AMGN/REGN/SNY: Top-line ODYSSEY results do not show a mortality benefit

Print Friendly
Share on LinkedIn0Tweet about this on Twitter0Share on Facebook0


Richard Evans / Scott Hinds

203.901.1631 /.1632

revans@ /


March 20, 2018

AMGN/REGN/SNY: Top-line ODYSSEY results do not show a mortality benefit


  • ODYSSEY top-line results show an 0.026 nominal p-value for all-cause mortality for all comers (all baseline LDL-C’s); the ‘nominal’ caveat implies the standard, pre-specified test for significance on all-cause mortality failed
  • Post-hoc analysis shows that hazard ratios for CV, CHD, and all-cause mortality are highest for the subgroup with baseline LDL-C’s at or above 100 mg/dl. Concluding that results differ by sub-group is only valid if differences across sub-groups are large enough to reject the default interpretation that results are in truth equivalent across subgroups. ODYSSEY fails to reject this null hypothesis, which invalidates the implied finding of a mortality benefit in patients with higher baseline LDL-C’s. The net result is that as presented, the ODYSSEY top-line results failed to prove a mortality benefit for Praluent
  • With no mortality benefit to jump-start the class, the PCSK9 market only gets rolling if the companies agree to a smaller target patient population, and/or significantly lower prices. ODYSSEY also fails to show that reduction in major adverse cardiovascular events (MACE) is greater in patients with higher baseline LDL-C’s, though the results are directionally consistent with this conclusion. We doubt FDA will ever agree that efficacy (MACE or mortality) is better at higher baseline LDL-C’s; however, payors may be willing to agree to coverage restricted to patients above the 100 mg/dl threshold. This would reduce the potential PCSK9 patient population from 5.6M US persons (those who meet the FOURIER inclusion criteria) to 2.4M (those who meet the ODYSSEY inclusion criteria, and also have LDL-C’s ≥ 100 mg/dl
  • REGN/SNY expressed a willingness to negotiate prices in line with ranges suggested by ICER; however, this willingness may be predicated on an expectation of claiming a mortality benefit in patients at or above LDL-C of 100 mg/dl, which we see as unrealistic. More likely is that ODYSSEY only supports a claim of reduced MACE without regard to baseline LDL-C, an outcome for which ICER calculates a value-based pricing range of $2,306 – $3,441 per patient per year. Current US net pricing for the PCSK9’s averages approximately $8,175
  • At the ICER price range, the PCSK9’s probable high COGS becomes a potential rate limit; this may open the door to MDCO/ALNY’s inclisiran (an RNA interference based inhibitor of PCSK9), if this product can be made more efficiently than Praluent and Repatha

The ODYSSEY mortality signal appears too weak to support a labelled claimThe cardiovascular outcomes trial for REGN/SNY’s PCSK9 inhibitor Praluent (ODYSSEY) reported an identical primary endpoint hazard ratio (0.85, Exhibit 1) to AMGN’s PCSK9 Repatha in its pivotal outcomes trial (FOURIER). The two trials used similar primary endpoints (time to first major adverse cardiovascular event, or MACE); however, MACE definitions and patient populations differed somewhat across the trials[1],[2]. Based only on primary endpoints, at the very least ODYSSEY puts Praluent on equal footing with Repatha

The far more critical question is whether ODYSSEY shows mortality benefit – and we care about this less from the perspective of Praluent v. Repatha market share, and more from the perspective of whether ODYSSEY’s mortality signal might be sufficient to create momentum for the PCSK9 class overall

FOURIER showed no mortality benefit for Repatha, which no one disputes. In contrast REGN/SNY framed the top-line ODYSSEY results as providing evidence of a mortality benefit for Praluent; however, the strength of ODYSSEY’s mortality signal is questionable

On an all-comers basis ODYSSEY failed to show improved CV or CHD mortality specifically but is being presented as having shown evidence of improved all-cause mortality. The problem is that the p-value provided for all-cause mortality (0.026) is footnoted as ‘nominal’, as compared to the standard log-rank p-values given for all other hazard ratios (HRs). The clear implication is that the log-rank p-value for all-cause mortality is not significant, though we have no way of knowing this with certainty until and unless the companies release this value – a disclosure which will be necessary in the context of regulatory applications, and highly likely in the context of peer-reviewed publication(s)

