Best Practice & Research Clinical Endocrinology & Metabolism 30 (2016) 205e218

Contents lists available at ScienceDirect

Best Practice & Research Clinical

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

Diabetes insipidus: Differential diagnosis and

management
Gary L. Robertson, MD, Emeritus Professor of Medicine *
Feinberg Medical School of Northwestern University, Chicago, IL, USA

a r t i c l e i n f o
Diabetes insipidus (DI) is a syndrome characterized by the excre-
Article history: tion of abnormally large volumes of dilute urine. It can be caused
Available online 18 February 2016 by any of 4 fundamentally different defects that must be distin-
guished for safe and effective management. They are: (1) pituitary
Keywords: DI, due to inadequate production and secretion of antidiuretic
diabetes insipidus hormone, arginine-vasopressin (AVP); (2) gestational DI due to
vasopressin, desmopressin (DDAVP) degradation of AVP by an enzyme made in placenta; (3) primary
differential diagnosis polydipsia, due to suppression of AVP secretion by excessive fluid
magnetic resonance imaging intake; and (4) nephrogenic DI due to renal insensitivity to the
vasopressin neurophysin gene
antidiuretic effect of AVP. This review describes several methods of
aquaporin-2 gene
differential diagnosis, indicates the advantages and disadvantages
vasopressin receptor V2 receptor (AVPR2)
gene of each and presents a new approach that is simpler and less costly
pregnancy but just as reliable as the best of the older methods. The various
treatments for the different types of DI and recent findings on the
genetic basis of the familial forms of DI are also discussed with
emphasis on their contributions to improved diagnosis and
management.
© 2016 Elsevier Ltd. All rights reserved.

Introduction

Diabetes insipidus (DI) is a syndrome characterized by chronic excretion of abnormally large vol-
umes of dilute urine [1,2]. The polyuria is associated with a commensurate increase in fluid intake that
is often but not always due to increased thirst. Other signs and symptoms include urinary frequency,
incontinence, nocturia and enuresis. It is distinguished from the polyuria and polydipsia of

* Tel.: þ1 608 437 8999.

E-mail address: glr388@northwestern.edu.

http://dx.doi.org/10.1016/j.beem.2016.02.007
1521-690X/© 2016 Elsevier Ltd. All rights reserved.
206 G.L. Robertson / Best Practice & Research Clinical Endocrinology & Metabolism 30 (2016) 205e218

uncontrolled diabetes mellitus or other forms of solute diuresis by the absence of glucosuria and a
relatively normal rate of total urinary solute excretion.
This review will describe the clinical criteria for the diagnosis of DI, the four fundamentally different
types of defect that can cause it, the methods for differentiating between them and the safest and most
effective way to manage each.

Diagnosis

The symptoms of DI (Table 1) are similar to those in several other disorders caused by funda-
mentally different abnormalities. Therefore, the diagnosis should be based on the measurement of 24 h
volume, osmolarity and glucose collected while the patient is eating and drinking normally and is not
taking any medication that can interfere with or cause a water diuresis. In an adult or child over 2 years
of age, a volume of more than 40 mL/kg body weight, an osmolarity of less than 300 mosms/L and a
negative test for glucose is diagnostic of DI. In infants or children below the age of 2 years, the upper
limit of normal for urine volume is slightly higher (Table 1) due to the greater water content of their
diet.
Other methods of diagnosis may be required in certain patients or clinical settings. In infants or
children who are not toilet trained, daily urine output can be estimated by weighing diapers before and
immediately after each change. Alternatively, spontaneous fluid intake can be measured. It is normally
about 10e30 mL/kg/day higher than urine output owing to the need to replace insensible as well as
urinary losses (Table 1). Diagnosing DI is also problematic if the patient is unconscious or otherwise
unable to drink or requires intravenous fluids for other reasons. In that setting, it is often advisable to
defer diagnostic tests until the patient is able to eat and drink normally.

Pathophysiology

There are four different types of DI each of which is due to a fundamentally different defect and
must be managed differently (Table 2).
The most common type of DI is due to a deficiency of AVP production and secretion caused by an
acquired or genetic defect in the neurohypophysis [1]. It is variously referred to as pituitary, neuro-
hypophyseal, hypothalamic, neurogenic, central or cranial DI. It occurs when the amount of AVP
secreted at normal basal levels of osmotic stimulation is too low to concentrate the urine. The polyuria
that results produces a slight decrease in body water and a commensurate rise in plasma osmolarity/
sodium that stimulates thirst and a compensatory increase in water intake (polydipsia) which prevents
further decline in body water. As a result, plasma osmolarity/sodium are maintained a level that is only
1e2% above normal and usually within the normal range [3]. However, if water intake is restricted,
there is a further decline in body water and a rise in plasma osmolarity/sodium that intensifies the

Table 1
Criteria for diagnosis of DI. All values from 24 h collection. Fluid intake is always higher than urine output due to the need to
replace insensible water loss. The difference shown here apply to afebrile subjects at rest in a temperature controlled envi-
ronment. They may be much larger when physical activity or temperature are increased.

Age years Symptoms 24 h Urine osmolarity 24 h Urine volume 24 h Fluid intake Urine glucose
mOsm/L mL/kg mL/kg

<2 Heavy wet diapers <300 >60 >100 Negative

Urinary frequency
Irritability
Increased thirst
Hypernatremia
>2 Polyuria, <300 >40 >70 Negative
Nocturia
Incontinence
Enuresis
Polydipsia, þ/ Thirst
 G.L. Robertson / Best Practice & Research Clinical Endocrinology & Metabolism 30 (2016) 205e218 207

Table 2
Types of DI, their causes and treatment. The acquired and genetic causes of pituitary and nephrogenic DI are listed in Table 3.

