Measurement of glomerular filtration rate and renal blood flow

This chapter is relevant to the aims of Section H4(i) from the 2017 CICM Primary Syllabus, which expects the exam candidate to "describe the principles of measurement of glomerular filtration rate and renal blood flow". Question 3 from the second paper of 2016 is the only representative question for this syllabus item, and even there the GFR measurement section was worth only 30% of the marks, trivialising the importance of this topic for the exam-savvy candidate. In the interest of time, those mainly interested in packing their short term memory can limit their reading to the contents of the grey box below:

There is of course much more to this, and quite a large amount of ink has been spilt on it over the years, making it easy to come up with several good recommendations from official peer-reviewed literature. Gaspari et al (1997) is probably the clearest of these, but it is unfortunately rendered inaccessible by paywalls, and even with institution access it is difficult to get hold of because Kidney International Supplements don't seem to have a digitised copy. Levey & Inkler (2017) is the next best option. Another great resource is this 2011 dissertation by Margareta Gref. Additonally, for the reader with abundant time resources, just about anything by Homer Smith will also make an interesting read, even if it is a semifictional novella about hunting lungfish in Africa.

Measurement of glomerular filtration

To be clear, what's going to be discussed here is a series of measurement techniques, rather than the more commonly resorted-to estimates of glomerular filtration. Yes, on occasion, for a variety of reasons, it may become necessary to have a measurement here. For example it is important for the diagnosis of early kidney disease, where hyperfiltration or a rapid decline in GFR could be detected by direct measurement but not by estimates (Bjornstad et al, 2018).

Glomerular filtration is the work of turning glomerular blood flow into glomerular ultrafiltrate, i.e. the fluid which enters the proximal tubule. As such, in all honesty there is actually no way to measure the glomerular filtration rate directly in the living human organism, as it would require one to measure and add every drop of ultrafiltrate coming out of all the millions of individual glomeruli. Fortunately, we can determine this rate of ulrafiltrate formation by measuring the behaviour of other body fluids and various marker solutes. Specifically, you can measure it by:

• Renal (urinary) clearance of a marker solute, where the rate of urinary clearance is used as the surrogate for glomerular filtration
• Plasma clearance of a marker solute,  where the disappearance of a marker from the blood is used as a measure of glomerular filtration

Measurement of glomerular filtration by urinary clearance

When something is cleared from the blood, we imply that is has been eliminated from the body, and by the rate of its elimination we can infer something about the blood flow to the organ that does the eliminating. At a basic level, the principles of making that inference can be outlined in the following graphical manner:

In case this picture is not worth a thousand words, here is a pointform explanation of the same process:

• You give a known quantity of substance X intravenously
• You measure the concentration of substance X in the plasma, in mg/ml
• Some of this substance is filtered by the glomerulus
• That filtered amount of substance X therefore disappears from the circulation and appears in the urine
• If you measure the urine flow rate (volume per unit time) and the concentration of substance X in that urine (in mg/ml), you can figure out the amount of substance X which has been cleared during that unit of time (urine concentration multiplied by urine flow rate gives the amount in mg).
• The amount of substance X dumped into the urine can then be used to figure out what volume of blood got "cleared" of this substance, because you already know its concentration in the blood. Thus, if the urinary excretion is 100mg/min, and the blood concentration is 100mg/L, that means 1L of blood was cleared.
• If that substance is 100% cleared from the blood by the glomerulus, this volume of cleared blood therefore represents the blood flow though the glomerulus

This is usually represented as an equation, such as this:

CL_x  = (U_x × Q_u ) / P_x

where

• CL_x  is the clearance rate of substance X in ml/min
• U_x  is the urinary concentration of substance X
• Q_u is the urine flow rate, in ml/min
• P_x is the plasma concentration of substance X in mg/ml

In other words, to borrow a turn of phrase from Levey & Inker, 

"Clearance does not represent an actual volume; rather, it is a virtual volume of plasma that is completely cleared of the solute per unit of time"

Those au fait with pharmacokinetics will immediately recognise this equation and this concept because it is common to the discussion of drug clearance. So, which marker substance to use? In his 1935 PhD thesis, James Shannon outline the desirable properties of such a substance, and this list seems to have been definitive (in the sense that every subsequent author seems to have reproduced it with minimal modifications. Not to be outdone by official textbooks, the exact same list is quoted here:

