- Email: [email protected]
Cost-effectiveness of adding bevacizumab to first line therapy for patients with advanced ovarian cancer
YGYNO-975349; No. of pages: 7; 4C: Gynecologic Oncology xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Gynecologic Oncology journal homepage: www.elsevier.com/locate/ygyno
Cost-effectiveness of adding bevacizumab to first line therapy for patients with advanced ovarian cancer Darshan A. Mehta a, Joel W. Hay b,⁎ a b
Leonard D. Schaeffer Center for Health Policy and Economics, School of Pharmacy, University of Southern California, USA Clinical Pharmacy and Pharmaceutical Economics & Policy, Leonard D. Schaeffer Center for Health Policy and Economics, School of Pharmacy, University of Southern California, USA
H I G H L I G H T S • Three-state Markov model for GOG-218 and ICON-7 protocol showed adding bevacizumab is cost-ineffective, at willingness-to-pay threshold of $150,000/QALY. • Strategy is cost-effective in high-risk population using biosimilar bevacizumab as per ICON-7 protocol. • PSA results show negligible probability for cost-effective ICER for the GOG-218 trial, and approximately 50% probability for ICON-7 trial.
a r t i c l e
i n f o
Article history: Received 29 November 2013 Accepted 14 January 2014 Available online xxxx Keywords: Ovarian cancer Bevacizumab Cost-effectiveness Subgroup Platinum-resistant Platinum sensitive
a b s t r a c t Objective. To evaluate, from a societal perspective, the cost-effectiveness of adding bevacizumab to first-line therapy based on outcomes from the GOG-218 and ICON-7 trials. Methods. A three-state Markov model was used. The time horizon was until the death of 99% of the initial cohort of 1000 individuals. Costs and quality-adjusted life-years (QALYs) were discounted at an annual rate of 3%. All costs were adjusted to 2013 USD. The incremental cost-effectiveness ratio (ICER) was reported as incremental cost per QALY gained. The robustness of the result was checked with one-way sensitivity analyses and for relevant clinical situations (i.e. varying the drug of choice to treat cancer recurrence). Subgroup analysis was conducted to identify subgroup of population for whom the strategy could be cost-effective. The potential impact of biosimilar bevacizumab was considered, using a 30% price reduction. Results. For the GOG-218 study protocol, widely followed in US, the addition of bevacizumab results in an ICER of $2,420,691/QALY. For the ICON-7 study protocol, the ICER is $225,515/QALY. The results of the model were sensitive to the quality of life (QoL) and the median progression free survival (PFS). Biosimilar bevacizumab didn't reduce cost sufficiently to change conclusions. First-line augmentation is cost-effective, with biosimilar bevacizumab, for stage IV patients ($126,169/QALY), ECOG PS1 patients ($116,575/QALY) and for patients with suboptimal residual disease ($122,822/QALY) as per the ICON-7 protocol. Conclusion. Addition of bevacizumab, by in large, is cost-ineffective. It can become cost-effective with the ICON-7 protocol, in patients at high risk of progression using biosimilar bevacizumab. © 2014 Elsevier Inc. All rights reserved.
Introduction Epithelial ovarian cancer, a common form of gynecological malignancy, is the fifth most frequent cause of cancer death in women. Since the underlying pathophysiology leading to ovarian cancer is unknown, it isn't easily diagnosed. Patients usually present at stage III and higher when the tumor has already metastasized. At this stage the 5-year relative survival rate is 26.9% [1]. ⁎ Corresponding author at: USC Schaeffer Center for Health Policy and Economics, University Park Campus, UGW-Unit A, Los Angeles, CA 90089-7273, USA. Fax: +1 213 740 3460 (Attn: Jacky). E-mail address: [email protected] (J.W. Hay). URL: http://healthpolicy.usc.edu/ListItem.aspx?ID=4, http://pharmacyschool.usc. edu/faculty/profile/?id=374 (J.W. Hay).
The National Comprehensive Cancer Network (NCCN) recommends debulking surgery on presentation followed by adjuvant chemotherapy [2]. Clinicians consider the extent of primary debulking surgery as an important prognostic factor for the survival of patient. Bristow et al. report that for each 10% increase in maximal cytoreduction there is a 5.5% increase in median survival time [3]. The objective of adjuvant chemotherapy is to lower the risk of cancer recurrence and hence increase the overall survival (OS). NCCN recommends multiple treatment regimens post-primary cytoreductive surgery [2]. Clinical trials have compared the safety and efficacy of the different drug combinations and routes of administration for adjuvant first-line therapy [4,5]. The current gold standard is intravenous (IV) paclitaxel and carboplatin after primary debulking surgery. In patients with microscopic residual tumors (b1 cm in size) increased OS
0090-8258/$ – see front matter © 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ygyno.2014.01.021
Please cite this article as: Mehta DA, Hay JW, Cost-effectiveness of adding bevacizumab to first line therapy for patients with advanced ovarian cancer, Gynecol Oncol (2014), http://dx.doi.org/10.1016/j.ygyno.2014.01.021
2
D.A. Mehta, J.W. Hay / Gynecologic Oncology xxx (2014) xxx–xxx
is achieved by administering the drugs intraperitoneally (IP) [4]. Based on pathophysiological studies clinical trials have evaluated the effect of adding a VEGF inhibitor like bevacizumab to the current first-line therapy. Notable among these trials are, GOG-218, ICON-7 and the current ongoing GOG-262 [6–8]. The GOG-218 trial reports a median progression free survival (PFS) of 10.3 months with current first line therapy and 14.1 months for the bevacizumab throughout the group. The trial does not report significant difference in OS among the comparative arms [6]. The ICON-7 trial reports a median PFS of 22.4 months for the standard therapy and 24.1 months with bevacizumab. In patients at high risk of progression, the trial reports a median PFS of the standard therapy of 14.5 months and 18.1 months with bevacizumab added; with respective median OS of 28.8 and 36.6 months [7]. Based on the results, the latest guideline of NCCN recommends adding bevacizumab to the current gold standard therapy among other treatment options [2]. Around 80% of patients diagnosed with ovarian cancer relapse after first-line chemotherapy [9]. The relapse is either platinum resistant or platinum sensitive based on time of recurrence post-cessation of firstline therapy. Relapse within 6 months of cessation is defined as platinum resistant and post 12 months as platinum sensitive [9].
Model structure To evaluate the cost-effectiveness of adding bevacizumab to the first-line therapy, a three state Markov model was used (see supplement S1). The states were; stable state, progression state and death. The clinical information for the model was based on the results of the GOG-218 and the ICON-7 trial [6,7]. The population cohort matches the inclusion criteria of the GOG-218 trial considered for the model [6]. This included patients with median age of 60 and ECOG status of 0 to 2, previously untreated, FIGO stage 3/ stage 4 epithelial ovarian cancer, having undergone a primary debulking surgery, with either microscopic or macroscopic residual lesions. A separate analysis was conducted on a high-risk population cohort modeled on the ICON-7 trial [7]. For this, the patient cohort included those with high risk of progression, characterized by N1.0 cm residual disease after debulking surgery. Other criteria remain the same as the cohort for the GOG-218 trial. The model is developed from the US societal perspective. The cycle length is 3 weeks, with an initial cohort of 1000 patients having characteristics as defined for the two trials respectively. The time horizon considered for the model is until death of 99% of the cohort. Costs and health benefits are discounted at a societal discount rate of 3% [10].
Patients enter the model in the stable disease state. The disease progression state captures patients who have failed to respond to the first-line therapy and have cancer recurrence. Both these disease states also reflect the QoL and cost associated with medical treatment, including drug therapy and associated side-effects. The death state is the absorbing state in the model.
Base case For the base case, the treatment pattern was modeled from the clinical trials [6,7]. In the stable disease state, patients in the carboplatin and paclitaxel (CP) arm received six cycles of CP. For the carboplatin, paclitaxel and bevacizumab (CPB) arm, patients received 6 cycles of CPB followed by 18 additional cycles of bevacizumab in the GOG-218 trial and 12 additional cycles in the ICON-7 trial. In line with the clinical trials, bevacizumab costs were incorporated in the model beginning from cycle 2, as the drug is known to delay surgical healing [6,7]. After completing chemotherapy, patients with stable disease were assumed to have disease assessment every 3 months for 2 years and then every 6 months until the end of the model. Based on expert inputs, platinum resistant progression was assumed to be treated with pegylated doxorubicin, 50 mg/m2 administered every 4 weeks. This clinical assumption was later tested during sensitivity analysis with 40 mg/m2 pegylated doxorubicin and with different lines of therapy. During chemotherapy cycles, for the stable state, it is assumed that disease assessment occurs every 6 cycles, with a single office visit with the oncologists per 3-week cycle. For the progression phase disease assessment is assumed at every 3 cycles and 2 office visits per cycle. Our base case societal willingness-to-pay ICER threshold was assumed to be $150,000/QALY; roughly three time the US GDP per capita [11].
Transition probabilities For calculating the transition probabilities, the median PFS and OS, were obtained from the clinical trials [6,7]. Under the declining exponential approximation of life expectancy (DEALE) method assumption of constant proportional hazard, median PFS and OS were converted into rates of progression and death and subsequently into respective transition probabilities for the three week cycles [12]. For the base case analysis, only mortality risk due to ovarian cancer was considered (see Table 1).
