Enhancement of cytotoxic T-lymphocyte responses in patients with gastrointestinal malignancies following vaccination with CEA peptide–pulsed dendritic cells

Abstract

Carcinoembryonic antigen (CEA) is strongly expressed in a vast majority of gastrointestinal carcinomas. Recently, epitope peptides of CEA were identified. We have demonstrated HLA-A24–restricted peptide, CEA652[9] (TYACFVSNL), was capable of eliciting specific cytotoxic T lymphocytes (CTLs) which could lyse tumor cells expressing HLA-A24 and CEA. HLA-A24 is the most applicable MHC class I allele in the Japanese population. In this pilot study, we have used the peptide-pulsed dendritic cells (DCs) generated from peripheral blood mononuclear cells (PBMCs) supplemented with GM-CSF and IL-4 as the source of the vaccine. Eight patients with advanced CEA-expressing gastrointestinal malignancies received subcutaneous injections every 2 or 3 weeks. Immunomonitoring was performed by ELISpot (enzyme-linked immunosorbent spot) assay to measure the precursor frequency of CTLs and their capacity to elicit antitumor CTLs in vitro. Four of seven patients have developed their CTL response after vaccinations. DTH reaction was observed in one of eight patients at the DC-injected site. Skin biopsy at the injected site showed the infiltration of the lymphocytes. Furthermore, A24/CEA peptide tetramer assay revealed an increase in peptide-specific T-cell precursor frequency in vaccinated patients. No significant toxic adverse effects were observed, except for mild diarrhea in one case after three vaccinations. Three patients have shown stabilization of the disease after vaccinations. In conclusion, our results clearly demonstrated that our vaccination protocol was safe and might develop a CEA-specific CTL response in cancer patients.

Introduction

It is known that specific tumor antigens must be targeted to develop effective immunotherapy. Carcinoembryonic antigen (CEA) is thought to be a good candidate, especially in solid epithelial tumors. CEA is a 180-kDa oncofetal antigen overexpressed in the vast majority of gastrointestinal malignancies, and including lung cancer, breast cancer, and thyroid cancer. CEA belongs to the immunoglobulin superfamily, and one of its functions was reported as acting as intracellular adhesion molecules, therefore it might play an important role in invasion and metastasis [1, 2]. The induction of anti-CEA cytotoxic T lymphocytes (CTLs) was previously thought improbable, since CEA was one of the self-antigens. Recent studies have clearly revealed that anti-CEA CTLs are inducible even in humans [3, 4]. Furthermore, the identification of several HLA-restricted epitope peptides clearly confirmed that anti-CEA CTLs truly exist [5, 6]. We also have identified HLA-A24–restricted CTL epitope (CEA652[9]) from CEA by primary in vitro immunization with peptide-pulsed DCs from healthy volunteers [7].

DCs are the potent antigen-presenting cells which can trigger CTL responses in vitro and in vivo [8, 9]. After DCs have been successfully generated in vitro [10], it becomes possible to utilize them for immunotherapy. The findings of clinical trials have been reported on melanoma [11, 12], gastrointestinal malignancies [13, 14], prostate cancer [15], renal cell carcinoma [16], and B-cell lymphoma [17]. Furthermore, the immunogenicity of DCs was clarified in the injection of DCs into healthy subjects [18].

We have performed a pilot study in which CEA652[9]-pulsed monocyte-derived DCs are injected into eight patients with advanced CEA-expressing gastrointestinal malignancies.

Immunomonitoring is of significant for immunotherapy, especially cancer vaccine. In this study, we measured whether the precursor frequency of anti-CEA652[9] CTLs increases after the vaccinations even in advanced cancer patients. ELISpot assay was performed for this purpose. Furthermore, it was also performed to establish which CTLs in immunized patients can lyse tumor cells in a CEA-specific and HLA-restricted fashion. Herein, we demonstrated the possibility of peptide-based immunotherapy in gastrointestinal cancer patients. Three of six patients increased their ability of CTL induction, and two of six patients responded in ELISpot assay. DTH reaction was seen in one of eight patients. These data suggest that CEA652[9]-specific CTL precursors might be augmented in vaccinated patients. Here we demonstrated that CEA peptide-based immunotherapy was safe and might be useful even in gastrointestinal cancer patients.

