Radiation therapy has played an important role in the treatment of patients with head and neck cancer . Head and neck malignancies are often in close proximity to critical normal tissues, such as the spinal cord, brain stem, salivary glands, and optic structures. This spatial relationship makes radiation therapy for head and neck cancers a challenging task. Some efforts have been made to use three-dimensional conformal radiation therapy (3DRT)  to preserve normal organ function while giving the tumor a tumoricidal dose. This goal may be possible with the development of an advanced form of 3DRT, called intensity-modulated radiation therapy (IMRT) [3–6]. IMRT is capable of generating complex 3D dose distributions to conform closely to the target volume even in tumors with concave features. With IMRT, the beam intensity (fluence) is optimized using computer algorithms, as it is oriented around the patient. This form of computer algorithm considers not only the target and normal tissue dimensions but also user-defined constraints such as dose limits to targets and critical organs.
Although nasopharyngeal cancer (NPC) is rare in the Unites States and Western Europe, it is one of the most common head and neck cancers in Asia and Africa [7, 8]. Patients with NPC are typically treated with radiation therapy rather than surgery because of the anatomically challenging/difficult location and a demonstrated favorable response to irradiation and chemotherapy [9–11]. Because of the proximity to surrounding critical structures, NPC is an ideal disease site to evaluate implementation of IMRT. In this study we used three different treatment modalities, conventional 3DRT and two IMRT systems in 12 patients with NPC to compare the differences in tumor coverage and normal tissue sparing.
MATERIALS AND METHODS
Patient Population and Computed Tomography (CT) Simulation
Twelve patients with newly diagnosed primary NPC treated at the Radiation Oncology Center, Mallinckrodt Institute of Radiology, Washington University Medical Center, were included in this study. Patients were immobilized in the supine position using thermoplastic masks, and a 3D volumetric CT scan (PQ 2000, Marconi, Cleveland OH) was acquired using 3-mm thick contiguous slices. Contrast medium was routinely used for the CT studies. The CT images were acquired and transferred to a virtual simulation workstation (AcuSim, Marconi, Cleveland, OH) for structure delineation. The CT scans were used to outline the extent of the primary tumor and involved neck lymph nodes. The senior author (K.S.C.) contoured the target volumes and critical structures. The gross tumor volume (GTV) included gross nasopharyngeal disease (GTVNP) and involved lymph nodes of more than 1 cm diameter (GTVLN). The clinical target volume (CTV) modeled regions considered to be at risk of microscopic nodal involvement and consisted of the cervical lymphatic drainage and included lymph nodes of less than 1 cm in diameter . Critical structures, including the brain stem, spinal cord, parotid glands, pituitary gland, optic chiasm, optic nerves, eyes, and mandible, were contoured on axial CT slices throughout the volume of interest before transferring to the two different treatment planning systems. CT images and target contours of each individual patient were transferred to both 3D and IMRT treatment planning systems. This imaging transfer process eliminated contouring variations when the procedure was performed separately on the two treatment planning systems. A 3-mm margin was designated for setup variability.
Conventional beam arrangements were designed following the guidelines of Intergroup (SWOG, RTOG, ECOG) NPC Study 0099 in which the dose was prescribed to the central axis with the goal of delivering the prescribed dose ± 10% to the target volumes. Treatment plans were carried out using isocentric 6 MV photon beams. The nasopharynx and upper neck were planned using parallel-opposed lateral 6 MV isocentric photon beams to cover the GTV and CTV with a margin. The superior margin of the primary field encompassed at least 2 cm beyond what was visible on the CT scan and included the entire base of skull and the sphenoid sinus. Posteriorly, the field included at least 2 cm beyond the mastoid process, and the field was extended further posteriorly by at least a 1.5-cm margin beyond palpable nodal disease. Anteriorly, the field included the posterior third of the maxillary sinus and nasal cavity. Blocking was used to exclude as much of the retrobulbar structures as possible without compromising the margins around the tumor. Inferiorly, the portals extended to the thyroid notch. Photon beams were used until the spinal cord reached 45 Gy, when the posterior boundary was moved to avoid further irradiation of the spinal cord. Lateral electron beams were used to continue irradiation of the posterior triangle region. The energy of electron beam was selected based on the size and depth of lymph node under the constraint of spinal cord tolerance, which was set not to exceed 48 Gy.
