Relationship Between Blood Glucose Variability in Ambulatory Glucose Prole and Standardized Continuous Glucose Monitoring Metrics; A Pilot Study

Background: Treatment indexes using continuous glucose monitoring (CGM) have become standardized internationally, and the use of ambulatory glucose prole (AGP) is currently recommended. However, the relationship between AGP indexes and standardized CGM metrics has not been investigated. Using ash glucose monitoring (FGM), this retrospective study served to evaluate the association of the inter-quartile range (IQR) of AGP with standardized CGM metrics. Methods: The study subjects were 30 patients with type 2 diabetes mellitus (T2DM) and 23 non-diabetic patients (control group). We evaluated average IQR (AIQR) and standardized CGM metrics. The primary endpoint was the relationship between AIQR and Time in range (TIR) in a 24-hour period. Results: In the T2DM group, the AIQR was notably high and correlated negatively with TIR, and positively with Time above range, average glucose level, SD, CV, and MODD. For the T2DM group, the AIQR was notably lower in patients who achieved TIR>70%, compared to those who did not. The AIQR cutoff value, as determined by ROC analysis, was 28.3 mg/dL for those who achieved TIR>70%. No association was detected between the presence of hypoglycemia and AIQR. Conclusions: Our study is the rst to provide the AIQR cutoff value for achieving the TIR target value. The range of blood glucose variability in AGP was associated with indexes of intra- and interday variations and hyperglycemia. Our results provide new perspectives in the yet-to-be established methods for evaluation of AGP in practical clinical settings. Sex, χ 2 test. Age, Welch test. BMI: Body Mass SBP: Systolic blood pressure, DBP: Diastolic blood pressure, eGFR: estimated glomerular ltration rate, SU: sulfonylureas, DPP-4 inhibitor: dipeptidyl peptidase-4 inhibitor, SGLT-2 inhibitor: sodium-glucose transporter-2 inhibitor, GLP-1 receptor: Glucagon-like peptide-1 receptor.


Background
Continuous glucose monitoring (CGM) and ash glucose monitoring (FGM) are becoming standard devices used for management of glucose levels in diabetic patients. A consensus report on CGM metrics was published for the rst time last year, and the time-in-range (TIR) and blood glucose levels of ≥ 180 and < 70 mg/dl have become international standards 1 as diabetes treatment indexes. A study using DCCT research data indicated that TIR correlates with retinopathy and nephropathy 2 , while the percentage of low blood glucose under 70 is associated with the risk of serious hypoglycemia 3 .
On the other hand, the use of the ambulatory glucose pro le (AGP) as a standard CGM report has been recommended 1 . The AGP proposed by Mazze et al. 4 is an analytical method that allows easy and visual understanding of various factors, such as the time during the day when there is a high probability of development of hypoglycemia or hyperglycemia, and times during the day with large swings in blood glucose levels; the AGP can also indicate movements and trends in glucose variability (Supplemental Fig. 1). Although indexes, such as the inter-quartile range (IQR), inter-decile range (IDR) and median are the leading AGP indicators, their relationships with the existing standardized CGM metrics have not been identi ed; furthermore, the target values and cutoff values in T2DM have not been proposed.
In this pilot study, we used FGM and evaluated the inter-quartile range (IQR) of AGP and its association with standardized CGM metrics. We also examined the target values and cutoff values in T2DM.

Subjects
We conducted a retrospective study that included 30 patients with type 2 diabetes mellitus (T2DM) and 23 non-diabetic patients who attended the outpatient clinic of the University of Occupational Medicine Hospital and University of Occupational Medicine Wakamatsu Hospital, between September 2018 to January 2019. At the time of the study, the T2DM patients were being treated for T2DM and monitored with the ash glucose monitoring system (FGMS® System FreeStyle Libre Pro System, Abbott Diabetes Care, Inc.) for at least eight days. The following inclusion criteria were applied in the selection of the T2DM group: 1) Patients aged between 30-80 years at the time of consent to the study; 2) Outpatients with T2DM; 3) Patients who had not changed (added, switched, or discontinued) their glucose-lowering medications or changed the doses of these medications within four weeks before the start date of monitoring with the FreeStyle Libre Pro sensor. We also applied the following exclusion criteria: 1) Type 1 or secondary diabetes mellitus; 2) Patients with severe infections or serious trauma, and pre-and postoperative patients; 3) Patients on dialysis; 4) Patients with severe liver dysfunction (AST 100 IU/l or greater); 5) Patients with moderate or serious heart failure (NYHA/New York Heart Association Classi cation III or higher stage); 6) Pregnant, lactating, or planning to become pregnant patients.
Patients and hospital staff con rmed to be non-diabetic and had no glucose intolerance were recruited as the non-diabetic control group.
The study protocol and opt-out method of informed consent were approved by the ethics committee of the University of Occupational and Environmental Health (Trial registration: H27-186, Registered 25 Dec 2015).
We used the following baseline de nition of diabetic microangiopathy: The earliest clinical evidence of nephropathy is the appearance of low but abnormal levels (≥ 30 mg/day or 20 µg/min) of albumin in the urine, referred to as microalbuminuria 5 . The urinary albumin excretion rate is presented as the albumin-tocreatinine ratio (mg/g creatinine).
In this study, diabetic nephropathy was de ned as an albumin-to-creatinine ratio of ≥ 30 mg/g creatinine.
Diabetic retinopathy was de ned as simple or more severe retinopathy based on funduscopic examination by ophthalmologists. Diabetic neuropathy was diagnosed by the presence of two or more clinical symptoms (e.g., bilateral spontaneous pain, hypoesthesia, paresthesia of the legs), absence of ankle tendon re exes, and decreased vibration sensations using a C128 tuning fork.

