heart failure: torsemide vs furosemide

 A recent open-label, pragmatic randomized trial of patients with heart failure found no difference if using torsemide versus furosemide as the loop diuretic (see chf torsemide vs furos JAMA Card2023 in dropbox, or https://jamanetwork.com/journals/jama/fullarticle/2800428?guestAccessKey=5afad846-ee09-457e-83c5-1cef2f98e0ad&utm_source=silverchair&utm_medium=email&utm_campaign=article_alert-jamacardiology&utm_content=olf&utm_term=012523 )

 

Details: 

-- 2859 participants, all hospitalized with heart failure at 60 hospitals in the United States, were recruited from June 2018 through March 2022 

   -- they had either newly diagnosed heart failure (30% of them), the rest had chronic heart failure and were hospitalized for heart failure in the past year

-- mean age 64, 35% female, 58% white/33% Black/5% Hispanic 

-- left ventricular ejection fraction>50% in 24% of the patients/  <40% in 70%/41-49% in 6% (but the latter had only 81 people) 

-- loop diuretic prior to randomization: furosemide 53%/torsemide 10%/ bumetanide 5% 

-- comorbidities: atrial fibrillation/flutter 44%, diabetes 48%, chronic kidney disease 35% (eGFR 59) 

-- systolic blood pressure 118 mmHg, heart rate 81, BMI 32 

-- baseline NT-proBNP 3994, BNP 982 

-- baseline meds: beta blocker 80%, ACE/ARB 45%, mineralocorticoid receptor antagonist (MRA) 37%, sacubitril-valsartan 18%, SGLT-2 inhibitor 6%

-- in this pragmatic study, patients were then randomly assigned to furosemide vs torsemide in a 1:1 ratio (at baseline, furosemide was used in 53%, torsemide in 10%)

 

-- Main outcome: all-cause mortality 

-- Secondary outcomes: all-cause mortality, all-cause hospitalization assessed over 12 months, or a combination of these two (there were also two other outcomes dealing with quality-of-life, to be published later) 

 

-- median follow-up 17 months 

 

Results: 

-- both groups were on roughly 80 mg equivalent of loop diuretic (this is assuming a 2:1 conversion for furosemide to torsemide, though this is not rigorously established as an equivalency) 

-- crossover therapy (clinicians or patients decided to change the medicine by the time of hospital discharge: 

    -- 96 patients (7%) switched from torsemide to furosemide; 53 patients (3.8%) switched from furosemide to torsemide [ie, twice as many swithced off of the torsemide]

 

-- all-cause mortality: 

    -- torsemide: 373 of 1431 patients (26.1%) 

    -- furosemide: 374 of 1428 patients (26.2%) 

        -- no statistically significant difference, with HR 1.02 (0.89-1.18), p=0.76 

 

-- all-cause mortality plus all-cause hospitalization:

    -- torsemide: 677 patients (47.3%) 

    -- furosemide: 704 patients (49.3%) 

        -- a nonsignificant, but strong trend to torsemide benefit by intention-to-treat analysis, HR 0.92 (0.83-1.02), p=0.11 

        -- analysis by patients who were actually on the prescribed treatment, a somewhat stronger trend, HR 0.91 (0.81-1.01), p=0.082 

 

-- subgroup analyses: 

    -- no statistically significant difference was found in different patient groups: stratifed by age, sex, race/ethnicity, left ventricular ejection fraction, loop diuretic prior to the hospitalization, NYHA (New York Heart Association heart failure class), systolic blood pressure, eGFR, diabetes, use of MRA at randomization, being in a teaching hospital, or duration of heart failure (worsening of chronic heart failure or newly diagnosed) 

  

Commentary: 

-- this pragmatic trial did not demonstrate any difference in all-cause mortality (the primary outcome) or in all-cause hospitalizations or in the combination of these two (though there was a pretty strong but not statisticially significant trend to torsemide superiority in the secondary analysis). 

-- this ambitious trial was set to identify significant differences between these two loop diuretics, given that in many different prior studies (which were largely observational) torsemide seemed to be a better choice 

    -- as compared to furosemide, torsemide has increased bioavailability and a longer half-life (3.5 hours versus 1 to 2 hours). 

    -- as opposed to furosemide, torsemide reduces aldosterone production, decreases sympathetic activation, reduces myocardial fibrosis, improves ventricular remodeling, and reduces blood natriuretic peptides (all of which seem to be a good thing...)

 

Limitations: 

-- there are huge pluses and minuses of pragmatic trials, as noted in the recent blog http://gmodestmedblogs.blogspot.com/2023/01/hypertension-hctz-vs-chlorthalidone.html. A few points in particular:

    -- the big plus is that a pragmatic trial is a ”real world” trial of patients actually being seen in clinical practice, is easily done with access to a good database through an electronic medical record, can involve a large network of providers and patients and reflect more diversity of both groups, is not based on patients willing to participate in a structured randomized controlled trial, and does not have the many accoutrements of an RCT (e.g. study nurses etc who tracked the patient and offer support and guidance) which is not available in usual practice settings

    -- but there are big minuses as well:

        -- several of them are noted by the authors, for example:

            -- “interpretation of [the] findings is limited by loss to follow-up and participant crossover and nonadherence”

            -- the targeted sample size was 6000 participants, but this was decreased to 2859 because the number of the primary endpoint of all-cause mortality was achieved at this lower number, but this might decrease the statistical validity of all of the secondary analyses assessed

