Sulfonylurea Therapy Benefits Neurological and Psychomotor Functions in Patients With Neonatal Diabetes Owing to Potassium Channel Mutations

Neonatal diabetes is a rare condition that develops during the first months of life with an estimated incidence of 1 in 90,000 newborns (1,2). Neonatal diabetes can be permanent or transient, relapsing around puberty after a period of remission. We recently reported that 42% of patients in a large cohort had a heterozygous activating mutation in the coding sequences of the KCNJ11 or ABCC8 gene (3), both of which encode the Kir6.2 subunit and SUR1 subunit, respectively, of the K_ATP channel. In β-cells, this channel induces membrane depolarization, thereby triggering insulin granule exocytosis (4,5). Sulfonylureas (SUs), which are widely used to treat type 2 diabetes, bind specifically to SUR1, closing the K_ATP channel via an ATP-independent mechanism and therefore increasing the release of insulin. We and others have shown that SUs are effective when used instead of subcutaneous insulin in children and adults with Kir6.2- or SUR1-activating mutations (5,6). SU therapy provided excellent metabolic control without the hypoglycemic episodes commonly seen with insulin.

Studies have established that ∼20% of patients with mutations in K_ATP genes have abnormalities of the standard neurological evaluation ranging from mild to severe developmental delay. The concomitant presence of treatment-resistant epilepsy and muscle weakness is known as developmental delay, epilepsy and neonatal diabetes (DEND) syndrome (3,7); intermediate DEND is a less severe phenotype without epilepsy. However, we recently reported that appropriate testing methods detected developmental impairments in >70% of patients with K_ATP gene mutations (3). These impairments adversely affect academic performance, social functioning, and quality of life. They might be improved by SU therapy, since K_ATP channels are found in many tissues, including the brain and muscle (8,9), and play a role in membrane polarization and cell functions. Anecdotal case reports support this possibility (2,7,10).

We hypothesized that a successful switch from insulin to SU in patients with neonatal diabetes owing to K_ATP channel mutations would improve developmental parameters. We conducted a prospective single-center cohort study of patients successfully switched from insulin to SU therapy. In-depth neurodevelopmental assessments were performed just before the switch and then 12–18 months later.

The appropriate ethics committee (CPP Île-de-France 1) approved the study. SUs are not licensed for use in children in France, and we therefore obtained approval for SU therapy in our patients from both the CPP Île-de-France 1 and the French Healthcare Agency (ANSM). Written informed consent was obtained from the parents or patients before study inclusion.

We prospectively included 19 patients seen at the Pediatric Endocrinology and Diabetology Unit of Hôpital Universitaire Necker Enfants Malades Paris between 10 July 2006 and 4 February 2009 who had neonatal diabetes owing to documented mutations in the coding sequence of KCNJ11 or ABCC8. Exclusion criteria were known hypersensitivity to SUs, severe renal failure (creatinine clearance <30 mL/min), severe hepatic failure (prothrombin rate <70%), porphyria, imidazole therapy, pregnancy, or no coverage by the statutory health care insurance system. Of the 19 patients, 18 completed the study and 1 was withdrawn when the parents decided to decline the switch from insulin to SU.

The switch from insulin to oral SU (glibenclamide) was performed as previously described (5,6). Patients were evaluated 2, 6, 12, and 18 months after inclusion.

The French Neuromotor Functions in Children (NP-MOT) battery (11) was performed at baseline and then after 12 months to evaluate development via qualitative (movements) and quantitative (speeds) assessments of muscle tone, gross motor control, laterality, praxis, gnosopraxis (12), digital and manual dexterity, body spatial integration, rhythmic tasks, and an auditory attentional task (12,13). (See Supplementary Data for details.) The NP-MOT battery is a standardized normative instrument with identical subtests for all ages (and expected saturation for patients aged 8 years or older) developed and validated by L.V.-D. For most of the tests, the cutoffs vary with age. Test scores are standardized according to scoring guidelines and expressed as SD of the population mean (failure if <1 SD) or as the percentile (failure if <20th percentile).

Overall test-retest reliability of the NP-MOT has been reported to range from 70 to 98% (11), and correlation coefficients with the Lincoln-Oseretsky motor development scale (similar to the Bruininks-Oseretsky Test of Motor Proficiency [14] for upper-limb coordination, balance, and bilateral coordination subtests) were 0.72 and 0.84 in two studies (15,16).

