Locoregional Cellular Immunotherapy for Patients with Advanced Esophageal Cancer¹

Cancer in the esophagus is one of most common malignant neoplasms in the world, particularly in the Pacific countries. Surgery remains the standard approach for patients with locoregional advanced disease that is resectable. Curative resection is feasible in only 50% of cases, and local or distant failure is common after resection (1, 2, 3). The 5-year survival is only ∼30% for stage-III and -IV patients undergoing surgery. Some adjuvant multimodality therapies have been attempted to control both local and systemic disease (4, 5, 6). However, unresectable and relapsed esophageal cancers are still resistant to the presently available chemotherapy or radiation therapy regimens, and there is almost no clear advantage from these regimens for overall survival. Consequently,the development of a new effective therapeutic approach such as immunotherapy could be valuable to expand treatment modalities (7, 8, 9). Recently several reports presented the clinical efficacy of immunotherapy for advanced cancer in the digestive tract,but little clinical experience has been reported for advanced esophageal cancer (10, 11, 12). We have reported the presence of precursors of HLA class I-restricted and SCC³-specific CTLs in both PBMCs and TILs of patients with esophageal cancer (13, 14, 15). In the present study, we investigated the clinical toxicity and clinical response of locoregional administration of PBMCs stimulated with autologous tumor cells in advanced and recurrent esophageal cancer patients. Laboratory markers for the prediction of toxicity or response were also determined.

Patients with unresectable primary or recurrent metastatic esophageal cancer were eligible to this pilot study. Patients were required to have disease assessable by physical or radiographic examination and life expectancies of at least 2 months. Patients’ characteristics are shown in a Table 1. No patients had been receiving corticosteroids or any prior immunotherapy; however, six patients had received prior chemotherapy and radiotherapy. The intervals between these prior treatments and immunotherapy were at least 1 month (range, 1–6 months). The protocol was approved by the Institutional Review Committee of the Kurume University. The protocol was explained to each patient, and written informed consent to participate in the study was obtained from all of the patients who entered this study. Thirteen patients fulfilled the eligibility criteria, but two patients were eliminated before treatment because of inadequate growth of cultured PBMCs. Eleven patients with unresectable esophageal cancer (n = 4) or recurrent metastatic cancer (n = 7) received this treatment (Table 2). These tumors were histologically confirmed as SCCs by pathological examination. The mean age of the patients was 67.7 years. Six of the 11 patients (cases 4–8 and 10) received prior radiotherapy (48, 50, 52,50, 50, and 50 Gy, respectively; mean, 50 Gy) combined with two cycles of the following chemotherapy: 110 mg/m²/day of cisplatin and 700 mg/m²/5 days of 5-fluorouracil. The remaining five patients (cases 1–3, 9, and 11) did not receive any chemotherapy or radiotherapy before the immunotherapy.

Tumor samples used for the stimulation of PBMCs were obtained as follows: original esophageal tumors that were biopsied through an endoscope (n = 4), resected metastatic left supraclavicular LNs (n = 2), carcinomatous pleural effusion (n = 4), and carcinomatous ascitic fluid (n = 1; Table 2). The mean weight of the biopsied specimens was 0.29 ± 0.43 g, and each contained the mean of 0.5 to 2.0 × 10⁷ viable tumor cells. The tumor cells were irradiated with a dosage of 50 Gy over 10 min. Autologous tumor cells were prepared from biopsied samples by mincing and by enzymatic digestion with stirring in 50 ml of PBS containing 10 mg of type IV collagenase and 5 mg of DNase type I (Sigma) for 3–4 h at room temperature, followed by filtration through a layer of 100 nylon mesh and then by washing twice with PBS. The samples and effusion were applied on Ficoll-Hypaque solution and were centrifuged to isolate live cells from dead cells and aggregates. Tumor cells were discriminated from inflammatory mononuclear cells on the basis of the size and the other morphological features under the microscope, as reported previously (15). These cells were also cryopreserved in 90% human AB serum (Blood Center of Japanese Red Cross) plus 10% DMSO (Sigma) at −178°C in liquid nitrogen for restimulation and subsequent immunological assay. Heparinized peripheral blood samples (50–100 ml; mean, 78 ml) were collected from patients for 2 weeks to prepare cells for the treatment. The yield of PBMCs that were obtained from the blood samples with the Ficoll-Hypaque gradient method (14) was 0.38–1.3 ×10⁶ cells/ml (mean, 0.54) or 0.5–1.5 ×10⁶ cells (mean, 1.0) in patients or healthy volunteers, respectively. The tumor cell lines and normal cell lines used in these studies and their HLA class I alleles have been reported previously (13, 14).

