METHODS

Patients

Tumour specimens were obtained from 47 consecutive patients with craniopharyngioma, operated on between January 1990 and December 2000 at our hospital. The study included 21 female patients (45%) and 26 male patients (55%) with a mean (SE) age of 34.3 (2.8) years, ranging from 6 to 78 years. Sixteen patients (34.0%) were ⩽18 years of age at the time of surgery. Ten patients had previously been operated (two by us and eight in other hospitals). The reason for surgery was recurrence of the tumour in three cases and unsuccessful removal in the remaining seven patients. Only one patient had received radiation therapy before surgery.

Presenting symptoms, hormonal, and neuroradiological data were prospectively recorded for each patient. Follow up included first postoperative magnetic resonance imaging (MRI) two to six months after surgery and then at variable intervals, depending on the clinical and neuroradiological situation. Recurrence of craniopharyngioma during follow up was defined as evidence at repeat MRI of pathological tissue not detected earlier or growth of residual neoplastic tissue in comparison with the previous MRI.

Histological analysis

Surgically removed specimens were fixed immediately in 10% buffered formalin and subsequently embedded in paraffin wax. Standard haematoxylin and eosin sections were used for diagnosis. Craniopharyngiomas were subdivided into adamantinomatous and papillary tumours. Tumours with mixed characteristics were classified as adamantinomatous type.

MIB-1 and cyclin A immunostaining

Ki-67 antigen was detected by means of the monoclonal antibody MIB-1, which can be used in formalin fixed paraffin wax embedded specimens. MIB-1 and cyclin A immunostaining were performed using the avidin-biotin-peroxidase complex method on 4 μm thick sections serially obtained from the paraffin wax embedded tissue. Antigen retrieval was obtained by microwave treatment (two passages at 850 Watts of five minutes each). MIB-1 monoclonal antibody diluted 1:50 (Immunotech SA, Marseille, France) or anticyclin A monoclonal antibody diluted 1:100 (Amersham) were used as primary antibodies. No MIB-1 or cyclin A antisera were added on sections used as negative controls. The Ki-67 and cyclin A LIs were determined by a single pathologist (FM), who was unaware of the clinical characteristics of the patients. A total of at least 1000 neoplastic nuclei, subdivided in 10 fields chosen randomly at ×400 magnification, were counted.

Statistical analysis

Ki-67 LI and cyclin A LI were transformed logarithmically before statistical analysis, because they were not normally distributed. However, in the text and figures Ki-67 LI and cyclin A LI are given in the more usual decimal format. Continuous data are expressed as mean (SE), except for Ki-67 LI and cyclin A LI, which are presented as median values with interquartile ranges. Student’s t test for unpaired data was used to compare continuous variables among groups. Correlation coefficient between Ki-67 LI and cyclin A LI was calculated by the method of least squares. Stepwise regression analysis was used to test the association between Ki-67 LI and clinical and histological characteristics of the patients. The Kaplan-Meier method was used to analyse the primary end point of recurrence of craniopharyngioma during follow up. Recurrence free survival was measured from the date of surgery to the date of relapse and was censored at the date of the last MRI follow up.

A p value of less than 0.05 was considered to indicate significance.

RESULTS

Ki-67 and cyclin A LIs

Table 1 summarises the clinical characteristics of the study population. The median Ki-67 LI in all patients was 8.6% (4.4%–14%). Table 2 shows the univariate analysis of the associations between Ki-67 LI and several clinical characteristics of the patients. Because of the possible relations among some of the variables, we then construed a multivariate analysis, including in the model those variables that had a p<0.10 in the univariate analysis plus age and maximum tumour diameter. Stepwise regression analysis showed that only heavy flogistic reaction (p<0.01) and diabetes insipidus at presentation (p<0.01) were independently and positively associated with a high Ki-67 LI, whereas age, maximum tumour diameter, and histological subtype had no independent value.

Cyclin A LI could not be evaluated in three samples because of technical failure. The median cyclin A LI in the remaining 44 patients was 3.6% (1.5%–6.6%). There was a strong linear correlation between Ki-67 LI and cyclin A LI (r = 0.77; p<0.0001; fig 1). Because of the strong correlation between the two variables, cyclin A LI showed the same relations described for Ki-67 LI with clinical and histological variables.

