Although LEC markers are not specific to LECs, they can differentiate them from blood vessel endothelial cells. The most widely used markers are podoplanin, the LYVE-1, vascular endothelial growth factor receptor 3 (VEGFR-3), and the prospero-related homeo-box transcription factor 1 (Prox1) (Seeger et al., 2012). Of these, podoplanin, a mucin-type transmembrane protein, is the most reliable in human kidneys and can be localized in paraffin-embedded tissue using podoplanin antibody (Kalof and Cooper, 2009).

Renal lymphatic anatomical studies, whether by microscopic or microradiographic technique, have previously required removal of the kidney. However, techniques for visualizing larger lymphatics in vivo have been developed and are currently in use (e.g., pedal lymphangiography, lymphangiography and lymphoscintigraphy) (Itkin and Nadolski, 2018), although current methods do not allow visualization of renal lymphatics. It is, however, possible to image the central lymphatic system by injection of contrast into inguinal lymph nodes (intranodal lymphangiography), and direct trans-abdominal catheterization of the cisterna chyli (Itkin and Nadolski, 2018). Minimally invasive lymphatic procedures have also been described. Injection of embolization material, such as N-butyl cyanoacrylate (N-BCA) glue (TRUFILL, Codman Neuro, Rayhnam, MA, United States), into the interstitium of lymph nodes can lead to propagation and embolization downstream (Itkin and Nadolski, 2018). Liver lymphatics can be imaged by contrast injection into the peri-portal space (Itkin et al., 2017). Taken together, this indicates renal interstitial injection of contrast (or probe) material for imaging renal lymphatics could be feasible in the future.

The lymphatic system probably evolved around the time aquatic organisms moved onto land, as there is no well-developed lymphatic system in fish but there is in amphibians (Brown, 2005). Reptiles, many amphibians and bird embryos have lymph ‘hearts’ to assist with moving lymph. These are pulsating chambers, located at points where lymph vessels enter veins. They are highly variable in wall thickness, contain a valve and beat independently of the blood heart rhythm (Satoh and Nitatori, 1980). The kidneys of fish and anuran amphibians have an accessory role as a lymphoid organ, whereas this becomes less so in the kidneys of amniotes (Manning, 1979). In general, reports of reptile (e.g., Davis et al., 1976) and avian (e.g., Morild et al., 1985) renal anatomy fail to mention renal lymphatics.

The majority of research on renal lymphatic anatomy has been done on mammals, particularly dogs. In a seminal study, Pierce (1944) injected dog, rabbit and guinea pig kidneys with India ink, as well as giving intravenous tryptan blue, in order to delineate the renal lymphatics microscopically. He found that renal lymphatics begin in the cortex as intralobular lymphatics, which are sparse, blind-ended tubes in close proximity to the renal tubules. These pass close to, but do not enter, a renal corpuscle, before joining the interlobular lymphatics, which connect to the larger arcuate and interlobar lymphatics. These then drain into hilar lymphatics (Figure 1). There are 6–8 hilar lymphatics and 4–6 lymphatic channels that leave the renal capsule in the dog (Cockett, 1977). Capsular lymphatics, after leaving the capsule, appear to connect to hilar lymphatics in the renal sinus and hilum (Yazdani et al., 2014). The total volume of cortical lymph in a dog kidney is approximately equivalent to only 1% of the volume of blood in the cortical peritubular capillaries (O’Morchoe and Albertine, 1980).

Anatomy between different species is very similar with a few exceptions; rabbit kidneys appear to lack intralobular lymphatics (O’Morchoe and O’Morchoe, 1988) and the sheep kidney lacks a capsular system (McIntosh and Morris, 1971) (Table 1). The species with the least extensive intra-cortical lymphatics (the rabbit) has the lowest urine concentrating ability and the species with the most extensive system (the golden hamster) has the highest concentrating ability (Niiro et al., 1986).

The basic mammalian renal lymphatic anatomy outlined above is the same as that found in humans (Figure 1). This was shown with microscopy at autopsy (Rawson, 1949) or using radiographic studies (Cuttino et al., 1989), and has been confirmed using LEC markers. Ishikawa et al. (2006), using podoplanin antibody (Figure 2), studied normal human kidney tissue obtained at autopsy, and confirmed that lymphatic vessels were most abundant in the interstitium surrounding interlobular, arcuate and interlobar arteries and veins.

From the left kidney lymphatic vessels enter preaortic, para-aortic and retroaortic lymph nodes, and from the right kidney, they enter paracaval, precaval, retrocaval and interaortocaval lymph nodes (Karmali et al., 2014). Both kidneys also send lymphatics posterior to the aorta, which can travel directly to the thoracic duct (Assouad et al., 2006). The surface of the upper pole may also drain through the diaphragm into posterior mediastinal lymph nodes. Renal lymphatics may reach very distant nodes, but always eventually connect to the origin of the thoracic duct, usually via the cisterna chyli (Assouad et al., 2006) (Figure 1).

