⁸⁹Zr-anti-CD8 immunoPET visualizes heterogeneous intratumoral CD8⁺ immune responses to treatment with radiation and anti-CTLA4

Abstract

Anti-CD8 immunoPET facilitates non-invasive, whole-body visualization of immune responses, and syngeneic preclinical models are a crucial tool for studying tumor infiltration of T cells in response to cancer therapies. This study characterized longitudinal CD8⁺ T cell responses in an orthotopic mouse model of breast cancer treated with radiation and anti-CTLA4 by immunohistochemistry and anti-CD8 immunoPET, confirming an early but heterogeneous response induced by combination treatment that is detectable by imaging.

Anti-CD8 immuno-positron emission tomography (CD8 immunoPET) is a useful tool to non-invasively study localization of CD8⁺ T cells during anti-tumor immune responses. In the preclinical setting, radiation treatment (RT) has demonstrated CD8⁺ T cell-mediated synergy with checkpoint inhibitors^1–3, and CD8 immunoPET has been studied in the context of treatment with checkpoint inhibitors as well as the combination of anti-CTLA4 with low-dose RT. One limitation of previous studies is the reliance on subcutaneous tumor models which may not fully recapitulate the natural tumor microenvironment^4–6. In early phase clinical trials with CD8 immunoPET, radiotracer uptake at the site of metastases at baseline or early after treatment with immunotherapy has been associated with clinical responses^7–9. However, considerable heterogeneity has been observed, underscoring the utility of whole-body imaging to better understand anti-tumor immune responses as well as the need for complementary preclinical models.

We have previously developed a CD8-specific cys-diabody that binds CD8α across multiple mouse strains and detects tumor-infiltrating CD8⁺ T cells^10,11. Here, we characterize the intratumoral CD8⁺ T cell response to RT combined with anti-CTLA4 and demonstrate feasibility of applying CD8 immunoPET in an orthotopic 4T1 mouse tumor model. Advantages of this model include implantation into a mammary fat pad for better approximation of the natural tumor microenvironment and spontaneous development of metastases to facilitate studies of abscopal effects¹².

The combination of radiation treatment and anti-CTLA4 induces an early CD8⁺ T cell immune response in an orthotopic 4T1 tumor model

Tumor growth was followed for 30 days in mice with orthotopic 4T1 tumors treated with RT and anti-CTLA4, RT alone, anti-CTLA4 alone, or no treatment. (Fig. 1A). Combination treatment was sequenced as previously published¹; anti-CTLA4 alone or RT alone was given such that tumors were of equivalent size when treatment started. Of 10 mice receiving combined treatment, 3 developed complete tumor regression about 1 week after treatment, while the rest achieved partial response with eventual tumor outgrowth (Fig. 1B). Consistent with previously published data, anti-CTLA4 alone did not induce tumor responses¹. Tumor volume with RT alone was significantly lower than anti-CTLA4 alone starting at day 12 (p < 0.001). The combination of CTLA4 with RT also showed a statistically significant difference (p < 0.05) at the final timepoint only, indicating a potential interaction that augments tumor regression when compared to RT alone.

Fig. 1. Tumor responses to combined treatment with anti-CTLA4 and 8 Gy x3 in an orthotopic 4T1 model.

A Experimental schema. B Left, growth curves for treatment groups; right, growth curves for individual tumors (untreated/anti-CTLA4 alone, n = 5; RT, n = 6; RT + anti-CTLA-4, n = 10). Error bars represent standard deviation of the mean.

As spontaneous lung metastases develop in the 4T1 tumor model ~2 weeks after implantation¹², lungs were evaluated at time of euthanasia. H&E staining demonstrated variable metastatic burden with no qualitative differences in CD8 staining, and lung tumors were found in all treatment groups (Supplemental Fig. 1).

To profile CD8⁺ T cell responses, tumors were harvested for IHC at specified timepoints during and after treatment (Fig. 2A). These included 1 day after completing RT and 1 dose of anti-CTLA4 as well as 1 day, 5 days, and 14 days after completing combination treatment with 3 doses of anti-CTLA4. Tumors for controls with anti-CTLA4 and RT alone were harvested relative to the corresponding timepoint 1 day after completing combination therapy with 3 doses of anti-CTLA4. Compared to untreated tumors, tumors that received combined treatment displayed a statistically significant increase in CD8⁺ cell density 1 day after completing 3 doses of anti-CTLA4 (Fig. 2B, C) compared to untreated tumors. However, staining was highly variable (Fig. 2C). No differences compared to untreated tumors were observed at all other time points (Fig. 2B).

