#NephJC Chat
Tuesday November 10 at 9 pm Eastern
Wednesday November 11 at 9 pm Indian Standard Time
Wednesday November 11 at 9pm GMT
NEJM October 23, 2020 DOI: 10.1056/NEJMoa2025845
Effect of Finerenone on Chronic Kidney Disease Outcomes in Type 2 Diabetes
George L. Bakris, M.D., Rajiv Agarwal, M.D., Stefan D. Anker, M.D., Ph.D., Bertram Pitt, M.D., Luis M. Ruilope, M.D., Peter Rossing, M.D., Peter Kolkhof, Ph.D., Christina Nowack, M.D., Patrick Schloemer, Ph.D., Amer Joseph, M.B., B.S., and Gerasimos Filippatos, M.D. for the FIDELIO-DKD Investigators
NEJM Link
Introduction
Diabetes is exploding!
“Diabetes is one of the fastest growing health challenges of the 21st century, with the number of adults living with diabetes having more than tripled over the past 20 years.”
(International Diabetes Federation, IDF diabetes Atlas, ninth edition 2019)
The global pandemic of diabetes is here to stay, and here to grow. Type II diabetes accounts for about 90% of the 460 million people with diabetes in the world. About half of these patients will develop kidney disease. No wonder, that diabetic kidney disease is the most frequent cause of end-stage kidney failure worldwide (approximately 25-50% of all patients on dialysis have diabetic kidney disease). Interventions to slow down the progression of diabetic kidney involvement are worth their weight in gold!
As Narva and Norton discuss in an editorial in the September issue of JASN, early detection of kidney involvement in diabetes is only going to be useful if we also promote better management (and prevention of progression) once diabetic kidney disease (DKD) is detected. Among the interventions to slow the progression of kidney disease in diabetes, few have been as successful as inhibition of the renin-angiotensin-aldosterone system (RAAS). Major clinical trials using ACE inhibitors (ACEi) or angiotensin receptor blockers (ARB) have yielded positive results that have become the cornerstone of therapeutic approaches to diabetic kidney disease. These include the Collaborative study (captopril), RENAAL (losartan), IRMA (irbesartan) and the IDNT studies (irbesartan). These agents are useful in diabetic kidney disease because they abrogate the rise in intra-glomerular pressure, the release of aldosterone (and subsequent effects of fluid and sodium retention) and activation of the sympathetic nervous system - all consequences of activation of the RAAS seen in diabetes.
The inhibition of multiple steps in the RAAS pathway has been attractive to researchers trying to maximise the protective effects of RAAS inhibition. Previous attempts at ‘dual’ inhibition have not been particularly successful - one still remembers the problems with combined ACEi and ARB treatment or the trials with the addition of renin inhibitors to usual care. The FIDELIO-DKD trial represents another attempt at this treatment, but this time chooses a ‘downstream’ target - the mineralocorticoid receptor (MR), which is the receptor for aldosterone, involved in the final, effector steps of the RAAS pathway.
Mechanisms of cardiac and Kidney damage induced by Aldosterone excess -
The Mineralocorticoid receptor and antagonists
The MR is part of a sub-family of steroid receptors, whose other members are the androgen, glucocorticoid, progesterone and estrogen receptors. MRs are involved in fluid / electrolyte balance and ‘hemodynamic homeostasis’, of course, and often in concert with glucocorticoid receptors, which are also found in similar tissues. The original MR antagonist (MRA) was spironolactone, a therapeutic agent since 1960. Over the past 60 years, multiple benefits of the MRAs have been described - see below for a timeline of the history of MR antagonism and clinical trials.
