- Email: [email protected]
The association between health-related quality-of-life scores and clinical outcomes in metastatic castration-resistant prostate cancer patients: Exploratory analyses of AFFIRM and PREVAIL studies
European Journal of Cancer 87 (2017) 21e29
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.ejcancer.com
Original Research
The association between health-related quality-of-life scores and clinical outcomes in metastatic castration-resistant prostate cancer patients: Exploratory analyses of AFFIRM and PREVAIL studies Tomasz M. Beer a,*, Kurt Miller b, Bertrand Tombal c, David Cella d, De Phung e, Stefan Holmstrom e, Cristina Ivanescu f, Konstantina Skaltsa g, Shevani Naidoo h a
OHSU Knight Cancer Institute, Oregon Health & Science University, 3303 S.W. Bond Avenue, CH14R, Portland, OR 97239, USA b Department of Urology, Charite´ e Universita¨tsmedizin Berlin, Charite´platz 1, 10117 Berlin, Germany c Service D’Urologie, Cliniques Universitaires Saint-Luc, Avenue Hippocrate 10, B-1200 Brussels, Belgium d Feinberg School of Medicine and Robert H. Lurie Comprehensive Cancer Center, Northwestern University, 633 N. St. Clair St., 19th Floor, Chicago, IL 60611, USA e Astellas Pharma Inc, Sylviusweg 62, 2300 AH Leiden, The Netherlands f QuintilesIMS, Siriusdreef 10, 2132 WT Hoofddorp, The Netherlands g QuintilesIMS, Sardenya 537, 08024 Barcelona, Spain h Astellas Pharma Inc, Global Health Economic Outcomes Research (HEOR), 2000 Hillswood Dr., Chertsey, Surrey KT16 0RS, UK Received 30 August 2017; accepted 25 September 2017
KEYWORDS Castration-resistant prostate cancer; Enzalutamide; Health-related quality of life; Survival
Abstract Background: Our exploratory analysis examined the association between healthrelated quality of life (HRQoL) (baseline and change over time) and clinical outcomes (overall survival [OS]/radiographic progression-free survival [rPFS]) in metastatic castration-resistant prostate cancer (mCRPC). Methods: HRQoL, OS and rPFS were assessed in phase III trials comparing enzalutamide with placebo in chemotherapy-naı¨ve (PREVAIL; NCT01212991) or post-chemotherapy (AFFIRM; NCT00974311) mCRPC. HRQoL was assessed using the Functional Assessment of Cancer Therapy-Prostate (FACT-P). Multivariate analyses evaluated the prognostic significance of baseline and time-dependent scores after adjusting for treatment and clinical/
* Corresponding author. E-mail addresses: [email protected] (T.M. Beer), [email protected] (K. Miller), [email protected] (B. Tombal), d-cella@ northwestern.edu (D. Cella), [email protected] (D. Phung), [email protected] (S. Holmstrom), cristina.ivanescu@ quintilesims.com (C. Ivanescu), [email protected] (K. Skaltsa), [email protected] (S. Naidoo). https://doi.org/10.1016/j.ejca.2017.09.035 0959-8049/ª 2017 Elsevier Ltd. All rights reserved.
22
T.M. Beer et al. / European Journal of Cancer 87 (2017) 21e29
demographic variables. Hazard ratios (HRs) and 95% confidence intervals (CIs) represented the hazard of rPFS or OS per minimally important difference (MID) score change in HRQoL variables. Results: In baseline and time-dependent multivariate analyses, OS was independently associated with multiple HRQoL measures across both studies. In time-dependent analyses, a 10point (upper bound of MID range) increase (improvement) in FACT-P total score was associated with reductions in mortality risk of 19% in AFFIRM (HR 0.81 [95% CI 0.78e0.84]) and 21% in PREVAIL (HR 0.79 [0.76e0.83]). For baseline analyses, a 10-point increase in FACTP total score was associated with reductions in mortality risk of 12% (HR 0.88 [0.84e0.93]) and 10% (HR 0.90 [0.86e0.95]) in AFFIRM and PREVAIL, respectively. rPFS was associated with a subset of HRQoL domains in both studies. Conclusion: Several baseline HRQoL domains were prognostic for rPFS and OS in patients with mCRPC, and this association was maintained during treatment, indicating that changes in HRQoL are informative for patients’ expected survival. ª 2017 Elsevier Ltd. All rights reserved.
1. Introduction Health-related quality of life (HRQoL) is an important end-point in oncology studies [1]. Patient-reported outcomes (PROs) can be associated with overall survival (OS) or progression-free survival (PFS) in various cancers [2e5]. Thus, the patient’s perspective on changes in HRQoL, and more specifically PROs, is an important determinant of the value of cancer treatment. The Prostate Cancer Clinical Trials Working Group 3 acknowledged the need to optimise assessment and analysis of PRO data and recommended evaluating the association between early changes in individual outcome measures (e.g. PROs) and later events such as radiographic PFS (rPFS) or OS [6]. In metastatic castrationresistant prostate cancer (mCRPC), improvement in HRQoL has been associated with improved clinical outcomes [7,8]. In addition, PRO data from 19 mCRPC clinical trials indicated that HRQoL and pain status augmented clinical efficacy data by providing better understanding of treatment impact [9]. However, information on relationships between commonly used efficacy outcome measures, such as rPFS or OS, and HRQoL in mCRPC is limited. The AFFIRM (ClinicalTrials.gov, NCT00974311) [10] and PREVAIL (ClinicalTrials.gov, NCT01212991) [11] phase III trials showed that, compared with placebo, enzalutamide significantly prolonged OS and rPFS and displayed HRQoL benefits in men with mCRPC in postchemotherapy and chemotherapy-naı¨ve settings. These exploratory analyses investigate the association between HRQoL and rPFS and OS in the AFFIRM and PREVAIL studies at baseline and during treatment. 2. Methods 2.1. Study populations and design PREVAIL and AFFIRM were phase III, randomised, double-blind, placebo-controlled trials comparing oral
enzalutamide 160 mg/day with placebo in postchemotherapy (AFFIRM) or chemotherapy-naı¨ve (PREVAIL) mCRPC patients. The trials’ design, patient populations and results have been described previously [10,11]. Both trials were approved by local independent review boards and conducted according to the Declaration of Helsinki and Good Clinical Practice guidelines. All patients provided written informed consent. 2.2. Health-related quality of life In both studies, HRQoL was assessed using the Functional Assessment of Cancer Therapy-Prostate (FACTP), version 4, a 39-item instrument validated for use in mCRPC [12e14]. HRQoL data were collected at baseline (i.e. treatment day 1), at week 5 (PREVAIL only), at week 13 and every subsequent 12 weeks until study drug discontinuation. 2.3. Statistical analyses In this post hoc exploratory analysis, we investigated the relationship between rPFS and OS and the following HRQoL measures: prostate cancerespecific subscale (PCS), PCS pain-related score, physical well-being (PWB), emotional well-being (EWB), social/family wellbeing (SWB), functional well-being (FWB), FACT-P total score, FACT-General (FACT-G) total score (the sum of PWB, FWB, SWB and EWB), FACT Advanced Prostate Symptom Index (FAPSI) and trial outcome index (TOI) (the sum of PWB, FWB and PCS). The analyses were performed on intent-to-treat patient populations (i.e. all patients who were randomised into the studies). Cox proportional hazard models with baseline (i.e. fixed) or time-dependent covariates were fitted to timeto-event data (OS and, separately, rPFS). OS was defined as the time from randomisation to death from
T.M. Beer et al. / European Journal of Cancer 87 (2017) 21e29
any cause. rPFS was defined as the time from randomisation to first objective evidence of radiographic progression or death due to any cause. Demographic, clinical and HRQoL variables were evaluated in a univariate analysis to determine which parameters showed individual prognostic value for rPFS or OS. Note that here, P-values were not used to reject a null hypothesis but served only as a criterion for inclusion in multivariate analyses, so multiplicity adjustment was not applied. In multivariate analyses, joint prognostic significance of selected demographic and clinical variables (using a significance level of 0.05), and all baseline HRQoL or time-dependent HRQoL variables, were investigated. To minimise instability of the final multivariate model potentially resulting from high multicollinearity, separate models were fit for each of the following domains: FACT-P total score, FACT-G total score, FACT-P TOI, FAPSI and FACT-P PCS pain-related score. The remaining sub domains (PWB, FWB, SWB, EWB and PCS) were introduced in a separate single model and no other domains were included in that model. The complete list of covariates included in each model is listed in Appendix A. Since a single hypothesis is not being tested, but rather a number of hypotheses about HRQoL measures are individually explored, the multiplicity issue does not arise with regard to those individual hypotheses. Conclusions regarding the significance of association for individual HRQoL measures may refer to the nominal alpha level for the statistical test (i.e. a P-value < 0.05 has a 5% probability of being due to chance). However, we observe the usual caution with regard to drawing a single general conclusion about the association between HRQoL and the end-points of interest, when any of the multiple tests of the individual HRQoL measures are used as evidence of this general association between HRQoL and the end-points. The hazard ratio (HR) with associated 95% confidence intervals (CIs) for the HRQoL variable was the key measure and was calculated as the hazard of rPFS or OS per minimally important difference (MID) score change (upper bound; Table A.1) [3,15] in the HRQoL variable. Data were analysed using SAS, version 9.3 (SAS Institute, Cary, NC). Data sets for the analyses correspond to data cut-offs of 16 September 2013 for PREVAIL [11] and 25 September 2011 for AFFIRM [10].
3. Results In AFFIRM, 1199 patients were randomised 2:1 to receive enzalutamide (n Z 800) or placebo (n Z 399); in PREVAIL, 1717 patients were randomised 1:1 to receive enzalutamide (n Z 872) or placebo (n Z 845). Baseline demographic and disease characteristics [12,13] and
23
HRQoL scores [16,17] were comparable between treatment groups in both studies. Reflecting the different trial populations, baseline FACT-P scores were generally higher overall in PREVAIL [17] than in AFFIRM [16]. 3.1. Univariate analyses Univariate analyses for AFFIRM and PREVAIL are summarised in Tables A.2eA.7. In AFFIRM, most demographic and baseline factors and baseline HRQoL scores were prognostic of rPFS and OS. In analyses of change in HRQoL score, all FACT-P scores, except SWB, were prognostic of OS and rPFS. In PREVAIL, baseline visceral disease, bone metastases at screening and type of disease progression at entry were prognostic of rPFS and OS; time since diagnosis, baseline Eastern Cooperative Oncology Group (ECOG) performance score and baseline pain were prognostic of OS only. The majority of baseline HRQoL scores were prognostic of rPFS and OS. In analyses of change in HRQoL score, all FACT-P scores were prognostic of OS and rPFS (except SWB for rPFS). 3.2. Multivariate analyses 3.2.1. AFFIRM 3.2.1.1. Association of baseline HRQoL with rPFS and survival. FACT-P total, FACT-G total, FACT-P TOI
and EWB scores were prognostic of rPFS (HR; 0.92e0.95) [Fig. 1A]. FACT-P total, FACT-G total, FAPSI, TOI, PCS pain-related score and FWB scores were prognostic of OS (Fig. 1B), with higher values in these scores reducing risk of death by 6e17%. Higher baseline SWB showed a negative association with OS (risk of death increased by 11%; HR Z 1.11 [95% CI 1.05e1.19]). 3.2.1.2. Association of changes in HRQoL with rPFS and survival. All FACT-P domains (except FWB, EWB
and PCS) were associated with rPFS (HR; 0.86e1.07) [Fig. 2A]. All domains were prognostic of OS, except EWB (HR; 0.75e1.15) [Fig. 2B]. There was a negative association between improved SWB and rPFS (HR Z 1.07 [95% CI 1.01e1.12]) and OS (HR Z 1.15 [95% CI 1.08e1.23]). Point estimates for HRs were smaller in the timedependent model for all statistically significant scores, revealing a stronger association between changes in HRQoL throughout the study and rPFS and OS, compared with the simpler model including baseline HRQoL score only. 3.2.2. PREVAIL 3.2.2.1. Association of baseline HRQoL with rPFS and survival. Only baseline FACT-P TOI (HR Z 0.94 [95%
CI 0.90e0.99]) and SWB (HR Z 1.06 [95% CI
24
T.M. Beer et al. / European Journal of Cancer 87 (2017) 21e29 A
QoL variable
HR (95% CI) 0.99 1.00 0.92 1.00 1.00 0.98 0.98 0.95 0.95 0.95
Physical well-being score Functional well-being score Emotional well-being Social well-being Prostate cancer subscale score Prostate cancer subscale pain-related score FACT Advanced Prostate Symptom Index (FAPSI) FACT-P Trial Outcome Index FACT-G total score FACT-P total score 0.7
0.8
0.9
1.0
1.1
(0.97-1.01) (0.98-1.02) (0.87-0.97) (0.99-1.02) (0.99-1.02) (0.94-1.03) (0.93-1.02) (0.90-0.99) (0.92-0.99) (0.91-0.99)
1.2
HR (95% CI) B
HR (95% CI)
QoL variable
0.94 0.90 0.98 1.11 0.96 0.92 0.94 0.83 0.90 0.88
Physical well-being score Functional well-being score Emotional well-being Social well-being Prostate cancer subscale score Prostate cancer subscale pain-related score FACT Advanced Prostate Symptom Index (FAPSI) FACT-P Trial Outcome Index FACT-G total score FACT-P total score 0.7
0.8
0.9
1.0
1.1
(0.88-1.02) (0.84-0.96) (0.91-1.05) (1.05-1.19) (0.91-1.01) (0.87-0.97) (0.90-0.97) (0.78-0.88) (0.87-0.95) (0.84-0.93)
1.2
HR (95% CI)
Fig. 1. The association between baseline PROs and rPFS (A) and OS (B) in a multivariate model in the AFFIRM trial. For rPFS, the model included the following baseline covariates: treatment group, age, time since diagnosis, ECOG performance status at baseline, baseline fatigue severity, baseline pain, presence of visceral disease at baseline, bone metastases at screening and type of disease progression at study entry. For OS, the model included the following baseline covariates: treatment group, time since diagnosis, ECOG performance status at baseline, number of prior chemotherapies, baseline fatigue severity, baseline pain, presence of visceral disease at baseline, bone metastases at screening and type of disease progression at study entry. CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; FACT-G, Functional Assessment of Cancer Therapy-General; FACT-P, Functional Assessment of Cancer TherapyProstate; HR, hazard ratio; OS, overall survival; PROs, patient-reported outcomes; QoL, quality of life; rPFS, radiographic progressionfree survival.
1.01e1.10]) were prognostic of rPFS (Fig. 3A). For OS, except for FWB and SWB, all other baseline FACT-P scores were prognostic of OS (HR; 0.82e1.17) [Fig. 3B]. There was a negative association between improved EWB and OS (HR Z 1.17 [95% CI 1.08e1.26]). 3.2.2.2. Association of changes in HRQoL with rPFS and survival. Increases in six domain scores (PWB, PCS
pain-related score, FAPSI, FACT-P TOI, FACT-G and FACT-P total) were associated with improved rPFS (HR; 0.91e0.95) [Fig. 4A]. Increases in all domain scores were associated with reduced risk of death (HR; 0.74e0.90), except for EWB and SWB [Fig. 4B]. A 10point increase in FACT-P total score reduced risk of death by 21% (HR Z 0.79 [95% CI 0.76e0.83]). Similar to AFFIRM, point estimates for HRs were, in general, smaller in the time-dependent model versus the model including baseline values only.