The same pediatric neurologist conducted a thorough neurological evaluation at baseline and then 6 and 12 months after SU initiation. An electroencephalogram was recorded at baseline in all patients.

Intellectual performance was evaluated at baseline and then after 12 months by the same examiner. All children completed all subtests of a standard measure of intelligence (Brunet-Lézine test, Wechsler Preschool and Primary Scale of Intelligence–Revised [17], Wechsler Intelligence Scale for Children–Fourth Edition [18], or Wechsler Adult Intelligence Scale–Third Edition). Other specific standardized tests consisted of visual-perceptual-motor tests assessing visual construction skills (reproduction of a block design [19]), visual-spatial structuring (manual copy followed by visual-spatial memory of a complex geometric figure [20]), and the Development Test of Visual-Motor Integration (21), which involves manually copying 24 geometric drawings of progressively increasing complexity. Visual-spatial attention was assessed using a bell-crossing test (22) similar to that developed by Gauthier, Dehaut, and Joanette. Mental executive function was evaluated using the Porteus Maze Test (23) and the Tower of London test (24). In addition, the patients performed visual perception (visual gnosis) tasks (recognizing tangled lines and naming animals seen in outline from the rear) (25) and a language screening battery (22). Hyperactivity was defined using the DSM criteria. Children’s behavior was observed by the same examiner at inclusion and 12 months after the switch. The examiner recorded the symptoms characterizing the disorder: inattentiveness, impulsivity, and overactivity. Parents were administered an interview covering a broad range of child behaviors. As a part of this interview, parents were asked about the presence/absence of hyperactivity symptoms.

Each patient underwent an electroretinogram and visual evoked potential recordings, as recommended by the International Society for Clinical Electrophysiology of Vision (26,27), at baseline and then 6 months later. Flash visual evoked potentials were recorded in younger patients (monocular, patient no. 10, or binocular if patching an eye was not possible, patient nos. 1, 6, 18, 15, 7, 12, and 14) and pattern reversal visual evoked potentials in older patients (100% contrast; reversal at 1.0 Hz; 60’, 30’, or 15’ squares, patient nos. 4, 9, 11, 17, 3, 5, and 16). No recordings were performed in patient nos. 2 and 13.

Electrophysiological testing was performed at baseline and then after 6 and 12 months in children older than 6 years of age and carrying a KCNJ11 mutation (as SUR1 expressed by ABCC8 is not expressed in the muscle). A standardized protocol ensuring reproducible and painless electrophysiological testing of skeletal muscle excitability was used (28). Briefly, compound muscle action potentials were recorded from the right and left abductor digiti minimi muscles after supramaximal electrical stimulation of the ulnar nerves at the wrists. Recordings were repeated before and after voluntary contraction of the recorded muscle. If the response changed after exercise, care was taken to check that the electrode positions were unchanged and that nerve stimulation remained supramaximal.

Two provocative tests were performed. The first was a repeated short exercise test at room temperature on the left hand (three maximal, isometric, 10-s abductor digiti minimi contractions separated by 50-s rest periods during which compound muscle action potentials were recorded every 5–10 s). Then, a long exercise test was performed at room temperature on the right hand (maximal, isometric, 5-min abductor digiti minimi contraction followed by compound muscle action potential recording every 5 min for 40 min). Compound muscle action potential amplitude, total duration, and total area were expressed as percentages of pre-exercise values.

Needle electrode recordings were obtained from five muscles (deltoid, extensor digitorum communis, first interosseus dorsalis, vastus medialis, and tibialis anterior) to look for myotonic discharges or evidence of myopathy. Finally, motor conduction of the ulnar and peroneal nerves and sensory conduction of the right and left superficial peroneal nerves were measured. Amplitude, latency, and conduction velocity of the electrophysiological signals were compared with normal values for the laboratory.

