Current status of immunotherapy against gastrointestinal cancers and its biomarkers: Perspective for precision immunotherapy

Abstract Immunotherapy has shown encouraging results for some types of tumor. Although enormous efforts have been made toward the development of specific immunotherapeutic strategies against gastrointestinal cancers, such as adoptive T‐cell transfer, peptide vaccines, or dendritic cell vaccines, the efficacy of immunotherapies prior to the introduction of immune checkpoint inhibitors was not substantial. This article reviews immunotherapy for gastrointestinal malignancies, including cell therapy, peptide vaccine, and immune checkpoint inhibitors, and attempts to resolve the immunosuppressive conditions surrounding the tumor microenvironment, and to construct novel combination immunotherapies beyond immune checkpoint inhibitors.


| INTRODUCTION
Gastrointestinal (GI) cancers are the most common human tumor worldwide, and the incidence and mortality are increasing every year. 1 Several treatment strategies have been developed for GI cancers, including surgery, chemotherapy, radiotherapy, and molecularly targeted therapy. However, the overall survival (OS) of patients with GI cancer remains poor. Novel approaches to the treatment of GI cancer are thus needed. 2 Immunotherapy is a novel treatment strategy that is emerging as an effective and promising treatment option against several types of cancer. 3 The first immunological treatments were carried out by Coley using a bacterial immunotoxin to patients with malignancy in 1891. 4 The first notion of a role of immunity in cancer was postulated in 1909 by Smith,5 speculating that the immune system could repress the growth of carcinomas by recognizing tumor cells as foreign. In 1970, the concept of immunological surveillance was presented by Burnet,6 and an antigen recognized by cytolytic T lymphocytes on a human melanoma was finally reported by van der Bruggen et al. 7 Progress in this field is largely attributable to the identification of new immune-based targets, based on continued advances in the understanding of tumor immunology and the tumor microenvironment. 8 Many types of immunomodulatory therapies have been demonstrated in the treatment of GI cancers, including non-specific biological response modifiers (OK432, 9 lentinan, 10 PSK 11 ), interleukin (IL)-2-activated lymphocytes, 12 tumor-specific reactive CD8 + T-lymphocyte transfer, 13 dendritic cell (DC) vaccines, 14,15 and tumor-associated antigen (TAA)-derived peptides. [16][17][18][19] These immunotherapies have shown a certain degree of efficacy, but not durable objective responses. 20 Confidence in the efficacy of immunotherapies was given a boost with the advent of immune checkpoint inhibitors, which was selected as a "Breakthrough of the Year 2013" by Science. 21 Immunotherapy is now becoming mainstream as a treatment for GI cancer.
In the present study, we review immunotherapies for GI malignancies, including immune cell transfer therapy, peptide vaccine, immune checkpoint inhibitors, and combination immunotherapy beyond immune checkpoint inhibitors, by clarifying suppressive immune biomarkers surrounding the tumor microenvironment.

