The pattern and clinicopathological correlates of programmed death-ligand 1 expression in cutaneous squamous cell carcinoma

Dear Editor, the incidence of cutaneous squamous cell carcinoma (cSCC) is rising,1 and as there is a lack of effective therapies for inoperable cSCC, new therapeutic options are needed. Immuno‐oncological agents targeting checkpoints such as programmed death‐1 (PD‐1) or its ligand, PD‐L1, are very promising new anticancer drugs, while efficacy correlates with PD‐L1 expression in various tumour types. In order to provide a translational basis for the possible use of PD‐1/PD‐L1 inhibitors in cSCC, we examined the expression pattern of PD‐L1 in tumour cells and tumour infiltrating leucocytes (TILs), as well as the proportion of CD8+ T cells and correlated these findings with the clinicopathological characteristics of patients. We therefore performed immunohistochemical dual staining against PD‐L1 (clone E1L3N; Cell Signaling Technology, Danvers, MA, U.S.A.) and CD8 (clone C8/144B; Dako, Hamburg, Germany) (Fig. S1; see Supporting Information) of 75 formalin‐fixed, paraffin‐embedded specimens [68 primary cSCC and seven cutaneous metastases; for details see Appendix S1 (Supporting Information)]. Applying a cut‐off of ≥ 1%, 26% of the primary cSCC were positive for PD‐L1 expression in the tumour and 60% in TILs. Utilizing a cut‐off ≥ 5%, 10% of tumours and 43% of TILs were PD‐L1 positive. The severity of inflammation (graded as absent, mild, moderate and severe) was positively correlated with PD‐L1 expression in both tumour and TILs (Table 1, Fig. S2; see Supporting Information). Moreover, PD‐L1 expression in tumour cells was associated with the presence of intratumoral CD8+ cells (Table S1; see Supporting Information). The seven metastases displayed a similar staining pattern as the primary tumours (Table S2; see Supporting Information). Additionally, the location of the tumour had an impact on PD‐L1 expression. cSCC located in sun‐exposed areas (66%) showed a higher expression of PD‐L1 in TILs than cSCC excised in nonsun‐exposed areas of the body. Other clinicopathological parameters, including sex, age, tumour size, tumour grading or the risk of developing metastases, did not correlate with PD‐L1 expression in primary cSCC (Table 1). Correlation between programmed death‐ligand 1 (PD‐L1) expression and clinicopathological features of patients with primary cutaneous squamous cell carcinoma (cut‐off ≥ 1%) Statistical analyses were performed with the χ2 test. TIL, tumour‐infiltrating leucocytes. aSun‐exposed locations: head, face, lower arms and hands; nonsun‐exposed locations: trunk, upper arms and legs, and feet. Correlation between programmed death‐ligand 1 (PD‐L1) expression and clinicopathological features of patients with primary cutaneous squamous cell carcinoma (cut‐off ≥ 1%) Statistical analyses were performed with the χ2 test. TIL, tumour‐infiltrating leucocytes. aSun‐exposed locations: head, face, lower arms and hands; nonsun‐exposed locations: trunk, upper arms and legs, and feet. Herbst et al. recently showed in multiple cancer types that a response to treatment with a PD‐L1 inhibitor was observed, especially when PD‐L1 was expressed by TILs;2 therefore, the expression of PD‐L1 in the majority of TILs in cSCC provides a preclinical rationale for the use of PD‐1 inhibitors. However, it has to be taken into account that PD‐L1 is a dynamic marker affected by the inflammatory cell infiltrate, limiting the value of this marker. For laryngeal SCC, melanoma, nonsmall cell lung carcinoma and renal cell carcinoma, it has been demonstrated, that PD‐L1 expression correlated with severity of inflammation.3,4 Thus, our observations showing a strong link between TILs and PD‐L1 expression further confirm these findings and provide the first evidence for a tumour‐specific immune response in cSCC. As we have shown that the SCC cell line A431 upregulated PD‐L1 protein in response to interferon (IFN)‐γ (Fig. S3; see Supporting Information), and intratumoral CD8+ cells were associated with PD‐L1 expression, we anticipate that IFN‐γ plays a prominent role in triggering PD‐L1 expression in tumour as well as in TILs in cSCC, as has been shown previously for other entities.3,5 The reason for the correlation of PD‐L1 expression on TILs in sun‐exposed areas is unclear at the moment. In melanoma, an increased mutational burden has been described for tumours in sun‐exposed vs. sun‐protected areas.6 As the mutational load, which is associated with a higher number of neoantigens, has been described to correlate with PD‐L1 expression in lung cancer,7 it is tempting to speculate that there might also be a correlation in cSCC that remains to be determined. The prognostic implications of PD‐L1 expression, independent of treatment with a PD‐1 blocker, are controversial. While PD‐L1 is generally regarded as an immunosuppressive molecule, retrospective studies in different cancer types, such as Merkel cell carcinoma, melanoma and laryngeal SCC, described PD‐L1 expression as a positive prognostic factor.3,4,5 In contrast, in oesophageal SCC PD‐L1 expression was linked with poor prognosis.8 Our findings revealed that PD‐L1 expression does not correlate with prognostic features of the primary cSCC or the occurrence of metastases during follow‐up, which corresponds to data obtained from pulmonary SCC and head and neck cancer.9,10 However, this is in contrast to a recently published study investigating 45 cSCC, which found a correlation of PD‐L1 expression and the risk of metastasis.11 At this point in time the explanation for this discrepancy is unclear and might require the investigation of further tumours. Moreover, no standardized staining or evaluation scheme for PD‐L1 expression currently exists, limiting the comparability of PD‐L1 expression among individual studies. Taken together, our findings indicate the expression of PD‐L1 in a relevant proportion of cSCC. First case reports and case series described a response of locally advanced or metastatic cSCC to PD‐1 inhibitors,12,13,14 providing the rationale for the systematic investigation of PD‐1/PD‐L1 inhibitors in inoperable cSCC within clinical trials (e.g. NCT02760498; see www.clinicaltrials.gov). Conflicts of interest: R.G. has served as a consultant for Bristol‐Myers Squibb, Roche Pharma, Merck Sharp & Dohme, Novartis, GlaxoSmithKline, Amgen, Almirall Hermal, LEO and Pfizer; has received honoraria for lectures from Bristol‐Myers Squibb, Roche Pharma, Merck Sharp & Dohme, Novartis, GlaxoSmithKline, Amgen, Merck Serono, Almirall Hermal, Galderma and Boehringer Ingelheim; has received travel support from Bristol‐Myers Squibb and Roche Pharma; and has received research grants from Pfizer and Johnson & Johnson.