Document resource
Purpose To evaluate the efficacy and safety of flow diversion for direct (Barrow A) carotid-cavernous fistula (DCCF) using pooled individual-patient data (IPD), and to test whether a higher flow-diverter (FD) device count shortens time-to-occlusion and improves clinical outcomes.Methods We performed a retrospective pooled IPD analysis of DCCFs treated with FD stents from two sources: (1) a consecutive single-centre series (January 2018-September 2025) and (2) published reports identified by systematic searches of PubMed/MEDLINE, Embase, and CENTRAL (inception-September 2025), with reference/citation hand-searching. Inclusion required angiographically confirmed direct CCF, endovascular treatment using an FD (primary or adjunct/secondary), and extractable patient-level core outcomes. Prespecified variables included age, sex, etiology (traumatic/iatrogenic/spontaneous), treatment sequence (primary vs secondary), FD count, and adjuncts (FD alone vs coils/liquid embolic/covered stent graft). Primary endpoints were complete angiographic occlusion and time to occlusion (days); secondary endpoint was clinical improvement or complete symptom resolution; safety endpoint was any periprocedural or delayed complication. Associations between FD count and outcomes were tested using Pearson/Spearman correlations, ordinary least-squares regression for occlusion time, and logistic regression plus chi-square testing for clinical improvement (two-tailed α=0.05).Results Seventy-two patients were included (mean age 43 years; 30 female, 27 male, 15 unspecified). Etiology was traumatic in 42/72 (58%), spontaneous in 16/72 (22%; including 8 ruptured cavernous carotid aneurysms), and iatrogenic in 14/72 (19%). Flow diversion was primary therapy in 47/72 (65%) and secondary/adjunct in 25/72 (35%). FD alone was used in 31/72 (43%); FD followed by cavernous sinus coiling in 18/72 (25%); coiling first then FD in 16/72 (22%); FD plus liquid embolic in 4/72 (5%); FD plus covered stent graft in 2/72 (3%); and 2/72 (3%) ultimately required parent-artery occlusion. FD count was reported in 70 patients (151 devices; mean 2.1; range 1-8). Complete angiographic occlusion occurred in 67/72 (93%), including immediate closure in 5/72 (7%). Occlusion time was available in 62 cases (range 0-720 days; mean 133.5 days; median 105 days), with nearly two-thirds occluding within 6 months. Clinical improvement/resolution occurred in 53/72 (73%). No complications were reported in 56/72 (78%); adverse events included subarachnoid haemorrhage, focal neurological deficits, internal carotid artery dissection/occlusion, and bleeding related to dual antiplatelet therapy. FD count was not associated with occlusion time (Pearson r=0.075, p=0.56; Spearman r=0.238, p≈0.062; slope 5.34 days per FD, 95% CI −13.1 to 23.8, p≈0.56) or with clinical improvement (odds ratio 0.82 per FD, 95% CI 0.58-1.14, p=0.23; chi-square p=0.26).Conclusion In pooled IPD, flow diversion achieved a high angiographic cure rate for DCCF with generally favourable clinical recovery, but with a characteristic latency to closure (median ~3.5 months). Increasing FD device count did not shorten time-to-occlusion nor improve clinical outcomes, suggesting device number reflects lesion complexity rather than a modifiable determinant of success. These findings support carotid-preserving reconstruction with counselling regarding delayed occlusion and vigilance for uncommon but serious complications.Disclosures D. Varma: None.