|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 26, No 6 (February 20), 2008: pp. 925-933 © 2008 American Society of Clinical Oncology. DOI: 10.1200/JCO.2006.10.4190 Docetaxel in Combination With Doxorubicin and Cyclophosphamide As Adjuvant Treatment for Early Node-Positive Breast Cancer: A Cost-Effectiveness and Cost-Utility Analysis
From RTI-Health Solutions, Manchester; Department of Oncology, Western General Hospital, Edinburgh; New House Farm, Purton End, Debden, Saffron Walden Essex, UK; and Prescribing Research Group, The University of Liverpool Management School, Liverpool, United Kingdom Corresponding author: Sorrel E. Wolowacz, PhD, RTI-Health Solutions, Williams House, University of Manchester Science Park, Manchester M15 6SE, United Kingdom; e-mail: swolowacz{at}rti.org
Purpose To estimate the cost effectiveness of TAC (docetaxel, doxorubicin, and cyclophosphamide) compared with FAC (fluorouracil, doxorubicin, and cyclophosphamide) when administered as adjuvant therapy to women with node-positive early breast cancer in the United Kingdom (UK), both with and without primary prophylaxis with granulocyte colony-stimulating factor (G-CSF). Methods A standard health economic Markov model estimated the cost and outcome for node-positive early breast cancer patients, from initiation of adjuvant chemotherapy to death. Patient-level data were used from the Breast Cancer International Research Group (BCIRG) 001 trial for estimates of the effect of chemotherapy on toxicity and outcome, and an observational data set collected from a UK university hospital provided estimates of resource use and outcome for patients with relapsed disease. Results Over a 10-year analysis timeframe, the incremental cost per life-year saved associated with the use of TAC rather than FAC was estimated as £15,418 (95% CI, £13,734 to £17,997) and the incremental cost per quality-adjusted life-year gained (IC/QALY) was £18,188 (95% CI, £14,161 to £32,422). The addition of primary G-CSF (lenograstim or filgrastim) to the TAC regimen resulted in an IC/QALY of £20,432. The results were most sensitive to the quality-of-life (QOL) score for patients in remission postchemotherapy. However, even if QOL was assumed to be as poor as for patients with metastatic disease, the IC/QALY estimate rose only to £32,430. Conclusion The use of adjuvant TAC rather than FAC for node-positive early breast cancer patients is cost effective, despite the increased drug and toxicity treatment costs, and when primary G-CSF prophylaxis is given to all patients.
Docetaxel is an antineoplastic agent that binds beta tubulin, resulting in stabilization of microtubules, cell-cycle arrest, and cell death.1 Docetaxel has an extensive history of use since its approval in the mid-1990s in the treatment of locally advanced or metastatic breast cancer, non–small-cell lung cancer, and hormone-refractory prostate cancer.
Recently, docetaxel has been investigated as an adjuvant treatment for early breast cancer (EBC) in combination with anthracycline-containing regimens. Two phase III trials have demonstrated significant relative reductions in the risk of relapse and death of between 17% and 30% in comparison to robust anthracycline-containing regimens. In trial BCIRG001,2 six cycles of docetaxel in combination with doxorubicin and cyclophosphamide (TAC) was significantly more effective than six cycles of fluorouracil in combination with doxorubicin and cyclophosphamide (FAC; hazard ratio [HR] for relapse = 0.72, P = .001; HR for death = 0.70, P = .008; median follow-up, 55 months). In trial PACS 01,3 three cycles of fluorouracil, epirubicin, and cyclophosphamide (FEC) followed by three cycles of docetaxel (FEC
The superior survival outcome expected from docetaxel-containing regimens is, however, acquired at an additional cost. At 2005 values, the drug acquisition cost for TAC in the United Kingdom (UK) is £1,196 per cycle (
We present an economic analysis that estimates the cost effectiveness of TAC versus FAC as adjuvant therapy for node-positive EBC. The analysis takes the perspective of the UK National Health Service (NHS) for the cost year 2005 and estimates the incremental cost per life-year saved (IC/LYS) and the incremental cost per quality-adjusted life-year gained (IC/QALY) for TAC versus FAC. The implications of adverse events in terms of cost and HRQoL, and the impact of providing all patients with primary G-CSF prophylaxis to prevent neutropenia, are explored.
