S

CiAiaccvr

HMm6

pE

Aj

A

A

PECIAL ARTICLE

Proteinuria as a Surrogate Outcome in CKD: Report of a ScientificWorkshop Sponsored by the National Kidney Foundation and the US

Food and Drug Administration

Andrew S. Levey, MD,1 Daniel Cattran, MD,2 Aaron Friedman, MD,3 W. Greg Miller, PhD,4

John Sedor, MD,5 Katherine Tuttle, MD,6 Bertram Kasiske, MD,7 and Thomas Hostetter, MD8

Changes in proteinuria have been suggested as a surrogate outcome for kidney diseaseprogression to facilitate the conduct of clinical trials. This report summarizes a workshop sponsoredby the National Kidney Foundation and US Food and Drug Administration (FDA) with the followinggoals: (1) to evaluate the strengths and limitations of criteria for assessment of proteinuria as apotential surrogate end point for clinical trials in chronic kidney disease (CKD), (2) to explore thestrengths and limitations of available data for proteinuria as a potential surrogate end point, and (3)to delineate what more needs to be done to evaluate proteinuria as a potential surrogate end point.We review the importance of proteinuria in CKD, including the conceptual model for CKD,measurement of proteinuria and albuminuria, and epidemiological characteristics of albuminuria inthe United States. We discuss surrogate end points in clinical trials of drug therapy, including criteriafor drug approval, the definition of a surrogate end point, and criteria for evaluation of surrogacybased on biological plausibility, epidemiological characteristics, and clinical trials. Next, the reportsummarizes data for proteinuria as a potential surrogate outcome in 3 broad clinical areas: earlydiabetic kidney disease, nephrotic syndrome, and diseases with mild to moderate proteinuria. Weconclude with a synthesis of data and recommendations for further research. At the present time,there appears to be sufficient evidence to recommend changes in proteinuria as a surrogate forkidney disease progression in only selected circumstances. Further research is needed to defineadditional contexts in which changes in proteinuria can be expected to predict treatment effect. Werecommend collaboration among many groups, including academia, industry, the FDA, and theNational Institutes of Health, to share data from past and future studies.Am J Kidney Dis 54:205-226. © 2009 by the National Kidney Foundation, Inc.

INDEX WORDS: Proteinuria; glomerular filtration rate (GFR); chronic kidney disease (CKD).

tpltikcd

c

hronic kidney disease (CKD) is now widelyacknowledged as a public health problem

n the United States and around the world.1,2

pproximately 13% of the US adult populations estimated to have CKD, corresponding topproximately 26 million individuals.3 Out-omes of CKD include complications of de-reased kidney function, increased risk of cardio-ascular disease, and progression to kidney failureequiring treatment by dialysis or transplanta-

From 1Tufts Medical Center, Boston, MA; 2Toronto Generalospital, Toronto, ON, Canada; 3University of Minnesota,inneapolis, MN; 4Virginia Commonwealth University, Rich-ond, VA; 5Case Western Reserve University, Cleveland, OH;

University of Washington, Seattle and Spokane, WA; 7Henne-in County Medical Center, Minneapolis, MN; and 8Albertinstein College of Medicine, Bronx, NY.Received February 7, 2009. Accepted in revised form

pril 13, 2009. Originally published online as doi: 10.1053/.ajkd.2009.04.029 on July 6, 2009.

Because an author of this manuscript is an editor for

JKD, the peer-review and decision-making processes were

merican Journal of Kidney Diseases, Vol 54, No 2 (August), 2009

ion. Earlier stages of CKD can be detected inopulations at increased risk by using simpleaboratory tests; however, there are few therapieso slow the progression to kidney failure. Thencreasing incidence and prevalence of treatedidney failure, with poor outcomes and highost, highlight the urgent need to facilitate theevelopment of therapies for CKD.Nephrology lags behind other fields in medi-

ine in the conduct of clinical trials.4,5 One

andled entirely by an Associate Editor (Rulan Parekh, MD,S, Johns Hopkins Medical Institutions) who served ascting Editor-in-Chief. Details of the journal’s procedures

or potential editor conflicts are given in the Editorialolicies section of the AJKD website.Reprint requests to: Kerry Willis, PhD, National Kidney

oundation, 30 E 33rd St, New York, NY 10016. E-mail:erryw@kidney.org

© 2009 by the National Kidney Foundation, Inc.0272-6386/09/5402-0006$36.00/0

hMAfP

Fk

doi:10.1053/j.ajkd.2009.04.029

: pp 205-226 205

pdoamtgdurksd

aoifeeircckcsd

twOtsmp

lpcadg

co(Cfmaprfh

taeocpmawstrctwio

2

dtdlwb

C

F

Levey et al206

roblem in conducting clinical trials in CKD isefining end points. The progression of CKD isften slow, and until late stages, it is oftensymptomatic. Thus, end points for clinical trialsay be long delayed from disease onset and the

ime that interventions may be effective. Surro-ate end points may provide an opportunity toetect early evidence of effectiveness. Protein-ria is an accepted marker for kidney damage; iselated to diagnosis, prognosis, and treatment inidney disease; and has been suggested as aurrogate outcome for clinical trials of kidneyisease progression (Box 1).6-8

The US Food and Drug Administration (FDA)pproves drugs for use in the United States basedn demonstration of efficacy on clinically mean-ngful end points in pivotal clinical trials. Kidneyailure requiring dialysis or transplantation is anxample of an accepted clinically meaningfulnd point in CKD. The FDA also accepts changen kidney function, assessed as doubling of se-um creatinine level, as a surrogate end point forlinical trials of kidney disease progression be-ause it is expected to predict the development ofidney failure. Except in a very limited sense,hange in proteinuria has not been accepted as aurrogate end point for pivotal clinical trials forrug approval.In May, 2008, the National Kidney Founda-

ion (NKF) and FDA cosponsored a scientificorkshop entitled “Proteinuria as a Surrogateutcome in Chronic Kidney Disease.” The objec-

ives of the conference were to: (1) evaluate thetrengths and limitations of criteria for assess-ent of proteinuria as a potential surrogate end

Box 1. Importance of Proteinuria as a Biomarker

Marker of kidney damageClue to the diagnosis of CKDRisk factor for progression (causal in animal models)Modifier for efficacy of ACE-inhibitor therapy in nondia-

betic kidney diseaseHypothesized marker of vascular permeability (“gener-

alized endothelial dysfunction”)Risk factor for CVD at low levels (less than the

threshold for the definition of CKD)Hypothesized surrogate outcome for kidney disease

progression and CVD risk reduction

Abbreviations: ACE, angiotensin-converting enzyme;KD, chronic kidney disease; CVD, cardiovascular disease.Reproduced with permission of the National Kidney

oundation.6

oint in CKD; (2) explore the strengths and k

imitations of available data for proteinuria as aotential surrogate end point, focusing on spe-ific clinical circumstances and therapeuticgents; and (3) delineate what more needs to beone to evaluate proteinuria as a potential surro-ate end point.

1. SCOPE OF WORKSHOP

This conference built upon a prior workshoposponsored by the NKF and National Institutef Diabetes and Digestive and Kidney DiseasesNIDDK), “Proteinuria and Other Markers ofhronic Kidney Disease.”9 Recommendations

rom that workshop included standardization ofeasurement and terminology for proteinuria

nd albuminuria, periodic clinical assessment ofroteinuria as a marker of kidney damage, andesearch on proteinuria as a surrogate end pointor kidney disease progression. Since then, thereas been substantial progress.Topics for the current conference included: (1)

he importance of proteinuria in CKD, (2) evalu-tion of surrogacy in clinical trials, and (3)valuation of change in proteinuria as a surrogateutcome in kidney disease progression in 3 broadlinical areas: early diabetic kidney disease, ne-hrotic syndrome, and diseases with mild tooderate proteinuria. The conference agenda

nd slide presentations are posted on the NKFebsite10; attendees are listed in the online

upplementary material (Item S1; available withhis article at www.ajkd.org). This report brieflyeviews the conference proceedings and dis-usses some of the research recommendationshat resulted from the conference. The synthesisas prepared by the planning committee, with

nput from representatives from the FDA andther attendees.

2. THE IMPORTANCE OF PROTEINURIAIN CKD

.1. ConceptualModel for CKD

CKD is a heterogeneous condition, with manyifferent causes, manifestations, comorbid condi-ions, and factors affecting prognosis.11 It isefined as kidney damage or decreased glomeru-ar filtration rate (GFR) for 3 months or more,ith assessment of kidney damage from kidneyiopsy or markers of damage or a history of

idney transplantation and with estimation of

GaadoStshaswd

dnama3fttawp

blihutanac

flcprtrm

2a

ts

dserk mL/mp vens e

Proteinuria as a Surrogate Outcome in CKD 207

FR from equations using serum creatinine level,ge, sex, and race. Stages of CKD are definedccording to level of GFR, with kidney failureefined as GFR less than 15 mL/min/1.73 m2

r treatment by dialysis or transplantation.ymptoms caused by decreased kidney func-

ion generally appear at or shortly before thetage of kidney failure. Figure 1 shows aypothetical example of the change in GFRnd proteinuria during the course of progres-ive kidney disease.8 In practice, there is aide range for protein excretion and rate ofecrease in GFR in CKD.Proteinuria is one of many markers of kidney

amage. As described next, urine contains a largeumber of proteins, including albumin. Urinelbumin, rather than total protein, is now recom-ended for early detection of kidney damage in

dults, with albumin-creatinine ratio greater than0 mg/g as the threshold level.9,11 The rationaleor this threshold is that it is 2 to 3 times greaterhan the level in young healthy adults, is indica-ive of kidney damage in patients with diabetesnd other glomerular diseases, and is associatedith risk of progression to greater levels of

Figure 1. Hypothetical example of change in glomerularisease before kidney failure. Stages of chronic kidney disehown on the inner left vertical axis. Total protein-creatininelapsed time since the onset of kidney failure on the horizontaemains increased throughout. GFR remains normal for approidney failure after 30 years. Conversion factor for GFR inermission of the American Society of Nephrology from Ste

roteinuria and subsequent GFR decrease in dia- w

etic and nondiabetic individuals. Levels eveness than this threshold also are associated withncreased risk of cardiovascular disease. Someave suggested that very low levels of albumin-ria may reflect generalized endothelial dysfunc-ion rather than kidney disease per se. This latterrgument is difficult to resolve because the kid-ey is a highly vascular organ and appears to beffected prominently in disorders of the microvas-ulature.

Specific proteins other than albumin may re-ect damage to renal tubules in some diseaseonditions.12 In particular, low-molecular-weightroteins may originate from failure of tubulareabsorption of filtered proteins. In addition, pro-eins with molecular weight greater than albumineflect greater severity of kidney damage in glo-erular disease.

