Various forms of decision health analytic modeling have been used for the last 3 decades to assess potential health economic impact of new therapies. As initial gene and cellular therapies come to market, debate is growing as to the best ways to evaluate the associated economic impact, specifically in the context of the shift to value-based reimbursements. We share some of our observations and early learnings in modeling the economic impact.
Science fiction author Arthur C Clarke once said, “New ideas pass through three periods: It can’t be done. It probably can be done, but it’s not worth doing. I knew it was a good idea all along!” Over the last 3 to 4 years cellular and gene therapies have reached the third phase of Mr Clarke’s innovation paradigm with a number high-profile therapies being approved by the Food and Drug Administration (FDA), including:
- Voretigene neparvovec-rzyl indicated for inherited retinal disease due to mutations in both copies of the RPE65 gene (gene therapy)1
- Tisagenlecleucel indicated for the treatment of B-cell precursor acute lymphoblastic leukemia (chimeric antigen receptor T-cell [CAR-T] therapy)2
- Axicabtagene ciloleucel indicated for the treatment of adult patients with relapsed or refractory large B-cell lymphoma after 2 or more lines of systemic therapy (CAR-T therapy)3
Now that these “good ideas” and clinical advances are coming to fruition, there is growing debate regarding how best to evaluate the economic impact associated with gene and cellular therapies, especially in health systems evolving to value-based payments.
For over 30 years, different forms of decision health analytic modeling have been used to assess health economic impact of new therapies, undergoing a number of refinements along the way.4-6 For the most part, methodologies have been dictated by 2 core elements: (1) cost of treatment in the form of drug cost and cost offset, including pharmacy and medical costs; and (2) clinical (life years gained) and quality-of-life (quality-adjusted life years [QALY]) benefits. Taken together, these considerations result in an incremental cost-effectiveness ratio (ICER); the output of an ICER is cost/life year gained or cost/QALY gained. Through the years, the use of the ICER has become the mainstay of health technology appraisal organizations such as the UK’s National Institute for Clinical Excellence (NICE), the United States’ Institute for Clinical and Economic Review, and others.
With experience in developing health economic models for several of the initial gene and cellular therapies that have come to market or are projected to do so within the next few years, we share some of our observations and early learnings in modeling the economic impact.
Differences in Development and Commercialization of Gene and Cellular Therapies
Small Patient Populations
Most gene and cellular therapies are being developed for diseases with relatively small patient populations, frequently for refractory disease. A small patient population has multiple repercussions. First, it can lead to clinical trials with small sample sizes that generate nonrobust clinical data and, in some circumstances, may result in single-arm clinical trials; such findings impact how modeling is approached (eg, requiring indirect comparisons to be made). Second, the small overall treatment population could influence pricing decisions, which also impacts cost effectiveness. Last, treatment for these rare conditions often is focused in centers of excellence, which can impact which patients actually receive treatment as well as potential time and cost factors associated with traveling to a center of excellence.
Intensive Manufacturing Process
It used to be said that it took $800 million dollars to produce the first pill, and subsequent pills cost 1 cent. This is not the case with gene and cellular therapies. In fact, CAR-T therapies are manufactured specifically for each individual patient, since the individual’s own white blood cells serve as the “raw material” for developing the final product. Thus, there are both time and cost elements for these types of therapies that need to be fully accounted for in economic analyses.
Value May Vary by Perspective
The value of a potential cure is extremely high for patients as well as for the clinicians treating them. However, a payer will always be concerned about assuming financial risk without confidence in duration of effect or that the individual may switch to a different benefit provider. For providers, therapy requiring an inpatient hospitalization also elevates their financial risk, since the cost will surely exceed typical reimbursement.
