1 Recommendations

Non-small-cell lung cancer (NSCLC) is the most common type of lung cancer. The KRAS oncogene is the most commonly mutated gene in lung cancer. The KRAS G12C mutation occurs in 11% of NSCLC tumours in the UK. This mutation is more common in non-squamous NSCLC and does not usually occur with other known mutations such as EGFR, ALK and ROS-1. People with previously treated KRAS G12C mutation-positive locally advanced or metastatic NSCLC usually have chemotherapy, which is a non-targeted treatment associated with adverse effects that affect health-related quality of life. Sotorasib is the only available targeted treatment for the KRAS G12C mutation. It is only recommended for use within the Cancer Drugs Fund (see NICE's technology appraisal guidance on sotorasib for previously treated KRAS G12C mutation-positive advanced non-small-cell lung cancer). The patient and professional organisations and clinical expert submissions highlighted that people with KRAS G12C mutation-positive NSCLC have a poor prognosis. They noted that there is an unmet need for effective and tolerable treatments in this population. The patient organisation also highlighted that the lack of targeted treatment options and having symptoms that are difficult to treat, can be distressing for patients and their carers. The company submission also emphasised the psychological and financial impact on patients and their carers. The committee concluded that there is an unmet need for additional targeted treatments for KRAS G12C mutation-positive locally advanced NSCLC, and that these would have physical, psychological and financial benefits.

According to the conditional marketing authorisation (see section 2.1), people having adagrasib must have previously had, or intolerance to, 'platinum-based chemotherapy and/or anti-PD-1/PD-L1 immunotherapy'. The company submission compared adagrasib with docetaxel and docetaxel plus nintedanib in line with the NICE scope. Sotorasib is not a relevant comparator to adagrasib because it is recommended within the Cancer Drugs Fund and so is not considered to be part of routine practice. A clinical expert submission highlighted that people either have first-line platinum-based chemotherapy with immunotherapy, or single-agent immunotherapy based on PD-L1 expression, disease performance status and co-morbidities. People with a poor performance status may have platinum-based chemotherapy only. The clinical expert submission further highlighted that on disease progression after first-line treatment, a KRAS G12C inhibitor treatment can be considered. Sotorasib and adagrasib are KRAS G12C inhibitors. After first-line single-agent immunotherapy, platinum doublet therapy is used before KRAS G12C inhibitor treatment. After platinum-based chemotherapy, people could have docetaxel with or without nintedanib. The clinical experts at the committee meeting agreed that adagrasib would be used instead of docetaxel with or without nintedanib if it was available. They agreed that the company's positioning of adagrasib in the treatment pathway for KRAS G12C mutation-positive NSCLC is broadly reflective of NHS clinical practice. The committee acknowledged that because sotorasib is recommended for use within the Cancer Drugs Fund, it is not available in routine practice and so is not a relevant comparator for this evaluation. The committee concluded that docetaxel and docetaxel plus nintedanib are the relevant comparators to adagrasib for this evaluation.

The main clinical-effectiveness evidence for adagrasib came from KRYSTAL-12. This is an ongoing phase 3, multicentre, open-label clinical trial for KRAS G12C mutation-positive advanced NSCLC in adults who have had platinum-based chemotherapy and anti-PD-1 or anti-PD-L1 immunotherapy. It compares adagrasib (n=301) with docetaxel (n=152). People in the trial had up to 4 previous lines of systemic treatments, measurable disease according to RECIST 1.1 criteria and an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1. The company presented KRYSTAL-12 data from the 31 December 2023 data cut. The median duration of follow up was 9.43 months. Adagrasib significantly reduced the risk of progression by 42% (hazard ratio 0.58, 95% confidence interval [CI] 0.45 to 0.76) compared with docetaxel. For the objective response rate, adagrasib was associated with a significantly higher likelihood of response (odds ratio 4.68, 95% CI 2.56 to 8.56) compared with docetaxel. Progression-free survival 2 (PFS2) was also reported as an exploratory outcome in KRYSTAL-12. This provides information in cases when analysis of overall survival could be confounded by subsequent treatments. The results for PFS2 closely overlapped between adagrasib and docetaxel arms following subsequent treatment (the exact results are confidential and cannot be reported here). The company did not present the results of the interim overall-survival analysis because it considered these highly immature and inconclusive.

