DMICC Meeting on July 20 - 21, 1999 : NIDDK

DMICC Meeting on July 20 - 21, 1999

Diabetes Mellitus Interagency Coordinating Committee
Type 2 Diabetes in Children
Radisson Barcelo Hotel

Speaker Roster 


Tuesday, July 20


8:15 - 8:30 a.m.


Dr. Richard Eastman,


Dr. Charles A. Wells, Executive Secretary,


8:30 a.m. - 2:50 p.m.

Descriptive Epidemiology

Dr. Anne Fagot-Campagna, Chair

8:30 - 9:00 a.m.

Obesity in Children

Dr. Katherine Flegal

9:00 - 9:15 a.m.



9:15 - 10:00 a.m.

Hispanic Americans

Dr. Daniel Hale, Chair

10:00 - 10:15 a.m.



10:15 - 10:30 a.m.



10:30 - 11:50 a.m.

African Americans

Dr. Lawrence Dolan
Dr. Rebecca Lipton
Dr. Ingrid Libman
Dr. George Burghen

Dr. Lawrence Dolan, Chair

11:50 - 12:05 a.m.



12:05 - 1:15 a.m.



1:15 - 2:35 p.m.

Native Americans

Dr. Kelly Moore
Dr. William Knowler
Dr. Heather Dean
Dr. Valerie Cook

Dr. Kelly Moore, Chair

2:35 - 2:50 p.m.



2:50 - 5:05 p.m.

Metabolic Phenotyping

Dr. Silva Arslanian
Dr. Steven Willi
Dr. Joan DiMartino-Nardi
Dr. Philip Zeitler
Dr. Kenneth Lee Jones
Dr. William Winter

Dr. Arlan Rosenbloom, Chair

5:05 - 5:20 p.m.



Wednesday, July 21


9:00 - 9:15 p.m.

Epidemiology Summary

Dr. Anne Fagot-Campagna

9:15 - 9:30 p.m.

Phenotyping Summary

Dr. Arlan Rosenbloom

9:30 a.m. - 12:00 p.m.


Dr. Daniel Hale, Moderator

12:00 p.m.



Speaker Roster

Silva A. Arslanian, M.D.

Professor of Pediatrics
Children's Hospital of Pittsburgh
3705 Fifth Avenue
Pittsburgh, PA 15213
Phone: (412) 692-6565
Fax: (421) 692-5834

George Burghen, M.D.

Professor of Endocrinology
Pediatric Diabetes Specialist
University of Tennessee
50 N. Dunlap Street
Memphis, TN 38103
Phone: (901) 572-3292
Fax: (901) 572-3122

Valerie Cook, Ph.D.

Child and Adolescent Health Coordinator
Gila River Indian Community
Public Health Department
P.O. Box 7
Sacaton, AZ 85247
Phone: (602) 528-1231 (520) 562-3321
Fax: (602) 528-1266

Heather Dean, M.D.

Department of Pediatrics
University of Manitoba
FE-Suite 25
685 William Ave
Winnipeg, Manitoba R3E 0Z2
Phone: (204) 787-7435
Fax: (204) 787-1655

Joan DiMartino-Nardi, M.D.

Associate Professor
Pediatric Endocrinology
Montefiore Medical Center
111 E. 210th Street
Bronx, NY 10467
Phone: (718) 920-4664
Fax: (718) 405-5609

Lawerence M. Dolan, M.D.

Professor of Pediatrics
Children's Hospital Medical Center
3333 Burnet Street
Cincinnati, OH 45229
Phone: (513) 636-4744
Fax: (513) 636-7486

Anne Fagot-Campagna, M.D., Ph.D.

Medical Epidemiologist
Centers for Disease Control and Prevention
Division of Diabetes Translation
4770 Buford Highway, NE.
Atlanta, GA 30341
Phone: (770) 488-1053
Fax: (770) 488-1148

Katherine Flegal, Ph.D.

Senior Research Scientist
National Center for Health Statistics
Centers for Disease Control and Prevention
6525 Belcrest Road
Hyattsville, MD 20872
Phone: (301) 436-7075 ext. 202
Fax: (301) 436-3436

Daniel E. Hale, M.D.

Associate Professor of Pediatrics
University of Texas Health Science Center at San Antonio
Department of Pediatrics
7703 Floyd Curl Drive
San Antonio, TX 78284
Phone: (210) 567-4298
Fax: (210) /567-6921

Kenneth Lee Jones, M.D.

Professor of Pediatrics
University of California, San Diego
9500 Gilman Drive, 0831
La Jolla, CA 92093
Phone: (619) 543-5238
Fax: (619) 543-3575
E-mail :

William C. Knowler, M.D., Dr. PH

Chief, Diabetes and Arthritis Epidemiology Section
National Institute of Diabetes and Digestive and Kidney Diseases
1550 E. Indian School Road
Phoenix, AZ 85014
Phone: (602) 200-5206
Fax: (602) 200-5225

Ingrid M. Libman, M.D., Ph.D.

Resident, 2nd Year Pediatrics
Children's Hospital of Pittsburgh
3705 Fifth Avenue
Pittsburgh, PA 15213
Phone: (412) 682-2344
Fax: (412) 682-8051

Rebecca Lipton, Ph.D.

Associate Professor of Epidemiology
University of Illinois at Chicago
2121 W. Taylor Street, M/C 922
Chicago, IL 60612
Phone: (312) 413-0301
Fax: (312) 996-0064

Kelly Moore, M.D.

Area Diabetes Control Officer
Office of Healthcare Programs
Billings Area Indian Health Service
P.O. Box 2143
2900 Fourth Avenue North
Billings, MT 59103
Phone: (406) 247-7111

Arlan L. Rosenbloom, M.D.

Distinguished Service Professor Emeritus
University of Florida
Children's Medical Services Center
1701 SW. 16th Avenue
Gainesville, FL 32608
Phone: (352) 334-1393
Fax: (352) 378-3382

Steven M. Willi, M.D.

Associate Professor of Pediatrics
Medical University of South Carolina
171 Ashley Avenue
Charleston, SC 29464
Phone: (843) 792-6807
Fax: (843) 792-0548

William E. Winter, M.D.

Professor of Pathology and Pediatrics
University of Florida
1600 SW. Archer Road, Room D6-3
Gainesville, FL 32610-0275
Phone: (352) 392-4495
Fax: (352) 846-2149

Philip Scott Zeiter, M.D., Ph.D.

Assistant Professor of Pediatrics
The Children's Hospital
1056 E. 19th Avenue
Denver, CO 80218
Phone: (303) 526-5829
Fax: (303) 834-5679 


    In Attendance:

    Kelly Acton, M.D., M.P.H.
    Nell Armstrong, Ph.D., R.N.
    Sue BlissGeorge M. Bright, M.D.
    Joan Chamberlain
    Geoffrey Cheung, Ph.D.
    Valerie Cook, Ph.D.
    Jane DeMouy, Ph.D.
    Peter Dudley, Ph.D.
    Richard Eastman, M.D.
    Judy Fradkin, M.D.
    Joanne Gallivan, M.S., R.D.
    Sanford Garfield, Ph.D.
    Gilman D. Grave, M.D.
    Colleen Guay-Broder
    Joan T. Harmon, Ph.D.
    Richard Kahn, Ph.D.
    Mary Beth Kester
    N. Krish Krishnan, Ph.D.
    John Lanigan
    Barbara Linder, M.D., Ph.D.
    Mimi Lising, M.P.H.
    Louis Emmet Mahoney, M.D., Dr.P.H.
    Sandra Raff
    Peter J. Savage, M.D.
    Nancy Stancic
    Dan Stryer
    Rebecca Troisi
    Frank Vinicor, M.D., M.P.H.
    Charles A. Wells, Ph.D.
    Faye Wong, M.P.H., R.D.
    Elaine Young, Ph.D.


    Silva A. Arslanian, M.D.
    Children's Hospital of Pittsburgh

    George Burghen, M.D.
    University of Tennessee

    Heather J. Dean, M.D.
    University of Manitoba

    Joan DiMartino-Nardi, M.D.
    Montefiore Medical Center

    Lawrence M. Dolan, M.D.
    Children's Hospital Medical Center of Cincinnati

    Anne Fagot-Campagna, M.D., Ph.D.
    Centers for Disease Control and Prevention

    Carol Feld
    National Institute of Diabetes and Digestive and Kidney Diseases

    Katherine Flegal, Ph.D.
    Centers for Disease Control and Prevention

    Daniel E. Hale, M.D.
    University of Texas Health Science Center at San Antonio

    Kenneth Lee Jones, M.D.
    University of California, San Diego

    William C. Knowler, M.D., Dr.P.H.
    National Institute of Diabetes and Digestive and Kidney Diseases

    Ingrid M. Libman, M.D., Ph.D.
    Children's Hospital of Pittsburgh

    Rebecca Lipton, Ph.D.
    University of Illinois at Chicago

    Kelly Moore, M.D.
    Billings Area Indian Health Services

    Arlan L. Rosenbloom, M.D.
    University of Florida Children's Medical Services Center

    Steven M. Willi, M.D.
    Medical University of South Carolina

    William E. Winter, M.D.
    University of Florida

    Philip Scott Zeitler, M.D., Ph.D.
    The Children's Hospital of Denver


    Richard Eastman, M.D., Director of the Division of Diabetes, Endocrinology, and Metabolic Diseases at the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) and Chairman of the Diabetes Mellitus Interagency Coordinating Committee (DMICC), welcomed attendees to the meeting and introduced Charles A. Wells. Ph.D., Executive Secretary of the DMICC, as the meeting's emcee. Barbara Linder, M.D., Ph.D., Program Director at NIDDK, and the organizer of the meeting, then previewed the day's upcoming sessions and defined the meeting's goals as pinpointing the critical deficiencies in scientific knowledge about pediatric type 2 diabetes mellitus and prioritizing the research agenda.

    Descriptive Epidemiology

    Anne Fagot-Campagna, M.D., Ph.D., who is Medical Epidemiologist in the Division of Diabetes Translation at the Centers for Disease Control and Prevention (CDC), served as chair of this session.

    Obesity in Children

    Katherine Flegal, Ph.D., of the National Center for Health Statistics (NCHS) of the CDC, presented data the CDC has collected in several cross-sectional national surveys: the National Health Examination Surveys (NHESs) - NHES I in 1959-1962, NHES II in 1963-1965, and NHES III in 1966-1970, and the National Health and Nutrition Examination Surveys (NHANESs) - NHANES I in 1971-1974, NHANES II in 1976-1980, and NHANES III in 1988-1994. These surveys are representative samples of the entire United States.

    Concerning the issue of what should be used to define and measure obesity in children, Dr. Flegal presented data that use body mass index (BMI) - weight (kg)/height (m)- for determining obesity. Obesity is then defined statistically; the reference population is from the NCHS surveys that went into the CDC formulation of pediatric growth charts. Children who fall between 85% and 95% on the NCHS growth chart are defined as "at risk," and children who fall at or above 95% on the NCHS growth charts are defined as "obese." Among pre-school-aged children, there is no real trend for increasing obesity in boys aged 2-3 and 4-5; however, there is an increase in obesity incidence in girls aged 2-3 (from 2.0% obese in NHANES I to 4.8% obese in NHANES III) and particularly in girls aged 4-5 (from 5.8% obese in NHANES I to 10.8% obese in NHANES III). Dr. Flegal presented data for older children showing that for all 6-11 year-olds, obesity increased in males from 3.9% (NHES II) to 11.4% (NHANES III), and for females from 4.3% (NHES II) to 9.9% (NHANES III). For all 12-17 year-olds, obesity increased in males from 4.6% (NHES II) to 11.4% (NHANES III) and in females from 4.5% (NHES II) to 9.9% (NHANES III). The increase is especially striking in the more recent surveys (e.g., NHANES III). As expected, inasmuch as the earlier studies looked at populations close to the reference population, the prevalence of obesity in earlier studies was always close to 5%, which is expected for a definition of obesity that captures everyone at or above the 95 percentile.

