Dendritic cells can be rapidly expanded ex vivo and safely administered in patients with metastatic breast cancer

Abstract

Purpose

Immunotherapy using either dendritic cells (DCs) or expanded cytotoxic T cells (CTLs) has received increased interest in the treatment of specific malignancies including metastatic breast cancer (MBC). DCs can be generated ex vivo from monocytes or CD34⁺ precursors. The ability to expand and safely administer CD34-derived DCs in patients with MBC that have received prior cytotoxic chemotherapy has not been evaluated.

Methods

We enrolled ten patients with MBC that had received prior chemotherapy for the treatment of metastatic disease on a phase I/II trial designed to test the safety and feasibility of administering ex vivo expanded DCs from CD34⁺ progenitor cells.

Results

Using a cocktail of multiple different cytokines, we could expand DCs 19-fold compared to the initial CD34-selected product, which allowed the administration of as many as six vaccine treatments per patient. Patients received three to six injections i.v. of DCs pulsed with either the wild type GP2 epitope from the HER-2/neu protein or an altered peptide ligand, isoleucine to leucine (I2L). Toxicity was mild, with no patients demonstrating grade III toxicity during the treatment. Two patients with subcutaneous disease had a partial response to therapy, while IFN-γ-producing CD8⁺ T cells could be found in two other patients during treatment.

Conclusions

This approach is safe and effective in generating a significant quantity of DCs from CD34-precursors.

Introduction

The outcome for patients with metastatic breast cancer (MBC) is quite poor, with few patients alive 10 years after the occurrence of metastatic disease [1–3]. The use of high-dose chemotherapy and stem cell transplantation has not resulted in an improved survival for patients with MBC [4]. A combination of the biological response modifier trastuzumab with chemotherapy has modestly improved the outcome of patients with MBC [5]. However, for patients with MBC to achieve a better outcome, newer approaches to therapy need to be explored. Immunotherapy is one potential treatment option for this population of patients.

Dendritic cells (DCs) are antigen-presenting cells that are capable of stimulating antigen-specific naïve and memory T cells. Human DCs can be differentiated based on their origin, surface phenotype and function into myeloid and plasmacytoid DCs [6, 7]. Within the subset of myeloid DCs are a significant number of phenotypically distinct DCs such as Langerhans cells. DCs have been used by multiple different investigators to generate tumor-specific T cell responses [8–11]. In most of these trials, DCs were generated ex vivo from CD14⁺ monocytes. DCs can also be generated from CD34⁺ precursors. Previous work from our laboratory and others has suggested that CD34-derived DCs are more effective in stimulating peptide-specific T cells than are monocyte-derived DCs [12] perhaps due to the greater number of Langerhan’s-like cells in this product. However, CD34-derived DCs may be difficult to expand from patients treated with marrow suppressive cytotoxic chemotherapy.

CD8⁺ cytotoxic T cells (CTLs) can be induced against breast cancer cells that overexpress self-proteins such as HER-2/neu [13–16]. The HER-2/neu protein contains several peptides that have been shown to bind to HLA-A0201 [17, 18]. However, peptides that bind HLA-A0201 from HER-2/neu often possess amino acids that lead to suboptimal binding of the peptide to the class I MHC complex [19]. Our group has focused on the interaction of the peptide GP2 (IISAVVGIL), from HER-2/neu, with HLA-A0201. Previously, we found that altering position two in this peptide, from isoleucine to leucine (I2L), increased the stability of the peptide/MHC (p/MHC) complex and increased its immunogenicity when pulsed onto DCs in an animal model [20]. The half-life of the p/MHC complex increased from 24 min to 108 min by substituting a leucine for an isoleucine in position two.

Here, we investigated the feasibility of administering multiple cycles of peptide-pulsed DCs from CD34⁺ progenitor cells to patients with MBC using either the GP2 or I2L epitope.

