Multidisciplinary Management of Radiation-Induced Salivary Gland Carcinomas in the Modern Radiotherapy Era

Simple Summary Etiopathogenesis of salivary gland cancers [SGCs] is largely unknown, even if exposition to ionizing radiation is a recognized risk factor for SGCs development. To date, exhaustive data to guide clinicians in managing patients with radiation-induced [ri] SGCs are scarce and their treatment remains challenging. The purpose of this work is to describe and to analyze clinical and histopathological features, delivered treatments, and outcome of a series of patients with ri-SGCs treated at two Italian cancer referral sites. Given the rarity of ri-SGCs, this retrospective analysis conducted on a case series of 13 patients adds further knowledge to the paucity of literature. The management of these malignancies is extremely complex requiring a multidisciplinary treatment approach. Abstract Clinical data of ri-SGCs patients treated between 2015 and 2019 at a tertiary cancer center and a national hadron therapy facility were reviewed. Latent time (LT) from first RT to ri-SGCs diagnosis, overall (OS), and disease-free survival (DFS) were assessed. Thirteen patients developed 14 ri-SGCs (one patient had 2 synchronous ri-SCGs), after a median LT of 23 years (range 16–34). Parotid was the primary site in 8 cases (57%) and salivary duct carcinoma was the most frequent histotype (29%). Nine patients (69%) underwent surgery (Sx). Among them, 4 patients (31%) underwent Sx alone, 5 received post-operative treatments: 3 (23%) photon-based (X) reRT, one (8%) protons and carbon ions, one (8%) carbon ions only. One patient (8%) received definitive XRT. The remaining 3 patients (23%) received androgen deprivation therapy. With a median follow-up of 48 months (range 24–72), median OS and PFS were 74 and 24 months, respectively. In the subgroup of AR+ ri-SGCs, median PFS and OS were 12 and 74 months, respectively. Given the rarity of ri-SGCs, this work adds further knowledge to the paucity of literature. The management of these malignancies is extremely complex requiring a multidisciplinary treatment approach.


Introduction
Malignant tumors of salivary glands (SGCs) comprise less than 0.5% of all cancers and constitute about 2-8.5% of head and neck cancers (HNCs) [1]. Worldwide annual incidence ranges from 0.4 to 2.6 cases over 100,000 [2] with a male to female ratio of 1.5:1, a higher prevalence in males compared to females during the 6th-7th decade for the salivary duct carcinoma (SDC), being exceedingly uncommon in children [1][2][3].
Etiopathogenesis of SGCs is largely unknown, even if exposition to ionizing radiation is a recognized risk factor for SGCs development. This was reported for the first time in Japanese survivors of the atomic bomb: the frequency of SGCs, mucoepidermoid carinoma (MEC) in particular, was disproportionately high at high radiation doses. In Japanese survivors, out of 145 SGCs registered from 1950 to 1987, 41 were malignant and the proportion of MEC raised with the increasing of radiation dose (p = 0.004 for linear trend) [4]. However, even doses of 2 Gy could raise the risk of developing salivary gland tumors [4,5].
Further evidence supporting the role of radiation in salivary gland carcinogenesis derives from data of children treated with scalp irradiation for tinea capitis, or who received radiotherapy (RT) to the head and neck area to reduce the size of the tonsils and adenoids [6,7]. These children had a 4.5-fold incidence of SGCs compared to untreated people with a mean latency period until tumor development of 11 years [6].
Globally, the standardized incidence ratio of HNCs in survivors of childhood cancer is 13.6 and SGCs represent 65% of these cases [8]. The risk has been associated with the primary tumor type, gender, type of chemotherapy, etc. Radiation has been associated with an increased risk of all carcinomas, most marked for HNCs (SIR 18.5), and the site of the second tumor had arisen in a previous radiation field in 85% of HNCs. Besides, a large study on children survivors after RT (with or without chemotherapy) for a first cancer, showed a linear correlation between dose exposure and the risk of developing solid tumors including SGCs, with the second malignancy occurring at least eight years after the end of the first oncological treatment [9].
Exhaustive data to guide clinicians in managing patients with ri-SGCs are scarce and their treatment remains challenging, especially in the era of targeted systemic treatments and innovative radiation therapies, namely intensity modulated RT (IMRT) and hadron therapy [9][10][11][12][13].
Ideally, surgery with wide margins remains the mainstay of treatment followed by RT in high-risk cases [14,15]. However, radical salvage surgery is often not feasible, essentially because of technical difficulties when operating within an irradiated area [9,12,16] or due to the local extension of disease that prevents a radical resection.