    Dr. Flegal said that these increases in the incidence in obesity occurred in both African Americans and whites. In NHANES III, where African Americans were oversampled in order to increase the statistical significance, the prevalence of obesity in African American girls has reached 15.6% among 6-11-year olds (vs. 9.1% for white girls) and 15.7% among 12-17-year-olds (vs. 9.4% for white girls). Obesity in boys over time has increased about the same in both African Americans and whites (to 10-12%). Among Mexican Americans (NHANES III also oversampled this group) are several populations in which obesity is especially pronounced: males aged 6-11, 17.4% obese (compared with 11.2% obese for the total U.S. population of males aged 6-11) and males aged 12-17, 14.6% obese (compared with 11.3% obese for the total U.S. population of males aged 12-17).

    Dr. Flegal found that socioeconomic status, as measured by family income as compared with poverty-line income, showed only inconsistent correlations with prevalence of obesity: an inverse correlation was most pronounced in white males and white females aged 6-11 and 12-17. No discernible pattern was seen in non-Hispanic African Americans, whereas a positive relationship was seen in Hispanics.

    Dr. Flegal also presented data on the relationship between educational attainment of the household's "reference person" (i.e., the breadwinner who pays the rent) and obesity in household children. For white males aged 6-11 and 12-17, there is a fairly marked inverse gradient, with lesser inverse relationships seen in females and in nonwhites in the same age groups.

    Comparing the earliest data (NHES II and III) with the most recent data (NHANES III), the mean BMI has increased since the 1960s for all age groups between 6 and 17 years and for all races. Dr. Flegal also demonstrated the utility of analyzing the changes in BMI percentile distribution over time. For instance, the absolute weight of 6-year-olds at the 98 percentile has increased between the NHES II and NHANES III: the top 2% are much more overweight now than in the past. Analytically this can be described as an increase in the skewness of the distribution curve. Dr. Flegal also described the utility of transforming the data into a mean-difference plot, which demonstrates that the increases in BMI that have occurred between NHES and NHANES II have been largely at the highest BMIs and percentiles. The largest increases are seen in the extremely obese, and, conversely, very little changes are seen over time in the mean BMI of those children at the lowest percentiles. For many of the age and sex distributions surveyed, at the high end of the mean-difference plots (96 percentile-98 percentile) the increases over time have been around 5 BMI units, a very sizable change in obesity.

    Dr. Flegal summarized her presentation by stating that over the last few years, obesity has increased for all ages (both in children and in adults), for both sexes, and for all races studied. Overweight varies inversely with socioeconomic status for non-Hispanic whites, both by income and by education. Similar results have also been seen in non-US studies of obesity.

    Speculating on the causes of these increases in obesity, Dr. Flegal used the epidemiologic categories of agent, host, and environment. The agent is energy imbalance, a combination of greater calorie intake and less physical activity. The host implies differences, such as differences in genetic susceptibility. Environment refers to causal factors, such as social organization, cultural values, and economic factors. Research is needed to elucidate how changes in social environment have led to the trend toward increasing obesity.

    From a public health standpoint, three questions arise from these data: (1) What are the health implications of these recent increases in childhood obesity? (2) What should proposed remedies focus on? (3) What can we realistically accomplish? For instance, should efforts be targeted to the most overweight children (e.g., those with BMI of 45 or above)? Dr. Flegal proposed the following priorities: careful monitoring of trends in weight, to better detect sudden changes among populations; research to understand the causes of these changes; and a focus that is less on diet and activity per se and more on the social, cultural, and economic factors that determine them.

    During the question period, the issues were raised as to whether NHES and NHANES captured much data on physical activity in children and whether there are many data originating from schools, such as changes in rates of school-based physical activity and changes in fat content of school lunches. Dr. Flegal answered that the surveys did not capture data for activity in children and that there is a dearth of good national data on trends in physical activity and diet in children.

    Dr. Flegal was then asked what the definition of reference population would be in the soon-to-be released CDC growth charts. She stated that the most recent NHANES III data for children above age 6 was excluded, inasmuch as they reflect the most recent large increases in obesity, but that the earlier NHES and NHANES surveys would be used.

    There was also a discussion on the merits of using BMI as a gauge of obesity; for example, a highly muscular, athletic adolescent male might present with a high BMI, yet have low levels of adiposity. Dr. Flegal stated that clinicians are seeing children at the very high BMI (e.g., 35-45) who are almost surely obese and that it is these extremely obese children who appear to be most at risk for morbidity. She also mentioned that NHANES III has captured some skin-fold data that could be used to measure adiposity. An attendee offered that there is some evidence that waist circumference better correlates with risk and also excludes the highly athletic false positives that BMI identifies.

    Hispanic Americans

    Daniel E. Hale, M.D., Associate Professor of Pediatrics at the University of Texas Health Science Center at San Antonio, spoke next on type 2 diabetes in Mexican American youth, using San Antonio as a model.

    Dr. Hale first reviewed the demographics of the site. San Antonio is the tenth largest city in the United States, with just under 500,000 children (evenly divided between Mexican Americans and whites, with a small population of African Americans) in the catchment area. There is significant poverty in both the Mexican American and the African American communities, with about 60% of these households having an annual income of less than $25,000.

    Dr. Hale next described the pediatric diabetic population of San Antonio. He is one of only two pediatric endocrinologists in the city, who together have seen about 669 patients with pediatric diabetes in the last 9 years. Dr. Hale's investigations have led him to believe that these cases of the disease represent the preponderance of all such cases in San Antonio. Of the 669 patients, 512 (77%) have type 1 diabetes; 123 (18%) have type 2 diabetes; and 34 (5%) have other types, such as unclassified, atypical, and maturity-onset diabetes of youth. During this time, there has been a steady rise in the number of Mexican American type 2 pediatric patients, from approximately 2 per year to 20 per year. A similar rise in type 2 incidence in the African American and, only most recently, in the white communities has occurred. For the last 3 years, about 40% of the new pediatric diabetes diagnoses have been type 2, a substantial increase.

    Within San Antonio's Mexican American community is a growing percentage of diabetic children who have type 2 diabetes, especially among older children. Although 100% of diabetic Mexican American children aged 5 or under have type 1 diabetes, 74% of all new diagnoses among Mexican American children over 15 years old are for type 2 diabetes. A similar type 1/type 2 distribution is seen among African Americans; a lesser trend is seen among whites.

    Weight is a useful differentiating factor for distinguishing pediatric type 1 from type 2 diabetes: 84% of Mexican American children with type 1 diabetes have BMI at time of diagnosis under 20 (after rehydration); in contrast, 82% of Mexican American children with type 2 diabetes have a BMI of greater than 25. Pubertal status also distinguishes the two types of pediatric diabetes: 78% of children with type 1 diabetes are pre-pubertal or in very early puberty, whereas 65% of children with type 2 diabetes are fully pubertal at time of first diagnosis. Finally, about 93% of children with type 2 diabetes have acanthosis nigricans, whereas only 2% of children with type 1 diabetes have acanthosis.

    There is a strong familial trend for type 2 diabetes in children: 70% of children with type 2 diabetes have one or more parents who have diagnosed diabetes (either type 1 or type 2). In contrast, 12% of parents who have a child with type 1 diabetes have either type 1 or 2 diabetes. In contrast,.

    The severity of the diabetic illness can be roughly estimated by the rate of hospitalization at time of diagnosis. In general, far fewer children with type 2 diabetes (22%) require hospitalization at first diagnosis than do children with type 1 diabetes (87%); however, about half of the hospitalized children with type 2 diabetes presented with ketoacidosis (i.e., were seriously ill).

    Dr. Hale has undertaken rough estimates of the rate of incidence of pediatric diabetes in San Antonio: type 1 diabetes has an incidence of 8-9 per 100,000 (all races). This puts San Antonio between the type 1 diabetes incidences of 7.2 per 100,000 in San Diego (where there is also a substantial Mexican American population) and 15 per 100,000 in the Allegheny registry region of Pennsylvania. Total pediatric diabetes incidence rates in San Antonio have increased substantially in the last several years to the present rate of around 15 per 100,000 (again similar to the experiences of the Allegheny registry) because of a sharp increase in type 2 diabetes, with a largely flat incidence trend seen for type 1 diabetes.

    Dr. Hale next focused on the incidence rates of pediatric type 1 and type 2 diabetes by ethnicity. Within San Antonio's Mexican American community as a whole, the most recent incidence rate for pediatric type 2 diabetes is about 10 per 100,000. However, for Mexican American children over the age of 10, the incidence rate is 20 per 100,000, and for Mexican American children below the poverty line the incidence rate is about 30 per 100,000. The mean age at the time of diagnosis among Mexican American children is about 13.5 years. For type 1 diabetes, the incidence rate is 5 per 100,000 for Mexican American children, 8 per 100,000 for white children, and 12 per 100,000 for African American children.

    Dr. Hale next presented data from a small survey obtained through screening of about 1,500 children at three public middle schools in San Antonio. Nearly 60% of children surveyed had a family history of diabetes; 40% of the boys and 27% of the girls had a BMI greater than the 95 percentile; 29% had fasting hyperinsulinism; 19% had acanthosis nigricans; 7% had impaired fasting glucose; and 1% had undiagnosed diabetes (presumably type 2). These children also had an adverse risk profile for coronary artery disease. Nearly 50% had elevated total cholesterol; 29% had a family history of hyperlipidemia; 27% had elevated levels of triglycerides; 11% had a family history of myocardial infarction under the age of 50 in a parent or grandparent; 8% had low HDL cholesterol; 7% had high LDL cholesterol; and 3% of the girls and 15% of the boys had diastolic blood pressures greater than the 95 percentile for their age.

    Dr. Hale summarized by stating that, given these data, it is no surprise that he sees a rising incidence in pediatric diabetes not only in those populations that are particularly most at risk (i.e., Mexican American, poor) but also to a lesser but rising extent in more affluent, predominantly white children.

    An attendee asked Dr. Hale how convincing the distinction is between type 1 and type 2 diabetes in the newly diagnosed patients. Dr. Hale recounted the five questions that he typically asks when presented with a new case of pediatric diabetes: (1) How old is the child (type 2 diabetes tends to occur in children who are pubertal or postpubertal)? (2) Is there a family history of diabetes? (3) Is the child obese? (4) Does the child have acanthosis nigricans? (5) Does the child have a Spanish-surname? These five questions reveal two distinctly different populations. In addition, Dr. Hale described two biochemical distinctions between type 1 and type 2 diabetes: (1) Children with type 2 are less likely to have ketoacidosis at the time of diagnosis, and (2) generally, children with type 2 diabetes are less likely to have significant recognizable symptoms at time of diagnosis.

    In response to another question, Dr. Hale stated that his white population self-identifies as being English, Irish, or Scottish (rather than Eastern European, for example). Another attendee asked how the largely asymptomatic children with type 2 diabetes were being identified and was there, for example, a screening program for fasting hyperglycemia. Dr. Hale replied that identification was not through a screening program. Rather, blood sugar is measured during routine checkups or, in girls, after recurring vaginal or urinary tract infections. He cited the important role of school nurses in bringing certain issues to the parent's attention (e.g., child's falling asleep during class) which triggers a visit to the doctor, who then diagnoses diabetes.

