Joint #IPNA & #NephJC Chat
Tuesday January, 24th, 2023 at 9 pm Eastern (AEST = June 8th, 11am)
Wednesday January 25th, 2023, at 9 pm Indian Standard Time and 3:30 pm GMT (AEST = June 9th, 2am)
Clin J Am Soc Nephrol 2022 Dec;17(12):1763-1774.doi: 10.2215/CJN.02400222. Epub 2022 Nov 21.
Effects of Bardoxolone Methyl in Alport Syndrome
Bradley A Warady, Pablo E Pergola, Rajiv Agarwal, Sharon Andreoli, Gerald B Appel, Sripal Bangalore, Geoffrey A Block, Arlene B Chapman, Melanie P Chin, Keisha L Gibson, Angie Goldsberry, Kazumoto Iijima, Lesley A Inker, Clifford E Kashtan, Bertrand Knebelmann, Laura H Mariani, Colin J Meyer, Kandai Nozu, Megan O'Grady, Michelle N Rheault, Arnold L Silva , Peter Stenvinkel, Roser Torra , Glenn M Chertow
PMID: 36411058
Introduction
Until now there have been more editorials on Alport syndrome than patients in the largest RCT for the disease, despite it being the second most prevalent inherited cause of chronic kidney disease. Mutations in the genes encoding alpha-3, alpha-4, and alpha-5 of type 4 collagen cause progressive loss of kidney function. By ages 25 and 40, respectively around 50% and 90% of untreated men with X-linked Alport syndrome proceed to ESKD. Unfortunately, there are no disease-modifying medications currently available. The only medications which retard the progression are inhibitors of the renin angiotensin system (RASi). The field has moved to increasingly early use of ACEi, even before the onset of proteinuria, backed up by small RCT evidence (Gross et al, KI 2020) and reflected in the new clinical practice guidelines (Kashtan et al, Pediatric Nephrology 2021). Clinical studies on mineralocorticoid receptor blockers (MRAs) and sodium-glucose cotransporter-2 inhibitors (SGLT2i) recruited very few patients with Alport syndrome (for example, DAPA-CKD had five), and thus no definitive recommendations can be made about their use at this time.
Bardoxolone methyl is a newcomer to the arena. Bardoxolone methyl is a semisynthetic triterpenoid that activates the nuclear factor erythroid 2-related factor 2 (Nrf2). Nrf2 is a transcription factor that modulates the expression of various genes involved in inflammation, oxidative stress, and cellular energy metabolism (figure 1). Based on the hypothesis that bardoxolone can facilitate resolution of inflammation by activating Nrf-2 and inhibiting the NF-light-chain enhancer of activated B cells, it has been tried as a potential intervention to delay progression of kidney disease in a number of diseases (Zoja C et al 2014). The results of murine models of ischemic AKI (Wu Q et al, 2011) revealed enhanced expression of Nrf2 in the glomerular endothelial cells and cortical peritubular capillaries. Heme oxygenase 1 expression was also seen to increase in the tubules and interstitial leukocytes upon the administration of bardoxolone methyl. Additionally, there was a notable decrease in inflammation and blood urea nitrogen concentration, coupled with an improvement of tubular and glomerular damage. However, it is unknown if these are involved in the pathway of progression in Alport syndrome.
Figure from (Zoja C et al 2014)
Due to the potential for it to exhibit inhibiting effects on tumor proliferation, metastasis, angiogenesis, and tumor-linked myeloid-derived suppressor cells, bardoxolone methyl was initially identified as a potential treatment for advanced solid tumors and lymphoma in multiple animal models. Intriguingly, during the first-in-human clinical trial with patients with cancer, an increase in eGFR was noticed (Hong D et al 2012). Phase 1 studies in patients with diabetes and CKD were already underway and showed increased eGFR with bardoxolone methyl, furthering interest (Pergola P et al 2011). However, the subsequent phase 3 BEACON trial in 2185 patients with diabetes and CKD stage 4 was halted prematurely due to an increased risk of hospitalisation or death from heart failure (Zeeuw et al 2013). Given this context, it is reasonable to wonder if bardoxolone treatment for Alport syndrome would be effective - since there is no clear mechanism for why this particular molecule would work (apart from it’s supposed GFR increasing ‘mechanism’) and the clear harm demonstrated in diabetic nephropathy. This brings us to the discussion of the CARDINAL trial.
