Purpose: Breast cancer patients with brain metastasis have a dismal prognosis. We determined the ability of immunostimulatory CpG oligodeoxynucleotides (ODN) to treat or prevent brain metastasis in a mouse model.

Experimental Design: Mice bearing orthotopic breast carcinoma with or without concurrent i.c. tumors were treated by injections of CpG ODN at the primary tumor. Immunologic memory was tested by tumor rechallenge and immune responses were assessed by flow cytometry, delayed-type hypersensitivity, and CTL assays.

Results: Orthotopic tumors regressed in treated mice regardless of whether concurrent i.c. disease was present. In mice bearing only orthotopic tumors, CpG ODN rendered 50% tumor-free and they rejected tumor rechallenge in breast and brain. In mice with concurrent i.c. disease, there was no difference in brain tumor growth compared with saline controls, despite regression of the primary tumor. Flow cytometry revealed that treated mice that died from i.c. disease exhibited a significant increase in brain-infiltrating T and natural killer cells relative to saline controls. CTLs from these mice were able to kill tumor in vitro and extend survival of naive mice bearing less-established brain tumors by adoptive transfer.

Conclusions: The lack of survival benefit in mice with appreciable brain metastasis was not explained by a deficit in lymphocyte trafficking or function because CTLs from these mice killed tumor and inhibited microscopic brain metastasis by adoptive transfer. These results indicate that CpG ODN might be beneficial as a preventative adjuvant to initial therapy preceding brain metastasis or to inhibit progression of microscopic brain metastases.

In the United States, breast cancer remains the most common malignancy in women today. About 212,920 new breast cancer cases and 40,970 deaths from breast cancer were expected in the United States in 2006 (1). The most important factor influencing the outcome of patients with invasive breast cancer is whether the tumor has spread regionally or systemically. Central nervous system (CNS) metastases account for the majority of malignant brain tumors and may engraft into the brain parenchyma or along the leptomeninges. Breast cancer is the second leading cause of CNS metastases. The prognosis for breast cancer patients with CNS involvement is very poor, with a 1-year survival rate of only 20% (2). The incidence of patients with brain metastases was estimated to be 10% to 16%, but this number is probably an underestimate considering that 30% of patients have CNS metastases at autopsy (3). CNS metastases typically occur late in disease progression, often being preceded by metastases in bone, liver, or lung. Standard treatment for brain metastases includes corticosteroids to reduce intracranial pressure, and chemotherapy, surgery, radiosurgery, or whole-brain radiotherapy to kill/remove tumor cells. Treatment-associated toxicity is significant and often debilitating, especially when multiple metastases are treated in the brain parenchyma and leptomeninges simultaneously. Therefore, studies that specifically address the challenges of treating metastatic brain tumors are clearly needed.

Immunotherapy is one approach being considered for the treatment of disseminated cancer. In the case of brain tumors, the appeal of immunotherapy is that tumor-reactive lymphocytes might infiltrate the brain to eradicate microscopic metastases that are undetectable by current imaging technologies. Immunotherapy of brain tumors has unique challenges because the brain is an immunologically specialized organ shielded by the blood-brain barrier and lacking lymphatics in the parenchyma. Toll-like receptors (TLR) recognize a set of conserved pathogen-associated molecular patterns, which allows them to sense pathogens and initiate innate and adaptive immune responses (46). Synthetic CpG oligodeoxynucleotides (ODN) specifically interact with TLR9 on dendritic cells and B cells, leading to their activation, maturation, and eventual priming of cellular and humoral immunity, respectively (7). Direct intratumoral or peritumoral injection of CpG ODN has been studied in animals, with the expectation that the immune system will select the appropriate tumor antigens. This approach showed antitumor activity in animal models of glioblastoma, fibrosarcoma, lung carcinoma, melanoma, and neuroblastoma (810). Several CpG ODN sequences are being tested in patients with glioblastoma (11), non–small cell lung carcinoma (7), and metastatic melanoma (12). Antitumor activity was documented in these clinical trials, albeit only in select patients. CpG ODN has been used as an effective vaccine adjuvant to treat an orthotopic model of breast cancer, but unfortunately, the CNS was ignored in these studies (13).

CpG ODN and dendritic cell vaccines are being tested on a variety of extracranial and primary brain tumors in clinical trials (7, 14), yet very little is known about the efficacy of immunotherapy in the prevention and treatment of brain metastasis. In this study, we sought to characterize the antitumor immune response evoked by CpG ODN in mouse models of premetastatic and postmetastatic disease of the CNS. The results provide insight into the trafficking and function of tumor-reactive lymphocytes in the CNS and indicate that CpG ODN may be an effective adjuvant to standard therapies in the treatment or prevention of microscopic brain metastasis.

Cell line and cell culture. EMT6 and 4T1 are aggressive mammary carcinoma cell lines isolated from BALB/c mice that were obtained from the American Type Culture Collection. Luciferase-stable EMT6-Luc cells were constructed by transfection with a plasmid encoding firefly luciferase and a green fluorescent protein/blasticidin resistance fusion gene: pKT2/PGK-BSD:GFP-clp-Luc (kindly provided by Dr. Andy Wilber, University of Minnesota, Minneapolis, MN). The procedure used to transfect and isolate blasticidin-resistant clones was done as described (15). 4T1, EMT6, and EMT6-Luc were maintained in RPMI 1640 (Life Technologies) with 10% fetal bovine serum (HyClone) and 1% penicillin-streptomycin at 37°C in a humidified incubator with 5% CO2.

Mouse model. Female BALB/c mice (6-8 wk of age) were purchased from The Jackson Laboratory and maintained in the Research Animal Resource facility at the University of Minnesota. All experiments were done in compliance with Research Animal Resource and Institutional Animal Care and Use Committee protocols.

