Lenvatinib plus pembrolizumab for untreated advanced renal cell carcinoma: a systematic review and cost-effectiveness analysis

Incidence of disease

Between 2015 and 2017, there were 13,055 new cases of kidney cancer in the UK (England: 10,759; Wales: 631). 12 Worldwide, kidney cancer is twice as common in men than in women. 4 In the UK, between 2015 and 2017, there were 1.7 times more new cases in men (62.8%) than in women (37.2%);12 a quarter (25.5%) of cases were diagnosed in people aged 60–69 years, with nearly half of the cases (49.2%) diagnosed in people aged ≥ 70 years. 12

Incidence and death rates by stage of disease

In England, between 2013 and 2017, 43.0% of all cases of kidney cancer with a known stage at diagnosis were classified as being advanced, that is Stage 3 or Stage 4 (see Table 1). During this period, the 5-year relative survival rates by stage of disease were markedly lower for patients with Stage 4 (metastatic) disease than for patients with the other stages of kidney cancer, including Stage 3 (locally advanced) disease (see Table 1).

Incidence and death rates by disease risk status

Two models commonly used to classify risk status are the Memorial Sloan-Kettering Cancer Center (MSKCC) risk stratification model13,14 and the International Metastatic Renal Cell Carcinoma Database Consortium (IDMC) model. 15,16 As highlighted in the Eisai company submission (CS),1 the former ‘was originally the gold standard method for assessing risks associated with targeted treatment in metastatic RCC, and is still considered relevant by UK clinicians today to estimate patient prognosis’ and the latter ‘was developed to extend the MSKCC criteria to increase concordance, and is primarily applied in UK clinical practice’.

Both models13–16 calculate patients’ risk of progression taking into consideration a number of specific prognostic risk factors. The following risk factors are common to both models:13–16 time from diagnosis to treatment, haemoglobin levels, calcium levels and Karnofsky Performance Status (KPS). The MSKCC model also includes lactate dehydrogenase concentration, and the IMDC model also considers absolute neutrophil count and platelet count. 13–16 Both models13–16 classify risk as favourable (no adverse prognostic risk factors), intermediate (one or two adverse prognostic risk factors) or poor (three or more adverse prognostic risk factors). In a study to validate the IMDC, Heng et al. 16 reported that 83% of patients were classified into the same risk subgroup by both models.

The proportions of patients with metastatic RCC who belong to each risk subgroup in eight population-based studies16–23 are presented in Table 2.

The OS estimates are reported by risk subgroup in six population-based studies16–21 of patients with metastatic RCC who received sunitinib as a first-line treatment (see Table 3). Three19,21,23 of the four most recently published studies included in Table 2 also considered prognosis based on whether patients with intermediate-risk status had one or two prognostic factors.

Some drugs are recommended only by NICE24,25 for patients with IMDC intermediate or poor (intermediate/poor) risk. Only one of the population studies (Savard et al. 21) listed in Table 3 reported OS for the combined IMDC intermediate-/poor-risk subgroup. The reported median OS for this subgroup was 23.2 [95% confidence interval (CI) 21.0 to 25.8] months. In the total (all-risk) population, median OS was 28.6 (95% CI 25.9 to 31.0) months, whereas median OS for the IMDC favourable-risk population was 52.1 (95% CI 43.4 to 61.2) months. Information on treatment options for patients in different IMDC risk subgroups is provided in Current service provision.

Surgery

Surgery is usually possible, and is the preferred treatment, for patients with early RCC and patients with locally aRCC29 and is usually curative. However, results from two studies30,31 that have explored disease progression following surgery suggest that approximately 30% of patients who have received surgery subsequently develop metastatic RCC. Surgery is rarely a treatment option for patients with metastatic RCC.

National Institute for Health and Care Excellence guidance for first-line drug treatment

At the time of this appraisal, the NICE-recommended treatments (see Table 4) are systemic vascular endothelial growth factor receptor (VEGFR)-targeted tyrosine-kinase inhibitor (TKI) agents (sunitinib,32 pazopanib,33 tivozanib34 and cabozantinib24). Two-drug combination treatments have been made available to patients via the Cancer Drugs Fund (CDF): avelumab plus axitinib35 [a programmed-death ligand 1 (PD-L1) checkpoint inhibitor in combination with a VEGFR-TKI] and nivolumab plus ipilimumab25 [a programmed death cell protein 1 (PD-1) inhibitor and a cytotoxic T-lymphocyte antigen 4 (CTLA-4) checkpoint inhibitor]. Nivolumab plus ipilimumab was subsequently recommended by NICE as a routine treatment option for patients with intermediate-/poor-risk aRCC (TA78036) on 24 March 2022. Although licensed for treating patients with aRCC, pembrolizumab plus axitinib is not recommended by NICE37 and so is not used in NHS clinical practice. Treatment options that are now rarely used due to their associated toxicities6 are cytokines (interferon alpha and high-dose interleukin-2).

European clinical guidelines for first-line drug treatment

Clinical practice guidelines published in 2021 by the European Association of Urology38 and the European Society for Medical Oncology (ESMO)39 recommend four combination treatments for the first-line treatment of metastatic clear cell RCC: lenvatinib plus pembrolizumab, pembrolizumab plus axitinib and nivolumab plus cabozantinib for intermediate-/poor-risk or favourable-risk disease and nivolumab plus ipilimumab for intermediate-/poor-risk disease only. For patients who cannot tolerate immune checkpoint inhibitors, the European Association of Urology38 recommend sunitinib, pazopanib and cabozantinib for intermediate-/poor-risk disease and sunitinib and pazopanib for favourable-risk disease. The AG highlights that pembrolizumab plus axitinib is not recommended by NICE37 and nivolumab plus cabozantinib has not been appraised by NICE, the planned single technology appraisal (STA) being suspended. 41

NHS first-line treatment options

Clinical advice to the AG is that in NHS clinical practice, patients with aRCC receive the treatments recommended in NICE guidance24,25,32–35 (see Table 4) and that treatment decisions are made based on histological subtype, IMDC disease risk category, patient age and comorbidities, patient fitness, disease aggressiveness/biology and patient preference.

In line with recommendations in NICE guidance,24,36 at the time of this appraisal, the clinical advice to the AG is that, in general, nivolumab plus ipilimumab is the preferred first-line treatment option for patients with intermediate-/poor-risk disease and that cabozantinib is the preferred treatment option for fitter patients in this subgroup who have rapidly progressing disease (approximately 20%). The clinical advice to the AG is also that patients unable to tolerate either of these treatments receive sunitinib, pazopanib or tivozanib.

The treatment options available in NHS clinical practice to patients with favourable-risk disease at the time of this appraisal are sunitinib, pazopanib or tivozanib and, via the CDF, avelumab plus axitinib. 35 The clinical advice to the AG is that, where available, avelumab plus axitinib is the preferred first-line treatment option for patients with favourable-risk disease who can tolerate this combination, and tivozanib is the favoured treatment option for patients who are able to tolerate only VEGFR-TKI monotherapy.

Subsequent lines of drug treatment

The NICE has recommended five treatment options24,25,32–34 for previously treated patients with aRCC (Table 5). All of these subsequent treatments are recommended for patients regardless of their risk status. The clinical advice to the AG is that cabozantinib and nivolumab monotherapy are the most commonly used second-line treatments; lenvatinib plus everolimus is not a treatment option for patients who have previously received lenvatinib.

The ESMO39 recommends axitinib, cabozantinib and lenvatinib plus everolimus, which are all recommended by NICE,42,43,44 and sunitinib, pazopanib and tivozanib.

Focus of the systematic reviews of lenvatinib plus pembrolizumab

Six of the reviews53–57,59 focused on the efficacy and safety of treatment, and one review58 focused only on safety. One review56 compared lenvatinib plus pembrolizumab versus other combination therapies and versus sunitinib. Six other reviews53–55,57–59 assessed the evidence for lenvatinib plus pembrolizumab and other combination therapies versus sunitinib; three reviews54,55,59 presented only pooled results and two reviews57,58 compared lenvatinib plus pembrolizumab versus other combination therapies by ranking the probability of maximal efficacy.

The therapies included in the seven reviews53–59 were a combination of PD-1 and CTLA-4 checkpoint inhibitors (nivolumab plus ipilimumab),54,56–59 a PD-L1 checkpoint inhibitor in combination with an angiogenesis inhibitor (atezolizumab plus bevacizumab54,55,57–59), a PD-L1 checkpoint inhibitor in combination with VEGFR-TKI (avelumab plus axitinib53–55,57–59) or a PD-1 checkpoint inhibitor in combination with VEGFR-TKI (pembrolizumab plus axitinib53–59 or nivolumab plus cabozantinib53–59). Three reviews55,57,59 included subgroup analyses by risk subgroup and one review53 included only favourable-risk patients.

Results from the systematic reviews of lenvatinib plus pembrolizumab

Patient population

In previous NICE appraisals of treatments for untreated aRCC,25,35 NICE appraisal committees (ACs) noted that there was a lack of evidence to guide treatment decisions for patients with non-clear cell RCC. This is primarily due to non-clear cell RCC being (1) heterogeneous (up to 15 different subtypes are listed in the most recent World Health Organization classification of RCC9) and (2) less common9,10 than clear cell RCC. The AG made no attempt to provide evidence separately for patients with clear cell and non-clear cell histologies.

As noted in Current service provision, decisions about the most appropriate first-line treatments for patients with aRCC are now typically made based on patient risk subgroup. Therefore, the AG conducted subgroup analyses for intermediate-/poor-risk and favourable-risk subgroups.

Unless otherwise stated, risk subgroup within this report refers to IMDC model risk stratification subgroups.

Comparators

Four of the five comparators listed in the final scope29 issued by NICE (sunitinib, pazopanib, tivozanib, and cabozantinib for patients with intermediate-/poor-risk aRCC) are all used in current NHS clinical practice. Nivolumab plus ipilimumab is also listed as a comparator; however, at the time of writing this AG report, nivolumab plus ipilimumab was subject to an ongoing CDF review25 and was not available for routine use in the NHS. Following advice from the NICE technical team, the AG has included nivolumab plus ipilimumab as a relevant comparator. Nivolumab plus ipilimumab was subsequently recommended by NICE as a routine treatment option for patients with intermediate-/poor-risk aRCC (TA78036) on 24 March 2022.

Subgroup analyses

In line with the final scope29 issued by NICE, the AG carried out clinical and cost-effectiveness analyses of lenvatinib plus pembrolizumab for the subgroup of patients with intermediate-/poor-risk disease. While it is stated in the AG protocol that analyses would be undertaken separately for the two subgroups, the AG has carried out analyses only for the combined intermediate-/poor-risk subgroup; clinical advice to the AG is that, in line with NICE guidance,24,36 treatment decisions are based on the combined intermediate-/poor-risk disease category (one category, not two categories). If a patient does not have intermediate-/poor-risk disease then, by definition, the patient has favourable-risk disease; hence the AG has carried out subgroup analysis for the subgroup of patients with favourable risk.

Search strategies

The clinical effectiveness search strategy was designed to identify randomised controlled trials (RCTs) that met the inclusion criteria for the review of direct clinical effectiveness evidence and to identify RCTs that could potentially be used to populate the AG NMAs. The AG identified clinical effectiveness studies by searching relevant major medical databases, trial registries, conference abstracts, the NICE technology appraisal (TA) website listed in Appendix 2, Table 50. and grey literature websites. The search terms used to search the database are given in Appendix 2.

As part of the MTA process, companies were invited to submit evidence to NICE to inform this appraisal. Two companies provided direct and indirect evidence: Eisai,1 the manufacturer of lenvatinib, and MSD,2 the manufacturer of pembrolizumab. The AG screened the reference lists of the Eisai CS1 and the MSD CS2 alongside all other included reports for relevant studies and consulted the AG clinical experts to identify any relevant studies that may have been missed.

A database of identified published literature was compiled. MEDLINE, EMBASE, PubMed, CENTRAL, International Health Technology Assessment (INAHTA), ClinicalTrials.gov and International Clinical Trials Registry Platform (ICTRP) data were collated in a bibliographic database (Endnote X9 software package63) and exported to a specialist systematic review management system (Covidence Systematic Review software64). Conference abstracts results were screened on organisations’ websites. The search terms used to search each of the databases and the websites are given in Appendix 2.

Inclusion and exclusion criteria: direct evidence

The eligibility criteria used to identify studies for the review of direct clinical effectiveness are listed in Table 8.

Titles and abstracts identified through electronic searches were uploaded to Covidence and screened by two reviewers (NF and either JG or KE). Full-text articles of any titles and abstracts that were considered potentially eligible for inclusion were obtained via online resources, or through the University of Liverpool libraries, and uploaded to Covidence. These full-text articles were assessed for inclusion by two reviewers (NF and either JG or KE). Discrepancies at each stage of screening were resolved via discussion between the three reviewers. Full-text articles that did not meet the inclusion criteria were excluded with reasons for exclusion noted.

In addition to screening the articles exported to Covidence, two out of three reviewers (RH, JG and KE) screened the conference proceedings independently following the eligibility criteria shown in Table 8.

Data extraction and quality assessment strategy: direct evidence

Data relating to study characteristics, population characteristics and outcomes were extracted by one reviewer (NF) into tables and independently checked for accuracy by a second reviewer (SN or KE). Data from multiple publications of the same study were extracted and reported as a single study.

Study quality was assessed independently by two reviewers (JG and KE) using the criteria published in the CRD Guidance for Undertaking Reviews in Healthcare. 60 Disagreements were resolved through discussion and, when necessary, a third reviewer (SN) was consulted.

Statistical approaches for the conduct and analysis of randomised controlled trials: direct evidence

The AG assessed the prespecified statistical approach of the only included RCT. 66 This assessment considered:

• analysis populations

• trial design and sample size

• amendments to the protocol and statistical analysis plan

• definition and analysis approach for primary and secondary efficacy outcomes

• definition and analysis approach for patient reported outcomes (PROs)

• definition and analysis approach for safety outcomes and AEs

• validity of modelling assumptions [e.g. proportional hazards (PH)]

• approach to handling missing data

• subgroup and sensitivity analyses.

The AG also performed an assessment of specific statistical approaches, where appropriate for any relevant study (e.g. analyses to adjust for treatment switching).

Data analysis/synthesis: direct evidence

Progression-free survival results from the CLEAR trial

Key PFS results from the CLEAR trial are summarised in Table 13.

Exploratory subgroup analyses of progression-free survival assessed by Blinded Independent Review Committee

All results from CLEAR trial PFS subgroup analyses for the comparison of lenvatinib plus pembrolizumab versus sunitinib were statistically significantly in favour of lenvatinib plus pembrolizumab (Motzer et al. 2021,66 figure 1B). The AG highlights that these subgroup analyses were not powered to detect statistically significant differences between the two treatment arms.

FIGURE 1.

Structure of MSD and MSD/AG company model. PD = progressed disease.

Subgroup results by MSKCC and IMDC risk subgroups for PFS assessed by BIRC, using both the FDA and European Medicines Agency (EMA) preferred censoring methods, were provided by Eisai and MSD in their CSs (appendices D2.4.21 and D1.1, respectively). The AG highlights that these subgroup analyses were not powered to detect statistically significant differences between the two treatment arms. The data are marked as academic-in-confidence and cannot be presented here.

Overall survival results from the CLEAR trial

Key OS results from the CLEAR trial are presented in Table 14.

Treatment on disease progression and impact on overall survival in the CLEAR trial

In addition to the effect of the study drug, OS results may be influenced by subsequent anticancer treatment(s) received on disease progression. Just under half of all patients in the CLEAR trial received subsequent treatment [IA3 data cut-off (45.4%) and updated OS analysis (49.6%)]. Compared with patients in the lenvatinib plus pembrolizumab arm, at the IA3 data cut-off, 1.7 times as many patients in the sunitinib arm (57.1%) than in the lenvatinib plus pembrolizumab arm (33.0%) received subsequent treatment (71.0% and 54.9%, respectively, of patients who discontinued treatment). At the updated data cut off, the proportion of patients receiving subsequent treatment was 61.9% of all sunitinab arm patients and 37.2% of all lenvatinib plus pembrolizumab patients.

Eisai1 presented analyses of updated OS to attempt to take into account additional treatments received for the all-risk population. Eisai presented a comparison of OS data in each treatment arm for patients who received subsequent treatment, and a comparison of OS data in each treatment arm for patients who did not receive subsequent treatment. All the results are academic-in-confidence and so cannot be presented here. However, the AG highlights that the PH assumption was violated for the analysis of OS data from patients who received subsequent treatment and so the OS HR should not be used to infer magnitude of treatment effect or statistical significance for this comparison. Nonetheless, for patients who did not receive subsequent treatment, the K-M data suggested an OS benefit for patients treated with lenvatinib plus pembrolizumab. However, for patients who did receive subsequent treatment, the K-M data suggested an OS benefit for patients treated with lenvatinib plus pembrolizumab up to approximately 33 months, at which point the curves cross. Eisai also conducted prespecified analyses to adjust OS for the effect of any subsequent anticancer treatment (FAS population, updated OS analysis). These analyses were conducted using the two-stage estimation method with different models [log-normal acceleration factor (AF) with and without re-censoring; log-logistic AF with and without re-censoring; Weibull AF with and without re-censoring]. A summary of the AG checks of the treatment-switching analysis methods used by Eisai is provided in Appendix 3 (see Table 55). The results derived from the analysis were marked as academic-in-confidence.

Objective tumour response results from the CLEAR trial

Key tumour response results, including ORR results, from the CLEAR trial all-risk population are presented in Table 15. All subgroup data were marked as academic-in-confidence and cannot be presented.

Safety results

Safety data from the CLEAR trial were reported (IA3 data cut-off). The AEs were graded using common terminology criteria for adverse event (CTCAE) version 4.03. 87 The safety population included all patients who received at least one dose of either study drug.

The median duration of treatment was longer in the lenvatinib plus pembrolizumab arm than in the sunitinib arm (17.0 vs. 7.8 months).

A summary of treatment emergent adverse events (TEAEs) is presented in Table 16. Patients in the lenvatinib plus pembrolizumab arm experienced more AEs (of any type) than patients in the sunitinib arm. While 37.2% of patients discontinued either lenvatinib and/or pembrolizumab due to TEAEs, 13.4% of patients discontinued both lenvatinib and pembrolizumab and 14.4% of patients discontinued sunitinib due to TEAEs.

The AEs of any cause (any grade in ≥ 25% of patients and Grade ≥ 3 in ≥ 5% of patients) that emerged or worsened during the CLEAR are summarised in Tables 17 and 18, respectively. Nearly all patients in both arms experienced at least one all-grade AE with more Grade ≥ 3 AEs reported in the lenvatinib plus pembrolizumab arm (82.4%) than in the sunitinib arm (71.8%).

The most commonly occurring all-grade AEs in both arms were diarrhoea (61.4% vs. 49.4%) and hypertension (55.4% vs. 41.5%). Hypertension was also the most common Grade ≥ 3 AE in both arms (27.6% vs. 18.8%). The other most common Grade ≥ 3 AEs in the lenvatinib plus pembrolizumab arm were increased lipase (12.8% vs. 8.8%), diarrhoea (9.7% vs. 5.3%), increased amylase (9.1% vs. 2.9%), decreased weight (8.0% vs. 0.3%), proteinuria (7.7% vs. 2.9%) and asthenia (5.4% vs. 4.4%).

MSD2 (p. 69) reported a ‘higher than expected’ incidence of Grade ≥ 3 hepatic AEs. From data presented by the companies [Eisai CS1 (table 20) and MSD CS2 (appendix F, see table 3)], incidences of Grade ≥ 3 alanine aminotransferase increased and Grade ≥ 3 aspartate aminotransferase increased were 4.3% and 3.1%, respectively, in the lenvatinib plus pembrolizumab arm versus 2.4% and 0.9%, respectively, in the sunitinib arm. Grade ≥ 3 blood bilirubin increased in 1.1% of patients treated with lenvatinib plus pembrolizumab and in 0.6% of patients treated with sunitinib. It is reported in the summary of product characteristics (SmPC) for lenvatinib that Grade 3 liver-related reactions occurred in 9.9% of patients in the lenvatinib plus pembrolizumab arm and in 5.3% of patients in the sunitinib arm. 45

MSD2 reported that the most common non-fatal serious adverse events (SAEs) in the lenvatinib plus pembrolizumab arm were diarrhoea (3.4%), vomiting (2.8%), pneumonitis (2.6%), acute kidney injury (2.3%) and hypertension (2.3%), each of which occurred with an incidence ≤ 1.2% in the sunitinib arm [MSD CS2 (appendix F and table 3)]. Pyrexia was the most common SAE in the sunitinib arm (2.1% vs. 1.7% in the lenvatinib plus pembrolizumab arm).

Eisai1 reported that adverse events of special interest (AEOSI) for pembrolizumab were experienced by 60.8% of patients in the lenvatinib plus pembrolizumab arm and 30.9% of patients in the sunitinib arm [Eisai CS1 (appendix F3.2)]. According to the CSR,70 for the comparison of lenvatinib plus pembrolizumab versus sunitinib, the most common AEOSI was hypothyroidism; other AEOSIs reported by ≥5% of patients in the lenvatinib plus pembrolizumab arm were hyperthyroidism, pneumonitis, adrenal insufficiency and severe skin reactions. 70

Health-related quality of life results from the CLEAR trial

In the CLEAR trial, HRQoL was assessed as a secondary end point using the following validated questionnaires: (1) the Functional Assessment of Cancer Therapy Kidney Symptom Index-Disease-Related Symptoms (FKSI-DRS), (2) the European Organisation for the Research and Treatment of Cancer (EORTC) Quality of Life Questionnaire (QLQ-C30) and (3) the European Quality of Life-5 Dimensions-3 Levels Version (EuroQoL EQ-5D-3L). In summary:

1. The FKSI-DRS consists of nine items designed to assess the frequency/severity of symptoms specific to advanced kidney cancer, including fatigue, pain, bone pain, lack of energy, shortness of breath, fevers, weight loss, coughing and blood in the urine. Scores are measured using a 5-point Likert scale, and higher total scores correspond to better HRQoL.

2. The EORTC is a cancer-specific questionnaire consisting of function and symptom scales, which are scored from 0 to 100. Higher scores on the functional scales reflect better HRQoL, and higher scores on the symptom scales reflect worse symptoms.

3. The EQ-5D-3L is used to assess general HRQoL in five domains (mobility, self-care, usual activities, pain/discomfort and anxiety/depression) with three levels of response. Responses are used to generate health state index scores, with higher scores indicating better health. The second part of this questionnaire consists of the visual analogue scale, where patients rate their perceived health on a scale of 0 (worst imaginable health) to 100 (best imaginable health).

Health-related quality of life assessments were performed at baseline, day 1 of each subsequent treatment cycle and at the off-treatment visit (30 days after final dose of study drug). As stated in the Eisai HRQoL outcomes study report,72 completion rates (at least one complete score; FAS population) for all HRQoL instruments were notably different for the two trial arms. The completion rates for any instrument declined below 50% at Cycle 26 for patients treated with lenvatinib plus pembrolizumab and at Cycle 12 for patients treated with sunitinib. The completion rates at the off-treatment visit were 40.0% for patients treated with lenvatinib plus pembrolizumab and 55.7% for patients treated with sunitinib. Compliance was generally greater than 90% in both trial arms during early cycles of treatment; however, at the off-treatment visit, compliance had dropped to approximately 80%.

Studies assessed by the assessment group for potential inclusion in network meta-analysis

Any study identified by the AG searches for direct evidence that appeared to be designed as a RCT of any drug used to treat adults with untreated aRCC was tagged as ‘RCT’ within Covidence (n = 1129 records). These records were then examined by SN to confirm that the study design and the study population were of interest (i.e. RCTs of adults with untreated aRCC) and to identify the drug treatments included in the studies.

In addition, any study previously identified by the AG searches that appeared to be a NMA of RCTs of drugs used to treat adults with untreated aRCC was tagged as a ‘NMA’ within Covidence (n = 36, published from 2009 to 2021). The AG examined the reference lists and network structures of recently published NMAs (n = 1057,88–96), that is, those published since 2020, to assess the feasibility of constructing suitable networks for each outcome listed in the final scope29 issued by NICE.

In total, the AG identified 10 RCTs27,66,97–104 of drug treatments for adults with untreated aRCC that were potentially eligible for inclusion in the AG NMAs.

Assessment group consideration of specific networks

The AG’s assessment of the feasibility of constructing specific networks considered the following:

• the feasibility of constructing a ‘suitable connected network’ of relevant treatments for each outcome and for each risk subgroup

• the clinical and methodological heterogeneity of the included studies in terms of (1) study population, (2) interventions and comparators, (3) outcome measures (OS, PFS, ORR, safety and HRQoL) and (4) study quality.

For each outcome listed in the final scope29 issued by NICE, the AG initially considered a ‘suitable connected network’ to be a network that included only RCTs of comparators listed in the final scope29 issued by NICE for the following risk groups, as defined in the IMDC criteria:15

• intermediate-/poor-risk subgroup (network nodes: lenvatinib plus pembrolizumab, cabozantinib and nivolumab plus ipilimumab)

• favourable-risk subgroup (network nodes: lenvatinib plus pembrolizumab, sunitinib, pazopanib and tivozanib)

• the all-risk population (network nodes: lenvatinib plus pembrolizumab, sunitinib, pazopanib and tivozanib).

However, where it was not possible to construct a connected network using only the comparators listed in the final scope29 issued by NICE, the AG considered introducing additional treatments (i.e. nodes), such as interferon-alpha and sorafenib to form connections. The AG considered that it was not appropriate to attempt to connect comparators listed in the final scope29 issued by NICE via two or more non-relevant treatments as more uncertainty is introduced with the addition of each irrelevant node.

Following assessment of suitable network structures and consideration of the availability of outcome data from each of the 10 RCTs,27,66,97–104 the AG excluded two trials27,99 (reasons are listed in Table 19) in at least one of the AG NMAs.

Details about the comparators and a list of the RCTs that provided information to inform the AG PFS, OS and ORR NMAs for the intermediate-/poor-risk and favourable-risk subgroups and all-risk population are presented in Table 20. The AG PFS, OS and ORR network diagrams are presented in Figures 4–7 and the outcome data used to populate the AG PFS, OS and ORR NMAs are presented in Tables 58–60 (see Appendix 4).

The AG considered that the different definitions of AEs reported within the trials (i.e. treatment-emergent, treatment-related or all-cause AEs for Grade ≥ 3 AEs and discontinuations due to AEs) made it difficult to interpret any relative differences between treatments. Furthermore, safety data were not reported separately for subgroups of interest, most notably for the intermediate-/poor-risk subgroup in the CheckMate 214 trial,100 and for the favourable-risk subgroup in any trials other than the CLEAR trial. AE data were unavailable for the previously untreated patients in the TIVO-1 trial. 102

Nonetheless, the AG performed NMAs for Grade ≥ 3 AEs where either treatment-emergent or all-cause AEs were reported (see Appendix 4, Figures 5 and 7 for network diagrams and Table 61 for outcome data used to populate these NMAs). The AG also considered performing NMAs for discontinuations due to AEs comparing (1) discontinuations of both lenvatinib and pembrolizumab and (2) discontinuations of either lenvatinib or pembrolizumab with relevant comparators. However, it appeared that only data for the latter were available from the CLEAR trial for risk subgroups. Further, when summing the total of AEs from the two subgroups, there were still many AEs in the all-risk population that appeared to be unaccounted for according to subgroup, that is summing the numbers of discontinuations due to AEs in the intermediate/poor and favourable-risk subgroups from Eisai CS,1 (Appendix F, see tables 64 and 65) did not sum to the total reported for the all-risk population in Table 16. Therefore, the AG considered it inappropriate to conduct NMAs for discontinuations due to AEs.

It was not possible for the AG to perform any HRQoL NMAs due to the heterogeneity of the HRQoL outcome scales used in the included trials and the sparsity of reported data (i.e. 95% CIs not reported and data not reported separately for subgroups of interest).

Assessment group methodological approach: intermediate-/poor-risk subgroup

The AG was able to construct a suitable network for PFS, OS and ORR including the two relevant comparators for this subgroup (cabozantinib and nivolumab plus ipilimumab); these networks also included sunitinib, a comparator common to the three included RCTs66,97,100 (see Figure 4 in Appendix 4). Safety data were not reported for the intermediate-/poor-risk subgroup in the CheckMate 214 trial,100 therefore the AG networks for Grade ≥3 AEs due to AEs for this subgroup included only cabozantinib (and sunitinib) as comparators (see Figure 5 in Appendix 4).

Assessment group methodological approach: International Metastatic Renal Cell Carcinoma Database Consortium/Memorial Sloan-Kettering Cancer Center favourable-risk subgroup

The AG PFS and OS networks included only sunitinib and pazopanib as comparators (see Figure 6). It was not possible to connect tivozanib to the PFS and OS networks as the only identified trial of tivozanib (TIVO-1 trial102) recruited a mixed population of untreated and previously treated patients with metastatic RCC and did not report PFS and OS data separately for the subgroup of untreated patients.

Only the CLEAR trial reported ORR and safety data for the favourable-risk subgroup; therefore, it was not possible to carry out NMAs of ORR or safety outcomes for lenvatinib plus pembrolizumab versus pazopanib or tivozanib.

Assessment group methodological approach: all-risk population

The AG PFS all-risk population network included all relevant comparators (sunitinib, pazopanib and tivozanib). This network was constructed by including sorafenib as a node and by using PFS data relating to first-line treatment from two trials (CROSS-J-RCC104 and SWITCH98) of sunitinib versus sorafenib that used a sequential design to connect tivozanib to the network (see Figure 7 in Appendix 4).

It was not possible to connect tivozanib to the OS network as OS data from patients receiving first-line treatment were not available from the CROSS-J-RCC104 and SWITCH98 trials and no trials were identified that allowed tivozanib to be included in the OS network via a single additional treatment node. The AG did not consider that it was appropriate to attempt to connect tivozanib to the OS network via two or more non-relevant treatments, which were not relevant comparators because of the increased level of uncertainty.

The AG was also unable to connect tivozanib to the ORR network as the only identified tivozanib trial (TIVO-1 trial102) recruited a mixed population of untreated and previously treated patients with metastatic RCC and did not report ORR data separately for the subgroup of untreated patients.

Therefore, for the all-risk population, the AG OS, ORR, Grade ≥3 AEs networks included only sunitinib and pazopanib as comparators (see Figure 6 in Appendix 4). The AG was not able to indirectly compare the clinical effectiveness of lenvatinib plus pembrolizumab with tivozanib for OS, ORR or Grade ≥ 3 AEs for patients in the all-risk population.

Quality assessment of the trials included in assessment group network meta-analysis

The quality assessment of the CLEAR trial and the seven other RCTs97,98,100–104 included in the AG NMAs is presented in Appendix 4 (see Table 62).

The AG considers that most of the trials included in the AG NMAs were of good methodological quality. However, due to insufficient information available, the AG was unable to assess the robustness of the randomisation procedures and whether robust procedures were in place to prevent patients or investigators predicting allocation to treatment in one trial. 103 All of the trials were open-label; however, the CLEAR trial and four other trials97,100–102 reported the use of blinded independent review of radiologic outcomes.

Assessment group summary of patient and trial characteristics and assessment of heterogeneity

Summaries of the design, demographic characteristics and the IMDC and MSKCC risk subgroups of patients enrolled in the CLEAR trial and other seven RCTs97,98,100–104 included in the AG NMAs are provided in Appendix 4 (see Tables 63 and 64).

In addition to the CLEAR trial, five of the trials were also phase III RCTs98,100,101,103,104 and two were phase II RCTs. 97,102 Three trials98,103,104 used a sequential design in which patients were randomised to first-line treatment, and patients who discontinued first-line treatment due to disease progression or toxicity received the alternative trial treatment as a second-line therapy; data from only these trials relating to first-line treatment were extracted. All of the RCTs were designed as open-label trials (see Table 62 in Appendix 4); the CLEAR trial and four other RCTs97,100–102 used blinded independent review for radiologic outcomes (i.e. PFS and ORR), two RCTs98,104 used unblinded investigator assessment, and the authors of one RCT103 did not report method of radiologic outcome assessment.

Two trials102,104 recruited patients with metastatic RCC only. The CLEAR trial and six other RCTs97,98,100,101,103,104 recruited untreated patients only, while one trial (TIVO-1102) recruited a mix of untreated patients (70.0%) and patients who had received previous systemic therapy (29.8%); data were extracted from the TIVO-1102 trial for the untreated subgroup only.

The CLEAR trial and five other RCTs97,100,102–104 recruited patients with clear cell RCC only, while 12.9% and 13.0% of recruited patients in the other two trials98,103 had non-clear cell histology. Results were not reported separately in the SWITCH trials98,103 for patients with clear cell histology.

The ages of recruited patients were similar across the RCTs (see Tables 63 and 64 in Appendix 4); across trial arms, the median age ranged from 61 years (in the CLEAR trial and two other trials100,101) to 68 years. 103 All trials recruited a majority of male patients (72.4%103 to 82.5%104).

In addition to the CLEAR trial, three RCTs100–102 recruited patients irrespective of disease risk according to IMDC or MSKCC criteria. However, data from the CheckMate 214 trial100 (nivolumab plus ipilimumab vs. sunitinib) were available for the intermediate-/poor-risk population and were used in the AG NMAs. The cabozantinib RCT97 recruited patients only with intermediate or poor-risk disease. Three RCTs98,103,104 were designed to recruit patients only with favourable- or intermediate-risk disease by MSKCC criteria.

Only the CLEAR trial reported disease risk classifications according to both IMDC and MSKCC risk criteria (see Table 64 in Appendix 4). Two other RCTs97,100 reported the proportion of patients classified by IMDC risk subgroup and four other RCTs98,101,103,104 reported the proportion of patients classified by MSKCC risk subgroup. The remaining RCT (TIVO-1102) did not report risk of disease according to IMDC or MSKCC criteria for the subgroup of untreated patients. The proportions of patients classified within each disease risk subgroup according to either IMDC or MSKCC criteria varied across RCTs (see Table 64 in Appendix 4).

The following differences between RCTs may have introduced heterogeneity into the AG NMAs:

• population characteristics (see Table 63 and Table 64 in Appendix 4)

• PFS and ORR assessment methods (BIRC, investigator or not reported) and types of AEs (all-cause AE or TEAE)

• differences in median PFS, OS, ORR and Grade ≥ 3 follow-up times (see Tables 58–61 in Appendix 4).

The AG is not aware of any statistical methods that can be used to adjust for these differences in patient baseline characteristics and trial design.

Assessment group assessment of proportional hazards assumptions

For time-to-event outcomes presented as HRs (i.e. PFS and OS), the AG assessed the validity of the within-trial PFS and OS PH assumptions for each of the groups (intermediate-/poor-risk and favourable-risk subgroups and all-risk population). The AG PH assessments were carried out by examining the figures (Schoenfeld residuals plots or log cumulative hazard plots) and statistical test results (e.g. Grambsch–Therneau test86) presented in the Eisai CS1 (sections 5.3.1 and 5.3.2) and in the Eisai response to clarification (questions A1 and A2).

Data from the CheckMate214 trial100 (nivolumab plus ipilimumab vs. sunitinib) were not included in the company NMAs. The AG, therefore, digitised the published intermediate-/poor-risk subgroup PFS and OS 42-month K-M data100 and assessed proportionality by plotting Schoenfeld residuals and performing a Grambsch–Therneau test. 86 The AG OS and PFS PH assessments are presented in Appendix 4 (Proportional hazards assessments for trials included in the assessment group network meta-analysis). Violations of the PH assumption within the studies included in the AG NMAs are listed in Table 21.

If the PH assumption holds, a HR represents an average of the relative treatment effect during the trial follow-up period106 (or trials, in the context of a NMA) and the HR is proportional over time. 107 When the PH assumption is violated, this means that the HR (whether from a trial or from a NMA including data from one or more trials with PH violations) is not applicable to all time points across the trial follow-up periods. If the PH assumption holds, then it may not be unreasonable to assume that the estimated HRs is valid beyond the trial follow-up periods. However, when the PH assumption is violated, estimated HRs may not produce accurate projections of relative survival between treatment arms beyond the observed trial follow-up periods.

Some PH test results showed (see Table 21) that PFS and OS outcome hazards were not proportional. Within any network, if any within-trial hazards are not proportional, then Bayesian HR NMA results [i.e. the HRs and 95% credible intervals (CrIs)] should not be used to infer statistically significant differences (or lack of statistically significant difference) between treatments.

Where violations of the PH assumption are demonstrated, alternative flexible modelling approaches for NMA, which relax the PH assumption, including FP NMAs, have been proposed to aid decision-making. 108,109 However, interpretation of the estimates provided by these complex modelling techniques can be difficult and often are not intuitive. 108,109

The ‘best-fitting’ FP model (or alternative flexible model) for a NMA, which is defined according to model fit statistics, such as the deviance information criterion (DIC), reflects the model that most closely captures the shape of the observed data. However model fit statistics do not provide information about whether a model is a good fit to the data or whether the estimates generated by the model, including projections of results beyond the follow-up times of trials included in the NMA, are clinically plausible. 109 Furthermore, flexible models that appear similar according to model fit (i.e. according to DIC statistics) may generate very different long-term survival estimates; advice from the Medical Research Council Biostatistics Unit110 is that, ‘if the difference in DIC is, say, less than 5 and the models make very different inferences, then it could be misleading just to report the model with the lowest DIC.’ Because of these limitations, the AG does not consider that it is appropriate to use the results of FP NMAs for clinical decision-making.

The AG considers that the limitations associated with the interpretation of results from FP NMAs are greater than the limitations of interpretation of the Bayesian HR NMA results when the PH assumption is violated. In addition, for the intermediate-/poor-risk subgroup (the largest of the two risk subgroups considered), there was no violation of the OS PH assumption within any of the trials included in the AG OS network.

The AG carried out PFS, OS and ORR NMAs for the intermediate-/poor-risk and the favourable-risk subgroups and all-risk population. However, the AG emphasises that where violations of the PH assumption were demonstrated, HRs and 95% CrIs should not be used to infer any statistically significant difference (or lack of statistically significant difference) for the treatment comparisons.

Assessment group statistical approach to Bayesian hazard ratio network meta-analysis

The AG performed PFS, OS and ORR NMAs using a Bayesian framework. These were carried out using the multinma R package. 111 This approach is in line with Decision Support Unit (DSU) guidance (documents 2, 3 and 4112–114). All results were generated using 100,000 iterations on 3 chains after a burn-in of 100,000 and vague prior distributions were used for intercept, treatment and heterogeneity [for random-effects (RE) models only] parameters.

The AG performed NMAs using fixed-effects (FE) and RE models. As convergence issues occurred due to sparse data, RE NMA results were unusable. Because of the lack of published information to select informative prior distributions to improve convergence of RE models, the AG has only presented results from FE models in the main body of this report. The AG has described where important clinical or statistical heterogeneity between RCTs included in the NMA may have had an impact on how NMA results can be interpreted.

For PFS, the only outcome with a closed loop present within the network, the AG assessed inconsistency in the NMAs by applying an unrelated mean effects model114 and by comparing model fit statistics of inconsistency models with consistency models.

Treatment effect estimates for direct and indirect clinical effectiveness evidence are presented as HRs for time-to-event data (i.e. PFS and OS) and ORs for dichotomous data (i.e. ORR). All treatment effect estimates are presented with 95% CrIs.

An example of the statistical code used by the AG to perform PFS, OS, ORR and safety NMAs is provided in Appendix 4 (see Example statistical code for assessment group network meta-analysis).

Progression-free survival: assessment group fixed-effects network meta-analysis

The AG PFS NMA results for all pairs of treatments for the intermediate-/poor-risk subgroup and the IMDC/MSKCC favourable-risk subgroup and all-risk population are presented in Table 22.

The AG NMAs included PFS data that were assessed using FDA censoring rules. The AG PFS NMA sensitivity analysis included CLEAR trial PFS data assessed using the EMA censoring rules and data from all other included trials using FDA censoring rules (see Table 69 in Appendix 4). Results from the two AG PFS NMAs were similar.

Because of PH violations or uncertainty regarding the validity of the PH assumption, the HRs and 95% CrIs shown in Table 22 cannot be used to infer any statistically significant difference (or lack of statistically significant difference) for any of the treatment comparisons (see Assessment group assessment of proportional hazards assumptions).

Overall survival: assessment group fixed-effects network meta-analysis

The OS FE NMA results for all pairs of treatments for the intermediate-/poor-risk subgroup and the IMDC/MSKCC favourable-risk subgroup and all-risk population are presented in Table 23.

In the intermediate-/poor-risk subgroup, a numerical advantage in terms of OS was shown for lenvatinib plus pembrolizumab versus cabozantinib (HR = 0.78, 95% CrI 0.47 to 1.28) and versus nivolumab plus ipilimumab (HR = 0.94, 95% CrI 0.66 to 1.32). However, neither of these numerical advantages was statistically significant. No violations of the PH assumption were observed for OS in this subgroup (see Assessment group assessment of proportional hazards assumptions).

Because of PH violations or uncertainty regarding the validity of the PH assumption, the AG OS NMA HRs and 95% CrIs for the IMDC/MSKCC favourable-risk subgroup and all-risk population (see Table 23) cannot be used to infer any statistically significant difference (or lack of statistically significant difference) for any of the treatment comparisons (see Assessment group assessment of proportional hazards assumptions).

Objective response rate: assessment group fixed-effects network meta-analysis

The AG ORR NMA results for all pairs of treatments for the intermediate-/poor-risk subgroup and all-risk population and are presented in Appendix 4 (see Table 67).

In the intermediate-/poor-risk subgroup, ORR was statistically significantly higher for lenvatinib plus pembrolizumab compared to nivolumab plus ipilimumab (OR = 3.19, 95% CrI 1.95 to 5.26); however, no statistically significant difference was shown between lenvatinib plus pembrolizumab and cabozantinib (OR = 2.46, 95% CrI 0.84 to 6.82). In the all-risk population, ORR was statistically significantly higher for lenvatinib plus pembrolizumab compared to sunitinib (OR = 4.35, 95% CrI 3.16 to 5.99) and compared to pazopanib (OR = 3.22, 95% CrI 2.14 to 4.85).

Grade ≥ 3 adverse events: assessment group fixed-effects network meta-analysis

The AG Grade ≥ 3 FE NMA results for all pairs of treatments for the intermediate-/poor-risk subgroup and the all-risk population are presented in Appendix 4 (see Table 70).

In the intermediate-/poor-risk subgroup, for the comparison of lenvatinib plus pembrolizumab with cabozantinib, there were no statistically significant differences in Grade ≥ 3 AEs (OR = 1.80, 95% CrI 0.79 to 4.10). In the all-risk population, there were statistically significantly more Grade ≥ 3 AEs for lenvatinib plus pembrolizumab compared to sunitinib (OR = 1.84, 95% CrI 1.28 to 2.66) and compared to pazopanib (OR = 1.86, 95% CrI 1.17 to 2.94).

Assessment group sensitivity analysis network meta-analysis: favourable-risk subgroup

The COMPARZ trial101 reported PFS and OS results (including a separately reported final OS analysis105) for the MSKCC favourable-risk subgroup (not for the IMDC favourable-risk subgroup). Therefore, the AG performed sensitivity analyses including MSKCC favourable-risk subgroup data from the CLEAR trial and the COMPARZ trial101 for the PFS (FDA and EMA censoring rules) and the OS NMAs (using COMPARZ trial final OS analysis105). Results of the MSKCC/MSKCC favourable-risk subgroup PFS and OS NMAs are presented in Appendix 4 (see Table 68). Numerical results (i.e. HRs and 95% CrIs) from the sensitivity analyses of PFS and OS (Appendix 4, Table 70) were similar to the results presented in Table 23 and Table 24, respectively.

Assessment of inconsistency for overall survival, progression-free survival and objective response rate network meta-analysi

The AG assessments of inconsistency for PFS in the all-risk population, the only NMA with a closed loop present within the network, are presented in Appendix 4 (see Assessment group assessment of inconsistency in the network meta-analysis). Although a model which accounts for inconsistency in the NMA provides a better statistical model fit compared to a model which assumes consistency, results of AG FE NMAs which assumed consistency or accounted for inconsistency were very similar. Therefore, any inconsistency present between direct and indirect evidence for PFS in the all-risk population does not seem to have had an important impact on AG PFS NMA results.

Because of the lack of closed loops in any of the other AG networks, the consistency of indirect estimates of OS, ORR and AEs are unknown.

Additional assessment group network meta-analysis sensitivity analysis

Additional sensitivity analyses were conducted for reasons described in Appendix 4 (see Additional assessment group network meta-analysis sensitivity analyses). In summary, updated results from the CheckMate 214 trial115,116 were incorporated into the intermediate-/poor-risk subgroup NMA. The AG found that, as with the original NMAs, PH is violated for PFS data, but not for OS data. Including the updated data from the CheckMate 214 trial had little impact on the results.

Assessment group search strategy

The AG searched the electronic sources listed in Appendix 2, Table 51. Full search strategies are presented in Appendix 2. As lenvatinib was first approved for the treatment of aRCC by the FDA in 2016, the AG considered that searching databases from 2006 onwards would allow all relevant economic evidence to be identified. In addition, the reference lists of all included publications were assessed for relevance. The results of the searches were entered into an Endnote (X9 software package63) library, de-duplicated, and then exported into Covidence Systematic Review software. 64

Assessment group study selection and inclusion criteria

Records were selected for inclusion in the review on the basis of the criteria shown in Table 24. The criteria were developed to ensure that the included studies would provide information to help address the AG decision problem which aligns to the final scope29 issued by NICE, that is to assess the cost-effectiveness of treatment with lenvatinib plus pembrolizumab for patients with untreated aRCC versus treatment with sunitinib, pazopanib, tivozanib, cabozantinib and nivolumab plus ipilimumab.

Two reviewers (RH/DB) independently screened the titles and abstracts of all records identified by the searches. Full-text versions of all studies considered potentially relevant were obtained. The same two reviewers then independently assessed the relevance of these full-text publications and reasons for exclusion were assigned based on the hierarchical order as shown in Table 24. Disagreements about inclusion were resolved through discussion and, in all cases, a consensus was reached.

Quantity of cost-effectiveness evidence

The AG searches identified 3127 records. Of these, 2742 records were obtained from the database searches and 385 records were identified from other sources, that is from conference proceedings (n = 129) and website searches (n = 256). After duplicates were removed, 1899 records remained. Following screening of titles and abstracts, 47 full-text publications were retrieved (one potentially relevant report could not be retrieved) and checked for eligibility using prespecified inclusion criteria. The AG study selection process is shown in the section Sources searched of Appendix 2, Figure 3.

Assessment group data extraction

A data extraction form was designed in MS Excel. Extracted data included bibliographic information (e.g. authors and title) and details of the type of analyses conducted. Details about the economic model were also extracted (e.g. parameters used and their sources, results of the analyses, authors’ conclusions and limitations reported by the authors). Information from the included study was extracted independently by two reviewers (RH/DB).

Quality of cost-effectiveness evidence

The AG assessed the quality of the included cost-effectiveness study (i.e. Li et al. 2021117) using the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) checklist118 (see Table 75 in Appendix 5). Two reviewers (RH/DB) independently carried out the quality assessment. The reviewers agreed that, except for resource use items, the included study117 had transparently reported the methods used to conduct their CEA.

Key information from the included cost-effectiveness study

The data extracted by the AG from the included cost-effectiveness study117 are provided in Table 74 in Appendix 5.

The cost-effectiveness results generated by Li et al. 2021117 showed that lenvatinib plus pembrolizumab generated more life-years (LYs) and more quality-adjusted life-years (QALYs) in comparison to sunitinib. However, incremental costs were high and the base-case incremental cost-effectiveness ratio (ICER) for this comparison was more than US$100,000 per QALY – a level that the authors report is an acceptable willingness to pay (WTP) threshold.

Assessment group systematic review conclusions

The Li et al. 2021117 cost-effectiveness study included estimates of the comparative cost-effectiveness of lenvatinib plus pembrolizumab versus sunitinib. However, the study was undertaken from the perspective of the US healthcare system and, therefore, the extent to which resource use and results are generalisable to the NHS is unclear. Further, the study was limited to the all-risk population and included comparators that are not recommended by NICE for patients with untreated aRCC.

Assessment group summary of companies’ economic models

Key information about the models submitted by the companies is presented in Table 25.

Critical appraisal of the companies’ economic analyses

The AG critical appraisal of the companies’ economic analyses was carried out using the Drummond checklist (see Appendix 6, Table 77) and the NICE Reference Case checklist (see Appendix 6, Table 78).

Prognostic risk subgroups

The IMDC prognostic risk subgroup data are available from the CLEAR trial:

• intermediate/poor risk (n = 472, 66.3%)

• intermediate risk (n = 402, 56.5%)

• poor risk (n = 70, 9.8%)

• favourable risk (n = 234, 32.9%).

Previous NICE TAs24,25,32–34,36 have produced treatment recommendations for patients with untreated aRCC for the combined intermediate/poor-risk subgroup (TA542,24 TA58125 and TA64535 for use within the CDF; TA78036 which superseded TA58125 for use in routine practice) and all-risk population (TA169,32 TA21533 and TA51234). As some treatments are only available for the intermediate/poor-risk subgroup, the AG considers that cost-effectiveness results for the all-risk population (CLEAR trial FAS/ITT population) are not relevant to this appraisal. The AG has therefore conducted separate CEAs for the intermediate-/poor-risk and favourable subgroups by using relevant comparator data for each subgroup (i.e. intermediate/poor risk: cabozantinib or nivolumab plus ipilimumab; favourable risk: sunitinib, pazopanib or tivozanib). For completeness, cost-effectiveness results for the all-risk population are provided in Appendix 7. As cabozantinib and nivolumab plus ipilimumab are only recommended by NICE for treating patients with intermediate-/poor-risk disease, the AG does not consider that cost-effectiveness results for the poor-risk subgroup only are relevant and so has not generated any cost-effectiveness results for this subgroup.

Discounting, time horizon and perspective

In line with the NICE Reference Case,119 in the Eisai and MSD (and MSD/AG) models, costs and benefits were discounted at a rate of 3.5%. In the MSD model, discounting was incorrectly applied from the first cycle; in the MSD/AG model, this error was corrected and discounting now starts at the beginning of the second year. Scenario analyses were performed by the AG using annual discount rates of 0% and 6% for costs and benefits.

The time horizon used in the Eisai, MSD and MSD/AG models is 40 years. The AG considers that this is sufficient to capture all relevant costs and benefits. The perspective of all three models is the NHS and personal and social services (PSS).

Populating the model with clinical effectiveness data: general methods

Direct clinical effectiveness evidence (PFS, OS and TTD) is only available from the CLEAR trial for the comparison of the efficacy of lenvatinib plus pembrolizumab with sunitinib.

In line with DSU guidance,120 Eisai, MSD and the AG assessed the goodness-of-fit to PFS, OS and TTD K-M data of standard distributions (exponential, gamma, generalised gamma, Gompertz, log-logistic, log-normal, Weibull) using the Akaike information criterion (AIC) and the Bayesian information criterion (BIC) statistics. The distribution producing the lowest AIC and BIC statistics is considered the best fitting (i.e. highest ranking); however, Eisai suggests that other distributions may be as good as the highest-ranking distribution. The AG highlights that, for PFS and OS, Eisai only provided AIC and BIC statistics for the all-risk population.

As well as the visual fit of the seven distributions to the K-M data, the AG assessed the:

• clinical plausibility of long-term projections (i.e. whether the mortality rate rapidly fell below background mortality)

• whether the distribution used to model PFS led to higher mortality than the distribution chosen to model OS

• whether survival projections for the intermediate-/poor-risk subgroup were more/less optimistic than those for the favourable-risk subgroup.

Populating the Merck Sharp & Dohme/assessment group model: progression-free survival

Eisai and MSD fitted distributions to CLEAR trial BICR assessed PFS data (FDA censoring rules). The PFS distributions chosen by Eisai, MSD and the AG are shown in Table 32. The PFS distributions chosen by the AG cannot be shown graphically as these data are confidential.

Intermediate-/poor-risk subgroup (progression-free survival)

Favourable-risk subgroup (progression-free survival)

Eisai did not generate any cost-effectiveness estimates for the favourable-risk subgroup.

Assessment group scenario analyses: intermediate-/poor-risk and favourable-risk subgroups (progression-free survival)

Intermediate-/poor-risk subgroup (overall survival)

Favourable-risk subgroup (overall survival)

For patients in the favourable-risk subgroup, there was considerable uncertainty around the validity of the CLEAR trial OS estimates due to the low number of events experienced by these patients.

Assessment group scenario analyses: intermediate-/poor-risk and favourable-risk subgroups (overall survival)

Intermediate-/poor-risk subgroups (time to treatment discontinuation)

The AG considered that TTD for patients receiving lenvatinib should be modelled by fitting a distribution to CLEAR trial TTD K-M data, and for patients receiving pembrolizumab, the CLEAR trial TTD K-M data should be used directly.

Favourable-risk subgroup

Of the two companies, only MSD provided cost-effectiveness results for the favourable-risk subgroup.

Assessment group scenario analyses: intermediate-/poor-risk and favourable-risk subgroups (time to treatment discontinuation)

Assessment group scenario analyses (utility values)

The AG carried out two scenario analyses. One scenario analysis used the Eisai treatment dependent health state utility values and the other used the MSD treatment independent health state utility values. The utility values are confidential and therefore cannot be reported.

Lenvatinib

Eisai and MSD estimated drug acquisition costs for lenvatinib and pembrolizumab on the basis of dosing schedules for each drug as described in the CLEAR trial protocol. Eisai calculated the cost of lenvatinib using a weighted average cost per mg on the basis of average dose received by CLEAR trial patients, and MSD used weekly CLEAR trial dosing data. These data were provided for the all-risk population and not separately by risk subgroups. Clinical advice to the AG was that dosing was unlikely to vary by risk subgroup.

Lenvatinib tablets are available in two strengths (4 and 10 mg); the cost of a 30-tablet pack is the same irrespective of dose. Clinical advice to the AG was that, in NHS clinical practice, a patient’s dose of lenvatinib varies in line with the CLEAR trial protocol descriptions, that is a patient will start on a dose of 20 mg per day and then their dose will be reduced to 14 mg, then to 10 mg and finally to 8 mg, with reductions ceasing once a level that the patient can tolerate has been reached. Further, clinical advice to the AG was as follows:

• A dose of 8 mg per day was quite rare as patients unable to tolerate a 10 mg per day dose were unlikely to be able to tolerate an 8 mg per day dose.

• In the short term, 14 mg per day was the dose that most patients were titrated to from 20 mg.

• In the longer term, approximately 25% of patients were prescribed a 10 mg per day dose.

As the cost per pack of lenvatinib is the same for a 20 mg per day dose and a 14 mg per day dose, the proportion of people prescribed a 10 mg dose (i.e. one capsule) is important.

The AG has used the weekly lenvatinib CLEAR trial dosing data (available from the MSD model). The AG highlights that after 120 weeks, patient CLEAR trial data are limited and, therefore, are unreliable. The AG estimated the cost of lenvatinib using CLEAR trial data (tablets per week) over the first 120 weeks and, for the remainder of the model timeframe, used the average weekly number of lenvatinib tablets patients received between weeks 94 and 120 (i.e. 6 months prior to the end of the reliable data). This approach meant that use of a relative dose intensity (RDI) multiplier was not relevant.

Pembrolizumab

In the CLEAR trial, treatment with pembrolizumab was available for a maximum of 2 years. On the basis of CLEAR trial data, Eisai and MSD used a RDI multiplier (based on all-risk population data) to account for ‘delays in drug administration’. Eisai and MSD used the same methods to estimate RDI values and therefore it is unclear why the values presented by Eisai and MSD differ. Eisai did not provide the values used in their calculation; however, MSD did provide this detail and the AG was able to verify the MSD RDI value. Therefore, the AG used the MSD value in the MSD/AG model.

Sunitinib

Eisai, MSD and the AG estimated the cost of sunitinib using the CLEAR trial dosing schedule. Eisai and MSD used a RDI multiplier (estimated using CLEAR trial data) to adjust the cost of sunitinib. Eisai used a mean value (confidential information has been removed) and MSD used the published median value of 83.2%. 66 The AG has used the Eisai mean value.

Pazopanib, tivozanib, cabozantinib and nivolumab plus ipilimumab

Eisai and MSD estimated the costs of treatment with pazopanib, tivozanib and cabozantinib using dosing schedules published in the relevant SmPCs (Table 37). Eisai and MSD used RDI multipliers published in previous NICE TAs to adjust the costs of pazopanib (86%), tivozanib (94%) and cabozantinib (94%). The AG considered that the approach followed by the companies were appropriate and used the same dosing schedules and RDI values in the MSD/AG model.

The AG used the published dosing schedule for nivolumab plus ipilimumab52 (Table 37). No RDI multiplier information was available for nivolumab plus ipilimumab and therefore the AG used the MSD pembrolizumab RDI multiplier (confidential information has been removed), based on CLEAR trial data, to adjust the cost of nivolumab plus ipilimumab.

For all first-line treatments (intervention and comparators), costs per cycle were calculated using published British National Formulary prices (online database) (Table 38).

Drug administration costs

Drug administration costs are presented in Table 39. Eisai and MSD estimated chemotherapy administration costs using the National Schedule of NHS Costs 2019–20 (SB12Z Simple parenteral chemotherapy at first attendance). 121 However, the costs associated with this code differ as Eisai has assumed that administration is an outpatient appointment (£221.35) and MSD has assumed that administration is a day case appointment (£299.61). Clinical advice to the AG is that chemotherapy infusions are delivered as part of an outpatient appointment and, therefore, the AG has used the same administration cost as Eisai (£221.35) for first attendance and SB15Z Deliver Subsequent Elements of a Chemotherapy Cycle for all other attendances (£253.77).

Eisai and MSD assumed that the cost of administering oral drugs was zero. The AG considered that this was a conservative assumption and therefore included the cost of the delivery of oral chemotherapy for the first cycle and the cost of a hospital-based pharmacist dispensing the drugs for the subsequent cycles. These assumptions are the same as the assumptions used in TA64535 (Table 39).

As nivolumab and ipilimumab are both intravenous drugs, the AG assumed that for the period patients received both drugs (first four cycles), the most appropriate administration cost was Deliver Complex Chemotherapy at First Attendance (SB14Z) – outpatient. For the subsequent cycles, when patients received only nivolumab, the administration cost used was Deliver Simple Parenteral Chemotherapy at First Attendance (SB12Z) – outpatient.

Assessment group scenario analysis (adverse events)

The AG carried out two scenario analyses: one in which AE costs were set to zero and one in which AE costs were doubled.

Assessment group sensitivity analyses (subsequent treatment costs)

The AG carried out sensitivity analyses that varied the costs of subsequent treatments by ± 20%.

Intermediate-/poor-risk subgroup

For the intermediate-/poor-risk subgroup, the AG base-case cost-effectiveness results suggest that treatment with lenvatinib plus pembrolizumab generates more QALYs than treatment with cabozantinib or nivolumab plus ipilimumab but at a greater overall cost (list prices for all drugs). For the comparison of lenvatinib plus pembrolizumab with cabozantinib, the ICER per QALY gained is £133,362, and for the comparison of lenvatinib plus pembrolizumab with nivolumab plus ipilimumab, the ICER per QALY gained is £166,249. Detailed results are presented in Tables 40 and 41.

Favourable-risk subgroup

For the favourable-risk subgroup, the AG OS NMA results and the CLEAR trial data suggest that treatment with sunitinib generates improved OS compared to treatment with lenvatinib plus pembrolizumab. The AG base-case cost-effectiveness results suggest that treatment with sunitinib generates more QALYs than treatment with lenvatinib plus pembrolizumab at a lower overall cost (list prices for all drugs), that is treatment with lenvatinib plus pembrolizumab is dominated by treatment with sunitinib. Detailed results are presented in Tables 42 and 43.

Intermediate-/poor-risk subgroup

The mean probabilistic ICERs per QALY gained for the comparison of lenvatinib plus pembrolizumab with cabozantinib (£169,019) and with nivolumab plus ipilimumab (£134,253) are slightly higher than the deterministic cost-effectiveness results. In all iterations, lenvatinib plus pembrolizumab was the most expensive treatment option and generated the most QALYs. At a WTP threshold of £50,000 per QALY gained, in 100% of iterations cabozantinib was the most cost-effective treatment option. At a WTP threshold of £100,000 per QALY gained, in 0.8% of iterations lenvatinib plus pembrolizumab was the most cost-effective treatment option.

Favourable-risk subgroup

The mean probabilistic results were almost identical to the deterministic cost-effectiveness results. Lenvatinib plus pembrolizumab was dominated by sunitinib, pazopanib and tivozanib, and sunitinib was the most cost-effective treatment option. In all iterations, lenvatinib plus pembrolizumab was the most expensive treatment option and generated the fewest QALYs. As the QALYs generated for sunitinib, pazopanib and tivozanib are always the same in each iteration, the cost effectiveness acceptability curve shows horizontal lines for these, that is the probability of any of these three treatments being cost-effective does not vary with the WTP for a QALY threshold. For the majority (85.9%) of iterations, sunitinib was the cheapest option and therefore also the most cost-effective option. In 14.1% of iterations, pazopanib was the cheapest option and therefore the most cost-effective option. Lenvatinib plus pembrolizumab or tivozanib were not the most cost-effective options at any WTP threshold.

Assessment group one-way deterministic sensitivity analysis results

Assessment group deterministic scenario analysis results (intermediate-/poor-risk subgroup)

Assessment group deterministic scenario analysis results (favourable-risk subgroup)

The AG has presented deterministic scenario results for the comparison of lenvatinib plus pembrolizumab with sunitinib (see Table 46), with pazopanib (see Table 47) and with tivozanib (see Table 48) for the favourable-risk subgroup. Lenvatinib plus pembrolizumab was dominated by sunitinib, pazopanib and tivozanib across all scenarios considered.

Additional assessment group sensitivity analyses

In response to errors identified during the NICE appraisal consultation comments, and to incorporate the results from the updated intermediate-/poor-risk subgroup NMA, the AG produced additional sensitivity analyses to address two modelling errors. The details of the errors and the results are presented in Appendix 10. In summary, using the updated costs and results from the updated intermediate-/poor-risk subgroup NMA in the model had little impact on results (Appendix 10, Tables 91–98) and the same conclusions could be drawn.

Direct clinical effectiveness results

The AG systematic review of clinical effectiveness evidence identified only one RCT of lenvatinib plus pembrolizumab versus sunitinib for patients with untreated aRCC, the CLEAR trial. Results from this trial demonstrated improved PFS and ORR for lenvatinib plus pembrolizumab in the intermediate/poor and favourable-risk subgroups and all-risk population. CLEAR trial results from the updated OS analysis showed a statistically significant improvement for patients treated with lenvatinib plus pembrolizumab versus patients treated with sunitinib for the intermediate-/poor-risk subgroup and the all-risk population; there were too few events in the favourable-risk subgroup for robust OS conclusions to be drawn. Generally, the AEs experienced by patients treated with lenvatinib plus pembrolizumab were consistent with the known safety profile of the two drugs. When compared to treatment with sunitinib, treatment with lenvatinib plus pembrolizumab appears to neither improve nor worsen HRQoL.

Indirect clinical effectiveness results

The AG carried out Bayesian HR NMAs for the three patient disease risk groups. However, due to limited data availability, it was not possible to carry out NMAs for all outcomes for all three patient risk groups. Further, as networks were sparse, it was possible to generate meaningful results only using FE NMAs.

The AG PFS NMA results for the intermediate-/poor-risk subgroup, the favourable-risk subgroup and the all-risk population should not be used to infer any statistically significant difference (or lack of statistically significant difference) for any of the treatment comparisons owing to within-trial PH violations or uncertainty regarding the validity of the PH assumption.

The AG OS NMA results for the intermediate-/poor-risk subgroup suggested that there was a numerical, but not a statistically significant, improvement in OS for patients treated with lenvatinib plus pembrolizumab compared with patients treated with cabozantinib or nivolumab plus ipilimumab. Because of within-trial PH violations or uncertainty regarding the validity of the PH assumption, the AG OS NMA results for the favourable-risk subgroup and the all-risk population should not be used to infer any statistically significant difference (or lack of statistically significant difference) for any of the treatment comparisons.

The AG ORR NMA showed a statistically significantly improved ORR for lenvatinib plus pembrolizumab versus nivolumab plus ipilimumab and a non-statistically significant numerical advantage for lenvatinib plus pembrolizumab versus cabozantinib in the intermediate-/poor-risk subgroup. Lenvatinib plus pembrolizumab also resulted in statistically significant improvements versus sunitinib and pazopanib in the all-risk population. Evidence was unavailable for lenvatinib plus pembrolizumab versus tivozanib in the all-risk population or versus any relevant comparator in the favourable-risk population.

Results from the AG AE NMAs in the intermediate-/poor-risk subgroup showed non-statistically significant evidence that lenvatinib plus pembrolizumab resulted in an increase in Grade ≥ 3 AEs versus cabozantinib. In the all-risk population, there were statistically significantly more Grade ≥ 3 AEs for patients treated with lenvatinib plus pembrolizumab versus sunitinib and versus pazopanib.

It was not possible for the AG to perform any HRQoL NMAs due to the heterogeneity of the HRQoL outcome scales used in the included trials and limited reported data (i.e. 95% CIs not reported, data not reported separately for risk subgroups).

Cost-effectiveness results

For the intermediate-/poor-risk subgroup, AG base-case cost-effectiveness results (list prices) suggested that treatment with lenvatinib plus pembrolizumab generated more QALYs than cabozantinib and more QALYs than nivolumab plus ipilimumab, but at a greater overall cost than either of these two treatments. Using list prices, the ICERs per QALY gained for the comparison of lenvatinib plus pembrolizumab versus cabozantinib and versus nivolumab plus ipilimumab exceeded £100,000.

For the favourable-risk subgroup, AG base-case cost-effectiveness results (list prices) suggested that treatment with sunitinib generated more QALYs than lenvatinib plus pembrolizumab at a lower overall cost, that is treatment with lenvatinib plus pembrolizumab was dominated by treatment with sunitinib (and, using the assumption of equivalent effectiveness, by pazopanib and tivozanib).

The AG base-case cost-effectiveness results for the intermediate-/poor-risk and favourable-risk subgroups were robust over most of the assumptions used in the AG PSA, sensitivity and scenario analyses.

Patient and public involvement

There was no PPI regarding the production of the protocol or report for this systematic review and CEA. However, as the analyses were conducted to inform a NICE appraisal, NICE received input from experts and stakeholders in addition to the evidence presented by the AG and companies. All stakeholders and the public were able to comment on the preliminary guidance issued by NICE. For this appraisal, NICE received a written submission from the following patient organisations: Action Kidney Cancer and Kidney Cancer Support Network. In addition, patient experts attended the NICE AC meeting and offered valuable insight into living with the disease.

Equality, diversity and inclusion

Strengths

Weaknesses

Uncertainties

Favourable-risk population

NICE24,36 has recommended aRCC treatments for the all-risk population and for the intermediate-/poor-risk subgroup. If a patient does not have intermediate-/poor-risk disease, then, by definition, the patient has favourable-risk disease. The AG has, therefore, carried out clinical and cost-effectiveness analyses for the favourable-risk subgroup. Efficacy results from a recent population-based study21 showed that median OS for the all-risk population was approximately half the length of that for the favourable-risk subgroup [all-risk population: 28.6 (95% CI 25.9 to 31.0) months; favourable-risk subgroup: 52.1 (95% CI 43.4 to 61.2) months]. These results suggest that it is informative to consider the favourable-risk subgroup separately alongside results for the intermediate-/poor-risk subgroup.

While there were few events, favourable-risk subgroup CLEAR trial results show no statistically significant OS benefit for lenvatinib plus pembrolizumab versus sunitinib; these results are consistent with previously published reviews53,55,59 of combination therapies, including lenvatinib plus pembrolizumab.

It was beyond the scope of this appraisal to compare lenvatinib plus pembrolizumab with avelumab plus axitinib. Clinical advice to the AG is that treatment with avelumab plus axitinib is the preferred option for patients with favourable-risk aRCC.

Issues identified during the National Institute for Health and Care Excellence appraisal

After the NICE AC Meeting, the AG conducted additional clinical and cost-effectiveness sensitivity analyses. The additional clinical effectiveness sensitivity analyses were PFS and OS NMAs for the intermediate-/poor-risk subgroup using updated information from the CheckMate 214 trial. 100 The additional cost-effectiveness analyses were to correct for two modelling errors identified in the tivozanib engine for AE costs and application of oral administration costs. Using the revised costs and updated NMA data had relatively little impact on the clinical and cost-effectiveness results and the same conclusions could be drawn as from the original analyses.

The AG considers that it is important to reiterate that the cost-effectiveness analyses presented in this appraisal are based on list prices only. As patient access scheme (PAS) discount prices are in place for lenvatinib, pembrolizumab, sunitinib, pazopanib, tivozanib, cabozantinib, nivolumab, ipilimumab, everolimus and axitinib, the cost-effectiveness comparisons presented using list prices in this report cannot be used as the basis for NHS decision-making. The AG provided cost-effectiveness results generated using the discounted prices for lenvatinib and pembrolizumab in a confidential appendix presented to NICE. The NICE AC concluded that, when using PAS prices for all drugs, in the intermediate-/poor-risk subgroup:

• the cost-effectiveness estimates were above the range that NICE considers an acceptable use of NHS resources when lenvatinib plus pembrolizumab was compared with cabozantinib

• the cost-effectiveness estimates were within the range that NICE considers acceptable when lenvatinib plus pembrolizumab was compared with nivolumab plus ipilimumab.

The NICE AC concluded that, when using PAS prices for all drugs, in the favourable-risk subgroup, all the cost-effectiveness estimates were above the range that NICE considers an acceptable use of NHS resources. 3

MEDLINE (via Ovid)

Ovid MEDLINE(R) ALL <1946–7 October 2021>

The Cochrane Library (CENTRAL)

www.cochranelibrary.com/

Cochrane Central Register of Controlled Trials

Issue 10 of 12 October 2021

Searches terms with and without hyphen, that is same results for clear-cell as for ‘clear cell’.

EMBASE (via Ovid)

EMBASE <1974–7 October 2021>

PubMed

https://pubmed.ncbi.nlm.nih.gov/

((((‘Carcinoma, Renal Cell’[Mesh]) OR (‘Kidney Neoplasms’[Mesh]) OR (‘renal cancer*’[Text Word] OR ‘renal carcinoma*’[Text Word] OR ‘renal adenocarcinoma*’[Text Word] OR ‘renal tumor*’[Text Word] OR ‘renal tumour*’[Text Word] OR ‘renal malignanc*’[Text Word]) OR (‘kidney cancer*’[Text Word] OR ‘kidney carcinoma*’[Text Word] OR ‘kidney adenocarcinoma*’[Text Word] OR ‘kidney tumor*’[Text Word] OR ‘kidney tumour*’[Text Word] OR ‘kidney malignanc*’[Text Word]) OR (‘clear-cell cancer*’[Text Word] OR ‘clear-cell carcinoma*’[Text Word] OR ‘clear-cell adenocarcinoma*’[Text Word] OR ‘clear-cell tumor*’[Text Word] OR ‘clear-cell tumour*’[Text Word] OR ‘clear-cell malignanc*’[Text Word]) OR (‘non-clear cell cancer*’[Text Word] OR ‘non-clear cell carcinoma*’[Text Word] OR ‘non-clear cell adenocarcinoma*’[Text Word] OR ‘non-clear cell tumor*’[Text Word] OR ‘non-clear cell tumour*’[Text Word] OR ‘non-clear cell malignanc*’[Text Word]) OR (hypernephroma[Text Word]) OR (hypernephroid carcinoma*[Text Word]) OR (grawitz tumor*[Text Word] OR grawitz tumour*[Text Word]) OR (rcc[Text Word])) AND ((advanced[Text Word] OR metastatic[Text Word] OR mRCC[Text Word] OR m-RCC[Text Word] OR aRCC[Text Word] OR a-RCC[Text Word] OR ‘first-line’[Text Word] OR ‘first line’[Text Word] OR metastasize[Text Word] OR metastasis[Text Word] OR metastases[Text Word] OR ‘stage iii’[Text Word] OR ‘stage 3’[Text Word] OR ‘stage 4’[Text Word] OR ‘stage iv’[Text Word] OR recurrent[Text Word] OR ‘non resectable’[Text Word] OR inoperable[Text Word] OR ‘non operable’[Text Word] OR unresectable[Text Word]) OR (‘Neoplasm Metastasis’[Mesh]))) OR (mrcc[Text Word] OR arcc[Text Word])) AND ((((randomized controlled trial [pt] OR ‘controlled clinical trial’[Publication Type] OR ‘randomized’[Title/Abstract] OR ‘randomised’ [Title/Abstract] OR ‘placebo’[Title/Abstract]) OR (‘clinical trials as topic’ [mesh: noexp]) OR (randomly [tiab] OR trial [ti] OR RCT [ti])) NOT (animals [mh] NOT humans [mh]))) Filters: English

Note: Cannot search in abstract only field in PubMed [RCT filter].

ClinicalTrials.gov

https://clinicaltrials.gov/

((advanced OR metastatic OR secondary OR EXPAND[Concept] ‘first-line’ OR EXPAND[Concept] ‘first line’ OR metastasis or mRCC or m-RCC OR aRCC OR a-RCC OR metastasize OR metastasis OR metastases OR EXPAND[Concept] ‘stage iii’ OR EXPAND[Concept] ‘stage 3’ OR EXPAND[Concept] ‘stage 4’ OR EXPAND[Concept] ‘stage iv’ OR recurrent OR EXPAND[Concept] ‘non resectable’ OR EXPAND[Concept] ‘non-resectable’ OR inoperable OR EXPAND[Concept] ‘non operable’ OR EXPAND[Concept] ‘non-operable’ OR unresectable) AND AREA[ConditionSearch] (EXPAND[Concept] ‘Renal cell’ OR EXPAND[Concept] ‘renal clear cell’ OR EXPAND[Concept] ‘renal clear-cell’ OR EXPAND[Concept] ‘renal non-clear cell’ OR EXPAND[Concept] ‘renal non clear cell’ OR RCC OR EXPAND[Concept] ‘renal carcinoma’ OR EXPAND[Concept] ‘renal cancer’ OR EXPAND[Concept] ‘renal tumor’ OR EXPAND[Concept] ‘renal tumour’ OR EXPAND[Concept] ‘renal adenocarcinoma’ OR EXPAND[Concept] ‘renal malignancy’ OR EXPAND[Concept] ‘kidney cancer’ OR EXPAND[Concept] ‘kidney carcinoma’ OR EXPAND[Concept] ‘kidney adenocarcinoma’ OR EXPAND[Concept] ‘kidney tumor’ OR EXPAND[Concept] ‘kidney tumour’ OR EXPAND[Concept] ‘kidney malignancy’ OR EXPAND[Concept] ‘clear-cell cancer’ OR EXPAND[Concept] ‘clear cell cancer’ OR EXPAND[Concept] ‘clear-cell carcinoma’ OR EXPAND[Concept] ‘clear cell carcinoma’ OR EXPAND[Concept] ‘clear-cell adenocarcinoma’ OR EXPAND[Concept] ‘clear cell adenocarcinoma’ OR EXPAND[Concept] ‘clear-cell tumor’ OR EXPAND[Concept] ‘clear cell tumor’ OR EXPAND[Concept] ‘clear-cell tumour’ OR EXPAND[Concept] ‘clear cell tumour’ OR EXPAND[Concept] ‘clear-cell malignancy’ OR EXPAND[Concept] ‘clear cell malignancy’ OR EXPAND[Concept] ‘non-clear cell cancer’ OR EXPAND[Concept] ‘non clear cell cancer’ OR EXPAND[Concept] ‘non-clear cell carcinoma’ OR EXPAND[Concept] ‘non clear cell carcinoma’ OR EXPAND[Concept] ‘non-clear cell adenocarcinoma’ OR EXPAND[Concept] ‘non clear cell adenocarcinoma’ OR EXPAND[Concept] ‘non-clear cell tumor’ OR EXPAND[Concept] ‘non clear cell tumor’ OR EXPAND[Concept] ‘non-clear cell tumour’ OR EXPAND[Concept] ‘non clear cell tumour’ OR EXPAND[Concept] ‘non-clear cell malignancy’ OR EXPAND[Concept] ‘non clear cell malignancy’ OR hypernephroma OR EXPAND[Concept] ‘hypernephroid carcinoma’ OR grawitz)) OR (aRCC OR mRCC or a-RCC OR m-RCC)

International Clinical Trials Registry Platform

https://trialsearch.who.int/

Search 1:

TITLE: advanced OR metastatic OR metastasis OR metastasize OR secondary OR ‘first line’ OR ‘first-line’ recurrent OR non-resectable OR ‘non resectable’ OR ‘stage 3’ OR ‘stage 4’ OR ‘stage iii’ OR ‘stage iv’ OR mRCC OR aRCC OR inoperable OR ‘non operable’ OR unresectable

AND

CONDITION: ‘renal cell’ OR ‘clear-cell’ OR ‘non-clear cell’ OR RCC OR ‘kidney cancer*’ OR ‘renal cancer*’ OR ‘renal carcinoma*’ OR ‘renal adenocarcinoma’ OR ‘renal tumor*’ OR ‘renal tumour*’ OR hypernephroma OR ‘hypernephroid carcinoma’ OR grawitz

Search 2:

aRCC OR mRCC or a-RCC OR m-RCC

Note: Parentheses (brackets) cannot be used to determine the order in which terms are combined.

Searches automatically include synonyms generated using the UMLS metathesaurus.

Searches are restricted to 256 character spaces, truncated search strategies used.

With/without hyphen retrieves same numbers.

International Health Technology Assessment Database

https://database.inahta.org/

((‘Neoplasm Metastasis’[mhe]) OR (advanced OR metastatic OR mRCC OR m-RCC OR aRCC OR a-RCC OR ‘first-line’ OR ‘first line’ OR metastasize OR metastasis OR metastases OR ‘stage iii’ OR ‘stage 3’ OR ‘stage 4’ OR ‘stage iv’ OR recurrent OR ‘non resectable’ OR inoperable OR ‘non operable’ OR unresectable)) AND ((‘renal cancer*’ OR ‘renal carcinoma*’ OR ‘renal adenocarcinoma*’ OR ‘renal tumor*’ OR ‘renal tumour*’ OR ‘renal malignanc*’ OR ‘kidney cancer*’ OR ‘kidney carcinoma*’ OR ‘kidney adenocarcinoma*’ OR ‘kidney tumor*’ OR ‘kidney tumour*’ OR ‘kidney malignanc*’ OR ‘clear cell cancer*’ OR ‘clear cell carcinoma*’ OR ‘clear cell adenocarcinoma*’ OR ‘clear cell tumor*’ OR ‘clear cell tumour*’ OR ‘clear cell malignanc*’ OR ‘non clear cell cancer*’ OR ‘non clear cell carcinoma*’ OR ‘non clear cell adenocarcinoma*’ OR ‘non clear cell tumor*’ OR ‘non clear cell tumour*’ OR ‘hypernephroma’ OR ‘hypernephroid carcinoma*’ OR ‘grawitz tumor*’ OR ‘grawitz tumour*’ OR ‘rcc’) OR (‘Kidney Neoplasms’[mhe]) OR (‘Carcinoma, Renal Cell’[mhe])) OR mRCC OR m-RCC or aRCC or a-RCC

MEDLINE (via Ovid)

Ovid MEDLINE(R) ALL <1946–7 October 2021>

The Cochrane Library (CENTRAL)

www.cochranelibrary.com/

Cochrane Central Register of Controlled Trials

Issue 10 of 12 October 2021

EMBASE (via Ovid)

EMBASE <1974–7 October 2021>

PubMed

https://pubmed.ncbi.nlm.nih.gov/

((((‘carcinoma, renal cell’[MeSH Terms] OR ‘Kidney Neoplasms’[MeSH Terms] OR (‘renal cancer*’[Text Word] OR ‘renal carcinoma*’[Text Word] OR ‘renal adenocarcinoma*’[Text Word] OR ‘renal tumor*’[Text Word] OR ‘renal tumour*’[Text Word] OR ‘renal malignanc*’[Text Word]) OR (‘kidney cancer*’[Text Word] OR ‘kidney carcinoma*’[Text Word] OR ‘kidney adenocarcinoma*’[Text Word] OR ‘kidney tumor*’[Text Word] OR ‘kidney tumour*’[Text Word] OR ‘kidney malignanc*’[Text Word]) OR (‘clear cell cancer*’[Text Word] OR ‘clear cell carcinoma*’[Text Word] OR ‘clear cell adenocarcinoma*’[Text Word] OR ‘clear cell tumor*’[Text Word] OR ‘clear cell tumour*’[Text Word] OR ‘clear cell malignanc*’[Text Word]) OR (‘non clear cell cancer*’[Text Word] OR ‘non clear cell carcinoma*’[Text Word] OR ‘non clear cell adenocarcinoma*’[Text Word] OR ‘non clear cell tumor*’[Text Word] OR ‘non clear cell tumour*’[Text Word]) OR ‘hypernephroma’[Text Word] OR ‘hypernephroid carcinoma*’[Text Word] OR (‘grawitz tumor*’[Text Word] OR ‘grawitz tumour*’[Text Word]) OR ‘rcc’[Text Word]) AND (‘advanced’[Text Word] OR ‘metastatic’[Text Word] OR ‘mRCC’[Text Word] OR ‘m-RCC’[Text Word] OR ‘aRCC’[Text Word] OR ‘a-RCC’[Text Word] OR ‘first-line’[Text Word] OR ‘first line’[Text Word] OR ‘metastasize’[Text Word] OR ‘metastasis’[Text Word] OR ‘metastases’[Text Word] OR ‘stage iii’[Text Word] OR ‘stage 3’[Text Word] OR ‘stage 4’[Text Word] OR ‘stage iv’[Text Word] OR ‘recurrent’[Text Word] OR ‘non resectable’[Text Word] OR ‘inoperable’[Text Word] OR ‘non operable’[Text Word] OR ‘unresectable’[Text Word] OR ‘Neoplasm Metastasis’[MeSH Terms])) AND (‘Economics’ OR ‘Costs and Cost Analysis’[mh] OR ‘Economics, Nursing’[mh] OR ‘Economics, Medical’[mh] OR ‘Economics, Pharmaceutical’[mh] OR ‘Economics, Hospital’[mh] OR ‘Economics, Dental’[mh] OR ‘Fees and Charges’[mh] OR ‘Budgets’[mh] OR budget*[tiab] OR economic*[tiab] OR cost[tiab] OR costs[tiab] OR costly[tiab] OR costing[tiab] OR price[tiab] OR prices[tiab] OR pricing[tiab] OR pharmacoeconomic*[tiab] OR pharmaco-economic*[tiab] OR expenditure[tiab] OR expenditures[tiab] OR expense[tiab] OR expenses[tiab] OR financial[tiab] OR finance[tiab] OR finances[tiab] OR financed[tiab] OR value for money[tiab] OR monetary value*[tiab] OR ‘models, economic’[mh] OR economic model*[tiab] OR ‘markov chains’[mh] OR markov[tiab] OR ‘monte carlo method’[mh] OR monte carlo[tiab] OR ‘Decision Theory’[mh] OR decision tree*[tiab] OR decision analy*[tiab] OR decision model*[tiab])) AND ((english[Filter]) AND (2006:2021[pdat])))

NHS Economic Evaluation Database via Centre for Reviews and Dissemination

www.crd.york.ac.uk/CRDWeb/

1. MeSH DESCRIPTOR Carcinoma, Renal Cell EXPLODE ALL TREES

2. MeSH DESCRIPTOR Kidney Neoplasms EXPLODE ALL TREES

3. (‘renal cancer*’)

4. (‘renal carcinoma*’)

5. (‘renal adenocarcinoma*’)

6. (‘renal tumor*’)

7. (‘renal tumour*’)

8. (‘renal malignanc*’)

9. (‘kidney cancer*’)

10. (‘kidney carcinoma*’)

11. (‘kidney adenocarcinoma*’)

12. (‘kidney tumor*’)

13. (‘kidney tumour*’)

14. (‘kidney malignanc*’)

15. (‘clear-cell cancer*’)

16. (‘clear-cell carcinoma*’)

17. (‘clear-cell adenocarcinoma*’)

18. (‘clear-cell tumor*’)

19. (‘clear-cell tumour*’)

20. (‘clear-cell malignanc*’)

21. (‘non-clear cell cancer*’)

22. (‘non-clear cell carcinoma*’)

23. (‘non-clear cell adenocarcinoma*’)

24. (‘non-clear cell tumor*’)

25. (‘non-clear cell tumour*’)

26. (‘non-clear cell malignanc*’)

27. (hypernephroma)

28. (hypernephroid carcinoma*)

29. (grawitz tumor*)

30. (grawitz tumour*)

31. (rcc)

32. #1 OR #2 OR #3 OR #4 OR #5 OR #6 OR #7 OR #8 OR #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15 OR #16 OR #17 OR #18 OR #19 OR #20 OR #21 OR #22 OR #23 OR #24 OR #25 OR #26 OR #27 OR #28 OR #29 OR #30 OR #31

33. (advanced)

34. (metastatic)

35. (mRCC)

36. (m-RCC)

37. (aRCC)

38. (a-RCC)

39. (‘first-line’ or ‘first line’)

40. (metastasize)

41. (metastasis)

42. (metastases)

43. (‘stage iii’)

44. (‘stage 3’)

45. (‘stage 4’)

46. (‘stage iv’)

47. (recurrent)

48. (‘non resectable’)

49. (inoperable)

50. (‘non operable’)

51. (unresectable)

52. MeSH DESCRIPTOR Neoplasm Metastasis EXPLODE ALL TREES

53. #33 OR #34 OR #35 OR #36 OR #37 OR #38 OR #39 OR #40 OR #41 OR #42 OR #43 OR #44 OR #45 OR #46 OR #47 OR #48 OR #49 OR #50 OR #51 OR #52

54. #32 AND #53

55. (mrcc)

56. (m-rcc)

57. (arcc)

58. (a-rcc)

59. #55 OR #56 OR #57 OR #58

60. #54 OR #59

EconLit (via EBSCOhost)

1. S1 TI ((renal N2 (cancer* or carcinoma* or adenocarcinoma* or tumo#r* or malignanc*))) OR AB ((renal N2 (cancer* or carcinoma* or adenocarcinoma* or tumo#r* or malignanc*))) OR SU ((renal N2 (cancer* or carcinoma* or adenocarcinoma* or tumo#r* or malignanc*)))

2. S2 TI ((kidney N1 (cancer* or carcinoma* or adenocarcinoma* or tumo#r* or malignanc*)) OR AB ((kidney N1 (cancer* or carcinoma* or adenocarcinoma* or tumo#r* or malignanc*)) OR SU ((kidney N1 (cancer* or carcinoma* or adenocarcinoma* or tumo#r* or malignanc*))

3. S3 TI ((clear-cell N3 (cancer* or carcinoma* or adenocarcinoma* or tumo#r* or malignanc*))) OR AB ((clear-cell N3 (cancer* or carcinoma* or adenocarcinoma* or tumo#r* or malignanc*))) OR SU ((clear-cell N3 (cancer* or carcinoma* or adenocarcinoma* or tumo#r* or malignanc*)))

4. S4 TI ((‘clear cell’ N3 (cancer* or carcinoma* or adenocarcinoma* or tumo#r* or malignanc*))) OR AB ((‘clear cell’ N3 (cancer* or carcinoma* or adenocarcinoma* or tumo#r* or malignanc*))) OR SU ((‘clear cell’ N3 (cancer* or carcinoma* or adenocarcinoma* or tumo#r* or malignanc*)))

5. S5 TI ((non-clear-cell N3 (cancer* or carcinoma* or adenocarcinoma* or tumo#r* or malignanc*))) OR AB ((non-clear-cell N3 (cancer* or carcinoma* or adenocarcinoma* or tumo#r* or malignanc*))) OR SU ((non-clear-cell N3 (cancer* or carcinoma* or adenocarcinoma* or tumo#r* or malignanc*)))

6. S6 TI ((‘non clear cell’ N3 (cancer* or carcinoma* or adenocarcinoma* or tumo#r* or malignanc*))) OR AB ((‘non clear cell’ N3 (cancer* or carcinoma* or adenocarcinoma* or tumo#r* or malignanc*))) OR SU ((‘non clear cell’ N3 (cancer* or carcinoma* or adenocarcinoma* or tumo#r* or malignanc*)))

7. S7 TI hypernephroma OR AB hypernephroma OR SU hypernephroma

8. S8 TI ‘hypernephroid carcinoma*’ OR AB ‘hypernephroid carcinoma*’ OR SU ‘hypernephroid carcinoma*’

9. S9 TI grawitz tumo#r* OR AB grawitz tumo#r* OR SU grawitz tumo#r*

10. S10 TI rcc OR AB rcc OR SU rcc

11. S11 S1 OR S2 OR S3 OR S4 OR S5 OR S6 OR S7 OR S8 OR S9 OR S10

12. S12 TI (advanced or metastatic or mRCC or m-RCC or aRCC or a-RCC or ‘first-line’ or ‘first line’ or metastasize or metastasis or metastases or ‘stage iii’ or ‘stage 3’ or ‘stage 4’ or ‘stage iv’ or recurrent or ‘non resectable’ or inoperable or ‘non operable’ or unresectable) OR AB (advanced or metastatic or mRCC or m-RCC or aRCC or a-RCC or ‘first-line’ or ‘first line’ or metastasize or metastasis or metastases or ‘stage iii’ or ‘stage 3’ or ‘stage 4’ or ‘stage iv’ or recurrent or ‘non resectable’ or inoperable or ‘non operable’ or unresectable) OR SU (advanced or metastatic or mRCC or m-RCC or aRCC or a-RCC or ‘first-line’ or ‘first line’ or metastasize or metastasis or metastases or ‘stage iii’ or ‘stage 3’ or ‘stage 4’ or ‘stage iv’ or recurrent or ‘non resectable’ or inoperable or ‘non operable’ or unresectable)

13. S13 S11 AND S12

14. S14 TI (mRCC OR m-RCC or aRCC or a-RCC) OR AB (mRCC OR m-RCC or aRCC or a-RCC) OR SU (mRCC OR m-RCC or aRCC or a-RCC)

15. S15 S13 OR S14

16. S16 S13 OR S14

Narrow by Language: - English, Published: 20060101-20211231

CEA Registry

https://cevr.tuftsmedicalcenter.org/databases/cea-registry

advanced renal cell

metastatic renal cell

advanced kidney

metastatic kidney

mRCC

aRCC

first-line renal cell

first-line kidney

first line renal cell

first line kidney

lenvatinib

sunitinib

pazopanib

tivozanib

cabozantinib

nivolumab

Note: Basic search only with free version of CEA Registry. No Boolean. No download function. Screened on website

ClinicalTrials.gov

https://clinicaltrials.gov/

(((advanced OR metastatic OR secondary OR EXPAND[Concept] ‘first-line’ OR EXPAND[Concept] ‘first line’ OR metastasis or mRCC or m-RCC OR aRCC OR a-RCC OR metastasize OR metastasis OR metastases OR EXPAND[Concept] ‘stage iii’ OR EXPAND[Concept] ‘stage 3’ OR EXPAND[Concept] ‘stage 4’ OR EXPAND[Concept] ‘stage iv’ OR recurrent OR EXPAND[Concept] ‘non resectable’ OR EXPAND[Concept] ‘non-resectable’ OR inoperable OR EXPAND[Concept] ‘non operable’ OR EXPAND[Concept] ‘non-operable’ OR unresectable) AND AREA[ConditionSearch] (EXPAND[Concept] ‘Renal cell’ OR EXPAND[Concept] ‘renal clear cell’ OR EXPAND[Concept] ‘renal clear-cell’ OR EXPAND[Concept] ‘renal non-clear cell’ OR EXPAND[Concept] ‘renal non clear cell’ OR RCC OR EXPAND[Concept] ‘renal carcinoma’ OR EXPAND[Concept] ‘renal cancer’ OR EXPAND[Concept] ‘renal tumor’ OR EXPAND[Concept] ‘renal tumour’ OR EXPAND[Concept] ‘renal adenocarcinoma’ OR EXPAND[Concept] ‘renal malignancy’ OR EXPAND[Concept] ‘kidney cancer’ OR EXPAND[Concept] ‘kidney carcinoma’ OR EXPAND[Concept] ‘kidney adenocarcinoma’ OR EXPAND[Concept] ‘kidney tumor’ OR EXPAND[Concept] ‘kidney tumour’ OR EXPAND[Concept] ‘kidney malignancy’ OR EXPAND[Concept] ‘clear-cell cancer’ OR EXPAND[Concept] ‘clear cell cancer’ OR EXPAND[Concept] ‘clear-cell carcinoma’ OR EXPAND[Concept] ‘clear cell carcinoma’ OR EXPAND[Concept] ‘clear-cell adenocarcinoma’ OR EXPAND[Concept] ‘clear cell adenocarcinoma’ OR EXPAND[Concept] ‘clear-cell tumor’ OR EXPAND[Concept] ‘clear cell tumor’ OR EXPAND[Concept] ‘clear-cell tumour’ OR EXPAND[Concept] ‘clear cell tumour’ OR EXPAND[Concept] ‘clear-cell malignancy’ OR EXPAND[Concept] ‘clear cell malignancy’ OR EXPAND[Concept] ‘non-clear cell cancer’ OR EXPAND[Concept] ‘non clear cell cancer’ OR EXPAND[Concept] ‘non-clear cell carcinoma’ OR EXPAND[Concept] ‘non clear cell carcinoma’ OR EXPAND[Concept] ‘non-clear cell adenocarcinoma’ OR EXPAND[Concept] ‘non clear cell adenocarcinoma’ OR EXPAND[Concept] ‘non-clear cell tumor’ OR EXPAND[Concept] ‘non clear cell tumor’ OR EXPAND[Concept] ‘non-clear cell tumour’ OR EXPAND[Concept] ‘non clear cell tumour’ OR EXPAND[Concept] ‘non-clear cell malignancy’ OR EXPAND[Concept] ‘non clear cell malignancy’ OR hypernephroma OR EXPAND[Concept] ‘hypernephroid carcinoma’ OR grawitz)) OR (aRCC OR mRCC or a-RCC OR m-RCC)) AND (economic OR economics OR cost OR costs OR costly OR costing OR budget OR price OR prices OR pricing OR pharmacoeconomics OR pharmaco-economics OR expenditure OR expenditures OR expense OR expenses OR financial OR finance OR finances OR financed OR EXPAND[Concept] ‘value for money’ OR EXPAND[Concept] ‘monetary value’ OR EXPAND[Concept] ‘economic model’ OR EXPAND[Concept] ‘economic models’ OR markov OR monte carlo OR EXPAND[Concept] ‘Decision Theory’ OR EXPAND[Concept] ‘decision tree’ OR EXPAND[Concept] ‘decision analysis’ OR EXPAND[Concept] ‘decision model’)

International Clinical Trials Registry Platform

https://trialsearch.who.int/

Search 1:

TITLE: (economic OR economics OR cost OR costs OR costly OR costing OR budget OR price OR prices OR pricing OR pharmacoeconomics OR pharmaco-economics OR expenditure OR expenditures OR expense OR expenses OR financial OR finance OR finances OR financed OR ‘value for money’ OR ‘monetary value’ OR ‘economic model’ OR ‘ economic models’ OR markov OR monte carlo OR ‘Decision Theory’ OR decision tree OR decision analysis OR decision model)

AND

CONDITION: ‘renal cell’ OR ‘clear-cell’ OR ‘clear cell’ OR RCC OR ‘kidney cancer*’ OR ‘renal cancer*’ OR ‘renal carcinoma*’ OR ‘renal adenocarcinoma’ OR ‘renal tumor*’ OR ‘renal tumour*’ OR hypernephroma OR ‘hypernephroid carcinoma’ OR grawitz

Search 2:

TITLE: (economic OR economics OR cost OR costs OR costly OR costing OR budget OR price OR prices OR pricing OR pharmacoeconomics OR pharmaco-economics OR expenditure OR expenditures OR expense OR expenses OR financial OR finance OR finances OR financed OR ‘value for money’ OR ‘monetary value’ OR ‘economic model’ OR ‘ economic models’ OR markov OR monte carlo OR ‘Decision Theory’ OR decision tree OR decision analysis OR decision model)

AND

CONDITION: (aRCC OR mRCC or a-RCC OR m-RCC)

Note: Limited to 2006 onwards

Parentheses (brackets) cannot be used to determine the order in which terms are combined.

Searches automatically include synonyms generated using the UMLS metathesaurus.

Searches are restricted to 256 character spaces per line – truncated strategies used

International Health Technology Assessment Database

https://database.inahta.org/

((‘Neoplasm Metastasis’[mhe]) OR (advanced OR metastatic OR mRCC OR m-RCC OR aRCC OR a-RCC OR ‘first-line’ OR ‘first line’ OR metastasize OR metastasis OR metastases OR ‘stage iii’ OR ‘stage 3’ OR ‘stage 4’ OR ‘stage iv’ OR recurrent OR ‘non resectable’ OR inoperable OR ‘non operable’ OR unresectable)) AND ((‘renal cancer*’ OR ‘renal carcinoma*’ OR ‘renal adenocarcinoma*’ OR ‘renal tumor*’ OR ‘renal tumour*’ OR ‘renal malignanc*’ OR ‘kidney cancer*’ OR ‘kidney carcinoma*’ OR ‘kidney adenocarcinoma*’ OR ‘kidney tumor*’ OR ‘kidney tumour*’ OR ‘kidney malignanc*’ OR ‘clear cell cancer*’ OR ‘clear cell carcinoma*’ OR ‘clear cell adenocarcinoma*’ OR ‘clear cell tumor*’ OR ‘clear cell tumour*’ OR ‘clear cell malignanc*’ OR ‘non clear cell cancer*’ OR ‘non clear cell carcinoma*’ OR ‘non clear cell adenocarcinoma*’ OR ‘non clear cell tumor*’ OR ‘non clear cell tumour*’ OR ‘hypernephroma’ OR ‘hypernephroid carcinoma*’ OR ‘grawitz tumor*’ OR ‘grawitz tumour*’ OR ‘rcc’) OR (‘Kidney Neoplasms’[mhe]) OR (‘Carcinoma, Renal Cell’[mhe])) OR mRCC OR m-RCC or aRCC or a-RCC