Intermittently Scanned Continuous Glucose Monitoring Performance in Patients With Liver Cirrhosis

Abstract

Aim:

To evaluate the use of intermittently scanned continuous glucose monitoring (isCGM) in patients with liver cirrhosis (LC).

Methods:

Observational study including 30 outpatients with LC (Child-Pugh B/C): 10 without diabetes (DM) (G1), 10 with newly diagnosed DM by oral glucose tolerance test (G2), and 10 with a previous DM diagnosis (G3). isCGM (FreeStyle Libre Pro) was used for 56 days (four sensors/patient). Blood tests were performed at baseline and after 28 and 56 days.

Results:

No differences were found in the baseline characteristics, except for higher age in G3. There were significant differences between G1, G2 and G3 in glucose management indicator (GMI) (5.28 ± 0.17, 6.03 ± 0.59, 6.86 ± 1.08%, P < .001), HbA1c (4.82 ± 0.39, 5.34 ± 1.26, 6.97 ± 1.47%, P < .001), average glucose (82.79 ± 7.06, 113.39 ± 24.32, 149.14 ± 45.31mg/dL, P < .001), time in range (TIR) (70.89 ± 9.76, 80.2 ± 13.55, 57.96 ± 17.96%, P = .006), and glucose variability (26.1 ± 5.0, 28.21 ± 5.39, 35.31 ± 6.85%, P = .004). There was discordance between GMI and HbA1c when all groups were considered together, with a mean difference of 0.35% (95% SD 0.17, 0.63). In G1, the mean difference was 0.46% (95% SD 0.19, 0.73) and in G2 0.69% (95% SD 0.45, 1.33). GMI and HbA1c were concordant in G3, with a mean difference of −0.10 % (95% SD [−0.59, 0.38]).

Conclusion:

Disagreements were found between the GMI and HbA1c levels in patients with LC. isCGM was able to detect abnormalities in glycemic control that would not be detected by monitoring with HbA1c, suggesting that isCGM can be useful in assessing glycemic control in patients with LC.

Keywords

diabetes mellitus liver cirrhosis continuous glucose monitoring glycated hemoglobin A1c

Introduction

The interplay between type 2 diabetes mellitus (T2D) and chronic liver disease (CLD) is well-established.^1-4 Among patients with liver cirrhosis (LC), the prevalence of glucose intolerance is up to 80%, and 30% have T2D.^5,6 The liver plays a primary role in carbohydrate metabolism, and LC contributes to dysglycemia through various mechanisms.^1,6 Diabetes mellitus (DM) is a well-recognized metabolic risk factor for CLD that predisposes patients to liver disease progression, LC complications, and decompensation events.⁷

Glycated hemoglobin (HbA1c) is the most widely used marker for long-term glucose control in DM and represents the average glucose status over the last 2 to 3 months.^8,9 HbA1c level does not adequately represent the glycemic status of patients with cirrhosis, which is generally underestimated.¹⁰ The reasons for reduced HbA1c levels in LC are not fully understood. Possible explanations include shortened erythrocyte lifespan and anemia, which are frequently observed in patients with advanced liver disease, whether due to bleeding or hemolysis related mainly to hypersplenism.^8,11

Diabetes mellitus monitoring in patients with advanced cirrhosis can be challenging because alternative options for HbA1c can also be affected by the disease. Glycated albumin has higher values in relation to plasma glucose levels in these individuals due to the prolonged half-life of serum albumin, consequent to the reduced synthesis capacity.^12,13 Fructosamine levels, in turn, are strongly influenced by the concentration of serum proteins and low-molecular-weight substances that coexist in the plasma (e.g., bilirubin, uric acid, etc.).^13-15

Self-monitoring of blood glucose (SMBG), usually performed with finger-prick blood samples, is not affected by these limitations. However, it depends on the patient’s adherence, does not provide complete information about blood glucose fluctuations, and nocturnal hypoglycemia usually remains undetected.^16,17

Continuous glucose monitoring (CGM) provide continuous measurement of glucose concentrations in the interstitial fluid, and has become an important tool for the clinical management of DM.^18,19 Intermittently scanned continuous glucose monitoring (isCGM) is a subset of CGM that requires periodic scanning of the glucose sensor.^20,21 Current scientific evidence supports the use of the CGM in the population with DM, with reasonable accuracy and some advantages, such as convenience for the patient and a more significant amount of data available compared with SMBG.^21-25

Individuals with LC have insulin resistance and hyperinsulinemia with marked glycemic fluctuations. While they frequently present postprandial hyperglycemia, after an overnight fast, patients with LC have a metabolic profile similar to that found in normal individuals after 2 to 3 days of fasting due to their low hepatic glycogen supply.^26,27 Hence, this is easily undervalued by standard monitoring tools. For this reason and considering that the systemic changes caused by liver dysfunction generate imprecision in standard markers of glycemic control, isCGM has potential applicability in this population, although there are not enough data to support its use. Therefore, the aim of this study was to compare isCGM monitoring with traditional measures of serum glycemic control in patients with LC.

Materials and Methods

Study Design

Analytical observational, single center study.

Setting and Subjects

Adult outpatients (aged 18 years or older) with LC, classified as Child-Pugh B or C, were evaluated between July 2019 and May 2022 at the Gastroenterology Clinic of the University Hospital of the Federal University of Santa Catarina, Brazil. The diagnosis of LC was established by histology (when available) or by a combination of imaging findings (ultrasound, computed tomography, or magnetic resonance imaging), laboratory, and clinical findings.

Exclusion criteria were as follows: Child-Pugh classification A; hemoglobin level less than 9.0 g/dL; active oncological disease; chronic kidney disease (estimated glomerular filtration rate less than 45 mL/min/1.73 m²); previous splenectomy; serum B12 vitamin less than 200 pg/mL; pharmacological treatment with dapsone or ribavirin; decompensated hypo or hyperthyroidism; thalassemic, falciform or hemolytic anemia; hyperbilirubinemia more than 20 mg/dL; chronic use of aspirin or opioids; alcohol consumption in the last 12 months; hypertriglyceridemia more than 500 mg/dL; those that did not complete at least 50% of the follow-up and those that refused to participate.

