Presentation by Eric Ruggieri

1
Presentation by Eric Ruggieri December 18th, 2007
2
Outline

• Background and Terminology
• Helgason et al. Refining the impact of TCF7L2
  gene variants on type 2 diabetes and adaptive
  evolution
• Lyssenko et al. Mechanisms by which common
  variants in the TCF7L2 gene increase risk of type
  2 diabetes

3
Background and Terminology

• Two types of Diabetes Type 1 (Juvenile Diabetes)
  and Type 2
• Juvenile Diabetes is the result of the body
  failing to produce insulin
• Possible consequence Ketoacedosis (diabetic
  coma) dangerously high levels of ketones in the
  body, especially the bloodstream

4
What is Type 2 Diabetes?

• The more common of the two types
• If glucose builds up
• Cells become starved for energy
• High Sugar levels can damage kidneys, eyes,
  nerves, heart

5
Beta Cells and Islets of Langerhans

• Beta Cells make and release insulin for the body
  within the pancreas
• characterize the two types of diabetes in terms
  of problems related to Beta cells
• Type 1 destruction or dysfunction of the insulin
  producing Beta cells by the cells of the immune
  system.
• Type 2 the Beta cells have degraded over time.

6
Glucagon the balancer of insulin

• Hormone glucagon produced by alpha cells in
  pancreas
• Released by body when blood sugar levels are low
  (Hypoglycemia)
• Causes the liver to convert stored glycogen into
  glucose - released into the bloodstream.
• Process is known as Hepatic Glucose Production,
  or net release of glucose from the liver.

7
Incretins

• Gastrointestinal hormone that causes an increase
  in the amount of insulin released from the Beta
  cells
• Occurs after eating, but even before the blood
  glucose levels begin to elevate
• Also involved in inhibiting the release of
  glucagon from the alpha cells of the Islets
• GLP1 (Glucagon Like Protein) and GIP (Gastric
  Inhibitory Peptide)
• rapidly inactivated by the enzyme dipeptidyl
  peptidase 4 (DPP-4).

8
Outline

• Background and Terminology
• Helgason et al. Refining the impact of TCF7L2
  gene variants on type 2 diabetes and adaptive
  evolution
• Lyssenko et al. Mechanisms by which common
  variants in the TCF7L2 gene increase risk of type
  2 diabetes

9
Premise to the Study

• Previous study identifies three markers
  associated with type 2 Diabetes (T2D)
• composite allele X of micro satellite DG10S478
• T alleles of SNPs rs12255372 and rs7903146
• All three of these markers are located within a
  64kb block of strong Linkage Disequilibrium (LD)
  containing exon 4 and two large flanking introns
  of the 217kb TCF7L2 gene on chromosome 10
• highly correlated in populations of European
  ancestry.
• So which SNP has the best association?
• Larger study conducted than in past
• West African group included genetically diverse

10
Which SNP is it?

• Whats more interesting

11
Phylogenetic Reconstruction
12
Two tests for Positive Selection of HapA

• FST the observed proportion of the overall
  genetic variation due to difference between
  groups.
• Assumes neutral evolution
• allele and haplotype frequency differences
  between populations are shaped by the
  counteracting forces of genetic drift and
  mutation
• So, the range of outcomes is constrained by
  demographic history of the populations.
• The results of the test were consistent with a
  positive selective sweep of HapA in East Asians
  and no selection (or less intense selection) in
  the other two HapMap groups.

13

• (Long Range Haplotype (LRH) test - deals with the
  time required for background mutation and
  recombination to break down linkage.
• neutral model is assumed
• comparison is made to the diversity expected for
  a neutral variant with the same frequency
• Common alleles with unusually low diversity at
  linked sites and/or slow decay of LD with
  increasing physical distance represent likely
  candidates for recent positive selection.
• Test indicated positive selection for HapA in all
  three HapMap groups (East Asians, Europeans and
  Africans), when compared with the rest of the
  genome, especially in East Asians.

