The proportions were calculated using 78 as the denominator for treated tumors and 60 as the denominator for treatment-naive tumors.
A, Original magnification ×200. B, Original magnification ×100. C and D, The ratio of CD8+ to CD3+ cells was 1 in 4 (25%).
Customize your JAMA Network experience by selecting one or more topics from the list below.
Chang J, Zhu GA, Cheung C, Li S, Kim J, Chang ALS. Association Between Programmed Death Ligand 1 Expression in Patients With Basal Cell Carcinomas and the Number of Treatment Modalities. JAMA Dermatol. 2017;153(4):285–290. doi:10.1001/jamadermatol.2016.5062
What is the expression of programmed death ligand 1 (PD-L1) in basal cell carcinomas (BCCs) given that PD-L1 expression has been associated with response to programed death 1 inhibitor immunotherapy in other tumor types?
In this cross-sectional study of 138 BCCs from 62 patients, 89.9% were positive for PD-L1 immunohistochemical staining in tumor cells, and 94.9% were positive for PD-L1 expression in tumor-infiltrating lymphocytes. In a multivariable model, PD-L1 expression of treated BCCs was significantly increased in tumor cells and tumor-infiltrating lymphocytes compared with treatment-naive BCCs.
The high rate of PD-L1 positivity in BCCs and tumor-infiltrating lymphocytes suggests that programmed death 1 inhibitor immunotherapy may have activity against BCCs.
Response to programmed death 1 (PD-1) inhibitors has been associated with programmed death ligand 1 (PD-L1) expression levels in several cancers, but PD-L1 expression and its clinical significance in basal cell carcinoma (BCC) are unknown to date.
To assess PD-L1 expression in treatment-naive and treated BCCs.
Design, Setting, and Participants
This investigation was a cross-sectional study at a single academic tertiary referral center. Immunohistochemical staining on formalin-fixed BCCs from a dermatology clinic were examined in masked fashion by a dermatopathologist and a dermatologist. The study dates were March 31, 2014, to June 7, 2016.
Treated BCCs (including those recurrent after surgery, radiotherapy, systemic chemotherapy, or topical chemotherapy) vs treatment-naive BCCs.
Main Outcomes and Measures
Percentage of tumor cells and tumor-infiltrating lymphocytes (TILs) with PD-L1 expression, intensities of expression, and association with treatment modalities.
Among 138 BCCs from 62 patients (43 males and 19 females; mean [SD] age at biopsy, 61.6 [13.7] years), 89.9% (124 of 138) were positive for PD-L1 expression in tumor cells, and 94.9% (131 of 138) were positive for PD-L1 expression in TILs, defined as greater than 5% positive immunohistochemical staining in the respective cell populations. The PD-L1 immunohistochemical staining intensity of 78 treated BCCs compared with 60 treatment-naive BCCs was significantly different in tumor cells (32% vs 7%, P = .003) and TILs (47% vs 18%, P = .008) after adjusting for the age at diagnosis. In a multivariable model adjusting for age, sex, and BCC location, PD-L1 staining intensity in tumor cells increased with the number of distinct prior treatment modalities (median, 0.12; interquartile range, 0.03-0.20; P = .007).
Conclusions and Relevance
Our data suggest that PD-1 immunotherapy may have activity against BCCs, including in those that have been previously treated. This hypothesis needs to be tested in future clinical trials.
