| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
EXTENSIVE CLINICAL EXPERIENCE |
Endocrine Unit (G.C., B.P., R.Z., F.M., C.B.) and the 3rd Internal Medicine Unit (F.P., C.D.P.), Department of Medical and Surgical Sciences, and Department of Medicine and Public Health (C.C.), University of Padua Medical School, and Blood Bank (M.P.A.), Padua General Hospital, I-35128 Padua, Italy; and FIRS Laboratories, RSR Ltd. (S.C., J.F., B.R.S.), Cardiff CF14 5DU, United Kingdom
Address all correspondence and requests for reprints to: Professor Corrado Betterle, Department of Medical and Surgical Sciences, University of Padua Medical School, Via Ospedale Civile 105, I-35128 Padua, Italy. E-mail: corrado.betterle{at}unipd.it.
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and MethodsResultsDiscussionNote Added in ProofReferences |
|---|
Design: To assess the contribution of different clinical, immunological, genetic, and functional factors in the progression to AAD, we followed up 100 ACA-positive and 63 ACA-negative patients without AAD for a maximum of 21 yr (mean 6.0 yr, median 4.8). ACA were measured by immunofluorescence and 21-OH autoantibodies (Abs) by RIA. Adrenal function was assessed by measuring basal levels of cortisol, aldosterone, ACTH, renin activity, and cortisol response to ACTH. The risk of developing AAD was calculated using survival and multivariate analyses.
Results: AAD developed in 31 ACA-positive patients and one ACA-negative patient. The cumulative risk of disease in ACA-positive patients was 48.5% [95% confidence interval (CI) 40.8–56.1]. The cumulative risk was higher in children than adults (100 vs. 31.9%; P < 0.0001), males than females (68.6 vs. 42.7%; P = 0.006), patients with subclinical rather than normal adrenal function at entry (87.4 vs. 30.1%; P < 0.0001), patients with hypoparathyroidism and/or candidiasis than patients with other autoimmune or nonautoimmune diseases (100 vs. 29.7%; P < 0.0001), and patients with high rather than low-medium ACA titers (62.8 vs. 41.2%; P = 0.12). The presence of human leukocyte antigen (HLA)-DRB1 did not appear to contribute to the prediction of AAD. Adjusted hazard ratios by Cox model for the development of AAD were 3.37 for males (CI 1.38–8.24), 5.23 for hypoparathyroidism and/or candidiasis (CI 1.53–17.92), 3.33 for high antibody titers (CI 1.43–7.78), and 6.15 for impaired adrenal function at entry (CI 2.79–13.57).
Conclusions: These results were used to construct a risk algorithm for estimating the probability of developing AAD from the combination of gender, age, adrenal function, antibody titer, and associated autoimmune disorders at entry. The values of estimated risk could be used to decide appropriate follow-up intervals and future immunointervention strategies.
|
Introduction |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionNote Added in ProofReferences |
|---|
We carried out a long-term follow-up of 100 ACA-positive and 63 ACA-negative patients with no initial evidence for clinical AAD to assess the prognostic importance of various clinical, immunologic and genetic factors in disease progression.
