Management of maxillary defects is among the most challenging and controversial areas of head and neck oncologic reconstruction. Options include use of prosthetic obturators, pedicled flaps, and free flaps, sometimes combined with grafts or alloplasts. Although the popularity of pedicled flaps has declined because of limited reach and volume, obturators remain a good solution for select patients with limited palatal defects.1–4 However, for extensive defects, obturators may be difficult or impossible to retain, particularly in edentulous patients.1,4 Furthermore, obturators are usually inappropriate for defects that involve resection of the skull base, orbital floor, orbital contents, or soft tissues of the face.
Maxillary reconstructions with various bony and soft-tissue free flaps have been described, and the best technique is a subject of debate. One of the fundamental problems with reconstructing the maxilla is that defects created by oncologic resection are highly variable. In addition, because cancers involving the maxilla are rare, comprehensive studies have been limited to a few centers.4–8 Further guidance based on objective outcomes is needed, particularly for maxillary defect types that are infrequently encountered.
We conducted an analysis of all maxillary free flap reconstructions performed at our institution during a 10-year period. Defects were evaluated with respect to size and extent, and the types of reconstructions performed. Reconstructive outcomes, including short- and long-term complications and postoperative speech and swallowing function, were then examined. Based on these outcomes, recommendations are made regarding reconstruction depending on the specific type of defect.
PATIENTS AND METHODS
All free flap reconstructions for oncologic maxillary defects performed at our institution between 2001 and 2010 were reviewed. Purely orbital or nasal reconstructions were excluded from this study. Institutional review board approval was obtained before this study was undertaken.
The extent of each defect, including the amount of palate and alveolus, and whether or not the orbital floor or orbital contents were resected, was recorded. Figure 1 illustrates the various palatoalveolar defects encountered. Figure 2 illustrates orbital floor and orbital cavity defects, which may occur in combination with any of the palatoalveolar defects shown in Figure 1. References are made to previously described classification systems by Brown and Shaw9 and Cordeiro and Chen.10 to facilitate comparisons with other studies. Complications were classified based on time period, either perioperative (occurring ≤30 days after surgery) or late (occurring >30 days after surgery). Speech and diet at the time of last follow-up were recorded. Speech was classified as greater than 80 percent intelligible, 50 to 80 percent intelligible, or less than 50 percent intelligible.11 Diet was categorized as unrestricted, soft, pureed, liquid, partial oral, or nonoral.12
Continuous data (reported as mean ± SD) were compared using the Wilcoxon rank sum test. Categorical data were compared using the Fisher's exact test or chi-square test, as appropriate. All tests were two-tailed. Values of p < 0.05 were considered significant.
Two hundred forty-six maxillary free flap reconstructions were performed during the study period. Patients included 141 men and 105 women with a mean age of 55.2 ± 19.4 years. Tumor histopathologic diagnoses and sites of origin are listed in Table 1. Patient characteristics that might be expected to affect outcomes are listed in Table 2. Defects were analyzed according to missing structures and are listed in Table 3.
Reconstructive flap choices for each type of defect are listed in Table 4. Five patients underwent reconstruction with two simultaneous free flaps, including four patients who received fibula osteocutaneous free flaps for palatoalveolar reconstruction and anterolateral thigh flaps for external skin coverage, and one patient who received an fibula osteocutaneous free flap for palatoalveolar reconstruction and a serratus anterior muscle with rib free flap for orbital rim and floor reconstruction. Although fibula osteocutaneous flaps were the most frequently used bone free flap, serratus anterior muscle with rib flaps were used for restoring the orbital rim and floor in eight cases.
Some free flaps were used only rarely because of specific patient-related conditions: three latissimus dorsi myocutaneous flaps and two lateral arm fasciocutaneous flaps were used because of obese body habitus that would have resulted in a overly thick anterolateral thigh or rectus abdominis myocutaneous flap. Also, four rectus abdominis muscle flaps were used in patients in whom obesity precluded use of a myocutaneous flap for the same reason. An additional latissimus dorsi myocutaneous flap was selected over an anterolateral thigh flap because of arteriovenous malformations involving both thighs.
