Effect of bevacizumab plus XELOX (CapeOX) chemotherapy on liver natural killer cell activity in colorectal cancer with resectable liver metastasis

Abstract Aim We investigated the chemotherapy effect of resectable colorectal cancer with liver metastasis (CRLM) on the function of intrahepatic immune cells. Methods We classified patients into adjuvant chemotherapy (bevacizumab+CapeOX) after hepatectomy group (group A) and neoadjuvant chemotherapy followed by hepatectomy group (group B), and collected peripheral blood mononuclear cells (PBMC) and liver mononuclear cells (LMNC) to ascertain phenotypic and functional differences. Results There were no significant differences in lymphocyte fractions of either PBMC or LMNC between groups, except for the significantly lower percentage of natural killer (NK) cells in LMNC in group B than in group A. Significantly higher percentage of natural‐killer group 2, member D (NKG2D)‐ positive NK cells in PBMC and percentage of tumor necrosis factor‐related apoptosis‐inducing ligand (TRAIL)‐, NKp30‐, and signal regulatory protein β (SIRPβ)‐positive NK cells in LMNC were found in group B. Furthermore, significantly higher expressions of NKG2D and SIRPβ in peripheral blood NK cells and of NKp46 and CD122 in liver NK cells were found in group B. When LMNC were incubated with interleukin (IL)‐2 in vitro, no difference was observed in the expression of these molecules in NK cells between groups. Consistently, there was no difference in the cytotoxic activity of those LMNC against a colon adenocarcinoma cell line between groups. Conclusion Colorectal cancer with liver metastasis patients treated with neoadjuvant chemotherapy showed enhanced expression of activation markers on peripheral blood and liver NK cells in comparison with patients who did not receive therapy; however, the difference in those function remains unclear. These results suggest that neoadjuvant chemotherapy does not have a negative impact on intrahepatic immune cells in resectable CRLM patients.


| INTRODUCTION
The primary clinical complication of colorectal cancer (CRC) is the invasion of tumor cells into distant organs and outgrowth of metastases. CRC with liver metastasis (CRLM) is a major prognostic factor for CRC patients. 1,2 Disease-specific mortality of progressive CRC has decreased significantly in past decades with novel chemotherapeutic agents, such as oxaliplatin and irinotecan, and molecular targeted drugs, such as bevacizumab, a recombinant humanized monoclonal antibody that targets vascular endothelial growth factor (VEGF). 3 Adding bevacizumab to chemotherapy significantly reduces the residual viable tumor cell volume in resected tumors and increases the proportion of patients eligible for liver metastasis resection, when compared with chemotherapy alone. 4,5 Despite recent advancements in chemotherapy strategies for the treatment of advanced CRC patients, surgical resection of CRLM has been established as the treatment of choice and is the most effective and potentially curative therapy. [6][7][8][9][10] Furthermore, surgical resection for CRLM combined with systemic adjuvant chemotherapy has a potential benefit to be curative for CRLM patients. 11,12 Meanwhile, the advantage of neoadjuvant chemotherapy in initially resectable CRLM patients is the treatment of undetected distant micrometastasis, thereby reducing recurrence risk after surgery. 13 Although various studies have reported on the treatment strategy in resectable CRLM patients, the optimal treatment sequence remains unclear.
Considering the impact of both surgical resection and chemotherapy on host immunity is imperative in the treatment of resectable CRLM. Immune systems surrounding cancer cells are known to play crucial roles in regulating cancer cell proliferation, invasion, and metastasis through immunosurveillance. 14,15 Our group and other researchers have shown that the decreased activity of immune cells in the liver after hepatectomy leads to tumor growth in mice. 16,17 Although the influence of surgery on immune surveillance against cancer cells has been occasionally investigated, the influence of chemotherapy on that remains to be elucidated. Studies have reported that the presence of tumor-infiltrating immune cells, such as natural killer (NK) cells and T cells, in CRLM patients improved the overall survival (OS). 6,18 NK cells are part of the innate immune system and may provide a first line of defense against neoplastic cells by exerting an effector function without the necessity for priming. 19 In addition, NK cells are abundant in human liver, and liver NK cells have remarkably higher cytotoxic activity against neoplastic cells than peripheral blood NK cells. 20,21 Such unique anatomical distribution and functional property of NK cells in the liver prompt us to investigate the influence of chemotherapy, particularly neoadjuvant chemotherapy, on the immunity of NK cells in the liver of CRLM patients.
In a clinical setting, we are conducting a phase II/III randomized clinical trial in Hiroshima Surgical Study Group of Clinical Oncology (HiSCO) to determine whether neoadjuvant chemotherapy followed by hepatectomy is superior to adjuvant chemotherapy after hepatectomy in resectable CRLM patients regarding progression-free survival (PFS), OS, and time to treatment failure. As this clinical trial provided a good opportunity to obtain samples from resectable CRLM patients, who were either previously exposed or not exposed to chemotherapy, we additionally but separately designed an independent study to investigate the effect of neoadjuvant chemotherapy in resectable CRLM patients on liver NK cell activity as subanalysis of a clinical trial.

