The study compared the use of trastuzumab (Herceptin) in combination with adjuvant chemotherapy versus chemotherapy alone for the treatment of patients with early stage breast cancer. Patients received trastuzumab for either a 12-month or 9-week course.

Treatment.

The aim of the study was to estimate the impact of using trastuzumab in combination with standard treatment (i.e. chemotherapy), compared with standard treatment alone, on overall survival and costs for patients with human epidermal growth factor receptor 2 protein (HER2)-positive, resectable, early-stage breast cancer. As the outcome of interest occurred over a period longer than the duration of any clinical trial, the outcomes were modelled using a Markov model. The perspective adopted in the economic analysis was that of the Australian health care system.

Cost-effectiveness analysis and cost-utility analysis.

As this was a modelling study, the target population comprised a cohort of patients with HER2-positive breast cancer. The cohort had an age at diagnosis of 50 years.

A Markov model was developed to estimate average survival. The model was run until the patients reached 100 years of age. The health states, cycle length and transition probabilities were presented in full in the paper, along with a number of modelling assumptions.

The setting was not explicitly reported. However, it appears to have been secondary care. The economic study was carried out in Australia.

The effectiveness data used to populate the model came from studies published between 2001 and 2006. The price year was 2005. Resource use was based on the authors' clinical experience, while the cost data were obtained from official sources (Australian Department of Health) published in 2007.

The clinical parameters associated with the treatment interventions included the transition probabilities for different health states (including death from metastatic breast cancer and non-cancer causes). The transition probabilities were adjusted to reflect risk of relapse in treated breast cancer over time. The risk of New York Heart Association Grades III and IV heart failure during trastuzumab administration was also accounted for in the model.

The baseline transition probabilities were obtained from an international, multi-centre randomised trial and from two large published trials. The design of the latter studies was not discussed in detail in the current paper.

The process used to identify the data was not reported. However, the authors appear to have obtained the data from large, mainly Phase III, trials in which trastuzumab was administered for the period of interest (1 year and 9 weeks, respectively). No inclusion criteria were specified for any of the parameters. The adjustments the authors made to published transition probabilities were reported.

The measures of benefit used were the years of life saved (YOLS), quality-adjusted life-years (QALYs) and cancer deaths avoided. Average life expectancy (average survival) was derived from the model and the methods used for estimation were explicitly reported. Survival was adjusted for quality of life using utility weights from a published systematic review of cost-utility evaluations in oncology, while utility weights for patients who experienced heart failure were based on authors' assumptions. The utility weights used in the model were reported and were discounted at a rate of 3%.

The costs to the health service included:

the cost of trastuzumab;

metastases pre-palliative and palliative phase (palliative chemotherapy, ancillary drug therapy, radiotherapy, specialist and home or hospice nursing, and general practitioner (GP) visits, hospice care and transportation costs);

loco-regional recurrence costs (chemotherapy, radiotherapy, GP visits and transportation costs);

heart failure screening and treatment costs (echocardiography, electrocardiograph, specialist and GP visits, drug costs).

It was reported that annual treatment costs for diseases other than cancer were also accounted for in the analysis, but the annual costs of medical insurance were used to proxy these cost estimates. The cost of HER2 over-expression testing was excluded as it was common to all alternatives. Resource use was based on the authors' clinical experience. The unit costs were taken from national Australian published unit costs. The costs were reported as the average cost per patient and were reported separately from the quantities of resources used. The costs were discounted at an annual rate of 3% and were reported for the price year 2005.

Productivity costs were not included in the analysis.

Australian dollars (AUD).

The cost estimates were treated deterministically.

Parameter uncertainty was investigated by means of one-way and multi-way sensitivity analyses. The parameters tested were the discount rate, age at diagnosis, cost of treating metastases, relapse and other diseases, cost of heart failure, duration of trastuzumab effect and costs due to risk of distant recurrence. The ranges over which the parameters were tested were explicitly reported. In the multi-way sensitivity analysis, the discount rate, duration of treatment effect of trastuzumab and individual costs over 52 weeks were varied simultaneously. The same analysis was conducted taking into account a cohort at the age of 70 years. The authors gave detailed descriptions of the derivation of these ranges.

