Sanofi and Regeneron welcome the opportunity to comment on the ICER Draft Report on PCSK9 Inhibitors for Treatment of High Cholesterol: Effectiveness, Value, and Value-Based Price Benchmarks.

In summary, we have concerns with the report, which center upon three key domains:

1) Limitations in methods and assumptions.  There are limitations in the Care Value analysis, resulting in likely underestimation of cardiovascular disease (CVD) risk in approved patient populations for the PCSK9 inhibitor class and undervaluation of the potential clinical and economic benefits of these medications. Additionally, variations in key inputs can cause large swings in model outputs.
2) Inadequate visibility into the model architecture. This results in the inability to fully interpret and validate conclusions. Peer review of the model is desirable.
3) Imposition of a narrowly focused Health System Value analysis.  This analysis includes unrealistic estimates of budget impact and establishes arbitrary investment thresholds for societal investment in novel pharmaceuticals and ignores the value and risk of innovation.

Sanofi and Regeneron are committed to developing innovative medicines that make a difference in patients’ lives and ensuring that patients in the U.S. who are prescribed alirocumab are able to access the medicine and receive the support they may need.

We believe alirocumab provides meaningful value to those patients with the greatest unmet need, and carefully considered multiple parameters in establishing an appropriate price for alirocumab. It is also important to consider that specialty biologics require unique manufacturing & supply chain, together with intensive patient and physician support models, neither of which are included in the cost-effectiveness analysis.

We agree with the conclusions of the Comparative Clinical Effectiveness analysis that PCSK9 inhibitors provide substantial or incremental net health benefits while being well tolerated for all relevant patient subpopulations. We also agree with the statements that “the drugs improve intermediate risk factors for cardiovascular disease” and “substantially reduce LDL-C, total cholesterol, lipoprotein(a), and modestly elevate HDL-cholesterol.”

However, we disagree with certain aspects of the modeling assumptions that drive the outputs of the analysis. We are confident in the clinical and economic value that alirocumab contributes to the PCSK9 inhibitor class. We believe appropriate patients should have broad access to innovative medicines. As one input for determining an appropriate price for alirocumab in the U.S., we evaluated the potential value of treatment in our indicated populations. Utilizing our cost-effective modeling approach, for patients with clinical astherosclerotic CVD (ASCVD) and those with heterozygous familial hypercholesterolemia (HeFH), we made efforts to ensure that the net price of alirocumab to payers is cost effective.

Taken together, these limitations in the ICER approach raise significant concerns about the conclusions of the report. Below, we outline our concerns in greater detail.

1. Limitations with Methodology and Assumptions: The benefits associated with the PCSK9 inhibitor class are underestimated in the intended patient populations in the three key areas highlighted below: 1) cardiovascular (re)event rates among CVD patients and its association with LDL-C at baseline; 2) the lifetime risk profile for the familial hypercholesterolemia (FH) population; 3) the age distribution in the model.
Recommendation: Provide detailed descriptions of all data sources, assumptions and model input values with sources, and a complete description of the CVD Policy Model methodology.
a. Risk groups are defined using wide baseline LDL-C categories that underestimate the CVD risk faced by patients with higher levels of LDL-C within a given category. This is important because baseline LDL-C level is a key driver of CVD risk and ultimately of cost effectiveness estimates.
ICER’s analysis groups patients into one of three LDL-C categories: LDL-C <70 mg/dL, LDL-C between 70 and 99.9 mg/dL, and LDL-C ≥100 mg/dL. The third category is particularly broad toward the lower end of distribution. Since LDL-C as a risk factor enters the model categorically rather than continuously, the risk for CVD at baseline will be constant within each category. This is a restrictive assumption since CVD risk is increasing in LDL-C levels, and the assumption is particularly worrisome for the third category given the high variance of risk for the population within the category. Recommendation: Add discrete categories above 100 mg/dL.
b. Using LDL-C values alone as the proxy definition of FH significantly underestimates, the lifetime CVD risk profile of the FH population.
The ability to identify FH patients in databases is difficult and the approach taken is reasonable; however, it is unclear how the probability of the incident and subsequent CVD events are determined. FH patients tend to have earlier and more frequent CVD events and higher CVD mortality rates compared to non-FH populations with similar LDL-C levels [10, 11] due to lifetime exposure to high LDL-C [12]. Recommendation: Framingham risk estimates are not appropriate to estimate CV risk in FH patients. Further analyses should be undertaken using published standardized mortality ratios between FH and non-FH populations [11] to allow for the true mortality effects in the FH group.
c. A significant underestimation of deaths and CVD events averted by PCSK9 inhibitor use in the risk groups studied may be due to the inclusion of a younger lower risk population and truncation of the older age groups who are at highest risk of having a first or recurring CVD event.
The age distribution in the CVD Policy Model skews young, which ignores potentially large treatment benefits to the population over age 75. The baseline model cohort includes only individuals aged 35-74. The age distribution is particularly problematic since there are only half the number of 75-84 year olds that would be in a cohort that began with a representative population of 35-84. This is why a lifetime model would be preferable to truly represent the benefits of interventions such as PCSK9 inhibitors. The 2010 National Vital Statistics Report on mortality shows that 62% of CVD deaths occur in those aged 75+ [13, 14], suggesting that the absence of individuals in older age groups results in underestimation of potential benefits from PCSK9 inhibitor treatment compared to a natural cohort, or a life-course model. [15-17].
Recommendation: A lifetime model should be applied to a cohort representative of the U.S. age distribution to reduce risk underestimation. If the model architecture requires defined age cut-points (35-74 years), ensure the age distribution does not change over the 20 year period and that the upper age cut-off is at least 84 years.

