Customize your JAMA Network experience by selecting one or more topics from the list below.
KAREN H.CALHOUNMDRONALD B.KUPPERSMITHMD
When resecting a floor-of-mouth (FOM) squamous cell carcinoma that abuts a nonirradiated mandible, if neither clinical examination nor preoperative studies have shown evidence of bony tumor invasion and if the periosteum is histologically positive for cancer on frozen section and the underlying bone is grossly normal, a segmental resection of the mandible is required. A marginal resection is never appropriate.
Historically, when an FOM tumor abutted the mandible, a complete segment of mandibular bone was resected en bloc with the tumor.1 It was believed that the FOM lymphatics drained through the mandibular periosteum and bone to the neck, so that en bloc resection decreased the risk of later neck disease. Unfortunately, the segmental mandibulectomy left the patient with a significant functional and aesthetic deficit. Once Marchetta et al2 and Carter et al3 demonstrated that cancer spreads to the mandible by direct invasion rather than lymphatic spread, preservation or partial resection of the mandible became oncologically feasible. The question became how to select appropriate patients with oral cancer for segmental or partial-thickness (marginal) mandibular resection.
Mark K. Wax, MD
Oncologic surgeons agree that mandibular bone grossly eroded by tumor requires a segmental resection, and most agree that normal bone separated from an oral cancer by grossly normal mucosa can be safely left intact. When tumor abuts bone, however, or involves periosteum, the decision is less clear. The head and neck oncologic surgeon currently has available several options to evaluate mandibular involvement: (1) clinical evaluation, (2) preoperative radiologic examinations, (3) gross inspection intraoperatively, and (4) frozen section and/or touch-preparation evaluation of cancellous bone from the resected bony margins.
Larry L. Myers, MD
We conducted a MEDLINE literature search for the years 1980 through 2001. The following search string was used: "1) Mouth neoplasms OR 2) oral cavity neoplasms AND 3) squamous cell carcinoma AND 4) mandible." The search was limited to the English language. This strategy yielded 1472 references, from which the appropriate articles were chosen. Citations from each chosen article were also examined for appropriate references. A second search using the keywords "squamous cell carcinoma AND mandible" yielded 110 references. A third search using the keywords "oral cavity neoplasms OR mouth neoplasms OR mouth floor" yielded 31 959 references. When we cross searched for periosteum only, 18 English-language references were identified. The results of our literature analysis as they illuminate the hypothesis are given below.
Patient characteristics that influence the decision to perform a segmental vs a marginal mandibulectomy include tumor approximation and/or fixation to the underlying bony structure, the presence or absence of dentition, and previous mandibular irradiation. Weisberger and Kimmelman4 suggest that 33% of specimens with histologic evidence of mandibular involvement do not show clinical signs of preoperative bone invasion. These data were supported by Tsue et al,5 who evaluated the relationship between clinical assessment of tumor mobility and histologic evidence of bone invasion. They found that 34% of patients (11/32) judged clinically to have a freely mobile tumor were subsequently found to have histologic evidence of mandibular invasion. This must be considered in the context of only a 44% correlation (14/32) between the clinical assessment of tumor fixation and positive mandibular invasion. These findings suggest that clinical examination alone is an inaccurate method of predicting mandibular invasion by oral cavity cancer.
In the nonirradiated mandible, bony invasion takes place mainly through the occlusal plane, in dental interstices. When the mandible is edentulous, the bony defects that previously housed teeth provide many more potential entry ports for tumor into the mandible, so there is a higher rate of tumor spread into the edentulous mandible. Histologic examination of decalcified specimens (with or without teeth) generally shows only 1 focus of tumor entry into bone from which all other bony spread occurs. On the other hand, histologic examination of previously irradiated mandible often shows multifocal involvement of bone by tumor, suggesting that irradiation alters the bone's resistance to tumor spread. Most oncologic surgeons agree that a partial-thickness resection of previously irradiated mandible would carry too great a risk of leaving tumor behind. When oral cancer abuts a previously irradiated mandible, a segmental mandibulectomy is indicated.