Page 4/24
The average glucose level (AG), median, standard deviation (SD), coe cient of variation (CV), percent time at blood glucose level of 70-180 mg/dl (TIR; time-in-range), percent time with blood glucose level at > 180 mg/dl (TAR; time-above-range), percent time with blood glucose level at < 70 mg/dl (TBR; timebelow-range), maximum, minimum, glucose management indicator (GMI, new terms used for estimating HbA1c from CGM) 6 , mean of daily difference in blood glucose (MODD), low blood glucose index (LBGI), high blood glucose index (HBGI), and AGP 25-75th percentile width (IQR; interquartile range) were obtained from the data recorded with the FGM 7,8 . Hypoglycemia was de ned as glucose value < 70 mg/dl, as measured by FGM. The FGM recorded and stored blood glucose levels over seven consecutive days. Data collected over the rst day were discarded to avoid bias due to the insertion and removal of FGM, or insu cient stability of the monitoring system. We recorded the daily average value, and listed the average value of seven days.

Statistical analysis
Data are shown as mean ± standard deviation. The Shapiro-Wilk normality test was used to check the distribution of data. Differences between the mean values of various parameters in two groups were tested for statistical signi cance using the Student's t-test after con rming equal variance by the F test and Welch's t-test, or the presence of normal distribution. The Mann-Whitney U test was used for data with skewed distribution, and Spearman's correlation analysis was used for relationships between two variables. We also analyzed the cutoff values using the ROC curve. The required sample size in ROC analysis was calculated as 25 patients in total using an area under the curve (AUC) of 0.85, test power of 0.80, signi cance level of 5%, and a 5:1 ratio for the group that achieved TIR > 70% versus the group that did not achieve TIR > 70%. A p value < 0.05 denoted the presence of statistical signi cance. All statistical procedures were performed using the SPSS Statistical software version 25.0 (SPSS Inc., Chicago, IL).

Results
Clinical characteristics of study participants 43.0). For the diabetic group, 70% of the patients were using DPP-4 inhibitors and 50% were on biguanides.

Comparison of the diabetic and control groups
The FGM data of the diabetic group and control group are shown in Table 2. The average IQR (AIQR) was 17.3 ± 4.3 mg/dl for the control group, and signi cantly higher for the diabetic group (30.1 ± 11.7 mg/dl, p < 0.001). The TIR was signi cantly lower (p < 0.001), while TAR was signi cantly higher (p < 0.001) in the diabetic group. There were no notable differences between the two groups with respect to the minimum, TBR, and LBGI. Correlation between IQR and CGM index in the diabetic group Table 3 summarizes the results of analysis of the correlation between AIQR and CGM metrics in the diabetic group. The AIQR correlated negatively (r=-0.840, p < 0.001, Fig. 1) with TIR, and positively with maximum, SD, CV, MODD, HGBI, and TAR (p < 0.001, each). However, there was no correlation between AIQR and the hypoglycemia indexes of minimum, LBGI, and TBR.   Fig. 2).
Next, we used an AIQR of 28.3 mg/dL as the cutoff value and compared the data with TIR. The TIR was 94.1 ± 4.3% in the AIQR < 28.3 mg/dL group and 76.4 ± 12.7% (p < 0.001) in the AIQR ≥ 28.3 mg/dL group.
Comparison of the hypoglycemia and non-hypoglycemia subgroups Table 5 shows the results of comparison of CGM metrics in the presence of hypoglycemia. Hypoglycemia was noted in 20 of the 30 patients of the diabetic group, and the hypoglycemia indexes of TBR and LGBI were signi cantly higher in the hypoglycemia group than the non-hypoglycemia group (p < 0.001, each). However, there was no notable difference in AIQR in the presence of hypoglycemia. AG, GMI, and TAR were all signi cantly higher in the non-hypoglycemia group (p = 0.005, p = 0.005, p = 0.037).

