Blocking immune checkpoint molecules with monoclonal antibodies has recently emerged as a promising strategy for treating some malignancies (1–5). Currently, The immune regulatory programmed death-1 (PD-1)/programmed death-ligand 1 (PD-L1) axis has been used as a checkpoint target for immunotherapy (6). Anti-PD-1/PD-L1 therapy has also shown promising antitumor activity in gastric cancer (GC) (4, 7, 8). Although there are no established biomarkers for anti-PD-1/PD-L1 antibodies, PD-L1 expression, Epstein–Barr virus (EBV) status, and DNA mismatch repair (MMR) status have been proposed as predictive biomarkers for anti-PD-1 response (9, 10). The CheckMate-032 trial showed a greater association of PD-L1 expression by combined positive score (CPS) with anti-PD1 therapy efficacy (11). However, the percentages of CPS ≥1, ≥5, and ≥10 were 32%, 10%, and 8%, respectively. Additionally, all CPSs for GC are decided by tests on tissue from gastric mucosa biopsy. Whether the CPS characteristics of resectable GC are different from the CheckMate-032 trial remains unknown. There are still no predictive biomarkers for the clinical efficacy of anti-PD-1 therapy. Currently, the main indication for anti-PD-1 therapy in GC is positive PD-L1 expression. However, since immune therapy have reshaped the paradigm of cancer therapy and progress rapidly in GC treatment in reccent year, the immunohistochemical (IHC) for immune therapy has not yet been updated or popularized in most hospital. In China, the area with the highest GC incidence, most hospitals cannot adequately assess PD-L1 expression for GC patients. This phenomenon inhibits the use of anti-PD-1 therapy in GC. In addition, associations between PD-L1 expression and other clinicopathological features, which urgently need to be better understood, remain unclear. Thus, we investigated PD-L1 expression in GC and discrepancy in CPS between gastric mucosa biopsy and postoperative pathology. Finally, we assessed the clinical associations between PD-L1 expression and other clinicopathological features that are more accessible in Chinese hospitals.

The immune regulatory PD-1/PD-L1 axis can induce inhibitory immune signaling within activated T cells, destroying their antitumor immune response, and thus is an immune checkpoint target for immunotherapy in many malignancies including GC (4, 7). The KEYNOTE-059 (4) and ATTRACTION-2 (19) trials confirmed the favorable efficacy and tolerability of anti-PD-1/PD-L1 therapies as third-line treatment for advanced GC. The CheckMate-649 trial (20) suggested the superiority of immune checkpoint inhibitors as first-line treatment. More encouragingly, we noted that durable response and long-term benefits could only be achieved by checkpoint inhibitors such as anti-PD-1 therapy rather than chemotherapy (21, 22). However, predictive biomarkers for the efficacy of anti-PD-1 therapy are lagging behind these clinical data. Currently, the primary indication for anti-PD-1 therapy in GC is the expression of PD-L1. However, since immune therapy progress rapidly in GC treatment in reccent year, most hospitals are not equipped to reliably test for PD-L1 expression in routine work. This phenomenon has limited the use of anti-PD-1 therapy in GC. Thus, we investigated whether PD-L1 expression in GC was associated with other clinicopathological features that are accessible in most hospitals.

PD-L1 protein expression is assessed in GC using CPS, which is the number of PD-L1 positively-stained cells (i.e., tumor cells, lymphocytes, and macrophages) divided by the total number of viable tumor cells, multiplied by 100%. A specimen is considered to have positive PD-L1 expression when CPS≥1. Consistently, Pembrolizumab is approved by the USA FDA for GC patients with positive PD-L1 expression (CPS≥1). The CheckMate-649 trial (20) presented survival benefits in patients with CPS≥5 following nivolumab treatment (HR: 0.71, 98.4%CI, 0.59–0.86, p<0.0001). The CheckMate-032 trial showed that PD-L1 expression by CPS demonstrated a stronger association with overall survival at higher cutoffs than PD-L1 expression on tumor cells (response rates of PD-L1 cutoff: <1 vs. ≥1 vs. ≥5 vs. ≥10: 0% vs. 28% vs. 41% vs. 55%) (11).While the KEYNOTE-062 trial which enrolled 763 patients with untreated, locally advanced/unresectable or metastatic GC with PD-L1 CPS ≥1, also analysed the efficiency of the subgroup of CPS≥10 (23). Meanwhile, many ongoing trials (e.g., NCT04139135 and NCT04744649) are currently exploring the effectiveness of neoadjuvant immunotherapy for GC using the cut-off of CPS≥10 to make sure there is a response to anti-PD-1 antibodies. However, all the CPS results regarding advanced GC and neoadjuvant treatment were tested on tissue from gastric mucosa biopsy. While for resectable GC, whether the CPS characteristics are different from the CheckMate-032 trial remains unknown. Our data answered this question and showed the significant CPS discrepancy between gastric mucosa biopsy and postoperative pathology. Our analysis also revealed that CPS≥1 was seen in 62.3% of patients (579/930), CPS≥5 was seen in 49.2% of patients (458/930), and CPS≥10 was seen in 33.3% of patients (310/930). While in CheckMate-032 trial, the percentages of CPS ≥1, ≥5, and ≥10 were 32%, 10%, and 8%, respectively. Another Chinese cohort analysis showed that 37.3% of cases (205/550) presented PD-L1 expression in tumor cells or tumor-infiltrating immune cells (24). Another Asian cohort study revealed that PD-L1 IHC scores were positive in 22.8% of patients (25). These results suggested that the spatial heterogeneity of PD-1 expression resulted in the assessments of gastric mucosa biopsy, such as those performed in previous studies, inaccurate in advanced GC. Accordingly, the potential beneficiaries of immune checkpoint inhibitors are likely broader than what has been reported. This phenomenon indicated that in clinical practice we should obtained tumor tissue from as many sites as possible when performing forceps biopsy to assess the PD-L1 expression.