    It was pointed out that many surveys of diabetes prevalence (e.g., coming out of screening programs) have either ignored children under 18 or stratified groups by using wide age ranges. It was urged that narrow age-groups be used (e.g., 5-year windows, such as 5-9, 10-14, 15-19, etc.) to determine age-specific prevalence. There were also suggestions that pubertal status may be more important than age and should be accurately assessed during data collection.

    African Americans

    Lawrence M. Dolan, M.D., Professor of Pediatrics and Director of the Diabetes Program at Children's Hospital Medical Center of Cincinnati, was the chair for this session. Dr. Dolan stated that the following major issues would be addressed during the session:

    • Diagnostic criteria for type 2 diabetes and the data on pediatric diabetes prevalence in the survey regions.
    • Estimates of what the incidence and prevalence of pediatric type 2 diabetes are in the African American population.
    • Clinical characteristics within the African American population.
    • Strengths and weaknesses of the available data.
    • Ascertainment bias.
        African Americans: Chicago

      Rebecca Lipton, Ph.D., Associate Professor of Epidemiology at the University of Illinois at Chicago, was the next speaker. Dr. Lipton spoke about a type 1 diabetes registry she has worked on that has since expanded its scope to include early-onset type 2 diabetes and atypical diabetes.

      Dr. Lipton described a continuum of glucose tolerance in children. At one end are the vast majority of children who are at no increased risk for diabetes. At the other end are the classic type 1 diabetes children, who are insulin dependent and have autoantibodies. In between are the children who are currently healthy, yet who are at elevated risk for developing diabetes. These children may have certain characteristics, such an increased genetic propensity to type 2 diabetes or an unfavorable energy balance. Other groups of children are those that already have undiagnosed type 2 diabetes, those that have known type 2 diabetes who are receiving treatment (typically insulin), and those known variously as having atypical diabetes or pediatric diabetes. It is hypothesized that these may be children whose type 2 disease was not detected early; children who are no longer obese, because of a diabetes-related weight loss; and children who, in fact, present with typical type 1 diabetes symptoms (e.g., ketosis), yet who after several years no longer require insulin treatment. Dr. Lipton has evidence that, as might be expected, these cases cluster in populations that have the least access to primary medical care. There may be alternative explanations for pediatric atypical diabetes patients. Some patients might have what is sometimes called double diabetes and have the HLA-DQ marker propensity to type 1 diabetes, but who also have an increased genetic load for type 2 diabetes and thus have a clinical course different from the classic type 1 diabetes.

      Dr. Lipton stressed that she sees as one of the major research goals the quantification of prevalence of each of the types of diabetes she mentioned.

      Dr. Lipton's Chicago registry focuses on insulin-treated diabetic children. The total number of children in Chicago aged 0-17 is approximately 650,000-700,000. The registry attempted to capture from hospital records all patients up through age 17 who meet the following criteria: Chicago resident at time of diagnosis, discharged and taking insulin, and Hispanic or African American (by hospital records or Spanish surname). Capture analysis indicated that the registry was achieving 86% complete ascertainment. About 25% (194) of the captured patients have been interviewed, usually some time after discharge. Currently, approximately 800 children are in the registry.

      Chicago has a large, diverse, Hispanic population: 70% are Chicago Hispanics are of Mexican American origin, 23% are of Puerto Rican origin, 2% are of Cuban American origin, and the remaining 5% represent numerous other countries of origin.

      One of Dr. Lipton's chief goals was to determine which children had type 1 diabetes, type 2 diabetes, and other conditions. The first step was to use hospital records to flag those children who were not clearly type 1 diabetes. These children had notations in the medical record such as "unusual diabetes," "atypical diabetes," "possible type 2 diabetes," acanthosis nigricans, obesity (or a calculated BMI above 27 at diagnosis), and polycystic ovarian syndrome (in girls). The second step was to conduct follow-up interviews to collect data. Those diabetic children who reported taking oral antihyperglycemic agents or those who had stopped taking insulin for at least 3 months and who had not developed ketoacidosis more than 6 months after first diagnosis were also classified as possibly having type 2 diabetes. In total, 207 diabetic children were classified as possible type 2 diabetes or atypical diabetes or both; the remainder were typed as classic type 1 diabetes. The best data cover the years 1985-1994, when nearly all the Chicago pediatric cases were being captured, and this was the period that Dr. Lipton focused on. During this time, there was a slight year-by-year increase in the number of cases, and the proportion of these cases that were classified as possible early onset type 2 or atypical diabetes or both also significantly increased (from 10% to 36%).

      The overall pediatric incidence rate of diabetes in this minority population, averaged over 1985-1994, was 15 per 100,000 annually. The pattern of increase over time was similar for the African American and Hispanic populations. Although the incidence rate of early-onset type 2 diabetes increased both relatively and absolutely over the 1985-1994 period (as has been seen elsewhere), Dr. Lipton stressed that in these two populations there was still a significant proportion of pediatric diabetes that could be classed as putative type 2 diabetes even as early as 1985. This time is much earlier than the 1996-1997 pediatric type 2 diabetes epidemic that has been seen by others.

      This phenomenon of an increasing proportion over time of pediatric type 2 and atypical diabetes was seen in certain subpopulations: African American and Hispanic males and females.

      Comparing the onset signs and symptoms seen in the children on first diagnosis, substantial proportions of children with classic type 1 diabetes and, to a lesser extent, those with putative type 2 diabetes presented with ketoacidosis. Notably, more than 50% of African American children with type 2 diabetes had ketoacidosis at diagnosis. Most African American children in Chicago with diabetes, whether type 1 or type 2, had polyuria and polydipsia; a smaller but still significant proportion had polyphagia at diagnosis. A majority of both type 1 and type 2 patients reported weight loss prior to hospital admission. The age at onset, as expected, was older in the early-onset type 2 patients than in the type 1 patients.

      The male-to-female ratio at onset for type 1 diabetes in African American children was about 1:1, whereas for putative type 2 diabetes there were substantially more females. In contrast, Dr. Lipton reported that in the Hispanic type 2 diabetes group, there were more males than females.

      Hispanic children tended to have lower incidence of ketoacidosis than did the African American children, although incidence was still substantial.

      Discussing the strengths and weaknesses of her data, Dr. Lipton pointed out that, on one hand, since her population was defined as those receiving hospital treatment, her capture record is complete (as compared with data collection oriented toward private practices and clinics) and represents a strength. On the other hand, a weakness is that only children receiving insulin upon discharge were included in the registry, thus missing mild-symptom or prediabetic children or children classified at the outset as type 2. Another strength is the large number of children in the registry, allowing for statistical analysis, as well as the large, 10-year time period, allowing for trend analysis. A weakness is that only recently has interview data collection started to include such indices as C-peptide levels, HLA types, MODY gene markers, and diet and exercise history. No autoantibody data have been collected. Another weakness is that the dozens of hospitals surveyed all have different charting practices.

      African Americans: Memphis and Arkansas

      George Burghen, M.D., Professor of Pediatrics and Chief of the Division of Endocrinology and Metabolism at the University of Tennessee at Memphis, was the next speaker. His hospital-based survey used the National Diabetes Data Group diagnostic criteria. Until July 1997, he and his colleagues used classic symptoms of diabetes and a random plasma glucose greater than 200 mg/dl to diagnose diabetes; an oral glucose tolerance test or fasting plasma glucose greater than or equal to 140 mg/dl was rarely needed to diagnose diabetes. In July 1997, they started using the American Diabetes Association criteria of classic symptoms with or without weight loss with a casual plasma glucose of 200 mg/dl or more.

      Dr. Burghen presented data showing plasma insulin measurements in the patients. He noted that 71% of the type 1 diabetes patients had insulin levels less than or equal to 10 m U/mL, compared with 16% of the type 2 diabetes patients, and 29% of the type 1 diabetes patients had insulin levels greater than 10 m U/mL, compared with 84% of the type 2 diabetes patients. Similarly, most of the type 1 diabetes patients had C-peptide levels less than 1 m g/mL, whereas a large percentage of the type 2 diabetes patients had C-peptide levels greater than 1 m g/mL.

      Dr. Burghen also used his survey data to demonstrate how autoantibody data could be used to type the diabetes patients, especially islet cell autoantibodies (ICA) and glutamic acid decarboxylase (GAD) autoantibodies. He noted that 60% of the type 1 diabetes patients were either ICA or GAD positive, compared with 4% of the type 2 diabetes patients. These latter ICA-positive and GAD-positive patients had been classified as type 2 diabetes using clinical criteria; Dr. Burghen would now prefer to reclassify them as obese type 1 diabetes, given the antibody data.

      Dr. Burghen said that 27 obese diabetes patients between the ages of 10 and 16 were screened to see whether they should be classified as type 1 or type 2 diabetes. They included 17 males and 10 females, 18 African Americans and 9 Caucasians. All had acanthosis nigricans. C-peptide mean was 5.4 m g/mL, which was elevated and suggested type 2 classification (see above). The mean BMI of African American children was 37 and was 30 for the Caucasian children. Most of the children had a lipid abnormality. Interestingly, ICA was positive in 4 of these 27 patients (15%) who had diabetes 1-3 years. The ICA-positive children had a mean BMI of 29 and a C-peptide range of 1.7-4.9 ng/ml, each of which is more typical of type 2 diabetes than of type 1 diabetes. Three other patients who were ICA-negative had C-peptide levels less than 0.4 ng/ml 2-7 years after the onset of diabetes.

      Dr. Burghen then discussed an Arkansas study (Carla Scott, et al. Pediatrics 1997) ending in 1995 that compared children who had type 2 diabetes with children who had type 1 diabetes, of the same age range and gender and from the same geographic area. Overall, 88% were type 1; 12% were type 2. In the last 4 years of the study, the proportion of the total that was type 2 diabetes showed a large increase. A large percentage of the type 2 diabetic children were African American; 86% had acanthosis nigricans, compared with 0% for the type 1 diabetic children; 96% had a BMI greater than the 85 percentile, compared with 24% for the type 1 children. Hypertension at onset was found in nearly one-third of the type 2 group, compared with only 4% of the type 1 group.

      Dr. Burghen described biochemical differences between the two Arkansas groups.

      Hemoglobin A1c levels were the same in both groups; blood glucose levels were lower in the type 2 diabetes children; carbon dioxide levels were similar but a bit lower in the type 1 children, and insulin and C-peptide levels were significantly higher in the type 2 children. Although ketosis was more prevalent in the type 1 children, a significant subgroup of type 2 children presented with ketosis.

      Dr. Burghen then presented some of his data generated from the LeBonhear Children's Medical Center in Memphis. Between 1990 and 1998, he and his colleagues studied 456 children with type 1 diabetes (75% of all the diabetic children) and 149 children with type 2 diabetes (25%). Whereas only 24% of the type 1 group were African American, 89% of the children with type 2 diabetes were African American. Caucasian children made up 343 (76%) of the type 1 diabetes cases, but only 11% of the type 2. African American children made up 37% of the total cases of diabetes. In the last year, 22% of the children with type 2 diabetes were Caucasian, a significantly increasing trend. Dr. Burghen showed that the proportion of all pediatric diabetes that is diagnosed as type 2 has been steadily increasing over the 1990-1998 period, as has been seen in other studies. Currently, about 30% of Dr. Burghen's pediatric new-onset diabetes patients have type 2. Among the African American diabetic children, the male-to-female ratio is 1:2.4. In 1998, African American children with diabetes comprised 45% of the new cases of diabetes, type 1 and type 2 combined.

