RESUMEN
BACKGROUND: Tuberculous spondylodiscitis (TS) is the most common form of musculoskeletal tuberculosis. Currently, histology is widely used to distinguish tuberculous from nontuberculous disease. OBJECTIVES: The aim of the present study was to assess the accuracy of histology compared with bacteriology in the diagnosis of TS. METHODS: This is a single-center case series carried out from January 2014 to February 2018 in a pathology department. It included 121 discovertebral biopsies of infective spondylodiscitis. The measures of diagnostic accuracy of histology were determined taking bacteriology as criterion standard. RESULTS: Among the 121 cases, 55 (45.4%) were diagnosed as TS by histological and/or bacteriological findings, 17 (30.9%) were classified as definite TS by bacteriology, and the remaining 38 (69.1%) had positive histology and negative bacteriology. There were 2 false-negatives, which histologically displayed suppuration without granuloma, and 3 false-positives; in one case, histology displayed granulomas without necrosis and culture isolated Brucella. In the 2 others, histology revealed granulomas with caseous-like necrosis and microbiology isolated fungal species. The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of histology in the diagnosis of TS were 88.2%, 93.4%, 83.3%, 95.5%, and 92%, respectively. CONCLUSIONS: Histology is proved to be an accurate diagnostic tool in TS. Suppurative forms of TS without granuloma are rare and represent the main cause of false-negative histology. Suggestive histology of TS does not rule out fungal and brucellar spondylodiscitis. Caseous necrosis is not pathognomonic of tuberculosis. Fungal infection can also exhibit such type of necrosis.