Table 1 Base case parameters. Base case transition probabilities Allowable transitions
Stable to progressive Stable to death Stable to stable Progressive to death Progressive to progressive
GOG 218 trial
ICON 7 trial
CPB arm
CP arm
CPB arm
CP arm
0.0334 0.0120 0.9521 0.0525 0.9449
0.0455 0.0121 0.9399 0.0525 0.9449
0.0295 0.130 0.9550 0.0525 0.9449
0.0446 0.0165 0.9364 0.0525 0.9449
Base case utility weights States
Stable state chemotherapy cycles 1–6 Stable state: 12 months post-chemotherapy Stable state: post 12 months Progression phase — throughout
GOG 218 trial
ICON 7 trial
CPB arm
CP arm
CPB arm
CP arm
0.596,16,17,18 – 0.8516 0.6316,18,19
0.6710,16,17,18 0.7817,20 0.8516 0.6316,18,19
0.797,16,17,18 – 0.8516 0.6316,18,19
0.8112,16,17,18 0.8317,19 0.8516 0.6316,18,19
Please cite this article as: Mehta DA, Hay JW, Cost-effectiveness of adding bevacizumab to first line therapy for patients with advanced ovarian cancer, Gynecol Oncol (2014), http://dx.doi.org/10.1016/j.ygyno.2014.01.021
D.A. Mehta, J.W. Hay / Gynecologic Oncology xxx (2014) xxx–xxx
The base model was subsequently augmented in a sensitivity analysis to include patients' additional mortality risks [13]. The underlying age-specific mortality rates from the National Vital Statistics (NVS) report by the Center for Disease Control (CDC) were used. The ovarian cancer specific mortality rate was calculated and this was added to the age-specific mortality rates. For the time period post the median follow up period in the respective clinical trials, the transition probability was calculated based on the combined rate of death due to ovarian cancer and the underlying natural rate of death . The hazard ratios of progression for the two trials were used to determine the stable to progression transition probability for different prognostic factors of progression [6,7]. The study incorporated clinically relevant scenarios beyond the clinical trial findings. Based on expert input, treatment with 40 mg/m2 doxorubicin was modeled for patients experiencing disease progression. Since this reduced dosage is equally efficacious with the base case dosage, the progression to death transition probability was modeled to be the same as the base case [14]. Due to lack of prospective clinical trial evidence, multiple lines of chemotherapy for treating cancer recurrence were modeled, based on Griffith et al. [15]. The lines of therapy were selected based on widespread use. The method similar to the base case was adopted for calculating transition probabilities. During the sensitivity analysis best case and worst case scenarios were modeled based on published ranges of model parameters (see supplement S2). Quality of life assessment Table 1 shows the utility weights used for the model. The literature derived utility weights for the stable and progression states are adjusted for side-effects due to chemotherapy administration for the two arms [16–19]. Also as observed in Table 1, for the stable state different utility weights are associated with different time points. For the CP arm, the utility weight was adjusted for a period of 12 months for residual neurotoxicity due to chemotherapy cycles [20]. This was not replicated for the CPB arm. Occurrence of residual neurotoxicity is not documented with bevacizumab and hence residual neurotoxicity due to CP administration was assumed to wane off during the maintenance bevacizumab cycles. The utility weights used in the model were based on preferences elicited from the general population using the Time Trade-Off (TTO) method [21]. In line with the results of the QoL assessment done alongside the clinical trials, the utility weights associated with the CPB arm are lower than for the CP arm and gradually increase for both the arms [22,23]. During the sensitivity analysis, when the preferred drug for treating cancer recurrence is varied, the utility weight is adjusted based on incidence of side-effects from their respective clinical trials [24–27].
3
Costs A mixture of micro-costing and gross costing methodologies was used to determine the cost associated with each Markov cycle [28]. Direct costs included drug acquisition cost, laboratory costs for patient evaluation prior to each chemotherapy cycle, imaging costs for evaluating disease progression, cost of treating side-effects and office visits. The drug costs were derived from the US Veteran Affairs Federal Supply Schedule (FSS), as a closest approximation for fully discounted drug prices. The required dosage for bevacizumab was calculated based on mean patient body weight for 60 year old women [29]. The paclitaxel required dosage was based on body surface area calculated following the Mosteller formula [30]. For the carboplatin dosing, the normal renal function was assumed with glomerular filtration rate of 125 ml/min. The Calvert formula was made use of to calculate the dose required to attain concentration equivalent to area under curve (AUC) 6 [31]. This dosage was used to calculate the number of units required meeting the dose. In instances where multiple companies supplied a drug, the one with the lowest FSS price was utilized. For all the three drugs, based on expert opinion, no vial sharing was assumed. For the stable disease state, the model included only the cost of treating hypertension for the GOG-218 study as it was the only sideeffect with statistically significant difference between the two arms [6]. The cost of treating side-effects derived from the literature was weighted based on differential incidence of hypertension observed in the clinical trial [32,6]. The cost of treating side-effects was accounted for during the first cycle and no side effect cost was incorporated for the remaining cycles. For the progression state, the costs of treating side-effects were weighted by incidence rates that were adjusted to reflect incidence every 4 weeks [32,19]. Burudapakdee et al. reported hospital charges from the Healthcare Cost and Utilization Project (HCUP) and hence a nationwide Medicare inpatient mean cost-tocharge ratio of 0.33 (FY 2013) (0.165–0.709) was used to define costs for both stable and progression states. The lab technician time associated with disease assessment was derived from literature [33]. The nursing time associated with the chemotherapy administration was based on the time needed for infusion of drugs as per the drug labeling information. The pharmacist time for preparation of drug was the same for all the drugs as the procedure involved reconstitution of drugs as IV infusion by diluting the content of the single use vial in a 100 ml of 0.9% sodium chloride injection, USP. These were valued at the mean hourly compensation rates using the Bureau of Labor Statistics — National Compensation Survey. Indirect costs in the model included unpaid patient and caregiver time and costs. The patient time was derived from the literature [34]. This was valued at the mean hourly compensation rate of women with mean age of 60 years. The time associated with informal caregiving was assumed to be at 6.8 h per day for the stable state and 10 h
Table 2 Results. Results — base case and biosimilar bevacizumab GOG 218 trial CPB arm Life expectancy (weeks) Life expectancy (years) Incremental life expectancy (years) Total cost — base case (USD) Total cost — biosimilar bevacizumab (USD) Incremental cost (USD) Incremental cost — biosimilar (USD) QALY Incremental QALY ICER — base case ICER — biosimilar bevacizumab
108 1.97 201,423 162,500
1.306
ICON 7 trial CP arm 97.12 1.78 0.19 70,158 70,158 131,265 92,342 1.252 0.05 2,420,691/QALY 1,702,968/QALY
CPB arm 104.45 2.01 125,114 107,598
1.508
CP arm 86.84 1.67 0.34 63,311 63,311 61,803 44,286 1.234 0.27 225,515/QALY 161,603/QALY
USD: US dollars.
Please cite this article as: Mehta DA, Hay JW, Cost-effectiveness of adding bevacizumab to first line therapy for patients with advanced ovarian cancer, Gynecol Oncol (2014), http://dx.doi.org/10.1016/j.ygyno.2014.01.021
4
D.A. Mehta, J.W. Hay / Gynecologic Oncology xxx (2014) xxx–xxx
per day for the progression state [35]. This was valued at the mean hourly compensation of a home health aide. Results The base case results for both the clinical trials resulted in the CPB strategy being cost-ineffective at a cost-effectiveness willingnessto-pay threshold of $150,000/QALY. The base case ICER for GOG218 trial is $2,420,691/QALY and $225,515/QALY for ICON-7 trial. Considering the different patient populations for both the clinical trials, the strategy is cost-ineffective in the universal pool of patients of GOG-218 trial and in the high-risk subset of individuals of the ICON-7 trial (see Table 2). Sensitivity analysis One-way sensitivity analyses on the base case parameters, for both the trials, were done by increasing and decreasing values by 25%. In cases where a 25% increase would create nonsensical values (e.g., utility weights, transition probabilities), 1 was considered as the upper bound value for the parameter. For the cohort modeled on the GOG-218 trial, the CPB strategy was dominated by CP with 25% parameter decreases in the median PFS, OS, QoL for the CPB arm, or with increases in the median PFS, OS and QoL for the CP arm. The study didn't find the ICER sensitive to the cost of the drugs (see Fig. 1a). For the high-risk sub-group modeled on the ICON-7 trial the ICERs were sensitive to stable state utility weight for the CPB strategy and
stable to progressive state transition probabilities for both the arms. The CPB strategy became cost-effective, when the median PFS increased to 21.43 months, and the utility weight for the stable state increased to 0.9875 (see Fig. 1b). At the assumed willing-to-pay threshold, probabilistic sensitivity analysis (PSA) on the base case parameters resulted in a miniscule probability of ICER becoming cost-effective for the GOG-218 trial, and approximately 50% probability for ICON-7 trial (see Supplements S2, S3). A threshold sensitivity analysis was done on the cost of bevacizumab, median PFS and OS for both clinical trials. For the ICON-7 trial, at constant median PFS for CP, the CPB strategy becomes cost-effective when the median PFS exceeds 20.15 months. If the median PFS for CP is also varied, the CPB strategy becomes cost-effective at a combination of 20.59 months for the CPB arm and 10.74 months for the CP arm. The CPB strategy becomes cost-effective on reduction of per cycle cost to 38% of the current per cycle cost. Similarly, for GOG-218 study, under constant median PFS for the CP arm, the CPB strategy becomes cost-effective when median PFS exceeds 64.5 months. The GOG-218 protocol doesn't become cost-effective over the range of costs considered. However when the median PFS is increased to 49 months and the price of bevacizumab reduced to 58% of the current per unit cycle cost, the CPB strategy becomes cost-effective. Sensitivity analyses on the clinical assumptions resulted in the CPB strategy remaining cost-ineffective for 40 mg/m 2 dosage of doxorubicin, and for multiple lines of therapy for the progression phase (see supplement S1). Also, when platinum sensitive and platinum resistant recurrences were considered, the CPB strategy remained costineffective [36].
