Talimogene laherparepvec for treating unresectable metastatic melanoma

4.2 The marketing authorisation for talimogene laherparepvec is for unresectable melanoma that is regionally or distantly metastatic (stage 3B, 3C and 4M1a) with no bone, brain, lung or other visceral disease. The committee noted that this is based on evidence from a post-hoc subgroup within the OPTiM trial (57% of the overall trial population with no visceral metastatic disease). The clinical experts stated that in clinical practice, treatment with talimogene laherparepvec would be suitable for approximately 10 to 15% of people with unresectable metastatic melanoma.

4.3 The patient expert stated that talimogene laherparepvec might be a particularly valuable option for people with visible skin tumours, which can be a source of great anxiety. The clinical experts considered the main benefits of talimogene laherparepvec to be that the method of administration is acceptable to patients, and that it has an improved toxicity profile compared to currently available systemic treatments (particularly ipilimumab). They stated that patients with melanoma that is suitable for treatment with talimogene laherparepvec may have multiple small lesions, which make surgical resection impractical, and that other localised therapies such as isolated limb perfusion are not widely available. Having a choice of effective treatments would be particularly valuable to people with this condition. The committee concluded that the availability of a new treatment option with a novel mechanism of action and improved tolerability would be valuable for people with metastatic melanoma, if it was shown to be as clinically effective as other available treatments.

4.4 The comparators in the final scope were the immunotherapy agent ipilimumab, and the BRAF inhibitors vemurafenib and dabrafenib. The newer systemic immunotherapy agents, pembrolizumab and nivolumab, were not included as comparators in the scope of this appraisal. However, the committee noted the recent NICE technology appraisal guidance on pembrolizumab for treating advanced melanoma after disease progression with ipilimumab and on pembrolizumab for advanced melanoma not previously treated with ipilimumab and also on nivolumab for treating advanced (unresectable or metastatic) melanoma, which had been published by the time of the final meeting of the committee. The clinical experts noted that these treatments would be considered for the same group of patients as talimogene laherparepvec. For patients with BRAF negative or wild type disease the only alternative therapy in routine clinical practice would be systemically administered immunotherapy agents. However, for people with the BRAF‑V600 mutation, the disease can be treated either with immunotherapy or BRAF‑specific agents. This choice would be influenced by the overall burden of disease, and whether it is slowly or rapidly progressing. A BRAF inhibitor, such as vemurafenib or dabrafenib, is likely to be the preferred treatment for people with BRAF‑V600 mutations whose disease is progressing rapidly, while immunotherapies such as ipilimumab, pembrolizumab and nivolumab will be used for people with BRAF‑V600 mutations with more slowly progressive disease or a lower tumour burden. The committee heard that in the light of emerging evidence of long‑term benefit experienced by some people having immunotherapy, this would generally be used in preference to the BRAF inhibitors whenever clinically possible. For practical purposes, the group of patients considered for immunotherapy, and in particular talimogene laherparepvec (who have earlier stage disease and no visceral metastases) would not correspond with those for whom a BRAF inhibitor would be the first choice of treatment. The committee concluded that the most clinically relevant comparator within the scope for this appraisal was ipilimumab. The committee noted that the newer immunotherapies, pembrolizumab and nivolumab, were not included in the scope and therefore could not be considered as direct comparators as part of the appraisal process. However it was reasonable for the committee to acknowledge their increasing use in clinical practice, particularly since they had shown superior short‑term outcomes to ipilimumab in clinical trials and had lower toxicity than ipilimumab.

