Pediatric Central Diabetes Insipidus: Brain Malformations Are Common and Few Patients Have Idiopathic Disease

Context:

Pediatric cohorts of central diabetes insipidus (CDI) have shown varying prevalences for the different causes of CDI, including idiopathic.

Objective:

The objective of the study was to determine the causes of CDI at a pediatric tertiary care center and to characterize their clinical outcomes.

Design and Setting:

All patients with CDI at Seattle Children's Hospital were identified and retrospectively analyzed.

Patients:

From 2000 to 2013, 147 patients with CDI were encountered (mean age 7 y at diagnosis, mean follow-up 6.2 y).

Outcome Measures:

The different causes of CDI were grouped, and age of diagnosis, anterior pituitary hormone deficiencies (APHDs), and presence of the posterior pituitary bright spot (PPBS) were analyzed. Patients with idiopathic CDI had infundibular thickening measured using a systematic method.

Results:

Brain malformations caused 24% of CDI cases, and 12.2% were idiopathic. Four of 22 patients with initially idiopathic CDI were diagnosed with an underlying condition, none occurring later than 2.5 years from diagnosis. APHDs were as common in the brain malformation group as they were in the tumor/infiltrative group (72% vs 85%; P = .09). The PPBS was present in at least 13% of patients and in 19% of those with brain malformations. Patients with idiopathic CDI and stalk thickening on the initial magnetic resonance imaging were more likely to have an underlying diagnosis (40% vs 0%; P = .03).

Conclusions:

Brain malformations were a more common cause of pediatric CDI than previously reported. These patients have a high rate of APHDs, and many have persistence of the PPBS. Idiopathic CDI is an uncommon diagnosis, and none of our patients were diagnosed with Langerhans cell histiocytosis or germinoma for more than 3 years from CDI diagnosis. Providers can consider less frequent magnetic resonance imaging after this time point. A systematic method of infundibular measurement on the initial magnetic resonance imaging may predict an underlying germinoma or Langerhans cell histiocytosis.

Central diabetes insipidus (CDI) is a rare but potentially dangerous and often difficult-to-manage dysregulation of water balance secondary to arginine vasopressin [AVP; or antidiuretic hormone (ADH)] deficiency, with an annual incidence approaching 4 per 100 000 (1–3). CDI can have serious underlying causes including brain tumors (4–7) and midline malformations (8–10) as well as infiltrative conditions such as Langerhans cell histiocytosis (11), sarcoidosis (12), tuberculosis, and nonspecific hypophysitis, some of which are treatable and require unique management. Patients with a new diagnosis of CDI should undergo a diagnostic work-up including neuroimaging, and those with no cause identified are often considered idiopathic (13). A few groups have reported the epidemiology of pediatric CDI either at their own institution (14–17) or combined across multiple institutions (18). Many of these studies were limited by small sample sizes, and the last cohort described in the United States was more than 20 years ago (19). The studies have shown a wide variation in the underlying causes of CDI, with most showing a 30%–50% prevalence of idiopathic diabetes insipidus but others finding a much lower rate (14, 20). With the reasons for the variation in these different reports unclear, there is no well-established true prevalence of idiopathic CDI in the current literature.

Brain magnetic resonance imaging at diagnosis is an important tool to determine the underlying cause of CDI and often helps to confirm the diagnosis of CDI via absence of the posterior pituitary bright spot (PPBS) (13). Previous investigators have shown that absence of the PPBS, considered as a surrogate marker for structural abnormalities or decreased ADH production, is nearly universal in patients with CDI, with the sensitivity of this magnetic resonance imaging (MRI) finding typically reported as between 90% and 100% (18, 21, 22). One exception has been cases of familial CDI, where the PPBS is present more frequently (23, 24). Thickening of the infundibulum is known to be associated with Langerhans cell histiocytosis (LCH), germinoma, and hypophysitis and is followed longitudinally in patients with idiopathic CDI, but it is unclear how long repeat imaging is necessary to exclude these underlying diagnoses (25).

CDI is a rare disorder and it is difficult to assemble a large pediatric cohort to effectively examine outcomes and improve care. We have identified all patients in the Seattle Children's Hospital health system with CDI over a 13-year period. Based on clinical experience at our institution, we suspected a lower prevalence of idiopathic CDI and a higher prevalence of brain malformations than was previously reported, and we sought to test this hypothesis in this large cohort. Here we report the causes of CDI in our cohort as well as relevant outcomes for anterior pituitary hormone deficiency and brain MRI.