The companies further disclosed mortality HR’s by baseline LDL-C; numerically superior HR’s were given for all-cause mortality (0.71 (0.56, 0.90)), CHD mortality (0.72 (0.53, 0.98)) and for CV mortality (0.69 (0.52, 0.92)) in patients with baseline LDL-C’s of ≥ 100 mg/dl. Because the high-ends of these ranges all fall below 1.00, the inference is that ODYSSEY shows statistically significant evidence of a mortality benefit in patients with LDL-C’s ≥ 100 mg/dl. However, if we interpret ODYSSEY according to the statistical standards we believe FDA will follow (which are enforced by at least one major medical journal[3]), ODYSSEY failed to prove a mortality benefit, at least according to the top-line results thus far disclosed

In a slide showing the all-cause mortality curves for the three LDL-C sub-groups, the companies disclosed the p-value (0.12) for interaction among the three groups. If this p-value reflects the result of the standard test for interaction among groups, then it shows that the null hypothesis (all-cause mortality across all groups is the same) cannot be ruled out. And, if the null hypothesis cannot be ruled out, this means that the test for significance including all baseline LDL-C’s – which apparently failed (‘nominal’ p-value 0.026) – is the one that matters. The companies did not disclose p-values for interaction among groups for CHD or CV mortality. We speculate that the companies have shown the mortality data in their best light, the inference being that the CHD and/or CV mortality findings in the ≥ 100 mg/dl LDL-C subgroup are likely to be on even weaker statistical footing than the all-cause mortality finding

On net, our read of the ODYSSEY result is that Repatha and Praluent have similar effects on the primary MACE endpoint, and that Praluent’s label is likely to be updated for reduction of MACE along lines similar to Repatha’s. The mortality signal in ODYSSEY appears too weak to result in a mortality claim of any type, including one limited in scope to patients with baseline LDL-C’s of ≥ 100 mg/dl

For the record we thought that the PCSK9’s LDL-C lowering power was enough to drive demand when they were first approved; and we subsequently thought that the FOURIER outcomes result (reduction in MACE, but no mortality effect) was enough to kick-start the class. Both predictions were wrong. We’d love to argue that ODYSSEY has added further outcomes evidence for PCSK9’s, but it simply has not – the mortality signal is almost certainly too weak to find its way into the label, and/or to serve as a motive for widespread formulary adoption

How to get PCSK9 sales moving: shrink the patient population and/or lower the price

If ODYSSEY kickstarts the PCSK9 class, it won’t be by proving mortality, it will be because SNY/REGN and payors use ODYSSEY to define a limited scope of patients for whom PCSK9’s are suitable. In any statistically rigorous sense ODYSSEY does not support a conclusion that Praluent is more effective in patients with baseline LDL-C’s ≥ 100 mg/dl (whether for MACE or mortality), though the findings are directionally consistent with that interpretation. This is loose-and-fast with trial results in a way that will never fly at FDA, but payors might accept the ≥ 100 mg/dl threshold as a coverage limitation, simply because this this meaningfully limits the patient population and reduces financial exposure. We estimate that 5.6M insured US persons meet the inclusion criteria for Repatha’s FOURIER trial (3.7M Medicare, 1.9M private; recall that FOURIER showed efficacy regardless of baseline LDL-C). In contrast, just 2.4M persons[4] meet the inclusion criteria for ODYSSEY, and have a baseline LDL-C ≥ 100 mg/dl (1.4M of whom are ≥ 65 years old). The PCSK9’s have a much better shot at getting payors to greenlight the drug for a population of 2.4M than for a population of 5.6M

Shifting to price, ICER released a new evidence update to its PCSK9 value findings; assuming no mortality benefit is found, the report supports a value-based price benchmark range of $2,306 – $3,441 for all ODYSSEY eligible patients, and $4,460 – $6,578 for patients with baseline LDL-C’s at or above 100 mg/dl. The higher value-based price range for patients at or above 100 mg/dl presumes that Praluent has a greater impact on MACE at or above the 100 mg/dl threshold, something ODYSSEY does not support, at least in any statistically rigorous sense. The companies reported a p-value for interaction among LDL-C subgroups of 0.09 for MACE HR’s, which plainly argues that ODYSSEY did not show a statistically meaningful difference in MACE efficacy according to baseline LDL-C. This weakens the argument that the $4,460 – $6,578 range should apply, even if the drugs are limited to patients with baseline LDL-C’s ≥ 100 mg/dl. ICER further calculated a value-based price benchmark range of $5,234 – $7,975 assuming a reduction in all-cause mortality for patients with baseline LDL-C ≥ 100 mg/dl, though as we’ve argued we don’t see a mortality claim as realistic

