Neoadjuvant Vaccination Provides Superior Protection against Tumor Relapse following Surgery Compared with Adjuvant Vaccination

Tumors that recur following surgical resection of melanoma are typically metastatic and associated with poor prognosis. Using the murine B16F10 melanoma and a robust antimelanoma vaccine, we evaluated immunization as a tool to improve tumor-free survival following surgery. We investigated the utility of vaccination in both neoadjuvant and adjuvant settings. Surprisingly, neoadjuvant vaccination was far superior and provided ∼100% protection against tumor relapse. Neoadjuvant vaccination was associated with enhanced frequencies of tumor-specific T cells within the tumor and the tumor-draining lymph nodes following resection. We also observed increased infiltration of antigen-specific T cells into the area of surgery. This method should be amenable to any vaccine platform and can be readily extended to the clinic. [Cancer Res 2009;69(9):3979–85]

Surgery is the leading treatment modality for melanoma and most solid tumors. More than 90% of primary melanomas can be cured by surgical resection if diagnosed early. However, resection alone is rarely curative for advanced tumors due to either local tumor recurrence or outgrowth of micrometastases. In that regard, patients with stage II melanoma are at high risk for recurrent disease and are reported to have only 40% to 60% chance of survival 5 years after the surgery (1).

Cancer vaccines offer an appealing strategy to improve disease-free survival following surgical resection. Preclinical studies have shown that cancer vaccines can routinely confer protection in prophylactic settings, but the same vaccines usually exhibit only limited therapeutic efficacy (2, 3). We have observed that existing tumors can suppress immunity against tumor-associated antigens, limiting the activity of cancer vaccines (4). Furthermore, in patients with advanced-stage disease, tumor progression can occur within the window of time required for induction of an immune response following vaccination. Based on these data, it seems logical that cancer vaccines would be most useful as an adjuvant treatment in the setting of minimal residual disease. Alternatively, cancer vaccines could be employed in the neoadjuvant setting. Under these circumstances, surgical resection could be timed to coincide with the peak of the immune response. Indeed, such a strategy has been evaluated in the clinic (5, 6). These reports showed evidence of vaccine immunogenicity, but these studies were not sufficiently powered to make conclusion regarding efficacy. To date, however, the issue of whether neoadjuvant immunization provides an advantage over adjuvant immunization has not been addressed. Thus, the optimal vaccination schedule remains unknown.

With regards to melanoma, a limited number of randomized clinical trials have tested the efficacy of adjuvant vaccination (7–9). All of those trials were halted early because vaccination did not affect improve either disease-free or overall survival. It was unclear from those studies whether the lack of clinical benefit was due to the strategy (adjuvant vaccination) or the vaccine. We have found that replication-defective adenovirus (rAd)-based cancer vaccines produce robust antitumor immunity. In particular, a rAd vector expressing human dopachrome tautomerase (AdhDCT) provided superior protection against the murine B16F10 melanoma compared with rAd vectors expressing other melanoma-associated antigens (10–13).¹

DCT (also called tyrosinase-related protein 2) is a promising candidate for the development of melanoma vaccines, as it is widely expressed in melanoma and is naturally recognized by melanoma-specific CTLs (14, 15). Despite the robust activity of AdhDCT in prophylactic models (10, 11), we have observed that vaccination with AdhDCT was ineffective even when the vaccine was administered as little as 1 day after seeding the tumor (4), suggesting that AdhDCT may only be useful in the setting of minimal tumor burden. Given the robust protective immunity provided by AdhDCT in prophylactic, but not therapeutic, settings, we reasoned that this vaccine would provide a stringent test for our hypothesis that immunization against tumor antigen(s) may improve disease-free survival following surgery. To this end, we investigated the effect of neoadjuvant and adjuvant immunization with AdhDCT on tumor relapse in the B16F10 melanoma model.

Mice. Six- to 8-week-old female C57BL/6 mice were obtained from Charles River Breeding Laboratories. All of our investigations have been approved by the McMaster Animal Research Ethics Board.

rAd and immunizations. All rAd vectors employed in this study contain deletions of E1 and E3 regions (16). The expression cassettes were inserted into the E1 region under the control of the murine cytomegalovirus (CMV) promoter and the SV40 polyadenylation sequence. The rAd vectors were propagated using 293 cells and purified using CsCl gradient centrifugation as described previously (17). AdhDCT encodes the full-length human DCT (11). AdLCMV GP expresses the sequence corresponding to residues 33 to 41 and 61 to 80 of the lymphocytic choriomeningitis virus glycoprotein (18). For immunizations, 10⁸ pfu Ad vector was prepared in sterile PBS and injected into both rear thighs (50 μL/thigh).

