Tumour Recurrence, Depth of Invasion, and Temple Location as Independent Prognostic Parameters of Lymph Node Metastases of Head and Neck Cutaneous Squamous Cell Carcinomas

The excellent survival rate of cutaneous squamous cell carcinoma (cSCC) exceeding 90% is reduced by the presence of nodal metastases by over 50%. We analysed various risk parameters of cSCC to predict the incidence of nodal metastases. A total of 118 patients with the head cSCC were included in a single-institution retrospective study covering the period from 2008 to 2020. Tumour recurrence, temple location, and tumour infiltration depth were found to be independent predictors of nodal metastases (increasing the probability of metastases by 8.0, 8.1, and 4.3 times, respectively). Furthermore, univariate analysis shows that the tumour size and T stage are significant factors increasing the risk of metastases. Several independent risk factors for the development of metastases in the head cSCC have been confirmed. These findings might help identify at-risk patients who require additional attention for adequate radical treatment and close follow-up. In contrast, elective treatment of lymph nodes is not recommended due to the low incidence of regional metastases.


Introduction
Basal cell carcinomas primarily represent nonmelanoma skin cancers and only one third of this oncological group comprises squamous cell carcinomas (SCC) [1,2].Teir incidence has increased signifcantly in the last three decades [3][4][5].Te reason for this is the increasing age of the population, higher UV radiation exposure due to ozone depletion [4,5], and a lifestyle with more outdoor activities.
Identifying predictors of a higher risk of local recurrence and metastases has implications for tumour staging.
Individual authors have identifed several tumour and patient characteristics that increase the risk of metastatic incidence.Since 2009, the TNM classifcation system has been adjusted to account for the presence of the following two or more additional risk factors: >4 mm depth, Clark level V, perineural invasion, angioinvasion, localisation on the ear or lip, and poorly diferentiated or undiferentiated tumours.Te presence of these risk factors upstages the T value [29,30].
However, the results of various independent studies have not always been consistent, and the predictive values of the factors tested and their combinations have not been strong enough to serve as independent prognostic markers.
In our retrospective study, we analysed a group of 118 patients with cSCC of the face, auricular, and frontotemporal regions to identify predictors of nodal metastases and patient prognosis.

Patient and Methods
Database.A total of 118 patients with cSCC were included in the retrospective study covering the period from 2008 to 2020.Of these patients, 78 were already treated with the primary cSCC at our clinic, and the remaining 40 were referred from other institutions due to disease recurrence involving regional lymph nodes (22 cases), the site of the primary tumour (5 cases), or both (13 cases).
Te clinical stage of the disease was determined in all patients before initiating their treatment.Te skin tumour was evaluated by a dermatologist, and cervical and parotid nodes were examined clinically using imaging methods (US, CT, MRI, and PET/CT).Te histopathological diagnosis of squamous cell carcinoma was verifed postoperatively by examining surgical specimens or biopsies of inoperable tumours.Te lymph node status was assessed in clinically evident nodal metastases using aspiration cytology (only in cases where the diagnosis was later confrmed through histology after surgical treatment) or core needle biopsy.
In our study, we included only those risk factors listed in the NCCN recommendation [24] for which we had a sufcient amount of relevant data necessary for valid statistical analysis, i.e., the clinical stage of the tumour, tumour size (measured in millimetres at the greatest diameter) and depth of its invasion (measured in millimetres) or invasion into the subcutaneous layer, grade of diferentiation (G1-G3), perineural invasion, and resection margins (R).In Rx and R1 cases, the completeness of resection, i.e., R, was assessed through a histological review of the surgical specimen harvested during revision procedures performed immediately after the primary operation.Te predictive power of the R parameter was evaluated only in those 35 cases where metastases appeared as a relapse.R0 tumours were categorised based on resection margins using a 4-mm cutof.Other factors (angioinvasion, perineural spread, and similar) were not analysed.
Within the entire set of 118 tumours, 76 were evaluated as cN0.Tis group consisted of 69 men and 7 women, aged 41-92 years (average of 74 years).Te tumours were located on the skin of the auricle in 50 patients, on the external auditory canal in 7 patients, frontotemporally in nine patients and on the nose in 10 patients (Figure 2).Te followup period ranged from 3 to 192 months (average of 48 months).
Metastatic nodal involvement was found in 42 patients (34 men and 8 women) aged 41-95 years (average of 79 years).Only three cases showed involvement of cervical nodes (the areas I-III), 24 cases involved only the parotid gland, and 15 cases involved both localities (5 times in level II, 3 times in levels II and III, once in levels II and V, and 6 times in 4 and more neck levels) (Figure 2).
Te stage of regional nodal metastases was as follows: cN1 in two patients, cN2 in fve patients, and cN3b, all with the extracapsular extension, in 35 patients.
In 35 patients, nodal metastases occurred within the range of 3-24 months (average of 8 months), and 10 of these cases also presented with local tumour recurrence (Figure 3).In 7 cases, nodal metastases were already present during the clinical presentation of the primary tumour.Te patients were followed up for 2-125 months (an average of 31 months).Te primary tumour was located frontotemporally in 21 patients, on the auricle in 15 patients, in the external auditory canal in 3 patients, and on the external nose, neck, and lower lip in 1 patient each (Figure 2).
Clinical and demographic data of all the patients, including the tumour stage, tumour size and depth of invasion, histopathological grading, quality of resection margins, perineural invasion, other skin pathology (keratic acanthosis), and immunosuppression are presented in Table 1 (Table 1).

