Guidance on the use of trastuzumab for the treatment of advanced breast cancer

One randomised controlled trial (RCT; n=469) of first-line trastuzumab combination therapy was available. All individuals had HER2 overexpression at least at level 2 or more. Patients who had not previously received anthracyclines were randomised to an anthracycline in combination with cyclophosphamide with or without trastuzumab. Patients who had previously received an anthracycline as adjuvant therapy (n=188) were randomised to paclitaxel (n=96) or paclitaxel plus trastuzumab (n=92). The primary endpoint was time to progression and the median duration of follow-up was 30 (range 30 to 51) months. Patients in all arms of the trial were given the option of receiving trastuzumab monotherapy at the time of disease progression, meaning that allocation to this further treatment was non-random.

When only the 2 relevant treatment arms involving paclitaxel were considered, there was no significant difference in overall survival between the group treated with trastuzumab plus paclitaxel and the group treated with paclitaxel alone (22 versus 18 months, p=0.17). However, the addition of trastuzumab resulted in longer median times to disease progression (7 versus 3 months, p<0.001), duration of response (11 versus 5 months, p<0.01) and time to treatment failure (6 versus 3 months, p<0.001). There was no statistically significant difference between the 2 treatments in terms of complete response (relative risk [RR] 3.65, 95% CI 0.89 to 15.22), but overall tumour response (RR 2.48, 95% CI 1.49 to 4.12), disease progression (RR 0.38, 95% CI 0.27 to 0.53) and treatment failure (RR 0.46, 95% CI 0.33 to 0.63) favoured treatment with trastuzumab. For patients with HER2 level 3 or more, trastuzumab and paclitaxel was associated with a longer median survival than paclitaxel alone (25 versus 18 months, no p-value provided).

Quality of life, assessed using the pain and dyspnoea domains and the breast cancer module of the EORTC QLQ-C30 questionnaire, was higher in the group receiving trastuzumab plus chemotherapy than in the group receiving chemotherapy alone.

The most important adverse event seen in the trial was cardiac dysfunction, which occurred in 27% of the group given an anthracycline, cyclophosphamide and trastuzumab, 8% of the group given an anthracycline and cyclophosphamide alone, 13% of the group given paclitaxel and trastuzumab, and 1% of the group given paclitaxel alone. The cardiotoxicity was potentially severe, and in some cases life threatening, but the symptoms were reported generally to improve with standard medical management.

Although not originally observed during clinical trials, serious infusion-related reactions have been reported in 74 from a total of approximately 25,000 patients who received trastuzumab, with the reactions leading to the death of 15 patients. These reactions, which include anaphylaxis and severe dyspnoea, usually occur within 24 hours of the infusion, although delayed reactions have also been reported. Allergic or hypersensitivity reactions, haematological toxicity, hepatic and renal toxicity, diarrhoea and an increased risk of infections have also been noted.

No comparative RCTs of trastuzumab monotherapy (versus systemic therapy without trastuzumab) were available. Of the studies that were identified, 2 were case series of trastuzumab (H0551g, n=46; H0649g, n=222) and 1 was an RCT of trastuzumab monotherapy, which was concerned with an unlicensed indication and was essentially a dose ranging study (H0650g, n=113). Women in H0650g received first-line monotherapy with trastuzumab, which is an unlicensed indication.

In H0649g, 4% of women experienced a complete response to treatment and 12% experienced a partial response. The overall response rate was 15%. Smaller proportions of women responded to treatment in study H0551g.

The median duration of survival in H0649g was 13 months for all women and 16 months in a sub-group analysis of women overexpressing HER2 at levels of 3 or more. In the manufacturer's submission, survival reported in H0649g was indirectly compared with survival reported in 2 RCTs of vinorelbine monotherapy. Both RCTs contained a treatment arm consisting of vinorelbine monotherapy as second-line or salvage therapy for MBC. In 1 of these RCTs, 91% of participants had received prior treatment with anthracyclines for advanced breast cancer. The median duration of survival for people receiving vinorelbine monotherapy was 8 months. The reported median duration of survival in the other study was 10 months. Participants in these 2 RCTs were not selected on the basis of their HER2 status, which may suggest that their prognosis was better than those in H0649g.

