Renal effects of a sodium‐glucose cotransporter 2 inhibitor, tofogliflozin, in relation to sodium intake and glycaemic status

Abstract Aims Little is known about whether sodium intake is associated with the clinical effects of SGLT2 inhibitors (SGLT2is); however, SGLT2is may increase urinary sodium excretion. Thus, we investigated the impact of daily sodium intake on the estimated glomerular filtration rate (eGFR) via an SGLT2i, tofogliflozin (TOFO), in patients with type 2 diabetes (T2D). Methods Individual‐level data on 775 T2D patients in TOFO Phase 3 trials were analysed. Adjusted changes in variables during 52 weeks of TOFO therapy were compared according to basal daily salt intake (DSI), which was measured based on estimated daily urinary sodium excretion using the Tanaka formula. Multivariable analysis was used to investigate the impact of basal DSI on changes in eGFR at Weeks 4 and 52. Results Sixty‐six percent of participants were men; mean age, HbA1c, body mass index, eGFRMDRD and median DSI were 58.5 years, 8.0%, 25.6 kg/m2, 83.9 mL/min/1.73 m2 and 9.3 g/d, respectively. In all participants, eGFRMDRD sharply dipped during Week 4, and gradually increased by Week 52, showing a significant increase overall from baseline to Week 52. Multivariable analysis showed that basal DSI and HbA1c levels were independently correlated with eGFRMDRD changes at Weeks 4 and 52. Additionally, lower baseline HbA1c and DSI levels were significantly correlated with a greater increase in eGFRMDRD at Week 52. Conclusions Dietary salt intake, in addition to glycaemic control, correlates with changed eGFRMDRD via TOFO. Thus, an appropriate dietary approach to therapy should be considered before treatment of T2D patients with an SGLT2i.


| INTRODUCTION
Type 2 diabetes (T2D) is the leading cause of renal and cardiovascular disease in the world. 1 Hyperglycaemia as the result of diabetes is thought to be exaggerated by hyper-reabsorption of renal glucose in the proximal tubule. However, the etiology of the increased glycosuria threshold in T2D is still unclear. One of the mechanisms of hyperreabsorption of glucose is an increased expression of sodium/glucose cotransporter-2 (SGLT2) in patients with T2D and diabetic nephropathy. 2,3 Increased SGLT2 induces increased proximal tubular reabsorption of, not only glucose, but also sodium, both of which are SGLT2 substrates. This reduces sodium chloride and fluid delivery from the proximal tubule to the downstream macula densa, causing glomerular hyperfiltration via impaired tubuloglomerular feedback (TGF). 4 Finally, glomerular hyperfiltration exaggerates the work of sodium transport and oxygen consumption in the kidney, particularly in the proximal tubules, 5 leading to subsequent kidney damage. Therefore, diabetes-induced glomerular hyperfiltration is one of the major risk factors for the subsequent development of diabetic kidney disease. 6 High intake of dietary sodium was associated with an elevated incidence of cardiovascular disease in patients with T2D, 7 while urinary sodium excretion was nonlinearly associated with all-cause mortality and the cumulative incidence of ESRD. 8 Lower urinary sodium excretion was reported to be associated with increased all-cause and cardiovascular mortality in T2D patients. 9 Therefore, sodium intake may be associated with risks of cardiovascular and kidney disease.
Also, dietary sodium intake was reported to influence renal haemodynamics. 10 However, little is known about the mechanism of the effects of sodium intake on renal haemodynamics.
The SGLT2 reabsorbs, not only filtered glucose, but also sodium.