      Dr. Burghen discussed the prevalence of type 2 diabetes in children (less than 20 years old) in greater Memphis (Shelby County), an area with an estimated 300,000 children, 56% of whom are African American. He estimated that there are 100 children with type 2 diabetes in this population, given the 80 children currently being followed in his clinic and assuming an 80% capture rate. Thus, the prevalence rate for type 2 diabetes for the population of children under age 20 in greater Memphis is currently estimated to be 1 per 3,000 (0.03%). For African American children the current prevalence is estimated to be 1 per 2000 (0.05%).

      Dr. Burghen reported that in 1998, 27 newly diagnosed cases of type 2 diabetes were recorded; using this statistic, he estimated that the overall incidence rate is 11 per 100,000 per year (again assuming an 80% capture rate). For the African American community, the projected incidence rate is higher: 14 per 100,000 each year.

      Focusing on children with type 1 diabetes, Dr. Burghen showed that a high percentage (75%) of them are less than 11 years old, and cited similar data for African American children (71% less than 11 years) and Caucasian children (78% less than 11 years). In contrast, the age distribution for type 2 diabetes was 26% among children less than 11 years old, 67% among children 11 to 15 years old, and 7% among children 16-19 years old.

      Dr. Burghen next described some data from the Diamond project in Shelby County, Tenn., which showed that the mean BMI did not substantially increase for the Caucasian children with diabetes over the years 1992-1995 but was increasing during these years in the African American children with diabetes: This trend presumably is reflected in an increasing incidence of type 2 diabetes in African American children during this time.

      In his discussion of diastolic blood pressure, Dr. Burghen said that the percentage of patients with diastolic blood pressure greater than the 90 percentile was similar for both type 1 and type 2 diabetes at the onset. Among children with type 1 diabetes, more white children than African American children were in this category. Among children with type 2 diabetes, more African American children than Caucasian children were in this category.

      In his discussion of blood lipids, Dr. Burghen noted the following American Diabetes Association target guidelines for clinical care of diabetes: total cholesterol less than 200 mg/dL, HDL cholesterol greater than 45 mg/dL, LDL cholesterol less than 100 mg/dL, triglycerides less than or equal to 150 mg/dL, and BMI less than 25. In his sample, Dr. Burghen saw that 15% of children with type 1 diabetes and 17% of children with type 2 diabetes had triglycerides greater than 150 mg/dL; 14% of children with type 1 and 43% of children with type 2 diabetes had total cholesterol levels greater than 180 mg/dL; 22% of type 1 and 42% of type 2 diabetic children had HDL cholesterol less than 40 mg/dL; 6% of type 1 and 31% of type 2 diabetic children had LDL cholesterol greater than 120 mg/dL. Furthermore, 6% of Caucasian children and 26% of African American children with diabetes had LDL cholesterol greater than 120 mg/dL. In addition, 18% of the type 1 diabetic children and 42% of the type 2 diabetic children had a total cholesterol-to-HDL cholesterol ratio greater than 4.0, 6% of the type 1 diabetic children and 29% of the type 2 diabetic children had an LDL cholesterol-to-HDL cholesterol ratio greater than 3.0.

      African Americans: Pittsburgh and Allegheny County

      The next speaker was Ingrid M. Libman, M.D., Ph.D., of the Department of Pediatrics at the Children's Hospital of Pittsburgh, who presented data from the Allegheny County Diabetes Registry in Pennsylvania. Allegheny County has a population of 1,400,000 (1990), of which 150,000 (11%) are nonwhites, 330,000 (24%) are under 20 years old, and 53,000 (16%) of the population under age 20 are nonwhite. The county's largest city is Pittsburgh. This population-based registry collected data from 1965 to 1994 and had the following criteria for inclusion:

      • Diagnosis of insulin-dependent diabetes (IDDM) by a physician
      • Treatment with insulin at time of diagnosis
      • Age at onset between 0 and 19 years
      • Resident of Allegheny County at time of diagnosis
      • Diabetes not secondary to other conditions such as cystic fibrosis

      The primary sources of case information were hospitals (23-25 in number), using a retrospective review of patient records every 5 years. The secondary sources were pediatricians and endocrinologists, who were surveyed by mail or fax every 5 years. Over time, the response rate to the surveys was 65-90%. The capture-recapture method was used to estimate the completeness of the registry and was determined to be approximately 95%.

      The following information was collected during the retrospective surveys of hospital records:

      • Name
      • Gender
      • Race
      • Date of birth
      • Date of diagnosis
      • Date of first insulin administration
      • Residence at diagnosis
      • Onset characteristics (not including weight and height)
      • Family history of diabetes
      Dr. Libman reviewed the data from the first 25 years of the registry, from 1965-1989. Overall, 10% of the patients were African American (15% of the 15-19-year-olds). For the whole 0-19 age group, the incidence rate of diabetes has been steadily increasing since the 1970s, both for whites and especially for African Americans; however, the absolute incidence rate for whites always exceeds the incidence rates for African Americans. Similar data were seen for the 15-19 subgroup during this 1965-1989 period.

      For cases identified between 1990-1994, Dr. Libman pointed out that African Americans overall made up 18% of the newly diagnosed diabetes cases (compared with 10% during 1965-1989); and that 15-19-year-old African Americans made up 38% of the cases (compared with 10% during 1965-1989). In fact, for this time period, the incidence rate for African Americans exceeds that for whites, for both males and females. Dividing the children into age groups, Dr. Libman showed that for the 0-4, 5-9, and 10-14 age groups, the incidence rate for whites still exceeds that for African Americans. For the 15-19 age group, however, the incidence rate for African Americans significantly exceeds the incidence rate for whites, having greatly increased since the 1965-1989 period in African American males and females.
  • To help answer the question as to whether this recent increase was for type 1 or type 2 diabetes, autoantibody data were collected from patients from one of the largest local hospitals, Children's Hospital of Pittsburgh. ICA, GAD65, and IA-2 autoantibodies were determined for approximately 441 children with clinically diagnosed IDDM. The prevalence of these antibodies was significantly lower in African Americans than in whites. About 90% of whites (both younger and older) had autoantibodies (and thus presumably had type 1 diabetes), whereas African Americans, especially the older African American children, had a substantially lower prevalence of autoantibodies. The percentage of those patients aged 0-19 with no evidence of any autoantibodies was 25% for African Americans compared with 5.5% for whites. Dividing now by age group, Dr. Libman showed that 3.1% of whites less than 11 years and 8.7% 11 years old and older had no autoantibodies, whereas 40% of African Americans 11-19 had none (and presumably thus had type 2 diabetes).

    Among a group of 40 African American diabetic children (67.5% autoantibody-positive and 32.5% autoantibody-negative), the mean age of the autoantibody-positive group was 8.7, whereas the mean age of the autoantibody-negative group was the significantly greater 12.9; 30% of the autoantibody-positive group was obese, whereas 62% of the autoantibody-negative group was obese (greater than 85 percentile). A group worth investigating further is the 38% of the autoantibody-negative group that were not obese (i.e., that were below the 85 percentile). These children are putatively non-type 1 (because they are autoantibody negative) and also putatively non-type 2 (because they are not obese).

    In summary, African Americans that were autoantibody-negative were older at diagnosis, had a higher prevalence of obesity, and had a greater tendency to have a parent with diabetes. These studies continue at present.

    African Americans: Cincinnati

    Lawrence M. Dolan, M.D., of the Children's Hospital Medical Center of Cincinnati and the chair of this session, spoke next. Dr. Dolan discussed his data on 128 pediatric type 2 diabetes patients diagnosed between 1992 and 1998, who are predominantly African American and female. He stressed the importance of autoantibody data in pediatric diabetes classification and noted that 96 of these patients had been tested for the presence of autoantibodies and that had all been autoantibody-negative. By Dr. Dolan's criteria, the presence of autoantibodies is sufficient to classify the disease as type 1, even if the patients present with other type 2 indications (e.g., obesity, in which case their records are flagged with an asterisk for special followup attention).

    Dr. Dolan stated that his study is not looking at any genetic markers but rather for possible autosomal-dominant inheritance (e.g., MODY), by screening family history for diabetic relatives. In addition, although he notes in his patients the presence of acanthosis nigricans and levels of insulin and C-peptide, these are neither inclusion nor exclusion criteria for type 1 or type 2 classification.

    Dr. Dolan calculated prevalence and incidence rates using the 124 records of patients within the eight-county Greater Cincinnati service area, which demographically is 87% non-Hispanic white, 12% African American, and 1% other (including Hispanics). As of December 1998, the overall prevalence of type 2 diabetes was 0.14 per 1,000 (0.014%) for 0-19-year olds. The children with type 2 diabetes tend to be female, African American, and in older adolescence (aged 15-19). Incidence rate determinations for type 1 pediatric diabetes were not possible before 1990, given that there were then only 1 or 2 patients per year in Greater Cincinnati. The overall incidence rate of type 2 diabetes increased greatly during the years 1992-1994; since then the incidence rate has been relatively stable.

    Overall, the average age for children with type 2 diabetes is 15.1 years, which does not vary much by gender or race. The average BMI is 36.3 (again no real difference by gender or race). Acanthosis nigricans is present in 71% of the population (African Americans 82%, whites 54%).

    Dr. Dolan next described the complications that he has seen in this type 2 diabetes population. High blood pressure, above the 90 percentile (corrected for age and height), was seen in a fair number of the 66 patients for whom data were available. High blood pressure was determined within 3 months of initial diagnosis but only once the patient was metabolically stable. There was a statistically significant relationship between systolic pressure and BMI.

    Prevalence of hyperlipidemia was determined by fasting lipid profiles also obtained no later than 3 months after diagnosis (mean = 29 days). Using 95 percentile limit for total cholesterol, triglycerides and LDL cholesterol, and an analogous 5 percentile limit for HDL cholesterol, Dr. Dolan noted that among 27 patients tested, 23 (85%) had at least one lipid abnormality: 14 among 17 of the African Americans (82%) and 9 among 10 (90%) of the non-Hispanic whites.

    Ophthalmologic abnormalities, such as retinopathy, were not seen. Dr. Dolan remarked that one group of Japanese investigators used fluorescein angiography to identify retinopathy and found up to a 38% risk for this disorder within 2 years of the onset of type 2 diabetes. Using the less sensitive method of direct ophthalmoscopy, Dr. Dolan could not detect any abnormalities; however, he believes that this is an important issue that needs to be addressed.

    Albumin excretion rate was checked to screen for diabetic nephropathy. Among the 11 children with type 2 diabetes who underwent testing, 3 (27%) had elevated excretion rates (greater than 20 m g/min as measured by timed overnight urine collection), signaling possible incipient nephropathy.

    Finally, Dr. Dolan addressed the complication of ketoacidosis, defined as serum bicarbonate level less than or equal to 15 milliequivalents/L. Among the 128 patients with type 2 diabetes, 17 presented in ketoacidosis (13%): divided equally by gender but heavily weighted toward the African Americans (19% of whom presented with ketoacidosis).

    Dr. Dolan continued by discussing the issue of ascertainment bias; whether his group of pediatric diabetes patients is a referral population. He said that it was particularly important to determine whether the sharp increase in the incidence rate of diabetes seen in Cincinnati during 1992-1994 was an artifact of ascertainment bias. Dr. Dolan and colleagues have not attempted the capture-recapture method, which is considered to be the gold standard for determination of ascertainment bias. The Cincinnati Children's Hospital, however, is, and has been since the 1970s, the only hospital in the eight-county region where pediatric diabetes cases are referred, thus ensuring a high level of completeness. Evidence for this is the hospital's calculated incidence rates for pediatric type 1 diabetes - 15-18 per 100,000 - which are similar to rates seen elsewhere (e.g., Allegheny County, Pennsylvania). An exhaustive 1995 survey of pediatric diabetes caseloads of Cincinnati diabetes educators leads Dr. Dolan to estimate that he is seeing 75-80% of children with type 2 diabetes.

