Cytohistologic correlation of suspected Cervicofacial Head & Neck Extra-Pulmonary Tuberculosis in children: A retrospective case series

Background Tuberculosis (TB), especially extrapulmonary TB, is a difficult diagnosis to make in children due to the paucibacillary nature of paediatric disease and difficulty in obtaining sputum and tissue samples for microbiology confirmation. Lymphadenopathy in children with suspected cervicofacial TB are amenable to FNA or surgery for further cytological and histological assessment. It is therefore important to understand how well the morphologic features (from cytology and histology) correlate with defined reference standards (TB culture and molecular evidence of MTb) for the diagnosis of TB and the reliability of these features. Aim The aim of the study is to determine how well the cytology and histology-made TB diagnoses in children with suspected cervicofacial EPTB correlates with TB culture and MTb PCR results. Materials and methods This is a descriptive retrospective study that involved a re-appraisal of all patients with suspected cervicofacial EPTB who had histology and cytology performed at Red Cross Children's Hospital identified from the National Health Laboratory Service (NHLS) Trakcare system over a 5 year period (2012-2017). Following identification of histopathology accession numbers, histopathology reports and slides were retrieved from the archive of the Division of Anatomical Pathology/ National Health Laboratory Service, Red Cross Children's Hospital, Cape Town for evaluation. In addition, results for Genexpert testing and TB culture were identified using the National Health Laboratory Service (NHLS) Trakcare system. In patients that did not have either of the above, MTb PCR testing was performed. Results Data from the reports of 76 children with suspected cervicofacial TB were included in this study. More biopsies were submitted for histology (48) than for cytology (22). Six children had biopsies for both cytology and histology done. Most children had suspected and confirmed TB involvement of the cervical lymph nodes. On histology, the feature that correlated the best with proven TB was necrotising granulomatous inflammation (79.5% of cases had confirmed TB). On cytology, necrotising inflammation, necrotising granulomatous and non-necrotising granulomatous inflammation correlated well with proven TB. The sensitivity of cytology was 77.3% against TB culture and 81.8% against GXP for TB diagnosis. Whilst for histology the sensitivity was 82.5% against TB culture and 90.3% against GXP as reference standards for TB diagnosis. Conclusion FNA for cytology is a safer procedure with less complications than biopsy for histology. Also, the use of cytology together with a GXP renders a rapid and accurate diagnosis of TB and our findings are supportive for the combined use of these modalities as first line investigations. However, every attempt should still be made to obtain a sample for TB culture (as the WHO recommended gold standard for TB confirmation)