University of Groningen
Kidney, Cardiovascular, and Safety Outcomes of Canagliflozin according to Baseline
Albuminuria
Jardine, Meg; Zhou, Zien; Lambers Heerspink, Hiddo J; Hockham, Carinna; Li, Qiang;
Agarwal, Rajiv; Bakris, George L; Cannon, Christopher P; Charytan, David M; Greene, Tom
Published in:
Clinical Journal of the American Society of Nephrology
DOI:
10.2215/CJN.15260920
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from
it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date:
2021
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Jardine, M., Zhou, Z., Lambers Heerspink, H. J., Hockham, C., Li, Q., Agarwal, R., Bakris, G. L., Cannon,
C. P., Charytan, D. M., Greene, T., Levin, A., Li, J-W., Neuen, B. L., Neal, B., Oh, R., Oshima, M., Pollock,
C., Wheeler, D. C., de Zeeuw, D., ... Perkovic, V. (2021). Kidney, Cardiovascular, and Safety Outcomes of
Canagliflozin according to Baseline Albuminuria: A CREDENCE Secondary Analysis.
Clinical Journal of the
American Society of Nephrology
,
16
(3), 384-395. https://doi.org/10.2215/CJN.15260920
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the
author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
The publication may also be distributed here under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license.
More information can be found on the University of Groningen website: https://www.rug.nl/library/open-access/self-archiving-pure/taverne-
amendment.
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the
number of authors shown on this cover page is limited to 10 maximum.
Article
Kidney, Cardiovascular, and Safety Outcomes of
Canagliflozin according to Baseline Albuminuria
A CREDENCE Secondary Analysis
Meg Jardine,
1,2
Zien Zhou,
1,3
Hiddo J. Lambers Heerspink,
1,4
Carinna Hockham ,
1
Qiang Li,
1
Rajiv Agarwal ,
5,6
George L. Bakris,
7
Christopher P. Cannon,
8
David M. Charytan ,
9,10
Tom Greene,
11
Adeera Levin,
12
Jing-Wei Li,
1
Brendon L. Neuen ,
1
Bruce Neal ,
1,13,14
Richard Oh,
15
Megumi Oshima ,
1
Carol Pollock,
16
David C. Wheeler,
1,17
Dick de Zeeuw,
4
Hong Zhang,
18
Bernard Zinman,
19
Kenneth W. Mahaffey,
20
and Vlado Perkovic
1,21
Abstract
Background and objectives The kidney protective effects of renin-angiotensin system inhibitors ar e greater in
people with higher levels of albuminuria at trea tment initiation. Whether this applies to sodium-glucose
cotr ansporter 2 (SGLT2) inhibito rs is uncertain, p articu larly in patients with a very high urine albumin-to-
creatinine ratio (UACR; $3000 mg/g). We examined the association between baseline U ACR and the effects of the
SGLT 2 inhibitor, canagliflozin, on efficacy an d safety outcomes in the Canagliflozin and Renal Endpoints in
Diabetes with Established Nephropathy Cli nical Eval uation (CREDENCE) randomized controlled trial.
Design, setting, pa rticipants, & measurements The study enrolled 4401 participants with type 2 dia betes, an
eGFR of 30 to ,90 ml/min per 1.73 m
2
, and UACR of .300 to 5000 mg/g. Using Cox proportional hazards
regression, we examined the relative and absolute ef fects of canagliflo zin on kidney, cardiovascular, and
safety outcome s ac cording to a bas eline UACR of #1000 mg/g (n52348), .1000 to ,3000 mg/g (n51547),
and $3000 mg/g (n5506). In addition, we examined the effects of canagliflozin on UACR i tself, eGFR slope,
and the intermediate outcomes of glycated hemoglobin, body weight, and systo lic BP.
