| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Diabetes Unit, Department of Clinical and Biological Sciences, University of Turin, San Luigi Gonzaga Hospital (F.C., M.Tra., K.B., E.F., M.C., G.A., M.Tro.), and Department of Public Health, University of Turin (A.P., G.C.), 10043 Orbassano, Turin, Italy
Address all correspondence and requests for reprints to: Prof. Mariella Trovati, Diabetes Unit, Department of Clinical and Biological Sciences, University of Turin, San Luigi Gonzaga Hospital, 10043 Orbassano (Turin), Italy. E-mail: mariella.trovati{at}unito.it.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Methods: In a population of 529 (284 men and 245 women) consecutive type 2 diabetic patients attending our diabetes clinic, we evaluated the relationships, corrected for cardiovascular risk factors and type of treatment, between cardiovascular events in a 5-yr follow-up and baseline values of hemoglobin A1c (HbA1c) and blood glucose measured: 1) after an overnight fast, 2) after breakfast, 3) after lunch, and 4) before dinner. Continuous variables were categorized into tertiles.
Results: We recorded cardiovascular events in 77 subjects: 54 of 284 men (19%) and 23 of 245 women (9.4%). Univariate analysis indicated that cardiovascular events were associated with increasing age, longer diabetes duration, and higher HbA1c and fibrinogen in men, and higher systolic blood pressure, albumin excretion rate, HbA1c, and all blood glucose values in women. Smoking was more frequent in subjects with events. When all blood glucose values and HbA1c were introduced simultaneously in the models, only blood glucose after lunch predicted cardiovascular events, with hazard ratio of the third tertile vs. the first and the second tertiles greater in women (5.54; confidence interval, 1.45–21.20) than in men (2.12; confidence interval, 1.04–4.32; P < 0.01).
Conclusions: Postprandial, but not fasting, blood glucose is an independent risk factor for cardiovascular events in type 2 diabetes, with a stronger predictive power in women than in men, suggesting that more attention should be paid to postprandial hyperglycemia, particularly in women.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The role of postprandial BG as an independent contributor to diabetes complications and the need to target it for prevention of cardiovascular events are a matter of intense debate. As exhaustively reviewed (13, 14, 15, 16), studies carried out mainly in the general population show that postchallenge BG predicts the incidence of cardiovascular events and mortality more than fasting BG: however, results obtained measuring BG after an oral glucose tolerance test (i.e. postchallenge or postload BG) cannot be extrapolated to the postprandial (i.e. after a meal) condition. The extent at which postchallenge BG reflects BG after a mixed meal is not well understood (14); therefore, postprandial and postload glucose concentrations should be kept clearly distinct (17).
As far as we know, in only one study, the Diabetes Intervention Study (DIS), the role of postprandial BG in the prediction of cardiovascular events in type 2 diabetes has been addressed: BG after breakfast, but not fasting BG, has been found to predict myocardial infarction and mortality in newly diagnosed type 2 diabetic patients (11). In 2001, the American Diabetes Association stated that whether postprandial hyperglycemia is an independent risk factor for cardiovascular disease is still controversial and requires additional studies (18). Because the equivalence between postchallenge and postprandial BG has been criticized, it is of major interest to provide additional evidence on the predictive role of postprandial BG in the diabetic population.
In the general population, cardiovascular mortality rate is two to five times greater in men than in women (19, 20). In contrast, hyperglycemia seems to influence cardiovascular mortality more strongly in women than in men. Actually, many studies show that both diabetes (4, 20) and asymptomatic hyperglycemia (3, 5, 7, 21) confer a greater relative risk of cardiovascular disease to women than to men, whereas other studies do not show significant gender-related differences (6). Some meta-analyses state that hyperglycemia is a stronger cardiovascular risk factor in women than in men (22, 23), whereas other state that gender-related differences disappear after adjustment for the main cardiovascular risk factors (24). Recently, the DECODE Study Group, by analyzing both fasting and postchallenge glucose concentrations from 14 prospective European cohorts, found that the multivariate adjusted hazard ratios for cardiovascular mortality in subjects who met the criteria for diabetes were 1.40 in men and 3.29 in women (25).
