Diabetes Technology and Therapy in the Pediatric Age Group

Pump therapy with rapid-acting insulin allows for a more physiologic insulin replacement and may thus contribute to improved metabolic control, thereby reducing the risk of long-term complications. Previous large registry data (5–7) already demonstrated better glycemic control in children with T1D treated with insulin pump therapy compared with children treated with multiple daily injections. However, the association of pump therapy compared to injection therapy with the risk of acute diabetes complications has not been studied comprehensively. Several studies reported an increased risk of DKA associated with insulin pump therapy in pediatric patients with diabetes (8,9) raising concerns about the safety of pump therapy. The results of this study favor pump therapy, with lower rates of acute complications (severe hypoglycemia, hypoglycemic coma, and DKA) and with lower glycated hemoglobin (HbA1c) levels reflecting improved metabolic control.

In the current study, all patients continuously used either pump therapy or injection therapy during the entire observation period of 12 months, thus excluding treatment crossover. The present study strengths include its prospective design, the very large sample size of a population of more than 30,000 young patients with T1D, and the statistical methodology that included a matched-pair approach that allowed a direct comparison of hypoglycemia and DKA frequencies in pump users and injection users. Whereas previous randomized clinical trials have been too small to assess the risk of these short-term diabetes complications, this study provides outcome data in clinical use that are likely representative of pediatric patients with T1D.

In the present study, the reduced risk of severe hypoglycemia with pump therapy was associated with lower total daily insulin dose. The use of rapid-acting insulin analogues with pump therapy allows for more flexible therapy with lower glycemic variability, leading to lower rates of acute and long-term diabetes complications. Prevention of DKA is an integral part of diabetes education and should be emphasized in patients treated with pump therapy, thereby reducing their incidence of DKA.

The study limitations include its nonrandomized, observational design that may be prone to selection bias.

Another limitation is lack of other data relevant to glycemic control, hypoglycemia, and DKA as: the intensity of diabetes education, data about the sociodemographic characteristics of the study population and family support, and the lack of data regarding the use of continuous glucose monitoring.

There is some evidence for long-term negative neurodevelopmental outcomes in neonates with severe hypoglycemia (10). Very preterm infants have an even greater risk of poor neurological outcomes. To reduce hypoglycemia as a possible risk factor in these patients, Galderisi and colleagues performed a randomized controlled trial at the University Hospital of Padua of a system that uses CGM to control the delivery rate of intravenous (i.v.) dextrose via a proportional-integrative-derivative (PID) controller. Infants born before 32 weeks of gestation or with birth weight below 1500 grams and less than 48 hours old were randomized to receive 7 days blinded CGM and standard-of-care i.v. dextrose or an unblinded CGM with alarms and i.v. dextrose delivered according to the PID controller. Outcomes were based on CGM data from a Dexcom G4P.

An important issue brought up in the paper is that detachment of the device more than once occurred in two subjects who discontinued the intervention as per protocol. This is very important for informed consent in other neonatal CGM studies.

The authors conclude that delivery of dextrose by a controller using CGM data improved measures of glucose by the same CGM. While these results demonstrate technology proof of concept, it must be noted that there was no effect on the short-term clinical outcomes the team evaluated. As the Dexcom G6 no longer requires calibration, use of this technology in the neonatal intensive care unit may significantly decrease the burden of managing glucose for health-care providers. Larger studies will be needed to assess whether there are long-term clinical benefits for these children.

There is a tremendous need to address glycemic control in adolescents and young adults (11,12). The BE IN CONTROL study assessed the effect of financial incentives on adherence to glucose checks and glycemic controls. During the first 3-month period, daily adherence to glucose checks was significantly greater in the incentive group. The difference was not maintained after the incentive was removed, and HbA1c did not significantly differ between groups at 3 or 6 months. The incentivized behavior did increase, but the authors conclude that increasing glucose checks in isolation did not improve glycemic control.

