Role of crush smear cytology in the diagnosis of gastrointestinal malignancy.

2014 
: Crush smear cytology, commonly used for central nervous system lesions was reported to be useful in the diagnosis of GI malignancies. This study was designed to see the accuracy of crush smear cytology in detection of gastrointestinal malignancy in relation to histopathological examination. First 4 or 5 bits of pinch biopsy specimens from each of the consecutive patients having endoscopic findings suggestive of carcinoma of gastro-intestinal tract were examined by conventional paraffin embedding and H-E staining by a cytologist. Crush smears stained with Papanicolaou's stain were prepared with the last bit of specimen and were examined by another cytologist. The diagnostic accuracy was examined by correlating with clinical and histological data. Out of 100 cases of suspected oesophageal malignancies, 99 were diagnosed as carcinoma by histopathology and 84 (sensitivity 83.83%, accuracy 83%, K - 0.14) cases were positive for malignancy by crush smear cytology. Out of 60 cases of gastric lesion, 54 and 44 cases were proved to be malignant by histopathology (sensitivity 76%, accuracy 73.3%, K - 0.2) and crush smear cytology respectively. Fifty four of the 57 cases of colonic lesions were proved to be malignant by histopathology and 50 (sensitivity 91%, accuracy 89.5%, K - 0.34) were malignant by crush smear cytology. Combining two methods the accuracy was 100%, 95% and 96.5% in detecting oesophageal, gastric and colonic malignancies respectively. Concordance rate of both the methods in diagnosing oesophageal, gastric and colonic lesions were 883.83%, 73.3% and 89.5% respectively. Crush smear cytology is a cheap, easy and rapidly performing technique. The diagnostic yield is very high when the technique is combined with histopathology. It may be used as a useful adjunct to conventional histopathology.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    0
    References
    2
    Citations
    NaN
    KQI
    []