Expression of PD-L1 in keratoacanthoma and different stages of progression in cutaneous squamous cell carcinoma

Abstract

Background

Programmed cell death 1 (PD-1) and its ligands (PD-L1) play a major role in the immune responses of a variety of cancers.

Objectives

To investigate the expression of PD-L1 in different progression forms of cutaneous squamous cell carcinoma (cSCC) and keratoacanthoma (KA).

Methods

We performed immunohistochemical staining of 21 KA, 26 actinic keratoses (AK), 20 Bowen´s diseases (BD), and 26 high-risk cSCC. The staining patterns were assessed using the tumour proportion score and staining intensity evaluation. Immunohistology scores were statistically analysed.

Results

PD-L1 expression of tumour cells as well as tumour-infiltrating cells (TILs) was significantly higher in KA and cSCC when compared to AK and BD (P = 0.00028 and P = 0.00033, respectively). We observed a very strong positive correlation between the PD-L1 protein expression of tumour cells of KA and the PD-L1 protein expression of TILs (r = 0.97; P < 0.0001). A similar correlation was also found for cSCC (r = 0.86; P < 0.0001). The percentage of PD-L1 + tumours was 33.3% for KA and 26.9% for cSCC. Similarly, the percentage of PD-L1 + TILs in KA and cSCC was 33.3 and 34.6%, respectively.

Conclusions

PD-L1 is differently expressed in cSCC and closely related non-melanoma skin cancer. cSCC exhibit PD-L1 expression in a fourth of cases, indicating that PD1/PD-L1 inhibitors might be beneficial in a proportion of patients with an inoperable or metastatic cSCC. Unlike AK and BD, TILs and tumour cells of KA and cSCC present very similar PD-L1 expression profiles indicating a common immune escape mechanism.

Background

Cutaneous squamous cell carcinoma (cSCC) belongs to the common group of non-melanoma skin cancer (NMSC). In the past decades, the incidence of cSCC seems to have increased by 50 and up to 200% with stabilisation trends or slower rates of increase [1]. Nevertheless, cSCC reveals to be the second highest cancer entity in the Caucasian population with about 20% of all cutaneous malignancies. The potential of metastasis is estimated to be within a range of 2–5% which differs with various subgroups. Thus, lesions with a high histologic grade, an expansion larger than 2 cm in diameter, and especially, greater than 6 mm in thickness, are known to have a higher risk for metastasis. In most cases, cSCC primarily metastasise in regional lymph nodes and subsequently infiltrate distant tissues dramatically worsening the survival prognosis [2, 3].

It is evident that the tumour microenvironment, which is orchestrated by inflammatory cells, plays a central role in immunologic response and tumour escape. Thus, the engagement of programmed death 1 (PD-1) receptor with its two ligands (PD-L1 and PD-L2) is associated with suppression of tumour-reactive T cells and the recruitment of immunosuppressive T regulatory cells. This interaction provides contrary stimuli in regulating T cell activation in many immunological processes, including autoimmune diseases, pregnancy, and malignant transformation. Hence, the differentiation of regulatory T cells is promoted, whereas the cytotoxic program including the proliferation of CD8+ T cells is inhibited. Moreover, the expression of PD-1 and PD-L1 varies in different cells and tissues. While PD-1 is expressed on regulatory T cells, activated lymphoid cells and overexpressed on exhausted T cells, PD-L1 is mainly expressed on activated T and B cells, antigen-presenting cells, placenta, and different tumours [4–7].

To the best of our knowledge, there are no reports on the PD-L1 expression in different progression forms of cSCCs as well as keratoacanthomas (KAs). We have to consider actinic keratosis (AK) and Bowen´s disease (BD) as part of a disease continuum in a multi-step carcinogenesis process which can progress into invasive cSCC with definitive metastasizing potential. The risk of progression of a single lesion is estimated to range between 0.025 and 16% per year [8]. Thus, it is not possible to predict which precursor lesion will progress into an invasive tumour. By contrast, KA is considered a benign skin tumour clinically and histopathologically resembling well-differentiated cSCC [9–12]. In the present study, we aimed to compare the PD-L1 expression profiles in KA and subtypes of cSCC, and to correlate clinicopathological data with PD-L1 expression.

