J Neurooncol DOI 10.1007/s11060-014-1386-5
CLINICAL STUDY
The pituitary stalk effect: is it a passing phenomenon? Marvin Bergsneider • Leili Mirsadraei • William H. Yong Noriko Salamon • Michael Linetsky • Marilene B. Wang • David L. McArthur • Anthony P. Heaney
•
Received: 25 November 2013 / Accepted: 21 January 2014 Springer Science+Business Media New York 2014
Abstract Most patients with large pituitary tumors do not exhibit hyperprolactinemia as a result of pituitary lactotroph disinhibition (stalk effect). Studies have demonstrated that increased intrasellar pressure is associated with both ‘‘stalk effect’’ hyperprolactinemia and pituitary insufficiency. Our primary hypothesis was that, despite continued disinhibition, lactotroph failure is responsible for normoprolactinemia in patients with large macroadenomas. As a corollary, we proposed that the hyperprolactinemia phase, which presumably would precede the insufficiency/ normoprolactinemic state, would more likely be discovered in premenopausal females and go unnoticed in males. Prospective, consecutive surgical series of 98 patients of clinically nonfunctional pituitary adenomas. Lactotroph insufficiency was inferred by the coexistence of insufficiency in another pituitary axis. The existence of preoperative lactotroph disinhibition was inferred based on comparison of pre- versus post-operative prolactin levels. 87 % of patients with tumor size [20 mm and
normoprolactinemia had pituitary insufficiency. Pre-operative prolactin in patients with pituitary insufficiency were lower than those with intact pituitary function. Prolactin levels dropped in nearly all patients, including patients with normoprolactinemia pre-operatively. Premenopausal women had smaller tumors and higher pre-operative prolactin levels compared to males. No premenopausal female exhibited evidence of pituitary insufficiency. Our study provides suggestive evidence that the ‘‘stalk effect’’ pathophysiology is the norm rather than the exception, and that the finding of normoprolactinemia in a patient with a large macroadenoma is likely a consequence of lactotroph insufficiency. In males, the hyperprolactinemia window is more likely to be missed clinically due to an absence of prolactin-related symptoms.
M. Bergsneider (&) Department of Neurosurgery, David Geffen School of Medicine at UCLA, Gonda 3357, BOX 951761, Los Angeles, CA 90095-1761, USA e-mail: [email protected]
M. B. Wang Department of Head and Neck Surgery, David Geffen School of Medicine at UCLA, 200 Medical Plaza, Ste 550, BOX 956959, Los Angeles, CA 90095-6959, USA
L. Mirsadraei W. H. Yong Division of Neuropathology, Department of Pathology, David Geffen School of Medicine at UCLA, 18-161A NPI, BOX 951732, Los Angeles, CA 90095-1732, USA N. Salamon M. Linetsky Division of Diagnostic Neuroradiology, Department of Radiology, David Geffen School of Medicine at UCLA, RRUMC 1621, BOX 957437, Los Angeles, CA 90095-7437, USA
Keywords Hyperprolactinemia Pituitary adenoma Dopamine Prolactin Pituitary insufficiency
D. L. McArthur Department of Neurosurgery, David Geffen School of Medicine at UCLA, 18-266 NPI, BOX 957039, Los Angeles, CA 90095-6901, USA A. P. Heaney Division of Endocrinology, Department of Medicine, David Geffen School of Medicine at UCLA, 24-130 WH, BOX 957073, Los Angeles, CA 90095-7073, USA
123
J Neurooncol
Introduction The ‘‘pituitary stalk effect’’ [1–3] is an elegant explanation describing the modest hyperprolactinemia associated with large pituitary tumors [4–7]. Compression of the pituitary stalk was proposed to diminish dopaminergic inhibition of normal lactotrophs, resulting in elevation of prolactin (PRL) levels. To date, however, it remains an enigma why the majority of patients with large clinically nonfunctional (CNF) pituitary tumors do not exhibit ‘‘stalk effect’’ hyperprolactinemia. The fact that tumor size poorly correlates with serum PRL levels [8, 9] suggested alternative mechanisms for the hyperprolactinemia [10], including pharmacological side-effects or presence of certain medical conditions [11]. Other studies have demonstrated that, rather than tumor size, hyperprolactinoma correlates well with increased intrasellar pressure [12, 13]. At some critical intrasellar pressure threshold, the portal flow to the pituitary was presumed to be disrupted, leading to the disinhibition of lactotroph PRL production. Interestingly, these same studies found that increased intrasellar pressure is also strongly associated with pituitary insufficiency [14–16]. The aim of our study was to gain an understanding of why the majority patients with large CNF macroadenomas do not exhibit hyperprolactinemia. Extrapolating from the above studies focused on intrasellar pressure [14–16], we hypothesized that the combination of pituitary insufficiency and portal flow disruption could create a state in which hyperprolactinemia can no longer be sustained. Specifically, a critical reduction in the number of lactotrophs, the manifestation of lactotroph insufficiency, resulted in normo- (or hypo-) prolactinemia even despite continued disinhibition by portal flow disruption. A challenge in studying this hypothesis was the inability to directly assess lactotroph insufficiency clinically. As a surrogate, we inferred that insufficiency in other pituitary axes would increase the likelihood of concomitant lactotroph failure. The validity of this approach has been demonstrated with growth hormone deficiency [17–19]. Moreover, absolute hypoprolactinemia is strongly associated with severe pituitary insufficiency [20–23]. Importantly, however, here we are proposing that normoprolactinemia can be the manifestation of lactotroph insufficiency in a patient for whom ‘‘stalk effect’’ hyperprolactinemia would be expected. This reasoning predicts that patients with large CNF macroadenomas and normal PRL levels have a higher prevalence of pituitary insufficiency. Conversely, patients with ‘‘stalk effect’’ hyperprolactinemia are less likely to exhibit pituitary insufficiency. By including lactrotroph failure into the concept of ‘‘stalk effect,’’ we are proposing a multiple state scenario that includes normal (or low) PRL levels in addition to the customary hyperprolactinemia ‘‘stalk effect’’ state. Logically,
123
lactotroph insufficiency (and hence normoprolactinemia) should occur later in the clinical course, whereas one would expect hyperprolactinemia to occur earlier, being more common with relatively ‘‘smaller’’ CNF macroadenomas. We further postulated that a patient presenting with a combination of normoprolactinemia, a large CNF macroadenoma, and pituitary insufficiency, would have likely transitioned through a hyperprolactinemic phase while the tumor was smaller in size. This would raise the question of why the hyperprolactinemic phase would have gone unnoticed. From a clinical standpoint, we reasoned that males and post-menopausal females would have been that group as they would have been less likely to be diagnosed with hyperprolactinemia compared with premenopausal females. Conversely, premenopausal females would more likely present with hyperprolactinemia-related symptoms, and therefore be diagnosed with relatively smaller clinically nonfunctional macroadenomas as they enter the ‘‘stalk effect’’ phase.
Clinical materials and methods Over the 4-year period from 2008 to 2012, 312 consecutive patients undergoing endoscopic, endonasal resection of suspected pituitary adenomas by the senior author (M.B.). An institutional review board approved informed consent was obtained for the prospective collection of clinical, imaging measurements, and quantitative immunohistochemistry data. Exclusion criteria (1) diagnosis of acromegaly, Cushing’s disease, Nelson’s disease, and thyrotropinoma; (2) diagnosis of probable/possible prolactinoma based on pre-operative PRL[200 ng/mL or immunohistochemistry PRL staining[10 % of the tumor area; (3) history of prior pituitary surgery; (4) apoplexy; (5) indeterminant immunohistochemistry or nondiagnostic pathology; (6) intake of any of the following medications: dopamine receptor antagonists, metoclopramide, tricyclic antidepressants, reserpine, methyl-dopa, or oral contraceptives; and (7) presence of renal failure, severe liver disease, or the diagnosis of polycystic ovary syndrome. After applying these exclusionary criteria, data from 98 patients with CNF pituitary adenomas were analyzed. Serum hormone measurements The pre-operative PRL value, measured by standard commercial two-site immunoassay, was recorded. For all macroadenomas with modestly elevated PRL levels, a PRL determination using the dilution technique was also
J Neurooncol
performed to rule-out a possible ‘‘Hook effect’’ [24–27]. Normal ranges of PRL levels varied among clinical laboratories, with the two standard deviation (SD) upper limit of normal (ULN) ranging from 13 to 30 ng/mL. Most PRL values (68 %) were obtained at UCLA laboratories. For the purposes of this study, a PRL level greater than the ULN was considered as hyperprolactinemia and below the lower limit of normal as hypoprolactinemia. Pre-operative IGF-1 values were available in 93 (95 %) of the patients. In 84 patients (86 %), pre-operative free T4 (fT4) values were available. In six (6 %) otherwise healthy patients without evidence of hyperalbuminemia, total T4 instead of fT4 had been ordered by the referring physicians. TSH, cortisol and ACTH values were obtained in all patients. To standardize the data sets for comparison across patients, we calculated z-scores for the pre-operative PRL, IGF-1, and T4 (free or total) values using the general formula: z = (value–mean)/SD. For PRL and IGF-1, z scores were based on the log values [28], whereas T4 had showed a normal Gaussian distribution [29]. With the exception of one patient (PRL 24.1, ULN 15.2 ng/mL), a PRL z score [3.0 corresponded to a PRL [35 ng/mL. IGF-1 z scores were based on age and gender controlled norms. The fT4, or total T4 if the fT4 was not available, was used to calculate a T4 z score. There were nine patients for whom a T4 z score could not be determined: three were on longterm thyroid replacement (etiology unclear), and no records could be obtained to verify ‘‘normal’’ thyroid function for the others.
