Most surgeons consider intraoperative frozen section examination as an important part of pituitary surgery.1 2Frozen section is used to establish a histological diagnosis (adenoma, other tumours, hyperplasia, and non-neoplastic processes) and to examine the margins of resection and determine its adequacy.1 2 A reliable intraoperative histological diagnosis will help the surgeon with achieving the ultimate goal in pituitary surgery—namely, selective removal of the lesion with preservation of endocrine function.3 With the increased popularity of the transsphenoidal approach for microsurgical removal of pituitary lesions, mainly microadenomas, the reliability of the frozen section technique becomes even more important. It has been stated that the trans-sphenoidal route provides the surgeon with suboptimal exposure of the gland and therefore smaller pieces of tissue are submitted for evaluation by frozen section.4 5 Some authors suspect that this may result in a lesser degree of accuracy of the intraoperative rapid diagnostic techniques.4 5

There are only a few papers in the English literature studying the accuracy rate of frozen section diagnosis of pituitary lesions.1-3 6 None of these compared the accuracy rate in trans-sphenoidal surgery with that of the subfrontal craniotomy approach.

We studied 55 cases of pituitary surgery. Twenty four operations were done before 1991 via the subfrontal craniotomy approach. Since August 1991 almost all pituitary surgery at our institution is done via the trans-sphenoidal route. Between 1991 and 1993 we performed 31 trans-sphenoidal pituitary operations. The patients’ medical records, operative reports, and pathological reports were studied. The histological slides of the specimens submitted for rapid intraoperative diagnosis, as well as those submitted for permanent section, were reviewed. All slides of frozen sections, touch preparations, and permanent sections, were stained with haematoxylin and eosin. Additional stains and immunohistochemical studies were performed on the permanent sections only. In addition, the size of the intraoperative specimen as measured by the examining pathologist was recorded for every case. Three cases were excluded from the study. In two cases the diagnosis could not be made intraoperatively and was deferred. Permanent section diagnosis was pituitary adenoma. The specimens submitted for frozen section during a third case were considered non-diagnostic. The patient was eventually found to have a pituitary abscess.

Of the 28 cases of transsphenoidal surgery included in the study, the diagnosis given by frozen section agreed with the permanent section diagnosis in 26 patients. None of the 24 cases of subfrontal craniotomy were excluded from the study. The frozen section diagnosis agreed with the permanent section diagnosis in 22 cases. The table illustrates the cases with discrepancy between frozen section and permanent section diagnosis. The accuracy rate for rapid intraoperative diagnosis for the transsphenoidal route was 92.9%. The accuracy rate in subfrontal surgery was 91.7%.

The size of the intraoperative specimen ranged from 0.4×0.4×0.4 cm to 1.2×1.0×0.6 cm in subfrontal surgery, and from 0.1×0.1×0.1 cm to 1.5×0.5×0.1 cm in transsphenoidal surgery. Nine trans-sphenoidal samples (32.1%) were smaller than 0.4×0.4×0.4 cm which was the smallest subfrontal sample size. However, both misdiagnosed trans-sphenoidal samples were larger than 0.4×0.4×0.4 cm. All smaller samples were accurately diagnosed.

There are a few studies available in the literature describing frozen section accuracy rates in pituitary surgery. Overall accuracy rates were found to be between 83.1 to 94.3%.1-3 6 Langet al 1 emphasised that these rates are consistently lower than accuracy rates of frozen section in other types of surgery7 8 including breast surgery.9 10Farmer et al 11 reported 91.0% accuracy of frozen section diagnosis in neuropathology. Lang et al 1 and other authors indicate that two main factors contribute to lower accuracy rates in pituitary surgery: Firstly, the size of the specimen. They state that extremely small specimens present technical difficulties which result in difficult interpretation. Secondly, the difficult task of histological differentiation between normal pituitary tissue, hyperplasia, and adenoma. Many authors indicate that trans-sphenoidal pituitary surgery provides a limited exposure and therefore results in smaller pieces submitted for intraoperative diagnosis.4 5 Therefore, it has been suspected that trans-sphenoidal surgery results in lower frozen section accuracy rates.4 5

The results of our study indicate that the surgical approach did not play a part in determining the accuracy rates. Accuracy rates of frozen section diagnosis were essentially the same for both the trans-sphenoidal and the subfrontal routes. Based on these findings and review of the literature,1 2 5 we conclude that problems with the quality of the specimen, as well as problems with histological characteristics of the frozen section, play a more important part than the size of the specimen as reasons for inaccurate diagnosis.

Touch imprint is a very helpful histological technique for intraoperative diagnosis.2 It is a very rapid procedure which provides a microscopical sample with less artifacts and with better cellular details. Its disadvantages include some loss of architectural details and the inability to evaluate invasion and margins.1