14
What is the age of HapA

• Researchers were able to ballpark the age of HapA
• 11,933, 8,401, and 4,051 years for CEU, East
  Asian, and YRI HapMap groups respectively
• Dates coincide with the onset of agriculture in
  the geographic regions represented by the HapMap
  groups
• But why?

15
An Association between TCF7L2 and BMI?

• BMIbody mass index
• Previous report of a nonsignificant negative
  association between HapBT2D and BMI in T2D
  patients
• Performed a test of association on both HapBT2D
  and HapA with BMI in 5 diabetes groups and 6
  control groups.
• Results of test are not independent as 80 of all
  haplotypes in populations of European ancestry
  are one of these two alleles
• For the 20 of alleles that are not one of the
  two tested, the estimated effect on BMI is
  intermediate, but there is not sufficient
  statistical power to distinguish it from the
  effects of HapA or HapBT2D
• Overall, the association between TCF7L2 variants
  and BMI appears real, but modest in the controls
  and the researchers had no way to distinguish
  whether the effect is driven by HapA, HapBT2D, or
  whether they exert opposing effects on BMI.

16
(No Transcript)
17
Key Points

• HapA is associated with increased BMI (1.4 per
  copy, P0.0014) whereas HapBT2D is associated
  with decreased BMI (-1.3 per copy, P0.0016).
• Results of the control groups show weaker
  association in the same direction as the affected
  individuals
• Estimated effects of the haplotypes on BMI are
  consistently stronger for male affected
  individuals and male controls

18
An Apparent Paradox?

• HapBT2D is associated with both T2D and reduced
  BMI
• However, BMI is a known major risk factor for T2D
• Is it possible that individuals who contract
  diabetes through HapBT2D may have a different
  disease than those who acquire it through
  obesity?
• Since HapA is associated with increased BMI
• Are these two variants are pulling people towards
  opposite ends of the BMI spectrum, but to the
  same end in Type 2 Diabetes?
• i.e. Do there exist two distinct routes to a
  common end, routes that have opposing
  physiological effects?
• At this time, HapBT2D remains the only clear-cut
  risk factor for T2D at this locus

19
Metabolic Effect of Genetic Variants of TCF7L2?

• Recent study indicated a reduced secretion of
  insulin in HapBT2D carriers with T2D
• Therefore, analyze the effects of HapA and
  HapBT2D on 14 metabolic traits
• Two of the 14 traits, the fasting concentrations
  of the hormones ghrelin (decrease) and leptin
  (increase), showed nominally significant
  differences by HapA copy number in males.
• No such associations were seen for HapBT2D.
• Ghrelin and Leptin are involved in the short-term
  neuroendocrine regulation of appetite and the
  long-term regulation of fat storage and energy
  metabolism.
• In the long-term, the fasting levels of these two
  hormones can also be strongly influenced by BMI
  as weight gain can decrease the basal level of
  ghrelin and increase the basal level of leptin.
• So why was HapA positively selected?
• Variation in energy metabolism is evidently
  important for the survival of the human race
• HapA began its sweep through the population
  around the same time as the transition to
  agriculture in each of the geographic regions
  represented. Coincidence?
• More studies are needed to confirm these
  findings

20
(No Transcript)
21
Methods

• Type of Association Study Candidate Polymorphism
• There was no mention of Hardy-Weinberg
  Equilibrium Testing
• Phasing of haplotypes for the HapMap groups was
  done via the EM algorithm, in combination with
  family trio information for the CEU and YRI
  groups
• Likelihood ratio tests were used to calculate
  two-sided P-values for single-marker association
  to T2D.
• There was no need to tag SNPs, since only three
  were being tested.
• The relationship between BMI and HapA or HapBT2D
  was analyzed using multiple regression after
  adjusting for age and sex, with log(BMI) taken as
  the response variable.

