4 Evidence and interpretation

The Appraisal Committee (appendix A) considered evidence from a number of sources (appendix B).

4.1 Clinical effectiveness

Bevacizumab

4.1.1 Three randomised controlled trials (RCTs) have investigated the effectiveness of bevacizumab as a first-line treatment for metastatic colorectal cancer.

  • One study (n = 813; median age 59 years) investigated the effect of irinotecan, bolus 5-FU and leucovorin (calcium folinate) (IFL) with and without the addition of bevacizumab.

  • The other two studies (one n = 71, median age 64 years; one n = 209, median age 71 years) investigated the effect of bolus 5-FU and leucovorin (5-FU/LV) with and without bevacizumab.

    For two of the studies the primary end point was overall survival, while in the smaller study that used 5-FU/LV as the comparator the primary end points were time to disease progression and best tumour response. In all three studies participants tended to have a good performance status (Eastern Cooperative Oncology Group [ECOG] status 0 or 1; unrestricted by disease or only restricted in strenuous physical activity), although in the larger study that used 5-FU/LV as a comparator, 7% had an ECOG status of 2 (ambulatory and capable of all self-care but unable to carry out any work activities).

4.1.2 Data taken from the manufacturer's submission are based on analyses carried out using data from the clinical trials database, which is subject to updates and revisions. Therefore in some instances the results presented here differ from the results in earlier published journal articles.

4.1.3 The addition of bevacizumab to IFL led to a statistically significant difference in median overall survival compared with IFL alone (20.3 months vs 15.6 months, respectively; hazard ratio [HR] 0.66, 95% confidence interval [CI] 0.54 to 0.81). In the studies that used 5-FU/LV as comparator there were no statistically significant differences in median overall survival. In the larger study the median overall survival in the bevacizumab-containing arm was 16.6 months compared with 13.2 months in the control arm (HR 0.77, 95% CI 0.56 to 1.05). In the smaller study the median overall survival in the bevacizumab-containing arm was 17.7 months compared with 13.6 months in the control arm (HR 0.52, 95% CI not reported; p = 0.07).

4.1.4 Progression-free survival (which was defined as time from randomisation until tumour progression or death) was measured in two of the studies. In both, there was a statistically significant difference in median progression-free survival. In the study comparing bevacizumab and IFL with IFL alone, the median progression-free survival was 10.6 months in the bevacizumab arm and 6.2 months in the control arm (HR 0.54, 95% CI 0.45 to 0.66). In the larger of the two studies comparing bevacizumab and 5-FU/LV with 5-FU/LV alone, the median progression-free survival was 9.2 months in the bevacizumab arm and 5.5 months in the control arm (HR 0.50, 95% CI 0.34 to 0.73). The smaller study that used 5-FU/LV as a comparator reported the median time to disease progression. There was a statistically significant difference favouring the bevacizumab arm over the control arm (9.0 months vs 5.2 months, respectively [HR 0.44, 95% CI not reported; p=0.005]).

4.1.5 All three studies measured tumour response rate (as partial or complete reduction in tumour size). In two studies, the differences in tumour response rate reached statistical significance. In the study with IFL as a comparator, the tumour response rate in the bevacizumab arm was 44.8% compared with 34.8% in the control arm (incremental difference 10.0%, 95% CI 3.3 to 16.7). In the smaller study that used 5-FU/LV as a comparator, the tumour response rate was 40.0% in the bevacizumab arm and 16.7% in the control arm (incremental difference 23.3%, 95% CI not reported; p = 0.029). In the larger study that used 5-FU/LV as a comparator, the difference in tumour response rate did not reach statistical significance (26.0% and 15.2% for treatment and control arms, respectively [incremental difference 10.8%, 95% CI not reported; p = 0.055]).

4.1.6 In all the studies there was a higher incidence of grade 3 and 4 adverse events in the groups receiving bevacizumab compared with the control groups:

  • 84.9% vs 74.0%, respectively, with IFL as the comparator

  • 74.3% vs 54.3% in the smaller study with 5-FU/LV as the comparator

  • 87% vs 71% in the larger study with 5-FU/LV as the comparator.