In addition, further RT could be hardly given since normal tissues surrounding secondary SGCs have been generally treated to near their dose tolerance. Finally, in some cases, patients who suffer from secondary SGCs are young and potentially candidates to develop a third cancer, in particular in the presence of a genetic susceptibility [17]. In this scenario, modern RT techniques, including IMRT and particle therapy, may play a principal role in the management of these patients. Indeed, thanks to their physical properties, they have an excellent sparing of normal tissue outside the target and they are able to overcome an enhanced hypo-oxygenation that can also be present in ri-SGCs [14].
With the limitations of the descriptive nature of our work and the limited patient cohort, the purpose of our review is to describe and to analyze clinical and histopathological features, delivered treatments, and outcome of a series of patients with ri-SGCs.
Immunohistochemical (IHC) research of androgen receptor has been done only in cases with diagnosis of SDC, adenocarcinoma NOS, and carcinoma ex pleomorphic adenoma. Seven out of 8 patients had androgen receptors (ARs) overexpression. HER2 overexpression was not identified. NGS was performed in 6 out of 13 (46%) cases, in 3 cases somatic tumor mutations have been found: BRCA2 mutation in one SDC, TP53 mutation in one adenocarcinoma NOS, PIK3CA and HRAS mutations in another SDC.
Concerning the relationship with the previous treated volumes, data were available for 9 (64%) ri-SGCs (one patient had two synchronous ri-SGCs, (Table 1). Six ri-SGCs developed within the previously treated volumes and in 5 of them, ri-SGC was marginal to high-dose volumes. Three ri-SGCs occurred within the initial target volume.

Treatment and Outcome for Secondary SGCs
Nine patients (69%) underwent surgery, either alone (n = 3), in combination with RT (n = 5), or with postoperative systemic treatment (n = 1); resection margins were positive in 8 cases. One patient received definitive RT. The remaining 3 patients received palliative treatment.
Seven patients (54%) out of 13 received RT as part of their multimodality treatment. Post-operative photon-based RT (XRT) (n = 3) doses ranged from 64 Gy to 70 Gy (median of 67 Gy); one patient received a palliative course of 30 Gy for an unresectable parotid gland tumor. Particle therapy was delivered in two cases: one patient was treated by post-operative Proton Therapy (PT) with a boost of Carbon Ion Radiotherapy (CIRT), receiving 6 Gy(RBE) with CIRT and 59.4 Gy with PT. The other one was treated with post-operative CIRT after microscopic residual surgery, with a total dose of 68.8 Gy(RBE). In the subgroup treated with post-operative RT, median age at primary tumor diagnosis was 12 years (range 3-27) and median LT was 17 years (range 5-41 years). The patient treated with radical RT was 8 years old at the time of the diagnosis of the primary tumor, with an LT of 37 years. With regard to acute toxicities, all of the patients who underwent post-operative RT developed mild to moderate xerostomia and oral mucositis (≤G2); as regards late toxicities, three patients treated with surgery and XRT developed buccal spasms (n = 1), fibrosis (n = 1), trismus and neuropathy of the facial nerve (n = 2). Both the 2 patients treated with post-operative particle therapy developed G1 trismus, G1 xerostomia, and G1 neuropathy of the facial and trigeminal nerve.

Discussion
We reported a small series of 13 patients with ri-SGCs with a more favorable outcome than previously published older series. Median PFS and OS were 24 months (95% CI: 6 months-not reached) and 74 months (95% CI: 36-74), respectively, despite high-grade histology in 77% of cases and advanced stage (III-IV) in 10 out 13 cases. These results are very encouraging, if we consider that 9 patients (69%) received surgery, 1 patient (8%) was treated with definitive XRT and concomitant androgen deprivation therapy (ADT), 1 patient (8%) received palliative XRT and subsequent ADT, whereas the remaining 2 patients (15%) received only systemic therapy. In the past, a 5-year OS higher than 90% was reported, however, mucoepidermoid was the most common histotype and almost all patients received surgery combined or not with RT [9,12]. In the analysis by Bhattacharyya et al., for example, patients with MECs exhibited a higher 5-year OS (81.5%), compared to other histotypes such as ACC (70.7%) and carcinoma ex pleomorphic adenoma (40.2%) [18]. On the other hand, Mallik et al. reported 47 submandibular gland cancers, mostly ACCs, with a 5-year DFS of 71.8%, 12.8% (n = 6) of loco-regional failure, and 12.8% (n = 6) of development of distant metastases. Compared to the past, availability of new effective systemic treatments (e.g., androgen deprivation therapy, ADT; anti-HER2 monoclonal antibody), improvement in the re-irradiation (reRT) [19] techniques and the opening of heavy ions facility in Italy in 2011 have offered new therapeutic opportunities.
