Clinical Oncology xxx (2011) 1e8. Contents lists available at SciVerse ScienceDirect. Clinical Oncology

Transcription

1 Clinical Oncology xxx (2011) 1e8 Contents lists available at SciVerse ScienceDirect Clinical Oncology journal homepage: Original Article Adaptive Radiotherapy Using Helical Tomotherapy for Head and Neck Cancer in Definitive and Postoperative Settings: Initial Results L. Capelle *, M. Mackenzie y, C. Field y, M. Parliament *, S. Ghosh z, R. Scrimger * * Division of Radiation Oncology, Cross Cancer Institute, Edmonton, Canada y Division of Medical Physics, Cross Cancer Institute, Edmonton, Canada z Division of Experimental Oncology, Cross Cancer Institute, Edmonton, Canada Received 12 April 2011; received in revised form 7 September 2011; accepted 9 November 2011 Abstract Aims: To assess whether routine mid-treatment replanning in head and neck squamous cell carcinoma patients results in meaningful improvements in target or normal tissue dosimetry and to assess which patients derive the greatest benefit. Materials and methods: Twenty patients treated with either postoperative chemoradiotherapy or definitive chemoradiotherapy with primary or nodal disease 3 cm in size were included in this prospective pilot study. Seven patients received adjuvant chemoradiotherapy and 13 received definitive chemoradiotherapy. Patients were planned and treated on a helical tomotherapy system. All patients had a second computed tomography scan after 15 fractions and a new plan based on this was initiated from fraction 20. Results: Relative volume changes between computed tomography scans were: GTV 29%; CTV60 (adjuvant patients) 4%; parotid volume 17.5%; median reduction in neck separation 6e7 mm; weight loss 3%. For the group overall and for the definitively treated patient cohort, respectively, adapted plans resulted in reductions in PTV66 D 1 (0.3 Gy, P ¼ 0.01 and 0.5 Gy, P ¼ 0.01); PTV54 D 1 (0.6 Gy, P < and 0.9 Gy, P ¼ ); spinal cord maximum (0.5 Gy, P ¼ and 0.6 Gy, P ¼ 0.04) and volume of skin receiving 50 Gy (16 cm 2, P ¼ 0.01 and 19 cm 2, P ¼ 0.001). Definitively treated patients also had a reduction in mean parotid dose (0.6 Gy, P ¼ 0.046) and volume of normal tissue receiving 50 Gy (67 cm 3, P ¼ 0.02). Patients with nasopharyngeal carcinoma received the greatest benefits with treatment adaptation with reduction in spinal cord maximum 1.2 Gy, mean parotid dose 1.2 Gy and parotid V %. There was no significant benefit for adjuvant patients. Other factors associated with greater benefits were greater weight loss and greater reduction in neck separation and higher T stage. Conclusions: There is minimal benefit to routine adaptive replanning in unselected patients, and no benefit in adjuvantly treated patients. Patients with nasopharyngeal carcinoma or with greater weight loss or reduction in neck separation did have clinically significant benefits. These patients should be targeted for adaptive strategies. Ó 2011 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved. Key words: Adaptive radiotherapy; head and neck cancer; radiotherapy planning; tomotherapy; toxicity Introduction Radiotherapy is an integral part of treatment for head and neck squamous cell carcinoma (HNSCC), both in the definitive and adjuvant settings. It is, however, associated with significant morbidity due to effects on normal tissues. Use of intensity-modulated radiotherapy, with its ability to deliver dose in a highly conformal way, allows greater normal tissue sparing than traditional three-dimensional conformal radiotherapy and improves quality of life in Author for correspondence: R. Scrimger, Department of Radiation Oncology, Cross Cancer Institute, University Ave, Edmonton, AB T6G 1Z2, Canada. Tel: þ ; Fax: þ address: rufus.scrimger@albertahealthservices.ca (R. Scrimger). HNSCC patients [1]. With this highly conformal treatment there is the potential for changes in dosimetry to occur due to changes in patient anatomy during the course of therapy. It is currently not routine practice to account for these changes during treatment. Changes in patient anatomy may occur either from a reduction in tumour bulk, weight loss or a reduction in postoperative oedema, and may vary between patients. Adaptive radiotherapy, in this context, refers to reimaging the patient during their radiotherapy course and generating a new radiotherapy plan based on this new imaging, which takes into account changes in anatomy. Decisions on whether to modify (adapt) a treatment plan during the treatment course are generally made empirically based on weight loss, change in neck separation or poor immobilisation shell fit. To /$36.00 Ó 2011 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved. doi: /j.clon

