Research ArticleCancer

Vaccination during myeloid cell depletion by cancer chemotherapy fosters robust T cell responses

See allHide authors and affiliations

Science Translational Medicine  13 Apr 2016:
Vol. 8, Issue 334, pp. 334ra52
DOI: 10.1126/scitranslmed.aad8307

Vaccinating cancer away

Cervical cancer, a common killer of women worldwide, is most often caused by human papillomavirus type 16 (HPV16). Although a vaccine targeting this virus is available and very effective at preventing cervical cancer, it does not work once cancer is already established, and advanced cervical cancer is very difficult to treat. Welters et al. have developed a method of therapeutic vaccination, where they synthesize long peptides mimicking key oncogenic proteins from HPV16 and use them to treat patients. Although it is too early to tell how the new vaccine will affect patient survival, combining it with chemotherapy helped strengthen patients’ immune responses against the cancer, so it is a promising candidate for further clinical development.


Therapeutic vaccination with human papillomavirus type 16 synthetic long peptides (HPV16-SLPs) results in T cell–mediated regression of HPV16-induced premalignant lesions but fails to install clinically effective immunity in patients with HPV16-positive cervical cancer. We explored whether HPV16-SLP vaccination can be combined with standard carboplatin and paclitaxel chemotherapy to improve immunity and which time point would be optimal for vaccination. This was studied in the HPV16 E6/E7–positive TC-1 mouse tumor model and in patients with advanced cervical cancer. In mice and patients, the presence of a progressing tumor was associated with abnormal frequencies of circulating myeloid cells. Treatment of TC-1–bearing mice with chemotherapy and therapeutic vaccination resulted in superior survival and was directly related to a chemotherapy-mediated altered composition of the myeloid cell population in the blood and tumor. Chemotherapy had no effect on tumor-specific T cell responses. In advanced cervical cancer patients, carboplatin-paclitaxel also normalized the abnormal numbers of circulating myeloid cells, and this was associated with increased T cell reactivity to recall antigens. The effect was most pronounced starting 2 weeks after the second cycle of chemotherapy, providing an optimal immunological window for vaccination. This was validated with a single dose of HPV16-SLP vaccine given in this time window. The resulting proliferative HPV16-specific T cell responses were unusually strong and were retained after all cycles of chemotherapy. In conclusion, carboplatin-paclitaxel therapy fosters vigorous vaccine-induced T cell responses when vaccination is given after chemotherapy and has reset the tumor-induced abnormal myeloid cell composition to normal values.


Most cervical cancers are induced by human papillomavirus type 16 (HPV16) (1). Up to 70% of the advanced cancers relapse (2, 3). One of the preferred treatments for patients with recurrent, metastatic, or advanced cervical carcinoma is the combination of carboplatin with paclitaxel (CarboTaxol) (4), but this is rarely curative (5).

The two HPV16-encoded oncoproteins E6 and E7 are required for the transformation of epithelial cells (6) and constitute excellent targets for the immune system. HPV16-specific T cell reactivity is frequently detected in healthy individuals but usually not in patients with premalignant anogenital lesions or cancer (7). Installment of robust HPV16-specific immunity by vaccination with therapeutic HPV16 overlapping synthetic long peptides (HPV16-SLPs) admixed with Montanide ISA-51 resulted in regressions of HPV16-induced premalignant lesions of the vulva in two independent studies (810). In contrast, therapeutic vaccination of patients with advanced or recurrent HPV16-positive cervical cancer partly installed HPV16-specific T cell reactivity, particularly in patients with a less suppressed immune status, but had no clinical effect (11).

Chemotherapeutic agents act on cancer cells (12), but many of them mediate part of their therapeutic effects through immune mechanisms (13, 14). In murine models, the combination of chemotherapy with activation of T cells resulted in improved treatment of tumors (1316). Therefore, we investigated whether CarboTaxol could be successfully combined with HPV16-SLP vaccination, first in a mouse model (13) and then in an open-label observational study with cervical cancer patients.


Combined chemoimmunotherapy improves the eradication of HPV16-positive tumors in mice

To test the effects of CarboTaxol with HPV16-SLP vaccination, HPV16 E6– and E7–positive TC-1 tumor-bearing mice were treated when tumors were palpable at day 8 (~4 mm2; Fig. 1A). CarboTaxol had little effect on tumor growth, whereas vaccination induced a temporary decrease in tumor size (Fig. 1, B and C). The combined treatment had the strongest antitumor effect (Fig. 1, B and C). None of the treatments affected the percentages of circulating CD8+ and CD4+ T cells (Fig. 1, D and E, and fig. S1, A and B). Vaccination induced HPV16-specific CD8+ T cells, and this was not influenced by co-treatment with CarboTaxol (Fig. 1F and fig. S1C).

Fig. 1. CarboTaxol improves the clinical outcome of therapeutic peptide vaccination.

(A) C57BL/6 mice were injected with 1 × 105 TC-1 tumor cells and treated systemically with carboplatin (C) and paclitaxel (P) with or without injection of the HPV16 E743–77 peptide in Montanide vaccine (V) in the flank opposite of the tumor as shown in the schematic diagram. (B) Kaplan-Meier survival plots show the combined data from several experiments (number of mice is indicated). Peptide-treated versus peptide-CarboTaxol–treated group (P = 0.004). (C) Tumor growth data from two pooled individual experiments with eight mice per group. (D to F) Quantification of the percentage of (D) CD4+ T cells, (E) CD8+ T cells, and (F) the vaccine-specific CD8+ T cells as determined by H2-Db E749–57 (RAHYNIVTF) tetramer (TM) staining. Column 3 versus columns 1 (P = 0.03), and 2 and 4 (P = 0.005). Column 5 versus columns 2 and 4 (P = 0.03). n = 8 mice in the tumor-bearing groups; n = 4 in the naïve group; data are representative of two individual experiments and expressed as means + SEM. Data were analyzed by one-way analysis of variance (ANOVA) followed by Tukey’s post hoc analysis.