4. Discussion Using trial data for enzalutamide versus placebo in chemotherapy-naı¨ve and post-chemotherapy patients with mCRPC, we explored the association of survival outcomes with both baseline and longitudinally collected HRQoL scores. Several HRQoL domains were significantly associated with rPFS and OS, not only at baseline but also over the course of treatment. Generally, patients with higher baseline scores, or scores improving during treatment, had a reduced risk of death. For example, in the time-dependent model, 10point (upper bound of MID) increases (improvements) in FACT-P total score reduced the risk of death by 19% in AFFIRM and 21% in PREVAIL. While caution must be exercised in moving from the significance of individual HRQoL outcomes to a general conclusion regarding the association between HRQoL and OS,
T.M. Beer et al. / European Journal of Cancer 87 (2017) 21e29 A
25
QoL variable
HR (95% CI) 0.94 0.96 0.95 1.07 0.96 0.89 0.87 0.86 0.90 0.89
Physical well-being score Functional well-being score Emotional well-being Social well-being Prostate cancer subscale score Prostate cancer subscale pain-related score FACT Advanced Prostate Symptom Index (FAPSI) FACT-P Trial Outcome Index FACT-G total score FACT-P total score 0.7
0.8
0.9
1.0
1.1
1.2
(0.88-1.00) (0.91-1.01) (0.89-1.01) (1.01-1.12) (0.92-1.00) (0.86-0.92) (0.84-0.91) (0.82-0.89) (0.87-0.93) (0.86-0.92)
1.3
HR (95% CI) B
HR (95% CI)
QoL variable
0.92 0.86 1.04 1.15 0.91 0.83 0.79 0.75 0.84 0.81
Physical well-being score Functional well-being score Emotional well-being Social well-being Prostate cancer subscale score Prostate cancer subscale pain-related score FACT Advanced Prostate Symptom Index (FAPSI) FACT-P Trial Outcome Index FACT-G total score FACT-P total score 0.7
0.8
0.9
1.0
1.1
1.2
(0.86-0.98) (0.81-0.92) (0.97-1.11) (1.08-1.23) (0.86-0.95) (0.80-0.87) (0.75-0.82) (0.72-0.79) (0.81-0.87) (0.78-0.84)
1.3
HR (95% CI)
Fig. 2. The association between longitudinal PROs and rPFS (A) and OS (B) in a multivariate model in the AFFIRM trial. For rPFS, the model included the following covariates: treatment group, age, time since diagnosis, ECOG performance status at baseline, baseline fatigue severity, baseline pain, presence of visceral disease at baseline, bone metastases at screening and type of disease progression at study entry. For OS, the model included the following covariates: treatment group, time since diagnosis, ECOG performance status at baseline, number of prior chemotherapies, baseline fatigue severity, baseline pain, presence of visceral disease at baseline, bone metastases at screening and type of disease progression at study entry. CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; FACTG, Functional Assessment of Cancer Therapy-General; FACT-P, Functional Assessment of Cancer Therapy-Prostate; HR, hazard ratio; OS, overall survival; PROs, patient-reported outcomes; QoL, quality of life; rPFS, radiographic progression-free survival.
these findings might provide an additional rationale for the incorporation of HRQoL assessment into selected clinical trials because HRQoL outcomes may also be sensitive indicators for survival. The survival benefit of several mCRPC therapies is accompanied by HRQoL benefits [18e21], reduced hospitalisation days [22] or an impact on mCRPCassociated pain [23]. Our findings support previous studies showing that HRQoL scores can be independently prognostic for survival in prostate cancer. An analysis of three phase III trials in CRPC patients showed a significant association between pain and survival [24]. Also, baseline insomnia and appetite loss were identified as HRQoL predictors of survival in mCRPC patients, using the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ-C30) [25]. Predicted survival and time to disease progression were also improved in mCRPC patients with better versus worse baseline HRQoL [8]. In addition, in a mixed prostate cancer population (mostly stage II), patients with better baseline fatigue, or whose
dyspnoea and cognitive functioning improved within 3 months, had a significantly increased probability of survival [7]. Most studies assessed baseline HRQoL. However, since HRQoL scores are likely to change over time, failure to assess trends across numerous time points could underestimate true associations between HRQoL and clinical outcomes [26]. Furthermore, the prognostic value of HRQoL measured over the course of disease could reveal information on complex pathways between physical and psychological disease manifestations [26]. Therefore, we examined the relationship between longitudinal change in HRQoL and survival. Similar to baseline analyses, both trials showed a significant association between several FACT-P domains and rPFS and OS in longitudinal analyses. This timedependent association confirms previous observations in mCRPC patients, where 12-week change scores in EORTC QLQ-C30 and FACT-P were significant predictors of survival [8], and in localised disease, where time-dependent analysis showed that higher levels of physical function, role physical and general health were
26
T.M. Beer et al. / European Journal of Cancer 87 (2017) 21e29 A
QoL variable
HR (95% CI)
Physical well-being score Functional well-being score Emotional well-being Social well-being Prostate cancer subscale score Prostate cancer subscale pain-related score FACT Advanced Prostate Symptom Index (FAPSI) FACT-P Trial Outcome Index FACT-G total score FACT-P total score
1.00 0.96 1.04 1.06 0.96 0.99 0.97 0.94 0.99 0.98 0.7
0.8
0.9
1.0
1.1
1.2
(0.92-1.08) (0.91-1.00) (0.99-1.11) (1.01-1.10) (0.92-1.00) (0.95-1.03) (0.92-1.01) (0.90-0.99) (0.95-1.03) (0.94-1.02)
1.3
HR (95% CI) B
QoL variable
HR (95% CI)
Physical well-being score Functional well-being score Emotional well-being Social well-being Prostate cancer subscale score Prostate cancer subscale pain-related score FACT Advanced Prostate Symptom Index (FAPSI) FACT-P Trial Outcome Index FACT-G total score FACT-P total score
0.85 0.97 1.17 1.04 0.89 0.88 0.88 0.82 0.94 0.90 0.7
0.8
0.9
1.0
1.1
1.2
(0.77-0.93) (0.91-1.03) (1.08-1.26) (0.99-1.10) (0.84-0.94) (0.83-0.94) (0.84-0.91) (0.77-0.87) (0.89-0.98) (0.86-0.95)
1.3
HR (95% CI)
Fig. 3. The association between baseline PROs and rPFS (A) and OS (B) in a multivariate model in the PREVAIL trial. For rPFS, the model included the following baseline covariates: treatment group, bone metastases at screening, presence of visceral disease at baseline and type of disease progression at study entry. For OS, the model included the following baseline covariates: treatment group, bone metastases at screening, ECOG performance status at baseline, pain at baseline, time since diagnosis, presence of visceral disease at baseline and type of disease progression at study entry. CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; FACT-G, Functional Assessment of Cancer Therapy-General; FACT-P, Functional Assessment of Cancer Therapy-Prostate; HR, hazard ratio; OS, overall survival; PROs, patient-reported outcomes; QoL, quality of life; rPFS, radiographic progression-free survival.
significantly associated with better survival [26]. As in other studies [27], point estimates for HRs were, in general, smaller in the time-dependent model, revealing a stronger association between HRQoL (self-rated health) changes throughout the study and survival outcomes, versus the simpler model including baseline HRQoL only. In studies of abiraterone acetate plus prednisone exploring temporal relationships between PRO changes and clinical outcomes in mCRPC, in patients progressing after chemotherapy, improved PROs were associated with increased OS and rPFS [28]; in patients without previous chemotherapy, deteriorating PROs were associated with worse rPFS [29]. We found negative associations between improved SWB and OS (baseline and time-dependent) and rPFS (time-dependent analysis) in AFFIRM and improved EWB and OS (baseline analysis) in PREVAIL. Our analysis does not provide an explanation for these findings. Stronger social relationships have been linked to increased likelihood of survival [30], and higher EWB may also benefit recovery and survival in physically ill patients [31]. However, not all studies show a positive
relationship: Coyne et al. [32] found that EWB was not associated with survival among patients with head and neck cancer, and Brown et al. [33] found that positive mood was associated with shorter survival. A large study including 2733 cancer cases and 1548 deaths also showed no link between personality traits and cancer survival [34]. A systematic review of breast cancer studies found little consistent evidence that psychological coping styles play an important part in cancer survival or recurrence [35]. In light of the exploratory nature of our study that examined multiple relationships, it is also possible that our results represent a statistical artefact, and further study is needed to understand this finding. HRQoL is a multidimensional concept that provides a picture of patient status and well-being beyond clinical characteristics. Identification of prognostic HRQoL factors for survival could contribute to modification of treatment regimens and help identify patient groups for interventions [20]. The prognostic value of HRQoL may reflect patients’ experiences beyond the reach of conventional techniques (e.g. clinical characteristics).
