Abstract
Background:
A proprietary continuous glucose monitoring (CGM) system that uses third-generation sensor technology has previously been shown to exhibit stable, single-digit mean absolute relative difference (MARD) and 15-day wear life in adults with diabetes.
Method:
This was a prospective, single-arm study of pediatric participants aged 2 to 17 years that were enrolled at 3 clinical centers. Participants wore 2 sensors, one on each side of the abdomen, and were randomized to assessment of device performance on days 1 and 2, days 7 to 9 or days 15 and 16. Device performance across a range of metrics was assessed by comparison with venous blood glucose values obtained using a laboratory reference device. Per-protocol results are reported for the sensor with the inferior MARD.
Results:
The per-protocol set comprised 75 participants, of whom 16 were aged 2 to 5 years. The overall MARD was 8.89%, and MARDs at days 1 and 2, 7 to 9 and 15 and 16 were 8.65%, 8.70% and 9.30%, respectively. DTS error grid analyses showed that 100.0% of data pairs fell in clinically acceptable zones A+B. True alarm rates for hypoglycemia and hyperglycemia were high, at 98.6% and 98.8%, respectively. Mean sensor wear life was 14.4 days, and participant/guardian assessments of sensor usability revealed high satisfaction with respect to system assembly, sensor insertion and overall comfort. No device-related or skin-related adverse events were reported.
Conclusions:
In a pediatric population, the novel CGM system demonstrated accurate and stable performance across its 15-day wear life and showed high usability.
Keywords
Introduction
Continuous glucose monitoring (CGM) systems that use a minimally invasive approach to measure glucose in interstitial fluid have led to significant advances in the management of diabetes. As a result, management guidelines now recommend the integration of CGM with multiple daily insulin injections or to guide continuous insulin infusion as the standard of care for outpatient diabetes management.1-4 The American Diabetes Association guidelines also recommend CGM for patients on basal insulin. 2 These clinical advances have been enabled by developments in sensor technology, which have afforded increased sensor accuracy, reliability and wear times. 5
The iCan sensor is a factory-calibrated 15-day CGM sensor that uses third-generation, direct electron-transfer technology with modified glucose dehydrogenase (GDH) with a flavin adenine dinucleotide (FAD) cofactor as the electron acceptor. 6 This novel sensor has been tested in a total of 507 people with diabetes in trials conducted at clinical centers in the United States, Europe and Asia, with a total of 295 adults7-9 and 212 children aged 2 to 17 years enrolled.10,11 Here, we report the results on the performance, safety and usability of the sensor in a pediatric population aged 2 to 17 years.
Methods
Study Design
This was a prospective, multicenter, single-arm study that was conducted complying to Good Clinical Practice and the declaration of Helsinki at 3 investigational sites in China (registered on ClinicalTrials.gov as NCT06570551). Approval was obtained from the ethics committees of each of the participating institutions, and written informed consent was obtained from each subject and/or his/her guardian.
The main study inclusion criteria were: (1) age 2 to 17 years; (2) clinical diagnosis of diabetes; (3) willingness to wear the device continuously for 15 days and conduct blood collection for blood glucose testing in accordance with the study requirements; (4) body weight ≥10.0 kg. The main exclusion criteria were: (1) need to undergo magnetic resonance imaging (MRI) during the study; (2) diffuse subcutaneous nodules; (3) acute complications of diabetes; (4) abnormal coagulation function (activated partial thromboplastin time and/or prothrombin time >1.5× the upper limit of normal or less than the upper limit of normal).
Participants wore 2 sensors, one on each side of the abdomen, and the device performance was assessed by comparison with a point of care, enzymatic-amperometric glucoseoxidase-based, calibratable laboratory reference device (Biosen C-Line glucose and lactate measuring instrument, EKF-diagnostic GmbH). For data consistency, the sensor placed on the right side of the abdomen with the poorer overall performance was chosen for data reporting. Subjects were randomly allocated to assessment of device performance on days 1 and 2 (Group A), days 7 to 9 (Group B) or days 15 and 16 (Group C). During these intervention days, the subjects were provided with a meal to induce glucose changes, but no intentional delay or increase of insulin doses was applicable where applied. On day 16, the devices were removed, and subjects were asked to complete a feedback questionnaire (completed by the subject’s guardian in case of children aged ≤6 years).
Statistical Analysis
In accordance with the latest requirements from the Center for Medical Device Evaluation of the National Medical Products Administration, the study planned to enroll a total of 78 subjects, with no less than 60 subjects aged 6 to 17 years and at least 15 subjects aged 2 to 5 years. The full analysis set (FAS) included all cases that expressed the intention to accept CGM and sign the informed consent form, had used the investigational device and had post-CGM evaluation data. The per-protocol set (PPS) was a subset of the FAS and consisted of subjects having worn sensors as specified in the study protocol, with available observational data on the primary endpoints and no major protocol violation. The safety set comprised all subjects who applied the investigational device at least once.
