Diuretics are a bit like vasopressors. We use them all the time, so we feel that we know them pretty well. However, the amount of RCT-level evidence on them is shockingly low. So, whenever a RCT on diuretics shows up, it’s worth careful examination.
the 3T trial: basics
This is a single-center, double-blind trial involving 60 patients admitted with acute heart failure and found to be resistant to high-dose IV furosemide.1 Patients were randomized into three groups: oral metolazone, intravenous chlorothiazide, or oral tolvaptan. All patients continued to receive a titrated furosemide infusion. The primary outcome was 48-hour weight loss. The study was funded by Otsuka Pharmaceuticals, manufacturers of tolvaptan.
Inclusion criteria required heart failure with volume overload, which was determined by one these two criteria:
- Swan-Ganz catheterization with wedge pressure >19 mm plus hypervolemia on physical examination (peripheral edema, ascites, or rales on auscultation)
- At least two of the following:
- Peripheral edema
- Jugular venous pressure >10 mm
- Pulmonary edema on chest x-ray
Additionally, patients were required to be diuretic resistant. This was defined as having a urine output of <2 liters over 12 hours, while on a dose of furosemide >240 mg/day (or equivalent dose of bumetanide).
Exclusion criteria included:
- Need for dialysis or ultrafiltration
- Glomerular filtration rate <15 ml/min
- Systolic blood pressure <85 mm
- Potassium <3 mEq/L
- Sodium outside the range of 130-145 mEq/L
- Advanced liver disease
- Use of strong CYP3A4 inducers or inhibitors (which may interact with tolvaptan)
Baseline characteristics are shown above. Patients were indeed very diuretic resistant (producing an average of 1 liter of urine in the 12 hours prior to inclusion, despite receiving a whopping ~600 mg/day furosemide). Most patients had severe systolic heart failure, often due to coronary artery disease. 93% of patients were on loop diuretics prior to admission, which may partially explain their degree of diuretic resistance.
Patients were randomized to receive oral metolazone (5 mg PO BID), IV chlorothiazide (500 mg IV BID), or tolvaptan (30 mg daily). The use of placebo tablets and infusions allowed the study to be double-blinded (despite the use of both oral and intravenous medications).
In addition to study medication, patients received aggressive doses of furosemide, according to the below protocol. The goal urine output was 3-5 liters over 24 hours.
primary endpoint: 48-hour weight loss
The primary endpoint was weight loss over 48 hours (measured using a standing scale). This was equivalent across all groups, with an average loss of about 5 kg (below figure, left panel).
This primary endpoint has massive limitations, however. There are extremely different ways that patients can lose weight, and these have different clinical consequences.
As a thought experiment, let’s imagine starting out with an 80-kg patient, who might have the following fluid volumes:
First, let’s imagine that we remove five liters of electrolyte-free water (top panel, below). Water is freely distributed throughout all three compartments, so this will be removed proportionally from all three components. Removing electrolyte-free water therefore mostly reduces the volume of the intra-cellular compartment, with relatively little decrease in the interstitial fluid and plasma (plasma volume only decreases by 400 ml).
Now, let’s imagine removing five liters of isotonic fluid from the body (for example, via hemodialysis). Removal of isotonic fluid will drain fluid out of the extracellular space (bottom panel). This will result in much greater reductions in the interstitial and plasma volumes.
So, it’s clear that removing the same volume of fluid may have different clinical consequences, depending on how it’s removed. In heart failure, the real problem is excessive interstitial and plasma volume – so what we really want to do is remove isotonic fluid. Alternatively, removal of free water will cause relatively little reduction in plasma and interstitial volumes, so this will cause much less clinical decongestion.
In short, using total weight change as a primary endpoint will make it easier for tolvaptan to look good. Vaptans are great at causing water loss, so they may be successful at inducing a large volume reduction due to free water excretion. However, there are two major clinical problems with this:
- Free water removal may cause less decongestion (as explained above).
- As soon as vaptan therapy is discontinued, there may be water retention, which will erase any volume losses due to elimination of free water (more on this below).
uncontrolled water loss due to tolvaptan
Let’s take a closer look at the right side of that panel, which shows the cumulative urine output over 48 hours. There is enormous variation in the urine output among patients treated with vaptans (with an inter-quartile range spanning from 8 liters to 15 liters).
Of course, this is what we expect with vaptan therapy. Vaptans induce a state of nephrogenic diabetes insipidus, leading to uncontrolled water loss. The results are unpredictable. Some patients lost moderate amounts of water, but others lost almost a third of their total body water.
Massive water shifts aren’t safe. If close attention isn’t paid to this, it may cause hypernatremia and potentially even cerebral demyelination. In this study, it seems that the treating clinicians gave back most of the lost water (so that the net fluid change in the vaptan group was ~5 liters). Or maybe patients got insanely thirsty and cheated on their fluid restriction by drinking water surreptitiously. So these patients did OK. But in a situation where free water wasn’t returned to the patients, suddenly losing 15 liters of water would not be awesome.
clinical secondary endpoints
Secondary endpoints were generally similar between oral metolazone versus intravenous chlorothiazide (table below). IV chlorothiazide might have been a wee bit more powerful, with a greater achievement of decongestion and higher rates of hyponatremia. However, on the whole, the outcomes of the metolazone and chlorothiazide groups were very similar.
Patients did require supplemental potassium and magnesium, but these requirements weren’t very high. On average, the cumulative supplementation was ~80 mEq of potassium and ~1 gram of magnesium. This is somewhat reassuring, indicating that a combined nephron blockade with a loop diuretic and thiazide can be used without causing enormous electrolyte losses. However, 45% of the patients were on an aldosterone inhibitor, which likely minimized these losses.
