3069 Background: CD8 T cells mediate antitumor effects of immune checkpoint blockade (ICB); their histologically-determined abundance prior to or during therapy (Rx) appears to correlate with ICB response in melanoma (MEL) and other solid tumors. We performed this Phase IIb study to assess CD8 targeted PET imaging with 89 Zr-crefmirlimab berdoxam prior to and during ICB Rx to study its potential as a marker for Rx selection, enhancement or even replacement of existing tumor measurement systems, and correlation of clinical outcomes. Methods: 70 patients (pts) with MEL or Merkel cell cancer (MCC) (27), renal cell cancer (RCC) (35) or lung cancer (NSCLC) (8) who were eligible for first or second-line single- or double-agent ICB or ICB plus an oral kinase inhibitor (TKI) were enrolled. 89 Zr-crefmirlimab berdoxam (1 mCi,1.5 mg protein), was administered intravenously ≤2 weeks prior to cycle 1 of ICB Rx, followed by a PET/CT scan 24 (+/-3) hours later (baseline, BL). The 2nd tracer injection and associated PET/CT scan occurred 4-6 weeks after ICB Rx initiation, prior to cycle 3 (on treatment, OT). Standard of Care imaging and RECIST 1.1 were used to assess response to Rx. Results: All 70 pts were evaluable for safety analysis; none experienced an SAE related to imaging agent. 65 pts (21 MEL, 2 MCC, 34 RCC and 8 NSCLC) were included in the primary endpoint analysis of correlating CD8 PET scans with best confirmed overall response (BOR) by RECIST 1.1. MEL and MCC pts received single (6/23) or double ICB (17/23), RCC pts received single ICB with/without TKI (21/34) or dual ICB (13/34), and all NSLCL pts received single ICB. In the RCC cohort, multiple CD8 PET metrics showed statistically significant association with either individual BOR or Binary Response at BL, OT or as delta, including: Tumor Standard Uptake Value (SUV), Lymph Node (LN) SUV, relative percentage of CD8 negative or positive lesions and normal organ SUV. Normalization to reference organs improved the correlation with BOR for the combined cohorts and for the MEL/MCC cohort. Depending on clinical objectives, different classification rules to predict BOR were applied, prioritizing specificity for non-responders and sensitivity for responders (Table). These results were cohort-dependent, as each cohort exhibited a distinct distribution of SUVs. Conclusions: PET imaging can quantitate CD8 T cells with sufficient specificity and sensitivity to support further study in selection of pts for Rx regimens and assessing early response to therapy. Clinical trial information: NCT05013099 . Prediction of BOR by CD8 PET. Cohort Response Parameter Timepoint AUC 95% CI Sensitivity Specificity MEL + MCC CR+PR vs SD+PD Spleen SUVmean BL 0.67 0.44,0.90 87.5 60.0 MEL + MCC CR+PR vs SD+PD Tumor SUVmax normal. to thyroid Delta 0.70 0.46, 0.95 85.7 64.3 RCC CR+PR vs SD+PD Hottest LN SUVmax OT 0.84 0.65, 1.00 71.4 90.9 RCC CR vs PR+SD+PD % CD8 negative lesions Delta 0.96 0.87, 1.00 100.0 92.3
Margolin et al. (Wed,) studied this question.