PD-L1 Expression with Epithelial Mesenchymal Transition of Circulating Tumor Cells Is Associated with Poor Survival in Curatively Resected Non-Small Cell Lung Cancer

In addition to the FDA-approved definition of a circulating tumor cell (CTC), various CTC phenotypes have been discovered. Epithelial-mesenchymal transition (EMT) of cancer cells is directly linked to PD-L1 upregulation. The goal of the study was to investigate PD-L1 expression and EMT in CTCs of non-small cell lung cancer (NSCLC) patients, and perform an outcome analysis. Prospectively, 7.5 mL peripheral blood was collected from 30 NSCLC patients that underwent surgery and 15 healthy controls. CTCs were enriched by size-based microfilter and immunofluorescence stainings performed (cytokeratin (CK) 8/18/19, EpCAM, CD45, PD-L1, EMT markers vimentin, and N-Cadherin, DAPI). Patient-matched NSCLC tissues were also stained. CTC staining intensity was quantified with a software and correlated with patient-matched NSCLC tissues and survival. PD-L1 and EMT markers were expressed at significantly higher proportions in CTCs than patient-matched NSCLC tissues (p < 0.05); ≥3 PD-L1pos/EMTposCTCs were associated with significantly poorer survival after curative surgery (p < 0.05). No CTCs were detected in 15 healthy controls. This study shows that PD-L1 expression and EMT of CTCs is a negative survival predictor for NSCLC patients. The therapeutic role of the molecular linkage of PD-L1 and EMT will need to be further investigated, as linked pathways could be targeted to improve NSCLC outcome.


Introduction
Lung cancer is by far the leading cause of cancer-related deaths, with non-small cell lung cancer (NSCLC) being the most common subtype. Circulating tumor cells (CTCs) have detached from a tumor and are found in the blood of cancer patients. After FDA clearance of the CellSearch ® system for CTC enumeration in 2007, CTCs are defined as cells that express cytokeratins (CKs) 8/18 and/or 19 and as vimentin and N-Cadherin, are upregulated. In different solid cancers, we and other groups have identified CTCs that have undergone EMT [4,6,7,11]. Checkpoint inhibition of PD-1/PD-L1 interaction has proven to be a successful immunotherapeutic approach in metastatic lung cancer treatment [12,13]. PD-L1 pos CTCs can be detected in cancer patients, and are associated with a poorer prognosis in NSCLC patients undergoing anti-PD-1 treatment [14,15]. There are data demonstrating that EMT of cancer cells is associated with PD-L1 upregulation, inducing immune tolerance towards cancer cells [16][17][18][19][20][21]. In NSCLC patients, EMT-CTCs have also been observed to co-express PD-L1 [22,23]. These findings led to our hypothesis that PD-L1 pos /EMT pos CTCs are associated with poorer survival of NSCLC patients after curative surgery.
In a prospective trial of 30 surgically treated NSCLC patients, CTC and matched tumor tissue expression of PD-L1 and EMT markers vimentin and N-Cadherin was compared, and CTC phenotypes correlated with survival after curative treatment of NSCLC patients. Results suggest that PD-L1 is significantly upregulated in EMT-CTCs, and PD-L1 pos /EMT pos CTCs are associated with poor survival of NSCLC patients.

CTCs Express PD-L1 and EMT Markers Vimentin and N-Cadherin at Higher Rates than Matched Primary NSCLC Tissue
Patient-matched NSCLC tumor tissues (n = 30) were harvested at the time of surgical resection and stained for PD-L1, EMT markers vimentin, and N-Cadherin (Table 2; Figures 2, 3). Positive expression of PD-L1 was noted in 14/30 (46.7%), whereas EMT markers were observed in lower frequencies: Vimentin in 2/30 (6.7%) and N-Cadherin in 4/30 (13.3%) of NSCLC tissues (Table 2). No NSCLC tumor tissue was found to be triple PD-L1 pos /vimentin pos /N-Cadherin pos . Expression proportion scores (%) of PD-L1 pos CTCs, vimentin pos CTCs, and N-Cadherin pos CTCs or tissue tumor cells of all CK pos /EpCAM pos /CD45 neg CTCs or all tissue tumor cells were determined (Table 2)    Different CTC expression patterns with regard to PD-L1, vimentin, and N-Cadherin are shown.