Orbital Floor Reconstruction
The orbital floor was reconstructed in 59 patients (23.4 percent). It was reconstructed with titanium mesh in 30 patients (50.8 percent), bone grafts in 20 patients (33.9 percent), porous polyethylene in four patients (6.8 percent), bony free flaps in four patients (6.8 percent), and a fascia lata graft in one patient (1.7 percent). The proportion of patients undergoing preoperative or postoperative irradiation was not significantly different between those reconstructed with a bone graft compared with those reconstructed with an alloplast (95.0 percent versus 82.4 percent, respectively; p = 0.28). Patients who received bony free flap reconstruction had a history of prior irradiation and had failed prior alloplastic reconstruction.
Complications related to orbital floor reconstruction were analyzed separately from other complications because a minority of patients underwent orbital floor reconstruction (Table 5). The material used was not a significant predictor of a complication (p = 0.18). In addition, three patients who underwent soft-tissue free flap maxillary reconstruction but did not receive separate orbital floor reconstruction all developed dystopia and enophthalmos.
Perioperative complications, excluding complications related to orbital floor reconstruction, are listed in Table 6. None of the potential risk factors listed in Table 1 were associated with a significantly higher rate of recipient, donor, medical, or overall complications (data not shown). There was no significant difference in the proportion of patients experiencing a perioperative complication between patients undergoing soft-tissue free flap reconstruction and patients undergoing bony free flap reconstruction (36.0 percent versus 43.5 percent, respectively; p = 0.35).
There were eight free flap losses (3.3 percent) with causes that included two venous thromboses caused by compression of the pedicle, one venous thrombosis caused by propagation of internal jugular vein thrombosis, two arterial thromboses associated with hypercoagulability disorders, and one arterial thrombosis associated with purulent infection. There were two losses in which flap compromise was recognized late and the cause could not be determined. Three patients with posterior palatomaxillectomy defects and no orbital floor/orbital exenteration defect that experienced flap losses were treated with obturators, and the remaining five patients (three hemipalatomaxillectomy and two bilateral palatomaxillectomy defects) underwent a successful second free flap.
Late complications are listed in Table 7. No late-occurring fistula healed spontaneously, in contrast to two fistulas that occurred during the perioperative period that did. Oroantral fistulas were treated with prosthetic obturation. Nasocutaneous fistulas, occurring along prior incision lines, were treated with readvancement of the free flap (n = 1) or additional free flaps (n = 4), or left open with a permanent fistula (n = 5). Two patients with nasocutaneous fistulas that were treated with a second free flap later refistulized and required a third free flap. None of the potential risk factors listed in Table 1 were associated with a significantly higher rate of late complications (data not shown), nor was there a significant association between late complications and bony versus soft-tissue free flap reconstruction (p = 0.47).
Speech and swallowing outcomes were analyzed for patients with a successful reconstruction and an assessment that occurred at least 3 disease-free months after reconstruction (Table 8). Of 229 patients meeting these criteria with regard to speech assessment, only 11 (4.8 percent) had less than 80 percent intelligibility, including nine patients with 50 percent or more of their soft palate resected and two patients who had additional resection involving their tongue. Of 223 patients meeting these criteria with regard to swallowing assessment, 12 (5.4 percent) were partially or totally tube feed dependent, including nine patients with resections that involved other oral cavity structures and three patients with aspiration related to stroke, vocal cord paralysis, or progressive dementia (one patient each).
Outcomes involving dental restoration for patients with at least 6 months of follow-up are listed in Table 9. Dental restoration was provided in 80 patients (40.0 percent), including 24 patients who received osseointegrated implants (all with osteocutaneous free flap reconstruction) and 56 patients who received conventional prostheses (four with osteocutaneous free flap reconstruction and 52 with soft-tissue free flap reconstruction). Some patients did not desire dental restoration (e.g., because missing teeth were posterior and not easily visible), and others may have pursued obtaining a prosthesis after leaving our care.
No patients received osseointegrated implants following postoperative radiation therapy. There were four patients who were initially treated with obturators and reconstruction with osteocutaneous free flaps in a delayed manner following the conclusion of radiation therapy and two patients who underwent reconstruction immediately with an osteocutaneous free flap and underwent osseointegrated implant placement less than 2 weeks later, before beginning radiation therapy.
