Economic evaluation of rituximab in addition to standard of care chemotherapy for adult patients with acute lymphoblastic leukemia

Abstract

Aims: Acute lymphoblastic leukemia (ALL) is an aggressive form of leukemia with a poor prognosis in adult patients. The addition of the monoclonal antibody rituximab to standard chemotherapy has been shown to improve survival in adults with ALL. However, it is unknown whether the addition of rituximab is cost-effective. The objective was to determine the economic impact of rituximab in addition to standard of care (SOC) chemotherapy vs SOC alone in newly-diagnosed Philadelphia chromosome-negative, CD20-positive, B-cell precursor ALL.

Methods: A decision analytic model was constructed, based upon the Canadian healthcare system. It included the following health states over a lifetime horizon (max ≈60 years): event-free survival (EFS), relapsed/resistant disease, cure, and death. SOC was either hyper-CVAD or the Dana Farber Cancer Institute (DFCI) ALL consortium. EFS, overall survival, and serious adverse event (SAE) rates were derived from a large randomized controlled trial. Costs of the model included: first-line treatment and administration, disease management, second-line and third-line treatment and administration, palliative care, and SAE-related treatments. Inputs were sourced from provincial and national public data, the literature, and cancer agency input.

Results: Quality-adjusted life-years (QALYs) increased by 2.20 QALYs with rituximab in addition to SOC. The resulting mean Incremental Cost-Effectiveness Ratio (ICER) was C$21,828/QALY. At a willingness-to-pay threshold of C$100,000/QALY, the probability of being cost-effective was 98%. Decision outcomes were robust to the probabilistic and deterministic sensitivity analyses, including the SOC backbone as either hyper-CVAD or DFCI.

Limitations: The results of this analysis are limited by generalizability of the chemotherapy backbone to Canadian practice.

Conclusions: For adults with ALL, rituximab in addition to SOC was found to be a cost-effective intervention, compared to SOC alone. The addition of rituximab is associated with increased life years and increased QALYs at a reasonable incremental cost.

Keywords:

• Leukemia
• cost-effectiveness
• acute lymphoblastic leukemia
• rituximab

Introduction

Acute lymphoblastic leukemia (ALL) comprises 10% of newly-diagnosed leukemias¹. While only 40% of new ALL cases are diagnosed in adult patients, adults disproportionally account for 80% of ALL-related deaths². Current treatment options include prolonged multi-agent chemotherapy, with or without blood and marrow transplantation. For Philadelphia chromosome-negative (Ph-ve) patients, novel or targeted therapies have not been conventionally used as part of initial therapy. With standard of care (SOC) multi-agent chemotherapy, 80–90% of patients are expected to reach complete remission (CR), but roughly half will later experience relapse³.

The cell surface antigen CD20 is expressed in ∼40% of patients with Ph-ve B-cell ALL⁴. The efficacy of targeting CD20 was recently examined in the GRAALL-R 2005 study⁵. GRAALL-R 2005 was a phase III randomized open-label trial evaluating the efficacy of the addition of the CD20-targeted monoclonal antibody, rituximab, to the GRAALL chemotherapy protocol compared to the GRAALL regimen alone in chemotherapy-naïve adult CD20 + ve, Ph-ve, B-cell precursor (BCP) ALL. The trial demonstrated significant improvement in event-free survival (EFS) in the rituximab arm (hazard ratio (HR) = 0.66; 95% confidence interval [CI] = 0.45–0.98; p = .04), which was likely attributable to the lower incidence of relapse (HR = 0.52; 95% CI = 0.31–0.89; p = .02). The overall incidence of severe adverse events was not significantly different between the two treatment arms.

For health technology assessment countries like Canada, clinical evidence can be insufficient to obtain access for novel therapies. In Canada, both the health and cost consequences are assessed in a framework of health technology assessment. There are no economic evaluations for adult ALL in the frontline in a Canadian context, regardless of treatment management strategy. The current analysis is the first to examine the cost-effectiveness of adding rituximab to standard of care for CD20 + ve, Ph-ve BCP-ALL.

Methods

Model structure

A decision-analytic model was built in Microsoft Excel 2010 to represent the lifetime pathway of the patient. The model is a health state transition model using partitioned survival analysis to populate transition probabilities. Figure 1 presents the conceptual model. The “EFS state” resembles a complete remission state, whilst a relapsed/resistant disease state represents leukemia that has relapsed following or is resistant to treatment. All newly-diagnosed patients began in EFS state and progressed to relapsed/resistant, corresponding to the EFS curve. Following 60 months of EFS, the patient was considered cured, with evidence in childhood and adult ALL using a comparable definition^6–8. Death was possible from either the cured state or the relapsed/resistant state. A cycle length of 1 week was used, and a half-cycle correction was applied to any transition; however, costs and quality-adjusted life-years (QALYs) during first induction were applied at the beginning of the cycle.

Figure 1. Model structure. Patients start in an EFS state and move to relapse/resistant upon an event, and death upon a death event. After 60-months in the EFS state, patients move to a cured state.

Table 1 summarizes model parameter unit values. Table 2 summarizes drug acquisition costs. Table 3 summarizes non-drug resource use. Table 4 summarizes adverse event resource use. Table 5 aggregates all information and presents costs by treatment phase. The Appendix summarizes complete drug dosing for each regimen and statistical output of time-to-event modeling.

Table 1. Summary of model parameters of unit values.

