Noninvasive Control of Uremic Bleeding

We frequently receive requests to dialyze patients with moderate renal failure who are bleeding due to the concern that uremia may be contributing to failure to control blood loss.

Patients with renal failure do have an increased tendency to bleed, usually manifested by prolonged bleeding time. I was recently reminded by one of my colleagues in conference that while dialysis is an option for treatment of uremic bleeding, it is important to recall (ideally prior to placing the large and inflexible central line) that there are several noninvasive medical therapies for uremic bleeding, some of which work quite well and with minimal side effects.

Uremic bleeding is multifactorial, involving dysfunctional Von Willebrand factor, accumulation of many uremic toxins, particularly L-arginine, increased levels of cyclic AMP and cyclic GMP (cGMP) both of which reduce levels of thromboxane A2 (TxA2) and ADP, anemia, which causes platelets to travel midstream through the blood vessels, farther away from the endothelium. Red blood cells also scavenge nitric oxide and release ADP and TxA2. Lower levels of TxA2 and ADP lead to decreased platelet aggregation. Nitric oxide (NO) also plays a central role. L-arginine induces NO synthesis, which stimulates guanylyl cyclase, increasing cGMP levels, and thus decreasing TxA2 and ADP. TNF-alpha and IL 1beta are also increased in uremia, both of which induce NO synthase.

Treatment options for uremic bleeding include: erythropoetin, cryoprecipitate, desmopressin, conjugated estrogens (premarin, explaining the picture associated with the post), and dialysis. Various doses of erythropoetin have been studied, but the success is dependent not on the dosing regimen but the achievement of a hematocrit of 30%. Cryoprecipitate contains factor VIII, vWF, and fibrinogen, likely increasing the proportion of functional clotting factors in uremic patients' plasma. Desmopressin likely releases factor VIII from storage sites, increasing the concentration of factor VIII and minimizing the effect of dysfunctional vWF. The dose is 0.3 to 0.4mcg/kg IV or subcutaneously. It acts within an hour, but tachyphylaxis develops after one dose, likely due to depleted vWF from endothelial stores.

Estrogens can also safely and effectively improve bleeding time in uremic males and females. The dose needed is 0.6mg/kg IV conjugated estrogens over 30-40 minutes once daily for 5 days. The onset of action is about 6 hours, with maximum effect at 5-7 days and a duration of action of 14-21 days. The mechanism is thought to be reduction of L-arginine, which is a precursor to NO. This results in less guanylyl cyclase stimulation and less production of cGMP, leading to increased TxA2 and ADP and improved platelet aggregation. This theory is supported in rat studies. 17beta estradiol given to rats made uremic by reduction of renal mass significantly reduced bleeding time within 24 hours and normalizes plasma concentration of NO metabolites, nitrites and nitrates, and of NO synthase catalytic activity. Endothelial NOS and inducible NOS immunoperoxidase staining in the endothelium of uremic aortas of untreated rats was significantly more intense than in control rats. Uremic rats receiving 17beta estradiol had NOS staining comparable to controls. Estrogens have been used for patients with uremic GI bleeding, with some controversy and a negative RCT in angiodysplasia. Nate Hellman also talked about estrogens in GI bleeding here.

Nature Clinical Practice published a nice review on treatment recommendations for uremic bleeding, with a summary of recommendations and the strength of the evidence behind each recommendation, along with an algorithm for treatment of bleeding patients. Special thanks to Nirupama Ramkumar for finding and summarizing this article in clinical case conference.

Alimentary Azotemia Redux: A Quantitative Approach

The issue of whether a marked elevation in the BUN when compared with the creatinine might represent gastrointestinal bleeding was nicely covered previously on RFN. One of our attendings recently had our group of first year fellows review the issue using a quantitative approach that highlighting the relevant physiology.


Consider a 72kg male in steady state eating 90grams of protein per day with a creatinine clearance of 120ml/min and a Urea clearance of 60ml/min.


Remembering that a male will produce about 20mg/kg of creatinine a day, our 72kg male will produce 1440mg of creatinine in a day…


72kg x 20mg/kg = 1440mg


A person in steady state must excrete what they produce (a key nephrology concept). So if our man makes 1440mg of creatinine he must excrete 1440mg of creatinine (if he fails to excrete it all his plasma creatinine concentration will rise and he has fallen out of steady state).


We can additionally estimate the amount of BUN produced by remembering that urea nitrogen production is approximately 1/6th of protein intake. So our man eating 90grams of protein produces 15grams of urea nitrogen each day (90grams x 1/6 = 15grams) which in steady will be excreted.


With the above we can now calculate the plasma creatinine and BUN concentrations using the clearance equation…


clearance (C) = [urine concentration (U) x urine volume (V)] / plasma concentration (P)


C = UV/P


Plug in the numbers correcting the units along the way for Cr…


120ml/min = (1440mg/day) / P
P = (1440mg/day) / 120ml/min
P = (1440mg/day) / 172,800ml/day
P = 0.0083mg/ml
P = 0.83 mg/dl


Same deal for BUN…


60ml/min = (15g/day) / P
P = (15g/day) / 60ml/min
P = (15,000mg/day) / 86,400ml/day
P = 0.17mg/ml
P = 17 mg/dl


A final thing we can sort out from what was provided is the fractional excretion of urea which by convention is expressed in percent. This is just what it says it is, the fraction of filtered urea (we'll approximate GFR with CrCl) that gets excreted in the urine. As urea is freely filtered this is…


FeUr = (Urea clearance / GFR) * 100
FeUr = (Urea clearance / CrCl) * 100
FeUr = [(60 ml/min) / (120 ml/min)] * 100
FeUr = 50%


So here’s what we know in table form…



Now imagine that our man starts feeling unwell, stops eating and has a one liter bleed from a peptic ulcer into his GI tract. For arguments sake lets say this occurs with no drop GFR (the “it’s the blood not the renal function” argument).


