| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Studies |
Departments of Internal Medicine (B.B., E.M.E.), Occupational and Environmental Medicine (R.A., L.H.), Oncology (P.M.), and Neurosurgery (C.-H.N.), University Hospital, Lund, Sweden
Address all correspondence to: Dr. Eva Marie Erfurth, Department of Internal Medicine, University Hospital, S-221 85 Lund, Sweden.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Treatment of craniopharyngioma is still controversial. The choice of therapy has mainly been radical or nonradical surgery with or without radiotherapy. Several researchers advocate radiotherapy when a radical operation is not appropriate or possible (3, 4, 5, 6, 7). To assess the effectiveness of a given therapy, long follow-up is required. In most previous studies the follow-up time has been short; in only a few studies has it been 10 yr or longer (6, 8).
Patients with craniopharyngioma often present with pituitary insufficiency (9), and hormone substitution is needed. GH substitution to children, operated on for craniopharyngioma, has not been shown to increase tumor recurrence (10). As GH substitution nowadays is recommended also for GH-deficient adults, a firm knowledge of the tumor recurrence rates in unsubstituted adults is needed for a future risk assessment of GH substitution.
The primary aim of the present investigation was to examine long term survival and recurrence rates in patients operated for craniopharyngioma and to compare mortality and causes of death with a reference population. In addition, we aimed to assess the importance of predictors for survival and recurrence, especially radicality at surgery and postoperative radiotherapy. A consecutive series of patients from the South Medical Region in Sweden, operated on for craniopharyngioma during a 38-yr period, was considered an appropriate study group.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Between 1951 and 1988, 69 patients were treated for
craniopharyngioma at the Department of Neurosurgery, Lund University
Hospital (Lund, Sweden). This department is the only neurosurgical
clinic in the South Medical Region in Sweden and has during the studied
period served a relatively homogeneous population of 1.6 million. In
all cases, the diagnosis was based on radiological findings as well as
histological examinations of tissue specimens. Nine patients were
excluded due to missing records or incomplete identification. The final
group thus consisted of 60 patients, 36 males and 24 females. The
median age at the initial operation was 28 yr (range, 3–71 yr), and at
the initial operation 26 patients were under 20 yr of age (median, 9
yr; range, 3–18 yr), and 34 patients were older (median, 49 yr; range,
21–71) (Fig. 1
).
|
Fifty-nine of the patients have been followed with respect to survival and tumor recurrence from date of surgery until June 1994. One patient emigrated 2 yr after the initial operation. Twenty-seven patients were deceased at the end of the follow-up period. The median follow-up time was 12.5 yr (range, 1 day to 40 yr). Information on cause of death was obtained from the Swedish National Bureau of Statistics. The general population constituted the reference population, and expected mortality for the period 1952–1992 was obtained using cause-, gender-, county-, calendar year-, and 5-yr age group-specific death rates. Death certificates were based on autopsies for 48% of the cohort compared with 46% in the general population in the same geographical area during the same observation period. The death certificates for all deceased patients were examined.
Presenting symptoms
The median duration of symptoms before diagnosis was 6 months (range, 2 weeks to 12 yr).
In the young patients (
20 yr of age), the duration of symptoms was
shorter (median, 3 months; range, 2 weeks to 3 yr) than in the older
group (>20 yr; median, 10 months; range, 1 week to 12 yr). Main
presenting symptoms are shown in Table 1.
|
A neuroopthalmological examination was performed preoperatively in all patients who could participate. Forty-seven of 58 examined patients (81%) had reduced visual acuity, and 42 of 53 examined patients (79%) had visual field defects. Among those under 20 yr of age at diagnosis, 50% had papilloedema compared with 29% among the older patients.