Methods

Eligibility and apheresis

Patients with MBC with stable disease off chemotherapy for 30 days who expressed HLA-A0201 by single-stranded oligonucleotide primer DNA analysis were eligible for this study. Patients with central nervous system metastases were excluded. Hormonal therapy and bisphosphonate treatment for bony metastases during vaccine treatment were permitted. Trastuzumab therapy was not permitted during vaccine treatment. All patients gave written informed consent according to federal and institutional guidelines before treatment.

For mobilization, patients were treated for 4 days with G-CSF (Amgen, Thousand Oaks, CA, USA) at 10 μg/kg/day subcutaneously, followed by a 15L apheresis collection using a Cobe-Spectra machine (GambroBCT, Lakewood, CO, USA). CD34⁺-selection was performed using a CliniMacs (Miltenyi Biotec, Auburn, CA, USA) device. DCs were expanded from CD34⁺-precursors in a 5%-CO₂ humidified incubator at 37°C in the presence of GM-CSF (Immunex, Seattle, WA, USA), TGF-β1, FLT3-ligand, IL-4, stem cell factor, and TNF-α (remaining cytokines from Peprotech, Rocky Hill, NJ, USA) using AIMV media (Gibco, Grand Island, NY, USA) with 10% human AB+ serum (BioWhittaker, Walkersville, MD, USA).

Peptides

Peptides used for the clinical trial GP2 (IISAVVGIL) and I2L (ILSAVVGIL) were synthesized by Multiple Peptide Systems (Multiple Peptide Systems, San Diego, CA, USA). The FMP peptide (GILGFVFTL) was synthesized by New England Peptide (Fitchburg, MA, USA).

Generation of DCs

DCs were generated in the human applications laboratory in the General Clinical Research Center of the University of North Carolina Hospitals using Good Manufacturing Practice guidelines. Expanded DCs were pulsed with wild-type GP2 or I2L peptide ligand [20–22] overnight at 37°C, washed, resuspended, and checked for viability and contamination with mycoplasma, bacteria, endotoxin, and fungus using commercially available assays, or standard gram stain, KOH stain and culture. All DC products had to be at least 80% viable by trypan blue staining, negative for contamination by endotoxin testing, as well as negative for the presence of bacteria and fungi using conventional staining. DCs had to express greater than 2 logs compared to isotype controls of class II MHC and CD80 or CD86 prior to infusion. DC samples from two patients were analyzed for the expression of the chemokine receptor, CCR7, prior to infusion by RT-PCR using the following primers: (sense), 5′-CGTGCTGGTGGTGGCTCTCCT-3′; 5′-ATCGGTCATGGTCTTGAGCCTCTTG-3′(antisense). DC activity was measured in a standard mixed lymphocyte assay using responder cells from five different individuals. ELISA assays were performed according to the manufacturer’s instructions (R and D Systems, Minneapolis, MN, USA).

Infusions

Allocation of patients to treatment with either the GP2 or I2L epitope was performed by the Cancer Center statistician using a random 2×2 block design. The clinicians monitoring responses to therapy were blinded to the treatment given to each patient. The peptide-pulsed autologous DCs were infused i.v. over 4–8 min every 21 days for a maximum of six infusions in the outpatient treatment area of the General Clinical Research Center of the University of North Carolina Hospital. Toxicity was monitored weekly and response was monitored every two cycles. Patients who did not develop dose-limiting toxicity were eligible for second and subsequent infusions. Toxicity was graded according to the NCI Common Toxicity Criteria Version 2.0. Dose-limiting toxicity was defined as grade 3 neurologic, pulmonary, cardiac, hepatic, or renal toxicity or any grade 4 toxicity that occurred within 7 days of the infusion and was clearly unrelated to progressive disease. Response was measured by the WHO criteria.

Isolation and cryopreservation of peripheral blood mononuclear cells

Peripheral blood mononuclear cells (PBMCs) were isolated from anticoagulated blood from each patient 7 days after each infusion and 56–90 days after the last infusion. PBMCs were separated from erythrocytes and platelets by Ficoll/Paque (Amersham Pharmacia Biotech AB, Uppsala, Sweden) density gradient centrifugation. PBMCs were cryopreserved at 10–20×10⁶/ml in 90% fetal bovine serum (FBS) and 10% DMSO (Sigma Chemical Co., St. Louis, MO, USA) in a liquid nitrogen storage tank.