The results are fascinating and do suggest that incentivizing desired actions is efficacious while the intervention persists. Economic factors are often highlighted in health-care disparities research, but targeted research on interventions is more limited. One criticism of the current study is the way behavior was incentivized. It is frequently more useful to positively reinforce good behaviors than punish undesired behaviors. The phrasing here was that the subjects would lose money if they did not check. It may have been more useful to suggest a $2 reimbursement for the desired behavior. Additionally, it is possible the wrong behavior was incentivized. Since the primary outcome was HbA1c, incentivizing mealtime boluses or glucose reading within target may have had a more direct effect on glycemic control. It is not clear how long a behavior must be incentivized to persist. If the correct behavior is incentivized and improves glycemic control, it may be cost feasible for payers.

FSL offers a factory-calibrated sensor that can be worn for 14 days without manual calibration by using blood testing, a reader combined with a standard glucometer, and dedicated software. The user receives a current glucose measurement along with historic results from the preceding 8 hours by scanning the sensor with the FreeStyle reader. The glucose trends are displayed as arrows on the reader along with alerts. The device requires active scanning by the user, and it does not report alerts in real time. So far, only one study tested the accuracy of FSL in a pediatric population that examined the use of FSL in home-based conditions, and the data obtained by isCGM were hidden from participants (13).

Initial observations suggested that FSL users may rely solely on flash glucose monitoring (FGM) and not perform SMBGs. This raises the question whether isCGM is accurate enough to replace standard glucometers, or should it be just a complementary addition to SMBG. The current study demonstrated that most FSL readings were localized in zone A or B of the error grids for more than 98% of cases, which means that they would not impact clinical decisions in a severe manner. Most of the children expressed willingness to continue using this device. It is possible that the good reception of isCGM may result from its minimal invasiveness when compared with standard blood glucose measurements and added value in the form of easily interpretable glucose trend arrows. Therefore, it can be a good compromise in those young patients who do not perform frequent SMBG and do not want to use the CGM. However, the main limitations of this device are the requirement of compliance of active scanning by the user and the lack of alarms, which may be crucial for those patients with hypoglycemic unawareness and during the night hours. The main limitations of the study are its short (12 days) duration and comparing FSL readings only with the Contour Plus One glucometer measurements. Comparing results of the FSL with Yellow Springs Instrument (YSI), which is widely accepted as a method for reference measurements, and system calibration by most manufacturers of blood glucose monitoring systems would be more accurate. Studies with longer observation times are needed to evaluate whether FSL could provide lasting improvement in glycemic outcomes.

We have to consider that when glucose levels were falling rapidly, the FSL reading differed by up to 20% from blood glucose measurements. This difference may be due to the lag time between glucose concentrations in the interstitial fluid reaching those in the blood. Therefore, patients and parents have to be instructed to confirm the measurements with SMBG in such situations and also in situations of glucose levels in the hypoglycemic range or when the patient's sensations are inadequate to the device readings.

Progress in diabetes technology has been rapid and relentless, although still slower than people with T1D who desire a fully automated, burden-free, inexpensive, and reliable technology would wish. Predictive low-glucose suspend systems mark an important advance that sensibly shuts off insulin when hypoglycemia is predicted (14). The MiniMed 640G system available in many parts of the world preceded the 670G hybrid closed-loop system (15) and adds protection against hypoglycemia. Data such as these by Abraham and colleagues add important data from a longer study (6 months) to support the use of predicted low-glucose suspend. Although systems that dose insulin will increasingly be used, for some people this step on the pathway may be a start as the technologies develop and user experience expands.

In the past decade, artificial pancreas research has progressed from single day, hospital-based, single participant studies to the first U.S. FDA approval of a first-generation hybrid closed-loop system for people with T1D. Future iterations of artificial pancreas systems will improve usability and performance. One challenge for current systems is exercise (16,17). While much work is needed to improve current systems performance, this study makes clear that existing systems do better than open-loop control as far as hypoglycemia prevention and time in range. This study design placed the bar high with a control group of sensor-augmented pump therapy that was remotely monitored by a physician 24 h/day for the duration of the study. How the CLC system (71% time in range) would compare with typical glucose control in a free-living skiing situation is uncertain, but undoubtedly the glucose control would not be as tight as in this comparison group (64% time in range during the study), as these participants had a mean HbA1c of 8.5% at entry to the study. Moreover, studies such as these are required to test artificial pancreas systems in settings that mimic user's real-life activity. Future iterations of these devices will increasingly be tested in challenging situations to provide insight on how to further improve their ability to benefit people with diabetes.