|
Patients and Methods |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionNote Added in ProofReferences |
|---|
In an organ-specific autoantibody screening of about 60,000 patients performed over the period 1980–2003 in our Autoimmune Endocrine Disease Laboratory, 152 clinically euadrenal ACA-positive subjects from northern Italy were identified. Patients underwent autoantibody screening because they were suspected of having or being affected by autoimmune disorders or hepatitis C virus (HCV)-related hepatitis or because they were first-degree relatives of autoimmune disease patients. One hundred subjects (97 positive for ACA at their enrollment and three initially ACA-negative patients who seroconverted during follow-up) agreed to enter our study on the basis of written informed consent. Approval for the study was obtained from the local ethical committee. There were 80 females and 20 males (ratio 4:1), 20 patients were children (16 yr or younger), and 80 were adults; the mean age at the beginning of the follow-up was 31.6 yr (range 4–72, median 31). Autoantibody status and adrenal function were assessed in each patient at entry and once a year during follow-up; however, both the patients and their general practitioners were advised to contact us whenever the signs or symptoms of AAD developed. Seventy-four patients were diagnosed with potential autoimmune polyglandular syndrome (APS) type 2 (i.e. autoimmune thyroid diseases and/or type 1 diabetes mellitus), according to the revised classification of APS (3). Seventeen patients had clinical or potential APS type 1 (i.e. idiopathic hypoparathyroidism and/or candidiasis) (3); in 13 of them, the genetic analysis of AIRE locus was carried out (five had R257X homozygous mutation, two had W78R/Q358X mutation, one had homozygous del13, one had homozygous R139X, one had homozygous R203X, one had del13/delGT mutation, one had del13/R257X mutation, and in the case of one patient the mutations have not been found). Two female patients presented with other autoimmune disorders (one autoimmune premature ovarian failure and one autoimmune chronic hepatitis), five patients had HCV-positive chronic hepatitis, whereas two were first-degree relatives of patients with organ-specific autoimmune diseases (Table 1
).
|
-interferon for 6 months.
Sixty-three patients found ACA-negative in our antibody screening were matched with the ACA-positive group for disease, sex, and age and entered the study as controls (Table 1
). In these patients clinical and biochemical evaluation of adrenocortical function (rapid ACTH test) (8, 16) and autoantibody status were assessed every 1–3 yr. At diagnosis of clinical AAD, the adrenals were examined by computerized tomography or nuclear magnetic resonance.
Adrenal autoantibodies
ACA were detected by the classical indirect immunofluorescence technique on normal human adrenal tissue using goat antihuman IgG conjugated to fluorescein isothiocyanate (6, 8). Sera showing staining of adrenal tissue at 1:2 dilution were considered as positive. Titers of ACA were defined by doubling dilution up to the end point.
21-OHAbs were measured using an immunoprecipitation assay based on 125I-labeled recombinant human 21-OH (RSR Ltd., Cardiff, UK) as described before (4). In this assay, 21-OHAbs levels above 1 U/ml were considered positive (4). In the case of some patients, 21-OHAbs could not be measured at entry because the assay had not yet become available. However, the serum samples obtained at enrollment were stored and tested for 21-OHAbs later when possible.
At enrollment, the patients were divided according to ACA titers and 21-OHAb levels into two groups: the first group included patients with low-medium autoantibody levels (1:2–1:32 for ACA, and > 1–100 U/ml for 21-OHAbs), whereas the second included patients with high autoantibody levels (titers > 1:32 for ACA and/or > 100 U/ml for 21-OHAbs). The patients with high titer ACA and low-medium 21-OHAbs levels or vice versa were assigned to the high-level group.
Adrenal function assessment
Adrenal cortex function was assessed at 0800 h by measuring basal plasma levels of cortisol (normal range 138–550 nmol/liter), aldosterone (normal range 277–831 pmol/liter), ACTH (normal range 4–22 pmol/liter), and plasma renin activity (PRA) (normal range 0.2–6 ng/liter per 3 h). Cortisol concentration was also measured in the morning 60 min after iv injection of 250 µg synthetic ACTH (rapid ACTH test) (8). ACTH was measured by an immunoradiometric assay (Euro-Diagnostic, Apeldoorn, Holland), and cortisol and aldosterone were detected by RIA (Diagnostic Products Corp., Los Angeles, CA, and Sorin, Vercelli, Italy, respectively). PRA was determined according to Stockigt et al. (17) with the omission of the boiling step. Adrenal function tests were performed yearly from the time of entry into the study. Adrenal function was assessed using a five-stage scale according to Betterle et al. (8). Briefly, stage 0 (normal adrenal function) was characterized by normal levels of plasma ACTH, aldosterone, PRA, and basal cortisol and cortisol response of more than 200 nmol/liter after ACTH bolus. Stage 1 was defined by an increase in PRA along with a normal or low serum aldosterone levels and appropriate cortisol response to ACTH, while stage 2 by a decreased cortisol response to ACTH. Stage 3 was characterized by increased ACTH levels, basal cortisol levels at the lower end of the normal range, and no cortisol response to ACTH. Finally, stage 4 was identified by additional decrease in basal cortisol level with evidence of signs and symptoms of adrenocortical failure.