Tsue et al5 retrospectively evaluated 64 patients who had undergone composite resection to determine predictors of mandibular invasion. Thirty-nine percent of the mandibular specimens had tumor involvement. There was no correlation among T stage, N stage, or tumor differentiation and mandibular invasion, but bone invasion seen on computed tomography (CT) accurately predicted bone involvement. Patients with anterior FOM tumors and CT scans showing no bony involvement had tumor in the mandible 50% of the time. No mention was made in this study of periosteal invasion. Brown et al6 found that plain radiographs had a 14% false-negative rate and a 3% false-positive rate in the prediction of bone invasion.
The practical question is whether tumor in the periosteum on frozen section is a sensitive and specific indicator of tumor in the mandible. Brown et al6 in a comparison of multiple methods to assess bone invasion found that an exploratory stripping of the periosteum at the time of resection is a good predictor of the presence of mandibular invasion. In their evaluation of 35 cases of oral cavity squamous cell carcinoma (SCCA), there was only 1 false assessment of mandibular invasion. Based on these data, in the patient without clear evidence of mandibular invasion, a marginal resection followed by periosteal stripping and inspection could be performed. If macroscopic evidence of invasion at the rim is present or equivocal, the marginal mandibulectomy can be converted to a segmental resection. The degree of periosteal stripping on the in situ mandible should be limited because stripping of the mandibular periosteum may devitalize the remaining cortical bone and predispose the mandible to pathologic fractures.7
Schusterman et al8 evaluated 182 patients undergoing segmental mandibulectomy. Forty-five percent had tumor in the resected mandible on histologic examination. Only 2% of the patients had positive bone margins on final histologic analysis. This study demonstrated (1) the efficacy of segmental resection in completely removing the tumor, (2) the surgeon's inability to determine clinically whether mandibular bone is involved with tumor, and (3) the ability of the surgeon to obtain adequate bone margins.
Dubner and Heller9 retrospectively evaluated 130 patients undergoing marginal or segmental mandibulectomy SCCA of the oral cavity. Fifty patients had FOM lesions. Of these, 40 had marginal mandibulectomy and 10, segmental. Although periosteal invasion was not directly addressed, histologic mandibular invasion was examined. Most of the marginal lesions were early stage and involved the occlusal surface of the bone. Thirteen percent of the patients undergoing marginal mandibulectomies demonstrated histologic evidence of tumor invasion of the bone, whereas 66% of patients with segmental resection demonstrated bony involvement. The local recurrence rate was 6% in patients who had a segmental mandibulectomy and 19% following marginal mandibulectomy. Of the patients developing recurrent tumor after marginal mandibulectomy, surgical salvage brought the overall failure rate down to 6%. This study demonstrates the effectiveness of segmental resection when bone and/or periosteum is involved. Marginal mandibulectomy was effective when there was no invasion. Patients with bone invasion who had a marginal resection demonstrated a significantly higher local recurrence rate (22%).
O'Brien et al10 studied the histopathologic characteristics of 27 mandibles invaded by SCCA. The soft tissue around the mandible was usually fibrotic and infiltrated with inflammatory cells. The bone had clefts along the surface with foci of new bone formation on each side, sometimes projecting as spurs into the surrounding soft tissues. These clefts, apparently from previous local inflammation, provided access for tumor cell spread into the mandible. The first stage of bone destruction is mediated by local host osteoclasts, eroding bone in front of the advancing tumor edge. When this response declines, direct tumor cell invasion continues the bone destruction (second stage). These 2 stages often coexist in different parts of the same specimen. This information sheds doubt on the classic thinking that fixation of the tumor to the mandible means that there is tumor in the periosteum, and vice versa. In some specimens, without tumor mass fixation to the mandible, bone invasion was nevertheless present.
McGregor and MacDonald11 evaluated 16 irradiated mandibles with histologic evidence of tumor invasion. The roots of tumor entry were more variable than in the nonirradiated mandible. Multiple foci of tumor invasion of the bone were present wherever tumor and the adjacent soft tissues approached the bone.
Segmental resection leaves the patient with a potential functional defect. Reconstruction with a locoregional or microvascular free flap has an excellent chance of successfully restoring form and function.