Statistical
Evaluation.Tis study compared 5-year overall and disease-free survival in subgroups of patients with and without nodal metastases.To predict nodal metastases of primary tumours, we assessed the importance of the following fve parameters: T stage according to the current WHO TNM classifcation [27], tumour size, depth of skin invasion (greater than or equal to 4 mm), histological diferentiation (G1/2 vs. G3/4), microscopic resection margins (negative), and local recurrences (more than three months after remission).Negative resection margins were determined based on the absence of microscopic evidence of malignant cells, with close resection margins defned as equal to or smaller than 4 mm.
Fisher's exact test was used to compare the incidence of risk factors in patients with metastases and those with nonmetastatic tumours.Te study also utilised Kaplan-Meier analysis with the log-rank test to compare OS and disease-free interval (DFI) between the two groups and to evaluate the impact of risk factors on OS and DFI.

International Journal of Clinical Practice
Te signifcance of individual parameters for lymphogenic spread was evaluated using univariate analysis.An algorithm was defned using Cox stepwise logistic regression analysis to determine the predictive ability of signifcant parameters to calculate the risk of nodal metastases in individual tumours.Due to the patient group size (n � 118), only the three most signifcant parameters from the univariate analysis were evaluated.
IBM SPSS Statistics version 23 (Armonk, NY: IBM Corp.) was used for data analysis.All tests were conducted at a signifcance level of 0.05.

Treatment of cN0
Patients.In two patients, the primary tumour was treated with radiotherapy.Furthermore, 76 patients underwent surgery, with 51 achieving adequate resection margins and 25 having close or positive margins.Five patients with positive or close margins underwent reresection (three patients) or radiotherapy (two patients).Local relapse occurred in six patients, and dissemination occurred in one patient.No nodal recurrence was recorded.

