Article Text

Download PDFPDF

Clinical Predictors of Progressive Beta-Cell Failure in Type 2 Diabetes
  1. Concetta Irace, PhD*,
  2. Cesare Tripolino, MD*,
  3. Claudio Carallo, MD*,
  4. Faustina Barbara Scavelli, PhD*,
  5. Elisabetta Della Valle, MD,
  6. Claudio Cortese, MD,
  7. Agostino Gnasso, MD*
  1. From the *Department of Clinical and Experimental Medicine, University Magna Græcia, Catanzaro; †Department of Public Health, Federico II University, Naples; and ‡Department of Experimental Medicine and Surgery, Tor Vergata University, Rome, Italy.
  1. Received November 5, 2014, and in revised form March 21, 2015.
  2. Accepted for publication April 7, 2015.
  3. Reprints: Agostino Gnasso, MD, Department of Clinical and Experimental Medicine, University Magna Græcia, Catanzaro Viale Europa, Località Germaneto, 88100 Catanzaro, Italy. E-mail: gnasso{at}


Objectives The aim of the study was to identify factors associated with progressive beta-cell failure in a cohort of nonselected subjects with type 2 diabetes.

Methods Two hundred twenty-four medical records were evaluated. Progressive beta-cell failure was defined as the following: glycated hemoglobin is higher than 7.5% despite combined drug therapy and appropriate diet (ie, isocaloric or hypocaloric diet depending on body weight) and absence of any illness causing acute hyperglycemia. The following factors were considered as possible predictors: diabetes-related symptoms, fasting plasma glucose at the onset of disease, family history of type 2 diabetes, number of visits per year, and residency. Further potential predictors were disease duration, age, body mass index, estimated glomerular filtration rate, and hypertension and/or hyperlipidemia at the enrollment in the study.

Results The prevalence of beta-cell failure was 41%. Independent predictors of failure were longer disease duration (hazard ratio [HR] for each year of diabetes, 1.03; confidence intervals (CIs), 1.01–1.05; P = 0.03), history of hypertension (HR, 1.90; CIs, 1.73–2.89; P = 0.04), hyperlipidemia (HR, 1.65; CIs, 1.06–2.58; P = 0.03), residence in suburb (HR, 1.78; CIs, 1.06–3.01; P = 0.03), and presence of symptoms at the onset of disease (HR, 2.47; CIs, 1.51–4.03; P = 0.0001).

Conclusions Patients with long disease duration, hypertension, and hyperlipidemia who are residents in suburbs and had diabetes-related symptoms at diagnosis might deserve intensive treatment to obtain adequate and stable glycemic control.

Key Words
  • diabetes
  • beta cell
  • secondary failure
  • symptoms
  • oral hypoglycemic agent

Statistics from

Key Words

Type 2 diabetes is a complex disease of polygenic and environmental origin, mainly characterized by defective insulin secretion and peripheral insulin resistance.1

Subjects with type 2 diabetes are suggested to start lifestyle modification (nutritional habits and physical activity) and, if not contraindicated and well tolerated, are prescribed metformin.2Early treatment might slow the deterioration of insulin secretion and the onset of beta-cell failure.3–5Interventional studies have provided evidence of loss of glycemic control over time. In the UK Prospective Diabetes Study, only 25% of subjects undergoing intensive treatment were in the therapeutic range of glycated hemoglobin (HbA1c) after 9 years.6

In the more recent A Diabetes Outcome Progression Trial study, the cumulative incidence of progressive beta-cell failure ranged from 15% to 34% after 5 years of follow-up.7

Normally, pancreatic beta cells respond to initial insulin resistance by increasing insulin secretion to preserve a near-normal glucose level. Gradually, the reduction of beta-cell mass and the decline of insulin secretion develop.8Genetic predisposition, poor dietary control, glucotoxicity and lipotoxicity, liver steatosis, and therapy itself influence beta-cell failure and then the onset of progressive beta-cell failure.7,9–12

At the time of beta-cell failure, insulin therapy is the most appropriate therapy to restore a good metabolic control.13

Currently, we do not have reliable, unequivocal, and easily detectable predictors for progressive beta-cell failure, and the decision to start insulin regimen is based on clinical grounds and/or persistent elevated HbA1c.

The aim of this study was to possibly identify predictors associated with progressive beta-cell failure in a sample of subjects attending our metabolic clinic.


A retrospective cohort study was conducted using medical records of consecutive outpatients with type 2 diabetes who visited at our clinic from December 1, 2006 to December 1, 2007. Subjects who met the following inclusion criteria at screening were enrolled in the study: age, 35 to 70 years, HbA1c, less than 7.5%; noninsulin treatment; disease duration, more than 1 year; and at least 1 recorded visit through December 2013. Subjects with type 1 diabetes, Maturity Onset Diabetes of the Young, contraindication to hypoglycemic agents, history of gestational diabetes, and alcohol consumption exceeding 20 grams per day and subjects requiring insulin treatment during follow-up due to conditions causing acute hyperglycemia such as infections, cancer, surgery, myocardial infarction, pancreatitis, and corticosteroid treatment were excluded.