Many unresectable or metastatic basal cell carcinomas (BCCs) are responsive to targeted therapies, such as those inhibiting the Hedgehog signaling pathway.1 However, a substantial subset do not respond,2 and alternative treatment options are limited. In many solid tumors, including skin cancers (eg, melanoma or Merkel cell carcinoma), programmed death 1 (PD-1) inhibitors have demonstrated good results and acceptable safety profiles.3
Quiz Ref IDProgrammed death ligand 1 (PD-L1) is a cell surface glycoprotein normally expressed by immune cells that binds its receptor, PD-1, thereby inhibiting the immune response in inflammatory reactions and promoting peripheral tolerance in normal physiological signaling.4 Many cancers exploit this pathway by upregulating PD-L1 on tumor cells and tumor-infiltrating lymphocytes (TILs), thereby escaping immune surveillance.5 The PD-L1 expression levels have been associated with response to treatment, with PD-1 inhibition in numerous tumor types.6-8 For instance, the degree of PD-L1 expression in tumor cells predicts response in melanoma,6-8 and the degree of PD-L1 expression in TILs predicts response in non–small cell lung cancer and metastatic bladder cancer.9-11 The role of the PD-1 and PD-L1 pathway in BCCs is not well known, thereby precluding investigation of potential associations between PD-1 and PD-L1 immunophenotypes and tumor behavior. In this initial study, we performed immunohistochemical staining on formalin-fixed BCCs, assessed the percentage and intensity of PD-L1 expression on tumor cells and TILs, and explored whether expression correlates with clinical outcomes.
After approval from the Stanford University Human Subjects Panel (institutional review board), consecutive BCCs biopsied at an academic dermatology clinic specializing in nonmelanoma skin cancer were included for analysis if there was adequate tissue for immunohistochemical staining (detailed below). The Stanford University Human Subjects Panel waived the need for patient consent. Demographic and clinical data at the time of biopsy were recorded through review of medical records. If more than 1 subtype of BCC was noted on the pathology report, the predominant form (usually the first subtype) was recorded. Treatment modalities before biopsy were recorded and included surgery (with each distinct surgery counted as a separate treatment modality), radiotherapy, systemic chemotherapy (with each chemotherapy drug counted as a separate treatment modality), and topical chemotherapy (with each chemotherapy drug counted as a separate treatment modality).
The PD-L1 immunohistochemical staining was performed on formalin-fixed, paraffin-embedded tissue. Glass slides were subjected to heat-induced antigen retrieval and peroxide blocking before incubation with the primary antibody (mouse anti–PD-L1 monoclonal antibody, clone 2B11D11; Proteintech) at a dilution of 1:1000. Antigen-antibody binding was visualized using 3,3′-diaminobenzidine tetrahydrochloride. To identify TILs, the tumor sections were also immunostained with anti-CD8 antibody (clone 144B; Dako) and anti-CD3 antibody (product No. CMC10317040; Cell Marque) at dilutions of 1:400 and 1:100, respectively. Tonsillar tissue was the positive control, and normal human skin was the negative control.
The percentage of PD-L1 expression in tumor cells and PD-L1, CD8, and CD3 expression in TILs for each specimen was assessed by a dermatopathologist (J.K.) and a dermatologist (A.L.S.C.) masked to clinical data. Discrepancies in assigning percentages for each specimen were reconciled by consensus. Each specimen was examined under low magnification to identify representative fields of PD-L1, CD8, and CD3 staining. The percentage of PD-L1 staining was visually assessed under high magnification after examining multiple representative fields. The percentage of PD-L1 staining in tumor cells or TILs was determined by dividing the number of cells staining positive by the total number of cells for each cell type. The PD-L1 staining intensity was ranked on a numerical scale, where 0 is none, 1 is weak, 2 is moderate, and 3 is strong.12,13
The ratio of CD8+ to CD3+ cells in TILs was assessed by examining each tumor section at scanning magnification to identify a representative field of CD3+ immunohistochemical staining. Total CD3+ cells and CD8+ cells in the same field were counted under high-power magnification, and the ratio of CD8+ to CD3+ cells was calculated.
The mean values with SDs or the medians with interquartile ranges (IQRs) were reported in univariate analyses. Bivariate comparisons were performed with χ2 tests or Fisher exact tests for categorical variables and Wilcoxon rank sum tests for continuous variables. Multivariable analyses were performed using a linear mixed regression model or a generalized estimating equation model adjusted for age and repeated measures. Spearman rank correlation coefficients between immunohistochemical staining variables of intensity and percentage and the number of distinct treatment modalities were calculated, and a linear repeated mixed model adjusting for age, sex, and anatomic location of BCC was used. All statistical analyses were performed using a software program (SAS, version 9.4; SAS Institute Inc).
