The prognostic significance of programmed cell death protein 1 and its ligand on lymphoma cells and tumor-immune cells in diffuse large B-cell lymphoma, not otherwise specified

Diffuse large B-cell lymphoma, not otherwise specified (DLBCL, NOS) is the most prevalent type of non-Hodgkin’s lymphoma (NHL) arising from a complex interplay of genetic and molecular factors. This heterogeneity results in approximately 35% of DLBCL, NOS cases not responding to standard treatment method that combines rituximab with anthracycline-based chemotherapy. As a result, patients who do not respond to standard therapy experience relapsed or refractory disease, which remains the leading cause of mortality.^1,2 In recent years, a variety of novel therapies, including immunotherapies, have emerged that may provide effective treatment strategies for DLBCL, NOS patients, particularly for relapsed or refractory disease. Identifying additional biomarkers and carefully assessing them to precisely define the effectiveness of immunotherapy are therefore critical for improved treatment outcomes, which are being investigated in several ongoing studies.^3,4,5

The programmed cell death protein 1 (PD-1)/programmed cell death ligand 1 (PD-L1) immune checkpoint pathway, which is crucial for maintaining self-tolerance and excessive immune responses, has emerged as a novel biomarker target for various malignant neoplasms to suppress the anti-tumor immune response and evade immune surveillance.⁶ Immunotherapy based on the PD-1/PD-L1 signaling pathway has already been included in standard treatment guidelines for various carcinomas such as melanoma, non-small cell lung cancer, urothelial carcinoma, triple negative breast cancer and many others.⁷ Moreover, it has also attracted wide attention for the treatment of lymphomas, with the greatest success achieved in classical Hodgkin’s lymphoma and T-cell lymphoma.^{8,9,10,11,12,13} However, in DLBCL, NOS, immunotherapy based on the PD-1/PD-L1 pathway has not yet become part of the standard treatment approach. Expression of PD-1 and PD-L1 in DLBCL, NOS is often elevated, leading to impaired immune cell function and tumor growth. The expression of PD-1 has been reported mainly on immune cells, emphasizing tumor-immune cells (TICs), while PD-L1 expression on lymphoma cells (LCs) and TICs. Some studies have already shown that increased PD-1 expression on TICs is associated with favorable overall survival (OS) of DLBCL, NOS patients.¹⁴ In contrast, increased PD-L1 expression on LCs has been associated with poorer prognosis and increased resistance to chemotherapy⁸, which also correlates with the DLBCL, NOS non-germinal center B-cell like (non-GCB) DLBCL, NOS subtype defined by the Hans algorithm.¹⁵ These data confirm their pivotal role in the tumor microenvironment of DLBCL, NOS and also their contribution to the poor clinical outcomes.^8,14 However, to date, reported data have yielded conflicting results, particularly in relation to PD-1 expression on TICs, which has been by some authors associated with better progression-free survival (PFS) and OS, while some other studies have found no association with survival.^{1,6,16,17,18,19,20} On the other hand, some findings revealed an association of PD-L1 expression on LCs with worse OS^{1,6,8,14,16,18,19,20,21}, although there are data disputing the prognostic significance of PD-L1 or even showing a correlation with better rather than worse outcomes. It is also worth noting that most studies on the expression of PD-1 and PD-L1 in DLBCL, NOS have focused on the Asian population, where there is a higher prevalence of non-GCB DLBCL, NOS subtypes²², which has led to a lack of comprehensive research in the European population. Furthermore, almost all published studies have generally included only a very small number of DLBCL, NOS patients in their analyses.^{1,18,19,23,24,25}

Moreover, the majority of published research on DLBCL, NOS has merely focused on analyzing PD-1 and PD-L1 expression on LCs or TICs, and when this has been the case, the analyses had been based on only a single immunohistochemical (IHC) staining assessment. To our knowledge, the use of markers to accurately identify LCs, such as PAX5, has rarely been reported.¹⁴ PAX5, a member of the paired box gene family of transcription factors, is a B cell-specific activator protein that plays an important role during B lymphopoiesis. It shows consistent expression across various stages of B-cell maturation and can be identified in the majority of B-cell neoplasms, even in cases where mature B-cell markers are not expressed.²⁶ Furthermore, no studies have investigated the simultaneous expression of PD-1 and PD-L1 on both LCs and TICs and their prognostic significance.

