Mortality in patients with non-functioning pituitary adenoma is increased: systematic analysis of 546 cases with long follow-up

Introduction

Non-functioning pituitary adenomas (NFAs) have a prevalence of 7–22/100 000 people (1, 2), a standardized incidence rate of 1.02/100 000 (3), and comprise 15–37% of all pituitary adenomas (1, 2, 3). They are found more commonly than previously thought, mainly due to the increased awareness of pituitary disease, as well as due to the advances and the easier access to diagnostic tools. As they are not associated with hormonal hypersecretion, they usually escape early diagnosis, and are mostly recognized when they are large enough to exert pressure effects to surrounding tissues; thus, at the time of detection, 67–90% of them are macroadenomas with the median age of the patients ranging between 51.5 and 65.5 years (1, 2, 3). A significant number of patients with NFAs suffer from morbidities related to the tumor, possible recurrence(s), and the treatments utilized; visual deterioration and hypopituitarism are the most frequent (4, 5, 6). Studies assessing patients with NFAs following surgery and using various diagnostic criteria have shown growth hormone (GH) deficiency in 78–88%, follicle-stimulating hormone (FSH)/luteinizing hormone (LH) deficiency in 46–90%, thyroid-stimulating hormone (TSH) deficiency in 12–57%, adrenocorticotropic hormone (ACTH) deficiency in 13–60% (7), and persistent diabetes insipidus (DI) in 15% (6); a further increase in these rates is expected if pituitary radiotherapy is added (8).

Studies assessing the long-term mortality of patients with NFA are very limited, and provide conflicting outcomes reporting no difference or a higher mortality compared with the general population (9, 10, 11, 12, 13, 14). Furthermore, the data on predictors of mortality are either conflicting (e.g. sex (9, 11, 12, 14) and radiotherapy (11, 13, 14, 15)) or are not available, and importantly, the impact of hypopituitarism and its management (particularly with the currently-used hormone substitution protocols) on survival is not yet clearly defined in this group of patients with non-secreting adenomas. Drawbacks of the published reports include the relatively small number of patients studied (10, 11), selection bias with assessment only of subjects who received post-operative radiotherapy (13), and inclusion of other types of hypothalamic–pituitary tumors in the analysis (9, 12, 13).

The aim of the present study was to assess the mortality and its causes in a large series of non-selected patients with macro NFAs diagnosed between January 1963 and December 2011 and followed up in a tertiary referral center in the UK, as well as parameters predicting survival with particular emphasis on possible roles of radiotherapy, recurrence, and hypopituitarism.

Subjects and methods

Results

Predictive factors of mortality

Clinical, hormonal, and imaging follow-up data (until date of death or if alive, until date the database was frozen) were available for 436 patients (269 males and 167 females; median age at diagnosis 58.5 years (range: 16.1–94.2)) (Table 3). This group was used for the estimation of SMRs related to long-term outcomes/management of the NFA and predictive factors of mortality. Transsphenoidal surgery had been offered to 423 (97%) and craniotomy to 13 (3%) as primary surgical treatment; adjuvant radiotherapy has been subsequently used for 123 patients (28.2%). The median follow-up was 6.9 years (1 month–48.5 years); during this interval, regrowth of the tumor had been detected in 110 patients (25.2%). Detailed data on the pituitary function and the treatment of hypopituitarism are shown in Table 3.

Table 3

Characteristics and long-term outcomes of patients with NFAs and adequate follow-up data (until date of death or, if alive, until date the database was frozen).