Peptides. The immunodominant MHC class I peptide of hDCT (DCT_180-188) and MHC class II peptide (hDCT_88-102) were purchased from Biomer Technologies. Peptides were dissolved in DMSO and stored at −20°C.

Tumor cell lines and tumor model. B16F10 cells was cultured in F11 medium supplemented with 10% fetal bovine serum, 2 mmol/L l-glutamine, 100 units/mL penicillin, and 100 μg/mL streptomycin. Tumors were induced by intradermal injection of 10⁵ tumor cells in 30 μL sterile PBS on the shaved back of the mice. Tumor size was measured twice a week using calipers.

Surgical tumor excision. Tumors were surgically excised 14 days following tumor inoculation. Mice were anesthetized with isofluorane and tumors were removed with a 2 mm perimeter of healthy skin. Incisions were closed with steel wound clips (Harvard Apparatus) and mice were given 50 μL of 1:10 diluted temgesic for pain relief.

Preparation of tissues for flow cytometry. Lymphocytes from the spleen and lymph nodes were isolated by teasing the tissues apart between two sterile microscope slides. Vaccine-draining lymph nodes consisted of inguinal, ileac, and popliteal, whereas tumor-draining lymph nodes (TDLN) were brachial and axillary. For isolation of lymphocyte from the skin, tissue sections (∼1.5 × 1.5 cm) were finely minced with scissors and digested in HBSS containing 0.1% collagenase type I (Invitrogen Life Technologies) for 1 h at 37°C. Following the digestion, released cells were filtered through a 70 μm strainer and washed twice with PBS before further use. For isolation of tumor-infiltrating lymphocytes, tumors were dissected, cut into small pieces, and digested with a mixture of collagenase type I (0.5 mg/mL), DNase (0.2 mg/mL), and hyalorunidase (0.02 mg/mL) for 1 h at 37°C. Cells were then passed through 70 μm cell strainers, centrifuged, and further purified by positive selection using Thy1.2-PE selection cocktail (EasySep; Stem Cell Technologies).

Flow cytometry reagents, acquisition, and analysis. Anti-CD8α, anti-CD4, anti-IFN-γ, anti-tumor necrosis factor-α, and anti-CD3 were purchased from BD Biosciences. Anti-FoxP3 was purchased from eBioscience. Data from stained samples were acquired using either a LSRII or a FACSCanto (BD Biosciences) and analyzed using FlowJo (Treestar).

Intracellular cytokine staining. Aliquots of 1 × 10⁶ to 2 × 10⁶ cells splenocytes were placed in 96-well U-bottomed plates and restimulated with or without specific peptide (1 μg/mL) for 5 h at 37°C, 5% CO₂ in the presence of brefeldin A (BD Pharmingen) at concentration of 5 μg/mL. At the end of stimulation, cells were washed with fluorescence-activated cell sorting buffer (0.5% bovine serum albumin in PBS), blocked with Fc Block (BD Pharmingen) for 15 min at 4°C, and stained with surface antibodies at 4°C for 30 min. Cells were then washed and permeabilized with Cytofix/Cytoperm (BD Pharmingen) for 20 min at 4°C. Finally, the cells were stained with IFN-γ and tumor necrosis factor-α for 30 min at 4°C, washed in Perm/Wash buffer (BD Pharmingen), and resuspended in fluorescence-activated cell sorting buffer for further analysis. Antigen-specific T cells were identified as IFN-γ⁺ or IFN-γ⁺/tumor necrosis factor-α⁺.

Statistical analysis. Student's t tests were conducted using Microsoft Excel, and differences were considered significant at P < 0.05. Data are presented as mean ± SE. Differences in the survival of mice were analyzed using Kaplan-Meier method, and groups were compared using the log-rank test (GraphPad Prism 5).

Analyzing the efficacy of vaccination with AdhDCT in combination with surgical resection. We observed that, following surgical resection of bulky B16F10 melanomas [average size, 1,346.11 ± 109.10 mm³ (n = 20)], tumors relapsed at a rate of 50% to 100% depending on the experiment (Fig. 1B). To determine whether adjuvant vaccination with AdhDCT would improve disease-free survival, mice were immunized with AdhDCT 1 day after the surgery (Fig. 1A). Surprisingly, this strategy had no significant effect on disease-free survival (Fig. 1B). Because surgical intervention can temporarily induce a state of general immunosuppression (19, 20), mice were immunized 1 week after the surgery with either AdhDCT to allow time for recovery from the surgery. As a negative control, a group of mice were immunized with AdLCMV GP, which encodes an irrelevant antigen. We found that 75% of tumors recurred in both groups of mice regardless of the vector used (Fig. 1C), indicating that delaying adjuvant vaccination does not benefit disease-free survival.

