PD-L1 Polymorphism Can Predict Clinical Outcome of Brazilian Non-Small Cell Lung Cancer Patients After Postoperative Adjuvant Treatment

Although the incidence and mortality of non-small cell lung cancer (NSCLC) remains high with poor prognosis, programmed death ‐ 1 (PD ‐ 1) and programmed death ‐ ligand 1 (PD ‐ L1) pathway are promising prognostic and predictive biomarker of NSCLC. The polymorphisms on PD ‐ L1 gene may be associated with their protein expressions and affect the adjuvant treatment and outcome of NSCLC patients. The goal of this study was to evaluate the clinicopathologic values of PD-L1 expression and single-nucleotide polymorphisms (SNPs) in the PD-L1 gene in lung adenocarcinoma (ADC), squamous cell carcinoma (SqCC), and large cell carcinoma (LCC). The 70 NSCLC samples consisted of 33 samples of ADC, 24 of SqCC and 13 of LCC. All tissue microarray paran blocks were used for PD-L1 multiplex immunouorescence assays with Cell Signaling E1L3N clone. Fifteen polymorphisms in the PD-L1 gene were investigated by deep targeted sequencing. respectively). The rs7041009 AG genotype was inversely associated with E1L3N clone PD-L1 labeling in NSCLCs but did not reach statistical signicance. Kaplan-Meier survival curves showed that the prognosis of patients with NSCLC was strongly dependent on the alternative allele G (genotype AG or GG) of rs4742098, alternative allele C (genotype CG or of rs4143815 and individuals with GG genotype of rs7041009 (P=0.02; P=0.05 and P<0.01, respectively). Multivariate analysis the Cox Regression model showed genotype GG of rs7041009 associated with cancer recurrence and patients without adjuvant radiotherapy (hazard ratio [HR] = 7.59, 95% condence interval [CI] = 1.06-54.03, and HR = 9.24, CI = 1.13-75.16). Samples were histologically reviewed by pathologists before nucleic acid extraction in order to select samples with at least 30% of lung cancer cells. We extracted DNA from snapped frozen samples obtained from 70 of the 164 patients in our study. According to the pathological reports, the samples included adenocarcinoma (ADC, n = 33), squamous cell carcinoma (SqCC, n = 24), and large cell carcinoma (LCC, n = 13). The clinical information and basic characteristics of the patients were recorded by physicians. rs4143815 and rs7041009 signicantly associated with tumor recurrence; 3) different frequencies of genotypes rs4742098 A/A, rs4143815 G/G, and rs7041009 A/G or A/A genotypes; 4) G allele of rs7041009 signicantly correlated with younger patients and status; 5) rs7041009 A > G genotype inversely associated with E1L3N PD-L1 labeling in NSCLCs; 6) prognosis of patients with NSCLC strongly dependent on the genotype; and 7) genotype G/G of PD-L1 rs7041009 independently associated with risk of cancer recurrence and patients without postoperative radiotherapy.