Characteristics	Values
n	436
Males/females	269/167
Median age at surgery (years) (range)	58.5 (16.1–94.2)
Presentation with acute apoplexy (%)	61/434 (14.1%)
Surgical approach
Transsphenoidal (%)	423/436 (97%)
Craniotomy (%)	13/436 (3%)
Patients operated on before or after 1990 (%)	36/400 (8.3/91.7%)
Extent of removal
No of intrasellar remnant (%)	202/431 (46.9%)
Extrasellar remnant (%)	229/431 (53.1%)
Post-operative adjuvant radiotherapya (%)	123/436 (28.2%)
Regrowth of tumor (%)	110/436 (25.2%)
Radiotherapy (adjuvant post-operative or for relapse)a (%)	188/436 (43.1%)
Repeat surgery for relapse (%)	41/436 (9.4%)
Median follow-up (years) (range)	6.9 (1 month–48.5 years)
Deaths	64/436 (14.7%)
Median age of death (years) (range)	76.6 (36.4–90.4)
GH statusb (%)
GH replete	16/436 (3.7%)
GH deficient untreated	222/436 (50.9%)
GH deficient treated	101/436 (23.2%)
Unknown	97/436 (22.2%)
FSH/LH status (%)b
FSH/LH replete
Males	43/269 (16%)
Femalesc	131/167 (78.4%)
FSH/LH deficient untreated
Males	34/269 (12.6%)
Femalesd	3/167 (1.8%)
FSH/LH deficient treated
Males	189/269 (70.3%)
Females	32/167 (19.2%)
Unknown
Males	3/269 (1.1%)
Females	1/167 (0.6%)
ACTH status (%)b
ACTH replete	224/436 (51.4%)
ACTH deficient treatede	209/436 (47.9%)
Unknown	3/436 (0.7%)
TSH status (%)b
TSH replete	207/436 (47.5%)
TSH deficient treated	225/436 (51.6%)
Unknown	4/436 (0.9%)
ADH status (%)b
No DDAVP	397/436 (91.1%)
On DDAVP	36/436 (8.2%)
Unknown	3/436 (0.7%)

^aMedian dose 45 Gy in 25 fractions. Median interval between radiotherapy (primary or for relapse) and death or if alive, last assessment: 7.5 years (range: 0.3–48.5). Median age when radiotherapy (primary or for relapse): 56.7 years (range: 16.1–85.1).

^bAt last assessment.

^cAlso includes post-menopausal females who developed FSH/LH deficiency after the mean age of natural menopause.

^dAlso includes females who at last follow-up were aged above the age of natural menopause but had untreated FSH/LH deficiency during their reproductive life.

^eOn hydrocortisone: total daily dose: 10–<20 mg 18/209 patients, 20 mg 170/209 patients, 25–30 mg 15/209 patients, and unknown 6/209 patients.

During the follow-up period, 64 patients (52 males and 31 females) died at a median age of 76.6 years (range: 36.4–90.4). The SMR estimated for this group was 3.8 (95% CI, 2–4.7; Table 4). In patients with tumor regrowth, this was 3.5 (95% CI, 2.1–5.5) and in those without, 3.9 (95% CI, 2.9–5.1). SMRs for subjects treated by radiotherapy (adjuvant post-operative or for relapse) or not were 4.1 (95% CI, 2.7–6.0) and 3.6 (95% CI, 2.5–4.7) respectively. Those with untreated GH deficiency had SMR 3.9 (95% CI, 2.8–5.2) and with treated GH deficiency 3.5 (95% CI, 1.8–6.3). Patients with intact ACTH axis had SMR 3.3 (95% CI, 2.1–4.8) and with treated ACTH deficiency 4.0 (95% CI, 2.9–5.4). Hydrocortisone replacement showed the following dose distribution pattern: 10–<20 mg 18/209 patients, 20 mg 170/209, and 25–30 mg 15/209. SMRs for subjects with intact TSH axis or with treated TSH deficiency were 4.2 (95% CI, 2.7–5.7) and 3.4 (95% CI, 2.4–4.8), respectively, and for those not on desmopressin or on desmopressin were 3.7 (95% CI, 2.9–4.80 and 1.7 (95% CI, 0.08–8.5) respectively (Table 4).

Table 4

SMRs based on the group of patients with adequate follow-up data (until date of death or if alive, until date the database was frozen).

Groups of patients	Observed deaths	Expected deaths	SMR (95% CI)	P value
Total group	64	17	3.8 (2.0–4.7)	<0.001
Patients with tumor regrowth	16	4.6	3.5 (2.1–5.5)	<0.001
Patients without tumor regrowth	48	12.3	3.9 (2.9–5.1)	<0.001
Patients with radiotherapy (adjuvant post-operative or for relapse)	25	6.1	4.1 (2.7–6.0)	<0.001
Patients without radiotherapy	39	10.8	3.6 (2.5–4.7)	<0.001
Patients with untreated GH deficiency	42	10.9	3.9 (2.8–5.2)	<0.001
Patients with treated GH deficiency	10	2.8	3.5 (1.8–6.3)	<0.001
Patients with intact ACTH axis	24	7.4	3.3 (2.1–4.8)	<0.001
Patients with treated ACTH deficiency	38	9.5	4.0 (2.9–5.4)	<0.001
Patients with intact TSH axis	29	7.2	4.2 (2.7–5.7)	<0.001
Patients with treated TSH deficiency	33	9.6	3.4 (2.4–4.8)	<0.001
Patients not on desmopressin	61	16.3	3.7 (2.9–4.8)	<0.001
Patients on desmopressin	1	0.6	1.7 (0.08–8.5)	0.6