Genetic studies
Human leukocyte antigens (HLA)-DRB1 were determined in 79 ACA-positive patients. One hundred twenty healthy adult subjects randomly selected from the Bone Marrow Donor Register PD02 served as controls. DNA extraction was performed using a QIAamp DNA minikit (QIAGEN, Crawley, UK) from 400 µl of peripheral blood. HLA typing was determined using PCR sequence-specific primers (GenoVision, Philadelphia, PA). For each sample, 23 PCRs were performed and the amplification products stained with ethidium-bromide and analyzed on 2% agarose gels using standard procedures (18, 19).
Statistical analysis
Actuarial survival rates were used to estimate the likelihood of progression to AAD according to the method of Cutler and Ederer (20). All patients entered the life-table when ACA were first assessed. The follow-up ended when the diagnosis of overt AD was made (stage 4) or immunologic and biochemical evaluations were last performed for unaffected individuals. The results of the survival analysis were plotted as curves of cumulative risk (CR) of morbidity against years of follow-up. A log-rank test was used to compare the estimates between selected categories (21). If two groups did not show the same length of observation time, the significance level was calculated at the last time interval common to both groups.
Differences in the prevalence of HLA-DRB1 between ACA-positive subjects and healthy controls were assessed using 
2 test, and P < 0.05 was considered significant.
The relationship between survival (time between first ACA-positive test and diagnosis of AD) and each of the variables was analyzed using a Cox proportional hazard model (22). The Cox proportional hazard regression model is expressed as: 
(t) = 0(t)exp(
ßkzk) (t > 0), where (t) is the hazard at time t, 0(t) is the baseline hazard (when the covariates zk are all zero), and ßk (k = 1, ... . , p) are regression coefficients. The selection of variables in the multivariate model was based on a significant level of 0.1 in the univariate analysis. Risk score for developing AAD at 5 yr was calculated from the beta coefficients of Cox proportional hazard models (see Table 3). The bootstrap was used to assess model stability and for internal validation of the model, including 50 samples with replacement to provide unbiased estimates of predictive accuracy (23, 24). Statistical analyses were performed using the Stata software (version 8; Stata Corp., College Station, TX).
|
|
Results |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionNote Added in ProofReferences |
|---|
). At the end of follow-up, 31 patients developed clinical AAD (11 of 70 at stage 0, nine of 19 at stage 1, two of two at stage 2, and nine of nine at stage 3) within a mean period of 3.6 yr (range 3 months to 11 yr, median 2.1 yr), 15 had impaired and 54 had normal adrenal function. None of the patients who were at stage 2 or 3 at entry improved their adrenal function during follow-up, whereas four of 19 patients who were at stage 1 recovered to stage 0 (Fig. 1). Sixty-three ACA-negative patients were followed up for a mean period of 5.4 yr (range 6 months to 18 yr, median 4.3 yr). All 31 Addisonian patients had normal or small adrenals by imaging. During follow-up, three initially ACA-negative subjects were found to be positive for ACA and 21-OHAbs and were reassigned to the antibody-positive group to monitor them more closely (every 12 months rather than 1–3 yr). One ACA-negative subject, a female with APS type 1, developed AAD after 12 yr of follow-up at 32 yr of age. She was ACA-negative in the last test carried out 3 yr before the diagnosis of AAD but at diagnosis of AAD was found to be ACA-positive.
|
).
|
).
Ten of the 20 males and 21 of the 80 females progressed to clinical AAD during follow-up. The CR of AAD at 9 yr of follow-up was significantly higher in males (68.6%; 95% CI 53.3–83.8), compared with females (42.7%; 95% CI 34.4–51.0) (P = 0.003, log-rank test) (Fig. 2C). The CR of AAD at 17 yr of follow-up in patients with initially impaired adrenal function was 87.4% (95% CI 77.0–97.7), compared with 30.1% (95% CI 21.4–38.9) in patients with normal adrenal function (P < 0.0001, log-rank test) (Fig. 2D).