Radiological evaluation
Thirty-eight patients were examined with encephalography, in most cases complemented with angiography. Twenty patients were examined with computerized tomographic scanning or magnetic resonance imaging. Two patients who were operated on in the 1950s were only examined with a scull radiograph before operation. The radiographic evaluation showed that the tumors in all cases had a suprasellar growth, and in 53 cases the third ventricle was affected. Thirty-eight percent had an enlarged or eroded sella. Forty percent had hydrocephalus at the time of operation. Among those under 20 yr of age at diagnosis, 54% had hydrocephalus compared with 29% among the older patients.
Endocrine evaluation
Pituitary function was evaluated according to tests given at each time period and in combination with symptoms and signs of pituitary failure. Preoperatively, 7 of 38 tested patients had insufficient thyroid function. Eight of 39 patients tested for adrenocorticotropic hormone function were insufficient, 16 of 23 tested for gonadal function were insufficient, and all 4 tested for GH function were insufficient. Eight patients had diabetes insipidus at the time of diagnosis. Fifty-one of 54 patients (95%) retested postoperatively had at least 1 pituitary hormone insufficiency. The median time to the diagnosis of the first postoperative pituitary hormone insufficiency was 1.6 months (range, 0–120 months). Ten patients had been treated postoperatively with GH during their childhood. Within the follow-up period, 1 childhood-onset GH deficiency patient was substituted with GH for 3 months during adulthood, and 2 more adult patients were treated for 17 months with GH. None of these patients had a recurrence or was deceased at follow-up. Ten of the 16 women, who were 48 yr or younger at diagnosis of postoperative gonadal insufficiency, were treated with estrogens. None of the 6 women who were 49 yr or older at operation had had a history of long standing amenorrhea before surgery.
Recurrence
Recurrence was defined as a reoperation due to symptomatic growth of the tumor. In all cases tumor growth was confirmed by radiological and histological examinations.
Surgical treatment
Fifty-seven patients were operated transcranially, and three
were treated by aspiration of cyst fluid. Microsurgical techniques were
used from the beginning of the 1970s. A radical excision was attempted
in all patients who were operated upon transcranially. According to the
neurosurgeon, total excision of the tumor was achieved in 30 patients,
and subtotal resection was achieved in the remaining 30 patients (Fig. 2). Among the younger patients, 73% were
regarded as radically operated, whereas only 32% of the patients over
20 yr were radically operated.
|
Twenty-two patients received radiotherapy, 11 after primary
operation and 11 after recurrence. Before 1964, 2 patients were treated
with ortovoltage and given 42.5–47 Gy with a 4-field technique. From
1964–1970, 6 patients were treated using a cobalt source with 2
opposing fields and were given 35–60 Gy. From 1971, the remaining 14
patients were treated with 50–56 Gy using linear accelerators and a 2-
or 3-field technique. However in the latter group, in 1 patient the
postoperative radiotherapy had to be interrupted due to the subject’s
poor condition. This patient therefore only received 11 Gy. The median
absorbed dose of all treated patients was 50 Gy (range, 11–60 Gy). All
patients who received radiotherapy had finished this treatment within 6
months after operation of the tumor. No case of damage to the optic
nerve or brain necrosis was reported. Six patients received local
injection of yttrium-90: one patient as the first choice of treatment,
followed by radiation within 6 months, and another 5 patients at
recurrence of the tumor. Figure 2
shows a flow chart of patients by
type of operation, radiotherapy, and recurrence.
Statistical methods
The 95% confidence intervals (CIs) for cause-specific standardized mortality ratios (SMRs) were calculated by treating the observed number as a Poisson variable or a normal variable if the observed value was greater than 15.
