SHBC1258

D.X.C.YIN¹, S.M.CHIOW¹, A.KARANDIKAR¹, J.P.N.GOH¹, J.Y.J.GAN¹, W.Z.E.FU¹, H.LI¹, M.Y.LIM¹

Tan Tock Seng Hospital¹

A landmark study by William Wei et. al. showed that the preoperative impression based on clinical examination and CT scan was a gross underestimate of true nodal involvement. Our study sought to determine if modern MRI imaging may be better able to predict nodal involvement, to allow selected levels of neck dissection to be preserved, thus reducing morbidity of treatment.

All NPC patients in our tertiary center from 2002 to 2017 with isolated regional failure after completion of primary RT/CRT and who received a (1) preoperative MRI scan (2) RND or modified RND (MRND) were included. 35 cases met these criteria.

We calculated the sensitivity and specificity of preoperative MRI neck. Overall sensitivity was 76% and overall specificity was 90%. In William Wei’s landmark study, the total number of lymph nodes detected with or without CT was 59, with a range of 1 to 5 (mean, 1.5) per patient. The number of lymph nodes containing tumor was 294, with a range of 1 to 62 (mean, 7.4) per patient. In our study, the total number of lymph nodes detected by MRI was 105, with a range of 0 to 10 (mean, 3.0) per patient. The number of nodes that contained tumor was 91, with a range of 0 to 5 (mean, 2.6) per patient.

Based on an overall sensitivity of 76%, we will be missing 10 (5.7%) diseased lymph node levels. MRI alone does not allow safe selective preservation of neck levels in surgical salvage of neck recurrences in NPC.