Subjects and Methods

This study was approved by the Seattle Children's Hospital (SCH) Institutional Review Board. SCH is a tertiary pediatric medical center with a catchment area covering several states in the Pacific Northwest region of the United States. Patients were identified from an International Classification of Diseases, ninth revision, search of the electronic medical record at SCH. International Classification of Diseases, ninth revision, codes for central diabetes insipidus, as well as panhypopituitarism and other relevant diagnoses (other disorders of the pituitary gland, malignant histiocytosis, malignant neoplasms of the pituitary gland and craniopharyngeal duct, disorders of the optic chiasm, etc) were used to identify a large cohort of patients with a possible diagnosis of CDI seen at SCH from January 2000 through July 2012. These records were then reviewed to find those patients with a diagnosis of CDI, and they were included in our CDI database. A few additional patients diagnosed after July 2012 were added to the database by the authors as they were encountered in clinical practice. All diagnoses were made by a pediatric endocrinologist.

To exclude short-term temporary CDI, patients with less than 2 months of follow-up time after diagnosis of CDI were excluded (n = 2). In total, all but two patients had at least 6 months of follow-up with persistent CDI. All patients were being treated with desmopressin at the end of their follow-up time for this analysis.

One hundred fifty patients with a diagnosis of CDI on their medical record were identified. Two patients were excluded because very little diagnostic information was available and the reported causes of CDI (microcephaly and demyelinating process) did not fit known causes. Another patient was excluded because their endocrine care was at an outside institution, and we did not have records regarding their CDI diagnosis, treatment, or follow-up. This left 147 patients for our analysis.

For the 106 patients who had available information, 19% of patients were diagnosed with a water deprivation test, 15% with a single measurement of fasting laboratory results, 31% based on single measurement laboratory results while in the pediatric intensive care unit recovering from intracranial surgery, and 32% with nonfasting laboratory results or symptoms in the context of a significant risk factor for CDI (eg, known brain malformation or intracranial surgery). Where the data were available (n = 89), we evaluated serum sodium, serum osmolality, urine osmolality, ADH level, and urine specific gravity to confirm the diagnosis of CDI.

Data collection and analysis

The causes of CDI were grouped into the following categories: 1) anatomic malformations: septooptic dysplasia, holoprosencephaly, or other brain midline malformation (encephalocele, frontonasal dysplasia, isolated absence of the septum pellucidum); 2) acquired, infiltrative, or tumor: LCH, craniopharyngioma, germ cell tumor, other brain mass/cyst, infection/inflammation, or trauma; 3) genetic or familial; or 4) idiopathic. Any patient with no known predisposing condition at the time of the CDI diagnosis and with no cause identified on their initial MRI was considered to have initially idiopathic CDI. All patients later found to have underlying diagnoses were assigned to the appropriate diagnosis groups for further analysis.

First, to check the accuracy of case ascertainment using this method, charts for all patients seen in our clinic during a randomly determined 3-month period (August 30, 2013 through November 30, 2013) were reviewed for cases of CDI. Patients with CDI identified in this manner were then searched for in our database. There was 100% ascertainment of all patients with CDI seen in this time period.

Second, charts for patients with CDI were reviewed for demographic information, date and method of diagnosis, cause of CDI, anterior pituitary deficiencies, and surgical history. Pituitary deficiencies were counted if an endocrinologist stated the patient was deficient. These diagnoses were based on a combination of clinical factors and laboratory testing including IGF-1, IGF binding protien-3 levels, and provocative testing for GH deficiency, estradiol levels for gonadotropin deficiency, T₄ levels for thyroid hormone deficiency, and low morning cortisol or response to cortrosyn stimulation testing for ACTH deficiency.

Neuroimaging

For the purpose of this study, each patient's initial MRI report was reviewed for a description of the presence of the posterior pituitary bright spot (PPBS) in its normal location. Some patients with no comment on the PPBS on their MRI report had their images reviewed by a coauthor (F.A.P.). PPBS information was missing for 33 patients. Those with a PPBS present had all subsequent MRI reports investigated, and if the PPBS was later absent, the patient was considered to have absence of the PPBS. For the PPBS outcome, all patients with data missing for the PPBS were assumed to have absence of the PPBS, thereby providing a minimum estimate of the PPBS prevalence in that subgroup.