Material and methods

Generation of DCs

Peripheral blood from four patients was used, and from others leukapheresis was performed to obtain peripheral blood mononuclear cells (PBMCs). There were no phenotypic and characteristic differences among these DCs. After 7 days of culture, DCs were generated from each patient. Numbers of DCs were from 1.9% to 11.6% (average: 7.2%) of the starting number of PBMCs. Flow cytometric analysis of large cell population showed typical phenotype HLA-DR^high, CD80^low, CD83^moderate, CD86^high, and CD14^low (Table 1).

Table 1.

FACS analysis of the in vitro generated DCs (%). NT not tested

	HLA class I	HLA-DR	CD14	CD80	CD83	CD86
KT	99.7	95.5	2.0	2.7	5.5	93.0
TT	97.8	96.7	1.6	5.5	NT	94.5
SH	98.8	98.8	1.8	2.6	NT	95.5
KO	96.5	99.6	2.7	9.6	8.8	97.7
HK	99.7	97.5	0.5	7.0	10.2	96.6.
MS	98.8	946	1.8	5.5	14.5.	94.4
YK	97.5	99.2	0.8	7.6	7.7	98.8
UY	97.6	97.6	0.6	9.8	18.9.	97.7

Patients

Patients with gastrointestinal malignancies who were HLA-A24–positive and had positive serum CEA level (>5 μg/ml) or had a metastatic carcinoma expressing CEA as defined by immunohistochemical analysis were enrolled with written informed consent and with the approval of the ethical committee of Wakayama Medical University. Any previous therapy such as surgery, radiation therapy, and chemotherapy had been discontinued for more than 4 weeks. The enrolled patients consisted of seven males (code with age: KT, 37; TT, 54; SH, 65; KO, 66; HK, 60; YK, 65; UY, 66) and one female (MS, 42), and the average patient’s age was 57 (range 37 to 66). Seven patients had colorectal cancer, and one (UY) had advanced gall bladder cancer.

Synthetic peptides

Peptides were synthesized according to the standard solid phase method and purified by high-performance liquid chromatography (HPLC). The purity (>90%) and identity of peptides were determined by mass spectrometry analysis. This peptide was synthesized at Takara Shuzo, Japan.

Patient treatment

The cryopreserved PBMCs isolated by Ficoll-Paque (Pharmacia) were thawed and plated on a plastic tissue culture flask for 2 h to enrich the monocyte fraction. The adherent cells were then cultured in the presence of 1,000 U/ml of IL-4 (Ono Pharmaceutical) and 1,000 U/ml GM-CSF (Kirin) in RPMI 1640 (BioWhittaker) containing 5% autologous plasma. After 7 days, the DCs were harvested and pulsed with 50 μg/ml CEA652[9] (TYACFVSNL) in 2-ml saline for 2 h at 20°C. Before injection, DCs were washed in sterile PBS and resuspended in a total volume of 1 ml of the saline. Vaccinations were injected into the inguinal region subcutaneously every 2 or 3 weeks. DTH reactions were evaluated 2 days after the injections. Computed tomography scans and hematological examinations including tumor markers (CEA and CA19-9) were performed before and after vaccinations to determine the clinical response.

Flow cytometric analysis

The CEA expression of tumor cells was determined by monoclonal antibody (MAb) NCL-CEA-2 (Novocastra Lab., UK). The surface markers of DCs were measured following incubation with MAb: FITC-conjugated HLA-A, HLA-B, HLA-C, HLA-DR, CD14, and PE-conjugated CD80 (Becton Dickinson), CD83 (Immunotech), and CD86 (Pharmingen). CTLs were analyzed on a FACS Calibur (Becton Dickinson) using CellQuest software (Becton Dickinson).

Cell lines

The human B-lymphoblastoid cell lines TISI (HLA-A24/24) and EHM (HLA-A03/03), the gastric cancer cell lines MKN1 (HLA-A24/26) and MKN45 (HLA-A24/24), the colon cancer cell line WiDr (HLA-A1/24) were provided by Takara Shuzo (Shiga, Japan). The colon cancer cell line KM12LM (HLA-A2/24) which was defective in the class I molecule was kindly provided by Prof. Itoh (Kurume University, Japan). The colon cancer cell line HT29 (HLA-A24/24) was provided by Shionogi Pharmaceutical Company (Osaka, Japan). The HLA-A*2402 transfected class I–defective human T-leukemia cell line C1R/A24 was kindly provided from Prof. Takiguchi (Kumamoto University, Japan).

