The cost-effectiveness of alectinib in anaplastic lymphoma kinase-positive (ALK+) advanced NSCLC previously treated with crizotinib

Abstract

Introduction Anaplastic lymphoma kinase (ALK) targeting drugs provide an important option for advanced non-small cell lung cancer patients with this distinct tumor type; however, there is considerable uncertainty as to which drug provides the optimal value after crizotinib treatment. This study estimated the cost-utility of alectinib vs ceritinib from a US payer perspective.

Methods A cost-utility model was developed using partition survival methods and three health states: progression-free (PF), post-progression (PP), and death. Survival data were derived from the key clinical trials (alectinib: NP28761 & NP28673, ceritinib: ASCEND I and II). Costs included drugs, adverse events, and supportive care. Utilities were based on trial data and the literature. One-way and probabilistic sensitivity analyses (PSA) were performed to assess parameter uncertainty.

Results Treatment with alectinib vs ceritinib resulted in increases of 2.55 months in the PF state, 0.44 quality adjusted life-years (QALYs), and $13,868, yielding a mean cost/QALY of $31,180. In the PSA, alectinib had a 96% probability of being cost-effective at a willingness-to-pay of $100,000/QALY. Drivers of model results were drug costs and utilities in the PF health state. The ICER ranged from $10,600–$65,000 per QALY in scenario analyses, including a sub-group analysis limited to patients with prior chemotherapy and crizotinib treatment.

Conclusions Treatment with alectinib in ALK + crizotinib-treated patients increased time progression-free and QALYs vs ceritinib. The marginal cost increase was driven by longer treatment durations with alectinib. This model demonstrates that alectinib may be considered a cost-effective treatment after progression on crizotinib.

Keywords:

• Non-small cell lung cancer
• ALK
• Alectinib
• Ceritinib
• Cost-effectiveness

Introduction

Advances in our understanding of the etiology and molecular characteristics of cancer have improved our ability to target therapy according to the unique molecular characteristics of tumor tissue^1–4. Notably, there have been many developments in the treatment of non-small cell lung cancer (NSCLC) with a number of targeted drugs developed, approved, and being used to improve patients’ length and quality-of-life. Specifically, the development of agents targeting anaplastic lymphoma kinase fusion-gene positive (ALK+) tumors has provided an important option for patients with advanced stage ALK + NSCLC. Approximately 4% of advanced NSCLC patients have ALK + tumors, which translates into 2,300 new patients (previously treated with an ALK inhibitor) expected in 2016^5–7. Crizotinib is approved for treatment of ALK + metastatic NSCLC⁸. However, the majority of patients progress within the first year of treatment or develop intolerance^9–11. Two recently approved agents, alectinib and ceritinib, have been shown to provide clinical benefit for patients who progressed on or were intolerant to crizotinib^12–16. These next-generation ALK inhibitors fill an unmet clinical need for ALK + patients. A comparative evaluation of these two agents can inform subsequent treatment decisions for patients in this setting.

Alectinib is a selective inhibitor of ALK that received approval in 2015 by the US Food and Drug Administration (FDA) for the treatment of patients with ALK + metastatic NSCLC who have progressed on or are intolerant of crizotinib treatment. Two recent phase II studies (NP28761 & NP28673) in advanced ALK + NSCLC whose disease progressed on crizotinib (600 mg orally twice daily) demonstrated the efficacy and safety of alectinib in these patients^12,13,17. The NP28761 study was a single-arm, open-label, multi-center trial that evaluated 87 patients, and the NP28673 study was a single-arm, open-label, multi-center trial that evaluated 138 patients. Both studies met their primary end-points with objective response rates (ORR) of 52% (95% CI = 40–65) and 51% (95% CI = 42–60) (at the time of the updated data cut-off), respectively^12,13,17,18. Median progression-free survival for these two trials were 8.1 (95% CI = 6.2–12.6) and 8.9 (95% CI = 5.6–12.8) (see Table 1)^13,17,19. Median overall survival (OS) was not reached for either trial after 21 months, but the overall survival at 12 months was 71% (95% CI = 61–81) in the NP28761 study¹³.

