Risk profile for osteoradionecrosis of the mandible in the IMRT era

Background The risk for osteoradionecrosis (ORN) of the mandible is positively related to bone volume exposed to > ~ 60 Gy. We hypothesized that in combined treatment, surgery may also be a risk factor. Patients and methods The impact of mandibular surgery on ORN in locally disease-free IMRT cohorts was retrospectively analyzed. Results Between October 2002 and October 2013, 531 of 715 patients with oral cavity cancer (OCC), mesopharyngeal cancer (MC), or salivary gland tumor were treated with the mandible bone exposed to ~ > 60 Gy (mean follow-up, 38 months; 7–143 months). Of the 531 patients, 36 developed ORN (7 %; 1.5 % with grade 3–4). The ORN rate in definitive IMRT MC (16/227) and in postoperative IMRT OCC patients with no mandibular surgery (3/46) was 7 % each; in OCC patients with mandibular surgery the rate was 29 % (15/60, p = 0.002). Marginal or periosteal bone resection was found to be a high risk factor (39 %, vs. 7 % followed by segmental or no resection, p < 0.0001). Conclusion Marginal or periosteal bone resection of the mandible was identified as the highest ORN risk factor in our IMRT cohort.


Introduction
The risk for osteoradionecrosis ORN) of the mandible is known to be positively related to radiation doses of > 60 Gy to the bone [1,2]. Patients at the highest risk for ORN are expectedly those who undergo definitive radiation therapy with doses of ~ 70-72 Gy for oral cavity cancer (OCC) or large central or lateralized mesopharyngeal cancer (MC) close to the mandible.
In our OCC cohort, a higher ORN rate was observed in the postoperative subgroup (60-66 Gy in 2.0 Gy/fraction) than in the definitive IMRT subgroups as well as in the postoperative IMRT MC cohort. This observation motivated us to analyze the impact of previous mandibular surgery on the risk for ORN.
Our hypothesis was that pre-IMRT surgery of the mandible may also be a risk factor for ORN.

Patients and methods
We retrospectively analyzed a single-center OCC/MC/ salivary gland tumor IMRT cohort with respect to ORN. Data were prospectively acquired in a StatView ® database. Patients with ≥ 60 Gy to the mandible were included. The following potential risk cofactors were analyzed: surgical technique (no mandibular surgery vs. periosteal resection vs. segmental resection vs. marginal (horizontal or sagittal) resection), nicotine and alcohol abuse, carotid artery calcification depicted in the planning computed tomography, comorbidities [diabetes, chronic obstructive pulmonary disease (COPD), peripheral arterial occlusive disease, coronary heart disease, immunosuppression], age, gender, and T stage.
Approval from the local ethics committee was obtained for data evaluation of our IMRT cohort. The analysis start point was defined as the start of radiation.

Radiation dose
Postoperative IMRT prescription doses ranged between 60 and 66 Gy in 2.0 Gy/fraction, definitive IMRT prescription doses ranged between 68 and 72 Gy, mostly 70 Gy or 69.6 Gy in 2.0 Gy or 2.11 Gy/fraction, five fractions per week. Attention was paid to avoid/minimize dose hot spots to the mandible. Early analyses of mandible dose-volumes of our first 73 IMRT patients at risk (included in the present analysis) were previously reported [3]. Dose-volume histograms (DVHs) of the mandibles exposed to > 60.0 Gy were calculated. The maximum dose (Dmax) is defined as the highest dose in 1 %.
Details of the IMRT schedules used have been provided in previous reports [10,11].

Systemic concomitant therapy
If indicated, cisplatin was concomitantly given (40 mg/ m 2 /week). Since April 2006, cetuximab has been used for patients with contraindications for cisplatin chemotherapy (400 mg/m 2 loading dose, followed by 250 mg/m 2 1 day/ week). Age and/or comorbidity or early-stage disease were reasons not to add systemic therapy (15 %).

Follow-up
All patients were regularly seen in our joint clinics 3-6 weeks after completion of IMRT, at the Departments of Otorhinolaryngology and Head and Neck Surgery or the Department of Craniomaxillofacial and Oral Surgery. Our Head Neck Cancer Center standards for patient assessment include physical examination with additional flexible fiberoptic endoscopy approximately every 2-3 months (MC) in the first follow-up (FU) year, every 3 months in the second to third year, and every 6 months in the fourth to fifth year.

ORN grading
• Out of several systems grading ORN, we used the system proposed by Glanzmann and Graetz [1]:Grade 1: Exposed bone without signs of infection for at least 3 months • Grade 2: Exposed bone with signs of infection or sequester, but not grades 3-5 • Grade 3: ORN treated with mandibular resection, with satisfactory result G. Studer et al.