4.5 The evidence underpinning the marketing authorisation for talimogene laherparepvec came solely from an exploratory post-hoc subgroup analysis of people in the OPTiM trial who had melanoma with no visceral metastases. The committee was aware that the comparator arm in the trial was granulocyte macrophage colony-stimulating factor (GM-CSF), which in the view of the clinical experts is clinically ineffective, effectively equivalent to placebo, and is not used in clinical practice. The committee noted that, in common with ipilimumab, pembrolizumab and nivolumab, talimogene laherparepvec is a disease-modifying immunotherapy and some patients who have a complete or sustained response may require no further treatment for melanoma. The clinical experts stated that although durable response rate is a new, non-validated endpoint in clinical trials of advanced melanoma, it is considered to be more clinically meaningful than overall response rate because of its association with a reduced risk of recurrence. In the OPTiM trial, talimogene laherparepvec showed a statistically significant improvement of 25.3 months in overall survival (p value 0.0008), a durable response rate of 25.2% (compared with 1.2% for GM-CSF) and a complete response rate of 16.6% (compared with 0% for GM-CSF). The ERG raised concerns about the potential for bias in the trial because of limited blinding, differences in the withdrawal rates in the 2 arms, and the use of a non-validated primary endpoint, all of which made it difficult to interpret the efficacy results. The committee accepted that talimogene laherparepvec was clinically effective when compared with GM-CSF, although it also acknowledged that this was based on a post-hoc analysis of a subgroup in the trial, using a comparator that was considered ineffective and is not in clinical use in the NHS.

4.6 The committee acknowledged that it was not feasible for the company to carry out a network meta-analysis because of the lack of a common comparator in the trial network. It also understood that the population in the subgroup in OPTiM for which the licence was granted (stage 3B to 4M1a disease) was not directly comparable with the population in the ipilimumab trials, because there were substantial differences in the patient characteristics. In particular, only 11% to 17% of patients in the ipilimumab trials had stage 3B to 4M1a disease; the others had more advanced melanoma. Also, it was not clear what proportion of the small number of patients with stage 3B to 4M1a disease in the ipilimumab trials had injectable lesions that could have been treated with talimogene laherparepvec.

4.7 The clinical experts stated that there is a lack of evidence on the effectiveness of any melanoma treatments for stage 3B to 4M1a advanced melanoma, and that the OPTiM trial represents the best evidence for this stage of disease. The committee also heard that the disease trajectory of stage III melanoma is likely to differ from that of stage 4M1a, with a different life expectancy, and also noted the clinical expert's comment that as a general rule, earlier-stage disease with a smaller tumour burden is likely to respond better to treatment than later-stage disease.

4.8 The committee noted that the company had explored ways in which talimogene laherparepvec could be compared with ipilimumab for stage 3B to 4M1a disease using the modified and 2‑step Korn methods to correct for differences in patient characteristics between the ipilimumab trials and OPTiM. These adjusted the progression-free and overall survival data from the pooled ipilimumab trials by stage of disease and lactate dehydrogenase (LDH) level in the modified Korn method, and also adjusted for a better disease response in earlier-stage disease in the 2‑step Korn method. The 2 different estimates of ipilimumab efficacy were then used to calculate the relative effectiveness of talimogene laherparepvec compared with ipilimumab. When the modified Korn method was used (the best case), the adjusted survival estimates for ipilimumab were lower than for talimogene laherparepvec. However, the committee noted that the confidence intervals around the adjusted ipilimumab data overlapped with the talimogene laherparepvec trial results. The committee noted that when the 2‑step Korn method was used, which the company considered to be the 'worst case', the overall survival estimates for ipilimumab were very similar to those for talimogene laherparepvec in the OPTiM trial. The committee acknowledged that the company had made efforts to make a comparison with ipilimumab but noted the uncertainty of that comparison, largely because of the lack of efficacy data for ipilimumab in the relevant population.

4.9 Of the 2 methods used by the company, the committee considered that the modified Korn (best case) was the less reliable because it had heard from the clinical expert that treatment response was likely to be better in early‑stage than in later‑stage disease, and the method did not take this into account. In the 2‑ step Korn method talimogene laherparepvec had not been shown to be superior to ipilimumab. The committee noted the ERG's comment that the company should be complimented on their thorough approach to the problem of defining an appropriate comparison with ipilimumab from the available trial data. However, it accepted the underlying concern of the ERG that the Korn method was flawed for modelling progression in stage 3B to 4M1a disease because the algorithm was developed using data from people with predominantly stage 4M1b and stage 4M1c disease, which have different disease trajectories. It also questioned the inclusion of an adjustment for LDH level in the modified Korn method, because this is of limited relevance for people with stage 3B, stage 3C or stage 4M1a disease. Also, the LDH adjustment had the effect of reducing the influence of other prognostic adjustment factors, leading to a potential overestimate of the efficacy of talimogene laherparepvec compared with ipilimumab. The committee agreed that the modifications to the Korn method (the modified and 2-step Korn) further compounded the underlying issues with the Korn method. The committee concluded that the evidence presented was not sufficient to draw any firm conclusions about the relative clinical effectiveness of talimogene laherparepvec compared with ipilimumab in this patient population.

4.11 The company's model compared talimogene laherparepvec with ipilimumab in people with stage 3B to stage 4M1a melanoma. The committee considered that the 3‑state model structure was similar to models used in other melanoma appraisals and therefore accepted that it was appropriate for decision‑making. The company had used a multi‑stage approach to modelling overall survival based on different data sources. The committee noted the ERG's comments that, in principle, the multi-stage approach (using Kaplan–Meier data directly followed by modelled projections of overall survival) was generally appropriate. However, the ERG questioned the sudden change in the shape of the curve at 62.1 months, and also the removal of any melanoma-related mortality after 10 years. The committee accepted the basic structure of the company's model, but gave further consideration to the assumptions used in the modelling of survival.

4.12 The committee discussed the extrapolation of overall survival data in the talimogene laherparepvec arm of the company's model (based on the entire Kaplan–Meier curve to 60 months) and the ERG's exploratory analysis (which used a 2-part exponential model from 9 to 47 months, when the last death was recorded). These different approaches led to 2 divergent survival trends resulting in very different estimates of long-term survival for patients who had talimogene laherparepvec. The committee heard from the company's representative that it considered the entire Kaplan–Meier curve to be most relevant for the purposes of extrapolation because it uses the full extent of the trial follow‑up data. The committee noted that 24% of patients in the talimogene laherparepvec arm were alive at 47 months (when the estimated overall survival was 49%) and remained so at 60 months. The ERG stated that the Kaplan–Meier method estimates survival only for those time points when a death occurred, and therefore only the survival estimates at the time of these events can be legitimately used for fitting projective trends to trial data. Extending the data used for survival estimation beyond the last recorded death, as the company had done, involves assuming that, across an extended time period in which no deaths occur and beyond, any patient still alive can be expected to remain indefinitely free of the risk of death from any cause (not just melanoma). The ERG did not consider this method of extrapolating survival beyond 47 months to be plausible. The ERG also referred to the results of the Kaplan–Meier analysis from OPTiM, which indicated that, following the last recorded death at 47 months, 39 patients remained alive and at risk. These were all censored due to the termination of the trial on a particular date, even though the patients were recruited at different times. This means that the true time of death of these patients cannot be determined. The ERG's approach resulted in a reduction in mean overall survival from 108.5 months, as calculated by the company, to 73 months. This was lower than the overall survival in the ipilimumab trials, indicating that overall survival with talimogene laherparepvec could be less favourable than with ipilimumab. The committee noted the ERG's comment that the company had overestimated overall survival with talimogene laherparepvec by between 49% and 59%. The committee expressed concern that it had not seen enough evidence to be confident that talimogene laherparepvec was as clinically effective as ipilimumab or other currently available therapies in people with stage 3B to stage 4M1a melanoma. The committee concluded that, because of the lack of suitable effectiveness inputs in the economic model, it had not been presented with a plausible incremental cost‑effectiveness ratio (ICER) for talimogene laherparepvec compared with ipilimumab.