For the primary orthotopic tumor model, 4 × 106 of EMT6-Luc cells were injected into the right mammary fat pad in 100 μL of RPMI 1640. Mice were anesthetized using 2% isoflurane before inoculation. For orthotopic rechallenge, mice cured by CpG ODN and age-matched naive mice were used. Two million EMT6 cells were implanted into the left mammary gland. When the tumor volume exceeded 1,000 mm3, the mice were sacrificed. For i.c. rechallenge, mice cured by CpG ODN or age-matched naive mice were anesthetized with a ketamine (60 mg/kg) and xylazine (7.5 mg/kg) cocktail. Anesthetized mice were placed into a stereotactic frame. A 1-mm burr hole was introduced in the region of the coronal suture using a D#60 drill bit (Plastics One). Stereotactic coordinates used were +0.5 mm anterior, 2.2 medial from bregma, and 3 mm deep from the cortical surface. A total of 5,000 EMT6-Luc cells in 1 μL of RPMI 1640 were delivered using a 10-μL Hamilton syringe. To minimize backflow, the needle was left in place for 1 min following the injection. After injection, the burr hole was covered by bone wax (Ethicon, Inc.), and the incision was closed with a 5-O Dexon suture.

CpG ODN injection, monitoring, and bioluminescent imaging. The timing of the animal procedures is summarized in Figs. 1A and 4A. Mice were imaged for luciferase activity on the indicated days as described (16). Light emitted from the tumors was quantified using Living Image software (Xenogen) and expressed as photons per second per centimeter squared per steradian (p/s/cm2/sr).

Fig. 1.

A, CpG ODN therapy and tumor challenge schedule. Mice were injected into the right mammary fat pad with EMT6-Luc cells on day 0. On days 4, 6, and 8, mice were treated by peritumoral injection of CpG ODN or saline. On day 95, surviving mice and the same number of naive mice were rechallenged with wild-type EMT6 cells. On day 135, surviving mice and the same number of naive mice were challenged i.c. with EMT6-Luc cells into the right striatum. B and C, tumor volume of saline-treated or CpG ODN–treated mice graphed on a per mouse basis; each line represents the tumor volume of an individual mouse. D, Kaplan-Meier survival graph from the mice bearing orthotopic EMT6-Luc tumors that were treated with CpG ODN or saline. Mice treated with CpG ODN survived significantly longer than controls (P < 0.001).

Fig. 1.

A, CpG ODN therapy and tumor challenge schedule. Mice were injected into the right mammary fat pad with EMT6-Luc cells on day 0. On days 4, 6, and 8, mice were treated by peritumoral injection of CpG ODN or saline. On day 95, surviving mice and the same number of naive mice were rechallenged with wild-type EMT6 cells. On day 135, surviving mice and the same number of naive mice were challenged i.c. with EMT6-Luc cells into the right striatum. B and C, tumor volume of saline-treated or CpG ODN–treated mice graphed on a per mouse basis; each line represents the tumor volume of an individual mouse. D, Kaplan-Meier survival graph from the mice bearing orthotopic EMT6-Luc tumors that were treated with CpG ODN or saline. Mice treated with CpG ODN survived significantly longer than controls (P < 0.001).

Close modal

CpG ODN 1826 was used in all CpG ODN treatment experiments: 5′-TCCATGACGTTCCTGACGTT-3′ (Invivogen). Tumor-bearing mice were treated by peritumoral injection (17) of 20 μg CpG ODN or 0.9% saline; a 100-μL final volume was delivered by subdermal injection into four sites around the tumor (25 μL each site) on the indicated days. Tumor volume was determined as described (15).

Delayed-type hypersensitivity and brain histology. On day 13 after i.c. tumor rechallenge, a delayed-type hypersensitivity (DTH) test was done. Briefly, 1 million irradiated (3,000 Rads) EMT6 cells were resuspended in 50 μL of RPMI 1640 and injected into the right hind footpad. As a control, the left hind footpad was injected with 50 μL of RPMI 1640. The footpad thickness was measured with a caliper before and 48 h after injection. The data are graphed as the net increase in thickness: thickness (μm) = postinjection measurement minus preinjection.

On day 15 after i.c. rechallenge, mice were sacrificed for histologic analysis of their brains. The animals were terminally anesthetized by i.p injection of a ketamine (80 mg/kg) and xylazine (10 mg/kg) cocktail solution. Brains were processed for H&E staining as described (16).

Brain-infiltrating lymphocyte and flow cytometry analysis. Intracellular IFN-γ flow cytometry was conducted as reported (18). Briefly, the spleens of mice were harvested and dissociated into a single-cell suspension before perfusion on day 15 after i.c. rechallenge (Fig. 1A). Five million splenocytes were cocultured with 5 × 105 irradiated EMT6 cells in complete growth medium containing 10 μg/mL brefeldin A (Sigma-Aldrich) for 12 h. As a positive control, splenocytes were stimulated with phorbol 12-myristate 13-acetate and ionomycin (Sigma-Aldrich). The nonadherent cells were then analyzed by flow cytometry using the following monoclonal antibodies: phycoerythrin (PE)-Cy5–conjugated anti-mouse CD4 (L3T4, clone RM4-5; BD PharMingen), allophycocyanin-conjugated anti-mouse CD8α (Ly-2, clone 53-6.7; BD PharMingen), PE-conjugated anti-mouse IFN-γ (clone XMG1.2; BD PharMingen), and isotype control PE-conjugated rat IgG1. Stained cells were analyzed on a FACSCanto (BD Biosciences) with a minimum of 500,000 events collected using BD FACSDiva software. The data were imported and analyzed further using FlowJo software (Tree Star).

Brain-infiltrating lymphocytes (BIL) were prepared by perfusing five mice in saline and CpG ODN groups with PBS 22 d after double-tumor inoculations (Fig. 4A). Brains were made into a single-cell suspension by mincing into small pieces and then subjected to a 20-min room temperature digestion with 0.05% trypsin-EDTA (Invitrogen) followed by filtration through a 40-μm cell strainer (BD Bioscience Discovery Labware). BILs were isolated by three-layer Percoll gradient centrifugation (70%, 37%, and 30%, 500 × g, 20 min). Viable BILs from each mouse were counted using a hemocytometer and trypan blue exclusion. Samples from the same group were pooled together and stained after Fc blocking with anti-CD16/32 with the following antibodies: anti-CD4-FITC (clone GK1.5), anti-CD25-allophycocyanin (clone pc61.5), anti-FoxP3-PE (clone Fjk-16s), and rat IgG2α,κ-PE (isotype control) for regulatory T cell (Treg) intracellular staining following the manufacturers' protocol (eBioscience). The following antibodies were used for additional staining: anti-CD4-PE (clone GK1.5), anti-CD8-Alexa Fluor 700 (clone 53-6.7), anti-CD11c-Cy5-PE (clone N418), anti-CD69-Cy5-PE (clone H1.2F3), anti-panNK-FITC (clone DX5), hamster IgG-Cy5-PE, and rat IgMκ (isotype control). All antibodies were from eBioscience. The total number of each BIL subset was quantified by multiplying the percentage of stained cells times the total number of BILs that were counted using a hemocytometer. BIL number is plotted as the total number of cells per brain (Fig. 5).