Model A cohort of 1,000 women with node-positive EBC (median age, 49 years; 55% premenopausal; 76% estrogen or progesterone receptor positive), who were disease free after locoregional surgery, were entered into a Markov model (Fig 1). Patients remained in the remission health state unless they had a relapse (locoregional or distant) or died as a result of other causes. The number of patients exiting the remission state at the end of each monthly cycle was governed by time-dependent transition probabilities of relapse or death resulting from other causes. Patients remaining in the remission state were attributed the survival and quality-adjusted survival for that month, and the cost of any monitoring in that month.
Because there are no data suggesting that outcomes after first relapse are affected by the adjuvant chemotherapy received (confirmed for survival outcome in patients receiving TAC or FAC in BCIRG001 [treatment group did not have a significant effect on survival from locoregional recurrence (log-rank P = .48) or distant recurrence (log-rank P = .83), Sanofi-aventis data on file]), the analysis of events postrelapse was simplified to a series of three pay-offs: the total expected survival, QALYs, and cost postrelapse. These represented the average experience of relapsing patients, including those achieving long-term remission after first locoregional recurrence and those developing further locoregional and/or metastatic disease. A decision tree was used to estimate the costs of adjuvant chemotherapy, and the impact of adverse events on costs and HRQoL, which allowed the proportion of patients receiving primary G-CSF prophylaxis, experiencing adverse events, and discontinuing therapy as a result of adverse events to be varied. Patients discontinuing as a result of adverse events received fewer cycles of the planned regimen and appropriate supplementary adjuvant chemotherapy (FAC or CMF [cyclophosphamide, methotrexate and fluorouracil]). The time horizon of the analysis was varied from 5 years to 40 years. The primary data source was trial BCIRG001,2 in which women with node-positive EBC were randomly assigned to TAC (n = 745) or FAC (n = 746) and followed up for a median of 55 months. The primary end point was disease-free survival (DFS). Patient-level data were analyzed to derive estimates of DFS and overall survival, chemotherapy usage, and adverse event probabilities. An observational data set provided by the Western General Hospital (WGH; Edinburgh, UK), consisting of women with node-positive EBC who had received adjuvant chemotherapy and subsequently relapsed, was analyzed to estimate health care resource use and outcomes postrelapse. Key model parameters are summarized in Tables 1 and 2. Probabilistic and univariate sensitivity analyses were performed (described in the Appendix, online only). Details of probabilistic parameter estimation are presented in Tables 1 and 2. Costs and outcomes were discounted at 3.5% per annum.
Cost of Adjuvant Chemotherapy and Support Drug costs assumed an average body surface area4 of 1.7 m2, an average weight of 70 kg and that unused drug in opened vials is discarded (Table 1). Administration costs were adapted from a NHS report.5 The number of cycles was estimated by treatment group from drug use in trial BCIRG001. Although primary prophylaxis with G-CSF was not permitted in BCIRG001, a small proportion of patients received it (7.0% and 7.7% in the TAC and FAC arms, respectively). Because the adverse events recorded reflect this use of primary prophylaxis, the cost was included. The impact of assuming that all TAC patients received primary prophylaxis was investigated in sensitivity analysis. The cost of adjuvant hormone and radiotherapy was excluded because they were used equally in the TAC and FAC arms of BCIRG001 (68.4% and 68.9% of TAC and FAC patients received hormone, respectively; 68.8% and 71.9% received radiotherapy, respectively).
Adverse Events Probabilities were derived from trial BCIRG001. Febrile neutropenia and neutropenic infection were combined because the impact on costs and HRQoL are similar. For analyses in which TAC patients received primary prophylaxis with G-CSF, the probability of febrile neutropenia was assumed to be 7.5% for lenograstim or filgrastim (from GEICAM98056) and 1% for pegfilgrastim (from Vogel et al7). Cost estimates (Table 2) were derived from the published literature8-10 with the exception of stomatitis, which was based on clinical opinion (from five UK consultant oncologists). Utility decrements were derived from the published literature.11-13 The duration of HRQoL impairment was estimated from published data and clinical opinion (five UK consultant oncologists). G-CSF use was estimated from trial BCIRG001. Costs assumed 50% of patients receive lenograstim, and 50% receive filgrastim.