| Adoptive T-cell transfer
The concept of adoptive immunotherapy (AIT) for cancer treatment was presented by Mule et al 22 in the form of IL-2 generated lymphokine-activated killer (LAK) cells combined with repeated injections of recombinant IL-2 (Table 1). Although LAK cells are nonspecific killer cells that were considered effective against various types of tumor, the efficacy of LAK cells combined with high-dose IL-2 proved limited against metastatic GI cancer. The objective response rate (ORR) including complete response (CR) or partial response (PR) for colorectal cancer (CRC) was 11% (3 of 27 patients), and 0% (0/1) for esophageal cancer. Furthermore, severe toxicities were observed as a result of high-dose IL-2, which induces a vascular permeability leak that leads to fluid retention and interstitial edema, and results in circulatory failure, lung edema, and renal dysfunction. Hence, they made the shift to tumor-infiltrating lymphocytes (TIL), which are specific to tumor antigens and appear to offer far greater therapeutic potency than LAK cells. 28 Takayama et al conducted a randomized study to evaluate the efficacy of autologous lymphocytes activated in vitro with recombinant IL-2 and solid-phase antibody to CD3 as adjuvant therapy for curatively resected HCC. A total of 150 patients who had undergone curative resection for HCC were assigned to receive either AIT (n = 76) or no adjuvant treatment (n = 74). The immunotherapy group showed significantly longer DFS (P = .01) and disease-specific survival (P = .04) than the control group. No patients experienced grade 3 or 4 adverse events. 23 These results suggested that transfer of non-specific-activated killer cells might be effective in preventing the intrahepatic recurrence of cancer.
The next advance was antigen-specific cytotoxic T lymphocytes (CTL) for the management of effector cells as treatment. 24 In Japan, Aruga  with metastatic liver cancer), two CR, three PR, and four minor responses were observed without any severe treatment-associated systemic adverse events. 29 We have assessed the efficacy of CTL against pancreatic ductal adenocarcinoma (PDAC). Patients with curatively resected PDAC received AIT with CTL stimulated using MUC1-expressing human cell lines (MUC1-CTL), and the results indicated that MUC1-CTL might prevent liver metastasis. 30 For the next step, combination therapy using MUC1-CTL and gemcitabine was carried out. A total of 43 patients who underwent radical pancreatectomy received treatment with MUC1-CTL and gemcitabine after surgery. MUC1-CTL were induced and given i.v. three times, and gemcitabine was given according to the standard regimen for 6 months. No severe treatment-associated systemic adverse events were encountered in the 43 treated patients. In the adequate treatment group (n = 21) in which the relative dose intensity of gemcitabine was ≥50% and ≥2 MUC1-CTL treatments were provided, disease-free survival (DFS) was 15.8 months, and OS was 24.7 months. Liver metastasis was found in seven patients only (33%), and local recurrence occurred in four patients (19%). Combination therapy with AIT and GEM might prevent liver metastasis and local recurrence. 25 As described above, adoptive immunotherapies have shown a certain degree of efficacy (Table 1) neoantigen recognized TIL or genetically engineered T cells such as T-cell receptor (TCR) T cells and chimeric antigen receptor (CAR) T

cells. 31
So-called CAR-T therapy was also selected as a "Breakthrough of the Year 2013" by Science. 21

CAR-modified T cells (CAR-T) targeting
CD19 showed durable effects against leukemia, achieving complete remission. 32 For gastrointestinal tumor, CAR-T therapies remain experimental. 33 In clinical studies of CRC 34 and biliary tract and pancreatic cancers, 35

| Dendritic cell vaccines
Dendritic cells are antigen-presenting cells specialized for the induc- Another target of DC therapy is HCC. A phase I trial was conducted on the basis of a previous basic study that reported overexpression of heat-shock protein (HSP) 70 in HCC using proteomic profiling and immunohistochemical staining. 38 DC transfected with HSP70 mRNA (HSP70-DC) by electroporation were injected intradermally. Patients were treated three times every 3 weeks, and the number of HSP70-DC injected was dose-escalated in a three-patient method, from 1 9 10 7 to 2 9 10 7 , and finally to 3 9 10 7 . No adverse effects at grade III/IV were observed, except for one case of grade III liver abscess at the 3 9 10 7 dose, and three patients were therefore added to confirm the safety of the 3 9 10 7 dosage. CR without any recurrence was achieved in two patients (for at least 44 and 33 months) and SD in five patients. That study indicated that HSP70-DC therapy is both safe and effective in patients with HCC. 14 DC vaccines might gain a place in novel combination immunotherapy.

| Peptide vaccines
Since the first clinical trial of a melanoma antigen gene-1-derived peptide-based vaccine was reported in 1995, 39 various types of next-generation peptide vaccine are currently under development. 16 Here, we present some successful reports from among these numerous studies.
We conducted phase I and phase II trials using HLA-A*24:02- patients treated using gemcitabine alone as a prospective control group that did not meet eligibility criteria as a result of HLA-A type only, for whom the median DFS was 12.0 months. No significant difference was seen between the two groups (P = .504). Significant differences in DFS were apparent between patients with and without KIF20A-specific CTL responses (P = .027), and between patients with and without KIF20A expression (P = .014). In addition, all four patients who underwent R0 resection with KIF20A expression showed no recurrence with KIF20A-specific CTL responses. 18

