Granzyme B PET Imaging of Combined Chemotherapy and Immune Checkpoint Inhibitor Therapy in Colon Cancer

Purpose Chemotherapeutic adjuvants, such as oxaliplatin (OXA) and 5-fluorouracil (5-FU), that enhance the immune system, are being assessed as strategies to improve durable response rates when used in combination with immune checkpoint inhibitor (ICI) monotherapy in cancer patients. In this study, we explored granzyme B (GZB), released by tumor-associated immune cells, as a PET imaging-based stratification marker for successful combination therapy using a fluorine-18 (18F)-labelled GZB peptide ([18F]AlF-mNOTA-GZP). Methods Using the immunocompetent CT26 syngeneic mouse model of colon cancer, we assessed the potential for [18F]AlF-mNOTA-GZP to stratify OXA/5-FU and ICI combination therapy response via GZB PET. In vivo tumor uptake of [18F]AlF-mNOTA-GZP in different treatment arms was quantified by PET, and linked to differences in tumor-associated immune cell populations defined by using multicolour flow cytometry. Results [18F]AlF-mNOTA-GZP tumor uptake was able to clearly differentiate treatment responders from non-responders when stratified based on changes in tumor volume. Furthermore, [18F]AlF-mNOTA-GZP showed positive associations with changes in tumor-associated lymphocytes expressing GZB, namely GZB+ CD8+ T cells and GZB+ NK+ cells. Conclusions [18F]AlF-mNOTA-GZP tumor uptake, driven by changes in immune cell populations expressing GZB, is able to stratify tumor response to chemotherapeutics combined with ICIs. Our results show that, while the immunomodulatory mode of action of the chemotherapies may be different, the ultimate mechanism of tumor lysis through release of Granzyme B is an accurate biomarker for treatment response. Supplementary Information The online version contains supplementary material available at 10.1007/s11307-021-01596-y.


Introduction
Tumors exploit immune checkpoint receptors to evade the immune system. Therapeutic immune checkpoint inhibitors (ICIs) activate lymphocytes, including T cells and NK cells to mount an effective immune response. Activated CD8 T cells and NK cells release Granzyme B leading to apoptosis of the tumor cells [1]. However, due to resistance and suppression mechanisms, the majority of patients do not show a durable response to ICIs [2][3][4]. Many combination clinical trials are currently underway driven by the hypothesis that cytotoxic chemotherapies may enhance responsiveness to ICIs by increasing tumor immunogenicity. Chemotherapeutics were originally thought to be immunosuppressive; however, recent studies show that many agents enhance antitumor effects by activating the immune system. Chemotherapies may promote tumor immunogenicity either by inducing immunogenic cell death as part of their therapeutic effect or by enhancing tumor antigen presentation or upregulating co-stimulatory molecules/downregulating co-inhibitory molecules expressed on the tumor cell surface, such as PDL-1 [5,6]. 5-fluorouracil (5-FU) and oxaliplatin (OXA) are commonly used as first-line chemotherapeutics for the treatment of colorectal cancer [7] and have been shown to modulate the immune system [8][9][10]. Recent clinical trials have shown that chemotherapy combined with ICIs leads to an improvement in overall survival compared to ICI monotherapy alone (KEYNOTE trials 048, 189, 407) [11][12][13], potentially due to synergies in their immune-related mechanisms of action.
Currently, there is a lack of specific biomarkers capable of providing a readout in situ of immune responses to different treatment strategies, complicating interpretation of clinical trials comparing different treatment strategies such as chemotherapy and ICIs. In the current study, we have assessed the ability of [ 18 F]AlF-mNOTA-GZP (Fig. 1A), a peptide probe targeting granzyme B to serve as a PET imaging biomarker of combined ICI/chemotherapy in a syngeneic mouse model of colon cancer. PET imaging results were paired with multicolour flow cytometry analysis of the tumorassociated immune cell subsets for an in-depth comparative assessment of which immune cell types expressing GZB are best associated with tumor response.

Materials and Methods
General H-Asp(OtBu)-H NovaSyn TG resin (0.21 mmol/g) was obtained from Merck. Fmoc-amino acids, HATU and HOAt were obtained from Advanced Chemtech. Fmoc-glutamic acid was t-butyl protected. (p-SCN-Bn)-NOTA was purchased from Boc Sciences and Macrocyclics. Glacial acetic acid was purchased from JT Baker.  18 F] nuclear reaction (GE PETtrace 860 cyclotron). Quality control analytical radio-HPLC was performed on a UFLC Shimazdu HPLC system equipped with dual-wavelength UV detector and a NaI/ PMT-radiodetector (Flow-Ram, LabLogic). Radioactivity measurements were made with a CRC-55tPET dose calibrator (Capintec, USA).

Animal Procedures
All animal procedures were carried out following the Institutional Animal Care and Use Committee Singapore (IACUC No. 181399) and conformed to the US National Institutes of Health (NIH) guidelines and public law. BALB/ c mice aged 6-8 weeks were purchased from In Vivos (Singapore). Mice were housed in specific-pathogen-free (SPF) environment during the experiments, at room temperature with a 12-h light-dark cycle and had free access to food and water.
The murine colon tumor cell line CT26 was acquired from ATCC and cultured in RPMI supplemented with 10% foetal bovine serum, 100 U/mL penicillin and 100 μg/mL streptomycin, at 37°C in a humidified atmosphere at 5% CO 2 . CT26 cells (2 × 10 5 ) were prepared in a 1:1 (v:v) ratio in Matrigel (Sigma) and injected subcutaneously into the right shoulder of Balb/c mice. In vivo subcutaneous tumors were measured by callipers on days 6,9,12,15,19, and 21 after tumor inoculation. Tumor volume was then calculated using the modified ellipsoid formula 1/2(Length × Width 2 ) [16].
In order to accurately assess tumor response to therapy tumor growth inhibition (%TGI) was determined using the formula %TGI = (V c -V t )/(V c -V o ) × 100, where V c and V t are the mean tumor volumes of control and treated groups on day 21 and V o is the tumor volume at the start of the study (Supplementary Table S3).

PET-CT Imaging
Animals were imaged 14 days after tumor inoculation using a Siemens Inveon PET-CT. Briefly, animals were anesthetized using inhalational isoflurane (maintained at 1.5% alveolar concentration) and injected with [ 18 F]AlF-mNOTA-GZP (~10MBq) via the lateral tail vein. Static PET acquisitions were performed at 60-80 min postinjection (p.i.) and CT scans were used for co-registration. Animals were monitored for maintenance of body temperature and respiration rate during imaging studies using the Biovet physiological monitoring system. Post-analysis of reconstructed calibrated images was performed with FIJI and Amide software (version 10.3 Sourceforge). Uptake of radioactivity in tissues was determined by the placement of volumes of interest (VOI) delineated by CT imaging. Data are expressed as % of the injected dose per gram (%ID/g) of tumor tissue in the VOI.

Dimension Reduction Analysis
Time-gated, size-gated, Live, singlet, CD45 positive cells from 46 fcs files were exported from FlowJo and used for dimension reduction analysis. t-Distributed Stochastic Neighbor Embedding (t-SNE) was used for unbiased dimension reduction and Rphenograph was used for clustering. t-SNE, clustering and overlay with t-SNE maps were performed with the cytofkit package in RStudio [17] (https:// github.com/JinmiaoChenLab/cytofkit). The default cytofkit parameters were used for the analysis on 1000 cells from each fcs file, for a total of 46,000 cells. The following markers were used for the Rphenograph clustering: CD3, CD4, CD8, CD11b, CD11c, CD206, F4/80, Granzyme B, I-A/I-E, Ly6C, Ly6G, Nkp46 and Siglec-F.

Statistical Analysis
Data were analyzed using a Kruskal Wallis 1-way ANOVA with a Dunn's post-test using GraphPad Prism version 8.0.0 for Windows, GraphPad Software, San Diego, California USA, www.graphpad.com, PG0.05 was considered statistically significant. Data are expressed as mean ± S.D. unless otherwise indicated.
Successful therapy response in this preclinical model was determined by the comparison of day 6 baseline tumor volumes with day 21 post therapy tumor volumes separating the treatment arms into two groups, treatment responders (combining complete responders and partial responders into a single group, TR) and treatment non-responders (TNR). This reductionist approach has been used previously to the assessment of the utility of imaging to stratify responders from non-responders but may introduce bias [15,[18][19][20][21]. TRs were identified as final tumor volumes less than 850 mm 3 and include tumors with stable or decreased volumes (tumor volumes are shown in Supplementary Table S1). The volume of 850 mm3 was chosen as this is more than 2 standard deviations from the mean volume of the control group on day 21, using this approach there is only a 5% chance for a TR to be incorrectly assigned. Treatment response varied between treatment arms (Supplementary  Table S2) with the combination treated groups exhibiting a greater response rate than the monotherapy groups.

Discussion
Chemotherapy response is classically defined by evaluating morphologic changes in tumor volume defined by CT or MRI following RECIST criteria [22]. However, tumors treated with ICIs may remain stable or even increase in size before ultimately responding to therapy [23]. This divergence adds complexity for on-going clinical trials attempting to quantify tumor responsiveness to ICIs when combined with chemotherapy [24]. The expectation is that tumors will respond better to the combination than to monotherapy; however, often combination therapy is no more effective than successful monotherapy, administering multiple drugs simply increases the chance of experiencing a meaningful anti-tumor response to any single drug in the combination [25]. While tumor growth inhibition (TGI) may provide a simple way to determine the rate of response to a therapy, it provides no information on whether therapies are working alone or synergistically. Both 5-FU and OXA are well-known to function as effective chemotherapeutic adjuvants inhibiting tumor cell proliferation; however, they can also exert immunomodulatory effects in numerous ways. 5-FU can facilitate antigen uptake by dendritic cells (DCs) and selectively kills monocyte-derived suppressor cells (MDSCs) while sparing other lymphocyte subtypes [9,10]. OXA upregulates PD-L1 expression on tumors and DCs and can induce immunogenic cell death (ICD), a form of apoptotic cell death associated with the release of damage-associated molecular patterns (DAMPs) [26]. These DAMPs, in combination with cancer antigens, induce maturation of dendritic cells and can lead to an adaptive immune response against tumor cells [27][28][29]. The immune effects for both 5-FU and OXA are observed at lower doses than typically used clinically, the doses chosen in the current study were designed to mimic this, the equivalent dose in humans has been shown to be minimally symptomatic [30] and in the animals displayed no side effects associated with high-dose chemotherapy. The effect of chemotherapy-induced changes in the immune environment and how they affect ICIs when given in combination is difficult to quantify using the standard measure of tumor growth inhibition.
Non-invasive molecular imaging with radiolabelled GZBtargeting peptides has been shown to stratify response to ICIs administered alone or in combination [14,15,20,31]. The current study, however, is the first to show that GZB targeting peptides can stratify response to chemotherapies that exert an immunostimulatory effect either alone or in combination with ICIs. Tumor uptake of [ 18 F]AlF-mNOTA-GZP was well correlated to tumor growth inhibition across the treatment arms and when separated by treatment response, uptake of [ 18 F]AlF-mNOTA-GZP was significantly higher in PD1, OXA and 5-FU responsive tumors compared to TNRs (Table 1, Fig. 3B). Tumors responsive to chemo-ICI combinations PD1+OXA or PD1+5-FU showed even greater uptake than those treated with mono therapy alone ( Table 1, Fig. 3B, # P90.05). This additive effect may be caused by the recruitment of different immune cell types. Tumors that responded to OXA monotherapy showed significant increases in CD8+ and CD8+GZB+ TILs and decreases in F4/80 myeloid cells, a similar immune cell profile observed in responders to αPD1 monotherapy ( Table 2, Fig. 4). Tumors that responded to combined OXA and αPD1 therapy showed even greater increases in CD8+ and CD8+GZB+ TILs, a profile mirrored by increases in tumor retention of [ 18 F]AlF-mNOTA-GZP. 5-FU treatment, however, worked via a different mechanism of action, inducing no change in CD8+ TILs but instead a significant increase in tumor infiltrating NK+ and GZB+ NK+ cells. When administered as a combination, 5-FU and αPD1 showed no additive effect on NK+ cell infiltration ( Table 2, Fig. 4) but still showed increases in tumor retention of [ 18 F]AlF-mNOTA-GZP. The data indicate that in both cases response to therapy is driven by GZB+ cell infiltration, increases in GZB+ CD8+ TILs caused by both parts of the therapy when OXA and PD1 are combined and increases in GZB+ NK+ cells and GZB+ CD8+ cells independently induced by 5-FU and PD1 when used in combination. Further studies will be needed to determine whether stratification of tumor response is possible using other chemo-ICI combinations.
Much work is needed to determine whether the preclinical data outlined in the current study are clinically translatable. Previous studies have shown that a simple amino acid substitution (IEFD to IEPD) confers selectivity for human Granzyme B [20] versus other granzymes; however, it is not known whether IEPD is selective for Granzyme B over caspases. Typically, clinical assessment of therapy response is based on tumor size reduction, whether granzyme B imaging peptides will be able to distinguish between progressive disease and partial response is difficult to determine from studies based on tumor growth inhibition. Capsases, which also cut peptides with Asp/Glu at position 1, may be released when tumors debulk rapidly due to therapy response or necrosis and this may further complicate clinical assessment of these imaging peptides. In the current study, the tumor to blood ratio in TRs is 91.5 and the tumor to muscle ratio is 91.5 primarily due to rapid clearance from the blood; however, the overall tumor uptake is still low. Low tumor uptake will hamper the utility of granzyme B peptides in tumors located in organs with higher background such as those organs associated with peptide excretion including the bladder and kidneys and to a lesser extent the liver and intestines. Furthermore, granzyme B targeting peptides ability to stratify response has been shown to be affected by the background immune phenotype [15], patient tumors tend to be large and heterogeneous, with areas of necrosis introducing variability that may complicate clinical assessment.
Overall, the preclinical data in the current study suggests that while the immunomodulatory mode of action of the chemotherapies may be different, they both induce tumor lysis in part through the release of granzyme B and that the detection of this granzyme B release acts as biomarker for efficacy in the syngeneic tumors studied. If granzyme B targeting peptides can be successfully translated to the clinic they may provide information on the efficacy of chemotherapy-ICI combinations aiding in trials striving to enhance responsiveness to ICIs.