Dynamic biological and metabolic response by EBV DNA and ¹⁸ F-FDG-PET-CT to induction chemotherapy (IC) to enhance prognostication in endemic nasopharyngeal carcinoma (NPC).

Abstract

6112 Background: IC followed by concurrent chemoradiotherapy (CCRT) is the current standard of care for locoregionally-advanced NPC (LA-NPC). Nonetheless, evidence suggests that IC response is a strong predictor for risk of relapse in LA-NPC. Here, we characterized the dynamic metabolic and biological responses to IC, and investigated their associations with survival in patients with high-risk LA-NPC. Methods: Newly diagnosed patients with non-metastatic, biopsy-proven NPC were enrolled into two prospective ongoing studies (NCT04340024 and NCT06093061). For this analysis, patients who received 2-3 cycles of gemcitabine-cisplatin/carboplatin IC were included; all patients had plasma EBV DNA assessed following every IC cycle, and 18 F-FDG-PET-CT performed pre-IC and 1 week before the end of IC. For the latter, standardized uptake values (SUVs) were recorded for the primary tumor (SUVp) and individual nodal lesion(s) (SUVn). Biological (bCR) and metabolic complete response (mCR) were defined as EBV DNA =0 copy/mL and SUV ≤2.5, respectively. Prediction accuracy of 2y disease-free survival (DFS) was assessed by area under the receiver operating characteristic curve (AUC). Results: 117 patients diagnosed between Jan 2019 to Apr 2025 were included in this analysis; of these, 112 and 106 had SUVp and SUVn at pre- and post-IC. For SUVn, we performed lesion-level analysis for 259 LNs. Median follow-up was 20.5 (interquartile range [IQR]: 11.6-41.3) mo and TNM-8 stage distribution for stage III and IVA were 43% (50/117) and 57% (67/117), respectively. Median SUVp and SUVn pre-IC were 15.0 (IQR: 10.4-18.2) and 8.5 (IQR: 4.5-12.3), respectively, while median EBV DNA level was 4800 (IQR: 914-20598) copies/mL. Post-IC, we recorded mCR at the primary tumor and LNs for 33/112 (29%) and 49/106 (46%), respectively; while 14/117 (12%), 30/117 (26%) and 40/117 (34%) patients manifested bCR post-IC1, IC2, and IC3, respectively. Metabolic response was not correlated with biological response; of 33 patients with primary tumor mCR, 20/33 (61%) had bCR; while for LNs, only 25/49 (51%) patients with mCR manifested bCR post-IC. Additionally, we observed significant intrapatient heterogeneity of mCR between LN lesions post-IC; in 57/106 patients with non-mCR for ≥1 of the LNs, SUVn ranged between 0-12.9. Finally, incorporating both post-IC bCR and mCR to TNM-8 stage and pre-IC EBV DNA enhanced AUC of 2y DFS prediction; from 0.58 (95%CI:0.44-0.73) [TNM-8+pre-IC EBV DNA] to 0.77 (95%CI:0.60-0.94). Conclusions: Metabolic and biological responses to IC provide unique information on response phenotypes of patients with LA-NPC, and improved the prediction accuracy of 2y DFS compared with TNM-8 stage and pre-IC EBV DNA. Combinatorial 18 F-FDG-PET-CT and EBV DNA post-IC may enhance the selection of these patients for treatment intensification.