CarboTaxol treatment alters circulating and intratumoral myeloid cell populations

To understand the mechanism underlying these improved therapeutic outcomes, immune cells in blood and tumors were analyzed 3 to 4 days after CarboTaxol treatment (and/or 9 to 10 days after peptide vaccination) (Fig. 2A), which is at the start of the tumor regression phase (fig. S2). In untreated tumor-bearing mice, the percentage of circulating myeloid cells increased (fig. S3A) because of the increase in circulating CD11bhi cells, in particular, CD11bhiGr-1hi cells. However, their numbers decreased markedly in tumor-bearing animals treated with CarboTaxol (Fig. 2, B and C, and fig. S3). The frequencies of CD4+ and CD8+ T cells, antigen-specific CD8+ T cells, monocytes, and dendritic cells in the blood were not affected by CarboTaxol treatment (fig. S3). Thus, CarboTaxol treatment normalized the myeloid cell populations in the blood of tumor-bearing mice, making them more similar to those of naïve mice. This effect could not be ascribed to one individual chemotherapeutic compound because the effect on circulating CD11bhi cells was particularly pronounced in animals treated with both compounds (fig. S4).

Fig. 2. Chemotherapy normalizes systemic tumor-induced myeloid subsets, but intratumoral T cells are not affected.

(A) C57BL/6 mice were injected with TC-1 tumor cells (TC-1) and treated with HPV16 E743–77 peptide in Montanide vaccine (V) in the flank opposite of the tumor and with carboplatin (C) and paclitaxel (P) as indicated in the schematic diagram. B and T indicate the time points for blood and tumor analysis, respectively. (B) Flow cytometry analysis of the total percentage of myeloid cells [left; column 2 versus columns 1 (P = 0.007), 4 (P = 0.003), and 5 (P < 0.0001)], the CD11bhiGr-1hi cells [middle; column 2 versus columns 1 (P = 0.0004), 3 (P = 0.03), and 4 and 5 (P < 0.0001); column 3 versus columns 4 (P = 0.02) and 5 (P = 0.002)], and the CD11bhiGr-1int (right) cells in the blood. (C) Representative flow cytometry plots for each treatment gated on live (7AAD) cells (top). Distribution of Gr-1hi–, Gr-1int–, and Gr-1neg–expressing cells within the total CD11bhi population is indicated in the pie charts (bottom). (D to F) Tumor samples were collected and analyzed to determine the percentage of (D) CD45+ cells within the live gate [column 1 versus columns 2 and 4 (P < 0.0001) and 3 (P = 0.02); column 3 versus columns 2 and 4 (P < 0.0001)], (E) CD8+ T cells in the leukocyte gate [column 1 versus columns 2 and 4 (P < 0.0001); column 3 versus columns 2 and 4 (P < 0.0001)], (F) and vaccine-specific T cells determined by H2-Db E749–57 (RAHYNIVTF) tetramer staining [column 1 versus columns 2 and 4 (P < 0.0001); column 3 versus columns 2 and 4 (P < 0.0001)]. (G) Single cell suspensions of tumors were co-incubated with HPV16 E743–77 peptide–pulsed D1 cells and stained for intracellular tumor necrosis factor–α (TNFα) and interferon-γ (IFN-γ). Representative flow cytometry plots (left) and quantification (right) show the frequency of cytokine-producing CD8+ T cells [IFN-γ graph: column 1 versus columns 2 (P = 0.009) and 4 (P = 0.0001); column 3 versus columns 2 (P = 0.004) and 4 (P < 0.0001). TNFα graph: column 1 versus columns 2 (P = 0.006) and 4 (P = 0.0009); column 3 versus columns 2 (P = 0.008) and 4 (P = 0.001)]. n = 5 to 7 mice per group; data shown are representative of two individual experiments. Data are expressed as means + SEM and analyzed by one-way ANOVA followed by Tukey’s post hoc analysis.

Next, we assessed the effects of CarboTaxol-vaccine combination treatment on the tumor microenvironment (13). The percentage of intratumoral leukocytes increased upon treatment with CarboTaxol and/or vaccine (Fig. 2D). In vaccinated mice, a markedly high percentage of these leukocytes were CD8+ T cells (Fig. 2E), half of which were vaccine-specific (Fig. 2F) and capable of producing IFN-γ and TNFα (Fig. 2G). There was no direct effect of CarboTaxol treatment on the presence and function of these lymphocytes.

We then focused on the intratumoral CD11bhi myeloid cells because the Gr-1hi subtype of this cell population was increased in the blood of untreated tumor-bearing mice. The Gr-1hi cells in the tumors strongly expressed the granulocytic marker Ly6G and decreased amounts of the macrophage marker F4/80 and the dendritic cell marker CD11c. In contrast, Gr-1int cells had a higher expression of F4/80, CD11c, CD80, CD86, and major histocompatibility complex class II, but not Ly6G (Fig. 3A), suggesting a superior immune stimulatory capacity. Treatment with either CarboTaxol or vaccine resulted in a predominance of the CD11bhiGr-1int population over the Gr-1hi population (Fig. 3, B and C). Together, these data demonstrate that treatment of tumor-bearing mice with CarboTaxol results in a relative loss of myeloid cell–associated immunosuppression in both tumor and blood.

Fig. 3. Gr-1hi cells are depleted from the tumor by CarboTaxol treatment.