The committee noted that adagrasib is associated with central nervous system (CNS) penetration and may potentially have some CNS benefits. But it also noted that both KRYSTAL-12 and KRYSTAL-1 excluded people with active brain metastases. The company highlighted that KRYSTAL-12 evaluated intracranial efficacy of adagrasib in people with treated, neurologically stable CNS metastases only (n=78 in the adagrasib arm and n=36 in the docetaxel arm). But phase 1b of KRYSTAL-1 evaluated intracranial efficacy in people with untreated, neurologically stable, asymptomatic CNS metastases (n=25). Adagrasib showed some intracranial response. But the company acknowledged that none of the patients in the trials had active brain metastases. A clinical expert at the committee meeting highlighted that the intracranial response seen with adagrasib from KRYSTAL-1 was consistent with that seen for tyrosine kinase inhibitors penetrating the brain. The NHS England Cancer Drugs Fund clinical lead highlighted that there is good biological plausibility for adagrasib to penetrate the brain because adagrasib is a small-molecule medicine. The committee concluded that adagrasib likely penetrates the blood-brain barrier and provides some intracranial response in stable brain metastases.

KRYSTAL-12 has an open-label design so people know which intervention they are having. Early withdrawals before having any treatment were significantly lower in the adagrasib arm (n=3, 1.0%) than in the docetaxel arm (n=12, 7.9%), with 11 withdrawals in the docetaxel arm being 'withdrawal by subject'. After treatment initiation, 'withdrawal by subject' was nearly twice as high in the docetaxel arm (n=21, 13.8%) compared with the adagrasib arm (n=21, 7.0%). Some people in both treatment arms also had subsequent treatment before their cancer progressed (the exact numbers are confidential and cannot be reported here). The company assessed a low risk of bias in KRYSTAL-12. It highlighted that the lack of blinding was unlikely to have substantially affected the interpretation of response or progression because these endpoints were assessed by blinded independent central review. The EAG highlighted that people's knowledge of the intervention that they were having (and possible expectation of higher benefit with adagrasib over docetaxel), may have affected patient retention in the docetaxel arm. But the company provided additional sensitivity analyses to explore the potential impact of early asymmetric dropout and informative censoring (people in the docetaxel arm of KRYSTAL-12 who crossed over to adagrasib) on progression-free survival. The results of these analyses are confidential and cannot be reported here. The EAG considered that it is unclear if the prognosis of people who remained in the docetaxel arm was balanced with those remaining in the adagrasib arm. So, it is uncertain if withdrawal in KRYSTAL-12 may have biased overall survival, objective response rates and safety outcomes. Adagrasib was associated with reduced symptom burden and improved health-related quality of life compared with docetaxel. But the EAG said that despite this, the subjective nature of patient-reported outcomes and lack of blinding means that health-related quality-of-life outcomes in KRYSTAL-12 may have been overestimated. The US Food and Drug Administration (FDA) requested a protocol amendment in KRYSTAL-12, in line with CodeBreaK 200. CodeBreaK 200 was a phase 3, open-label randomised controlled trial of the safety and efficacy of sotorasib compared with docetaxel in people with previously treated KRAS G12C mutation-positive NSCLC. The FDA's protocol amendment permitted crossover to adagrasib in people assigned to the docetaxel arm. At clarification the company reported that a proportion of people in the docetaxel arm had a KRAS G12C inhibitor as subsequent treatment. The subset of people in the docetaxel arm who crossed over to adagrasib or had another KRAS G12C inhibitor had shorter mean time since metastatic diagnosis (11.5 months compared with 15.0 months). The EAG highlighted that it is unclear how many people in the docetaxel arm crossed over to adagrasib or were given another subsequent KRAS G12C inhibitor before confirmed disease progression. The EAG also highlighted that people in both treatment arms may have benefited from adagrasib because of the crossover. They said that a subset of people in the docetaxel arm may also have benefited from subsequent sotorasib or an unlicensed KRAS G12C inhibitor. The committee acknowledged the limitations associated with the design of KRYSTAL-12, including lack of blinding, risk of bias, crossover and subsequent treatments in the docetaxel arm. It noted that this may introduce uncertainty in the overall survival, objective response rates, safety outcomes and health-related quality-of-life outcomes from KRYSTAL-12. But it acknowledged that KRYSTAL-12 is the most robust source of evidence for adagrasib for previously treated KRAS G12C mutation-positive NSCLC. The committee concluded that, despite limitations with the trial design, KRYSTAL-12 provides the best available evidence to inform the clinical effectiveness of adagrasib.

