- Email: [email protected]
Lean versus obese diabetes mellitus patients in the United States minority population
Journal of Diabetes and Its Complications xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Journal of Diabetes and Its Complications journal homepage: WWW.JDCJOURNAL.COM
Lean versus obese diabetes mellitus patients in the United States minority population☆ Nathaniel J. Coleman a, Jadwiga Miernik a, Louis Philipson b, Leon Fogelfeld a,⁎ a b
Division of Endocrinology, John H. Stroger, Jr Hospital of Cook County and Rush University Medical Center Section of Endocrinology, University of Chicago
a r t i c l e
i n f o
Article history: Received 18 May 2013 Received in revised form 19 November 2013 Accepted 20 November 2013 Available online xxxx Keywords: Lean diabetes Pancreatitis Diabetes in minorities
a b s t r a c t Objective: To identify special characteristics in large group of lean diabetes minority patients in comparison to obese type 2 diabetes. Methods: 1784 lean (BMI b 25) diabetes patients were identified and compared with 8630 obese (BMI ≥ 30) patients. Patients with Type 1 Diabetes (N = 523) were excluded. Patient data, including demographics, psychosocial factors, insulin use, and complications was analyzed. Results: In lean compared to obese, there was male predominance (62% vs 48%, p b 0.001), higher prevalence of insulin use (49% vs 44%, p = 0.001), lower TG/HDL (2.28 vs 3.4, p b 0.001), and higher prevalence of alcoholism (5.7% vs 2.4%, p b 0.001) and pancreatitis (3.6% vs 0.9%, p b 0.001). In both groups, African Americans and Latinos were the prevalent ethnicities (38%, 34% vs. 53%,31%). When comparing patients within the lean group who were on insulin (49%) to those on oral medications, there were more males (65% vs. 59%, p b 0.001), earlier age of onset (40 ± 14 vs. 47 ± 12, p b 0.001), lower BMI (22.1 ± 2 vs.22.6 ± 1.7, p b 0.001) and lower TG/HDL (2.18 vs. 2.42, p = 0.021). Conclusions: A subset of diabetes patients in the United States minority population are lean and may have rapid beta cell failure. The etiology is not clear and acquired factors, genetics, and autoimmunity may be contributory. © 2013 Elsevier Inc. All rights reserved.
1. Introduction The worldwide prevalence of diabetes has risen drastically over the past two decades, from an estimated 30 million cases in 1985 to 177 million in 2000. Based upon current trends, greater than 360 million individuals will have diabetes by the year 2030 (Powers, 2010). This worldwide explosion of diabetes has been attributed to the increased prevalence of obesity. However, the contribution of nontraditional forms of diabetes to the increased prevalence of diabetes has not been addressed. We discovered at the Cook County Diabetes Center (CCDC), a large diabetes center serving an uninsured and underinsured minority population in Chicago, that 13% of 18,000 patients with diabetes mellitus are lean (BMI b 25 but N 17). An internet search did not identify any publications addressing this phenomenon in lean, minority patients in the United States. To determine if these lean
☆ Conflict of interest statement: We do not have any financial disclosures or conflicts of interest to report. ⁎ Corresponding author at: 1900 West Polk Street, Chicago, IL 60612. Tel.: +1 312 864 0539; fax: +1 312 864 9735. E-mail address: [email protected] (L. Fogelfeld).
diabetes patients have special clinical and pathophysiological characteristics, we compared them to obese diabetes patients after excluding the type 1 diabetes patients. 2. Methods 2.1. Design and Study Population 1784 lean (BMI b 25 but N17) diabetes patients were identified and compared with 8630 obese (BMI ≥ 30) diabetes patients. Data was obtained from the database of the CCDC. The database was initiated in the year 2000 and consists of ~ 18,000 patients for a total of 90,000 visits. The data from individual patient visits was recorded then electronically extracted and transferred to SPSS (Statistical Package for the Social Sciences) for statistical analysis. Patient data at presentation, including demographics, psychosocial factors, insulin use, and complications was analyzed. For patients in whom viable follow-up data were available, longitudinal analysis was performed. Patients were divided into two groups based upon available BMI: Lean diabetes group: BMI b 25 but N17 and obese diabetes group: ≥ 30. Patients with an intermediate BMI of 25 to 29 were excluded to allow creation of two study groups with distinct BMI differences.
1056-8727/$ – see front matter © 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jdiacomp.2013.11.010
Please cite this article as: Coleman, N.J., et al., Lean versus obese diabetes mellitus patients in the United States minority population, Journal of Diabetes and Its Complications (2014), http://dx.doi.org/10.1016/j.jdiacomp.2013.11.010
2
N.J. Coleman et al. / Journal of Diabetes and Its Complications xxx (2014) xxx–xxx
Patients with Type 1 Diabetes (N = 523) were excluded. Patients were classified as having Type 1 diabetes by their endocrinologists if they had the combination of laboratory presence of islet cell antibodies and/or clinical characteristics of young age at diagnosis, lean body mass, history of diabetic ketoacidosis, and insulin dependence at diagnosis. The presence of autoimmunity in the lean group was evaluated by reviewing the clinical charts of the lean patients for the presence of anti-islet cell antibodies including GAD-65 and IA-2.
normally distributed values and by medians. Comparisons between groups were made using t-tests. For non-parametric data, the Mann-Whitney test was used. Chi-square was used for analysis of categorical variables. For multivariate analysis, binary logistic regression was used.
2.2. Statistical Analysis
The average BMI was 22.4 ± 1.8 in the lean group compared to 37.4 ± 6.9 in the obese group (p-value b 0.001). The mean waist circumference (available in half the study group proportionately) was 85.5 cm in the lean and 114.5 cm for the obese. Within the lean, the
All analyses were performed with SPSS 21.0. (SPSS Inc., Chicago, IL). Baseline characteristics are presented as means ± SD for
3. Results 3.1. Patient Characteristics in the Lean vs Obese (Table 1)
Table 1 Lean versus Obese Diabetes Patients.
N BMI (median)a Waist circumference (median in cm)a Males/Females Central obesity (%)b Males (N102 cm)/Females (N88cm) Male/Female %b Age at Diagnosis (y) Age at Presentation (y)c Duration of DM at presentation Median (y)a % diagnosed within 1 yeara AA/Latino/Asian/Other (%)b DM in first degree relatives (%)b 3 or more 1st degree relatives (%) Insulin Use (%)b Insulin Use (%) in patients diagnosed within 1 yearb Insulin daily dose in units(median)a Insulin daily dose per kg (units/kg)c TG/HDL ratio (median)a SBP/DBP (mmHg)c AIC at presentation (%)c Complications of DMb Microvascularb Retinopathy (%) Neuropathy (%) Foot lesions (%) Albuminuria (%) Macrovascularb Peripheral vascular disease (%) Coronary Heart Disease (%) Stroke Social Factorsb Current Smoking% Males/females % Alcohol% Males/females % Drugs% Males/females % Depression% Males/females % History of Pancreatitis %b Males/females %
Lean
Obese
1784 23 22.4 ± 1.8 85.5 86.6/85.6
8630 35.6 37.4 ± 6.9 114.5 116.5/114.4
P value
3.1/37.1 62/38 43.6 ± 13.6 50.4 ± 12.9
82.7/98.2 48/52 43.6 ± 12.2 49.1 ± 11.8
P b 0.001f,g P b 0.001 NS P b 0.001
4 30 38/34/17/11 56 1.2 49 35% 38 0.6 ± 0.37 2.28 129 ± 23/74 ± 12 10.6 ± 2.9
3 34 53/31/4/12 62 1.3 44 25.8 57 0.57 ± 0.43 3.4 137 ± 22/76 ± 12 9.9 ± 2.5
P b 0.001 P b 0.001 P b 0.001 NS P = 0.001 P = 0.001 P b 0.001 NS P b 0.001 P b 0.001 P b 0.001
21.4 29.4 21.3 38
18.8 28.7 20.7 35
NS NS NS NS
5.6 8.5 2.5
6.4 11.9 2.7
NS P b 0.001 NS
30.5 28.8/16.4 5.7 8.4/1.7 3.3 3.9/2.3 4.6 4.2/5.9
22 21.4/13.6 2.4 4.2/1.0 2.6 3.8/1.7 7.1 5.5/8.8
P b 0.001
3.6 4.5/2
0.9 1.3/0.6
P b 0.001 P b 0.001f,g
d,e,f,g
P b 0.001d,e,f NSg NS d,f,g P b 0.001e P b 0.001e NSd,f P = 0.014g P b 0.001f P b 0.007 d P b 0.002e p = 0.001g
Data are % or means ± SD. a Mann–Whitney analysis (nonparametric). b Pearson’s Chi-Square (categorical variables). c Two sample t test (normally distributed). d Comparison of males vs females in lean group (Chi-Square). e Comparison of males vs females in obese group (Chi-Square). f Comparison of lean males vs obese males (Chi-Square). g Comparison of lean females vs obese females (Chi-Square).