The terminal thoracic duct variably drains into the left internal or external jugular vein, the left subclavian vein or the venous angle (Ratnayake et al., 2018). The ostial valve is found at this junction and prevents backflow of venous blood (Ratnayake et al., 2018).

Pierce (1944) found a connection between cortical and capsular/perirenal lymphatics in dogs suggesting that lymph can leave the kidney through an alternative route, rather than the renal hilum. Evans blue dye injected into the renal parenchyma appears in hilar, or capsular lymphatics, or both (Sugarman et al., 1942). This lymphatic communication has been confirmed in other species (Bell et al., 1968; O’Morchoe, 1985), including humans (Rawson, 1949), and recently using LEC markers (Seeger et al., 2012), but Kaiserling (1940) was unable to detect this communication in rabbits.

Holmes et al. (1977a) described two sets of connecting lymphatics in dogs, which are differentiated based on where they penetrate the capsule (Figure 1). The first are perforating lymphatics, which are the majority. They penetrate the capsule at any point, either alone or with a small vein, from the subcapsular plexus. They form the primary pathway for lymph from the most superficial parts of the cortex, as opposed to the hilar route (the hilar route is the primary path for the majority of renal lymph). The second are communicating lymphatics, which form the main connecting link between the hilar and capsular systems. These are a continuation of interlobular lymphatics and are closely associated with occasional interlobular arteries and veins that penetrate the capsule. Each kidney was found to contain only 5–10 interlobular arteries that penetrated the capsule like this, and of these, 60% were accompanied by a communicating lymphatic. Where it pierced the capsule the lymphatic was single, but either side of the capsule it quickly branched to form either the interlobular lymphatic plexus or the capsular plexus. For excellent histological sections of communicating and perforating lymphatic vessels in a dog, see Holmes et al. (1977a).

Pierce (1944) found no evidence of medullary lymphatics in dogs, consistent with the lack of arterial supply in the medulla, and concluded they do not exist. Since Pierce, most animal studies have failed to detect medullary lymphatics. McIntosh and Morris (1971), using light and electron microscopy, as well as physiological studies, found no evidence of medullary lymphatics in the sheep. Bell et al. (1968), Albertine and O’Morchoe (1980), and Eliska (1984) could not detect medullary lymphatics in dogs. In contrast to these findings, Cuttino et al. (1985) reported a medullary lymphatic system in pigs, Cockett (1977) in dogs and Ruszniják et al. (1960) in rabbits. Animal studies examining content of renal lymph have shown that medullary interstitial fluid must contribute to hilar lymph (Keyl et al., 1972), suggesting the presence of medullary lymphatics. However, fluid movement within the medullary interstitium to juxtamedullary lymphatics (in the cortex) can also cause this result (Albertine and O’Morchoe, 1980; O’Morchoe and O’Morchoe, 1988).

Rawson (1949) mapped lymphatic distribution in a human kidney autopsy specimen. The specimen was from a patient with a gastric carcinoma that had metastasized in a retrograde manner into the renal lymphatics. This dilated the lymphatics and made them visible under the microscope. He concluded that renal lymphatics begin blindly in two areas; near Bowman’s capsule in the cortex, and beneath the mucosa of the papilla in the medulla. Those in the cortex drain down toward the arcuate vessels and those in the medulla drain up to meet them. Rawson’s findings have been controversial due to the presence of cancer in the renal lymphatics, and recent review articles have not relied on his work and instead have suggested, based on animal studies, that no lymphatic vessels are present in normal human renal medulla (Lee et al., 2011; Seeger et al., 2012). However, Ishikawa et al. (2006) found medullary lymphatics in normal human renal tissue in four out of ten cases (Figure 2), but only near the cortex, with none seen in the central area of the medulla.

Bell et al. (1968) found numerous small lymphatic vessels intimately associated with glomeruli, but none that penetrated into the glomerulus. These vessels either completely (horse) or partially (dog) surrounded Bowman’s capsule. Eliska (1984) was able to detect rare and very small lymphatics near tubules between interlobular arteries, and noted that lymphatics surrounding glomeruli become rarer with increasing distance from an interlobular artery. Ishikawa et al. (2006) confirmed that lymphatic capillaries sporadically surround the glomerulus in humans but do not penetrate it (Figure 2).

