- Email: [email protected]
Cost-effectiveness analysis of LHRH agonists in the treatment of metastatic prostate cancer in Italy
VALUE IN HEALTH 14 (2011) 80-89
available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/jval
Cost-effectiveness analysis of LHRH agonists in the treatment of metastatic prostate cancer in Italy S. Iannazzo, EngD, MBAa,*, L. Pradelli, MDa, M. Carsi, MSca, M. Perachino, MDb a b
AdRes Health Economics & Outcomes Research, Torino, Italy Ospedale Santo Spirito, Casale Monferrato, Alessandria, Italy
A B S T R A C T
Keywords:
Objectives: Luteinizing hormone-releasing hormone (LHRH) agonists represent one of the
Cost-effectiveness
main cost factors in the management of patients with metastatic prostate cancer. We com-
Hormone therapy
pared the cost-effectiveness of the five different 3-month formulations of LHRH agonists
Prostate cancer
currently available for advanced prostate cancer in Italy, because these differ both in their
Testosterone
capacity to suppress testosterone and in their acquisition costs. Methods: A probabilistic, patient-level simulation model was developed to compare the cost-effectiveness, from the perspective of the Italian National Health Service (INHS), of leuprorelin 11.25 mg and 22.5 mg, triptorelin 11.25 mg, buserelin 9.9 mg, and goserelin 10.8 mg. The model incorporated testosterone-dependent progression-free and cancer-specific survival functions, LHRH agonist effectiveness data, and national costs and tariffs. Cox’s proportional hazard models were used to compute total and progression-free survival functions based on clinical data from 129 patients with metastatic prostate cancer treated in an Italian center. Bayesian random effects models were employed to summarize evidence from published literature on testosterone suppression obtained with the available LHRH agonists. Results: Estimated total survival was ⬇5 years, with a maximum difference between treatment options of ⬇2 months. There was a mean difference of almost €2,500 in lifetime total costs between the least costly option (leuprorelin 22.5 mg) and the most expensive (goserelin). In the incremental cost-effectiveness analysis, leuprorelin 22.5 mg dominated all alternatives except buserelin, which had an incremental cost-effectiveness ratio versus leuprorelin 22.5 mg of ⬇€12,000 per life-month gained. Conclusions: Based on modelling with meta-analysis of comparative survival data, leuprorelin 22.5 mg was the most cost-effective treatment of the available depot formulation LHRH agonists. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc.
Funding: L. Pradelli and Sergio Iannazzo are employees and owners of AdRes Srl, Torino, Italy, an independent health economics consulting and research company. Astellas Pharma Italia supported the development of the model through a consulting service contract with AdRes. M. Perachino cooperated with AdRes for the development of the model through a consulting agreement. The authors all contributed to the paper and approved it. A poster of this study was displayed (presentation code: PCN122) at the ISPOR 12th Annual European Congress, October 24 –27, 2009, Paris, France. * Address correspondence to: Sergio Iannazzo, AdRes Health Economics & Outcomes Research, P.zza Carlo Emanuele II, 19, I-10123 Torino, Italy. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2010.10.023
81
VALUE IN HEALTH 14 (2011) 80-89
Fig. 1 – Simplified structure of the model, represented as a decisional tree, with evaluated treatment options. In each branch, the simulation is conducted using a Markov chain. For simplicity, only the leuprorelin 22.5-mg branch is fully represented. pCancerDeath, probability of cancer-related death; pDeath, probability of death; pProgression, probability of disease progression.
Introduction Prostate carcinoma is the second most common tumor in Italy, second only to lung cancer, with an estimated incidence of 42,804 cases in 2005 (total male population of 28,376,804, of which 4,717,170 were older than 65 years of age), representing
17% of all diagnosed cancers. Furthermore, it is the third most common cancer-related cause of death, after lung and colorectal cancers. In 2005 there were 9070 deaths from prostate cancer in Italy, corresponding to 7% of total cancer mortality [1,2]. Although, to the best of our knowledge, no comprehensive cost-of-illness study is available for prostate cancer in Italy, a
Table 1 – Clinical parameters included in the cost-effectiveness model. Variable Mean baseline characteristics of the patient cohort PSA at 6 months, ng/ml Gleason score Age, years Cox’s proportional hazard model 1: coefficients of time to biochemical relapse Gleason score* Ln PSA at 6 months† Testosterone level at nadir2* Cox’s proportional hazard model 2: coefficients (SD) of cancer-specific survival after relapse Gleason score† Ln PSA at 6 months† Testosterone at nadir2†
LHRH agonist Meta-analysis results: nadir testosterone concentrations (ng/dl) Goserelin Leuprorelin 3.75/11.25 mg (Enantone®) Leuprorelin 7.5/22.5 mg (Eligard®) Buserelin Triptorelin Ln, logarithm; PSA, prostate-specific antigen; SD, standard deviation. * P ⬍ 0.05. † P ⬍ 0.01.
Mean or coefficient 185.8 7.0 74.6
SD 344.1 1.5 6.7
0.173 0.336 2.346
0.081 0.046 1.153
0.291 0.208 3.998
0.101 0.051 1.30
Mean
SD
18.41 13.45 9.09 8.48 10.72
6.95 3.02 2.86 3.97 4.85
82
VALUE IN HEALTH 14 (2011) 80-89
recent multinational study assessed the direct initial health care costs during the first year after diagnosis. For Italy, the estimate was approximately €5200 per patient, leading to an estimated annual cost of €202 million for newly diagnosed patients alone [3]. In cases of metastatic prostate cancer, luteinizing hormone-releasing hormone (LHRH) agonist monotherapy is the first-line treatment recommended by Italian guidelines, given that Italian patients generally refuse surgical castration and that there is no evidence that combined androgen blockade is superior [4].
The rationale for anti-androgen therapy in prostate cancer is the observation that testosterone represents an essential growth stimulus for prostatic cancer cells during much of the biological history of this tumor. LHRH agonists modulate the physiological feedback mechanism along the hypothalamicpituitary-gonadal axis, with consequent suppression of testicular testosterone production. The judgment on the optimum testosterone level to be reached for suppressing the growth of prostatic cancer cells has evolved alongside the detection ability of the assays used to determine serum testosterone concentrations: whereas the
Table 2 – Analysis of comparative efficacy of different LHRH agonists in lowering serum testosterone levels in patients with metastatic prostate cancer. Authors [reference]
LHRH agonist
N
Serum testosterone, ng/dl Mean Median
Waxman et al. [10] Chander et al. [25] Pettersson et al. [26] Kaisary et al. [11] Fernandez del Moral et al. [12] Fernandez del Moral et al. [12] Debruyne et al. [13] Debruyne et al. [13] Debruyne et al. [13] Debruyne et al. [13] Yri et al. [28] Fontana et al. [21] Fujii et al. [29] Sarosdy et al. [16] Fujii et al. [29] Crawford et al. [27] Oefelein [14] Oefelein [15] Oefelein and Cornum [17] Fowler et al. [18] Chu et al. [19] Perez-Marreno et al. [20] Sartor et al. [22] Heyns et al. [23] Heyns et al. [23] Teillac et al. [24] Teillac et al. [24] Yri et al. [28] Fujii et al. [29] Fujii et al. [29]
SD Min Max Percentile Percentile Cutoff 2.5% 97.5%
Buserelin‡ Buserelin† Buserelin‡ Goserelin* Goserelin*
4 69 21 148 83
6.2 13.6 5.5 36.3 17.3
NR NR 4.0 NR NR
NR NR 5.3 NR 11.2
0 NR 2.0 NR NR
15.0 NR 26.0 NR NR
NR NR 2 NR NR
NR NR 19 NR NR
Goserelin‡
77
17.9
NR
14.1
NR
NR
NR
NR
Goserelin*,# Goserelin‡,# Goserelin*,** Goserelin‡,** Goserelin‡ Goserelin‡ Goserelin* Goserelin‡ Goserelin‡ Leuprorelin (Eligard)㛳 Leuprorelin (Eligard)‡ Leuprorelin (Eligard)‡ Leuprorelin (Eligard)‡
38 42 39 41 25 120 50 59 74 103 13 32 38
15.3 19.4 19.2 18.8 NR 11.3 10.0 17.5 12.0 12.3 NR NR NR
NR NR NR NR 26.0 NR 7.2 33.9 10.5 NR 10.0 10.0 NR
NR NR NR NR NR NR NR NR NR 2.1 11.0 NR NR
NR NR NR NR 14.4 NR NR NR NR NR 1.0 10.0 NR
NR NR NR NR 63.5 NR NR NR NR NR 130.0 40.0 NR
NR NR NR NR NR NR 5.1 NR 5.1 NR 4 10 NR
NR NR NR NR NR NR 13.1 NR 14.9 NR 49 20 NR
NR NR NR NR 81 58 50 50 50 50–20 20 20 50–20
NR NR NR NR 100 100 100 100 98.6 99–88.3 92.0 NR 95–87
Leuprorelin (Eligard)¶ Leuprorelin (Eligard)‡ Leuprorelin (Eligard)*
80 111 118
9.3 10.1 6.1
NR NR 4.3
NR 0.1 NR
NR NR 3.0
NR NR 27.0
NR NR NR
NR NR NR
50 50–20 50–20
98.8 100–94 100–97.5
Leuprorelin (Eligard)§ Leuprorelin (Eligard)* Triptorelin* Triptorelin* Triptorelin‡ Leuprorelin (Enantone)‡ Leuprorelin (Enantone)* Leuprorelin (Enantone)‡
90 140 137 68 63 40 40 68
12.4 4.1 4.3 13.9 13.9 NR 11.0 12.0
NR NR NR NR NR 17.3 8.2 9.9
0.8 NR NR NR NR NR NR NR
NR 0 0 NR NR 14.4 NR NR
NR 10.3 8.3 NR NR 271.1 NR NR
NR NR NR 7.2 6.0 NR 5.0 5.9
NR NR NR 20.8 20.8 NR 13.8 14.8
50–20 50 50 50 50 81 50 50
98–90 97.3 98.8 96.0 98.0 90.0 95.0 98.5
Max, maximum; Min, minimum; NR, not reported; SD, standard deviation. * 1-month depot † 2-month depot ‡ 3-month depot § 4-month depot 㛳 6-month depot ¶ 1-year depot # Study I ** Study II
72.1 NR 50–20 NR 58
Patients within cutoff, %
58
100 NR 100–95 NR 100 98.8
83
VALUE IN HEALTH 14 (2011) 80-89
Table 3 – Pharmaceutical costs. LHRH agonist
Brand
Public price [24], €/package
Ex-factory price, €/package
Leuprorelin 11.25 mg
Enantone 3-month depot
460.14
278.80
Leuprorelin 22.5 mg
Eligard 3-month depot
387.49
234.78
Goserelin
Zoladex 3-month depot
592.35
358.91
Triptorelin
Decapeptyl 3-month depot
524.59
317.85
Buserelin
Suprefact 3-month depot
409.29
247.99
traditional threshold was set at 50 ng/dl (“castration level”), with the introduction of more sensitive assays, it was lowered to 20 ng/dl. Nowadays, immunoassays are able to detect much lower concentrations, allowing investigating more accurately the relationship between circulating testosterone levels and clinical outcomes in patients with metastatic prostatic cancer. Recently, a continuous relationship between the risk of cancer-related death and serum testosterone levels has been demonstrated [5]. Five different depot formulations of LHRH agonists are currently available in Italy, for administration either once monthly or every 3 months. A recent review suggests that these have differing capacities to lower testicular production of testosterone, but all induce high rates of “castration” levels (⬍50 ng/dl) [6]. Once-monthly formulations have a higher acquisition cost per milligram than the corresponding 3-month doses, which have been approved on the basis of noninferiority studies. An expert roundtable discussion concluded that “LHRH agonists should be considered first-choice testosterone-lowering therapy for the treatment of prostate cancer, with the 3-month depot formulation providing optimal convenience and flexibility” [7]. In order to establish the most cost-effective option among LHRH agonists, an economic analysis was performed to compare the 3-month options: leuprorelin 11.25 mg and 22.5 mg, triptorelin 11.25 mg, buserelin 9.9 mg, and goserelin 10.8 mg. The aim of this work was the pharmacoeconomic comparative evaluation of 3-month LHRH agonist treatments for metastatic prostate cancer, from the per-
Table 4 – Costs considered in the model. Item LHRH agonist ex-factory acquisition costs, €/package Leuprorelin 11.25 mg Leuprorelin 22.5 mg Goserelin Triptorelin Buserelin Follow-up examinations, €/year Medical examination plus DRE PSA, hematology, blood chemistry Bone densitometry Cardiology evaluation (visit plus ECG) Abdominal sonography Bone scintigraphy (total body) Therapy for the hormone-refractory phase, € Chemotherapy and adverse events management
Cost 278.80 234.78 358.91 317.85 247.99 41.32 82.76 31.50 32.28 120.86 226.20 9475.58
DRE, digital rectal examination; ECG, electrocardiogram; PSA, prostate-specific antigen.