Please cite this article as: Coleman, N.J., et al., Lean versus obese diabetes mellitus patients in the United States minority population, Journal of Diabetes and Its Complications (2014), http://dx.doi.org/10.1016/j.jdiacomp.2013.11.010
N.J. Coleman et al. / Journal of Diabetes and Its Complications xxx (2014) xxx–xxx
3
Table 2 Diabetes medications usage, changes in A1C and kidney function in a subset of patients with available longitudinal data.
Diabetes medications The group that started oral medications Started on oral medications and needed insulin addition (%)b Time in weeks until insulin added to oral medications (median)a Insulin and oral medications usage during the follow up periodb Started and stayed on oral medications (%) Started on oral medications and needed insulin addition (%) Oral medications and insulin started at the same time (%)a Started insulin and orals added later (%) Started and stayed on insulin only (%) Glycemic Control AIC at presentation (%)c A1C at last follow-up (%)c A1C at target of b7% at presentation(%)b A1C at target of b7% at last follow-up (%)b A1C follow-up time in weeks (median)a Kidney Function (median)a eGFR at presentation (mL/min/1.73 m2) eGFR at last follow-up (mL/min/1.73 m2) eGFR follow-up time (in weeks) UACR at presentation (mg/g) UACR at last follow-up (mg/g) Albuminuria at presentation (%)b Albuminuria at last follow-up (%)b UACR follow-up time in weeksa a b c
Lean
Obese
P value
21 16
15 30
P = 0.001 NS P = 0.001
46.6 12.2 11.2 3.1 26.9
50.9 9.4 17.2 4.6 17.9
10.5 ± 2.6 8.8 ± 2.5 8.7 26 70
9.9 ± 2.5 8.4 ± 2.2 12.6 29 61
P = 0.001 P = 0.001 P = 0.007 NS NS
96.5 94.5 59 18.1 20.6 41 41.9 81
96.2 96.2 56 15.6 14.7 37.6 35.5 78
NS NS NS NS P = 0.014 NS NS NS
Mann–Whitney analysis (nonparametric). Pearson’s Chi-Square (categorical variables). Two sample t test (normally distributed).
mean waist circumference for males was 86.6 cm and 85.6 for females. Within the obese group, mean waist circumference for males was 116.5 cm and 114.4 cm for females. There was a male predominance in the lean group with 62% of patients being male compared to 48% in the obese (p-value b0.001). There was no difference in the age of diagnosis between the two groups (43.6.0 ± 13.6.5 in the lean versus 43.6 ± 12.2 in the obese). At presentation to the diabetes clinic, the age of the patients was 50.4 ± 12.9 versus 49.1 ± 11.8 (p b 0.001) and median duration of diabetes was 4 years vs 3 years (p b 0.001). The percentage of patients with newly diagnosed diabetes within 1 year was 30% vs 34% (p b 0.001). African-Americans and Latinos were the predominant ethnicities in both lean and obese. African-Americans comprised 38% of lean patients versus 53% of obese patients and Latinos 34% of lean versus 31% of obese. Asians had a fivefold higher prevalence in the lean group (17% of lean versus 4% of obese). Ethnicities that did not fall into any of
these categories was classified as “other” and made up 11% of the lean patients and 12% of obese patients. When comparing lean diabetes patients to their obese counterparts, 56% of lean patients reported having a first degree relative with diabetes compared to 62% in the obese diabetes group (p b 0.001). In the lean group, 30.5% were current smokers and 5.7% had a history of alcoholism which was much higher compared to 22% and 2.4%, respectively, in the obese population (p b 0.001). There was a higher prevalence of pancreatitis in the lean diabetes patients with 3.6% of patients having a history of pancreatitis compared to 0.9% in obese patients (p b 0.001). Of lean patients, 3.3% reported illicit drug use compared to 2.6 of obese patients (NS). Depression was more common in obese patients with 7.1% of patients reporting depression compared to 4.6% in the lean group (p b 0.001). When the groups were divided by gender, the males within the lean and obese groups had higher rates of smoking, alcohol use, and history of pancreatitis compared to the
Table 3 Comparison of Lean Patients on Insulin versus Lean Patients on oral medications.
N Male/Female %b Age at Diagnosis c AA/Latino/Asian/Others (%)b DM in first degree relatives (%)b Duration of Dm at presentation (%)a BMIc A1C at presentation (%)b TG/HDL ratio (median)a Social Factorsb Smoking% Alcohol% Drugs%
Insulin Users
OAD Users
P value
869 65/35 39.8 ± 14 41/34/11/14 57 7 22.1 ± 2 10.79 ± 2.85 2.18
915 59/41 47 ± 12 34/33/22/11 56 2 22.6 ± 1.7 10.45 ± 2.89 2.42
P b 0.001 P b 0.001 P b 0.001 NS P b 0.001 P b 0.001 p = 0.025 P = 0.021
30.9 6.9 3.4
30.2 4.8 3.4
NS NS, P = 0.084 NS
Data are % or means ± SD. a Mann–Whitney analysis (nonparametric). b Pearson’s Chi-Square (categorical variables). c Two sample t test (normally distributed).
Please cite this article as: Coleman, N.J., et al., Lean versus obese diabetes mellitus patients in the United States minority population, Journal of Diabetes and Its Complications (2014), http://dx.doi.org/10.1016/j.jdiacomp.2013.11.010
4
N.J. Coleman et al. / Journal of Diabetes and Its Complications xxx (2014) xxx–xxx
Table 4 Factors associated with use of insulin in univariate and multivariate logistic regression model.
For the lean study group Age at diagnosis Being Male Having Pancreatitis BMI For the obese study group Age at diagnosis Being depressed Having Pancreatitis DM in first degree relatives
Units increase
Unadjusted OR (95% CI)
P Value
Final model Adjusted OR (95% CI)a
P Value
years Yes vs No Yes vs No kg/m2
0.96 1.29 2.61 0.911
(0.95–0.97) (1.005–1.67) (1.34–5.12) (0.85–0.97)
b0.001 0.046 0.005 0.005
0.96 1.29 2.67 0.912
(0.95–0.97) (1.013–1.64) (1.39–5.14) (0.85–0.97)
b0.001 0.039 0.003 0.006
years Yes vs No Yes vs No Number of relatives
0.968 1.308 2.024 1.19
(0.96–0.97) (1.068–1.6) (1.153–3.55) (1.15–1.27)
b0.001 0.009 0.014 b0.001
0.968 1.29 1.974 1.89
(0.96–0.97) (1.058–1.58) (1.13–3.45) (1.11–1.27)
b0.001 0.012 b0.001 b0.001
a OR (95% CI): odds ratio and 95 % confidence interval. Other variables for which the model was adjusted using the backward selection were history of smoking, alcoholism, and drug addiction.
females in the respective groups. The males in the lean group had the highest rates of smoking, alcohol use, and history of pancreatitis which was significantly higher than males in the obese group. The females in the lean group had higher rates of smoking and history of pancreatitis but not alcohol use compared to females in the obese group. Of the 1784 lean patients, 53 patients had islet cell antibodies drawn. 89% of these patients had negative antibodies. A higher percentage of lean (49%) versus obese (44%) patients required the use of insulin for glycemic control (p = 0.001). The percentage was even higher in patients with recent diagnosis of diabetes (35% vs 26%, p b .001). However, obese patients required a higher daily dose of insulin with a median daily dose of 57 units compared to 38 units in lean patients (p b 0.001). When comparing daily insulin/kg dose between the two groups, there was no difference (0.6 vs 0.57). The TG/HDL correlated with waist circumference for both females and males in both lean and obese groups. The TG/HDL was lower in the lean group as opposed to the obese with a ratio of 2.28 and 3.4, respectively (p b 0.001). Central obesity defined by NCEP as waist circumference of N102 cm and N 88 cm was present in 3.1% of lean males and 37.1% of lean females. The TG/HDL was higher in lean females with central obesity versus females without (median 2.7 vs 2.0, p b 0.01). Lean diabetes patients had lower blood pressure readings but worse glycemic control at presentation. The average systolic and diastolic blood pressure readings were 129 and 74 for lean patients versus 137 and 76 for obese patients (p b 0.001). A1C at presentation was 10.6% for lean patients and 9.9% for obese patients (p b 0.001). Coronary complications were more prevalent in the obese group (8.5% vs 11.9%, P b 0.001). There was no difference in microvascular complications when comparing the two groups. In a multivariate logistic regression analysis, the significant factors (p b 0.05) associated with the lean vs. obese group were male gender (OR 1.685, 95% CI 1.48-1.915), history of pancreatitis (OR 3.5, 95% CI 3.5-5.2), alcoholism (OR 2.2, 95% CI 1.63-2.98), and treatment with insulin (OR 1.94, 95% CI 1.05-1.35). Depression (OR 0.662, 95% CI 0.50.88) and having more first degree relatives with diabetes (OR 0.90, 95% CI 0.83-0.98) were associated with the obese group. When ethnic groups were included, the Pacific-Asians were associated with the lean group (OR 3.4, 95% CI 2.18-4.54). Longitudinal data in relation to diabetic medications, A1C, and kidney function changes over time were available for a subset of patients (Table 2). In the lean vs. the obese group that started with oral medications, a higher fraction needed insulin during follow-up (21% vs 15%) and a higher fraction started and stayed on insulin only (26.9 vs 17.9%). While the A1C at baseline was higher in the lean group and slightly higher at the end of follow-up, there was no difference in the achievement of the target A1C b7% (26% vs 29%). The kidney function progression was not significantly different between lean and obese groups except for slightly higher albumin/creatinine ratio at the end of follow-up in the lean group.
