Adaptive radiotherapy in head and neck cancer is required to avoid tumor underdose

Introduction

During the course of intensity-modulated radiotherapy (IMRT) for head and neck cancer (HNC), the delivered dose may differ from the planned dose due to significant anatomical variations. Several studies have particularly reported parotid gland (PG) overdose [1–3]. On the other hand, clinical target volume (CTV) coverage is usually thought to be unaffected due to the planning target volume (PTV) margin combined with tumor shrinkage. Studies investigating differences between planned and delivered doses to the target volume have, however, produced controversial results [2–14]. Adaptive radiotherapy (ART), involving one or several replannings, has therefore so far been primarily focused on decreasing the dose to the PGs [1,3,6,14–16], albeit with concerns about the risk of tumor underdose.

In a context of definitive RT for locally-advanced oropharynx carcinoma, this dosimetric study focused on both the CTV and PG, aiming to quantify: (i) the dose differences between the planning and delivered IMRT without ART (non-ART), particularly CTV underdose and PG overdose; (ii) the benefit of weekly systematic replanning (ART) in terms of correcting these dose differences.

Material and methods

Thirty-seven patients with locally-advanced oropharyngeal cancer were prospectively included in this study. All tumors corresponded to T3-4 or N2-3 Stage (Stage III or IV, AJCC 7th edition). All patients underwent IMRT at a total dose of 70 Gy (2 Gy/fraction/day, 35 fractions), with a simultaneously integrated boost technique [17] and concomitant chemotherapy (cetuximab or platinum).

Each patient underwent six weekly CTs using the same protocol as CT0 over the treatment course. Anatomical structures were propagated from the planning to the weekly CT using deformable image registration (MIM Software 6.6^©) and were manually corrected by the same radiation oncologist for each patient.

Two weekly dose distributions were calculated on each weekly CT dataset, corresponding to two situations, either IMRT without replanning or IMRT with replanning (ART). Weekly IMRT replanning was performed in accordance with the dose constraints described for the initial planning. The cumulated dose, with or without replanning, was considered using the mean of the weekly dose for each dosimetric endpoint. The dose objectives were considered to be reached when the D98 exceeded 66.5 Gy (95% of the prescribed dose) for the CTV₇₀ [18], and the mean dose for the contralateral PG was less than 30 Gy.

Results

The median D98_CTV decreased from 69.1 Gy at the planning to 68 Gy without ART in the whole population (p < .01). CTV underdose was exhibited in 76% of patients to a median value of 1.6 Gy. CTV underdose exceeding 1 Gy and 2 Gy was reported in 57% and 28%, respectively. Figure 1 presents the population distribution based on D98_CTV at the planning and without ART.

Figure 1. Population distribution of clinical target volume (CTV) D98: at the planning, with and without adaptive radiotherapy (ART). The cumulated dose (without or with ART) was estimated by considering the mean of the weekly doses. Compared to the planned dose (blue area), the cumulated dose to 98% of the CTV (D98_CTV) was decreased without ART (red area). Weekly ART (green area) enabled tumor underdose correction.

The mean contralateral PG dose slightly increased from a median of 27.8 Gy at the planning to 27.9 Gy without ART (p < .01). Contralateral PG overdose of a median of 2.3 Gy was reported in 67% of patients. In total, 95% of patients suffered either CTV underdose, contralateral PG overdose or both.

ART achieved a median D98_CTV increase from 68 Gy without ART to 69.2 Gy for the whole population (p < .01). Figure 1 shows the population distribution of D98_CTV at the planning with and without ART. ART achieved a median contralateral PG dose decrease from 27.9 Gy to 25.9 Gy (p < .01) (Figure 1).

Figure 2 shows the individual benefit of ART to correct CTV underdose and PG overdose compared to the dose without replanning. ART improved the percentage of patients achieving the dosimetric objectives for the CTV from 86% without ART to 97%, those for the PG from 67% to 83%, and those for both the CTV and PG from 60% without ART to 83% with ART.

Figure 2. Individual benefit of the adaptive radiotherapy (ART) strategy in terms of correcting clinical target volume (CTV) underdose and parotid gland (PG) overdose. The dose endpoints are the D98 for the CTV and the mean dose for the parotid gland. The delivered dose was estimated by considering the mean of the weekly doses. The dose variations were calculated as the delivered dose minus the planned dose. The green area corresponds to the dosimetric constraints for the CTV (D98 > 95% of prescribed dose) and parotid glands (mean dose <30 Gy). Without ART, dose constraints were achieved for 60% of patients, compared to 83% with ART.

Discussion

CTV dose variations between non-ART and planning, assessed by D95 or D99, appear to represent a controversial issue in the literature. Three studies have reported underdoses of 1.4 Gy [13], 2 Gy [10] and 3 Gy [12] for unspecified numbers of patients, while four others reported no difference [2,3,8,9] and three more reported overdoses of 0.5 Gy [4], 1 Gy [5] and 2 Gy [11] for unspecified numbers of patients. However, these studies suffered from several limitations, related to their limited numbers of patients (less than 30), unspecified tumor locations [2,6–9] and designs including both definitive and post-operative radiotherapy [14,19].

In all studies, ART was performed with the goal of correcting OAR overdose (mostly the PGs) rather than dose escalation in the CTV. This strategy enables lower mean PG doses (ranging from 0.6 Gy to 4.1 Gy) compared to non-ART treatments [1,2,7,14]. In our study, ART enabled mean PG dose decrease in 89% of the patients. Moreover, our dose objectives for the PG (mean dose <30 Gy) were achieved for 83%, compared to 66% when using IMRT without ART.

In both our study and the literature, an ART strategy designed to spare PGs appears safe in terms of CTV coverage [3,6,12,15,16]. We found that weekly replanning-based ART increased the CTV dose by 1.2 Gy for 67% of the patients. The percentage of patients who achieved the CTV coverage objectives also increased from 80% without ART to 97% (Figures 1 and 2). Only one study, involving 11 patients, has investigated the benefit of weekly replanning, achieving a 10% decrease in the mean PG dose. However, no difference was reported between the CTV doses with and without ART, suggesting that replanning may not be necessary for the CTV [2].

Our study has some limitations. Considering the mean of the weekly values of the dosimetric endpoints to estimate the cumulated doses is subject to discussion. This calculation assumes that dose differences between planning and fraction doses are always located in the same area during the course of treatment, which is not necessarily the case. On the other hand, using deformable image registration to cumulate the dose, as some studies do [2,3,9], carries its own risk due to the loss of tissue observed during the treatment for both tumors and PGs [20]. Finally, we used an online approach in our study simulation to ensure the maximum benefit of ART, while an offline approach could slightly reduce the benefit due to delays in generating replanning.

Conclusion

During IMRT without ART, a tumor underdose was observed for a majority (76%) of the patients, justifying the use of ART. In total, all patients benefited from ART: 89% in terms of reduced dose to the PGs, 67% for increased dose to the CTV, and 56% for both benefits, while all benefitted from at least one of these advantages. Clinical results are now required to confirm the benefit of ART in decreasing xerostomia and potentially increasing localized control.

Acknowledgments

This study was approved by the institutional review board (ARTIX study NCT01874587).

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability

This work was a part of an ongoing clinical study. Until the final analysis of this study, the data will not be shared.

Additional information

Funding

This work was funded in part by the French INCa within the PAIR VADS program and by the French ANR within the Investissement d’Avenir program (Labex CAMI) under reference ANR-11-LABX-0004.