    Dr. Dolan then discussed whether there had been a recent change in the referral pattern to the diabetes center at the Cincinnati Children's Hospital. He said there had been no change over time in the diagnostic criteria for type 2 diabetes; the market share of the hospital; the ethnic pattern of referrals; the referral patterns by the practicing physicians; and the geographic distribution by ZIP code of either physicians or referrals. Nonetheless, Dr. Dolan asserted that his patient database is probably a significant underestimate of the whole type 2 pediatric diabetes disease process. Evidence for this includes Dr. Dolan's review of the chief complaint of patients on hospitalization: only 59% had the classic symptoms of polyuria and polydipsia; 40% had nonclassic symptoms, and 13% presented only with a positive urinalysis for glucose, obtained during a routine physical examination.

    African Americans: Panel Discussion

    Dr. Dolan asked Dr. Lipton to discuss her contention that the pediatric type 2 epidemic among minorities did not start suddenly in the mid-1990s, but has been steadily growing for a much longer period. Dr. Lipton reiterated her point that a substantial proportion of diabetic children in Chicago had type 2 diabetes (or at least not type 1) in 1985-1987. She also stated that she still suspects that there may be some ascertainment bias in the sudden big jumps in type 2 incidence that Dr. Libman and Dr. Dolan saw in the early 1990s. An example of ascertainment bias that Dr. Lipton saw in her own Chicago studies was her identification of 23 African Americans who had died over a period of 8 years at onset of symptoms; some of these patients were only diagnosed at autopsy. Dr. Lipton thus suggested to the other attendees that they check the death records from county medical examiners for unexplained deaths that were later related to type 2 diabetes after the fact at autopsy. Dr. Lipton predicted that by taking account of these early, hitherto missed cases of type 2 diabetes, one might sufficiently increase the incidence in the early years to smooth out the "sudden" increases in incidence seen by some in the mid-1990s.

    Dr. William Knowler made the point that what is being called complete ascertainment of cases of type 2 diabetes is really complete ascertainment of cases that have been diagnosed and what should be focused on instead is how much total diabetes there is, both diagnosed and undiagnosed. Dr. Knowler offered, and the other attendees concurred, that true, complete ascertainment can never be achieved by using registries or hospital case series and that population samples should be tested and screened. Dr. Lipton then reviewed the Japanese experience, where school-based universal screening of children by urinalysis was used to determine that the proportion of all pediatric diabetes classified as type 2 is 66-75% (i.e., the type 2 rates are two to three times as high as the type 1 rates). Dr. Knowler predicted that U.S. proportions would be similar.

    An attendee made a further point that in the registries described earlier, cases of girls with type 2 diabetes who were also pregnant were not being categorized as type 2 but were often categorized instead as having gestational diabetes.

    As to autoantibody tests, an attendee stated that there are no U.S. data comparing the performance of the four common autoantibody reagents, nor have antibody protocols been sufficiently standardized to allow comparisons of data taken at two different times in two different locales.

    Another attendee commented that elevated serum insulin and elevated C-peptide levels are strong indicators of insulin resistance and are thus powerful tools in confirming the diagnosis of type 2 diabetes. These tools must be used cautiously, however. Serum glucose must first be brought under control; otherwise, the insulin and C-peptide determinations are unreliable because their values may be suppressed.

    Dr. Burghen commented that the presence in children of syndrome X symptoms, such as hyperinsulinemia, is indicative of insulin resistance and that these are children at high risk of developing type 2 diabetes. Even if children never progress to full-blown diabetes, syndrome X symptoms may be dangerous to their health, and these children should be monitored. There was a consensus that longitudinal studies are needed to learn the natural history of type 2 diabetes in children.

    NIDDK Strategic Plan

    Carol Feld, Associate Director of NIDDK for Scientific Programs and Policy Analysis, discussed the implementation process for the NIDDK Strategic Plan. The recommendation that NIDDK, along with all the other Institutes within NIH, promulgate a Strategic Plan came from a congressionally directed Institute of Medicine (IOM) study on improving NIH's setting of priorities and public input. Among the IOM recommendations that have been put in motion are that all NIH Institutes establish an Office of Public Liaison, that a Council of Public Representatives to the NIH Office of the Director is established, and that Strategic Plans are formulated (they are due to be submitted in draft form by December 31, 1999).

    The NIDDK Strategic Plan will have the following features:

    • Organization by cross-cutting scientific themes, rather than by diseases within the NIDDK research mission
    • A catalog of scientific opportunities and needs
    • Implementation strategies (what tools, reagents, techniques, and personnel are needed)
    • Five-year time horizon
    • Public involvement
    • A final document written in lay language
    • A widely distributed plan, both electronically and in hard copy
  • Requests for public input have been made to more than 70 voluntary and professional organizations with interests within NIDDK's disease mission, and a request for public input has also been posted on the NIDDK web site. In addition, all of the Interagency Coordinating Committees (e.g., DMICC) have been enlisted to supply input into the NIDDK Strategic Plan.
  • The cross-cutting working groups are:

    • Genes and Disease - Regulation, Expression, Screening
    • Cells - Integration of Biological Mechanisms in Health
    • Causes and Mechanisms of Disease and Injury
    • Prevention and Treatment of Disease - Epidemiologic and Clinical Investigation
    • Infrastructure - Human Resources, Technology, and Research Facilities

    Native Americans

    Kelly Moore, M.D., the Acting Chief Medical Officer/Area Diabetes Consultant at the Billings Area Indian Health Services, was the chair for this session, as well as the first speaker. Dr. Moore noted that the Native American/Alaska Native population is a growing one, with a 1994 population of 2.2 million and a projected population of 4.3 million by 2050. More than 550 tribal groups are federally recognized and reside within 34 States; about one-half live in urban areas. About one-third of the Native American/Alaska Native population live in poverty, as compared with about 13% for the total U.S. population.

    Dr. Moore referred to Canadian aboriginal data from Dr. Heather Dean's studies, in which the average age at diagnosis of type 2 diabetes is 10-14 years, similar to data from other populations. Also, BMI for males in this Canadian aborginal group is nearly always above the 95 percentile. In addition, Type 2 diabetes tends to be more common in females in this population, whereas type 1 diabetes tends to occur equally in males and females. Finally, 76% of the children with type 2 diabetes have mothers with diagnosed diabetes.

    Next, Dr. Moore compared the Canadian data on diabetes with data collected from the Pima Indians in Arizona. She noted that most of the cases are being first diagnosed in children between 10-19 years and again the gender ratio for type 2 diabetes tends to favor females, although recent data show relative increases in male incidence. Also, the Pima diabetes population shows a high percentage of exposure to diabetes in utero, which is proposed to be a major risk factor in this population for the development of type 2 diabetes.

    In a discussion of fasting insulin levels, Dr. Moore presented data from the Canadian study showing that about 62 of 103 (60%) study participants have increased insulin levels. Hyperinsulinemia has been shown to be a good predictor for the development of type 2 diabetes in Pima children; the 10-year diabetes incidence rate is much higher in those children with elevated fasting insulin. Frequency distribution data of fasting insulin and fasting glucose levels show that Pima children have markedly higher levels than do white children, with a marked rightward shift of the distribution curves.

    Dr. Moore next reviewed data on diabetes in nationwide American Indian/Alaska Native populations from 1991-1997 collected by the Indian Health Service (IHS) and analyzed by the CDC. The data were collected from 151 IHS service units. Persons with diabetes were identified according to the ICD-9 code 250. The denominators in these prevalence rate determinations were taken from U.S. Census data covering each of these IHS service units; 46 IHS service units (representing 16% of the HIS population) were excluded because of incomplete or missing data.

    It has long been known that adults in the American Indian/Alaska Native population have high prevalence rates of diabetes. Dr. Moore showed that the data revealed an increase during 1991-1997 in diabetes in all of the age groups studied, particularly in adolescents aged 15-19 (up 32%), young adults aged 20-24 (up 36%), and in adults aged 25-34 (up 28%). The increase seen in young men was about twice that seen among young women. Dividing up the American Indian/Alaska Native data by region, it was apparent that the Alaskan region is the region with the lowest prevalence of diabetes in children, adolescents, and young adults. All of the regions showed a steady increase in diabetes over the period 1991-1997.

    Dr. Moore listed three limitations of the data: (1) the case definition does not distinguish between type 1 and type 2 diabetes, (2) there may be inaccuracies in the estimations of the populations, and (3) only data on American Indians treated within the IHS are collected (e.g., many urban Indians do not use the IHS and thus these cases are not captured).

    Canadian First Nations People

    Heather J. Dean, M.D., Professor of Pediatric Endocrinology and Metabolism at the University of Manitoba, was the next speaker. Dr. Dean discussed several regional studies on diabetes prevalence in aboriginal populations in Canada (also known as First Nation). The largest language grouping is the Algonquin, which is Cree-Ojibwa; they also have the highest rate of diabetes. Diabetes prevalence is lowest in the Haida community in the west and in the Inuit communities in the north. In contrast to the U.S. Native American community, 70% of the First Nation inhabitants live on reserves, not in urban areas. In Manitoba, there are 60,000 First Nation inhabitants, 70% of whom live in small isolated villages in the north.

    Dr. Dean reviewed 1990s prevalence data from three regional diabetes registries, which all show type 2 diabetes prevalence between 1.7 per 1000 and 2.5 per 1000. For one of the registries - a Manitoba region registry - Dr. Dean showed by capture-recapture techniques that there is a 95% ascertainment rate of pediatric diabetes.

    Next, Dr. Dean presented data from diabetes screening programs. Among three youth populations screened in the mid-1990s, prevalence rates averaged around 16 per 1000, that is, about 10 times higher than the prevalence rates from the registry data.

    Dr. Dean then outlined the following research priorities:

    • Development of standard diagnostic criteria
    • Use of standard age groups (e.g., 0-4, 5-9, 10-14, 15-19)
    • Universal differentiation between type 1 and type 2 diabetes in registry databases
    • Acceptance of standardized, uniform methods for large-scale population-based screening
      Dr. Dean stressed that in screening initiatives in large populations, the protocols may have to be more practical (e.g., capillary blood glucose vs. drawn blood, random blood glucose instead of fasting blood glucose, etc.).
  • Pima Indians

    William C. Knowler, M.D., Dr.P.H., Chief of the Diabetes and Arthritis Epidemiology Section, NIDDK, was the next speaker. Dr. Knowler remarked that diabetes prevalence in Pima Indian adolescents, which is historically high, has dramatically increased in 1976-1996. The type 2 diabetes prevalence rate is now 8-10 times higher than rates in comparable white populations from Rochester, Minnesota.

    Dr. Knowler then discussed the risk factors for type 2 diabetes in the Pima population. Family history of diabetes is a strong risk factor; interestingly, if only one parent has diabetes, then the risk for the child is greater if the mother, rather than the father, has diabetes. The strongest family risk factor, in fact, is in utero exposure to a diabetic mother, rather than simply inheritance of diabetes susceptibility genes. Dr. Knowler showed that much of the recent increase in type 2 diabetes prevalence is attributable to the increase in the proportion of Pima women who are diabetic while pregnant.

    Another risk factor for type 2 diabetes is birth weight. Either very low or very high birth weight markedly increases the risk for diabetes. An independent risk factor for type 2 diabetes is bottle feeding.

    Metabolic Phenotyping

    Arlan L. Rosenbloom, M.D., Distinguished Service Professor Emeritus of the University of Florida, was the chair for this session.