Survival Estimates
Survival postrelapse. Mean overall survival postrelapse was estimated from the BCIRG001 patient-level data pooled for both treatment arms using Kaplan-Meier survival analysis (Table 1). The uncertainty in these estimates was explored in sensitivity analysis. All-cause mortality. The probability of death resulting from other causes was estimated from age-specific mortality rates for women.15 Probabilities were calculated for the age distribution of the BCIRG001 population, and adjusted as patients aged in subsequent model cycles.
QoL Utility weights for health states postrelapse were obtained from the literature.17 QALYs postrelapse were estimated by combining these utility weights with the probability of experiencing each health state and the mean time spent in each health state, estimated from the WGH data set.
Costs of Monitoring and Care Postrelapse The NICE Final Appraisal Determination was issued on July 24, 2006, and the Scottish Medicines Consortium advice was issued on September 9, 2005. In both cases, docetaxel was recommended for use in the NHS in combination with doxorubicin and cyclophosphamide as an option for the adjuvant treatment of early (operable), node-positive breast cancer. The mean cost of hospital care postrelapse was estimated from patient-level resource use data provided by the WGH (Appendix),20 including diagnostics, out-patient visits, day case procedures, in-patient stay, surgery, chemotherapy, radiotherapy, and hormonal treatments. Unit costs from national sources22,23 were applied to the patient-level data to calculate total costs for individual patients, and mean costs were estimated by bootstrapping (1,000 simulations with replacement). The cost of palliative care in the community was estimated from published data.21 The cost of hospice care was excluded because it is not funded by the NHS. Unit costs were obtained from national sources22,23 and inflated to 2005 values where necessary.23
Base-Case Analysis The base-case results are presented in Table 3. Over a timeframe of 10 years, patients receiving TAC benefited from an average additional 0.37 years of life, and 0.32 QALYs compared with those receiving FAC. The impact of grade 3 or 4 adverse events on HRQoL was estimated as an average loss per patient of 2.8 quality-adjusted life days in the TAC arm compared with 1.4 in the FAC arm.
The mean number of cycles of chemotherapy received was 5.75 for TAC and 5.91 for FAC. The estimated mean cost of chemotherapy and administration was £7,190 in the TAC arm and £1,263 in the FAC arm; a difference between arms of £5,927. The average additional cost of managing adverse events and supportive G-CSF in the TAC arm were £514 and £609 respectively. The average cost of treatments for disease relapse was £1,322 less in the TAC arm than in the FAC arm. The cost effectiveness of TAC versus FAC was estimated as £15,418/LYS and £18,188/QALY.
Sensitivity and Subgroup Analysis The results of other sensitivity analyses are summarized in Table 4. The timeframe of the analysis had the greatest impact on the results. A timeframe of 5 years yielded an estimate of £58,201/QALY, whereas the lifetime analysis longer than 40 years (on completion of which all patients had died) yielded an estimate of £9,865 (95% CI, £7,864 to £15,891). The various methods of extrapolation of outcomes beyond the end of trial follow-up resulted in estimates varying from £15,588 to £28,782/QALY. The lowest estimates were generated by the treatment-specific loglogistic model, which assumed continuation of treatment effect beyond trial follow-up, and the highest by the pooled loglogistic model which assumed no continuation of treatment effect.
The sensitivity analysis also explored the costs applied for the management of adverse events as well as their impact on HRQoL (Table 4). Increasing the cost of all events occurring at a higher rate in the TAC arm (including febrile neutropenia) to £3,000 per episode resulted in a modest increase in the IC/QALY from £18,188 to £20,754. Increasing the impact of events on HRQoL resulted in estimates of less than £19,000/QALY. The sensitivity of the base-case analysis to other input parameter estimates was investigated by varying all parameters by ± 50% of their base-case value (including the cost of chemotherapy). The results were most sensitive to the value of the utility weight for patients in remission. If the estimate were reduced to 0.5 (similar to estimates reported for metastatic disease)17 the IC/QALY increased to £32,431. If all TAC patients were assumed to receive primary G-CSF prophylaxis (with filgrastim or lenograstim), the total cost of supportive G-CSF was estimated as £1,966 (including primary and secondary prophylaxis and treatment of febrile neutropenia). The cost of adverse events was reduced from £1,014 to £402 per patient as a result of a reduction in the incidence of febrile neutropenia. The number of patients discontinuing chemotherapy because of adverse events was reduced, resulting in an increase in the mean cost of chemotherapy and administration. The overall impact on the IC/QALY estimates was a modest increase from £18,188 to £20,432. If primary prophylaxis was with pegfilgrastim, the average cost of G-CSF was £4,176 and the IC/QALY was £57,320 (95% CI, £34,806 to £113,676). Subgroup analysis suggests that TAC is more cost effective in patients who are younger, estrogen-receptor–negative, and with fewer positive nodes and lower tumor grades (Table 4; Appendix).
