AI Research Rounds – The Facial Artery Musculomucosal (FAMM) Flap — Anatomical Basis and Technique

AI Research Rounds: The Facial Artery Musculomucosal (FAMM) Flap — Anatomical Basis and Technique

Research Date: 2026-09-25
Research Question: What is the anatomical (vascular and muscular) basis of the facial artery musculomucosal (FAMM) flap, and what are the technique details governing its design, elevation, insetting and donor-site management?
Method: AI-assisted systematic literature review (Ralph Loop + Multi-Agent architecture, ai-research-rounds v2.0)
Report Version: 1.0
Intended audience: Oral and maxillofacial surgeons, head and neck surgeons, reconstructive surgeons, surgical residents

Executive Summary

Clinical Bottom Line

The FAMM flap is a facially-based composite myomucosal flap whose reliability depends far more on its venous drainage than on its arterial input — it behaves clinically as an arterialized flap with a substantial random submucosal-plexus component, not as a classic end-artery axial flap, and every hard design rule that governs it (≥2 cm pedicle base, ≥1 cm behind the commissure, ≥5 mm below Stensen’s duct orifice, avoid skeletonising onto the artery alone) follows from that single fact.

Key Numbers

Metric Value 95% CI GRADE
Pooled flap success (all sites, 407 flaps) 99.47% 98.26–100.00 ++○○ Low
Total flap failure 1.7% (7/407) — ++○○ Low
Partial flap necrosis 5.7% (23/407) — ++○○ Low
Wound dehiscence 4.4% (18/407) — ++○○ Low
Venous congestion (pooled) 3.2% (13/407) — +○○○ Very Low ⚠
Pooled any complication 30.18% 16.97–43.38 +○○○ Very Low ⚠
Complication rate, tunnelized vs conventional FAMM (n=141) 37% vs 7% — ++○○ Low
5-yr progression-free survival, BMF vs free flap (n=615) HR 0.87 (non-inferiority met) 0.43–1.78 +++○ Moderate
Facial artery diameter at mandibular border 2.14 mm 1.74–2.54 +○○○ Very Low ⚠

⚠ The two flagged rows are flagged deliberately. Pooled venous congestion of 3.2% is a statistical artefact of pooling: congestion is reported in essentially every small FAMM series (8/8, 17/17) but is self-limiting over 24 h–1 week without intervention, so pooled estimates under-count the true clinical experience by an unknown margin. The facial-artery diameter has an I² of 99.2% across studies — a single number does not exist, and any quoted diameter must carry its modality and measurement level.

Scope Statement

What This Is

  • An AI-assisted literature synthesis using systematic methodology, with external, auditable state (research plan, append-only findings log, independent citation verification).
  • A rapid evidence review with a transparent, reproducible process — every identifier in this report was independently re-resolved against PubMed/Europe PMC by a separate agent from the one that collected it.
  • A clinical decision-support and orientation document, with explicit flagging of what is measured, what is conventional practice, and what is simply unknown.
  • A data-integrity audit: this topic contains a retracted paper that has been absorbed into secondary literature, and this report identifies it and quantifies its influence.

What This Is NOT

  • A formally registered systematic review. (The 2025 JPRAS meta-analysis this report cites is registered — CRD42024529989.)
  • A meta-analysis. No new pooled statistics were computed by this review; pooled figures are quoted from an existing meta-analysis.
  • A clinical practice guideline.
  • Peer-reviewed.
  • A replacement for clinical judgment or institutional protocol.

Main Findings

Finding 1 — The vascular basis: what the flap actually is

Conclusion: The FAMM is a composite myomucosal flap comprising buccal mucosa, submucosa, and a portion of buccinator, carried on the facial artery and its buccal branches. It is not a simple end-artery axial flap: its venous outflow depends largely on a submucosal venous plexus, and this is the anatomical reason for its characteristic congestion behaviour.

The facial artery. From the external carotid, the artery runs a cervical/submandibular course (medial to sternocleidomastoid, behind the posterior belly of digastric, grooving the submandibular gland, giving the submental artery), then crosses the inferior mandibular border at the antegonial notch at a mean 49.8° to the mandible (CTA, 255 facial arteries; PMID 30721996), giving off a premasseteric branch in 51.4% (Ø 0.8 mm) (PMID 35811399). It then ascends as the buccal segment — the FAMM-carrying segment — before becoming labial and angular.

Which branches supply the flap. The clinically relevant supply to buccinator is the buccal branch group: a posterior buccal branch, 1–3 inferior buccal branches supplying the lower half of the muscle, and 3–5 anterior buccal branches supplying its anterior half (PMID 10597674; PMID 38745516). Importantly, the posterior half of buccinator is also supplied by the buccal artery, a branch of the internal maxillary artery, which anastomoses with the posterior vestibular branch of the facial artery.

Nomenclature correction — a real finding of this review. "Inferior buccinator artery" is not a verified anatomical entity. No primary anatomical study retrieved by this review uses the term; exhaustive searching returned no source. The verified structures are inferior buccal branches of the facial artery and the buccal artery (internal maxillary origin). This phrase should not be used in clinical or academic writing. (PMID 10597674; PMID 38745516; searches recorded in raw-search-results/T1_*)

The facial artery diameter is not a single number. Retrieved values span a three-fold range and the heterogeneity is structural, not incidental:

Measurement Value Modality / population Source
Pooled, mandibular border 2.14 mm (95% CI 1.74–2.54) Doppler, 7 studies / 649 FAs, I² = 99.2% PMID 40874402
Mandibular border 1.56 ± 0.36 mm Doppler PMID 37327736
Mandibular border 1.26 / 1.04 mm CE-CT PMID 34448258
Main trunk 3.05 ± 0.69 mm Cadaveric dissection PMID 41366554
Cheilion → lateral nasal ala 1.71 → 1.46 mm Doppler PMID 40874402

Sex and side asymmetry is documented (male > female at the mandibular border; right > left) but no retrieved source addresses the implication for harvest laterality. (PMID 40874402; PMID 37327736; PMID 41366554)

Buccinator. Origin from the mandible and maxillary alveolar processes and the pterygomandibular raphe, inserting into orbicularis oris at the modiolus. Its main neurovascular bundle runs approximately 1 cm anterior to the pterygomandibular raphe (PMID 41837171). A previously undescribed deepest oblique layer immediately beneath the parotid duct was characterized in 2025 (50 hemifaces; PMID 41491248), and an inferior "fourth band" variant occurs in 35% (PMID 21239921). Motor supply is CN VII; sensory supply is the long buccal nerve (CN V3), which enters posterolaterally — with pronounced CN V–VII sensorimotor anastomoses (PMID 39508330).

Pedicle dimensions (with an important caveat). The figures most often quoted for "FAMM pedicle length" — 4.8 ± 0.6 cm (right) / 4.9 ± 0.6 cm (left), extensible to 7.8 ± 0.7 / 7.7 ± 0.6 cm by dissecting the facial artery through the submandibular gland, with 2–3 perforators (13 fresh cadavers / 26 sides; PMID 24035128) — are measured on the inferiorly based buccinator myomucosal island flap, which is a different pedicle. Likewise the median maximum pedicle of 106 mm (PMID 38894868) is for the contralateral island FAMM. The usable pedicle length of the standard pedicled FAMM has never been directly measured in the retrieved literature. This substitution is rarely flagged in the reviews that recycle these numbers, and it is the single most common unacknowledged data-transplant in this literature.

Distance from the oral commissure to the facial artery: 1.45 ± 0.26 cm (cadaveric; PMID 41889531) or 21.3 ± 2.2 mm (intraoral; PMID 39610359) — the latter was measured for microvascular anastomosis planning, not FAMM harvest, and is the closest available surrogate.

Evidence: Predominantly cadaveric and imaging anatomical studies, n = 8–28 specimens, plus one CT-angiographic series of 255 arteries. GRADE: ++○○ Low for descriptive anatomy (consistent across independent dissection series), +○○○ Very Low for any single quantitative diameter.

Clinical Implication: Size the pedicle expectation from the dissection, not from the literature — and if a specific calibre matters (e.g. planning a free or supermicrosurgical variant), measure it in your own patient by Doppler or CTA, because the published spread is three-fold.

Finding 2 — The axial-versus-arterialized controversy: the most consequential unresolved question

Conclusion: Pribaz designed the flap as an axial musculomucosal flap, and essentially every review since has called it axial. The only dedicated cadaveric study of its vascular pattern concluded the opposite in part — that it is "more an arterialized flap than an axial-pattern flap." These two positions have never been tested against each other, and the difference is not academic: it predicts the flap’s congestion behaviour and the failure mode of artery-only island designs.

The evidence, both sides:

Position Source Nature
Axial musculomucosal flap on the facial artery, "combining the principles of nasolabial and buccal mucosal flaps" Pribaz 1992, PMID 1513887 Primary — original description, 18 flaps / 15 patients
Repeatedly described as axial-pattern PMID 40940987; PMID 41837171; PMID 35196693 Reviews / technique papers (secondary)
"The flap is more an arterialized flap than an axial-pattern flap" Dupoirieux 1999, PMID 10203218 Primary — cadaveric, latex injection, 10 cadavers

Clinical correlates for each reading:

Favouring "arterialized / fragile":

  • Artery-only island flaps in a dog split-mouth model: all flaps with the facial vein preserved survived; all with the vein ligated and divided necrosed (100%) — the authors concluded an island FAMM on a skeletonised facial artery is "not biologically acceptable for clinical use" (PMID 27648649).
  • An arterialized tunnelised island series (a-FAMMIF, 5 tongue-cancer patients) reported 1 total and 2 partial necroses; authors call the variant unreliable and recommend avoiding tunnelling/islanding when the vein is excluded (PMID 29771835).
  • Explicit caution that harvesting island flaps on the artery alone is "fraught with a very high rate of flap loss if the artery is dissected to its origin" (PMID 38440437).
  • The n=141 comparative cohort: vein-omitting tunnelised flaps had total necrosis 11% vs 2% and partial necrosis 19% vs 2% (p=0.048 / p=0.003) (PMID 42026291).