    Metabolic Phenotyping I

    The first speaker was Silva A. Arslanian, M.D., Professor of Pediatrics at Children's Hospital of Pittsburgh. She described a hyperglycemic clamp experiment, in which healthy African American adolescent volunteers were matched with healthy white adolescents for BMI, pubertal development, and hormonal development. The hyperglycemic clamp experiment measures the amount of insulin secretion over120 minutes while blood glucose is clamped at a constant, hyperglycemic level of 225 mg/dL. Dr. Arslanian demonstrated that, for the same glucose level, African Americans on average secrete significantly more insulin than do whites. From these experiments, Dr. Arslanian derived an insulin sensitivity index for each of the volunteers; insulin sensitivity was approximately 35-40% lower in the African American adolescents than in their matched white peers.

    Dr. Arslanian noted that the mean age of onset of pediatric type 2 diabetes ranges from 13 to 15 years (midpuberty). Puberty is normally associated with insulin resistance ¾ i.e., a plot of age vs. fasting insulin level rises before adolescence, peaks at around age 15, and then declines again to prepubertal levels. She measured in vivo insulin sensitivity by the hyperinsulinemic euglycemic clamp technique and found that adolescents have approximately 30% lower insulin sensitivity than either preadolescents or adults. Her recent data implicate a spurt of growth hormone at the time of puberty as the cause of this phenomenon. Insulin sensitivity (controlling for BMI) is less in girls than in boys and is inversely proportional both to BMI and to total percent body fat. Fasting insulin levels, a marker of insulin resistance, also as expected, increase with increasing BMI, even when measured in normal nondiabetic children. Gender differences in insulin sensitivity need further evaluation with careful attention to body composition differences.

    Dr. Arslanian then presented data on the relative contributions of visceral vs. subcutaneous fat, which she measured by CAT scan. As seen for total percent body fat, insulin sensitivity declines (and fasting insulin levels increase) with increasing percentages of either visceral or subcutaneous fat; however, the slopes are much more steep for visceral fat. The impact of excessive visceral fat is thus more deleterious than the impact of subcutaneous fat. Dr. Arslanian hypothesized that if, as some have said, the waist circumference is a better predictor of type 2 diabetes risk than BMI, this is because waist circumference is a proxy measurement for the percentage of visceral fat.

    Dr. Arslanian mentioned that hyperandrogenism and related syndromes, such as polycystic ovarian syndrome and hirsutism, are associated with increased risk for type 2 diabetes. She conducted a study comparing obese (BMI = 33) hyperandrogenic adolescents with nonhyperandrogenic adolescents matched for BMI, percent body fat, lean body mass, and abdominal adiposity. The fasting insulin levels were twofold higher with similar hepatic glucose product and the insulin sensitivity 30-40% lower in the hyperandrogenic adolescents. Thus, here data indicate both hepatic and skeletal muscle insulin resistance. This correlates well with the increased risk for developing type 2 diabetes seen in hyperandrogenic adolescents. Dr. Arslanian also detected evidence of a decrease in pancreatic beta-cell function in these hyperandrogenic adolescents who have impaired glucose tolerance. Young children (10 years old) with a family history of type 2 diabetes had lower insulin sensitivity than peers without family history.

    Metabolic Phenotyping II

    The next speaker was Steven M. Willi, M.D., Associate Professor of Pediatrics at the Medical University of South Carolina. Dr. Willi described a study of 97 South Carolina African American children and young adults (53 females and 44 males) with diabetes who required insulin therapy before the age of 20. Comprehensive islet cell autoantibody testing was undertaken, including ICA, anti-GAD antibodies, and antityrosine phosphatase antibodies. In addition, Dr. Willi measured C-peptide after withholding intermediate-acting insulin for at least 24 hours and short-acting insulin for at least 12 hours. Of these 97 patients, only 50 were classified as having type 1 diabetes, with antibodies directed at one or more islet cell antigens, and a fasting C-peptide level less than 0.6 ng/mL (the Diabetes Control and Complications Trial [DCCT] standard). The remaining 47 patients were diagnosed as having pediatric type 2 diabetes: they were autoantibody-negative and had significantly elevated C-peptide levels (mean = 2.8 ng/mL).

    Similar to what others presented at this meeting, the peak age at onset for the type 2 diabetes children was 13-14 years. This age coincides with the peak age of insulin resistance in adolescents and was much later than the peak age at onset for the type 1 diabetes children. The type 2 children were also much heavier; half of them had a BMI above 30. A family history of diabetes was much more common in the children with type 2 diabetes.

    Acanthosis nigricans was also much more prevalent in the type 2 children (more than 50%) than in the type 1 children (14%). A substantial proportion (two-thirds) of these type 2 children had a history of significant ketoacidosis.

    HLA-DR/DQ typing was also performed and compared with an unrelated African American control population from the region. Not surprisingly, the clearest associations were observed in the children with type 1 diabetes, with the anticipated increased incidence of DR3 and DR4 as well as increases in the DQ types 0201 and 0302. There was also a significantly increased incidence of DR5 seen in the children with type 2 diabetes.

    As seen elsewhere, the trend in South Carolina over the last several years has been for pediatric type 2 diabetes to increase in prevalence. Currently, type 2 diabetes is the predominant form among African American children in South Carolina.

    Dr. Willi next described the 4-day inpatient metabolic studies that were undertaken on some type 2 patients in the medical school's General Clinical Research Center, along with young adult African American controls and a group of type 1 patients matched for age and race. The protocol included a total body dual-energy x-ray absorptiometry (DEXA) scan, determination of glycohemoglobin (hemoglobin A1c assay), and urinary C-peptide secretion. The patients with type 2 diabetes, who were evaluated only after a period of improved metabolic control at the time the study (as evidenced by their mean hemoglobin A1c = 8.4%), were quite obese (mean BMI = 33). Insulin secretion and sensitivity were measured by conducting a 0.5 g/Kg intravenous (IV) glucose challenge followed by IV insulin injection at 20 minutes. The insulin secretory response of IV glucagon was also evaluated, and a euglycemic hyperinsulinemic glucose clamp experiment was performed in conjunction with indirect calorimetry.

    Dr. Willi presented data showing that the fasting insulin level of the pediatric type 2 diabetes group was elevated at 50.2 m units/mL compared with 14.6 m units/mL in the control population. Despite this elevation in fasting insulin level among the type 2 patients, their urine C-peptide levels were significantly lower.

    Insulin response to IV glucose in the patients with type 2 diabetes was paradoxical: the first phase of insulin secretion, which typically occurs within five minutes of glucose exposure, is completely absent. In contrast to IV glucose, the response of insulin to IV glucagon in the type 2 patients was quite pronounced and was greater than in the controls. This insulin response to glucagon was rapid, and the kinetics suggest that a relatively large pool of stored insulin granules exist within the beta cells, yet cannot be released in response to glucose.

    Dr. Willi evaluated the second phase of insulin secretion by monitoring C-peptide levels over a 3-hour time course after IV glucose challenge. In the predominantly African American pediatric patients with type 2 diabetes this second phase is quite blunted as compared to the control group. As expected, the C-peptide response to glucose in the matched patients with type 1 diabetes was absent.

    Dr. Willi next reviewed the results of the euglycemic hyperinsulinemic glucose clamp experiment. The patients with pediatric type 2 diabetes exhibited insulin resistance, as reflected in a mean glucose disposal rate of 6 mg · kg lean body mass -1 · min -1.

    In summary, Dr. Willi stated that his studies suggest that a significant subset of young African Americans with diabetes have a form of type 2 disease that may be ketosis prone. These patients present during adolescence, are frequently obese, and have acanthosis nigricans. After metabolic stabilization, fasting insulin levels appear to be elevated. Despite a high fasting insulin level, first-phase glucose-stimulated insulin release is nearly absent; however, insulin response to glucagon is preserved. The euglycemic hyperinsulinemic glucose clamp experiments demonstrate severe insulin resistance in these patients, even after correcting for their obesity.

    Furthermore, Dr. Willi said that obese African American adolescents with ketosis-prone type 2 diabetes manifest a combination of abnormal beta-cell function and insulin action. This selective defect in glucose-insulin coupling in these patients suggests an abnormality in either uptake or metabolism of glucose by the beta cell, as well as in the periphery.

    Metabolic Phenotyping III

    The next speaker was Joan DiMartino-Nardi, M.D., Associate Professor of Pediatric Endocrinology at the Montefiore Medical Center, Bronx, NY. Dr. DiMartino-Nardi discussed her research using patient records for 73 patients with type 2 diabetes diagnosed between 1990 and 1999. Test results for type 1 diabetes-specific autoantibodies were available for 70% of them; any patient who had positive titres of antibodies was excluded from the study. Dr. DiMartino-Nardi was able to identify 68 patients with type 2 diabetes; 35 were African American, 29 were Caribbean Hispanic, and 4 were Asian Indians. There was a preponderance of females to males.

    Looking back over the past several years, Dr. DiMartino-Nardi has also detected an increasing trend in the percentage of pediatric patients with diabetes who are type 2. She noted a tenfold increase in the number of pediatric patients with type 2 diabetes since the early 1990s.

    Of the pediatric patients with type 2 diabetes, more than 80% had acanthosis nigricans, 50% had polyuria/polydipsia, 20-25% had weakness, and 25-30% had weight loss. Of particular concern to Dr. DiMartino-Nardi was that 30% of the patients were completely asymptomatic at the time of diagnosis. Although most of the patients were morbidly obese, a substantial number had normal or near-normal BMI (some due to recent diabetes-related weight loss). The mean age for the African American patients with type 2 diabetes was 13.7, and for the Caribbean Hispanics it was 14.8 years. Most of the patients were in mid-late puberty (Tanner IV/Tanner V), 15-25% were in early puberty, and none were prepubertal. At the time of diagnosis, all of the diabetic children had fasting blood glucose above 126 mg/dl, 5 children presented with severe ketoacidosis, 10% had normal hemoglobin A1c levels, and the vast majority had elevated insulin levels.

    Dr. DiMartino-Nardi next discussed the role of the fasting glucose-to-insulin ratio in the assessment of insulin sensitivity in girls with premature adrenarche, which recent reports indicate may be a risk factor for the development of polycystic ovarian syndrome.

    Many of Dr. DiMartino-Nardi's patients with premature adrenarche have several of the features of syndrome X: obesity, acanthosis nigricans, and strong family history of diabetes. Many patients also have marked hyperandrogenism (e.g., ACTH-stimulated 17-hydroxypregnenolone levels greater than 2 standard deviations above the mean for early pubertal children). Furthermore, more than 50% of her patients have decreased insulin sensitivity, as measured by the frequently sampled IV glucose tolerance test (FSIVGTT). Interestingly, the more insulin-resistant children were also the children to have the more severe hyperandrogenism.

    Dr. DiMartino-Nardi next discussed the recent reports of using the fasting glucose-to-insulin ratio as an indicator of insulin sensitivity, as an alternative to the more involved standard tests: FSIVGTT and the euglycemic hyperinsulinemic glucose clamp procedure. She reviewed recent published reports by R.S. Legro and A. Dunaif that a low fasting glucose-to-insulin ratio (less than 4.5) in obese women with polycystic ovarian syndrome was a very sensitive and specific tool for identifying women with insulin resistance when compared with the FSIVGTT.

    The FSIVGTT has several advantages: it is a well-validated method for measuring insulin sensitivity, and it correlates very closely with insulin sensitivity as determined by the euglycemic insulin clamp. However, FSIVGTT is time consuming, is labor and cost intensive, and is not amenable to large-scale studies or to routine assessment or followup of patients with insulin resistence.

