Effects of Empagliflozin on Fluid Overload, Weight and Blood Pressure in Chronic Kidney Disease

Structured Abstract Background Chronic kidney disease (CKD) is associated with fluid excess which can be estimated by bioimpedance spectroscopy. We aimed to assess effects of sodium glucose co-transporter 2 inhibition on bioimpedance-derived “Fluid Overload” and adiposity in a CKD population. Methods EMPA-KIDNEY was a 6609-participant double-blind placebo-controlled trial of empagliflozin 10 mg once daily in patients with CKD at risk of progression. In a 660-participant substudy, bioimpedance measurements were added to the main trial procedures at randomization, 2- and 18-month follow-up visits. The substudy’s primary outcome was the study-average difference in absolute “Fluid Overload” (an estimate of excess extracellular water) analyzed using a mixed-model repeated measures approach. Results The 660 substudy participants were broadly representative of the 6609-participant trial population. Substudy mean baseline absolute “Fluid Overload” was 0.4±1.7 L. Compared to placebo, the overall mean absolute “Fluid Overload” difference among those allocated empagliflozin was -0.24 L (95%CI -0.38, -0.11), with similar-sized differences at 2- and 18-months, and in pre-specified subgroups. Total body water differences comprised between-group differences in extracellular water of -0.49 L (95%CI -0.69, -0.30, including the -0.24 L “Fluid Overload” difference); and a -0.30 L (95%CI -0.57, -0.03) difference in intracellular water. There was no significant effect of empagliflozin on bioimpedance-derived adipose tissue mass (-0.28 [95%CI -1.41, 0.85] kg). The between-group difference in weight was -0.7 kg (95%CI -1.3, -0.1). Conclusions In a broad range of patients with CKD, empagliflozin resulted in a sustained reduction in a bioimpedance-derived estimate of fluid overload, with no statistically significant effect on fat mass.

METHODS: EMPA-KIDNEY was a double-blind placebo-controlled trial of empagliflozin 10 mg once daily in patients with CKD at risk of progression.In a substudy, bioimpedance measurements were added to the main trial procedures at randomization and at 2-and 18-month follow-up visits.The substudy's primary outcome was the study-average difference in absolute "Fluid Overload" (an estimate of excess extracellular water) analyzed using a mixed-model repeated measures approach.

RESULTS:
The 660 substudy participants were broadly representative of the 6609participant trial population.Substudy mean baseline absolute "Fluid Overload" was 0.4±1.7 L. Compared to placebo, the overall mean absolute "Fluid Overload" difference among those allocated empagliflozin was -0.24 L (95%CI -0.38, -0.11), with similar-sized differences at 2-and 18-months, and in pre-specified subgroups.

CONCLUSIONS:
In a broad range of patients with CKD, empagliflozin resulted in a sustained reduction in a bioimpedance-derived estimate of fluid overload, with no statistically significant effect on fat mass.