The primary endpoints were the proportion of CGM values that fell within ±20% or ±20 mg/dL of the reference values (agreement rate [AR]); the proportion of measurement points falling within zones A+B according to the consensus error grid (C-EGA) 12 and mean absolute relative difference (MARD). A post-hoc analysis evaluated the proportion of measurement points falling within zones A+B according to the Diabetes Technology Society error grid (DTS EG). 13 Secondary endpoints included: (1) true notification rates of hypoglycemia and hyperglycemia; (2) sensor stability; (3) sensor repeatability; (4) sensor life and (5) product usability. Statistical analyses were performed using SAS 9.4 (SAS).
Results
Participant flow is summarized in Figure 1. A total of 78 subjects were enrolled, and ultimately 76 subjects completed the clinical trial. The FAS consisted of 76 participants, of whom 16 were aged 2 to 5 years, and 60 were aged 6 to 17 years. Most participants had type 1 diabetes (86.8%). Baseline data are summarized in Table 1.
Study flowchart.
Baseline Data (Full Analysis Set).
Primary Endpoints
Per-protocol analysis demonstrated strong sensor performance with an overall 20/20% AR of 95.96% (95% confidence interval: 95.69%-96.23%). Based on 1879 data pairs, the overall MARD was 8.89%, with stable MARD results across a 15-day wear time (days 1-2, 8.65%; days 7-9, 8.70%; days 15-16, 9.30%) (Figure 2). The device demonstrated single-digit MARD/MAD across multiple different glucose levels (Table 2). The C-EGA analysis and a post-hoc DTS EG analysis both showed that 100.0% of data pairs fell in clinically acceptable zones A+B (Table 3 and Figure 3).
MARD stability over wear life. The horizontal dashed line indicates the 10% MARD threshold for eligibility for replacement of fingerstick glucose tests (nonadjunctive use).
Sensor Accuracy Across Glucose Ranges: Per-Protocol Set.
Error-Grid Data: Per-Protocol Set.
Error grid analyses for subjects with type 1 diabetes (per-protocol sets). (a) Consensus error grid. (b) Diabetes Technology Society error grid.
Secondary Endpoints
The sensor demonstrated reliable true notification rates for hypoglycemia and hyperglycemia. Of 146 alarms issued for hypoglycemia (80 mg/dL, 4.4 mmol/L), 144 were preceded or followed within 15 minutes by a reference glucose measurement above the threshold level. Of 248 alarms issued for hyperglycemia (200 mg/dL, 11.1 mmol/L), 245 met the criteria for a true alarm. Thus, the true alarm rates for hypoglycemia and hyperglycemia were 98.6% and 98.8%, respectively.
For the right abdomen, the 20/20% AR results with the reference value were 95.83% (95% CI: 95.47%-96.20%) at days 1 and 2, 97.34% (95% CI: 97.00%-97.67%) at days 7 to 9 and 94.76% (95% CI: 94.14%-95.38%) at days 15 and 16. Post-hoc analysis showed that mean sensor wear life was 14.41 days; 92.0% of right sensors in the PPS had a wear life of 15 days.
Regarding sensor insertion, 86.7% of the subjects reported no pain, 13.3% reported moderate pain and none reported severe pain (PPS). The insertion process was considered very convenient for 97.3% of participants, while it was considered generally acceptable for 2.7% of participants (PPS). Similarly, the assembly convenience of the glucose signal transmitter and glucose sensor was rated as very convenient in 97.3% of participants, and generally acceptable in 2.7% of participants. When evaluating the overall comfort of use during the 16-day study period, the device was reported as comfortable for 93.3% of participants, and acceptable in 6.7%.
Safety
Of the 78 subjects in the safety set, 51 experienced 82 adverse events, of which 4 subjects received corrective treatment. Of note, there were no device-related adverse events or skin-related adverse events reported; most adverse events were attributable to expected events (Table 4). There was 1 serious adverse event of hypertension (unrelated to the study device).
Adverse-Event Data (Safety Set; N = 78).
