Surgical Management of Patients with Parotid Involvement from Non-Melanoma Skin Cancer of the Head and Neck

We conducted a retrospective, longitudinal study on a single-center series of patients who underwent parotidectomy in the management of advanced head and neck non-melanoma skin cancer (hnNMSC). The aim of this study was to identify prognostic factors associated with worse outcomes. Forty-one men and nine women were included. The mean age at the time of surgery was 78.9 years. The 5-year overall survival, disease-specific survival, locoregional recurrence-free survival, and distant metastasis-free survival calculated with Kaplan–Meier curves were 39.9%, 56.3%, 58.6%, and 82.1%, respectively. A univariate analysis showed that the status of the margins, facial nerve direct involvement, lymph vascular invasion, and histological grading were associated with worse outcomes (p < 0.05). Positive margins were associated with worse disease-specific survival also in a multivariate analysis (p = 0.001, HR = 32.02, and CIs 4.338 to 351.3). Because the resection in free margins is the most important prognostic factor, early diagnosis or, in the case of advanced disease, extensive surgical resection with concomitant reconstruction is needed. Adjuvant therapy is indicated in selected cases.


Introduction
Skin cancers are classified in two major categories: melanoma and non-melanoma skin cancer (NMSC) [1].They accounted for 330,000 and 1,200,000 new cases worldwide, respectively, in 2022 [2].
Chronic solar ultraviolet radiation is the main etiological factor; as a consequence, most lesions develop from the head and neck in elderly patients [3,4].
Compared to melanoma, NMSC has a lower propensity to distant metastasis and an overall better survival [5].
NMSC usually presents as a small lesion, and simple surgical excision in free margins guarantees a high cure rate, with acceptable cosmetic and functional results.
Approximately 1%-5% of patients with NMSC present with locally advanced disease, with lymph node metastasis, and/or infiltration of contiguous structures [7].In these patients, survival is significantly reduced [8].
In 1963, Conley and Arena [13] first described a series of 81 patients who underwent surgical treatment for parotid involvement from non-salivary gland tumors.The survival rate at five years was 12.5%.It was previously considered that non-salivary parotid tumors were rare [14] until O'Brien in 1993 [15] described a series of 242 consecutive parotidectomies, in which metastasis from melanoma and hnNMSC were the main malignant histotypes.Nine years later, in 2002, O'Brien et al. [16] observed that intraparotid metastasis from cutaneous SCC had distinct prognostic behavior.This led to the introduction of a prognostic classification, which distinguished intraparotid lymph node metastasis (PM) from neck lymph node metastasis.Although this classification improves prognostic accuracy, there is still an insufficient correlation with pathological data.The N1S3 classification [17] was thus introduced to improve these aspects.Despite the research in the prognostic value of PM, definitive evidence is still lacking.The AJCC-TNM classification [1], currently the most widely used, does not distinguish between PM and neck metastases, but in its eighth edition, it differentiates between hnNMSC and lesions arising from other areas.
Nowadays, the management of the parotid gland represents the cornerstone of treatment in advanced hnNMSC.
In the case of parotid involvement with delayed diagnosis, patients must undergo extensive resection surgery followed by immediate reconstruction.Adjuvant radiotherapy (RT) can improve locoregional control and survival in these advanced cases [4,18,19].
Despite the improvement of a multidisciplinary approach for advanced hnNMSC, the importance of parotid involvement is still not clear, and many patients experience relapse and die as a consequence of uncontrolled disease [4].The present study analyzes a single-center series of patients who underwent parotidectomy for advanced hnNMSC.The objective of this study was to describe the experience of a tertiary center in the diagnosis and surgical management of patients with advanced hnNMSC over a ten-year period and to help clinicians tailor the treatment protocol for patients with parotid involvement, together with giving advice to the primary surgeon as to planning rigorous follow-up in the case of resection of hnNMSC.

Study Design, Data Extraction, and Variables
We performed a retrospective, longitudinal analysis approved by the Ethical Committee on Human Clinical Research of our hospital (PG/2016/4730) of all the consecutive patients who underwent parotidectomy for hnNMSC at the Department of Otorhinolaryngology-Head and Neck Surgery of Cagliari State University between January 2011 and May 2022.