The target volume, which was the entire tumor (GTVNP or GTVLN), prescribed to at least 90% or greater of the mid-plane central axis dose. Variation within the target volumes did not exceed ± 10% of the target dose except in the photon-electron junction regions. The total prescription dose to GTVNP and GTVLN was 70 Gy at 2 Gy/d, whereas the prescription doses to the CTV and the uninvolved cervical lymphatics were 60 and 50 Gy, respectively. A separate anterior 6 MV supraclavicular field with a spinal cord shield, matched to the bilateral opposed fields, was used for the low neck and supraclavicular fossa below the level of the thyroid notch. The dose to the supraclavicular nodes was prescribed to a depth of 3 cm.
Serial Tomotherapy IMRT
Treatment plans were computed using a commercial serial tomotherapy planning system (Corvus, NOMOS Corporation, Sewickley, PA). Each plan was designed to deliver radiation with 6-MV photons, and the beam was intensity modulated by a multivane intensity modulating collimator (MIMiC, NOMOS Corporation). The MIMiC consists of two adjacent rows of 20 leaves, each leaf projecting to a 1 × 0.84 cm2 field at isocenter. The system is designed to deliver different doses to different targets simultaneously. Lower neck nodes were treated with a separate anterior supraclavicular field and a spinal cord shield similar to that used in the conventional 3DRT plan.
A commercial IMRT treatment planning system (Corvus) was used to provide IMRT treatment plans using conventional multileaf collimators (MLCs) using a full-field modulation technique. Five fixed-gantry (0°, 60°, 120°, 240°, 300°, Varian convention) angles were designated, and the fluence was delivered using a step-and-shoot leaf sequence and 6-MV beams. The use of more than five beams did not add any benefit to either target coverage or parotid sparing. Lower neck nodes were treated with a separate anterior supraclavicular field and a spinal cord shield similar to that used in the conventional 3DRT and tomotherapy plans.
The dosimetric goal of the treatment plan for the two IMRT techniques was to achieve adequate coverage of the target volumes while limiting dose to the critical structures. The dose prescribed and delivered to GTVNP and GTVLN was 70 Gy at 2 Gy/d; 60 Gy was given to the CTV. The limiting doses for the critical structures are shown in Table 1. Cumulative dose-volume histograms (DVHs) were obtained from the three planning systems for quantitative evaluation and comparison. Because identical lower neck portals were used in these three techniques, DVHs of tumor targets and normal structures in the lower neck and supraclavicular regions were not compared.
Table 1. Dose Prescription for Critical Structures
Maximum point dose (Gy)
A statistical analysis of the DVHs of target volumes and critical organs was performed. The paired Student t-test was used to compare the difference in the average fractional volume with selected dose levels of target volumes and normal organs in 12 patients with the three evaluated techniques. The level P < 0.05 was defined as having statistical significance.
Three-dimensional RT and serial tomography IMRT achieved similar coverage of the primary nasopharyngeal tumor. The fraction volume of GTVNP receiving 70 Gy ± 10% was 98.9% and 97.8% for 3DRT and serial tomotherapy IMRT, respectively (P = 0.06). Both techniques had significantly better GTVNP coverage than with fixed-field IMRT, in which 92.1% received 70 Gy ± 10% (P < 0.014). Serial tomotherapy IMRT had the best coverage of the gross cervical lymph nodes, and both IMRT techniques were superior to 3DRT in treating GTVLN. The fraction volume of GTVLN receiving 70 Gy ± 10% was 96.1%, 87.7%, and 46.2% for serial tomotherapy IMRT, fixed-field IMRT, and 3DRT, respectively. Similarly, the two IMRT techniques did better than 3DRT in treating the CTV. The fraction volume of CTV receiving 60 Gy ± 10% was 94.8%, 96.9%, and 41.1% for serial tomotherapy IMRT, fixed-field IMRT, and 3DRT, respectively (Fig. 1).