In diabetes, increased expression and activity of SGLT2, and fully activated SGLT1, account for almost 50 g of sodium, which may represent over 10% of the filtered sodium load, may be reabsorbed via SGLTdependent pathways. 11 Although post meal urinary sodium excretion, in addition to urinary glucose excretion, was increased from baseline, both acutely and chronically, by administration of an SGLT2 inhibitor (SGLT2i), 12 little is known about the association of sodium intake with the clinical effects of SGLT2is. Fundamental experiments indicated that genetic and pharmacological inhibition of SGLT2 attenuated primary proximal tubule hyper-reabsorption of sodium and glucose in diabetic models and, thereby, lowered glomerular hyperfiltration via TGF. 13,14 Additionally, lowering of the glomerular filtration rate (GFR) via the SGLT2i, empagliflozin, was reported in patients with type 1 diabetes. 1 Recently, striking reductions in the relative risk of, not only cardiovascular, but also renal, outcomes with use of SGLT2is in patients with T2D were observed in the EMPA-REG OUTCOME trial, the CANVAS Program and the DECLARE-TIMI 58 study. [16][17][18][19] However, the renal effects, particularly those on the estimated glomerular filtration rate (eGFR) at different levels of baseline sodium intake estimated from urinary sodium excretion, have not been investigated. We therefore investigated the impact of basal salt intake on changes in the eGFR in patients with T2D using an SGLT2i, tofogliflozin (TOFO), focusing on early and chronic effects, as well as effects two weeks after the termination of treatment.

| RESEARCH DESIGN AND METHODS
A pooled analysis was conducted on two Phase 3 studies (Table S1) 20 Individual-level data from the 52-week core treatment and 2-week termination of treatment periods of each study were used for analysis. Each included study was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice. Protocols were reviewed and approved by the institutional review boards of each participating center. All patients provided written informed consent prior to enrollment.
The following laboratory variables were measured at baseline: HbA1c, fasting plasma glucose (FPG), sodium, potassium, uric acid, brain natriuretic peptide (BNP), urine creatinine, urine sodium, urine potassium, urine albumin-to-creatinine ratio (ACR), creatinine and cystatin C. The eGFR of creatinine (eGFR MDRD ) was estimated using the Modification of the Diet in Renal Disease (MDRD) formula for the Japanese population 21,22 and the eGFR CKD-EPI was estimated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) formula for the Japanese population. 23,24 In addition, the eGFR CRE + CYS from the Japanese equation 24,25 was derived from both serum creatinine and cystatin C values. The other variables measured were systolic blood pressure (SBP), diastolic blood pressure (DBP) and body weight. To determine predicted 24-hour urinary sodium and potassium at baseline, we used the Tanaka  To assess the effects of basal DSI on changes in metabolic variables during the use of TOFO, participants were divided into four groups according to quartiles of basal DSI (1st quartile, DSI <7.9; 2nd quartile, 7.9 to <9.3; 3rd quartile, 9.3 to <11.0; 4th quartile, ≥11.0).
Adjusted assessments of HbA1c, FPG, sodium, potassium, body weight, SBP, DBP, eGFR MDRD , BNP and ACR were analysed using an analysis of covariance (ANCOVA) model, with the baseline values age, sex and eGFR MDRD as covariates to determine changes across quartiles. In the evaluation of adjusted assessments of cystatin C, eGFR CKD-EPI and eGFR CRE + CYS , changes were analysed using the ANCOVA model, with baseline values, age and sex as covariates. The adjusted assessment of creatinine was also analysed, using the ANCOVA model with baseline values, age and sex as covariates across quartiles of DSI.
Changes in eGFR MDRD were assessed in participants receiving, and in those not receiving, renin-angiotensin system (RAS) inhibition drugs (ARB and/or ACE inhibitor [ACEI]) as concomitant antihypertensive therapy and in participants with or without hyperfiltration. Participants were divided into two groups according to whether they received RAS inhibition drugs and into another two groups according to basal eGFR MDRD (without hyperfiltration [eGFR MDRD < 120 mL/min/1. To identify baseline clinical factors that might independently influence changes in eGFR MDRD
Participants were divided into four groups according to quartiles of estimated basal DSI (Table 1). Age was greater, and BMI, eGFRs (eGFR MDRD , eGFR CKD-EPI , eGFR CRE + CYS ) and the proportion of participants with albuminuria (ACR = >30 mg/g Cre) were higher, based on increased basal DSI, while glycemic status and blood pressure were consistent among all quartiles. Changes in variables, glycaemic status, body weight and blood pressure, at both Weeks 4 and 52 were consistent across quartiles (Table S2). Moreover, differences in BNP level and urine ACR were insignificant across quartiles at Week 52. On the contrary, at Week 4, the reductions in eGFR MDRD and eGFR CKD-EPI were higher, based on increased basal DSI. At Week 52, the increase in eGFR MDRD and eGFR CKD-EPI tended to be smaller, according to the increased basal DSI, while the reduction in eGFR CRE + CYS and increase in cystatin C were greater, based on increased basal DSI.
