SGLT2 inhibitors, hemodynamics, and kidney protection

For decades, when teaching students and nephrology fellows about tubuloglomerular feedback (TGF), I was met with polite silence, making it clear that the subject was perceived as arcane, complicated, and clinically irrelevant. TGF, not to be confused with glomerulotubular balance, is the phenomenon wherein the macula densa senses some correlate of flow and regulates glomerular filtration in response. Its discovery is typically attributed to Thurau and Schnermann, who performed the meticulous experiments that were required to identify it. Although of physiological interest among a small group of dedicated scientists, TGF was not a focus for students and physicians alike. Fast forward to today, and its centrality to the remarkable kidney protection afforded by sodium-glucose cotransporter-2 (SGLT2) inhibitors in diabetic kidney disease and other conditions has turned the tables. Although “pleiotropic” effects within and outside the kidney are frequently invoked to explain the salutary effects of SGLT2 inhibitors, the most compelling mechanistic models involve activation of TGF (1). A study by Thomson and Vallon recently published in the American Journal of Physiology-Renal Physiology (2) adds several important pieces to the puzzle that is the remarkable effectiveness of these drugs.

Hyperglycemia increases SGLT2 activity because glucose is freely filtered and glucose reabsorption along the proximal tubule (together with sodium) is load dependent. Additionally, the sustained exposure to excess glucose that occurs in diabetes can lead to proximal tubule hyperplasia and hypertrophy. For this reason, sodium delivery out of the proximal tubule to the macula densa is reduced by hyperglycemia. Sodium (and chloride) delivery to the macula densa is the canonical signal that activates TGF, thereby reducing glomerular filtration rate (GFR). Thus, relatively low salt delivery to the macula densa is believed to be a central contributor to hyperfiltration in diabetes by inhibiting TGF (3), as shown schematically in Fig. 1.

Figure 1.

A: relation between single-nephron glomerular filtration rate (SNGFR) and late proximal flow rate, where late proximal flow = SNGFR − proximal reabsorption. Note that late proximal flow is lower in diabetes at any SNGFR and that this relationship may be curvilinear, as shown in Thomson and Vallon (2). B flips A (to reorient the axes) and superimposes it and C. C: relation between late proximal flow rate and SNGFR, which is a manifestation of tubuloglomerular feedback (TGF). In B, the operating point of the system is the intersection between the blue and green or red lines. Note that SNGFR is higher in diabetes than in controls, but that SGLT2 administration normalizes it. This analysis was adapted from Palmer and Schnermann (9). SGLT2, sodium-glucose cotransporter-2.

Although the existence of glomerular hyperfiltration in diabetes mellitus has been recognized for many years, it was not always viewed as an attractive therapeutic target, as more attention, until recently, has been on metabolic factors, inflammation, and fibrosis. Yet, gradual progress, first from physiology laboratories, then from medicinal chemists, and more recently from clinical trialists, has now led to what have been characterized as “breakthrough discoveries,” which are transforming the prognosis of diabetes mellitus.

This story began in 1835, when a chemical found in nature, phlorizin from apple tree bark, was isolated and found to cause glucosuria in humans. Although it was recognized that this substance inhibited renal and intestinal glucose transport, poor intestinal absorption and other features meant that it was relegated to the physiology laboratory for more than 100 years. Among those who used this molecule to characterize the transport protein activity was Ernest Wright, whose group later went on to clone the SGLT SL5A2 gene, providing molecular insight and pushing the field dramatically.

The remarkable effects of these drugs to slow the rate of kidney disease progression in diabetes, and perhaps other diseases, is now well recognized. Many of us would not have predicted this efficacy because nodular glomerulosclerosis, the hallmark of established diabetic kidney disease, is a lesion in the glomerulus, and not the tubules, where SGLT2 inhibitors have their actions. Yet, tubules communicate with glomeruli via the almost magical juxtaglomerular apparatus, which is responsible for renin secretion, for TGF, and for the production of signaling molecules, such as nitric oxide, ATP, and adenosine. In the case of diabetes, the tubules send the “wrong” message to the glomerulus, telling it that there is volume contraction and forcing GFR to rise! This is an important cause of diabetic hyperfiltration.

The experiments in this study (2) strongly support this model, as the pressure in the glomerular capillaries was reduced by acute SGLT2 inhibitor treatment only when measured directly, but not when estimated using the “stop-flow” method. In the latter, forward flow along the proximal tubule is blocked, by placing a wax block in the tubule lumen. The pressure that develops when flow stops is an index of, although not equal to, the pressure in the glomerular capillary. When this is done, because fluid cannot pass the wax block into further distal nephron segments, TGF is inactive.

A second major finding from this work is that the responses to SGLT2 inhibition likely involve both afferent and efferent arterioles. It has become axiomatic, since the work of Briggs and Wright (4), that TGF responses result primarily from adjustments in afferent arteriolar tone. This would imply tight coupling between single-nephron GFR and the pressure in the glomerular capillary, meaning that knowing one would allow the other to be predicted. Despite a clear role for the afferent arteriole, it has also been recognized, however, that efferent responses may participate under some circumstances (5). Surprisingly, Thomson and Vallon (2) found that single-nephron GFR and the pressure in the glomerular capillary did not change in lock step, a sure sign that the two arterioles work in concert.

Another important finding here was the impact of dietary salt intake on GFR and glomerular capillary pressure. It has been observed that the typical relationship between salt intake and GFR is disturbed in individuals with diabetes and in animals with experimental hyperglycemia; this has been called the “salt paradox,” in which higher salt intake reduces GFR and vice versa. The authors suggested that this may account for the positive association between salt intake and kidney survival in the FinnDiane study (6). This anomaly has been explained because proximal salt reabsorption appears to be more sensitive to inhibition by dietary salt loading in the setting of diabetes than in those without it. Sodium reabsorption along the proximal tubule is dominated by sodium-proton exchange via sodium/hydrogen exchanger isoform 3; it appears that this protein interacts with SGLT2, thereby amplifying the effects of SGLT2 inhibition.

The results presented by Thomson and Vallon remind us to remember the lessons of the past. More than 35 years ago, Brenner and colleagues argued that hemodynamic, rather than metabolic, factors are the dominant drivers of diabetic kidney disease (7). We have previously been down this road before; bardoxolone, which activates nuclear factor erythroid 2-related factor 2 (Nrf2), a transcription factor regulating antioxidant genes, increased GFR in short-term studies of diabetics, leading the community to anticipate that it could slow diabetic kidney disease progression (8). It was also, however, noted to increase albuminuria, a frequent correlate of glomerular hyperfiltration, and, in longer-term studies, there was no evidence for kidney benefit. The development of cardiovascular side effects led to premature study cessation. This adverse cardiovascular signal is reminiscent of the inverse signal observed with SGLT2 inhibitors. Ascribing improved cardiovascular outcomes to a glomerular hemodynamic cause may be excessively “nephrocentric,” but this ongoing story is a stark reminder that we ignore TGF at our, and our patients’, risk.

GRANTS

This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants DK054196 and DK54983, by Veterans Affairs Merit Award 1I01BX002228, and by NCATS UL1TR002369.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

D.H.E. drafted manuscript; D.H.E. approved final version of manuscript.