Part 1:Characterization And Implications Of The Initial Estimated Glomerular Filtration Rate ‘dip’ Upon Sodium-glucose Cotransporter-2 Inhibition With Empagliflozin in The EMPA-REG OUTCOME Trial

Contact: Audrey Hu Whatsapp/hp: 0086 13880143964 Email: audrey.hu@wecistanche.com

Bettina J. Kraus]23, Matthew R. Weir', George L.Bakris5, Michaela Mattheus, David Z.I. Cherney', Naveed Sattar, Hiddo J.L. Heerspink,10,11, lvana Ritter2, Maximilian von Eynatten,13, Bernard Zinman4, Silvio E. Inzucchi', Christoph Wanner2 and Audrey Koitka-Weber'2,16

Treatment with sodium-glucose co-transporter-2 inhibitors induces an initial 3–5 ml/min/1.73 m2 decline in estimated glomerular ﬁltration rate (eGFR). Although considered to be of hemodynamic origin and largely reversible, this ‘eGFR dip’ may cause concern in clinical practice, which highlights the need to better understand its incidence and clinical implications. In this post hoc analysis of the EMPA-REG OUTCOME trial, 6,668 participants randomized to empagliﬂozin 10 mg, 25 mg or placebo with eGFR available at baseline and week four were categorized by initial eGFR change into three groups; over 10% decline (‘eGFR dipper’), over 0 and up to 10% decline (‘eGFR intermediate’), no eGFR decline (‘eGFR non-dipper’). Baseline characteristics of ‘eGFR intermediate’ and ‘eGFR non-dipper’ were generally comparable. An initial ‘eGFR dip’ was observed in 28.3% of empagliﬂozin versus 13.4% of placebo-treated participants; odds ratio 2.7 [95% Conﬁdence Interval 2.3– 3.0]. In multivariate logistic regression, diuretic use and higher KDIGO risk category at baseline were independently predictive of an ‘eGFR dip’ in empagliﬂozin versus placebo. Safety and beneﬁcial treatment effects with empagliﬂozin on cardiovascular and kidney outcomes were consistent across subgroups based on these predictive factors. The initial ‘eGFR dip’ did not have a major impact on the treatment effect of empagliﬂozin on subsequent cardiovascular death, hospitalization for heart failure, and incident or worsening kidney disease. Thus, patients with type 2 diabetes with more advanced kidney disease and/or on diuretic therapy were more likely to experience an ‘eGFR dip’ of over 10% with empagliﬂozin, but the reduction in cardiovascular and kidney outcomes was not relevantly modiﬁed by such ‘eGFR dip.’

cistanche-kidney disease

Cistanche can help with kidney function

treatment effect of empagliﬂozin on subsequent cardiovascular death, hospitalization for heart failure, and incident or worsening kidney disease. Thus, patients with type 2 diabetes with more advanced kidney disease and/or on diuretic therapy were more likely to experience an ‘eGFR dip’ of over 10% with empagliﬂozin, but the reduction in cardiovascular and kidney outcomes was not relevantly modiﬁed by such ‘eGFR dip.’

KEYWORDS: acute kidney injury; cardiovascular disease; diabetes; estimated glomerular filtration rate; sodium-glucose co-transporter-2 inhibition Copyright ª 2020, International Society of Nephrology. Published by Elsevier Inc. This is an open-access article under the CC BY-NC-ND license

cistanche-kidney disease-4(52)

Sodium-glucose co-transporter-2 inhibitors (SGLT2i) were developed as glucose-lowering agents but represent a new treatment option for cardiovascular (CV) and kidney disease in patients with type 2 diabetes (T2D). I'm- improvements in CV and kidney outcomes have been observed across several SGLT2i outcomes trials.1-6 Owing to its renal mechanism of action, SGLT2i is associated with a transient decrease in estimated glomerular ﬁltration rate (eGFR), also termed the ‘eGFR dip,’ shortly after treatment initiation.2,7,8 Although considered largely hemodynamic and reversible, this initial ‘eGFR dip’ has raised concerns in clinical practice, as it may predispose patients to acute kidney injury (AKI). Adverse event (AE) post-marketing reporting on AKI led the U.S. Food and Drug Administration to issue warnings Correspondence: Bettina J. Kraus, University Hospital Würzburg, Depart- ment of Internal Medicine I, Oberdürrbacher Str 6, 97080 Würzburg, Ger- many. E-mail: bk@doktorkraus.de

Received 11 June 2020; revised 30 September 2020; accepted 15 October 2020; published online 10 November 2020 for SGLT2i to be used with caution in patients at risk of AKI.9 However, data from clinical trials1,3,4,6 and large observational cohorts10-13 show a reduced AKI risk with SGLT2i.