Then, we searched for associations between PD-L1 expression and other clinicopathological features. The aim of this work was to identify new biomarkers associated with PD-L1 expression that are available in Chinese hospitals. Our analysis confirmed that CPS for PD-L1 was linearly correlated with age, Lauren classification, Ki-67 index, and EBV status. The Ki-67 index and Lauren classifications are both very accessible assays in most Chinese hospitals and might also be potential biomarkers for the efficacy of anti-PD-1 therapy. Thus, it is worth exploring their potential relationship with outcomes of anti-PD-1 therapy.

The multiplex immunofluorescence staining of non-small cell lung cancer (NSCLC) has revealed that Ki-67 index, along with cytokeratin, PD-L1, PD1, CD8, and CD68 are key components of the immune response to NSCLC (26). This finding was provided to assist others to apply similar methods to further understand the immune response to NSCLC. Zhao et al. (27) also showed that compared with those with negative PD-L1 expression, NSCLC patients with positive PD-L1 expression had significantly higher rates of lymph node metastasis (64.9% vs. 27.5%, p<0.01), more advanced tumor stage (p<0.01) and Ki-67 index (P<0.01), and thus concluded that positive PD-L1 expression was associated with more aggressive pathological features and poorer prognosis in advanced-stage NSCLC. Similarly, Pawelczyk et al. (28) also found that PD-L1 expression was associated with increased tumor proliferation and aggressiveness. In line with these NSCLC data, it was also found that PD-L1 expression was correlated with clinicopathologic parameters in breast cancer, including lymphovascular invasion and Ki-67 index (29). Accordantly, another study showed that PD-L1 expression was significantly associated with age and high Ki-67 index in breast cancer (30). Furthermore, some studies have even indicated that high Ki-67 index is a strong predictor of pathologic complete response in HER2+ breast cancer (31). Recently, it has been shown that positive Ki-67 and PD-L1 expression in post-neoadjuvant chemotherapy radical cystectomy samples was associated with inferior overall survival and the absence of tumor downstaging. IHC of Ki-67 and PD-L1 could help select patients for adjuvant therapy in post-neoadjuvant chemotherapy muscle-invasive bladder cancer (32). Wang et al. also demonstrated that high Ki-67 index was associated with a higher TNM stage and was an independent predictor of unfavorable prognosis in colorectal cancer (33). Thus, whether the Ki-67 index can improve the efficiency of predicting anti-PD-1 response in GC should be explored. Regarding Lauren classification, previous report has also demonstrated a significant association between PD-L1 status and Lauren classification (34). Histology pattern of Lauren classification included intestinal type, diffuse type and mixed type. The intestinal-type maintains the glandular appearance, which is concerned in an environmental factor; While the diffuse type shows diffusely infiltrating cells without glandular architecture which is used to be concerned in genetic factors. And the mixed type presents both intestinal type and diffuse type in the tumor specimen (35). Of coures, the Lauren classification is associated with biological behavior in GC (35); thus, finding ways to apply this information to identify tumor subsets and develop molecularly tailored, individualized immunotherapy benefits is goal for future studies.

EBV positivity has been proposed to be a predictive biomarker for anti-PD-1 response in GC patients (9, 10). Our analysis showed that EBV-positive GC is more prone to high PD-L1 expression. Consistently, previous studies have also shown that PD-L1 expression by tumor cells appears to be more common in EBV-positive GC (36). Another Asian cohort also demonstrated that PD-L1 expression was associated with distinct clinicopathological features, including dMMR and EBV positivity (25). Separate follow-up studies have also shown that EBV+ tumors exhibit robust PD-L1 expression both in cancer cells and immune cells (37). The mechanism might be that EBV-positive GCs are characterized by marked intra- or peri-tumoral immune cell infiltration and often exhibit genomic amplification of the chromosome 9 locus containing the genes encoding PD-L1 and PD-L2 (38). In 2018, Kim et al. provide insight into the molecular features associated with response to pembrolizumab in patients with metastatic GC and provided potentially relevant biomarkers for selecting patients who may derive greater benefit from PD-1 inhibition (10). Their results showed that dramatic responses to pembrolizumab were observed in patients with EBV-positive tumors (100% overall response rate in EBV-positive metastatic GC).