The company did not present the results of the overall-survival analysis in its original submission. This is because it considered the interim overall-survival results from KRYSTAL-12 to be highly immature and inconclusive, and the presence of the potential confounding effect of crossover had not been adjusted for. But the company did provide the results following a request from the EAG at clarification. The results indicated that adagrasib is associated with a non-statistically significant hazard ratio compared with docetaxel (the exact results are confidential and cannot be reported here). The EAG highlighted that using immature overall-survival data to quantify the treatment effect for adagrasib compared with docetaxel leads to substantial uncertainty in the cost-effectiveness analysis. It advised that in the absence of crossover-adjusted analyses, the potential impact of crossover and subsequent KRAS G12C inhibitors in the docetaxel arm on overall survival is uncertain. The EAG also advised that more mature evidence was needed from KRYSTAL-12 to address the concerns around the overall-survival results from KRYSTAL-12. To explore alternative scenarios around overall survival benefit for adagrasib, the EAG looked at evidence from CodeBreaK 200 (see section 3.5). CodeBreaK 200 showed that sotorasib (a biologically similar treatment to adagrasib licensed for previously treated KRAS G12C mutation-positive advanced NSCLC), had a statistically significant progression-free survival benefit compared with docetaxel (hazard ratio 0.66, 95% CI 0.51 to 0.86). But it did not show a statistically significant overall-survival benefit for sotorasib (hazard ratio 1.01, 95% CI 0.77 to 1.33). CodeBreak 200 allowed for crossover of patients from the comparator arm to the intervention arm upon disease progression. However, when different crossover-adjustment methods were applied, the treatment benefit varied. To explore the uncertainty in the overall-survival results for KRYSTAL-12, the EAG used 2 alternative estimates of overall survival in its base case:

• EAG base case 1 assumed no effect of adagrasib on overall survival compared with docetaxel (hazard ratio 1)

• EAG base case 2 assumed the same effect of adagrasib as sotorasib (the only other KRAS inhibitor with data in the relevant population) on overall survival. This was based on the effect from the 2-step crossover-adjusted analysis for sotorasib compared with docetaxel from CodeBreaK 200 (hazard ratio 0.89).

The company did a post-hoc exploratory surrogacy analysis using overall-survival data from KRYSTAL-1 to:

• assess the relationship between time to progression and time to death at the individual patient level for KRAS G12C mutation-positive NSCLC

• predict overall survival for KRYSTAL-12 in censored patients for both adagrasib and docetaxel using the surrogacy relationship derived from KRYSTAL-1 applied to progression data from KRYSTAL-12.

In the absence of a direct comparison between adagrasib and docetaxel plus nintedanib, the company did an indirect treatment comparison. For progression-free survival and safety outcomes it used available data from the most recent data cut-offs from KRYSTAL-12 (adagrasib), LUME-Lung1 (docetaxel plus nintedanib) and CodeBreaK 200 (sotorasib). For overall-survival outcomes, the company used the most recent data cut-off from LUME-Lung1 and CodeBreaK 200. It also used overall survival estimates from the surrogacy analysis of KRYSTAL-1 to provide estimates of overall survival for KRYSTAL-12. For both progression-free survival and overall survival, the company used the fixed-effects time-varying network meta-analysis using the gamma curve in its base case. The EAG raised several concerns with the validity of the network meta-analysis results because of limitations in the clinical evidence, including:

• the validity of the overall-survival data from the company's surrogacy analysis (see section 3.7).

• the risk of bias in KRYSTAL-12 (see section 3.5).

• the population from LUME-Lung 1, which had limited comparability with KRYSTAL-12 and CodeBreaK 200, most notably in prior immunotherapy exposure and histology. People in LUME-Lung 1 were recruited between 2008 and 2011, so may not have benefited from the improvements in treatment and managing tolerability and safety over the past decade. So, it was uncertain whether prior immunotherapy exposure may affect the relative overall-survival benefits of docetaxel plus nintedanib compared with docetaxel monotherapy. Also, adenocarcinoma histology might be a treatment-effect modifier for network meta-analysis comparisons between docetaxel plus nintedanib and adagrasib or docetaxel monotherapy, although the evidence is limited and uncertain.