2 2 L. Capelle et al. / Clinical Oncology xxx (2011) 1e8 date, a small number of studies investigating the benefits of adaptive radiotherapy in the treatment of HNSCC, generally in patients with marked changes in anatomy, have been carried out. These have shown an overall benefit in terms of reducing dose to normal tissues, particularly the parotid gland, and improved tumour target coverage [2e4]. However, it remains unclear whether all patients derive a significant benefit from adaptive replanning. With this pilot study, we aimed to assess the feasibility of routinely offering adaptive radiotherapy to HNSCC patients in our department; to determine whether this results in clinically significant benefits in either tumour target coverage or normal tissue sparing; and to determine which patients receive the greatest benefit, with a view to selecting these patients for routine adaptive therapy. Materials and Methods Study Design and Eligibility Two cohorts of patients were enrolled. Cohort 1 consisted of postoperative patients requiring adjuvant radiotherapy. Any patient receiving adjuvant radiotherapy/ chemoradiotherapy was eligible. Cohort 2 consisted of patients receiving radical radiotherapy/chemoradiotherapy for HNSCC. Patients were eligible for cohort 2 if they had primary tumour size 3 cm on imaging or examination and/or cervical nodal lymphadenopathy 3 cm in size on clinical examination, and had not undergone surgery. This trial was approved by the Alberta Cancer Research Ethics Committee, and all patients were required to give informed consent. Interventions Patients underwent standard department computed tomography (CT) simulation, supine with a head rest and a thermoplastic (Orfit Ò -Orfit Industries, Jericho, NY, USA) immobilisation shell. Normal tissues and target volumes were contoured in the Eclipse TM treatment planning system (Varian Oncology Systems, Palo Alto, CA, USA). Helical tomotherapy (TomoTherapy, Inc, Madison, WI, USA) was used for generation of the intensity-modulated radiotherapy plan and treatment delivery. All target volumes were determined by the radiation oncologist. In cohort 1, the CTV60 encompassed the pathologically involved regions at the time of surgery; in cohort 2, the GTV was defined as the combined gross primary and nodal disease. For both cohorts, the CTV54 was the clinically/pathologically negative anatomical regions at risk for microscopic disease determined by knowledge of the natural history of the disease. A 5 mm volumetric expansion was used to generate the PTV60 and PTV54 from the CTV60 and CTV54, and the PTV66 from the GTV, with tissue less than 3 mm below the skin surface excluded. These volumes were treated to 66, 60 or 54 Gy over 30 fractions using a simultaneous integrated boost technique with once-daily fractionation. One patient in the definitive group and one in the adjuvant group were treated with 64 Gy to the high dose PTV. Standard department dose constraints and optimisation parameters were used. Treatment goals are shown in Table 1. A second planning CT scan was carried out after fraction 15. The two planning scans were manually co-registered primarily based on the cervical spine and other bony landmarks d and secondarily on the external patient contour. The original organs at risk and target structure sets were copied onto the new scan before being manually modified, by the radiation oncologist who determined these structures on the first planning scan, based on changes in patient anatomy. The second planning CT and structure set were exported into the helical tomotherapy planning system and a new plan was generated. This plan was started from fraction 20. Standard helical tomotherapy quality assurance procedures were carried out and patients were reviewed in multidisciplinary rounds to ensure consistency in contouring. Dosimetric Measurements The target and normal tissue dose parameters recorded were: PTV60/66 D 95 (minimum dose that covers 95% of the high dose PTV); PTV60/66 D 1 (maximum dose received by 1% of the high dose PTV). The same parameters were collected for PTV54. Organ at risk parameters were: spinal cord maximum dose; combined parotid mean dose; combined parotid V 26 (percentage volume of parotid receiving 26 Gy); the volume of normal tissue outside the PTVs receiving 50 Gy (normal tissue V 50 ); and the volume of skin (defined as a 5 mm strip beneath the surface of the body contour) receiving 50 Gy (skin V 50 ). We selected these parameters because of the information each provides. For the target parameters, D 95 reflects the adequacy of target coverage and D 1, the clinically significant tissue maximum within the PTV or D 1. Both the mean dose to the parotid and the volume of parotid receiving 26 Gy have been found to be predictive of xerostomia and were included in this analysis [5e7]. 50 Gy was judged to be a dose that would probably cause significant mucositis, Table 1 Treatment planning goals Parameter PTV 66 PTV 60 PTV54 Parotid Spinal cord Brainstem Optic apparatus Unspecified tissue Treatment goal 90% to receive 66 Gy; 1% to receive 60 Gy; 20% to receive 72.6 Gy 90% to receive 60 Gy; 1% to receive 54 Gy; 20% to receive 66 Gy 90% to receive 54 Gy; 3% to receive 47 Gy; 10% to receive 66 Gy (with PTV60) or 72.6 Gy (with PTV66) 50% of one or both parotids to receive 30 Gy Maximum dose 45 Gy Maximum dose 54 Gy Maximum dose 45 Gy Maximum dose 68 Gy

