The syndrome of inappropriate antidiuretic hormone (SIADH) is characterized by euvolemic hyponatremia. Patients with SIADH continue to drink normal amounts of fluid, despite plasma osmolalities well below the physiological osmotic threshold for onset of thirst. The regulation of thirst has not been previously studied in SIADH. We studied the characteristics of osmotically stimulated thirst and arginine vasopressin (AVP) secretion in eight subjects with SIADH and eight healthy controls and the nonosmotic suppression of thirst and AVP during drinking in the same subjects. Subjects underwent a 2-h infusion of hypertonic (855 mmol/l) NaCl solution, followed by 30 min of free access to water. Thirst rose significantly in both SIADH (1.5 ± 0.6 to 8.0 ± 1.2 cm, P < 0.0001) and controls (1.8 ± 0.8 to 8.4 ± 1.5 cm, P < 0.0001), but the osmotic threshold for thirst was lower in SIADH (264 ± 5.5 vs. 285.9 ± 2.8 mosmol/kgH2O, P < 0.0001). SIADH subjects drank volumes of water similar to controls after cessation of the infusion (948.8 ± 207.6 vs. 1,091 ± 184 ml, P = 0.23). The act of drinking suppressed thirst in both SIADH and controls but did not suppress plasma AVP concentrations in SIADH compared with controls (P = 0.007). We conclude that there is downward resetting of the osmotic threshold for thirst in SIADH but that thirst responds to osmotic stimulation and is suppressed by drinking around the lowered set point. In addition, we demonstrated that drinking does not completely suppress plasma AVP in SIADH.
- syndrome of inappropriate antidiuretic hormone
- arginine vasopressin
in healthy humans, the osmotic threshold for thirst is very similar to that of vasopressin secretion, such that at plasma osmolalities above ∼285 mosmol/kgH2O, there is simultaneous stimulation of drinking and vasopressin release, the magnitude of which depends on the degree of hyperosmolality (16). The stimulation of thirst, which promotes fluid intake, and the actions of vasopressin to reduce water excretion, are synergistic in returning plasma osmolality to normal. At plasma osmolalities below the osmotic thresholds, both thirst and vasopressin release are inhibited, allowing the development of a hypotonic diuresis. The relationship between plasma vasopressin, thirst, and plasma osmolality is highly reproducible within individuals, with close concordance of osmotic thresholds for both thirst and vasopressin release on repeated testing (19). In patients with cranial diabetes insipidus, who rely on water intake to maintain water homeostasis in the absence of osmoregulated vasopressin release, the osmotic threshold for thirst is entirely normal (15), whereas in compulsive water drinking, there is lowering of the osmotic threshold for thirst to well below that of vasopressin release (9, 18).
The syndrome of inappropriate antidiuretic hormone (SIADH) is characterized by euvolemic hyponatremia, with plasma osmolalities that are consistently below the normal threshold for thirst and vasopressin release (1). The characteristics of vasopressin secretion in SIADH have been well documented, and four distinct patterns of abnormal vasopressin secretion have been described (11). However, there has been little information on the characteristics of thirst in SIADH. At the typical plasma osmolalities seen in SIADH, which are well below the physiological osmotic threshold for the onset of thirst, the desire to drink should be suppressed. In clinical practice, patients with SIADH appear to have thirst and daily fluid intake similar to that of healthy individuals. Robertson (10) has argued that the persistence of thirst at such low plasma osmolalities is pathological in itself and is the prime reason for the hyponatremia that characterizes SIADH (10).
In this study, therefore, we have assessed the osmotic stimulation and the nonosmotic suppression of thirst appreciation in patients with SIADH.