The CR of AAD in 86 patients positive for both ACA and 21-OHAbs was 50.7% at 7 yr (95% CI 43.2–58.3), compared with 0% of 14 ACA-positive/21-OHAb-negative patients (P = 0.06, log-rank test) (Fig. 2E). Specifically, all 31 patients who developed clinical AAD were positive for both ACA and 21-OHAbs (16 at high titers and 15 at low-medium titers). At the end of follow-up, 54 of the 55 ACA-positive and 21-OHAbs-positive patients without AAD were still positive for both autoantibodies (22 had high autoantibody titers, whereas 32 had low-medium titers), whereas one remained positive for 21-OHAbs but lost ACA positivity (this patient had undergone immunosuppressive therapy for kidney-pancreas transplantation). Four of the 14 ACA-positive/21-OHAbs-negative patients became ACA-negative (one spontaneously, two under treatment with methimazole, and one 5 yr after treatment with -interferon).
The CR of AAD in patients with high autoantibody titers at enrollment was 62.8% at 17 yr (95% CI 49.6–76.0), compared with 41.2% (95% CI 31.8–50.7) in patients with medium-low autoantibody titers (P = 0.12, log-rank test) (Fig. 2F).
Fourteen of the 17 patients with APS type 1, 16 of the 74 patients with APS type 2, and one of the nine patients with other diseases developed clinical AAD, respectively. The CR of AAD for patients with APS type 1 at 11 yr was 100%, and this was significantly higher than in patients with other autoimmune and nonautoimmune conditions (29.7%; 95% CI 15.5–43.8) (P < 0.0001, log-rank test) (Fig. 2G).
HLA-DRB1 typing was performed in 79 antibody-positive patients (26 developing clinical AAD and 53 not progressing to overt adrenal failure). Twenty-three of the 79 patients (29.0%) were HLA-DRB1*03, compared with 18 healthy controls (15.0%) (P = 0.02); 26 (33.0%) were HLA-DRB1*04, compared with 17 healthy controls (14.0%) (P = 0.003); and seven (9.0%) were HLA-DRB1*13, compared with 30 healthy controls (25.0%) (P = 0.005). Similarly, in the 26 patients who developed AAD, HLA-DRB1*03 was found in nine (34.6%) and HLA-DRB1*04 in eight (30.7%) (P = 0.03 and P = 0.08 vs. healthy controls, respectively), whereas HLA-DRB1*13 was found in three (11.5%) (P = 0.19 vs. healthy controls). The CR of AAD in patients with HLA-DRB1*03 and/or -DRB1*04 was 43.8% (95% CI 33.9–53.7), compared with 56.2% (95% CI 43.2–69.3) with other -DRB1 haplotypes at 17 yr of follow-up (P = 0.92, log-rank test) (Fig. 2H).
The results of univariate and multivariate analyses using Cox proportional hazards model are summarized in Table 2. According to the univariate analysis, age, male gender, impaired adrenal function, high antibody titers, and coexistence of hypoparathyroidism and/or candidiasis were significantly associated with the development of AAD, whereas HLA status was not relevant for the progression to AAD.
|
Consequently, the output of multivariate analysis by Cox model (including age, gender, adrenal function at initial examination, antibody titer, and disease at enrollment) could be used to produce a prediction algorithm at 5 yr (Table 3).
The likelihood of developing AAD was arbitrarily regarded as low risk, moderate risk, or high risk when the estimated probabilities were 10% or less, 11–39%, or 40% or more, respectively.
Using our prediction algorithm, of the 31 patients who finally developed AAD, 24 (77.4%) were initially assigned to be at high risk, three (9.7%) at moderate risk, and four (12.9%) at low risk. In contrast, 11 (91.7%) of 12 patients followed up for more than 12 yr who did not develop AAD were initially assigned to be at low risk, one (8.3%) at moderate risk, and none at high risk.
|
Discussion |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionNote Added in ProofReferences |
|---|
), as reported previously (10, 11). However, when multivariate analysis was used, the child to adult HRs did not maintain statistical significance (HR 1.47, Table 2).