Fig. 2. Longitudinal intratumoral CD8⁺ responses by IHC.

A Experimental schema. B Two-dimensional CD8⁺ cell density for tumors harvested at the specified timepoints. Mean cell count/mm² for tumors harvested 1 day after RT + anti-CTLA4 was 980 (n = 15) vs 94 for untreated tumors (n = 11, p < 0.0001). C Representative intratumoral CD8 staining from tumors for each timepoint. Two examples of tumors treated with RT + anti-CTLA4 (#1 and #2) illustrate variability in this group.

⁸⁹ Zr-Anti-CD8 immunoPET visualizes tumor-infiltrating T cells after treatment with radiation and anti-CTLA

Anti-CD8α cys-diabody was radiolabeled with ⁸⁹Zr for imaging. Radiolabeling efficiency across 3 experiments was >97%, and specific activity ranged from 0.23 to 0.37 MBq/µg (6.3 to 10.1 µCi/µg). As IHC suggested maximal CD8⁺ T cell infiltration early after the last dose of anti-CTLA4, imaging experiments were performed within 24–48 h after treatment, using untreated mice with size-matched tumors as controls (Fig. 3A). CD8⁺ T cells in tumors and lymphoid tissues were visualized with CD8 immunoPET in all mice (Fig. 3B). Consistent with IHC, PET signal in tumors was variable, ranging from 3.8 to 7.0%ID/cc, but the mean for treated versus untreated tumors was higher at 5.6 versus 4.2%ID/cc (Fig. 3C).

Fig. 3. CD8 immunoPET imaging of orthotopic 4T1 tumors.

A Experimental schema. B Representative coronal maximum intensity projections. Two examples (top and bottom scans) are shown for each observed pattern in the treated group—higher vs lower tumor PET signal—and in the control group. Pink arrow, tumor; yellow arrowheads, ipsilateral cervical, axial, brachial, inguinal lymph nodes; t, thymus; k, kidney; s, spleen; b, bladder; %ID, percent injected dose. C Mean %ID/cc was 5.6 versus 4.2 for treated and untreated tumors (n = 9 each). Error bars represent standard deviation of the mean. D Biodistribution studies. Mean %ID/g was 3.6 vs 2.3 for treated vs untreated tumors (p = 0.0008, n = 9 and 8, respectively). Whiskers represent minimum to maximum. Note: ipsilateral/contralateral axial/brachial nodes were only collected from 5 untreated and 5 treated mice; nodes from 1 untreated mouse and cervical nodes from 1 treated mouse not included due to presence of adipose tissue confounding weight.

After imaging, tumors and organs were harvested immediately for ex vivo biodistribution studies. Tumor uptake (%ID/g) was variable but higher on average for treated versus untreated tumors (Fig. 3D). There were no statistically significant differences in other organs, including lymphoid tissue. IHC was used to assess correlation of two-dimensional CD8⁺ T cell density of central tumor slices with %ID/cc for the entire tumor region of interest (ROI) on PET imaging (Fig. 4A, B).

Fig. 4. IHC for imaged tumors.

A CD8 staining for peripheral and central areas of tumors corresponding to PET images in Fig. 3. Each row includes two examples (left two columns and right two columns) from each observed pattern in the treated group—higher vs lower tumor PET signal—and in the control group. Each scan represents a single IHC slice taken from the central axis of each tumor. B %ID/cc for tumor ROIs on imaging plotted against two-dimensional quantification of CD8⁺ cell density. Quantification was performed on the entire tumor slice.

Early tumor uptake of ⁸⁹Zr-anti-CD8 1 day after completing treatment may not predict tumor response to radiation and anti-CTLA in an orthotopic 4T1 model

In a pilot study to investigate correlation of early PET signal after treatment with tumor response as a potential predictive biomarker, a cohort of 12 mice treated with RT and anti-CTLA4 was imaged 1 day after treatment (Supplemental Fig. 2A, B). Tumor size was monitored for 20 days after imaging based on expected tumor growth trajectories identified in the previous experiments; 2 of 12 mice experienced complete tumor regression.

