Clinical outcome after pencil beam scanning proton therapy and dysphagia/xerostomia NTCP calculations of proton and photon radiotherapy delivered to patients with cancer of the major salivary glands

Abstract

Objectives

The purpose of this study is to report the oncological outcome, observed toxicities and normal tissue complication probability (NTCP) calculation for pencil beam scanning (PBS) PT delivered to salivary gland tumour (SGT) patients.

Methods

We retrospectively reviewed 26 SGT patients treated with PBSPT (median dose, 67.5 Gy(RBE)) between 2005 and 2020 at our institute. Toxicities were recorded according to CTCAEv.4.1. Overall survival (OS), local control (LC), locoregional control (LRC) and distant control (DC) were estimated. For all patients, a photon plan was re-calculated in order to assess the photon/proton NTCP.

Results

With a median follow-up time of 46 months (range, 3–118), 5 (19%), 2 (8%), 3 (12%) and 2 (8%) patients presented after PT with distant, local, locoregional failures and death, respectively. The estimated 4 year OS, LC, LCR and DC were 90%, 90%, 87 and 77%, respectively. Grade 3 late toxicity was observed in 2 (8%) patients. The estimated 4 year late high-grade (≥3) toxicity-free survival was 78.4%. The calculated mean difference of NTCP-values after PBSPT and VMAT plans for developing Grade 2 or 3 xerostomia were 3.8 and 2.9%, respectively. For Grade 2–3 dysphagia, the grade corresponding percentages were 8.6 and 1.9%. Not using an up-front model-based approach to select patients for PT, only 40% of our patients met the Dutch eligibility criteria.

Conclusion

Our data suggest excellent oncological outcome and low late toxicity rates for patients with SGT treated with PBSPT. NTCP calculation showed a substantial risk reduction for Grade 2 or 3 xerostomia and dysphagia in some SGT patients, while for others, no clear benefit was seen with protons, suggesting that comparative planning should be performed routinely for these patients.

Advances in knowledge

We have reported that the clinical outcome of SGT patients treated with PT and compared IMPT to VMAT for the treatment of salivary gland tumour and have observed that protons delivered significantly less dose to organs at risks and were associated with less NTCP for xerostomia and dysphagia. Noteworthy, not using an up-front model-based approach, only 40% of our patients met the Dutch eligibility criteria.

Introduction

Major salivary gland tumours (SGTs) are a histologically heterogenous group of tumours with an annual incidence of approximately 1 per 100,000 in Europe and the US. ^1,2 The 5-year overall survival of initially non-metastasized disease is 83–93%. ³ In most cases, surgical resection is the treatment of choice. Adjuvant radiotherapy is recommended in patients with locally advanced tumours (T3-T4) and high-grade tumours, as well as close or incomplete resection margins (R1, R2), vascular (V1) or perineural (Pn1) invasion and lymph node metastases (N+). In the case of inoperable tumours or surgery being refused, primary radio(chemo-)therapy should be considered. ^4,5 Irradiation doses from 60 to 74 Gy are needed for effective locoregional tumour control. However, due to close anatomical relationship to several delicate organs at risk, the delivery of these doses can cause significant acute and long-term toxicities including xerostomia, dysgeusia, trismus, swallowing dysfunction and rarely necrosis in the temporal lobe. ^6,7

The dose distribution of proton irradiation with its unique physical qualities allows for a minimisation of dose delivered to normal tissue. Especially the finite beam range and thus the avoidance of exit dose is a great advantage of protons. This has additional benefits especially in cases with unilateral neck involvement, as is usually the case in SGTs. In regard to the substantially increased long-term survival, ³ proton therapy may be superior to photon irradiation considering adverse events and the risk of secondary malignancy induction.

Comparisons of photon versus proton therapy in head and neck cancers support this thesis. ^8–10 For major salivary gland cancers, Grant et al ¹¹ have shown a significant benefit of proton therapy over conventional radiotherapy in acute toxicities and dosimetric characteristics in children. Chuong et al ¹² have found substantially lower acute toxicities compared to reported intensity-modulated radiation therapy (IMRT) outcomes. However, there is little to no data available for long-term toxicities and local control rates of proton irradiation of major salivary gland cancers.