As a point of reference, we estimate current average net pricing per year of therapy in the (US) PCSK9 class to be $8,175, with current class-leader Repatha at $8,842, and Praluent at $7,120 (Exhibit 2) – all well above the ICER value-based benchmark range ($2,306 – $3,441) that best fits a statistically rigorous interpretation of the ODYSSEY data (reduction of MACE independent of baseline LDL-C)

We’re not suggesting ICER actually sets the pricing for PCSK9’s, but they do put forward credible value-based pricing ranges, and the gap between current pricing and the applicable ICER range argues net pricing will have to come down considerably before payors start approving PCSK9 claims

As price falls, COGS becomes A LOT more important; this may shift the near-term advantage to Praluent, and the longer-term advantage to inclisiran (MDCO/ALNY)

Both Praluent and Repatha are monoclonal antibodies (MAbs) and are given at relatively high milligram doses (1,950 mg to 3,900 mg annually for Praluent; 3,640 mg to 5,460 mg annually for Repatha). MAbs are traditionally somewhat expensive to manufacture on a per-mg basis; we have no way of knowing the production economics of these two specific products, but we can at least look for approved MAbs whose COGS are disclosed. Using company disclosures, we were able to closely estimate COGS for Erbitux ($0.30/mg), Soliris ($1.59/mg), and Adcetris ($11.09/mg). If COGS/mg for the PCSK9’s approach the levels of either Soliris or Adcetris then the products wouldn’t be commercially viable at their current prices, much less the lower ranges proposed by ICER. So entirely for the sake of argument we’ll assume the Erbitux COGS/mg is at least crudely informative. Admittedly we have nothing else to go on, but at least the molecular weight of Erbitux (152 kDa) is on par with the molecular weights of the PCSK9’s (Praluent 146 kDa, Repatha 144 kDa)

At the Erbitux COGS of $0.30/mg, COGS/sales for Praluent would be 8% to 16% for the low and high doses, respectively, at current prices. COGS for Repatha would be 12% and 20% at current prices for the low and high doses, respectively. At the $5,519 mid-point of ICER’s range ($4,460 – $6,578) for MACE efficacy at or above 100 mg/dl, COGS/sales ratios for low and high doses rise to 11% and 21% for Praluent, and to 20% and 27% for Repatha. At the $2,874 mid-point of ICER’s range ($2,306 – $3,441) for MACE efficacy independent of baseline LDL-C, COGS/sales ratios for low and high doses rise to 20% and 41% for Praluent, and to 38% and 53% for Repatha (Exhibit 3a). Of interest, again assuming the Erbitux COGS level is applicable, the brands should be nearly indifferent between a market with 2.4M potential patients at $5,519 net price per year (the number of patients qualifying for ODYSSEY and having LDL-C’s ≥ 100mg, and the ICER price assuming MACE efficacy at or above 100 mg/dl LDL-C), and one with 5.6M patients at an annual net price of $2,874 (the number of patients qualifying for FOURIER across all baseline LDL-C’s, and the ICER price for MACE efficacy across all LDL-C’s). Figures in Exhibit 3b assume Erbitux COGS, an 80/20 split across low and high doses for each brand, and equal market shares

Assuming the PCSK9 market can’t get rolling without significantly lower prices, this raises the question of whether MAbs may simply be too expensive to manufacture at practical PCSK9 price points. MDCO/ALNY’s inclisiran is an RNA interference (RNAi) based means of reducing PCSK9 levels, dosed at 300mg every six months, or 600mg annually[5]. Inclisiran produces LDL-C reductions on the order of 50%, just shy of the reductions achieved by Praluent and Repatha. We have no immediate means of estimating COGS for inclisiran and are wary of concluding it’s cheaper to manufacture an annual dose of inclisiran simply because fewer milligrams are involved. As an RNAi inclisiran manufacturing is completely distinct from the techniques used to produce MAbs such as Praluent and Repatha. However, IF inclisiran can be made more cheaply than Praluent and/or Repatha, inclisiran might be more viable at likely PCSK9 prices. Inclisiran is in pivotal phase 3 trials having estimated primary completion dates in 2H19

ESPR’s bempedoic acid also (ETC-1002) is in late stage clinical testing, with results from its largest safety / efficacy trial also due in 2H19. Bempedoic acid produces LDL-C reductions on the order of 25%. Because both ODYSSEY and FOURIER have shown that larger LDL-C reductions are worthwhile, we see inclisiran (50% LDL-C reduction) being far more competitive than bempedoic acid, assuming it can be manufactured efficiently