The presence of impaired adrenal function at entry was confirmed to be an important risk factor for developing AAD both by multivariate and univariate analysis (HR 6.15 and 6.10, respectively, Table 2). A comparable observation has been found in other autoimmune conditions, such as in patients with thyroid autoantibodies and subclinical hypothyroidism who develop overt thyroid dysfunction more frequently than those without subclinical hypothyroidism (25). Also, the risk of overt diabetes has been reported to be higher in first-degree relatives of diabetic patients with islet cell autoantibodies and impaired iv glucose tolerance test, compared with antibody-positive relatives with normal iv glucose tolerance test (26). During follow-up, not all of our patients with adrenal cortex autoantibodies progressed to AAD, and four of 19 who were at stage 1 regained normal adrenal function. In contrast, none of the 11 patients who were found at stage 2 or 3 of subclinical hypoadrenalism recovered to earlier stages. Our study suggests that stage 2, when patients do not yet have clinical symptoms, represents a critical phase in the deterioration of the adrenal cortex and could be considered as a point of no return in adrenal failure. Consequently patients at stage 2 could be considered for early substitutive therapy. This is a consequence of the observations made during the current study because at the start of our follow-up program (about 25 yr ago), therapy was commenced at stage 4.
Our patients with higher titers of ACA/21-OHAbs showed an increased rate of progression to AAD, compared with those with low-medium titers (HR 3.33, Table 2). This suggests that levels of adrenal autoantibodies can be considered as a measure of the (auto)immune aggression toward the adrenal cortex. This can be compared with observations that the risk of developing diabetes in first-degree relatives of diabetic patients has been also related to the titers of islet cell antibodies (27, 28). In this study we used the titers of both ACA and/or 21-OHAbs as one of our variables for the prediction equation. However, it is of interest that none of the 14 ACA-positive/21-OHAbs-negative patients developed AAD during an observation period of 6 yr. These patients probably represent a subgroup with a reactivity against a not-yet-identified adrenal cortex autoantigen, and it is still unclear whether they are at risk of AAD. Consequently, 21-OHAbs may be a more appropriate marker than ACA for the assessment of the risk of progression to AAD. In contrast, only one of 63 ACA-negative patients developed AAD during the follow-up. This patient was a female with chronic hypoparathyroidism and candidiasis who was found ACA-positive at the onset of AAD. We were unable to detect the time of her seroconversion between the last biochemical assessment (3 yr before) and the disease onset. This observation indicates that the cumulative risk of developing AAD in ACA-negative patients is significantly lower, compared with ACA-positive patients (P < 0.0001, Fig. 2A), but also that APS type 1 patients need a serological follow-up, even if initially ACA-negative.
None of our patients who underwent immunosuppression during follow-up lost adrenal autoantibody positivity or improved in stage of adrenal dysfunction, and this contrasts to previous reports (9, 13). The reasons for this apparent discrepancy are unclear at present but may be related to differences in immunosuppressive protocols. In the present study, adrenal autoantibodies disappeared in two patients with Graves’ disease treated with methimazole only, and one of them also recovered normal adrenal function. Methimazole has been demonstrated to exert an immunosuppressive effect in patients with Graves’ disease (29, 30). Therefore, an immunosuppressive action on adrenal autoimmunity cannot be completely excluded in patients with adrenal antibodies taking methimazole for a coexisting Graves’ disease (29, 30). Interestingly, none of our four ACA-positive patients treated with -interferon showed increases in autoantibody titers and progression to AAD, although in some patients such treatment may result in an increase in 21-OHAbs as well as thyroid and islet cell autoantibody levels (15, 31, 32).
In our study the preexistence of hypoparathyroidism and/or candidiasis (i.e. APS type 1) in adrenal antibody-positive patients conferred a significant risk of progression to AAD (HR 5.23, Table 2). Patients with APS type 1 differ from other patients with AAD in that they are usually diagnosed at an early age, have particular associated autoimmune diseases, and have mutations in the AIRE gene. However, over 90% of all APS type 1 patients are ACA-positive at diagnosis of AAD (7, 10, 11), and our Cox model analysis was carried out on all 100 ACA-positive patients, irrespective of the associated diseases, and this included 17 patients with clinical APS type 1.
Our study also showed that the risk of progression to AAD does not appear to change between 12 and 21 yr of follow-up (see flattening of the morbidity curve in Fig. 2A). None of our 12 patients followed up for more than 12 yr (of whom 11 could be assigned to the low-risk group) developed clinical hypoadrenalism. The reasons for this are not clear, but it suggests that autoimmune destruction of the adrenal tends to occur over a definite time interval. This observation can be compared with similar findings in individuals who are positive for thyroid autoantibodies or for diabetes-associated autoantibodies but do not develop clinical signs of disease for many years (25, 26, 27).
An important goal in the prevention and management of organ-specific autoimmunity lies in identification of individuals at high risk of developing clinical disease. Our study provides a new approach to assigning a risk score for the development of AAD in adrenal antibody-positive patients, taking into account five easily obtainable factors. Three of these are clinical (age, gender, preexisting disease), one immunological (autoantibody titers), and one functional (ACTH test). Our approach gives useful information about the level of risk and recommended timing for clinical and laboratory evaluation as indicated in Table 4. The differences in the disease progression among the three groups of patients according to their baseline risk assigned at recruitment are shown using survival curves in Fig. 3. Regular assessment of adrenal autoantibodies and adrenal function should clearly be performed most frequently in high-risk individuals so that the risk of developing a life-threatening adrenal crisis can be minimized by early substitutive therapy. For example, as shown in Fig. 3, in the high-risk group, the development of new cases of AAD was observed almost every 12 months over 11 yr of follow-up. Consequently, we propose a clinical evaluation every 6–12 months for patients at high risk and every 24–36 months for patients assigned to low risk (Table 4). Five of 31 patients (16%), indeed, developed AAD in less than 1 yr of follow-up, and all of them (four adults and one child) were assigned with a high-risk score for developing AAD (ranged from 45 to 99%).
|
|
Overall, our study allows estimation of the risk of AAD developing according to five easily obtainable factors and assignment of different assessment intervals for patients positive for ACA and/or 21-OHAbs. Patients assessed to be at high risk of AAD might be considered for inclusion in future intervention trials to halt, or at least delay, the progression of adrenal cortex destruction.
|
Note Added in Proof |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionNote Added in ProofReferences |
|---|
|
Acknowledgments |
|---|
|
Footnotes |
|---|
First Published Online March 7, 2006
Abbreviations: AAD, Autoimmune AD; ACA, adrenal cortex autoantibodies; AD, Addison’s disease; APS, autoimmune polyglandular syndrome; CI, confidence interval; CR, cumulative risk; HCV, hepatitis C virus; HLA, human leukocyte antigen; HR, hazard ratio; 21-OHAbs, 21-hydroxylase autoantibodies; PRA, plasma renin activity.
Received April 20, 2005.
Accepted February 27, 2006.
|
References |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionNote Added in ProofReferences |
|---|
therapy for chronic hepatitis C. Gut 48:378–383 and autoimmune thyroid disease. Thyroid 13:547–551[CrossRef][Medline]This article has been cited by other articles:
 |
|
 |
|
  C. K. WELT Autoimmune Oophoritis in the Adolescent Ann. N.Y. Acad. Sci., June 1, 2008; 1135(1): 118 - 122. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. DE BELLIS, A. COLAO, R. PIVONELLO, A. SAVOIA, M. BATTAGLIA, G. RUOCCO, G. TIRELLI, G. LOMBARDI, A. BELLASTELLA, and A. BIZZARRO Antipituitary Antibodies in Idiopathic Hyperprolactinemic Patients Ann. N.Y. Acad. Sci., June 1, 2007; 1107(1): 129 - 135. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Bensing, S. O. Fetissov, J. Mulder, J. Perheentupa, J. Gustafsson, E. S. Husebye, M. Oscarson, O. Ekwall, P. A. Crock, T. Hokfelt, et al. Pituitary autoantibodies in autoimmune polyendocrine syndrome type 1 PNAS, January 16, 2007; 104(3): 949 - 954. [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 |
