Neonatal diabetes mellitus around the world: Update 2024

Abstract

Neonatal diabetes mellitus (NDM), defined as diabetes with an onset during the first 6 months of life, is a rare form of monogenic diabetes. The initial publications on this condition began appearing in the second half of the 1990s and quite surprisingly, the search for new NDM genes is still ongoing with great vigor. Between 2018 and early 2024, six brand new NDM‐genes have been discovered (CNOT1, FICD, ONECUT1, PDIA6, YIPF5, ZNF808) and three genes known to cause different diseases were identified as NDM‐genes (EIF2B1, NARS2, KCNMA1). In addition, NDM cases carrying mutations in three other genes known to give rise to diabetes during childhood have been also identified (AGPAT2, BSCL2, PIK3R1). As a consequence, the list of NDM genes now exceeds 40. This genetic heterogeneity translates into many different mechanism(s) of disease that are being investigated with state‐of‐the‐art methodologies, such as induced pluripotent stem cells (iPSC) and human embryonic stem cells (hESC) manipulated with the CRISPR technique of genome editing. This diversity in genetic causes and the pathophysiology of diabetes dictate the need for a variety of therapeutic approaches. The aim of this paper is to provide an overview on recent achievements in all aspects of this area of research.

Between 2018 and early 2024, six brand new NDM‐genes have been discovered and 3 genes known to cause different diseases were identified as NDM‐genes. Currently NDM genes' list exceeds forty.This genetic heterogeneity translates into many different mechanism(s) of disease that are investigated with state‐of‐the‐art methodologies, such as induced pluripotent stem cells (iPSC) manipulated with CRISPR technique of genome editing. This diversity in genetic causes and pathophysiology of diabetes dictate the need of a variety of therapeutic approaches.

Introduction

Neonatal diabetes mellitus (NDM), is diabetes appearing within the first 6 months of life and a type of monogenic diabetes. NDM is caused by many genes, most of which have an impact on pancreatic β cell formation or/and function, leading to extremely reduced or absent plasma insulin levels. In a small percentage of patients NDM (often in alternation with hypoglycemia) is caused by mutations in genes affecting insulin action. In addition, NDM is also linked to genes regulating the immune system, representing a distinct subtype. Consequently, the mechanisms of disease are numerous, with important implications for therapeutic approaches. This review focuses on advances in the field of NDM from 2018 to early 2024. To favor an easier reading the review has been divided in two main sections: (a) the discovery of new genes causing NDM and recent findings on the mechanisms of disease, (b) the progress in NDM diagnosis and therapy around the world.

New NDM genes leading to insulin deficiency

Between 2018 and 2024 six new NDM genes were identified: CNOT1 (OMIM *604917) (dominant) ¹ , ² , YIPF5 (OMIM #619278) (recessive) ³ , PDIA6 (OMIM *611099) (recessive) ⁴ , ⁵ , ONECUT1 (previously known as HNF6) (OMIM *604164) (recessive) ⁶ , ⁷ , FICD (OMIM *620875) (recessive) ⁸ , and ZNF808 ⁹ (NM_001321425.2) ¹⁰ (recessive). These genes cause the permanent form of the disease (PNDM) in patients with extrapancreatic features such as holoprosencephaly (CNOT1) ¹ , ² microcephaly (YIPF5) ³ , small thorax, microcephaly, polycystic kidney (PDIA6) ⁴ , ⁵ , gallbladder agenesis and musculoskeletal abnormalities (ONECUT1) ⁶ , ⁷ , neurodevelopmental abnormalities (FICD) ⁸ . Of much interest, mutations in ZNF808, a primate‐specific gene encoding for a transcription factor, were found to cause pancreas agenesis in 18 patients belonging to 14 different families ⁹ , ¹⁰ . The other five represent an exceedingly rare cause (1–6 probands described) of PNDM.

Insulinopenic NDM due to genes previously associated with other conditions

Ten NDM patients carrying mutations in genes previously associated with other diseases or a different age of diabetes onset were described: five cases with heterozygous mutations in EIF2B1 ¹¹ , three cases with biallelic NARS2 mutations ¹² , ¹³ , and a case carrying a KCNMA1 variant ¹⁴ .

Previously, recessive variants of EIF2B1, encoding the eIF2B complex alpha subunit, were identified as the cause of leukoencephalopathy with white vanishing matter (OMIM #603896). Instead, EIF2B1 mutations found in PNDM patients are de novo (heterozygous) located in the binding surface for phosphorylated eIF2α and likely impair the eIF2‐eIF2B regulation of protein translation ¹¹ . Consequently, a fundamental mechanism of pancreatic β cell homeostasis is disrupted, with likely severe and sustained endoplasmic reticulum (ER)‐stress leading to β‐cell apoptosis and diabetes. Interestingly, in these patients (transient) liver dysfunction has been also observed, an extrapancreatic feature found in a much more severe form in most patients with EIF2AK3 biallelic variants; EIF2AK3 encodes for PERK, a crucial component of protein translation control.

NARS2 encodes mitochondrial asparaginyl‐tRNA synthetase 2 and it is known to cause autosomal recessive deafness (OMIM #618434) and combined phosphorylation deficiency 24 (COXPD24) (OMIM #616239). Three patients with NDM, hearing impairment and epilepsy, therefore categorized as developmental delay, epilepsy, neonatal diabetes, i.e. DEND, an acronym used to describe a subtype of NDM with developmental delay and epilepsy caused by severe pathogenic variants in K_ATP (KCNJ11 and ABCC8 genes) were found to bear biallelic NARS2 mutations ¹² , ¹³ . KCNMA1 encodes for the pore forming subunit of the large‐conductance voltage‐ and Ca²⁺‐activated K channel. Heterozygous, spontaneous KCNMA1 mutations cause the Liang‐Wang syndrome (OMIM #618729), characterized by developmental delay, facial dysmorphism, visceral and connective tissue abnormalities. A patient carrying the same missense, spontaneous variant found in some patients with Liang‐Wang syndrome (p.Gly375Arg) presented with diabetes on day 3 of life, along with severe congenital abnormalities ¹⁴ .