Treatment of cN +
Patients.Among 42 patients with nodal metastases, the primary tumour was treated with radiotherapy in six cases.Surgical resection was performed on 36 patients, 21 with insufcient resection margins.Consequently, 20 patients underwent reoperation, and two received adjuvant radiotherapy.Te remaining patient underwent radiotherapy.Of the three patients with solely cervical metastases, one underwent comprehensive neck dissection, another patient underwent selective (II-IV) neck dissection, and the remaining patient received radiotherapy.On 17 out of 24 patients with isolated parotid metastases, near-total or total parotidectomy was performed along with elective neck dissection of levels II-IV.Microscopic examination of all patients' specimens revealed no evidence of subsequent metastases.Te remaining seven patients received only palliative radiotherapy.On 9 out of 15 patients with concurrent parotid and cervical metastases, parotidectomy and curative neck dissection were performed (3x comprehensive and 6x selective).Adjuvant radiotherapy was performed on 12 patients.Primary radiotherapy (5x) or symptomatic treatment (1x) was indicated due to poor general condition in six patients.Remission was achieved in 23 out of 26 patients who received curative treatment for metastatic disease (i.e., 9x surgery, 17x surgery with adjuvant radiotherapy (RT), or chemoradiotherapy (ChRT)).Eight patients experienced relapse (six in the parotid, one in the cervical nodes, and one with distant metastasis).
No diference was observed when comparing early stages (N1 + 2) to advanced stages (e.g., N3b) with the extracapsular spread (p � 0.727).However, there was an obvious imbalance in patient distribution between the two groups, with the most advanced stage of nodal metastases with the extracapsular spread being predominant (35 patients).In contrast, early stages (N1-2) were diagnosed in only seven patients.
No signifcant diference was found in terms of histopathological grading of cSCC in both groups (p � 0.118), presence of actinic keratosis (p � 0.564), or incidence of multiple previous skin malignant tumours (both basaliomas and spinaliomas) in other locations of the head and neck (p � 0.061), although the latter parameter was nearly signifcant (Table 2).1.62-10.6),close and positive resection margins (p � 0.026, OR: 2.62), and perineural invasion (p � 0.038, OR: 10.13, and CI: 1.14-89.9).Te statistical signifcance of the T stage and tumour size was the most substantial (p � 0.0001 and p � 0.003) for auricular tumours.Te importance of histological grading of cancer was suggestive but it did not reach a signifcant level (p � 0.121, OR: 1.93, and CI: 0.84-4.42)(Table 3).
Te negative predictive value (NPV) and the positive predictive value (PPV) of the signifcant risk parameters for the prediction of metastases are summarised in Table 5.

Aetiology and Diagnostics of cSCC.
Cutaneous SCC arises from actinic keratosis (AK) through an uncontrolled proliferation in a long-term, multistep process.Typical histological features of SCC include parakeratosis, hyperkeratosis, nuclear pleomorphism, atypical mitoses, and multinucleated cells through all epidermal layers [49,50].
Te diagnosis of cSCC must always be established histologically through biopsy or excision.Depending on the tumour size, biopsy may be performed before radical extensive surgical treatment of a large tumour.International Journal of Clinical Practice Histological examination with H&E staining is used to confrm the diagnosis; immunohistochemical testing is added only in rare cases of uncertain diagnosis.Various histological variants of cSCC with diferent prognoses and metastatic incidences can be distinguished.Desmoplastic, acantholytic, and adenosquamous variants of SCC are known to have worse prognosis with highly infltrative growth, perineural and perivascular spread, and a high incidence of metastases [42,51,52].

International Journal of Clinical Practice
However, as these types were rarely identifed in our set of metastatic tumours, their signifcance was not analysed.
SCC is usually diagnosed based on its typical appearance; dermoscopy can be helpful for small lesions that may be confused with keratoacanthoma.Bowen disease is another specifc entity, an intraepidermal carcinoma with atypia of keratinocytes at all levels of the epidermis (carcinoma in situ).Te early tumour stage can be missed if the tumour develops within a chronically infamed area accompanied by pseudoepitheliomatous hyperplasia in the surrounding tissue [53].
In our study, the histology of the skin tumour was confrmed in all cases by examining the resection specimen.A pretreatment diagnostic biopsy was performed in only a few advanced cases.Parotid and neck nodal metastases were always diagnosed by standard HE histological examination, which was performed either on tissue samples retrieved after open biopsy or through preoperatively obtained core cuts, and the diagnosis was subsequently confrmed postoperatively.