For patients with initially idiopathic CDI, MRI studies were analyzed by a fellowship-trained, subspecialty board-certified neuroradiologist without knowledge of the clinical data. Sagittal and coronal T1 MRI sequences of the brain were performed using standard institutional clinical scanning protocols with slice thicknesses of 1–3 mm on both 1.5T and 3T systems. Pituitary stalk diameter was measured on sagittal and coronal images at the following time points: 1) at the level of the optic chiasm (near its origin from the median eminence), 2) at its insertion on the pituitary gland, and 3) at the level of maximum thickness along its length. The pituitary stalk was considered thickened if the average of the maximum measurements on sagittal and coronal images was 3.5 mm or greater [modified from established criteria (26)].

Results

CDI diagnosis and underlying disorders

Our record review identified 147 children with CDI evaluated at Seattle Children's Hospital from January 2000 through December 2013 who were analyzed for this report. The children in our cohort were followed up an average of 6.2 years (SD 5.0 y) from diagnosis to the end of data collection for this study.

The causes of CDI in our population are displayed in Table 1. The most common single diagnosis was craniopharyngioma (25.2%), followed by septooptic dysplasia (14.3%), and LCH (12.2%).

Table 1.

Causes of Central Diabetes Insipidus at Seattle Children's Hospital, 2000–2013 (n = 147)

Cause	Number	Percentage
Craniopharyngioma	37	25.2
Septooptic dysplasia	21	14.3
LCH	18	12.2
Idiopathic	18	12.2
Germ cell tumor	15	10.2
Genetic/familial	11	7.5
Holoprosencephaly	8	5.4
Other brain mass or cyst	8	5.4
Other brain and/or midline malformation	6	4.1
Infection/inflammation	3	2.0
Trauma	2	1.4
Total	147	100.0

Grouped causes of CDI and their characteristics are displayed in Table 2. Almost one in four patients (24%) had a brain malformation as their underlying cause of CDI. The average age of diagnosis for those with a congenital anatomic cause was 3.7 years (SD 5.0, median 1.2 y), with a wide range spanning infancy to adolescence (Table 2). Thirty-three percent of patients with a congenital anatomic cause were diagnosed with central DI after 2 years of age.

Table 2.

Patient Characteristics and Frequency of Anterior Pituitary Hormone Deficiencies by Diagnostic Groupings, Seattle Children's Hospital, 2000–2013

	n, %	Average Age at Diagnosis, y (range)	Idiopathic Diabetes Insipidus at Diagnosis, %	Any Anterior Hormone Deficiency	GH Deficiency, %	FSH/LH Deficiency, %	TSH Deficiency, %	ACTH Deficiency, %	Normal PPBS, %a
Anatomic brain malformations	36 (24%)	3.7 (0.02–17.2)	8 (3/36)	72 (26/36)	56 (20/36)	8 (3/36)	56 (20/36)	67 (24/36)	19 (n = 7)
Acquired infiltrative or tumor	82 (56%)	8.7 (1.2–24.4)	7 (6/82)	85 (70/82)	73 (60/82)	44 (36/82)	74 (61/82)	70 (57/82)	5 (n = 4)
Familial/genetic	11 (7%)	5.3 (0.2–10.8)	18 (2/11)	0 (0/11)	0 (0/11)	0 (0/11)	0 (0/11)	0 (0/11)	27 (n = 3)
Idiopathic	18 (12%)	6.1 (0.2–13.3)	100 (18/18)	39 (7/18)	44 (8/18)	17 (3/18)	28 (5/18)	17 (3/18)	28 (n = 5)
Total	147	7.1 (0.02–24.4)	20% (29/147)	70% (103/147)	60% (88/147)	29% (42/147)	59 (86/147)	57 (84/147)	13 (n = 19)

^a

A minimum estimate of PPBS prevalence with patients having missing data counted as having absence of the PPBS.

Acquired infiltrative and tumor etiologies were the most common cause of CDI (56.7%) (Table 2). Eleven patients (7%) had familial CDI or a presumed genetic cause. One pair of siblings was found to have a mutation in the neurophysin II moiety, preventing linkage with AVP (27). One patient with rapid-onset obesity, hypothalamic dysfunction, hypoventilation, autonomic dysregulation, and neuroendocrine tumor syndrome (28) was considered to have genetic CDI. A total of 18 patients (12.2%) were considered to have idiopathic CDI at their last assessment, and the average follow-up time from diagnosis was 6.2 years for these patients.