The sample was defined by convenience. The inclusion of patients occurred consecutively until the pre-established number of 30 patients was reached (10 in each study group). Initially, 10 patients with a previous diagnosis of DM according to American Diabetes Association criteria were selected.²⁸ Patients without an established diagnosis of DM underwent the oral glucose tolerance test (OGTT). Of these, 10 patients diagnosed with DM based on the OGTT (recently diagnosed group) and 10 patients without DM were included. In cases of DM diagnosed by the OGTT, patients were advised about lifestyle changes and referred for specialized follow-up.

Data Collection

The following clinical variables were collected: age, sex, previous diagnosis of DM, medications used, etiology of cirrhosis, presence of ascites, encephalopathy,²⁹ peripheral edema, and current or past alcohol consumption. Height was obtained with a stadiometer to the nearest 0.5 cm and weight was measured to the nearest 0.1 kg using a digital scale wearing light clothes and no shoes. Dry weight was calculated after adjusting for ascites and lower limb edema, as previously described.³⁰

Blood tests included hemoglobin, C-peptide, fructosamine, insulin, fasting glucose, HbA1c, total bilirubin, albumin, prothrombin time (INR), sodium, and creatinine. HbA1c was measured by high-performance liquid chromatography (Bio-Rad D-10, Bio-Rad Laboratories, USA), with reference values: <5.7% normoglycemia; 5.7 to 6.4% prediabetes or increased risk for DM; and ≥ 6.5% established diabetes. Fructosamine was measured by colorimetric assay (BioTécnica, Brazil), with a reference value of 205 to 285 mcmol/L. The remaining tests were performed using commercially available kits from the Clinical Analysis Laboratory of the University Hospital. Patients were classified according to the Child-Pugh score, and the model for end-stage liver disease (MELD) score was calculated.^31,32 Insulin resistance was evaluated using the homeostasis model assessment-insulin resistance (HOMA-IR) test.³³

The subcutaneous interstitial glucose levels were monitored using the isCGM FreeStyle Libre Pro^TM system (Abbott Diabetes Care, Alameda, CA, USA). This tool includes a reader device and a small disposable sensor applied on the back of the upper arm for up to 14 days, according to the manufacturer’s instructions. It is factory-calibrated, with data automatically stored on the sensor and is blind to the patient. Data were transferred to the reader after 14 days, and the results were downloaded to the LibreView software to generate summary glucose reports. The following parameters were analyzed: (1) average glucose (AG): average of all glucose measurements recorded in the period; (2) time in range (TIR): average percentage of time that individuals remained with glucose within the target range of 70 to 180 mg/dL; (3) time below range (TBR): average percentage of time that individuals remained with glucose below 70 mg/dL; (4) time below 54 mg/mL: mean percentage of time that individuals remained with glucose below 54 mg/dL; (5) time above range (TAR): average percentage of time that individuals remained with glucose above 180 mg/dL; (6) glucose variability: percent coefficient of variation (%CV), i.e. the ratio of standard deviation to mean glucose ([%CV] = standard deviation [SD]/mean interstitial glucose), with a target set to ≤ 36%³⁴; (7) Glucose Management Indicator (GMI): equation developed to estimate HbA1c based on mean interstitial glucose: (GMI [%] = 3.31 + 0.02392 x mean glucose [mg/dL])^35,36; and glycemia risk index (GRI): a number that summarizes the quality of glycemia, through a model that uses the percentage of time in the glycemia ranges 54 to <70 mg/dL (Low), <54 mg/dL (VLow), >180 to 250 mg/dL (High) and >250 mg/dL (VHigh) (GRI = [3.0 × VLow] + [2.4 × Low] + [1.6 × VHigh] + [0.8 × High]).³⁷

Assessments and follow-ups were performed exclusively by the main researcher over five visits. Initially, a date was scheduled for clinical evaluation and placement of the first glucose monitoring sensor on an outpatient basis. Every 14 days, the researcher scanned the sensor, removed it, and placed a new sensor up to four sensors per patient, totaling a follow-up of 56 consecutive days. The reader data were subsequently transferred to the respective software. Peripheral blood samples were collected after fasting for 8 to 12 hours on the day of placement of the first sensor (D0), after 28 (D28) and 56 days (D56).

Statistical Analysis

Numerical variables are expressed as mean and SD and qualitative variables are presented as absolute (N) and relative (%) frequencies. The normality of the distribution of variables was assessed using the Shapiro-Wilk test. In cases with normal distribution, the comparison between the means was performed using the analysis of variance (ANOVA) test, and the difference between the groups was evaluated using the Bonferroni correction (post hoc analysis). In cases of non-normal distribution, the Kruskal-Wallis test was used, followed by the Mann-Whitney test in case of statistical difference in the initial analysis. The correlation between variables was evaluated using Spearman’s correlation coefficient. Agreement between different methods with the same outcome was assessed using the Bland-Altman test, with analysis of proportion bias performed using simple regression. Descriptive levels (p) lower than .05 were considered statistically significant. All tests were performed using SPSS software (version 23.0, Chicago, Illinois, USA).

Ethical Approval

The study was conducted in accordance with the principles of the Declaration of Helsinki and approved by the Ethics Committee of the Federal University of Santa Catarina (approval number: CAAE 01062212.4.0000.0121). All participants provided written informed consent prior to entry into the study.

Sponsorship

FreeStyle Libre Pro^TM devices (sensors and readers) were supplied by Abbott Diabetes Care, Alameda, CA, USA. No research grants were provided.

Results

Patients were divided into three groups according to their DM diagnosis (group 1 [G1] = without DM, group 2 [G2] = recent diagnosis of DM by OGTT; group 3 [G3] = previous diagnosis of DM). Among patients in G3, eight (80%) were on a basal-bolus insulin regimen, with a mean insulin dose of 0.85 IU/Kg. Two patients were not treated with insulin but were treated with metformin plus sulfonylurea. One patient from G3 was excluded during the study, as he died before completing at least 50% of the follow-up and was therefore replaced to reach the proposed number of 10 patients per group.

The clinical and laboratory characteristics of each group at baseline are summarized in Table 1. Age was different between groups and was higher in G3, with no statistical difference between G1 and G2. There were no differences in BMI, sex, or variables associated with liver disease severity. Only two patients in G1 were classified as Child-Pugh C, and all other patients were classified as Child-Pugh B. There was also no difference across the groups in relation to pancreatic reserve assessed by serum C-peptide, as well as insulin resistance, estimated by HOMA-IR. The fasting glucose, HbA1c and fructosamine levels progressively increased from G1 to G3.