Bi-leaflet valves are seen in pre-collector (arcuate and interlobar lymphatics) and collecting (hilar) lymphatics, and facilitate unidirectional flow toward the hilum and aortic lymph nodes (Seeger et al., 2012) (Figure 1). For communicating and perforating lymphatics, often a single valve is present as they traverse the capsule, oriented to prevent flow from the surface back into the renal parenchyma (Holmes et al., 1977a). These valves continue into capsular lymphatics and are numerous there. No valves are present in interlobular lymphatics (Holmes et al., 1977a). Thus, lymph forming in the cortex is permitted to exit the kidney in either direction.

The renal interstitium is defined as the intertubular, extraglomerular, extravascular space of the kidney (Zeisberg and Kalluri, 2015). It is bound on all sides by tubular and vascular basement membranes. The cortical renal interstitium is divided into (a) peritubular interstitium, (b) periarterial interstitium, and (c) glomerular and extra-glomerular mesangium (Lemley and Kriz, 1991). The lymphatics lie in the periarterial interstitium, which is a fluid-rich loose connective tissue sheath. It is seen surrounding afferent arterioles and larger arteries, but it is less distinct in efferent arterioles and tubular capillaries.

Renal lymphatic capillaries are considered part of the interstitium because they lack a basement membrane. They are differentiated from blood vessels in that they are blind-ended and lack pericytes. LECs of capillaries are single-layered, ‘oak-leaf’-shaped, partly overlapping cells, interconnected via discontinuously arranged ‘button-like’ junctions, which result in interjunctional gaps (Seeger et al., 2012). These junctions are composed of vascular endothelial cadherin (VE-cadherin) and various tight junction-associated proteins (e.g., occludin, zonula occludens-1) (Baluk et al., 2007). LECs are connected to the perivascular matrix by filaments. This arrangement leads to an increase in diameter of the vessel lumen and width of intercellular gaps in the presence of tissue edema, which facilitates fluid entry (Seeger et al., 2012). Precollectors (arcuate and interlobar lymphatics) exhibit some perivascular smooth muscle cells, and collectors (hilar lymphatics) contain continuous endothelial junctions, a basement membrane and a smooth muscle cell layer (Figure 1). Lymphatic vessels usually transport lymph via both extrinsic pumping, secondary to surrounding tissue forces, and intrinsic pumping, via these smooth muscle cells, with the latter likely predominating in renal lymphatics (Zawieja, 2009).

Ishikawa et al. (2006) showed that, in humans, lymphatics around interlobular veins were more developed than lymphatics around interlobular arteries. They saw abundant lymphatics in the interstitium around interlobar arteries but noted that lymphatics surrounding interlobar veins were distributed not just in the interstitium, but also in the wall itself, in both the media and the intima just under the endothelium. Although lymphatics travel close to renal vessels, Eliska (1984) and others have been unable to detect any anatomical intra-renal lymphatic-venous shunts. Lymphovenous communications have been seen, however, at the level of the renal vein in rats and primates, but not in human autopsy studies (Karmali et al., 2014).

Lymph from the most outer part of the cortex is primarily drained via the capsular lymph vessels, in contrast to lymph from the medulla and remainder of the cortex, which is primarily drained via hilar lymphatics (O’Morchoe et al., 1975). Although most renal lymph is formed in the cortex, flow rate from hilar lymph has been measured at 4–8 times greater than capsular lymph (Keyl et al., 1965), suggesting that most cortical lymph drains via hilar, and not capsular, lymphatics. This is further evidenced by the fact that the electrolyte composition of hilar lymph is very similar to plasma, which is because the electrolyte composition of the cortical interstitium (where most hilar lymph is formed) also closely resembles plasma. These solutes are found in high concentrations in medullary interstitium (due to the requirements of the counter-current mechanism for concentrating urine), and renal hilar lymph would reflect these concentrations if it were primarily draining the medulla (Keyl et al., 1965).

Studies have confirmed that renal lymph is derived from both capillary filtrate and tubule reabsorbate (Bell and Lowsitisukdi, 1983; Bell, 1984). Kaplan et al. (1942) gave intravenous inulin to dogs and found a renal inulin lymph/arterial plasma ratio of 0.68. If renal lymph where derived exclusively from renal tubular reabsorbate, lymphatic inulin content would be close to zero as inulin is not reabsorbed. Other studies have found a lymph/plasma ratio of 0.80 for inulin (Keyl et al., 1965), 0.78 for creatinine (Cockett et al., 1969) and 0.58 for para-aminohippurate (PAH) (Keyl et al., 1965). The mixed source of renal lymph can also be shown by measuring the ratio of labeled glucose (reabsorbed by the proximal tubule) and mannitol (not reabsorbed) in renal lymph compared with plasma (Cook et al., 1982). The labeled glucose appears in renal lymph in a higher proportion compared with mannitol after reabsorption from the proximal tubule, indicating that tubule reabsorbate contributes to the formation of renal lymph.