INTRODUCTION
4][5] These risks increase progressively as estimated glomerular filtration rate (eGFR) decreases, 6 with risk of death from cardiovascular disease exceeding risk of progression to kidney failure for many people with CKD.Fluid excess is common in CKD, especially when heart failure coexists, 7 and can be quantified using bioimpedance spectroscopy. 8Bioimpedance can estimate a number of fluid and adiposity-related parameters, including the excess constituent of total body extracellular water (ECW) over and above what is considered normohydration.
We refer to this parameter as "Fluid Overload" (see Figure 1 and the Supplemental Methods for more details about bioimpedance spectroscopy and a glossary of fluidrelated terms). 9"Fluid Overload" can be used to guide dialysis prescription, 10 and epidemiologically there are positive associations between bioimpedance-measured "Fluid Overload" with cardiovascular outcomes and mortality in patients on dialysis, with non-dialysis CKD, or with heart failure. 8e double-blind international multicenter EMPA-KIDNEY trial demonstrated that, compared to matching placebo, empagliflozin 10 mg once daily reduced the risk of kidney disease progression or cardiovascular death by 28% (95% CI 18-36%) in 6609 patients with CKD at risk of progression. 11A meta-analysis of large placebocontrolled trials extended these findings and showed that in people with CKD, heart failure, or type 2 diabetes at high cardiovascular risk, SGLT2 inhibitors safely reduce the risk of kidney disease progression by about two-fifths and acute kidney injury by about a quarter, with consistent effects irrespective of diabetes status. 12SGLT2 inhibitors also reduce the risk of cardiovascular outcomes, particularly hospitalization for heart failure. 12These absolute cardiovascular benefits are particularly large in patients with pre-existing heart failure, 12,13 but smaller numbers of cardiovascular events in patients with CKD without diabetes and at low levels of eGFR mean effects are less certain in these populations. 11,12The amount of glycosuria induced by SGLT2 inhibition falls with decreasing eGFR and with ambient normoglycemia, 14 so it is reasonable to hypothesize that other effects of SGLT2 inhibitors could also be attenuated in such patients. 11,15To address uncertainty about the effects of SGLT2 inhibitors on fluid status and adiposity in CKD, we embedded a bioimpedance-based substudy within the EMPA-KIDNEY trial. 11The primary aim was to assess the effects of empagliflozin 10 mg once daily versus placebo on fluid status using the bioimpedance-derived parameter of absolute "Fluid Overload" (i.e.estimated excess extracellular water).We also aimed to assess effects on this "Fluid Overload" parameter over time and in different types of patients with CKD.In this report, we also put the substudy findings with respect to empagliflozin's effects on bioimpedance-derived fluid and adiposity parameters in the context of its potentially related effects on weight, blood pressure, glycated hemoglobin and hematocrit (as observed in the full trial cohort).

Substudy design and population
The full methods of the EMPA-KIDNEY trial and the main results have been reported elsewhere (ClinicalTrials.govnumber, NCT03594110; EudraCT number, 2017-002971-24). 11,16Briefly, patients with CKD at risk of progression were identified based on historical and screening local laboratory measurements of an eGFR ≥20 but <45 mL/min/1.73m 2 , or an eGFR ≥45 but <90 mL/min/1.73m 2 with a urinary albumin-to-creatinine ratio (uACR) ≥200 mg/g.This report details the results of an optional substudy conducted in a subset of sites in the United Kingdom (UK) and Germany which added bioimpedance measurements at the randomization, 2-and 18-month follow-up visits to the trial's main protocol-specified procedures (Substudy Protocol Supplement available in the Supplemental Materials).All participants provided written informed consent.Regulatory authorities, as well as ethics committees at each center, approved the trial and the substudy which adhere to the Declaration of Helsinki.

Bioimpedance measurements
Bioimpedance spectroscopy is a tool used in the clinical care of patients requiring dialysis to monitor fluid status. 179][20] The device passes low level electrical current at frequencies of 5-1000 kHz (with results extrapolated from zero to infinity kHz) between electrodes attached to patients' hands and feet. 8All substudy bioimpedance measurements were performed by trained local research Copyright © 2023 The Author(s).Published by Wolters Kluwer Health, Inc. on behalf of the American Society of Nephrology coordinators.Body fluid and adiposity indices were then derived centrally using age, sex, a paired weight measurement, and height data combined with bioimpedance measurements of electrical resistance, and a validated three-compartment model formula using proprietary coefficients. 9,21e primary outcome was based on the bioimpedance-derived estimate of excess extracellular water which we refer to as absolute "Fluid Overload" (sometimes referred to as "overhydration").It is reported in Litres and can have positive or negative values (Figure 1).Its reference range estimated from the 10 th and 90 th centiles of a reference general population distribution is -1.1 L to +1.1 L. 22 "Fluid Overload" can be indexed to extracellular water volume and referred to as percentage relative "Fluid Overload".An absolute value of +1.1 L approximately corresponds to relative "Fluid Overload" of +7%. 23Values above this threshold have been consistently associated with increased risk of death and cardiovascular events, 8 and we refer to it as moderate "Fluid Overload" (>7%, ≤15%) or severe "Fluid Overload" (>15%). 8,23,24Bioimpedance measurements were also used to derive estimates of extracellular and intracellular water volume; lean tissue index (LTI) and fat tissue index (FTI) (see Supplemental Methods for more details).
Local research coordinators were trained to repeat measurements when the BCM device's automated quality score (the Q value) was below 80 (out of 100).Visual inspection of reactance versus resistance plots (known as Cole-Cole plots) were additionally used to assess data quality. 25It was not always possible to obtain a Q value ≥80, so any measurement with a Q value <80 had its Cole-Cole plot assessed independently by two researchers to determine data quality and inclusion in the primary assessment using pre-specified rules blind to treatment allocation (see prespecified Data Analysis Plan for details).Absolute "Fluid Overload" values lower than -5 L were consistently associated with low quality bioimpedance measurement and were considered invalid.

Outcomes
The substudy's pre-specified primary outcome was the effect of empagliflozin versus placebo on mean absolute "Fluid Overload" averaged over time, with effects on relative "Fluid Overload" provided for completeness.It was estimated that at least 382 participants would provide >90% power (at a two-sided P value of 0.05) to detect at least a 0.3 L difference in absolute "Fluid Overload" between treatment groups.
The key secondary outcome was the effect of empagliflozin versus placebo on time to the first event of a cardiovascular composite defined as death from heart failure, heart failure hospitalization, or development of new moderate or severe "Fluid Overload" (in participants without this level of "Fluid Overload" at baseline).The other secondary outcomes were the effects of empagliflozin versus placebo on "Fluid Overload" at the different measurement time points.Tertiary assessments are detailed in the Supplemental Methods and include analyses of the effects of empagliflozin versus placebo on all extracellular (of which "Fluid Overload" is a constituent) and intracellular water.In addition, the effects of empagliflozin versus placebo on total body water (the sum of all extracellular and intracellular water) were assessed as a post-hoc analysis to contextualise effects on "Fluid Overload".
In order for inferences from the bioimpedance substudy to be put in the context of findings from all the available EMPA-KIDNEY data, additional analyses included assessments of the effects of empagliflozin versus placebo on weight, body mass index (BMI), waist-to-hip ratio, glycated hemoglobin, hematocrit and blood pressure (systolic and diastolic) in the full trial cohort.Analyses emphasized results of studyaverage effects including all available measurements from routine trial visit time points (with effects at 2 and 18 months also presented).The full cohort results are emphasized due to greater statistical power and wider generalizability than the substudy.Substudy results were compared to results from the full cohort using standard statistical tests of heterogeneity.Analyses of weight and systolic blood pressure also considered results for the same subgroups as the substudy (plus selfreported raceto explore effects by race in the full trial cohort since the substudy took place in the UK and Germany only).Pre-specified sensitivity analysis for the primary outcome included three analyses assessing any effect of data quality assessments.
Analyses of effects of empagliflozin on diuretic use were included post-hoc.

Statistical analysis
Substudy analyses followed the intention-to-treat principle and required a consenting participant to have provided at least one valid bioimpedance measurement.The primary outcome was pre-specified to be assessed using a mixed model repeated measures (MMRM) approach adjusted for age, sex, prior diabetes, eGFR, and uACR in the categories used in the minimized randomization algorithm. 11The MMRM model also included fixed categorical effects of time (to avoid assuming a linear association between treatment allocation and "Fluid Overload" over time), treatment allocation, treatment-by-time interaction, and continuous effects of baseline (randomization) measurements, and baseline-by-time interaction.The within-person error correlations were assumed to be unstructured.Analyses of the full trial cohort were additionally adjusted for region. 11Effects at each follow-up time point were estimated and used to derive study-average effects (with weights proportional to the amount of time between visits).All between-group differences are reported as empagliflozin minus placebo.To assess effect modification, subgroup-specific treatment effects were estimated by fitting interaction terms in the MMRM models.
The null hypothesis was that the treatment effect is the same across all subgroups.
This was tested by calculating a heterogeneity or trend statistic from subgroupspecific means and standard errors, without correction for multiplicity of testing.
The key secondary outcome and its components were analysed using an adjusted Cox proportional hazards regression using the same covariates in the minimization algorithm (age, sex, prior diabetes, eGFR and uACR) and included the complete substudy population of 660 participants (i.e. it included participants without a valid follow-up bioimpedance measurement who were excluded from MMRM analyses but were at risk of clinical outcomes).Tertiary analyses used the same MMRM approach as described for the primary outcome and assessed effects on ECW, ICW, LTI, FTI, body weight and BMI.Waist and hip circumference measurements were obtained at a single follow-up time point (18 months) and were therefore analysed by analysis of covariance (ANCOVA), adjusted for the baseline value and minimization variables.
Handling of missing data is outlined in the Supplemental Methods.P values for hypothesis testing for outcomes are limited to the primary outcome.P values for testing for any evidence of effect modification between subgroups, and between treatment effect and effects by time are provided.The pre-specified Data Analysis Plan is provided in the Supplemental Materials.Analyses were performed using R