spective of the Italian National Health Service (INHS). The evaluation was performed by developing a model to simulate testosterone levels obtained with these therapies and the correlation between these levels and the risk of cancerrelated death.
Methods Model description and transition probabilities Clinical data used to compute total and progression-free survival were extracted from the database of the Santo Spirito Hospital in Casale Monferrato, Alessandria, Italy. In this administrative database all parameters related to patients followed in the Department of Urology are prospectively recorded. At the time we performed the analysis, data from 129 patients with metastatic prostate cancer (M1) were available. An economic model was developed with the TreeAge Pro 2009 decision analysis software (TreeAge Software, Inc., Williamstown, MA), using a patient-level, microsimulation technique. Each branch of the decisional tree represents a therapeutic option, and all branches have identical structures (Fig. 1). In the probabilistic, patient-level simulation, each patient is defined with unique characteristics of initial age, prostate-specific antigen (PSA) blood concentration, and Gleason score. Their initial values are defined in order to reproduce the mean and standard deviation (SD) of the studied patient’s cohort (Table 1). Five identical “clones” are sent through the five treatment branches. It is assumed that the simulation starts as soon as the testosterone nadir is reached, so that for each patient this value is constant and totally dependent on the LHRH agonist treatment. The clinical evolution of each single patient is defined by the progression between health states in a simple Markov chain (states of alive and dead). Until biochemical relapse occurs, the patient is subjected only to natural mortality. After biochemical relapse, a specific cancer-related mortality is added to the general population mortality. Current modelling guidelines recommend that the “cycle length of the model should be short enough so that multiple changes in pathology, symptoms, treatment decisions, or costs within a single cycle are unlikely” [8]. Clinical outcomes data used (disease progression and death; see next section) were reported on a monthly basis, whereas other relevant timedependent parameters change more slowly (LHRH agonists were given every 3 months, with follow-up examinations performed every 6 months). Thus, we chose 1 month as the Markov cycle length.
84
VALUE IN HEALTH 14 (2011) 80-89
Table 5 – Tests and examinations costs. Test/Examination Medical examination plus DRE PSA and hematochemical examinations Bone densitometry Cardiology evaluation (visit plus ECG) Abdominal echography Bone scintigraphy (total body) Total
Frequency
Unit cost [25], €
Annual cost, €
Every 6 months Every 6 months Yearly Yearly Every 6 months Every 6 months
20.66 41.38 31.50 32.28 60.43 113.1
41.32 82.76 31.50 32.28 120.86 226.20 534.92
DRE, digital rectal examination; ECG, electrocardiogram; PSA, prostate-specific antigen.
cally significant. In both models, the assumption of proportional hazard for each included covariate was also verified. The two Cox models were incorporated into the simulation model. For each simulated patient, the probability of disease progression and of cancer-related death was calculated as a function of Gleason score, PSA, and testosterone levels. Natural mortality was factored as a competing risk as a function of age, and was derived from the mortality tables of the general Italian population [9].
Survival functions We derived the progression-free and total survival functions by analyzing the outcomes data of the patient cohort using Cox semiparametric models. The Cox regression model analyzes the influence of independent variables on event-free survival, under the assumption of proportional hazards, which means that the hazard ratio must be constant over time. We analyzed the data to find relationships between baseline characteristics and the time of biochemical relapse and between those parameters and the total survival time after relapse. The first Cox regression model was based on data from the cohort of 129 metastatic patients (M1) and evaluated the time to PSA elevation (biochemical relapse). The Gleason score, testosterone level at nadir, PSA concentration (on a logarithmic scale) 6 months after the start of LHRH agonist therapy, and the age were considered as potential covariates. Except for initial age, all were significantly associated with the time to progression (Table 1). The second Cox model was developed on a subset of 89 metastatic patients (M1) who had biochemical relapse, in order to evaluate cancer-related survival. The same set of covariates was tested for significant correlations, and initial age alone was again discarded because it was not statisti-
Treatment effectiveness To identify relevant publications for the estimate of the effectiveness of LHRH agonists in lowering testosterone levels, in May 2009 we performed a literature search in Medline using the keywords “Gonadotropin-Releasing Hormone” [Mesh] “Testosterone/blood” [Mesh], with the limits “Clinical trial” and “Humans.” Abstracts of the 276 articles identified by this strategy were checked for relevance by two of the authors (LP and MP), and full texts were obtained for 35 articles that were judged by at least one of the authors to possibly contain information regarding testosterone blood levels achieved. Manually searching the reference lists of these latter articles yielded one additional title. After examining the full text of these 36 papers, 20 studies conducted from 1990 to 2007 [10 –29] were
Table 6 – Base-case results: average (SD) effectiveness, costs, and cost-effectiveness of LHRH agonists in the treatment of metastatic prostate cancer.
Effectiveness, months Overall survival (SD) Progression-free survival* (SD) Costs, € Hormone therapy (SD) Follow-up (SD) Chemotherapy (SD) Total (SD) Cost-effectiveness, €/month lived Cost-effectiveness Incremental cost-effectiveness, €/month of life gained ICER vs. leuprorelin 22.5 mg
Leuprorelin 22.5 mg
Leuprorelin 11.25 mg
Goserelin
Triptorelin
Buserelin
60.31 (34.09) 40.49 (41.42)
59.49 (33.75) 39.86 (40.86)
57.93 (32.93) 38.40 (39.22)
59.87 (33.99) 40.19 (41.17)
60.35 (34.15) 40.57 (41.55)
4857 (2670) 2809 (1522) 6315 (3116) 13,981 (5150)
5977 (3139) 2772 (1507) 6364 (3101) 15,114 (5502)
7405 (3943) 2703 (1470) 6472 (3058) 16,579 (6133)
6804 (3604) 2789 (1517) 6341 (3108) 15,935 (5909)
5428 (2825) 2811 (1525) 6307 (3121) 14,546 (5277)
232
254
286
266
241
NA
Dominated
Dominated
Dominated
11,700
ICER, incremental cost-effectiveness ratio; NA, not applicable; SD standard deviation. * Undiscounted. Dominated: More costly and less effective than leuprorelin 22.5 mg.
85
VALUE IN HEALTH 14 (2011) 80-89
Table 7 – Average (SD) life expectancy and progression-free survival with the different considered treatments.
Months lived (SD) Months free from biochemical relapse* (SD)
Leuprorelin 22.5 mg
Leuprorelin 11.25 mg
Goserelin
Triptorelin
Buserelin
60.31 (34.09) 40.49 (41.42)
59.49 (33.75) 39.86 (40.86)
57.93 (32.93) 38.40 (39.22)
59.87 (33.99) 40.19 (41.17)
60.35 (34.15) 40.57 (41.55)
SD, standard deviation. * Not discounted.
identified as reporting detailed data on the subject, some of them previously identified in a recent review by Novara et al. [6]. These studies evaluated goserelin (n ⫽ 7), leuprorelin 7.5 or 22.5 mg (Eligard®; n ⫽ 9), leuprorelin 3.75 or 11.25 mg (Enantone®; n ⫽ 2), triptorelin (n ⫽ 2), and buserelin (n ⫽ 3). All data (mean, median, SD, minimum-maximum, and percentile values) relating to the distribution of testosterone values at nadir and the percentage of patients who reached the testosterone cutoff point (almost always 50 ng/dl) were extracted from these studies (Table 2) and meta-analyzed using a Bayesian random effects model. In a random effects model it is assumed that the effects observed in different trials for the same drug can be ascribed to a common effect, which is then influenced by peculiar conditions of each single study. The results of the meta-analysis, summarized in Table 1, were incorporated into the model to simulate the levels of testosterone reached at nadir for each LHRH agonist.
Costs Costs considered in the model include acquisition costs of LHRH agonists (Table 3)., based on ex-factory prices [30] and doses recommended in the summary of product characteristics (SPC), costs of monitoring tests and examinations, and costs of chemotherapy for the hormone-refractory phase in patients with biochemical relapse (Table 4). Consumption of follow-up procedures was modeled according to national guideline recommendations [4], and their costs were assessed based on the national price lists for ambulatory services, which also include diagnostic examinations (Table 5) [31]. The cost of managing toxicities related to LHRH agonist treatment was not included, assuming that this cost would not differ between strategies. Resources consumed during chemotherapy were based on data from the randomized, controlled trial on hormone-refractory prostate cancer patients reported by Tannock et al. [32]. Resources include drugs and administration of the chemotherapy regimen (docetaxel plus prednisolone, for an average of 9.5 cycles), premedication (corticosteroid plus antiemetic agent), and drugs used for managing grades III to IV
anemia and neutropenic fever (darbepoetin alfa 500 g every 3 weeks for the first 9 weeks and 300 g every 3 weeks subsequently, and pegfilgrastim 0.6 mg every 3 weeks, respectively, as recommended by the SPCs). The resulting total cost for treating the terminal phase was estimated at €9475.58 and was applied to all patients dying of cancer.
