Metastatic Status Modifies the Survival Association of Neoadjuvant Radiotherapy in Malignant Peripheral Nerve Sheath Tumors: A Nationwide Analysis
Shilp Shah, Shivam Singh, Vikas N. Vattipally, David Barreto, Taha Khalilullah, Melanie Alfonzo Horowitz, Liam P. Hughes, Joseph M. Dardick, Jawad M. Khalifeh, Yuanxuan Xia, Ali Bydon, Nicholas Theodore, Daniel LubelskiABSTRACT
Purpose
Malignant peripheral nerve sheath tumors (MPNSTs) are rare, aggressive sarcomas with poor overall survival (OS). Radiotherapy timing is extrapolated from other soft‐tissue sarcomas, and no prior MPNST‐specific analysis has formally tested whether neoadjuvant radiotherapy (NRT) effects are modified by metastatic status. We evaluated NRT's association with R0 resection and OS by metastatic status.
Methods
Retrospective cohort of MPNST patients in the National Cancer Database (2005−2023). Multivariable logistic regression evaluated NRT and R0 resection. Cox proportional hazards models tested an NRT‐by‐metastatic status interaction, with hospital‐level frailty terms addressing inter‐facility correlation. Propensity score matching (1:1) reduced confounding. Sensitivity analyses addressed guarantee‐time bias and confounding. Causal mediation evaluated R0 as a mediator of the NRT−survival relationship.
Results
Of 2241 patients, 24.2% received NRT. NRT increased the odds of R0 resection (OR 2.17, 95% CI 1.60−2.98; p < 0.001) but was not associated with improved survival overall. The NRT‐survival association was modified by metastatic status (interaction p < 0.001): metastatic patients receiving NRT had better survival (matched HR 0.41, 95% CI 0.23−0.72; p = 0.004), whereas no survival benefit was seen in localized disease (worse in some analyses). This signal was consistent across guarantee‐time sensitivity analyses but was not mediated by margins.
Conclusion
NRT improves the odds of R0 resection in MPNST, but improved margins do not yield a uniform survival benefit. The NRT‐survival association is modified by metastatic status; the metastatic signal, though robust to guarantee‐time bias, was not mediated by margins and rests on a small subgroup. These findings are hypothesis‐generating and warrant a prospective study.