September 5th, 2014
Perspectives on PARADIGM-HF
CardioExchange Editors, Staff
The post titled “Let’s Take a Close Look at PARADIGM-HF” generated a lot of discussion. We present that discussion here in a separate post (edited to better showcase the substantive points of members who offered their insightful comments).
Felix Valencia, MD, PhD Almeria, Spain
A key issue in PARADIGM-HF is blood pressure control as a determinant of outcomes. Treatment with LCZ696 was associated with lower blood pressure and more symptomatic hypotension. Patients who were less sick seem to derive greater benefit in the trial, as suggested by the subgroup analyses by age, NYHA class, prior hospitalization for heart failure, and prior treatment with aldosterone antagonists. These HF patients would have better BP at randomization and greater potential benefit of lowering BP from LCZ696. The significant interaction between treatment effect and NYHA class was in the primary outcome but not in death, so the main interaction was in hospitalization, an endpoint very sensitive to appropriate BP control. We hope that we have a new drug that may help our HF patients to live better and longer. But before accepting that, I think that the role of BP difference in outcomes should be completely clarified.
Milton Packer, MD Dallas, TX
1. Drug Dosing
Dr. Prasad proposes that the dose of enalapril was too low, and we should have used 40 mg daily of enalapril as a comparator. However, when 40 mg of enalapril daily has been used in a clinical trial (CONSENSUS), these extremely high doses were poorly tolerated due to hypotension and renal insufficiency. As a result, even though the target dose may have been 40 mg daily, the actual dose achieved in the CONSENSUS trial was 18.4 mg. However, in the PARADIGM-HF trial, the average dose of enalapril achieved was 18.9 mg. Therefore, the dose of enalapril used as the comparator in the PARADIGM-HF trial was the highest dose of the drug ever used in a clinical trial of heart failure.
There is no evidence that valsaratan 160 mg BID produces more blockade of the renin-angiotensin system than enalapril 10 mg BID. In fact, valsartan 160 mg BID does not even have a mortality effect when compared with placebo, whereas enalapril 10 mg BID does have a survival benefit.
Dr. Prasad contends that the neprilysin inhibitor could have been a placebo, and the difference between treatments was due to more blockade of the renin-angiotensin system in the LCZ696 group. However, more angiotensin blockade would have led to more renal dysfunction and more hyperkalemia, but the LCZ696 had less renal dysfunction and less hyperkalemia.
2. Drug Run-In Periods
Drug run-in periods are not a controversial study-design choice. In fact, this type of design is strongly preferred because it closely mimics clinical practice. In clinical practice, patients who do not tolerate a drug are not maintained on it, and that is the intent of the run-in period. Indeed, the FDA strongly supported our decision to have a double-drug run-in period because it would enhance the credibility of the trial.
3. Stopping What Already Works
As to the claim that it makes little sense to stop a drug that is working to enter a trial, I would ask “How does one conclude that the patients’ prior use of an ACE inhibitor or ARB was “working”? And how effective could ACE inhibitors have been if — when patients were given enalapril 10 mg BID (a dose higher than what most were receiving before entering the trial) — the patients in the PARADIGM-HF trial died earlier, were hospitalized more frequently, and had a worse quality of life?
The real lesson of PARADIGM-HF is that combined angiotensin receptor neprilsyin inhibition is superior to inhibition of the renin-angiotensin system alone in patients with chronic heart failure. That is the conclusion of our paper, which passes stringent peer review in The New England Journal of Medicine.
Vinay Prasad, MD MPH Rockville, MD
The VAL-HeFT trial — where valsartan 160 BID was no better than placebo — occurred in a setting where 92% of patients were already on an ACE inhibitor. As such, it cannot be used to say what the effect of valsartan is among patients not taking an ACE inhibitor, as was the case in PARADIGM-HF.
CONSENSUS shows a benefit for enalapril when doctors attempt to titrate the dose up to 40 mg. The protocol for PARADIGM does not ask providers to replicate this drug goal, settling for 20 mg.
I believe the persistent difference in BP makes the case that the dose matters, particularly as neprilysin inhibition is thought not to confer much BP effect by itself. I think it is dangerous to draw too many conclusions from the rates of hyperkalemia and renal dysfunction, particularly because there is a drug run-in period that may mitigate some of these adverse events. I wonder why the run-in period was twice as long for the study drug than enalapril. The role of drug run-in periods has a vast literature. In this case, it was not necessary, and I am not convinced it was a good design simply because the FDA has said they will accept it.