Issues With Assessing Gene and Cellular Therapies Using Health Economic Modeling
Duration of Effect
Most gene and cellular therapies are one-time interventions with a biologically credible potential to produce a long-term remission or “cure.” But long-term data following patients for 30 or 40 years do not exist. As the joint ICER-OHE (Office of Health Economics) points out in a recent white paper on gene therapies, uncertainties regarding the duration of effect only complicate questions about how “society” values a cure relative to typical incremental gains observed with other therapies. From a practical standpoint, lack of long-term data also necessitates the employment of projection functions that are notorious for under or overestimating length of a “cure.”7
Assessing Utility/Quality of Life
Utility measures summarize both positive and negative effects of an intervention into one value between 0 (equal to death) and 1 (equal to perfect health). Utilities are usually assessed using a patient report outcome instrument (eg, EQ5D8 or SF-369). In many gene therapies, the earlier it is administered, the quicker the progression of the disease can be halted. For example, the earlier voretigene neparvovec-rzyl is administered in the progression of blindness, the greater the effect. The result is that the therapy is often administered in children as early as 4 to 6 years of age and utility measures with children of that age are challenging to gauge. This imprecision could greatly affect the cost per QALY. Additionally, impacts on nonclinical outcomes, such as productivity, educational attainment, and caregiver burden, are difficult to assess.
Factoring in Disease-Related Costs
Most models do not factor in the cost of the underlying disease and the cost associated with the added years of life. There is credible belief that many of the gene and cellular therapies will greatly extend life and thereby impact the utilization of unrelated medical costs—and that the added years of life and conventional cost-effectiveness methods may implicitly penalize therapies that add “expensive” life years.10
Recommendations and Considerations
Do we need to abandon the use of health economics models for some other way to assess value? Based on our experiences, we believe the answer to this question is no. But we do believe that the modeling techniques might need refinements, and the lenses through which we view the results should be different. Below we posit 3 considerations:
1. Based on some of the imperfections inherent in modeling “cures,” we suggest that cost-effectiveness ratios be developed and a more flexible idea of “threshold” adopted. For example, NICE and the British National Health Services usually use between 20,000 to 30,000 £/QALY, and ICER has a stated threshold of $100,000 to $150,000/QALY as therapy that is cost-effective.
2. Expand scope of purview to capture unrelated medical costs in cost-effectiveness analyses. Unrelated costs would include the cost for treatment for other medical conditions; for example, cost for treatment for a myocardial infarction when modeling a cancer therapy that may substantially extend life.
3. Employ multiple types of modeling techniques, not just cost-effectiveness. For example, cost-consequence models could be developed that report the clinical benefit separately from costs. These types of models can also demonstrate where averted cost exists as a result of the therapy in a transparent manner.
As these scientific breakthroughs bring newfound hope to many with previously untreatable conditions, and may even potentially lead to cures, we must also achieve breakthroughs in economic modeling to best measure the economic value of these treatments. We want to say with conviction, “We knew it was a good idea all along!” Yet, without these breakthroughs, the clinical promise and availability of these therapies could be thwarted by a health system striving for greater value.
1. Luxturna (voretigene neparvovec-rzyl) intraocular suspension for subretinal injection [package insert]. Philadelphia, PA: Spark Therapeutics Inc; 2017.
2. Kymriah (tisagenlecleucel) suspension for intravenous infusion [package insert]. East Hanover, NJ: Novartis Pharmaceuticals Corp; 2017.
3. Yescarta (axicabtagene ciloleucel) suspension for intravenous infusion [package insert]. Santa Monica, CA: Kite Pharma Inc; 2017.
4. Torrance GW, Thomas WH, Sackett DL. A utility maximization model for evaluation of health care programs. Health Serv Res. 1972;7(2):118-133.
5. Zeckhauser R, Shepard D. Where now for saving lives?” Law Contemp Problems. 1976;40(4):5-45.
6. Pliskin JS, Shepard DS, Weinstein MC. Utility functions for life years and health status. Operations Res. 1980;28:206-224.
7. Marsden G, Towse A, Pearson SD, Dreitlein B, Henshall C. Gene therapy: understanding the science, assessing the evidence, and paying for value. Office of Health Economics website. https://www.ohe.org/publications/gene-therapy-understanding-science-assessing-evidence-and-paying-value. Published March 2017. Accessed March 20, 2018.
8. Rabin R, de Charro F. EQ-5D: a measure of health statues from the EuroQol Group. Ann Med. 2001;33(5):337-343.
9. Ware JE Jr, Sherbourne CD. The MOS 36-item short-form health survey (SF-36). I. conceptual framework and item selection. Med Care. 1992;30(6):473-483.
10. Olchanski N, Zhong Y, Cohen JT, Saret C, Bala M, Neumann PJ. The peculiar economics of life-extending therapies: a review of costing methods in health economic evaluations in oncology. Expert Rev Pharmacoecon Outcomes Res. 2015;15(6):931-934.