ResultsOverall, higher UACR was associated with higher rates of kidney and cardiovascular events. Canagliflozin
reduced efficacy outcomes for all UACR levels, with no evidence that relative benefits varied between levels. For
example, canagliflozin red uced the primary composite outcome by 24% (hazard ratio [HR], 0. 76; 95% confidence
interval [95% CI], 0.56 to 1.04) in the lowest UACR subgroup, 28% (HR, 0.72; 95 % CI, 0.56 to 0.93) in the UACR
subgroup .1000 to ,3000 mg/g, and 37% (HR, 0.63; 95% CI, 0.47 to 0.84) in the highest subgroup
(P
heterogeneity
50.55 ). Absolute risk reductions for kidney outcomes were greater in par ticipants with higher
baseline albuminuria; the number ofprimary composite events prevented across ascending UACRcategories were
17 (95% CI, 3 to 38), 45 (95% CI, 9 to 81), and 119 (95% CI, 35 to 202) per 1000 treated participants over 2.6 years
(P
heterogeneity
50.02 ). Rates of kidney-related adverse events were lower with canagliflozin, with a greater relative
reduction in higher UACR categories.
Conclusions Canagliflozin safely reduces kidney and cardiovascular events in people with ty pe 2 diabetes
and severely increa sed albuminuria. In this popu lation, the relative kidney benefits were consisten t over a range
of albuminuria levels, with greatest absolute kidney benefit in those with an UACR $3000 mg/g.
Clinical Trial registry name and registration number: ClinicalTrials.go v: CREDENCE, NCT02065791.
CJASN 16: 384–395, 2021. doi: https://doi.org/10.221 5/CJN.15260920
Introduction
Agen ts that offer kidney protection often have greater
relative benefits in those with higher albuminuria
(or proteinuria) at treatment initiation. For example,
the protective effect of renin-angiotensin system (RAS)
inhibitors on the progression of CKD is modified by
baseline proteinuria in people with (1,2) and without
(3,4) diabetes. Similarly, the relative benefits of tolvaptan,
a vasopressin v2 receptor antagonist, on eGFR decline
in people with autosomal-dominant polycystic kidney
disease increases with baseline albuminuria (5). The
relationship between albuminuria and treatment effects
in these studies was demonstrated in populations with
normal-to-moderate albuminuria. Whether this holds
true at very high levels (including nephrotic-range) and
whether albuminuria modifies the effects of sodium-
glucose cotransporter 2 (SGLT2) inhibitors are unclear.
Before the demonstration of their benefits for kidney
and cardiovascular outcomes (6,7), it was clear that
SGLT2 inhibitors reduced albuminuria in patients with
Due to t he number o f
contribu t ing authors,
the affiliations are
listed at the end of
this article.
Correspondence:
Dr. Carinna Hockham,
Th e George Institute
for Global Health, PO
Box M201, Missenden
Road, Sydney, NSW
2050, Australia. Email:
chockham@
georgeinstitute.org.au
www.cjasn.org Vol 1 6 March, 2021384 Copyright © 2021 by the American So ciety of Nephrology
type 2 diabetes (8). Albuminuria is a strong predictor of
kidney disease progression and cardiovascular disease (9–11)
and, together with eGFR, is the foundation for the Kidney
Disease Improving Global Outcomes (KDIGO) kidney dis-
ease risk classification system (9,12). Consequently, people
with higher albuminuria might derive greater absolute
benefit from albuminuria-lowering treatments.
SGLT2 inhibitors prevent kidney and cardiovascular
events in people with type 2 diabetes (13–16). In the kidney
outcome trial, Canagliflozin and Renal Endpoints in Di-
abetes with Established Ne phropathy Clinical Evaluation
(CREDENCE), cana glifloz in reduced the risk of the pri-
mary composite o utcome of kidney failure, a doubling of
serum creatinine, or kidney or cardio vascular deat h by 30%
(hazard ratio [HR], 0.70 ; 95% confiden ce interval [95% CI],
0.59 to 0.82). Canagliflozin also reduced the risk of
numerous kidney- and cardiov ascula r-specific outcomes
(e.g., kidney failure and the composite outcome of myo-
cardial infarction, stroke, or cardiovascular death).