In this retrospective study, we compare the acute and long-term toxicities of proton beam therapy, as well as the disease control rates to historical data of conventional radiotherapy. Moreover, we use the NTCP models developed by Langendijk et al. ^13–16 to compare dosimetric profiles of proton and photon plans as well as the predicted toxicities.

Material and methods

Patient cohort

We retrospectively reviewed all patients who had received PBSPT for primary cancers of the major salivary glands between 2005 and 2020 at our institution. We identified 29 cases with malignant tumours either in the parotid, submandibular or sublingual glands with at least 3 months of follow-up. For our analysis, we excluded two patients who received a combination of photon irradiation and PBSPT and one patient who underwent re-irradiation for a local recurrence. Of the remaining 26 patients, 25 (96%) had undergone tumour resection and were referred for adjuvant PBSPT due to high risk factors (R1/2 type resection, Pn1, N+), and another unresectable patient received PBSPT only. Of note, one (4%) patient had a single bone metastasis at time of first diagnosis. Ethics Committee approval (EKNZ 2020–02077) was obtained for this study.

Proton therapy

PT was delivered using PBS with an energy-degraded beam from a 250 MeV cyclotron for all patients. Treatment planning was performed either on the in-house planning system PSIplan or Eclipse^® (Varian Medical Systems, Palo Alto, CA, USA). Multi-field optimisation (MFO) and single-field optimisation (SFO) techniques were used, usually applying 2 to 3 ipsilateral fields to the target volumes. In post-surgical patients, the initial gross tumour was delineated in pre-surgery images (CT, MRI) that were fused with the PT planning CT. In the patients with persisting gross tumour after surgery, and undergoing definitive PT, gross tumour volume (GTV) was defined by the tumour visualised on the planning CT and MRI. Based on individualised risk assessment, two or three clinical target volumes (CTV) were used. The boost CTV covered the GTV or the tumour bed with an additional margin of 5 mm. Boost dose was dependent from resection status as well as from histology. CTV_60 Gy(RBE) was further extended with a margin of 5–10 mm and additionally included the levels of positive lymph nodes from neck dissection and perineural spread pathways. A CTV_54 Gy(RBE) included elective lymph node areas in the neck (level Ib- V) in patients with adjuvant or prophylactic neck irradiation and/or treatment along the perineural pathways to the skull base. Inclusion of the neck lymph nodes was dependent on post-operative N-status (all N + patients received adjuvant neck irradiation) and the estimated risk for lymph node involvement was dependent on the tumour histology, grading and T-status. The planning target volumes (PTV) were defined as the CTV plus a 5-mm safety margin. For the proton treatment planning, a constant relative biological effectiveness (RBE) value of 1.1 was used.

Comparative planning and dosimetric evaluation

For the 26 patients included in this analysis, the comparative planning was performed in Eclipse TPS transferring all planning CTs and treatment plans. For equal comparability, the parametrics for the PTV coverage was set to a minimum of 98% coverage of 95% of the prescribed total dose (98/95) for each plan, according to the study of van der Laan et al. ¹⁰ Photon plans were calculated according to this parameter with volumetric arc therapy (VMAT) using 2 (n = 23; 88.5%) to 3 (n = 3; 11.5%) arcs and 6 MV photon beam data of a Varian True Beam LINAC (Varian Medical System; Palo Alto, CA). PT plans not fulfilling the demanded criteria were modified and optimised to meet these parameters for comparability reasons. OARs were checked and adapted if necessary by three experienced radiation oncologists. The OARs evaluated for each patient included bilateral parotid and bilateral submandibular glands, oral cavity, pharyngeal constrictor muscles (PCM), spinal cord, larynx and thyroid glands. Prioritisation of OARs regarding dose reduction optimisation was set according to above order of OARs for maximal risk reduction for the development of xerostomia and dysphagia. The delineated OAR volumes were adapted wherever necessary to be consistent according to the contouring guidelines of Brouwer et al. ¹⁷ and Christianen et al. ¹⁸ The planners could waive the dose-constraints for planning purposes, with the exception of the spinal cord constrained at D2% < 54 Gy RBE. Mean and maximum doses were evaluated for each structure. Additionally, for calculating the integral dose for photon- and PBSPT-plans, the body contour was adapted to cover the volume 2 cm above and below the PTV.