Consistent with previous experiments, the intratumoral PET signal was variable, ranging from 3.0 to 5.6%ID/cc. No difference in tumor uptake was observed between complete (n = 2) and partial responders (n = 10) (Supplemental Fig. 2B, C), and did not correlate with tumor volume at the final endpoint (Supplemental Fig. 2), suggesting no relationship between early tracer uptake and eventual tumor outcome. In the two mice with apparent complete responses by caliper measurement, small residual tumor tissue was found during dissection. IHC staining showed higher CD8⁺ cell density compared to tumors harvested from the 10 partial responders (Supplemental Fig. 2D).

Although studies have shown that combining RT with immunotherapy induces tumor infiltration of CD8⁺ T cells^1–3,6, the time course of this response has not been well-characterized, particularly in an orthotopic setting. While checkpoint inhibitors targeting PD1/PDL1 have demonstrated improved clinical efficacy, particularly in combination with anti-CTLA4, anti-CTLA4 + RT was chosen for this study as it is a well-established preclinical model that does not exhibit complete tumor inhibition^1,13. This heterogeneity allows for characterization of CD8 immunoPET across a spectrum, facilitating future exploration of combinatorial strategies. Additionally, the optimal radiation dose and fractionation regimen for immunomodulation is not well defined and may be context-dependent; the regimen used in this study has previously demonstrated synergy with anti-CTLA4¹, and future studies may be able to apply CD8 immunoPET to investigate the effects of different regimens on CD8⁺ T cell distribution.

In the 4T1 model described here, increased intratumoral CD8⁺ T cells were seen by IHC about 1 week after completing RT when combined with anti-CTLA4, consistent with the timeframe for T cell priming and activation. Significant CD8⁺ T cell infiltration was not seen in tumors harvested at later timepoints, suggesting a transitory response. However, in two mice with gross tumor response, residual tumor tissue was found to have higher CD8⁺ cell density than tumors with partial response. Longer treatment duration could potentially sustain T cell responses, as has previously been demonstrated with low-dose RT and immunomodulating treatment¹⁴. This underscores the importance of understanding longitudinal CD8⁺ T cell kinetics, facilitated by non-invasive CD8 immunoPET imaging.

This study has demonstrated the feasibility of using CD8 immunoPET in a 4T1 orthotopic mouse model. Of note, there was high variability of T cell infiltration after treatment demonstrated both by IHC and immunoPET imaging, in alignment with previous preclinical studies and emerging clinical data^7,8. Due to logistical considerations, PET imaging was performed within 24–48 h of completing treatment; a limitation of these results is that IHC experiments had identified the peak of CD8 + T cell infiltration 1 day after completing treatment. Although PET imaging was indeed able to detect an increase in tracer uptake in treated vs untreated tumors over this wider experimental window, it cannot be determined what portion of the high signal variability could be explained by this difference in timing.

A pilot experiment could not demonstrate correlation of tumor responses with early CD8 immunoPET signal, pointing to additional factors that influence tumor control, such as the functional status of T cells or T cell invigoration to tumor burden ratio¹⁵. Tumors were size-matched at the imaging timepoint to reduce confounding of PET signal measurements, based on the hypothesis that an early CD8 response precedes changes in tumor size and thus could be used as a predictive biomarker. However, a single imaging timepoint may be inadequate to capture the highly dynamic and transitory CD8⁺ T cell response which may also persist in actively responding tumors; serial imaging on this timescale would require a radioisotope with a shorter half-life, e.g. ¹⁸F. CD8 immunoPET imaging with ¹⁸F could also offer improved spatial resolution and is being investigated in the clinic (clinicaltrials.gov #NCT05629689), but logistical constraints for optimization of imaging timepoints must be taken into consideration. This study did not look at repeat imaging after the initial ⁸⁹Zr radiotracer injection; while this approach would not capture CD8⁺ T cell proliferation, in future studies it could be used to investigate dynamic trafficking of the initially labeled T cells, for example to specifically look at intratumor influx of CD8⁺ T cells and drainage to regional lymph nodes.

