Background:
Optically labeled therapeutic antibodies (OLTAs) enable intraoperative visualization of solid tumors while providing a unique opportunity to directly assess antibody biodistribution and target engagement in patients. The FDA’s Project Optimus initiative has highlighted the need for biologically optimized dosing strategies; however, optimization of therapeutic antibody dosing remains challenging because antibody behavior within human tumors cannot be directly quantified.

Hypothesis/Objective:
We used panitumumab-IRDye800CW (Pan800), an anti-EGFR OLTA, to directly measure tumor receptor occupancy (RO) in patients and identify the optimal biological dose of anti-EGFR therapy.

Methods:
A prospective surgical window-of-opportunity study was conducted to determine the optimal biological dose of panitumumab (NCT06819228). All patients received a fixed 50-mg dose of Pan800 before surgery. To achieve total antibody doses corresponding to 100% (6 mg/kg; n=3), 80% (n=4), 60% (n=4), or 40% (n=3) of the standard clinical dose, additional unlabeled panitumumab was co-administered. A separate cohort received 50 mg Pan800 alone (approximately 10% of the standard clinical dose; n=10; NCT05945875 and NCT04511078). All patients underwent surgery within 3 days after infusion. Plasma samples and tumor specimens were collected on the day of surgery. Plasma panitumumab concentrations and tumor antibody concentrations were quantified. FFPE sections were used to evaluate antibody distribution within tumor and stromal compartments, whereas matched frozen sections were used to quantify RO. In cohorts receiving unlabeled panitumumab, total intratumoral panitumumab levels were estimated from measured Pan800 signals and the known labeling dilution ratio.

Results:
Total panitumumab dose correlated more strongly with plasma concentration than body-weight–normalized dosing (R²=0.78 vs. 0.53; Fig A, B). Increasing doses increased antibody accumulation in both tumor and stromal compartments and elevated the stromal-to-tumor signal ratio (R²=0.50, Fig C), consistent with stromal overflow of unbound antibody. This stromal signal was largely removed by frozen-section washing (Fig D). RO was quantified by comparing fluorescence signals before and after ex vivo receptor saturation (Fig E, F). Near-complete EGFR saturation was achieved despite a 60% reduction from the standard clinical dose (Fig G, H). RO exceeded 80% in all evaluable tumors (14/14) and 90% in 79% (11/14). Tumors failing to achieve 90% RO exhibited exceptionally high EGFR expression. Plasma panitumumab concentration demonstrated a strong Emax relationship with RO (R²=0.78, AICc=92.2), whereas tumor concentration showed weaker predictive performance (R²=0.62, AICc=102.8), likely because measured tumor concentrations were influenced by stromal overflow of unbound antibody and interpatient variability in EGFR expression. Incorporation of EGFR expression significantly improved model performance, and an Emax model incorporating both variables (plasma concentration/EGFR^0.6) accurately predicted RO (R²=0.84, AICc=86.6; Fig I). The model estimated that achieving 90% RO in a 75-kg patient would require 1.7 mg/kg for EGFR 1+ tumors (54% of cases), 2.5 mg/kg for EGFR 2+ tumors (33%), and 6.0 mg/kg for EGFR 3+ tumors (13%) (Fig J), suggesting that the current uniform 6 mg/kg dose substantially exceeds the biologically required dose for most patients at least at early post-infusion time points.

Conclusion:
OLTA revealed dose-dependent stromal overflow of unbound antibody and near-complete tumor RO at substantially reduced doses. Tumor EGFR expression and plasma panitumumab concentration were major determinants of RO. These findings support RO–guided optimization of therapeutic antibody dosing and establish fluorescence imaging as a tool for direct assessment of target engagement in human solid tumors.

Image/Figure:
Click to View

Image/Figure Caption:
Figure. Direct measurement of tumor receptor occupancy identifies biologically optimized panitumumab dosing. (A–B) Plasma panitumumab concentration correlated more strongly with total administered dose than with mg/kg dosing. (C) Higher plasma concentrations increased stromal-to-tumor signal ratios, consistent with antibody overflow into the stroma. (D) Representative frozen sections before and after washing, demonstrating removal of unbound stromal antibody. (E) Schematic of receptor occupancy (RO) quantification by comparing in vivo–bound and ex vivo–saturated Pan800 signals. (F) Representative tumor images from control, low-dose, and high-dose cohorts. (G) RO exceeded 80% in all evaluable tumors and 90% in 79% of tumors. (H) Plasma panitumumab concentration exhibited an Emax relationship with RO. (I) Incorporation of tumor EGFR expression improved RO prediction (plasma concentration/EGFR^0.6; R² = 0.84). (J) Model-predicted panitumumab doses required for 90% RO in a 75-kg patient according to EGFR expression level. EGFR 1+, 2+, and 3+ tumors required 1.7, 2.5, and 6.0 mg/kg, respectively.

Author
Hidenori Tanaka
Vanderbilt University Medical Center