Progressive beta-cell failure was defined as the following: HbA1c is higher than 7.5% despite combined drug therapy and appropriate diet (ie, isocaloric or hypocaloric diet depending on body weight) and no apparent external reason for hyperglycemia.11,12Actually, a clear cutoff value for defining beta-cell failure has not been established; however, we chose an HbA1c value slightly higher than the target recommended by American Diabetes Association and European Association for the Study of Diabetes to be sure that the increase was durable.13Variables included in the study as potential predictors of progressive beta-cell failure were the following: diabetes-related symptoms (polyuria, polydipsia, fatigue, weight loss, blurred vision), fasting plasma glucose (FPG) at the onset of disease (cut off 200 mg/dL), family history of type 2 diabetes, number of visits per year, and residency (town or suburb). Furthermore, disease duration, age, body mass index (BMI), estimated glomerular filtration rate (eGFR), and concomitant hypertension and/or hyperlipidemia at the enrollment in the study were also evaluated. Body mass index was calculated according to the formula kg/m2, and obesity was defined as BMI 30 kg/m2 or greater. Estimated GFR was calculated with a shortened version of the Modification of Diet in Renal Disease.14Hypertension was defined as systolic and/or diastolic blood pressure greater than 139/89 mm Hg or treatment with antihypertensive drugs15; hyperlipidemia, as total cholesterol and/or triglycerides 200 mg/dL or greater or therapy with lipid-lowering drugs.16We also recorded smoking habit and hypoglycemic treatment. Subjects who were smoking regularly during the previous 12 months were classified as smokers. Pharmacological treatment was recorded at the last available visit for subjects who did not fail during the observation time and before starting insulin for subjects who failed.

Statistical Analysis

All analyses were performed using SPSS 17.0. The subjects included in the study were divided into the following 2 groups: those with beta-cell failure and those without beta-cell failure. Continuous variables were normally distributed, and then, t test for unpaired data was applied to detect differences between the 2 groups. The χ2 test was used to evaluate difference between expected and observed frequencies of hypoglycemic treatment. Medications were classified as follows: metformin or sulphonylurea in monotherapy, metformin plus sulphonylureas, incretin diabetes medications (glucagon-like peptide 1 analogs or dipeptidyl peptidase-4 inhibitors) plus metformin and/or sulphonylureas, and PPARγ agonist plus metformin and/or sulphonylureas. Survival analyses, life table, and Kaplan-Meier curve were applied to estimate the average time to failure from the date of the enrolment and from the onset of diabetes. The probability for the event from 1-year intervals, in the setting of failure from the enrolment, was also estimated. The Cox proportional hazard regression model was used to assess the relationship between progressive beta-cell failure and covariates, including diabetes-related symptoms and FPG at the onset of the disease, sex, family history of type 2 diabetes, number of visits per year, and residency, and furthermore age disease duration, BMI, eGFR, hypertension, and hyperlipidemia at the enrollment. Results have been displayed as HR and confidence intervals (CIs) 95%. The HR for continuous variables refers to each unit increase.


Two hundred twenty-four subjects met the inclusion criteria and were enrolled in the study. Of these, 91 (41%) developed beta-cell failure in the following 6 years of follow-up. Figure 1 outlines the results of the Kaplan-Meier analysis. The cumulative proportion of subjects free from beta-cell failure for each of the 6 follow-up year was 0.88, 0.83, 0.79, 0.69, 0.60, and 0.59. The estimated mean time (SE) free from the event was 4.8 (0.1)years for 6 years. The estimated mean time (SE) from event measured from the time of diabetes diagnosis was 19(1)years.


Percentage of subjects free from the progressive beta-cell failure for 6-year follow-up.

Table 1 shows the characteristics of subjects with and without progressive beta-cell failure. Disease duration was significantly higher in subjects with beta-cell failure, whereas age at the enrollment was comparable between the 2 groups. The prevalence of male sex, hypertension at the enrollment, the prevalence of subjects with both symptoms and FPG greater than 200 mg/dL at the onset of diabetes, and the prevalence of subjects living in suburban areas was significantly higher in progressive beta-cell failure group. The HbA1c at the enrollment was comparable between the groups.


Characteristics of Patients With and Without Beta-Cell Failure

The prevalence of subjects taking different agents (metformin, sulphonylureas, metformin plus sulphonylureas, incretin diabetes medications [GLP-1 analogs or DPP4 inhibitors] plus metformin and/or sulphonylureas, and PPARγ agonist plus metformin and/or sulphonylureas) divided according to presence or absence of progressive beta-cell failure is displayed in Table 2. The medical regimen is referred to last available follow-up visit for subjects who did not experience beta-cell failure and to the time of failure for subjects who experienced beta-cell failure. The prevalence of subjects taking metformin alone was significantly higher among those who did not fail, whereas the prevalence of subjects taking the combination of metformin and sulphonylureas was significantly higher among those who failed.