Other limitations of CD8 immunoPET include potential confounding factors such as vascularity, tumor necrosis, and tumor size. Despite this, increased signal in treated vs size-matched untreated tumors was consistent with higher CD8 staining by IHC on 2-dimensional central axis tumor slices. A stronger correlation may have been possible to detect with 3-dimensional estimation of CD8⁺ T cell density using many parallel tumor cross sections, although this would be a more time and cost-intensive approach. Inter-experiment variability was also observed; the tumor microenvironment may be influenced by factors such as the gut microbiome or pathogen exposure that can differ between cohorts of co-housed mice. Still, heterogeneity in this model may be exploited in future studies to probe whether other immunomodulatory regimens shift the overall percentage of tumors with robust CD8⁺ T cell infiltration. For example, radiosensitization with ATR (Ataxia-Telangiectasia Rad3-related) inhibition, acting through suppression of the DNA damage response, has been demonstrated in other tumor models to synergize with RT and immunotherapy and modulate the immune response at least in part through increased CD8⁺ T cell infiltration of tumors^16–18. CD8 immunoPET imaging could enable in vivo preclinical screening of optimal dosing strategies for this and other novel treatment combinations that may mediate tumor responses though CD8⁺ T cell proliferation and/or trafficking.

In summary, CD8 immunoPET can visualize an early, transitory, and heterogenous CD8⁺ T cell response after RT and anti-CTLA4 in a 4T1 orthotopic mouse tumor model. Ultimately, it will be necessary to better understand the heterogeneity of CD8⁺ T cell responses and how baseline and/or on-treatment CD8⁺ T cell infiltrations influences tumor regression. CD8 immunoPET provides a unique tool for non-invasive, whole-body monitoring of this complex biology. Evaluation of how agents alter the immune contexture within the tumor microenvironment with this tool may support clinical translation of immunomodulatory agents.

Methods

Mice and cell culture

Female Balb/c mice (Jackson Laboratories) were housed and maintained at City of Hope. Mice were acclimatized for a minimum of 3 days prior to experimental procedures. Animal study protocols were approved by the Institutional Animal Care and Use Committee (IACUC) at City of Hope. Humane endpoints for euthanasia prior to pre-determined experimental endpoints were per institutional guidelines and included maximum tumor diameter >15 mm or tumor necrosis >80% of tumor surface area. Since this was a proof-of-concept study, sample size was determined by the number of mice that could be feasibly treated with RT or imaged in a single day. Mice were randomly allocated to unblinded treatment groups, stratified for size-matched tumors. For localized irradiation studies, sulfatrim feed or water was provided starting a few days prior to radiation and for at least 2 weeks after; hydrogel was provided post-irradiation. Animals were monitored at least 2–3 times per week (more frequently near endpoints) for radiation sickness, body condition, and tumor size.

4T1 tumor cells were obtained from ATCC (CRL-2539) and cultured per manufacturer’s instructions in Roswell Park Memorial Institute (RPMI, Corning Cat # 10-041-CV) supplemented with 10% fetal bovine serum (Corning Cat # 35-011-CV). Cells were passaged no more than 6 times prior to orthotopic injections and were regularly tested for Mycoplasma contamination.

Orthotopic tumor implantation and treatment

For orthotopic injections, tumor cells were suspended in 50 μl of Matrigel matrix (Corning Cat # 356237) and Hanks’ Balanced Salt Solution (HBSS, Corning Cat # 21-022-CV). Anti-CTLA4 (InVivoMab 9H10, Bio X Cell) was diluted in phosphate-buffered saline (PBS, Corning Cat # 21-031-CV). To generate size-matched tumors for untreated mice at the imaging timepoint, orthotopic injections were performed 6 days prior to radiotracer injection for the untreated cohort.

50 ×103 4T1 tumor cells were injected into the right fourth mammary fat pad of anesthetized 6–8 week-old female BALB/c mice; treatment began after 6 days when tumors were ~5 mm in diameter. Tumor growth was monitored by caliper measurement ((length of longest dimension) * (perpendicular width)²)/2.

For RT, tumors were treated conformally with 8 Gy fractions for 3 consecutive days to a total of 24 Gy. Radiation was delivered with a 1 cm collimator to anesthetized mice on the X-RAD SmART system using CT-based treatment planning. One day after completing RT, mice started treatment with anti-CTLA4 antibody; 200 µg was injected intraperitoneally every 3 days for a total of 3 doses¹.

Anesthesia and euthanasia

Inhaled isoflurane up to 5% was used for anesthesia during tumor implantation, imaging, and radiation delivery. Depth was monitored by respiratory rate/color and loss of toe-pinch reflex. Euthanasia methods included CO₂ from compressed gas cylinders and/or isoflurane overdose, followed by a secondary physical method (bilateral thoracotomy or cervical dislocation) to confirm death.

