Why Is the GFR So High?: Implications for the Treatment of Kidney Failure

Abstract

The high GFR in vertebrates obligates large energy expenditure. Homer Smith’s teleologic argument that this high GFR was needed to excrete water as vertebrates evolved in dilute seas is outdated. The GFR is proportional to the metabolic rate among vertebrate species and higher in warm-blooded mammals and birds than in cold-blooded fish, amphibians, and reptiles. The kidney clearance of some solutes is raised above the GFR by tubular secretion, and we presume secretion evolved to eliminate particularly toxic compounds. In this regard, high GFRs may provide a fluid stream into which toxic solutes can be readily secreted. Alternatively, the high GFR may be required to clear solutes that are too large or too varied to be secreted, especially bioactive small proteins and peptides. These considerations have potentially important implications for the understanding and treatment of kidney failure.

The GFR is impressively high. In normal humans, about 20% of total cardiac output is channeled to the kidneys to sustain filtration of over ten times the extracellular fluid volume each day. The process is metabolically expensive, with approximately 10% of resting energy production expended on reabsorption of filtered solutes and water as the glomerular filtrate is converted into urine.

Why is the GFR so high when glomerular filtration is so expensive? This question may be recast teleologically: What purpose does glomerular filtration serve? We suppose that a high GFR is required to remove noxious wastes. The kidney has other functions, including the production of renin, erythropoietin, and active vitamin D and the metabolism of amino acids and glucose. None of these processes, however, require glomerular filtration’s fluid flow and metabolic cost.

If glomerular filtration exists to remove waste, the question “Why is the GFR so high?” becomes “What substances require a high GFR for their removal?” Homer Smith (1) proposed that glomerular filtration evolved to provide rapid removal of water. In From Fish to Philosopher: The Story of Our Internal Environment (1), he famously depicted glomerular filtration appearing with the evolution of fish in fresh water (Supplemental Figure 1). He proposed that water unavoidably absorbed from a dilute sea was excreted by the addition of an ultrafiltering glomerulus to an existing secretory tubule, which adapted to recover valuable filtered solutes. This established a pattern for kidney structure and function that persisted in most vertebrates over the succeeding 500 million years.

The Relation of Glomerular Filtration Rate to Metabolic Rate

Smith’s suggestion that glomerular filtration developed to allow primordial vertebrates to excrete water is outdated. We know now that many invertebrates have a fluid excretory system consisting of an ultrafilter at the head of an epithelial tubule (2–4). This ultrafilter contains cells similar to glomerular podocytes and was presumably present in the creatures from which vertebrates evolved. More importantly, the GFR among vertebrates is more closely related to the metabolic rate than to the need to excrete water. Singer and Morton (5) reported that the GFR in mammals increases with body weight raised to the 0.77 power, as illustrated in Figure 1. Thus, a human weighing 300 times as much as a rat has a GFR only 80 times as high. Singer (6) noted that the resting metabolic rate bears a similar relation to body weight in mammals and suggested that the metabolic rate determines the GFR. The proportionality of GFR to metabolic rate extends beyond mammals, as previously described by Calder (7) and illustrated in Figure 2. Warm-blooded mammals and birds have a higher GFR than amphibians, reptiles, and fish of the same body size, and their higher GFR is proportional to their higher metabolic rate. Freshwater fish have a higher GFR than marine fish of the same size, and fish that live in both salt and fresh water lower their GFRs in salt water (8,9). These findings can be accounted for by Smith’s hypothesis that freshwater fish increase their GFR to excrete an obligate water load. However, the GFR in mammals and birds is higher in proportion to their body weight than the GFR in freshwater fish, and this cannot be attributed to an evolutionary precursor’s need to excrete water. The relation of GFR to metabolic rate suggests rather that a high GFR is required to clear metabolic waste solutes. We do not know which solutes are so toxic that a high GFR is required to clear them, but the question is of obvious importance to the treatment of uremia.

Figure 1.

