Primary Polydipsia: update

Abstract

In primary polydipsia pathologically high levels of water intake physiologically lower arginine vasopressin (AVP) secretion, and in this way mirror the secondary polydipsia in diabetes insipidus in which pathologically low levels of AVP (or renal responsiveness to AVP) physiologically increase water intake. Primary polydipsia covers several disorders whose clinical features and significance, risk factors, pathophysiology and treatment are reviewed here. While groupings may appear somewhat arbitrary, they are associated with distinct alterations in physiologic parameters of water balance. The polydipsia is typically unrelated to homeostatic regulation of water intake, but instead reflects non-homeostatic influences. Recent technological advances, summarized here, have disentangled functional neurocircuits underlying both homeostatic and non-homeostatic physiologic influences, which provides an opportunity to better define the mechanisms of the disorders. We summarize this recent literature, highlighting hypothalamic circuitry that appears most clearly positioned to contribute to primary polydipsia. The life-threatening water imbalance in psychotic disorders is caused by an anterior hippocampal induced stress-diathesis that can be reproduced in animal models, and involves phylogenetically preserved pathways that appear likely to include one or more of these circuits. Ongoing translational neuroscience studies in these animal models may potentially localize reversible pathological changes which contribute to both the water imbalance and psychotic disorder.

Characteristic populations and concurrent diagnoses

Severe mental illness (SMI) and neurodevelopment disorders

Primary polydipsia (PP) is common in patients with neurodevelopmental disorders (such as autism and intellectual disabilities) and chronic psychotic disorders (such as schizophrenia, schizoaffective disorder, bipolar disorder and psychotic depression) (Table 1). In chronic schizophrenia, it is found in 11–20% of patients who generally consume 5 to 15 liters of fluid/day (1). Up to one third of the latter (3–6% of total with chronic schizophrenia) experience episodes of symptomatic hyponatremia (plasma sodium typically below 125 mEq/L) and episodic water intoxication (plasma sodium typically below 120 mEq/L) a condition labeled by Vieweg as psychosis-intermittent hyponatremia-polydipsia syndrome(PIP) (2).

Table 1:

Characteristics of major categories of primary polydipsia

Disorders	Clinical features^	Treatment*	Physiologic findings	Proposed Mechanism
Psychosis-intermittent hyponatremia-polydipsia syndrome (includes polydipsic chronic psychotics who do not become hyponatremic)	Not thirsty: report drinking reduces dysphoria. Other stereotypies. Risk of life-threatening water intoxication with acute psychosis	Clozapine, target weight program for institutionalized PIP patients, vaptans for symptomatic PIP patients.	Reset osmostat for AVP varies with psychosis and acute psychological stress, proportional to anterior hippocampal (AH) deformations.	Disruption of normal AH restraint of stress impacts lateral hypothalamic and nucleus accumbens non-homeostatic modulators of water intake and motivated behaviors
Autism, intellectual disability, bipolar, psychotic depression	Hyponatremia and water intoxication are rare, thirst uncommon complaint.	None, target weight program for rare institutionalized patient with symptomatic hyponatremia	Not studied	Disruption of non-homeostatic influences on non-thirst element of water intake
Compulsive water drinking/psychogenic polydipsia	Depression, anxiety, OCD, anorexia, alcoholism. Excess thirst in some. Females more common.	Psychopharmaco- logic or behavioral treatment specific to psychiatric disorder	Reset osmostat for thirst, enhanced intake for degree of thirst with delayed satiation.	Disruption of non-homeostatic influences on water intake
Health enthusiasts/athletes	Believe drinking in excess of need is healthy, promotes cognition and performance	Education	Not studied	Habitual behavior, i.e. intact water balance overridden by cognitive beliefs
Dipsogenic DI	Thirst-driven, no apparent psychologic factors	Desmopressin	Reset osmostat for thirst, relative increased threshold for AVP, high basal osmolality	Disruption of homeostatic influences on thirst.
Brain injury or disease (Sarcoid, TBI, hypothalamic pathology)	Polyphagia more common, AVP regulation may be intact	None	Disrupted congruence of AVP/thirst set points in some.	Disruption of homeostatic or non-homeostatic influences on water intake

^#All patients with primary polydipsia have an increased risk of urinary tract abnormalities which may evolve to renal disease. Also long standing polydipsia is a risk factor for osteoporosis and pathological fractures.

^*It is critical to avoid medications that impair water excretion, especially desmopressin except for dipsogenic DI.

Other psychiatric and medical conditions and general population

Other individuals with non-psychotic Axis I psychiatric disorders also have PP, a condition typically called compulsive water drinking (CWD) or psychogenic polydipsia (1). The prevalence of CWD is higher in women than men (3), and many of the patients have depression, anxiety, obsessive-compulsive disorder, anorexia nervosa or alcoholism (4).

Other persons with PP become habituated to drinking water owing to fear of dehydration or the belief that water improves their health (5) and may or may not have a diagnosable psychiatric disorder. Indeed, the prevalence of PP is increasing in the general population and is perhaps a learned behavior attributable to the increased pressures of adult life and the popularity of lifestyle programs emphasizing fluid intake (6). Many athletes, particularly marathon runners drink water in advance of athletic events. Dipsogenic DI appears unrelated to psychiatric symptoms, overvalued beliefs or social norms (7). Finally, hypothalamic sequelae of traumatic brain injuries, vascular or infiltrative diseases (e.g., sarcoidosis) may induce PP rather than DI (8).

