Fasting Glucose Levels

1
Fasting Glucose Levels Incident Diabetes
Mellitus in Older Non-Diabetic Adults Randomized
to Three Different Classes of Antihypertensive
TreatmentA Report from ALLHATJ. Barzilay, M.
Alderman, B. R. Davis, J. A. Cutler, S. L.
Pressel, P. K. Whelton, J. Basile, K. L.
Margolis, S. T. Ong, L. S. Sadler, J. Summerson
Archives of Internal Medicine In Press
2
Context

• Elevated glucose levels have been reported with
  use of diuretic therapy in the treatment of
  hypertension.
• The clinical significance of this is uncertain.

3
Objective

• Among participants who are non-diabetic at
  baseline
• Compare the effects of 1st-step antihyper-tensive
  drug therapy with chlorthalidone, amlodipine, or
  lisinopril on fasting glucose (FG) levels and
  incident diabetes
• Determine risks for CV and renal disease
  associated with elevated FG and incident diabetes
  in the three treatment groups.

4
Design Population

• Post hoc analyses of ALLHAT population
  (hypertensive, age ?55 years, gt1 other CVD risk
  factor)
• Subgroup that was nondiabetic by history at
  baseline, plus
• FG lt 126 mg/dl, or
• Random glucose lt110 mg/dl
• Follow-up mean 4.9 years

5
Derivation of Cohortfor Analysis
42,418
Total ALLHAT Participants
33,357
Randomized to C, A, or L
21,294
Nondiabetic by history
18,411
FGlt126 or RGlt110 mg/dl
14,005
1 follow-up blood samples
(fasting or nonfasting)
9,802
1 follow-up FG values
Duration of at least 8 hours
6
Baseline Characteristics
7
Fasting Glucose




plt.05 compared to chlorthalidone
8
Changes in Fasting Glucose




plt.05 compared to chlorthalidone
9
Follow-up Fasting Glucose126 mg/dL




plt.05 compared to chlorthalidone
10
Potential Confounders and Mediators

• ß-blockers decrease insulin sensitivity and
  therefore may increase the risk of DM.
• Potassium depletion appears to be a major
  intervening factor between thiazide treatment and
  dysglycemia.
• Statin therapy may decrease risk of incident DM.

11
Medication at 2 Years
12
Diabetes Incidence Logistic Regressions
13
Effect of Change in Fasting Glucose on ALLHAT
Endpoints(Cox Regressions Beginning at 2 Years)
14
Effect of Change in Fasting Glucose on ALLHAT
Endpoints(Cox Regressions Beginning at 2 Years)
15
Effect of Incident Diabetes on ALLHAT
Endpoints(Cox Regressions Beginning at 2 Years)
16
Effect of Incident Diabetes on CHD Heart
Failure by Treatment Group(Cox Regressions
Beginning at 2 Years)
17
Effect of Incident Diabetes on Combined CVD
ESRD by Treatment Group(Cox Regressions
Beginning at 2 Years)
18
Effect of Incident Diabetes on Total Mortality by
Treatment Group(Cox Regressions Beginning at 2
Years)
19
Incident Diabetes in ALLHAT Summary

• FG increased in all 3 treatment groups
• Differences between treatment groups were small
• For incident DM to 2 years, mean increase was 52
  mg/dl
• Follow-up FG and incident diabetes were highest
  in chlorthalidone, lowest in lisinopril
• Chlorthalidone has detrimental effect on FG?
• Lisinopril / amlodipine have neutral / protective
  effect on FG?

20
Effect of ?FG Incident Diabetes on Outcomes
Summary

• No significant overall effect of change in FG on
  any of the study endpoints in the combined
  treatment groups or the chlorthalidone group
  separately
• Incident DM increased risk of CHD
• Statistically significant for total group
  lisinopril
• In chlorthalidone group, increase in risk was
  smallest and not significant

21
Discussion of ALLHAT Findings

• Lisinopril group ?FG associated with ? risk of
  CCHD and CCVD incident DM associated with ? risk
  of CHD
• Lisinopril generally prevents ?FG
• Amlodipine group Incident DM associated with ?
  risk of total mortality
• Amlodipine does not generally raise glucose
  levels
• ?Participants with ?FG in these groups may have
  been very insulin resistant and at high risk for
  CV events

22
Discussion of ALLHAT Findings

• Low potassium did not significant increase the
  odds of developing DM
• Use of K supplements doubled from year 2 to year
  5
• Treatment differences in FG and DM decreased at
  years 4 and 6
• Sustained low K not captured in dataset
  prescription of K supplement may indicate this,
  and tends to be associated with DM
• Recent review thiazide-induced hyperglycemia
  should be anticipated and prevented by measures
  to preserve normokalemia and total body K.
  (Zillich et al. Hypertension. 200648219-224.)