    Dr. DiMartino-Nardi described a study she conducted that looked at whether the fasting glucose-to-insulin ratio, compared with the FSIVGTT, was useful as a screening tool for assessing insulin sensitivity in children with premature adrenarche. She studied 33 prepurbertal girls with premature adrenarche (22 Caribbean Hispanic, 11 African American; mean age 6.8 years). Among them, 15 were identified as insulin resistant using the FSIVGTT; 13 of these girls had a fasting glucose-to-insulin ratio less than 7, giving a true positive rate or sensitivity of 87% and a specificity of 89%. The degree of insulin resistance in the girls with premature adrenarche was, as expected, a function of their BMI. This dependent relationship was seen whether the fasting glucose-to-insulin ratio or the FSIVGTT was used; the data were superimposible, again validating the utility of the fasting glucose-to-insulin ratio test.

    The girls who had premature adrenarche and were insulin resistant also had several other abnormal biochemical markers: lower insulin-like growth factor binding protein 1 (IGF BP-1); significantly higher levels of ACTH-stimulated 17-hydroxypregnenolone; higher free testosterone; lower sex hormone-binding globulin levels; and lower HDL levels. Again, these relationships held whether insulin resistance was assayed using the fasting glucose-to-insulin ratio assay or the FSIVGTT.

    In conclusion, Dr. DiMartino-Nardi stated that she hopes the fasting glucose-to-insulin ratio assay will identify which girls with premature adrenarche also have syndrome X-type metabolic abnormalities and are thus at risk for developing attendant possible complications, such as polycystic ovarian syndrome and pediatric type 2 diabetes.

    Metabolic Phenotyping IV

    The next speaker was Philip Scott Zeitler, M.D., Ph.D., Assistant Professor of Pediatrics at the Children's Hospital of Denver. Dr. Zeitler presented data that were gathered while he was in Cincinnati to analyze the characteristics of adolescents with type 2 diabetes and their families. A goal of the study was to refine a profile of at-risk individuals and families, to allow the development of early-intervention strategies for children at risk for type 2 diabetes.

    Dr. Zeitler studied 11 families that had both parents present and an adolescent with type 2 diabetes. All family members, including siblings, were studied, for a total of 42 participants. Among them, type 2 diabetes had already been diagnosed in 45% of the mothers and 40% of the fathers before the study, and Dr. Zeitler diagnosed type 2 diabetes in 27% more of the fathers during the medical examination. In 27% of the families, both parents had type 2 diabetes. The diabetic parents tended to have poor metabolic control; for example, the mothers with type 2 diabetes had elevated mean hemoglobin A1c levels of 13%.

    All of the family members were obese (e.g., skinfold measurements were all greater than 95 percentile) and the diets were very high in fats and low in fiber, even if the mother had known she had type 2 diabetes and had been educated on good dietary practices. Many of the family members showed signs of binge-eating behavior. None of the patients participated in structured exercise programs, and 85% engaged in no physical activity; 45% of the siblings also reported no physical activity. Family members, however, reported watching television and/or playing video games and/or using computers 3-5 hours per day.

    Although none of the siblings had frank diabetes, all had elevated fasting insulin and C-peptide levels, suggesting insulin resistance, syndrome X, or both.

    In conclusion, Dr. Zeitler stated that the siblings and parents of adolescents with type 2 diabetes exhibit many of the risk factors for the disease and should be screened. Furthermore, given the frequency of lifestyle risk factors in these families, the entire family needs to be considered as the target for both screening and intervention.

    Metabolic Phenotyping V

    The next speaker was Kenneth Lee Jones, M.D., Professor of Pediatrics at the University of California at San Diego. Dr. Jones described the ethnic characteristics of the San Diego population as it relates to diabetes incidence. As is true in most localities, the incidence of type I diabetes in San Diego is largely the same in all ethnic groups except for Asian Americans, who have a low incidence. An interesting observation about the ethnic data on type 2 diabetes is that in the San Diego Asian community many more males than females have the disease. A similar male predominance has been seen in Asian communities in Hong Kong.

    Dr. Jones then focused on those rare type 2 diabetes patients who are not obese. In the San Diego population that Dr. Jones studied, all African Americans with type 2 diabetes were obese, whereas some Mexican Americans and Asian Americans with type 2 diabetes were not obese. Total mean BMI for type 2 patients also varied by ethnicity, with mean BMI ranked as follows from highest to lowest: African Americans, Mexican Americans, whites, Asian Americans.

    Dr. Jones did not observe acanthosis nigricans in any of the Asian American or white females but it was seen in both males and females among the African Americans and Mexican Americans.

    Dr. Jones next discussed a study of the predictive value of C-peptide levels at time of diagnosis in 41 patients who had type 1 diabetes and in 38 patients who had type 2 diabetes. The predictive value of fasting C-peptide analysis as a criterion for type 2 diabetes was 91%.

    Dr. Jones then discussed a study by Peter Reaven and colleagues that followed both Mexican American and white school populations and showed that by age 11 the mean of the BMI for the Mexican Americans was already higher. Study results also showed a significantly higher fasting insulin level among Mexican Americans. Both findings suggest that syndrome X develops disproportionately in the Mexican American school population. The researchers developed a rank score for syndrome X, involving a large number of test results values, and showed that, in general, Mexican American children in San Diego have significantly higher syndrome X scores than do matched white controls.

    Dr. Jones pointed out that these data show that, as early as age 11, it is possible to identify in large populations of nonaffected individuals those children who are at risk for type 2 diabetes because they display syndrome X factors. These children would then be candidates for potential early intervention strategies.

    Metabolic Phenotyping VI

    The final speaker for this session was William E. Winter, M.D., Professor of Pathology and Pediatrics at the University of Florida. Dr. Winter discussed maturity-onset diabetes of youth (MODY), which is a familial, youth-onset form of diabetes in individuals who are not obese that results primarily from insulinopenia. MODY was first described by Dr. S.S. Fajans in the 1960s, who defined what Dr. Winter calls classic MODY as follows:

    • Diabetes onset before age 25
    • Correction of fasting hyperglycemia without insulin for at least 2 years after diagnosis
    • Nonketotic disease
    • Autosomal-dominant mode of inheritance
      In 1987, Dr. Winter described what he called an atypical form of diabetes in African Americans younger than age 40. He used this term because the patients' symptoms were similar to symptoms of type 1 diabetes (i.e., ketoacidosis, severe weight loss, polyuria/polydipsia), but after a number of months to years, the disease developed a non-insulin-dependent course. This atypical diabetes is transmitted in families with an autosomal-dominant mode of inheritance. Dr. Winter's atypical diabetes differs from classic MODY in several respects¾ MODY has not been described in African Americans; classic MODY is not a disease of acute onset; and, in general, patients with classic MODY never have ketoacidosis.
  • Dr. Winter's atypical diabetes of African Americans represents at least 10% of diabetes in African Americans under 40; classic MODY represents less than 5% of diabetes in whites.

    Five different genetic loci have been associated with MODY. Four of the genetic abnormalities involve transcription factors that regulate insulin production; one gene (MODY 2) codes for glucokinase, which is involved in the beta-cell sensing of glucose elevations. Dr. Fajan's classic MODY is MODY 1, a genetic defect in hepatic nuclear factor-4 (HNF-4 ). MODY 1 causes progressive insulinopenia and is an uncommon cause of type 2 diabetes in adults. MODY 2 is the most common form of MODY, but is also an uncommon cause of type 2 diabetes in adults. MODY 3 also causes a progressive insulinopenia due to gradual loss of beta-cell function and is the most common cause of MODY in the British population. MODY 4 regulates beta-cell development in utero. The latest MODY to be described, MODY 5, is very rare.

    Dr. Winter has described two mutations in atypical diabetes of African Americans: (1) a mutation in glucokinase, and (2) a mutation in NADH dehydrogenase subunit 1.

    In terms of severity of illness at onset, atypical diabetes of African Americans is the most severe; then MODY 1 and MODY 3, with progressive insulinopenia. The least severe is the nonprogressive MODY 2 glucokinase mutation.

    Atypical diabetes of African Americans over time is characterized by relatively unchanging fasting C-peptide levels, which is evidence that this is not a progressive disease.

    Metabolic Phenotyping: Discussion

    Dr. Hale asked Dr. Winter to comment on the physical phenotype of MODY and atypical diabetes of African American patients. Dr. Winter responded that the white MODY patients, as described by Dr. Fajans, were lean and lacked acanthosis nigricans. In contrast, approximately 40% of the atypical diabetes of African American patients are obese (likely reflecting the prevalence of obesity in the African American population generally). There was no dyslipidemia among the patients with atypical diabetes.

    Dr. Knowler questioned Dr. DiMartino-Nardi on her advocacy of the fasting glucose-to-insulin ratio for classifying patients with insulin resistance, and recommended that she also look at the glucose-to-insulin product, which he thought might have a greater correlation with insulin resistance.

    Dr. Rosenbloom commented that early kidney failure in young Japanese patients with type 2 diabetes for many years reinforced the point that children who develop type 2 diabetes in adolescence will have many complications, such as nephropathy, that will manifest when these individuals are in their 30s or 40s.

    Dr. Barbara Linder of NIDDK said that a lot of interesting information had been presented during the day and that it was now the task of the attendees to assimilate it and to return the next day to formulate the critical key pieces of information that are missing as related to epidemiology and diagnostic criteria. Dr. Linder suggested that the attendees set priorities for the research community.

    Epidemiology Summary

    Dr. Fagot-Campagna conducted a summary presentation of the epidemiology of pediatric type 2 diabetes. She noted that the disease was first described 20 years ago by Dr. P.J. Savage and co-workers reporting on the Pima Indians.

    Dr. Fagot-Campagna reviewed 9 case series, with a total of 578 cases of pediatric type 2 diabetes: 94% of the patients with type 2 diabetes were members of minority groups, and 56-92% of the patients had acanthosis nigricans. Elevated hemoglobin A1c levels, which are indicative of insufficient metabolic control of diabetes, were quite high: 10-13%.

    Next, Dr. Fagot-Campagna reviewed data on pediatric type 2 diabetes in the Pima Indians. At an average of 9 years after initial diagnosis, followup data showed little improvement and increased incidence of complications, such as elevated hemoglobin A1c, hypercholesterolemia, microalbuminuria, and macroalbuminuria.

    A review of the screening data for pediatric type 2 diabetes from population-based studies of American Indian youth revealed that there are many communities in which the prevalence is 1.4-5.1% of all those screened. Alarmingly, reported prevalence data for American Indians from registries, clinics, and hospitals tend to be tenfold lower, around 0.12-0.45%.

    Dr. Fagot-Campagna summarized data, including some presented at the meeting, demonstrating that the proportion of newly diagnosed pediatric diabetes cases that are type 2 ranges up to 46% in some minority communities.

    There was then a discussion about the need to capture data on Asian American children (e.g., Korean Americans in Chicago, Chinese Americans in San Francisco) given the anecdotal evidence of increased type 2 diabetes and obesity in these populations.

    Metabolic Phenotyping Summary

    Dr. Rosenbloom summarized the presentations on metabolic phenotyping of pediatric type 2 diabetes. He noted that as long ago as 1970 obesity in children was recognized as being associated with hyperinsulinism, glucose intolerance, and a form of diabetes different than typical type 1. He noted that the Bogalusa Heart Study has documented relative hyperinsulinemia in African-American compared to white youngsters.

    The key points in the various presentations on metabolic phenotyping were reviewed by Dr. Rosenbloom as follows:

  • Silva Arslanian, MD

    • Nondiabetic AA youth (prepubertal and pubertal) are insulin resistant compared to Caucasians (pre-pubertal & pubertal).

