The use of group sequential, information-based sample size re-estimation in the design of the PRIMO study of chronic kidney disease

Yili Pritchett, Yannis Jemiai, Yuchiao Chang, Ishir Bhan, Rajiv Agarwal, Carmine Zoccali, Christoph Wanner, Donald Lloyd-Jones, Jorge B. Cannata-Andía, Taylor Thompson, Evan Appelbaum, Paul Audhya, Dennis Andress, Wuyan Zhang, Scott Solomon, Warren J. Manning, Ravi Thadhani

Research output: Contribution to journalArticle

13 Citations (Scopus)

Abstract

Background: Chronic kidney disease is associated with a marked increase in risk for left ventricular hypertrophy and cardiovascular mortality compared with the general population. Therapy with vitamin D receptor activators has been linked with reduced mortality in chronic kidney disease and an improvement in left ventricular hypertrophy in animal studies. Purpose: PRIMO (Paricalcitol capsules benefits in Renal failure Induced cardia MOrbidity) is a multinational, multicenter randomized controlled trial to assess the effects of paricalcitol (a selective vitamin D receptor activator) on mild to moderate left ventricular hypertrophy in patients with chronic kidney disease. Methods: Subjects with mild-moderate chronic kidney disease are randomized to paricalcitol or placebo after confirming left ventricular hypertrophy using a cardiac echocardiogram. Cardiac magnetic resonance imaging is then used to assess left ventricular mass index at baseline, 24 and 48 weeks, which is the primary efficacy endpoint of the study. Because of limited prior data to estimate sample size, a maximum information group sequential design with sample size re-estimation is implemented to allow sample size adjustment based on the nuisance parameter estimated using the interim data. An interim efficacy analysis is planned at a pre-specified time point conditioned on the status of enrollment. The decision to increase sample size depends on the observed treatment effect. A repeated measures analysis model, using available data at Week 24 and 48 with a backup model of an ANCOVA analyzing change from baseline to the final nonmissing observation, are pre-specified to evaluate the treatment effect. Gamma-family of spending function is employed to control family-wise Type I error rate as stopping for success is planned in the interim efficacy analysis. Limitations: If enrollment is slower than anticipated, the smaller sample size used in the interim efficacy analysis and the greater percent of missing week 48 data might decrease the parameter estimation accuracy, either for the nuisance parameter or for the treatment effect, which might in turn affect the interim decision-making. Conclusions: The application of combining a group sequential design with a sample-size re-estimation in clinical trial design has the potential to improve efficiency and to increase the probability of trial success while ensuring integrity of the study.

Original languageEnglish
Pages (from-to)165-174
Number of pages10
JournalClinical Trials
Volume8
Issue number2
DOIs
StatePublished - Apr 2011

Fingerprint

Chronic Renal Insufficiency
Sample Size
Left Ventricular Hypertrophy
Calcitriol Receptors
Social Adjustment
Cardia
Mortality
Therapeutics
Capsules
Renal Insufficiency
Decision Making
Randomized Controlled Trials
Placebos
Magnetic Resonance Imaging
Observation
Clinical Trials
Morbidity
Population
paricalcitol

ASJC Scopus subject areas

  • Medicine(all)
  • Pharmacology

Cite this

The use of group sequential, information-based sample size re-estimation in the design of the PRIMO study of chronic kidney disease. / Pritchett, Yili; Jemiai, Yannis; Chang, Yuchiao; Bhan, Ishir; Agarwal, Rajiv; Zoccali, Carmine; Wanner, Christoph; Lloyd-Jones, Donald; Cannata-Andía, Jorge B.; Thompson, Taylor; Appelbaum, Evan; Audhya, Paul; Andress, Dennis; Zhang, Wuyan; Solomon, Scott; Manning, Warren J.; Thadhani, Ravi.

In: Clinical Trials, Vol. 8, No. 2, 04.2011, p. 165-174.

Research output: Contribution to journalArticle

Pritchett, Y, Jemiai, Y, Chang, Y, Bhan, I, Agarwal, R, Zoccali, C, Wanner, C, Lloyd-Jones, D, Cannata-Andía, JB, Thompson, T, Appelbaum, E, Audhya, P, Andress, D, Zhang, W, Solomon, S, Manning, WJ & Thadhani, R 2011, 'The use of group sequential, information-based sample size re-estimation in the design of the PRIMO study of chronic kidney disease', Clinical Trials, vol. 8, no. 2, pp. 165-174. https://doi.org/10.1177/1740774511399128
Pritchett, Yili ; Jemiai, Yannis ; Chang, Yuchiao ; Bhan, Ishir ; Agarwal, Rajiv ; Zoccali, Carmine ; Wanner, Christoph ; Lloyd-Jones, Donald ; Cannata-Andía, Jorge B. ; Thompson, Taylor ; Appelbaum, Evan ; Audhya, Paul ; Andress, Dennis ; Zhang, Wuyan ; Solomon, Scott ; Manning, Warren J. ; Thadhani, Ravi. / The use of group sequential, information-based sample size re-estimation in the design of the PRIMO study of chronic kidney disease. In: Clinical Trials. 2011 ; Vol. 8, No. 2. pp. 165-174.
@article{98deaac112684f5dad334850529fd3e4,
title = "The use of group sequential, information-based sample size re-estimation in the design of the PRIMO study of chronic kidney disease",
abstract = "Background: Chronic kidney disease is associated with a marked increase in risk for left ventricular hypertrophy and cardiovascular mortality compared with the general population. Therapy with vitamin D receptor activators has been linked with reduced mortality in chronic kidney disease and an improvement in left ventricular hypertrophy in animal studies. Purpose: PRIMO (Paricalcitol capsules benefits in Renal failure Induced cardia MOrbidity) is a multinational, multicenter randomized controlled trial to assess the effects of paricalcitol (a selective vitamin D receptor activator) on mild to moderate left ventricular hypertrophy in patients with chronic kidney disease. Methods: Subjects with mild-moderate chronic kidney disease are randomized to paricalcitol or placebo after confirming left ventricular hypertrophy using a cardiac echocardiogram. Cardiac magnetic resonance imaging is then used to assess left ventricular mass index at baseline, 24 and 48 weeks, which is the primary efficacy endpoint of the study. Because of limited prior data to estimate sample size, a maximum information group sequential design with sample size re-estimation is implemented to allow sample size adjustment based on the nuisance parameter estimated using the interim data. An interim efficacy analysis is planned at a pre-specified time point conditioned on the status of enrollment. The decision to increase sample size depends on the observed treatment effect. A repeated measures analysis model, using available data at Week 24 and 48 with a backup model of an ANCOVA analyzing change from baseline to the final nonmissing observation, are pre-specified to evaluate the treatment effect. Gamma-family of spending function is employed to control family-wise Type I error rate as stopping for success is planned in the interim efficacy analysis. Limitations: If enrollment is slower than anticipated, the smaller sample size used in the interim efficacy analysis and the greater percent of missing week 48 data might decrease the parameter estimation accuracy, either for the nuisance parameter or for the treatment effect, which might in turn affect the interim decision-making. Conclusions: The application of combining a group sequential design with a sample-size re-estimation in clinical trial design has the potential to improve efficiency and to increase the probability of trial success while ensuring integrity of the study.",
author = "Yili Pritchett and Yannis Jemiai and Yuchiao Chang and Ishir Bhan and Rajiv Agarwal and Carmine Zoccali and Christoph Wanner and Donald Lloyd-Jones and Cannata-And{\'i}a, {Jorge B.} and Taylor Thompson and Evan Appelbaum and Paul Audhya and Dennis Andress and Wuyan Zhang and Scott Solomon and Manning, {Warren J.} and Ravi Thadhani",
year = "2011",
month = "4",
doi = "10.1177/1740774511399128",
language = "English",
volume = "8",
pages = "165--174",
journal = "Clinical Trials",
issn = "1740-7745",
publisher = "SAGE Publications Ltd",
number = "2",

}

TY - JOUR

T1 - The use of group sequential, information-based sample size re-estimation in the design of the PRIMO study of chronic kidney disease

AU - Pritchett, Yili

AU - Jemiai, Yannis

AU - Chang, Yuchiao

AU - Bhan, Ishir

AU - Agarwal, Rajiv

AU - Zoccali, Carmine

AU - Wanner, Christoph

AU - Lloyd-Jones, Donald

AU - Cannata-Andía, Jorge B.

AU - Thompson, Taylor

AU - Appelbaum, Evan

AU - Audhya, Paul

AU - Andress, Dennis

AU - Zhang, Wuyan

AU - Solomon, Scott

AU - Manning, Warren J.