Landmark advances have been made in the past few years in the commercial availability of diabetes technology to the pediatric population. The MiniMed 670G system was the first HCL system available and lowered HbA1c in both adult and pediatric participants of the clinical trial. This paper by Messer adds to our understanding of how to initiate these systems in this population and provides anticipatory guidance on what education and dose adjustments will maximize usability (18,19). Providers should consider initial C-to-I adjustments to dose more insulin with meals and snacks and to shortening active insulin time. It will be especially important that users of the 670G have realistic expectations about this first-generation system and what efforts are required to remain in Auto Mode (i.e., closed loop) such as required calibration, bolusing for meals and snacks, and how to troubleshoot system challenges. The 670G marks a tremendous advance in diabetes technology, yet the 15% decline in time in Auto Mode over 3 months while in a fully supported, industry-sponsored clinical trial raises questions about what real-world clinical use might be in this patient population. Given the clear benefit with respect to increased time in range and lower HbA1c and hypoglycemia, improving usability remains a primary goal for next generation artificial pancreas systems.

Despite evidence of the benefits of exercise for those with T1D (20), there remains a risk of hypoglycemia both during the activity and the following night (21). Dovc and colleagues tested the performance of the Glucositter fuzzy-logic controller on young people with T1D engaging in unannounced exercise routines conducted in the hospital. The study was well designed with objective measurements of exercise intensity. If hypoglycemia can be prevented during exercise, weight loss becomes more likely as the individual does not have to consume additional calories to treat lows. Additionally, the decrease in insulin with greater time in range implies improved insulin sensitivity or increased glucose utilization by insulin-independent mechanisms. One can hypothesize that this profile improves the efficacy of exercise in those with T1D.

Following FDA approval of the Medtronic 670G in September 2016 (22), both Insulet and Tandem have been at work creating closed-loop insulin delivery systems. Insulet's system utilizes the OmniPod patch pump, Dexcom CGM, and a model predictive control algorithm running on a tablet for automated insulin delivery. This was a proof of concept for the algorithm and not the commercial system, which will undoubtedly be designed very differently on final release. As the first human trial of this system, the initial group of 24 adult subjects received a reduced 80% meal bolus, while another 10 adults, 12 adolescents, and 12 pediatric age patients received full meal boluses.

Importantly, carbohydrate intake was restricted to 30–90 grams and exercise was limited. While these are very restrictive conditions, it is a necessary first step in assessing safety. The investigation is significant, as some individuals prefer the patch pump design to a tethered pump. Additionally, it is the first test of a commercial model predictive control insulin delivery system. Currently, both Insulet and Tandem are paring with Dexcom, who is now producing a sensor that does not require calibration. This will provide significant competition for the Medtronic 670G system, wherein the Guardian 3 sensor requires at least one calibration every 12 hours.

The study hypothesis was that adolescents with high levels of albumin excretion have an increased glomerular filtration rate, dyslipidemia, and increased arterial stiffness, aortic intima-media thickness, and cardiac autonomic dysfunction. Therefore, they might benefit from ACE inhibitors and statins that may lower the rate of cardiovascular risk factors as hypertension and dyslipidemia. The present study demonstrated that neither ACE-inhibitor nor statin therapy over a relatively short period of time (2–4 years) altered the change in repeated measures of the albumin-to-creatinine ratio among adolescents with T1D.

Recent data from the Diabetes Control and Complications Trial and the Epidemiology of Diabetes Interventions and Complications study indicate that even intermittent microalbuminuria may predict cardiovascular risk (23). In the current study, lower rates of progression to microalbuminuria were observed with the ACE inhibitor than with placebo, and although this finding was not significant, it may be of clinical relevance, since it was related to reduction in the variability of albumin-to-creatinine ratio.

