[18F]FDG PET/CT for predicting neoadjuvant PD-L1 blockade monotherapy treatment response in patients with locally advanced esophageal squamous cell carcinoma: a preliminary study

Abstract

Purpose

To investigate the predictive value of 2-[18F]-fluoro-2-deoxy-D-glucose ([¹⁸F]FDG) PET/CT for evaluating primary tumor (PT) and lymph node (LN) responses after neoadjuvant programmed death-ligand 1 (PD-L1) blockade monotherapy in patients with locally advanced esophageal squamous cell carcinoma (LA-ESCC).

Methods

In the single-arm phase 1b NATION-1907 trial (NCT04215471), 23 patients with LA-ESCC received two cycles of neoadjuvant PD-L1 blockade Adebrelimab followed by surgery. Among these, 18 patients underwent [¹⁸F]FDG PET/CT scans both before immunotherapy and prior to surgery. Standardized uptake value corrected for lean body mass (SUL)-derived parameters, including SUL_max and SUL_peak, were documented for PTs and LNs. Lesions > 1cm³ were segmented using thresholds of 41% and 50% of SUL_max, respectively, following European Association of Nuclear Medicine (EANM) guidelines, with metabolic tumor volume (MTV) and total lesion glycolysis (TLG) calculated. Percentage changes of all metabolic parameters were also recorded. Residual viable tumor ≤ 33% were classified as well-responders, whereas residual viable tumor > 33% were classified as poor-responders based on histological evaluation.

Results

In the PT analysis, 10 patients were classified as PT well-responders and 8 as PT poor-responders. All post-treatment metabolic parameters, except MTV, were significantly lower in well-responders compared to poor-responders. The %ΔMTV, %ΔTLG were significantly higher in the poor-responder group (all P < 0.05). ROC curves indicated %ΔMTV₄₁ exhibited optimum performance in predicting well-responders, with an AUC of 0.875 (cut-off: -31.01). Furthermore, %ΔMTV₄₁ significantly predicted patients' recurrence-free survival (RFS) (P < 0.1). In the LN analysis, 7 LNs were classified as well-responders and 10 as poor-responders. Pre-treatment SUL_max, SUL_peak were significantly lower in poor-responders compared to well-responders. Post-treatment MTV₅₀ and all percentage changes in parameters were significantly higher in the poor-responder group (all P < 0.05). Receiver operating characteristic curve (ROC) analysis indicated %ΔTLG₅₀ had excellent predictive performance for well-responders, with an AUC of 1.000 (cut-off: -7.5). However, there was no significant correlation between the metabolic response evaluations for PTs and LNs.

Conclusion

The metabolic parameters of [¹⁸F]FDG PET/CT, particularly %ΔMTV and %ΔTLG, could effectively predict well-responders among both PTs and LNs to neoadjuvant PD-L1 blockade monotherapy in LA-ESCC, which may facilitate personalized immunotherapy and serve as a stratification tool in future larger-scale studies.