Quantification of immunotherapy response by granzyme B PET imaging is dependent on specific activity.

Objectives: While cancer immunotherapy has delivered stunning results, it remains impossible to predict which patients will respond to therapy using biopsy or standard imaging techniques. Early assessment of patient response to immunotherapy could reduce unnecessary side effects which can be associated with these therapies. As granzyme B, a downstream effector of tumoral cytotoxic T cells, has been implicated in immunotherapy response, we developed a peptide-based PET imaging probe directed against granzyme B. We investigated its ability to quantitatively assess granzyme B levels in vivo, and analyzed its effectiveness in predicting response to immunotherapy. We then sought to understand the role of injected peptide specific activity in the probe’s ability to assess granzyme level, as any specific activity dependence should be taken into account if the probe were to be optimized for clinical translation.

Methods: Murine CT26 cancer cells were injected subcutaneously into Balb/C mice. All mice were treated by injection of saline (vehicle), anti-PD1 (monotherapy), or anti-PD1 and anti-CTLA-4 (combination therapy) days 3, 6, and 9 following innoculation. NOTA-GZP was labeled with ⁶⁸Ga eluted from a ⁶⁸Ge/^G8Ga generator and mice were injected intravenously with 350MBq/mg, 185 MBq/mg, or 85 MBq/mg of ⁶⁸Ga-NOTA-GZP on either day 12 or day 14 following tumor innoculation. PET images were taken at 1 hour following injection and specific uptake was calculated by dividing tumoral uptake by left ventricle uptake to derive a tumor to blood ratio (TBR). For correlation studies, mice were sacrificed immediately following imaging. For survival studies, mice were imaged at Day 12 and measured every 2 days from Day 10 until the end of the study, or when volumes reached greater than 1500mm².

Results: NOTA-GZP was radiolabeled with high yield (67+-11%) and purity (>95%). Peptide uptake was shown to correlate with Western Blot with a highly significant correlation (P<0.0001) and an R²=0.71, indicating that GZP PET imaging reflects tumor granzyme B expression. For survival studies, treated mice were classified based on TBR into high uptake (range=1.25-2.86) and low uptake (range=.57-1.08) groups. The high-uptake group had significantly different TBRs compared to the low-uptake (P<.05) and control (P<.001) groups (Figure A, B). These groupings were 100% predictive of response by Day 18 (Figure C). To investigate specific activity-related limitations to the predictive ability of GZP PET imaging, data from mice given combination therapy and imaged on Day 12 from correlation and survival studies were pooled, and the effect of intravenous specific activity on TBR was analyzed (Figure D). While TBRs from 18.5MBeq and 35MBeq conditions were not significantly different, TBRs from the 8.5MBeq condition were significantly lower than those from high-specific activity conditions (p<0.0001) and had the lowest standard deviation (Figure E). Therefore, individual mice from these studies could not be differentiated into low-uptake and high-uptake groups.

Conclusion: PET granzyme imaging using NOTA was shown to be an accurate technique for granzyme quantification in mice. In addition, granzyme B levels, as measured by GZP PET, were highly predictive of response to PD-1 blockade and combination therapy. However, the accuracy of GZP PET is highly dependent on injected specific activity, such that at low injected specific activity this predictive ability is lost. This specific activity dependence should be taken into account in future studies which utilize GZP, and especially as granzyme B PET imaging is optimized for clinical translation.

• Download figure
• Open in new tab
• Download powerpoint