Journal of Clinical Endocrinology and Metabolism Copyright © 1978 by The Endocrine Society

Vol. 46, No. 4 Printed in U.S.A.

Prolactin in Human Cerebrospinal Fluid* JOHANNA ASSIES,f ANDRIES P. M. SCHELLEKENS, AND JAN L. TOUBER Division of Endocrinology, Department of Medicine, University Hospital "Wilhelmina Amsterdam, The Netherlands ABSTRACT. PRL was measured radioimmunologically in plasma and cerebrospinal fluid (CSF) samples obtained simultaneously in 31 patients with various neurological or infectious, but non-endocrine diseases (group A), 12 patients (7 pregnant women and 5 newborns) with physiological hyperprolactinemia (group B), 10 psychiatric patients with pharmacologically induced hyperprolactinemia (group C), 12 normoprolactinemic patients with pituitary adenoma and suprasellar extension (SSE) (group D), and 14 hyperprolactinemic patients with pituitary adenoma with and without SSE (group E). Plasma PRL and CSF PRL concentrations (ng/ml, mean and range in brackets) of the various groups were: group A, 6.2 (1.3-14.5) and 1.3 (0.6-4.7); group B, 85.2 (31-200) and 13.2 (3-28); group C, 54.3 (3.5-160) and 6.5 (0.7-18); group D, 17.2 (5.4-30) and 9.7 (2.7-34); and group E, 2,529 (115-10,000) and 1,449 (6-13,000). The


HE PRESENCE of adenohypophyseal hormones in the cerebrospinal fluid (CSF) of patients with pituitary tumors and suprasellar extension (SSE) of the pituitary mass has recently been reported for GH in acromegalic patients, ACTH in Nelson's syndrome, PRL in patients with PRL-producing tumors, and TSH, LH, and FSH in the CSF of patients with chromophobe adenoma (1-9). With the exception of ACTH and /3-MSH (10, 11), these hormones were not found in the CSF of individuals without endocrine disease, or their concentration was so low that the results of assays were difficult to interpret (1, 2, 3, 5, 9). Indeed, it has been claimed that in the normal individual the bigger adenohypophyseal hormones do not cross the blood/CSF barrier and that detectable levels of these hormones in the CSF are indicative of supraReceived April 30, 1976. * This work was supported by the Netherlands Foundation for Medical Research (FUNGO). This work was presented, in part, at the 58th Annual Meeting of The Endocrine Society, San Francisco, June 1976 (Abstract 116). f To whom requests for reprints should be addressed.


plasma to CSF concentration ratios (mean and range in brackets) were: group A, 5.2 (1.4-13.0); group B, 7.0 (2.9-10.3); group C, 7.3 (3.9-11.3); group D, 2.6 (0.9-7.1); and group E, 10.9 (0.2-34.9). The ratio was > 3 in 87% of the non-tumor patients; in 42% of the tumor patients the ratio was < 3. The correlation between plasma and CSF PRL levels of all 53 subjects without a pituitary tumor (groups A, B, and C) was positive (r = 0.9097; P = 0.00001); in the 26 tumor patients (groups D and E) the correlation was also postive (r = 0.7141; P = 0.00002). These results indicate that 1) PRL is a normal constituent of CSF, 2) the CSF PRL level is a function of the plasma level, 3) detectable, or even high, CSF PRL levels per se are not indicative of the presence of a pituitary tumor, with or without SSE, and 4) abnormally low ratios may be found in patients with a pituitary tumor with SSE. (J Clin Endocrinol Metab 46: 576, 1978)

sellar extension of a pituitary tumor with a breakdown of the normal blood/CSF barrier (7-9). However, the results of these studies did not rule out the possibility that the bigger adenohypophyseal hormones are normal constituents of the CSF, although at a lower concentration than in plasma. We, therefore, decided to investigate this possibility and the assumption that measurable hormone levels in the CSF are indicative of suprasellar extension of a pituitary tumor. Inasmuch as hyperprolactinemia in the presence and in the absence of a pituitary adenoma occurs frequently (12-21), we measured PRL in plasma and CSF samples taken simultaneously from normoand hyperprolactinemic individuals with or without a pituitary tumor. Materials and Methods Patients Blood samples from an antecubital vein and CSF samples (taken either by lumbar, suboccipital, or ventricular puncture) were obtained simultaneously from: 1) 31 patients with various neurological or infectious, but non-endocrine diseases (group A); 2) 12 patients (7 pregnant women and 5 newborns)


The Endocrine Society. Downloaded from by [${individualUser.displayName}] on 15 November 2015. at 22:19 For personal use only. No other uses without permission. . All rights reserved.

