The study included 107 skin specimens from 98 patients (42 female, 56 male) with skin tumors or nevi, and 9 anonymous donors who provided healthy skin samples as controls. Tumor specimens included AK (n = 18), SCC (n = 23), BCC (n = 19), and primary melanoma (n = 22). To compare malignant and benign melanocytic lesions, an additional set of nevi (n = 16) was analyzed. The mean patient age was 61.4 years (range 19–92 years), and the head and neck region was the most frequently affected site (41% of cases). Detailed demographic and clinicopathological data, including tumor thickness (Breslow depth for melanoma) and histological subtypes, are summarized in Supplementary Tables 1–5.

To analyze the distribution of CD1a+ DC and CD3+ T cells, four compartments were defined for each entity: intratumoral/intralesional, tumor margin/lesion margin, intraepidermal, and intradermal (Fig. 1). For each sample, five to eight representative images per region were acquired using a fluorescence microscope. This approach ensured systematic coverage of each compartment while maintaining consistency across all samples.

Fig. 1The alternative text for this image may have been generated using AI.

Spatial overview of a melanoma sample with region-specific IF analysis. Representative IF images of thefour analyzed compartments are shown: (a) intratumoral, (b) tumor margin, (c) intraepidermal, and (d) intradermal.The locations of the immunofluorescence images are indicated within an overview histology section of the samemelanoma sample. (e) Hematoxylin and eosin–stained overview section and (f) corresponding clinical presentation. Scale bars, 50 μm (all IF images) and 1000 μm (histology)

In healthy skin, we analyzed the epidermal and dermal compartments to provide a baseline for comparison with cancerous tissue. Fluorescence images were processed using ImageJ software to allow precise quantification of immune cells. CD1a+ DC were identified by staining with a secondary antibody conjugated to Alexa Fluor 488 (green fluorescence), while CD3+ T cells were visualized with Alexa Fluor 594 (red fluorescence). Average cell counts were calculated for each region based on multiple images and used for statistical comparisons between tumor types and regions.

CD1a+ DC and CD3+ T cell staining in skin cancer and healthy controls

To validate the IF staining protocol, we performed staining on healthy control skin. As shown in Fig. 2, in the representative healthy skin sample, CD1a+ DC were visualized using Alexa Fluor 488 (green fluorescence) and CD3+ T cells using Alexa Fluor 594 (red fluorescence). CD1a+ DC were predominantly localized in the epidermis, while CD3+ T cells were concentrated in the dermis, close to the epidermal layer. Fluorescence microscopy revealed occasional close proximity between CD1a+ DC and CD3+ T cells in the merged images, suggesting potential cell-cell interactions.

Fig. 2The alternative text for this image may have been generated using AI.

IF staining of healthy adult human skin. Representative images showing nuclear staining with DAPI (blue; a), CD1a⁺ DC (green, Alexa Fluor 488; b), CD3+T cells (red, Alexa Fluor 594; c), and the merged image displaying allfl uorochromes (d). Scale bar, 50 μm

Figure 3 presents representative images providing a comprehensive overview of the clinical, histopathological, and immunological characteristics of the analyzed tumor types. The top panel illustrates the clinical presentation of AK, SCC, BCC, and melanoma, showing their distinct morphological features. The middle panel shows hematoxylin and eosin (H&E)-stained sections, revealing tumor-specific architectural and cytological features, such as epidermal dysplasia in AK, keratinocyte atypia in SCC, basaloid nests in BCC, and melanocytic proliferation in melanoma. The bottom panel shows IF staining of CD1a+ DC (green) and CD3+ T cells (red) in representative tumor sections. Green arrows indicate CD1a+ DC, and red arrows indicate CD3+ T cells, highlighting their distribution within the TME. These images illustrate the morphological and immunological heterogeneity across tumor entities.

Fig. 3The alternative text for this image may have been generated using AI.

