Relationship Between HbA1c and Chronic Glycemia in Patients With Cirrhosis

Glycosylated hemoglobin A1c (HbA1c), a reliable test to assess chronic glycemia and recommended both for the diagnosis and monitoring of T2DM, is neither accurate nor reliable in patients with cirrhosis. A validation study has not been performed to define its true usefulness in the setting of cirrhosis. The study aims to determine the level of HbA1c that better corresponds to the diagnosis of T2DM - as determined by an oral glucose tolerance test (OGTT) - and to correlate the levels of HbA1c with the average glucose over a 12-week period in patients with cirrhosis and known T2DM, in cirrhotic patients with different degrees of liver impairment. This study is designed as a pilot, observational, cross-sectional (Group A - cirrhotics with no T2DM) and longitudinal study (Group B - cirrhotics with known diabetes). For our first aim (Group A) the diagnostic yield of HbA1c for T2DM will be compared to that of an OGTT. Regarding the second aim (Group B) HbA1c will be compared to glucose reading from a continuous glucose monitor. In both groups, the following study labs will be performed: fructosamine, Complete Blood Count (CBC), liver function tests, chemistry, International Normalized Ratio (INR), and pre albumin. A total of 26 patients per group will be included.

STUDY AIMS

The study will look into three primary aims:

1. To determine the level of HbA1c that better corresponds to the diagnosis of T2DM - as determined by an OGTT - in cirrhotic patients with different degrees of liver impairment.

• Will test the hypothesis of lower levels of HbA1c being found in T2DM in patients with decompensated cirrhosis when compared to those with compensated disease.

1. To correlate the levels of HbA1c with the average glucose (AG) over a 12-week period in patients with cirrhosis and known T2DM, according to different degrees of liver impairment.

• Will test the hypothesis of progressively lower levels of HbA1c being associated with hyperglycemia in diabetic patients with decompensated cirrhosis when compared to those with compensated disease.

1. To compare the performance of continuous glucose monitoring, particularly the level of agreement between the AG over a 12-week period and HbA1c, in patients with cirrhosis and known T2DM with that of controls with T2DM but no known chronic liver disease (including cirrhosis).

As secondary aims will:

1. Determine the prevalence of T2DM in cirrhotics with different degrees of liver disease according to use of fasting glucose, HbA1c, or OGTT.
2. Determine the level of HbA1c that better defines well-controlled diabetes in patients with cirrhosis and known T2DM, according to different degrees of liver impairment.
3. Glucose variability in patients with cirrhosis and T2DM with that of patients with T2DM but no chronic liver disease (including cirrhosis).

Study Procedures

All patients will be identified and recruited from any of the Liver or GI clinics in the Outpatient Center, UAMS. To test the first aim (Group A) - the cross-sectional part of the study - patients with cirrhosis and varying degrees of liver dysfunction but no prior diagnosis of T2DM will have an OGTT performed after overnight fasting (8-12 hours). The OGTT consists of taking a baseline glucose determination, followed by a second one 2 hours after swallowing a 75-g glucose load. HbA1c will be obtained before OGTT (with baseline glucose) along with the following tests: HbA1c, fructosamine, insulin, CBC, HFP, GGT, OP Chem 7, INR, and prealbumin. Anthropometrics (height, weight, and waist circumference) will be obtained as well (Research Visit 1, see calendar).

All patients will be identified and recruited from any of the Liver or GI clinics in the Outpatient Center, UAMS. For the second aim (Group B) - the longitudinal part of the study - patients with known diabetes and cirrhosis with varying degrees of liver dysfunction will wear a continuous glucose monitor (Dexcom, San Diego, CA) for ≥3 continuous days during week days and week ends, every 4 weeks for 12 consecutive weeks, according to the manufacturer's recommendations. The continuous glucose monitor will record patient's glucose from the odd to the even research visit (i.e. from 1 to 2, 3 to 4, 5 to 6, and 7 to 8). On the last day the continuous glucose monitor is retrieved by the investigators, patients will have the following blood tests performed after overnight fasting (8-12 hours): HbA1c, fructosamine, CBC, HFP, GGT, BMP, INR, and prealbumin. Anthropometrics (height, weight, and waist circumference) will be obtained as well (Research Visit 8, see calendar). In order not to affect glucose control during the study period glucose reads from the continuous glucose monitor will not be disclosed to the patient until the end of the study.

All controls will be identified from the Endocrinology or Internal Medicine clinics in the Outpatient Center, UAMS. Controls will wear a continuous glucose monitor (Dexcom, San Diego, CA) for ≥3 continuous days during week days and week ends, every 4 weeks for 12 consecutive weeks, according to the manufacturer's recommendations. The continuous glucose monitor will record patient's glucose from the odd to the even research visit (i.e. from 1 to 2, 3 to 4, 5 to 6, and 7 to 8). On the last day the continuous glucose monitor is retrieved by the investigators, patients will have the following blood tests performed after overnight fasting (8-12 hours): HbA1c, CBC, and a metabolic profile; as per standard of care recommendations. Anthropometrics (height, weight, and waist circumference) will be obtained as well (Research Visit 8, see calendar - this research visit will occur on the same day of a diabetes follow up outpatient visit). In order not to affect glucose control during the study period glucose reads from the continuous glucose monitor will not be disclosed to the patient until the end of the study.