Substudy baseline characteristics and adherence
Between  1).Severity of "Fluid Overload" mirrored established markers of fluid excess: heart failure was twice as common in those with severe "Fluid Overload" compared to the normohydrated group, and NT-proBNP was five-fold higher (Supplemental Table 2).Additionally, participants with "Fluid Overload" were more likely to be older, be male, to have prior diabetes, and have a lower eGFR (Supplemental Table 2).The substudy cohort characteristics were broadly representative of the full trial cohort, 11 although were less racially diverse due to being conducted only in the UK and Germany (Supplemental Table 3).
Substudy adherence to study treatment was consistent with adherence in the full trial population. 11At 12 months of follow-up (the approximate midpoint of the trial), of substudy participants who remained alive, 282/318 (88.7%) in the empagliflozin group and 292/320 (91.3%) in the placebo group reported taking at least 80% of their allocated study treatment.

Effects on bioimpedance-derived parameters
The primary assessment found that the study-average mean absolute "Fluid Overload" was 0.24 L lower in those allocated empagliflozin compared to placebo (absolute difference in means -0.24 L, 95% CI -0.38, -0.11), with similar differences at 2 months (-0.23 L, 95% CI -0.37, -0.08) and 18 months (-0.26L, 95% CI -0.46, -0.06) (Table 2, Figure 2).Findings were robust in sensitivity analyses assessing the effect of data quality assessments (Supplemental Table 4).The effect of empagliflozin on the primary outcome was similar in subgroups by sex, diabetes status, and across the spectrum of NT-proBNP and eGFR studied (p-values for heterogeneity or trend >0.3, Figure 3 & Supplemental Table 5).Neither was there any evidence of heterogeneity in post-hoc exploratory subgroups divided by baseline fluid status (fluid depletion, low-and high-normohydration, moderate and severe Copyright © 2023 The Author(s).Published by Wolters Kluwer Health, Inc. on behalf of the American Society of Nephrology "Fluid Overload"; p=0.71), diuretic use (p=0.07) or urinary albumin-to-creatinine ratio (p=0.33,Supplemental Figure 2).
The number of outcomes was low, limiting statistical power: development of new moderate "Fluid Overload" occurred in 7.8% of substudy participants allocated empagliflozin versus 10.1% allocated placebo; and development of new severe "Fluid Overload" occurred in 2.6% versus 1.3% of empagliflozin and placebo groups, respectively.The tertiary outcome of regression of moderate or severe "Fluid Overload" did not differ significantly between the empagliflozin and placebo groups (54.8% versus 48.6%;Table 3).Heart failure events were also infrequent; there were no deaths due to heart failure in the substudy population.In the full trial cohort, hospitalization for heart failure occurred in 2.7% and 3.2% of participants allocated empagliflozin and placebo, respectively (HR 0.80, 95% CI 0.60-1.06);and findings from the substudy cohort considered in isolation were consistent (empagliflozin 3.3% versus placebo 4.9%; HR 0.67, 95% CI 0.31-1.46;Table 3).

Effects on anthropometry, blood pressure and relevant laboratory values in the full trial cohort
In the full trial cohort, the between-group difference in weight was -0.9 kg (95% CI -1.2, -0.6) (Figure 4, Supplemental Table 8) and the effect of empagliflozin on weight did not vary significantly over time (interaction p value by time=0.47,Supplemental Table 8).In the full cohort, there was no evidence of heterogeneity of the effect of empagliflozin on weight in subgroups by sex, baseline eGFR or diabetes (Figure 4, or in post-hoc analyses by race: Supplemental Figure 3).Waist-to-hip ratio at 18 months was also not significantly different between the empagliflozin versus placebo groups (Supplemental Table 9).The study-average difference in HbA1c in the full cohort was -0.4 mmol/mol (95% CI -0.8, -0.0), with a -0.9 mmol/mol (95% CI -1.6, -0.1) difference in HbA1c in participants with diabetes at randomization and no significant difference in participants without diabetes (0.0 mmol/mol, 95% CI -0.2, 0.2; Supplemental Table 10).The full trial cohort average between-group difference in hematocrit at 18 months post-randomization was 2.3% (95% CI 1.9, 2.7).
The study-average between-group differences in systolic and diastolic blood pressure were -2.6 mmHg (95% CI -3.3, -1.9) and -0.5 mmHg (95% CI -0.9, -0.1), respectively.In the full trial cohort, there was no evidence of heterogeneity of the effect of empagliflozin on systolic blood pressure when subdivided by sex, baseline eGFR, NTpro-BNP (Figure 4) or race (Supplemental Figure 3), but there was some evidence to suggest a larger systolic blood pressure difference in patients with diabetes (Figure 4).Effects on anthropometry, HbA1c, hematocrit and blood pressure in the substudy were approximately consistent with the full trial cohort results (Supplemental Tables 8-11).Supplemental Figure 4 shows the change in weight (relative to baseline) with change in different bioimpedance-indices at the 2 month follow-up visit.

Effects on diuretic use
Among those participants in the full trial cohort who were not taking a loop diuretic at randomization, 159/2453 (6.5%) in the empagliflozin group compared to 212/2409 (8.8%) in the placebo group started such medication during follow-up, representing a 26% lower likelihood of a new loop diuretic prescription among the empagliflozin group (risk ratio 0.74, 95% CI 0.60-0.90).

DISCUSSION
In the EMPA-KIDNEY substudy of 660 patients with CKD, empagliflozin resulted in a sustained reduction in bioimpedance-derived "Fluid Overload" for at least 18 months, irrespective of diabetes status or level of kidney function.Using the threecompartment model, we observed a -0.24 L between-group difference in "Fluid Overload" but no significant differences in normally-hydrated lean or adipose tissue compartments.Fluid volume differences consisted of ~0.8 L less total body water of which ~0.5 L was extracellular and ~0.3 L intracellular water (with the ~0.5 L total extracellular water difference including the -0.24L between-group difference in excess extracellular water referred to as "Fluid Overload").These data raise a hypothesis that an important determinant of the substudy -0.7 kg weight difference was due to effects on fluid status.Along with other mechanisms, 26 this effect may contribute to the cardiovascular benefits of SGLT2 inhibitors.
7][28] In patients with type 2 diabetes without kidney disease, mechanistic trials have reported plasma volume reductions by SGLT2 inhibitors, 30 and raised a hypothesis that SGLT2 inhibitors reduce interstitial volume more than plasma volume. 282][33][34] To the best of our knowledge, the 16-week DECREASE trial provides the only peer-review published randomized evidence on the effects of SGLT2 inhibitors on bioimpedance parameters to date.It found that, in 66 participants with type 2 diabetes -CKD status not reporteddapagliflozin reduced extracellular fluid by ~1 L and systolic blood pressure by ~4 mmHg at 10 days versus placebo. 35EMPA-KIDNEY now substantially extends these previous findings by studying longer term effects (over 18 months) in a much larger number of participants in a placebo-controlled trial.
7][38] Despite this, we found consistent effects on "Fluid Overload" across the eGFR-based subgroups.
Similarly, effects did not vary by baseline fluid status, diuretic use or albuminuria.
These findings are analogous to results from large randomized trials in heart failure populations that included a large proportion of patients with CKD and low eGFR and demonstrated consistent effects of SGLT2 inhibitors on cardiovascular death or hospitalization for heart failure irrespective of sex, diabetes, eGFR or NTpro-BNP at baseline. 13 is also relevant that the effect of empagliflozin on fluid loss in EMPA-KIDNEY was achieved safely.Although estimates of extracellular water reduction reflected loss of extracellular water that is not considered to be in excess by the three-compartment model, there was no increased risk of participant reports of symptomatic dehydration in the full trial or substudy cohorts (Supplemental Table 12), nor any increased risk of acute kidney injury. 39 also report assessments of the effects of empagliflozin on anthropometry, blood pressure, HbA1c and hematocrit for the full trial and substudy cohorts, with the full trial data providing better statistical power to assess for any effect modification between subgroups of participant.The effects of empagliflozin on weight and HbA1c in EMPA-KIDNEY are generally consistent with results from other CKD trials.