Background
Lung cancer is still the leading cause of cancer-related mortality worldwide with a poor prognosis [1]. Most of the patients with non-small cell lung cancer (NSCLC), which represents 85% of all patients with lung cancer, are diagnosed at advanced stages or are presenting with recurrent disease after initial locoregional treatment [2]. Over the past few decades, conventional chemotherapy, mainly platinum-based, was the only therapeutic option for those not eligible for radical intent treatment, with limited e cacy and very few long-term survivors-5year overall survival < 15% -, and highlighting the fact that therapeutic options beyond rst-line treatment remained limited for patients [3].
The blockade of immune checkpoints has recently been applied as an antitumor immune therapy. Moreover, various immune checkpoint molecules related to tumor immune evasion have recently been identi ed. Tumor cells express the programmed death-1-ligand 1 (PD-L1) as an adaptive, resistant mechanism to suppress the inhibitory receptor, namely programmed cell death-1 (PD-1) in order to evade host immunosurveillance [4]. Because PD-L1 up-regulation by tumor cells can protect tumor cells from antitumor immune response, blockade of PD-L1/ PD-1 interactions have been recently selected for antitumor immune therapy [5]. Furthermore, Agents targeting the PD-1/PD-L1 signaling have shown promising response in NSCLC treatment. PD-L1 protein expression may be predictive for a blockade of the PD-L1/PD-1 interactions [6]. However, this biomarker is not su ciently robust and demonstrates many challenges.
In this context, some studies demonstrated that patients with more than 50% PD-L1 positive tumor cells are non-responders to anti-PD-1/PD-L1 treatment. In contrast, others shown that patients whose tumors do not express PD-L1 are good responders [7][8][9]. To explain the con ict that affects the PD-L1 expression, some studies have considered that the heterogeneity of axis expression and response to PD-1/PD-L1 treatment in NSCLC depends on alterations in other factors, such as more precise methods to investigate immune evasion mechanisms and immune microenvironment, as well as immune checkpoint genomic pro le and genetic variants regarding PD-L1 gene [10][11][12][13].
Genetic variants might affect the normal process of gene activation and transcriptional initiation, hence in uence the quantity of mRNA and encoded protein [14]. Thus, several studies investigated the association between genetic polymorphisms of PD-1 and PD-L1 genes and the risk of various cancers, but the nding was not completely elucidated yet [15].
A recently reported study suggested that PD-L1 polymorphism may predict chemotherapy response and survival of advanced stage NSCLC patients after rst-line paclitaxel-cisplatin [16]. The occurrence of polymorphisms in genes that are involved in immune checkpoints may affect the antitumor immune activity, thereby prognosis of patients with early stage NSCLC after surgery.

Patient Samples
We obtained archival formalin-xed para n-embedded histologic tumor sections from 164 patients diagnosed with NSCLC who had undergone surgical resection between January 1, 1995, and December 31, 2015, and who had not received neoadjuvant therapy. Patients had been treated at the Hospital das Clinicas of the State University of São Paulo Medical School (HC-FMUSP), the Heart Institute of the University of São Paulo (INCOR), the Cancer Institute of São Paulo (ICESP), and the A.C. Camargo Cancer Center, São Paulo, Brazil.
Samples were histologically reviewed by pathologists before nucleic acid extraction in order to select samples with at least 30% of lung cancer cells. We extracted DNA from snapped frozen samples obtained from 70 of the 164 patients in our study. According to the pathological reports, the samples included adenocarcinoma (ADC, n = 33), squamous cell carcinoma (SqCC, n = 24), and large cell carcinoma (LCC, n = 13). The clinical information and basic characteristics of the patients were recorded by physicians.

Dna Extraction
Genomic DNA (gDNA) was extracted from frozen NSCLC tissue using the QIAamp DNA Mini Kit (Qiagen) following the manufacturer's recommendations. DNAs were quanti ed by Qubit (Life Technologies) and integrity was assessed by BioAnalyzer 2100 (Agilent Technologies).
NGS data analysis was performed on Molecular Genetics and Bioinformatics Laboratory, Experimental Research Unit (UNIPEX) at Medical School of São Paulo State University (UNESP). Sequencing quality was assessed by FastQC. Reads were aligned to the human genome (hg19, GRCh37) with BWA software and SAMtools was used to convert the alignment results to BAM format [17]. Mapped reads underwent variant calling for SNP with GATK command line tools, including HaplotypeCaller, SelectVariants, and VariantFiltration programs using default parameters. After calling step, the variants were annotated using the VEP [18] software. Coverage depth was a priori settled up at 100×. Filtered retained variants had to have > 10 reads of position depth (PD) and/or > 6 reads of allele depth (AD) and/or an AD/PD ratio of > 0.05 and/or a population frequency higher than 1% (popfreq_all > 0.01).