Statistical analyses also included Kaplan-Meier estimation of the survival curves (all causes of death) and log-rank tests comparing patients according to type of operation (radical excision vs. not), radiotherapy (none vs. postoperative vs. after recurrence), sex, age at operation and calendar period. Multivariate analysis was undertaken using Cox proportional hazards regression modelling. In these models, the instantaneous death rate ratio is estimated as a measure of comparison among the different groups. This is called the hazard ratio (HR) and is assumed in the model to be constant during the follow-up time. In certain regression models, radiotherapy and recurrence were treated as time-dependent variables. For example, when comparing survival according to recurrence status (yes vs. no), a person remains in the risk set for the no recurrence group until the time that the recurrence occurs, and then enters the risk set for the recurrence group. To test the sensitivity of the models to the proportionality assumptions, models with stratification with respect to age at the initial operation were also fitted. Similar analyses were also conducted on the incidence of recurrence rather than death. In these models, the date of first recurrence was taken as the end point. Subjects who died without a recurrence were treated as censored at their time of death. The statistical analyses were carried out in SPSS (SPSS Inc., Chicago, IL) and EGRET (Cytel Software Corp., Seattle, WA).
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The overall mortality rate was more than 5 times higher than that
of the general population (SMR, 5.55; 95% CI, 3.68–8.22). Females had
higher excess mortality (SMR, 11.4; 95% CI, 4.93–22.5) than males
(SMR, 4.79; 95% CI, 2.85–7.79). After exclusion of 4 deaths due to
craniopharyngioma, which had been misclassified as malignant tumors, no
increased mortality in malignant disease was observed (SMR, 0.78). We
found an increased risk for cardiovascular, including cerebrovascular,
mortality (SMR, 3.21; 95% CI, 1.29–6.61). Exclusion of early
postoperative deaths, i.e. starting the follow-up period 1
month after the initial operation, did not change the relative risk
estimates for either overall mortality (SMR, 4.50; 95% CI, 2.84–6.98)
or cardiovascular mortality (SMR, 3.22; 95% CI, 1.29–6.63). Both the
underlying and contributing causes of death, according to death
certificates, are presented in Table 2
for all 27 deceased patients. Four patients had postoperative cerebral
hemorrhages that contributed to their deaths, but they are not included
in this risk estimate (Table 2). Four other patients died also within 1
month after the initial operation or reoperation for recurrence. The
contributing causes of death in those patients were in 2 cases
pneumonia, in 1 case meningitis, and in 1 case pulmonary embolism. This
immediate postoperative mortality was not related to age, calendar year
of operation, or radicality of surgery. In 8 of those patients who
survived the immediate postoperative period, acute infection was the
underlying cause (1 patient) or the contributing cause (7 patients) of
their deaths later during the follow-up period.
|
The median survival time after the initial operation was 24 yr
(95% CI, 14–34). The cumulative survival rates at 10 and 15 yr were
68% (95% CI, 54–78) and 59% (95% CI, 30–63), respectively (Fig. 3).
|
and Table 2). In contrast, no patient older than 50 yr at surgery
survived more than 14 yr (median, 1.5 yr; range, 1 day to 14.8 yr;
Fig. 1). Consideration of the Kaplan-Meier curves showed that patients with radical excisions appeared to have better survival than others, but this was not statistically significant (P = 0.2). Further analyses showed, however, that radically operated patients were younger (median age, 12 yr; n = 30) than nonradically operated patients (median age, 46 yr; n = 30).
There were no statistically significant differences in survival
(P > 0.5) between the patients when divided into 3
groups according to radiotherapy (none, 38 patients; postoperative, 11
patients; after recurrence, 11 patients). Postoperative radiotherapy
and radicality of excision were also related. Only 3 of 30 with radical
excisions were given primary radiotherapy (10%) compared with 8 of the
other 30 with nonradical excisions (27%; Fig. 2).
The importance of considering time varying factors is illustrated in the comparison of survival of those who sustained a recurrence and those who did not. A standard comparison between these two groups was not statistically significant and had a HR (recurrence vs. not) of 0.7 (P = 0.4). However, when the timing of the recurrence was taken into account, the HR increased to 2.2 (P = 0.1), thus indicating a doubling of the death rate when a recurrence was diagnosed (which was not apparent from a consideration of standard Kaplan-Meier curves).