CTL assay. On day 22 after double-tumor implantations (Fig. 4A), spleens from five mice in each group were harvested. Pooled splenocytes from each group of mice were seeded in six-well plates with complete RPMI 1640 (Invitrogen) supplemented with mouse recombinant interleukin-2 (20 IU/mL; R&D Systems). Splenocytes were cocultured with mitomycin C–treated wild-type EMT6 at a 10:1 E:T ratio. Five days later, CTLs were collected for cytotoxicity testing with EMT6 or 4T1 target cells at E:T ratios of 5:1, 10:1, 20:1, 40:1, and 80:1. Five hours later, measurement of lactate dehydrogenase activity was done according to the manufacturers' protocol (Roche Applied Science). Specific cell lysis was calculated as follows: [(effector-target cell mix − effector cell control) − low control] / (high control − low control) × 100.

Adoptive transfer. Twenty-two days after tumor implantations described in Fig. 4A, splenocytes were harvested and pooled together from saline-treated or CpG ODN–treated mice. Splenocytes were expanded in complete medium with 20 IU/mL for 3 d before adoptive transfer. Mice were infused with 5 million splenocytes via tail vein injection 3 d after i.c. EMT6 tumor challenge.

Reverse transcription-PCR, Western blot, and 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium salt assay. For reverse transcription-PCR, 5 million mouse splenocytes or EMT6-Luc cells were harvested for RNA extraction. RNeasy mini kit and on-column digestion reagents were used for RNA extraction according to the manufacturers' protocol (Qiagen). The following primers were used to detect a 195-bp band specific to mouse TLR9: 5′-CCTGGCTATGCGTACACTGGAG (forward) and 5′-CGGGAGTTCTTGTAGTAGCAGTTCCCG (reverse). A Qiagen one-step reverse transcription-PCR kit was used for PCR according to the manufacturers' protocol. PCR conditions were as follows: 50°C × 30 min as reverse transcription step followed by 95°C × 15 min and then 94°C × 30 s, 54°C × 30 s, and 72°C × 45 s for 40 cycles. Samples without RNA template and started after the initial 50°C reverse transcription step were used as water and no-reverse transcription controls, respectively.

Western blot was conducted using standard methods (19). Briefly, 100 μg of protein lysate from EMT6, 293T, or mouse intestine were run on an 8% polyacrylamide gel and transferred to a nitrocellulose membrane. The membrane was cut in half at about the 50-kDa molecular weight marker. The upper half was incubated with rabbit anti-TLR9 polyclonal antibody (0.5 μg/mL; eBioscience), and the other half was incubated with rabbit anti-glyceraldehyde-3-phosphate dehydrogenase (1:200 dilution; Santa Cruz Biotechnology) overnight at 4°C. The membranes were washed with TBS-Tween 20 and incubated with the horseradish peroxidase–conjugated secondary antibody (anti-rabbit horseradish peroxidase, 1:10,000 dilution; Pierce) for 1 h at room temperature. Detection of antigen was done using chemiluminescence (SuperSignal West Pico Substrate, Pierce).

For the 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium salt assay, EMT6 cells were seeded in a 96-well plate with 2,500 per well and treated with varying doses of CpG ODN 1826. After 48 h, cell viability was measured by using the CellTiter 96 Aqueous Non-Radioactive Cell Proliferation Assay according to the manufacturers' protocol (Promega); absorbance was read at 490 nm with a BioTek Synergy 2 plate reader (BioTek).

Statistical analysis. All data are plotted as the mean value ± SD. Differences between experimental groups were determined by two-way ANOVA and ad hoc comparison using Student's t test. Animal survival was analyzed using the log-rank test. In all cases, a P value of <0.05 was considered statistically significant. All statistical analyses were conducted using Prism 4 statistical software (GraphPad Software).

Peritumoral injection of CpG ODN eradicates primary orthotopic breast tumors. To establish an orthotopic tumor model, BALB/c mice were inoculated with luciferase-stable EMT6 cells (EMT6-Luc) by direct injection into the mammary fat pad. Mice were treated by peritumoral injection with CpG ODN or saline (control) on days 4, 6, and 8 after inoculation (Fig. 1A). Mice treated with CpG ODN exhibited a noticeable but insignificant (P > 0.5) decrease in body weight that persisted for 20 days (Supplementary Fig. S1). Treatment with CpG ODN caused a 63% reduction in tumor growth by 20 days after inoculation that was significant (Supplementary Fig. S2).

All animals that were treated with saline became moribund due to tumor burden between 20 and 40 days and were sacrificed (Fig. 1B). In contrast, 50% of mice treated with CpG ODN were rendered tumor-free as assessed by caliper measurement (Fig. 1C) and exhibited increased survival (P < 0.001; Fig. 1D). Tumors that initially regressed did not recur for 100 days (Fig. 1C). Taken together, these results show that CpG ODN immunotherapy inhibited or eradicated murine breast carcinoma.

Because CpG ODN has been shown to directly cause tumor cell death by reacting with TLR9 in the tumor cell (20), we investigated if EMT6 cells expressed TLR9 or were killed by CpG ODN in vitro. The expression of TLR9 mRNA was detected by reverse transcription-PCR (Fig. 2A) and TLR9 protein was detected by Western blot (Fig. 2B). Despite the expression of TLR9, CpG ODN did not significantly affect the viability of EMT6 cells at 48 h after treatment (Fig. 2C).

Fig. 2.

A, agarose gel electrophoresis of reverse transcription-PCR products using mouse TLR9-specific primers detected expression of TLR9 mRNA (arrow) in splenocytes (positive control) and EMT6 breast carcinoma cells. B, Western blot was used to detect TLR9 protein in EMT6 at 120 kDa and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) at 37 kDa as a loading control. 293T and mouse intestine cell lysates were used as negative and positive controls, respectively. C, EMT6 cells were treated with increasing amounts of CpG ODN 1826 for 48 h and cell viability was assessed by 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium salt assay. There was no significant difference in cell viability at any dose of CpG ODN (P > 0.05).