DVHs of the mandible
DVHs of the 531 patients at risk were calculated for all subgroups ( Fig. 1 ORN of the mandible that occurred in relation to invasive interventions (dental implants, tooth extractions, reconstructive/rehabilitative osseous or soft tissue surgery) on gingiva and/or mandible regions previously irradiated with doses > 50 Gy was termed "postintervention ORN." Exposed hardware after surgery was not considered as ORN, except for cases with secondary exposure after previous coverage.
Pre-IMRT surgical interventions on the mandible that were regarded as inclusion criteria for the term "mandibular surgery" (MS) were: The time delay between surgery and radiation ranged between 4 weeks to 7.5 weeks (mean 5.5).
Dental care before/during/post radiation therapy All patients underwent standardized dental care according to our in-house dental care protocol, which is characterized by its individual risk-adjusted approach [7].

Statistical analysis
Statistical calculations were carried out using the statistics program implemented in StatView ® (version 4.5; SAS Institute, Cary, NC). Univariate analyses were performed with a Cox proportional hazards regression model in StatView ® . Actuarial survival data were calculated using Kaplan-Meier curves and log-rank tests implemented in StatView ® ; p values of < 0.05 were considered statistically significant.
Multivariate analysis was performed using the StatView ® calculation program (Mantel-Cox log-rank test). DVHs curves were calculated using the Excel ® (Office 2013) program.

Cohort
We assessed 715 locally disease-free OCC/OC/SGT IMRT patients who were treated at our department between October 2002 and October 2013. The mean FU was 38 months (7-143; Table 1).

3
Risk profile for osteoradionecrosis of the mandible in the IMRT era operative IMRT subgroups with lower IMRT doses ( Fig. 1, dotted lines), with 17 ORN events in 282 patients (6 %) vs. 19 of 249 operated patients ( 7.6%, ns; Table 2), suggesting an additional risk factor impacting the risk for ORN other than radiation dose only.
Of 36 ORN events, eight (22 %) developed as a consequence of invasive manipulation on previously with > 50 Gy irradiated bone areas (postintervention ORN): Four of eight events translated into complicated grade 4 ORN (pathological fracture, osteocutaneous fistula, osteomyelitis, persisting pain) following segmental resection or limited surgery for ORN. Two patients were treated with segmental resection (grade 3) with uncomplicated FU. Four of the eight postintervention ORN patients underwent postoperative IMRT (marginal resection in two patients, segmental resection in two patients).
All grade 3 ORN occurred in the definitive IMRT MC cohort: in two of four patients after 70 Gy in 2.0 Gy/fraction (D max point doses 71.3 and 71.6 Gy, respectively), and in two of four following post-IMRT interventions.
The mean time from IMRT completion to ORN was 20 months (range, 1-104); the corresponding time for postintervention ORN was also 20 months (2-50).

Osteoradionecrosis
The definitive IMRT subgroups (characterized by higher IMRT doses to the mandible, Fig. 1, continuous lines) showed a tendency to have lower ORN rates than the post-   mobile tongue and severe pre-existing comorbidity (nicotine ± alcohol abuse, COPD, polyneuropathy, postoperative wound healing problems, low body mass index, etc.). The D max to the mandible was 66.7 Gy, 64.7 Gy, and 69.1 Gy, respectively.
Surgical reconstruction in 65 patients at risk who underwent mandibular surgery ( Table 3)

ORN in MC patients
The rate of grade 1-3 ORN in MC patients was 4 % (16/441): 5 % (16/344) in the entire definitive IMRT MC cohort, 7 % (16/227) in patients with some aspect of the bone exposed to > 60 Gy (see Table 2). D max point doses to the mandible of the 16 ORN-affected patients were a mean 73.6 Gy (range, 69.6-76.2 Gy). No ORN developed in 97 postoperative IMRT MC patients.

ORN in SGT patients
Of 69 postoperative patients treated with > 60 Gy to the ascending ramus or angle of the mandible, one patient was diagnosed with grade 2 ORN: In this patient IMRT following periosteal resection and rarefication of the covering soft tissues resulted in a symptomatic necrosis of the condylar head (Table 2).

Other potential risk cofactors
Age and gender were equally distributed in the ORN and non-ORN patients. T stage in ORN patients was distributed

ORN in definitive IMRT OCC patients
In this subgroup ( Table 2) at highest risk (highest dose to largest mandible volume, Fig. 1), no ORN was diagnosed.

