Diagnostic accuracy of imaging findings in tuberculous meningitis

Background Tuberculous meningitis (TBM) is difficult to diagnose because of its variable clinical presentation, the absence of specific imaging or cerebrospinal fluid (CSF) biomarkers, and the lack of a single diagnostic test. Focal sulcal nodular enhancing lesion (FSNEL), the imaging correlate of the Rich focus, has been suggested to aid in diagnosis, but systematic data on its diagnostic accuracy are lacking. We aimed to determine the sensitivity, specificity, and predictive value of FSNEL for diagnosing TBM in patients with chronic meningitis. Methods In this diagnostic accuracy study conducted at a tertiary care center in Northern India, consecutive patients aged ≥ 14 years with clinically suspected chronic meningitis (headache for more than five days, with or without fever, nuchal stiffness, and constitutional symptoms) were screened on clinical grounds alone before neuroimaging or CSF results. Chronic meningitis and TBM were diagnosed using composite clinical, CSF, and imaging criteria. A radiologist, blinded to the final diagnosis, assessed FSNEL on post-contrast MRI. Diagnostic accuracy measures were calculated with 95% confidence intervals, including subgroup analyses restricted to patients without characteristic imaging findings. Results Of 146 patients screened, 113 had index test and reference standard available for analysis (91 with TBM, 22 with other chronic meningitides), yielding a TBM prevalence of 80.5%. FSNEL had a sensitivity of 39.6% (95% CI 29.5–50.4%), specificity of 86.4% (95% CI 65.1–97.1%), positive likelihood ratio of 2.90 (95% CI 0.98–8.56), negative likelihood ratio of 0.70 (95% CI 0.55–0.89), positive predictive value of 92.3% (95% CI 79.1–98.4%), and negative predictive value of 25.7% (95% CI 16.2–37.2%). In patients without parenchymal lesions (n = 75), the findings remained highly specific (89.5%) with lower sensitivity (32.1%). In a stricter subgroup without tuberculomas, hydrocephalus, or basal exudates (n = 41), specificity remained similar (88.2%) with lower sensitivity (12.5%). Conclusion FSNEL is a specific, noninvasive imaging marker for TBM with a modest rule-in value and remains useful even in the absence of parenchymal lesions. Given its limited sensitivity, this finding should be interpreted alongside clinical, CSF, and microbiological assessments rather than used alone. Larger multicentre studies are warranted to validate these findings.