At least 2 arguments can be made for less-than-segmental resection of the mandible in this situation. First, if the underlying bone is grossly normal, it is unlikely to be invaded by tumor. Second, if there is tumor invasion into grossly normal bone, the tumor load would probably be very small and amenable to sterilization by postoperative radiation.
Jones et al12 examined mandibular invasion in 82 patients with SCCA of the oral cavity or oropharynx. Periosteal invasion was not a variable measured in this study. However, the authors did find that clinical examination had a sensitivity of 91% and a negative predictive value of 93% in the detection of mandibular invasion. This is in contrast to radiographic assessment where the sensitivity and specificity was 84% and 88%, respectively. These data suggest that clinical examination is an accurate means of stratifying patients to the marginal vs segmental resection procedures.
Brown et al,6 in a multifactorial evaluation of different methods of predicting bone invasion, found that a combination of orthopantogram and bone scan can provide a useful screening tool. In their study of 35 consecutive patients with SCCA of the oral cavity, 21 patients were found to have invasion of the mandible, 7 of whom were treated with marginal resection. In this study the false-negative rates were 14.3% (5/35) based on orthopantogram, 2.9% (1/34) for bone scan, and 28% (7/25) for CT scan. Smyth et al13 attempted to assess the role of pantomography, CT, and intraoperative assessment in predicting the presence and extent of mandibular invasion in 40 patients with SCCA of the oral cavity or oral pharynx. They concluded that there is currently no radiographic evaluation that can accurately predict periosteal invasion and that intraoperative assessment is the best adjunct to a preoperative orthopantogram or CT scan. Computed tomography was found to predict periosteal invasion with a sensitivity of 52% and had a negative predictive value of 48%.
Unfortunately, most of the literature that describes and supports marginal mandibulectomy describes patients who undergo resection of a rim of mandible for clinically suspect tumor or to establish better soft tissue margins. In most of these instances, histologic evidence of periosteal and/or bony involvement is not present.
Ord et al14 studied 46 patients, 26 of whom had marginal resection. Of these, 54% had early-stage tumors, and 70% of these in the floor of the mouth. Histologically, 7.6% of the patients who underwent marginal resection demonstrated bone invasion, and 18% demonstrated extension to bone. While this was not defined, periosteal invasion was implied. Twenty patients underwent segmental resection. Ninety-five percent of the tumors were advanced stage, with 65% demonstrating histologic invasion of bone and a further 22% demonstrating periosteal invasion but no bone destruction. Survival rates were equal between the 2 groups, with 10% of the segmental group dying with local disease compared with 20% of the marginal group.
McGregor and MacDonald11 reviewed 46 nonirradiated mandibles with histologic evidence of tumor invasion. Forty-one were invaded through the occlusal surface. Small areas of involvement with limited bone or periosteal invasion were amenable to marginal mandibulectomy. This study also demonstrated that when tumor extended beyond the clinically obvious mucosal margin, it never extended more than 4 or 5 mm. Thus, a 5- to 10-mm margin of normal bone would be sufficient. Barttelbort et al15 evaluated 38 patients who underwent mandibulectomy. Twenty-one underwent marginal mandibulectomy while 17 underwent segmental resections. Half of the patients in the marginal group had early-stage tumors, while only 28% in the segmental group had early-stage tumors. Three patients in the marginal group demonstrated bone invasion, while 5 patients in the segmental group did so. No patient experienced recurrence at the site of the bone resection.
One of the largest barriers facing the oncologic surgeon in the assessment of bone involvement is the ability to obtain intraoperative frozen-section analysis of bone margins. The standard method of histopathologic analysis of bone margins requires a 7- to 10-day period of decalcification. Forrest et al16 reported the results of curettage of the cancellous bone in 16 patients, a total of 33 surgical margins. Using this method, the authors were able to predict the surgical margin in 97% of the specimens. More recently in their analysis of 30 patients and 61 surgical margins, Forrest et al17 found that frozen-section analysis of cancellous bone had a sensitivity of 88.9% and a specificity of 100%. Moreover, their results did not seem to be influenced by pervious irradiation. Theoretically, this technique could be used as an operative guide to the surgeon considering marginal mandibulectomy. If curettage of the cancellous bone underlying the site of the marginal resection yields positive findings, then a segmental resection is warranted. This technique has not been reported in the literature but may merit further evaluation. Weisberger and Kimmelman4 describe a method of processing bone margins using microwave technology to prepare the bone for histopathologic analysis. They found a 100% correlation between margins analyzed using the microwave technique with those prepared using traditional decalcification and permanent sections. Neither technique specifically addresses periosteal invasion. Moreover, all of the patients in both studies underwent segmental or composite resection.