•
•
•
Assessment of pituitary insufficiency Clinical pituitary insufficiency was routinely assessed as part of the routine pre-operative evaluation based on a combination of clinical symptomatology and laboratory evaluation. Provocative stimulation testing was not performed, and therefore we could not make definitive determinations of pituitary insufficiency in many cases. •
Central hypogonadism: Probable central hypogonadism was designated for normoprolactinemic males with total testosterone \LLN and post-menopausal females with nonelevated LH and FSH values. The clinical diagnosis of central hypogonadism in the setting of hyperprolactinemia required special consideration due to the challenge of distinguishing primary pituitary failure from hypothalamic suppression of GnRH by PRL. More than half of the 44 patients with a PRL z score [2.0 exhibited biochemical evidence of hypogonadism. Based on the report of Falaschi et al. [30] that revealed that testosterone levels remained normal with PRL elevations up to 50 ng/mL, we judged male
•
patients with low testosterone and modest hyperprolactinemia (2.0 [ PRL z B 3.0) to have probable central hypogonadism. For patients with PRL z [ 3.0 the etiology of hypogonadism was less clear, and therefore this axis was not included in the calculation of total number of axes with insufficiency. Thyroid axis: central hypothyroidism was determined based on T4 z score\-2.0 in combination with low or inappropriately normal TSH levels [20]. One of the six patients with total T4 values had a z score \-2.0. Replacement therapy was initiated pre-operatively for all patients diagnosed with central hypothyroidism. Hypothalamic–pituitary–adrenal axis: Probable central adrenal insufficiency was diagnosed in one patient with insufficiencies in three other axes and an early a.m. serum cortisol \5 mcg/dL. GH axis: According to Consensus Guidelines for the Diagnosis and Treatment of Adults with Growth Hormone Deficiency [31], a serum IGF-I below the normal range is suggestive of GH deficiency. Five of 93 (5 %) patients had two or more pituitary hormone deficiencies combined with a low serum IGF-I level. Eleven patients had an IGF-1 z score \-2.0 plus one other pituitary axis deficiency, and additional two with no other identified deficiency. Together, these 17 (18 %) patients were considered to have probable GHD. It has been suggested that, for patients with pituitary disease, the incidence of GHD increases from 45 % for patients with no other deficits to 100 % if three or more deficits are present [31]. Nine patients. with IGF-1 z score between -2.0 and 0.1 plus one other pituitary axis deficiency, were considered as possible GHD for the purposes of this study. Lactotroph insufficiency was a dependent variable in this study and therefore not included in the calculation of the total number of pituitary axes exhibiting insufficiency. Eight patients were missing either IGF1 or T4 measurements. To error on the conservative side, the total number of pituitary axes for each patients was calculated using only the determinations were designated as probable.
Assessment of lactotroph disinhibition We reasoned that if lactotroph disinhibition was present in all patients with macroadenomas regardless of pituitary insufficiency status, then PRL levels should drop postoperatively. To assess for iatrogenic surgically-induced pituitary damage, we compared pre- and post-operative IGF-1 levels. Post-operative PRL (measured [48 h, \18 months) and IGF-1 values ([4 weeks, \18 months) were available in 72 (73 %) and 71 (72 %) patients,
123
J Neurooncol Table 1 Study demographics, imaging analysis, and laboratory evaluation All
A: hypo-prolactinemia Pre-operative PRL z \ -2.0
B: normo-prolactinemia Pre-operative PRL -2.0 B z B 2.0
C: hyper-prolactinemia Pre-operative PRL z [ 2.0
p value B versus C
Demographics n
98
1
53
44
Age (years) Gender (F:M)
59 ± 12 38:60
71 M
60 ± 10 17:36
58 ± 15 21:23
n.s. n.s.
Neuroimaging Tumor size (mm)
28 ± 10
46
27 ± 11
29 ± 10
n.s.
SSE (mm)
12 ± 8
18
11 ± 8
13 ± 7
n.s.
PRL range (ng/dL)
1.0
4.9–27.9
17.8–200
IGF-1 z score
-1.7
-0.9 ± 1.6 (n = 49)
-1.5 ± 3.1 (n = 42)
n.s.
T4 z score
-3.5
-1.2 ± 1.5 (n = 49)
-1.1 ± 1.2 (n = 39)
n.s.
Hormone levels (Preop)
Chief complaint Visual defect
1
19 (36 %)
21 (47 %)
1 (2 %)
5 (11 %)
Hypogonadism (male)
3 (5 %)
3 (7 %)
Incidental (includes headache evaluation)
27 (51 %)
14 (32 %)
Other
3 (5 %)
1 (2 %)
Menstrual/galactorrhea
SSE suprasellar extension, PRL prolactin, IGF-1 insulin-like growth factor-1, T4 either free T4 or total T4, n.s. non significant
respectively. In 59 patients, both PRL and IGF-1 values were available post-operatively. When comparing the 72 patients with post-operative PRL levels with the 26 without, we found no statistical difference in pre-operative PRL z scores, IGF-1 z scores, any pituitary insufficiency score, or tumor size.
or Fisher’s Exact Test were used for categorical data analyses when individual cell sizes were low. Median confidence intervals were obtained by bootstrapping. Computations were made using R version 3.0.0; p = 0.05 was set as threshold for statistical significance before adjustment.
Neuroimaging analysis Results A one-dimensional estimate of tumor size was determined as the linear measurement of the maximum tumor dimension in the coronal, sagittal, or axial plane. We also measured suprasellar extension (SSE) of the tumor as a surrogate of stalk distortion/compression. Using the midline sagittal T1-weighted MRI image, SSE was estimated as the orthogonal distance above the line between the tuberculum sella and the dorsum sella projecting in the direction of the hypothalamus. Two board-certified neuroradiologists (N.S., M.L.), blinded to the other data, independently made the SSE measurements. Statistical analysis For continuous measures, conventional and robust analyses of variance with appropriate adjustments for post hoc comparisons, and conventional and robust correlation coefficients were computed. Chi square tests, Kruskal test,
123
Table 1 shows an overview of study demographics, imaging analysis, and laboratory evaluation. Mean tumor size was 28 ± 10 mm for the entire cohort. Consistent with prior studies, hyperprolactinemia was observed pre-operatively in the minority (44/98 or 45 %) of patients. Only one patient presented with hypoprolactinemia. As shown in the Table, simply comparing the heterogeneous groups with normoprolactinemia versus hyperprolactinemia revealed insignificant differences in age, gender distribution, tumor size indices, and IGF-1 and T4 z scores. In general, patients with hyperprolactinemia were more likely to present with menstrual irregularities and/or galactorrhea, and less commonly as incidental discoveries. The scatterplot distribution of tumor size versus PRL z score (Fig. 1) is similar in appearance to prior studies [8] and shows a lack of correlation between these variables (Pearson correlation coefficient of r = 0.03). The SSE
J Neurooncol
Fig. 1 Scatterplot revealing lack of correlation (r = 0.03) between pre-operative serum prolactin log z scores versus tumor size. The z scores between -2.0 and 2.0 correspond to standard clinical laboratory norms based on healthy populations. Pituitary insufficiency occurred less commonly among patients with tumor sizes \20 mm and/or prolactin z scores [3.0. Conversely, 87 % of patients with tumor size [20 mm and prolactin z scores \2.0 exhibited pituitary insufficiency
Fig. 2 Mean and standard deviation ranges of serum prolactin log z scores versus number of pituitary axes showing insufficiency. As the number of pituitary axis insufficiencies increased, the mean prolactin z score decreased
versus PRL z score correlation was similarly weak at r = 0.06.
of pituitary axis insufficiencies. As the number of pituitary axis insufficiencies increased, the mean PRL z score decreased (leveling out at two or more axis insufficiencies). Lumping all patients with pituitary insufficiency together, the mean PRL z score for patients exhibiting pituitary insufficiency was lower than for those patients with intact pituitary function (PRL z score 1.7 ± 1.4 vs. 2.8 ± 1.5, p = 0.01).