Shannon and Homer Smith initially explored simple sugars such as xylose (a monosaccharide found in wood) and raffinose (the trisaccharide responsible for some of the flatulence associated with eating beans). These did not meet all of the abovestated criteria, for a variety of reasons (eg. it soon became clear that xylose is in fact metabolised), and the authors ultimately settled on inulin, a fructan polysaccharide originally found in the roots of Inula helenum. Being a polysaccharide (i.e. a rosary made from a random number of fructose molecules), its molecular weight can be a bit unpredictable, to the point where Westfall & Landis (1936) literally gave a range from 164 Da to 164 kDa, spanning three orders of magnitude. Obviously it would be pointless to inject people with impossibly large molecules which the glomerulus cannot filter, and apparently modern commercially available preparations all average at around 6.1 kDa.

Measurement of glomerular filtration rate by inulin clearance in the urine

Inulin clearance is the "gold standard" for glomerular filtration measurement, and it remains the reference method against which all other methods are measured, for example in the systematic review of GFR measurement methods by Soveri et al (2014). It has been that way ever since Shannon & Smith published on it in the 1930s. In fact, prior to that work, there was no prior use for inulin in human biological experiments, to the point that James Shannon had to inject himself with 160g of inulin (20% solution) as a means of demonstrating its safety.

Those first experiments consisted of giving the subject 100g of inulin as an infusion over 30 minutes and then measuring urine flow and inulin concentration over 10-20 minute time periods for the subsequent several hours (1-4, depending on the experiment). These days, the process is somewhat more standardised, although it is so rarely called for that even your huge quaternary institution probably does not have a protocol. In case a protocol is ever required, Jung et al (1992) published a description of a representative procedure that could act as a guideline for anybody who has to reinvent the wheel locally, or for some sort of experiment. It would probably only get dusted off for situations where somebody has developed an alternative method and is trying to validate it against a recognised benchmark.

Apart from being unable to get hold of sterile inulin for IV injection (nobody seems to stock it), what are the limitations of this technique? Well:

• Inulin needs to be injected into the patient (requiring IV access)
• The amount of inulin which needs to be injected is not trivial (it is about 100g); and as an osmotically active solute, this could have significant volume-expanding effects, which may not be tolerated by fragile little ventricles.
• The concentration of inulin needs to be maintained at a constant rate, i.e. a continuous infusion must carry on while urine collection is taking place
• The technology for measuring plasma and urine inulin concentrations is not available everywhere
• Theoretically, you could have an anaphylactic reaction to inulin.

Moreover, a major limitation of this technique is that - to be completely accurate - it measures inulin clearance, rather than the glomerular filtration rate. It's just that the two are thought to be closely related. So much so that inulin clearance is assigned the value of 1.0, and the clearance of other substances is indexed against this.

Measurement of glomerular filtration by creatinine clearance in the urine

Rather than injecting hundreds of grams of sugar into people, one might consider using some endogenously available solute for the purpose of GFR measurement. This bypasses the need to give an injection and greatly simplifies the process, albeit at the cost of decreased accuracy.

Creatinine should probably have some of the spotlight here, in the event that there is no chance to discuss it in any other chapter, and because upon closer questioning many people will not be able to articulate exactly what the hell it is.  In case anybody needs a detailed explanation of where creatine and creatinine fit into the overall picture of muscle metabolism, Wyss & Kaddurah-Daouk (2000) drag through the subject over eighty pages of grinding detail.  

 Creatinine is the breakdown product of creatine, a small lactam which is produced by the cyclocondensation of creatine. This is a nonenzymatic reaction which takes place in muscle under most normal conditions, and textbooks usually say that a normal person is expected to produce about 2g of creatinine per day, which is apparently about 2% of the total body creatine store. This small molecule is nontoxic, undergoes no clearance other than renal, and filters freely at the glomerulus because it is only 113 Da in mass. Unfortunately, it is not a completely "perfect" marker substance, because it is excreted actively by the pars recta of the proximal tubule, and therefore the amount of it which ends up in the urine is slightly larger than the amount filtered though the glomerulus, overestimating the GFR.