Table 1.

Distribution of Clinical and Laboratory Variables in Each Group.

Variables	Group 1cirrhosis without DM(n = 10)	Group 2cirrhosis with recent DM by OGTT(n = 10)	Group 3cirrhosis with DM(n = 10)	P value
Age (years)	50.8 ± 11.3	54.6 ± 8.9	67.2 ± 3.8	.001 ^a
BMI (kg/m²)	28.41 ± 4.73	24.83 ± 4.58	26.67 ± 3.48	.196
Male, n (%)	6 (60%)	6 (60%)	8 (80%)	.549
Child-Pugh	7.5 ± 0.71	8.10 ± 1.37	7.70 ± 0.82	.599
MELD	12.31 ± 2.71	12.25 ± 3.91	10.56 ± 2.69	.384
Hemoglobin (g/dL)	13.41 ± 2.13	12.82 ± 2.07	12.40 ± 2.18	.572
Creatinine (mg/dL)	0.89 ± 0.20	0.99 ± 0.18	0.92 ± 0.22	.492
Total Bilirubin (mg/dL)	1.91 ± 1.07	1.79 ± 0.74	1.48 ± 1.08	.252
Albumin (g/dL)	3.04 ± 0.39	3.10 ± 0.54	3.13 ± 0.32	.893
Sodium (mEq/L)	139.60 ± 2.41	138.30 ± 3.53	138.60 ± 3.81	.659
Prothrombin time (INR)	1.34 ± 0.16	1.40 ± 0.48	1.24 ± 0.14	.352
C-peptide (ng/mL)	2.39 ± 0.95	2.66 ± 1.23	2.37 ± 1.30	.830
HOMA-IR	4.33 ± 3.34	4.83 ± 2.77	N/A	.579
Insulin (mU/L)	18.63 ± 14.22	18.97 ± 10.64	N/A	.684
Fasting glucose (mg/dL)	94.10 ± 10.05	107.20 ± 21.99	148.50 ± 37.98	.001 ^b
HbA1c (%)	4.75 ± 0.43	5.39 ± 0.95	6.79 ± 1.41	<.001 ^c
Fructosamine (mcmol/L)	275.90 ± 36.90	297.90 ± 24.95	365.59 ± 76.19	.004 ^d

Abbreviations: DM, diabetes mellitus; OGTT, oral glucose tolerance test; BMI, body mass index; MELD, model for end-stage liver disease; INR, international normalized ratio; HOMA-IR, homeostasis model assessment of insulin resistance; N/A, not applicable; HbA1c, glycosylated hemoglobin A1c.

P = .997 for G1 versus G2; P = .001 for G1 versus G3, and P = .009 for G2 versus G3.

P = .069 for G1 versus G2; P = .001 for G1 versus G3, and P = .006 for G2 versus G3.

P = .508 for G1 versus G2; P = .000 for G1 versus G3, and P = .014 for G2 versus G3.

P = .059 for G1 versus G2; P = .006 for G1 versus G3, and P = .016 for G2 versus G3.

The most common cause of cirrhosis was alcohol-associated liver disease (ALD) (36%), followed by cryptogenic and chronic hepatitis C infection (HCV) (17% each); 13% autoimmune, 10% HCV associated with ALD and 7% chronic hepatitis B infection (HBV) with ALD.

During the study, there were a total of six sensor losses, two in each group: in G1 and G3, there was a loss of the third and another of the fourth sensor, in different participants, and in G2, there was a loss of the second and one of the fourth sensor, also in different participants.

There were differences between the three groups concerning GMI, average TIR, average TBR, glycemic variability and average GRI, as shown in Table 2.

Table 2.

Comparison of isCGM Data Between Groups.

Variables	Group 1Cirrhosis without DM(n = 10)	Group 2Cirrhosis with recent DM by OGTT(n = 10)	Group 3Cirrhosis with DM(n = 10)	P value
isCGM
GMI (%)	5.28 ± 0.17	6.03 ± 0.59	6.86 ± 1.08	<.001 ^a
Average glucose (mg/dL)	82.79 ± 7.06	113.39 ± 24.32	149.14 ± 45.31	<.001 ^b
TIR (%)	70.89 ± 9.76	80.20 ± 13.35	57.96 ± 17.96	.006 ^c
TBR (%)	30.95 ± 11.70	11.43 ± 10.16	12.48 ± 12.51	.001 ^d
Glucose variability (%)	26.10 ± 5.00	28.20 ± 5.39	35.31 ± 6.85	.004 ^e
GRI	72.40 ± 20.79	35.90 ± 24.19	55.30 ± 17.60	.030 ^f

Abbreviations: isCGM, intermittently scanned continuous glucose monitoring; DM, diabetes mellitus; OGTT, oral glucose tolerance test; GMI, glucose management indicator; TIR, time in range; TBR, time below range; GRI, glycemia risk index.

P = .081 for G1 versus G2; P< .001 for G1 versus G3; and P = .043 for G2 versus G3.

P = .092 for G1 versus G2; P< .001 for G1 versus G3; and P = .038 for G2 versus G3.

P = .264 for G1 versus G2; P = .330 for G1 versus G3; and P = .006 for G2 versus G3.

P = .002 for G1 versus G2; P = .004 for G1 versus G3; and P = 1.000 for G2 versus G3.

P = 1.000 for G1 versus G2; P = .004 for G1 versus G3; and P = .033 for G2 versus G3.

P = .002 for G1 versus G2; P = .249 for G1 versus G3;s and P = .147 for G2 versus G3.

Table 3 shows Spearman’s correlation analysis between GMI and clinical, laboratory, and other isCGM parameters after 56 days of follow-up. For this analysis, the last HbA1c (D56), average fructosamine of the second (D28) and third (D56) blood test, and average fasting glucose of all blood tests (D0, D28 e D56) were used.

Table 3.

Correlations Between GMI and Clinical, Laboratory and isCGM Data Variables.

Average GMI	Group 1		Group 2		Group 3
Average GMI	r	P	r	P	r	P
Age (years)	−.252	.482	.596	.069	−.238	.508
BMI (kg/m²)	.681	.030	.316	.374	.442	.200
MELD	.583	.077	.043	.906	−.237	.510
HbA1c (%)	.203	.574	.618	.057	.827	.003
Fasting glucose (mg/dL)	.203	.574	.924	.000	.806	.005
Fructosamine (mcmol/L)	.472	.168	.401	.25	.697	.025
isCGM: Average glucose (mg/dL)	.988	.000	.997	.000	1.000	.000
TIR (%)	.583	.077	−.267	.455	−.758	.011
TBR (%)	−.583	.077	−.462	.179	−.661	.038
Time below 54 mg/dL (%)	−.542	.106	−.506	.136	−.745	.013
∆ (GMI × HbA1c)	.105	.774	−.462	.179	−.236	.511
Glucose variability (%)	.693	.026	.036	.920	.03	.934
GRI	−.615	.058	−.134	.712	−.273	.446

Abbreviations: GMI, glucose management indicator; isCGM, intermittently scanned continuous glucose monitoring; BMI, body mass index; MELD, model for end-stage liver disease; HbA1c, glycosylated hemoglobin A1c; TIR, time in range; TBR, time below range; GRI, glycemia risk index.

As expected, there was a positive correlation between the GMI and AG in all groups. In G1, there was also a correlation with BMI and mean glycemic variability. In G2, the correlation was positive for fasting glucose and HbA1c. In G3, there was a positive correlation with fasting glucose and fructosamine levels. A negative correlation was observed in G3 with mean TIR, mean TBR and mean time below 54 mg/dL.

Average GMI of the 56 days and last HbA1c (D56) for each group is presented in Table 4 and a graph representation of the discordance between GMI and HbA1c is presented in Figure 1. A numerical difference between these variables was found, with HbA1c tending to be lower than GMI, except for G3. A significant discrepancy (≥0.5%) was observed in at least half of the patients in each group. In G1, five patients (50%) had higher GMI than HbA1c, with a difference greater than 0.5%. In G2, this difference was observed in six patients (60%), five of whom had an HbA1c value lower than GMI. In G3, six patients also showed a difference of 0.5% or more, but only half of them had HbA1c levels lower than GMI.

Table 4.

Average GMI of the 56 Days and Last HbA1c (D56).

	GMI	HbA1c
All groups	6.06 ± 0.95	5.71 ± 1.44
Group 1	5.28 ± 0.17	4.82 ± 0.39
Group 2	6.03 ± 0.59	5.34 ± 1.26
Group 3	6.86 ± 1.08	6.97 ± 1.47

Abbreviations: GMI, glucose management indicator; HbA1c, glycosylated hemoglobin A1C.

Figure 1.

Number of individuals in each group according to relative difference thresholds between GMI and HbA1c.

The Bland-Altman comparison between GMI and HbA1c showed disagreement when considering all groups together, with a mean difference of 0.35% (95% SD 0.17, 0.63). In group 1, the mean difference was 0.46% (95% SD 0.19, 0.73) and in group 2 0.69% (95% SD 0.45, 1.33), as shown in Figure 2. Conversely, there was agreement between GMI and HbA1c values in group 3, with a mean difference of −0.10 % (95% SD −0.59, 0.38), as shown in Figure 2. There was no proportion bias, as evaluated using simple regression (P = .075).

Figure 2.

Bland-Altman plots for comparison between GMI and HbA1c. The dots represent the differences between the paired measurements from the two methods plotted against their averages. The dashed lines indicate the 1.96 standard deviations of the differences, from the mean difference.

Figure 3 shows the percentage of time groups spent in each glycemic interval: time below 54 mg/dL, TBR (<70 mg/dL), TIR (70–180 mg/dL), and TAR (>180 mg/dL). G1 remained 69.05% of the TIR and 30.95% of the time in hypoglycemia, with 5.67% of values below 54 mg/dL. G2 spent 80.2% of TIR, 8.42% of time above 180 mg/dL and 11.38% below range, with 2.37% of the time below 54 mg/dL. While G3 remained 57.96% between 70 and 180 mg/dL, 29.56% above range and 12.48% below 70 mg/dL, with 3.44% below 54 mg/dL.

Figure 3.

Percentage of time groups spent in each glycemic interval.

A visual analysis was performed, and a pattern of low glucose values was observed during the overnight fasting period in patients in Groups 1 and 2, as shown in Figure 4. However, those in G3, of which 80% were in insulin users, did not show a well-defined pattern of hypoglycemia throughout the day (Figure 5).

Figure 4.

Visual analysis of the glycemic distribution pattern by glucose management indicator (GMI) in a single patient exemplifying the behavior typically observed in groups 1 and 2.

Figure 5.

Visual analysis of the glycemic distribution pattern by glucose management indicator (GMI) in a single patient exemplifying the behavior typically observed in group 3.

There were no serious complications related to sensor permanence. Minor skin problems were observed in five patients, including pruritus, erythema, hematoma, and mild pain, which improved after removal.

Discussion

This study, with outpatients with LC Child-Pugh classification B or C with and without DM, found a significant disagreement between the GMI and HbA1c parameters, which was higher in patients with a recent diagnosis of DM by OGTT. In this group, HbA1c levels were within the reference values for individuals without diabetes, whereas GMI values were higher. isCGM showed alterations in glycemic control that could not be detected by traditional follow-up based on HbA1c levels. The small sample size does not allow us to draw definitive conclusions, but the results suggest possible advantages of isGGM over conventional markers in this group of patients. To our knowledge, this is the first study of patients with decompensated LC (Child-Pugh B and C) comparing isCGM data with HbA1c.

Baseline characteristics were similar between the groups, except that the group with a previous diagnosis of diabetes was older than the others. A previous study evaluated the performance of isCGM compared to capillary blood glucose in patients aged 18 to 71 years with DM and found no age-related differences in sensor accuracy.²¹ Another study that specifically included patients with LC and DM also found no differences in the analytical accuracy of isCGM when comparing patients aged < 60 years with those aged 60 years and older.³⁸ Therefore, it is unlikely that this factor significantly influenced the results.