An initial ‘eGFR dip’ has been reported with renin-angiotensin-aldosterone system (RAAS) inhibition.14,15 Until recently, RAAS inhibition was the sole nephroprotective treatment for patients with kidney disease, and it remains widely used. Nevertheless, the predictive value of the ‘eGFR dip’ associated with RAAS inhibition on CV and kidney outcomes remains controversial.16-19

The initial ‘eGFR dip’ with SGLT2i on top of RAAS inhibition may limit its clinical use, especially in patients within the lower eGFR range. Therefore, its incidence and clinical implications need to be better understood. We characterized EMPA-REG OUTCOME participants with various degrees of the initial change in eGFR and investigated whether the initial ‘eGFR dip’ observed with empagliﬂozin was inﬂuenced by baseline characteristics and/or had an impact on safety and CV and kidney outcomes.

METHODS

The design and methods of the double-blind, placebo-controlled, multinational EMPA-REG OUTCOME trial (ClinicalTrials.gov identiﬁer: NCT01131676) have been described previously.1 The study population included 7020 treated participants with T2D, established CV disease, and eGFR $30 ml/min per 1.73 m2 (MDRD; modiﬁcation of diet in kidney disease). Participants were assigned at random to receive empagliﬂozin 10 mg or 25 mg or placebo (1:1:1) once daily, in addition to standard care. The median duration of treatment was 2.6 years, and the median observation time was 3.1 years. The primary CV outcome and prespeciﬁed secondary kidney outcome (deﬁned as an incident or worsening nephropathy) have been reported previously.1,2

For this post hoc analysis, 6668 participants who received $1 dose of study drug and had baseline and week 4 eGFR values available were categorized by percent eGFR (equation developed by the Chronic Kidney Disease Epidemiology Collaboration [CKD-EPI]) change from baseline at week 4 into 3 categories: >10% decline (‘eGFR dipper’), >0% to #10% decline (‘eGFR intermediate’), and no decline (‘eGFR non-dipper’). This pragmatic categorization, using clinically relevant, memorable, and easy-to-apply cutoffs, resulted in 3 similar-sized groups in the empagliﬂozin treatment arm. For each category, we described baseline characteristics and eGFR change over time for the duration of treatment and after treatment discontinuation. Additional sensitivity analysis was performed for an eGFR decline of >30% from baseline. For all analyses, we compared placebo and pooled empagliﬂozin (10 mg and 25 mg) groups. Serum creatinine and albumin, and urinary albumin in spot urine, were measured in central laboratories to calculate the urine-albumin-to- creatinine ratio (UACR). Kidney Disease: Improving Global Outcomes (KDIGO) categorization was conducted according to the KDIGO heat map, a 2-dimensional classiﬁcation system, identifying patients with low eGFR and higher UACR levels, who are at elevated risk of adverse kidney and CV outcomes.20 For kidney function over time, we used the CKD-EPI creatinine equation. Mixed-model, repeated-measures analysis was used to evaluate changes in eGFR over time, including glycated hemoglobin (HbA1c) level and eGFR (CKD-EPI) at baseline as linear covariates and geographic region, baseline body mass index (BMI), treatment, visit, visit-by-treatment interaction, the interaction between baseline HbA1c level and visit, and interaction between baseline eGFR and visit as ﬁxed effects. Changes in eGFR per year (i.e., eGFR slope) were obtained using a random intercept/random coefﬁcient model, as described previously.21 The model was applied by each ‘eGFR dipping’ category separately, and only data for participants on treatment were used. Participants who also had eGFR values available after treatment discontinuation were assessed for absolute and percent changes in eGFR between the last value on treatment (LVOT) and ﬁrst value after treatment discontinuation (follow-up).