NTCP-model calculation

NTCPs for National Cancer Institute’s Common Terminology Criteria for Adverse Events (CTCAE) Grades 2 and 3 xerostomia and dysphagia were calculated for each proton and photon plan. The employed models were developed and validated at the University Medical Center Groningen (UMCG) and form the basis of the Dutch model-based approach (MBA) ¹⁹ for head and neck cancer which allows for systematic patient stratification for proton or photon treatments. Model parameters for xerostomia were associated with mean doses to the parotid and submandibular glands as well as to the degree of baseline xerostomia. Models for dysphagia integrate the degree of baseline dysphagia, mean doses to the oral cavity and PCM and the location of the primary tumour. In case of surgical removal of a salivary gland prior to radiotherapy, only mean doses to the remaining glands were considered in the model application. Eligibility criteria for proton therapy based on differences of NTCP calculation between photon and proton plans were directly taken from the Dutch MBA, which are: >10% probability difference for Grade 2 (xerostomia or dysphagia), > 5% probability difference for Grade 3 (xerostomia or dysphagia), > 15% probability difference for Grade 2 xerostomia+Grade 2 dysphagia and >7,5% probability difference for Grade 3 xerostomia+Grade 3 dysphagia.

Follow-up

Acute toxicity was recorded weekly during PT. All subsequent institutional and outside-facility clinical notes were collected by our study and research office and reviewed during our weekly follow-up meeting to identify disease status, toxicity and survival data. Local failure (LF) was defined as the development of residual tumour progression or new nodular contrast enhancement and/or FDG-uptake in the surgical bed compared with the baseline images. Distant failure (DF) was defined as the development of new distant lesions in either MRI-, CT- or PET/CT follow-up or histologically proven by biopsy or resection. Toxicity scores were retrospectively assigned according to the CTCAE v4.1.

Statistical analysis

Time to event data were calculated from the last day of PT to the date of event, death or censored at last follow-up using the Kaplan-Meier method. Death from any cause, LF and DF were the defined events for the calculation of overall survival (OS), local control (LC) and distant control (DC), respectively. A paired t-test analysis was performed for the pairwise comparison of OAR dose parameter differences between PBSPT and VMAT treatment plans. A p-value ≤ 0.05 was considered statistically significant. All statistical analyses were computed using SPSS version 26 (IBM, USA).

Results

Patient characteristics

The patients (n = 26) tumour and treatment characteristics are shown in Table 1. Median age was 42.6 years (range, 11.5–74.4) with 17 (59%) female and 12 (41%) male patients. Two patients were younger than 18 years at the time of PT.

Table 1.

Patient and Tumour and Treatment Characteristics (n = 26)

Age (years)
Median	42,6
Range	11.5–74.4
Gender	n (%)
Female	16 (62)
Male	10 (38)
Primary tumour gland	n (%)
Parotid gland	20 (77)
Submandibular gland	5 (19)
Sublingual gland	1 (4)
Histology	n (%)
Adenoid cystic carcinoma	13 (50)
Acinic cell carcinoma	6 (23)
Mucoepidermoid carcinoma	3 (11)
Salivary duct carcinoma	2 (8)
Adenocarcinoma	1 (4)
Myoepithelial carcinoma	1 (4)
Tumour resection before PT	n (%)
yes	25 (96)
no	1 (4)
Resection status	n (%)
R0	7 (28)
R1	14 (56)
R2	4 (16)
Clinical T stage	n (%)
T1	3 (12)
T2	8 (31)
T3	8 (31)
T4	7 (26)
Clinical N stage	n (%)
N0	20 (78)
N1	3 (11)
N2	3 (11)
Clinical M stage	n (%)
M0	25 (96
M1	1 (4)
Total treatment dose Gy(RBE)
Median	67,5
Range	60.0–75.6
Number of total fractions
Median	33
Range	30–42
Proton therapy concept	n (%)
Sequential	17 (65)
Simultaneous integrated boost	9 (35)
Proton therapy combined with chemotherapy	n (%)
Yes	2 (8)
No	24 (92)