NDM due to mutations in genes usually presenting with impaired insulin action and diabetes outside the neonatal period

For decades INSR, encoding for the insulin receptor, has been the only known gene involved in diabetes associated with congenital severe insulin resistance (c.SIR). Recently, NDM cases with c.SIR have been reported bearing mutations in genes different from INSR: PIK3R1, BSCL2, and AGPAT2 ¹⁵ , ¹⁶ , ¹⁷ ; this was an interesting finding because carriers of mutations in these genes usually present diabetes well outside the neonatal period. Heterozygous mutations of PIK3R1 cause the SHORT syndrome that features short stature, sensineural hearing loss, facial dysmorphism, and partial lipodystrophy. In the three papers describing 24 patients with SHORT syndrome in 2013, seven presented with diabetes in childhood, but none with NDM. In a 2021 paper, a patient carrying a previously described PIK3R1 variant (p.Arg649Trp) was diagnosed with diabetes at 5 days from birth. Hyperglycemia ceased at 6 weeks of life and the patient was diagnosed as having transient NDM (TNDM) ¹⁵ .

In a 2024 reappraisal of NDM in Italy, Rapini et al. ¹⁶ reported the results of 104 patients with NDM, seven of which were diagnosed with c.SIR. In six patients c.SIR was due to biallelic mutations of INSR. The seventh patient with distinctive features of congenital lipodystrophy, including hypertriglyceridemia, was diagnosed with BSCL2‐c.SIR. He was the fourth patient carrying biallelic BSCL2 mutations who presented with diabetes in the neonatal period ¹⁷ , ¹⁸ , ¹⁹ . In another 2024 paper, Hassan described 37 patients with NDM; notably, in two patients with generalized lipodystrophy, a homozygous splice site mutation of AGPAT2 was identified, further extending the time of onset of diabetes in different types of c.SIR ²⁰ .

With the latest additions the NDM subtypes caused by genes impacting on β cell function now number more than 40 (this review and Refs. [21, 22]) (Figure 1; Table S1).

Figure 1.

Neonatal diabetes genes. Most gene products are located in the nucleus (transcription factors) or in the endoplasmic reticulum (ER). Gene in black font in the nucleus and in red font in the ER are recently identified genes. Genes responsible of glucose‐induced insulin secretion are on the beta cell membrane (e.g. KCNJ11, ABCC8) or in the cytoplasm (GCK). Large deletions of mitochondrial DNA (m.DNA) have been recently reported to give rise to neonatal diabetes; NARS2 is also expressed in mitochondria. INSR causes severe insulin resistance and neonatal diabetes by impairing insulin action in all major insulin‐sensitive tissues (skeletal muscle, liver and adipose tissue). Genes responsible for congenital generalized lipodystrophies AGPAT2, BSCL2) or partial loss of subcutaneous fat (SHORT syndrome, PIK3R1) are depicted in adipose tissue, though all 3 genes are expressed in many organs/tissues. Please note that chromosomal abnormalities of 6q24 leading to transient neonatal diabetes mellitus and trisomy 21 leading to autoimmune permanent neonatal diabetes not linked to HLA are not included in the figure. KCNMA1 is not shown because only one patients has been reported so far.

MECHANISM(S) OF DISEASE

KCNJ11

K_ATP channel genes (KCNJ11, ABCC8) are expressed in multiple tissues, including the pancreatic β cell, skeletal muscle, and neurons. As a result, patients bearing KCNJ11 gain‐of‐function mutations and – more rarely – in ABCC8 may show a neurologic phenotype described as DEND syndrome or incomplete (or intermediate) DEND (iDEND) when epilepsy is not present ²³ . Cerebral organoids derived from induced Pluripotent Stem Cells (iPSC) carrying the KCNJ11 Val59Met variant, a frequent iDEND mutation, showed a neuronal activity that was defective in bursting synchronization and other alterations ²⁴ . This finding gave a mechanistic explanation of the phenotypic features invariably found in KCNJ11‐V59M carriers and other pathogenic variants in K_ATP genes leading to iDEND ²⁵ , ²⁶ .

INS

Biallelic, loss‐of‐function and heterozygous, gain‐of‐function variants of INS gene cause PNDM with two different mechanisms. In the first case transcription of the gene is hampered by homozygous variants in the INS promoter region or insulin protein is not translated or degraded as a consequence of mutations in the first ATG (signal peptide) or premature stop codons ²⁷ . Differently, more common heterozygous, proteotoxic variants seem to cause beta cell failure mainly through β cell apoptosis ²⁸ .

Experiments on iPSC derived from a patient with the homozygous c.3G>A, p.0? INS variant showed a lack of proinsulin after differentiation into β cell‐like cells ²⁹ . Conversely, correction of the mutation to the wild type sequence by CRISPR‐Cas9 determined the translation of proinsulin and the activation of unfolded protein response ²⁹ . Of note, corrected β cell‐like cells (i.e. with ATG encoding Met1 of preproinsulin) kept glucose levels in the normal range when transplanted in diabetic mice.

A study on iPSCs derived from patients with gain‐of‐function, heterozygous INS Cys96Arg and Cys109Tyr variants confirmed that mutant insulins cause endoplamic reticulum stress in β cell‐like cells, but failed to detect apoptosis. Instead, reduced proliferation and impaired insulin secretion were detected after β cell‐like cell transplantation in mice ³⁰ .

Finally, experiments on iPSC derived from INS‐PNDM mutations p.Leu39_Tyr40 delinsHis (also described as L^B15Y^B16delins) ²⁸ and Tyr50Cys (or Y^B26C) ³¹ have revealed that in addition to early apoptosis, beta cell dedifferentiation with a loss of beta cell identity is a possible explanation for the defective insulin production that deteriorates over time in mutation carriers ²⁸ , ³¹ , ³² .

CNOT1

Four patients with pancreas agenesis and holoprosencephaly, bearing the same de novo missense mutation (p.Arg535Cys) have been described in two separate papers that provided some important clues about this association ¹ , ² . In one paper describing three of these patients, the results of the analysis of Cnot1^{p.(Arg535Cys)/p.(Arg535Cys)} mouse (created utilizing CRISPR) were reported. These mice did not survive beyond E14.5 and analysis of the embryos revealed exencephaly, eye defects, and small pancreas. In contrast, heterozygous animals did not show any phenotype ¹ . In another paper (one patient with PNDM and exocrine insufficiency) in situ hybridization in mouse embryo during early neurolation (GD8.25) revealed that Cnot1 expression is detectable in the neuroectoderm and this could possibly determine disruption of sonic hedgehog signaling leading to holoprosencephaly ² .

YIPF5

Six patients with homozygous missense or single amino acid deletion in YIPF5 were described, four of which with diabetes with onset within 6 months of life (4–23 weeks) ³ . Functional analysis of the mutations was performed on EndoC‐βH1 cells (human pancreatic β cell line, YIPF5 silencing), human embryonic stem cells (YIPF5 knockout and mutation knockin), and iPSC from a patient. Of note, iPSC carrying the Ile98Ser mutations could be differentiated β cell‐like cells that showed a normal proinsulin and insulin content ³ . However, β cell‐like cells YIPF5^Ile98Ser were more prone to undergo apoptosis than cells with corrected mutation when ER‐stressor (e.g. tunicamycin, thapsigargin) were applied for 48–72 h ³ suggesting that the metabolic phenotype is likely due to a progressive loss of β‐cells due to ER‐stress apoptosis.

NEUROG3

A thorough investigation on six different NEUROG3 mutations leading to malabsorptive diarrhea and diabetes has been published in 2019 ³³ . The authors created a tetracycline‐inducible NEUROG3 in NUROG3‐deficient hESCs in order to express WT or mutant NEUROG3 as close as possible to physiologic level. The authors showed that three NEUROG3 mutations, namely Arg93Leu, Arg107Ser, and Ser171fsTer68, weakly activated three (out of six known, including RFX6, see below) downstream transcription factors important for endocrine pancreas development such as NEUROD1, NKX2‐2, and PAX4, while variants Glu28Ter, Glu123Ter, and Leu135Pro were incapable of inducing target genes. By playing with the time (i.e. extending it) of the doxycycline treatment the authors demonstrated that Arg93Leu, Arg107Ser, and Ser171fsTer68 variants were in fact hypomorphic and capable of rescuing endocrine cell formation if expressed for a longer time. However, while different NEUROG3 mutations have a different impact on endocrine pancreas (childhood vs NDM), all cause malabsorptive diarrhea. Other experiments showed that NEUROG3 protein is unstable (has a shorter half‐life) in enteroendocrine cells when compared with pancreas and this might explain why the intestine is more sensitive to mutations ³³ .

RFX6

Three papers investigated the role of RFX6 in pancreas development. The first one was based on a conditional knockout mouse in which Rfx6 function was disrupted in the β‐cell of adult animal (Rfx6^Δbeta) and compared with the effects of Rfx6 absence in fetal β‐cell (Rfx6^ΔEndo). This elaborate strategy circumvented the problem of Rfx6^ΔEndo mice dying of diabetes 2–3 days after birth because of the lack of insulin producing cells ³⁴ and helped to explain which is the role of Rfx6 in the adult β cell. As a matter of fact, a lack of Rfx6 in adult β‐cells (Rfx6^Δbeta) causes impaired insulin secretion, largely due to the decreased expression of Gck and Abcc8 in Rfx6^Δbeta mouse. In addition, derepression of disallowed genes in Rfx6^Δbeta indicates that Rfx6 is crucial for maintaining the adult β cell functional identity ³⁴ . These experiments contribute to explain how biallelic mutations of RFX6 cause Mitchell‐Riley syndrome, which include NDM, while heterozygous stop codon mutations may cause a maturity onset diabetes of the young (MODY)‐like phenotype. The differentiation attempt of iPSC carrying a RFX6 homozygous premature stop codon proceeded normally until the generation of the pancreatic endoderm which turned out to be significantly reduced ³⁵ . Further experiments showed that RFX6 is normally present in human PDX1‐expressing pancreatic endoderm confirming its essential role for efficient pancreas development ³⁵ . Finally, RFX6 involvement in differentiation of pancreatic endoderm at its early stage has been recently demonstrated exploiting a cleverly engineered RFX6 ^eGFP hiPSC cell line ³⁶ .

GATA6

A seminal paper published in 2020 exploits iPSC technology to show how a non‐coding variant at 3′ of GATA6 gene, by decreasing GATA6 expression, affects the phenotype, i.e. acts as a disease modifier. Carriers of coding sequence mutation(s) plus homozygous 3′ SNP minor frequence allele A/A (rs12953985) have pancreas agenesis (and NDM) in addition to heart defects, while the parent with coding GATA6 mutations and homozygous 3′ SNP G/G ('normal' SNP), have adult‐onset diabetes. In addition, some patients with pancreas agenesis inherited the coding region mutation from the parent with adult‐onset diabetes and the 3′ A minor allele in trans from the other parent with wild type GATA6 coding sequence and normal glycemia (e.g. parents of patient Exeter_28 with pancreas agenesis: father with adult‐onset diabetes carrying GATA6/Gly323Ter and SNP G/G and mother with GATA6 SNP A/A) ³⁷ . This paper is of general interest because it investigates the genetic bases of different phenotypes in carriers of identical variants in the coding region.

ONECUT1

A thorough analysis of the physiologic ONECUT1 role in pancreas development showed that this transcription factor is mainly expressed during the pancreatic endoderm (PE) stage throughout the pancreatic progenitors (PP) stage, when the ONECUT1 locus (chromatin) opened ³⁸ . Experiments suggested that ONECUT1 regulates pancreas progenitors probably through co‐binding with PDX1, GATA6, and FOXA2 to specific enhancers and prepares PP for endocrine specification ³⁸ . These results matched those obtained investigating differentiation into β cell‐like cells from iPSC derived from fibroblasts of a patient with pancreas hypoplasia carrying ONECUT1/Glu231Ter homozygous mutation ⁶ . In those experiments PP formation from PE was reduced. In addition, truncated ONECUT1 failed to transactivate specific enhancers of NXK6.1/NKX2‐2, two transcription factors crucial for pancreatic islet formation ⁶ .