Informed consent was obtained from all the subjects involved in this study.The current study included patients who underwent parotid surgery for PM, those who had undergone surgery for NMSC where the parotid parenchyma was directly infiltrated, and those who had undergone parotidectomy as part of the surgical management of NMSC where the final histology did not demonstrate parotid involvement.
The clinical data were obtained by the review of clinical charts, direct clinical reexamination and imaging of surviving patients, and phone calls to the relatives of deceased patients.All the data were organized with Microsoft Excel (Microsoft; Redmond, WA, USA).The demographic and clinical data included sex, age at surgery, age-adjusted Charlson Comorbidity Index, and physical examination at diagnosis.The histopathological and clinical variables used included the histotypes, site of disease, skin involvement, status of margins, sites of metastasis, extranodal extensions (ENEs), facial nerve (FN) status, perineural invasion (PNI), lymphovascular invasion (LVI), histological grading (G), TNM stage, O'Brien stage, N1S3 stage, and adjuvant treatments.The main outcomes were overall survival (OS), disease-specific survival (DSS), locoregional relapse-free survival (LRFS), and distant metastasis-free survival (DMFS).The outcomes were calculated from the day of surgery to, respectively, the day of death for other causes, day of death disease-related, date of clinical or radiological diagnosis of locoregional relapse, and date of radiological diagnosis of distant metastasis.For censored subjects, the date of the last follow-up was used as the endpoint.

Preoperative Work-Up and Treatment Principles
The preoperative work-up included ultrasonography (US), computer tomography (CT), magnetic resonance imaging (MRI), and histological assessment by fine-needle aspiration cytology (FNAC) and/or incisional biopsy.
The comorbidities were evaluated using the age-adjusted Charlson Comorbidity Index [20].
The parotidectomies were classified according to the European Salivary Gland Society Classification (ESGSC) [21].
The FN dissection was always performed under operative microscopic view or with the 3D exoscope.The systematic intraoperative FN monitoring was always performed using a four-channel free-running electromyography nerve integrity monitor system (NIM 3.0, Medtronic USA, Inc., Jacksonville, FL, USA).
The preoperative and postoperative FN function was evaluated with the House-Brackmann grading system (HB) [22].A total or partial FN resection was performed in the case of nerve infiltration.In the case of FN resection, intraoperative neurorrhaphy and/or static suspension of the facial palsy was executed to reduce the impact of the FN postoperative deficit.
According to the NCCN Guidelines [23,24], elective neck dissection was always indicated in patients with a clinical diagnosis of parotid or lymph node metastasis.
The patients with cT3 and cT4 cN0 lesions of the auricular or periauricular region underwent prophylactic neck dissection associated with type I-II parotidectomy (superficial parotidectomy).
The surgical defects were classified according to the Rosenthal classification [25].In the case of extensive resections, the reconstruction was performed with free or pedicled flaps.The extent of surgery, age, comorbidities, and performance status influenced the reconstructive choice.
In the patients with a high risk of postoperative airway impairment (pedicle or free flap reconstruction and bilateral neck dissection), a concomitant elective tracheostomy was performed.
The Clavien-Dindo classification was used for the evaluation of postoperative complications [26].
According to the Royal College of Pathologists guidelines, the margin status was evaluated as follows: complete excision (the tumor is fully excised and surrounded by a margin of normal tissue of at least 1 mm), involved margin (the specimen analyzed shows cancer present at the margin), and close margin (the specimen analyzed shows a fully excised tumor with, however, a ≤1 mm margin of normal tissue at one or more sites of the specimen) [27,28].
SCC, BCC, and MCC were staged according to the 8th edition of the AJCC-TNM [1].The patients with intraparotid metastasis of SCC were also staged with the O'Brien classification [16] and N1S3 classification [17].