It’s hard to argue that consensus is too low, but it’s at least too uneven …

Current 2023E consensus (WW sales) for the PCSK9’s totals $4.8B, made up of $2.5B for Repatha, $1.5B for Praluent, and $700M for inclisiran. As reference, 55% penetration (the rate of statin uptake among patients who qualify for a statin) of the 2.4M US persons who meet the ODYSSEY inclusion criteria and have LDL-C’s of ≥ 100 mg/dl yields 1.3M PCSK9 treated patients. At the ICER pricing range (mid-point $5,519 per year) which assumes MACE efficacy but no mortality benefit for patients with baseline LDL-C’s at or above 100 mg/dl, these 1.3M US patients would drive net sales of approximately $7.3B annually in the US alone (Exhibit 4). Alternatively if we choose the lower ICER price reference but expand the patient population to all baseline LDL-C’s (5.6M persons, 3.1M treated at 55% penetration), US net sales potential expands to $8.9B; however as we show in Exhibit 3b gross profit potentials across these two revenue scenarios are comparable for the existing PCSK9 brands

We’re hesitant to imply that consensus is too low, simply because it’s unclear whether price concessions in the wake of ODYSSEY will be sufficient to get the market rolling, and even more unclear what this means for gross profit potentials. And, because inclisiran’s possible launch (late 2019) has drawn nearer as the PCSK9 market has languished, the existing brands face the prospect of a very low (relative) cost competitor that could limit the incumbents’ sales potential. At the very least we’d argue that estimates for Praluent should be at least as high as those for Repatha, and that inclisiran appears to have much more potential to surprise on the upside than either of the existing brands


  1. ODYSSEY defined MACE as coronary heart disease (CHD) death, non-fatal myocardial infarction (MI), ischemic stroke (fatal or non-fatal), or unstable angina (UA) requiring hospitalization. FOURIER defined MACE as cardiovascular (CV) death, MI, stroke, hospitalization for UA, or coronary revascularization 
  2. ODYSSEY patients had a recent (1-12 months preceding randomization) acute coronary syndrome (ACS), and inadequate control of lipids (LDL-C ≥ 70mg/dl, non-HDL_C ≥ 100mg/dl, or apoB≥ 80mg/dl). FOURIER patients had atherosclerotic cardiovascular disease (ASCVD, defined as history of MI, non-hemorrhagic stroke, or symptomatic peripheral arterial disease (PAD)), plus at least one major risk factor (Diabetes, age ≥ 65, MI or non-hemorrhagic stroke w/in 6 months of screening, additional MI or non-hemorrhagic stroke excluding qualifying event, daily smoking, history of symptomatic PAD if eligible based on MI / stroke history, or at least two minor risk factors (History of non-MI related coronary revascularization, coronary artery disease with ≥ 40% stenosis in ≥ 2 large vessels, most recent HDL_C < 40 mg/dl for men or <50 mg/dl for women, most recent hsCRP > 2 mg/dl, most recent LDL-C ≥ 130 mg/dl or non-HDL_C ≥ 160 mg/dl, metabolic syndrome) 
  4. 2005 – 2010 NHANES; US persons with ACS and on statin therapy 
  5. The first year of therapy involves three 300mg doses, at day 1, day 90, and then every six months thereafter 

©2018, SSR, LLC, 225 High Ridge Rd, 2nd Floor, Stamford, CT 06905. All rights reserved. The information contained in this report has been obtained from sources believed to be reliable, and its accuracy and completeness is not guaranteed. No representation or warranty, express or implied, is made as to the fairness, accuracy, completeness or correctness of the information and opinions contained herein. The views and other information provided are subject to change without notice. This report is issued without regard to the specific investment objectives, financial situation or particular needs of any specific recipient and is not construed as a solicitation or an offer to buy or sell any securities or related financial instruments. Past performance is not necessarily a guide to future results. In the past 12 months, through a wholly-owned subsidiary SSR Health LLC has provided paid advisory services to BioPharmX (BPMX), Pfizer Inc (PFE), Gilead Sciences (GILD), Bristol-Myers Squibb (BMY) and Sanofi (SNY) on both securities-related and non-securities-related topics. One or more of SSR Health’s analysts owns long positions in the following stocks: ACOR, AGEN, AGIO, AKAO, ALKS, ALNY, ANAB, ARRY, BMY, DERM, DOVA, DPLO, DVA, ESALY, ESRX, FOLD, GWPH, INCY, IONS, ITCI, KALA, LOXO, MDCO, NSTG, PFSCF, PGNX, PRTK, PTLA, RARE, RHHBY, SRPT, TBPH, TRVN, TSRO, TTPH, TXMD, VRTX, VSTM

Print Friendly