Immunotherapy improves the survival in patients with advanced lung cancer. Data suggest that a high expression of PD-L1 on tumour cells is associated with a superior response to treatment with ICI [2–6]. Therefore, evaluation of PD-L1 expression on tumour cells may help to select patients who will benefit from immunotherapy providing an individualized therapeutic approach.
As most patients with lung cancer present with advanced inoperable disease at the time of diagnosis, PD-L1 testing is performed mainly on biopsy specimens. Therefore, our study examined whether the small amount of tissue analysed in biopsies my represent the PD-L1 expression of the tumour depending on biopsy technique and neoadjuvant treatment. Our results indicated a statistically significant but not necessarily clinically relevant correlation between PD-L1 expression on the tumour cells of diagnostic biopsy of the primary tumour with corresponding resected surgical specimens. This is particularly evident when looking at the comparison using a cut-off of ≥50% PD-L1 TPS that resulted only in a moderate agreement. Focusing on biopsy technique, a superior concordance for PD-L1 testing was observed for endobronchial biopsies, an inferior concordance for TBB.
Various studies comparing PD-L1 status of biopsy and surgical specimen resulted in controversial results. Kitazono and colleagues who compared PD-L1 expression between biopsy from the primary tumour and matched surgical specimens in 79 lung cancer patients, confirmed a significant association of the PD-L1 expression in tumour cells between biopsies and resected tumours [11]. A concordance rate of PD-L1 status was found to be 92.4% that is superior to our reported concordance rate of 78%. However, while Kitazono et al. used a hybrid score to describe PD-L1 expression on tumour cells and differed between PD-L1 positivity and PD-L1 negativity, we used a cut-off value of 50% PD-L1 TPS. It must be noted that the interpretation of PD-L1 staining is a challenge for pathologists. PD-L1 immunohistochemistry using a cut-off value is considered to be more challenging than a simple positive or negative result.
In another study which compared PD-L1 expression on tumour cells in different cytology, biopsy and surgical specimens of primary tumours or metastases (lymph node metastases, distant metastases, pleural effusions) in 23 patients independent of treatment in-between, no significant difference between the percentage of PD-L1 positive cytology and histology specimens was found suggesting a good concordance. However, this result was somehow surprising and contrary to findings by Xu et al. who described a poor correlation for PD-L1 expression between primary tumours and metastatic lymph nodes in 76 NSCLC patients [17]. In our study, only biopsy specimens from primary tumours were compared to corresponding resected tumours.
In contrast to our finding, other studies revealed a poor concordance for PD-L1 expression between biopsy and surgical specimen. Ilie and colleagues who used a scoring system from 0 to 3 depending on PD-L1 expression (≥50%, ≥5% and <50%, ≥1% and <5%, <1%) reported a poor correlation between the PD-L1 expression assessed in biopsy specimens and the corresponding resected tumour in 160 patients with NSCLC [9]. In this study, biopsy samples were derived from bronchoscopy including EBUS-TBNA or CT-guided transthoracic biopsy, but it remains unclear, whether the biopsy was sampled from the primary tumour or lymph node metastases. Also Li et al. found only a moderate correlation of PD-L1 expression between whole sections from NSCLCs and the corresponding tissue microarrays (TMAs) serving as surrogate biopsy specimens [10]. Of note, in this study, no real biopsy specimens but TMA were used for comparison.
Previously, different studies reported that a cisplatin-gemcitabine combination, paclitaxel-based regimen or TKI-based therapy led to a downregulation of the PD-L1 expression of tumour cells [15; 16]. In our study however, a statistical significant correlation of PD-L1 expression between biopsy and surgical specimen was not only found in treatment-naïve surgical patients, who did not undergo a treatment in-between but also in patients who received neoadjuvant therapy.
The limitation of this study is its retrospective character and the relatively small sample size. Particularly the number of patients who underwent neoadjuvant treatment between the biopsy and the surgical resection is low and does not allow a general statement about concordance of PD-L1 expression between biopsy and surgical samples in this subgroup. Another limitation is, that the histological workup of the biopsy specimen was performed in different pathology units that used different antibodies for staining (BioSite, Klon BSR90, SP263, QR1, Biocare CAL10 and 22C3 pharmDx). However, a good agreement rate in interpretation of the PD-L1 expression using various antibodies is described [18]. Moreover, similar results were obtained when regarding only at PD-L1 testing of biopsy and surgical specimens using the same antibody in one pathological department. Therefore, the use of various antibodies for staining in different pathology departments seem not to have substantial impact. It must be noted, that our study in comparison to the other study reflects the real-life situation. The demonstration of a statistically significant correlation in this study is even more significant because it is known that interpretation of PD-L1 expression does have its peculiarities with varying interobserver agreement.
Summarizing, this study found a statistically significant correlation for PD-L1 expression on tumour cells between biopsy and surgical specimen, but of uncertain clinical relevance. Particular the use of a cut-off value of ≥50% PD-L1 TPS resulted only in a moderate agreement. Therefore, the interpretation of biopsy based PD-L1 status should be considered with caution when deciding therapeutic approach for a patient.