Tumour characteristics

The majority of tumours (n = 20; 77%) involved the parotid gland (Table 1). The submandibular and sublingual glands were involved in five (19%) and one patient (4%), respectively. Adenoid cystic carcinoma (ACC) was the most common tumour histology (n = 13, 50%), followed by acinic cell carcinoma (n = 6, 23%) and mucoepidermoid carcinoma (n = 3, 11.5%; Table 1). Of the 25 patients who underwent surgical primary tumour resection, the majority (n = 17, 68%) also underwent an ipsilateral neck dissection. In six (35%) of them, histopathological workup revealed lymph node metastases. Resection status in the operated patients was R0 (n = 7, 28%), R1 (n = 14, 56%) and R2 (n = 4, 16%), respectively.

Treatment characteristics

The median RT dose to the primary tumour region was 67.5 Gy (RBE) in 30 fractions (range, 60.0–75.6). Eleven patients (42%) received additional ipsilateral neck irradiation, six (23%) as adjuvant irradiation for positive lymph nodes and five (19%) as prophylactic irradiation. In 16 (62%) patients, PBSPT was applied as sequential treatment, nine (35%) patients received a simultaneous integrated boost (SIB) and one (3%) patient had a single series treatment. Two (8%) patients received concomitant chemotherapy to PT.

Outcome

After a median follow-up time of 46 months (range, 3–118), 2 (8%) local failures (one in-field, one marginal) 12 and 27 months (median, 20) and 1 (4%) regional failure (lymph node metastasis) 8 months after PBSPT were observed. All failures were biopsied and were thus histologically proven. The estimated 4 year local and locoregional control rate was 90.3% (95%CI:77.6–100%) and 86.7% (95%CI:72.6–100%), respectively. Distant failures were observed in five (19%) patients, including the oligo-metastatic patient at diagnosis. Both patients with local failure also developed metastatic disease. The median time to distant failure was 14 months (range, 2–27). The estimated 4-year distant control rate was 77.3% (95% CI: 59.6–94.8%; Figure 1). Two (8%) patients died, 22 and 32 months (median, 27) after PBSPT, both of uncontrolled disease. The estimated 4 year OS was 90.2% (95%CI:77.3–100%; Figure 1).

Figure 1.

Local tumour (a) control and overall survival (b) in 26 patients with major salivary gland cancers treated with pencil beam proton beam therapy

Toxicity

The treatment-related acute toxicity was recorded from treatment beginning until 90 days after PT. Acute Grade 2 toxicity was observed in 16 (61.4%) patients mainly as dermatitis and mucositis. Acute Grade 2 xerostomia and dysphagia were observed in 2 (8%) and 3 (12%) patients, respectively. Five (19.2%) patients developed Grade 3 acute toxicity as dermatitis, mucositis and otitis media. Late Grade 2 toxicities were observed in 5 (20%) patients with xerostomia in 1 (4%) patient. Late Grade 3 toxicity was observed in 2 (8%) patients with persisting wound dehiscence in one patient (40 months after the end of PT) and persisting otitis media with mastoiditis in another patient (11 months after the end of PT). There were no Grade 4 or 5 toxicities. Acute or late Grade 2 xerostomia and dysphagia were both captured in three (12%) patients each. There were no acute or late Grade 3 xerostomia or dysphagia observed. The radiation-induced toxicities are detailed in Table 2. Noteworthy, the rate of observed Grade 2 dysphagia and xerostomia in three (11.5%) patients in our cohort was lower than the estimated mean risk in the NTCP model for these toxicities with 23,4% for xerostomia and 16,8% for dysphagia, respectively. In addition, for Grade 3 dysphagia and xerostomia, the models predicted risks of 6,4% and 4,7%, whereas in no patients in our cohort have been observed with these toxicities. The estimated 4-year late high-grade (≥ 3) toxicity-free survival was 78.4% (95% CI: 58.6–98.2%).

Table 2.

