Arden RL, Rachel JD, Marks SC, Dang K. Volume-Length Impact of Lateral Jaw Resections on Complication Rates. Arch Otolaryngol Head Neck Surg. 1999;125(1):68–72. doi:10.1001/archotol.125.1.68
To study the relationship between soft tissue volume loss and bone resection length following lateral segmental mandibulectomy with plate reconstruction and complication rates.
Retrospective case review of 31 patients (1989-1996), with average follow-up of 37.2 months, who were treated by lateral composite resection for oral cavity and/or oropharyngeal malignancy with primary reconstruction by defect-bridging plates.
Academic tertiary care referral center.
Thirty patients had stainless steel and 1 patient a vitallium reconstruction plate to restore mandibular continuity. Soft tissue defects were repaired with pectoralis myocutaneous flaps (n = 25), skin grafts (n = 4), a radial forearm free flap (n = 1), or primary closure (n = 1). All patients received preoperative (n = 6) or postoperative (n = 25) radiation therapy.
Main Outcome Measures
Overall and hardware-related complications.
All 31 initial soft tissue repairs were successful. Subsequent complications occurred in 14 patients (45%), which included plate exposure (29%), loosened screws requiring hardware removal (29%), fistula (14%), local wound infection (14%), osteomyelitis (7%), and plate fracture (7%). Average time to complication was 7.7 months. Complication rates were 81% for bone defects greater than 5.0 cm, and 7% for those less than 5.0 cm. Bivariate analysis indicated bone resection lengths greater than 5.0 cm to be a significant predictor of both hardware-related (P = .02) and overall complications (P = .005), whereas soft tissue volume resections greater than 240 cm3 were found only to be marginally significant (P = .04) for overall complications.
Extirpative losses involving more than 5 cm of bone, or tissue volume greater than 240 cm3, are associated with unacceptably high complication rates when reconstructed with solid screw stainless steel plates and this warrants consideration of alternative techniques for long-term stability.
SINCE THEIR inception clinically in 1976,1 3-dimensional bendable mandibular reconstruction plates (3-D MRPs) have afforded a means of providing immediate restoration of mandibular continuity without imparting additional donor site morbidity. This feature, coupled with generalized availability, ease of application, and biocompatibility, fostered widespread use of this stabilization system for both anterior and lateral mandibular defects. Recent evidence, however, suggests high complication rates (30%-50%)2- 4 when this technique is applied to anterior segmental losses, which has been related to (1) absent or dysfunctional mandibular depressors leading to upward pressure on the overlying soft tissue cover; (2) denervation or ptosis of the lower lip, in conjunction with gravity and scar contraction, favoring increased pressure of the tissues against the plate; and (3) tenuous blood supply to the distal end of a regional intraoral flap favoring ischemic necrosis at the suture line.
By contrast, acceptable complication rates for lateral mandibulectomy defects following MRP application are typically reported at between 5% and 10%.2- 4 Other smaller series, however, have reported plate-related complications of 25%5 and 40%,6 both of which exclusively included combined body-ramus defects. It has been suggested that the larger mandibulectomy defects would more permissively favor medialization of the overlying cheek skin during wound contraction over a greater dead space, thereby promoting pressure necrosis about the MRP.6
Support for a relationship between plate-exposure rates and the extent of soft tissue resection has been reported; however, statistical analysis was lacking.5,7 Similar deficiencies exist in the literature as it pertains to quantitating the bony component of composite oromandibular defects.
Instead, classification schemas are based on anatomical, functional, and aesthetic considerations.
With both the HCL system (hemi-, central, and lateral)4 and the RBSP system (ramus, body, symphysis, and palate),8 variations in defect length within a given subsite are not considered, and therefore rarely reported in the literature.
The purpose of this study was to assess the relationship between soft tissue volume loss and mandibular resection length following lateral composite resection with primary plate reconstruction and complication rates. Furthermore, we hoped to determine whether these variables, either singly or together, can be predictive of postoperative complications and, therefore, guide the reconstruction technique.
This series includes 31 patients who underwent lateral segmental resection (posterior to mental foramen) as part of a composite procedure for oral cavity and/or oropharyngeal squamous cell cancer performed between 1989 and 1996, with available follow-up extended to June 1997.