3
Risk profile for osteoradionecrosis of the mandible in the IMRT era resection may take place with or without bone removal. This will lead to a denuded mandibular cortical plate with compromised healing capacity. In segmental resection cases the surgery is more extensive and mostly requires a composite free vascularized graft with a well-nourished bone flap with an unharmed periosteum and adequate soft tissue coverage. Table 4  The observation of no ORN event (but one postintervention ORN) in the cohort at highest risk, i.e., definitively irradiated OCC, seems to be sufficiently explained by the fact of the survival time being too short to experience ORN (only ~ 35 % surviving 20 months, Fig. 2). Poor outcome following definitive irradiation of OCC also in the IMRT era was previously reported; however, data on this topic are scant [13], which may partly be explained by a negative selection of OCC patients referred for definitive radiation (elderly, comorbid patients with large/inoperable tumors). For the small subgroup (n = 14) who survived 2 years, we cannot exclude an ORN rate of approximately 22 % (according to the statistical "rule of three" to estimate the probability of adverse events in small sample sizes with few events, giving the upper limit of the 95 % confidence interval of the probability: 3/14 = 22 %).

Discussion
Previous surgery to the mandible, and mainly the type of surgery, was found to be the most important ORN risk parameter following IMRT.
In addition, this analysis confirmed that doses of < ~ 60-65 Gy to the mandible represent the lower threshold for any risk of ORN. An ORN rate (grade 2-3) of 39 % was found for postoperative IMRT patients at risk with floor of the mouth carcinoma or mandible infiltrating carcinoma who underwent previous periosteal or marginal mandibular resection (Table 3). This was statistically significantly different compared with postoperative IMRT OCC patients with no or segmental resection (7 %, each). A plausible explanation for this finding is that the mandible is more dependent on the periosteum for its blood supply than on the inferior alveolar neurovascular bundle, especially in older individuals [12]. During surgery, significant periosteal stripping or Table 4 ORN rates related to mandibular surgery (MS) and DVH in OCC/MC patients at risk (> 60 Gy to the bone, n = 459; see also corresponding DVHs, Fig. 1

Conclusion
Periosteal or marginal mandibular resection was the statistically significantly highest ORN risk factor, translating into a 39 % ORN rate grade of 2-3, vs. 7 % following definitive IMRT or postoperative IMRT with no or segmental resection. Consequently, the radiation dose to the mandible should be minimized in patients at high risk for ORN.
Open Access This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.
The positive relationship between radiation dose-volumes and the risk for ORN is known, and the reduced ORN risk by using mandible-sparing IMRT techniques was confirmed in several reports [3,[5][6][7][8]. Only scant information is available on the ORN risk in the postoperative setting, which is characterized by lower radiation doses to the mandible than in the definitive radiation setting. Table 5 gives an overview of the ORN events in our patients related to the treatment sequence.
Korean researchers reported on OCC/MC patients irradiated postoperatively with conventional three-dimensional radiation techniques. The authors found a significantly higher ORN rate in 59 patients with previous mandibular surgery than in 139 patients with no mandibular surgery (13.6 vs. 3.6 %, p = 0.01), with a latency from radiation to onset of ORN of a mean 22 months-comparable to our own results-and an ORN grade 3 rate of 38 % (5/13) [14].
Comorbidity was not found to be statistically significantly influencing the ORN risk in our cohort, which may be due to sample size imbalance. Gevorgyan et al. found no association between the severity of 14 ORN cases (grade 1-3, similar to our classification, 8/14 following definitive radiation) and gender, age smoking, alcohol abuse, TN stage, RT technique, or performance status [8]; they reported a significantly lower ORN incidence following IMRT compared with conventional techniques (p < 0.015).
With respect to ORN events in postoperatively irradiated parotid gland tumor patients, temporal bone ORN after surgery was reported by Leonetti et al. [15], who assessed 221 patients divided into the following groups: (1) parotidectomy only; (2) parotidectomy with mastoidectomy; and (3) parotidectomy with subtotal petrosectomy. The overall incidence of temporal bone ORN in group 1 was 2/106 (2 %); in group 2, 8/64 (13 %); and in group 3, 0/51 (0 %; radiation techniques not indicated). Because the horizontal and angular aspect of the mandible bone is at highest risk, ORN is a rare event following postoperative parotid gland tumor irradiation. This low susceptibility to ORN may be partially explained by the fewer surgical interventions performed in this area, the absence of teeth, and adequate soft tissue coverage by the masticatory muscles.
In summary, the following conclusions can be drawn from the present results: