Dr G. T. Everson, Director of Hepatology, University of Colorado Health Sciences Center, 4200 East 9th Avenue, B-154, Denver, CO 80262, USA. E-mail: firstname.lastname@example.org
Background Measurement of portal inflow and portal-systemic shunt using cholate clearances could be useful in monitoring patients with liver disease.
Aim To examine relationships of cholate clearances and shunt to cirrhosis and varices and to define minimal sampling requirements.
Methods Five hundred forty-eight studies were performed in 282 patients enrolled in the Hepatitis C Antiviral Long-term Treatment to prevent Cirrhosis (HALT-C) trial. Stable, non-radioactive isotopes of cholate were administered intravenously and orally, clearances (Cliv and Cloral) were calculated from [dose/area under curve (AUC)] and cholate shunt from [(AUCoral:AUCiv) x (Doseiv:Doseoral) x 100%].
Results Cholate Cloral and cholate shunt correlated with prevalences of both cirrhosis and varices (P < 0.0001 for all). Peripheral venous sampling at 5, 20, 45, 60 and 90 min defined the minimal model. Linear regression of cholate shunt determined from five points within 90 min vs. the standard method of 14 points over 3 h yielded slope of 1.0 and intercept 0.5% (r2 = 0.98, P < 0.0001). Results were identical in the 189 validation studies (slope 1.0, intercept 0.5%, r2 = 0.99, P < 0.0001).
Conclusions Cholate Cloral and cholate shunt may be useful in monitoring patients with liver disease. The 5-point model enhances application of cholate Cloral and cholate shunt in the non-invasive assessment of the portal circulation.
Chronic liver disease affects over 10 million Americans, and is the 12th leading cause of death in the United States.1 Chronic hepatitis C (CHC) alone currently affects 4 million Americans and is responsible for 10 000 deaths annually.2, 3 By 2020, the number of patients with HCV-related cirrhosis will double, incidence of hepatocellular carcinoma will increase by 81%, and HCV-related death will increase by 180%.2 Although chronic liver disease is a major cause of morbidity and mortality, methods for tracking disease progression are woefully inadequate.
Major clinical complications of chronic liver disease, such as variceal haemorrhage, ascites and encephalopathy, are related to portal hypertension and portal-systemic shunting. Despite this central role, the portal circulation is incompletely defined using standard tests.4 Platelet count is the one standard test that may correlate with portal hypertension as platelets may be sequestered in the spleen as portal pressure increases. However, platelet count relates only indirectly to portal hypertension and many other factors, such as diminished thrombopoeitin production, intercurrent disease, medications and impaired bone marrow function, may cause thrombocytopenia in patients with chronic liver disease. Furthermore, many patients with advanced fibrosis or cirrhosis, who have portal hypertension, impaired portal blood flow and portal-systemic shunt, lack thrombocytopenia. Other tests of the portal circulation, such as ultrasonographic measurement of spleen size, are insensitive, and still others, such as transjugular measurement of portal pressure, are invasive.
Cholate shunt directly measures first pass hepatic extraction of cholate, a function influenced primarily by portal blood flow and portal-systemic shunt. It is defined as the percentage of orally administered cholate that escapes hepatic extraction from the portal circulation. In recent studies primarily of patients with CHC, cholate shunt correlated with clinical manifestations of portal hypertension and response to anti-viral therapy,5–8 and was more sensitive than standard laboratory tests in detecting fibrotic stages of liver disease.5 However, the method for measuring cholate shunt required a sampling period of 3 h and 14 samples of blood.
The primary objectives of the current study were to examine relationships of cholate shunt to cirrhosis and varices and to define the minimal sampling requirements for accurate measurement of cholate shunt in man.
Patients and methods
This study and associated consent forms were approved by the National Institute of Digestive and Kidney Disease, US Food and Drug Administration (FDA), institutional review boards, General Clinical Research Centers (GCRCs) and other regulatory bodies within the participating centres. The study was conducted according to the principals of the Declaration of Helsinki regarding the proper procedures for human research. All subjects participating in this study had signed individual informed consents for participation in this study.
For this analysis we used 548 studies of 282 patients with CHC enrolled in the Hepatitis C Antiviral Long-term Treatment against Cirrhosis (HALT-C) Trial9 who participated in the quantitative liver function test (QLFT) ancillary study.5 HALT-C patients are characterized by hepatic fibrosis (Ishak fibrosis scores 2–4) or compensated cirrhosis (Ishak fibrosis score 5 or 6), non-response to prior courses of anti-viral therapy, and exhibit a broad range of functional impairment5 and clinical manifestations.10 These results were compared with results from 32 healthy controls. The first 359 studies of HALT-C patients were used for definition of the minimal model, and the model was validated in the next 189 HALT-C studies.
2,2,4,4-2H cholate (CDN Isotopes Inc., Quebec, Canada, product # D-2452) was administered orally (40 mg) and studied under FDA Investigational New Drug (IND) application no. 65,123. A solution (20 mg in 5 cm3 NaHCO3 1 mmol/mL) of 24-13C cholate (CDN Isotopes Inc., product # C-3448) was studied under FDA IND 65,121. The solution of 24-13C cholate was passaged through micropore filter and transferred to sterile glass vials; sterility and absence of pyrogens were confirmed prior to use.
24-13C cholic acid solution, 20 mg, was first mixed with 5 mL human serum albumin (25% w/v), and then administered intravenously through an indwelling intravenous catheter over 2 min. 2,2,4,4-2H cholic acid, 40 mg, was dissolved in water, mixed in juice and taken orally at the same time as the intravenous injection. Peripheral venous blood samples for measurement of serum concentrations, were drawn through the indwelling catheter and obtained prior to and 5, 10, 15, 20, 30, 45, 60, 75, 90, 105, 120, 150 and 180 min after administration of cholate isotopes.
Aliquots of serum (0.5 mL) were dispensed, 1.5 μg of unlabelled cholate was added as internal standard, and cholates were isolated from serum using C18 liquid chromatographic cartridges. The eluate was acidified with concentrated HCl and cholates extracted from the eluate with diethyl ether, methylated, and derivatized to trimethyl-silyl ethers. Isotopes were quantified by isotope dilution – mass spectrometry and selected ion monitoring (m/z 458, 459, 462; Agilent GC 6890 MS 5973N, Agilent Technologies, Paulo Alto, CA, USA), equipped with an HP-1 MS 30 m x 25 mm column (Agilent Technologies).11, 12
A full description of the methods and mathematical models used in curve fitting, measurement of AUC and analysis of models is provided in Supplementary Material – Appendix S1.13–15
Clearances and shunt fraction
Areas under serum concentration curves (AUCs) were measured and clearances and shunt were calculated. Intravenous clearance (Cliv) was defined by dose24-13C cholate/AUC24-13C cholate, oral clearance (Cloral) by dose2,2,4,4-2H cholate/AUC2,2,4,4-2H cholate and cholate shunt by [AUCoral:AUCiv] x [Doseiv:Doseoral] x 100%.
Distributions of test results were defined by mean, median, standard deviation and quartiles spanning the full range of results. Models using reduced sampling were evaluated for accuracy in determination of cholate shunt, by comparison with results from the full 14 samples obtained over 3 h. Correlations of results of models with the full 14-point sampling period were evaluated by linear regression analysis and Spearman correlation coefficients.16–18
Serum concentrations of administered isotopes
Mean serum concentrations (±s.d.) after oral and intravenous administration of cholate isotopes in the first 359 studies are illustrated in Figure 1. The mean values and standard deviations of serum concentrations of cholate isotopes reflect the variability of liver disease within this cohort with CHC and fibrosis or compensated cirrhosis. Healthy controls have lower mean values and standard deviations (data not shown). Despite the interindividual variation in serum concentrations, due to varying severity of underlying liver disease, the overall shape of both intravenous and oral clearance curves was generally similar from patient to patient and in healthy controls.
Baseline cholate clearances and shunt
The full range and boundaries for quartiles of results of cholate clearances and shunt in the 282 patients at baseline, prior to entry into the Trial, are shown in Table 1. Clearances progressively declined and shunt progressively increased as results ranged from ‘best’ to ‘worst’. Cholate shunts spanned the entire range of expected result, from the low end of the normal range, 10% (‘Best’), to nearly complete shunting, 91% (‘Worst’).
Table 1. Range of results for cholate clearances and cholate shunt in study patients
Boundaries for quartiles of test results*
Cliv, clearance after intravenous administration of cholate; Cloral, clearance after oral administration of cholate.
* The ranges of results of baseline tests in the 282 study patients are defined by boundaries for the quartiles of results. The column headed ‘Best’ lists the maximum values for cholate clearances and minimum value for cholate shunt; the column headed ‘Worst’ lists the minimum values for cholate clearances and maximum value for cholate shunt. Clearances decline and shunt increases from ‘Best’ to ‘Worst’ quartiles. Cliv is the clearance after intravenous administration of cholate and Cloral is the clearance after oral administration of cholate.
