MRI of combination immunotherapy in an epithelial ovarian cancer preclinical model

Abstract

Immunotherapies such as checkpoint inhibitors (i.e. anti-PD-1) and peptide-based therapies (DPX-Survivac) have strong potential for treatment of epithelial ovarian cancer, the most lethal gynecological malignancy. Magnetic resonance imaging (MRI) can be used to track tumor growth and iron-labeled immune cells longitudinally at the individual level. We studied MRI immune cell tracking in a murine model of ovarian cancer using a clinically relevant treatment combination of DPX-Survivac, anti-PD-1, and an intermittent low dose of Cyclophosphamide (CPA). HHD-DR1 mice were orthotopically implanted with mouse ovarian surface epithelial (MOSE) cancer cells. Myeloid and CD8⁺ cells were isolated from matched donor mice, labeled with superparamagnetic iron oxide (SPIO) and were scanned using MRI on days 42, 49 and 56. Tumor volumes in the treatment group as measured by MRI were significantly lower than in the control group (p < 0.01). The density of SPIO-labeled myeloid and CD8⁺ T cells in tumors was higher in the treatment group than in the control group. This study provides insights into how MRI can be used in concert with biological assays to study how immunotherapy and chemotherapy combinations exert their antitumor effects.

Introduction

Epithelial ovarian cancer is the most lethal gynecological cancer¹. It is highly aggressive and commonly advanced when diagnosed at stages III and IV when symptoms have spread to the abdominal area^2,3, such as ascites and pelvic discomfort. Ascites are often associated with intraperitoneal metastasis caused by cell shedding from the primary tumor⁴. Late-stage diagnoses are primarily responsible for the global five-year survival rate between 40 and 50%^1,2,5. One of the most significant challenges associated with ovarian cancer is that many patients do not respond to current treatment methods. Even if advanced ovarian cancer cases respond to traditional procedures, such as chemotherapy, radiation, and surgery, an estimated 75% of patients will relapse².

Considering these challenges, immunotherapies are a promising class of cancer therapies. Immunotherapies aim to harness and enhance the immune system’s natural ability to detect and fight off cancerous cells. Some therapies currently in clinical use include immune checkpoint inhibitors and peptide-based T cell-activating molecules. Various checkpoint inhibitors have been FDA-approved and have shown success in clinical trials for skin, lung, and pancreatic cancers^6,7. Immune checkpoint receptors act as “gatekeepers” of the immune response, preventing overstimulation of the immune system⁸. However, in tumor microenvironments, cancer can hijack this interaction, resulting in suppressive effects on the activity of T cells, which in turn can lead to the exhaustion of their immune function^8–10. Checkpoint inhibitors work by blocking surface-expressed immune cell checkpoint molecules that regulate the duration and intensity of immune responses upon interaction with their binding partner. One significant checkpoint protein is the programmed cell death-1 (PD-1) receptor¹¹. Immunotherapeutic checkpoint inhibitors, such as anti-PD-1, are monoclonal antibodies (mAb) that bind and block the interaction between PD-1 and its ligand PD-L1, enhancing T cell antitumor activity by preventing PD-L1 tumor-mediated shutoff of T cells¹². PD-1 receptors are upregulated on the surface of T cells following activation and modulate cell function, survival, and proliferation upon ligand binding.

Peptide-based therapies enhance the immune system’s ability to recognize and respond to specific tumor antigens. The novel immunotherapy DPX^TM (at the time of experiments, this was owned by IMV Inc., Halifax, NS, Canada, but is now owned by BioVaxys, Vancouver, BC, Canada) is a specialized formulation that stimulates strong and specific immune responses¹³. DPX’s oil-based, water-free formulation enhances immune responses by providing the immune system with prolonged exposure to antigens¹³. Specifically, DPX-Survivac combines the DPX formulation and a mixture of survivin peptides. Survivin, a protein necessary for inhibiting apoptosis, is highly expressed in many cancer types with solid tumors, including ovarian cancer^13,14. DPX-Survivac therefore enhances antitumor immune responses by stimulating host immune systems^15,16.

One of the most significant limitations of single-agent therapies is the increased likelihood of acquired therapeutic resistance in malignant cells². Combination therapies avoid acquiring resistance by enhancing antitumor activity through multiple mechanisms¹⁷. For example, positive outcomes have been demonstrated in ovarian cancer patients treated with a combination of DPX-Survivac and low-dose cyclophosphamide (CPA)^13,18. Low-dose CPA is a chemotherapeutic agent regimen that has been shown to selectively reduce the number of regulatory T cells (Tregs), thereby removing suppressive cells and subsequently enhancing targeted immune responses^19,20.

Studies demonstrated that low-dose CPA enhanced antigen-specific immune responses induced by DPX-Survivac in ovarian cancer patients^13,18. Furthermore, Weir et al.²¹ found that triple combination therapy with DPX, low-dose CPA, and anti-PD-1 provided better long-term control of established tumors via slower growth in a derived preclinical model of human papillomavirus (HPV) cancer compared to treatment with anti-PD-1 alone. Clinical studies have investigated DPX-Survivac (or a modified version of the DPX therapy now known as Maveropepimut-S) with pembroluzimab (anti-PD-1) and low-dose cyclophosphamide in the context of Relapsed-Refractory Diffuse Large B Cell lymphoma²² and ovarian cancer²³.

Studying novel combination therapies like this triple combination therapy in preclinical models is vital for their translation into a clinical setting, given the long time and high cost commitments needed for clinical trials. In particular, tumor progression and cell recruitment can be tracked using magnetic resonance imaging (MRI). Once acquired, anatomical MR images can be used to identify the volume and properties of specific regions of interest (ROI), such as tumors and lymph nodes. Quantification and recruitment of cells to particular ROIs can be performed by labeling cells with superparamagnetic iron oxide (SPIO) in vitro and then tracking them in vivo using MRI. SPIO nanoparticles can be used to label a wide variety of immune cells, including dendritic cells, T cells, and natural killer (NK) cells, amongst others^15,24–28. SPIO causes signal dephasing due to magnetic perturbations, which can be detected using both R₂* and R₂ contrast, with R₂* being particularly sensitive to these effects. Both R₂ and R₂* measure transverse relaxation effects (R₂ = 1/T₂ and R₂* = 1/T₂*), with R₂* being sensitive to larger time invariant field effects (such as those due to SPIO) and R₂ being more sensitive to time varying effects (i.e., due to motion such as diffusion).