Clinical features, risk factors, morbidity, and differential diagnosis

Severe mental illness with (i.e. PIP) and without hyponatremia

PP patients with psychotic disorders rarely complain of thirst but instead provide delusional explanations for their excessive drinking or state that drinking reduces their anxiety and makes them feel better (9, 10) (Table 1). Because the polydipsia appears related to an internal state distinct from thirst (9), this state is typically assessed by asking patients about their ‘desire for water’ (in cups) (11), which can be validated by measuring ad libitum intake over a fixed period (e.g. 10 minutes) (12). These patients can often be identified in community settings because they have a cup in their hand at all times and tend to gather around radiators during cold weather (10). Compared to non-polydipsic patients they drink three times as often and consume five times more fluid (13). They frequently drink from non-potable sources (e.g. toilets) and rarely drink at night compared with patients with DI. They generally weigh less than other patients and along with their first-degree relatives, have a higher incidence of alcoholism and smoking than others with schizophrenia (14,15). Their water drinking often coincides with other stereotypies (such as smoking, pacing and mannerisms) as well (16).

Worsening hyponatraemia and water intoxication in the PIP subset typically coincide with exacerbations of the underlying psychotic illness (17). PIP and, to a lesser extent, other polydipsic schizophrenics (i.e. those who are consistently normonatraemic) suffer increased morbidity and mortality. The median age at death is 57 years old compared to 68 for non-polydipsic patients (an age much younger than the general population due primarily to metabolic syndrome and suicide) (18). Severe hyponatremia is life-threatening as are some of its sequala (e.g. osmotic demyelination syndrome). Even mild levels of hyponatremia have been associated with osteoporotic fractures, ataxia and falls (19); as well as cognitive deficits (3,20), which also likely contribute to the elevated morbidity or mortality (21). Other associated disorders, e.g. rhabdomyolysis (22) and stress cardiomyopathy (23), contribute to morbidity.

Hyponatraemia in PIP patients is dilutional and must be distinguished from iatrogenic causes: psychotropic medications (such as antipsychotic drugs, SSRIs or the anticonvulsant carbamazepine), acute nicotine ingestion, or medications prescribed for hypertension (such as thiazides), diabetes mellitus (such as chlorpropamide), diabetes insipidus (e.g. desmopressin) or psychosis itself (antipsychotics such as haloperidol, thorazine) all of which can impair water excretion (24). The dilutional hyponatremia induced by antipsychotic drugs can generally be distinguished from that attributable to psychotic exacerbations because antipsychotic drugs induce stable, severe impairments in water excretion while impairments in PIP patients are mild and transient, and typically precede (rather than follow) changes in antipsychotic therapy (25). Diabetes mellitus, cardiac and renal diseases themselves contribute to dilutional hyponatremia by their adverse impact on water excretory capacity. While PIP is clearly associated with transient impairments in water excretion, water intake may rarely overwhelm the diluting capacity of normal functioning kidneys (≥ 20L/day) even in the presence of sufficient osmotic intake (e.g. > 600 mOsm) and absence of aberrant influences.

Other psychiatric and medical and non-medical conditions

Unlike patients with psychotic disorders, others with PP are more likely to complain of excessive thirst. CWD are often heavy smokers, drinkers of alcohol, and exhibit eating disorders all of which link the drinking to addiction and other ‘oral fixations’ (4,6,26) (Table 1). Habitual polydipsia occurs in lifestyle conscious men and women who believe that excess water helps detoxify the body and improve cognition (3,20). PP is common in runners, perhaps reflecting a learned behavior and can even cause water intoxication probably in association with a solute deficiency (27). Dipsogenic DI is inevitably associated with increased thirst in the absence of the above conditions or factors (7). Differential diagnosis of these disorders from diabetes insipidus is generally more difficult than for patients with chronic psychosis and is discussed elsewhere in this series of papers.

While thirst may be a prominent symptom in many with CWD, homeostatic-initiated thirst (e.g. hypernatremia, hypovolemia, hypotension) is rarely the cause but instead likely the product of psychological or cognitive influences on thirst intensity. Increased thirst in anorexia nervosa, for instance, appears to have little to do with water intake per se, but a means of compensating for low food intake and diminishing hunger (11). PP associated with organic brain disorders often occurs in conjunction with polyphagia and is not necessarily linked to altered AVP regulation. Hypothalamic sarcoidosis, for instance, is more likely to enhance water intake than to impair AVP secretion (3).

PP itself may induce urinary tract abnormalities and promote infection that can ultimately cause renal damage. Long standing polydipsia is a risk factor for osteoporosis and pathological fractures (17).

Management of PP

Severe mental illness with (i.e. PIP) and without hyponatremia

Unlike other antipsychotic agents, clozapine has consistently been reported to normalize sodium levels in PIP patients (Table 1). Clozapine appears to primarily decrease fluid intake rather than normalize AVP regulation (30), and thus may improve PP in normonatraemic psychotic polydipsic patients as well. It seems less likely that it would benefit others with PP given the evidence that the pathophysiology in this population, discussed further below, is specific to their underlying psychotic disorder. None of the published studies were placebocontrolled and clozapine carries unique, potentially life-threatening risks, making a risk-benefit analysis problematic but still essential. Other medications reported to reduce polydipsic behavior in SMI are phenytoin, propranolol, bupropion, SSRIs and other treatments for OCD (31,32), though salutary effects have been inconsistent.

The most severe risk of the water imbalance is episodic water intoxication, a common occurrence in PIP. An alternative or adjunct to clozapine is the vaptans, AVPR2 antagonists. These agents rapidly normalize serum sodium concentration in PIP patients but require close monitoring to avoid dehydration and renal damage (33). Studies of other interventions, i.e. electrolyte-containing beverages, reducing the dose or switching to another antipsychotic agent, adjunctive angiotensin inhibitors, acetazolamide, demeclocycline, α-adrenergic receptor antagonists, opioid antagonists, have been inconclusive (24,34).

Measuring diurnal weight gain and imposing brief fluid restriction is an effective nonpharmacologic intervention capable of preventing water intoxication in monitored settings (e.g. hospital, nursing or group homes) (2). An acute increase in body weight can predict symptomatic hyponatremia (recall the hyponatremia is dilutional) if patients are weighed regularly and when exhibiting prodromal signs of water intoxication (e.g. ataxia, nausea, confusion, sedation). Generally, the acute increase must exceed 5 to 8 Kg for the patient to be at risk of water intoxication, making regular scales a more than sufficient tool. Patients are weighed each morning (when they are generally at their basal plasma sodium level) and in the afternoon (when episodes of water intoxication are most likely to occur) as well as anytime they exhibit prodromal signs. A brief fluid restriction (1–3 h) is imposed if the body weight exceeds the pre-determined threshold, and generally produces a marked diuresis restoring AM body weight since the impairment in water excretion is relatively mild. Prolonged restriction must be avoided to prevent osmotic demyelination syndrome.

Other psychiatric and medical conditions and general population

Management of PP outside of SMI depends largely on the psychiatric diagnosis of the patient as CWD typically resolves with treatment of the core psychiatric disorder (e.g. depression, anxiety or obsessive-compulsive disorder, anorexia nervosa). However, no effective treatment may be available, leaving patients at risk of structural urinary tract abnormalities and other medical complications. Ice chips may reduce polydipsia in some, perhaps by suppressing drinking behavior through recognized thermoregulatory mechanisms as part of oropharyngeal regulation (19). Desmopressin has successfully treated dipsogenic DI, which is not surprising since it is attributable to an impairment in homeostatic regulation of thirst (diminished osmotic set point) that can be overcome by water retention (7). Desmopressin is generally contraindicated in those whose emotional state appears to be the primary driver of the excessive intake because of the risk of symptomatic hyponatremia. Standard behavioral treatments, such as relaxation, response prevention, and cognitive behavioral therapy have been reported to be successful in small case studies (35,36,37). A single case study in a hyponatremic schizophrenic patient with diabetes insipidus demonstrated the behavioral treatment was so effective the person could subsequently be safely treated with desmopressin (19).

Physiologic findings and pathophysiology

Severe mental illness with (i.e. PIP) and without hyponatremia

Identifying a reset osmostat and its worsening by acute psychosis in PIP patients

Studies carried out over the past century have identified aberrant influences on water balance regulation in chronic psychosis that appear linked to the underlying mental disorder (Table 2). In the 1980s, carefully conducted studies confirmed PIP patients have a reset osmostat (type C SIADH), that is the osmotic set point for the threshold for desire for water and for AVP secretion are both lower in polydipsic patients than matched patients or healthy controls (38) (Table 1). Reset osmostat typically reflects the impact of a non-osmotic influence (e.g. hypotension, hypovolemia) on thirst and AVP, but none of the recognized physiologic or pharmacologic modulators were apparent.

Table 2:

Progression of observations that have tied polydipsia and symptomatic hyponatremia in chronic psychosis to neural structures and functions implicated in water balance regulation and psychosis*

Studies:
-in the first decades of the20th century.	Water excretion diminishes during acute psychosis Polydipsia occurs in 25% of patients with chronic psychosis Water intoxication coincides with exacerbations of psychosis
- linking water imbalance to non-osmotic stimuli	Reset osmostat for AVP and for desire for water in PIP patients Reset osmostat for desire for water only in normonatremic polydipsic patients Desire for water rebounds rapidly following intake in polydipsic patients
-linking reset osmostat to acute psychosis.	Acute psychosis increases AVP secretion Induced psychotic exacerbations lower AVP set point in PIP patients to level capable of inducing water intoxication. No effect of acute psychosis on AVP secretion in normonatremic polydipsics
-linking psychosis to AH-mediated stress diathesis.	Reset osmostat appears to normalize in PIP patients with habituation to research setting. Polydipsic patients have higher incidence of cortisol resistance to dexamethasone AH volume smaller in schizophrenia and linked to stress diathesis Anterior hippocampal (AH) volume smaller in PIP, not normonatremic polydipsics AH equivalent in rodents restrains HPAA and AVP responses to psychological stress Enhanced HPAA and AVP responses to psychological stess reproduced in animal model of schizophrenia stress diathesis Psychological, but not physical, stress enhances AVP and HPAA responses and reproduces psychosis-induced lower set point for AVP in PIP patients
-showing normonatremic polydipsics exhibit similar, albeit less marked, AH pathology	AVP and HPA responses in non-polydipsic psychotic patients blunted consistent with intact neuroendocrine responses to chronic stress. Responses not blunted in normonatremic polydipsics, albeit not heightened. Deformations on lateral AH in PIP and normonatremic polydipsics, only medial AH in nonpolydipsics patients. Size and position of deformations predicted AVP responses in all three groups.
-linking AH pathology to psychotic disorder	Social dysfunction greater in polydipsic than nonpolydipsic psychotic patients. Diminished oxytocin linked to social dysfunction in schizophrenia AH projection which modulates AVP, HPAA responses to stress appears to influence oxytocin secretion Oxytocin levels are lower in polydipsic than nonpolydipsics patients in proportion to deficits in social cognition and predicted by size and position of AH deformations. Supplemental oxytocin restores social cognition in polydipsic but not nonpolydipsics patients.

The findings accounted for the basal hyponatremia in PIP patients but were insufficient to account for episodes of water intoxication which typically coincided with exacerbations of the psychotic disorder and were reported to further diminish the osmotic threshold for AVP secretion (17, 38). A study confirmed that psychotic exacerbations lowered the threshold for AVP secretion to a level capable of inducing water intoxication in PIP, but not normonatremic, polydipsic patients (12), though again none of the recognized non-osmotic stimuli were associated with the acute exacerbation, leaving the mechanism unclear.

Evidence that an anterior hippocampal-induced stress diathesis is responsible for PIP findings

Other systematic investigations demonstrated that other physiologic influences on water balance regulation were intact (39), but also provided clues to the likely mechanism of the resetting and its relationship to the psychotic disorder (40). Thus, in one of the studies the severity of the resetting appeared to diminish as PIP subjects became habituated to the research setting, a finding resembling the physiologic response seen with other neurohormones whose levels increase due to the psychologic stress of acute hospitalization and fall back to normal with time. The finding did not make sense at first, since psychological stress normally has no consistent effect on AVP secretion, and the AVP set points did not vary in non-PIP patients (31). It soon became apparent, however, that PIP patients do not exhibit normal responses to psychological stress.

During this period, reports began accumulating that the volume of the anterior hippocampus (AH) was reduced in schizophrenia as part of a stress diathesis (increased vulnerability to adverse impact of stress on emotional stability) that underlay the psychotic disorder (Table 2). These structural findings were particularly prominent in patients with PIP (38). The rodent analogue of the AH (ventral hippocampus) restrains hypothalamic pituitary adrenal axis (HPAA) responses to psychological stress (41) via a relay in the anterior hypothalamus that terminates in the paraventricular nucleus (PVN) (Figure 1). Enhanced HPAA activity in PIP patients appeared consistent with dysfunction of this AH projection (38). Furthermore, other studies demonstrated the projection innervates adjacent PVN magnocellular neurons which secrete AVP and oxytocin into the peripheral circulation. A subsequent study showed that disrupting the pathway causes AVP levels to abnormally rise during psychological stress (42). The possibility this was relevant to neuroendocrine findings in PIP was substantiated by showing that disruption of the neurodevelopment of the ventral hippocampus, an animal model of the stress diathesis in schizophrenia, reproduced these findings (Figure 1) (43). Finally, a series of studies in humans provided direct support by showing that hippocampal modulation of HPA activity was markedly disrupted in PIP patients, and they exhibited enhanced AVP and HPAA responses to psychological but not physical stress. Most striking was that the AVP response replicated the effects of psychotic exacerbations on the AVP set point (44). Hence, the findings demonstrated that the AH pathology in PIP disrupts the normal restraint of neuroendocrine responses to psychological stress and can account for the psychosis-induced water intoxication in these patients (Figure 1).

Figure 1. Anterior hippocampal disruption of neuroendocrine secretion in hyponatremic polydipsic patients.

1) Neuroendocrine secretion from hypothalamic magnocellular (MAGN) neurons in the supraoptic (not shown) and paraventricular (PVN) nuclei release vasopressin (AVP) and oxytocin (OXY) directly into the peripheral circulation via the posterior pituitary (PP) where they enhance water resorption and lactation, respectively. Stress hormone secretagogues are released from adjacent parvocellular (PARV) PVN neurons into the hypophyseal circulation where they modulate cortisol secretion by inducing adrenocorticotropin (ACTH) release from the anterior pituitary (AP). 2) Projections from the anterior hippocampus relay in the PVN surround and terminate in the PVN where they normally restrain AVP and stress hormone activity during psychological stress, and, we speculate, modulate OXY release. 3) The hippocampal restraint of stress hormone activity is partly regulated by glucocorticoid negative feedback in the hippocampus which is also disrupted in the polydipsic group. 4) The dendrites on MAGN neurons release AVP and OXY directly into the brain. Dendritic OXY may account for most of OXY’s actions in the CNS including social behaviors and stress reactivity which occur in part by binding to the AH, amygdala and anterior hypothalamus. Dendritic secretion appears diminished in polydipsic patients and may contribute to their particularly impaired social functioning. Not shown is the putative disruption of AH modulation of dopamine activity in the NAC and its proposed enhancement of behavioral responses to stress and its impairment of the ability of coping behaviors to blunt these responses. Evidence supports the possibility that this disruption in conjunction with the altered AH influence on hypothalamic function underlies the polydipsia and other features of the psychotic disorder.

The link to the stress diathesis model of schizophrenia was further expanded by a study showing that another animal model of schizophrenia which relies on disrupting hippocampal development, disrupts the ability of coping behaviors to blunt psychologic stress responses while also increasing dopamine responses in the nucleus accumbens (NAC), the subcortical structure most frequently linked to psychotic symptoms (45). What was befuddling however was that the structural AH findings and neuroendocrine dysfunction (apart from the HPAA abnormalities) were not apparent in the normonatraemic polydipsic patients, raising the issue that perhaps the AH-induced stress diathesis was relevant only to the neuroendocrine findings and psychotic disorder but not the polydipsia in schizophrenia (38, 46).