Cholate Cliv (mL/min)
Cholate Cloral (mL/min)
Cholate shunt (%)
Results in 32 healthy controls were (mean ± s.d., range): Cliv 390 ± 136, 155–873 mL/min; Cloral 2173 ± 677, 1369–3856 mL/min; and shunt 18.5 ± 5.5, 8.0–28.5%. The range of Cliv of HALT-C patients (Table 1) completely overlapped with Cliv for these healthy controls. In contrast, approximately 70% of HALT-C patients exceeded the normal limits for Cloral of 1300 mL/min and shunt of 30% (Table 1). Consequently, both Cloral and shunt, but not Cliv, may be useful for defining risk of cirrhosis and varices.
Correlations of cholate clearances and shunt with cirrhosis and varices
One hundred and thirteen of the 282 study patients had cirrhosis. The numbers of patients and percentages of patients with cirrhosis within quartiles of test results are given in Table 2, panel A. Cloral was ≤1300 mL/min and shunt was ≥30% in 87% and 88% of patients with cirrhosis. The prevalence of cirrhosis increased as Cloral decreased (P < 0.0001) and shunt increased (P < 0.0001).
Table 2. Prevalence of cirrhosis and varices within quartiles of cholate clearances and shunt
Number of patients within each quartile of test result
Percentage of patients with cirrhosis within each quartile of test result
Cliv, clearance after intravenous administration of cholate; Cloral, clearance after oral administration of cholate; Q1, the quartile of patients with least impaired function; Q2 and Q3, the quartiles of patients with intermediate functional impairment; Q4, the quartile of patients with most impaired function; CMH, Cochran-Mantel-Haenszel test of significance of the trend in percentage of patients with cirrhosis from Q1 to Q4.
* Two hundred and eighty-two patients underwent testing and 113 of these had cirrhosis. Panel A displays the numbers of patients and prevalences of cirrhosis within quartiles of test results. Prevalences of cirrhosis were lowest in Q1, the quartiles with best test results, and highest in Q4, the quartiles with worst test results. Cholate Cloral and cholate shunt, but not cholate Cliv, correlated significantly with prevalence of cirrhosis, which increased progressively from Q1 to Q4; † Two hundred and twenty-two patients underwent endoscopy and 75 had varices. Panel B shows the numbers of patients and prevalences of varices within quartiles of results. Prevalences of varices were lowest in Q1 and highest in Q4. Cholate Cloral and cholate shunt correlated best with prevalence of varices, which increased progressively from Q1 to Q4.
Panel A: Prevalence of cirrhosis*
Percentage of patients with varices within each quartile of test result
Panel B: Prevalence of varices†
Seventy-five of 222 study patients that underwent endoscopy had varices. Sixty-five of these 75 patients (87%) had Cloral≤1300 mL/min and 66 (88%) had shunt ≥30%. In contrast, only 48 of the 75 patients with varices (64%) had biopsy-proven cirrhosis. The varices of patients with normal Cloral (n = 10) and shunt (n = 9) were all classified as small. The latter patients tended to have higher platelet counts, normal spleen size, and only one of the 10 with normal Cloral and two of the nine with normal shunt had cirrhosis on biopsy.
Prevalence of varices increased as Cloral decreased (P < 0.0001) and shunt increased (P < 0.0001; Table 2, panel B). Twenty-two of the 75 patients with varices had medium (n = 15) or large (n = 7) varices. All had Cloral≤1300 mL/min and shunt ≥30% and all of the seven patients who had large varices had shunt ≥45%. In contrast, only 17 of the 22 patients with medium to large varices (77%) had biopsy-proven cirrhosis. Altogether, these results suggest that Cloral and shunt may be more sensitive than stage of fibrosis on liver biopsy in detecting patients with varices.
Cloral and shunt also correlated with other measures of severity of liver disease, including Ishak fibrosis score, variceal size, splenomegaly, portal hypertensive gastropathy and standard laboratory tests, such as bilirubin, INR, albumin and platelet count (data not shown).