Acute and late radiation-induced toxicity observed in 26 patients with SGTs treated with PT

Highest toxicity	Acute			Late
Grade	0/1	2	3	0/1	2	3
	n (%)	n (%)	n (%)	n (%)	n (%)	n (%)
Xerostomia	24 (92)	2 (8)	0	25 (96)	1 (4)	0
Dysphagia	23 (88)	3 (12)	0	26 (100)	0	0
Mucositis	17 (80)	7 (12)	2 (8)	26 (100)	0	0
Dermatitis	10 (38)	14 (54)	2 (8)	26 (100)	0	0
Otitis	23 (88)	2 (8)	1 (4)	25 (96)	0	1 (4)
Fibrosis	26 (100)	0	0	25 (96)	1 (4)	0
Trism	26 (100)	0	0	25 (96)	1 (4)	0
Wound dehiscence	26 (100)	0	0	25 (96)	0	1 (4)
Alopecia	26 (100)	0	0	24 (92)	2 (8)	0

PT: proton therapy;SGT: major salivary gland tumour.

Comparison of dose to OARs using PBS or VMAT

A representative comparison of dose distribution between PBSPT and VMAT is shown in Figure 2. Of note, no significant difference was observed for mean doses to the ipsilateral salivary glands, the ipsilateral parotid (60.7 (range, 9.6–72.8) vs 61.0 Gy(RBE) (range, 9.2–74.1), p = 0.529) or the ipsilateral submandibular gland (50.6 (range, 2.1–75.3) vs 50.7 Gy(RBE) (range, 0.2–76.4), p = 0.855). The integral dose was reduced on average by about 40% in the PBSPT plans compared to VMAT. Also, for all other OARs, we found significant lower mean doses in the PBSPT plans as can be seen in Table 3.

Figure 2.

NTCP analysis for xerostomia Grade 2/3 and dysphagia Grade 2/3. Each dot respresents a patient, the red dots represent patients meeting the Dutch eligibility criteria for PT. Ph = photon, Pr = proton

Table 3.

Organs at risk mean dose differences for VMAT vs PBSPT

mean dose
Organ	Photon Gy[RBE]	Proton Gy(RBE)	δ (Gy/Gy)RBE)	δ (%)	p-value
Parotid gland (ipsi)	60,7	61	−0,3	0,5	0,529
Parotid gland (contra)	10,6	0	10,6	−100	<0.001
Oral cavity	29,5	13,5	16	−56	<0.001
Submandibular gland (ipsi)	50,6	50,7	−0,1	0,2	0,855
Submandibular gland (contra)	18,5	1,1	17,4	−94	<0.001
PCM superior	41,1	31,3	9,8	−23,8	<0.001
PCM middle	35,7	26,4	9,3	−26,1	<0.001
PCM inferior	22,1	14,1	8	−36,2	<0.001
Spinal cord	20,8	4,2	16,6	−79,8	<0.001
Larynx	18,7	12,7	6	−32,1	<0.001
Integral dose	17,8	10,4	7,4	−41,6	<0.001

NTCP-calculation comparison results

The calculated mean risks after PBSPT and VMAT plans in our cohort for developing Grade 2 xerostomia were (23.4 (range, 19.6–28.7) vs 27.2% (range, 21.9–40.0); ΔNTCP 3.8%), for Grade 3 xerostomia (4.7 (range, 2.7–7.5) vs 7.6% (range, 3.8–12.6); ΔNTCP 2.9%), for Grade 2 dysphagia (16.8 (range, 4.1–55.4) vs 25.4% (range, 9.6–63.6); ΔNTCP 8.6%) and for Grade 3 dysphagia (6.4 (range, 0.5–14.0) vs 8.3% (range, 2.7–16.4); ΔNTCP 1.9%). The Dutch eligibility criteria for PT were met in 11 (42.3%) patients. One patient reached four out of six eligibility criteria, and nine patients (34%) could meet two and one patient (4%) met one eligibility criterion (Figure 3). The Box plots of δ NTCPs for Xerostomia and Dysphagia grade II/III for 26 SGT patients planned with VMAT and PT are detailed in the Supplemental Figure.

Figure 3.