                -- and this all-cause mortality outcome may have been imprecise and not truly reflect the heart failure specific outcomes associated with these two different loop diuretics

            -- the study design did not allow for assessment of benefits or harms from these therapies, including kidney function, electrolyte abnormalities, and non-hospitalization events (e.g. emergency department visits, outpatient intravenous diuretics, thiazide use). And, these events could affect the results

            -- loop diuretic dose was at the discretion of the clinician, and there might have been inherent biases by the physician and perhaps transmitted to the patient related to this;  this might have been reflected in the fact that more patients shifted medications from torsemide to furosemide in this open-label study (ie, everyone knew what they were taking)

        -- and, I would add the very likely issue of placebo/nocebo effects:

            -- the placebo effect can be quite powerful and is likely playing a significant effect in this trial, since patients who were already on furosemide (the vast majority were treated with furosemide beforehand) and were used to/comfortable with that medication: these patients were potentially on furosemide for a long time, stayed on the furosemide because there were no real problems with it, then half of them were randomized to the new medication torsemide and were more likely to have its attendant nocebo effect (i.e., the patient being more skeptical and likely having more adverse effects from the new medication).  

                -- of tangential interest to the placebo effect: it might well have a significant genetic component, the “placebome”  see http://gmodestmedblogs.blogspot.com/2015/04/placebo-genetics-and-placebome.html 

-- at baseline, of the 70% who had heart failure prior to this study, these patients were not on what we might consider optimal medical therapy at that point, including very few on sacubitril-valsartan (18%), which seems to have been the best choice at the time of this study, as found in the PARADIGM-HF trial of 2014 

-- we do not know how non-loop-diuretic medications were changed during the course of the study: including sacubitril-valsartan, beta blockers, SGLT-2 inhibitors, and MRA antagonists. All of these changes would likely affect clinical outcomes  and there might have been significant differences between the groups in terms of their heart failure (since this was not a randomized trial)

-- the median follow-up of 17 months may be inadequate to assess their outcomes, especially with the primary target being mortality, and life expectancy with heart failure is significantly better than 17 months with our current medications, though we do not know how sick these patients were with their heart failure, including their functional status or NYHA class on therapy (they did comment on a subgroup analysis of this, but no mention of this in the paper or supplement of this stratification), or their combinations of their comorbidities

-- also, 7% of the patients either were not discharged from the hospital on a loop diuretic or had unknown status, which may have diminished the statistical validity of the study; also there was an unusually high percent of patients who were not taking loop diuretics at discharge from the hospital (3%) 

-- and, this type of data-mining study does not allow for clear causal conclusions, just an association between the diuretic and outcome (since there were many unknown variables, as noted above, but also there could be unexpected/unmeasured other baseline confounders, which a prospective RCT would randomize out

 

So, how should we integrate the results of the study into our practice?? After all, this study did not find any benefit of torsemide over furosemide (though there was this pretty strong trend to benefit for the secondary outcomes, but not statistically significant), and many clinicians are likely more comfortable with furosemide (it is by far the number one loop diuretic prescribed). So it is totally reasonable to continue with furosemide based on this study. 

-- I personally changed to using torsemide as the loop diuretic of choice for those with heart failure in all of my patients, based on the consistent prior studies over the past 10+ years that were largely observational 

    -- which does bring up the well-known issue of the need for real randomized controlled trials as opposed to observational studies, which cannot definitively answer the question. And, based on impressive observational studies, I personally have been swayed incorrectly by the results of observational studies (eg homocysteine as a target to decrease cardiovascular disease, or estrogen replacement therapy, as just 2 examples)

-- in this case, I do feel there were a lot of unknowns that could have affected the results as mentioned above: 

    -- the fact that we don’t know medication changes after the initial hospitalization (were one group of clinicians more aggressive at prescribing more evidence-based treatments) 

    -- the study halved the number of patients assigned to each therapy because of higher mortality than anticipated, and this might have decreased the statistical ability to show differences in other outcomes (and these were almost statistically significant)

    -- the fact that they used all-cause hospitalizations and mortality as their outcomes instead of more heart failure-specific outcomes (one would likely need a larger cohort followed for longer to have these more specific outcomes); and we do not know what the reasons for the all-cause hospitalizations or mortality were in the study, which may have had nothing to do with their heart failure or treatment

    -- the mechanisms of action of torsemide versus furosemide certainly favor torsemide (less myocardial fibrosis, less left ventricular remodeling, lower sympathetic hormones and aldosterone, lower BNP levels)

-- so, i personally will keep on my merry trail of using torsemide..... (i can think of no down-side for this)

geoff

-----------------------------------

If you would like to be on the regular email list for upcoming blogs, please contact me at gmodest@bidmc.harvard.edu

  

to get access to all of the blogs (2 options):

1. go to https://www.bucommunitymedicine.org/ , a website from the Community Medicine section at Boston Medical Center.  This site does have a very searchable and accessible list of my blogs and is the easiest to view blogs and displays more at a time.

2. go to http://gmodestmedblogs.blogspot.com/ to see the blogs in reverse chronological order

  -- click on 3 parallel lines top left, if you want to see blogs by category, then click on "labels" and choose a category​

  -- or you can just click on the magnifying glass on top right, then type in a name in the search box and get all the blogs with that name in them

please feel free to circulate this to others. also, if you send me their emails, i can add them to the list