Cell‐blocks are suitable material for programmed cell death ligand‐1 immunohistochemistry: Comparison of cell‐blocks and matched surgical resection specimens in lung cancer

Abstract Objective Programmed cell death ligand‐1 (PD‐L1) has emerged as a predictive biomarker in lung cancer. PD‐L1 immunohistochemistry (IHC) assay predicts the response to immunotherapy, but cytology specimens are often the only samples available in a considerable proportion of advanced lung cancer patients. We delineate practical feasibility and efficacy of cytology cell‐block (CB) specimens for PD‐L1 expression and concordance between cytology CBs and surgical resection specimens. Methods In total, 58 eligible patients with primary lung cancer who received computed tomography‐guided percutaneous needle aspiration and surgery were included. PD‐L1 IHC (clone SP263) was performed on CBs prepared from residual liquid‐based cytology material and matched surgical resection specimens. PD‐L1 positive tumour cell proportion was categorised in four score groups: (a) <1%; (b) ≤1% to <10%; (c) ≤10% to <50%, (d) ≥50%. Results Comparison of PD‐L1 expression in cytology CBs and matched surgical resection specimens showed a high concordance (κ value 0.65). According to the therapeutic guideline of immunotherapeutic agents, a positive percent agreement was 94.34%, and a negative percent agreement was 100% at a cut‐off value for positivity of 1% PD‐L1 expression. There was a significant difference observed with regard to rates of PD‐L1 positivity when comparing smoking history (P = 0.02), age (P = 0.04) and pathological TNM stage (P = 0.04). Conclusions The results show that cytology CBs evaluated for PD‐L1 IHC assay have high concordance with matched surgical resection specimens and can be used for assessing PD‐L1 expression. Also, we propose that CBs are suitable materials for evaluating PD‐L1 expression while simultaneously performing both diagnostic and molecular tests.


| INTRODUC TI ON
Despite recent advances in molecular targeted therapies, lung cancer is the most common cause of cancer-related death. 1 The 5-year survival rates of lung cancer patients were reported as 4%-17% according to tumour stage and location. 2 Despite an initial dramatic response, almost all non-small cell lung cancer (NSCLC) patients harbouring an epidermal growth factor receptor (EGFR) mutation eventually showed disease progression due to acquired resistance to EGFR tyrosine kinase inhibitor treatment within 1 year. 3 In the absence of activating mutation of EGFR or translocation/gene rearrangements of anaplastic lymphoma kinase and ROS proto-oncogene 1, platinum-based chemotherapy remains the first-line treatment for advanced stage NSCLC. 4 It seems that ceaseless effort to overcome the limitation of conventional cytotoxic agents and target therapies may finally bear fruit. Recently, many clinical trials using immunotherapeutic agents, particularly immune checkpoint inhibitors, have shown signs of improvement. 2,[4][5][6][7][8] Programmed cell death ligand-1 (PD-L1) expression is a predictive biomarker for anti-PD-L1 immunotherapy in lung cancer. 9 Clinical trials with anti-PD-1 immune checkpoint inhibitors have shown an association between the magnitude of clinical efficacy and the level of PD-L1-expression as evaluated by PD-L1 immunohistochemistry (IHC). 4,6,7,10-12 PD-L1 IHC assay is a suitable diagnostic method for precise targeted therapy and is an easily performed, relatively inexpensive, and widely accessible method. According to recent reports, anti-PD-1 immune checkpoint inhibitors including pembrolizumab, nivolumab, atezolizumab and durvalumab resulted in significantly longer overall survival for patients with advanced NSLCL. [4][5][6][7][8]10 However, it is often difficult to obtain tumour tissue in patients with advanced NSLCL because of tumour location and patient comorbidity. 13 Therefore, diagnostic lung cancer samples are often very small biopsy tissues or cytology samples. In a considerable proportion of lung cancer patients, cytology specimens are the only samples available for diagnosis. In this setting, PD-L1 IHC on cytology specimens has become critical. Expert consensus is that cytology specimens are suitable not only for diagnostic purposes but also for molecular testing. 14,15 However, the use of cytology specimens for assessing PD-L1 expression is not advocated, because little is validated for this purpose.
In this study, we delineate the practical feasibility and efficacy of cytology cell-block (CB) specimens for PD-L1 expression and concordance between cytology CB and surgical resection specimens.
Finally, we would like to determine whether the PD-L1 IHC on cytology CBs can be a suitable method for making future immunotherapy treatment decisions.