a) One-way sensitivity analysis for GOG-218 trial Change in ICER from baseline (2,490,691/QALY). -50
-40
-30
-20
-10
0
10
20
30
40
50
Follow up utility weight - CP Utility weight - Stable - CP Bevacizumab Cost Stable to Death (CP) Follow up utility weight - CPB Patient Body weight Progressive to Death (CP) Stable to Death (CPB)
CPB is dominated by CP
12 moth post first line - CP Stable to progressive (CP) Utility weight - Stable - CPB Utility weight - Progressive - CPB Progressive to Death (CPB) Utility weight - Progressive - CP Stable to progressive (CPB) Doxorubicin Cost Paclitaxel Cost Carboplatin Cost
X Axis represents change in ICER from baseline (2,490,691/QALY), values in USD 100,000s Fig. 1. a: One-way sensitivity analysis for GOG-218 trial. X axis represents change in ICER from baseline (2,490,691/QALY), values in USD 100,000s. 1b: One-way sensitivity analysis for the ICON-7trial. X axis represents change in ICER from the baseline (225,515/QALY), values in USD 1,000s.
Please cite this article as: Mehta DA, Hay JW, Cost-effectiveness of adding bevacizumab to first line therapy for patients with advanced ovarian cancer, Gynecol Oncol (2014), http://dx.doi.org/10.1016/j.ygyno.2014.01.021
D.A. Mehta, J.W. Hay / Gynecologic Oncology xxx (2014) xxx–xxx
5
b) One-way sensitivity analysis for the ICON-7trial Change in ICER from the baseline (225,515/ QALY) (150)
(100)
(50)
-
50
100
150
200
250
300
Utility weight - Stable - CPB Stable to progressive (CP) Stable to progressive (CPB) Utility weight - Progressive - CP Follow up utility weight - CPB Utility weight - Progressive - CPB Stable to Death (CPB) Progressive to Death (CP) 12 moth post first line - CP Stable to Death (CP) Patient Body weight Bevacizumab Cost Utility weight - Stable - CP Progressive to Death (CPB) Follow up utility weight - CP Doxorubicin Cost Paclitaxel Cost Carboplatin Cost
X axis represents change in ICER from the baseline (225,515/ QALY), values in USD 1,000s Fig. 1 (continued).
Impact of biosimilar bevacizumab With the product patent on bevacizumab expiring in July 2019 in the US and 2018 in the EU, currently 7 companies are developing biosimilar bevacizumab. Though the Biologics Price Competition and Innovation (BPCI) Act, 2009 defined an abbreviated biosimilar approval process, it resulted in many unanswered questions regarding the pathway to FDA approval. Considering the complex regulatory hurdles and uncertain data requirements for biosimilar approval, reduced competition due to high entry barrier; similarly high price discounts as for small molecule generics isn't expected for biosimilars in the US. As there is no precedence of a biosimilar price for the US, a 30% discount on the current FSS price of bevacizumab was assumed. This is in line with the pricing of the biosimilar Remicade by Celltrion and Hospira, launched in June 2013 in the EU (see Table 2). Subgroup analysis Using hazard ratios of progression, ICERs were calculated for prognostic subgroups [6]. The results for risk factors, as indicated in
Table 3 and supplement S4, show that CPB strategy following GOG218 treatment protocol is cost-ineffective across all subgroups. The hazard ratios of progression for ICON-7 were derived from the entire population set of the ICON-7 trial [6]. Since the median OS for the entire population hasn't been attained, these hazard ratios were adjusted, using the adjustment factor of 0.839, to reflect the hazard for a high risk population (see Table 3 and supplement S5) [7]. From the resulting ICERs we infer that the strategy is cost-ineffective for any of the subgroups as per the base case scenario. The strategy does become cost-effective with biosimilar bevacizumab, in patients at high risk of progression with FIGO stage IV disease, suboptimal residual disease or ECOG PS 1. Discussion From the societal perspective, the strategy of adding bevacizumab to the first-line therapy is cost-ineffective at a willingness-to-pay threshold of $150,000/QALY. However, probabilistic sensitivity analysis varying the base case parameters results in CPB therapy following the ICON-7 protocol having a 50% chance of being cost-effective at a $150,000/QALY threshold, while the GOG-218 protocol remains predominantly cost-
Please cite this article as: Mehta DA, Hay JW, Cost-effectiveness of adding bevacizumab to first line therapy for patients with advanced ovarian cancer, Gynecol Oncol (2014), http://dx.doi.org/10.1016/j.ygyno.2014.01.021
6
D.A. Mehta, J.W. Hay / Gynecologic Oncology xxx (2014) xxx–xxx
Table 3 Subgroup analysis. Risk factor
Hazard ratio
Hazard ratio adjusted for high risk population
Rate of progresion
Transition probability
ICER (cost/QALY)
ICER with biosimilar bevacizumab (cost/QALY)
ICON-7 trial ECOG PS 1 FIGO IV Suboptimal residual disease
0.66 0.69 0.68
0.554 0.579 0.571
0.0365 0.0382 0.0376
0.0250 0.0261 0.0257
165,429 178,231 173,844
116,575 126,169 122,822
GOG 218 trial ECOG PS 1 or 2 FIGO stage IV FIGO III and macroscopic b= 1 cm FIGO III and macroscopic N 1 cm
0.69 0.70 0.62 0.76
– – – –
0.0455 0.0461 0.0407 0.0503
0.0310 0.0313 0.0278 0.0342
1,408,864 1,504,219 874,929 3,133,968
988,171 1,055,531 610.896 2,206,730
The table depicts ICERs for risk factors which make strategy cost effective as per ICON 7 trial. ICERs for comparative risk factors as per GOG 218 trial protocol depicted. For ICERs for all risk factors see supplements S5 and S6.
ineffective at any willingness-to-pay threshold. Prior cost effectiveness studies have defined cost-effectiveness threshold varying from $100,000/QALY to $200,000/QALY [17,37,38]. For clinical oncologist, minor effectiveness of intervention is considered of good value. Hence in clinical practice, oncologists demonstrate a willingness-to-pay threshold of $300,000/QALY [39]. Oncologists demonstrate a higher willingnessto-pay threshold if the intervention provides an increment in length of life ($245,972/QALY) as opposed to when an intervention provides an increment in quality of life ($119,082/QALY) [40]. The addition of bevacizumab as per GOG-218 trial protocol is deemed cost-ineffective even if the highest defined willingness-to-pay threshold of $300,000/ QALY is considered. However the addition is cost effective as per ICON-7 protocol at this willingness-to-pay threshold of the clinicians. Bevacizumab first approved by the US FDA in the year 2004, with its patent expiring in the year 2019; this indication extension to ovarian cancer could be seen as a strategy to extend the exclusivity period by the manufacturer. The United States Patent and Trademark Office (USPTO) grants an additional 2 year data exclusivity for a new indication of an approved biologic, delaying the introduction of biosimilars in the market and leading to maintenance of higher drug prices. Our analysis suggests that the societal value of this extended exclusivity may be questionable. This study has certain limitations. In our model constant proportional hazard is assumed for progression and death. Though this is consistent with the data from the GOG-218 trial, the ICON-7 trial has shown evidence of non-proportional hazards among the patient population (test of proportional hazard p b 0.001) [7]. There is lack of prospective trial evidence and widespread consensus about the clinical value of multiple lines of therapy for patients with cancer recurrence. Hence, our results, based on retrospective data from a single institution, may have a limited perspective on subsequent therapy; although this is unlikely to alter our key findings [14]. Also our analysis didn't take into consideration the costs associated with end-of-life palliative care or hospice care in patients. However it is likely that such costs would be similar for both treatment arms. Future avenues of research are suggested. The current results can be updated once the median OS for the ICON-7 trial is attained. Such results can further establish a value-based price for bevacizumab for ovarian cancer patients. The current study found ICERs sensitive to QoL of patients, affected by choice of drugs and route of administration. Hence cost-effectiveness of a shift to intraperitoneal strategy and use of docetaxel rather than the current IV CPB regimen should be evaluated. The clinical effectiveness of continued use of bevacizumab in platinumsensitive recurrent patients may also impact its cost-effectiveness.
Conclusion The strategy of adding bevacizumab to first-line therapy is not costeffective. In contrast to previous findings, the ICERs are sensitive to the
median PFS and QoL associated with the use of bevacizumab. The strategy doesn't become cost-effective with biosimilar bevacizumab, or under typical clinical scenarios. However, the strategy does become cost-effective with biosimilar bevacizumab, for patients, with stage IV disease, ECOG PS 1 and at high-risk of disease progression. Conflict of interest statement None of the authors have any conflict of interest to declare.
Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx.doi. org/10.1016/j.ygyno.2014.01.021. References [1] Surveillance, epidemiology, and end result program of the National Cancer Institute. Accessed July 2013. Available at http://seer.cancer.gov/statfacts/html/ ovary.htm. [2] Morgan Jr RJ, Alvarez RD, Armstrong DK, et al. Ovarian cancer, version 3.2012. J Natl Compr Canc Netw 2012;10(11):1339–49. [3] Bristow RE, Tomacruz RS, Armstrong DK, et al. Survival effect of maximal cytoreductive surgery for advanced ovarian carcinoma during the platinum era: a meta-analysis. J Clin Oncol 2002;20(5):1248–59. [4] Armstrong DK, Bundy B, Wenzel L, et al. Intraperitoneal cisplatin and paclitaxel in ovarian cancer. N Engl J Med 2006;354(1):34–43. [5] Ozlos RF, Bundy BN, Greer BE, et al. Phase III Trial of carboplatin and paclitaxel compared with cisplatin and paclitaxel in patients with optimally resected Stage III ovarian cancer: A gynaecologic oncology group study. J Clin Oncol 2003;21(17):3194–200. [6] Burger RA, Brady MF, Bookman MA, et al. Incorporation of bevacizumab in the primary treatment of ovarian cancer. N Engl J Med 2011;365(26):2473–83. [7] Perren TJ, Swart AM, Pfisterer J, et al. A phase 3 trial of bevacizumab in ovarian cancer. N Engl J Med 2011;365(26):2484–96. [8] Clinical trial registry maintained by US National Institutes of Health. Accessed July 2013. Available http://clinicaltrials.gov/show/NCT01167712. [9] National Cancer Institute at the National Institute of Health. Accessed July 2013. Available: http://www.cancer.gov/cancertopics/pdq/treatment/ovarianepithelial/ HealthProfessional/page6. [10] Lipscomb J, Weinstein MC, Torrance GW. Time preference. In: Gold M, Siegel J, Russell LB, et al, editors. Cost-effectiveness in health and medicine. Oxford University Press; 1996. p. 214–35. [11] Choosing interventions that are cost effective (WHO-CHOICE). Accessed July 2013. Available: http://www.who.int/choice/costs/CER_thresholds/en/index.html. [12] Keeler E. Decision tree and Markov models in cost-effectiveness research. In: Sloan F, editor. Valuing health care: costs, benefits, and effectiveness of pharmaceuticals and other medical technologies. Cambridge University Press; 1995. p. 185–206. [13] Baldwin LA, Huang B, Miller RW, et al. Ten-year relative survival for epithelial ovarian cancer. Obstet Gynecol 2012;120(3):612–8. [14] Rose PG. Pegylated liposomal doxorubicin: optimizing the dosing schedule in ovarian cancer. Oncologist 2005;10(3):205–14. [15] Griffiths RW, Zee YK, Evans S, et al. Outcomes after multiple lines of chemotherapy for platinum-resistant epithelial cancers of the ovary, peritoneum, and fallopian tube. Int J Gynecol Cancer 2011;21(1):58–65. [16] Greving JP, Vernooij F, Heintz AP, et al. Is centralization of ovarian cancer care warranted? A cost-effectiveness analysis. Gynecol Oncol 2009;113(1):68–74. [17] Lesnock JL, Farris C, Krivak TC, et al. Consolidation paclitaxel is more cost-effective than bevacizumab following upfront treatment of advanced epithelial ovarian cancer. Gynecol Oncol 2011;122(3):473–8.
Please cite this article as: Mehta DA, Hay JW, Cost-effectiveness of adding bevacizumab to first line therapy for patients with advanced ovarian cancer, Gynecol Oncol (2014), http://dx.doi.org/10.1016/j.ygyno.2014.01.021
D.A. Mehta, J.W. Hay / Gynecologic Oncology xxx (2014) xxx–xxx [18] Havrilesky LJ, Broadwater G, Davis DM, et al. Determination of quality of life-related utilities for health states relevant to ovarian cancer diagnosis and treatment. Gynecol Oncol 2009;113(2):216–20. [19] Gordon AN, Fleagle JT, Guthrie D, et al. Recurrent epithelial ovarian carcinoma: a randomized phase III study of pegylated liposomal doxorubicin versus topotecan. J Clin Oncol 2001;19(14):3312–22. [20] Piganta S, De Placido S, Biamonte R, et al. Residual neurotoxicity in ovarian cancer patients in clinical remission after first-line chemotherapy with carboplatin and paclitaxel: The Multicenter Italian Trial in ovarian cancer (MITO-4) retrospective study. BMC Cancer 2006;6:5. [21] Mandelblatt JS, Fryback DG, Weinstein MC, et al. Assessing the effectiveness of health interventions. In: Gold M, Siegel J, Russell LB, et al, editors. Cost-effectiveness in health and medicine. Oxford University Press; 1996. p. 135–68. [22] Stark D, Nankivell M, Pujade-Lauraine E, et al. Standard chemotherapy with or without bevacizumab in advanced ovarian cancer: quality-of-life outcomes from the International Collaboration on Ovarian Neoplasms (ICON7) phase 3 randomised trial. Lancet Oncol 2013;14(3):236–43. [23] Monk BJ, Huang HQ, Burger RA, et al. Patient reported outcomes of a randomized, placebo-controlled trial of bevacizumab in the front-line treatment of ovarian cancer: a gynecologic oncology group study. Gynecol Oncol 2013;128(3):573–8. [24] van der Burg ME, deWit R, van Putten WL, et al. Weekly cisplatin and daily oral etoposide is highly effective in platinum pretreated ovarian cancer. Br J Cancer 2002;86(1):19–25. [25] Tewari D, Monk BJ, Hunter M, et al. Gemcitabine and cisplatin chemotherapy is an active combination in the treatment of platinum-resistant ovarian and peritoneal carcinoma. Invest New Drugs 2004;22(4):475–80. [26] Parmar MK, Ledermann JA, Colombo N, et al. Paclitaxel plus platinum based chemotherapy versus conventional platinum based chemotherapy in women with relapsed ovarian cancer: the ICON 4/AGO-OVAR 2.2 trial. Lancet 2003;361(9375):2099–20106. [27] Cantu MG, Buda A, Parma G, et al. Randomized controlled trial of single-agent paclitaxel versus cyclophosphamide, doxorubicin, and cisplatin in patients with recurrent ovarian cancer who responded to first-line platinum-based regimens. J Clin Oncol 2002;20(5):1232–7.
7
[28] Luce BR, Manning WG, Siegel JE, et al. Estimating costs in cost-effectiveness analysis. In: Gold M, Siegel J, Russell LB, et al, editors. Cost-effectiveness in health and medicine. Oxford University Press; 1996. p. 176–209. [29] Demographic data of the respondent (NHANES survey1999–2002). Accessed July 2013. Available: http://www.cdc.gov/nchs/nhanes/nhanes_questionnaires.htm. [30] Mosteller RD. Simplified calculation of body-surface area. N Engl J Med 1987;317(17):1098. [31] Carboplatin dosing. Accessed July 2013. Available: http://www.fda.gov/ AboutFDA/CentersOffices/OfficeofMedicalProductsandTobacco/CDER/ucm228974. htm. [32] Burudpakdee C, Zhao Z, Munakata J, et al. Economic burden of toxicities associated with metastatic colorectal cancer treatment regimens containing monoclonal antibodies. J Med Econ 2012;15(2):371–7. [33] Shinder GA, Paradis PE, Posman M, et al. Patient and workflow and costs associated with staff time and facility usage at a comprehensive cancer centre in Quebec, Canada — a time and motion study. BMC Health Serv Res 2012;12:370. [34] Yabroff KR, Davis WW, Lamont EB, et al. Patient time costs associated with cancer care. J Natl Cancer Inst 2007;99(1):14–23. [35] Hayman JA, Langa KM, Kabeto MU, et al. Estimating the cost of informal caregiving for elderly patients with cancer. J Clin Oncol 2001;19(13):3219–25. [36] Lee CK, Simes RJ, Brown C, et al. A prognostic nomogram to predict overall survival in patients with platinum-sensitive recurrent ovarian cancer. Ann Oncol 2013;24(4):937–43. [37] Barnett JC, Alvarez Secord A, Cohn DE, et al. Cost effectiveness of alternative strategies for incorporating bevacizumab into the primary treatment of ovarian cancer. Cancer 2013;119(20):3653–61. [38] Cohn DE, Kim KH, Resnick KE, et al. At what cost does a potential survival advantage of bevacizumab make sense for the primary treatment of ovarian cancer? A costeffectiveness analysis. J Clin Oncol 2011;29(10):1247–51. [39] Nadler E, Eckert B, Neumann PJ. Do oncologists believe new cancer drugs offer good value? Oncologist 2006;11(2):90–5. [40] Kozminski MA, Neumann PJ, Nadler ES, et al. How long and how well: oncologists' attitudes toward the relative value of life-prolonging v. quality of life-enhancing treatments. Med Decis Making 2011;31(3):380–5.