| Study design and procedures
We have been conducting a phase II/III randomized clinical trial in

| Collection of mononuclear cells
We analyzed the contents of immune cells and phenotype using peripheral blood mononuclear cells (PBMC) and liver mononuclear cells (LMNC) obtained from eligible patients in both groups. We collected blood samples at the time of hepatectomy, and PBMC were isolated by gradient centrifugation with Separate-L (Muto Pure Chemicals Co., Ltd, Tokyo, Japan) from 40 mL heparinized peripheral blood. In addition, LMNC were obtained by ex vivo perfusion through the portal vein of resected livers from CRLM patients as previously described. 20 Effluents were condensed by centrifugation, and LMNC were isolated by gradient centrifugation with Separate-L.
Cultured cells were harvested and used for phenotypic analyses and cytotoxicity assays, which were carried out using FACSAriaII (BD Biosciences) and FlowJo 7.6.5 software. Furthermore, we used DLD-

| Real-time polymerase chain reaction
Total RNA was isolated from resected tumor tissues of CRLM using RNeasy Mini kit (Qiagen, Limburg, the Netherlands) and reverse-T A B L E 1 Eligibility criteria for patients with colorectal cancer with liver metastasis Eligibility criteria (1) Primary lesion histologically diagnosed as CRC (2) Presence of CRLM of stage H1 or H2 (3) Metastatic lesion in the liver required less than 60% of liver resection and allowed resection with a microscopically negative margin (4) No distant/peritoneal metastasis other than CRLM (5) The primary lesion had already been or could be resected with a microscopically negative margin (6) No history of local therapy such as radiofrequency ablation or chemotherapy/radiotherapy for the CRLM (7) No history of chemotherapy involving the use of oxaliplatin (8) The liver disease could be classified as Child-Pugh class A

| Statistical analysis
Data are presented as mean ± standard deviation (SD   Figure S1).

| Potential augmentation of NK cell activity in the liver of patients receiving neoadjuvant chemotherapy followed by hepatectomy
We assessed phenotypic differences in NK cells, which play a pivotal role in tumor surveillance, to investigate the effect of neoadjuvant chemotherapy on the innate-immune system in CRLM patients.
Mean fluorescent intensity (MFI) of NKG2D and SIRPβ, which is associated with the small transmembrane adapter protein DAP12, transduce stimulatory signals, 26 and are expressed on activated NK cells 27 on peripheral blood NK cells, was significantly higher in group B than in group A (P = 0.03 and P = 0.04, respectively; Figure 2A).  Figure 3A,B). Consequently, the cytotoxic activity of IL-2stimulated LMNC obtained from both groups showed no significant differences ( Figure 3C).

| Neoadjuvant chemotherapy did not affect the expression levels of NKG2D ligands in liver tumors
We then evaluated the influence of chemotherapy on the expression of tumor-specific antigens that were theoretically recognized by NK cells on tumors from CRLM patients. Ligands for NKG2D comprised human class I-like molecules MICA, MICB, and ULBP, which are stress-induced molecules expressed by tumors of epithelial origin and activate NK cell cytotoxicity through their NKG2D receptor. 29,30 Although we carried out immunohistochemistry using frozen sections  we could not exclude the possibility of inconsistencies arising from this limitation, no significant differences were observed in the expression levels of these three targets between the two groups ( Figure 4).

| DISCUSSION
Effect of chemotherapeutic agents on immune cell function is considered to be involved in the immunosuppression associated with myelosuppression or drug cytotoxicity. 31   CapeOX directly influences the existence of NK cells in the liver. To address this concern, further investigation may be required. Recently, a complete pathological response was shown to be correlated with high rates of the OS and PFS in advanced CRC patients who had undergone neoadjuvant and conversion chemotherapy before resection of CRLM. 47 In the present study focusing on the early limited phase until hepatectomy, we found no significant