.2. Measurement of ProteinuriandAlbuminuria

Normal urine contains a large number of pro-eins. In healthy individuals, albumin makes up amall fraction of total protein, but in individuals

rate (GFR) and proteinuria during 30-year course of kidneyD) are indicated on the outer left vertical axis, and GFR isshown on the right vertical axis. Time in years is shown asroteinuria appears early in the course of kidney disease andly 15 years, then decreases, reaching levels associated within/1.72 m2 to mL/s/1.72 m2, �0.01667. Reproduced witht al.8

filtrationase (CKratio isl axis. Pximate

ith increased urine total protein levels, albumin

imnmamtumbtwihTcpu

tcavaap

anaCdainm(sutmtcaktct

rit

T

2S

A

2S

a

a

v

Levey et al208

s predominant. Traditional methods for measure-ent of proteinuria detect all proteins, but are

ot sensitive enough to detect quantities of albu-in just greater than the normal range. Specific

ssays for albumin are now available that areore sensitive, and levels of urine albumin de-

ected by using these assays, but not assays forrine total protein, have been termed “microalbu-inuria.” However, this term is a misnomer

ecause urine albumin detected by using theseests is not “small.” Microalbuminuria originallyas studied in diabetes, but has now been stud-

ed in patients with other conditions, such asypertension, and in general population studies.able 1 lists threshold levels for definitions oflinically relevant abnormalities in urine totalrotein and albumin according to methods forrine collection.11,13

Methods for measurement of urine total pro-ein are inexpensive and widely used in clini-al medicine, but are limited by the absence of

gold-standard reference method and largeariability among methods. Currently avail-ble methods for urine albumin measurementre less variable than methods for urine total

Table 1. Definitions of To

otal Protein*

Assay No

4-h Excretion �200pot urine protein-creatinine ratio† �200

lbumin

Assay Normal Low

4-h Excretion �10 mg/d 10-29 mg/dpot urine ACR†(sex-specificvalues§)

�10 mg/g 10-29 mg/g

Note: Correspondence among terms is inexact; thereflbumin-creatinine ratio in mg/g to mg/mmol, �0.113.Abbreviations: ACR, albumin-creatinine ratio; NA, not ap*Values vary according to assay.†Average creatinine excretion exceeds 1,000 mg/d and

verage value for this table is 1,000 mg/d.‡Threshold values for children are 500 mg/g in children 6§Sex-specific cutoff values are from a single study.13 Us

alues of prevalence for women than men.Source: National Kidney Foundation.11

rotein measurement, but are not standardized m

cross clinical laboratories. The National Kid-ey Disease Education Program of the NIDDKnd the International Federation of Clinicalhemistry and Laboratory Medicine have un-ertaken a standardization program for urinelbumin measurement and issued the follow-ng recommendations14: urine albumin-creati-ine ratio should be reported as either “mg/mol” or “mg/g,” albumin concentration

miligrams per liter) is difficult to interpret andhould not be reported alone, first morningrine sample has lower biological variabilityhan a random collection, albumin should beeasured in fresh (nonfrozen) urine, and the

erm “urine albumin” should replace “mi-roalbumin.” The implications for clinical tri-ls are to use a central laboratory and, for mostidney diseases, measure urine albumin ratherhan urine total protein unless there is a spe-ific rationale for measuring nonalbumin pro-eins.

Possible alternatives for terms to define catego-ies of albumin-creatinine ratio and correspond-ng quantities would be as follows: normal, lesshan 10 mg/g; low, 10 to 29 mg/g; high, 30 to 300

teinuria and Albuminuria

Elevated (formerly termed clinical proteinuria)

�200 mg/d�200 mg/g‡

igh (formerly termedmicroalbuminuria)

Very High (formerly termedmacroalbuminuria)

mg/d �300 mg/dmg/g (17-250 mg/g for

, 25-355 mg/g foren)

�300 mg/g (�250 mg/g formen, �355 mg/g forwomen)

reshold levels are not consistent. Conversion factor for

.

according to age, sex, and race. For simplicity, assumed

onths of age and 200 mg/g in children 2 years or older.e same cutoff value for men and women leads to greater

tal Pro

rmal

mg/dmg/g‡

H

30-30030-300

menwom

ore, th

plicable

varies

to 24 me of th

g/g; and very high, greater than 300 mg/g

(attkrs

2U

lucacu3pipabswnoomsrcm

emdp

misHfetrcnf

aogc

3

onil

biaataWpbac

Proteinuria as a Surrogate Outcome in CKD 209

Table 1). It would be reasonable to create andditional category above the very high range forhe equivalent of nephrotic syndrome. Differinghreshold values by age, sex, and race based onnown variation in urinary creatinine excretionates have been investigated; however, consen-us on utility has not been reached.15,16

.3. Epidemiological Data forAlbuminuria in thenited States

Variations in methods for measurement andack of standardized definitions for levels ofrine total protein have hampered epidemiologi-al studies. Data from the US National Healthnd Nutrition Examination Surveys show an in-rease in prevalence of high levels of albumin-ria (defined as spot albumin-creatinine ratio �0 mg/g) from 7.1% to 8.2% during the surveyeriods 1988 to 1994 and 1999 to 2004.3 Thencrease was attributed to the older age of theopulation, greater proportion of minority groups,nd greater prevalence of hypertension and dia-etes and greater body mass index. Figure 2hows associations of high levels of albuminuriaith older age and greater prevalence of diag-osed hypertension, diabetes, or both. A wealthf epidemiological data shows that greater levelsf albuminuria, even less than the threshold of 30g/g, are associated with lower estimated GFR,

ubclinical cardiovascular disease, and increasedisk of subsequent kidney failure and cardiovas-ular and all-cause mortality, even after adjust-

Figure 2. Distribution of al-umin-creatinine ratios (ACRs)

n the United States. Numbernd percentage of participantsccording to ACR in the Na-ional Health and Nutrition Ex-mination Survey 1988-2004.ithin each category, the

revalence of diagnosed dia-etes and hypertension, bothnd neither, are shown. Figureourtesy of Josef Coresh.

ent for confounding factors.11,17-19 Some have c

ven proposed that a decrease in albuminuriaight be a surrogate marker for cardiovascular

isease events, as well as for kidney diseaserogression.

3. EVALUATION OF CANDIDATE SURROGATEEND POINTS FOR CLINICAL TRIALS

The growth of biotechnology has created novelethods to measure and monitor disease, includ-

ng the use of biomarkers that can serve asurrogate end points for pivotal clinical trials.owever, important lessons have been learned

rom other fields in which initial beliefs about thefficacy of drug treatment on the basis of trialshat used surrogate outcomes subsequently wereeversed with additional evidence based on clini-al end points.20,21 Thus, it is critical that anyew potential surrogate be rigorously tested be-ore it is accepted.

In this section, we review criteria for drugpproval by the FDA, definitions of types ofutcomes for clinical trials, criteria for surro-acy, and applications of these concepts to clini-al trials of kidney disease progression.

.1. Criteria forDrugApproval

Approval of a drug by the FDA requires dem-nstration of “substantial evidence” of effective-ess consisting of “adequate and well-controllednvestigations.” Phases of drug development areisted in Box 2.22,23 To establish a drug’s effi-

acy, a development program must establish that

aetebopp

3ckctP

Aaeo“Aatotetcio4kfptb

3

fcepaqc

mte

Levey et al210

drug has an effect on a clinically meaningfulnd point.24,25 Alternatively, a surrogate end pointhat reliably predicts a clinically meaningful ben-fit can be used. Thus, although a number ofiomarkers commonly are used in the early phasesf drug development, clinically meaningful endoints or reliable surrogates must be used forhase 3 clinical trials.Established surrogates are widely used in phaseclinical trials in lieu of direct measures of

linical benefit. In certain settings, other biomar-ers also can be used as end points in phase 3linical trials even if there is uncertainty abouthe relation of the outcome to clinical benefit.

Box 2. Phases of Drug Development

Preclinical Studies

Preclinical studies include evaluation of the drug’stoxic and pharmacological effects through in vitro and invivo laboratory animal testing. Genotoxicity screening isperformed, as well as investigations on drug absorptionand metabolism, the toxicity of the drug’s metabolites,and the speed with which the drug and its metabolites areexcreted from the body.

Clinical Studies

The clinical investigation of a previously untested druggenerally is divided into 3 phases. Although in general,the phases are conducted sequentially, they may over-lap.

Phase 1 includes the initial introduction of an investiga-tional new drug into humans. These studies usually areconducted in healthy volunteer participants. These stud-ies are designed to determine the metabolic and pharma-cological actions of the drug in humans, the side effectsassociated with increasing doses, and, if possible, togain early evidence on effectiveness. Phase 1 studiesalso evaluate drug metabolism, structure-activity relation-ships, and the mechanism of action in humans.

Phase 2 includes the early controlled clinical studiesconducted to obtain some preliminary data on the effec-tiveness of the drug for a particular indication or indica-tions in patients with the disease or condition. This phaseof testing also helps determine the common short-termside effects and risks associated with the drug.

Phase 3 studies are intended to gather the additionalinformation about effectiveness and safety that is neededto evaluate the overall benefit-risk relationship of thedrug. Phase 3 studies also provide an adequate basis forextrapolating the results to the general population andtransmitting that information in the physician labeling.

Source: Food and Drug Administration.22,23

assed in the late 1990s, the FDA Modernization t

ct (Subpart H) gave the FDA authority topprove drugs for serious or life-threatening dis-ase with no good available therapy on the basisf an effect of a drug on an end point that isreasonably likely” to predict clinical benefit.26

pproval under this expanded authority requiresphase 4 commitment to verify that the effect on

he surrogate translates into improved clinicalutcomes. An important premise of subpart H ishat approval can be granted for acceleratedntry of a drug to the market, but a confirmatoryrial then is required, showing an effect on thelinical end point. Subpart H has been usednfrequently in drug development, partly becausef the inherent difficulties of completing a phasecommitment after a drug has entered the mar-

et. Although Subpart H may be a viable optionor some development programs, such an ap-roval pathway may not be feasible for long-erm therapies for which the anticipated clinicalenefits may not manifest for many years.

.2. Definitions of Types ofOutcomes

Figure 3 shows the overlapping relationshipsor various potential outcome measures for clini-al trials. In practice, a “clinically meaningful”nd point refers to a direct measure of how aatient feels, functions, or survives. Mortalitynd measures of morbidity, functional status, anduality of life are accepted end points for phase 3linical trials.

A surrogate end point is a laboratory measure-ent or physical sign that is used in therapeutic

rials as a substitute for a clinically meaningfulnd point and is expected to predict the effect of

Figure 3. Definitions and relationships. For explana-ions, see text. Figure courtesy of Tom Greene.

tsEodvmfossft

tncpapkt

iiGibsdkpct

3

tpo

dtagtidascf

beeofstcpcekko

wueumdco

N

Proteinuria as a Surrogate Outcome in CKD 211

he therapy.21 Box 3 lists rationales for the use ofurrogate end points in phase 3 clinical trials.xamples of accepted surrogate end points inther fields include blood pressure and low-ensity lipoprotein cholesterol level for cardio-ascular diseases. In CKD, surrogate end pointsay be useful to establish the efficacy of a drug

or treating early stages of CKD, in which it istherwise difficult to show a benefit. Becauseurrogate end points can reduce the duration,ize, and cost of clinical trials, they also canacilitate the development of drugs intended toreat later stages of CKD.

A biomarker is a characteristic that is objec-ively measured and evaluated as an indicator oformal biological processes, pathogenic pro-esses, or pharmacological responses to a thera-eutic intervention.27 By definition, proteinuriand decreased GFR are biomarkers for CKD andotentially could be surrogate end points foridney failure because in general, they precedehe development of kidney failure.