2. Inadequate Visibility to the Model Architecture.  Transparency around the Markov model architecture and assumptions is critical to interpreting results in context. In this report, readers are left to infer model details related to the estimation of CVD risk in the secondary prevention population, one of the important target populations for PCSK9 inhibitors. The CVD Policy Model presents parameters for individuals with prior CVD events, but ICER’s comparative value appendix discusses calculation of heart disease and stroke based on the Framingham Heart Study. It is unclear whether values from Framingham are applied to the entire model population, but the limitations of Framingham data in this context have been highlighted by ACC/AHA.[1] The risk profile for the ASCVD population should utilize event probabilities associated with an ASCVD population.

Despite the reference to the original CVD Policy Model publication, [2] the ICER report still has considerable information gaps. The report lacks detailed summary statistics and sample sizes for age, sex, and the 8 CVD risk factors, which makes the baseline population characteristics indeterminable. Moreover, without a baseline reference point, especially with respect to key parameters such as baseline LDL-C distribution, it is difficult for readers to put the results into context.  In addition, missing relevant life tables make it difficult to understand the source of life years gained, which is critical because the results in appendix 7, tables 3 and 4 imply that an average of 6 life years are gained from a death averted. This is half the size of similar estimates in other published literature – for example, the recent U.S. burden of disease report from the Institute of Health Metrics and Evaluation estimates values closer to 12 life years [3].

A wider understanding of these assumptions is essential to testing the validity of results. For example, we compared the report’s calculated incremental cost-effectiveness ratios for ezetimibe against a wider literature. While the ICER report estimates a cost per QALY value of $373,000 for ezetimibe plus statin therapy for secondary prevention patients, multiple published peer-reviewed international cost-effectiveness studies cite cost per QALY values ranging between $20,000 and $60,000 in 2015 USD [4-9]. This suggests that the ICER model is estimating considerably higher costs per QALY gained compared to other studies. Recommendation: Provide further information on the CVD Policy Model, include life tables in the appendixes, and provide clarification on rationale for QALY assumptions.

3. Imposition of a Narrowly Focused Health System Value Analysis. The ‘new drug’ thresholds suggested by ICER are based on unreliable assumptions and unrealistic estimates of uptake that significantly overestimate the budget impact, creating a chilling effect on innovation and disincentives for healthcare investment.  Results from this section should be de-emphasized since they do not consider the benefits to patients.

The health system value analysis, which drives ICER’s budgetary recommendations, focuses on drug cost, not value, and the “GDP+1” approach uses arbitrary thresholds that are subject to measurement error. Caps on pharmaceutical spending separate the assessment of total cost from the more relevant assessment of net benefit to patients, which can lead to erroneous decision-making, because low cost is often a poor measure of value. The approach outlined by ICER sets a budget threshold of $904 million in total annual costs for a new drug. Applying this threshold to past innovations, such as statins and anti-retrovirals, would have limited access to these drugs at the time they were introduced to the market [18, 19].  The underlying logic of a cap is flawed if only applied to one health care component.

Further, the uptake levels used for the analysis were 10%, 25%, 50%, and 75% after 5 years. In contrast, a recent study examining retrospective uptake of statins [18] showed that after 20 years, the uptake in the U.S. amongst those in the statin benefit groups had reached just 16%.

