Best Practice & Research Clinical Endocrinology & Metabolism xxx (xxxx) xxx

Contents lists available at ScienceDirect

Best Practice & Research Clinical

Endocrinology & Metabolism
journal homepage: www.elsevier.com/locate/beem

Acquired forms of central diabetes insipidus:

Mechanisms of disease
Joseph G. Verbalis, MD, Professor of Medicine and Chief,
Endocrinology and Metabolism
Georgetown University, Washington, DC, 20007 USA

a r t i c l e i n f o
Most cases of acquired central diabetes insipidus are caused by
Article history: destruction of the neurohypophysis by: 1) anatomic lesions that
Available online xxx destroy the vasopressin neurons by pressure or infiltration, 2)
damage to the vasopressin neurons by surgery or head trauma, and
Keywords: 3) autoimmune destruction of the vasopressin neurons. Because
adipsic diabetes insipidus the vasopressin neurons are located in the hypothalamus, lesions
central diabetes insipidus confined to the sella turcica generally do not cause diabetes
neurohypophysis insipidus because the posterior pituitary is simply the site of the
neuroinfundibulohypophysitis axon terminals that secrete vasopressin into the bloodstream. In
osmoreceptors
addition, the capacity of the neurohypophysis to synthesize vaso-
vasopressin
pressin is greatly in excess of the body's needs, and destruction of
80e90% of the hypothalamic vasopressin neurons is required to
produce diabetes insipidus. As a result, even large lesions in the
sellar and suprasellar area generally are not associated with
impaired water homeostasis until they are surgically resected.
Regardless of the etiology of central diabetes insipidus, deficient or
absent vasopressin secretion causes impaired urine concentration
with resultant polyuria. In most cases, secondary polydipsia is able
to maintain water homeostasis at the expense of frequent thirst
and drinking. However, destruction of the osmoreceptors in the
anterior hypothalamus that regulate vasopressin neuronal activity
causes a loss of thirst as well as vasopressin section, leading to
severe chronic dehydration and hyperosmolality. Vasopressin
deficiency also leads to down-regulation of the synthesis of
aquaporin-2 water channels in the kidney collecting duct principal
cells, causing a secondary nephrogenic diabetes insipidus. As a
result, several days of vasopressin administration are required to
achieve maximal urine concentration in patients with CDI.
Consequently, the presentation of patients with central diabetes

E-mail address: verbalis@georgetown.edu.

https://doi.org/10.1016/j.beem.2020.101449
1521-690X/© 2020 Published by Elsevier Ltd.

Please cite this article as: Verbalis JGAcquired forms of central diabetes insipidus: Mechanisms of disease,
Best Practice & Research Clinical Endocrinology & Metabolism, https://doi.org/10.1016/
j.beem.2020.101449
2 J.G. Verbalis / Best Practice & Research Clinical Endocrinology & Metabolism xxx (xxxx) xxx

insipidus can vary greatly, depending on the size and location of

the lesion, the magnitude of trauma to the neurohypophysis, the
degree of destruction of the vasopressin neurons, and the presence
of other hormonal deficits from damage to the anterior pituitary.
© 2020 Published by Elsevier Ltd.

Etiologies of acquired central diabetes insipidus

Central diabetes insipidus (CDI) is caused by inadequate secretion of arginine vasopressin (AVP)
from the posterior pituitary gland in response to osmotic stimulation. In most cases, this is due to
destruction of the neurohypophysis by a variety of acquired anatomic lesions that destroy or damage
the neurohypophysis by pressure or infiltration (Table 1). The neurohypophysis includes the magno-
cellular neurons that synthesize AVP and oxytocin in the hypothalamic paraventricular and supraoptic
nuclei, the axons that transport the synthesized AVP down the pituitary stalk, and the axon terminals of
the posterior pituitary. The severity of the resulting hypotonic diuresis depends on the degree of
destruction of the neurohypophysis, leading to either complete or partial deficiency of AVP secretion.
Despite the wide variety of lesions that can potentially cause CDI (Table 1), it is much more common
to not have CDI in the presence of sellar and/or suprasellar lesions than to actually produce the syn-
drome. This apparent inconsistency can be understood by considering several common principles of
neurohypophyseal physiology and pathophysiology that are relevant to all of these etiologies. The first
is that the synthesis of AVP occurs in the hypothalamus (Fig. 1); the posterior pituitary simply

Table 1
Etiologies of hypotonic polyuria.

Central (neurogenic) diabetes insipidus

Congenital (congenital malformations, autosomal dominant arginine vasopressin (AVP)eneurophysin gene mutations)
Drug-/toxin-induced (ethanol, diphenylhydantoin, snake venom)
Granulomatous (histiocytosis, sarcoidosis)
Neoplastic (craniopharyngioma, germinoma, lymphoma, leukemia, meningioma, pituitary tumor, metastases)
Infectious (meningitis, tuberculosis, encephalitis)
Inflammatory/autoimmune (lymphocytic infundibuloneurohypophysitis)
Trauma (neurosurgery, deceleration injury)
Vascular (cerebral hemorrhage or infarction, brain death)
Idiopathic
Osmoreceptor dysfunction (Adipsic diabetes insipidus)
Granulomatous (histiocytosis, sarcoidosis)
Neoplastic (craniopharyngioma, pinealoma, meningioma, metastases)
Vascular (anterior communicating artery aneurysm/ligation, intrahypothalamic hemorrhage)
Other (hydrocephalus, ventricular/suprasellar cyst, trauma, degenerative diseases)
Idiopathic
Increased AVP metabolism
Pregnancy
Nephrogenic diabetes insipidus
Congenital (X-linked recessive AVP V2 receptor gene mutations, autosomal recessive or dominant aquaporin-2 water
channel gene mutations)
Drug-induced (demeclocycline, lithium, cisplatin, methoxyflurane)
Hypercalcemia
Hypokalemia
Infiltrating lesions (sarcoidosis, amyloidosis)
Vascular (sickle cell anemia)
Mechanical (polycystic kidney disease, bilateral ureteral obstruction)
Solute diuresis (glucose, mannitol, sodium, radiocontrast dyes)
Idiopathic
Primary polydipsia
Psychogenic (schizophrenia, obsessive-compulsive behaviors)
Dipsogenic (downward resetting of thirst threshold, idiopathic or similar lesions as with central DI)