Now let’s take a look at differences between thiazides vs. tolvaptan. First, remember the different physiologic effects of these drugs:
- Tolvaptan causes free water excretion, which removes volume mostly from the intracellular volume.
- Thiazides promote loss of NaCl and water, which removes volume mostly from the extracellular space.
This explains the observed differences perfectly:
- Tolvaptan increases serum sodium due to free water loss. Thiazides don’t.
- Thiazides cause a greater contraction alkalosis (due to preferential loss of NaCl).
- Thiazides cause a greater increase in creatinine and BUN, with more reduction in the glomerular filtration rate (because they’re decreasing the intravascular volume more).
At first blush, it might seem that an increase in creatinine and BUN in the thiazide group might be a bad thing (they’re hurting the kidneys!). However, this could just be a signal that thiazides are causing more effective vascular decongestion. Some studies have found that creatinine rise correlates with improved outcomes, perhaps since this reflects effective diuresis.2
There’s another key bit of information hiding here. After discontinuing tolvaptan, the serum sodium abruptly drops below baseline (aqua arrow above). In heart failure, this can mean only one thing – after the tolvaptan wore off, patients avidly retained free water (rapidly regaining all the weight lost following tolvaptan use). Unfortunately, the study didn’t report weight loss at the time of discharge (my guess is that these would be superior for thiazides than for tolvaptan).
Overall adverse event rates were similar. Two patients in the tolvaptan group experienced at least 12 mM increases in sodium over 24 hours, producing a potential risk of osmotic demyelination. This figure may be artificially low, however, as most current studies use >10 mM or even >8 mM as the threshold for osmotic demyelination risk.
The study wasn’t technically powered for noninferiority analysis (it was designed as a superiority trial). The error margins are wide, which could have caused the study to miss small differences between diuretics. However, it’s doubtful that the study missed clinically relevant differences.
About half of patients in the study were on a mineralocorticoid receptor antagonist (e.g. spironolactone). It’s unclear to what extent this may have affected outcomes.
conclusions on tolvaptan as a diuretic
Vaptans are drugs in search of an indication. They are extremely expensive, leading to vigorous industry support (including the funding of this trial). They’ve been aggressively promoted for use in hyponatremia, but they really don’t work well for that (certainly not in the ICU). Perhaps they might fare better in heart failure?
This study re-demonstrates the physiology of vaptans: they induce aggressive and often uncontrolled loss of free water. This action is probably not beneficial in heart failure, for three reasons:
- Uncontrolled excretion of water may cause rapid elevation of serum sodium levels, creating a risk of osmotic demyelination.
- Removal of free water predominantly causes intracellular dehydration, rather than extracellular decongestion.
- As soon as vaptan therapy is discontinued, patients will rapidly retain water and regain lost volume.
parting thoughts on metolazone vs. chlorothiazide
Chlorothiazide is more expensive than metolazone and it's administered intravenously, so chlorothiazide seems like a more dramatic intervention. This often leads to the assumption that chlorothiazide must be superior.
Of course, that's not necessarily true. Metolazone has an ace up it's own sleeve – a longer half life (~14 hours, compared to chlorothiazide's measly 2 hours). Metolazone is the glargine of diuretics: quietly sticking around, it diligently combats sodium retention after all the other diuretics have left. Metolazone's long half-life could be especially useful in patients being treated with intermittent boluses of furosemide. For example:
- A regimen of IV furosemide boluses plus IV chlorothiazide together Q12hr could expose patients to several hours without any diuretic on board.
- A regimen of IV furosemide boluses plus oral metolazone Q12hr would not expose patients to any diuretic-free period.
Ultimately, IV chlorothiazide and oral metolazone are both terrific diuretics, but they may be best suited for different purposes. IV chlorothiazide is great in emergent situations where immediate action is needed (e.g. emergent hyperkalemia, as part of the Nephron Bomb). Metolazone may be better for deresuscitation, if your goal is to gently remove several liters of fluid over the course of 24 hours. The good news from this study is that they both seem to be safe and effective.
- Thiazide diuretics are effective in relieving resistance to loop diuretics. This supports the traditional place of thiazides as second-line agents for patients who are refractory to loop diuretics.
- Oral metolazone appears to be equally effective as intravenous chlorothiazide. In practice, metolazone may have an advantage due to reduced cost and longer half-life.
- Aggressive combined diuresis with a thiazide plus loop diuretic was well tolerated (without sustained changes in renal function or major electrolyte shifts). Potassium and magnesium supplementation were required, but not in enormous quantities.
- Tolvaptan does result in excretion of large volumes of water. Based narrowly on this single metric, tolvaptan might be considered a success. However, tolvaptan doesn’t seem ready for routine clinical use for several reasons (e.g. uncontrolled water removal and rapid rebound water retention after discontinuation).
- 1.Cox ZL, Hung R, Lenihan DJ, Testani JM. Diuretic Strategies for Loop Diuretic Resistance in Acute Heart Failure. JACC: Heart Failure. December 2019. doi:10.1016/j.jchf.2019.09.012
- 2.Griffin M, Rao VS, Fleming J, et al. Effect on Survival of Concurrent Hemoconcentration and Increase in Creatinine During Treatment of Acute Decompensated Heart Failure. The American Journal of Cardiology. December 2019:1707-1711. doi:10.1016/j.amjcard.2019.08.034
- PulmCrit Theoretical Post – The COVID Severity Index (CSI 1.0) - April 2, 2020
- PulmCrit wee – Why the SCCM/AARC/ASA/APSF/AACN/CHEST joint statement on split ventilators is wrong. - March 29, 2020
- PulmCrit- Is Lopinavir/Ritonavir down and out? - March 19, 2020