The CREDENCE trial recruited participants with se-
verely increased albuminuria (urine albumin-to-creatinine
ratio [UACR] .300 to 5000 mg/g), including .500 partic-
ipants with nephrotic-range albuminuria wh o were already
stabi lized on RAS blockade. In this population of people at
high risk of progressive kidney and cardiovascular disease,
we assess ed the r elative and absolute effects of canagli-
flozin according to baseline UACR.
Materials and Methods
CREDENCE was an event-driven, double-blind, ran-
domized controlled trial whose design and mai n res ults
have been previously described (15,17). E thical approval
was o btained at each participating site before commence-
ment of recruitment. The trial was conducted in ac cordance
with the principles of the Declaration of Helsinki.
Participants and Albuminuria Assessment
Trial el igibil ity criteria were designed to recruit partic-
ipants at high risk of progression of diabetic kidney disease.
Participants were aged $30 years, with type 2 diabetes, a
glycated hemoglobin (HbA1c) level of 6.5%–12.0%, an
eGFR of 30–,90 ml/min per 1 .73 m
2
(calculated using
the CKD Epidemiology Collaboration formula) (18), and
aUACRof.300 to 5000 mg/g. Key ex clusion crite ria
included nondiabe tic kidney disease, type 1 diabetes,
and prior trea tment of kidney disease with immuno-
suppression or KRT. Participants were required to have
received treatment with a stable maximum-labeled/tolerated
dose of an angiotensin-converting enzyme inhibitor
or angiotensin receptor blocker for $4 weeks prior to
randomization.
In the CREDENCE study, albuminuria was assessed at
multiple timepoints. First, to be eligible for screening,
participan ts were required to have a UA CR .300 mg /g
(.33.9 m g/mmol) or equivalent, confirmed by a local
laboratory result within 6 months of screening. At screen-
ing, a UACR of .300 to 5000 mg/g (.33.9 to 565.6 mg/mmol)
on cent ral laborator y measure ment was required. Third,
albuminuria was meas ured at randomizat ion t hrough
a central laboratory, but, n otably, this was not used to
judge eligibility. Thus, participants with a UACR ,300 mg/g
by randomization could be enrolled.
Participan ts were randomi zed in a 1:1 ratio to receive
double-blinded oral canagliflozin 10 0 mg or placebo daily,
until initiation of KRT (dialysis or kidney transplantation),
occur rence of diabetic ketoac idosis, pregnancy, receipt of
disallowed therapy , or study end.
Outcomes
The efficacy outcomes for the current analyses were the
same as those reported for the overall trial (15). All efficacy
outcomes and selected safety outcomes were indepen-
dently adjudicated by b linded expert committees.
The primary outcome was the composite of kidney
failure (initiation of dialysis for $30 days, kidney trans-
plantation, or eGFR,15 ml/min per 1.73 m
2
sustained for
$30 days by central laboratory assessment), a doubling of
serum creatinine from baseline (average of randomization
and prer andomization value) sustained for $30 days by
central laboratory assessme nt, or deat h due to kidney o r
cardiovascular di sease. Secondary kidney and cardiovas-
cular efficacy outcomes are shown in Table 1.
Safety outcomes with ten or more events in each
albuminuri a subgroup were examined, and included all
kidney-r elated advers e events combined, AKI, vo lume
depletion, hyperkalemia, urinar y tract infections (UTIs),
and hypoglycemia (Table 1). S imilar to other CREDENCE
secondary analyses, kidney-related adverse events were
define d as those that were coded as primarily involving the
kidney ac cording to Medical Diction ary for Re gulatory
Activities terminology, an d which were investigator-
reported (Table 1).
