Sepsis-Associated AKI – Bellomo Kidney – Implications for Management
“Rather than love, than money, than fame, give me truth.”
A 56 year old man with non-ischemic cardiomyopathy [LVEF 40% and mitral regurgitation] is admitted with severe sepsis due to appendicitis. One month prior to admission, his outpatient cardiologist saw him and noted a dry weight of 88 kg. On admission to the ICU, the patient’s weight is 90 kg. He is with acute kidney injury with a very elevated BUN-to-Creatinine ratio, oliguria and mild hypotension. His ACE-inhibitor is held and his serum chemistry is interpreted as ‘pre-renal azotemia’ so he is given 3 litres of lactated ringers in addition to broad spectrum antimicrobials. 48 hours later, he is greatly improved with normalization of his renal function. His weight on transfer to the floor is 96 kg and he is dyspneic.
There is a paucity of data on the underlying mechanism of renal insufficiency in early sepsis. Often renal ischemia is invoked as an underlying etiology followed by intrinsic, tubular functional embarrassment. However, much of what is known about human septic kidney injury flows from post-mortem analysis which is fraught with selection bias.
Figure 1A: The normal glomerulus. MAP is the mean arterial pressure which is the pressure head for flow through the afferent arteriole. Pgc is the pressure in the glomerulus which drives GFR [the blue arrow] into Bowman's Capsule, into the nephron. PCT is proximal convoluted tubule.Additionally, it is usually exceptionally difficult to measure renal blood flow in humans during sepsis. However, this has been endeavoured in the past. The legendary Dr. Homer Smith attempted this in the 1940s at Bellevue Hospital in a remarkable study of healthy volunteers given ‘pyrogens’ – rendering the study subjects hypotensive and rather toxic whilst measuring renal hemodynamics. Interestingly, renal blood flow rose, despite there being kidney injury. Other, early human data mirror the results of Dr. Smith. But how might this work? How could the kidneys receive more flow, but still demonstrate indices consistent with ‘pre-renal azotemia’?
Recall from basic physiology that glomerular filtration rate [GFR] stems from the pressure generated within the glomerulus relative to the pressure within Bowman’s space [as well as the ‘leakiness’ of the glomerular membrane - see figure 1A].
Thus the pressure within the glomerulus is determined by the flow into the kidney and the relative resistances of the afferent to efferent arterioles. In response to a falling MAP, the efferent arteriole will constrict to maintain Pgc [see figure 1B]. This occurs, classically, in hypovolemic hypotension.
Figure 1B: Hypovolemic hypotension. MAP falls, thus flow into the glomerulus falls and Pgc falls. To maintain GFR, the efferent arteriolar resistance rises in response to angiotensin II. This defends GFR, but at a lower rate. The diminished filtrate results in increased urea nitrogen reabsorption [BUN] at the PCTThe picture is a little different in sepsis, however. For example, Pgc will rise if the afferent arteriole dilates [which will increase renal blood flow]. However, if the efferent arteriole dilates to a greater extent, renal blood flow will remain high, but the glomerular filtration pressure and therefore GFR will fall. The proximal tubule will see less filtrate, and BUN will be reabsorbed to a greater proportion such that the BUN-to-creatinine ratio is augmented [see figure 2].
An additional, more interesting, phenomenon is also brought to the forefront here. There is a possibility that renal resistance falls [i.e. renal conductance rises] in response to intra-renal shunting. There is an anatomical basis for this. Shunting within the kidneys is the passage of blood from the afferent arteriole directly to the efferent arteriole/vasa recta; that is, the glomerulus is bypassed. In sepsis, this may be mediated by nitric oxide, but there is scanning electron micrographic evidence of this effect. Thus renal blood flow may intensify, but if it is not passing through the glomerulus, GFR will fall, and the kidneys will behave in a 'pre-renal' fashion.
Figure 2: Sepsis-associated AKI, note there is no increase in efferent resistance despite the falling MAP, additionally, there is shunting from afferent to efferent arteriole. The result is that the Pgc falls and so too does GFR, even in the face of preserved - or increased - renal blood flow. The PCT will reabsorb urea nitrogen & increase the BUN:Creatinine ratio.
Renal Flow-Function Mismatch
Very recent data in an ovine model of sepsis echoes the aforementioned. Notably, this comes from Dr. Rinaldo Bellomo’s group in Australia and reveals that in the first 48 hours of sepsis, renal blood flow does not fall in the face of diminished mean arterial pressure. Thus, renal resistance attenuates [or conductance rises]. Despite maintained renal blood flow, there is an abrupt loss of renal function with anuria. Importantly, renal biopsy during this period of impaired function, did not reveal histological changes consistent with cellular death and destruction. In fact, a blinded pathologist found no significant changes in underlying renal histology save for an enlarged glomerular mesangium of uncertain significance.
Given these ovine findings, ostensibly, early septic renal injury is, predominantly, a micro-vascular redistribution of blood flow and not ischemic or histological devastation. Early investigation implied that the increase in renal blood flow occurred both in the renal cortex and the medulla; though more recent data suggest a redistribution of blood flow away from the medulla.
If the underlying abnormality in early, septic renal dysfunction is microvascular, it calls for a reappraisal of macrovascular, goal-directed therapies. Rather than giving excessive, chloride-containing crystalloids, it may be of merit to focus on the pressure within the glomerulus [Pgc]. Raising mean arterial pressure with norepinephrine should help increase Pgc, but also avoiding things that constrict the afferent arteriole [e.g. NSAIDs, excessive chloride]. Further, there may be benefit to increasing the resistance of the efferent arteriole; this too would raise Pgc. Means to accomplish this are the administration of vasopressin or intravenous angiotensin II – which has been recently evaluated. In the recently-published VANISH Trial - which compared vasopressin to norepinephrine in early septic shock - vasopressin was associated with a 10% absolute decrease in the need for renal replacement therapy.
For more information on management, please read this excellent post by Josh Farkas on this topic [which includes a link to Dr. Bellomo’s must-watch grand rounds on his data] and Josh's excellent review of VANISH.
Return to the Case
On presentation, our patient may not have been hypovolemic, in fact, he was probably slightly hypervolemic. His kidneys were likely hyperemic, but with diminished Pgc, and therefore an attenuated GFR and pre-renal azotemia on his chemistry. This underscores the important fact that a 'pre-renal' BUN-to-creatinine ratio can be seen in a patient of any volume status. The onus is on the clinician to determine if the patient is hypovolemic or hypervolemic from the history and physical examination [nor can the clinician determine volume status from the ultrasonographic appearance of the IVC!]
Given the patient’s high renal blood flow yet low-glomerular capillary pressure, his diminished GFR should be viewed as a consequence of renal microvascular flow redistribution; additionally, he likely has little or no gross abnormalities in renal histology. The lactated ringers that he received increased his MAP and, therefore, GFR, but simply treating his sepsis and holding his ACE-inhibitor also increased Pgc and GFR.
Following resolution of renal septic physiology, he required active diuresis with intravenous furosemide and went to the operating room for successful appendectomy.