3 L. Capelle et al. / Clinical Oncology xxx (2011) 1e8 3 desquamation and late soft tissue changes. We hypothesised that plans with greater volumes of skin receiving 50 Gy may have higher risks of moist desquamation, as suggested by historical estimates of skin tolerance [8]. We also hypothesised that plans treating larger volumes of normal tissue to doses 50 Gy would probably have greater rates of late toxicity, potentially including subcutaneous fibrosis, impairment of swallowing function and large vessel changes depending on where this dose is distributed. The volume of normal tissue receiving 50 Gy is analogous to the irradiated volume defined in ICRU 62 [9]. Comparison of Plans The dose accumulated in the first 19 fractions is identical for both the non-adapted and adapted distributions, and is generated by applying the first plan (plan 1) to the pretreatment planning scan (CT1). Two dose distributions were generated for the second phase of treatment. The cumulative adapted dose distribution combines the dose statistics from the first 19 fractions with 11 fractions of dose determined by delivery of the second plan (plan 2) on the mid-treatment planning CT scan (CT2). This was carried out by using manual registration of the two planning scans as described above. The cumulative non-adapted dose distribution combines the dose statistics from the first 19 fractions with the final 11 fractions of dose determined by delivery of plan 1 on CT2. Doseevolume histogram statistics were simply summed taking into account the number of fractions in each plan. Potential Predictive Factors Several factors were hypothesised to potentially predict the magnitude of benefit from adaptive replanning. Patient factors included change in lateral neck separation at C1 vertebral level, mid-ptv60/66 and thyroid notch level (measured in mm); and change in weight. The initial GTV volume (cm 3 ), T and N stage and treatment cohort were also investigated. One technical factor was also considered. This was the quality of registration between CT1 and CT2. Marked changes in patient position between planning scans would result in relative differences in position between target and normal tissue volumes, and changes in dosimetry between adapted and non-adapted plans could be in part due to this. We accounted for this possibility by including a registration fit parameter, which was a semi-quantitative assessment of how much change in dosimetry may be due to a change in patient positioning based on the closeness of match in patient positioning between the two scans. This was determined by comparing the volume of the patient body encompassed by the body contour over the region of the PTV54 between the two scans using rigid manual registration. Excellent registration would result in 100% of the body volume over this region in CT2 falling within the body contour of CT1. Less optimal registration would result in a percentage of the CT2 body volume falling outside the original body contour. A cut-off value of 95% was used to differentiate between a very good and a moderately good match in patient position between the two scans. Statistical Analysis Comparisons between the paired CT volume measurements were obtained using a Wilcoxon signed-rank test. Comparisons between the non-adapted and adapted dosimetric parameters were carried out using the Manne Whitney U test. The correlation between the potential predictive factors on dosimetric end points was carried out using Spearman s rho correlation test. A P-value of <0.05 was considered for all statistical significance. All statistical analyses were conducted using SPSS (originally Statistical Package for the Social Sciences) statistical software version 15. Results Patient Demographics The first 20 patients accrued to this ongoing pilot study, between December 2008 and February 2010, are presented for preliminary analysis. The patient and tumour characteristics are shown in Table 2. Patient and Tumour Parameter Changes during Radiotherapy Table 3 shows the changes in patient and plan parameters between CT1 and CT2. On average, lateral neck separation reduced by 6e7 mm (range e5 to 18 mm) between scans. The median parotid volume loss was 17.5% (range e1.0 to Table 2 Patient and tumour characteristics Median age 58 (range 27e75) Median weight (kg) 74.9 (range 51e118) Median initial GTV volume (cm 3 ) 60.4 (range 15.2e308.5) Diagnosis Adjuvant 7 Nasopharyngeal 3 carcinoma Other definitive 10 T stage Tx N stage Stage I 0 II 2 III 8 IVa 8 IVb 2 Concurrent chemotherapy* 19/20 * One patient with T1N0 supraglottic SCC received radiotherapy alone.

4 4 L. Capelle et al. / Clinical Oncology xxx (2011) 1e8 Table 3 Median patient and tumor dimensions at the time of initial (CT1) and mid-treatment (CT2) scans CT1 CT2 Lateral neck separation C1 level (cm) Lateral neck separation mid PTV60/ level (cm) Lateral neck separation thyroid notch level (cm) GTV volume (cm 3 ) CTV60 volume (cm 3 ) PTV60/66 volume (cm 3 ) PTV54 volume (cm 3 ) Right parotid volume (cm 3 ) Left parotid volume (cm 3 ) Weight (kg) %) with a median reduction in PTV60/66 of 16% (range 0 to 45%) and PTV54 of 6.8% (range e1.2 to 19%). Median reductions in GTV and CTV60 were 28.8% (range 1.6 to 60%) and 4.1% (range e0.1 to 10%), respectively. The median weight loss between planning scans was 2.7% of preradiotherapy weight (range e4.2 to 10.2%). The median weight on completion of treatment was 67.2 kg, with a median weight loss of 6.0% of preradiotherapy weight (range e1.2 to 13.6%). Comparison between Non-adapted and Adapted Doses Table 4 shows the comparison between the non-adapted and adapted doses. Overall there was a statistically significant benefit to adaptive therapy in improving PTV60/66 D 1 (P ¼ 0.01); PTV54 D 95 and D 1 (P ¼ 0.02 and P < , respectively); spinal cord maximum (P ¼ 0.006); and volume of skin receiving 50 Gy (P ¼ 0.002). Generally these benefits were small in magnitude, with the largest being an absolute 0.5 Gy benefit in spinal cord maximum dose and a 26 cm 2 reduction in skin V 50. Benefit of Adaptation by Cohort The level of benefit to replanning was broken down by cohort, as shown in Table 5. Cohort 1 (adjuvantly treated) patients had significant, but small benefits in absolute reductions in PTV54 D 1 of 0.3 Gy (P ¼ 0.02) and a spinal cord maximum dose of 0.5 Gy (P ¼ 0.04), with a trend for an improvement in PTV54 D 95 of 0.6 Gy (P ¼ 0.08). In contrast, cohort 2 (definitively treated) patients had more clinically significant changes, with reductions in PTV66 and 54 D 1 (0.5 Gy, P ¼ and 0.9 Gy, P ¼ 0.003, respectively); spinal cord maximum (0.6 Gy, P ¼ 0.04); combined mean parotid dose (0.6 Gy, P ¼ 0.045); normal tissue V 50 (67 cm 3, P ¼ 0.02); skin V 50 (33 cm 2, P ¼ 0.002). The differences in dose distribution between an adapted and non-adapted plan for a definitively treated patient are shown in Figure 1. Potential Predictive Factors A summary of the relationship of relevant potential predictive factors and dosimetric outcomes is given in Table 6. The most consistently predictive factors were changes in neck separation. Greater reduction in neck separation at the level of the thyroid notch was predictive of a greater reduction in PTV54 D 1 (Spearman correlation r s ¼ 0.51, P ¼ 0.02); spinal cord maximum dose (r s ¼ 0.73, P < ); normal tissue V 50 (r s ¼ 0.71, P < ). The extent of reduction in neck separation at the mid-ptv level was also predictive of the magnitude of benefit in the reduction in spinal cord maximum (r s ¼ 0.55, P ¼ 0.01) and in the combined mean parotid dose (r s ¼ 0.64, P ¼ 0.002). Greater weight loss was predictive of a reduction in PTV54 D 1 (r s ¼ 0.60 P ¼ 0.006) and a higher T stage in definitively treated patients was predictive of greater benefit in normal tissue V 50 (median improvement with adaptation 46 cm 3 Table 4 Comparison between median non-adapted and adapted doses Parameter Treatment component 1 (19 fractions) Treatment component 2 (11 fractions) Non-adapted dose y Received adapted dose z P-value* PTV60/66 D 95 (Gy) PTV60/66 D 1 (Gy) PTV54 D 95 (Gy) PTV54 D 1 (Gy) < Spinal cord max (Gy) Combined parotid mean (Gy) Combined parotid V 26 (%)z,x N tissue 50 Gy (cm 3 ) x Skin 50 Gy (cm 2 ) x,jj * Wilcoxon rank test P-value was used to compare the non-adapted and received adapted dose. y Plan 1 delivered on CT2. z Plan 2 delivered on CT2. x Figures represent parameter adjusted for number of fractions delivered. jj Skin reported as square area of 5 mm thickness.