TurboSPI is used to create R₂* maps^29–31 that can be overlaid onto anatomical images. The specific acquisition properties of TurboSPI make it relatively insensitive to free iron, but it remains sensitive to magnetic perturbations caused by SPIO that is encapsulated within cells^29,30. Our lab has previously demonstrated the use of TurboSPI in vivo to semi-quantitatively evaluate SPIO-labeled T cells and myeloid-derived suppressor cells^15,26. In contrast with more traditional R₂* mapping techniques that capture a small number of points along a decay curve, TurboSPI acquires hundreds of points at high temporal resolution^29,32, which improves its sensitivity to compartmentalized iron. Previous work has been done by our lab³⁰ to minimize other potential confounders such as fat. MRI and molecular imaging in general have proven to be particularly useful in better understanding the roles of treatment and interaction with the immune system on tumor growth and individual outcomes⁹.

The objective of this study was to use MRI in combination with biological assays to evaluate the effects of a clinically relevant combination therapy. Specifically, this study will investigate survival, tumor progression, tumor and treatment draining lymph node swelling, immune cell recruitment to tumor and lymph nodes, and cellular composition of tumor infiltrating lymphocytes and ascites. We hypothesized that mice treated with DPX-Survivac, anti-PD-1, and low-dose CPA would have better tumor control (i.e., slower tumor growth) and increased immune cell recruitment to tumors, as compared to untreated mice.

Results

Imaging demonstrates the success of triple combination therapy

The timeline for imaging and treatment is shown in Fig. 1. The average tumor volume of mice treated with DPX-Survivac, anti-PD-1, and CPA appeared smaller than the average tumor volume of untreated mice at all three imaging timepoints (Fig. 2), but was only statistically significantly different on day 56 post-implant (p < 0.01, Fig. 2B). Additionally, the tumor growth percentage was calculated for all mice imaged across the three-time points (Fig. 2C). Percent tumor growth between days 42–56 post-implant in untreated mice was significantly increased compared to the treated group (335% increase vs 93% increase, **p < 0.01, Fig. 2C). Taken together, these results suggested that a combination therapy of DPX-Survivac, anti-PD-1, and CPA slows the growth of primary tumors, resulting in reduced tumor volumes. There was no significant change in survival due to treatment (Supplementary Fig. 1) during the study period; however, all mice were terminated at a set endpoint for tissue processing, restricting a true evaluation of long-term survival.

Fig. 1. Treatment and imaging timeline.

Timeline indicates dates for all treatments and imaging for both donor and recipient mice. MOSE are ovarian cancer cells, CPA is the low-dose cyclophosphamide, MC and CTL are the injected bone marrow derived myeloid cells and cytotoxic T lymphocytes, respectively. Donor mice were used exclusively for the isolation of immune cells and therefore did not receive low-dose cyclophosphamide, as internal work revealed that immune cells were too depleted for sufficient yields. Recipient mice received injections of SPIO-labeled immune cells 24 h prior to MRI scans.

Fig. 2. Tumor volumes and growth.

Tumor volumes were quantified at each time point (days 42, 49, and 56 post-MOSE-implant) by hand-drawing the region of interest. A Representative MR images of tumor growth across the three imaging timepoints in a treated and untreated mouse (L: left; R: right). Ovarian tumors are shown outlined in red. B Quantified tumor volumes for treated and untreated mice at each time point. C Percent tumor growth for mice that were imaged at all three time points. Data were pooled from three separate experiments, shown as average ± standard error (SEM), n = 5–14, Student’s t test, **p $\le$ 0.01.

In addition to quantifying primary tumor volumes, the volumes of both inguinal lymph nodes were quantified at each imaging time point (Fig. 3A, quantitative values in Supplementary Fig. 2) as a potential marker of response^33,34. The average volume of the DPX-draining LN (i.e., the right inguinal lymph node) trended higher than the tumor-draining LN (i.e., the left inguinal lymph node) within the treated group. We then calculated the volumetric ratio between DPX-draining and tumor-draining lymph nodes by dividing the volume of the DPX-draining LN by the tumor-draining LN. Therefore, a ratio >1 indicated that the DPX-draining LN was swelling, and presumably more active relative to the tumor-draining LN. At each imaging time point, the average volumetric ratio was >1 in the treated group, whereas it was <1 in the untreated group (Fig. 3B). This is an indirect indicator of increased immune activity in the lymph node due to treatment as opposed to tumor-specific activity. Using a two-way ANOVA, we found there were significant group-level differences due to treatment (p < 0.0001).

Fig. 3. Inguinal lymph node analysis.

The volumes of the tumor-draining (left) and DPX-draining (right) inguinal lymph nodes were quantified at each time point by hand-drawing the regions of interest. A Representative MR images of the inguinal lymph nodes from a treated and untreated mouse on day 42 post MOSE-implant with LNs colored in red (tumor-draining) and blue (DPX-draining). B The volumetric ratio of the DPX-draining to tumor-draining inguinal lymph nodes (i.e., right lymph node volume divided by left lymph node volume). Statistics were done using 2-way ANOVA, n = 9–13. Treatment was found to have a significant effect on the volumetric ratio. C Lymph node volumetric ratio versus tumor volume for the treated (n = 35) and untreated (n = 30) groups at each time point. Statistics performed using linear regression and the slopes of the lines were found to be significantly different (p < 0.05). Average ± (SEM), *p $\le$ 0.05, **p $\le$ 0.01, ****p $\le$ 0.0001. ns indicates data with 0.1 < p > 0.05. Data were pooled from three separate experiments. L and R indicate left and right, respectively.

As DPX-draining lymph nodes were generally more swollen than the tumor-draining LNs, we then assessed whether increased swelling (i.e., larger treatment-induced LN volumes) correlated with tumor volume. The lymph node volumetric ratio of the treated animals negatively correlated with tumor volume, whereas the opposite was observed in the untreated group (Fig. 3C). This means that for the treated group, mice with lower tumor volumes had higher amounts of swelling in treatment-draining LNs. However, this was not true for untreated mice, indicating that the effect was treatment driven and not a systemic cancer effect. We found there was a significant difference (p < 0.05) between the slope of the untreated and treated mice. These results suggested that DPX-Survivac induces lymph node swelling in the DPX-draining lymph node, which correlates with a smaller tumor volume.

Cell culture phenotyping and SPIO labeling

Flow cytometry was used on samples not tagged with SPIO to assess the purity and activity of cell cultures. Cultures of marrow-derived myeloid cells from treated mice appeared to have a higher proportion of macrophages and fewer monocytes and dendritic cells. Still, they were not significantly different than those in untreated mice (data not shown). Furthermore, the percentage of myeloid cells cultured from treated mice appeared to be higher than that of untreated mice expressing MHCII. There were no evident differences between treated and untreated cultures in the CD11b/CD11c subsets of macrophages, monocytes, and dendritic cells. Cultured CTLs were very pure: >90% of CD3⁺ cells were CD8⁺ (Supplementary Fig. 3A). There were no differences in the expression of PD-1, CTLA-4, or TIM3 between CTLs cultured from treated and untreated mice (Supplementary Fig. 3B). However, CTLs cultured from untreated mice appeared to express higher levels of Ki67 (Supplementary Fig. 3B).