Parameter	Unit	Mean	SE	Comments
Miscellaneous clinical
SOC: HyperCVAD proportion	Percentage	52%		CCMB expert opinion
Reinduction failure: RTX	Prob	0.08	0.03	GRAALL-R
Reinduction failure: SOC	Prob	0.10	0.03	GRAALL-R
Proportion of SCT of CR in RTX arm	Percentage	27%	6%	GRAALL-R weighted average
Proportion of SCT of CR in SOC arm	Percentage	27%	6%	GRAALL-R weighted average
SMR for OS: cure vs life tables	HR	8.2	0.10	Martin et al.^17
Relapse/resistant disease
Proportion of 2L novel agent, of relapsed/resistant	Prob	0.67		CCMB expert opinion
Proportion of 3L novel agent, of 2nd relapsed	Prob	0.25		CCMB expert opinion
Proportion of 3L palliative, of 2nd relapsed	Prob	0.25		CCMB expert opinion
Median time until 2nd relapse, following 1st relapse	Months	6	0.51	PETHEMA^11
1-year DFS, following 1st relapse	Percentage	34%	0.4%	PETHEMA^11
Prob of completing 2L induction	Prob	0.90	0	PETHEMA 6m DFS^11, converted to 4 weeks
Prob of completing 2L consolidation	Prob	0.73	0	PETHEMA 1Y DFS^11, converted to 39 weeks
Prob of completing 2L maintenance	Prob	0.25	0	PETHEMA 1Y DFS^11, converted to 174 weeks
Utilities
Complete remission		0.86	0.01	Complete remission. Aristides et al.^28
Complete remission w/partial hematological recovery		0.75	0.02	Complete remission w/partial hematological recovery. Aristides et al.^28
Remission health state		0.85	0.02	CCM expert opinion. 95% complete remission: 5% complete remission w/partial hematological recovery
Relapsed health state		0.30	0.04	Progressive disease. Aristides et al.^28
Cure health state		0.86	0.01	Assumed same as EFS
Chemotherapy disutility		0.18	0.02	Matza et al.^29
Costs
Chair time	$/min	2.97	0.59	Tam et al.^22 Inclusive of chair, nursing, pharmacy, prep time
Hospitalization	$/d	457	54	CAT indirect costs for ALL (C910) divided by avg LOS (22.8)
LMS fee	$/LMS unit	0.517		Zaric et al.^25
Stem cell transplant	$/procedure	68,868	3,887	CAT. Homologous SCT (1LZ19HHU7J)^26. Weighted avg of 2009–2011
Physician consult	$/visit	105.25		ON SOB A446. Repeat consultation^23
Physician fee for induction supervision	$/induction	262.40		ON SOB G390. Supervision of induction phase of acute leukemia^23
Physician fee for IV admin	$/admin	75.00		ON SOB G345. Complex single agent or multi-agent therapy^23
Physician fee for oral mgmt	$/admin	20.50		ON SOB G388. Management of special oral chemotherapy, for malignant disease^23
Physician fee for SCT	$/procedure	831.71		ON SOB Z425 + Z426 + G390^23
Chair time
Induction: hyperCVAD	min/visit	45.417		CCO regimen monograph HyperCVAD^10
Induction: hyperCVAD + RTX	min/visit	53.125		CCO regimen monograph HyperCVAD^10
Induction: Dana Farber	min/visit	52.806		CCO Nursing RIW/Workload, Intensification. Assumed same as nursing time^37
Induction: Dana Farber + RTX	min/visit	60.514		Assumed to increase by same amount as HyperCVAD
Maintenance: hyperCVAD	min/visit	31.222		Assumed to decrease by same amount as DF
Maintenance: hyperCVAD + RTX	min/visit	38.930		Assumed to increase by same amount as HyperCVAD
Maintenance: Dana Farber	min/visit	38.611		CCO Nursing RIW/Workload, Intensification. Assumed same as nursing time^7^,^37
Maintenance: Dana Farber + RTX	min/visit	46.319		Assumed to increase by same amount as HyperCVAD
2L palliative	min/visit	45.417		CCM expert opinion. Palliative care requires IV vincristine. Assumed same chair as HyperCVAD
Hospitalization
Induction: all backbones	d/cycle	28		CCM expert opinion
Consolidation: hyperCVAD cycle 2A, 3A, 4A	d/cycle	4		CCM expert opinion
Consolidation: hyperCVAD cycle 1B, 2B, 3B, 4B	d/cycle	5		CCM expert opinion
Intensification: all backbones	d/cycle	0		CCM expert opinion
Maintenance: all backbones	d/cycle	0		CCM expert opinion
Physician visits & cost
Induction: all backbones	visits/cycle	28		CCM expert opinion
Other phases: all backbones	visits/cycle	1		CCM expert opinion
Chest X-ray	$/view	14.90		ON SOB. Chest single view. X090^23
Specialist consult	$/consult	157.00		ON SOB. Infectious disease consult A465. Or hematology consult. A615^23
Spinal CT	$/scan	97.50		ON SOB. Spinal CT (X416)^23
ECG	$/admin	11.05		ON SOB. ECG: technical and professional component. G310 + G313^23
Cardioversion	$/procedure	92.45		ON SOB. Cardioversion. Z437^23
Procedures
Stem cell transplant	$/procedure	68,868	3,887	CAT. Homologous SCT (1LZ19HHU7J). Weighted avg of 2009–2011^26
Cardioversion	$/procedure	14,663	2,487	CAT. Stimulat heart cardiovert/defib. 1HZ09JAFS^26
SAE
Pulmonary embolism	$/event	8,731	346	CAT. Pulmonary embolism. I269^26
Arrhythmia	$/event	7,144	174	CAT. Atrial fibrillation I480^26
Constipation	$/event	3,591	150	CAT. Constipation. K59^26
Pulmonary event: % embolism	Percentage	20%		CCMB expert opinion. Pulmonary event on average = embolism or pneumonia
GI event: pancreatitis	Percentage	50%		Assumption

CCMB, CancerCare Manitoba; CCO, CancerCare Ontario; CAT, cost analysis tool; ON, Ontario; RIW, resource intensity weight; SOB, Schedule of Benefits; SCT, stem-cell transplant; STFM, systemic treatment funding model; IV, intravenous.

Table 2. Treatment Rx costs. Package size, unit cost, and total drug cost per cycle of treatment phase (under no vial sharing assumption).

Rituximab phases	Product	Package size	Quantity of drug	Package price	# Packages	Total cost
			per package			per cycle
Induction	Rituximab	10 mg/mL in 50 mL vial	500 mg	$2,331.61	2.00	$6,548.86
Reinduction		10 mg/mL in 20 mL vial	200 mg	$932.64	3.00	$6,548.86
Consolidation						$6,548.86
Late Intensification						$6,548.86
First-Year Maintenance						$6,548.86
Consolidation A Cycles 1, 3, 5, 7 (28 day cycle)	Cyclophosphamide	2000 mg vial	2,000 mg	$152.35	2.00	$253.41
	Mesna	100 mg/mL in 150 mL vial	15,000 mg	$599.15	1.00	$132.88
	Dexamethasone	10 mg/mL in 10 mL vial	100 mg	$12.83	32.00	$410.56
	Doxorubicin	2 mg/mL in 100 mL vial	200 mg	$770.00	1.00	$355.77
4 cycles	Vincristine	1 mg/mL in 1 mL vial	1 mg	$30.60	4.00	$122.40
Total						$1,275.01
Consolidation B Cycles 2, 4, 6, 8 (28 day cycle)	Methotrexate	25 mg/mL in 40 mL vial	1,000 mg	$225.00	2.00	$415.83
	Cytarabine	100 mg/mL in 10 mL vial	1,000 mg	$153.25	23.00	$3,398.74
	Methylprednisolone	500 mg in 5 vials	2,500 mg	$93.00	1.00	$11.16
4 cycles	Leucovorin	10 mg/mL in 50 mL vial	500 mg	$350.19	1.00	$108.56
Total						$3,934.29
POMP Maintenance Months 1–5, 8–17, 20–30	6-Mercaptopurine	50 mg tablets 60 tablets per pack	3,000 mg	$171.66	1.40	$240.32
	Methotrexate	2.5 mg tablets 100 tablets per pack	250 mg	$63.25	0.32	$20.24
	Vincristine	1 mg/mL in 1 mL vial	1 mg	$30.60	2.00	$61.20
26 months	Prednisone	50 mg tablets 100 tablets per pack	5,000 mg	$17.35	0.20	$3.47
Total						$325.23
Intensification A Months 6, 18	Cyclophosphamide	2000 mg vial	2,000 mg	$152.35	2.00	$253.41
	Vincristine	1 mg/mL in 1 mL vial	1 mg	$30.60	4.00	$122.40
	Doxorubicin	2 mg/mL in 100 mL vial	200 mg	$770.00	1.00	$355.77
2 months	Dexamethasone	10 mg/mL in 10 mL vial	100 mg	$12.83	3.20	$41.06
Total						$772.63
Intensification B Months 7, 19	Methotrexate	2.5 mg tablets 100 tablets per pack	250 mg	$63.25	2.96	$187.03
2 months	L-asparginase	10,000 IU	10,000 IU	$134.79	8.00	$1,078.32
Total						$1,265.35
Dana Farber Phases		Vial size
DFCI (Induction) (1 cycle) (28 day cycle)	Vincristine	1 mg/mL in 1 mL vial	1 mg	$30.60	8.00	$244.80
	Prednisone	50 mg tablets 100 tablets per pack	5,000 mg	$17.35	1.00	$7.18
	Doxorubicin	2 mg/mL in 100 mL vial	200 mg	$770.00	1.00	$426.92
	Methotrexate	25 mg/mL in 40 mL vial	1,000 mg	$225.00	8.00	$1,663.33
	Leucovorin	10 mg/mL in 50 mL vial	500 mg	$350.19	1.00	$258.88
		10 mg/mL in 50 mL vial	500 mg	$350.19	1.00	$62.13
	L-asparginase	10,000 IU	10,000 IU	$134.79	5.00	$622.78
	Methotrexate	25 mg/mL in 40 mL vial	1,000 mg	$225.00	1.00	$8.10
	Cytarabine	100 mg/mL in 10 mL vial	1,000 mg	$153.25	1.00	$18.39
	Allopurinol	200 mg tablets 100 tablets per pack	20,000 mg	$13.00	0.42	$5.46
	Cotrimoxazole (Septra)	1 tab	6 tablets	$0.12	4.00	$0.48
Total						$3,318.45
DFCI (Intensification) (3-week cycle × 10 cycles)	Vincristine	1 mg/mL in 1 mL vial	1 mg	$30.60	20.00	$612.00
	Dexamethasone	10 mg/mL in 10 mL vial	100 mg	$12.83	17.00	$213.41
	6-Mercaptopurine	50 mg tablets 60 tablets per pack	3,000 mg	$171.66	4.67	$801.08
	Doxorubicin	2 mg/mL in 100 mL vial	200 mg	$770.00	2.00	$1,155.00
	Methotrexate	25 mg/mL in 40 mL vial	1,000 mg	$225.00	2.00	$374.25
	L-asparginase	10,000 IU	10,000 IU	$134.79	70.00	$9,341.67
10 cycles	Methotrexate	25 mg/mL in 40 mL vial	1,000 mg	$225.00	1.00	$4.50
	Cytarabine	100 mg/mL in 10 mL vial	1,000 mg	$153.25	1.00	$10.22
Total						$12,512.12
DFCI (Continuation) (3-week cycle until 2 years after CR) (74 weeks per DJ DeAngelo et al; 3-week cycle ×24.67 cycles)	Vincristine	1 mg/mL in 1 mL vial	1 mg	$30.60	2.00	$61.20
	Dexamethasone	10 mg/mL in 10 mL vial	100 mg	$12.83	1.00	$7.11
	6-Mercaptopurine	50 mg tablets 60 tablets per pack	3,000 mg	$171.66	0.46	$79.31
	Methotrexate	25 mg/mL in 40 mL vial	1,000 mg	$225.00	1.00	$37.42
25 cycles	Methotrexate	25 mg/mL in 40 mL vial	1,000 mg	$225.00	1.00	$0.45
	Cytarabine	100 mg/mL in 10 mL vial	1,000 mg	$153.25	1.00	$1.02
Total						$186.52
Palliative Care (until death)	Vincristine	1 mg/mL in 1 mL vial	1 mg	$30.60	2.00	$61.20
	Prednisone	50 mg tablets 100 tablets per pack	5,000 mg	$17.35	1.00	$3.47
	6-Mercaptopurine	50 mg tablets 60 tablets per pack	3,000 mg	$171.66	0.35	$60.08
1 cycle = 3 weeks	Methotrexate	2.5 mg tablets 100 tablets per pack	250 mg	$63.25	0.24	$15.18
						$139.93
SAE-related drugs
Infection SAE	Vancomycin	250 mg cap	0.25 g	$11.25	64.00	$720.00
	Meropenem	500 mg per vial	0.50 g	$26.35	64.00	$1,686.40
Total						$2,406.40
Allergic SAE	Benadryl	50 mg/mL × 10 mL	500 mg	$40.40	0.10	$4.04
	Hydrocortisone	1000 mg/vial	1,000 mg	$8.60	0.25	$2.15
	Epinephrine	0.3 mg autoinjector pen	0 mg	$83.35	1.00	$83.35
	Salbutamol	2.5 mg	3 mg	$0.45	1.00	$0.45
Total						$89.99
Pulmonary SAE	Levofloxacin	500 mg × 100 tabs	50,000 mg	$350.70	0.07	$24.55
	Dalteparin	25,000 IU per unit	25,000 IU	$159.00	0.08	$12.04

Table 3. Treatment-related resource use by cycle of treatment phase.