ELISpot assay

CTLs were stimulated with the peptides before the assay. The PBMCs were plated in 24-well plates (4×10⁶ cells/ml) in the presence of 10 μg/ml of peptide and 10 ng/ml of IL-7. IL-2 (10 U/ml) was added on day 2 and 5. On day 7, lymphocytes were restimulated by the peptide-pulsed autologous plastic adherent cells. On day 14, lymphocytes were tested for their ability to secrete IFN-γ. Nitrocellulose 96-well plates were coated with antihuman IFN-γ monoclonal antibody (Pharmingen) and incubated overnight at 4°C. Wells were washed with PBS containing 0.05% Tween 20 and incubated 2 h with the blocking buffer at room temperature. The peptide-pulsed TISI cells (10⁵) and nonpulsed TISI cells (10⁵) were plated to the wells. Graded doses of the lymphocytes (1×10⁵ diluted 1:3 per well) were added for a total volume of 200 μl/well. The plates were incubated overnight at 37°C with 5% CO₂, and after the plates were washed, the wells were incubated with biotinylated mouse antihuman IFN-γ monoclonal antibody (Pharmingen) for 2 h. Then the plates were washed extensively, and streptavidin alkaline phosphatase was added and incubated for 1 h. After washing the plates, 5-bromo-4-chloro-3-indolyl phosphate reagent was added and incubated overnight at room temperature. Blue spots indicating IFN-γ–secreting cells were counted through a microscope. Specific spots were determined by the equation: TISI⁺ spots − TISI⁻ spots. Increased reactivity after vaccination of the patient was considered positive if measurements were obtained at least two times after vaccination and both were over twenty spots more than those recorded before vaccination.

Induction of CTL responses

In an initial experiment of CTL induction, the thawed PBMCs were cultured in 24-well plates (4×10⁶ cells/ml; 2 ml/well) in RPMI 1640 containing HEPES with 2-mM l-glutamine, 1-mM sodium pyrborate, 0.1-mM nonessential amino acid solution, 5% heat-inactivated human AB serum (all reagents from BioWhittaker) in the presence of 10 μg of CEA652[9] peptide. On days 2 and 5, 20 U/ml of rIL-2 (Shionogi Pharmaceutical Company) was added. On day 7, the irradiated autologous (33 Gy) PBMCs were plated in 24-well plates (4×10⁶ cells/ml; 1 ml/well ) for 1.5 h at 37°C. After incubation, the nonadherent cells were washed out, and the adherent cells were incubated for 2 h with 20 μg/ml of the peptide and 3 μg/ml of β₂-microglobulin (Scripps, San Diego, CA) in a final volume of 0.5 ml/well. After the removal of the excess peptide from the adhered cells, responder lymphocytes were harvested and re-plated in 24-well plates (2×10⁶ cells/ml; 2 ml/well). The cultures were fed with 20 U/ml rIL-2 on days 9 and 12, and the cytotoxicity was tested on day 14. In the second experiment of CTL induction, PBMCs were stimulated by the methods that had previously been described [19]. The monocyte derived DCs in culture supplemented with GM-CSF (1,000 U/ml) and IL-4 (1,000 U/ml) were used as an APC to generate CTL responses. Briefly, after 7 days DCs were pulsed with 40 μg/ml peptide and 3 μg/ml of β₂-microglobulin for 4 h at 20°C. Then, the peptide-pulsed DCs were irradiated (55 Gy) and mixed on a 1:20 ratio with the positively selected autologous CD8⁺ T cells (Dynabeads M-450 and Detachabeads; Dynal). These cultures were set up in 48-well plates (Beckton Dickinson) by mixing 0.25 ml of DCs (1×10⁵ cells/ml) and 0.25 ml of CD8⁺ T cells (2×10⁶ cells/ml) in the presence of 10 ng/ml IL-7 (Genzyme). Each well was restimulated with the autologous peptide-pulsed monocyte, and IL-10 (Genzyme; final concentration is 10 ng/ml) was added on day 1 after the restimulation. After two and four rounds of restimulations, cytotoxicity was tested using the peptide-pulsed TISI cells as a target. The effector cells which had the cytotoxic activity for the peptide-pulsed target were expanded by the similar method as described before [20]. Briefly, CTLs were cultured with the irradiated (33 Gy) allo-PBMCs and irradiated (80 Gy) EHM in the presence of 30 ng/ml of anti-CD3 monoclonal antibody (Pharmingen). Cytotoxicity for various tumor cell lines was measured on day 14 after the CTL expansion.