3.2. Comparison of lean patients treated with insulin vs lean patients treated with oral antidiabetic medications (Table 3) Within the lean group, we compared patients using insulin (49%) to those using oral medications. The average BMI was 22.1 ± 2 in the lean group using insulin versus 22.6 ± 1.7 in the lean group on oral meds (p b 0.001). There was also a male predominance seen when comparing these two groups as 65% of lean patients using insulin were male compared to 59% on oral medications (p b 0.001). Lean diabetes patients who required insulin were diagnosed at a younger age of 40 ± 14 as opposed to 47 ± 12 in those on oral anti-diabetes medications. Within the lean group, those patients who were using insulin had a lower TG/HDL of 2.18 compared to 2.42 in the lean patients on oral medications (p 0.021). In a multivariate logistic regression analysis (Table 4), treatment with insulin vs. oral medications in the lean group was associated with younger age at diagnosis, male gender, lower BMI, and history of pancreatits. A similar analysis in the obese group showed that treatment with insulin was associated with younger age at diagnosis, depression, pancreatitis, and having more first degree relatives with diabetes. In the lean group, when ethnic groups were included in the analysis, the Pacific-Asians were associated with the oral medications subgroup (OR 0.539, 95% CI 0.38-0.76) while African-Americans and Caucasians were associated with the insulin users subgroup (OR 1.459, 95% CI 1.3-1.64 and OR 1.858, 95% CI 1.53-2.25). 4. Discussion This study is the first description of a large cohort of lean diabetes patients in the United States minority population that do not appear to have characteristics of Type 1 Diabetes or classic monogenic diabetes as defined by young age of diagnosis and strong family history of diabetes (3 or more 1st degree relatives with diabetes). Lean diabetes populations have been described in other parts of the world, particularly underdeveloped countries (Balasubramanyam, Yajnki, et al., 2011). Lean patients who had insulin requiring but ketosis resistant diabetes in young adulthood (KRDY) were originally described as “J [Jamaica] type diabetes”, a term used to represent about 5% of Caribbean patients who could not be categorized as either insulin dependent or non-insulin dependent (Hugh-Jones, 1955). Subsequently, similar clinical syndromes were described in regions of South Asia, India, and Africa (Abu-Bakare, Taylor, Gill, & Alberti, 1986; Ahuja, 1985; Bajaj, 1982). The clinical characteristics of these patients include a history of childhood malnutrition, poor socioeconomic status, lean or “wasted” at time of presentation (BMI b18), age at onset usually b30 years, and absence of ketosis on withdrawal of insulin (Ahuja, 1980; Balasubramanyam et al., 2011). In comparison to the KRDY patients, the lean diabetes patients in this study had a higher age of diagnosis with a mean of 43.6 and a higer BMI at
Please cite this article as: Coleman, N.J., et al., Lean versus obese diabetes mellitus patients in the United States minority population, Journal of Diabetes and Its Complications (2014), http://dx.doi.org/10.1016/j.jdiacomp.2013.11.010
N.J. Coleman et al. / Journal of Diabetes and Its Complications xxx (2014) xxx–xxx
presentation. Very few patients had a BMI of b 18 and were not included in the study. When comparing lean diabetes patients to their obese counterparts in the minority population in Chicago, there is a possibility that the lean had more rapid beta cell failure than the obese. This assumption may be supported by a higher prevalence of insulin use among the lean, which is more apparent when comparing patients with recent onset of diabetes, and is also shown in multivariate analysis. In addition, lean patients using insulin were diagnosed at an earlier age compared to those lean on orals medications, which may indicate a subgroup of patients with more pronounced and earlier beta cell failure. In addition, the lean in comparison to the obese had lower TG/ HDL ratio that may indirectly be viewed as a marker for a lower hepatic insulin resistance (Boizel et al., 2000; Giannini et al., 2011; Gonzalez-Chavez et al., 2011). Among the lean females, ~ 40% had central obesity as defined by waist circumference. St-Onge, Janssen, and Heymsfield (2004) found that waist circumference was a good indicator of insulin resistance and the development of the metabolic syndrome, particularly in nonobese patients. These nonobese patients with increased waist circumference were defined as metabolically obese normal weight (MONW). In this study, the lean females with central obesity had higher TG/HDL ratio in comparison to their counterparts without central obesity. One can postulate that insulin resistance may be playing some contributory role to the development of diabetes in lean females with central obesity. However, in males that constituted the majority of the lean group, insulin resistance does not appear to play such an important role since most (96.9%) did not have central obesity. Therefore, one can postulate that the possible mechanism of diabetes development in the majority of lean patients appears to reside in failure of the beta cells. This may also explain the higher use of insulin observed in the follow-up analysis of a subset of patients in the lean group who started on oral diabetes medications as compared to the obese. There were not striking differences in follow-up between the groups in relation to glycemic control and kidney function. The reason for the putative rapid beta cell failure in the lean diabetes patients is unknown. In the KRDY population, it has been suggested that malnutrition and intrauterine growth retardation could affect beta cell development and function (Balasubramanyam et al., 2011; Jones & Ozanne, 2009). A study that examined women exposed to the Dutch famine during World War II found that those who experienced moderate and severe famine during childhood or adolescence had increased risk for developing adult-onset type 2 diabetes compared with unexposed women (Van Abeeleen et al., 2012). In this study, the history of childhood malnutrition is not available in our lean diabetes mellitus population. However, many of our patients are from low socioecomic status and similar factors of food deprivation during the intrauterine and postnatal periods may have played a role in the development of diabetes. Other acquired factors may have contributed to beta cell failure in this group. There was a higher prevalence of alcoholism and pancreatitis in the lean group. The difference in alcohol consumption was two-fold higher in the lean group, which is likely an underestimation due to underreporting. Kim et al. (2012) have recently reported that chronic ethanol consumption induced pancreatic β-cell dysfunction and apoptosis. In addition, there was a higher prevalence of smoking in the lean group. A study conducted by Zhang et al. (2011) suggests that exposure to passive and active smoking are positively and independently associated with the risk of diabetes. The prevalence of males in the lean group, particularly those using insulin, was much higher in comparison to obese diabetes patients. Gale and Gillespie (2001) have reported that although Type 2 diabetes showed a pronounced female excess in the first half of the last century it is now equally prevalent among men and
5
women in most populations, with some evidence of male preponderance in early middle age. Whether the pronounced male prevalence among the lean is due to inherent genetic differences or unhealthy behaviors that can promote beta cell failure in comparison to females is unknown. In this study, the highest prevalence of alcoholism was in the lean males at 8.5% and the lowest was 1% in obese females. In a study of ketosis prone diabetes (KPD) with insulin deficiency in West Africans, there was a strong predominance of males and association with the beta cell related gene variant, NGN3 199F (Louet et al., 2008). In another study there was a male predominance in “unprovoked” KPD patients with preserved beta cell function (Nalini et al., 2010). In a recent study of autoantibody-negative KPD patients, stronger islet-specific T-cell response and higher percentages of proinflammatory monocytes were found in 7 patients that may have some similarity to the lean patients described here (Brooks-Worrell et al., 2013). Among the ethnic groups, there was a higher prevalence of Asians in the lean group while the other groups had similar distribution. One potential explanation is that the BMI definition in the Asians may underepresent overweight and obesity Wang et al., 1994. Thus the “lean Asians” in our study were behaving like obese. In fact there was a strong association with use of oral medications in this group. In addition to acquired and behavioral factors, genetic factors as a cause of diabetes is a consideration. It is possible that a small percentage of patients in the study have MODY (Philipson, Murphy, Ellard, et al., 2010). However, it is unlikely that MODY accounts for diabetes for the majority of these patients given the higher age of diagnosis and higher need for insulin. Perry et al. (2012) recently found significant differences in genetic enrichment between lean and obese patients with type 2 diabetes from European descent, suggesting that leaner patients have a larger genetic disposition to diabetes compared with their heavier counterparts. They found that genetic variants near the LAMA1 gene were associated with type 2 diabetes in leaner individuals only. Our analysis had several limitations. First, islet cell antibodies were analyzed in only a small percentage of the lean patients. Therefore, we can not exclude that an immune mediated process is contributing to diabetes in these patients. However, it is unlikely that the majority of the patients in this group have Type 1 Diabetes or LADA given the older age of diagnosis, strong family history of diabetes, high proportion of patients using only oral medications, and negative antibodies in the small group of patients tested. Second, information regarding psychosocial factors, such as alcoholism, was obtained through history taking. Thus, the overall prevalence of alcoholism may have been underreported by patients. Despite this, the prevalence of alcoholism was more than two-fold higher in the lean group which may be a pathogenic factor in the development of lean diabetes. Another limitation is the use of the TG/HDL as a measure of insulin resistance. This measure may be less accurate in patients on a high carbohydrate diet and with elevated A1C. Finally, there is a lack of experimental data in describing the pathogenesis of diabetes in these patients. The pathogenic hypotheses proposed in this study are relying mainly on clinical and laboratory inferences. Additional studies are needed to characterize diabetes in this group of patients, including perinatal history, hemodynamic, and genetic testing. In conclusion, a large subset of lean patients, not yet described in the US minority population, may represent a variant of Type 2 Diabetes where rapid beta cell failure appears to be the predominant mechanism. Acquired factors, genetics, and autoimmunity may be contributing to beta cell dysfunction. Identifying and understanding non-traditional factors that lead to diabetes would be particularly relevant to the United States minority population and could lead to innovative ways to prevent and treat diabetes in these patients.