22
Methods cont.

• Two statistical tests for positive selection
  performed FST and LRH. Both provided evidence
  for positive selection of HapA.
• The age of a haplotype was determined based on
  the formula EHH e -2rg
• Simcol simulation software used to generate
  samples under the neutral coalescent model in
  order to evaluate the statistical significance of
  values obtained from selection tests from the
  observed data.
• One major complication to the use of coalescent
  simulations is that a populations demographic
  history has a major impact in the expected
  patterns of diversity within the gene pool. To
  counteract this hurdle, the researchers simulated
  genetic data under a wide range of demographic
  scenarios (ex. Constant population size,
  population bottleneck, expansion, etc.).

23
Outline

• Background and Terminology
• Helgason et al. Refining the impact of TCF7L2
  gene variants on type 2 diabetes and adaptive
  evolution
• Lyssenko et al. Mechanisms by which common
  variants in the TCF7L2 gene increase risk of type
  2 diabetes

24
Premise of the Study

• Genetic variants of TCF7L2 have been associated
  with T2D and impaired Beta cell function
• mechanisms have remained unknown.
• Conducted a prospective study regarding the
  ability of common variants of TCF7L2 to predict
  future T2D
• Sought out mechanisms by which this occurs
• Scandinavian subjects were followed for up to 22
  years, genotyped for three SNPs and a subset of
  them underwent extensive metabolic studies.

25
Goals

• Can the variants of TCF7L2 predict T2D in the two
  study groups?
• Prospective study groups of 7061 individuals from
  Sweeden (Malmo Preventative Project) and 2651
  individuals from Finland (Botnia study)
• Explore Genotype-Phenotype correlations
• Study the influence on insulin, glucagon, GIP
  levels, incretin effects and hepatic and
  peripheral insulin sensitivity
• Does the risk genotype influence the
  transcription of TCF7L2 and insulin or glucagon
  secretion?
• Since transcription increases (See Goal 3)
  manually overexpress the gene via adenoviral
  vector in human islets and compare the results.

26
Results

• Of the 7,061 persons included in the Swedish MPP
  study, 1,422 developed diabetes
• Two SNPs were studied rs12255372 and rs7903146
• but since the two SNPs were in strong LD
  (D0.86) they capture essentially the same
  genetic information, the results were presented
  only for rs7903146
• Frequency of the risk T allele was significantly
  higher in diabetes converters than in
  non-converters (31.3 vs 25.2 Plt0.0001)
• Carriers of the risk genotypes CT/TT had a higher
  risk of future T2D than CC carriers (odds ratio
  1.58, 95 CI 1.38-1.81, Plt0.0001).
• In Botnia prospective study, similar results were
  found (odds ratio 1.61 95CI 1.14-2.27, P0.007).
  

27
(No Transcript)
28
Haplotype Notation

• Similar to the notation in previous paper
• HapA allele A at rs10885406 and allele C at
  rs7903146
• HapAB allele A at rs1088546 and either allele T
  or C at rs7903146
• HapB allele G at rs10885406 and allele T at
  rs7903146
• Claim Complete Linkage Disequilibrium

29
Risk Genotypes of TCF7L2 associated with Impaired
Insulin Secretion

• Early insulin response to an oral glucose
  tolerance test (OGTT), also called the
  insulinogenic index
• 1,038 individuals from Malmo (A)
• carriers of the risk CT/TT genotype (n481) had
  lower insulinogenic index than did carriers of
  the CC genotypes (9.3 5.5 vs 10.2 5.3,
  P0.0006).
• Insulin secretion adjusted for insulin
  sensitivity
• 7.15.5 vs 8.16.2, P0.005
• Results replicated in the Botnia study (B)
• Furthermore, the risk CT/TT genotypes showed a
  more severe deterioration in insulin secretion
  over time (C)

30
Arginine-Stimulated Insulin Response

• 250 subjects from MPP the acute insulin response
  (AIR) to arginine was significantly reduced at
  both 14mmol/l (6760 vs 8670mU/l, P0.02) and
  28mmol/l (107113 vs 155138, P0.009) of glucose
  in CT/TT versus CC genotype carriers with
  abnormal glucose tolerance
• No difference was noted in those with normal
  glucose tolerance.