Materials and methods

Patients

We studied 21 KA, 26 AK, 20 BD, and 26 invasive (tumour thickness >2 mm) cSCC. All aforementioned tumours were located on the head/face. We conducted this study in the Skin Cancer Center of the Department of Dermatology, Ruhr-University (Bochum, North-Rhine-Westphalia, Germany). Patients with immunosuppression were excluded from the investigation. The study was approved by the ethics review board of the Ruhr-University, Bochum, and conducted according to the Declaration of Helsinki principles.

Immunohistology

Immunohistochemistry was performed according to the following protocol: 4 µm paraffin sections were placed on Dako (Hamburg, Germany) FLEX glass slides and stored in an oven at 60 °C for 30 min. Thereafter, the section was dewaxed and heat-induced epitope retrieval was performed using DAKO Envision Flex Retrival Solution at low pH (pH 6) for 20 min. After a cooling period in Dako Wash Buffer, the sections were treated with Endogenous Enzyme Block (Dako) for 5 min. A washing step was followed by the incubation with the primary antibody (anti-PD-L1 rabbit monoclonal, 1:500, 60 min, room temperature, Abcam, ab205921 (Cambridge UK)). The following steps (secondary antibody, alkaline phosphatase, and chromogen) were carried out using the Dako REAL Detection System, Alkaline Phosphatase/RED, Rabbit/Mouse according to the manufacturer’s procedure, including the haematoxylin counterstaining. For the staining, a Dako Autostainer Link Instrument was used. The sections were dehydrated using an ascending alcohol series and mounted with Eukitt.

Microscopic evaluation

The immunohistochemical expression of PD-L1 was examined using light microscopy. PD-L1 protein expression was evaluated by means of the tumour proportion score, defined as the percentage of all viable tumour cells showing partial or complete membrane staining (intensity ≥1+) [13]. Sections of tumour were confirmed to have at least 100 viable tumour cells. Staining intensity was semi-quantitatively evaluated: 0 negative, 1+ weak, 2+ moderate or 3+ strong. According to the H-score [14], staining intensity was multiplied with the TPS resulting in a score ranging from 0 to 300. Tumour proportion score was classified as follows: negative PD-L1 expression (<1%), positive PD-L1 expression (≥1%). Only the membrane staining of tumour cells was evaluated. Cytoplasmic staining was not included. Tumour-associated lymphocytes were separately assessed; all other non-tumour cells were excluded from scoring: normal cells, necrotic cells, and debris. Tumour-infiltrating lymphocytes (TILs) were also assessed using the H-score.

Statistics

Data analysis was performed using the statistical package MedCalc Software version 15.2 (Ostend, Belgium). Distribution of data was assessed by the D`Agostino–Pearson test. Data were analysed using the Chi-square test, Kruskal–Wallis ANOVA including the Conover post hoc test for pairwise comparisons, Spearman or Kendall’s Tau correlation procedures.

Results

The mean tumour thickness of cSCC was 5.9 ± 2.3. One (3.8%) cSCC had Clark level II, one (3.8%) had Clark level III, 16 (61.5%) had Clark level IV, and 8 (30.8%) had Clark level V. Histologic grading was I in 15 (57.7%) cSCC, II in 10 (38.5%) cSCC, and III (3.8%) in 1 cSCC. Perineural invasion was observed only in one (3.8%) tumour. 24 (92.3%) cSCC patients had tumour stage II, 1 patient had stage I, and 1 patient had stage III. The PD-L1 H-score of all tumours was compared using the Kruskal–Wallis ANOVA for non-parametric data. We have observed that PD-L1 expression of tumour cells was significantly higher in KA and cSCC when compared to AK and BD (P = 0.00028; Fig. 1). Moreover, PD-L1 expression of TILs was also significantly higher in KA and cSCC when compared to AK and BD (P = 0.00033; Fig. 2). We observed a very strong positive correlation between the PD-L1 protein expression of tumour cells of KA and the PD-L1 protein expression of TILs (r = 0.97; P < 0.0001). A very similar correlation was also found for cSCC (r = 0.86; P < 0.0001). As shown in Table 1 the percentage of PD-L1+ tumours (immunoreactive cells ≥1%) was 33.3% (7/21) for KA and 26.9% (7/26) for cSCC (Fig. 3). Similarly, the percentage of PD-L1+ TILs was 33.3% (7/21) and 34.6% (9/26). In AK and BD, there was no detectable immunoreactivity for PD-L1. The results of the tumour proportion score and staining intensity are collected in Table 1. Clinicopathological data such as tumour thickness, Clark level, perineural invasion, grading, and tumour stage of cSCC patients did not significantly correlate with PD-L1 immunoreactivity in tumour cells and tumour-infiltrating cells (P > 0.05).