If renal lymph were derived solely from capillary filtrate, then its electrolyte composition would be identical to plasma (Cockett, 1977), and while renal lymph biochemistry does resemble plasma, it is not identical. LeBrie and Mayerson (1959) found capsular lymph had an 11.3 and 26.9% higher concentration of Na⁺ and Cl^-, respectively, compared with plasma. They concluded that the distal tubule is able to reabsorb solute independently of water, and this concentrated solution of Na⁺ and Cl^- is mixed with capillary filtrate to make renal lymph. This was not confirmed by Mayerson (1963) who suggested that the sodium was erroneously high and not significant. Other studies have found similar concentrations of Na⁺, K⁺, Mg²⁺, osmolality and pH in plasma and renal lymph (Swann et al., 1958; LeBrie and Mayerson, 1959; Keyl et al., 1965). The concentration of urea in renal lymph varies between studies, with lymph/plasma ratios of 1.4 (Swann et al., 1958), 1.3 (Sugarman et al., 1942) and 0.89 (Keyl et al., 1965), and this discrepancy may represent a difference in methodologies. Keyl et al. (1965) speculated that a raised urea lymph/plasma ratio is due to clamping of capsular lymphatics in order to engorge them for cannulation. This impairs normal renal lymph drainage, which leads to variations in interstitial urea. These studies all suggest that capillary filtrate rather than tubule reabsorbate, is the major source of renal lymph.

Protein concentration in renal lymph varies greatly and has been measured at 20% (rat) (Hargens et al., 1977), 34% (dog) (Keyl et al., 1965), 49% (dog) (Atkins et al., 1988), 50% (dog) (LeBrie and Mayerson, 1960), 60% (dog) (LeBrie and Mayerson, 1959), and 66% (dog) (Henry et al., 1969) of systemic plasma levels, and 70% of thoracic duct lymph levels (LeBrie and Mayerson, 1959). Protein content of renal lymph varies with flow rate (Sugarman et al., 1942; LeBrie and Mayerson, 1960; Bell, 1985), with a lower protein concentration associated with an increased lymph flow rate. It also depends on whether the animal is anesthetized or not, with a higher protein concentration with anesthesia, although the mechanism for this is unknown (Henry et al., 1969).

LeBrie and Mayerson (1959, 1960) found no difference in relative proportions of different proteins (albumin, alpha 1, alpha 2, beta, and gamma globulins) between renal lymph and plasma, but others found a greater proportion of albumin and alpha globulins in lymph (Swann et al., 1958; Keyl et al., 1965; Hargens et al., 1977). The renal lymphatic clearance of albumin has been calculated to be quite high (0.16 cm³/min/100 g kidney) which is consistent with the relatively high permeability of renal capillaries (Atkins et al., 1988) and the high levels of interstitial albumin (Slotkoff and Lilienfield, 1967).

Renal lymph contains a relatively high concentration of renin, which is not found in lymph from other organs or in renal vein blood (Lever and Peart, 1962; Moffat, 1981). This may be explained by lymphatics lying in close proximity to the juxtaglomerular apparatus, where renin is synthesized and acts in a paracrine fashion (Moffat, 1981). Likewise, levels of angiotensin II were found to be higher in renal lymph than in renal arterial or venous samples in dogs (Bailie et al., 1971). It was noted that partial occlusion of the renal artery (renal artery stenosis) causes an increase in the concentration of renin in renal lymph (Lever and Peart, 1962). Apolipoproteins (specifically ARP and A-IV), which presumably have a lipid transport role in the renal interstitium, are also present in renal lymph in relatively high concentrations (Roheim et al., 1976).

LeBrie and Mayerson (1960) studied the composition of renal lymph after partial occlusion of the inferior vena cava (mimicking right heart failure). They found no change in the concentration of Na⁺ and Cl^-, even though overall lymphatic flow (and therefore total Na⁺ and Cl^- content) increased. Profuse diuresis causes decreased concentrations of Na⁺ and Cl^- in renal lymph, a change which is unaffected by initial plasma electrolyte concentrations (O’Morchoe et al., 1970; O’Morchoe et al., 1978).

Increased venous pressure (from 22 to 30–35 cmH₂O in dogs) causes a markedly disproportionate rise in renal lymphatic protein content (LeBrie and Mayerson, 1960). This is thought to be due to increased capillary protein loss when venous pressure is raised to this level (LeBrie and Mayerson, 1960). Protein content of renal lymph decreases during diuresis, which is a direct result of increased lymphatic flow rates (LeBrie, 1968).