Fig. 2.

A, agarose gel electrophoresis of reverse transcription-PCR products using mouse TLR9-specific primers detected expression of TLR9 mRNA (arrow) in splenocytes (positive control) and EMT6 breast carcinoma cells. B, Western blot was used to detect TLR9 protein in EMT6 at 120 kDa and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) at 37 kDa as a loading control. 293T and mouse intestine cell lysates were used as negative and positive controls, respectively. C, EMT6 cells were treated with increasing amounts of CpG ODN 1826 for 48 h and cell viability was assessed by 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium salt assay. There was no significant difference in cell viability at any dose of CpG ODN (P > 0.05).

Close modal

Mice cured by CpG immunotherapy rejected orthotopic and i.c. tumor rechallenge. We postulated that long-term survival of mice in the CpG ODN group (50% tumor-free) was due to an antitumor immune response and that luciferase in the EMT6-Luc cells was not a critical antigen for tumor eradication. If true, we expected induction of immunologic memory that would inhibit the growth of the parental EMT6 cells that did not express luciferase. To test this hypothesis, we implanted 2 million parental EMT6 cells in the left mammary fat pad (contralateral to original tumor site) in five naive mice and five mice that were cured from the primary tumor challenge. Parental EMT6 cells grew quickly in naive mice, exceeding 1,000 mm3 on day 20 after rechallenge, whereas the same tumor cells did not form palpable tumors in the mice cured by CpG ODN (Supplementary Fig. S3). Accordingly, all mice cured by CpG ODN (five of five) survived beyond 40 days and naive animals did not survive beyond 25 days (Supplementary Fig. S4).

Although these results were consistent with induction of immunologic memory, the potential for protection within the CNS remained unclear. To address this, mice that had rejected orthotopic rechallenge with parental EMT6 cells were i.c. inoculated with EMT6-Luc cells. Naive mice were used as controls. In vivo bioluminescent imaging revealed that all mice had appreciable tumor in the brain 1 day after imaging (Fig. 3A and B). As soon as 5 days after inoculation, mice that had rejected the orthotopic tumors exhibited a 61.2% reduction in tumor viability compared with naive controls (P = 0.022; Fig. 3B). By 10 days after i.c. inoculation, there was no detectable tumor in all mice cured by CpG ODN, whereas the tumor in naive mice continued to grow (Fig. 3A and B).

Fig. 3.

Evidence for immunologic memory in CpG ODN–treated mice. A, mice treated as described in Fig. 1A were i.c. challenged with EMT6-Luc cells. Representative in vivo images are shown of naive mice or mice that were cured of orthotopic tumor by CpG ODN. B, the average photons in each group are plotted at the indicated times after i.c. tumor inoculation. Columns, mean (n = 5 per group); bars, SD. **, P = 0.032, comparing CpG ODN–cured mice with naive mice. C, representative brain sections from mice in A and B above. The brains were excised on day 15 after i.c. challenge. Images are photographed at magnification of ×1 (top) and ×200 (bottom). D, 15 d after intracranial inoculation, mice treated with CpG ODN and naive controls were sacrificed and their splenocytes were stimulated by coculture with EMT6 cells for 12 h. Intracellular (IFN-γ staining) staining was conducted and the cells were analyzed by flow cytometry. Irrelevant isotype-matched antibodies and cells from naive mice were used to set gates. The IFN-γ–expressing population was compared with each isotype control. Columns, mean; bars, SD. ***, P = 0.025; **, P = 0.04, comparing CpG ODN–cured mice with naive mice.

Fig. 3.

Evidence for immunologic memory in CpG ODN–treated mice. A, mice treated as described in Fig. 1A were i.c. challenged with EMT6-Luc cells. Representative in vivo images are shown of naive mice or mice that were cured of orthotopic tumor by CpG ODN. B, the average photons in each group are plotted at the indicated times after i.c. tumor inoculation. Columns, mean (n = 5 per group); bars, SD. **, P = 0.032, comparing CpG ODN–cured mice with naive mice. C, representative brain sections from mice in A and B above. The brains were excised on day 15 after i.c. challenge. Images are photographed at magnification of ×1 (top) and ×200 (bottom). D, 15 d after intracranial inoculation, mice treated with CpG ODN and naive controls were sacrificed and their splenocytes were stimulated by coculture with EMT6 cells for 12 h. Intracellular (IFN-γ staining) staining was conducted and the cells were analyzed by flow cytometry. Irrelevant isotype-matched antibodies and cells from naive mice were used to set gates. The IFN-γ–expressing population was compared with each isotype control. Columns, mean; bars, SD. ***, P = 0.025; **, P = 0.04, comparing CpG ODN–cured mice with naive mice.

Close modal

Antitumor immunity in the CNS and detection of tumor-reactive lymphocytes in mice treated with CpG ODN. A DTH assay was then conducted to determine if the antitumor effects of CpG ODN therapy correlated with the induction of tumor-reactive lymphocytes. DTH is a T-cell–mediated inflammatory reaction that occurs when a previously encountered antigen is recognized. At 13 days after i.c. inoculation, the tumor-bearing naive mice and CpG ODN–cured mice were inoculated with irradiated EMT6 cells in the hind footpad or medium in the contralateral foot pad as a control. The footpad thickness was measured before and 48 h after inoculation. Mice that had been previously treated with CpG ODN exhibited significantly greater footpad thickness compared with naive controls (Supplementary Fig. S5).

By 15 days after inoculation, naive tumor-bearing mice became moribund and were sacrificed along with the putatively tumor-free mice. Brain and spleen were collected from all mice to analyze histologically and immunologically, respectively. Consistent with the imaging data, histologic analysis revealed that naive mice had massive brain tumors that were likely responsible for their neurologic symptoms (Fig. 3C). No evidence of tumor could be detected in the brains of CpG ODN–treated mice, thereby showing the induction of protective immunity in the CNS.