Fig. 1.

PD-L1 protein expression of tumour cells. Showing the PD-L1 protein expression of tumour cells in keratoacanthoma (KA), actinic keratosis (AK), Bowen’s disease (BD), and invasive cutaneous squamous cell carcinoma (cSCC). PD-L1 expression is significantly higher in KA and cSCC when compared to AK and BD (Kruskal–Wallis test P = 0.00028)

Fig. 2.

PD-L1 protein expression of tumour-infiltrating lymphocytes. Showing the PD-L1 protein expression of tumour-infiltrating lymphocytes in keratoacanthoma (KA), actinic keratosis (AK), Bowen’s disease (BD), and invasive cutaneous squamous cell carcinoma (cSCC). PD-L1 expression is significantly higher in KA and cSCC when compared to AK and BD (Kruskal–Wallis test P = 0.00033)

Table 1.

PD-L1 protein expression in keratoacanthoma, actinic keratosis, Bowen’s disease, and invasive cutaneous squamous cell carcinoma (cSCC)

PD-L1 expression	Keratoacanthoma	Actinic keratosis	Bowen’s disease	Invasive cSCC
H-score—tumour median (range)	0 (0–90)	0 (0–0)	0 (0–0)	0 (0–173)
Tumour (% of expression)a median (range)	5 (2.4–30)	0 (0–0)	0 (0–0)	4 (1.4–57.5)
Tumour (intensity of expression; score 1-3)a median (range)	2 (1–3)	0 (0–0)	0 (0–0)	2 (1–3)
H-score TILsb median (range)	150 (0–260)	198.8 (0–243.8)	185.8 (0–256.5)	186 (0–268.5)
Proportion of positivea tumours	7/21 (33.3%)	–	–	7/26 (26.9%)
Proportion of positivea TILs	7/21 (33.3%)			9/26 (34.6%)

^aOnly PD-L1 positive cases (≥1% immunoreactivity); ^b Tumour-infiltrating lymphocytes

Fig. 3.

Immunohistochemical staining for PD-L1 in cSCC and KA. Invasive cutaneous squamous cell carcinoma (a, b) and keratoacanthoma (c, d) with tumour-infiltrating lymphocytes (TILs) positive for programmed death ligand 1 [original magnification ×20 (a, c), ×40 (b, d)]. PD-L1 reactivity can be detected along tumour cell membranes (*) with moderate intensity (a, b) and stronger intensity (c, d), respectively. In contrast, TILs show a higher reactivity to PD-L1 (>)

Discussion

Monoclonal antibodies against PD-1 can block the PD-1 immune checkpoint pathway and enhance anti-tumour T cell responses. PD-1 inhibitors have shown durable responses in different tumour entities such as non-small cell lung carcinoma, renal cell carcinoma, cutaneous melanoma, and Merkel cell carcinoma. The membranous PD-L1 expression of tumour cells and/or TILs has been worked out as biomarker to predict the response to PD-1 antibody treatment in several cancers [5, 14–17]. First case reports have already documented a response of locally advanced or metastatic cSCC to PD-1 inhibitors [18–20].