Clinical presentation and histopathological features of skin tumor types. Representative images of four different skin tumors are shown: AK, SCC, BCC, and melanoma. The top panel (a) shows the clinical appearance of each tumor type with characteristic macroscopic features. The middle panel (b) shows corresponding hematoxylin and eosin (H&E)-stained histological sections of each tumor type, highlighting their distinct architectural and cytological features. The bottom panel (c) shows IF staining for CD1a+ DC (green) and CD3+ T cells (red) in a representative region of interest (as indicated in the H&E staining). Green arrows indicate CD1a+ DC, and red arrows indicate CD3+ T cells. Scale bars, 1000 μm for histological overview images (500 μm for AK; b), and 50 μm for IF images

Localization of CD1a+ DC in different types of skin cancer

CD1a+ DC, which play a key role in antitumor immunity by presenting antigens to T cells, were present in all tumor types. MANOVA revealed significant differences among regions (F3,82 = 70.396, p < 0.001) and among entities (Pillai's trace: F12,252 = 2.771, p = 0.001), suggesting tumor entity-specific spatial distribution patterns (Fig. 4).

Fig. 4The alternative text for this image may have been generated using AI.

Distribution of CD1a+ DC in AK (pink, n = 18), SCC (purple, n = 23), BCC (orange, n = 19), and melanoma (dark blue, n = 22) compared to nevi (light blue, n = 16) and healthy skin (green, n = 9). Data are presented asfl oating bar charts showing the minimum to maximum values, with the mean indicated as the central line. Individualdata points are shown as black squares. Each data point represents one biological sample, calculated as the meanvalue of 5–8 regions of interest (ROIs) per sample. The counts were assessed separately in four regions as defined in Figure 1: intratumoral/intralesional, tumor margin/lesion margin, intraepidermal, and intradermal. Statistical analysiswas performed using MANOVA to assess overall effects, followed by one-way ANOVA for region-specific comparisons and post hoc testing where applicable

In AK samples, CD1a+ DC were most abundant in the intratumoral area (7.73 ± 4.48 cells/field), with moderate numbers at the tumor margin (5.07 ± 4.20 cells/field) and lower counts in the intraepidermal (3.79 ± 3.54 cells/field) and intradermal regions (2.32 ± 4.00 cells/field).

In SCC samples, the highest number of CD1a+ DC was observed in the intratumoral region (9.17 ± 5.68 cells/field), followed by the tumor margin (4.98 ± 4.16 cells/field). The intraepidermal area had a relatively higher density of CD1a+ DC compared to AK (7.57 ± 4.53 cells/field), whereas the intradermal area had a notably low number of CD1a+ DC (0.47 ± 1.02 cells/field).

BCC showed the highest intratumoral density among all tumor types (9.40 ± 7.22 cells/field), with moderate densities at the tumor margin (4.41 ± 2.99 cells/field) and in the epidermis (7.27 ± 4.61 cells/field). The dermis was sparsely populated (1.06 ± 1.76 cells/field).

In contrast, melanoma samples exhibited a distinct distribution pattern. Similar to other skin cancers, the intratumoral region had the highest concentration of CD1a+ DC (7.21 ± 6.80 cells/field). However, the tumor margin had a lower density (2.01 ± 1.75 cells/field) than that observed in NMSC. The intraepidermal region showed a moderate presence of CD1a+ DC (6.48 ± 3.27 cells/field), whereas the dermal region contained a very low number (0.55 ± 1.34 cells/field).

In nevi, CD1a+ DC were most abundant intralesionally (7.07 ± 3.07 cells/field), as well as in the epidermis (5.69 ± 2.86 cells/field), lesion margin (3.19 ± 2.60 cells/field), and dermis (0.90 ± 1.06 cells/field). Compared to melanoma, nevi exhibited a higher density of DC at the lesion margin, although this difference was not statistically significant (p = 0.089, η2 = 0.083). Pairwise comparisons between melanoma and nevi revealed no statistically significant differences in CD1a+ DC density across the regions investigated (p > 0.1).

In healthy skin, CD1a+ DC were primarily located in the epidermis, with a mean of 11.53 (± 3.37 cells/field), and only a few were found in the adjacent dermis (0.89 ± 1.08 cells/field). In contrast to the decreased numbers observed across all four tumor entities (AK, SCC, BCC, melanoma), healthy skin showed the highest average number of CD1a+ DC in the epidermis.