An intermediate end point is a biomarker thats intermediate in the causal pathway between anntervention and a clinical end point. DecreasedFR also is an intermediate end point because it

s on the causal pathway to kidney failure. Dou-ling of serum creatinine level is accepted as aurrogate end point because it reflects a largeecrease in GFR and predicts the development ofidney failure. As shown in Fig 1, the increase inroteinuria potentially could be a surrogate out-ome for a large decrease in GFR in clinicalrials.

.3. Criteria for Surrogacy

Although there is no formally defined “eviden-iary standard,” the surrogate must be able toredict reliably the effect of a treatment on an

Box 3. Rationale for Using Surrogate Outcomesas End Points in Phase 3 Clinical Trials

Earlier measurementEasier or more frequent measurementGreater measurement precisionLess subject to competing risksLess affected by other treatment modalitiesReduced sample size requirementsFaster decision making

Reprinted with permission of the American Society ofephrology from Stevens et al.8

utcome of interest; this generally means the a

emonstration that the clinical outcomes trackhe marker regardless of how the marker isffected by the various interventions. There are 3eneral lines of evidence that support the accep-ance of a surrogate end point (Box 4).28 Thesenclude biological plausibility, epidemiologicalata (observational studies and clinical data),nd evidence from clinical trials (interventiontudies). The evidence supporting claims forhanges in proteinuria as a surrogate end pointor kidney disease progression is described next.

3.3.1. Biological Plausibility

Biological plausibility refers to the biologicalasis that a putative surrogate will predict theffect of an intervention on an outcome of inter-st. Such evidence usually derives from in vitror animal studies, but sometimes may arise fromavorable clinical responses in extreme circum-tances. Biological plausibility is greatest whenhe marker is a necessary intermediate on theausal pathway of disease. For proteinuria as aotential surrogate for CKD progression, biologi-al plausibility is based on the hypothesizedffects of drugs on established mechanisms ofidney damage that cause proteinuria and lead toidney failure, but it is not usually believed to ben the causal pathway.Figure 4 shows a variety of mechanisms by

hich drug treatment could reduce protein-ria.29-31 The heterogeneity among kidney dis-ases in the biological mechanisms for protein-ria suggest that response to drug treatment alsoay be diverse. Currently, there is a wealth of

ata relating proteinuria in experimental andlinical diabetic kidney disease and for somether glomerular diseases. These relationships

Box 4. Criteria for Surrogacy

Biological plausibility: sometimes intuitive, sometimessupported by animal data or by favorable responses inextreme cases

Epidemiological data: increases (or decreases) in theputative surrogate are correlated with unfavorable (orfavorable) clinical outcomes

Clinical trials: changes in the putative surrogate result-ing from at least 1 type of intervention, and preferablymany types, working by different mechanisms, affectclinical outcomes in a predictable manner that is fullyaccounted for by the effect on the surrogate

Source: Desai et al.28

re less well known for other kidney diseases,

skfpthtormin

dibrotorcatciseefefo

tomta

(

sstusswtdlsapdaTmttmso

Levey et al212

uch as vascular, tubulointerstitial, and cysticidney diseases, and there is little informationor kidney transplant recipients. In addition, whenroteinuria appears, it may cause further damageo the glomerular mesangium or tubules, therebyastening progression of kidney disease.32 Thus,he response to treatment may differ dependingn the stage of kidney disease. Heterogeneity inesponse to treatment suggests that proteinuriaay be an acceptable surrogate for progression

n some kidney diseases and some settings, butot in others.The time course of response of proteinuria to

rugs is critically important for the design andnterpretation of clinical trials. Biological plausi-ility is considerably strengthened if long-termeduction in proteinuria persists after withdrawalf the drug. Of interest, in some kidney diseases,he mechanisms for proposed beneficial effectsf drugs on proteinuria reflect hemodynamicather than structural mechanisms.33,34 In prin-iple, alterations in hemodynamic effects couldmeliorate proteinuria without reducing struc-ural damage. Although reduction in glomerularapillary hydrostatic pressure may be beneficialn kidney disease, reversing or halting progres-ion of the underlying structural damage is anven greater clinical benefit. Evidence for a ben-ficial effect of a drug on structure rather thanunction can be inferred from persistence of theffect on proteinuria after withdrawal of the drugor several months. However, the exact duration

f persistence of the effect after withdrawal of t

he drug is uncertain because beneficial effectsn structure may also diminish over time. Further-ore, drug therapy may be intended for long-

erm use in practice, rather than for withdrawalfter short-term use.

3.3.2. EpidemiologicalDataobservational studies)

Epidemiological data refer to observationaltudies or, in some circumstances, clinical ob-ervations of the relationship of the surrogateo the clinical outcome of interest. For protein-ria as a potential surrogate for CKD progres-ion, epidemiological data refer to the relation-hip of increases or decreases in proteinuria toorsening or improving GFR, respectively. In

he statistical literature, these relationships areescribed as “individual-level associations.” Aarge body of evidence is now available tohow that a greater initial level of proteinuriand an increase over time are independentredictors of both faster GFR decrease andevelopment of kidney failure in patients withwide variety of types of kidney disease.11

hese results appear strong across all seg-ents of the population and all levels of pro-

einuria, including levels of albuminuria lesshan the threshold for the definition of CKD. Inany studies, proteinuria is associated more

trongly with kidney disease outcomes than allther factors tested. Although these associa-

Figure 4. Possible mecha-nisms for proteinuria reduction.Schematic view of a glomerulus(left), nephron (center), andcross-section of nephron seg-ments (right). Abbreviations:[Alb], albumin; ESL, endothelialcell surface layer (glycocalyx);GBM, glomerular basementmembrane; GFR, glomerular fil-tration rate; Qp, plasma flow.29,30

Left panel reproduced with per-mission of the American Physio-logical Society from Haraldssonet al.30 Right panel courtesy ofLutz Slomianka, University ofWestern Australia.

ions provide preliminary support, clinical tri-

attd

fuampttlflepictefsSaproddl(eisfaw

ttptscFkcm

ptoetsttpts2ghhsmt

vpatrtiwc5zgOPosetcawstewcfluit

Proteinuria as a Surrogate Outcome in CKD 213

ls are required to show that the effect ofreatments on change in proteinuria predictshe effect of the same treatment on kidneyisease progression.

3.3.3. Clinical Trials (intervention studies)

The strongest evidence for surrogacy comesrom clinical trials, in which changes in protein-ria resulting from at least 1 type of intervention,nd preferably many types, working by differentechanisms would affect clinical outcomes in a

redictable manner that is fully accounted for byhe effect on the surrogate.28 There are 2 limita-ions to the current body of evidence. First, noarge-scale clinical trial with clinically meaning-ul end points has specifically targeted differentevels of proteinuria as the intervention; how-ver, a number of trials have evaluated change inroteinuria as a secondary end point of othernterventions, such as levels of blood pressure,lasses of antihypertensive agents, glycemic con-rol in diabetes, lipid-lowering therapy, and di-tary modification. Thus, all the evidence derivesrom secondary analysis of interventions de-igned to affect a different pathway of disease.econd, because of the variability in proteinuriamong patients in most studies and the slowrogression of most kidney diseases, analyseselating change in proteinuria to the occurrencef clinically meaningful end points can be con-ucted in only large clinical trials with a longuration of follow-up. Although there are somearge trials of angiotensin-converting enzymeACE) inhibitors and angiotensin receptor block-rs (ARBs), there are few large trials of othernterventions. Given these limitations, as de-cribed next, no method of analysis is likely toulfill strict definitions of the terms “predictable”nd “fully accounted for”; some interpretationill be required.The statistical community has repeatedly cau-

ioned that simply observing a correlation be-ween the putative surrogate and clinical endoint is not sufficient to conclude that the puta-ive surrogate is valid.35 Figure 5 shows possiblecenarios for the relationship among treatment,hange in proteinuria, and clinical end points.8 Inig 5A, proteinuria is on the causal pathway toidney disease progression and the treatmentauses a change in proteinuria; however, treat-

ent also affects progression by a separate causal p

athway. Depending on the size of the effecthrough this second pathway, the treatment effectn progression may differ markedly from itsffect on proteinuria. Alternatively, in Fig 5C,reatment affects both proteinuria and progres-ion, but these effects are unrelated. Statisticalreatment of the relationships between effects ofreatment on the surrogate and the clinical endoint has been reviewed.8 In addition to evalua-ion of the individual-level associations de-cribed previously, statisticians have focused onother general approaches to validation of surro-ate end points in the clinical trials reviewedere. Other statistical treatments, not reviewedere, have considered the problem of validatingurrogate end points by using formal causalodels that incorporate counterfactual variables

o express causal effects.36,37

Prentice Criteria. Formal statistical criteria foralidation of surrogate end points first wereroposed by Prentice.8 In practice, these criteriare evaluated by testing whether an estimate ofhe treatment effect on the clinical end point iseduced to zero after statistical adjustment forhe surrogate. Under certain models for causalnference, fulfillment of this criterion would,ith strong assumptions, show the absence of a

ausal pathway independent of proteinuria (FigB). Unfortunately, reduction in the effect size toero holds only rarely, even for end points thatenerally have been accepted as valid surrogates.ther investigators have proposed relaxing therentice criterion to stipulate that the proportionf the treatment effect (PTE) that remains aftertatistical adjustment for the surrogate should notxceed a designated threshold (for example, 0.5o 0.75). Substantial variation in PTE acrosslinical trials can arise from different analyticpproaches and likely reflects low precision withhich the PTE can be determined within a single

tudy. Another limitation of applying the Pren-ice criteria is undercorrection for measurementrror in the surrogate. Most important, recentork under formal statistical frameworks for

ausal modeling has clarified that even exactulfillment of the Prentice criteria fails to estab-ish the validity of the surrogate except under thenlikely assumption that there are no confound-ng factors that influence both the surrogate andhe clinical end point (Fig 5D). In the case of

roteinuria, it is plausible that other factors influ-

erca

dmf

qroep(t

Teei

Levey et al214

nce both the initial change in proteinuria andate of progression. All these considerations dis-ourage sole reliance on the Prentice criteria tossess proteinuria as a surrogate outcome.Trial-Level Association. The trial-level approach

irectly evaluates the association between treat-ent effects on the surrogate and treatment ef-

Figure 5. Possible scenarios for relationships amongreatment affects both change in proteinuria and clinical enffect is mediated only through reduction in proteinuria. (Cffects are unrelated. (D) Treatment affects both proteinu

nfluence both. Reproduced with permission of the America

ects on clinical end points. This approach re- o

uires joint analysis or meta-analysis of multipleandomized trials. A regression model is devel-ped from previous trials to predict the treatmentffect and confidence interval on the clinical endoint from the treatment effect on the surrogateFig 6).8,38 The most important limitation of therial-level approach is the assumption that previ-

ent, change in proteinuria, and clinical end points. (A)ts, but there are separate causal pathways. (B) Treatmenttment affects both proteinuria and progression, but these

progression, and there are no confounding factors thatety of Nephrology from Stevens et al.8

treatmd poin) Trearia andn Soci

us studies are representative of a new study to

wEtibmaia

riIutsc

acsc

pAatmd

ompscmdauarearpcarsttpntptcpse

Proteinuria as a Surrogate Outcome in CKD 215

hich the surrogate outcome is to be applied.xtrapolation to studies with features substan-

ially different from the original validation stud-es will entail an additional level of uncertaintyeyond that captured by error in the statisticalodel and must rely primarily on biological

rguments. This limitation is critically importantn studies of new populations or new therapeuticgents.