We analyzed our experience involving 246 midfacial reconstructions, which is the largest experience reported to date. Tumors requiring midfacial resections included a large and heterogenous group of histopathologic entities that arose from the sinonasal cavities, oral cavity, skin, orbit, and skull base. As a result, the variety of defects resulting from oncologic resection was also large. In general, complication rates were on par with free flap reconstructions of other head and neck sites.11 The flap loss rate was 3.3 percent and there did not appear to be a consistent cause that would suggest specific risks for free flap loss in the midface. One difference from other head and neck reconstructions is that a flap loss for a palatal defect can sometimes be treated with an obturator (see below).
Suboptimal speech outcomes were associated with loss of 50 percent or more of the soft palate.12 Many patients with soft palatal defects still attained greater than 80 percent intelligibility, but this required either closure of a partial palatal defect or obturation of the soft palate defect in addition to free flap reconstruction. The need for dental restoration and type of prosthetic retention required was predicted by the degree of palatoalveolar resection. All but a few patients, who also had oral cavity resections or neurologic deficits, achieved a soft or unrestricted diet following reconstruction.
As with all reconstructions, stable wound closure, especially when dura or brain is exposed, and restoring aesthetic appearance are major goals. Goals specific to maxillary reconstruction depend on the extent of the defect and may include separating the oral cavity from the sinonasal cavities, facilitating dental restoration with either conventionally supported or implant-retained prosthetics, and supporting the orbit(s) to prevent diplopia and enophthalmos, or separating the orbit from the sinonasal cavities if exenterated.13,14 Besides assessing overall outcomes, our objective was to evaluate reconstructive techniques, including flap and graft choices, used for the various defects we encountered to arrive at a treatment algorithm based on lessons learned from both our successes and our failures at achieving these reconstructive goals.
According to Archibald et al.,15 the ideal maxillary treatment algorithm should “use a functional approach that defines the defect-related challenges, prioritizes the reconstructive goals, and identifies how and to what extent each of the microsurgical free flaps could meet these goals.” Several landmark articles have proposed classification schemes for midfacial defects.7–10 What these articles share in common is an understanding that there are basically three major components to maxillary defects, each with specific reconstructive goals, that must be assessed and considered in the surgical plan: the extent of the palatoalveolar defect, the status of the orbital floor, and whether or not an orbital exenteration is performed in conjunction with a maxillectomy.8
A reconstructive algorithm based on addressing each of these three maxillary components is presented in Figure 3. Rather than recommending a specific flap for each defect type, possible flap choices are given, because flap selection is based on a number of factors, including donor-site availability, donor-site morbidity, body habitus, and surgeon preference, as we noted in the present series. For soft-tissue free flaps, the anterolateral thigh and vertical rectus abdominis myocutaneous flaps were assumed to be thick or bulky free flaps and the radial forearm fasciocutaneous flap was assumed to be a thin free flap. Specific reconstructive needs of each type of defect are discussed below.
Maxillary resections include those that do not involve the palate at all, involve a portion of the palate, or involve the entire palate. Although any number of partial palatoalveolar defects are possible, Okay et al.4 have recommended distinguishing defects based on whether function can be satisfactorily restored with an obturator or if a free flap is required. Palatoalveolar defects that spare both canine teeth can usually be successfully treated with an obturator. In these cases, cantilever forces resulting in unstable prosthetic retention are minimized because of the favorable root morphology of the canine adjacent to the obturator and the generous arch length provided by the remaining alveolus. Thus, defects including unilateral posterior palatomaxillectomy defects and anterior defects limited to the premaxilla, which bears the four incisor teeth, can be obturated and should be considered separately from those that cannot, including those that involve half the palate and those that involve the entire anterior arch or whole palate.
Suprastructure maxillectomies result in defects that do not involve the palate. Defects that do not include the orbital floor or orbital contents usually do not need reconstruction. Exceptions include when facial soft tissues are included in the resection, and when intracranial contents at the skull base have been exposed, as occurred in five cases in our series (data not shown). In the later case, a bulky soft-tissue free flap that obliterates the maxillary sinus was successful in isolating the intracranial cavity from the nasal cavity by creating a watertight seal against the dura or brain, thereby preventing cerebrospinal fluid leaks and meningitis.16
Premaxillary defects were rare in our series, probably in part because many of these defects received obturation alone.4 We treated these defects with either soft-tissue free flaps or a radial forearm osteocutaneous free flap. Soft-tissue reconstruction can be combined with a dental prosthesis that clasps to the remaining teeth to provide midfacial projection. The advantage of an osteocutaneous flap in this area would be maintenance of upper lip and nose support without a prosthesis and the possibility of osseointegrated implant dental restoration.6
Unilateral Posterior Palatomaxillectomy
As mentioned, unilateral posterior palatomaxillectomy defects can usually be treated with an obturator. However, because of inadequate stability of the obturator in some patients, especially edentulous ones, and long-term costs associated with periodic adjustment and replacement of the prosthesis, autologous tissue may still be preferred.1–3 In addition, exposure of the intracranial contents, loss of the orbital floor or orbital contents, and resection of the facial soft tissues are indications for free flap reconstruction.