All patients had immediate reconstruction of the resultant bony defect with a rigid reconstruction plate (30 A-O stainless steel, 1 Luhr vitallium MRP). Surgical application of the MRPs in this series was distributed between 6 academic otolaryngology–head and neck surgeons. Data were obtained retrospectively by reviewing patient's hospital and clinic charts, operative reports, and pathology reports. Direct ruler measurements of bone length (before decalcification) and soft tissue resection volume (inclusive of bone segment) of the primary tumor (before formalinization) were provided in the gross pathologic descriptions. Calculations of bone length were made along the midlongitudinal axis of the lateral aspect of the entire mandibulectomy segment in each case. Soft tissue volumes were grossly derived from measurements obtained from length, width, and height of the resected specimen (exclusive of neck). The patient composition included 21 men and 10 women whose tumors were classified as stages III and IV in 13 and 15 patients, respectively. Three patients presented with tumor recurrence following irradiation failure. Patient ages ranged from 41 to 76 years (mean age, 57 years).
In all cases of MRP application, either 3 or 4 solid screws (8-14 mm) were applied bicortically to the mandibular remnants for plate stability. Maxillomandibular relationships were maintained in most cases by predrilling holes in the buccal cortex prior to segmental resection after plate contouring. In those patients in whom this was not performed, a best-fit application based on surface contour and dental relationships was used.
Repair of the soft tissue defects was achieved through the use of 25 myocutaneous flaps (all pectoralis major), 4 skin grafts (3 split thickness, 1 dermal), and 1 radial forearm free flap, and by primary closure in 1 case. The choice of repair varied among surgeons; however, a clear bias toward using skin grafts for smaller volume defects (<200 cm3) was evident and applied to lining defects of the lateral pharynx or floor of the mouth. In each case the skin graft was stabilized to its recipient bed using a tie-over bolster technique. Myocutaneous flap insetting was achieved by marginal approximation to the defect edges without the use of suspension or fixation sutures to the MRP.
External beam radiation therapy was delivered to all patients in this series. Six patients received radiation treatment preoperatively, and 25 received irradiation postpoperatively, with an average dose of 58.4 Gy (range, 45-70 Gy) to the primary site. Brachytherapy and hyperfractionation techniques were not used in any of these treatment regimens.
Statistical analysis was carried out by initially identifying baseline factors that could impact rates of postoperative complications. These univariate descriptions included resection characteristics (bone length and primary resection volume), dental status, maximal diet achieved, and nutritional status. Bivariate screening of all independent variables, with both hardware-related and overall complications as dependent variables, were used to identify those significant variables that were then analyzed by conjunctive consolidation (cluster technique).9 To conduct a cluster analysis, length and volume were transformed from a continuous to a binary scale. Length was categorized as those resections 5.0 cm or less or greater than 5.0 cm. Primary resection volume was delineated as either 240 cm3 or less or greater than 240 cm3.
Patient characteristics, TNM classification, tumor location, resected lengths and volumes of the primary specimen, methods of treatment, and outcomes are listed in Table 1.
The most common neoplasm in this series was squamous cell carcinoma of the tonsil (52%), followed by oral tongue (19%), floor of mouth (13%), base of tongue (10%), and retromolar trigone (6%). Primary tumor resection volumes ranged from 95 to 629 cm3 (mean, 226 cm3). Segmental mandibulectomies ranged from 3.0 to 10.0 cm (mean, 5.4 cm). Most of the bony resections were combined body-ramus subunits (55%), none of which included the condylar head or neck. Body resections constituted 42%, and a single ramus resection accounted for 3% of the total. All patients received radiation therapy as part of their treatment and no significant difference was found in radiation dosage between complicated cases (average dose, 56.7 Gy) and uncomplicated cases (average dose, 59.5 Gy). Similarly, no significant difference was noted between the incidence of complications and the timing of irradiation (ie, preoperative vs postoperative). Uncomplicated cases (55%) have been followed up an average of 37 months (range, 10 months to 7 years 9 months).