Study limitations
EMPA-KIDNEY demonstrated the clear benefits of SGLT2 inhibition on kidney disease progression in a wide range of patients with CKD at risk of progression, including about a one-third reduction in the risk of needing to start kidney replacement therapy. 11This large EMPA-KIDNEY substudy benefits from its sample size, long duration, systematic measurements and randomized double-blind design.
These help ensure between-group differences are unbiased and reliable.The BCM device has some technical limitations.For example, BCM parameters are derived and not direct measurements and based on formulae normalized to healthy reference populations and estimations may be less accurate at extremes of "Fluid Overload" (although extremes of levels were uncommon in the substudy population).
Furthermore, imprecision in fat mass estimates mean the lack of statistical effect on fat mass does not exclude some effect (Supplementary Figure 4).BCM also does not reliably assess subtypes of adiposity (e.g.visceral versus peripheral).Follow-up was affected by COVID-19 restrictions resulting in some missed bioimpedance measurements, and the pre-specified key secondary composite analysis was underpowered due to lower cardiovascular risk in the trial population than was predicted during its design.Nevertheless, this substudy collected sufficient data to provide reliable and clear results for the primary and other continuously measured outcomes.Due to the regions contributing to the substudy, Asian, Black, Mixed and Other races were under-represented, but effects on weight, HbA1c, and blood pressure for the full trial cohort were broadly similar to the substudy results across the studied races, suggesting our conclusions are likely to be generalizable.Lastly, use of other diuretics was determined by local doctors and not controlled by the protocol.We observed more new use of loop diuretics among those allocated to placebo, so the presented estimates of effects on fluid parameters, weight and blood pressure may be slight underestimates of the full effect of empagliflozin.
In summary, the EMPA-KIDNEY bioimpedance substudy found that fluid excess is common in a broad population of patients with CKD at risk of progression, and that empagliflozin resulted in sustained reductions in "Fluid Overload", weight and blood pressure in patients with CKD with and without diabetes, even in patients with low levels of kidney function.