NTCP analysis for xerostomia grade 2/3 and dysphagia grade 2/3. Each dot respresents a patient, the red dots represent patients meeting the Dutch eligibility criteria for PT. Ph = photon, Pr = proton

Discussion

In our study, we have observed high LC and OS rates after PBSPT for major SGTs and limited high-grade toxicities. Furthermore, we show a significant reduction of the integral dose to the normal tissues of patients (Table 3) with relevant reduction of dose burden to contralateral OARs that could potentially impact their quality of life. Using the Dutch model for head and neck tumours, ²⁰ we were able to demonstrate for some SGT patients a substantial reduction of toxicity probability for Grade 2 dysphagia after PBSPT (Figure 2).

We report a 4-year LC, OS and DC rate of 91%, 89% and 77%, respectively. These results are comparable with recently published data from photon and proton studies. ^21–26 No special selection criteria were used in selecting patients for PT in our cohort; hence, indications for post-operative PT corresponded to the general criteria depending on risk factors. ²⁷ Nevertheless, we have an overrepresentation of ACCs in our cohort (Table 1). In photon series, the percentage of ACCs over all histologies is reported in the range of 19–28%. ^24,28 The fact that 50% of our patients presented with ACCs expresses the awareness of the referring physicians, radiation oncologists and ENT surgeons, that proton therapy might be beneficial in this challenging tumour entity. From different publications, it is known that a perineural tumour invasion (PTI) is detected histopathologically and/or radiologically in at least 50% of the investigated ACCs. ^29,30 PTI of the affected nerves (facial and trigerminal nerves) makes it necessary to expand the treatment volume along these nerves to the skull base and therefore the risk for high grade toxicity induction is increased. PTI has been identified as a risk factor for impaired outcome. ³¹ We have reported recently our own results regarding PBSPT of head and neck ACCs with high LC and low toxicity rates. ³² Nevertheless, distant metastases showed to be the main pattern in that study. The rather high rate of 23% of patients with distant metastases in our SGT cohort is also an expression of the before discussed histological distribution.

The good treatment outcome in patients with SGTs in general makes it important to deliver the most organ-sparing radiation therapy modality possible, minimising the dose to healthy tissue. The comparison of the mean doses in involved OARs revealed significantly less radiation to the majority of organs with PBPST, when compared to VMAT (Table 3). Of note, exceptions of this dose-sparing were ipsilateral OARs that often are at least partially involved in the target volume, where we have not observed significant differences in the mean doses delivered to those with protons or photons (Table 3). As demonstrated by other comparative planning studies, it is not the high dose regions that profit from the physical advantage of protons, but rather a reduction of the mid-to-low dose bath. Treatments that require ipsilateral irradiation seem to benefit significantly more by the use of protons. This is confirmed by other studies comparing proton- and photon-based planning which analysed the dose to involved OARs in SGT patients or ipsilateral-treated patients with H&N tumours. ^8,11,33 All studies showed a significant reduction of integral dose, as well as mean doses in the contralateral OARs. Not only does this result in the reduction of potential side effects but also, especially in children, adolescent and young adults, in the reduction of radiation induced secondary malignancies. ^34,35

Most higher grade acute toxicities in our cohort were mucositis (7,7%) or dermatitis (7,7%). The toxicity of radiation dermatitis is not surprising due to the superficial location of the target volumes, often extending to the skin in these patients. Similar findings were reported from a multi-institutional registry REG001-09 trial (NCT01255748) that investigated the acute toxicity from PT for SGTs in 105 patients. ¹² The authors reported a Grade 3 dermatitis rate of 10,5%. Memorial Sloan Kettering Cancer Center published comparable results in their study irradiating 68 SGT patients with PT with an acute Grade 3 dermatitis rate of 13,2%. ²⁴ The overall rate of Grade 3 toxicity observed in another study from MD Anderson Cancer Center treating 72 patients with major salivary gland cancers was 21%. ²⁶