Ki-67 and cyclin A LIs and the risk of recurrence

One patient died one week after operation because of hypothalamic derangement. Mean (SE) follow up for the remaining 46 patients was 60.4 (5.2) months, ranging from 3 to 141 months. The first postoperative MRI showed no residual tumour in 29 patients (63.0%), whereas pathological tissue was present in the remaining 17 patients (37.0%).

Recurrence of craniopharyngioma occurred in 13 of the 46 patients (28.3%). The disease free interval at five years was 69.6% (95% confidence intervals, 55.0% to 84.3%). The median Ki-67 and cyclin A LIs in the 13 recurrent craniopharyngiomas (9.0% and 2.9%, respectively) were not significantly different from that of non-recurring tumours (7.9% and 5.0%, respectively). Both Ki-67 and cyclin LIs had no independent prognostic value on the risk or recurrence either when computed as continuous or categorical variables (fig 2).

DISCUSSION

Despite their benign histological nature, craniopharyngiomas may behave aggressively and recur even after apparent complete surgical removal. The median Ki-67 LI in our study population was 8.6%, with a wide range of values, from 0.1% to 40%. Other studies found a mean (SD) Ki-67 LI of 10.8 (8.0)% and 7.8 (7.9)% in 17 and 37 patients, respectively.^10,11 In another study involving only children, the mean (SD) Ki-67 LI was 4.7 (2.3)%.¹² To put these values into perspective, pituitary adenomas, the most frequent benign tumour of the pituitary gland, seem to have a lower proliferation activity.^13–15

Among several characteristics, only presence of a strong inflammatory reaction and diabetes insipidus at presentation were independently associated with high Ki-67 LI. It is possible that craniopharyngiomas with a high proliferative index may induce a strong inflammatory reaction. It is more difficult to explain the apparent association between diabetes insipidus and Ki-67 LI. One possible explanation is that diabetes insipidus is a surrogate marker for an infiltrating and more aggressive tumour behaviour. However, we caution against overinterpretation of these results because of the small sample size of the study population.

Cyclin A LI was almost 50% lower than Ki-67 LI and strongly correlated with it. This finding implies that in craniopharyngiomas most cells entering the mitotitc cycle proceed until the late stages of the cycle with few losses. Therefore, measurement of cyclin A does not add more information than that obtained by Ki-67 LI determination alone.

As for other benign tumours of the sellar region, the factors contributing to the long term prognosis are complex. Tumour recurrence is one of the most fearful complications in patients operated for a craniopharyngioma and has strong adverse consequences on survival and quality of life. Recurrence rates from 5% to 57% have been reported in large series of craniopharyngiomas.^3,12,16–26 The variability of the reported recurrence rates is probably attributable to various factors, such as different definitions of recurrence (clinical or neuroradiological), whether surgery was aimed at total or partial removal of the tumour and whether radiation therapy was advised or not postoperatively. The recurrence rate in our study was 28% and almost all recurrences occurred within three years from surgery. The tendency to early recurrence has already been noted in other series.^12,16

Our study failed to show any association between the proliferation activity of the tumour, assessed either by means of Ki-67 LI or cyclin A LI, and the risk of relapse. Our results are in agreement with those of others,^11,12 but in contrast with those of Nishi and coworkers¹⁰ who, in their series of 17 patients, identified a Ki-67 LI greater than 7% as strongly predictive of tumour recurrence. However, in the second study¹⁰ there was an unusually high rate of recurrence (76.5%) compared with our data (28.3%). It is noteworthy that the same failure of Ki-67 LI to predict tumour recurrence has occurred in a similarly designed study involving patients with non-functioning pituitary adenomas¹³ and in odontogenic tumours, which share several histological and behavioural characteristics with craniopharyngiomas.²⁷

In conclusion, our study confirms the great variability of proliferative activity in craniopharyngiomas. Both Ki-67 LI and cyclin A LI were associated with the presence of inflammatory reaction and diabetes insipidus, but did not correlate with the long term risk of tumour regrowth. Therefore, our data do not support the routine use of both variables to identify patients at higher risk of recurrence.