PTF1A

In his paper Miguel‐Escalada demonstrates that mice with homozygous deletions of Ptf1a enhancer encompassing known human mutations (Ptf1a^enhΔ/enhΔ) recapitulate the human phenotype, i.e. pancreas hypoplasia and low birth weight ³⁹ . Elegant experiments also showed that in Ptf1a^enhΔ/enhΔ mice PTF1A is silenced in multipotent pancreatic progenitors, while it is normally expressed in acinar cells and in central nervous system (cerebellar agenesis is a feature of PTF1A mutations in the coding sequence, but not of mutations in distal enhancer). In addition, the authors demonstrate that intact PTF1A enhancer allows transient PTF1A expression in multipotent progenitor cells determining a permissive chromatin state. This in turn sets in motion an epigenetic cascade that allows full activation of NEUROG3 in endocrine progenitors and the formation of endocrine cells ³⁹ .

ZNF808

De Franco et al. ⁹ first described ZNF808 biallelic variants as a cause of pancreas agenesis. A peculiar finding about ZNF808 that sets it apart from other PNDM genes is that it can be found only in primates, but not in Old World monkeys or any other mammal ¹⁰ . Studying iPSC derived from two probands with homozygous ZNF808 deletion (and inactivating ZNF808 in hESC by CRISPR‐Cpf1 creating a ZNF808 knockout, ZNF808KO), the authors found that a lack of ZNF808 function during differentiation leads to an increasing number of dysregulated genes with repression of dorsal pancreas genes and activation of hepatic cord genes during the primitive gut tube stage onwards. This aberrant activation of liver genes expression during pancreas differentiation is the likely mechanism for pancreas agenesis ¹⁰ .

2018–2024: NDM CASE SERIES

PNDM

Europe and Americas

Surprisingly, recent reports of the NDM case series from a single country from this geographic area are few. A 2021 survey of the Monogenic Diabetes Registry of University of Chicago (established in 2008) ⁴⁰ confirms that NDM in USA (as already shown in Europe) is mostly caused by KCNJ11 (35%) and INS (16%) pathogenic variants but does not disclose other information. More detailed case series from Italy, Ukraine, and Brazil and have been published between 2023 and 2024 (Table 1) ¹⁶ , ⁴¹ , ⁴² . The clinical application of next generation sequencing (NGS) in these three studies yield similar results: KCNJ11 pathogenic variants were consistently found as the main cause of PNDM (34.8–44.4% of cases) followed by heterozygous, gain‐of‐function INS mutations (17.7–26%) in Ukraine and Italy ¹⁶ , ⁴¹ . Recessive and semidominant genes (e.g. PDX1, RFX6, GLIS3, PTF1A enhancer, GCK) were infrequent. Of interest, monogenic PNDM of autoimmune origin, previously considered extremely rare, represents a sizeble percentage (6.6–12.5) in these series (Table 1) ¹⁶ , ⁴¹ , ⁴² .

Table 1.