In addition to DTH, we investigated IFN-γ elaboration from T cells as a further measure of tumor-reactive lymphocytes. Splenocytes were harvested from naive mice or CpG ODN–cured mice that had been i.c. inoculated. Splenocytes were cocultured with irradiated EMT6 cells. Twelve hours later, IFN-γ–expressing CD4+ and CD8+ T cells were detected using flow cytometry. A 2.5-fold increase in IFN-γ–positive CD4+ T cells was observed in CpG ODN–cured mice relative to naive mice bearing brain tumors (P < 0.001; Fig. 3D). Similarly, there was a 1.5-fold increase in IFN-γ–positive CD8+ T cells observed in CpG ODN–cured mice relative to naive mice bearing brain tumors (P = 0.025; Fig. 3D).

CpG ODN immunotherapy caused regression of primary tumors but not brain tumors. Although peritumoral CpG ODN protected mice from i.c. tumor engraftment, the effects in the setting of concurrent primary and i.c. disease remained unclear. To investigate this, mice were inoculated with EMT6-Luc cells into the mammary fat pad and brain on the same day. Peritumoral injections of CpG ODN or saline were given only at the primary tumor site (Fig. 4A). CpG ODN administration caused regression or significantly inhibited growth of tumors at the primary site relative to saline treatment (P = 0.003; Supplementary Fig. S6; Fig. 4C and D). In contrast, there was no significant difference in brain tumor growth in mice treated with CpG ODN or saline (P > 0.05; Supplementary Fig. S7). Accordingly, survival between the two groups was nearly identical (Supplementary Fig. S8), despite the significant difference in primary tumor growth (Supplementary Fig. S6).

Fig. 4.

Effects of CpG ODN in concurrent tumor model. A, 20 mice were inoculated with EMT6-Luc cells into the mammary fat pad and i.c. on day 0. CpG ODN or saline was given peritumorally around the primary breast tumor at the indicated days (n = 10 per group). Splenocytes and BILs were harvested from 50% of animals in each group on day 22, whereas the remaining five animals were monitored for survival. B and C, tumor volume of saline-treated or CpG ODN–treated mice graphed on a per mouse basis; each line represents the tumor volume of an individual mouse. D, representative in vivo bioluminescent images of i.c. and primary tumor growth (“brain” refers to light emitted from the skull and “breast” refers to light emitted from the breast tumor).

Fig. 4.

Effects of CpG ODN in concurrent tumor model. A, 20 mice were inoculated with EMT6-Luc cells into the mammary fat pad and i.c. on day 0. CpG ODN or saline was given peritumorally around the primary breast tumor at the indicated days (n = 10 per group). Splenocytes and BILs were harvested from 50% of animals in each group on day 22, whereas the remaining five animals were monitored for survival. B and C, tumor volume of saline-treated or CpG ODN–treated mice graphed on a per mouse basis; each line represents the tumor volume of an individual mouse. D, representative in vivo bioluminescent images of i.c. and primary tumor growth (“brain” refers to light emitted from the skull and “breast” refers to light emitted from the breast tumor).

Close modal

CpG ODN immunotherapy was associated with a marked increase in BILs. To analyze BIL quantity and phenotype, BILs were harvested from mice when they became moribund 22 days after tumor inoculations (presumably due to brain tumor burden; Fig. 4A). Flow cytometry analyses revealed a 5- to 8-fold increase in the absolute number of CD4+ and CD8+ T cells and NK1.1+ cells in CpG ODN–treated mice relative to mice treated with saline (Fig. 5A). Triple staining showed that CpG ODN–treated mice had a significantly greater number of activated (CD69+ or CD25+) brain-infiltrating T cells and a noticeable but statistically insignificant increase in CD4+CD25+FoxP3+ Tregs (21) relative to saline-treated mice (Fig. 5B and C). However, we noted that the relative portion of Tregs to total CD4+ T cells was over 1-fold higher in the BILs of saline-treated mice, revealing a putatively more proinflammatory population of CD4+ T cells in CpG ODN–treated mice (Fig. 5D). Collectively, these data show that although CpG ODN treatment failed to significantly affect the growth of well-established i.c. tumors, it facilitated a massive infiltration of innate and adaptive lymphocyte effectors into the brain.

Fig. 5.

CpG ODN induced an increase in BILs. A, animals were treated as described in Fig. 4A. BILs were collected from five mice in each group 22 d after tumor inoculations and analyzed by flow cytometry. The total number of each cell type is shown. Columns, mean; bars, SD. *, P = 0.007; **, P = 0.058; ***, P = 0.005. B, triple staining of BIL for CD4, CD25, and FoxP3 was used to measure Tregs. *, P = 0.018; **, P = 0.055, not significant (ns). C, double staining was used to quantify activated (CD69+) CD4+ and CD8+ BILs. *, P = 0.006; **, P = 0.004. D, the mean number of Tregs from five mice per group was divided by the total number of CD4+ T cells and is graphed as a percentage.

Fig. 5.

CpG ODN induced an increase in BILs. A, animals were treated as described in Fig. 4A. BILs were collected from five mice in each group 22 d after tumor inoculations and analyzed by flow cytometry. The total number of each cell type is shown. Columns, mean; bars, SD. *, P = 0.007; **, P = 0.058; ***, P = 0.005. B, triple staining of BIL for CD4, CD25, and FoxP3 was used to measure Tregs. *, P = 0.018; **, P = 0.055, not significant (ns). C, double staining was used to quantify activated (CD69+) CD4+ and CD8+ BILs. *, P = 0.006; **, P = 0.004. D, the mean number of Tregs from five mice per group was divided by the total number of CD4+ T cells and is graphed as a percentage.

Close modal

Lymphocytes from mice that failed therapy have adequate function in vitro and in vivo. The increase in BILs following CpG ODN administration without any corresponding effect on brain tumor growth led us to question whether the CTLs from these mice had an inherent deficit in function. Alternatively, the brain tumors in these mice may have been simply too large for an antitumor immune response to significantly affect. To examine these possibilities, splenocytes were harvested from mice treated with CpG ODN or saline when they became moribund from tumor burden (on day 22; Fig. 4A). The splenocytes from mice treated with saline or CpG ODN were pooled. The splenocytes were either adoptively transferred into naive mice bearing 3-day-old i.c. EMT6 tumors or used in CTL assays to determine their ability to kill tumor in vitro.