Uptake of SPIO by labeled CTLs and MCs was evaluated using a Prussian Blue assay measured on a spectrophotometer, validated against a known concentration curve³⁵. CTLs were found to have approximately 4 pg of iron/cell, and MCs were found to have approximately 7 pg of iron/cell. Viability for all labeled cells was >90%.

Semi-quantitative analysis of SPIO-labeled cells in the tumor and inguinal lymph nodes

To quantify the recruitment of SPIO-labeled MCs and CTLs, R₂* maps were generated with the TurboSPI MRI pulse sequence and overlaid onto anatomical MR images. R₂* values within the specific ROI were converted into the number of SPIO-labeled cells per mm³ of ROI using R₂* relaxivity curves generated from cell phantoms made with SPIO-labeled MCs and CTLs (same isolation and labeling procedures as in vivo studies).

Treatment with DPX-Survivac, low dose CPA, and anti-PD-1 significantly increased MC recruitment to the tumor, as measured by MCs per mm³ of tumor (p = 0.0211, two-way ANOVA; Fig. 4A). Tumor MCs were recruited to the tumors of treated mice at a density of approximately 750 cells/mm³ (standard error of mean/SEM = 193 cells/mm³; day 42), 590 cells/mm³ (SEM = 209 cells/mm³; day 49), and 540 cells/mm³ (SEM = 433 cells/mm³; day 56). MCs were recruited to the tumors of untreated mice at a density of 185 cells/mm³ (SEM = 134 cells/mm³; day 42), 40 cells/mm³ (SEM = 25 cells/mm³; day 49), and 55 cells/mm³ (SEM = 55 cells/mm³; day 56). Overall, for individual mice, there was no clear trend in cell recruitment of MCs over time, with some mice having large increases in the recruitment of cells and some mice having large decreases. There were no statistically significant differences in MC recruitment. There were however many untreated mice that had no detectable MCs at each time point, with four of six mice on day 42, two of five mice on day 49 and two of three mice on day 56 having no detectable MCs. For untreated mice, however, there was only one mouse of six on day 49 and two mice of five on day 56 with no detectable MCs, even though tumors were considerably smaller in treated mice by day 56.

Fig. 4. Quantification of SPIO-labeled cell recruitment to tumors.

Recruitment of bone marrow-derived myeloid cells (A) and cytotoxic T lymphocytes (B) per mm³ of tumors in both treated and untreated mice. Total cell recruitment was quantified and then divided by tumor volumes for each individual mouse at days 42, 49, and 56 post-MOSE-implant. Data were pooled from three separate experiments, average ± SEM, n = 3–7, 2way ANOVA, *p $\le$ 0.05, **p $\le$ 0.01. Treatment was found to have a significant effect on both myeloid cell recruitment (**p < 0.01) and cytotoxic T lymphocyte recruitment (*p < 0.05) to tumors.

Given that lymph nodes are immune cell infiltration and priming sites, it was hypothesized that the combination therapy might increase the recruitment of MCs to lymph nodes. However, treatment did not significantly impact MC recruitment to the DPX-draining lymph node (i.e., left LLN, p = 0.1598; two-way ANOVA) (Fig. 5A). In addition, treatment did not substantially affect MC recruitment to the tumor-draining lymph node (p = 0.8903; two-way ANOVA; Fig. 5B).

Fig. 5. Quantification of SPIO-labeled cell recruitment to the inguinal lymph nodes.

Total recruitment of myeloid cells (A, B) and cytotoxic T lymphocytes (C, D) to the DPX- and tumor-draining inguinal lymph nodes on days 42, 49, and 56 post-MOSE-implant. Data were pooled from three separate experiments, average ± SEM, n = 3–8, 2-way ANOVA.

CTL recruitment was also quantified using the same methods. Treatment with DPX-Survivac, anti-PD-1 and low dose CPA significantly increased CTL recruitment to the tumor, as measured by CTLs per mm³ (p = 0.0155, two-way ANOVA; Fig. 4B). Average CTL recruitment to the tumors of treated mice was 2500 cells/mm³ (SEM = 687 cells/mm³; day 42), 2300 cells/mm³ (SEM = 690 cells/mm³; day 49), and 2300 cells/mm³ (SEM = 844 cells/mm³; day 56; Fig. 4). Average CTL recruitment to tumors in the untreated group was 2200 cells/mm³ (SEM = 472 cells/mm³; day 42), 450 cells/mm³ (SEM = 450 cells/mm³; day 49), and 83 cells/mm³ (SEM = 58 cells/mm³; day 56; Fig. 4B). These results suggest that treatment with DPX-Survivac, low-dose CPA, and anti-PD-1 increases the recruitment of CTLs to the tumors of treated mice. In untreated mice, three of the four mice had no detectable cells on day 49 and one out of four mice had no detectable cells on day 56. Treated mice always had large numbers of detectable CTLs.

Overall, for individual mice, in the treated group there was no clear trend in cell recruitment over time, with some mice having large increases in the recruitment of cells and some mice had relatively stable levels. Interestingly, two of the three mice that needed to be terminated prior to day 56 due to increasing tumor volumes saw large decreases in the recruitment of CTLs between days 42 and 49. In the untreated group all of the mice saw large decreases in the numbers of CTLs recruited over time. Unfortunately, due to termination of mice over time, statistical power was limited.

We found that treatment did not significantly impact CTL recruitment to the DPX-draining lymph node (p = 0.2405, two-way ANOVA; Fig. 5C) or the tumor-draining lymph (p = 0.9820, two-way ANOVA; Fig. 5D). However, it did appear that the recruitment of CTLs to the DPX-draining lymph node decreased in untreated mice as the study progressed, whereas it remained relatively consistent in treated mice. This resulted in far more CTLs in the DPX-draining lymph node in treated mice at day 56 compared to untreated mice. There was a single outlier mouse in the untreated group, but the other 4 of 5 untreated mice had no detectable CTLs in the DPX-draining lymph node at day 56.