And we have noted that there is a synergetic effect on tumor reduction in some pre-clinical research when combining anti-HER2 and anti-PD-1 therapies (39). Moreover, the clinical benefit for the combination of anti-HER2 anti-PD-1 therapies, and chemotherapy for patients with HER2-positive GC have been revealed in clinical trial (40). On the basis of it, the phase III KEYNOTE-811 trial was conducted to investigate the efficacy whether pembrolizumab or placebo in coadministration with trastuzumab and the investigator`s choice of chemotherapy in participants with previously untreated unresectable or metastatic, HER2-positive gastric or gastro-oesophageal junction adenocarcinoma. And encouragingly, the results of KEYNOTE-811 trial suggested that adding pembrolizumab to trastuzumab and chemotherapy markedly enhanced the treatment efficacy (41).

These primary data indicate that HER-2 status is also the potential biomarker for anti-PD-1 therapy and encouraged us to investigate whether HER-2 status was correlated with PDL1 expression. Furthermore, another Chinese cohort suggested that PD-L1 expression was more common in HER2-negative tumors compared with HER-2-positive tumors (39.0% vs. 24.2%, P=0.020) (24). However, our data showed that PD-L1 expression was not associated with the HER-2 status in GC. Thus, the previous clinical synergetic effect of anti-HER-2 and anti-PD-1 may be attributed to other mechanisms. For example, preclinical study indicated that trastuzumab could upregulate expression of PD-L1 through engagement of immune effector cells may function as a potential mechanism (42). Also, it has been uncovered that combination of anti-HER-2 and anti-PD-1 therapies could improve HER-2-specific T cell responses, boost immune cell trafficking and weaken gathering of peripheral memory T cells (39, 43).

It is also noteworthy that our data showed that PD-L1 expression was not related to tumor stage. In contrast, many previous studies have indicated that PD-L1 expression is significantly associated with tumor stage. As early as 2014, Muenst et al. showed that PD-L1 expression was significantly associated with tumor size, AJCC primary tumor classification, tumor grade, and lymph node status in breast cancer (30). Similarly, Uhercik et al. (44) demonstrated significant transcript level reductions in PD-L1 in patients who developed metastases, as well as in those who had local recurrence compared with patients who remained disease-free. Consistently, PD-L1 expression in NSCLC was also increased in higher malignancy grades (p<0.001) and in higher lymph node status (p=0.043) (28). Furthermore, a systematic review and meta-analysis revealed that GC patients with deeper tumor infiltration, positive lymph node metastasis, and positive venous invasion were more likely to express PD-L1 (45). In contrast, our data showed that PD-L1 expression was not associated with the TNM stage in GC. Therefore, we propose that immunotherapy could also be explored for less advanced GC, rather than only for stage IV. However, until now, immunotherapy only been demonstrated in late-stage GC (4, 7). Whether locally advanced GC has enough PD-L1 expression to work with anti-PD-1/PD-L1 therapy remains unknown. Thus, our primary evidence provided perspective on this clinical question. Additionally, neoadjuvant chemotherapy or perioperative chemotherapy have been proposed to improve outcomes, which can increase the R0 resection rate by shrinking tumor size. Moreover, potential prognosis-related factors like micrometastases can be better addressed by chemotherapy prior to surgery (46, 47). Theoretically, neoadjuvant immunotherapy or perioperative immunotherapy for locally advanced GC could benefit patients. Therefore, the situation in which locally advanced GC has abundant PD-L1 expression similar to late-stage GC is supported by our data and provides a solid theoretical foundation for neoadjuvant immunotherapy or perioperative immunotherapy. Therefore, on the basis of our results showing locally advanced GC has a similar frequency of PD-L1 expression as late-stage GC, we designed a clinical trial to investigate the safety and effectivity of neoadjuvant immunotherapy for locally advanced GC (NCT04744649).

PD-L1 expression showed strong intratumoral heterogeneity in GC; thus, in clinical practice, multiple biopsies should be recommended for accurate reflection of PD-L1 expression in GC. Age, Ki67 index, and Lauren classification, which is popular and accessible in most hospitals, should be further explored as potential biomarkers for anti-PD-1 therapy.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Ethics Committee of Nanfang Hospital, Southern Medical University. The patients/participants provided their written informed consent to participate in this study.

LZ, and JY made substantial contributions to the conception and design, and interpretation of data. XC, HZ, and MW contributed in drafting the manuscript or critically revising it for important intellectual content. HL, YH, TL, HC, MZ, and TC collected and analyzed the data. Each author participated sufficiently in the work to take public responsibility for appropriate portions of the content and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All authors contributed to the article and approved the submitted version.

This work was supported by grants from the Guangdong Provincial Key Laboratory of Precision Medicine for Gastrointestinal Cancer (2020B121201004), and National Natural Science Foundation of China (81902444) Science, Technology Program of Guangzhou (201903010072) and Climbing Program, Special Fund of Guangdong Province(No.pdjh2022a0093).

The authors declare that the paper was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.