The adjuvant therapy was planned according to the NCCN Guidelines [23,24], following a discussion by a multidisciplinary tumor board.The RT field, dosage, modality, and duration was always planned by the radiation oncologist.Usually, a dose of 54-63 Gray was administered on the P and N, in 30-33 fractions, with conformal radiation therapy or intensity-modulated radiation therapy.Usually, three cycles of Cisplatin were administered by the medical oncologist as the chemotherapy (CHT) regimens.
All the patients were followed-up regularly at our institution for 5 years after surgery to rule out regional or distant relapses.Our protocol consists of a clinical examination every 2 months for the first 2 years, every 3 months for the third year, every 4 months for the fourth year, and every 5 months for the fifth year.Head and neck CT or MRI with contrast medium were scheduled every 6 months for the first 2 years and once every year for the remaining 3 years.Chest CT with contrast medium was performed once a year.Biopsy or cytology were executed in the case of suspected relapse.All the follow-up data were updated until April 2023.

Statistical Analysis
The quantitative variables (age and age-adjusted Charlson Comorbidity Index) were reported as the mean, range and standard deviations (SDs).The qualitative variables were reported as the absolute (N • ) and relative frequencies (%).A statistical analysis was performed using GraphPad Prism 9 (San Diego, CA, USA).The Kaplan-Meier curves were calculated for survival outcomes at 5 years (OS, DSS, LRFS, and DMFS).The univariate analysis results were reported as the survival proportion, hazard ratios (HRs), and confidence intervals (CIs).The log-rank, Mantel, and Wilcoxon tests were used to compare the survival curves.Cox proportional hazards regression was used for the multivariate analysis.A p-value < 0.05 was considered significant.
A total of 37 patients (74%) had PM at clinical presentation, while 13 patients (26%) underwent resection of the primaries of the periauricular region extended to the parotid gland to obtain free deep margins of resection.
Among the patients with PM, 23 patients (62.2%, 46% of all series) had a clinical history of previous hnNMSC resection without neck or parotid staging performed elsewhere, 13 patients (35.1%, 26% of all series) were treated for primary hnNMSC of the head and neck with concomitant PM, and 1 patient (2.7%, 2% of all series) developed PM 8 months after the resection of SCC of the submandibular skin associated with level I-V neck dissection and adjuvant chemotherapy (CHT-RT) performed at our institution.
The mean time between the treatment of the primary tumor and the diagnosis of PM (available for only 12 cases) was 8.9 months (range: 1-18 months).
The cutaneous primaries are detailed in Table 1.
A preoperative radiological evaluation was obtained for all the patients: 12 patients underwent preoperative work-up with MRI, 30 patients with CT, and 8 patients with MRI and CT.Distant metastases were excluded by chest CT with contrast medium.
The preoperative histology was obtained with incisional biopsy in 26 patients (52%) with a cutaneous lesion.The preoperative diagnosis of malignancy in the patients observed for a parotid lump without a cutaneous lesion was based on the FNAC in 10 patients (20%), and in 14 patients (28%), the diagnosis was based on the clinical history of a recent resection of hnNMSC.
In 23 cases (44.2%), the FN was involved by the tumor, but only 8 patients (16%) presented a preoperative FN impairment (HB > 1).The surgical resection was extended to a single branch of the FN in 12 cases (52.2%) and to the main trunk in 11 cases (47.8%), as shown in Table 2.In these cases, intraoperative neurorrhaphy and/or static rehabilitation of the facial palsy was performed to reduce the morbidity of the postoperative FN palsy.