The Study
Methods
This was an international, multicenter, randomised, placebo controlled, double blind phase 3 trial.
Inclusion criteria
Age 12–70 years with histologic or genetic confirmation of Alport Syndrome
eGFR of 30–90 ml/min/1.73m2, and urinary albumin-creatinine ratio (UACR) of ≤ 3500 mg/g (<400 mg/mmol)
Maximally tolerated stable doses of an ACE inhibitor or ARB, unless medically contraindicated
Exclusion criteria
Heart disease:
BNP >200 pg/mL
EF <40%, clinically significant valve disease or pericardial constriction on screening echo
Any prior angina, MI, PCI or admission for heart failure
Atrial fibrillation
Renal transplant recipient
HbA1c >11.0%
BP >160/100
Unwilling to practise acceptable methods of birth control (male or female), pregnancy and breastfeeding
Interventions
Adult patients were initiated on once-daily dosing with 5 mg, and the dose was increased every two weeks to their target dose of 20 or 30 mg (the latter for patients with a baseline UACR of >300 mg/g). During a 4-week withdrawal period between weeks 48 and 52, patients did not receive the study drug; treatment was then resumed at the same dose received at week 48 and continued until week 100. Patients were then re-assessed at week 104, after a 4 week washout period. Figure 1 depicts the dose-titration scheme and schedule of assessments.
Patients <18 years of age were started on 5 mg every other day during the first week, 5 mg daily during the second week, and then increased the dose every 2 weeks according to the dose-titration scheme as adults.
Randomization was stratified by baseline UACR (≤300 mg/g, >300 to ≤1000 mg/g, and >1000 to ≤3500 mg/g), 1:1 to bardoxolone methyl or placebo.
Outcomes
Primary efficacy endpoint
The change from baseline eGFR by randomised group after 48 and 100 weeks of treatment, while on treatment.
Key secondary efficacy endpoints
Change from baseline in eGFR at weeks 52 and 104, after an intended 4 weeks off treatment.
Analytic Plan
It was calculated that enrollment of 150 patients would provide approximately 80% power, assuming a between-group difference at 48 and 100 weeks in change from baseline eGFR of 3.1 ml/min per 1.73 m2. In case there was a significant treatment effect for both year 1 end points, then the significance level for year 1 (0.025) remained available to be carried forward to the year 2 testing sequence. Thus, if both year 1 end points were significant, the year 2 testing sequence was tested using a significance level of 0.05. The difference in GFR was performed using a mixed model repeated measures approach to analyze the primary end points, with eGFR through 48 or 100 weeks as the response, with baseline eGFR, baseline UACR strata, and geographic location (United States, yes/no; week 100 only) as covariates and the following fixed factors: treatment group, time, interaction between treatment group and time, and interaction between baseline eGFR and time. Missing data was imputed for primary analysis, with additional sensitivity analysis including tipping point analysis being performed.
Funding
The study was funded by Reata Pharmaceuticals, the maker of Bardoxolone. Three of the authors are employees of the company, and 4 other employees (not on author list) are acknowledged for their help in writing the paper. It is not stated who performed the statistical analysis and whether the academic authors had full access to the data. Interestingly, the rigorous CJASN disclosure policy means the authors' disclosures runs almost 1800 words.
Results
Study population
Recruitment ran from July 2017 to November 2018 at 48 sites in the USA, Europe, Japan, and Australia. As shown in the CONSORT diagram, 371 patients were screened and 157 subsequently randomised, with 77 patients to bardoxolone and 80 to placebo. Safety follow-up at 104 weeks was completed by 98%, and eGFR values at 104 weeks were obtained in 96% of patients.