Overall, all tumor types demonstrated an enrichment of CD1a+ DC within the tumor. Healthy skin exhibited the highest intraepidermal density of DC, while melanoma was characterized by a relative scarcity of DC at the tumor margin. This suggests an impairment in immune surveillance in this region. Statistical comparisons confirmed these observations. Although no entity-specific differences were detected in overall or intratumoral (p > 0.3) CD1a+ DC densities, significant variation emerged at the tumor margin (F3,78 = 3.68, p = 0.015, η2 = 0.12) and in the epidermis (F3,70 = 2.91, p = 0.040, η2 = 0.11). Post hoc analyses revealed that melanoma harbored significantly fewer CD1a+ DC at the tumor margin than BCC and SCC (p < 0.05), while AK showed intermediate values. Similarly, intraepidermal CD1a+ DC numbers were lower in melanoma than in SCC and BCC. There was a trend toward significance in the dermal compartment (p = 0.052), indicating low overall CD1a+ DC densities across entities. Effect size estimates indicated that the entity contributed moderately to the variance in tumor margins (η2 = 0.12) and the epidermis (η2 = 0.11), whereas the intratumoral and dermal compartments showed small effects only (η2 ≤ 0.03). No statistically significant differences in CD1a+ DC densities were detected when comparing melanoma and nevi across intratumoral/intralesional, intraepidermal, or intradermal compartments (all p > 0.3). At the tumor margin/lesion margin, however, there was a non-significant trend toward lower CD1a+ DC numbers in melanoma (p = 0.089, η2 = 0.083), which suggests an early impairment of DC-mediated immune surveillance in malignant versus benign melanocytic lesions. Consistent with these findings, post hoc analyses revealed significantly higher intraepidermal CD1a+ DC densities in healthy skin than in all tumor entities (p < 0.001).

Localization of CD3+ T cells in different types of skin cancer

CD3+ T cells, which are crucial for adaptive immune responses, were present in all tumor types and in all regions examined. The highest densities were consistently found at the tumor margin. A MANOVA confirmed a significant main effect of tumor type on CD3+ T cell densities (Pillai’s trace: F20,332 = 2.12, p = 0.004), indicating that distribution patterns varied across tumor types (Fig. 5).

Fig. 5The alternative text for this image may have been generated using AI.

Distribution of CD3+ T cells in AK (pink, n = 18), SCC (purple, n = 23), BCC (orange, n = 19), and melanoma (dark blue, n = 22) compared to nevi (light blue, n = 16) and healthy skin (green, n = 9). Data are presented as floating bar charts showing the minimum to maximum values, with the mean as a central line. Individual data points are shown as black squares within the bars. Each data point represents one biological sample, calculated as the mean value of 5–8 regions of interest (ROIs) per sample. The different regions were analyzed separately as shown in Figure 1: intratumoral/intralesional, tumor margin/lesion margin, intraepidermal, and intradermal. Statistical analysis was performed using MANOVA to assess overall effects, followed by one-way ANOVA for region-specific comparisons and post hoc testing where applicable

In AK samples, the tumor margin exhibited the highest concentration of CD3+ T cells (2.93 ± 1.43 cells/field). The dermal region displayed a lower density (1.47 ± 2.30 cells/field), while the intratumoral and intraepidermal regions contained a low number of T cells (0.98 ± 1.36 and 0.38 ± 0.58 cells/field, respectively).

In SCC, CD3+ T cells were likewise most abundant at the tumor margin (5.45 ± 1.97 cells/field), with a moderate density in the dermal region (2.53 ± 2.18 cells/field). There were fewer T cells in the intratumoral region (1.94 ± 2.03 cells/field) and almost no T cells in the epidermal layer (0.56 ± 1.05 cells/field).

The tumor margin of BCC samples exhibited the highest density of CD3+ T cells (6.08 ± 5.12 cells/field), followed by the dermal region (2.88 ± 4.70 cells/field). A similar number of T cells was observed in the intratumoral region (2.83 ± 2.25 cells/field). In contrast, the epidermis showed a very sparse T cell infiltration (0.82 ± 1.06 cells/field).

In melanoma samples, the tumor margin exhibited the greatest concentration of CD3+ T cells (7.17 ± 5.60 cells/field), followed by the dermal region (2.23 ± 1.97 cells/field). The intratumoral region showed a comparable T cell density (1.92 ± 2.05 cells/field). As with the other tumor types, the epidermis contained fewer T cells (0.87 ± 1.04 cells/field).

