Permanent neonatal diabetes: combining sulfonylureas with insulin may be an effective treatment

Permanent neonatal diabetes caused by mutations in the KCNJ11 gene may be managed with high‐dose sulfonylureas. Complete transfer to sulfonylureas is not successful in all cases and can result in insulin monotherapy. In such cases, the outcomes of combining sulfonylureas with insulin have not been fully explored. We present the case of a woman with diabetes due to a KCNJ11 mutation, in whom combination therapy led to clinically meaningful improvements.


Introduction
Permanent neonatal diabetes mellitus is a rare monogenic form of diabetes that presents within the first 6 months of life [1]. Mutations in the KCNJ11 and ABCC8 genes, encoding the Kir6.2 and SUR1 subunits, respectively, of the pancreatic b-cell ATP-sensitive potassium channel (K ATP channel) account for 50% of cases [2,3]. The K ATP channel plays a key role in insulin secretion by regulating the b-cell plasma membrane potential, and thereby insulin secretion, in response to the ambient glucose concentration [4]. At low glucose levels, the channel is open, keeping the membrane hyperpolarized and inhibiting insulin release. However, when glucose levels rise, the accompanying increase in metabolically generated ATP closes the K ATP channel, resulting in membrane depolarization and activation of voltage-gated calcium entry, which stimulates insulin secretion. Activating mutations in either KCNJ11 or ABCC8 impair the ability of the channel to respond to ATP and as a consequence, the K ATP channel remains open despite of elevated plasma glucose. This impairs insulin release and results in diabetes.
Sulfonylurea drugs bind to the SUR1 subunit of the K ATP channel and close it independently of ATP, thereby stimulating insulin release. They are now the therapy of choice in children with permanent neonatal diabetes caused by mutations in KCNJ11 or ABCC8 [5]. Prior to the discovery of the genetic cause of permanent neonatal diabetes, people with the condition were treated with insulin. However,~90% of people with neonatal diabetes due to a KCNJ11 mutation have been able to successfully transfer from insulin to highdose oral sulfonylurea therapy, which usually results in significant improvements in HbA 1c [5][6][7].
Unsuccessful transfer is determined by two factors: the sensitivity of the mutant K ATP channel to sulfonylurea inhibition, and the duration of diabetes prior to transfer [6,7]. Some mutant channels are less sensitive to sulfonylurea inhibition, and no cases with a mutation that shows < 60% block of the K ATP current by 0.5 mM tolbutamide in in vitro studies, have been able to transfer. In other cases, however, where some people have been able to transfer but others with the same mutation have been unable to do so, the ability to transfer is correlated with the duration of diabetes; an earlier age at transfer is more likely to result in success [6,7].
In those who have been unable to transfer completely to sulfonylureas, variable practice is observed with some reverting to insulin monotherapy, and others remaining on a combination of sulfonylureas and insulin or other agents [6][7][8]. The clinical benefits of remaining on combined insulin and sulfonylurea therapy have not been fully explored.
We report a case in which sulfonylurea therapy in combination with insulin treatment enabled a significant reduction in insulin dose, transition off insulin pump therapy and led to clinically meaningful improvements in HbA 1c , glycaemic variability and hypoglycaemia awareness. We undertook clinical and in vitro studies to investigate this phenotype further and show that the inability to transfer off insulin completely should not be considered a failure if the addition of sulfonylureas to the treatment regime results in improved glycaemia.

Molecular genetics
Genomic DNA was extracted from peripheral leukocytes using standard procedures. The coding regions and conserved splice sites of the ABCC8 and KCNJ11 genes were amplified by polymerase chain reaction (PCR) and the resulting amplicons sequenced using the Big Dye Terminator Cycler Sequencing Kit v3.1 (Applied Biosystems, Warrington, UK). The products were analysed on an ABI 3730 capillary sequencer (Applied Biosystems) and compared to the reference sequences (NM_000525.3 and NM_000352.3) using Mutation Surveyor v3.24 software (SoftGenetics, State College, PA, USA).