Discounting Italian guidelines for economic evaluations in health care issued in 2001 recommend applying a 3% discount rate to both future costs and outcomes [33]. The same value has been indicated in the more recent proposal of guidelines for economic evaluation elaborated and discussed in 2009 by the Italian Association of Health Economics (AIES) [34]. Presently, the European Commission advocates a 5% rate in its current cost-benefit guidance [35]. This rate is considered to be too high, however, by many economists [36], and the National Institute for the Social Impact of Economy (Istituto Nazionale per l’Impatto Sociale dell’Economia, INISE) recommends a discount rate approximately 1.5% lower than the official social discount rate [37]. Thus, we chose to apply an annual discount rate of 3.5% to both survival and costs.
Probabilistic sensitivity analysis The base case of the model is built to simulate the outcomes and costs for individual patients through the model. Results of the base case were calculated with 20,000 simulated patients. Characteristics of each patient (age, PSA blood concentration, and Gleason score) were extracted from distributions representing the mean (SD) from the studied cohort of 129 metastatic patients (M1) (Table 1). A probabilistic sensitivity analysis was performed to take into account the uncertainty surrounding estimated parameters of the two Cox models. This was accomplished by repeating 1000 cycles, each cycle with 5000 patient-level iterations. Each cycle was characterized by a single value for each of the parameters of the Cox models. These were randomly ex-
Table 8 – Average (SD) lifetime costs for each patient with the considered treatments. Cost type Hormone therapy (SD), € Follow-up (SD), € Chemotherapy (SD), € Total (SD), €
Leuprorelin 22.5 mg
Leuprorelin 11.25 mg
Goserelin
Triptorelin
Buserelin
4856.87 (2670.25) 2809.05 (1521.93) 6315.29 (3115.62) 13,981.21 (5150.10)
5977.07 (3138.54) 2772.40 (1506.50) 6364.41 (3101.18) 15,113.88 (5502.42)
7404.81 (3943.09) 2702.59 (1470.35) 6471.77 (3058.34) 16,579.18 (6133.08)
6804.14 (3603.69) 2789.34 (1517.25) 6341.11 (3108.41) 15,934.60 (5909.45)
5427.60 (2825.12) 2811.31 (1524.54) 6307.34 (3120.84) 14,546.25 (5277.24)
86
VALUE IN HEALTH 14 (2011) 80-89
Fig. 2 – Comparative cost-effectiveness of five LHRH agonist formulations in the treatment of metastatic prostate cancer. In this representation, the higher the slope, the more cost-effective the treatment: better clinical results are obtained at the same cost.
tracted from distributions representing the uncertainty of the estimate (Table 1).
cost-effectiveness ratio (ICER) of this drug (i.e., the cost required to obtain one additional unit of benefit) compared to leuprorelin 22.5 mg was about €12,000 per life-month gained (Table 9).
Results Probabilistic sensitivity analysis
Base case Our meta-analysis showed that the lowest testosterone nadirs were achieved with buserelin and leuprorelin 22.5 mg (⬇8 –9 ng/dl versus ⬇18 ng/dl with goserelin; Table 1). Estimated total survival for metastatic patients was approximately 5 years, with a maximum difference between treatment options of approximately 2 months (Table 6). The maximum difference between treatments for progression-free survival was almost 2 months (Table 7). The differences in the lifetime total costs of therapy were more pronounced, with a mean difference of nearly €2500 between the least costly option (leuprorelin 22.5 mg) and the most expensive (goserelin) (Tables 6 and 8). From the perspective of cost-effectiveness, leuprorelin 22.5 mg had the best cost/survival ratio, with a cost of approximately €230 per month (Table 6). In the incremental cost-effectiveness analysis (Fig. 2), leuprorelin 22.5 mg dominated leuprorelin 11.25 mg, goserelin, and triptorelin, having greater effectiveness and lower costs. Buserelin appeared more costly but slightly more effective than leuprorelin 22.5 mg. The incremental
Leuprorelin 22.5 mg was the least costly drug in all 1000 simulations, and, in terms of months of life gained, only buserelin was more effective in a certain percentage of iterations (Fig. 3). Therefore, leuprorelin 22.5 mg consistently dominated leuprorelin 11.25 mg, goserelin, and triptorelin. In Figure 3, the 1000 ICER values of buserelin versus leuprorelin 22.5 mg are represented as a scatterplot on a cost-effectiveness plane. Points are all found in the northeast and southeast quadrants. Points in the southeast quadrant represent a negative incremental survival gain of buserelin versus leuprorelin 22.5 mg, indicating that leuprorelin was superior to buserelin in terms of both costs and effectiveness (dominant). Buserelin never dominated leuprorelin, because its costs were always higher. ICER estimates of buserelin versus leuprorelin 22.5 mg were also plotted on a cost-effectiveness acceptability curve, in which the probability of being cost-effective is represented as a function of increasing levels of willingness to pay (WTP). As observed in Figure 4, there was a 100% chance of leuprorelin 22.5 mg being cost-effective when WTP was
Table 9 – Cost-effectiveness and incremental cost-effectiveness with respect to leuprorelin 22.5 mg, in the base case.
Cost-effectiveness, €/month lived ICER vs. leuprorelin 22.5 mg, €/month of life gained
Leuprorelin 22.5 mg
Leuprorelin 11.25 mg
Goserelin
Triptorelin
Buserelin
231.84
254.06 Dominated
286.20 Dominated
266.17 Dominated
241.01 11,700.22
ICER, incremental cost-effectiveness ratio. Dominated: more costly and less effective than leuprorelin 22.5 mg.
VALUE IN HEALTH 14 (2011) 80-89
Fig. 3 – Scatterplot of the 1000 incremental costeffectiveness ratio (ICER) estimates for buserelin versus leuprorelin 22.5 mg in the probabilistic sensitivity analysis. 〫 shows the mean value of the cost-effectiveness ratio. NE, northeast; SE, southeast. less than €28,000 per life-year gained. Buserelin was costeffective in more than 50% of cases only if the willingness to pay was greater than €280,000 per life-year gained.
87
In our analysis, estimated testosterone nadirs reached with different depot LHRH agonists were combined with survival functions obtained via Cox analysis of data from metastatic patients in order to compare their cost-effectiveness. Results obtained in the base case showed that there were some differences in mean overall survival and lifetime costs with the different treatments. Leuprorelin 22.5 mg on average dominated all alternative options except buserelin, being associated with lower costs and higher effectiveness. In a sensitivity analysis we explored the effect of the uncertainty on the main model parameters. The mean lower cost and higher effectiveness of leuprorelin 22.5 mg versus leuprorelin 11.25 mg, goserelin, and triptorelin was confirmed in all 1000 estimates. Buserelin, on average, was more costly but more effective than leuprorelin 22.5 mg, with an ICER of approximately €12,000 per life-month gained in the base case. An official WTP threshold for the INHS is lacking. To assess the value of a health intervention, results of pharmacoeconomic studies need, therefore, to be benchmarked against international or conventional thresholds. In the literature, there is a plethora of proposed cost-effectiveness thresholds, generally ranging from US$20,000/quality-adjusted life year (QALY) to the current £30,000/QALY applied in the United Kingdom [38 – 41]. The opportunity of raising this threshold is constantly under debate [42]; possibly the highest proposed cost-effectiveness threshold is the three times per capita gross domestic product (GDP) value suggested by the World
Discussion It is well known that the different available formulations of LHRH agonists induce different levels of testicular inhibition, but clinicians have not considered these differences to be clinically relevant, provided that the castration threshold is achieved. However, as an analysis of recently published clinical data suggests a continuous relationship between testosterone concentrations and disease progression, differences in the testosterone nadir achieved— even below the castration threshold—might determine variations in the rapidity of disease evolution and, therefore, in survival.
Health Organization (WHO) in the WHO-CHOICE program for efficient use of health resources [43]. In the probabilistic sensitivity analysis, leuprorelin 22.5 mg had an 80% probability of being cost-effective in comparison to buserelin even adopting this “extreme” threshold (equating to €78,000 based on Italy’s 2007 GDP per capita [44]). There is high variability in the differences in outcomes and costs associated with the evaluated treatments in the patientlevel simulations, which have distributions that are largely overlapping. In contrast, consistent differences in mean survival and costs were seen in the probabilistic sensitivity anal-
Fig. 4 – Incremental cost-effectiveness acceptability curve for buserelin versus leuprorelin 22.5 mg in the treatment of metastatic prostate cancer.
88
VALUE IN HEALTH 14 (2011) 80-89
ysis. We believe that this apparent discrepancy is related to the simulation technique that we used, in turn reflecting the different type of uncertainty that the two analyses are designed to reproduce. In the patient-level simulations, the variability reflects the interindividual heterogeneity and corresponds to the low predictability of the outcomes of the single patient. The much lower dispersion of mean estimates shown by the probabilistic sensitivity analysis, however, reflects the substantial low sensitivity of mean, cohort-level outcomes to the uncertainty in the parameters relating clinical characteristics and outcomes, as estimated by the survival functions. In other terms, the clinical effect of the treatments, and more generally the outcomes, of the single patient are very unpredictable. Indeed, it may well be possible that a patient treated with an LHRH agonist that is less effective than others on average does better than another similar patient treated with a drug that is on average more effective, whereas in a cohort of patients the difference will emerge. These considerations further imply that the present results are relevant for decision makers responsible for choices and budgets regarding cohorts of patients—from the head of the department and hospital drug purchaser to the policy makers at regional and national levels—while having little or no importance for the clinician deciding about the care of a single patient.
costs are common to all treatments and the maximal mean survival difference between treatments is approximately 2 months, so we are confident that the potential of these estimates to bias the final results is very limited.
Conclusions Despite some limitations, our analysis represents the first attempt to summarize the available evidence regarding the relative effectiveness of testosterone-lowering therapies in metastatic prostate cancer and their consequences on hard outcomes into one comprehensive conceptual scheme. The results suggest that leuprorelin 22.5 mg is the most cost-effective of the available depot LHRH agonists in the Italian setting. Although only tentative conclusions can be drawn, based on modeling with meta-analysis of comparative survival data, our findings should provide valuable guidance for decision makers who need to make rational choices regarding the use and price of LHRH agonists in Italian patients with metastatic prostate cancer —like the hospital pharmacist, the head of the urology department, or the regional agency deciding on hospital drug formularies.