cistanche-kidney disease

Baseline characteristics of the ‘eGFR intermediate’ and ‘eGFR non-dipper’ groups were generally comparable. Thus, further analyses were performed based on pooled data from these 2 categories to focus on one harmonized ‘eGFR dip’ event, deﬁned as the occurrence of an ‘eGFR dip’ >10% from baseline at week 4. Baseline characteristics were evaluated for the potential predictive effect of such an initial percent ‘eGFR dip’ from baseline at week 4 with empagliﬂozin versus placebo. We used logistic regression with baseline factors, treatment, and interaction of baseline factors with treatment to investigate potential interactions of baseline factors with treatment and hence predictive effects. Following that approach, we applied a multivariate logistic regression model using backward selection and applied a signiﬁcance level of P < 0.05 for the interaction of each baseline factor and treatment to be retained in the model. Baseline factors with a signiﬁcance level of P < 0.1 for interaction with treatment as determined from the ﬁrst step were included in the multivariate model. Relevant predictive factors for an ‘eGFR dip’ event were calculated from the multivariate logistic regression model.

To investigate the association of an ‘eGFR dip’ from baseline at week 4 with CV-related death, hospitalization for heart failure (HHF), or kidney outcomes after week 4 independent of treatment, we combined empagliﬂozin and placebo groups and used a Cox proportional hazards model with factors for the treatment group, baseline variables of age, sex, BMI, HbA1c, eGFR, region, and ‘eGFR dip’ at week 4 and with additional adjustment for baseline values and changes from baseline at week 4 in systolic blood pressure (SBP), diastolic blood pressure (DBP), and fasting plasma glucose.

While categorization by actual ‘eGFR dipping’ after randomization results in loss of randomization and thus does not allow for a comparison of empagliﬂozin treatment effect in ‘eGFR dipper’ versus other categories, we assessed the impact of an ‘eGFR dip’ at week 4 using 2 approaches. We ﬁrst analyzed the effect of Tampa- gliﬂozin on CV and kidney outcomes across relevant predictive baseline factors for an ‘eGFR dip’ event. These analyses were performed using a Cox proportional hazards model with factors for treatment, age, sex, baseline BMI, baseline HbA1c, region, subgroup, and treatment-by-subgroup interaction. In addition, we assessed the ‘eGFR dip’ as a potential mediator for the effect of empagliﬂozin on these outcomes. This analysis was done following the previously described concept of traditional mediation analysis proposed by Baron and Kenny,22 and similarly applied for time to occurrence of CV death.23 We compared the treatment effect of empagliﬂozin on outcomes from week 4 onward from the analysis using the primary model,1 with the treatment effect obtained using a model also adjusted for a >10%

Figure 1| Percentage of participants by 'estimated glomerular filtration rate (eGFR) dipping' category (a), eGFR overtime per 'eGFR dipping' category in empagliflozin-treated participants (b), and mean eGFR at last value on treatment (LVOT) and follow-up according to 'eGFR dipping category in empagliflozin-treated participants(c). (a)Percentage of participants who received at least 1 dose of study drug (empagliflozin vs. placebo) and had eGFR(Chronic Kidney Disease Epidemiology Collaboration [CKD-EPI) values at both baseline and week 4 categorized by percent eGFR change from baseline at week 4 to >10% decline ('eGFR dipper), >0% to ≤10% decline (eGFR intermediate'), or no decline ("eGFR non-dipper). (b)Mixed-model repeated-measures results of eGFR (CKD-EPL) on treatment over time by category of percent change in all participants treated with at least 1 dose of study drug who had baseline and week 4 eGFR values available. The model includes baseline eGFR and baseline glycated hemoglobin (HbA1c) as linear covariates and geographical regions. baseline body mass index category, treatment, visit visit-by-treatment interaction, baseline HbA1c-by-visit interaction, and baseline eGFR-by-visit interaction as fixed effects applied for each'eGFR dipping category. (c) Participants treated with at least 1 dose of study drug who had baseline and (continued)‘eGFR dip’ at week 4. The percent mediation was calculated as follows: mediation % ¼ 100 * ([lnHR - lnHRC]/lnHR), where HR (hazard ratio) denotes a comparison of treatment groups in the model with the treatment group alone and HRC denotes a comparison of treatment groups in the model adjusting for the ‘eGFR dip’. To be considered a mediator, the ‘eGFR dip’ should affect the studied outcome, and the effect of empagliﬂozin on outcome must be reduced in the analysis adjusted for ‘eGFR dip’. Mediation was indicated if the HR for outcome between treatment groups adjusted for ‘eGFR dip’ was closer to unity than the HR from the model with treatment group alone (primary model). Complete mediation would be indicated by an HR of 1.0 in the model adjusted for ‘eGFR dip’. A positive mediation value indicates that the effect of empagliﬂozin on outcome was partially mediated by the ‘eGFR dip,’ and a negative value indicates that the effect of empagliﬂozin was partially diminished by the ‘eGFR dip.’