Incidence, Location, and Prognostic Impact of Regional
Metastases.Te reported incidence of lymph node metastases in cSCC across all tumour sites is generally very low.In recent population studies, Veneblas [47] reported a 3-year cumulative incidence of metastases of 2% in a cohort of 76,977 patients, and Tokezs reported a 10-year cumulative incidence of only 1.9% in a study of 11,137 patients [48].However, in head and neck cSCC, the risk of metastases is usually higher, up to 6% [8,11,16,26,35].Te incidence reported in independent studies on a limited cohort of patients can vary widely, ranging from 10% to 21% [8,9,11,35].
In our study, metastases were present in 35.6% (42/118) cases.Tis high incidence compared to the previously published studies is probably due to selection bias.Our cohort of patients is not population based but reports on the results of a tertiary ENT centre to which most patients with cN+, advanced tumours, or otherwise complicated patients are referred.

Lymphatic Spread.
When assessing the anatomical localisation of lymph node metastases in cSCC, it is important to note that the pattern of lymph drainage can vary.As described by several authors, for most head skin locations (e.g., auricular, temporal, zygomatic, frontal, and anterior scalp), the frst lymph node basin is the parotid and periparotid regions.
Te involvement of the neck lymph nodes typically follows the parotid region.In addition, the external jugular nodes can be involved in either the parotid or II neck level, which is signifcant [12,[21][22][23]36].
Our results supported the fndings of these authors.In our set of cases, 39 out of 42 showed metastases to the parotid gland, whereas only three patients had metastases restricted to the neck lymph nodes, with clinically negative involvement of the parotid gland.Te primary skin tumours in these cases were located on the skin of the mental region, the apex of the external nose, and the skin of the lateral neck.
However, in 24 out of 39 cases, the metastases were confned to the parotid nodes alone, whereas parotid and neck lymph node metastases were present in only 15 cases.Regarding the neck levels, the metastases were most frequently located in the upper jugular nodes (level II), followed by the middle jugular nodes (level III).(Figure 1).
Te extracapsular extension (ECE) of nodal metastases is generally considered a negative prognostic predictor, which has been confrmed in several clinical studies using univariate analysis [23,37].
Our results did not confrm this assumption, as the univariate analysis did not reveal a signifcant association between ECE and worse prognosis.An uneven distribution between the groups may have infuenced the statistical results, as ECE was confrmed in most cases (35 out of 42).

Tumour Localisation.
In general, SCCs located on the skin of the head and neck area are more likely to metastasise compared to those on the trunk and extremities.Tis risk also varies depending on the specifc skin locations of the head, of which the ear and periauricular, temporal regions, lips, and vermilion have the highest risk of metastases.
When comparing tumours located in diferent areas, those on the skin of the temporal area had a higher risk of metastases in our cohort.Furthermore, according to regression analysis, the localisation of tumours in the temporal area was confrmed as an independent prognostic factor that increases the risk of metastases 8-fold; these fndings are consistent with independent studies by Haisma and Brougham [8,16] as well as a meta-analysis by Tompson [10].
However, our results did not confrm the previously observed high risk of metastases of tumours localised on the skin of the ear.We believe that this discrepancy is due to a tumour size bias.Te frequency of small T1 tumours was signifcantly higher on the skin of the auricle than in any other location (p � 0.0001) [8,10,16,26,31,35,44] 4.5.Tumour Stage.Our study confrmed the signifcant importance of the T stage, specifcally the tumour size, as an independent predictor of nodal metastases [9-12, 26, 28, 30-34, 44].
According to our results, any stage higher than T1 is considered signifcant.Te association between the T stage and nodal metastases was demonstrated in previous metaanalyses by Rowe [26] and, more recently, by Zeng [28], who evaluated 29,000 skin cSCC of the head and neck from 43 studies.A meta-analysis by Tompson [10] also obtained similar results in skin cancers of the head and neck regions.