Barttelbort et al7 initially explored the survival issue in a population of 38 patients, 21 of whom underwent rim mandibulectomy and 17, segmental resections. Bone invasion was identified in 3 of 21 patients in the rim mandibulectomy group and 5 of 17 patients in the segmental group. Local control rates were similar in the 2 populations, 75% and 64% for the rim and segmental groups, respectively.
Advocates of a marginal mandibulectomy in FOM carcinoma stress the importance of preserving the mandible for physiologic function and cosmetic appearance. While this is desirable, mandibular preservation must be oncologically sound and leave enough mandibular bone so that fractures do not occur. Barttelbort et al15 studied 10 fresh human mandibles. Using strained-gauge techniques, they determined that maintaining an inferior bone segment 10 mm thick eliminated fracture risk.
First, patients who have previously undergone radiotherapy demonstrate a different pattern of bone invasion. It is usually not through the occlusal surface, but rather at multiple points along the cortex of the mandible. Periosteal or bone invasion is an indication for segmental resection. In general, marginal mandibulectomy is contraindicated in an irradiated mandible.
Second, preoperative clinical or radiographic examination findings suggestive (or directly indicative) of mandibular involvement would indicate the need for segmental resection. On the other hand, if the preoperative clinical or radiographic findings do not suggest mandibular invasion, bone involvement must be evaluated intraoperatively. For this evaluation, the periosteum should be judiciously stripped from the mandible and the underlying bone examined. If there is definite evidence of bone invasion, the extent of the invasion will dictate whether a marginal or segmental resection is indicated. While not the focus of this discussion, adequate histologic examinations of perineural and medullary bone spread would indicate that segmental resection is required to obtain negative margins. If the bone appears grossly normal, the surgeon can obtain an intraoperative consultation on the bony periosteum. If the periosteum is negative, then either no resection or a marginal resection is indicated. If the periosteum is positive, marginal or segmental resection is indicated.
Third, notwithstanding the above, there remains a small, select group of patients who demonstrate periosteal and/or bony involvement and are candidates for marginal mandibulectomy. Multiple authors have demonstrated excellent local control rates with mandibular preservation.7,9,14,15 However, all of these studies are retrospective, with tremendous selection bias by extremely experienced surgeons. The criteria in the literature by which these authors decide at what point they convert from a marginal to a segmental procedure in patients with histologic evidence of periosteal and/or bone involvement is not defined. A safe oncologic resection of FOM tumors with histologic bony involvement leaves a minimum 1-cm rim of soft tissue and normal bone and a 1-cm inferior rim. The ability to obtain frozen sections on bone may help define which patients are candidates for marginal mandibulectomy.
Finally, there is no foolproof method to determine mandibular invasion. The surgeon has several methods, each with its limitations, to assist in surgical decision making. In the end, it is the combination of these methods, tempered by clinical judgment and experience, that will yield the best outcome.
Accepted for publication January 17, 2002.
Corresponding author: Mark K. Wax, MD, Department of Otolaryngology–Head and Neck Surgery, Oregon Health and Science University, 3181 SW Sam Jackson Park Rd, PV-01, Portland, OR 97201 (e-mail: email@example.com).
Wax MK, Bascom DA, Myers LL. Marginal Mandibulectomy vs Segmental Mandibulectomy: Indications and Controversies. Arch Otolaryngol Head Neck Surg. 2002;128(5):600–603. doi:10.1001/archotol.128.5.600
Browse and subscribe to JAMA Network podcasts!
Create a personal account or sign in to: