A Study to Assess the Dosimetric Impact of the Anatomical Changes Occurring in the Parotid Glands and Tumour Volume during Intensity Modulated Radiotherapy using Simultaneous Integrated Boost (IMRT‐SIB) in Head and Neck Squamous Cell Cancers

Abstract Background Anatomical variations in head and neck cancer during IMRT leads to volume shrinkage, results in dosimetric variations in tumour and normal tissue including parotid glands, with a risk of radiation toxicities. Methods 30 patients with a stage II–IV head and neck squamous cell carcinoma (HNSCC) were treated with definitive IMRT‐SIB and concomitant chemotherapy. Volumetric and dosimetric variations were evaluated during the period of IMRT by recalculating and obtaining dose‐volume histograms of re‐contoured target volumes and parotid glands on repeat CT scans taken multiple times during treatment (CT1, CT2, CT3 and CT4). Results Result showed significant (p < 0.001) mean decrease in both primary and nodal tumors volume with time whereas increase (p < 0.01 or p < 0.001) in respective V100 (%) and D2% (Gy). The mean parotid gland dose increased (p < 0.01 or p < 0.001) with time, whereas parotid gland volume and distance between plan isocenter and centre of mass of parotid glands decreased (p < 0.05 or p < 0.001) with time. Patient's mean weight and neck circumference both decrease (p < 0.001) with time whereas ECOG score increase (p < 0.001) with time. The mucosal toxicity increased significantly (p < 0.001) with time. The change in both weight and neck circumference showed significant (p < 0.001) and direct (positive correlation) association with change in parotid gland volume. Conclusion If the PTV and normal anatomy are changing with time, adaptive IMRT would be beneficial radiation dose delivery where possible.


| INTRODUCTION
Conformal radiotherapy techniques such as intensitymodulated radiotherapy (IMRT) in head and neck cancers (HNC) have allowed radiation oncologists to deliver curative radiation doses to the tumour with higher accuracy while restricting the dose to organs at risk (OARs), consequently reducing treatment-related morbidity.
However, steep dose gradients are produced in IMRT which imply that there should be no or minimal changes in the patient's anatomy, tumour volume and OARs position so that target volume coverage is not compromised and radiation overdose to critical and normal structures is prevented, thus resulting in enhanced response and reduced radiation toxicity. 1 Appearance of anatomical variations during the period of radiotherapy in HNC is routinely observed and is due to body weight loss, primary tumour shrinking, parotid gland volume reduction and variation in volume of normal tissue irradiated, which may result in discrepancy in planned dose and actual dose administered causing dosimetric variation of target volume and critical structures with a risk of compromised dose coverage to the target volumes and overdose to the parotid glands and normal tissue influencing treatment response and associated toxicities. [2][3][4] Therefore, our aim in this study was to evaluate anatomic and volumetric alterations in the parotid glands and tumour volume of HNC patients being treated with IMRT-SIB, and to study the dosimetric impact of these anatomic changes on dose variation to target volume and parotid glands.

| MATERIALS AND METHODS
30 newly diagnosed, biopsy proven patients with stage II-IV (AJCC Cancer Staging Manual, 8 th edition) Head and Neck Squamous Cell Carcinoma (HNSCC) registered at Radiotherapy Department, King George's Medical University, Lucknow India were prospectively enrolled between June 2019 and May 2020. All patients were treated with IMRT step-and-shoot modality and received concomitant chemotherapy. Study specific informed consent was taken from all the patients. Study was approved by Institutional Ethics Committee, King George's Medical University. The study was done in accordance with the Declaration of Helsinki and its subsequent amendments, good clinical practice guidelines, and other legal requirements.
Each patient underwent a planning kilo voltage computerized tomography scan (KVCT-scan) of the head-and neck region with 3-mm slice thickness. Patients were scanned in the supine position, immobilized on a flat table top with a customized five fixation points thermoplastic facemask and a head-and-neck immobilization board (AIO Board). The planning KVCT images were transferred to a treatment planning system (Monaco Treatment Planning System, Elekta), and contours for the target volumes and normal organs were drawn.
Initial planning CT1 (Plan1) with intravenous contrast agents was acquired from the vertex to the carina. Target volumes and normal structures were manually contoured on the axial slices of the planning CT scan. Gross tumour volume (GTV) was delineated to include primary tumour (GTV-P) and enlarged neck nodes (GTV-N) in the enhanced CT images. Three clinical target volume (CTVs), based on the current clinical practice at this institution, were used for each patient: (a) CTV high, which encompassed the GTVs plus a physician-determined planning margin, was prescribed 66 Gy (at 2.2 Gy per fraction) (b) CTV intermediate, which surrounded the lymph nodes that have a high probability of cancer involvement was prescribed 60 Gy (at 2 Gy per fraction) and (c) CTV low, which encompassed those lymph nodes with a relatively lower probability of cancer involvement and was prescribed 54 Gy (at 1.8 Gy per fraction).
For treatment planning, the PTVs encompassed the CTVs with a 5-mm margin. The IMRT beam arrangements consisted of seven/nine co-planar beams. A simultaneous integrated boost technique was used to deliver 66 Gy, 60 Gy and 54 Gy to PTV high, PTV intermediate and PTV low respectively, in 30 fractions over 6 weeks, and the following dose constraints were set on the OAR: maximum dose for the spinal cord, 45 Gy; maximum dose of the brain stem, 54 Gy; mean dose for at least one parotid gland, 26 Gy, although both parotid glands were tried to spare.
All patients received weekly chemotherapy with cisplatin (35mg/m 2 ) concurrent with radiotherapy. Patients were weighed and neck circumference of each patient was taken weekly during treatment. Patients were assessed weekly for treatment-related toxicities. During treatment period, repeat kVCT images with contrast were acquired after patients received 10, 20 and 29 fractions each with the same thermoplastic cast and following the same protocols as during the acquisition of initial CT1 to generate CT2, CT 3 and CT 4. The GTV primary and nodal were delineated as the mass shown in the enhanced CT images. Both the parotid glands were also contoured as seen on the repeat scans of each patient. The initial IMRT plan (Plan 1/CT1) was transferred to CT2, CT3 and CT4 based on carefully matched isocentre and bony alignment to make Plan2, Plan3 and Plan4 respectively. Dose distributions of these plans were recalculated to obtain dose-volume histograms (DVHs) of re-contoured target volumes and parotid glands. The changes in volume, distance and dose were analyzed for each patient. To quantify the positional shifts of the parotid glands, we calculated the distance from the centre of mass (COM) of the parotid glands to the matched isocentre for CT scan (CT1, CT2, CT3 and CT4).

| RESULTS AND OBSERVATIONS
The present study assesses the dosimetric impact of anatomical changes occurring in the parotid glands and tumour volume during IMRT-SIB for HNSCC. A total of 30 patients were recruited and evaluated. Patients were treated with radiotherapy 30 fractions over 6 weeks. The primary outcome measures of the study were primary and nodal tumour related volume and dosimetric variables and volume, mean dose and positional shift of parotid glands. The secondary outcome measures of the study were changes in weight, neck circumference and performance status of patients and correlation between these and the primary outcome measures. All measures were assessed at time of CT1, CT2, CT3 and CT4. We also assessed treatment related acute toxicities in patients during treatment. Most commonly involved site was oropharynx, followed similarly by larynx and oral cavity with oropharynx involvement accounting together for 70.0% of the cases. Patients with stage III and IVA disease made up 70.0% of the study population. 46.7% of the patients had moderately differentiated tumors.

| The effect of treatment on patient's weight, neck circumference and ECOG is summarised in table 3
Comparing the mean weight, neck circumference and ECOG score, ANOVA showed significantly different weight (F = 28.46, p < 0.001), neck circumference (F = 16.21, p < 0.001) and ECOG score (F = 11.00, p < 0.001) among the periods (Table 3).
Further, comparing the difference in mean weight, neck circumference and ECOG score between the periods ( Table 4), Newman-Keuls test showed significantly (p < 0.01 or p < 0.001) different and decreased weight and neck circumference both at CT3 and CT4 as compared to both CT1 and CT2. Furthermore, mean weight also decreased significantly (p < 0.05) at CT4 as compared to CT3. In contrast, mean ECOG score increased significantly (p < 0.001) at CT4 as compared CT1, CT2 and CT3 but not differ (p>0.05) between CT1, CT2 and CT3 i.e. found to be statistically the same. The net mean decrease (i.e. mean change from CT1 to CT4) in weight and neck circumference of patients was found to be 8.1% and 4.1% respectively whereas ECOG score increased by 20.9%.

| The effect of treatment on GTV primary tumour related variables [GTV P vol (cc), GTV P V100 (%), GTV P D100% (Gy), GTV P D98% (Gy) and GTV P D2% (Gy)] is summarised in table 5
The mean GTV P vol showed marked decrease with time. Other variables had increased with time.

| The effect of treatment on GTV nodal tumour related variables [GTV N vol (cc), GTV N V100 (%), GTV N D100% (Gy), GTV N D98% (Gy) and GTV N D2% (Gy)] is summarised in table 5
The mean GTV N vol showed marked decrease with time whereas both GTV N V100 and GTV N D2% showed marked increase with time (Table 5).
Further, for each GTV nodal tumour variable, comparing the difference in mean between periods (Table 6), Newman-Keuls test showed significant (p < 0.01 or p < 0.001) decrease in GTV N vol at CT2, CT3 and CT4 as compared to CT1. It also showed significant (p < 0.05) decrease at both CT3 and CT4 as compared to CT2. In contrast, GTV N V100 and GTV N D98% both showed significant (p < 0.05 or p < 0.01) increase at CT4 as compared to CT1, CT2 and CT3 but found similar (p>0.05) between other periods. Conversely, GTV N D2% showed significant (p < 0.05) increase at both CT3 and CT4 as compared to both CT1 and CT2 but found similar (p>0.05) between CT1 and CT2, and CT3 and CT4 i.e. did not differ significantly.

| Effect of treatment on the parotid glands
3.5.1 | The effect of treatment on the parotid gland which received higher mean dose at planning on CT1 relative to contralateral side [H-Parotid gland D mean (Gy), H-Parotid gland volume (cc) and Distance between plan isocenter and COM of H-Parotid gland (cm)] is summarised in table 7 The mean H-Parotid gland D mean showed linear increase with time whereas both H-Parotid gland volume and distance between plan isocenter and COM of H-Parotid gland showed linear decrease with time (Figures 1 and 2).
Further, for each H-Parotid gland related variable, comparing the difference in mean between periods (Table 8), Newman-Keuls test showed significant (p < 0.001) increase in H-Parotid gland D mean at CT4 as compared to other periods whereas it was found to be statistically the same (p > 0.05) between other periods. In contrast, both H-Parotid gland volume and distance between plan isocenter and COM of H-Parotid gland showed significant (p < 0.05 or p < 0.001) decrease at CT2, CT3 and CT4 as compared to CT1. Both variables also showed significant (p < 0.001) decrease at both CT3 and CT4 as compared to CT2. The H-Parotid gland volume showed significant (p < 0.05) decrease at CT4 as compared to CT3.
At final evaluation the H-Parotid gland shrank in volume by 31.6% and shifted medially by 9.2% from CT1 to CT4 with a net mean increase in D mean of 7.3%.  between plan isocenter and COM of L-Parotid gland showed linear decrease with time (Figures 1 and 2). For each L-Parotid gland variable, comparing the mean among periods, ANOVA showed significantly different L-Parotid gland D mean (F = 4.49, p = 0.006), L-Parotid gland volume (F = 84.13, p < 0.001) and distance between plan isocenter and COM of L-Parotid gland (F = 28.90, p < 0.001) among the periods (Table 7).
Further, for each L-Parotid gland variable, comparing the difference in mean between periods (Table 8), Newman-Keuls test showed significant (p < 0.05 or p < 0.01) increase in L-Parotid gland D mean at CT4 as compared to CT1, CT2 and CT3 whereas it was found to be statistically the same (p > 0.05) between CT1, CT2 and CT3. In contrast, both L-Parotid gland volume and distance between plan isocenter and COM of L-Parotid gland showed significant (p < 0.05 or p < 0.001) decrease at CT2, CT3 and CT4 as compared to CT1. Both variables showed significant (p < 0.01 or p < 0.001) decrease at both CT3 and CT4 as compared to CT2. Further, both variables also showed significant (p < 0.01) decrease at CT4 as compared to CT3.
At final evaluation the L-Parotid gland shrank in volume by 30.1% and shifted medially by 7.5% from CT1 to CT4 with a net mean increase in D mean of 7.8%.

| The effect of treatment on variables related to combined volume of both parotid glands of the patient [BOTH-Parotid glands D mean (Gy), BOTH-Parotid glands volume (cc) and Distance between plan isocenter and COM of BOTH-Parotid glands (cm)] is summarised in table 7
The mean BOTH-Parotid glands D mean showed linear increase with time whereas BOTH-Parotid glands volume and distance between plan isocenter and COM of BOTH-Parotid glands showed linear decrease with time.
For each, BOTH-parotid glands related variable, comparing the mean among periods, ANOVA showed significantly different BOTH-Parotid glands volume (F = 107.83, p < 0.001) and Distance between plan isocenter and COM of BOTH-Parotid glands (F = 3.82, p < 0.05) among the periods (Table 7). However, BOTH-Parotid glands D mean showed insignificant change among the periods (F = 2.40, p =0.073).
Further, for each, BOTH-parotid glands related variables, comparing the difference in mean between periods (Table 8), Newman-Keuls test showed significant (p < 0.05 or p < 0.001) decrease in BOTH-Parotid glands volume and distance between plan isocenter and COM of BOTH-Parotid glands at CT2, CT3 and CT4 as compared to CT1. Furthermore, BOTH-Parotid glands volume also showed significant (p < 0001) decrease at both CT3 and CT4 as compared to CT2. Moreover, it also showed significant (p < 0.01) decrease at CT4 as compared to CT3.
At final evaluation, BOTH-Parotid glands D mean showed net mean increase (i.e. mean change from CT1 to CT4) of 5.4% whereas BOTH-Parotid glands volume and Distance between plan isocenter and COM of BOTH-Parotid glands showed net mean decrease of 27.5% and 23.8% respectively.

| Correlation
The correlation of change in both weight and neck circumference with change in parotid gland (D mean , volume and distance) of patients over the periods (CT1+CT2+CT3+CT4, n = 120) is summarised in Table 9. The Pearson correlation analysis showed a significant and positive (direct) correlation between change in neck circumference and change in weight of patients (r = 0.70, p < 0.001) ( Table 9 and Figure 3). Further, change in H-Parotid gland volume (r = 0.51, p < 0.001), Distance between plan isocenter and COM of H-Parotid gland (r = 0.18, p < 0.05), L-Parotid gland volume (r = 0.64, p < 0.001) and BOTH-Parotid glands volume (r = 0.64, p < 0.001) showed a significant and positive correlation with change in weight (Table 9 and Figure 4A-D). In contrast, change in H-Parotid gland volume (r = 0.50, p < 0.001), Distance between plan isocenter and COM of H-Parotid gland (r = 0.25, p < 0.01), L-Parotid gland volume (r=0.64, p < 0.001), BOTH-Parotid glands volume (r = 0.61, p < 0.001) and Distance between plan isocenter and COM of BOTH-Parotid glands (r = 0.18, p < 0.05) showed a significant and positive correlation whereas L-Parotid gland D mean (r = −0.28, p < 0.01) showed a significant and negative (inverse) correlation with change in neck circumference (Table 9 and Figure 5A-F).