Pituitary insufficiency analysis
Post-operative hormonal analysis
Data points in Fig. 1 are labeled to reveal the distribution of patients with pituitary insufficiency (dark grey filled circles). With the exception of one patient, no patient with a tumor size B20 mm showed evidence of pituitary insufficiency. On the other hand, at least 87 % of patients (47/54) with tumor size [20 mm and normal prolactin values exhibited evidence of pituitary insufficiency in at least one axis. In Fig. 1, the distribution of hyperprolactinemic patients with and without pituitary insufficiency exhibited a variable pattern. Pituitary insufficiency was common for PRL z scores between 2.0 and 3.0 (15/23 or 65 %). For PRL z scores [3.0 the prevalence of pituitary insufficiency was 4 of 18 (22 %): with three patients having one (thyroid) axis deficiency and one patient with two axes (thyroid plus GHD). As noted above, hypogonadism was not included in the calculation of total number of axes for patients with PRL z scores[3.0. In this group, six patients demonstrated biochemical evidence of hypogonadism, four of which as the only possible deficiency. If these patients were included in the calculation of total number of axes, the percentage would increase to 44 % (8/18). Figure 2 shows the mean PRL z scores for those patients with tumor sizes [20 mm plotted against the total number
Of the 72 patients for whom post-operative PRL values were available, the post-operative PRL level was statistically lower (pre-op 25 ± 22 vs. post-op 9 ± 5 ng/mL, p \ 0.0001). In only four patients (6 %) was the postoperative PRL higher (average of only 2 ± 1 ng/mL). This is in contrast with post-operative IGF-1 levels (n = 71) which were statistically unchanged (pre-op 104 ± 47 vs. post-op 113 ± 61 ng/mL, p = 0.11). Among the 30 patients with tumor size [20 mm and normoprolactinemia, following surgery an additional four had IGF-1 z scores drop below -2.0 for a total of 13 (43 %). In contrast, patients with pre-operative PRL z scores [3.0, only one patient had a new subnormal IGF-1 value (total 1/15, 7 %) post-operatively. Gender-based analysis Of the 98 patients studied, only 11 (11 %) were premenopausal females. Five presented with oligomenorrhea or amenorrhea (three also with galactorrhea). The remaining were diagnosed incidentally, with no patient presenting with visual disturbances. Premenopausal women were more likely to present with relatively smaller tumors (21 ± 9 mm) as compared to either postmenopausal
123
J Neurooncol
females (28 ± 9 mm, p = 0.03) or males (30 ± 11 mm, p = 0.01). The pre-operative PRL z score was higher in premenopausal females compared to men (2.7 ± 1.6 vs. 1.6 ± 1.6, p = 0.03). No premenopausal female exhibited evidence of pituitary insufficiency based on biochemical measurements. For premenopausal females, the mean IGF-1 z score was -0.2 ± 0.8. The relative paucity of premenopausal females (11 %) among all patients studied surprised us. When reviewing the patients with CNF pituitary adenomas who were excluded based on the study criteria, 20 of 63 (32 %) were premenopausal females: 14 of which presented with menstrual irregularities and/or galactorrhea. Many were excluded due to birth control pill intake at time of presentation or history of prior pituitary surgery. The mean tumor size was 20 ± 10 mm, essentially identical to the comparative studied cohort. For patients with tumor size [20 mm, 59 % of postmenopausal females and 85 % of males exhibited pituitary insufficiency. Among males, the tumor was diagnosed in the majority (35/60 or 58 %) based on visual complaints referable to chiasmal compression and/or as a result of evaluation for hypogonadism. Half of postmenopausal females presented with chiasmal compression, with the remainder incidentalomas.
Discussion Our study demonstrates that for patients with large CNF macroadenomas (tumor size [20 mm) who present normoprolactinemia, the prevalence of pituitary insufficiency was at least 87 %. Given that we did not perform dynamic studies of GH secretion, this is likely an underestimate of pituitary insufficiency in this cohort. If general pituitary insufficiency is a predictor of lactotroph insufficiency, as it is for GHD, this would support our hypothesis that the normoprolactinemia could be a consequence of the inability of the remaining lactotrophs to produce hyperprolactinemia (at least PRL levels [35 ng/mL). We also predicted that patients presenting with hyperprolactinemia (classic ‘‘stalk effect’’) would be less likely to have pituitary insufficiency. As demonstrated in Fig. 2, our study results generally support this part of the central hypothesis. However, as revealed in Fig. 1, the group of patients with larger tumors (size [20 mm) and very mild hyperprolactinemia (PRL z score \3.0) appeared to cluster with the normoprolactinemic cohort based on pituitary insufficiency analysis. This may reflect that our operational definition of hyperprolactinemia was based on a statistical threshold (z [2.0) rather than a physiologic-based boundary. Another consideration is that this very mild
123
hyperprolactinemia zone is a ‘‘transit’’ area of either patients with smaller tumors who are entering the hyperprolactinemic zone (and hence less like to have pituitary insufficiency) or alternatively patients who previously had higher serum PRL levels but, at the time of discovery of the pituitary tumor, still maintained very mild hyperprolactinemia with presumed early lactotroph insufficiency. This supposition is supported by the fact that all of the patients in this very mild hyperprolactinemia zone (tumor size [20 mm) with evidence of pituitary insufficiency were males or postmenopausal females. Of the six patients with no evidence of pituitary insufficiency, three (50 %) were premenopausal females. A lower percentage (22 %) of patients with a tumor size [20 mm and a PRL z score [3.0 (PRL [* 35 ng/mL) demonstrated probable pituitary insufficiency (44 % if hypogonadism was included). One might wonder, if our central hypothesis is correct, why any of these patients exhibited pituitary insufficiency. Given that we were not directly measuring lactotroph insufficiency, one would anticipate a percentage of patients with 1–2 axis pituitary insufficiencies who still retain lactotroph function. We postulated that the physiological mechanism of lactotroph disinhibition should continue for patients with large macroadenomas: with or without hyperprolactinemia. Indeed, we found that PRL levels dropped post-operatively in virtually every patient with a macroadenoma, similar to the finding of Arafah [14]. Iatrogenic damage of normal lactotrophs as a result of surgery appears an unlikely explanation given the fact that post-operative IGF-1 levels as a whole were statistically unchanged. Our analysis did suggest, however, that the risk of worsened post-operative pituitary insufficiency was higher in patients with presumed lactotroph failure. There were several methodological weaknesses in our study. First, the lack of GH dynamic studies almost certainly led to the underestimation of the number of patients with GHD. With regard to influencing the interpretation of the results, however, this potential underestimation error would be most critical among patients with hyperprolactinemia in whom we postulated would be less likely to have pituitary insufficiency. As noted, patients with PRL z scores[3.0, 14 of 18 (78 %) patients had no evidence of pituitary insufficiency (with the caveat that 4 patients with only biochemical evidence of hypogonadism were not included for reasons stated previously). Clinically, none of these patients exhibited signs or symptoms that were suggestive of GHD. A second weakness related to the inability to directly assess the degree of lactotroph insufficiency. The use of pituitary insufficiency in other pituitary axes as a surrogate indicator is accepted for GHD, but nevertheless remains unproven for lactotroph function with the exception of hypoprolactinemia [20–23].
J Neurooncol
insufficiency in patients with normo- and even mild hyperprolactinemia (PRL \35 ng/mL). Acknowledgments Additional data acquisition performed by Saman Shabani, B.S. Medical Student, Stony Brook School of Medicine, NY, USA. No compensation received. Dr. Yong was supported in part by the Art of the Brain Fund and the Henry Singleton Brain Tumor Program. Conflict of interest of interest.