Characteristics of clearance of intravenously administered 24-13C cholate
The mean serum concentrations of intravenously administered 24-13C cholate for the 14 time points are illustrated in Figure 2. Three phases of clearance were identified: 0–20, 20–45 and 45–180 min. Equations were defined for each of the three phases of elimination, bracketed by the four time points of 5, 20, 45 and 90 min (see Supplementary Material – Appendix S1 for derivation of equations). The equation defining elimination rate between 45 and 90 min was extrapolated to 180 min. The area under the curve (AUC) of plasma concentration vs. time was determined from the four time points by integration of the resultant tri-exponential equation. The AUC of the 4-point model was within 98 ± 1.7% of the AUC calculated from all 14 points. In the final combined 5-point model defined below, only the four time points noted above are used for the intravenous curve.
Characteristics of clearance of orally administered 2,2,4,4-2H cholate
The mean serum concentrations of orally administered 2,2,4,4-2H cholate for the 14 time points are illustrated in Figure 3. The initial increase in systemic concentration of 2,2,4,4-2H cholate occurred at 5 min and peak concentrations occurred between 30 and 60 min. We analysed various combinations of time points for generation of the oral curve and concluded that sampling at 5, 20, 45, 60 and 90 min would be required to bracket all inflection points in clearance curves from every patient. Cubic spline functions were defined for all time intervals between 0 and 90 min of the oral curve (Supplementary Material – Appendix S1). The final interval from 90 to 180 min was defined by a single exponential function using the mean rate constant of elimination (kelim) for this interval from all 359 studies, 0.01295 min−1.
The AUC of plasma concentration vs. time was integrated using the respective spline and exponential functions. The AUC using the five time points was within 98 ± 1% of the AUC calculated from the 14 time points.
Primary end point: cholate shunt
Cholate shunts calculated from 5 and 14 time points for each individual were nearly identical – failing to match in only one study due to improper administration of the intravenous dose. Cholate shunt, calculated using five points, was 98 ± 1% of cholate shunt measured with 14 time points. Linear regression of cholate shunts measured by five points vs. 14 points yielded a straight line, indistinguishable from the line of identity (slope 1.0, intercept 0.5%, r = 0.99, P < 0.0001; Figure 4). Linear regression analysis of results from five and 14 point methods are given in Table 3. Regression analysis of five point vs. 14 point models for shunt and clearances of intravenously or orally administered cholates yielded slopes indistinguishable from the lines of identity and intercepts equivalent to zero.
Table 3. Comparison of clearances and shunt determined from the 5-point minimal model with results from the standard 14-point method
Cliv, clearance after intravenous administration of cholate; Cloral, clearance after oral administration of cholate; r, correlation coefficient; P, P-value of probability.
* The model-building set consisted of the first 359 studies of cholate clearances and shunt; 282 were performed at baseline and an additional 77 in follow-up studies of the same patients. All correlations of results from the 5-point model with results from the 14-point standard method were highly significant. The trendline of linear regression of Cliv, Cloral and shunt determined from the 5-point model vs. the standard 14-point method approximated the line of identity; † The validation set consisted of an additional 189 follow-up studies of cholate clearances and shunt performed in the same patients. All correlations of results from the 5-point model with results from the 14-point standard method were highly significant. Once again, the trendline of linear regression of Cliv, Cloral and shunt from the 5-point model vs. the standard 14-point method approximated the line of identity. The 5-point model of sampling for only 90 min accurately measures Cliv, Cloral and shunt.
Panel A: Model-building set: results of linear regression analysis*
Panel B: Validation set: results of linear regression analysis†
Evaluation of other models
After defining time period and sampling intervals, we then examined variations in the number of samples required to define cholate shunt by analysing models using 2, 3, 4, 6 and 7 time points. For 2-point models, we used the average k1 and k2 values for the exponential elimination rates for the intravenous curve. AUCs were calculated by integrating from the intersection of the two elimination rates to respective end points. For the 3-point model, the intravenous curve used one average k elimination rate and calculated the other k elimination rate depending on the points selected. The 4-, 6- and 7-point intravenous curve was generated as described above for the 5-point analysis. All models used the spline functions for the oral curve. The 2-, 3- and 4-point models used 5 min as one time point and all combinations of time points at 20, 45, 60 and 90 min. The 6- and 7-point models used time points at 5, 20, 45, 60 and 90 time points, but also different combinations of time points at 15, 75, 105, 120 and 180 min. The 5-point model proved to be optimal in the determination of cholate shunt (Figure 5).