Table 1. Comparison of alectinib and ceritinib trial data.

	ASCEND I^a^,^16	ASCEND II^c^,^14	NP28673^d^,^17^,^18	NP28761^e^,^13^,^21
	n (%; 95% CI)	n (%; 95% CI)	n (%; 95% CI)	n (%; 95% CI)
Study population characteristics
Study population size	163	140	138	87
Age (median, years)	52	51	52	54
Sex (% female)	88 (54%)	70 (50%)	77 (56%)	48 (55%)
ECOG or WHO performance status (≤1)	142 (87%)	120 (86%)	125 (91%)	78 (90%)
Prior chemotherapy	116 (71%)	140 (100%)	110 (80%)	64 (74%)
Baseline CNS metastases	98 (60%)	100 (71%)	84 (61%)	52 (60%)
Clinical outcomes
Objective response rate	75 (46%; 38–54)^b	50 (49%; 39–59)	62 (51%; 42–60)	35 (52%; 40-65)
Disease control rate	121 (74%; NA)^b	87 (85%; 77–92)	96 (79%; 70–86)	53 (79%; NA)
Median PFS (months)	7.0 (5.7–8.7)^b	7.2 (5.4–9.0)	8.9 (5.6–12.8)	8.1 (6.2–12.6)
Median duration of response (months)	8.8 (6.0–13.1)^b	9.7 (5.6–12.9)	14.1 (10.9–NR)	13.5 (6.7–NR)
Median OS (months)	16.7 (14.8–NR)	14.9 (13.5–NR)	NR	NR
OS at 12 months (%)	67 (59–74)	NA	NA	71 (61–81)
Median follow-up time, months (range)	11.1 (6.7–15.2)	11.3 (0.1–18.9)	14.5 (0.6–20.5)	9.9 (6.2–12.9)

ECOG, Eastern Cooperative Oncology Group; CNS, central nervous system; PFS, progression free survival; OS, overall survival; NA, not available.

^a Based on data from crizotinib pre-treated patients (data cut: April 14, 2014)¹⁶.

^b Blinded independent review committee assessed.

^c Based on blinded independent review committee assessed data, response outcomes limited to per protocol population (n = 102) (data cut: August 13, 2014)¹⁴.

^d Based on independent review committee assessed data from updated data set (data cut: April 27, 2015), response data limited to response evaluable population (n = 122)¹⁷.

^e Based on independent review committee assessed data from latest data cut (date: April 27, 2015), response data limited to response evaluable population (n = 69)^13,21.

Ceritinib was approved in 2014 based on the results of a multi-center, single-arm, open-label clinical trial enrolling a total of 246 advanced ALK + NSCLC patients, 163 of whom had progressed on or were intolerant of crizotinib¹⁶. Another similar trial, ASCEND II, was a single-arm, open-label, multi-center phase II study of ceritinib in adult patients with ALK + non-small cell lung cancer previously treated with chemotherapy and crizotinib that enrolled 140 patients¹⁴. All patients received ceritinib at a dose of 750 mg once daily. The ORR for the ASCEND I prior crizotinib sub-group was 46% (95% CI = 38–54)^15,16. ASCEND II had an ORR of 49% (95% CI = 39–59)¹⁴. Progression-free and overall survival results were also reported with median PFS estimates of 7.0 (95% CI = 5.7–8.7) and 7.2 (95% CI = 5.4–9.0), respectively^14,15. Median OS was 16.7 (95% CI = 14.8–NR) months for the prior crizotinib sub-group in ASCEND I, with an overall survival at 12 months of 67% (95% CI = 59–74)¹⁶. ASCEND II reported a median OS estimate of 14.9 (95% CI = 13.5–NR) months (see Table 1)^14,15.

Alectinib and ceritinib are both approved for the same indication, have been evaluated in similar patient populations, and are recommended by clinical guidelines as options after crizotinib²⁰. However, they have not been compared head-to-head in clinical trials and, therefore, it is not yet clear which option is likely to provide the optimal choice for patients after crizotinib progression. Our objective was to estimate the clinical and economic impact of treatment with alectinib vs ceritinib from the US payer perspective.

Materials and methods

Markov model overview

We developed a model in Microsoft Excel (Redmond, WA) with three health states: progression-free (PF), post-progression (PP), and death (Figure 1). Time in each health state was estimated using partition survival methods (i.e. area under the survival curves). We applied a discount rate of 3% and used a US payer perspective. We used a cycle length of 1 month. Costs are reported in 2016 US dollars. For the cost-effectiveness analysis, the results are reported in terms of incremental cost per life year (LYs) gained and incremental costs per quality-adjusted life years (QALYs) gained.

Figure 1. Model structure.

Clinical inputs

There are no clinical trials directly comparing the efficacy of alectinib and ceritinib. A review of the clinical trial populations and outcomes suggest that a naïve comparison, although limited, may still be informative given the similarity of the patient populations and the clinical setting in which these agents are expected to be used. We, therefore, performed a naive comparison (i.e. a comparison of treatment outcomes across separate trials) using available clinical trial data. Data for alectinib and ceritinib progression-free survival (PFS) and overall survival (OS) were derived from the key clinical trials for these drugs in this setting (alectinib: pooled data from NP28761 & NP28673, ceritinib: ASCEND I and II)^{12–14,16,17,19,22}. We used the Kaplan Meier (KM) curves until the end of the study and extrapolated beyond the end of study using a Weibull parametric function (selected based on AIC values, visual fit, and clinical plausibility) (See Appendix Table 2). The estimates for the parametric functions used to extrapolate the alectinib curves were derived from clinical trial data²³. The estimates for ceritinib were derived using the available KM data and the methodology described previously^14,16,22,24. In the base case, we used data from ASCEND I for ceritinib PFS^16,22. As there is not yet publicly available KM OS data from ASCEND I, we used data from ASCEND II for ceritinib OS¹⁴. Although the ASCEND II population includes all patients with prior chemotherapy as well as prior crizotinib treatment, this population is still considered suitable for comparison given the similarity in the PFS data (ASCEND I: prior ALK inhibitor patients = median PFS of 6.93 months vs ASCEND II: median PFS of 7.2 months)^14,16,22. However, a sub-group analysis was also conducted which limited the alectinib PFS and OS data to comparable patients with both prior chemotherapy and prior crizotinib treatment and compared outcomes to a ceritinib treated population using ASCEND II data.

Adverse events

Adverse event rates were based on the clinical trial data and package inserts^23,25. Grade 3 and 4 adverse events (AEs) occurring in greater than 5.0% of patients or those considered to be resource intensive (i.e. cost per event greater than $1,000) were included. We did not include any AE costs related to alanine aminotransferase or aspartate aminotransferase laboratory levels. The costs of managing each adverse event were estimated using treatment assumptions based on clinical guidelines and expert opinion and drug and medical services unit costs^20,26–29. The AE rates and costs are provided in Table 2. Treatment assumptions are provided in Appendix Table A1.

Table 2. Grade 3/4 adverse event rates^a and costs^20,26–29.

Adverse event	Cost per event	Alectinib		Ceritinib
Anemia	$2,762	2%	$55.24	5%	$140.86
Fatigue	$208	1%	$2.50	5%	$10.63
Infection (including pneumonia)	$9,575	0%	$0.00	5%	$450.01
Nausea	$208	0%	$0.00	6%	$12.29
Diarrhea	$208	1%	$2.50	6%	$12.29
Vomiting	$4,339	0%	$17.36	5%	$203.93
Hyperglycemia	$208	2%	$4.17	6%	$12.29
Lipase increase	$208	0%	$0.00	6%	$13.13
Blood alkaline increase	$208	1%	$2.50	5%	$10.63
Total			$84.26		$866.07

^a Adverse events were assumed to occur during the first month of treatment.

Table 3. Model inputs.

	Input	Low	High	Distribution	Source
Utilities
Utility: Alectinib PF	0.79	0.71	0.87	Beta	^23
Utility: Ceritinib PF	0.73	0.66	0.80	Beta	^30
Utility: Disease progression	0.46	0.41	0.51	Beta	^31^,^32
Cost inputs
Alectinib drug cost	$51	$46	$56	Normal^a	^26
Ceritinib drug cost	$96	$86	$105	Normal	^26
Valsartan	$44	$39	$48	Normal	^26
Pegfilgrastim, 0.6 mL syringe	$5156	$4640	$5671	Normal	^26
Epoetin alfa, vial	$32	$29	$35	Normal	^26
Prednisone	$21	$19	$23	Normal	^26
Cost continuing care (per month)	$750	$675	$825	Normal	^34
Cost terminal care (per month)	$8956	$8060	$9851	Normal	^34
Cost of Chest CT (CPT: 71260)	$231	$208	$254	Normal	^28
Cost of Brain CT (CPT: 70460)	$105	$95	$116	Normal	^28
Follow-up Consultation (CPT code 99215)	$208	$131	$160	Normal	^28
Outpatient Consultation (CPT code 99205)	$208	$188	$229	Normal	^28
Platelet Transfusion (CPT code 36430)	$6427	$5784	$7070	Normal	^36
ER Visit (CPT code 99285)	$175	$158	$193	Normal	^28
IV infusion per hour (CPT code 96413)	$136	$123	$150	Normal	^28
Subcutaneous or intramuscular injection (CPT code 96372)	$25	$23	$28	Normal	^28
Inpatient Stay: Complications from treatment	$8261	$7435	$9087	Normal	^28^,^37
Inpatient Consultation on Admission Day (CPT code 99254)	$169	$152	$186	Normal	^28
Inpatient Physician Fee (CPT code 99252)	$76	$68	$83	Normal	^28
Inpatient Consultation on Discharge Day (CPT code 99217)	$73	$66	$81	Normal	^28
Inpatient Stay: Respiratory Illness with Complications	$11,177	$10,059	$12,294	Normal	^28^,^37
Inpatient Stay: Respiratory Illness with Major Complications	$7973	$7176	$8770	Normal	^28^,^37
Treatment for vomiting	$4339	$3905	$4773	Normal	^28^,^37

^a Values limited to positive values.

PF: progression-free; CT: computerized tomography; CPT: current procedural terminology; ER: emergency room; IV: intravenous.

Quality-of-life inputs

Health-related quality-of-life was incorporated into the model using utility values (Table 3). The PF utility estimates were calculated from patient level data for alectinib and were obtained from publicly available data for ceritinib. Both estimates were calculated by mapping EORTC data from the clinical trials for alectinib and ceritinib to EQ-5D^23,30. The PF utility estimates were 0.79 for alectinib and 0.73 for ceritinib^23,30. We also evaluated a scenario with equal utilities as part of the sensitivity analysis. The utility of the progression health state was the same for all comparators, and estimated at 0.46 based on Zhou et al.³¹ and Naffees et al.³².

Cost inputs

We included costs related to drug treatment, adverse events, and supportive care (Table 3). Treatment costs per month are calculated using dosing schedules and unit costs for the drugs based on the wholesale acquisition cost^23,25,33. The model calculates the total cost of treatment assuming that patients are treated until progression or death. Alectinib and ceritinib are both oral drugs and, therefore, no administration costs are included.

Supportive care costs were also included in the progression-free and post-progression health states. The additional cost of care for patients in the progression-free health state was assumed to be $261 per month based on routine monitoring of metastatic NSCLC patients (i.e. one physician visit per month, one chest CT scan every 8 weeks, and one brain/head CT scan every 8 weeks)²⁰. The monthly cost during progression is based on a study by Mariotto et al.³⁴, which estimated the cost of NSCLC in the initial, continuing, and terminal care (last 12 months of life) phases. The cost is highest for the terminal care phase and, therefore, the cost per month in progression is treatment-specific, depending on the average time spent in the post-progression health state, i.e. the longer the time spent in progression beyond 12 months, the lower the per month cost in the progression health state. The monthly cost of the progression-free and post-progression health states is then multiplied by the proportion of patients in each health state in each month in the model and then summed to calculate the total cost.

Sensitivity analysis

The uncertainty in the model was evaluated using one-way and probabilistic sensitivity analysis (PSA). Ranges were based on 95% confidence intervals or plausible ranges, and distributional assumptions were based on recommended guidelines³⁵. In addition, we performed the following scenario analyses: (1) Alectinib PFS and OS data limited to patients with prior exposure to both platinum-based therapy and crizotinib; (2) Utility values in the PF health state are equal for both alectinib and ceritinib; and (3) ceritinib overall survival is assumed to be the same as alectinib.

Results

In the base case, treatment with alectinib vs. ceritinib resulted in an increase of 2.55 months in the PF state, an additional 0.72 life years, an additional 0.44 quality adjusted life-years (QALYs), and an increase of $13,868. This yielded a mean incremental cost per life-year gained of $19,313, and incremental cost per QALY gained of $31,180 (Table 4).

Table 4. Results for cost utility analysis.

	Pre-progression cost	Post-progression cost	Total cost	LYs	QALYs
Alectinib	$153,189	$102,224	$255,413	2.39	1.42
Ceritinib	$143,662	$97,883	$241,545	1.67	0.98
Difference	$9527	$4341	$13,868	0.72	0.44
ICER				$19,313	$31,180

Sensitivity analyses

A probabilistic sensitivity analysis was performed (5,000 simulations) for the base-case analysis, and the cost-effectiveness acceptability curve is shown in Figure 2. The PSA demonstrated that alectinib has a 96% probability of being cost-effective at a willingness-to-pay of $100,000/QALY. In one-way sensitivity analysis, the main model drivers were drug costs and utilities in the PF health state (see Figure 3).

Figure 2. Cost-effectiveness acceptability curve.

Figure 3. Tornado diagram of one-way sensitivity analysis.

In the scenario analysis which used alectinib data limited to patients with prior chemotherapy and prior crizotinib treatment vs ASCEND II data, the ICER was $10,644/QALY. In the scenario with equal utility values (0.79) for alectinib and ceritinib in the PF health state, the ICER was $34,758/QALY. Finally, in the scenario in which the overall survival for ceritinib was set equal to alectinib, the ICER was $64,562/QALY.

Discussion

There is an important unmet need for ALK + metastatic NSCLC patients that progress or are intolerant to treatment with crizotinib. Two agents have been approved in this setting; however, no direct comparisons exist in terms of their clinical and economic impact. We evaluated alectinib vs ceritinib for the treatment of ALK + advanced NSCLC patients to estimate the incremental QALYs and treatment costs from the US payer perspective. We estimated that alectinib is likely to increase time progression free, increase quality adjusted life years, and moderately increased costs. At an estimated cost per QALY of $31,000, alectinib may be considered a cost-effective option compared to ceritinib in the US.

Continued innovation in the treatment of advanced NSCLC is improving outcomes for patients with this serious condition. Crizotinib was the first treatment option available specifically for patients with ALK + tumors. Although this treatment option continues to improve outcomes for these patients compared to traditional chemotherapy, there are limitations with its use, including acquired resistance and the ability to penetrate the blood–brain barrier, which results in modest efficacy preventing or shrinking brain metastases³⁸. Both alectinib and ceritinib have demonstrated the ability to meaningfully impact central nervous system (CNS) tumors (Table 1), however only Alectinib has data on CNS efficacy in its FDA approved prescribing information^39–41. The ability to impact CNS tumors is another advance in the treatment of advanced ALK + NSCLC patients and may confer additional improvements to the length and quality of patients’ lives. Also in this context, these agents may have a role earlier in the treatment sequence—a hypothesis currently under investigation in ongoing clinical trials⁴². For example, recent data from the J-ALEX study found a progression-free hazard ratio of 0.34 (95% CI = 0.17–0.70, p < .0001) for alectinib vs crizotinib in Japanese ALK + treatment-naïve NSCLC patients⁴³. A larger global study is due to be completed in 2017.