• Adagrasib, docetaxel plus nintedanib and sotorasib were only directly compared with docetaxel, so the absence of loops in the network and the limited number of studies prevents any assessment of consistency and heterogeneity.

• substantial quality issues with CodeBreaK 200 because of concerns with early asymmetric dropout, censoring and crossover, the duration of interval between assessments, and lack of blinding.
  
  So, the EAG advised that the results of the network meta-analysis should be interpreted with caution. It also highlighted that incorporating results from the subgroup of people with adenocarcinoma from LUME-Lung 1 may partially address concerns about the comparability of this trial. For progression-free survival and safety outcomes, the committee acknowledged that there was little the company could do to improve its approach to doing these network meta-analyses. So, the committee concluded that this was acceptable for decision making. But the committee noted the uncertainty introduced by using the overall-survival data from the company's surrogacy analysis. It concluded that the overall-survival results from the company's network meta-analyses to inform the economic model were not suitable for decision making.

In KRYSTAL-12, the safety population was defined as everyone who had any part of a dose of study medication. This was 298 people in the adagrasib arm and 140 people in the docetaxel arm. Everyone in the adagrasib arm and 98.6% of people in the docetaxel arm experienced treatment-emergent adverse events. Fatal treatment-emergent adverse events occurred in 48 (16.1%) people in the adagrasib arm and 10 (7.1%) people in the docetaxel arm. Grade 3 or higher treatment-emergent adverse events occurred in 213 (71.5%) people in the adagrasib arm and 93 (66.4%) people in the docetaxel arm. Serious treatment-emergent adverse events occurred in 149 (50.0%) people in the adagrasib arm and 50 (35.7%) people in the docetaxel arm. Treatment-emergent adverse events led to treatment discontinuation for 40 (13.4%) people in the adagrasib arm and 25 (17.9%) people in the docetaxel arm. Treatment-emergent adverse events that led to dose reductions or interruptions were reported in 237 (79.5%) people in the adagrasib arm and 67 (47.9%) people in the docetaxel arm.

The company used a partitioned survival model with 3 health states:

• progression-free

• progressed

• death.
  
  The cycle length was 1 week, and the time horizon was 20 years. Health-state occupancy for adagrasib, docetaxel and docetaxel plus nintedanib was informed by overall-survival and progression-free survival estimates from the company's time-varying network meta-analysis. For adagrasib and docetaxel, the company's surrogacy analysis used KRYSTAL-1 individual patient data to predict KRYSTAL-12 overall survival that was used in the overall-survival network meta-analysis. Treatment discontinuation was informed from extrapolated progression-free survival curves for each treatment and adjusted to reflect the relationship between time to treatment discontinuation and progression-free survival. The EAG advised the company's base-case model structure is consistent with previous NICE evaluations on advanced or metastatic NSCLC at the second line of treatment. The committee noted the uncertainty in the overall-survival data. It recalled its conclusion on overall survival that the EAG's base case 1 which assumes no effect of adagrasib on overall survival compared with docetaxel is the most plausible. But it would consider the company trying different methods to inform the long-term survival estimates (see section 3.6). Overall, the committee agreed the company's model was similar to previous models used in NSCLC. It concluded that the model structure was appropriate for decision making.

NICE's manual on health technology evaluations notes that, above a most plausible incremental cost-effectiveness ratio (ICER) of £20,000 per QALY gained, judgements about the acceptability of a technology as an effective use of NHS resources will consider the degree of certainty around the ICER. The committee will be more cautious about recommending a technology if it is less certain about the ICERs presented. But it will also consider other aspects including uncaptured health benefits. The committee noted the high level of uncertainty, specifically, the:

• limitations of KRYSTAL-12 and uncertainty in the overall survival, objective response rates, safety outcomes and health-related quality-of-life outcomes (see section 3.5)

• effect of adagrasib relative to docetaxel or docetaxel plus nintedanib on overall survival (see section 3.6)

• company's surrogacy analysis to predict overall survival for KRYSTAL-12 from overall survival from KRYSTAL-1 (see section 3.7)

• uncertainty in the company's network meta-analysis results, in particular overall survival (see section 3.8)

• serious adverse events associated with adagrasib that were factored into the decision making on the clinical effectiveness (see section 3.9)

• insufficient justification of the different utility values in the progressed-disease health state between the treatment arms (see section 3.11).
  