Cytotoxicity assays

Cytotoxic responses of CTLs were determined in a standard 4-h ⁵¹Cr-release assay. Peptide-pulsed targets were prepared by incubating the cells with 10 μg/ml of peptide overnight at 37°C. Various tumor cell lines were treated with 100 U/ml of IFN-γ (Genzyme) for 48 h at 37°C before the assay to increase the level of MHC class I expression. Targets were labeled with 100 μCi of ⁵¹Cr sodium chromate (Dupont, Wilmington, DE) for 1 h at 37°C. Labeled target cells (10⁴) were incubated at 37°C for 4 h with varying numbers of CTLs in a final volume of 200 μl in 96-well round-bottom plates. To eliminate the nonspecific lysis by natural killer cells, unlabeled K562 cells (2×10⁵) were added to the wells. MHC restriction was determined by testing the inhibitions of the cytotoxicity by anti-HLA class I MAb, anti-class II MAb, anti-CD4 MAb, and anti-CD8 MAb (DAKO, Denmark). After 4 h of incubation, 100-μl of supernatant was collected from each well and the percentage of specific lysis was determined by the formula: [(experimental CPM − spontaneous CPM) / (maximum CPM − spontaneous CPM)]×100. A result of 10% cytotoxicity above the control was considered positive.

HLA-tetramer assay

HLA-A24 tetrameric complexes were produced as described previously [21]. Briefly, purified HLA heavy chain and β₂-microglobulin were synthesized. The heavy chain, β₂-microglobulin, and peptide were refolded by dilution and biotinylated by BirA in the presence of biotin, adenosine 5’-triphosphate. The 45-kDa refolded product was isolated by chromatography. Streptavidin-PE conjugate was added in a 1:4 molar ratio. HLA-A24/CEA-tetramers were used at 50 μg/ml. This was kindly provided by S. Asabe (Jichi Medical University, Japan).

Results

ELISpot assay in vaccinated patients

At first, we attempted to measure antigen-specific IFN-γ–releasing effector cells from the freshly isolated PBMCs without any in vitro stimulations. However, we did not detect any spots, since it might be difficult to detect adequate amounts of IFN-γ for unstimulated PBMCs (data not shown). Therefore, PBMCs were stimulated twice ex vivo as described in “Materials and methods.” This assay did not directly reflect the precursor frequency of CTLs, however semiquantitatively it was possible to analyze the precursor frequency after minimal stimulations. In two (SH, UY) of six patients tested, it was clearly demonstrated that the precursor frequency of CTL increased after vaccination (Fig. 1). In patient SH, the specific spots increased after only a single injection; furthermore, serum CEA was dramatically decreased from 2,790 μg/ml to 1,300 μg/ml (normal threshold is ≦5 μg/ml). And in patient UY, the specific spots were increased after seven vaccinations; furthermore, by using PBMCs in this period, it was possible to establish tumor-specific CTL lines.

Fig. 1.

Number of specific IFN-γ–secreting spots by peptide-stimulated cells before and after vaccinations. Effector cells were stimulated as described in “Materials and methods.” Results are shown as the number of specific spots per 10⁵ ex vivo stimulated cells. Open squares indicate prevaccination, solid squares, postvaccination

CTL response

After two in vitro stimulations, CTL response was analyzed from PBMCs after vaccination in two (TT and UY) of the four patients (Fig. 2a, b). CTLs lysed the peptide-pulsed targets but not the non-peptide-pulsed targets.

Fig. 2a,b.