Case series

	Cases with genetic diagnosis	KCNJ11 (%)	ABCC8 (%)	INS dom. (%)	INS rec. (%)	EIF2AK3 (%)	PTF1A distal enhancer (%)	OTHER rare dom. (%)	OTHER rare rec./semidom. (%)	Auto‐immune PNDM (%)	6q24 (%)
PNDM
Europe/Americas
Globa E, Diabet Med 2023 (Ukraine)	23	8 (34.8)	1 (4.3) (biallel.)	6 (26)	0	3 (13)	0	0	3 (13) (GLIS3, PDX1, RFX6)	2 (8.7) (FOXP3, LRBA)	/
Costa‐Riquetto AD, Clin Genet 2023 (Brazil)	16	7 (43)	1 (6.2)	1 (6.2)	0	0	1 (6.2)	0	4 (25) (4 GCK)	2 (12.5) (2 FOXP3)	/
Rapini N, J Clin Endocrinol Metab 2024 (Italy)	35	20 (44.4)	1 (2.2)	8 (17.7)	0	0	0	1 (GATA6) (2.2)	2 (PDX1, RFX6) (4.4)	3 (6.6) (FOXP3, LRBA, IL2RA)	/
Middle East/Africa
Deeb A, Am J Med Genet, 2015 (Abu‐Dhabi, UAE)	15	1 (6.7)	1 (6.7)	0	5 (33.3)	8 (53.3)	0	0	0	0	/
Demirbilek H, Eur J Endocrinol 2015 (Turkey)	15	2 (13.3)	0	0	1 (6.7)	3 (20)	3 (20)	0	6 (GCK) (40)	0	/
Al‐Khawaga S, Mol Genet Genomic Med 2019 (Quatar)	9	0	0	1 (11.1)	1 (11.1)	1 (11.1)	3 (33.3)	1 (HNF1B) (11.1)	2 (GCK, SLC2A2) (22.2)	0	/
Abali ZY, Horm Res Paediatr 2020 (Turkey)	11§	1 (9)	4 (biallel.) (36.3)	1 (9.1)	0	1 (9.1)	3 (27.2)	0	1 SCL19A2 (9.1)	0	/
Öngen YD, J Clin Rs Pediatr Endocrinol, 2021 (Turkey)	6*	3 (50)	1 (16.6)	1 (16.6)	0	1 (16.6)	0	0	0	0	/
Mouler M, Pediatr Diabetes 2021 (Israel)	15	5 (33)	0	5 (33)	0	0	2 (14)	0	2 (IER3P1, RFX6) (14)	1 (FOXP3) (7)	/
Laimon WA, Acta Diabetol, 2021 (Egypt)	16	2 (12.5)	2 (12.5)	1 (6.2)	4 (25)	5 (31.2)	0	0	2 (1 GCK, 1 SLC19A2) (12.5)	0	/
Amaratunga SA, Diabetologia 2024 (Kurdistan, Iraq)	11	0	1 (9)	1 (9)	0	0	6 (54.5)	0	3 (2 GLIS3, ZNF808) (27.2)	0	/
Hassan SS, Pediatr Diabet 2024 (Sudan)	30	4 (13.3)	2 (1 biallelic) (6.7)	1 (3.3)	2 (1 heteroz.?) (6.7)	7 (23.3)	1 (3.3)	0	13 (3 ZNF808, 3 SLC19A2, 3 NARS2; 2 GCK, 2 SLC2A2) (43.3)	0	0
Asia/Pacific
Suzuki S, J Clin Endocrinol Metab, 2007 (Japan)	10	7 (70)	2 (20)	0	0	0	0	0	0	1 (FOXP3) (10)	/
Ibrahim MN, J Pediatr Endocrinol Metab, 2021 (Pakistan)	17 (age at onset ≤9 months of life^)	2 (11.8)	4 (3 homozygous) (23.5)	2 (11.8)	0	7 (41.2)	0	0	2 GCK (11.8)	0	/
Nayak S, Indian J Pediatr, 2021 (India)	7	0	0	0	0	2 (28.6)	0	0	4 (2 GCK, 2 SLC19A2) (57.1)	1 (FOXP3) (14.3)	/
Korula S, Indian J Endocrinol Metab, 2022 (South India)	6	0	0	1 (16.6)	0	2 (33.3)	0	0	1 PDX1 (16.6)	2 (1 FOXP3, 1 IL2RA) (33.3)	/
Ngoc CTB, Frontiers Endocrinol, 2021*§, 2022*§ (Vietnam)	39 *§	12 (30.7)	11 (28.2) (3 comp. heteroz.)	10 (25.6)	0	2 (5.1)	0	1 (EIF2B1) (2.6)	1 (GLIS3) (2.6)	2 (1FOXP3, 1 IL2RA) (5.1)	/
Lin Y, BMJ Open Diab Res Care, 2020, China (PNDM + TNDM)	31§§	KCNJ11 (21) + ABCC8 (3) 24 (77.4)		1 (3.2)	0	0	0	0	0	0	6 (19.3)
TNDM
Europe/Americas
Globa E, Diabet Med 2023 (Ukraine)	9	1 (11.2)	4 (44.4)	/	0	/	/	0	/	/	4 (44.4)
Costa‐Riquetto AD, Clin Genet 2023 (Brazil)	4	0	3 (75)	/	0	/	/	0	/	/	1 (25)
Rapini N, J Clin Endocrinol Metab 2024 (Italy)	40	11 (27.5)	13 (32.5) 1 biallel.	/	0		/	1 (HNF1B) (2.5)	/	/	15 (37.5)
Middle East/Africa
Deeb A, Am J Med Genet, 2015 (Abu‐Dhabi, UAE)	2	0	0		1 (50)				0		1 (50)
Demirbilek H, Eur J Endocrinol 2015 (Turkey)	5	0	1 (20)	0	1 (20)	/	/	0	0	/	3 (2 ZFP57) (60)
Mouler M, Pediatr Diabetes 2021 (Israel)	6	0	4 (66.6)	/	0	/	/	0	0	/	2 (33.4)
Hassan SS, Pediatr Diabet 2024 (Sudan)	5	0	0	1 (VUS)	0	/	/	0	1 SLC2A2	/	3 (60)
Asia/Pacific
Suzuki S, J Clin Endocrinol Metab, 2007 (Japan)	13	2 (15.4)	0	/	0	/	/	0	0	/	11 (84.6)
Ngoc CTB, Front Endocrinol 2021 and 2022 (Vietnam)	*§16	2 (12.5)	3 (18.7)	/	0	/	/	0	0	/	11 (68.8)

§One patient with TNDM (KCNJ11/E322K) excluded; *including only patient with diabetes onset within 6 months of age (2 EIF2AK3, 4 biallelic ABCC8 with diabetes onset >6 mesi excluded, 1 heterozygous ABCC8 with TNDM also excluded); ^ = all patients with presentation >6 months have been excluded; *§including patients with diabetes onset within 1 year; §§excluded 3 cases with hyperglycemia and heterozygous GCK variant that can not be considered PNDM.

Consang. (%), consanguinity % of total cases; dom., dominant mutations; rec., recessive mutations; enhan., enhancer; semidom., semidominant; NDM, neonatal diabetes mellitus; 6q24, aberrations of chromosome 6q24; PNDM, permanent neonatal diabetes mellitus; TNDM, transient neonatal diabetes mellitus; c.SIR, congenital severe insulin resistance.

Middle East and Africa

The incidence of PNDM only in the Middle East can be up to 8.5‐fold higher (1:23,000–1:29,000 live births) ⁴³ , ⁴⁴ than that found in Europe (about 1:200,000 live births) ¹⁶ , ⁴⁵ , this is likely due to the frequent occurrence of homozygous mutations in recessive genes (and likely founder effect) such as EIF2AK3 and PTF1A (distal enhancer) accounting for 11.1–53.3% and 14–54.5% of cases with PNDM in Abu‐Dhabi, Quatar, Egypt, and Kurdistan, respectively (Table 1) ⁴³ , ⁴⁴ , ⁴⁶ , ⁴⁷ . In contrast, heterozygous KCNJ11 and INS variants are common (66% combined) in Israel, while recessive causes are as rare as in Europe/Americas ⁴⁸ . In the South‐Eastern region of Turkey (Anatolia) the incidence is higher than in Europe, but lower than in other Middle East countries: 1:48,000 live births ⁴⁹ . In a group of 15 patients with PNDM from this geographic area recessive genes accounted for 86.7% of cases with homozygous variants identified in GCK (six patients; 40%), PTF1A enhancer (three patients) EIF2AK3 (three patients), and INS promoter (one patient), respectively. In two other studies from Turkey four out of 17 patients were diagnosed with KCNJ11‐PNDM (23.5%), five with ABCC8‐PNDM (biallelic in four patients) (29.4%), two with INS‐PNDM (heterozygous, proteotoxic), three with PTF1A distal enhancer/PNDM, two with EIF2AK3‐PNDM and one SCL19A2‐PNDM (TRMA), for a total of 64.8% of PNDM cases due to biallelic mutations ⁵⁰ , ⁵¹ . In a 2024 paper from Sudan, 30 patients were diagnosed with PNDM, in most cases caused by recessive mutations: seven EIF2AK3, three ZNF808, three NARS2, three SLC19A2, two GCK, two SLC2A2, one PTF1A enhancer ²⁰ . In addition, the incidence of PNDM over a 5 year period was 1 in 20,833 ²⁰ , i.e. about 10 times higher than in Europe ¹⁶ , ⁴¹ .