Type of DI Basic defect Causes Treatment

Pituitary DI Deficient production & Acquired or genetic AVP

secretion of AVP
Primary polydipsia Suppressed secretion of
AVP due to excessive
fluid intake
Gestational DI Increased degradation of
AVP during pregnancy
Nephrogenic DI Renal insensitivity to
antidiuretic effect of AVP
Desmopressin
Chlorpropamide
Tegretol
Abnormal thirst (dipsogenic) or Education, Psychotherapy?
Abnormal cognition (psychogenic)
Placental vasopressinase Desmopressin
Acquired or genetic Thiazide and/or Amiloride with
or without Indomethacin

osmotic stimulation of thirst and AVP secretion. The effect on plasma AVP and urine osmolarity varies
from patient to patient depending on the extent of the deficiency (Fig. 1). If it is severe, fluid restriction
produces little or no rise in plasma AVP or urine osmolarity. However, if the deficiency is partial, as it is
in many patients, the increase in osmotic stimulation may force the release of enough additional AVP to
concentrate the urine. The level of concentration produced varies but, it is almost always submaximal
and remains so even after the administration of AVP or desmopressin (DDAVP) (Fig. 2). The reason for
the decrease in urinary concentrating capacity is uncertain but it also occurs in healthy people after
water loading [4] and may result from decreased AVP stimulation of other tubular functions necessary
to maintain the medullary hypertonic gradient which drives the reabsorption of water from urine in
the collecting tubules [5,6].
A second type of DI is due to suppression of AVP secretion by excessive fluid intake. It is referred to
as primary polydipsia to distinguish it from the secondary polydipsia that occurs in response to water
loss in the other types of DI. It is divided into two subcategories: dipsogenic, in which the polydipsia
appears to be due to abnormal thirst; and psychogenic in which the polydipsia appears to be due to
various cognitive abnormalities. The abnormalities in primary polydipsia are the reverse of those in
pituitary DI. Increased fluid intake slightly reduces plasma osmolarity/sodium thereby reducing AVP
secretion. The decline in plasma AVP reduces urine osmolarity and increases urine volume thereby
producing a compensatory increase in water excretion that prevents over-hydration. Thus, basal
plasma osmolarity/sodium tends to be slightly lower than in patients with pituitary DI but is usually
within the normal range [3]. Fluid intake reverses these abnormalities resulting in normal rises in

Fig. 1. Relation of plasma AVP to plasma and urine osmolarity in patients with different types of DI. Open and closed squares indicate
partial or severe pituitary DI. Open or closed triangles indicate partial or severe nephrogenic DI. Closed circles indicate primary
polydipsia.
208 G.L. Robertson / Best Practice & Research Clinical Endocrinology & Metabolism 30 (2016) 205e218

Fig. 2. Urine osmolarity under basal conditions, during a dehydration test and after administration of AVP or desmopressin (DDAVP)
in patients with primary polydipsia, acquired or familial pituitary DI and acquired or familial nephrogenic DI. The shaded boxes
indicate the values in healthy adults under the same three conditions.

plasma osmolarity/sodium, plasma AVP and urine osmolarity (Fig. 1). The levels of urine osmolarity
achieved are submaximal and remain so even after administration of AVP or DDAVP (Fig. 2) probably
for the same reason as in patients with pituitary DI. Thus, on the basis of urine osmolarity alone it is
virtually impossible to differentiate primary polydipsia from partial pituitary DI by fluid restriction-
AVP/desmopressin challenge tests.
The third type of DI is also due to a deficiency of plasma AVP. However, it is caused by increased
degradation of the hormone by an enzyme made in the placenta. It is called gestational DI [7e10]. Renal
resistance to AVP [7] and reduction of the thirst threshold [11] have also been suggested as contributing
factors. In some cases, the patient is predisposed to gestational DI by a pre-existing sub-clinical defi-
ciency in AVP [12e15]. The DI typically begins in the third or fourth month of pregnancy and remits
4e6 weeks after delivery of the placenta. In other respects, the abnormalities are similar to those in
pituitary DI. Plasma osmolarity and sodium are usually within the normal range for pregnancy which is
about 3e4% lower than in the non-gravid state [8,10]. Vasopressinase-mediated DI developing
immediately after delivery has also been reported after placental abruption (Chapter 11).
The fourth type of DI is caused by renal insensitivity to the antidiuretic effect of normal levels of AVP
[16]. It is referred to as nephrogenic DI and is due to an acquired or genetic defect in renal mechanisms
for concentrating the urine. The clinical presentation is similar to pituitary DI except that basal plasma
AVP is elevated. Basal urine osmolarity is subnormal, urine volume is high, plasma osmolarity/sodium
is slightly elevated, thirst is stimulated and fluid intake is increased in proportion to urine output. Thus,
basal plasma osmolarity/sodium is usually within normal limits. If fluid intake is restricted for any
reason, body water decreases further, raising plasma osmolarity/sodium and plasma AVP to even
higher levels (Fig. 1). If the renal insensitivity to AVP is severe, as it is in many patients with familial
nephrogenic DI, the rise in plasma AVP produced by fluid deprivation causes little or no increase in
urine osmolarity. In other patients, however, the renal insensitivity to AVP is “partial” and is overcome
by the large increases in plasma AVP. In that event, the resultant increase in urine osmolarity is similar
to that in patients with partial pituitary DI or primary polydipsia even after administration of AVP or
DDAVP (Fig. 2). Consequently, it is not possible to differentiate reliably between these 3 types of DI by
determining the effect on urine osmolarity of fluid deprivation followed by administration of AVP or
DDAVP.
 G.L. Robertson / Best Practice & Research Clinical Endocrinology & Metabolism 30 (2016) 205e218 209