Please cite this article as: Verbalis JGAcquired forms of central diabetes insipidus: Mechanisms of disease,
Best Practice & Research Clinical Endocrinology & Metabolism, https://doi.org/10.1016/
j.beem.2020.101449
 J.G. Verbalis / Best Practice & Research Clinical Endocrinology & Metabolism xxx (xxxx) xxx 3

Fig. 1. Summary of the anterior hypothalamic pathways that mediate secretion of arginine vasopressin (AVP) and oxytocin (OT). The
vascular organ of the lamina terminalis (OVLT) is especially sensitive to hyperosmolality. Hyperosmolality also activates other
neurons in the anterior hypothalamus, such as those in the subfornical organ (SFO) and median preoptic nucleus (MnPO), and
magnocellular neurons, which are intrinsically osmosensitive. Circulating angiotensin II (Ang II) activates neurons of the SFO, an
essential site of Ang II action, as well as cells throughout the lamina terminalis and MnPO. In response to hyperosmolality or Ang II,
projections from the SFO and OVLT to the MnPO activate excitatory and inhibitory interneurons that project to the supraoptic
nucleus (SON) and paraventricular nucleus (PVN) to modulate direct inputs to these areas from the circumventricular organs.
Cholecystokinin (CCK) acts primarily on gastric vagal afferents that terminate in the nucleus of the solitary tract (NST), but at higher
doses it can also act at the area postrema (AP). Although neurons are apparently activated in the ventrolateral medulla (VLM) and
NST, most neurohypophyseal secretion appears to be stimulated by monosynaptic projections from A2-C2 cells, and possibly also
non-catecholaminergic somatostatin-inhibin B cells, of the NST. Baroreceptor-mediated stimuli, such as hypovolemia and hypo-
tension, are more complex. The major projection to magnocellular AVP neurons appears to arise from A1 cells of the VLM that are
activated by excitatory interneurons from the NST. Other areas, such as the parabrachial nucleus (PBN), may contribute multisynaptic
projections. Cranial nerves IX and X, which terminate in the NST, also contribute input to magnocellular AVP neurons. AC, Anterior
commissure; OC, optic chiasm; PIT, anterior pituitary (From Stricker EM, Verbalis JG: Water intake and body fluids. In Squire LR,
Bloom FE, McConnell SK et al., editors: Fundamental neuroscience, San Diego, 2003, Academic Press, pp 1011e1029.)

represents the site of storage and secretion of the neurosecretory granules that contain AVP. Conse-
quently, lesions contained within the sella turcica that destroy only the posterior pituitary generally do
not cause CDI because the cell bodies of the magnocellular neurons that synthesize AVP remain intact
and the site of release of AVP shifts more superiorly, typically into the blood vessels of the median
eminence at the base of the brain. Perhaps the best examples of this phenomenon are large pituitary
macroadenomas that destroy the anterior and posterior pituitary. DI is a distinctly unusual presen-
tation for such pituitary adenomas, because destruction of the posterior pituitary by such slowly
enlarging intrasellar lesions merely destroys the nerve terminals, but not the cell bodies, of the AVP
neurons. As this occurs, the site of release of AVP shifts more superiorly to the pituitary stalk and
median eminence. Sometimes this can be detected on non-contrast MRI as a shift of the pituitary
“bright spot” more superiorly to the level of the infundibulum or median eminence [1], but often it this
process is too diffuse to be detected in this manner. The occurrence of DI from a pituitary adenoma is so
uncommon, even with macroadenomas that completely obliterate sellar contents sufficiently to cause
panhypopituitarism, that its presence should lead to consideration of alternative diagnoses, such as
craniopharyngioma, which often causes damage to the median eminence by virtue of adherence of the
capsule to the base of the hypothalamus, more rapidly enlarging sellar/suprasellar masses that don't
allow sufficient time for shifting the site of AVP release more superiorly (e.g., metastatic lesions), or
granulomatous disease with more diffuse hypothalamic involvement (e.g., sarcoidosis, histiocytosis).

Please cite this article as: Verbalis JGAcquired forms of central diabetes insipidus: Mechanisms of disease,
Best Practice & Research Clinical Endocrinology & Metabolism, https://doi.org/10.1016/
j.beem.2020.101449
4 J.G. Verbalis / Best Practice & Research Clinical Endocrinology & Metabolism xxx (xxxx) xxx

With very large pituitary adenomas that cause ACTH deficiency, it is actually more likely that patients
will present with hypoosmolality from an SIADH-like picture because of the impaired free water
excretion that accompanies hypocortisolism.
A second general principle is that the capacity of the neurohypophysis to synthesize AVP is greatly
in excess of the body's daily needs for maintenance of water homeostasis. Carefully controlled studies
of surgical section of the pituitary stalk in dogs have clearly demonstrated that destruction of 80e90%
of the magnocellular neurons in the hypothalamus is required to produce polyuria and polydipsia in
this species [2]. Thus, even lesions that do cause destruction of the AVP magnocellular neuron cell
bodies must cause a large degree of destruction to produce CDI. The most illustrative example of this is
surgical section of the pituitary stalk in humans. Necropsy studies of these patients have revealed
atrophy of the posterior pituitary and loss of the magnocellular neurons in the hypothalamus [3]. This
loss of magnocellular cells presumably results from retrograde degeneration of neurons whose axons
were cut during surgery. As is generally true for all neurons, the likelihood of retrograde neuronal
degeneration depends on the proximity of the axotomy, in this case section of the pituitary stalk, to the
cell body of the neuron. This was shown clearly in studies of human subjects in whom section of the
pituitary stalk at the level of the diaphragm sellae (i.e., a “low” stalk section) produced transient but not
permanent DI, whereas section at the level of the infundibulum (i.e., a “high” stalk section) was
required to cause permanent DI in most cases [4].
The development of DI following surgical or traumatic injury to the neurohypophysis represents a
unique situation and can follow several different, well-defined patterns. In some patients, polyuria
develops 1e4 days after pituitary surgery and resolves spontaneously. The incidence of transient DI
varies with the surgical approach to the sella, from 5.1 ± 3.5% with transnasal endoscopic surgery to
25.2 ± 13% with transnasal microscopic surgery [5]. Less often, the DI is permanent. The incidence of
permanent DI is low following resection of pituitary microadenomas (0.8e1.8% [5]) and macro-
adenomas (3.8% [6]), but is more marked after resection of suprasellar lesions such as craniophar-
yngiomas (from 10% with subtotal resection to 25% with gross total resection of the lesion [7]). Most
interestingly, a “triphasic” response can occur because of pituitary stalk transection in which an initial
phase of DI is followed by a brief period of inappropriate AVP secretion and then a recurrence of DI. This
is discussed in more detail under Pathophysiology.
One of the most severe forms of CDI that occurs in a small subset of patients is osmoreceptor
dysfunction, also known as “adipsic DI”. There is an extensive literature in animals indicating that the
primary osmoreceptors that control AVP secretion and thirst are located in the anterior hypothalamus;
lesions of this region in animals, called the “AV3V” area, cause hyperosmolality through a combination
of impaired thirst and osmotically stimulated AVP secretion [8,9]. Initial reports in humans described
this syndrome as essential hypernatremia [10], and subsequent studies used the term adipsic hyper-
natremia in recognition of the profound thirst deficits found in most of the patients [11]. Based on the
known pathophysiology, all of these syndromes can be grouped together as disorders of “osmoreceptor
dysfunction” [12]. Although the pathologies responsible for this condition can be quite varied, all of the
cases reported to date have been due to various degrees of osmoreceptor destruction associated with a
variety of different brain lesions as summarized in Table 1. Many of these are the same types of lesions
that can cause CDI, but in contrast to CDI, these lesions usually occur more rostrally in the hypothal-
amus, consistent with the anterior hypothalamic location of the primary osmoreceptor cells (Fig. 1).
One lesion that is unique to this disorder is an anterior communicating cerebral artery aneurysm.
Because the small arterioles that feed the anterior wall of the third ventricle originate from the anterior
communicating cerebral artery, an aneurysm in this region [13], but more often following surgical
repair of such an aneurysm that may entail ligation of the anterior communicating artery [14], pro-
duces infarction of the part of the hypothalamus containing the osmoreceptor cells.
Next to destructive and traumatic lesions of the pituitary and hypothalamus, idiopathic forms of AVP
deficiency represent a large pathogenic category in both adults and children. A study in children
revealed that over half (54%) of all cases of CDI were classified as idiopathic [15]. These patients typically
do not have historical or clinical evidence of any injury or disease that can be linked to their DI, and MRI
of the pituitary-hypothalamic area generally reveals no abnormality other than absence of the posterior
pituitary bright spot and sometimes-varying degrees of thickening of the pituitary stalk. Several lines of
evidence have suggested that many of these patients may have had an autoimmune destruction of the

Please cite this article as: Verbalis JGAcquired forms of central diabetes insipidus: Mechanisms of disease,
Best Practice & Research Clinical Endocrinology & Metabolism, https://doi.org/10.1016/
j.beem.2020.101449
 J.G. Verbalis / Best Practice & Research Clinical Endocrinology & Metabolism xxx (xxxx) xxx 5

neurohypophysis to account for their DI. First, the entity of lymphocytic infundibulo-neurohypophysitis
has been documented to be present in a subset of patients with idiopathic DI [16,17]. Lymphocytic
infiltration of the anterior pituitary, lymphocytic hypophysitis, has been recognized as a cause of
anterior pituitary deficiency for many years, but it was not until an autopsy called attention to a similar
finding in the posterior pituitary of a patient with DI that this pathology was recognized to occur in the
neurohypophysis as well. Since that initial report, a number of similar cases have been described,
including cases in the postpartum period, which is characteristic of lymphocytic hypophysitis. With the
advent of magnetic resonance imaging, lymphocytic infundibulo-neurohypophysitis can diagnosed
based on the appearance of a thickened stalk and/or absence of the posterior pituitary bright spot on
non-contrast T1-weighted images in a patient with new onset CDI. The enlargement of the stalk can so
mimic a neoplastic process that some of these patients were operated based on a suspicion of a pituitary
tumor, but with the finding of lymphocytic infiltration of the pituitary stalk. Since then, a number of
patients with a suspicion of infundibulo-neurohypophysitis and no other obvious cause of DI have been
followed and have shown regression of the thickened pituitary stalk over time [18]. Several cases have
been reported with the coexistence of CDI and adenohypophysitis, and these presumably represent
cases of combined lymphocytic infundibulo-neurohypophysitis and hypophysitis [19]. A second line of
evidence supporting an autoimmune etiology in many cases of idiopathic DI stems is the finding of AVP
antibodies in the serum of an many as 1/3 of patients with idiopathic DI and 2/3 of those with Lang-
erhans cell histiocytosis X, but not in patients with DI caused by tumors [20]. A more recently recognized
form of infundibulo-neurohypophysitis occurs in middle-aged to older men and is associated with
immunoglobulin G4 (IgG4)-related systemic disease [21]. In this disease, various organs, especially the
pancreas, are infiltrated with IgG4 plasma cells, and neurohypophysitis is only one manifestation of a
multiorgan disease that may include other endocrine glands.