| Patients and samples
After institutional review board approval was obtained, a retrospective review of pathology reports was done. A total of 58 patients underwent computed tomography (CT)-guided percutaneous needle aspiration (PCNA) and surgical resection from December 2011

| Cell-block preparation
All cytology specimens were obtained by CT-guided PCNA and fixed with CytoRich™ Red (Thermo Fisher Scientific). Each sample was processed for both one liquid-based cytology (LBC) and one CB slide. LBC slides were prepared using SurePath™ (Thermo Fisher Scientific, Becton, Dickinson and Co.) according to the manufacturer's instructions and were stained with Papanicolaou stain. The residual SurePath™ samples for CB slides were immediately centrifuged at 689g for 5 minutes and embedded into paraffin after fixation in 10% buffered neutral formalin for 6 hours at room temperature. Tissue processing was conducted by Leica PELORIS II Rapid Tissue Processor (Leica Microsystems). CBs were cut into 4μm thick sections and stained with haematoxylin and eosin using the standard method. All pathological slides were reviewed independently by two pulmonary pathologists. If there was a discrepancy in independent opinions, the slides were reviewed together to achieve a consensus. The consensus judgments were adopted as the final results.

| PD-L1 IHC
The scoring was performed in a blinded fashion for each other and for the relationship between cytology and histology pairs. Viable tumour cells exhibiting partial or complete membranous staining with any staining intensity were considered as positive.
Cytoplasmic staining without any membranous staining was excluded. PD-L1 expression was categorised in four score groups according to the following cut-offs: 1. score 0: no staining cells or less than 1% of tumour cells; 2. score 1: at least 1% of tumour cells but less than 10%; 3. score 2: at least 10% of tumour cells but less than 50%; 4. score 3: more than 50% of tumour cells.

| Statistical analysis
Statistical analyses were conducted using the statistical software

Details of clinical and pathological factors are shown in
The median follow-up after surgery was 28.6 months (range, 2-61 months). Cancer-related deaths occurred in two patients (3.45%), and one patient died of an unrelated cause. Kaplan-Meier survival analysis with log-rank test showed no statistically significant difference according to the expression rates of PD-L1.

| Comparison of PD-L1 expression between cytology CBs and matched surgical resection specimens
All cytology samples were obtained from the same anatomical site before surgical resection. The median time interval between two specimens was 42 days (range, 7-93 days). All patients received no treatment during the interval. Comparison of PD-L1 expression in cytology CBs and matched surgical resection specimens showed high concordance (κ value 0.65, Table 2). Examples of PD-L1 expression in cytology CBs and matched histology slides are shown in

| D ISCUSS I ON
The neoplastic process including tumour initiation, proliferation and migration of tumour cells depends not only on the evolving genomics and molecular properties of the tumours but also on their interaction with the tumour microenvironment, which is composed of immune cells, tumour cells, stromal cells and the extracellular matrix. 2,19 Recent studies have revealed that the interaction between tumour cells, and the immune system is especially worthy of notice. 12

| Comparison of PD-L1 expression
Objective assessment of PD-L1 expression has three major theoretical and practical issues; inter-assay variability because of different antibody affinities, tumour heterogeneity within the tumour, and different results that depend on the timing of the biopsy.
First, although inter-assay variability that results from different PD-L1 antibody affinities is known, 21 ported that the accuracy of the antibody SP263 was ,higher than that of other antibodies on formalin-fixed paraffin-embedded and cytology samples. 22 The PD-L1 expression of tumour cells within cytology specimens was found to be consistent with the PD-L1 expression of matching histology specimens, but the PD-L1 expression of immune cells within cytology specimens cannot replace results in matched histology specimens. 26 Here we investigated concordance rates of PD-L1 expression in tumour cells between cytology CBs and matched surgical resection specimens using an SP263 assay, based on findings of previous reports. [21][22][23][24][25][26] Second, recent studies reported cytology-histology correlation in lung cancer, but we have doubts about the rigorous control of heterogeneity in tumours. [25][26][27][28][29][30][31] These studies were done using samples collected by various methods. Cytology specimens were collected by fine needle aspiration, bronchial washing and bronchial brushing.
Surgical specimens were obtained by endobronchial biopsy, core needle biopsy, wedge resection or lobectomy. This study included only cytology cases derived from CT-guided PCNA and processed by the same method. Surgical specimens were all lobectomies.
Samples from CT-guided PCNA have fewer normal cells than do trans-bronchial processes, so they provide an easier way to recognise tumour cells and PD-L1expression despite CB preparation.
Although the tumour sections from lobectomies showed no even PD-L1 expression because of tumour heterogeneity, PD-L1 expression can be measured more accurately than from small biopsy specimens. Previous studies included comparative samples from different anatomical sites although they used paired cytology and histology specimens. Although various sampling methods have been used, substantial agreement was achieved between matched CBs and surgical specimens when CBs contain more than 100 tumour cells. 32 However, we used matched cytology and histology specimens from This study showed a high concordance of PD-L1 expression between cytology CBs and matched surgical resection specimens. We assume that the concordance results from the concept of this experiment, of using comparative samples obtained from the same anatomic site and by the same method. As shown in Table 2 We presumed that the cytology specimen happened to be obtained from the region with negative PD-L1 expression of the tumour with severe heterogeneity. Although there are a few exceptions of poor cyto-histology concordance, tumours with low rates of PD-L1 positive cells or severe heterogeneity, we provide early evidence that cytology CBs can be successfully used for PD-L1 immunohistochemical detection at least using the cut-off of 1%. Furthermore, the recent study revealed that cytology CB samples showed slightly higher sensitivity for PD-L1 immunohistochemical staining on tumour cells as compared to surgical specimens, the utility of CBs on PD-L1 IHC is expected to increase in the future. 34 The third potential issue is different PD-