Favouring "robust axial":

  • Pooled success 99.47% and pooled venous congestion of only 3.2% across 407 flaps (PMID 39914242).

GRADE: +○○○ Very Low — this is conflicting evidence between a primary cadaveric study and a body of secondary literature, never directly tested. This review deliberately does not resolve it.

Clinical Implication: Design and dissect as if the flap needed its venous outflow preserved by tissue, not just by an artery. Concretely: include the facial vein where possible, or keep a ≥2 cm soft-tissue base; do not skeletonise onto the artery alone; avoid unnecessary tunnelling; and treat intraoperative venous congestion as an expected finding rather than a technical error.

Finding 3 — Venous drainage is the governing constraint, not arterial inflow

Conclusion: Every practical design limit on this flap traces back to venous drainage.

Anatomical facts:

  • At the base of the mandible the facial artery was anterior to the facial vein in 100% of cases, mean separation 6.2 mm (range 0–15 mm) (90 halved viscerocrania, latex injection; PMID 38792988). By high-resolution ultrasound the facial vein lay lateral to the artery in 100% right / 95% left, at 5.6 / 4.4 mm (41 volunteers; PMID 37691033).
  • In facial-artery perforator flaps for intraoral reconstruction, venous drainage was classified into three types: type A → internal jugular (47%), type B → external jugular (37%), type C → anterior jugular (16%) — and type C was a risk factor for flap survival, with venous congestion in those cases. Mean artery-to-vein distance was 2.79 ± 0.51 mm at the pedicle and 10.24 ± 0.70 mm at the flap tip (PMID 27998737).

The operating rule, and its epistemic status. The facial vein is usually not included in the flap; drainage is held to depend on a submucosal venous plexus, which is said to require a pedicle at least 2 cm wide (PMID 40940987; PMID 34440964; PMID 40104415; PMID 41837171). This 2 cm threshold appears to be clinical empiricism propagated through technique papers and reviews rather than a measured anatomical threshold — no retrieved study measured it. The submucosal venous plexus itself has no quantitative anatomical description at all (no diameter, density, or correlation with facial vein flow).

GRADE: ++○○ Low for the artery–vein topographic relationships (multiple concordant primary dissection and imaging studies); +○○○ Very Low for the 2 cm rule (empirical, unreplicated); no evidence grade assignable for the venous plexus itself.

Clinical Implication: Treat ≥2 cm as a conservative practical floor rather than a proven boundary, and prefer including the facial vein when the design permits — it is the one intervention with consistent supporting data.

Finding 4 — The design envelope

Conclusion: Length 7–9 cm typical (maximum reported 10 cm); width 2–3 cm; three obligatory anatomical landmarks; and the width limit is set by donor-site closure and the parotid duct, never by the pedicle.

Parameter Value Attribution
Length, typical 7–9 cm PMID 34484602; secondary attributions
Length, maximum reported 10 cm PMID 27330928 (extended retrograde FAMM + masseteric fascia, n=1)
Pedicle-only length (anatomically measured) 4.8–4.9 → 7.7–7.8 cm PMID 24035128 — different flap (BMMF island) ⚠
Width, typical working range 2–3 cm PMID 40940987; PMID 34440964
Width permitting primary donor closure ≤2.5 cm (conservative) to ≤3 cm (accepted) PMID 24163549; PMID 37602111; PMID 40940987
Width, maximum reported paddle 4 cm (55 × 40 mm) PMID 41890649
Largest measured flap 70 × 50 mm without duct translocation PMID 38894868
Minimum soft-tissue base if the facial vein is excluded 2 cm PMID 37602111; PMID 40940987

The three obligatory landmarks (highest-consensus findings in this entire review):

  1. Anterior limit: 1 cm posterior to the oral commissure. The single most consistently reported landmark — stated in nine independent sources (PMID 36470675, PMID 37602111, PMID 34440964, PMID 42022315, PMID 39957871, PMID 38894868, PMID 40104415, PMID 40940987, PMID 40142773). Rationale, quoted: "A distance of almost 1 cm should be maintained from the labial commissure anteriorly… to avoid fibrotic retraction and trismus during the healing process" (PMID 36470675); exceeding it "leads to an anaesthetic scar retraction of the commissure" (PMID 40142773).
  2. Superior limit: at least 5 mm (0.5 cm) below the parotid duct orifice. "The upper edge was positioned at least 5 mm away from the orifice of the parotid duct" (PMID 42022315); "the upper margin is at least 0.5 cm below the Stensen’s duct papillae, to avoid the risk of an obstructive sialoadenitis" (PMID 40142773); concordant in PMID 38894868 and PMID 34440964.
  3. Posterior limit: the pterygomandibular raphe, where the vascular pedicle enters the flap (PMID 36470675; PMID 42022315; PMID 40142773). For the inferiorly based design the duct orifice itself caps the posterior/central width (PMID 40940987; PMID 39957871).

Additional constraint: only about half of the buccinator is usable, given the 0.5–1 cm duct margin (PMID 24163549). The conventional flap is harvested centred over the facial artery in an oblique axis extending from the retromolar trigone toward the gingivolabial sulcus near the alar margin (PMID 37741079).

Width is not limited by vascularity. No retrieved source reports pedicle failure from excess flap width. The limit is donor-site closure (trismus) and duct position — a distinction the literature generally blurs.

Orientation — antegrade versus retrograde:

Inferiorly based (antegrade) Superiorly based (retrograde)
Pedicle Facial artery at the mandibular border Angular artery / distal FA / superior labial artery (sources unreconciled)
Reach Floor of mouth, lateral/ventral tongue, tongue base, buccal mucosa, retromolar trigone, lower lip, tonsillar fossa, lateral pharynx Hard and soft palate, alveolus, antrum, nasal floor/septum/lining, upper lip, orbit/conjunctiva, nasopharynx, skull base
Reliability More reliable — significantly less necrosis and dehiscence than buccinator flaps in n=134 cleft patients (PMID 40910620) Depends on facial↔intracranial anastomoses via the angular artery, absent in 5–10% (PMID 24163549)
Reported failure — 16 flaps in 14 children: 2 partial + 2 complete failures (PMID 17760492)

GRADE: ++○○ Low for the landmark rules (nine concordant sources, but all observational/technique-level), +○○○ Very Low for the dimension ranges (heterogeneous, partly secondary attributions).

Finding 5 — Operative technique, step by step

Conclusion: The harvest is reproducible and has a small number of genuinely critical steps; the technique literature is now good on landmarks and planes but has explicit, identifiable holes (notably Stensen’s duct transposition detail and anti-congestion management).

Preoperative

  • Defect selection: 4–10 cm, ideally a single functional unit; assess dentition, level I nodal status, and any prior cheek surgery (PMID 36470675).
  • Doppler marking of the facial artery from the mandibular border to the nasal ala is the standard adjunct (PMID 36470675; PMID 39957871; PMID 40104415; PMID 40142773) — though one source calls Doppler unnecessary (PMID 40940987). The duct orifice is identified and marked preoperatively (PMID 37602111; PMID 39957871). No retrieved source describes CTA-based mapping for the conventional FAMM — and no source defines a preoperative mouth-opening threshold for eligibility. (One adjacent flap, the facial artery perforator myomucosal flap / FAPMF, does use CT angiography plus Doppler: PMID 40948003, 8 patients, Chinese-language; see Addendum A.)
  • Airway: nasal/nasotracheal intubation in most reports (PMID 36470675; PMID 40142773; PMID 37741079). Tracheostomy is not routine — 0/27 in the islanded-FAMM arm vs 14 free-flap cases in one comparative series (PMID 31980317). In the n=59 comparative cohort, tracheostomy was performed in 3/41 FAMM (7.3%) vs 3/18 free flap (16.7%) (PMID 41464662, full text verified).
  • Antibiotic practice is heterogeneous: amoxicillin/clavulanate 875/125 mg × 72 h (PMID 36470675) versus cefuroxime plus metronidazole × 5–7 days (PMID 42026291).

Marking — the three landmarks above, plus the facial artery marked as a surface line.

Incision and elevation. Two described strategies for finding the pedicle:

  • (A) Distal-first: cut the distal flap through mucosa, submucosa and buccinator; once the facial artery is located it is clipped and sectioned distally (PMID 37602111).
  • (B) Anterior-first: incise 1 cm behind the oral commissure through mucosa, submucosa and orbicularis oris to locate the superior labial artery, then trace it retrograde to the facial artery (PMID 37602111; PMID 39957871). A Y-shaped junction of the facial, superior labial and lateral nasal arteries is often encountered — and the superior labial artery should be ligated only once the angular artery has been identified (PMID 37602111).