Asia/Pacific

A 2007 paper by Suzuki et al. ⁵² reported 10 PNDM patients from Japan with a final genetic diagnosis. Most patients were diagnosed with KCNJ11‐PNDM (70%) followed by heterozygous ABCC8‐PNDM and FOXP3. It has to be mentioned that at that time only 11 genetic defects causing NDM were known. Thus, in five patients no genetic cause was found ⁵² . In three recent papers from Pakistan and India ⁵³ , ⁵⁴ , ⁵⁵ reporting the genetic cause of PNDM in 30 individuals (some of them diagnosed within 9 months of age) ⁵³ , the relative contribution of recessive genes was prevalent (73.3%). Eleven patients carried homozygous or compound heterozygous EIF2AK3 mutations (36.7%), four homozygous GCK variants (13.3%), three with homozygous ABCC8 (10%), two with homozygous SLC19A2, one each with homozygous PDX1 and homozygous IL2RA ⁵³ , ⁵⁴ , ⁵⁵ . Among heterozygous or hemizygous mutations there were three KCNJ11, two INS, two with FOXP3 and one ABCC8. In two separate papers Ngoc reports the results of genetic screening of 70 individuals with diabetes onset within 1 year of age. In 55 patients a mutation was identified: 28 with KCNJ11 and ABCC8 variants (25.4% each), 10 with heterozygous INS mutations (18.1%), 11 (20%) with 6q24 abnormalities. In the remaining six, two had homozygous EIF2AK3 mutations, one compound heterozygous GLIS3, one with heterozygous EIF2B1, one with compound heterozygous IL2RA, and one hemizygous for FOXP3 ⁵⁶ , ⁵⁷ .

These results from different continents demonstrate that the genetic etiology (i.e. recessive vs dominant genes) and incidence of PNDM is strongly influenced by consanguinity (Middle‐East, India, Pakistan). This observation is further reinforced by the sizeable number of PNDM cases caused by biallelic INS (loss‐of‐function) and ABCC8 (mild, often found in heterozygous state in TNDM) mutations ⁴³ , ⁴⁶ , ⁵⁰ , ⁵³ as opposed to prevalent heterozygous, dominant INS (proteotoxic), and ABCC8 (severe) variants found in Europe and Vietnam ¹⁶ , ³⁸ , ⁵⁶ , ⁵⁷ .

TNDM

Europe/Americas

The results from three studies ¹⁶ , ³⁸ , ³⁹ confirm that most cases with TNDM are caused by 6q24 chromosomal abnormalities (25–44.4%) and ABCC8 (32.5–75%) variants, followed by KCNJ11 (11.2–27.5%). Only one patient out of a total of 53 was found to bear a heterozygous HNF1B defect ¹⁶ .

Middle East/Africa

Patients with TNDM are less frequently reported from Middle East/Africa and the data set is small. Nevertheless rare, recessive causes such as homozygous variants of the INS promoter, ZFP57 and SLC2A2 are found in Abu Dhabi, Turkey and Sudan ²⁰ , ⁴³ , ⁴⁹ . Differently, TNDM in Israel is caused by ABCC8 variants (66.6%) and 6q24 aberrations (33.4) as in Europe/Americas ⁴⁸ .

Asia/Pacific

In a paper from Japan published in 2007, just 1 year after the discovery of ABCC8 as a major cause of TNDM, most of the patients were diagnosed with 6q24 chromosomal defects (84.6%) ⁵² . In two papers by Ngoc et al. ⁵⁶ , ⁵⁷ , a methylation defect in 6q24 was found in 11 out of 16 patients with TNDM (68.7%), with the remaining five carrying ABCC8 (3 patients) and KCNJ11 mutations.

c.SIR and NDM

As mentioned previously AGPAT2, BSCL2, and PIK3R1 mutations have been found in patients with c.SIR and NDM, even in its transient form ¹⁹ . Therefore these three genes, in addition to INSR, should be routinely screened in patients with NDM.

2018–2024 CASE SERIES ON SINGLE GENES

Europe and Americas

m.DNA (large deletions)

In 2023 Björkmann reported the clinical features of 80 patients with large mitochondrial DNA deletions. Notably 20 patients were diagnosed with 'type 1' diabetes with the age at onset from 0.1 to 34.4 years. Therefore we can assume that at least one patient fell in the definition of NDM, although the exact number of cases with diabetes onset within 6 months of age is not reported in the paper ⁵⁸ . This indicates that m.DNA should be investigated in patients with NDM negative to NGS panels (Table 2).

Table 2.