One million PBMCs were incubated in 2 ml of the culture medium containing 10⁵ irradiated autologous tumor cells in each well of a 24-well culture plate at 37°C in 5%CO₂. The culture medium consisted of RPMI 1640 with 10% heat-inactivated human AB serum, 0.1 mm MEM nonessential amino acids solution, 100 IU/ml recombinant IL-2, 100 units/ml penicillin, 100 μg/ml streptomycin, 50 μg/ml gentamicin,and 0.5 μg/ml fungizone. PBMCs stimulated with autologous tumor cells were restimulated on day 7 of culture, followed by washing and preparation for clinical use on day 14. PBMCs were also cultured with IL-2 alone and were used as a control for the in vitroanalysis. Contamination in the cultured lymphocytes was checked by the Department of Laboratory Medicine according to the guideline for cellular therapy of our university. Endotoxin was not checked in this study, although the cells were repeatedly washed before injection.

The activated PBMCs were harvested after the second stimulation at day 14 of culture and were characterized for their phenotypes and cytotoxicity. Autologous tumor cells were separated from the single cell suspensions of biopsied tumor samples, pleural effusion, and ascites as mentioned above. These autologous tumor cells were cryopreserved in 90% human AB serum (Blood Center of Japanese Red Cross) plus 10% DMSO (Sigma) at −178°C in liquid nitrogen for restimulation and subsequent immunological assay. These cells were thawed, cultured for several days, and used as stimulator or target cells. A 6-h ⁵¹Cr-release assay was used to measure cytotoxicity of these activated PBMCs against autologous tumor cells or other tumor cell lines by the methods previously reported (14). The PBMCs were also measured for their IFN-γ production in response to various tumor cells by incubation of cells for 18 h with target cells at an E:T ratio of 3:1. The amounts of IFN-γ in cell-free supernatants were measured by an IFN-γ ELISA kit, and the limit of sensitivity of ELISA was 5 pg/ml as reported previously (14). The number of WBCs per mm³ was counted by the Department of Laboratory Medicine. Heparinized blood was applied for Ficoll-Hypaque solution and was centrifuged to obtain PBMCs, as reported previously (14). Viability of PBMCs was determined by a trypan blue dye-exclusion test. The surface phenotypes of PBMCs were tested by the two-color immunostaining technique with anti-CD3, -CD4, -CD8, and -CD16 monoclonal antibodies and FACScan flow cytometry at 4-week intervals, as reported previously (14). We used the term “CTL activity” in this study if the activated PBMCs produced significantly higher levels of IFN-γproduction or percentage of cytotoxicity (P <0.05 by a two-tailed Student’s t test) by recognition of the autologous tumor cells compared with those in response to the allogenic tumor cell lines.

The activated PBMCs were washed in PBS three times, resuspended in 5–10 ml of 0.9% saline, and administered by endoscopic intratumoral injection or direct regional injection (Table 2). Biopsy for preparation of the tumor cells was carried out before injection of the activated PBMCs in cases 1–3 and 8. These injections were repeated at least two times at 2-week intervals and for up to nine times until disease progression or severe toxicity was seen. The first clinical evaluation was performed 14 days after the second injection. The clinical outcome was evaluated by the following methods:esophagoscopy, esophagography and ultrasonography at 2-week intervals;CT scan at 4-week intervals; and serum CEA at 4-week intervals. A CR was evaluated as disappearance of all of the measurable tumor mass without the appearance of new lesions, and a PR was defined as a reduction of all of the measurable disease by 50% of the sum products of the two greatest perpendicular diameters. A SR was defined as <50%reduction in all of the measurable lesions. Adverse effects were evaluated by history and physical examination and graded according to the National Cancer Institute common toxicity scale.

Statistical analyses for lymphocyte phenotypic markers and IFN-γproduction were performed using the two-tailed Student’s ttest or the Fisher’s exact test.