Another limitation of this approach is theequirement for sufficient variation across trialsn the effect of the intervention on the surrogate.f treatment effect on the surrogate is relativelyniform, it will require a large number of studieso relate the magnitude of the effect on theurrogate to the effect on the hard clinical out-

Figure 6. Trial-level analysisf the relationship between treat-ent effect on the clinical endoint and treatment effect on theurrogate. (Top panel) Hypotheti-al relationship between treat-ent effect on end-stage renalisease (ESRD) (vertical axis)nd percentage of change inrine protein (%�UP) (horizontalxis). (Bottom panel) Observedelationship between treatmentffect on survival (vertical axis)nd treatment effect on tumoresponse. In both panels, eachlot character represents a singlelinical trial. In the top panel, therrow on the horizontal axis rep-esents the hypothetical ob-erved effect on �UP in a newrial, and the vertical line reflectshe confidence interval for theredicted effect on ESRD in theew trial. In the bottom panel,he size of plot characters is pro-ortional to the sample size ofhe clinical trial.8,38 Top panelourtesy of Tom Greene. Bottomanel reproduced with permis-ion of the Royal Statistical Soci-ty from Burzykowski et al.39

ome. In principle, including trials of different c

gents or different kidney diseases could in-rease the variation in the treatment effect on theurrogate, but there may be reluctance to makeonclusions from such comparisons.

4. APPLICATION TO SPECIFICCLINICAL CIRCUMSTANCES AND

THERAPEUTIC AGENTS

A large number of studies were reviewed by theresenters for discussion during the conference.lthough the searches were not systematic, the

rticles reviewed permit some general observa-ions. Table S1 (provided as online supplementaryaterial available with this article at www.ajk-

.org) shows the studies grouped according to clini-

al context and level of proteinuria.40-89 Except for

ef

4

dwavfckburoa

hhwcatacfttsdftfrr

fwmmdcAwri

afGlafdiAt

iQlmfmatrdntraim

dsrict

wcp

Levey et al216

arly diabetic kidney disease,40-54 the conferenceocused predominantly on clinical trials.

.1. EarlyDiabetic KidneyDisease

4.1.1. Biological Plausibility

Early structural changes in the glomerulus iniabetic kidney disease include podocyte loss,idening of the glomerular basement membrane,

nd expansion of mesangial and total glomerularolume.90-92 Of these changes, an increase in theraction of mesangial to glomerular volume isonsidered the most specific marker for diabeticidney disease, with a reasonable correlationetween increased mesangial-glomerular vol-me ratio and microalbuminuria. However, theelationship can be variable, with some examplesf structural changes in the absence of increasedlbumin excretion.

4.1.2. Epidemiological Characteristics

The natural history of diabetic kidney diseaseas been well characterized, although it mayave changed since the introduction of treatmentith ACE inhibitors and ARBs. The earliest

linical abnormality is microalbuminuria, usu-lly between 5 and 15 years after the onset ofype 1 or type 2 diabetes. Thereafter, worseninglbuminuria, increasing blood pressure, and de-reasing GFR give rise to symptomatic kidneyailure by approximately 20 years (as shown inhe hypothetical example in Fig 1). Observa-ional studies show a strong and graded relation-hip between level of albuminuria and risk ofecreasing GFR and the development of kidneyailure in patients with type 1 and type 2 diabe-es.46,93-99 In type 2 diabetes, the competing riskrom cardiovascular disease is greater than theisk of kidney failure and may obscure thiselationship.

Data from large cohorts presented at the con-erence showed that 70% to 90% of participantsho develop an estimated GFR less than 60L/min/1.73 m2 had prior macroalbuminuria oricroalbuminuria. In 1 study, this proportion has

ecreased in the more recent era, possibly be-ause of treatment with ACE inhibitors andRBs.43 In all studies, virtually all individualsho developed kidney failure had prior mac-

oalbuminuria. The main limitation of these stud-

es is that few patients beginning with normal

lbumin excretion were followed up long enoughor the development of low levels of estimatedFR. Other limitations are related to variation in

aboratory methods and definitions, such as urinelbumin and serum creatinine assays, definitionsor microalbuminuria, imprecision and bias (un-erestimation of measured GFR) of GFR-estimat-ng equations, and effects of ACE inhibitors andRBs on GFR, independent of their effects on

he underlying kidney disease.

4.1.3. Clinical Trials

A comprehensive review of clinical trials isncluded in the NKF’s Kidney Disease Outcomesuality Initiative (KDOQI) clinical practice guide-

ines on diabetes and CKD.7 Briefly, better glyce-ic control reduces the risk of transitions

rom normal albumin excretion to microalbu-inuria and from microalbuminuria to macro-

lbuminuria. Better blood pressure control andreatment with agents that interfere with theenin-angiotensin system (ACE inhibitors, ARBs,irect renin inhibitors, and aldosterone antago-ists) reduce urinary albumin excretion or slowhe progression from microalbuminuria to mac-oalbuminuria. When studied, withdrawal ofntihypertensive agents was followed by anncrease in albuminuria, but often not to pretreat-ent levels.The main limitation of these studies is that the

uration of follow-up was not sufficient to ob-erve clinical end points. Thus, it is difficult toelate the reduction in albuminuria to reductionn clinical end points, as required by the Prenticeriteria or evaluation of the trial-level associa-ion.

4.1.4. Conclusions

The NKF-KDOQI work group concluded thereas not sufficient evidence for acceptance of

hanges in proteinuria as a surrogate outcome forrogression of early diabetic kidney disease7:

Interventions that reduce albuminuria or delay its in-crease may be promising as potential therapies fordiabetic kidney disease. However, in the opinion of theWork Group, there currently is insufficient evidence toassume that lowering albuminuria levels will necessar-ily lead to improvements in clinical outcomes, such asprogression to CKD stage 5, cardiovascular diseaseevent, or death. Conversely, the failure to reduce albu-minuria does not preclude a beneficial clinical effect on

diabetic kidney disease from a potential intervention.

tKetmndditmlpamtdaPtdsnanpqaqsimcstrc

4

ddt

sttpagcnofubtatd

tarocerdwocp

spdcmt9swdplsouaig

Proteinuria as a Surrogate Outcome in CKD 217

Therefore, to be considered efficacious, potential treat-ments for diabetic kidney disease must demonstratebenefits not only on albuminuria reduction, but also onsuch clinical end points as CKD stage 5, cardiovasculardisease events, or death.

Although there was general agreement duringhe conference with the conclusion of the NKF-DOQI work group, there was support for

xtending this conclusion to acceptance ofransitions from normal albumin excretion toacroalbuminuria or from macroalbuminuria to

ormal albumin excretion as surrogates for theevelopment and remission of diabetic kidneyisease, respectively. Justification for this opinions based on the greater risk of decreasing GFR andhe development of kidney failure in patients withacroalbuminuria than microalbuminuria and the

ower risk of misclassification of response in partici-ants with greater changes in albuminuria. Thispproach is consistent with the guidelines in thatacroalbuminuria in the setting of either type 1 or

ype 2 diabetes is an acceptable criterion for theiagnosis of diabetic kidney disease unless therere features to suggest another form of disease.ersistence of the response after withdrawal of

herapy also would be important to have confi-ence that the drug has had an effect on underlyingtructural damage, rather than a transient hemody-amic effect, although the optimal interval forssessing durability is not established. Unfortu-ately, the development of macroalbuminuria inarticipants with normal albumin excretion re-uires a long follow-up, and complete disappear-nce of macroalbuminuria appears to occur infre-uently. Therefore, adoption of these transitions asurrogate outcomes may not have a meaningfulmpact on clinical trial design for drug develop-ent in early diabetic kidney disease. Possibly,

onsidering intermediate changes on a continuouscale may have more utility. More work is neededo determine the magnitude of change that willeliably predict the effect of a treatment on out-ome.

.2. Nephrotic Syndrome

4.2.1. Biological Plausibility

Nephrotic syndrome55-74 is an infrequent, butramatic, constellation of signs and symptoms,efined as urine total protein excretion greater

han 3.5 g/d, low serum albumin level, increased o

erum cholesterol level, and edema. In 1 study,he corresponding urine albumin level was greaterhan 2.2 g/d.100 Urinary loss of large amounts ofrotein represents a major disruption of the sizend charge-selective filtration barrier of a normallomerulus and is central to the pathophysiologi-al process of other signs and symptoms ofephrotic syndrome.98,101,102 Patient-reportedutcomes, such as discomfort from edema oratigue or side effects of drugs used to treat thenderlying disease or symptoms, are common,ut generally not well quantified. Deep-veinhrombosis, infections, and acute kidney injuryre life-threatening complications from persis-ent nephrotic syndrome. Kidney failure mayevelop over months to years.Initially, corticosteroids were the mainstay of

herapy for adults and children, and there is a largemount of literature documenting some impressiveesponses, particularly in children. With the devel-pment of percutaneous kidney biopsy, pathologi-al and etiologic classification of glomerular dis-ases became increasingly sophisticated. It now isecognized that there are a large variety of kidneyiseases associated with nephrotic syndrome, aide range of responses to corticosteroids andther therapies, and a wide range of rates of de-rease in kidney function according to the cause ofathological types of kidney disease.

4.2.2. EpidemiologicalData

Observational studies show a strong relation-hip between the presence of nephrotic-rangeroteinuria and increased risk of future GFRecrease irrespective of the cause and pathologi-al state of kidney disease, with the exception ofinimal change disease. The risk of progression

o kidney failure during 5 to 10 years is 20% to0% in adults with primary kidney diseases. Theame general pattern is observed for patientsith nephrotic syndrome caused by systemiciseases. Conversely, complete remission of ne-hrotic syndrome (to normal urine total proteinevels) is associated with a greater likelihood oftability of GFR and freedom from developmentf kidney failure. Partial remission (to abnormalrine total protein levels) is not strictly defined,lthough most studies have required a decreasen urine total protein excretion to less than 3.5/d. The relationship of partial remissions to risk

f kidney failure is not clear, in large part be-

craac

tdsistamietrdt

rpdfceraptsotsopmtm

4

mpu

rdbeuecpll

pptTnhmdgtmsdgtotnsu

suoamrydcatcppad

Levey et al218

ause of variability in the definitions for partialemissions and GFR decrease. Both completend partial remissions (to � 3.5 g/d) are associ-ted with improvement in symptoms, biochemi-al abnormalities, and quality of life.

4.2.3. Clinical Trials

There have been a large number of clinicalrials in adults and children with nephrotic syn-rome caused by primary kidney diseases andystemic diseases involving the kidneys, includ-ng a wide variety of interventions and compari-on groups (usually not placebo controlled). Mostrials have focused on decrease in proteinuriand subsequent GFR decrease. Unfortunately,ost trials have been relatively small, and there

s little firm evidence of benefit for most dis-ases. Complete remissions are not common, andhere is not a standardized definition for partialemissions. Altogether, these limitations make itifficult to apply the Prentice criteria or evaluatehe trial-level association.

