Small molecule tyrosine kinase inhibitors modulated blood immune cell counts in patients with oncogene-driven NSCLC

Background Lack of biomarkers and in vitro models has contributed to inadequate understanding of the mechanisms underlying the inferior clinical response to immune checkpoint inhibitors (ICIs) in patients with oncogene-driven non-small cell lung cancer (NSCLC). Methods The effect of small molecule tyrosine kinase inhibitors (TKIs) on peripheral blood mononuclear cells (PBMCs) in 34 patients with oncogene-driven NSCLC (cohort A) was compared with those from 35 NSCLC patients without oncogene-driven mutations received ICI (cohort B) or from 22 treatment-naïve NSCLC patients (cohort C). Data for each blood biomarker were summarized by mean and standard deviation and compared by Wilcoxon rank sum tests or Kruskal-Wallis tests with significance at 2-sided p value < 0.05. Co-culture of PBMCs and pleural effusion-derived tumor cells from individual patients with oncogene-driven NSCLC was used to determine the in vitro cytotoxicity of TKI and ICI. Results Except for low CD3% in cohort A, there were no significant differences in other 12 blood biomarkers among the 3 cohorts at baseline. TKI treatment in cohort A was associated with significant increase in CD3% and decrease in total and absolute neutrophils (p < 0.05). In cohort B, patients with good clinical response to ICI treatment (N = 18) had significant increases in absolute lymphocyte counts (ALCs), CD4 and/or CD8 cell counts. Conversely, those patients with poor clinical response to ICI (N = 17) had significant decreases in these cell counts. Of the 27 patients with pre- and post-treatment blood samples in cohort A, 11 had poor clinical response to TKIs and decreased lymphocyte counts. Of the remaining 16 patients who had good clinical response to TKI therapy, 10 (62.5%) patients had decreased, and 6 (37.5%) patients had increased lymphocyte counts. Multicolor immunophenotyping of PBMCs revealed ICI treatment activated additional immune cell types that need further validation. We confirmed that TKI treatment could either antagonize or enhance the effect of ICIs in the co-culture assay using patient’s tumor cells and PBMCs. Conclusions To the best of our knowledge, this is the first study showing that TKIs can have various effects on blood immune cells, which may affect their response to ICIs. Further validation of the blood biomarker and in vitro assay is warranted.


Conclusions:
To the best of our knowledge, this is the first study showing that TKIs can have various effects on blood immune cells, which may affect their response to ICIs. Further validation of the blood biomarker and in vitro assay is warranted.
Keywords: Tyrosine kinase inhibitor, Peripheral blood mononuclear cells, Immune cells, Oncogenic-driven, NSCLC, In vitro cytotoxicity, Malignant pleural effusion Background Immune checkpoint inhibitors (ICIs) have revolutionized the diagnosis and treatment for patients with locally advanced or metastatic non-small cell lung cancer (mNSCLC). However, ICIs have low or inferior clinical efficacy compared to chemotherapy in patients with epidermal growth factor receptor (EGFR)-mutant or anaplastic lymphoma kinase (ALK)-rearranged mNSCLC [1][2][3]. This low clinical efficacy of ICIs in EGFR-mutant or ALK-rearranged mNSCLC remains even when their tumors had high Programmed death-ligand 1 (PD-L1) immunohistochemistry (IHC) expression [4]. Furthermore, ICI treatment has been associated with increased incidence and severity of interstitial lung disease and immune-mediated adverse effects (including pneumonitis, colitis and hepatitis) when they are in sequential or concurrent use with small molecule tyrosine kinase inhibitors (TKIs) in patients with EGFR-mutant or ALKrearranged mNSCLC [5]. Thus, mNSCLC patients with oncogene-driven mutations have been excluded in the first-line ICI trials except in the case of atezolizumab in IMpower 150 study [6]. The addition of atezolizumab to carboplatin, paclitaxel and bevacizumab (CPB) had superior clinical activity compared to CPB in a small cohort of patients with EGFR-mutant or ALK-rearranged cohorts after first-line TKI therapy. The key effect of angiogenesis inhibitor bevacizumab is postulated, which has synergism with each of the other 3 components. Recently, the updated report of PACIFIC [7] and retrospective analysis on durvalumab consolidation for patients with stage III NSCLC [8,9] suggested that durvalumab might have limited clinical efficacy in the small subset of patients with EGFR-or HER2-mutant NSCLC.