Fig. 3 Schematic plot depicting our concept of progressive tumor growth leading to phasic changes in prolactin levels. The traditional ‘‘stalk effect’’ representing the earlier hyperprolactinemia phase. If undiagnosed, prolactin levels in this phase peak and then recede as a result of progressive lactotroph insufficiency until normoprolactinemia is again encountered. Profound lactotroph failure manifests as hypoprolactinemia
Taken together, our data supports the concept that ‘‘stalk effect’’ may be a dynamic phenomenon (Fig. 3). A growing tumor first transitions into the hyperprolactinemia phase. Premenopausal females are more often discovered during this phase, whereas males and post-menopausal females may progress with continued tumor growth owing to minimal or no symptomatology arising from hyperprolactinemia. For these latter patients, at some point the degree of hyperprolactinemia begins to subside as lactotroph insufficiency ensues and the patient may eventually become normoprolactinemic again. This descent phase coincides with pituitary insufficiency of other pituitary axes. Normoprolactinemia is therefore bimodal: first among pre-stalk effect patients with smaller tumors, and then later in post-hyperprolactinemia patients (male and postmenopausal females) with large tumors and pituitary insufficiency. The terminal phase is hypoprolactinemia, typically associated with panhypopituitarism. As predicted, our study indeed demonstrated the mean tumor size for males and postmenopausal females was significantly larger than premenopausal females. Because this study was cross-sectional in nature, however, we lack longitudinal data for any given patient to support our phasic theory. Following a patient and documenting their descent into pituitary insufficiency while a tumor grows, however, would be ethically untenable. In summary, our study provides suggestive evidence that the ‘‘stalk effect’’ pathophysiology is not necessarily absent in patients with normoprolactinemia. The finding of normoprolactinemia in a patient with a large macroadenoma may be a consequence of lactotroph insufficiency. While the existence of pituitary insufficiency should be investigated in all patients with macroadenomas, our study calls for a more comprehensive, detailed assessment of pituitary
The authors declare that they have no conflict
References 1. Murai I, Garris PA, Ben-Jonathan N (1989) Time-dependent increase in plasma prolactin after pituitary stalk section: role of posterior pituitary dopamine. Endocrinology 124(5):2343–2349 2. Vaughan L, Carmel PW, Dyrenfurth I, Frantz AG, Antunes JL, Ferin M (1980) Section of the pituitary stalk in the rhesus monkey. I. Endocrine studies. Neuroendocrinology 30(2):70–75 3. Turkington RW, Underwood LE, Van Wyk JJ (1971) Elevated serum prolactin levels after pituitary-stalk section in man. N Engl J Med 285(13):707–710. doi:10.1056/NEJM197109232851302 4. Grossman A, Besser GM (1985) Prolactinomas. Br Med J (Clin Res Ed) 290(6463):182–184 5. Molitch ME (2001) Disorders of prolactin secretion. Endocrinol Metab Clin North Am 30(3):585–610 6. Randall RV, Scheithauer BW, Laws ER Jr, Abboud CF (1983) Pseudoprolactinomas. Trans Am Clin Climatol Assoc 94:114–121 7. Randall RV, Scheithauer BW, Laws ER Jr, Abbound CF, Ebersold MJ, Kao PC (1985) Pituitary adenomas associated with hyperprolactinemia: a clinical and immunohistochemical study of 97 patients operated on transsphenoidally. Mayo Clin Proc 60(11):753–762 8. Smith MV, Laws ER, Jr. (1994) Magnetic resonance imaging measurements of pituitary stalk compression and deviation in patients with nonprolactin-secreting intrasellar and parasellar tumors: lack of correlation with serum prolactin levels. Neurosurgery 34(5):834–839; discussion 839 9. Kruse A, Astrup J, Gyldensted C, Cold GE (1995) Hyperprolactinaemia in patients with pituitary adenomas. The pituitary stalk compression syndrome. Br J Neurosurg 9(4):453–457 10. Skinner DC (2009) Rethinking the stalk effect: a new hypothesis explaining suprasellar tumor-induced hyperprolactinemia. Med Hypotheses 72(3):309–310. doi:10.1016/j.mehy.2008.08.030 11. Karavitaki N, Thanabalasingham G, Shore HC, Trifanescu R, Ansorge O, Meston N, Turner HE, Wass JA (2006) Do the limits of serum prolactin in disconnection hyperprolactinaemia need redefinition? A study of 226 patients with histologically verified non-functioning pituitary macroadenoma. Clin Endocrinol (Oxf) 65(4):524–529 12. Arafah BM, Prunty D, Ybarra J, Hlavin ML, Selman WR (2000) The dominant role of increased intrasellar pressure in the pathogenesis of hypopituitarism, hyperprolactinemia, and headaches in patients with pituitary adenomas. J Clin Endocrinol Metab 85(5):1789–1793 13. Lees PD, Pickard JD (1987) Hyperprolactinemia, intrasellar pituitary tissue pressure, and the pituitary stalk compression syndrome. J Neurosurg 67(2):192–196. doi:10.3171/jns.1987.67. 2.0192 14. Arafah BM (1986) Reversible hypopituitarism in patients with large nonfunctioning pituitary adenomas. J Clin Endocrinol Metab 62(6):1173–1179
123
J Neurooncol 15. Arafah BM, Kailani SH, Nekl KE, Gold RS, Selman WR (1994) Immediate recovery of pituitary function after transsphenoidal resection of pituitary macroadenomas. J Clin Endocrinol Metab 79(2):348–354 16. Arafah BM, Nekl KE, Gold RS, Selman WR (1995) Dynamics of prolactin secretion in patients with hypopituitarism and pituitary macroadenomas. J Clin Endocrinol Metab 80(12):3507–3512 17. Kwan AY, Hartman ML (2007) IGF-I measurements in the diagnosis of adult growth hormone deficiency. Pituitary 10(2): 151–157. doi:10.1007/s11102-007-0028-8 18. Baum HB, Biller BM, Katznelson L, Oppenheim DS, Clemmons DR, Cannistraro KB, Schoenfeld DA, Best SA, Klibanski A (1996) Assessment of growth hormone (GH) secretion in men with adult-onset GH deficiency compared with that in normal men–a clinical research center study. J Clin Endocrinol Metab 81(1):84–92 19. Biller BM, Samuels MH, Zagar A, Cook DM, Arafah BM, Bonert V, Stavrou S, Kleinberg DL, Chipman JJ, Hartman ML (2002) Sensitivity and specificity of six tests for the diagnosis of adult GH deficiency. J Clin Endocrinol Metab 87(5):2067–2079 20. Ascoli P, Cavagnini F (2006) Hypopituitarism. Pituitary 9(4):335–342. doi:10.1007/s11102-006-0416-5 21. Chahal J, Schlechte J (2008) Hyperprolactinemia. Pituitary 11(2):141–146. doi:10.1007/s11102-008-0107-5 22. Toledano Y, Lubetsky A, Shimon I (2007) Acquired prolactin deficiency in patients with disorders of the hypothalamic-pituitary axis. J Endocrinol Invest 30(4):268–273 23. Mukherjee A, Murray RD, Columb B, Gleeson HK, Shalet SM (2003) Acquired prolactin deficiency indicates severe hypopituitarism in patients with disease of the hypothalamic-pituitary axis. Clin Endocrinol (Oxf) 59(6):743–748. doi:10.1046/j.1365-2265. 2003.01916.xCEN1916
123
24. Dahlmann N, Brensing KA, Klingmuller D, Bidlingmaier F (1990) ‘‘Hook-effect’’ in a patient with a gonadotropin-secreting tumor. Clin Chem 36(1):168 25. Delgrange E, de Hertogh R, Vankrieken L, Maiter D (1996) Potential hook effect in prolactin assay in patients with giant prolactinoma. Clin Endocrinol (Oxf) 45(4):506–507 26. Petakov MS, Damjanovic SS, Nikolic-Durovic MM, Dragojlovic ZL, Obradovic S, Gligorovic MS, Simic MZ, Popovic VP (1998) Pituitary adenomas secreting large amounts of prolactin may give false low values in immunoradiometric assays. The hook effect. J Endocrinol Invest 21(3):184–188 27. St-Jean E, Blain F, Comtois R (1996) High prolactin levels may be missed by immunoradiometric assay in patients with macroprolactinomas. Clin Endocrinol (Oxf) 44(3):305–309 28. Wang DY, de Stavola B, Bulbrook RD, Allen D, Kwa HG, Verstraeten AA, Moore JW, Fentiman IS, Chaudary M, Hayward JL (1987) The relationship between blood prolactin levels and risk of breast cancer in premenopausal women. Eur J Cancer Clin Oncol 23(10):1541–1548 29. Beckett GJ, Toft AD (2008) Thyroid dysfunction. In: Marshall WJ, Bangert SK (eds) Clinical biochemistry. Metabolic and clinical aspects. Churchill Livingstone, Philadelphia, p 406 30. Falaschi P, Frajese G, Sciarra F, Rocco A, Conti C (1978) Influence of hyperprolactinemia due to metoclopramide on gonadal function in men. Clin Endocrinol (Oxf) 8(5):427–433 31. Consensus guidelines for the diagnosis and treatment of adults with growth hormone deficiency: summary statement of the Growth Hormone Research Society Workshop on Adult Growth Hormone Deficiency (1998) J Clin Endocrinol Metab 83(2):379–381
CLINICAL STUDY
The pituitary stalk effect: is it a passing phenomenon? Marvin Bergsneider • Leili Mirsadraei • William H. Yong Noriko Salamon • Michael Linetsky • Marilene B. Wang • David L. McArthur • Anthony P. Heaney
•
Received: 25 November 2013 / Accepted: 21 January 2014 Springer Science+Business Media New York 2014
Abstract Most patients with large pituitary tumors do not exhibit hyperprolactinemia as a result of pituitary lactotroph disinhibition (stalk effect). Studies have demonstrated that increased intrasellar pressure is associated with both ‘‘stalk effect’’ hyperprolactinemia and pituitary insufficiency. Our primary hypothesis was that, despite continued disinhibition, lactotroph failure is responsible for normoprolactinemia in patients with large macroadenomas. As a corollary, we proposed that the hyperprolactinemia phase, which presumably would precede the insufficiency/ normoprolactinemic state, would more likely be discovered in premenopausal females and go unnoticed in males. Prospective, consecutive surgical series of 98 patients of clinically nonfunctional pituitary adenomas. Lactotroph insufficiency was inferred by the coexistence of insufficiency in another pituitary axis. The existence of preoperative lactotroph disinhibition was inferred based on comparison of pre- versus post-operative prolactin levels. 87 % of patients with tumor size [20 mm and
normoprolactinemia had pituitary insufficiency. Pre-operative prolactin in patients with pituitary insufficiency were lower than those with intact pituitary function. Prolactin levels dropped in nearly all patients, including patients with normoprolactinemia pre-operatively. Premenopausal women had smaller tumors and higher pre-operative prolactin levels compared to males. No premenopausal female exhibited evidence of pituitary insufficiency. Our study provides suggestive evidence that the ‘‘stalk effect’’ pathophysiology is the norm rather than the exception, and that the finding of normoprolactinemia in a patient with a large macroadenoma is likely a consequence of lactotroph insufficiency. In males, the hyperprolactinemia window is more likely to be missed clinically due to an absence of prolactin-related symptoms.
