The aim of the present study was to investigate the potential clinical role of intensity-modulated arc therapy (IMAT) for patients with nasopharyngeal cancer (NPC). In contrast to the seven fixed gantry beams used in intensity-modulation, IMAT is based on volumetric intensity-modulated arc delivery. The IMAT technique is based on an investigation by K. Otto [1]. It aims to improve sparing of organs at risk (OARs) and healthy tissue as compared to other solutions; to maintain or improve the degree of target coverage and to reduce beam-on time per fraction. Intensity-modulated radiation therapy (IMRT) enables delivery of highly conformal dose distributions. It also increases the therapeutic ratio, as target volumes are often large, concave, and in the proximity of critical normal tissues [2, 3, 4, 5, 6, 7, 8]. Clinical target volumes (CTVs) for cancer of the nasopharynx lie in close proximity to the optic nerves, optic chiasm, orbits, brainstem, spinal cord, cochlea and parotid glands.

Xerostomia often occurs as a side effect of high-dose head and neck irradiation. Xerostomia encompasses a wide range of symptoms that vary from inconvenience experienced during eating and speaking, to a debilitating condition. Reduction in salivary flow per se is not life threatening, but alterations in eating function and deterioration of dental and oral health have a significant impact on quality of life [4, 5]. Saliva also possesses antibacterial functions and a reduction in saliva flow may lead to frequent opportunistic infections in the oral cavity [9, 10].

IMAT is a novel radiation treatment technique based on volumetric modulated rotational delivery, as opposed to the fixed gantry beams used in classical IMRT [11, 12]. By varying the gantry rotation speed and multileaf collimator (MLC) shape, as well as continually changing the fluence (dose rate), IMAT delivers highly conformal IMRT plans in a short time [4, 5, 6, 7, 8, 11, 12, 13].

Patients and methods

Between January and June 2010, computed tomography (CT) data of 10 patients with NPC were enrolled in a prospective study comparing two IMRT techniques at our institute. These patients had been irradiated with a seven-field fixed IMRT technique. Their IMAT plans were only generated for the purpose of comparing two radiotherapy plans for the same tumor. Only five OARs (optic nerve, optic chiasma, brainstem, spinal cord and parotid glands) were taken into account during plan optimization.

Patient immobilization and imaging acquisition

A noninvasive immobilization method was used. The patient was placed in the supine position on a customized head and neck support with a reinforced thermoplastic nine-point fixation immobilization mask around the head. A set of volumetric CT images was acquired from our dedicated CT simulator with the patient immobilized in the treatment position. The data was then transferred to the inverse- or the forward treatment planning system (TPS). The scan slice thickness was 2.5 mm throughout the tumor target region and other areas.

Target definition

The definition of target volume follows the description in the International Commission on Radiation Units (ICRU) Reports 50, 62 and 83. [14, 15, 16]. The gross tumor volume (GTV70) is defined as all known gross disease determined from CT, magnetic resonance imaging (MRI), positron-emission tomography (PET)-CT, physical examination and endoscopic findings. Three different CTVs are described: CTV70: GTV + 5 mm margin; CTV60: entire uninvolved nasopharyngeal area and CTV54: areas of possible spread of the tumor, i.e. the posterior third of the nasal cavity and maxillary sinuses, sphenoid sinus, clivus, posterior ethmoid and cavernous sinuses and elective nodal areas [17, 18, 19]. A 3 mm margin was added to the CTVs to create the planning target volumes (PTVs). The lymph node groups or neck levels at risk were determined according to image-based nodal classification [10, 11]. Levels II-IV were included in CTV54 in all cases; levels Ib and/or V were also included in CTV54 where a neighboring level was involved [17, 18, 19]. OARs for radiation complications were delineated on each axial CT slice and included both parotid and submandibular glands, the spinal cord, brainstem, optic nerves and optic chiasma. All of the organs in the head and neck were delineated. Parotid glands were delineated separately as the entire gland and the parotid gland outside of the PTV (parotid-PTV). Dose constraints were applied to parotid-PTV, rather than to the entire parotid gland.

Planning objectives and techniques

Dose prescriptions were set to 54 Gy with 1.63 Gy/fraction to PTV54; at 60 Gy with 1.81 Gy/fraction to PTV60 and at 70 Gy with 2.12 Gy/fraction to PTV70. All volumes were to be simultaneously treated according to the simultaneous integrated boost (SIB) approach. All plans were normalized to the mean dose to PTV70 resulting in at least 95 % of the PTV being covered with the prescribed dose.