We reconstructed all posterior palatomaxillectomy defects with soft-tissue rather than bony free flaps (Fig. 4). The aesthetic challenge is usually to provide adequate volume to the cheek to support the facial soft tissues and avoid a hollow appearance. An analogous situation is present in the mandible, where posterior mandibular reconstruction with soft-tissue flaps can often achieve good results with regard to both function and appearance, provided the flap has adequate bulk.17,18 As with the mandible, restoration of posterior maxillary dentition, which is not easily visible even when smiling, is not a priority to many patients.
Unlike unilateral posterior palatomaxillectomy defects, unilateral hemipalatomaxillectomy defects are difficult to obturate because of the greater cantilever forces acting on the prosthesis, which must rely on less dentition for retention.4 This is consistent with our experience that many patients who desire dental restoration were unable to accommodate a conventionally retained prosthesis, in contrast to patients who underwent a posterior palatomaxillectomy. Our reconstructive flap choices were mixed for hemipalatomaxillectomy defects. Soft-tissue flaps are usually more straightforward surgically. However, they do not provide a rigid skeletal framework, which can result in a loss of anterior maxillary projection on the side of the defect, and cannot accept osseointegrated implants to help retain dental prostheses. To accommodate a conventional prosthesis, the soft-tissue flap must not protrude excessively into the oral cavity. However, achieving a concave palatal reconstruction with soft-tissue flaps can be technically challenging, especially if the lateral portion of the defect includes some or all of the buccal mucosa.19
Although we treated only a few hemipalatal defects with osteocutaneous free flaps, we now favor their use in highly functional patients with a good prognosis. Besides providing better anterior projection, osteocutaneous free flaps offer the possibility of osseointegrated implants for dental restoration. For these reasons, other authors20,21 have advocated osteocutaneous free flap reconstruction for these types of defects. A caveat is that postoperative radiation therapy may render placement of osseointegrated implants risky, defeating one of the main purposes of bony reconstruction. In such cases, options include initial placement of an obturator followed by delayed osteocutaneous free flap reconstruction after the conclusion of irradiation, immediate free flap reconstruction with early osseointegrated implant placement before beginning radiation therapy, or immediate osseointegrated implant placement at the time of free flap reconstruction (not performed in our series).
Bilateral palatomaxillectomy defects that involve loss of the anterior maxillary alveolar arch, including the canine teeth, require bony reconstruction to maintain midfacial height, width, and projection.4 They also require bony reconstruction for dental restoration with osseointegrated implants, which were necessary to retain a prosthesis in all but a few patients in our experience.22 Although other osteocutaneous free flaps have been advocated,13,20,21,23 the fibula free flap is our preferred flap for bilateral palatomaxillectomy reconstruction for the same reasons it is preferred for mandibular reconstruction in most institutions: acceptable donor-site morbidity, ability to perform simultaneous harvest and recipient-site preparation using two surgical teams, and a long length of bone with adequate thickness to accept osseointegrated implants (Fig. 5).24–28 Bilateral palatomaxillectomy defects can rarely, if ever, be treated with obturation alone.4
Orbital Floor Defects
Our findings suggest that, when supported by a soft-tissue free flap, the orbital floor can usually be successfully reconstructed with bone grafts or alloplasts, even when radiation therapy is administered (Fig. 6). Although many surgeons feel that bone grafts are relatively more resistant to radiation-associated complications than are alloplasts, our results suggest that complication rates are similar regardless of the material used.
In a recent review of orbital floor reconstruction for trauma, Kirby et al.29 found that autologous bone reconstructions were more likely to be complicated by orbital dystopia and enophthalmos compared with titanium mesh and porous polyethylene, possibly because of increased difficulty in shaping the reconstructed orbital floor, irregular thickness, and unpredictable resorption. Our experience with three patients in whom a rigid orbital floor reconstruction was not performed who all developed orbital dystopia or enophthalmos demonstrates that a soft-tissue free flap alone is not adequate to support the orbital contents.