  The committee concluded that given the considerable uncertainties, and to minimise risks posed by opportunity cost of other treatments and services being displaced, an acceptable cost-effectiveness estimate would not be higher than £20,000 per QALY gained.

Because of confidential commercial arrangements for adagrasib and some of the comparators, the exact cost-effectiveness results are confidential and cannot be reported here.

The committee's preferred assumptions included:

• Assuming no effect of adagrasib on overall survival compared with docetaxel (hazard ratio of 1). Alternatively, the committee would consider the company trying different methods to inform the long-term survival estimates including:

  • interim overall survival from KRYSTAL-12 extrapolated for the model time horizon, using methods described in NICE's Technical Support Document 14

  • interim overall survival from KRYSTAL-12 adjusted for treatment switching using methods described in NICE's Technical Support Document 16, and overall-survival outcome from each crossover-adjustment method extrapolated for the model time horizon (see section 3.6).

• In the progressed-disease health state, using the same health-state utility value for all treatments (based on the confidential value of the docetaxel arm). The committee would consider scenarios using the same progressed-disease health-state utility value (based on the confidential value reported for the adagrasib arm) and scenarios with differential utility values with further justification (see section 3.11).
  
  The exact ICERs are confidential and cannot be reported here. The company's base case incorporated the surrogacy analysis, which the committee found unsuitable for decision making (see section 3.7). The EAG's base case 1 assumed there is no effect of adagrasib on overall survival (hazard ratio of 1) compared with docetaxel and docetaxel plus nintedanib. It also used the same progressed-disease health-state utility value for all treatments using the value in the adagrasib arm (which is confidential so cannot be reported). Using the EAG's base case 1, the ICER for adagrasib compared with docetaxel was above the range that NICE considers a cost-effective use of NHS resources. The EAG's base case 2 assumed that adagrasib has the same effect on overall survival as sotorasib (hazard ratio 0.89). It also used the same progressed-disease utility value for all treatments using the value in the adagrasib arm (which is confidential so cannot be reported). Using the EAG's base case 2, the ICER for adagrasib compared with docetaxel was above the range that NICE considers a cost-effective use of NHS resources. In both the EAG's base cases, docetaxel with nintedanib provided fewer health benefits at higher cost compared with a combination of adagrasib and docetaxel monotherapy. So, the committee concluded that adagrasib could not be recommended for routine commissioning.

Having concluded that adagrasib could not be recommended for routine use in the NHS, the committee then considered if it could be recommended for use during a managed-access period for treating KRAS G12C mutation-positive advanced NSCLC. But the committee noted that the company had not made a managed-access proposal for adagrasib for inclusion in the Cancer Drugs Fund. The committee noted that for it to be considered for the Cancer Drugs Fund adagrasib needs to have plausible potential to be cost effective. But the committee recalled the uncertainties in the clinical evidence and economic model. It also noted that the most likely cost-effectiveness estimates using its preferred assumptions are uncertain but are most likely above the range that NICE considers a cost-effective use of NHS resources. So, the committee concluded that, in the absence of a managed-access proposal, adagrasib could not be considered for use with managed-access. The committee also considered that if a managed-access proposal was made during consultation, adagrasib would still not be considered for use with managed-access because currently it is not plausibly cost effective.

The committee did not identify any equality issues.

The committee considered whether there were any uncaptured benefits of adagrasib. It did not identify additional benefits of adagrasib not captured in the economic modelling. So, the committee concluded that all additional benefits of adagrasib had already been taken into account.

The clinical-effectiveness evidence for adagrasib is uncertain because of the design of KYSTAL-12, immature interim overall-survival results and uncertainties in the company's surrogacy analysis. There are uncertainties in the economic model, including with the utility values for adagrasib and the comparators used in the modelling. The most likely cost-effectiveness estimates for adagrasib are substantially above the range that NICE considers an acceptable use of NHS resources. So, adagrasib should not be used.