Pathophysiology of acquired central diabetes insipidus

The normal inverse relationship between urine volume and urine osmolality means that initial
decreases in maximal AVP secretion will not cause an increase in urine volume sufficient to be detected
clinically by polyuria. In general, basal AVP secretion must fall to <10e20% of normal before basal urine
osmolality decreases to <300 mOsm/kg H2O and urine flow increases to symptomatic levels (i.e.,
>50 ml/kg BW/d). This resulting loss of body water produces a slight rise in plasma osmolality that
stimulates thirst and induces a compensatory polydipsia. The resultant increase in water intake re-
stores balance with urine output and stabilizes the osmolality of body fluids at a new, slightly higher
but still normal level. As the AVP deficit increases, this new steady state level of plasma osmolality
approximates the osmotic threshold for thirst (Fig. 2). It is important to recognize that the deficiency of
AVP need not be complete for polyuria and polydipsia to occur; it is only necessary that the maximal
plasma AVP concentration achievable at or below the osmotic threshold for thirst is inadequate to
concentrate the urine [22]. The degree of neurohypophysial destruction at which such failure occurs
varies considerably from person to person, largely because of individual hereditary differences in the
set point and sensitivity of the osmoregulatory system [23]. In general, functional tests of AVP levels in
patients with DI of variable severity, duration, and cause indicate that AVP secretory capacity must be
reduced by at least 75e80% for significant polyuria to occur, which also agrees with neuroanatomical
studies of cell loss in the supraoptic nuclei of dogs with experimental pituitary stalk section [2], and of
patients who had undergone pituitary surgery [3].
Because renal mechanisms for sodium conservation are unimpaired with decreased or absent AVP
secretion, there is no accompanying sodium deficiency. Although untreated DI can lead both to
hyperosmolality and volume depletion, until the water losses become severe, volume depletion is
minimized by osmotic shifts of water from the intracellular fluid (ICF) compartment to the more
osmotically concentrated extracellular fluid (ECF) compartment. This phenomenon is not as evident
following increases in ECF sodium concentration ([Naþ]), since such osmotic shifts result in a slower
increase in the serum [Naþ] than would otherwise occur. However, when non-sodium solutes such as
mannitol are infused, this effect is more obvious due to the progressive dilutional decrease in serum
[Naþ] caused by translocation of intracellular water to the ECF compartment. Because patients with DI
do not have impaired urine sodium conservation, the ECF volume is generally not markedly decreased

Please cite this article as: Verbalis JGAcquired forms of central diabetes insipidus: Mechanisms of disease,
Best Practice & Research Clinical Endocrinology & Metabolism, https://doi.org/10.1016/
j.beem.2020.101449
6 J.G. Verbalis / Best Practice & Research Clinical Endocrinology & Metabolism xxx (xxxx) xxx

Fig. 2. Relationship of plasma osmolality, plasma AVP concentrations, urine osmolality, and urine volume in humans. The osmotic
threshold for AVP secretion defines the point at which urine concentration begins to increase, but the osmotic threshold for thirst is
somewhat higher and approximates the point at which maximal urine concentration has already been achieved. Note also that
because of the inverse relation between urine osmolalty and urine volume, changes in plasma AVP concentrations have much larger
effects on urine volume at low plasma AVP concentrations than at high plasma AVP concentrations (Adapted from Robinson AG:
Disorders of antidiuretic hormone secretion. J Clin Endocrinol Metab 14:55e88, 1985.)

and regulatory mechanisms for maintenance of osmotic homeostasis are primarily activated: stimu-
lation of thirst and AVP secretion (to whatever degree the neurohypophysis is still able to secrete AVP).
In cases where AVP secretion is totally absent (complete DI), patients are dependent entirely on water
intake for maintenance of water balance. However, in cases where some residual capacity to secrete
AVP remains (partial DI), plasma osmolality can eventually reach levels that allow moderate degrees of
urinary concentration (e.g., 200e500 mOsm/kg H2O).
A characteristic triphasic response can occur following transection or damage to the pituitary stalk
(Fig. 3A) [4]. The initial DI (first phase) is due to axon shock and lack of function of the damaged neurons.
This phase lasts from several hours to several days, and is followed by a second, antidiuretic phase that is
due to the uncontrolled release of AVP from the disconnected and degenerating posterior pituitary gland
[24]. Overly aggressive administration of fluids during this second phase does not suppress AVP
secretion and can lead to hyponatremia. The antidiuresis can last 2e14 days, after which DI recurs
following depletion of the AVP from the degenerating posterior pituitary gland (third phase) [25].
Transient hyponatremia without preceding or subsequent DI has been reported following trans-
sphenoidal surgery for pituitary microadenomas [26], which generally occurs 5e10 days post-
operatively. The incidence may be as high as 30% when such patients are carefully followed, although
majority of cases are mild and self-limited [27,28]. This occurs because of inappropriate AVP secretion
via the same mechanism as in the triphasic response, except that in these cases only the second phase
occurs (“isolated second phase”) because the initial neural lobe/pituitary stalk damage is not sufficient to
impair AVP secretion sufficiently to produce clinical manifestations of DI (Fig. 3B [29]) [30].
Once a deficiency of AVP secretion has been present for more than a few days or weeks, it rarely
improves even if the underlying cause of the neurohypophysial destruction is eliminated. The major
exception to this occurs in patients with postoperative DI, where spontaneous resolution usually occurs.
Although recovery from CDI that persists more than several weeks postoperatively is less common,
nonetheless well-documented cases of long-term recovery have been reported [25]. The reason for
amelioration and resolution is apparent from pathological and histological examination of neurohy-
pophyseal tissue following pituitary stalk section [31]. Neurohypophyseal neurons that have intact
perikarya are able to regenerate axons and form new nerve terminal endings capable of releasing AVP