Percentage and absolute change in albuminuria was
calcu lated as the difference between baseline UACR and
the average of all UACR measurements to week 182. eGFR
slope w as assessed as the acute change in eGFR from
baseline to week 3 (acute slop e), the annualized change in
eGFR from week 3 until treatm ent end (chronic slope), and
the annualized change in eGFR from baseline to week 130
(total slope). Finally, we assessed the intermediate out-
comes of HbA1c, body weight, and sys tolic BP.
Statistical Analyses
The effects of canagliflozin were analyzed according to
the baseline UACR categories #1000, .10 00 to ,3000, and
$3000 mg/g. These broadly equate to a urine protein-to-
creatinine ratio of #1920 mg/g, .1920 to ,5000 mg/g, and
$5000 mg/g, albeit with some uncertainty around these
values (http://ckdpcrisk.org/pcr2acr/; accessed on 24
July 2020) (19). Bas eline UACR was used in this analysi s
as it represents the pretreatment measurement at which
all participants had been treate d with a stable dose of
maximally tolerated angiotensin-converting enzyme inhib-
itor/angiotensin receptor blocker.
For all event-based outcomes , an intention-to-treat ap-
proach was used. Annualized incidence rates were calcu-
lated per 1000 patient-years of follow-up. HRs and 95% CIs
were es timated using a Cox proportional hazards regres -
sion model, stratified by screening eGFR (30 to ,45 , 45 to
,60, and 60 to ,90 ml/min per 1.73 m
2
). The heterogeneity
of relative effects across UACR subgroups was assessed by
CJASN 16: 384–395, March, 2021 Canagliflozin Effects in UACR Subgroups, Jardine et al. 385
including UACR group as a model covariate, together with
an interaction term for treatment and baseline UACR. To
calculate absolute risk differences, the number of partici-
pants with an outcome (per 1000 patie nts over median
follow-up) in those assigned to canaglifl ozin was subtrac-
ted from the corresponding number in those assigned to
placebo. The heterogeneity in absolute risk reduction
was estimated using a fixed-effect meta-analysis, with a
chi-squared test.
To assess the relative effects of canagliflozin on albu-
minuria, HbA1c, body weight, and systolic BP, linear
mixed-effects models for repeated measures were used to
analyze the percentage change in the outcome (log-trans-
formed for UACR) over time. Mode ls were adjusted for
baseline value and trial visit. Tim e was included as a
categorical factor such that the geo metric means were
modeled for each visit separately. The residuals from the
mean model were assumed to have an unstructured co-
variance matrix.
eGFR slope analyses were conducted using on-treatment
eGFR measurements only. This was to avoid the expected
distortions from modifications o f the hemodynamic effect
after cessation of study drug. On-treatment eGFR mea-
sure ments comprised all measurements available between
day 1 an d the last dose of study medicati on (12days),from
a central laborat ory. To estimate the ef fects of canagliflozin
on the mean eGFR slope, a two-slope, mixed-effects, linea r
spline model was fitted to eGFR measurements (with a
knot at week 3, the first post randomization eGFR measure),
with a random interce pt and random slopes for treatment.
Similar to previous CREDENCE subg roup analyses (20),
the mean total slope was computed as a weighted combi-
nation of the acute and chronic slopes. Heterogeneity in the
effect of canaglifl ozin on acute, chronic, and total eGFR
slop e between UACR subgroups was estimated by com-
paring the subgroup-level effects, using a chi-squared test
with two degrees of freedom, accounting for th e standard
error of the mean (SEM) in each subgroup. Change in mean
eGFR according to treatment and baseline UACR is
graphically presented using a restricted maximum likeli-
hood, repeated measures approach.
No adjustment for multiplicity of testing was made.
Importantly, given the post hoc nature of these anal yses, the
pres ented P values should be interpret ed with caution and
have been presented for descriptive rather than inferential
purposes. Analyses were performed usin g SAS version 9.4
(SAS Institute Inc., Cary, NC).
Results
The 4401 participants o f the CREDENCE trial were
followed for a me dian of 2.6 years (range , 0. 0–4.5) (15) .