ICIs are designed to activate exhausted tumor-reactive T cells, which are responsible for killing tumor cells. Current data suggest that the presence of high membranous PD-L1 IHC staining on tumor cells and the presence of intratumoral PD-1 expressing tumor infiltrating lymphocytes (TILs) in the tumor microenvironment (TME) are favourable prognostic factors and the best predictive factors of clinical response to ICIs [10]. The presence of a T-cell inflamed gene expression profile (GEP) in addition to PD-L1 IHC has improved the prediction of favourable clinical response to ICIs [10]. ICIs could increase the number of absolute lymphocyte counts (ALCs), restore the function in exhausted CD8+ T cells and induce phenotypically and functional changes of effector immune cells [11]. Several mechanisms have been postulated for the underlying mechanisms by which patients with oncogene-driven NSCLC do not derive clinical benefit from ICI [12]. These include low PD-L1 expression on tumor cells and TILs, low tumor mutation burden (TMB) and immune escape using other immune checkpoints in patients with EGFRmutant NSCLC [13][14][15][16]. However, these mechanisms do not explain the inferior clinical response observed in patients with oncogene-driven NSCLC and high PD-L1 expression [17]. Further, functional studies have been hampered by insufficient paired tumor specimens before and after treatment. Along with, a lack of relevant human NSCLC models that can simulate the interaction and delineate the mechanisms of ICI with TKI, chemotherapy, and/or bevacizumab.
Increasingly, liquid biopsy with blood and malignant body fluids have been used to provide a minimally invasive way to study tumor biology and monitor dynamic changes of molecular and immune biomarkers during cancer treatment [18]. Previous studies have shown that pre-treatment low ALCs (< 600-1200 cells/μL) were associated with decreased progression-free survival (PFS) and overall survival (OS) to ICI in NSCLC patients [19][20][21][22]. Dynamic changes of ALCs after ICI treatment were also associated with clinical response. Post-ICI treatment low ALCs (< 700-900 cells/μL) were associated with decreased PFS and OS in patients with advanced solid cancer types including NSCLC [20,23]. Immunophenotypic analysis of circulating immune cells revealed increases in circulating proliferating CD4+ and CD8+ T cells at 2 weeks after durvalumab treatment [24]. Assessing the function of peripheral T-cell subclones, particularly the T-cell receptor (TCR) clonality and activity to clonal neoantigens, have also been explored as a predictive biomarker for response to ICI [24,25]. Furthermore, high derived neutrophil-to-lymphocyte ratio (dNLR) was associated with poor prognosis in patients with advanced NSCLC treated with durvalumab [26]. However, the effect of small molecule TKIs on these blood immune cells in patients with oncogene-driven NSCLC are unknown. The objective of this study was to determine the effect of small molecule TKIs on blood immune cells in patients with oncogene-driven NSCLC. We also explored the feasibility of using tumor cells from malignant pleural effusion and patient's own peripheral blood mononuclear cells (PBMCs) for in vitro evaluation of the effect of targeted therapy and ICIs.

Study patients and biospecimen collection
Lung cancer patients receiving care at an academic institution between March 2017 and March 2021 were retrospectively identified through chart review if their tumors had been tested for tumor genomic profiling by a clinical next generation sequencing (NGS) assay under an Institutional Review Board (IRB) approval protocol (University of California, Davis Protocol No. 937274). Patients whose tumors had at least one driver oncogene, defined as EGFR, MET exon 14 skip or ERBBR2 (HER2) mutation, ALK, ROS1 or RET fusions, were defined as oncogene-driven NSCLC according to National Comprehensive Cancer Network (NCCN) guidelines. Demographic information, clinical genomic sequencing results, and complete blood cell counts with differentials were abstracted from electronic medical records. Fresh biofluids (blood and malignant pleural effusion if available) were collected via an IRB approved protocol (University of California, Davis Protocol No. 226210). Multiple samples from the same patients at different time points were collected during the disease course. Cell pellets from malignant pleural effusion were washed three times in 1x phosphate-buffered saline (PBS) buffer containing 0.2% BSA and 10 mM ethylenediamine tetraacetic acid (EDTA) and resuspended in 10 ml of the 1x PBS buffer for use. The samples were handled under strict operating procedures for collection, processing, and storage to minimize the variation in handling of samples.