As mentioned, mucositis is another critical side effect occurring in these patients treated with radiation therapy due to involvement of oral and pharyngeal mucosa. Together with the affection of the major salivary gland, RT impairs the secretory potential, causing xerostomia, depending on dose and volume. ³⁶ Additionally, the irradiation of the PCM can severely impact the swallowing function resulting in dysphagia. ¹⁴ It was shown that these side effects are crucial concerning the quality of life of patients with oropharyngeal cancers. ²⁰ To focus more deeply on these two notable and QoL-impairing side effects, we did an extended analysis using the NTCP model for the risk of xerostomia and dysphagia induction developed by the Dutch group. Generating additional comparison plans with VMAT allows for a quantitative risk calculation of these toxicities for the two treatment modalities, hence selecting patients who profit most from proton therapy. ³⁷ Based on this NTCP model analysis, 42% of our unselected cohort would benefit from proton therapy and did meet indeed retrospectively the Dutch eligibility criteria. In a study using the same model investigating 50 patients with head and neck tumours of different localisations, 32% of their patients met the Dutch criteria. ³⁸ Interestingly, their analysis showed a significant benefit of patients with lateralised tumours. This fact could explain our higher rate of patients benefitting from PT.

With the used model-based approach, additional benefits from decreased integral dose in the involved OARs and tissues are not respected in the selection of treatment modality. As shown in the study of Romesser et al ⁸ , additional acute side effects as mucositis, nausea and dysgeusia can be reduced when using PBRT instead of IMRT. The same group showed in an earlier study treating patients with oropharyngeal cancer with IMRT, that it can result in increased rates of malnutrition, treatment breaks, prolonged recovery and consequently lead to increased occurrence of worse late toxicity. ³⁹ These acute side effects potentially can lead to treatment interruptions resulting in decreased outcome and additional need for medical interventions.

Interestingly, the rate of observed Grade 2/3 dysphagia and xerostomia in our cohort was lower than the estimated mean risk in the NTCP-model for these toxicities using the model-based approach. This higher predicted/observed toxicity ratio was also observed in the Dutch group and this group (Hans Langendijk, personal communication) will soon publish this observation.

One explanation for these findings could be the fact that the model-based data used were generated from IMRT patients. The Groningen group have recently reported that a lower rate of proton radiation-induced toxicity was observed than predicted. Possibly additional factors as dose-volume-toxicity relationships and delivery paradigms (discrete vs continuous radiation delivery) might differ between photons and protons and therefore may play an additional role which is not fully captured in the model yet. As suggested by the Dutch group, the addition of H&N patients treated with protons will further optimise the model for this treatment modality. Blanchard et al investigated their results of H&N cancer patients treated with protons comparing them with different photon-derived NTCP models. ⁴⁰ The authors found a drop in model performance with around a 10% decrease in the area under the curve value. Another explanation for the clinical toxicity results of our cohort compared with the NTCP model results is of course the limitations of our study. These are mainly the retrospective nature of data collection with its uncertainties and the small number of patients. The latter limited the statistical power for the comparison with the model-based results. The combination of clinical outcome in SGT patients treated with PBSPT and the additional analysis of the benefit of this radiation modality using the NTCP model-based approach in comparison with VMAT bear however useful results for decision making for the treatment of these patients. In principle, these data suggest that not all SGT patients may benefit from protons and comparative planning could be recommended prior to treatment delivery, although counterarguments of not delivering protons to these usually young patients with a favourable prognosis could be the integral dose reduction achievable with protons when compared to photons techniques.

There were several limitations of our study assessing patient’s clinical outcome. First, the study design was retrospective in nature and thus lacked complete data for certain variables such as objective measurements of salivary flow and swallowing function. Second, the small sample size of 26 patients limited the outcome implication for the general cognoscenti and the clinical relevance of the dose- and NTCP-comparison between VMAT and PT. This being said, patient follow-up times were substantial and the outcome was reported after pencil beam only proton therapy which is the new standard for this treatment modality and has rarely been reported for this tumour entity.

Conclusion

This study demonstrates excellent clinical outcome for SGT patients treated with PBSPT. Significantly lower contralateral OAR doses were observed with PBSPT plans compared to VMAT plans. NTCP retro-calculation identified more than 40% of treated patients qualifying for PT with a substantial risk reduction of developing grade 2 xerostomia or dysphagia. Comparative planning for these SGT patients is recommended.