Intra-operative frozen section analysis for suspected early-stage ovarian cancer: 11 years of Gateshead Cancer Centre experience.

OBJECTIVE: In centres in which intra-operative frozen section (FS) analysis is not performed, 'apparent' early-stage ovarian cancer diagnosed after surgery on paraffin section may require further restaging laparotomy or adjuvant chemotherapy. Previous studies on FS analysis have reported high sensitivity, specificity and overall accuracy. The objective of this article is to present the largest published dataset on the accuracy of FS analysis over an 11-year period from a single institution. DESIGN: Diagnostic test accuracy. SETTING: Northern Gynaecological Oncology Centre and Department of Cellular Pathology, Gateshead, UK. POPULATION: 1439 intra-operative FS analyses performed between January 2000 and December 2010 for suspected ovarian cancer. METHODS: Prospectively collected data on FS analysis were compared with gold standard paraffin section. MAIN OUTCOME MEASURES: Sensitivity, specificity, likelihood ratios and post-test probability. RESULTS: The overall sensitivity and specificity of FS analysis were 91.2% and 98.6%, respectively. Positive and negative likelihood ratios were 64.7% and 0.09%, respectively. The pre-test probability of an ovarian tumour being borderline or malignant was 45.8%. When FS analysis was reported to be positive, the post-test probability increased to 98% (confidence interval, 97-99%). Conversely, when FS analysis was reported to be negative, the post-test probability decreased to 7% (confidence interval, 6-9%). The majority of false test results were either borderline tumours or of mucinous differentiation. CONCLUSIONS: Intra-operative FS analysis has excellent diagnostic test accuracy and assists gynaecological oncologists to perform the appropriate surgery in 95% of cases, thereby preventing the morbidity of surgical staging in benign cases and the morbidity of restaging procedures or chemotherapy in early-stage malignant tumours.

Comparison of examination of the entire uterine cervix with routine cervical sampling in hysterectomy specimens from women with endometrial cancer.

BACKGROUND: Cervical involvement by endometrial cancer alters the FIGO stage and determines clinical management, but there are no accepted guidelines for cervical sampling of these cases. AIM: To assess whether sampling more than two "routine" blocks of the cervix (anterior and posterior) alters the pathological staging of hysterectomy specimens for endometrial cancer. METHODS: Histological involvement of the cervix was prospectively compared in hysterectomies performed for proven endometrial cancer (n = 61). Specimens had two "routine" blocks taken from anterior and posterior cervix; all of the remaining cervix was also processed for histological assessment. RESULTS: 61 cases of endometrial cancer had the entire uterine cervix processed. There were 54 cases of endometrioid adenocarcinoma and 7 special types. Twelve cases had cervical involvement (stage 2A or 2B), and seven cases were stage 3A or above, of which three also had cervical involvement. In none of the 61 cases did the additional cervical blocks (n = 544) taken alter the staging made on the "routine" blocks. CONCLUSION: Sampling of two blocks from the cervix appears sufficient for histological staging of endometrial cancer in hysterectomy specimens.

Prognostic indicators for gastrointestinal stromal tumours: a clinicopathological and immunohistochemical study of 108 resected cases of the stomach.

AIMS: Whether immunohistochemical markers increase accuracy in predicting prognosis for gastrointestinal stromal tumours (GISTs) remains uncertain. However, past studies have used only small, heterogeneous patient groups. Our aim was to test previously studied and more novel morphological features as well as four immunohistochemical markers as prognostic indicators amongst a large cohort of surgically resected, gastric GISTs. METHODS AND RESULTS: Tissues from 127 gastric mesenchymal tumours were collected retrospectively and subjected to repeat histological assessment and immunophenotyping. Further immunohistochemistry was performed for Ki67, p53, Bcl-2 and cyclin D1. Complete follow-up data were collected for 108 patients with immunophenotyped diagnoses of GIST (i.e. c-kit+ tumours). At the census point, 52 patients were alive, 24 had died from their GISTs and the remainder of other causes. Univariate analysis showed the following predicted for shorter disease-specific survival: size > or =50 mm; necrosis, no intratumoral lymphocytes; mitotic count > or =5/50 high power fields; Ki67 labelling index > or =5%; p53 immunopositivity. Of these variables, multivariate analyses showed only mitotic count and, to a lesser extent, Ki67 labelling to be independent prognostic indicators. CONCLUSIONS: Mitotic count remains the best predictor of outcome following surgical resection of gastric GISTs. Ki67 immunohistochemistry does not provide better prognostication and p53, Bcl-2 and cyclin D1 immunohistochemistry provide no additional prognostication.