For all PTVs, the plans aimed to achieve a minimum lowest dose greater than 95 % of the prescribed dose, and a maximum dose of less than 107 % of the prescribed dose. The plan parameters aimed to deliver 98 % of the PTV not less than 95 % of the prescribed dose (D98 % near min) and no more than 2 % of the PTV more than 107 % of the prescribed dose (D2 % near max) [9]. As explained above, we described parotid glands as the entire parotid gland and the proportion of it excluded from the PTV (parotid-PTV). The plans aimed at keeping the mean parotid gland dose below 26 Gy and the dose to 50, 33 and 66% of the parotid gland volume (D50 %, D33 %, D66 %) under 30, 50 and 15 Gy, respectively [16, 17, 18].

The two sets of plans compared in this study were all designed on the Varian Eclipse TPS (Varian Medical Systems Inc., Palo Alto, CA, USA). For all plans, the isocenter of the beams was set at the center of the PTV54. The protocols employed 6 MV photon beams from a Varian Clinac equipped with a Millennium MLC with 120 leaves (spatial resolution of 5 mm at the isocenter for the central 20 cm, and 10 mm in the outer 2 × 10 cm; maximum leaf speed 2.5 cm/s and leaf transmission of 1.6 %; maximum gantry speed of 5.54°/s). Plans for IMAT were optimized by selecting a variable dose rate of 0–600 MU/min; a fixed dose rate of 300 MU/min was selected for IMRT plans. The anisotropic analytical algorithm (AAA, version 8.6.15) photon dose calculation algorithm was used in all cases. The dose calculation grid was set to 2.5 mm.

Details of the IMAT plan delivery process were as described previously [2, 3]. In brief, two complementary 358-degree coplanar arcs were used: from 181 to 179° for the clockwise rotation (CW); 179 to 181° for the counter clockwise rotation (CCW). A sequential approach was applied, in which the first arc plan was used as a base dose plan for the second arc plan. This compensated for possible under- or over dosage in the first arc plan, leading to a homogeneous dose within the PTV. The application of two coplanar arcs aims to increase the modulation factor during optimization. Geometrical optimization was used to optimize field size. The collimator angle was set to 45° for all arc plans. The normal tissue objective was used to protect healthy tissue. The same optimization constraints as used in IMRT plans were applied initially and then adjusted in real-time during the optimization process.

Details of reference of IMRT plans were computed by selecting the conventional intensity-modulation approach as a benchmark, with fixed beam gantry and intensity-modulated beams delivering the dose by means of a sliding window approach [20, 21, 22, 23, 24]. Plans were optimized individually using seven fixed beam angles (0, 51, 102, 153, 207, 258 and 309°) of 6 MV fields. Dose calculation was performed using the Helios/Eclipse version 8.6.15 (Varian) with the AAA algorithm.

Quantitative evaluation of plans was performed by means of dose–volume histograms. For the optimization, PTVs were reduced to 3 mm under the surface of the skin to prevent optimization problems in the buildup region.

Dose constrains for OARs were described according to the recommendations in Emami and Quantec tables [25, 26]. Conformity- (CI) and homogeneity index (HI) values were calculated in all cases and used to compare the quality of the two plans [27, 28]. A t-test was used for statistical evaluation.

Results

Dose distribution and dose–volume histograms of two plans for the same patient are shown in Fig. 1 (axial, coronal and sagital views).

Fig. 1
figure 1

Dark blue PTV70, Light blue PTV54, pink left parotid, green right parotid, lines linked by triangles IMRT, lines linked by squares IMAT, dark blue line PTV70, light blue line PTV54, pink line left parotid gland, green line right parotid gland

Full size image

Target coverage

Target coverage and dose homogeneity were identical for both planning techniques. Mean PTV70 dose was 72.7 Gy for arc plans and 73.01 Gy for IMRT plans. The mean, near max and near min doses to PTV70, PTV60 and PTV54 for arc and IMRT plans are summarized in Tab. 1. IMAT and IMRT plans were shown to be equivalent in terms of CI and HI: mean CI of the 10 patients was 1.05 ± 0.08 for both arc and IMRT plans. The mean HIs of the 10 patients were 1.08 ± 0.02 and 1.07 ± 0.01 for arc and IMRT plans, respectively (Tab. 2).