T.M. Beer et al. / European Journal of Cancer 87 (2017) 21e29 A
27
QoL variable
HR (95% CI)
Physical well-being score Functional well-being score Emotional well-being Social well-being Prostate cancer subscale score Prostate cancer subscale pain-related score FACT Advanced Prostate Symptom Index (FAPSI) FACT-P Trial Outcome Index FACT-G total score FACT-P total score
0.91 0.99 0.98 0.99 1.01 0.95 0.93 0.92 0.94 0.93 0.7
0.8
0.9
1.0
1.1
(0.85-0.97) (0.95-1.04) (0.92-1.03) (0.95-1.04) (0.97-1.05) (0.92-0.99) (0.90-0.97) (0.88-0.96) (0.91-0.97) (0.90-0.97)
1.2
HR (95% CI) B
QoL variable
HR (95% CI)
Physical well-being score Functional well-being score Emotional well-being Social well-being Prostate cancer subscale score Prostate cancer subscale pain-related score FACT Advanced Prostate Symptom Index (FAPSI) FACT-P Trial Outcome Index FACT-G total score FACT-P total score
0.88 0.90 1.06 1.04 0.90 0.83 0.77 0.74 0.83 0.79 0.7
0.8
0.9
1.0
1.1
(0.82-0.94) (0.84-0.96) (0.99-1.14) (0.98-1.10) (0.85-0.95) (0.80-0.87) (0.74-0.81) (0.70-0.77) (0.79-0.86) (0.76-0.83)
1.2
HR (95% CI)
Fig. 4. The association between longitudinal PROs and rPFS (A) and OS (B) in a multivariate model in the PREVAIL trial. For rPFS, the model included the following covariates: treatment group, bone metastases at screening, presence of visceral disease at baseline and type of disease progression at study entry. For OS, the model included the following covariates: treatment group, bone metastases at screening, ECOG performance status at baseline, pain at baseline, time since diagnosis, presence of visceral disease at baseline and type of disease progression at study entry. CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; FACT-G, Functional Assessment of Cancer Therapy-General; FACT-P, Functional Assessment of Cancer Therapy-Prostate; HR, hazard ratio; OS, overall survival; PROs, patient-reported outcomes; QoL, quality of life; rPFS, radiographic progression-free survival.
HRQoL reports may act as a marker for otherwiseundetected prognostic factors or as an indicator of the need to pay increased attention to patient symptoms and well-being [26]. In a multivariate analysis of PREVAIL data, the only baseline HRQoL variables that were significant predictors of rPFS were SWB and FACT-P TOI. The lower symptom burden and limited number of progression events may have precluded capture of a sufficiently broad range of patients to detect a relationship between HRQoL and rPFS. The study has several strengths, including the large population analysed, use of a validated instrument, high compliance with the questionnaire completion and the availability of mature and reliable time-to-event data (OS and rPFS). Importantly, we used models with baseline and time-dependent variables; previous studies in this area seem to have focused on baseline/pretreatment HRQoL only [7]. Potential study limitations include the absence of control for multiple comparisons; however, this may be considered acceptable for hypothesis-generating studies [36]. In addition, better
prognostic models may have been possible by making the model specification more sophisticated. However, our goal was to investigate the association between HRQoL domains and survival and disease progression and to compare baseline and change scores in predicting these outcomes. Other limitations include the exploratory, post hoc nature of the analyses, the selected population and the lack of a control for sociocultural effects that might affect HRQoL assessments. In conclusion, our exploratory analysis found that several baseline HRQoL domains were prognostic factors for rPFS and OS in patients with mCRPC. This association was maintained or strengthened over the disease course, indicating that changes in HRQoL are informative for expected survival. Moreover, these results indicate the prognostic value of PROs, in addition to that obtained from performance status, and provide quantitative estimates of these associations. Associations between HRQoL and clinical outcomes may help inform future decision-making and influence clinicians’ treatment approaches in mCRPC. Additional studies are required to confirm these observations and
28
T.M. Beer et al. / European Journal of Cancer 87 (2017) 21e29
demonstrate that therapy selection based on these variables helps patients.
[4]
Funding This study was funded by Astellas Pharma, Inc and Medivation, Inc (Medivation, Inc was acquired by Pfizer, Inc in September 2016), the co-developers of enzalutamide. The study concepts and design were proposed, developed and refined by the authors. Astellas was involved with the design, analysis and interpretation of the studies; Quintiles performed the analyses for this specific work. All authors participated in the writing, review and the decision to submit this manuscript for publication. Conflict of interest statement KS and CI are employees of QuintilesIMS, who received consulting fees from Astellas for the analysis and interpretation of data from the AFFIRM and PREVAIL trials. DP, SH and SN are employees of Astellas. TB received research funding from Astellas, Medivation and Janssen; engaged in consulting with Astellas and Janssen and participated in a certified nursing education programme supported by Astellas and Medivation. KM received fees for consultancy from Astellas, Medivation, Janssen and Bayer. BT was a principal investigator on the PREVAIL trial and an advisor for Astellas. DC reports personal fees from Astellas, outside the submitted work.
Acknowledgements The authors would like to thank Thomas Lavelle of Bioscript Science for assistance with writing and revising the draft manuscript, based on detailed discussion and feedback from all authors, and Stephanie Rippon and Rebecca Cocking of Complete HealthVizion for editorial assistance, all of which was funded by the study sponsors. Appendix A. Supplementary data
[5]
[6]
[7] [8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
Supplementary data related to this article can be found at https://doi.org/10.1016/j.ejca.2017.09.035.