We hypothesized that vaccine efficacy might be hindered by the emergence of the relapsed tumor. In this circumstance, neoadjuvant vaccination could be more beneficial, as there will be more time for the immune response to develop. Our previous experience showed that the T-cell response following immunization with rAd peaks between 8 and 12 days following immunization (17). Therefore, we immunized mice 9 days before surgery to coordinate the resection with the peak of the immune response to the vaccine. Strikingly, we observed that 21 of 22 mice that received neoadjuvant AdhDCT remained tumor-free following surgical resection (Fig. 1D). Surviving mice were also protected from a secondary lethal challenge with B16F10 cells 3 months after the surgery, showing long-term immunity (Supplementary Fig. S1A). By contrast, we observed that 60% of the tumors recurred in mice that received neoadjuvant vaccination with AdLCMV GP, confirming that tumor protection in this model was antigen-specific (Fig. 1D). Neoadjuvant vaccination with a rAd vector expressing human gp100 also failed to improve disease-free survival (Supplementary Fig. S1B), consistent with our previous observations that this vector does not provide effective protection against challenge with B16F0 (10, 11). Thus, to get the maximal benefit of combining vaccination with surgical resection, it is necessary to consider both timing and antigen selection.

To determine how the interval between immunization and surgery affects long-term survival, additional mice were immunized 6 and 3 days before surgery (Fig. 1E). In both cases, disease-free survival was significantly improved compared with untreated mice, although the effect does seem to diminish as the interval gets smaller. Nevertheless, there appears to be a substantial window of time in which neoadjuvant vaccination can be administered and provide significantly improved disease-free survival.

Because treatment of established tumors with AdhDCT vaccination does slow tumor growth (4), it is possible that decreased rate of tumor recurrence is related to the tumor size at the time of resection. On examination of all the mice that received neoadjuvant immunization (n = 42), we observed no relationship between tumor size and recurrence (Fig. 2). In fact, the few tumors that relapsed were smaller than the average tumor size in those groups. Furthermore, examination of the data from untreated mice and animals immunized with AdLCMV GP also failed to reveal a correlation between tumor size and recurrence (Fig. 2). Thus, the benefit of the neoadjuvant vaccination is not simply a result of delayed tumor growth but is due to the presence of enhanced antitumor immunity at the time of tumor resection.

Characterizing the effect of surgery on DCT-specific immunity. We have observed that tumor growth can suppress the magnitude of the immune response following AdhDCT immunization (4); therefore, surgical debulking may reverse tumor-induced immunosuppression and instigate a more robust immune response. To address this possibility, we examined the magnitude of the DCT-specific T-cell response in mice immunized with AdhDCT 9 days before surgery and compared the results to tumor-free and tumor-bearing animals immunized with the same vaccine but left without the surgery (Fig. 3). Similarly to our previous results (4), we observed that total numbers of CD8⁺ antigen-specific T cells were significantly reduced in the spleens of tumor-bearing mice compared with the tumor-free animals (Fig. 3B, days 10 and 14). However, there was no difference in the total numbers of DCT-specific CD8⁺ T cells between tumor-bearing mice and those who underwent resection (Fig. 3B,, compare closed columns with open columns). Similar results were obtained while analyzing DCT-specific CD8⁺ T cells isolated from vaccine-draining lymph nodes (Fig. 4A), suggesting that removal of the tumor does not reverse tumor-induced suppression when vaccination precedes the surgery and therefore cannot explain the superior protection in a neoadjuvant setting. Interestingly, examination of the TDLN revealed that significantly higher numbers of the DCT-specific CD8⁺ T cells were transiently detected in the resected mice compared with the other two groups of animals (Fig. 4B, day 10). To determine whether the observed increase in antigen-specific T cells was a consequence of tumor debulking or was a consequence of the surgery, tumor-free mice was vaccinated with AdhDCT and treated with a sham surgery. Similar to the observation in tumor-bearing animals, surgery produced an increase in DCT-specific CD8⁺ T cells in the draining lymph nodes compared with a matched group of mice that did have surgery (Fig. 4C), indicating that the elevation in DCT-specific T cells in the TDLN following surgery is driven by the surgical process rather than removal of the tumor. We have also examined DCT-specific CD4⁺ T cells in all of our samples but found no significant differences between the tumor-bearing mice with and without the surgery in any of the tissues examined (Supplementary Fig. S2).

We also examined changes in regulatory T cells following surgery. The frequencies of CD4⁺FoxP3⁺ regulatory T cells were decreased in the TDLN of the resected animals 1 day post-surgery compared with the other two groups (resected, 3.47 ± 0.17%; tumor-free, 3.92 ± 0.11%; tumor-bearing, 4.25 ± 0.23%; P < 0.05). However, this decrease in regulatory T-cell frequency was due to an overall increase in cellularity of the TDLN following surgery. The actual number of regulatory T cells in the resected mice was unchanged relative to the unresected mice (resected, 6.51 × 10⁵ ± 4.28 × 10⁴; tumor-bearing, 6.02 × 10⁵ ± 6.83 × 10⁴).