The relation of GFR to body size as derived by Singer and Morton (5) from data collected by Renkin and Glimore (10). The line describes the relation GFR=0.0234×BW^0.77. Singer and Morton (5) noted that this relation is similar to the allometric relation of resting metabolic rate (M) to body weight (BW), which can be written “M is proportional to BW^b,” with the exponent b having been found to be near 0.77 among mammals (11). Redrawn from ref. 5, with permission.

Figure 2.

The relation of GFR to body size in different groups of animals as derived by Calder (7) from data collected by Yokota et al. (12). The relation depicted for mammals is very close to that derived by Singer and Morton (5) and depicted in their figure 1. The supposition that GFR is proportional to metabolic rate (M) accounts for the lines being nearly parallel. The relation of M to body weight (BW) can be expressed as M=a×BW^b not only in mammals but also in other groups of animals. The lines are nearly parallel because b has been found to be around 0.7 not only among vertebrates but also among many invertebrates (11,13). This remarkable similarity in the relation of M to body size among different animal groups remains to be explained (13). Values for a differ widely among animal groups and are much higher in warm-blooded mammals and birds than in amphibians, reptiles, and teleosts (bony fishes). The GFR is, therefore, much higher in mammals and birds than in other vertebrates. Of note, cardiac output and liver oxygen consumption have nearly the same relation to body size and M as GFR, whereas liver size tends to increase to a slightly greater degree with body size in mammals (14,15). In mammals, the GFR remains relatively stable over a wide range of physiologic circumstances. The GFR is more variable in other vertebrates and is often reduced in response to water deprivation (9). F.W., fresh water. Redrawn from ref. 7, with permission.

Solutes Cleared by Secretion

The clearance of many solutes is raised above the GFR by proximal tubular secretion (16–19). We presume that secretion evolved to reduce plasma levels of substances that are particularly toxic. As Smith (1) noted, referring to secretion as “excretion”:

The reason why some substances are excreted by the tubule and others are not is poorly understood, but it appears that this process helps to rid the body of certain types of substances which cannot be utilized and the accumulation of which in the blood would be injurious.

Going further, we would consider the possibility that glomerular filtration serves mainly to provide a fluid stream into which toxic solutes can be secreted. Secretory clearance is limited by the fluid flow at the end of the proximal tubule and by the transepithelial concentration gradient that secretion can achieve. If a lower GFR provided a lower fluid flow, tubular cells would have to secrete solute against a higher gradient to maintain the same clearance.

Knowledge of the transepithelial concentration gradients achieved by tubular secretion is limited. However, we can estimate the transepithelial concentration gradients required for tubular secretion in a hypothetical nephron as depicted in Figure 3. The most extensively studied solute secreted by the proximal tubule is para-aminohippurate (PAH). Smith used PAH to measure kidney plasma flow because tubular secretion removes it almost completely from the peritubular capillary blood. As shown in Figure 3, left panel, our hypothetical nephron could clear 90% of the PAH from the kidney plasma flow by secreting PAH against a concentration gradient of 126:1 at the end of the proximal tubule. For any given rate of production, the plasma level of a solute that is cleared efficiently by secretion is lower than it would be if the solute was cleared only by filtration. The combination of binding to plasma proteins and tubular secretion can keep the free, unbound solute levels to which tissues are exposed even lower, as shown in Figure 3, right panel. Rapidly reversible dissociation of solute from the binding protein in the peritubular capillaries allows the tubule to secrete solute at a rate that exceeds the free plasma concentration multiplied by the kidney plasma flow. Moreover, remarkably, as shown in the figure, protein binding allows the free, unbound plasma levels of secreted solutes to be kept very low, with the tubule achieving a lesser concentration gradient than is required to provide a high clearance of unbound solutes like PAH. To achieve equally low effective plasma levels without protein binding would require both a greater kidney plasma flow and a higher transepithelial concentration gradient.

Figure 3.