References [1] Osoba D. Health-related quality of life and cancer clinical trials. Ther Adv Med Oncol 2011;3:57e71. [2] Cella D. Quality of life considerations in patients with advanced lung cancer. Semin Oncol 2004;31:16e20. [3] Cella D, Cappelleri JC, Bushmakin A, Charbonneau C, Li JZ, Kim ST, et al. Quality of life predicts progression-free survival in
[18]
[19]
patients with metastatic renal cell carcinoma treated with sunitinib versus interferon alfa. J Oncol Pract 2009;5:66e70. Maione P, Perrone F, Gallo C, Manzione L, Piantedosi F, Barbera S, et al. Pretreatment quality of life and functional status assessment significantly predict survival of elderly patients with advanced non-small-cell lung cancer receiving chemotherapy: a prognostic analysis of the multicenter Italian lung cancer in the elderly study. J Clin Oncol 2005;23:6865e72. Quinten C, Coens C, Mauer M, Comte S, Sprangers MA, Cleeland C, et al. Baseline quality of life as a prognostic indicator of survival: a meta-analysis of individual patient data from EORTC clinical trials. Lancet Oncol 2009;10:865e71. Scher HI, Morris MJ, Stadler WM, Higano C, Basch E, Fizazi K, et al. Trial design and objectives for castration-resistant prostate cancer: updated recommendations from the Prostate Cancer Clinical Trials Working Group 3. J Clin Oncol 2016;34:1402e18. Gupta D, Braun DP, Staren ED. Prognostic value of changes in quality of life scores in prostate cancer. BMC Urol 2013;13:32. Sullivan PW, Nelson JB, Mulani PM, Sleep D. Quality of life as a potential predictor for morbidity and mortality in patients with metastatic hormone-refractory prostate cancer. Qual Life Res 2006;15:1297e306. Nussbaum N, George DJ, Abernethy AP, Dolan CM, Oestreicher N, Flanders S, et al. Patient experience in the treatment of metastatic castration-resistant prostate cancer: state of the science. Prostate Cancer Prostatic Dis 2016;19:111e21. Scher HI, Fizazi K, Saad F, Taplin ME, Sternberg CN, Miller K, et al. Increased survival with enzalutamide in prostate cancer after chemotherapy. N Engl J Med 2012;367:1187e97. Beer TM, Armstrong AJ, Rathkopf DE, Loriot Y, Sternberg CN, Higano CS, et al. Enzalutamide in metastatic prostate cancer before chemotherapy. N Engl J Med 2014;371:424e33. Cella D, Nichol MB, Eton D, Nelson JB, Mulani P. Estimating clinically meaningful changes for the functional assessment of cancer therapy-prostate: results from a clinical trial of patients with metastatic hormone-refractory prostate cancer. Value Health 2009;12:124e9. Esper P, Mo F, Chodak G, Sinner M, Cella D, Pienta KJ. Measuring quality of life in men with prostate cancer using the functional assessment of cancer therapy-prostate instrument. Urology 1997;50:920e8. Victorson DE, Beaumont JL, Rosenbloom SK, Shevrin D, Cella D. Efficient assessment of the most important symptoms in advanced prostate cancer: the NCCN/FACT-P symptom index. Physco-oncology 2011;20:977e83. Yost KJ, Eton DT. Combining distribution- and anchor-based approaches to determine minimally important differences: the FACIT experience. Eval Health Prof 2005;28:172e91. Cella D, Ivanescu C, Holmstrom S, Bui CN, Spalding J, Fizazi K. Impact of enzalutamide on quality of life in men with metastatic castration-resistant prostate cancer after chemotherapy: additional analyses from the AFFIRM randomized clinical trial. Ann Oncol 2015;26:179e85. Loriot Y, Miller K, Sternberg CN, Fizazi K, de Bono JS, Chowdhury S, et al. Effect of enzalutamide on health-related quality of life, pain, and skeletal-related events in asymptomatic and minimally symptomatic, chemotherapy-naive patients with metastatic castration-resistant prostate cancer (PREVAIL): results from a randomised, phase 3 trial. Lancet Oncol 2015;16: 509e21. Basch E, Autio K, Ryan CJ, Mulders P, Shore N, Kheoh T, et al. Abiraterone acetate plus prednisone versus prednisone alone in chemotherapy-naive men with metastatic castration-resistant prostate cancer: patient-reported outcome results of a randomised phase 3 trial. Lancet Oncol 2013;14:1193e9. Fizazi K, Scher HI, Miller K, Basch E, Sternberg CN, Cella D, et al. Effect of enzalutamide on time to first skeletal-related event, pain, and quality of life in men with castration-resistant prostate
T.M. Beer et al. / European Journal of Cancer 87 (2017) 21e29
[20]
[21]
[22]
[23]
[24]
[25]
[26]
cancer: results from the randomised, phase 3 AFFIRM trial. Lancet Oncol 2014;15:1147e56. Harland S, Staffurth J, Molina A, Hao Y, Gagnon DD, Sternberg CN, et al. Effect of abiraterone acetate treatment on the quality of life of patients with metastatic castration-resistant prostate cancer after failure of docetaxel chemotherapy. Eur J Cancer 2013;49:3648e57. Nilsson S, Cislo P, Sartor O, Vogelzang NJ, Coleman RE, O’Sullivan JM, et al. Patient-reported quality-of-life analysis of radium-223 dichloride from the phase III ALSYMPCA study. Ann Oncol 2016;27:868e74. Parker C, Zhan L, Cislo P, Reuning-Scherer J, Vogelzang NJ, Nilsson S, et al. Effect of radium-223 dichloride (Ra-223) on hospitalisation: an analysis from the phase 3 randomised Alpharadin in Symptomatic Prostate Cancer Patients (ALSYMPCA) trial. Eur J Cancer 2017;71:1e6. ¨ zgu¨roglu M, Hansen S, Kocak I, Bahl A, Oudard S, Tombal B, O et al. Impact of cabazitaxel on 2-year survival and palliation of tumour-related pain in men with metastatic castration-resistant prostate cancer treated in the TROPIC trial. Ann Oncol 2013; 24:2402e8. Halabi S, Vogelzang NJ, Kornblith AB, Ou SS, Kantoff PW, Dawson NA, et al. Pain predicts overall survival in men with metastatic castration-refractory prostate cancer. J Clin Oncol 2008;26:2544e9. Collette L, van Andel G, Bottomley A, Oosterhof GO, Albrecht W, de Reijke TM, et al. Is baseline quality of life useful for predicting survival with hormone-refractory prostate cancer? A pooled analysis of three studies of the European Organisation for Research and Treatment of Cancer Genitourinary Group. J Clin Oncol 2004;22:3877e85. Sadetsky N, Hubbard A, Carroll PR, Satariano W. Predictive value of serial measurements of quality of life on all-cause mortality in prostate cancer patients: data from CaPSURE (cancer of the prostate strategic urologic research endeavor) database. Qual Life Res 2009;18:1019e27.
29
[27] Lyyra TM, Leskinen E, Jylha¨ M, Heikkinen E. Self-rated health and mortality in older men and women: a time-dependent covariate analysis. Arch Gerontol Geriatr 2009;48:14e8. [28] Traina S, Li T, Johnson K, Ho KF, Molina A, Cella D. Is there a relationship between patient-reported outcomes (Pros) and clinical outcomes in metastatic castration-resistant prostate cancer (Mcrpc) post-docetaxel? Value Health 2015;18:A470. [29] Traina S, Li T, Johnson K, Ho KF, Molina A, Cella D. Analysis of the relationship between patient-reported outcomes (Pros) and clinical outcomes in metastatic castration-resistant prostate cancer (Mcrpc) patients without prior chemotherapy. Value Health 2015;18:A335e6. [30] Holt-Lunstad J, Smith TB, Layton JB. Social relationships and mortality risk: a meta-analytic review. PLoS Med 2010;7: e1000316. [31] Lamers SM, Bolier L, Westerhof GJ, Smit F, Bohlmeijer ET. The impact of emotional well-being on long-term recovery and survival in physical illness: a meta-analysis. J Behav Med 2012;35:538e47. [32] Coyne JC, Pajak TF, Harris J, Konski A, Movsas B, Ang K, et al. Emotional well-being does not predict survival in head and neck cancer patients: a Radiation Therapy Oncology Group study. Cancer 2007;110:2568e75. [33] Brown JE, Butow PN, Culjak G, Coates AS, Dunn SM. Psychosocial predictors of outcome: time to relapse and survival in patients with early stage melanoma. Br J Cancer 2000;83:1448e53. [34] Nakaya N, Bidstrup PE, Saito-Nakaya K, Frederiksen K, Koskenvuo M, Pukkala E, et al. Personality traits and cancer risk and survival based on Finnish and Swedish registry data. Am J Epidemiol 2010;172:377e85. [35] Petticrew M, Bell R, Hunter D. Influence of psychological coping on survival and recurrence in people with cancer: systematic review. BMJ 2002;325:1066. [36] Fayers PM, Machin D. Exploring longitudinal data. Quality of life: the assessment, analysis and interpretation of patientreported outcomes. 2nd ed. John Wiley & Sons, Ltd; 2007. p. 305e27.