Complications occurred in 14 (45%) of 31 patients with an average postoperative interval of 7.7 months (range, 1-18 months), which are listed in Table 2. A dominant trend between a particular surgeon and postoperative complications was not found. In order of descending incidence, the complications included plate exposure and loose screws (28.6% each), fistula formation and local wound infection (14.3% each), and a single case each of plate fracture and osteomyelitis (7.1%). All plate exposures in this series developed intraorally in the intermediate or late postoperative period. Of the patients who developed hardware failure or wound-related complications, comorbid factors were identified as diabetes mellitus in 29%, moderate-severe peripheral vascular disease in 21%, and absent in 50%. In no case was tumor recurrence found to be the cause of these complications. A χ2 goodness of fit test at P<.05 did not demonstrate a significant relationship between the incidence of complications and postoperative distribution intervals of 0 to 6, 6 to 12, and 12 to 18 months.
Table 3 lists the baseline factors and related complications for the study group that considers both overall and hardware-related events. Initial bivariate screening of all independent variables with hardware complications as the dependent variable yielded only one variable (resected bone length) to be significant at the .05 level (P = .02). However, primary resection volume demonstrated borderline insignificance (P = .06). Since it was significant when overall complications were used as the dependent variable, it was included in subsequent analysis for hardware complications as well.
Table 4 depicts a comparison of effect size between aggregating length and volume as opposed to the individual strengths of their relationship to hardware complications. It is apparent through this table that a 3-level split with conjunctive consolidation yielded a concise and interpretable representation of the relation between the length and volume variables and the outcome. The exact χ2 value calculated for the χ2 test for trend was highly significant (χ2 = 8.18; P = .004).
The same analytic process was used to evaluate the outcomes of all complications. Preliminary analysis of all independent variables from bivariate analysis eliminated most independent variables because they did not exhibit a significant effect at the .05 level. Only 2 variables were found to be significant: volume (P = .04), which was marginally significant, and length (P = .005). Table 5 shows the conjoined effect of length and volume, as well as their individual effects on overall complications. The χ2 for linear trend was also found to be highly significant (χ2 = 11.24; P = .001).
There are many studies in the literature reviewing MRPs as an option for immediate restoration of continuity following resection for malignant tumors. The potential for complications is well documented and variably related to the area of resection (anterior vs lateral), type of defect-bridging system used (stainless steel vs titanium hollow osteointegratin reconstruction plate [THORP]), extent of the tumor resection, delivery of radiation to the primary field, and design of the soft tissue reconstruction. More favorable outcomes are usually anticipated with smaller, nonirradiated, lateral defects that allow for either primary closure or placement of a well-vascularized soft tissue flap. Owing to the reportedly high success rates of MRPs in lateral defect repairs, and the potential confounding interplay of these variables, decisions regarding preferable methods of repair remain elusive when extracted from the literature.
Kellman and Gullane11 reported a 17% failure rate (4 of 23 A-O MRPs) for lateral defect repairs following tumor ablative surgery, noting an average plate retention interval of 8 to 10 months. The details of this series were not reported other than with mention of a "generally subcondylar" defect that did not include the symphysis. Schusterman et al2 reported a 7% failure rate (single extrusion related to flap loss) among 14 posterior/lateral repairs with A-O MRPs. Of note, the patient group as a whole was relatively young (average age, 48 years), and only 3 patients in this series underwent postoperative radiotherapy to the plate. In a well-described study that included 20 L-type defects, Boyd et al4 reported a 5% failure rate. In this select group, 80% of patients received radiation therapy, all underwent free fasciocutaneous flap reconstructions for their soft tissue defect, and had mandibular continuity reestablished with either stainless steel or THORP MRPs (trend toward using latter for larger defects). More recently, Blackwell et al6 reported a 40% delayed failure rate in 10 patients (3 external plate exposures and 1 plate fracture), reconstructed with THORP plates and a soft tissue free flap, who were followed up for a minimum of 12 months.
Our patient population consisted of a relatively homogeneous composition with advanced stage oral cavity and/or oropharyngeal malignancies who underwent similar surgical exposures and methods of stabilization and soft tissue reconstruction, and were subjected to similar doses of radiation therapy. The plate-related failure rates reported herein were similar to other reported series. A plate fracture rate of 7.1% is consistent with the 3% to 10% reported in the literature.6 Although our screw fixation failure rate (13% overall incidence, 28.6% of complications) compares with the findings of Soderholm et al12 of 30%, it seems high relative to the same plate exposure rate. This may be explained by the number of different surgeons included in this series and the possibility of technical errors (ie, overdrilling holes or inadequate cooling during drilling) leading to suboptimal bone-screw contact. This supposition is supported by the trend toward early development of this complication relative to the plate exposure rates.