Tubular secretion provides solute clearances that are greater than the GFR. The values illustrated assume a kidney plasma flow of 600 ml/min and a GFR of 120 ml/min, with two-thirds of the filtrate reabsorbed by the end of the proximal tubule. The left panel depicts the clearance of para-aminohippurate (PAH), which is assumed not to bind to plasma proteins. The clearance is 540 ml/min or 90% of the kidney plasma flow. A lumen-plasma concentration gradient of 126:1 must be achieved by the end of the tubule if secretion is assumed to occur all along its length. The right panel depicts the clearance of p-cresol sulfate (PCS), which is approximately 98% bound to plasma proteins in humans. Values for its plasma concentrations are given as the total concentration [PCS]_T and the free, unbound concentration [PCS]_F. When expressed in accordance to standard convention in terms of the total concentration [PCS]_T, the PCS clearance of 23 ml/min is much lower than the PAH clearance. When expressed in terms of the free concentration [PCS]_F to which body cells are exposed, however, the PCS clearance of 1150 ml/min is more than twice that of PAH. Of note, this higher effective clearance can be achieved with the tubular epithelium maintaining a lesser lumen-plasma concentration gradient of 28:1 at the end of the tubule. To achieve an equally low plasma concentration without protein binding while excreting the same amount of PCS daily would require much greater kidney plasma and blood flows. A higher transepithelial concentration gradient would also be required unless the GFR was greatly increased. In addition to facilitating solute excretion by the kidney, protein binding can blunt changes in solute levels in response to changes in clearance as depicted in Figure 5. Plasma concentrations and clearances for PCS in normal humans are adapted from Sirich et al. (20).

Because the number of secreted waste solutes is large, individual secretory transporters must carry a wider variety of solutes than the transporters responsible for recovering valuable solutes from the glomerular filtrate. Molecular characterization of the secretory transporters and their binding characteristics has, therefore, so far not brought us much closer to identifying which of the myriad secreted solutes are most toxic (18,21). It is logical to look for toxic solutes among those with the highest clearance rates. This group includes several compounds that are produced by colon microbes and have high effective clearance rates due to a combination of tubular secretion and protein binding (16,22,23). Indoxyl sulfate and p-cresol sulfate have been the most often studied (24). Production rates of these solutes are highly variable among individuals and not clearly related to the metabolic rate. It should be emphasized, however, that the number of colon-derived solutes cleared by secretion is large and that solutes that were identified later because their concentrations are lower may ultimately prove more toxic than those that have been most frequently studied (23).

The question of the extent to which a high GFR is required to allow tubular secretion thus remains unsettled. For birds, a high GFR may be required to prevent secreted uric acid from obstructing the tubules (25). In mammals, however, we do not have sufficient knowledge of the toxic secreted solutes and the limitations of their transepithelial concentration gradients to say that a high GFR is required to clear them rapidly.

Solutes Cleared by Filtration

Our high GFR may also be necessary to remove solutes that cannot be secreted. Circulating low mol wt proteins fall in this category. They are filtered and taken into proximal tubule cells via the scavenging proteins megalin, cubilin, and amnionless on the brush border (26,27). They are then degraded in lysosomes, and their constituent amino acids are returned to the circulation. Solutes cleared by this process include numerous molecules derived from signaling, inflammatory, and enzymatic pathways, with β₂-microglobulin being the best known to nephrologists (26,28). Studies in patients with aberrant proximal tubule endocytosis have revealed that the human kidney clears >1 g of low mol wt proteins each day (29). Some of these proteins are degradation products, while others are intact proteins such as the protease inhibitor cystatin C and bioactive components of the complement system (30).

Although low mol wt proteins are taken up by endocytosis and hydrolyzed in lysozomes, smaller peptides are hydrolyzed by brush border proteases/peptidases and then reabsorbed as amino acids or di- and tripeptides (31–33). The characteristics that determine whether an individual solute is handled by one or the other of these processes are not well defined. Brush border proteases, however, have been shown to degrade peptides with molecular mass up to 3500 D (34). Recent mass spectrometric analyses have identified large numbers of such peptides in the plasma (35,36). We can suppose that a high GFR serves to prevent injurious accumulation of peptides degraded by brush border proteases as well as low mol wt proteins taken up by endocytosis (37).