Differential diagnosis

Effective management of DI depends on accurate determination of the type. Differentiation can be

relatively simple or challenging depending on the severity of the defect and the methods used. The
clinical setting in which DI occurs may be sufficient to determine the type. For example, DI that begins
during pregnancy and remits several weeks after delivery is almost certainly gestational. However,
several disorders such as neurosarcoid or head trauma have been associated with more than one type
of DI and some associations may be incidental rather than causative. Even a clear-cut family history of
DI does not distinguish between the pituitary and nephrogenic types because each of them can be
inherited in either an autosomal dominant, x-linked recessive or autosomal recessive mode (see
below). Thus, in most cases, a reliable method for differentiating between pituitary DI, primary
polydipsia and nephrogenic DI should be employed before deciding on treatment or embarking on a
search for the underlying cause.
The method used for differential diagnosis varies depending on the clinical presentation and the
resources available. In the rare patient who presents with an abnormal elevation in plasma osmolarity/
sodium, primary polydipsia as well as partial pituitary and partial nephrogenic can be excluded and the
remaining possibilities, severe pituitary or severe nephrogenic DI can be differentiated simply by
determining the effect on urine osmolarity of injecting standard test doses of AVP (0.05 units of
Aqueous Pitressin) or desmopression (1 mg of DDAVP).
In most patients with DI, however, basal plasma osmolarity/sodium on unrestricted fluid intake is
within normal limits. Therefore, it has been customary to perform a fluid deprivation test followed by
the injection of AVP or DDAVP (Fig. 3). Because of the risk of producing severe dehydration in a patient
with severe pituitary or severe nephrogenic DI, the fluid deprivation should be started in the morning
with close monitoring by trained staff in a facility with access to a laboratory that can measure and
report plasma osmolarity and/or sodium as well as urine osmolarity with a turn-around time of less
than 30 min. The details of the method vary but the general approach is to stop all fluid intake and
monitor body weight, plasma sodium and urine osmolarity every hour until urine concentration occurs
or hypertonicity/hypernatremia develops, whichever comes first. If fluid deprivation does not raise
urine osmolarity above 300 mosms/L before hypernatremia develops, primary polydipsia as well as
partial pituitary DI and partial nephrogenic DI are excluded. In that case, the fluid restriction can and
should be stopped to prevent further dehydration and the injection of AVP or DDAVP followed by
repeat measurements of urine osmolarity every hour for 2 h can distinguish between severe pituitary
and severe nephrogenic DI.

Fig. 3. Algorithm for differential diagnosis of DI by modified fluid deprivation test.

210 G.L. Robertson / Best Practice & Research Clinical Endocrinology & Metabolism 30 (2016) 205e218

The traditional fluid deprivation AVP/DDAVP challenge test is less reliable for differential diagnosis
when fluid deprivation results in concentration of the urine before plasma osmolarity and/or sodium
rise above the normal range. In that event, the change in urine osmolarity produced by injecting AVP or
DDAVP does not distinguish clearly between primary polydipsia, partial pituitary DI and partial
nephrogenic DI (Fig. 2). Hence, some other approach to differential diagnosis is needed in this situation.
Primary polydipsia, partial pituitary DI and partial nephrogenic DI can be differentiated reliably by
adding measurements of plasma AVP to the fluid deprivation test [1] (Fig. 1). With this method, the
patient is fluid deprived as in the traditional test but, if urinary concentration occurs, the subsequent
procedures differ. Instead of injecting AVP or DDAVP and repeating the measurements of urine os-
molarity, blood samples for the measurement of plasma AVP as well as plasma osmolarity/sodium are
collected and the fluid deprivation is continued while hypertonic (3%) saline is infused at a rate of
0.05 mL per kilogram body weight per minute. During the infusion, plasma osmolarity/sodium is
measured every 30 min. When the values rise above the upper limits of normal, usually 60e90 min
after the start of the infusion, additional blood is collected for the measurement of plasma AVP and the
test is terminated. The results are interpreted by plotting them on nomograms that depict the normal
relationship of plasma AVP to plasma osmolarity/sodium or urine osmolarity (Fig. 1). This method
differentiates reliably between primary polydipsia, partial pituitary and partial nephrogenic DI but it
also has practical disadvantages. Like the traditional method, it should be performed only with close
monitoring by trained staff in a hospital or other facility with access to a laboratory capable of rapidly
measuring and reporting plasma osmolarity or sodium as well as urine osmolarity. It also requires a
reference laboratory that can assay plasma AVP and provide method specific nomograms to interpret
the results. Therefore, the modified fluid deprivation test is best reserved for difficult cases referred to
consultants with the special resources required.
A new approach that eliminates the need for fluid deprivation provides a simpler but equally
reliable way to differentiate between the 3 major types of DI regardless of the severity of the underlying
defect (Fig. 4). It is based on two observations; first, that basal plasma AVP is always normal or elevated
in nephrogenic DI but is low or undetectable in pituitary DI and primary polydipsia [1]; second, the
hyper-intense signal (“bright spot”) normally emitted by the posterior pituitary on T-1 weighted MRI of
brain is absent or abnormally small in pituitary DI [17] but is present or abnormally large in primary
polydipsia (Fig. 5). Thus, the 3 types of DI can be differentiated in two steps (Fig. 4). The first is to
measure basal plasma AVP and the concurrent urine osmolarity under conditions of unrestricted fluid
intake. If plasma AVP is normal or elevated (>2.5 pg/mL) when urine osmolarity is low (<300 mosms/
L), the patient has nephrogenic DI and the only additional evaluation required is a search for the cause.
However, if plasma AVP is low or undetectable (<1 pg/mL) when urine osmolarity is also low, a brain
MRI can be performed to determine if the posterior pituitary bright spot is absent or small, indicating
pituitary DI, or is normal or enlarged, indicating primary polydipsia. As a bonus, the MRI may also
reveal pathology responsible for either disorder.