(Source: Agarwal R, et al EHJ, 2020)
The beneficial effects of spironolactone and eplenerone (both steroidal MRAs) in patients with heart disease - particularly HFrEF - are now well established. Inhibition of the MR could protect against injury in the kidney as well: in a mouse model of glomerulonephritis, knockout of the MR in myeloid (inflammatory) cells conferred renoprotection. Turning to clinical trials in patients with CKD, systematic reviews have pointed out that while these drugs reduce proteinuria and blood pressure, the effects on renal function were more ‘imprecise’. All trials, particularly in those with advanced kidney disease, have been characterised by a high incidence of hyperkalemia (and often, a drop in the eGFR). Unfortunately a large efficacy trial of spironolactone (or even eplerenone) in diabetic kidney disease has never been done.
(Adapted from Bauersachs, J; Eur Heart J, 2013, Editorial)
The incidence of side-effects such as hyperkalemia and gynaecomastia spurred research into alternate MRAs. Nonsteroidal MRAs were first described about a decade ago - currently, finerenone and esaxerenone are in active development. These agents are theoretically different from steroidal MRAs in several respects, as summarised below.
(Source: Agarwal R, et al EHJ 2020
Early Studies with Finerenone
These are theoretical considerations, but what do we know about what it actually does in humans? In an early, industry-funded, phase II study (ARTS Trial, Pitt et al, Eur Heart J, 2013), finerenone was assessed in two groups - 65 patients with heart failure (HFrEF) and 392 patients with HFrEF and CKD ( all eGFR > 30, mean eGFR 47 ), over a study duration of 4 weeks. The latter group also had a subset (n=63) who received spironolactone as a comparator. In this short trial, finerenone was deemed safe (less hyperkalemia, especially compared to spironolactone), and useful at reducing levels of natriuretic peptides and albuminuria.
This led to the larger ARTS-DN trial (Bakris et al, JAMA 2015) which enrolled ~ 800 patients with diabetes, albuminuria(UACR > 30mg/g) and eGFR>30, who were on RAAS blockade, and demonstrated that finerenone (at doses from 1.25 to 20 mg/day for 90 days) did not substantially reduce GFR or cause hyperkalemia compared to placebo. It is worth noting that 60% of patients here started with an eGFR>60ml/min. So finerenone seemed safe in this short-duration trial, but was it efficacious? It reduced albuminuria significantly, in a dose-dependent fashion, but had only modest BP-lowering effects. These safety and efficacy signals were deemed sufficient to merit further investigation into outcomes that matter to the renal community, leading to the FIDELIO-DKD (Finerenone in Reducing Kidney Failure and Disease Progression in Diabetic Kidney Disease) Trial.
The Study
Methods
On the theory that, the non-steroidal MRA, finerenone would have greater anti-inflammatory and anti-fibrotic activity than earlier, non-steroidal agents, the FIDELIO-DKD trial was conceived to test the hypothesis that finerenone slows CKD progression and reduces cardiovascular morbidity and mortality among patients with advanced CKD and type 2 diabetes. The study was a multi-centre (1062 sites), international (48 countries, 40% of patients in Europe), randomized, double-blinded, interventional trial.
Population
Inclusion criteria were designed to include those with significant CKD and/or proteinuria. `This had an interesting definition:
Adult patients with type 2 diabetes were included if they were already on treatment with an ACEi or ARB, had a potassium ≤ 4.8 and had:
a urine ACR between 30 and 300 mg/g; eGFR ≥ 25 but < 60; and diabetic retinopathy OR
urine ACR between 300 and 5000 mg/g and eGFR ≥ 25 but < 75 ml/min/1.73m2.
Thus lower levels of albuminuria were accepted if GFR < 60 and participants had diabetic retinopathy. This makes the definition of retinopathy important. Sadly this was only defined as ‘presence of diabetic retinopathy in the medical history’ (and yes, we combed through both the protocol and the design and baseline characteristics publication. It is always just defined as “diabetic retinopathy in the medical history”).