Immunohistochemistry (IHC)

Tumors were harvested at multiple timepoints for analysis by hematoxylin and eosin (H&E) staining and IHC: 1 day after anti-CTLA4 x 3, 1 day after 8 Gy x 1, 1 day after 8 Gy x 3, 1 day after 8 Gy x3 + anti-CTLA4 x 1, 1 day after 8 Gy x 3 + anti-CTLA4 x 3, 5 days after 8 Gy x 3 + anti-CTLA4 x 3, 14 days after 8 Gy x 3 + anti-CTLA4 x 3. Except where indicated, control mice were treated according to the same schedule as mice treated with RT + anti-CTLA4, such that RT alone controls were harvested 8 days after completing RT, and anti-CTLA4 alone controls were harvested 1 day after the last dose of anti-CTLA4. For Visiopharm analysis, non-tumor adipose tissue was excluded by an independent reviewer, when present, to accurately characterize intratumoral lymphocytes.

Tumors were fixed in formalin and paraffin-embedded for IHC. CD8⁺ cell density on central tumor slices (cell count/mm²) was quantified with Visiopharm software using the same protocol across all samples. Two-dimensional cell density was calculated as (CD8⁺ cells for an entire tumor slice)/(total tumor area). Antibodies: anti-CTLA4 (InVivoMab 9H10, Bio X Cell), CD8α (D4WD2Z, Cell Signaling), CD4 (D7D2Z, Cell Signaling) and CD3 (2GV6, Ventana).

CD8 immunoPET imaging

Purified CD8α-specific cys-diabody was lightly reduced, conjugated with maleimide-desferrioxamine and radiolabeled with ⁸⁹Zr as previously described¹⁰. Approximately 10 µg (3.7 MBq) of radiolabeled CD8 cys-diabody (⁸⁹Zr-Df-CD8cDb) was administered by tail vein injection to each mouse 15 days after orthotopic injection of tumors treated with RT. Static PET/CT images were acquired 24–48 h after radiotracer injection. Images were acquired on a GNEXT small animal PET/CT scanner (SofieBiosciences). Imaging analysis was performed with Velocity software (V4.1, Varian Medical Systems, Inc. Palo Alto, CA).

Average %injected dose/cc (%ID/cc) was determined for the entire tumor contoured on axial CT images with treatment group blinded. For biodistribution studies, mice were euthanized after imaging, and tissues were dissected, weighed, and counted in a gamma counter.

Statistics

Comparisons for cell density by IHC were made by Kruskal-Wallis ANOVA with p-values adjusted for multiple comparisons, or unpaired two-tailed t-test when comparing two groups. Unpaired two-tailed t-test was used for comparisons of %ID/cc or %ID/g for imaging and biodistribution studies. Pearson correlation coefficient was computed for assessing linear correlation between %ID/cc for imaging and tumor volume or CD8⁺ cell density by IHC. GraphPad Prism v7.0 was used for statistical analysis. N-way analysis of variance (ANOVA) at each timepoint was used to evaluate statistical significance of tumor regression including interaction with CTLA4 (MATLAB, The MathWorks Inc., Natick, Massachusetts). Three treatment groups were considered: (1) CTLA4 alone, (2) RT alone, (3) CTLA4 + RT.

Author contributions

E.A.G., K.A.Z., T.W., H.M.M., T.M.W., and A.M.W. contributed to experimental design. E.A.G., K.A.Z., T.W., B.Y.C., D.L., F.B.S., J.C. and S.L. performed experiments. All authors contributed to data analysis and interpretation. A.M.W. supervised the project. E.A.G. drafted the manuscript; all authors provided revisions and have read and approved the manuscript.

Data availability

All data generated and analyzed in this study are included within the paper and its Supplementary Information. Raw data are available from the corresponding author on request.

Competing interests

A.M.W. is a shareholder and consultant to ImaginAb, Inc. but declares no non-financial competing interests. All other authors declare no financial or non-financial competing interests. E.A.G. is currently employed at Genentech, Inc., D.L. and B.Y.C. are currently employed at AbbVie, Inc., and S.L. is currently employed at AskGene Pharma, Inc.; all work was completed at City of Hope and no current employers have been involved in this publication.

Supplementary information

The online version contains supplementary material available at 10.1038/s44303-025-00128-5.