The prognosis of lung cancer is much improved during the last decade, but despite progress, it is still the cancer which takes most lives yearly (1). With the advent of immunotherapy, patients with advanced or metastatic non-small cell lung cancer (NSCLC) may now achieve durable responses and extended overall survival when treated with an immune checkpoint inhibitor (ICI) in monotherapy or in combination with chemotherapy in first line (2–5). However, not all patients benefit from immunotherapy. Oncogenic driven tumors with mutations in the gene encoding the epidermal growth factor receptor (EGFR) seem particularly irresponsive to checkpoint inhibitors (6, 7). Thus, first line treatment for advanced or metastatic disease is targeted treatment with EGFR tyrosine kinase inhibitors (EGFR-TKIs). Modern EGFR-TKIs are superior to older generation drugs in terms of progression free survival (PFS) and overall survival (OS) (8, 9), but resistance develops in virtually all patients. Osimertinib, a third generation TKI is currently the first choice in many countries globally. However, it was originally developed as a second line drug to target the EGFR T790M resistance mutation. Studies of osimertinib in the second line setting have demonstrated a PFS of around 10 months when T790M is detected at re-biopsy at progression on earlier generation EGFR-TKIs (10). Furthermore, there are indications that osimertinib to some extent exerts activity also in T790M-negative cases resistant to previous EGFR-TKIs (11).

Osimertinib is also highly active in the first line setting in the presence of sensitizing mutations (deletions in exon 19, del19, or the point mutation L858R in exon 21), yielding prolonged PFS of around 18 months when given in monotherapy and 25.5 months as combinatory treatment with chemotherapy (8, 12). However, given the poor responsiveness to immunotherapy, treatment options for EGFR-mutated NSCLC are limited when the disease becomes resistant to the available targeted therapies. Extensive efforts have been made to map resistance mechanisms to osimertinib with the aim of identifying new targets for treatment (13–17).

The ectoenzymes CD73 and CD39 are regulators of the adenosine pathway which is involved in inducing an immune-evasive tumor microenvironment (TME). Extracellular CD39 converts ATP to ADP and AMP and CD73 converts AMP to the immune suppressant adenosine which in turn binds to the A2A adenosine receptor (18). EGFR-mutated NSCLC express higher levels of CD73 compared to EGFR WT tumors (19) and upregulation of CD73 has been proposed as a resistance mechanism to EGFR-TKIs (20). In a phase I study, oleclumab, an antibody targeting CD73, displayed some clinical activity when given in combination with osimertinib to patients with T790M-negative disease after progression on first- or second-generation EGFR-TKIs (21), however the role of CD73-inhibition is still unclear.

In this study, we explored the expression of CD73, CD39 and related immune markers in tissue samples from different time points in the treatment course of patients with EGFR-mutation positive NSCLC during several lines of EGFR-directed therapies, with the aim of identifying potential biomarkers for future CD73-based treatment options.

In this study we investigated the expression of CD73 and related immune markers in tumor tissue at different time points during the disease course of patients with EGFR-mutated NSCLC, including across different lines of treatment with EGFR-TKIs. All patients received osimertinib after progression of first- or second line EGFR-TKIs, regardless of resistance mechanism (11). In the specimens from the patients included in this biomarker study, we found a differential expression pattern of the various markers, with high expression in tumor epithelium of CD73 and HLA-E, whereas CD39 and NKp46 were higher expressed in the stroma. This expression pattern was consistent at different time points in the sequence of treatment. As only four patients had paired samples pre- and post-osimertinib, conclusions regarding the temporal expression pattern during osimertinib treatment cannot be drawn. However we observed that two out of the four had an increase in CD73 expression and three of four had an increase in both NKp46 and HLA-E upon osimertinib exposure.

The adenosine pathway regulating ectoenzyme CD73 is expressed in many cancer types including lung cancer and is known to induce an immune-evasive TME, but also contribute to tumor progression and metastasis (24). Within the different types of lung cancers, the expression of CD73 is higher in TTF1 positive adenocarcinomas and especially EGFR-mutated lung adenocarcinomas (25). Interestingly, we found that the expression of CD73 was significantly higher in EGFR-mutation positive tumor specimens with non-T790M resistance to first- and second-generation EGFR-TKIs than in specimens with the T790M-mutation. To our knowledge, this has previously not been shown. The high expression in T790M-negative cases might suggest increased CD73 as a resistance mechanism to EGFR-directed drugs, possibly mediated through several inhibitory effects to the immune response, or more directly via interference with the EGFR signaling pathway. CD73-induced excess adenosine production leads to less effective antitumor activity by affecting a number of different immune cells. Adenosine signaling via the A2A receptor inhibits activation of T cells and stimulates the expression of other co-inhibitory receptors like PD-1 (26, 27). Likewise, dendritic cells are modulated by adenosine exposure to engage in pro-tumor activity (28, 29). Furthermore, there is evidence suggesting an interplay between adenosine levels and the activity of natural killer (NK) cells. Some studies demonstrate dysregulation of NK cells by increased levels of adenosine and even induction of CD73 on NK cells by tumor cells leading to NK cell dysfunction which affects their proliferation, cytotoxic effect and cell trafficking (30–33). Previous studies have shown that CD73 and CD39 are involved in resistance to chemotherapy by suppressing the adaptive immune response (34, 35), further supporting the hypothesis that CD73 expression might lead to resistance to other types of cancer therapies. In addition, some studies indicate a more direct effect of CD73 on EGFR signaling by cross-activation of EGFR through which resistance to EGFR-TKIs could evolve (36, 37) and also through altered phosphorylation of the EGFR by CD73 (20). Moreover, a preclinical study demonstrated that MET-amplification upregulate CD73 in EGFR-TKI resistant cells, creating a tumor-promoting TME and thus providing another possible explanation for the link between CD73 expression and therapy resistance (38).

As CD73 expression has been shown to be associated with poor prognosis in NSCLC (25, 39), we speculate that the high expression in T790M-negative cases in our study might contribute to the modest PFS to osimertinib for these patients in second or later lines (11). Additionally, we also found a trend towards shorter overall survival from the time of diagnosis and irrespective of number of treatment lines for patients with high expression of CD73 in their diagnostic biopsy, supporting the notion that CD73 is a prognostic factor.

Intriguingly, in our study, patients with non-T790M-resistance to EGFR-TKIs and high expression of NKp46 before commencing osimertinib as next-line treatment had a shorter PFS than patients with low expression of NKp46. The reason for the observation of less clinical benefit of osimertinib despite the abundant presence of NK cells is not known, however, a possible explanation is the adenosine-induced dysregulation of NK cells, as discussed above (33). However, patients with high expression of CD73/CD39 at start of osimertinib treatment showed at trend towards a larger clinical benefit of the drug in terms of PFS.

Preclinical studies have identified the CD73/adenosine pathway as a potential therapeutic target to sensitize EGFR-positive tumors to immunotherapy (40). In a phase I study investigating the combination of the CD73 antibody oleclumab and durvalumab, 9.5% of 42 patients with EGFR -mutated NSCLC achieved a partial response and 21% had stable disease (41). There have also been different attempts of targeting the CD73/adenosine pathway without including immune checkpoint inhibitors. Oleclumab demonstrated promising activity in combination with osimertinib in patients with T790M-negative disease after first- or second-generation TKIs (21). Among other potentially relevant agents are the MEK1/2 inhibitors selumetinib and PD318088, as for these a correlation of drug sensitivity and the protein levels of CD73 has been shown (39). Nevertheless, there is still a large unmet need to develop effective treatment options for patients with EGFR-mutated NSCLC resistant to established treatments and thus also a need for identifying biomarkers to guide the development of new drugs and patient selection.

The treatment landscape of EGFR-mutated NSCLC has been rapidly moving forward, and this study was designed when first- or second-generation EGFR-TKIs were standard of care as first line treatment. However, increasing knowledge of clinical activity and resistance mechanisms to osimertinib given in later lines might still be of relevance as many patients, especially in some territories, still receive sequenced EGFR-directed therapies. Furthermore, as upregulation of CD73 is known to contribute to modulation of the TME and immune evasion in multiple cancers and in different treatment settings, there is a rationale to investigate the role of CD73 within the present treatment algorithm for EGFR-mutated NSCLC at resistance to osimertinib given as first line treatment. The FLAURA-2-study demonstrated superior efficacy of osimertinib in combination with chemotherapy in the first line setting (12), and some retrospective studies also suggest that adding chemotherapy to osimertinib at the time of progression on osimertinib monotherapy might be of clinical benefit for patients (42). Furthermore, the MARIPOSA-trial recently demonstrated an overall survival benefit in favor of combination therapy with the bispecific MET/EGFR-antibody amivantamab and a third-generation EGFR-TKI compared to osimertinib as monotherapy (43). The different combination therapies come with an increased burden of toxicity, and how to best select which patients who benefit from treatment enhancement is still unclear. Future studies should be done to elucidate whether CD73 could be utilized as a marker for not only resistance to treatment, but also to guide upfront treatment. Therefore, it is of great importance that clinical studies are designed to incorporate well conducted translational analyses to identify biomarkers for patient selection in the choice of treatment at different stages of the disease.

There are several limitations to this work. Progression samples were scarce, and thus the sample size small, precluding firm interpretation of the results. We did not have any information on PD-L1-expression which could be relevant in the context of evaluating the immunogenicity of the tumors, especially as data regarding the role of PD-L1 in EGFR-mutant NSCLC are conflicting (44, 45). Furthermore, more extensive analyses to reveal other resistance mechanisms to TKI-treatment at the different time points could have been informative. Taken together, this study is descriptive and should be considered hypotheses generating, stimulating further research in this field.

In conclusion, in this explorative biomarker study we found that CD73 is expressed in higher degree in tumor samples with non-T790M-mediated resistance after treatment with first- or second-generation EGFR-TKIs. Although the study design and limited sample size do not allow for causality assessments, these findings contribute to the concept of possible involvement of immune markers in the development of resistance to EGFR-TKIs. Further research to pursue the identification of resistance mechanisms, predictive biomarkers and efforts to enhance immunogenicity of EGFR-mutated NSCLC should be undertaken.