Fig. 4. Algorithm for differential diagnosis of DI by plasma AVP-MRI method.

 G.L. Robertson / Best Practice & Research Clinical Endocrinology & Metabolism 30 (2016) 205e218 211

Fig. 5. T-1 weighted mid-saggital MR images in patients with different types of DI. Note that the hyperintense signal normally
emitted by the posterior pituitary is absent in pituitary DI (A), faint in nephrogenic DI (B) and present in primary polydipsia (C).

The main advantages of the AVP-MRI method of differential diagnosis are that it can be performed
on an outpatient basis, does not require specially trained staff, is less stressful for the patient and is
probably more cost-effective than the other methods. Nevertheless, a few caveats are in order. One is
the possibility that a powerful non-osmotic stimulus such as a vasoevagal reaction provoked by
phlebotomy could elevate plasma AVP enough to result in a false diagnosis of nephrogenic DI in a
patient with primary polydipsia or even partial pituitary DI. To guard against this error, the phlebot-
omist should pay particular attention to any symptoms or other reactions of the patient. Another
potential problem is misinterpretation of the MRI. On T-1 weighted images, fat in the dorsum sella also
emits a hyperintense signal that may be misinterpreted as coming from the posterior pituitary. If there
is doubt in this regard, the use of a fat suppression method may help. The MRI is also unreliable for
differential diagnosis in patients with empty sella since they lack the posterior pituitary bright spot
even when they do not have DI. The bright spot is also absent, faint or small in most patients with
nephrogenic DI (Fig. 5) presumably because the high rate of AVP secretion reduces pituitary stores of
the hormone [17]. Consequently, the presence or absence of the posterior pituitary bright spot on MRI
is not sufficient for differential diagnosis of DI unless accompanied by measurement of AVP or other
tests of AVP function.
Another approach to differential diagnosis is a closely monitored therapeutic trial of AVP or DDAVP
(Fig. 6). Properly performed, this method distinguishes with approximately 90% accuracy between
primary polydipsia, pituitary DI and nephrogenic DI regardless of their severity. In pituitary DI,
treatment eliminates the polyuria and produces a slight decrease in plasma osmolarity/sodium that
eliminates thirst and polydipsia and stabilizes water balance within the normal range (Fig. 6A). In
primary polydipsia, the treatment also abolishes the polyuria but the resultant decrease in plasma
osmolarity/sodium has little or no effect on the polydipsia. Therefore, excess water is retained resulting
in the development of hypoosmolemia and hyponatremia, sometimes severe, within 24e48 h (Fig. 6C).
In nephrogenic DI, the treatment has little or no effect on any of these variables (Fig. 6B). Though simple
in concept, the cost, ambiguities and risks of this approach tend to be greater than with the other
methods of differential diagnosis. The trial should be conducted in the hospital with close monitoring
of urine volume, urine osmolarity and plasma osmolarity/sodium. In addition, to differentiate unam-
biguously between primary polydipsia and pituitary DI, the optimum treatment dose should be
determined by titration so as to raise 24 h urine osmolarity to only about 400e700 mosms/L because
doses large enough to produce maximum antidiuresis (urine osmolarity > 1000 mosms/L) produce
water intoxication in patients with pituitary DI as well as in about 20% of healthy subjects [18].

Management

DI can be a life-threatening disorder if the age of the patient or other circumstances interfere with
the compensatory mechanisms that otherwise prevent severe disturbances in hydration. It may also be
212 G.L. Robertson / Best Practice & Research Clinical Endocrinology & Metabolism 30 (2016) 205e218

3600 150 3600 150 3600 150

RFI
Urine Volume, Urine Osmolarity &

3000 145 3000 145 3000 145

Fluid Intake mL/8 hours

Plasma Sodium (mmol/L)

2400 140 2400 140 2400 140

1800 135 1800 135 1800 135

1200 130 1200 130 1200 130

600 125 600 125 600 125

0 120 0 120 0 120

8 32 56 80 104 8 32 56 80 104 8 32 56 80 104
A Hours B Hours C Hours

Fig. 6. Effect of desmopressin therapy in 3 adults with different types of DI. The left panel (A) is a patient with partial pituitary DI.
The middle panel (B) is partial nephrogenic DI. The right panel (C) is a patient with the dipsogenic form of primary polydipsia. The
shaded bars indicate urine volume and the open bars fluid intake in mL/8 h. The shaded triangles show the osmolarity of each 8 h
urine collection. The black circles indicate the plasma sodium concentration at the beginning of each collection period. The stippled
boxes at the top indicate the period of desmopressin therapy, 200 mg by mouth every 8 h. The box labeled FR in panel C indicates
fluid restriction. In all 3 patients, fluid intake tended to be higher than urine volume due to insensible water loss. In the patient with
pituitary DI, desmopressin normalized urine osmolarity and volume within 8 h. At the same time, it also reduced thirst and fluid
intake but the decrease was smaller than urine output. Consequently, body water increased and plasma sodium fell, suppressing
thirst and water intake to bring it into balance with urine output and stabilize plasma sodium within the normal range. In the
patient with nephrogenic DI, desmopressin had no appreciable effect on any of these variables. In dipsogenic DI, desmopressin also
normalized urine osmolarity and volume within 8 h but the reduction in thirst and fluid intake was smaller and remained relatively
high for the next 32 h, resulting in progressive water retention and the development of hyponatremia that was corrected by dis-
continuing desmopressin and restricting fluid intake (RFI).

the harbinger of a more serious underlying disorder that needs to be identified and treated. Usually,
however, DI is simply a bother to the patient. The polyuria, nocturia, enuresis, thirst and polydipsia are
irksome, sometimes embarrassing and often limiting in ordinary activities. Fortunately, most of these
irritations can be eliminated or minimized by treatments appropriate for the type of DI and other
characteristics of the patient.