Finally, we must consider solutes cleared by filtration that are smaller than the low mol wt proteins and peptides degraded by the proximal tubule. We have assumed that toxic small solutes can be secreted, and are reluctant to suggest that there are chemical structures for which evolution could not provide secretory transport. Perhaps, however, the very multiplicity of small organic waste solutes makes filtration the most efficient means to dispose of the majority, with secretion superimposed only to lower levels of the most toxic. This raises an interesting question in comparative physiology. How do aglomerular species dispose of solutes that glomerular species clear by filtration? Insects likely clear low mol wt proteins in nephrocytes, which contain analogs of cubilin and amnionless that move captured proteins on to lysosomal degradation (38). How aglomerular fish clear such solutes remains unknown.

Clues to the Nature of Solutes that Require a High Glomerular Filtration Rate for Their Removal

We suppose that the waste solutes that require a high GFR for their removal, whether cleared by filtration or secretion, are breakdown products of compounds normally found in the body. In this, we discount the suggestion of Schmidt-Nielsen (39) that the GFR is kept high to allow excretion of new compounds, which animals encounter as they explore different environments and change their diets. A potential need to excrete new compounds cannot account for the similar relation of GFR to metabolic rate among animal species whose diets are widely different and individually sometimes very restrictive. Because the GFR is proportional to the metabolic rate, we suppose further that waste solutes that require a high GFR for their removal are produced in proportion to the metabolic rate. Such solutes include post-translationally modified amino acids and post-transcriptionally modified ribonucleosides. Whole-body turnover is proportional to resting metabolic rate, and protein degradation loads the extracellular fluid with numerous modified amino acids, which are cleared by the kidney (40–44). RNA turnover generally parallels protein turnover and loads the extracellular fluid with modified ribonucleosides, which are also cleared by the kidney (43–45).

Conditions that cause the GFR to vary may provide further clues as to the nature of waste solutes that require a high GFR for their removal. A particularly notable example is protein feeding, which is the only dietary maneuver consistently shown to increase the GFR independent of the metabolic rate (46–48). Single-protein meals cause transient increases in the GFR, and a high-protein diet causes a sustained increase in the GFR (46–50). The GFR does not increase, however, in proportion to the amount of protein consumed. In humans, for example, the GFR was found to be about 20% higher when protein intake averaged 2.6 g/kg per day compared with 0.3 g/kg per day over 2 weeks (51). This rise in GFR is not sufficient to limit plasma levels of urea or other compounds derived from catabolism of amino acids in the dietary protein. It may, however, be accounted for by the hypothesis that the GFR serves to clear solutes produced in proportion to the rate of protein turnover throughout the body, which is usually much greater than the dietary protein intake (52). In humans, for example, approximately 300 g of body proteins are degraded to and synthesized from amino acids each day when the diet contains 100 g of protein each day. Increasing the dietary protein intake increases the whole-body protein turnover, but the increase in turnover is modest relative to the increase in intake (53,54).

The GFR is also increased in poorly controlled diabetes (55). The increase in GFR in this setting is associated with an increase in both metabolic rate and protein turnover rate (56,57). High and low GFR values seen in hyperthyroidism and hypothyroidism are likewise associated with variations in metabolic rate and, less certainly, protein turnover rate (6,58). Pregnancy affords an interesting exception to this general pattern. The GFR is increased by 30%–40% at the end of the first trimester of human pregnancy, while the metabolic rate and protein turnover rate remain nearly stable (59–62). We can only speculate that this high GFR keeps the extracellular fluid exquisitely clean during early fetal development.