Please cite this article as: Verbalis JGAcquired forms of central diabetes insipidus: Mechanisms of disease,
Best Practice & Research Clinical Endocrinology & Metabolism, https://doi.org/10.1016/
j.beem.2020.101449
 J.G. Verbalis / Best Practice & Research Clinical Endocrinology & Metabolism xxx (xxxx) xxx 7

Fig. 3. Mechanisms underlying the pathophysiology of the triphasic pattern of diabetes insipidus (DI) and the isolated second phase.
A, In the triphasic response, the first phase of DI is initiated following a partial or complete pituitary stalk section, which severs the
connections between the AVP neuronal cell bodies in the hypothalamus and nerve terminals in the posterior pituitary gland, thus
preventing stimulated AVP secretion (1 ). This is followed in several days by the second phase of SIADH, which is caused by un-
controlled release of AVP into the bloodstream from the degenerating nerve terminals in the posterior pituitary (2 ). After all of the
AVP stored in the posterior pituitary gland has been released, the third phase of DI returns if more than 80%e90% of the AVP
neuronal cell bodies in the hypothalamus have undergone retrograde degeneration (3 ). B, In the isolated second phase, the pituitary
stalk is injured, but not completely cut. Although the maximum AVP secretory response will be diminished as a result of the stalk
injury, DI will not result if the injury leaves intact at least 10%e20% of the nerve fibers connecting the AVP neuronal cell bodies in the
hypothalamus to the nerve terminals in the posterior pituitary gland (1 ). However, this is followed in several days by the second
phase of SIADH, which is caused by uncontrolled release of AVP from the degenerating nerve terminals of the posterior pituitary
gland that have been injured or severed (2 ). Because a smaller portion of the posterior pituitary is denervated, the magnitude of
AVP released as the pituitary degenerates will be smaller and of shorter duration than with a complete triphasic response. After all
the AVP stored in the damaged part of the posterior pituitary gland has been released, the second phase ceases, but clinical DI will
not occur if less than 80%e90% of the AVP neuronal cell bodies in the hypothalamus undergo retrograde degeneration (3 ) (From Loh
JA, Verbalis JG: Disorders of water and salt metabolism associated with pituitary disease. Endocrinol Metab Clin North Am
37:213e234, 2008.)

into nearby capillaries. In animals this may be accompanied by a bulbous growth at the end of the
severed stalk which represents a new, albeit small, neural lobe. In humans, the regeneration process
appears to proceed more slowly, and formation of a new neural lobe has not been detected. Nonetheless,
histological examination of a severed human stalk from a patient 18 months after hypophysectomy
demonstrated reorganization of neurohypophyseal fibers with neurosecretory granules in close prox-
imity to nearby blood vessels, closely resembling the histology of a normal posterior pituitary [31].
Recognition that most patients with CDI retain a limited capacity to secrete some AVP allows an
understanding some otherwise perplexing features of the disorder. For example, in many patients,
restricting water intake long enough to raise plasma osmolality by only 1e2% induces sufficient AVP
secretion to concentrate the urine. As the plasma osmolality increases further, some patients with
partial DI can even secrete enough AVP to achieve near maximal urine osmolalities. However, this
should not cause confusion about the diagnosis of DI, since in such patients the urine osmolality will
still be inappropriately low at plasma osmolalities within normal ranges, and they will respond to
exogenous AVP administration with further increases in urine osmolality. These responses to dehy-
dration illustrate the relative nature of the AVP deficiency in most cases, and underscore the impor-
tance of the thirst mechanism to restrict the necessity to use residual AVP secretory capacity under
basal conditions of ad libitum water intake.

Please cite this article as: Verbalis JGAcquired forms of central diabetes insipidus: Mechanisms of disease,
Best Practice & Research Clinical Endocrinology & Metabolism, https://doi.org/10.1016/
j.beem.2020.101449
8 J.G. Verbalis / Best Practice & Research Clinical Endocrinology & Metabolism xxx (xxxx) xxx