Immunophenotyping of PBMCs by flow cytometry
Fresh or thawed PBMCs were immunophenotyped for T cell subsets and reported as percentages of total PBMCs and as percentages of total T cells in the case of T cell subtypes. Specifically, PBMCs were stained with wellcharacterized antibodies against markers of interest, including CD3, CD4 plus CD8 according to standard protocols. Results were analyzed using a BD Fortessa multicolor flow cytometer and FlowJo 7.6.1 program (Ashland, OR). Further, immunophenotypic changes of major innate and adaptive immune cells of ICI were evaluated using a minimum of a million PMBCs collected before and after cancer therapy for individual lung cancer patients by a 24-color antibody panel using the "Aurora" Spectral cytometer (Cytek Biosciences, CA). The data analysis and statistical evaluations of this complex data set were performed using the Cytobank (Cytobank Inc., CA) [27].

Data collection and statistical analysis
Data were summarized according to frequency and percentage for qualitative variables, and by mean ± standard deviation (SD) for quantitative variables unless noted otherwise. The 95% confidence interval for survivals was calculated using the exact binomial distribution. For each of the 13 blood cell types, the cell counts were summarized using mean and SD for pre-and posttreatment in cohort and response subgroups. Wilcoxon rank sum tests were used to compare two groups (or Kruskal-Wallis tests for three cohorts). Two-sided P < 0.05 was used to determine statistical significance. Due to the exploratory nature, adjustment for multiplicity was not performed for the types of blood cells [28]. Statistical analyses were carried out using SAS version 9.4 (SAS Institute, Cary, NC).
According to the National Cancer Institute (NCI) Common Terminology Criteria for Adverse Events (CTCAE) Version 5.0 (http://ctep.info.nih.gov), lymphopenia (< 1000 cells/μl) was defined as grade 1: ALCs 800-999 cells/μL; grade 2: ALCs 500-799 cells/μL; grade 3: ALCs 200-499 cells/μL and grade 4: ALCs < 200 cells/μL. Best response to systemic therapies, defined as a complete or partial response (CR or PR), stable disease (SD) or progression disease (PD) achieved to cancer treatment, was assessed using Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 [29]. PFS was measured as the time from the first administration of a cancer therapy to progression defined by RECIST1.1, or death due to any cause. Patients alive without progression at the time of analysis were censored at the initiation of a new therapy or last follow-up. Good clinical response was defined in patients who achieved a CR, PR, PFS exceeding the reported median PFS for each targeted therapy. OS was measured as the time from the first administration of a cancer therapy to death due to any cause. Patients alive at the time of analysis were censored at the initiation of a new therapy or last follow-up. Survival data were estimated using the Kaplan-Meier method and compared using the log-rank test in each cohort and response subgroups. For the blood cell types significant in univariable survival analysis, multivariable survival analysis was further conducted using Cox proportional hazards models, by further adjusting for age, gender, race and histology in the model of each cell type.