Please cite this article as: Mehta DA, Hay JW, Cost-effectiveness of adding bevacizumab to first line therapy for patients with advanced ovarian cancer, Gynecol Oncol (2014), http://dx.doi.org/10.1016/j.ygyno.2014.01.021
Contents lists available at ScienceDirect
Gynecologic Oncology journal homepage: www.elsevier.com/locate/ygyno
Cost-effectiveness of adding bevacizumab to first line therapy for patients with advanced ovarian cancer Darshan A. Mehta a, Joel W. Hay b,⁎ a b
Leonard D. Schaeffer Center for Health Policy and Economics, School of Pharmacy, University of Southern California, USA Clinical Pharmacy and Pharmaceutical Economics & Policy, Leonard D. Schaeffer Center for Health Policy and Economics, School of Pharmacy, University of Southern California, USA
H I G H L I G H T S • Three-state Markov model for GOG-218 and ICON-7 protocol showed adding bevacizumab is cost-ineffective, at willingness-to-pay threshold of $150,000/QALY. • Strategy is cost-effective in high-risk population using biosimilar bevacizumab as per ICON-7 protocol. • PSA results show negligible probability for cost-effective ICER for the GOG-218 trial, and approximately 50% probability for ICON-7 trial.
a r t i c l e
i n f o
Article history: Received 29 November 2013 Accepted 14 January 2014 Available online xxxx Keywords: Ovarian cancer Bevacizumab Cost-effectiveness Subgroup Platinum-resistant Platinum sensitive
a b s t r a c t Objective. To evaluate, from a societal perspective, the cost-effectiveness of adding bevacizumab to first-line therapy based on outcomes from the GOG-218 and ICON-7 trials. Methods. A three-state Markov model was used. The time horizon was until the death of 99% of the initial cohort of 1000 individuals. Costs and quality-adjusted life-years (QALYs) were discounted at an annual rate of 3%. All costs were adjusted to 2013 USD. The incremental cost-effectiveness ratio (ICER) was reported as incremental cost per QALY gained. The robustness of the result was checked with one-way sensitivity analyses and for relevant clinical situations (i.e. varying the drug of choice to treat cancer recurrence). Subgroup analysis was conducted to identify subgroup of population for whom the strategy could be cost-effective. The potential impact of biosimilar bevacizumab was considered, using a 30% price reduction. Results. For the GOG-218 study protocol, widely followed in US, the addition of bevacizumab results in an ICER of $2,420,691/QALY. For the ICON-7 study protocol, the ICER is $225,515/QALY. The results of the model were sensitive to the quality of life (QoL) and the median progression free survival (PFS). Biosimilar bevacizumab didn't reduce cost sufficiently to change conclusions. First-line augmentation is cost-effective, with biosimilar bevacizumab, for stage IV patients ($126,169/QALY), ECOG PS1 patients ($116,575/QALY) and for patients with suboptimal residual disease ($122,822/QALY) as per the ICON-7 protocol. Conclusion. Addition of bevacizumab, by in large, is cost-ineffective. It can become cost-effective with the ICON-7 protocol, in patients at high risk of progression using biosimilar bevacizumab. © 2014 Elsevier Inc. All rights reserved.
Introduction Epithelial ovarian cancer, a common form of gynecological malignancy, is the fifth most frequent cause of cancer death in women. Since the underlying pathophysiology leading to ovarian cancer is unknown, it isn't easily diagnosed. Patients usually present at stage III and higher when the tumor has already metastasized. At this stage the 5-year relative survival rate is 26.9% [1]. ⁎ Corresponding author at: USC Schaeffer Center for Health Policy and Economics, University Park Campus, UGW-Unit A, Los Angeles, CA 90089-7273, USA. Fax: +1 213 740 3460 (Attn: Jacky). E-mail address: [email protected] (J.W. Hay). URL: http://healthpolicy.usc.edu/ListItem.aspx?ID=4, http://pharmacyschool.usc. edu/faculty/profile/?id=374 (J.W. Hay).
The National Comprehensive Cancer Network (NCCN) recommends debulking surgery on presentation followed by adjuvant chemotherapy [2]. Clinicians consider the extent of primary debulking surgery as an important prognostic factor for the survival of patient. Bristow et al. report that for each 10% increase in maximal cytoreduction there is a 5.5% increase in median survival time [3]. The objective of adjuvant chemotherapy is to lower the risk of cancer recurrence and hence increase the overall survival (OS). NCCN recommends multiple treatment regimens post-primary cytoreductive surgery [2]. Clinical trials have compared the safety and efficacy of the different drug combinations and routes of administration for adjuvant first-line therapy [4,5]. The current gold standard is intravenous (IV) paclitaxel and carboplatin after primary debulking surgery. In patients with microscopic residual tumors (b1 cm in size) increased OS
0090-8258/$ – see front matter © 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ygyno.2014.01.021
Please cite this article as: Mehta DA, Hay JW, Cost-effectiveness of adding bevacizumab to first line therapy for patients with advanced ovarian cancer, Gynecol Oncol (2014), http://dx.doi.org/10.1016/j.ygyno.2014.01.021
2
D.A. Mehta, J.W. Hay / Gynecologic Oncology xxx (2014) xxx–xxx
is achieved by administering the drugs intraperitoneally (IP) [4]. Based on pathophysiological studies clinical trials have evaluated the effect of adding a VEGF inhibitor like bevacizumab to the current first-line therapy. Notable among these trials are, GOG-218, ICON-7 and the current ongoing GOG-262 [6–8]. The GOG-218 trial reports a median progression free survival (PFS) of 10.3 months with current first line therapy and 14.1 months for the bevacizumab throughout the group. The trial does not report significant difference in OS among the comparative arms [6]. The ICON-7 trial reports a median PFS of 22.4 months for the standard therapy and 24.1 months with bevacizumab. In patients at high risk of progression, the trial reports a median PFS of the standard therapy of 14.5 months and 18.1 months with bevacizumab added; with respective median OS of 28.8 and 36.6 months [7]. Based on the results, the latest guideline of NCCN recommends adding bevacizumab to the current gold standard therapy among other treatment options [2]. Around 80% of patients diagnosed with ovarian cancer relapse after first-line chemotherapy [9]. The relapse is either platinum resistant or platinum sensitive based on time of recurrence post-cessation of firstline therapy. Relapse within 6 months of cessation is defined as platinum resistant and post 12 months as platinum sensitive [9].
Model structure To evaluate the cost-effectiveness of adding bevacizumab to the first-line therapy, a three state Markov model was used (see supplement S1). The states were; stable state, progression state and death. The clinical information for the model was based on the results of the GOG-218 and the ICON-7 trial [6,7]. The population cohort matches the inclusion criteria of the GOG-218 trial considered for the model [6]. This included patients with median age of 60 and ECOG status of 0 to 2, previously untreated, FIGO stage 3/ stage 4 epithelial ovarian cancer, having undergone a primary debulking surgery, with either microscopic or macroscopic residual lesions. A separate analysis was conducted on a high-risk population cohort modeled on the ICON-7 trial [7]. For this, the patient cohort included those with high risk of progression, characterized by N1.0 cm residual disease after debulking surgery. Other criteria remain the same as the cohort for the GOG-218 trial. The model is developed from the US societal perspective. The cycle length is 3 weeks, with an initial cohort of 1000 patients having characteristics as defined for the two trials respectively. The time horizon considered for the model is until death of 99% of the cohort. Costs and health benefits are discounted at a societal discount rate of 3% [10].
Patients enter the model in the stable disease state. The disease progression state captures patients who have failed to respond to the first-line therapy and have cancer recurrence. Both these disease states also reflect the QoL and cost associated with medical treatment, including drug therapy and associated side-effects. The death state is the absorbing state in the model.
Base case For the base case, the treatment pattern was modeled from the clinical trials [6,7]. In the stable disease state, patients in the carboplatin and paclitaxel (CP) arm received six cycles of CP. For the carboplatin, paclitaxel and bevacizumab (CPB) arm, patients received 6 cycles of CPB followed by 18 additional cycles of bevacizumab in the GOG-218 trial and 12 additional cycles in the ICON-7 trial. In line with the clinical trials, bevacizumab costs were incorporated in the model beginning from cycle 2, as the drug is known to delay surgical healing [6,7]. After completing chemotherapy, patients with stable disease were assumed to have disease assessment every 3 months for 2 years and then every 6 months until the end of the model. Based on expert inputs, platinum resistant progression was assumed to be treated with pegylated doxorubicin, 50 mg/m2 administered every 4 weeks. This clinical assumption was later tested during sensitivity analysis with 40 mg/m2 pegylated doxorubicin and with different lines of therapy. During chemotherapy cycles, for the stable state, it is assumed that disease assessment occurs every 6 cycles, with a single office visit with the oncologists per 3-week cycle. For the progression phase disease assessment is assumed at every 3 cycles and 2 office visits per cycle. Our base case societal willingness-to-pay ICER threshold was assumed to be $150,000/QALY; roughly three time the US GDP per capita [11].
Transition probabilities For calculating the transition probabilities, the median PFS and OS, were obtained from the clinical trials [6,7]. Under the declining exponential approximation of life expectancy (DEALE) method assumption of constant proportional hazard, median PFS and OS were converted into rates of progression and death and subsequently into respective transition probabilities for the three week cycles [12]. For the base case analysis, only mortality risk due to ovarian cancer was considered (see Table 1).