ICER’s analysis ignores important differences between the wholesale acquisition price (WAC) and the effective price in U.S. settings.  WAC price does not accurately reflect discounts and may bias estimates. Recommendation: Allow for a GDP growth range, number of new drug approvals, and more realistic uptake rates for an injectable medication. Sensitivity analyses should be conducted to incorporate a range of estimates on discounts offered to plans and co-pay assistance to patients in evaluating value based pricing.
 
References

1. Grundy, S.M., et al., Assessment of cardiovascular risk by use of multiple-risk-factor assessment equations: a statement for healthcare professionals from the American Heart Association and the American College of Cardiology. Journal of the American College of Cardiology, 1999. 34(4): p. 1348-1359.
2. Weinstein, M.C., et al., Forecasting coronary heart disease incidence, mortality, and cost: the Coronary Heart Disease Policy Model. American journal of public health, 1987. 77(11): p. 1417-1426.
3. Murray, C.J., et al., The state of US health, 1990-2010: burden of diseases, injuries, and risk factors. Jama, 2013. 310(6): p. 591-606.
4. Ara, M.R., et al., Cost Effectiveness of Ezetimibe in Patients with Cardiovascular Disease and Statin Intolerance or Contraindications. American journal of cardiovascular drugs, 2008. 8(6): p. 419-427.
5. Charles, Z., E. Pugh, and D. Barnett, Ezetimibe for the treatment of primary (heterozygous-familial and non-familial) hypercholesterolaemia: NICE technology appraisal guidance. Heart, 2008. 94(5): p. 642-643.
6. Kohli, M., et al., Cost effectiveness of adding ezetimibe to atorvastatin therapy in patients not at cholesterol treatment goal in Canada. Pharmacoeconomics, 2006. 24(8): p. 815-830.
7. Cook, J.R., et al., Cost-effectiveness of ezetimibe coadministration in statin-treated patients not at cholesterol goal. Pharmacoeconomics, 2004. 22(3): p. 49-61.
8. van Nooten, F., et al., Economic evaluation of ezetimibe combined with simvastatin for the treatment of primary hypercholesterolaemia. Netherlands Heart Journal, 2011. 19(2): p. 61-67.
9. Reckless, J., et al., Projected Cost‐Effectiveness of Ezetimibe/Simvastatin Compared with Doubling the Statin Dose in the United Kingdom: Findings from the INFORCE Study. Value in Health, 2010. 13(6): p. 726-734.
10. Knowles, J.W., et al., Reducing the burden of disease and death from familial hypercholesterolemia: A call to action. American heart journal, 2014. 168(6): p. 807-811.
11. Neil, A., et al., Reductions in all-cause, cancer, and coronary mortality in statin-treated patients with heterozygous familial hypercholesterolaemia: a prospective registry study. European heart journal, 2008. 29(21): p. 2625-2633.
12. Besseling, J., et al., Severe heterozygous familial hypercholesterolemia and risk for cardiovascular disease: a study of a cohort of 14,000 mutation carriers. Atherosclerosis, 2014. 233(1): p. 219-223.
13. Mozaffarian, D., et al., Heart disease and stroke statistics-2015 update: a report from the american heart association. Circulation, 2015. 131(4): p. e29.
14. Murphy, S.L., J. Xu, and K.D. Kochanek, National vital statistics reports. National vital statistics reports, 2013. 61(4).
15. Fleg, J.L., et al., Secondary Prevention of Atherosclerotic Cardiovascular Disease in Older Adults A Scientific Statement From the American Heart Association. Circulation, 2013. 128(22): p. 2422-2446.
16. Saunderson, C.E., et al., Acute coronary syndrome management in older adults: guidelines, temporal changes and challenges. Age and ageing, 2014. 43(4): p. 450-455.
17. Williams, M.A., et al., Secondary Prevention of Coronary Heart Disease in the Elderly (With Emphasis on Patients≥ 75 Years of Age) An American Heart Association Scientific Statement From the Council on Clinical Cardiology Subcommittee on Exercise, Cardiac Rehabilitation, and Prevention. Circulation, 2002. 105(14): p. 1735-1743.
18. Grabowski, D.C., et al., The large social value resulting from use of statins warrants steps to improve adherence and broaden treatment. Health Affairs, 2012. 31(10): p. 2276-2285.
19. Philipson, T.J. and A.B. Jena. Who benefits from new medical technologies? Estimates of consumer and producer surpluses for HIV/AIDS drugs. in Forum for Health Economics & Policy. 2006.