Introduction
This evidence review focuses on evidence for FDA approved implantable therapeutic continuous glucose monitors and whether this evidence is adequate to draw conclusions about health outcomes for the Medicare population. Health outcomes of interest include mortality, patient morbidity as a result of diabetes, harm from treatments, and patient quality of life and function. Additionally, proven intermediate outcomes in the causal pathway of health outcomes are of interest. The current clinical evidence based on trial data for I-CGMs includes one pivotal European trial, two pivotal U.S. clinical trials under one investigational device exemption (IDE), two post-market registry studies, and guideline and consensus statements.
Internal Technology Assessment
Pubmed was searched for clinical studies that included “implantable continuous glucose monitor” or “Eversense” (the only FDA approved implantable CGM as of March 1, 2020). Literature was reviewed for inclusion if it met these criteria, and all abstracts were reviewed. Studies must be published in English language and specific to the human population. Only devices approved by the FDA as of March 1, 2020 were included in this review.
Pivotal clinical studies U.S.
Christiansen M, et.al reported on two studies conducted under IDE #G150165, the PRECISE II and PRECISION studies. Data from both studies are included in the FDA summary of safety and effectiveness data (SSED). PRECISE II was a non-randomized, blinded, prospective single-arm study at 8 U.S. sites evaluating 90 patients with type 1 and type 2 diabetes in 2016.(6) The I-CGM glucose values and all glucose-related alerts were blinded to patients and investigators. Management decisions were based on self-monitored blood glucose. Of the initial 90 patients, 82 completed the 90 day visit with data. The primary endpoint was the Mean Absolute Relative Difference (MARD) between paired Eversense and Yellow Springs Instrument (YSI) reference measurements through 90 days for reference glucose values from 40 to 400 mg/dL.(6) The prospective, unblinded, multicenter PRECISION study evaluated 35 patients, with enrollment starting in July 2017 and subject completion in February 2018.(7) Study outcome measurement was similar to PRECISE II. In PRECISION, the glucose calculation algorithm was updated. Inclusion and exclusion criteria, follow-up schedules, and accountability of patients are included in the respective articles. The demographics of both trials are listed in Table 8 of the 2018 SSED.(13p15) The mean age of PRECISE II was 45 years (SD:16 years, minimum age: 18, maximum age: 77) and 52 years (SD: 16 years, minimum age: 18, maximum age: 75) for PRECISION. Table 9 of the SSED has diabetic history, with PRECISE II having 28.6% type II patients and PRECISION having 32.2% type II patients.(13p17) Types of insulin therapies for PRECISE II included: 14.3% none, 31.4% multiple daily injections, 54.3% continuous insulin infusion pump; PRECISION: 22.2% none, 26.7% multiple daily injections, 47.8% continuous insulin infusion pump. Safety endpoints for both studies are in Table 10 in the SSED and Table 11 reports the adverse events related or possibly related to the device or procedure, with most related to dermatological issues.(13p18) There were no infections in either study. Effectiveness results are tabulated in Tables 13 – 26 of the SSED.(13p20-35) In PRECISE II, the overall reported MARD value against reference glucose values was 8.8% (95% CI: 8.1 – 9.3%), which was lower than the performance goal of 20%. The system identified 93% of hypoglycemic events and 96% of hyperglycemic events per YSI. In PRECISION, the MARD value against reference was 9.6% (95% CI 8.9 – 10.4). Precision measurement results that compare data from two separate sensors from the same subject are available for both studies (Tables 27 and 28) as is alert performance (Tables 29 – 36).(13p36-39)
Pivotal Clinical Studies Outside the US
The Eversense implantable CGM study by Kropff J, et.al, studied 71 patients in a 180-day prospective trial at seven clinical sites in Europe November 2014 to November 2015. Baseline patient characteristics are reported in Table 1 and include 66 type 1 and 5 type 2 patients with diabetes.(7p65) Average age was 41.7 years (SD 12.6 years). The sensor accuracy compared to venous plasma glucose samples using YSI 2300 STAT PLUS glucose and lactate analyzer (YSI, Yellow Springs, OH) over the range of 2.2 – 22 mmol/L or 40 – 400 mg/dL is reported in Table 2.(7p66) The MARD compared to reference venous glucose values (>4.2mmol/L) was 11.1% (95% CI 10.5, 11.7).(7) Table 3 reports the accuracy of the sensor for glycemic ranges and for rates of change.(7p66) In particular, in the range of <75 mg/dL, the MARD was 21.7%, SD 21.5 and 95% CI 20.4, 23.0.(7) No device related serious adverse events occurred. Two cases of incision site infection were reported; one received antibiotic treatment. The authors note, “Results from the questionnaire data indicated high participant acceptance of the system but did not register improved perceived generic quality of life, as assessed per SF-36 questionnaire.”(7p66)
Post-Market Registry Studies
A study by Sanchez P et.al included deidentified data from the first 205 patients to reach a 90-day wear period in the Eversense Data Management System (DMS).(8)This data was analyzed for mean sensor glucose (SG), median interquartile range, coefficient of variation, glucose measurement index, and percent and time across glucose ranges for various time periods. Data was collected from August 1, 2018 to May 11, 2019. Patients were identified as either type 1 (129), type 2 (18), or unreported (58). Only limited demographic data was obtained. SG data and “patient-entered inputs such as meals, exercise, and insulin are automatically uploaded” into the DMS through smartphone connection. Results are listed in Table 1.(8p3) The MARD compared to home blood glucose meters was 11.2% (SD 11.3%).(8) Patients chose to have the sensor reinserted 78.5% of the time after the completion of the first sensor wear.(8) The transmitter wear time median was 83.6%. No serious adverse events were reported with 2% of patients reporting mild infection at the insertion site (3 treated with oral antibiotics), 1.5% reporting hypoglycemia that was self-treated, and 2% reporting failure to remove the sensor on first attempt. The authors state, “In conclusion, the 90-day implanted Eversense I-CGM system appears to be a valuable and safe tool for management of diabetes with patients reinserting and using the device the majority of the time.”(8p4)