Please cite this article as: Coleman, N.J., et al., Lean versus obese diabetes mellitus patients in the United States minority population, Journal of Diabetes and Its Complications (2014), http://dx.doi.org/10.1016/j.jdiacomp.2013.11.010
6
N.J. Coleman et al. / Journal of Diabetes and Its Complications xxx (2014) xxx–xxx
Acknowledgements The abstract of this manuscript was published in abstract form and presented at the ADA 72 Annual Scientific Session in Philadelphia in June 2012 (Coleman, Miernik, & Fogelfeld, 2012). References Abu-Bakare, A., Taylor, R., Gill, G. V., & Alberti, K. G. (1986). Tropical or malnutritionrelated diabetes: a real syndrome? Lancet, I(8490), 1135. Ahuja, M. M. (1980). Diabetes-special problems in developing countries. Bulletin: Delivery of Health Care for Diabetics in Developing Countries, 1, 5–6. Ahuja, M. M. (1985). Heterogeneity in tropical pancreatic diabetes. Diabetologia, 28, 708. Bajaj, J. S. (1982). Diabetes mellitus the third dimension. In E. N. Mngola (Ed.), Diabetes (pp. 11–17). Amsterdam: Excerpta Medica. Balasubramanyam, A., Yajnki, C. S., & Tandon, N. (2011). Non-Traditional Forms of Diabetes Worldwide: Implications for Translational Investigation. In R. P. Robertson, & A. C. Powers (Eds.), Translational Endocrinology and Metabolism: Type 2 Diabetes Update (pp. 43–67). Chevy Chase: The Endocrine Society. Boizel, R., Benhamou, P. Y., Lardy, B., Laporte, F., Foulon, T., & Halimi, S. (2000). Ratio of Triglycerides to HDL Cholesterol Is an Indicator of LDL Particle Size in Patients With Type 2 Diabetes and Normal HDL Cholesterol Levels. Diabetes Care, 23, 1679–1685. Brooks-Worrell, B. M., Iyer, D., Coraza, I., Hampe, C. S., Nalini, R., Ozer, K., et al. (2013). Isletspecific T-cell responses and proinflammatory monocytes define subtypes of autoantibody-negative ketosis-prone diabetes (KPD). Diabetes Care, 36, 4098–4103. Coleman, N., Miernik, J., & Fogelfeld, L. (June 8-12th). High Prevalence of Lean NonClassical Type 1 and Type 2 Diabetes Patients with Special Characteristics. American Diabetes Association 72nd Annual Scientific Sessions (Philadelphia, PA). Gale, E. A., & Gillespie, K. M. (2001). Diabetes and Gender. Diabetologia, 44(1), 3–15. Giannini, C., Santoro, N., Caprio, S., Kim, G., Lartaud, D., Shaw, M., et al. (2011). The Triglyceride-to-HDL Cholesterol Ratio Association with insulin resistance in obese youths of different ethnic backgrounds. Diabetes Care, 34(8), 1869–1874. González-Chávez, A., Simental-Mendía, L. E., & Elizondo-Argueta, S. (2011). Elevated triglyceride/HDL ratio associated with insulin resistance. Cirugia y Cirujanos, 79, 115–119.
Hugh-Jones, P. (1955). Diabetes in Jamaica. Lancet, I, 891–897. Jones, R. H., & Ozanne, S. E. (2009). Fetal programming of glucose-insulin metabolism. Molecular and Cellular Endocrinology, 297(1–2), 4–9. Kim, J. Y., Kang, J. K., Park, K. J., Kim, J. E., Kim, D. H., Song, J., et al. (June 8-12th). Role of Activating Transcriptional Factors-3 on the Induction of Type 2 Diabetes via the Impairment of Glucose Metabolism in the Chronic Ethanol-Fed Mice. ADA 72nd Scientific Sessions (Philadelphia, PA). Louet, J. F., Smith, S. B., Gautier, J. F., Molokhia, M., Virally, M. L., Kevorkian, J. P., et al. (2008). Gender and neurogenin3 influence the pathogenesis of ketosis-prone diabetes. Diabetes, Obesity and Metabolism, 10, 912–920. Nalini, R., Ozer, K., Maldonado, M., Patel, S. G., Hampe, C. S., Guthikonda, A., et al. (2010). Presence or absence of a known DKA precipitant defines distinct syndromes of “A-β+” Ketosis-Prone Diabetes (KPD) based on long-term beta cell function, HLA class II alleles, and gender predilection. Metabolism, 59(10), 1448–1455. Perry, J. R., Voight, B. F., Yengo, L., Amin, N., Dupuis, J., Ganser, M., et al. (2012). Stratifying Type 2 Diabetes Cases by BMI Identifies Genetic Risk Variants in LAMA1 and Enrichment for Risk Variants in Lean Compared to Obese Cases. PLoS Genetics, 8(5), 1–14. Philipson, L. H., Murphy, R., Ellard, S., Hattersley, A. T., Støy, J., Greeley, S. A., et al. (2010). Genetic testing in diabetes mellitus: a clinical guide to monogenic diabetes. In R. E. Weiss, & S. Refetoff (Eds.), Genetic Diagnosis of Endocrine Disorders (pp. 17–25). London: Elsevier. Powers, A. C. (2010). Diabetes Mellitus. In J. L. Jameson (Ed.), Harrison’s Endocrinology (pp. 267–270). New York: McGraw Hill Medical. St-Onge, M. P., Janssen, I., & Heymsfield, S. B. (2004). Metabolic Syndrome in NormalWeight Americans. Diabetes Care, 27(9), 222–227. van Abeelen, A. F., Elias, S. G., Bossuyt, P. M., Grobbee, D. E., van der Schouw, Y. T., Roseboom, T. J., et al. (2012). Famine Exposure in the Young and the Risk of Type 2 Diabetes in Adulthood. Diabetes, 61, 2255–2260. Wang, J., Thornton, J. C., Russell, M., Burastero, S., Heymsfield, S., & Pierson, R. N., Jr. (1994). Comparison of BMI and anthropometric measures among South Asian Indians using standard and modified criteria. American Journal of Clinical Nutrition, 60, 23–28. Zhang, L., Curhan, G. C., Hu, F. B., Rimm, E. B., & Forman, J. P. (2011). Association between Passive and Active Smoking and Incident Type 2 Diabetes in Women. Diabetes Care, 34(4), 892–897.