Outcomes of initially idiopathic CDI patients:

Twenty-nine patients were diagnosed with CDI in the absence of any medical conditions known to be associated with CDI and were evaluated primarily on the basis of polyuria and polydipsia (Table 2). Seven of the 29 patients (24%) had an abnormal initial MRI or clinical evaluation that immediately suggested an underlying cause of CDI. Of these seven patients, two patients were diagnosed with septooptic dysplasia on their initial MRI. One patient was diagnosed clinically with rapid-onset obesity, hypothalamic dysfunction, hypoventilation, autonomic dysregulation, and neuroendocrine tumor syndrome within months of their initial presentation. Another patient had multiple family members diagnosed with CDI after our initial evaluation and therefore was presumed to have a familial form of CDI. Three other patients were diagnosed with a germinoma, juvenile xanthogranuloma of the pituitary, or ectopic posterior pituitary based on their initial MRIs.

The remaining 22 patients had no known cause of CDI after their initial MRI and were considered initially idiopathic (Table 3). Ten of these patients (45%) had abnormal infundibular thickening. Three of these patients were later diagnosed with LCH (Table 3). One patient with an infundibular thickness at the upper limit of normal was unfortunately lost to follow-up after their initial MRI and was found to have a 19-mm germinoma on their second MRI 1.8 years later (Figure 1). The initial MRI met criteria for stalk thickening based on our criteria for this study. The other six patients with initial abnormal stalk thickening remain idiopathic with no underlying cause identified after an average of 5.8 years of follow-up. Notably, three of these six patients have had resolution or improvement of the stalk thickening over time.

Table 3.

Cause of Central Diabetes Insipidus in Those Patients With Initially Idiopathic CDI, by Initial Brain MRI Result (n = 22)

Diagnosis	Initial MRI Findings	Follow-Up to Diagnosis or End of Analysis	Remarks
Initial MRI with increased stalk thickness (n = 10)a
LCH (n = 3)	Infundibular thickening of 5.0–5.9 mm	0.5–2.3 y (mean 1.5)	Biopsy of pituitary stalk (n = 1), lumbar spine (n = 1), and skin (n = 1)
Germinoma (n = 1)	Infundibular thickening of 4.1 mm	1.8 y	Germinoma 1.8 y after CDI diagnosis
Idiopathic (n = 6)	Infundibular thickening of 3.5–5.1 mm	1.3–13.5 y (mean 5.8)	Stalk thickening resolved or decreased in three of six patients
Initial MRI with normal stalk thickness (n = 12)a
Idiopathic (n = 12)	PPBS present (n = 4), absent (n = 8)	0.6–15.3 y (mean 6.5)

Abbreviations: CDI, central diabetes insipidus; PPBS, posterior pituitary bright spot.

^a

P = .03 for association between initial stalk thickness and having an underlying diagnosis identified.

Figure 1.

Postcontrast sagittal T1 images demonstrate mild thickening of the pituitary stalk on initial MRI (A), with development of marked thickening on a follow-up study 18 months later in the same patient (B).

There were 12 patients (55%) with no initial cause of CDI and normal stalk thickness on the initial brain MRI (Table 3). These patients were followed up for an average of 6.5 years (range 0.6–15.3 y), and none have developed LCH, germinoma, or any other underlying cause.

In summary, of the 22 patients with initially idiopathic CDI, four were found to have an underlying cause (three with LCH, one with germinoma) an average of 1.6 years after the CDI diagnosis. By our stalk assessment methodology, the percentage of patients with an underlying cause was higher in those with initial pituitary stalk thickening compared with those without thickening [40% (4 of 10) vs 0% (0 of 12); Fisher's exact P = .03].

Pituitary function and the posterior pituitary bright spot

Anterior pituitary hormone deficiencies (APHDs) were present as frequently in those with anatomic malformations as those with tumor/infiltrative forms of CDI (Table 2; 72% vs 85% P = .09). APHDs were less common in the idiopathic group (39%) and completely absent in those with genetic/familial CDI. Of all patients with CDI, 60% had GH deficiency, 59% had TSH deficiency, and 57% had ACTH deficiency (Table 2). In our cohort at least 13% of our patients had a normal PPBS despite having a diagnosis of CDI (Table 2).

Discussion

Our study presents the largest cohort of pediatric CDI patients in the literature and is the first report on pediatric CDI from a large pediatric tertiary care center in the United States in more than 20 years (19).

We report more patients with anatomic brain malformations than almost all other previously published pediatric CDI cohorts (Table 4) (14–18). These patients represent an important subgroup of CDI patients because in addition to the possibility of impaired thirst mechanism they may also have limited access to water due to developmental delays. They may also have comorbidities that can make management of CDI more difficult such as seizure disorder and feeding tube requirement for failure to thrive, and we have demonstrated that their rate of APHD is similar to the high rates seen with tumor/infiltrative causes of CDI.