Phase	Resource use	Cost	Notes
Rituximab phases
Induction/Consolidation/Intensification (w/Hyper-CVAD)	2 infusions @ 53 min per infusion 2 IV admin fees (All other costs part of backbone)	C$497	CCO hyper-CVAD^10 ON SOB IV admin (G345)^23
Maintenance (w/Hyper-CVAD)	2 infusions @ 39 min per infusion 2 IV admin fees (All other costs part of backbone)	C$404	Hyper-CVAD maintenance + same difference between RTX and hyper-CVAD of 7.7 min ON SOB IV admin (G345)^23
Induction/Consolidation/Intensification (w/DF)	2 infusions @ 61 min per infusion 2 IV admin fees (All other costs part of backbone)	C$546	DF induction + same difference between RTX and hyper-CVAD of 7.7 min ON SOB IV admin (G345)^23
Maintenance (w/DF)	2 infusions @ 46 min per infusion 2 IV admin fees (All other costs part of backbone)	C$453	CCO STFM RIW, DF Continuation^37. Nurse intensity time + same difference between RTX and hyper-CVAD of 7.7 min ON SOB IV admin (G345)^23
Hyper-CVAD phases
Induction A1	4 infusions @ 45 min per infusion 1 induction admin fee 4 IV admin fees 28 days inpatient LOS 4 physician consults 28 CBC, 28 INR, 28 fibrinogen, 28 lipase	C$17,184	CCO hyper-CVAD^10 ON SOB IV admin (G345), supervision of induction (G390), consultation (A446)^23 ON SOB CBC (L393), INR (L445), fibrinogen (L402), lipase (L155)^24 CCMB expert opinion
Consolidation A2–4	4 infusions @ 45 min per infusion 4 IV admin fees 4 days inpatient LOS 1 physician consult 4 CBC, 1 biochem, 3 INR, 3 fibrinogen, 3 lipase	C$3,261	CCO hyper-CVAD^10 ON SOB IV admin (G345), consultation (A446)^23 ON SOB CBC (L393), INR (L445), fibrinogen (L402), lipase (L155), biochemistry (86 LMS)*^,^24 CCMB expert opinion
Consolidation B	3 infusions @ 45 min per infusion 3 IV admin fees 5 days inpatient LOS 1 physician consult 4 CBC, 1 biochem, 3 INR, 3 fibrinogen, 3 lipase	C$3,529	CCO hyper-CVAD^10 ON SOB IV admin (G345), consultation (A446)^23 ON SOB CBC (L393), INR (L445), fibrinogen (L402), lipase (L155), biochemistry (86 LMS)*^,^24 CCMB expert opinion
POMP Maintenance	1 infusion @ 31 min 1 IV admin fee 1 oral admin fee 0 days inpatient LOS 4 CBC, 1 biochem, 3 INR, 3 fibrinogen, 3 lipase	C$597	CCO STFM RIW, DF Continuation^37. Nurse intensity time—difference btw DF and hyper-CVAD induction times of 14 min ON SOB IV admin (G345), oral chemo (G388), consultation (A446)^23 ON SOB CBC (L393), INR (L445), fibrinogen (L402), lipase (L155), biochemistry (86 LMS)*^,^24 CCMB expert opinion
Intensification A	4 infusions @ 31 min per infusion 4 IV admin fees 0 days inpatient LOS 1 physician consult 4 CBC, 1 biochem, 3 INR, 3 fibrinogen, 3 lipase	C$1,294	CCO STFM RIW, DF Continuation^37. Nurse intensity time—difference btw DF and hyper-CVAD induction times of 14 min ON SOB IV admin (G345), consultation (A446)^23 ON SOB CBC (L393), INR (L445), fibrinogen (L402), lipase (L155), biochemistry (86 LMS)*^,^24 CCMB expert opinion
Intensification B	4 infusions @ 31 min per infusion 4 IV admin fees 1 oral admin fee 0 days inpatient LOS 1 physician consult 4 CBC, 1 biochem, 3 INR, 3 fibrinogen, 3 lipase	C$1,314	CCO STFM RIW, DF Continuation^37. Nurse intensity time—difference btw DF and hyper-CVAD induction times of 14 min. ON SOB IV admin (G345), oral chemo (G388), consultation (A446)^23 ON SOB CBC (L393), INR (L445), fibrinogen (L402), lipase (L155), biochemistry (86 LMS)*^,^24 CCMB expert opinion
Dana Farber phases
Induction	5 infusions @ 53 min per infusion 4 IV admin fees 1 induction admin fee 28 days inpatient LOS 4 physician consults 28 CBC, 28 INR, 28 fibrinogen, 28 lipase	C$17,528	CCO STFM RIW, DF Intensification. Nurse intensity time^37 ON SOB IV admin (G345), supervision of induction (G390), consultation (A446)^23 ON SOB CBC (L393), INR (L445), fibrinogen (L402), lipase (L155)^24 CCMB expert opinion
Intensification	1 infusion @ 53 min 1 IV admin fee 1 oral admin fee 0 days inpatient LOS 1 physician consult 4 CBC, 1 biochem, 3 INR, 3 fibrinogen, 3 lipase	C$668	CCO STFM RIW, DF Intensification. Nurse intensity time^37 ON SOB IV admin (G345), oral chemo (G388), consultation (A446)^23 ON SOB CBC (L393), INR (L445), fibrinogen (L402), lipase (L155), biochemistry (86 LMS)*^,^24 CCMB expert opinion
Continuation	2 infusions @ 39 min per infusion 1 IV admin fee 1 oral admin fee 0 days inpatient LOS 1 physician consult 4 CBC, 1 biochem, 3 INR, 3 fibrinogen, 3 lipase	C$823	CCO STFM RIW, DF Continuation. Nurse intensity time^37 ON SOB IV admin (G345), oral chemo (G388), consultation (A446)^23 ON SOB CBC (L393), INR (L445), fibrinogen (L402), lipase (L155), biochemistry (86 LMS)*^,^24 CCMB expert opinion
Palliative care	1 infusion @ 45 mins 1 IV admin fee 1 physician consult	C$329	CCO hyper-CVAD^10 ON SOB IV admin (G345), consultation (A446)^23

* Biochemistry = blood urea nitrogen, serum creatinine, AST, ALT, bilirubin, GGT, alkaline phosphatase, magnesium, calcium, albumin, potassium, sodium, phosphate.

