PD-L1 expression, tumor-infiltrating lymphocytes, and mismatch repair proteins status in digestive neuroendocrine neoplasms: exploring their potential role as theragnostic and prognostic biomarkers

Gastroenteropancreatic (GEP) neuroendocrine neoplasms (NENs) encompass a heterogeneous group of malignant epithelial neoplastic proliferations including relatively indolent neuroendocrine tumors (NETs) and highly aggressive neuroendocrine carcinomas (NECs) [1]. NETs and NECs are two different diseases, with distinct morphological, molecular, and clinical features [2]. Based on their distinct biological features, they are treated differently in the context of clinically advanced disease. Generally speaking, platinum-based chemotherapy schedules are employed in patients with advanced NECs, whilst NETs can be managed with integrated approaches including surgery, somatostatin receptors targeted therapy, peptide receptor radionuclide therapy (PRRT), mTOR inhibitors (i.e., everolimus), antiangiogenic drugs (i.e., sunitinib), or chemotherapy depending on tumor type, grade, and stage [3‐11]. However, despite the availability of multiple treatment options and guidelines from several scientific societies, there are still areas of controversy in advanced NENs treatment, for which there is limited guidance [12]. In this complex scenario, the role of immune checkpoint inhibitors (ICIs) therapy in NENs management has been preliminarily explored with conflicting and partially non-conclusive results and it still remains to be elucidated [13]. It is worth noting that the proper identification of patients to treat with ICIs reduces the cost/benefit balance, also in terms of adverse effects of these drugs, which may cause severe and potentially lethal conditions [14, 15]. Consequently, the evaluation of predictive biomarkers is needed to predict clinical benefit. High tumor mutational burden, the presence of mismatch repair protein (MMR) deficiency or of microsatellite instability, the estimate of tumor infiltrating lymphocytes (TILs), and the immunohistochemical analysis of PD-L1 expression on tumor cells and immune cells of the microenvironment are used as biomarkers for selecting patients receiving ICIs therapy [16].

Although some data have been published regarding lung NENs, studies exploring the clinical efficacy of PD-L1 inhibitors in extrapulmonary NENs have given conflicting results and the use of a combined or, mainly, single-agent does not seem to be particularly active for unselected patients [13]. In terms of biomarkers, it is worth noting that, although some studies suggest that PD-L1 is more frequently expressed by tumor cells in NECs than in NETs [17, 18], there are only few investigations exploring PD-L1 expression in association with the evaluation of TILs and microsatellite instability or mismatch repair (MMR) deficiency status, which are all associated with potential effectiveness of ICIs in re-activating anti-cancer immune response [19‐22].

In this study we performed an integrated evaluation of PD-L1 and mismatch repair protein (MMR) immunohistochemical expression, together with TILs characterization in a series of well characterized digestive NENs with the aim of better defining the potential role of ICIs therapy in these diseases.

ICIs-based therapeutic approach is now accepted for several different cancers including melanoma, head and neck squamous cell carcinoma, lung cancers including small cell carcinoma, renal cell carcinoma, urothelial carcinomas, and others [14]. The clinical efficacy of ICIs therapy reflects the immunomodulatory effect mediated by the binding with their specific antigens involved in the regulation of immune response, consequently increasing host response to tumor cells [16]. It clearly appears that ICIs performance depends on the correct selection of patients who can benefit from the therapy, excluding those potentially not responding for whom the cost/benefit of ICIs therapy, which is not free of adverse symptoms (including, but not limited to, hypophysitis, thyroiditis, and colitis), is not favorable [15].

The usefulness of ICIs for the therapy of digestive NENs remains to be established, as published results are conflicting [13, 23]. The objective response rates in patients treated with pembrolizumab, spartalizumab, avelumab or combined nivolumab/ipilimumab, durvalumab/tremelimumab, or atezolizumab/bevacizumab are variable from 0% to 3.7% in unselected patients. It has been observed that only 10–30% of digestive NENs express PD-L1 [17‐19, 24] and the need for a proper selection of patients undergoing ICIs treatment may be advocated on these bases. Indeed, besides PD-L1 expression, also MMRd and intense intratumor lymphoid infiltrate have been used as predictors of good response to ICIs [25‐32].

In our series, we have found PD-L1 expression in 28 out of 68 NENs, representing 41.2% of cases. Considering studies published in the literature, PD-L1 expression has been reported in 34.2% GEP-NENs, with a range from 6.1% to 75% (Table 2). Although the specific comparison of our findings with those of other studies is not easy due to the different antibodies used, different methods of evaluation employed, and different NEN classifications considered, the percentage of positive cases we observed is generally comparable with the literature data. For the first time in GEP-NENs, we evaluated and compared both TPS and CPS scores, which represent two different systems for reporting PD-L1 expression in tumors. TPS score refers to PD-L1 expression only in tumor cells and is mainly used in the lung to select patients eligible for ICIs therapy [33‐35]. The CPS score includes the evaluation of PD-L1 expression in both neoplastic and inflammatory cells and is generally used in cancers arising in the head and neck region or in urothelial mucosa [36‐38]. A comparison between TPS and CPS scores has never previously been reported in digestive NENs and the use of ICI has only been used in advanced with limited response and poor connection with PD-L1 expression in immunohistochemistry [31, 39‐44]. In our series, we have found more tumors with CPS > 1 than with TPS > 1% and it is conceivable that CPS may be proposed as the standard score for determining PD-L1 positivity in NENs for treatment with ICIs, although specific clinical trials are needed to determine which of the two scoring systems is truly more effective. As expected, the number of PD-L1 positive cases (considering both TPS and CPS) was higher among NECs than NETs (p = 0.003). Additionally, we observed a direct correlation between PD-L1 positive cases and proliferation grade of in NETs, in line with previously published studies [45‐48].

For the analysis of tumor infiltrating lymphocytes (TILs), we counted both intra-epithelial and intra-stromal CD3⁺ T lymphocytes. In addition, the number of CD4⁺ and CD8⁺ T-cells was evaluated. The CD3⁺ TILs counts were significantly higher in NECs than in NETs (p = 0.00001 for intra-stromal TILs and p = 0.001279 for intra-epithelial TILs). CD4⁺/CD8⁺ ratio was not significantly different between NECs and NETs; moreover, we did not find a definite predominance of CD4⁺ or CD8⁺ cells in TILs. The more abundant lymphocytic infiltration observed in NECs than in NETs is in line with previous findings that also demonstrated a direct correlation with PD-L1 expression, not observed in our study [45, 54, 59]. Because of the immunological implications, high TIL count has been considered a prognostic marker to select patients responding to checkpoint inhibitors therapy, although a consensus has not been still reached, especially for digestive NENs [19, 56]. However, the results of our study demonstrating that NECs generally show a more abundant lymphocytic infiltration than NETs suggest that immunological therapies may be considered only for this aggressive NEN subtype.

Intratumor inflammatory infiltration has more frequently been observed in MMRd and/or MSI NENs [60, 61] and for this reason we evaluated the expression of mismatch repair proteins in our NENs using immunohistochemistry. Seven out of 68 cases showed MMRd and were predominantly represented by NECs (p < 0.05). In our series, we did not find any correlation between MMRd and other clinico-pathological parameters when considering both NETs and NECs. The observation that MMRd is more frequently observed in NECs, which also show a higher intratumor lymphocytic infiltration, and that is associated with better prognosis is in line with previous data [21] supports the hypothesis that immunological therapies should be employed in a subset of NECs.

Finally, we correlated PD-L1 expression with other clinico-pathological parameters (age, gender, histological subtype, etc.…) and we did not find any relevant correlation. Survival was not related to PD-L1 expression, which was one of our starting hypotheses. A correlation between PD-L1 expression and survival has been reported in only one study [18], while it was not found in others [48, 62, 63]. Although PD-L1 expression alone cannot be considered as a prognostic marker, its prognostic role may change in future if it can be used to select patients potentially responding to ICIs therapy. Consequently, its predictive role may influence the prognosis of patients with NECs. This may open a new therapeutic strategy, considering that digestive NECs are currently treated with platinum-based chemotherapy, which has not significantly increased patient’s survival in the last years [64].

We are well aware that, like all retrospective studies, our investigation shows potential limitations. Specifically, staging system procedures (i.e., radiology and/or nuclear medicine investigations, and laboratory tests) and therapeutic protocols were not the same for all patients since the diagnoses were performed in a range of 10 years, a long period in which new clinical and therapeutic approaches have been developed. In addition, not all patients were treated in the same hospital or by the same team. Nevertheless, to reduce biases, all tumors were reclassified according to the 2019 WHO classification of digestive neuroendocrine neoplasms.

In conclusion, this study, which should be considered as a preliminary investigation, has demonstrated that PD-L1 expression is not correlated with clinical and pathological data. Thus, it cannot be considered a significant biomarker on its own. However, the heterogeneous expression of PD-L1 in NENs, with a clear predominance in NECs and in high grade NETs, may suggest selecting specific patients for whom targeted therapy with checkpoint inhibitors may have a beneficial effect. In our hands, the CPS score proved to be more sensitive than TPS in identifying PD-L1 positive cases and a value of CPS > 1 may be suggested, at this stage, as potentially predictive of a good response to ICIs. Furthermore, the T-cell infiltrate and MMRd, more frequently observed in NECs than in NETs, further support the higher efficacy of immunological therapy in NECs.

This study, although with the limitations above discussed, compares for the first time PD-L1 expression (using both CPS and TPS scores), tumor infiltrating lymphocytes, and MMR status in NENs of the digestive tract, including NETs and NECs. Our results provide a conceptual rationale for restricting the employment of immune therapy to NECs, in which this kind of therapeutical approach is likely to be more effective.