The bottom line: If sacubutil really adds something to ARB therapy, why not conduct a clean trial of randomizing patients already on an ARB to sacubutil or placebo?
Dismissing the evidence from Val-HeFT eliminates all of the evidence that supports the use of valsartan in heart failure. With Val-HeFT set aside, what evidence is there that valsartan 160 mg BID does anything for patients with heart failure? Second, it is not the dose you plan to achieve that matters; it is the dose you actually achieve. One can design a trial with enalapril with a target dose of even 80 or 200 mg daily, but it would mean very little, because few patients would ever be able to take it. Even if a trial allowed patients to receive 40, 80 or 200 mg daily, it would not provide any useful information about these doses. Third, should we really ignore the differences in the rates of hyperkalemia and renal dysfunction between the two treatments over a period of 27 months, simply because patients had to be able to tolerate both drugs for a few weeks?
My argument about VAL-HeFT, valsartan, and dosing is that it is very possible that the difference observed in PARADIGM-HF was because valsartan was given at 320 mg and enalapril was given at 20 mg. There is no study where these two drugs were compared at these doses to say for sure that they perform similarly.
Pointing out that enalapril had a mortality benefit over placebo in RCT, but that valsartan did not — and, therefore, enalapril by itself is better — is a misleading argument. Enalapril is surely better than not taking an ACE inhibitor, but I disagree that valsartan has no effect by itself. The trial in which it did the same as placebo was one in which most patients were already on an ACE inhibitor. As we know from the ONTARGET trial, the combination of ARB and ACE adds toxicity, but no benefit. I think most doctors believe that taking valsartan is better than not taking either an ACE inhibitor or an ARB, as made clear by the fact that 397 patients were taking valsartan at the screening visit (see PARADIGM-HF’s supplementary appendix).
Now the question becomes: Do we know, with any degree of certainty, who would win in a head-to-head comparison of valsartan 320 mg and enalapril 20 mg? I, and many others, think we don’t know for sure. PARADIGM-HF’s protocol does not attempt to push the dose of enalapril to 40 mg. This is unbalanced.
There is a common misunderstanding that drug run-in periods better reflect the “real” world of clinical practice. They do not. Here is why: Let’s say 4 drugs (A. B, C, and D) are used fairly interchangeably for a medical condition. Let’s also say that Drug A and B are used more often than C and D. A clinician wants to know – maybe she suspects – that Drug B is better than Drug A. How do you test this question to mirror real-world practice? You randomize patients to start with Drug B or Drug A, and allow patients in both arms to switch to other drugs in the class, per investigator discretion, if they are intolerant. Then you look at the intention-to-treat analysis to see which arm does better. This tests the relevant clinical question: If I start with Drug B, do I save lives?
Instead, here is how a run-in period trial would test the question: All patients take Drug A for a couple weeks. Intolerant patients withdraw from the trial. All patients then take Drug B for a few weeks. More patients withdraw. Whoever remains is randomized to Drug A or Drug B, and then you do more or less the same from that point.
One might think that the second scenario (drug run-in) is the same as the first, but more efficient. Unfortunately, it isn’t. It is testing this clinical question: Is it better to stick with Drug B, if you have taken and tolerated A and B, or switch back to A? For many reasons, this is not the same clinical question — and a clinical question that does not mimic real-world practice. Moreover, we have no way to advise (or identify) the people who dropped out.
It’s still not clear why the PARADIGM-HF run-in period was twice as long for LCZ696. More important, why not randomize patients with HF who are already on an ARB to sacubutil or placebo? A good HF trial is not necessarily one that is positive, but one that picks a single clinical question and answers it. PARADIGM-HF may be positive, but it does not answer the clinical question of whether sacubutil does anything.
If the claim that valsartan 320 mg daily may be more effective than enalapril 20 mg daily in preventing cardiovascular death in patients with heart failure were true, the results from PARADIGM-HF could be entirely due to valsartan’s superiority over enalapril rather than an incremental contribution of neprilysin inhibition. Such a proposal is at best quite imaginative; it is shared by no one in the cardiovascular field. Adequately powered head-to-head comparator studies have shown a variety of ARBs at a variety of doses (including valsartan 320 mg daily) to be no better than an ACE inhibitor and generally somewhat inferior to an ACE inhibitor. Therefore, the suggestion that the results of PARADIGM-HF are explicable by the superiority of valsartan alone is at odds with the totality of available evidence.
We have used run-in periods in many heart failure trials, and it was used in the SOLVD Treatment Trial, which originally showed the survival effects of enalapril. If trials with run-in periods were inherently flawed, no trials would use them.