5 Table 5 Comparison of absolute level of benefit by cohort. difference between adapted and non-adapted plans. Positive figures indicate benefit, negative detriment, to replanning Parameter Cohort 1 Cohort 2 Relative* benefit (%) benefit P-valuey Relative* benefit (%) benefit PTV60/66 D 95 (Gy) PTV60/66 D 1 z (Gy) PTV54 D 95 (Gy) PTV54 D 1 (Gy) z Spinal cord maximum (Gy) Combined parotid mean (Gy) e2.0 e Combined parotid V 26 (%) x e6.1 e N tissue 50 Gy (cm 3 ) x e5.3 e Skin 50 Gy (cm 2 ) x,jj * Relative difference between adapted and non-adapted plans for final 11 fractions. y P-values represent the Wilcoxon rank test values. z 1% dose maximum. x Figures represent parameter adjusted for number of fractions delivered. jj Skin reported as square area of 5 mm thickness. L. Capelle et al. / Clinical Oncology xxx (2011) 1e8 5 P-value y for T3/4 versus 17 cm 3 for T0e2 tumours, P ¼ 0.01). There was no correlation between initial GTV volume, lateral separation at the level of C1, cohort or initial N stage and the level of benefit to adaptive replanning. The accuracy of CT registration as described by registration fit (defined above) was generally good, with 14 patients having a very good fit. The median score was 95.2%, with range 85.2e99.0%. There was no correlation Fig 1. Changes in the dose distribution for the non-adapted versus the adapted plan. Left panels: axial and coronal images of the non-adapted plan. Right panels: axial and coronal images of the adapted plan. Blue contour ¼ parotid glands; magenta contour ¼ PTV66. In the adapted plan there is reduced dose to the left parotid (white arrows) and improved PTV66 coverage (black arrow heads). The right parotid receives a lower dose in the non-adapted plan (white arrowhead), but this is due to inadequate PTV66 coverage. The spinal cord dose is unchanged (black arrow). Overall the high dose volume is smaller in the adapted plan.

6 6 L. Capelle et al. / Clinical Oncology xxx (2011) 1e8 Table 6 Factors predictive of benefit to adaptive radiotherapy Skin V 50 Combined PTV60 D 1 PTV54 D 1 SC maximum Normal Tissue V 50 mean parotid dose GTV volume* Spearman s rho e e0.26 e0.55 e0.45 e0.20 P-value Weight change* Spearman s rho P-value Change in separation C1* Spearman s rho e e0.01 e P-value Change in separation Spearman s rho e mid-ptv* P-value <0.001 Change in separation Spearman s rho thyroid notch* P-value < < T stage (T0e2 versus T3/4)y Median difference 0 (Gy) e0.4(gy) 0.2 (Gy) 51 (cm 3 ) 29 (cm 2 ) e1.3(gy) P-value N stage (N0e2 versus N3)y Median difference e0.3(gy) 0.4 (Gy) 0.6 (Gy) 19.5 (cm 3 ) 15 (cm 2 ) 1.3 (Gy) P-value * Spearman correlation between continuous parameters and dose outcomes. y Difference in median benefit between the two groups for discrete variables using the ManneWhitney U test. Positive values show greater benefit for higher T or N stage. between registration fit and any of the dose parameters calculated (data not shown). Discussion We have shown a modest benefit to adaptive radiotherapy in these relatively unselected patients. For the group overall we found significant, although small, benefits to adaptive therapy with a reduced PTV D 1, a reduced maximum spinal cord dose and a reduced volume of skin receiving 50 Gy. For definitively treated patients there were significant benefits also in a reduced mean parotid dose and volume of normal tissue outside the PTV receiving 50 Gy. The absolute magnitude of difference was also greater for definitively treated patients than that for the group overall. One of the strengths of this study was that it included a relatively unselected population, including adjuvantly treated patients. The purpose of the study was to assess the role of routine replanning in patients commonly seen in practice, in comparison with other trials that looked at more selected patients with large changes in PTV volumes or weight, where the benefits to adaptive replanning are clearer. We have shown that there is no benefit to adaptive replanning in adjuvantly treated patients, and that the benefit to routine adaptive replanning in unselected definitively treated patients is minimal. We assessed a