Patients excluded were those with
Known non-diabetic kidney renal disease
UACR ≥ 5000 mg/g
Poor diabetes control (HbA1c ≥ 12%)
Uncontrolled hypertension or systolic BP<90
HFrEF with NYHA class II-IV symptoms
Dialysis for AKI in the previous 3 months or those with a transplant, or transplant scheduled within 12 months
Additional exclusions were for participants who had suffered a stroke, TIA, acute coronary syndrome or hospitalization for heart failure in the previous month
(Source : Bakris GL, et al Am J Nephrol 2020)
The study design was published separately. All participants who consented had a mandatory run-in period (4 to 16 weeks) during which ACEi/ARB therapy was optimised to maximum tolerated doses. Screening was again repeated after this, prior to randomisation. Patients were randomised to receive either the placebo or the study drug. Patients with a GFR ≥ 60 ml/min received 20 mg of the study drug, those with eGFR < 60 received 10 mg. While in the trial, dosages of the study drug could be adjusted upwards (after the first month, if GFR was stable and K < 4.8) or downwards (at any time) according to the site’s discretion.
Study visits occurred at month 1, 4, and then every 4 months until completion. In order to monitor for adverse events, potassium and creatinine were also measured at a local laboratory (in addition to this study’s central monitoring) so that investigators could make changes in dosages as required. Quality of life questionnaires with KDQOL-36, EQ-5D-5L were recorded at baseline and annually (however, QOL outcomes are not included with the final results in this manuscript).
Pre-specified study outcomes and statistical methods.
The primary outcome was renal, and a composite of:
Death from renal causes
Kidney failure (dialysis ≥ 90 days, transplantation, or eGFR < 15 ml/min).
Sustained decrease of at least 40% in the eGFR from the baseline (over at least 4 weeks).
The key secondary outcome was cardiovascular, and a composite of:
Death from cardiovascular causes
Nonfatal myocardial infarction
Hospitalization for heart failure
Non-fatal stroke
Other secondary outcomes:
Sequential hierarchy of testing was used to study other secondary outcomes. This technique is often used in event-driven trials when there is a multiplicity of possible outcomes and the study cannot be sufficiently powered for each outcome. Only if the results for the initial outcome reject the null hypothesis are subsequent outcomes reported.
The hierarchical order of secondary outcomes was as follows:
Death from any cause
Hospitalization for any cause
Change in the urine ACR from baseline to month 4
A different renal composite, consisting of:
Death from renal causes
Kidney failure
Sustained decrease of at least 57% in the eGFR from baseline over at least 4 weeks (this corresponds to a doubling of creatinine)
The primary and the key secondary outcome were analysed in a time-to-event analysis, following the intention-to-treat protocol. On the basis that 1068 patients could have a primary outcome event, this trial was designed to have 90% power to detect a 20% lower risk of this primary outcome with finerenone than placebo. After stratification for region and eGFR category, stratified log-rank tests were used. Treatment effects were shown as hazard ratios (with confidence intervals) from stratified Cox proportional-hazards models.
Funding source and roles
The study was sponsored by Bayer, the maker of finerenone. The executive committee in collaboration with the sponsor designed and amended the trial protocol and supervised the conduct of the trial; an independent data monitoring committee conducted one planned interim efficacy analysis and oversaw patient safety. Bayer conducted the analyses, though all the authors had access to the data and participated in the interpretation of the data. There was medical writing assistance, funded by Bayer.
Results
13,911 patients underwent screening, as follows:
Fig S1 from Bakris et al, NEJM 2020
Since the enrollment for the cardiovascular (CV) trial (FIGARO) was on going concurrently and its enrollment criteria were less stringent than FIDELIO-DKD. So patients who were enrolling in the CV trial but also fulfilled the CKD trial were transferred. Thus 1374 patients were switched from FIGARO to FIDELIO-DKD (here). Likewise, patients who failed screening for FIDELIO-DKD were eligible for the CV trial and 1,152 patients went from screening for FIDELIO-DKD to enrollment in the FIGARO CV study.