PRL IN CSF with "physiological" hyperprolactinemia (group B); 3) 10 patients with various psychiatric, but nonendocrine diseases on treatment with psychotropic drugs known to induce hyperprolactinemia (group C); 4) 12 patients with chromophobe adenoma and SSE and normal or slightly elevated plasma PRL levels (group D); and 5) 14 patients with chromophobe adenoma and hyperprolactinemia, with (9) or without (5) SSE (group E). The diagnosis of chromophobe adenoma was based on visual field examination, sella tomography, pneumencephalography, and carotid angiography. In all patients with SSE except one, the diagnosis was confirmed by surgery and histological examination of the tumor tissue. The majority of the tumor patients had clinical and laboratory evidence of hypopituitarism; all 9 female patients and 5 of the 17 males had galactorrhea. In all but four patients, plasma PRL was measured on more than one occasion. All CSF samples were obtained from patients in whom a lumbar, suboccipital, or ventricular puncture was performed for diagnostic purposes related to their disease. In the tumor patients the CSF and plasma samples were obtained before surgery and/ or irradiation. PRL RIA The PRL preparations V-L-S 1 and 2 (National Pituitary Agency, NIH) were used for iodination and standard solutions. PRL (2 jug) was radioiodinated with 0.5 mCi 125I (Radiochemical Centre, Amersham, UK) by using the lactoperoxidase method (22, 23); iodinated PRL was purified by gel filtration on Sephadex G-75. The specific activity of the labeled hormone varied from 9.2-43.2 juCi/ jug. Antibodies specific for PRL were detected in the serum of a rabbit immunized more than a decade ago with an impure GH preparation (6). This antiserum was used in a final dilution of 1:10,000 in the PRL assay (1:250,000 in the GH assay). Radioactive hormone (approximately 0.2 ng), diluted antiserum, standard PRL (0-2 ng), or unknown plasma or CSF samples were incubated in phosphate buffer (0.02 M, pH 7.6, with 100 mg bovine serum albumin/100 ml) for 3-7 days at 4 C. Antibody-bound and free hormone were separated with dextran-coated charcoal. All samples were incubated in a final plasma concentration of 10% (vol/ vol); total incubation volume was 0.5 ml. Porcine plasma was added to standard and CSF samples [porcine PRL does not react with antibodies to human PRL (24)]; plasma samples were diluted with porcine plasma if necessary. All unknown spec-


imens were assayed in duplicate at more than one dose level, if feasible; plasma samples at a final dilution of 10% (v/v) and less, and CSF samples at 20% and less. The protein concentration of the clear CSF samples was normal or at most slightly elevated (< 80 mg/100 ml); CSF and plasma specimens from patients with meningitis taken during convalescense were assayed only. Heparinized plasma and centrifuged CSF samples were kept at —20 C until assayed. The assay is specific for PRL (6). High concentrations (1 /xg/tube) of human chorionic somatomammotropin (hCS; Medical Research Council, UK) did not register in the assay and highly purified GH (first international reference preparation, Medical Research Council, UK) even when added in high concentrations (1 /xg/tube) showed only negligible reactivity, presumably caused by some (0.1% or less) contamination with PRL. With this assay, normal PRL concentrations were found in plasma of acromegalic patients with high GH levels, whereas plasma GH was normal in patients with PRL-producing tumors and hyperprolactinemia. A parallel relationship between dilutions of plasma (6) and CSF samples and the standard curves was obtained, indicating the absence of non-specific interference (Fig. 1). PRL added to plasma (6) or CSF was recovered quantitatively (Table 1). The sensitivity of the normal assay (defined as twice the SD of the zero standard) was 0.125 ng/tube or 2.5 ng/ml plasma or CSF. The sensitivity could be increased 5-fold by using a preincubation method

B/B 0 100.

.003 .006



10 .012 .025" /u\ CSF



' fi\ CSF

human prolactin (ng/tube)

FIG. 1. Parallelism of diluted CSF samples and standard PRL. • , Standard PRL (± SD); • and • , tumor patients (no. 14, Table 7; no. 8, Table 6); O, pregnant subject (no. 3, Table 4); A, psychiatric patient (no. 7, Table 5); A and *, non-endocrine control subjects (nos. 23 and 27, Table 3).

The Endocrine Society. Downloaded from by [${individualUser.displayName}] on 15 November 2015. at 22:19 For personal use only. No other uses without permission. . All rights reserved.

J C E & M i i 1978 Vol 46 I No 4



TABLE 1. Recovery of PRL in three different CSF samples Sample I r n i j aaaea to sam- • Observed pie (ng/ml) % Recovery (ng/ml) 0 (< 0.5) 0 0.5 1.0 2.0 3.0

0.5 0.9 2.1

Sample III

Sample II Observed (ng/ml)

100 90 105

with addition of the tracer on the fourth and charcoal on the seventh day. When 100 /xl CSF was added in the preincubation system, the mean sensitivity ± SD was 0.52 ± 0.12 ng PRL/ml, as calculated from the results of seven different assays in which the three lowest concentrations of the standard curve (0, 0.5, and 0.75 ng/ ml) were assayed 10-fold. The intra-assay variation of the 100-/il CSF system was calculated for three different PRL concentrations (Table 2). The interassay variation coefficient at levels of 4 and 16 ng/ ml were 18 and 11%, respectively; in the preincubation procedure the variation coefficients were 17 and 9% at the 2 and 4 ng/ml levels, respectively. In this laboratory the following plasma PRL concentrations have been found: normal women (n = 25), 2.5-22 ng/ml (mean 7.4); normal men (n = 34), < 2.5-15 ng/ml; pregnant women in the third trimester (n = 12), 56-290 ng/ml (mean 134); and postpartum women after nursing (n = 11), 68-480 ng/ml (mean 162).

Results The results of the measurements in the various groups are presented in Tables 3-7. All patients without endocrine disease of group A (table 3) had normal plasma PRL levels (mean, 6.2 ng/ml; range, 1.3-14.5); CSF PRL levels were low (mean, 1.3 ng/ml; range, 0.6-4.7). The pregnant women and the premature infants of group B (table 4) were hyperprolactinemic (mean, 85.2 ng/ml; range, 31-200); their CSF PRL concentrations ranged from 3-28 ng/ml with a mean of 13.2. The psychiatric patients of group C (table 5) had plasma levels ranging from 3.5-160 ng/ml (mean, 54.3) and a mean CSF level of 6.5 ng/ ml (range, 0.7-18). Plasma concentrations ranging from 5.4-30 ng/ml (mean, 17.2) and CSF concentrations ranging from 2.7^-34 ng/ ml (mean, 9.7) were found in the tumor patients of group D (table 6), whereas very high

1.6 2.2 2.7 4.0 4.5

% Recovery 120 110 120 97

Observed (ng/ml)

% Recovery

4.2 4.6 5.6 6.0 7.0

80 140 90 93

TABLE 2. Intraassay variation of CSF samples (100 /xl) with low PRL levels CSF PRL concentration No. of different samples assayed in duplicate Variation coefficient (%)

1 ng/ ml 27 21.9

1-2 nj ml 20 14

2-3nj ml 19 14

plasma levels (mean, 2,529 ng/ml; range, 115-10,000) and CSF levels (mean, 1,449 ng/ ml; range, 6-13,000) were measured in the tumor patients with hyperprolactinemia of group E (table 7). The correlation between plasma and CSF PRL levels of all subjects without a pituitary tumor (groups A, B, and C) is depicted in Fig. 2. The regression equation (calculated by the method of the least squares) is: y = 0.122 x + 0.953. The correlation is positive (r = 0.9097) and highly significant (P = 0.00001). In the 26 tumor patients (groups D and E), the correlation between plasma and CSF PRL levels is again positive (r = 0.7141, P = 0.00002). The plasma/CSF concentration ratios (mean + range in parentheses) in the various groups were: group A, 5.2 (1.4-13.0); group B, 7.0 (2.9-10.3); group C, 7.3 (3.9-11.3); group D, 2.6 (0.9-7.1); and group E, 10.9 (0.2-34.9). None of the 53 patients without a pituitary tumor (groups A, B, and C) had a plasma to CSF ratio 3.0. In contradistinction, CSF PRL levels equalled or exceeded plasma levels in 4 (15%) of the 26 tumor patients of groups D and E, resulting in plasma to CSF ratios of 1.1 or less. In 8 patients (31%) the ratio was ^2.0 and in 11 (42%) the ratio was 2 X 106 mol wt) have been reported (25) and there is a correlation between the concentration ratio and the molecular size (or hydrodynamic volume) of the various proteins. Also, Linfoot et al. (1) reported clearly detectable levels of GH in the CSF of acromegalic patients with high plasma GH levels but without SSE, which suggested that the larger peptide hormones can cross the blood/ CSF barrier in the absence of detectable encroachment of the tumor in the third ventricle. Thus, the possibility existed that the CSF level of the adenohypophyseal hormones simply reflects the height of the plasma level. In order to test this hypothesis, the measurement of PRL seemed appropriate, because CSF samples taken for various diagnostic purposes from normo- or hyperprolactinemic patients with or without a pituitary tumor were readily available. Moreover, the sensitivity of the PRL assay could be increased rather easily, possibly enabling the assay of the hormone in CSF of normoprolactinemic subjects. The results presented in Tables 3-5 and in Fig. 2 indicate quite clearly that, in non-tumorous subjects, the level of PRL in CSF is indeed a function of the plasma concentration

The Endocrine Society. Downloaded from by [${individualUser.displayName}] on 15 November 2015. at 22:19 For personal use only. No other uses without permission. . All rights reserved.



and that the normal plasma to CSF concentration ratio is in the order of 6. PRL has a smaller molecular size (21,000 mol wt) than pre-albumin, and the fact that PRL has a lower plasma to CSF ratio than pre-albumin is consistent with the theory that the concentration of serum proteins in CSF is largely determined by their molecular size (25). It seems likely, therefore, that PRL enters the CSF by the same route (via the circulation) and by the same mechanism (filtration at the chorioid plexuses) as the other protein constituents of the blood. The validity of our data rests, of course, on the specificity of the PRL RIA in CSF, especially at the low levels found in the subjects of group A. However, the absence of non-specific interference in CSF, the essentially quantitative recovery of PRL added to CSF, and the results of assays in CSF diluted more than 1 in 10 (if feasible) and in less diluted CSF, all attest to the specificity of the assay. Furthermore, strong corroborative evidence is provided by the results of CSF measurements in the subjects with physiological and pharmacologically induced hyperprolactinemia (groups B and C), where the substance measured in various dilutions of CSF was unquestionably immunologically indistinguishable from standard PRL and where essentially the same plasma to CSF concentration ratios were found. Theoretically, one might argue that pregnancy, neonatal life, and meningitis are not physiological but abnormal states, in which the normal blood/CSF relationship does not prevail. During pregnancy the pituitary gland may enlarge greatly (26), possibly mimicking a pituitary tumor. Newborns are known to have an incomplete blood/CSF barrier (27, 28) and the integrity of the barrier may be altered in meningitis (29). Thus, it would not have been surprising to find abnormally low plasma to CSF concentration ratios in these patients. However, the mean ratio of the 12 subjects of group B (pregnant women and neonates, four of whom had meningitis) was 7.0 (range 2.9-10.3) and the mean ratio of the seven patients with meningitis of group A was 5.1 (range 3.3-6.5), which is not different from

J C E & M • 1978 Vol 46 • No 4

the ratios obtained in the other patients. It is possible that in the acute stage of meningitis the ratios might have been different; we have not assayed such samples because of the greater risk of non-specific factors influencing the assay. In the tumor patients of groups D and E, a possitive correlation between plasma and CSF PRL levels could be demonstrated, indicating again that it is the plasma level and not the presence of a tumor (with or without SSE) which mainly determines the CSF concentration. However, whereas none of the non-tumor patients had a CSF level equal to the plasma level and almost 90% had ratios >3, equal plasma and CSF levels were found in one sixth of the tumor patients and almost half had a ratio 4%) of big PRL in the CSF of tumor patients with SSE in whom big PRL constituted 11-25% of the total plasma concentration. Secondly, it is possible that PRL may also reach the CSF, apart from filtration at the chorioid plexuses, by an alternative and more direct route such as retrograde transport via the vessels of the stalk and median eminence (36-38). This possibility has been suggested for ACTH because immunoreactive ACTH has been found in CSF of normal individuals

The Endocrine Society. Downloaded from by [${individualUser.displayName}] on 15 November 2015. at 22:19 For personal use only. No other uses without permission. . All rights reserved.

PRL IN CSF in equal or even higher concentration than in plasma (10). Our results indicate that even when such more direct secretion occurs, it is still governed by the blood levels of the hormone. It is highly unlikely that, at least in the normal individual, a transport mechanism dissociated from and independent of the circulation and the normal blood/CSF barrier is operative in the regulation of the PRL content of CSF. However, distortion of pituitary architecture and/or alterations in blood flow by tumor growth could conceivably lead to more retrograde transport which, together with the high PRL concentration of the blood in the vicinity of the pituitary gland, could explain the finding of CSF PRL levels equal to, or even exceeding, the PRL concentration of peripheral plasma. Thirdly, if retrograde transport (or leakage) from the pituitary to the CSF under normal and/or abnormal conditions does indeed occur, it would not be surprising to detect higher hormone concentrations in CSF samples obtained from regions closest to the pituitary. It is of interest, therefore, that of the 21 tumor patients with SSE, the mean plasma to CSF ratio was lower (although statistically not significant, P < 0.05) in the 15 patients in whom CSF was obtained by suboccipital puncture (mean, 3.5; range, 0.2-8.5) than in the 6 patients whose CSF samples were obtained by lumhar puncture (mean, 5.5; range, 1.7-14.5); all 4 patients whose CSF PRL levels were equal to or higher than the plasma concentration underwent suboccipital puncture. From the results presented in Tables 3-7, it can be seen that in the various groups where lumbar or ventricular CSF was measured (including five tumor patients without SSE), the mean plasma to CSF ratio was 5.2 or higher. Thus, the data suggest that, at least in tumor patients, but perhaps also in the normal individual, suboccipital CSF may contain more PRL than peripheral (lumber or ventricular) CSF. Recently, several studies on the presence of adenohypophyseal hormones in CSF have been reported. Schroeder et al. (39) measured PRL in serum and CSF of control subjects, pregnant women, and patients with pituitary disease. In their 30 control subjects, the mean


serum PRL concentration (7.0 ng/ml) and the mean CSF PRL concentration (1.2 ng/ml) are surprisingly similar to the values found in our control subjects (6.2 and 1.3 ng/ml, respectively). Again, as in our tumor patients, a significant relationship (r = 0.953, P < 0.001) was demonstrated between the serum and CSF PRL levels of 12 hyperprolactinemic patients with pituitary tumors. Also, 3 pregnant women were found to have elevated CSF PRL levels, whereas in 15 patients with the primary empty sella syndrome and a defective diaphragma sellae, low CSF PRL concentrations were found. The latter finding would argue against the postulate that disruption of the diaphragma sellae by a pituitary tumor significantly influences the pituitary hormone concentrations of CSF. The authors conclude that the CSF PRL concentration is influenced by the serum PRL level, but that another mechanism besides passive diffusion may have been operative in determining the CSF PRL levels of two of their tumor patients (with SSE) in whom a higher PRL level in CSF than in serum was found, and in three normoprolactinemic tumor patients (two of whom had SSE) with elevated CSF PRL levels. They mention the possibility that decreased clearance of PRL from the CSF of these patients may have contributed to the elevated CSF levels. Although this is certainly a possibility, our results indicate that the finding of abnormally low plasma to CSF ratios in tumor patients may also be due to a higher PRL content of suboccipital CSF. Jordan et al. (9) reported CSF concentrations of GH, TSH, PRL, LH, and FSH in patients with various neurological disorders, patients with pituitary tumors with and without SSE, and patients with craniopharyngiomas. They found elevations of one or more adenohypophyseal hormones in the CSF of 21 of 22 patients with suprasellar extension of their pituitary tumor and in 2 of 6 patients with craniopharyngioma. They concluded that the finding of an elevated adenohypophyseal hormone concentration is a sensitive indicator of SSE of a pituitary tumor. Surprisingly enough, the authors were unable to detect PRL in the CSF of seven patients without

The Endocrine Society. Downloaded from by [${individualUser.displayName}] on 15 November 2015. at 22:19 For personal use only. No other uses without permission. . All rights reserved.



SSE, whose plasma PRL varied from 20-100 ng/ml, although they did find a CSF PRL concentration of 17 ng/ml in a patient with pseudotumor cerebri and a plasma PRL concentration of 89 ng/ml. Commenting on the findings of Jordan et al., Login (40) reported clearly elevated CSF PRL levels in five patients with PRL-secreting pituitary adenomas, despite the fact that the tumor was totally intrasellar in four of the five patients. In retrospective review, Jordan and Kendall (40) conceded that absolute levels of CSF PRL may not be a useful indicator of suprasellar extension; in their series of hyperprolactinemic patients without SSE, only one patient had sustained hyperprolactinemia and only in this patient was PRL CSF clearly elevated. This explanation may well be true; it has been shown for ACTH (10) and insulin (41) that CSF hormone levels respond slowly to an increase in the plasma level and do not mirror the moment to moment changes in hormone secretion. We have found very low (2 in two patients with SSE and

Prolactin in human cerebrospinal fluid.

Journal of Clinical Endocrinology and Metabolism Copyright © 1978 by The Endocrine Society Vol. 46, No. 4 Printed in U.S.A. Prolactin in Human Cereb...
1MB Sizes 0 Downloads 0 Views