Forty-nine patients (98%) underwent concomitant neck dissection.A total of 46 patients underwent ipsilateral neck dissection, 2 patients with bilateral parotid involvement underwent bilateral neck dissection (n = 4 neck dissection), and 1 patient previously treated with unilateral neck dissection underwent contralateral neck dissection.In total, 51 neck dissections were performed: 28 selective neck dissections (54.9%), 12 functional neck dissections (23.5%), and 11 modified radical neck dissections (21.6%), as shown in Table 2.One patient (2%) did not undergo neck dissection because of the histology (BCC).In 28 cases, the surgical resection was extended to the surrounding structures (skin or ear lobe in 21 cases, skin and cortical bone from the mastoid and/or zygomatic process in 6 cases, and skin and mastoid in 1 case).According to the Rosenthal classification [25] of the defects of the lateral region, 12 patients (42.9%) had a type I defect, 12 patients (42.9%) had a type II defect, and 4 patients (14.3%) had a type III defect.The reconstruction was made with a free flap (n = 20, 71.4%) or pedicled flap (n = 8, 28.6%), as reported in Tables 2 and 3.The recipient artery for microvascular anastomosis was the facial artery in 17 cases (85%), the superior thyroid artery in 2 cases (10%), and the ascending pharyngeal artery in 1 case (5%).In all the cases, the recipient vein for the microanastomosis was one of the branches of the thyro-lingual-facial trunk.Venous drainage was obtained with a single anastomosis in most cases (n = 18, 90%), using the Microvascular Anastomotic Coupler Device (MACD) produced by GEM Synovis MCA [29] in all the cases.The mean time of anastomosis (TA), defined as the operative time of anastomosis of both the arterial and venous sides, was 42.3 min (range 20-60 min).The mean time of ischemia (TI), defined as the period between the detachment of the flap from the donor site and the end of the vascular anastomosis, was 58.8 min (range 30-100 min).We observed one free flap failure (5%) and one pedicle flap failure (12.5%).These two cases were reconstructed with local rotational flaps.Eighteen patients (36%) had concomitant elective tracheostomy to prevent respiratory distress in the postoperative period.Six patients (12%) had dissection and identification of the trachea, without tracheotomy (pre-tracheostomy), to facilitate airway management in case of emergency.One patient (2%) had an emergency tracheostomy performed for postoperative bleeding.No one patient required a permanent tracheostomy.
Seventeen patients (34%) were admitted in the Intensive Care Unit (ICU) of our hospital after surgery (mean stay in ICU was 1.6 days).
The mean hospitalization time was 14.8 days (range 5-42 days).
According to Clavien-Dindo [26], we considered grade I and grade II as minor complications (n = 14, 28%) and grade III and IV as major complications (n = 8, 16%), see Table 4.At definitive histology, PM was confirmed in 37 patients (74%), in 10 patients (20%) the parotid parenchyma was involved by the neoplasm by direct infiltration, and in 3 cases (6%) the parotid was not involved by the tumor, see Table 5.
The primaries in the patients with PM (37 patients, 74% of the whole series) originated from the periauricular region in 15 cases (40.5%), forehead and temple in 7 cases (18.9%), scalp in 5 cases (13.5%), unknown in 3 cases (8.1%), cheek in 2 cases (5.4%), neck in 2 cases (5.4%), periorbital region in 2 cases (5.4%), and nose in 1 patient (2.7%).The definitive histology showed free margins (R0, >1 mm) in 38 cases (73.1%), close margins (R close, <1 mm) in 8 cases (15.4%), and positive margins (R+) in 6 cases (11.5%).An R close was observed on the T in one patient and on the P in seven patients.Three patients had an R+ on the T not amenable to surgical enlargement: in one case, the lesion was in the frontal region and the deep margin was represented by the frontal bone, while in two patients with parotid infiltration who underwent superficial parotidectomy, the deep margin was represented by the facial nerve.Three patients had an R+ on the N with infiltration of the neck structure not suitable for surgical resection.
After the multidisciplinary evaluation, 30 patients (60%) underwent adjuvant treatment: RT alone in 21 cases, and RT and concomitant CHT in 9 cases.Fifteen patients (30%) could not undergo adjuvant treatment despite indications because of the advanced age (n = 8), comorbidities (n = 5), and personal reasons (n = 2).Adjuvant therapies were not indicated in five patients (10%).
The mean time of follow-up was 2.6 years (range 0.7-10 years).Sixteen patients (34%) experienced a locoregional relapse.One patient (6.25%) underwent salvage neck dissection and is still alive after 16 months.Fifteen patients (93.7%) were not fit for aggressive salvage surgery or with unresectable disease and underwent palliative treatment (only one patient is still alive).Two patients (4%) experienced locoregional relapse and distant metastasis (one of them underwent immunotherapy and is still alive, while the other patient underwent palliative treatment and died of disease 16 months after the recurrence).Four patients (8%) experienced distant metastasis and died of disease.