Validation of 5-point model
To validate the 5-point model, we applied the model prospectively to the next 189 studies. Linear regression analysis of results from five and 14 point methods using this validation set are given in Table 3. Regression analysis of five point vs. 14 point models for shunt and clearances of intravenously or orally administered cholates yielded slopes indistinguishable from the line of identical and intercepts equivalent to zero. Shunt and clearances of intravenously or orally administered cholates were, once again, accurately defined using the 5-point model.
A major deficiency of the current clinical assessment of chronic liver disease is the inability of standard laboratory tests to adequately and non-invasively assess hepatic function, the portal circulation and portal-systemic shunt.4 Quantification of hepatic physiology and pathophysiology by use of test compounds, such as cholate, may allow better definition of the portal circulation and enhance the evaluation of patients with chronic liver disease.19–22
In this study, we demonstrated that cholate Cloral, using a cut-off of 1300 mL/min, and shunt, using a cut-off of 30%, may be useful in detecting patients with cirrhosis and varices – both tests were more sensitive than liver biopsy in detection of patients with varices. Previously we reported that cholate shunt and cholate Cloral correlated with Ishak fibrosis score, variceal size and grade of portal hypertensive gastropathy at endoscopy, spleen size on ultrasonography and routine laboratory tests of liver dysfunction.5 In addition, other analyses have indicated that cholate shunt and cholate Cloral may be useful in prediction of response to anti-viral therapy,6, 7 and risk of future decompensation.8 They may also define improvements in hepatic function that are undetected by routine laboratory tests.7 These findings suggest that cholate Cloral and shunt could have potential clinical utility.
Some patients with small varices had normal Cloral and shunt. These patients exhibited few signs of portal hypertension – platelet counts and spleen size were generally normal and most had non-cirrhotic liver biopsies. This suggests that in the early stages of portal hypertension, hepatic extraction of cholate may remain normal via compensatory mechanisms that improve portal and hepatic blood flow or enhance the hepatocellular uptake of cholate.
Standard methods for measuring cholate Cloral and shunt, requiring 14 samples of blood collected over 3 h, are clinically impractical. For these reasons we used mathematical methods to define the minimal sampling requirements – five time points within 90 min of administration of cholates. Our 5-point model greatly simplifies performance of the test increasing its potential for broader application in the assessment of patients with liver disease. Switching from 14 time points to five time points reduces patient phlebotomies by 64%, patient sampling time by 50%, and laboratory analysis and sample preparation by 64%.
The stable isotope-labelled cholates offered several unique advantages over radioisotope-labelled cholate or ursodeoxycholate used in other studies.23, 24 First, use of stable isotopes avoids exposure of human subjects to radioactivity. Secondly, quantification by mass spectrometry–isotope ratiometry is inherently more accurate because of the added capability for compound identification. Other problems inherent with use of radiolabelled compounds, such as sequential study of intravenous and oral clearances which prolongs sampling and increases phlebotomies,23, 24 or use of multiple radioisotopes which increases radiation exposure,23 are avoided. With stable isotopes, because test compounds are separated by mass, multiple stable isotopes of the same compound may be administered simultaneously.
The 5-point model is generally applicable to all test compounds that exhibit certain characteristics comparable to cholate. Key characteristics include relatively high first-pass hepatic elimination, rapid and complete intestinal absorption, retention of the test compound in the intravascular space, lack of dependency on hepatic metabolism, lack of renal excretion and lack of direct effects of the test compound on the cardiovascular system or portal circulation. Cholate fulfils all of these criteria. Further validation of cholate as a appropriate test compound for assessment of portal blood flow and portal-systemic shunt is the fact that our measurement of mean cholate shunt in healthy intact human subjects (18%, or 82% first-pass hepatic elimination from the portal vein)8 is identical to expected, based upon other studies of the liver or liver cells.
Non-invasive tests have been used for prediction of fibrosis or cirrhosis.4 Methods include models using standard laboratory tests and other blood components,25–29 clearance of test compounds,19–22, 30–33 Doppler ultrasonography,34, 35 transient ultrasonographic elastography36–38 and magnetic resonance elastography.39 However, none of these methods accurately quantifies the portal circulation or portal-systemic shunting. In contrast, our cholate method not only quantifies the portal circulation and portal-systemic shunting, but also correlates well with cirrhosis, varices and variceal size.