TurboSPI validation

IHC was used to validate that the injected cells labeled with SPIO were the cells of interest in the terminal tissues. Lymph nodes were utilized as the stains were more homogeneous. Supplementary Fig. 4 shows lymph nodes from mice that received either CD8 + T cells (top) or CD11+ myeloid cells (bottom). Immune cells were stained with Alexa Fluor 633 Avidin and biotinylated anti-CD8 (top) or biotinylated anti-CD11 (bottom). The SPIO had a rhodamine B tag visible with IHC. All of the rhodamine B-positive cells are also CD8 or CD11 positive in the lymph nodes (as indicated by the green arrows), indicating that the cells visualized with MRI are the cells of interest.

Characterization of ascites

The total proportion of immune cells in the ascitic fluid was determined using the CD45 marker (common leukocyte antigen). On average, treated mice had a higher proportion of leukocytes than untreated mice, albeit not statistically significant (p = 0.09; unpaired t-test, Fig. 6A). The cellular composition of ascitic fluid collected from treated mice had an average of 97.16% CD45+ cells, while the untreated mice had an average of only 83.3% CD45+ cells (Fig. 6A). The variances of the two groups were significantly different (p < 0.001) with the coefficient of variation being only 1% for the treated group, versus 30% for the untreated group. Within the CD45+ population in ascites from treated mice, an average of 22.02% expressed F4/80, 27.90% expressed CD11c, and 16.76% expressed CD3ε (Fig. 6B). In the untreated ascites samples, 34.89% of the CD45+ population expressed F4/80, 32.88% expressed CD11c, and 17.39% expressed CD3ε (Fig. 6B). There were no statistical differences in the percentage of F4/80 (p = 0.1712), CD11c (p = 0.6212), and CD3ε (p = 0.9310) populations between treated and untreated samples (unpaired t-test, Fig. 6B).

Fig. 6. Characterization of the cellular composition of ascites.

Flow cytometry was used to assess the cellular composition of ascites fluid collected from both treated and untreated mice. Red blood cells were lysed immediately prior to preparation for flow cytometry. Cells were gated to remove debris, as indicated above. A Percentage of leukocytes (CD45⁺ cells). Treated mice appeared to have consistently higher proportions of CD45⁺ cells, though not statistically significant. B Percentage of F4/80 vs. CD11c populations. C Percentage of CD3⁺ T cells. D Percentage of CD4⁺CD8^-, CD4^-CD8⁺, and CD4⁺CD8⁺ T cell populations. Average ± SEM, n = 2-5.

The CD3ε marker encompasses all T cell populations; therefore, we also examined the proportions of CD4 and CD8 cells within the CD3ε population (Fig. 6D). There were more CD4 + T cells than CD8 + T cells in the ascitic fluid collected from treated and untreated mice (Fig. 6D). Interestingly, a population of T cells also expressed both CD4 and CD8 (double positive, or DP). In the treated ascites samples, an average of 44.37% of the CD3ε population was DP (Fig. 6D). Only 15.42% of the CD3ε population in the untreated samples was DP (Fig. 6D). There was no significant difference between the average percentage of DP cells in the treated and untreated samples (p = 0.1250; unpaired t-test) (Fig. 6C). However, the coefficient of variation in the CD4 + CD8+ cells was only 12% for the treated group and 82% for the untreated group. Together, these results suggest that treatment with DPX-Survivac, low-dose CPA, and anti-PD-1 increases the presence of CD45+ immune cells in ascitic fluid and may increase DP T cells but does not have an impact on the presence of other specific immune cell populations.

Tumor-infiltrating lymphocyte assay

Tumor-infiltrating lymphocytes were assessed using flow cytometry (Fig. 7). The tumors of mice treated with DPX-Survivac/CPA and anti-PD-1 appeared to have fewer CD45+ immune cells, CD3⁺CD4⁺ T cells, CD19⁺ B cells, and granulocytic myeloid-derived suppressor cells (MDSCs) (CD11b⁺Ly6G⁺Ly6c^neg; Fig. 7A, B, D, E) than tumors from untreated mice. Furthermore, there were no significant differences in the percentage of CD3⁺CD8⁺ T cells, transitional MDSCs (CD11b⁺Ly6G⁺Ly6c^high), monocytic MDSCs (CD11b⁺Ly6G^negLy6c^high), classical or myeloid-derived dendritic cells, macrophages, and F4/80-expressing cells (Fig. 7C, E, F, G) between untreated and treated mice. However, there was a trend of increased CD3 + CD8 + T cells in treated mice. Overall, the ovarian tumors had few monocytic and transitional MDSCs and very few macrophages or myeloid-derived DCs (Fig. 7E, F).

Fig. 7. Tumor-infiltrating lymphocyte assay.

Tumors were resected from remaining treated and untreated mice following the last MRI scan. Tumors were digested and processed for flow cytometry to assess the proportions of tumor-infiltrating lymphocytes: A CD45⁺ immune cells. B CD4⁺ T cells. C CD8⁺ T cells. D CD19⁺ B cells. E Myeloid-derived suppressor cells and F, G dendritic cells and macrophage populations. Average ± SEM, n = 3 (per treatment group), statistics by Student t test, *p $\le$ 0.05, ns = not significant, but p < 0.1.

Discussion

In this study, we used MRI and biological therapies to evaluate a clinically relevant combination therapy for epithelial ovarian cancer. Our primary findings matched our hypothesis and MRI indicated that the combination therapy studied in the project slows ovarian tumor growth (Fig. 2) and increases the recruitment of CTLs to tumors and MCs to the DPX-draining lymph node (Fig. 4). We found no statistical difference in survival (Supplementary Fig. 1). We did find statistically significant increases in volume of the treatment draining inguinal lymph nodes in treated mice. We also found using flow cytometry that treated mice had significantly decreased amounts of CD19 + B cells, CD4 + T cells and granulocytic MDSCs (Fig. 7). Using flow cytometry we found no significant differences in ascites composition, although there was a very high degree of variability in the proportion of double-positive T cells in untreated mice (Fig. 6).

Late-stage diagnoses account for the high mortality of patients with epithelial ovarian cancer: the 5-year survival rate is less than 50%^1,36, with advanced cases having a survival rate of only ~30%⁵. Finding effective therapies is imperative for increasing the survival rate of this disease. However, it is challenging to study orthotopic ovarian cancer models in any depth without using imaging to look at the primary tumors located in and around the ovaries. We therefore used anatomical MRI to evaluate the effect of combination therapy with DPX-Survivac, low-dose CPA, and anti-PD-1 on ovarian cancer survival and tumor growth. While there was no statistical significance between the survival of treated and untreated mice (Supplementary Fig. 1), this is likely due to the duration of the study. One limitation of assessing survival was that all remaining mice were euthanized on day 60 post-implant for tissue collection. Future studies will extend to further time points to assess actual survival more accurately.

Compared to the untreated group, treated mice had significantly smaller tumors by the end of the study (Fig. 2). These results suggested that combination therapy with DPX-Survivac, low-dose CPA, and anti-PD-1 slowed the growth of established ovarian tumors. Due to the latent nature of ovarian cancer, patients are typically diagnosed at stages when the primary tumor has already been established. Therefore, novel treatments must be effective with more advanced cancer models. This clinically relevant treatment schedule replicates current clinical trials^23,37 and builds upon previous use in other cancer models^13,18,38. In addition, slowing the growth of primary tumors may help prolong survival and enhance the efficacy of other treatments, such as debulking surgeries. While this combination shows some promise in slowing tumor growth, there remain a number of questions about the mechanisms of action of the therapeutics.

A primary objective of this study was to evaluate whether MRI immune cell tracking could be used to monitor and quantify the recruitment of adoptively transferred SPIO-labeled immune cells at two different timepoints. While cell tracking of these two cell types has been done previously in a subcutaneous tumor model¹⁵, this is the first time it has been done in a more clinically relevant orthotopic ovarian model, which can be more difficult due to increased motion artifacts and increased fat in the lower abdomen. These orthotopic models of ovarian cancer, particularly when used in combination with humanized mice with intact immune systems and clinically relevant treatment regimens, are critical for correctly understanding immunological responses.

Using MRI, we found that overall, combination therapy increased the recruitment of both SPIO-labeled MCs and SPIO-labeled CTLs to the tumor (Fig. 4), although there were no significant differences at each timepoint due to the large variability, particularly in treated mice. For SPIO-labeled CTLs, this increase is likely linked to the slower tumor growth seen in treated mice. Although flow cytometry results did not find a significant difference, likely due to a small N, there was a trend of higher levels of CD8 + T cells as measured by FC in treated mice. It should be noted that freshly SPIO-labeled CTLs and/or MCs are being injected 24 h prior to each scan session due to the approximate seven-day lifespan of 50 nm SPIO particles. There may be some retention of cells week to week, although given the stability of cell numbers in treated mice, this retention is likely minimal, otherwise we would see increases over time. However, this is a limitation of this technique and must be taken into account when interrogating the results.

Interestingly, more untreated mice experienced a drop in the cellular density of recruited CTLs throughout the study with three of four mice at day 49 and one of four mice at day 56 having no detectable CTLs in the tumor, and with remaining mice at day 56 having very low numbers of CTLs. This could be due to an increasingly immunosuppressive environment. As indicated in Fig. 7, untreated mice had increased numbers of granulocytic myeloid cells, which can be myeloid-derived suppressor cells, a highly suppressive immune cell type. Given the broad range of cells that can be classified as MCs, it is not clear if the increases seen with SPIO-labeled cells represent a more inflammatory or suppressive cell type. Future studies would benefit from further sorting of MCs prior to labeling with SPIO and implanting them to understand this phenotype better.

We also quantified cell recruitment to the tumor-draining and DPX-draining inguinal lymph nodes, which are sites of immune cell infiltration and priming. Although we found no significant differences in cell recruitment, we did notice that CTL recruitment to the draining lymph node in untreated mice decreased throughout the study similar to tumors. In contrast, treated mice had similar cell densities throughout. Though this difference was not statistically significant, this may have been due to our limited sample size and an outlier point at day 56 for untreated mice (4/5 untreated mice had no CTLs in the right, or DPX-draining, lymph node at day 56). This data suggested that the combination therapy may have enhanced the recruitment of immune cells, such as CTLs, to the DPX-draining lymph node for priming adaptive immune cells against antigens. Future studies would benefit from looking at cellular recruitment to other tumor-draining lymph nodes such as the para-aortic LNs.

The increase of CTLs to the DPX draining lymph node in treated mice may also be linked to changes in the volumetric ratio (of the DPX-draining: tumor-draining inguinal lymph node) between treated and untreated mice. The average volumetric ratio was consistently increased in mice treated with the combination therapy. In fact, few untreated mice had a volumetric ratio greater than 1 (Fig. 3). These results suggest that DPX-Survivac increased swelling in the DPX-draining lymph node relative to the contralateral lymph node, as the other individual therapies in the combination are systemic. Given that lymph nodes are sites of immune cell priming, the observed increase in size, along with CTL cell recruitment to the DPX-draining lymph node, suggested that DPX-Survivac may increase lymph node size due to immune cell infiltration. Our results were similar to those found in studies^33,34,39 that described the inguinal lymph node volumetric ratio as a potential biomarker for successful therapy with DPX. Using a cervical cancer (C3) tumor model, they^33,34,39 found that treatment with a DPX peptide-based therapy increased the volumetric ratio of the DPX-draining lymph node to the tumor-draining lymph node. Furthermore, this volume increase was associated with better outcomes, namely, decreased tumor growth, indicating its utility as a potential indirect indicator of treatment response.

Weir et al.^19,21 assessed the tumor infiltration of antigen-specific (R9F-specific) CTLs in the C3 cervical tumor model using a similar combination therapy: anti-PD-1, DPX-R9F, and low-dose CPA. The study found that combination therapy with all three agents enhanced the infiltration of R9F-specific CTLs to tumors compared to treatment with DPX-R9F and low-dose CPA alone, using flow cytometry. Similarly, we found that the tumor infiltration of CTLs was increased when mice were treated with the three therapies combined compared to untreated.

Upon analysis of the ascitic fluid, mice treated with the combination therapy had consistently higher percentages of CD45+ cells (leukocytes) (Fig. 6). The remaining cells were believed to be free-floating tumor cells. In ovarian cancer, malignant ascites contribute to transcoelomic metastasis (metastasis through the peritoneal cavity) by providing primary tumor cells with a medium for dissemination^5,40,41. Therefore, the increased percentage of CD45+ cells observed in ascites from treated mice indicates fewer tumor cells in the ascitic fluid. Conversely, the increased presence of putative tumor cells in the ascites of untreated mice suggests that these mice are more likely to have metastases. As the presence of tumor cells within ascites is associated with poor prognosis, this could serve as a potential biomarker for the efficacy of this combination therapy to reduce tumor burdens. Future studies would benefit from imaging at longer timepoints and monitoring regions like the lungs for potential metastases.

We also noted the presence of a CD4 + CD8+ (DP) T cell population in the ascitic fluid of both treated and untreated mice (Fig. 6). While untreated mice had consistently decreased proportions of DP T cells, some mice in the treated group had an increased percentage of DP T cells (Fig. 6). DP T cells have been reported in healthy and diseased individuals⁴². The roles of this unconventional T cell population are not fully understood, with conflicting reports describing cytotoxic or suppressive roles for these cells⁴³. Several studies have reported the role of DP T cells in mediating the tumor response and favoring immune escape in many cancer types, such as urological cancer, renal carcinoma, metastatic colorectal cancer, melanomas, and breast cancer lesions^42,44–46. Menard et al.⁴⁵ have observed in renal carcinoma patients that DP T cells express high levels of PD-1, which will make them a suitable target for checkpoint inhibitor therapies. This would explain the reduced tumor volume in mice treated with combination therapy by restricting the immunosuppressive role or enhancing the cytotoxicity of this cell population. Tracking these DP T cells in future experiments would allow for future study of their specific role.

It should be noted that R₂* quantification has limitations. It assumes that the amount of iron per cell that is measured in vitro prior to injection, and which is used to generate the calibration curve, remains constant until it is completely degraded. This therefore assumes no spilling of iron or cell division with iron split between cells. This is mitigated somewhat by acquiring the images 24 h post-injection, i.e., before iron content should change dramatically. We are also assuming that the amount of iron per cell is reasonably constant between days and even across cells within an experiment. This is the highest cause of variability in all cell tracking experiments, because it is almost impossible to guarantee that cells are homogenously taking up iron (or any other contrast agent for that matter). We have found internally that the average mass of iron/cell varied by approximately ±0.5 pg across days, which is reasonably stable. However, it is almost impossible to assess variability across a population, so we are limited by our assumption of “equal iron” across all cells. Because TurboSPI is less sensitive to free iron due to the differences in the R₂/R₂* relationships between free and cellular SPIO (see refs. ^29,30 for more details), these R₂* maps are more specific to cellular SPIO than traditional multi-echo gradient echo MRI scans. Compared to SPIO, background levels of intracellular iron should be negligible. We have done limited validation using IHC to demonstrate that iron is co-localized with the appropriate cell type, but the precise quantification has not been extensively validated. However, the most important information to take away from the semi-quantitative results is not the absolute numbers, but the relationships over time and between treated and untreated groups. Even if there were a significant scaling factor missing, the overall conclusions would remain unchanged.

Collectively, our results demonstrated that molecular imaging allowed the study of a combination therapy with DPX-Survivac, anti-PD-1, and low-dose CPA in epithelial ovarian cancer, demonstrating that the combination decreases the tumor burden of mice. As evidenced by cell tracking, the combination therapy may exert its effects by inhibiting tumor-mediated immunosuppression and subsequently increasing the infiltration of cytotoxic T lymphocytes into the primary tumor. Additionally, the increased percentage of immune cells shown in the ascitic fluid may indicate the combination therapy’s efficacy against ovarian tumors. This study demonstrates that immune cell tracking can be used to probe the longitudinal mechanics of SPIO-labeled cells in orthotopic cancer models. In future work, we plan on investigating more immune cell subtypes, particularly subtypes of myeloid cells such as monocytes, granulocytes, and myeloid-derived suppressor cells, in single immunotherapies and at more timepoints, to understand individual therapeutic mechanisms of action.

Methods

Mice

All preclinical experiments were conducted under ethics protocols approved by the University Committee on Laboratory Animals at Dalhousie University, Halifax, N.S., Canada (project ID: 20-124). Female humanized transgenic mice (HLA-A2.I-HLA-DRI-transgenic (HHD) H-2 class I/II knockout), 6–23 weeks old, were obtained from Charles River Laboratory (France) and bred in-house. All mice were housed in filter-top cages and were provided food and water ad libitum. Mice were randomized into the two groups. Originally we had aimed to use 15 mice per group, but due to issues related to surgery, there were n = 10 recipient mice in the untreated group and 14 mice in the treated group that received DPX-Survivac, anti-PD-1, and low-dose CPA. Upon study completion or upon reaching humane endpoint, mice were anesthetized via inhaled isofluorane (5%) until loss of pedal reflex and terminated via subsequent cervical dislocation.

Cancer cell line and implant

Mouse ovarian surface epithelial (MOSE) cancer cells were provided by Dr. Vanderhyden and were genetically modified in-house via plasmid to express the survivin epitope and confer puromycin resistance. Cells were cryopreserved in Calf Bovine Serum with 10% dimethyl sulfoxide (DMSO), thawed and cultured in DMEM (Dulbecco’s Modification of Eagle’s Medium; Corning, Corning NY) supplemented with 100 U/mL Penicillin, 100 µg/mL streptomycin (Gibco, Burlington, ON), and 10% Calf Bovine Serum (Hyclone) and grown at 37 °C in an atmosphere of 5% CO₂. After the first passage, cells were selected for survivin by adding 10 µg/mL of puromycin.

Prior to cell injections, mice were given a subcutaneous injection of 5 mg/kg meloxicam (Boehringer Ingelheim, Burlington, ON, Canada). At the same time, mice were also given 0.5 mL of warmed Lactated Ringer’s solution (Baxter, Mississauga, ON, Canada). Mice were then anesthetized using isofluorane, with depth assessments done regularly using pedal reflex. The interscapular and left flank region were shaved and a slow-release buprenorphine 1 mg/mL (Chiron Pharmacy, Guelph, ON, Canada) was given via subcutaneous injection on the left side. The flank was massaged to minimize potential buprenorphine-induced lesions, and the surgical site was cleaned with a chlorohexidine:alcohol:betadine scrub three times. Small incisions were made in both the skin and peritoneum to enable access to the ovarian fat pad. The ovary was pulled out through the incision and clamped. A Hamilton syringe (Hamilton, Reno, NV, USA) was loaded with 2 µL of 1 × 10⁴ MOSE cells in 1X PBS (Phosphate Buffered Solution; Corning), and using a surgical microscope, cells were injected into the oviductal-bursal junction. Care was taken to avoid leaking through the injection site. The ovary was then pushed back into the abdomen and the peritoneum and skin were sutured, with wound clips used to ensure closure.

Animals received frequent detailed clinical examinations (DCEs) until the pre-surgery weight was reached. Stitches and wound clips were removed between 5 and 10 days post implant. Once mice regained weight (approximately 3 weeks post-surgery), DCEs were conducted twice/week. Tumors were monitored throughout the study by palpating the abdomen and at certain timepoints via MR imaging (more details below) to obtain precise tumor volumes. Mice were terminated if they lost 15% of their pre-surgery weight, presented severe ascites or ulcerations, or showed signs of pain and lethargy. Implant and treatment timelines can be found in Fig. 1.

DPX-Survivac treatment

DPX-Survivac was prepared at IMV Inc. using their proprietary DPX formulation described elsewhere¹³. Briefly, DPX-Survivac was formulated by combining the following epitopes: SurA1.T (FTELTLGEF), SurA2.M (LMLGEFLKL), SurA3.K (RISTFKNWPK), SurA24 (STFKNWPFL), and SurB7 (LPPAWQPFL), and the tetanus toxin universal T-helper peptide TT_830–843(AQYIKANSKFIGITEL; A16L). DPX-Survivac was resuspended in Montanide ISA 51 VG (SEPPIC, France). Doses of 50 µL were given subcutaneously on the right flank to each mouse on days 28 and 49 post-MOSE implants (Fig. 1). The dosing schedule for all treatments was chosen based on previous work done by Weir et al.²¹ demonstrating that this dosing combination and schedule worked well in an HPV model.

Cyclophosphamide (CPA; MilliporeSigma, Oakville, ON, Canada) was reconstituted in PBS and delivered to mice in drinking water at 0.133 mg/mL for seven consecutive days, beginning on days 14, 28, and 42 post-MOSE implants (Fig. 1). Mice received 20 mg/kg/day, assuming a 20 g mouse consumes 3 mL water/day.

Anti-PD-1 (clone RPM1-14; BioXCell; West Lebanon, NH, USA) was diluted to 2 mg/mL in 1x PBS. Treated donor mice received six intraperitoneal injections of anti-PD-1 at 200 µg per dose on days 21, 24, 27, 42, 45, and 48 post-MOSE implants. Treated recipient mice received injections on days 28, 31, 34, 49, 52, and 55 post-MOSE implants. (Fig. 1).

CD8⁺ cytotoxic T lymphocyte (CTL) isolation and labeling

Inguinal, axial, brachial, and mesenteric lymph nodes were collected from disease-matched and treatment-matched donor mice for isolation of CD8⁺ T cells. This ensured that the cells used for adoptive transfer were exposed to the same cancer and immune environment as recipient mice. CD8⁺ T cells were isolated from lymph nodes using the EasySep™ Mouse CD8⁺ T cell Isolation Kit (Stemcell Technologies, Cambridge, MA, United States). Cells were suspended at 5 × 10⁵ cells/mL in complete RPMI (cRPMI) media (RPMI 1640 (Corning), 10% FBS (Hyclone), 1% penicillin/streptomycin (Gibco), and 55 μM β-mercaptoethanol (Gibco)) and incubated in a CD3-coated cell culture plate with human IL-2 (20 U/mL), mouse IL-12 (100 ng/mL), hamster anti-mouse CD28 (1 µg/mL), and Gentamicin (5 µg/mL). Cells were monitored and kept at a density of 0.5–1 × 10⁶ cells/ml. Fresh cRPMI with IL-2 (20 U/mL) was added as required. Four days following CTL isolation, antigen-presenting cells (APC) were isolated from the spleens of disease- and treatment-matched donor mice. Splenocytes were incubated with LPS (10 µg/mL) in media (DMEM + 10% FBS + 55 μM β-mercaptoethanol +1% Penicillin–Streptomycin + 1% L-glutamine) for 48 h. Non-adherent cells were then treated with mitomycin-c (50 µg/mL) for 20 min, washed, and added to the CTL culture at a ratio of 1:6 (APC: CTL) with survivin peptides (10 µg/mL). Following a 48 h incubation with the APCs, a sample of cells was collected to assess cell purity via flow cytometry, and the other cells were used for in vivo studies. Activated CTLs were collected, washed, and incubated with SPIO-Rhodamine B Molday ION (0.075 mg/mL; Biopal Inc., Worchester, MA, USA) and IL-2 (100 U/mL) at 4 × 10⁶ cells/mL in cRPMI for 22–24 h. Cell viability of CTLs post-SPIO labeling was found to be >90%. Previous work³⁵ demonstrated that labeling CTLs using this methodology does not affect the cytotoxicity ability of CTLs. A subset of labeled CTLs was removed for flow cytometry analysis to assess the effects of labeling on functionality.

Bone marrow-derived myeloid cell isolation and labeling

Bone marrow was collected from the femurs and tibias of disease- and treatment-matched donor mice for isolation of bone marrow-derived myeloid cells (described as MCs hereafter). Red blood cells were lysed with 1x RBC lysis buffer (Tonbo Biosciences, San Diego, CA, USA). Three million cells were incubated in Petri dishes with 10 mL of media containing RPMI, 10% FBS, 1% penicillin-streptomycin, 20 mM HEPES (MilliporeSigma), and 20 ng/mL of granulocyte monocyte-colony stimulating factor for 72 h (GM-CSF; Peprotech, Rocky Hill, NJ, USA). After incubation, 10 mL of media with 20 ng/mL of GM-CSF was added to each plate. Following another 72 h, cells were collected, centrifuged, resuspended in fresh media with 20 ng/mL of GM-CSF, and returned to the plates. The next day, cells were stimulated overnight with survivin peptide SurA2.M (20 µg/mL). Forty-eight hours later, cells were collected, resuspended at 4 × 10⁶ cells/mL, and a sample of cells was assessed for cell purity using flow cytometry, with the remaining cells being used for in vivo studies. The purified myeloid cells were incubated with 0.030 mg/mL of SPIO-Rhodamine B Molday ION for 18–20 h. Cell viability of MCs post SPIO labeling was found to be >90%. A subset of labeled MCs was also removed for flow cytometry analysis to assess the effects of labeling on functionality.

Cell injection and preparation

SPIO-labeled cells (CTLs or MCs) were collected, washed twice with 1x PBS, twice with HBSS++ (Hank’s Balanced Salt Solution; Corning), and then resuspended in HBSS++ with 20 mM HEPES at 5 × 10⁶ cells/mL (MCs) and 25 × 10⁶ cells/mL (CTLs). All mice received 200 µL of CTLs or MCs through intravenous tail vein injections. Cells were injected on days 41, 48, and 55 post-implant (24 h before MRI). Iron loading was assessed in the remaining cells using a Prussian blue assay³⁵; cells were lysed overnight in 100 µL of 1 M HCl, and then 100 µL of K₄Fe(CN)₆ was added to each sample. The absorbance ($\lambda$ = 620 nm) was recorded on SpectraMax i3 (Molecular Devices, San Jose, CA, USA) and compared to a standardized in vitro calibration curve.