Co-culture of patient's tumor cells and PBMCs and growth inhibition by the MTS assay H1975 and primary tumor cells isolated from the malignant pleural effusion of NSCLC patients with oncogenedriven mutations were seeded in 96-well plates at a density of 5 × 10 3 cells/well overnight. Human PBMCs from 20 ml of blood from the same patients were first purified by Ficoll-Paque and washed twice in 1x PBS. The acquired PBMCs were counted and added in a 2:1 ratio to the seeded tumor cells on the 96-well plates for co-culturing. After 8-12-h incubation, the cells were treated with various concentrations (0, 0.001, 0.01, 0.1, 1, 10 μM) of an TKI (osimertinib or alectinib) and/or an ICI (nivolumab (10 μg/mL), or atezolizumab (10 μg/mL) as indicated. The MTS assay was performed and analyzed as described previously [30,31]. Specifically, after 72-h incubation, the MTS solution (3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium) was added and the cell viability was measured by the absorbance at 490 nm using a microplate reader (SpectraMax M3, Molecular Devices, USA). Untreated cells served as a control. Results were shown as the average cell viability ± SD [(ODtreat −ODblank)/(ODcontrol−ODblank) × 100%] of triplicate wells. Data were presented as the mean ± SD unless noted otherwise. All the experiments were performed in triplicate. Statistical analysis was performed using Graph Prism software (Version 8.21). Two-sided, P < 0.05 was considered statistically significant.
Immunophenotypic changes of PBMCs in a patient who received ICI combination Figure 7a summarizes the clinical course of a patient with EGFR exon 19 deletion who received ICIs at diagnosis when tumor genomic profiling test did not identify the EGFR mutation. Patient subsequently received platinum-based chemotherapy for 2 cycles and has been having clinical PR to EGFR TKI osimertinib. The changes in ALCs, CD4, CD8, and CD4 plus CD8 cell counts during the treatment course were shown in Fig. 7c. Although CD4+ and CD8+ T cells account for the majority of ALCs, ALCs include a few rare yet important lymphocyte subtypes such as natural killer (NK) cells and monocytes. In addition, changes in these lymphocytes might affect other immune cell types such as B cells in the PBMCs. Further, the immunophenotypic analysis of PBMCs collected before and after ICI

Discussion
Our study has several clinical implications. First, lymphocyte counts and immunophenotyping of T-cell and B-cell have been used to assess the immune status and are prognostic biomarkers. Low levels of lymphocytes in the blood could indicate an increased risk for death [32]. The American Society of Hematology (ASH)-American Society for Transplantation and Cellular Therapy (ASTCT) defined patients with neutropenia ANC < 500 cells/μL and lymphopenia ALCs< 200 cells/μL as immunodeficiencies and recommended against COVID-19 vaccination [33]. We found that all study patients had ALCs > 200/μL during almost the entire disease course (except 3 patients at imminent dying stage), which is safe for receiving COVID-19 vaccination [34]. It is well known that many cancer treatments for hematological malignancies are immunosuppressive, which include cytotoxic chemotherapy, monoclonal antibodies against CD20, CD38 and CD52, calcineurin inhibitors, Mammalian Target of Rapamycin (mTOR), bortezomib and Bruton's tyrosine kinase (BTK) inhibitors [35,36]. A recent study showed that severe lymphopenia (ALCs < 500 cells/μL) before starting the consolidation durvalumab in patients with unresected locally advanced NSCLC after Tick marks indicate censored data. Groups were compared using the Z test. P < 0.05 for statistical significance Fig. 5 Kaplan-Meier PFS and OS estimates according to ALCs, CD4 plus CD8 counts, PD-L1 IHC and TMB in cohort A. Post-treatment low ALCs (< 800 cells/μL, red) were associated with shorter PFS and OS compared to high ALCs (≥800 cells/μL, blue) in patients with oncogene-driven NSCLC (cohort A) (A and B), respectively. Post-treatment low CD4 plus CD8 cell counts (< 500 cells/μL, red) were associated with shorter PFS and OS compared to high CD4 plus CD8 cell counts (≥500 cells/μL, blue) (cohort A) (A and B), respectively. PD-L1 IHC expression (C) and TMB status (D) did not affect the PFS and OS in NSCLC patients with oncogene driven NSCLC. Tick marks indicate censored data. Groups were compared using the log-rank test. P < 0.05 for statistical significance   6 Kaplan-Meier PFS and OS estimates according to ALCs, CD4 plus CD8 counts, PD-L1 IHC and TMB in cohort B. In patients without oncogene-driven NSCLC (cohort B), post-treatment low ALCs (< 800 cells/μL, red) were associated with statistically significant shorter PFS (A) but not OS (B) compared to high ALCs (≥800 cells/μL, blue). Post-treatment low CD4 plus CD8 cell counts (< 500 cells/μL, red) were associated with shorter PFS and OS compared to high CD4 plus CD8 cell counts (≥500 cells/μL, blue) (cohort B) (A and B), respectively. Patients with NSCLC expressing PD-L1 IHC < 50% had shorter PFS and shorter OS compared to those patients with NSCLC expressing PD-L1 IHC ≥50% (C). Patients with NSCLC expressing TMB ≥10 Mut/Mb had shorter PFS but not OS compared to those patients with NSCLC expressing TMB < 10 Mut/Mb (D).