AU - Thadhani, Ravi

PY - 2011/4

Y1 - 2011/4

N2 - Background: Chronic kidney disease is associated with a marked increase in risk for left ventricular hypertrophy and cardiovascular mortality compared with the general population. Therapy with vitamin D receptor activators has been linked with reduced mortality in chronic kidney disease and an improvement in left ventricular hypertrophy in animal studies. Purpose: PRIMO (Paricalcitol capsules benefits in Renal failure Induced cardia MOrbidity) is a multinational, multicenter randomized controlled trial to assess the effects of paricalcitol (a selective vitamin D receptor activator) on mild to moderate left ventricular hypertrophy in patients with chronic kidney disease. Methods: Subjects with mild-moderate chronic kidney disease are randomized to paricalcitol or placebo after confirming left ventricular hypertrophy using a cardiac echocardiogram. Cardiac magnetic resonance imaging is then used to assess left ventricular mass index at baseline, 24 and 48 weeks, which is the primary efficacy endpoint of the study. Because of limited prior data to estimate sample size, a maximum information group sequential design with sample size re-estimation is implemented to allow sample size adjustment based on the nuisance parameter estimated using the interim data. An interim efficacy analysis is planned at a pre-specified time point conditioned on the status of enrollment. The decision to increase sample size depends on the observed treatment effect. A repeated measures analysis model, using available data at Week 24 and 48 with a backup model of an ANCOVA analyzing change from baseline to the final nonmissing observation, are pre-specified to evaluate the treatment effect. Gamma-family of spending function is employed to control family-wise Type I error rate as stopping for success is planned in the interim efficacy analysis. Limitations: If enrollment is slower than anticipated, the smaller sample size used in the interim efficacy analysis and the greater percent of missing week 48 data might decrease the parameter estimation accuracy, either for the nuisance parameter or for the treatment effect, which might in turn affect the interim decision-making. Conclusions: The application of combining a group sequential design with a sample-size re-estimation in clinical trial design has the potential to improve efficiency and to increase the probability of trial success while ensuring integrity of the study.

AB - Background: Chronic kidney disease is associated with a marked increase in risk for left ventricular hypertrophy and cardiovascular mortality compared with the general population. Therapy with vitamin D receptor activators has been linked with reduced mortality in chronic kidney disease and an improvement in left ventricular hypertrophy in animal studies. Purpose: PRIMO (Paricalcitol capsules benefits in Renal failure Induced cardia MOrbidity) is a multinational, multicenter randomized controlled trial to assess the effects of paricalcitol (a selective vitamin D receptor activator) on mild to moderate left ventricular hypertrophy in patients with chronic kidney disease. Methods: Subjects with mild-moderate chronic kidney disease are randomized to paricalcitol or placebo after confirming left ventricular hypertrophy using a cardiac echocardiogram. Cardiac magnetic resonance imaging is then used to assess left ventricular mass index at baseline, 24 and 48 weeks, which is the primary efficacy endpoint of the study. Because of limited prior data to estimate sample size, a maximum information group sequential design with sample size re-estimation is implemented to allow sample size adjustment based on the nuisance parameter estimated using the interim data. An interim efficacy analysis is planned at a pre-specified time point conditioned on the status of enrollment. The decision to increase sample size depends on the observed treatment effect. A repeated measures analysis model, using available data at Week 24 and 48 with a backup model of an ANCOVA analyzing change from baseline to the final nonmissing observation, are pre-specified to evaluate the treatment effect. Gamma-family of spending function is employed to control family-wise Type I error rate as stopping for success is planned in the interim efficacy analysis. Limitations: If enrollment is slower than anticipated, the smaller sample size used in the interim efficacy analysis and the greater percent of missing week 48 data might decrease the parameter estimation accuracy, either for the nuisance parameter or for the treatment effect, which might in turn affect the interim decision-making. Conclusions: The application of combining a group sequential design with a sample-size re-estimation in clinical trial design has the potential to improve efficiency and to increase the probability of trial success while ensuring integrity of the study.

UR - http://www.scopus.com/inward/record.url?scp=79959252941&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79959252941&partnerID=8YFLogxK

U2 - 10.1177/1740774511399128

DO - 10.1177/1740774511399128

M3 - Article

C2 - 21478328

AN - SCOPUS:79959252941

VL - 8

SP - 165

EP - 174

JO - Clinical Trials

JF - Clinical Trials

SN - 1740-7745

IS - 2

ER -