Overall, median baseline UACR was 927 mg/g (105 mg/
mmol). Around half (53%) had a baseline UACR
#1000 mg/g, 35% had a UACR between .1000 and
,3000 mg/g, and 12% had a UACR $3000 mg/g (Table 2).
Mean age was lower in patient s with a highe r baseline
UACR compared with those with a l ower UACR. The
proportion of women and patients of Asian ethnicity
were higher among participants with a higher UAC R
compared with those with a lower UACR, whereas the
opposite was true for the proportion of White patients.
Table 1. Efficacy and safety end points of the CREDENCE study that are included in this study
Primary end point Compos ite of kidney failure, a doubling of serum creatinine from baseline, or death due to kidney or
cardiovascular disease
Secondary kidney
end points
Compos ite of kidney failure, a doubling of serum creatinine, or kidney death
Compos ite of kidney failure or kidney death
Compos ite of KRT initiation (dialysis for $30 days or kidney transplantation) or kidney deat h
Kidney failure
Doubling of serum creatinine
Compos ite of kidney failure, or kidne y or cardiovascular death
Secondary cardiovascular
end points
Compos ite of cardiovascular death or hospitalization for heart failure
Major adverse cardiovascular events composite of cardiovascular death, nonfatal myocardial infarction,
or nonfatal st roke
Hospitalizati on fo r heart failure
Cardiovascular death
Death from any caus e
Compos ite of cardiovasc ular death, myocardial infarcti on, stroke, or hospitalization for he art failure or
unstable angina
Intermediate outcomes Change in urine albumin-to-creatinine ra tio
Acute, chronic, and total eGFR slope
Change in glyca ted hemoglobin
Change in body weight
Change in systolic BP
Safety outcomes
Kidney-related
safety outcomes
Compos ite of AKI, anuria, azotemia, blood creatinine increased, blood urea increased, eGFR decreased,
nephropathy toxic, renal failure, and renal impairment
AKI
Other safety outcomes Volume depletion
Hyperkalemia
Urinary tract infections
Hypoglycemia
Table adapted fr om efficacy and end points of Canagliflozin and Renal Endpoints in Diabetes with Established Nephropathy Clinical
Evaluation (CREDENCE).
386 CJASN
Higher UACR subgroups were also more likely to be using
glucose-lowering regimens reflecting higher diabetes sever-
ity (i.e., more receiving insulin and fewer receiving a sulpho-
nylurea or biguanide), and more likely to have microvascular
disease, higher BP, and lower eGFR, compared with lower
UACR subgroups (Table 2).
Higher baselin e UACR was consistently asso ciated with
a higher rate of kidney and cardiovascular events in both
the placebo and ca nagliflozin groups (Figures 1 and 2,
Supp lemental Table 1). The rates at which parti cipants with
baseline UACR $3000 mg/g randomized to placebo expe-
rienced at least one event was 201.5 events per 1000 patient-
years for the prima ry outcome, and 126.9 events per
1000 patient-years for the composite of kidney failure or
kidney death. The rates at which this same population
experienced at least one cardiovascular or fatal event was