4.2.4. Conclusions

There was general support at the conference foremission of nephrotic syndrome as a surrogate endoint based on the association of nephrotic syn-rome with important symptoms and high risk ofuture GFR decrease. These symptoms could beonsidered as clinically meaningful outcomes fornd points in phase 3 clinical trials. Completeemission is relatively well defined and uniformlyssociated with good outcomes. Partial remission isoorly defined, and standardization of the defini-ion would clarify the relationship of partial remis-ion to resolution of symptoms and future coursef GFR decrease. Another suggestion to facili-ate testing of new drugs would be to define atandard for the efficacy of corticosteroids andther therapies in inducing remission of ne-hrotic syndrome in patients with more com-on glomerular diseases, such as lupus nephri-

is, to define an acceptable noninferiorityargin for the evaluation of new drugs.

.3. DiseasesWithMild toModerate Proteinuria

4.3.1. Biological Plausibility

Many kidney diseases are characterized byild to moderate proteinuria54,74-89 (urine total

rotein from a lower level of 0.5 to 1.0 g/d to an

pper level of 3.5 g/d [just less than nephrotic f

ange]). Diabetic and various nondiabetic kidneyiseases are considered together in this sectionecause they share certain common features,tiologic and pathological diagnoses are oftenncertain, and they often are grouped together inpidemiological studies and clinical trials. Theorresponding level of albuminuria has not beenrecisely defined, but probably corresponds to aower level greater than 300 mg/g and an upperevel of approximately 2 g/d.

Irrespective of the cause of kidney damage,rogression of kidney disease leads to uniformathophysiological, pathological, and clinical fea-ures, suggesting a “common final pathway.”103

here are numerous theories for the progressiveature of CKD, giving rise to a number ofypothesized treatments to slow progression. Inany experimental models, the course of kidney

isease is reflected in the proportion of scleroticlomeruli and magnitude of proteinuria, and inhese models, drug effects on proteinuria oftenirror drug effects on glomerular pathological

tates and survival. Some of the strongest evi-ence linking proteinuria to kidney disease pro-ression derives from experiments in which pro-einuria is induced by intravenous administrationf high-molecular-weight proteins, leading toransglomerular passage of proteins into the uri-ary space, tubulointerstitial damage, glomerularclerosis, and the development of kidney fail-re.104

4.3.2. EpidemiologicalData

Reliable data about the prevalence and diagno-is of kidney diseases with this level of protein-ria are not readily available. Figure 2 shows thatnly approximately 1.2% of US adults havelbuminuria with albumin greater than 300g/g (which is likely to correspond to this

ange of proteinuria), with a mean age of 60ears, and the vast majority have hypertension,iabetes, or both. In most series, the mostommon diagnoses are diabetic kidney diseasend other glomerular disease, although hyper-ensive nephrosclerosis, tubulointerstitial andystic diseases, and kidney disease in trans-lant recipients occasionally may have urinaryrotein levels this high. These diseases aressociated with a wide range in rates of GFRecrease and risk of the development of kidney

ailure. As discussed, despite this heterogene-

ibprm

rdctitctaftdGdiguacodptchpocaAtnu

eusRnmhda

fPtiftbcoa

AtsiecgmtktAOWaodwbdfrwupCddcdMte

tn

Proteinuria as a Surrogate Outcome in CKD 219

ty, a wealth of epidemiological data relateaseline level of proteinuria and changes inroteinuria during follow-up to subsequentate of GFR decrease and risk of the develop-ent of kidney failure.

4.3.3. Clinical Trials

A large number of clinical trials across a wideange of types of kidney disease and types ofrugs have shown strong relationships betweenhanges in proteinuria and GFR decrease duringreatment.11 However, there have been few stud-es relating treatment effects of drugs or otherreatments on changes in proteinuria and laterhanges in GFR. Most of the largest clinicalrials have used agents that act on the renin-ngiotensin system. These studies have uni-ormly shown a beneficial effect of ACE inhibi-ors and ARBs to decrease blood pressure,ecrease proteinuria, and slow the decrease inFR in both diabetic and nondiabetic kidneyiseases. The beneficial effects appear to bendependent of the antihypertensive effects andreater in patients with greater baseline protein-ria.74,78,80 In several of these studies, the controlrm was treated with a dihydropyridine calciumhannel blocker, which increases proteinuria. In 2f 3-arm studies,75,81 treatment with a dihydropyri-ine calcium channel blocker led to an increase inroteinuria and faster GFR decrease compared withhe �-blocker arm. In 1 study not reviewed at theonference,105 combination therapy with ACE in-ibitors and ARBs led to a greater decrease inroteinuria than either agent alone and a slower ratef GFR decrease, although a recent letter of con-ern raises questions about this study.106 Despitedequate blood pressure control and treatment withCE inhibitors and ARBs, many patients continue

o show GFR decrease and progression to kid-ey failure proportionate to residual protein-ria.74,75,78,80,107,108

Despite the relative uniformity of treatmentffects of ACE inhibitors and ARBs on protein-ria and GFR decrease, a review of selectedtudies in 2006 showed a wide range of PTE.8

eview of additional data at the conference didot suggest a strong trial-level association; theagnitudes of the treatment effects of ACE in-

ibitors and ARBs on clinical end points acrossifferent randomized trials were not predicted

ccurately by the magnitudes of treatment ef- a

ects on proteinuria. However, as noted, theTE can be confounded by other factors, and

he uniformity of treatment effects of ACEnhibitors and ARBs may make it difficult toulfill the trial-level association criterion. Addi-ional analyses are necessary, according toaseline level of proteinuria and magnitude ofhange early in follow-up, and after inclusionf trials of agents other than ACE inhibitorsnd ARBs.

In contrast to the kidney disease studies, thentihypertensive and Lipid-Lowering Treatment

o Prevent Heart Attack Trial (ALLHAT), a largetudy of high-risk hypertensive patients, ACEnhibitors did not reduce the risk of kidney dis-ase progression compared with the diuretic andalcium channel blocker arm, even in the sub-roup with CKD.109-111 Urine protein was noteasured in this study, and it was hypothesized

hat the discrepancy between ALLHAT and theidney disease studies cited previously was at-ributable to lower levels of proteinuria inLLHAT.112 Subsequent to the conference, thengoing Telmisartan Alone and in Combinationith Ramipril Global Endpoint Trial (ONTARGET),large study comparing ACE inhibitors, ARBs,

r both for the treatment of cardiovascularisease, showed a greater decrease in proteinuriaith the combination therapy, but no additionalenefit on cardiovascular disease, and a fasterecrease in GFR and increased risk of kidneyailure and death.113,114 In this study, the geomet-ic mean baseline urine albumin-creatinine ratioas approximately 10 mg/g, with microalbumin-ria and macroalbuminuria in 13% and 4% ofarticipants, respectively. In subgroups withKD, including “overt diabetic nephropathy”efined as macroalbuminuria in the setting ofiabetes or CKD stage 3 or higher, no benefit ofombination therapy was observed for kidneyisease end points individually or combined.oreover, in subgroups without diabetes, hyper-

ension, or CKD, the risk of these kidney diseasend points was increased.

4.3.4. Conclusions

CKD with mild to moderate proteinuria consti-utes a heterogeneous set of diseases. The largeumber of trials and uniformity of responses

cross a range of kidney diseases suggests that a

raetnptttutdpdp5waminc

Aoawwiwtdodm

fisiahctA

(

(

(

(

(

(

(

(

sp

s

Levey et al220

eduction in mild to moderate proteinuria may besuitable surrogate for changes in kidney dis-

ase progression in clinical trials with ACE inhibi-ors and ARBs. However, there is not a sufficientumber of trials of other interventions or otheropulations to extend this conclusion beyondhis narrow setting. For other agents, innovativerial designs to fulfill criteria for FDA approvalhrough Subpart H may allow the use of protein-ria as a surrogate outcome. For example, in therial of the glycosaminoglycan sulodexide iniabetic kidney disease, 2 concurrent trials werelanned; a study of patients with less severeisease (microalbuminuria) with a primary endoint of decrease in albuminuria to normal or by0% of the baseline level that persisted after drugithdrawal, and a study of patients with more

dvanced disease (urine total protein � 900g/d) with a primary end point of decrease in

ncidence of doubling of baseline serum creati-ine level or kidney failure. In both studies, theontrol group received an ACE inhibitor or

Table 2. Conclusions for Use of Change in ProteinuriaClin

Conditions

Sufficient evid1) Preventing development of kidney disease (progressio

normal albuminuria to macroalbuminuria) in diabetes,persistence of effect after drug withdrawal

2) Complete remission of macroalbuminuria in diabetes,persistence of effect after drug withdrawal

3) Complete remission of nephrotic syndrome, with persiwithdrawal of drug

4) Reduction in mild to moderate proteinuria, with persisteffect after drug withdrawal

Surrogate is reasonably li1) Slowing progression from microalbuminuria to macroa

in diabetes, with persistence of effect after drug withdr2) Complete remission of microalbuminuria in diabetes, w

persistence of effect after drug withdrawal3) Partial remission of nephrotic syndrome, with persisten

drug withdrawal4) Reduction in mild to moderate proteinuria, with persist

effect after drug withdrawal

Note: Conclusions do not necessarily represent Food auch as specific kidney diseases, duration of kidney disroteinuria, and duration for persistence of effect after drugAbbreviations: ACE, angiotensin-converting enzyme;

ystem.

*Consider approval under Subpart H with phase 4 commitment.

RB.54 A significant beneficial effect in the trialf patients with less severe disease would enableccelerated entry of the drug into the markethile the trial of patients with more severe diseaseas continuing. One disadvantage of this approach

s that an early “null” effect in the trial in patientsith less severe disease may diminish enthusiasm

o complete the trial in patients with more severeisease, thus overlooking the possible later benefitf an agent that has a beneficial effect on kidneyisease progression without decreasing proteinuriaore than an ACE inhibitor or ARB.Additional studies are necessary to establish more

rmly the validity of changes in proteinuria as aurrogate for kidney disease progression. Random-zed trials comparing treatment regimens targetinghigher versus lower level of proteinuria would beelpful. A recent study by Ruggenenti et al115

omparing kidney disease progression in 2 cohortsreated with a multimodality therapy (includingCE inhibitors and ARBs) to achieve either target

urrogate Outcome for Kidney Disease Progression inials

Drugs

or surrogacyAny

Any

after Any

f ACE inhibitors or ARBs

predict treatment effect*uria Any

Any

r Any

f ACE inhibitors or ARBs in combination; RAS drugsother than ACE inhibitors or ARBs (direct renininhibitors, aldosterone inhibitors)

g Administration opinion. Additional details not specified,glomerular filtration rate, absolute or relative change inawal.angiotensin reuptake blockers; RAS, renin-angiotensin

as a Sical Tr

ence fn fromwith

with

stence

ence o

kely tolbuminawalith

ce afte

ence o

nd Druease,withdr

ARBs,

bp

tCscndd

fs

msntwkpcn

A

E

N

M

A

Proteinuria as a Surrogate Outcome in CKD 221

lood pressure or target urine protein excretionrovides promising preliminary data.