The IMRT treatment plans offered the most dose reduction in the parotid glands. The fraction volume receiving more than 30 Gy was 48.4%, 76.6%, and 93.2% for fixed-field IMRT, serial tomography IMRT, and 3DRT, respectively (P < 0.05). Similar results were seen for the other critical structures. No significant difference was observed between the two IMRT techniques in sparing the other critical structures, except that serial tomography IMRT spared more brain stem than fixed-field IMRT (Fig. 1).
Radiation therapy has been one of the most important treatment modalities for patients with head and neck cancer. Use of combined radiation therapy and chemotherapy has significantly improved local control and disease-free survival for patients with high-staged NPC [10, 11]. Because of its success, long-term quality of life issues become important. Radiation from conventional beam arrangement impairs the function of salivary glands and frequently results in permanent xerostomia . The decreased secretion of saliva is responsible for altered taste sensation, impaired mastication and deglutition, compromised oral hygiene, and reduced quality of life [14, 15]. Parotid glands account for approximately one-half to two-thirds of the total salivary output . Some effort has been made to preserve salivary function after radiation therapy for patients with head and neck malignancies, including the use of pilocarpine , surgical transfer of the submandibular gland , customization of beam arrangements using 3D conformal therapy , and the implementation of IMRT [19–21]. In this study we further demonstrated that better coverage of nasopharyngeal tumor/cervical nodes and sparing of critical structures can be achieved with serial tomotherapy or fixed-field IMRT techniques.
IMRT offers increased target dose conformality while reducing dose to sensitive normal structures. The feasibility of IMRT has been tested and confirmed in the treatment of prostate cancer , head and neck malignancies , and recurrent tumors after radiation therapy . Experiences from Mallinckrodt Institute of Radiology have shown that tumor control is superior to the conventional beam arrangement and salivary glands were spared with subsequent improvement of quality of life . NPC is an ideal disease to be treated with IMRT, due to the centralized location of the primary tumor with its close relationship to the surrounding critical structures.
The two IMRT techniques—serial tomography IMRT and fixed-field IMRT—had better sparing of normal organs than conventional 3DRT. Significant dose reduction to the parotid glands was obtained with either IMRT technique in comparison with 3DRT. Doses in excess of 30 Gy have been reported to cause permanent decrease in saliva output in patients with head and neck cancer . With IMRT, the dose to half of the parotid volume was less than 30 Gy with the fixed-field IMRT plans. In these plans, the parotid gland dose was lower with fixed-beam IMRT than with tomotherapy. Verhey  reported the best sparing of brain stem, cord, and parotids by using fixed-field static IMRT with five hand-selected field directions.
In the present study, conventional 3DRT and serial tomography IMRT both achieved satisfactory coverage of nasopharyngeal tumor, while fixed-field IMRT did a little worse. This result was partly due to the limited number of treatment fields used in fixed-field IMRT. Homogeneous dose coverage of target volume has been emphasized in biological models to obtain the adequate tumor control probability. The impact of underdosage within the tumor should be dealt with great caution . In the head and neck, as the number of fields increases with fixed-field IMRT, the target dose will increase but the dose tends to increase to neighboring critical structures . In this study, both IMRT techniques were superior to 3DRT in covering the cervical lymph nodes due to the limited distal coverage possible with posterior triangle electron beams.
In summary, these encouraging dosimetric results substantiate the theoretical advantage of inverse-planning IMRT in the management of NPC. We showed that target coverage of the primary tumor was adequately maintained and nodal coverage was improved, as compared with conventional beam arrangements. The ability of IMRT to spare the parotid glands is exciting, and may translate to improve quality of life in patients with NPC. Prospective clinical studies are underway at our institution in an effort to accurately define the relation between clinical xerostomia and the extent of parotid dose-volume sparing.