On univariate regression analysis, significant predictors of mortality were age at diagnosis (HR 1.10/1 year increase; 95% CI, 1.08–1.13, P<0.001), radiotherapy (adjuvant post-operative or for relapse) (HR 2.21; 95% CI, 1.31–3.73, P<0.001), regrowth of the tumor (HR 1.99; 95% CI, 1.11–3.56, P=0.02), and untreated GH deficiency (HR 0.32; 95% CI, 0.16–0.64, P=0.01), whereas sex, presentation with acute apoplexy, cavernous sinus invasion, extent of adenoma removal, repeat surgery for relapse, untreated FSH/LH deficiency, ACTH deficiency, TSH deficiency, and treatment with desmopressin were not. Multivariate analysis including the variables found to be significant showed that only age at diagnosis remained an independent predictor of mortality (HR 1.10/1 year increase; 95% CI, 1.07–1.13, P<0.001). There was a significant difference in the rate of survival between those diagnosed before the age of 50 years and after the age of 50 (at 10 years follow-up 98.5% vs 80.9% respectively P<0.01; Fig. 1).

Cumulative probability of survival between those diagnosed before and after the age of 50 years; a significant difference was found (log-rank test; at 10-year follow-up 98.5% vs 80.9% respectively P<0.01). A full colour version of this figure is available at http://dx.doi.org/10.1530/EJE-15-0967.

Citation: European Journal of Endocrinology 174, 2; 10.1530/EJE-15-0967

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Cumulative probability of survival between those diagnosed before and after the age of 50 years; a significant difference was found (log-rank test; at 10-year follow-up 98.5% vs 80.9% respectively P<0.01). A full colour version of this figure is available at http://dx.doi.org/10.1530/EJE-15-0967.

Citation: European Journal of Endocrinology 174, 2; 10.1530/EJE-15-0967

• Download Figure
• Download figure as PowerPoint slide

Cumulative probability of survival between those diagnosed before and after the age of 50 years; a significant difference was found (log-rank test; at 10-year follow-up 98.5% vs 80.9% respectively P<0.01). A full colour version of this figure is available at http://dx.doi.org/10.1530/EJE-15-0967.

Citation: European Journal of Endocrinology 174, 2; 10.1530/EJE-15-0967

• Download Figure
• Download figure as PowerPoint slide

Discussion

This is the first study assessing the SMR of patients with NFAs based on the largest series of non-selected subjects followed-up in a single UK tertiary referral center during a long period (median 8 years). All our subjects had histopathologically confirmed diagnosis and underwent a very detailed, comprehensive clinical characterization, particularly on the tumor treatment, the degree of hypopituitarism and its management. We found that mortality was 3.6 times higher than that of the general population and that the main causes of death were cardio/cerebrovascular, infections, and malignancies. Despite the improvement in mortality in the group of patients operated after 1990, the SMRs remained high; of the number of factors assessed, only age at diagnosis was an independent significant predictor and, importantly, pituitary hormone deficits managed with mostly current protocols did not affect mortality.

Studies assessing the long-term mortality of patients with NFAs are scarce and provide conflicting data. Lindholm et al. (10), in a series of 160 patients with NFAs operated between 1985 and 1996 (18% had adjuvant radiotherapy) and followed-up for a median period of 12.4 years, found an SMR of 1.18 (95% CI, 0.87–1.60). Similar results were reported by Nielsen et al. (11) in 192 patients treated between 1985 and 1996 (22% received adjuvant radiotherapy; SMR 1.21 (95% CI, 0.93–1.59) – median follow-up for the surviving subjects 13.1 years and for those who died 4.6 years). The relatively small number of patients remains a limitation of both reports. Bates et al. (9), in a group of 348 patients with pituitary tumors, most of which were NFAs, treated between 1970 and 1992 (28% had adjuvant radiotherapy), found an SMR of 1.2 (95% CI, 0.95–1.55). Notably, a reduction in cardiovascular mortality was identified; a rather surprising finding given that 94% of the subjects had ACTH deficiency and 91% of them were treated with hydrocortisone doses of 30 mg/day, which would be considered overtreatment according to modern management criteria. On the other hand, Chang et al. (14), in a series of 663 patients with NFAs managed between 1975 and 1995 (51% received adjuvant radiotherapy) and followed up for a median period of 14 years, reported that the study population had a significantly higher mortality rate than the general population; however, no SMRs were provided. Finally, Brada et al. (13) in a group of 334 patients with pituitary adenomas, two-thirds of which were NFAs treated with radiotherapy (following surgery or as primary approach, usual dose 40–50 Gy in 20–30 fractions with conventional three-field techniques between 1962 and 1986), found a relative risk of death of 1.58 (95% CI, 1.32–1.90) compared with the general population. Notably, the relative risk of cerebrovascular deaths in patients with NFAs was 3.65 (95% CI, 2.26–5.58). This study suffers from selection bias, as it included only patients referred for radiotherapy and had no details on pituitary function and treatment of possible hormone deficits. It should be noted that the high SMR we identified may also relate to the extensive search of deaths we performed (through both National Health Service Central Register and the hospital records) which allowed us capture all deceased patients.