| DISCUSSION
IMRT in the HNC was specifically introduced to minimize irradiation of the parotid glands and to improve the patient's quality of life after radiotherapy. 5 In the present study, the patients experienced a significant decrease in weight and neck circumference after having received twenty fractions of radiotherapy. Moreover, decrease in neck circumference was significantly associated with decrease in weight. We found a significant correlation between decrease in patient's weight with decrease in volume of both the parotid glands as well as medial shift of the parotid gland which received higher mean dose at initial planning. Decrease in neck circumference correlated well with decrease in volume of both parotid glands and their medial shift as well as increase in mean dose to the parotid gland which received lower mean dose at initial planning. The reduction of the head thickness leads consequently to the occurrence of dose hotspot in the neck, close or within the parotid glands as observed by Castelli et al. 6 You et al found that patients with significant reduction of the neck diameter and/or weight loss showed significantly frequent grade 2 acute xerostomia. 7 9 patients were seen to have a decline in ECOG status, mostly after the fourth week of treatment. Decline in performance status was also noted in a study by Lohia et al. 8 This could be associated with IMRT related fatigue and other treatment related toxicities. We observed a decrease in the volumes of GTV P and GTV N by 65.5% and 78.2% respectively. Similarly, Barker et al reported a median total relative loss of 69.5% of the initial GTV on the last day of treatment. 2 The amount of normal mucosa around the gross tumour volume that needs to be included in the clinical target volume is unclear, but even in the IMRT era most primary-tumour failures typically occur in the gross tumour volume and not in the surrounding mucosal area. 9 Dosimetric coverage of the target volumes tends to be robust during radiotherapy. The current study found no difference in GTV P D98% from start to end of treatment, while there was a slight but significant increase in GTV P D2%, GTV N D98% and GTV N D2%. Wu et al. reported  no change in the delivered dose to the primary CTV, with small a small increase in the minimum dose delivered to the nodal CTV, likely caused by the larger volume and anatomic changes experienced by the nodal CTV. 10 Similarly, Nishi et al also reported a slight increase in dose to the primary GTV in their study of 20 patients who underwent a repeat CT scan partway through treatment. They reported no changes in the minimum delivered dose to the nodal GTV. 4 Castadot et al who also investigated the impact of anatomic changes on target coverage reported that the dose to the primary and nodal CTVs remained unchanged as a result of anatomic changes throughout radiotherapy. 11 This study showed that the parotid glands decreased in volume by about 30% by end of treatment. Likewise, Bhide et al and Ho et al reported a contraction of the parotid gland volumes by 35% and 25% respectively through the course of treatment. 12,13 The medial shift of parotid glands on either side and the linear increase in their mean dose with time as observed in our study, correlated well with other published literature. 2,4,6,[10][11][12]14,15 The anatomic changes observed over time, as quantified in this study, are particularly important, because the parotid glands move medially towards the highdose region (Figure 1 and 2). This implies that most of the radiation dose was delivered to a deviated anatomy compared with the original treatment plan.
Despite advancements in the RT technique, acute toxicities continue to be a major challenge in HNC radiotherapy. Mucositis and xerostomia were the most common acute toxicities seen in our patients. The strength of this study is that we have taken multiple CT images of the same patient during treatment period in the treatment position and compared these with the initial simulation images with respect to the anatomic changes of bilateral parotid glands and the primary as well as nodal tumor volumes to assess the dosimetric changes on the same. We have also correlated these changes with change in patient's weight loss and changes in neck circumference. We have also monitored the acute treatment related toxicities. Limitation of this study is that in view of limited resources, patients had not undergone midcourse  replanning to compensate for the anatomical changes that they underwent, it may have resulted in optimum dose distributions and reduced long term toxicities for some patients.

| CONCLUSION
With temporally changing anatomy of both tumour and normal tissue, delivery of radiotherapy should be temporally changing to match the observed anatomic changes where possible, however, needs further validation on larger population.

AUTHOR CONTRIBUTIONS
Arunima Ghosh: Collected Data, Contributed in paper writing Seema Gupta: Formulated and designed the manuscript and analysis, Collected Data, Contributed in data and data analysis tools, Contributed in paper writing Danial Johny: Contributed in data and data analysis tools Vivek Vidyadhar Bhosale: Contributed in paper writing Mahendra Pal Singh Negi: Contributed in data and data analysis tools, Performed data analysis.

MESSAGE OF THE MANUSCRIPT
If the planning target volume and normal tissue anatomy are changing with time during IMRT, adaptive IMRT would be beneficial radiation dose delivery where possible to minimize normal tissue toxicity without influencing therapeutic outcome.

CONFLICT OF INTEREST
There is none.

DATA AVAILABILITY STATEMENT
Data available on request from the authors.