The main study strengths are the large sample size and double-blind, randomized, placebo-controlled trial design. The main study limitation is the relatively short duration of exposure to the trial drugs. Despite the absence of response with respect to the carotid intima-media thickness in this trial, long-term follow-up of the cohort will be important to assess whether early intervention during this critical period of adolescence with ACE inhibitors or statins will result in a later effect, as reported in other trials of patients with type 2 diabetes (T2D) of glucose-lowering, lipid-lowering, or blood-pressure–lowering interventions in which reduced vascular complications emerged beyond the duration of the original trials (24,25). The concept of a “metabolic memory,” meaning that diabetic vascular stresses persist after glucose normalization, has been supported for both T1D and T2D. Mechanisms for propagating this memory appear focused on the nonenzymatic glycation of cellular proteins and lipids and on an excess of cellular reactive oxygen and nitrogen species, in particular originating at the level of glycated mitochondrial proteins and perhaps acting in concert with one another to maintain stress signaling independent of glucose levels. Thus, there is a need for early aggressive treatment aiming to “normalize” metabolic control together with the addition of agents that reduce cellular reactive species and glycation in order to minimize long-term diabetic complications. Therefore, follow-up of the present cohort will be essential to evaluate the long-term potential benefits of early intervention with ACE inhibitors and statins.

Data regarding complementary therapies (used in addition to insulin) in adults with T1D that do not attain the target glycemic control have already been reported (26,27). In pediatric T1D patients, however, there are limited data on this topic. Studies in the adult patients with type 2 diabetes showed that metformin treatment improves glycemic control and reduces cardiovascular mortality without weight gain or risk of hypoglycemia (28,29). Previously it was reported that among overweight adolescents with T1D, the addition of metformin to insulin did not improve glycemic control after 6 months (30). A recent systematic review and meta-analysis of randomized controlled trials conducted on children aged 6 to 19 years diagnosed with T1D examined the effect of adding metformin to standard insulin therapy. The evidence did not support use of metformin in T1D adolescents to improve HbA1c. However, metformin provided modest reduction in the daily insulin dose and BMI (31).

The current study was for longer duration (1 year) and demonstrated improved glycemic control as well as a good safety profile during the trial. This study contributes important knowledge of metformin use in children with T1D—in addition to improved glycemic control, reduction in insulin, and an increase in estimated insulin sensitivity, it has shown an effect on vascular function and support of CVD benefit of metformin in adolescent T1D. Since most of the patients were not overweight, the results can be applied to patients of a wider weight range. The main limitation was that the duration of only 12 months did not provide the opportunity to detect changes in vascular structure, only changes in vascular function. Change in vascular structure likely requires several years of follow-up in T1D (32,33).

The SEARCH for Diabetes in Youth Study demonstrated that 27% of youth with T2D participating in that study had poor glycemic control with HbA1c ≥9.5% (≥80 mmol/mol) (34), and the Pediatric Diabetes Consortium T2D Registry has shown that rescue therapy with insulin, in metformin treatment failures, rarely lowers HbA1c to <7.5% (<58 mmol/mol) (35), the target level recommended by the American Diabetes Association and the International Society for Pediatric and Adolescent Diabetes. Therefore, new therapies in addition to metformin and insulin are required in youth with T2D below 18 years of age.

Overall, the results from this trial are consistent with the clinical efficacy and safety profile that has been reported for linagliptin in adult patients with T2D, favoring linagliptin 5 mg over 1 mg. Although the number of patients enrolled in the current study was sufficient for the assessment of the key PD outcome of DPP-4 inhibition, the main limitation is the small sample size, which makes the generalization of data difficult. However, the findings of the current study support the need for evaluation of the long-term safety and efficacy (including further PK data) of linagliptin 5 mg in a phase 3 study in pediatric patients with T2D.

Although the results of this investigation were disappointing in that they do not provide hope for prevention therapy for those who are antibody positive, much was learned from this well-performed study. Congratulations are in order for the research teams and study participants for the effort required for this decade-long project, which demonstrates feasibility for such multinational research in early prevention of T1D. Over 130,000 participants were screened to find the 560 randomized participants. While the overall findings were negative, the intervention was safe. Moreover, there will be ongoing debate and further analyses regarding the prespecified secondary stratum in which there was a statistically significant delay to T1D of 31 months. More mechanistic studies on samples from the trial promise to add insight into results and help refine hypotheses for future study.

TrialNet and others have developed a stage of T1D schema (36) that emphasizes the need to intervene early in the pathophysiologic course of T1D to preserve beta-cells and avoid the acute and long-term complications that come once T1D develops. This study highlights the importance of developing shorter-term markers of T1D progression to allow for more rapid studies. Future research may also take a multidrug approach with a goal of a personalized approach based on better understanding of what is increasingly being understood as a heterogeneous disease process requiring tailored approaches.