Kidney Size and Glomerular Filtration Rate

Given that the majority of metabolic work necessitated by glomerular filtration is devoted to tubular solute reabsorption, it is not surprising that kidney size generally parallels GFR. The ratio of kidney weight to body weight across a wide variety of mammalian species is similar to the ratio of GFR to body weight depicted in Figure 1, possibly with a slight tendency toward a relatively greater kidney weight in larger species (63). Potentially interesting deviations from this pattern in select species remain to be accounted for (63,64). If we suppose that the GFR is normally set high to remove noxious metabolic waste compounds, it is logical to suppose that rising concentrations of these compounds in the plasma stimulate remnant nephron hyperfiltration and hypertrophy when nephron number is reduced. In adult animals, reduction in nephron number increases the plasma levels of both substances cleared by secretion and substances cleared by filtration. We, therefore, obtain no clue as to the nature of the substance(s) that we have supposed promotes remnant nephron hyperfiltration and hypertrophy. Some indication of the nature of these substances may, however, be gleaned from remnant kidney growth following reduction of functioning nephron number in fetal animals (65,66). For technical reasons, the best information on fetal kidney growth derives from studies in sheep. Both unilateral ureteral obstruction and unilateral nephrectomy cause growth of the other kidney in fetal sheep. With uninephrectomy, nephron number increases, but with obstruction, kidney mass and DNA content rise without any increase in nephron number (65,67). The same response appears to occur in humans. Infants born with unilateral kidney agenesis or ureteral obstruction demonstrate enlargement of the unaffected kidney (68,69). Thus, kidney size in the fetus depends at least in part on the functioning fetal kidney mass. However, a reduction in maternal nephron number in sheep and rats has no effect on the kidney weights of the offspring, suggesting that the stimulating factor does not cross the placenta (70). Because we assume small solutes cross the placenta more easily that proteins and peptides, these observations suggest that the responsible solute is a protein or peptide degraded by the fetal proximal tubule. The lack of correlation between fetal and maternal blood levels for small proteins, such as β ₂-microglobulin, supports the view that these solutes are generated by the fetus but not cleared by the maternal kidney (71).

Why Is Glomerular Filtration on All the Time?

Glomerular filtration is remarkable not only for its high rate but also for its stability. Filtration and, presumably, secretion proceed throughout the day with only modest circadian variation (72–74). We might expect the GFR to decline during exercise to conserve energy. Early reports found the GFR remained stable with moderate exercise but declined with exercise to exhaustion in humans (75–78). Subsequent reports have shown further that the GFR remains stable during distance running (79,80). The GFR is also relatively well maintained even in seals swimming underwater. Early reports of a marked reduction in GFR were obtained when diving was simulated by forced immersion or asphyxia (81,82). Later studies showed the GFR remained stable during voluntary dives lasting as long as 10 minutes and declined only when underwater exercise continued to the point of lactic acid accumulation (83). We presume that glomerular filtration is maintained stable to avoid transient elevations in solute levels. Elevations in solute levels in response to reductions in clearance would be particularly prominent for secreted solutes that have a high clearance relative to their volume of distribution as shown in Figure 4.

Figure 4.

The predicted effect on plasma solute concentrations of a transient reduction in their kidney clearances. Solute concentrations relative to their normal levels are depicted around a period during which their clearances are reduced to 20% of normal for 3 hours as indicated by the shaded band. The rapidity with which a solute's level will change is determined by the ratio of its normal clearance to its volume of distribution (K/V) assuming there is no change in its production. The level of a solute that is cleared by secretion and restricted to the extracellular fluid (K/V=(540 ml/min)/(14,000 ml) or approximately 0.4 min⁻¹) would rise to four-fold normal over 3 hours and then return rapidly toward normal, as depicted by the red line. In contrast, the level of urea with K/V=(60 ml/min)/(42,000 ml) or approximately 0.0014 min⁻¹ would rise by only 20% and then return slowly toward normal, as depicted by the blue line. Protein binding can blunt variations in plasma levels caused by changes in the clearance of secreted solutes. The yellow line depicts predicted changes in the level of a bound solute with K/V=(23 ml/min)/(13,000 ml) or approximately 0.002 min⁻¹, similar to values for PCS obtained by Sirich et al. (20). Protein binding can likewise blunt change in solute levels in response to changes in solute production, as shown in Supplemental Figure 2.

Medical Considerations

We assume that performance is somehow limited in people with a half-normal GFR but cannot say how. Indeed, we assure transplant donors that removal of one kidney does not have significant ill effects. Perhaps only tests requiring sustained physical or mental activity would reveal the limitations of a lower GFR. The value of having an extracellular fluid free of some solute(s) might also become apparent only over a long time. We can imagine, for instance, that a half-normal GFR would impair reproductive performance or increase susceptibility to infection. Defects of this nature might be important over generations without being apparent clinically.