Mice were treated as in Fig. 2. Tumor samples were isolated and analyzed by flow cytometry. (A) Leukocytes from resected untreated tumors were analyzed for the expression of Gr-1 and CD11b. The histograms show the expression of class II, CD80, CD86, F4/80, Ly6G, and CD11c on the Gr-1hi (gray lines) and Gr-1int (black lines) subsets. (B) Four days after chemotherapy or 10 days after peptide vaccination, leukocytes from resected tumors were analyzed for the expression of Gr-1 and CD11b (top). Distribution of Gr-1hi–, Gr-1int–, and Gr-1neg–expressing cells within the total CD11bhi population is shown in the pie charts (bottom). (C) Percentages (mean + SEM) of Gr-1hi [column 1 versus columns 2 (P = 0.006), 3 (P = 0.0004), and 4 (P = 0.0008)] and Gr-1int [column 4 versus columns 1 (P = 0.0006) and 2 (P = 0.02)] subsets were analyzed for untreated and treated tumors. n = 5 to 7 mice per group; data shown are representative of two individual experiments. Data were analyzed by one-way ANOVA followed by Tukey’s post hoc analysis.

A specific time window is associated with increased immunity in patients on chemotherapy

On the basis of the above observations, we performed a study in patients with advanced, recurrent, or metastatic cervical carcinoma. The trial was designed to study the impact of chemotherapy on vaccine-induced immunity, and therefore patients were not required to have an HPV16+ tumor. Patients were screened between January 2011 and January 2013 in four Dutch hospitals, and their characteristics are listed in table S1. In the first cohort of six patients, the number and function of lymphoid and myeloid cells were studied in blood samples taken at different time points during and after CarboTaxol treatment (Fig. 4A). CarboTaxol treatment was associated with a decrease in the otherwise high frequency of myeloid cells (median of 32% at baseline), which reached its nadir at 1 to 2 weeks after the second chemotherapy cycle (median of 6% at 1 to 2 weeks after chemotherapy cycle 2; Fig. 4, B and C) and coincided with an increase in the percentages of lymphoid cells (Fig. 4D). Although the relative frequencies of CD4+ and CD8+ T cells (fig. S5) remained unchanged, T cell function was improved, as evidenced by the increase in their proliferation against a bacterial recall antigen mixture [memory response mix (MRM)] in the same time window (Fig. 4E). T cell responses to phytohemagglutinin (PHA) stimulation were strong at all time points, indicating that there were no intrinsic problems with the T cells’ response to mitogens (fig. S5). The capacity of antigen-presenting cells (APCs) to stimulate allogeneic T cell proliferation was slightly improved (Fig. 4F). Thus, the observations in mice are mirrored by the findings in patients. Furthermore, the results revealed a specific time window throughout CarboTaxol treatment, during which antigen-specific T cell responses were optimal. This time window, starting at 1 to 2 weeks after the second cycle of CarboTaxol, appeared attractive for the generation of strong T cell responses by vaccination. We used this observation to select the time window for the application of a single dose of vaccine in the second patient cohort.

Fig. 4. CarboTaxol induces changes in cellular immunity in advanced-stage cervical cancer patients.

(A) Blood draws (B) and CarboTaxol cycles (C) for the six patients in cohort 1 are indicated in days (D) and weeks (W) in the schematic outline. (B) Representative flow cytometry plots show the myeloid cell gate and lymphocyte gate in the blood of a patient at baseline and after one to two CarboTaxol cycles in comparison to the blood of a healthy donor. The percentages of myeloid cells and lymphoid cells within the total number of cells are indicated. (C and D) To determine the relative percentage of each population, the sum of the events in the lymphoid and myeloid cell gates in the forward and side scatter plots was set to 100, and then the frequency of (C) myeloid cells [column 1 versus columns 4 (P = 0.0002), 5 (P = 0.02), and 6 (P = 0.005)] and of (D) lymphocytes [column 1 versus columns 4 (P = 0.0002), 5 (P = 0.02), and 6 (P = 0.005)] was determined. (E) Proliferation of T cells upon recognition of recall antigens (MRM) shown as stimulation index. Column 1 versus columns 3 and 6 (P = 0.02), 4 (P = 0.001), and 5 (P = 0.005). (F) The ability of APCs to stimulate T cells in a mixed lymphocyte reaction (MLR) shown for the four tested patients. Column 1 versus columns 2 (P = 0.005) and 4 (P = 0.049). Data (shown as median + interquartile range) were analyzed by repeated-measures model.

CarboTaxol mediates normalization of circulating immune cell frequencies

The second cohort consisted of 13 patients (table S1). One patient (ID6002) died of progressive disease before vaccination and was substituted by ID6102. Compared to 19 healthy donors, the patients from both cohorts displayed an increased frequency of circulating myeloid cells before chemotherapy (fig. S6A), confirming that the progressive tumor growth–induced myeloid changes in mice are mirrored in patients with advanced cervical cancer (Fig. 4). Throughout the CarboTaxol treatment, the absolute numbers of lymphocytes remained similar (Fig. 5A), but the absolute number of circulating leukocytes was strongly reduced (median, −4.7 × 109/liter) as measured by leukocyte differentiation analyses (Fig. 5B). This reduction reached its nadir after two cycles of chemotherapy and was retained during the remainder of the chemotherapy cycles. Flow cytometry analysis again revealed a decrease in myeloid (CD45+CD3CD19) and a relative increase in lymphoid (CD45+CD3+CD19) cells (fig. S6, B and C). The frequency of these populations almost normalized to the levels observed in healthy donors (Fig. 5, C and D), and this correlated with an increased T cell responsiveness to bacterial (MRM; Fig. 5E) and viral antigens (FLU; Fig. 5F). Similar to the first cohort, the blood samples of cohort 2 showed no overt changes in APC function or the response to PHA stimulation (fig. S5).