All analyses were performed at the nominal level of 0.05 without correction for multiple hypothesis testing.

Safety analyses of kidney and overall AEs were descriptive across predictive baseline factors for an ‘eGFR dip’. Kidney AEs represent reporting of the narrow Standardised Medical Dictionary for Regulatory Activities (MedDRA) Query for acute renal failure (ARF) by study investigators, which included the preferred term AKI. Statis- tical analyses were conducted using SAS version 9.4 (SAS Institute, Cary, NC).

cistanche-kidney disease

RESULTS

There was wide interindividual variability in the initial eGFR change among the participants. Whereas the median (IQR) eGFR change from baseline at week 4 was –0.05 (IQR, –4.04 to þ4.27)ml/minper1.73m2 in the placebo-treated participants (–21.4andþ21.4forthe1stand99thpercentiles, respectively), it shifted toward a median reduction of –2.69 (IQR, –7.87, þ1.30) ml/minper1.73m with empagli2 ﬂozin(–24.9and þ17.7for the 1st and 99th percentiles, respectively).

Baseline characteristics of ‘eGFR dipping’ categories

Categorization resulted in 28.3% and 13.4% of ‘eGFR dipper,’ 41.1% and 39.5% of ‘eGFR intermediate,’ and 30.5% and 47.1% of ‘eGFR non-dipper’ participants in the empagliﬂozin and placebo groups, respectively (Figure 1a). Baseline characteristics of ‘eGFR dip’ categories are shown in Table 1. Among empagliﬂozin-treated participants, most baseline characteristics were comparable in the ‘eGFR non-dipper’ and ‘eGFR intermediate’ groups, except for SBP and eGFR, which were slightly higher in the latter group. In contrast, there were some relevant differences in baseline characteristics between empagliﬂozin-treated ‘eGFR dippers’ and ‘eGFR non-dippers’; ‘eGFR dippers’ were older, had a longer-standing history of diabetes, and had higher rates of impaired kidney function and albuminuria and hence a higher KDIGO risk category. Hemoglobin, hematocrit, and albumin levels were slightly lower in ‘eGFR dippers.’ The ‘eGFR dippers’ were more likely to have suboptimal SBP control, even though they were taking more antihypertensive medications. In addition, more ‘eGFR dippers’ were treated with insulin and fewer were treated with metformin, while study participants using insulin had a lower baseline eGFR compared with nonusers, and metformin users had a higher baseline eGFR compared with nonusers.24 Characteristics of the participant subset that experienced an initial eGFR decline of >30% on initiation of empagliﬂozin (1.4%; n ¼ 64) are summarized in Supplementary Table S1. Overall, their baseline characteristics were comparable to those of the EGR dippers,’ but they tended to have more comorbidities and CV risk factors.

eGFR over time and after treatment discontinuation according to the ‘eGFR dipping’ category

The baseline means SD eGFR values in the empagliﬂozin- treated participants were 68.3 18.1, 79.5 22.9, and 72.9 20.6 ml/min per 1.73 m2 for the ‘eGFR dipper,’ ‘eGFR intermediate,’ and ‘eGFR non-dipper’ groups, respectively. The respective mean SD eGFR changes from baseline at week 4 in these 3 groups were – 12.6 5.7, –3.3 2.4, and þ5.4 5.7 ml/min per 1.73 m2. Few participants experienced an eGFR decline >30% at week 4 (1.4% [n ¼ 64] on empagliﬂozin and 0.9% [n ¼ 20] on placebo). Among these, 1 patient on empagliﬂozin and no patients on placebo discontinued the study following week 4.

In participants receiving empagliﬂozin treatment, the mean eGFR remained stable from week 12 onward in all ‘eGFR

dipping’ categories (Figure 1b), as well as in the subset with an initial empagliﬂozin-induced eGFR decline >30% (Supplementary Figure S1). In contrast, mean eGFR levels in placebo recipients declined across all categories (Supplementary Figure S2A): the mean eGFR slope from week 12 to the LVOT was – 1.598 (95% conﬁdence interval [CI], –2.076 to – 1.121) ml/min per 1.73 m2/year in ‘eGFR non- dippers,’ –1.278 (95% CI, – 1.752 to –0.804) ml/min per 1.73 m2/year in the ‘eGFR intermediate’ group, and –0.718 (95% CI, – 1.713 to 0.278) ml/min per 1.73 m2/year in ‘eGFR dippers.’