| Clinicopathological parameters
This study indicated that PD-L1 expression has a positive correlation with smoking history in cytology and a positive tendency in surgical resection. It is consistent with previous findings. Smoking might skew host immune response to promote lung cancer, and consequently influence the tumour microenvironment and the prognosis in NSCLC patients. 35,36 PD-L1 expression was associated with smoking status, current smokers with more pack-years showed higher expression rates and intensity than did never smokers. 37 Rizvi et al have reported that the clinical efficacy of pembrolizumab correlates with a molecular signature characteristic of tobacco carcinogen-related mutagenesis in NSCLC patients. 38 There is no statistical significance between overall survival and PD-L1 expression in this study, probably because of the short follow-up period and the small number of enrolled patients who died. However, we confirmed that PD-L1 expression is strongly correlated with smoking status in this study, so it might be useful for selecting patients for PD-L1 immunotherapy in the clinic.  It was observed that PD-L1 expression was significantly related to age and TNM stage. In recent literature, the relationship between age and PD-L1 expression has been controversial. There are studies suggesting a positive correlation, 39,40 whereas others find no statistical correlation. 26,27 The relationship between pathological TNM stage and PD-L1 expression has been also controversial.
Some reports have the same conclusions as this study, 25,37,39,40 and another suggests no association. 26,41 The biology of an association between PD-L1 expression and these clinicopathological factors is not well understood. We assume that older age and advanced TNM stage may be associated with a higher tumour-mutation burden.

| Limitations
This study has a key limitation. We have no clinical outcome data on the responses to anti-PD-1 immunotherapy. We did not include patients prospectively treated with immunotherapeutic agents.
Another limitation is the limited number of paired cases. However, it was challenging to obtain matched cytology CBs and resection specimens with sufficient tumour cells remaining in both following routine diagnosis and work-up. We evaluated for only the concordance rate of PD-L1 expression between cytology CBs and matched surgical resection specimens with a limited number of cases in this study. We will make every endeavour to perform further clinical concordance studies in the larger number of patients treated with PD-1 axis therapies.

| CON CLUS ION
Herein, we found that cytology CB sections showed a strong positive correlation with resection specimens of the same anatomic site for PD-L1 assessment using the SP263 IHC assay. Taken together with our findings and a few studies of PD-L1 concordance between cytology and histology, 26-31 lung cancer patients with positive PD-L1 expression in the preoperative or diagnostic cytology do not need to repeat the test using the surgical resection specimens, at least using the cut-off of 1%. We propose that CBs are suitable materials for evaluating PD-L1 expression when simultaneously performing both diagnostic and molecular tests. There is no need to perform further biopsies for other molecular tests for management of advancedstage lung cancer. In clinical practice, cytology specimens may represent the only specimens available, especially from patients with advanced disease. CBs prepared from residual LBC material improve the diagnostic accuracy and tumour subclassification 15 and provide the results of a PD-L1 test and molecular tests rapidly.
We believe that CBs prepared from residual LBC material have an important role on the application of anti-PD-L1 therapy and expand the pool of PD-L1 examinations in the clinic. For therapeutic interventions, inhibition of PD-1/PD-L1 is expected to become a more powerful therapeutic alterative for NSCLC than ever before.

ACK N OWLED G M ENTS
This work was supported by the Dong-A University research fund.

CO N FLI C T O F I NTE R E S T
The authors have no conflicts of interest to report.

DATA AVA I L A B I L I T Y
The data that support the findings of this study are available from the corresponding author upon reasonable request. The data are not publicly available due to privacy or ethical restrictions.