Polydipsic patients without hyponatremia exhibit less severe, but otherwise similar, findings

A closer assessment of the neuroendocrine responses to the psychological stressor in the non-polydipsic patients and a more sensitive analysis of the structural findings resolved the confusion. First, responses to psychological stress in the polydipsic normonatraemic patients closely resembled those of the healthy controls while responses in non-polydipsic patients were blunted and resembled those of chronically stressed healthy controls with intact coping responses. Indeed, previous reports suggested that blunted HPAA responses were commonly seen in most schizophrenia patients and were predicted by individual patient’s coping behaviors consistent with intact responses to psychological stress, while other patients exhibited elevated HPAA responses and derived no benefit from their coping behaviors consistent with a stress diathesis (38).

A shape analysis of the AH helped reconcile the data. Polydipsic groups had deformations limited to the lateral AH surface, which is the surface which projects to the anterior hypothalamus, while deformations in the non-polydipsic patients were restricted to the medial surface (47). The deformations on the lateral surface were substantially larger in the PIP subset and predicted the intensity of their AVP responses, as well as the (non-elevated but not blunted) responses in the normonatremic polydipsics. Hence, the AH pathology was found in both polydipsic groups, and it seemed plausible that the more severe changes in the PIP patients accounted for their more severe water imbalance. This view was substantiated and further linked to the underlying psychotic illness when the investigation extended to oxytocin (48).

Oxytocin studies show AH pathology directly relevant to psychotic disorder in polydipsics as well

Oxytocin is a nonpeptide closely related to AVP and implicated in the pathophysiology of schizophrenia. Oxytocin secretion from the PVN also appears to be influenced by the same projection that restrains AVP and HPAA responses to stress, and diminished CNS oxytocin has been implicated in the impaired social behaviors in schizophrenics that are particularly impaired in the PIP subset (Figure 1) (38). Plasma concentrations of oxytocin were also predicted by the size of the lateral AH indentions and proportional to impaired social functioning in both polydipsic subsets, consistent with other data showing clinic features are predicted by differences in AH shape (48). This interpretation was further substantiated by a study showing deficits in social cognition were ameliorated by increasing CNS oxytocin levels in polydipsic but not non-polydipsic schizophrenic patients (38). Together these studies provided diverse evidence that AH pathology produces a stress diathesis that disrupts neuroendocrine regulation and contributes to the psychotic illness in polydipsic chronic psychotic patients. Table 2 summarizes the studies and observations that underlay the evolution of thinking about the pathophysiology of the water imbalance and its relationship to the underlying psychotic disorder, while Figure 1 summarizes the structural and functional circuit linking the neuroendocrine dysfunction to the water imbalance and psychiatric disorder.

Plausible model linking the polydipsia to the AH-induced stress diathesis

These studies did not, however, reveal how the stress diathesis could contribute to the primary polydipsia. A plausible explanation comes from studies showing psychological stress in the presence of enhanced NAC dopamine activity produces polydipsia (6) as well as an array of stereotypic behaviors common to the polydipsic subset (49). In particular, these behaviors closely resemble the scheduled-induced behaviors in mammals that arise from NAC dopamine hyperactivity (27) and are greatly enhanced by hippocampal lesions. The behaviors appear to reduce arousal associated with the schedule-induction paradigm, and the hippocampal lesions may perhaps promote these behaviors by impairing the efficacy of other coping behaviors (26,27,50). The validity of this animal model is supported by evidence that clozapine preferentially diminishes polydipsia and other stereotypic behaviors in both patients and animals (38). Hence, the animal model might recreate a disorder closely aligned to that seen in patients, i.e. increased drinking and other stereotypies that do not serve homeostatic needs but rather function as ‘primitive’ coping behaviors (26). Together these studies support that view that the AH-induced stress diathesis produces a distinct mental disorder in polydipsic schizophrenic patients. Further, they raise the possibility that the pathophysiology of their impaired water balance and psychotic illness are one and the same, and that only the downstream pathways differ.

Other psychiatric and medical conditions and general population

Others with PP also exhibit changes in the physiologic regulation of water intake that are in many cases related to emotional or, at least non-homeostatic, influences on water balance (Table 1). The PP in CWD is associated with a lower osmotic threshold and enhanced response to thirst (i.e. greater intake at any level of thirst). The latter is in turn associated with delayed satiation following water intake and appears attributable to impaired oropharyngeal regulation (5), a finding distinct from that in the polydipsic psychotic patients. Patients with dipsogenic DI also exhibit a low osmotic threshold for thirst but appear to have a higher osmotic set point for AVP and a higher basal plasma osmolality than those with CWD (7). Hypothalamic sarcoidosis seems to disrupt the normal congruence of the set points in AVP and desire for water (the latter generally 10 mOsm/Kg higher than the former thus blunting need for consume water) by an unknown mechanism (8).

