The Impact of Brain Metastases and Associated Neurocognitive Aspects on Health Utility Scores in EGFR Mutated and ALK Rearranged NSCLC: A Real World Evidence Analysis

With improved outcomes, more lung cancer patients are being diagnosed with brain metastases. This article evaluates the correlations between brain metastasis and health utility scores in lung cancer patients.

Abstract

Background.

In lung cancer, brain metastases (BM) and their treatment are associated with high economic burden and inferior health‐related quality of life. In the era of targeted therapy, real world evidence through health utility scores (HUS) is critical for economic analyses.

Materials and Methods.

In a prospective observational cohort study (2014–2016), outpatients with stage IV lung cancer completed demographic and EQ‐5D‐3L surveys (to derive HUS). Health states and clinicopathologic variables were obtained from chart abstraction. Patients were categorized by the presence or absence of BM; regression analyses identified factors that were associated with HUS. A subset of patients prospectively completed neurocognitive function (NCF) tests and/or the FACT‐brain (FACT‐Br) questionnaire, which were then correlated with HUS (Spearman coefficients; regression analyses).

Results.

Of 519 patients with 1,686 EQ‐5D‐3L‐derived HUS, 94 (18%) completed NCF tests and 107 (21%) completed FACT‐Br; 301 (58%) never developed BM, 24 (5%) developed first BM during study period, and 194 (37%) had BM at study entry. The sample was enriched (46%) for EGFR mutations (EGFRm) and ALK‐rearrangements (ALKr). There were no HUS differences by BM status overall and in subsets by demographics. In multivariable analyses, superior HUS was associated with having EGFRm/ALKr (p < .0001), no prior radiation for extracranial disease (p < .001), and both intracranial (p = .002) and extracranial disease control (p < .01). HUS correlated with multiple elements of the FACT‐Br and tests of NCF.

Conclusion.

Having BM in lung cancer is not associated with inferior HUS in a population enriched for EGFRm and ALKr. Patients exhibiting disease control and those with oncogene‐addicted tumors have superior HUS.

Implications for Practice.

In the setting of EGFR mutations or ALK rearrangement non‐small cell lung cancer (NSCLC), a diagnosis of brain metastases no longer consigns the patient to an inferior health state suggesting that new economic analyses in NSCLC are needed in the era of targeted therapies. Additionally, the EQ‐5D questionnaire is associated with measures of health‐related quality of life and neurocognitive scores suggesting this tool should be further explored in prospective clinical studies.

Introduction

Despite lung cancer having high mortality, morbidity, and poor prognosis [1], [2], molecular and immunohistochemical subtyping (programmed cell death ligand 1 expression) has led to survival gains with targeted therapies and immune checkpoint inhibitors [3]. With improved outcomes and the prevalent use of sensitive imaging techniques such as magnetic resonance imaging (MRI), more patients with lung cancer are diagnosed with brain metastases (BM). Patients with targetable driver mutations in the epidermal growth factor receptor (EGFR) gene (EGFRm) and gene rearrangements in anaplastic lymphoma kinase (ALK; ALKr) and ROS1 may predispose to BM development [4]; however, increases in incidence and prevalence may simply reflect a longer survival in these subgroups.

Multiple BM are typically treated with whole brain radiation (WBRT), which can be associated with cognitive decline [7], [8], [9]; however, feasibility of stereotactic radiosurgery (SRS) has also been demonstrated [5]. SRS has potential advantages over WBRT, including shorter treatments and less cognitive impairment [6]. Although most EGFR and ALK tyrosine kinase inhibitors (TKIs) can cross the blood‐brain barrier, central nervous system (CNS) progression will be the first site of disease progression in one third of EGFRm patients; resistance mechanisms may differ from those causing systemic progression [7], [8]. More recent targeted therapies such as osimertinib and alectinib have been shown to be highly effective in treating CNS disease [9], [10].