5. SUMMARY OF CONCLUSIONSAND RECOMMENDATIONS

There is an urgent need to facilitate the inves-igation of new drugs to slow the progression ofKD. Because most kidney diseases progress

lowly and are not symptomatic until late in theourse, validation of surrogate markers for kid-ey disease progression would greatly facilitaterug development. The criteria for surrogacy

Table 3. Research Recommendations for Use of CDisease Progres

Population

Exisll kidney disease populations Standardize indice

and excretion raprotein-creatinin

Relate absolute ancontinuous scalechanges that areand onset of kidn

arly diabetes Determine if theremacroalbuminurmicroalbuminuria

ephrotic syndrome Compile data on prdiseases and his

Standardize definitRelate present leveDefine a standard

nephrotic syndrodefining acceptadrugs

ild to moderate proteinuria Compile data fromheterogeneity duduring follow-up

New cll clinical populations Use a central labor

Measure urine protmeasures of albu

Assess persistenceMeasure proteinur

points; relate basRelate measures oConcurrent trials inCompare intervent

Abbreviations: GFR, glomerular filtration rate; SLE, syste

eveloped for other disease appear reasonable c

or evaluation of changes in proteinuria as aurrogate for disease progression in CKD.

Proteinuria is a biomarker. It is not a directeasure of how a patient feels, functions, or

urvives (a clinical end point), and it is notecessarily an intermediate end point on the patho kidney failure. It is unlikely that proteinuriaill be useful as a surrogate in all settings ofidney disease. At the present time, there ap-ears to be sufficient evidence to recommendhanges in proteinuria as a surrogate for kid-ey disease progression in only selected cir-

in Proteinuria as a Surrogate Outcome for KidneyClinical Trials

Recommendations

atateinuria; relate levels of urine total protein to albumin,med urine collections to ratios of concentrations of totallbumin-creatinine in untimed spot urine samplesve changes in urinary albumin and total protein on asequent GFR decline; determine absolute and relativeredictive of GFR decline, doubling of serum creatinine

ure

termediate level between microalbuminuria andis more predictive of subsequent GFR decrease than

is and response to treatment from studies of specificologic patternsemission and partial remission of nephrotic syndromene protein to patient reported outcomestment efficacy for remission and partial remission ofpecific diseases (for example, SLE) for purpose ofinferiority margin for new drugs in comparison to older

g clinical trials for trial level analysis; investigateseline levels of proteinuria, relative or absolute changesecific kidney diseases

trialsnd standardize methods for urine albumin assayher than albumin if there is a specific rationale; relateonalbumin proteins, and total proteinnge in proteinuria after drug withdrawal in clinical trials

nical trials exploring effects of therapy on clinical endevels, absolute and relative changes to clinical end pointsinuria to other markers of kidney damage and biomarkers

and mild study populationsrgeting different urine protein levels

us erythematosus.

hangesion in

ting ds of protes in tie and ad relatito submost pey fail

is an inia that

ognostopathion of rl of uri

for treame in sble non

existine to baand sp

linicalatory aeins otmin, nof cha

ia in clieline lf protesevere

ions ta

mic lup

umstances (Table 2).

ceAdartocttitotbbnceitpt

afcrwFerNtchmtfteiebtFFif

su

iNTS

mh

LDPK

fzlpeam

f

aa

dip

DscpP

c2

h2

io

ti

Levey et al222

Further research is needed to define limitedontexts in which changes in proteinuria can bexpected to predict treatment effect (Table 3).ttention perhaps should focus on particularisease states in which, given what is knownbout the pathogenesis of disease, there is greatereason to believe that proteinuria, even if not onhe causal pathway, is a consistent intermediaten the pathway to kidney failure. Research shouldontinue to explore the consistency of the rela-ionship between changes in proteinuria (direc-ion and magnitude) and clinical outcomes ofnterest. Analyses also should attempt to definehe magnitude of change in proteinuria (relativer absolute) that reliably predicts outcome. Sys-ematic search strategies are necessary to avoidias, and meta-analysis to quantify effects woulde desirable. Additional intervention trials areeeded to establish the relationship betweenhanges in proteinuria and clinically meaningfulnd points. As research on proteinuria continues,t also is important to search for other biomarkerso aid in the diagnosis and prognosis in CKD anderhaps serve as surrogate end points for clinicalrials in the future.

Progress in this field will require collaborationmong investigators to share data from past anduture studies. Based on the experience with thisonference, it appears that substantial data al-eady are available and investigators may beilling to share data from past clinical trials. TheDA is in a unique position to facilitate thesefforts. The FDA should be involved in collabo-ative efforts with academia, industry, and theational Institutes of Health to collect and share

hese data. Some of the trials presented at thisonference are on file at the FDA, and the FDAas data on file from other trials that should beade available to investigators for research on

his topic. However, many of the trials reviewedor this conference were small and not submittedo the FDA for drug approval. There are otherxamples of successful collaborative efforts tomprove the tools used to evaluate the safety andfficacy of drugs.116,117 Governance of the data-ase is an important issue for such a collabora-ion, and to address confidentiality issues, theDA could act as a trusted party to hold the data.or future clinical trials, it will be important to

nclude patients with CKD and assess drug ef-

ects on proteinuria so that we can better under- 2

tand the relationship between changes in protein-ria and clinically meaningful end points.39

ACKNOWLEDGEMENTSRepresentatives from the FDA provided valuable advice

n planning the conference and reviewing the manuscript:orman Stockbridge, MD, PhD, Aliza Thompson, MD,homas Marciniak, MD, Federico Goodsaid, PhD, andhaAvhree Buckman MD, PhD.Staff of the the NKF assisted with meeting logistics andanuscript preparation: Kerry Willis, PhD, Donna Finger-

ut, Maggie Goldstein, and Tom Manley.Members of the planning committee were Andrew S.

evey, MD (co-chair), Thomas Hostetter, MD (co-chair),aniel Cattran, MD, Aaron Friedman, MD, W. Greg Miller,hD, John Sedor, MD, Katherine Tuttle, MD, and Bertramasiske, MD.Support: The NKF received support for the conference

rom Abbott, Amgen, Novartis, Merck, Pfizer, and Gen-yme. Contributions from industry supported travel andodging of conference attendees, including speakers and thelanning committee (some of whom are coauthors). How-ver, the NKF did not provide fees to speakers or coauthors,nd neither the NKF nor industry had involvement in deter-ining the content of the conference or the manuscript.Financial Disclosure: None.

SUPPLEMENTARY MATERIALSItem S1: List of Meeting Participants.Table S1: Compilation of Studies Reviewed for the Con-

erence Report.Note: The supplementary material accompanying this

rticle (doi:10.1053/j.ajkd.2009.04.029) is available at www.jkd.org.

REFERENCES1. Levey AS, Atkins R, Coresh J, et al: Chronic kidney

isease as a global public health problem: Approaches andnitiatives—A position statement from Kidney Disease Im-roving Global Outcomes. Kidney Int 72:247-259, 20072. Levey A, Schoolwerth A, Rios Burrows N, Williams

, Stith K, McClellan W: Comprehensive public healthtrategies for preventing the development, progression, andomplications of chronic kidney disease: Report of an expertanel convened by the Centers of Disease Control andrevention. Am J Kidney Dis 55:522-535, 20093. Coresh J, Selvin E, Stevens LA, et al: Prevalence of

hronic kidney disease in the United States. JAMA 298:2038-047, 20074. Himmelfarb J: Chronic kidney disease and the public

ealth: Gaps in evidence from interventional trials. JAMA97:2630-2633, 20075. Strippoli GF, Craig JC, Schena FP: The number, qual-

ty, and coverage of randomized controlled trials in nephrol-gy. J Am Soc Nephrol 15:411-419, 20046. National Kidney Foundation: K/DOQI Clinical Prac-

ice Guidelines on Hypertension and Antihypertensive Agentsn Chronic Kidney Disease. Am J Kidney Dis 43:S1-S290,

004 (suppl 1)

tD4

fS

asND

nKtp

tcS

LDL

odr

iC

tS

Hi2

ac1

ddCPa

nd9

a1

c

N

a2

tRcD

ftAa

(

pciaga

at

aA

WwJ

tR

b1

pp

MA

AlS

c6

c

s

sW

di

Proteinuria as a Surrogate Outcome in CKD 223

7. National Kidney Foundation: KDOQI Clinical Prac-ice Guidelines and Clinical Practice Recommendations foriabetes and Chronic Kidney Disease. Am J Kidney Dis9:S12-S154, 2007 (suppl 2)8. Stevens LA, Greene T, Levey AS: Surrogate end points

or clinical trials of kidney disease progression. Clin J Amoc Nephrol 1:874-884, 20069. Eknoyan G, Hostetter T, Bakris GL, et al: Proteinuria

nd other markers of chronic kidney disease: A positiontatement of the National Kidney Foundation (NKF) and theational Institute of Diabetes and Digestive and Kidneyiseases (NIDDK). Am J Kidney Dis 42:617-622, 200310. Proteinuria as a Surrogate Outcome in Chronic Kid-

ey Disease: A workshop co-sponsored by the Nationalidney Foundation and the U.S. Food and Drug Administra-

ion. Available at: http://www.kidney.org/professionals/hysicians/proteinuria.cfm. Accessed July 9, 200811. National Kidney Foundation: K/DOQI Clinical Prac-

ice Guidelines for Chronic Kidney Disease: Evaluation,lassification, and stratification. Am J Kidney Dis 39:S1-266, 2002 (suppl 1)12. Stevens LA, Lafayette LA, Perone RD, Levey AS:

aboratory evaluation of kidney disease, in Schrier RW (ed):iseases of the Kidney and Urinary Tract. Philadelphia, PA,ippincott Williams & Wilkins, 2007, pp 299-36613. Warram JH, Gearin G, Laffel L, Krolewski AS: Effect

f duration of type I diabetes on the prevalence of stages ofiabetic nephropathy defined by urinary albumin/creatinineatio. J Am Soc Nephrol 7:930-937, 1996

14. Miller GW, Burns DE, Hortin G, et al: Current issuesn measurement and reporting of urinary albumin excretion.lin Chem 55:24-38, 200915. de Jong PE, Curhan GC: Screening, monitoring, and

reatment of albuminuria: Public health perspectives. J Amoc Nephrol 17:2120-2126, 200616. Forman JP, Fisher ND, Schopick EL, Curhan GC:

igher levels of albuminuria within the normal range predictncident hypertension. J Am Soc Nephrol 19:1983-1988,00817. Hillege HL, Fidler V, Diercks GF, et al: Urinary

lbumin excretion predicts cardiovascular and noncardiovas-ular mortality in general population. Circulation 106:1777-782, 200218. Sarnak MJ, Levey AS, Schoolwerth AC, et al: Kidney

isease as a risk factor for development of cardiovascularisease: A statement from the American Heart Associationouncils on Kidney in Cardiovascular Disease, High Bloodressure Research, Clinical Cardiology, and Epidemiologynd Prevention. Circulation 108:2154-2169, 2003