Deiss D et.al reports on the post-market clinical follow-up (PMCF) registry in 534 centers in Europe and South Africa from June 2016 to August 2018.(9) The initial device approval in Europe occurred in May 2016 for the 90-day sensor and September 2017 for the 180-day sensor. All patients with implanted sensors were enrolled prospectively, a total of 3,023 patients.(9) Patient inclusion/exclusion criteria were consistent with the device label. A total of 969 patients had used the system for at least 6 months and 173 patients had used the system for a minimum of 1 year. Adverse events were defined as, “any untoward medical occurrence, unintended disease or injury, or untoward clinical signs (including abnormal laboratory findings) in patients, users, or other persons, whether or not related to the medical device.(9) A serious adverse event (SAE) was defined as an AE that led to death, led to serious deterioration in the health of the patient requiring medical assistance including emergency medical services and/or hospitalization, or led to fetal distress, fetal death, or a congenital abnormality or birth defect. AEs were recorded at each visit and also between visits by patient report. Events were adjudicated by a medical monitor. The primary safety endpoint was the rate of SAEs through 4 sensor cycles. No SAEs were reported. The discontinuation rate was 11% (337 patients), with the most common reason listed as unknown and next most common listed as lack of medical reimbursement. Sensor location site infection was reported at a rate of 0.96% and inability to remove the sensor upon first attempt at 0.76%.(9) Sensor life was 91% for 90-day sensors and 75% for 180-day sensors. The authors conclude, “The PMCF registry provides real-world evidence that the Eversense I-CGM system is safe over multiple cycles of use.”(9p2)
Evidence-Based Guidelines
According to information from 2019 Chamberlain JJ et.al, the American Diabetes Association (ADA) added this section on diabetes technology to the Standards of Medical Care in Diabetes. The Professional Practice Committee of the ADA who developed this guideline is comprised of an expert committee whose appointment is based on excellence in research or clinical care. For these recommendations, members systematically search the literature. The recommendations are rated as A, B, C, or E.(10) Those with an A rating are based on large, well-designed clinical trials or high-quality meta-analyses. Recommendations with a B rating are based on well-conducted cohort studies, while uncontrolled studies are given a C rating. An E rating is assigned when there is no evidence from clinical trials, clinical trials might be impractical, or evidence is conflicting. “The ADA funds development of the standards from its general revenues, with no industry involvement or support.”(10)
Recommendations for self-monitoring of blood glucose, continuous glucose monitors, and automated insulin delivery are included in this document. Several of the recommendations are listed below:
Continuous Glucose Monitors
“Sensor-augmented pump therapy can be considered for children, adolescents, and adults to improve glycemic control without increasing overall or severe hypoglycemia. Benefits correlate with ongoing consistent use of the device. (Grade A recommendation)”(10p74)
Real-Time Continuous Glucose Monitor Use in Adults
“When used properly, real-time CGM in conjunction with intensive insulin regimens is a useful tool to lower HbA1C levels in adults with type 1 diabetes who are not meeting glycemic targets. (Grade A recommendation).”(10p76)
“Real-time CGM may be a useful tool in those with hypoglycemia unawareness or frequent hypoglycemic episodes. (Grade B recommendation)”
Professional Society Recommendations/Consensus Statements/Other Expert Opinion
In 2016, the American Association of Clinical Endocrinologists and the American College of Endocrinology convened a conference to review available CGM data. They felt the evidence supported the use of CGM in patients with type 1 diabetes and the benefits are likely to apply, whenever, intensive insulin therapy is used, whether the patients has type 1 or type 2. Fonseca V et.al reported findings from this conference, and felt that patients older than 65 years with comorbidities and/or risk for severe hypoglycemia may have significant benefits, though at that time, additional studies were needed. “(11) The primary purpose of CGM is to identify glucose patterns, hypoglycemia, and hyperglycemia.”(11) The goal of management is to maximize time in the desired glucose range. The authors concluded, CGM improves glycemic control, reduces hypoglycemia, and may reduce overall costs of diabetes management. Expanding CGM coverage and utilization is likely to improve the health outcomes of people with diabetes.”
The Advanced Technologies and Treatments for Diabetes Congress convened an international panel of experts in CGM technologies to make recommendations on the use on CGMs. Danne T et.al indicated the panel made recommendations on these topics: limitations of HbA1c; use of glucose monitoring methodologies to guide management and assess outcomes in different patient populations; minimum requirements for CGM performance; definition and assessment of hypoglycemia in clinical studies; assessment of glycemic variability; time in range; and visualization, analysis, and documentation of key CGM metrics.(12)