When using the fibula osteocutaneous free flap for reconstructing hemipalatomaxillectomy and bilateral palatomaxillectomy defects that include resection of the orbital floor, we include some soleus or flexor hallucis longus muscle with the fibula osteocutaneous free flap to support a bone graft or alloplastic orbital floor reconstruction.25 A double-barreled design to reconstruct both the floor and the hard palate is possible26 but challenging because of the limited space in the midface. Patients who have had a complication following nonvascularized reconstruction can usually undergo reconstruction secondarily with a bony free flap, such as the serratus anterior muscle free flap.
Orbital Exenteration Defects
Orbital exenteration performed in concert with a maxillectomy adds another reconstructive priority: closure of the orbital wound to prevent the escape of air and nasal drainage.13,30 Because the formation of a nasocutaneous fistula can be difficult to treat, every effort should be made to obtain a secure closure around the orbit. We reserve thin flaps that result in a concave orbital cavity for patients with isolated orbital exenteration defects that wish to have an orbital prosthesis.30 For patients with orbitomaxillectomy defects, we recommend bulkier flaps that obliterate the orbital cavity to minimize the chance of a fistula and to maintain cheek contour (Fig. 7). Defects involving both an orbital defect and a palatal defect are best reconstructed with flaps that allow for multiple skin paddles to resurface the two defects separately.
Midfacial reconstruction remains one of the most challenging areas of reconstructive surgery because of the complex anatomy and varied nature of oncologic resections performed for tumors occurring in this region. We performed an analysis of outcomes following 246 consecutive midfacial reconstructions to ultimately arrive at a defect-oriented reconstructive algorithm that is based on restoring the various functions of the midface with microvascular free flaps.
Patients gave written informed consent for the use of their images.
The authors thank their colleagues who contributed patients to this study: Drs. David Adelman, Donald Baumann, Elisabeth Beahm, Charles Butler, David Chang, Melissa Crosby, Patrick Garvey, Steven Kronowitz, Scott Oates, Gregory Reece, Geoffrey Robb, Jesse Selber, and Mark Villa.
1. Moreno MA, Skoracki RJ, Hanna EY, Hanasono MM. Microvascular free flap reconstruction versus palatal obturation for maxillectomy defects. Head Neck 2010;32:860–868.
2. Genden EM, Okay D, Stepp MT, et al.. Comparison of functional and quality-of-life outcomes in patients with and without palatomaxillary reconstruction: A preliminary report. Arch Otolaryngol Head Neck Surg. 2003;129:775–780.
3. Rogers SN, Lowe D, McNally D, Brown JS, Vaughan ED. Health-related quality of life after maxillectomy: A comparison between prosthetic obturation and free flap. J Oral Maxillofac Surg. 2003;61:174–181.
4. Okay DJ, Genden E, Buchbinder D, Urken M. Prosthodontic guideline for surgical reconstruction of the maxilla: A classification system of defects. J Prosthet Dent. 2001;86:352–363.
5. Brown JS, Rogers SN, McNally DN, Boyle M. A modified classification for the maxillectomy defect. Head Neck 2000;22:17–26.
6. Cordeiro PG, Santamaria E. A classification system and algorithm for reconstruction of maxillectomy and midfacial defects. Plast Reconstr Surg. 2000;105:2331–2346; discussion 2347–2348.
7. Davison SP, Sherris DA, Meland NB. An algorithm for maxillectomy defect reconstruction. Laryngoscope 1998;108:215–219.
8. Futran ND, Mendez E. Developments in reconstruction of midface and maxilla. Lancet Oncol. 2006;7:249–258.
9. Brown JS, Shaw RJ. Reconstruction of the maxilla and midface: Introducing a new classification. Lancet Oncol. 2010;11:1001–1008.
10. Cordeiro PG, Chen CM. A 15-year review of midface reconstruction after total and subtotal maxillectomy: Part I. Algorithm and outcomes. Plast Reconstr Surg. 2012;129:124–136.
11. Zafereo ME, Weber RS, Lewin JS, Roberts DB, Hanasono MM. Complications and functional outcomes following complex oropharyngeal reconstruction. Head Neck 2010;32:1003–1011.