CDI is also associated with changes in the renal response to AVP. The most apparent change is a
reduction in maximal concentrating capacity caused by chronic polyuria. The severity of this defect is
proportional to the magnitude of the polyuria and is independent of its cause. Because of this, the level
of urine concentration achieved at maximally effective levels of plasma AVP is reduced in all types of DI.
Information about the renal concentration mechanism from studies of AQP2 expression in experi-
mental animals has suggested that a form of nephrogenic DI (NDI) is likely associated with all types of
DI, as well as with primary polydipsia. Brattleboro rats that have genetic CDI have been found to have
low levels of kidney aquaporin-2 (AQP2) expression compared to Long-Evans control rats; AQP2 levels
are corrected by treatment with AVP or desmopressin, but this process takes 3e5 days, during which
time urine concentration remains subnormal, despite pharmacologic concentrations of AVP [32].
Similarly, physiologic suppression of AVP by chronic over-administration of water produces a down-
regulation of AQP2 in the renal collecting duct [33]. Clinically, it is well known that patients with both
CDI and primary polydipsia often fail to achieve maximally concentrated urine when they are given
desmopressin during a water deprivation test to differentiate among the various causes of DI. This
effect has long been attributed to a washout of the medullary concentration gradient as a result of the
high urine flow rates in polyuric patients; however, based on the results of animal studies, it seems
likely that at least part of the decreased response to AVP is due to a downregulation of kidney AQP2
expression. Consequently, patients with untreated CDI also have a component of NDI, which is
reversible following treatment with AVP or AVP analogues that allow restoration of normal AQP2 levels
in the kidney collecting ducts. This explains why it takes time, typically several days, to restore
maximal urine concentration after patients with CDI are treated with antidiuretic therapy [34].
The cardinal defect of patients with osmoreceptor dysfunction, a.k.a. adipsic DI, is lack or dysfunction
of the osmoreceptors that regulate thirst. With rare exceptions, the osmoregulation of AVP is also
impaired, although the hormonal response to non-osmotic stimuli remains intact [35]. Four major
patterns of osmoreceptor dysfunction have been described as characterized by defects in thirst and/or
AVP secretory responses: (a) upward resetting of the osmostat for both thirst and AVP secretion (normal
AVP and thirst responses but at an abnormally high plasma osmolality), (b) partial osmoreceptor
destruction (blunted AVP and thirst responses at all plasma osmolalities), (c) total osmoreceptor
destruction (absent AVP secretion and thirst regardless of plasma osmolality), and (d) selective
dysfunction of thirst osmoregulation with intact AVP secretion [12]. Regardless of the actual pattern, the
hallmark of this disorder is an abnormal thirst response in addition to variable defects in AVP secretion.
Because of this, such patients fail to drink sufficiently as their plasma osmolality rises, and as a result, the
new set point for plasma osmolality rises far above the normal thirst threshold. Unlike patients with CDI
whose polydipsia maintains their plasma osmolality within normal ranges, patients with osmoreceptor
dysfunction typically have osmolalities in the range of 300e340 mOsm/kg H2O. This again underscores
the critical role played by normal thirst mechanisms in maintaining body fluid homeostasis; intact renal
function alone is insufficient to maintain plasma osmolality within normal limits in such cases.
The rate of development and the severity of hyperosmolality and hypertonic dehydration in patients
with osmoreceptor dysfunction are influenced by a number of factors. First, is the ability to maintain
some degree of osmotically stimulated thirst and AVP secretion, which will determine the new set
point for plasma osmolality. Second, are environmental influences that affect the rate of water output.
When physical activity is minimal and ambient temperature is not elevated, the overall rates of renal
and insensible water loss are low and the patient's diet may be sufficient to maintain a relatively
normal water balance for long periods. Anything that increases perspiration, respiration, or urine
output greatly accelerates the rate of water loss and uncovers the patient's inability to mount an
appropriate compensatory increase in water intake. Under these conditions, severe and even fatal
hypernatremia can develop relatively quickly. When the dehydration is only moderate (plasma
osmolality 300e330 mOsm/kg H2O), the patient is usually asymptomatic and signs of volume deple-
tion are minimal, but if the dehydration becomes severe the patient can exhibit symptoms and signs of
hypovolemia, including weakness, postural dizziness, confusion, coma, azotemia, hyperglycemia, and
secondary hyperaldosteronism with hypokalemia. In severe cases, there may also be rhabdomyolysis
with marked serum elevations in muscle enzymes and occasionally acute renal failure.
A third factor also substantially influences the degree of hyperosmolality and dehydration present
in these patients. For all cases of osmoreceptor dysfunction, it is important to remember that afferent

Please cite this article as: Verbalis JGAcquired forms of central diabetes insipidus: Mechanisms of disease,
Best Practice & Research Clinical Endocrinology & Metabolism, https://doi.org/10.1016/
j.beem.2020.101449
 J.G. Verbalis / Best Practice & Research Clinical Endocrinology & Metabolism xxx (xxxx) xxx 9

pathways from the brainstem to the hypothalamus remain intact; therefore, these patients will usually
have normal AVP and renal concentrating responses to baroreceptor-mediated stimuli such as hypo-
volemia and hypotension, or to other non-osmotic stimuli such as nausea. This has the effect of pre-
venting severe dehydration, since as hypovolemia develops this will stimulate AVP secretion via
baroreceptive pathways through the brainstem (Fig. 1). Although protective, this effect often causes
confusion, since at sometimes these patients appear to have classic CDI yet at other times they can
concentrate their urine to normal levels. Nonetheless, the presence of refractory hyperosmolality with
absent or inappropriate thirst should alert clinicians to the presence of osmoreceptor dysfunction
regardless of apparent normal urine concentration at different times.
In a few patients with osmoreceptor dysfunction, forced hydration has been found to lead to
hyponatremia in association with inappropriate urine concentration [11]. This paradoxical defect re-
sembles that seen in the syndrome of inappropriate antidiuretic hormone secretion and has been
postulated to be due to two different pathogenic mechanisms. One is continuous or fixed secretion of
AVP because of loss of the capacity for osmotic inhibition and stimulation of hormone secretion. These
observations, as well as electrophysiological data strongly suggest that the osmoregulatory system is
bimodal (i.e., it is composed of inhibitory as well as stimulatory input to the neurohypophysis [36]). The
other cause of the diluting defect appears to be hypersensitivity to the antidiuretic effects of AVP,
because in some patients, urine osmolality may be elevated even when the hormone is undetectable in
plasma [37].
Hypodipsia is also a common occurrence in elderly persons in the absence of any overt hypotha-
lamic lesion [38]. In such cases, it is not clear whether the defect is in the hypothalamic osmoreceptors,
in their projections to the cortex, or in some other regulatory mechanism. However, in most cases the
osmoreceptor is likely not involved, because both basal and stimulated plasma AVP levels have been
found to be normal, or even hyper-responsive, in relation to plasma osmolality in aged humans, with
the exception of only a few studies that showed decreased plasma levels of AVP relative to plasma
osmolality [39].