Multiplex Immuno uorescence Staining
Multiplex immuno uorescence (mIF) staining was performed using similar methods that have been previously described and validated [19,20]. Four micrometer-thickness consecutive TMA sections were stained using an automated staining system (BOND-RX; Leica Biosystems, Buffalo Grove, IL) to characterize PD-L1 (clone E1L3N, dilution 1:100; Cell Signaling Technology, Danvers, MA). The marker was stained in sequence using their respective uorophore containing in the Opal 7 kit (catalogue #NEL797001KT; Akoya Biosciences/PerkinElmer, Waltham, MA) after completely validation using immuno uoresce (IF) to obtain an uniform, speci c, and correct signal across all channel to be well balanced staining pattern during the multiplex staining [19,20]. The correct signal from the uorophores, were also de ned and optimized between 10 to 30 counts of intensity, to maintain a good balance and similar thresholds of intensity across all the antibodies.
In parallel, to detect possible variations in staining and optimal separation of the signal, positive and negative (auto-orescence) controls were included during the staining to make sure, especial with the internal positive control that all the antibodies are working together. Autouorescence controls with an expected spectral of 488 nm will be able to accurate remove the auto-uorescence from all the label signals during the analysis process. The stained slides were scanned using a multispectral microscope, Vectra Polaris 3.0 imaging system (Akoya Biosciences/PerkinElmer PerkinElmer), under uorescence conditions.

Multiplex Immuno uorescence Quantitation
Multispectral images of tumor sections from each core were analyzed with inForm 2.2.1 (Akoya/PerkinElmer) software Individual cells, which were de ned by nuclei staining and identi ed by the InForm cell segmentation tool, were subjected to a phenotyping patternrecognition learning algorithm to characterize co-localization of the various cell populations using panel labeling [20,21]. The panel labeling was as follows: Malignant cells, (MCs) with the AE1/AE3 + marker, including those with and without PD-L1 expression (AE1/AE3 + PD-L1 + and AE1/AE3 + PD-L1-, respectively).
The individual cell phenotype report produced by the InForm software was processed using Excel 2010 (Microsoft. Houston, TX), and a nal summary of the data, which contained the median of each individual phenotype (given as number of cells/mm 2 ) and the percentage of macrophages and MCs expressing PD-L1, was created for statistical analysis. If the percentage of MCs or macrophages expressing PD-L1was greater than the median value, the PD-L1 expression was considered positive. If the percentage of macrophages or MCs expressing PD-L1 was lower than or equal to the median, the PD-L1 expression was considered negative.

Statistical Analysis
The statistical analysis of allelic and genotypic frequencies of PD-L1 polymorphisms in NSCLC group was calculated by Hardy Weinberg equilibrium ([1 _ (hC 2H)]/2N, where "h'' represents as heterozygous genotype, "H'' for homozygous and "N'' represents the number of samples. Associations between polymorphisms, PD-L1 protein expression, and clinicopathological parameters of NSCLC patients were investigated by Fisher exact test. Prognostic value was assessed by survival analysis using the Kaplan-Meier method with the log-rank test for statistical signi cance. In addition, we used Cox's proportional hazards regression models were used in a multivariate analysis to test the association of the SNPs with OS. OS was de ned as the time from curative surgery to the date of death or censored on the last date known to be alive. The statistical software programs IBM SPSS (version 22; Armonk, NY, USA) was used to perform all analyses. Differences were considered statistically signi cant at P < 0.05.

Clinicopathologic characteristics
The clinical characteristics of the patients in our cohort are summarized in Table 1, strati ed by histologic types. Of 70 patients included in the study, 24 (32.9%) had SqCC, 13 (17.8%) had LCC, and 33 (45.2%) had ADC. Among 24 SqCC, 9 (22.5%) were more frequent in males, while the 33 ADC cases were equally distributed among the genders 10 (25.0%). All histological types were more frequent in patients with 63 years or younger. Regarding the tobacco smoking history, 8 (30.8%) occurred both in patients with ADC and patients with SqCC and 3 (11.5%) in patients with LCC. All the histological subtypes presented more cases with an advanced stage of disease (9 cases to ADC, 6 cases to SqCC, and 4 cases to LCC). Most of the patients do not receive chemotherapy (12 cases to ADC, 8 cases to SqCC, and 6 cases to LCC) or radiation therapy (18 cases to ADC,10 cases to SqCC and LCC) as adjuvant treatment. According to PD-L1 protein expression in malignant cells, LCC with 7 (13.2%) presented more relevant expression, above the median, when compared to other histological subtypes. The median follow-up of the patients was 66 (12-144) months. Follow-up information about survival was missing for 45 patients, 15 (20.5%) patients died after disease progression. In all analyzes, there were no signi cant differences between the histological types (P > 0.05).