MRI was performed at baseline. High-resolution images were acquired using a 1.5-T Signa System machine (GE Healthcare, Milwaukee, WI) with a three-dimensional T1-weighted fast spoiled gradient recalled imaging sequence (repetition time [TR]/echo time [TE]/inversion time/NEX: 10.5/2.2/600/1, 10°, matrix 256 · 192; 124 axial slices, 1.2-mm thickness, 124 contiguous slices, 22 cm field of view), an axial fast spin echo T2-weighted sequence (TR/TE: 6,000/120, 4-mm slices, 0.5 mm gap, 22 cm field of view), coronal fluid-attenuated inversion recovery (FLAIR) sequences (TR/TE/TI: 10,000/150/2,250, 4-mm slices, 1-mm gap, 24 cm field of view), and ¹H-MRS.

Plasma glibenclamide concentrations were determined after standard liquid-liquid extraction by liquid chromatography–ion-trap tandem mass spectrometry as previously described (29).

Data were described as median (range or interquartile range where indicated) for quantitative variables and as number (percentage) for qualitative variables. Comparisons of data at different time points were performed with the paired Wilcoxon or McNemar test. The nonlinear mixed-effect modeling program NONMEM (version VII, release 1) was used to compute the area under the curve of glibenclamide concentrations over 24 h.

We studied 18 patients aged 5 months to 18 years; eight were younger than 4 years old. Metabolic control improved after the switch to glibenclamide therapy, and no patients experienced hypoglycemia. One patient had a remission allowing glibenclamide discontinuation after the 12-month evaluation (Table 1).

Median glibenclamide dosage was 0.37 mg/kg/day (range 0–1.4) at month 18. Median plasma glibenclamide concentration was 50 μg/L (interquartile range 21.5–118), and median area under the time curve over 24 h was 1,335 μg · L/h (511–2,122).

Neurological impairments were found in a single patient (patient no. 17) and consisted of a global pyramidal syndrome with spasticity and mild walking disability, mild mental retardation, and seizures (DEND). The baseline electroencephalogram was normal in 15 patients and abnormal in three (patient nos. 10 and 12, frontal spikes, and no. 17, temporal spikes).

As previously reported, NP-MOT showed developmental coordination disorder, attention deficit, or both in 17 of the 18 patients. Impairments in the planning or programming of movements were combined with hypotonia in 15 patients. Tables 2 and 3 report the impairments at baseline and the improvements after 12 months of SU therapy. Individual results are reported in Supplementary Tables 1 and 2.

These abnormalities improved markedly after 12 months of SU therapy. Thus, in young children, tone was closer to the normal for chronological age and attention disorders had resolved, with normalization of gross motor skills in all children but one and of fine motor skills in three children. In children older than 4 years of age, tone, visual attention, and laterality improved. These changes benefited gesture conception and realization by ameliorating sensory integration and central gesture planning. Thus, body spatial integration, gesture imitation, and visual-spatial integration were improved. Motor skills were the area with the greatest improvements during SU therapy.

Language disorders or delay were found in six patients at inclusion. They were not improved after 12 months of therapy (Tables 4 and 5).

Social skills were altered at inclusion in four children younger than 4 years of age. Two children presented with hyperactivity. Sociability was not significantly improved after 12 months of SU, whereas hyperactivity had resolved (Tables 1, 4, and 5).

Development was moderately delayed in five children younger than 2.5 years of age. In older children, intelligence scores were altered. Total IQ score was decreased in four patients. None of these items improved significantly with SU therapy in either age-group (Tables 4 and 5).

Ophthalmoscopy, ocular anatomy, and electroretinography were normal in all patients. Visual evoked potentials were delayed in all but one of the eight youngest patients. Pattern reversal potentials were normal in all but one patient. After 5 months of SU therapy, follow-up examinations in five of eight patients showed normalization of visual evoked potentials in two patients.

The 10 patients who underwent electrophysiological testing had normal motor and sensory nerve conduction studies at baseline. The short and long exercise tests induced minor and nonsignificant changes in compound muscle action potential amplitude and area, suggesting absence of membrane excitability impairments (Fig. 1).

MRI was performed in 17 patients (one family refused MRI) and was abnormal in 12 (71%). Nonspecific findings included cerebellar venous angioma (n = 1), posterior periventricular white matter abnormalities (n = 4), and Virshow-Robin space dilation (n = 2). Eight patients had multiple punctate white matter hyperintensities on T2 and FLAIR sequences (Fig. 2B), and two had hyperintensities in the nucleus raphe pontis (Fig. 2). ¹H-MRS was consistently normal.