Practice Points

Central diabetes insipidus (CDI) is usually caused by destruction of the neurohypophysis by

anatomic lesions that destroy the vasopressin neurons by pressure or infiltration.
Because the vasopressin neurons are located in the hypothalamus, lesions confined to the
sella turcica generally do not cause CDI because the posterior pituitary is simply the site of the
axon terminals that secrete vasopressin into the bloodstream.
The capacity of the neurohypophysis to synthesize vasopressin is greatly in excess of the
body's needs, and destruction of 80e90% of the hypothalamic vasopressin neurons is required
to produce CDI.
Even large lesions in the sellar and suprasellar area generally are not associated with impaired
water homeostasis until they are surgically resected with destruction of >80e90% of the hy-
pothalamic vasopressin neurons.
After destructive lesions and damage to the neurohypophysis from surgery or head trauma,
autoimmune destruction of the hypothalamic vasopressin neurons (e.g., lymphocytic infun-
dibuloneurohypophysitis) is the next most common cause of CDI.
Destruction of the osmoreceptors in the anterior hypothalamus that regulate vasopressin
neuronal activity causes a loss of thirst as well as vasopressin section (“adipsic diabetes
insipidus”), leading to severe chronic dehydration and hyperosmolality.
Vasopressin deficiency leads to down-regulation of the synthesis of aquaporin-2 water
channels in the kidney collecting duct principal cells causing a secondary nephrogenic diabetes
insipidus, which explains why it takes several days of vasopressin administration to achieve
maximal urine concentration in patients with CDI.
The presentation of patients with CDI can vary greatly, depending on the size and location of
the lesion, the magnitude of trauma to the neurohypophysis, the degree of destruction of the
vasopressin neurons, and the presence of other hormonal deficits from damage to the anterior
pituitary.

Please cite this article as: Verbalis JGAcquired forms of central diabetes insipidus: Mechanisms of disease,
Best Practice & Research Clinical Endocrinology & Metabolism, https://doi.org/10.1016/
j.beem.2020.101449
10 J.G. Verbalis / Best Practice & Research Clinical Endocrinology & Metabolism xxx (xxxx) xxx

Research Agenda

Identification of the immunological precipitants of lymphocytic neurohypophysitis.

Development of diagnostic tests for autoimmune destruction of vasopressin neurons.
Elucidation of neural pathways regulating thirst following activation of hypothalamic
osmoreceptors.
Predictors of the development of hyponatremia (“isolated second phase”) following pituitary
surgery or trauma.
Etiology of the reduced thirst perception of older individuals.

References

[1] Root AW, Martinez CR, Muroff LR. Subhypothalamic high-intensity signals identified by magnetic resonance imaging in
children with idiopathic anterior hypopituitarism. Evidence suggestive of an 'ectopic' posterior pituitary gland. Am J Dis
Child 1989;143(3):366e7.
[2] Heinbecker P, White HL. Hypothalamico-hypophyseal system and its relation to water balance in the dog. Am J Physiol
1941;133:582e93.
[3] Maccubbin DA, Van Buren JM. A quantitative evaluation of hypothalamic degeneration and its relation to diabetes
insipidus following interruption of the human hypophyseal stalk. Brain 1963;86:443.
*[4] Lipsett MB, Maclean JP, West CD, et al. An analysis of the polyuria induced by hypophysectomy in man. J Clin Endocrinol
Metab 1956;16(2):183e95.
[5] Schreckinger M, Szerlip N, Mittal S. Diabetes insipidus following resection of pituitary tumors. Clin Neurol Neurosurg
2013;115(2):121e6.
[6] Laws Jr ER, Iuliano SL, Cote DJ, et al. A benchmark for preservation of normal pituitary function after endoscopic trans-
sphenoidal surgery for pituitary macroadenomas. World Neurosurg 2016;91:371e5.
[7] Clark AJ, Cage TA, Aranda D, et al. Treatment-related morbidity and the management of pediatric craniopharyngioma: a
systematic review. J Neurosurg Pediatr 2012;10(4):293e301.
[8] Buggy J, Jonhson AK. Preoptic-hypothalamic periventricular lesions: thirst deficits and hypernatremia. Am J Physiol 1977;
233(1):R44e52.
[9] Thrasher TN, Keil LC, Ramsay DJ. Lesions of the organum vasculosum of the lamina terminalis (OVLT) attenuate
osmotically-induced drinking and vasopressin secretion in the dog. Endocrinology 1982;110(5):1837e9.
[10] DeRubertis FR, Michelis MF, Davis BB. "Essential" hypernatremia. Report of three cases and review of the literature. Arch
Intern Med 1974;134(5):889e95.
[11] Halter JB, Goldberg AP, Robertson GL, et al. Selective osmoreceptor dysfunction in the syndrome of chronic hypernatremia.
J Clin Endocrinol Metab 1977;44:609e16.
*[12] Baylis PH, Thompson CJ. Osmoregulation of vasopressin secretion and thirst in health and disease. Clin Endocrinol 1988;
29(5):549e76.
[13] Takaku A, Shindo K, Tanaka S, et al. Fluid and electrolyte disturbances in patients with intracranial aneurysms. Surg Neurol
1979;11(5):349e56.
[14] McIver B, Connacher A, Whittle I, et al. Adipsic hypothalamic diabetes insipidus after clipping of anterior communicating
artery aneurysm. Br Med J 1991;303(6815):1465e7.
*[15] Maghnie M, Cosi G, Genovese E, et al. Central diabetes insipidus in children and young adults. N Engl J Med 2000;343(14):
998e1007.
[16] Kojima H, Nojima T, Nagashima K, et al. Diabetes insipidus caused by lymphocytic infundibuloneurohypophysitis. Arch
Pathol Lab Med 1989;113(12):1399e401.
*[17] Imura H, Nakao K, Shimatsu A, et al. Lymphocytic infundibuloneurohypophysitis as a cause of central diabetes insipidus.
N Engl J Med 1993;329(10):683e9.
*[18] Schaefers J, Cools M, De WK, et al. Clinical presentation and outcome of children with central diabetes insipidus asso-
ciated with a self-limited or transient pituitary stalk thickening, diagnosed as infundibuloneurohypophysitis. Clin
Endocrinol 2017;87(2):171e6.
[19] Johnston PC, Chew LS, Hamrahian AH, et al. Lymphocytic infundibulo-neurohypophysitis: a clinical overview. Endocrine
2015;50(3):531e6.
[20] Scherbaum WA, Bottazzo GF, Czernichow P. Role of autoimmunity in central diabetes insipidus. In: Czernichow P,
Robinson AG, editors. Diabetes insipidus in man. Basel: Karger; 1985. p. 232e9.
*[21] Shimatsu A, Oki Y, Fujisawa I, et al. Pituitary and stalk lesions (infundibulo-hypophysitis) associated with immuno-
globulin G4-related systemic disease: an emerging clinical entity. Endocr J 2009;56(9):1033e41.
[22] Zerbe RL, Robertson GL. A comparison of plasma vasopressin measurements with a standard indirect test in the differ-
ential diagnosis of polyuria. N Engl J Med 1981;305:1539e46.
[23] Zerbe RL, Miller JZ, Robertson GL. The reproducibility and heritability of individual differences in osmoregulatory function
in normal human subjects. J Lab Clin Med 1991;117:51e9.
[24] Hollinshead WH. The interphase of diabetes insipidus. Mayo Clin Proc 1964;39:92e100.
[25] Verbalis JG, Robinson AG, Moses AM. Postoperative and post-traumatic diabetes insipidus. In: Czernichow P, Robinson AG,
editors. Diabetes insipidus in man. Basel: Karger; 1984. p. 247e65.