At Week 4, baseline HbA1c, fasting plasma glucose, SBP, BNP, urinary sodium and potassium excretion, and DSI were significantly correlated with the change in eGFR MDRD (Table S3). There were significant differences in the change in eGFR MDRD at Week 4 according to dosage of TOFO, use of concomitant antihypertensive drugs including ARB and RAS inhibition drugs, and ACR category (<30 mg/g Cre or ≥ 30) (Table S4). At Week 52, age, baseline HbA1c, fasting plasma glucose, cystatin C and urinary sodium excretion, and DSI were correlated with change in eGFR MDRD (Table S3). At Week 52 there were significant differences in the change in eGFR MDRD, according to the use of concomitant antihypertensive drugs, including the use of ARB (Table S4). Multivariable analysis demonstrated that higher basal DSI and HbA1c levels, and use of RAS inhibition drugs, were negatively correlated with changes in eGFR MDRD at Week 4 ( Table 2), but were positively correlated with changes in creatinine levels (  Figure 2).

| DISCUSSION
The present study is the first to show an independent correlation between basal DSI and changes in eGFR MDRD at both Weeks 4 and 52 during treatment with the SGLT2i, TOFO, in patients with T2D.
We also found that basal DSI did not correlate with improved hyperglycaemia, decreased body weight and blood pressure, and changes in ACR. Moreover, participants with both lower basal HbA1c and DSI experienced greater increases in eGFR MDRD from baseline to Week 52. Based on these findings, we concluded that changes in eGFR MDRD as the result of TOFO indicated the attenuated glomerular relativehyperfiltration and subsequent improved renal function, which might be attributed to the handling of reabsorption of, not only urinary glucose, but also sodium, through SGLT2 in the proximal tubule.
eGFR MDRD levels initially dipped at Week 4, which is consistent with results of previous studies. 18,27 However, the clinical factors that influenced the initial reduction in eGFR MDRD remain to be elucidated.
In our participants, the basal eGFR MDRD increased with higher levels of basal DSI, suggesting that the proportion of participants with glomerular relative-hyperfiltration might be higher following increases in DSI across quartiles. Multivariable analyses also indicated that basal DSI was correlated with eGFR MDRD and changes in creatinine levels, independent of basal eGFR MDRD , creatinine level, HbA1c level and use of RAS inhibition drugs. Therefore, these differences in the degree of initial dip in eGFR MDRD as the result of TOFO was attributed to the correction of TGF action via SGLT2i. These findings suggested that salt intake-induced glomerular relative-hyperfiltration might be caused by SGLT2 in the proximal tubule and that the inhibition of SGLT2 might attenuate it in patients with T2D.
Our results also indicated that the change in eGFR MDRD   However, the monitored period of TOFO treatment was limited and histological confirmation was not performed to evaluate structural changes in the kidney. Therefore, further investigation will be needed to evaluate the effects of RAS inhibition drugs on changes in eGFR as the result of long-term SGLT2i treatment.
