurrence-risk tier (low/intermediate/high) for confirmed DTC; low-risk T1aN0M0 PTMC → active-surveillance candidacy (Brito/MSK three-domain framework; age-stratified Kuma progression) (Tessler 2017; Cibas/Ali 2017; Remonti 2015; Woliński 2014; Lai 2017; Abbasgholizadeh 2021; Steward 2019; Patel 2018; Brito 2016; Ito 2014; ATA 2015)
   inputs: acr_ti_rads_features, fna_bethesda_category, molecular_test_result, age
   actions: calc.acr_ti_rads, calc.bethesda_thyroid
   advance: Composite malignancy probability + ATA recurrence tier + AS candidacy documented
9. 9TREATMENT
   Benign (Bethesda II / molecular-negative III-IV): interval US surveillance, no LT4-suppression for benign nodules (AACE/ACE/AME 2016 Gharib). Low-risk PTMC T1aN0M0: active surveillance is a guideline-endorsed alternative (immediate surgery if clinical N1/M1, RLN paralysis, tracheal protrusion, aggressive cytology, critical location). Bethesda V/VI or molecular-positive: lobectomy vs total thyroidectomy by size/risk; RAI risk-stratified to ATA recurrence risk; post-thyroidectomy levothyroxine replacement ± risk-stratified TSH-suppression → route post-op hypothyroidism to endo.hypothyroidism.core.v1. Medullary: total thyroidectomy + central neck dissection, RET-guided (ATA 2015 Haugen; Ito 2014; Sugitani/JAES 2020; Brito 2016; Wells ATA MTC 2015)
   inputs: fna_bethesda_category, molecular_test_result, serum_calcitonin, age
   actions: workup.thyroid_nodule
   advance: Surveillance / active-surveillance / surgery / RAI / LT4 plan documented
10. 10DISPOSITION
    Almost all managed outpatient (endocrinology ± endocrine surgery). Route suppressed-TSH/autonomous nodule → endo.hyperthyroidism.core.v1. Route post-thyroidectomy hypothyroidism → endo.hypothyroidism.core.v1. Anaplastic/invasive red-flag → urgent ENT/head-neck oncology. Pediatric / MEN2 → specialist multidisciplinary centre (ATA 2015 Haugen; ATA pediatric 2015 Francis; ATA MTC 2015 Wells)
    inputs: serum_tsh, invasive_red_flag_features
    advance: Setting set and autonomy / post-op-hypothyroid / oncology routing executed
11. 11MONITORING
    Benign: interval US by TR level (lower-risk longer intervals; re-FNA if significant growth or new suspicious features). Active surveillance: serial neck US (size + nodal) on a defined cadence; intervene on documented progression. Post-thyroidectomy DTC: serum thyroglobulin + anti-Tg antibody + TSH (titrated to a risk-stratified target — suppression for higher recurrence risk) + neck US; on prolonged TSH-suppression add periodic BMD + bone-turnover surveillance and export the suppression depth/duration/menopausal status to endo.osteoporosis.core.v1. Medullary: serial calcitonin/CEA (ATA 2015 Haugen; Ito 2014; Sugitani/JAES 2020; Wells ATA MTC 2015)
    inputs: acr_ti_rads_features, fna_bethesda_category, serum_calcitonin
    actions: panel.tsh, panel.thyroid, panel.tumor, panel.bone_turnover
    advance: Monitoring cadence individualised by TR level / Bethesda / post-op risk tier / medullary status
12. 12FOLLOWUP
    Benign stable: lengthening US intervals, discharge to primary care once stable. Active surveillance: lifelong periodic US with documented intervention triggers and shared decision-making. Treated DTC: long-term thyroglobulin/TSH/US surveillance per ATA recurrence tier, recurrence/return precautions (new neck mass, voice change). Pre-conception/pregnancy timing counselling. Pediatric/MEN2: lifelong specialist follow-up (ATA 2015 Haugen; Ito 2014; ATA pediatric 2015 Francis; ATA MTC 2015 Wells)
    inputs: pregnancy_status
    actions: panel.tsh, panel.tumor
    advance: Long-term surveillance / discharge / pre-conception plan booked