HbA1c is the most important test for monitoring blood glucose control in DM.²⁸ GMI represents the mean glucose derived from isCGM data and usually shows good agreement with HbA1c if there are no factors that significantly interfere with these parameters.^35,38,39 GMI and HbA1c can differ in any person owing to the non-glycemic factors involved in each method.⁴⁰ In addition, HbA1c can be influenced, along with other factors, by glycation kinetics and lifespan of red blood cells.⁸ Liver cirrhosis is considered an important factor related to HbA1c levels reduction. Cacciatore et al⁴¹ demonstrated that patients with LC and glucose intolerance or DM diagnosed by OGTT had HbA1c levels similar to those of patients with chronic hepatitis and controls without DM. Kanda et al⁴² observed lower HbA1c values in patients with comorbid LC and DM than in those with DM without LC. The reasons for underestimation of HbA1c in patients with LC are still not fully understood. A possible explanation is the short half-life of erythrocytes and anemia, frequently observed in patients with advanced liver disease, due to overt or occult bleeding related to portal hypertension and hemolysis associated with hypersplenism.⁸

Patients with LC and DM often have HbA1c levels that are similar to those of healthy individuals. Hence, this test is not recommended for DM diagnosis or monitoring in patients with advanced liver disease. OGTT is the method of choice for diagnosis.⁴ The results of the present study are in line with this recommendation since the group of patients with DM recently diagnosed by OGTT had episodes of hyperglycemia detected by isCGM during follow-up, despite normal HbA1c levels, both at baseline and at the end of the study. The group that presented normal OGTT and, therefore, was considered as not having DM did not present episodes of hyperglycemia according to isCGM data, which corroborates the discriminatory capacity of the OGTT. In addition, we did not observe significant anemia in our sample, suggesting the contribution of other factors to the reduction in HbA1c levels.

Discordant values were found between GMI and HbA1c when all participants were evaluated together and in the analysis of groups 1 and 2, with higher values of GMI in relation to HbA1c. The greatest discrepancy was observed in group 2, with a mean difference of 0.69%. There was agreement only in group 3, in which HbA1c levels approached the GMI. This was also the only group that showed a significant correlation between GMI and fructosamine levels. The reasons are unknown but may be related to the increase in HbA1c levels due to greater glycemic variability. Kuenen et al⁴³ observed that glycemic variability influenced the association between mean blood glucose and HbA1c in patients with type 1 DM, with greater glycemic variability leading to higher HbA1c values for the same mean glucose. In contrast, Liu et al⁴⁴ reported that the correlation coefficient between GMI and HbA1c was lower in patients with greater glycemic excursion amplitudes.

A recent study evaluated 20 patients diagnosed with LC and showed agreement between GMI and serum HbA1c results, with a tendency toward higher HbA1c values, unlike the results of the present study. However, the sample consisted predominantly of patients classified as Child-Pugh A (90%).⁴⁵ Another study, which primarily included patients with Child-Pugh A cirrhosis, also found an association between CGM results and HbA1c, but this was lower than that in patients with DM without LC.⁴⁶ Honda et al⁴⁷ evaluated patients with cirrhosis according to liver functional reserve compared to patients with chronic hepatitis, and found lower levels of HbA1c only in the group with LC Child-Pugh B and C. Finally, a study carried out by Ogawa et al⁴⁸ compared estimated HbA1c by CGM data with serum HbA1c, observing an increase in discrepancy with greater impairment of the hepatic reserve, with lower levels of serum HbA1c in patients with LC Child-Pugh C. Taken together, the results of these studies suggest that the severity of liver disease influences the magnitude of underestimation of HbA1c levels in this population.

A difference greater than or equal to 0.5% was observed between GMI and HbA1c in 56.6% of our sample. This frequency was higher than that one described by Bergenstal et al,³⁵ in which 28% of individuals with DM but without LC showed a difference greater than 0.5%. The result of the present study is, however, closer to that of a recent real-life study, which Perlman et al⁴⁹ observed a disagreement greater than 0.5% in 50% of individuals with DM. In individuals with chronic kidney disease and DM, in whom the interpretation of HbA1c is also limited, a discrepancy of more than 0.5% occurred in 68% of the sample.⁵⁰

The group of patients with a previous diagnosis of DM, despite the HbA1c within the target for adult individuals with DM (less than 7%), remained only 57.96% of the time in range, below the recommended of more than 70% of time, as well as 12.48% of the time in hypoglycemia (glucose below 70 mg/dL), which is above the target of less than 5% of the time. In this sense, the information provided by the isCGM can be beneficial for optimizing glycemic control.

On the other hand, patients in the recent diagnosis of DM by OGTT group remained at 80.2% of the time in range, but with 11.38% of the time below 70 mg/dL. Additionally, patients without DM were only 69.05% of the time between 70 and 180 mg/dL, as the rest of the time they remained with glucose below 70 mg/dL. When only the time with glucose < 54 mg/dL was evaluated, the three groups presented percentages above the expected of 1%: 5.67% for group 1; 2.37% for group 2, and 3.44% for group 3. The high percentage of time spent below the target in patients without DM is noteworthy, especially at night. The occurrence of blood glucose levels below 70mg/dL is not uncommon in patients with LC. Although insulin resistance is present in these patients, after overnight fasting, they usually have a metabolic profile similar to that found in normal individuals after two or three days of fasting due to impaired gluconeogenesis and low hepatic glycogen stores.⁴⁷ Hypoglycemia is related to the severity of cirrhosis. Ogawa et al⁴⁸ reported higher rates of nocturnal hypoglycemia in patients with LC classified as Child-Pugh C. In the present study, 20% of the group without DM had greater impairment of liver function (Child-Pugh C) in contrast to other groups composed exclusively of patients classified as Child-Pugh B. This scenario may explain, at least in part, the longest time below the range observed in this group.