In unselected patients with metastatic lung cancer, the life expectancy after BM is variable, ranging between 3–15 months [11]. The presence of EGFRm or ALKr offers an improved prognosis, with an impressive median overall survival of 46 months in EGFRm patients treated with SRS, when compared with 30 and 25 months in cohorts treated with WBRT and an EGFR‐TKI, respectively [12]. Similarly ALKr non‐small cell lung cancer (NSCLC) has a reported median survival of 49.5 months from discovery of BM [13]. Factors such as age, number of lesions, performance status, and extracranial control have been consistent influencing factors [11], [12], [13].

As novel therapies and drug sequencing for BM continue to emerge and the quantity of life (i.e., overall survival) improves, consideration of quality of life becomes important in medical decision‐making. In addition, economic analyses may fail to account for the recent paradigm shifts in the way molecular subsets of lung cancer are managed and its subsequent consequences [14]. Health utility scores (HUS) depict a patient's preference for a specific health‐related outcome, where “0” represents death and “1” represents perfect health [15]. HUS can be used to generate quality‐adjusted life years required in cost‐effectiveness analyses. The EQ‐5D‐3L is one of the most commonly used HUS tools in patients with cancer, with specific Canadian values reported [16]. Although we have presented HUS results in patients with advanced lung cancer [17], HUS in patients with lung cancer specifically with BM have not been reported. This study aims to evaluate the impact of BM on HUS in lung cancer patients, enriched for EGFRm and ALKr.

Three questions were addressed in this study. First, we evaluated whether having BM in this new era of targeted therapy alone, after adjusting for other clinico‐demographic factors, consigned a patient to a worse HUS. Second, we evaluated whether the development of new BM in patients who had not previously developed BMs leads to worsening HUS, at least temporarily. Third, we evaluated whether neurocognitive symptoms, assessed either through neurocognitive function testing or through a survey of neurocognitive symptoms and brain‐related quality of life, correlated to any extent with HUS, that is, whether a component of the HUS is driven partly by neurocognitive issues that can occur in lung cancer patients with BM. Knowing how BM affects HUS in the modern therapy era can help guide the translation of future therapies, both systemic and local, into the clinical setting.

Materials and Methods

Study Design, Patient Sample, and Procedures

A research‐ethics approved, prospective observational study of HUS, as measured by EQ‐5D‐3L, in specific health states of patients with advanced lung cancer began in November 2014 at the Princess Margaret Cancer Centre (Toronto, Canada). Patients could enrol at any time point. Questionnaires were completed at each outpatient visit provided consent was given. The last datapoint analyzed occurred in August 2016.

Eligible patients had histologically confirmed lung cancer, were able to provide consent, and had no significant cognitive deficits limiting consent or capacity to fill out questionnaires. Patients on study completed, either electronically or on paper, a demographic survey together with the EQ‐5D‐3L questionnaire, which focused on five dimensions (mobility, self‐care, usual activity, pain or discomfort, and anxiety or depression) and a visual analog scale of 0–100 rating of health [18]. Patients either had to speak or read English or have an interpreter translate all parts of the study or the demographic questionnaire, with the exception of the main EQ‐5D‐3L survey, which could be provided in multiple languages and had to be completed by the patient; if illiterate, the EQ‐5D‐3L could be administered orally, but the patient had to supply answers. Recruitment was enriched for EGFRm and ALKr patients as outpatient lists of EGFRm/ALKr patients directed coordinators to specific clinics; however, once in that specific clinic, all additional patients with NSCLC and small‐cell lung cancer (SCLC) were approached for recruitment to avoid additional selection bias.

Clinical data were extracted from medical records as previously reported [17]. At first diagnosis, computed tomography (CT) imaging of the thorax and abdomen and brain imaging (typically MRI brain) were standard. In assessing response to therapy or at follow‐up, all patients had at least one radiologic test (typically chest radiograph or CT scans) and bloodwork available. Abstracted data included date of BM diagnosis, symptoms at diagnosis and at each encounter (headache, seizure, or focal neurological deficits), treatment received, and state of CNS disease at encounter (controlled, uncontrolled, unknown). Dates of last MRI or CT brain relative to encounter were documented. Control was defined as evidence of stability or any reported BM response on imaging and/or absence of symptoms. Patients were stratified by encounter according to BM status: those with a known diagnosis of BM (BM+) and those without a diagnosis of BM (NBM) at encounter. In those patients who developed BM on study, the mean HUS (mHUS) pre‐ and post‐BM were interrogated. Longitudinal analyses of HUS were conducted pre‐ and post‐WBRT as well as in patients remaining on EGFR‐TKI post‐BM diagnosis to ascertain the impact of CNS progression on HUS despite continuing therapy. Patients were grouped according to subtypes: EGFRm, ALKr, other NSCLC (i.e., not EGFRm or ALKr), and SCLC.

From May to August 2016, patients were additionally invited to undergo two substudies by completing the Functional Assessment of Cancer Therapy‐Brain (FACT‐Br) questionnaire [19] and a battery of neurocognitive function (NCF) tests. The BM+ patients were selected at random, whereas age, gender, smoking distribution‐matched patients with NBM lung cancer were also evaluated for comparison purposes, selected after enrollment of the BM+ cohort. The FACT‐Br questionnaire, originally developed for patients with primary brain tumors, is also relevant in patients with BM [20]. Total scores range from 0–200 (higher values indicate better health‐related quality of life [HRQoL]), and include the brain cancer‐specific (FACT‐Br) subscale and the FACT‐general (FACT‐G), which itself includes physical well‐being, social/family well‐being, emotional well‐being, and functional well‐being subscales.

Tests of NCF included the Hopkins Verbal Learning Test‐Revised (HVLT‐R) [21], [22], Trail Making Test parts A and B (TMT‐A/B) [23], and Controlled Oral Word Association Test (COWAT) [24], [25]. The HVLT‐R assesses learning and memory for verbal information (a list of words) with age‐normed scores including total immediate recall, delayed recall, retention (i.e., percentage of learned words that are recalled at the delay), and recognition (i.e., ability to discriminate list words from new words). The TMT‐A assesses information processing speed. The TMT‐B assesses cognitive flexibility, and the COWAT assesses verbal fluency; both are used as measures of executive functioning. Raw scores from both the TMT and COWAT were converted to demographically corrected T scores based on age, education, and/or sex according to published criteria.

Derived Variables and Statistical Analysis

The combination of answers to the five EQ‐5D‐3L questions were converted to a single HUS based upon a set of Canadian preference weights [16]. Descriptive summary statistics utilized Kruskal‐Wallis tests and Fisher's exact tests. For exploratory multiple regression analyses, we applied standard multiple least squares regression on health utilities adjusting for known or putative confounders: age, gender, and histo‐molecular subtype. For most analyses, only the first encounter (baseline) HUS for each patient was analyzed. For patients who developed BM during the study, all the HUS measurements were compared for pre‐ and post‐BM development using longitudinal analyses with compound symmetry assumption. The same was applied for longitudinal analyses in patients pre‐ and post‐WBRT and in those remaining on an EGFR‐TKI post‐BM development. Compound symmetry assumption was chosen because of smallest AIC [26]. Correlation between FACT‐Br or NCF tests and HUS was measured using Spearman rank correlation coefficients and regression analyses were used to assess interactions. All p values < .05 were considered significant. Analyses were conducted using SAS version 9.2 (SAS Institute, Cary, NC).

Results

Patient Characteristics

Of 1,686 encounters in 519 patients with lung cancer, less than 2% were illiterate (and had evaluations administered orally); the EQ‐5D‐3L was completed in English in over 92% of patients. Study participation rate in approached patients was 87%. At first encounter, 194 (37%) had a prior diagnosis of BM (BM+); of the remainder, during the study period, 301 (58%) patients never developed BM, and 24 (5%) patients developed BM and therefore were evaluated in both the BM+ and NBM groups (first encounter for NBM and first encounter post‐BM development). The mean number of assessments across all patients was 3.24 (SD, 2.88; range 1–17). The median time from lung cancer diagnosis to first encounter was 13 months in the BM+ group, and for NBM, 10 months. Patient characteristics at first encounter according to BM status are presented in Table 1 and according to the subset who developed BM during the study period in supplemental online Table 1. In brief, the BM+ and NBM groups had similar clinico‐demographics, except that BM+ patients were older, more likely to be female, and more likely to have EGFRm tumor subtypes than the NBM cohort.