Limitations
Acknowledgment
The model relationships between testosterone levels and cancer mortality and progression are based on the analysis of data from one clinical center. The use of a continuous analytical function even for very low levels of the hormone implies the absence of a threshold below which testosterone stops stimulating cancer cell growth. Such a threshold is referred to as castration level in earlier literature and its non existence might need to be further investigated and confirmed by larger clinical studies. The retrieval of the evidence on testosterone nadirs achieved with the available treatments might have been more formally rigorous, although we believe that we have not omitted any crucial information. The findings of the review by Novara et al. [6] support this belief. Regarding the cost side of the evaluation, some simplifications and approximations have been used. The national price list used for estimating the costs of follow-up diagnostic procedures and examinations dates back to 1996; therefore, costs are likely underestimated. The updating of price lists is the responsibility of the regions, but many of them have not complied with this duty to date. For example, for the most expensive of the diagnostic procedures considered in the present model (i.e., total body bone scintigraphy), only 10 of the current 20 regional price lists have an updated value. The reported mean of regional prices is of limited representativeness, because it neglects any weighting based on activity volumes or target population. It is, however, higher than the old national price list [45]. The consequent underestimation of follow-up costs in the model may have slightly biased the results in favor of the strategies associated with longer survival. The cost of chemotherapy for the hormone-refractory phase used in the calculations is also an aggregate estimate that may not reflect costs accurately. Nevertheless, all of these
Editorial support was provided by Julia Balfour, Consultant Medical Writer, Kilconquhar, United Kingdom. REFERENCES
[1] Grande E, Inghelmann R, Francisci S, et al. Regional estimates of all cancer malignancies in Italy. Tumori 2007; 93:345–51. [2] Inghelmann R, Grande E, Francisci S, et al. Regional estimates of prostate cancer burden in Italy. Tumori 2007;93:380 – 6. [3] Fourcade RO, Benedict A, Black LK, et al. Treatment costs of prostate cancer in the first year after diagnosis: a short-term cost of illness study for France, Germany, Italy, Spain and the UK. BJU Int 2010;105:49 –56. [4] AURO.it (Associazione Urologi Italiani—Association of Italian Urologists) AUROLINE 10: Guidelines on prostate cancer: diagnosis, staging and therapy, 2008. Available from: http://www.auro.it/wp-content/uploads/al10.pdf [Accessed May 4, 2010]. [5] Perachino M, Cavalli V, Bravi F. Testosterone levels in patients with metastatic prostate cancer treated with luteinizing hormone-releasing hormone therapy: prognostic significance? BJU Int 2010;105:648 –51. [6] Novara G, Galfano A, Secco S, et al. Impact of surgical and medical castration on serum testosterone level in prostate cancer patients. Urol Int 2009,82:249 –55. [7] de Jong IJ, Eaton A, Bladou F. LHRH agonists in prostate cancer: frequency of treatment, serum testosterone measurement and castrate level: consensus opinion from a roundtable discussion. Curr Med Res Opin 2007;23:1077– 80. [8] Weinstein MC, O’Brien B, Hornberger J, et al. Principles of good practice of decision analytic modeling in health care evaluation: Report of the ISPOR Task Force on Good Research Practices—Modeling Studies. Value Health 2003;6: 9 –17.
VALUE IN HEALTH 14 (2011) 80-89
[9] ISTAT (Italian National Institute of Statistics) Table 2.16: Mortality Table by Age and Sex, 2005. Annuario Statistico Italiano, ISTAT, 2008. [10] Waxman J, Sandow J, Abel P, et al. Three-monthly GnRH agonist (buserelin) for prostatic cancer. Br J Urol 1990,65:43–5. [11] Kaisary AV, Tyrrell CJ, Peeling WB, Griffiths K. Comparison of LHRH analogue (Zoladex) with orchiectomy in patients with metastatic prostatic carcinoma. Br J Urol 1991;67:502– 8. [12] Fernandez del Moral P, Dijkman GA, Debruyne FM, et al. Three-month depot of goserelin acetate: clinical efficacy and endocrine profile. Dutch South East Cooperative Urological Group. Urology 1996;48:894 –900. [13] Debruyne FM, Dijkman GA, Lee DC, et al. A new long acting formulation of the luteinizing hormone-releasing hormone analogue goserelin: results of studies in prostate cancer. J Urol 1996;155:1352– 4. [14] Oefelein MG. Time to normalization of serum testosterone after 3-month luteinizing hormone-releasing hormone agonist administered in the neoadjuvant setting: implications for dosing schedule and neoadjuvant study consideration. J Urol 1998;160:1685– 8. [15] Oefelein MG: Serum testosterone-based luteinizing hormone-releasing hormone agonist redosing schedule for chronic androgen ablation: a phase I assessment. Urology 1999;54:694 –9. [16] Sarosdy MF, Schellhammer PF, Soloway MS, et al. Endocrine effects, efficacy and tolerability of a 10.8-mg depot formulation of goserelin acetate administered every 13 weeks to patients with advanced prostate cancer. BJU Int 1999;83:801– 06. [17] Oefelein MG, Cornum R. Failure to achieve castrate levels of testosterone during luteinizing hormone releasing hormone agonist therapy: the case for monitoring serum testosterone and a treatment decision algorithm. J Urol 2000;164:726 –9. [18] Fowler JE, Flanagan M, Gleason DM, et al. Evaluation of an implant that delivers leuprolide for 1 year for the palliative treatment of prostate cancer. Urology 2000;55:639 – 42. [19] Chu FM, Jayson M, Dineen MK, et al. A clinical study of 22.5 mg. La-2550: a new subcutaneous depot delivery system for leuprolide acetate for the treatment of prostate cancer. J Urol 2002;168:1199 –203. [20] Perez-Marreno R, Chu FM, Gleason D, et al. A six-month, open-label study assessing a new formulation of leuprolide 7.5 mg for suppression of testosterone in patients with prostate cancer. Clin Ther 2002;24:1902–14. [21] Fontana D, Mari M, Martinelli A, et al. 3-month formulation of goserelin acetate (“Zoladex” 10.8-mg depot) in advanced prostate cancer: results from an Italian, open, multicenter trial. Urol Int 2003;70:316 –20. [22] Sartor O, Dineen MK, Perez-Marreno R, et al. An eight-month clinical study of LA-2575 30.0 mg: a new 4-month, subcutaneous delivery system for leuprolide acetate in the treatment of prostate cancer. Urology 2003;62:319–23. [23] Heyns CF, Simonin M, Grosgurin P, et al. Comparative efficacy of triptorelin pamoate and leuprolide acetate in men with advanced prostate cancer. BJU Int 2003;92:226 –31. [24] Teillac P, Heyns CF, Kaisary AV, et al. Pharmacodynamic equivalence of a decapeptyl 3-month SR formulation with the 28-day SR formulation in patients with advanced prostate cancer. Horm Res 2004;62:252– 8. [25] Chander S, Choo R, Danjoux C, et al. Effect of androgen suppression on hemoglobin in prostate cancer patients undergoing salvage radiotherapy plus 2-year buserelin acetate for rising PSA after surgery. Int J Radiat Oncol Biol Phys 2005;62:719 –24. [26] Pettersson B, Varenhorst E, Petas A, Sandow J. Duration of testosterone suppression after a 9.45 mg implant of the GnRH-analogue buserelin in patients with localised
[27]
[28]
[29]
[30] [31] [32]
[33]
[34] [35]
[36]
[37]
[38]
[39]
[40] [41] [42] [43]
[44]
[45]
89
carcinoma of the prostate: a 12-month follow-up study. Eur Urol 2006;50:483–9. Crawford ED, Sartor O, Chu F, et al. A 12-month clinical study of LA-2585 (45.0 mg): a new 6-month subcutaneous delivery system for leuprolide acetate for the treatment of prostate cancer. J Urol 2006;175:533– 6. Yri OE, Bjoro T, Fossa SD. Failure to achieve castration levels in patients using leuprolide acetate in locally advanced prostate cancer. Eur Urol 2006;49:54 – 8. Fujii Y, Yonese J, Kawakami S, et al. Equivalent and sufficient effects of leuprolide acetate and goserelin acetate to suppress serum testosterone levels in patients with prostate cancer. BJU Int 2008;101:1096 –100. Italian Drug Formulary: L’Informatore Farmaceutico, 69th ed. Italy: Elsevier Masson, 2009. Italian Specialist Tariffs: Nomenclatore Tariffario Nazionale, D.M. 22, July 1996. Tannock IF, de Wit R, Berry WR, et al. Docetaxel plus prednisone or mitoxantrone plus prednisone for advanced prostate cancer. N Engl J Med 2004;351:1502–12. Capri S, Ceci A, Terranova L, et al. Guidelines for economic evaluations in Italy: recommendations from the Italian group of pharmacoeconomic studies. Drug Inf J 2001;35:189–201. Guide to Cost-Benefit Analysis of Investment Projects. Evaluation Unit, DG Regional Policy, European Commission, 2002. Italian Health Economics Association (AIES). A proposal for guidelines for the economic evaluation of health interventions in Italy. PharmacoEconomics Italian Research Articles 2009;11: 83–93. Evans D. Social discount rate for the European Union: new estimates. In: Florio M, ed. Cost-Benefit Analysis and Incentives in Evaluation: The Structural Funds of the European Union. Edward Elgar Publishing, 2007. Istituto Nazionale per l’Impatto Sociale dell’Economia (INISE). L’analisi costi-benefici. Available from: http://www.bilanciosociale.info/public/documenti/ Libro/capitolo%204.doc [Accessed May 2010]. Laupacis A, Feeny D, Detsky AS, Tugwell PX. How attractive does a new technology have to be to warrant adoption and utilization? Tentative guidelines for using clinical and economic evaluations. CMAJ 1992;146:473– 81. Tosteson AN. Quality of life in the economic evaluation of osteoporosis prevention and treatment. Spine 1997;22(24 Suppl.):58S– 62S. Kanis JA, Jonsson B. Economic evaluation of interventions for osteoporosis. Osteoporos Int 2002;13:765–7. Raftery J. NICE: faster access to modern treatments? Analysis of guidance on health technologies. BMJ 2001;323:1300–3. Towse A, Raftery J. Should NICE’s threshold range for cost per QALY be raised? BMJ 2009;338:268 –9. World Health Organization. Commission on Macroeconomics and Health Report. Macroeconomics and Health: Investing in Health for Economic Development. Geneva, 2001. Available from: http://whqlibdoc.who.int/publications/2001/924154550x.pdf [Accessed November 2010]. ISTAT (Italian National Institute of Statistics) Historical Series of Per Capita Gross Domestic Product (GDP) (1970 –2007). Available from: http://www.istat.it/conti/ [Accessed March 2010]. Agenzia Nazionale per i Servizi Sanitari Regionali. Tariffe delle prestazioni specialistiche ambulatoriali aggiornato al 31 dicembre 2008. Nomenclatore delle Prestazioni specialistiche ambulatoriali— confronto tra le tariffe nazionali ex DM 1996 e le tariffe regionali relative all’anno 2008. Available from: http://www.agenas.it/ [Accessed February 2010].