We can conclude that adjuvant TAC is cost effective compared with FAC in the UK. The IC/QALY over a 10-year timeframe was £18,188 (95% CI, £14,161 to £32,422 [
These estimates are comparable with recent estimates for other adjuvant treatments for EBC for the UK (using lifetime analysis timeframes), for example for trastuzumab (£18,000/QALY),24 anastrozole (£12,310/QALY)25 and letrozole (£9,325/QALY),25 and is more favorable than for estimates for gemcitabine combination therapy for advanced breast cancer (£42,800/QALY).26
TAC supported by primary G-CSF prophylaxis was also cost effective. The IC/QALY for TAC supported by primary lenograstim or filgrastim was estimated as £20,432 (
Although FAC is rarely used in the UK, we believe it provides a reasonable surrogate for the commonly used regimen in the UK, FEC (usually administered at epirubicin doses of 60 to 75 mg/m2 in the UK).27 Direct comparison of the two regimens in metastatic breast cancer has demonstrated equivalence.28,29 In addition, the recent Oxford overview30 performed an indirect comparison of FAC and FEC in EBC and concluded that these two most widely studied regimens "appear to be of comparable efficacy." This analysis is not applicable to TAC compared with FEC at higher epirubicin doses (more than 75 mg/m2). The primary uncertainty in the analysis is the extrapolation of outcomes beyond the available trial follow-up. Assuming the same risk of relapse in both FAC and TAC cohorts after trial end resulted in a gradual convergence of the DFS curves beyond 5 years (Fig 2B). Since the recent Oxford overview30 suggests that DFS curves for alternative adjuvant regimens continue to diverge until at least 10 years post–random assignment, it is likely that this analysis provides a conservative estimate of the long-term benefit of TAC. Using several alternative methods of estimating the long-term risk of relapse, the IC/QALY varied from £15,588 to £28,782. Thus, the considerable uncertainty inherent in the extrapolation is unlikely to alter the cost effectiveness of TAC over FAC. In univariate sensitivity analysis, the results were most sensitive to the utility weight for remission. Reducing the value to 0.5 (similar to estimates reported for metastatic disease)17 increased the IC/QALY estimate to £32,430. We believe a utility of 0.5 to be implausible because the HRQoL of EBC patients in remission after their chemotherapy has been demonstrated to return to levels similar to that of the general population.31 Truncation of the model timeframe to 5 years yielded an estimate of £58,201. This analysis effectively assumes that there are no differences in costs and benefits between the patient cohorts after 5 years, which is inconsistent with long-term follow-up data for other chemotherapy regimens.30 The analysis provides useful insight into the impact of serious adverse events associated with TAC on costs and HRQoL. The average cost of managing adverse events, including G-CSF, was £1,123 more per patient receiving TAC than for patients receiving FAC, whereas the impact on HRQoL of grade 3 and 4 adverse events was estimated as an average loss of only 1.4 quality-adjusted life days. Although the utility decrements applied to the periods of time that patients experience adverse events were large (0.35 to 0.38, placing them at a lower utility value than distant disease), the difference in HRQoL experienced by patients receiving TAC and FAC is small when averaged over the whole population, because QoL is impaired by serious adverse events for a relatively short time and for a small additional number of patients.