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.ejcancer.com
Original Research
The association between health-related quality-of-life scores and clinical outcomes in metastatic castration-resistant prostate cancer patients: Exploratory analyses of AFFIRM and PREVAIL studies Tomasz M. Beer a,*, Kurt Miller b, Bertrand Tombal c, David Cella d, De Phung e, Stefan Holmstrom e, Cristina Ivanescu f, Konstantina Skaltsa g, Shevani Naidoo h a
OHSU Knight Cancer Institute, Oregon Health & Science University, 3303 S.W. Bond Avenue, CH14R, Portland, OR 97239, USA b Department of Urology, Charite´ e Universita¨tsmedizin Berlin, Charite´platz 1, 10117 Berlin, Germany c Service D’Urologie, Cliniques Universitaires Saint-Luc, Avenue Hippocrate 10, B-1200 Brussels, Belgium d Feinberg School of Medicine and Robert H. Lurie Comprehensive Cancer Center, Northwestern University, 633 N. St. Clair St., 19th Floor, Chicago, IL 60611, USA e Astellas Pharma Inc, Sylviusweg 62, 2300 AH Leiden, The Netherlands f QuintilesIMS, Siriusdreef 10, 2132 WT Hoofddorp, The Netherlands g QuintilesIMS, Sardenya 537, 08024 Barcelona, Spain h Astellas Pharma Inc, Global Health Economic Outcomes Research (HEOR), 2000 Hillswood Dr., Chertsey, Surrey KT16 0RS, UK Received 30 August 2017; accepted 25 September 2017
KEYWORDS Castration-resistant prostate cancer; Enzalutamide; Health-related quality of life; Survival
Abstract Background: Our exploratory analysis examined the association between healthrelated quality of life (HRQoL) (baseline and change over time) and clinical outcomes (overall survival [OS]/radiographic progression-free survival [rPFS]) in metastatic castration-resistant prostate cancer (mCRPC). Methods: HRQoL, OS and rPFS were assessed in phase III trials comparing enzalutamide with placebo in chemotherapy-naı¨ve (PREVAIL; NCT01212991) or post-chemotherapy (AFFIRM; NCT00974311) mCRPC. HRQoL was assessed using the Functional Assessment of Cancer Therapy-Prostate (FACT-P). Multivariate analyses evaluated the prognostic significance of baseline and time-dependent scores after adjusting for treatment and clinical/
* Corresponding author. E-mail addresses: [email protected] (T.M. Beer), [email protected] (K. Miller), [email protected] (B. Tombal), d-cella@ northwestern.edu (D. Cella), [email protected] (D. Phung), [email protected] (S. Holmstrom), cristina.ivanescu@ quintilesims.com (C. Ivanescu), [email protected] (K. Skaltsa), [email protected] (S. Naidoo). https://doi.org/10.1016/j.ejca.2017.09.035 0959-8049/ª 2017 Elsevier Ltd. All rights reserved.
22
T.M. Beer et al. / European Journal of Cancer 87 (2017) 21e29
demographic variables. Hazard ratios (HRs) and 95% confidence intervals (CIs) represented the hazard of rPFS or OS per minimally important difference (MID) score change in HRQoL variables. Results: In baseline and time-dependent multivariate analyses, OS was independently associated with multiple HRQoL measures across both studies. In time-dependent analyses, a 10point (upper bound of MID range) increase (improvement) in FACT-P total score was associated with reductions in mortality risk of 19% in AFFIRM (HR 0.81 [95% CI 0.78e0.84]) and 21% in PREVAIL (HR 0.79 [0.76e0.83]). For baseline analyses, a 10-point increase in FACTP total score was associated with reductions in mortality risk of 12% (HR 0.88 [0.84e0.93]) and 10% (HR 0.90 [0.86e0.95]) in AFFIRM and PREVAIL, respectively. rPFS was associated with a subset of HRQoL domains in both studies. Conclusion: Several baseline HRQoL domains were prognostic for rPFS and OS in patients with mCRPC, and this association was maintained during treatment, indicating that changes in HRQoL are informative for patients’ expected survival. ª 2017 Elsevier Ltd. All rights reserved.
1. Introduction Health-related quality of life (HRQoL) is an important end-point in oncology studies [1]. Patient-reported outcomes (PROs) can be associated with overall survival (OS) or progression-free survival (PFS) in various cancers [2e5]. Thus, the patient’s perspective on changes in HRQoL, and more specifically PROs, is an important determinant of the value of cancer treatment. The Prostate Cancer Clinical Trials Working Group 3 acknowledged the need to optimise assessment and analysis of PRO data and recommended evaluating the association between early changes in individual outcome measures (e.g. PROs) and later events such as radiographic PFS (rPFS) or OS [6]. In metastatic castrationresistant prostate cancer (mCRPC), improvement in HRQoL has been associated with improved clinical outcomes [7,8]. In addition, PRO data from 19 mCRPC clinical trials indicated that HRQoL and pain status augmented clinical efficacy data by providing better understanding of treatment impact [9]. However, information on relationships between commonly used efficacy outcome measures, such as rPFS or OS, and HRQoL in mCRPC is limited. The AFFIRM (ClinicalTrials.gov, NCT00974311) [10] and PREVAIL (ClinicalTrials.gov, NCT01212991) [11] phase III trials showed that, compared with placebo, enzalutamide significantly prolonged OS and rPFS and displayed HRQoL benefits in men with mCRPC in postchemotherapy and chemotherapy-naı¨ve settings. These exploratory analyses investigate the association between HRQoL and rPFS and OS in the AFFIRM and PREVAIL studies at baseline and during treatment. 2. Methods 2.1. Study populations and design PREVAIL and AFFIRM were phase III, randomised, double-blind, placebo-controlled trials comparing oral
enzalutamide 160 mg/day with placebo in postchemotherapy (AFFIRM) or chemotherapy-naı¨ve (PREVAIL) mCRPC patients. The trials’ design, patient populations and results have been described previously [10,11]. Both trials were approved by local independent review boards and conducted according to the Declaration of Helsinki and Good Clinical Practice guidelines. All patients provided written informed consent. 2.2. Health-related quality of life In both studies, HRQoL was assessed using the Functional Assessment of Cancer Therapy-Prostate (FACTP), version 4, a 39-item instrument validated for use in mCRPC [12e14]. HRQoL data were collected at baseline (i.e. treatment day 1), at week 5 (PREVAIL only), at week 13 and every subsequent 12 weeks until study drug discontinuation. 2.3. Statistical analyses In this post hoc exploratory analysis, we investigated the relationship between rPFS and OS and the following HRQoL measures: prostate cancerespecific subscale (PCS), PCS pain-related score, physical well-being (PWB), emotional well-being (EWB), social/family wellbeing (SWB), functional well-being (FWB), FACT-P total score, FACT-General (FACT-G) total score (the sum of PWB, FWB, SWB and EWB), FACT Advanced Prostate Symptom Index (FAPSI) and trial outcome index (TOI) (the sum of PWB, FWB and PCS). The analyses were performed on intent-to-treat patient populations (i.e. all patients who were randomised into the studies). Cox proportional hazard models with baseline (i.e. fixed) or time-dependent covariates were fitted to timeto-event data (OS and, separately, rPFS). OS was defined as the time from randomisation to death from
T.M. Beer et al. / European Journal of Cancer 87 (2017) 21e29
any cause. rPFS was defined as the time from randomisation to first objective evidence of radiographic progression or death due to any cause. Demographic, clinical and HRQoL variables were evaluated in a univariate analysis to determine which parameters showed individual prognostic value for rPFS or OS. Note that here, P-values were not used to reject a null hypothesis but served only as a criterion for inclusion in multivariate analyses, so multiplicity adjustment was not applied. In multivariate analyses, joint prognostic significance of selected demographic and clinical variables (using a significance level of 0.05), and all baseline HRQoL or time-dependent HRQoL variables, were investigated. To minimise instability of the final multivariate model potentially resulting from high multicollinearity, separate models were fit for each of the following domains: FACT-P total score, FACT-G total score, FACT-P TOI, FAPSI and FACT-P PCS pain-related score. The remaining sub domains (PWB, FWB, SWB, EWB and PCS) were introduced in a separate single model and no other domains were included in that model. The complete list of covariates included in each model is listed in Appendix A. Since a single hypothesis is not being tested, but rather a number of hypotheses about HRQoL measures are individually explored, the multiplicity issue does not arise with regard to those individual hypotheses. Conclusions regarding the significance of association for individual HRQoL measures may refer to the nominal alpha level for the statistical test (i.e. a P-value < 0.05 has a 5% probability of being due to chance). However, we observe the usual caution with regard to drawing a single general conclusion about the association between HRQoL and the end-points of interest, when any of the multiple tests of the individual HRQoL measures are used as evidence of this general association between HRQoL and the end-points. The hazard ratio (HR) with associated 95% confidence intervals (CIs) for the HRQoL variable was the key measure and was calculated as the hazard of rPFS or OS per minimally important difference (MID) score change (upper bound; Table A.1) [3,15] in the HRQoL variable. Data were analysed using SAS, version 9.3 (SAS Institute, Cary, NC). Data sets for the analyses correspond to data cut-offs of 16 September 2013 for PREVAIL [11] and 25 September 2011 for AFFIRM [10].