CCMB, CancerCare Manitoba; CCO, CancerCare Ontario; CBC, complete blood count; DF, Dana Farber; INR; international normalized ratio; LOS, length of stay; ON, Ontario; RIW, resource intensity weight; SOB, Schedule of Benefits; STFM, systemic treatment funding model; IV, intravenous.

Table 4. Resource utilization and cost per SAE.

	Resource use	Cost	Notes
Infection	Vancomycin 1g IV q12h × 8d, Meropenem 1g IV q6h × 8d, Specialist consult X-ray 1 urine analysis, 3 blood cultures	C$2,662	Already in hospital. ON SOB specialist consult (A465), single chest view (X090)^23 ON SOB urine (L079), blood cultures (L624)^24
Allergy	Benadryl 50 mg IV × 1 Hydrocortisone 250 mg IV × 1 Epinephrine 0.3 mg subcut × 1 Salbutamol 2.5 mg via nebulizer Specialist consult	C$195	Already in hospital. ON SOB specialist consult (A465)^23
Pulmonary embolism	Total direct and indirect cost	C$9,399	Could happen anytime. CAT pulmonary embolism (I269)
Pneumonia	Levofloxacin 500 mg PO per day × 7 days Specialist consult X-ray	C$196	Already in hospital. ON SOB specialist consult (A465), single chest view (X090)^23
Arrhythmia	Total direct and indirect cost	C$7,690	Could happen anytime. CAT atrial fibrillation (I480)
Constipation	Total direct and indirect cost	C$3,866	Could happen anytime. CAT constipation (K59)
Pancreatitis	Specialist consult X-ray Lipase	C$223	Already in hospital. ON SOB specialist consult (A465), single chest view (X090)^23 ON SOB lipase (L155)^24

PO, orally; CAT, Ontario Case Costing Initiative; ON, Ontario; q6h, every 6 hours; q12h, every 12 hours; SOB, Schedule of Benefits.

Table 5. Treatment costs by cycle of treatment phase.

	Infusion visits	Costs (C$)
		Drug^21	Admin^22^,^23	Monitoring^24^,^25	Hospitalization^38	Total
Rituximab phases
Induction: (w/Hyper-CVAD)	2	$6,549	$497	NA	NA	$7,046
Induction: (w/DF)	2	$6,549	$546	NA	NA	$7,094
Maintenance: (w/Hyper-CVAD)	2	$6,549	$404	NA	NA	$6,953
Maintenance: (w/DF)	2	$6,549	$453	NA	NA	$7,002
Hyper-CVAD phases
Induction (A1)	4	$1,275	$1,156	$2,258	$13,770	$18,459
Consolidation A (2,3,4)	4	$1,275	$894	$400	$1,967	$4,536
Consolidation B	3	$3,934	$670	$400	$2,459	$7,463
POMP maintenance	1	$325	$198	$400	$0	$923
Intensification A	4	$773	$894	$400	$0	$2,066
Intensification B	4	$1,265	$914	$400	$0	$2,580
DFCI phases
Induction	5	$3,318	$1,500	$2,258	$13,770	$20,847
Intensification	1	$12,512	$268	$400	$0	$13,180
Continuation	2	$187	$423	$400	$0	$1,009
3L blinatumomab	28	$39,601	$7,666	$2,258	$13,770	$63,295
Palliative care	1	$140	$223	$105	NA	$469

C$, Canadian dollars.

Treatment strategies

In Canada, we used CancerCare Manitoba (CCMB) as the data source from which we derived relevant epidemiological and therapeutic data for this analysis. CCMB is responsible for setting provincial guidelines, standards, and recording outcomes for all patients with malignant neoplasms in the province of Manitoba, Canada⁹.

Our analysis compares rituximab in combination with SOC to SOC alone for adults (aged 18 years and over). In Canada, this patient population commonly receives either “hyper-CVAD”¹⁰ or the Dana Farber Cancer Institute (DFCI) ALL consortium regimen for adults¹¹. Both regimens have a number of phases—induction, consolidation, maintenance and/or intensification—as well as a number of drugs for each phase. A comprehensive description of each regimen used in the model is detailed in the Appendix. Patients received 16–18 infusions of rituximab across the consolidation, intensification, and maintenance phases of hyper-CVAD or DFCI chemotherapy backbones.

Of the target population under the age of 50 years, we estimated that 80% would receive DFCI chemotherapy, whilst 20% would receive hyper-CVAD. For patients over 50, 66% (2/3) would receive hyper-CVAD, 20% DFCI therapy, and the remaining 14% a mix of other treatments.

Using data from Cancer Research UK for age distribution¹², and CANSIM sex distribution of ALL¹, we determined that the split of ages for both sexes to determine a base case of 52:48 (Male: Female) mix of hyper-CVAD and DFCI.

Clinical effectiveness

We assumed that treatment effects of GRAALL-R were generalizable to Canadian practice settings, as the drugs found in the DFCI, GRAALL, or hyper-CVAD protocols are very similar¹³. We also assumed that the incremental effects of adding rituximab to either hyper-CVAD or DFCI regimens are well represented by the GRAALL-R trial⁵.

Perspective, time horizon, discounting

Our model took the perspective of the Canadian publicly funded healthcare payer. The present analysis used a lifetime horizon (max ∼60 years) in light of long survival in a considerable proportion of patients⁵. An annual discount rate of 1.5% was used for both costs and effects¹⁴.

EFS

The EFS state was modeled using EFS data from the GRAALL-R trial⁵. Parametric models were applied to both SOC and SOC plus rituximab. Kaplan-Meier (KM) curves were digitized using DigitizeIt, and the individual patient data (IPD) was reconstructed using Guyot et al.’s¹⁵ algorithm. Parametric functions were fit using the flexsurv package in R 3.3.1. Goodness-of-fit was assessed using the Akaike Information Criterion (AIC) and the log-likelihood. Given a cure threshold time of 60 months, the GRAALL-R EFS curves may be considered complete. Thus, the best fitting distribution was selected by goodness-of-fit because no extrapolation was required.