A total of 138 BCCs from 62 patients were included for analysis in this cross-sectional study at a single academic tertiary referral center, of which 43.5% (60 BCCs from 40 patients) were treatment naive and 56.5% (78 BCCs from 29 patients) were previously treated at the time of biopsy. Some BCCs in the treated groups were from the same patient. Treatment modalities before biopsy included surgery (with biopsy confirming recurrent disease) (n = 25), radiotherapy (n = 9), systemic chemotherapy (n = 58), and topical chemotherapy (n = 22). Systemic agents included smoothened inhibitors (SIs) (n = 47), platinum agents (n = 10), and gefitinib (n = 5), and topical agents included fluorouracil (n = 21) and imiquimod (n = 1). Nineteen percent (5 of 26) of BCCs (26 from 12 patients) were treated with more than 1 modality.
Previously treated BCCs were exposed to a mean (SD) of 1.7 (1.4) modalities. Tumors were most commonly found on the head and neck (64.5% [89 of 138]), followed by the trunk (30.4% [42 of 138]) and extremities (5.1% [7 of 138]) (Table 1). The most common histological subtype was nodular (42.0% [58 of 138]), followed by superficial (21.0% [29 of 138]). In bivariate analyses, treated tumors were associated with younger age (58.6 vs 65.5 years, P = .001) and anatomic location, with a greater proportion occurring on the trunk (43.6% [34 of 78] vs 13.3% [8 of 60], P < .001) compared with untreated tumors.
Quiz Ref IDPositive PD-L1 staining in tumor cells and TILs was observed in 89.9% (124 of 138) and 94.9% (131 of 138) of specimens, respectively, as defined by a greater than 5% expression threshold.6,14 Among all BCCs, the median percentage of PD-L1 immunohistochemical staining was 50% (IQR, 30%-75%) in tumor cells and 50% (IQR, 30%-70%) in TILs. High PD-L1 staining intensity (defined as moderate or strong [intensity score of 2 or 3 of 3]) of tumor cells and TILs was noted in 21% (29 of 138) and 34.8% (48 of 138) of specimens examined, respectively. The median ratio of CD8+ to CD3+ cells was 30% (IQR, 20%-50%).
The distribution of PD-L1 immunohistochemical staining percentage and intensity in tumor cells and TILs for treatment-naive and previously treated BCCs is shown in Figure 1. Quiz Ref IDTreated BCCs had significantly higher PD-L1 immunohistochemical staining intensity in tumor cells (32% vs 7%, P = .003) and TILs (47% vs 18%, P = .008) compared with treatment-naive BCCs.
To explore whether a particular type of treatment was associated with increased PD-L1 expression, staining characteristics of treated BCCs exposed to an SI (n = 47) and treated BCCs not exposed to an SI (n = 31) were compared with treatment-naive BCCs (n = 60) (Table 2). The PD-L1 staining intensity of tumor cells and TILs was increased in SI-exposed BCCs and SI-naive BCCs compared with treatment-naive BCCs (P < .05 for all comparisons).
Because the density of CD8+ TILs has been reported to be associated with response to anti–PD-1 therapy in metastatic melanoma,5 we assessed the ratios of CD8+ to CD3+ cells in the TILs of treatment-naive and treated BCCs. No significant differences in the median ratios of CD8+ to CD3+ cells were found (Table 2).
Next, we examined whether the number of treatment modalities before biopsy was associated with PD-L1 expression in tumor cells and TILs in multivariable models adjusted for age, sex, and BCC location. Treatment modalities included surgery, radiotherapy, systemic chemotherapy, and topical chemotherapy. The number of treatment modalities was not associated with the PD-L1 staining percentage in tumor cells or TILs, nor was it associated with the ratio of CD8+ to CD3+ cells in TILs. Quiz Ref IDHowever, the number of treatment modalities was significantly associated with the PD-L1 intensity score in tumor cells (median, 0.12; IQR, 0.03-0.20; P = .007), with each additional treatment modality increasing the score by 0.12 (Table 3). Examples of immunohistochemical staining of tumor cells and TILs for PD-L1, CD8, and CD3 are shown in Figure 2.