ReRT should be usually proposed as post-operative treatment, since it is related to prolonged survival rates, especially for ACC [39]: in our cohort, 5 patients out of 13 (38%) received a post-operative RT [any energy]. Indications for post-operative RT include advanced stage (T3, T4, n+), positive surgical margins, high tumor grade, perineural invasion, recurrent diseases [40,41]. Moreover, reRT should be taken into account whenever surgery is not feasible, even though no data from prospective randomized data are available in the literature. Nowadays, state-of-the-art RT techniques, such as IMRT and stereotactic body RT (SBRT) allow steep dose gradients with optimal coverage even for complex target volumes, while allowing a good spare of close healthy tissues. Following NCCN recommendations, reRT should take place at least 6 months after the end of the first RT course and target volumes should be taken as narrower as possible, avoiding elective nodes coverage. Results are encouraging: Karam et al. re-treated 18 consecutive patients with SBRT with a median dose of 30 Gy delivered in 5 fractions. With a median cumulative dose of 91.1 Gy, 2-years local control and OS were 53% and 39%, respectively, even if 4 patients eventually experienced soft tissue necrosis as a late severe toxicity [42].
In regards to IMRT, most of the experience in literature refers to mixed series where the number of SGCs is significantly lower than pharyngeal or laryngeal squamous cell carcinoma cases [19,43,44]. Recently, Orlandi et al. analyzed reRT outcomes on a cohort of 159 patients, of which 71 (45%) were not squamous cell carcinomas (in particular, 25 (16%) were ACC and 8 (5%) were affected by SGCs). With a median follow up of 49.9 months, 5-years OS and PFS were 43.5% (95% CI, 34.6-54.8%) and 20.9% (95% CI, 14.7-29.6%), respectively. It is hard to make comparisons with our current results, due to the limited number of patients and the mixed cohort of the cited study. To our knowledge, there are no data available in literature concerning reRT on an exclusive ri-SGCs cohort.
Treatment with hadrontherapy represents a new approach considering that most of these tumors arose in a previously operated and irradiated field. PT and CIRT seem to be a reasonable choice to overcome dosimetric thresholds in unresectable patients. Two patients (one teenager) received PT and CIRT (delivered doses: PT 59.4 Gy(RBE) + CIRT 6 Gy(RBE); CIRT 68.8 Gy(RBE)). The physical properties of PT can be exploited in pediatric patients. Through PT nearly the same physical dose of XRT can be delivered, with a neat fall of dose outside the target: Grant et al. reported a 53% of grade 2 − grade 3 (G2-G3) dermatitis, 0% of G2-G3 dysphagia, and a 46% of G2-G3 mucositis in a cohort of 13 pediatric patients receiving 60 Gy(RBE), compared to 11 patients treated with XRT (60 Gy, 54%, 27% and 91% of toxicities rates, respectively). Takagi et al. compared PT and CIRT in a cohort of 80 pts treated for head and neck ACC: no differences were found in terms of local control (66% for PT, 68% for CIRT) for T4 unresectable cases when a dose of 65.0 Gy(RBE) was delivered. CIRT combines the ballistic properties and superior biological equivalent dose, allowing to retreat unresectable radio-resistant tumors. With a median CIRT dose of 51 Gy(RBE) (27 × 3 Gy(RBE)) delivered 61 months after the first radiotherapy course, Jensen et al. reported an objective response rate of 57% and a local control (median follow-up 14 months) of 70%; in that cohort, median cumulative dose was 128 Gy(RBE), with acceptable toxicity rates [45,46]. The Italian experience concerning reRT of SGCs with CIRT has been recently published [47]: from 2013 to 2016, 51 patients were treated with a median dose of 60 Gy(RBE) at 3-5 Gy(RBE) per fraction. After a median follow-up of 19 months, the local control rate was 41.2%; one-year OS and PFS were 90.2% and 71.7%, respectively.
In line with other papers, the treatment was well tolerated, with no G4 or G5 toxicities and only 17.5% of G3 events. Moreover, compared to other series with CIRT [46,48], reRT at CNAO did not suffer from soft tissue necrosis or carotid blow-out syndrome events [47]. This was probably related to the lower biologically equivalent doses (BED) delivered-155.2-167.4 Gy(RBE)-which could likely be increased in order to improve the outcomes. Interestingly, Vischioni et al. reported a shorter median LT, 6.33 years (range 1.08-20) compared to our results.