As to the idea that we should have designed the trial as a comparison of valsartan in one arm and valsartan plus a neprilyin inhibitor in the other arm, most patients with heart failure are not taking an ARB, and those receiving valsartan are generally taking a dose less than 320 mg daily. Such a trial would apply to fewer than 10% of patients with chronic heart failure and therefore would have little practical value.
As a general rule, when you develop a new drug, it is a good idea to conduct a trial that keeps everything else the same, except for the addition of the novel drug. PARADIGM-HF does not follow this model and, as such, will always be open to interpretation and hard questions.
The facts are: Valsartan was escalated to 320 mg, whereas enalapril was given at 20 mg (and not pushed to 40 mg). Blood pressure was different between arms, favoring the ARB. These were predominately patients with NYHA class I and II HF, so the enalapril dose could likely have been pushed to 40 mg to make the comparison fair. In any case, the protocol could have called for it. Unfortunately, the protocol did not ask providers to do this, while using the maximum HF dose of valsartan.
Readers can decide if 320 mg of valsartan versus 20 mg of enalapril is fair. Consider this trial on the safety and efficacy of valsartan versus enalapril in heart failure.
CONSENSUS enrolled only NYHA class IV patients, while these were fewer than 1% of patients in PARADIGM-HF. Many patients in PARADIGM-HF might well have been able to tolerate and benefit from enalapril 40 mg. Perhaps that would have resulted in similar BP between arms.
I want to see a study of whether sacubutil does anything by itself. This is typically the sort of study done when a novel agent seeks regulatory approval.
The fact that many (and often industry sponsored) studies use drug run-in periods is not a justification for their use.
I would not approve a drug lacking a fundamental demonstration of efficacy with a legitimate control arm. But I can be convinced that sacubutil is a good drug. In fact, I hope it is, for the sake of our patients. However, PARADIGM-HF is not the trial that should convince the regulators.
It appears that Dr. Prasad and I agree that the PARADIGM-HF trial demonstrates that LCZ696 is superior to enalapril 10 mg BID. The major point of our disagreement is why this was so. My coauthors and I attribute the superiority to neprilysin inhibition. If the superiority were due solely to the dose of valsartan that we used, we would have seen more adverse effects related to inhibition of the renin-angiotensin system in the LCZ696 group than in the enalapril group, not fewer.
David Powell, MD, FACC Springfield, NJ
Valsartan alone (without ACEI) to my knowledge has a minimal evidence base for chronic CHF (as opposed to post-MI). A small subgroup of ValHeFT is all we have (and this was limited to ACEI-intolerant patients). Combined ACEI/ARB has more side effects and negative data from ONTARGET. And spironolactone has an excellent evidence base, excluding my use of ACEI/ARB combination.
The run-in mimics real life with one exception: Patients needed to tolerate ACEI to be enrolled. But is there any reason to think that ACEI intolerance would negate the benefit of ARB/neprilysin inhibitor? Certainly not the ARB benefit. Do patients with ACEI intolerance (mostly cough) have higher levels of bradykinin or other neurohormones that offset the benefit of neprilysin inhibition? I don’t know, but unless convincingly shown, it wouldn’t affect my usage of the new drug. So the tolerance of the new drug tips the scale even further in favor.
I have a reservation about African Americans with respect to angioedema risk. Is this an issue? Is there a way to predict angioedema risk (bradykinin levels)? Is the African American subgroup large enough for safety?
Sure, we should have had a good ACEI vs. ARB trial. I am not certain why we didn’t. But now we have one thing that is better than enalapril with fewer side effects. I think an ARB vs. ARB/neprilysin inhibitor trial would have to be confined to ACEI-intolerant patients.
Harsh Golwala, MD Louisville, OK
I am wondering is there any reason to compare ARB+sacubutil vs. ACE inhibitor — instead of ARB+ sacubutil vs. ARB (with ARB in equivalent doses in both the groups) or ACE inhibitor + sacubutil vs. ACE inhibitor alone (again with ACE inhibitor in equivalent doses). In the end, we all know that enalapril trumps everything, as it has been studied and has shown a mortality benefit in HFrEF patients. So if LCZ696 trumps enalapril (either due to higher dose of valsartan in itself or due to addition of newer drug or combination of both), LCZ696 will be the frontline treatment in the near future.
Enrique Guadiana, MD Monterrey, Mexico
There is no such thing as a perfect trial. Every answer raises new questions. Only for the investor the True is absolute. We are physicians, and we don’t work with absolutes. The results of this study are very interesting — it could be a new door. Why the hurry? Respond to the new inquiries and advance medicine. Remember we are physicians, not investors. I echo all the questions my colleagues have and will wait for the answers.