Please cite this article as: Coleman, N.J., et al., Lean versus obese diabetes mellitus patients in the United States minority population, Journal of Diabetes and Its Complications (2014), http://dx.doi.org/10.1016/j.jdiacomp.2013.11.010
Contents lists available at ScienceDirect
Journal of Diabetes and Its Complications journal homepage: WWW.JDCJOURNAL.COM
Lean versus obese diabetes mellitus patients in the United States minority population☆ Nathaniel J. Coleman a, Jadwiga Miernik a, Louis Philipson b, Leon Fogelfeld a,⁎ a b
Division of Endocrinology, John H. Stroger, Jr Hospital of Cook County and Rush University Medical Center Section of Endocrinology, University of Chicago
a r t i c l e
i n f o
Article history: Received 18 May 2013 Received in revised form 19 November 2013 Accepted 20 November 2013 Available online xxxx Keywords: Lean diabetes Pancreatitis Diabetes in minorities
a b s t r a c t Objective: To identify special characteristics in large group of lean diabetes minority patients in comparison to obese type 2 diabetes. Methods: 1784 lean (BMI b 25) diabetes patients were identified and compared with 8630 obese (BMI ≥ 30) patients. Patients with Type 1 Diabetes (N = 523) were excluded. Patient data, including demographics, psychosocial factors, insulin use, and complications was analyzed. Results: In lean compared to obese, there was male predominance (62% vs 48%, p b 0.001), higher prevalence of insulin use (49% vs 44%, p = 0.001), lower TG/HDL (2.28 vs 3.4, p b 0.001), and higher prevalence of alcoholism (5.7% vs 2.4%, p b 0.001) and pancreatitis (3.6% vs 0.9%, p b 0.001). In both groups, African Americans and Latinos were the prevalent ethnicities (38%, 34% vs. 53%,31%). When comparing patients within the lean group who were on insulin (49%) to those on oral medications, there were more males (65% vs. 59%, p b 0.001), earlier age of onset (40 ± 14 vs. 47 ± 12, p b 0.001), lower BMI (22.1 ± 2 vs.22.6 ± 1.7, p b 0.001) and lower TG/HDL (2.18 vs. 2.42, p = 0.021). Conclusions: A subset of diabetes patients in the United States minority population are lean and may have rapid beta cell failure. The etiology is not clear and acquired factors, genetics, and autoimmunity may be contributory. © 2013 Elsevier Inc. All rights reserved.
1. Introduction The worldwide prevalence of diabetes has risen drastically over the past two decades, from an estimated 30 million cases in 1985 to 177 million in 2000. Based upon current trends, greater than 360 million individuals will have diabetes by the year 2030 (Powers, 2010). This worldwide explosion of diabetes has been attributed to the increased prevalence of obesity. However, the contribution of nontraditional forms of diabetes to the increased prevalence of diabetes has not been addressed. We discovered at the Cook County Diabetes Center (CCDC), a large diabetes center serving an uninsured and underinsured minority population in Chicago, that 13% of 18,000 patients with diabetes mellitus are lean (BMI b 25 but N 17). An internet search did not identify any publications addressing this phenomenon in lean, minority patients in the United States. To determine if these lean
☆ Conflict of interest statement: We do not have any financial disclosures or conflicts of interest to report. ⁎ Corresponding author at: 1900 West Polk Street, Chicago, IL 60612. Tel.: +1 312 864 0539; fax: +1 312 864 9735. E-mail address: [email protected] (L. Fogelfeld).
diabetes patients have special clinical and pathophysiological characteristics, we compared them to obese diabetes patients after excluding the type 1 diabetes patients. 2. Methods 2.1. Design and Study Population 1784 lean (BMI b 25 but N17) diabetes patients were identified and compared with 8630 obese (BMI ≥ 30) diabetes patients. Data was obtained from the database of the CCDC. The database was initiated in the year 2000 and consists of ~ 18,000 patients for a total of 90,000 visits. The data from individual patient visits was recorded then electronically extracted and transferred to SPSS (Statistical Package for the Social Sciences) for statistical analysis. Patient data at presentation, including demographics, psychosocial factors, insulin use, and complications was analyzed. For patients in whom viable follow-up data were available, longitudinal analysis was performed. Patients were divided into two groups based upon available BMI: Lean diabetes group: BMI b 25 but N17 and obese diabetes group: ≥ 30. Patients with an intermediate BMI of 25 to 29 were excluded to allow creation of two study groups with distinct BMI differences.
1056-8727/$ – see front matter © 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jdiacomp.2013.11.010
Please cite this article as: Coleman, N.J., et al., Lean versus obese diabetes mellitus patients in the United States minority population, Journal of Diabetes and Its Complications (2014), http://dx.doi.org/10.1016/j.jdiacomp.2013.11.010
2
N.J. Coleman et al. / Journal of Diabetes and Its Complications xxx (2014) xxx–xxx
Patients with Type 1 Diabetes (N = 523) were excluded. Patients were classified as having Type 1 diabetes by their endocrinologists if they had the combination of laboratory presence of islet cell antibodies and/or clinical characteristics of young age at diagnosis, lean body mass, history of diabetic ketoacidosis, and insulin dependence at diagnosis. The presence of autoimmunity in the lean group was evaluated by reviewing the clinical charts of the lean patients for the presence of anti-islet cell antibodies including GAD-65 and IA-2.
normally distributed values and by medians. Comparisons between groups were made using t-tests. For non-parametric data, the Mann-Whitney test was used. Chi-square was used for analysis of categorical variables. For multivariate analysis, binary logistic regression was used.
2.2. Statistical Analysis
The average BMI was 22.4 ± 1.8 in the lean group compared to 37.4 ± 6.9 in the obese group (p-value b 0.001). The mean waist circumference (available in half the study group proportionately) was 85.5 cm in the lean and 114.5 cm for the obese. Within the lean, the
All analyses were performed with SPSS 21.0. (SPSS Inc., Chicago, IL). Baseline characteristics are presented as means ± SD for
3. Results 3.1. Patient Characteristics in the Lean vs Obese (Table 1)
Table 1 Lean versus Obese Diabetes Patients.
N BMI (median)a Waist circumference (median in cm)a Males/Females Central obesity (%)b Males (N102 cm)/Females (N88cm) Male/Female %b Age at Diagnosis (y) Age at Presentation (y)c Duration of DM at presentation Median (y)a % diagnosed within 1 yeara AA/Latino/Asian/Other (%)b DM in first degree relatives (%)b 3 or more 1st degree relatives (%) Insulin Use (%)b Insulin Use (%) in patients diagnosed within 1 yearb Insulin daily dose in units(median)a Insulin daily dose per kg (units/kg)c TG/HDL ratio (median)a SBP/DBP (mmHg)c AIC at presentation (%)c Complications of DMb Microvascularb Retinopathy (%) Neuropathy (%) Foot lesions (%) Albuminuria (%) Macrovascularb Peripheral vascular disease (%) Coronary Heart Disease (%) Stroke Social Factorsb Current Smoking% Males/females % Alcohol% Males/females % Drugs% Males/females % Depression% Males/females % History of Pancreatitis %b Males/females %
Lean
Obese
1784 23 22.4 ± 1.8 85.5 86.6/85.6
8630 35.6 37.4 ± 6.9 114.5 116.5/114.4
P value
3.1/37.1 62/38 43.6 ± 13.6 50.4 ± 12.9
82.7/98.2 48/52 43.6 ± 12.2 49.1 ± 11.8
P b 0.001f,g P b 0.001 NS P b 0.001
4 30 38/34/17/11 56 1.2 49 35% 38 0.6 ± 0.37 2.28 129 ± 23/74 ± 12 10.6 ± 2.9
3 34 53/31/4/12 62 1.3 44 25.8 57 0.57 ± 0.43 3.4 137 ± 22/76 ± 12 9.9 ± 2.5
P b 0.001 P b 0.001 P b 0.001 NS P = 0.001 P = 0.001 P b 0.001 NS P b 0.001 P b 0.001 P b 0.001
21.4 29.4 21.3 38
18.8 28.7 20.7 35
NS NS NS NS
5.6 8.5 2.5
6.4 11.9 2.7
NS P b 0.001 NS
30.5 28.8/16.4 5.7 8.4/1.7 3.3 3.9/2.3 4.6 4.2/5.9
22 21.4/13.6 2.4 4.2/1.0 2.6 3.8/1.7 7.1 5.5/8.8
P b 0.001
3.6 4.5/2
0.9 1.3/0.6
P b 0.001 P b 0.001f,g
d,e,f,g
P b 0.001d,e,f NSg NS d,f,g P b 0.001e P b 0.001e NSd,f P = 0.014g P b 0.001f P b 0.007 d P b 0.002e p = 0.001g
Data are % or means ± SD. a Mann–Whitney analysis (nonparametric). b Pearson’s Chi-Square (categorical variables). c Two sample t test (normally distributed). d Comparison of males vs females in lean group (Chi-Square). e Comparison of males vs females in obese group (Chi-Square). f Comparison of lean males vs obese males (Chi-Square). g Comparison of lean females vs obese females (Chi-Square).
