A phase I clinical trial of RNF43 peptide-related immune cell therapy combined with low-dose cyclophosphamide in patients with advanced solid tumors

The objective of this study was to investigate the safety and the tolerability of combined cellular immunotherapy with low-dose cyclophosphamide (CPA) in patients with advanced solid tumors. This study targeted a novel tumor-associated antigen, ring finger protein 43 (RNF43). Eligible patients were resistant to standard therapy, HLA-A*24:02- or A*02:01-positive and exhibiting high RNF43 expression in their tumor cells. They were administered 300 mg/m2 CPA followed by autologous lymphocytes, preliminarily cultured with autologous RNF43 peptide-pulsed dendritic cells (DCs), RNF43 peptide-pulsed DCs and systemic low dose interleukin-2. The primary endpoint was safety whereas the secondary endpoint was immunological and clinical response to treatment. Ten patients, in total, were enrolled in this trial. Primarily, no adverse events greater than Grade 3 were observed. Six out of 10 patients showed stable disease (SD) on day 49, while 4 other patients showed progressive disease. In addition, one patient with SD exhibited a partial response after the second trial. The frequency of regulatory T cells (Tregs) in patients with SD significantly decreased after CPA administration. The ratio of interferon-γ-producing, tumor-reactive CD8+ T cells increased with time in patients with SD. We successfully showed that the combination of immune cell therapy and CPA was safe, might induce tumor-specific immune responses and clinical efficacy, and was accompanied by a decreased ratio of Tregs in patients with RNF43-positive advanced solid tumors.

Introduction a valid target for anti-tumor, immunotherapeutic treatment [18,19] and the combination with DC vaccination is expected to synergistically increase the antitumor immunity targeting RNF43.
Here we report a phase I clinical trial of combined immune cell therapy consisting of autologous RNF43 peptide-pulsed DCs and DC-activated killer lymphocytes (DAKs) with CPA and systemic low-dose IL-2 in patients with advanced solid tumors. This study was designated as a dose-escalation study of DAKs, targeting the novel HLA-A Ã 02:01-or HLA-A Ã 24:02-restricted TAAs of RNF43 [18]. The purpose of this trial was to investigate the safety and tolerability of the proposed method as primary endpoint in addition to efficacy, including overall survival and immune response, as the secondary endpoints.
The following exclusion criteria were applied: 1. Patients with severe pre-existing diseases; 2. Presence of autoimmune disease, active infectious disease, cardiovascular disorders, respiratory disorders, renal dysfunction, immunodeficiency, and hematological disorders; 3. Pregnant, lactating, or possibly pregnant women, or those willing to be pregnant, or willing male partner; 4. Presence of brain metastases; 5. Patients who required systemic administration of steroid or immunosuppressive agents; 6. Patients who were inappropriate for study entry, as judged by the attending physician.

Study design and treatment
This was a nonrandomized, open-label, phase I clinical trial, with dose escalation of the DAKs in patients with advanced solid tumors, to assess safety of the treatment. Subjects were enrolled into 2 cohorts (levels 1 and 2) in this dose-escalation study, wherein 5 patients each in levels 1 and 2 respectively received 5 × 10 7 and 2 × 10 8 DAKs (see Fig 1 for CONSORT diagram). The protocol for this trial and supporting TREND checklist are available as supporting information; see S1-S3 Files. Immunological responses, clinical responses, and overall survival were examined as secondary endpoints. Leukapheresis was performed 23 days prior to the administration of CPA. Low-dose CPA at 300 mg/m 2 was administered to patients on Day 1. DAKs were administered on Day 6 intravenously and 1 × 10 7 DCs were injected subcutaneously once a week, on Days 6, 13, and 20. Low-dose IL-2, at 3.5 × 10 5 IU, was injected subcutaneously on Days 6, 7, and 8, sequentially after each DC vaccination (Fig 2). All of these procedures were  done by physicians in charge. The evaluation was assessed on Days 1, 28, and 49 (Fig 2). Additional courses of the treatment were allowed after receiving approval from the case management committee in case the patient showed no severe adverse events and a clinical response higher than stable disease (SD