Fig. 5. The frequencies of circulating immune cells normalize upon CarboTaxol treatment.

(A and B) Blood samples of the 12 patients in cohort 2 were analyzed for leukocyte differentiation, showing a shift from baseline for the counts of (A) lymphocytes and (B) leukocytes × 109/liter. Column 1 versus columns 3 (P = 0.007) and 4 to 6 (P < 0.0001). Column 3 versus columns 4 (P < 0.0001), 5 (P = 0.04), and 6 (P = 0.003). NT, not tested. (C and D) The frequency of (C) myeloid cells [column 1 versus columns 2 (P < 0.0001), 3 (P = 0.002), 4 (P = 0.003), 5 (P = 0.006), 6 (P = 0.008), and 7 (P = 0.009); column 2 versus columns 3 (P = 0.04), 4 (P = 0.02), 5 (P = 0.007), and 6 and 7 (P = 0.004)] and (D) lymphocytes [column 1 versus columns 2 (P < 0.0001), 3 (P = 0.0004), 4 (P = 0.0009), 5 (P = 0.002), 6 (P = 0.005), and 7 (P = 0.006); column 2 versus columns 3 (P = 0.02), 4 (P = 0.002), 5 (P = 0.0009), and 6 and 7 (P = 0.0002)] were determined in the forward and side scatter plots of these blood samples after acquisition by flow cytometry. Blood samples from healthy donors (HD; n = 19) were included for comparison. Data (shown as median + interquartile range) were analyzed by repeated-measures model. (E and F) The fold change in stimulation index (SI), which is SI in a sample during/after chemotherapy divided by that of the baseline sample, of the blood samples stimulated with (E) recall antigens (MRM) or (F) influenza matrix 1 peptides (FLU) is shown versus the shift in percentage of myeloid or lymphoid cells from baseline. Repeated-measures regression analysis was conducted to determine whether there is a slope significantly different from 0, represented with the P value.

Further dissection of the changes within the myeloid (CD45+CD3CD19CD1a) cell population was performed on the basis of HLA-DR (human leukocyte antigen–DR) expression to distinguish macrophages and dendritic cells (HLA-DR+) from myeloid-derived suppressor cells (MDSCs; HLA-DR−/low) and further subdivide them on the expression of CD14 and CD11b within the HLA-DR+ myeloid cell population (fig. S7). Of the five identified subpopulations, population 1 (CD14+CD11b+) and population 2 (CD14intCD11bint) were most abundant and increased before CarboTaxol treatment when compared to healthy donors (Fig. 6, A and B, and fig. S8). The other three populations each constituted 0.2 to 2.6% of the myeloid cell fraction. During chemotherapy, the frequencies of populations 1, 3, and 5 dropped (Fig. 6B and fig. S8). The treatment-induced decrease in population 1 coincided with improved T cell reactivity against MRM and FLU (Fig. 6C). Extended analysis of the various subsets by flow cytometry (fig. S7) revealed that population 1 was composed of M1 monocytes/macrophages (CD45+CD3CD19CD1aHLA-DR+CD14+CD11b+CD206CD163CD16CD11c+) and M2c monocytes/macrophages (CD45+CD3CD19CD1aHLA-DR+CD14+CD11b+CD206CD163+CD16CD11c+). The frequency of both subpopulations was increased in patients but normalized upon treatment (Fig. 6, D and E). The frequency of 10 distinct circulating MDSC populations (17) was not different between patients and healthy donors. Only the main MDSC population (CD45+CD3CD19CD1aHLA-DRlow) displayed a slight decrease during chemotherapy (Fig. 6F and fig. S7).

Fig. 6. CarboTaxol treatment induces a decline in all subsets of circulating myeloid cells.

Peripheral blood mononuclear cells (PBMCs) from the 12 patients in cohort 2 were subjected to multiparameter flow cytometry analysis. (A) Representative dot plots of the five subpopulations within the CD45+CD3CD19CD1aHLA-DR+ population defined by expression of CD11b and CD14 in the baseline blood sample and 2 weeks after the second cycle of chemotherapy. (B) The frequency of (CD45+CD3CD19CD1aHLA-DR+) CD11b+CD14+ (population 1) as a percentage of the CD45+ cells in healthy donors (HD; n = 19) and in the patients over time. The time points of blood sampling (x axis) were 2 weeks after the first (1st), second (2nd), and third (3rd) cycle of chemotherapy and 3 weeks after the sixth (6th) or last chemotherapy cycle. Column 1 versus columns 2 (P < 0.0001), 3 (P = 0.01), 4 (P = 0.02), 5 (P = 0.009), and 6 (P = 0.04). Column 2 versus columns 3 (P = 0.04), 4 (P = 0.01), 5 (P = 0.02), and 6 (P = 0.005). (C) The fold change in stimulation index of the blood samples stimulated with recall antigens (MRM) or influenza matrix 1 peptides (FLU) is shown versus the absolute shift in percentage of population 1 cells from baseline. Repeated-measures regression analysis was conducted to determine whether there is a slope significantly different from 0, represented with the P value. (D and E) The frequencies of (D) CD163CD206CD16CD11c+ [M1-like cells; column 1 versus columns 2 (P = 0.001) and 3 (P = 0.01); column 2 versus column 5 (P = 0.01)] and (E) CD163+CD206CD16CD11c+ [M2c-like cells; column 1 versus columns 2 (P = 0.004) and 5 (P = 0.01)] within population 1 are shown for the healthy donors and patients over time. (F) The frequency of MDSCs (CD45+CD3CD19-CD1aHLA-DR−/low) is depicted. Column 1 versus column 3 (P = 0.03). Column 2 versus column 4 (P = 0.04). Data (shown as median + interquartile range) were analyzed by repeated-measures model.

