Introduction

Transient neonatal diabetes mellitus (TNDM) is a clinically defined form of neonatal diabetes mellitus that presents soon after birth, undergoes spontaneous remission during infancy but may relapse to a permanent form of diabetes mellitus in childhood or adolescence [1]. While 26% of TNDM patients have mutations of KCNJ11 (OMIM 601374 [http://omim.org; accessed 04 May 2012]), ABCC8 (OMIM 600509), INS or HNF1B, almost 70% (OMIM 601410) have genetic and epigenetic aberrations at the TNDM locus on chromosome 6q24, causing overexpression of two imprinted genes, PLAGL1 and HYMAI [2, 3]. The three reported causes of PLAGL1 and HYMAI overexpression are: (1) paternal uniparental disomy of chromosome 6 (UPD6pat); (2) paternally inherited duplication of 6q24 (duplication) and (3) maternal hypomethylation of the differentially methylated region (DMR) at 6q24. In a proportion of patients the hypomethylation appears to be purely epigenetic, without any detectable underlying genetic cause, and exclusively affects the DMR in TNDM. In other cases, hypomethylation of multiple imprinted loci (HIL) is observed, with a portion of these cases associated with genetic mutations of ZFP57 (OMIM 612192; ZFP57-HIL) [4, 5].

The rarity of TNDM (1:200,000 to 1:400,000 live births) poses challenges for data collection about clinical features, outcome and management. Until now the clinical features of 6q24 TNDM have been defined in small case studies, some including patients without a molecularly confirmed diagnosis [6, 7]; therefore trends in birthweights, presentation, remission and clinical features, particularly comparing different 6q24 TNDM aetiologies, has been limited by low statistical power. Here we describe the clinical presentation of the largest worldwide cohort of confirmed 6q24 TNDM cases, the majority of whom have not been previously reported, which enables us for the first time to quantify genotype–phenotype correlations.

Methods

Patients

Patients positively diagnosed with 6q24 TNDM at the Wessex Genetics Service (www.wrgl.org.uk; accessed 11 September 2012) were ascertained to be from Europe, the Americas, Asia and Australia but ethnicity was not recorded. They were identified through the British Paediatric Association Surveillance Unit, British Diabetic Association, or after referral by endocrinologists, clinical geneticists and paediatricians to either the Peninsula Genetics Service or the Wessex Genetics Service. As part of the diagnostic process referring physicians completed a clinical questionnaire recording: conception, pregnancy history, gestation, birthweight, age of presentation and remission, treatment and number and nature of congenital abnormalities (electronic supplementary material [ESM] questionnaire). Consent to include clinical data in the referral was obtained by the referring physician.

Genetic analysis

DNA was extracted from whole blood using standard procedures. Methylation-specific PCR was used to detect hypomethylation of the 6q24 locus, followed by microsatellite analysis to discriminate UPD6pat from isolated hypomethylation at 6q24, as described [4]. Extent of paternal duplication was not routinely determined since it was incidental to molecular diagnosis of TNDM, and extent of uniparental disomy could not always be definitively determined where microsatellite data were uninformative. Samples with 6q24 hypomethylation but not UPD6pat were tested for hypomethylation at other imprinted loci and for ZFP57 mutations, as described [5].

Data handling and analysis

Information from referral questionnaires was recorded on an in-house clinical database. Birthweight, gestation and sex were used to calculate adjusted birthweight standardised deviation scores (SDS) using the LMSgrowth application (version 2.76. www.healthforallchildren.co.uk/; accessed 4 May 2012). Statistical calculations were performed using SPSS (version 19: http://spss-mac.en.softonic.com/mac, accessed 4 May 2012).

Results

One-hundred and sixty-three patients with a molecular diagnosis of TNDM were analysed: 87 (53%) male and 76 (47%) female. Sixty-six (41%) had UPD6pat, 54 (33%) paternal 6q24 duplication and 43 (26%) maternal 6q24 hypomethylation. Of hypomethylation patients 18 (11%) were isolated, 12 (7%) were non-ZFP57-HIL, 12 (7%) were ZFP57-HIL and 1 (1%) were unclassified due to insufficient sample for complete analysis; because standard molecular diagnostic methods did not unequivocally determine the extent of either UPD6 or chr6 duplication, these patients were not further subclassified.

The majority of patients in our cohort were born small for gestational age, with a mean weight and adjusted birthweight SD of 2,001 g and −2.5 respectively (Table 1). Forty of the 133 patients for whom data were available were born at <37 weeks of gestation (30.1%), an incidence significantly higher than in the general population (e.g. the 6.2% quoted by the UK Office of National Statistics) [p = 0.02, paired t test; www.ons.gov.uk/ons/publications/re-reference-tables.html?edition=tcm%3A77-50818 (archived)] or the global incidence of 9.6% estimated by the World Health Organization (www.who.int/bulletin/volumes/88/1/08-062554/en/; accessed 13 November 2012).

Table 1 Clinical features of the 6q24 TNDM patients, divided according to aetiology
Full size table

The cohort presented with hyperglycaemia at a modal age of 1 day, but at markedly greater median and mean ages (4 and 8 days, respectively; Table 1, ESM Fig. 1). Likewise, the modal age of remission was 2 months but the median and mean ages were 3 and 4.5 months, with the longest recovery recorded at 48 months. Age of presentation with diabetes was significantly correlated with gestational age (r = 0.244, p = 0.005). In addition, age of remission was negatively correlated with adjusted birthweight SD (r = −0.188, p = 0.046, ESM Table 1). The removal of the 48-month outlier increased this significance further (r = −0.199, p = 0.036, data not shown).