Preparation of DCs and DAKs
Leukapheresis was performed using a cell separator (COBE1 Spectra Apheresis System) (TERUMOBCT, Lakewood, CO). PBMCs were isolated from resulting peripheral blood with Ficoll-Hypaque plus (GE Healthcare, Buckinghamshire, England). PBMCs were suspended in X-VIVO™ 15 medium (Lonza, MD) and allowed to adhere on to plastic plates for 2 h. Nonadherent cells were collected and frozen at -80˚C in KM Banker II (Cosmo Bio, Tokyo, Japan) for later use as effector cells. The adherent cells were subsequently cultured for 5 days with 1,000 U/ml granulocyte macrophage colony-stimulating factor (GM-CSF; Bayer HealthCare, Leverkusen, Germany) and 1,000 U/ml CellGro1 Good Manufacturing Practice (GMP)grade IL-4 (CellGenix, Freiburg, Germany) in GMP serum-free DC medium (CellGenix, Freiburg, Germany). On Day 6, 50% of the DC medium was removed and fresh DC medium containing 50 ng/μl CellGro1 GMP tumor necrosis factor-α (TNF-α; CellGenix) and 25 μg/ml keyhole limpet hemocyanin (Merck Millipore Darmstadt, Germany) was added. On day 7, 0.1 KE/ml OK432 (Chugai Pharmaceutical, Osaka, Japan) was added to the culture medium to induce terminal maturation of the DCs. On day 8, harvested cells were pulsed with 20 μg/ml of HLA-A Ã 24:02-restricted RNF43 peptide (RNF43-A24-9-721, sequence: NSQPVWLCL) for HLA-A Ã 24:02-positive patients and 20 μg/ml of HLA-A Ã 02:01-restricted RNF43 peptide (RNF43-A02-10-11, sequence: ALWPWLLMAT) for HLA-A Ã 02:01-positive patients, for 2 h. The RNF43 peptides were synthesized by PolyPeptide Laboratories (San Diego, CA, USA). Mature, RNF43-pulsed DCs were obtained from each patient and prepared as described above, washed, and frozen until use for in vitro stimulation of non-adherent cells or injection into patients. A part of the DCs were analyzed for phenotype using flow cytometry. An adequate quantity of DCs were prepared so that >4.5 × 10 7 cells were available for each course of therapy. The remaining portion of the 5 × 10 6 RNF43-pulsed DCs were used to stimulate 1×10 8 non-adherent cells, which were then co-cultured with cytokines such as 100 IU/ml IL-2 (Novartis, Basel, Switzerland), 25 ng/ml IL-7 (R&D Systems, Minneapolis, USA), 400 pg/ml IL-12 (R&D Systems, Minneapolis, USA), and 25 ng/ml IL-15 (Thermo Fisher Scientific, MA, USA) every 2 to 3 days in X-VIVO™ 10 (Lonza, MD) to induce long-term survival of DAKs [20,21]. After three cycles of weekly stimulation, more than 2.0 × 10 8 DAKs were harvested and analyzed phenotypically by flow cytometry. All cells were prepared as per GMP grade at the Kyushu University Molecular and Cell Processing Center (KU-MCPC) and endotoxin levels as well as bio-burden of these products were confirmed as within acceptable levels of GMP-grade immune cell therapy.

Flow cytometry
Cells including patients' PBMCs, manufactured DCs, and DAKs were analyzed using the following mouse anti-human monoclonal antibodies: fluorescein 5-isothiocyanate (FITC)-

Detection of Tregs and Th17 cells in patient PBMCs
PBMCs (2.5 × 10 5 cells) collected from each patient at the four time points (days 1, 6, 28, and 49) were incubated with anti-CD4 (APC-H7) and anti-CD25 (PE) for 20 min at room temperature. Samples were permeabilized, and fixed with human Foxp3 buffer set (BD Biosciences, San Jose, CA, USA). The cells were then stained with anti-Foxp3 and anti-IL-17A for 40 min at room temperature. Activated Tregs were defined as CD4 + CD25 high Foxp3 + (S5 Fig).

Immunohistochemistry
To detect RNF43 protein and infiltrating lymphocytes in tumor tissues, formalin-fixed, paraffinembedded tissue was cut to obtain 3-μm sections. Antigen retrieval was carried out by boiling the slides with 10 mM sodium citrate (pH 6.0) or Target

Statistical analysis
All of the measurements were performed by specialists at the Kyushu University hospital who were unaware of the objective of this trial (S4 File). Based on the RECIST criteria, the patients were categorized into 2 groups showing different therapeutic effects at the end of treatment (Day 49), namely the treatment-effective group (complete response [CR], PR, and SD) and the treatment-ineffective group (PD). Changes in biomarker levels from the baseline were assessed with paired t-tests, unpaired t-tests for clinical response and logistic-regression analysis was performed to evaluate corrective factors for clinical responses. Because these statistical analyses were conducted for explorative purposes, the significance levels were not adjusted for multiple comparisons. All statistical analyses were performed using JMP 11 software (SAS Institute Inc., Cary, NC) by specialists.

Patient characteristics
To meet the goal of 10 patients completing this trial, 12 patients were recruited from January, 2006 to February, 2015. Ten patients completed the 10-week run-in period of the trial, comprising of 6 males and 4 females. The ages of patients were between 38 and 68 years old, with a median age of 57 years. Of the 10 enrolled patients, 7 had colorectal cancer and the other 3 had either small cell lung cancer, esophageal cancer, or cervical cancer. A total of 6 out of 10 patients achieved SD and 5 out of 10 patients lived more than 300 days with or without chemotherapy. Patient characteristics are presented in Table 1. Two of the twelve enrolled patients did not complete the trial and therefore were not evaluable by CT imaging and immunological analyses. One of the two discontinued because the one chose other treatments, and the other died from rapid disease progression.  Table 2A and 2B. These results showed that the central memory CD8 + T cells (CD45RA -CD62L + ) in DAKs tended to be expanded in patients with SD more so than in patients with PD, whereas the effector memory CD8 + T cells (CD45RA -CD62L -) in DAKs tended to be expanded in patients with PD   (Table 2B).

Feasibility and safety
Our results demonstrated the feasibility of producing and administering at least a total of 3× 10 7 DCs and 5 × 10 7 to 2 × 10 8 DAKs in adequate quantity for a 2-dose escalation study, from patients with advanced solid tumors, using clinically relevant and safe cell manufacturing procedures at KU-MCPC. Adverse events were evaluated according to the Common Terminology Criteria for Adverse Events v4.0 after the administration of CPA. The results are summarized for all patients in Table 3. No severe adverse events more than Grade 3 were observed. All patients showed Grade 1 or 2 dermal reaction at the injection site of DC or IL-2. No treatment-related hematologic, hepatic, renal, or neurological toxicities were observed. Clinical evaluation SD was observed in 6 out of the 10 patients on day 49 (Table 1). Of these, 2 patients (KU-4 and KU-5) experienced a decrease of serum tumor marker levels without any change of tumor size during the observation period (S3 and S4 Figs). In addition, 1 patient (KU-10) achieved a partial response against lung metastasis after the second course of DCs and DAKs, which lasted more than 1250 days (Fig 3 and Table 1). This patient was evaluated as showing SD on day 49, followed by an increase in a part of the tumor burden for a short period, but subsequently  evaluated as having achieved an objective response (Fig 3). Four other patients, however, showed PD at day 49 (Table 1).  (Fig 4A). There was no statistically significant change in the frequency of Th17 cell between before and after treatment, SD and PD.

Cytokine production by T cells after stimulation with the RNF43-peptide
The ICS assay has been reported as a potentially sensitive and functional assay that can generate quantitative phenotypic and polyfunctional data about responding T cell populations, Phase I clinical trial of RNF43 in immunotherapy making the assay valuable for functional profiling of low-level vaccine-induced T cell responses [23]. We performed ICS assays to detect tumor-reactive CD4 + or CD8 + T cells. It was found that the ratio of IFN-γ produced by tumor-reactive CD8 + T cells in patients with SD was significantly higher than those with PD between pre-CPA (day 1) and day 49 (p = 0.018; 95% CI, -3.922 to -0.580) (Fig 4B).

Correlation between the serum IL-6 level and clinical outcome
Among serum inflammatory cytokine levels examined, the serum IL-6 level was significantly increased in patients with PD at day 49 compared with that at day 1 (p = 0.020; 95% CI, -37.74 to -6.66; Fig 4C). In addition, the level of IL-6 was higher in patients with PD than those with SD at day 49 (p = 0.0018; 95% CI, 13.025 to 39.531; Fig 4C).

Immunohistochemical analysis of biopsied tumor specimens
Immunohistochemical analysis of RNF43 expression in tumor biopsies obtained before the trial demonstrated abundant expression of RNF43 in 8 of the 10 patients (Table 4 and Fig 5). The tumors in the remaining 2 patients (KU-1 and KU-6) demonstrated relatively lower RNF43 expression levels and were evaluated clinically as PD (Tables 1 and 4). Additionally, very low numbers of CD3 + , CD4 + , CD8 + , and Foxp3 + lymphocytes were noted (Table 4). In patient KU-10, the tumor was immunohistochemically evaluated before and after the trial. Tumor biopsy was collected on Day 230 and showed that the population of CD8 + lymphocytes in the tumor increased compared with that prior to initial treatment (Table 4 and Fig 5). The logistic model was applied to determine the ability of CD8/Foxp3 ratio as a biomarker to correlate the CD8/Foxp3 ratio with the efficacy of this treatment (PD/PD+SD). The odds ratio was estimated to be 0.667, with the 95% confidence interval between 0.335 and 1.080 (Fig 6).

DNA sequencing of RNF43 transcripts from tumors
Because of the paucity of patient's tumor samples for genomic sequencing in our clinical study, we performed RNF43 transcript analysis by RT-PCR of tumor mRNA focusing on two recurrent hotspot mutations of G659fs and R117fs, accounting for 41.7% to 48.0% and 8.3% to 12.0% of RNF43 frameshift mutations identified in colon cancer and endometrial cancer [26]. The cDNA sequencing results showed no frameshift mutations in all of ten samples (Table 4).