In nevi, CD3+ T cells were detected in all regions, with the greatest numbers observed at the lesion margin (2.75 ± 2.06 cells/field) and within the intralesional region (2.38 ± 2.13 cells/field). Lower densities were observed in the epidermis (1.25 ± 0.81 cells/field) and dermis (0.70 ± 0.57 cells/field). Compared to nevi, melanoma samples displayed markedly higher T cell infiltration, particularly at the tumor margin/lesion margin (7.17 ± 5.60 vs. 2.75 ± 2.06 cells/field) and in the dermis (2.23 ± 1.97 vs. 0.70 ± 0.57 cells/field). Intratumoral T cell numbers were slightly lower in melanoma samples (1.92 ± 2.05 vs. 2.38 ± 2.13 cells/field). Epidermal densities were similar (0.87 ± 1.04 vs. 1.25 ± 0.81 cells/field). Direct comparison of melanoma and nevi revealed significantly higher CD3+ T cell densities in melanoma at the tumor margin/lesion margin (F1,34 = 8.52, p = 0.006, η2 = 0.20) and in the dermis (F1,34 = 8.06, p = 0.008, η2 = 0.19). However, intratumoral/intralesional and epidermal densities were not significantly different (p = 0.43 and p = 0.26).

In healthy skin, CD3+ T cells were rarely found in the epidermis (0.39 ± 0.56 cells/field), whereas most T cells were located in the dermis (4.50 ± 3.00 cells/field).

Across entities, ANOVA confirmed highly significant differences at the tumor margin (F5,100 = 111.25, p < 0.001, η2 ≈ 0.85). Here, melanoma harbored significantly more CD3+ T cells than AK, SCC, and BCC (all p < 0.01). Nevi showed intermediate values. Dermal infiltration also varied (F5,97 = 3.21, p = 0.01, η2 = 0.14), with SCC and BCC showing higher values than melanoma and AK. No significant variation was observed intratumorally (p = 0.18) or in the epidermis (p = 0.12).

Correlation between CD1a+ DC and CD3+ T cells

Correlation analyses revealed a weak but statistically significant positive association between CD1a+ DC and CD3+ T cell densities in the intratumoral region (Spearman’s ρ = 0.27, p = 0.005) and at the tumor margin (ρ = 0.26, p = 0.007). In contrast, no significant correlations were observed in the epidermal (p = 0.29) or dermal (p = 0.32–0.51) regions. These findings suggest that the spatial coupling of DC and T cell infiltration primarily occurs within the tumor core and at the invasive front, rather than in the surrounding tissue layers.

Sex- and age-related differences in immune cell infiltration

No significant sex-related differences were observed in the overall densities of CD1a+ DC or CD3+ T cells. Independent-samples t tests confirmed no significant variation in total CD1a+ DC (20.15 ± 12.76 vs. 18.53 ± 11.67, p = 0.55) or CD3+ T cells (10.94 ± 8.74 vs. 9.75 ± 5.50, p = 0.73) between female and male patients.

Region-wise analyses likewise showed no sex-related effects for either cell type (CD1a+ DC: intratumoral p = 0.82, tumor margin p = 0.93, intraepidermal p = 0.55, intradermal p = 0.71; CD3+ T cells: intratumoral p = 0.19, tumor margin p = 0.27, intraepidermal p = 0.62, intradermal p = 0.30, Supplementary Table 10). Entity-specific analyses confirmed this overall pattern: AK, SCC, and BCC showed no sex-dependent differences in either cell type (all p > 0.2), except for a significant difference in the epidermis of SCC, where male patients displayed slightly higher CD3+ counts (p = 0.029). In melanoma, there were no significant differences in CD1a+ DC densities between sexes (all p > 0.1), though a non-significant intratumoral increase was seen in male patients (9.6 vs. 4.8 cells/field, p = 0.10), while CD3+ T cell densities remained unaffected (all p > 0.25).

A non-significant trend was observed for total CD3+ T cell densities (p = 0.054). Further confirmation of the absence of sex-related effects on CD1a+ DC or CD3+ T cell densities was provided by correlation analyses (Pearson’s and Spearman’s; all r < 0.12, all p > 0.18).