After testing proportionality (log-transformed residuals vs time), it was determined that the proportional hazards assumption held. The fit statistics and statistical output of parametric models to EFS are presented in the Appendix. Curves are presented in Figure 2(a). Sensitivity analyses tested all models.

Figure 2. Kaplan-Meier and parametric model curves: (a) event-free survival; (b) overall-survival.

The GRAALL-R EFS curves were only necessary until the cure threshold time. The cure threshold was set at 60 months^6–8. Patients in the cure state exited only due to a mortality event. Mortality hazards were derived using Canadian Life Tables¹⁶, and further adjusted using a standardized mortality ratio (SMR) to reflect a poorer prognosis compared to the general population. A SMR of 8.2 was derived from a study in patients who survived more than 5-years without recurrence of original disease after allogeneic hematopoietic cell transplant (HCT)¹⁷.

Relapsed/resistant

Relapsed or resistant disease represents eponymous events and the ensuing progressive nature of disease. Transitions to the relapsed/resistant and death disease states were governed by EFS and OS curves, respectively. Model fitting of the GRAALL-R OS data was conducted using the same method described for EFS. Because extrapolation was required, the best model was selected using goodness-of-fit statistics and external validity, as judged by CCMB. After testing proportionality (log-transformed residuals vs time), it was determined that the proportional hazards assumption did not hold. RTX and SOC were modeled independently without proportionality. The fit statistics and statistical output for each distribution for each trial arm are shown in the Appendix. Although the Gompertz seemed to fit the data, it fit it too well; survival reached an asymptote equal to the last death event in the Kaplan-Meier. After judging external plausibility, the best model for the relevant time horizon was the log-logistic model. Curves are presented in Figure 2(b). Sensitivity analyses tested all models.

Newly relapsed

Newly relapsed was captured at the time of the event to account for subsequent treatment costs and health-related quality-of-life (HRQoL) changes; however, newly relapsed was not an actual health state. Costs in newly relapsed depend on the eligibility/preference and duration of response. The present analysis estimates that the proportion of patients who receive second-line treatment would be 67%, whilst the remaining 33% would receive palliative care only. Second-line treatment would generally require a chemotherapy backbone (such as FLAG [fludarabine, high dose cytarabine, GCSF]), which would require similar administration/hospitalization as hyper-CVAD.

The present model estimates the probabilities of being disease-free at the end of induction, consolidation, and first-year maintenance using a combination of published evidence and assumptions. Median time until 2nd relapse was 6 months, and 1-year disease-free survival was 34%¹⁸. Of those who fail second-line treatment, we assume that 25% would go on to receive third-line treatment, whilst 75% would receive palliative care only. Figure 3 summarizes the aforementioned literature into a treatment pathway that was reflected in the present model.

Figure 3. Treatment pathway for relapsed/resistant disease. Reported probabilities are conditional upon the higher level event.

Hematopoietic cell transplantation (HCT)

As use is not expected to differ on the basis of the receipt of rituximab, especially if complete response rates after induction/reinduction are no different, the present study used the overall average HCT rate of GRAALL (27%).

Serious adverse events (SAEs)

Serious adverse events were based on the GRAALL-R trial¹⁹. The model only used the SAEs, for which there was a measurable difference between the two treatment arms: infection, allergy, cardiac, pulmonary, and gastrointestinal (GI) events.

Treatment duration and dosing

The treatment duration was modeled using a time to treatment discontinuation (TTD) curve. For the SOC backbone in both arms, the model assumed 95% of patients who were event-free at the completion of any phase (induction, consolidation, intensification, and/or maintenance) were healthy to continue to the next phase, based on CancerCare Manitoba opinion.

Rituximab discontinuation was reported in the GRAALL-R trial⁵, but were only used in a sensitivity analysis for the present model, as a conservative assumption.

For both arms, discontinuation due to HCT was modeled at week 40, just after the completion of consolidation. The Appendix presents the TTD curves.

Drug costs

Table 2 summarizes all drug acquisition costs for treatment. The model implemented 100% dose intensity, assuming no reductions in dose. The base case assumed 100% vial sharing for IV drugs and, for rituximab only, wastage of 1.3% per vial was assumed²⁰. Package sizes and prices were taken from QunitilesIMS’ Delta PA²¹. Rituximab costs $2,331.61 for a 50 mL vial of 10 mg/mL (500 mg of drug). The total drug cost per cycle of treatment is summarized in Table 1.

Non-drug medical costs

Table 3 summarizes non-drug resource use. Administration costs included chair, nurse, pharmacy, and preparation times, as well as physician billing-related to drug administration. The unit cost of chair time was taken from the published literature²², and physician fees were taken from the Ontario Schedule of Benefits for Physician Services²³.

Monitoring costs included physician assessments and laboratory testing. We selected commonly ordered laboratory tests for each cycle of treatment phase. Laboratory test billing was taken from the Ontario Schedule of Laboratory Services^24,25.

Hospitalization costs included the inpatient length of stay (LOS) indirect costs. Direct medical costs were captured mostly by the administration and monitoring costs; thus, only the “hotel” costs of inpatient stay were required. LOS was estimated by the CCMB tumor group representatives (MD, KP). The daily cost of inpatient stay was estimated to be C$491.78, taken from the Ontario Cost Analysis Tool (CAT)²⁶.

For post-treatment supportive care and supportive care for those who discontinued treatment, a supportive care cost of physician visits every 3 weeks was applied until an event occurred. The cost of HCT of $68,868 was taken as the total direct and indirect costs of inpatient care from CAT²⁶.

Subsequent treatments

For second-line-treatment, the model used the costs of hyper-CVAD induction, consolidation, and 1st-year maintenance. The costs of hyper-CVAD induction were the same as described above. For third-line induction, the model assumed the costs of blinatumomab^8,27, whose induction costs are displayed in Table 1.

The probabilities of palliative treatment were derived from the failures or discontinuation at each phase of treatment. For palliative care costs, the model used a treatment regimen informed by CCMB tumor group expert opinion (Appendix). The drug costs were C$139.93 every 3 weeks. Patients would also incur the same per visit chair time as hyper-CVAD; an IV administration fee; and a physician consult every 3 weeks.

SAE costs

Table 4 presents the SAE costs. SAEs reported by GRAALL-R which were not specific included: pulmonary event, cardiologic event, and GI event. CCMB expert opinion expected, on average, a pulmonary event to be 80%;20% pneumonia:pulmonary embolism; a cardiologic event to be arrhythmia; and a GI event to be 50%:50% constipation:pancreatitis.