available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/jval
Cost-effectiveness analysis of LHRH agonists in the treatment of metastatic prostate cancer in Italy S. Iannazzo, EngD, MBAa,*, L. Pradelli, MDa, M. Carsi, MSca, M. Perachino, MDb a b
AdRes Health Economics & Outcomes Research, Torino, Italy Ospedale Santo Spirito, Casale Monferrato, Alessandria, Italy
A B S T R A C T
Keywords:
Objectives: Luteinizing hormone-releasing hormone (LHRH) agonists represent one of the
Cost-effectiveness
main cost factors in the management of patients with metastatic prostate cancer. We com-
Hormone therapy
pared the cost-effectiveness of the five different 3-month formulations of LHRH agonists
Prostate cancer
currently available for advanced prostate cancer in Italy, because these differ both in their
Testosterone
capacity to suppress testosterone and in their acquisition costs. Methods: A probabilistic, patient-level simulation model was developed to compare the cost-effectiveness, from the perspective of the Italian National Health Service (INHS), of leuprorelin 11.25 mg and 22.5 mg, triptorelin 11.25 mg, buserelin 9.9 mg, and goserelin 10.8 mg. The model incorporated testosterone-dependent progression-free and cancer-specific survival functions, LHRH agonist effectiveness data, and national costs and tariffs. Cox’s proportional hazard models were used to compute total and progression-free survival functions based on clinical data from 129 patients with metastatic prostate cancer treated in an Italian center. Bayesian random effects models were employed to summarize evidence from published literature on testosterone suppression obtained with the available LHRH agonists. Results: Estimated total survival was ⬇5 years, with a maximum difference between treatment options of ⬇2 months. There was a mean difference of almost €2,500 in lifetime total costs between the least costly option (leuprorelin 22.5 mg) and the most expensive (goserelin). In the incremental cost-effectiveness analysis, leuprorelin 22.5 mg dominated all alternatives except buserelin, which had an incremental cost-effectiveness ratio versus leuprorelin 22.5 mg of ⬇€12,000 per life-month gained. Conclusions: Based on modelling with meta-analysis of comparative survival data, leuprorelin 22.5 mg was the most cost-effective treatment of the available depot formulation LHRH agonists. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc.
Funding: L. Pradelli and Sergio Iannazzo are employees and owners of AdRes Srl, Torino, Italy, an independent health economics consulting and research company. Astellas Pharma Italia supported the development of the model through a consulting service contract with AdRes. M. Perachino cooperated with AdRes for the development of the model through a consulting agreement. The authors all contributed to the paper and approved it. A poster of this study was displayed (presentation code: PCN122) at the ISPOR 12th Annual European Congress, October 24 –27, 2009, Paris, France. * Address correspondence to: Sergio Iannazzo, AdRes Health Economics & Outcomes Research, P.zza Carlo Emanuele II, 19, I-10123 Torino, Italy. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2010.10.023
81
VALUE IN HEALTH 14 (2011) 80-89
Fig. 1 – Simplified structure of the model, represented as a decisional tree, with evaluated treatment options. In each branch, the simulation is conducted using a Markov chain. For simplicity, only the leuprorelin 22.5-mg branch is fully represented. pCancerDeath, probability of cancer-related death; pDeath, probability of death; pProgression, probability of disease progression.
Introduction Prostate carcinoma is the second most common tumor in Italy, second only to lung cancer, with an estimated incidence of 42,804 cases in 2005 (total male population of 28,376,804, of which 4,717,170 were older than 65 years of age), representing
17% of all diagnosed cancers. Furthermore, it is the third most common cancer-related cause of death, after lung and colorectal cancers. In 2005 there were 9070 deaths from prostate cancer in Italy, corresponding to 7% of total cancer mortality [1,2]. Although, to the best of our knowledge, no comprehensive cost-of-illness study is available for prostate cancer in Italy, a
Table 1 – Clinical parameters included in the cost-effectiveness model. Variable Mean baseline characteristics of the patient cohort PSA at 6 months, ng/ml Gleason score Age, years Cox’s proportional hazard model 1: coefficients of time to biochemical relapse Gleason score* Ln PSA at 6 months† Testosterone level at nadir2* Cox’s proportional hazard model 2: coefficients (SD) of cancer-specific survival after relapse Gleason score† Ln PSA at 6 months† Testosterone at nadir2†
LHRH agonist Meta-analysis results: nadir testosterone concentrations (ng/dl) Goserelin Leuprorelin 3.75/11.25 mg (Enantone®) Leuprorelin 7.5/22.5 mg (Eligard®) Buserelin Triptorelin Ln, logarithm; PSA, prostate-specific antigen; SD, standard deviation. * P ⬍ 0.05. † P ⬍ 0.01.
Mean or coefficient 185.8 7.0 74.6
SD 344.1 1.5 6.7
0.173 0.336 2.346
0.081 0.046 1.153
0.291 0.208 3.998
0.101 0.051 1.30
Mean
SD
18.41 13.45 9.09 8.48 10.72
6.95 3.02 2.86 3.97 4.85
82
VALUE IN HEALTH 14 (2011) 80-89
recent multinational study assessed the direct initial health care costs during the first year after diagnosis. For Italy, the estimate was approximately €5200 per patient, leading to an estimated annual cost of €202 million for newly diagnosed patients alone [3]. In cases of metastatic prostate cancer, luteinizing hormone-releasing hormone (LHRH) agonist monotherapy is the first-line treatment recommended by Italian guidelines, given that Italian patients generally refuse surgical castration and that there is no evidence that combined androgen blockade is superior [4].
The rationale for anti-androgen therapy in prostate cancer is the observation that testosterone represents an essential growth stimulus for prostatic cancer cells during much of the biological history of this tumor. LHRH agonists modulate the physiological feedback mechanism along the hypothalamicpituitary-gonadal axis, with consequent suppression of testicular testosterone production. The judgment on the optimum testosterone level to be reached for suppressing the growth of prostatic cancer cells has evolved alongside the detection ability of the assays used to determine serum testosterone concentrations: whereas the
Table 2 – Analysis of comparative efficacy of different LHRH agonists in lowering serum testosterone levels in patients with metastatic prostate cancer. Authors [reference]
LHRH agonist
N
Serum testosterone, ng/dl Mean Median
Waxman et al. [10] Chander et al. [25] Pettersson et al. [26] Kaisary et al. [11] Fernandez del Moral et al. [12] Fernandez del Moral et al. [12] Debruyne et al. [13] Debruyne et al. [13] Debruyne et al. [13] Debruyne et al. [13] Yri et al. [28] Fontana et al. [21] Fujii et al. [29] Sarosdy et al. [16] Fujii et al. [29] Crawford et al. [27] Oefelein [14] Oefelein [15] Oefelein and Cornum [17] Fowler et al. [18] Chu et al. [19] Perez-Marreno et al. [20] Sartor et al. [22] Heyns et al. [23] Heyns et al. [23] Teillac et al. [24] Teillac et al. [24] Yri et al. [28] Fujii et al. [29] Fujii et al. [29]
SD Min Max Percentile Percentile Cutoff 2.5% 97.5%
Buserelin‡ Buserelin† Buserelin‡ Goserelin* Goserelin*
4 69 21 148 83
6.2 13.6 5.5 36.3 17.3
NR NR 4.0 NR NR
NR NR 5.3 NR 11.2
0 NR 2.0 NR NR
15.0 NR 26.0 NR NR
NR NR 2 NR NR
NR NR 19 NR NR
Goserelin‡
77
17.9
NR
14.1
NR
NR
NR
NR
Goserelin*,# Goserelin‡,# Goserelin*,** Goserelin‡,** Goserelin‡ Goserelin‡ Goserelin* Goserelin‡ Goserelin‡ Leuprorelin (Eligard)㛳 Leuprorelin (Eligard)‡ Leuprorelin (Eligard)‡ Leuprorelin (Eligard)‡
38 42 39 41 25 120 50 59 74 103 13 32 38
15.3 19.4 19.2 18.8 NR 11.3 10.0 17.5 12.0 12.3 NR NR NR
NR NR NR NR 26.0 NR 7.2 33.9 10.5 NR 10.0 10.0 NR
NR NR NR NR NR NR NR NR NR 2.1 11.0 NR NR
NR NR NR NR 14.4 NR NR NR NR NR 1.0 10.0 NR
NR NR NR NR 63.5 NR NR NR NR NR 130.0 40.0 NR
NR NR NR NR NR NR 5.1 NR 5.1 NR 4 10 NR
NR NR NR NR NR NR 13.1 NR 14.9 NR 49 20 NR
NR NR NR NR 81 58 50 50 50 50–20 20 20 50–20
NR NR NR NR 100 100 100 100 98.6 99–88.3 92.0 NR 95–87
Leuprorelin (Eligard)¶ Leuprorelin (Eligard)‡ Leuprorelin (Eligard)*
80 111 118
9.3 10.1 6.1
NR NR 4.3
NR 0.1 NR
NR NR 3.0
NR NR 27.0
NR NR NR
NR NR NR
50 50–20 50–20
98.8 100–94 100–97.5
Leuprorelin (Eligard)§ Leuprorelin (Eligard)* Triptorelin* Triptorelin* Triptorelin‡ Leuprorelin (Enantone)‡ Leuprorelin (Enantone)* Leuprorelin (Enantone)‡
90 140 137 68 63 40 40 68
12.4 4.1 4.3 13.9 13.9 NR 11.0 12.0
NR NR NR NR NR 17.3 8.2 9.9
0.8 NR NR NR NR NR NR NR
NR 0 0 NR NR 14.4 NR NR
NR 10.3 8.3 NR NR 271.1 NR NR
NR NR NR 7.2 6.0 NR 5.0 5.9
NR NR NR 20.8 20.8 NR 13.8 14.8
50–20 50 50 50 50 81 50 50
98–90 97.3 98.8 96.0 98.0 90.0 95.0 98.5
Max, maximum; Min, minimum; NR, not reported; SD, standard deviation. * 1-month depot † 2-month depot ‡ 3-month depot § 4-month depot 㛳 6-month depot ¶ 1-year depot # Study I ** Study II
72.1 NR 50–20 NR 58
Patients within cutoff, %
58
100 NR 100–95 NR 100 98.8
83
VALUE IN HEALTH 14 (2011) 80-89
Table 3 – Pharmaceutical costs. LHRH agonist
Brand
Public price [24], €/package
Ex-factory price, €/package
Leuprorelin 11.25 mg
Enantone 3-month depot
460.14
278.80
Leuprorelin 22.5 mg
Eligard 3-month depot
387.49
234.78
Goserelin
Zoladex 3-month depot
592.35
358.91
Triptorelin
Decapeptyl 3-month depot
524.59
317.85
Buserelin
Suprefact 3-month depot
409.29
247.99
traditional threshold was set at 50 ng/dl (“castration level”), with the introduction of more sensitive assays, it was lowered to 20 ng/dl. Nowadays, immunoassays are able to detect much lower concentrations, allowing investigating more accurately the relationship between circulating testosterone levels and clinical outcomes in patients with metastatic prostatic cancer. Recently, a continuous relationship between the risk of cancer-related death and serum testosterone levels has been demonstrated [5]. Five different depot formulations of LHRH agonists are currently available in Italy, for administration either once monthly or every 3 months. A recent review suggests that these have differing capacities to lower testicular production of testosterone, but all induce high rates of “castration” levels (⬍50 ng/dl) [6]. Once-monthly formulations have a higher acquisition cost per milligram than the corresponding 3-month doses, which have been approved on the basis of noninferiority studies. An expert roundtable discussion concluded that “LHRH agonists should be considered first-choice testosterone-lowering therapy for the treatment of prostate cancer, with the 3-month depot formulation providing optimal convenience and flexibility” [7]. In order to establish the most cost-effective option among LHRH agonists, an economic analysis was performed to compare the 3-month options: leuprorelin 11.25 mg and 22.5 mg, triptorelin 11.25 mg, buserelin 9.9 mg, and goserelin 10.8 mg. The aim of this work was the pharmacoeconomic comparative evaluation of 3-month LHRH agonist treatments for metastatic prostate cancer, from the per-
Table 4 – Costs considered in the model. Item LHRH agonist ex-factory acquisition costs, €/package Leuprorelin 11.25 mg Leuprorelin 22.5 mg Goserelin Triptorelin Buserelin Follow-up examinations, €/year Medical examination plus DRE PSA, hematology, blood chemistry Bone densitometry Cardiology evaluation (visit plus ECG) Abdominal sonography Bone scintigraphy (total body) Therapy for the hormone-refractory phase, € Chemotherapy and adverse events management
Cost 278.80 234.78 358.91 317.85 247.99 41.32 82.76 31.50 32.28 120.86 226.20 9475.58
DRE, digital rectal examination; ECG, electrocardiogram; PSA, prostate-specific antigen.