Pituitary DI

The polyuria, thirst and polydipsia of pituitary DI can be controlled completely by treatment with
the hormone pitressin (AVP) or desmopressin (DDAVP, Ferring), an analogue specific for the V2 re-
ceptor [19]. The choice depends mainly on the clinical setting. If the patient is hospitalized, obtunded,
acutely ill or otherwise unable to regulate fluid intake via thirst, intravenous infusion of aqueous
pitressin may be preferable because it has a shorter duration of action than DDAVP and, as a result, can
be reduced more quickly if hyponatremia develops or intravenous fluids need to be increased for other
therapeutic reasons. In adults and children over the age of 2 years, infusing AVP at a rate of about 2 mU/
kg body weight per hour usually maintains urine osmolarity and volume in the normal range of about
400e800 mosms/L and 0.5 to 1.0 mL/kg/h, respectively. DDAVP produces similar levels of antidiuresis
when given intravenously or subcutaneously at a dose of 1 mg once or twice a day.
In patients who are alert and able to regulate their fluid intake via the thirst mechanism, the
preferred treatment for pituitary DI is DDAVP. In addition to having a longer duration of action than
AVP, DDAVP is available in parenteral, intranasal, oral or sublingual formulations. The bioavailability of
each of the formulations differs in a ratio of about 1:10:100:60, respectively. In adults and children over
the age of 2 years, the oral or sublingual formulations are usually preferred for long-term management.
The recommended doses range from 100 to 200 mg one to three times a day. However, the optimum
dose should be determined for each patient by titration because bioavailability varies. The aim should
 G.L. Robertson / Best Practice & Research Clinical Endocrinology & Metabolism 30 (2016) 205e218 213

be to maintain 24 h urine osmolarity and volume in the normal range ei.e. 400e800 mosms/L and
20e30 mL/kg/day - because at that level of antidiuresis thirst is minimized and most patients reduce
total fluid intake enough to prevent the development of water intoxication (Fig. 6A). As an additional
precaution against water intoxication, patients on antidiuretic therapy should be advised to drink only
enough to satisfy thirst since larger amounts cannot be excreted promptly as they would be if urine
output was regulated normally by changes in plasma osmolarity/sodium.
Pituitary DI can also be treated with chlorpropamide (Diabinese), carbamazepine (Tegretol) and
clofibrate (Atromid) [20e26]. They tend to be less effective than DDAVP and have more undesirable
side effects but should be considered when other options are unavailable.
The underlying cause of pituitary DI also should be investigated. Most are acquired but a substantial
number are genetic and nearly half are still idiopathic (Table 3). Most of the acquired causes are evident
from the history or brain MRI at the time of presentation. However, others, such as histiocytosis, may not
manifest until later. Thus, repeat MRI may be worthwhile, especially in children, if the initial MRI is negative

Table 3
Causes of pituitary DI.

Acquired
Trauma (including pituitary surgery)
Neoplasm
Primary
Craniopharyngioma
Pituitary adenoma (suprasellar)
Dysgerminoma
Meningioma
Metastatic
Lung
Breast
Hematologic
Lymphoma
Leukemia
Granuloma
Neurosarcoid
Histiocytosis
Xanthoma disseminatum
Infection
Meningitis
Viral encephalitis
Toxoplasmosis
Inflammation
Lymphocytic neurohypophysitis
Granulomatosis with polyangiitis
Lupus erythematosus
Vascular
Sheehan's syndrome
Aneurysm of internal carotid
Aortocoronary bypass
Hypoxic encephalopathy
Toxin
Tetrodotoxin
Snake venom
Congenital malformations
Septo-optic dysplasia
Holoprosencephaly
Idiopathic
Genetic
Autosomal dominant (AVP gene)
Autosomal recessive
Type A (AVP gene)
Type B (AVP gene)
Type C (WFS-1 gene)
X-linked recessive (unknown gene, Xq28)
214 G.L. Robertson / Best Practice & Research Clinical Endocrinology & Metabolism 30 (2016) 205e218