We conducted a systematic prospective evaluation of potential effects of SU therapy in patients with neonatal diabetes owing to K_ATP channel mutations. In keeping with our previous work (3), severe neurological deficiencies were uncommon, whereas developmental coordination disorders, language impairments, and attention deficiencies were often detected when appropriate tests were used. Electrophysiological testing showed no evidence of the membrane excitability alterations reported in muscle channelopathies, suggesting a neurological origin to the hypotonia and motor impairments. The detection by cerebral MRI of white matter abnormalities in most patients supports a central neurological origin to the motor and developmental impairments. The electrophysiological assessment of visual function also argued for a central origin, as it showed alterations in central sensory integration. Motor abnormalities were noticeably improved after 12 months of SU therapy. In the younger patients, hypotonia and attention deficits were greatly improved or corrected, contributing to ameliorations in motor skills. In patients older than 4 years, tone, laterality, and visual functions showed the greatest improvements, and these contributed to measurable improvements in gesture conception and realization. Neither intelligence scores nor language impairments improved within the study period, whereas hyperactivity was corrected in the affected patients. SU plasma concentrations and area under the curve values were within the therapeutic range for adults (29). Our findings support an effect of SU therapy on the central nervous system in patients with permanent or transient neonatal diabetes owing to KCNJ11 or ABCC8 mutations. This effect seems to target specific brain regions, as the greatest improvements were for cerebellar functions (tone, gesture imitation, and body spatial integration) and thalamic functions (tone, laterality, gesture imitation, and visual-spatial construction) involved in various steps of gesture conception and realization such as sensory integration and motor planning.

Several anecdotal case reports are consistent with our findings. A 23-month-old boy with a KCNJ11 mutation and marked developmental delay experienced substantial improvements in motor function and attention with no change in intelligence scores (30). SU therapy started at 12 years of age in a boy who had a KCNJ11 mutation induced marked improvements in mental and motor function, with an increase in the Mental Developmental Index from 29 to 39 months within 6 months (7). Attention and gross and fine motor skills improved. MRI showed white matter abnormalities similar to those in our population. In a 6-year-old girl with a KCNJ11 mutation, improvements in muscle tone, motor function, attention, and hyperactivity were apparent 7 months after switching to SU therapy (10). A study of 19 children with KCNJ11 mutations focused on visuomotor performance, which correlated negatively with age at SU initiation; however, the patients were tested on a single occasion: after SU initiation (31). In our study, SU therapy in the younger patients was associated with a direct neuronal effect, improving visual conduction and thereby sensory integration and visual attention. This effect probably participated in the motor gains by improving the integration of visual sensory signals.

The neurological and developmental abnormalities seen in patients with neonatal diabetes owing to KCNJ11 or ABCC8 mutations are the result of the genetic abnormality and not of diabetes. Furthermore, they are ascribable to the effects of the mutation in central nervous system cells rather than to muscle excitability impairments. The Kir6.2 and SUR1 subunits are widely expressed in all brain regions in rodents and probably form the K_ATP channel pores in most neurons (8,9). K_ATP channels are also found in muscle. However, we found no evidence of abnormalities in muscle responses to exercise in patients with mutations in the KCNJ11 gene that encodes the Kir6.2 subunit. In contrast, exercise-induced decreases in the amplitude and area of compound muscle action potentials have been reported in patients with periodic paralysis due to mutations in sodium (SCN4A), calcium (CACNA1S), or Kir2.1 potassium (KCNJ2) channel genes (28,32). Thus, the muscle disorders in patients with KCNJ11 mutations do not seem ascribable to changes in muscle membrane excitability. Furthermore, they are ascribable to the effects of the mutation in central nervous system cells rather than to muscle excitability impairments (33). None of our patients had a history of recurrent severe hypoglycemia. The MRI changes in our patients differ from those reported in children with type 1 diabetes (34,35). These last two facts support the direct effect on the central nervous system rather than an indirect effect due to improved metabolic control.

Limitations of our study include the small sample size, although we included most of the patients in France who had neonatal diabetes owing to potassium-channel subunit mutations and had not yet been switched to SU. A control group would have been ethically unacceptable. The short study period was sufficient to see substantial coordination disorders improvement allowing complex tasks such as handwriting to be better performed. If we found no major changes in language, intelligence, or sociability, these functions might improve with time, as this short study was sufficient to correct hyperactivity signs, or, alternatively, SUs may fail to target the brain regions involved. A major strength of our study is the use of a standardized normative neurodevelopmental test battery validated in normal French children and performed by a single examiner. This battery allowed us to obtain accurate information about SU effects on developmental parameters.