CTLs generated from mice treated with saline did not show any significant ability to kill EMT6 cells in vitro (Fig. 6A). In contrast, CTLs from CpG ODN–treated mice readily killed EMT6 cells. Furthermore, CTLs from these mice did not kill 4T1 cells (an irrelevant syngeneic mammary carcinoma cell line), thereby showing specificity (Fig. 6B). Consistent with the CTL data, mice that were given an i.v. injection of splenocytes from saline-treated mice rapidly succumbed to i.c. EMT6 tumors, whereas adoptive transfer of splenocytes from CpG ODN–treated mice provided a significant extension of survival (Fig. 6C).

Fig. 6.

Splenocytes from CpG ODN–treated mice have tumoricidal effects on EMT6. A, CTL assay on EMT6 and 4T1 (control) cells using saline-primed E:T cell ratios of 80:1 to 5:1. B, same as in A, only the effectors were harvested from CpG ODN–treated mice. C, naive mice were i.c. inoculated with EMT6 cells and, 3 d later, received a tail vein injection of splenocytes from mice treated with saline or CpG ODN. Mice treated by CpG ODN–primed splenocytes survived significantly longer relative to mice treated by saline-primed splenocytes (n = 6-8 per group).

Fig. 6.

Splenocytes from CpG ODN–treated mice have tumoricidal effects on EMT6. A, CTL assay on EMT6 and 4T1 (control) cells using saline-primed E:T cell ratios of 80:1 to 5:1. B, same as in A, only the effectors were harvested from CpG ODN–treated mice. C, naive mice were i.c. inoculated with EMT6 cells and, 3 d later, received a tail vein injection of splenocytes from mice treated with saline or CpG ODN. Mice treated by CpG ODN–primed splenocytes survived significantly longer relative to mice treated by saline-primed splenocytes (n = 6-8 per group).

Close modal

Significant progress has been made in treating extracranial breast cancer metastases with the implementation of tumor-targeted drugs such as trastuzumab, a monoclonal antibody with high affinity for human epidermal growth factor receptor-2 (reviewed in ref. 22). However, trastuzumab and many other drugs used for control of extracranial disease do not efficiently cross the blood-brain barrier. Furthermore, the incidence of brain metastases in patients treated with trastuzumab is 28% to 43%, alarmingly higher than historical series of 10% to 16% (reviewed in ref. 3). This apparent increase in brain metastases might be related to the aggressive nature of human epidermal growth factor receptor-2–positive tumors or patient selection (23), yet it is not surprising for several reasons. First, as the survival of breast cancer patients is extended by effective treatment of extracranial disease, CNS metastases have more time to develop and become symptomatic. Second, metastasis to the brain may provide a selective growth/survival advantage to tumor cells because the blood-brain barrier inhibits delivery of certain drugs into tumor cells. Therefore, new approaches need to be explored to treat brain metastases, including manipulation of the immune system.

The brain has been termed an immunologically specialized organ due to several curious observations (i.e., that antibodies are typically too large to penetrate the blood-brain barrier, the complete lack of lymphatic vessels in the brain parenchyma, and that the blood-brain barrier hinders, in part, the trafficking of lymphocytes; reviewed in ref. 24). Consistent with these observations, studies have shown sustained transgene expression in the brain following viral vector-mediated gene transfer, despite a systemic antivector immune response (2527). Paradoxically, studies have shown the role of T-cell–mediated immune response in clearance of viral infection (28, 29) and eradication of brain tumors treated by local gene therapy (30) or dendritic cell vaccines (31). Therefore, immunotherapy seems a viable approach in the treatment of brain tumors.

The antitumor activity of CpG ODN has been shown by many investigators experimentally (7) and, more recently, in select cancer patients (7, 11, 12). In the current study, we assessed the efficacy of CpG ODN immunotherapy against murine EMT6 breast carcinoma. Although the treatment of this kind of animal model with CpG ODN is not particularly novel, the data about the CNS are both novel and potentially important. Consistent with other studies, we found that peritumoral injection of CpG ODN either inhibited or completely eradicated preestablished breast tumors. In contrast to the reported cytotoxicity of CpG ODN on mouse glioma cells (20), CpG ODN did not directly kill EMT6 breast carcinoma at doses up to 600 μg/mL, despite the fact that EMT6 expressed TLR9 (Fig. 3A-C). Therefore, we attribute the antitumor activity of peritumoral CpG ODN injection to the indirect immunologic effects involving generation of tumor-reactive CD4+ and CD8+ T cells. Additionally, it is well known that dendritic cells, natural killer cells, B cells, and macrophages are typically involved in priming the complete antitumor immune response induced by CpG ODN in mice (7, 10, 32).

We previously reported that depletion of CD4+, CD8+, and NK1.1+ cells abolished cure of brain tumors in mice that were vaccinated with tumor lysate plus CpG ODN s.c. (33). In this study, we found a dramatic increase in these three populations of BILs in mice treated by CpG ODN, consistent with their putative role in immune surveillance of brain tumors that is exacerbated by CpG ODN. Importantly, more of the brain-infiltrating CD4+ and CD8+ cells expressed CD25 and CD69 activation markers in the CpG ODN group (Fig. 5B and C), revealing that the increased quantity of BIL was also better suited for tumor eradication. We also noted appreciable numbers of brain-infiltrating Tregs in CpG ODN–treated and saline-treated groups, which are known to inhibit productive antitumor immune responses in murine glioma models (21, 34, 35) and likely had a similar function in the EMT6 brain metastasis model. It will be important to address the effect of Treg depletion in future studies.

The rejection of brain tumor challenge in mice that were cured of their orthotopic tumor shows that administration of extracranial TLR agonists primes antitumor immunity capable of inhibiting brain tumor growth. However, the direct injection of tumor cells into the brain is not a perfect model of brain metastases because it induces an acute inflammatory response caused by the needle, a very different circumstance than natural metastasis through the blood capillaries. We attempted to address this, at least in part, by adoptively transferring splenocytes from CpG ODN–treated animals into naive mice bearing 3-day-old brain tumors, in which the acute inflammation caused by the needle damage should be waning. In this setting, CpG ODN–primed splenocytes were still able to significantly inhibit the growth of i.c. EMT6 tumors (Fig. 6C), an extremely aggressive brain tumor with a median survival of 21 days in control animals.