spective of the Italian National Health Service (INHS). The evaluation was performed by developing a model to simulate testosterone levels obtained with these therapies and the correlation between these levels and the risk of cancerrelated death.
Methods Model description and transition probabilities Clinical data used to compute total and progression-free survival were extracted from the database of the Santo Spirito Hospital in Casale Monferrato, Alessandria, Italy. In this administrative database all parameters related to patients followed in the Department of Urology are prospectively recorded. At the time we performed the analysis, data from 129 patients with metastatic prostate cancer (M1) were available. An economic model was developed with the TreeAge Pro 2009 decision analysis software (TreeAge Software, Inc., Williamstown, MA), using a patient-level, microsimulation technique. Each branch of the decisional tree represents a therapeutic option, and all branches have identical structures (Fig. 1). In the probabilistic, patient-level simulation, each patient is defined with unique characteristics of initial age, prostate-specific antigen (PSA) blood concentration, and Gleason score. Their initial values are defined in order to reproduce the mean and standard deviation (SD) of the studied patient’s cohort (Table 1). Five identical “clones” are sent through the five treatment branches. It is assumed that the simulation starts as soon as the testosterone nadir is reached, so that for each patient this value is constant and totally dependent on the LHRH agonist treatment. The clinical evolution of each single patient is defined by the progression between health states in a simple Markov chain (states of alive and dead). Until biochemical relapse occurs, the patient is subjected only to natural mortality. After biochemical relapse, a specific cancer-related mortality is added to the general population mortality. Current modelling guidelines recommend that the “cycle length of the model should be short enough so that multiple changes in pathology, symptoms, treatment decisions, or costs within a single cycle are unlikely” [8]. Clinical outcomes data used (disease progression and death; see next section) were reported on a monthly basis, whereas other relevant timedependent parameters change more slowly (LHRH agonists were given every 3 months, with follow-up examinations performed every 6 months). Thus, we chose 1 month as the Markov cycle length.
84
VALUE IN HEALTH 14 (2011) 80-89
Table 5 – Tests and examinations costs. Test/Examination Medical examination plus DRE PSA and hematochemical examinations Bone densitometry Cardiology evaluation (visit plus ECG) Abdominal echography Bone scintigraphy (total body) Total
Frequency
Unit cost [25], €
Annual cost, €
Every 6 months Every 6 months Yearly Yearly Every 6 months Every 6 months
20.66 41.38 31.50 32.28 60.43 113.1
41.32 82.76 31.50 32.28 120.86 226.20 534.92
DRE, digital rectal examination; ECG, electrocardiogram; PSA, prostate-specific antigen.
cally significant. In both models, the assumption of proportional hazard for each included covariate was also verified. The two Cox models were incorporated into the simulation model. For each simulated patient, the probability of disease progression and of cancer-related death was calculated as a function of Gleason score, PSA, and testosterone levels. Natural mortality was factored as a competing risk as a function of age, and was derived from the mortality tables of the general Italian population [9].
Survival functions We derived the progression-free and total survival functions by analyzing the outcomes data of the patient cohort using Cox semiparametric models. The Cox regression model analyzes the influence of independent variables on event-free survival, under the assumption of proportional hazards, which means that the hazard ratio must be constant over time. We analyzed the data to find relationships between baseline characteristics and the time of biochemical relapse and between those parameters and the total survival time after relapse. The first Cox regression model was based on data from the cohort of 129 metastatic patients (M1) and evaluated the time to PSA elevation (biochemical relapse). The Gleason score, testosterone level at nadir, PSA concentration (on a logarithmic scale) 6 months after the start of LHRH agonist therapy, and the age were considered as potential covariates. Except for initial age, all were significantly associated with the time to progression (Table 1). The second Cox model was developed on a subset of 89 metastatic patients (M1) who had biochemical relapse, in order to evaluate cancer-related survival. The same set of covariates was tested for significant correlations, and initial age alone was again discarded because it was not statisti-
Treatment effectiveness To identify relevant publications for the estimate of the effectiveness of LHRH agonists in lowering testosterone levels, in May 2009 we performed a literature search in Medline using the keywords “Gonadotropin-Releasing Hormone” [Mesh] “Testosterone/blood” [Mesh], with the limits “Clinical trial” and “Humans.” Abstracts of the 276 articles identified by this strategy were checked for relevance by two of the authors (LP and MP), and full texts were obtained for 35 articles that were judged by at least one of the authors to possibly contain information regarding testosterone blood levels achieved. Manually searching the reference lists of these latter articles yielded one additional title. After examining the full text of these 36 papers, 20 studies conducted from 1990 to 2007 [10 –29] were
Table 6 – Base-case results: average (SD) effectiveness, costs, and cost-effectiveness of LHRH agonists in the treatment of metastatic prostate cancer.
Effectiveness, months Overall survival (SD) Progression-free survival* (SD) Costs, € Hormone therapy (SD) Follow-up (SD) Chemotherapy (SD) Total (SD) Cost-effectiveness, €/month lived Cost-effectiveness Incremental cost-effectiveness, €/month of life gained ICER vs. leuprorelin 22.5 mg
Leuprorelin 22.5 mg
Leuprorelin 11.25 mg
Goserelin
Triptorelin
Buserelin
60.31 (34.09) 40.49 (41.42)
59.49 (33.75) 39.86 (40.86)
57.93 (32.93) 38.40 (39.22)
59.87 (33.99) 40.19 (41.17)
60.35 (34.15) 40.57 (41.55)
4857 (2670) 2809 (1522) 6315 (3116) 13,981 (5150)
5977 (3139) 2772 (1507) 6364 (3101) 15,114 (5502)
7405 (3943) 2703 (1470) 6472 (3058) 16,579 (6133)
6804 (3604) 2789 (1517) 6341 (3108) 15,935 (5909)
5428 (2825) 2811 (1525) 6307 (3121) 14,546 (5277)
232
254
286
266
241
NA
Dominated
Dominated
Dominated
11,700
ICER, incremental cost-effectiveness ratio; NA, not applicable; SD standard deviation. * Undiscounted. Dominated: More costly and less effective than leuprorelin 22.5 mg.
85
VALUE IN HEALTH 14 (2011) 80-89
Table 7 – Average (SD) life expectancy and progression-free survival with the different considered treatments.
Months lived (SD) Months free from biochemical relapse* (SD)
Leuprorelin 22.5 mg
Leuprorelin 11.25 mg
Goserelin
Triptorelin
Buserelin
60.31 (34.09) 40.49 (41.42)
59.49 (33.75) 39.86 (40.86)
57.93 (32.93) 38.40 (39.22)
59.87 (33.99) 40.19 (41.17)
60.35 (34.15) 40.57 (41.55)
SD, standard deviation. * Not discounted.
identified as reporting detailed data on the subject, some of them previously identified in a recent review by Novara et al. [6]. These studies evaluated goserelin (n ⫽ 7), leuprorelin 7.5 or 22.5 mg (Eligard®; n ⫽ 9), leuprorelin 3.75 or 11.25 mg (Enantone®; n ⫽ 2), triptorelin (n ⫽ 2), and buserelin (n ⫽ 3). All data (mean, median, SD, minimum-maximum, and percentile values) relating to the distribution of testosterone values at nadir and the percentage of patients who reached the testosterone cutoff point (almost always 50 ng/dl) were extracted from these studies (Table 2) and meta-analyzed using a Bayesian random effects model. In a random effects model it is assumed that the effects observed in different trials for the same drug can be ascribed to a common effect, which is then influenced by peculiar conditions of each single study. The results of the meta-analysis, summarized in Table 1, were incorporated into the model to simulate the levels of testosterone reached at nadir for each LHRH agonist.
Costs Costs considered in the model include acquisition costs of LHRH agonists (Table 3)., based on ex-factory prices [30] and doses recommended in the summary of product characteristics (SPC), costs of monitoring tests and examinations, and costs of chemotherapy for the hormone-refractory phase in patients with biochemical relapse (Table 4). Consumption of follow-up procedures was modeled according to national guideline recommendations [4], and their costs were assessed based on the national price lists for ambulatory services, which also include diagnostic examinations (Table 5) [31]. The cost of managing toxicities related to LHRH agonist treatment was not included, assuming that this cost would not differ between strategies. Resources consumed during chemotherapy were based on data from the randomized, controlled trial on hormone-refractory prostate cancer patients reported by Tannock et al. [32]. Resources include drugs and administration of the chemotherapy regimen (docetaxel plus prednisolone, for an average of 9.5 cycles), premedication (corticosteroid plus antiemetic agent), and drugs used for managing grades III to IV
anemia and neutropenic fever (darbepoetin alfa 500 g every 3 weeks for the first 9 weeks and 300 g every 3 weeks subsequently, and pegfilgrastim 0.6 mg every 3 weeks, respectively, as recommended by the SPCs). The resulting total cost for treating the terminal phase was estimated at €9475.58 and was applied to all patients dying of cancer.