Therefore, our aim was to evaluate the simultaneous expression of PD-1 and PD-L1 on LCs and TICs in a Slovenian cohort of DLBCL, NOS patients using double IHC staining in combination with PAX5 and to investigate their association with prognosis.

In the present study, we investigated the expression of PD-1 and PD-L1 on LCs and TICs in the tumor microenvironment of DLCBL, NOS patients in relation to non-GCB and GCB subtypes and patients’ survival.

The expression of PD-1 and PD-L1 has drawn great attention to the impact of lymphoma treatment, particularly in aggressive lymphomas such as DLBCL, NOS, where they evade immune response and drive aggressiveness.²¹ Identification of patients for PD-1/PD-L1 immunotherapy, possibly through IHC evaluation, holds promise for better patient outcomes and further research in checkpoint inhibitor treatment. Currently, there are few ongoing clinical trials investigating the use of anti-PD-1 and anti-PD-L1 treatments in patients with relapsed or refractory DLBCL, NOS. Preliminary results in small patient cohorts show promising results for prolonged disease-free intervals.²⁹ However, the selection criteria for patient enrollment are not dependent on PD-1 or PD-L1 expression on either TICs or LCs. One of the explanations for this could be the lack of consensus in the evaluation criteria, especially for PD-1 expression, as it is less frequently assessed and less defined. Different evaluation criteria are used to assess cell positivity, and we have used the most commonly used ones.^{8,14,16,17,25,29,32} We found only a few studies investigating whether there is a correlation between the expression of PD-1 or PD-L1 on TICs or LCs and patient outcomes. For the assessment of PD-1 and PD-L1 on LCs and TICs in our study, we performed double staining for PD-1/PAX5 and PD-L1/PAX5 to simplify the assessment of PD-1 and PD-L1 expression and obtain more reliable results. Indeed, PAX5 was required for accurate identification of B cells, including LCs, because PAX5 is expressed in mature B cells and LCs.^14,24,25,33 The use of PAX5 increased the accuracy of LC identification, which in combination with the simultaneous staining of PD-1 and PD-L1 is one of the major advantages of this study. This double staining was so far reported in Kiyasu’s study of 1091 patients with DLBCL¹⁴, NOS and Chen’s study of various lymphoma subtypes, including 66 patients with DLBCL, NOS³³, but both studies used only PD-L1/PAX5 staining. To the best of our knowledge, our study is the first to perform PD-L1 and PD-1 staining simultaneously with PAX5.

In this way, we confirmed a PD-1 expression of 38.4% on TICs and 8.8% on LCs. Our results were similar to the already published data, where PD-1 expression on TICs ranged from 22.2–60.0%. For PD-1 on LCs, we observed a slightly lower expression rate compared to the expression levels of PD-1 reported in the literature (22.2–65.0%).^{16,24,25,29,34} Using the same assessment method as in the other published studies^{14,16,17,25,29,30}, we reported 62.4% PD-L1 expression on TICs and 6.5% on LCs. Our data again differed from already reported PD-L1 expression levels, where expression ranged from 15.3–37.0% on TICs and 8.9–61.1% on LCs.^14,17,24,29 In summary, we observed a lower expression of PD-L1 and PD-1 on LCs and a higher expression of PD-L1 on TICs. We speculate that the major reason for the low expression on LCs is the additional staining with PAX5. With the double staining, we were able to recognize PD-1/PD-L1 positive LCs more precisely, so the numbers are probably more reliable than in other studies where other cells such as macrophages could be misinterpreted as LCs and resulted in higher expression of PD-L1 and PD-1 on LCs. Two already published studies using PAX5 staining reported 8.9%¹⁴ and 11% PD-L1 positivity on LCs³³, which is comparable to our results.

We also speculate that a possible reason for this discrepancy in PD-L1 and PD-1 expression rates could be a consequence of several other factors. One reason might be the different inclusion criteria of the patients. Almost all published studies included patients with Epstein-Barr virus (EBV) positive DLBCL, NOS, which typically exhibit PD-L1 positive status due to the association between PD-L1 expression and EBV infection. EBV positive patients were not a part of our patient cohort, so this could be one of the reasons for the discrepant results. In addition, different algorithms were used to define the DLBCL, NOS subtypes. The majority of the studies have used Hans’ algorithm^{14,16,17,25,29,30}, but some have applied Choi’s algorithm.³⁵ The choice of one of these algorithms may contribute to different percentages of GCB and non-GCB subtypes, as well as different percentages of PD-L1 positive LCs and TICs within each subtype.³⁵ Furthermore, different antibody clones were used in different studies, and in some cases the clones used were not clearly stated.^8,25 Additionally, some studies lacked a clear description of the criteria used for the assessment of PD-L1 positivity.²⁴ Since the major focus was on the Asian population, where non-GCB subtypes of DLBCL, NOS, were more common than GCB subtypes, a higher number of PD-L1 positive cases was expected.²²

The second aim of our study was to investigate the association of PD-1 and PD-L1 expression on TICs or LCs with clinicopathological characteristics. We investigated the potential differences between various clinicopathological characteristics of patients such as age at diagnosis, gender, Ann Arbor stage, extranodal organ involvement, involvement of the spleen, presence of B symptoms, and IPI score, but our results showed no difference in PD-1 and/or PD-L1 expression on LCs and/or TICs between these groups. Additionally, we examined the histological subtypes determined by the Hans algorithm and found that PD-L1 positive expression was more frequent in the non-GCB subtype, as already reported.^14,15,17,19 However, in a univariate analysis, the Hans algorithm showed no significant impact on PFS or OS. Regarding patient survival, we found that patients with PD-L1 expression on the LCs are correlated with shorter PFS, which also had a significant impact in the multivariate analysis including the IPI score. This indicates that although the number of patients with PD-L1 positive expression on the LCs is small, their survival is worse in comparison with the patients with no PD-L1 expression, regardless of the IPI score. Since the IPI score is still one of the strongest and most reliable prognostic markers in DLBCL, NOS, even in the era of novel therapies, we believe this is an interesting conclusion that requires further research.

In the multivariate analysis for PFS, where the Ann Arbor stage, presence of B symptoms, IPI score, and PD-L1 expression on LCs for PFS were included as significant parameters from the univariate analyses, PD-L1 expression on LCs was found to be an independent prognostic marker for PFS. As expected, the IPI score also proved to be an independent prognostic marker for PFS. In the multivariate analysis of OS, where the age, Ann Arbor stage, B symptoms and IPI score were included, we showed that age remained as an individual marker for OS. This result was consistent with already published data based on 5-year follow-up analyses for PFS and OS of DLBCL, NOS patients.⁸

Noteworthy, there are some limitations of our study that need to be considered. For example, due to the retrospective nature of the study, the longer archiving time of FFPE tissue blocks may potentially influence the staining results. In addition, it is still unclear whether the expression of PD-1 and PD-L1 on tumors and TICs is a key factor for the clinical prognosis of DLBCL, NOS patients treated with PD-1/PD-L1 blockade therapy. On the other hand, our study has certain advantages, such as centralized evaluation of specimens, homogeneous treatment and long follow-up time. To the best of our knowledge, we are the first to simultaneously assess the expression of PD-1 and PD-L1 on both TICs and LCs from the same cohort of patients using a double immunostaining approach and their impact on PFS and OS, as well as their association with other clinicopathological characteristics. Our results were consistent with individual studies on the European population as well as studies on the Asian population.

In conclusion, we demonstrated that PD-L1 expression on LCs was associated with shorter PFS and was more frequently observed in the non-GCB subtype. Double IHC staining with PAX5 proved to be a feasible method to assess PD-1 and PD-L1 expression in tissue samples. Further research and clinical studies are required to assess the importance of assessing PD-1 and PD-L1 in DLBCL, NOS patients as well as methods to determine their expression, particularly with regard to planning immunotherapy treatments. It is also important to understand the mechanisms of tumor immune evasion induced by PD-1/PD-L1 and to explore approaches to modulate the host immune response accordingly, which requires further research.