Please cite this article as: Verbalis JGAcquired forms of central diabetes insipidus: Mechanisms of disease,
Best Practice & Research Clinical Endocrinology & Metabolism, https://doi.org/10.1016/
j.beem.2020.101449
 J.G. Verbalis / Best Practice & Research Clinical Endocrinology & Metabolism xxx (xxxx) xxx 11

[26] Cusick JF, Hagen TC, Findling JW. Inappropriate secretion of antidiuretic hormone after transsphenoidal surgery for pi-
tuitary tumors. N Engl J Med 1984;311:36e8.
*[27] Olson BR, Rubino D, Gumowski J, et al. Isolated hyponatremia after transsphenoidal pituitary surgery. J Clin Endocrinol
Metab 1995;80(1):85e91.
*[28] Olson BR, Gumowski J, Rubino D, et al. Pathophysiology of hyponatremia after transsphenoidal pituitary surgery.
J Neurosurg 1997;87(4):499e507.
[29] Loh JA, Verbalis JG. Diabetes insipidus as a complication after pituitary surgery. Nat Clin Pract Endocrinol Metabol 2007;
3(6):489e94.
[30] Ultmann MC, Hoffman GE, Nelson PB, et al. Transient hyponatremia after damage to the neurohypophyseal tracts.
Neuroendocrinology 1992;56(6):803e11.
[31] Daniel PM, Prichard MM. The human hypothalamus and pituitary stalk after hypophysectomy or pituitary stalk section.
Brain 1972;95(4):813e24.
*[32] DiGiovanni SR, Nielsen S, Christensen EI, et al. Regulation of collecting duct water channel expression by vasopressin in
Brattleboro rat. Proc Natl Acad Sci U S A 1994;91:8984e8.
[33] Terris J, Ecelbarger CA, Nielsen S, et al. Long-term regulation of four renal aquaporins in rats. Am J Physiol 1996;271(2 Pt
2):F414e22.
[34] Harrington AR, Valtin H. Impaired urinary concentration after vasopressin and its gradual correction in hypothalamic
diabetes insipidus. J Clin Invest 1968;47:502.
[35] Smith D, McKenna K, Moore K, et al. Baroregulation of vasopressin release in adipsic diabetes insipidus. J Clin Endocrinol
Metab 2002;87(10):4564e8.
[36] Leng G, Brown CH, Bull PM, et al. Responses of magnocellular neurons to osmotic stimulation involves coactivation of
excitatory and inhibitory input: an experimental and theoretical analysis. J Neurosci 2001;21(17):6967e77.
[37] Andersen LJ, Andersen JL, Schutten HJ, et al. Antidiuretic effect of subnormal levels of arginine vasopressin in normal
humans. Am J Physiol 1990;259(1 Pt 2):R53e60.
[38] Phillips PA, Bretherton M, Johnston CI, et al. Reduced osmotic thirst in healthy elderly men. Am J Physiol 1991;261(1 Pt 2):
R166e71.
*[39] Cowen LE, Hodak SP, Verbalis JG. Age-associated abnormalities of water homeostasis. Endocrinol Metab Clin N Am 2013;
42(2):349e70.

Please cite this article as: Verbalis JGAcquired forms of central diabetes insipidus: Mechanisms of disease,
Best Practice & Research Clinical Endocrinology & Metabolism, https://doi.org/10.1016/
j.beem.2020.101449