These short-term disadvantages of TAC are small compared with the long-term benefits to be realized from the lower risk of relapse with this regimen. Although the incidence of febrile neutropenia was high in trial BCIRG001, where primary G-CSF prophylaxis was not permitted, there were no septic deaths and fewer noncancer deaths than have been reported for alternative effective adjuvant regimens such as E
Although all authors completed the disclosure declaration, the following authors or their immediate family members indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. For a detailed description of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors. Employment: N/A Leadership: N/A Consultant: Sorrel E. Wolowacz, Sanofi-aventis; David A. Cameron, Sanofi-aventis; Helen C. Tate, Sanofi-aventis; Adrian Bagust, Sanofi-aventis Stock: N/A Honoraria: David A. Cameron, Sanofi-aventis; Adrian Bagust, Sanofi-aventis Research Funds: Sorrel E. Wolowacz, Sanofi-aventis; David A. Cameron, Sanofi-aventis Testimony: N/A Other: N/A
Conception and design: Sorrel E. Wolowacz, David A. Cameron Provision of study materials or patients: David A. Cameron Collection and assembly of data: Sorrel E. Wolowacz, Helen C. Tate Data analysis and interpretation: Sorrel E. Wolowacz, David A. Cameron, Helen C. Tate, Adrian Bagust Manuscript writing: Sorrel E. Wolowacz, David A. Cameron, Helen C. Tate Final approval of manuscript: Sorrel E. Wolowacz, David A. Cameron, Helen C. Tate, Adrian Bagust
Disease-Free Survival Estimates To maximize the power of trial data and optimize the efficiency of economic models, survival data used in economic analyses are often fitted to a parametric model that describes the hazard of relapse with time and its dependence on each of the risk factors of interest. Our analysis of the patient-level data of trial BCIRG001 stepped through a series of parametric models, starting with simple Weibull models, through loglogistic models, and finally to a partitioned model to establish a model that closely predicted the risk of relapse observed in BCIRG001. Because the hazard of relapse is well known to differ for patients with differing clinical characteristics (for example by node status [Early Breast Cancer Trialists’ Collaborative Group: Lancet 365:1687-1717, 2007] and hormone receptor status[Berry DA, Cirrincione C, Henderson IC, et al: JAMA295:1658-1667, 2006 (Erratum: JAMA 295:2356, 2006)]), key clinical characteristics were included as covariates in the survival analysis (node status, hormone-receptor status, tumor grade, and age). Weibull models demonstrated a poor fit to the relapse event (Kaplan-Meier) data. An example of this is shown for the survival to distant relapse in the FAC arm in Figure A1. The Kaplan-Meier DFS distribution function exhibited inflection points, and a "reverse S" shape, indicating that early in the follow-up, the hazard is increasing, reaching a peak at around 20 months, after which it decreases again. The Weibull function is able to describe only monotonic changes in hazard with time (ie, an increasing rate of change or a decreasing rate of change in hazard), and cannot describe the peak in hazard that the data seem to exhibit. We therefore explored the loglogistic function, which is able to describe a hazard that increases to a peak and then decreases. However, loglogistic models also demonstrated a poor fit to the data, as shown in Figure A2). Because neither of these simple parametric approaches were able to achieve an acceptable fit to the data, we explored the possibility that the trial population may represent a mixed population exhibiting different hazards of relapse, and that a lag or event-free period exists early in the follow-up. Because node status, hormone-receptor status, tumor grade, and age were accounted for in our analyses, differing hazards of relapse would be attributable to some other unknown prognostic factor or factors. The lag may result from the fact that all patients entering the trial were clear of obvious evidence of distant metastases by conventional staging techniques. Subsequent relapse will have been mostly diagnosed by symptomatic presentation, and it will have taken time for subclinical metastases to have grown sufficiently to present with symptoms. A partitioned model in which relapse in a subgroup of patients at higher risk was described by a loglogistic function was fitted to the data; remaining patients were subject to a lower risk of relapse described by an exponential function, and a short event-free period was imposed on both groups immediately after random assignment. The long-term exponential function was constrained to be equivalent in both treatment arms, so that the hazard beyond trial follow-up was the same in both treatment cohorts. The model parameters and survival function are presented in Table A1, and the model fit to the Kaplan-Meier data in Figure 2A. The partitioned model was adopted for the base-case analysis because it provided the closest fit to the observed relapse data. In addition to this base-case analysis, alternative methods of describing the survival-to-relapse were investigated to explore the sensitivity of the cost-effectiveness ratios to these estimates. In these alternative methods, relapse during the first 5 years of the model (the duration of follow-up in BCIRG001), was described nonparametrically. Monthly probabilities were derived directly from the number of relapses observed in the trial. The risk of relapse was then extrapolated beyond the end of the trial using one of three alternative methods: (1) using a series of loglogistic models fitted to pooled data for both TAC and FAC arms to apply the same risk of relapse to both cohorts beyond trial end; (2) using a series of loglogistic models with treatment group as a factor to apply a chemotherapy regimen specific risk of relapse to each cohort beyond trial end; and (3) using published natural history data for a cohort of patients treated with CMF and followed for a period of more than 20 years (Weiss R, Woolf S, Demakos E, et al: J Clin Oncol 21:1825-1835, 2003).