Multivariate models, including radicality at surgery, radiotherapy (yes
vs. no), and recurrence as a time-dependent factor with and
without a broad age stratification (<29 yr; 
29 yr), were fitted
(Table 3). Statistical significance was
achieved for radiotherapy and recurrence with age stratification. The
models indicate the importance of recurrence as a risk factor (HR, 4.4;
95% CI, 1.4–14) and a protective effect of radiotherapy (HR, 0.3;
95% CI, 0.1–0.8). The hazard ratios for both of these factors were
relatively consistent with and without age stratification. However, the
apparent increased risk for mortality for nonradically operated
patients (HR, 2.0) disappears with age stratification due to the
confounding described above.
|
). Tumor recurrence
Fifteen patients had a recurrence leading to a second operation,
and five patients were operated on for recurrence twice during the
follow-up period. More than half of these appeared within 5 yr; the
median time to recurrence was 26 months (range, 3–198 months). The
cumulative frequency of recurrence after 10 yr was 33% (95% CI,
22–48), and that after 15 yr was 40% (95% CI, 28–56; from the
Kaplan-Meier estimation; Fig. 4). The
incidence of recurrence did not differ statistically significantly
according to age, sex, calendar period, or radicality at surgery
(P = 0.3). The Kaplan-Meier estimation gave some
evidence for a protective effect of radiotherapy (P =
0.09). However, in the Cox regression model, the confidence interval
around the estimated HR for radiotherapy vs. none (0.3; 95%
CI, 0.1–1.3) reflects the small number of patients on which it was
based (Table 4).
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Compared to the general population, the patients operated on for craniopharyngioma in the present study had a 5-fold increase in mortality. Female patients were at especially high risk (SMR, 11.4). As expected, there was a relatively high immediate postoperative mortality (13%) caused by surgical and postoperative complications. More interesting, adult patients also suffered from a more than 3-fold increase in cardio- and cerebrovascular mortality. Mainly due to this increased risk, patients over 50 yr of age survived, on the average, only 1.5 yr postoperatively, and no one survived more than 14 yr.
These data are in accordance with the results from a recent study of 344 patients with hypopituitarism (12), in which 42 patients of the present study group were included. In this previous study, females were also at greater risk than males. In the present study operation for a craniopharyngioma was the sole inclusion criteria (and not pituitary failure per se). However, the probability of an early postoperative pituitary insufficiency is very high (in both children and adults) after surgical extirpation and, in some cases, additional radiation therapy (9, 13). Most frequently, the insufficiency affects GH and gonadotropins (9). In the present study the median time to the diagnosis of the first postoperative pituitary insufficiency was 1.6 months.
In adults with hypopituitarism receiving conventional hormone treatment, but with unsubstituted GH deficiency, an increased prevalence of premature atherosclerosis (14) as well as cardiovascular risk factors have been shown (14, 15, 16). In the present study only three adult patients, all alive at the end of follow-up, had for a limited period of time been substituted with GH. We suggest that the enhanced cardio- and cerebrovascular mortality of adults in the present study could be due to an unsubstituted GH insufficiency alone or in combination with a long-standing unsubstituted or inadequately substituted insufficiency of other pituitary hormones. Not all female patients of fertile age received adequate estrogen treatment. However, seven of the eight women who were deceased during follow-up were 48 yr or older at the time of operation, and none of them had had a history of long standing preoperative amenorrhea. It seems therefore unlikely that unsubstituted gonadal insufficiency has had any significant impact on the enhanced mortality observed in the present study.