4.13 The company submitted additional analyses in response to the appraisal consultation document, intended to address uncertainty in the relative clinical effectiveness of talimogene laherparepvec compared with ipilimumab. This included the use of the Korn method for adjusting the effectiveness of ipilimumab data in the intention to treat population from the OPTiM trial (stage 3B to 4M1c). The committee reconsidered the use of Korn methodology for adjusting the baseline characteristics of the ipilimumab trial, including the results of the Korn adjustment in the intention to treat population (stage 3B to 4M1c), which was a broader population than the marketing authorisation of talimogene laherparepvec. These results suggested that talimogene laherparepvec was at least as effective as ipilimumab. The committee noted that these analyses did not address the underlying methodological concern that the Korn algorithm (which was based predominantly on patients with later‑stage disease) was not valid because it had not been calibrated against patient‑level data from ipilimumab trials in a similar population to the OPTIM trial (see section 4.9).

4.14 In response to consultation the company also submitted a 'naive' indirect comparison of talimogene laherparepvec with ipilimumab in which GM‑CSF, dacarbazine and gp100 were assumed to be equally ineffective in the treatment of metastatic melanoma. But the committee did not consider it to be a reliable method of establishing the relative effectiveness of these agents. The committee appreciated that the company had made every reasonable effort to adjust the ipilimumab data, but there is no methodologically valid way of comparing talimogene laherparepvec with ipilimumab in stage 3B to 4M1a melanoma. The committee noted the proven long‑term survival benefit in a proportion of patients who had ipilimumab (based on 5‑year overall survival data) and concluded that it is not possible to resolve the uncertainty about the relative effectiveness of talimogene laherparepvec compared with ipilimumab (and other newer systemically administered therapies). The committee considered that it needed to be very confident that talimogene laherparepvec is at least as effective as ipilimumab before recommending it as an option for all patients in the licensed population, given that ipilimumab monotherapy has been increasingly replaced by newer therapies that have shown better short‑term effectiveness in clinical trials, with lower toxicity.

4.15 The committee considered additional analyses on the cost effectiveness of talimogene laherparepvec compared with dacarbazine (which has not been shown to prolong overall survival), and best supportive care that were submitted as part of the company's response to consultation. The committee noted that the ICERs for talimogene laherparepvec compared with dacarbazine and best supportive care were approximately £23,900 and £24,100 per QALY gained, and were substantially lower than the corresponding ICERs for ipilimumab compared with dacarbazine and best supportive care (approximately £47,900 and £42,200 per QALY gained, respectively, in NICE technology appraisal guidance on ipilimumab for previously untreated advanced melanoma and on ipilimumab for previously treated advanced melanoma). The committee noted that these figures applied to patients at different stages of disease and were not directly comparable, and so could not be used to draw conclusions about the relative cost effectiveness of these agents.

4.16 The committee considered whether there may be a subgroup of patients for whom talimogene laherparepvec would be particularly beneficial, in particular whether there was a group of patients for whom talimogene laherparepvec might be the only effective option, such as those for whom systemic immunotherapy was contraindicated. The clinical expert, in response to consultation, had highlighted that there were people with BRAF‑negative disease for whom systemically administered immunotherapy is not suitable and who currently had no other effective treatment options. The committee noted the cost-effectiveness analyses presented comparing talimogene laherparepvec with dacarbazine and best supportive care. While these analyses did not specifically relate to a population with melanoma for whom systemically administered immunotherapies were not suitable, the committee was satisfied that they gave an indication of the cost effectiveness of talimogene laherparepvec in this situation. It concluded that talimogene laherparepvec is a clinically and cost-effective option for people with unresectable non-visceral metastatic melanoma for whom systemically administered immunotherapies are not suitable.

4.17 The company stated that talimogene laherparepvec is innovative and a step change in the management of advanced melanoma because it has a novel mechanism of action, in that it produces local tumour control and leads to a systemic anti-tumour immune response. Also, it is the only treatment approved specifically for people with regionally or distantly metastatic melanoma with no visceral disease (stage 3B to stage 4M1a) and is associated with fewer treatment-related grade 3 and 4 adverse events compared with existing treatments. The committee agreed that intra-lesion injections are an innovative approach to the treatment of melanoma, although the marketing authorisation did not support the systemic action of talimogene laherparepvec. The committee also noted that talimogene laherparepvec is being investigated as a combination therapy with other agents, which it considered may be important in the future. However, the committee could not identify any specific health‑related benefit that had not already been captured in the QALY calculation.