As far as we know, no study addressed the influence of gender in the predictive power of postprandial (not postchallenge) BG in patients with an established diagnosis of type 2 diabetes.
In the present study, carried out in patients affected by type 2 diabetes in regular clinical follow-up at our diabetes clinic, we investigated whether postprandial BG, either after breakfast or after lunch, predicts cardiovascular events more than fasting BG, and whether the relationships between BG and cardiovascular events differ in the two genders after correction for classical cardiovascular risk factors and type of treatment.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
HbA1c was evaluated in our laboratory by HPLC (Diamat, Bio-Rad Laboratories, Inc., Milan, Italy; normal range, 3.8–5.9%; mean value in a cohort of 200 healthy subjects, 4.8 ± 0.05%). In our clinic, BG profiles were measured as a part of the clinical routine in all type 2 diabetic patients. According to patient preference, they were carried out either at the hospital by trained nurses on the same days of the scheduled visits or at home by the patients themselves; patients on self-BG monitoring were asked to perform also one BG profile in a day very close to the scheduled visits.
Patients were consecutively enrolled in the study when they came to the hospital for a routine follow-up visit; on this occasion, we measured together all the main cardiovascular risk factors known at that time. To get BG values truly reflecting the real glycemic control, we considered BG profiles available for clinical purposes at the routine visit in which the cardiovascular risk factors and HbA1c were measured, i.e. 1) BG profiles measured by nurses at hospital in that visit day for patients not performing self-BG monitoring, and 2) BG profiles measured by patients themselves at home on one of the days immediately preceding the visit for patients performing self-BG monitoring. We did not consider BG measurements opportunistically performed when patients were worried for unexpectedly high or low BG values.
Profiles carried out in the hospital or at home contained the same number of BG measurements and, in particular, consisted of FBG, BGAB, BGAL, and BGBD. In 1995, the reflectance meter used both in the hospital and at home was Reflolux II (Roche, Mannheim, Germany), whose reliability in our laboratory has been previously described (27).
All remaining laboratory tests were determined at the central laboratory of San Luigi Gonzaga Hospital, which participates in the regional quality control protocol. In particular, total cholesterol and triglycerides were determined by standard automated method: HDL cholesterol after separation of non-HDL cholesterol and AER on 24-h urine collections by nephelometry (Beckman Coulter, Milan, Italy). Patients with signs of infection at the urine analysis had a urine culture test performed and AER determined again after adequate treatment of the infection. Fibrinogen was determined by the Clauss method. White blood cell counts were determined in a Coulter counter automated hemocytometer.
Table 1
shows the types of treatment. The 529 patients enrolled in the study in 1995 received during the follow-up the same treatment and care as the other patients of our diabetes clinic; in the year 2000, the follow-up was completed for 484 patients (91.4%). The patients gave informed consent for anonymous use of their clinical data for research purposes.
|
4,200,000 inhabitants), the region in northwest Italy that includes Turin (the capital, with about 1,000,000 inhabitants), was conducted from the date of enrollment until the end of 2000, and the first admission for the same causes recorded during active follow-up was evaluated (n = 11). 2) An individual record linkage using the individual fiscal code as the key linkage was conducted with the Turin historical population register on the subjects that were lost to active follow-up or who did not undergo a hospital admission in the Piedmont region (step 1) during the period considered (n = 34) to ascertain cases of death before access to hospital admission only for subjects resident in Turin. No death was observed in 12 subjects (of 34) resident in Turin during the period considered for which person-time was taken into account for the purpose of the study. For the purpose of the analysis, the remaining 22 subjects were considered lost at follow-up. A total amount of 77 cardiovascular events were observed: 66 identified from the clinical documentation and 11 by record linkage with hospital discharge database.