8
International Journal of Clinical Practice In our study, the risk of developing lymph node metastases in T2 and T3/T4 carcinomas was signifcantly increased compared with T1 tumours.4.6.Tumour Size.Tumour size, as one of the parameters included in the TNM staging system, is considered a risk factor related to the occurrence of lymph node metastases.Several studies have already confrmed this assumption [8,26,32,35].Te critical tumour size estimated by several authors [8,11,26] and also accepted by the NCCN panel is 20 mm for all skin sites [24].
In head and neck cSCCs, even smaller tumours are considered high risk; the critical diameter was assumed to be 15 mm by Quaedvlieg and 6 mm by Bratsch [32,35].Our study confrmed that any tumour exceeding 10 mm in its greatest diameter was associated with a signifcantly higher risk of developing nodal metastases.

Tumour Recurrence.
Recurrence is strongly associated with the risk of metastases.Cox regression as an independent predictor confrmed this relationship's strength, which increased the risk of metastases by 7.6 times.Numerous authors have extensively documented similar results supporting a strong association between tumour recurrence and the development of metastases [9,11,34].

Depth of Tumour Invasion.
Our study demonstrated that a depth of tumour invasion above 4 mm and local recurrence are independent factors associated with nodal metastases, increasing the risk by 4.3 and 7.4 times, respectively.Tese fndings correspond to previously published studies [8-11, 16, 26, 27, 31, 35].
In their meta-analyses, Tompson [10], Haisma [8], and Cherpelis [11] stated that the risk of regional lymph node involvement increases with the depth of invasion, starting from 2 mm and beyond.Tis fnding is consistent with the results of Brantsch's analysis [35], which indicates that tumours (regardless of the T stage) with an invasion of 2 mm do not establish nodal metastases, while tumours infltrating to a depth of 2-6 mm and above 6 mm metastasise in 4% and 16%, respectively.Our study confrms that the depth of invasion is a crucial factor in nodal metastases, even in earlystage (T1 and T2) cancers.
Te depth of tumour invasion was also classifed by Clark by the infltrated layer of the dermis or subdermis.Te risk of metastases increases substantially with the invasion of cartilage, subcutaneous fat, or bony structures.
Univariate analysis confrmed the expected relationship at a signifcant level (p � 0.003) although the factor was not estimated as an independent predictor.Numerous previous studies have also shown a positive correlation between the involvement of subcutaneous layers and the risk of metastases [9,11,12,30,32,34].

Histological Diferentiation.
A low degree of histopathological diferentiation was found to be signifcantly associated with metastases in T1 and T2 tumours (p � 0.016) in the univariate analysis but not in the multivariate analysis.In more advanced cancers (T3/T4), there was only an indication of a higher risk of lymph node spread.However, the results of published studies on this subject are controversial.While Clark [12], similar to our fndings, did not confrm the association between the grade of diferentiation and the occurrence of nodal metastases, other authors have demonstrated it at statistically signifcant levels [8, 9, 16, 22, 26, 30-32, 34, 44].
Tese diferences might be explained by the grading heterogeneity in large volumes of advanced tumours, which can contain areas with varying degrees of diferentiation.Standards for determining grading in these carcinomas need to be provided.
In general, poor diferentiation (G3) signifcantly increases the risk of metastases.However, an association for moderate grade (G2) has also been reported [8].