During the run-in, or after screening, 8177 patients were excluded. Interestingly, among the 5734 participants who underwent randomisation, 60 were excluded for “critical GCP violations”. This meant that after a median follow-up period of 2.6 years, 5674 patients were included in the final analysis.
Fig S1 from Bakris et al, NEJM 2020
The study population was predominantly white (63.3%) male (70%) participants with diabetes for a mean duration of 16.6 years and a mean eGFR of 44. Just over half the participants, 52.5%, had a mean eGFR between 25 and 45 ml/min, 33.5% had an eGFR 45 to < 60 ml/min and 11.6% had GFR > 60 ml/min.
The majority (87.5%) of patients had macroalbuminuria (UACR ≥ 300 mg/g). Interestingly, some patients snuck in despite missing data (2 for eGFR and 3 for ACR), and 23 patients had ACR less than the eligibility cutoff of 30 mg/g. A higher number, 135 (2.4%) also had a GFR lower than the eligibility cutoff of 25 ml/min/1.73m2. Baseline systolic BP was ~ 138 mm Hg, and from table S1 the mean diastolic BP was 75.8 mm Hg.
Baseline medication use was intriguing, since several of these medications could also affect potassium or eGFR. ARB use was more common at 65.7% compared to ACEi at 34.2%. Fourteen patients were not on an ACEi or an ARB and 7 patients were on dual RAS blockade. 56.6% were already on a diuretic. Just over 50% were on a beta-blocker and a quarter on alpha-blockers, with CCB use at 63%. Reassuringly, a high proportion, 74.3%, were on statins. Given the recent interest in SGLT2 inhibitors, only 4.6% of the population overall were on these drugs, understandable, given the trial enrollment was 2015 to 2018, before the results of the recent flozin trials were known. Though table 1 only provides data on insulin and GLP-1RAs and flozins, from the supplementary table S1, we can see that metformin was used in 44%, sulfonylurea in 23%, and DPP-4 inhibitors in 27%.
Table S1 from Bakris et al NEJM 2020
Primary and Secondary Kidney-related Outcomes:
Analysing the pre-specified primary composite outcome further, this occurred in 17.8% of patients in the finerenone group and 21.1% in the placebo group. The absolute, between-group difference was 3.4% after 3 years; hazard ratio 0.82 (0.73-0.93; P = 0.001). This yields a number needed to treat value of 29 over 3 years ( 95% CI, 16 to 166), with all the caveats in interpreting this value. As can be seen from the graphs, beneficial effects appear to accrue after 12 months of treatment. If you breakdown the composition of the primary outcome (which was renal death OR kidney failure defined as dialysis, transplantation, or eGFR < 15 ml/min OR a sustained decrease of at least 40% in the eGFR from the baseline), it is clear that the outcome was mostly driven by the 40% GFR decline criterion. The difference for ESKD was only 20 patients (14% relative and 0.7% absolute risk reduction, NNT 143 over the study period) whereas for the 40% GFR decrease, the NNT was 98 patients (19% relative and 3.4% absolute risk reduction).
Fig 1 from Bakris et al, NEJM 2020
The key secondary outcome event - a MACE of cardiovascular complications - occurred in 13% and 14.8% in the finerenone and placebo groups respectively - an absolute risk reduction of 1.8%, hazard ratio 0.86 (0.75-0.99; P = 0.03). A forest plot showing the results for other outcomes is shown below:
Fig 2 from Bakris et al, NEJM 2020
The prespecified hierarchical sequence of testing meant that among other secondary outcomes, since the first outcome in the hierarchy, “death from any cause,” was not significantly different between the groups, subsequent secondary outcomes can only be considered exploratory. When analysed, however, these secondary outcomes - reduction in the UACR by month 4, average UACR value maintained throughout the study and the secondary kidney outcome composite were all lower in the finerenone group.