and other investigations, including genetic analysis are also unrevealing. The lack of a family history is not a
contraindication to genetic analysis because some of the mutations that cause pituitary DI occur de novo or
are recessive. Identification of the mutation is useful not only for general family counseling but also, in
many cases, for determining at birth whether offspring of an affected parent will or will not have DI.
There are at least 5 different forms of familial pituitary DI [27,28]. By far the most common, with
more than 100 kindreds reported to date, is inherited in an autosomal dominant completely penetrant
mode. It is attributed to at least 70 different mutations in one allele of the AVP gene on chromosome
20p13 that encodes the AVP-neurophysin II-copeptin precursor. The mutations have been found in all
coding regions except the part of exon 3 that encodes copeptin. All of them replace or delete one or
more amino acid residues believed to be critical for correct processing and/or folding of the precursor
protein. Hence, they are thought to cause the mutant precursor to misfold, accumulate and eventually
destroy the neuron, thereby preventing expression of the normal allele. The age at which signs of DI
begin has not been defined with certainty but it appears to vary from 6 months to 6 years or even older
in a few cases. In the beginning, the deficiency of AVP is partial and may remain so for several years.
Eventually, it almost always progresses to a severe if not complete lack of AVP. Autopsy studies in adults
with autosomal dominant familial pituitary DI show selective degeneration of AVP producing mag-
nocellular neurons. Oxytocin producing neurons are spared. Anterior pituitary function is also normal.
The other 4 types of familial pituitary DI appear to be relatively rare [27]. Two of them are autosomal
recessive and also result from mutations in the AVP gene. In one, the mutation is in the part of the gene
that encodes for position 7 in the AVP moiety. It apparently does not interfere with folding or pro-
cessing of the precursor but results in the production and secretion of a mutant AVP that is biologically
inactive. Homozygotes produce only the biologically inactive mutant of AVP. Heterozygotes do not
manifest DI presumably because the normal allele is able to produce enough normal AVP to maintain
urinary concentration. The second type of autosomal recessive pituitary DI is due a very large deletion
of the AVP gene and the intergenic region between the AVP and oxytocin precursor genes. This mutation
would be expected to abolish the transcription of the AVP gene. Homozygotes have pituitary DI but
heterozygotes do not, indicating again that one normal allele can produce enough AVP to prevent DI.
The third type of autosomal recessive pituitary DI occurs in Wolfram's syndrome, a rare, complex
and progressive neurodegenerative disease that also includes early onset insulin dependent diabetes
mellitus, progressive optic atrophy, sensorineural deafness and various other neurological abnormal-
ities. The DI typically begins in adolescence but can occur at almost any age and is associated with a
deficiency of AVP secretion, a good response to DDAVP, histologic evidence of posterior pituitary
degeneration and defective processing of the AVP precursor. Most patients are homozygous or com-
pound heterozygous for mutations of the WFS1 gene on chromosome 4p16.1. Heterozygotes appear to
be unaffected.
The fifth type of familial pituitary DI is inherited in an X-linked recessive mode. It is characterized by
onset of pituitary DI in males during infancy or early childhood, variable progression in the AVP
deficiency and full control of the DI by standard doses of DDAVP. Obligatory female carriers secrete AVP
normally and do not have DI. The gene responsible for the DI has not been identified but it links to
chromosome Xq28, the same region that contains the gene for the V2 receptor. However, the AVPR2
gene as well as the AVP gene are both normal.

Gestational DI

This type of DI also should be treated with DDAVP because it is more resistant to degradation than
AVP [10]. The same precautions apply as in pituitary DI. The only differences are [1] the target level of
plasma osmolarity/sodium is about 2e3% lower than in the non-gravid state and [2] the treatment
probably can be discontinued 4e6 weeks after delivery. If the DI recurs, the cause should be investi-
gated further as in pituitary DI.

Nephrogenic DI

The treatment of nephrogenic DI varies depending on the cause. AVP and DDAVP are ineffective
(Fig. 6B) except when given in very high doses to patients with partial resistance to the hormone.
 G.L. Robertson / Best Practice & Research Clinical Endocrinology & Metabolism 30 (2016) 205e218 215

However, the cost and inconvenience of this treatment makes it impractical for long-term use. Some
acquired forms of nephrogenic DI are cured by eliminating the cause. For example, lithium-induced
nephrogenic DI may remit completely if the drug is stopped within the first year or two of treat-
ment. Similarly, the nephrogenic DI caused by hypokalemia remits when the potassium deficiency is
corrected. In most patients, however, the basic cause cannot be corrected and other less effective
methods of treatment must be used. Reducing dietary sodium and protein reduces the polyuria and
polydipsia by decreasing the urinary solute load, principally sodium and urea. However, the effects of
this approach are often limited by dislike of the dietary restrictions and, in children, by the re-
quirements for normal growth. Administration of standard doses of chlorothiazide (Diuril) and/or
amiloride (Midamor) increases urine osmolarity and decreases urine volume. The mechanism is most
likely a reduction in sodium reabsorption in the loop of Henle (which impairs urinary dilution) and a
compensatory increase the reabsorption of sodium, chloride and water in the proximal tubule (which
reduces urine volume) [29]. The magnitude of the antidiuretic effect varies but, on average, it doubles
urine osmolarity and halves urine volume [27]. The principal side effects are hypokalemia (chloro-
thiazide) or hyperkalemia (amiloride). Hence, the two drugs are often given together to neutralize the
effects on potassium excretion. Treatment with the prostaglandin synthetase inhibitor indomethacin
(Indocin) [27] or a cyclooxogenas-2 inhibitor [30,31] also increases urine osmolarity and reduces urine
volume in nephrogenic DI. The antidiuretic effects also vary and are similar in magnitude to those of
chlorothiazide and/or amiloride. However, the effects are additive and they are often given together to
enhance antidiuresis. In some patients, particularly those with partial nephrogenic DI, the combination
reduces urine output and fluid intake to normal. The principal adverse effect of indomethacin is
gastritis, which can be minimized by taking the medication with meals. In most cases, the existing
treatments provide only partial relief of polyuria, thirst and polydipsia and alternative approaches are
under consideration [27].
The cause of nephrogenic DI also should be investigated. The acquired causes are usually evident
from other clinical findings (Table 4). If no cause is found, a genetic analysis is indicated even if the
family history is negative since most of mutations responsible for nephrogenic DI are recessive. There
are several forms of familial nephrogenic DI [16,27]. The most common, accounting for approximately
90% of all patients, is transmitted in an X-linked recessive mode. It is due to a mutation of the AVPR2
gene on chromosome Xq28 that encodes the V2 receptor. Over 200 different mutations have been
identified to date. They cause DI by interfering with the intracellular trafficking or binding properties of
the V2 receptor. Males with the mutation have nephrogenic DI from birth. It is usually severe but may
be partial. Female carriers are usually unaffected but may also exhibit mild partial nephrogenic DI due
to skewed inactivation of the X chromosomes. The X-linked recessive form of familial nephrogenic DI
can be confused with the X-linked recessive form of familial pituitary DI because the inheritance
patterns are identical and the genes for both are located on chromosome Xq28.
Familial nephrogenic DI can also be caused by a mutation in the AQP2 gene on chromosome 12q13
that encodes aquaporin-2. Aquaporin-2 is a homotetrameric protein which acts as a water channel,
When activated by the action of AVP, it is inserted into the luminal surface of the principle cell where it
permits reabsorption of water from dilute urine in renal collecting ducts. Approximately 50 different
mutations in the AQP2 gene have been identified to date. They are located throughout the coding region
and most are recessive. They cause the mutant aquaporin-2 to misfold and be retained in the endo-
plasmic reticulum. In the homozygous state, they cause severe nephrogenic DI in males and females. A
few of the mutations are dominant. They are located in the C-terminal intracellular tail of aquaporin-2
and appear to cause nephrogenic DI by forming tetramers with wild type aquaporin-2 causing the
water channels to be misrouted to the wrong surface of the principal cell. In these patients, the
nephrogenic DI is usually partial or less severe than in those with the recessive mutations presumably
because a small fraction of the water channels are formed completely by tetramers of wild-type
aquaporin-2 and, as a consequence, are routed correctly to the luminal surface of the principal cell.