3. Results In AFFIRM, 1199 patients were randomised 2:1 to receive enzalutamide (n Z 800) or placebo (n Z 399); in PREVAIL, 1717 patients were randomised 1:1 to receive enzalutamide (n Z 872) or placebo (n Z 845). Baseline demographic and disease characteristics [12,13] and
23
HRQoL scores [16,17] were comparable between treatment groups in both studies. Reflecting the different trial populations, baseline FACT-P scores were generally higher overall in PREVAIL [17] than in AFFIRM [16]. 3.1. Univariate analyses Univariate analyses for AFFIRM and PREVAIL are summarised in Tables A.2eA.7. In AFFIRM, most demographic and baseline factors and baseline HRQoL scores were prognostic of rPFS and OS. In analyses of change in HRQoL score, all FACT-P scores, except SWB, were prognostic of OS and rPFS. In PREVAIL, baseline visceral disease, bone metastases at screening and type of disease progression at entry were prognostic of rPFS and OS; time since diagnosis, baseline Eastern Cooperative Oncology Group (ECOG) performance score and baseline pain were prognostic of OS only. The majority of baseline HRQoL scores were prognostic of rPFS and OS. In analyses of change in HRQoL score, all FACT-P scores were prognostic of OS and rPFS (except SWB for rPFS). 3.2. Multivariate analyses 3.2.1. AFFIRM 3.2.1.1. Association of baseline HRQoL with rPFS and survival. FACT-P total, FACT-G total, FACT-P TOI
and EWB scores were prognostic of rPFS (HR; 0.92e0.95) [Fig. 1A]. FACT-P total, FACT-G total, FAPSI, TOI, PCS pain-related score and FWB scores were prognostic of OS (Fig. 1B), with higher values in these scores reducing risk of death by 6e17%. Higher baseline SWB showed a negative association with OS (risk of death increased by 11%; HR Z 1.11 [95% CI 1.05e1.19]). 3.2.1.2. Association of changes in HRQoL with rPFS and survival. All FACT-P domains (except FWB, EWB
and PCS) were associated with rPFS (HR; 0.86e1.07) [Fig. 2A]. All domains were prognostic of OS, except EWB (HR; 0.75e1.15) [Fig. 2B]. There was a negative association between improved SWB and rPFS (HR Z 1.07 [95% CI 1.01e1.12]) and OS (HR Z 1.15 [95% CI 1.08e1.23]). Point estimates for HRs were smaller in the timedependent model for all statistically significant scores, revealing a stronger association between changes in HRQoL throughout the study and rPFS and OS, compared with the simpler model including baseline HRQoL score only. 3.2.2. PREVAIL 3.2.2.1. Association of baseline HRQoL with rPFS and survival. Only baseline FACT-P TOI (HR Z 0.94 [95%
CI 0.90e0.99]) and SWB (HR Z 1.06 [95% CI
24
T.M. Beer et al. / European Journal of Cancer 87 (2017) 21e29 A
QoL variable
HR (95% CI) 0.99 1.00 0.92 1.00 1.00 0.98 0.98 0.95 0.95 0.95
Physical well-being score Functional well-being score Emotional well-being Social well-being Prostate cancer subscale score Prostate cancer subscale pain-related score FACT Advanced Prostate Symptom Index (FAPSI) FACT-P Trial Outcome Index FACT-G total score FACT-P total score 0.7
0.8
0.9
1.0
1.1
(0.97-1.01) (0.98-1.02) (0.87-0.97) (0.99-1.02) (0.99-1.02) (0.94-1.03) (0.93-1.02) (0.90-0.99) (0.92-0.99) (0.91-0.99)
1.2
HR (95% CI) B
HR (95% CI)
QoL variable
0.94 0.90 0.98 1.11 0.96 0.92 0.94 0.83 0.90 0.88
Physical well-being score Functional well-being score Emotional well-being Social well-being Prostate cancer subscale score Prostate cancer subscale pain-related score FACT Advanced Prostate Symptom Index (FAPSI) FACT-P Trial Outcome Index FACT-G total score FACT-P total score 0.7
0.8
0.9
1.0
1.1
(0.88-1.02) (0.84-0.96) (0.91-1.05) (1.05-1.19) (0.91-1.01) (0.87-0.97) (0.90-0.97) (0.78-0.88) (0.87-0.95) (0.84-0.93)
1.2
HR (95% CI)
Fig. 1. The association between baseline PROs and rPFS (A) and OS (B) in a multivariate model in the AFFIRM trial. For rPFS, the model included the following baseline covariates: treatment group, age, time since diagnosis, ECOG performance status at baseline, baseline fatigue severity, baseline pain, presence of visceral disease at baseline, bone metastases at screening and type of disease progression at study entry. For OS, the model included the following baseline covariates: treatment group, time since diagnosis, ECOG performance status at baseline, number of prior chemotherapies, baseline fatigue severity, baseline pain, presence of visceral disease at baseline, bone metastases at screening and type of disease progression at study entry. CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; FACT-G, Functional Assessment of Cancer Therapy-General; FACT-P, Functional Assessment of Cancer TherapyProstate; HR, hazard ratio; OS, overall survival; PROs, patient-reported outcomes; QoL, quality of life; rPFS, radiographic progressionfree survival.
1.01e1.10]) were prognostic of rPFS (Fig. 3A). For OS, except for FWB and SWB, all other baseline FACT-P scores were prognostic of OS (HR; 0.82e1.17) [Fig. 3B]. There was a negative association between improved EWB and OS (HR Z 1.17 [95% CI 1.08e1.26]). 3.2.2.2. Association of changes in HRQoL with rPFS and survival. Increases in six domain scores (PWB, PCS
pain-related score, FAPSI, FACT-P TOI, FACT-G and FACT-P total) were associated with improved rPFS (HR; 0.91e0.95) [Fig. 4A]. Increases in all domain scores were associated with reduced risk of death (HR; 0.74e0.90), except for EWB and SWB [Fig. 4B]. A 10point increase in FACT-P total score reduced risk of death by 21% (HR Z 0.79 [95% CI 0.76e0.83]). Similar to AFFIRM, point estimates for HRs were, in general, smaller in the time-dependent model versus the model including baseline values only.