In our study, the SMR of the patients operated after 1990 was found 3.5, whereas for those treated before this period it was 4.7. This improvement is in agreement with Nilsson et al. (12) and the meta-analysis by Nielsen et al. (18), and possibly relates to the advances in the diagnosis and management of pituitary disease including hormonal replacement. Amongst a number of factors assessed and following multivariate analysis, only age at diagnosis remained an independent predictor of mortality (shorter survival for older patients) and the probability of survival was significantly lower in subjects aged more than 50 years. Nilsson et al. (12) and Chang et al. (14) have also confirmed the negative impact of age at presentation. Furthermore, acute apoplexy and extent of resection have not been found as independent predictors in previous series (14, 19).

The causal link between pituitary irradiation and mortality (particularly cerebrovascular) remains a subject of debate (20) and relevant randomized trials are not available. In our study, the median follow-up after radiotherapy was 7.5 years (range: 0.3–48.5), and this treatment modality did not affect survival. Although, Brada et al. (13) reported high SMRs in patients with various types of irradiated pituitary adenoma, this study lacked a control group. On the other hand, series including only patients with NFAs are in agreement with our results. Thus, Nielsen et al. (11) reported that the SMR of patients with NFAs who had received radiotherapy was similar to that of those treated solely by surgery. Similarly, Chang et al. (14) showed that radiotherapy was not an independent predictor of survival, while Tomlinson et al. (15) in a series of 573 patients with NFAs included in the West Midlands Hypopituitary database found no effect of radiotherapy on mortality.

In our series, based on a non-selected group of patients which, importantly, did not include any type of functioning adenomas, we found that hypopituitarism is not an independent predictor of mortality. The advances in the management of pituitary hormone deficits (earlier diagnosis, more optimal hormone substitution, and closer monitoring) may have contributed to this. Previously published observational studies on this topic have a number of drawbacks, making the interpretation of their results rather difficult; thus, they rely on series of small number of patients (10, 21), with occasionally rather outdated diagnostic criteria (21, 22, 23) and different or suboptimal hormonal regimes (often higher doses of steroids than the ones currently used, non-routine sex hormone substitution or GH) (9, 15, 21, 22, 23, 24). These factors may influence the reported outcomes and the conclusions of studies drawn mainly from early calendar periods do not necessarily reflect the prognosis of patients managed in the latest years (9, 21, 22, 23). Furthermore, a number of series include heterogeneous groups of underlying causes of hypopituitarism (e.g. functioning adenomas and craniopharyngiomas) managed with various modalities having the potential to affect mortality independently of pituitary hormone deficits (9, 21, 22, 23, 24, 25). Tomlinson et al. (15) in a series of 573 patients with NFAs included in the West Midlands Hypopituitary database found an SMR of 1.70 (99% CI, 1.34–2.15) with the excess mortality explained by an increase in respiratory and vascular deaths. Patients with untreated gonadotropin deficiency had a worse outcome compared with those with treated deficiency; notably, the percentage of subjects with untreated gonadotropin deficiency was 37% compared with only 14% in our study. It should be also pointed out that selection bias may be present in this series, as it does not include consecutive NFAs but only those with hypopituitarism.