19. Tuttle KR, Puhlman ME, Cooney SK, Short R: Uri-ary albumin and insulin as predictors of coronary arteryisease: An angiographic study. Am J Kidney Dis 34:918-25, 199920. Psaty BM, Lumley T: Surrogate end points and FDA

pproval: A tale of 2 lipid-altering drugs. JAMA 299:1474-476, 200821. Temple R: Are surrogate markers adequate to assess

ardiovascular disease drugs? JAMA 282:790-795, 199922. US Food and Drug Administration: Investigational

ew Drug Process. Available at: http://www.fda.gov/cder/ c

bout/smallbiz/faq.htm#Introduction.Accessed December 28,00823. FDA-CDER: Drug Review and Related Activities in

he United States. FDA-CDER Forum for Internationalegulatory Authorities. Available at: http://www.fda.gov/der/audiences/iact/forum/200609_kremzner.pdf. Accessedecember 28, 200824. Code of Federal Regulations-21CFR314.125: Re-

usal to approve an application. Food and Drug Adminis-ration, Department of Health and Human Resources.vailable at: http://edocket.access.gpo.gov/cfr_2002/prqtr/21cfr314.125.htm. Accessed June 22, 2009

25. Warner-Lambert Co v Heckler, 1986 787 F.2d 1473d Cir, 1986)

26. Code of Federal Regulations-21CFR314.510: Ap-roval based on a surrogate endpoint or on an effect on alinical endpoint other than survival or irreversible morbid-ty. Food and Drug Administration, Department of Healthnd Human Resources. Available at: http://edocket.access.po.gov/cfr_2002/aprqtr/21cfr314.510.htm. Accessed Janu-ry 15, 2009

27. Biomarkers Definitions Working Group: Biomarkersnd surrogate endpoints: Preferred definitions and concep-ual framework. Clin Pharmacol Ther 69:89-95, 2001

28. Desai M, Stockbridge N, Temple R: Blood pressures an example of a biomarker that functions as a surrogate.APS J 8:E146-E152, 200629. Slomianka L: Tubules of the Nephron. University ofestern Australia. Available at: http://www.lab.anhb.u-a.edu.au/mb140/corepages/urinary/urinary.htm. Accessed

anuary 15, 200930. Haraldsson B, Nystrom J, Deen WM: Properties of

he glomerular barrier and mechanisms of proteinuria. Physiolev 88:451-487, 200831. Hewitt SM, Dear J, Star RA: Discovery of protein

iomarkers for renal diseases. J Am Soc Nephrol 15:1677-689, 200432. Remuzzi G, Benigni A, Remuzzi A: Mechanisms of

rogression and regression of renal lesions of chronic ne-hropathies and diabetes. J Clin Invest 116:288-296, 200633. Bohrer MP, Deen WM, Robertson CR, Brenner BM:echanism of angiotensin II-induced proteinuria in the rat.m J Physiol 233:F13-F21, 197734. Lapinski R, Perico N, Remuzzi A, Sangalli F, Benigni

, Remuzzi G: Angiotensin II modulates glomerular capil-ary permselectivity in rat isolated perfused kidney. J Amoc Nephrol 7:653-660, 199635. Fleming TR, DeMets DL: Surrogate end points in

linical trials: Are we being misled? Ann Intern Med 125:605-13, 199636. Frangakis CE, Rubin DB: Principal stratification in

ausal inference. Biometrics 58:21-29, 200237. Joffe M, Greene T: Related causal framework for

urrogate outcomes. Biometrics 65:530-538, 200938. Burzykowski T, Chmielarczyk W: Predicted and ob-

erved thyroid cancer incidence in Poland after year 1986.iad Lek 57:306-310, 200439. Burzykowski T, Molenberghs G, Buyse M: The vali-

ation of surrogate end points by using data from random-zed clinical trials: A case-study in advanced colorectal

ancer. J R Statist Soc A 167:103-124, 2004

ITttt

Mds

NaI2

Kai

bpit1

bvP

CaD

md5

Dm2

itR2

Mf1

a

rmvD

mdd

Rto

Ll

Ait

Pp2

nNd2

rolt

nn

Lit

Crt

ra1

tt

cwn

aA

Fc5

tt

ra

Levey et al224

40. Molitch M, Rutledge B, Steffes M, Cleary P: Renalnsufficiency in the Absence of Albuminuria in Adults withype 1 Diabetes in the Epidemiology of Diabetes Interven-

ions and Complications (EPIC) Study. Poster presented athe 66th Annual Meeting of the American Diabetes Associa-ion, June 9-13, 2006, Washington, DC

41. Nelson RG, Knowler WC, Pettitt DJ, Saad MF, CharlesA, Bennett PH: Assessment of risk of overt nephropathy in

iabetic patients from albumin excretion in untimed urinepecimens. Arch Intern Med 151:1761-1765, 1991

42. Pavkov ME, Knowler WC, Hanson RL, Bennett PH,elson RG: Predictive power of sequential measures of

lbuminuria for progression to ESRD or death in Pimandians with type 2 diabetes. Am J Kidney Dis 51:759-766,00843. Pavkov ME, Mason CC, Bennett PH, Curtis JM,

nowler WC, Nelson RG: Change in the distribution oflbuminuria according to estimated glomerular filtration raten Pima with type 2 diabetes. Diabetes Care (in press)

44. UK Prospective Diabetes (UKPDS) Group: Intensivelood-glucose control with sulphonylureas or insulin com-ared with conventional treatment and risk of complicationsn patients with type 2 diabetes (UKPDS 33). UK Prospec-ive Diabetes Study (UKPDS) Group. Lancet 352:837-853,99845. UK Prospective Diabetes (UKPDS) Group: Tight

lood pressure control and risk of macrovascular and micro-ascular complications in type 2 diabetes: UKPDS 38. UKrospective Diabetes Study Group. BMJ 317:703-713, 199846. Adler AI, Stevens RJ, Manley SE, Bilous RW, Cull

A, Holman RR: Development and progression of nephrop-thy in type 2 diabetes: The United Kingdom Prospectiveiabetes Study (UKPDS 64). Kidney Int 63:225-232, 200347. Retnakaran R, Cull CA, Thorne KI, Adler AI, Hol-an RR: Risk factors for renal dysfunction in type 2

iabetes: U.K. Prospective Diabetes Study 74. Diabetes5:1832-1839, 200648. Perkins BA, Ficociello LH, Silva KH, Finkelstein

M, Warram JH, Krolewski AS: Regression of microalbu-inuria in type 1 diabetes. N Engl J Med 348:2285-2293,

00349. Perkins BA, Ficociello LH, Roshan B, et al: Decline

n renal function to end stage renal disease begins soon afterhe onset of microalbuminuria (MA) in type 1 diabetes:esults of a 12 year follow-up. J Am Soc Nephrol 18:44A,007 (abstr)50. Perkins BA, Ficociello LH, Ostrander BE, et al:icroalbuminuria and the risk for early progressive renal

unction decline in type 1 diabetes. J Am Soc Nephrol8:1353-1361, 200751. Perkins BA, Nelson RG, Krolewski AS: Cystatin C

nd the risk of death. N Engl J Med 353:842-844, 200552. Mann JF, Gerstein HC, Yi QL, et al: Progression of

enal insufficiency in type 2 diabetes with and withouticroalbuminuria: Results of the Heart Outcomes and Pre-

ention Evaluation (HOPE) randomized study. Am J Kidneyis 42:936-942, 200353. Parving HH, Lehnert H, Brochner-Mortensen J, Go-is R, Andersen S, Arner P: The effect of irbesartan on the

evelopment of diabetic nephropathy in patients with type 2

iabetes. N Engl J Med 345:870-878, 2001 1

54. Lambers Heerspink HJ, Fowler MJ, Volgi J, et al:ationale for and study design of the sulodexide trials in

ype 2 diabetic, hypertensive patients with microalbuminuriar overt nephropathy. Diabet Med 24:1290-1295, 200755. Chen YE, Korbet SM, Katz RS, Schwartz MM,

ewis EJ: Value of a complete or partial remission in severeupus nephritis. Clin J Am Soc Nephrol 3:46-53, 2008

56. Grootscholten C, Ligtenberg G, Hagen EC, et al:zathioprine/methylprednisolone versus cyclophosphamide

n proliferative lupus nephritis. A randomized controlledrial. Kidney Int 70:732-742, 2006

57. Moroni G, Quaglini S, Gallelli B, Banfi G, Messa P,onticelli C: The long-term outcome of 93 patients withroliferative lupus nephritis. Nephrol Dial Transplant 22:531-2539, 200758. Houssiau FA, Vasconcelos C, D’Cruz D, et al: Immu-

osuppressive therapy in lupus nephritis: The Euro-Lupusephritis Trial, a randomized trial of low-dose versus high-ose intravenous cyclophosphamide. Arthritis Rheum 46:121-2131, 200259. Houssiau FA, Vasconcelos C, D’Cruz D, et al: Early

esponse to immunosuppressive therapy predicts good renalutcome in lupus nephritis: Lessons from long-term fol-owup of patients in the Euro-Lupus Nephritis Trial. Arthri-is Rheum 50:3934-3940, 2004

60. Ginzler EM, Dooley MA, Aranow C, et al: Mycophe-olate mofetil or intravenous cyclophosphamide for lupusephritis. N Engl J Med 353:2219-2228, 200561. Chan TM, Tse KC, Tang CS, Mok MY, Li FK:

ong-term study of mycophenolate mofetil as continuousnduction and maintenance treatment for diffuse prolifera-ive lupus nephritis. J Am Soc Nephrol 16:1076-1084, 2005

62. Boumpas DT, Austin HA III, Vaughn EM, et al:ontrolled trial of pulse methylprednisolone versus two

egimens of pulse cyclophosphamide in severe lupus nephri-is. Lancet 340:741-745, 1992

63. Troyanov S, Wall CA, Miller JA, Scholey JW, Catt-an DC: Idiopathic membranous nephropathy: Definitionnd relevance of a partial remission. Kidney Int 66:1199-205, 200464. Ponticelli C, Villa M, Banfi G, et al: Can prolonged

reatment improve the prognosis in adults with focal segmen-al glomerulosclerosis? Am J Kidney Dis 34:618-625, 1999

65. Jha V, Ganguli A, Saha TK, et al: A randomized,ontrolled trial of steroids and cyclophosphamide in adultsith nephrotic syndrome caused by idiopathic membranousephropathy. J Am Soc Nephrol 18:1899-1904, 200766. Cattran DC, Rao P: Long-term outcome in children

nd adults with classic focal segmental glomerulosclerosis.m J Kidney Dis 32:72-79, 199867. Rydel JJ, Korbet SM, Borok RZ, Schwartz MM:

ocal segmental glomerular sclerosis in adults: Presentation,ourse, and response to treatment. Am J Kidney Dis 25:34-542, 199568. Ponticelli C, Villa M, Banfi G, et al: Can prolonged

reatment improve the prognosis in adults with focal segmen-al glomerulosclerosis? Am J Kidney Dis 34:618-625, 1999

69. Troyanov S, Wall CA, Miller JA, Scholey JW, Catt-an DC: Focal and segmental glomerulosclerosis: Definitionnd relevance of a partial remission. J Am Soc Nephrol

6:1061-1068, 2005

it

mc

own3

ifp1

oanE

ttN

pf2

ud6

tt1

pdd

cbA

beSM

oD

seI

i3

pA1

dr1

Sn2

rL

ti2

gd

Pd

Met

Cpp

ulsD

Td1

Vif

Pan2

mwR1

oco

Proteinuria as a Surrogate Outcome in CKD 225

70. Cattran DC, Appel GB, Hebert LA, et al: Cyclospor-ne in patients with steroid-resistant membranous nephropa-hy: A randomized trial. Kidney Int 59:1484-1490, 2001

71. Praga M, Barrio V, Juarez GF, Luno J: Tacrolimusonotherapy in membranous nephropathy: A randomized

ontrolled trial. Kidney Int 71:924-930, 200772. Alexopoulos E, Papagianni A, Tsamelashvili M, Le-

ntsini M, Memmos D: Induction and long-term treatmentith cyclosporine in membranous nephropathy with theephrotic syndrome. Nephrol Dial Transplant 21:3127-132, 200673. Cattran DC, Appel GB, Hebert LA, et al: A random-

zed trial of cyclosporine in patients with steroid-resistantocal segmental glomerulosclerosis. North America Ne-hrotic Syndrome Study Group. Kidney Int 56:2220-2226,99974. GISEN Group: Randomised placebo-controlled trial

f effect of ramipril on decline in glomerular filtration ratend risk of terminal renal failure in proteinuric, non-diabeticephropathy. The GISEN Group (Gruppo Italiano di Studipidemiologici in Nefrologia). Lancet 349:1857-1863, 199775. Lewis EJ, Hunsicker LG, Clarke WR, et al: Renopro-

ective effect of the angiotensin-receptor antagonist irbesar-an in patients with nephropathy due to type 2 diabetes.