      - Increased first phase insulin release (FPIR)
      - Increased insulin secretion
    • Normally with puberty, insulin sensitivity declines (oxidative and non-oxidative).

    • Insulin sensitivity is inversely proportional to BMI/adiposity.

    • Increasing visceral adiposity decreases insulin sensitivity more rapidly than increases in subcutaneous adiposity.

    • Insulin sensitivity is decreased with hyperandrogenism (HA).

    • In nondiabetic children, family history of type 2 DM is associated with lower insulin sensitivity.

  • Joan DiMartino-Nardi, MD
    • Insulin sensitivity in girls with premature adrenarche who may be a model for syndrome X:

      - High frequency of obesity

      - acanthosis nigricans (AN)
      - Positive family history of DM
      - Decreased insulin sensitivity
      - Decreased HDL
    • G/I <7 correlated with reduced insulin sensitivity by FSIR (minimal model)

    Phil Zeitler, MD

    • Adolescents with type 2 diabetes display:
      - central adiposity
      - high fat/low fiber diet
      - little/no exercise
      - binge eating
    • Families of children with type 2 diabetes display:
      - high risk of overweight with central adiposity
      - high fat/low fiber diet
      - little/no exercise
      - binge eating
      - increased prevalence of insulin resistance
    • Importance of dealing with type 2 diabetes in children by addressing the family environment and habits

    Ken Jones, MD

    • C-peptide at diagnosis may help differentiate type 1 from type 2 diabetes.
    • Differences between Anglo American and Mexican American children provide evidence of syndrome X by age 11 in the Mexican American children.
      NOTE: One important issue in non-type 1 diabetes in children concerns phenotypic heterogeneity: some children diagnosed with type 2 DM are asymptomatic while other children may present in frank DKA. Dr. Willi has studied insulin-requiring diabetes in AA children that presents with ketosis/DKA

    Steven Willi, MD

    • Type 2 DM in AA youth may be ketotic or present with DKA.
    • After stabilization, fasting insulin is elevated; yet glucose-stimulated FPIR is absent, while the response to glucagon is preserved or above normal.
    • Euglycemic hyperinsulinemic clamp studies display insulin resistance even when diabetic and nondiabetic children are matched for obesity.

    William Winter, MD

    • MODY is differentiated from type 2 diabetes on the basis of autosomal dominant pattern of early-onset diabetes in relatives.
    • Classic MODY is nonketotic, mild, and is often diagnosed only by OGTT.
    • Atypical diabetes mellitus of African Americans is a subtype of MODY that presents identically to type 1 diabetes and progresses to a non-insulin dependent type 2 -diabetes-like state.
    • Five molecular mutations have been described that can cause MODY:
      - four are transcription factor mutations
      - one is due to glucokinase mutations
    • MODY2 and ADM are not progressive whereas MODY1 and MODY3 display a progressive decline in beta cell function with advancing age.
    • Studies using the stepped-glucose infusion protocol demonstrate variable metabolic phenoptyes among MODY 1, 2, and 3. Sustacal tolerance testing in ADM demonstrates that this disease involves defective beta cell secretion of insulin. IVGTT in ADM shows loss of FPIR.
    • All forms of MODY are characterized by insulinopenia and a lack of insulin resistance in the absence of obesity. This contrasts with the high frequency of insulin resistance in children diagnosed with type 2 diabetes.


    Dr. Hale led the discussion period. He asked Dr. Linder, the program organizer, what her goals were for this discussion period. Dr. Linder stated that the attendees should reach consensus on important gaps in knowledge.

    Dr. Hale then presented a series of candidate consensus statements and asked the attendees to comment on them.

    Consensus Statement 1.

    Prevention of diabetes is essential, and treatment is a major concern; another conference is needed to address the issues of treatment and prevention.

    Dr. Jones said that with patient and family compliance, this disease is very easily treatable, as evidenced by normal hemoglobin A1c. Dr. Zeitler said that, in his experience, there is a real problem with patient and family noncompliance. Dr. Lipton noted a real lack of understanding of this disease in the general practitioner/family physician community.

    Consensus Statement 2.

    Metabolic phenotyping is essential to our understanding of pediatric diabetes and the related co-morbidities.

    Dr. Winter called for a long-term longitudinal prospective study, one that starts with a group of nondiabetic children who may nonetheless be at risk for diabetes and defines the natural history of the disease. Metabolic phenotyping will play a key role in this prospective study.
    An attendee observed that heterogeneity is not only just in the phenotype but also in the disease stage and that more effort should be put into characterizing the stages of the disease, through characterization of the natural history.

    Consensus Statement 3.

    Standardization of the metabolic phenotyping protocol across ethnicities is (a) desirable and (b) feasible, but not mandatory.

    Dr. Rosenbloom commented that there might be a role here for an NIH multicenter study of various ethnic populations using standard protocols. A comment was made that standardization also depends on the efforts of the reagent manufacturers and of organizations such as the American Diabetes Association. It was also recommended that proinsulin testing be considered, in that changes in the proinsulin-to-insulin ratio are indicative of beta-cell failure.

    Consensus Statement 4.

    Standards are needed for anthropometric measurements in youth.

    There was also consensus on the need for more research on the development of standard tables and curves (standard deviations, percentiles, etc.) for these anthropometric measurements (e.g., percentage visceral fat by age, sex, and pubertal status).

    Consensus Statement 5.

    There is a need for standardized terminology.

    What is meant by "obesity," "adolescent," "pubertal?" Ethnic descriptors are also problematic and need to be standardized. The need for better genetic markers was also expressed. The observation was made that Tanner staging of puberty can be accurately determined by self-reporting and is just as reliable as physical examination.

    Consensus Statement 6.

    A large registry of children with diabetes would be a useful research tool.

    Comments were made that a registry is useful because it identifies patients who then can be enrolled in research studies that can further help describe the natural history of the disease. Dr. Knowler expressed doubts about large registries; for instance, they are often based on insulin treatment, and they thus exclude many nontreated and undiagnosed diabetic children. Dr. Rosenbloom remarked that a large national registry can be quite expensive and is probably not cost-effective.

    There were several comments that cohort registries would be very useful and perhaps more practical than national registries. Dr. Dean in particular advocated a cohort registry of the small percentage of type 2 cases that are prepubertal.

    Consensus Statement 7.

    Longitudinal studies would substantially enhance our understanding of diabetes and comorbidities in youths.

    Several attendees endorsed the continued tracking of diabetic children once they become adults.

    Consensus Statement 8.

    Risk factors for type 2 diabetes should be defined, publicized to both physicians and the lay public, and used in screening.

    Dr. DiMartino-Nardi proposed screening for gestational diabetes, given the data demonstrating that in utero exposure to diabetes is an important risk factor. Dr. Rosenbloom asserted that if "screening" is really case-finding through testing of at-risk populations, it will be considered cost-effective by HMOs. There was also the concern that unless case discovery and early intervention are proved to lower morbidity, the HMOs will not pay for them.

    Dr. Rosenbloom then read a list of questions on pediatric type 2 diabetes that a committee of the American Diabetes Association will address at a meeting in October 1999:
      • What is the classification of diabetes in children and youth?
      • What is the epidemiology of diabetes in children and youth?
      • What is the pathophysiology of type 2 diabetes in children and youth?
      • Who should be tested?
      • How should children and youth with type 2 diabetes be treated?
      • Can type 2 diabetes in children and youth be prevented?
      Dr. Lipton remarked that screening and case-finding for hypercholesterolemia in at-risk children has wide acceptance and that these activities will also identify children with type 2 diabetes, who share the same risk factors.

    Consensus Statement 9.

    True prevalence and true incidence rates for pediatric diabetes must be obtained.

    Dr. Fagot-Campagna commented on the large number of missed cases (e.g., adolescent gestational diabetes) and misclassified cases and the unreliability of prevalence data. She also stated that prevalence rates for diseases such as pediatric type 2 diabetes tend to be very low (e.g., 0.2%) and have large error bars, thus limiting their utility.

    A recommendation was made to identify and validate risk factors that bring the prevalence rate up to 3-5%, which would be more acceptable. Dr. Lipton then recommended two-stage screening, in which the first stage would use an inexpensive questionnaire based on validated risk factors and the second stage would involve diagnostic tests. Only those individuals shown to be at risk in the first stage would proceed to the second stage.

    An observation was made that hemoglobin A1c determination might be more appropriate for prevalence studies.

    Dr. Flegal asserted that prevalence rates based on a small number of confirmed cases (12, for example) are simply not statistically significant.

    Dr. Jones predicted that cost-effective screening and case-finding in children can have an important effect on combating type 2 diabetes in adults, who have prevalence rates much higher than those of children but whose illness may start in childhood.

    Dr. Fagot-Campagna recommended screening the offspring of parents with pediatric type 2 diabetes. Dr. Arslanian recommended screening all children with BMI greater than the 85 percentile as a first step and use this group to get valid prevalence data. Others suggested screening for hyperlipidemia in the same population at the same time.

    Dr. Hale then directed the discussion toward identifying which blood tests make the most sense to use in large-scale screening. Dr. Rosenbloom asked the attendees if hemoglobin A1c would be a useful screening test for pediatric type 2 diabetes. Dr. Hale responded that in an in-school study, he found hemoglobin A1c to be worthless and that glucose determination was a better marker. Dr. Knowler reported some recent data of his that suggested that hemoglobin A1c is an extremely good preliminary test for diabetes, using a cutoff of greater than 7% hemoglobin A1c as indicative of 90% probability of diabetes in his Native American population. Dr. Burghen recommended routine office-based urinalysis for screening. As Dr. Hale mentioned, the Japanese screen their children by using universal urine testing.

    Dr. Fagot-Campagna recommended that the attendees consider using a commercial dry blood spot technique that will soon be readily available on the market.

    Dr. Hale then directed the attendees to discuss fasting vs. random glucose testing. They agreed that for many routine blood tests (glucose, hemoglobin, blood lipids) random testing is as good or better than fasting testing and is certainly less problematic.

    Dr. Hale asked for the attendees' opinion on the oral glucose tolerance test for use in population-based screening. The consensus was that the test is very useful for detecting diabetes cases that are missed by other methods but that there are problems with cost and patient acceptance. Alternatives such as Sustacal need validation data.

    Dr. Lipton predicted that using any screening test more involved than a random finger stick will result in an unacceptable loss of test participants.

    Dr. Hale opined that one of the main gaps in current knowledge is what are normal (e.g., 50 percentile) values for these tests for various pediatric age groups.

    Dr. Hale concluded by saying that population screening will help identify a pre-diabetic group of children who can then be observed longitudinally.

    Dr. Hale thanked Dr. Linder and Dr. Wells and their staff for organizing such a successful meeting.

    The meeting adjourned at 12 p.m. on July 21, 1999.

    Approved by:___________________________________ Date:__________________
    Richard Eastman, M.D., Chairman
    Diabetes Mellitus Interagency Coordinating
    Committee, NIDDK

    Approved by:___________________________________ Date:__________________
    Charles A. Wells, Ph.D., Executive Secretary
    Diabetes Mellitus Interagency Coordinating
    Committee, NIDDK

    Page last updated: October 02, 2007

    General inquiries may be addressed to: Office of Communications & Public Liaison
    Building 31. Rm 9A06
    31 Center Drive, MSC 2560
    Bethesda, MD 20892-2560
    For information about NIDDK programs: 301.496.3583

    The National Institutes of Health   Department of Health and Human Services is the U.S. government's official web portal to all federal, state, and local government web resources and services.  HONcode Seal - Link to the Health on the Net Foundation