In our study, ACTH deficiency was not a predictor of mortality; notably, in around 90% of our patients the total daily dose of hydrocortisone was 20 mg. This is in agreement with Zueger et al. (26) who, in a series of 105 patients with NFAs and ACTH deficiency, found that the HR for mortality was one for those treated with a total dose between 5 and 19 mg daily but increased to 2.03 for total dose in the range of 20–29 mg. Furthermore, based on data of hypopituitary patients from the KIMS cohort (Pfizer International Metabolic Database) (27), ACTH deficient patients on hydrocortisone-equivalent doses of <20 mg/day do not differ in their metabolic profile from the ACTH deplete. Given our small number of patients on total daily dose >20 mg, we did not analyze the impact of total dose on mortality. Interestingly, 67% of our patients who died of infections had ACTH deficiency, and a contribution of adrenal insufficiency cannot be excluded.

Multivariate analysis did not confirm GH status as an independent predictor of mortality. It has been shown that GH replacement improves cardiovascular risk factors but to date, only a limited number of studies (mainly post-marketing surveillance) have investigated its effect on mortality in GH deficient adults and long-term controlled trials on mortality (including cardiovascular) are lacking (28). Svensson et al. (29) in a series of 1411 adult hypopituitary patients not on GH replacement found mortality higher than that of the normal population (risk ratio 3.36; 95% CI, 2.93–3.83) with increased rates of myocardial infarctions, cerebrovascular events, and malignancies. In the same report, patients on GH replacement had overall mortality similar to the normal population. On the other hand, data from the Dutch National Registry of Growth Hormone Treatment in Adults have suggested that mortality in patients treated with GH was high in females mainly due to cardiovascular disease, whereas, in the control group of untreated subjects this was not different from the background population (30). In both studies however, information on the underlying causes of hypopituitarism and its management, as well as details on the presence of other pituitary hormone deficits and their treatment are not included. It should be noted that, in our center, GH-deficient patients are offered GH replacement after assessment of their quality of life with a standardized questionnaire (Assessment of Growth Hormone Deficiency in Adults (AGHDA)). Those with optimal scores are not candidates for treatment and, therefore, there is necessarily a selection bias introduced by the pre-screening of candidates for GH replacement according to the NICE guidelines; this needs to be taken into account when considering the overall impact of GH replacement on mortality in this series.

Very recently, while this manuscript was being prepared for submission, a study of 2795 patients with NFA followed in nationwide health registries in Sweden between 1987 and 2011 was published reporting SMR of 1.10 (95% CI, 1.00–1.20) (31). The validity of the diagnosis is a drawback of this report as the identification criteria checked in a sample of 467 subjects confirmed NFA in 424 (90.8%) and importantly, 1172 patients diagnosed between 1997 and 2011 had received no surgical treatment or radiotherapy, thereby lacking histological confirmation (notably, information on surgical treatments and radiotherapy were not available for those diagnosed before 1997). Furthermore, the SMR for patients with hypopituitarism (defined by entry of this term in the patient registry) was found 1.06 (95% CI, 0.94–1.19); nonetheless, data on the diagnosis and degree of hypopituitarism, the percentage of subjects tested, as well as its management – factors of major importance when assessing outcome – are not available. Finally, in contrast to our detailed clinical characterization which allowed assessment of the impact of factors like acute apoplexy, extent of NFA removal and regrowth, the recently published series evaluates the effects of the total number of treatments offered for the NFA without providing or taking into account details on the natural history of each tumor and the reason the treatments were used (for primary adenoma or for relapse).

The large sample size of non-selected patients with histologically confirmed NFAs followed up for a long period in one center and managed with similar (and mostly current) protocols in terms of hormone substitution are the strengths of our study, whereas limitations include its retrospective nature, the selection bias in GH treated patients, and the unknown duration of pituitary hormone deficits prior to their diagnosis (which is however, practically not possible to be clarified).

In conclusion, patients operated for an NFA have increased mortality and factors related with the management/outcome of the tumor including presentation with acute apoplexy, extent of adenoma removal, radiotherapy, and recurrence do not seem to be independent predictors. We have also shown that pituitary hormone deficits managed with the currently used substitution protocols do not adversely affect mortality, although the impact of GH replacement on survival remains to be assessed with more robust studies specifically designed to address this question. The only parameter affecting mortality in the patients followed-up in our center was age at diagnosis, a finding also supported by previous studies. The pathophysiological mechanisms of this or alternative explanations for the increased mortality remain to be elucidated. However, it is reassuring that current treatment modalities and replacement therapies for these tumors do not adversely affect survival.

Declaration of interest

The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.

Funding

Dr G Ntali has been rewarded a Clinical Department Visit Grant from the Society for Endocrinology (UK).