| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Lipids, Lipoproteins, and Cardiovascular Risk Factors |
1 Department of Cardiovascular Research, Istituto di Ricerche Farmacologiche "Mario Negri", Milan, Italy.
2 Department of Medicine, New York Medical College, Valhalla, NY.
3 Cardiology Section, Veterans Affairs Medical Center, Minneapolis, MN.
4 Laboratorio di Endocrinologia, Ospedale "Luigi Sacco", Milan, Italy.
5 Cardiovascular Division, Department of Medicine, University of Minnesota Medical School, Minneapolis, MN.
6 Laboratory of Cardiovascular Endocrinology and Cell Biology, Consiglio Nazionale delle Ricerche, Institute of Clinical Physiology, Pisa, Italy.
7 Consorzio Mario Negri Sud, S. Maria Imbaro, Chieti, Italy.
aAddress correspondence to this author at: Istituto di Ricerche Farmacologiche "Mario Negri", via Eritrea 62, 20157 Milan, Italy. Fax 39-023-320-0049; e-mail masson{at}marionegri.it.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: We measured BNP and NT-proBNP at baseline in 3916 patients enrolled in the Valsartan Heart Failure Trial. To identify the variables associated with both peptides, we conducted simple and multivariable linear regression analyses. We used Cox multivariable regression models to evaluate the independent prognostic value for all-cause mortality, mortality and morbidity, and hospitalization for HF. Prognostic performance was assessed by pairwise comparisons of the area under the curve of receiver-operator characteristic curves.
Results: NT-proBNP and BNP had similar relationships with age, left ventrical ejection fraction, and internal diameter and creatinine clearance. Either peptide ranked as the first independent predictor of outcome after adjustment for major confounding clinical characteristics. ROC curves were almost superimposable for all-cause mortality (area under the curve (SE): BNP 0.665 (0.011) vs NT-proBNP 0.679 (0.011); P = 0.0734), but NT-proBNP was superior to BNP for predicting mortality and morbidity (P = 0.032) or hospitalization for HF (P = 0.0143). Overall sensitivity and specificity ranged from 0.590 to 0.696.
Conclusions: The natriuretic peptides BNP and NT-proBNP showed subtle differences in their relation to clinical characteristics and prognostic performance in a large population of patients with chronic and stable HF. They were the most powerful independent markers of outcome in HF.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
BNP and the amino-terminal fragment of BNP (pro-hormone NT-proBNP) are secreted in equimolar amounts in response to mechanical or neurohormonal stimulation. Both peptides are reliably measured in clinical practice. Whereas BNP is a natriuretic and diuretic hormone, the biological role of NT-proBNP, if any, is not well understood. These peptides also differ with regard to their biological half-life, in vitro stability, and mechanisms involved in their clearance (5). However, there is no compelling evidence in favor of either peptide in terms of indicators of clinical severity (6)(7)(8), screening for left ventricular (LV) dysfunction in the general population (9), diagnosis (10)(11)(12)(13), or prognosis (8)(14)(15). In particular, a study in a large and representative population of patients with chronic HF is required to examine and compare the prognostic value of BNP and NT-proBNP in a setting that would allow sufficient statistical power to examine subgroups of patients according to their clinical characteristics.
In the Valsartan Heart Failure (Val-HeFT) trial, blood samples were collected at the time of entry into the study in a large population of patients with chronic and symptomatic HF, yielding one of the largest plasma banks collected for this disease. The clinical utility of several neurohormones, including BNP, has already been evaluated in this trial (16)(17)(18)(19). In this study, we carried out a direct comparison of the 2 natriuretic peptides in this large population in relation to the severity of the disease and clinical outcome.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
2.9 cm/m2, were enrolled in the study. Results of the main trial have been presented in detail previously (20).
measurement of natriuretic peptide concentrations
Blood samples were collected at study entry by venipuncture in evacuated plastic tubes containing EDTA dipotassium salt. Blood samples were centrifuged at 3500g for 10 min at 4 °C immediately after collection, and plasma samples were stored at –70 °C. BNP (IRMA Shionogi) and NT-proBNP (ECLIA Elecsys 2010 analyzer, Roche Diagnostics) were measured by commercially available assays in plasma samples never thawed before. Norepinephrine, plasma renin activity, and C-reactive protein were determined, as previously described (17)(19). The intraassay CVs were 2.9% and 3.9% for NT-proBNP and BNP, respectively. Corresponding interassay CVs were 3.9% and 4.7%. In 81 age-matched, healthy volunteers, the 95th percentile concentrations were 121 and 76 ng/L, for NT-proBNP and BNP, respectively.
statistical methods
We performed all statistical analyses both NT-proBNP and BNP concentrations measured at baseline in all patients (n = 3916). We compared concentrations of NT-proBNP or BNP according to clinical characteristics by means of the Wilcoxon rank-sum test because the normality of the distribution was not satisfied. We assessed the correlation between baseline concentrations of BNP and NT-proBNP with the Pearson correlation coefficient test.
Simple linear regression analyses were carried out to evaluate the relationship between baseline concentrations of BNP or NT-proBNP (expressed as log-transformed data) and clinical characteristics as continuous variables when appropriate. The assumptions underlying the linear regression models, such as linearity, normality, and homoscedasticity, were satisfied for all the independent continuous variables, except for creatinine and bilirubin, which were transformed on a logarithmic scale.
Multivariable linear regression analyses were carried out to determine the variables independently associated with NT-proBNP or BNP, expressed as continuous log-transformed concentrations. In the final models, only variables statistically significant at the univariate analysis (P <0.05) were included. ß regression coefficients and their significance from multiple linear regression analysis are reported.
The relationship between NT-proBNP or BNP and these variables was also explored by dividing them into quartiles and assessing differences by means of the Kruskal–Wallis test.
Cox multivariable regression models were used to evaluate the independent prognostic value of baseline NT-proBNP and BNP (entered as categorical variables below and above the median concentration) on all-cause mortality, on mortality and morbidity (M&M), and on hospitalization for HF alone. Mean follow-up time was 23 months. The model included all the demographic, clinical, and echocardiographic variables, considered as categorical, that had a significant univariate relationship with outcome. To check for possible biases resulting from the arbitrary choice of median concentration, we also carried out the same Cox multivariate analysis with continuous variables where appropriate. We checked the assumption of proportional hazard with the log[-log(survival)] plot and by the time-dependent covariate test. Also, we used the martingale residuals plot to evaluate whether a specific time-independent continuous covariate might be entered directly into the model, or if a transformation was necessary. To ease the interpretability of these results, hazard ratios have been computed for intervals of 50 ng/L of BNP and 500 ng/L of NT-proBNP.
To assess the incremental prognostic value of NT-proBNP or BNP with respect to clinical variables, we evaluated a separate addition of NT-proBNP or BNP to a model based on demographic, clinical, and echocardiographic variables with the likelihood-ratio test with 1 degree of freedom.
To compare the predictive value of BNP and NT-proBNP, ROC curves were generated for the main outcomes. Exploratory analyses were also carried out to compare the prognostic value of both natriuretic peptides for the 2 main adjudicated causes of death, sudden death from cardiac causes and death because of pump failure (20). Pairwise comparisons of the areas under the curve (AUC) were conducted, following the procedure of Hanley and McNeil (21).
All endpoints were adjudicated by an independent Endpoint Committee, blind to study treatment. All tests were made at a 2-sided, 5% significance level. Statistical analyses were performed with SAS 8.2 (SAS Institute, Inc.).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Plasma concentrations of NT-proBNP and BNP according to baseline clinical characteristics are presented in Table 1
. NT-proBNP and BNP were significantly higher in older patients, and in patients with more symptomatic HF, more dilated LV and more severe LV dysfunction, lower body mass index (BMI), lower sitting arterial blood pressure, HF of ischemic etiology, or atrial fibrillation at study entry. The concentration of both peptides was also higher in patients with increased laboratory markers (creatinine and bilirubin) and neurohormones (plasma renin activity, norepinephrine, and C-reactive protein). There were no relevant differences between the univariate clinical correlates of NT-proBNP or BNP, except for beta-blocker use, sitting heart rate, and plasma aldosterone concentration (all nonsignificant for BNP).
|
Baseline log-transformed NT-proBNP or BNP concentrations correlated with age, LVEF, and LVIDd, with r2 ranging from 0.02 to 0.12 (P <0.0001, variables as continuous data). Fig. 1
compares the relationship between median NT-proBNP or BNP and these variables, categorized by quartiles. The trend was similar for both natriuretic peptides.
|
The strongest continuous relationship was found between log-transformed estimated glomerular filtration rate (eGFR) and NT-proBNP (Pearson CV –0.34, P <0.0001) or BNP (–0.21, P <0.0001). The relationship between median concentrations of NT-proBNP or BNP and eGFR showed a steeper increase of NT-proBNP compared with BNP in patients with reduced renal function (Fig. 2
). However, the number of patients with eGFR <30 mL/min/m2 was limited in this study (n = 59, 1.5% of the population).
|
The clinical variables that were independently associated with baseline concentrations of BNP or NT-proBNP (expressed as log-transformed data) were largely common to both peptides, with the exception of the use of angiotensin-converting enzyme (ACE) inhibitors and beta-blockers (Table 2
). Whereas NT-proBNP was independently associated with ACE inhibitors (positive relation) and beta-blockers (negative relation), no such associations were found with BNP. The factors with the strongest association were age, ischemic etiology, presence of atrial fibrillation, LV ejection fraction, and internal diameter.
|
prognostic value of bnp and nt-PRObnp
The adjusted hazard ratios for death, M&M, or hospitalization for HF according to baseline deciles of NT-proBNP and BNP are presented in Fig. 3
. There was a similar progression in the probability of death with increasing plasma concentration of NT-proBNP or BNP, with hazard ratios in the tenth decile of 4.02 (2.63–6.11) and 4.02 (2.70–5.99), respectively. Corresponding values for M&M were 4.74 (3.36–6.70) and 3.67 (2.70–4.98). An apparent superiority of NT-proBNP over BNP for predicting the risk of hospitalization for patients with HF was observed across the whole range of concentrations. For instance, the relative risk of hospitalization for HF was higher in the tenth decile of NT-proBNP [7.51 (4.30–13.11)] compared with that of BNP [3.86 (2.56–5.84)], although 95% confidence intervals (CIs) always overlapped.
|
The independent prognostic values of baseline BNP or NT-proBNP were tested separately in Cox multivariable models that included the demographic, clinical, and echocardiographic variables with a statistically significant association with outcome at univariate analysis (P <0.05). First, the natriuretic peptides and the other covariates were entered as categorical variables, dichotomized above or below the median value. In these models, NT-proBNP was the first predictor of outcome with hazard ratios (95% CI) of 2.07 (1.76–2.46) for mortality, 2.20 (1.92–2.51) for M&M, and of 2.66 (2.19–3.22) for hospitalization for HF (Fig. 4
). Similarly, BNP resulted as the strongest independent predictor of mortality [1.87 (1.59–2.20)], M&M [2.05 (180–2.34)], and hospitalization for HF [2.48 (2.06–2.98)]. When the natriuretic peptides and appropriate covariates were considered as continuous variables, their prognostic value was consistent with the analysis based on median cutoff values. For instance, an increment of 500 ng/L of baseline NT-proBNP concentration corresponded to an increase of risk of 3.8% for mortality (P <0.0001), whereas an increment of 50 ng/L of baseline BNP concentration corresponded to an increase of risk of 5.7% (P <0.0001). Similarly, increases of risk for M&M were 3.5% and 5.4%, and 3.0% and 5.7% for hospitalization for HF, respectively (P <0.0001 for all). Finally, the likelihood-ratio test showed that BNP or NT-proBNP provided significant incremental prognostic value for the prediction of all-cause mortality, M&M, and hospitalization for HF (all with P <0.0001) compared with a model based on demographic, echocardiographic, and clinical variables.
|
The predictive values of NT-proBNP and BNP were compared by means of ROC curves (Table 3
). The AUC (SE) for NT-proBNP (0.679 [0.011]) was marginally higher than for BNP (0.665 [0.001]; P = 0.0734) for predicting all-cause mortality. For M&M, NT-proBNP (AUC = 0.688 [0.009]) had a statistically significant higher predictive value than BNP [0.674 (0.009); P = 0.0332]. The same was true for hospitalization for HF [0.685 (0.011) vs 0.665 (0.012); P = 0.0143]. For a given peptide, optimum cutoff points were similar for main outcomes (Table 3
). The concentrations of BNP at 95% of sensitivity were 18, 17, and 14 ng/L for mortality, M&M, and hospitalization for HF, respectively (Table 3
). Corresponding values for NT-proBNP were 202, 208, and 250 ng/L. The concentrations of BNP at 95% of specificity were 531, 490, and 546 ng/L for mortality, M&M, and hospitalization for HF, respectively. Corresponding values for NT-proBNP were 4643, 4367, and 5015 ng/L. In exploratory analyses, the prognostic values of both peptides were compared according to adjudicated causes of death: whereas NT-proBNP was similar to BNP in predicting sudden death [399 events, AUC = 0.619 (0.015) vs 0.627 (0.015); P = 0.47, Fig. 5
], results superior to BNP for the prediction of death from pump failure [191 events, AUC = 0.749 (0.017) vs 0.703 (0.019); P = 0.0003].
|
|
When patients were divided into clinical subgroups, NT-proBNP retained statistical superiority over BNP in males (P = 0.018 for all-cause mortality; P = 0.044 for hospitalization for HF, Table 4
), in patients older than 70 years (P <0.0001 for all-cause mortality; P = 0.023 for hospitalization for HF), or in patients with eGFR of 30 and 59 mL/min/1.73 m2 (P = 0.017 for all-cause mortality; P = 0.036 for hospitalization for HF). BNP and NT-proBNP, however, performed similarly in females, in patients <70 years, in patients with eGFR 60–89 mL/min/1.73 m2, in patients with or without diabetes, and in 3 classes of BMI index (Table 4
).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
relationships between nt-probnp, bnp, and clinical severity
In this study we investigated the association of various demographic, clinical, and echocardiographic variables and baseline concentrations of natriuretic peptides with univariate and multivariable models. We concluded that the set of variables associated with plasma concentrations of NT-proBNP or BNP was essentially the same, with the exception of the use of ACE inhibitors and beta-blockers. The reason for this difference is unknown at present; however, recent studies have shown that the introduction of a beta-blocker (metoprolol) in patients with mild and stable HF raised the concentration of both C-terminal and N-terminal natriuretic peptides (22)(23). The strongest associations were found with age, ischemic etiology of HF, presence of atrial fibrillation, LVEF, and LV internal diameter, as previously described (24)(25)(26). Our data also concur with the conclusion of a comparative study performed in a large cohort of patients with stable ischemic heart disease in which LVEF, age, and renal function were the most powerful independent predictors of BNP and NT-proBNP (8).
Among the continuous variables, the strongest association was found with renal function. Median concentrations of the natriuretic peptides increased progressively with renal dysfunction estimated as eGFR and categorized according to the classification of the National Kidney Foundation. This trend has been observed in patients presenting in emergency departments for dyspnea (27). Decreased clearance from the kidney has been put forward to explain increased BNP in patients with HF (28). Although the mechanisms of elimination of these 2 peptides might differ (BNP is mainly cleared by neutral endopeptidases and specific receptors, whereas renal excretion is currently regarded as the main clearance mechanism for NT-proBNP (5), the increase of both peptides with lower eGFR suggests that the susceptibility of NT-proBNP concentrations to renal dysfunction was not, apparently, greater than that of BNP. We must remember, however, that the number of patients with severely compromised renal function (eGFR <30 mL/min/1.73 m2) was limited in our study population, so this conclusion does not apply in such patients. This extends to previous observations of patients with myocardial infarction and compensated renal dysfunction (7) and confirms recent studies that showed similar renal extraction of the 2 peptides in healthy volunteers and in patients with essential arterial hypertension (29)(30).