Supplemental Figure 2 from Warady et al, CJASN 2022; CONSORT diagram.
Baseline characteristics of participants can be seen in Table 1. Over 90% had a genetic diagnosis of Alport syndrome, and females made up 58% of the population. The mean age was 40, and the mean age of Alport diagnosis was 30, but 15% of participants were <18 years old. Mean baseline eGFR was 63 ml/min/1.73m2, and mean urine ACR was 142 mg/g (16 mg/mmol). Only ~75% of patients were on an ACEi or ARB.
Table 1 from Warady et al, CJASN 2022
Primary & secondary outcomes
The primary outcome of ‘on treatment’ eGFR at 48 weeks showed a higher eGFR in the bardoxolone treatment group versus placebo by 9.2 ml/min/1.73m2 (97.5% CI 5.1-13.4 ml/min/1.73m2; P<0.001). At week 100, the ‘on treatment’ bardoxolone group had a higher eGFR of 7.4 ml/min/1.73m2 (95% CI, 3.1 to 11.7 ml/min/1.73 m2; P<0.001 - Figure 1A).
Figure 1A from Warady et al, CJASN 2022.
The mean difference in eGFR between treatment groups for the primary end point at 48 weeks (year 1) and 100 weeks (year 2) are shown above.
Key secondary outcomes included eGFR 4 weeks after temporary drug discontinuation at week 48, and after drug discontinuation at week 100 when active treatment ended. Results after washout at week 52 and week 104 are shown in Figure 1D, demonstrating reversal of much of the eGFR gains. This analysis did not include patients who had discontinued drug in year 1, and any eGFR values taken at ≥2 week intervals were accepted as adequate for the ‘4 week’ washout period. Further sensitivity analysis performed by the FDA are presented in the discussion.
Figure 1D from Warady et al, CJASN 2022. Observed mean eGFR values through the 104 weeks of the study.
Subgroup analysis
Subgroup analysis was reported ‘on treatment’ at week 100 rather than ‘off treatment’, and showed mostly consistency of results across pre-specified subgroups, despite what the text of results in the paper might say. The p values of interaction seem to be significant for race/ethnicity and baseline GFR - though the 95% CI overlap quite a bit, and it is unknown if these are chance or whether biological plausibility exists, given the mechanism of action of bardoxolone itself on eGFR is unclear.
Figure 2 from Warady et al, CJASN 2022, showing subgroup analysis of eGFR change at week 100 while on treatment.
Safety
There were multiple reasons to be concerned about drug safety.
Ten (13%) of patients on bardoxolone stopped treatment early due to an adverse event, versus four (5%) on placebo. A further 5 patients on bardoxolone stopped treatment when a pre-specified threshold for ALT rise was met. In total by week 100, 26 (34%) on bardoxolone had discontinued treatment, versus 13 (16%) in the placebo arm. They report that three patients in each group developed “kidney failure” (no specific definition), but there was no consistency in the supplement as to whether this was coded as a discontinuation criterion or a serious adverse event.
Rises in liver enzymes below the threshold for discontinuation were much more common in the bardoxolone group versus placebo (ALT rise occurred in 47% on bardoxolone versus 3% on placebo, AST rise 25% versus 1%).
Supplemental Figure 6 from Warady et al, CJASN 2022. In panel D the vertical line corresponds to 3x upper limit normal (ULN) for ALT, and the horizontal line corresponds to 2x ULN for total bilirubin - no patients met Hy’s criteria in the upper-right quadrant.
Various further safety outcomes are presented in Figure 3. A decrease in weight from baseline was seen in 13% of patients on bardoxolone, versus 1% with placebo. The weight loss was greater at higher starting BMI.
Proteinuria is presented as having equal frequency, but absolute values rose with bardoxolone in proportion to eGFR rise, and this was reversible with discontinuation of bardoxolone (supplemental figure 5A).
Figure 3 from Warady et al, CJASN 2022, showing different safety parameters over time.