From our results, we could not explain the association of the Na +/H + -exchanger 3(NHE3), which contributed to reabsorption of urinary sodium, with the effects of long-term TOFO treatment on eGFR MDRD . SGLT2 is co-expressed with NHE3, which reabsorbs approximately 30% of filtered sodium. Recent studies provide evidence that SGLT2 may be functionally linked to NHE3, such that SGLT2 inhibition may also inhibit NHE3 in the proximal tubule. 4,32,33 Our study indicated that baseline DSI, which was estimated from urinary sodium excretion, might independently influence changes in eGFR MDRD during long-term TOFO treatment. The interaction between SGLT2 and NHE3, and change in the reabsorption of sodium leading to increased delivery of sodium to the macula densa after Two weeks after termination of treatment with TOFO, eGFR MDRD levels had further increased from Week 52, which was consistent with previous reports. 18 These results supported the opinion that, with long-term administration of SGLT2i, the attenuation of glomerular relative-hyperfiltration might be maintained. Recently, Cherney et al.
reported that, 30 days after termination of treatment with empagliflozin, a greater reduction in urine ACR was maintained with emphagliflozin than with a placebo. 34 In our study, the positive correlation between change in eGFR MDRD and urine ACR from Week 52 to Week 54 was observed in participants with albuminuria ( Figure S4). This result suggested that attenuation of glomerular relativehyperfiltration might contribute, in part, to the reduction in urine ACR. The report by Cherney et al. also suggested that long-term administration of empagliflozin might contribute to functional improvement of the kidney. 34 Results of our correlation analysis indicated that increased eGFR MDRD during the termination period was not significantly correlated with the initial decrease, which might be caused by the attenuation of glomerular relative-hyperfiltration ( Figure S5). Moreover, DSI was not significantly associated with increased eGFR MDRD during the termination period, based on results of multivariable analysis (Table S7). These findings suggested that increased eGFR MDRD after the termination of TOFO treatment might be caused, not only by the released attenuation of glomerular relative-hyperfiltration, but possibly by other factors as the result of long-term use of the medication, such as functional changes in the kidney.
Although study participants had a mean eGFR MDRD value at baseline of 83.9 mL/min/1.73m 2 , 30 of these participants had eGFR MDRD levels ≥120 mL/min/1.73m 2 (hyperfiltration). The proportion of participants with hyperfiltration tended to increase in association with increased basal DSI. We further investigated the effects on eGFR MDRD , according to whether participants had hyperfiltration during the study period ( Figure S6). Reduction in eGFR MDRD at Week 4 was greater in participants with hyperfiltration than in those without   39 Of course, further investigation of the effects of DSI on eGFR as the result of SGLT2i treatment, according to the use of concomitant hypertensive drugs, especially diuretics, will be needed.
Importantly, measuring and monitoring actual daily urinary sodium excretion and investigating its association with the effect of longterm SGLT2i treatment would be desirable.
This study has several other limitations. We could not perform a comparison between a placebo and TOFO, and the baseline eGFR MDRD level in almost all participants was less than 120 mL/min/1.73m 2 . We could not measure the actual GFR and variables related to intrarenal haemodynamic function. Histological confirmation was not performed to evaluate structural changes in the kidney. This study was the first to elucidate baseline predictors of changes in eGFR MDRD at Weeks 4 and 52 from the results of multivariable analyses. However, the contribution of these predictors to change in eGFR MDRD according to the multivariable analyses might be small; more studies to determine greater contributions to eGFR as the result of SGLT2i treatment are required. Also, our results were obtained in Japanese participants with T2D, whose DSI might be higher than that in other populations. 40 Thus, investigation of the impact of salt intake on the kidney as the result of SGLT2i treatment in areas other than those addressed here is needed. Finally, prospective long-term placebo-controlled randomized studies with larger cohorts, in vitro molecular actions and pharmacological factors are required to confirm conclusions based on the results of this study.
In conclusion, basal DSI was independently correlated with changes in eGFR MDRD after long-term treatment with an SGLT2i, TOFO, in patients with T2D. Moreover, both low basal HbA1c and DSI levels were associated with a greater increase in eGFR MDRD at Week 52. The effects of the SGLT2i, TOFO, on eGFR might be associated with the handling of, not only glucose, but also sodium, through SGLT2 in the proximal tubule.