Schaper et al. [21] recently showed that when applying a cut-off of ≥1, 26.5% of the primary cSCC were positive for PD-L1 expression in the tumour and 60.3% in TILs. Their clinicopathological parameters, including sex, age, tumour size, tumour grading or the risk of developing metastasis did not correlate with PD-L1 expression in primary cSCC [21]. By contrast, Slater and Googe [13] could demonstrate a correlation between PD-L1 expression in cSCC tumour cells and its risk of metastatic disease. In their study, PD-L1 expression was present (≥1%) in 4 of 20 low risk tumours (20%), all of which showed low expression [13]. PD-L1 expression was present in 14 of 20 (70%) high-risk tumours (diameter ≥2 cm, histologic grade ≥2, and/or tumour thickness ≥4 mm) [13]. Moreover, Slater and Googe [13] observed PD-L1 expression in 5 of 5 (100%) metastases, with 3 showing low expression and 2 showing high expression. Together, Slater and Googe [13] demonstrated PD-L1 expression in low- and high-risk cSCC and showed a positive correlation between the degree of PD-L1 expression and pathologic findings related to the risk of metastasis. Jiao et al. [14] found that 25 of 61 (40.98%) cSCC cases (stage I–IV) exhibited positivity for PD-L1. The expression rate of PD-L1 was associated with UICC (Union Internationale Contre le Cancer) stage of cSCC [14]. Previous studies on other tumour entities revealed controversial data with respect to the PD-L1 expression and its association with prognosis [22–29].

With regard to PD-L1 expression in cSCC, our present data is in line with the above discussed studies investigating invasive cSCC of different tumour stages [13, 17, 21]. Hence, we observed in about a fourth and a third positive PD-L1 immunoreactivity in tumour cells and TILs, respectively. Unlike the aforementioned researchers studying PD-L1 expression in cSCC [13, 17, 21], we also assessed BD and AK representing in situ cSCC and cSCC precursor lesions, respectively. Interestingly, neither AK nor BD exhibited positive PD-L1 expression in tumour cells or TILs. We have to concede, however, that the differences between KA/cSCC and AK/BD observed in the extent of PD-L1 expression of lymphocytes may be based on the fact that AK/BD are not invasive tumours, and thus there is a significantly reduced likelihood for immune cells to come and attack these lesions. By contrast, we found PD-L1 expression in tumour cells and TILs in about a third of KA cases. Frequency of PD-L1 positivity and median PD-L1 expression levels in KA and cSCC were not significantly different, and in both tumour types we observed a very strong correlation between the PD-L1 expression of tumour cells and PD-L1 expression of TILs. A positive correlation of PD-L1 expression by tumour cells with PD-L1 expression by adjacent immune cells has recently been demonstrated in Merkel cell carcinoma [27]. Nevertheless, it is still controversially discussed whether KA is a well-differentiated variant of cSCC or a separate entity [9, 10, 12]. By contrast, KA and cSCC are also considered biologically distinct lesions with distinct clinicopathological properties. KA is usually considered a benign neoplasm that is characterized by spontaneous regression. In very rare cases, however, otherwise typical KA can behave aggressively showing signs of perineural and perivascular invasion, and metastases in regional lymph nodes [12]. Our results indicate that KA and cSCC are also closely related regarding their tendency to express the immunosuppressive molecule PD-L1 on the surface of the tumour cells and TILs. PD-L1 expression in KA as well as cSCC might represent an immune escape mechanism induced by TILs which frequently consist of cytotoxic T lymphocytes. The latter have been shown to be abundant in KA and cSCC [9, 10]. In KA, CD8 + TILs and granzyme B expression seem to play a crucial role in spontaneous tumour regression [9, 10]. Though cytotoxic T lymphocytes are commonly found in cSCC as well, spontaneous regression is a rare event in this tumour. This may be due to other more effective immune escape mechanisms of “really” malignant tumour cells in cSCC when compared to KA tumour cells usually not having a sufficient metastatic potential [9, 10, 12].

In conclusion, PD-L1 is differently expressed in cSCC and closely related to NMSC. cSCC exhibit PD-L1 expression in a fourth of cases indicating that PD1/PD-L1 inhibitors might be beneficial in a proportion of patients with inoperable or metastatic cSCC [18–20]. Unlike AK and BD, TILs and tumour cells of KA and cSCC exhibit very similar PD-L1 expression profiles indicating a common immune escape mechanism.

Abbreviations

AK

Actinic keratosis

BD

Bowen´s disease

cSCC

Cutaneous squamous cell carcinoma

KA

Keratoacanthoma

NMSC

Non-melanoma skin cancer

TPS

Tumour proportion score

UICC

Union Internationale Contre le Cancer

Compliance with ethical standards

Conflicts of interest

The authors declare that they have no conflict of interest.