Positive margins of resection (R+) were associated with a worse prognosis in the univariate analysis in all the patients of the present series (see Tables 6 and 7).The univariate analysis highlighted that the margins' status, FN infiltration, LVI, and G are prognostic risk factors in patients with parotid metastasis from SCC (see Tables 8 and 9).
The multivariate analysis of these risk factors showed that only positive margins of resection were associated with a significantly worse DSS (p = 0.001), as shown in Table 11.

Discussion
The majority of NMSCs are small superficial skin lesions, associated with a good prognosis, with a DSS at 5 years greater than 90% [30].Treatment modalities reported in the literature are surgical excision, cryotherapy, topical pharmacotherapy, laser, and radiotherapy [31].A small percentage of these patients (5%) develop advanced locoregional disease that requires more extensive surgical procedures and multi-modality treatments.
Patients treated for even early hnNMSC should always undergo close follow-up so as to detect the onset of parotid or neck metastasis.A parotid swelling observed within two years after the resection of a potentially metastatic hnNMSC should be considered as highly suspicious for metastasis as reported in the literature (PM occurs within 10 to 19 months of the treatment of the primary lesion) [32] and as observed in our series (8.9 months, range of 1-18 months).
Risk factors for advanced disease are tumor diameter, depth of invasions, histological grading, recurrent tumor, and immunosuppression [33]; these characteristics are often neglected at primary surgery.Lesions arising from the preauricular region have also been identified as a high risk for locoregional involvement [27].
A diagnostic work-up of advanced hnNMSC is based on imaging, histology, and cytology [34].MRI is generally considered more accurate in the diagnostic assessment of parotid lesions [35].In our series, most of the patients were evaluated by MRI (24%), and patients with highly suspicious intraparotid metastasis referred to our institution with a CT performed elsewhere did not undergo further imaging (60%) unless the parotid or temporal areas required a finer assessment of the deep extension (16%).
The histological diagnosis of lesions arising or involving the skin is confirmed by a biopsy (52% of our patients), while the diagnosis of metastasis is routinely made by FNAC (as observed in 10% of our patients).In our series, we did not consider a preoperative histological or cytological assessment necessary in 28% of patients with a clear clinical correlation between the parotid swelling and the resection of the primary (in 13 cases the first treatment was performed elsewhere and in 1 case in our center).As for parotid lymphoma [36], a standardized diagnostic work-up is still lacking for secondary parotid lesions, with ultrasound representing an emerging diagnostic tool.Recently, the administration of gas contrast medium has enhanced the accuracy of ultrasound in differentiating between benign and malignant lesions [37].
Parotidectomy is the gold standard in the case of hnNMSC metastatic to the parotid.SCC is the most common lesion that can metastasize to the parotid lymph nodes.MCC is an aggressive but uncommon lesion originating from neuroendocrine cells of the skin [38] with an incidence of nodal metastasis of 26% [39].Accordingly, with their incidence, in our series, 89.2% of the PM were from SCC and 8.1% from MCC. BCC rarely involves regional lymph nodes (from 0.003% to 0.1% as reported in the literature) [39].In our series, we observed only one case (2.7%) of PM from metatypical BCC, which is considered the most aggressive form of BCC in the literature [40].
Parotidectomy is also indicated in the case of close proximity or direct involvement of the gland from the primary lesion.In our series, 13 patients with periauricular NMSC (n = 12 SCC and n = 1 MCC) underwent wide resection with concomitant parotidectomy.In 10 of these patients (76.9%), the histology showed the gland's infiltration, while only in 3 cases (23.1%) the parotid was not involved.
Parotidectomies should be extended to the deep lobe of the gland only when it is macroscopically involved, because lymph nodes are rarely observed in the deep lobe [34, [41][42][43].
In our series, neck dissection was indicated in all cN+ patients with SCC and MCC.Because of the potential risk of occult metastasis, all the patients with PM but without clinical node involvement (cP+N0) of our series underwent neck dissection.The risk of occult neck disease observed in the literature in cP+N0 patients is 22.5% [44].This is similar to the rate of occult neck metastasis observed in our series of patients (21%).As a consequence, unilateral selective neck dissection (levels II-V) is indicated in these patients [45].Contralateral neck dissection is still not mandatory [23], and in our series, we never observed contralateral metastasis during the follow-up.