Survival Postrelapse Mean survival postrelapse was estimated from the BCIRG001 patient-level data pooled for both treatment arms using Kaplan-Meier survival analysis. The Kaplan-Meier survival estimate is likely to underestimate mean survival because not all patients had died within the follow-up. Because survival data are usually skewed, the survival times of patients who had not died within follow-up would be expected to skew the mean towards a longer survival estimate. The impact of increasing the mean survival estimate was therefore explored in the univariate sensitivity analysis.
Subgroup Analysis However, outcomes for patients with early breast cancer are known to vary according to several risk factors. Younger patients, those with higher numbers of positive nodes and those with ER-negative or higher-grade tumors are known to be at greater risk of relapse (Early Breast Cancer Trialists’ Collaborative Group: Lancet 365:1687-1717, 2007; Berry DA, Cirrincione C, Henderson IC, et al: JAMA295:1658-1667, 2006. [Erratum: JAMA 295:2356, 2006]). Older patients, on the other hand, are at a greater risk of death resulting from other causes. The interaction between the competing risks of breast cancer relapse and death resulting from other causes is an important determinant in the overall benefit gained from chemotherapy. For example, given a constant risk reduction for recurrence with age, older patients are likely to gain less benefit from chemotherapy than are younger patients because they are less likely to have a recurrence and more likely to die as a result of other causes. We have explored the cost effectiveness of TAC versus FAC in subgroups of the overall population defined by age, number of positive nodes, ER status, and tumor grade. This analysis takes account of any nonsignificant differences in treatment effect reflected in the trial data as well as differences in the underlying hazard of relapse and death from other causes in patient subpopulations. Appropriate groupings were identified with assistance of a Clinical Advisory Panel and are displayed in Table A2. A partitioned survival model was fitted to the DFS data stratified by treatment group and risk group. This approach ensured that any nonsignificant differences in the efficacy of TAC were taken into account, as well as the difference in the baseline hazard of relapse. As in the base-case analysis, the treatment effect was assumed not to continue beyond the end of trial follow-up. All-cause mortality rates were adjusted for the relevant age group. The expected postrelapse survival was estimated for each subpopulation by Kaplan-Meier analysis of the patient-level data from BCIRG001. This analysis therefore takes account of any nonsignificant differences in treatment effect reflected in the trial data as well as differences in the underlying hazard of relapse and death from other causes in patient subpopulations. The results of the subgroup analysis (life-time analysis) are presented in Table A3. The analysis suggests that TAC is expected to be cost effective relative to FAC regardless of age, node status, ER status, and tumor grade. The incremental cost per QALY gained ranged from £6,736 in patients with one to three positive nodes to £16,656 in patients with four or more positive nodes. TAC was less cost effective in patients age 50 years or older than in younger patients, principally because of the greater risk of death resulting from other causes in the older group. Although the difference in treatment effect was not statistically significant, TAC was less cost effective in patients with four or more positive nodes than in those with one to three positive nodes, primarily because the treatment effect was smaller in this population. TAC was more cost effective in ER-negative patients than in the ER-positive group despite the treatment effect being smaller (not significantly), because the prognosis in the ER-negative group was predicted to be poor. TAC was more cost effective in patients with grade 1 or 2 tumors than in those with grade 3 tumors because the treatment effect was (not significantly) smaller.
Cost of Postrelapse Hospital Care The total expected cost postrelapse was estimated by bootstrapping (1,000 simulations with replacement), discounted to the point of first relapse and then applied in the Markov model as a payoff attributed to patients in the cycle in which they relapse. Where patients relapsed shortly before the end of the model timeframe and would have survived beyond the end of the timeframe, the cost postrelapse was truncated to reflect the survival within the model timeframe. As expected, the bootstrap mean cost was normally distributed. The uncertainty around the estimate of the cost postrelapse was reflected in the probabilistic analysis by random sampling from a normal distribution of the parameter as defined by the bootstrap mean and SE. The data collected by the WGH included care provided in the hospital setting only. No data were available to describe primary care received or hospice care.
