Dosimetric impact of target definitions on normal structures in head and neck cancer

As treatments for locally advanced head and neck cancer (LAHNC) have become more aggressive through the use of altered fractionation radiotherapy and concomitant chemotherapy, concern has grown regarding the risk of late toxicities in normal tissue.

The use of intensity-modulated radiation therapy (IMRT) makes it possible to more effectively spare normal tissues, such as the parotid gland. Contouring and construction of targets is extremely important in planning IMRT. Currently, no consensus exists regarding the appropriate extent of margins for the clinical target volume (CTV).

In a recent paper, we showed that there was no significant difference in locoregional control between anatomic and volumetric expansions of gross tumor volumes (GTV) [1]. Additionally, smaller volumetric expansions appeared to result in similar locoregional control as larger volumetric expansions. A number of the patients reported in that study had a direct GTV to planning target volume (PTV) expansion of 4-6 mm without the use of a CTV-high dose; many of these employed the use of a CTV-intermediate dose (CTV-ID) to account for possible microscopic extension. No significant difference in locoregional control was found for PTV-high dose (PTV-HD) expansions of 4-6 mm, 10-15 mm, or >15 mm.

Theoretically, the definition of radiation targets should impact the ability of a radiation oncologist to spare normal tissues. The recently completed Radiation Therapy and Oncology Group trial 0522 (RTOG 0522) for LAHNC employed target volumes intended to provide a regimen that was based on historical knowledge and could be implemented in a large cooperative group trial. The target volumes stipulated by the RTOG 0522 protocol tend to render volumes that would be at the upper limits of the volumes that were studied in the above mentioned investigation of Caudell et al. In the current report, we compare the RTOG 0522 target definitions with smaller target volumes and their outcome on normal tissue sparing.

Previous work from our group has suggested that early assumptions that called for large volumetric expansions on the high dose GTV may not improve locoregional control of tumors. This retrospective study found that volumetric expansions of ≤ 1.5 cm resulted in similar locoregional control as greater expansions [1]. In the report, direct GTV to PTV-HD expansions of 4-6 mm, without the use of a high dose clinical target volume, resulted in similar locoregional control as 10 - 15 mm expansions or > 15 mm expansions. A recent abstract from Sultanem et al., also reports satisfactory locoregional control and patterns of failure analysis without the use of a high dose clinical target volume [3]. Different institutional IMRT protocols may have an important impact on these findings regarding the high dose GTV-PTV expansion. For instance, the use of an intermediate dose PTV is an important factor to consider.

In this study we examined the possible dosimetric benefits of smaller target volumes. Utilizing a three dose level plan resulted in a more homogenous dose to the PTV-HD and PTV-ED, evidenced by improved 93% coverage of each volume with concomitant decreases in mean target doses and target volumes receiving 110% of prescription doses. This may be due to the smaller total target volume or the effect of the PTV-ID limiting dose to the periphery of PTV-HD and central portion of PTV-ED.

PTV-ID as used in the 3Dose plans serves a more important role than just improving heterogeneity, however. As shown in Table 2 the intermediate volume was identical to PTV-HD in RTOG plans. Any subclinical disease beyond the GTV is, by definition, microscopic. By replacing a high dose CTV with an intermediate dose CTV that receives a "microscopic dose" of 63 Gy, marginal failures may be avoided while providing improved toxicity outcomes. Additionally, Kashibatla et al. suggested that the use of concurrent chemotherapy provides a biological equivalent dose equal to a 12 Gy dose escalation in 2 Gy daily fractions [4]. Therefore, the delivery of 63 Gy with concurrent chemotherapy may result in comparable tumor control probability. For example, data from EORTC 24954 in advanced larynx and hypopharynx cancers would further substantiate this dose as 60 Gy split course radiotherapy with alternating chemotherapy provided equivalent outcomes to induction chemotherapy followed by 70 Gy of conventional radiotherapy [5]. Future investigations will be required to assess clinical outcomes with these proposed volumes and doses.

Blanco et al. and Chao et al. estimated that 4 - 5% of salivary function was lost for each additional 1 Gy in mean dose to the parotid [6, 7]. Thus the finding that mean dose to the contralateral parotid was reduced 2.1 Gy in the 3Dose plans, from 28.4 Gy to 26.3 Gy (Table 5), may be a clinically significant amount. Four patients achieved a mean dose less than 26 Gy with RTOG plans, while the same 4 and one additional patient were under this threshold in the 3Dose plans. Additionally, the volume of the contralateral parotid receiving 30 Gy was reduced from 33.4% to 29.8%, though all patients met the < 50% parameter specified by RTOG 0522.

Currently, there is little data available regarding dose-volume parameters for the mandible in order to avoid osteoradionecrosis (ORN). Both Ben-David et al. and Studer et al. found very low rates of ORN (<1%) in a population treated with prophylactic dental care and IMRT [8, 9]. However, Eisbruch et al. reported a risk of 6% in early stage oropharyngeal patients treated with IMRT on RTOG 0022 [10]. In two patients for whom dosimetry was available, ORN occurred in the sites of maximal dose, which approached 70 Gy in 30 fractions. It is therefore reasonable to assume that reducing the total dose and dose per fraction to the mandible will result in reduced rates of osteoradionecrosis. Using smaller target volumes, it is possible to significantly reduce the mean mandible dose (absolute decrease of 3 Gy) as well as higher doses to the mandible (absolute V₇₀ decrease of 5%, Table 5).

Doses to both the GSL and IPC have been associated with late swallowing toxicity [2, 11–15]. Mean doses to the GSL and IPC were reduced in the 3Dose plans, although these structures were not constrained in our planning process for either the RTOG or 3Dose plans. Further reductions should theoretically be possible in appropriate patients. For example, in the 6 patients without a primary site in the larynx or hypopharynx, the mean dose to the GSL was reduced 4.5 Gy, and the mean dose to the IPC was reduced 5.3 Gy using 3Dose target definitions.

Reductions in dose to the brainstem may reduce the incidence of nausea and vomiting during treatment [16]. In the current study, the mean dose to the brainstem was significantly less, though only by 90 cGy. The maximum dose to the brainstem was also reduced approximately by a mean of 2 Gy. These reductions likely are not clinically significant, though an improvement according to the "as low as reasonably achievable" principle.

Investigations have suggested that patients may achieve equivalent locoregional control with wide target volumes as stipulated in various protocol regimens or more restrictive target volumes with additional dose levels as outlined herein. The present communication suggests that target volume reductions (with an additional intermediate dose level) may provide an enhanced therapeutic ratio as critical structures may be spared and patients may avoid debilitating toxicities. The refinement of IMRT treatments, with or without chemotherapy, for locoregionally head and neck cancer will require consideration of the ideal target volumes that allow tumor control while minimizing toxicity.