TA145: cetuximab for locally advanced squamous head and neck cancer²¹⁵

The i.v. infusion of cetuximab was apparently costed as a medical oncology outpatient visit using NHS reference costs: £178.66 for the initial visit and £124.66 thereafter. The ERG noted the i.v. infusion administration cost, but did not particularly comment on it.

TA172: cetuximab for recurrent and/or metastatic squamous head and neck cancer²²⁰

A weighted average of inpatient and outpatient administration was drawn from Hopper et al.²³² These were costed using the NHS 2007–8 tariff for X99OST: Solid Tumour Cancer Chemotherapy: All Drugs at £296.00, and the NHS reference costs 2004 for a medical oncology outpatient visit of £124.66.

TA145: cetuximab for locally advanced squamous head and neck cancer²¹⁵

The cost of radiotherapy and cetuximab was drawn from IPD in the pivotal trial. Three radiotherapy regimens were possible, with cetuximab being added to these. The maximum duration of treatment was 8 weeks.

TA172: cetuximab for recurrent and/or metastatic squamous head and neck cancer²²⁰

The cost of cetuximab was drawn from IPD in the pivotal trial. The pivotal trial permitted up to six 21-day cycles of therapy. Some of the regimen drugs could be withdrawn if not tolerated. Study treatment was discontinued early if there was unacceptable toxicity or PD.

TA145: cetuximab for locally advanced squamous head and neck cancer²¹⁵

For those deemed to be cured, UK life tables coupled with a proportionate hazard of 2.786 were used to model mortality. For the remainder, PFS and OS was modelled by fitting log-normal parametric curves to the Kaplan–Meier data.

The ERG comments on the modelling of survival appeared to mainly be with its treatment in the probabilistic modelling, the large uncertainty around the extrapolated survival not be reflected within this and the probabilistic results.

TA172: cetuximab for recurrent and/or metastatic squamous head and neck cancer²²⁰

Parameterised Weibull curves were fitted to the trial Kaplan–Meier data for PFS and OS.

TA145: cetuximab for locally advanced squamous head and neck cancer²¹⁵

The manufacturer commissioned a utility valuation study from M-TAG Ltd. This aimed to estimate utility values for a series of health states describing a range of side effects and post-treatment outcomes among patients with locally advanced squamous cell H&N cancer. A literature search identified how the AE profiles changed with the addition of cetuximab to radiotherapy. This informed the choice of AEs that should be included in the study:

• stomatis/mucus membrane disorders
• nausea/vomiting
• haematological toxicities
• rash/acne
• late onset peripheral neuropathy
• late onset ototoxicity.

Seven health states were used to describe different toxicity grades, based on the NCI Common Toxicity Criteria. Two further health states described the late toxicities of peripheral neuropathy and ototoxicity, and a further two health states described the final outcomes of treatment success and treatment failure.

A total of 50 UK oncology nurses were recruited for the study, as it was thought unethical to recruit patients. Nurses were screened before being accepted into the study in order to ensure they were familiar with the area, having:

• a minimum of 2 years working as an oncology nurse
• a minimum of 11 patients in clinic with locally advanced squamous cell H&N cancer in the last 3 months; and,
• experience in treating patients with radiotherapy, chemotherapy or concomitant chemoradiation therapy.

They rated the 11 health states using the EQ-5D and the EQ-5D visual analogue scale (VAS), and also ranked the various health states from 1 to 11. This resulted in the following utility estimates (Table 90).

TABLE 90

TA145 EQ-5D nurse estimates of QoL for H&N cancer

These values needed to be coupled with the mean times in the health states. For treatment duration as defined by the acute phase this was differentiated by arm. For the AEs for health states B through to G, these were calculated based on the average time spent with the AE pooled across the arms. To do this, health states B–G were ordered according to the ranking of the utility study. For patients experiencing more than one of the health states B–G, the QALY impact of the AEs was determined by the mean EQ-5D utility for the worst health state experienced multiplied by the mean duration of the AE during the pivotal trial. The submission is not quite clear about how the mean durations of the AEs were calculated, but it appears to be across all events observed during the trial without any similar ranking for multiple events being applied. The ERG report, in its assessment of the costing of AEs, further suggests that overlapping AEs were ignored in the calculation of the mean durations of AEs (Table 91).

TABLE 91

TA145 AE QALY decrements

The ERG noted that given the absence of other studies for locally advanced SCC of the H&N, the nurses were reasonable patient proxies given their experience. The ERG also noted that not taking into account multiple AEs could have tended to lessen the estimated impact of AEs within the modelling, though it is not clear whether or not this is a criticism of the estimated utilities and QALY losses for the given health states per se. The ERG also noted that censoring could have tended to reduce the estimated duration of AEs and so their estimated QALY impact.

TA172: cetuximab for recurrent and/or metastatic squamous head and neck cancer²²⁰

Individual patient EORTC QLQ-C30 data were mapped onto EQ-5D scores using the algorithm developed by Kind²²¹ study among pancreatic cancer patients where:

HRQoL = 0.633 + 0.047 × Q29 –0.124 × Q3 –0.167 × Q5 – 0.102 × Q20 – 0.082 × Q26

This resulted in HRQoL values of 0.69 for stable/response with cetuximab, 0.65 for stable/response with standard treatment and 0.52 for PD. Unfortunately, the detail of this is given in an appendix to the submission that is not publicly available. QoL values for AEs were not separately calculated.

The ERG noted the uncertainty inherent in the mapping function, and the lack of any statistically significant difference for health states between the arms. The ERG also noted that AEs had not been explicitly considered, and also that these would not have been captured within the mapping function of Kind²²¹ It felt that some elements of the mapping function for EORTC QLQ-C30 in lung cancer of Baghust et al.²³³ could have proxied for those elements not within Kind.²²¹

TA145: cetuximab for locally advanced squamous head and neck cancer²¹⁵

Some AEs were grouped into a single category for costing purposes: mucositis/stomatitis/dysphagia, acne/rash and nausea/vomiting. An expert panel was convened to estimate the proportion of AEs that would result in a hospital admission, and the medication that would be administered for both those who were and were not admitted. NHS reference costs were applied to the proportion that were assumed to be admitted, these costs being assumed to cover all relevant procedures. Thrombocytopenia was associated with the cost of a platelet transfusion, the estimate for this being drawn from Varney and Guest²³⁴ Medication costs were conditioned by the duration of events as estimated from trial data, and described in greater detail in Quality-of-life values for disease states and adverse events. No primary care costs were included presumably due to it being assumed that ongoing routine hospital follow-up identified and prescribed medication for the AEs.

An ERG expert suggested that the grouping of AEs for costing purposes into mucositis/stomatitis/dysphagia, acne/rash and nausea/vomiting was reasonable. The ERG also noted the possibility of bias arising from censored data and questioned the elimination of overlapping AEs from the analysis in order to estimate the mean durations of individual AEs.

TA172: cetuximab for recurrent and/or metastatic squamous head and neck cancer²²⁰

The AE costs as estimated in TA145²¹⁵ were applied (Table 92).

TABLE 92

TA145 and TA172 AE costs