SAEs were micro-costed by drug, monitoring, and lab test costs if the event was likely to occur during an induction stay and/or not likely to increase inpatient stay, as the patient was already in-hospital and “hotel” costs would not increase. If, however, the SAE was not thought to be associated with an induction, CAT was consulted for the total direct and indirect costs associated with the respective diagnosis.

For SAEs likely to occur during an induction stay (infection, allergy, pneumonia, and pancreatitis), total costs were micro-costed. All other SAEs (pulmonary embolism, arrhythmia, and constipation) were derived from the CAT²⁶.

Utilities

Table 1 summarizes the health state utilities used in the model. Health state utilities were derived from an Aristides et al.²⁸ study of population preference values for health states in adults with relapsed or refractory BCP-ALL in the UK. According to CCMB tumor group expert opinion, the present model states of EFS would relate to the Aristides et al. study of a 95%:5% mix of CR:CR with partial hematological recovery, whilst relapsed/resistant related to progressive disease. The cure state was estimated to be no different to the EFS state.

Because patients may achieve second CR following second-line-treatment, the model attributed a utility of EFS for the duration of time in second CR. This was implemented using the pathway probabilities previously described.

All patients were given a disutility of 0.175 for chemotherapy²⁹, as long as they remained on treatment in the EFS state.

Sensitivity

A probabilistic sensitivity analysis (PSA) was used to model joint uncertainty. Gamma distributions were used for cost data, beta distributions for utility data, and normal distributions for [log-transformed] ratios. All variables in Table 1 with standard error values were probabilistically sampled in addition to the entire EFS and OS distribution, which subsequently affected the treatment discontinuation as well. A Monte Carlo simulation used 2,000 iterations to generate the probabilistic results.

Scenario analyses were also conducted for structural parameters whose value depended on external validity. Scenarios were run for alternative distributions of EFS and OS. As well, the duration of treatment effect was cut at 48 months (by setting probabilities of RTX + SOC equal to SOC). Discontinuation and compliance were both explored. SOC was set at hyperCVAD and DFCI. Stem-cell transplant rates were allowed to differ based on GRAALL-R. Also, high and low values were explored for the cure threshold, utilities, and several unit costs. Finally, shorter time horizon was also explored.

Results

Rituximab added to SOC led to both greater life-years (LYs) and QALYs in the EFS and Cure states, compared to the SOC alone. Greater mean times in these aforementioned states meant that rituximab added to SOC led to lesser relapsed/resistant time. Rituximab added to SOC led to 2.63 greater LYs and 2.20 greater QALYs compared to the SOC alone (Table 6).

Table 6. Cost-effectiveness analysis.

	Total RTX + SOC	Total SOC	Incremental
Effectiveness results
LYs
EFS	3.19	2.61	0.58
Cure	8.07	6.02	2.05
Relapse/resistant	1.49	1.49	0.00
Total/incremental LYs	12.75	10.12	2.63
QALYs
EFS	2.50	2.04	0.46
Cure	6.94	5.18	1.76
Relapse/resistant	0.54	0.55	–0.01
Total/incremental QALYs	9.97	7.77	2.20
Cost results (C$)
EFS
Drug: RTX	34,911	0	34,911
Drug: SOC	59,045	55,048	3,997
Administration	39,659	34,168	5,491
Re-induction	2,044	1,846	198
SAE	2,032	1,828	204
HCT	20,265	20,265	0
Supportive care	2,465	2,085	380
Mean cost of EFS	160,422	115,240	44,802
Cure
Mean cost of cure	14,768	11,024	3,744
Relapsed/resistant
Palliative	10,157	10,188	–31
2L Treatment	3,951	4,539	–588
3L Treatment	1,339	1,539	–199
Mean cost of progression	15,447	16,266	–818
Total/incremental costs	190,637	142,529	48,108

C$, Canadian dollars; EFS, event-free survival; LY, life year; QALY, quality-adjusted life year; RTX, rituximab; SOC, standard of care; 2L, second-line; 3L, third-line.

Rituximab added to SOC led to greater costs in the EFS state vs SOC alone (+$44,802), driven by increased drug and administration usage. Re-induction and supportive care had a negligible impact on incremental costs during the EFS state (+$198). Rituximab added to SOC led to fewer costs in the relapsed/resistant state vs SOC alone (–$818), driven by palliative care and second-line treatment savings. Total costs were higher with rituximab added to SOC vs SOC alone ($190,637 vs $142,529; difference = $48,108).

Rituximab added to SOC led to an incremental cost-effectiveness ratio (ICER) of C$18,327/LY and C$21,828/QALY.

Figures 4 and 5 present the cost-effectiveness plane and cost-acceptability curves. When joint uncertainty was modeled, rituximab added to SOC led to an ICER of C$18,843/LY and C$22,239/QALY. Rituximab added to SOC was estimated to be cost-effective, with a probability of 98% at a WTP of C$100,000/QALY (Figure 5). The ICERs of the scenario analyses are presented in the Appendix. The time horizon and select EFS distribution choices had a large impact on the ICER. A time horizon of 10 years increased the ICER to $46,936/QALY, and a Gamma distribution on EFS increased the ICER to $40,175/QALY. Limiting the duration of treatment effect to only 48 months had negligible impact ($22,405/QALY). Decision results were robust to most of the scenario analyses.

Figure 4. Cost-effectiveness plane of the probabilistic sensitivity analysis. C$, Canadian dollars; QALYs, quality-adjusted life years; RTX, rituximab; SOC, standard of care.

Figure 5. Cost-effectiveness acceptability curves.

Discussion

The objective of the present analysis was to conduct an economic evaluation of rituximab in addition to standard of care for the treatment of newly-diagnosed, untreated BCP-ALL in adults who are Ph-ve and CD20 + ve, compared to standard of care alone. To our knowledge, this is the first economic evaluation of any treatment in newly diagnosed, untreated ALL in adults in Canada and elsewhere.

The incremental health benefits of rituximab added to SOC are driven by the greater time in CR and the lower time in the relapsed/resistant state. This led to greater life years. The difference in LYs is further distinguished in a HRQoL difference between complete remission and progressive disease. CR entails no signs of ALL, the chance to return to work, and the normalization of social activities, among other characteristics²⁸. These contrast greatly with progressive disease, where outlook is negative, signs and symptoms are worsening, the risk of serious life-threatening infections causes a reduction in social activity, and depression is common²⁸. Rituximab added to SOC led to greater EFS and lower relapsed/resistant disease, which drove incremental LYs and QALYs. Rituximab + SOC had a relatively comparable safety profile to SOC alone, with the exception of infection. However, an increased rate of infection did not lead to excess mortality.