Physiologic disruption of water balance may be a reflection of non-homeostatic influences associated with one appetite impacting another. As previously mentioned, the increased thirst in anorexia nervosa may be compensating for low food intake and diminishing hunger. Eating and drinking are mechanistically inter-related and are markedly influenced by the same environmental and especially social contexts (28,29). In addition to substituting for hunger, drinking may be a particularly effective means of addressing other emotional states not only because of the pleasurable effects of fluid ingestion but because of the intense non-specific arousal induced by thirst. In addition, the marked renal excretory capacity for water limits the acute physiologic impact of the excess intake. These factors together may promote learning to substitute drinking for more appropriate coping responses, particularly in the presence of altered hippocampal and ventral striatal functioning (11, 26). To the extent this only occurs with particular pathologic states versus being inherent properties of the neural systems is unknown.

Technological advances provide potential mechanisms at neural circuit level

Opportunity to characterize previously inaccessible emotional influences on CNS function

Technological advances within the past decade (e.g. optogenetics, chemo genetics, calcium imaging) have enabled isolation of specific neural structures, circuits and functions underlying appetitive and consumptive behaviors in mammals (51). The ability to identify, map and manipulate homogenous sets of neurons has disentangled a multitude of homeostatic and non-homeostatic influences, though the exact mechanisms and circuits by which most non-homeostatic influences, particularly those related to emotional stimuli, modulate appetitive behaviors have yet to be determined (51,52,53).

One or more non-homeostatic pathways are likely to contribute to primary polydipsia, and at least in the case of chronic psychosis, the underlying pathologic process appears likely to be relevant to the pathophysiology of the psychiatric condition as well. To the extent this is correct, clarification of the mechanism of the water imbalance could provide otherwise inaccessible insights into the pathophysiology and treatment of severe mental illness. Figure 2 illustrates homeostatic and non-homeostatic circuits which modulate intake and satiety. We briefly summarize what has been learned about influences on water balance regulation and what is not yet known but is likely to be revealed soon. Non-homeostatic components that appear most likely to prove to PP are colored purple.

Figure 2. Functional neurocircuitry modulating fluid intake and potentially relevant to primary polydipsia.

A. Homeostatic influences driving thirst and vasopressin release (hypernatremia, hypotension, hypovolemia) are conveyed via the peripheral blood stream to glutamatergic neurons () lamina terminalis nuclei (SFO, OVLT) outside the blood brain barrier and via relays in the brainstem which project to the third LT nucleus (MnPO). The MnPO integrates signals from the SFO and OVLT and is the primary structure which coordinates fluid intake, such that any output from glutamatergic MnPO neurons may be sufficient to over-ride satiety signals (solid purple efferents). B. Non-homeostatic influences that anticipate water intake and protect from overhydration act largely via gabaergic projections () in the LT nuclei. Of particular interest is the oropharyngeal projection which seems to regulate the dopamine release that occurs with swallowing and somehow can operate independently of satiation, likely through poorly understood projections to the lateral hypothalamic area (dashed purple line with ?). C. Non-homeostatic projections also promote intake in advance of need for water. These include eating where there are multiple overlapping pathways between the two appetitive systems. D. Efferents from the MnPO relay via the LHA and PVT to the cortical and subcortical structures that regulate drinking behavior. Both pathways appear capable of enhancing intake even in the presence of satiety signals. Potential pathways and mechanisms relevant to the effects of psychologic stress on intake are discussed in the text. Abbreviations: AH: Anterior hypothalamus; LHA: Lateral hypothalamic area; MnPO: median preoptic nucleus; NTS: nucleus of the solitary tract; OVLT: organum vasculosum of the lateral terminalis; PBN: parabrachial nucleus; PP: posterior pituitary; PVN: paraventricular nucleus; PVT: paraventricular thalamus; SCN:suprachiasmatic nucleus SFO: sub-fornical organ; SON:supraoptic nucleus.

The lamina terminalis (LT) coordinates water intake

Two lamina terminalis (LT) nuclei outside the blood brain barrier, and a third LT nucleus within the BBB, integrate the multitude of influences on water ingestion (Figure 2A) via specific subsets of neurons dedicated to water balance. The three nuclei are directly tied to the major systemic influences that drive thirst and AVP secretion (blood tonicity, volume and pressure) (54). Specifically, a distinct subset of glutamatergic neurons in the sub-fornical organ (SFO) bind angiotensin II which is peripherally produced in response to hypovolemia/ hypotension while a distinct subset of glutamatergic neurons in the organum vasculosum of the lamina terminalis (OVLT) are stimulated via adjacent osmoreceptors by increases in blood tonicity/hypernatremia. The third nucleus, the median preoptic nucleus (MnPO), receives afferents from the parabrachial nucleus (PBN) which convey homeostatic signals from cardiac baroreceptors relevant to maintaining cardiac output. MnPO glutamatergic neurons also integrate the signals from the SFO and OVLT that drive thirst and directly or indirectly influence the homeostatic and non-homeostatic signals that diminish and ultimately terminate the desire for water. Activity of this specific subset of MnPO glutamatergic projections is the preeminent driver regulating water intake and appears responsible for the aversive drive that is ultimately perceived as thirst. The MnPO also tracks the total volume of consumed liquid (55) and determines the extent CNS resources are dedicated to finding and consuming water (56). Thus, any factor that pathologically activates MnPO efferents despite satiety signals appears capable of inducing primary polydipsia (58).