The regression of the primary breast tumor in animals with simultaneously progressing brain tumors suggests that different factors or cell types may be available for control of tumor growth extracranially as opposed to in the brain. This could be due to a variety of factors, including poor penetration of antibodies into the brain parenchyma and lower concentration of inflammatory cytokines induced by CpG ODN in the brain relative to the breast tumor site. Because the adoptive transfer of CpG ODN–primed splenocytes was able to inhibit the growth of 3-day-old brain tumors and CTLs from mice that died of progressing brain tumor lysed EMT6 intro, it is unlikely that failure to affect brain tumor growth in the concurrent model was due to a deficit in effector cell function. A more likely explanation may be that immunotherapy had its limits, and the brain tumors in these animals had progressed beyond a point that could be effected by the BILs. We observed a significant lag time between CpG ODN injection and regression of the primary tumor, taking 20 to 40 days before complete disappearance of the tumor was noted by caliper measurement (Figs. 1B and 4C). This response time in which tumor-reactive T-cell populations are presumably proliferating, infiltrating tumor tissue, and killing tumor cells is probably too slow to affect EMT6 brain tumors that kill 50% of mice in less than 25 days. In addition, the combination of tumor load at multiple sites in the concurrent model may have contributed to enhanced immunosuppression relative to animals with a single tumor. Therefore, we conclude that CpG ODN is probably best suited for treatment of minimal disease burden or prevention of brain metastasis in this model. Although one must be cautious about extrapolating to human patients, these data indicate that CpG ODN may be most rationally used as an adjuvant to front-line therapies before appreciable brain metastasis engrafts.

In this study, we also reported the creation of EMT6-Luc cells as an immunocompetent breast cancer model that can be noninvasively quantified. Animals that had undetectable i.c. tumor by histology exhibited no detectable tumor by bioluminescence. One potential disadvantage of this model might be that luciferase is recognized as a xenoantigen, leading to tumor rejection or enhanced immunogenicity. However, we did not find evidence for this. First, EMT6-Luc cells formed lethal tumors. Second, when CpG ODN–cured mice were challenged with parental EMT6 cells, these tumors were still rejected, thereby showing that luciferase was not providing critical antigens for tumor eradication. Likewise, the target cells used in the DTH and CTL assays were the parental EMT6 cells, which provided evidence of tumor-reactive lymphocytes in CpG ODN–treated mice. Despite the potential limitations of the EMT6-Luc model, it may prove useful to study tumor metastases in living animals longitudinally over time.

Recent studies showed a positive correlation between the induction of DTH in response to autologous tumor cell injection and a prolonged survival in cancer patients treated with immunotherapy (36, 37). Consistent with this finding, we found that mice exhibiting strong DTH response were able to reject intracranial tumor challenge within 10 days (Fig. 3A; Supplementary Fig. S5). The results of this study suggest that peritumoral CpG ODN administration warrants clinical investigation as a metastasis prevention strategy. There is evidence in murine models (32) and in non–small cell lung carcinoma patients (7) that systemic CpG ODN has a strong synergy with chemotherapy, probably due to chemotherapy-induced tumor cell death causing release of tumor antigens while antigen-presenting cells are simultaneously being activated by CpG ODN. It would seem feasible to administer CpG ODN peritumorally before surgical excision of the primary tumor. Alternatively, CpG ODN could be administered i.v. with concurrent chemotherapy followed by monitoring the incidence of intracranial metastases compared with patients receiving chemotherapy only. CpG ODN immunotherapy may offer an attractive adjuvant to the treatment of cancer patients at high risk of developing brain metastases.

No potential conflicts of interest were disclosed.

Grant support: American Cancer Society grant IRG-58-001-46-IRG35, NIH/National Cancer Institute grant 5 P30 CA077598 (J.R. Ohlfest and W. Chen), NIH/National Institute of Neurological Disorders and Stroke grant 1R21-NS055738-01A2 (J.R. Ohlfest), Randy Shaver Foundation (J.R. Ohlfest), and Graduate School of the University of Minnesota (J.R. Ohlfest).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).

