Evaluating the Impact of Adjunctive Partial Cryoablation on Dual Checkpoint Inhibitor Immunotherapy Response in a Murine Model

Abstract

Purpose

To evaluate the impact of adjunctive partial cryoablation on checkpoint inhibitor (CPI) immunotherapy response.

Materials and Methods

One hundred fifty-six mice (equal number of male and female animals) with dual-implanted tumor models were treated with dual CPI or a vehicle and randomized to treatment of a single tumor with partial cryoablation. Tumors were followed for 60 days following cryoablation for response assessment. In additional groups, the tumor microenvironment was characterized via flow cytometry, cytokine analysis, and immunohistochemistry. Statistical comparison was made between the different treatment groups regarding T-cell infiltration and activation characteristics within the noncryoablated tumor and cytokine levels within the partially ablated tumor. Additionally, qualitative assessment of T-cell activation within the cryoablated and noncryoablated tumors at immunofluorescence was carried out.

Results

At 60 days following treatment, CPI and adjunctive cryoablation-treated MC-38 mice had a significantly increased survival rate (79%) compared with mice treated with CPI alone (61%; P < .001). CT-26 mice also had an increased survival rate (57% vs 35%, respectively; P = .04). Following cryoablation, increases in inflammatory cytokines and chemokines within the treated tumors were observed. Flow cytometry of noncryoablated tumor showed increased CD8 T-cell activation. Immunofluorescence and histologic evaluation following cryoablation further demonstrated a robust CD8 T-cell and myeloid infiltrate.

Conclusion

Adjunctive cryoablation significantly increased the response to dual CPI in multiple cancer models at both partially ablated and distant (nonablated) tumor sites. Immune analysis suggests cryoablation promotes a vigorous immune response within the partially cryoablated tumor that increases activation of the adaptive immune system within distant tumor sites.

Keywords: Cancer, Cryoablation, Checkpoint Inhibitor Immunotherapy, Tumor Response

Supplemental material is available for this article.

© RSNA, 2024

See also commentary by Rivera Rodríguez and Mouli in this issue.

Summary

Adjunctive partial cryoablation of a single tumor site significantly increased durable complete response to dual checkpoint inhibitor therapy in a preclinical model, both locally and for distant nonablated tumor sites.

Key Points

• ■ Adjunctive partial cryoablation significantly increased the 60-day survival rate in two dual checkpoint inhibitor–treated models of multifocal murine cancer (79% vs 61% [P < .001] in MC-38 mice; 57% vs 35% in CT-26 mice [P = .04]).

• ■ Adjunctive partial cryoablation induced significant increases in multiple cytokines and chemokines that were correlated with the checkpoint inhibitor response in other studies, including CCL-3 (P = .028) and CXCL-10 (P = .023).

• ■ Activated cytotoxic CD8 T-cell levels were increased (30.3% vs 5.9%; P = .006) and exhausted CD8 T-cell levels decreased (7.9% vs 20.4%; P = .005) in distant tumors.

Introduction

Checkpoint inhibitors (CPIs), which prevent tumoral inhibition of the cytotoxic T-cell response and inhibit T-cell exhaustion, have greatly advanced cancer care. A minority of patients develop highly durable responses indistinguishable from cure. However, overall response rates to these therapies remain modest, with between 5% and 50% of patients responding depending on tumor type and therapy line (1–3). Even among CPI-responsive tumor types, there remains a substantial number of patients who experience only partial responses and respond only briefly. Extending the benefits of checkpoint inhibition by increasing the degree and duration of response across all tumor types would continue to advance the promise of immunotherapies, and combination or adjunctive regimens that can achieve this are of intense clinical interest.