Probabilistic Sensitivity Analysis
 b is the incremental benefit, ICERt is the ICER threshold, and c is the incremental cost. The probability of cost effectiveness at each ICER threshold is estimated as the percentage of the 1,000 simulations for with NMB more than 0. The probabilistic estimate of the ICER is generated by solving for the ICER threshold at which the mean net benefit is zero (using the GoalSeek function in Excel (Microsoft Corp, Redmond, WA). The 95% CIs are estimated in an analogous way, by solving for the ICER threshold at which the 95% CI is zero. The individual results of each of the 1,000 simulations are plotted on the cost-utility plane in presented in Figure A3A and the cost effectiveness acceptability curve is presented in Figure A3B. Equivalent results for the 10-year time-frame are presented in Figure A4.
We thank the following contributors: Gill Kerr and Morag Gibson (resource utilization data set); Paul Kind (EORTC QLQ-C30 algorithm); Neil Roskell, Lori McLeod, and Meghan Wills (statistical analysis); and Steven Beard, Fiona Maciver, Louise McCrink, and Angela Christie (model development).
Supported by a grant from Sanofi-aventis to Western General Hospital, Edinburgh, United Kingdom, to cover costs of data retrieval. RTI-HS was commissioned by Sanofi-aventis Inc to perform the modeling work. Contractual arrangements between RTI-HS and Sanofi-aventis did not prevent RTI-HS from independently publishing this work. This manuscript was reviewed by Sanofi-aventis. The model reported herein was used to support submissions to the UK National Institute for Health and Clinical Excellence (NICE) and the Scottish Medicines Consortium (SMC). Presented in part at the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) 8th Annual European Congress, November 6-8, 2005, Florence, Italy. Authors’ disclosures of potential conflicts of interest and author contributions are found at the end of this article.
1. Montero A, Fossella F, Hortobagyi G, et al: Docetaxel for treatment of solid tumours: A systematic review of clinical data. Lancet Oncol 6:229-239, 2005[CrossRef][Medline] 2. Martin M, Pienkowski T, Mackey J, et al: Adjuvant docetaxel for node-positive breast cancer. N Engl J Med 352:2302-2313, 2005 3. Roché H, Fumoleau P, Spielmann M, et al: Sequential adjuvant epirubicin-based and docetaxel chemotherapy for node-positive breast cancer patients: The FNCLCC PACS 01 Trial. J Clin Oncol 24:5664-5671, 2006 4. Dooley MJ, Singh S, Michael M: Implications of dose rounding of chemotherapy to nearest vial size. Support Care Cancer 12:653-656, 2004[Medline] 5. Hodgetts A, Raffle A: Prescribing Issues in Breast Cancer, a report on behalf of the Avon Somerset & Wiltshire Cancer Drug Policy Forum. Bristol, United Kingdom, Avon, Somerset and Wiltshire, 2001 6. Mart 7. Vogel CL, Wojtukiewicz MZ, Carroll RR, et al: First and subsequent cycle use of pegfilgrastim prevents febrile neutropenia in patients with breast cancer: A multicenter, double-blind, placebo-controlled phase III study. J Clin Oncol 23:1178-1184, 2005 8. Smith DH, Adams JR, Johnston SRD, et al: A comparative economic analysis of pegylated liposomal doxorubicin versus topotecan in ovarian cancer in the USA and the UK. Ann Oncol 13:1590-1597, 2002 9. Twelves C, Boyer M, Findlay M, et al: Capecitabine (Xeloda) improves medical resource use compared with 5-fluorouracil plus leucovorin in a phase III trial conducted in patients with advanced colorectal cancer. Eur J Cancer 37:597-604, 2001[CrossRef][Medline] 10. Aristides M, Jackson D, Liepa AM, et al: Costs of managing toxicities in non-small cell lung cancer with pemetrexed compared with docetaxel as second-line chemotherapy: Application to the UK setting. Presented at the 6th Annual European Congress of the International Society for Pharmacoeconomics and Outcomes Research, November 9-11, 2003, Barcelona, Spain 11. Ossa D, Briggs A, Cowell W, et al: Utility associated with severity of cancer-related anaemia (CRA): A societal valuation. Presented at the 7th Annual European Congress of the International Society for Pharmacoeconomics and Outcomes Research, October 24-26, 2004, Hamburg, Germany 12. Launois R, Reboul-Marty J, Henry B, et al: A cost-utility analysis of second-line chemotherapy