Age-related analysis (< 65 vs. ≥ 65 years) revealed no differences in CD1a+ DC densities overall (p = 0.98) or by region (all p > 0.1). Patients aged ≥ 65 years had significantly higher dermal CD3+ T cell counts (2.9 vs. 1.4 cells/field, p = 0.02), with a non-significant trend toward higher numbers at the tumor margin (p = 0.19). No differences were observed in the intratumoral or intraepidermal regions. When analyzed across all regions combined, CD3+ T cells were also more abundant in older patients (11.5 vs. 8.5 cells/field, p = 0.05, d = 0.43) (Supplementary Table 11). Entity-specific analyses confirmed this pattern. In AK, older patients showed significantly higher CD3+ T cell densities (7.8 ± 4.8 vs. 3.8 ± 2.8 cells/field, p = 0.040), with no effect on CD1a+ DC (p = 0.77). SCC showed a similar, albeit non-significant trend (11.3 vs. 8.0 cells/field, p = 0.075), while BCC demonstrated numerically higher CD3+ T cell counts in older patients (15.4 vs. 10.6 cells/field, p = 0.32), with no change in CD1a+ DC (p = 0.81). There were no age-related differences in melanoma for either cell population (all p > 0.6). Correlation analyses using age as a continuous variable confirmed these findings. Although there was no significant association between patient age and CD1a+ DC densities (all p > 0.1), there was a positive correlation between CD3+ T cell densities and age, particularly in the dermis (Spearman’s ρ = 0.30, p = 0.007) and across all regions combined (ρ = 0.27, p = 0.013). For nevi, age-related analyses were not feasible, as all patients were under 63 years of age. For healthy skin samples, stratification by sex or age was not possible, as these data were anonymized and therefore unavailable.

Comparison of IF and immunohistochemistry data

Figure 6 shows a comparison of the IF and immunohistochemistry (IHC) staining methods, focusing on a representative melanoma sample. This comparison highlights the consistency between the two techniques in detecting CD1a+ DC and CD3+ T cells within the TME. In the IHC sections, CD1a+ DC cells (Fig. 6a) cells  and CD3+ (Fig. 6b) are stained pink, with arrows indicating representative cells + . In the IF section (Fig. 6c), the CD1a+ DC are stained green (green arrow) and the CD3+ T cells are stained red (red arrow).

Fig. 6The alternative text for this image may have been generated using AI.

Comparison of IF and IHC staining methods in a melanoma tissue specimen. Representative images comparing the immunostaining patterns of CD1a and CD3 by IHC (a: CD1a+ DC in pink, b: CD3+ cells in pink, with arrows indicating representative cells), and by IF (c: CD1a+ DC in green and CD3+ cells in red, arrows indicate representative cells). Scale bars: 500 μm for IHC images (a and b)

To compare IF and IHC across tumor entities, mixed-effects models were applied (Supplementary Table 11). For CD1a+ DC, no significant effect of staining method was detected in any tumor type, and no significant region × method interactions were observed, indicating comparable quantification between IF and IHC (Fig. 7). For CD3+ T cells, no significant method effect was observed in AK, BCC, or melanoma. However, SCC showed a significant main effect (p = 0.0245), with higher counts detected by IHC. Region × method interactions were identified in SCC, BCC, and melanoma, indicating compartment-dependent variability (Fig. 8). Overall, IF and IHC demonstrated largely comparable quantitative results across tumor entities, with limited marker- and region-specific differences (Supplementary Table 12).

Fig. 7The alternative text for this image may have been generated using AI.

Comparison of CD1a⁺ DC distribution across different types of skin cancer using IF and IHC. The distribution of CD1a⁺ DC across four predefined tissue compartments (intratumoral, tumor margin, intraepidermal, and intradermal) is shown for actinic keratosis (AK; n = 6, a), squamous cell carcinoma (SCC; n = 6,b), basal cell carcinoma (BCC; n = 5, c), and melanoma (n = 5, d). IF (green circles) and IHC (orange squares) were performed on serial sections of the same tumor specimens, enabling direct paired comparison. Individual paired measurements are connected by lines

Fig. 8The alternative text for this image may have been generated using AI.

Comparison of CD3⁺ T cell distribution across different types of skin cancer using IF and IHC. The distribution of CD3⁺ T cells across four predefined tissue compartments (intratumoral, tumor margin, intraepidermal, and intradermal) is shown for actinic keratosis (AK; n = 6, a), squamous cell carcinoma (SCC; n = 6, b), basal cell carcinoma (BCC; n = 5, c), and melanoma (n = 5, d). IF (green circles) and IHC (orange squares) were performed on serial sections of the same tumor specimens, enabling direct paired comparison. Individual paired measurements are connected by lines