Cytotoxicity after patients were vaccinated by initial method of CTL induction. a The CEA652[9]-specific bulk CTL culture from vaccinated patient (TT). CTL culture after two rounds of restimulations in vitro was used as effector to test the lysis of the following targets: C1R/A24-pulsed CEA652[9] (open squares); C1R/A24 without peptide (solid squares). b CTL culture from vaccinated patient (UA). TISI-pulsed CEA652[9] (open circles); TISI without peptide (solid circles). Target cells were pulsed with 10 μg/ml peptide overnight

It is important to analyze whether the established CTL lines lyse the tumor cells (which are CEA and HLA-A24–positive) as well as the peptide-pulsed target cells. To our knowledge, there are few reports that CTL induced from vaccinated patients lysed the tumor cells in an antigen-specific and HLA-restricted fashion [22, 23]. Therefore, CTLs were established by using peptide-pulsed DCs. This had been used successfully to generate CTLs from CEA, HER-2/neu, and gp100 [5, 6, 7, 19]. After four in vitro stimulations, the established CTL lines derived from patients before and after vaccination were tested to lyse peptide-pulsed TISI cells (HLA-A24⁺ human lymphoblastoid cell lines). In two patients (UY, KT), the CTL responses increased after vaccination (Fig. 3). Of these patients, one (UY) could develop the CTL response against not only the peptide-pulsed targets cells but also the tumor cells. Furthermore, cytotoxicity was blocked by anti-class I and anti-CD8 MAb, but not by class II and anti-CD4 MAb (Fig. 4). The cytotoxicity against HT29 was specifically blocked by the cold target cells that were pulsed with the CEA652[9] (data not shown). Thus, this CTL line lysed CEA-expressing tumor cells in an HLA-restricted manner.

Fig. 3.

Cytotoxic activity against tumor cell lines. Cytotoxicity was tested after vaccinations. CTL was established with four rounds of restimulations by primary in vitro immunization with peptide-pulsed DCs. CEA-specific CTL line was used as an effector to test the lysis of the targets cell lines: open squares indicate MKN45 (stomach, A24⁺, CEA⁺), open circles HT29 (colon, A24⁺, CEA⁺), open triangles WiDr (colon, A24⁺, CEA⁺), solid squares KM12LM (colon, A24⁻, CEA⁺), solid circles MKN1 (A24⁺, CEA⁻). Tumor cells were pretreated with IFN-γ for 48 h

Fig. 4.

Inhibition of specific cytotoxicity of anti-CEA CTL lines by MAbs. ⁵¹Cr-labeled HT29 tumor cell (HLA-A24⁺, CEA⁺) were preincubated with anti–HLA class I MAbs and anti-HLA class II MAbs, or the CTL line was preincubated with anti-CD4 and CD8 MAbs. CTL line and target were mixed at an E/T ratio of 1:50. Results are indicated as percentage of specific lysis

DTH response

One of eight patients showed DTH reaction against the peptide-loaded DC after three injections. Erythema and induration (30×30 mm) were repeatedly observed at the injected site 1 day later (Fig. 5a). Skin biopsy at injected site was evaluated by H&E staining, and demonstrated the massive infiltration of the lymphocytes (Fig. 5b).

Fig. 5a.

DTH reaction at the vaccinated site. Photograph shows the vaccinated inguinal site of the patient (MS) after 2 days of vaccination. Erythema and induration 30 mm in diameter were observed at the vaccinated site. b Histological analysis of the skin biopsy specimen. Massive infiltration of the lymphocytes was observed at the vaccinated site

Clinical development

No objective tumor regressions were observed during this trial. Serum CEA in one patient (SH) was dramatically decreased. Three patients (KO, TT, UY) showed no disease progression for 3, 4, and 5 months, respectively. In the other patients, the progression continued. There were no obvious adverse effects except in patient MS who showed mild watery diarrhea 1 day after the third injection. The summary of this trial is shown in Table 2.

Table 2.

Summary of this trial. NT not tested

Patient	No. of average OC doses (106)	No. of vaccinations	CTL responses	ELISpot responses	Time to progression (months)	DTH
KT	1.8	5	+	−	2	−
TT	3.3	8	+	−	5	−
SH	4.0	2	NT	+	1	−
KO	5.7	7	−	NT	5	−
HK	5.8	4.	NT	NT	1	−
MS	6.8	7	−	−	2	+
YK	12.1	5	−	−	2	−
UY	17.6	7	+	+	9	−

HLA tetramer assay

In one patient (UY), T-cell precursor frequency was monitored throughout treatment with vaccination. Strong enhancement (from 0.41% to 0.80%) was noted 2 weeks after the seventh vaccination (Fig. 6), at which time he showed no disease progression. At this point, tumor-cell specific CTL lines were established.