Tab. 1 Comparison of mean, D98 % (near min) and D2 % (near max) doses to PTVs
Full size table
Tab. 2 Comparison of CI, HI and MU
Full size table

Parotid glands

For both sides, parotid glands were separately delineated as the entire organ and the parotid gland outside of the PTV (parotid-PTV). The analysis was carried out for ipsi- and contralateral parotid glands separately. Contralateral parotid gland sparing was significantly better in arc plans. The mean contralateral parotid doses were 25.7 and 27.7 Gy for arc and IMRT plans, respectively. Mean contralateral parotid doses for the 10 patients are shown in Fig. 2 a, b, c, d The mean ipsilateral parotid doses were 30.6 and 32.5 Gy for arc and IMRT plans, respectively, although this difference was not statistically significant (Tab. 3).

Fig. 2
figure 2

a Contralateral parotid mean doses. Doses to 50, 33 and 66% of the contralateral parotid gland volume (D50 %, D33 %, D66 %): b Contralateral parotid D50 %. c Contralateral parotid D33 %. d Contralateral parotid D66 %

Full size image
Tab. 3 Comparison of contralateral and ipsilateral parotid glands and parotid gland outside the PTV54 (parotid-PTV54) doses
Full size table

Spinal cord

All plans respected the planning objective of 46 Gy as the maximum dose to the spinal cord; the IMAT plan allowed for better sparing of the spinal cord in terms of D2 % (Tab. 4).

Tab. 4 Comparison of doses to other critical structures
Full size table

Brainstem

As for the spinal cord, IMRT and IMAT plans successfully achieved the D2 % planning objective of 54 Gy. A slightly better dose distribution was observed for arc plans, although this was not statistically significant (Tab. 4). Dose comparisons for other OARs are summarized in Tab. 4.

Healthy tissue

Healthy tissue was described as the body outside of PTV54, where the aim was to keep doses as low as possible. Mean healthy tissue doses were 9.4 and 10.1 Gy for IMAT and IMRT plans, respectively. The volume of healthy tissue receiving 10 Gy (V10 Gy) was also examined: V10 Gy values of 26.9 and 29.5 % were found for IMAT and IMRT plans, respectively. Although the difference was not statistically significant, more healthy tissue sparing was achieved with arc plans than with IMRT plans (data not shown).

Monitor units

MU values were significantly better in arc plans: mean MUs were 540 ± 130 and 1288 ± 197 for IMAT and IMRT, respectively (p < 0.001). IMRT plans had MU values that were roughly double those of arc plans. The multiple field arrangement and spilt fields used in IMRT mean that the delivery times for these plans are significantly longer than for IMAT, since they include“dead times” such as the time needed to reposition the gantry and reprogram the linear accelerator (LINAC) at every field. However, plan optimization times were longer IMAT than for IMRT plans (3 h verses 1 h).

Discussion

NPC is a unique type of tumor for which radiotherapy is the only curative treatment option. For many years, NPC has been successfully treated with radiotherapy alone for early stages and in combination with chemotherapy in locally advanced disease. Radiotherapy planning for NPC has evolved during the years. Until the introduction of three-dimensional radiotherapy planning, two laterally opposed fields were used to cover the primary site and neck nodes. During the two-dimensional planning era it was impossible to spare the parotid glands. This meant that all patients suffered varying degrees of xerostomia, which resulted serious dental problems and difficulty in swallowing [1, 2, 3, 4, 5]. The complex anatomy of this region and various routes of NPC spread render radiotherapy planning extremely difficult. This anatomical area contains critical structures such as the spinal cord, brain stem, optic- and cranial nerves, temporal lobe of the brain, eyes, salivary glands (including parotid, sub-mandibular and sub-lingual glands, as well as the minor salivary glands), temporomandibular joint, mandible, major vessels, swallowing muscles, cochlea, etc. Each structure has its own radiosensitivity and corresponding dose constraint [8, 9, 10]. The proximity of these structures to the target volume is dependent upon tumor spread and thus differs patient to patient. IMRT planning allows us to spare critical structures while simultaneously achieving good target coverage [29, 30, 31, 32, 33].