Primary polydipsia

Treatment of primary polydipsia with AVP or DDAVP eliminates the polyuria but has little or no
effect on the polydipsia even though it reduces plasma osmolarity/sodium. As a result, the excessive
216 G.L. Robertson / Best Practice & Research Clinical Endocrinology & Metabolism 30 (2016) 205e218

Table 4
Causes of nephrogenic DI.

Acquired
Pharmacologic
Lithium
Demeclocycline
Methoxyflurane
Ampohotericin B
Aminoglycosides
Cisplatin
Rifampin
Foscarnet
Metabolic
Hypokalemia
Hypercalcemia, hypercalciuria
Obstructive
Ureter or urethra
Vasacular
Sickle cell disease or trait
Acute tubular necrosis
Granuloma
Sarcoid
Neoplasm
Sarcoma
Infiltrative
Amyloidosis
Idiopathic
Genetic
X-linked recessive (AVP receptor-2 gene)
Autosomal Recessive (aquiaporin-2 gene)
Autosomal dominant (aquaporin-2 gene)

intake of water continues but can no longer excreted due to the antidiuretic therapy. Hence, water
intoxication, sometimes severe, develops rapidly (Fig. 6C). Other medications, like chlorothiazide,
which also impair urinary free water excretion have the same effect and should be avoided as well. The
only safe and effective way to treat primary polydipsia is to eliminate the patients desire to consume
large amounts of fluid. Education may succeed at least when the polydipsia is motivated by a popular
misconception about the health benefits of a high fluid intake. However, it usually accomplishes little
when the urge to drink is driven by psychological factors or abnormal thirst. In that event, the best
approach is to warn the patient and/or caregivers about the causes, signs and symptoms of water
intoxication and the importance of seeking immediate medical care if they occur.
The cause of primary polydipsia should be sought when it is motivated by abnormal thirst. In addition
to a careful history about head injuries and other pathology known to be associated with dipsogenic DI
(Table 5), the investigation should include a brain MRI. Genetic causes have not been identified.

Summary

Differentiation between pituitary DI, primary polydipsia and nephrogenic DI is essential because
safe and effective management of each one differs. Accurate diagnosis by measurements of urine os-
molarity alone during fluid-deprivation e DDAVP challenge tests is problematic because the defects in
AVP secretion or action are often partial. Adding measurements of plasma AVP improves the accuracy
of the test but it is still stressful, of limited availability and costly owing to need for special facilities and
staff. A new method based on a measurement of basal plasma AVP followed, if necessary, by MRI of
brain is much easier and less costly but equally reliable even in partial forms of pituitary and neph-
rogenic DI. Recent discoveries of the genetic basis and modes of inheritance of new as well as previ-
ously known forms of familial pituitary and nephrogenic DI also provide better methods of early
diagnosis and family counseling as well as new directions in the search for more effective methods of
treatment.
 G.L. Robertson / Best Practice & Research Clinical Endocrinology & Metabolism 30 (2016) 205e218 217

Table 5
Causes of primary polydipsia.

Acquired
Psychogenic
Schizophrenia
Obsessive compulsive disorder
Dipsogenic (abnormal thirst)
Head trauma
Neurosarcoid
Tuberculous meningitis
Multiple sclerosis
Lithium

Practice points

 It is essential to differentiate accurately between pituitary DI, primary polydipsia and neph-
rogenic DI because these three types of DI must be managed differently.
 Traditional fluid-deprivation-DDAVP challenge tests are of limited utility for diagnosis
because they are stressful, costly because of the need for special facilities and staff and do
not distinguish reliably between primary polydipsia, partial pituitary DI and partial nephro-
genic DI.
 Assay of plasma AVP during fluid deprivation-hypertonic saline infusion tests improves the
accuracy of differential diagnosis but the procedure is still stressful, costly and of limited
availability.
 A new method of diagnosis based on a measurement of basal plasma AVP followed, if
indicated, by a brain MRI is just as reliable but much easier, more available and less costly
than the fluid deprivation-AVP assay method.
 Both pituitary and nephrogenic DI can be inherited in an autosomal dominant, autosomal
recessive or X-linked recessive mode.
 Genetic analysis is indicated in patients with idiopathic pituitary or nephrogenic DI regardless
of age or family history.

Future research

 Identify the gene in chromosome Xq28 that is responsible for the X-linked recessive form of
familial pituitary DI.
 Investigate methods to prevent post-natal degeneration of neurohypophyseal neurons in
patients with autosomal dominant pituitary DI.
 Develop better treatments for nephrogenic DI and primary polydipsia.
 Define the role of copeptin in the evaluation of patients with osmoregulatory disorders.

Research support

Otsuka Pharmaceuticals: Aquaretic efficacy of tolvaptan.