23
Total Mortality () 14.3 yrs Follow up
plt 0.05 vs no diabetes
SHEP-X Systolic Hypertension in the Elderly
Program extended follow-up. Kostis, et al. Am
J Cardiol. 20059529-35
24
Cardiovascular Death () 14.3 yrs Follow up
SHEP-X Systolic Hypertension in the Elderly
Program extended follow-up. Kostis, et al. Am
J Cardiol. 20059529-35
plt 0.05 vs no diabetes
25
New diabetes and CVD risk Verdecchia 2004

• 795 treated HTs, median FU 6 yrs.
• Diuretic rx (low-mod dose HCTZ or CLTD)
  independently predictive of new diabetes.
• Adjusted RR (95 CI) of CVD-renal event (n63)
• --BL DM, 3.57 (1.65, 7.73)
• --New DM, 2.92 (1.33, 6.41)
• Results for specific regimens not given, only
  11 on diuretic/ß blocker alone.

Verdecchia et al. Hypertension 200443963-69.
age, 24h SBP, LVH.
26
Evidence from Previous Studies

• 15-y follow-up of 686 middle-age hypertensive
  adults treated with diuretic
• Diabetes at baseline significantly associated
  with CHD--RR 2.1 (1.1, 4.1)
• Incident diabetes was not significantly related
  with CHDRR 1.5 (0.4, 6.0).
• Samuelsson O, et al. Brit Med J 1996 313660-63.

27
Evidence from Previous Studies

• Cessation of long-term use of thiazide diuretics
  is associated with prompt improvement in FG
  levels
• Suggests that diuretics lead to elevated glucose
  levels by mechanisms different from those
  associated with DM
• Murphy MB, et al. Lancet 19822(8311)1293-95.

28
Evidence from Previous Studies

• Meta-analysis of ACE inhibitors ARBs
• Both decrease the risk of DM
• Neither reduces the odds of mortality, CV events,
  or cerebrovascular events vs control therapy
  e.g., thiazides and beta blockers
• Gillespie EL, et al. Diabetes Care 2005
  282261-66.

29
Strengths

• ALLHAT much larger than other studies
• ? statistical power
• Use of central biochemical laboratory
• Variety of practice environments

30
Limitations

• Misclassification of incident diabetes and not
  identifying impaired glucose tolerance could have
  diluted findings
• FU measures in ½ of cohort were non-fasting, and
  not used
• Data on diabetes medication use not collected
• Conclusions cannot be extrapolated beyond about 5
  years
• Other measures of glucose metabolism (e.g.,
  HbA1c, insulin levels) may have been helpful

31
Conclusions

• Treatment of hypertension with chlorthalidone was
  associated with small initial increase in FG
  increased risk of DM compared with amlodipine
  lisinopril.
• Differences in FG diminished over 5 years.
• No corresponding increase in risk of stroke,
  combined CVD, total mortality or ESRD over the
  period of follow-up.
• ? risk of CHD associated with ? DM not clearly
  identified in chlorthalidone arm

32
Perspectives on Incident Diabetes

• Assuming CCB is metabolically neutral, 85 (9.3
  vs 11.0) of DM at 4 years on chlorthalidone was
  not due to chlorthalidone
• Lifestyle intervention remains paramount

33
Conclusion

• Neither amlodipine- nor lisinopril-based
  treatment led to superior outcomes for any CVD
  endpoint.
• Both were inferior for prevention of heart
  failure
• While clinicians need to be aware of, and monitor
  patients for hyperglycemia, the totality of the
  evidence still supports the use of thiazide
  diuretics as preferred agents for prevention of
  cardiovascular disease in hypertensive patients.
• The relatively small detrimental metabolic
  effects of thiazide-type diuretic should not
  affect their preferred use in the management of
  hypertension.

34
EXTRA SLIDES
35
Diabetes and Hypertension Links

• Common antecedents
• Obesity
• Insulin resistance
• Treatment of one may impact the other

36
Diabetes Incidence - 4 Years -All
Participants(lt126 mg/dL at baseline)


plt.05 compared to chlorthalidone
JAMA 20022882981-2997
37
BP Meds and Glucose inRandomized Clinical Trials
Diuretic vs Placebo, Diuretic vs Beta-blocker
Padwal and Laupaci (Diabetes Care 27247-256)
38
BP Meds and Glucose inRandomized Clinical Trials
ACEI or ARB vs Placebo
Padwal and Laupaci (Diabetes Care 27247-256)
39
BP Meds and Glucose inRandomized Clinical Trials
ACEI or ARB vs Diuretic/Beta-blocker
Padwal and Laupaci (Diabetes Care 27247-256)
40
BP Meds and Glucose inRandomized Clinical Trials
Diuretic Beta-blocker vs CCB vs () ACEI
Padwal and Laupaci (Diabetes Care 27247-256)
41
CAVEATThe criterion for defining DM( gt 125
mg/dl) was chosen not based on CVD risk (a
complication not specific to DM). Rather the
criterion was based on a microvascular
complication specific to DM - retinopathy.
42
Fasting Glucose at 4 Yearsin Nondiabetic
Participants
43
Diuretic Useat 2, 4, and 6 Years