Elective neck dissection is also indicated in the case of NMSC directly infiltrating the parotid gland because the advanced T stage (at least T3) is associated with a higher risk of nodal metastasis [46].In our series, 4 out of 10 patients with parotid infiltration (40%) presented neck metastasis at definitive histology.
Sentinel node biopsy could be a treatment option in NMSC with a risk of occult nodal metastasis higher than 10% [47], but it was never indicated in our series of patients because of the advanced stage of disease (see Table 5).
Another major challenge in patients with hnNMSC with involvement of the parotid region is the management of the FN, because its sacrifice has a negative impact on quality of life.Direct nerve infiltration with preoperative palsy is a clear indication for nerve sacrifice [34], while in patients without preoperative nerve impairment, it can be difficult to predict when a partial or complete sacrifice of the nerve is necessary to achieve an oncologically sound resection [48].Conservative nerve-sparing approaches have been reported in the literature in the cases of intraoperative evidence of FN involvement with preoperative normal function, leaving microscopic residual disease involving the nerve, followed by timely postoperative RT [49,50].However, in such cases, postoperative unpredictable events that delay the RT could have a dramatic impact on oncologic outcomes.According to O'Brian et al. [16], our surgical strategy aimed to preserve the FN only when its function was clinically normal preoperatively and macroscopic clearance of the tumor could be achieved at the time of surgery.The dissection of the FN was performed with the aid of an intraoperative microscope or a 3D exoscope [34,43] to improve the intraoperative assessment of the FN neoplastic involvement and to avoid improper FN sacrifices (all the resected branches of the FN showed neoplastic involvement at definitive histology).An intraoperative macroscopic involvement of the nerve was observed in 36.6% of patients with preoperative normal FN function (15 out of 41 patients).In total, 17 patients (45.9%) with parotid metastasis required the sacrifice of the FN (12 patients underwent resection of at least one branch of the nerve and 5 patients underwent total nerve resection), followed by intraoperative neurorrhaphy (in the case of sacrifice of one or a few branches) or static rehabilitation of the facial palsy (in the case of resection of the main trunk).These data are similar to those reported in the literature [19,42,46].
In the case of extended head and neck resections, a reconstruction is generally required.In our series, most of the reconstructions were performed with microvascular free flaps (antero-lateral tight free flap in 70% of cases, forearm free flap in 25% of cases, and rectus abdominis free flap in 5% of cases).A pectoralis major pedicled flap was chosen only in four patients with a high risk for free flap failure (previous head and neck RT, severe chronic kidney disease, and an age-adjusted Charlson Comorbidity Index higher than nine), while the platysma pedicled flap was used in four patients with an age-adjusted Charlson Comorbidity Index higher than eight who underwent minor skin resection.Our success rate of free tissue transfer (95%) is in line with other studies from the literature [25].
The Rosenthal classification [25] showed a good correlation between the surgical defect and the modality of reconstruction in 89.3% of patients (see Table 3).A total of 3 out of 28 patients (10.7%) underwent a different reconstruction from what was indicated by the Rosenthal classification.Two patients with a type I defect (parotidectomy extended to the skin with preservation of the auricle) were reconstructed with an ALT flap because of the extensive resection of the check and neck skin, and one patient with a type III defect (parotidectomy extended to the ear and mastoidectomy) was reconstructed with a radial forearm free flap because of the limited surface of the area of resection.
According to Clavien-Dindo [26], we observed a minor complication rate of 28% and a major complication rate of 16% (see Table 4).The relatively high complication rate of our series is mainly due to the advanced mean age of our patients and the advanced stage of disease and consequent aggressive treatment required, as already reported in the literature [51].