Functional studies
We used human Kir6.2 (GenBank NM000525, with E23 and I337) and rat SUR1 (GenBank L40624). Site-directed mutagenesis of Kir6.2 was performed as described previously [6,9]. Defolliculated Xenopus laevis oocytes were injected with 0.8 ng wild-type (or mutant) Kir6.2 mRNA and 4 ng SUR1 mRNA, and were incubated in Barth's solution at 18°C for 1-4 days. To simulate the heterozygous state of the heterozygous phenotype, we co-injected a 1 : 1 mixture of mutant and wild-type Kir6.2, together with SUR1. The resulting channel population (referred to here as hetG334V) contains a variable number of mutant subunits (between zero and four) in the Kir6.2 tetramer.

Clinical investigations
All studies were undertaken on the clinical investigation unit. Sulfonylurea test dose and oral glucose tolerance tests were undertaken according to the Exeter team protocol [10]. Briefly, following cessation of bolus insulin and a 75-g oral glucose tolerance test, a test dose of~0.1 mg/kg of glibenclamide was administered. Subsequent dose titration was carried out according to the transfer protocol.

Clinical data
A 22-year-old woman of South East Asian ethnicity was diagnosed with diabetes at the age of 3 weeks in her home country. This was assumed to be Type 1 diabetes. She was delivered following a caesarean section (breech position) at term weighing 3.2 kg. She developed normally during childhood with no delay in achieving milestones, no epilepsy and only a diagnosis of mild dyslexia. She was treated for 17 years with multiple daily injections of insulin, before commencing insulin pump therapy at the age of 18 years. She

What's new?
• KCNJ11 mutations causing permanent neonatal diabetes are treated with sulfonylureas, but not all individuals are able to transfer completely from insulin to sulfonylureas.
• Our data highlight that combining sulfonylurea treatment with insulin in those who are unable to fully transfer may still lead to clinically meaningful outcomes.
• We demonstrate improvements in endogenous insulin production, HbA 1c , glycaemic variability and hypoglycaemia awareness.
• These changes were not observed at initial doses of glibenclamide and improvements required higher sustained doses of glibenclamide.
• Combining insulin and sulfonylureas should be considered in those with permanent neonatal diabetes who are not able to transfer to sulfonylurea therapy alone.
had several documented episodes of diabetic ketoacidosis during her childhood and teenage years. Neither of her parents or her only sibling was affected with diabetes.

Genetic data
Having moved to the UK, genetic testing for permanent neonatal diabetes was requested on the basis of her diagnosis before 6 months of age. Genetic testing revealed a heterozygous KCNJ11 missense mutation, p.G334V (c.1001G > T), presumed to have occurred de novo, given the absence of diabetes in either parent. At the time of genetic testing, the woman had an above target HbA 1c of 88 mmol/mol, impaired awareness of hypoglycaemia, background retinopathy and significant abdominal lipohypertrophy (see Table 1 for baseline characteristics). There was no evidence of nephropathy or neuropathy. She required between 80 and 85 units of insulin per day via a pump, delivered with variable basal rates and bolus mealtime insulin. Continuous glucose monitoring (CGM) was undertaken in view of her frequent hypoglycaemia and revealed significant glycaemic variability (SD 4.0 mmol/l) with hyperglycaemic excursions and frequent hypoglycaemia.
Functional studies (see below) were undertaken to determine if the Kir6.2-G334V mutant channels were sensitive to sulfonylurea inhibition, with positive results. Consequently, a glibenclamide test dose was undertaken following a day admission, as part of clinical care (see Fig. 1). This revealed an immediate and significant rise in C-peptide following a 5 mg glibenclamide test dose, indicating that the b cells were capable of releasing endogenous insulin. A formal glibenclamide trial was therefore undertaken using the Exeter protocol [10] (see Fig. 2 for a summary of the dose titration), and insulin basal rates were reduced in parallel with increasing glibenclamide doses.
Within 8 weeks of commencing treatment, she was receiving a glibenclamide dose of 20 mg twice daily (0.67 mg/kg/day; 40 mg/day total) and her basal insulin (insulin glargine, Lantus) requirements had reduced to a total basal dose of 9.6 units/day via her insulin pump (0.4 units/ h). Her median blood glucose, as assessed by frequent home blood glucose monitoring, was 8.4 mmol/l. Basal insulin was further reduced, but glucose levels rose to a median of 12 mmol/l. Glibenclamide was therefore increased to a dose of 70 mg/day (1.2 mg/kg/day in three divided doses of 30, 10 and 30 mg) and basal insulin was reduced to 0.1 units/h (total basal dose 2.4 units/day). At this point, her median blood glucose was 8.6 mmol/l and an attempt to stop insulin completely was undertaken. However, following disconnection of the pump, glucose levels rapidly rose to between 20 and 25 mmol/l within 6 h (capillary blood ketones were 0.4 mmol/l), insulin was therefore recommenced the following day.
Following discussion, a higher dose of glibenclamide was given (40 mg three times daily; 2 mg/kg/day) for several months, however, insulin was still required to maintain  normoglycaemia. This was achieved through once-daily injections of long-acting insulin, enabling discontinuation of pump therapy. After commencement of glibenclamide, her HbA 1c fell to 52 mmol/mol (6.9%) and her random Cpeptide was 637 pmol/l ( Table 1). Her hypoglycaemia awareness was also restored. Repeat CGM revealed some hypoglycaemia and her glargine dose was reduced accordingly. The standard deviation (SD) of blood glucose measurements was significantly lower (at 2.7 mmol/l) than before sulfonylurea therapy. She is currently maintained on 40 mg glibenclamide three times daily and 10 units insulin glargine once daily, with no mealtime insulin bolus.