Average survival was 25.3 years in the trastuzumab (52-week) group and 21.2 years in the standard treatment group. This resulted in an incremental survival of 4.1 years.

The incremental cases of cancer avoided were 136 per 1,000 patients (relative risk reduction of 28%).

Estimated QALYs were not reported.

The average total lifetime discounted costs (including drug acquisition costs of trastuzumab) were AUD 87,818,858 in the trastuzumab (52-week) group and AUD 31,513,284 in the current therapy group.

An incremental cost-effectiveness analysis was performed.

The 52-week trastuzumab intervention resulted in an incremental cost of AUD 13,720 per additional YOLS and AUD 22,793 per additional QALY gained. When the costs of future disease were excluded, the incremental cost-effectiveness ratio was AUD 21,771 per QALY gained.

The 9-week trastuzumab course resulted in an incremental cost of AUD 1,016 per additional YOLS and AUD 1,700 per QALY gained compared with standard treatment alone.

One-way and multi-way sensitivity analyses demonstrated that the results of the model were most sensitive to variations in the drug acquisition costs, duration of treatment benefit of trastuzumab and the discount rate. The model validation processes demonstrated that the model compared well with the results of recent literature.

The combination of trastuzumab with adjuvant chemotherapy constituted a cost-effective option. However, the economic impact of the 52-week course on the overall budget is considerable, making the 9-week course a more attractive option.

The selection of the comparators was justified by reference to Phase III and smaller trials that reported considerable benefits from the use of adjuvant trastuzumab over standard treatment alone. The relative merits of the treatment options in relation to overall survival were unknown. Standard treatment seems to have represented the most commonly used approach in the authors' setting. You should decide if these represent valid comparators in your own setting.

The model structure was described. Model parameters, their sources and the modelling assumptions employed were reported in full. The authors evaluated the internal validity of the model and investigated uncertainty in the model parameters through one-way and multi-way sensitivity analyses. The ranges over which the parameters were tested and the results of the sensitivity analysis were reported.

The parameters were taken from published research. No systematic search for the data was reported and no inclusion criteria were specified for any of the parameters. However, the authors mainly used data from an international, multi-centre, randomised Phase III trial and a smaller randomised trial, which potentially have a satisfactory level of internal validity. In addition, adjustments made by the authors to published estimates were reported and justified.

The authors used YOLS, QALYs and cancer deaths avoided as measures of benefit in the economic analysis. The estimation of average survival was derived appropriately from the model. The utility weights were derived from published literature, but no details of the valuation method were specified. The benefits were appropriately discounted.

The analysis of the costs was performed from the perspective of the Australian National Health Service. It appears that all the relevant categories of costs have been included in the analysis. Although the cost of HER2 over-expression testing was excluded from the analysis, it was assumed to have been common to all alternatives and its omission is unlikely to have affected the authors' conclusions. Annual medical insurance costs were used to proxy the annual cost of non-cancer illness, which was inappropriate given the health service perspective of the study. The authors evaluated uncertainty in the specific cost categories, but did not investigate uncertainty in the resources used. The costs and the quantities were reported separately, which will enhance the generalisability of the results to other settings. The costs were appropriately discounted and the price year was reported.

The time profile during which health changes materialised was quite complex and appears to have been dealt with in an appropriate manner. Although the authors reported results for the 9-week treatment course, the reporting of the results mainly focused on the 52-week treatment course and this choice was explicitly justified. The results of the study were compared with those from other published studies and methodological differences were discussed. The authors also considered the effect of adjuvant trastuzumab in younger patient groups and evaluated the impact of this on the economic results in a sensitivity analysis. The authors appear to have provided a balanced discussion about the limitations of their study and the methods employed to address them.

The authors did not make explicit recommendations for changes in policies or practice. Although not explicitly recommended by the authors, it can be inferred from their discussion that further research is needed to determine the effectiveness of trastuzumab relative to the age of diagnosis.

None stated.

This record was compiled by CRD commissioned reviewers according to a set of guidelines developed in collaboration with a group of leading health economists.