Critical technical points:

  • Flap composition: mucosa, submucosa, part of buccinator, plus a small portion of orbicularis oris at the commissure (PMID 36470675; PMID 40142773; PMID 39957871).
  • Raised off the buccopharyngeal fascia, which must be preserved to prevent buccal fat-pad herniation and facial-nerve branch injury (PMID 40142773; PMID 34440964).
  • The facial artery must remain connected to surrounding tissue throughout its length; collateral vessels are clipped distal→proximal as dissection proceeds (PMID 37602111).
  • The arterial stump is sutured to the buccinator muscle and mucosal edge to prevent the artery being torn from the flap (PMID 34440964).
  • The flap must be centralised over the artery; failure to do so is a recognised cause of partial and complete necrosis (PMID 37602111).
  • Venous drainage: "Venous drainage relies on a submucosal plexus, and hence including the facial vein is not necessary. However, if the facial vein is not included, a 2-cm soft tissue base should be maintained for adequate venous drainage" (PMID 37602111). Where included, the vein is ligated, divided and sewn into the upper flap edge (PMID 34440964).
  • Posterior dissection proceeds to the raphe following the buccal artery, with the buccal artery, vein and buccal nerve arising at the posteroinferior buccinator (PMID 40142773).
  • Vasoconstrictor infiltration of the flap margins is described in only one retrieved operative paper (local anaesthetic + 1% adrenaline, with 2/0 silk stay sutures at the commissure and lips; PMID 40142773). This should not be presented as universal practice.

Stensen’s (parotid) duct. The duct orifice is displayed and excluded from the flap design (PMID 36470675 and others). The only retrieved description of protective cannulation uses a 22-gauge angio-catheter (PMID 37741079). To gain width, the distal duct may be divided and transposed, described as analogous to the Bozola technique — the existence and purpose of this manoeuvre is doubly sourced (PMID 34440964; PMID 40940987) but the operative detail is not retrievable (Bozola 1989 is paywalled; only the abstract was obtained). Duct re-anastomosis or re-implantation is not described anywhere in the retrieved literature. Consequence of crowding the duct: obstructive sialoadenitis (PMID 40142773). In a 22-flap buccinator donor-site morbidity series, duct injury was an explicitly measured outcome and did not occur (PMID 26786664). Underpinning anatomy: the parotid–masseteric and buccopharyngeal fasciae are firmly attached to buccinator and stabilise the duct against kinking or occlusion during mastication (PMID 41491248).

Structures at risk. The buccal artery runs very close to the buccal nerve at the posteroinferior buccinator (PMID 2748738; PMID 40142773). The marginal mandibular nerve should be identified and exposed before vessel dissection and kept under vision throughout (PMID 42022315); islanded flaps are tunnelled medial to it (PMID 37602111). Persistent marginal-mandibular palsy: 4% conventional (4/107) vs 11% tunnelized (3/27), p=0.15 (PMID 42026291). Using a high Doppler signal check on the facial artery is specifically recommended in patients who have had neck surgery or radiotherapy (PMID 37602111).

Insetting. The conventional design uses Duranceau’s single-stage modification — prolongation of the anterior incision with direct inset over the alveolar ridge (PMID 22109756; PMID 42026291). Islanded designs require two tunnels (external mandibular → neck, internal mandibular → floor of mouth), bluntly enlarged, kept close to the medial mandible (PMID 36470675; PMID 38894868). Alternatives include a lateral maxillary sinus window with 180° rotation (PMID 42250969), a Caldwell-Luc approach (PMID 37602111), and transposition via the mylohyoid, tonsillar pillar or nasal floor. The pedicle must not be folded or twisted; tension is relieved by extending the dissection at the base, not by stretching the pedicle (PMID 40142773) — a kinked pedicle caused complete flap loss in one report (PMID 39957871). Suture: Vicryl 3/0 (PMID 40142773); suturing to thin mucosa predisposes to dehiscence (PMID 37602111).

Donor site. Primary closure of muscle and mucosa if width <2.5 cm (PMID 24163549) or <3 cm (PMID 37602111); primary closure is not recommended for wider defects because of trismus. The commonest alternative is buccal fat pad advancement, which epithelialises in about 3 weeks, with iodoform gauze removed at 2 weeks (PMID 38894868). Skin graft, secondary granulation and masseter flap are also described. Consequences: trismus is the most reported donor-site complication. In the t-FAMMIF cohort (n=23), maximum interincisal opening fell from 44.00 ± 1.05 mm preoperatively to 27.11 ± 2.05 mm at 1 month (p=0.000, moderate trismus), then recovered to 37.33 ± 2.84 mm at 3 months (p=0.055), 40.33 ± 1.36 mm at 6 months (p=0.056) and 43.11 ± 1.02 mm at 12 months (p=0.452, no longer significantly different from preoperative) — with mouth-opening exercises begun at 2 weeks, 30–60 min/day, until MIO exceeds 35 mm (PMID 42022315, full text verified). In oronasal fistula repair, 86% scar contracture (6/7) with 4 needing release has been reported (PMID 41837171, very small series).

Adjacent-flap note (do not conflate). The only randomised trial in this anatomical region is PMID 42578092 — a prospective RCT (n=40) of posterior-based buccinator myomucosal flaps, i.e. a buccal-artery-pedicle flap, not a FAMM. It found tunnelling superior for oral opening at 6 months (39 mm vs 33 mm, p=0.008) and for hospital stay (2.0 vs 4.0 days, p=0.002) despite higher venous congestion (25% vs 5%) and trismus (15% vs 5%), with comparable flap survival (95% vs 100%, p=1.000). This is the best available randomised evidence on tunnelling in this reconstructive family, but it does not directly answer the FAMM tunnelling question, because the pedicle is a different vessel.

Postoperative care and monitoring. Nasogastric feeding 48 h then soft diet for one month (PMID 36470675; PMID 40142773), or 10–14 days until fat-pad epithelialisation (PMID 34440964). Venous congestion is expected and managed expectantly — resolving in 24–48 h (PMID 37602111), within a week without intervention (PMID 40104415), or 3 days (PMID 39957871). No leech, anticoagulation or pharmacological protocol for congestion is described in any retrieved source. Mouth-opening exercises start at 2 weeks, 30–60 min/day until MIO exceeds 35 mm (PMID 42022315). Pedicle division in staged designs at 3 weeks (PMID 39957871; PMID 40104415) or 3–4 weeks under local anaesthesia (PMID 40142773).

Explicit technique gaps (no source in the retrieved literature): CTA/CT-angiographic mapping of the conventional FAMM (the adjacent FAPMF flap does use CTA — PMID 40948003); consent content specific to FAMM; a preoperative mouth-opening threshold for eligibility; an airway algorithm defining tracheostomy indications; Stensen’s duct transposition operative detail and any re-anastomosis technique; an explicit mandibular periosteum preservation technique; anti-kink or pedicle-wrap devices or drain protocols for the tunnelized pedicle; a defined flap-monitoring schedule or any active anti-congestion treatment; and exact suture size/spacing/knotting for inset.

GRADE: ++○○ Low — consistent across multiple primary technique reports and open-access full texts, but no comparative evidence for most technical choices, and several steps rest on a single source.

Finding 6 — Modifications and variants: the family is organised by failure mode, not by creativity

Conclusion: Sixteen variant groups exist, and they were not generated by surgical creativity but by five recurring failure modes — the flap is too short, the pedicle is ligated, dentition/staging is unacceptable, the defect is through-and-through, or the flap is congested. Every variant is a response to one of those five. Understanding this makes the family predictable rather than a list of acronyms. The complete taxonomy is in Addendum A.

The five generative failure modes:

# Failure mode Solution family
1 Too short — defect beyond the flap’s reach FAMME (masseter extension, PMID 27330928); pedicle dissection through the submandibular gland; transverse-FAMM (axis rotated 90° across the vessels, PMID 27257582); Duranceau single-stage (PMID 22109756); Stensen’s relocation (PMID 26218177); trans-sinus access (PMID 42250969); iFOMM bone-bearing island (PMID 33257300)
2 Facial artery ligated / unusable dpFAMM (double-pedicled, PMID 34440964); BAMMIF (buccal artery); Bozola flap (PMID 22154378); random-pattern partial-thickness buccinator (PMID 27429953); contralateral designs (PMID 29616169; PMID 38894868)
3 Dentition / staging unacceptable FAMMIF; t-FAMMIF (tunnelized, PMID 38440437); Matros modification (design unverified)
4 Through-and-through defect (both nasal and oral lining) Folded t-FAMMIF (PMID 22705225); f-BMMF (PMID 25289306); Kite-t-FAMMIF (unfolded, PMID 36262678); Spacer FAMM (PMID 26710029); NAMMC (PMID 23997862)
5 Congestion / uncertain perfusion a-FAMMIF (unreliable — do not use, PMID 29771835); facial-vein inclusion; ICG fluorescence video-angiography-guided trimming (PMID 35867210); IRTI (PMID 41739500)

The single most useful organising document in this literature is Massarelli’s 2024 defect-oriented classification and algorithm for functional soft-palate reconstruction (PMID 39768688), which arranges the variants as a graded "ladder" from small to total soft-palate defects (FAMMIF → t-FAMMIF → folded t-FAMMIF → Kite-t-FAMMIF). If you want one paper to orient yourself in the variant family, that is it.