    Higher incidences of grade 3 and 4 hypertension were also reported in the groups receiving bevacizumab compared with the control groups:

  • 11.0% vs 2.3%, respectively, with IFL as the comparator

  • 8.6% vs 0% in the smaller study with 5-FU/LV as the comparator

  • 16% vs 3% in the larger study with 5-FU/LV as the comparator.

    For other grade 3 and 4 toxicities there were no consistent patterns of effects. An increased incidence of diarrhoea was reported in the study that used IFL as the comparator (32.4% vs 24.7%), and there was an increased incidence of thrombotic events in the smaller study that used 5-FU/LV as the comparator (14.3% vs 2.9%).

Cetuximab

4.1.7 The assessment group identified no studies that compared cetuximab with current standard treatments (which in the case of second- and subsequent-line treatment are FOLFOX and active/best supportive care [ASC/BSC], respectively). One RCT, the BOND study, was identified in which cetuximab combined with irinotecan was compared with cetuximab monotherapy (n = 329). In this study participants in the monotherapy arm could have irinotecan added to their treatment regimen upon disease progression. Three single-arm studies were also identified, of which two measured the effect of cetuximab monotherapy (one with 346 participants and one with 57 participants) and one measured the effect of cetuximab combined with irinotecan (n = 138). The primary outcome for all studies was tumour response rate. A median age of 56 years was reported in two of the trials and a median age of 59 years in the other two. In all four studies the populations tended to have good performance status (ECOG 0 to 1 or Karnofsky 80 to 100).

4.1.8 In the RCT there was no statistically significant difference in median overall survival between treatment groups. The median overall survival was 8.6 months in the cetuximab plus irinotecan arm and 6.9 months in the cetuximab monotherapy arm (HR 0.91, 95% CI 0.68 to 1.21). In the single-arm studies of cetuximab monotherapy, the median survival duration was 6.6 months (95% CI 5.6 to 7.6) in the larger and 6.4 months (95% CI 4.1 to 10.8) in the smaller study. In the single-arm study of cetuximab plus irinotecan, median overall survival duration was 8.4 months (95% CI 7.2 to 10.3).

4.1.9 In the RCT there was a statistically significant difference in median time to progression between treatment groups. The median time to progression was 4.1 months in the cetuximab combined with irinotecan arm and 1.5 months in the cetuximab monotherapy arm (HR 0.54, 95% CI 0.42 to 0.71). Median time to progression was reported in two of the single-arm studies: 1.4 months (95% CI 1.3 to 2.8) in the larger cetuximab monotherapy study and 2.9 months (95% CI 2.6 to 4.1) for cetuximab combined with irinotecan.

4.1.10 All four cetuximab studies measured tumour response rate. In the RCT there was a statistically significant difference between treatment groups. The tumour response rate was 22.9% in the cetuximab combined with irinotecan arm and 10.8% in the cetuximab monotherapy arm (incremental difference 12.1%, 95% CI 4.1 to 20.2). The rates of response in the single-arm studies were 8.8% (95% CI 2.9 to 19.3) and 12.0% (95% CI 8.4 to 15.4) in the two cetuximab monotherapy studies and 15.2% (95% CI 9.7 to 22.3) in the study that combined cetuximab with irinotecan. The Institute also received data, following completion of the assessment report, from three additional single-arm studies of cetuximab. In two of these studies all patients had received two prior chemotherapy regimens. Results from these studies confirmed the effect seen in other studies of cetuximab.

4.1.11 Data from the manufacturer's submission suggest that the response to cetuximab may be associated with an acne-like rash. Post hoc analyses of pooled data from the two studies in which patients received cetuximab combined with irinotecan (combined total of 339 patients) show 153 patients had stable disease at 6 weeks, of whom 50% had an acne-like rash of grade 2 or above (n = 76). Of these, 26% (n = 20) went on to have a partial response compared with 13% (n = 10) of those without an acne-like rash of grade 2 or above (p = 0.043).

4.1.12 Data from the RCT show that patients in the cetuximab plus irinotecan arm with an acne-like rash of grade 2 or above had an overall survival of 10.8 months compared with 5.8 months for those with either no rash or a grade 1 rash. In the single-arm study of cetuximab plus irinotecan, patients who had a grade 3 acne-like rash had a median survival of 13.1 months, compared with 10.6 months for those with a grade 2 rash, 6.2 months for those with a grade 1 rash and 4.3 months for those with no rash (p = 0.0008, grade 0 vs grade 1−3).

4.1.13 In the RCT the incidence of some adverse events was higher in patients receiving cetuximab plus irinotecan compared with those receiving cetuximab alone: grade 3 and 4 adverse events (65.1% vs 43.5%); diarrhoea (21.2% vs 1.7%); neutropenia (9.4% vs 0%); grade 3 or 4 acne-like rash (9.4% vs 5.2%).

4.2 Cost effectiveness

4.2.1 No published economic analyses of either bevacizumab or cetuximab were identified. The manufacturers of bevacizumab and cetuximab both submitted cost-effectiveness models, and the assessment group developed two models for each drug.

Bevacizumab – manufacturer's models

4.2.2 The manufacturer submitted two simple-state transition models with three health states: pre-progression, post-progression and death. Each model was based on data from a different bevacizumab study. The first was based on the study that compared bevacizumab plus IFL with IFL, while the second was based on the larger of the two studies that compared bevacizumab plus 5-FU/LV with 5-FU/LV. In both models the analysis was carried out from the perspective of the NHS. Data on progression-free survival for the treatment and control arms were taken from trial data, and an equal risk of death was applied following progression irrespective of treatment group. The models assumed equivalent utility scores for both the intervention and control groups, with a utility of 0.80 given to the pre-progression health state and 0.50 to the post-progression health state. Utility decrements associated with adverse events were not included. Pre-progression costs were calculated from the trials, augmented with data from other published sources. For post-progression costs an assumption of £2000 a month was used, applied equally to both arms.

4.2.3 With discounting of 6% for costs and 1.5% for benefits, the cost per quality-adjusted life year (QALY) gained was £88,364 for bevacizumab combined with IFL compared with IFL alone. With the same discounts, the cost per QALY gained for bevacizumab combined with 5-FU/LV was £56,628 compared with 5-FU/LV alone. One-way sensitivity analyses resulted in estimates of cost per QALY gained of between £82,577 and £106,770 for bevacizumab combined with IFL, and between £39,136 and £69,439 for bevacizumab combined with 5-FU/LV. Probabilistic sensitivity analyses suggest that the likelihood of cost effectiveness at a willingness-to-pay of £30,000 per QALY is 0.16 for bevacizumab combined with IFL, and 0.24 for bevacizumab combined with 5-FU/LV.

Bevacizumab – assessment group models

4.2.4 The methods used for the models produced by the assessment group were similar to those used in the NICE appraisal of irinotecan, oxaliplatin and raltitrexed (NICE technology appraisal 93). The assessment group presented two models based on the same trials as used in the manufacturer's models. The models were simple-state transition models with costs and effects calculated from the perspective of the NHS. Unlike the manufacturer's models the outcome data were based on published overall survival curves from the two studies. The utility value for pre-progression was the same as was used in the manufacturer's models (0.80), whereas that for post-progression was slightly higher (0.60). Data on second-line and subsequent therapies were taken from a study that investigated the optimal sequence of FOLFOX and FOLFIRI as first- and second-line therapies, and were applied equally to treatment and control groups. Costs were calculated from study data and augmented from a range of sources including published literature and personal communications. Discounting was not used because the distribution of costs incurred over time was unknown and was not considered relevant by the assessment group because of the short time horizon in the model.

4.2.5 The base-case costs per QALY gained for the assessment group models were £62,857 for bevacizumab combined with IFL and £88,436 for bevacizumab combined with 5-FU/LV, compared with IFL or 5-FU/LV alone, respectively. One-way sensitivity analyses of the base case produced a cost per QALY gained of £60,430–£76,831 for bevacizumab combined with IFL, and £51,355 and higher for bevacizumab combined with 5-FU/LV. Probabilistic sensitivity analyses suggest that, with a willingness-to-pay threshold of £30,000, the likelihood of bevacizumab being cost effective is zero.

4.2.6 The differences in the cost per QALY gained between the assessment group's model and the manufacturer's model are likely to have been caused by the difference in the methods used to calculate survival. The manufacturer's model resulted in more favourable estimates of the cost per QALY than did the assessment group's model when the comparator was 5‑FU/LV because the difference in progression-free survival was greater than the difference in mean overall survival. Conversely, the assessment group's model resulted in more favourable cost per QALY estimates when the comparator was IFL, because the difference in overall survival was greater than the difference in progression-free survival.