In this feasibility study, we compared a seven-field sliding window IMRT technique and a double arc dynamic IMRT technique in terms of parotid gland sparing and target coverage. OARs were also compared, as were total MUs delivered and treatment times in 10 patients with NPC. Patients were not treated with the arc technique. Similar investigations for various clinical settings have been published in recent years. Palma et al. [32] investigated an IMAT progenitor in prostate cancer and showed that variable dose-rate volumetric arc modulation is beneficial compared to IMRT or a constant dose-rate. Cozzi et al. [33] and Fogliata et al. [34] appraised the behavior of IMAT on cervix uteri cancer and small benign brain tumors. Aydogan et al. [35] showed much better organ sparing with IMAT in total marrow irradiation. In these studies, IMAT proved to be at least as good as IMRT in terms of target coverage while simultaneously demonstrating improved OAR sparing. We specifically selected NPC as the primary site–rather than head and neck cancer in general—because its unique location and patterns of spread result in a close proximity between high dose areas and OARs that need to be spared. This makes radiotherapy planning particularly challenging.

Target coverage was identical for both IMAT and IMRT techniques. Mean PTV70 doses were 72.7 and 73.7 Gy; PTV70 D98 % doses were 69.2 and 69.7 Gy and PTV70 D2 % were 74.9 and 74.7 Gy for arc and IMRT plans, respectively. In accordance with the results of other studies, we demonstrated identical CI and HI values for both plans and also found target coverage to be ideal in both cases. Vanetti et al. [12] compared IMRT with single arc and double arc treatments and showed a slight improvement in target dose homogeneity and coverage using the double arc method.

Although mean dose to the ipsilateral parotid gland was 2 Gy lower in the arc plan, this difference did not reach statistical significance. Most of the time, the majority of the ipsilateral parotid volume is inside the PTV and adequate sparing is thus impossible. Mean contralateral parotid gland doses were significantly lower with double arc plans (25.7 Gy versus 27.7 Gy, p = 0.008). The 50, 33 and 66 % contralateral parotid gland doses were also lower in arc plans, and here the difference did reach statistical significance (Tab. 3). Vanetti et al. [4] showed a reduced dose to the parotid glands on both sides with double arc plans.

While maintaining similar OAR sparing, Verbakel et al. [11] achieved the best PTV dose homogeneity using a double arc plan as compared to single arc and seven-field sliding window IMRT plans. In the multi-institutional comparative planning study by Holt et al. [36], better sparing for almost all OARs was shown with VMAT when compared to step-and-shoot IMRT and the average effective delivery time was also shorter with VMAT. Similarly, Wiehle et al. [37] compared step-and-shoot IMRT and VMAT in 15 patients with head and neck cancer and achieved higher doses for the PTV and lower doses for other structures using VMAT. Pasler et al. [38] compared IMRT and VMAT plans at two different gantry speeds. Their results were in favor of VMAT with a slower gantry speed (120 deg/s). The similarity between the results of our study and those of other institutions not only eliminates a potential bias due to different optimization strategies, but also confirms that VMAT can achieve similar results with different industrial products.

Parotid function starts to diminish at a mean organ dose of 15–20 Gy and considerable salivary dysfunction occurs at a mean dose of 20–40 Gy. At 40 Gy, moderate to severe xerostomia is inevitable. The generally accepted threshold mean dose for the parotid gland is 26 Gy. Eisbruch et al. suggested that 66 % of the parotid gland should receive 15 Gy or less [18]. In our study, parotid gland D66 % was 14.87 Gy for arc- and 17.44 Gy for IMRT plans. Obviously we were able to meet dose volume criteria for ipsilateral parotid gland with the arc plan. Although it may be arguable whether or not a 2 Gy reduction in parotid gland mean dose is of clinical benefit, keeping parotid gland doses as low as possible is the key to avoiding long term xerostomia.

Both IMRT and IMAT plans resulted in doses below dose constraints for the spinal cord, brainstem, optic chiasma and optic nerves. Although statistically insignificant, lower doses were achieved with IMAT as compared to IMRT plans for these OARs. The most significant improvement with arc planning was the MU per fraction. Arc planning yielded a mean MU value of 540, compared to 1288 in IMRT plans. American Association of Physicists in Medicine (AAPM) task group test studies confirmed reduced MU using VMAT as mentioned in the study by Sathiyen et al [39]. In addition, Szu Huai Lu et al compared step-and-shoot IMRT, VMAT and helical tomotherapy plans. These authors showed a VMAT mean delivery time of 5.7 min, compared to 9.5 min for helical tomotherapy and 9.2 min for step-and shoot-IMRT [40]. Lower MUs and shorter treatment times are consistently reported with VMAT [11, 36, 37, 38, 39, 40].

Conclusion

IMAT planning with two arcs achieves a slight reduction of dose to the parotid glands and other OARs while maintaining good target coverage as compared to static seven-field dynamic IMRT. Moreover, total MU per fraction and therefore beam-on times are dramatically reduced in the arc technique.