Statistical analysis
The association among baseline parameters, gender, and the occurrence of cardiovascular events was evaluated using the criteria described below. For numerical variables, the assumption of normality was firstly tested by the Shapiro-Wilkes test, which is the most appropriate test for small sample size. The Wilcoxon sum-rank nonparametric test was applied in the case of significance of the Shapiro-Wilkes test. Otherwise, after application of the folded F test to evaluate homogeneity of variances, a t test was applied and eventually corrected by the Cochrane and Cox tail modification for heterogeneity of variances (29). A Pearson 
2 test was applied to evaluate association between categorical factors. For tables where at least one cell showed a frequency of 5 or less, the Fisher exact test was applied. Associations between BG values were explored using 1) Wilcoxon rank-sum test to compare mean values, and 2) Pearson correlation coefficient to investigate the linear correlation between the measures.
Cox proportional hazards models (30) were used to analyze the independent role of each factor on cardiovascular outcomes. Sixteen subjects who died from causes other than cardiovascular events were considered for survival analysis by contributing to the total amount of person-time until the date of death. HbA1c, FBG, BGAB, BGAL, and BGBD were introduced into the models as determinants.
Age, known diabetes duration, previous cardiovascular events, smoking habit, type of treatment, BMI, systolic and diastolic blood pressure, ratio between total and HDL cholesterol (cholesterol ratio), triglycerides, AER, fibrinogen, and white blood cell count were used as possible confounding factors. Because one of the major aims of the study was to evaluate gender differences on the impact of BG on cardiovascular events, the analyses were conducted independently for the two genders.
The following strategy was used for model building. 1) The roles of factors not significantly associated with the outcomes in the bivariate analysis were evaluated by comparing models in which all the covariates, including these factors, were used in the models with models in which only factors significantly associated with outcomes were considered. No significant improvement in the likelihood ratio was observed; thus, the models were fitted with the covariates significantly associated with the outcome in the bivariate analysis (see Table 2) with the addition of type of treatment. Consequently, all models were fitted with age, known diabetes duration, previous cardiovascular events, systolic blood pressure, AER, fibrinogen, smoking habit, and type of treatment as covariates.
|
All variables were categorized into tertiles and introduced into the models as factors by the creation of dummy variables and comparing top tertiles of BG and HbA1c with tertiles 1 and 2 combined. In particular, median, interquartile range, and cutoff point for the third tertile of BG values (milligrams per deciliter) and HbA1c (percentage) were: for FBG, 197, 39, and greater than 172.1 in men and 204, 58, and greater than 175.0 in women; for BGAB, 195, 37, and greater than 168.1 in men and 205, 41, and greater than 170.1 in women; for BGAL, 205, 44, and greater than 177.1 in men and 221, 52, and greater than 187.0 in women; for BGBD, 179, 52, and greater than 142.0 in men and 188, 37, and greater than 158.6 in women; and for HbA1c, 8.60, 1.00, and greater than 7.80 in men and 9.30, 2.42, and greater than 8.30 in women.
Goodness of fit was evaluated using the likelihood ratio test. Possible interactions among BG values and among BG values and HbA1c were tested by introducing the relative parameters into the models and evaluating the statistical significance of the increase in log-likelihood. Moreover, models were fitted to test differences in the effect of BGAL between men and women by introduction of the specific interaction terms. Assumption of proportionality of hazards was tested using the log cumulative hazard plot, that is, the plot of log (–log [S(t)]) vs. log of survival time, where S(t) represents the survival function. Residual analysis was conducted using log-cumulative hazard plots of Cox-Snell residuals to ascertain the adequacy of linear prediction in the Cox models.
All analyses were performed using the SAS System (SAS Institute, Cary, NC).
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
shows the characteristics of the population according to gender. Men and women differed significantly for age, diabetes duration, previous cardiovascular events, smoking habit, hypoglycemic therapy, BMI, total/HDL cholesterol ratio, fibrinogen, AER, and HbA1c.