One adjunctive treatment that may have tremendous potential for activating the antitumoral immune response and increasing CPI response rates is cryoablation, a treatment already clinically approved and widely used as a minimally invasive alternative to surgery for multiple tumor types as well as palliation of metastatic disease (4–6). The potential for cryoablation to increase CPI response rates lies in the mechanisms by which it induces tumoral cell death. Cryoablation of a single tumor focus induces necrotic cell death through lethal cooling that brings the tumor temperature below −40 °C. Alternating freeze and thaw cycles induce cell lysis, resulting in release of tumoral antigens and inflammatory intracellular contents. Release of these intracellular tumoral contents induces dendritic cell maturation and activation (7–9). Dendritic cells in turn traffic to lymph nodes where they activate cytotoxic T cells to recognize tumor elsewhere in the body. Dendritic cell activation is a critical aspect of the antitumoral immune response, and increased intratumoral dendritic cell infiltration has been associated with tumoral immunotherapy response (10–12). Through this dendritic cell activation, it is hypothesized that cryoablation can lead to increased CPI response.

The use of cryoablation as an adjunctive cancer immunotherapy has only been studied preclinically in a limited fashion (13–16), evaluating the role of the adjunctive technique only in the context of single-agent CPI (either anti–cytotoxic T-lymphocyte–associated protein 4 [CTLA-4] monotherapy or anti–programmed cell death protein 1 [PD-1] monotherapy). There is long-standing clinical interest in inducing the abscopal effect which has been noted rarely when cryoablation is used in isolation (17). Prospective clinical evaluation of cryoablation and checkpoint inhibition has recently begun, with recent small-scale studies reported in the neoadjuvant setting for breast cancer (18), as well as in metastatic prostate cancer (19) and renal cell carcinoma (20). The recent clinical trial of cryoablation combined with CPI (the CTLA-4 inhibitor tremelimumab) has reported results in metastatic renal cell carcinoma, demonstrating safety of the treatment and showing an increased immune infiltrate with the clear cell subtype. However, the study did not demonstrate superior efficacy of the combined treatment relative to tremelimumab alone.

We hypothesized that adjunctive cryoablation may be of particular benefit in tumors with high mutation rates due to its ability to expose tumor antigens to antigen-presenting cell surveillance. Furthermore, we hypothesized that cryoablation is most likely to be beneficial in the context of dual-agent CPI, which promotes both adaptive immune system priming and continued cytotoxic T-cell activity. In a dual-tumor murine model using the hypermutated MC-38 murine tumor line (21), we aimed to evaluate the impact of partial cryoablation on the CPI response rate in treated nonablated tumors. Identical analysis was also performed in the CT-26 murine model. We also characterized the postcryoablation immune response in both the ablated and nonablated tumors using flow cytometry, cytokine analysis, and immunohistochemistry analyses.

Materials and Methods

Material support in the form of a Visual Ice Cryoablation System and animal cryoprobes were provided by Boston Scientific. The authors maintained complete control of the data and the information submitted for publication.

Cell Culture

MC-38 (Kerafast), a murine colon adenocarcinoma line was maintained in Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum and 1 mmol of l-glutamine. An additional murine colon adenocarcinoma line, CT-26 (ATCC), was maintained in Roswell Park Memorial Institute medium with 10% fetal bovine serum media. Cells were tested for mycoplasma by the polymerase chain reaction method and used within 6 months of purchase. Cells (~5 × 10⁵ cells [MC-38] or ~1 × 10⁶ cells [CT-26]) were implanted in each of the bilateral upper flanks of mice.

Animal Studies

All mice were housed and maintained by the Center for Comparative Medicine, and all experiments were approved by the Massachusetts General Hospital institutional animal care and use committee. Equal portions of male and female 6–8-week-old C57BL6 mice implanted with MC-38 tumors were randomized to treatment with a vehicle or 200 µg of anti–PD-1 (clone RMP1–14; Bio X Cell) and 100 µg of anti–CTLA-4 (clone 9H10; Bio X Cell) (referred to as CPI here on) and additional treatment with single tumor cryoablation (Fig 1A). An identical treatment protocol was performed for BALB/c mice implanted with CT-26 tumors. Investigators (E.W.K., E.A., A.M., and P.H.) were aware of group allocation at the time of allocation and during conduct of the experiment.