Please cite this article as: Coleman, N.J., et al., Lean versus obese diabetes mellitus patients in the United States minority population, Journal of Diabetes and Its Complications (2014), http://dx.doi.org/10.1016/j.jdiacomp.2013.11.010
N.J. Coleman et al. / Journal of Diabetes and Its Complications xxx (2014) xxx–xxx
3
Table 2 Diabetes medications usage, changes in A1C and kidney function in a subset of patients with available longitudinal data.
Diabetes medications The group that started oral medications Started on oral medications and needed insulin addition (%)b Time in weeks until insulin added to oral medications (median)a Insulin and oral medications usage during the follow up periodb Started and stayed on oral medications (%) Started on oral medications and needed insulin addition (%) Oral medications and insulin started at the same time (%)a Started insulin and orals added later (%) Started and stayed on insulin only (%) Glycemic Control AIC at presentation (%)c A1C at last follow-up (%)c A1C at target of b7% at presentation(%)b A1C at target of b7% at last follow-up (%)b A1C follow-up time in weeks (median)a Kidney Function (median)a eGFR at presentation (mL/min/1.73 m2) eGFR at last follow-up (mL/min/1.73 m2) eGFR follow-up time (in weeks) UACR at presentation (mg/g) UACR at last follow-up (mg/g) Albuminuria at presentation (%)b Albuminuria at last follow-up (%)b UACR follow-up time in weeksa a b c
Lean
Obese
P value
21 16
15 30
P = 0.001 NS P = 0.001
46.6 12.2 11.2 3.1 26.9
50.9 9.4 17.2 4.6 17.9
10.5 ± 2.6 8.8 ± 2.5 8.7 26 70
9.9 ± 2.5 8.4 ± 2.2 12.6 29 61
P = 0.001 P = 0.001 P = 0.007 NS NS
96.5 94.5 59 18.1 20.6 41 41.9 81
96.2 96.2 56 15.6 14.7 37.6 35.5 78
NS NS NS NS P = 0.014 NS NS NS
Mann–Whitney analysis (nonparametric). Pearson’s Chi-Square (categorical variables). Two sample t test (normally distributed).
mean waist circumference for males was 86.6 cm and 85.6 for females. Within the obese group, mean waist circumference for males was 116.5 cm and 114.4 cm for females. There was a male predominance in the lean group with 62% of patients being male compared to 48% in the obese (p-value b0.001). There was no difference in the age of diagnosis between the two groups (43.6.0 ± 13.6.5 in the lean versus 43.6 ± 12.2 in the obese). At presentation to the diabetes clinic, the age of the patients was 50.4 ± 12.9 versus 49.1 ± 11.8 (p b 0.001) and median duration of diabetes was 4 years vs 3 years (p b 0.001). The percentage of patients with newly diagnosed diabetes within 1 year was 30% vs 34% (p b 0.001). African-Americans and Latinos were the predominant ethnicities in both lean and obese. African-Americans comprised 38% of lean patients versus 53% of obese patients and Latinos 34% of lean versus 31% of obese. Asians had a fivefold higher prevalence in the lean group (17% of lean versus 4% of obese). Ethnicities that did not fall into any of
these categories was classified as “other” and made up 11% of the lean patients and 12% of obese patients. When comparing lean diabetes patients to their obese counterparts, 56% of lean patients reported having a first degree relative with diabetes compared to 62% in the obese diabetes group (p b 0.001). In the lean group, 30.5% were current smokers and 5.7% had a history of alcoholism which was much higher compared to 22% and 2.4%, respectively, in the obese population (p b 0.001). There was a higher prevalence of pancreatitis in the lean diabetes patients with 3.6% of patients having a history of pancreatitis compared to 0.9% in obese patients (p b 0.001). Of lean patients, 3.3% reported illicit drug use compared to 2.6 of obese patients (NS). Depression was more common in obese patients with 7.1% of patients reporting depression compared to 4.6% in the lean group (p b 0.001). When the groups were divided by gender, the males within the lean and obese groups had higher rates of smoking, alcohol use, and history of pancreatitis compared to the
Table 3 Comparison of Lean Patients on Insulin versus Lean Patients on oral medications.
N Male/Female %b Age at Diagnosis c AA/Latino/Asian/Others (%)b DM in first degree relatives (%)b Duration of Dm at presentation (%)a BMIc A1C at presentation (%)b TG/HDL ratio (median)a Social Factorsb Smoking% Alcohol% Drugs%
Insulin Users
OAD Users
P value
869 65/35 39.8 ± 14 41/34/11/14 57 7 22.1 ± 2 10.79 ± 2.85 2.18
915 59/41 47 ± 12 34/33/22/11 56 2 22.6 ± 1.7 10.45 ± 2.89 2.42
P b 0.001 P b 0.001 P b 0.001 NS P b 0.001 P b 0.001 p = 0.025 P = 0.021
30.9 6.9 3.4
30.2 4.8 3.4
NS NS, P = 0.084 NS
Data are % or means ± SD. a Mann–Whitney analysis (nonparametric). b Pearson’s Chi-Square (categorical variables). c Two sample t test (normally distributed).
Please cite this article as: Coleman, N.J., et al., Lean versus obese diabetes mellitus patients in the United States minority population, Journal of Diabetes and Its Complications (2014), http://dx.doi.org/10.1016/j.jdiacomp.2013.11.010
4
N.J. Coleman et al. / Journal of Diabetes and Its Complications xxx (2014) xxx–xxx
Table 4 Factors associated with use of insulin in univariate and multivariate logistic regression model.
For the lean study group Age at diagnosis Being Male Having Pancreatitis BMI For the obese study group Age at diagnosis Being depressed Having Pancreatitis DM in first degree relatives
Units increase
Unadjusted OR (95% CI)
P Value
Final model Adjusted OR (95% CI)a
P Value
years Yes vs No Yes vs No kg/m2
0.96 1.29 2.61 0.911
(0.95–0.97) (1.005–1.67) (1.34–5.12) (0.85–0.97)
b0.001 0.046 0.005 0.005
0.96 1.29 2.67 0.912
(0.95–0.97) (1.013–1.64) (1.39–5.14) (0.85–0.97)
b0.001 0.039 0.003 0.006
years Yes vs No Yes vs No Number of relatives
0.968 1.308 2.024 1.19
(0.96–0.97) (1.068–1.6) (1.153–3.55) (1.15–1.27)
b0.001 0.009 0.014 b0.001
0.968 1.29 1.974 1.89
(0.96–0.97) (1.058–1.58) (1.13–3.45) (1.11–1.27)
b0.001 0.012 b0.001 b0.001
a OR (95% CI): odds ratio and 95 % confidence interval. Other variables for which the model was adjusted using the backward selection were history of smoking, alcoholism, and drug addiction.
females in the respective groups. The males in the lean group had the highest rates of smoking, alcohol use, and history of pancreatitis which was significantly higher than males in the obese group. The females in the lean group had higher rates of smoking and history of pancreatitis but not alcohol use compared to females in the obese group. Of the 1784 lean patients, 53 patients had islet cell antibodies drawn. 89% of these patients had negative antibodies. A higher percentage of lean (49%) versus obese (44%) patients required the use of insulin for glycemic control (p = 0.001). The percentage was even higher in patients with recent diagnosis of diabetes (35% vs 26%, p b .001). However, obese patients required a higher daily dose of insulin with a median daily dose of 57 units compared to 38 units in lean patients (p b 0.001). When comparing daily insulin/kg dose between the two groups, there was no difference (0.6 vs 0.57). The TG/HDL correlated with waist circumference for both females and males in both lean and obese groups. The TG/HDL was lower in the lean group as opposed to the obese with a ratio of 2.28 and 3.4, respectively (p b 0.001). Central obesity defined by NCEP as waist circumference of N102 cm and N 88 cm was present in 3.1% of lean males and 37.1% of lean females. The TG/HDL was higher in lean females with central obesity versus females without (median 2.7 vs 2.0, p b 0.01). Lean diabetes patients had lower blood pressure readings but worse glycemic control at presentation. The average systolic and diastolic blood pressure readings were 129 and 74 for lean patients versus 137 and 76 for obese patients (p b 0.001). A1C at presentation was 10.6% for lean patients and 9.9% for obese patients (p b 0.001). Coronary complications were more prevalent in the obese group (8.5% vs 11.9%, P b 0.001). There was no difference in microvascular complications when comparing the two groups. In a multivariate logistic regression analysis, the significant factors (p b 0.05) associated with the lean vs. obese group were male gender (OR 1.685, 95% CI 1.48-1.915), history of pancreatitis (OR 3.5, 95% CI 3.5-5.2), alcoholism (OR 2.2, 95% CI 1.63-2.98), and treatment with insulin (OR 1.94, 95% CI 1.05-1.35). Depression (OR 0.662, 95% CI 0.50.88) and having more first degree relatives with diabetes (OR 0.90, 95% CI 0.83-0.98) were associated with the obese group. When ethnic groups were included, the Pacific-Asians were associated with the lean group (OR 3.4, 95% CI 2.18-4.54). Longitudinal data in relation to diabetic medications, A1C, and kidney function changes over time were available for a subset of patients (Table 2). In the lean vs. the obese group that started with oral medications, a higher fraction needed insulin during follow-up (21% vs 15%) and a higher fraction started and stayed on insulin only (26.9 vs 17.9%). While the A1C at baseline was higher in the lean group and slightly higher at the end of follow-up, there was no difference in the achievement of the target A1C b7% (26% vs 29%). The kidney function progression was not significantly different between lean and obese groups except for slightly higher albumin/creatinine ratio at the end of follow-up in the lean group.