Sawada et al identified in glypican-3 (GPC-3) an HLA-A*24,
HLA-A*02 restriction peptide with extreme cancer specificity. In a phase I study, they reported safety, and immunological and clinical responses. 40 A subsequent trial showed a durable effect against giant HCC, although the patients died from circulatory failure as a result of tumor thrombus, which occupied most of the right atrium. 41 In the next phase II study of GPC3 peptide vaccine as an adjuvant therapy for HCC, no significant difference in recurrence rate was found between 35 patients treated with surgery plus vaccination and 33 patients who underwent surgery alone ( Although tumor-associated antigen-derived peptide vaccines have been shown to provide effective induction of antigen-specific immunity, the clinical efficacy has not proven durable ( Table 1). As a result, no peptides are covered by the National Health Insurance.
Pooled results of clinical trials show a very weak clinical response rate of <1% for the active specific immunization procedures currently available for advanced CRC. 20 Rosenberg et al reported that the objective response rate was low (2.6%) in their cancer vaccine trials of 440 patients, even though the main target was melanoma, which is highly immunogenic. 42 Combination immunotherapy appears needed for peptide vaccinations such as immune checkpoint inhibitors, 43 novel immune adjuvants, 44 COX-2 inhibitors, 45 and anti-epidermal growth factor receptor (EGFR) antibodies. 46 50 Tumors shrunk by about half or more in 31% of those with melanoma, in 29% with kidney cancer, and in 17% with lung cancer. Immune checkpoint inhibitors have also been applied to gastrointestinal cancers (Tables 2-4).

| Esophageal cancer
Compared with other solid tumors, esophageal squamous cell carcinoma (ESCC) has a very high somatic mutation rate. 61,62 The high mutation load in esophageal tumors has been associated with the clinical benefit of PD-1 blockade. 63 Nivolumab is a human monoclonal immunoglobulin (Ig)G4 antibody that seals PD-1 expressed on activated T cells. This drug was applied to treatment-refractory esophageal cancer in an open-label, multicenter, phase II trial (Table 2).
Nivolumab showed promising activity with a manageable safety profile. 51 PD-1/PD-L1 blockade alone or in combination with radiotherapy and chemotherapy will be a direction for future research in the treatment of advanced esophageal cancer (Table 4).

| Gastric cancer
To assess the efficacy and safety of nivolumab in patients with advanced gastric cancer (GC) or gastroesophageal junction cancer (GEJC) refractory to, or intolerant of, two or more previous regimens of chemotherapy, a randomized, double-blind, placebo-controlled, phase III trial was carried out ( pembrolizumab at 10 mg/kg once every 2 weeks (Table 2). Pembrolizumab showed a manageable toxicity profile and promising antitumor activity. 53 The FDA approved pembrolizumab for previously treated patients with recurrent locally advanced or metastatic GC or GEJC whose tumors express PD-L1. This decision was based on data from a global multicohort trial, KEYNOTE-059, which indicated a superior response in patients with tumors that expressed PD-L1 (Tables 2,3). 54 Ongoing trials are investigating various settings and earlier treatment lines for GC or GEJC (Table 4).

| Colorectal cancer
In early-phase studies, responses of CRC to PD-1/PD-L1 inhibitors were not promising. 60  Patients were given nivolumab at 3 mg/kg every 2 weeks until disease progression or unacceptable toxic effects (Table 2). 57 The FDA approved nivolumab use in the treatment for MSI-H or dMMR metastatic colorectal cancer that has progressed following treatment ( Table 3). Studies of immune checkpoint inhibitors are ongoing to indicate potential efficacy as first-line agents (Table 4).

| Hepatocellular carcinoma
The only evidence-based systemic treatment option is sorafenib, a small-molecule multikinase inhibitor, for patients with advanced  showed a manageable safety profile, including acceptable tolerability.
Incidence of treatment-related adverse events did not seem to be associated with dose and no maximum tolerated dose was reached, and nivolumab 3 mg/kg was chosen for dose expansion. Durable objective responses show the potential for nivolumab in the treatment of advanced HCC (Table 2), 59 and nivolumab received FDA approval for the treatment of hepatocellular carcinoma patients previously treated with sorafenib (Table 3).   70 These results clearly indicate that suppressive immunity should be controlled using a multidisciplinary approach ( Table 5) Figure 1D,E). 8 One way to achieve successful immunotherapies is to establish biomarkers for excluding those patients unlikely to respond to immunotherapy ( Figure 1E). The old-new biomarker is NLR, which is prognostic in many oncological settings. NLR kinetics in patients with advanced solid tumors treated with PD-1/PD-L1 inhibitors showed that the median OS for patients with a high NLR was 8.5 months, compared to 19.4 for patients with low NLR (P = .01). 88 The importance of NLR was also reported in a study of a peptide vaccine against CRC. 17  Another way to achieve successful immunotherapies is to alter tumor microenvironments and host immunosurveillance. So-called "hot tumor" with massive infiltration of CD8 + cells and without suppressive immunity could respond well to immune checkpoint inhibitor alone ( Figure 1A,D). Each of "dark tumor" (Figure 1B), which is highly immunogenic but with suppressive immunity, "cold tumor"