The present study has some limitations, including the small sample size, which may have been insufficient to detect other differences between the evaluated groups. In addition, the diagnosis of recent diabetes was based on a single OGTT without retesting. It is also necessary to mention the limitation of the device itself because its accuracy is lower when there is a rapid change in glucose concentration, at lower glucose levels, and in the first 24 hours after the insertion of the sensor.^20,51 Capillary blood glucose measurements were not performed during follow-up, given that the isCGM device used in the study was blinded to the patient, precisely to guarantee that there was no interference from changes in eating, activity, or medication-related behavior during this period. Therefore, it was not possible to determine the accuracy of isCGM compared to a reference method. However, a recent study evaluated isCGM performance in patients with DM and LC. The results showed a strong agreement between the isCGM readings and capillary blood glucose. A 12.68% MARD was found in the LC group (31 patients) against 10.55% in the control group (30 patients without LC, but with DM), similar to what was observed in previous studies with different populations. Furthermore, 80.36% of the results were found in zone A and 99.83% in zones A + B of the Consensus Error Grid in LC and DM group. The accuracy of the device remained acceptable across the 14 days and through different glucose levels, and it was not affected by factors such as stage of cirrhosis, edema, and ascites. Although small, this study suggested that the use of isCGM has satisfactory performance in patients with LC.⁵²

In conclusion, we found disagreements between HbA1c and GMI levels in patients with LC, which were greater in patients with a recent diagnosis of DM by OGTT. isCGM was able to detect abnormalities in glycemic control that would not be detected by follow-up using HbA1c. Although further studies are needed, this study suggests that isCGM can be a useful method for assessing glucose control in patients with LC for whom the use of HbA1c is known to be inaccurate.

Research Data

sj-xlsx-1-dst-10.1177_19322968241232686 – Supplemental material for Intermittently Scanned Continuous Glucose Monitoring Performance in Patients With Liver Cirrhosis

sj-xlsx-1-dst-10.1177_19322968241232686 for Intermittently Scanned Continuous Glucose Monitoring Performance in Patients With Liver Cirrhosis by Fernanda Augustini Rigon, Marcelo Fernando Ronsoni, Alexandre Hohl, André Gustavo Daher Vianna, Simone van de Sande-Lee and Leonardo de Lucca Schiavon in Journal of Diabetes Science and Technology

This article is distributed under the terms of the Creative Commons Attribution 4.0 License (https://creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage).

Footnotes

Abbreviations

%CV, percent coefficient of variation; AG, average glucose; ALD, alcohol-associated liver disease; BMI, body mass index; CGM, Continuous glucose monitoring; CLD, chronic liver disease; D0, day of placement of the first sensor; D28, after 28 days; D56, after 56 days; DM, diabetes mellitus; G1, group 1; G2, group 2; G3, group 3; GMI, Glucose Management Indicator; GRI, glycemia risk index; HbA1c, glycated hemoglobin; HBV, chronic hepatitis B infection; HCV, chronic hepatitis C infection; HOMA-IR, homeostasis model assessment-insulin resistance; INR, prothrombin time; isCGM, Intermittently scanned continuous glucose monitoring; LC, liver cirrhosis; MELD, model for end-stage liver disease; OGTT, Oral glucose tolerance test; SD, standard deviation; SMBG, Self-monitoring of blood glucose; T2D, type 2 diabetes mellitus; TAR, time above range; TBR, time below range; TIR, time in range.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Fernanda Augustini Rigon

Alexandre Hohl

References

Picardi

D’Avola

Gentilucci

, et al. Diabetes in chronic liver disease: from old concepts to new evidence. Diabetes Metab Res Rev. 2006;22(4):274-283. https://onlinelibrary.wiley.com/doi/10.1002/dmrr.636. Accessed August 16, 2022.

Kishimoto

Noda

. Verification of glycemic profiles using continuous glucose monitoring: cases with steroid use, liver cirrhosis, enteral nutrition, or late dumping syndrome. J Med Invest. 2015;62(1-2):1-10. https://www.jstage.jst.go.jp/article/jmi/62/1.2/62_1/_article. Accessed August 16, 2022.

Kawaguchi

. Insulin resistance and chronic liver disease. World J Hepatol. 2011;3(5):99. http://www.wjgnet.com/1948-5182/full/v3/i5/99.htm. Accessed August 16, 2022.

Elkrief

Rautou

Sarin

Valla

Paradis

Moreau

. Diabetes mellitus in patients with cirrhosis: clinical implications and management. Liver Int. 2016;36(7):936-948. http://doi.wiley.com/10.1111/liv.13115. Accessed February 22, 2019.

García-Compeán

Jáquez-Quintana

Lavalle-González

, et al. The prevalence and clinical characteristics of glucose metabolism disorders in patients with liver cirrhosis. A prospective study. Ann Hepatol. 2012;11(2):240-248.

Garcia-Compean

Jaquez-Quintana

Gonzalez-Gonzalez

Maldonado-Garza

. Liver cirrhosis and diabetes: risk factors, pathophysiology, clinical implications and management. World J Gastroenterol. 2009;15(3):280. http://www.wjgnet.com/1007-9327/full/v15/i3/280.htm. Accessed April 22, 2021.

Liu

Trogdon

Weinberger

Fried

Barritt

4th . Diabetes is associated with clinical decompensation events in patients with cirrhosis. Dig Dis Sci. 2016;61(11):3335-3345. http://link.springer.com/10.1007/s10620-016-4261-8. Accessed August 16, 2022.

Silva

Ronsoni

Schiavon

. Challenges in diagnosing and monitoring diabetes in patients with chronic liver diseases. Diabetes Metab Syndr. 2018;12(3):431-440. https://linkinghub.elsevier.com/retrieve/pii/S1871402117304435. Accessed March 2, 2019.

Lenters-Westra

Schindhelm

Bilo

Slingerland

. Haemoglobin A1c: historical overview and current concepts. Diabetes Res Clin Pract. 2013;99(2):75-84. https://linkinghub.elsevier.com/retrieve/pii/S0168822712003749. Accessed July 10, 2022.

10.

Nadelson

Satapathy

Nair

. Glycated hemoglobin levels in patients with decompensated cirrhosis. Int J Endocrinol. 2016;2016:8390210. https://www.hindawi.com/journals/ije/2016/8390210/. Accessed July 10, 2022.

11.

Koga

Kasayama

Kanehara

Bando

. CLD (chronic liver diseases)-HbA1C as a suitable indicator for estimation of mean plasma glucose in patients with chronic liver diseases. Diabetes Res Clin Pract. 2008;81(2):258-262. https://linkinghub.elsevier.com/retrieve/pii/S0168822708001940. Accessed February 16, 2019.

12.

