← Back to Previous Page

 

S209: MULTI-INSTITUTIONAL MACHINE-LEARNING PREDICTOR OF GROSS-TOTAL RESECTION IN SKULL-BASE CHONDROSARCOMA
Juan P Zuluaga Garcia, MD, MSc¹; Franco Rubino, MD²; Francisco Call Orellana, MD¹; Esteban Ramirez-Ferrer, MD¹; Geroge Zenonos, MD³; Paul Gardner, MD³; Hanna Algattas, MD³; Juan C Fernadez-Miranda, MD⁴; Vigo Vera⁴; Franco DeMonte, MD¹; Shaan M Raza, MD¹; ¹The University of Texas MD Anderson Cancer Center; ²Baptist Medical Center; ³University of Pittsburg Medical Center; ⁴Stanford School of Medicine

Objective: Develop and validate an anatomy-driven model that predicts gross-total resection (GTR) for skull-base chondrosarcomas (SCBs).

Methods: We analyzed 179 consecutive SBCs resected at three academic centers. Thirteen preoperative variables were abstracted: tumor location, internal carotid artery (ICA) encasement, compartmental extensions, cranial-nerve involvement, prior radiotherapy, and approach. Data were split 75/25 into training (n=135) and validation (n=44). Five algorithms were tuned for validation. For interpretability, multivariable logistic GLM and nomogram were refitted. Performance was summarized with AUC, accuracy, sensitivity/specificity, and Brier score.

Results: Anatomic burden varied by zone, cavernous-sinus invasion clustered in peri-lacerum (66.7%) and petroclival (57.0%); jugular-foramen extension mainly petroclival (37.2%); sinonasal/orbital spread characterized midline lesions. Approach selection mirrored: EEA predominated in midline (80%) and common in petroclival (62%), whereas lateral tumors were mostly open (84.6%). GTR was 56% (101/179) overall. In petroclival disease, corridor choice was decisive (p<0.001): ETPA 70.3% GTR vs open 25.7% and midline EEA 36.4%.

Operative Approach, EOR, and Residual Disease

	Petroclival (N=121)	Peri-lacerum (N=30)	Lateral (N=13)	Midline (N=15)	p-value
EOR					0.116
GTR	61 (50%)	20 (67%)	10 (77%)	10 (67%)
STR	60 (49%)	10 (35%)	3.0 (23%)	5 (33%)
GTR% x approach					<0.001
Open	9/35 (26%)	12/19 (63%)	9/11 (82%)	2/3 (67%)
EE-Midline	4/11 (36%)	4/4 (100%)		5/8 (63%)
ETPA	45/64 (70%)	2/4 (50%)	0/1 (0%)	3/4 (75%)
Combined/stage	3/11 (27%)	2/3 (67%)	1/1 (100%)
Residual disease					<0.001
Petrous Apex	21 (17%)	3 (10%)	2 (15%)
Meckel’s-cave	11 (8%)	3 (11%)
Cavernous-sinus	9 (7%)	6 (23%)		1 (7%)
Cavernous-ICA	3 (2%)	1 (3%)
Petrous-ICA	4 (3%)	2 (8%)
Jugular-Foramen	7 (6%)	1 (3%)

Multivariable analysis showed significant lower odds of GTR with graded ICA encasement (90–180° OR=3.12; 181–270° 7.41; 271–359° 9.76; 360° 8.52), prior radiotherapy (OR=4.04), petroclival location (OR=4.72) and  infratemporal extension (OR=2.75). ETPA—associated with higher odds of GTR (OR=0.22, p<0.001)—with a favorable trend for midline-EEA (OR=0.37, p=0.063). The GLM achieved AUC=0.756, Brier=0.19 (accuracy=0.818). 

Conclusions: Anatomical determinants—particularly petroclival origin, ICA encasement, and lower-cranial-nerve corridors—are the principal barriers to complete resection in SBC. The proposed ML provides a reproducible preoperative tool that aligns corridor choice with individual anatomy, improves likelihood of GTR, and rationalizes use of adjuvant therapy.

 

← Back to Previous Page