Optical PD-L1 imaging using ultrasound-guided quantitative fluorescence molecular endoscopy combined with durvalumab-680LT in locally advanced esophageal cancer patients

Aims Locally advanced esophageal cancer (EC) is treated with neoadjuvant chemo(radio)therapy (nCRT) and surgery, but only 16-43% achieve a complete response. The addition of immune checkpoint inhibition (ICI) targeting PD-1/PD-L1 is explored to increase response rates, yet patient selection is difficult. PD-L1 expression’s predictive value in biopsies is inconsistent due to heterogeneity, and nCRT's effect on ICI response is unclear. Novel imaging methods are crucial to understand PD-L1 heterogeneity in order to ultimately select patients for ICI. This ongoing study assesses the safety and feasibility of the innovative technique ultrasound guided quantitative fluorescence molecular endoscopy (US-qFME) using the fluorescently labelled ICI drug durvalumab-680LT before and after nCRT in EC patients to visualize, for the first time, PD-L1 drug distribution from the macroscopic to the cellular level.

Methods A durvalumab-680LT dose-optimization was performed in which twenty EC patients scheduled for nCRT were included in either 0 mg, 4.5 mg, 15 mg or 25 mg cohort. US-qFME procedures were performed both before and after nCRT, consisting of in vivo fluorescence signal visualization, quantification by mucosal and ultrasound-guided spectroscopy measurements of healthy esophageal tissue, tumor and/or lymph nodes, and collection of biopsies. Subsequently, ex vivo analyses were performed for quantification of fluorescence signals, correlation to histology and visualization of tissue durvalumab-680LT tissue distribution and target cells.

Results Fluorescence signals were generally higher in tumor tissue compared to healthy tissue across the 4.5 mg, 15 mg, and 25 mg dose groups, though there was considerable variability within the tumor tissue, with some patients showing signal intensities similar to healthy tissue. Specifically, the median (IQR) fluorescence signals were: 4.5 mg: 0.020 Qµ^f _a,x [0.012 – 0.020], 15 mg: 0.024 Qµ^f _a,x [0.012 – 0.035], and 25 mg: 0.052 Qµ^f _a,x [0.040 – 0.053]. Signals in the control cohort were negligibly low, with a median (IQR) of 0.001 Qµ^f _a,x [0.0005 – 0.002]. A preliminary comparison of pre- and post-nCRT signals showed lower and less heterogeneously distributed signals post-treatment. Additionally, fluorescence microscopy in mucosal biopsies from the 15 mg and 25 mg cohorts detected durvalumab-680LT binding, with fluorescence signals observed irrespective of PD-L1 expression. No adverse events were reported.

Conclusions This study shows a broad range of durvalumab-680LT signal in tumor tissue across patients, with preliminary fluorescence microscopy data indicating durvalumab binding independent of PD-L1 expression. In conclusion, US-qFME combined with durvalumab-680LT is safe and enables visualization and quantification of durvalumab-680LT at both macroscopic and microscopic levels, suggesting its potential for future ICI patient selection.

Publication History

Article published online:
27 March 2025