Wright

Hirsch

. Metrics beyond hemoglobin A1C in diabetes management: time in range, hypoglycemia, and other parameters. Diabetes Technol Ther. 2017;19(suppl 2):S16-S26. http://www.liebertpub.com/doi/10.1089/dia.2017.0029. Accessed February 16, 2019.

13.

Constanti

Simo

Joven

Camps

. Serum fructosamine concentration in patients with nephrotic syndrome and with cirrhosis of the liver: the influence of hypoalbuminaemia and hypergammaglobulinaemia. Ann Clin Biochem. 1992;29(pt 4):437-442. https://journals.sagepub.com/doi/10.1177/000456329202900412?url_ver=Z39.88-2003&rfr_id=ori:rid:crossref.org&rfr_dat=cr_pub%20%200pubmed. Accessed February 17, 2019.

14.

Armbruster

. Fructosamine: structure, analysis, and clinical usefulness. Clin Chem. 1987;33(12):2153-2163.

15.

Koga

Kasayama

. Clinical impact of glycated albumin as another glycemic control marker. Endocr J. 2010;57(9):751-762. http://joi.jlc.jst.go.jp/JST.JSTAGE/endocrj/K10E-138?from=CrossRef. Accessed February 16, 2019.

16.

Dunn

Hayter

Ajjan

. Real-world flash glucose monitoring patterns and associations between self-monitoring frequency and glycaemic measures: a European analysis of over 60 million glucose tests. Diabetes Res Clin Pract. 2018;137:37-46. https://linkinghub.elsevier.com/retrieve/pii/S016882271731104X. Accessed August 16, 2022.

17.

Heinemann

. Finger pricking and pain: a never ending story. J Diabetes Sci Technol. 2008;2(5):919-921. http://journals.sagepub.com/doi/10.1177/193229680800200526. Accessed August 16, 2022.

18.

Haak

Hanaire

Ajjan

Hermanns

Riveline

Rayman

. Use of flash glucose-sensing technology for 12 months as a replacement for blood glucose monitoring in insulin-treated type 2 diabetes. Diabetes Ther. 2017;8(3):573-586. http://link.springer.com/10.1007/s13300-017-0255-6. Accessed October 10, 2019.

19.

Ish-Shalom

Wainstein

Raz

Mosenzon

. Improvement in glucose control in difficult-to-control patients with diabetes using a novel flash glucose monitoring device. J Diabetes Sci Technol. 2016;10(6):1412-1413. http://journals.sagepub.com/doi/10.1177/1932296816653412. Accessed August 16, 2022.

20.

Leelarathna

Wilmot

. Flash forward: a review of flash glucose monitoring. Diabet Med. 2018;35(4):472-482. https://onlinelibrary.wiley.com/doi/10.1111/dme.13584. Accessed August 16, 2022.

21.

Bailey

Bode

Christiansen

Klaff

Alva

. The performance and usability of a factory-calibrated flash glucose monitoring system. Diabetes Technol Ther. 2015;17(11):787-794. http://www.liebertpub.com/doi/10.1089/dia.2014.0378. Accessed February 23, 2019.

22.

Al Hayek

Robert

Al Dawish

. Differences of freestyle libre flash glucose monitoring system and finger pricks on clinical characteristics and glucose monitoring satisfactions in type 1 diabetes using insulin pump. Clin Med Insights Endocrinol Diabetes. 2019;12:1179551419861102. http://journals.sagepub.com/doi/10.1177/1179551419861102. Accessed August 16, 2022.

23.

American Diabetes Association Professional Practice Committee; ElSayed

Aleppo

Bannuru

, et al. Diabetes technology: standards of care in diabetes—2024. Diabetes Care. 2024;47(suppl 1):S126-S144. https://diabetesjournals.org/care/article/47/Supplement_1/S126/153939/7-Diabetes-Technology-Standards-of-Care-in. Accessed January 12, 2024.

24.

Ying

Wang

Han

Zhang

. Effects of continuous glucose monitoring on glycaemic control in type 2 diabetes: a systematic review and network meta-analysis of randomized controlled trials. Diabetes Obes Metab. 2024;26(1):362-372. https://dom-pubs.pericles-prod.literatumonline.com/doi/10.1111/dom.15328. Accessed January 17, 2024.

25.

Teo

Hassan

Tam

Koh

. Effectiveness of continuous glucose monitoring in maintaining glycaemic control among people with type 1 diabetes mellitus: a systematic review of randomised controlled trials and meta-analysis. Diabetologia. 2022;65(4):604-619. https://link.springer.com/10.1007/s00125-021-05648-4. Accessed Jan 12, 2024.

26.

Owen

Trapp

Reichard

Jr , et al. Nature and quantity of fuels consumed in patients with alcoholic cirrhosis. J Clin Invest. 1983;72(5):1821-1832. http://www.jci.org/articles/view/111142. Accessed February 17, 2019.

27.

Rigon

Ronsoni

Vianna

AGD

, et al. Flash glucose monitoring system in special situations. Arch Endocrinol Metab. 2022;66(6):883-894. https://www.aem-sbem.com/article/flash-glucose-monitoring-system-in-special-situations/. Accessed July 2, 2022.

28.

American Diabetes Association. Standards of medical care in diabetes—2022. Abridged for primary care providers. Clin Diabetes. 2022;40(1):10-38. https://diabetesjournals.org/clinical/article/40/1/10/139035/Standards-of-Medical-Care-in-Diabetes-2022. Accessed July 16, 2022.

29.

Weissenborn

. Hepatic encephalopathy: definition, clinical grading and diagnostic principles. Drugs. 2019;79(suppl 1):5-9. http://link.springer.com/10.1007/s40265-018-1018-z. Accessed July 24, 2023.

30.

Duarte

Castellani

. Semiologia Nutricional. Rio de janeiro: Axcel Books, 2002.

31.

Child

Turcotte

. Surgery and portal hypertension. Major Probl Clin Surg. 1964;1:1-85.

32.

Kamath

Kim

. The model for end-stage liver disease (MELD). Hepatology. 2007;45(3):797-805. https://onlinelibrary.wiley.com/doi/10.1002/hep.21563. Accessed July 12, 2022.

33.