In HALT-C patients, cholate Cloral and cholate shunt compared favourably to other quantitative tests of liver function. They correlated better with cirrhosis, varices, standard laboratory tests and response to anti-viral therapy than caffeine elimination, antipyrine elimination and clearance, galactose elimination capacity, MEGX generation from lidocaine and methionine breath test.5–7 In a study of other patients, cholate Cloral and cholate shunt were better than caffeine elimination or antipyrine elimination and clearance in identifying patients at risk of future decompensation.8
The portal circulation may also be assessed by measurement of portal pressure. Elevation of hepatic venous portal pressure gradient (HVPG), calculated from the difference between free and wedged hepatic venous pressure, identifies patients at risk for complications of portal hypertension and may predict survival in patients with cirrhosis.40 However, the transjugular technique for measuring HVPG is invasive, requiring percutaneous puncture of the internal jugular vein. The technique is expensive and time and labour intensive requiring procedure room, fluoroscopy, skilled medical personnel and postprocedure monitoring. Significant complications, such as carotid artery puncture, neck haematoma, venous thrombosis, pneumothorax and hemothorax, although rare, may occur in up to 1–6% of cases.40 In our studies we have demonstrated that cholate Cloral and cholate shunt correlate strongly with prevalence of cirrhosis and varices. Advantages of the cholate method over transjugular measurement of portal pressure are the non-invasive method, lack of requirement for specialized rooms, equipment, personnel and training, and absence of any known complications, except those associated with phlebotomy from a peripheral vein. Although comparative studies are needed, characterization of the portal circulation by cholate Cloral and cholate shunt could complement or possibly replace portal pressure measurements.
We conclude that cholate Cloral and cholate shunt may be useful in the assessment of patients with chronic liver disease. The 5-point model represents a new and accurate method that reduces expenses, minimizes phlebotomy and time commitment by patients, and limits laboratory and analytical processing. These benefits could broaden application of the 5-point method for measuring cholate shunt and cholate Cloral in non-invasive assessment of the portal circulation.
Declaration of personal interests: Financial relationships of the authors with Hoffmann-La Roche, Inc., are as follows: G. T. Everson is a consultant, on the speaker’s bureau, and receives research support; M. L. Shiffman is a consultant, on the speaker’s bureau, and receives research support; T. R. Morgan is on the speaker’s bureau, a consultant and receives research support; J. C. Hoefs is on the speaker’s bureau; and R. K. Sterling is a consultant, on the speaker’s bureau, and receives research support;
Other financial relationships related to this project are: G. T. Everson, M. L. Shiffman and T. R. Morgan received research support from Metabolic Solutions, Inc.; R. K. Sterling is a consultant and receives support from Wako Diagnostics.
G. T. Everson and M. A. Martucci hold intellectual property rights under US Patent Application No. 60/647,689, ‘Methods for Diagnosis and Intervention of Hepatic Disorders’, 26 January 2005, and International Application Number PCT/US2006/003132 as published under the Patent Cooperation Treaty, World Intellectual Property Organization, International Patent Classification A61 K 49/00 (2006.01), International Publication Number WO 2006/081521 A2, 3 August 2006. The minimal model for measurement of cholate shunt is included in this patent. At the time of this submission there has been no commercial development of either the methods or models presented in this manuscript.
Declaration of funding interests: This study was supported by the National Institute of Diabetes, Digestive & Kidney Diseases (contract numbers are listed below). Additional support was provided by the National Institute of Allergy and Infectious Diseases, the National Cancer Institute, the National Center for Minority Health and Health Disparities and by General Clinical Research Center grants from the National Center for Research Resources, National Institutes of Health (grant numbers are listed below). Additional funding to conduct this study was supplied by Metabolic Solutions, Inc. and by Hoffmann-La Roche, Inc., through a Cooperative Research and Development Agreement with the National Institutes of Health.
The authors wish to acknowledge the contributions of our co-investigators, study coordinators and staff at each of the participating institutions as follows:
University of Colorado School of Medicine, Denver, CO: (Contract N01-DK-9-2327, Grant M01RR-00051) Marcelo Kugelmas MD, Jennifer DeSanto RN, Carol McKinley RN, Brenda Easley RN, Shannon Lauriski BS, Stephanie Shea BA, Seda Ozdemir BS and Michelle Jaramillo.
Virginia Commonwealth University Medical Center, Richmond, VA: (Contract N01-DK-9-2322, Grant M01RR-00065) Charlotte Hofmann RN, Paula Smith 22RN.
University of California – Irvine, Irvine, CA: (Contract N01-DK-9-2320, Grant M01RR-00827) Muhammad Sheikh MD, Norah Milne MD, Choon Park RN, William Rietkerk, Richard Kesler-West.