Pre-systemic and other non-homeostatic influences on LT activity also modulate water intake

Pre-systemic modulators of water intake anticipate satiation as well as deficits in water balance, and largely act via gabaergic neurons in the three LT nuclei (Figure 2B). In particular, the MnPO distinguishes fluids from solids, and, when fluids are detected, immediately inhibits intake by diminishing SFO glutamatergic neurons (58). This inhibitory effect is initiated by oropharyngeal stimulation (e.g. gulping) and is non-specific to the type of fluid, i.e. even inhibiting fluid intake (and AVP release) when the ingested fluid is hypertonic. The circuit is relayed via the vagus nerve to the nucleus of the solitary tract (NTS) and parabrachial nucleus. Of potential interest is that oxytocin secreting neurons in the PBN appear to be a critical part of this inhibitory circuit, which is also activated by a set of oxytocin secreting neurons in the PVN (69), raising the possibility that deficient oxytocin activity, such as that seen in polydipsic psychotic patients, may contribute to the polydipsia. Downstream from the oropharynx the GI tract contains osmoreceptors and provides a more specific indicator of the tonicity of ingested water that is conveyed to SFO gabaergic neurons which in turn dampen SFO glutamatergic activity.

The two pathways appear to contribute to satiety in a different manner, and interestingly only the first appears strongly linked to NAC dopamine activity that likely contributes to the reduction of the aversive drive to drink (59). The associated dopamine activity that accompanies drive reduction has a marked effect on shaping drinking behavior, and likely accounts for why the specific fluid consumed while thirsty is much more likely to be selected in a less thirsty state (11). Of interest is that the increase in dopamine activity is not recreated by stimulating the inhibitory pathway in the LT which reduces drinking behavior in response to oropharyngeal stimulation. (59). This in turn suggests a mechanism by which ongoing drinking could be rewarding without necessarily diminishing fluid intake perhaps by bypassing the MnPO and directly innervating the lateral hypothalamic area (LHA) which is part of the circuit that regulates the aversive drive (see below). Alternatively, it provides a mechanism by which drinking might reduce aversive states of arousal that have nothing to do with fluid deficits (26).

Other pathways anticipate satiation even before any fluid has been ingested. Thus, AVP secretion in thirsty rats is dampened simply by cues of water availability that involve learning, and hence rely on hippocampal and NAC activity (60). The pathway is unknown but is a powerful demonstration of how non-homeostatic cognitive factors impact water balance. Other effects of drinking on dopamine activity appear to be directly linked to the pleasurable effects of water intake when dehydrated and appear to be driven by taste receptors (not shown) (61). While clearly implicated in non-homeostatic regulation of water balance, much more remains to be learned about how motivated behaviors and learning shaped by dopamine activity interact with innate and cognitively derived factors to determine drinking behavior (62) as well as their dysregulation (see below for example on eating) (26).

Stimulatory effects on SFO glutamatergic neurons also occur in anticipation of water deficits and provide another potential mechanism for excessive non-homeostatic water intake (Figure 2C). Thus, eating generates prandial drinking; an example of food intake directly modulating water intake (19). Analogous effects on intake arise from a projection from the suprachiasmatic nucleus (SCN) to OVLT glutamatergic neurons which increase water intake in rodents prior to sleep.

Two major efferent projections from MnPO glutamergic neurons

MnPO glutamatergic thirst neurons project to multiple subcortical sites, but at present the lateral hypothalamic area (LHA) and paraventricular thalamic (PVT) nucleus appear most relevant to water balance (62) (Figure 2D). LHA and PVT neurons associated with water balance project to the cerebral cortex, particularly the insula and anterior cingulate gyri and appear to produce the conscious sensation of thirst and modulate the affective tone, arousal and decision-making processes that further contribute to drinking behavior (29). The extent that these downstream modulatory influences operate independently of the MnPO is unknown, but obviously likely relevant to PP. Once satiety signals have terminated the MnPO glutamatergic efferent activity, drinking itself becomes aversive and swallowing liquids is activity inhibited (not shown) (63). Impairment in this function could also theoretically contribute or at least promote non-regulatory drinking.

For decades it has been recognized non-specific activation of the LHA produces both ingestive and non-ingestive behaviors resembling the stereotypic behaviors seen in polydipsic schizophrenic patients and animals exhibiting schedule-induced behaviors. A distinct subset of gabaergic LHA neurotensin neurons, however, is specific to driving water intake and this and similar recent studies have helped clarify much of the confusion regarding the specificity of the LHA in appetive behaviors (53). The function of the LHA and PVT projections differs. Thus, the LHA projection appears to convey the aversive signal that drives thirst, since activation of the dipsogenic MnPO to LH, but not the dipsogenic MnPO to PVT, pathway causes rodents to lever press to stop the activation (a core characteristic of aversive stimuli). Less is known about the specific influences of the PVT on intake, though it may be responsible for conveying signals from the MnPO to the cortex that are distinct from the aversive drive yet are capable of driving drinking in the presence of intact satiety signals (57).

Neurocircuits modulating food intake may provide insights into mechanisms of polydipsia

The components of the neural systems and the underlying principles of brain organization (e.g. top-down cognitive vs bottom-up reward based; appetitive drive vs. satiation; anticipatory vs systemic influences; learned vs. innate responses; hierarchical organization) that determine fluid and food intake share many features, although the specific stimuli, CNS nuclei, circuit connections and time scales often differ (64). In addition, energy and water balance modulators directly influence each other, providing other mechanisms by which pathways regulating non-homeostatic (e.g. affective) influences on food may directly influence fluid intake (11,28). This is particularly relevant since because more is currently known about the neural mechanisms underlying food than fluid intake.

Knowledge of the competing roles of reward-based and cognitive influences on non-homeostatic influences on food intake are instructive and provide potential targets for diminishing the impact of psychological stress on appetitive behaviors and psychopathology. Thus, the PVT conveys appetitive drives from the hypothalamus to both the cortex and NAC, which in turn appear to gate the influence of cognitive and reward-driven factors on eating. Some animals are more likely than others to become ‘stimulus bound’ when stressed and react to cues (unconditioned stimuli) as if they were the reward itself, as well as exhibit deficits in cognitive control normally generated by the prefrontal cortex. Food cues in these animals during hunger induce activity in the PVT that correlates with NAC dopamine activity, while those animals not ‘stimulus bound’ (‘goal-oriented’) show PVT activity that correlates within PFC activity (65). Furthermore, stimulating the PFC projection to the PVT in ‘stimulus bound’ animals reduces their cue focused behavior, while turning off the same projection in goal-directed animals makes them appear stimulus bound. Together the results identify the PVT as a target that can potentially reduce the impact of arousing stimuli on primitive coping behaviors while promoting greater cognitive controls.

Can translational neuroscience reveal how a stress diathesis induces water imbalance and psychosis?

The life-threatening water imbalance in psychotic disorders is caused by an anterior hippocampal (AH) induced stress-diathesis that can be reproduced in animal models, and involves phylogenetically-preserved pathways that appear likely to involve one or more of the hypothalamic non-homeostatic circuits illustrated in Figure 2. Disruptions in this circuitry activity may convey the adverse effects of psychological stress on water balance as well as other hypothalamically-modulated functions central to psychosis. If so, the circuits are likely discoverable and may provide an otherwise inaccessible means of isolating the pathophysiology for severe mental illness (38).

While a pathway underlying AH modulation of psychological stress on neuroendocrine activity has been identified, we do not know the analogous pathway conveying emotional influences on water balance. We know something however of how the AH modulates the impact of anxiety on food intake. The ventral hippocampus (the rodent equivalent of the AH) appears to directly modulate the influence of anxiety on appetitive behaviors via a projection to the LHA (66). Specific cells signal levels of anxiety that in turn regulate approach and avoidance behaviors. There is also pilot evidence that the AH-hypothalamic circuit implicated in the impact of psychologic stress on neuroendocrine function and behavior (67) is disrupted in polydipsic patients and may provide a signature of the circuit dysfunction. Thus pilot data indicate that the resting state functional connectivity of the AH-hypothalamic circuit differs significantly in polydipsic schizophrenic patients from non-polydipsic schizophrenics and healthy controls (68) in proportion to the severity of the polydipsic patients’ social deficits. Initial translational studies, could, for instance, explore if non-homeostatic circuit influences on water imbalance are altered in animal models of schizophrenia and in animal models of schedule-induced behaviors, whether these alterations are similar and involve changes in AH modulation of LHA anxiety neurons, and if the alterations are linked to changes in NAC dopaminergic reactivity.

Summary

Primary polydipsia (PP) is found in a broad range of people, including those with the most severe mental disorders and health enthusiasts. For some there are effective treatments, though these may carry significant risks and require more rigorous evaluation particularly to establish their efficacy in groups that vary from the original subject population. For many the most critical issue is simply to avoid medications and to actively treat other disorders that could impair excretory capacity and place patients at risk of symptomatic dilutional hyponatremia. Unlike secondary polydipsia, most with PP do not have increased thirst, and even fewer are likely to have disruptions in the homeostatic regulation of water balance. Many of the different groupings of PP, however, show distinct physiologic alterations in water balance that likely reflect the impact of cognitive and emotional factors on water intake and satiation. Hypothalamic pathways modulating these non-homeostatic influences have recently been identified in laboratory animals, are likely to be phylogenetically preserved, and may help isolate the mechanisms of the PP and how these pathologic processes contribute to psychiatric illnesses.

BEST PRACTICE POINTS.

• Medications that can impair water excretion, thereby placing PP patients at risk of life-threatening water intoxication, must be avoided.

• PP must therefore be definitively excluded before initiating treatment for DI. This can be difficult in all but those with the most severe psychiatric disorders (e.g. PIP)

• Unlike others with PP, desmopressin can be an effective treatment for Dipsogenic DI.

• Clozapine, unlike other antipsychotic agents, reduces hyponatremia in PIP patients and may reduce polydipsia in normonatremic polydipsic psychotic patients. Risks of clozapine are substantial. Vaptans are also effective in psychotic patients with symptomatic hyponatremia but again risks must be taken into account.

• Primary polydipsia in compulsive water drinkers frequently responds to treatment of the underlying psychiatric illness (e.g depression, anxiety, OCD, anorexia nervosa). Other populations with PP are often resistant to treatment.

RESEARCH AGENDA.

• Establish if clozapine diminishes PP in normonatremic polydipsic psychotic patients with acceptable risks.

• Determine what behavioral treatments are effective for the different catagories of PP.

• Leverage the increased understanding of neurocircuit mechanisms of non-homeostatic influences on water intake, examine how psychologic stress and disruption of ventral striatal dopamine regulation contribute to increased intake/diminished satiety.

• Determine if and how non-homeostatic emotional influences on water imbalance also contribute to the associated psychiatric disorders, particularly in polydipsic chronic psychotic patients whose anterior hippocampal-induced stress diathesis appears to overlap structures and circuits implicated in both psychosis and primary polydipsia.