Cytomorphological tissue reaction patterns in lymph node tuberculosis and their correlation with bacterial density

Abstract

Background- Diagnosis of extra pulmonary lymph node tuberculosis is made by demonstration of different cyto-morphological tissue reaction patterns on FNA smears however definitive diagnosis is made by demonstration of AFB by culture or Ziehl Neelsen stained smears. This procedure is technically demanding and time consuming and is liable to fail on occasion for unexplained reasons. If identification of cytomorphological patterns could predict bacterial density, it would help to improve diagnostic accuracy and also serve as a control on the acid-fast staining procedure. Therefore this study was being undertaken to determine the correlation between bacterial density and cytomorphological patterns in lymph node tuberculosis. Methods- FNA was performed on 505 clinically suspected lymph node tuberculosis patients. May Grunwald Giemsa stained smears were used to analyse cytomorphological patterns and ZN stained smears for AFB detection. Bacterial density (BI) was calculated by utilizing Ridley’s logarithmic scale. Results- Seven distinct cytomorphological tissue reaction patterns were observed. Pattern 1 was predominantly exudative response comprising of neutrophils and mononuclear phagocytes and was the most common tissue reaction pattern 160/505 (31.6%) followed by pattern 4 i.e. epithelioid cell granulomas with necrosis 148/505 (29.3%). This study showed that multibacillary lesion (BI>1) was more often associated with pattern 1. Although the BI varied significantly across different cytomorphological tissue reaction patterns (p value=0.004), no trend was observed as both paucibacillary as well as multibacillary was seen with various proportions with all patterns. Conclusion- Present study showed that multibacillary disease is more frequently associated with pattern 1 compare to pattern 4. However more studies are needed to establish a trend among different cytomorphological tissue reaction patterns.