3.2. Comparison of lean patients treated with insulin vs lean patients treated with oral antidiabetic medications (Table 3) Within the lean group, we compared patients using insulin (49%) to those using oral medications. The average BMI was 22.1 ± 2 in the lean group using insulin versus 22.6 ± 1.7 in the lean group on oral meds (p b 0.001). There was also a male predominance seen when comparing these two groups as 65% of lean patients using insulin were male compared to 59% on oral medications (p b 0.001). Lean diabetes patients who required insulin were diagnosed at a younger age of 40 ± 14 as opposed to 47 ± 12 in those on oral anti-diabetes medications. Within the lean group, those patients who were using insulin had a lower TG/HDL of 2.18 compared to 2.42 in the lean patients on oral medications (p 0.021). In a multivariate logistic regression analysis (Table 4), treatment with insulin vs. oral medications in the lean group was associated with younger age at diagnosis, male gender, lower BMI, and history of pancreatits. A similar analysis in the obese group showed that treatment with insulin was associated with younger age at diagnosis, depression, pancreatitis, and having more first degree relatives with diabetes. In the lean group, when ethnic groups were included in the analysis, the Pacific-Asians were associated with the oral medications subgroup (OR 0.539, 95% CI 0.38-0.76) while African-Americans and Caucasians were associated with the insulin users subgroup (OR 1.459, 95% CI 1.3-1.64 and OR 1.858, 95% CI 1.53-2.25). 4. Discussion This study is the first description of a large cohort of lean diabetes patients in the United States minority population that do not appear to have characteristics of Type 1 Diabetes or classic monogenic diabetes as defined by young age of diagnosis and strong family history of diabetes (3 or more 1st degree relatives with diabetes). Lean diabetes populations have been described in other parts of the world, particularly underdeveloped countries (Balasubramanyam, Yajnki, et al., 2011). Lean patients who had insulin requiring but ketosis resistant diabetes in young adulthood (KRDY) were originally described as “J [Jamaica] type diabetes”, a term used to represent about 5% of Caribbean patients who could not be categorized as either insulin dependent or non-insulin dependent (Hugh-Jones, 1955). Subsequently, similar clinical syndromes were described in regions of South Asia, India, and Africa (Abu-Bakare, Taylor, Gill, & Alberti, 1986; Ahuja, 1985; Bajaj, 1982). The clinical characteristics of these patients include a history of childhood malnutrition, poor socioeconomic status, lean or “wasted” at time of presentation (BMI b18), age at onset usually b30 years, and absence of ketosis on withdrawal of insulin (Ahuja, 1980; Balasubramanyam et al., 2011). In comparison to the KRDY patients, the lean diabetes patients in this study had a higher age of diagnosis with a mean of 43.6 and a higer BMI at
Please cite this article as: Coleman, N.J., et al., Lean versus obese diabetes mellitus patients in the United States minority population, Journal of Diabetes and Its Complications (2014), http://dx.doi.org/10.1016/j.jdiacomp.2013.11.010
N.J. Coleman et al. / Journal of Diabetes and Its Complications xxx (2014) xxx–xxx
presentation. Very few patients had a BMI of b 18 and were not included in the study. When comparing lean diabetes patients to their obese counterparts in the minority population in Chicago, there is a possibility that the lean had more rapid beta cell failure than the obese. This assumption may be supported by a higher prevalence of insulin use among the lean, which is more apparent when comparing patients with recent onset of diabetes, and is also shown in multivariate analysis. In addition, lean patients using insulin were diagnosed at an earlier age compared to those lean on orals medications, which may indicate a subgroup of patients with more pronounced and earlier beta cell failure. In addition, the lean in comparison to the obese had lower TG/ HDL ratio that may indirectly be viewed as a marker for a lower hepatic insulin resistance (Boizel et al., 2000; Giannini et al., 2011; Gonzalez-Chavez et al., 2011). Among the lean females, ~ 40% had central obesity as defined by waist circumference. St-Onge, Janssen, and Heymsfield (2004) found that waist circumference was a good indicator of insulin resistance and the development of the metabolic syndrome, particularly in nonobese patients. These nonobese patients with increased waist circumference were defined as metabolically obese normal weight (MONW). In this study, the lean females with central obesity had higher TG/HDL ratio in comparison to their counterparts without central obesity. One can postulate that insulin resistance may be playing some contributory role to the development of diabetes in lean females with central obesity. However, in males that constituted the majority of the lean group, insulin resistance does not appear to play such an important role since most (96.9%) did not have central obesity. Therefore, one can postulate that the possible mechanism of diabetes development in the majority of lean patients appears to reside in failure of the beta cells. This may also explain the higher use of insulin observed in the follow-up analysis of a subset of patients in the lean group who started on oral diabetes medications as compared to the obese. There were not striking differences in follow-up between the groups in relation to glycemic control and kidney function. The reason for the putative rapid beta cell failure in the lean diabetes patients is unknown. In the KRDY population, it has been suggested that malnutrition and intrauterine growth retardation could affect beta cell development and function (Balasubramanyam et al., 2011; Jones & Ozanne, 2009). A study that examined women exposed to the Dutch famine during World War II found that those who experienced moderate and severe famine during childhood or adolescence had increased risk for developing adult-onset type 2 diabetes compared with unexposed women (Van Abeeleen et al., 2012). In this study, the history of childhood malnutrition is not available in our lean diabetes mellitus population. However, many of our patients are from low socioecomic status and similar factors of food deprivation during the intrauterine and postnatal periods may have played a role in the development of diabetes. Other acquired factors may have contributed to beta cell failure in this group. There was a higher prevalence of alcoholism and pancreatitis in the lean group. The difference in alcohol consumption was two-fold higher in the lean group, which is likely an underestimation due to underreporting. Kim et al. (2012) have recently reported that chronic ethanol consumption induced pancreatic β-cell dysfunction and apoptosis. In addition, there was a higher prevalence of smoking in the lean group. A study conducted by Zhang et al. (2011) suggests that exposure to passive and active smoking are positively and independently associated with the risk of diabetes. The prevalence of males in the lean group, particularly those using insulin, was much higher in comparison to obese diabetes patients. Gale and Gillespie (2001) have reported that although Type 2 diabetes showed a pronounced female excess in the first half of the last century it is now equally prevalent among men and
5
women in most populations, with some evidence of male preponderance in early middle age. Whether the pronounced male prevalence among the lean is due to inherent genetic differences or unhealthy behaviors that can promote beta cell failure in comparison to females is unknown. In this study, the highest prevalence of alcoholism was in the lean males at 8.5% and the lowest was 1% in obese females. In a study of ketosis prone diabetes (KPD) with insulin deficiency in West Africans, there was a strong predominance of males and association with the beta cell related gene variant, NGN3 199F (Louet et al., 2008). In another study there was a male predominance in “unprovoked” KPD patients with preserved beta cell function (Nalini et al., 2010). In a recent study of autoantibody-negative KPD patients, stronger islet-specific T-cell response and higher percentages of proinflammatory monocytes were found in 7 patients that may have some similarity to the lean patients described here (Brooks-Worrell et al., 2013). Among the ethnic groups, there was a higher prevalence of Asians in the lean group while the other groups had similar distribution. One potential explanation is that the BMI definition in the Asians may underepresent overweight and obesity Wang et al., 1994. Thus the “lean Asians” in our study were behaving like obese. In fact there was a strong association with use of oral medications in this group. In addition to acquired and behavioral factors, genetic factors as a cause of diabetes is a consideration. It is possible that a small percentage of patients in the study have MODY (Philipson, Murphy, Ellard, et al., 2010). However, it is unlikely that MODY accounts for diabetes for the majority of these patients given the higher age of diagnosis and higher need for insulin. Perry et al. (2012) recently found significant differences in genetic enrichment between lean and obese patients with type 2 diabetes from European descent, suggesting that leaner patients have a larger genetic disposition to diabetes compared with their heavier counterparts. They found that genetic variants near the LAMA1 gene were associated with type 2 diabetes in leaner individuals only. Our analysis had several limitations. First, islet cell antibodies were analyzed in only a small percentage of the lean patients. Therefore, we can not exclude that an immune mediated process is contributing to diabetes in these patients. However, it is unlikely that the majority of the patients in this group have Type 1 Diabetes or LADA given the older age of diagnosis, strong family history of diabetes, high proportion of patients using only oral medications, and negative antibodies in the small group of patients tested. Second, information regarding psychosocial factors, such as alcoholism, was obtained through history taking. Thus, the overall prevalence of alcoholism may have been underreported by patients. Despite this, the prevalence of alcoholism was more than two-fold higher in the lean group which may be a pathogenic factor in the development of lean diabetes. Another limitation is the use of the TG/HDL as a measure of insulin resistance. This measure may be less accurate in patients on a high carbohydrate diet and with elevated A1C. Finally, there is a lack of experimental data in describing the pathogenesis of diabetes in these patients. The pathogenic hypotheses proposed in this study are relying mainly on clinical and laboratory inferences. Additional studies are needed to characterize diabetes in this group of patients, including perinatal history, hemodynamic, and genetic testing. In conclusion, a large subset of lean patients, not yet described in the US minority population, may represent a variant of Type 2 Diabetes where rapid beta cell failure appears to be the predominant mechanism. Acquired factors, genetics, and autoimmunity may be contributing to beta cell dysfunction. Identifying and understanding non-traditional factors that lead to diabetes would be particularly relevant to the United States minority population and could lead to innovative ways to prevent and treat diabetes in these patients.