Discussion
The major ndings of the present study using the FGM were the following: 1) AIQR correlated negatively with TIR, 2) the longer the time blood glucose level is between 70 and 180 mg/dL, the smaller the variation range in AGP, 3) AIQR correlated positively with SD, CV, and MODD, and 4) Lack of relationship between AIQR and hypoglycemia.
Our study also established the AIQR cutoff values for achieving the TIR target values. AGP is an excellent tool that clearly and visually presents the trends in blood glucose variability for the individual patient 9 ; our study showed that AIQR is a valuable index for use in AGP.
Our data showed that AIQR correlates negatively with TIR, and that the longer the time blood glucose is maintained within the range of 70 to 180 mg/dL, the smaller the variation range in AGP. TIR refers to the percent time during the 24-hour period when blood glucose is within the range of 70-180 mg/dL. TIR provides a more complete picture of blood glucose than HbA1c, and hence makes it possible to offer personalized treatment options 10 . Similar to HbA1c, TIR time is reported to correlate strongly with the risk of retinopathy and/or onset of microalbuminuria 2 . Other studies found that TIR is independent of HbA1c and that it is associated with cardiovascular autonomic neuropathy 11 . Furthermore, TIR correlates strongly with HbA1c 12 , and was recently recommended by the American Diabetes Association (ADA) for use along with HbA1c for targeted blood glucose control.
IQR is in the 25-75th percentile range on the AGP, and is an excellent visual indicator of glucose variability 13,14 ; although there is no reported relationship between IQR and standardized CGM metrics. To our knowledge, our study is the rst to de ne the AIQR-TIR relationship. Importantly, setting the cutoff values for AIQR makes it easier to evaluate treatment goals especially achieving the TIR targets. Following ADA's target of TIR > 70% in T2DM patients, we established the AIQR cutoff values of 28.3 mg/dL for TIR > 70% and 22.9 mg/dL for a stricter adherence to TIR > 90%. These indexes are useful markers to use when setting target treatment values using the AGP.
The IQR has also been discussed as a marker of interday variations 15 . Another strong aspect of our study was the strong correlation between AIQR and the SD and CV (which are markers of intraday variations), in addition to the correlation between AIQR and MODD (which is a marker of interday variation). In this regard, Monnier et al. 16 proposed the use of indexes that can evaluate blood glucose variability in addition to chronic hyperglycemia and hypoglycemia, as a strategy for monitoring blood glucose to inhibit progression of cardiovascular disorders. The results of the present study suggest that IQR is useful index for evaluation of intra-and interday variations in blood glucose levels. In practical clinical settings from hereon, a large IQR will require analysis of the factors behind interday variations, as well as necessitate evaluation of the factors responsible for intraday variations, including, for example, postprandial glucose and low blood glucose levels.
Our analysis showed the lack of any association between AIQR and presence of hypoglycemia, and that low blood glucose is associated with AG and LBGI. These results suggest that AIQR is not suitable for use to predict the extent of hypoglycemia; and thus, we advise instead to visually evaluate each case when assessing low blood glucose in AGP. It is noteworthy that a number of investigators found close associations between low blood glucose and certain markers, such as AG and LBGI 17,18 , similar to our ndings. Therefore, AG, CV, LBGI and other CGM metrics can also be used with other indexes to evaluate objectively the risk of hypoglycemia in clinical settings.
There were two limitations in this study. The rst is we did not record glucose density of ≥ 500 mg/dL based on the use of FGM, and accordingly, or study did not include patients with poor blood glucose control. This is an important limitation to mention here because the results for such patients may be different from those obtained in the present study. The second limitation is that this research work was a cross-sectional study conducted at two facilities with only a relatively small number of patients. We need to proactively continue with our research on a larger scale in order to substantiate the results of this study.

Conclusion
This study noted, for the rst time, a strong correlation between TIR and blood glucose variability range that can be understood visually using the AGP. The study also provided the AIQR cutoff values for achieving TIR target values. Furthermore, the blood glucose variability range in AGP was strongly associated with indexes of intra-and interday variations; though there was no association with LBGI. Our results may provide new perspectives in the yet-to-be established methods for evaluating AGP in practical clinical settings. Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. TokutsuSupplFigure2.tiff TokutsuSupplFigure1.tiff