Most of the deaths in the present study occurred shortly after operation in both the younger and the older patients. Although the present study covers a considerable period of time, the frequency of these operative deaths is comparable with more recent experiences (17). One problem with a long study inclusion period is the time-dependent discrepancy with respect to operation techniques and other therapies. However, we found no statistically significant difference in recurrence rate according to calendar year of operation. In the present study the 10-yr survival rate for the whole group was 68%. For comparison, in a group of children and adults treated with either surgery alone or combined surgery and radiotherapy, the 5- and 10-yr survival rates were 66.7% and 52.9%, respectively (7). The figures for those treated with surgery plus radiotherapy were 82.2% and 71.0%, respectively (7); the latter is close to the present results. For the youngest patients all 3 deaths occurred within 4 months postoperatively and were caused by complications to surgery (cerebral hemorrhage and acute infection), and the remaining 23 patients in this age group all were alive at the end of the follow-up. This is in contrast to the long term mortality in adults (above). It has been suggested that there is a difference in the natural history of craniopharyngiomas between young and adult patients. Some authors have described that craniopharyngiomas grow more rapidly in young subjects (18) and that older patients have a better disease-related survival (6). In contrast, others have proposed that the chronic leakage of cyst fluid over the years may create dense fibrous adhesions, making total removal difficult (19), which might contribute to the higher mortality seen in the older patients (7, 20).
In 50% of the operations, tumor removal was regarded as radical by the surgeon. In the literature, the percentage of operations classified as radical varies from 24–90% (17, 21). A total removal of tumor tissue may sometimes be hazardous due to adherence to surrounding structures, with a risk of damage to the optic nerve, the carotid artery, and the hypothalamus. Since the early 1970s, microsurgical techniques have permitted complete excision in a larger number of cases. In our material we did not find a higher incidence of immediate postoperative mortality in the radically operated group (2 of 6 who died within 1 month after surgery were thought to be radically operated) than in the nonradically operated group. Moreover, in the present material we did not find a significant difference in long time survival between the radically and the nonradically operated group. This may indicate that radicality is not such an important predictor of survival as previously suggested (4, 7). Another possibility is that the subjective assessment of radicality has introduced a misclassification bias, especially for operations performed in the premicrosurgical era. On the other hand, our data indicate an increased risk for tumor recurrence among those not radically operated. However, due to a limited number of observations, this difference was not statistically significant. It has been claimed that the critical period for recurrence is within the first postoperative years (4, 20, 21, 22), although late recurrences have been reported (23). In our study, 20 of the 60 patients had at least 1 recurrence. More than half of these appeared within 5 yr after surgery. No statistically significant difference between the recurrence rates of those younger and those older than 29 yr was seen in the present study. None of the 10 patients who received GH substitution during childhood were reoperated due to a recurrence. In agreement with the result of a previous study (10), this indicates that GH substitution during childhood does not increase the risk for tumor recurrence later in life.
Due to the long inclusion period, radiation techniques improved, and thus, there was no uniform approach to radiotherapy. Three radically and 8 nonradically operated patients were treated with radiation. Moreover, radiotherapy was given to 11 patients after recurrence. It is possible that the patients with the most aggressively growing tumors were offered radiotherapy and that this has influenced the results of the survival analysis. Several previous reports have indicated that radiotherapy may improve survival rates in patients with incompletely resected tumors (4, 5, 7). However, as mortality was highest during the early postoperative period, several patients may have died too early to receive radiotherapy. These early deaths are, of course, not explained by lack of radiotherapy. The effects of radiotherapy on survival can accordingly be properly evaluated only if patients with a limited postoperative survival are excluded. In some previous studies this exclusion has been performed (5, 7), but in most studies no such considerations have been taken.
In the present study all patients given radiotherapy had completed this treatment within 6 months. This time period was accordingly chosen for exclusion in the survival analysis. As could be expected, the difference in survival rate between the irradiated and the nonirradiated group decreased in the present study after exclusion of deaths during the first 6 postoperative months. Thus, in contrast to Rajan et al. (6), we believe that patients suitable for radiotherapy were in better condition than those who were not given radiotherapy. A randomized trial of the effect of postoperative radiotherapy would have been desirable, but the low incidence rate of craniopharyngiomas makes this a very difficult task.