Cetuximab – manufacturer's model

4.2.7 The manufacturer's model for cetuximab used survival modelling to estimate the lifetime costs and benefits for patients receiving cetuximab combined with irinotecan compared with ASC/BSC. Two sets of analyses were presented. The first was based directly on survival data from the RCT, whereas in the second analysis adjustments were made to the survival data to reflect a proposed continuation rule. Under the continuation rule patients would only continue to receive cetuximab beyond 6 weeks if there were either a partial or complete tumour response or an acne-like rash of grade 2 or above.

4.2.8 The duration of survival of patients receiving cetuximab combined with irinotecan was extrapolated from data in the RCT. The survival of patients receiving ASC/BSC was calculated from the survival of the patients in the cetuximab monotherapy arm of the RCT. The data from the monotherapy arm were adjusted to remove the impact of cetuximab using an HR taken from an RCT of second-line irinotecan compared with ASC/BSC. Therefore the model assumes that the relative hazard of overall survival between cetuximab monotherapy and ASC/BSC as second-line and subsequent treatment is exactly equivalent to the relative survival hazard between irinotecan and ASC/BSC as second-line treatment. The modelling was carried out from the perspective of the NHS, with costs and resource data taken from the RCT comparing cetuximab monotherapy with cetuximab plus irinotecan and augmented from the published literature. Outcomes were presented as life years gained, with sensitivity analyses to examine the impact of quality of life using alternative utilities for metastatic colorectal cancer of 0.95 and 0.71, both constant over the lifetime and based on published data. Additional data were presented using a utility of 0.73, constant over the lifetime, based on data collected as part of a single-arm study that investigated the effectiveness of cetuximab as a second- and subsequent-line treatment for patients with metastatic colorectal cancer (MABEL). Costs and benefits were discounted at an annual rate of 3.5%.

4.2.9 The base-case analysis suggests a cost per life year gained of £33,263 if the continuation rule were applied. One-way sensitivity analyses with the application of the continuation rule result in cost per QALY estimates of £35,014 with a utility value of 0.95 and £45,566 with a utility value of 0.73. Without the continuation rule, cost per QALY estimates are £45,237 and £58,870 respectively. Cost-effectiveness acceptability curves suggest that at a willingness-to-pay threshold of £30,000 per life year gained the likelihood of cost effectiveness is 0.10.

Cetuximab assessment group models

4.2.10 In the absence of direct comparisons of cetuximab plus irinotecan with ASC/BSC or FOLFOX, the assessment group developed two models. The first was a threshold analysis considering the incremental benefit that cetuximab combined with irinotecan would have to provide over ASC/BSC in order to be considered cost effective. The second model was an indirect comparison of data from the arm receiving cetuximab and irinotecan in the RCT with data from other published studies of second-line ASC/BSC.

4.2.11 In both models overall survival for patients receiving cetuximab was estimated from patient-level data in the RCT. In the threshold analysis, the survival of patients receiving ASC/BSC was held as an unknown variable, whereas in the indirect comparisons different values for overall survival, ranging from 6 to 9 months, were taken from three published studies. Health-related quality of life was estimated in the same way as in the bevacizumab model, applying a utility of 0.80 to pre-progression disease states and 0.60 to post-progression states. For the cetuximab arm, measures of the duration of pre-progression survival as a proportion of overall survival were estimated using data from the RCT. For the comparator arm, they were derived from a trial that compared tipifarnib (a farnesyl transferase inhibitor) with BSC in refractory advanced colorectal cancer. In this study the duration of pre-progression survival was approximately 37% of overall survival. Resource use and costs were taken from the RCT as reported in the manufacturer's submission and augmented from the published literature and personal communication with clinical experts. Discounting was not used in the model because the distribution of costs incurred over time was unknown and was not considered relevant by the assessment group because of the short time horizon in the model.