Westby G Fisher, MD Evanston, IL
Drs. Prasad and Packer both make excellent points. Dr. Prasad’s concerns (which I share) have been made more relevant by our experience with Zetia (another combination drug) that was marketed with much fanfare, only to be later shown to have little benefit over simvastatin alone.
By the way, I’m an EP, not a CHF specialist — but I know enough to understand the incredible influence of media on perceptions of a trial, and the gushingly positive spin this trial has received without critical challenge to stock market and drug pricing. (Cynical, I know, but when every cardiovascular news organization is in bed with pharma these days to assure their survival, we are appropriately cautious.) Time will tell who’s right here, but one thing social media has provided is an educational forum to get these real concerns of this trials design aired and rebutted quickly.
I applaud both Dr. Prasad and Packer’s bravery to discuss their respective perspectives openly and I hope they each understand the value of their discussion to those of us on the (real) front lines of heart failure care.
Larry Husten, PhD New York, NY
Wes, I don’t understand the comparison with Zetia. First of all, this was a drug that got on the market (and was indeed heavily marketed) with absolutely no outcomes data. It was approved on the basis of a surrogate endpoint. LCZ696 isn’t on the market yet, and there has been zero marketing so far — yet it already has more outcomes data than 99% of all the approved drugs.
While I am not a CHF trial expert, neprilysin inhibitors are a new class of drug for the treatment of CHF (and it would seem a fairly narrow CHF population at that). While the PARADIGM-HF results are impressive, the question remains: How much of the CHF improvement was due to the new class of drug vs. its co-drug? While the number of patients studied was impressive, it is unfortunate that we really can’t see how much effect this new class of drug has on HF treatment. Instead, the investigators chose to study the med against an entirely different class of drug (albeit one that’s effective in CHF rx) for the reasons Dr. Packer explains. I think more skeptical types (like me) would like to know what additional value a drug like this adds to our HF treatment options, since this drug is likely to be very expensive.
Increasingly, the cost of any new drug must be weighed against its clinical benefit since our patients are absorbing more and more drug costs themselves. Gone are the days where we can freely jump from one drug to the other without directly determining a drug’s benefits to our patients. It would seem to me that the “value” of such a new drug should be demonstrated before we abandon old therapies that have stood the test of time (like enalapril) for new ones.
Please add your comments to this discussion!
As you can imagine we have considered the possibility that valsartan 160mg bid gives more RAS blockade than enalapril 10mg bid. There is no evidence that this is the case – this dose of valsartan has been compared with enalapril 10mg bid and shown no difference (Lee YS et al Effect of valsartan on N-terminal pro-brain natriuretic Peptide in patient with stable chronic heart failure: comparison with enalapril. Korean Circ J. 2011;41:61-7). Valsartan 160mg bid has, of course, also been compared with captopril 50mg tid in VALIANT and found to be similar in efficacy (Pfeffer MA, et al Valsartan in Acute Myocardial Infarction Trial Investigators. Valsartan, captopril, or both in myocardial infarction complicated by heart failure, left ventricular dysfunction, or both. N Engl J Med. 2003;349:1893-906). Even, if valsartan 160mg bid did give more RAS blockade than enalapril 10mg bid, this couldn’t really explain the benefits seen – the dose response studies with RAS blockers have shown little incremental benefit with higher doses and even the trials adding an ARB to an ACE inhibitor did not show the magnitude of benefit we saw in PARADIGM-HF.
Lastly, as Dr Prasad will know, the original CONSENSUS target dose of enalapril was 20 mg bid but this caused unacceptable intolerance. Only 22% achieved this target and the mean daily dose achieved was 18.4 mg (less than the 18.9 mg in PARADIGM-HF). There has even been a study comparing 60mg enalapril daily to 20mg (Nanas J, et al. J Am Coll Cardiol. 2000;36:2090-2095). In that trial, 32.5% and 72.5% of patients, respectively, reached target dose by 3 months. There was no statistically significant or clinically meaningful difference between groups for change in blood pressure, heart rate, LVEF or NYHA class between dose groups. In other words, the evidence we have says that not many patients tolerate more than enalapril 10mg bid and that higher doses don’t seem to have a greater pharmacological effect.