According to the NCCN [23], adjuvant radiotherapy in SCC patients is indicated in the case of PNI, multifocal nerve invasion, recurrent tumor, a maximum diameter >6 cm, close margins when further surgery is not indicated, and desmoplastic or infiltrative tumors in immunosuppressed patients.In patients with MCC, the NCCN Guidelines [24] recommend RT in the case of a maximum diameter >1 cm, LVI, chronic T-cell immunosuppression, HIV, chronic lymphocytic leukemia, solid organ transplant, and a head and neck primary site.The NCCN Guidelines [23] do not recommend adjuvant CHT for most cases of fully resected regional SCC.Adjuvant CHT is used in a clinical trial with advanced locoregional SCC, but the supporting data are limited [23].There is no indication for adjuvant CHT in fully resected primary MCC [24].CHT is indicated in the case of recurrent disease [24].
We compared the OS, DSS, LRFS, and DMFS of 10 patients with direct parotid infiltration by NMSC with those of 37 patients with PM from NMSC, and we observed similar outcomes.
In the 8th edition of the TNM [1], the side, size, number of metastatic lymph nodes, and ENE are considered to differentiate the N1, N2, and N3 categories.According to the literature [11,46,53], in our study, N+ patients experienced a worse but not statistically significant prognosis compared to N0 patients, as shown in Table 7.
There is an emerging interest in the literature about the different prognostic behaviors between parotid (P) and neck (N) metastasis from NMSC [1,23,54].
O'Brien et al. in 2002 [16] described a classification that differentiates the parotid lymph node (P) from the neck lymph node (N).In 2003, Carsten et al. [55] observed a good statistical correlation between the extent of the parotid disease on the basis of the O'Brien classification and survival outcomes in patients with intraparotid metastasis from SCC.In our series, the O'Brien [16] classification showed a better correlation in the prediction of LRFS than the TNM [1] classification (see Table 8).
In 2010, Forest et al. [17] described a new classification (N1S3) in which the number and dimensions of neck and parotid metastases differentiate the N categories: stage I single metastasis measuring ≤3 cm; stage II single metastasis >3 cm or multiple measuring ≤3 cm; and stage III multiple metastases measuring >3 cm.In our series, the N1S3 classification showed a better correlation with DSS and LRFS than the other classifications (see Table 8), as observed by other authors [56].
We did not observe differences in DSS, LRFS, and DMFS between the patients with PM and the patients with neck metastasis from SCC (see Tables 8 and 9).
According to the literature [57,58], our analysis showed that the margins of resection, FN infiltration, LVI, and grading (G3-G4) were associated with worse outcomes in the patients with PM from SCC (see Tables 8 and 9).
Free margins of resection were associated with better outcomes than close or positive margins (DSS 65.7% versus 23.6%, p = 0.008; LRFS 72.7% versus 12.5%, p = 0.005; and DMFS 86.9% versus 52.5%, p = 0.01).The patients with close margins (<1 mm) experienced better outcomes than the patients with positive margins (DSS 40% versus 0%, p = 0.001; LRFS 20% versus 0%, p = 0.002).Our multivariate analysis confirmed that an R+ was associated with a worse DSS (p = 0.0011, see Table 11).These data show that radical resection is mandatory to improve outcomes.Unfortunately, the complex anatomy and the cosmetic implications of the head and neck are associated with the high incidence of close and positive margins of resection reported in the literature [27].
Myers et al. [46] reported that FN infiltration was associated with a reduced DSS and Shao et al. [59] reported a reduced OS.In our series, FN infiltration was associated with worse LRFS (80.1% versus 39.1%, p = 0.04), especially in the case of resection with close or positive margins, although without statistical significance (LRFS 75% versus 16.7%, p = 0.1).These data support the indication of the resection of the FN because a surgical approach aiming to its preservation at the cost of leaving macroscopic neoplasm may worsen the prognosis.
LVI is a well-identified risk factor in patients with head and neck SCC [29,57].In the univariate analysis, LVI was associated with a worse prognosis (DSS 68.2% versus 30%, p = 0.04; DMFS 84.2% versus 53.3%, p = 0.04).Other authors observed that LVI is associated with a worse DSS [11,46,52] and, as a consequence, adjuvant CHT could be indicated in these cases.