Based on these data, it appears that surgery is associated with modest immunologic effects. However, we did observe a localized increase in tumor-specific T cells that may facilitate the destruction of residual tumor cells.

Analyzing T-cell recruitment into the skin in AdhDCT-vaccinated mice. We observed that vitiligo developed in the vicinity of the incision and this pathology appeared more intense following neoadjuvant vaccination, suggesting that the immune response in the region of the tumor was enhanced when the vaccine was delivered before surgery (Supplementary Fig. S3). It should be noted that no vitiligo developed in the untreated mice or those that received AdLCMV GP showing that vitiligo was an antigen-driven process (Supplementary Fig. S3). To gain further insight into the events taking place within the surgical site, we investigated the presence of T cells in the skin around the incision. There was a dramatic ≥15-fold increase in the frequencies of CD8⁺ T cells and 4-fold increase in the CD4⁺ T-cell frequencies as a consequence of neoadjuvant vaccination with AdhDCT compared with the naive mice (Fig. 5). Similar T-cell infiltration of the skin was observed in tumor-bearing animals immunized with AdLCMV GP before surgery, indicating that the infiltration was not antigen-specific and likely due to increased migration of effector T cells to peripheral tissues (Fig. 5B,, compare closed and checkered columns). We also examined T-cell infiltration of the skin in tumor-free mice concurrently immunized with AdhDCT (Fig. 5B , gray columns). Frequencies of both CD8⁺ and CD4⁺ T cells were higher in the skin of immunized, tumor-free mice than in the skin of naive mice, consistent with the fact that, following immunization, effector T cells distribute throughout peripheral tissues. However, in immunized, tumor-bearing hosts, the frequencies of T cells in the skin were ∼3-fold higher, showing that the presence of tumor enhances local T-cell recruitment.

T-cell infiltration of the skin was further increased following the surgery in all mice (Fig. 5). Moreover, both CD8⁺ and CD4⁺ T cells remained elevated in the skin 3 months post-surgery in mice vaccinated with AdhDCT in a neoadjuvant setting (CD8⁺ T cells, 4.78 ± 0.55%; CD4⁺ T cells, 10.91 ± 2.80%). These results show that neoadjuvant immunization results in increased infiltration of T cells into the skin in the vicinity of the tumor, which is subsequently sustained for a long period following surgery. Therefore, in addition to increased frequencies of DCT-specific CD8⁺ T cells in the TDLN, mice that received neoadjuvant AdhDCT immunization also benefited from the immediate availability of effector T cells within the excision site.

As stated earlier, we hypothesized that neoadjuvant immunization would improve therapeutic outcome because elevated levels of circulating antigen-specific T cells would be available to clear residual tumor at the time of resection. To directly confirm this hypothesis, we tested tumor-infiltrating lymphocytes and skin-infiltrating lymphocytes for reactivity to DCT. Similar to our recent report (4), we measured high frequencies of DCT-specific CD8⁺ T cells within the tumor at the time of resection (Fig. 6A) in mice immunized with AdhDCT compared with control mice immunized with Ad LCMV GP. Furthermore, DCT-specific CD8⁺ T cells were found in the AdhDCT-immunized group at the time of surgery (Fig. 6A) and during the period following surgery (Fig. 6B), whereas no DCT-specific T cells were detected in the skins of mice vaccinated with AdLCMV GP.

Several recent preclinical studies have investigated the potential of vaccination in conjunction with surgical resection using various vaccination platforms (21–23). The overall conclusion was that multiple doses of the vaccine administered in the adjuvant setting were required in each case to provide protection from tumor recurrence. Only one of those reports considered initiating the vaccination before surgery (21). Those investigators did observe that commencing their immunization schedule before surgery was more effective than a similar strategy initiated after surgery (21). However, the “neoadjuvant” immunization protocol used by those authors also included two additional immunizations following surgery and the immunization schedule (every 14 days) was different from the adjuvant vaccination (every 7 days; ref. 21). Thus, it was not clear from those results whether neoadjuvant immunization truly provided any advantage. The present report shows clearly that neoadjuvant immunization provides superior protection compared with adjuvant immunization. The difference appears to result from the heightened frequency of tumor-specific T cells at the time of surgery and the increased infiltration of T cells into the tumor, the TDLN, and the area of resection. This method should be amenable to any vaccine platform and can be readily extended to the clinic.

No potential conflicts of interest were disclosed.

Grant support: Ontario Cancer Research Network and National Cancer Institute of Canada research studentship (N. Grinshtein). Y. Wan is a CIHR new investigator.

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.

We thank Carole Evelegh for preparing the viruses used in these experiments.