Northwestern NUgene Project: Type 2 Diabetes

The ability to correlate genetic variation with disease susceptibility and response to drug therapy depends on genotype or sequence analysis of large numbers of richly characterized DNA samples. Eight years agoWe are a part of NHGRI's electronic Medical Records and Genomics (eMERGE) Network, whose goal is to conduct genome-wide association studies in thousands of individuals using EMR-derived phenotypes and DNA from linked biorepositories. For eMERGE, Northwestern University (NU) is studying type 2 diabetes as a phenotype. In addition, in order to explore race differences in the prevalence of type 2 diabetes, NU collaborated with Vanderbilt University to study a mix of both Caucasian and African-Americans.

Northwestern University: In 2002, Northwestern committed to the development of a DNA repository to serve as a platform for the identification and validation of genotype-phenotype associations that will impact healthcare. The NUgene Project is a repository with longitudinal medical information from participating patients at affiliated hospitals and outpatient clinics from the Northwestern University Medical Center. Participants' DNA samples are coupled with data from a questionnaire (2 versions were used, 1 before and 1 after February 2006, both are included) and continuously updated data from our Electronic Medical Record (EMR) representing actual clinical care events. Northwestern has a state-of-the art, comprehensive inpatient and outpatient EMR system of over 2 million patients. NUgene has broad access to participant data for all outpatient visits as well as inpatient data via a consolidated data warehouse. NUgene participants consent to distribution and use of their coded DNA samples and data for a broad range of genetic research by third-party investigators.

Vanderbilt University: BioVU, Vanderbilt's DNA databank, is an enabling resource for exploration of the relationships among genetic variation, disease susceptibility, and variable drug responses, and represents a key first step in moving the emerging sciences of genomics and pharmacogenomics from research tools to clinical practice. BioVU acquires DNA from discarded blood samples collected from routine patient care. The biobank is linked to de-identified clinical data extracted from Vanderbilt's EMR, which forms the basis for phenotype definitions used in genotype-phenotype correlations.

• Study Weblinks:
  • The eMERGE Network
  • NUgene
• Study Design:
  • Case-Control
• Study Type:
  • Case-Control
• dbGaP estimated ancestry using GRAF-pop
• Total number of consented subjects: 3563
• Subject Sample Telemetry Report (SSTR)

• BioProject
• BioSample
• MeSH

T2DM CASE SELECTION FLOWCHART BELOW

Footnotes for T2D case diagram:
* Abnormal labs: Random glucose > 200mg/dl, Fasting glucose > 125 mg/dl, or hemoglobin A1c ≥ 6.5%.
** Clinician-entered diagnoses: Encounter or problem list diagnoses only (all other diagnoses in this diagram could also include diagnoses in the medical history)

T2DM CONTROL SELECTION FLOWCHART BELOW

Study Population
Suitable participant DNA samples were selected from the NUgene and BioVU biobanks, including both Caucasian and African-American populations.

Introduction
The cases and controls have been defined to avoid confounding by inclusion of cases with type 1 diabetes and as much as possible of controls at risk for type 2 diabetes which has not, as yet, manifested itself. By doing this, a potential source of bias has been introduced in that type 2 diabetic subjects who are treated with insulin alone have been excluded, although diabetic subjects on insulin together with one of the diabetes medications listed above are eligible for inclusion. This approach may select against type 2 diabetic subjects with more significant degrees of pancreatic beta cell failure.

Challenges

1. Potential case contamination with T1DM (Type 1 Diabetes Mellitus) and Mature Onset Diabetes of the Young (MODY) patients.
2. Potential control contamination with cases. However, an ICD9 code for T2D is likely for diet controlled patients. Also, the family history exclusion was added to reduce likelihood that patients were too young to have developed the disease yet.
3. Restrictions imposed by inclusion criteria for cases. One difficult area is the problems presented by patients on insulin alone with an ICD9 code for type 2 diabetes, as some of these patients could represent individuals with type 1 diabetes which has been misclassified as type 2 diabetes because of age of onset, etc. To address this, we have identified as cases, patients who are on insulin alone, AND: have been on a type 2 diabetes medication in the past, or do not have a type 1 DM diagnosis, but have at least two visits (on different dates) with the type 2 DM diagnosis in the problem list or in the encounter diagnosis.
4. Avoiding cases who have medication (e.g., steroid)-induced hyperglycemia.

Identification of T2D Cases: 2 groups
Neither group should have T1D diagnosis codes (ICD-9 250.x1 or 250.x3)

1. Identification of patients who already have a T2D diagnosis
   • Step 1: Include patients with Type 2 Diabetes diagnosis based on ICD9 codes (excluding those with ketoacidosis codes)
     
     Table 1: Type 2 Diabetes ICD9 codes meeting inclusion criteria.

     Description for type 2 diabetes codes used	ICD9 Code
     Diabetes with other coma	250.3
     	250.32
     Diabetes with hyperosmolarity	250.2
     	250.22
     Diabetes with unspecified complication	250.9
     	250.92
     Diabetes with other unspecified manifestation	250.8
     	250.82
     Diabetes with peripheral circulatory disorder	250.7
     	250.72
     Diabetes with neurological manifestations	250.6
     	250.62
     Diabetes with opthalmic manifestations	250.5
     	250.52
     Diabetes with renal manifestations	250.4
     	250.42
     Diabetes mellitus without mention of complication	250
     	250.02

   
   
   • Step 2: Exclude patients (currently) treated only with insulin AND have never been on a type 2 diabetes medication, and: diagnosed with T1DM, or even if not diagnosed with T1DM, diagnosed with T2DM on < 2 dates in an encounter or problem list.
     