Excess muscle spasms were identified in BEACON, and in CARDINAL were common in both groups, at 49% with bardoxolone and 34% with placebo.
Discussion
The publication has many industry authors, and they conclude the paper on an upbeat note with: “Treatment with bardoxolone methyl resulted in significant preservation in eGFR relative to placebo after a 2-year study period”. Meanwhile, back in reality, the drug efficacy and safety prolife is far from optimistic, despite a technically ‘positive’ primary outcome. As usual the data from CARDINAL was presented to the FDA long before publication in a scientific journal for public scrutiny. In December 2021 the assigned FDA committee voted unanimously against approving bardoxolone, stating lack of evidence that it slowed CKD progression and its unfavourable risk profile.
The major criticism related to the adequacy of the washout period used (FDA briefing document, Bardoxolone Methyl, 12/8/21). Within pharmacology a ‘reversible pharmacodynamic effect’ occurs when, after an initial eGFR rise, there is both no slope difference in CKD progression and then no sustained eGFR benefit after drug discontinuation. Look at the slopes in figure 1D as bardoxolone seems to display this general pattern - there’s minimal accrual of excess eGFR at 104 weeks that would indicate true impact upon disease progression. The paper states that the drug washout period used was adequate as >99% of the drug is cleared within 2 weeks, but the FDA asserted that it cannot be assumed that a pharmacodynamic effect on eGFR would follow the same resolution period as simple drug clearance. In fact, FDA modelling including use of inulin clearance data from the TSUBAKI trial (Nangaku et al, KI Reports 2020), predicted that at week 104 about 28% of a ‘reversible effect’ would remain. In fact, a washout period would need to be >8 weeks to ensure all temporary effects on eGFR had been reversed. Figure 5 shows the eGFR difference predicted by the FDA model if a longer washout period was observed. The matter was made worse in CARDINAL as 39 patients actually had their ‘104 week’ assessment blood test taken the week before, meaning only 71% of patients had their eGFR actually taken ≥4 weeks after the last dose of drug.
Figure 5. FDA model with solid line showing predicted eGFR with a longer washout period in CARDINAL. From FDA briefing document, Bardoxolone Methyl, Cardiovascular and Renal Drugs Advisory Committee Meeting 12/08/21.
The paper reported a 4.4 ml/min/1.73m2 eGFR difference at week 104 with bardoxolone, but accepted a ≥2 week blood test for the ≥4 week cutoff. Additionally, they excluded 25 patients from week 104 analysis who had discontinued treatment in year 1, even though eGFR data was available (9 in placebo group and 16 in bardoxolone group, making up 16% of the total trial population). The FDA performed their own sensitivity analysis and really took the last possible shine off the results by including all available eGFR data but only analysing the 111 patients who had bloodwork taken after a strict 4 week cutoff. This more stringent group only showed a 0.8 ml/min/1.73m2 eGFR difference - a statistically and clinically negligible value.
Table 2. From slide 26, FDA presentation on Bardoxolone Efficacy and Safety, Cardiovascular and Renal Drugs Advisory Committee Meeting December 8, 2021.
So why does the initial rise in eGFR with bardoxolone not translate into better outcomes? Well, with our rockstar drug classes, RASi and SGLT2i, we see the opposite pattern to bardoxolone: an initial fall in eGFR (which we are comfortable ignoring as long term strategists), with decreases in proteinuria and BP. This is interpreted as indicating that intraglomerular pressure is being lowered, and this phenomenon then translates into improved clinical outcomes over the patient’s lifetime. With bardoxolone the exact mechanism of the initial eGFR rise is unknown. Given the accompanying rise in proteinuria and BP (picked up in BEACON when 24 hour ambulatory monitors were used), and the absence of reassuring animal data, it is possible the mechanism is induction of hyperfiltration. If true, this would result in an acceleration of CKD progression from bardoxolone in the long-term. This was suggested in an insightful editorial (Baigent and Lennon, JASN 2018), even before the data from CARDINAL was presented.
