Intensity Modulated Radiotherapy (IMRT) With Carbon Ion Boost in the Multimodal Treatment of Salivary Duct Carcinoma

Background: To assess outcomes and treatment related toxicity following intensity-modulated radiotherapy (IMRT) and a Carbon Ion Radiotherapy (CIRT) boost for salivary duct carcinoma (SDC). Methods: Twenty-eight consecutive patients with SDC who underwent a postoperative (82%) or definitive (18%) radiation therapy between 2010 and 2017 were assessed in this retrospective single-center analysis. CIRT boost was delivered with median 18 Gy(RBE) in 6 daily fractions, followed by an TomoTherapy®-based IMRT (median 54 Gy in 27 daily fractions). Treatment-related acute toxicity was assessed according to CTCAE Version 4. Results: Tumors were most commonly located in the major salivary glands (n = 25; 89%); 23 patients (82%) received previous surgery (R0: 30%; R1: 57%; R2: 4%; RX: 19%). Median follow-up was 30 months. Four patients (14%) experienced a local relapse and 3 (11%) developed locoregional recurrence. The two-year local control (LC) and locoregional control (LRC) was 96 and 93%, respectively. Median disease-free survival (DFS) was 27 months, metastasis-free survival (MFS) was 69 months, and overall survival (OS) was 93 months. Acute grade 3 toxicity occurred in 11 patients (mucositis, dermatitis, xerostomia; n = 2 each (7%) were the most common) and 2 osteonecroses of the mandibular (grade 3) occurred. No patients experienced grade ≥4 toxicities. Conclusions: Multimodal therapy approaches with surgery followed by IMRT and CIRT boost for SDC leads to good local and locoregional disease control. However, the frequent occurrence of distant metastases limits the prognosis and requires optimization of adjuvant systemic therapies.

INTRODUCTION SDC were first described by Kleinsasser and colleagues in 1968 as a separate group of "adenocarcinomas" of the salivary gland displaying a histopathological resemblance to ductal carcinoma of the breast; the World Health Organization recognized these tumors as a distinct tumor entity in 1991 (1,2). Since then, SDC refer to rare but highly aggressive tumors originating from the ductal epithelium of major salivary glands (3). Malignant salivary gland tumors (MSGT) have an estimated annual incidence rate of 1 to 1.2 per 100,000 (4). SDC account for approximately 1% to 3% of all malignant salivary gland tumors and mostly occur during the fifth to seventh decade of life; men are predominantly affected (4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14). Current treatment options include surgery, systemic therapy, radiation and targeted therapy. Surgical management typically involves complete surgical resection and lymphadenectomy; depending on the tumor localization and stage, this can include parotidectomy, submandibular excision, ipsilateral, and contralateral neck dissection. Owing to its rarity and often poor response, the role of systemic therapies has only been investigated in case series and small clinical trials (15,16).
Despite the advancements in surgery, systemic therapy and radiotherapy, the prognosis of SDC remains meager (6). Despite showing trends toward improved local control, especially the role of adjuvant radiotherapy remains unclear (6,(17)(18)(19)(20). These tumors are currently treated in analogy to other MSGT. In the adjuvant setting, radiotherapy plays a role in patients with higher risk disease, e.g., perineural invasion (PNI), R+, T3/4 tumors. Additionally, definitive radiotherapy is a valuable alternative for unresectable cases. In the adjuvant setting, dose response relationships were described for both LRC (21) and LC (22) in MSGTs. Here, high-linear energy transfer (LET) radiation therapy (e.g., with charged particles such as carbon ions) can lead to improved tumor control rates in other head/neck malignancies as compared to standard photon therapy (22,23). The objective of this retrospective, single-institutional study is to provide further clinical data and prognostic factors regarding intensity modulated radiotherapy (IMRT) with carbon ion radiotherapy (CIRT) in patients with SDC.

Patient Population
Patient records, surgical reports, histological work-up, and radiotherapy treatment plans of patients with SDC who underwent IMRT-CIRT between August 2010 and November 2017 in the Department of Radiation Oncology, University Hospital and at the Heidelberg Ion-Beam Therapy Center (HIT) were evaluated retrospectively. A subset of patients (18%) with locoregional advanced disease or unresectable received a primary radiotherapy.