In vitro stimulation of lymphocytes

CD8⁺ T cells were isolated as described previously. FMP-specific T cells were expanded for 7–10 days by addition of the FMP peptide (10 μg/ml) to PBMCs, which were cultured in IL-2 and IL-7. All culture conditions were at 37°C, 5% CO₂. Expanded T cells were incubated with FMP tetramer (Beckman-Coulter, Miami, FL, USA) for 1 h followed by addition of 6 μl of anti-CD8-PerCP (BD Pharmingen, San Diego, CA, USA) at 4°C for 20 min. Cells were resuspended in 200 μl FACS wash and run immediately on a BD FACScan. Data were analyzed using FloJo software (TreeStar, Stanford, CA, USA).

Intracellular cytokine staining

1–2×10⁶ thawed PBMCs were stimulated in the presence of anti-CD28/49D mAb (BD Immunocytometry Systems, San Jose, CA, USA) with either peptide-pulsed PBMCs, 2 μg/ml PHA (Remel, Richmond, VA, USA), or without antigen for 6 h. GP2 or an irrelevant MART-1 peptide were used at 50 μg/ml. The intracellular cytokine assay was performed according to the manufacturer’s instructions. A total of 40,000 CD8⁺-PerCP gated events were collected. Data were analyzed using CellQuest data analysis software (BD Immunocytometry Systems). ELISPOT assays measuring the production of IFN-γ were performed according to the manufacturer’s instructions (Becton-Dickinson, Franklin Lakes, NJ, USA).

Results

Generation of DCs, patient characteristics and clinical outcome

DCs were generated from CD34⁺ precursors as described in Methods. CD34⁺ precursors were expanded ex vivo for 13–18 days, which resulted in a mean 19-fold increase in the number of DCs at these timepoints compared to the initial CD34⁺ progenitor cells (Fig. 1a,b). Expanded DCs expressed characteristic markers with high levels of HLA-DR, CD80, CD86 and 25–50% of cells expressing CD83 and CD1a (data not shown). In unique patient number (UPN) 3, the apheresis product was held overnight prior to selection. This patient had a modest threefold expansion ex vivo of DCs compared to the initial CD34 product, and because of this result, CD34 selection was performed on the day of apheresis in the remaining patients (Fig. 1b). DC expansion was sufficient to allow the treatment of all patients that did not have tumor progression on the trial with six cycles of therapy.

Fig. 1a, b.

Expansion of DCs from CD34 progenitor cells. DCs were expanded from CD34 progenitor cells as indicated in Methods. Immediately prior to infusion, the number of viable DCs from an aliquot of DCs was compared to the number of progenitor cells used to initiate the culture by flow cytometry. In (a) the increased number of DCs from CD34⁺ precursors is shown for each patient over the entire time of ex vivo expansion. In (b) the data are presented as the mean fold expansion for each patient on the trial. UPN 3 had progenitor cells incubated overnight in media prior to selection the following day. In the remaining patients, CD34⁺ cells were selected on the day of apheresis

Prior to infusion, DCs in two patients were analyzed for the expression of CCR7, a chemokine receptor found on mature DCs and necessary for migration of these cells into secondary lymphoid organs, by RT-PCR. There was a fivefold increased expression of CCR7 in TNF-α matured compared to immature DCs without TNF-α (data not shown). Additionally, we found that the expanded DCs from two of the patients were functional and elicited both a potent MLR response (Fig. 2a) with a stimulation index of 10 at a ratio of T cells to DCs of 10:1 and significant production of IFN-γ (Fig. 2b).

Fig. 2a, b.