number of potential predictive factors to determine which patients received the greatest benefit to treatment adaptation and, as expected, degree of weight loss and reduction in neck separation were found to be predictive. With only 13 definitively treated patients, our power to determine which pretreatment parameters for these patients might be predictive was limited. Numerically, patients with nasopharyngeal carcinoma seemed to receive the greatest benefit with adaptation, with a mean absolute reduction in spinal cord maximum dose of 1.2 Gy; a mean parotid dose and parotid V 26 of 1.2 Gy and 6.3%, respectively; a normal tissue V 50 of 150 cm 3 ; skin V 50 of 37 cm 2. We believe these are clinically significant differences, and suggest that electively scheduling adaptive replanning when initiating treatment for patients undergoing radiotherapy for nasopharyngeal carcinoma may be beneficial. Data regarding the biology of human papilloma virusrelated oropharyngeal HNSCC are accumulating [10] and these malignancies may have similar radioresponsiveness characteristics to nasopharyngeal carcinoma, and may receive similar levels of benefit from treatment adaptation. We are planning to assess this factor in the future with a larger patient group. The benefits we found in our study are less than reported in more highly selected populations. Hansen et al. [2] carried out a retrospective review of 13 HNSCC patients who underwent replanning during their radiotherapy course due to significant tumour shrinkage or weight loss. Rigid registration was carried out, with eight patients having a new shell made. In the absence of adaptive replanning, there was a significant reduction in PTV coverage, with a mean reduction in the high dose and low dose PTV D 95 of 2.2 and 3 Gy, respectively. Maximum doses to the spinal cord, brainstem and the mandible increased by 4, 2.6 and 1.7 Gy, respectively. The right parotid mean dose and V 26 increased by 2.9 Gy and 10%, respectively. However, these differences in dose may be overestimated if the CT registration (particularly in the eight patients with the new shell) is poor. Wang et al. [3] reported on a retrospective study of 28 patients with nasopharyngeal carcinoma who underwent replanning, using deformable registration, before fraction 25 (of 33). Significant benefits to replanning were found in CTV1 V 100 and nodal GTV V 100 (increased 4.9 and 1.8%, respectively); and spinal cord point maximum, left parotid mean dose and right parotid V 30 (decreased 5 Gy,

7 L. Capelle et al. / Clinical Oncology xxx (2011) 1e Gy and 3.2%, respectively). Again these patients would be expected to have greater benefits to replanning than the general definitive HNSCC population, particularly if these patients were replanned because of changes in anatomy. Wu et al. [4] conducted a complex dosimetry study investigating the effects of adaptive radiotherapy and of different PTV expansion sizes in various dosimetric parameters. Eleven patients with locally advanced HNSCC underwent weekly CTs and plans were generated using these scans with PTV expansions of 5, 3 or 0 mm, using deformable registration. The effects of replanning with a PTV expansion of 0 mm were reported. There were no significant differences in target coverage, or doses to the brainstem, spinal cord or mandible, but reductions in the mean parotid dose of 3, 5 and 8% with replanning once, twice or six times during treatment. The goal of our trial was to assess the potential benefits of adapting radiotherapy plans. A separate but related issue is what detriment is there to not replanning. This is a question that arises frequently in clinical practice in patients with significant weight loss or reduction in neck separation during treatment. By comparing the difference between the initial planned dose and the non-adapted doses for our patient group, we found that differences in PTV D 95,D 1 and doses to sensitive normal structures were minimal, as shown in Table 7. Again our figures are lower than for more highly selected populations [11e18]. Although it is reassuring that the absolute changes in dose related to anatomical changes