Figure S4B, showing difference in all cause mortality from Bakris et al, NEJM 2020
Subgroup analysis
These are buried in figure S3, and reveal some interesting differences in the hazard ratios within pre-specified sub-groups. The primary outcome occurred among ~15% of whites, ~30% of blacks (low numbers overall) and ~20% of asians - but the difference from placebo was significant mainly in Asia. Similarly, hazard ratios in those with a history of cardiovascular disease or a low baseline BMI seem to be somewhat different. The interactions for GLP-1RAs and SGLT2i are probably just a reflection of small numbers. Have a look at all the subgroups below.
Fig S3 from Bakris et al, NEJM 2020
Blood pressure and Albuminuria
There was a very small decrease in systolic blood pressure of about 3 mm at 1 month which was even lower at 2.1 mm after 1 year. They do not describe any changes in diastolic blood pressure in the manuscript or even the supplement.
As was seen in the prior trial (ARTS-DN), there was a significant decrease in the albuminuria which does get its own graph as you can see below.
Fig 3A from Bakris et al, NEJM 2020
Potassium
Investigator-reported hyperkalaemia was much more common in the finerenone group (18.3%) than placebo (9%). Serious hyperkalemia occured in 44 (1.6%) of patients on finerenone versus 12 (0.4%) on placebo - a difference of 32, greater than the number saved from dialysis. For hyperkalemia in the other category (see table 2), it occured in a whopping 446 patients (15.8%) compared to 221 in placebo (7.8%). However, only 2.3% of participants receiving the study drug discontinued the trial because of a high potassium; only 1.4% required hospitalisation for the same indication. However, from table S3, note that 307(10.8%) patients on finerenone required potassium lowering agents such as SPS or another resin or binding agent, compared to 184 (6.5%) on placebo. This is higher than the baseline usage which was about 70 in both groups.
GFR
Somewhat less impressive than the GFR graphs we have seen with SGLT2is recently
Fig S6 from Bakris et al, NEJM 2020
Other adverse events
GFR decrease was more common with finerenone as well (6.3% versus 4.7%), though investigator-reported AKI was similar (4.6 versus 4.8%). Peripheral edema also appeared to be less with finerenone (6.6 versus 10.7%). Interestingly enough, there was less bronchitis and pneumonia with finerenone as well as less hypoglycemia (5.3 vs 6.9%).
Table 2 from Bakris et al, NEJM 2020
Discussion
How do we interpret these results? Clearly, given the increasing incidence of diabetic end-stage kidney disease, patients who have diabetes and kidney involvement need all the help they can get. We already know that treatment with maximal doses of RAS inhibitors, in addition to other measures - including, but not limited to, blood pressure control, blood sugar control, lifestyle interventions, balanced diet, salt and protein restriction - can all act together to significantly improve the outlook for these patients. Can we do more? Are the results of FIDELIO impressive enough that we start using finerenone in all patients with diabetic nephropathy? Let us examine the results a little bit closely.
Firstly is 40% reduction in GFR a valid surrogate outcome? It is not a hard clinical outcome like the need for dialysis or death. The latter two were reduced somewhat but not significantly in FIDELIO. However, from some modeling studies, as well as an FDA-NKF workshop (Levey et al AJKD 2019), the consensus seems to be that 40% reduction in GFR is an acceptable surrogate (see more at Badve et al, NDT 2015). This is important, since, as we note above, the results in FIDELIO were mostly driven by this component of the composite outcome. Of note IDNT (Lewis 2001) also was not able to show a reduction in death or dialysis, though RENAAL (Brenner 2001) and the Collaborative Study Group (Lewis 1993) were able to clear this hurdle. It is also important to keep in mind that the individual patient benefit is small, especially as we compare them with the other blockbuster drugs in diabetic nephropathy which have come out recently, the flozins. Are the adverse effects worth this small reduction?