Functional analysis of expanded DCs. CD34-derived DCs from two different patients were incubated with allogeneic T cells from five different donors and analyzed by a a standard MLR assay and b by cytokine production using an ELISA assay. Representative data for UPN 6 are shown. CD34-derived DCs functioned similarly to monocyte-derived DCs in this assay (data not shown). For the MLR assay, DCs were incubated with allogeneic T cells at the ratio indicated in the figure in triplicate. Control samples included DCs and T cells analyzed alone. For the ELISA, supernatants were collected from the MLR and analyzed in triplicate. The data are shown as mean with standard error. Cytokine production by DCs and T cells alone was below the limit of detection (3.5 pg/ml) for the assay

Ten patients with MBC were treated with 40 courses using autologous CD34-derived DCs pulsed either with the GP2 epitope or I2L in doses ranging from 3×10⁶ to 3×10⁷ DCs. Patient characteristics are shown in Tables 1 and 2. All patients had prior chemotherapy and nine of ten had received chemotherapy for the treatment of MBC. Eight had prior hormonal therapy. As we have found in animals that tumor cells need to express, but not overexpress, HER-2/neu (Serody, unpublished data) to allow for a response to the vaccine, we enrolled six patients whose tumors did not overexpress HER-2/neu. Six patients (UPNs 1–6) were treated at the first dose level of 3×10⁶ DCs and four (UPNs 7–10) at the second dose level of 3×10⁷ DCs, with each patient receiving either the wild-type or APL for all vaccinations.

Table 1.

Demographic information for patients treated

Patient characteristics	No. of patients
Performance status
0	8
1	2
Prior chemotherapy regimens for metastatic disease (median 1, range 0–4)	9
Prior radiotherapy	7
Prior hormonal therapy	8
Prior BMT	0
Concomitant hormonal therapy	6

N=10. The median age is 48 years, the range 34–63 years.

Table 2.

Characterization of previous disease status and treatment for patients on the clinical trial

Patient	Age (years)	Disease site	Prior Rx	Therapy within 60 days of vaccine	Ongoing hormonal Rx	HER-2, status	Peptide	Number of cycles	Clinical response	Immune response
1	63	Soft tissue, bones	C	H	no	non-overexpression	WT	5	PR after cycle 2	IC:a
2	53	Soft tissue	C+H	H	yes	non-overexpression	APL	6	PR after cycle 3	IC:a
3	60	Pleural effusion	H	H	yes	non-overexpression	APL	3	PD	IC:a
4	46	Liver, bones	C	C	no	overexpression	WT	3	PD	IC:a
5	43	Ovary	C+H	H	yes	overexpression	WT	6	Not evaluable	IC:b
6	41	Lungs, liver	C	C	no	overexpression	APL	3	PD	IC:a
7	33	Soft tissue, lungs	C	C	no	non-overexpression	WT	4	PD	IC:a
8	52	Lungs, bones	C+H	H	yes	overexpression	APL	2	PD	IC:a
9	46	Liver, bones	C+H	H	yes	non-overexpression	WT	2	PD	IC:b
10	50	Bones	C+H	H	yes	non-overexpression	APL	6	SD	IC:a

C chemotherapy, H hormonal therapy, WT wild type, APL altered peptide ligand, PD progressive disease, SD stable disease, PR partial response

IC:^aNo detectable IFN-γ-producing CD8+ T cells above baseline, ^bfive- to eightfold increase

The DC vaccine treatment was well tolerated. No patients developed grade 3 or 4 toxicity during therapy. Four patients had grade 1–2 fatigue, four had grade 1 anemia, two had grade 1 lymphopenia, two had grade 1 hepatic enzyme transaminitis. One patient had transient grade 1 chest tightness following infusions and one patient a persistent cough. No infusional toxicity was found and only one patient (UPN 3) with a history of nausea and vomiting and pancreatitis was removed from therapy due to an increase in symptoms after treatment from tumor progression.