Please cite this article as: Coleman, N.J., et al., Lean versus obese diabetes mellitus patients in the United States minority population, Journal of Diabetes and Its Complications (2014), http://dx.doi.org/10.1016/j.jdiacomp.2013.11.010
6
N.J. Coleman et al. / Journal of Diabetes and Its Complications xxx (2014) xxx–xxx
Acknowledgements The abstract of this manuscript was published in abstract form and presented at the ADA 72 Annual Scientific Session in Philadelphia in June 2012 (Coleman, Miernik, & Fogelfeld, 2012). References Abu-Bakare, A., Taylor, R., Gill, G. V., & Alberti, K. G. (1986). Tropical or malnutritionrelated diabetes: a real syndrome? Lancet, I(8490), 1135. Ahuja, M. M. (1980). Diabetes-special problems in developing countries. Bulletin: Delivery of Health Care for Diabetics in Developing Countries, 1, 5–6. Ahuja, M. M. (1985). Heterogeneity in tropical pancreatic diabetes. Diabetologia, 28, 708. Bajaj, J. S. (1982). Diabetes mellitus the third dimension. In E. N. Mngola (Ed.), Diabetes (pp. 11–17). Amsterdam: Excerpta Medica. Balasubramanyam, A., Yajnki, C. S., & Tandon, N. (2011). Non-Traditional Forms of Diabetes Worldwide: Implications for Translational Investigation. In R. P. Robertson, & A. C. Powers (Eds.), Translational Endocrinology and Metabolism: Type 2 Diabetes Update (pp. 43–67). Chevy Chase: The Endocrine Society. Boizel, R., Benhamou, P. Y., Lardy, B., Laporte, F., Foulon, T., & Halimi, S. (2000). Ratio of Triglycerides to HDL Cholesterol Is an Indicator of LDL Particle Size in Patients With Type 2 Diabetes and Normal HDL Cholesterol Levels. Diabetes Care, 23, 1679–1685. Brooks-Worrell, B. M., Iyer, D., Coraza, I., Hampe, C. S., Nalini, R., Ozer, K., et al. (2013). Isletspecific T-cell responses and proinflammatory monocytes define subtypes of autoantibody-negative ketosis-prone diabetes (KPD). Diabetes Care, 36, 4098–4103. Coleman, N., Miernik, J., & Fogelfeld, L. (June 8-12th). High Prevalence of Lean NonClassical Type 1 and Type 2 Diabetes Patients with Special Characteristics. American Diabetes Association 72nd Annual Scientific Sessions (Philadelphia, PA). Gale, E. A., & Gillespie, K. M. (2001). Diabetes and Gender. Diabetologia, 44(1), 3–15. Giannini, C., Santoro, N., Caprio, S., Kim, G., Lartaud, D., Shaw, M., et al. (2011). The Triglyceride-to-HDL Cholesterol Ratio Association with insulin resistance in obese youths of different ethnic backgrounds. Diabetes Care, 34(8), 1869–1874. González-Chávez, A., Simental-Mendía, L. E., & Elizondo-Argueta, S. (2011). Elevated triglyceride/HDL ratio associated with insulin resistance. Cirugia y Cirujanos, 79, 115–119.
Hugh-Jones, P. (1955). Diabetes in Jamaica. Lancet, I, 891–897. Jones, R. H., & Ozanne, S. E. (2009). Fetal programming of glucose-insulin metabolism. Molecular and Cellular Endocrinology, 297(1–2), 4–9. Kim, J. Y., Kang, J. K., Park, K. J., Kim, J. E., Kim, D. H., Song, J., et al. (June 8-12th). Role of Activating Transcriptional Factors-3 on the Induction of Type 2 Diabetes via the Impairment of Glucose Metabolism in the Chronic Ethanol-Fed Mice. ADA 72nd Scientific Sessions (Philadelphia, PA). Louet, J. F., Smith, S. B., Gautier, J. F., Molokhia, M., Virally, M. L., Kevorkian, J. P., et al. (2008). Gender and neurogenin3 influence the pathogenesis of ketosis-prone diabetes. Diabetes, Obesity and Metabolism, 10, 912–920. Nalini, R., Ozer, K., Maldonado, M., Patel, S. G., Hampe, C. S., Guthikonda, A., et al. (2010). Presence or absence of a known DKA precipitant defines distinct syndromes of “A-β+” Ketosis-Prone Diabetes (KPD) based on long-term beta cell function, HLA class II alleles, and gender predilection. Metabolism, 59(10), 1448–1455. Perry, J. R., Voight, B. F., Yengo, L., Amin, N., Dupuis, J., Ganser, M., et al. (2012). Stratifying Type 2 Diabetes Cases by BMI Identifies Genetic Risk Variants in LAMA1 and Enrichment for Risk Variants in Lean Compared to Obese Cases. PLoS Genetics, 8(5), 1–14. Philipson, L. H., Murphy, R., Ellard, S., Hattersley, A. T., Støy, J., Greeley, S. A., et al. (2010). Genetic testing in diabetes mellitus: a clinical guide to monogenic diabetes. In R. E. Weiss, & S. Refetoff (Eds.), Genetic Diagnosis of Endocrine Disorders (pp. 17–25). London: Elsevier. Powers, A. C. (2010). Diabetes Mellitus. In J. L. Jameson (Ed.), Harrison’s Endocrinology (pp. 267–270). New York: McGraw Hill Medical. St-Onge, M. P., Janssen, I., & Heymsfield, S. B. (2004). Metabolic Syndrome in NormalWeight Americans. Diabetes Care, 27(9), 222–227. van Abeelen, A. F., Elias, S. G., Bossuyt, P. M., Grobbee, D. E., van der Schouw, Y. T., Roseboom, T. J., et al. (2012). Famine Exposure in the Young and the Risk of Type 2 Diabetes in Adulthood. Diabetes, 61, 2255–2260. Wang, J., Thornton, J. C., Russell, M., Burastero, S., Heymsfield, S., & Pierson, R. N., Jr. (1994). Comparison of BMI and anthropometric measures among South Asian Indians using standard and modified criteria. American Journal of Clinical Nutrition, 60, 23–28. Zhang, L., Curhan, G. C., Hu, F. B., Rimm, E. B., & Forman, J. P. (2011). Association between Passive and Active Smoking and Incident Type 2 Diabetes in Women. Diabetes Care, 34(4), 892–897.
Please cite this article as: Coleman, N.J., et al., Lean versus obese diabetes mellitus patients in the United States minority population, Journal of Diabetes and Its Complications (2014), http://dx.doi.org/10.1016/j.jdiacomp.2013.11.010