Tick marks indicate censored data. Groups were compared using the log-rank test. P < 0.05 for statistical significance completion of definitive chemoradiation was associated with worse PFS compared to those patients without severe lymphopenia [37]. Although immune modulation is well documented in targeted therapy with monoclonal antibody for solid tumors [38], there is no prior study to determine the effect of TKIs on blood immune cells in patients with oncogene-driven NSCLC. In large clinical trials of TKIs for patients with advanced oncogenedriven NSCLC, the incidence and severity of leukopenia, neutropenia, and lymphopenia are generally less than 5% (Table 6) [39][40][41][42][43][44][45][46][47][48][49]. We observed that at baseline the patients with oncogene-driven NSCLC in cohort A had lower CD3% compared to patients without oncogenedriven NSCLC in cohort B and control group in cohort C. After TKI treatment, and these patients in cohort A had significantly increases in CD3% and decreased WBC, ANC and dLNR (Table 4). Consistent with previous reports, we also found that cancer progression [50] cytoreduction by chemotherapy and/or radiation [51][52][53][54], infection [32], and steroids use [55] were common reasons associated with decreased lymphocyte counts in patients with mNSCLC. This could reverse promptly with the clinical improvement or stopping steroids.
While the changes of immune cell counts were associated with good or poor clinical responses to ICIs, targeted therapy with TKIs could modulate immune cell counts that mimic good or poor clinical responses to ICIs. The impact of our observation on patient's response to ICIs is unknown and needs further exploration. Second, small molecule TKIs targeting a growing number of gain-of-function molecular targets, such as EGFR, BRAF V600E, MET exon 14 skipping mutations, and ALK-, ROS1-, and neurotrophic receptor tyrosine kinase (NTRK)-or RET-gene fusions have been shown to improve PFS and OS with favorable toxicity profiles compared to platinum-containing combination chemotherapy as first-line systemic therapy in~25% of NSCL C patients (Table 8). It is important to understand the role of TKIs on the immune system and treatment with ICIs. Compared to molecular biomarkers, immune biomarkers are complex with many different components that are subjected to change during natural tumor progression and treatment. A coordinated response by both humoral immunity and cell-mediated immunity is important to the response to ICIs [64,65]. Using multiplex   1 μM), nivolumab (10 μg/ml), atezolizumab (10 μg/ml) or combination as indicated for 72 h. Growth inhibition was measured using the MTS assay using vehicle as 100% control. Alectinib-sensitive ALK-fusion NSCLC cells (C) and alectinib-resistant RET-fusion NSCLC cells (D) were cocultured with PBMCs from corresponding patients and treated with vehicle, alectinib (0.1 μM), nivolumab (10 μg/ml), atezolizumab (10 μg/ml) or combination as indicated for 72 h. Growth inhibition was measured using the MTS assay with vehicle as 100% control. All data are shown as mean of triplicate samples. Error bars indicate standard deviation (SD). Groups were compared by the Wilcoxon signed rank test. P < 0.05 was considered statistically significant. Abbreviations: MTS, 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium) CD8 + PD-1+ cells appeared similar. We observed that CD4 plus CD8 counts were an independent biomarker for PFS of NSCLC patients in both cohort A and B, and OS of NSCLC patients in cohort B. ALCs were an independent biomarker for OS in patients with oncogenedriven NSCLC. In cohort B, post-treatment CD4 plus CD8 cell counts performed better than the known immune biomarkers (PD-L1 IHC and TMB) in predicting response to ICIs. Further characterization of the effect of TKIs on the expression and function of immune cells in oncogene-driven NSCLC are warranted.
Upregulation of PD-L1 by EGFR activation mediates the immune escape in EGFR-driven NSCLC, implicating an optional immune targeted therapy for NSCLC patients with EGFR mutation. In genetically engineered mouse models (GEMMs), EGFR-driven tumors express higher levels of PD-L1 with a more immunosuppressive tumor microenvironment (increased FoxP3+ T-cells, decreased CD8+/CD4+ ratio). The addition of an EGFR-TKI in these EGFR-mutant GEMM models modulated PD-L1 expression and reversed EGFR-pathway mediated immunosuppression. However, ICIs either alone or in combination with small molecule TKIs have low or inferior effect in EGFR-mutant or ALK-rearranged NSCL C. Therefore, GEMMs are not good models to study the effect of ICIs on human NSCLC tumors. As PD-L1 expression is lower in the majority of oncogene-driven NSCLC compared to non-oncogene-driven NSCLC, this may explain why the results in GEMMs did not correlate with the clinical observation. We also did not observe Abbreviations CR complete response, mo month, mPFS median progression free survival, NA not available, ORR overall response rate, PR partial response significant predictive or prognostic association between lymphocyte counts and correlation of PD-L1 or TMB. Using multiplex flow cytometry, we evaluated the immunophenotypic changes of PBMCs in a patient with EGFR E19 deletion who received 3 cycles of ICIs. We found that the ICI treatment increased lymphocytes, CD4+, CD4 terminally differentiated effector memory (TEMRA; CD45RA+ CCR7-), and CD8 naïve cells, and decreased all other types of immune subtype cells, including PD1 + CD4 and PD1 + CD8 cells. Of note, NSCLC patients with oncogene-driven mutations, such as EGFR, have been associated with hyperprogression to ICIs [66]. The impact of our observation and the association of blood circulating immune cells with TILs in TME in patients with oncogene-driven NSCLC after ICI treatment is unknown and deserves further exploration. Malignant biofluids such as pleural effusion is a unique source for liquid biopsy that is currently underused for molecular diagnosis and tumor biology study. Pleural effusion affects at least 40% of patients with lung cancer. Up to 90% of these patients have confirmed malignant pleural effusion (MPE) and require palliative thoracentesis for symptomatic relief [67]. We explored the effect of TKIs on patient's tumor cells from malignant pleural effusion and PBMCs using an in vitro coculture model. We observed significant variations in individual's response to different ICIs that were not clearly associated with tumor PD-L1 expression or sensitivity to targeted therapy with TKIs. To the best of our knowledge, this is the first study to determine the effect of small molecule targeted therapy on lymphocyte cells using NSCLC patient's blood samples and malignant tumor cells. Ongoing study is determining the clinical utility of this in vitro assay in predicting clinical response to ICI therapy in patients with oncogene-driven NSCLC.
There are several limitations to this study, including its small sample size, it being a retrospective study, and no adjustment for multiplicity due to the exploratory nature. The potential selection bias and the imbalance of the baseline characteristics and treatment history in patients may have contributed to the treatment outcomes. Furthermore, dysregulation of various immunoregulatory cells and cytokines in the TME may be responsible for tumor response [68,69]. We did not perform the functional analysis of immune subtypes in patient PBMCs collected before and after TKI or ICI treatment. Further study is needed to confirm our findings and understand the effect of TKI modulation on "cold" or "hot" tumor microenvironment.

Conclusions
Molecularly targeted therapy by small molecule TKIs have various effects on modulating the blood immune cell count in patients with oncogene-driven NSCLC.
There are unmet needs to understand the underpinning mechanisms and develop predictive biomarkers and assays to select the appropriate patients for ICI therapy.