31
(No Transcript)
32
Impaired Incretin Response

• TCF7L2 proposed to bind to promoter region of
  proglucagon gene
• Refers to a better glucose dependent insulin
  response to oral than to i.v. glucose due to the
  additional stimulation of insulin secretion by
  intestinal hormones such as GLP1 and GIP
• 20 lower insulin response to oral than to i.v
  glucose in the risk CT/TT than in the CC genotype
  carriers
• As expected, carriers of HapB had a more severe
  incretin defect than did carriers of HapAB and
  HapA
• The correlation coefficients in the insulin
  response to oral and i.v. glucose were
  significantly different between risk and non-risk
  genotype carriers (r0.68 vs r0.52, P0.007).
• The correlation was strongest in HapB carriers
  (r0.72, Plt0.0001)
• No difference in plasma concentrations of
  glucagon or GIP were observed between different
  genotype carriers
• Measurements for GLP1 were unavailable

33
Hepatic Insulin Resistance

• No change in whole body glucose uptake between
  risk and non-risk genotype carriers (Peripheral
  Insulin Resistance) D
• The rate of endogenous glucose production (EGP)
  was significantly higher in the risk genotypes
  E
• suggests an impaired suppression of basal hepatic
  glucose production either by impaired insulin
  response or an enhanced glucagon response.
• However, recall that there was no change in
  glucagon levels between the different genotype
  carriers.
• Step-wise increase in EGP in haplotypes HapA,
  HapAB, and HapB.

34
Increased Gene Expression in Human Islets

• SNPs in TCF7L2 are in non-coding regions
• authors hypothesize altered expression of TCF7L2
  in pancreatic islets to obtain observed effects
• The mRNA levels in pancreatic islets measured in
  7 type 2 diabetics and 15 nondiabetic human
  cadaveric organ donors
• Patients with T2D had reduced amounts of insulin
  and increased amounts of glucagon secreted
  compared to non-diabetic donors
• mRNA levels of TCF7L2 were 5-fold higher in human
  pancreatic islets from T2D than non-diabetic
  donors
• expression increased with the number of T alleles
  (P0.006)
• Expression of TCF7L2 was lowest in HapA, then
  HapAB, and highest in HapB

35

• Interestingly, although expression of the insulin
  gene is decreased in islets from T2D subjects
  compared to non-diabetics, there is a positive
  correlation between mRNA levels of TCF7L2 and the
  insulin gene (r0.76, P0.01).
• The correlation was strongest in non-diabetic
  CT/TT genotype carriers (r0.81, P0.01)
• This means While there may be an inverse
  correlation between expression of TCF7L2 and
  glucose-stimulated insulin release, there is a
  positive correlation between the expression
  levels of the two genes.
• May suggest TCF7L2 influences post-transcriptional
  events of insulin expression (such as mRNA
  degradation)
• Verified findings via adenovirus system
• Confirmed that overexpression of TCF7L2 in human
  islets alters glucose-stimulated insulin but not
  glucagon secretion.

36
Methods

• No mention of testing for Hardy-Weinberg
  Equilibrium.
• Genotype data and not haplotype data is used
  throughout the paper and so there was no need for
  phasing.
• With respect to missing data, no mention is made.
  Claimed average genotyping success rate of 98
  and concordance rate of 99
• SNP tagging was unnecessary as this was a
  Candidate Polymorphism Association Study
• The effects of genetic variants (risk vs non-risk
  genotypes) on developing T2D were estimated using
  Kaplan-Meier survival curves. The odds ratios
  for risk of developing T2D were calculated using
  logistic regression analyses adjusted for age at
  entry and time of follow-up, sex, BMI, and family
  history of diabetes. To test the differences
  between group means, an analysis of covariance
  (ANCOVA) was used. Which has to do with an F
  test, although not explicitly stated in the
  paper
• All statistical analyses were performed with
  Statistical Package for the Social Sciences
  version 14.0 (SPSS), Number Crunching Statistical
  Systems version 2005 (NCSS).
• Did they even know what the numbers being
  produced mean statistically?