In participants who also had data available for the follow-up period, mean eGFR increased after empagliﬂozin treatment discontinuation compared with the LVOT (Figure 1c). After placebo discontinuation, eGFR did not change in any category compared with the LVOT (Supplementary Figure S2B).

Figure 1 | (continued) week 4 eGFR values available, as well as an eGFR value after treatment cessation (follow-up). Descriptive statistics for eGFR (CKD-EPI) at baseline, last value on treatment (LVOT), and follow-up. Median percent change at follow-up compared with LVOT: þ2.12% (interquartile range [IQR], –2.98 to þ8.94%) in the ‘eGFR non-dipper’ group, þ2.67% (IQR, –2.27 to þ10.64%) in the ‘eGFR intermediate’ group, and þ6.63% (IQR, –0.08 to þ16.07%) in the ‘eGFR dipper’ group. Predictive baseline factors for an ‘eGFR dip’ with empagliﬂozin treatment

To focus on one harmonized ‘eGFR dip’ event, and because the baseline characteristics of the ‘eGFR intermediate’ and ‘eGFR non-dipper’ groups were generally comparable, further analyses were performed based on pooled data from these 2 categories.

To deﬁne baseline characteristics associated with an initial empagliﬂozin-induced ‘eGFR dip’ >10%, and to identify participants particularly prone to experiencing such an ‘eGFR dip’ with empagliﬂozin, we performed logistic regression analyses on speciﬁed factors. Figure 2a presents the overall odds ratio (OR) for an initial ‘eGFR dip’ with empagliﬂozin treatment

Table 1| Baseline characteristics for empagliﬂozin-treated participants according to ‘eGFR dipping’ categories

Table 1 | (Continued) Baseline characteristics for empagliflflozin-treated participants according to ‘eGFR dipping’ categories

ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; ASA, acetylsalicylic acid; BMI, body mass index; CCB, calcium channel blocker; CHD, cor

onary heart disease; CKD, chronic kidney disease; CV, cardiovascular; DBP, diastolic blood pressure; eGFR, estimated glomerular fifiltration rate; HDL, high-density lipoprotein;

HF, heart failure; KDIGO, Kidney Disease: Improving Global Outcomes; LDL, low-density lipoprotein; PAD, peripheral artery disease; SBP, systolic blood pressure; T2D, type 2

diabetes; TG, triglycerides; UACR, urine albumin-to-creatinine ratio.

aP < 0.05 compared with the non-dipping group.

bP < 0.001 compared with the non-dipping group.

cP < 0.0001 compared with the non-dipping group.

dN value: 1258 for the >10% decline group.

eN values: 1352, 1818, and 1257 for the no, >0 to #10%, and >10% decline groups, respectively.

fN values: 1337, 1813, and 1245 for the no, >0 to #10%, and >10% decline groups, respectively.

gN values: 1356, 1826, and 1259 for the no, >0 to #10%, and >10% decline groups, respectively.

hN values: 1353, 1821, and 1257 for the no, >0 to #10%, and >10% decline groups, respectively.

iN value: 1259 for the >10% decline group.

jN values: 1339, 1807, and 1243 for the no, >0 to #10%, and >10% decline groups, respectively.

kN values: 1339, 1805, and 1242 for the no, >0 to #10%, and >10% decline groups, respectively.

compared with placebo (2.7; 95% CI, 2.3–3.0) and for each baseline factor included in further multivariate models individually. Diuretic use, higher KDIGO risk category, and impaired kidney function at baseline were associated with a further increased odds of an initial ‘eGFR dip’ with empagliﬂozin versus placebo at week 4 (P-value for interaction <0.05) (Figure 2a). Among subcategories of diuretic treatment, the OR for an ‘eGFR dip’ of empagliﬂozin versus placebo was consistent across participants with versus participants without baseline use of potassium-sparing agents and across participants with versus those without the baseline use of low-ceiling diuretics, excluding thiazides (Supplementary Figure S3A). However, treatment with loop diuretics or thiazides at baseline for an ‘eGFR dip’ with empagliﬂozin versus placebo (interaction P ¼ 0.0094 and P ¼ 0.0072, respectively) (Supplementary Figure S3A).

cistanche