M. Bergsneider (&) Department of Neurosurgery, David Geffen School of Medicine at UCLA, Gonda 3357, BOX 951761, Los Angeles, CA 90095-1761, USA e-mail: [email protected]
M. B. Wang Department of Head and Neck Surgery, David Geffen School of Medicine at UCLA, 200 Medical Plaza, Ste 550, BOX 956959, Los Angeles, CA 90095-6959, USA
L. Mirsadraei W. H. Yong Division of Neuropathology, Department of Pathology, David Geffen School of Medicine at UCLA, 18-161A NPI, BOX 951732, Los Angeles, CA 90095-1732, USA N. Salamon M. Linetsky Division of Diagnostic Neuroradiology, Department of Radiology, David Geffen School of Medicine at UCLA, RRUMC 1621, BOX 957437, Los Angeles, CA 90095-7437, USA
Keywords Hyperprolactinemia Pituitary adenoma Dopamine Prolactin Pituitary insufficiency
D. L. McArthur Department of Neurosurgery, David Geffen School of Medicine at UCLA, 18-266 NPI, BOX 957039, Los Angeles, CA 90095-6901, USA A. P. Heaney Division of Endocrinology, Department of Medicine, David Geffen School of Medicine at UCLA, 24-130 WH, BOX 957073, Los Angeles, CA 90095-7073, USA
123
J Neurooncol
Introduction The ‘‘pituitary stalk effect’’ [1–3] is an elegant explanation describing the modest hyperprolactinemia associated with large pituitary tumors [4–7]. Compression of the pituitary stalk was proposed to diminish dopaminergic inhibition of normal lactotrophs, resulting in elevation of prolactin (PRL) levels. To date, however, it remains an enigma why the majority of patients with large clinically nonfunctional (CNF) pituitary tumors do not exhibit ‘‘stalk effect’’ hyperprolactinemia. The fact that tumor size poorly correlates with serum PRL levels [8, 9] suggested alternative mechanisms for the hyperprolactinemia [10], including pharmacological side-effects or presence of certain medical conditions [11]. Other studies have demonstrated that, rather than tumor size, hyperprolactinoma correlates well with increased intrasellar pressure [12, 13]. At some critical intrasellar pressure threshold, the portal flow to the pituitary was presumed to be disrupted, leading to the disinhibition of lactotroph PRL production. Interestingly, these same studies found that increased intrasellar pressure is also strongly associated with pituitary insufficiency [14–16]. The aim of our study was to gain an understanding of why the majority patients with large CNF macroadenomas do not exhibit hyperprolactinemia. Extrapolating from the above studies focused on intrasellar pressure [14–16], we hypothesized that the combination of pituitary insufficiency and portal flow disruption could create a state in which hyperprolactinemia can no longer be sustained. Specifically, a critical reduction in the number of lactotrophs, the manifestation of lactotroph insufficiency, resulted in normo- (or hypo-) prolactinemia even despite continued disinhibition by portal flow disruption. A challenge in studying this hypothesis was the inability to directly assess lactotroph insufficiency clinically. As a surrogate, we inferred that insufficiency in other pituitary axes would increase the likelihood of concomitant lactotroph failure. The validity of this approach has been demonstrated with growth hormone deficiency [17–19]. Moreover, absolute hypoprolactinemia is strongly associated with severe pituitary insufficiency [20–23]. Importantly, however, here we are proposing that normoprolactinemia can be the manifestation of lactotroph insufficiency in a patient for whom ‘‘stalk effect’’ hyperprolactinemia would be expected. This reasoning predicts that patients with large CNF macroadenomas and normal PRL levels have a higher prevalence of pituitary insufficiency. Conversely, patients with ‘‘stalk effect’’ hyperprolactinemia are less likely to exhibit pituitary insufficiency. By including lactrotroph failure into the concept of ‘‘stalk effect,’’ we are proposing a multiple state scenario that includes normal (or low) PRL levels in addition to the customary hyperprolactinemia ‘‘stalk effect’’ state. Logically,
123
lactotroph insufficiency (and hence normoprolactinemia) should occur later in the clinical course, whereas one would expect hyperprolactinemia to occur earlier, being more common with relatively ‘‘smaller’’ CNF macroadenomas. We further postulated that a patient presenting with a combination of normoprolactinemia, a large CNF macroadenoma, and pituitary insufficiency, would have likely transitioned through a hyperprolactinemic phase while the tumor was smaller in size. This would raise the question of why the hyperprolactinemic phase would have gone unnoticed. From a clinical standpoint, we reasoned that males and post-menopausal females would have been that group as they would have been less likely to be diagnosed with hyperprolactinemia compared with premenopausal females. Conversely, premenopausal females would more likely present with hyperprolactinemia-related symptoms, and therefore be diagnosed with relatively smaller clinically nonfunctional macroadenomas as they enter the ‘‘stalk effect’’ phase.
Clinical materials and methods Over the 4-year period from 2008 to 2012, 312 consecutive patients undergoing endoscopic, endonasal resection of suspected pituitary adenomas by the senior author (M.B.). An institutional review board approved informed consent was obtained for the prospective collection of clinical, imaging measurements, and quantitative immunohistochemistry data. Exclusion criteria (1) diagnosis of acromegaly, Cushing’s disease, Nelson’s disease, and thyrotropinoma; (2) diagnosis of probable/possible prolactinoma based on pre-operative PRL[200 ng/mL or immunohistochemistry PRL staining[10 % of the tumor area; (3) history of prior pituitary surgery; (4) apoplexy; (5) indeterminant immunohistochemistry or nondiagnostic pathology; (6) intake of any of the following medications: dopamine receptor antagonists, metoclopramide, tricyclic antidepressants, reserpine, methyl-dopa, or oral contraceptives; and (7) presence of renal failure, severe liver disease, or the diagnosis of polycystic ovary syndrome. After applying these exclusionary criteria, data from 98 patients with CNF pituitary adenomas were analyzed. Serum hormone measurements The pre-operative PRL value, measured by standard commercial two-site immunoassay, was recorded. For all macroadenomas with modestly elevated PRL levels, a PRL determination using the dilution technique was also
J Neurooncol
performed to rule-out a possible ‘‘Hook effect’’ [24–27]. Normal ranges of PRL levels varied among clinical laboratories, with the two standard deviation (SD) upper limit of normal (ULN) ranging from 13 to 30 ng/mL. Most PRL values (68 %) were obtained at UCLA laboratories. For the purposes of this study, a PRL level greater than the ULN was considered as hyperprolactinemia and below the lower limit of normal as hypoprolactinemia. Pre-operative IGF-1 values were available in 93 (95 %) of the patients. In 84 patients (86 %), pre-operative free T4 (fT4) values were available. In six (6 %) otherwise healthy patients without evidence of hyperalbuminemia, total T4 instead of fT4 had been ordered by the referring physicians. TSH, cortisol and ACTH values were obtained in all patients. To standardize the data sets for comparison across patients, we calculated z-scores for the pre-operative PRL, IGF-1, and T4 (free or total) values using the general formula: z = (value–mean)/SD. For PRL and IGF-1, z scores were based on the log values [28], whereas T4 had showed a normal Gaussian distribution [29]. With the exception of one patient (PRL 24.1, ULN 15.2 ng/mL), a PRL z score [3.0 corresponded to a PRL [35 ng/mL. IGF-1 z scores were based on age and gender controlled norms. The fT4, or total T4 if the fT4 was not available, was used to calculate a T4 z score. There were nine patients for whom a T4 z score could not be determined: three were on longterm thyroid replacement (etiology unclear), and no records could be obtained to verify ‘‘normal’’ thyroid function for the others.