Levy

Matthews

Hermans

. Correct homeostasis model assessment (HOMA) evaluation uses the computer program. Diabetes Care. 1998;21(12):2191-2192. https://diabetesjournals.org/care/article/21/12/2191/21899/Correct-Homeostasis-Model-Assessment-HOMA. Accessed July 24, 2023.

34.

Danne

Nimri

Battelino

, et al. International consensus on use of continuous glucose monitoring. Diabetes Care. 2017;40(12):1631-1640. https://diabetesjournals.org/care/article/40/12/1631/37000/International-Consensus-on-Use-of-Continuous. Accessed August 16, 2022.

35.

Bergenstal

Beck

, et al. Glucose management indicator (GMI): a new term for estimating A1C from continuous glucose monitoring. Diabetes Care. 2018;41(11):2275-2280. https://diabetesjournals.org/care/article/41/11/2275/36593/Glucose-Management-Indicator-GMI-A-New-Term-for. Accessed July 11, 2022.

36.

Battelino

Danne

Bergenstal

, et al. Clinical targets for continuous glucose monitoring data interpretation: recommendations from the international consensus on time in range. Diabetes Care. 2019;42(8):1593-1603. http://care.diabetesjournals.org/lookup/doi/10.2337/dci19-0028. Accessed November 20, 2021.

37.

Klonoff

Wang

Rodbard

, et al. A glycemia risk index (GRI) of hypoglycemia and hyperglycemia for continuous glucose monitoring validated by clinician ratings. J Diabetes Sci Technol. 2023;17(5):1226-1242.

38.

Lin

Brown

Ekinci

. The ambulatory glucose profile and its interpretation. Med J Aust. 2022;217(6):295-298. https://onlinelibrary.wiley.com/doi/10.5694/mja2.51666. Accessed August 31, 2022.

39.

Dunn

Bergenstal

Ogawa

Ajjan

. Personalized glycated hemoglobin in diabetes management: closing the gap with glucose management indicator. Diabetes Technol Ther. 2023;25(suppl 3):S65-S74. https://www.liebertpub.com/doi/10.1089/dia.2023.0146. Accessed January 12, 2024.

40.

Gomez-Peralta

Choudhary

Cosson

Irace

Rami-Merhar

Seibold

. Understanding the clinical implications of differences between glucose management indicator and glycated haemoglobin. Diabetes Obes Metab. 2022;24(4):599-608. https://onlinelibrary.wiley.com/doi/10.1111/dom.14638. Accessed August 21, 2022.

41.

Cacciatore

Cozzolino

Giardina

, et al. Abnormalities of glucose metabolism induced by liver cirrhosis and glycosylated hemoglobin levels in chronic liver disease. Diabetes Res. 1988;7(4):185-188.

42.

Kanda

Nakamura

Iwasaki

Wada

Kurotobi

Shimizu

. Clinical significance of HbA1c, serum fructosamine and 1,5-AG in diabetic patients complicated with liver cirrhosis. J Japan Diabetes Soc. 1993;36:847-854.

43.

Kuenen

Borg

Kuik

, et al. Does glucose variability influence the relationship between mean plasma glucose and HbA 1c levels in type 1 and type 2 diabetic patients? Diabetes Care. 2011;34(8):1843-1847. http://care.diabetesjournals.org/lookup/doi/10.2337/dc10-2217. Accessed February 17, 2019.

44.

Liu

Yang

Deng

, et al. Impacts of glycemic variability on the relationship between glucose management indicator from iPro^™2 and laboratory hemoglobin A1c in adult patients with type 1 diabetes mellitus. Ther Adv Endocrinol Metab. 2020;11:2042018820931664. Published 2020 June 8. http://journals.sagepub.com/doi/10.1177/2042018820931664. Accessed August 25, 2022.

45.

Barron

CVM

Heenan

Thompson

Chan

Ngu

Lunt

. Detecting dysglycaemia in compensated liver cirrhosis: comparison of oral glucose tolerance test and glycated haemoglobin, with continuous glucose monitoring. Diabet Med. 2022;3939(3):e14778. https://onlinelibrary.wiley.com/doi/10.1111/dme.14778. Accessed August 26, 2022.

46.

Addepally

George

Martinez-Macias

Garcia-Saenz-de-Sicilia

Kim

Duarte-Rojo

. Hemoglobin A1c has suboptimal performance to diagnose and monitor diabetes mellitus in patients with cirrhosis. Dig Dis Sci. 2018;63(12):3498-3508. http://link.springer.com/10.1007/s10620-018-5265-3. Accessed February 22, 2019.

47.

Honda

Hiramatsu

Hyogo

, et al. Evaluation of glycemic variability in chronic liver disease patients with type 2 diabetes mellitus using continuous glucose monitoring. PLoS ONE. 2018;13(4):e0195028. https://dx.plos.org/10.1371/journal.pone.0195028. Accessed February 16, 2019.

48.

Ogawa

Nakahara

Ono

, et al. Underestimation of impaired glucose tolerance and usefulness of a continuous glucose monitoring system in chronic liver disease. J Gastroenterol Hepatol. 2022;37(3):592-599. https://onlinelibrary.wiley.com/doi/10.1111/jgh.15766. Accessed August 31, 2022.

49.

Perlman

Gooley

McNulty

Meyers

Hirsch

. HbA1c and glucose management indicator discordance: a real-world analysis. Diabetes Technol Ther. 2021;23(4):253-258. https://www.liebertpub.com/doi/10.1089/dia.2020.0501. Accessed August 22, 2022.

50.

Oriot

Viry

Vandelaer

, et al. Discordance between glycated hemoglobin A1c and the glucose management indicator in people with diabetes and chronic kidney disease. J Diabetes Sci Technol. 2022;17(6):1553–1562. http://journals.sagepub.com/doi/10.1177/19322968221092050. Accessed August 22, 2022.

51.

Blum

. Freestyle libre glucose monitoring system. Clin Diabetes. 2018;36(2):203-204. https://diabetesjournals.org/clinical/article/36/2/203/32874/Freestyle-Libre-Glucose-Monitoring-System. Accessed January 12, 2024.

52.

Costa

Lourenço

Monteiro

, et al. Clinical performance of flash glucose monitoring system in patients with liver cirrhosis and diabetes mellitus. Sci Rep. 2020;10(1):7460. http://www.nature.com/articles/s41598-020-64141-x. Accessed June 25, 2022.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB