In spite of high expectations, cancer vaccines have found little applicability in oncology clinical practice [1]. It is likely that vaccines will be more efficacious at preventing relapse in patients with minimal residual disease status as opposed to advanced cases with bulky and immunosuppressive disease [1, 2].

Dendritic cell vaccines harness the antigen-presenting functions of these leukocyte subsets to induce antitumor CD8 and CD4 T cell responses [3]. The only exception of a FDA-approved cancer vaccine is a monocyte-derived dendritic cell product pulsed with a prostate serum antigen chimeric protein (Sipuleucel, Provenge) which was approved based on overall survival benefit [4, 5] for castration resistant metastatic prostate cancer. The nature of the antigen is likely to be important with current preference for antigens encoded by non-synonymous mutations or other gene alterations in the tumor genome [6]. Presentation of such antigens can be attained by pulsing defined peptide sequences following their genomic identification [7] or transfecting total tumor mRNA into the DCs [8]. A simpler alternative is the use of autologous tumor lysates to load the DC, which are known to immunize to at least to some extent to those relevant neoantigens [9–13].

We have developed a clinical-grade product based on autologous CD14+ monocytes differentiated to DC by means of 7-day cultures in the presence of GM-CSF and IL-4. Such cultures are incubated with autologous tumor lysate and activated by a cocktail of agents combining TNFα, poly-ICLC (Hiltonol^tm) and IFNα [12]. We have previously reported immunogenicity and safety of the approach [2], that showed intriguing long relapse-free survival in a series of resected glioblastoma cases [14].

Colon cancer frequently metastasize to the liver and in some cases, rescue surgery is feasible [15]. Various studies recommend neoadjuvant and adjuvant chemotherapy to reduce relapse risk [16–19] that in the best series still remains over 50–60%. Previous attempts have been made to use cancer vaccines to prevent relapse, including positive trials with Bacillus Calmette Guerin (BCG) mixed with autologous irradiated tumor cells from the resection specimen. This preventive treatment to avoid relapse was effective in stage II colon carcinoma but not in stage III or IV [20].

In this disease scenario of surgically amenable colon cancer with liver metastases, minimal residual disease is expected following surgery plus chemotherapy. We have tested our previously reported DC formulations loaded with autologous tumor lysates [12] in a randomized fashion. Even if the trial had to be halted with only 15 randomized patients, our follow-up observation strongly suggests a beneficial effect of the vaccination scheme on disease-free survival (DFS).

We have previously reported that a cell therapy product consisting of monocyte-derived DC loaded with heat-treated autologous tumor lysate and matured with a cocktail of poly-ICLC, TNFα and IFNα was immunogenic [11].

An unmet clinical setting in which vaccines would be very useful is the prevention of relapse following potentially curative resections of colon adenocarcinoma liver metastases. A randomized clinical was designed to address a putative beneficial effect of this treatment versus standard-of-care at our institutions.

The original design of this academically sponsored trial (NCT01348256) included 56 patients but budget restrictions forced an early termination of recruitment, when only 19 patients had signed informed consent.

As indicated in Fig. 1a, three of the patients were excluded for evaluation due to positive resection margins (>R0) following neoadjuvant chemotherapy and surgery. One patient withdrew informed consent after neoadjuvant treatment, and thus was not randomized. Neoadjuvant chemotherapy consisted of standard chemotherapy 5-FU-based cycles (Table 1) .Fig1Fig. 1

Clinical trial design and treatment generation scheme. a Graphical summary of patient accrual, randomization and treatment. b Schematic representation of production of dendritic-cell vaccines loaded with autologous tumor lysate in patients undergoing complete surgical resection of colon cancer liver metastasis

Randomized evidence for efficacy of dendritic cell vaccination for cancer is yet missing. In cutaneous melanoma, DC pulsed with shared antigen peptides did not show evidence for benefit [25]. However, important adjuvant dendritic cell vaccination randomized trials for skin melanoma (NCT02993315) and uveal melanoma [26] are ongoing.

Our study chose resectable metastatic colon cancer because is an unmet clinical need in patients with potential minimal residual disease but at very high risk of relapse. Unfortunately, logistical/budgetary problems forced the trial to be halted with a short number of patients randomized to each of the two arms. However, these patients were kept in follow-up according to protocol and relapses evaluated as planned.

With all caveats due to patient sample size, current results are strongly supportive of a beneficial effect of DC vaccination that would warrant a confirmatory randomized clinical trial in a similar setting. It is worth mentioning that unbalanced presence of higher-risk of relapse patients (according to clinical scores) in the vaccination arm further supports the existence some degree of relapse-free survival benefit in the vaccinated arm. Reinforcing this notion, risk scales of relapse would have been predictive of higher probability of relapse in the vaccination arm. In this regard, there was also an unbalance in MSI-H status with three cases in the vaccination versus none in the observation arm. However, two of the MSI-H cases actually relapsed. This is important since MSI-H is related to less metastatic progression following surgery of the primary tumor [27] and higher antigenicity and more susceptibility to PD-1 blocking agents at metastatic stage [28]. However, once metastatic MSI-H patients show a similar progression pattern as non-MSI cases [29] as it is the case in our small series of patients.

An important question would be whether to go on with the same vaccination schedule or to provide some more boosting cycles of vaccination during the first and second year following surgery. The impression is that the vaccines may have prevented relapse in some patients, while in others vaccination might only have delayed relapse, thus advocating for extended boosting doses.

The quality and maturation of the DC product was good in relapsed and non-relapsed cases following treatment, particularly showing prominent IL-12 production. A tendency to more IL-8 production and brighter surface expression of CD11c were noted in relapsed cases, a finding that warrants confirmatory research in ongoing and subsequent clinical studies.

Other points to be considered would be to add check-point inhibitors circa vaccination dates [30] and to monitor patients for immunization against neo-antigens by IFNγ ELISPOT [12] or other techniques. Our series of IFNγ-ELISPOT assays is not sufficiently large to draw any conclusions but suggests interesting individual heterogeneity in the measurable response 4 weeks after the last vaccine.

There is recently published evidence supportive for checkpoint inhibitor immunotherapy approaches given in neoadjuvant schemes prior to surgery [31–33]. In our opinion, it makes sense to use anti PD-1 or PD-L1 mAbs prior to surgery as part of the neoadjuvant regimen. In this setting, postsurgical DC vaccinations would boost an already unleashed T-cell response. However, it must be taken into account that metastatic non-MSI colon cancer is mostly refractory to checkpoint inhibitors.

All considered, in spite of the reduced number of cases, results are very encouraging in favor the vaccination arm and indicative of the potential of DC vaccinations with autologous tumor antigens for patients with colon cancer liver metastasis amenable to complete surgical resections.