are small, these figures do not take into account the risk of changes in dose due to a reduction in the adequacy of immobilisation. We suggest that the adequacy of immobilisation may be a more pressing reason to replan a patient than anatomy changes in most HNSCC patients. One weakness of this study was that deformable registration of the two planning CT scans was not used. The accuracy of CT registration was not found to be predictive of the level of benefit for any target or normal tissues, so any errors related to the accuracy of CT registration do not seem to have significantly affected our results. Without the Table 7 Differences between the initial plan and the non-adapted plan Parameter Combined group Cohort 1 Cohort 2 difference difference difference PTV60/66 D 95 (Gy) PTV60/66 D 1 (Gy) e0.3 e0.1 e0.3 PTV54 D 95 (Gy) PTV54 D 1 (Gy) e0.4 e0.3 e0.4 Spinal cord e0.2 e0.2 e0.2 maximum (Gy) Combined parotid e0.2 e0.1 e0.2 mean (Gy) Combined parotid V 26 (%) e0.3 e0.2 e0.3 Positive ¼ initial plan dose greater than non-adapted plan. deformation process matching each dose voxel between the two CT scans, it is not possible to accurately sum the dose statistics from the two parts of the treatment, although the non-adapted and adapted portions of the treatment can be directly compared. Conversely there are some limitations to deformation software, and although there have been significant improvements over recent years, it should still be used with caution, particularly with respect to computer generated automatic segmentation, until a centre has experience with it in their context [19,20]. In our study, the relevant tumour and normal tissue volumes were drawn by a radiation oncologist on each scan, with the same radiation oncologist redrawing the volumes on the second planning scan, which we consider the gold standard in target and normal tissue volume delineation. In routine clinical practice this can be onerous, and more automated approaches are being actively investigated, including a current randomised controlled trial at the MD Anderson Cancer Center investigating the role of adaptive radiotherapy in HNSCC using automated segmentation [21]. Finally, the safety of adaptive radiotherapy should be considered. In our study, although the adapted PTV volumes were smaller than the original volumes, all regions deemed to be at risk of disease in the initial scan were included in the subsequent PTVs. With a median follow-up of 18 months, one patient with T1N3 nasopharyngeal carcinoma recurred locally at the base of the skull at 9 months. The local recurrence was at the superior edge of the treatment field. There was no reduction in field at this region and the recurrence was not a result of the modification of treatment volumes. A second patient with T3N3 nasopharyngeal carcinoma relapsed distantly at 9 months. Conclusion Adaptive radiotherapy proved feasible. Modest benefits in reduction in PTV D 1, low dose PTV coverage, spinal cord maximum dose and volume of skin receiving 50 Gy were found in this relatively unselected population. Definitively treated patients also had benefits in a reduction in mean parotid dose and the volume of normal tissue outside the PTV receiving 50 Gy. Patients with nasopharyngeal carcinoma, initial stage T3/4 and greater weight loss or a reduction in lateral neck separation received the greatest benefit. No local recurrences occurred related to PTV modification. There is no significant benefit to adaptive radiotherapy in patients receiving adjuvant radiotherapy. Conflict Of Interest Statement Lisa Capelle - No conflicts of interest to declare Marc Mackenzie - No conflicts of interest to declare Colin Field - No conflicts of interest to declare Matthew Parliament - No conflicts of interest to declare Sunita Ghosh - No conflicts of interest to declare Rufus Scrimger - No conflicts of interest to declare