The manufacturer also submitted evidence from the Imperial Cancer Research Fund database at Guy's and St Thomas' Hospital NHS Trust on patients who had received third-line chemotherapy for MBC and who were not treated on the basis of their HER2 expression status. Analysis of these data showed that the median survival for these patients was 6.3 months.

Trastuzumab monotherapy appeared to have a relatively low toxicity level. For study H0649g, the common adverse events (occurring in approximately 40% of people) were infusion-related fever and/or chills that usually occurred only during the first infusion. The most clinically significant adverse event was cardiac dysfunction, which occurred in 10 people (5%). However, only 1% of participants in this study discontinued treatment because of treatment-related adverse events. In study H0551g, toxicity was reported as minimal, although 2 patients died as a result of cardiac dysfunction. In study H0650g, the adverse events recorded were mainly mild to moderate in nature and were mostly associated with the higher dose regimen; only 1 person experienced cardiac dysfunction.

The manufacturer evaluated the cost-effectiveness of trastuzumab in combination with paclitaxel versus paclitaxel alone for patients with HER2 level 3 or more, based on results from the RCT. Estimates of direct medical and social care costs were included in the evaluation, including the costs of HER2 testing (£21 for a single test) and cardiac testing (£520 to £640 for 4 tests).

The manufacturer estimated the incremental cost-effectiveness ratio to be £37,500 per quality-adjusted life-year (QALY) gained (and substantially less per life-year gained). This survival benefit used to estimate the QALY gain was based on a weighting of case-mix reflecting the selection of patients in the pivotal trial who, after the trial, crossed over to trastuzumab monotherapy. After extrapolating the trial results for this selection of patients, approximately 10 months' mean survival advantage was imputed into the economic evaluation. A number of other sources, in particular 2 non-controlled studies that examined the use of taxane monotherapy as first-line treatment for metastatic breast cancer, suggest a survival advantage of combination therapy compatible, or even better, than this.

One economic evaluation of monotherapy comparing trastuzumab with vinorelbine monotherapy was available to the committee. Direct information relating to clinical effectiveness was unavailable (as there were no RCTs of trastuzumab other than a dose ranging study). Health outcomes were expressed in terms of life years and QALY by extrapolating survival from non-controlled studies. Direct medical and social care costs were included in the evaluation. Information on clinical effectiveness was imputed from H0649g and an RCT of vinorelbine versus melphalan that contained patients who were at an earlier stage of the disease and who were not selected on the basis of HER2 expression status. Patients who received vinorelbine in the RCT survived for approximately a median of 8 months. The manufacturer referenced 2 other non-controlled studies in an attempt to validate this period of survival for patients receiving vinorelbine. One of these studies examined median survival in patients at a similar disease stage who received vinorelbine monotherapy; in this study, median survival was approximately 6 months.

The estimated incremental cost-effectiveness ratio was approximately £7,500 per life-year gained if trastuzumab was used instead of vinorelbine, when it was assumed that the additional survival attributable to trastuzumab monotherapy was 8 months. The manufacturers also provided a cost per QALY of approximately £19,000 by assuming that the 8 months of additional survival was equivalent to 2.6 quality-adjusted months.

The appraisal committee considered that a survival gain of approximately 10 months used in the economic evaluation was likely to be an underestimate of the true survival gain attributable to combination therapy given that patients in the non-controlled studies of taxane monotherapy were not HER2 selected. The appraisal committee also concluded that the utility weights used to adjust survival for changes in quality of life were low.

Based on these factors, the appraisal committee believed that trastuzumab combination therapy was likely to be lower than the estimate of £37,500 per QALY gained provided by the manufacturer.

The appraisal committee also noted that improvements in survival of this magnitude due to therapeutic intervention have rarely been recorded in women with metastatic breast cancer.

The evidence for the effectiveness of trastuzumab monotherapy was limited to 2 case-series and 1 RCT, which was concerned with an unlicensed indication and was essentially a dose ranging study. The report for the first case-series (H0551g) did not state the line of therapy being assessed or the length of follow-up. The second case-series (H0649g), which was relatively well reported, suggested that in terms of response rate trastuzumab monotherapy was an effective treatment in patients with MBC and HER2 overexpression at levels 3 or more.

Although the appraisal committee had some reservations about the quality and robustness of the economic evaluation for trastuzumab monotherapy, as survival was not based on results from controlled studies, it was believed that these misgivings would not increase the ratio sufficiently to suggest that it is not cost-effective.