The three routes that fluid can leave the kidney are venous, ureteral and lymphatic. These three systems bear relationships to each other and operate reciprocally (Katz and Cockett, 1959). Both obstruction of the upper ureter and the renal vein results in increased renal lymph pressure and flow, and ligation of lymphatic vessels results in increased urine flow (Cockett, 1977). The lymphatic system has been said to function as a ‘safety valve’ mechanism protecting the kidney from high intra-renal pressures, such as in raised IRVP or hydronephrosis (Goodwin and Kaufman, 1956).

Schmidt and Hayman (1929) found that the discrepancy between renal arterial and venous flow during profuse diuresis in dogs was not fully accounted for by urinary production, and concluded the remainder was leaving the kidney via lymphatics. They showed that the kidney, during diuresis, was capable of producing a substantial amount of lymph. O’Morchoe et al. (1970) found increased lymph flow with mannitol diuresis but not mersalyl or furosemide. They concluded the effect of mannitol on lymph flow was due to the general increase in interstitial fluid rather than through any specific intra-renal consequence of the diuresis itself.

Increasing systemic venous pressure causes a decrease in urinary flow and urinary sodium concentration, however, this is likely due to changes in renal blood flow, resulting in a lower glomerular filtration rate (GFR), rather than a direct consequence of the venous pressure (LeBrie and Mayerson, 1960).

Increased systemic venous pressure causes an increase in renal lymph flow. Katz (1958) measured the pressure of capsular lymphatics in dogs during occlusion of the renal vein or ureter. The lymphatic pressure began to rise sharply 5 s after renal vein occlusion, and declined 7 s after its release (Figure 4). A rise in interstitial pressure preceded the rise in lymphatic pressure. These changes also occurred in partial renal vein occlusion, a situation resembling the raised venous pressure seen in cardiac failure. LeBrie and Mayerson (1960) estimated that renal lymph flow increased up to 2400 ml/24 h from both kidneys under conditions of partial inferior vena cava (IVC) occlusion. This represents a volume 3–4 times greater than the plasma volume in a 15 kg dog.

IRVP and interstitial fluid volume are the main drivers of renal lymph formation (Bell et al., 1972, 1974; Keyl et al., 1972). Clamping the renal artery causes decreased capsular lymph pressure that reflects decreased IRVP and tissue pressure (Figure 5) (Bell et al., 1972). The IRVP quickly recovers when the artery is unclamped; however, the tissue pressure and lymphatic pressure recover more slowly. Bell et al. (1972) reasoned that tissue pressure was a result of an interstitial fluid volume that had to be replaced after the artery was unclamped. These experiments are relevant for patients with severe acute disease who have poor renal perfusion (e.g., circulatory shock). However, renal lymph and tissue pressures are maintained even in decreased renal blood flow as long as venous pressure is not allowed to fall, showing that IRVP is the major factor in renal lymph production (Bell et al., 1974).

Hydronephrosis also causes increased renal lymphatic flow (Goodwin and Kaufman, 1956). However, the rise in lymphatic pressure is more gradual after occlusion of the ureter, compared with occlusion of the renal vein (Katz, 1958) (Figure 4). Bell et al. (1974) noted that renal pelvic pressure is relatively ineffective at increasing lymphatic pressure and flow under conditions of decreased renal blood flow. Therefore, the increase in lymphatic pressure and flow seen in increased pelvic pressure is probably actually due to intra-renal venous obstruction secondary to the increased pelvic pressure. Increased pelvic pressure has been observed to obstruct venous outflow from the kidney due to the intimate intra-renal anatomical relationship between renal veins and pelvis (Keyl et al., 1972; Holmes et al., 1977b). In dogs, ureteric obstruction causes a diversion of hilar lymph to capsular lymph within 3 days due to compression of hilar lymphatics in the renal sinus by the distended renal pelvis (Holmes et al., 1977b).

Figure 4 shows a sudden decrease in capsular lymphatic pressure when the kidney was visibly very tense. The author concluded this was from occlusion of the capsular lymphatic vessels as they exited the kidney. This demonstrates a way that capsular renal lymphatics can fail in an oedematous kidney. This would presumably lead to worsening renal oedema and renal function, although the authors did not investigate this further (Katz, 1958).

A further explanation for the increased renal lymph flow seen in hydronephrosis is pyelolymphatic backflow. This has been shown experimentally in pigs and rabbits, and observationally in humans (Morrison, 1929; Narath, 1951; Cuttino et al., 1978; Bidgood et al., 1981). Retrograde ureteral injection of contrast under pressure was seen to rupture the fornices, fill the interstitium and drain via the cortical and cortico-medullary lymphatics (Cuttino et al., 1978). This has been portrayed as further evidence of the ‘safety valve’ mechanism of renal lymph (Goodwin and Kaufman, 1956). As opposed to lymphatico-venous communications, this is not a direct anatomical connection (unless there is a pathological fistula, e.g., parasitic chyluria) but rather a two-stage process where fluid is moved first into the interstitium and then into the lymphatics (Goodwin and Kaufman, 1956; Threefoot et al., 1987). The importance of lymphatic outflow in hydronephrosis can also be seen where metastatic obstruction of both the ureter and the renal lymphatic outflow results in ureteric rupture (Fergusson, 1943). O’Morchoe and O’Morchoe (1988) argued against pyelolymphatic backflow acting as a ‘safety valve’ in hydronephrosis, citing the fact that renal hilar lymph and obstructed pelvic urine have very different compositions (Holmes et al., 1977b), and the difference in composition becomes greater with prolonged obstruction, which is the opposite to what would be predicted if there was a ‘safety valve’ effect.