12. Hanasono MM, Weinstock YE, Yu P. Reconstruction of extensive head and neck defects with multiple simultaneous free flaps. Plast Reconstr Surg. 2008;122:1739–1746.
13. Yoshida H, Michi K, Yamashita Y, Ohno K. A comparison of surgical and prosthetic treatment for speech disorders attributable to surgically acquired soft palate defects. J Oral Maxillofac Surg. 1993;51:361–365.
14. McCarthy CM, Cordeiro PG. Microvascular reconstruction of oncologic defects of the midface. Plast Reconstr Surg. 2010;126:1947–1959.
15. Archibald S, Jackson S, Thoma A. Paranasal sinus and midfacial reconstruction. Clin Plast Surg. 2005;32:309–325.
16. Hanasono MM, Silva AK, Skoracki RJ, et al.. Skull base reconstruction: An updated approach. Plast Reconstr Surg. 2011;128:675–686.
17. Mosahebi A, Chaudry A, McCarthy CM, et al.. Reconstruction of extensive composite posterolateral mandibular defects using nonosseous free tissue transfer. Plast Reconstr Surg. 2009;124:1571–1577.
18. Hanasono MM, Zevallos JP, Skoracki RJ, Yu P. A prospective analysis of bony versus soft-tissue reconstruction for posterior mandibular defects. Plast Reconstr Surg. 2010;125:1413–1421.
19. Cordeiro PG, Santamaria E, Kraus DH, Strong EW, Shah JP. Reconstruction of total maxillectomy defects with preservation of the orbital contents. Plast Reconstr Surg. 1998;102:1874–1884; discussion 1885–1887.
20. Genden EM, Wallace D, Buchbinder D, Okay D, Urken ML. Iliac crest internal oblique osteomusculocutaneous free flap reconstruction of the postablative palatomaxillary defect. Arch Otolaryngol Head Neck Surg. 2001;127:854–861.
21. Brown JS, Jones DC, Summerwill A, et al.. Vascularized iliac crest with internal oblique muscle for immediate reconstruction after maxillectomy. Br J Oral Maxillofac Surg. 2002;40:183–190.
22. Chang YM, Coskunfirat OK, Wei FC, Tsai CY, Lin HN. Maxillary reconstruction with fibula osteoseptocutaneous free flap and simultaneous insertion of osseointegrated dental implants. Plast Reconstr Surg. 2004;113:1140–1145.
23. Yamamoto Y, Kawashima K, Sugihara T, Nohira K, Furuta Y, Fukuda S. Surgical management of maxillectomy defects based on the concept of buttress reconstruction. Head Neck 2004;26:247–256.
24. Peng X, Mao C, Yu GY, Guo CB, Huang MX, Zhang Y. Maxillary reconstruction with the free fibula flap. Plast Reconstr Surg. 2005;115:1562–1569.
25. Yazar S, Cheng MH, Wei FC, Hao SP, Chang KP. Osteomyocutaneous peroneal artery perforator flap for reconstruction of composite maxillary defects. Head Neck 2006;28:297–304.
26. Rodriguez ED, Martin M, Bluebond-Langner R, Khalifeh M, Sing N, Manson PN. Microsurgical reconstruction of posttraumatic high-energy maxillary defects: Establishing the effectiveness of early reconstruction. Plast Reconstr Surg. 2007;120(Suppl 2):103S–117S.
27. Hanasono MM, Skoracki RJ. The omega-shaped fibula osteocutaneous free flap for reconstruction of extensive midfacial defects. Plast Reconstr Surg. 2010;125:160e–162e.
28. Hanasono MM, Jacob RF, Bidaut L, Robb GL, Skoracki RJ. Midfacial reconstruction using virtual planning, rapid prototype modeling, and stereotactic navigation. Plast Reconstr Surg. 2010;126:2002–2006.
29. Kirby EJ, Turner JB, Davenport DL, Vasconez HC. Orbital floor fractures: Outcomes of reconstruction. Ann Plast Surg. 2011;66:508–512.
30. Hanasono MM, Lee JC, Yang JS, Skoracki RJ, Reece GP, Esmaeli B. An algorithmic approach to reconstructive surgery and prosthetic rehabilitation after orbital exenteration. Plast Reconstr Surg. 2009;123:98–105.