•
•
•
Assessment of pituitary insufficiency Clinical pituitary insufficiency was routinely assessed as part of the routine pre-operative evaluation based on a combination of clinical symptomatology and laboratory evaluation. Provocative stimulation testing was not performed, and therefore we could not make definitive determinations of pituitary insufficiency in many cases. •
Central hypogonadism: Probable central hypogonadism was designated for normoprolactinemic males with total testosterone \LLN and post-menopausal females with nonelevated LH and FSH values. The clinical diagnosis of central hypogonadism in the setting of hyperprolactinemia required special consideration due to the challenge of distinguishing primary pituitary failure from hypothalamic suppression of GnRH by PRL. More than half of the 44 patients with a PRL z score [2.0 exhibited biochemical evidence of hypogonadism. Based on the report of Falaschi et al. [30] that revealed that testosterone levels remained normal with PRL elevations up to 50 ng/mL, we judged male
•
patients with low testosterone and modest hyperprolactinemia (2.0 [ PRL z B 3.0) to have probable central hypogonadism. For patients with PRL z [ 3.0 the etiology of hypogonadism was less clear, and therefore this axis was not included in the calculation of total number of axes with insufficiency. Thyroid axis: central hypothyroidism was determined based on T4 z score\-2.0 in combination with low or inappropriately normal TSH levels [20]. One of the six patients with total T4 values had a z score \-2.0. Replacement therapy was initiated pre-operatively for all patients diagnosed with central hypothyroidism. Hypothalamic–pituitary–adrenal axis: Probable central adrenal insufficiency was diagnosed in one patient with insufficiencies in three other axes and an early a.m. serum cortisol \5 mcg/dL. GH axis: According to Consensus Guidelines for the Diagnosis and Treatment of Adults with Growth Hormone Deficiency [31], a serum IGF-I below the normal range is suggestive of GH deficiency. Five of 93 (5 %) patients had two or more pituitary hormone deficiencies combined with a low serum IGF-I level. Eleven patients had an IGF-1 z score \-2.0 plus one other pituitary axis deficiency, and additional two with no other identified deficiency. Together, these 17 (18 %) patients were considered to have probable GHD. It has been suggested that, for patients with pituitary disease, the incidence of GHD increases from 45 % for patients with no other deficits to 100 % if three or more deficits are present [31]. Nine patients. with IGF-1 z score between -2.0 and 0.1 plus one other pituitary axis deficiency, were considered as possible GHD for the purposes of this study. Lactotroph insufficiency was a dependent variable in this study and therefore not included in the calculation of the total number of pituitary axes exhibiting insufficiency. Eight patients were missing either IGF1 or T4 measurements. To error on the conservative side, the total number of pituitary axes for each patients was calculated using only the determinations were designated as probable.
Assessment of lactotroph disinhibition We reasoned that if lactotroph disinhibition was present in all patients with macroadenomas regardless of pituitary insufficiency status, then PRL levels should drop postoperatively. To assess for iatrogenic surgically-induced pituitary damage, we compared pre- and post-operative IGF-1 levels. Post-operative PRL (measured [48 h, \18 months) and IGF-1 values ([4 weeks, \18 months) were available in 72 (73 %) and 71 (72 %) patients,
123
J Neurooncol Table 1 Study demographics, imaging analysis, and laboratory evaluation All
A: hypo-prolactinemia Pre-operative PRL z \ -2.0
B: normo-prolactinemia Pre-operative PRL -2.0 B z B 2.0
C: hyper-prolactinemia Pre-operative PRL z [ 2.0
p value B versus C
Demographics n
98
1
53
44
Age (years) Gender (F:M)
59 ± 12 38:60
71 M
60 ± 10 17:36
58 ± 15 21:23
n.s. n.s.
Neuroimaging Tumor size (mm)
28 ± 10
46
27 ± 11
29 ± 10
n.s.
SSE (mm)
12 ± 8
18
11 ± 8
13 ± 7
n.s.
PRL range (ng/dL)
1.0
4.9–27.9
17.8–200
IGF-1 z score
-1.7
-0.9 ± 1.6 (n = 49)
-1.5 ± 3.1 (n = 42)
n.s.
T4 z score
-3.5
-1.2 ± 1.5 (n = 49)
-1.1 ± 1.2 (n = 39)
n.s.
Hormone levels (Preop)
Chief complaint Visual defect
1
19 (36 %)
21 (47 %)
1 (2 %)
5 (11 %)
Hypogonadism (male)
3 (5 %)
3 (7 %)
Incidental (includes headache evaluation)
27 (51 %)
14 (32 %)
Other
3 (5 %)
1 (2 %)
Menstrual/galactorrhea
SSE suprasellar extension, PRL prolactin, IGF-1 insulin-like growth factor-1, T4 either free T4 or total T4, n.s. non significant
respectively. In 59 patients, both PRL and IGF-1 values were available post-operatively. When comparing the 72 patients with post-operative PRL levels with the 26 without, we found no statistical difference in pre-operative PRL z scores, IGF-1 z scores, any pituitary insufficiency score, or tumor size.
or Fisher’s Exact Test were used for categorical data analyses when individual cell sizes were low. Median confidence intervals were obtained by bootstrapping. Computations were made using R version 3.0.0; p = 0.05 was set as threshold for statistical significance before adjustment.
Neuroimaging analysis Results A one-dimensional estimate of tumor size was determined as the linear measurement of the maximum tumor dimension in the coronal, sagittal, or axial plane. We also measured suprasellar extension (SSE) of the tumor as a surrogate of stalk distortion/compression. Using the midline sagittal T1-weighted MRI image, SSE was estimated as the orthogonal distance above the line between the tuberculum sella and the dorsum sella projecting in the direction of the hypothalamus. Two board-certified neuroradiologists (N.S., M.L.), blinded to the other data, independently made the SSE measurements. Statistical analysis For continuous measures, conventional and robust analyses of variance with appropriate adjustments for post hoc comparisons, and conventional and robust correlation coefficients were computed. Chi square tests, Kruskal test,
123
Table 1 shows an overview of study demographics, imaging analysis, and laboratory evaluation. Mean tumor size was 28 ± 10 mm for the entire cohort. Consistent with prior studies, hyperprolactinemia was observed pre-operatively in the minority (44/98 or 45 %) of patients. Only one patient presented with hypoprolactinemia. As shown in the Table, simply comparing the heterogeneous groups with normoprolactinemia versus hyperprolactinemia revealed insignificant differences in age, gender distribution, tumor size indices, and IGF-1 and T4 z scores. In general, patients with hyperprolactinemia were more likely to present with menstrual irregularities and/or galactorrhea, and less commonly as incidental discoveries. The scatterplot distribution of tumor size versus PRL z score (Fig. 1) is similar in appearance to prior studies [8] and shows a lack of correlation between these variables (Pearson correlation coefficient of r = 0.03). The SSE
J Neurooncol
Fig. 1 Scatterplot revealing lack of correlation (r = 0.03) between pre-operative serum prolactin log z scores versus tumor size. The z scores between -2.0 and 2.0 correspond to standard clinical laboratory norms based on healthy populations. Pituitary insufficiency occurred less commonly among patients with tumor sizes \20 mm and/or prolactin z scores [3.0. Conversely, 87 % of patients with tumor size [20 mm and prolactin z scores \2.0 exhibited pituitary insufficiency
Fig. 2 Mean and standard deviation ranges of serum prolactin log z scores versus number of pituitary axes showing insufficiency. As the number of pituitary axis insufficiencies increased, the mean prolactin z score decreased
versus PRL z score correlation was similarly weak at r = 0.06.
of pituitary axis insufficiencies. As the number of pituitary axis insufficiencies increased, the mean PRL z score decreased (leveling out at two or more axis insufficiencies). Lumping all patients with pituitary insufficiency together, the mean PRL z score for patients exhibiting pituitary insufficiency was lower than for those patients with intact pituitary function (PRL z score 1.7 ± 1.4 vs. 2.8 ± 1.5, p = 0.01).
Pituitary insufficiency analysis
Post-operative hormonal analysis
Data points in Fig. 1 are labeled to reveal the distribution of patients with pituitary insufficiency (dark grey filled circles). With the exception of one patient, no patient with a tumor size B20 mm showed evidence of pituitary insufficiency. On the other hand, at least 87 % of patients (47/54) with tumor size [20 mm and normal prolactin values exhibited evidence of pituitary insufficiency in at least one axis. In Fig. 1, the distribution of hyperprolactinemic patients with and without pituitary insufficiency exhibited a variable pattern. Pituitary insufficiency was common for PRL z scores between 2.0 and 3.0 (15/23 or 65 %). For PRL z scores [3.0 the prevalence of pituitary insufficiency was 4 of 18 (22 %): with three patients having one (thyroid) axis deficiency and one patient with two axes (thyroid plus GHD). As noted above, hypogonadism was not included in the calculation of total number of axes for patients with PRL z scores[3.0. In this group, six patients demonstrated biochemical evidence of hypogonadism, four of which as the only possible deficiency. If these patients were included in the calculation of total number of axes, the percentage would increase to 44 % (8/18). Figure 2 shows the mean PRL z scores for those patients with tumor sizes [20 mm plotted against the total number
Of the 72 patients for whom post-operative PRL values were available, the post-operative PRL level was statistically lower (pre-op 25 ± 22 vs. post-op 9 ± 5 ng/mL, p \ 0.0001). In only four patients (6 %) was the postoperative PRL higher (average of only 2 ± 1 ng/mL). This is in contrast with post-operative IGF-1 levels (n = 71) which were statistically unchanged (pre-op 104 ± 47 vs. post-op 113 ± 61 ng/mL, p = 0.11). Among the 30 patients with tumor size [20 mm and normoprolactinemia, following surgery an additional four had IGF-1 z scores drop below -2.0 for a total of 13 (43 %). In contrast, patients with pre-operative PRL z scores [3.0, only one patient had a new subnormal IGF-1 value (total 1/15, 7 %) post-operatively. Gender-based analysis Of the 98 patients studied, only 11 (11 %) were premenopausal females. Five presented with oligomenorrhea or amenorrhea (three also with galactorrhea). The remaining were diagnosed incidentally, with no patient presenting with visual disturbances. Premenopausal women were more likely to present with relatively smaller tumors (21 ± 9 mm) as compared to either postmenopausal
123
J Neurooncol
females (28 ± 9 mm, p = 0.03) or males (30 ± 11 mm, p = 0.01). The pre-operative PRL z score was higher in premenopausal females compared to men (2.7 ± 1.6 vs. 1.6 ± 1.6, p = 0.03). No premenopausal female exhibited evidence of pituitary insufficiency based on biochemical measurements. For premenopausal females, the mean IGF-1 z score was -0.2 ± 0.8. The relative paucity of premenopausal females (11 %) among all patients studied surprised us. When reviewing the patients with CNF pituitary adenomas who were excluded based on the study criteria, 20 of 63 (32 %) were premenopausal females: 14 of which presented with menstrual irregularities and/or galactorrhea. Many were excluded due to birth control pill intake at time of presentation or history of prior pituitary surgery. The mean tumor size was 20 ± 10 mm, essentially identical to the comparative studied cohort. For patients with tumor size [20 mm, 59 % of postmenopausal females and 85 % of males exhibited pituitary insufficiency. Among males, the tumor was diagnosed in the majority (35/60 or 58 %) based on visual complaints referable to chiasmal compression and/or as a result of evaluation for hypogonadism. Half of postmenopausal females presented with chiasmal compression, with the remainder incidentalomas.
Discussion Our study demonstrates that for patients with large CNF macroadenomas (tumor size [20 mm) who present normoprolactinemia, the prevalence of pituitary insufficiency was at least 87 %. Given that we did not perform dynamic studies of GH secretion, this is likely an underestimate of pituitary insufficiency in this cohort. If general pituitary insufficiency is a predictor of lactotroph insufficiency, as it is for GHD, this would support our hypothesis that the normoprolactinemia could be a consequence of the inability of the remaining lactotrophs to produce hyperprolactinemia (at least PRL levels [35 ng/mL). We also predicted that patients presenting with hyperprolactinemia (classic ‘‘stalk effect’’) would be less likely to have pituitary insufficiency. As demonstrated in Fig. 2, our study results generally support this part of the central hypothesis. However, as revealed in Fig. 1, the group of patients with larger tumors (size [20 mm) and very mild hyperprolactinemia (PRL z score \3.0) appeared to cluster with the normoprolactinemic cohort based on pituitary insufficiency analysis. This may reflect that our operational definition of hyperprolactinemia was based on a statistical threshold (z [2.0) rather than a physiologic-based boundary. Another consideration is that this very mild
123
hyperprolactinemia zone is a ‘‘transit’’ area of either patients with smaller tumors who are entering the hyperprolactinemic zone (and hence less like to have pituitary insufficiency) or alternatively patients who previously had higher serum PRL levels but, at the time of discovery of the pituitary tumor, still maintained very mild hyperprolactinemia with presumed early lactotroph insufficiency. This supposition is supported by the fact that all of the patients in this very mild hyperprolactinemia zone (tumor size [20 mm) with evidence of pituitary insufficiency were males or postmenopausal females. Of the six patients with no evidence of pituitary insufficiency, three (50 %) were premenopausal females. A lower percentage (22 %) of patients with a tumor size [20 mm and a PRL z score [3.0 (PRL [* 35 ng/mL) demonstrated probable pituitary insufficiency (44 % if hypogonadism was included). One might wonder, if our central hypothesis is correct, why any of these patients exhibited pituitary insufficiency. Given that we were not directly measuring lactotroph insufficiency, one would anticipate a percentage of patients with 1–2 axis pituitary insufficiencies who still retain lactotroph function. We postulated that the physiological mechanism of lactotroph disinhibition should continue for patients with large macroadenomas: with or without hyperprolactinemia. Indeed, we found that PRL levels dropped post-operatively in virtually every patient with a macroadenoma, similar to the finding of Arafah [14]. Iatrogenic damage of normal lactotrophs as a result of surgery appears an unlikely explanation given the fact that post-operative IGF-1 levels as a whole were statistically unchanged. Our analysis did suggest, however, that the risk of worsened post-operative pituitary insufficiency was higher in patients with presumed lactotroph failure. There were several methodological weaknesses in our study. First, the lack of GH dynamic studies almost certainly led to the underestimation of the number of patients with GHD. With regard to influencing the interpretation of the results, however, this potential underestimation error would be most critical among patients with hyperprolactinemia in whom we postulated would be less likely to have pituitary insufficiency. As noted, patients with PRL z scores[3.0, 14 of 18 (78 %) patients had no evidence of pituitary insufficiency (with the caveat that 4 patients with only biochemical evidence of hypogonadism were not included for reasons stated previously). Clinically, none of these patients exhibited signs or symptoms that were suggestive of GHD. A second weakness related to the inability to directly assess the degree of lactotroph insufficiency. The use of pituitary insufficiency in other pituitary axes as a surrogate indicator is accepted for GHD, but nevertheless remains unproven for lactotroph function with the exception of hypoprolactinemia [20–23].
J Neurooncol
insufficiency in patients with normo- and even mild hyperprolactinemia (PRL \35 ng/mL). Acknowledgments Additional data acquisition performed by Saman Shabani, B.S. Medical Student, Stony Brook School of Medicine, NY, USA. No compensation received. Dr. Yong was supported in part by the Art of the Brain Fund and the Henry Singleton Brain Tumor Program. Conflict of interest of interest.