Functional studies
The latest structure of the K ATP channel reveals that residue G334 forms part of the ATP-binding site and may reduce the ability of ATP to bind to Kir6.2, either by steric interference or electrostatic repulsion [11]. We therefore first examined the sensitivity of wild-type (WT) and mutant channels to inhibition by MgATP. Figure S1 shows that WT channels were half maximally blocked (IC 50 ) by 20 AE 1 lM MgATP (n = 7). The ATP sensitivity of heterozygous (hetG334V) Kir6.2-G334V/SUR1 channels was significantly decreased (IC 50 = 163 AE 42 lM, n = 6; P < 0.01 vs. WT) and that of homozygous (homG334V) channels was reduced even further (IC 50 = 2.6 AE 0.5 mM, n = 3). There was also a substantial increase in the current amplitude at physiological levels of MgATP: at 3 mM, the percentage of unblocked current was 1.2 AE 0.2% (n = 7) for WT, whereas it was 12 AE 2% (n = 6) and 48 AE 4% (n = 3) for hetG334V and homG334V, respectively. This increase in current can account for the diabetes in the presented case and demonstrates that the mutation is pathological. We next analysed the effects of the Kir6.2-G334V mutation on the metabolic regulation and sulfonylurea sensitivity of the K ATP channel by measuring whole-cell currents. When WT K ATP channels are expressed in Xenopus oocytes they are normally closed, due to the high intracellular ATP concentration ([ATP] i ). However, they can be opened by lowering [ATP] i using a metabolic inhibitor such as Na-azide (Fig. 3a). Mutations that reduce the channel ATP sensitivity normally increase the whole-cell current in control solution, reflecting the fact that they are less blocked by resting [ATP] i. Both hetG334V and homG334V currents were substantially larger than WT both in control solution (Fig. 3a), as expected from their lower ATP sensitivity. The sulfonylurea tolbutamide, at a dose that maximally inhibits the WT channels (0.5 mM) blocked all three types of channel: by 97.2 AE 0.3% (n = 10) for WT, 91 AE 2% (n = 6) for hetG334V and 78 AE 2% (n = 7) for homG334V (Fig. 3b). These results predict that the individual should be sensitive to sulfonylurea therapy.