Renal lymphatic dysfunction, defined as a failure in adequate drainage, leading to renal interstitial edema, would be expected to decrease renal function due to pressure changes within the encapsulated kidney (Mullens et al., 2017). Indeed, increased pressure within the renal capsule leads to a decreased GFR and renal perfusion (Jaffee et al., 2018). Feder and McDonald (1967) measured renal function after obstructing all hilar and capsular lymphatic in dogs. As expected, the kidney became swollen and tense, and urinary output and urinary sodium both increased. Surprisingly, they also found a slightly increased creatinine clearance and p-aminohippurate clearance in the obstructed kidney. They speculated that this is due to the autoregulation of renal blood flow, where increased tissue pressure causes compression of the renal vascular bed and subsequent reflex hyperemia.

O’Morchoe and O’Morchoe (1988) suggested there was no specific evidence to indicate a special interdependent relationship between renal function and the renal lymphatic system. They cited the practical problem of occluding renal hilar lymphatics without damage to renal nerves. They also highlighted the fact that most previous studies have simply investigated the flow and composition of renal lymph under various conditions (i.e., diuresis, ureteric obstruction, raised renal venous pressure), and suggested all changes could simply be attributable to altered hemodynamic effects. They concluded that the lymphatic system functions in the kidney as it does elsewhere in the body, to remove accumulated protein and fluid from the interstitial spaces, but it remains unknown as to what effect removing the lymphatic system has on renal function.

Bell and Parry (1974) also do not consider that the renal lymphatic system has a specific impact on renal function. In canine kidneys, they ligated all observable hilar lymphatics as well as the fatty tissue at both poles of the kidney to obstruct the capsular lymphatics. In contrast to the experiments described earlier, neither subcapsular pressure nor IRVP increased after lymphatic ligation. Furthermore, they repeated the experiment after partial obstruction of the renal vein and noted no difference between lymphatic ligation and controls when IRVP was raised. They concluded that renal lymphatic vessels are not primarily involved with regulation of intra-renal pressures. They conceded that lymphatic ligation leads to interstitial edema but suggested this may be more pronounced in the rabbit (where there is no capsular lymphatic communication) than the dog. The histological changes seen with lymphatic ligation, they suggest, are due to interstitial accumulation of toxic products of metabolism rather than pressure related changes. The findings of Bell and Parry seem to contradict other studies (e.g., Wilcox et al., 1984) and it is possible they were not able to ligate all lymphatics. Feder and McDonald (1967) found many accessory lymphatic channels that were previously unrecognized opened up after ligating all observable hilar and capsular lymphatics in dogs.

In studying the interplay between renal venous pressure, interstitial pressure and lymph flow, Rohn et al. (1996) devised a simple experimental model in dogs, where lymph flow (Q_L) was measured at various outflow pressures (P_O). From this they were able to calculate the single pressure source (P_L) pushing renal lymph through a single resistance (R_L). As expected they found a higher P_L when the renal venous pressure increased. Importantly they found a plateau in the Q_L versus P_O relationship, where there was no change in lymph flow as outflow pressure decreased from 5 to -5 cmH₂O. This plateau extended when venous pressure was increased (up to 27.2 cmH₂O), such that there was no change in lymph flow when outflow pressure decreased from 20 to -5 cmH₂O. Because of this extension in the Q_L vs. P_L plateau, they reasoned that renal interstitial pressure may partially collapse intra-renal collecting lymphatics (not initial capillary lymphatics which are tethered via anchoring filaments to the interstitium), which in turn may compromise lymph flow. This has the important implication of suggesting that beyond a critical point of interstitial edema, lymph flow will not increase and edema, and renal function, will significantly worsen. Interestingly the same plateau was seen in the lymphatic flow of the lung in the dog (Drake et al., 1982) and sheep (Drake et al., 1985) but the effect was less marked. This may be a reflection of the less distensible nature of the renal capsule. The other important finding of their paper is that increased pressure at the outflow end of renal lymphatics (mimicking raised systemic venous pressure), significantly decreases renal lymph flow, which may have implications for patients with acute and critical illness who have a raised central venous pressure.