DISCLOSURES
The EMPA-KIDNEY bioimpedance substudy was initiated, designed, conducted Data are presented as mean (SD) or median (Q1-Q3) for continuous variables and n (%) for categorical variables.*Bioimpedance measurements are presented for 644/660 participants with a baseline measurement (missing for 16/660) irrespective of validity for inclusion in the primary analysis.Abbreviations: GFR = glomerular filtration rate; HbA1c = glycated hemoglobin; NTpro-BNP = N-terminal pro-brain-type natriuretic peptide; RAS = renin-angiotensin system.Study-average differences are adjusted for baseline values of the dependent variable (in continuous form) and for any differences in key baseline characteristics (categories of age, sex, diabetes, estimated glomerular filtration rate, urinary albumin-to-creatinine ratio and region) between treatment groups and weighted in proportion to the amount of time between follow-up visits (see Supplemental Methods).Each analysis includes all individuals with at least one follow-up measurement of the outcome variable with mean imputation of missing baseline measurements.For comparison, between-group differences in the substudy cohort were -0.7 (95% CI -1.3, -0.1) kg and -3.3 (-5.5, -1.2) mmHg for weight and systolic blood pressure, respectively.
corporation, or department) and its members Study Group/Organization Name: as follow-up to "Study Group Does your paper include study group(s)?If yes, please provide a list of study group(s) and members that have contributed to or participated in the submitted work in some way.This list may contain either a collaboration of individuals (e.g., investigators) and/or the name of an organization (e.g., a laboratory, educational institution, corporation, or department) and its members" EMPA-KIDNEY Collaborative Group Study Group Members' Names (Members' names should be entered as first name and last name, with individual names separated by commas.If the list of group members' names exceeds 4250 characters, the group members' names will appear in the Supplemental Material but will be indexed in PubMed.) as follow-up to "Study Group Does your paper include study group(s)?If yes, please provide a list of study group(s) and members that have contributed to or participated in the submitted work in some way.This list may contain either a collaboration of individuals (e.g., investigators) and/or the name of an organization (e.g., a laboratory, educational institution, corporation, or department) and its members" Listing the complete EMPA-KIDNEY Collaborative Group exceeds the character count, these individuals are listed within the Supplemental Material file provided.Clinical Trials Registration: My study was a clinical trial.Yes If yes, I have included the registration number here, and the registration number and date of registration are included within the Methods section of my NCT03594110 Copyright © 2023 The Author(s).Published by Wolters Kluwer Health, Inc. on behalf of the American Society of Nephrology manuscript.Registration Number: as follow-up to "Clinical Trials Registration: My study was a clinical trial."Date of Registration: as follow-up to "Clinical Trials Registration: My study was a clinical trial."20-Jul-2018 The ASN Journals require that authors deposit data in a community-approved public repository.If this action has not yet been completed, any data sets can be directly deposited to the Wolters Kluwer/Lippincott Data Repository (powered by FigShare) during the submission process by selecting the content type "Supplemental Data Set."This option is indicated separately below within the section titled, "Repository Name" as "Figshare: Lippincott Data Repository."If your manuscript is accepted for publication, the data set will be made publicly available with reciprocal linking to the published article.Data Sharing You must complete this section.[Select all that apply.]Partial restrictions to the data and/or materials apply Include a Detailed Explanation for Partial Restrictions: as follow-up to "The ASN Journals require that authors deposit data in a community-approved public repository.If this action has not yet been completed, any data sets can be directly deposited to the Wolters Kluwer/Lippincott Data Repository (powered by FigShare) during the submission process by selecting the content type "Supplemental Data Set."This option is indicated separately below within the section titled, "Repository Name" as "Figshare: Lippincott Data Repository."If your manuscript is accepted for publication, the data set will be made publicly available with reciprocal linking to the published article.Data Sharing The complete de-identified patient data set used for presented analyses will be available in due course and the application system to apply to use data will open 6 months after publication.Departmental policy details can be found here: https://www.ndph.ox.ac.uk/data-access.In adherence with the Boehringer Ingelheim Policy on Transparency and Publication of Clinical Study Data, scientific and medical researchers can request access to clinical study data after publication of the primary manuscript and secondary analyses in peer-reviewed journals and regulatory and reimbursement activities are completed, normally within 1 year after the marketing application has been granted by major Regulatory Authorities.Researchers should use the https://vivli.org/link to request access to study data and visit https://www.mystudywindow.com/msw/datasharingfor further information.

Figure 1 :
Figure 1: Relationship of the derived "Fluid Overload" parameter to body

Figure 2 :
Figure 2: Effects of empagliflozin on mean bioimpedance-derived absolute "Fluid Overload" by timeThe value at time 0 is the average across all randomized participants.Follow up means (and their CIs) are derived from a repeated measures mixed model adjusted for baseline values, age, sex, diabetes, eGFR and uACR.Follow-up values are plotted at the median follow-up day in each time window.There was no significant interaction between treatment allocation and time (p=0.11).The study average is the between-group difference (empagliflozin minus placebo) in weighted averages of both time points (see Supplemental Methods).Analyses excluded 40 consenting participants with no valid follow-up measurements.Median (Q1-Q3) follow-up since randomization for empagliflozin vs placebo groups at the 2-month visit: 64 (57-74) vs 64 (57-75) days, Wilcoxon rank sum p = 0.871; and at the 18-month visit: 540 (519-555) vs 532 (505-554) days, p = 0.026.
Copyright © 2023 The Author(s).Published by Wolters Kluwer Health, Inc. on behalf of the American Society of Nephrology 22 nd May 2019 and 14 th April 2021, 668 participants consented to join the substudy.One was excluded due to a metal knee implant and no usable bioimpedance measurement at baseline excluded a further seven, leaving 660 included in analyses (Supplemental Figure1, Supplemental Material).MMRM analyses excluded 40 consenting participants with no valid follow-up bioimpedance measurement (empagliflozin versus placebo: 21 versus 19 respectively; 3 due to death before first follow-up measurement, 28 with no follow-up measurement performed [e.g.due to COVID-19 precluding visits] and 9 due to low data quality).