Case series: single gene/genetic defect

	Gene	Number of cases	Clinical features	Laboratory features 1	Laboratory features 2	Therapy: insulin (# of cases)	Therapy: SU (# of cases)	Therapy: Other (# of cases)
Europe/Americas
Björkman K	m.DNA (large deletions)	80 (48 F)	PEO (23 cases) KSS (40 cases) PS (17 to >8 KSS)	Elevated lactate (32 out of 50)	Diabetes (25%) Age at onset 0.1–57.4 years. NDM% = ?	n.a.	n.a.	n.a.
Middle East/Africa
Habeb AM, Diabetologia 2018	SLC19A2 (TRMA)	32	Megaloblastic anemia, deafness, diabetes	All 32 diabetic NDM: 11 pat. (including 2 with onset at 7 m.)	Thiamine responders: improvement of HbA1c	28/32 = 87.5%; reduction of insulin dose in thiamine responders	///	Thiamine responsive: 11/15 = 73.3% (4 off insulin); 5 with NDM
Demirbilek H, J Clin Endocrinol Metab 2020	PTF1A enhancer	30	Pancreas agenesis/ hypoplasia; cholestasis. B.W.(mean): 1510 g	Biochemical exocrine pancreas insufficiency (low elastase, pancreatic enzymes): 19/30	Elevated transaminases: 25/30 Anemia: 25/30	30/30	///	Pancreatic enzymes
Aldrian D, Liver Int 2024	EIF2AK3	189	NDM in 186/189. Liver abnormality, growth abnormality: 75%	Acute liver failure		189/189	///	Liver transplantation in 6
Asia/Pacific
Gopi S, Pediatr Diabetes 2021	KCNJ11	20	B.W. (mean): 2,562 g Age at onset (mean): 3.2 m.	DKA: 81%		1/20 = 5%	SU positive response: 19/20 = 95%	///
Gopi S, Pediatr Diabetes 2021	ABCC8	19	B.W. (mean): 2,517 g Age at onset (mean): 3 m.	DKA: 62%		5/19 = 26.3%	SU positive response: 14/19 = 73.7%	///
Gopi S, J Diabetes Complications 2021	INS	8	B.W. (mean) 2,600 g Age at onset (mean): 4 m.	DKA: ?		8/8 = 100%	///	///

B.W., birth weight; DKA, diabetic ketoacidosis; KSS, Kearns‐Sayre syndrome; m.DNA, mitochondrial DNA; n.a., not available; NDM, neonatal diabetes mellitus; PEO, progressive external ophthalmoplagia; PS, Pearson syndrome; SU, sulfonylureas; TRMA, thiamine responsive megaloblastic anemia.

Middle East/Africa

TRMA

Habeb et al. studied 32 patients with thiamine responsive megaloblastic anemia with NDM. Twenty‐four different mutations were found. The authors recommended prompt recognition as early treatment with thiamine can result in improved gycemic control and even for some patients to become insulin‐independent ⁵⁹ .

PTF1A distal enhancer

Dermibilek et al. ⁶⁰ reported the results of 30 patients bearing biallelic variants of PTF1A distal enhancer. All but two were diagnosed with PNDM featuring low birth weight and intrauterine growth retardation; of interest, postnatal growth was also impaired. All patients had exocrine pancreas insufficiency and 25 (83.3%) elevated transaminases ⁶⁰ . One important finding of this paper is the lack of catch‐up growth in the postnatal period that is usually observed in non‐syndromic, isolated PNDM subtypes such as K_ATP (KCNJ11/ABCC8), INS and GCK.

EIF2AK3

Aldrian et al. undertook a systematic review on Wolcott‐Rallison syndrome. The authors focused on the spectrum and natural history, genotype–phenotype correlations, long term outcomes, and liver‐related complications. They concluded that WRS has a variable clinical course with a better survival in patients with missense mutation. The report also reviewed liver and multi‐organ transplantation and showed it is a feasible treatment option ⁶¹ .

Asia/Pacific

K_ATP genes (KCNJ11, ABCC8)

Gopi et al. ⁶² , ⁶³ performed two studies aimed at understanding the genetic causes of PNDM in Indian children. The first study concentrates on KCNJ11 and ABCC8 (N = 181) ⁶² , and subsequently the second study investigates INS gene mutations for the same cohort adding eight cases (N = 189). Out of 181 patients, 39 (22%) had diabetes due to KCNJ11 or ABCC8 mutations, with 20 cases bearing KCNJ11 mutations and 19 cases ABCC8 mutations. Developmental delays were observed in some patients, with three KCNJ11 mutation patients showing DEND syndrome, and seven out of 19 ABCC8 mutation patients showing developmental delays. A significant portion (84%) of these patients successfully transitioned to sulfonylurea therapy (glibenclamide), indicating a higher success rate for patients with KCNJ11 mutations compared with those with ABCC8 mutations.

INS

Among the 189 children studied, eight (4.2%) were found to have mutations in the INS gene ⁶³ . The study underscores that while the frequency of K_ATP channel mutations in Indian PNDM patients is lower compared with Caucasian and Japanese populations, these mutations are a significant cause of PNDM in India.

In summary, mutations in K_ATP channel genes and INS have been confirmed as major causes of NDM across various ethnicities. Recent evidence from the Asia‐Pacific region corroborates this, revealing clinical manifestations consistent with previously reported findings.

THERAPY

Advances of insulin‐pump therapy for NDM

In managing NDM, insulin therapy is often required initially, irrespective of the underlying cause. Insulin pump therapy offers precise control over insulin dosing, crucial for infants with small body sizes, which can make diabetes management more challenging ⁶⁴ . Fukuda et al. ⁶⁵ reported on three NDM patients treated with sensor augmented pump therapy (SAP). SAP can automatically suspend insulin delivery to prevent hypoglycemia based on predicted glucose levels, providing comprehensive management for patients with fluctuating blood glucose levels common in NDM. Recorded data such as glucose trends and total daily insulin doses simplify diabetes management and reduces the burden on caregivers.

Wanaguru et al. ⁶⁶ reported the introduction of Advanced Hybrid Closed‐Loop systems, an evolution from SAP, for managing TNDM with diluted insulin, meeting low daily total dose requirements in infants. This approach successfully managed the fluctuating insulin needs without causing significant hypoglycemia or other adverse effects.

SU treatment for NDM

New information on SU treatment for KATP‐NDM

Suzuki et al. ⁶⁷ reported a case of KCNJ11‐NDM, where sulfonylurea treatment began to fail during puberty. Typically, patients with KCNJ11‐NDM can switch from insulin to high‐dose sulfonylurea, but this particular case showed deteriorating glycemic control during adolescence. Contrary to the usual practice of reintroducing insulin, substantially increasing the sulfonylurea dose improved the patient's condition, highlighting the importance of re‐evaluating sulfonylurea dosage in adolescence.

