This study included patients with histologically saved ecSCC and an patient age of more than 18 years. Localizations of ecSCC were: retroauricular/posterior side, helix/lobules, cavum concae/anthelix/tragus and more than one region. All patients had a preoperative staging. All patients received staging according to the German guidelines for cutaneous squamous cell carcinoma (cSCC). For T1 / stage I this was ultrasound of the head an neck lymph drainage area. From T2 / stage II on all patients received CT of head and neck, CT thorax, ultrasound of the head and neck area and sonography of the abdomen.

Initial, primary therapy of all patients was surgery after case discussion in an interdisciplinary tumor board. In this study (matched pairs study population!) non-surgical cases were not included. Adjuvant radiotherapy was administered after tumor board decision according to guideline in cases with positive nodes (N +), close margin or R1/R2, perineural growth (Pn1) or invasion into blood vessels or lymph vessel (V1, L1).

Follow up was every 3 months in the first two years including ultrasound of the head an neck and CT or MRI scan once a year. For year 3 to 5 a follow-up interval of 6 months was chosen (also according to German guidelines).

Patients were excluded from analysis with incomplete data sets and other head and neck cancers. Tumor characteristics and follow-up data were collected retrospectively from the authors’ institutional database.

The IHC implementation and analysis were performed at the dermatohistopathologic laboratory of the Fachklinik Hornheide (Muenster, Germany). This study method oriented towards the study of M. Klein et al. [11].

For the first overview on tumor and peritumoral environment every tissue sample (primary tumor, no biopsy) was analyzed with HE staining. The used microscope was the Olympus BX51 microscope (Hamburg, Germany) with magnification of 400 × . All pictures were made with the Olympus UC30 microscope camera (Hamburg, Germany) and the Olympus cell sense entry programme (Hamburg, Germany) was used.

Formalin Fixed and Paraffin embed (FFPE) ecSCC were cut in slices (4 μm) and evacuated on coated slides. Reaction with primary antibody were performed in the Autostainer Plus (Dako REAL DETECTION SYSTEM K5005, Glostrup, Denmark). Table 1 gives an overview of the primary antibodies used and their application.

The tumor slices were then reacted (exposure time 15 min) with a secondary antibody (REAL Link Biotinylated secondary antibody (AB2)). After that the tissue slices were incubated (exposure time 15 min) with Dako Real Streptavidin Alkaline Phosphatase (AP) (Glostrup, Denmark) and reacted (8 min) with chromogen (Dako RED Chromogen, Glostrup, Denmark). As a nucleus counterstain all slices were colored with Haematoxylin (exposure time 8 min).

All slices were dehydrated with increasing alcohol concentrations (70%, 90% and 100%), processed with xylol. All colored slices were covered with Coverslip tape.

The analysis were performed by two independent investigators. Anti-CD4, anti-CD8, anti-FOXP3 and anti-PD-1 were analyzed in a semiquantitative analysis mainly in the subepithelial compartment in a hot spot analysis. Each cut was counted in five high power fields with 50 cells each, totalling 250 cells.

Formalin Fixed and Paraffin embed (FFPE) ecSCC were cut in slides (4 μm) and evacuated on coated slides. The sections were then pretreated using the pretreatment system PT link Dako (Glostrup, Denmark) at a pH of 6 and subsequently stained with the PD-L1 panel (PD-L1 ICH 22C3 pharmDx, Agilent/Dako, Glostrup, Denmark) in the Autostainer link 48 (Agilent/Dako, Glostrup, Denmark). Finally, the material was covered with a coverslip.

In addition, a color control was performed. Skin was used as a negative control and placenta as a positive control.

For analysis of anti-PD-L1 the recommendation of the PD-L1 IHC 22C3 pharmCx Interpretation Manual (Agilent/Dako, Glostrup Denmark) was used.

Two independent investigators analyzed the tissue samples with the tumor proportion score (TPS). PD-L1 expression was categorized into three groups. First no expression = 0% (0 pts.); second positive weak to moderate = 1% to 49% (1 pt.); and third positive strong with expression of ≥ 50% (2 pts.). Besides, expression and intensity of PD-L1 staining were combined to generate a PD-L1 score summing up the above-given points. The PD-L1 score therefore had a theoretical range from 0 to 5 points.

For PD-L1, intensity of staining was analyzed in an ordinal scale: none (0 points [pts.]), weak (1 pt.), moderate (2 pts.) and strong (3 pts.) colouring.

All statistical analyses were conducted by a statistician using the Statistical Package for Social Sciences (SPSS) version 22.0 for Windows® (SPSS Inc., Chicago, Illinois, USA).

Data were tested for normal distribution with the Kolmogorov–Smirnov test. The t test was used for normally distributed variables. For categorical variables, Fisher's exact test and the chi-square test were applied. For metric parameters, the Kruskal–Wallis H test or Mann–Whitney U test were used as non-parametric tests for not normally distributed data. All tests were two-sided. Survival time periods (time from first diagnosis until event; data on patients without event were censored at the last follow-up) were calculated using the Kaplan–Meier method, and group differences were analyzed using the log-rank test.

A special focus of our study was set on the predictive testing of the markers (CD4, CD8, FoxP3 PD-1, PD-L1) for the risk of LNM. The matched pair approach is suitable for this purpose. The matched pair approach can minimise the influence of known risk factors. For this purpose, two homogeneous groups were formed. In one group were patients with LNM + (n = 32) and in the other group patients with LNM- (n = 32). The groups were matched with the following parameters with the smallest possible differences: gender, age, immunosuppression, comorbidities, primary localization, T-stage, tumor thickness, grading and perineural growth (Pn). Comparability and homogeneity of this stratification were checked statistically using matched pairs analysis.

Randomization and blinding were used for the IHC analysis. The researcher analyzing the IHC staining did not know which group the patient belonged to.

The subgroups were determined by dividing the patients according to primary tumor location, American Joint Committee on Cancer stage (AJCC stage) and lymph node metastasis.