Two patients had modest decreases in metastatic lesions and one patient decreased bone pain after vaccine therapy. UPN 1 had a 50% reduction in the size of both subcutaneous lesions (initially 1.5×1 cm) after the second vaccine treatment that persisted for 2 months. She received three treatments and then a break in therapy of 4 months. Subsequent vaccine treatments were not effective in controlling new tumor growth. UPN 2 had a 70% reduction in the size of the largest subcutaneous lesion (initially 11×2 cm) with the remaining lesions either reduced in size or stable after the third vaccine treatment. Her response persisted for 8 months. UPN 10 had stable bone metastases, when evaluated by bone scan and MR scan of the affected areas, and an improvement in range of motion and decreased pain during vaccination. Interestingly, all of the patients with a clinical response had tumors that did not overexpress HER-2/neu, corroborating our in vivo animal data. UPNs 3, 4, and 6–9 had progression of disease either during or immediately after cessation of therapy. One patient was not evaluable because all metastatic tumor had been removed prior to vaccine therapy (UPN 5).

Immune response to therapy

All patients had PBMCs isolated 1 week after each vaccination and at the end of therapy. T cell responses after vaccination were analyzed using intracellular cytokine (IC) (Fig. 3) and ELISPOT assays.

Fig. 3a, b.

Intracellular cytokine staining for the presence of IFN-γ-producing T cells. Dot plots are shown for the response to the GP2 epitope from the two patients that generated immune responses serially post vaccination. The cells circled at upper right for each dot plot represent the GP2-specific CD8+ T cells generating IFN-γ. Production of IFN-γ was confirmed in UPN 5 using ELISPOT (mean number of spots GP2 peptide 185±20 vs. control 7±2). MART-1 pulsed DCs were used as control stimulators in (a); empty DCs were used for control stimulators in (b).

In eight of the patients treated, we did not find antigen-specific CD8⁺ T cell responses during or after vaccination. One of the patients, UPN 5, generated IFN-γ producing CD8+ T cells in response to the GP2 peptide post vaccination (Fig. 3). The peak response was found after the sixth vaccination and represented a sevenfold increase compared to the prevaccine values. Interestingly, this was the only patient on the trial who did not have active metastatic disease at the time of vaccination. A second patient, UPN 9 who also received the wild-type epitope GP2, had a sevenfold increase in IFN-γ-producing T cells at the end of vaccination (Fig. 3). We confirmed the presence of IFN-γ-producing T cells in UPN 5 by ELISPOT assays that demonstrated 185 spots using the GP2 epitope compared to seven spots in control samples.

To determine if we could generate T cells to a non-self antigen from vaccinated patients that did not generate a GP2-specific immune response, we evaluated the immune response to the FMP epitope from influenza. PBMCs were isolated pre vaccine therapy and after vaccination from UPN 9, an immune responder, and compared to UPN 1 (clinical response), 8, 10 (no response) and a healthy HLA-A0201-expressing individual. While the number of FMP-specific T cells isolated ex vivo and after 1 week of stimulation was much greater in the healthy control, we were able to expand FMP-specific T cells from all of the patients (Fig. 4). Thus, we were unable to document a global problem with immune function as a reason for the inability to expand T cells post vaccination.

Fig. 4.

FMP-specific tetramer positive T cells from UPN 6 and 8 compared to responding vaccinated patients and a healthy control. T cells were analyzed for the presence of FMP-specific tetramer-positive T cells either ex vivo or after 1 week of in vitro stimulation using FMP-pulsed PBMCs. The number of tetramer-positive T cells is shown. IC cytokine assays (not shown) confirmed functional activity of these cells

Discussion

We have evaluated the safety and efficacy of administering autologous DCs pulsed with the wild-type epitope, GP2, or an APL, I2L, to women with MBC that expressed HLA-A0201. DCs could be easily expanded from patients with MBC that had received multiple cycles of chemotherapy previously. Peptide-pulsed DCs were safe in this patient population as none of the patients treated had greater than grade II toxicity. We found no difference in toxicity using wild type or APL-pulsed DCs. Similarly there was no difference in toxicity based on the number of peptide-pulsed DCs infused. Two of the patients treated on this trial generated an immune response post vaccination. Interestingly, two other patients had a modest decrease in the size of subcutaneous metastatic lesions, suggesting that this approach may be beneficial in the treatment of small non-visceral tumors.