Despite the clinical challenges facing by any HNC specialist (i.e., radiation oncologist, head and neck surgeons, oro-maxillofacial surgeons, medical oncologist) Gs, a multidisciplinary approach is needed when dealing with ri-SC. This approach has been strongly recommended to achieve the best oncologic outcome and prevent or adequately treat any adverse effects [49][50][51]. Indeed, it should be considered as mandatory for SGCs as well [52]. In our series, most patients did not receive radical surgery; this is not an unexpected finding since post-treatment fibrosis and distortion of anatomy due to the treatment[s] for the previous tumor may increase the rate of positive margins. In this scenario, post-operative RT can almost double local control rate [53], and our results confirm that among 5 patients receiving marginal Sx and post-operative RT, 4 (80%) had no evidence of disease at last follow up; the remaining one did, but he actually received CIRT. Data on the radiobiological response to this type of RT are still being collected, and a stable disease does not always translate into treatment failure [54]. Nevertheless, all of our patients treated by Sx and post-operative RT were alive at last follow up. In this context, a dedicated pathologist is advisable as SGCs show the greater discrepancy between the initial and definitive pathological diagnosis [55]; moreover, pathology must guide systemic therapy by identifying druggable targets. This approach allowed almost half of our patients to benefit from ADT, warranting at least 12 months of survival even without a curative treatment. Lastly, a multidisciplinary approach allows late toxicities to be more promptly identified and managed; this could explain the low number of severe sequelae among our patients and could result in a better quality of life.
RT is a well-known causative agent in the development of secondary solid tumors and the issue of second cancers following therapeutic radiation for a wide variety of malignancies is currently receiving increasing attention. Indeed, it is well recognized that patients receiving radiation therapy have a higher long-term risk for developing second primary cancers compared with patients who do not [13].
In pediatric patients, who constitute most of the cases originally treated in this study, germline mutations and hereditary conditions may place one at risk for developing a secondary malignancy, and may not be related solely to chemotherapy [25]. Considering that 10 patients in our series received radiation and chemotherapy compared to 3 patients who received radiation alone, our series suggests that the association of chemo-radiation may increase the risk of second SGCs compared to radiation alone. This may lead to the idea of treatment-induced second SGCs, rather than simply radio-induced secondary malignancies.
The main limitations of our research are its descriptive nature and the limited patient cohort. Beal KB et al. reported 18 radiation-induced salivary gland tumors, 15 of which were ri-SGCs [11]. The main strengths of this work are central pathologic review of the 12 non-ACC tumor specimens, the prolonged follow-up, and the details about treatments for ri-SGCs.

Materials and Methods
We retrospectively analyzed clinical data of consecutive patients with ri-SGCs referring to Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy, and to Centro Nazionale di Adroterapia Oncologica (CNAO, National Center of Oncological Hadrontherapy), Pavia, Italy, between 2015 and 2019. Patients were managed by head and neck and pediatric multidisciplinary teams established at both the former Institutions and involving dedicated radiation oncologists. We considered as eligible adult and pediatric patients with a confirmed pathological diagnosis of SGC, who received for previous tumors curative RT in HN areas, lung, and upper mediastinum, with or without concurrent or sequential chemotherapy. Multidisciplinary discussion has been considered within the inclusion criteria as well. Patients with primary leukemia, benign lesions, and patients who received only chemotherapy as a curative treatment have been excluded.
Patients' characteristics, first tumor diagnosis, and secondary SGCs features, latent time (LT) from initial treatment to the development of SGCs, treatment, and outcome for secondary SGCs were analyzed.
Revision of all specimens was performed by a dedicated HN pathologist (PQ). Immunohistochemical (IHC) profile including androgen receptors and HER2 has been done in SDCs and adenocarcinoma not otherwise specified (NOS). Next-generation sequencing (NGS) has been performed on the tumor specimen until it was available for free (September 2019). RT plan of the primary tumor, when possible, was reconstructed on CT planning of the secondary SGCs with the aim of searching for a relationship between the dose distribution of the first RT and the second malignant tumor. Median overall survival (OS) and progression-free survival (PFS) and their 95% confidence interval (CI) were estimated using the Kaplan-Meier method. Further subgroup analyses were realized according to histological features of SGCs.
This study was approved on June 2020 by the Ethical Committee of Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy (internal study identification number: INT 127-20).

Conclusions
Our findings underline the benefit of a multidisciplinary approach and intraspeciality care for the management of this complex clinical scenario. This is realized through tailored treatments which are properly balanced between tumor control and toxicity, as to allow similar outcome to ex novo SGCs.