Twenty-three participants died: five from cardiovascular events, 10 from neoplasia, and eight from other reasons; death was preceded by a cardiovascular event in two subjects who died from neoplasia or other causes. We recorded cardiovascular events in 77 patients, 54 in 284 men (19%) and 23 in 245 women (9.4%); 38 of 77 (49.3%) were coronary (29 in men and nine in women), 26 of 77 (33.8%) were cerebrovascular (14 in men and 12 in women), and 13 of 77 (16.9%) were peripheral vascular (11 in men and two in women). Fourteen subjects were affected by fatal or nonfatal myocardial infarction, 24 by angina pectoris (15 of whom treated by coronary angioplasty or coronary artery bypass graft),14 by fatal or nonfatal stroke, seven by transient ischemic attacks, six presented intermittent claudication and/or lower limb amputation of any extent associated with documented ischemia, five were submitted to carotid thromboendoarterectomy, two to surgical interventions for abdominal aortic aneurism, and five to percutaneous angioplasties or bypass surgical interventions at the iliac or lower limbs vessels. Table 2 shows the characteristics of the population stratified according to gender and presence of cardiovascular events in the follow-up period. Men without and with cardiovascular events differed for age, known diabetes duration, previous cardiovascular events, fibrinogen, and HbA1c; women differed for previous cardiovascular events, systolic blood pressure, AER, HbA1c, and all BG values. When both men and women were pooled together, the effect of smoking habit was significant (2 = 7.8; P = 0.02). The effect of hypoglycemic treatment was not significant when men and women were considered separately or pooled together.
As far as BG profiles are concerned, BGAL was higher than BGAB (P = 0.02 in men and P = 0.007 in women; Table 1), with greater differences in patients with events (Table 2).
HbA1c was correlated (P < 0.0001) with all BG values in both men and women. In particular, linear correlation coefficients between HbA1c and FBG, BGAB, BGAL, and BGBD were, respectively, in men, r = 0.388, r = 0.425, r = 0.483, and r = 0.390; and in women, r = 0.452, r = 0.512, r = 0.458, and r = 0.421. When HbA1c and BG values from both genders were pooled together, multiple regression analysis showed that all values remained significantly correlated with HbA1c (r = 0.104, P < 0.046 for FBG; r = 0.216, P < 0.0001 for BGAB; r = 0.229, P < 0.0001 for BGAL; and r = 0.097, P = 0.049 for BGBD).
Table 3 shows the results of the Cox proportional hazards models for the risk of the first cardiovascular event in the follow-up according to gender after adjustment for cardiovascular risk factors and type of treatment. In men, when models were fitted with single BG values (models A–D) and the third tertile was compared with the first and the second tertiles, a significantly increased risk of cardiovascular events was observed only for BGAL. The association of cardiovascular events with BGAL remained significant when all BG values were added together (model E) or when HbA1c was added (model H). Moreover, when all BG values were added together and with HbA1c, the risk for men in the third tertile of BGAL was twice that for subjects in the other tertiles (model J). In women, when each BG value was considered independently of the others (models A–D), HR for FBG, BGAL, and BGBD were significant; the results were confirmed when HbA1c was included in the models (models F–I). When all BG values were considered together, only HR for BGAL remained statistically significant both without and with the addition of HbA1c in the models (model E and J). In model J, HR for BGAL was greater in women (5.54; confidence interval, 1.45–21.20) than in men (2.12; confidence interval, 1.04–4.32; for the difference between genders, P < 0.01).
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Furthermore, our study confirms that when BG values at all times and HbA1c are introduced in the models simultaneously, postprandial, but not fasting, BG is an independent risk factor for cardiovascular events in type 2 diabetic patients, as previously observed in the DIS (11). BG after breakfast predicted cardiovascular events in the DIS, but not in our study; this apparent discrepancy is not surprising because of the striking differences between the caloric and carbohydrate contents of breakfast in Germany and that in northwestern Italy. Actually, breakfast in the Piedmont is frequently a virtual meal, consisting only of a cup of coffee with, sometimes, a small piece of bread. For this reason and because hypoglycemic drugs are usually given in the morning, BG after breakfast is even lower than FBG. Thus, it is not surprising that the postprandial BG predicting cardiovascular events in our study is BG after lunch. As we described in a previous study, in type 2 diabetes the BG baseline on which postmeal excursions are superimposed (what we called the preprandial baseline) is not stable as it is in nondiabetic subjects, but decreases from morning to evening, probably because of waning of the early morning increase in counterregulatory hormones (31). Thus, because postprandial excursions are superimposed to a falling preprandial BG baseline, glycemic profiles look rather flat in our patients (31).