Analysis of the T cell populations (fig. S7) revealed no changes in CD4+ and CD8+ T cell frequencies (fig. S9, A and B), confirming our findings in mice. The frequency of TIM3 (T cell immunoglobulin domain and mucin domain-3) and/or PD-1 (programmed cell death protein 1)–expressing CD4+ or CD8+ T cells (fig. S9, C and D) and CD4+CD25+CD127Foxp3+ regulatory T cells (fig. S9E) was higher in patients when compared to healthy controls. The percentage of CD4+TIM3+PD1 and that of regulatory T cells slightly decreased during chemotherapy (fig. S9, C and E).

Together, these results showed that CarboTaxol treatment strongly affected myeloid cells but not lymphocytes. CarboTaxol treatment normalized the amounts of different myeloid cell populations found to be increased in the blood of cervical cancer patients. This normalization of myeloid cell numbers coincided with improved T cell reactivity to antigens from common pathogens, suggesting a relief from general immunosuppression.

Timed vaccination during chemotherapy results in a strong and sustained HPV16-specific T cell response

Twelve patients received a single vaccination with the HPV16-SLP vaccine (8, 9, 11) at 2 weeks (13 to 17 days) after the second (n = 11) or third cycle of chemotherapy (n = 1; ID6008). None of the patients had a demonstrable preexisting response to HPV16 E6/E7. Vaccination with the HPV16-SLP vaccine induced proliferative T cell responses in 11 patients (Fig. 7A). The median stimulation index to all six peptide pools was 25.0 (range, 4.3 to 133.4) at 3 weeks after vaccination in these responders. Vaccine-induced HPV16-specific proliferation was retained after six cycles of chemotherapy and even increased in some cases (median, 21.0; range, 5.0 to 141.5; Fig. 7A, black bars versus gray bars). The vaccine-specific proliferative T cell response in the seven HPV16+ patients was not statistically higher than in the other patients (Fig. 7B). For six patients, enough PBMCs were available to analyze the vaccine-induced T cell response by intracellular staining for IFN-γ, IL-2 (interleukin-2), and TNFα. A polyfunctional cytokine response to HPV16 E6 was measured in five and to E7 in four of the six patients. One patient (ID6004) was anergic (fig. S10), confirming the results of the proliferation assay (Fig. 7A).

Fig. 7. HPV16-SLP vaccination during CarboTaxol treatment results in a strong immune response in patients.

The patients in cohort 2 received a single vaccination with HPV16-SLP subcutaneously at 2 weeks after the second cycle of chemotherapy. (A) The proliferative responses of T cells in the lymphocyte stimulation test (LST) are shown as a stimulation index and depicted versus the indicated peptide pools used for stimulation of the cells in the blood sample at baseline (hatched bar), 2 weeks after the second cycle of chemotherapy and before vaccination (white bar), 3 weeks after this single vaccination (black bar), and 3 weeks after the sixth or last cycle of chemotherapy (gray bar). (B) The patients are grouped by HPV16 status (HPV16+, n = 7; HPV16, n = 5), and the proliferative response (stimulation index) is plotted versus the indicated blood samples. Each dot represents one response against HPV16 E6 and E7. In total, six peptide pools were tested per blood sample. Data were analyzed by linear mixed-model analysis and showed no statistically significant difference.

Previously, patients with recurrent HPV16+ cervical cancer were vaccinated at least 1 month after chemotherapy (11). In comparison to the responses seen during the earlier trial, the proliferative responses obtained by vaccination during chemotherapy were of far greater magnitude (fig. S11).

Myeloid cell depletion improves the response of PBMCs to stimulation in vitro

To recapitulate in vitro the association between a reduced myeloid cell population and improved T cell reactivity to recall antigens and HPV16 vaccination, we depleted myeloid cells from the PBMCs of two patients displaying relatively high frequencies of myeloid cells before chemotherapy and stimulated these PBMCs with autologous monocytes pulsed with a mix of recall antigens, a mix of E6 and E7 peptides, or a mix of p53 peptides as control for 11 days before the antigen-specific T cell response was tested. As a control, we used nondepleted PBMCs. Not only was the T cell response to recall antigens much higher in the culture started with myeloid cell–depleted PBMCs, but the HPV16-specific response was also more efficiently boosted during these 11 days. As expected, no reactivity was detected in the cultures stimulated with the control p53 peptides (fig. S12).

Combination of chemotherapy with vaccination is safe in advanced cervical cancer patients

Safety was assessed according to the Common Terminology Criteria for Adverse Events (CTCAE) v3.0. Most of the observed adverse events (AEs) were disease-related or chemotherapy-related. All patients developed chemotherapy-related anemia, thrombocytopenia, leucopenia, neutropenia, and alopecia. There were seven AEs, all in different patients, related to the advanced stage of the disease: shortness of breath, pulmonary embolism, abdominal pain (lymphedema), gastroenteritis, erysipelas, hydronephrosis. One patient (ID6002) died before vaccination could take place, and one patient (ID6004) died 11 weeks after vaccination. The cause of death in both cases was progressive disease. Vaccine-related AEs were largely localized to the vaccination site (table S2). One patient developed an ulcer at the injection site, which persisted for more than 6 weeks and required antibiotic treatment.