Allele And Genotype Distributions
All NSCLC patients who were underwent a surgical resection were successfully genotyped for the fteen PD-L1 SNPs rs76805387T > C, rs4742098A > G, rs47946526A > G, rs10217310G > T, rs7864231G > A, rs41280725C > T, rs573692330A > G, rs1011769981G > A, rs41280723T > C, rs138135676T > C, rs4143815G > C, rs2297136G > A, rs148242519G > A, rs41303227C > T and rs7041009G > A. The Supplementary Table 1 shows the SNP identi cation numbers, allele and genotype frequencies and P-value for HWE. Of the fteen polymorphisms studied, eleven were monomorphic. Similar ndings could also be concluded in their allele frequencies. The allele frequency of our cohort was compared to different populations of the 1000 Genomes Project (Supplementary Table 2). The monomorphic SNPs were excluded from further analysis. For the polymorphic SNPs, all were found to be in equilibrium (P > 0.05) for HWE.

Correlation between PD-L1 gene polymorphisms and clinicopathological characteristics in NSCLC
Strati ed analyses of the associations between clinical characteristics and four PD-L1 polymorphisms presented different genotypic distributions rs4742098A > G, rs4143815G > C, rs2297136G > A and rs7041009G > A were performed. Table 2 shows the SNP genotype frequency and the associated clinicopathological characteristics. Three of the four PD-L1 gene polymorphisms rs4742098A > G, rs4143815G > C, and rs7041009G > A were signi cantly associated with relapse (P = 0.01; P = 0.05; P = 0.02, respectively). According to the recurrence of the disease, for the rs4742098, carriers of the G allele (individuals with the AG or GG genotype) were less likely to relapse (P = 0.01) compared to the homozygous genotype AA. As with rs4143815, individuals with the alternative C allele (genotypes CG or CC) were also less likely to relapse (P = 0.05). As for the rs7041009, carriers of the alternative allele A (genotypes AG or GG) more likely to exhibit relapse. The genotype GG (reference) of rs7041009 also showed a signi cant correlation with age, among younger patients, 16 (41.0%) and status, among patients that are alive, 11 (39.3%) compared to carriers of the A allele (P = 0.02 and P < 0.01, respectively). In these patients a marginal signi cance was found for smokers (P = 0.09) and adjuvant chemotherapy (P = 0.07). No was observed signi cant association between rs2297136 genotypes and clinicopathological variants.

Correlation between PD-L1 protein expression and PD-L1 gene polymorphisms
The correlation between PD-L1 protein expression and PD-L1 gene polymorphisms are shown in Table 2. No statistically signi cant association between PD-L1 protein expression in malignant cells and PD-L1 gene polymorphisms was observed.
In our cohort, the four PD-L1 gene polymorphisms, apparently, are not interfering in the expression of PD-L1 in malignant cells of NSCLC. However, when we correlate the PD-L1 protein expression among the ADC, SqCC and LCC histological subtypes, we observed that the expression was below the median for malignant cell of all histological subtypes.

Survival Analysis
In the rst statistical test, were examined the individual effect of patient characteristics, such as age, sex, tobacco history, histology, clinical stage, treatment and polymorphisms to estimate statistical differences using Kaplan-Meier survival curves ( Table 3). The results of this analysis showed that the prognosis of patients with NSCLC is strongly dependent on the alternative allele G (genotype AG or GG) of rs4742098, alternative allele C (genotype CG or CC) of rs4143815 and individuals with GG genotype of rs7041009 (P = 0.02; P = 0.05 and P < 0.01, respectively). Age presented a signi cant marginal value (P = 0.07), and patients younger than 63 years seemed to have an increased overall cancer survival than patients older than 63 years. Radiotherapy also presented marginally signi cant value (P = 0.09), and patients treated with radiotherapy presented lower survival than patients without treatment. Table 3 Survival analysis with strati cation of the variables in optimal upper and lower binary cut-off limits by the Kaplan-Meier method strati ed for relapse occurrence and the differences by means of the log-rank test.  [17].
As observed in Fig. 1 Table 4). Thus, patients with NSCLC presenting PD-L1 rs7041009 polymorphism who did not receive treatment with radiotherapy were associated with the risk of cancer recurrence.