In the literature, a high-grade tumor (G3-G4) is reported as a negative prognostic factor in cutaneous SCC [57].In the univariate analysis, we also observed that the high-grade tumor (G3-G4) of the patients with PM from SCC was associated with a poorer outcome (DSS 84.4% versus 37.5%, p = 0.04), supporting the indication of adjuvant therapies in the case of poorly differentiated and undifferentiated SCC.
Patients with ENE+ experienced a worse although not statistically significant prognosis, as reported by other authors [11,52,53].
We did not observe a correlation between the size of the PM and the prognosis (see Tables 8 and 9).We considered the 3 cm cut-off as reported by the TNM classification system [1] and by O'Brien [16] and Forest [17].Laxague et al. [52] reported a worse DSS in patients with parotid metastasis > 5 cm 3 , but none of our patients had metastasis of this size.
Other authors associated PNI with worse outcomes [11,46], but in our series, we did not find any correlation.
Immunodepression is an important risk and prognostic factor reported in the literature [59,60].In our cohort of patients, only three patients (6%, n = 2 transplant and n = 1 chronic hematological malignancies) were considered immunosuppressed according to the NMSC United Kingdom Guideline [60].In these patients, the univariate analysis was not possible because of the low number of cases.Two patients died of other causes after 1 and 6 months of the treatment.One patient is still alive after 3.6 years despite a second primary SCC of the oropharynx treated with surgery.
Adjuvant RT or CHT can improve locoregional control and survival outcomes in selected patients with PM of NMSC [9].Some authors [11,53,61,62] observed that multimodality treatments are associated with better outcomes, but our results did not show benefits from RT and CHT (see Table 10).Probably, the small number of patients and the high comorbidity index (mean Charlson Comorbidity Index was 6.7) of our series could be a relevant bias in the analysis of the effectiveness of adjuvant treatments.
Immunotherapy has been recently introduced in patients with unresectable or metastatic SCC, but further studies are required to evaluate the real effectiveness [63,64].
Regardless of the oncological outcome, extensive surgical resection inevitably results in a reduction in patients' quality of life.As a consequence, in addition to survival, immunotherapy will also have to compete with surgery in terms of residual quality of life.
Our study shows several limitations.Primarily, it is a monocentric study based on a review of medical charts.The statistical analysis has limited power due to the small sample size.Additionally, most of our patients were observed after the resection of the primary tumor performed in another institution, which means that data about the primary lesion are lacking.Furthermore, our population was associated with a high prevalence of comorbidities and advanced age.

Conclusions
Our results confirm that the involvement of the parotid gland by metastasis or direct infiltration of NMSC is associated with a decreased prognosis.
Our finding leads to two important conclusions.Firstly, it is imperative to perform a preoperative imaging examination to rule out any early involvement of the parotid or neck lymph nodes in patients diagnosed with hnNMSC.Secondly, a rigorous follow-up program should be planned for patients treated for hnNMSC, even in cases of early stage disease, mostly for lesions originating from the ear, temple, and anterior scalp.
Because the resection in free margins is the most important prognostic factor, early diagnosis, extensive surgical resection, and concomitant reconstruction are required.Subsequent adjuvant therapy is indicated in selected cases.
We found that N1S3 was the more accurate system to predict the prognosis in patients with parotid metastasis from SCC.In these patients, positive margins of resection, FN infiltration, LVI, and high grade (G3-G4) represent negative prognostic factors.

Table 1 .
Demographics and clinical data of patients who underwent parotidectomy for NMSC.

Table 3 .
Reconstruction according to Rosenthal's classification for temporal region defect.

Table 7 .
Hazard ratios, p-value, and 95% confidence intervals for univariate analysis of margin status; significative values are highlighted in bold.

Table 9 .
Hazard ratios, p-value, and 95% confidence intervals for univariate analysis of patients with intraparotid metastases from SCC; significative values are highlighted in bold.

Table 10 .
Univariate analysis of adjuvant treatment outcomes.

Table 11 .
Multivariate analysis of patients with SCC intraparotid metastasis.# Value not reported by the program.Level of significance is reported as ** (p < 0.01).