Eleven patients with advanced (n = 4) or recurrent(n = 7) esophageal cancer received the locoregional cellular immunotherapy of the PBMC-activated autologous tumor cells and IL-2 every 2 weeks for 2–9 times (mean, 5.6 times). The mean number of administered cells per treatment was 0.8 ×10⁹ cells (range, 0.5–2 ×10⁹ cells). The patients’ profiles are summarized in a Table 1, and the treatment schedule, CTL activity of injected cells, adverse effects, and clinical response are summarized in Table 2. Six of the 11 patients developed a transient febrile reaction (<38°C) within 72 h after the treatment. One patient(case 1) developed grade III hypotension that required a Neo-Synephrine pressure drip for recovery, and the other patient (case 11) had transient grade II nausea and diarrhea. No adverse effect was observed in the remaining three patients (Table 2). No toxicity was associated with the intrapleural administration (cases 6, 7, and 10) of activated cells. Transient grade II nausea and diarrhea were observed by the intra-abdominal administration of the cells in case 11. The clinical outcome was as follows: 1 CR, 3 PR, 2 SR, and 5 PD. Clinical findings of the four cases that showed major tumor regression (1 CR and 3 PR)are shown in the next paragraphs.

One patient (case 1) had a CR with no evidence of disease for >20 months after the last treatment. Esophagography of the lower esophagus before treatment (Fig. 1,part 1a) and after the third treatment (Fig. 1, part 1b) showed a marked tumor regression for tumor size. Endoscopic appearance before treatment (Fig. 1, , part 2a), after the first treatment (Fig. 1, part 2, b and c), and after the third treatment (Fig. 1, part 2d) also clearly indicated a significant reduction in tumor size followed by the complete disappearance of tumor. Moderately differentiated SCC cells observed in the biopsied sample before treatment (Fig. 1, part 3a) were no longer observed in the biopsied sample after the third treatment (Fig. 1, part 3b). Instead, the marked infiltration of mononuclear cells was seen in the later sample. Similar histological changes were also observed in the samples after the treatment in the other responders (data not shown).

Three patients (cases 3, 5, and 9) had PRs. Esophagography before (Fig. 2,part 1a) and after treatment (Fig. 2, part 1b)in case 3 demonstrated a marked reduction in tumor size. Similarly,endoscopic appearance before (Fig. 2, part 2, a-1and a-2) and after treatment (Fig. 2, part 2, b-1 and b-2) also clearly showed a marked reduction in tumor size. Marked infiltration of CD3⁺CD8⁺ T cells was observed in the biopsied sample of case 3 after treatment (Fig. 2, part 3, a–c). Injection of the activated cells into metastatic tumors at the left supraclavicular LN resulted in complete disappearance after the second treatment (data not shown). Furthermore, the tumor size of liver metastases in the S2 (Fig. 3,a-1) and S4 (Fig. 3,a-2) areas of case 5 was markedly reduced after the fourth treatment (Fig. 3, b-1 and b-2). This patient received regional injections of the activated cells into metastatic LNs in the left supraclavicular area but not into the liver. A significant decrease in size of the metastatic tumor on the neck in the supraclavicular LN was observed in case 9 after the seventh treatment (Fig. 4,a and b). The duration of PR was 11, 9, and 7 months in cases 3, 5, and 9, respectively. No additional or subsequent treatment was given to these responders while they were in response.

In addition, the serum CEA levels largely decreased from 21.9 ng/ml before treatment to 9.8 ng/ml after the fifth treatment in case 9, who showed PR (Fig. 5). The levels also significantly decreased from 334 ng/ml before treatment to 170 ng/ml after the fourth treatment in case 11, who showed SR (Fig. 5). The decrease in the size of the para-aortoarterial LN after the fifth treatment of case 11 as measured by CT is shown in Fig. 6.

PBMCs stimulated with autologous tumor cells and IL-2 were measured for their activity to produce IFN-γ in response to autologous tumor cells and three allogeneic tumor cells (KE3, A2/A24; TE10, A24/A26; K562). PBMCs cultured with IL-2 alone that were not used for the treatment were also measured for their activity as the control. The PBMCs stimulated with autologous tumor cells and IL-2 produced a significantly higher level of IFN-γ in response to the autologous tumor cells than those in response to any of the other three target cells (case 1) or to those in response to K562 target cells (cases 3, 5, 6, and 9; Fig. 7,a). The stimulated PBMCs of case 3 (HLA-A24/A26),case 5 (HLA-A2/), and case 9 (HLA-A24/A11) also produced significantly higher levels of IFN-γ by recognition of allogeneic but HLA-A locus-matched tumor cells than those produced by recognition of K562 target cells. Four (cases 1, 3, 5, and 9) of these five patients responded to the treatment, whereas the remaining patient (case 6) had PD. There was no significant difference among the levels of IFN-γ production by the stimulated PBMCs in response to the four target cells tested in the remaining six patients who had either SR (n = 2) or PD (n = 4;data not shown). Similarly, there was no significant difference among the levels of IFN-γ production by the control PBMCs by recognition of the four target cells tested in any of the 11 patients. The results of PBMCs of the responders are shown in Fig. 7 a.