Engl J Med 345:851-860, 200176. Shahinfar S, Dickson TZ, Ahmed T, et al: Losartan in

atients with type 2 diabetes and proteinuria: Observationsrom the RENAAL Study. Kidney Int Suppl 82:S64-S67,00277. de Zeeuw D, Remuzzi G, Parving HH, et al: Protein-

ria, a target for renoprotection in patients with type 2iabetic nephropathy: Lessons from RENAAL. Kidney Int5:2309-2320, 200478. Ruggenenti P, Perna A, Gherardi G, et al: Renoprotec-

ive properties of ACE-inhibition in non-diabetic nephropa-hies with non-nephrotic proteinuria. Lancet 354:359-364,99979. Ruggenenti P, Perna A, Loriga G, et al: Blood-

ressure control for renoprotection in patients with non-iabetic chronic renal disease (REIN-2): Multicentre, ran-omised controlled trial. Lancet 365:939-946, 200580. Jafar TH, Schmid CH, Landa M, et al: Angiotensin-

onverting enzyme inhibitors and progression of nondia-etic renal disease. A meta-analysis of patient-level data.nn Intern Med 135:73-87, 200181. Lea J, Greene T, Hebert L, et al: The relationship

etween magnitude of proteinuria reduction and risk ofnd-stage renal disease: Results of the African Americantudy of Kidney Disease and Hypertension. Arch Interned 165:947-953, 200582. Hunsicker LG, Adler S, Caggiula A, et al: Predictors

f the progression of renal disease in the Modification ofiet in Renal Disease Study. Kidney Int 51:1908-1919, 199783. Peterson JC, Adler S, Burkart JM, et al: Blood pres-

ure control, proteinuria, and the progression of renal dis-ase. The Modification of Diet in Renal Disease Study. Annntern Med 123:754-762, 1995

84. Bartosik LP, Lajoie G, Sugar L, Cattran DC: Predict-ng progression in IgA nephropathy. Am J Kidney Dis

8:728-735, 2001 2

85. Descamps-Latscha B, Witko-Sarsat V, et al: Earlyrediction of IgA nephropathy progression: Proteinuria andOPP are strong prognostic markers. Kidney Int 66:1606-612, 200486. Praga M, Gutierrez E, Gonzalez E, Morales E, Hernan-

ez E: Treatment of IgA nephropathy with ACE inhibitors: Aandomized and controlled trial. J Am Soc Nephrol 14:1578-583, 200387. Kobayashi Y, Hiki Y, Kokubo T, Horii A, Tateno S:

teroid therapy during the early stage of progressive IgAephropathy. A 10-year follow-up study. Nephron 72:237-42, 199688. Pozzi C, Bolasco PG, Fogazzi GB, et al: Corticoste-

oids in IgA nephropathy: A randomised controlled trial.ancet 353:883-887, 199989. Yoshikawa N, Honda M, Iijima K, et al: Steroid

reatment for severe childhood IgA nephropathy: A random-zed, controlled trial. Clin J Am Soc Nephrol 1:511-517,00690. Najafian B, Kim Y, Crosson JT, Mauer M: Atubular

lomeruli and glomerulotubular junction abnormalities iniabetic nephropathy. J Am Soc Nephrol 14:908-917, 200391. Pagtalunan ME, Miller PL, Jumping-Eagle S, et al:

odocyte loss and progressive glomerular injury in type IIiabetes. J Clin Invest 99:342-348, 199792. Toyoda M, Najafian B, Kim Y, Caramori ML, Mauer: Podocyte detachment and reduced glomerular capillary

ndothelial fenestration in human type 1 diabetic nephropa-hy. Diabetes 56:2155-2160, 2007

93. Chan JC, Cheung CK, Cheung MY, Swaminathan R,ritchley JA, Cockram CS: Abnormal albuminuria as aredictor of mortality and renal impairment in Chineseatients with NIDDM. Diabetes Care 18:1013-1016, 199594. Ibsen H, Olsen MH, Wachtell K, et al: Does albumin-

ria predict cardiovascular outcomes on treatment withosartan versus atenolol in patients with diabetes, hyperten-ion, and left ventricular hypertrophy? The LIFE Study.iabetes Care 29:595-600, 200695. Mathiesen ER, Ronn B, Storm B, Foght H, Deckert T:

he natural course of microalbuminuria in insulin-depen-ent diabetes: A 10-year prospective study. Diabet Med2:482-487, 199596. Messent JW, Elliott TG, Hill RD, Jarrett RJ, Keen H,

iberti GC: Prognostic significance of microalbuminuria innsulin-dependent diabetes mellitus: A twenty-three yearollow-up study. Kidney Int 41:836-839, 1992

97. Miettinen H, Haffner SM, Lehto S, Ronnemaa T,yorala K, Laakso M: Proteinuria predicts stroke and othertherosclerotic vascular disease events in nondiabetic andon-insulin-dependent diabetic subjects. Stroke 27:2033-039, 199698. Nelson RG, Bennett PH, Beck GJ, et al: Develop-ent and progression of renal disease in Pima Indiansith non-insulin-dependent diabetes mellitus. Diabeticenal Disease Study Group. N Engl J Med 335:1636-642, 199699. Valmadrid CT, Klein R, Moss SE, Klein BE: The risk

f cardiovascular disease mortality associated with mi-roalbuminuria and gross proteinuria in persons with older-nset diabetes mellitus. Arch Intern Med 160:1093-1100,

000

paK2

Tpi

on

pTti3

tC

Ttnc3

E2

elcM

ow8

cabLH

cana(

lb1

i2

ohr5

oJ

rS

cNN

(A

Levey et al226

100. Stoycheff N, Stevens LA, Schmid C, et al: Ne-hrotic syndrome in diabetic kidney disease: An evaluationnd update of the definition [e-pub ahead of print]. Am Jidney Dis doi:10.1053/j.ajkd.2009.04.016, accessed June5, 2009101. Appel GB, Blum CB, Chien S, Kunis CL, Appel AS:

he hyperlipidemia of the nephrotic syndrome. Relation tolasma albumin concentration, oncotic pressure, and viscos-ty. N Engl J Med 312:1544-1548, 1985

102. Myers BD, Nelson RG, Tan M, et al: Progression ofvert nephropathy in non-insulin-dependent diabetes. Kid-ey Int 47:1781-1789, 1995103. Brenner BM, Meyer TW, Hostetter TH: Dietary

rotein intake and the progressive nature of kidney disease:he role of hemodynamically mediated glomerular injury in

he pathogenesis of progressive glomerular sclerosis in ag-ng, renal ablation, and intrinsic renal disease. N Engl J Med07:652-659, 1982104. Zoja C, Benigni A, Remuzzi G: Cellular responses

o protein overload: Key event in renal disease progression.urr Opin Nephrol Hypertens 13:31-37, 2004105. Nakao N, Yoshimura A, Morita H, Takada M, Kayano

, Ideura T: Combination treatment of angiotensin-II recep-or blocker and angiotensin-converting-enzyme inhibitor inon-diabetic renal disease (COOPERATE): A randomisedontrolled trial [Erratum in: Lancet 361:1230, 2003]. Lancet61:117-124, 2003106. Kunz R, Wolbers M, Glass T, Mann JF: The COOP-

RATE trial: A letter of concern. Lancet 371:1575-1576,008107. Appel LJ, Wright JT Jr, Greene T, et al: Long-term

ffects of renin-angiotensin system-blocking therapy and aow blood pressure goal on progression of hypertensivehronic kidney disease in African Americans. Arch Interned 168:832-839, 2008108. Brenner BM, Cooper ME, de Zeeuw D, et al: Effects

f losartan on renal and cardiovascular outcomes in patientsith type 2 diabetes and nephropathy. N Engl J Med 345:861-

69, 2001 c

109. ALLHAT Collaborative Research Group: Major out-omes in high-risk hypertensive patients randomized tongiotensin-converting enzyme inhibitor or calcium channellocker vs diuretic: The Antihypertensive and Lipid-owering Treatment to Prevent Heart Attack Trial (ALL-AT). JAMA 288:2981-2997, 2002110. Rahman M, Pressel S, Davis BR, et al: Renal out-

omes in high-risk hypertensive patients treated with anngiotensin-converting enzyme inhibitor or a calcium chan-el blocker vs a diuretic: A report from the Antihypertensivend Lipid-Lowering Treatment to Prevent Heart Attack TrialALLHAT). Arch Intern Med 165:936-946, 2005

111. Rahman M, Pressel S, Davis BR, et al: Cardiovascu-ar outcomes in high-risk hypertensive patients stratified byaseline glomerular filtration rate. Ann Intern Med 144:172-80, 2006112. Levey AS, Uhlig K: Which antihypertensive agents

n chronic kidney disease? Ann Intern Med 144:213-215,006113. Mann JF, Schmieder RE, McQueen M, et al: Renal

utcomes with telmisartan, ramipril, or both, in people atigh vascular risk (the ONTARGET study): A multicentre,andomised, double-blind, controlled trial. Lancet 372:47-553, 2008114. Yusuf S, Teo KK, Pogue J, et al: Telmisartan, ramipril,

r both in patients at high risk for vascular events. N EnglMed 358:1547-1559, 2008115. Ruggenenti P, Perticucci E, Cravedi P, et al: Role of

emission clinics in the longitudinal treatment of CKD. J Amoc Nephrol 19:1213-1224, 2008116. USFoodandDrugAdministration:FDA,EuropeanMedi-

inesAgency to ConsiderAdditional Test Results WhenAssessingew Drug Safety. Available at: http://www.fda.gov/bbs/topics/EWS/2008/NEW01850.html.Accessed January 31, 2009117. Clinical Data Interchange Standards Consortium

CDISC): Cardiovascular and Tuberculosis Data Standards.vailable at: http://cdisc.org/standards/cardio/index.html. Ac-

essed January 31, 2009