Here, we observed that tumors expressing the HPV16 oncoproteins E6 and E7 cause the numbers of circulating myeloid cells to be abnormally high in TC-1–challenged mice and in HPV-positive cervical cancer patients. Treatment with CarboTaxol normalizes the numbers of circulating myeloid cells but has no negative effect on the number and function of lymphocytes. In mice, CarboTaxol treatment had a similar effect on the myeloid cell composition within the tumors as in the blood. The effects of CarboTaxol are, therefore, not limited to circulating immune cells, and it is likely that similar effects occur within the tumors of cervical cancer patients. The CarboTaxol-mediated normalization of circulating myeloid cells was associated with increased T cell–mediated tumor control in mice and with higher T cell reactivity against common microbial recall antigens and response to HPV16-SLP vaccination in patients. This suggests a causal relationship between the normalization of abnormally high myeloid cell frequencies and improved T cell responsiveness. Because the combination of HPV16-SLP vaccination plus CarboTaxol improved the cure rate of mice with established TC-1 tumors, we expect that the robust and sustained HPV16-specific T cell responses seen with this combination improve the efficacy of treatment in patients with advanced cervical cancer. This needs to be studied in a future randomized clinical trial.

CarboTaxol is a standard chemotherapeutic treatment not only in cervical cancer but also in patients with other cancer types, including lung cancer and ovarian cancer. Its effect on the immune system, however, has not been widely studied. Carboplatin and paclitaxel are both known to cause dose-limiting myelotoxicity (18, 19). This is likely a direct effect on precursor cells in the bone marrow, as observed in different animal models (20). White bone marrow cells display impaired in vitro capacity to proliferate when treated with carboplatin (21). Furthermore, the number of myeloid cells reaching its nadir at 2 weeks and a rebound at 3 weeks after CarboTaxol treatment is in line with the mechanistic models for the development and maturation of leukocytes and drug susceptibility in the bone marrow (22, 23). Lymphopenia has not been reported. We observed an increase in T cell reactivity 1 to 2 weeks after the second and subsequent cycles of chemotherapy. This was not a result of changes in absolute lymphocyte counts or strong alterations in the number or phenotype of CD4+, CD8+, or regulatory T cells. Similar observations were made in ovarian cancer patients. Those patients who responded to CarboTaxol displayed a stronger IFN-γ–producing CD8+ T cell response during treatment 12 to 14 days after chemotherapy (24, 25). Here, we show that the positive effect of CarboTaxol on the immune response results from the normalization of abnormal myeloid cell numbers, which are initially high in the presence of larger tumor burden. Leukocytosis has been described in patients and animals with HPV-associated cancers (26, 27), but the composition of the increased leukocyte populations was not analyzed in detail. An in-depth analysis of the myeloid cell subsets affected by CarboTaxol revealed that these effects were found across all subsets that are elevated in patients or in tumor-bearing animals. This includes tumor growth–suppressing myeloid cells, but more importantly the tumor-promoting myeloid cell populations, which can suppress the function of antitumor effector T cells. Apparently, the balance among these subsets and in particular the decline in immunosuppressive myeloid cells within the tumor microenvironment appears to be important for successful implementation of immunotherapy and improved clinical efficacy. The change in the proportions of myeloid cells and lymphocytes allowed the latter population to respond to antigenic stimulation, most likely through a relief from myeloid cell–mediated immunosuppression. This notion is sustained by the unexpectedly high proliferative responses after timed application of a single vaccination and our in vitro experiment showing that removal of excessive CD14+ myeloid cells from prechemotherapy PBMC samples of two cancer patients allowed the tumor-specific T cells to react to antigenic stimulation. This seems to be a general phenomenon, and we observed this also in the context of lung cancer (28). A recent phase II trial in patients with extensive small-cell lung cancer reported that ipilimumab treatment beginning with the third cycle of CarboTaxol produced better clinical outcomes than giving the drugs during cycles 1 to 4 (29). The effect of CarboTaxol on myeloid cells may have relieved myeloid cell–mediated suppression of T cells, as in our study, providing ipilimumab the opportunity to release the brakes on activated T cells in the later phase of treatment. The effects of CarboTaxol on myeloid cells are clear in patients with cancers where myeloid cells have prognostic value (30, 31), of which cervical and ovarian carcinomas are prime examples. Other types of cancer in which myeloid cells play an important immunosuppressive and prognostic role are thus also candidates for timed immunotherapy.

Our study has some limitations. First, although abnormal numbers of myeloid cells are found both in the mouse model and in patients, their phenotype differs. In mice, the chemotherapy-related reduction in circulating CD11bhiGr-1hi cells reflected their depletion in the tumor. In patients, a number of circulating myeloid cell subsets were reduced, but whether this also occurs in the tumor remains to be established. Second, in comparison to the T cell responses obtained in our earlier studies, the current ones were of far greater magnitude. Although the tests were performed by the same laboratory according to the same standard operating procedures, we did not perform a formal head-to-head comparison, and future trials should confirm these findings. Finally, both the strength of the vaccine-induced immune response and the reduction in circulating myeloid cells were retained for up to 2 weeks after the sixth cycle of CarboTaxol. It is not clear if stopping chemotherapy will coincide with a quick rebound of the myeloid cells, how this affects the vaccine-induced immune response, and whether the phenotype of myeloid cells will be altered under the influence of vaccine-activated T cells. These should all be subjects of future investigations.

In conclusion, we have shown that CarboTaxol chemotherapy not only is devoid of immunosuppressive effects on tumor-specific T cells but also vigorously stimulates tumor-specific immunity by normalizing the abnormal numbers of the immunosuppressive myeloid cell populations. Additional studies will have to demonstrate whether CarboTaxol and adequately timed HPV16-SLP vaccination also produce clinical benefit in patients with advanced cervical cancer. A larger clinical trial is already under way to test this (NCT02128126). If successful, this immunotherapeutic approach should be easy to implement because it combines smoothly with the preferred chemotherapy treatment for advanced cervical cancer.