The most commonly reported congenital abnormalities were macroglossia and umbilical hernia, in 54/123 (44%) and 24/114 (21%) of patients, respectively. Less frequently reported congenital abnormalities included dysmorphic facial appearance 21/114 (18%), renal tract abnormalities (duplex kidneys, hydronephrosis, dilated renal pelvis and vesicoureteral reflux) 11/117 (9%), cardiac anomalies (ductus arteriosus, tetralogy of Fallot, atrial–septal defects and persistent foramen ovale) 10/114 (9%), clinodactyly, polydactyly, nail and short finger abnormalities 9/116 (8%) and hypothyroidism 4/103 (4%). No other significant congenital abnormalities were observed in our modest sample size (Table 2).

Table 2 Congenital anomalies in patients with 6q24 TNDM divided by genetic abnormality subgroup
Full size table

Congenital abnormalities occurred significantly less frequently in the 6q24 duplication subgroup, at 0.52/patient compared with 1.15 (p = 0.032) and 1.27 (p = 0.017) for UPD6pat and hypomethylation subgroups, respectively (Table 1, ESM Table 2). The duplication subgroup had reduced frequency across several congenital abnormalities (Table 2). Within the hypomethylation subgroup, patients with hypomethylation confined to 6q24 also had reduced frequency of congenital abnormalities, averaging 0.46/patient, compared with those with non-ZFP57-HIL or ZFP57-HIL who had an average of 1.73 and 1.83 congenital abnormalities, respectively (Table 1). Notably, macroglossia was the only anomaly recorded in the isolated hypomethylation subgroup (ESM Table 3).

Of 65 cases with data on conception (16 duplication, 23 UPD6pat, 26 maternal 6q24 hypomethylation), four were conceived after assisted reproductive technology (ART). All were hypomethylation patients, three non-ZFP57-HIL and one unclassified. The recorded incidence of ART in the hypomethylation subgroup was 15%.

Discussion

In this study we gathered information from clinicians worldwide on patients molecularly diagnosed with 6q24 TNDM at the Wessex Genetics Service. This is the largest cohort reported to date, containing more extensive clinical details than previous studies, and permits statistical analysis of 6q24 TNDM at presentation.

The principal findings of this study were: (1) the previously unreported relationship between age of presentation of 6q24 TNDM and gestation and the relationship between age of remission and adjusted birthweight SD; (2) the reduced frequency of congenital abnormalities among duplication and isolated hypomethylation patients; and (3) the elevated incidence of ART (15%; 4/26) within the hypomethylation group. Previous observations on severe intrauterine growth retardation and mean age of remission were confirmed [7]. These findings underline the lower birthweight and earlier presentation in 6q24 TNDM that is caused by potassium channel mutations (<1st vs 12th centile, and <1 week vs 4 weeks, respectively). However, the relatively low birthweight previously reported in duplication patients [8] was not supported by this study.

While birthweight (adjusted for gestation) was normally distributed, the ages of presentation and remission of diabetes were markedly skewed, with modes at 1 day and 2 months, but means of 8 days and 4.5 months. The limited clinical data available and the wide variety of healthcare settings in which these patients were treated makes it uncertain whether these variations represent primary variations in clinical history, or variations in diagnosis and management (e.g. delayed recognition of hyperglycaemia or delayed withdrawal of exogenous insulin). TNDM symptoms such as dehydration and failure to thrive are non-specific, so delayed diagnosis in full-term neonates may simply reflect a delay in recognition of neonatal diabetes among other potential diagnoses. The negative correlation between adjusted birthweight SD and age of remission of 6q24 TNDM may be accounted for by earlier remission in the subset of patients with residual insulin secretion and therefore higher birthweight. The correlation of gestation with age at presentation may reflect the prompt testing of blood glucose in premature babies. The high prevalence of preterm birth (30% <37 weeks gestation) may reflect early medical intervention to deliver infants on detection of growth restriction; detailed assessment of clinical history is required to determine whether there is an underlying trend to prematurity.

Stratified analysis of aetiological subgroups was limited by low patient numbers, but some interesting observations emerged. The increased incidence of congenital anomalies in UPD6pat and the largely consanguineous ZFP57-HIL cases may reflect the potential for unmasking of recessive traits among affected individuals. The increased prevalence of congenital abnormalities in the non-ZFP57-HIL group is hitherto unreported, probably because of the extreme rarity of these patients, and may stem from gene dysregulation at other loci affected by their wide-ranging epimutations. The ART frequency observed in the hypomethylation patients, though of limited power due to low cohort size, is in keeping with the incidence of ART in Beckwith–Wiedemann syndrome (4–10%) and is significantly higher than levels in the normal population [9].

In conclusion, 6q24 TNDM may be distinguished from other types of neonatal diabetes by birthweight, with congenital malformations indicating an aetiological subgroup. Emerging genotype–phenotype relationships may predict prognosis for patients in the future. Since TNDM is a relatively newly defined disorder, generally diagnosed in infancy, and therefore the majority of patients are not yet adults, long-term follow-up remains rare (e.g. [10]), but TNDM registries have been established in the UK and USA to aid this process (www.soton.ac.uk/geneticimprinting and http://monogenicdiabetes.uchicago.edu/neonatal-registry/; accessed 11 September 2012).