Defining continuous glucose monitor time in range in a large community-based cohort without

Context: Continuous glucose monitor (CGM) companies are beginning to market these sensors to populations


Introduction
New continuous glucose monitor (CGM) sensors are hitting the wearables market such as Abbott's Libre Lingo and Dexcom's Stelo, that can be used by individuals without diabetes.(1,2) To date, there have been no large studies describing normative CGM values in this population.Previous studies reporting normative CGM metrics in individuals without diabetes have been smaller (<100 adults), only included non-obese individuals, and asked participants to perform daily manual calibration using finger-stick blood glucose measurements.(3) Many of the large CGM brands have made their CGM sensors factory-calibrated, removing the need for manual calibration, and making them more user-friendly to an individual who is not familiar with self-monitoring blood glucose.
Despite a lack of research showing that using CGM sensors improves health outcomes in individuals without diabetes, there seems to be a growing interest in CGM use in this population.(4) We anticipate that there will be an increase in the number of CGM reports from patients without diabetes that healthcare providers are asked to interpret, for which there are currently no guidelines for a healthy patient population.So the ability for healthcare providers to compare patient reports to an established range of physiological CGM metrics from individuals without diabetes will be incredibly useful.
For individuals with diabetes, CGM metrics are also increasingly relevant to both clinical practice and research methodologies, including the well-established glucose ranges, time in range (70-180mg/dL), (5) and more recently, time in tight range (70-140mg/dL).(5-7)Therefore, it is important to understand the physiological time spent in specific ranges of CGM glucose for individuals without diabetes, which could serve as targets for individuals with diabetes or prediabetes.

Study Population
Our study included grandchildren of the original Framingham Heart Study (FHS) cohort who were enrolled in the Third Generation (n=4095), New Offspring Spouses (n=103), or Omni 2 (n=410) cohorts in 2002.(8) Our study included participants from these cohorts who attended the fourth examination from September 2022 to December 31, 2023 (n=1699) and wore a Dexcom G6 Pro CGM sensor for at least 7 full days on their upper arm or abdomen, with an available HbA1c and fasting glucose measurement (n=1175, Supplemental Figure 1).(9) Diabetes was defined using self-report (current diagnosis or history of diabetes), HbA1c ≥6.5% (≥48 mmol/mol), venous fasting glucose ≥126mg/dL, or taking glucose-lowering medications at the time of their study visit (n=152), prediabetes was defined if participants had either HbA1c 5.7-6.4% (39-47mmol/mol) and/or venous fasting glucose 100-125 mg/dL (n=463), and remaining participants did not meet criteria for prediabetes or diabetes, and will be described as having "normoglycemia" for this report (n=560).All participants provided written informed consent, and the institutional review board at Boston University Medical Center approved the study protocols.

CGM measurement
When participants came to the FHS Research Center for their study visit, a research technician applied a single Dexcom G6 Pro CGM sensor (Dexcom, Inc., San Diego, CA) to their upper arm or abdomen.
Participants were asked to wear the sensor for up to 10 days and mail the CGM sensor and transmitter back to the study site after the wear period.

CGM data cleaning procedures
We removed each participant's first and last partial days of data, including the first 12 hours of wear.Days with less than 70% complete data (<200 of the possible 288 values) were also removed (totaling 9 days from 7 participants combined).Subsequently, participants who did not wear the CGM sensor for ≥7 full days (midnight-to-midnight) were excluded from the dataset.The maximum and minimum possible recorded values were 400 and 40 mg/dL, with "HIGH" or "LOW" messages recorded for data falling outside that range.We imputed "HIGH" values as 401mg/dL and "LOW" values as 39mg/dL.We did not clean our data further because CGM data reports that patients may bring to their healthcare providers are not cleaned.

HbA1c, fasting venous plasma glucose, and covariates
Venous whole blood was collected in 15% EDTA tubes (Monoject Blood) from participants who fasted 10h overnight.An aliquot was taken for HbA1c measurement (Tina-quant Hemoglobin A1c, Roche Diagnostics, Germany).Blood glucose was analyzed using a Roche cobas c311 (GLUC3 Glucose HK, Roche Diagnostics, Germany), almost immediately after isolating plasma by centrifugation at 2500g at 4°C.