Percentage of Subjects Assuming Different Medication, Divided According to the Presence or Absence of Progressive Beta-Cell Failure

Table 3 shows the results of Cox proportional hazard regression model. Variables significantly influencing progressive beta-cell failure were disease duration, hypertension, and hyperlipidemia at the enrollment, presence of symptoms at the onset of the disease, and residence in suburban areas. Other variables included in the model and not significantly associated with beta-cell failure were FPG at the onset of the diabetes, family history of type 2 diabetes, number of visits per year, sex, and eGFR and age at enrollment.


Predictors of Progressive Beta-Cell Failure (Cox Proportional Hazard Regression Model)


Type 2 diabetes is a multifactorial progressive disease, and typically, the onset of the disease predates the diagnosis. Newly diagnosed subjects are instructed to modify their lifestyle and usually started on monotherapy with metformin. With disease progression insulin resistance increases and insulin delivery decreases, further hypoglycemic agents are needed. The transition to insulin therapy is usually defined as beta-cell failure. Cross sectional data have demonstrated that only 30% of orally treated patients are able to achieve HbA1c level 6.5% or less.17In the large UK Prospective Diabetes Study, less than 20% of all subjects included in the trial and on monotherapy was able to maintain advisable HbA1c at 9-year follow-up.6A national survey carried out in 2011 in a sample of approximately 550,000 subjects with type 2 diabetes revealed a 45% prevalence of subjects in good metabolic control.18The prevalence of beta-cell failure in our cohort is similar to that reported in other studies.

The possible causes of beta-cell failure are widely investigated. Medications seem to influence the rate of beta-cell failure. In the ADOPT study, monotherapy with metformin, rosiglitazone, and glyburide all failed over time, although with some difference. Specifically, at 5 years, the cumulative incidence of failure was 15% with rosiglitazone, 21% with metformin, and 34% with glyburide.7In our sample, most of the subjects who experienced progressive beta-cell failure were taking a combination of metformin and sulphonylureas. Furthermore, the comparison between subjects with and without progressive beta-cell failure revealed significant differences in sex, disease duration, presence of symptoms and FPG level at the onset, prevalence of hypertension, and residency in suburban areas. Cox regression analysis confirmed the independent statistically significant association between failure and disease duration, symptoms at the onset of the disease, hypertension, hyperlipidemia, and residency. Symptoms at the onset of disease, in absence of any condition causing sudden hyperglycemia, would suggest the presence of long-time undiagnosed diabetes or alternatively the presence of a rapidly progressing disease. Similarly, a long disease duration (∼10 years) should suggest a more intensive treatment to obtain acceptable and long-lasting metabolic control.

In the paper by Brown et al.,19both long disease duration and high HbA1c at baseline characterized subjects who experienced early metformin failure. In the same study, blood pressure, lipids, and complications were not predictive of failure.

In our study, we found more hypertensive subjects among those with beta-cell failure. Hypertension is approximately twice as frequent in subjects with diabetes, and these 2 pathological entities might exacerbate each other.20It is noteworthy that the presence of essential hypertension might be regarded as a marker of longer undetected diabetes duration. Indeed, insulin resistance and compensatory hyperinsulinemia, characterizing prediabetes, might have a role in the development of hypertension as well. As a matter of fact, it has been proposed that insulin resistance decreases the ability of insulin to promote vascular relaxation and glucose transport in vascular and skeletal muscle tissue.21Furthermore, it could be also hypothesized that some antihypertensive agents may have a significant impact in the development of diabetes.22

Abnormalities in lipid metabolism affect glucose metabolism. The excess of lipids may result in a lipid spill over to nonadipose tissue as pancreatic islet inducing cellular dysfunction and lipoapoptosis.23Furthermore, statin treatment increases the risk of type 2 diabetes in selected population, likely decreasing insulin sensitivity and secretion.24We might argue that in subjects with diabetes, hypolipidemic agents promote the progressive decline in beta-cell function.

We found that subjects living in the suburbs were prone to develop progressive beta-cell failure. There is evidence that people living in more compact area are less prone to develop obesity and diabetes. Indeed, they walk more and drive less. We might hypothesize that subjects with diabetes and living in suburb may have a worse metabolic control.25

In conclusion, the scanty available data on reliable predictors of progressive beta-cell failure might promote therapeutic inertia. Any effort toward a better identification of clinical characteristics and/or markers able to predict metabolic worsening and treatment failure is advisable. According to our results, albeit limited by the number of subjects enrolled and the study design, we would suggest to consider patients' diabetic history, comorbidities such as hypertension and hyperlipidemia at the time of the first visit to tailor therapy and obtain adequate and steady glycemic control.


The present study has some limitations that need to be discussed.

The definition of beta-cell failure was arbitrary. Unfortunately, there is no single definition of beta-cell failure. The limit of HbA1c greater that 7.5% that we have chosen seems reasonable to identify most patients who do not respond adequately to treatment with oral hypoglycemic agents. The choice of a value slightly higher or lower may have influenced the prevalence of failure, but not the identification of predictors.

The number of subjects was relatively small. However, they were all attending the same metabolic clinic and resident in the same geographical area. Therefore, they were a homogeneous group, followed and classified according to standard criteria.

Finally, the percentage of subjects taking new antidiabetic agents, as incretin medications, was low. Some recent evidence suggests a beta-cell protection action by these new drugs, but this could not be tested in the present study.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.