Tumor dissociation for tumor-infiltrating lymphocyte assay

Tumors were removed from mice following their final MRI scans. In Petri dishes, the tumor was chopped into smaller pieces with a scalpel and then incubated in digestion buffer [1 mg/mL collagenase type 1 (Gibco) + 0.1 mg/mL DNase I (MilliporeSigma) + 5% FB Essence in HBSS++] at 37 °C for 30 min. Samples were then filtered through a 70 µm strainer with separation buffer [2% FB essence + 1 mM EDTA in 1x PBS (Gibco)]. Red blood cell (RBC) lysis was performed on the suspension as required. Cells were washed with 1x PBS and used for flow cytometry.

Ascites sample preparation

At the endpoint, mice were euthanized, and ascites fluid was collected using a 25G needle and a 5 mL syringe. Red blood cells were lysed with an equal volume of 1x RBC lysis buffer, and samples were centrifuged to collect cells. These remaining cells were washed thoroughly with 1x PBS and used for flow cytometry.

Flow cytometry

Cell samples were blocked in 5% normal rat serum (NRS) for 10 min and then incubated with antibody cocktails at 4 °C for 20 min (Supplementary Tables 1). The same staining procedure was followed as in ref. ³⁵. After staining, samples were fixed with 4% paraformaldehyde (PFA). OneComp eBeads (eBioscience) were used for controls. Data were acquired with a FACS Celesta or FACS Canto II equipped with FACSDiva software (BD Biosciences, Franklin Lakes, NJ, USA) at the Dalhousie University Flow Cytometry Core Facility. Samples were analyzed using FlowJo v10.6.2 (Vicro, Torrance, CA).

Immunohistochemistry

Samples (spleen, tumor, and lymph nodes) were frozen immediately after termination in an Optical cutting temperature (OCT; Fisher)/Sucrose (1:1) solution and stored in a −80 °C freezer until sectioning. Samples were taken to the Dalhousie immunohistochemistry core for processing; tissues were sectioned on the cryostat and placed on slides. Slides were then fixed in −20 °C cold acetone for 2 min, dried, and stored in a −20 °C freezer until staining. For IHC staining, slides were brought to RT, dried, and fixed in cold acetone for 10 min and air-dried again for 30 min. They were then washed in a Tris-buffered saline (TBS)/Bovine serum albumin (BSA) wash and blocked with 20% horse serum for 1 h. They were rinsed again in the TBS/BSA before staining with the Avidin-Biotin Vector kit for 15 min. The samples were stained with the biotinylated primary antibodies overnight (CD8-biotin for the CD8 T cells and CD11-biotin for the MCs, 1:50 dilution). Samples were washed with the TBS/BSA and stained with the Avidin-Alexa Fluor 633 fluorophore (1:200) for 1 h, washed with TBS/BSA and TBS alone. Slides were mounted with antifade mounting media with 4’,6-diamidino-2-phenylindole (DAPI) and visualized on the Zeiss LSM 710 (upright) laser-scanning confocal microscope at the Dalhousie microscopy core facility.

MRI acquisition

Mice were imaged with MRI using a 3T preclinical Agilent MRI (Varian Inc., Santa Clara, CA, USA). The MRI contained a 21-cm inner-diameter gradient coil (200 mT/m; Magnex Scientific, Oxford, UK) and was interfaced with a Varian DD Console (Varian Inc.). Mice were anesthetized with 2–3% isofluorane and secured in an animal holder immediately before imaging. Temperature and respiration rates were monitored throughout imaging using a rectal probe and breathing monitor.

Anatomical images were acquired using a balanced steady-state free precession (bSSFP) pulse sequence. The bSSFP parameters were set at a repetition time of 8 ms, echo time of 4 ms, and a flip angle of 30°. The field of view (FOV), 256 × 170 × 170 matrix, was set at an isotropic resolution of 200 μm and was centered over the torso. The TurboSPI parameters were set with a FOV of 32 × 32 × 32 mm and a slab size of 30 mm^15,29. The repetition time (TR) was 250 ms, the echo train length was 8, and the echo spacing (ESP) was 10 ms. Mice were scanned on days 42, 49, and 56 post-implant, approximately 24 h after SPIO-labeled cell injections.

Imaging analysis

MRI Images were loaded in VivoQuant (InVicro, Ma, US), and regions of interest (ROI) were drawn on the tumor and lymph nodes using the bSSFP image for each imaging time point as in¹⁵. Cell density in tumors and lymph nodes was obtained by extracting frequency histograms of the R₂* signal from the ROIs, and converted from R₂* values per voxel to cell density per mm³ using the calibration curve for either CTLs or MCs. Each voxel was then summed over the ROI, resulting in total cell density for each tumor and lymph node ROI (same methods as refs. ^15,26). Data were then imported into GraphPad Prism 8 (San Diego, CA, USA) for statistical analysis.

R₂* calibration

The same CTL and MC isolation procedures and SPIO labeling were done to isolate sufficient numbers of cells to prepare calibration samples. Six samples for each cell type were prepared by suspending 0, 1 million, 2 million, 3 million, 4 million or 5 million cells in 1.5 mL of a 4% gelatin/PBS mixture in 5 mm NMR tubes. These tubes were then scanned using the same MRI procedures described previously in the methods to obtain R₂* values for each tube. A calibration curve was then plotted in Prism and the slope was used to convert relaxation rate (s⁻¹) to cellular density (cells/mL) for the in vivo data.

Statistical analysis

To compare the results between the two treatment groups, we used a Student’s t test with Bonferroni correction for multiple comparisons, and to evaluate group-level results across both time and treatment groups, we used a two-way ANOVA. Significance is designated as *p $\le$ 0.05, **p $\le$ 0.01, ***p $\le$ 0.001.

Author contributions

Experiment planning and oversight were done by K.D.B., M.S., G.W., O.H. and A.M.-J. Data acquisition was done by C.S.S., M.L.T., H.W., V.G., A.N., C.D., B.Da and A.V.-L. Data analysis was done by C.S.S., H.W., J.T.G., B.Di and K.D.B. Animal procedures were done by C.D., A.M., A.W. and K.B. Cells were provided by B.V. and modified by O.H. Manuscript writing was done by J.G., C.S.S., B.D. and K.D.B. The manuscript was reviewed by all authors.

Data availability

The datasets generated and/or analyzed during the current study are not publicly available due to intellectual property agreements surrounding biological data and size considerations for imaging data, but are available from the corresponding author on reasonable request.

Competing interests

At the time of this study, M.L.T., J.T.G., B.D., A.M., K.Bo, A.W., G.W., A.M.J., O.H. and M.S. were employees of IMV Inc. K.D.B. had a research contract with IMV on studying DPX-Svv in ovarian cancer. K.B. is an Editorial Board Member of NPJ Imaging. K.B. was not involved in the journal’s review of, or decisions related to, this manuscript. The other authors have no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s44303-026-00157-8.