Discounting Italian guidelines for economic evaluations in health care issued in 2001 recommend applying a 3% discount rate to both future costs and outcomes [33]. The same value has been indicated in the more recent proposal of guidelines for economic evaluation elaborated and discussed in 2009 by the Italian Association of Health Economics (AIES) [34]. Presently, the European Commission advocates a 5% rate in its current cost-benefit guidance [35]. This rate is considered to be too high, however, by many economists [36], and the National Institute for the Social Impact of Economy (Istituto Nazionale per l’Impatto Sociale dell’Economia, INISE) recommends a discount rate approximately 1.5% lower than the official social discount rate [37]. Thus, we chose to apply an annual discount rate of 3.5% to both survival and costs.
Probabilistic sensitivity analysis The base case of the model is built to simulate the outcomes and costs for individual patients through the model. Results of the base case were calculated with 20,000 simulated patients. Characteristics of each patient (age, PSA blood concentration, and Gleason score) were extracted from distributions representing the mean (SD) from the studied cohort of 129 metastatic patients (M1) (Table 1). A probabilistic sensitivity analysis was performed to take into account the uncertainty surrounding estimated parameters of the two Cox models. This was accomplished by repeating 1000 cycles, each cycle with 5000 patient-level iterations. Each cycle was characterized by a single value for each of the parameters of the Cox models. These were randomly ex-
Table 8 – Average (SD) lifetime costs for each patient with the considered treatments. Cost type Hormone therapy (SD), € Follow-up (SD), € Chemotherapy (SD), € Total (SD), €
Leuprorelin 22.5 mg
Leuprorelin 11.25 mg
Goserelin
Triptorelin
Buserelin
4856.87 (2670.25) 2809.05 (1521.93) 6315.29 (3115.62) 13,981.21 (5150.10)
5977.07 (3138.54) 2772.40 (1506.50) 6364.41 (3101.18) 15,113.88 (5502.42)
7404.81 (3943.09) 2702.59 (1470.35) 6471.77 (3058.34) 16,579.18 (6133.08)
6804.14 (3603.69) 2789.34 (1517.25) 6341.11 (3108.41) 15,934.60 (5909.45)
5427.60 (2825.12) 2811.31 (1524.54) 6307.34 (3120.84) 14,546.25 (5277.24)
86
VALUE IN HEALTH 14 (2011) 80-89
Fig. 2 – Comparative cost-effectiveness of five LHRH agonist formulations in the treatment of metastatic prostate cancer. In this representation, the higher the slope, the more cost-effective the treatment: better clinical results are obtained at the same cost.
tracted from distributions representing the uncertainty of the estimate (Table 1).
cost-effectiveness ratio (ICER) of this drug (i.e., the cost required to obtain one additional unit of benefit) compared to leuprorelin 22.5 mg was about €12,000 per life-month gained (Table 9).
Results Probabilistic sensitivity analysis
Base case Our meta-analysis showed that the lowest testosterone nadirs were achieved with buserelin and leuprorelin 22.5 mg (⬇8 –9 ng/dl versus ⬇18 ng/dl with goserelin; Table 1). Estimated total survival for metastatic patients was approximately 5 years, with a maximum difference between treatment options of approximately 2 months (Table 6). The maximum difference between treatments for progression-free survival was almost 2 months (Table 7). The differences in the lifetime total costs of therapy were more pronounced, with a mean difference of nearly €2500 between the least costly option (leuprorelin 22.5 mg) and the most expensive (goserelin) (Tables 6 and 8). From the perspective of cost-effectiveness, leuprorelin 22.5 mg had the best cost/survival ratio, with a cost of approximately €230 per month (Table 6). In the incremental cost-effectiveness analysis (Fig. 2), leuprorelin 22.5 mg dominated leuprorelin 11.25 mg, goserelin, and triptorelin, having greater effectiveness and lower costs. Buserelin appeared more costly but slightly more effective than leuprorelin 22.5 mg. The incremental
Leuprorelin 22.5 mg was the least costly drug in all 1000 simulations, and, in terms of months of life gained, only buserelin was more effective in a certain percentage of iterations (Fig. 3). Therefore, leuprorelin 22.5 mg consistently dominated leuprorelin 11.25 mg, goserelin, and triptorelin. In Figure 3, the 1000 ICER values of buserelin versus leuprorelin 22.5 mg are represented as a scatterplot on a cost-effectiveness plane. Points are all found in the northeast and southeast quadrants. Points in the southeast quadrant represent a negative incremental survival gain of buserelin versus leuprorelin 22.5 mg, indicating that leuprorelin was superior to buserelin in terms of both costs and effectiveness (dominant). Buserelin never dominated leuprorelin, because its costs were always higher. ICER estimates of buserelin versus leuprorelin 22.5 mg were also plotted on a cost-effectiveness acceptability curve, in which the probability of being cost-effective is represented as a function of increasing levels of willingness to pay (WTP). As observed in Figure 4, there was a 100% chance of leuprorelin 22.5 mg being cost-effective when WTP was
Table 9 – Cost-effectiveness and incremental cost-effectiveness with respect to leuprorelin 22.5 mg, in the base case.
Cost-effectiveness, €/month lived ICER vs. leuprorelin 22.5 mg, €/month of life gained
Leuprorelin 22.5 mg
Leuprorelin 11.25 mg
Goserelin
Triptorelin
Buserelin
231.84
254.06 Dominated
286.20 Dominated
266.17 Dominated
241.01 11,700.22
ICER, incremental cost-effectiveness ratio. Dominated: more costly and less effective than leuprorelin 22.5 mg.
VALUE IN HEALTH 14 (2011) 80-89
Fig. 3 – Scatterplot of the 1000 incremental costeffectiveness ratio (ICER) estimates for buserelin versus leuprorelin 22.5 mg in the probabilistic sensitivity analysis. 〫 shows the mean value of the cost-effectiveness ratio. NE, northeast; SE, southeast. less than €28,000 per life-year gained. Buserelin was costeffective in more than 50% of cases only if the willingness to pay was greater than €280,000 per life-year gained.
87
In our analysis, estimated testosterone nadirs reached with different depot LHRH agonists were combined with survival functions obtained via Cox analysis of data from metastatic patients in order to compare their cost-effectiveness. Results obtained in the base case showed that there were some differences in mean overall survival and lifetime costs with the different treatments. Leuprorelin 22.5 mg on average dominated all alternative options except buserelin, being associated with lower costs and higher effectiveness. In a sensitivity analysis we explored the effect of the uncertainty on the main model parameters. The mean lower cost and higher effectiveness of leuprorelin 22.5 mg versus leuprorelin 11.25 mg, goserelin, and triptorelin was confirmed in all 1000 estimates. Buserelin, on average, was more costly but more effective than leuprorelin 22.5 mg, with an ICER of approximately €12,000 per life-month gained in the base case. An official WTP threshold for the INHS is lacking. To assess the value of a health intervention, results of pharmacoeconomic studies need, therefore, to be benchmarked against international or conventional thresholds. In the literature, there is a plethora of proposed cost-effectiveness thresholds, generally ranging from US$20,000/quality-adjusted life year (QALY) to the current £30,000/QALY applied in the United Kingdom [38 – 41]. The opportunity of raising this threshold is constantly under debate [42]; possibly the highest proposed cost-effectiveness threshold is the three times per capita gross domestic product (GDP) value suggested by the World
Discussion It is well known that the different available formulations of LHRH agonists induce different levels of testicular inhibition, but clinicians have not considered these differences to be clinically relevant, provided that the castration threshold is achieved. However, as an analysis of recently published clinical data suggests a continuous relationship between testosterone concentrations and disease progression, differences in the testosterone nadir achieved— even below the castration threshold—might determine variations in the rapidity of disease evolution and, therefore, in survival.
Health Organization (WHO) in the WHO-CHOICE program for efficient use of health resources [43]. In the probabilistic sensitivity analysis, leuprorelin 22.5 mg had an 80% probability of being cost-effective in comparison to buserelin even adopting this “extreme” threshold (equating to €78,000 based on Italy’s 2007 GDP per capita [44]). There is high variability in the differences in outcomes and costs associated with the evaluated treatments in the patientlevel simulations, which have distributions that are largely overlapping. In contrast, consistent differences in mean survival and costs were seen in the probabilistic sensitivity anal-
Fig. 4 – Incremental cost-effectiveness acceptability curve for buserelin versus leuprorelin 22.5 mg in the treatment of metastatic prostate cancer.
88
VALUE IN HEALTH 14 (2011) 80-89
ysis. We believe that this apparent discrepancy is related to the simulation technique that we used, in turn reflecting the different type of uncertainty that the two analyses are designed to reproduce. In the patient-level simulations, the variability reflects the interindividual heterogeneity and corresponds to the low predictability of the outcomes of the single patient. The much lower dispersion of mean estimates shown by the probabilistic sensitivity analysis, however, reflects the substantial low sensitivity of mean, cohort-level outcomes to the uncertainty in the parameters relating clinical characteristics and outcomes, as estimated by the survival functions. In other terms, the clinical effect of the treatments, and more generally the outcomes, of the single patient are very unpredictable. Indeed, it may well be possible that a patient treated with an LHRH agonist that is less effective than others on average does better than another similar patient treated with a drug that is on average more effective, whereas in a cohort of patients the difference will emerge. These considerations further imply that the present results are relevant for decision makers responsible for choices and budgets regarding cohorts of patients—from the head of the department and hospital drug purchaser to the policy makers at regional and national levels—while having little or no importance for the clinician deciding about the care of a single patient.
costs are common to all treatments and the maximal mean survival difference between treatments is approximately 2 months, so we are confident that the potential of these estimates to bias the final results is very limited.
Conclusions Despite some limitations, our analysis represents the first attempt to summarize the available evidence regarding the relative effectiveness of testosterone-lowering therapies in metastatic prostate cancer and their consequences on hard outcomes into one comprehensive conceptual scheme. The results suggest that leuprorelin 22.5 mg is the most cost-effective of the available depot LHRH agonists in the Italian setting. Although only tentative conclusions can be drawn, based on modeling with meta-analysis of comparative survival data, our findings should provide valuable guidance for decision makers who need to make rational choices regarding the use and price of LHRH agonists in Italian patients with metastatic prostate cancer —like the hospital pharmacist, the head of the urology department, or the regional agency deciding on hospital drug formularies.