It should be underlined that in type 2 diabetes we strongly recommend performing BG profiles by measuring BG before and after breakfast, after lunch, and before dinner, without as strongly recommending BG measurement before lunch. Thus, because only a minority of our patients measured BG before lunch, we could not evaluate the lunch-induced BG excursions in all of the cohort. In a study carried out on diet-treated type 2 diabetic patients, we showed that BG values before lunch were significantly lower than BG measured 2 h after breakfast (31). The differences could be even greater in the patients investigated in the present study, because many of them were taking hypoglycemic drugs usually administered before breakfast. Furthermore, in patients developing cardiovascular events, the differences between BG values after lunch and after breakfast (which underestimate the after lunch glycemic increases) are greater than those in patients not developing events, a fact suggesting the occurrence of higher lunch-induced BG excursions. In this light, our study could support the predictive role of so-called glucose spikes (32, 33, 34).
When baseline features are related to the development of cardiovascular events at the follow-up, the role played by glycemic control can be easily recognized. Actually, HbA1c values, incorporating information derived from both pre- and postprandial BG, were higher in men and women developing events. This is even more relevant when we consider that other well-known cardiovascular risk factors were not significantly different in the two groups. In contrast, when we tried to determine the predictive role of each component of BG control, we observed a low predictive value of FBG and HbA1c, in agreement with other observations concerning the prediction of atherosclerosis in subjects at risk for type 2 diabetes mellitus (33).
Our study indicates the relevance of postprandial blood glucose control in the prevention of cardiovascular events in type 2 diabetic patients. For this purpose, it is interesting to mention that the STOP-NIDDM trial, in which patients affected by impaired glucose tolerance were treated with acarbose for an average of 3.3 yr to reduce postprandial hyperglycemia, showed a 49% relative risk reduction of major cardiovascular events (34). Acarbose also reduced the risk of myocardial infarction in type 2 diabetic patients (35) and slowed the progression of intima media thickness of carotid arteries in subjects with impaired glucose tolerance (36). Furthermore, correction of postprandial BG is more effective than correction of fasting BG on the regression of carotid intima media thickness and the decrease in circulating inflammatory markers in type 2 diabetic patients (37). Many biochemical pathways could be responsible for the influence exerted by postprandial BG on the occurrence of cardiovascular events (13, 16, 38, 39).
Furthermore, our study shows that BG influences the occurrence of cardiovascular events much more strongly in women than in men after correction for cardiovascular risk factors and type of therapy and when all daily BG values and HbA1c are considered in the models. In particular, HR for cardiovascular events conferred by postlunch BG was 5.54 in women vs. 2.12 in men, the difference being statistically significant. The simple observation of baseline features in subjects developing or not developing cardiovascular events at the follow-up shows that differences in BG control at the various time points are much greater in women. In our study, cardiovascular events were much more frequent in men; thus, it could be speculated that male gender provides such a strong cardiovascular risk per se that it cannot be easily influenced by blood glucose, whereas hyperglycemia dramatically worsens the cardiovascular prognosis in women, who bear per se a much lower gender-related risk. As far as the other cardiovascular risk factors are concerned, it should be underlined that women were older, had a longer diabetes duration, and had greater BMI, HbA1c and fibrinogen values, whereas men presented a more frequent smoking habit, greater total/HDL cholesterol ratio and AER, and a more frequent history of previous cardiovascular events. A possible limitation of our study is the presence of a limited number of cardiovascular events in women.
Since we have been able to provide these results because the policy we have used for many years at the San Luigi Gonzaga Hospital to measure BG profiles in all type 2 diabetic patients (27, 31), a policy that is rather unique to our diabetes unit, we decided to use for the present paper the subtitle: Lessons from the San Luigi Gonzaga Diabetes Study.