These results suggested that the CTL activity was observed in the stimulated PBMCs of 5 cases (cases 1, 3, 5, 6, and 9), whereas LAK cell activity was observed both in the stimulated PBMCs of the remaining 6 cases and in the control PBMCs cultured with IL-2 alone in all of the 11 cases. A 6-h ⁵¹Cr-release assay was used in cases 1 and 2, in which a relatively large number of cells were available for the study (Fig. 7 b). The PBMCs of case 1(HLA-A24/A26) showed significantly higher levels of cytotoxicity against HLA-A24⁺esophageal cancer cells than those against MKN28 (HLA-A31/)stomach cancer cells or K562 target cells. In contrast, there was no significant difference among the levels of cytotoxicity either by the stimulated PBMCs in case 2 or by the control PBMCs in cases 1 and 2.

Eleven patients were divided into the three groups as follows:responder group (n = 4, 1 CR + 3 PR; cases 1, 3, 5, and 9), SR group (n = 2; cases 8 and 11), and PD group(n = 5; cases 2, 4, 6, 7, and 10) to find laboratory markers useful for predicting the clinical response. Freshly isolated PBMCs (n = 57 from 11 cases) consisted of 71 ±4% CD3⁺ T cells, 44 ± 3%CD4⁺ T cells, 18 ± 5%CD8⁺ T cells, and 5 ± 2%CD16⁺ NK cells (Table 3). The PBMCs stimulated with autologous tumor cells and IL-2(n = 57 from 11 cases) consisted of 86 ± 5%CD3⁺ T cells (P = 0.0007 versus freshly isolated PBMCs), 28 ± 3%CD4⁺ T cells (P < 0.0001),42 ± 4% CD8⁺ T cells (P <0.0001), and 18 ± 4% CD16⁺ NK cells(P = 0.0031). The percentages of surface markers of the control PBMCs (n = 30 from 11 cases) cultured with IL-2 alone were similar to those of stimulated PBMCs, although the percentages of CD8⁺ T cells and of CD16⁺ NK cells in the stimulated PBMCs were slightly higher than those in the control PBMCs. In the responder group, the percentage of CD8⁺ T cells in the stimulated PBMCs (55 ± 20%) was significantly higher than that of freshly isolated PBMCs (22 ± 3%), whereas that in the stimulated PBMCs of the SR group or the PD group was not significantly different from that of fresh isolated PBMCs. The percentage of CD16⁺ NK cells in stimulated PBMCs of the responder group (17 ± 4%) and SR group (17 ± 5%)was significantly higher than that in the freshly isolated PBMCs of the responder group (7 ± 1%) and of the SR group (4 ± 2%),respectively, whereas that in the stimulated PBMCs of the PD group (15 ± 8%) was not significantly different from that of the freshly isolated PBMCs (6 ± 3%).

The mean number of administrated lymphocytes per case was 8.2 × 10⁹ cells in the responder group,4.2 × 10⁹ cells in the SR group, and 2.7 × 10⁹ cells in the PD group. The mean number of administrated lymphocytes per injection was 1.2 ×10⁹ cells in the responder group, 0.7 ×10⁹ cells in the SR group, and 0.7 ×10⁹ cells in the PD group. A kinetic study showed no significant difference in the mean number of WBCs, PBMCs, or CD3⁺, CD4⁺, or CD8⁺ cell counts that were measured before treatment and after the second and fourth treatments in these three groups (Fig. 8). In contrast, the mean number of CD16⁺lymphocytes in the peripheral blood after the second and fourth treatment in the responder group significantly increased compared with that before treatment in the responder group (Fig. 8). The number of CD16⁺ lymphocytes was also significantly higher than those after the second and fourth treatments in the PD group.