Statistical Analysis
Mean and standard deviation (SD), median and interquartile range (IQR), or frequency of participants with certain demographic factors and CGM metrics were reported, stratified by glycemic status, age, and obesity status.CGM metrics were also presented in histograms, by glycemic status.
All CGM metrics increased across glycemic status groups, further evidenced by increasing proportions of time spent in higher glucose ranges (Table 1).Mean glucose had a similar distribution and was only 8.6mg/dL higher in prediabetes compared to normoglycemia (123.1 vs. 114.5mg/dL);a majority of participants from both groups had a mean glucose between 100-140mg/dL (Figure 1).In normoglycemic participants, almost 98% of time was spent in the 70-180mg/dL range, but <87% of time was in 70-140mg/dL, with 11% time 140-180mg/dL and 1.3% time (>15 minutes/day) >180mg/dL (Table 1).Hence, normoglycemic participants (without elevated fasting venous glucose or HbA1c) spent ~3 hours/day (12.3% time) with CGM glucose >140mg/dL.Most normoglycemic participants had a maximum glucose level >180mg/dL, but rarely >250mg/dL (Figure 2).Normoglycemic individuals ≥60 years spend ~1% less time in 70-180mg/dL range compared to those <60 years, with the additional time split between being spent above and below that range (Table 2).But there was a larger difference in CGM time in tight range (70-140mg/dL) between non-obese individuals <60 years (89.0%time) and non-obese individuals ≥60 (85.8%).Having obesity at either age was also associated with lower time spent in tight range, with older, obese adults having the lowest time spent in tight range, on average (83.7%),compared with other normoglycemics.Unsurprisingly, participants with prediabetes and diabetes spend substantially less time in tight range (76.9 and 42.4%, respectively, Table 1).

Discussion
Our results demonstrate that the physiologic range of CGM glucose in a cohort of middle-aged, non-Hispanic white individuals without diabetes or prediabetes is 70-180mg/dL (approximately 97-98% of their time).An Further, investigators directed participants to manually calibrate the (factory-calibrated) Dexcom G6 sensors, requiring participants to have access to their glucose levels (unblinded).In our study, participants wore the same sensors in a blinded mode without manual calibration.Although participants in both studies likely modified their behavior, as is true in any short-term observational study, blinding minimizes that behavior.
Clinicians should expect that many patients without elevated HbA1c or fasting glucose could have CGM reports showing glucose values >180mg/dL, but that their reports may differ by sensor and by the wearing protocol (e.g.manual calibration).It is also possible that individuals experiencing substantial time >180mg/dL should not be viewed as "normal" physiologically; and certain glycemic patterns may be an early warning sign of metabolic and glycemic dysregulation that may be present in other clinical tests, such as an oral glucose tolerance test.(14) Our study design, integrating CGM sensor wear into an ongoing cohort study like The Framingham Heart Study, sets us up for answering important research questions, such as whether CGM metrics can predict the development of diabetes, as has been suggested in the A Estrada Glycation and Inflammation Study (AEGIS) in Spain of 499 individuals without diabetes.(14) We will also be able to elucidate the determinants of CGM metrics/glycemic patterns among individuals with and without diabetes.Several studies have demonstrated that age, sex, body composition, diet, sleep, genetics and gut microbiome are associated with CGM glucose metrics in individuals without diabetes.(12,(15)(16)(17)(18) The personalized responses to dietary composition trials (PREDICT), conducted in generally healthy individuals in the UK and US, showed that genetics and the meal time of day explained the most inter-individual variation in CGM postprandial glucose responses to standardized meals, compared to other predictors including age, sex, or body composition.(12) In Israel, the 10K study has reported that daily carbohydrate intake and visceral adipose tissue are also determinants of CGM measures of glycemic variability.(15) Another interesting finding they reported was that BMI appeared to be more strongly associated with the overall glycemic burden during sleep (measured by mean CGM glucose), but not the glycemic burden during waking hours.We hope to contribute to the understanding of these determinants of CGM patterns in subsequent analyses.
The strength of our CGM study lies in the nature of our participant pool, which were largely individuals without diabetes (n>1000).As discussed previously, prior studies reporting on the physiological range of CGM levels across non-diabetic populations were much smaller, healthier, and included manual CGM sensor calibration by participants.(3)Because we do not expect most individuals without diabetes to conduct manual sensor calibration while wearing the new CGM sensors that are slated to hit the wearables market this year (Abbott's Libre Lingo and Dexcom's Stelo), (1,2) our study provides the more "real world" design.
Our study was limited by a cross-sectional design, which cannot determine whether time spent in various glucose ranges among normoglycemic participants may be related to a higher risk of diabetes.We are also limited by lack of contemporaneous blood glucose measures, the use of only a single type of CGM sensor (Dexcom G6 Pro), and by the over-representation of non-Hispanic white individuals in our study population, who have been shown to have lower rates of dysglycemia compared to other races and ethnicities in the United States.(19)Future studies will be needed to explore normative CGM metrics in more diverse study populations for increased generalizability.There are many important questions that can be answered in subsequent analyses, including assessing the influence of sex and menopausal status on glycemic patterns, such as potential hypoglycemic events.(20)Another important future direction is standardizing CGM data cleaning methods and identifying biologically implausible glucose values or patterns.Finally, exploring whether high glucose levels can be avoided among normoglycemic individuals with behavior modification will also be important.
In conclusion, time in "tight" range (70-140mg/dL) has been growing in interest as a potential target for clinicians to help their patients with diabetes achieve better glucose control.(5-7)However, our study provides evidence that individuals with normoglycemia spend ~3 hours/day above that tight range.The physiologic glucose ranges we report from a sample of >1000 individuals without diabetes may also be important for clinicians to reference as CGM sensors become more widely accessible to individuals without diabetes.