We didn’t get convincing evidence for efficacy, but the real damning data relates to the drug’s safety profile. The cardiac exclusion criteria were rigorous after the bardoxolone arm in BEACON had more heart failure events (HR 1.8, P<0.001). The exclusion of patients with even a whiff of cardiac disease in CARDINAL did ameliorate heart failure events, in this much younger cohort without diabetes, though some BNP rises were still observed.
However, cardiac concerns were replaced with liver liability, with 47% of patients having a rise in ALT with bardoxolone. However, bilirubin did not rise to x2 upper limit of normal in any patient, so no patient met Hy’s criteria (the concept that if both ALT and bilirubin rise due to a drug in a small clinical trial, then when a large population is exposed there will be cases of severe liver failure - concerns nephrologists will remember from TEMPO 3:4 and high dose tolvaptan [Torres at al, NEJM 2012]).
Figure slightly modified from FDA presentation slide 32, Bardoxolone Efficacy and Safety, Cardiovascular and Renal Drugs Advisory Committee Meeting December 8, 2021
In BEACON a subpopulation was monitored with 24 hour ABPM, and showed a BP increase of 5.3/2.6 mmHg. No such signal was seen in CARDINAL, though only clinic BP was recorded.
Excess GI symptoms, muscle spasms, and proteinuria due to bardoxolone remain additional concerns. Though adult nephrologists may think that ~4% weight loss is benign, it is concerning that the small number of paediatric patients (total n=23) did not put on weight when on bardoxolone. The observation of weight loss is consistent with animal models and BEACON data, though the exact mechanism is again unknown. The cause might be as simple as altered taste, vomiting and decreased appetite seen in the bardoxolone arm. Given the fact that patients considered for bardoxolone treatment would contain many children and young adults, any plausible impact on growth and development is disquieting.
Figure from FDA presentation slide 41, Bardoxolone Efficacy and Safety, Cardiovascular and Renal Drugs Advisory Committee Meeting December 8, 2021
So given these disappointing results, can we just forget that bardoxolone methyl ever existed and move on? Well, not quite yet - there is an ongoing phase 3 ADPKD trial (FALCON, approximate N=850), a phase 3 study in Japan looking at diabetic nephropathy (AYAME, N=1013), and the CARDINAL trial and others will continue as an open-label, extended access trial called EAGLE.
CARDINAL investigators must be praised for performing by far the largest RCT yet in Alport syndrome. Without industry backing the nephrology community has largely relied on animal models and observational studies to assess treatment options. Seeing what can be achieved in only 16 months of recruitment here is heartening. The only previous trial that randomised patients with Alport syndrome did so in children, and managed to randomise 22 patients and follow them over 6 years (EARLY PRO-TECT Gross et al, KI 2020). This small study added to the evidence that ACE inhibition in children with Alport syndrome was safe and highly beneficial, even prior to the onset of proteinuria. Unfortunately, this story doesn’t fill anyone with high hopes that an SGLT2i trial in a specific Alport population will be organised quickly. For now practitioners are using flozins based on extrapolation, backed up by a few case series (Boeckhaus Cells 2021, Liu KI 2022). However, it is hard to imagine that trying an SGLT2i off-label in proteinuric patients is riskier than the current situation of allowing them to continue in an extended trial of bardoxolone.
Conclusion
CARDINAL did not demonstrate that bardoxolone methyl meaningfully slowed the progression of CKD in patients with Alport syndrome. When we assess the ongoing trials of bardoxolone we must pay careful attention to adequacy of the washout period deployed. After factoring in the concerns regarding proteinuria, hypertension, stunted growth, liver enzyme derangement, heart failure, and possible hyperfiltration injury, it is unfortunately difficult to maintain any optimism that this class of medication will be joining our armamentarium for slowing CKD, regardless of aetiology.
Summary prepared by
Priti Meena
Assistant professor, Nephrology. AIIMS Bhubaneswar
#NephJC Associate Editor
And
Jamie Willows
Renal registrar, Sunderland Royal Hospital, UK
#NephJC Associate Editor