The data from this study provide an initial step in describing the landscape of PD-L1 immunohistochemical staining in BCCs. Quiz Ref IDCompared with other epithelial tumors, untreated BCCs in this study demonstrated high levels of PD-L1 immunohistochemical staining. For instance, 40% of untreated lung adenocarcinomas15 are positive for PD-L1 when using the same positivity threshold of greater than 5% tumor cells and the same PD-L1 primary antibody clone as in this study. In addition, a study16 of oral squamous cell carcinomas (in which 93% were untreated tumors) found that 64% were positive for PD-L1 using the greater than 5% positivity threshold (although the exact primary antibody clone was not specified).
Moreover, our finding that treated BCCs express higher levels of PD-L1 compared with untreated BCCs is consistent with studies on PD-L1 expression in other tumor types. For example, PD-L1 immunohistochemical staining in paired tumor specimens increased after acquired gefitinib resistance in non–small-cell lung cancer.17 The PD-L1 expression in serum dendritic cells was increased in patients with androgen receptor inhibitor–resistant prostate cancer compared with treatment-naive patients.18 In addition, PD-L1 immunohistochemical staining has been reported to increase after radiotherapy or platinum-based chemotherapy in mice with breast and ovarian tumors.19,20
Our study has several limitations. First, the ability to analyze PD-L1 expression in BCCs exposed to different treatment modalities compared with treatment-naive BCCs was limited by the sample size.
Another caveat of our study is that the BCCs in our study were derived from a specialty nonmelanoma skin cancer clinic enriched for advanced or high-risk BCCs. Therefore, PD-L1 expression in this study may be higher than if the BCCs were drawn from a lower-risk, general dermatology clinic. This study population limits generalizability of our findings to other patient groups.
An additional limitation pertains to our lack of paired samples obtained from the same BCC before and after each treatment exposure. Therefore, while PD-L1 expression is associated with the number of treatment modalities, we cannot comment on the direction of causality. Specifically, high expression levels may be a consequence of treatment exposure or may reflect the underlying intrinsic aggressive behavior of the BCC. In other tumor types, such as melanoma21 and non–small-cell lung cancer,22 PD-L1 expression is independently associated with more aggressive phenotypes and worse outcomes. If the same is true in BCCs, then PD-L1 immunophenotyping may have substantial prognostic value with regard to response to immunotherapy and long-term outcomes.
Because PD-L1 expression has been positively associated with response to PD-1 inhibition in multiple cancers,6-8,23 our data suggest the potential usefulness of directed immunotherapy, such as PD-1 inhibitors, in the treatment of unresectable or metastatic BCCs. This concept remains to be tested in future clinical trials.
Accepted for Publication: October 25, 2016.
Corresponding Author: Anne Lynn S. Chang, MD, Department of Dermatology, Stanford University School of Medicine, 450 Broadway St, Redwood City, CA 94063 (firstname.lastname@example.org).
Published Online: March 4, 2017. doi:10.1001/jamadermatol.2016.5062
Author Contributions: Drs Kim and A. L. S. Chang had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: J. Chang, Zhu, A. L. S. Chang.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: J. Chang, Zhu, Cheung, A. L. S. Chang.
Critical revision of the manuscript for important intellectual content: J. Chang, Zhu, Li, Kim, A. L. S. Chang.
Statistical analysis: J. Chang, Zhu, Li, A. L. S. Chang.
Obtained funding: A. L. S. Chang.
Administrative, technical, or material support: Kim, A. L. S. Chang.
Study supervision: A. L. S. Chang.
Conflict of Interest Disclosures: Dr A. L. S. Chang reported that she is a clinical investigator for studies sponsored by Genentech, Novartis, Merck, and Eli Lilly. No other disclosures were reported.
Funding/Support: This study was funded by the Stanford University School of Medicine Department of Dermatology (Dr A. L. S. Chang) and Department of Pathology’s Dermatopathology Service (Dr Kim).
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Meeting Presentation: This study was presented at the annual Meeting of the American Academy of Dermatology; March 4, 2017; Orlando, Florida.