Congestive heart failure (CHF) causes an increase in systemic venous pressure, which has multiple effects on renal lymph. Elevated venous pressures place a double burden on renal lymphatic vessels by increasing post-glomerular capillary filtration and opposing thoracic lymphatic drainage in the neck, which can cause backpressure effects, which decrease renal lymphatic outflow (Drake et al., 1998). As previously outlined, increased IRVP leads to increased renal lymphatic pressure and flow due to increased capillary filtration (Dumont et al., 1963). It is possible impaired lymphatic drainage in the neck occurs because central venous congestion causes a functional obstruction at the thoracic duct ostial valve, although this is unproven at present (Ratnayake et al., 2018). The excess demand on the lymphatics can lead to failure of adequate lymphatic drainage (lymphatic dysfunction). Cuttino et al. (1989), in studying renal lymphatic anatomy, found they had to use autopsy specimens from patients with CHF because the lymphatics had dilated and the valves become incompetent. This was the only way they were able to inject hilar lymphatics with contrast material in a retrograde manner and study the lymphatic anatomy using microradiographic techniques. Overall, this suggests renal lymphatic drainage may become compromised in CHF, through impaired outflow and valvular incompetence.

Renal lymphatics also contribute to sodium retention and worsening of generalized edema in CHF. Partial occlusion of the IVC (resembling CHF) increases renal lymph flow and Na⁺ content, and decreases urinary Na⁺ and flow in a reciprocal manner (Katz and Cockett, 1959). This mechanism is thought to contribute to sodium and fluid retention in CHF by recycling Na⁺ and fluid rather than excreting them in the urine. The increased renal lymphatic flow also washes out interstitial proteins, which decreases colloid osmotic pressure in the renal interstitium and further promotes sodium reabsorption (Mullens et al., 2017).

External drainage of lymph in the presence of an elevated central venous pressure has been shown to reduce edema formation, without reducing venous pressure in the gut (Drake et al., 1998). This suggests that external thoracic duct drainage may be of therapeutic benefit to the kidney (as well as other organs) in CHF (Dumont et al., 1963).

The clinical entities of AKI and renal interstitial edema are closely and reciprocally linked (Prowle et al., 2010). Renal interstitial edema can result from many factors, but is especially associated with renal lymphatic dysfunction (Figure 6). As well as being a consequence of AKI (due to renal inflammation and increased capillary permeability), renal interstitial edema can in itself lead to AKI. As the kidney is encapsulated, excess interstitial fluid leads to a disproportionate rise in intra-renal pressure and a subsequent decrease in renal blood flow and GFR (Firth et al., 1988). Renal decapsulation has been shown to reduce the incidence of AKI in patients with hemorrhagic shock requiring massive resuscitation (Stone and Fulenwider, 1977). Renal interstitial edema also impairs oxygen and metabolite diffusion by increasing diffusion distances (Prowle et al., 2010). Furthermore, in the encapsulated kidney, it may compress lymphatic drainage, commencing a viscous cycle that leads to worsening edema and renal function (see Relationship of Intra-renal Pressure to Renal Lymphatic Function). This all highlights the key role that renal lymphatics play in prevention of both AKI and renal interstitial edema.

Many studies have shown that renal venous congestion is also an important factor in the development of AKI, and even possibly more important than arterial insufficiency (Li et al., 2011; Jaffee et al., 2018). Typically, venous congestion results in an ischemic-type kidney injury (Jaffee et al., 2018). Renal venous congestion increases interstitial fluid and renal lymphatic flow (see Renal Lymph Flow During Increased Venous Pressure). Therefore, renal lymphatic dysfunction, where lymphatic drainage fails to meet demand for whatever reason, can be expected to worsen AKI in the setting of venous congestion (e.g., heart failure, aggressive fluid resuscitation, intra-abdominal hypertension) (Prowle et al., 2010).

As is the case for AKI, renal lymphatics have a complicated relationship to the prevention, pathogenesis and resolution of chronic kidney diseases (CKDs). Most, if not all, CKDs are characterized by tubulointerstitial inflammation and widening of the interstitial space through extracellular matrix deposition, leading in turn to fibrosis (Seeger et al., 2012). Renal lymphatics are thought to contribute to the development of fibrosis itself (Yazdani et al., 2014). CCL21 is constitutively expressed by renal LECs and can attract CCR7-positive fibrocytes. Blocking this pathway has been shown to reduce renal fibrosis (Sakai et al., 2006), suggesting a causative role of renal lymphatics in the development of CKDs.