SUs have already been suggested as a neuroprotective drug after strokes (36,37). Our findings strongly suggest that SU therapy improves neurodevelopmental parameters in patients with neonatal diabetes owing to potassium-channel subunit mutations and acts via a central mechanism. The greater improvements in younger patients indicate a need for establishing the diagnosis early to allow prompt initiation of SU therapy. Further follow-up of our patients will provide information about the kinetics of SU effects on neurodevelopmental parameters.

The GlidKir Study Group members are as follows: Claire Le Tallec and Nicole Ser, Departement de Pédiatrie, CHU Toulouse, Toulouse, France; Sylvie Nivot-Adamiak and Marc de Kerdanet, Service de Pédiatrie, CHU Rennes, Rennes, France; Maryse Cartigny and Jacques Weill, Service de Pédiatrie, CHU Jeanne de Flandre, Lille, France; Sabine Baron and Emmanuelle Ramos-Caldagues, Service de Pédiatrie, CHU Nantes, Nantes, France; Henri Bruel, Service de Pédiatrie, Hopital de Le Havre, Le Havre, France; Anne Lienhardt-Roussie, Service de Pédiatrie, CHU Limoges, Limoges, France; Guy-André Loeuille, Service de Pédiatrie, Hopital de Dunkerque, Dunkerque, France; Berthe Razafimahefa, Pédiatrie-Nouveaux nés, Hopital Georges Sand, La Seyne sur Mer, France; and Rachel Reynaud and Gilbert Simonin, Service de Pédiatrie, Hopital La Timone, Marseilles, France.

Acknowledgments. The authors thank Paul Czernichow and Jean Jacques Robert for their long-standing support to studies on neonatal diabetes and Philippe Froguel, Martine Vaxillaire, and Amélie Bonnefont, Institut Pasteur, Lille, France, for their cooperation with work on identifying genes responsible for neonatal diabetes. The authors thank Myriam Faivre and the nursing team of the Pediatric Endocrinology and Diabetology Department, as well as Sandra Colas for her help in managing the protocol and the clinical research unit team—both at Hôpital Universitaire Necker Enfants Malades Paris.

Funding. This study was sponsored by Assistance Publique-Hôpitaux de Paris and received a government grant managed by Agence Nationale de la Recherche under the “Investments for the Future” program (reference ANR-10-IAHU-01). The work was performed within Département Hospitalo-Universitaire AUToimmune and HORmonal diseaseS. It was partly funded by Agence Nationale de la Recherche–Maladies Rares Research Program grant ANR-07-MRAR-000 (to M.P.), Transnational European Research Grant on Rare Diseases grant ERANET-09-RARE-005 (to M.P.), and Société Francophone du Diabète–Association Française du Diabète (to M.P.). K.B. received a CIFRE grant from the French government and was supported by the French Ministry of Higher Education and Research and Société Française de Pédiatrie. Support was also received from LabEx Revive and from the Bettencourt-Schueller Foundation (R.S.) and Aide aux Jeunes Diabétiques (to M.P.).

Duality of Interest. K.B. was supported by HRA Pharma. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. J.B., K.B., and A.S. collected data. J.B., C.E., K.B., and M.P. wrote the manuscript. J.B., C.E., M.B., and J.M.-T. analyzed data. K.B., A.S., I.F., and M.P. designed the study. E.F. performed the electromyography and analyzed the results. N.B. performed the MRI and analyzed the results. N.B.-B. performed the neurological examination. M.V. performed the intelligence tests. E.B.-Q., I.I.-M., and M.B. performed the electrophysiological assessment of visual function and analyzed the results. C.G. and Z.D. performed the pharmacological assessment. R.S. reviewed the manuscript. H.C. performed the genetic analysis. L.V.-D. performed the neuropsychomotor tests. M.P. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the Annual Meeting of the European Society for Paediatric Endocrinology, Dublin, Ireland, 18–21 September 2014, and at the 40th International Society for Pediatric and Adolescent Diabetes Conference, Toronto, ON, Canada, 3–6 September 2014.