in metastatic breast cancer: Docetaxel versus paclitaxel versus vinorelbine. Pharmacoeconomics 10:504-521, 1996[Medline] 13. Brown R, Hutton J: Cost-utility model comparing docetaxel and paclitaxel in advanced breast cancer patients. Anticancer Drugs 9:899-907, 1998[Medline] 14. Weiss R, Woolf S, Demakos E, et al: Natural history of more than 20 years of node-positive primary breast carcinoma treated with cyclophosphamide, methotrexate and fluorouracil-based adjuvant chemotherapy: A study by the Cancer and Leukemia Group B. J Clin Oncol 21:1825-1835, 2003 15. Office for National Statistics 2004: Life expectancy at birth: Methodological options for small populations. http://www.statistics.gov.uk/methods_quality/publications.asp 16. Kind P: Measuring the value of quality of life in cancer: An index based on EORTC QLQC-30. Presented at the 41st Annual Meeting of the American Society of Clinical Oncology, May 13-17, 2005, Orlando, FL 17. Hillner BE, Smith TJ: Efficacy and cost-effectiveness of adjuvant chemotherapy in women with node-negative breast cancer. N Engl J Med 324:160-168, 1991[Abstract] 18. National Institute for Health and Clinical Excellence: Guidance on cancer services: Improving outcomes in breast cancer. Manual Update, August 2002. http://www.nice.org.uk/nicemedia/pdf/Breast_cancer_info_public.pdf 19. Scottish Intercollegiate Guidelines Network, Management of Breast Cancer in Women: A draft for the National Meeting, Edinburgh, United Kingdom, 2003 20. Wolowacz S, Roskell N, Christie A, et al: Per-patient cost of hospital care for advanced breast cancer in the UK based on a large patient-level resource utilisation dataset. Value Health 8:A42, 2005 (abstr; PCN29) 21. Brown RE, Hutton J, Burrell A: Cost-effectiveness of treatment options in advanced breast cancer in the UK. Pharmacoeconomics 19:1091-1102, 2001[CrossRef][Medline] 22. British National Formulary, BNF50, September 2005. http://www.bnf.org/bnf/bnf/current/BNF50 23. Curtis L, Netten A: Unit costs of health & social care, 2004. Manchester, United Kingdom, Personal Social Services Research Unit, 2004 24. NICE Guidance No. 107: Trastuzumab for the adjuvant treatment of early stage HER2-positive breast cancer (2006). http://www.nice.org.uk/guidance/index.jsp?action=download&o=33459 25. NICE Guidance No. 112: Hormonal therapies for the adjuvant treatment of early oestrogen-receptor-positive breast cancer (2006) http://www.nice.org.uk/guidance/index.jsp?action=download&o=33642 26. NICE Final Appraisal Determination: Gemcitabine for the treatment of metastatic breast cancer (2006). http://www.nice.org.uk/page.aspx?o=389951 27. IMS Healthcare Oncology Analyzer, MAT, Q2, 2005 28. Italian Multicentre Breast Study with Epirubicin: Phase III randomized study of fluorouracil, epirubicin, and cyclophosphamide v fluorouracil, doxorubicin, and cyclophosphamide in advanced breast cancer: An Italian multicentre trial. J Clin Oncol 6:976-982, 1988 29. French Epirubicin Study Group: A prospective randomized phase III trial comparing combination chemotherapy with cyclophosphamide, fluorouracil, and either doxorubicin or epirubicin. J Clin Oncol 6:679-688, 1988[Abstract] 30. Early Breast Cancer Trialists’ Collaborative Group: Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: An overview of the randomised trials. Lancet 365:1687-1717, 2005[CrossRef][Medline] 31. Hall P, Walkington J, Douglas S, et al: Quality of life after 5 years tamoxifen for women with early breast cancer. Presented at the British Cancer Research Meeting, July 2-5, 2003, Bournemouth, United Kingdom 32. Poole CJ, Earl HM, Hiller L, et al: Epirubicin and cyclophosphamide, methotrexate, and fluorouracil as adjuvant therapy for early breast cancer. N Engl J Med 355:1851-1862, 2006 Submitted January 3, 2007; accepted October 24, 2007.
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2008 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
T) was significantly more effective than six cycles of FEC (HR for relapse = 0.82, P = .0434; HR for death = 0.73; median follow-up, 60 months). Because the number of cycles was equal in both treatment arms, improved outcomes in these trials may be directly attributed to the substitution of docetaxel into these regimens. 
n M, Lluch A, Segu
