Muscle-Specific MRI Grading of Soft Tissue Involvement Provides Additional Prognostic Value Beyond Skull Base Criteria in Nasopharyngeal Carcinoma: A Retrospective Study

Background In the 9th edition of the AJCC/UICC staging system for nasopharyngeal carcinoma (NPC), soft tissue extension to, but not beyond, the lateral pterygoid (LP) muscle is classified as stage T2 disease. Invasion beyond the LP into the masticator space or infratemporal fossa is designated as T4, while any skull base bone erosion is assigned to T3. However, this framework may oversimplify the prognostic spectrum of soft tissue involvement (STI). This study was formulated to investigate whether a muscle-specific magnetic resonance imaging (MRI) grading of offers incremental prognostic value over current skull base-based staging criteria. Methods Patients with newly diagnosed NPC treated with definitive intensity-modulated radiotherapy (IMRT) between 2014 and 2019 were retrospectively analyzed. Pretreatment MRIs were used to categorize STI severity as mild (tensor or levator veli palatini), moderate (prevertebral muscles), or severe (medial/lateral pterygoid or infratemporal fossa). Skull base invasion was classified as either limited (LSBI) or extensive (ESBI). Survival endpoints included local failure-free survival (LFFS), distant metastasis-free survival (DMFS), progression-free survival (PFS), and overall survival (OS). Kaplan–Meier analysis and log-rank tests assessed survival, and Cox proportional hazards models identified independent prognostic factors. Results Of 391 patients (median follow-up 88 months; interquartile range, 68–105), 42.9% exhibited mild, 19.4% moderate, and 37.6% severe STI. Five-year survival rates were 84.1% (LFFS), 90.0% (DMFS), 81.3% (PFS), and 80.3% (OS). Survival declined in a stepwise fashion with increasing STI severity (log-rank P ≤ 0.0001 for all endpoints); the 5-year OS was 92.9% for mild, 73.7% for moderate, and 68.0% for severe invasion. On multivariable analysis, moderate and severe STI were associated with a 3- to 4-fold increased risk of adverse outcomes, including disease progression and mortality (severe vs. mild OS: HR 4.55, 95% CI 2.47–8.37, P < 0.001). Induction chemotherapy was independently protective, reducing the hazard of death by approximately 45% (OS: HR 0.56, 95% CI 0.32–0.97, P = 0.04). In contrast, skull base invasion status was not prognostically significant in either univariate or multivariate models. When directly compared, OS for moderate STI and LSBI was similar (81% vs. 78%, P = 0.98), while severe STI showed a non-significant trend toward poorer OS compared with LSBI (68.0% vs. 76.1%, P = 0.24). Conclusions Muscle-specific MRI grading of STI serves as a more robust predictor of treatment outcomes than conventional skull base bone invasion in NPC patients receiving IMRT. This grading system may facilitate refined risk stratification and inform decisions on treatment escalation or de-intensification.