1
Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2006.
CA Cancer J Clin
2006
;
56
:
106
–30.
2
Engel J, Eckel R, Aydemir U, et al. Determinants and prognoses of locoregional and distant progression in breast cancer.
Int J Radiat Oncol Biol Phys
2003
;
55
:
1186
–95.
3
Lin NU, Bellon JR, Winer EP. CNS metastases in breast cancer.
J Clin Oncol
2004
;
22
:
3608
–17.
4
Schroder K, Sweet MJ, Hume DA. Signal integration between IFNγ and TLR signalling pathways in macrophages.
Immunobiology
2006
;
211
:
511
–24.
5
Sester DP, Brion K, Trieu A, et al. CpG DNA activates survival in murine macrophages through TLR9 and the phosphatidylinositol 3-kinase-Akt pathway.
J Immunol
2006
;
177
:
4473
–80.
6
Steinman RM, Hemmi H. Dendritic cells: translating innate to adaptive immunity.
Curr Top Microbiol Immunol
2006
;
311
:
17
–58.
7
Krieg AM. Therapeutic potential of Toll-like receptor 9 activation.
Nat Rev Drug Discov
2006
;
5
:
471
–84.
8
Carpentier AF, Chen L, Maltonti F, Delattre JY. Oligodeoxynucleotides containing CpG motifs can induce rejection of a neuroblastoma in mice.
Cancer Res
1999
;
59
:
5429
–32.
9
Carpentier AF, Xie J, Mokhtari K, Delattre JY. Successful treatment of intracranial gliomas in rat by oligodeoxynucleotides containing CpG motifs.
Clin Cancer Res
2000
;
6
:
2469
–73.
10
Kawarada Y, Ganss R, Garbi N, Sacher T, Arnold B, Hammerling GJ. NK− and CD8(+) T cell-mediated eradication of established tumors by peritumoral injection of CpG-containing oligodeoxynucleotides.
J Immunol
2001
;
167
:
5247
–53.
11
Carpentier A, Laigle-Donadey F, Zohar S, et al. Phase 1 trial of a CpG oligodeoxynucleotide for patients with recurrent glioblastoma.
Neuro-oncol
2006
;
8
:
60
–6.
12
Pashenkov M, Goess G, Wagner C, et al. Phase II trial of a toll-like receptor 9-activating oligonucleotide in patients with metastatic melanoma.
J Clin Oncol
2006
;
24
:
5716
–24.
13
Nava-Parada P, Forni G, Knutson KL, Pease LR, Celis E. Peptide vaccine given with a Toll-like receptor agonist is effective for the treatment and prevention of spontaneous breast tumors.
Cancer Res
2007
;
67
:
1326
–34.
14
Ridgway D. The first 1000 dendritic cell vaccinees.
Cancer Invest
2003
;
21
:
873
–86.
15
Ohlfest JR, Demorest ZL, Motooka Y, et al. Combinatorial antiangiogenic gene therapy by nonviral gene transfer using the sleeping beauty transposon causes tumor regression and improves survival in mice bearing intracranial human glioblastoma.
Mol Ther
2005
;
12
:
778
–88.
16
Wu A, Oh S, Ericson K, et al. Transposon-based interferon γ gene transfer overcomes limitations of episomal plasmid for immunogene therapy of glioblastoma.
Cancer Gene Ther
2007
;
14
:
550
–60.
17
Fallarino F, Puccetti P. Toll-like receptor 9-mediated induction of the immunosuppressive pathway of tryptophan catabolism.
Eur J Immunol
2006
;
36
:
8
–11.
18
Cohen AD, Diab A, Perales MA, et al. Agonist anti-GITR antibody enhances vaccine-induced CD8(+) T-cell responses and tumor immunity.
Cancer Res
2006
;
66
:
4904
–12.
19
Geurts AM, Yang Y, Clark KJ, et al. Gene transfer into genomes of human cells by the sleeping beauty transposon system.
Mol Ther
2003
;
8
:
108
–17.
20
El Andaloussi A, Sonabend AM, Han Y, Lesniak MS. Stimulation of TLR9 with CpG ODN enhances apoptosis of glioma and prolongs the survival of mice with experimental brain tumors.
Glia
2006
;
54
:
526
–35.
21
Grauer OM, Nierkens S, Bennink E, et al. CD4+FoxP3+ regulatory T cells gradually accumulate in gliomas during tumor growth and efficiently suppress antiglioma immune responses in vivo.
Int J Cancer
2007
;
121
:
95
–105.
22
Gasparini G, Longo R, Torino F, Morabito A. Therapy of breast cancer with molecular targeting agents.
Ann Oncol
2005
;
16
Suppl 4:
iv28
–36.
23
Gabos Z, Sinha R, Hanson J, et al. Prognostic significance of human epidermal growth factor receptor positivity for the development of brain metastasis after newly diagnosed breast cancer.
J Clin Oncol
2006
;
24
:
5658
–63.
24
Ransohoff RM, Kivisakk P, Kidd G. Three or more routes for leukocyte migration into the central nervous system.
Nat Rev Immunol
2003
;
3
:
569
–81.
25
Thomas CE, Schiedner G, Kochanek S, Castro MG, Lowenstein PR. Peripheral infection with adenovirus causes unexpected long-term brain inflammation in animals injected intracranially with first-generation, but not with high-capacity, adenovirus vectors: toward realistic long-term neurological gene therapy for chronic diseases.
Proc Natl Acad Sci U S A
2000
;
97
:
7482
–7.
26
Thomas CE, Schiedner G, Kochanek S, Castro MG, Lowenstein PR. Preexisting antiadenoviral immunity is not a barrier to efficient and stable transduction of the brain, mediated by novel high-capacity adenovirus vectors.
Hum Gene Ther
2001
;
12
:
839
–46.
27
Abordo-Adesida E, Follenzi A, Barcia C, et al. Stability of lentiviral vector-mediated transgene expression in the brain in the presence of systemic antivector immune responses.
Hum Gene Ther
2005
;
16
:
741
–51.
28
Cheeran MC, Gekker G, Hu S, Min X, Cox D, Lokensgard JR. Intracerebral infection with murine cytomegalovirus induces CXCL10 and is restricted by adoptive transfer of splenocytes.
J Neurovirol
2004
;
10
:
152
–62.
29
Cheeran MC, Gekker G, Hu S, Palmquist JM, Lokensgard JR. T cell-mediated restriction of intracerebral murine cytomegalovirus infection displays dependence upon perforin but not interferon-γ.
J Neurovirol
2005
;
11
:
274
–80.
30
King GD, Curtin JF, Candolfi M, Kroeger K, Lowenstein PR, Castro MG. Gene therapy and targeted toxins for glioma.
Curr Gene Ther
2005
;
5
:
535
–57.
31
Fecci PE, Mitchell DA, Archer GE, et al. The history, evolution, and clinical use of dendritic cell-based immunization strategies in the therapy of brain tumors.
J Neurooncol
2003
;
64
:
161
–76.
32
Weigel BJ, Rodeberg DA, Krieg AM, Blazar BR. CpG oligodeoxynucleotides potentiate the antitumor effects of chemotherapy or tumor resection in an orthotopic murine model of rhabdomyosarcoma.
Clin Cancer Res
2003
;
9
:
3105
–14.
33
Wu A, Oh S, Gharagozlou S, et al. In vivo vaccination with tumor cell lysate plus CpG oligodeoxynucleotides eradicates murine glioblastoma.
J Immunother
2007
;
30
:
789
–97.
34
Fecci PE, Sweeney AE, Grossi PM, et al. Systemic anti-CD25 monoclonal antibody administration safely enhances immunity in murine glioma without eliminating regulatory T cells.
Clin Cancer Res
2006
;
12
:
4294
–305.
35
El Andaloussi A, Han Y, Lesniak MS. Prolongation of survival following depletion of CD4+CD25+ regulatory T cells in mice with experimental brain tumors.
J Neurosurg
2006
;
105
:
430
–7.
36
Yamanaka R, Homma J, Yajima N, et al. Clinical evaluation of dendritic cell vaccination for patients with recurrent glioma: results of a clinical phase I/II trial.
Clin Cancer Res
2005
;
11
:
4160
–7.
37
de Vries IJ, Bernsen MR, Lesterhuis WJ, et al. Immunomonitoring tumor-specific T cells in delayed-type hypersensitivity skin biopsies after dendritic cell vaccination correlates with clinical outcome.
J Clin Oncol
2005
;
23
:
5779
–87.

Supplementary data