4.2.12 The base-case threshold analysis suggests it is not possible for cetuximab combined with irinotecan to have a cost per QALY gained of less than £20,000, irrespective of the application of the continuation rule. When the proposed continuation rule is applied, cetuximab plus irinotecan must provide 0.65 additional life years (7.8 months) compared with ASC/BSC in order to achieve a cost per QALY gained of £30,000. This would imply that survival for patients receiving ASC/BSC would have to be 0.14 life years (1.7 months) or less. It was not possible to achieve a cost per QALY gained of less than £30,000 without the continuation rule. A sensitivity analysis using utility values from the MABEL study suggested that with the continuation rule applied cetuximab plus irinotecan must provide 0.60 additional life years (7.2 months) compared with ASC/BSC in order to achieve a cost per QALY gained of £30,000. Indirect comparisons are associated with a high level of uncertainty, but they yielded estimates of the cost per QALY gained that ranged from £77,210 to £370,044.

4.3 Consideration of the evidence

4.3.1 The Committee reviewed the data available on the clinical and cost effectiveness of bevacizumab and cetuximab for metastatic colorectal cancer, having considered evidence on the nature of the condition and the value placed on the benefits of bevacizumab and cetuximab by people with metastatic colorectal cancer, those who represent them, and clinical experts. It was also mindful of the need to take account of the effective use of NHS resources.

4.3.2 The Committee heard from clinical specialists that it is accepted that multiple chemotherapy treatments lead to incremental gains in survival if patients are sufficiently fit to receive therapy. The Committee further heard from clinical specialists that treatment regimens frequently involve combination therapy of 5-FU/FA either with irinotecan or with oxaliplatin as first-line therapies, followed, where patients are sufficiently fit, by irinotecan as a second-line therapy where oxaliplatin has been given and by oxaliplatin plus 5-FU/FA as a second-line therapy where irinotecan has been given. The experts suggested that it would be of merit to add further options and lines of therapy to this sequence.

Bevacizumab

4.3.3 The Committee reviewed the clinical effectiveness evidence from the bevacizumab studies and noted that the age of the population in the largest study was lower than the age of patients normally receiving chemotherapy in England and Wales. However, clinical specialists agreed that fitness, rather than age, is the primary factor when considering whether chemotherapy is appropriate. The Committee therefore agreed that the patients included in the bevacizumab trials could be considered to reflect a population of relatively fit patients with metastatic colorectal cancer in England and Wales.

4.3.4 The Committee noted that all bevacizumab studies demonstrated statistically significant gains in progression-free survival and that some studies also demonstrated statistically significant gains in overall survival and tumour response rate. However, the Committee noted that the comparators in the studies cannot be considered current standard practice in the NHS in England and Wales because the 5-FU treatment schedules involved administration by bolus rather than administration by infusion. It heard from clinical specialists that the benefits associated with bevacizumab in combination with bolus 5-FU are expected to be seen in infusional regimens because the two drugs have different mechanisms of action and their effects are therefore likely to be independent. This was also said to be supported by interim results from ongoing clinical studies. The Committee was therefore persuaded that the results seen in the studies could be considered generalisable to NHS practice in England and Wales.

4.3.5 The Committee then considered the estimates of cost effectiveness of bevacizumab. It noted that the models from the manufacturer and from the assessment group were similar in their methods and data sources; the models differed chiefly in their use of progression-free survival and overall survival, respectively, as the primary outcome data. The Committee noted that this difference resulted in different estimates of cost effectiveness from the manufacturer and the assessment group. However, the Committee considered that neither source resulted in a cost-effectiveness estimate that was compatible with the best use of NHS resources. The Committee noted that a proposal from the manufacturer for a registry programme could not be taken into further consideration because the Committee had been informed of imminent additional license extensions for bevacizumab, which would need to be taken into account of in any such scheme. It concluded that bevacizumab in combination with 5-FU/FA, with or without irinotecan as a first-line treatment for metastatic colorectal cancer would not be a cost-effective use of NHS resources.

Cetuximab

4.3.6 The Committee considered the clinical effectiveness evidence from the cetuximab studies. The Committee understood that cetuximab plus irinotecan could be given in its current licensed indication either as a second-line treatment following failure of an irinotecan-containing regimen or as a subsequent-line treatment following failure of both irinotecan- and oxaliplatin-containing regimens. The Committee recognised that in both cases, some patients may still have a high performance status, meaning that further chemotherapy regimens could be considered appropriate.

4.3.7 The Committee was concerned that there were no studies that compared cetuximab with current standard care either in second- or subsequent-line therapy, or with any therapy not including cetuximab. The Committee noted that there were currently no clinical studies available comparing cetuximab with FOLFOX, and therefore the relative clinical effectiveness of cetuximab as a second-line treatment could not as yet be determined. The Committee heard testimonies from clinical specialists that subsequent to second-line treatment progression-free survival and tumour response would be negligible if further active treatment was not available. Therefore the results seen in the single-arm cetuximab studies for these outcomes could be interpreted as an effect of the drug. It also heard that clinical specialists believed that cetuximab, in this situation, where no other active treatment was available, could prolong survival for a number of months if the disease responded to the drug. The Committee was therefore persuaded that cetuximab in this situation demonstrated some evidence of effectiveness. However, the relative effectiveness against current standard care remains uncertain.

4.3.8 The Committee heard from patient experts and clinical specialists about the impact of the acne-like rash associated with treatment on patients' quality of life. They heard from patient experts that, for some patients, the side effects of the drugs were tolerated willingly because of the perceived benefit, whereas for others the side-effect profile may be more of a consideration. The Committee heard from clinical specialists that the acne-like rash was generally well managed with antibiotics, treatment breaks or dose reduction. The Committee was therefore satisfied that the side-effect profile of cetuximab should not be a determining factor in its deliberations.

4.3.9 The Committee noted the contradiction that although the UK marketing authorisation stipulates that patients need to be tested for the presence of EGFR, a positive test for the presence of EGFR did not predict response to treatment. The Committee heard from clinical specialists that there is increasing knowledge of the mechanism of action of cetuximab and that it is now thought that the antibody identified through EGFR testing is different from the one targeted by cetuximab. The Committee noted the difficulties in identifying patients who were likely to respond to cetuximab, but was fully aware that decisions about its use in the NHS would have to be based on the current marketing authorisation.

4.3.10 The Committee considered the continuation rule proposed by the manufacturer. The Committee expressed concern about conflicts with the SPC, but it agreed that this would only be an issue for the small proportion of patients who experience grade 3 and 4 acne rash. Although it acknowledged that there was some evidence to suggest that the presence of a rash may predict response to cetuximab, the Committee had reservations about using it to decide whether to continue or discontinue treatment because no studies have so far tested prospectively this continuation rule.

4.3.11 The Committee considered the cost-effectiveness evidence for cetuximab. It noted that the economic modelling from both the manufacturer and the assessment group had been completed using effectiveness data from the RCT of cetuximab where approximately 80% of patients received cetuximab plus irinotecan as a third-line or subsequent therapy. It was also aware that the comparator used in both models was ASC/BSC, which meant the modelled scenario and corresponding estimates of cost effectiveness more closely resembled third-line or subsequent use of cetuximab rather than second-line use.

4.3.12 The Committee discussed the uncertainties around the estimates of utility for patients with metastatic colorectal cancer. The manufacturer had provided estimates between 0.95 and 0.71, both constant over the lifetime of the patient. The Committee considered that the utility for a patient with metastatic colorectal cancer was more likely to reflect the lower end of this range, based on additional data submitted by the manufacturer from the MABEL study. The Committee concluded that, using the most realistic utility estimates, the cost-effectiveness estimates provided by both the manufacturer and the assessment group were not compatible with the best use of NHS resources. The Committee also noted that these estimates were associated with a high level of uncertainty because they were based on indirect comparisons.

4.3.13 The Committee therefore considered threshold analyses completed by the assessment group, where the survival in the comparator arm was held as unknown. The base-case threshold analysis suggested that, with the application of the continuation rule, a cost per QALY gained of £30,000 could only be achieved if survival with ASC/BSC is less than 2 months. A sensitivity analysis adjusting the assumptions to reflect utility values from the MABEL study did not materially alter the results. The Committee noted that the manufacturer had provided an estimate of mean survival of 5.6 months for patients receiving ASC/BSC in their economic model, while studies of ASC/BSC identified in the assessment report provided estimates of median survival ranging from 6 to 9 months. The Committee therefore considered that an estimate of mean survival while receiving ASC/BSC of approximately 2 months was an unrealistic underestimate. Considering all the available evidence on clinical and cost effectiveness, the Committee therefore concluded that cetuximab, either as a second-line or a subsequent-line treatment for metastatic colorectal cancer would not be a cost-effective use of NHS resources.