Limitations
Acknowledgment
The model relationships between testosterone levels and cancer mortality and progression are based on the analysis of data from one clinical center. The use of a continuous analytical function even for very low levels of the hormone implies the absence of a threshold below which testosterone stops stimulating cancer cell growth. Such a threshold is referred to as castration level in earlier literature and its non existence might need to be further investigated and confirmed by larger clinical studies. The retrieval of the evidence on testosterone nadirs achieved with the available treatments might have been more formally rigorous, although we believe that we have not omitted any crucial information. The findings of the review by Novara et al. [6] support this belief. Regarding the cost side of the evaluation, some simplifications and approximations have been used. The national price list used for estimating the costs of follow-up diagnostic procedures and examinations dates back to 1996; therefore, costs are likely underestimated. The updating of price lists is the responsibility of the regions, but many of them have not complied with this duty to date. For example, for the most expensive of the diagnostic procedures considered in the present model (i.e., total body bone scintigraphy), only 10 of the current 20 regional price lists have an updated value. The reported mean of regional prices is of limited representativeness, because it neglects any weighting based on activity volumes or target population. It is, however, higher than the old national price list [45]. The consequent underestimation of follow-up costs in the model may have slightly biased the results in favor of the strategies associated with longer survival. The cost of chemotherapy for the hormone-refractory phase used in the calculations is also an aggregate estimate that may not reflect costs accurately. Nevertheless, all of these
Editorial support was provided by Julia Balfour, Consultant Medical Writer, Kilconquhar, United Kingdom. REFERENCES
[1] Grande E, Inghelmann R, Francisci S, et al. Regional estimates of all cancer malignancies in Italy. Tumori 2007; 93:345–51. [2] Inghelmann R, Grande E, Francisci S, et al. Regional estimates of prostate cancer burden in Italy. Tumori 2007;93:380 – 6. [3] Fourcade RO, Benedict A, Black LK, et al. Treatment costs of prostate cancer in the first year after diagnosis: a short-term cost of illness study for France, Germany, Italy, Spain and the UK. BJU Int 2010;105:49 –56. [4] AURO.it (Associazione Urologi Italiani—Association of Italian Urologists) AUROLINE 10: Guidelines on prostate cancer: diagnosis, staging and therapy, 2008. Available from: http://www.auro.it/wp-content/uploads/al10.pdf [Accessed May 4, 2010]. [5] Perachino M, Cavalli V, Bravi F. Testosterone levels in patients with metastatic prostate cancer treated with luteinizing hormone-releasing hormone therapy: prognostic significance? BJU Int 2010;105:648 –51. [6] Novara G, Galfano A, Secco S, et al. Impact of surgical and medical castration on serum testosterone level in prostate cancer patients. Urol Int 2009,82:249 –55. [7] de Jong IJ, Eaton A, Bladou F. LHRH agonists in prostate cancer: frequency of treatment, serum testosterone measurement and castrate level: consensus opinion from a roundtable discussion. Curr Med Res Opin 2007;23:1077– 80. [8] Weinstein MC, O’Brien B, Hornberger J, et al. Principles of good practice of decision analytic modeling in health care evaluation: Report of the ISPOR Task Force on Good Research Practices—Modeling Studies. Value Health 2003;6: 9 –17.
VALUE IN HEALTH 14 (2011) 80-89
[9] ISTAT (Italian National Institute of Statistics) Table 2.16: Mortality Table by Age and Sex, 2005. Annuario Statistico Italiano, ISTAT, 2008. [10] Waxman J, Sandow J, Abel P, et al. Three-monthly GnRH agonist (buserelin) for prostatic cancer. Br J Urol 1990,65:43–5. [11] Kaisary AV, Tyrrell CJ, Peeling WB, Griffiths K. Comparison of LHRH analogue (Zoladex) with orchiectomy in patients with metastatic prostatic carcinoma. Br J Urol 1991;67:502– 8. [12] Fernandez del Moral P, Dijkman GA, Debruyne FM, et al. Three-month depot of goserelin acetate: clinical efficacy and endocrine profile. Dutch South East Cooperative Urological Group. Urology 1996;48:894 –900. [13] Debruyne FM, Dijkman GA, Lee DC, et al. A new long acting formulation of the luteinizing hormone-releasing hormone analogue goserelin: results of studies in prostate cancer. J Urol 1996;155:1352– 4. [14] Oefelein MG. Time to normalization of serum testosterone after 3-month luteinizing hormone-releasing hormone agonist administered in the neoadjuvant setting: implications for dosing schedule and neoadjuvant study consideration. J Urol 1998;160:1685– 8. [15] Oefelein MG: Serum testosterone-based luteinizing hormone-releasing hormone agonist redosing schedule for chronic androgen ablation: a phase I assessment. Urology 1999;54:694 –9. [16] Sarosdy MF, Schellhammer PF, Soloway MS, et al. Endocrine effects, efficacy and tolerability of a 10.8-mg depot formulation of goserelin acetate administered every 13 weeks to patients with advanced prostate cancer. BJU Int 1999;83:801– 06. [17] Oefelein MG, Cornum R. Failure to achieve castrate levels of testosterone during luteinizing hormone releasing hormone agonist therapy: the case for monitoring serum testosterone and a treatment decision algorithm. J Urol 2000;164:726 –9. [18] Fowler JE, Flanagan M, Gleason DM, et al. Evaluation of an implant that delivers leuprolide for 1 year for the palliative treatment of prostate cancer. Urology 2000;55:639 – 42. [19] Chu FM, Jayson M, Dineen MK, et al. A clinical study of 22.5 mg. La-2550: a new subcutaneous depot delivery system for leuprolide acetate for the treatment of prostate cancer. J Urol 2002;168:1199 –203. [20] Perez-Marreno R, Chu FM, Gleason D, et al. A six-month, open-label study assessing a new formulation of leuprolide 7.5 mg for suppression of testosterone in patients with prostate cancer. Clin Ther 2002;24:1902–14. [21] Fontana D, Mari M, Martinelli A, et al. 3-month formulation of goserelin acetate (“Zoladex” 10.8-mg depot) in advanced prostate cancer: results from an Italian, open, multicenter trial. Urol Int 2003;70:316 –20. [22] Sartor O, Dineen MK, Perez-Marreno R, et al. An eight-month clinical study of LA-2575 30.0 mg: a new 4-month, subcutaneous delivery system for leuprolide acetate in the treatment of prostate cancer. Urology 2003;62:319–23. [23] Heyns CF, Simonin M, Grosgurin P, et al. Comparative efficacy of triptorelin pamoate and leuprolide acetate in men with advanced prostate cancer. BJU Int 2003;92:226 –31. [24] Teillac P, Heyns CF, Kaisary AV, et al. Pharmacodynamic equivalence of a decapeptyl 3-month SR formulation with the 28-day SR formulation in patients with advanced prostate cancer. Horm Res 2004;62:252– 8. [25] Chander S, Choo R, Danjoux C, et al. Effect of androgen suppression on hemoglobin in prostate cancer patients undergoing salvage radiotherapy plus 2-year buserelin acetate for rising PSA after surgery. Int J Radiat Oncol Biol Phys 2005;62:719 –24. [26] Pettersson B, Varenhorst E, Petas A, Sandow J. Duration of testosterone suppression after a 9.45 mg implant of the GnRH-analogue buserelin in patients with localised
[27]
[28]
[29]
[30] [31] [32]
[33]
[34] [35]
[36]
[37]
[38]
[39]
[40] [41] [42] [43]
[44]
[45]
89
carcinoma of the prostate: a 12-month follow-up study. Eur Urol 2006;50:483–9. Crawford ED, Sartor O, Chu F, et al. A 12-month clinical study of LA-2585 (45.0 mg): a new 6-month subcutaneous delivery system for leuprolide acetate for the treatment of prostate cancer. J Urol 2006;175:533– 6. Yri OE, Bjoro T, Fossa SD. Failure to achieve castration levels in patients using leuprolide acetate in locally advanced prostate cancer. Eur Urol 2006;49:54 – 8. Fujii Y, Yonese J, Kawakami S, et al. Equivalent and sufficient effects of leuprolide acetate and goserelin acetate to suppress serum testosterone levels in patients with prostate cancer. BJU Int 2008;101:1096 –100. Italian Drug Formulary: L’Informatore Farmaceutico, 69th ed. Italy: Elsevier Masson, 2009. Italian Specialist Tariffs: Nomenclatore Tariffario Nazionale, D.M. 22, July 1996. Tannock IF, de Wit R, Berry WR, et al. Docetaxel plus prednisone or mitoxantrone plus prednisone for advanced prostate cancer. N Engl J Med 2004;351:1502–12. Capri S, Ceci A, Terranova L, et al. Guidelines for economic evaluations in Italy: recommendations from the Italian group of pharmacoeconomic studies. Drug Inf J 2001;35:189–201. Guide to Cost-Benefit Analysis of Investment Projects. Evaluation Unit, DG Regional Policy, European Commission, 2002. Italian Health Economics Association (AIES). A proposal for guidelines for the economic evaluation of health interventions in Italy. PharmacoEconomics Italian Research Articles 2009;11: 83–93. Evans D. Social discount rate for the European Union: new estimates. In: Florio M, ed. Cost-Benefit Analysis and Incentives in Evaluation: The Structural Funds of the European Union. Edward Elgar Publishing, 2007. Istituto Nazionale per l’Impatto Sociale dell’Economia (INISE). L’analisi costi-benefici. Available from: http://www.bilanciosociale.info/public/documenti/ Libro/capitolo%204.doc [Accessed May 2010]. Laupacis A, Feeny D, Detsky AS, Tugwell PX. How attractive does a new technology have to be to warrant adoption and utilization? Tentative guidelines for using clinical and economic evaluations. CMAJ 1992;146:473– 81. Tosteson AN. Quality of life in the economic evaluation of osteoporosis prevention and treatment. Spine 1997;22(24 Suppl.):58S– 62S. Kanis JA, Jonsson B. Economic evaluation of interventions for osteoporosis. Osteoporos Int 2002;13:765–7. Raftery J. NICE: faster access to modern treatments? Analysis of guidance on health technologies. BMJ 2001;323:1300–3. Towse A, Raftery J. Should NICE’s threshold range for cost per QALY be raised? BMJ 2009;338:268 –9. World Health Organization. Commission on Macroeconomics and Health Report. Macroeconomics and Health: Investing in Health for Economic Development. Geneva, 2001. Available from: http://whqlibdoc.who.int/publications/2001/924154550x.pdf [Accessed November 2010]. ISTAT (Italian National Institute of Statistics) Historical Series of Per Capita Gross Domestic Product (GDP) (1970 –2007). Available from: http://www.istat.it/conti/ [Accessed March 2010]. Agenzia Nazionale per i Servizi Sanitari Regionali. Tariffe delle prestazioni specialistiche ambulatoriali aggiornato al 31 dicembre 2008. Nomenclatore delle Prestazioni specialistiche ambulatoriali— confronto tra le tariffe nazionali ex DM 1996 e le tariffe regionali relative all’anno 2008. Available from: http://www.agenas.it/ [Accessed February 2010].