⚠️ The "folded FAMM" question — and a correction to this report’s own earlier draft

An earlier iteration of this review concluded that a folded FAMM could not be verified. That was wrong, and the correction matters. The evidence:

Design Exists? First report Mechanism
Folded t-FAMMIF YES Massarelli 2013 · PMID 22705225 · 10.1016/j.joms.2012.03.030 Island folded so one surface lines the nose and the other the mouth
f-BMMF (folded buccal musculomucosal) YES Kobayashi 2014 · PMID 25289306 Two denuded, separated mucosal islands (distal → oral, proximal → nasal)
Kite-t-FAMMIF YES — explicitly unfolded PMID 36262678 Muscle surface mucosalises (~3 weeks)
"Double-folded FAMM" NO — No retrieved description

Corroboration that folding is a recognised design category in this field, not an invention: Massarelli 2024 (PMID 39768688, open access) describes harvesting "in a folded fashion"; the Kite-t-FAMMIF paper’s own selling point is "without the need to fold the flap" — which only makes sense as a response to an existing folded technique; and an independent 2026 review lists "tunneled, folded, or super-thin flaps" as the emerging repertoire (PMID 41640131). How the earlier draft failed: one task’s table said "folded — unverified" while another task’s findings file already recorded it, and the discrepancy was only caught when a later iteration was made to read the siblings’ files. This is exactly why the review cross-checks tasks against each other.

⚠️ Three naming corrections (important if you read this literature)

  1. "e-FAMM" is not a real acronym. A dedicated search returns zero relevant records. The genuine primary acronym is FAMME — but it appears only in a figure legend of PMID 27330928 ("…flap with an island extension to the masseter muscle (FAMME flap)"), so title/abstract searching will never find it. "e-FAMM" is review shorthand that has drifted into currency.
  2. Acronym collision — "t-FAMM" means two different things. t-FAMM = transverse-FAMM (Pompei, PMID 27257582); t-FAMMIF = tunnelized FAMM island flap. Massarelli’s own comparison table even writes "t-FAMM" when meaning t-FAMMIF. Always check which is meant.
  3. NAMMC = nasal artery musculomucosal cutaneous flap (PMID 23997862) — it is not the nasolabial flap.

Also: "shuttle" and "fish-mouth" are not separate variants. They exist only as descriptive shape adjectives inside Zhao 2003 (PMID 12742209). Treat that as one design with three shape descriptors, not three variants — a distinction the secondary literature routinely blurs.

Where the variants were superseded by later evidence

Tunnelling was introduced to spare dentition and avoid a second stage. The best subsequent evidence argues against it: the n=141 cohort found tunnelized flaps more complicated (37% vs 7%, p<0.001) with no QoL gain (PMID 42026291, vein excluded in the tunnelized arm), and the randomised study found higher venous congestion (25% vs 5%) and trismus (15% vs 5%) (PMID 42578092, adjacent buccal-artery flap). Variant proliferation is not the same as variant superiority — a caution that applies to the whole family, most of which rests on case reports.

GRADE: +○○○ Very Low to ++○○ Low — the great majority of variants rest on case reports (many n=1) or small retrospective series. The two exceptions with comparative data are the conventional-vs-tunnelized comparison (PMID 42026291, Low) and the paediatric tissue-adjunct meta-analysis (PMID 37797378, Low–Moderate) — and the latter is a negative result (see Addendum A, Group N).

Finding 7 — Outcomes and complications

Conclusion: Pooled success is high and failure low, but the pooled congestion figure understates real-world congestion, and the largest comparative cohort shows that how the flap is designed matters more than the fact that it is a FAMM.

Pooled outcomes (PMID 39914242 — JPRAS 2025, 27 studies, 407 flaps in 402 patients, registry CRD42024529989):

Outcome Rate n
Pooled success 99.47% (95% CI 98.26–100.00, I² = 0%) —
Total failure 1.7% 7/407
Non-functional complications 26.0% 106/402
Partial necrosis 5.7% 23/407
Wound dehiscence 4.4% 18/407
Venous congestion 3.2% 13/407
Re-operation — 35/407
Pooled total complications 30.18% (95% CI 16.97–43.38, I² = 91%) —

⚠️ Do not add these two reviews. The 407-flap pool above and Ayad & Xie (PMID 24798986; 38 articles, 441 flaps, 1992–2013) draw on overlapping primary series. They are not independent confirmations of each other, and their totals must never be summed.

Congestion is systematically under-reported. Small series report congestion in essentially every flap — 8/8 and 17/17 — while the pooled figure is 3.2%. Since congestion is self-limiting over 24 h–1 week and no intervention is described, the discrepancy reflects reporting/reporting-threshold behaviour, not biology. Plan for congestion in most patients and reassure accordingly.

Other complications: haematoma 5/141 (4%); infection 1/141 (1%); Stensen’s duct injury 0/22 (PMID 26786664); persistent marginal mandibular palsy 4% vs 11% (PMID 42026291); scar contracture 6/7 needing release in 4 (PMID 41837171, tiny series).

Functional outcomes: regular diet 85% and functional speech 93% (PMID 17972306); normal speech 20/22 (90.9%) and normal mouth opening 19/22 (86.3%) (PMID 34441924); QoL global 79.27 with radiotherapy as the QoL driver (PMID 36556074). Notably, no speech benefit versus primary closure for defects ≤⅓ of the tongue (PMID 41890649) — i.e. the flap is not automatically justified by function alone in small defects.

Comparative effectiveness — versus free flaps (functionally equivalent, resource-sparing):

Parameter FAMM Free flap p Source
Operative time 196.7 ± 94.9 min 427.1 ± 129.8 min <0.001 (−54%) PMID 41464662
Hospital stay 12.7 ± 6.0 d 21.1 ± 8.0 d <0.001 PMID 41464662
ICU admission 7.3% (3/41) 83.3% (15/18) <0.001 PMID 41464662
Tracheostomy 7.3% (3/41) 16.7% (3/18) 0.357 PMID 41464662
Cost per patient USD 10,404 USD 23,607 <0.001 (−56%) PMID 41464662
Complications 1/13 patients 15 complications in 10/18 0.008 PMID 34861896
Tracheostomy 0 14 <0.001 PMID 31980317

All the PMID 41464662 figures above were re-verified by the reviewing agent directly against the open-access full text, including the coincidence that ICU admission and tracheostomy both equal 7.3% in the FAMM arm.

Oncologic non-inferiority (PMID 41127923, 615 patients: buccinator-myomucosal 390 vs free fasciocutaneous 225; cT1–T3 cN0 tongue/floor-of-mouth SCC): 5-yr PFS 69.8% vs 66.2%, adjusted HR (free flap vs BMF) 0.87, 95% CI 0.43–1.78 — non-inferiority met; 5-yr OS 77.9% vs 73.5%; local recurrence 7.5% vs 8.3%; regional 6.3% vs 6.1%; distant 2.1% vs 1.7%. Preserving the facial artery and vein during selective neck dissection did not compromise oncologic outcomes. Validity is conditional: small-to-medium defects, cN0, meticulous level I clearance, no R+ margins, no pull-through resection.

Conventional vs tunnelized — the largest comparative cohort (n=141; PMID 42026291): complications 37% tunnelized vs 7% conventional (p<0.001); total necrosis 11% vs 2% (p=0.048); partial necrosis 19% vs 2% (p=0.003); tooth extraction needed in 72% vs 4% (p<0.001); long-term QoL similar. In multivariate analysis flap type was the only significant risk factor. The authors’ technique excluded the facial vein in the tunnelized group — which is very likely the mechanism.

Other comparisons: inferiorly based FAMM had significantly less necrosis and dehiscence than buccinator flaps in n=134 cleft fistulas (PMID 40910620); "equally good" vs nasolabial flap in n=33 with no statistics (PMID 32523260); versus tongue flap in n=39, pain 3 vs 7 (p<0.096) and operative time 155 vs 242 min (p<0.002), with all tongue-flap patients reporting speaking/eating difficulty versus 2/16 for FAMM (PMID 27526234).

A discordant signal: one systematic review (PMID 41098423) asserts free flaps give better function for tongue reconstruction. This is not reproduced by any FAMM comparative series, and the supported claim is equivalent function at materially lower cost and morbidity — not superiority of the FAMM.

GRADE: ++○○ Low for pooled success/necrosis/dehiscence; ++○○ Low for conventional-vs-tunnelized (single-centre retrospective but large effect and multivariable-adjusted); +++○ Moderate for oncologic non-inferiority (large multicentre case-control, though observational and with wide confidence intervals); +○○○ Very Low for the pooled total-complication rate (I² = 91%).

Finding 8 — Contraindications, limitations and oncologic safety

Conclusion: The flap’s contraindications are almost entirely vascular and oncological, not anatomical. Two are absolute in practice; several commonly cited ones turn out to be unsupported.

Near-absolute (do not use the standard flap):

Contraindication Mechanism Evidence GRADE
Facial artery absent or hypoplastic No arterial inflow Two independent reviews name it; documented agenesis, Möbius-associated agenesis (p=0.03), bilateral hypoplasia; Doppler absence 0%/3.48%/10.46% by level +++○ Moderate
Facial artery ligated at prior surgery (standard single-pedicle FAMM only) No inflow. "The FAMM flap is an axial flap based on the facial artery and is useless in case of facial vessel ligation. In such cases, the reconstruction should be converted to a Bozola flap" (PMID 35279186) Grade C + B; escape routes documented: dpFAMM succeeded after ligation + 30 Gy; contralateral t-FAMMIF and buccal-artery myomucosal flap are the alternatives +++○ Moderate
cN+ neck with level IB perifacial nodes Oncological: perifacial nodal clearance conflicts with pedicle preservation. "In patients with cN+ necks, especially with level IB metastases, the use of facial-vessel-pedicled BMFs should be considered with great caution" (PMID 41127923) ++++ High (from the n=615 study’s own inclusion logic). See nuance below — this applies to the ipsilateral flap.
Skeletonised artery-only island / tunnelised pedicle Venous outflow failure Dog model: 100% necrosis with vein ligated vs 0% preserved (PMID 27648649); n=141: 37% vs 7% complications (PMID 42026291); "consider the a-FAMMIF an unreliable flap" (PMID 29771835) +++○ Moderate
Oral submucous fibrosis "OSMF is contra-indication for this flap" (PMID 38312956) Expert opinion +○○○ Very Low
Reverse-flow design with absent perinasal anastomoses (5–10%) No retrograde route PMID 24163549 (review-level) +○○○ Very Low

Relative — proceed with mitigation: prior radiotherapy (evidence conflict: an n=22 series calls it a contraindication [PMID 21535983]; the n=141 series refuted this, finding flap type and not RT to be the significant predictor, and three feasibility reports succeed in irradiated fields); defects larger than small-to-medium (a reach and donor-closure limit, not vascular — no cm threshold is published; largest defect closed was 8 × 7 cm but 5/24 needed an additional submental flap, PMID 29798613); dentition/prostheses blocking passage (tooth extraction needed in 72% of conventional cases; the Matros modification targets exactly this but its design detail is unverifiable); donor width >3 cm; pedicle kinking; paediatric/cleft cases (100% closure but high contracture; 1 total loss in 20 children).

Explicitly NOT supported — do not assert as a FAMM limitation:

  • Obesity or cheek thickness — no retrieved source mentions it at all.
  • Buccal fat pad herniation — the only retrieved statement is negative: the pad "could not herniate" (PMID 24163549).

Oncologic safety — SAFE with conditions. The 615-patient multicentre case-control study supports non-inferiority provided there is meticulous level I/perifacial clearance, cN0 disease, no R+ margins and no pull-through. Supporting series: n=50 FAMM (T1–3 N0) with occult metastases in 10/50, OS 88% and DSS 92% at 41 months (PMID 26286219 — margin and local-recurrence percentages not retrievable, 2-page brief report); n=21 with 5% level-IV recurrence and no increased locoregional risk (PMID 39662102). Field cancerisation is acknowledged but no recurrence within a FAMM has been reported in any retrieved study.

Important nuance on the cN+ contraindication — it applies to the ipsilateral flap. The contralateral FAMM island flap (C-FAMMIF) was developed specifically "to avoid the oncologic risks of ipsilateral regional flaps" and is explicitly positioned for patients with an ipsilateral cN+ neck: in 24 patients with T2–T3 oral defects compared against a propensity-matched ALT-flap control group (n=47), 2-year PFS was 88.2% vs 84.6% (p=0.6358), with 41.7% and 40.4% of patients respectively having ipsilateral pN+ necks; donor sites healed by primary closure without trismus or permanent facial palsy (PMID 38894868). That study also identified three anatomical patterns of the relationship between the pedicle and the marginal mandibular nerve and described the surgical management of each, and noted that the superficial venous drainage pattern makes contralateral-neck harvest safer when the ipsilateral neck has been dissected. So: a cN+ ipsilateral neck contraindicates the ipsilateral flap and is an indication for the contralateral design — not for abandoning the FAMM concept.

GRADE: +++○ Moderate for the oncologic safety claim and the vascular contraindications; +○○○ Very Low for the expert-opinion contraindications.

Finding 9 — Data-integrity findings (reported because they affect what clinicians may read elsewhere)

Conclusion: This topic’s secondary literature contains a retracted paper that has been absorbed into review-level claims, and two citation errors were found and corrected during this review.

The retraction. "The outcomes of Facial Artery Musculomucosal (FAMM) flaps in oral oncologic reconstructions" (PMID 40510716, DOI 10.4103/njms.njms_16_24, Natl J Maxillofac Surg 2025) is RETRACTED. The retraction notice is PMID 41019681 (DOI 10.4103/njms.njms_93_25), whose full abstract reads: "[This retracts the article on p. 24 in vol. 16, PMID: 40510716.]". Verified directly in NLM XML: 40510716 carries PublicationType = Retracted Publication; the notice carries Retraction Notice. The PubMed record states no reason for the retraction, and the article’s full text nonetheless remains live on PMC — which is why it keeps being read and cited. Downstream influence is numerically small (2 citations total), but it had already propagated into the working drafts used for this review, which is exactly how such papers cause harm.

What survives from it, and what does not:

Claim circulating via 40510716 Verdict Independent source
"3 cm maximum width" ✅ Survives PMID 37602111; PMID 40940987; PMID 24163549; PMID 39957871
"37% overall / 62% floor-of-mouth pedicle division rate" ⚠️ Unverifiable — do not use Attributed to Ayad & Xie, but their abstract contains no such rate, the paper is paywalled, and exhaustive full-text searching found no source other than 40510716 itself. Use instead: 27% (8/29) vs 0% (0/26) PMID 29481692; 7% (8/114) vs 15% (4/27) PMID 42026291
Its own 12-patient cohort results ❌ Drop Superseded by n=141 and n=615
"Pribaz 1.5–2 cm width" ⚠️ Downgrade No primary-source verification; appears to be a secondary attribution

Two citation errors found by independent verification (neither appears in this report):

  1. PMID 20502660 was cited for FAMM nasal-floor/nasal-septum reach; it resolves to "Risk factors for SARS transmission from patients requiring intubation" (Raboud J, PLoS One 2010). Almost certainly a digit transposition of the intended PMID 2052660 (Carstens 1991).
  2. PMID 42739192 was cited for the premasseteric-branch origin distance; it resolves to "Course Patterns of the Marginal Mandibular Branch of the Facial Nerve" (Vrapciu & Rusu, Diagnostics 2026), which contains neither "premasseteric" nor the quoted figure.

Also corrected: two sibling research drafts in this project used the term "inferior buccinator artery" (non-existent), and one cited PMID 37602111 for salivary fistula (which that paper does not mention) — see the progress.md log.

PRISMA Flow Diagram

Numbers below were computed programmatically from the 277 raw retrieval files actually on disk, not estimated.

IDENTIFICATION
├── Europe PMC REST queries executed ...................... 55 query sets
├── PubMed E-utilities queries executed ................... 24 query sets
├── Unique PMIDs identified across all sources ............ 1,078
└── Total identified ..................................... 1,078

SCREENING
├── Duplicates removed (cross-source overlap) ............. 336
├── Records screened (title/abstract) ................... ~742
└── Records excluded at title/abstract .................. ~616

ELIGIBILITY
├── Full-text / OA full text assessed .................... 126
│   (retrieved via Europe PMC fullTextXML; 24 for T3 alone)
└── Full-text excluded ................................... ~0
    (all retrieved full texts were relevant by construction)

INCLUDED
├── Citations independently verified ...................... 126
│   ├── verified, non-retracted, no caveat ............... 108
│   ├── verified with non-blocking caveat ................ 15
│   ├── RETRACTED ........................................... 1  (PMID 40510716)
│   └── MISMATCHED (excluded from report) ................... 2  (20502660, 42739192)
└── Studies cited in this report .......................... ~95

Evidence Tables

Study characteristics — the load-bearing studies

Study Design Setting Population n Key result
PMID 39914242 (2025) Systematic review + meta-analysis (registered CRD42024529989) Multi-database Oral cavity & tongue defects 407 flaps / 402 patients, 27 studies Success 99.47%; failure 1.7%
PMID 41127923 (2026) Retrospective multicentre case-control Multicentre cT1–T3 cN0 tongue/FOM SCC 615 (390 BMF / 225 free) PFS HR 0.87 (0.43–1.78), non-inferior
PMID 42026291 (2026) Retrospective comparative cohort Single centre, 2018–2025 Oral cavity cancer 141 (114 conventional / 27 tunnelized) 7% vs 37% complications, p<0.001
PMID 41464662 (2025) Retrospective comparative cohort Single centre, 2022–2025 Medium intraoral defects 59 (41 FAMM / 18 free) −54% operative time; −56% cost
PMID 24798986 (2015) Systematic review (narrative) PubMed Head & neck reconstruction 441 flaps, 38 articles, 1992–2013 Versatile for small/medium defects
PMID 24035128 (2014) Cadaveric anatomical — Fresh cadavers 13 / 26 sides Pedicle 4.8–4.9 → 7.7–7.8 cm ⚠ different flap
PMID 10203218 (1999) Cadaveric anatomical (latex) — Cadavers 10 "More an arterialized flap than an axial-pattern flap"
PMID 1513887 (1992) Primary clinical description — Oral/intraoral defects 18 flaps / 15 patients Original FAMM; 1 failure, 2 partial
PMID 2748738 (1989) Cadaveric anatomical + clinical — Cadavers 14 cadavers / 38 patients Buccal artery = major buccinator pedicle (antecedent flap)
PMID 27648649 Animal experimental (dog, split-mouth) — Buccal mucosa island flaps 3 animals 100% necrosis with facial vein ligated
PMID 40874402 Systematic review of Doppler studies — Facial artery calibre 7 studies / 649 FAs 2.14 mm (1.74–2.54), I² = 99.2%
PMID 41491248 (2025) Cadaveric anatomical — Buccinator layers + duct 50 hemifaces New deepest oblique layer beneath the duct

Risk of bias assessment (Newcastle-Ottawa Scale, observational studies)

Study Selection Comparability Outcome Total Quality
PMID 41127923 ★★★★ ★★ ★★★ 9/9 High
PMID 42026291 ★★★ ★★ ★★ 7/9 High
PMID 41464662 ★★★ ★ ★★ 6/9 Moderate
PMID 34861896 ★★ ★ ★★ 5/9 Moderate
PMID 26786664 ★★ ★ ★★ 5/9 Moderate
PMID 41837171 ★ ★ ★ 3/9 Low
PMID 29771835 ★ ★ ★ 3/9 Low
PMID 27648649 ★★ (animal) ★ ★ 4/8 Low–Moderate (n=3 animals)

Note on the evidence base as a whole: the FAMM literature is dominated by retrospective single-centre case series and technical reports. Only one meta-analysis and two large comparative cohorts exist. There are no randomised controlled trials identified for any FAMM design question, including the conventional-versus-tunnelized comparison that matters most.

Data-integrity assessment (all 126 verified citations):

Integrity metric Result
PMIDs resolved 126 / 126
DOIs resolved 108 / 108
Retracted records 1 (PMID 40510716) — excluded
Hard mismatches 2 (20502660, 42739192) — excluded
DOIs attached to the wrong paper 0
Duplicate records (same DOI, two PMIDs) 0

References

Included core studies

  1. Pribaz JJ, Stephens W, Crespo L, Gifford G. A new intraoral flap: facial artery musculomucosal (FAMM) flap. Plast Reconstr Surg. 1992;90(3):421–9. (PMID 1513887; DOI 10.1097/00006534-199209000-00009)

    • Study type: Primary clinical description · Population: oral/intraoral defects, 18 flaps/15 patients · Key finding: original FAMM; reliable superiorly and inferiorly based; 1 failure, 2 partial losses · Quality: Low (case series) · Note: the flap’s own dimensions are not present in any retrieved abstract

  2. Dupoirieux L, Plane L, Gard C, Penneau M. Anatomical basis and results of the facial artery musculomucosal flap for oral reconstruction. Br J Oral Maxillofac Surg. 1999;37(1):25–8. (PMID 10203218; DOI 10.1054/bjom.1998.0301)

    • Study type: Cadaveric anatomical (latex injection), 10 cadavers + clinical · Key finding: "more an arterialized flap than an axial-pattern flap" · Quality: Low–Moderate · Note: the single most on-point anatomical study for this research question; full text not retrievable

  3. Bozola AR, Gasques JA, Carriquiry CE, Cardoso de Oliveira M. The buccinator musculomucosal flap: anatomic study and clinical application. Plast Reconstr Surg. 1989;84(2):250–7. (PMID 2748738; DOI 10.1097/00006534-198908000-00010)

    • Study type: Cadaveric (14 cadavers) + clinical (38 patients), antecedent flap · Key finding: buccal artery is the major buccinator pedicle, supplies the posterior half, runs very close to the buccal nerve; venous drainage richer than arterial supply · Quality: Moderate · Note: different pedicle from the FAMM

  4. Zhao Z, Li S, Yan Y, et al. New buccinator myomucosal island flap: anatomic study and clinical application. Plast Reconstr Surg. 1999;104(1):55–64. (PMID 10597674; DOI 10.1097/00006534-199907000-00007)

    • Study type: Cadaveric dissection + 12 clinical cases · Key finding: buccal branch group (posterior, 1–3 inferior buccal, 3–5 anterior buccal); buccal artery ↔ posterior buccal branch anastomosis; buccal venous plexus → facial vein (2–4 tributaries) or pterygoid plexus/internal maxillary vein · Quality: Moderate

  5. Ayad T, Xie L. Facial artery musculomucosal flap in head and neck reconstruction: a systematic review. Head Neck. 2015;37(10):1525–36. (PMID 24798986; DOI 10.1002/hed.23734)

    • Study type: Systematic review (narrative), 38 articles 1992–2013, 441 flaps · Key finding: versatile for small/medium head-and-neck defects; floor of mouth most common site; functional/aesthetic outcomes excellent · Quality: Moderate · Note: abstract contains no pedicle-sectioning rate — the "37%/62%" figure attributed to it is unverifiable

  6. FAMM flap in oral cavity and tongue defect reconstruction: a systematic review and meta-analysis. J Plast Reconstr Aesthet Surg. 2025. (PMID 39914242; DOI 10.1016/j.bjps.2024.12.009)

    • Study type: Systematic review + meta-analysis, registry CRD42024529989, 27 studies · Population: 407 flaps / 402 patients · Key finding: success 99.47% (98.26–100.00); failure 1.7%; partial necrosis 5.7%; dehiscence 4.4%; congestion 3.2% · Quality: Moderate · ⚠ overlaps PMID 24798986 — do not sum

  7. Vaira LA, Massarelli O, Polesel J, et al. Oncologic safety of facial-vessel-based buccinator myomucosal flaps for tongue and oral floor reconstruction: a retrospective multicenter case-control study. Head Neck. 2026. (PMID 41127923; DOI 10.1002/hed.70082)

    • Study type: Retrospective multicentre case-control, cT1–T3 cN0 · Population: 615 (390 BMF / 225 free flap) · Key finding: 5-yr PFS 69.8% vs 66.2%; adjusted HR 0.87 (0.43–1.78), non-inferiority met; facial vessel preservation did not compromise oncologic outcome · Quality: High (9/9 NOS)

  8. Conventional vs. tunnelized facial artery myomucosal flaps in oral cavity reconstruction. Eur Arch Otorhinolaryngol. 2026. (PMID 42026291; DOI 10.1007/s00405-026-10214-y)

    • Study type: Retrospective comparative cohort, 2018–2025 · Population: 141 (114 conventional / 27 tunnelized) · Key finding: complications 37% vs 7% (p<0.001); total necrosis 11% vs 2%; partial necrosis 19% vs 2%; flap type the only significant multivariate risk factor; QoL similar · Quality: High (7/9 NOS) · Note: tunnelized technique excluded the facial vein

  9. Lean management in medium-sized oral cavity defect reconstruction: FAMM flaps versus free flaps. J Clin Med. 2025. (PMID 41464662; DOI 10.3390/jcm14248760)

    • Study type: Retrospective comparative cohort, 2022–2025 · Population: 59 (41 FAMM / 18 free flap) · Key finding: operative time −54% (196.7 vs 427.1 min), hospital stay 12.7 vs 21.1 d, ICU 7.3% vs 83.3%, tracheostomy 7.3% vs 16.7%, cost USD 10,404 vs 23,607 (−56%); functional outcomes equivalent · Quality: Moderate (6/9 NOS)

  10. Duranceau M, Ayad T. The facial artery musculomucosal flap: modification of the harvesting technique for a single-stage procedure. Laryngoscope. 2011. (PMID 22109756; DOI 10.1002/lary.22343)

    • Study type: Technique modification · Key finding: single-stage harvest avoiding a second-stage pedicle division

  11. Animal experimental: Buccal mucosa island flap on the facial artery alone. J Craniofac Surg. (PMID 27648649; DOI 10.1097/SCS.0000000000002947) — dog split-mouth, 3 animals; all flaps with the vein preserved survived; all with the vein ligated necrosed.

  12. Anatomical — facial artery calibre: Systematic review/meta-analysis of Doppler measurements. (PMID 40874402) — 2.14 mm (95% CI 1.74–2.54), 7 studies / 649 facial arteries, I² = 99.2%.

  13. Anatomical — artery/vein relationship: 90 halved viscerocrania, latex injection. (PMID 38792988) — facial artery anterior to facial vein in 100%, mean 6.2 mm.

  14. Anatomical — buccinator and parotid duct, 2025: Sci Rep. (PMID 41491248; DOI 10.1038/s41598-025-34600-4) — 50 hemifaces; newly described deepest oblique buccinator layer beneath the duct; parotid–masseteric and buccopharyngeal fasciae stabilise the duct against kinking.

  15. Cadavec pedicle length: Int J Oral Maxillofac Surg. 2014. (PMID 24035128) — 13 fresh cadavers / 26 sides; pedicle 4.8–4.9 → 7.7–7.8 cm; this is the inferiorly based BMMF island flap, not the standard FAMM.

  16. ⚠️ RETRACTED — excluded from all findings: PMID 40510716 (DOI 10.4103/njms.njms_16_24), retraction notice PMID 41019681 (DOI 10.4103/njms.njms_93_25).

The remaining ~80 cited sources are itemized with verified PMID + DOI in the project’s findings-T1.md through findings-T6.md and machine-readably in verified-citations.json.

Research Process Documentation

Methodology

Research type: AI-assisted systematic literature review.

Architecture: Ralph Loop (iterative fresh-context execution, spec as source of truth, append-only external memory) with multi-agent parallelism at each step.

Framework applied:

  • Phase 0 — Research Plan Generation: the question was decomposed into 6 discrete, prioritized single-question tasks and written to research_plan.json as the binding source of truth.
  • Phase 1 — Ralph Research Loop: 3 iterations. Each iteration loaded the plan from file (not from conversation), executed its task(s) through parallel subagents, and appended findings to findings-TN.md and learnings to progress.md — append-only, never rewritten.
  • Phase 2 — Synthesis & Evidence Grading: cross-task synthesis, GRADE assessment, Newcastle-Ottawa risk-of-bias scoring for the eight load-bearing observational studies, and PRISMA compilation computed programmatically from the raw corpus.
  • Phase 3 — Report Generation: this document, plus verification against the quality checklist.

Verification design (the distinguishing feature of this review): citation verification was performed by a separate agent from the one that collected each citation, against live PubMed/Europe PMC APIs, over all 126 identifiers. This was deliberate and it was necessary: two sibling agents had self-reported "all PMIDs/DOIs verified" while both citing a retracted paper. Self-reported verification is not verification.

AI tools: Hermes Agent with the ai-research-rounds skill (v2.0), multi-agent delegation (parallel leaf subagents, then an independent verification pass).

Review process:

  • Single reviewer (AI-assisted), with a separate independent AI verification pass over all citations
  • Dual human reviewer verification — not performed; this is the review’s principal methodological limitation

Search Strategy

Databases / interfaces searched:

  • Europe PMC REST API (primary discovery source; 55 query sets; 1,024 unique PMIDs)
  • PubMed/MEDLINE via E-utilities (esearch, esummary, efetch; 24 query sets; 390 unique PMIDs)
  • PubMed Central / Europe PMC open-access full texts (126 full texts retrieved)

Date range: 1989–2026 (from Bozola’s buccinator anatomical study to 2026 publications)
Search date: 2026-09-25
Language: English (plus English-indexed non-English records)

Representative search terms:

PubMed:
("facial artery musculomucosal"[tiab] OR FAMM[tiab] OR "buccinator myomucosal"[tiab])
AND (anatomy[tiab] OR cadaver[tiab] OR "vascular anatomy"[tiab] OR technique[tiab]
     OR complication*[tiab] OR "case series"[tiab] OR outcome*[tiab])

Europe PMC field-tagged queries, e.g.:
(TITLE:"facial artery musculomucosal") AND (anatom* OR cadaver* OR vascular)
(TITLE:FAMM) AND (modification OR variant OR island OR folded OR chimeric)
AUTH:"Pribaz" AND (musculomucosal OR buccinator OR intraoral)
AUTH:"Dupoirieux" AND (musculomucosal OR "oral reconstruction")

Search volume check: PubMed exact phrase "facial artery musculomucosal" = 96 records; Europe PMC loose phrase = 231 records. The corpus is small and bounded, so saturation is reachable — a genuinely useful property for a topic this narrow.

Selection Criteria

Inclusion: human studies; any design (cadaveric/imaging anatomy, technique description, case series, comparative cohort, systematic review, meta-analysis); FAMM or directly adjacent buccinator-myomucosal flaps with clearly attributed vascular pedicle; English or English-indexed; 1989 onwards.

Exclusion: conference abstracts without retrievable data; retracted records; non-human studies except where explicitly flagged as animal evidence (PMID 27648649 is retained and labelled because it is the only direct experimental test of the facial-vein question); records whose identifiers could not be independently verified.

Source Registry

Database Platform Search date Query sets Records identified Full texts retrieved Included
Europe PMC REST API 2026-09-25 55 1,024 unique PMIDs (995 unique DOIs) 126 ~95
PubMed/MEDLINE E-utilities 2026-09-25 24 390 unique PMIDs — "
PubMed Central / Europe PMC OA fullTextXML 2026-09-25 — 126 full texts 126 "
Total (deduplicated) 79 1,078 126 ~95

Limitations & Evidence Gaps

Methodological Limitations

  1. No human second reviewer. All screening, extraction and synthesis was AI-performed. Citation verification was independently replicated by a separate agent (126/126 identifiers), but content interpretation was not.
  2. The evidence base is weak in design. The FAMM literature consists predominantly of retrospective single-centre case series and technical reports. No randomised controlled trial was identified for any FAMM design question — including conventional-versus-tunnelized, which is the most decision-relevant comparison. The nearest randomised evidence is a 40-patient RCT of posterior-based buccinator myomucosal flaps (a buccal-artery pedicle, not a FAMM; PMID 42578092) — informative about tunnelling in this reconstructive family, but not a direct answer for the FAMM.
  3. Two paywalled key papers limited detail. Bozola 1989 (the duct-transposition technique) and Ayad & Xie 2015 (whose pedicle-sectioning figure is the subject of the retraction audit) could not be fully retrieved. The Bozola duct-transposition operative detail is therefore a stated gap rather than an omission.
  4. Double-counting is intrinsic to this literature. The two main reviews share primary series. Pooled figures quoted here must be read as describing overlapping cohorts.
  5. English-language bias.
  6. The "FAMM" label is applied inconsistently. Many studies include buccinator myomucosal flaps with a buccal-artery pedicle under FAMM statistics. Outcome figures should be read as "facial-vessel-based buccal flaps" rather than strictly facial-artery flaps.

Evidence Gaps (specific and actionable)

  1. The standard pedicled FAMM’s usable pedicle length has never been directly measured. Published figures come from a different flap (BMMF island) or from contralateral island designs.
  2. The venous anatomy is essentially unmeasured. The submucosal venous plexus — the sole presumed outflow — has no quantitative description. The 2 cm minimum pedicle width is clinical empiricism, not a measured threshold.
  3. No Doppler or CTA study has mapped the facial artery in the conventional FAMM’s own harvest field (buccal mucosa over buccinator, between the parotid papilla and retromolar region) — the imaging studies retrieved measure the artery cutaneously or at the mandibular border. The single exception in the whole family is the facial artery perforator myomucosal flap (FAPMF), which uses CT angiography plus Doppler (PMID 40948003, n=8, Chinese-language), i.e. CTA planning exists for the perforator variant but not for the classic flap.
  4. The axial-versus-arterialized question has never been tested. A single 1999 cadaveric study (n=10) contradicts the entire secondary literature, and no controlled anatomical or perfusion study has followed.
  5. Buccinator thickness and the muscle:mucosa proportion of the flap are unmeasured — which matters for bulk and contracture.
  6. Stensen’s duct transposition is described but never detailed, and duct re-anastomosis is described nowhere. Duct injury did not occur in the one series that measured it (0/22), so its true incidence is unknown.
  7. No anti-congestion protocol exists. Congestion is managed expectantly by common consent, with no leech, anticoagulation or pharmacological protocol described anywhere.
  8. The Matros modification for intact dentition has no retrievable design detail — despite addressing the fact that 72% of conventional cases required tooth extraction.
  9. Right-left and sex asymmetry in facial artery calibre is documented but its implication for harvest laterality is unaddressed.

Future Research Needs

  • A prospective comparative study of conventional vs tunnelized FAMM with the facial vein included in both arms, to separate the effect of tunnelling from the effect of venous exclusion. The n=141 cohort strongly implicates the vein, but the two variables were confounded.
  • A perfusion/cadaveric study (e.g. CT angiography or fluorescent perfusion) directly testing the axial-versus-arterialized question at the flap’s own harvest field.
  • A quantitative anatomical study of the buccal submucosal venous plexus, and a measured test of the 2 cm pedicle-width threshold.

Disclosures

Researcher: Liming Gou, D.M.D., Ph.D. — Oral and Maxillofacial Surgeon, Stomatological Hospital of Chongqing Medical University.
Disclosures: none declared.

AI tool: Hermes Agent (independent AI assistant platform), using the ai-research-rounds v2.0 skill and multi-agent delegation.

  • No conflicts of interest. No commercial funding.

Funding: None.

Important caveat on AI-assisted synthesis: every identifier in this report was independently re-resolved against live PubMed/Europe PMC records by a verification pass separate from the collecting pass, and quantitative claims were traced to retrieved text. Nevertheless, this report has not been human-peer-reviewed and should be read as a structured evidence map and orientation document rather than as an authority. Statements labelled NOT FOUND / unverifiable are genuine gaps, deliberately left visible rather than filled.

Version History

Version Date Changes
1.0 2026-09-25 Initial publication

Addendum A — Full variant taxonomy (Group A–P)

Organised by the clinical problem solved. Every identifier below was re-resolved against PubMed/Europe PMC. [P] = primary report; [R] = review-level naming only; [NF] = not found.

Group A — "The flap is too short"

Variant First report Structural difference Reported numbers
FAMME (masseter island extension) Xie 2016 · PMID 27330928 · 10.1055/s-0036-1579630 Island extension onto masseter muscle/fascia 10 × 2.5 cm, n=1; longest reported FAMM
Transverse-FAMM (t-FAMM) Pompei 2016 · PMID 27257582 · 10.1097/GOX.0000000000000632 Flap axis rotated ~90° across the vessels 8 × 3 cm, n=1
Pedicle dissection through submandibular gland PMID 24035128 Extends pedicle 4.8 → 7.8 cm 13 cadavers / 26 sides
Duranceau single-stage PMID 22109756 Prolonged anterior incision Avoids staged division
Stensen’s duct relocation PMID 26218177 Duct repositioned to widen flap Technique
Trans-sinus / maxillary window access Manfuso 2026 · PMID 42250969 Access route; 180° rotation n=1
iFOMM (osseous/osteo-myomucosal island) PMID 33257300 Bone-bearing island n=3 — the only composite variant
EULI flap Sarukawa 2006 · PMID 17113537 Wide mucosal island incl. artery and vein; no transoral pedicle Sibling design
FAPMF (facial artery perforator myomucosal flap) Yi 2025 · PMID 40948003 · 10.7507/1002-1892.202507082 Perforator-based; uses CT angiography + Doppler 8 patients; the only CTA-based planning in the family; Chinese-language

Group B — "The facial artery is ligated / unusable"

Variant First report Structural difference
dpFAMM PMID 34440964 Facial + buccal artery pedicles — succeeded after ligation + 30 Gy
BAMMIF PMID 39768688 Buccal-artery pedicle
Bozola flap PMID 22154378 Buccinator musculomucosal, buccal-artery pedicle
Random-pattern partial-thickness buccinator PMID 27429953 Non-axial design
Pedicled buccal backup PMID 35072841 Salvage design

Group C — Dentition / staging

FAMMIF (PMID 10597674); t-FAMMIF (PMID 38440437); Matros modification (PMID 20124849) — title verified, design NOT FOUND after three retrieval routes.

Group D — Through-and-through defects (both linings)

Variant First report Mechanism
Folded t-FAMMIF Massarelli 2013 · PMID 22705225 Island folded: one surface nasal, one oral; total soft palate
f-BMMF Kobayashi 2014 · PMID 25289306 Two denuded separated islands
Kite-t-FAMMIF Massarelli 2022 · PMID 36262678 Explicitly unfolded; muscle mucosalises
Spacer FAMM PMID 26710029 Oronasal fistula closure + palatal lengthening
NAMMC Rossell-Perry 2012 · PMID 23997862 Nasal artery musculomucosal cutaneous flap

Group E — Congestion / perfusion verification

a-FAMMIF — unreliable, 1 total + 2 partial necroses in 5 (PMID 29771835); facial-vein inclusion; ICG fluorescence video-angiography-guided trimming (PMID 35867210); IRTI (PMID 41739500, 10% vs 35%).

Group F — "Wider than one flap"

Bilateral / double BMMIF (PMID 20307776); trilobed (PMID 18835756); shuttle / fish-mouth / three-leaf — one design, three descriptive names (Zhao 2003 · PMID 12742209); dual "kite" (PMID 36752413); chimeric designs (PMID 38561137, 36608097, 39342787, 39855302).

Group G — Palatal lengthening

Spacer FAMM (PMID 26710029); buccinator sandwich pushback (Hill 2004 · PMID 15151445, 16 patients, 93% improved); nasal-side-only BMMF (PMID 40232256).

Group H — Cleft palate / alveolar cleft / bone coverage

Yang 2013 bone-coverage BMMF (PMID 24002491, 15 patients, 90° turn over raw bone to prevent maxillary growth restriction); Rahpeyma 2021 anterior palatal fistula (PMID 35223651, 10 patients); MacIsaac 2026 (PMID 41837171, 13 patients: BMMF closure 66.7% vs FAMM 100% but 86% scar contracture) — proposes the first defect-location algorithm; Fang 2014 (PMID 24469368, 22 fistulas, 18/22 satisfactory); Masseter–buccinator periosteal flap (PMID 42390680, n=1 — a sibling concept, masseter + buccinator together, not a FAMM).

Group I — Lip / vermilion / commissure

Pribaz 2000 (PMID 10724244, 16 flaps/13 patients); EULI (PMID 17113537); FAMMIF + skin chimeric (PMID 39342787); chimeric nasolabial + buccal-artery (PMID 39855302, n=7); motor-innervated chimeric BMM–submental island (PMID 38561137); LVMF commissuroplasty for Tessier 7 (PMID 40540842, 12 patients).

Group J — Nasal lining / septum / nasopharynx

Superiorly based FAMM for nasal lining (PMID 26566439); large septal perforation (PMID 16258293); reversed-flow superiorly based FAMM, 12 patients across palate/lip/septum/conjunctiva (PMID 25795189); post-surgical nasopharyngeal stenosis — 8 patients, 5 two-stage, 4 complete relief (PMID 25328699); bilateral FAMM for chemotherapy-induced NPS (PMID 32983793); nasopharyngeal/oropharyngeal reconstruction (PMID 32397756); reverse buccinator musculomucosal flap on retrograde anterior buccal artery via oro-nasal tunnel, 3 patients (PMID 16248520, Chinese).

Group K — Orbit

FAMM for lower eyelid fornix deepening in socket contracture (PMID 33503749, n=1) — the paper itself calls it a new indication.

Group L — Skull base

FAMME (PMID 27330928); reverse-flow facial artery buccinator (FAB) flap — cadaveric reach 7/8 sellar & planum, 4/8 contralateral cribriform, 0/8 middle clivus (PMID 26682122) → the middle clivus is out of reach, and that bounds the indication; cadaveric surface area 15.90 cm² (PMID 23666761); trans-sinus (PMID 42250969).

Group M — Contralateral harvest (ipsilateral pedicle unusable)

Contralateral islanded FAMM for FOM, n=1 (PMID 29892333); C-FAMMIF, median max pedicle 106 mm, 2-yr PFS 88.2% vs ALT 84.6% (PMID 38894868); foci contralateral FAMF, 10 patients, all flaps survived, flap 2.5 × 3 to 5 × 5 cm, universal post-op mouth-opening limitation with 9/10 normal by 1 year (PMID 29616169); Zhao 2003 the classic fallback statement (PMID 12742209).

Group N — Paediatric (application, not geometry)

Nasal-side-only BMMF in two-flap palatoplasty (PMID 37203179 / 40232256); IVV + BMMF (PMID 41460679); double-opposing BMMF (PMID 38527345); buccinator sandwich pushback (PMID 15151445, 93% improved); Furlow + BMMF comparative, 30 patients (PMID 41420755); largest paediatric series — 123 secondary cleft fistulas, mean age 6 y 11 mo, largest fistula 6 × 7 cm, no haemorrhage/infection/necrosis, significant speech improvement (PMID 42556038).

⚠️ Counterweight — the harm side (PMID 40892050): after BMMF in cleft palate, mucoperiosteal flap necrosis 23.8% (10/42), VPI 50% (21/42), palatal fistula 26% (11/42).
⚠️ The most important negative result in the paediatric domain (PMID 37797378, SR, 1,332 patients): BMMF 45.4% vs buccal fat pad 40.8% vs acellular dermal matrix 14% — no significant difference in fistula (p=0.79) or VPI (p=0.14) between adjuncts. Adjunct choice did not change outcomes; BMMF was simply used more often in severe Veau III/IV clefts. No paediatric-specific geometry change exists — paediatric use is a population, not a design.

Group O — Donor-site morbidity solutions

Buccal fat pad advancement — "a method for harvesting a broader flap than is traditionally described" (PMID 21535983); masseter flap for donor coverage where the fat pad cannot be used in irradiated tissue (PMID 24163549); standardised donor-site morbidity instruments (PMID 20813563 — mouth opening 8.6/9, commissure symmetry 8.6/9; PMID 27452415 — 3 patients with ipsilateral upper-lip anaesthesia, 5 with slight orbicularis dysfunction).

Group P — Sensory / functional quality

Neurovascular island flap with preserved buccal nerve — 36 flaps, mean follow-up 15.3 months, 11 bilateral, flap always reached the contralateral pillar, buccal nerve always preserved (PMID 39903875); EMG evidence of flap-muscle reinnervation with active tongue motion (PMID 12742209, n=1); sensory-recovery evidence base — 52/57/17-flap series showing recovery "better than reinnervated microvascular free flaps", tactile threshold 0.76 ± 1.58 g/mm (PMID 29130554; PMID 29396784; PMID 29225008) ⚠ but note all three are from a single Italian group — a co-location risk for this specific claim.

Saturation check — variant families that do NOT exist (honest negatives)

# Does not exist Basis
N1 "e-FAMM" as a literature acronym 0 relevant records; correct term is FAMME (figure legend only)
N2 "Double-folded FAMM" 19 hits, none describing such a design
N3 FAMMIF with a lingual tunnel 70 hits, no match — the route is paramandibular + anterior tonsillar pillar
N4 Standalone "shuttle FAMM" or "fish-mouth FAMM" Words exist only as adjectives inside Zhao 2003 — one design, three names
N5 Bilobed FAMM No hit (trilobed exists)
N6 Prefabricated / prelaminated / tissue-expanded FAMM 33 hits, none FAMM-related
N7 Balloon / expander / silastic sheet in the FAMM harvest 78 hits, none FAMM-specific — the only catheter in this literature is the 22G duct cannula
N8 FAMM flap "delay" Only palatal-flap delay combined with a FAMM (PMID 40104415) — do not describe a "FAMM delay"
N9 Supercharged / AV-loop / venous-coupled FAMM 22 hits, none — this family moves away from microsurgery
N10 Robotic / TORS-assisted FAMM harvest Appears inside TORS reconstruction series, but no robotic harvest or robot-specific design
N11 Any second source for Stensen’s-duct cannulation Single-source (PMID 37741079, 22G angio-catheter)
N12 FAMM in oral submucous fibrosis surgery Not found — and the one explicit statement is a contraindication (PMID 38312956)
N13 Cartilage- or mesh-bearing FAMM Only the iFOMM bone-bearing island (PMID 33257300)
N14 A paediatric-only geometry change Paediatric use is a population, not a design
N15 Any FAMM variant named for a "premasseteric branch" The identifier previously cited for this (PMID 42739192) is a marginal-mandibular-nerve paper — a mismatch
N16 A single "maximum FAMM defect size" number Largest defect (8 × 7 cm) ≠ largest flap (5.5 × 4 cm) ≠ largest design claim (9 × 4 cm). Any single number would be fabricated

About AI Research Rounds

AI Research Rounds uses AI-assisted literature review to synthesize evidence on clinical questions. It follows systematic methodology with transparent documentation and explicit, auditable external state. Powered by Ralph Loop architecture (Geoffrey Huntley) for iterative, fresh-context execution with multi-agent parallelism.

Project artifact index

Not shown.

本文仅代表作者个人观点，不代表作者所在单位立场。内容仅供医学教育参考，不构成诊疗建议；具体诊疗请咨询专业医生。详见免责声明。