additional 1 - 1 .
5% time (>15min/day) is spent >180mg/dL, but most do not reach 250mg/dL.Comparatively, individuals with prediabetes spend roughly 30 minutes more time per day in the >180mg/dL glucose range and less than a quarter reach 250mg/dL.Previous studies reported that normoglycemic adults spent between 93-97% of time, on average, in the 70-140mg/dL range.(3,10)However, our results demonstrate that only 87% of time is spent in this "tight range," varying by age and obesity status, and most of the remaining time (~3h/day) CGM glucose is >140mg/dL.Participants in our study spent almost 97.8% of time in the 70-180mg/dL range instead.These data are relevant to the current debate about the potential value of time in tight range (70-140 mg/dL) as a glycemic target.It is important for clinicians treating patients with diabetes to be aware that the physiologic CGM range among individuals with normoglycemia may include substantial time above 140mg/dL and even acute periods above 180mg/dL.Our observed proportion of time spent >180mg/dL was not shown in some smaller studies of normoglycemic adults (n<100) wearing Dexcom G6 sensors byShah et al.(3)  or Abbott's Freestyle Libre Pro factory-calibrated sensors.(10)However, other studies have observed participants without diabetes achieving glucose levels >180mg/dL, using the Freestyle Libre and iPro2 (Medtronic).(11)(12)(13)Direct comparisons are Downloaded from https://academic.oup.com/jcem/advance-article/doi/10.1210/clinem/dgae626/7754867 by guest on 14 September 2024difficult because these studies did not always report the average times spent in those ranges.Important differences in study design between our study and Shah et al. may explain differing results despite using the same CGM sensor.(3)Shah's study excluded individuals with obesity, and participants were relatively younger.

Figure 1 .
Figure 1.Ascending mean CGM glucose (green) for each participant within each glycemic status group (n=1175).Maximum CGM glucose (red) and minimum CGM glucose (blue) also displayed.

Figure 2 .
Figure 2. Histograms of CGM mean glucose, maximum glucose, and time spent in glucose ranges as a proportion of populations with normoglycemia, prediabetes, and diabetes (n=1175).Histogram colors are partially transparent to show overlap.
Figure 2 160x190 mm (DPI) A C C E P T E D M A N U S C R I P T R I P T