Table 1 Base case parameters. Base case transition probabilities Allowable transitions
Stable to progressive Stable to death Stable to stable Progressive to death Progressive to progressive
GOG 218 trial
ICON 7 trial
CPB arm
CP arm
CPB arm
CP arm
0.0334 0.0120 0.9521 0.0525 0.9449
0.0455 0.0121 0.9399 0.0525 0.9449
0.0295 0.130 0.9550 0.0525 0.9449
0.0446 0.0165 0.9364 0.0525 0.9449
Base case utility weights States
Stable state chemotherapy cycles 1–6 Stable state: 12 months post-chemotherapy Stable state: post 12 months Progression phase — throughout
GOG 218 trial
ICON 7 trial
CPB arm
CP arm
CPB arm
CP arm
0.596,16,17,18 – 0.8516 0.6316,18,19
0.6710,16,17,18 0.7817,20 0.8516 0.6316,18,19
0.797,16,17,18 – 0.8516 0.6316,18,19
0.8112,16,17,18 0.8317,19 0.8516 0.6316,18,19
Please cite this article as: Mehta DA, Hay JW, Cost-effectiveness of adding bevacizumab to first line therapy for patients with advanced ovarian cancer, Gynecol Oncol (2014), http://dx.doi.org/10.1016/j.ygyno.2014.01.021
D.A. Mehta, J.W. Hay / Gynecologic Oncology xxx (2014) xxx–xxx
The base model was subsequently augmented in a sensitivity analysis to include patients' additional mortality risks [13]. The underlying age-specific mortality rates from the National Vital Statistics (NVS) report by the Center for Disease Control (CDC) were used. The ovarian cancer specific mortality rate was calculated and this was added to the age-specific mortality rates. For the time period post the median follow up period in the respective clinical trials, the transition probability was calculated based on the combined rate of death due to ovarian cancer and the underlying natural rate of death . The hazard ratios of progression for the two trials were used to determine the stable to progression transition probability for different prognostic factors of progression [6,7]. The study incorporated clinically relevant scenarios beyond the clinical trial findings. Based on expert input, treatment with 40 mg/m2 doxorubicin was modeled for patients experiencing disease progression. Since this reduced dosage is equally efficacious with the base case dosage, the progression to death transition probability was modeled to be the same as the base case [14]. Due to lack of prospective clinical trial evidence, multiple lines of chemotherapy for treating cancer recurrence were modeled, based on Griffith et al. [15]. The lines of therapy were selected based on widespread use. The method similar to the base case was adopted for calculating transition probabilities. During the sensitivity analysis best case and worst case scenarios were modeled based on published ranges of model parameters (see supplement S2). Quality of life assessment Table 1 shows the utility weights used for the model. The literature derived utility weights for the stable and progression states are adjusted for side-effects due to chemotherapy administration for the two arms [16–19]. Also as observed in Table 1, for the stable state different utility weights are associated with different time points. For the CP arm, the utility weight was adjusted for a period of 12 months for residual neurotoxicity due to chemotherapy cycles [20]. This was not replicated for the CPB arm. Occurrence of residual neurotoxicity is not documented with bevacizumab and hence residual neurotoxicity due to CP administration was assumed to wane off during the maintenance bevacizumab cycles. The utility weights used in the model were based on preferences elicited from the general population using the Time Trade-Off (TTO) method [21]. In line with the results of the QoL assessment done alongside the clinical trials, the utility weights associated with the CPB arm are lower than for the CP arm and gradually increase for both the arms [22,23]. During the sensitivity analysis, when the preferred drug for treating cancer recurrence is varied, the utility weight is adjusted based on incidence of side-effects from their respective clinical trials [24–27].
3
Costs A mixture of micro-costing and gross costing methodologies was used to determine the cost associated with each Markov cycle [28]. Direct costs included drug acquisition cost, laboratory costs for patient evaluation prior to each chemotherapy cycle, imaging costs for evaluating disease progression, cost of treating side-effects and office visits. The drug costs were derived from the US Veteran Affairs Federal Supply Schedule (FSS), as a closest approximation for fully discounted drug prices. The required dosage for bevacizumab was calculated based on mean patient body weight for 60 year old women [29]. The paclitaxel required dosage was based on body surface area calculated following the Mosteller formula [30]. For the carboplatin dosing, the normal renal function was assumed with glomerular filtration rate of 125 ml/min. The Calvert formula was made use of to calculate the dose required to attain concentration equivalent to area under curve (AUC) 6 [31]. This dosage was used to calculate the number of units required meeting the dose. In instances where multiple companies supplied a drug, the one with the lowest FSS price was utilized. For all the three drugs, based on expert opinion, no vial sharing was assumed. For the stable disease state, the model included only the cost of treating hypertension for the GOG-218 study as it was the only sideeffect with statistically significant difference between the two arms [6]. The cost of treating side-effects derived from the literature was weighted based on differential incidence of hypertension observed in the clinical trial [32,6]. The cost of treating side-effects was accounted for during the first cycle and no side effect cost was incorporated for the remaining cycles. For the progression state, the costs of treating side-effects were weighted by incidence rates that were adjusted to reflect incidence every 4 weeks [32,19]. Burudapakdee et al. reported hospital charges from the Healthcare Cost and Utilization Project (HCUP) and hence a nationwide Medicare inpatient mean cost-tocharge ratio of 0.33 (FY 2013) (0.165–0.709) was used to define costs for both stable and progression states. The lab technician time associated with disease assessment was derived from literature [33]. The nursing time associated with the chemotherapy administration was based on the time needed for infusion of drugs as per the drug labeling information. The pharmacist time for preparation of drug was the same for all the drugs as the procedure involved reconstitution of drugs as IV infusion by diluting the content of the single use vial in a 100 ml of 0.9% sodium chloride injection, USP. These were valued at the mean hourly compensation rates using the Bureau of Labor Statistics — National Compensation Survey. Indirect costs in the model included unpaid patient and caregiver time and costs. The patient time was derived from the literature [34]. This was valued at the mean hourly compensation rate of women with mean age of 60 years. The time associated with informal caregiving was assumed to be at 6.8 h per day for the stable state and 10 h
Table 2 Results. Results — base case and biosimilar bevacizumab GOG 218 trial CPB arm Life expectancy (weeks) Life expectancy (years) Incremental life expectancy (years) Total cost — base case (USD) Total cost — biosimilar bevacizumab (USD) Incremental cost (USD) Incremental cost — biosimilar (USD) QALY Incremental QALY ICER — base case ICER — biosimilar bevacizumab
108 1.97 201,423 162,500
1.306
ICON 7 trial CP arm 97.12 1.78 0.19 70,158 70,158 131,265 92,342 1.252 0.05 2,420,691/QALY 1,702,968/QALY
CPB arm 104.45 2.01 125,114 107,598
1.508
CP arm 86.84 1.67 0.34 63,311 63,311 61,803 44,286 1.234 0.27 225,515/QALY 161,603/QALY
USD: US dollars.
Please cite this article as: Mehta DA, Hay JW, Cost-effectiveness of adding bevacizumab to first line therapy for patients with advanced ovarian cancer, Gynecol Oncol (2014), http://dx.doi.org/10.1016/j.ygyno.2014.01.021
4
D.A. Mehta, J.W. Hay / Gynecologic Oncology xxx (2014) xxx–xxx
per day for the progression state [35]. This was valued at the mean hourly compensation of a home health aide. Results The base case results for both the clinical trials resulted in the CPB strategy being cost-ineffective at a cost-effectiveness willingnessto-pay threshold of $150,000/QALY. The base case ICER for GOG218 trial is $2,420,691/QALY and $225,515/QALY for ICON-7 trial. Considering the different patient populations for both the clinical trials, the strategy is cost-ineffective in the universal pool of patients of GOG-218 trial and in the high-risk subset of individuals of the ICON-7 trial (see Table 2). Sensitivity analysis One-way sensitivity analyses on the base case parameters, for both the trials, were done by increasing and decreasing values by 25%. In cases where a 25% increase would create nonsensical values (e.g., utility weights, transition probabilities), 1 was considered as the upper bound value for the parameter. For the cohort modeled on the GOG-218 trial, the CPB strategy was dominated by CP with 25% parameter decreases in the median PFS, OS, QoL for the CPB arm, or with increases in the median PFS, OS and QoL for the CP arm. The study didn't find the ICER sensitive to the cost of the drugs (see Fig. 1a). For the high-risk sub-group modeled on the ICON-7 trial the ICERs were sensitive to stable state utility weight for the CPB strategy and
stable to progressive state transition probabilities for both the arms. The CPB strategy became cost-effective, when the median PFS increased to 21.43 months, and the utility weight for the stable state increased to 0.9875 (see Fig. 1b). At the assumed willing-to-pay threshold, probabilistic sensitivity analysis (PSA) on the base case parameters resulted in a miniscule probability of ICER becoming cost-effective for the GOG-218 trial, and approximately 50% probability for ICON-7 trial (see Supplements S2, S3). A threshold sensitivity analysis was done on the cost of bevacizumab, median PFS and OS for both clinical trials. For the ICON-7 trial, at constant median PFS for CP, the CPB strategy becomes cost-effective when the median PFS exceeds 20.15 months. If the median PFS for CP is also varied, the CPB strategy becomes cost-effective at a combination of 20.59 months for the CPB arm and 10.74 months for the CP arm. The CPB strategy becomes cost-effective on reduction of per cycle cost to 38% of the current per cycle cost. Similarly, for GOG-218 study, under constant median PFS for the CP arm, the CPB strategy becomes cost-effective when median PFS exceeds 64.5 months. The GOG-218 protocol doesn't become cost-effective over the range of costs considered. However when the median PFS is increased to 49 months and the price of bevacizumab reduced to 58% of the current per unit cycle cost, the CPB strategy becomes cost-effective. Sensitivity analyses on the clinical assumptions resulted in the CPB strategy remaining cost-ineffective for 40 mg/m 2 dosage of doxorubicin, and for multiple lines of therapy for the progression phase (see supplement S1). Also, when platinum sensitive and platinum resistant recurrences were considered, the CPB strategy remained costineffective [36].