CTCs Express PD-L1 and EMT Markers Vimentin and N-Cadherin at Higher Rates than Matched Primary NSCLC Tissue
Patient-matched NSCLC tumor tissues (n = 30) were harvested at the time of surgical resection and stained for PD-L1, EMT markers vimentin, and N-Cadherin (Table 2; Figures 2, 3). Positive expression of PD-L1 was noted in 14/30 (46.7%), whereas EMT markers were observed in lower frequencies: Vimentin in 2/30 (6.7%) and N-Cadherin in 4/30 (13.3%) of NSCLC tissues (Table 2). No NSCLC tumor tissue was found to be triple PD-L1 pos /vimentin pos /N-Cadherin pos . Expression proportion scores (%) of PD-L1 pos CTCs, vimentin pos CTCs, and N-Cadherin pos CTCs or tissue tumor cells of all CK pos /EpCAM pos /CD45 neg CTCs or all tissue tumor cells were determined (Table 2)

Commented [m8]
in this manuscript

Discussion
Despite successful implementation of promising novel targeted agents for NSCLC therapy, overall five-year survival of lung cancer remains at approximately 18% only [24]. After FDA clearance of the CellSearch ® system for CTC detection in 2007, CTCs were traditionally defined as cells that are CK pos /EpCAM pos /CD45 neg with a DAPI pos nucleus [1]. However, these criteria may not include CTCs that have undergone EMT [2][3][4]. In a prospective trial of 30 NSCLC patients that underwent curative lung surgery, we demonstrated that PD-L1 and EMT markers vimentin and N-Cadherin are

Discussion
Despite successful implementation of promising novel targeted agents for NSCLC therapy, overall five-year survival of lung cancer remains at approximately 18% only [24]. After FDA clearance of the CellSearch ® system for CTC detection in 2007, CTCs were traditionally defined as cells that are CK pos /EpCAM pos /CD45 neg with a DAPI pos nucleus [1]. However, these criteria may not include CTCs that have undergone EMT [2][3][4]. In a prospective trial of 30 NSCLC patients that underwent curative lung surgery, we demonstrated that PD-L1 and EMT markers vimentin and N-Cadherin are upregulated in CTCs in comparison to tumor tissue, and that ≥3 PD-L1 pos /EMT pos CTCs in 7.5 mL blood drawn before curative surgery are associated with poor overall survival.
In the present study, we applied a validated microfilter CTC isolation method that has proven to provide consistent CTC detection results in our own and other investigator's studies [25][26][27]. As initially described by Adams and colleagues, this microfilter analysis allows quenching of fluors, and restaining of CTCs for additional biomarkers using a cost-effective and straightforward protocol with borohydride [9]. CTCs have very heterogeneous pheno-and genotypes, with certain CTC subgroups having more metastatic potential than others [3]. EMT is a critical metastatic phenotype change required for detachment of cancer cells from the primary epithelial tumor [10,28]. During EMT, mesenchymal markers, such as vimentin, N-Cadherin, SNAIL, fibronectin, or β-catenin are upregulated. In different solid cancers, we and other groups have identified CTCs that carry an EMT phenotype [6,7,29,30]. Presence of ≥5 EMT-CTCs is associated with progressive disease in metastatic colorectal cancer patients [11]. Checkpoint inhibition of PD-1/PD-L1 interaction has proven to be a successful immunotherapeutic approach in metastatic lung cancer treatment [12,13]. In a seminal study, CTC PD-L1 expression was initially described in breast cancer patients [15]. Anti-PD-1 treatment responses in NSCLC are also noted if tumor tissues lack PD-L1 expression, and our present and other recent studies have confirmed that PD-L1 expression is higher in CTCs than in tumor tissue [31,32]. Also, pre-treatment PD-L1 pos CTCs are associated with a poorer prognosis in NSCLC patients undergoing anti-PD-1 therapy with nivolumab [14]. These findings suggest that PD-L1 upregulation in CTCs leads to a survival benefit of metastatic cells [15]. Unsurprisingly, the response to PD-1 inhibitor therapy correlates with a reduction in PD-L1 pos CTC [33,34].
It has been demonstrated that EMT is associated with increased expression of multiple immune checkpoints leading to immune tolerance [19,20]. Induction of EMT in epithelial cells leads to a PI3K/AKT pathway-dependent PD-L1 upregulation [16]. PD-L1-mediated CD8 pos tumor-infiltrating lymphocyte immunosuppression and subsequent metastasis is inhibited by EMT suppressing microRNA-200 [21]. NSCLC tissue analyses demonstrated association of PD-L1 expression with an EMT phenotype [18,35]. In our NSCLC cohort, we have now confirmed other investigators' findings that CTCs coexpress PD-L1 and EMT markers vimentin and N-Cadherin as a possible mechanism for immune escape [23]. As a novelty, our study shows a significant association of PD-L1 pos /EMT pos CTCs with NSCLC overall survival after curative surgical treatment. As it has been described by other investigators, we also observed that cellular PD-L1, vimentin, and N-Cadherin expression patterns in CTCs is differential (membranous, cytoplasmic, nuclear) [23,29,[36][37][38]. In our analysis, PD-L1 and EMT marker positivity was defined independent from membranous, cytoplasmic, and nuclear localization. But in a recent publication, it was reported that PD-L1 expressed in the nucleus of vimentin-positive CTCs predicts poor prognosis in colorectal and prostate cancer patients [22]. The significance of differential cellular expression has to be further clarified, but investigators suggested that nuclear translocation of PD-L1 might cause resistance to T cell induced cytotoxicity and inhibit drug-induced apoptosis [36,39]. Future mechanistic studies will have to shed further light on the biology of these EMT and immune checkpoint interactions on a molecular level.