Our interest in deriving DCs from CD34-precursors is based on several different concerns. First, as stated previously, there is data from our group and others that CD34-derived DCs are better able to stimulate T cells when pulsed with peptides as compared to monocyte-derived DCs. However, the administration of cytotoxic chemotherapy may be associated with diminished bone marrow reserve [23–25]. Furthermore, the use of 4 days of growth factor to mobilize CD34-progenitor cells for stem cell transplantation may not be adequate to generate an adequate number of ex vivo expanded DCs. While there is little information demonstrating that the administration of greater numbers of peptide-pulsed DCs improves response rates, our data and those from other groups suggest that clinical and immune responses following vaccination requires the administration of several vaccine treatments. In this study, we did not identify either peptide-specific T cells or clinical responses post vaccination until 6 weeks after treatment. The peak responses were found after the fourth vaccine treatment, and responses did not persist after the completion of therapy. Thus, our data imply that vaccine therapy lasting at least 6 months and perhaps longer may be critical for optimal clinical response. To administer this type of therapy requires the ability to generate approximately 1×10⁸ DCs. Here we show (1) that DCs can be easily generated after 4 days of g-CSF administration in patients with MBC and (2) that ex vivo expansion of CD34-progenitors using multiple different cytokines is effective in generating a large quantity of functional mature DCs, which can be used for vaccine therapy. However, a larger trial will be required to determine the number of vaccinations required to optimize both a clinical and an immune response.

While this trial was not designed to test the efficacy of vaccine therapy in the treatment of patients with MBC, we were somewhat encouraged that both clinical and immune responses were found in patients treated on this trial. Objective clinical responses were found in the two patients treated for subcutaneous disease. Interestingly, there was one patient on this trial without detectable metastatic disease at the time of vaccination and this individual had the most robust immune response during vaccination. One other patient generated a functional immune response post vaccination, although this was only present at the completion of treatment. Thus, this patient had clinical progression of her bone disease prior to the onset of the immune response. Our data suggest that peptide-pulsed DCs may have activity in the treatment of individuals with limited or micrometastatic breast cancer.

One question regarding the outcome of patients on this trial is the absence of a functional immune response from the two patients that had clinical responses. We were unable to show that these cells generated IFN-γ or IL-4 specific to GP2. Unlike antibody, T cells function in tissue not in the bloodstream. Previous investigators have demonstrated a difference in the quality of the immune response during viral infection [26] and in patients with melanoma between T cells found in the bloodstream and locally [27]. Thus, our inability to document a functional immune response from T cells isolated from the bloodstream does not indicate that these cells were not present locally. Our future studies will attempt to biopsy local tissue to evaluate the local immune response during vaccine therapy.

Our preclinical work suggested that unlike trastuzumab [28, 29], DCs pulsed with peptides from HER-2/neu would be active against tumors that expressed HER-2/neu but did not require overexpression of this proto-oncogene. The outcome from this clinical trial supported this observation. Overexpression of HER-2/neu from primary tumor samples was not required for either a clinical or immunological response post vaccination. The two patients that had clinical regression of subcutaneous lesions did not have overexpression of HER-2/neu by immunohistochemistry.

Our goal in this trial was to demonstrate that the administration of DCs pulsed with either the wild-type peptide or the altered peptide ligand, I2L, was safe and could be achieved in patients with MBC. Our previous data in an animal model suggested that the I2L peptide would be more effective in inducing GP2-specific T cells as compared to the GP2 peptide. We were unable to confirm this finding in the present study. However, this trial was not powered to demonstrate that one of the peptides was more efficacious in stimulating a clinical or immune response. Thus, a much larger trial will be needed to determine differences in the efficacy of the I2L and GP2 epitopes.

In summary, we have shown that CD34-derived DCs can be generated from patients with MBC. These cells are potent stimulators of T cells responses and express chemokine receptors necessary for migration into secondary lymphoid organs after ex vivo expansion. After pulsing with peptides from HER-2/neu, DCs could be administered safely to all of the patients treated. This treatment resulted in brief measurable responses in two patients and an immune response in two other patients. This approach appears promising as a method to generate significant quantities of mature functional DCs for vaccine therapy.