Given that the study only attempts to investigate the accuracy of HbA1c by comparing it to glucose and other biomarkers, no specific therapeutic interventions or recommendations concerning treatment of diabetes will be required during the study.

Sample Size

Since there is no precedent in the literature showing preliminary data to allow for sample calculation, will include 30 patients for each of the CTP stages A, B, and C, for both groups. This number should allow more than the minimum needed to perform correlation analyses (between and within groups) with confidence.

Sample Size

Since there is no precedent in the literature showing preliminary data to allow for sample calculation, will aim to have complete data on 26 patients per group (10 CTP A, 8 CTP B, and 8 CTP C in each group). However, given that the estimated attrition rate in Group B is 30%, will aim to recruit up to 34 patients for this group. We are planning on recruiting one control for each subject in control B, so that the number of controls would not exceed 34. Thus, the total number of subjects in study will be no more than 94. This number should allow more than the minimum needed to perform correlation analyses (between and within groups) with confidence.

Data Analysis

All glucose determinations from the continuous glucose monitoring will be summarized as an arithmetic mean. Shapiro-Wilk will be used to determine normal distribution of data. Receiver operation characteristic (ROC) curves will be constructed to determine the best cut-off to diagnose T2DM, using OGTT results as the gold standard. Spearman's or Pearson's correlation coefficient will be used to compare HbA1c to average glucose (eAG). Bland-Altman will be used to determine agreement. The estimated average glucose from HbA1c will be derived from a linear regression model, as previously described. Linear regression models will be performed to identify factors affecting HbA1c for both Groups A and B, such as hemoglobin, CTP, MELD, total protein, and prealbumin. CTP will be transformed to dummy variables for this analysis.

Data collection and handling Apart from the data generated by the study, demographics and clinical information to properly classify the hepatic functional and diabetes of each participant will be retrieved from the electronic medical record (EMR). Such information will include results from standard of care laboratories, imaging, and other special studies (EGD), to calculate Child-Turcotte-Pugh (CTP) and MELD scores.

A database will be created to accumulate all of the clinical information generated during this study. No identifiers, apart from name and Date of Birth(DOB), will be collected. All data will be stored in databases kept in a study folder created at PI's computer (password-protected). Main database will include identification (ID) assigned, and once an analytic table is created, all PHI will be removed from the table (after approval by the PI). Only this de-identified dataset with unique study ID's will be used for statistical analyses. Therefore, the risk to the privacy of the individuals will be minimized.

The access to the research folder and its contents will be restricted to the research staffs listed in this submission form. The PHI will be kept in the main database for no more than 10 years. The main database containing PHI will be kept in the study folder and will be never transferred out without de-identification of the data. No hard copy research data will be generated in this project.

Risks and Benefits

The risks associated with venipuncture and repeated blood collection from a vein are minor bleeding, bruising, infection (a small risk any time the skin is broken), or needing more than one attempt to locate veins.

There are no risks to the OGTT, although subjects may not like the taste of the highly concentrated glucose beverage. Occasionally someone may have low blood sugar or high blood sugar after the test. Blood glucose will be monitored before and during the test.

Some risks associated with the Dexcom G4 System Device have been reported: placing the devices in your skin may cause some anxiety, pain, irritation, redness, and swelling. There is also the risk of infection, excessive bleeding, or hematoma (swelling filled with blood caused by a broken blood vessel) - these risks are rare. Removing the sensor may also cause some discomfort or irritation because of the tape used. This reaction usually goes away within a few hours, but may last up to a week. There may be some itching in the area, which is a normal healing response.

In rare cases, the sensor may break while inserted. If this happens, the study team and will examine the area. Sometimes the study team may decide to leave the sensor fragment in place. If there is redness, pain, or infection, the study team will decide how to treat it. You should not try to remove the sensor yourself if it is broken unless the study team tells you to.

the University of Arkansas (UAMS) firewall and only used for specific research purposes in this project. Aggregated data will be analyzed and published, but specific data elements will not be made available. Further, subjects will not forfeit any rights by participating - their health care and eligibility for health benefits will not be affected in any way. Although a breach in confidentiality is always a risk when collecting personal information from patients, we believe this is small given the system that has been developed to obtain and handle data.

Ethics This study will be conducted according to US and international standards of Good Clinical Practice (FDA regulations) for all studies. Applicable government regulations, University of Arkansas for Medical Sciences research policies and procedures will also be followed. This protocol and any amendments will be submitted and approved by the University of Arkansas for Medical Sciences Institutional Review Board (IRB) to conduct the study. All subjects for this study will be provided a consent form describing this study and providing sufficient information in language suitable for subjects to make an informed decision about their participation in this study. The formal consent of a subject, using the IRB-approved consent form, will be obtained before that subject is submitted to any study procedure. This consent form must be signed by the subject or legally acceptable surrogate, and the investigator-designated research professional obtaining the consent.