Page 1
Acc
epte
d A
rtic
le
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/cen.12400 This article is protected by copyright. All rights reserved.
Received Date : 18-Sep-2013 Returned for Revision: 24-Oct-2013 Finally Revised Date : 22-Dec-2013 Accepted Date : 31-Dec-2013 Article type : D FOCAL CONGENITAL HYPERINSULINISM MANAGED BY MEDICAL
TREATMENT: A DIAGNOSTIC ALGORITHM BASED ON MOLECULAR
GENETIC SCREENING
Short title: Diazoxide-responsive focal hyperinsulinism
1Arianna Maiorana, 2Fabrizio Barbetti, 1Arianna Boiani, 3Vittoria Rufini, 4 Milena
Pizzoferro, 5Paola Francalanci, 6Flavio Faletra 7Colin G. Nichols, 8Chiara Grimaldi,
8Jean de Ville de Goyet, 9Jacques Rahier, 10Jean-Claude Henquin and 1Carlo Dionisi-
Vici
1Metabolic Unit , Department of Pediatrics, 2Laboratory of Mendelian Diabetes, 4Department
of Nuclear Medicine , 5Department of Pathology, 8Liver Surgery and Transplantation Unit,
IRCCS Bambino Gesù Children’s Hospital, Piazza S. Onofrio 4, 00165, Rome, Italy;
2Department of Experimental Medicine and Surgery, University of Tor Vergata, Section D,
Room 118, Viale Oxford 81, 00133, Rome, Italy; 3Department of Nuclear Medicine,
Catholic University of the Sacred Heart, Via Eugenio Tansi 67, 00135, Rome, Italy; 6Institute
for Maternal and Child Health, IRCCS Burlo Garofalo, Via dell'Istria 65, 34137, Trieste,
Italy; 7Department of Cell Biology and Physiology, Washington University School of
Medicine, St. Louis, Missouri, USA; 9Department of Pathology and 10Unit of Endocrinology
Page 2
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
and Metabolism, Faculty of Medicine, University of Louvain UCL 55.30, Avenue Hippocrate
55, B-1200, Brussels, Belgium.
Corresponding author: Arianna Maiorana, Metabolic Unit, Bambino Gesù Children's
Hospital, Piazza S.Onofrio 4, 00165, Rome, Italy; fax +390668592791: e-mail address:
[email protected]
Key Words: Congenital Hyperinsulinism; Hypoglycemia; Diazoxide; KATP-channel
Declaration of interest and financial disclosure: Nothing to declare
Abstract
Objective: Congenital hyperinsulinism (CHI) requires rapid diagnosis and treatment to avoid
irreversible neurological sequelae due to hypoglycemia. Etiological diagnosis is instrumental
in directing the appropriate therapy. Current diagnostic algorithms provide a complete set of
diagnostic tools including 1) biochemical assays, 2) genetic facility and 3) state-of-the-art
imaging. They consider the response to a therapeutic diazoxide-trial an early, crucial step
before proceeding (or not) to specific genetic testing and eventually imaging, aimed at
distinguishing diffuse versus focal CHI. However, interpretation of the diazoxide-test is not
trivial and can vary between research groups, which may lead to inappropriate decisions.
Objective of this report is proposing a new algorithm in which early genetic screening, rather
than diazoxide trial, dictates subsequent clinical decisions.
Patients, Methods and Results. Two CHI patients weaned from parenteral glucose infusion
and glucagon after starting diazoxide. No hypoglycemia was registered during a 72-h
Page 3
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
continuous glucose monitoring (CGMS) or hypoglycemic episodes were present for no
longer than 3% of 72-h. Normoglycemia was obtained by low-medium dose diazoxide
combined with frequent carbohydrate feeds for several years. We identified monoallelic,
paternally inherited mutations in KATP-channel genes, and 18F-DOPA PET-CT revealed a
focal lesion that was surgically resected, resulting in complete remission of hypoglycemia.
Conclusions: Although rare, some patients with focal lesions may be responsive to
diazoxide. As a consequence, we propose an algorithm that is not based on a “formal”
diazoxide response but on genetic testing, in which patients carrying paternally inherited
ABCC8 or KCNJ11 mutations should always be subjected to 18F-DOPA PET-CT.
Introduction
Congenital hyperinsulinism (CHI), characterized by uncontrolled or excessive insulin
secretion for the prevailing glucose levels, is the most frequent cause of severe and persistent
hypoglycemia in the neonatal period and early infancy. Patients present with recurrent
episodes of profound hypoglycemia requiring rapid and intensive treatment with dextrose
infusions and intravenous glucagon to prevent neurological sequelae1-4. Mutations in 9 genes
(ABCC8, KCNJ11, GCK, GLUD1, HADH, HNF1A, HNF4Α, SLC16A1, UCP2) have so far
been associated with CHI3-6. Loss-of-function mutations in ABCC8 and KCNJ11, which
respectively encode the SUR1 and Kir6.2 subunits of the pancreatic ATP-sensitive potassium
(KATP) channel, account for more than 50% of CHI cases and are associated with two
histological aspects of the endocrine pancreas: a diffuse form, which is inherited as either an
autosomal recessive or dominant trait and affects all beta cells, and a focal form, which
results from the combination of a paternally inherited germinal mutation and a somatic loss of
heterozygosity of the maternal allele in a restricted group of beta cells7.
[18F]dihydroxyphenylalanine positron emission tomography (18F-DOPA PET) imaging is
Page 4
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
useful to differentiate between the two forms, and to precisely localize focal lesions in the
pancreas8 or in ectopic sites9. Chronic medical treatment to prevent recurrence of
hypoglycemia may include drugs, such as diazoxide (a KATP channel opener) or octreotide (a
long-lasting somatostatin analogue), sometimes in association with frequent carbohydrate-
enriched feeding1-3. Diffuse CHI is first approached with conservative treatment, but when
medical measures are ineffective, near-total pancreatectomy is required, and is associated
with increased risk of developing diabetes. Focal CHI has so far been considered
unresponsive to diazoxide1-3,10,11, and the outcome of a therapeutic diazoxide trial, whose
modalities differ slightly between research groups, represents the critical decision step of
current diagnostic algorithms, that avoids 18F-DOPA PET-CT in those patients whose
hypoglycaemia can be prevented by diazoxide1-3,11. Here, we report two patients with focal
CHI due to paternally inherited mutations in KATP genes, in whom low-medium dose
diazoxide succeeded - as detected by repeated continuous glucose monitoring evaluations
(CGMS) - in maintaining normoglycemia before 18F-DOPA PET-CT localization and
elective resection of the focal lesion. Therefore, we suggest a new algorithm in which genetic
screening, instead of the response to diazoxide, dictates the subsequent decision tree.
Materials and Methods
Ethics
The study was approved by the Ethical Committees of Bambino Gesù Children's Hospital of
Rome, Italy, the Catholic University of Sacred Heart of Rome, Italy, and the University of
Louvain Faculty of Medicine, Brussels, Belgium.
Page 5
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
Continuous glucose monitoring system
Blood glucose control was assessed by pre-meal glucometer measurements, and by 72-h
continuous glucose monitoring (CGMS) (Medtronic; data analysis by MiniMed software)
during the first hospital admission, then at the beginning of treatment, monthly during the
following 3 months, and every 3-6 months afterwards. In general, maintenance of pre- and
post-meal glucose levels between 50 and 150 mg/dl (2.8-8.2 mmol/l), with an average of 70
mg/dl (3.9 mmol/l), was considered as good glycemic control. For this study, we defined a
subject as responsive to diazoxide if his/her glucose was below 50 mg/dl (2.8 mmol/l) for less
than 3% of the time during a 72-h CGMS session (i.e. equal or less than 130 minutes in 3
days) and less than 30 minutes consecutively per single episode. Because CGMS sensitivity
depends on the type of glucometer used for sampling, our Centre compared the accuracy of
blood glucose determination by different glucometers versus plasma glucose (automatically
corrected for the hematocrit) and identified the more sensitive and accurate device that was
selected for the use in our ward. Performing more than 900 GCMS in the last 5 years, we
acquired a fairly good level of performance and of interpretation of data, which makes us
confident about the efficacy and safety of CGMS.
Genetic analysis
Germline mutations were searched in patients and their parents on DNA extracted from
peripheral leukocytes by DNeasy Tissue Isolation kit (Qiagen, Valencia, CA).
ABCC8,KCNJ11,GCK, HNF4A genes were amplified by PCR and directly sequenced.
Genetic analysis of KCNJ11 gene has been performed as described in ref. 16 and analysis of
ABCC8 gene as described in ref. 20.
Page 6
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
18F-DOPA PET-CT scan
Anesthetized patients (fasted for at least 6h, and receiving i.v. glucose) underwent a PET
scan of the abdomen using a hybrid machine (Gemini GXL, Philips Medical Systems) 45 min
after i.v. administration of 18F-DOPA (4 MBq/Kg). A contrast-enhanced CT was also
sequentially performed to accurately localize the focal lesion on the basis of the vascular
map. To back up the visual identification of the focus, a standardized uptake value (SUV)
was obtained. A SUV ratio >1.2 was considered indicative of a focal lesion12.
Histology
Pancreatic specimens were obtained intra-operatively and prepared as previously described13.
Immunohistochemistry for CDKN1C peptide (p57) was subsequently performed on formalin-
fixed, paraffin-embedded tissue14.
In vitro functional studies of pancreas
On the basis of intra-operative examinations, the pathologist sampled small fragments of
healthy pancreas and focal lesion for functional studies. The two fragments, immersed in
culture medium at 4°C, were transported from Rome to Brussels within 10 hours. They were
then treated as recently described in detail, to measure the dynamics of insulin secretion and
estimate the insulin content of the tissue15.
Results
Patient 1
Clinical history
The clinical history of patient 1 has been reported16. Briefly, she was born with normal
weight at 38 weeks of gestation from unrelated parents. At the age of 2 months, she displayed
Page 7
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
cyanosis and tremor. The diagnosis of CHI was made on the basis of hypoketotic (<0.05
µmol/l) fasting hypoglycaemia (40 mg/dl, 2.2 mmol/l), relative hyperinsulinemia (31 µU/ml),
low free fatty acids (220 µmol/l), and good response to glucagon. At presentation, she needed
intravenous glucose infusion, glucagon infusion and enteral feeding to maintain
normoglycemia. Treatment with diazoxide (15 mg/kg/day, in three daily doses) and low-dose
chlorothiazide (2 mg/day, to avoid fluid retention) was started, allowing immediate weaning
from glucagon and parenteral glucose infusions. The patient was discharged on a diazoxide
dose of 7.5 mg/kg/day and a low amount of milk in enteral nutrition (12 ml/h) to reassure the
family. After about two months, nocturnal enteral feeding was stopped and free diet was
started. During that period, glucometer measurements were performed before each meal,
during night and before the first meal of the next day (i.e. after nocturnal fasting), for more
than five consecutive days, by parents at home, while the patient was on diazoxide, and by
nurses during subsequent hospitalization. Glycaemic values were consistently >3 mmol/L. In
addition, more refined assessment of glycemic control was obtained by 72-h CGMS (about
200 blood glucose determinations per day) during hospitalization; after discharge, the
procedure was repeated as stated in Methods. Four months after hypoglycaemia onset,
CGMS showed no hypoglycemic episodes (diazoxide dose 7 mg/kg/day), even between
nocturnal meals (Fig.1). This result was confirmed at 72-h CGMS sessions performed 12 and
24 months after discharge, while the patient was on diazoxide. Thus, the patient was
responsive at almost half the recommended initial diazoxide dose3,17,18, i.e. 10 mg/kg/day for
an infant. Subsequently, only single asymptomatic episodes of borderline hypoglycemia
(between 40-50 mg/dl, 2.2-2.8 mmol/l ) were recorded in each 3-days session of CGMS
performed 12, 8 and 3 months before surgery and which prevented us from withdrawing
diazoxide, but which progressively tapered off to 4.0 mg/kg/day (Fig. 1 suppl).
Page 8
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
Mutational analysis
A heterozygous KCNJ11/V290M mutation was found in the proband and in her unaffected
father16. DNA analysis of the mother was negative. No mutations in ABCC8, GCK and
HNF4Α genes were detected.
18F-DOPA PET-CT scan, surgery and histology
18F-DOPA PET-CT, performed at the age of 4 years and 11months, documented a focal
lesion in the pancreatic body, which was surgically removed 2 months later16. Histology on
intraoperative frozen pieces and postoperative fixed specimens was typical of a focal lesion.
CDKN1C (p57) was negative within the lesion, but positive in islets outside the lesion. This
pattern14,19 is consistent with loss of maternal 11p15.5 as the cause of a focal islet cell
adenomatous hyperplasia7,14. After operation the patient no longer required medical therapy
as attested by CGMS16.
In vitro insulin secretion
In healthy pancreas, insulin secretion rate was low in 1mM glucose, and 15mM glucose
induced a large biphasic secretory response that was abolished by subsequent addition of
diazoxide and eventually restored by tolbutamide (Fig.2). In tissue from the focal lesion, the
rate of insulin secretion was very high in 1mM glucose and paradoxically decreased slightly
on stimulation with 15 mM glucose, whereas 100 μM diazoxide and 100 μM tolbutamide had
no effect. The estimated insulin content of the focal lesion was ~50-fold higher than that of
normal pancreas (246.0 vs 5.0 ng/mg).
Page 9
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
Patient 2
Clinical history
Patient 2 was a female, born with normal weight at 41 weeks of gestation, from unrelated
parents. At the age of 3 months she first manifested generalized seizures. On the basis of
hypoketotic (<0.05 µmol/l) hypoglycemia (32 mg/dl,1.75 mmol/l), associated with
suppressed free fatty acids (108 µmol/l), detectable insulin (3.8 µU/ml) and appropriate
glycemic response to glucagon injection, a diagnosis of CHI was made. Intravenous infusion
providing 15 mg/kg/min of glucose was initially necessary to maintain euglycemia.
Thereafter, diazoxide (12 mg/kg/day divided in three doses) was started, allowing a gradual
withdrawal of parenteral treatment. The patient also received low-dose chlorothiazide (2
mg/day) to prevent fluid retention. During the first seven months after diagnosis the child
manifested feeding difficulties, gastroesophageal reflux, with frequent vomiting, which
complicated management of disease because meals were not completed and administration of
therapy was unsafe. This situation prompted the use of drip feeding with a carbohydrate rich
diet to minimize the risk of hypoglycaemia.
Follow up during long-term diazoxide treatment and on a free diet was performed by
CGMS, with the same scheme as in patient 1. As shown by a representative CGMS profile
obtained after 1.8 years of treatment (Fig.3), average glycemia was 90 mg/dl (5 mmol/l),
with a single recorded episode just below the threshold of 50 mg/dl (48 mg/dl, 2.6 mmol/l)
of 10 minutes duration (corresponding to <0.25% of 3-days CGMS registration) while
receiving diazoxide at a dose of 10 mg/kg/day. Over 2.5 years of follow-up, blood glucose
remained unchanged as detected at CGMS, and diazoxide was tapered to 6.6 mg/kg/day right
before surgery, whilst the child was on a free diet.
Page 10
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
Genetic analysis
DNA sequencing identified a frameshift mutation c.1580_1581dupGG, p.K528Gfs*4 in the
ABCC8 gene20 which produces a premature stop codon with generation of a truncated SUR1
protein. The same mutation was found in the asymptomatic father whereas the mother
showed a wild type ABCC8 sequence. No mutations were found in KCNJ11, GCK and
HNF4A genes.
18F-DOPA PET-CT scan, surgery and histology
18F-DOPA PET-CT performed at the age of 2 years and 3 months showed a large focal lesion
of the pancreatic head (Fig.4A). Removal by elective surgery 4 months later led to rapid
recovery of the patient to euglycemia, as ascertained by CGMS performed at 1.5 months after
surgery, while she was on normal diet and free of diazoxide. The resected lesion appeared as
a polyp at the surface of the pancreas (Fig.4B), and intra-operative examination of frozen
sections showed a typical picture of focal CHI13 (Fig.4C). The lesion, poorly delimited,
seemed to result from the confluence of apparently normally structured islets separated by a
few exocrine acini maintaining an organoid pattern. Nuclei appeared polymorphous with size
variations and hyperchromatism. Outside the lesion, islets were small, containing beta-cells
with a small nucleus and scanty cytoplasm, both characteristic of non-stimulated beta cells.
Post-operative immunohistochemistry on fixed tissue showed that CDKN1C peptide labelled
outside but not within the lesion.
In vitro insulin secretion
In vitro insulin secretion by healthy pancreatic tissue adjacent to the focal lesion was
stimulated ~2-fold when the glucose concentration was increased from 1 to 15mM (Fig.2).
Addition of diazoxide (to open KATP channels) inhibited the effect of glucose, and subsequent
Page 11
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
addition of tolbutamide (to close KATP channels) antagonized this inhibition in a reversible
manner. Again, characteristic of tissue lacking KATP channels, insulin secretion was not
stimulated by high glucose or affected by any of the two drugs (Fig.2), and the estimated
insulin content of the focal lesion was ~100-fold higher than that of healthy pancreas (426.0
vs 4.3 ng/mg).
Discussion
We describe two patients with focal CHI, both of whom achieved good metabolic control
with low-medium dose diazoxide therapy, according to our criteria of diazoxide
responsiveness. Such medical treatment is currently deemed adequate in mild, diffuse forms
of CHI caused by dominant mutations in ABCC8 and KCNJ11 genes, which lead to partial
inactivation of KATP channels, or by mutations in other genes1. In contrast, patients with focal
CHI frequently do not respond to even a maximal diazoxide dosage (as shown in Fig.5),
which most of the leading research groups on CHI have set at 15 mg/kg/day for neonates and
10 mg/kg/day for infants1-4. Clinical criteria to assess diazoxide responsiveness may vary, but
a plasma glucose constantly equal to or above 60-70 mg/dl (3.3-3.8 mmol/l) without support
of carbohydrate-enriched diet appears to be the gold standard for most groups; in addition,
some researchers advocate the use of fasting for a variable duration at the end of a 5-days
diazoxide trial to definitely classify patients as responders1,3,4,11. Conversely, non-responders
are defined as those with at least two confirmed blood glucose values below 3 mmol (i.e. 54
mg/dl) in a 24h period, after 5 days of diazoxide4,10. Algorithms proposed by these groups
prescribe that diazoxide-responders should not be subjected to 18F-DOPA imaging2,3,11. These
algorithms do not prescribe the frequency at which blood glucose determinations should be
performed during the diazoxide-test and we believe that rare (say 1-2 per day) glucose values
below 60 mg/dl (3.3 mmol/l), or even short, asymptomatic, severely hypoglycemic (< 50
Page 12
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
mg/dl, < 2.8 mmol/l) episodes could have been missed when glucose determinations are
performed by standard frequency sampling. For this reason we have implemented the use of
CGMS in patients with CHI in our metabolic ward in 2003. CGMS, with over 200
determinations/day, allows detection of low glucose values of quite short duration,
independent of meals or period of the day (i.e. sleep hours, fed/fasting state). As a result,
CGMS allows a more detailed determination of daily glucose profiles and better assessment
of the outcome of diazoxide therapy. Relying on “frequent sampling”, instead of on CGMS to
monitor glucose, might easily miss asymptomatic and/or short hypoglycemic episodes when
evaluating CHI patients at follow-up, but in our two patients, numerous 72-h CGMS sessions
either certified no hypoglycemic episodes (defined as glucose < 50 mg/dl, < 2.8 mmol/l) or
borderline low glucose values (50-54 mg/dl, 3-5 mmol/l) that represented less than 3% of the
total glucose determinations, with single episodes no longer than 30 minutes. On these bases,
we considered these two focal CHI patients to be diazoxide-responsive. Their therapeutic
outcome was fully divergent from other patients with focal lesion we observed, who failed to
respond at all to maximal doses of diazoxide (Fig.5).
These observations raise questions regarding definition of clinical responsiveness to
diazoxide. Our managing approach to the 86 CHI patients that we have followed during the
last 20 years was inspired by Glaser et al.21,22, who advocated medical treatment for all CHI
patients whenever possible. To this end, we have aggressively pursued diazoxide therapy, and
have considered all patients who weaned from parenteral glucose infusion and/or i.v.
glucagon, who stop presenting sustained hypoglycaemia after five days of diazoxide at initial
maximal dose combined, if needed, with nutritional measures, to be diazoxide-sensitive.
Responsiveness to diazoxide was then confirmed by CGMS. With this approach, only 21% of
our patients with instrumental and genetic diagnosis of diffuse CHI underwent near-total
pancreatectomy (Maiorana A. and Dionisi-Vici C. unpublished observations), in keeping with
Page 13
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
the 28% reported by Glaser et al.22. We also consider that, owing to its side-effects, diazoxide
should be tailored to the minimal effective dose. Hence, in the second patient, the initial
treatment combined a lower dose of diazoxide and an enriched carbohydrate diet, eliciting a
satisfactory clinical outcome. The efficacy of the treatment was then assessed by CGMS,
rendering fasting tolerance tests unnecessary. It also avoided repeated blood sampling to
measure free fatty acids and ketone bodies23, both of which may be of questionable utility
during diazoxide therapy and are of discomfort for patients, especially in infancy and
childhood. These considerations are important since, according to current algorithms, the
clinical determination of diazoxide-responsiveness is instrumental for subsequent execution
of 18F-DOPA PET-CT2,3,11. Only recently, the group from Great Ormond Street Hospital for
Children in London, UK, has slightly changed the decision tree, advocating 18F-DOPA PET-
CT for diazoxide-responsive patients, but only when paternally inherited heterozygous
mutations of KATP are detectable24.
The in vitro secretory behaviour of pancreatic fragments from our two patients was compared
with that of 18 diazoxide-resistant focal cases17. This large series showed that control
pancreas adjacent to the lesion secretes insulin in similar pattern to normal islets25 in spite of
the presence of a mutated allele, a result that fits in with the lack of symptoms in the
heterozygous fathers. In our two cases insulin secretion was also qualitatively normal in
healthy pancreatic tissue but secretion and content were very different in tissue from within
the lesion. In both patients, insulin secretion from the focal lesion was not stimulated (even
slightly decreased) in high glucose, and was unaffected by diazoxide and tolbutamide. These
characteristics have been observed in many previously reported diazoxide-resistant focal
cases with mutations in ABCC8 or KCNJ1117. The insulin content of both focal lesions was
very high, within the range of previously reported cases, whereas the insulin content of the
adjacent healthy pancreas was much lower17.
Page 14
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
The rapidity of blood glucose control at onset of the treatment can be interpreted as
efficient overall inhibition of insulin secretion from the intact pancreas by diazoxide in both
patients since the only other drug present was chlorothiazide, which does not affect insulin
secretion from normal beta-cells or from beta-cells lacking KATP channels26.
The discrepancy between these clinical findings and the lack of inhibition of secretion by
diazoxide in focal tissue in vitro is thus the reverse of situations in which diazoxide was
ineffective in vivo at maximally tolerable doses but able to open mutated KATP channels at
high concentrations in vitro29. In children (0.5 – 5 years)27 taking diazoxide at doses of 4-7
mg/kg, plasma concentrations of the drug remained below 200 µmol/l, of which >85% is
bound to albumin28. As we used diazoxide at 100 µmol/l in the presence of a ~40-fold lower
concentration of albumin than in plasma, the free drug concentration was not plausibly higher
in vivo than in vitro. Notably, the expected phenotype of the mutation, at least in patient 1, is
reduced – but not complete lack of - functional KATP channel activity16. It is therefore
possible that, owing to their complex regulation by intracellular nucleotides and other factors,
mutated channels maintained a weak sensitivity to diazoxide in vivo, which escaped detection
during strong glucose stimulation in vitro. One should also bear in mind that the global
therapeutic response depends on the balance between insulin secretion from diazoxide-
sensitive normal islets and abnormal β-cells within the lesion. Our findings cannot be
considered totally unusual. In a previous cohort of 35 diazoxide-sensitive patients, half were
found to exhibit focal forms of CHI at surgery17. Diazoxide-sensitive CHI displaying a focal
aspect by PET imaging has recently been described30. However, in contrast to our two cases,
no mutation in ABCC8 or KCNJ11 or loss of heterozygosity was found in these patients.
Their β-cell hyperfunction could also be reversed by diazoxide in vitro, because it was caused
by an activating mutation in GCK or undue expression of hexokinase I in a subset of islets
concentrated in a few lobules of their pancreas31. In addition, Kapoor et al. recently described
Page 15
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
partial or complete diazoxide responsiveness in some patients carrying compound
heterozygous, and recessive heterozygous ABCC8 mutations32. Remarkably, focal CHI may
be clinically heterogeneous, with sometimes responsiveness or resistance to diazoxide in
patients with the same mutation in KATP channels33. Alternatively, unidentified genetic
factors may modulate the expression of the phenotype, or off-target actions of diazoxide may
be responsible for the in vivo responsivity.
In an effort to rationalize the genetic diagnostic screening of patients with CHI, most
algorithms rely on the diazoxide test to distinguish between responders and non-responders1-
3,11, and hence whether or not to perform 18F-DOPA PET-CT. Nevertheless, at least two
major research groups (re)introduced KATP genes sequencing at a subsequent step of their
algorithms2,3. All groups, including ours, have found that most patients with CHI carry a
KATP mutation20,32,34 and it is conceivable that most groups agree in principle regarding the
usefulness of molecular genetic studies in all patients with CHI, both for appropriate
therapeutic decisions and genetic counselling in familial cases. For these reasons, we propose
a revised algorithm, based on a reasoned combination of metabolic investigation and genetic
analysis, that does not utilize diazoxide response at the beginning of the decision tree (Fig.6).
According to this algorithm, we sequenced the ABCC8 and KCNJ11 genes first, and then
sequenced the GCK and HNF4A genes in those negative for KATP gene mutations, in 51
patients that were negative to specific metabolic screening indicative of GLUD1, HADH or
CDGs genes defects. We identified 23 KATP mutations16,20 ( Maiorana A., Dionisi-Vici C.,
Barbetti F., unpublished observations). Among these 23, 6 heterozygous, paternally inherited
mutations were found, including the two described in the present study. In all 6 a focal lesion
was identified by 18F-DOPA PET-CT, and all underwent successful surgery and were
effectively cured16,20 ( Maiorana A., Barbetti F., Dionisi-Vici C., unpublished observations).
Page 16
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
We acknowledge that sometimes patients with a focal form of CHI may experience a
progressive decrease in the severity and an eventual cessation of hypoglycemic episodes after
several years of medical treatment, an improvement attributed to apoptosis of beta-cells
within the lesion3,35. However, we would recommend that surgery, in the hands of an
experienced surgeon, should be the elective indication for focal lesions regardless of medical
responsiveness because it allows complete remission of the disease, reduces the clinical
burden and, by relieving the discomfort of chronic medical therapy (and attendant side-
effects) requiring strict monitoring, improves the quality of life of patients and their families.
In conclusion, some patients with focal lesions can be responsive to diazoxide. CGMS
is a very useful tool in evaluating and monitoring therapy outcome, demonstrating the
existence of “boarder-line” situations, with fluctuations of metabolic control during
diazoxide treatment. Defining “partial” sensitivity to diazoxide is not trivial and may lead to
inappropriate diagnostic and therapeutic decisions, whereas responsiveness to diazoxide does
not exclude focal forms17,30-32. Based on our clinical observations, we propose a new
algorithm that is not based on a “formal” diazoxide trial but on genetic testing, one in which
patients carrying paternal ABCC8 or KCNJ11 mutations should then be subjected to 18F-
DOPA PET-CT, as well as those showing a wild type pattern of the known genes associated
with CHI.
Acknowledgments
We are grateful to M. Nenquin for performing the in vitro experiments.
Page 17
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
References
1. De León D.D., Stanley C.A. (2007) Mechanisms of Disease: advances in diagnosis and
treatment of hyperinsulinism in neonates. Nature Clinical Practice. Endocrinology &
Metabolism, 3, 57-68.
2. Kapoor R.R., James C., Hussain K. (2009) Advances in the diagnosis and management of
hyperinsulinemic hypoglycaemia. Nature Clinical Practice. Endocrinology & Metabolism, 5,
101-112.
3. Arnoux J.B., Verkarre V., Saint-Martin C., et al. (2011) Congenital hyperinsulinism:
current trends in diagnosis and therapy. Orphanet Journal of Rare Disease, 6, 63.
4. Arnoux JB, De Lonlay P, Ribeiro MJ, et al. (2010). Congenital hyperinsulinism. Early
Human Development, 86, 287-294.
5. Stanescu D.E., Hughes N., Kaplan B., et al. (2012) Novel Presentations of Congenital
Hyperinsulinism due to Mutations in the MODY genes: HNF1A and HNF4A. Journal of
Clinical Endocrinology and Metabolism, 97(10):E2026-30. doi: 10.1210/jc.2012-1356.
6. Marquard J., Palladino A.A., Stanley C.A., et al. (2011) Rare forms of congenital
hyperinsulinism. Seminars in Pediatric Surgery, 20, 38-44.
7. De Lonlay P., Fournet J.C., Rahier J., et al. (1997) Somatic deletion of the imprinted 11p15
region in sporadic persistent hyperinsulinemic hypoglycemia of infancy is specific of focal
adenomatous hyperplasia and endorses partial pancreatectomy. Journal of Clinical
Investigation, 100, 802–807.
8. Otonkoski T, Näntö-Salonen K., Seppänen M., et al. (2006) Noninvasive diagnosis of focal
hyperinsulinism of infancy with [18F]-DOPA positron emission tomography. Diabetes, 55,
13-18.
Page 18
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
9. Hussain K., Seppanen M., Nanto-Salonen K., et al. (2006) The diagnosis of ectopic focal
hyperinsulinism of infancy with [18F]-dopa emission tomography. Journal of Clinical
Endocrinology and Metabolism, 91, 2839-2842.
10. Bellanné-Chantelot C., Saint-Martin C., Ribeiro M.J., et al. (2010) ABCC8 and KCNJ11
molecular spectrum of 109 patients with diazoxide-unresponsive congenital hyperinsulinism.
Journal of Medicine Genetics, 47, 752-759.
11. Palladino A.A., Stanley C.A. (2011) A specialized team approach to diagnosis and
medical versus surgical treatment of infants with congenital hyperinsulinism. Seminars in
Pediatric Surgery, 20, 32-37.
12. Ribeiro M.J., Boddaert N., Bellanné-Chantelot C., et al. (2007) The added value of
[18F]fluoro-L-DOPA PET in the diagnosis of hyperinsulinism of infancy: a retrospective
study involving 49 children. European Journal of Nuclear Medicine and Molecular Imaging,
34, 2120-2128.
13. Rahier J., Sempoux C., Fournet J.C., et al. (1998) Partial or near-total pancreatectomy for
persistent neonatal hyperinsulinaemic hypoglycaemia: the pathologist's role. Histopathology,
32, 15-19.
14. Kassem S.A., Ariel I., Thornton P.S., et al. (2001) p57(KIP2) expression in normal islet
cells and in hyperinsulinism of infancy. Diabetes, 50, 2763-2769.
15. Henquin J.C., Nenquin M., Sempoux C., et al. (2011) In vitro insulin secretion by
pancreatic tissue from infants with diazoxide-resistant congenital hyperinsulinism deviates
from model predictions. Journal of Clinical Investigation, 121, 3932-3942.
16. Loechner K.J., Akrouh A., Kurata H.T., et al. (2011) Congenital hyperinsulinism and
glucose hypersensitivity in homozygous and heterozygous carriers of Kir6.2 (KCNJ11)
mutation V290M mutation: K(ATP) channel inactivation mechanism and clinical
management. Diabetes, 60, 209-217.
Page 19
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
17. Touati G., Poggi-Travert F., Ogier de Baulny H., et al. (1998) Long-term treatment of
persistent hyperinsulinaemic hypoglycaemia of infancy with diazoxide: a retrospective
review of 77 cases and analysis of efficacy-predicting criteria. European Journal of
Pediatrics, 157, 628-633.
18. MacMullen C.M., Zhou Q., Snider K.E., et al. (2011) Diazoxide-unresponsive congenital
hyperinsulinism in children with dominant mutations of the β-cell sulfonylurea receptor
SUR1. Diabetes, 60, 1797-1804.
19. Sempoux C., Guiot Y., Dahan K., et al. (2003) The focal form of persistent
hyperinsulinemic hypoglycemia of infancy: morphological and molecular studies show
structural and functional differences with insulinoma. Diabetes, 52, 784–794.
20. Faletra F., Athanasakis E., Morgan A., et al. (2013) Congenital hyperinsulinism: clinical
and molecular analysis of a large Italian cohort. Gene, 521,160-165.
21. Glaser B., Ryan F., Donath M., et al. (1999) Hyperinsulinism caused by paternal-specific
inheritance of a recessive mutation in the sulfonylurea-receptor gene. Diabetes, 48, 1652-
1657.
22. Glaser B., Landau H., Permutt M.A. (1999) Neonatal Hyperinsulinism. Trends in
Endocrinology and Metabolism, 10, 55-61.
23. Wolfsdorf J.I., Sadeghi-Nejad A., Senior B. et al (1984) Ketonuria does not exclude
hyperinsulinemic hypoglycemia. American Journal of Diseases of Children, 138, 168-171.
24 . Senniappan S., Shanti B., James C., et al. (2012) Hyperinsulinemic hypoglycaemia:
genetic mechanisms, diagnosis and management. Journal of Inherited Metabolic Disease,
35(4):589-601. doi: 10.1007/s10545-011-9441-2.
25. Henquin J.C., Dufrane D., Nenquin M. (2006) Nutrient control of insulin secretion in
isolated normal human islets. Diabetes, 55, 3470-3477.
Page 20
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
26. Szollosi A., Nenquin M., Henquin J.C. (2010) Pharmacological stimulation and inhibition
of insulin secretion in mouse islets lacking ATP-sensitive K+ channels. British Journal of
Pharmacology, 159, 669-677.
27. Pruitt A.W., Dayton P.G., Patterson J.H. (1973) Disposition of diazoxide in children.
Clinical Pharmacology and Therapeutics, 14, 73-82.
28. Panten U., Burgfeld J., Goerke F., et al. (1989) Control of insulin secretion by
sulfonylureas, meglitinide and diazoxide in relation to their binding to the sulfonylurea
receptor in pancreatic islets. Biochemical Pharmacology, 38, 1217-1229.
29. Dunne M.J., Cosgrove K.E., Shepherd R.M., et al. (2004) Hyperinsulinism in infancy:
from basic science to clinical disease. Physiological Reviews, 84, 239-275.
30. Sempoux C., Capito C., Bellanné-Chantelot C., et al. (2011) Morphological mosaicism of
the pancreatic islets: a novel anatomopathological form of persistent hyperinsulinemic
hypoglycemia of infancy. Journal of Clinical Endocrinology and Metabolism, 96, 3785-3793.
31. Henquin J.C., Sempoux C., Marchandise J., et al. (2013) Congenital hyperinsulinism
caused by hexokinase I expression or glucokinase-activating mutation in a subset of β-cells.
Diabetes, 62(5):1689-1696. doi: 10.2337/db12-1414.
32. Kapoor R.R., Flanagan S.E., Arya V.B., et al. (2013) Clinical and molecular
characterization of 300 patients with congenital hyperinsulinism. European Journal of
Endocrinology, 168, 557-64.
33. Ismail D., Kapoor R.R, Smith V.V., et al. (2012) The heterogeneity of focal forms of
congenital hyperinsulinism. Journal of Clinical Endocrinology and Metabolism, 97(1), E94-
99. doi: 10.1210/jc.2011-1628.
34. Snider K.E., Becker S., Boyajian L., et al. (2013) Genotype and phenotype correlations in
417 children with congenital hyperinsulinism. Journal of Clinical Endocrinology and
Metabolism, 98(2), E355-363. doi: 10.1210/jc.2012-2169.
Page 21
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
35. Hussain K.(2005) Congenital hyperisulinism. Seminars in Fetal & Neonatal Medicine,
10, 369-376.
Figures
Figure 1. Near normal blood glucose levelsachieved by medical treatment in patient 1. Blood
glucose was monitored using the 72-h continuous glucose monitoring system (CGMS) 1
month after the first discharge. CGMS showed a glycemic range of 56-106 mg/dl (3.1-5.8
mmol/l) and an average glucose value of 75 mg/dl (4.1 mmol/l) with no episode of
hypoglycemia. Diazoxide dosage was 7 mg/kg/day. Open stars indicate simultaneous
glucometer values; filled stars in indicate meal event.
Figure 2. Comparison of the effects of glucose and drugs acting on KATP channels on insulin
secretion by the focal lesion and the adjacent healthy pancreas in patient 1 and patient 2 .
The concentration of glucose was increased from 1 to 15 mM (G1 to G15), and 100 µM
diazoxide (Dz) and 100 µM tolbutamide (Tolb) were added as indicated at the top of the
figure. Forskolin (Fk), a direct activator of adenyl-cyclase, was present throughout at 1 µM.
Open circles: beta cells from the lesion; filled circles: beta cells from the healthy pancreas.
Figure 3. Near-normal glucose levels achieved by medical treatment in patient 2. Blood
glucose was monitored by 72-h CGMS, after 20 months of diazoxide treatment, at a
diazoxide dosage of 10 mg/kg/day. CGMS showed a glycemic range of 48-134 mg/dl (2.6-
7.4 mmol/l) and an average glucose value of 90 mg/dl (5 mmol/l) with one single event of
hypoglycemia of 10 minutes (< 0.25% in 3 days). Open stars indicate simultaneous
glucometer values.
Page 22
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
Figure 4. A. 18F-L-DOPA PET-CT scan (transaxial view) showing the presence of a focal
lesion in the medial-dorsal part of the pancreatic head of patient 2 (arrow). The uptake of 18F-
L-DOPA is increased in the head region (max SUV: 4.1) compared to the body and tail
regions (SUV ratio: 1.28). B. A focal lesion (exophytic polypoid mass) is easily identified at
the surface of the pancreatic head resected from patient 2. C. Intra-operative histology
showed adenomatous hyperplasia of endocrine cells within the mass.
Figure 5. Uncontrolled glucose oscillations in a patient with neonatal onset focal CHI and
paternal inherited ABCC8 mutation with no responsiveness to medical therapy. Blood glucose
was monitored using a120-h CGMS which showed hypoglycemia for >20% of the time at 4
months of age. At that time, after no responsiveness to diazoxide dosage of 15 mg/kg/day, the
patient was on therapy with diazoxide at the dosage of 4 mg/kg/day, octreotide at the dosage
of 19 µg/kg/day in combination with continuous enteral feeding for an amount of 13
mg/kg/min of glucose. Open stars indicate simultaneous glucometer values.
Figure 6. Suggested diagnostic and management algorithm for congenital hyperinsulinism
(CHI). Once the diagnosis has been established, patients should immediately be treated with
diazoxide (DZX) then, in case of unresponsiveness, with octreotide. A metabolic diagnostic
workup must be run in parallel with treatment, to identify specific disorders which include
Congenital Disorders of Glycosylation (CDGs), Hyperisulinism-Hyperammonemia syndrome
(HI-HA) due to gain of function mutation of glutamate dehydrogenase gene (GLUD1), and
3-Hydroxyacyl-CoA dehydrogenase (HAD) deficiency due to mutation of HADH gene.
According to familial history and suspected inheritance mode, ABCC8, KCNJ11, GCK,
HNF1α, HNF4α, HADH, INSR, SLC16A1or UCP2 genes can be sequenced (the last two
genes are sequenced only in a few laboratories). Patients with recessive biallelic mutations in
Page 23
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
ABCC8, KCNJ11, or HADH genes (mt/mt), or with dominant mutations (wt/mt) in ABCC8,
KCNJ1, or GCK, HNF1α, HNF4α, INSR, SLC16A1 and UCP2 genes exhibit a diffuse form
of CHI. Patients with a recessive mutation in ABCC8, KCNJ11 genes in the paternal allele
(wt/mt) exhibit a focal form of CHI and should undergo 18F-DOPA PET-CT and elective
partial pancreatectomy. Patients responsive to DZX with no detectable mutations (wt/wt)
should also undergo 18F DOPA-PET-CT to search for atypical focal forms of CHI potentially
treatable by partial pancreatectomy (i.e. bifocal forms). Wt: wild type; mt: mutant; wt/mt
dominant*: already reported in the literature or family tree strongly suggestive of dominant
negative mutation.
Figure 1 supp. Near-normal glucose levels achieved by medical treatment in patient 1. Blood
glucose was monitored by CGMS for 48h, three months before surgery, at the age of 4 years
and 10 months, at a diazoxide dosage of 4 mg/kg/day. CGMS showed a glycemic range of
44-115 mg/dl (2.4-6.3 mmol/l) and an average glucose value of 75 mg/dl (4.2 mmol/l) with
one single event of hypoglycemia of 5 minutes ( < 0.25% in 2 days). Open stars indicate
simultaneous glucometer values; filled stars in indicate meal event. This session lasted less
than 72h for family needs.
Page 24
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
Page 25
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
Page 26
Acc
epte
d A
rtic
le
This article is protected by copyright. All rights reserved.
Established diagnosis of CHIStart medical therapy (DZX ⇒Octreotide)
Mutational analysis
Metabolic investigations
Transferrin IEFabnormal
Hyperammonemia ⇑C4-OH-carnitine⇑3-OH-glutarate
HADHGLUD1
CDGs
HI-HAdiffuse HAD
diffuse
ABCC8/KCNJ11HNF1-4α/GK/HADH/
SLC16A1/UCP2wt/wt
DZX responsive
HNF1-4α/GK/SLC16A1/UCP2
wt/mt
ABCC8/KCNJ11mt/mt
wt/mt dominant*
diffuse
ABCC8/KCNJ11wt/mt paternal
18F DOPA PET/TC
Medical therapy (DZX ⇒ Octreotide)Diet
focalatypicaldiffuse
Partial pancreatectomy
responsive
General follow-up:growth, neurological andnutritional development
Non- responsive
Subtotalpancreatectomy
abnormal normal
HADHmt/mt
Mannosetherapy
MPI-CDG