Run-in periods are certainly much discussed but not always viewed so negatively (Pablos-Méndez et al Run-in periods in randomized trials: implications for the application of results in clinical practice. JAMA. 1998;279:222-5). Apart from reflecting what we do in clinical practice (i.e. start a treatment, see if patients tolerates it and continue it only if they do), there were several specific reasons why we had to do our study in this way and these are detailed in the design paper (McMurray et al Dual angiotensin receptor and neprilysin inhibition as an alternative to angiotensin-converting enzyme inhibition in patients with chronic systolic heart failure: rationale for and design of the Prospective comparison of ARNI with ACEI to Determine Impact on Global Mortality and morbidity in Heart Failure trial [PARADIGM-HF]. Eur J Heart Fail. 2013;15:1062-73). One was the regulatory necessity of ensuring the enalapril treated patients received an evidence-based dose, proven to reduce mortality. There was no point in testing LCZ696 against an unproven treatment.
I don’t understand the statement, “It makes little sense to stop what is working to enter this trial.” Why would one not be willing to change a patient to a potentially better treatment (either LCZ696 or a proven dose of a proven ACE inhibitor – or even an ACE inhibitor rather than an ARB, given the more robust evidence for an ACE inhibitor compared to an ARB)? Medicine will not make much progress if we can’t consider replacing old treatments with new ones. The other proposed designs either test invalid questions (e.g. A – valsartan 160mg bid is not a first line treatment for heart failure – enalapril is) or are impractical (e.g. B – the minority of patients are on an ARB and the key clinical question is can we do better than an ACE inhibitor – that is our gold standard?). The third proposal is effectively what we did.
I don’t know what providers will think, but guideline committees and regulators have already made it very clear that valsartan 160mg bid is an acceptable second line therapy to an ACE inhibitor (such as enalapril 10mg bid), but an ACE inhibitor is the preferred and first-line treatment and the one we are sure of about improving outcomes. I’m unaware of anyone previously suggesting valsartan 160mg bid is superior to an ACE inhibitor or that an ARB should be preferred to an ACE inhibitor – indeed, quite the contrary.
I thank Dr. McMurray for his comments.
I agree with the PIs of PARADIGM HF that the trial shows that LCZ696 given at a mean dose of 375mg outperforms enalapril when given at a mean dose of 18.9mg among patients who can tolerate enalapril 20 (for 2 weeks) followed by a dose escalation of LCZ over 4-6 weeks. This is a fact. However, this is not the important question that now faces clinicians and regulators.
LCZ696 is not the novel drug in PARADIGM. Instead, the novel drug is sacubutril. Sacubutril is an inhibitor of neprilysin, a class of medications of which no prior drug has received approval, and of which no prior drug has shown a meaningful benefit. PARADIGM HF seeks to justify the use of a new drug from this class among millions of Americans.
My position is this: The design of PARADIGM-HF cannot allow us to conclude what benefit, if any, is added by sacubutril. Neither can it allow us to provide a clear appreciation of the harms of therapy. This is not some arbitrary requirement, but the standard any new drug must meet before coming to market.
By itself, no single aspect of trial design leads to this unfortunate conclusion. Instead, a combination of choice of concomitant drug, comparator, dosing of drugs, use of a double run in period, and the disparate length of time and up titration during that run in—together make interpreting the role of this novel drug quite uncertain.
My burden is not to show that valsartan 320mg is unequivocally better than enalapril 20mg. In fact, at the outset, I agree, this cannot be shown. Instead it is to show that we cannot be certain it is not. The arguments of the study PIs rely on assumptions; I believe that for this question we cannot settle for assumptions but require scientific proof.
Enalapril and Valsartan dosing
The offered argument seeks to have it both ways. On one hand, Drs. Packer and McMurray argue that enalapril 40 should not have been the comparator, as it would not have been able to be tolerated (CONSENSUS), and on the other hand, they argue, that even if it had been tolerated there would be no additional benefit (Nanas J, et al.).
Here is how I interpret CONSENSUS. CONSENSUS enrolled only NYHeart Class IV patients and was conducted between 1985 and 1987. The patients in PARADIGM were mostly NY Heart Class II. 1 year survival in CONSENSUS was 60 percent, and ~94% in PARADIGM. For these reasons, we must conclude the population was different. I do not believe it is fair to extrapolate observations from CONSENSUS to the patients in PARADIGM. We cannot say they could not have tolerated dose escalation. In fact, evidence to the contrary…
Here is how I interpret Nanas. Nanas compared a mean dose of enalapril 17.9 mg/day, to 40.2 mg/day in 122 and 126 patients respectively. This trial better reflects the NY Heart status of patients as those in PARADIGM. Nanas shows precisely that PARADIGM patients COULD VERY WELL HAVE tolerate dose escalation. Regarding the lack of efficacy in that trial—the trial was not truly powered to look for differences in mortality.