The meaning of postprandial BG in the clinical practice is a matter of very intense debate (40). Our study supports the conclusion that it should be carefully considered in type 2 diabetic patients, because it plays a relevant predictive role for cardiovascular events, especially in women.
|
Footnotes |
|---|
Abbreviations: AER, Albumin excretion rate; BG, blood glucose; BGAB, BG 2 h after breakfast; BGAL, BG 2 h after lunch; BGBD, BG before dinner; BMI, body mass index; CHD, coronary heart disease; DIS, Diabetes Intervention Study; FBG, fasting blood glucose; HDL, high-density lipoprotein; HR, hazard ratio; ICD9-CM, International Classification of Diseases 9–Clinical Modification.
Received May 5, 2005.
Accepted December 5, 2005.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Find additional patient-related information at:
This article has been cited by other articles:
 |
|
 |
|
  C. Hohberg, T. Forst, M. Larbig, M. Safinowski, S. Diessel, S. Hehenwarter, M. M. Weber, T. Schondorf, and A. Pfutzner Effect of Insulin Glulisine on Microvascular Blood Flow and Endothelial Function in the Postprandial State Diabetes Care, May 1, 2008; 31(5): 1021 - 1025. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Esposito, M. Ciotola, D. Carleo, B. Schisano, L. Sardelli, D. Di Tommaso, L. Misso, F. Saccomanno, A. Ceriello, and D. Giugliano Post-Meal Glucose Peaks at Home Associate with Carotid Intima-Media Thickness in Type 2 Diabetes J. Clin. Endocrinol. Metab., April 1, 2008; 93(4): 1345 - 1350. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. L Pearce, M. Noakes, J. Keogh, and P. M Clifton Effect of carbohydrate distribution on postprandial glucose peaks with the use of continuous glucose monitoring in type 2 diabetes Am. J. Clinical Nutrition, March 1, 2008; 87(3): 638 - 644. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. K. Bothe, D. M. Franic, R. J. Ingham, and J. H. Davidow Pharmacological Approaches to Stuttering Treatment: Reply to Meline and Harn (2008) Am J Speech Lang Pathol, February 1, 2008; 17(1): 98 - 101. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Monnier and C. Colette Glycemic Variability: Should we and can we prevent it? Diabetes Care, February 1, 2008; 31(Supplement_2): S150 - S154. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Z. Milicevic, I. Raz, S. D. Beattie, B. N. Campaigne, S. Sarwat, E. Gromniak, I. Kowalska, E. Galic, M. Tan, and M. Hanefeld Natural History of Cardiovascular Disease in Patients With Diabetes: Role of hyperglycemia Diabetes Care, February 1, 2008; 31(Supplement_2): S155 - S160. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. H. O'Keefe, N. M. Gheewala, and J. O. O'Keefe Dietary Strategies for Improving Post-Prandial Glucose, Lipids, Inflammation, and Cardiovascular Health J. Am. Coll. Cardiol., January 22, 2008; 51(3): 249 - 255. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Giugliano, A. Ceriello, and K. Esposito Glucose metabolism and hyperglycemia Am. J. Clinical Nutrition, January 1, 2008; 87(1): 217S - 222S. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Zinman, B. J. Hoogwerf, S. Duran Garcia, D. R. Milton, J. M. Giaconia, D. D. Kim, M. E. Trautmann, and R. G. Brodows The Effect of Adding Exenatide to a Thiazolidinedione in Suboptimally Controlled Type 2 Diabetes: A Randomized Trial Ann Intern Med, April 3, 2007; 146(7): 477 - 485. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. M. Mariappan, D. Feliers, S. Mummidi, G. G. Choudhury, and B. S. Kasinath High Glucose, High Insulin, and Their Combination Rapidly Induce Laminin-{beta}1 Synthesis by Regulation of mRNA Translation in Renal Epithelial Cells Diabetes, February 1, 2007; 56(2): 476 - 485. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
