xRead - September 2022

Oakley et al.

expensive, and less accurate test than beta-2 transferrin or beta trace protein. Based on these facts, RNC should not be routinely employed to confirm the presence of a CSF leak (Supporting material 6). 24, 33–36 Aggregate grade of evidence: C (Level 3: 4 studies; Level 4: 1 study); Benefit: Moderately accurate for confirmation; localization limited to side of skull base that is leaking; Harm: Invasive study–potential risk of lumbar puncture and intrathecal contrast injection; Cost: High ($693.64 to $705.08); Benefits-harm assessment: Limited benefit with moderate potential harm and significant costs; Value judgments: Low level of evidence; less accurate, more invasive, and more expensive than other available tests; Recommendation level: Recommendation against; Intervention: Avoiding the routine use of RNC for confir mation of CSF leak. Localization Once a CSF leak has been confirmed, localization of the site of the leak is essential for preoperative planning. Various modalities are available for this purpose including HRCT, MRC, CTC, and IF, but it has been unclear which study or combination of studies is most effective with respect to accuracy, safety, and cost. Although HRCT is primarily a localization tool, the others are capable of simultaneous confirmation and localization. High resolution computed tomography HRCT is a common initial choice for localization given that it can familiarize the surgeon with the patient’s general sinonasal anatomy, as well as identify bony defects associ ated with CSF leak sites. Of the initial 762 articles from the 3 database searches, 16 met inclusion criteria and were rel evant to HRCT in the setting of CSF leaks. All of these 16 were level 3 evidence cross-sectional studies. In the setting of CSF leak localization, these studies reported a sensitivity and specificity for HRCT of 44% to 100% 13, 24, 34, 37–47 and 45% to 100%, 34, 46 respectively, with the majority of the studies being on the upper end of this scale. Reported pos itive predictive values and negative predictive values were 100% 13 and 50% to 70%, 13, 44 with an accuracy of 87% to 93%. 24, 44 Tahir et al. 46 and Eljamel et al. 40 reported the lower HRCT sensitivities of 43% and 48% with a specificity of 45%. Tahir et al. 46 report being limited by constrained resources in Pakistan and not having HRCT as a readily accessible study. They also noted that beta-2 transferrin is not available at all for leak confirmation. It is unclear whether HRCT or standard CT was used in this study. In addition to a low sensitivity, Eljamel et al. 40 also reported a 9.5% false positive and 67% false negative rate. In this study, they were assessing a specific population of patients who had inactive CSF fistulae. They considered demon stration of a bone defect or sinus fracture on HRCT as a

positive sign, but not pneumocephalus or air-fluid levels alone. Sillers et al. 45 also found a lower CSF fistula iden tification rate of 62% with HRCT, but 100% identifica tion of skull-base defects confirmed surgically. Their study includes more post–functional endoscopic sinus surgery (FESS) patients who may be more likely to have fine frac tures as leak sites rather than obvious skull-base defects, in contradiction to patients with leaks following closed head trauma. In some instances, HRCT may reveal findings consistent with CSF leak other than overt bony defects. Manes et al. 48 looked specifically at 15 patients with spontaneous CSF rhi norrhea and negative HRCT findings. They found that all 15 spontaneous CSF leak patients with no identifiable bony defect had an olfactory cleft opacification that marked the leak site with 100% accuracy (confirmed intraoperatively). Of the 14 studies that reported HRCT sensitivities, 12 re ported them to be over 80%, 13, 24, 34, 37–39, 41–44, 47, 49 with 1 study varying between 73% to 87% depending on location of leak. 43 In the La Fata et al. 41 study, even the size of the defect was accurately predicted on HRCT with multiplanar reformatting to within 2 mm in 75% of cases, with thinner cuts being more accurate. Some studies encourage the use of HRCT alone for lo calization. Stone et al. 34 found it to be 100% accurate in its series of patients (21/21 confirmed surgically) and recommended other modalities be reserved for instances in which a bony defect is not identified on HRCT. Sim ilarly, Zapalac et al. 24 recommended HRCT be used for localization in their diagnostic algorithm, as it was 87% accurate and cost $504 in their analysis, making it both more accurate and less costly than MRC or CTC. It should be noted, however, that dehiscences in the skull base oc cur naturally, so that not all defects represent CSF leaks. Because of this lack of specificity, other studies recom mend using a combination of HRCT and MRC, reporting a combined sensitivity or accuracy of 90% to 96%. 42, 44, 45 Whether or not to use further localization modalities would likely need to be determined on a case by case basis. However, given these findings in the literature, HRCT appears to be the best initial choice for localization and one that will be obtained in most cases simply to provide the surgeon with knowledge of the anatomy (Supporting material 7). 13, 24, 34, 37–50 Aggregate grade of evidence: C (Level 3: 16 studies); Benefit: Noninvasive, high accuracy; Harm: Low risk associated with radiation; Cost: Moderate ($280.32); Benefits-harm assessment: Preponderance of benefit over harm; Value judgments: Highly accurate test for knowledge of sinus anatomy and localization of a CSF leak is important for operative planning in nearly all cases; Recommendation level: Recommendation for; Intervention: Using HRCT for localizing CSF leaks.

International Forum of Allergy & Rhinology, Vol. 6, No. 1, January 2016

12