Discussion
The present work showed that the association between clinical data and single-nucleotide polymorphisms (SNPs) in the PD-L1 gene in uences the risk of disease recurrence and adjuvant treatment. Seventy diagnosed and surgically removed NSCLC, were used for PD-L1 multiplex immuno uorescence (mIF) assays and NGS sequencing. Under the conditions of the present investigation, we found a variety of associated effects between polymorphism in the PD-L1 gene pro le and behavior of NSCLC. They included: 1) PD-L1 protein expression higher in ADC than in SqCC and LCC; 2) three of the four PD-L1 gene polymorphisms rs4742098, rs4143815 and rs7041009 signi cantly associated with tumor recurrence; 3) different frequencies of genotypes rs4742098 A/A, rs4143815 G/G, and rs7041009 A/G or A/A genotypes; 4) G allele of rs7041009 signi cantly correlated with younger patients and status; 5) rs7041009 A > G genotype inversely associated with E1L3N PD-L1 labeling in NSCLCs; 6) prognosis of patients with NSCLC strongly dependent on the genotype; and 7) genotype G/G of PD-L1 rs7041009 independently associated with risk of cancer recurrence and patients without postoperative radiotherapy.
To better understand the role of SNPs in the PD-L1 gene on the risk of disease recurrence and adjuvant treatment, our study was carried out in a cohort of patients including metastatic and non-metastatic NSCLC. We analyzed the tumor pro le in 70 NSCLC patients for the occurrence of PD-L1 gene polymorphisms in fresh-frozen lung cancer tissue. Our investigation included ADC, SqCC, and LCC, and to our knowledge is the rst study comparing SNPs in the PD-L1 gene among the three major histological types of NSCLC, and its impact on disease progression.
In this study, we detected that positive malignant cells expressing PD-L1 (MCs PD-L1) using Cell Signaling E1L3N clone, mIF and image analysis were positively correlated with ADC above the median in 11 (20.8%) when compared to others histological subtypes. In fact, Shimoji et al [22] showed that PD-L1 expression using the Cell Signaling E1L3N clone was signi cantly correlated with a solid, vimentin expression, increased Ki-67 labeling index and poor prognosis in ADC but not in SqCC. Other studies have reported that IHC staining for PD-L1 was detected at signi cantly higher frequencies in SqCC than in ADC of the lung [22,23]. Cha et al. [24] found that PD-L1 expression of the Ventana Laboratory SP142 clone was signi cantly associated with an ADC solid subtype histology, p53 aberrant expression, and poor prognosis.
In our cohort, we also evaluated the relationship between genotype distributions of PD-L1 SNPs and NSCLC patients who underwent a surgical resection. Actually, gene polymorphism is one of the most common gene variations, and functional polymorphisms may affect mRNA and protein expression levels [4,16,25,26]. We found that PD-L1 polymorphisms presented different genotypic distributions rs4742098A > G, rs4143815G > C, rs2297136G > A and rs7041009G > A. Interestingly, three of the four PD-L1 gene polymorphisms rs4742098A > G, rs4143815G > C and rs7041009G > A were signi cantly associated with disease recurrence. According to the recurrence of the disease, for the rs4742098, carriers of the G allele (individuals with the AG or GG genotype) were less likely to relapse compared to the homozygous genotype AA. Similar ndings were reported by Du and colleagues [25] for the rs4742098 SNP, which AG genotype differed from the AA genotype on evaluation of the risk of NSCLC recurrence.
Concerning rs4143815, we found that patients with the alternative C allele (genotypes CG or CC) were also less likely to disease relapse in agreement with Nomizo and colleagues report [26]. Regarding rs7041009, carriers of the alternative allele A (genotypes AG or GG) were more likely to exhibit relapse. The genotype GG (reference) of rs7041009 also showed a signi cant correlation with age, among younger patients, 16 (41.0%) and status, among patients that are alive, 11 (39.3%) compared to carriers of the A allele. The SNP rs7041009 was in the region of the transcription factor-binding site of PD-L1. We expected that the polymorphism of rs7041009 might change the binding power between transcription factors and PD-L1, which leads to the up-regulation or downregulation of PD-L1 transcription and PD-L1 protein expression. Unfortunately, we did not nd a difference in the positive rate of PD-L1 protein expression in patients with different genotypes of PD-L1. Apparently, this nding suggests that genotypes of PD-L1 do not interfere in the expression of PD-L1 in the malignant cells of NSCLC. However, when we correlate the PD-L1 protein expression among the ADC, SqCC and LCC histological subtypes, we observed that the expression was below the median for malignant cell of all histological subtypes eventually related to downregulation of PD-L1 transcription. In addition, the specimens from our patients were analyzed with mIF staining of the PD-L1 protein and quanti ed by image analysis, while other authors used IHC staining and H-score, which may have limited the statistical power.