Radiation Therapy
Treatment planning was performed using native and contrast enhanced CT/MRI. Patients were immobilized with individualized thermoplastic head masks. Technical details of CIRT are described elsewhere (24,25). Treatment planning for CIRT was performed using Syngo PT Planning, Version 13 (Siemens, Erlangen, Germany) and TomoTherapy R -Planning Station (Accuray, Sunnyvale, CA, USA) for photon radiotherapy planning. Patients were treated with a fixed horizontal beam/gantry for CIRT utilizing 1-2 coplanar/non-coplanar beams.
All patients received combined IMRT and CIRT. The base plan was performed using a helical intensitymodulated radiotherapy (IMRT) with daily image guidance (TomoTherapy R , Accuray, Sunnyvale, CA, USA), with 5 daily fractions per week (Figure 1).

Target Volume Delineation and Dose Prescription
Target delineation was based on native/contrast enhanced CT scans fused with contrast-enhanced MRI. Two clinical target volumes (CTV1/CTV2) were outlined. CTV2 comprised the macroscopic tumor and/or tumor bed. CTV1 included CTV2 as well as local growth patterns; ipsilateral nodal levels II-III were included into the CTV1 as well. A 3 mm margin was added to the CTVs to generate the planning target volumes (PTVs).
Organs at risk such as the spinal cord, contralateral parotid gland, temporomandibular joints, and the optic system were constrained per QUANTEC data (26). CTV1 received a median dose of 54 Gy (range: 50-56 Gy) in 2 Gy daily doses (1.8 Gy to 2.0 Gy) (median equivalent dose in 2 Gy fractions (EQD2) of 50 Gy). CTV1 was to be covered by the 90% isodose line. A sequential CIRT boost was applied to the CTV2 utilizing an intensity-controlled active raster-scanning technique, in 3 Gy [2 + (α/β)] (where, D = total dose given in Gy, d = dose per fraction in Gy, and α/β = is assumed to be 2) of 78.5 Gy (range 78.5-80 Gy). We aimed for the CTV2 to be covered by the 95% isodose line. The following equation was used to calculate biologically effective dose (BED) = nd(1 + d α/β ) (where n is the number of fractions, d is fractional dose (in Gy), and α/β is assumed to be 2). The total CIRT dose of 18-24 Gy (RBE) corresponds to a BED of 45-60 Gy.

Follow Up
Patients were monitored on treatment weekly with toxicity assessments (CTCAE classification v.4). Follow-up included a clinical examination by an otorhinolaryngologist and contrast enhanced MR-imaging of the head and neck every 3 months for the first 2 years after radiotherapy, every 6 months until the fifth year after treatment, and annually thereafter. Staging CTs were performed yearly to exclude distant metastases.

Survival, Local, and Locoregional Control
Local control (LC) and locoregional control (LRC) rates were calculated by Kaplan-Meier estimates, from the start of therapy until local tumor progression/death and/or nodal failure. Patients without tumor progression and patients lost to follow-up were censored. FIGURE 1 | Bimodal radiotherapy treatment plan: (A) intensity modulated radiotherapy (IMRT) base plan with 50Gy in 2Gy/fraction and (B) active raster-scanning carbon ion radiotherapy (CIRT) boost plan with 28Gy (RBE) in 3Gy (RBE)/fraction. Treatment was delivered as a definite radiotherapy in a patient with a recurrent rcT2 rcN2b cM0 salivary duct carcinoma (SDC) of the right parotid gland. CIRT was applied with one lateral beam. Histopathological work up revealed Her2neu and androgen receptor (AR) positivity. Therefore, the patient received an adjuvant therapy with trastuzumab and bicalutamide.
Metastasis-free survival (MFS) and disease-free survival (DFS) were calculated by Kaplan-Meier estimates, defined from the start of therapy until distant metastases occurred or progression/relapse at any location, respectively. Patients without events and those lost to follow-up were censored.
Overall Survival (OS) was calculated by Kaplan-Meier estimates, from the start of therapy until death or last contact (alive subjects were censored).

Data Analysis
The log-rank test for univariate analysis was performed to assess prognostic factors for survival. Statistical analyses were performed using SigmaPlot TM (Systat Software GmbH, Germany) software, and a p-value of <0.05 was considered statistically significant.