Discussion
We conducted a phase I clinical trial of cellular immunotherapy consisting of intravenously administered DAKs and subcutaneous vaccination of DCs followed by systemic administration of low-dose IL-2, combined with administration of low-dose CPA in patients with advanced, solid tumors. The results showed that the treatment was safe. Several previous studies suggested that the combination of DC vaccination with adoptive T cell transfer was superior to individual administration [27][28][29]. Regulatory T cells were thought to dampen T-cell Phase I clinical trial of RNF43 in immunotherapy immune responses to TAAs and comprised the main hurdle to successful immunotherapy and active vaccination [30]. CPA was reported to eliminate Tregs without adversely affecting vaccine-induced CD8 + T-cell function in mice [31]. Our results also showed that CPA reduced the frequency of peripheral blood Tregs in association with good clinical responses and antitumor immunity. A single, pre-conditioning dose of 300 mg/m 2 CPA prior to immune cell administration was suggested to enhance antitumor immunity induced by multiple vaccinations using five tumor-specific antigen peptides [15]. CPA has long been used as treatment for cancer, with proven safety and is cost-effectively eliminates Tregs. The molecular basis explaining the selectively targeting of Tregs by low-dose CPA is unclear. Previous data demonstrated that the intracellular ATP level were much lower in Tregs compared with those in conventional T cells or other cell types [32]. The low levels of ATP attenuated glutathione synthesis, leading to decreased CPA detoxification and increasing sensitivity of Tregs to low-dose CPA [32]. Additionally, the administration of low-dose IL-2 maintained or further increased the number of antigen-specific CD8 + T cells [33]. IL-2 also increased the clinical response of the administered DC vaccine [34] and was not detrimental to the functional activities of vaccineprimed CD8+ T cell effectors despite the expansion of Tregs, because IL-2-boosted Tregs fraction was functionally modulated to a Th1-like phenotype in the vaccinated patients [35].
In this dose-escalation study of DAKs, all patients administered with both dose levels tolerated treatment well, showing toxicities less than Grade 2. We considered that the higher dose Phase I clinical trial of RNF43 in immunotherapy would be superior to the lower dose in terms of efficacy. Patient KU-10, who received the higher dose of DAKs, was evaluated as showing SD on day 49, followed by a transient increase in partial tumor mass, but subsequently showed a successive decrease in tumor size, thus demonstrating long-term objective response. Combined with pathological findings at the tumor site, tumor-specific immune responses were considered to be induced by the administration of immune cells.
The simultaneous administration of DC and IL-2 may have boosted the function of a relatively lower number of DAKs in vivo [36,37]. Conversely, the robust proliferation of DAKs was associated with the subsequent decrease in CD8+ T cell survival, leading to a disruption of the critical, proliferative hierarchy that is necessary to maintain antitumor cell populations over a long term [38]. The low number of DAKs incubated with common γ-chain cytokines such as IL-2, IL-7 and IL-15 in this study resulted in a low number of exhausted T cells (CD45RA + CD62L -) (Table 2A) [38].
To investigate the immunological changes and predictive biomarkers required to select patients for this treatment, PBMCs and serum from 10 patients were examined before and after administration of the immune cells. First of all, RNF43 peptide-reactive CD8 + T-cell responses correlated well with good clinical responses. Secondly, an association was observed between peripheral serum IL-6 level as well as frequency of Tregs with clinical response. Logistic regression analysis was performed, between pretreatment immunological parameters and clinical responses to identify the predictive factor. Although no significant results were observed in blood samples due to the small number of cases analyzed in this study, the odds ratio for clinically resistant patients with higher IL-6 ratios was 0.772 (0.54-1.11, 95% confidence interval, p = 0.157). Previous studies also suggested that serum IL-6 was a useful predictor for cancer immunotherapy [39,40]. IL-6 was considered to stimulate inflammatory cytokine production, tumor angiogenesis, and the tumor macrophage infiltrate as well as inhibit the differentiation of localized T cells to effector cells [40]. ATP also acts as the second messenger during inflammation and stimulates IL-6 synthesis. Extracellular ATP was able to activate Tregs and increase their suppressive capacity [41]. Thus, the patients with PD who showed higher serum levels of IL-6 without an increase in Th17 cells might be expected to augment serum ATP concentrations as a warning sign, followed by increased intracellular ATP levels in Tregs and decreased sensitivity of Tregs to low-dose CPA [35], resulting in tumor progression.
Moreover, the immunohistological CD8/Foxp3 ratio and RNF43 expression prior to the administration of immune cells might be a useful predictor of the clinical response, and our results suggested that an adequate balance between CD8 + T cells and Tregs is necessary to obtain strong antitumor effects.
Recent reports showed that RNF43 encoded an E3 ubiquitin ligase that negatively regulated Wnt signaling and played a crucial role in the development of various cancers [42][43][44]. Somatic mutations of RNF43 in 14% to 19% of colorectal, pancreatic, and endometrial cancers were recently reported [26,44]. These showed frameshift mutations encoding G659fs and R117fs, constituting insertions or deletions of 1 bp in homopolymeric tracts of the RNF43 gene. RNF43 was mutated in approximately 80% of microsatellite instability colorectal tumors and 4.8% of microsatellite stable (MSS) tumors [26]. Although RNF43-based immune therapies suggested its clinical potential [45,46], the precise role of RNF43 mutation in tumor cells in immunotherapy remains to be clarified. We targeted tumor cells expressing relatively high levels of RNF43 without major gene mutations, possibly included in MSS tumors. The function of ZNRF3/RNF43 is counteracted by R-spondin; R-spondin binds to LGR4/5 and ZNRF3/RNF43 and induces ubiquitination and degradation of ZNRF3/RNF43. To achieve high and sustained Wnt/β-catenin signaling, cancer cells need to overcome this strong negative feedback control, which can be achieved through mutations of ZNRF3/RNF43 or translocations/overexpression of R-spondin [47]. As we could detect no major ZNRF3/RNF43 gene mutations in our patients' tumor, the latter might be functional in our cases. Although the significance of higher RNF43 expression in our cases still unknown, tumor specific expression of RNF43 was considered to be a reliable target for immune therapy. Further examination is required to prove this hypothesis.
A deeper understanding of immune responses in the tumor microenvironment is required for recently proposed combination therapies to provide a survival benefit towards a greater number of cancer patients [48,49]. Namely, non-immunogenic tumor microenvironments that are resistant to immune-checkpoint therapy may require combination therapies to create an immunogenic tumor microenvironment [48,49]. Our new combination treatment was feasible in patients with advanced solid tumors, especially MSS tumors that were refractory to standard therapy, and may have the potential to induce an immunogenic tumor microenvironment. A clinical study of combined immune cell therapy with immune-checkpoint inhibitors for patients with refractory colorectal cancer is being planned [49].
In conclusion, we demonstrated that RNF43 peptide-related immune cell therapy combined with low-dose cyclophosphamide and IL-2 was safe, and that the reduction of peripheral blood Tregs and the immune response correlated well, resulting in good clinical response in patients with advanced solid tumors. Further immunological studies and a role of RNF43 mutation and translocations/overexpression of R-spondin in the progression of colon cancer is planned, to explore valid biomarkers that predict clinical efficacy of immune therapy.