Table 2. Baseline characteristics of participants in the CREDENCE trial, according to baseline UACR
Characteristic
Baseline UACR, mg/g
#1000 .1000 to ,3000 $3000
n52348 (53%) n51547 (35%) n5506 (12%)
Age, yr, mean (SD) 64 (9) 63 (9) 60 (9)
Women, n (%) 756 (32%) 534 (35%) 204 (40%)
Race, n (%)
Asian 422 (18%) 331 (21%) 124 (25%)
Black 138 (6.0%) 65 (4%) 21 (4%)
White 1596 (68%) 1018 (66%) 317 (63%)
Other
a
192 (8%) 133 (9%) 44 (9%)
Region, n (%)
North America 666 (28%) 381 (25%) 135 (27%)
Central/South America 523 (22%) 314 (20%) 104 (21%)
Europe 457 (20%) 328 (21%) 79 (16%)
Rest of the world 702 (30%) 524 (34%) 188 (37%)
Current smoker, n (%) 325 (14%) 242 (16%) 72 (14%)
History of hypertension, n (%) 2273 (97%) 1501 (97%) 486 (96%)
History of heart fa ilure, n (%) 319 (14%) 247 (16%) 86 (17%)
Duration of diabetes, yr, mean (SD) 16 (9) 16 (9) 15 (8)
Drug therapy, n (%)
Insulin 1463 (62%) 1057 (68%) 364 (72%)
Sulfonylurea 727 (31%) 427 (28%) 114 (23%)
Biguanide 1433 (61%) 865 (56%) 247 (49%)
GLP-1 receptor agonist 108 (5%) 56 (5%) 19 (4%)
DPP-4 inhibitor 419 (18%) 267 (17%) 65 (13%)
Statin 1628 (69%) 1077 (70%) 331 (65%)
Antithrombotic age nt
b
1448 (62%) 915 (59%) 261 (52%)
RAS inhibitor 2345 (100%) 1545 (100%) 505 (100%)
b-blocker 938 (340%) 631 (41%) 201 (40%)
Diuretic 913 (39%) 708 (46%) 261 (52%)
Microvascular disease history, n (%)
Neuropa thy 1106 (47%) 765 (50%) 276 (55%)
Retinopathy 913 (39%) 708 (46%) 261 (52%)
History of cardiovascular disease, n (%) 1198 (51%) 758 (49%) 264 (52%)
Body mass index, kg/m
2
,mean(SD)
c
31.4 (6.1) 31.3 (6.1) 31.2 (6.6)
Systolic BP, mm Hg, mean (SD) 138 (15) 142 (16) 143 (16)
Diastolic BP, mm Hg, mean (SD) 77 (9) 79 (9) 80 (9)
Glycated hemoglobin, %, mean (SD) 8.3 (1.3) 8.2 (1.3) 8.4 (1.5)
Triglycerides, mg/d l, mean (SD)
c
186 (133) 204 (159) 221 (142)
Cholesterol, mg/dl, mean (SD)
c
174 (46) 182 (50.3) 201 (58)
HDL cholesterol, mg/dl, mean (SD)
c
43 (12) 46 (12) 46 (16)
LDL cholesterol, mg/dl, mean (SD)
c
93 (39) 97 (39) 112 (50)
Ratio of LDL to HDL, mean (SD)
c
2.2 (1.0) 2.3 (1.1) 2.6 (1.3)
eGFR, ml/min per 1.73 m
2
, mean (SD) 58 (18) 55 (18) 53 (18)
UACR, mg/g, median (IQR)
d
489 (321–693) 1630 (1254–21 67) 3893 (3408–4765)
UACR, mg/mmol, median (IQR)
d
55 (36–78) 184 (142–245) 440 (385–539)
CREDENCE, Canagliflozin and Renal Endpoints in Diabetes with Established Nephropathy Clinical Evaluation; UACR, urine albumin-
to-creatinine ratio; GLP-1, glucagon-like peptide-1; DPP-4, dipeptidyl peptidase-4; RAS, renin-angiotensin system; IQR,
interquartile range.
a
Includ es American Indian or Alaska Native, Native Hawaiian or other Pacific Islande r, multiple, othe r, unknown, and not reported.
b
Includes anticoagulation and antiplatelet agents, including aspirin.
c
#1% missing data.
d
Eligibility was on thebasis ofa screening UACRof .300 to 5000mg/g (33.9–565.6 mg/mmol). Bybasel ine, 527 participants had aUACR
,300 mg/g, including 31 with normoalbum inuria (UACR, 30 mg/g, or ,3 mg/mmol) and 496 with microalbuminuria (UACR
30–300 mg/g, or 3–30 mg/mmol) (15).
CJASN 16: 384–395, March, 2021 Canagliflozin Effects in UACR Subgroups, Jardine et al. 387