Table 1. Baseline clinico‐demographic characteristics of the sample^a .

Significant p values are bolded.

^aValues are count, n, followed by percentage of total, in parentheses, unless otherwise stated.

Abbreviations: ALKR, anaplastic lymphoma kinase rearranged; BM, brain metastasis; ECOG, Eastern Cooperative Oncology Group; EGFRm, epidermal growth factor receptor mutated; met, metastases; SCLC, small cell lung cancer.

Does Having BM Alone Influence HUS, After Adjusting for Other Clinico‐Demographic Factors?

Using the first encounter data for each patient, the impact of clinical and treatment variables on HUS was evaluated. Univariable and multivariable analyses are presented in Table 2; mHUS by histo‐molecular subtype and BM status is presented in supplemental online Table 2. Patients with EGFRm or ALKr had overall higher mHUS compared with other patients with NSCLC and SCLC, confirming previous reports [17].

Table 2. Univariable analysis of factors associated with the health utility score (HUS).

Significant p values are bolded.

^aAdjusted for age, sex, histo‐molecular subtype.

Abbreviations: ALKR, anaplastic lymphoma kinase rearranged; EGFRm, epidermal growth factor receptor mutated; HUS, health utility score, NSCLC, non‐small cell lung cancer; SCLC, small cell lung cancer.

The presence of BM alone did not impact mHUS; neither did age, sex, or smoking status. There was a trend (p = .07) toward inferior HUS in those receiving chemotherapy (mHUS = 0.73, n = 104), when compared with targeted treatments (0.78, n = 176) and immunotherapy (0.80, n = 19). Patients who received prior palliative extracranial radiation had lower mHUS, whereas the absence of extracranial disease control also was associated with lower mHUS (both p < .001). Although a longer time between diagnosis and encounter was associated statistically with higher HUS, the magnitude of this last association was small and not clinically meaningful; for example, it would take a 86–110 month shorter time interval between diagnosis to encounter to have a similar effect on the mHUS as having prior palliative extracranial radiation or lack of extracranial disease control.

In multivariable analysis, the presence of an EGFR mutation/ALK rearrangement (p < .001), extracranial control at encounter (p < .001), and absence of prior radiotherapy for extracranial disease (p < .001) associated with higher HUS. Including the 24 patients who developed BM during the study (first encounter with BM), 218 assessments with BM+ patients were available for analysis. Univariable analyses are presented in Table 3, with stratification by histo‐molecular subtype in supplemental online Table 3. SCLC subtype and having BM symptoms at encounter were associated with inferior HUS (p = .004 and p = .01, respectively). There was a trend toward lower HUS in patients over 64 years with BM, mHUS 0.73 versus 0.77, (p = .05). The median time since WBRT in this group was 8 months (range, 0.5–166). Overall, individuals who had received WBRT at any point had marginally inferior HUS, (p = .06); however, longer time since first WBRT was associated with superior HUS (p = .02). In multivariable analysis, the ability to achieve CNS control was the only variable that remained significantly associated with higher HUS (p = .002).

Table 3. Univariable analysis of factors associated with health utility scores in patients with brain metastases.

Significant p values are bolded.

^aAdjusted for age, sex, histo‐molecular subtype.

Abbreviations: ALKR, anaplastic lymphoma kinase rearranged; EGFRm, epidermal growth factor receptor mutated; HUS, health utility score, NSCLC, non‐small cell lung cancer; SCLC, small cell lung cancer.

Longitudinal Analyses of NBM Patients Who Eventually Developed BM While on Study

In this study, HUS values over time were assessed in 24 original NBM patients who developed BM during the study period. The mean number of encounters pre‐BM was 2.83 and post‐BM was 2.46. In this small cohort of patients, the mHUS prediagnosis of BM was 0.80 (SD, 0.17) and after was 0.74 (0.15), with a difference of −0.06, p = .05, by compound symmetry method. Similarly, there was also a significant decline in HUS in patients who had received interval WBRT (n = 24; −0.071, p = .0026) compared with HUS before WBRT.

There were 41 patients who continued on an EGFR‐TKI post‐BM with at least two encounters. WBRT had been received by 22 (54%) patients at first encounter, and 4 (10%) had interval WBRT. Six (15%) patients had prior SRS, and one (2%) had interval SRS. Eight patients (19%) were treated with a TKI only. Most patients (n = 23, 56%) were treated with gefitinib, one patient each with erlotinib and afatinib, and 10 (24%) patients were receiving third‐generation TKIs. Six patients (15%) switched from first‐ to third‐generation TKIs during the study period. In longitudinal analysis, there was no association between mHUS and individual EGFR‐TKI (p > .20 for each TKI vs. non‐TKI). Time on EGFR‐TKI had a marginal effect on HUS (p = .058); however, longer time since BM diagnosis was associated with superior HUS (estimated HUS change per month +0.005, p = .01).

HUS and NCF

This subset analysis included 94 patients with HUS results who also consented to participate in NCF testing (54 BM+ and 40 NBM patients). Demographic information was similar for most variables between groups, including level of education completed; however, more EGFRm/ALKr patients were present in the BM+ group (supplemental online Table 4). In the BM+ patients, median time since BM diagnosis was 13 months (range, 0.5–178); 44% (n = 24) had received WBRT. mHUS were similar between groups: 0.77 for BM+ versus 0.78 for NBM (p = .86). Consistent with our overall sample, patients with stable BM had higher HUS than those with progressive BM (0.80 vs. 0.69; p = .045). Documented neurological symptoms alone did not impact NCF results. Performance on the COWAT TMT‐A and TMT‐B were inferior in BM+ compared with NBM patients, (mean T = 43.5 vs. 52.4, p = .017; 42.5 vs. 52.5, p = .02; and 41.1 vs. 54.1; p = .022, respectively, supplemental online Table 5).

Spearman rank correlations between HUS and NCF scores were variable (Table 4). Correlations were statistically significant for HVLT‐R total recall (rho = 0.35, p = .01) and recognition (rho = 0.32, p = .03) in BM+ patients only, for TMT‐B (rho = 0.30, p = .02) in NBM patients only, and for TMT‐A in NBM patients with a trend in BM+ (NBM rho = 0.39, p = .02, BM+ rho = 0.27, p = .06). Notably, in all patients, mean T scores for HVLT‐total recall and HVLT‐delayed recall were between 1–1.5 SD below population norms, indicating at least mild impairment regardless of BM status. In all cases, patients with better NCF endorsed higher HUS. In BM+ patients treated with WBRT (n = 24, 44%), the median time since WBRT was 9.5 months (0.5–172). Compared with BM+ patients not treated with WBRT (n = 30), those treated with WBRT (n = 24) had significantly poorer HLVT‐R total recall (mean T = 40.4 vs. 28.8, p < .0001) and delayed recall (mean T = 40.0 vs. 29.6, p = .006). WBRT was not associated with scores in TMT‐A/B or COWAT. Values are available in supplemental online Table 6. Mutational status had no bearing on these associations.

Table 4. Correlation between health utility scores and neurocognitive test scores according to brain metastases status.

rho, Spearman rank correlation; significant p values (<.05) are bolded.

Abbreviations: COWAT, Controlled Oral Word Association Test; HVLT‐R, Hopkins Verbal Learning Test‐Revised; TMT, Trail Making Test.

HUS and Symptoms/Brain‐Related Quality of Life as Measured by FACT‐Br

A cohort of 107 patients (65 BM+ and 42 NBM) completed the FACT‐Br questionnaire. Demographics in each cohort were similar. Except for the social well‐being subscale, all subscales (brain tumor‐specific questions, physical well‐being, emotional well‐being, and functional well‐being) and FACT‐G and total FACT‐Br scores were correlated with HUS (rho values ranged from 0.43 to 0.66, all p < .001) with similar strength of associations regardless of BM status (supplemental online Table 7).

Discussion

The treatment paradigm in lung cancer is rapidly evolving; in patients with advanced disease, a diagnosis of BM does not necessarily equate to lower HUS. Important determinants of HUS in this setting are histological subtype, prior palliative radiation for extracranial disease (a potential surrogate for prior disease bulk or prior symptomatic disease), and overall current disease control. This underscores the need for new utility studies in the era of precision medicine.

Superior HUS as derived by EQ‐5D was observed in patients with EGFRm/ALKr NSCLC (mHUS = 0.78) regardless of BM status, reflecting a general phenomenon inherent to oncogene‐addicted tumors. Similarly in large randomized trials, baseline and on treatment mHUS in patients with ALKr and EGFRm NSCLC ranged from 0.73–0.82, consistent with the real world values measured in our study [27], [28]. There was also a trend toward superior HUS in patients receiving targeted therapies in our analysis. The clear difference in HUS according to histo‐molecular subtype means that quality of life and economic data collected in the era prior to targeted agents may not be accurate for today's cost‐effectiveness analyses.

The importance of both systemic and intracranial disease control was also established in our study. Lack of disease progression has been correlated with superior HUS in several studies across multiple continents [29], [30]. We also identified that a longer time since first WBRT resulted in superior HUS, underscoring the proportion of patients who have sustained intracranial control and a prolonged survival. This result may be biased in that those patients who have died do not have HUS captured. Although numbers were small, an interval diagnosis of BM did impact HUS over a variable period of time and reflecting a change in health status, as did interval WBRT. Given the small numbers in the longitudinal analysis and lack of adjustment for confounding variables, these findings are considered hypothesis generating and require validation in large studies. HUS using EQ‐5D will be evaluated prospectively in a phase III trial of WBRT versus SRS in patients with 4 to 10 BM [31].

In our subset analysis, we found moderately strong correlations between HUS and several measures of NCF in patients with and without BM. This underscores the fact that function drives health utilities, rather than the simple act of having a BM. As MRI use becomes a standard baseline test in lung cancer, the presence of BM itself may reflect a new reality: the asymptomatic, small BM that does not affect HUS. Instead, the impact of BM may lie in its ability to affect neurocognitive function. Previously, it has been shown that NCF is associated with health‐related quality of life, as measured by FACT‐Br [32]. It is recognized that patients with cancer can have multiple reasons for impaired NCF, including extracranial or systemic factors such as chemotherapy use [33]. The presence of BM is consistently associated with NCF decline, including in patients prior to CNS‐directed treatment, as seen in a small pilot study investigating SRS in patients with newly diagnosed BM, 67% of whom had deficits in baseline (pretreatment) cognitive function [34]. In our study, lower HVLT‐R scores, particularly poorer learning (total recall scores), were seen in BM patients who had WBRT or lower HUS, extending data from three other studies that have highlighted neurocognitive toxicity associated with WBRT [6], [35], [36]. Although our study did not specifically investigate this, other studies have suggested that patients with EGFRm or ALKr lung cancer and BM may not develop the same level of neurotoxicity as other lung cancers [37]. Cognitive decline post‐WBRT is known to peak at 3–4 months, perhaps partly explaining why our patients who were further from first WBRT had relatively improved HUS; however, we did not have adequate numbers of patients to document whether the very long term patients were confounded by any decrease in cognition that affected insight into specific health states. Tumor control and CNS response have previously aligned with stable or improved NCF [38], [39].

Patients selected for WBRT tend to have worse baseline quality of life; in the Quartz trial of WBRT versus best supportive care, QALY difference and improvements in health‐related quality of life were minimal, questioning the benefit of treatment [40], [41], [42]. A number of other studies have in fact shown superior health‐related quality of life post‐WBRT. Consistently improved quality of life scores post‐WBRT were shown in a small study using FACT‐Br for up to 20 weeks post‐treatment [43]. Similarly, a prospective cohort study revealed improved palliative scores 1 month post‐WBRT [44]. These somewhat conflicting results suggest that when selected appropriately, patients benefit from WBRT especially if achieving durable control, and conceivably, such patients may have superior HUS as suggested by our study results.

Participation rate in our EQ‐5D analysis was 87% and for the additional two‐subset studies ranged between 85%–90%. This establishes the feasibility of using EQ‐5D in an observational, outpatient setting. EQ‐5D‐3L also associated with multiple domains of health‐related quality of life HRQoL, as measured by FACT‐Br, in patients with lung cancer with and without BM. The limitations of our study, however, include the targeted enrollment of patients with a known mutation. This was important to the aims of the study, as HUS in the era of molecularly targeted treatments have not been fully captured. We may also have failed to capture patients with BM who are too ill to complete questionnaires or those who have died, introducing potential bias. We had small numbers for the longitudinal analysis, which is key to evaluating changes in health states over time, and further longitudinal study in a larger sample is warranted.

Conclusion

Our results suggest that with the advent of newer, targeted therapies, having brain metastases in patients with advanced lung cancer no longer consigns the patient to inferior health states, as measured by health utility scores. Instead, HUS as measured by EQ‐5D are impacted by histo‐molecular subtype and extent of disease control. Their association with neurocognitive function suggests the latter is an important aspect of HRQoL, although further study is needed to understand the contribution of cognitive preservation or impairment to HRQoL and HUS in patients both with and without BM and how changes evolve over time. For economic analyses of novel drugs, regimens, and therapeutic approaches in patients with advanced lung cancer, many aspects of HRQoL may be captured reasonably by the EQ‐5D‐3L tool.

See http://www.TheOncologist.com for supplemental material available online.

Author Contributions

Conception/design: Grainne M. O'Kane, Hiten Naik, Nicole Mittmann, Nadine M. Richard, M. Catherine Brown, Wei Xu, Doris Howell, Geoffrey Liu

Provision of study material or patients: Grainne M. O'Kane, Hiten Naik, Penelope A. Bradbury, Natasha B. Leighl, Frances A. Shepherd, David Shultz, Geoffrey Liu

Collection and/or assembly of data: Grainne M. O'Kane, Brandon C. Tse, Vivian Tam, Tiffany Tse, Michael Borean, Emily Tam, Catherine Labbé, Mark K. Doherty, Kim Edelstein, Wei Xu, Geoffrey Liu

Data analysis and interpretation: Grainne M. O'Kane, Jie Su, Lin Lu, Catherine Labbé, Nadine M. Richard, Kim Edelstein, Wei Xu, Doris Howell, Geoffrey Liu

Manuscript writing: Grainne M. O'Kane, Michael Borean, Catherine Labbé, Nicole Mittmann, Mark K. Doherty, Penelope A. Bradbury, Natasha B. Leighl, Frances A. Shepherd, Kim Edelstein, David Shultz, M. Catherine Brown, Doris Howell, Geoffrey Liu

Final approval of manuscript: Grainne M. O'Kane, Jie Su, Brandon C. Tse, Vivian Tam, Tiffany Tse, Lin Lu, Michael Borean, Emily Tam, Catherine Labbé, Hiten Naik, Nicole Mittmann, Mark K. Doherty, Penelope A. Bradbury, Natasha B. Leighl, Frances A. Shepherd, Nadine M. Richard, Kim Edelstein, David Shultz, M. Catherine Brown, Wei Xu, Doris Howell, Geoffrey Liu

Disclosures

Catherine Labbé: AstraZeneca, Bristol‐Myers Squibb, Merck, Pfizer (C/A), AstraZeneca, Boehringer Ingelheim, Bristol‐Myers Squibb, Abbie, United Therapeutics, Xcovery, Roche (RF); Penelope A. Bradbury: Abbvie, Boehinger Ingelheim (C/A), Merck (H). The other authors indicated no financial relationships.

(C/A) Consulting/advisory relationship; (RF) Research funding; (E) Employment; (ET) Expert testimony; (H) Honoraria received; (OI) Ownership interests; (IP) Intellectual property rights/inventor/patent holder; (SAB) Scientific advisory board