Patterns of care, toxicity and outcome in the treatment of salivary gland carcinomas: long-term experience from a tertiary cancer center

Salivary gland carcinomas (SGC) account for 1–6% of all head and neck tumors and comprise a variety of 20 different histological subtypes as classified by the World Health Organization (WHO) in 2017 [1, 2]. Some of those subtypes are extremely rare, which prohibits clinical trials being conducted, and thus the development of evidence-based guidelines for treatment standardization [1]. Seventy percent of SGC arise from the parotid gland, while 20–25% derive from the submandibular and sublingual glands. Whereas acinic cell carcinoma (ACC, 15–17%), adenoid cystic carcinoma (AdCC, 16–27%) and mucoepidermoid carcinoma (MEC, 15–38%) represent frequent histopathological entities in the paired, major salivary glands, AdCC and MEC are also common in the minor salivary glands of the oral cavity [3]. While ACC depicts a low-risk malignancy, other histopathological subtypes such as AdCC, salivary duct carcinoma (SDC), or adenocarcinoma—not otherwise specified (ACA) are high-risk tumors with different prognosis [1]. By actual guidelines [4], surgical resection is the standard treatment for SGC including neck dissection (ND) in case of adverse features such as advanced tumor stages (T3–T4) and/or clinically positive lymph nodes of the neck. A common therapeutic challenge, even for high-volume centers, are the positive resection margins often described in the literature, which often lead to the decision for adjuvant/additive treatment [5‐7]. Postoperative radiotherapy (PORT) is recommended in case of adverse pathological features (e.g., advanced tumor stages T3–T4, positive resection margins), and AdCC [8]. Various irradiation modalities have been used in the past, including particle-radiotherapy characterized by a high linear energy transfer (LET) and relative biological efficacy (RBE), with heavy ions and neutrons showing some promising results even for radioresistant histopathological entities [9, 10]. Due to the lack of prospective, randomized clinical trials, the benefit of multimodal protocols has not been proven, yet. Data on the efficiency of chemotherapy (CTX), targeted therapy, and immunotherapy for SGC are rare, and so far, without prospective evidence for improved outcome [8, 11‐14]. This study aims to analyze patterns of care and the efficacy of surgery, radiotherapy (RT), and chemoradiotherapy (CRT) as single or combined modalities in the treatment of SGC with a focus on the outcome and adverse events from a longstanding experience of a university cancer center.

To date, only a few large retrospective studies regarding the treatment of SGC have been published and to the best of our knowledge, there are no prospective trials reporting results on primary treatment modalities including surgery or radiotherapy. Besides, most of the retrospective trials either report on major or minor salivary glands or distinct histopathological subtypes, exclusively [5, 22‐25].

The current study provides comprehensive data on the long-term outcome with a median follow-up period of 55 and 74 months for those patients who were alive by the time of their last appointment. We consider this to be a major strength of the study, as long-term follow-up has been recommended to detect local recurrences or metastases occurring beyond 60 months after initial treatment [26]. Furthermore, to prevent selection bias, all treated patients presented in the cancer center during the period analyzed were included in analysis.

The cohort of patients enrolled is largely heterogeneous, as expected for this diagnosis, however, representative of the daily clinical practice in a large university cancer center. The median age by the time of initial tumor diagnosis and the slight predominance of the male gender is in accordance with two large retrospective cohorts encompassing different entities of SGC, reporting overall median ages of 58 and 63 years and proportions of male patients to be 52.0% and 58.5%, respectively [6, 7]. Similar to one of the largest studies on SGC by Terhaard et al. on behalf of the Dutch Head and Neck Oncology Cooperative Group retrospectively including 565 patients [7], and a follow-up study by Westergaard-Nielsen et al. including 1601 SGC patients [27], the tumors in our cohort were more likely to be diagnosed at T1/T2-stage and N0-stage without signs of distant metastases. Furthermore, in this cohort, AdCC, ACA and MEC were the predominant histological subtypes, in accordance with the literature [7]. Interestingly, the percentage of SCC in our cohort is 11.8% and higher compared to previous reports (4%) [1, 28], so that false clinical diagnosis as an SGC primary instead of metastasis from other tumors at least for a part of these cases cannot be ruled out. Primary SCC of the salivary glands, mostly arising from Stensen’s and Wharton’s ducts is mostly caused by chronic inflammation as result of e.g., chronic sialolithiasis and has been described before [1].

Complete tumor resection in our study was achieved in 63.6% of the cases, slightly more often than reported by Terhaard et al. where complete primary tumor resection was achieved in 59% of the cases [7]. Incomplete primary tumor resection is not uncommon and complete resection rates as low as 31.3% have been reported even for experienced and high-volume centers [5, 6]. Parotid tumors in this study were more often diagnosed at advanced patient age (p = 0.006). ACA and SCC occurred more often in the parotid glands and their diagnosis has been associated with higher age before [7].