Figure 1:

Dual-tumor adjunctive cryoablation model. (A) Mice are implanted with a tumor in the bilateral upper flanks. Mice are treated with intraperitoneal (IP) injection of a programmed cell death protein 1 (PD-1) inhibitor and cytotoxic T-lymphocyte–associated protein 4 (CTLA-4) inhibitor on days 3, 6, and 9 with partial cryoablation on day 8. (B) Partial cryoablation of a single tumor proceeds under direct visual guidance after exposure of tumor to ensure that ~50% of tumor is cryoablated. Partial cryoablation is performed to allow for preservation of some vascular and lymphatic architecture to facilitate tumoral antigen sampling by antigen presenting cells. (C) Histologic image following cryoablation demonstrates the region of cryoablation partially encompassing the tumor to confirm the percentage of cryoablation achieved using this technique. (D) Analysis of six tumors demonstrates a median of 53% ± 15 (SD) tumoral cryoablation. SC = subcutaneous.

Cryoablation

Cryoablation of a single tumor occurred on day 8 following tumor inoculation, when tumors were 6–10 mm in size. Cryoablation was conducted using a Visual Ice Cryoablation System with a small animal cryoprobe (Boston Scientific). Mice were anesthetized with inhaled isoflurane. A single tumor was exposed. Subsequently, the small animal cryoprobe was inserted into the tumor under direct visual guidance. Cryoablation proceeded at full strength with a 30-second freeze cycle followed by 2-minute thaw cycle and then a repeat 30-second freeze cycle. In all instances, the cryoablation goal was to achieve approximately 50% freezing of the tumor as assessed by continuous visual inspection (Fig 1B), and the freeze cycle length could be adjusted as appropriate to achieve desired partial freezing. This cryoablation protocol was developed by the authors based on preliminary experience with the animal cryoablation probe, and two cycles was chosen to conform with the cryoablation protocols in most common clinical use. Our rationale for incomplete freezing was to preserve some degree of vascular and lymphatic architecture within the tumor to allow for efficient sampling of the cryoablated tumor by the immune system. All cryoablations were performed by E.W.K. with 5 years’ experience with cryoablation and P.H. with 1 year of experience with cryoablation. To quantify the degree of tumor freezing achieved with this protocol, a group of six mice were treated with cryoablation according to protocol. Mice were euthanized the same day, and their tumors were excised. Flash-frozen paraffin-embedded tumor slices were then stained with hematoxylin-eosin, and three tumor slices per tumor were analyzed for degree of tumor kill using regions of interest drawn with ImageJ software (version 1.52). Histologic confirmation of our partial cryoablation methodology is demonstrated in Figure 1C and 1D.

Tumor Growth Studies

Following treatment as described, tumor measurements were performed by caliper twice weekly for 60 days, with a subset of the MC-38 mice followed for 90 days. For mice undergoing partial cryoablation, growth of the contralateral tumor was followed until the end of the study or when volumes reached a size greater than 1500 mm³ using the formula volume = (width² × length)/2. The primary outcome was mouse survival, with the experiment conducted for 60 days following tumoral injection. A minimum sample size of 16 mice was chosen with the assumption of 50% effect size, SD of 10%, type 1 error rate of 5%, and power of 80%. Survival experiments in CPI only and CPI plus cryoablation arms were repeated. Survival studies were repeated in CT-26 mice.

Tumoral Cytokine Analysis

In groups of four, tumoral cytokine analysis was performed on day 8 prior to cryoablation or on the cryoablated tumor at 12 hours or 48 hours following cryoablation. Supernatant from tumoral single-cell suspensions was analyzed using a type 1 T helper, type 2 T helper, and type 17 T helper murine-specific ProcartaPlex Multiplex Immunoassay (Thermo Fisher Scientific), and the experiment proceeded as per manufacturer protocol. Samples were analyzed on a Luminex 100/200 instrument. The concentration of cell suspension samples was calculated by plotting the expected concentration of each standard against the mean fluorescent intensity generated by each standard. A four-parameter logistic algorithm was then used to generate the best curve fit, and sample values were interpolated accordingly.

Contralateral Tumor Flow Cytometry

Additional groups of four mice were randomized to either vehicle, cryoablation only, CPI only, or CPI with cryoablation. Four days following cryoablation, mice were euthanized, and contralateral tumors were excised. Tumors were processed into a single-cell suspension. Cell viability was assessed by trypan blue staining, and 1 × 10⁶ viable cells were stained for flow cytometry analysis. Cells were washed with phosphate-buffered saline and stained with a 1:1000 solution of Zombie Brilliant Violet (from Zombie Violet Fixable Viability Kit; BioLegend). Subsequently, the following extracellular antibodies were added: CD45.2-FITC (clone 104; BioLegend), with either CD8a-PE/Cy7 (clone 53–6.7; BioLegend) and PD-1-PE (clone RMP1–14; BioLegend) or CD4-PE/Cy7 (clone GK1.5) and CD25-APC (clone 3C7; BioLegend). Cells were then washed, fixed, and permeabilized. Next, fluorochrome-conjugated intracellular antibodies were added for the CD8 panel (granzyme B [GZB] PE-Dazzle [clone QA16A02; BioLegend] and eomesodermin [EOMES] eFluor660 [clone Dan11mag; Thermo Fisher Scientific]) and the CD4 panel (forkhead box P3 [FOXP3] PE [clone MF-14; BioLegend]). After intracellular staining, cells were resuspended in cell-staining buffer, and the samples were analyzed on a BD LSRFortessa X-20 flow cytometer (BD Biosciences).

Data were gated using FlowJo software (v10.8.1). Doublets and cell clumps were excluded by gating along the 1:1 line for forward scatter height versus forward scatter area. Cellular debris and dead cells were excluded by side scatter area versus forward scatter area and viability stain, and immune cells were then selected based on CD45 expression. Gating for individual markers was determined by fluorescent-minus-one control panels and unstained controls. Gates were then confirmed using backgating.

Ipsilateral (Cryoablated) and Contralateral Tumor Immunohistochemistry

Additional groups of four mice were treated with CPI plus cryoablation. Mice were euthanized on either day 8 immediately prior to cryoablation, day 12 (4 days following cryoablation), or on day 16 (8 days following cryoablation). Partially cryoablated and contralateral tumors were excised and preserved in formalin. Flash-frozen paraffin-embedded tumor samples were analyzed via hematoxylin-eosin or immunofluorescence. Immunofluoresence samples were stained with 4′,6-diamidino-2-phenylindole (DAPI) (Thermo Fisher Scientific), anti-GZB Alexa Fluor 546 (sc-8022; Santa Cruz Biotechnology), anti-CD8 Alexa Flour 647 (sc-1177; Santa Cruz Biotechnology), and adjacent cross-sections were stained with anti-CD11b AlexaFlour 546 (sc-53086 AF546; Santa Cruz Biotechnology).

Statistical Analysis

All statistical analysis was performed using GraphPad Prism software (version 8). For survival analyses, comparison was made using log-rank (Mantel-Cox) test. Flow cytometry comparisons were made using multiple t tests with false discovery rate correction for multiple comparisons using the Benjamini, Krieger, and Yekutieli method. P values are reported following false discovery rate correction, with P < .05 considered statistically significant. Cytokine and chemokine comparisons were made using two-way repeated measures analysis of variance with Tukey post hoc multiple comparison test correction. A Pearson correlation test was used when assessing correlations between cytokines. Statistical significance was set at P < .05.

Results

Adjunctive Cryoablation Significantly Increases Survival Relative to CPI Alone in Two Murine Adenocarcinoma Models

MC-38–injected mice had group sizes as follows: 16 mice for vehicle, 16 mice for unilateral cryoablation alone, 36 mice for CPI alone, and 28 mice for CPI plus adjunctive unilateral cryoablation. Median survival time for vehicle-treated mice was 14 days. Median survival time for mice treated with single cryoablation alone was 15 days. Median survival time was not reached for CPI only treated or CPI plus cryoablation treated mice. At 60 days, the survival rate of mice treated with vehicle or unilateral cryoablation alone was 0%, whereas CPI only was 61% (22 of 36) and CPI plus unilateral cryoablation was 79% (22 of 28) (Fig 2A). The difference in survival rates between CPI only and CPI plus unilateral cryoablation groups was significant by log-rank test (P ≤ .001). In one iteration of our survival experiment, mice alive at 60 days were observed for an additional 30 days. All mice remained alive at 90 days without recurrence of tumor (data not shown). Growth curves of contralateral tumors in CPI and CPI plus cryoablation mice are shown in Figure 2C and 2D, demonstrating durable tumor regression among surviving mice in both partially cryoablated and contralateral tumors. We note that tumor response appears binary. Tumors either respond completely to therapy with durable tumor regression or undergo progressive tumor growth without evidence of response.

Figure 2:

Adjunctive cryoablation synergizes with checkpoint inhibitors (CPI) to significantly improve the survival rate in murine dual-tumor models. (A) Graph shows that MC-38 injected mice treated with dual CPI and adjunctive partial cryoablation of a single tumor had a 79% survival rate over a 60-day study compared with a 61% survival rate in mice treated with dual CPI alone (P ≤ .001). The brief immune stimulus of cryoablation alone was insufficient to significantly improve survival relative to vehicle. Survival depends on both local control of the cryoablated tumor as well as contralateral (noncryoablated) tumor. (B) Graph shows that CT-26–injected mice treated with dual CPI plus partial cryoablation had a 57% survival rate at 60 days compared with a 35% survival rate in mice treated with dual CPI alone (P = .04). (C, D) Growth curves from a single repetition of the MC-38 survival experiment show durable regression in both partially cryoablated (C) and contralateral tumors (D). Cryo = cryoablation.

Mice injected with CT-26 cells were treated with CPI alone (31 mice) or CPI plus adjunctive unilateral cryoablation (30 mice) and followed for 60 days. Median survival time for mice treated with CPI alone was 27 days, and the median survival time for mice treated with CPI plus unilateral cryoablation was undefined. At 60 days, the survival rates were 35% (11 of 31) for mice treated with CPI only and 57% (17 of 30) for mice treated with CPI plus unilateral cryoablation. The difference between groups was significant by log-rank test (P = .04) (Fig 2B).

Cryoablation Increases Chemokines Central to Inducing Innate Immune Response

Cytokine analysis using a panel of 26 relevant cytokines and chemokines revealed multiple significant differences in proinflammatory and tolerogenic cytokines that developed in the first 12 hours and subsequently 48 hours following cryoablation, as shown in Figure 3 and the Table. At 48 hours, the fold change in mean chemokine concentration was most pronounced in CCL3 (1143-fold increase), CXCL2 (338-fold increase), CCL4 (253-fold increase), IL-4 (170-fold increase), CXCL10 (92-fold increase), and CXCL1 (62-fold increase). Additionally, the following cytokines demonstrated a greater than or equal to 25-fold increase: IL-6 (39-fold), IL-1b (34-fold), IL-17a (28-fold), CCL7 (26-fold), IL-13 (25-fold). The Pearson correlation coefficient demonstrated a high correlation coefficient between 12-hour and 48-hour cytokine and chemokine concentrations (r = 0.93) (Fig S1).

Figure 3:

Partial cryoablation induces significant changes in multiple inflammatory cytokines within remaining treated tumor. (A, B) Graphs show multiplex analysis of 26 cytokines within the partially cryoablated tumor at 12 hours and 48 hours following cryoablation that demonstrates significant increases in multiple cytokines and chemokines relative to control, predominantly among those responsible for initiating a type 1 T helper cell response. Simultaneous decreases in several tolerogenic cytokines were observed (A). The following cytokines demonstrated the greatest fold increase: CCL3, CXCL2, CCL4, IL-4, CXCL10, and CXCL1 (B). These cytokines and chemokines are primary chemoattractants that mediate the early innate immune response, with exception of the tolerogenic cytokine IL-4.