Treatment Setup and Tumor Characteristics
Detailed patient characteristics are depicted in Table 1. Overall 28 consecutive patients were included with a median age of 69 years (range 41-83 years). 79% (n = 22) of tumors were localized in the parotid gland.
The most frequent location of distant metastases was pulmonary (21%) and osseous (14%) areas. Metachronous distant metastases occurred in 4 patients (21%). In one patient, preexisting bipulmonary metastases were progressive. Overall, the median disease-free survival (DFS) was 27 months (range: 4-107 months). Median overall survival (OS) was 93 months (range: 9-109 months) (Figure 3). Five cases who underwent definitive radiotherapy did not experience a local relapse during followup, but 3 of 5 experienced distant metastases after 6, 7, and 15 months. Patient and tumor characteristics between definitive and postoperative treated patients did not differ significantly.
Here median DFS (p = 0.23) and OS (p = 0.58) did not show statistical differences, even though patient cohorts were rather small for comparison.

Prognostic Factors
On univariate analysis larger CTV and PTV dimension of the CIRT boost (both continuous variates) were prognostic for impaired DFS (p = 0.026 and p = 0.003), MFS (p = 0.006 and p = 0.007) and OS (p = 0.005 and p = 0.005). Nodal involvement was prognostic for poor DFS (p = 0.022), MFS (p = 0.044) (Figure 4) and showed a trend toward impaired OS (p = 0.059). LVI was associated with impaired DFS (p = 0.045) and OS (p = 0.041) (Figure 4). Other known prognostic factors like T-stage, Her2neu, PNI, age, adjuvant systemic therapies and resection status did not show a correlation with any endpoint.

Treatment Related Toxicities
Acute grade 1 and 2 fatigue, mucositis, xerostomia, and dermatitis were commonly observed in the study cohort. Ten acute grade 3 toxicities [two each (7%) of mucositis, dermatitis, xerostomia; and one each (3%) of dysphagia, odynophagia, dysgeusia, nausea/emesis] occurred in 7 patients (25%). In 11 patients (58%) a preexisting facial palsy remained stable during/after radiotherapy. Regarding late adverse event, two osteonecroses of the mandibular jaw occurred 24 and 32 months after radiotherapy. In one patient a surgical intervention was necessary (grade 3) and led to satisfactory long-term results. Overall, no acute or late grade ≥4 toxicities were reported.

