Real-World Diagnostic Phenotypes and Treatment Pathways in Trigeminal Pain: A Retrospective Tertiary-Center...

Background: Trigeminal neuralgia (TN) is clinically defined, but patients presenting to tertiary practice with trigeminal-region pain are often diagnostically heterogeneous and may follow prolonged medication, dental, imaging, and procedural pathways before a stable phenotype is established. We aimed to characterize diagnostic phenotypes, secondary causes, and treatment-escalation patterns in a large retrospective tertiary-center trigeminal pain cohort derived from routine free-text clinical documentation. Methods: We conducted a retrospective single-center cohort study based on a clinical dataset containing 18,007 note fragments linked to 672 unique patient records between 12 October 2010 and 21 April 2026. A rule-based natural-language-processing-assisted chart review framework was used to identify patients with trigeminal pain and to extract documentation-derived demographic features, pain distribution, secondary causes, dental pathway variables, imaging signals, medication exposure, procedures, and outcome language. Patients were grouped into primary/classical TN, secondary TN/trigeminal pain, and dental-first or mimic pathways using predefined operational criteria. Results: A total of 455 patients met criteria for the analytic trigeminal pain cohort; 311 (68.4%) carried explicit TN terminology. Mean age was 58.7 years, median age 60 years, and 267 of 428 patients with recoverable sex data (62.4%) were women. Trigeminal branch involvement could be extracted in 351 patients (77.1%), with V2 involvement documented in 256 (56.3%), V3 involvement in 218 (47.9%), and V1 involvement in 138 (30.3%). The final NLP-derived phenotypic distribution comprised 201 primary/classical TN cases (44.2%), 146 secondary TN/trigeminal pain cases (32.1%), and 108 dental-first or mimic presentations (23.7%). MRI was documented in 384 patients (84.4%), neurovascular conflict or vascular loop in 253 (55.6%), multiple-sclerosis-related disease in 69 (15.2%), and tumor-related trigeminal involvement in 84 (18.5%). Prior dental evaluation was identified in 169 patients (37.1%), and prior dental procedures in 114 (25.1%). Carbamazepine exposure was documented in 367 patients (80.7%), pregabalin in 221 (48.6%), gabapentin in 150 (33.0%), oxcarbazepine in 116 (25.5%), and phenytoin in 73 (16.0%). At least one invasive or image-guided procedure was documented in 390 patients (85.7%), including nerve blocks/injections in 355 (78.0%), radiofrequency procedures in 126 (27.7%), balloon compression in 90 (19.8%), microvascular decompression in 113 (24.8%), and stereotactic radiosurgery in 55 (12.1%). Dental-first patients were significantly more likely to have undergone prior dental procedures (65.7% vs. 3.5% in primary/classical TN and 24.7% in secondary TN; p < 0.001), whereas secondary TN/trigeminal pain was associated with higher use of radiofrequency procedures (36.3%; p = 0.017), higher use of stereotactic radiosurgery (19.9%; p = 0.002), higher recurrence documentation (70.5%; p = 0.001), and a higher rate of complete pain relief documented at last follow-up (46.6%; p = 0.004). Conclusions: In tertiary practice, trigeminal pain is substantially broader than a formal TN label. Secondary disease and dental-first pathways account for a large fraction of referrals, and management is characterized by heavy medication burden, frequent escalation, and recurrent retreatment. A structured phenotyping approach may help convert routine clinical documentation into a clinically meaningful framework for diagnostic triage and treatment selection, although imaging and outcome variables require cautious interpretation when derived from retrospective free text.