Although difficult to detect when the GFR is half normal, defects in performance become readily apparent when the GFR falls to low levels. Energy, appetite, and sexual function are impaired in people with GFR values of one-tenth normal. We presume that solutes normally cleared by the kidney cause much of this “uremic” illness. We know remarkably little, however, about the toxicity of individual solutes. Part of the problem is their large number. A pioneering 2003 study by the European Uremic Toxin Work Group (84) described 92 uremic solutes that had been identified up to that time. The number of uremic solutes has since risen to >250 and will continue to grow with the use of untargeted mass spectrometry (23,85–87). This analytic method allows simultaneous estimation of hundreds of solutes in individual plasma samples. However, the paucity of quantifiable clinical and physiologic end points remains a major hindrance to identifying toxins.

Better knowledge of which solutes cause uremic illness could improve KRT. Our current use of urea as a marker of dialysis adequacy has important limitations (88). The ratio of the dialytic clearance to the native kidney clearance is higher for urea than for any other solute. As a result, levels of other solutes remain higher relative to normal than urea levels in patients maintained on standard hemodialysis, as depicted in Figure 5. Clearances of many solutes are even lower relative to urea with peritoneal dialysis than hemodialysis (89–92). Yet, peritoneal dialysis controls uremic symptoms with a time-averaged urea clearance, which is lower rather than higher than that provided by conventional hemodialysis. It may be that long daily periods of peritoneal dialysis control the levels of most uremic solutes as well shorter thrice-weekly periods of hemodialysis, but this question has not been adequately examined. Potential means to increase nonkidney clearances of solutes within the body and to limit their production also remain largely unexplored.

Figure 5.

Conventional hemodialysis fails to replicate the control of solute levels achieved by the native kidney. Solute concentrations are presented as a multiple of normal throughout the weekly dialysis cycle for urea (blue line), a low mol wt protein with properties like β₂-microglobulin (green line), and a secreted solute that does not bind to plasma proteins (red line). Values are modeled for thrice-weekly treatment for 3.5 hours, providing the clearances for urea and β₂-microglobulin observed in the HEMO study and for the unbound, normally secreted solute phenylacetylglutamine on the basis of the report of Sirich et al. (20). Compartmental volumes and a nonkidney clearance for β₂-microglobulin are on the basis of the report of Ward et al. (93). Levels of a low mol wt protein that did not have the nonkidney clearance observed for β₂-microglobulin would rise much higher relative to normal in patients on dialysis.

Homer Smith’s original argument that a high GFR was needed for water excretion is outdated. The high mammalian GFR is presumably required to provide excretion by filtration and tubule secretion of largely unidentified toxic waste solutes. Knowledge of the toxic solutes normally removed by the kidney should provide a sounder basis for efforts to artificially replace it.

Disclosures

T.H. Hostetter reports consultancy agreements with the Center for Dialysis Innovation–University of Washington and Tricida; stock options in Tricida; receiving honoraria from Tricida; serving on the scientific advisory board of Tricida; and serving as a scientific advisor or member of the Center for Dialysis Innovation–University of Washington. T.W. Meyer reports serving as a consultant for Baxter and Daiichi Sankyo; reports serving on the JASN and Kidney International editorial boards; and has applied for a patent to improve the dialytic removal of uremic solutes that bind to plasma proteins.

Funding

This work was supported by National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases awards 5U01DK106965 (to T.H. Hostetter) and R01 DK101674 (to T.W. Meyer).

Supplemental Material

This article contains the following supplemental material online at http://cjasn.asnjournals.org/lookup/suppl/doi:10.2215/CJN.14300920/-/DCSupplemental.

Supplemental Figure 1. Homer Smith’s depiction of the evolution of the glomerular kidney in freshwater fish.

Supplemental Figure 2. The predicted effect of a 2-hour increase in solute production to three-fold normal on plasma solute levels.