prognostic value of nt-probnp and bnp in heart failure
BNP and NT-proBNP are strong markers of outcome in chronic HF (2)(17). The present analysis attests to their superiority compared with other variables, but also confirms the strength of several clinical variables (older age, diabetes, and lower LVEF) that have been powerful and independent predictors of mortality and morbidity in a large and contemporary population of patients with HF (31). Few studies have directly compared the prognostic value of the 2 brain natriuretic peptides in patients with heart disease. Some have concluded that the 2 markers are substantially equivalent (14)(15)(32). In patients with stable ischemic heart disease, Richards et al. (8) recently showed that both peptides independently predicted a composite clinical endpoint (1-year all-cause mortality or admission with HF) with almost overlapping, event-free survival curves. There are, however, some exceptions: BNP (but not NT-proANP or NT-proBNP) independently predicted sudden death in 382 patients with chronic HF (33), whereas NT-proBNP (but not BNP) predicted symptom-free survival and postoperative outcome in patients with severe aortic stenosis (34). Differences have also been observed in acute settings, for instance in response to short term exercise (35) or after intravenous infusion of Levosimendan (36). In the present study, NT-proBNP performed better than BNP to predict death from pump failure and hospitalization for HF (P = 0.0003 and 0.0143 by univariate pairwise comparison, respectively). When patients were divided into clinical subgroups, NT-proBNP retained a statistical superiority over BNP in males, in patients over 70 years, or with eGFR of 30 to 59 mL/min/1.73 m2. In univariate analysis, BNP and NT-proBNP performed similarly in females, in patients <70 years, or in patients with eGFR >60 mL/min/1.73 m2. In addition, the prognostic value of BNP and NT-proBNP was similar in patients with or without diabetes and in different classes of BMI (lean, overweight, or obese). Several factors might, in principle, explain these differences. First, the longer estimated biological half-life of NT-proBNP compared with BNP (70 min vs 
20 min) (37)(38) renders the former less prone to short-term fluctuations that may confound the prognosis. This difference is corroborated by the relatively smaller intraindividual biological variation of NT-proBNP observed in patients with HF (39). In the context of a large multicenter trial, preanalytical deviations from the protocol during blood collection, plasma processing, and storage would also have a greater detrimental impact on BNP than on NT-proBNP. Second, differences in the analytical performance of the methods used to measure BNP and NT-proBNP could influence their respective prognostic value, although the impact of assay imprecision on the performance of cardiac biomarkers for diagnosis seems not to be critical (40). The IRMA used in the present study to measure BNP showed slightly higher analytical variability than the automated method for NT-proBNP; this could further reduce the prognostic performance of BNP. It is important to note that immunoreactive moiety detected by any given immunoassay depends on the combined influence of circulating forms of NT-proBNP and BNP, the degree and rate at which they are degraded from amino and/or C-terminals, and (most importantly) the epitope(s) to which polyclonal or monoclonal antibodies have been raised (41). Indeed, recent studies demonstrated that the clinical results (including the reference interval and decisional cutoff values) of BNP and NT-proBNP assays are method-dependent (42). For these reasons, the analytical performance and clinical accuracy of any single immunoassay should be assessed in each laboratory. Third, it cannot be excluded that different relationships between NT-proBNP or BNP and some clinical characteristics not considered in the present analysis might explain our data. Finally, the large study population (3916) might have revealed subtle differences in the prognostic performance of the 2 peptides not detectable in smaller populations.
The relatively modest prognostic accuracy of BNP and NT-proBNP in the present study (AUCs of 0.66 to 0.75, sensitivity and specificity of 0.590 to 0.696) might result from the fact that our population included only patients with stable, chronic, and symptomatic mild-to-moderate HF, with LV systolic dysfunction (LVEF <40%). Effective neurohormone suppression with background therapy might also have limited the prognostic performance of these markers.