We recruited eight patients with SIADH (all males, median age 53.5 yr, range 34–66 yr), who attended the routine endocrine service and eight matched healthy controls (all males, median age 51.8 yr, range 32–71 yr). The patients were adjudged to have SIADH on the basis of euvolemic hyponatremia, with natriuresis and normal thyroid function and glucocorticoid secretion, fulfilling the diagnostic criteria of Verbalis (23). The etiology and baseline clinical data on the SIADH patients are outlined in Table 1. The use of hypertonic saline infusion for the study of patients with SIADH was approved by the Hospital Ethics Committee for patients whose baseline plasma sodium was > 120 mmol/l. Separate ethics committee approval was obtained for the use of hypertonic saline in control subjects for the purposes of the study. Patients and control subjects gave informed written consent after careful discussion of the potential side effects of hypertonic saline infusion.
Subjects were admitted to the investigation unit after overnight fast, including abstinence from caffeine, nicotine, and alcohol for at least 24 h. Free access to tap water was allowed on the morning of study up to the time of admission to the investigation unit. Cannulae were inserted into the brachial vein in each arm under local anesthesia (1% lidocaine), one for infusion of saline and one for withdrawal of blood samples. Subjects were rested recumbent for 30 min, after which two baseline blood samples separated by an interval of 15 min were withdrawn. Saline infusion was not performed in any patients with SIADH if they had evidence of symptomatic hyponatremia, if their baseline plasma sodium was <120 mmol, or if the patients were too systemically unwell to tolerate repeated blood testing.
Hypertonic saline (855 mmol/l NaCl) was then infused intravenously at 0.05 ml·kg−1·min−1 for 2 h in an adaptation of the method of Baylis and Robertson (2); after a 15-min equilibration period, subjects were then allowed free access to unlimited tap water at room temperature for 30 min.
Blood samples were withdrawn for plasma vasopressin, plasma osmolality, plasma sodium, and plasma glucose at 30-min intervals during the hypertonic saline infusion and at 5-min intervals during the period of free access to tap water. Thirst was assessed at blood sampling times, and the volume of water drunk in the period postinfusion was noted.
Free-running venous blood was withdrawn into cooled syringes and transferred to chilled lithium-heparin tubes. The blood was immediately centrifuged at 4°C and 2,000 g for 20 min, and the plasma supernatant was aspirated. On the day of the study, aliquots were separated for the measurement of plasma osmolality by the depression of freezing point method and of plasma sodium by an ion-specific electrode. The remaining plasma was then immediately frozen at −70°C for later measurement of plasma vasopressin. Vasopressin was extracted from plasma with magnesium silicate beads (Florisil), and vasopressin was measured by a sensitive and specific radioimmunoassay [inter- and intra-assay coefficients of variation, 9.7 and 15.3%, respectively; limit of detection, 0.3 pmol/l (13)].
Whole blood for measurement of blood glucose was withdrawn into fluoride oxalate tubes and analyzed using a glucose oxidase technique. Blood for plasma renin activity was taken into lithium heparin tubes for measurement using a commercial radioimmunoassay.
Blood pressure was recorded from the brachial artery with a manual sphygmomanometer, and mean arterial blood pressure was calculated by adding one-third of the pulse pressure to the diastolic blood pressure.
Thirst was recorded on a previously described visual analog scale (1). Subjects were presented with a 10-cm-long uncalibrated line and asked the question, “How thirsty are you?” They were invited to answer this question by placing a mark on the line between the extremes of “no thirst” at 0 cm and “very thirsty indeed” at 10 cm. For the purpose of statistical analysis, thirst ratings were defined as the distance in centimeters of the subject’s mark from the 0-cm extreme.
Results are presented as means ± SD. Changes in parameters with time were analyzed by one-way analysis of variance and between subject groups by analysis of variance with repeated-measures design or Student’s t-test, where appropriate. Linear regression analysis was calculated by the method of least squares. Results were analyzed using SPSS (for Windows 10.5; SPSS, Chicago, IL), and differences were regarded as statistically significant if P values were <0.05.
At baseline, plasma sodium was lower in the SIADH group (125.8 ± 2.7 mmol/l) than in the control group (140.3 ± 2.1 mmol/l, P < 0.01), as was plasma osmolality (SIADH 267.3 ± 4.7 vs. control 289.8 ± 2.2 mosmol/kgH2O, P < 0.001) and blood urea (SIADH 2.3 ± 0.8 vs. control 4.1 ± 1.0 mmol/l, P = 0.02). Plasma renin activity and mean arterial blood pressure (SIADH 90.6 ± 4.7 vs. control 90.3 ± 4.7 mmHg) were similar at baseline in the two groups. Serum albumin was lower at baseline in the SIADH group (38.4 ± 2.1 g/l) than in controls (41.1 ± 3.4, P = 0.017).
Infusion of hypertonic saline caused significant elevation in plasma osmolality in the SIADH group (267.3 ± 4.7 to 279.5 ± 6.5 mosmol/kgH2O, P < 0.0001) and the controls (289.8 ± 2.2 to 307.6 ± 4.3 mosmol/kgH2O, P < 0.0001), with plasma osmolality higher throughout the infusion period in the controls (P = 0.003, Fig. 1). The elevation in plasma osmolality stimulated a rise in plasma arginine vasopressin (AVP) concentration from 3.2 ± 3.3 to 6.7 ± 3.0 pmol/l (P = 0.03) in the SIADH group and from 0.9 ± 0.4 to 6.3 ± 2.4 pmol/l in the controls (P < 0.0001). There was no difference in plasma AVP concentrations between the two groups at baseline (P = 0.08) or during the infusion period (P = 0.068; Fig. 2). Urine osmolality at the end of the infusion period was similar in the two groups (SIADH 945 ± 108 mosmol/kgH2O and control 876 ± 142 mosmol/kgH2O, P = 0.65), with maximal urine concentration in each group.
Thirst rose from 1.5 ± 0.6 to 8.0 ± 1.2 cm (P < 0.0001) in the SIADH group and from 1.8 ± 0.8 to 8.4 ± 1.5 cm (P < 0.0001) in the control group, with no differences in thirst ratings between the two groups during the infusion period (P = 0.29; Fig. 3). SIADH subjects drank volumes of water similar to controls after cessation of the infusion (948.8 ± 207.6 vs. 1,091 ± 184 ml, P = 0.23).
Linear regression analysis showed a relationship between plasma AVP and plasma osmolality in only six of the eight subjects with SIADH. Subjects B and F had no relationship between plasma osmolality and plasma AVP, in a pattern typical of type A SIADH, whereas the other six subjects had a linear relationship between AVP and osmolality with a low osmotic threshold, typical of type B SIADH. The relationship between plasma osmolality and plasma AVP is shown graphically in Fig. 4, and the individual and mean regression lines are shown in Table 2. The mean osmotic threshold for AVP release was lower in SIADH subjects (263.8 ± 7.3 mosmol/kgH2O) than in controls (288.0 ± 3.6 mosmol/kgH2O, P < 0.001).
Linear regression analysis showed that there was a linear, close relationship between thirst ratings and plasma osmolality in all patients with SIADH (Fig. 5, Table 2). The osmotic threshold for thirst was significantly lower in the SIADH subjects (264.0 ± 5.5 mosmol/kgH2O) than in controls (285.9 ± 2.8 mosmol/kgH2O, P < 0.001). The osmotic thresholds for thirst and vasopressin release were not different in either SIADH (P = 0.84, Table 2) or controls (P = 0.58, Table 2).
Drinking led to an immediate fall in thirst ratings in both the SIADH and control group. In controls, drinking led to an immediate fall in thirst ratings, from 8.4 ± 1.5 to 4.6 ± 1.6 cm (P = 0.004) within 5 min, despite unchanged plasma osmolality (307.6 ± 4.3 to 307.0 ± 4.3 mosmol/kgH2O, P = 0.92). After a further 25 min of drinking, mean thirst ratings had fallen to the preinfusion mean (1.3 ± 0.5 cm) despite high ambient plasma osmolality (303.0 ± 2.7 mosmol/kgH2O), confirming the results of previous studies. In the SIADH group, thirst fell significantly after only 5 min of drinking, from 8.0 ± 1.2 to 4.7 ± 0.8 cm (P < 0.001), despite no significant change in plasma osmolality (279.5 ± 6.5 to 279.1 ± 6.9 mosmol/kgH2O, P = 0.94). Thirst had fallen to preinfusion values after 30 min of drinking (1.8 ± 0.5 cm) despite persistent elevation of plasma osmolality (274.9 ± 6.3 mosmol/kgH2O) above baseline values.
In controls, drinking was associated with a fall in plasma AVP concentrations, from 6.3 ± 2.4 to 3.5 ± 1.3 pmol/l (P = 0.012) within 5 min, despite no significant change in plasma osmolality (307.6 ± 4.3 to 307.0 ± 4.3 mosmol/kgH2O, P = 0.92). Plasma AVP had fallen to baseline values after 30 min of drinking (1.2 ± 0.5 cm) despite continued hyperosmolality (303.0 ± 2.7 mosmol/kgH2O). In contrast, plasma AVP remained unchanged in the SIADH group after 5 min (6.7 ± 3.0 to 5.9 ± 2.7 pmol/l, P = 0.78) and was significantly higher than in the control group (P = 0.024). Plasma vasopressin concentrations remained higher in the SIADH group throughout the remainder of the drinking period (P = 0.007) despite the SIADH group drinking volumes of water similar to the controls (SIADH 949 ± 208 vs. controls 1,091 ± 184 ml, P = 0.23). Drinking did not, therefore, suppress plasma AVP concentrations as effectively in the SIADH group.
This is the first study to formally examine the characteristics of the osmotic control of the stimulation of thirst and the nonosmotic suppression of thirst by drinking in subjects with SIADH. The results show that elevation of plasma osmolality stimulates the sensation of thirst in a linear manner, similar to the physiological control of thirst in healthy controls. However, although the slope of the mean regression line relating plasma osmolality and thirst was similar to that in controls, the osmotic threshold was lowered to a threshold of 20 mosmol/kgH2O below that in the control cohort, to a level of plasma osmolality where thirst would be completely suppressed in physiological conditions. A number of clinical studies have demonstrated that the threshold for thirst in healthy individuals is ∼285 mosmol/kgH2O (16, 12, 17), and the studies of Robertson (8) estimate even higher osmotic thresholds for thirst. Our control data showed osmotic thresholds for thirst very similar to those in previous studies; the lowered osmotic threshold for thirst in the SIADH group is a novel finding.
The osmotic thresholds for thirst (285.9 ± 2.8 mosmol/kgH2O) and vasopressin release (288.0 ± 3.6 mosmol/kgH2O) in the healthy control group were in keeping with previously reported data (16). In the six SIADH patients in whom there was a pattern of vasopressin secretion sufficient to construct a linear regression line, the osmotic threshold for thirst (264.0 ± 5.5 mosmol/kgH2O) was identical to that for vasopressin release (263.8 ± 7.3 mosmol/kgH2O), a situation similar to that in the control group in this study and previously published studies (16). Therefore, in SIADH the entire osmoregulatory set point is reset downward to a lower plasma osmolality.
It has previously been argued that patients with SIADH must have abnormally high rates of water intake to develop hyponatremia, on the basis that the administration of vasopressin to normal subjects does not lower plasma sodium concentrations in the absence of increased water intake (10). The results of our study do not support this theory, as the SIADH cohort did not drink more than the control cohort in response to osmotic stimulation, with a similar water intake per unit rise in plasma osmolality. The rate of water intake was, however, clearly inappropriate to the ambient plasma osmolality, as thirst should be suppressed. This is consistent with the scenario of a lowered osmoregulatory set point, with “normal” osmotically stimulated vasopressin secretion, thirst, and drinking behavior at plasma osmolalities above the set point.
Abnormal vasopressin secretion in SIADH is classified into four main types (11). Two patients had patterns of vasopressin secretion typical of type A, with random hypersecretion unrelated to plasma osmolality, whereas the other six had patterns of secretion typical of type B, which is characterized by a lowered osmotic threshold for vasopressin release. Interestingly, in these patients, the osmotic threshold for vasopressin release was almost identical to that for onset of thirst. This has never been previously reported in SIADH, although it is well described in physiological conditions.
The mechanism by which the osmotic thresholds for thirst and vasopressin release are reset in SIADH is not clear. Because of the diversity of etiology of SIADH in our patient group and the lowered osmotic threshold for thirst and vasopressin release, it seems most likely that there is a resetting of both the thirst and vasopressin osmoreceptors within the hypothalamus. It has been reported that small cell carcinoma of the lung express mRNA for vasopressin in tissue samples (4), and two of the four patients in this series showed dissociation between plasma osmolality and plasma vasopressin. It is possible that, in these two patients, basal vasopressin secretion was from the tumor itself rather than the posterior pituitary, but, given the similarity of the characteristics of osmotically stimulated thirst and AVP release, we postulate a hypothalamic control of stimulated thirst and AVP release.
The mechanism by which the hypothalamic osmoreceptors are reset in SIADH is not known and is beyond the scope of this paper. It is possible to hypothesize, however, that there may be a complex interaction between aquaporin water channels, cerebral osmolality, and the inappropriately elevated vasopressin levels characteristic of SIADH. Aquaporin water channels have been identified within the hypothalamus in areas known to be associated with the osmoregulation of vasopressin secretion and thirst (22) and are upregulated in the setting of systemic hyponatremia (21). Within the hypothalamus, the aquaporin water channels are expressed preferentially on astrocyte cell membranes, in particular on perivascular end foot processes (24), thereby in a perfect position to act as a link between the systemic circulation and the hypothalamic osmoreceptors. Vasopressin has been shown in vivo to stimulate neuronal activity within the astrocyte by controlling the flux of water through aquaporin-4 water channels (5). Therefore, we could hypothesize that in SIADH the presence of hypervasopressinemia, despite the hypoosmolality, promotes a signal through the hypothalamic aquaporin water channels to stimulate the hypothalamic osmoreceptors for thirst, causing resetting of the osmotic threshold to a lower set point. This hypothesis is, however, at present pure speculation and requires further research.
The results of the drinking phase were interesting and provide new insights into the suppression of thirst and vasopressin secretion during drinking. It is well established that drinking causes a rapid fall in plasma vasopressin concentrations and thirst ratings in humans (3, 17, 6) and a fall in vasopressin with rapid cessation of drinking in animals (20), despite a much slower rate of fall of plasma osmolality. The putative mechanism is a neural reflex initiated by oropharyngeal stretch (14). The suppression of thirst in the SIADH group was very similar to that in the control group, indicating that neural inhibition of thirst and drinking is preserved in this group. In contrast to the control group, however, the fall in plasma vasopressin concentrations in the SIADH was much slower, and although they fell to preinfusion levels by the end of the drinking period, they were still well above the suppressed level necessary to allow a diuresis after drinking. This may explain why hyponatremia is maintained in patients with SIADH despite there being no evidence from our data to support the hypothesis that excess drinking causes the drop in plasma sodium concentrations in the syndrome; if vasopressin secretion is never completely suppressed by drinking, to allow a hypotonic diuresis, excess fluid intake may not be necessary to provoke hyponatremia. This hypothesis could be tested by studies to examine whether water loading could suppress plasma vasopressin to enable a diuresis, but such studies would run the risk of lowering plasma sodium to a level that could provoke seizures. Studies in an animal model of SIADH may therefore be needed to advance this area of research.
Plasma vasopressin samples were measured in the laboratory of Prof. Peter Baylis in Newcastle-upon-Tyne.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2004 by American Physiological Society