Patient Selection Criteria
The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of the University of Missouri and Truman VA Hospital (Institutional Review Board (IRB) approval numbers: IRB2010166, IRB2004401-VA). All subjects gave their informed consent for inclusion before they participated in the study. Trials were registered at ClinicalTrials.gov (NCT02838836/NCT03551951). Recruitment timeframe was from July 2016 to May 2018. Thirty NSCLC patients that underwent surgery for AJCC stage I-IIIA were prospectively included for analysis. Fifteen healthy never-smokers were included as negative controls.

Clinicopathological Data
Data were collected by reviewing electronic records, including imaging results and other relevant medical findings. The staging manual of the American Joint Committee on Cancer (AJCC), 8th Edition, was used. All cases were reviewed in multidisciplinary thoracic oncology conferences. Clinical follow-up data were obtained. Survival and recurrence data were gathered by reviewing the hospital records, direct communication with the treating physicians, and from the Cancer Registry of the State of Missouri. Overall survival was calculated from the date of surgical excision of the tumor to the date of death or last follow-up. NSCLC recurrences (loco-regional or distant metastases) were calculated from the date of surgical excision of the tumor to the date of diagnosis of the recurrences.

CTC Detection and Staining PD-L1 and EMT Markers Vimentin and N-Cadherin
Blood draws were done at the initial patient encounters in the outpatient clinics via peripheral vein phlebotomy before biopsies or surgeries were performed to avoid CTC spillage by tissue manipulation. The first 6 mL blood were discarded to avoid epithelial contamination from skin puncture. Then, 7.5 mL of blood were collected in CellSave ® tubes (Menarini-Silicon Biosystems, Huntingdon Valley, PA, USA) (as recommended by the manufacturer of CellSieve ™ microfilters (Creatv MicroTech, Rockville, MD, USA)). Within 1 hour, CTCs were enriched by size with an established microfilter platform that has been successfully used by other investigators and our group [25,27,40,41]. Blood was passed through filters with a syringe pump (KD Scientific Legato 110 CMT, Analytical West Instruments, CA, USA). Enriched cells that were adherent to the filter were characterized by immunofluorescence staining on the filter with an antibody cocktail (CK8/18/19-FITC, EpCAM-PE, CD45-Cy5). Microfilters were then transferred on a microscopic slide, anti-fade mounting medium with DAPI (Cell Signaling Technology, Danvers, MA, USA) added, a coverslip placed on the filter, and incubated in the dark at room temperature for 8 hours. In alignment with the FDA-approved definition of a CTC, we defined a traditional CTC as CK pos /EpCAM pos /CD45 neg with a DAPI pos nucleus inside the cytoskeleton [1]. Immunostainings were standardized using human NSCLC cell line A549 (ATCC ® CCL-185™) that were spiked into whole blood of healthy volunteers.
For CTC PD-L1, vimentin, and N-Cadherin expression analysis, previously stained filters underwent fluorescence quenching with a simple, cost-effective, and established protocol using sodium borohydride [9]. This protocol allows multi-phenotype analysis of CTCs using sequential fluorescent quenching and restaining for additional biomarkers. Following quenching of immunofluorescence staining for CK/EpCAM/CD45, immunofluorescence restaining with an antibody cocktail containing anti-PD-L1 Alexa Fluor 488 (rabbit mAb; clone D8T4X; Cell Signaling), anti-vimentin Alexa 555 (rabbit mAb; clone D21H3; Cell Signaling), anti-N-Cadherin Alexa 647 (rabbit mAb; clone EPR1791-4; abcam), and DAPI was performed. PD-L1, vimentin, and N-Cadherin expression was analyzed on CTCs that had undergone quenching and restaining, but were previously identified as CK8/18/19 pos /EpCAM pos /CD45 neg with a DAPI pos nucleus. As recently described by other investigators, we also observed differential membranous, cytoplasmic, and/or nuclear expression patterns of PD-L1, vimentin, and N-Cadherin [23,29,[36][37][38]. The MetaMorph Microscopy Image Analysis Software (v.7.8.12) (Molecular Devices, San Jose, CA, USA) was used for fluorescence intensity quantification of the CTCs that were re-stained for PD-L1, vimentin, and N-Cadherin. CTC expression positivity was defined as ≥50% mean immunofluorescence intensity of the mean fluorescence intensity of all CTCs analyzed, independent of membranous, nuclear, or cytoplasmic expression localization. Proportions (in percentages) of PD-L1/vimentin/N-Cadherin pos CTCs of all CK pos /EpCAM pos /CD45 neg CTCs were calculated. CTCs were defined as EMT pos if they co-expressed both vimentin and N-Cadherin. Images were taken using Zeiss Axiovert 200M inverted microscope (Oberkochen, Germany) equipped with a Hamamatsu ORCA-ER CCD (Shizuoka, Japan) camera. All analyses were performed blinded to reduce observer bias.