Discussion
The glucose sensor that was evaluated in this pediatric study is a third-generation sensor that uses modified glucose dehydrogenase coupled with FAD as the electron acceptor. 6 Based on glucose measurements at days 1 and 2, 8 and 9 and 15 and 16 of the study, the overall MARD was 8.89%, which is consistent with the overall MARD of 8.71% observed with the same system in a multicenter study of adults with diabetes. 7 A sub-9% MARD for the device under study compares favorably with MARDs that have been reported for other contemporary CGM devices in pediatric populations. Previously reported MARDs are 9.7% for FreeStyle Libre 2 in subjects aged 4 to 17 years 14 ; 10.0% and 8.6% with FreeStyle Libre 3 in subjects aged 4 to 5 years and 6 to 17 years, respectively 15 ; 9.9% and 9.6% for Dexcom G6 in subjects aged 2 to 5 years and 6 to 17 years, respectively 16 ; 9.0% for abdomen-placed Dexcom G7 in subjects aged 7 to 17 years 17 and 11.6% for the Guardian 4 sensor in subjects aged 7 to 17 years. 18 The MARD that was obtained in the current study was complemented by the findings that 96.0% of CGM glucose readings were within 20/20% AR of the reference value, and that 100% of data pairs in error grid analyses fell in clinically acceptable zones A+B in both analyzed error grids.
Moreover, the MARD showed good stability over time, reflecting the reliability of the device over the 15-day wear life. This contrasts with a drop-off in accuracy toward the end of sensor life that has been seen with older CGM technology; thus, in a study of FreeStyle Libre 2 in children aged 4 to 17 years, mean MARD was 8.0% at days 7 and 8, 9.7% at days 9 to 12 and 10.2% at days 13 and 14. 14 In the present study, 92.0% of right sensors in the PPS had a wear life of 15 days. This compares favorably with the durability in children that has been reported for other devices, such as 75.7% sensor survival at 10 days for Dexcom G6 16 and 78.1% sensor survival at 14 days for FreeStyle Libre 2. 14
Prespecified secondary analyses also demonstrated good performance in terms of accuracy at different glucose concentrations and high rates of true alerts for hyperglycemia and hypoglycemia.
There were no device-related adverse events or skin-related adverse reported during the study, consistent with results obtained for the sensor in the study of adults with diabetes. 7 One serious adverse event in the present study, diagnosed as hypertension with a severity of Grade 1 (CTCAE, v5.0), was deemed to be unrelated to the device. Another potential safety consideration is the reduced need for sensor changes over the course of a year compared to sensors with targeted wear life of 10 days or 14 days and therefore a reduction in the length of time spent with sensors that are running in, since the run-in period is usually about a day.
Certain strengths and limitations of this study should be mentioned. Our study included children as young as 2 years to provide performance data on a broad range of ages. In addition, we adopted a conservative approach for data analysis, focusing on the performance characteristics of the poorer performing sensor, that is, the sensor applied to the right abdomen. No experimental glucose manipulations were performed during the study; glucose variations were those occurring naturally because of meals, drinks or routine medication. Ethical requirements prevented the participation of each subject at each performance evaluation visit. Therefore, the potential for bias was minimized by random allocation of subjects across the 3 performance evaluation visits.
Conclusion
Pivotal clinical trials of an innovative CGM system have enrolled more than 500 adult and pediatric participants with diabetes. In the current performance study in pediatric participants aged between 2 and 17 years, the novel CGM system demonstrated high overall accuracy and a single-digit MARD that is comparable to those of established systems across its 15-day wear time. This performance translates into high detection rates for both hypoglycemia and hyperglycemia. In addition, an outstanding average sensor wear life that is close to the intended wear time and no device-related adverse events – as already seen in an adult population – can be reported.
Footnotes
Acknowledgements
Dexcom is a registered trademark of Dexcom, Inc. in the United States and may be in other countries. FreeStyle, Libre and related brand marks are marks of Abbott. Guardian is a trademark of Medtronic. iCan is a registered trademark of Sinocare, Inc in the EU. Editorial support for the preparation of this article was provided by George Clarkson, PhD, funded by Sinocare Inc. According to the GCP requirements of the National Medical Products Administration (NMPA), clinical validation for premarketing products is subject to NMPA supervision and must be conducted as a clinical trial within a hospital.
Abbreviations
C-EGA, consensus error grid analysis; CGM, continuous glucose monitoring; DTS EG, Diabetes Technology Society error grid; MAD, mean absolute difference; MARD, mean absolute relative difference.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by Sinocare, Inc.
Declaration of Conflicting Interests
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Fengyun Wang is an employee of Children’s Hospital of Soochow University. Her position is a physician of the Pediatric Endocrinology Department. Frank Flacke is an employee and the owner of Flacke Consulting GmbH providing strategy advice, medical affairs and clinical affairs services and is or was within the past 3 years a contractor of Sinocare, BOYDSense, embecta, Sanofi and Glucoset. Shuxia Ding is an employee of the Women and Children’s Hospital of Ningbo University, her position is the Director of the Pediatric Endocrinology, Genetics and Metabolism Department. Wenxian Ouyang is an employee of Hunan Children’s Hospital, his position is the Director of the Hepatology and Endocrinology Department. Haiying Wu is an employee of Children’s Hospital of Soochow University, her position is the Director of the Pediatric Endocrinology Department.