Finally, Dr. McMurray cites a trial by Lee S, et al. This is an open label RCT where 602 patients were randomized to 320mg of valsartan or 20 of enalapril. Outcomes were reported for (246 and 199= 445 patients). Loss was mostly due to dropouts. The authors examined short term endpoints and found no real difference, as Dr. McMurray notes. Interestingly the mean administered dose of the medicines was 99mg for valsartan and 10.7mg for enalapril. In short the trial was underpowered, had high attrition, did not achieve similar dosing as PARADIGM. So I am not sure what to conclude from this study.
The problem with debating the dose of the medication is that there is no equivalence study of 320mg of Valsartan to 20mg of enalapril. Equivalence trials would need to be much larger to show that the two drugs have similar efficacy at these doses. In the absence of such studies, it is fair to believe as the PIs do that—they are unlikely to be very different. In many clinical contexts this might be fine, but when a novel drug seeks regulatory approval, it is not good enough. You have to prove that the reason the outcomes you observed were the way they were because the novel drug adds something.
Drug run in periods
We will just have to disagree about run in periods in general, which I have explained at length above. Let me focus here on the duration of the run in period. The length of time of the drug run in was twice as long for valsartan-sacubutril than enalapril (median 29 vs 15 days). This imbalance provides an advantage for the study drug, allowing dose escalation during run in, and giving more time for patients to declare themselves intolerant.
Renal dysfunction/ hyperkalemia
It is very hard to draw conclusions about the efficacy of doses from side effects of these drug, particularly given the run in, as I’ve described.
When Patients Ask
When a patient asks me what is the benefits and harms of sacubutril? I will honestly be unable to answer. This is because of the choice of design of PARADIGM HF. Further studies may help provide a better answer to this question.
When a patient asks what is the probability that sacubutril adds something? I will also be unable to answer as no statistical test in this trial addresses that crucial efficacy question.
The Right Trial
I continue to believe the best trial for this drug would be an RCT of sacubutril versus placebo among patients already on an ARB. This is no trivial population (or market share), 22% or nearly 2000 patients of those who made it as far as randomization (8442/10513) in PARADIGM were on such a drug at baseline. Thus, the trial I propose could easily have been conducted in this group, and would have answered a fundamental efficacy question before a larger market share was sought.
The Bottom line
In short LCZ696 at the dose it was given beat enalapril at the dose it was given after a group of patients already on an ACE and ARB were asked to stop, and undergo a double run in period.
Whether sacubutril is a good drug or bad drug cannot be answered by this trial—not without making assumptions. The study PIs believe those assumptions are reasonable. And, I don’t think we should have to rely on assumptions. Typically this is not the standard required before a new drug comes to market poised to be given to millions of patients.
The enalapril dose question still puzzles me. We thought long and hard about this when designing the trial. Since CONSENSUS, there have been 4 large trials using forced titration of enalapril to a target dose of 10mg bid. The two SOLVD trials had an active run-in and forced titration and still only managed mean daily achieved doses of 16.6mg (Treatment trial) and 1.7mg (Prevention trial). V-Heft II managed 15mg. In the OVERTURE trial it was 17.7. In a more recent but smaller trial, CARMEN with the same target dose, the mean achieved dose was 16.8 mg when enalapril was given alone but only 14.9 mg when combined with a beta-blocker.So I would still contend we achieved as large a dose of enalapril as would be possible in this patient population. Maybe more importantly, what evidence is there that more intense RAS blockade could account for our findings? Neither ATLAS nor HEAAL (and ACE inhibitor and ARB dose-response study showed any effect of higher dose therapy mortality. Even ADDING a very large dose of candesartan (target 32 mg daily) or the dose of valsartan deliver by LCZ696 200mb bid (160mg bid) to an ACE inhibitor did not lead to benefits like those we saw with combined neprilysin-RAS inhibition. For these reasons, I find it hard to accept the argument that the results of PARADIGM-HF could be explained by more RAS blockade in the LCZ696 arm.
Regarding adding sacubitril to a patients usual ARB therapy, it is highly unlikely such a study would impress any regulatory agency (or the clinical community). Firstly, no agency or guideline recognizes an ARB as the standard of care and secondly adding a new agent to a sub-optimal dose of an ARB (as would happen if the design was “add to usual therapy”) would also cut no ice – to be credible, we would have had to conduct a trial where we added sacubitril to valsartan titrated to a target dose of 160mg bid (or candesartan to 32 mg qd). If we had done this and beaten the ARB, the first question to us would have been “but could you beat an ACE inhibitor – that is the first-line standard of care?”