A number of clinical studies of adoptive immunotherapy using mostly LAK cells and TILs have shown a significant response rate in patients with melanoma, renal cell carcinoma, and some other cancers (16, 17, 18, 19, 20, 21). However, there has been no report on the clinical studies demonstrating the obvious tumor regression of advanced esophageal cancers, to the extent that we searched in the literature. This study showed that locoregional cellular immunotherapy resulted in marked tumor regression in 4 of 11 patients with advanced or recurrent esophageal cancer. Our report seems to be the first report showing clear evidence for the potential of the application of immunotherapy for esophageal cancer patients.

The adverse events in all of the 11 cases were moderate and tolerable. There was no relationship between the number of administered cells and adverse effects. Grade 1 fever was often observed and might have been attributable in part to production of cytokines by the administered lymphocytes (16). Other adverse effects included grade 3 hypotension in one case and grade 2 nausea and diarrhea in one other case. These adverse effects may also have been attributable to cytokine production from injected activated lymphocytes. These results suggest that this regimen of cellular immunotherapy was safe.

We provided the activated lymphocytes by stimulation of PBMCs with autologous tumor cells and IL-2 in vitro. These activated lymphocytes consisted of 42% CD8⁺ and 28%CD4⁺ cells. CTL activity was observed in these PBMCs from 5 (cases 1, 3, 5, 6, and 9) of the 11 cases, and 4 of them had significant tumor regression (1 CR and 3 PR), although the detailed studies of CTL activity, including the ⁵¹Cr-release assay against various target cells at different E:T ratios, were not carried out, mainly because of the limited number of cells for the in vitro analyses. LAK cell activity, instead of CTL activity, was observed in the PBMCs of the remaining six cases, and none of them showed major tumor regression (2 SR and 4 PD). LAK cell activity was observed in the control PBMCs cultured with IL-2 alone in all of the 11 patients tested. These results suggested that CTLs but not LAK cells were needed to achieve the tumor regression in this regimen of locoregional cellular immunotherapy. Consequently, the CTL activity of administered cells could be an appropriate laboratory marker to predict the clinical response. This phenomenon was supported by the fact that the mean percentage of CD8⁺ T cells (55 ± 20%) of the stimulated PBMCs of the responder group was highest among those of the different groups tested, and it was significantly higher than that of fresh PBMCs. The other laboratory marker for prediction of clinical response would be the percentage of CD16⁺ NK cells in the peripheral blood, which significantly increased in PBMCs of the responder group after treatment. These results suggest that both CTLs at the tumor sites and NK cells in peripheral blood were needed for tumor regression in this regimen. Therefore, both CTL activity in the injected cells and percentage of CD16⁺ cells in the peripheral blood might be useful laboratory markers for monitoring the clinical response to this locoregional cellular immunotherapy. These results, however, must be confirmed by a large-scale clinical study. The magnitude of CTL activity of the activated TILs was well correlated with clinical efficacy in metastatic melanoma patients in the regimens of adoptive cellular therapy with high-dose IL-2 (22). In contrast, no reliable laboratory markers for PBMCs, including the percentage of CD16⁺ cells, were found in the past decade,regardless of numerous trials in the field of immunotherapy using LAK cells, TILs, or cytokines.

It is of note that the injection of cells into the left supraclavicular LN in case 5 resulted in both the disappearance of tumors at the injection site and the decrease in tumor sizes at the uninjected site(liver). Some element of systemic immunity might been achieved by local therapy in this case, but this type of tumor regression was not observed in any other cases tested. It is also of note that regional treatment of pleural effusions and ascites was ineffective in 5 of the 11 patients. Subsequently, the application of this locoregional immunotherapy to the control of cancer cells in pleural or abdominal regions is not recommended. In addition, prior chemoradiotherapy is not recommended for this immunotherapy. Three of five patients who received no prior chemoradiotherapy responded to the immunotherapy, whereas only one of six patients who received prior chemoradiotherapy responded to the immunotherapy. This could be in part attributable to the number of cells available for the treatment. The mean number of administered cells per injection was 1.2 × 10⁹ cells in the responders, whereas it was 0.7 × 10⁹cells in the other groups. Prior chemoradiotherapy should suppress T-cell proliferation in response to IL-2 and autologous tumor cells.

In conclusion, we have shown that this treatment regimen was safe and resulted in tumor regression in 4 of 11 patients with advanced or recurrent esophageal cancer. A large-scale clinical trial is recommended to confirm the evidence from this Phase I clinical study.