The expectation of a high plate exposure rate in patients requiring more extensive ablative procedures is supported by the 4 patients evidencing this complication. In this series overall, soft tissue volume resections and bone resection lengths averaged 226 cm3 and 5.4 cm, respectively. The patients who developed plate exposures averaged considerably higher tissue losses (365 cm3 soft tissue, 7.35 cm bone). The 4 delayed plated exposures in this series manifested intraorally. While certain series detailed only external plate exposures,6 others have noted intraoral exposures only13 or combined exposures.3,5,14 Our study suggests that absolute bone loss, and possibly primary resection volume, is the main determinant(s) of plate-related failures using the A-O stainless steel system. The relative impact of soft tissue volume loss on complication rates could have been more strongly stated had volume analysis been carried out by displacement methodology that was not possible in this retrospective study.
Multiple factors can be offered to explain the increased plate failure rates seen with increased lateral mandibular losses.
Altered masticatory force vectors—lateral segmental resection usually leaves depressor muscles largely intact (although denervated). The stabilizing effect of the masseteric and medial pterygoid sling is usually lost owing to either resection or inability to functionally reattach to the MRP. The action of the lateral pterygoid favors medial and upward rotation of the condyle. Net effect is a rotational stress that favors inferior displacement of the distal plate and a lateral force on the proximal plate.
Difficulties in contouring and stabilizing plate 3 dimensionally—inability to account for lateral splay of the ramus relative to the sagittal plane of the mandibular body leads to further unbalanced force vectors. Contouring prior to mandibulectomy will minimize these problems, but high osteotomies may make contouring difficult, allowing room for only 2 stabilization screws. A more limited bony loss is easier to contour and stabilize.
Increased dead space—with greater mandibular losses, associated wound contraction and fibrosis, enhanced by radiation therapy, favors inward contraction of the overlying soft tissue against a rigid bar making plate exposure more likely.
Decreased vascularity at alloplastic interface—the extensive periosteal stripping required for MRP application, local tissue hypoxic environment created by radiation therapy, and interruption of the endosteal blood supply (inferior alveolar artery), favor bony resorption about the screw purchase sites. Predictably, the latter effect would be greatest with fixation screws placed proximal to the mental foramen, on the side of resection, secondary to creation of a "watershed zone." Blood supply in this zone is now reliant on a random blood supply of the overlying cheek soft tissues and hypoperfusion from the contralateral nutrient artery.
Lack of true osseointegration—this condition, associated with the solid screw A-O system, compounds the aforementioned direct factors relating to greater lateral bony losses.
The potential impact of nutritional status on complication rates has been considered and demonstrated to be insignificant in bivariate analysis. Clearly, protein malnutrition can negatively impact wound healing, primarily due to impaired collagen synthesis. Intuitively, one would predict related complications (ie, wound dehiscence, local infection, or fistula formation) in the early postoperative interval, since all patients were supported nutritionally postoperatively. Our series shows only 2 related complications (fistulas) in the early postoperative interval, both occurring 1 month after surgery. In each case, the orocutaneous fistula was small, healed spontaneously without surgical intervention, and was unassociated with other wound-healing problems. Also, the delayed onset of plate exposure, loose screws, and plate fracture would not be expected to be related to a negative nitrogen balance after primary wound healing had taken place.
Despite 20 years of experience with MRPs, significant controversy remains concerning their use. Part of this controversy may relate to the lack of predictable outcomes often seen with larger extirpative procedures. We believe that this study identifies 2 factors (resected bone length and tissue volume) strongly predictive of adverse outcomes at the reconstructive site. The results suggest that for cancer resection in which the mandibular defect is expected to be more than 5 cm, or tissue volume greater than 240 cm3, solid screw stainless steel plates are associated with unacceptably high complication rates and this warrants consideration of alternative techniques of reconstruction for long-term stability.
Accepted for publication July 16, 1998.
We acknowledge Joel Ager, PhD, for assistance in performing the statistical analysis for this project.
Reprints: Richard L. Arden, MD, Department of Otolaryngology–Head and Neck Surgery, Wayne State University, 540 E Canfield, 5E-UHC, Detroit, MI 48201.