Discussion
We report a young woman with a KCNJ11-G334V mutation who, despite being unable to stop insulin completely, showed a significant improvement in glycaemic control when highdose sulfonylurea therapy was combined with insulin. It is noteworthy that this combination therapy also led to a marked improvement in glycaemic variability and in hypoglycaemia awareness. Furthermore, mealtime bolus insulin was no longer required.
The G334V mutation has been described in one previous case, a male, in whom complete transfer was also attempted, around 20 years post diagnosis [8]. Despite a rise in Cpeptide concentration, insulin had to be recommenced after a 5-day trial of glibenclamide. The increase in C-peptide demonstrates that b cells were at least partially functional following K ATP channel closure. Based on our results, it seems possible that the reported case could also have been FIGURE 2 Changes in glibenclamide dose (grey line), median blood glucose (black line) and daily basal insulin (dashed line) dose during glibenclamide dose titration. Note that when the glibenclamide dose is increased much less insulin is required to maintain glycaemia. managed on a combination of a low-dose insulin supplemented with sulfonylurea treatment. It is important to titrate sulfonylurea doses gradually over time to ascertain responsiveness, as demonstrated in our case, where starting doses of glibenclamide had little effect on glycaemia.
In vitro functional studies revealed the G334V mutation led to a small reduction in sulfonylurea sensitivity. However, the extent of block (91%) was well within the range predicted to enable complete transfer off insulin, based on previous findings (60-75%) [4,6]. That our case was unable to do so likely reflects the late age at transfer, because transfer success rate declines with duration of diabetes [6,7]. Transfer is unsuccessful in~30% of those over the age of 18, compared with < 5% individuals with the same mutation who are aged under 2 years. Rodent models suggest one reason for a better outcome when sulfonylurea is commenced early may be that bcell exposure to chronic hyperglycaemia leads to loss of insulin content, impaired b-cell metabolism-secretion coupling and reduced b-cell mass [12]. Although it is possible that with further titration of glibenclamide, transition off insulin may be achieved, the 3-month trial at higher doses yielded no further reduction in insulin. Stopping insulin was characterized by fasting hyperglycaemia and so insulin was recommenced.
Sulfonylurea therapy led to a reduction in glycaemic variability and abolished the requirement for mealtime insulin. It is likely that both these effects result from restoration of the incretin response. Incretins are only effective when the K ATP channels are closed, which leads to membrane depolarization, calcium influx and insulin release [13]. Thus, incretin-based drugs maybe of greater benefit when combined with sulfonylurea therapy.
It is possible that combination therapy would also be of benefit in people with mutations that have shown a reduced blockade in response to tolbutamide. Further studies are needed to explore this possibility.
Two studies have reported treatment outcomes on numbers of individuals with permanent neonatal diabetes. Thurber et al. [7] reported that 2 of 57 individuals remained on both insulin and sulfonylurea therapy. Babiker et al. [6], defined successful transfer as the ability to stop insulin completely. Our study suggests that a reduction in insulin dose should also be considered a successful outcome as it is associated with clinical benefits. Indeed, although no clinical outcomes were reported, Babiker et al. also recommended trialling combination therapy.
In conclusion, our data show that in people with KNCJ11 mutations who are unable to transfer completely to sulfonylurea therapy, a combination of insulin and sulfonylurea treatment may be of clinical benefit, improving glycaemic control, hypoglycaemia awareness and glycaemic variability. Furthermore, there may be improvements in quality of life, for example, by cessation of insulin pump therapy and managing mealtimes without bolus insulin. Thus, the inability to fully transfer off insulin should be still considered a valuable treatment in combination with sulfonylureas.

Addendum
While this paper was in review, a paper appeared reporting that two individuals who were initially managed solely on sulfonylurea therapy failed to retain good glycaemic control after several years, and in these people combination therapy did not produce better glycaemic control [14]. This contrasts with our individual who showed an improved outcome and may be related to diabetes duration or other factors.   FIGURE 3 Metabolic sensitivity of the wild-type and mutant K ATP channel. (a) Whole-cell currents were measured in oocytes expressing wild-type (WT; n = 10), heteromeric (hetG334V; n = 7) or homomeric (homG334V; n = 6) K ATP channels. The currents were recorded in control solution (black bars), after metabolic inhibition with 3 mM sodium azide (white bars) and after the further addition of 0.5 mM tolbutamide (tolb, grey bars). (b) Mean AE SEM tolbutamide block for WT (n = 10) and hetG334V (n = 6) channels, compared with the mean value for permanent neonatal diabetes mutant channels (n = 19 mutations). Tolbutamide block is expressed as the percentage block of the azide-induced current. The horizontal grey bar depicts the level of block above below which no patients respond to sulphonylurea therapy.