There are certain situations where the treating clinicians are actively looking for more therapeutic options - for instance, resistant hypertension or proteinuria. Addition of spironolactone to existing therapy may already have a role in reducing proteinuria and blood pressure in such patients (although doubts remain - unclear effect on renal protection and mortality; side-effects such as hyperkalemia, decreasing GFR and gynecomastia). However, in this FIDELIO-DKD study, patients with uncontrolled hypertension were specifically excluded.
Hyperkalemia is a real concern, particularly in the patients such as those included in this trial, with a low baseline eGFR and already on maximal doses of ACEi or ARBs. An excellent review of hyperkalemia with RAAS inhibition was published in 2010. In recent times, the important trials and the hyperkalemia incidence is summarised below:
A Cochrane meta-analysis (revised in 2020) pooled data from 17 studies and 3000 participants to estimate the relative risk of hyperkalemia with aldosterone antagonists compared to usual treatment in proteinuric kidney diseases to be 2.17(1.47 to 3.22). In the FIDELIO-DKD study, the relative risk of investigator-reported hyperkalemia, seen in 15.8% of participants on finerenone, was 2.03 (1.76 - 2.33) compared to the group with maximal ACE/ARB treatment alone, calculated from data provided). (The “number needed to harm calculation, based on overall numbers of patients with hyperkalemia, was 12.4 patients for 3 years). The ACEi/ARB group also had an incidence of hyperkalemia of 7.8% - significantly higher than that in some other studies.
What is debatable is whether these results are sufficient to reassure physicians concerned about hyperkalemia. While it is reassuring that in this study, there was no attempt to restrict dietary potassium, it is worthwhile remembering that there have been attempts to use potassium-binding agents in order to enable dosing with agents such as aldosterone (the AMBER trial). Even in the present trial, more patients in the intervention arm needed potassium binding agents for hyperkalemia (307 versus 184 in placebo). This will be important to keep in mind when we do an analysis of cost. We see a small benefit, by using an expensive agent, which also requires increased use of more expensive agents to manage the adverse events. Is the long-term reduction in outcomes worth this domino of increasing costs?
Where do these results sit when compared to the data from the CREDENCE trial (Perkovic et al), and the DAPA-CKD trial (Heerspink et al) published recently?
The following table contrasts these studies:
Limitations
In the manuscript, the authors point out that the study population excluded participants without albuminuria and also had less than 5% of participants who identified as black, limiting the generalizability of the findings. In addition, we have pointed out a few other limitations - such as the exclusion of patients with uncontrolled hypertension or HbA1c> 12%. Several pre-specified sub-groups had variations in the statistical significance of the hazard ratios of the primary event versus placebo, and these differences could have been discussed in greater detail. Details of patient-centric outcomes - such as quality of life - have not been reported yet by the authors.
Final word
Where do these studies leave us? The past two years have brought these important trials in diabetic kidney disease; there is likely to be further evidence tying the use of these agents to chronic kidney disease in general or to cardiovascular outcomes. It remains to be seen if clinicians will take to prescribing finerenone and other MRAs, especially in this population - advanced kidney disease already treated with maximal doses of RAAS inhibitors. We know from experience that the real world in which our patients live and in which we prescribe these agents is very different from the carefully selected environs of the randomised clinical trial. The diabetic with chronic kidney disease often has multiple other complications, multiple other medications and a diet that may vary significantly from day-to-day. It may be difficult to maintain these agents safely in such a population.
In a carefully selected, predominantly middle-aged, white population without HFrEF or uncontrolled hypertension, the addition of finerenone produced, at 3 years, an absolute risk reduction of 3.4% (compared to those already on maximal ACEi/ARB therapy alone) in a composite renal outcome with a roughly 2-fold increase in the risk of hyperkalemia. Is this worth doing?
Summary by Rajesh Raj DM, FRACP, PhD
Clinical Associate Professor, University of Tasmania
Consultant Nephrologist
NSMC Intern, Class of 2020