Renal interstitial edema is seen in almost all CKDs, usually associated with tubulointerstitial damage and hypoproteinaemia secondary to proteinuria (Yazdani et al., 2014). Similar to AKIs, it can be considered a consequence of CKD, and one of the factors that promotes it. This is due to tissue architecture disruption, increased hydraulic conductivity and remodeling of the interstitial matrix (Yazdani et al., 2014). Therefore, lymphatic dysfunction can contribute to the development of CKDs by this mechanism. Consequently, lymphangiogenesis is activated in CKDs as an attempt at resolution (see Lymphangiogenesis). Further research is required to fully elucidate the roles of renal lymphatics in both AKI and CKD.

Following renal transplantation, where the lymphatics are inevitably divided, the kidney increases in size due to interstitial edema (Goodwin and Kaufman, 1956). Pedersen and Morris (1970) studied this phenomenon in sheep. They noticed lymph production increases in the post-transplant kidney, especially so in graft rejection, and the increase in kidney size could be avoided if renal lymph was drained artificially. Transplanted kidneys show active lymphangiogenesis in humans and rats, and LECs that actively organize inflammatory nodules through expression of the CCL21 chemokine (Yazdani et al., 2014). It is currently unknown whether lymphangiogenesis is detrimental or beneficial to the transplanted kidney and further research is needed in this area (Yazdani et al., 2014).

Newly formed lymph vessels (lymphangiogenesis) normally function as protection, helping to clear tissue edema and inflammatory infiltrates. Renal lymphangiogenesis is associated with renal fibrosis, inflammation and transplant rejection (Kerjaschki et al., 2004; Seeger et al., 2012; Yazdani et al., 2014). In CKDs, such as diabetic nephropathy, the degree of lymphangiogenesis appears to correlate with fibrosis more than inflammation (Sakamoto et al., 2009). In a rat proteinuric model, lymphangiogenesis occurred prior to macrophage influx, collagen deposition and interstitial fibrosis, suggesting a possible causal role in fibrosis (Yazdani et al., 2014). Lymphangiogenesis is also seen in rat models of hydronephrosis and hypertension (Suzuki et al., 2012; Kneedler et al., 2017). In humans, it is seen in end-stage renal failure and renal cell carcinoma (in tissue surrounding the tumor) (Ishikawa et al., 2006). It is also a feature of hydronephrosis in humans around interlobar and arcuate vessels, but not around interlobular vessels (Ishikawa et al., 2006). It is not seen in specimens of old or new renal infarction or acute tubular necrosis (Ishikawa et al., 2006). Therapies aimed at inducing lymphangiogenesis (e.g., 9-cis retinoic acid) have improved tissue swelling in animal models of lymphedema, and may have a role in treating renal interstitial edema as well (Choi et al., 2012; Yazdani et al., 2014). However, these therapies may potentially also be detrimental, as lymph vessels transport antigen-presenting cells to lymph nodes and LECs can initiate immune responses in and of themselves (e.g., following transplant) (Yazdani et al., 2014).

The long-term effects of ligating hilar lymph vessels in humans can be seen in renal transplantation, where the donor kidney’s lymphatic vessels are ligated (Ranghino et al., 2015). These effects are also seen following lymphatic ligation for the treatment of severe chyluria (Zhuo et al., 2013). It has been shown that the human kidney can survive long-term after lymphatic ligation, presumably due to new lymphatic connections. However, the true effects of this are not known. Zhang et al. (2007a,b) studied the effects of renal lymphatic ligation in rats over 8 weeks. They ligated both hilar and capsular lymphatic vessels and included animals where one kidney was ligated and the other removed to resemble a single transplanted kidney. They found severe proteinuria 1 week after ligation, and from 2 weeks, renal function began to deteriorate with elevated serum creatinine and reduced creatinine clearance. Histologically the kidneys showed tubular damage, tubulointerstitial fibrosis and expansion of the mesangium, possibly due to enhanced activation of the TGF-β1/Smad signaling pathway. These changes worsened over the 8-week study period. Renal lymphatic ligation has also been shown to cause enhanced renal cell apoptosis (Zhang et al., 2009; Cheng et al., 2013). Sclerosis of the renal parenchyma has also been seen in rats 3 months after ligation of the thoracic duct (Umbetov, 1989), although it is difficult to ascribe cause and effect in this situation. Renal interstitial fibrosis has been seen and thought to develop as a consequence of unabsorbed renal interstitial edema in a human retrospective long-term study (Wehrmann et al., 1989).

Primary lymphatic disorders are varied and the majority of reported cases do not have an identified mechanism. In most disorders, defects in relation to lymphatic valves are the cause of lymphatic dysfunction (Ballard et al., 2018). These patients generally present with bilateral lower limb lymphedema (e.g., Nonne-Milroy lymphedema, lymphedema praecox). There is a paucity of literature in regards to the renal effects of these various syndromes, but some are known to cause renal insufficiency and structural defects over time (Northup et al., 2003).