Cytokine Concentrations in Ablated Tumors

Cryoablation Increases Immune-cell Activation within Distant Tumor

The immune microenvironment of the distant (noncryoablated) tumor was analyzed on the day of cryoablation and 4 days following cryoablation by flow cytometry to assess for changes in the T-cell immune infiltrate that might suggest changes in antitumoral systemic immunity induced by cryoablation.

Four days following cryoablation, significantly more activated CD8 cells (GZB⁺/PD-1⁻/EOMES⁻) were observed within the contralateral tumors of mice that had undergone treatment with cryoablation and CPI relative to treatment with CPI alone (30.3% vs 5.9%; P = .006) (Fig 4). Conversely, within mice treated with cryoablation and CPI, significantly fewer CD8 cells were severely exhausted (GZB⁺/PD-1⁺/EOMES⁺) (22) within tumors of cryoablation- and CPI-treated mice relative to CPI only treated mice (7.8% vs 20.4%; P = .005). The addition of cryoablation did not significantly increase the percentage of regulatory T cells (CD4⁺/FOXP3⁺/CD25⁺) compared with CPI only treatment (11.0% vs 8.4%; P = .38).

Figure 4:

Graphs of flow cytometry results from distant tumors show highly active immune infiltrate following cryoablation. Four days following cryoablation, the immune infiltrate within the distant (noncryoablated) tumor was analyzed. Adjunctive cryoablation (A) significantly increased the percentage of active CD8⁺ T cells (granzyme B [GZB]⁺/protein cell death protein 1 [PD-1]⁻/eomesdermin [EOMES]⁻) relative to CPI treatment alone (30.3% vs 5.9%; P = .006) and (B) significantly decreased the percentage of highly exhausted (GZB⁺/PD-1⁺/EOMES⁺) CD8⁺ T cells (7.9% vs 20.4%; P = .005). Additionally, the percentage of regulatory T cells (CD4⁺/forkhead box protein P3 [FOXP3]⁺/CD25⁺) was not significantly increased with adjunctive cryoablation relative to CPI treatment alone (11.0% vs 8.4%; P = .38) (C). CPI = checkpoint inhibitor, Cryo = cryoablation.

Histologic characterization of cryoablated and distant tumors demonstrated robust myeloid and CD8 infiltrate.

Cryoablated and distant tumors were resected at multiple time points to illustrate the time course of immune infiltration (Fig 5). Immediately prior to cryoablation, comparison hematoxylin-eosin and immunofluorescence analyses showed a minimal immune infiltrate within tumor without a substantial portion of activated CD8 T cells, as evidenced by minimal evidence of GZB staining, and minimal myeloid cell presence.

Figure 5:

Adjunctive cryoablation induces robust immune response within cryoablation margin and contralateral tumor. Partially ablated and contralateral (nonablated) tumors shown on day of cryoablation (day 8), 4 days following cryoablation (day 12), and 8 days following cryoablation (day 16). Images (×40 magnification) show hematoxylin-eosin (H&E) stain (top row), CD8 and granzyme B immunofluorescence (middle row), and CD11b (bottom row). (A–C) Histologic images show rare CD8 infiltrate with granzyme B on day of cryoablation. (D–F) Images 4 days following cryoablation show a robust CD8 and myeloid infiltrate within the partially cryoablated tumor with significant expression of granzyme B. (G–I) Images show that cytotoxic activity appeared to continue, albeit at a less impressive pace, 8 days following cryoablation. Within the contralateral tumor, we also observed a robust cytotoxic CD8 T-cell infiltrate producing granzyme B involving most of the tumor (K) without associated myeloid infiltrate (L). Four days later, there were large areas of necrosis (M) and residual granzyme B activity (N), with an increased myeloid infiltrate along the areas of necrosis.

Within the cryoablated tumor, examination of the cryoablation margin on day 12 (Fig 5D–5F) showed necrosis within the cryoablated portion on the right side of Figure 5D. Within the residual tumor, there was a robust and highly active CD8 infiltrate secreting GZB, as well as a robust CD11b⁺ myeloid infiltrate. Four days later, the cryoablation margin continued to demonstrate cytotoxic activity but with a notable decrease in myeloid cell activity. Reduction in myeloid activity likely represents waning of the initial innate immune response to cryoablation.

Histologic analysis of the contralateral (noncryoablated) tumor on day 12 (Fig 5J–5L) also demonstrated a CD8 infiltrate throughout the tumor with evidence of cytotoxic activity, which appears subjectively less robust than the degree of activity within the partially cryoablated tumor at the same time point, and without significant myeloid infiltrate. By day 16 (Fig 5M–5O), cytotoxic T-cell activity within tumor had generated areas of necrosis (bottom left of Fig 5M–5O). At the margins of necrotic regions, there was a new influx of CD11b⁺ myeloid cells.

Discussion

The possible immune effects of cryoablation have been previously reported, but the ability of cryoablation to activate the antitumoral immune response in the context of checkpoint inhibition is understudied. A preclinical study reported on the use of cryoablation with concurrent use of a CTLA-4 inhibitor in a single tumor model and found that upon tumor rechallenge, 44% of combination-treated mice rejected the tumor, whereas all mice in other experimental groups died of the tumor (15). Recently, two additional preclinical reports have examined the efficacy of cryoablation and single-agent CPI in the context of dual-injected prostate and renal malignancies (13). Evaluating response in the nonablated tumor, the prostatic cancer model study showed no significant increase in survival relative to PD-1 inhibition alone, and the renal cell carcinoma model study showed a modest growth delay. In contrast to these studies, our study showed that the addition of cryoablation to dual-agent CPI in an MC-38 dual tumor model resulted in a significant improvement in efficacy, with improvement in survival rate from 61% with dual-agent CPI to 79% with the addition of adjunctive cryoablation (P < .001). In a CT-26 model, we also found a significant increase in survival rate, from 33% to 57% (P = .04).

There are several differences in study design that may explain the differences in response rate. First, our study examines the role of cryoablation in increasing response to CPI in models of moderate to highly CPI-responsive cancer. MC38 tumors are a microsatellite instability–high murine colon cancer cell line. Microsatellite instability–high tumors have intrinsic defects in DNA replication and thus produce more neoantigens that are correlated with response to CPI therapy (23,24). As cryoablation causes release of tumor neoantigens into the tumor microenvironment, it is more likely to increase immune system recognition of a tumor expressing a high number of such proteins. The other model investigated in our study, CT-26, is a microsatellite stable model assumed to present fewer neoantigens and is generally considered moderately responsive to checkpoint inhibition.

Additionally, we evaluated the value of cryoablation in the context of dual CPI addition (PD-1, CTLA-4), whereas all other studies have examined cryoablation with either single-agent CTLA-4 or PD-1 inhibitors. As the two inhibitors play a distinct role in T-cell priming and activation (25,26), treatment with both may increase the chance that the tumor antigens exposed by cryoablation lead to an increased T-cell repertoire that then can retain antitumoral activity. This regimen has previously demonstrated usefulness in distinguishing therapy responders from nonresponders in CT-26 mice and evaluating the ability of additional interventions to increase the proportion of responding tumors (27). It has also been valuable in multiple prior investigations using GZB PET imaging to distinguish responding from nonresponding tumors (28–30).

There is unresolved debate in the field as to the optimal timing of cryoablation relative to CPI administration. We chose to perform cryoablation after initiation of CPI for two reasons. First, to the extent that there may be rebound immunosuppressive signaling following the strong immune stimulus of cryoablation (31), we hypothesized that the presence of anti-PD1 therapy would blunt the effects of these signals upon cytotoxic T cells. We additionally hypothesized that it is also beneficial to have anti–CTLA-4 inhibition initiated prior to cryoablation to facilitate T-cell priming as tumoral antigens are released.