Study design

The aim of the study was to test whether CarboTaxol could be combined with HPV16-SLP vaccination. We first used the HPV16 E6/E7–expressing TC-1 tumor mouse model to define the impact of CarboTaxol on systemic and intratumoral immunological parameters as well as on the clinical efficacy of vaccination. After observing that CarboTaxol did not affect lymphocytes but had a strong effect on myeloid cells and improved tumor control by therapeutic vaccination, we started a multicenter, open-label, observational study, entitled “Immunological aspects of combined chemo-immunotherapy in patients with advanced cervical cancer” (EudraCT 2010-018841-76), consisting of two cohorts of patients. In the first cohort, six patients with advanced, recurrent, or metastatic cancer were treated with six cycles of CarboTaxol every 3 weeks, and the composition and function of the myeloid and lymphoid cells in peripheral blood were analyzed. After identifying a specific time window during chemotherapy potentially permitting the best T cell response, we studied the second cohort of patients. In this cohort, 12 patients were treated with CarboTaxol and one dose of an HPV16-SLP vaccine 2 weeks after the second cycle of CarboTaxol. Blood samples were drawn to validate the observations made in the first cohort as well as to study the vaccine-induced T cell response. The investigators performing and analyzing immunological assays were blinded to the clinical parameters of the patients. The data from the immunomonitoring studies are reported according to the recommended standard format “minimal information about T cell assays.”

Mice and tumor treatment

Female C57BL/6 mice (6 to 8 weeks old; Charles River Laboratories) were housed in individually ventilated cage systems under specific pathogen–free conditions. The experiments were approved by the Animal Experiments Committee of Leiden University Medical Center (LUMC), in line with the guidelines of the European Committee.

The tumor cell line TC-1 is of C57BL/6 origin and expresses HPV16 E6 and E7 (32). TC-1 tested negative for rodent viruses by polymerase chain reaction. Mice were subcutaneously inoculated with 1 × 105 TC-1 tumor cells. When a palpable tumor was present on day 8, mice were split into groups with comparable tumor size and treated with carboplatin [40 mg/kg, day 8, intraperitoneally], paclitaxel (20 mg/kg, days 8 and 9, intraperitoneally), and/or subcutaneous synthetic long HPV16 E743–77 peptide (SLP; GQAEPDRAHYNIVTFCCKCDSTLRLCVQSTHVDIR; 150 μg) dissolved in dimethyl sulfoxide (Sigma), diluted in phosphate-buffered saline (B. Braun), and emulsified in Montanide ISA-51 (Seppic). Chemotherapy was repeated 1 week later (day 15 for carboplatin and days 15 and 16 for paclitaxel), and vaccination was repeated 2 weeks after initial treatment (day 22). Detailed information on the immunomonitoring and statistics is given in the Supplementary Materials.


Patients with clinical and radiological evidence of advanced-stage, recurrent, or metastatic cervical cancer; with no curative treatment options; and scheduled for CarboTaxol were enrolled between January 2011 and January 2013. Other inclusion criteria were as follows: (i) mentally competent patients 18 years or older, (ii) no other active malignancy, (iii) no indication of active infectious disease such as HIV, (iv) and no other condition that may jeopardize the health status of the patient. Patients were followed until 2 to 3 weeks after they had received their last chemotherapy cycle and thereafter at standard visits. LUMC, Academic Medical Center (Amsterdam), Free University Medical Center (Amsterdam), and Netherlands Cancer Institute–Antoni van Leeuwenhoek Hospital (Amsterdam) were the participating hospitals.

HPV typing was performed on the tumor and/or smears taken at study entry (8), but it was not part of the inclusion criteria. The study was conducted in accordance with the Declaration of Helsinki (October 2008) and approved by the Central Committee on Research Involving Human Subjects (NL31572.000.10) in agreement with the Dutch law for medical research involving humans.

Treatment of patients

For the first cohort, six of the nine screened and eligible patients participated and were treated at their hospital with CarboTaxol, consisting of carboplatin (dose based on renal function; area under the curve of six regimen) and paclitaxel (175 mg/kg2) on day 1 of each cycle, every 3 weeks for a maximum of six cycles. The patients were subjected to serial blood sampling. According to the oncology protocols, routine premedication consisting of dexamethasone [20 mg, intravenously (iv)], ranitidine (50 mg in 100 ml of NaCl 0.9%, iv), granisetron (1 mg in 100 ml of NaCl 0.9%, iv), and clemastine (2 mg in 100 ml of NaCl 0.9%, iv) was administered immediately before chemotherapeutic treatment. In case of severe hematological toxicity, neurotoxicity, nephrotoxicity, or gastrointestinal toxicity, dose modifications of carboplatin and paclitaxel were made according to the following standard scheme: (i) if the absolute neutrophil count was <1.5 × 109/liter, platelet count was <100 × 109/liter, or other toxicities were higher than grade 2, then CarboTaxol treatment was postponed for at least a week (or longer until the patients had recovered), and the doses of both chemotherapeutic compounds were reduced by 25%; (ii) if the patient experienced neuropathy higher than grade 2, paclitaxel was stopped but carboplatin was continued. After the completion of immunomonitoring of these first 6 patients, a second group of 12 patients (cohort 2) received CarboTaxol at their hospital at the same schedule and dose, as well as a single subcutaneous vaccination of the HPV16-SLP vaccine (300 μg per peptide emulsified in Montanide ISA-51) consisting of two mixes of peptides injected separately in the left and right limb (either arm or leg) (10) at LUMC, 2 weeks after the second cycle of CarboTaxol. For cohort 2, 18 advanced cervical cancer patients were screened, 5 patients declined participation, and 1 patient (ID6002) died of her disease before receiving the vaccine.