     Table 2. Prescribed type 2 diabetes medications meeting patient inclusion criteria.

     Drug class	Brand name	Generic name
     Sulfonylureas		acetohexamide
     Sulfonylureas		tolazamide
     Sulfonylureas	Diabinese	chlorpropamide
     Sulfonylureas	Glucotrol	glipizide
     Sulfonylureas	Glucotrol XL	glipizide
     Sulfonylureas	Micronase	glyburide
     Sulfonylureas	Glynase	glyburide
     Sulfonylureas	Diabeta	glyburide
     Sulfonylureas	Amaryl	glimepiride
     Meglitinides	Prandin	repaglinide
     Meglitinides	Starlix	nateglinide
     Biguanides	Glucophage	metformin
     Thiazoldinediones	Avandia	rosiglitazone
     Thiazoldinediones	ACTOS	pioglitazone
     Thiazoldinediones		troglitazone
     Alpha-glucosidase inhibitors	Precose	acarbose
     Alpha-glucosidase inhibitors	Glyset	miglitol
     DPPIV inhibitor	Januvia	sitagliptin
     Injectables	Byetta	exenatide

     
     Table 3*. Prescribed medications meeting patient exclusion criteria unless one or more of the medications listed above is also prescribed:

     Drug class	Brand name	Generic name
     Injectables	Insulin	Insulin
     Injectables	Symlin**	Pramlintide
     Diabetic Insulin Supplies

     * Limits potential case contamination with T1D patients.
     ** Exclude if patient is on this alone or in combination with insulin only.
     
     Table 4. ICD9 codes to exclude type 1 diabetics.

     Description for type 1 (juvenile) diabetes codes used	ICD9 Code
     Diabetes with other coma	250.31
     	250.33
     Diabetes with hyperosmolarity	250.21
     	250.23
     Diabetes with unspecified complication	250.91
     	250.93
     Diabetes with other unspecified manifestation	250.81
     	250.83
     Diabetes with peripheral circulatory disorder	250.71
     	250.73
     Diabetes with neurological manifestations	250.61
     	250.63
     Diabetes with opthalmic manifestations	250.51
     	250.53
     Diabetes with renal manifestations	250.41
     	250.43
     Diabetes mellitus without mention of complication	250.01
     	250.03
     Diabetes mellitus with ketoacidosis	250.11
     	250.13

   
   
2. Identification of patients who do not yet have a T2D diagnosis
   • Step 1: Include patients with hemoglobin A1C lab value ≥ 6.5%, fasting glucose > 125 mg/dl or random glucose > 200 mg/dl AND prescribed one of the medications (or combinations thereof) listed in Table 2.

Identification of T2D Controls:

Patients must meet all of the following criteria:

1. Have had at least 2 clinic visits (face-to-face outpatient clinic encounters).
2. Have not been assigned an ICD9 code for diabetes (type 1 or type 2) or any diabetes-related condition (See codes from Tables 1, 4, and 5)

Table 5. ICD9 codes to exclude potential controls (in addition to Table 1).

Description	ICD9 Code
Diabetes mellitus type 1 & 2	250.xx
Impaired fasting glucose	790.21
Impaired oral glucose tolerance test	790.22
Abnormal glucose not otherwise specified	790.2, 790.29
Abnormal glucose during pregnancy	648.8x
Gestational diabetes	648.0x
Glycosuria	791.5
Dysmetabolic syndrome X	277.7
Family history of diabetes mellitus	V18.0
Screening for diabetes mellitus	V77.1



3. Have not been prescribed insulin or Pramlintide (See Table 3), or any medications for diabetes treatment (See Table 2), or diabetic supplies such as those for medication administration or glucose monitoring.
4. Do not have a reported (random or fasting) blood glucose ≥ 110mg/dl and have had at least 1 glucose measurement
5. Do not have a reported hemoglobin A1c ≥ 6.0%
6. Do not have a reported family history of diabetes (type 1 or type 2)

Type 2 Diabetes Time Line

• October 2002 - NUgene became operational
• January 2008 - Began T2DM phenotyping process
• May/June 2009 - Selected samples to be genotyped
• July 2010 - Samples shipped and received at The Broad Institute
• 2010 - Genotyping of samples complete

• Primary Phenotype: Diabetes Mellitus, Type 2

• Principal Investigator
  • Rex Chisholm, PhD. Northwestern University Feinberg School of Medicine, Chicago, IL, USA.
• Institute
  • National Human Genome Research Institute, National Institutes of Health, Bethesda, MD, USA.
• Funding Source
  • U01-HG004609. National Institutes of Health, Bethesda, MD, USA.
• Genotyping Center
  • Broad Institute of Harvard and MIT, Cambridge, MA, USA.
• Funding Source for Genotyping
  • U01-HG004424. National Human Genome Research Institute, National Institutes of Health, Bethesda, MD, USA.