Surgical evolution
Flaps were classified using the Brown classification (Table. 1).
Autologous composite reconstructions using chimeric flaps (bone, skin, muscle) 
Flaps most frequently employed for various midface defects (classified by Brown; Table. 1 ) are composite flaps comprising soft tissue and bone. Flaps harvested from the iliac crest5scapula6 and fibula7 have been well-described, each facilitating dental restoration. In addition, microvascular flaps have been reported from the forearm8, rectus abdominis9, latissimus dorsi10 and the anterolateral thigh flap11.
The very first microvascular midface reconstructions that we performed (1993) were flaps supplied by the subscapular axis. Our preference was a single pedicale scapula flap perfused on the angular artery, incorporating the latissimus dorsi and serratus anterior (with or without rib) with or without a fasciocutaneous paddle. The main advantage of this composite flap is its freedom of movement to suit most reconstructive needs. Brown type IV defects were reconstructed using this approach enabling us to reconstruct almost any complex defect.
Our technique has previously been described in more detail12. In brief, the scapula is positioned vertically, and osteotomies allow for the infraorbital rim and orbital floor to be reconstructed. The LD muscle was incorporated to separate the oral and nasal cavities. Patients had bespoke eye prostheses placed(Fig 1a,b,c,d,, Fig s4) . Although this reconstruction provided tremendous benefit for patients with Brown type IV maxillectomies, its drawbacks became evident over time. With the absence of 3D planning and rapid prototyping (unavailable at the time) to aid in scapular osteotomies intraoperatively, longer-term results were unpredictable. Moreover, tumor resection and reconstruction could not be performed simultaneously due to patient positioning, increasing complication rates (longer anaesthesia) and overall operative costs. Despite the operation becoming routine departmental practice, operative times exceeded 7-8 hours.
Combination of soft tissue or bone flaps and 3D shaped titanium mesh
Early in 2000, CT planning and 3D printing became available bringing about a shift in our reconstructive approach. A pedicled buccal fat flap with autologous bone using a prefabricated titanium mesh provided a functional reconstruction, similar to that described by Liu et al.13. In our case titanium mesh was used to support orbital contents. ALT flap was partly deepithelialised to cover the titanium mesh and to separate oral and nasal cavity. In cases were microvascular bone was used, afforded a stable construct for alveolar ridge reconstruction. All patients were classified as having Brown class III defects. Titanium mesh provides sufficient material to bridge most large midface defects.  Mesh was contoured preoperatively on 3D models to develop a stable platform to support overlying soft tissues14 (Fig 2a,b,c,d,e, s5) . Consequently, we were able to offer a simultaneous two team approach in which operative time was lowered to approximately 5-6 hours. Titanium mesh was also useful in trauma cases and skin tumor surgery when bone was resected. The difference between these cases and midface reconstruction was that these cases did not need postoperative irradiation whereas all midface reconstructions required adjunctive irradiation.
The majority of patients not receiving irradiation experienced no complications. Conversely, those that did need irradiation experienced complications, manifesting up to a year after surgery. More commonly we encountered fistulas and at the extreme end we encountered extrusion of the titanium mesh. In cases of prolong, repeated infection and potential implant extrusion lipofilling helped to mitigate this risk. In two instances, we had to utilize a pericranial flap and full thickness skin graft to reconstruct the infraorbital rim. In three cases we had to perform an additional free flap.
Computer assisted autologous 3D modelling utilizing the fibula rather than scapula
We have found that the adoption of 3D planning for autologous reconstruction has revolutionized head and neck reconstruction in our department and worldwide. Computer assisted 3D planning generates bespoke resection and cutting guides (Fig 3a,b,c,d)15. This was a welcome step, giving surgeons confidence in their reconstructive goals. We chose the fibula flap as our workhorse flap. The fibula tolerates multiple osteotomies, and the long pedicle avoids the need for vein grafts when anastomosing to vessels in the neck, lowering any thrombotic risk16. All of these cases were classified as Brown class II and III.
Indeed, the reconstructive gamechanger was CT planning for reconstruction (s1,2,3). After many years of practice, we realized we had to adapt to technological advances such as this. Its impact despite skepticism was far beyond our expectations. CT scans fed into computer software to generate a bespoke model to guide surgical resection (Brainlab AG, Germany). We use preplan STL files ( Standard Tessellation Language or STereoLithography) positioned according to a patients individualized length and angle. Data is submitted to the CAD Autodesk 123D and Autodex Fusion 360 software program whereby modeling of the resection and cutting guides is performed.
To enable 3D printing, preparations are made using an Autodesk Mashmixer + Ultramaker Cura software. Guides are printed using the 3D printer Ultimaker 2+ (Ultimaker, Netherlands). What is unique is that the majority of the plan and 3D printing is performed within the department. For this reason, patients can be prepared for surgery within 3 to 7 days. Usually, two osteotomies were performed to recreate the alveolar ridge for insertion of dental implants, and distally on the flap to recreate the zygomaticomaxillary buttress. The fibula was fixed with miniplates and 2.0mm screws for the alveolar ridge as well as to the body of the zygomatic bone. Additional stability was achieved using the free fibula bone fixed with miniplates to the bone flap itself and the medial aspect of the infraorbital rim when recreating the nasomaxillary buttress.
We felt that a free bone graft when we started out in 2017, would be a temporary measure and we expected that it would resorb over time. Encouragingly, we have seen that even after three years, bone graft following integration retains its volume. There was no significant inflammatory reaction nor any graft excursion. We position the pedicle in such a way that it is orientated distally, at the level of zygomatic body. Vein grafts are not used in our practice except in one instance when the flow via the facial artery was considered too weak to support the flap and a vein graft used as a precautionary measure. For Brown class III defects, the fibula flap was combined with titanium mesh to reconstruct the orbital floor and infraorbital rim.  The reporting guideline has been followed in this study.