Clinical evaluation of safety and tolerability

The safety and toxicity of treatment were evaluated according to the National Cancer Institute CTCAE v3.0. Well-known toxicities of CarboTaxol were classified as study-related events. Before the start of CarboTaxol and before vaccination, patients were physically examined and medical history was obtained. Vital signs were measured, and the injection site was inspected 15 min, 1 hour, and 4 hours after vaccination. Patients were followed with routine visits (every 3 months until progression) to monitor for AEs. For each vaccine-related AE, the relationship to HPV16-SLP was defined as definite, probable, or possible. Injection site reactions were classified as definitely vaccine-related. Injection site reaction grade 1 was defined as swelling, erythema, and tenderness (pain/itching). Injection site reaction grade 2 was defined as tenderness or swelling with inflammation or phlebitis. Injection site reaction grade 3 was defined as severe ulceration or necrosis. Venous blood samples were drawn for routine hematological analysis, including leukocyte differential counts and biochemistry assessments. Patients were followed up until date of death or loss to follow-up.

Immunomonitoring of clinical trial

Blood samples from patients were taken at the time points indicated in Fig. 4A. In addition, 19 healthy volunteers donated blood. PBMCs were isolated by Ficoll gradient centrifugation, and cells were subjected to LST (911). MRM and influenza peptide pools served as positive controls. The remaining cells were cryopreserved until use. Thawed PBMCs were tested for their response to PHA in a proliferation assay (33, 34), for their antigen-presenting capacity in an MLR (34), and for their HPV16-specific T cell responses by intracellular cytokine staining (9). The 11-day stimulated nondepleted and CD14-depleted PBMC samples were analyzed by a proliferation assay for antigen recognition (28). The supernatants of the proliferation assays were used for cytokine analysis by cytometric bead array (911). Immunophenotyping of the PBMC samples was performed by flow cytometry (34). Detailed information on immunomonitoring and statistics is given in the Supplementary Materials.


Materials and Methods

Fig. S1. T cells are not affected by CarboTaxol treatment.

Fig. S2. The delay in tumor growth does not differ between the treatment groups.

Fig. S3. Chemotherapy does not hamper T cells but decreases myeloid cell frequencies.

Fig. S4. The combination of carboplatin and paclitaxel results in the strongest reduction of circulating myeloid cells.

Fig. S5. CarboTaxol therapy does not influence general immune parameters.

Fig. S6. CarboTaxol therapy alters the relative frequencies of myeloid cells and lymphocytes.

Fig. S7. Myeloid cells and T cells in blood samples from cervical cancer patients were analyzed by flow cytometry.

Fig. S8. CarboTaxol treatment affects different subpopulations of CD11b+ and/or CD14+ myeloid cells.

Fig. S9. CarboTaxol therapy reduces the number of regulatory T cells in cervical cancer patients.

Fig. S10. HPV16-SLP vaccination induces polyfunctional T cells.

Fig. S11. The vaccine-induced HPV16-specific T cell response is stronger in patients vaccinated during chemotherapy.

Fig. S12. Myeloid cell depletion improves the response of PBMC to stimulation in vitro.

Table S1. Patient characteristics.

Table S2. AEs systemically and at vaccination site.

Reference (35)


  1. Acknowledgments: We thank the patients for participating in this study. M. J. G. Löwik and T. M. A. Berends-van der Meer are acknowledged for their help in the administration of the vaccination. The TC-1 tumor cell line was a gift from T. C. Wu, Johns Hopkins University, Baltimore, MD. The dendritic cell line D1 was provided by P. Ricciardi-Castagnoli, University of Milano-Bicocca, Milan, Italy. Funding: This study was financially supported by a grant from the Dutch Cancer Society (2009-4400) to C.J.M. and S.H.v.d.B., as well as by a grant from the Centre for Human Drug Research (Leiden, Netherlands) to H.v.M. Author contributions: G.G.K., J.B., C.J.M., M.J.W., R.A., A.F.C., and S.H.v.d.B. designed the study. H.v.M., M.I.v.P., G.G.K., and J.R.K. wrote and submitted the protocol and treated the patients. M.J.W., T.C.v.d.S., H.v.M., N.M.L., V.J.v.H., S.v.D., and S.J.S. performed the immunological experiments. M.L.d.K. conducted the statistical analysis. M.J.W., T.C.v.d.S., and S.H.v.d.B. analyzed and and all authors interpreted the data. M.J.W., T.C.v.d.S., H.v.M., R.A., A.F.C., C.J.M., J.B., and S.H.v.d.B. wrote the manuscript. All authors approved the final manuscript. Competing interests: C.J.M. has a stock appreciation right that is the equivalent of a stock option in 1% of the issued share capital of ISA Pharmaceuticals, is being named as an inventor on the patent for the use of SLPs as a vaccine, and is employed as chief scientific officer by ISA Pharmaceuticals, which exploits this patent. C.J.M. is also a member of the Scientific Advisory Board of Immatics, a cancer vaccine company. S.H.v.d.B. is one of the inventors on the patent for the use of SLPs as a vaccine but holds no financial interest. S.H.v.d.B. serves as a paid member of the strategy board of ISA Pharmaceuticals. S.H.v.d.B. also serves as an adviser for the immunomonitoring of spontaneous and immunotherapy-induced immune responses at the Dutch Cancer Institute, Amsterdam, Netherlands. No other potential conflict of interest relevant to this article was reported. Data and materials availability: The materials used and data obtained in this study are available on reasonable request and upon signing a material transfer agreement. ISA Pharmaceuticals provided the vaccine, and any request concerning this vaccine should be directed to ISA Pharmaceuticals.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article