There is limited existing data on the cost-effectiveness of alectinib or ceritinib. However, in one analysis reported at ISPOR 2015, Zhou et al.⁴⁴ found that ceritinib improved outcomes vs untargeted chemotherapy agents, pemetrexed and docetaxel. Importantly, these results are consistent with, and externally validate, the life expectancy and QALY estimates for ceritinib from our model (1.67 life years and 0.98 QALYs vs 1.77 life years and 0.94 QALYs, respectively). Another study evaluated the cost-effectiveness of certinib in Canada and used the same model structure and many of the same assumptions as we have herein. The authors did not, however, include alectinib, which is an appropriate comparator for ALK + patients after crizotinib. We note that the modeled life years and QALYs for ceritinib in their analysis are very similar to those in our model (life years = 1.61, QALYs = 0.86; life years = 1.67, QALYs = 0.98, respectively)⁴⁵. With the lack of cost-effectiveness data, our study provides the first clinical and economic comparison of alectinib and ceritinib.

There are several limitations to this model worth noting. First, there is no direct evidence comparing alectinib to ceritinib in this setting, and, therefore, we do not have information as to the relative clinical efficacy, safety, and resource utilization for the interventions in the same population. This issue was further complicated by the lack of Kaplan Meier data on overall survival for ceritinib from the ASCEND I trial. We, therefore, performed a naïve comparison and evaluated a number of different scenarios in our sensitivity analysis, including a sub-group analysis using the most comparable data to ASCEND 2 available from the alectinib trials, those with prior exposure to both crizotinib and chemotherapy. Across all the scenario analyses, the cost per QALY was below $100,000 per QALY. Our model results also align well with the 12-month overall survival rate provided in updated results from the ASCEND-1 trial (trial results: 67% vs modeled results: 65%)¹⁶. Ultimately, direct comparative trials are needed to determine if there are statistically significant or clinically meaningful differences in progression-free and overall survival between the treatment strategies. Another limitation was the absence of data on the actual costs of care in the progression health state. In the absence of this data, we used data from a widely cited retrospective cancer costing study that utilized SEER-Medicare claims data³⁴.

In conclusion, we estimated that treatment with alectinib in ALK + crizotinib-treated NSCLC patients increased time in the PF health state, increased life-years, and increased QALYs vs ceritinib. The marginal increase in costs was driven by longer treatment durations with alectinib. The results remained consistent across a number of scenarios. This model demonstrates that alectinib may be considered a cost-effective treatment after patients have progressed on crizotinib according to commonly used thresholds in the US (i.e. < $100,000 to $150,000/QALY).

Transparency

Declaration of funding

This work was funded by Genentech, Inc.

Declaration of financial/other relationships

JC has been a consultant for Genentech, Pfizer, and Seattle Genetics. WC has been a consultant for Astra Zeneca, MedImmune, and the National Pharmaceutical Council. AR is an employee of Genentech. WW is an employee of Genentech. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Appendix

Table A1. Healthcare resources and costs to treat adverse events^20,26–29.

Side effect	Treatment assumptions and cost of treatment	Total event cost
Anemia	1 outpatient physician office visit Cost of 1 outpatient physician visit = $208 50% of patients treated with 40K units of epoetin weekly for 8 weeks Cost of epoetin alfa 2K units = $32	$2762
Fatigue	1 outpatient physician office visit Cost of 1 outpatient physician visit = $208	$208
Infection (including pneumonia)	50% of patients have major complications DRG 178: Treatment of respiratory infection with complications = $7973 DRG 177: Cost of inpatient stay for respiratory stay with major complications = $$11,177	$9575
Nausea	1 outpatient physician office visit Cost of 1 outpatient physician visit = $208	$208
Diarrhea	1 outpatient physician office visit Cost of 1 outpatient physician visit = $208	$208
Vomiting	DRG 392: Esophagitis, gastroenteritis and miscellaneous digestive disorders without major complications or comorbidities	$4339
Hyperglycemia	1 outpatient physician office visit Cost of 1 outpatient physician visit = $208	$208
Lipase increase	1 outpatient physician office visit Cost of 1 outpatient physician visit = $208	$208
Blood alkaline increase	1 outpatient physician office visit Cost of 1 outpatient physician visit = $208	$208

Table A2. Parameter inputs.

Efficacy endpoint	Alectinib treatment arm	Ceritinib treatment arm
Progression free survival (Weibull)
Lambda	0.0673	0.0575
Gamma	1.0601	1.3087
Overall survival (Weibull)
Lambda	0.0192	0.0184
Gamma	1.1261	1.2804