a) One-way sensitivity analysis for GOG-218 trial Change in ICER from baseline (2,490,691/QALY). -50
-40
-30
-20
-10
0
10
20
30
40
50
Follow up utility weight - CP Utility weight - Stable - CP Bevacizumab Cost Stable to Death (CP) Follow up utility weight - CPB Patient Body weight Progressive to Death (CP) Stable to Death (CPB)
CPB is dominated by CP
12 moth post first line - CP Stable to progressive (CP) Utility weight - Stable - CPB Utility weight - Progressive - CPB Progressive to Death (CPB) Utility weight - Progressive - CP Stable to progressive (CPB) Doxorubicin Cost Paclitaxel Cost Carboplatin Cost
X Axis represents change in ICER from baseline (2,490,691/QALY), values in USD 100,000s Fig. 1. a: One-way sensitivity analysis for GOG-218 trial. X axis represents change in ICER from baseline (2,490,691/QALY), values in USD 100,000s. 1b: One-way sensitivity analysis for the ICON-7trial. X axis represents change in ICER from the baseline (225,515/QALY), values in USD 1,000s.
Please cite this article as: Mehta DA, Hay JW, Cost-effectiveness of adding bevacizumab to first line therapy for patients with advanced ovarian cancer, Gynecol Oncol (2014), http://dx.doi.org/10.1016/j.ygyno.2014.01.021
D.A. Mehta, J.W. Hay / Gynecologic Oncology xxx (2014) xxx–xxx
5
b) One-way sensitivity analysis for the ICON-7trial Change in ICER from the baseline (225,515/ QALY) (150)
(100)
(50)
-
50
100
150
200
250
300
Utility weight - Stable - CPB Stable to progressive (CP) Stable to progressive (CPB) Utility weight - Progressive - CP Follow up utility weight - CPB Utility weight - Progressive - CPB Stable to Death (CPB) Progressive to Death (CP) 12 moth post first line - CP Stable to Death (CP) Patient Body weight Bevacizumab Cost Utility weight - Stable - CP Progressive to Death (CPB) Follow up utility weight - CP Doxorubicin Cost Paclitaxel Cost Carboplatin Cost
X axis represents change in ICER from the baseline (225,515/ QALY), values in USD 1,000s Fig. 1 (continued).
Impact of biosimilar bevacizumab With the product patent on bevacizumab expiring in July 2019 in the US and 2018 in the EU, currently 7 companies are developing biosimilar bevacizumab. Though the Biologics Price Competition and Innovation (BPCI) Act, 2009 defined an abbreviated biosimilar approval process, it resulted in many unanswered questions regarding the pathway to FDA approval. Considering the complex regulatory hurdles and uncertain data requirements for biosimilar approval, reduced competition due to high entry barrier; similarly high price discounts as for small molecule generics isn't expected for biosimilars in the US. As there is no precedence of a biosimilar price for the US, a 30% discount on the current FSS price of bevacizumab was assumed. This is in line with the pricing of the biosimilar Remicade by Celltrion and Hospira, launched in June 2013 in the EU (see Table 2). Subgroup analysis Using hazard ratios of progression, ICERs were calculated for prognostic subgroups [6]. The results for risk factors, as indicated in
Table 3 and supplement S4, show that CPB strategy following GOG218 treatment protocol is cost-ineffective across all subgroups. The hazard ratios of progression for ICON-7 were derived from the entire population set of the ICON-7 trial [6]. Since the median OS for the entire population hasn't been attained, these hazard ratios were adjusted, using the adjustment factor of 0.839, to reflect the hazard for a high risk population (see Table 3 and supplement S5) [7]. From the resulting ICERs we infer that the strategy is cost-ineffective for any of the subgroups as per the base case scenario. The strategy does become cost-effective with biosimilar bevacizumab, in patients at high risk of progression with FIGO stage IV disease, suboptimal residual disease or ECOG PS 1. Discussion From the societal perspective, the strategy of adding bevacizumab to the first-line therapy is cost-ineffective at a willingness-to-pay threshold of $150,000/QALY. However, probabilistic sensitivity analysis varying the base case parameters results in CPB therapy following the ICON-7 protocol having a 50% chance of being cost-effective at a $150,000/QALY threshold, while the GOG-218 protocol remains predominantly cost-
Please cite this article as: Mehta DA, Hay JW, Cost-effectiveness of adding bevacizumab to first line therapy for patients with advanced ovarian cancer, Gynecol Oncol (2014), http://dx.doi.org/10.1016/j.ygyno.2014.01.021
6
D.A. Mehta, J.W. Hay / Gynecologic Oncology xxx (2014) xxx–xxx
Table 3 Subgroup analysis. Risk factor
Hazard ratio
Hazard ratio adjusted for high risk population
Rate of progresion
Transition probability
ICER (cost/QALY)
ICER with biosimilar bevacizumab (cost/QALY)
ICON-7 trial ECOG PS 1 FIGO IV Suboptimal residual disease
0.66 0.69 0.68
0.554 0.579 0.571
0.0365 0.0382 0.0376
0.0250 0.0261 0.0257
165,429 178,231 173,844
116,575 126,169 122,822
GOG 218 trial ECOG PS 1 or 2 FIGO stage IV FIGO III and macroscopic b= 1 cm FIGO III and macroscopic N 1 cm
0.69 0.70 0.62 0.76
– – – –
0.0455 0.0461 0.0407 0.0503
0.0310 0.0313 0.0278 0.0342
1,408,864 1,504,219 874,929 3,133,968
988,171 1,055,531 610.896 2,206,730
The table depicts ICERs for risk factors which make strategy cost effective as per ICON 7 trial. ICERs for comparative risk factors as per GOG 218 trial protocol depicted. For ICERs for all risk factors see supplements S5 and S6.
ineffective at any willingness-to-pay threshold. Prior cost effectiveness studies have defined cost-effectiveness threshold varying from $100,000/QALY to $200,000/QALY [17,37,38]. For clinical oncologist, minor effectiveness of intervention is considered of good value. Hence in clinical practice, oncologists demonstrate a willingness-to-pay threshold of $300,000/QALY [39]. Oncologists demonstrate a higher willingnessto-pay threshold if the intervention provides an increment in length of life ($245,972/QALY) as opposed to when an intervention provides an increment in quality of life ($119,082/QALY) [40]. The addition of bevacizumab as per GOG-218 trial protocol is deemed cost-ineffective even if the highest defined willingness-to-pay threshold of $300,000/ QALY is considered. However the addition is cost effective as per ICON-7 protocol at this willingness-to-pay threshold of the clinicians. Bevacizumab first approved by the US FDA in the year 2004, with its patent expiring in the year 2019; this indication extension to ovarian cancer could be seen as a strategy to extend the exclusivity period by the manufacturer. The United States Patent and Trademark Office (USPTO) grants an additional 2 year data exclusivity for a new indication of an approved biologic, delaying the introduction of biosimilars in the market and leading to maintenance of higher drug prices. Our analysis suggests that the societal value of this extended exclusivity may be questionable. This study has certain limitations. In our model constant proportional hazard is assumed for progression and death. Though this is consistent with the data from the GOG-218 trial, the ICON-7 trial has shown evidence of non-proportional hazards among the patient population (test of proportional hazard p b 0.001) [7]. There is lack of prospective trial evidence and widespread consensus about the clinical value of multiple lines of therapy for patients with cancer recurrence. Hence, our results, based on retrospective data from a single institution, may have a limited perspective on subsequent therapy; although this is unlikely to alter our key findings [14]. Also our analysis didn't take into consideration the costs associated with end-of-life palliative care or hospice care in patients. However it is likely that such costs would be similar for both treatment arms. Future avenues of research are suggested. The current results can be updated once the median OS for the ICON-7 trial is attained. Such results can further establish a value-based price for bevacizumab for ovarian cancer patients. The current study found ICERs sensitive to QoL of patients, affected by choice of drugs and route of administration. Hence cost-effectiveness of a shift to intraperitoneal strategy and use of docetaxel rather than the current IV CPB regimen should be evaluated. The clinical effectiveness of continued use of bevacizumab in platinumsensitive recurrent patients may also impact its cost-effectiveness.
Conclusion The strategy of adding bevacizumab to first-line therapy is not costeffective. In contrast to previous findings, the ICERs are sensitive to the
median PFS and QoL associated with the use of bevacizumab. The strategy doesn't become cost-effective with biosimilar bevacizumab, or under typical clinical scenarios. However, the strategy does become cost-effective with biosimilar bevacizumab, for patients, with stage IV disease, ECOG PS 1 and at high-risk of disease progression. Conflict of interest statement None of the authors have any conflict of interest to declare.
Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx.doi. org/10.1016/j.ygyno.2014.01.021. References [1] Surveillance, epidemiology, and end result program of the National Cancer Institute. Accessed July 2013. Available at http://seer.cancer.gov/statfacts/html/ ovary.htm. [2] Morgan Jr RJ, Alvarez RD, Armstrong DK, et al. Ovarian cancer, version 3.2012. J Natl Compr Canc Netw 2012;10(11):1339–49. [3] Bristow RE, Tomacruz RS, Armstrong DK, et al. Survival effect of maximal cytoreductive surgery for advanced ovarian carcinoma during the platinum era: a meta-analysis. J Clin Oncol 2002;20(5):1248–59. [4] Armstrong DK, Bundy B, Wenzel L, et al. Intraperitoneal cisplatin and paclitaxel in ovarian cancer. N Engl J Med 2006;354(1):34–43. [5] Ozlos RF, Bundy BN, Greer BE, et al. Phase III Trial of carboplatin and paclitaxel compared with cisplatin and paclitaxel in patients with optimally resected Stage III ovarian cancer: A gynaecologic oncology group study. J Clin Oncol 2003;21(17):3194–200. [6] Burger RA, Brady MF, Bookman MA, et al. Incorporation of bevacizumab in the primary treatment of ovarian cancer. N Engl J Med 2011;365(26):2473–83. [7] Perren TJ, Swart AM, Pfisterer J, et al. A phase 3 trial of bevacizumab in ovarian cancer. N Engl J Med 2011;365(26):2484–96. [8] Clinical trial registry maintained by US National Institutes of Health. Accessed July 2013. Available http://clinicaltrials.gov/show/NCT01167712. [9] National Cancer Institute at the National Institute of Health. Accessed July 2013. Available: http://www.cancer.gov/cancertopics/pdq/treatment/ovarianepithelial/ HealthProfessional/page6. [10] Lipscomb J, Weinstein MC, Torrance GW. Time preference. In: Gold M, Siegel J, Russell LB, et al, editors. Cost-effectiveness in health and medicine. Oxford University Press; 1996. p. 214–35. [11] Choosing interventions that are cost effective (WHO-CHOICE). Accessed July 2013. Available: http://www.who.int/choice/costs/CER_thresholds/en/index.html. [12] Keeler E. Decision tree and Markov models in cost-effectiveness research. In: Sloan F, editor. Valuing health care: costs, benefits, and effectiveness of pharmaceuticals and other medical technologies. Cambridge University Press; 1995. p. 185–206. [13] Baldwin LA, Huang B, Miller RW, et al. Ten-year relative survival for epithelial ovarian cancer. Obstet Gynecol 2012;120(3):612–8. [14] Rose PG. Pegylated liposomal doxorubicin: optimizing the dosing schedule in ovarian cancer. Oncologist 2005;10(3):205–14. [15] Griffiths RW, Zee YK, Evans S, et al. Outcomes after multiple lines of chemotherapy for platinum-resistant epithelial cancers of the ovary, peritoneum, and fallopian tube. Int J Gynecol Cancer 2011;21(1):58–65. [16] Greving JP, Vernooij F, Heintz AP, et al. Is centralization of ovarian cancer care warranted? A cost-effectiveness analysis. Gynecol Oncol 2009;113(1):68–74. [17] Lesnock JL, Farris C, Krivak TC, et al. Consolidation paclitaxel is more cost-effective than bevacizumab following upfront treatment of advanced epithelial ovarian cancer. Gynecol Oncol 2011;122(3):473–8.
Please cite this article as: Mehta DA, Hay JW, Cost-effectiveness of adding bevacizumab to first line therapy for patients with advanced ovarian cancer, Gynecol Oncol (2014), http://dx.doi.org/10.1016/j.ygyno.2014.01.021
D.A. Mehta, J.W. Hay / Gynecologic Oncology xxx (2014) xxx–xxx [18] Havrilesky LJ, Broadwater G, Davis DM, et al. Determination of quality of life-related utilities for health states relevant to ovarian cancer diagnosis and treatment. Gynecol Oncol 2009;113(2):216–20. [19] Gordon AN, Fleagle JT, Guthrie D, et al. Recurrent epithelial ovarian carcinoma: a randomized phase III study of pegylated liposomal doxorubicin versus topotecan. J Clin Oncol 2001;19(14):3312–22. [20] Piganta S, De Placido S, Biamonte R, et al. Residual neurotoxicity in ovarian cancer patients in clinical remission after first-line chemotherapy with carboplatin and paclitaxel: The Multicenter Italian Trial in ovarian cancer (MITO-4) retrospective study. BMC Cancer 2006;6:5. [21] Mandelblatt JS, Fryback DG, Weinstein MC, et al. Assessing the effectiveness of health interventions. In: Gold M, Siegel J, Russell LB, et al, editors. Cost-effectiveness in health and medicine. Oxford University Press; 1996. p. 135–68. [22] Stark D, Nankivell M, Pujade-Lauraine E, et al. Standard chemotherapy with or without bevacizumab in advanced ovarian cancer: quality-of-life outcomes from the International Collaboration on Ovarian Neoplasms (ICON7) phase 3 randomised trial. Lancet Oncol 2013;14(3):236–43. [23] Monk BJ, Huang HQ, Burger RA, et al. Patient reported outcomes of a randomized, placebo-controlled trial of bevacizumab in the front-line treatment of ovarian cancer: a gynecologic oncology group study. Gynecol Oncol 2013;128(3):573–8. [24] van der Burg ME, deWit R, van Putten WL, et al. Weekly cisplatin and daily oral etoposide is highly effective in platinum pretreated ovarian cancer. Br J Cancer 2002;86(1):19–25. [25] Tewari D, Monk BJ, Hunter M, et al. Gemcitabine and cisplatin chemotherapy is an active combination in the treatment of platinum-resistant ovarian and peritoneal carcinoma. Invest New Drugs 2004;22(4):475–80. [26] Parmar MK, Ledermann JA, Colombo N, et al. Paclitaxel plus platinum based chemotherapy versus conventional platinum based chemotherapy in women with relapsed ovarian cancer: the ICON 4/AGO-OVAR 2.2 trial. Lancet 2003;361(9375):2099–20106. [27] Cantu MG, Buda A, Parma G, et al. Randomized controlled trial of single-agent paclitaxel versus cyclophosphamide, doxorubicin, and cisplatin in patients with recurrent ovarian cancer who responded to first-line platinum-based regimens. J Clin Oncol 2002;20(5):1232–7.
7
[28] Luce BR, Manning WG, Siegel JE, et al. Estimating costs in cost-effectiveness analysis. In: Gold M, Siegel J, Russell LB, et al, editors. Cost-effectiveness in health and medicine. Oxford University Press; 1996. p. 176–209. [29] Demographic data of the respondent (NHANES survey1999–2002). Accessed July 2013. Available: http://www.cdc.gov/nchs/nhanes/nhanes_questionnaires.htm. [30] Mosteller RD. Simplified calculation of body-surface area. N Engl J Med 1987;317(17):1098. [31] Carboplatin dosing. Accessed July 2013. Available: http://www.fda.gov/ AboutFDA/CentersOffices/OfficeofMedicalProductsandTobacco/CDER/ucm228974. htm. [32] Burudpakdee C, Zhao Z, Munakata J, et al. Economic burden of toxicities associated with metastatic colorectal cancer treatment regimens containing monoclonal antibodies. J Med Econ 2012;15(2):371–7. [33] Shinder GA, Paradis PE, Posman M, et al. Patient and workflow and costs associated with staff time and facility usage at a comprehensive cancer centre in Quebec, Canada — a time and motion study. BMC Health Serv Res 2012;12:370. [34] Yabroff KR, Davis WW, Lamont EB, et al. Patient time costs associated with cancer care. J Natl Cancer Inst 2007;99(1):14–23. [35] Hayman JA, Langa KM, Kabeto MU, et al. Estimating the cost of informal caregiving for elderly patients with cancer. J Clin Oncol 2001;19(13):3219–25. [36] Lee CK, Simes RJ, Brown C, et al. A prognostic nomogram to predict overall survival in patients with platinum-sensitive recurrent ovarian cancer. Ann Oncol 2013;24(4):937–43. [37] Barnett JC, Alvarez Secord A, Cohn DE, et al. Cost effectiveness of alternative strategies for incorporating bevacizumab into the primary treatment of ovarian cancer. Cancer 2013;119(20):3653–61. [38] Cohn DE, Kim KH, Resnick KE, et al. At what cost does a potential survival advantage of bevacizumab make sense for the primary treatment of ovarian cancer? A costeffectiveness analysis. J Clin Oncol 2011;29(10):1247–51. [39] Nadler E, Eckert B, Neumann PJ. Do oncologists believe new cancer drugs offer good value? Oncologist 2006;11(2):90–5. [40] Kozminski MA, Neumann PJ, Nadler ES, et al. How long and how well: oncologists' attitudes toward the relative value of life-prolonging v. quality of life-enhancing treatments. Med Decis Making 2011;31(3):380–5.
Please cite this article as: Mehta DA, Hay JW, Cost-effectiveness of adding bevacizumab to first line therapy for patients with advanced ovarian cancer, Gynecol Oncol (2014), http://dx.doi.org/10.1016/j.ygyno.2014.01.021