| Pancreatic ductal adenocarcinoma
( Figure 1C), which shows low immunogenicity with suppressive immunity, and patients with immune exhaustion ( Figure 1E) might require a multidisciplinary approach ( Table 5).
The immunosuppressive environment surrounding PDAC might be one of the major obstacles to the development of successful therapies for this fatal disease. 69 (Table 5). 78 Other inhibitory factors in the tumor microenvironment, such as TGF-b 80 and PGE2 75,81 and its inhibitors, are summarized in Table 5.
Another problem is the presence of a uniquely desmoplastic stroma that functions as a barrier to T-cell infiltration ( Figure 1B,C).  Table 5.
Tumor tissues that lack expression of many immunological markers may indicate a non-immunogenic tumor microenvironment (Figure 1C), which may require combination therapies consisting of an agent to create an immunogenic tumor microenvironment plus an immune checkpoint agent to further enhance immune responses for clinical benefit. 86 Conventional cancer therapies such as chemotherapy or radiation may also lead to tumor cell death and release of antigens to initiate activation of T cells, which may then migrate into tumor tissues. Combination studies using conventional agents and immune checkpoint therapies should thus clarify the conditions needed to create an "immunogenic" tumor microenvironment with subsequent clinical benefit for patients. 86 We have reported that cetuximab strongly enhances immune cell infiltration into liver metastatic sites in CRC. 46 Cetuximab induces antibody-dependent cellmediated cytotoxicity and immunogenic cell death. We assessed immune cell infiltration into liver metastatic sites of 53 CRC patients treated with chemotherapy and cetuximab, chemotherapy without cetuximab, and no chemotherapy. Of note, inflammatory cells were found in intratumoral areas, and the destruction of cancer cell foci was observed in the cetuximab group. Moreover, higher infiltration of CD8 + (P = .003) and CD56 + (P = .001) cells was observed in the cetuximab group. The immune-related mechanism of cetuximab may enhance the efficacy of combination therapy using immune checkpoint inhibitors and/or therapeutic peptides. 46 Although no objective responder to pembrolizumab is seen in pMMR/MSS-CRC, some patients obtained SD lasting more than 3 months according to tumor markers as well as radiographical evaluation, 55 indicating that some groups of MSS-CRC could respond to T A B L E 5 List of suppressive immunity and its resolution methods CAF, cancer-associated fibroblast; COX, cyclooxygenase; FAK, focal adhesion kinase; IDO, indoleamine-2,3-dioxygenase; Ig, immunoglobulin; IL-6, interleukin-6; LAG, lymphocyte activation gene; MDSC, myeloid-derived suppressor cell; PD-1, programmed cell death 1; PD-L1, programmed cell death ligand 1; PGE2, prostaglandin E2; TAM, tumor-associated macrophage; TGF, transforming growth factor; TIM-3, T-cell immunoglobulin and mucin domain-containing protein-3; Treg, regulatory T cell.

Suppressive immunity Therapeutic strategies References
PD-1 blockade. Galon et al 89 attested that MSS colon cancer is divided into tumors with or without massive CD8 + T-cell infiltration.
The reason why metastatic MSS-CRC did not respond well to immune checkpoint inhibitors might be divided into two patterns, owing to the highly immunosuppressive state of the "dark tumor" microenvironment ( Figure 1B) and the low immunogenicity in the "cold tumor" microenvironment ( Figure 1C). For dark tumors, a combination of immune checkpoint inhibitors and the modalities summarized in Table 5 to resolve the suppressive immunity might prove effective. For cold tumors ( Figure 1C), additional use of methods to induce immune cells to the tumor site would be needed, as described above. The combination of neoantigen-derived vaccination and cetuximab might be one of the most promising strategies.
Hence, combination immunotherapy should be selected as a precision medicine based on comprehensive analyses using wholeexome sequencing and RNA sequences. Moreover, novel immune checkpoints might not yet have been detected. Absolutely effective combination strategies might be just around the corner.