Cost-effectiveness of Maintenance Capecitabine and Bevacizumab for Metastatic Colorectal Cancer

Abstract

Importance: Unregulated drug prices increase cancer therapy costs. After induction chemotherapy, patients with metastatic colon cancer can receive maintenance capecitabine and bevacizumab therapy based on improved progression-free survival, but whether this treatment's cost justifies its benefits has not been evaluated in the United States.

Objective: This study sought to determine the influence of capecitabine and bevacizumab drug prices on cost-effectiveness from a Medicare payer's perspective.

Design, setting, and participants: The incremental cost-effectiveness of capecitabine and bevacizumab maintenance therapy was determined with a Markov model using a quality-of-life penalty based on outcomes data from the CAIRO phase 3 randomized clinical trial (RCT), which included 558 adults in the Netherlands with unresectable metastatic colorectal cancer who had stable disease or better following induction chemotherapy. The outcomes were modeled using Markov chains to account for patients who had treatment complications or cancer progression. Transition probabilities between patient states were determined, and each state's costs were determined using US Medicare data on payments for capecitabine and bevacizumab treatment. Deterministic and probabilistic sensitivity analyses identified factors affecting cost-effectiveness.

Main outcomes and measures: Life-years gained were adjusted using CAIRO3 RCT quality-of-life data to determine quality-adjusted life-years (QALYs). The primary end point was the incremental cost-effectiveness ratio, representing incremental costs per QALY gained using a capecitabine and bevacizumab maintenance regimen compared with observation alone.

Results: Markov model estimated survival and complication outcomes closely matched those reported in the CAIRO3 RCT, which included 558 adults (n = 197 women, n = 361 men; median age, 64 and 63 years for patients in the observation and maintenance therapy groups, respectively) in the Netherlands with unresectable metastatic colorectal cancer who had stable disease or better following induction chemotherapy. Incremental costs for a 3-week maintenance chemotherapy cycle were $6601 per patient. After 29 model iterations corresponding to 60 months of follow-up, mean per-patient costs were $105 239 for maintenance therapy and $21.10 for observation. Mean QALYs accrued were 1.34 for maintenance therapy and 1.20 for observation. The incremental cost-effectiveness ratio favored maintenance treatment, at an incremental cost of $725 601 per QALY. The unadjusted ratio was $438 394 per life-year. Sensitivity analyses revealed that cost-effectiveness varied with changes in drug costs. To achieve an incremental cost-effectiveness ratio of less than $59 039 (median US household income) per unadjusted life-year would require capecitabine and bevacizumab drug costs to be reduced from $6173 (current cost) to $452 per 3-week chemotherapy cycle.

Conclusions and relevance: Antineoplastic therapy is expensive for payers and society. The price of capecitabine and bevacizumab maintenance therapy would need to be reduced by 93% to make it cost-effective, a finding useful for policy decision making and payment negotiations.

Figure 1.. Patient Counts and Incremental Cost-effectiveness Ratios (ICERs)

A, The model begins with 1000 patients in the alive-with-disease state. Their progression through the other model states by model iteration is indicated for the maintenance and observation treatment groups. The CAIRO3 randomized clinical trial reported times to first progression event of 8.5 and 4.1 months in the maintenance and observation groups, respectively, which correspond to 4.11 and 1.98 Markov iterations, while overall survival times of 21.6 and 18.1 months correspond to 10.44 and 8.74 iterations. B, ICERs per quality-adjusted life-year (solid line) and per unadjusted life-year (dashed line) are shown. The vertical line indicates the current price ($6173) paid by Medicare. For a willingness to pay of $200 000 per unadjusted life-year, the corresponding total drug cost would be approximately $2500 per 3-week cycle.

Figure 2.. Deterministic Sensitivity Analysis Tornado Plots Showing the Range of Model Outputs

A and B, Plots show mean per-patient cost ranges for the observation and maintenance therapy arms when each variable is tested at the high and low values listed in the eTable in the Supplement. Variables to which model costs were not sensitive are not shown. C, Plot shows the range in incremental cost-effectiveness ratio (ICER). The vertical lines in each plot indicate baseline values found in the model ($21.10 per patient for observation costs, $105 239 per patient for maintenance costs, and $725 601 per quality-adjusted life-year gained for ICER). Observation costs were most influenced by complications, while the transition probability for progression and costs of bevacizumab and capecitabine most influenced maintenance costs. Transition probabilities and quality-of-life (QoL) values create the greatest ICER differences. Changes to drug costs represent the most influential modifiable variables. Observation and maintenance QoL values are calculated by adding quality-of-life difference to baseline QoL, which is midway between the baseline values for patients in the observation and maintenance therapy groups. On progression, progression QoL is subtracted from baseline QoL. This allowed varying QoL assumptions for both treatment arms simultaneously.