In conclusion, we report a small, but statistically significant difference in the predictive value of 2 related natriuretic peptides in a well-characterized and large population of patients with chronic and symptomatic HF.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  S. Masson, R. Latini, I. S. Anand, S. Barlera, L. Angelici, T. Vago, G. Tognoni, J. N. Cohn, and for the Val-HeFT Investigators Prognostic Value of Changes in N-Terminal Pro-Brain Natriuretic Peptide in Val-HeFT (Valsartan Heart Failure Trial) J. Am. Coll. Cardiol., September 16, 2008; 52(12): 997 - 1003. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. M. Richards Serial Measurements of Plasma B-Type Natriuretic Peptides: What Do They Tell Us? J. Am. Coll. Cardiol., September 16, 2008; 52(12): 1004 - 1005. [Full Text] [PDF]
|
|
|
|
 |
|
 E. C H J Michielsen, J. A Bakker, R. R J V. Kimmenade, Y. M Pinto, and M. P V. Dieijen-Visser The diagnostic value of serum and urinary NT-proBNP for heart failure Ann Clin Biochem, July 1, 2008; 45(4): 389 - 394. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Vickery, M. C. Webb, C. P. Price, R. I. John, N. A. Abbas, and E. J. Lamb Prognostic value of cardiac biomarkers for death in a non-dialysis chronic kidney disease population Nephrol. Dial. Transplant., June 18, 2008; (2008) gfn341v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Braunwald Biomarkers in Heart Failure N. Engl. J. Med., May 15, 2008; 358(20): 2148 - 2159. [Full Text] [PDF]
|
|
|
|
 |
|
 R. Paniagua, D. Amato, S. Mujais, E. Vonesh, A. Ramos, R. Correa-Rotter, and W. H. Horl Predictive Value of Brain Natriuretic Peptides in Patients on Peritoneal Dialysis: Results from the ADEMEX Trial Clin. J. Am. Soc. Nephrol., March 1, 2008; 3(2): 407 - 415. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Schou, F. Gustafsson, C. N. Kistorp, P. Corell, A. Kjaer, and P. R. Hildebrandt Effects of Body Mass Index and Age on N-Terminal Pro Brain Natriuretic Peptide Are Associated with Glomerular Filtration Rate in Chronic Heart Failure Patients Clin. Chem., November 1, 2007; 53(11): 1928 - 1935. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. R. deFilippi, S. L. Seliger, S. Maynard, and R. H. Christenson Impact of Renal Disease on Natriuretic Peptide Testing for Diagnosing Decompensated Heart Failure and Predicting Mortality Clin. Chem., August 1, 2007; 53(8): 1511 - 1519. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Omland, M. S. Sabatine, K. A. Jablonski, M. M. Rice, J. Hsia, R. Wergeland, S. Landaas, J. L. Rouleau, M. J. Domanski, C. Hall, et al. Prognostic Value of B-Type Natriuretic Peptides in Patients With Stable Coronary Artery Disease: The PEACE Trial J. Am. Coll. Cardiol., July 17, 2007; 50(3): 205 - 214. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Tsutamoto, T. Tanaka, H. Sakai, C. Ishikawa, M. Fujii, T. Yamamoto, and M. Horie Total and high molecular weight adiponectin, haemodynamics, and mortality in patients with chronic heart failure Eur. Heart J., July 2, 2007; 28(14): 1723 - 1730. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Assmus, U. Fischer-Rasokat, J. Honold, F. H. Seeger, S. Fichtlscherer, T. Tonn, E. Seifried, V. Schachinger, S. Dimmeler, and A. M. Zeiher Transcoronary Transplantation of Functionally Competent BMCs Is Associated With a Decrease in Natriuretic Peptide Serum Levels and Improved Survival of Patients With Chronic Postinfarction Heart Failure: Results of the TOPCARE-CHD Registry Circ. Res., April 27, 2007; 100(8): 1234 - 1241. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
2.