Fig. 3 Schematic plot depicting our concept of progressive tumor growth leading to phasic changes in prolactin levels. The traditional ‘‘stalk effect’’ representing the earlier hyperprolactinemia phase. If undiagnosed, prolactin levels in this phase peak and then recede as a result of progressive lactotroph insufficiency until normoprolactinemia is again encountered. Profound lactotroph failure manifests as hypoprolactinemia
Taken together, our data supports the concept that ‘‘stalk effect’’ may be a dynamic phenomenon (Fig. 3). A growing tumor first transitions into the hyperprolactinemia phase. Premenopausal females are more often discovered during this phase, whereas males and post-menopausal females may progress with continued tumor growth owing to minimal or no symptomatology arising from hyperprolactinemia. For these latter patients, at some point the degree of hyperprolactinemia begins to subside as lactotroph insufficiency ensues and the patient may eventually become normoprolactinemic again. This descent phase coincides with pituitary insufficiency of other pituitary axes. Normoprolactinemia is therefore bimodal: first among pre-stalk effect patients with smaller tumors, and then later in post-hyperprolactinemia patients (male and postmenopausal females) with large tumors and pituitary insufficiency. The terminal phase is hypoprolactinemia, typically associated with panhypopituitarism. As predicted, our study indeed demonstrated the mean tumor size for males and postmenopausal females was significantly larger than premenopausal females. Because this study was cross-sectional in nature, however, we lack longitudinal data for any given patient to support our phasic theory. Following a patient and documenting their descent into pituitary insufficiency while a tumor grows, however, would be ethically untenable. In summary, our study provides suggestive evidence that the ‘‘stalk effect’’ pathophysiology is not necessarily absent in patients with normoprolactinemia. The finding of normoprolactinemia in a patient with a large macroadenoma may be a consequence of lactotroph insufficiency. While the existence of pituitary insufficiency should be investigated in all patients with macroadenomas, our study calls for a more comprehensive, detailed assessment of pituitary
The authors declare that they have no conflict
References 1. Murai I, Garris PA, Ben-Jonathan N (1989) Time-dependent increase in plasma prolactin after pituitary stalk section: role of posterior pituitary dopamine. Endocrinology 124(5):2343–2349 2. Vaughan L, Carmel PW, Dyrenfurth I, Frantz AG, Antunes JL, Ferin M (1980) Section of the pituitary stalk in the rhesus monkey. I. Endocrine studies. Neuroendocrinology 30(2):70–75 3. Turkington RW, Underwood LE, Van Wyk JJ (1971) Elevated serum prolactin levels after pituitary-stalk section in man. N Engl J Med 285(13):707–710. doi:10.1056/NEJM197109232851302 4. Grossman A, Besser GM (1985) Prolactinomas. Br Med J (Clin Res Ed) 290(6463):182–184 5. Molitch ME (2001) Disorders of prolactin secretion. Endocrinol Metab Clin North Am 30(3):585–610 6. Randall RV, Scheithauer BW, Laws ER Jr, Abboud CF (1983) Pseudoprolactinomas. Trans Am Clin Climatol Assoc 94:114–121 7. Randall RV, Scheithauer BW, Laws ER Jr, Abbound CF, Ebersold MJ, Kao PC (1985) Pituitary adenomas associated with hyperprolactinemia: a clinical and immunohistochemical study of 97 patients operated on transsphenoidally. Mayo Clin Proc 60(11):753–762 8. Smith MV, Laws ER, Jr. (1994) Magnetic resonance imaging measurements of pituitary stalk compression and deviation in patients with nonprolactin-secreting intrasellar and parasellar tumors: lack of correlation with serum prolactin levels. Neurosurgery 34(5):834–839; discussion 839 9. Kruse A, Astrup J, Gyldensted C, Cold GE (1995) Hyperprolactinaemia in patients with pituitary adenomas. The pituitary stalk compression syndrome. Br J Neurosurg 9(4):453–457 10. Skinner DC (2009) Rethinking the stalk effect: a new hypothesis explaining suprasellar tumor-induced hyperprolactinemia. Med Hypotheses 72(3):309–310. doi:10.1016/j.mehy.2008.08.030 11. Karavitaki N, Thanabalasingham G, Shore HC, Trifanescu R, Ansorge O, Meston N, Turner HE, Wass JA (2006) Do the limits of serum prolactin in disconnection hyperprolactinaemia need redefinition? A study of 226 patients with histologically verified non-functioning pituitary macroadenoma. Clin Endocrinol (Oxf) 65(4):524–529 12. Arafah BM, Prunty D, Ybarra J, Hlavin ML, Selman WR (2000) The dominant role of increased intrasellar pressure in the pathogenesis of hypopituitarism, hyperprolactinemia, and headaches in patients with pituitary adenomas. J Clin Endocrinol Metab 85(5):1789–1793 13. Lees PD, Pickard JD (1987) Hyperprolactinemia, intrasellar pituitary tissue pressure, and the pituitary stalk compression syndrome. J Neurosurg 67(2):192–196. doi:10.3171/jns.1987.67. 2.0192 14. Arafah BM (1986) Reversible hypopituitarism in patients with large nonfunctioning pituitary adenomas. J Clin Endocrinol Metab 62(6):1173–1179
123
J Neurooncol 15. Arafah BM, Kailani SH, Nekl KE, Gold RS, Selman WR (1994) Immediate recovery of pituitary function after transsphenoidal resection of pituitary macroadenomas. J Clin Endocrinol Metab 79(2):348–354 16. Arafah BM, Nekl KE, Gold RS, Selman WR (1995) Dynamics of prolactin secretion in patients with hypopituitarism and pituitary macroadenomas. J Clin Endocrinol Metab 80(12):3507–3512 17. Kwan AY, Hartman ML (2007) IGF-I measurements in the diagnosis of adult growth hormone deficiency. Pituitary 10(2): 151–157. doi:10.1007/s11102-007-0028-8 18. Baum HB, Biller BM, Katznelson L, Oppenheim DS, Clemmons DR, Cannistraro KB, Schoenfeld DA, Best SA, Klibanski A (1996) Assessment of growth hormone (GH) secretion in men with adult-onset GH deficiency compared with that in normal men–a clinical research center study. J Clin Endocrinol Metab 81(1):84–92 19. Biller BM, Samuels MH, Zagar A, Cook DM, Arafah BM, Bonert V, Stavrou S, Kleinberg DL, Chipman JJ, Hartman ML (2002) Sensitivity and specificity of six tests for the diagnosis of adult GH deficiency. J Clin Endocrinol Metab 87(5):2067–2079 20. Ascoli P, Cavagnini F (2006) Hypopituitarism. Pituitary 9(4):335–342. doi:10.1007/s11102-006-0416-5 21. Chahal J, Schlechte J (2008) Hyperprolactinemia. Pituitary 11(2):141–146. doi:10.1007/s11102-008-0107-5 22. Toledano Y, Lubetsky A, Shimon I (2007) Acquired prolactin deficiency in patients with disorders of the hypothalamic-pituitary axis. J Endocrinol Invest 30(4):268–273 23. Mukherjee A, Murray RD, Columb B, Gleeson HK, Shalet SM (2003) Acquired prolactin deficiency indicates severe hypopituitarism in patients with disease of the hypothalamic-pituitary axis. Clin Endocrinol (Oxf) 59(6):743–748. doi:10.1046/j.1365-2265. 2003.01916.xCEN1916
123
24. Dahlmann N, Brensing KA, Klingmuller D, Bidlingmaier F (1990) ‘‘Hook-effect’’ in a patient with a gonadotropin-secreting tumor. Clin Chem 36(1):168 25. Delgrange E, de Hertogh R, Vankrieken L, Maiter D (1996) Potential hook effect in prolactin assay in patients with giant prolactinoma. Clin Endocrinol (Oxf) 45(4):506–507 26. Petakov MS, Damjanovic SS, Nikolic-Durovic MM, Dragojlovic ZL, Obradovic S, Gligorovic MS, Simic MZ, Popovic VP (1998) Pituitary adenomas secreting large amounts of prolactin may give false low values in immunoradiometric assays. The hook effect. J Endocrinol Invest 21(3):184–188 27. St-Jean E, Blain F, Comtois R (1996) High prolactin levels may be missed by immunoradiometric assay in patients with macroprolactinomas. Clin Endocrinol (Oxf) 44(3):305–309 28. Wang DY, de Stavola B, Bulbrook RD, Allen D, Kwa HG, Verstraeten AA, Moore JW, Fentiman IS, Chaudary M, Hayward JL (1987) The relationship between blood prolactin levels and risk of breast cancer in premenopausal women. Eur J Cancer Clin Oncol 23(10):1541–1548 29. Beckett GJ, Toft AD (2008) Thyroid dysfunction. In: Marshall WJ, Bangert SK (eds) Clinical biochemistry. Metabolic and clinical aspects. Churchill Livingstone, Philadelphia, p 406 30. Falaschi P, Frajese G, Sciarra F, Rocco A, Conti C (1978) Influence of hyperprolactinemia due to metoclopramide on gonadal function in men. Clin Endocrinol (Oxf) 8(5):427–433 31. Consensus guidelines for the diagnosis and treatment of adults with growth hormone deficiency: summary statement of the Growth Hormone Research Society Workshop on Adult Growth Hormone Deficiency (1998) J Clin Endocrinol Metab 83(2):379–381
