Background:　In　recent　years,　significant　progress　has　been　made　in　the　treatment　of　nasopharyngeal　carcinoma　(NPC).　The　application　of　immunotherapy,　especially　the　use　of　Programmed　Cell　Death　Protein　1　inhibitors,　has　demonstrated　excellent　therapeutic　effects.　However,　inconsistent　immunotherapy　outcomes　due　to　individual　differences　in　patients　remain　a　major　challenge,　maki　ng　the　search　for　specific　biomarkers　to　screen　for　the　population　who　may　benefit　from　immunotherapy　a　priority.　Surface-enhanced　Raman　spectroscopy　(SERS)　has　shown　great　potential　as　a　highly　sensitive　and　specific　optical　analytical　tool　for　identifying　and　monitoring　tumor-related　markers　in　NPC.　Results:　In　this　study,　plasma　samples　from　Nasopharyngeal　cancer　patients　before　and　after　immunotherapy,　along　with　those　from　healthy　volunteers,　were　tested　using　label-free　SERS　combined　with　plasmapheresis.　Especially,　the　components　with　varying　molecular　weight　sizes　were　analyzed　via　the　separation　process,　thereby　preventing　the　potential　loss　of　diagnostic　information　that　could　result　from　competitive　adsorption.　Subsequently,　a　robust　machine　learning　algorithm　based　on　principal　component　analysis　and　linear　discriminant　analysis　(PCA-LDA)　was　used　to　extract　features　from　plasma　SERS　data　and　establish　an　effective　discriminant　model.　The　results　showed　that　in　the　upper　plasma　layer,　the　most　optimal　discrimination　was　found　between　normal　patients　and　those　post-treatment,　with　sensitivity　and　specificity　of　88.5　%　and　92.3　%,　respectively.　In　the　lower　plasma　layer,　the　most　optimal　discrimination　was　found　between　the　pre-treatment　and　post-treatment　patients,　with　sensitivity　and　specificity　of　84.6　%　and　80.8　%,　respectively.　Significance　and　novelty:　Plasmapheresis　can　reveal　latent　diagnostic　information　in　the　plasma　and　holds　promise　for　treating　NPC,　the　screening　of　the　population　who　may　benefit　from　immunotherapy,　and　the　postoperative　evaluation　of　immunotherapy.　Further　exploration　of　the　feasibility　of　using　SERS　to　detect　tumor　markers　in　the　population　who　may　benefit　from　immunotherapy　a　priority.　Surface-enhanced　Raman　spectroscopy　(SERS)　has　shown　great　potential　as　a　highly　sensitive　and　specific　optical　analytical　tool　for　identifying　and　monitoring　tumor-related　markers　in　NPC.　Results:　In　this　study,　plasma　samples　from　Nasopharyngeal　cancer　patients　before　and　after　immunotherapy,　along　with　those　from　healthy　volunteers,　were　tested　using　label-free　SERS　combined　with　plasmapheresis.　Especially,　the　components　with　varying　molecular　weight　sizes　were　analyzed　via　the　separation　process,　thereby　preventing　the　potential　loss　of　diagnostic　information　that　could　result　from　competitive　adsorption.　Subsequently,　a　robust　machine　learning　algorithm　based　on　principal　component　analysis　and　linear　discriminant　analysis　(PCA-LDA)　was　used　to　extract　features　from　plasma　SERS　data　and　establish　an　effective　discriminant　model.　The　results　showed　that　in　the　upper　plasma　layer,　the　most　optimal　discrimination　was　found　between　normal　patients　and　those　post-treatment,　with　sensitivity　and　specificity　of　88.5　%　and　92.3　%,　respectively.　In　the　lower　plasma　layer,　the　most　optimal　discrimination　was　found　between　the　pre-treatment　and　post-treatment　patients,　with　sensitivity　and　specificity　of　84.6　%　and　80.8　%,　respectively.　Significance　and　novelty:　Plasmapheresis　can　reveal　latent　diagnostic　information　in　the　plasma　and　holds　promise　for　treating　NPC,　the　screening　of　the　population　who　may　benefit　from　immunotherapy,　and　the　postoperative　evaluation　of　immunotherapy.　Further　exploration　of　the　feasibility　of　using　SERS　to　detect　tumor　markers　in　the　population　who　may　benefit　from　immunotherapy　for　NPC　will　help　improve　the　therapeutic　efficacy,　optimize　clinical　practice,　and　promote　the　development　of　individualized　treatment　strategies.