The duration of run-in argument isn’t fair – obviously with vast majority of patients having been on an ACE inhibitor for weeks, months or years prior to enrollment it was the ACE inhibitor that was favoured!
Sorry, I stupidly forgot CIBIS-3 which showed the mean achieved daily dose of enalapril was 17.2 mg when enalapril was started before bisoprolol; when bisoprolol was started first and enalapril later the mean daily achieved dose of enalapril was 15.8 mg. In other words this study shows the same as CARMEN – and suggests that in the modern era of heart failure treatment, when most patients also get a beta-blocker (unlike in the days of either CONSENSUS or SOLVD), it is harder to up-titrate the dose of enalapril to the levels achieved in the past.
Just a quick response to Dr Valencia. There really was no heterogeneity across subgroups. Please note the subgroup figure shows 36 statistical tests – so by the play of chance alone we might expect a spuriously “significant” finding. Our paper had a strict word limit and we couldn’t go into detail but if we looked at NYHA class at the start of run-in (as opposed to randomization, as in the manuscript), the p value for interaction was no longer “significant”. Similarly, if you look at duration of heart failure (time since diagnosis) no interaction is seen (it being likely that longer duration is associated with greater severity). Also, in an unpublished analysis we looked at baseline KCCQ score by tertile (as an alternative way at looking at symptom severity) and so no interaction.
As mentioned in the paper, we also carried out an analysis adjusted for both baseline BP and change in BP from baseline and neither changed the magnitude of effect of LCZ696.
Thanks Dr McMurray for your points.
I have done an approach to subgroup analysis as recommended by Rothman (Rothman KJ. Six persistent research misconceptions. J Gen Intern Med. 2014;29:1060-1064).
Using BP at randomization as the reference in the analysis, may have underestimated the true differences in BP (as suggested by figure in page 7 of the appendix) and their potential role in the results.
I think we will continue to disagree on whether the patients in PARADIGM could have tolerated more enalapril, and what this might have accomplished.
SOLVD, OVERTURE, CARMEN and V-HEFT II all pushed the dose to a cap of 20mg, as Dr. McMurray states.
In these trials, the subset of patients who achieved 20 might have been able to tolerate more, so the mean level might have been higher than 16.6 if you took away the upper cap.
Simply because the mean is lower than the target does not imply the mean could not be higher if you pushed the dose in the fraction of patients who are at target.
Which brings me back to the one trial that did remove the upper cap, Nanas shows one thing very clearly, and that is that a mean dose of enalapril of 40mg can be achieved in a comparable NY Heart Classification population. If the dose of enalapril achieved in PARADIGM-HF is as large as possible, then the protocol need not have specified enalapril 20mg as the target dose, and instead advised providers to escalate enalapril as tolerated.
Again, as I stated, I cannot unequivocally show that the dose of enalapril would have made a difference, or the benefit in PARADIGM would have been lost if you used valsartan 160 or 80. But, the point is, we cannot be sure. There is no equivalence trial of Valsartan 320 to Enalapril 20.
The open question
Dr. McMurray makes an excellent point. If the sponsors of PARADIGM-HF had conducted the trial I propose: an RCT of sacubitril or placebo in the population of HF patients already on an ARB. ARB is standard of care for patients who cannot tolerate an ACE (CHARM), and 22% of patients in PARADIGM fit this description– so this is not a trivial population. There would be different lingering questions. If the trial were negative, there would be none. But, if the trial were positive, the lingering question would be, as he correctly notes: well, thats fine for my patients on ARBs, but what about my patients on ACE inhibitors? My trial would not have had a run in– so we wouldn’t have had to debate that.
No clinical trial can answer every question, and here we see two sets of lingering questions from two different trials. If you conduct PARADIGM-HF the lingering questions are: How much did sacubitril add? How much did valsartan 320 add? How good is enalapril 20?— as we have been discussing. If instead, you conducted the trial I propose (and it was positive), we would be asking whether the results apply to broader populations?
The trial PIs and sponsor seem to believe that it is better to live with the former questions. And, I strongly believe it is better to live with the latter questions. When a novel drug (Sacubitril) seeks to be used in millions of patients for the first time, we should have confidence that its isolated benefits outweigh harms. The trial I propose applies to fewer patients, and doesn’t answer everything, but answers a fundamental biological question about sacubitril– and that is what doctors and patients need.
Hopefully my last comment because I think we have probably argued each of our perspectives as far as we can! I think your trial would still need a run-in or something like it because if your patients had not been forced to the target (or maximum tolerated) dose of valsartan or candesartan, the question would linger as to whether you could have achieved the same effect from maximizing the dose of ARB as you obtained from adding sacubitril to a sub-optimal dose of ARB (assuming the study was positive). Very much the debate following Val-HeFT and CHARM-Added, especially after VALIANT.
I must admit that I am a little dumbfounded by the degree of skepticism being hurled at PARADIGM-HF. You have a disease state that has a higher mortality rate than many cancers, with no non-device based therapeutic advances in decades, and a trial comparing a therapy to the standard of care being stopped early after showing a 20% reduction in risk of hard cardiovascular outcomes. With much of the pipeline at major pharma companies being switched over to oncology drugs where the path to approval requires FAR less convincing data, I think we should be congratulating the investigators and Novartis for moving forward with this study. Certainly, given the track record of trials in heart failure and the difficulty in getting medications approved for CVD, this require a massive leap of faith. One can monday morning quarterback the best of papers and the back and forth could fill many pages of this website, but I believe that this was a well-conduced and convincing study, and that it would be unethical to not offer this therapy to our systolic HF patients in the future based on the data presented.
The mortality rate seems very high given that so many patients were Class 1 or 2 compared to other trials. why is this and is this poulation different from what we generally see in America?
With some confidence, we know the following: sacubitril 40mg + valsartan 160mg BID is better than enalapril 10mg BID for patients who present like the PARADIGM HF population (i.e. NYHA I/II patients with a 3-year risk of death of ~20%); the effect of sacubitril alone versus ACE inhibition is arguably unknown; the effect of the new combination versus higher doses of ACE inhibition is unknown.
Given the above, what is sacubitril’s place in CHF therapy? Can sacubitril/ARB really be a First Line alternative to ACEI? Frankly, we know very little about the safety of sacubitril because PARADIGM-HF was stopped early and RCTs usually fail to detect important safety outcomes. How can a cardiologist reconcile a First Line sacubitril prescription with his “first do no harm” pledge?
Sacubitril 40mg + valsartan 160mg BID is probably more effective than enalapril 10mg BID for some CHF patients. I don’t know if it’s better than anything else. I know nothing about its long-term safety.
Disclaimer: not a cardiologist; a medical student and pharmacist
I would say the following:
Given the above, what is sacubitril’s place in CHF therapy?
– In many patients with systolic LCZ696 would be superior to ACE-I
Can sacubitril/ARB really be a First Line alternative to ACEI?
– Based on the trial, yes
Frankly, we know very little about the safety of sacubitril because PARADIGM-HF was stopped early and RCTs usually fail to detect important safety outcomes.
– Trials get stopped for multiple reasons that have been well worked out, including evidence of harm and clear evidence of benefit, its ethical to stop for both extremes.
– RCTs do pick up on “important safety data” in many cases; for rarer safety issues or those that will develop over a long period of time, registries might be needed. But that should not slow down approval of a therapy that passed the test of a large trial, as then no one would even develop any therapies and/or therapies would take decades to show they are safe
How can a cardiologist reconcile a First Line sacubitril prescription with his “first do no harm” pledge?
– This seems a little melodramatic given the data.
Even if the early termination is not grounds for questioning the trial’s safety data, it has to be a reason to question the treatment effect size, no? Shortened trials (even if they’re methodologically pristine; even if their event rate is high) usually exaggerate the treatment effect size.
Making both the trial data and the methods of determining the stop criteria available to independent investigators would ease some of the scrutiny, I think.
J Clin Oncol. 2009; 27(10) (cancer trials)
JAMA. 2005; 294(17) (general trials)
This study concerns me from several standpoints. The initial design is flawed in that 2 drugs (LCZ696 & valsartan) were compared to 1. Moreover, the compared enalapril is NOW an uncommonly prescribed ACE inhibitor in clinical practice. Why not use more clinically preferred drugs such as lisinopril, ramipril, or benazepril? Or why not choose as the comparator the insurmountable ARB’s such as telmisartan, azilsartan, or olmesartan. Finally, why not make it a “2 vs 2” comparison using either the just above mentioned ACE I’s or ARBs + epleronone or spironolactone in APPRORIATE doses. Frankly, this study strikes me as hysteria in the cardiological elite as well as in the editors for the presented and published studies. HRS, MD, FACC