His protein intake is now the protein content of 1L of blood. 40% is cells (mostly rbcs) and 60% is plasma. The major proteins in the cellular and plasma parts respectively are hemoglobin and albumin (there's a bit more protein around from globulins and so on but this will give us a rough estimate).


Normal hemoglobin and albumin concentrations would be 14 g/dl and 4 g/dl respectively. So from the above we can estimate the protein content of blood in the GI tract…


1L * 0.60 = plasma volume
0.6L = plasma volume


plasma volume * protein concentration = plasma protein content
0.6L * 4g/dl = plasma protein content
0.6L * 40g/L = plasma protein content
24g = plasma protein content


1L * 0.40 = cellular volume
0.4L = cellular volume


cellular volume * protein concentration = cellular protein content
0.4L * 14g/dl = cellular protein content
0.4L * 140g/L = cellular protein content
56g = cellular protein content


Total protein content = cellular protein content + plasma protein content
Total protein content = 24g + 56g
Total protein content = 80g


Using our previous calculations our table now looks like this…



Notice that in the above scenario the BUN drops a bit as the protein intake has decreased. What if we kept our man eating the same diet and had him bleed at the same time while holding kidney function stable?



If you almost double the protein intake you almost double the BUN (from 17 to 33 mg/dl). Now let’s try the stopped eating, 1 liter bleed scenario along with a 50% drop in GFR due to hypotension. Remember that in the volume depleted state the fractional excretion on urea is typically less than 35% and for arguments sake we’ll make it 20% in our man.



As compared with no renal dysfunction we now get an BUN/Cr ratio of 23 as compared to 18. How about we run scenario with continued eating, 1 liter bleed and now with 50% drop in GFR due to hypotension with the associated drop in urea clearance.



Pretty impressive, huh? With a bit of kidney dysfunction added into increased urea production we’ve now got a BUN/Cr ratio of 98 vs 40.


The point of all this is that the BUN and serum creatinine will vary based on:


1) Cr production
2) Cr clearance
3) BUN production
4) BUN clearance


The integration of these four things yields the BUN and serum creatinine values and the subsequent ratio between the two.


As noted by Ernest, the dog paper he reviewed and the math above the most impressive BUN/Cr ratio elevations are generated by a combination of increased urea nitrogen production and decreased urea clearance. The ratio is further accentuated by the proportionally greater drop in urea clearance vs creatinine clearance seen in volume depletion.

Alimentary Azotemia?

I am currently a clinical fellow, slogging through another consult month. As I am sure many of you have experienced, consult questions often come in batches. At least three times this week we have been called to see a patient with markedly elevated BUN out of proportion to creatinine with the team questioning uremia, need for renal replacement etc. In all three cases, there had been an associated drop in hematocrit, and so we suggested they consider an occult gastrointestinal bleed as a contributor to the azotemia. Having suggested the association several times this week, I decided to investigate the azotemia-GI bleed association further.

The association was first noted by Sanguinetti in 1934, and he postulated that the observation was due to the absorption of protein from the digestion of blood, subsequently dubbed alimentary azotemia.

Several other theories about the cause of this phenomenon exist, including:

- hemorrhagic shock causing pre-renal failure

- increased protein catabolism as a result of negative nitrogen balance

- a toxic effect of some molecule absorbed that led to renal dysfunction

- direct effects of anemia.

Some authors have been dogmatic and insist that azotemia from GI bleed would only occur in the presence of renal dysfunction, acute or chronic.

This interesting study attempted to tease out the relative contribution of each of these factors (http://archinte.highwire.org/cgi/content/summary/75/6/381). Their model involved bleeding dogs and then feeding them back the blood while varying the degree of hypotension and access to and resuscitation with fluid. They found that dogs fed blood or dietary protein (meat) developed a small degree of azotemia over a short duration of less than 12 hours. On the other hand, dogs bled to the point of hypotension were azotemic to a greater degree and for a longer period of time. When dogs were bled to the point of being hypotensive and then fed back the blood, they were more azotemic and for longer than when hypotensive or fed blood alone. Moreover, after feeding dogs blood without making them hypotensive, they found urea clearance to be unchanged, arguing against a nephrotoxic effect of the ingested blood. Finally, varying water intake, in the absence of hypotension, did not affect the subjects' BUN. Other investigators conducted similar studies in humans, with the investigators or their colleagues serving as subjects.

It's hard to draw conclusions from these studies, but it certainly seems that both intestinal reabsorption and impairment in renal function may play a role in azotemia associated with GI bleed. Further, I think few would disagree that while the cause of azotemia after GI bleed is likely "multifactorial," the finding of a patient who is acutely (or even subacutely) azotemic out of proportion to Cr with associated anemia out of proportion to any underlying CKD, should prompt the consideration of a GI bleed.