Fig. 6.

Tetramer analysis of patient UY. Left figure is the analysis of prevaccinated sample and right figure is the postvaccinated sample. Samples were double-stained with PE-labeled tetramer and FITC-labeled CD8

Discussion

In this pilot study, the aim was to investigate whether the epitope peptide from CEA was truly immunogenic in vivo. The precursor frequency for anti-CEA CTLs was considered low in circulation, and had been reported to be 1×10⁻⁵ to 1×10⁻⁶/PBMCs by limiting dilution analysis [4]. Therefore, it has been thought difficult to assess the precursor frequency of CTLs. In this study, we showed that CTL response was observed after the vaccinations in terms of the cytotoxic activity for peptide-pulsed targets, and the established CTL lines specifically recognized the various tumor cells dependent on the expression of CEA and HLA-A24. Furthermore, it was clear that the precursor frequency of specific CTLs increased after vaccination, as measured by ELISpot assay. Disease stabilizations were observed in three cases after immunization, and CTL responses were enhanced in two of these cases. In the peptide-based vaccination, several methods of immunomonitoring are utilized to evaluate the vaccine effect [24]. It is important to measure the CTL response before and after vaccination [22, 23, 25]. We have simply analyzed CTL responses as described in “Material and methods.” CTL responses against the peptide were detectable after the vaccination. Furthermore, to establish CTL lines, we utilized the DCs for in vitro stimulations. In patient UY, after seven injections, it was possible to establish CTL lines which recognized not only the peptide-pulsed targets but tumor cell lines in the tumor antigen–specific and HLA-restriction fashions. In the ELISpot assay, we attempted to measure the precursor frequency of CTLs to obtain the quantitative information. However, it was impossible to detect the specific increase after vaccinations from the circulating PBMCs. Therefore, in vitro stimulated PBMCs were used in this assay, and it became possible to detect the increased spots. This is consistent with previous reports [26]. HLA/peptide tetramer has been used in the analysis of peptide-specific T cells [21]. In the analysis of melanoma patients, the specific population could be detected by flow cytometry [27]. Though it was difficult to detect CEA652[9]-specific CTL precursor even in the immunized patients, we were able to detect CTL responses. This indicated that the precursor frequency of CEA652[9]-specific CTLs was low; however, it was clearly demonstrated that the vaccination might increase the specific CTLs within the detectable levels. Whereas it was controversial that the specific population was truly functional [27], we might demonstrate the increase of the functional CTL precursor in this report.

In recent vaccine therapy, clinical response has not necessarily been linked to the immunoresponse derived from PBMCs [22, 28, 29]. Rosenberg et al. showed that the clinical response was not related to the precursor frequency of CTLs [30]. We hypothesize that CTLs may immediately infiltrate to the tumor and, therefore, may not be found in the circulating peripheral blood. In fact, in one patient (MS) who showed DTH reaction and repeated mild diarrhea, CTL responses could not be detected in the PBMCs after the vaccinations. There might be induced CTL responses in vivo, and the CTLs might soon migrate into tumor or normal colon epithelium, and therefore might not be detected in PBMCs. Immunomonitoring in the tumor tissue will have to be targeted in future.

In this pilot study, antitumor response was not seen obviously even when CTLs were detected. One reason is the lack of HLA class I expression in the tumor cells. The lack of class I expression occurs in about 20% of invasive colon carcinomas [31]. The loss of HLA class I expression is a critical problem. Fas-ligand expression in tumor cells is also a critical problem [32]. It is thought to be important to induce strong CTL response, and it is also very important to overcome these problems. In peptide-based or DC-based vaccine therapy, there were no reports which showed these points [22, 28, 33].

To develop DC-based vaccine strategy, it may be effective to combine them with cytokines such as IL-2 or IL-12. Murine studies have demonstrated the efficacy of combination with IL-2 [34]. Another way of developing DC-based vaccine therapy may be the use of several peptides. We have already mapped the HLA-A24–restricted HER-2/neu [35] and p53 [36] epitope peptides. It may be effective to combine these epitope peptides if these antigens are overexpressed in the tumor.

In conclusion, our results have shown that peptide-pulsed DCs might induce measurable immunoresponses in vivo, and they provide some directions for specific immunotherapy.