Surgery is the standard treatment for patients with an early clinical stage and selected patients with stage IIIA disease with the possibility of complete tumor resection, but the overall 5-year survival remains below 50%. The addition of adjuvant cisplatin-based chemotherapy to surgery improved 5-year survival rates by 5-10%. Over the last decade, signi cant progress has been made in systemic therapies to improve the length of good-quality survival in patients with metastatic or recurrent NSCLC [27,28]. Following these systemic NSCLC treatments, patients with metastatic disease can be treated with the addition of local radiotherapy to improve treatment e cacy and patient survival compared to chemotherapy alone. More recently, adjuvant radiotherapy induced PD-L1 protein expression in initially PD-L1 negative tumor enabling, thereby subsequent effective immunotherapy [29]. However, not all NSCLC patients are good responders to immunotherapy, because the immune pro ling heterogeneity of expression probably depends on alterations in other factors, such as PD-L1 gene polymorphism.
In our cohort, Kaplan-Meier survival curves and log-rank tests showed a signi cant association between patients with NSCLC carrier of the alternative allele G (genotype AG or GG) of rs4742098, alternative allele C (genotype CG or CC) of rs4143815 and patients with GG genotype of rs7041009 with shorter disease-free survival. Similar ndings were reported by Lee and colleagues [16] who found PD-L1 rs4143815C > G signi cantly associated with worse survival outcomes in multivariate analyses. Moreover, patients with the CC and CG genotypes of PD-L1 rs4143815 showed slightly better responses to nivolumab [26]. In agreement with Lee and colleagues [16] we also found that patients with postoperative chemotherapy and radiotherapy presented a tendency of better survival compared to patients not treated.
Contrary to our result, another study found that rs822336 G to C substitutions was signi cantly associated with worse survival of earlystage non-small-cell lung cancer [16]. This indicated that the association between PD-L1 gene polymorphisms and survival in different tumors might not be consistent. Multivariate analysis of the survival risk (Hazard Ratio) based on signi cant and marginal signi cant factors at univariate analysis was examined by the Cox regression model. Initially, the model was constructed with patient age, radiotherapy and tumor relapse. In this situation, radiotherapy, and tumor recurrence were maintained as independent risk factors; age did not show a signi cant statistical value. Thus, PD-L1 polymorphisms were introduced into the model, controlled for age, radiotherapy and tumor relapse. This new situation showed that among the four PD-L1 polymorphisms genes, only genotype GG of PD-L1 rs7041009 was signi cant in predicting cancer recurrence in patients without adjuvant treatment with radiotherapy. Thus, patients with NSCLC presenting PD-L1 rs7041009 polymorphism who did not receive treatment with radiotherapy were associated with the risk of cancer recurrence.
Our analysis had several potential limitations. The rst and major limitation consisted that it was not possible to investigate the risk of lung cancer related to PD-L1 polymorphisms because we did not perform a case-control approach. Second, the modest sample size of 70 participants. However, the current study has strengths that should be mentioned; the analysis revealed that these SNPs had signi cant associations with clinicopathological features, including tumor relapse and overall survival of patients. Further studies are needed to provide additional insight that which gives strength to our ndings and help to understand the underlying mechanisms that cover SNPs of PD-L1 in determining the clinical outcome of relapse in patients with NSCLC. Kaplan-Meier survival curve according to PD-L1 rs4742098 G>A. Patients that were G allele carriers (AG+GG) had a higher survival rate compared to those who were AA genotype of rs4742098 (P=0.02).

Figure 2
Kaplan-Meier survival curve according to PD-L1 rs4143815 C>G. Patients that were C allele carriers (GG+CC) had a higher survival rate compared to those who were GG genotype of rs4143815 (P=0.05).