Conflict of interest

None.
218 G.L. Robertson / Best Practice & Research Clinical Endocrinology & Metabolism 30 (2016) 205e218

References

*[1] Robertson GL. Diabetes insipidus. Endocrinol Metab Clin N. Am 1995;24(3):549e72.

[2] Robertson GL. History. Transl Endocrinol Metab 2012;3(3):15e25.
*[3] Robertson GL. Differential diagnosis of polyuria. Annu Rev Med 1988;39:425e42.
*[4] Barlow ED, De Wardener E. Compulsive water drinking. Q J Med 1959;28. 235e258.l.
[5] Promeneur D, Bankir L, Hu MD, et al. Renal tubular and vascular urea transporters: influence of antidiuretic hormone on
messenger RNA expression in Brattleboro rats. J Am Soc Nephrol 1998;9(8):1359e66.
[6] Kim GH, Ecelbarger CA, Mitchell C, et al. Vasopressin increases Na-K-2Cl cotransporter expression in thick ascending limb
of Henle's loop. Am J Physiol 1999;276(1 Pt 2):F96e103.
[7] Barron WM, Cohen LH, Ulland LA, et al. Transient vasopressin resistant diabetes insipidus of pregnancy. N Engl J Med
1984;310(7):442e4.
[8] Barron WM, Davison JM, Lindheimer MD. Water metabolism in pregnancy. Semin Nephrol 1984;4(4):334e45.
[9] Durr JA, Hoggard JG, Hunt JM, et al. Diabetes insipidus in pregnancy associated with abnormally high circulating vaso-
pressinase activity. N Engl J Med 1987;316:1070e974.
*[10] Durr JA. Diabetes insipidus of pregnancy. Am J Kidney Dis 1987;9(4):276e83.
[11] Baylis PH, Thompson C, Burd J, et al. Recurrent pregnancy induced polyuria and thirst due to hypothalamic diabetes
insipidus: an investigation into possible mechanisms responsible for polyuria. Clin Endocrinol 1986;24:459e66.
[12] Iwasaki Y, Oiso Y, Kondo K, et al. Aggravation of subclinical diabetes insipidus during pregnancy. N Engl J Med 1991;
324(8):522e6.
[13] Williams DJ, Metcalfe KS, Skingle L, et al. Pathophysiology of transient cranial diabetes insipidus during pregnancy. Clin
Endocrinol 1993;38(6):595e600.
[14] Soule SG, Monson JP, Jacobs HS. Transient diabetes insipidus in pregnancy-a consequence of enhanced placental clearance
of arginine vasopressin. Hum Reprod 1995;10(12):3322e4.
[15] Hashimoto M, Ogura T, Otsuka F, et al. Manifestation of subclinical diabetes insipidus due to pituitary tumor during
pregnancy. Endocr J 1996;43(5):577e83.
*[16] Bichet DG. Renal actions of vasopressin. Transl Endocrinol Metab 2012;3(3):95e122.
*[17] Fujisawa I. Magnetic resonance Imaging of the hypothalamic-neurohypophyseal system. J Neuroendocrinol 2004;16:
297e302.
[18] Kovacs L, Rittig S, Robertson GL. Effects of sustained antidiuretic treatment on plasma sodium concentration and body
water homeostasis in healthy humans on ad libitum water intake. Clin Res 1992;40(2):165A.
*[19] Oiso Y, Robertson GL, Norgaard JP, et al. Treatment of neurohypophyseal diabetes insipidus. J Clin Endocrinol Metab 2013;
98(10):3958e67.
[20] Athar S, Robertson GL. Central diabetes insipidus and its management with oral antidiuretic agents. J Islamic Med Assoc
1981;13:91e5.
[21] Byun KY, Gaskill MB, Robertson GL. Clin Res 1984;32:483A.
*[22] Durr JA, Hensen J, Ehnis T, et al. Chlorpropamide upregulates antidiuretic hormone receptors and unmasks constitutive
receptor signaling. Am J Physiol Ren Physiol 2000;278:F799e808.
[23] Thompson P, Earll JM, Schaaf M. Comparisaon of clofibrate and chlorpropamide in vasopressin-responsive diabetes
insipidus. Metabolism 1977;26(7):749e62.
[24] Meinders AE, Cejka V, Robertson GL. The antidiuretic action of carbamazepine in man. Clin Sci Mol Med 1974;47:289e99.
[25] Stephens WP, Coe JY, Baylis PH. Plasma arginine vasopressin concentrations and antidiuretic action of carbamazepine. Br
Med J 1978;1:1445e7.
[26] Gold PW, Robertson GL, Ballenger JC, et al. Carbamazepine diminishes the sensitivity of the plasma arginine vasopressin
response to osmotic stimulation. J Clin Endocrinol Metab 1983;57(5):952e7.
*[27] Babey M, Kopp P, Robertson GL. Familial forms of diabetes insipidus: clinical and molecular characteristics. Nat Rev
Endocrinol 2011;7(12):701e14.
*[28] Christensen JH, Robertson GL. Deficiencies of vasopressin and thirst. Transl Endocrinol Metab 2012;3(3):57e94.
[29] Cutler RE, Kleeman CR, Maxwell MH, et al. Physiologic studies in nephrogenic diabetes insipidus. J Clin Endocrinol Metab
1962;22:827e38.
[30] Pattaragarn A, Alon US. Treatment of congenital nephrogenic diabetes insipidus by hydrochlorothiazide and
cyclooxygenase-2 inhibitor. Pediatr Nephrol 2003;18:1073e6.
[31] Soylu A, Kasap B, Ogun N, et al. Efficacy of COX-2 inhibitors in a case of congenital nephrogenic diabetes insipidus. Pediatr
Nephrol 2005;20:1814e7.