Furthermore, we have modified our cryoablation technique from that of conventional cryoablation. Rather than cryoablating the entire tumor as per the standard clinical technique, we chose to freeze only approximately 50% of the tumor. The rationale behind this choice was to leave partial vascular and lymphatic architecture in place, which should allow for more rapid trafficking of antigen-presenting cells and efficient sampling of tumor antigens than complete cryoablation. This approach is also similar to the clinical strategy employed by Duffy et al (32) in their clinical evaluation of the immune effect of subtotal radiofrequency ablation combined with the checkpoint inhibitor tremelimumab for the treatment of hepatocellular carcinoma. The robust cytotoxic and myeloid infiltrate at the ablation margin supports our hypothesis, showing robust trafficking of immune cells into the partially cryoablated tumor, with significant GZB secretion serving as evidence of the adaptive immune system (30).

Our investigation of biologic effects within the cryoablated tumor demonstrated marked changes in cytokine and chemokine expression in the first 48 hours following cryoablation. In the panel of 26 potentially relevant chemokines and cytokines studied, statistically significant changes were observed across multiple chemokines and cytokines, findings consistent with an early inflammatory immune response informing a type 1 T helper response. The magnitude of change in certain chemokine families is particularly notable. CCL3 (1143-fold increase), CCL4 (253-fold increase), and CXCL10 (92-fold increase) are related chemokines critical for recruitment of dendritic cells, natural killer cells, and T cells (33–35). We have previously shown that increases in tumoral CCL3 and CXCL10 are positively correlated with response to immunotherapy treatment in both MC-38 and CT-26 tumor lines (28). In addition to marked up-regulation of proinflammatory cytokines, we also observed an increase in the tolerogenic cytokines IL-4, IL-13, and IL-17 (36). The effect of increase in these cytokines must be weighed against the concurrently observed significant decrease in other tolerogenic cytokines following cryoablation, including IL-10, IL-22, and IL-23 (37,38). The balance of cytokine and chemokine production appears to favor the development of antitumoral immunity.

Our flow cytometry analysis suggests that changes in chemokine and cytokine expression induced by the combination of cryoablation and CPI results in a vigorous adaptive antitumoral immune response within the nonablated tumor. In comparison to CPI alone, the combination of cryoablation and CPI resulted in a higher percentage of active CD8 T cells (GZB⁺/PD-1⁻/EOMES⁻) and a smaller percentage of exhausted CD8 T cells (GZB⁺/PD1⁺/EOMES⁺). Furthermore, we did not see an increase in regulatory T cells (CD4⁺/CD25⁺/FOXP3⁺). The findings suggest that the addition of cryoablation to CPI can broaden the antitumor immune response as well as invigorate exhausted T cells without induction of a tolerogenic response.

Our study had important limitations. Studies of subcutaneously injected tumors have difficulty representing the complexity of antitumoral immune response. For example, in response to CPI, subcutaneous murine tumors have been shown to have a binary response of either regression or progression, a simplified response compared with the full spectrum of human tumoral response to immunotherapy (27). Our study is further limited by the use of a dual-injected tumor model to represent metastatic disease. This model has multiple sites of disease and is amenable both to minimally invasive percutaneous procedure and monitoring of treatment response, but it does not fully replicate the tumor immune microenvironment found within metastatic disease. Also, the technique of cryoablation with exposure of the tumor and treatment under visual inspection differs from the clinical scenario of image-guided treatment.

In conclusion, our preclinical results argue for more extensive evaluation of cryoablation’s potential to increase CPI response rate, as well as the degree and duration of response. The addition of a single minimally invasive intervention with low morbidity can potentially significantly increase response to CPI therapies in carefully selected scenarios. More work is necessary to better define the biology of cryoablation and CPI, the optimal sequencing of therapies, and biomarkers to identify those tumor types most likely to benefit from adjunctive cryoablation.