Just as the measurement of GFR with inulin, the measurement of GFR with creatinine involves the collection of urine over some time period  and the measurement of creatinine contained in that urine. The RCPA manual recommends a 24-hour urine collection. The creatinine concentration of that 24hr sample is then measured, and the creatinine clearance is calculated by the following formula:

CL_cr = (U_cr × V₂₄ × 16.7) / P_cr × T)

where:

• CL_cr  is the creatinine clearance in ml/min
• U_cr  is the urinary creatinine in mmol/L
• V₂₄  is the urine volume in ml, over 24 hrs
• 16.7 is the fudge factor required to convert the 24-hour result into ml/min
• P_cr  is the plasma creatinine concentration in mmol/L
• T is the duration of the urine collection (usually, 24 hrs)

The limitations of this technique are well described in the delightful article by R.B. Payne (1986), titled "Creatinine clearance: a redundant clinical investigation". In summary, they are as follows:

• Tubular secretion of creatinine contributes to urinary clearance (as mentioned above, creatinine is secreted by the proximal tubule). As the result, the GFR calculated by this method is about 10-20% higher than the actual GFR, or whatever ends up being measured by inulin excretion method. 
• Serum creatinine is not constant. It is assumed that the production of creatinine in the body occurs at a constant rate, but this is not necessarily so. Creatinine is a product of not only muscle energy metabolism but also protein breakdown, and in fact the rate of creatinine production can increase abruptly by the administration of exogenous protein. For example, Mayersohn et al (1983) were able to increase his subjects' serum creatinine by 50% simply by feeding them 225g of boiled beef.
• GFR itself is not constant. For example, a protein-rich meal can transiently increase the glomerular filtration rate. In the course of collecting urine for a creatinine clearance measurement, this needs to be taken into account (i.e. perhaps you have taken a sample during a period of uncharacteristically high or low GFR). One of the ways of controlling for this variability is to collect urine for a prolonged period (this is why a 24hr measurement is usually required).
• You need 24 hours of urine. The logistics of collecting, storing and processing litres of urine makes this an unpopular test. This also means your results are delayed by a minimum of 24 hours.

Measurement of glomerular filtration by plasma clearance

Some of us may have neither the time nor the patience to fuss with collecting urine for 24 hours. Fortunately there is a way to measure glomerular filtration by measuring the disappearance of the marker solute from the plasma, provided you can safely assume that marker has disappeared completely into the urine and was not sucked up into some metabolic pathway of alternative elimination route. 

This process, in the form of steps, would look like this:

• Once you have given a known amount of the marker solute, it will rapidly distribute to the circulating volume, and its concentration will initially drop.
• After this, the concentration will continue to fall at a slower rate, which will reflect its clearance by the kidneys (assuming the kidneys are the only organ doing the clearing). 
• This can be plotted along a concentration/time curve:
• Then, by calculating the area under this second β-elimination curve we can calculate the volume of blood cleared of the drug, using this equation:
  
• This gives you the clearance in ml/min or L/hr
• As long as the marker solute is close to being an "ideal" solute, this clearance rate is a good surrogate for the glomerular filtration rate.

One suitable marker for this is iohexol, known to many as Omnipaque. This iodinated benzenedicarboxamide is relatively inert, nontoxic (unless iodine allergy) and benign in relatively large doses (unless congestive heart failure). Gref & Karp (2007) describe the process of using this substance for measurement. They gave their patients 10ml of iohexol and then collected blood at 180, 225 and 270 minutes. Plasma samples were then tested for iohexol content using high-performance liquid chromatography. 

This technique also has its limitations. For one, you can never be completely sure that the substance you have given is not distributing to some weird fluid compartment outside of the circulatory system, in which case your GFR measurement will overestimate the GFR. And you need to be sure that it is not being reabsorbed by the renal tubule, because this would underestimate the GFR. Lastly, iodinated contrast media are not uniformly benign for the kidneys, to say nothing of the potential for anaphylaxis. Moreover, some people might justly point out that their department is not made of money and liquid chromatography columns don't just grow on trees. In short, the use of iohexol for the measurement of GFR does not have the widespread popularity that you might expect.

Iohexol is of course not the only possible marker solute. For example, one of the very first papers describing this method of GFR measurement used exogenous creatinine, making the argument that the equipment to measure it was already available (Sapirstein et al, 1955). However, these days most such measurements are performed using radiolabeled solutes, as discussed below.

Measurement of glomerular filtration by nuclear medicine

This, on close inspection, resembles the measurement of urinary or plasma clearance, but with the extra step of making the marker solute radioactive. The advantage here is that, instead of using super-expensive laboratory equipment to measure the concentration of a substance in the blood or urine, you can just wave a radioactivity detector over it, and calculate the concentration. It may be difficult to believe, but the bespoke isotopes and dosimeters used for this are somehow cheaper than analytical chemistry equipment.

Isotopes and compounds used in radionuclide measurement of GFR include the following list of substances. Rather than rave about their properties, veering into even greater irrelevance, this chapter will suffice to offer references in case somebody one day needs to know more.

• ⁵¹Cr-EDTA (Brändström et al, 1998)
• ⁹⁹Tc-DTPA (Russell et al, 1985)
• ¹⁶⁹Yb-DTPA  (Russell et al, 1985)
• ¹²⁵I iothalamate (Elwood et al, 1982)

There's probably many others but it would be unreasonable to expect anybody in ICU training to study them or to memorise long lists, as we are not ever expected to prescribe them. Apart from being radioactive, which is universally unattractive, these substances also have other non-charming features (for example the iodinated ones can harm the thyroid). 

Measurement of renal blood flow

Measurement of renal blood flow is related to, but not the same as, the measurement of glomerular filtration, given that the glomeruli do not let through 100% of the blood volume that passes through them (as that would be the opposite of filtration). The gold standard for this measurement would have to be some sort of horribly invasive direct technique, with catheters in renal arteries and veins, and that would be unpalatable to the public.   Instead, we usually resort to surrogate variables which are closely related to renal blood flow. One such measure is the renal plasma flow rate, which is related to renal blood flow by the same factor as plasma volume is related to blood volume. 

For this, you would need a marker which gets filtered readily and completely in the glomerulus, and which undergoes minimal reabsorption so that its concentration in post-glomerular blood is basically zero. Para-aminohippuric acid (PAH) is one such solute. This is a small organic acid which is rapidly and almost completely removed from the circulating blood volume by glomerular filtration and tubular secretion, such that almost none of it (well, only about 8%) makes it out of the renal microcirculation. This property is usually described as the renal extraction ratio; i.e. the renal extraction ratio of PAH is 0.92. Because PAH lives in the plasma compartment only (i.e not in the red cells), what is measured by this marker is actually renal plasma flow rather than renal blood flow, but there is no reason to be put off by this, because one can be inferred from the other.

The process of measuring renal blood flow using PAH was also pioneered by Homer Smith and his group in 1945. The modern version is described in broad brushstrokes in the FDA PI for PAH, and can be summarised as follows:

• For the duration of the procedure, you need to create and maintain a steady state concentration of PAH in your subject.
• To get there, you usually have to give 6-10mg/kg, and then run an infusion of 10-24 mg/min.
• The target concentration in the blood is about 20mg/L. 
• The plasma concentration is then measured
• Urine is collected over some timeframe, and the concentration of PAH is measured in the collected specimen.
• The renal plasma flow is then calculated from the following equation:
  
  eRPF = (U_PAH × V) / P_PAH 
  
  where 

  • eRPF is the estimated plasma blood flow,
  • U_PAH is the measured urinary concentration of PAH,
  • V is the flow of urine (i.e. volume of urine divided by the time over which it was collected), and
  • P_PAH is the plasma concentration of PAH.

• The renal plasma flow is said to be estimated because the PAH is not 100% cleared by a single pass (remember about the 8% that escape)
• The normal value is around 700ml/min (slightly more for boys, slightly less for girls).

As everything has its limitations, so does this method:

• If renal tubular secretion of PAH has decreased for some pathological reason (eg. recent ATN has damaged your tubules), this technique will underestimate the renal blood flow
• It depends on there being a steady state of plasma PAH concentration, which is not guaranteed
• It assumes nothing is competing with PAH for those organic anion transporters in the proximal tubule, which we know for a fact are always busy secreting other anions
• It assumes that the tubule is capable of excreting the delivered dose of PAH - but those transporters are saturable, which means larger concentrations of PAH will not be excreted and the technique will underestimate the renal plasma flow because the plasma concentration of PAH will be higher.