Actinic Keratosis, Prior Radiotherapy Site, Chronic
Infammation.AK is a nontumorous pathology that can progress into invasive cutaneous squamous cell carcinoma.Te progression of AK to SCC is unpredictable, and SCC may develop even without specifc signs of lesion progression [43].
AK, as well as chronic scarring, previously irradiated terrain, or infammation, increases the risk of metastases of cSCC that developed in these conditions.
We confrmed this association by observing a higher frequency of these pathologies in patients with metastatic tumours.Te frequency of patients with multiple previous cSCCs was almost signifcantly higher in the cN + SCC group [11,26,32,43,54,55].
Tere is a particularly high risk associated with the involvement of a nerve located deeper than the dermis or measuring ≥0.1 mm [46,56].
Our results confrmed this observation, showing a signifcant association between PNI and the incidence of metastases (p � 0.038) in the univariate analysis.However, due to the low frequency of PNI (12% in cN vs. 1.3% in cN0, p � 0.021), this factor was not included in the regression analysis.
Te involvement of lymphatic or blood vessels is another parameter in skin tumours that plays a role in the development of metastases [44,46,56].However, this factor was diagnosed in exceptional cases among our patients with metastases and could not be analysed.4.12.Immunosuppression.From the patient's perspective, immunosuppression worsens the prognosis and increases the incidence of metastases.Nowadays, immunosuppression, International Journal of Clinical Practice mostly of iatrogenic origin in transplant recipients, is becoming an important factor [10,26,30,47,48,57]. Nevertheless, our study included only one immunosuppressed patient in each group (cN0 and cN+), which limits further statistical analysis.

Resection Margin.
Univariate analysis showed that the positivity of microscopic resection margins was a signifcant predictor of nodal metastases in cSCC of the head and neck.Other authors [36,38] have obtained similar results for tumours in this and other locations.According to Mourouzis, this parameter is an independent predictor [39].
We did not fnd a relationship between close margins (less than 4 mm wide horizontal resection) and the incidence of nodal metastases.Grifths [58] and Quaedvlieg [32] have also reached the same conclusions, considering the vertical dimension, i.e., the bed of resection, separately.According to Jenkins [40] and Stratigos [41], tumours with a resection margin of less than 2 mm tend to metastasise to lymph nodes, making this a critical margin.
Distribution and frequency of 76 nonmetastatic (green) and 42 metastatic (red) primary cSCC in the frontotemporal, auricular region, external ear canal and face, and regional nodal metastases (blue) of 42 patients in superfcial and deep parotid lobes and neck regions were found (Table 6).

Conclusion
Te tumour recurrence, temple location, depth of invasion, size, and T stage were found to be the most relevant independent predictors of lymph node metastases in cSCC of the head and neck.Elective lymph node dissections might improve the poor prognosis of metastatic cancer patients.However, the predictive values of the clinical and histological parameters assessed are insufcient to justify elective surgery.We suggest that a more accurate examination of molecular markers should be performed to improve predictive accuracy.
Table 6: List of cSCC risk factors and their signifcance for incidence of metastases in our patients and in previously published studies.

Lymphatic invasion
Not evaluated Moore [34] 10 International Journal of Clinical Practice Zuzana Horakova, Ivo Starek, and Jana Zapletalova prepared the material and collected and analysed the data.Zuzana Horakova wrote the frst draft of the manuscript, and Richard Salzman commented on previous versions of the manuscript.

Figure 1 :
Figure 1: Flow diagram of tumour and patient risk parameters increasing the risk of metastases.

Figure 2 :
Figure 2: Distribution and frequency of primary skin carcinomas and their regional metastases.

Figure 3 :
Figure 3: Time between skin tumour excision and nodal metastases manifestation.

Figure 4 :
Figure 4: Overall survival prediction (a) and disease-free interval prediction (b) (Kaplan-Meier) in groups of patients with and without nodal metastases (cSCC N+ and cSCC N0).

Table 1 :
Characteristics of a patient group and tumour risk parameters.

Table 2 :
Diferences between the cSCC N0 and cSCC N+ groups of patients.

Table 4 :
Stepwise logistic regression of cSCC risk factors for nodal metastases.

Table 3 :
Univariate logistic regression: nonadjusted OR of cSCC risk factors for nodal metastases.

Table 5 :
Positive predictive values (PPVs) and negative predictive values (NPVs) of cSCC risk factors for nodal metastases.