Table 4.

Causes of Central Diabetes Insipidus From Published Pediatric Cohorts Containing Brain Malformations (2000–2015)

Cause	Maghnie et al (18) (2000, n = 79)	Santiprabhob et al (14) (2005, n = 67)	Bajpai et al (15) (2008, n = 59)	Catli (16) (2012, n = 34)	Liu (17) (2013, n = 62)	Average	Present Cohort (n = 147)
Idiopathic	47%	10%	32%	29%	8%	25%	12%
Histiocytosis	15%	5%	19%	12%	19%	14%	12%
Craniopharyngioma	15%	21%	27%	21%	24%	22%	25%
Germ cell tumor	8%	12%	3%	12%	32%	13%	10%
Brain malformation	5%	30%	12%	9%	3%	12%	24%
Genetic/familial	6%	0%	0%	0%	0%	1%	8%
Infection/inflammation	1%	6%	3%	3%	3%	3%	2%
Trauma	3%	0%	0%	6%	0%	2%	1%
Other brain mass or cyst	0%	9%	3%	9%	5%	5%	5%

Bold indicates categories that were unique about the patient population.

The reason for our higher prevalence of malformations is unclear. There may be differences in our underlying patient population, such as higher birth rates and/or higher neonatal survival rates for children with brain malformations, or perhaps other differences in case ascertainment. An additional consideration is whether the availability of newer brain MRI techniques with improved resolution has allowed a higher rate of malformation detection in newer cohorts. This may contribute to the differences seen between our cohort and that described by Maghnie et al (18), whose patients were diagnosed between 1970 and 1996. The only other cohort with a similar rate of brain malformations contained patients diagnosed between 1996 and 2003 (14).

It was also interesting to note that the average age of CDI diagnosis for patients with a brain malformation was 3.7 years of age, and 33% of the patients were diagnosed after age 2 years. This is separate from the diagnosis of the brain malformation itself, which is often made within the first year of life. Many of these patients have developmental delays and may be diapered later into childhood, which could make polyuria more difficult to appreciate. Additionally, they may have a later diagnosis of ACTH deficiency, with hydrocortisone replacement unmasking CDI.

The percentage of patients with an infiltrative or tumor cause of CDI was similar to that reported by other groups (Table 4). One exception is the higher percentage of germ cell tumors in one cohort from Taiwan, but this is consistent with the higher rates of germ cell tumors described in some Asian populations (17, 29).

Our cohort features a lower percentage of patients with idiopathic CDI (12%) than many previously published studies (15, 16, 18) (Table 4). Therefore, we agree with previous reports (14, 17) that idiopathic CDI is rare in children. We report a lower rate of idiopathic DI than a recent report from Di Iorgi et al (20) (12% vs 51%). This is likely because the cases reported by Di Iorgi et al appear to have been identified based on an initial presentation of polyuria and polydipsia and did not appear to include patients with predisposing conditions for CDI. In support of this, we report a higher percentage of patients with brain malformations (24% vs 3.5%) and craniopharyngioma (25% vs 7%) (20). It is less clear why we report a much lower rate of idiopathic CDI compared with Maghnie et al (18) (12% vs 48%), particularly because our cohorts had a similar follow-up time (6.2 y vs 7.6 y). Based on our data, we believe that idiopathic CDI represents a small proportion of all children with CDI.

Of the 22 patients with initially idiopathic CDI, a cause was later identified in only four (18%), the latest of which occurred 2.3 years after the diagnosis of CDI. The finding that no concerning MRI changes were detected after 3 years from initial CDI diagnosis is consistent with a recent report (20). We would propose that after 3 years, the surveillance MRI frequency can be decreased. However, the development of clinical practice guidelines on the long-term MRI surveillance of patients with idiopathic CDI may require data from a larger cohort of idiopathic patients.

In our cohort, none of the patients with normal stalk thickening on their initial MRI had a later diagnosis of LCH and germinoma. Whereas our cohort size is not large enough to suggest such patients require less frequent MRI, it does highlight the possibility that a well-defined rigorous analysis of the pituitary infundibulum on the initial MRI may be a helpful prognostic indicator for patients with idiopathic CDI. To this point, one patient with an upper limit of normal infundibular thickness on the initial MRI report had clear thickening based on our criteria. Eighteen months later this patient had a diagnosis of germinoma (Figure 1).