Several studies have shown a higher incidence of relapses when radiotherapy had not been given to nonradically operated patients (4, 7, 8, 21, 22, 24). In the present study this observation was not clearly confirmed. However, in our study only 18% of the patients had received radiotherapy after the initial operation, which limits the possibilities for a full evaluation. Another factor that may contribute to a less protective effect of radiotherapy is the somewhat lower radiation dose used (median, 50 Gy). It has been claimed that patients receiving doses less than 50 Gy have a higher recurrence rates and shorter survival rates than those who received higher doses (7). The Cox regression model indicated a protective effect of radiotherapy on tumor recurrence, but the confidence interval reflected the small number of patients on which it was based. Furthermore, when including all patients, the multivariate models indicated the importance of recurrence as a risk factor and a protective effect of radiotherapy on survival. However, when excluding the 10 patients who died within 6 months postoperatively, of whom only 1 had had a recurrence and none had obtained postoperative radiotherapy, the time-dependent recurrence terms decreased in magnitude, and most notably, no significant protective effect of radiotherapy remained. The complexity of the factors influencing long term outcome in patients with craniopharyngiomas is obvious, but altogether, our results support the opinion that radiotherapy has a protective effect against recurrence.
To summarize, in a consecutive series of 26 children and 34 adults operated on for craniopharyngioma with a median follow-up of 12 yr, there was a 5-fold increase in the overall mortality, with a higher risk for females. In particular, a more than 3-fold increase in cardio- and cerebrovascular mortality was observed in adults. Children had the best prognosis, with all deaths caused by postoperative complications. In contrast, patients over 50 yr of age at operation had a median survival of only 1.5 yr. Further studies with multivariate modelling and with greater concern for the criteria for the selection of patients for radiotherapy are needed for a better understanding of the complex factors that determine long term outcome in patients with craniopharyngiomas. In particular, the factors that determine prognosis in adults and elderly patients need more attention and further investigations.
|
Footnotes |
|---|
1 This work was supported by a grant from the Medical Faculty,
University of Lund (Lund, Sweden). 
Received May 7, 1998.
Revised July 23, 1998.
Accepted July 28, 1998.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  N. Karavitaki, S. Cudlip, C. B. T. Adams, and J. A. H. Wass Craniopharyngiomas Endocr. Rev., June 1, 2006; 27(4): 371 - 397. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 O M Dekkers, N R Biermasz, J W A Smit, L E Groot, F Roelfsema, J A Romijn, and A M Pereira Quality of life in treated adult craniopharyngioma patients. Eur. J. Endocrinol., March 1, 2006; 154(3): 483 - 489. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Verhelst, P. Kendall-Taylor, E. M. Erfurth, D. A. Price, M. Geffner, M. Koltowska-Haggstrom, P. J. Jonsson, P. Wilton, and R. Abs Baseline Characteristics and Response to 2 Years of Growth Hormone (GH) Replacement of Hypopituitary Patients with GH Deficiency due to Adult-Onset Craniopharyngioma in Comparison with Patients with Nonfunctioning Pituitary Adenoma: Data from KIMS (Pfizer International Metabolic Database) J. Clin. Endocrinol. Metab., August 1, 2005; 90(8): 4636 - 4643. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Kendall-Taylor, P. J Jonsson, R. Abs, E. M. Erfurth, M. Koltowska-Haggstrom, D. A. Price, and J. Verhelst The clinical, metabolic and endocrine features and the quality of life in adults with childhood-onset craniopharyngioma compared with adult-onset craniopharyngioma Eur. J. Endocrinol., April 1, 2005; 152(4): 557 - 567. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. C. Smith, H. A. Lane, J. Lewis, S. Dann, J. Goodfellow, P. Collins, L. M. Evans, M. F. Scanlon, and J. S. Davies Endothelial Function and Coagulant Factors in Growth Hormone-Treated Hypopituitary Adults Receiving Desmopressin J. Clin. Endocrinol. Metab., May 1, 2003; 88(5): 2152 - 2156. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
; n = 33) with respect
to age at operation in 60 patients operated for craniopharyngioma
between 1951 and 1988.