DISCUSSION
Although surgery combined with IMRT and CIRT resulted in appropriate LC and LRC, prognosis of patients with SDC is limited by the high rate of distant metastases underlined by a poor MFS in our cohort. However, even when definitively treated, SDCs are linked to a meager prognosis with most of the patients dying within 5 years of diagnosis (7,9,14,39,40,44). High rates of local recurrence (15-55%) and distant metastases (33-62%) account for the worse outcome (33). Local approaches should thus include radical surgical resection e.g., with parotidectomy and neck dissection whenever possible.
Postoperative radiotherapy in SDC is mainly performed as extrapolations from head and neck tumors including MSGT.
However, larger series that focus on the predictors and outcome after radiotherapy are lacking. Smaller series report of 5-year LC, DFS, and OS rates of 67, 45, and 47% after adjuvant radiotherapy with a median photon dose of 60Gy. The authors advise including nerves tracked to the skull base if PNI is presented (19). The addition of radiotherapy can reduce local recurrence rates from approximately 30 to 10% without impacting OS (21). Summarizing various single institution experiences, LC rates were encouraging after surgery and postoperative radiotherapy (17,20).
Overall a benefit for radiotherapy dose escalation for MSGTs has been shown, for instance in adenoid cystic carcinoma (ACC) (45,46). The data for the subgroup of SDC, however, is unclear. The current study presents the first data of advanced radiation techniques with IMRT and high-LET CIRT. With regard to local control in these relatively radioresistant tumors, high-LET radiotherapy seems to be beneficial. In this context, the biophysical advantages with its steep dose-gradient and superior relative biological effectiveness (RBE) allow for safer dose-escalation, like previously described in other tumor entities of the head and neck (47)(48)(49)(50)(51)(52) Furthermore the high physical conformity, compared to photons and decreased lateral scattering as with other particles lead to decreased dose to normal tissue (53). This potentially translates into improved local control by means of safer dose escalation combined with improved sparing of organs at risk. Despite negative prognostic factors in the majority of patients in our cohort, LC and LRC rates of 96 and 93% after 2 years were favorable compared to other reports in the literature (7,9,15,36,43). Our experiences of relatively low DFS and MFS are supported by previous series (6,7,14,39,40).
In the largest database analysis of 228 patients with SDC treated between 1973 and 2008, lymph node involvement, age, large tumor size, and tumor grade were associated with worse disease-specific survival (median OS was 79 months) (6). In another large national registry study in the Netherlands, OS, DFS, and MFS were 51, 23, and 26 months. Herein, the majority of patients (68%) initially presented with lymph node involvement, which is in line with our findings that greater boost volumes and nodal involvement were associated with inferior DFS and OS (14).
Clinical outcomes of 141 patients of a multi-institutional study cohort in Japan, where 59% of patients underwent postoperative radiotherapy, revealed that N+ was associated with lower OS and that the most common treatment failure was distant metastases in 39% (39). These results are consistent with the current study and underline the urgent need for improved systemic therapy.
A histopathological review of 75 cases, with the majority (81.3%) of patients receiving (chemo)radiotherapy, showed that PNI, LVI, and/or extracapsular spread were negative prognostic factors. The addition of chemotherapy to radiotherapy did not improve outcomes (40).
Additionally, there is no consensus on the role of systemic therapy in SDC in general (54)(55)(56). However, androgen receptors are found in 80 to 90% of SDC, as well as 30 to 70% expressing the human epidermal growth factor receptor (EGFR) and Her2neu, making the tumor a target for androgen deprivation therapy and monoclonal antibodies like cetuximab or trastuzumab, respectively (14,40,(57)(58)(59)(60)(61). Recently, adjuvant androgen deprivation in patients with androgen receptor positive SDC has been shown do have a positive impact on DFS and seems to influence OS (62). In a histopathologic study of 50 SDC cases, expression of Her2neu was associated with a more aggressive course of disease (7). In this study, a significant proportion of the assessed tumors were positive for Her2neu and a subset received non-standardized trastuzumab as adjuvant treatment. Furthermore, the majority of tumors assessed were positive for androgen receptors, and received bicalutamide. However, the treatment period, intervals, and combinations thereof were extremely heterogenous, likely why no effect of any systemic therapy in the current analysis could be shown. Moreover, 50 to 70% of tumors expressing the EGFR-receptor may show benefit to EGFR-targeted therapy (56,61,63,64). The high tendency for aggressive growth and patterns of failure demand the optimization of adjuvant treatment regimens. However, prospective trials remain elusive due to the rarity of the disease, even in a multicenter setting. A detailed list of series on surgical treatment and radiotherapy for SDC is provided in Table 2.
Toxicities herein were acceptable. In a retrospective analysis of patients with minor MSGT, several higher-grade toxicities were described, including dysphagia, xerostomia and also hearing loss, which were influenced by the target volume (65). Schulz-Ertner et al. described severe toxicity rates under 5% if radiotherapy is performed with modern techniques like IMRT combined with CIRT (66). Data of high-LET radiotherapy with neutrons produce late toxicities in approximately 10% (67), which is higher compared to these data, although follow up was relatively short herein. Furthermore, the retrospective design and the small patient sizes may add additional biases. In addition, adjuvant therapies were non-standardized and unmonitored herein. Despite these limitations, this is the first study to evaluate advanced radiation techniques using high-LET radiotherapy in SDC. Overall, the combination of surgical resection with neck dissection followed by dose-escalated radiotherapy with IMRT and CIRT leads to good LC. However, the high rate of distant metastases requires optimization of systemic therapies.

CONCLUSIONS
Overall, the combination of surgical resection with neck dissection followed by dose-escalated radiotherapy with IMRT and CIRT leads to good local control rates. Larger tumor size and nodal involvement were associated with inferior disease control and survival. However, the limiting factor in patients with SDC is the high rate of distant mestastases, which is why adjuvant therapy need to optimized.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article.

ETHICS STATEMENT
The study was conducted in accordance with the Declaration of Helsinki. The study was approved by the ethics committee University Heidelberg (S-421/2015). Due to the retrospective nature of the evaluation of the performed standard therapy and the sole use of anonymized data, no study specific informed consent was necessary according to the local ethical guidelines.

AUTHOR CONTRIBUTIONS
SAd and PW: conceptualization. SAd, PW, and FE: methodology. SAd: formal analysis. SAd, PW, SAk, MB, TH, and DB: investigation. SAd and FE: writing-original draft preparation. SAd, JD, and SR: supervision. All authors: writing-review and editing.

FUNDING
This work was supported by Heidelberg University young investigator grants to DB. We thank Thomas Mielke for excellent technical assistance.