NSCLC Tissue Immunostaining
Patient-matched NSCLC tumor tissues and adjacent normal lung tissue were snap-frozen in liquid nitrogen and stored at −70 • C. Tissue pieces were then placed in OCT Embedding Compound, and 5 µm thick sections were cut with a cryostat. Presence of tumor was confirmed after hematoxylin and eosin (H&E) staining by a pathologist. Slides were treated with citrate-based antigen unmasking solution (Vector Laboratories, Burlingame, CA, USA). Sections were blocked non-specific proteins with horse serum, and treated with BLOXALL Endogenous Peroxidase Blocking Solution (Vector). Anti-PD-L1 antibody (rabbit mAb; clone 73-10; abcam, Cambridge, MA, USA), anti-vimentin (mouse mAb; clone V9; abcam), and anti-N Cadherin (mouse mAb; clone 5D5; abcam) were incubated at 4 • C overnight. Rabbit or mouse IgG (Santa Cruz Biotechnology, Dallas, TX, USA) were used as isotype controls. ImmPRESS™ HRP Anti-Rabbit/-Mouse IgG Peroxidase Polymer Detection Kit (Vector) were used as secondary antibodies, ImmPACT DAB Peroxidase (HRP) substrate for detection, and hematoxylin (Vector) as counterstain. PD-L1, vimentin, and N-Cadherin expression were determined by using a tumor proportion score as applied in the diagnostic clinical setting: Percentage of tumor cells showing partial or complete membranous, cytoskeletal, and/or nuclear staining for PD-L1, vimentin, and N-Cadherin [13]. Normal lung tissues were stained for PD-L1, vimentin, and N-Cadherin as controls, and expected patterns of expression were observed [42].

Statistical Analysis
For this study, key elements of the prospective-specimen-collection, retrospective-blinded-evaluation (PRoBE) design were applied [43]. With this methodology, specimens and clinical data were collected prospectively, with the analytic personnel being without knowledge of the outcome (blinded). Statistical tests performed were the non-parametric Tukey's multiple comparison analysis, Mann Whitney test, Friedman test, non-parametric Wilcoxon signed-rank test for matched pairs, and Fisher's exact test. Survival was analyzed by the Kaplan-Meier curve method, and the log-rank test was used for univariate analysis. Significance statements refer to a p-value of <0.05. Statistical analyses were performed using software Prism version 5.00 (GraphPad, La Jolla, CA, USA).

Conclusions
In summary, data from this study support recent findings that malignant progression is driven by EMT and PD-L1, and is associated with poorer survival of NSCLC patients. These observations might open novel avenues in liquid biopsy profiling to select NSCLC patients for immunotherapies or even combinational treatments targeting EMT and the PD-1/PD-L1 axis. In hope of future improvement of currently disappointing NSCLC outcomes, molecular studies may clarify the underlying relationship of EMT and immune tolerance towards cancer cells.