8 8 L. Capelle et al. / Clinical Oncology xxx (2011) 1e8 References [1] Vergeer MR, Doornaert PAH, Rietveld DHF, et al. Intensitymodulated radiotherapy reduces radiation-induced morbidity and improves health-related quality of life: results of a nonrandomized prospective study using a standardized follow-up program. Int J Radiat Oncol Biol Phys 2009;74:1e8. [2] Hansen EK, Bucci MK, Quivey JM, et al. Repeat CT imaging and replanning during the course of IMRT for head and neck cancer. Int J Radiat Oncol Biol Phys 2006;64:355e362. [3] Wang W, Yang H, Hu W, et al. Clinical study of the necessity of replanning before the 25th fraction during the course of intensity-modulated radiotherapy for patients with nasopharyngeal carcinoma. Int J Radiat Oncol Biol Phys 2010;77:617e621. [4] Wu Q, Chi Y, Chen PY, et al. Adaptive replanning strategies accounting for shrinkage in head and neck IMRT. Int J Radiat Oncol Biol Phys 2009;75:924e932. [5] Blanco AI, Chao C, El Naqa I, et al. Dose-volume modeling of the salivary function in patients with head-and-neck cancer receiving radiotherapy. Int J Radiat Oncol Biol Phys 2005;62: 1055e1069. [6] Eisbruch A, Haken RKT, Kim HM, et al. Dose, volume, and function relationships in parotid salivary glands following conformal and intensity-modulated irradiation of head and neck cancer. Int J Radiat Oncol Biol Phys 1999;45:577e587. [7] M unter MW, Karger CP, Hoffner SG, et al. Evaluation of salivary gland function after treatment of head-and-neck tumors with intensity-modulated radiotherapy by quantitative pertechnetate scintigraphy. Int J Radiat Oncol Biol Phys 2004;58: 175e184. [8] Archambeau JO, Pezner R, Wasserman T. Pathophysiology of irradiated skin and breast. Int J Radiat Oncol Biol Phys 1995;31: 1171e1185. [9] International Commission on Radiation Units and Measurements (ICRU). Report No. 62, Prescribing, recording, and reporting photon beam therapy (supplement to ICRU report 50), vol. 62. Bethesda MD: ICRU; [10] Mannarini L, Kratochvil V, Calbrese L, et al. Human papilloma virus (HPV) in head and neck region: review of literature. Acta Otorhinolaryngol Ital 2009;29:119e126. [11] Bhide SA, Davies M, Burke K, et al. Weekly volume and dosimetric changes during chemoradiotherapy with intensity-modulated radiation therapy for head and neck cancer: a prospective observational study. Int J Radiat Oncol Biol Phys 2010;76:1360e1368. [12] Han C, Chen YJ, Liu A, et al. Actual dose variation of parotid glands and spinal cord for nasopharyngeal cancer patients during radiotherapy. Int J Radiat Oncol Biol Phys 2008;70: 1256e1262. [13] Vasquez Osorio EM, Hoogeman MS, Al-Mamgani A, et al. Local anatomic changes in parotid and submandibular glands during radiotherapy for oropharynx cancer and correlation with dose, studied in detail with nonrigid registration. Int J Radiat Oncol Biol Phys 2008;70:875e882. [14] Castadot P, Lee JA, Geets X, et al. Adaptive radiotherapy of head and neck cancer. Semin Radiat Oncol 2010;20:84e93. [15] Loo H, Fairfoul J, Chakrabati JC, et al. Tumour shrinkage and contour change during radiotherapy increase the dose to organs at risk but not the target volumes for head and neck cancer patients treat on the tomotherapy HiArt TM system. Clin Oncol 2011;23:40e47. [16] Lee C, Langen KM, Lu W, et al. Assessment of parotid gland dose changes during head and neck cancer radiotherapy using daily megavoltage computed tomography and deformable image registration. Int J Radiat Oncol Biol Phys 2008;71: 1563e1571. [17] Han C, Chen Y, Liu A, et al. Actual dose variation of parotid glands and spinal cord for nasopharyngeal cancer patients during radiotherapy. Int J Radiat Oncol Biol Phys 2008;70: 1254e1262. [18] O Daniel JC, Garden AS, Schwartz DL, et al. Parotid gland dose in intensity-modulated radiotherapy for head and neck cancer: is what you plan what you get? Int J Radiat Oncol Biol Phys 2007;69:1290e1296. [19] Castadot P, Lee JA, Parraga A, et al. Comparison of 12 deformable registration strategies in adaptive radiation therapy for the treatment for head and neck tumors. Radiother Oncol 2008;89:1e12. [20] Tsuji SY, Hwang A, Weinberg V, et al. Dosimetric evaluation of automatic segmentation for adaptive IMRT for head-and-neck cancer. Int J Radiat Oncol Biol Phys 2010;77:707e714. [21] Schwartz DL, Dong L. Adaptive radiation therapy for head and neck cancer d can an old goal evolve into a new standard? J Oncol (in press).