Recently investigators have proposed classifying many idiopathic cases as hypophysitis based on serial MRIs that showed thickening and then normalization of the pituitary stalk over time (20). In support of this, three of the six patients with idiopathic CDI and initial stalk thickening had resolution or decrease of this over time. This is consistent with a transient, possibly inflammatory process in these patients, and they are possibly at lower risk for subsequent development of LCH or germinoma. Further follow-up with a larger cohort is necessary to confirm this assertion.

Some previous investigators did not report patients with genetic or familial causes of CDI (Table 4) (14–17), possibly due to limited availability of testing. It is likely that some of their patients classified as idiopathic would be found to have a genetic cause of CDI if this were investigated and could account for 7%–8% of the CDI population. None of the patients with genetic/familial CDI had APHDs, and we believe that idiopathic CDI patients with no APHDs or infundibular thickening should be prioritized for genetic testing.

As has been shown in other studies, our cohort demonstrates that anterior pituitary deficiencies are common with CDI and that many patients have more than one deficiency (15, 16, 20). It should be noted that it is difficult to draw conclusions on the true prevalence of APHDs in many patients with CDI. One must consider that some families may opt not to treat GH or FSH/LH deficiency in severely developmentally delayed children, and this may lower the recorded rates of these deficiencies in our cohort. These children are likely at higher risk for death prior to an age when gonadotropin deficiency can be diagnosed. Some children have yet to attain a pubertal age, and gonadotropin deficiency is often unknown in these cases. Additionally, some providers and families may opt to withhold GH therapy to theoretically prevent tumor growth and/or recurrence (30), obviating the need to diagnose GH deficiency in these patients for at least some period of time.

Previous reports have shown near universal absence of the PPBS with CDI, leading others to conclude that the presence of the spot may be related to neurosecretory granules containing AVP (18, 21, 22, 31, 32). In contrast, the PPBS is present in at least 19% of our patients with CDI due to anatomic malformations. Previous investigators have noted a higher prevalence of the PPBS in those with familial CDI and almost 30% of the patients with familial CDI in our cohort have a PPBS (23, 24). These results demonstrate that although a present PPBS is uncommon in patients with CDI, its presence should not exclude the diagnosis.

There are limitations to this report that deserve consideration. First, this was a retrospective study, which prevented us from coordinating a standardized assessment of each patient and likely resulted in missing data and larger loss to follow-up. The retrospective nature of this study did not allow for the prospective MRI evaluations performed by Di Iorgi et al (20), and it is also possible that some of our patients considered idiopathic will have an underlying diagnosis identified after additional follow-up. We had limited laboratory and imaging data from the time of diagnosis for some patients because their initial work-up occurred at an outside institution or because it occurred before 2000 when our institution began keeping electronic medical records. Our data on the PPBS were taken from radiology reports primarily, but a neuroradiologist did review many of the studies with missing or equivocal PPBS reports. We have not completed genetic testing in all of our patients that we have categorized as idiopathic CDI, meaning that some of these may be misclassified as such; 7 of 14 patients (50%) with idiopathic CDI and no genetic testing had additional anterior pituitary hormone deficiencies, so these are unlikely to have a genetic etiology for their CDI. Finally, it is possible that there are patients in our catchment area with CDI who were not evaluated by our endocrinology department, and it is possible that these patients would be less complicated than patients seen at SCH. This selection bias is possible but we believe is unlikely based on regional referral patterns for similar conditions.

This is the largest single-center cohort of patients with CDI described in the literature, and this larger size provides a more precise estimate of the prevalence of the multiple causes of CDI. We have confirmed our suspicion that idiopathic CDI is relatively rare and anatomic brain malformations are a more common cause of pediatric CDI. Patients with anatomic brain malformations have a high rate of APHDs, and many have persistence of the PPBS. Our follow-up of patients with idiopathic CDI shows that providers can consider less frequent MRI after 3 years from diagnosis and that having a standardized assessment of the infundibulum on the initial MRI may help predict later development of LCH or germinoma.

Acknowledgments

The primary author's fellowship and research is partially supported by a donation from the Nordstrom family.

Disclosure Summary: The authors have nothing to disclose.

Abbreviations

 
• APHD

  anterior pituitary hormone deficiency

 
• AVP

  arginine vasopressin

 
• ADH

  antidiuretic hormone

 
• CDI

  central diabetes insipidus

 
• LCH

  Langerhans cell histiocytosis

 
• MRI

  magnetic resonance imaging

 
• PPBS

  posterior pituitary bright spot

 
• SCH

  Seattle Children's Hospital.