Ten year OS rate was 73.2% in our cohort, which is higher as reported in literature ranging from 43 to 52% [5‐7, 25, 27, 29, 30]. Independent prognostic factors for OS in the present study were parotid tumor site, advanced T- and positive N-stage. Yet, tumor site has been associated with OS in multivariate analysis by other authors as well [7, 31]. Moreover, similar to our data, Bjørndal et al. stressed the prognostic importance of tumor stage for recurrence and survival [30]. Possible explanations for the higher OS rate in our cohort might be the inclusion of predominantly high-risk subtypes by other authors [25], or older patient series ranging back to the 1940s with a lack of modern diagnostic imaging using different RT techniques [5‐7, 29].

Primary RT/CRT was performed when patients were ineligible for primary surgery. A review by Wang et al. revealed significant impairment of OS for primary RT when compared to PORT in five series with a total of 913 patients [8]. Due to patients’ risk profiles, ND was more often performed for parotid gland tumors than for other tumor locations (p = 0.008). Performing ND did not have a significant impact on OS. POCRT in this series was applied in cases of loco-regionally advanced ACA or SCC with additional risk factors like close or positive resection margins and extra-nodal extension and concomitant CTX was platin-based in combination with 5-FU in most cases. Despite the unfavorable tumor characteristics outcome was not impaired for these patients when compared to the group receiving only PORT. Indeed, from the cytotoxic regimens tested with or without RT for SGC, cisplatin and 5-FU were shown to be amongst the most efficient agents within various small patient series for locally advanced or recurrent/metastatic disease [8, 11]. While some authors reported good local control rates for POCRT, Hsieh et al. reported improved local control rates for POCRT (n = 58) in comparison to PORT (n = 33) [32‐34]. However, a SEER—Surveillance, Epidemiology, and End Results—Medicare database analysis revealed increased mortality and toxicity for the concomitant POCRT when compared to PORT alone [35]. Due to heterogeneous patient cohorts and individual indications for POCRT, interpretation of such data remains difficult. Taken together, no evidence exists for an additional benefit of CTX combined with RT in the postoperative setting. The main hematological toxicities are non-negligible and, therefore, the prescription of such regimens should be decided critically and reserved for selected, individual cases.

LPFS rate after 10 years was 66.9% in our cohort, which is in line with the literature. In analyses encompassing different histological subgroups and primary tumor sites, likewise this study, LPFS varied between 59 and 88% at 10 years [5‐7, 25, 27, 36]. Various series reported improved loco-regional control for the combination of surgery and PORT [7, 37, 38]. However, in our cohort, the outcome was not affected by the implementation of PORT in multivariate analysis, although PORT was associated with both improved DMFS and PFS in univariate analysis. In accordance with our multivariate analysis, primary tumor site (parotid gland), advanced T- and positive N-stage have been reported to impair LPFS [6, 7, 31].

In the literature, the most common failures are distant metastases. Reported rates range from 15 to 37% after 10 years which is in line with our results [5‐7, 29, 36, 37, 39, 40]. Primary tumor site, loco-regionally advanced tumor, and nodal stage were associated with DMFS in multivariate analysis, as reported before [6, 7, 36, 40‐42].

In our cohort, adverse events following surgical intervention often comprised injuries of the cranial nerves or their branches (33.1%) with facial nerve injuries being the most common (80.0%), whereas postoperative hemorrhage and wound-healing disorders were rare. Several authors reported on rates of facial nerve injuries from parotidectomies and numbers range between 29.2% and 45.0%, yet [29, 43‐45]. CRT/RT was well tolerated with low grade ≥ III toxicity rates compared to the literature (7.4–16.0%), which might be related to the broader implementation of IMRT and image-guidance in our cohort [5, 6, 24]. Dermatitis and mucositis of any grade were the most common RT-related sequelae and occurred as expected in almost any patient. No unexpected or grade IV–V acute toxicities occurred in this cohort and all patients but one completed the prescribed course, which may indicate that modern techniques can improve tolerability of this treatment modality [23, 29].

This study has several limitations: first, its retrospective character. Second, the heterogeneity of the patient cohort in terms of tumor site, histopathological entities and treatment hamper comparability and the derivation of treatment recommendations. Third, low patient numbers for rare tumor entities. Fourth, the pathological parameters vascular and lymphatic invasion, perineural invasion as well as extra-nodal extension were not analyzed. Fifth, possible referral bias. Sixth, the inclusion of some very advanced/palliative cases, which might have distorted the results. Although there are limitations to this study as mentioned before, this is one of the most recent series regarding the treatment of SGC from a tertiary tumor center, implementing all current techniques of surgery and radiotherapy with a comparably long follow-up period.

SGC are a heterogeneous group of carcinomas with varying outcome depending on factors such as primary tumor site and tumor stage. Optimizing multimodal treatment strategies for distinct histological entities, tumor localizations, and stages remains challenging due to low incidence rates. Radical surgery remains the mainstay of treatment, followed by radiotherapy in case of adverse pathological features, and results in fairly good survival rates, at least in high-volume centers. The role of chemotherapy remains controversial. However, to further improve outcome and reduce toxicities, prospective clinical trials are warranted. Large, international, prospective registers and systematic retrospective meta-analyses are necessary for overcoming recruiting problems for the rarest entities.