37
Comments and Criticisms

• Sample size from the cadaver study
• lucky to get such small standard deviations of
  their data which enables possibility of
  significant P values
• Were all assumptions with respect to the
  hypothesis test met?

38
Criticisms cont.

• data set used to gather data about reduced
  incretin effects has a sample size of 1,038. Of
  that, 951 are men and 87 are women.
• Recall that previous paper noted differences in
  responses of men and women
• Near the end of the methods section, authors
  remark that two SNPs are in almost complete LD as
  D0.99 and R2 1
• This is impossible as R2 lt D
• While on this topic

39

• The authors claim this when creating notation for
  the different haplotypes, Recall
• HapB allele G at rs10885406 and allele T at
  rs7903146
• Claim Complete Linkage Disequilibrium within
  main body of text
• Complete Linkage Disequilibrium means D1. D1
  only when one of the 4 haplotypes is missing from
  a contingency table.
• From their definition of the haplotypes, it is
  obvious that all 4 are present
• In fact, it appears that they tried to calculate
  D on genotypes rather than on haplotypes, for
  which the measure was actually developed.
• Note This table appears in the Supplement, it is
  correct in saying almost complete LD, although
  they still make the mistake about R2

40

• Figures in the paper do not correspond to the
  data in the text
• Ex. Impaired Insulin Secretion
• 1,038 individuals from Malmo (A)
• carriers of the risk CT/TT genotype (n481) had
  lower insulinogenic index than did carriers of
  the CC genotypes (9.3 5.5 vs 10.2 5.3,
  P0.0006).
• Insulin secretion adjusted for insulin
  sensitivity
• 7.15.5 vs 8.16.2, P0.005
• Look at the error bars
• Standard. deviation vs standard error

41
Summary Key Findings

• Helgason et al
• The T allele of rs7903146 was determined to be
  the risk allele for T2D
• Phylogenetic reconstruction of this part of the
  genome shows two distinct lineages. The
  structure of the graph suggests that there was a
  positive sweep of HapA in East Asians.
• The statistical tests FST and LRH confirm this
• Age of HapA haplotype is placed at the start of
  agriculture in each of the geographic regions
  represented by the HapMap groups
• There was an association between the variants of
  TCF7L2 and BMI
• HapA increased BMI, HapBT2D showed a negative
  association
• In both cases, the control groups showed a weaker
  association, but in the same direction
• The results were stronger for male subjects
• Could this mean that there are multiple paths to
  the same end (T2D)?
• Metabolic effects There was a change in the
  concentration of Ghrelin and Leptin in males only
  per HapA copy number
• These hormones are partially in control of
  appetite, fat storage, and energy metabolism.

42
Summary Key Findings, cont.

• The T allele of SNP rs7903146 of TCF7L2 strongly
  predicts T2D
• The increased risk of T2D is likely due to the
  impairment of insulin secretion
• impairment in both glucose and argentine-stimulate
  d insulin secretion
• Carriers of the risk T allele showed a weaker
  response to oral than to i.v. glucose.
• The normal incretin effect is stronger because of
  the additional hormones stimulated in the
  intestine.
• Risk genotype carriers showed and enhanced rate
  of EGP (endogenous glucose production) -
  primarily the liver
• A consequence of impaired insulin secretion or
  enhanced glucagon secretion
• However, glucagon concentrations did not differ
  between the differing genotype carriers
• Expression of the TCF7L2 gene was 5-fold higher
  in islets of patients with T2D
• non-diabetic carriers of the TT genotype had the
  highest expression of TCF7L2 in islets
• There was a strong positive correlation between
  the expression of TCF7L2 and the insulin gene,
  but a negative correlation with
  glucose-stimulated insulin secretion.
• relationship was strongest in TT carriers
• no relationship to the levels of glucagon.
• By over expressing the TCF7L2 gene in
  non-diabetic human islets through the use of an
  adenovirus system, support was found for the
  primary effect of this correlation with respect
  to the diabetic state.
• i.e. the correlation is not an effect of
  diabetes, but a possible cause.