Rituximab added to SOC led to C$48,108 incremental costs over SOC alone. This was primarily due to the additional EFS cost of C$44,802, dominated by medication and administration costs. However, the increased EFS time associated with rituximab + SOC also corresponded with an increased proportion of patients who reached and remained in the cure state. This life extension led to greater mean costs of C$3,744 in the cure state. The greater EFS also corresponded with reduced relapsed/resistant time. This reduction of palliative, second-, and third-line treatment costs amounting to C$818, modestly offsetting the increased costs accrued in longer EFS.

Rituximab added to SOC demonstrated cost-effectiveness across a range of sensitivity analyses. When joint uncertainty was modeled, rituximab + SOC showed a high probability of cost-effectiveness. The mean ICER was robust to a number of deterministic sensitivity analyses.

The strengths of this model stem from the underlying evidence. GRAALL-R was a large randomized trial conducted with prolonged follow-up. Regarding EFS, we have complete follow-up when considering a cure threshold at 5 years. From GRAALL-R, nearly 8 years of OS data are available. Although the numbers at risk did approach low numbers at the tails, proper digitization and reproduction of individual patient data (IPD) allowed the probabilistic sensitivity analysis to correctly model the inherent uncertainty by fitting parametric distributions. Thus, information size, or any potential lack thereof, has been properly characterized, analysed, and weighted in the probabilistic sensitivity analysis, whose decision outcomes were relatively robust and precise. The present study, although adapted for a Canadian perspective, is largely generalizable beyond Canada in its direction and relative magnitude of economic outcomes.

The pan-Canadian Drug Oncology Review (pCODR) recently recommended against reimbursing rituximab + SOC in adult Ph + ve BCP-ALL. One of the main criticisms was that the GRAALL-R trial used a different combination chemotherapy backbone than what would be commonly used in Canada. We recognize the limitation in generalizability. However, the DFCI, hyper-CVAD, and GRAALL-R (basically hyper-CVAD) chemotherapy backbones use similar drugs, and are largely substitutable when assessing the treatment effect of adding rituximab. There is, thus, reason to believe the incremental effects of adding rituximab, observed in the GRAALL-R trial, are generalizable to Canadian regimens. Finally, ALL patients are unlikely to ever be subjected to a large randomized trial using Canadian-specific regimens.

The present analysis is limited by generalizability of not only the comparator, but also potentially the jurisdiction and the trial design. European patients and practice may differ in other ways. Also, real-world effectiveness may differ to a randomized setting. pCODR’s other main criticism related to EFS; it was not believed to be a relevant or meaningful end-point. Failure to achieve complete remission, incident relapse of disease, or death are all extremely important patient relevant outcomes which can affect prognosis, quality-of-life, or costs. We emphasize that EFS is a clinically relevant and valuable end-point in frontline trials for ALL, and that EFS remains the standard, and preferred outcome for frontline RCTs in ALL. This includes many of the practice changing trials to date^30–35. An alternative modeling approach would be to use separate time-to-event curves specific to remission, relapse, and death, accounting for competing risks. As we did not have access to primary data, we could not complete this approach.

The present analysis differs from the pCODR submission in a few additional aspects. The present model uses a true lifetime horizon, whereas the pCODR submission uses only 15 years, because pCODR generally does not extend to a lifetime without evidence of complete survival (100% death), and most studies publish prior to that timepoint. Finally, we use a utility decrement whilst on treatment, identified by pCODR.

Incomplete evidence means complete estimates involve extrapolation. Predicting the future always carries uncertainty. The present analysis uses the data as observed by GRAALL-R to inform its estimates of the likely differences in EFS and OS. Decision results were robust to cutting the treatment effect at 48 months. As pCODR discarded the EFS end-point entirely and a non-significant OS result was translated into a conclusion of no net benefit, they employed a HR = 1.0 for both EFS and OS as a re-analysis of the most likely scenario. We believe estimates should be anchored with the best available evidence at our disposal, the GRAALL-R estimates. The best estimate is to use the EFS and OS data; both the means and their uncertainties. That is what we have presented in this analysis.

In GRAALL-R, more patients in the rituximab added to SOC arm received HCT compared to the SOC arm. However, HCT was not the cause of reduced relapse or improvement in EFS or OS. Maury et al.⁵ cite three reasons for this. First, most of the HCT-associated mortality was in the rituximab group. Second, when patients were censored at the time of HCT and introduced as a time-dependent covariate, EFS and OS improvements remained significant. In fact, results of GRAALL-R showed evidence of a possible negative bias of HCT; the confounding effect from HCT works against the observed treatment effect. This opposing confounding can be a reason to upgrade the quality of evidence³⁶. However, this analysis was a post-hoc analysis, and should be interpreted as descriptive.

The present economic evaluation assessed the impact of rituximab added to SOC to SOC alone in adult CD20 + ve, Ph-ve, BCP-ALL. Its scope included the health and cost consequences relevant to the public payer using a randomized controlled trial, the best available evidence at present. Rituximab + SOC improved EFS, OS, and cure, whilst reducing relapse/resistant disease. Rituximab added to SOC was found to be a cost-effective intervention, compared to SOC alone, with an average ICER of C$21,828/QALY.

Transparency

Declaration of funding

There was no sponsorship or funding to declare Hoffmann-La Roche Limited or otherwise.

Declaration of financial/other relationships

JN, RM, and SY are employees of Hoffmann-La Roche Limited. MG, KP, and MS did not receive compensation from Hoffmann-La Roche Limited for their work on this study and manuscript and have no other relevant disclosures. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Previous presentations

This manuscript is the first disclosure of the full methods and results of this economic evaluation. Some of the methods and results were disclosed by submission of the economic model to the Canadian Agency for Drugs and Technologies in Health (CADTH) as part of a reimbursement health technology assessment; funding recommendation is expected to be published August 31, 2017. In addition, the summary of methods and findings were presented at the CADTH Symposium (April 23–25, 2017; Ottawa, Canada) and the ISPOR 22nd Annual International Meeting (May 20–24, 2017; Boston, MA).