A study reports the successful discontinuation of insulin therapy in a 24‐year‐old Japanese patient with a long‐standing history of KCNJ11‐NDM, after the addition of sitagliptin to sulfonylurea ⁶⁸ . Previously, it was challenging to discontinue insulin in older patients with long‐term diabetes even after switching to sulfonylurea ⁶⁹ . The patient showed marked improvement in endogenous insulin secretion after adding sitagliptin, suggesting its effectiveness in enhancing insulin secretion. This case indicates that the combination of sulfonylurea and sitagliptin could improve glycemic control in older KCNJ11‐NDM patients, even those with long diabetes duration.

Early SU initiation for NDM before genetic testing

Li et al. ⁷⁰ presents an innovative approach to managing NDM in China by introducing sulfonylurea therapy before genetic test results are available. The study's novelty lies in its focus on initiating sulfonylurea therapy at a young age, potentially offering a more effective treatment alternative to insulin, especially where genetic testing results are delayed.

Durability of sulfonylurea

Sulfonylurea therapy is the primary treatment for PNDM resulting from mutations in the KATP channel genes. Key studies by Bowman in 2018 (KCNJ11) and 2021 (ABCC8) show high long‐term efficacy and safety. The efficacy of sulfonylurea therapy was notably high, with 93% of KCNJ11 mutation carriers and 86% of ABCC8 mutation carriers maintaining insulin independence at their final follow‐up. For KCNJ11‐PNDM, the median HbA1c levels dropped from 8.1% pre‐transition to 6.4% over a median duration of 10.3 years. The median sulfonylurea dose 1 year post‐transition was 0.30 mg/kg/day, which decreased to 0.23 mg/kg/day, demonstrating stable long‐term dosing with minimal adjustments ⁷¹ . Similarly, in the 2021 study focusing on ABCC8‐PNDM, 86% of the patients also maintained insulin independence over approximately 10 years. HbA1c improved from 7.2% to 6.5% over 10 years, with sulfonylurea doses reducing from 0.37 to 0.25 mg/kg/day ⁷² .

Despite the effective management of glycemic control, neurological outcomes varied; 47% of KCNJ11 patients showed initial improvement, but 64% had persistent CNS features. For ABCC8, 62% had neurological features, with 54% improving post‐sulfonylurea therapy. These findings indicate that while sulfonylurea therapy can significantly impact glycemic control, its effect on neurological symptoms is less pronounced and does not lead to complete resolution in most cases.

In summary, these key multicenter studies confirm that sulfonylurea therapy is highly effective in managing diabetes in patients with KCNJ11 and ABCC8 mutations, maintaining excellent long‐term glycemic control. However, the ongoing neurological challenges emphasize the importance of comprehensive treatment strategies that combine metabolic management with neurological care.

Support therapy in mitochondrial disease determined by NARS2

Mitochondria supportive treatment with vitamin B1, l‐carnitine, and coenzyme Q10 was applied in two NARS2 siblings. For the younger sibling, insulin therapy was stopped after 6 months due to early mitochondrial therapy ¹² . Conversely, while the older sibling continued to require insulin therapy up to the age of three, a reduction in insulin dosage was achieved ¹² . The maintenance of plasma C‐peptide levels and controlled HbA1c indicated effective glucose management, showing the potential benefits of timely mitochondrial supportive therapy for NDM with NARS2 mutations.

Insulin secretion and sensitivity of 6q24‐TNDM after remission

Little is known about insulin dynamics during the remission phase of 6q24‐TNDM. Sato et al. ⁷³ reported a patient from remission (185 days old) to relapse (14.5 years old) using annual oral glucose tolerance tests, β‐cell function, and insulin sensitivity indices. Key findings include impaired early insulin response, fluctuations in fasting insulin and β‐cell function, and varying insulin resistance. Uchida et al. ⁷⁴ reported on a rare case of 6q24‐TNDM with insulin resistance at relapse. Typically, 6q24‐TNDM relapses after puberty, characterized by insulin insufficiency. The 10‐year‐old Japanese girl, initially presented with hyperglycemia in the neonatal period and was treated with insulin until 2 months of age. Her diabetes relapsed at 10 years old with high HbA1c levels and significant insulin resistance. Metformin improved her glycemic control and normalized her insulin resistance, expanding the clinical spectrum of 6q24‐TNDM.

Case reports of rare syndromic cases

For the sake of space we included only references of recent case reports regarding patients with mutations in rare genes such as GLIS3, GATA6, NKX2‐2, MNX1, NEUROG3, PDX1, FOXP3, LRBA and ZFP57 ⁷⁵ , ⁷⁶ , ⁷⁷ , ⁷⁸ , ⁷⁹ , ⁸⁰ , ⁸¹ , ⁸² , ⁸³ , ⁸⁴ , ⁸⁵ , ⁸⁶ , ⁸⁷ , ⁸⁸ , ⁸⁹ , ⁹⁰ , ⁹¹ , ⁹² .

CONCLUSIONS

Recent advances in gene discovery and the mechanism of disease shed light on all aspects of β cell biology including pancreas/pancreatic islet organogenesis and the fine tuning of quality control of insulin biosynthesis. Moreover, the extensive clinical application of NGS in NDM gene screening confirmed previous findings on the differences in NDM incidence between countries and contributed to detailing the phenotype of rare genes. All these achievements provide an additional basis for the correct genetic diagnosis of all forms of NDM and further refine precision NDM therapy.

DISCLOSURE

The authors have no financial conflict of interest to declare.

Approval of the research protocol: N/A.

Informed consent: N/A.

Registry and the registration no. of the study/trial: N/A.

Animal studies: N/A.

Supporting information

Table S1. | Alphabetical order list of genes/chromosomal defects associated with neonatal diabetes mellitus