4. Discussion Using trial data for enzalutamide versus placebo in chemotherapy-naı¨ve and post-chemotherapy patients with mCRPC, we explored the association of survival outcomes with both baseline and longitudinally collected HRQoL scores. Several HRQoL domains were significantly associated with rPFS and OS, not only at baseline but also over the course of treatment. Generally, patients with higher baseline scores, or scores improving during treatment, had a reduced risk of death. For example, in the time-dependent model, 10point (upper bound of MID) increases (improvements) in FACT-P total score reduced the risk of death by 19% in AFFIRM and 21% in PREVAIL. While caution must be exercised in moving from the significance of individual HRQoL outcomes to a general conclusion regarding the association between HRQoL and OS,
T.M. Beer et al. / European Journal of Cancer 87 (2017) 21e29 A
25
QoL variable
HR (95% CI) 0.94 0.96 0.95 1.07 0.96 0.89 0.87 0.86 0.90 0.89
Physical well-being score Functional well-being score Emotional well-being Social well-being Prostate cancer subscale score Prostate cancer subscale pain-related score FACT Advanced Prostate Symptom Index (FAPSI) FACT-P Trial Outcome Index FACT-G total score FACT-P total score 0.7
0.8
0.9
1.0
1.1
1.2
(0.88-1.00) (0.91-1.01) (0.89-1.01) (1.01-1.12) (0.92-1.00) (0.86-0.92) (0.84-0.91) (0.82-0.89) (0.87-0.93) (0.86-0.92)
1.3
HR (95% CI) B
HR (95% CI)
QoL variable
0.92 0.86 1.04 1.15 0.91 0.83 0.79 0.75 0.84 0.81
Physical well-being score Functional well-being score Emotional well-being Social well-being Prostate cancer subscale score Prostate cancer subscale pain-related score FACT Advanced Prostate Symptom Index (FAPSI) FACT-P Trial Outcome Index FACT-G total score FACT-P total score 0.7
0.8
0.9
1.0
1.1
1.2
(0.86-0.98) (0.81-0.92) (0.97-1.11) (1.08-1.23) (0.86-0.95) (0.80-0.87) (0.75-0.82) (0.72-0.79) (0.81-0.87) (0.78-0.84)
1.3
HR (95% CI)
Fig. 2. The association between longitudinal PROs and rPFS (A) and OS (B) in a multivariate model in the AFFIRM trial. For rPFS, the model included the following covariates: treatment group, age, time since diagnosis, ECOG performance status at baseline, baseline fatigue severity, baseline pain, presence of visceral disease at baseline, bone metastases at screening and type of disease progression at study entry. For OS, the model included the following covariates: treatment group, time since diagnosis, ECOG performance status at baseline, number of prior chemotherapies, baseline fatigue severity, baseline pain, presence of visceral disease at baseline, bone metastases at screening and type of disease progression at study entry. CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; FACTG, Functional Assessment of Cancer Therapy-General; FACT-P, Functional Assessment of Cancer Therapy-Prostate; HR, hazard ratio; OS, overall survival; PROs, patient-reported outcomes; QoL, quality of life; rPFS, radiographic progression-free survival.
these findings might provide an additional rationale for the incorporation of HRQoL assessment into selected clinical trials because HRQoL outcomes may also be sensitive indicators for survival. The survival benefit of several mCRPC therapies is accompanied by HRQoL benefits [18e21], reduced hospitalisation days [22] or an impact on mCRPCassociated pain [23]. Our findings support previous studies showing that HRQoL scores can be independently prognostic for survival in prostate cancer. An analysis of three phase III trials in CRPC patients showed a significant association between pain and survival [24]. Also, baseline insomnia and appetite loss were identified as HRQoL predictors of survival in mCRPC patients, using the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ-C30) [25]. Predicted survival and time to disease progression were also improved in mCRPC patients with better versus worse baseline HRQoL [8]. In addition, in a mixed prostate cancer population (mostly stage II), patients with better baseline fatigue, or whose
dyspnoea and cognitive functioning improved within 3 months, had a significantly increased probability of survival [7]. Most studies assessed baseline HRQoL. However, since HRQoL scores are likely to change over time, failure to assess trends across numerous time points could underestimate true associations between HRQoL and clinical outcomes [26]. Furthermore, the prognostic value of HRQoL measured over the course of disease could reveal information on complex pathways between physical and psychological disease manifestations [26]. Therefore, we examined the relationship between longitudinal change in HRQoL and survival. Similar to baseline analyses, both trials showed a significant association between several FACT-P domains and rPFS and OS in longitudinal analyses. This timedependent association confirms previous observations in mCRPC patients, where 12-week change scores in EORTC QLQ-C30 and FACT-P were significant predictors of survival [8], and in localised disease, where time-dependent analysis showed that higher levels of physical function, role physical and general health were
26
T.M. Beer et al. / European Journal of Cancer 87 (2017) 21e29 A
QoL variable
HR (95% CI)
Physical well-being score Functional well-being score Emotional well-being Social well-being Prostate cancer subscale score Prostate cancer subscale pain-related score FACT Advanced Prostate Symptom Index (FAPSI) FACT-P Trial Outcome Index FACT-G total score FACT-P total score
1.00 0.96 1.04 1.06 0.96 0.99 0.97 0.94 0.99 0.98 0.7
0.8
0.9
1.0
1.1
1.2
(0.92-1.08) (0.91-1.00) (0.99-1.11) (1.01-1.10) (0.92-1.00) (0.95-1.03) (0.92-1.01) (0.90-0.99) (0.95-1.03) (0.94-1.02)
1.3
HR (95% CI) B
QoL variable
HR (95% CI)
Physical well-being score Functional well-being score Emotional well-being Social well-being Prostate cancer subscale score Prostate cancer subscale pain-related score FACT Advanced Prostate Symptom Index (FAPSI) FACT-P Trial Outcome Index FACT-G total score FACT-P total score
0.85 0.97 1.17 1.04 0.89 0.88 0.88 0.82 0.94 0.90 0.7
0.8
0.9
1.0
1.1
1.2
(0.77-0.93) (0.91-1.03) (1.08-1.26) (0.99-1.10) (0.84-0.94) (0.83-0.94) (0.84-0.91) (0.77-0.87) (0.89-0.98) (0.86-0.95)
1.3
HR (95% CI)
Fig. 3. The association between baseline PROs and rPFS (A) and OS (B) in a multivariate model in the PREVAIL trial. For rPFS, the model included the following baseline covariates: treatment group, bone metastases at screening, presence of visceral disease at baseline and type of disease progression at study entry. For OS, the model included the following baseline covariates: treatment group, bone metastases at screening, ECOG performance status at baseline, pain at baseline, time since diagnosis, presence of visceral disease at baseline and type of disease progression at study entry. CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; FACT-G, Functional Assessment of Cancer Therapy-General; FACT-P, Functional Assessment of Cancer Therapy-Prostate; HR, hazard ratio; OS, overall survival; PROs, patient-reported outcomes; QoL, quality of life; rPFS, radiographic progression-free survival.
significantly associated with better survival [26]. As in other studies [27], point estimates for HRs were, in general, smaller in the time-dependent model, revealing a stronger association between HRQoL (self-rated health) changes throughout the study and survival outcomes, versus the simpler model including baseline HRQoL only. In studies of abiraterone acetate plus prednisone exploring temporal relationships between PRO changes and clinical outcomes in mCRPC, in patients progressing after chemotherapy, improved PROs were associated with increased OS and rPFS [28]; in patients without previous chemotherapy, deteriorating PROs were associated with worse rPFS [29]. We found negative associations between improved SWB and OS (baseline and time-dependent) and rPFS (time-dependent analysis) in AFFIRM and improved EWB and OS (baseline analysis) in PREVAIL. Our analysis does not provide an explanation for these findings. Stronger social relationships have been linked to increased likelihood of survival [30], and higher EWB may also benefit recovery and survival in physically ill patients [31]. However, not all studies show a positive
relationship: Coyne et al. [32] found that EWB was not associated with survival among patients with head and neck cancer, and Brown et al. [33] found that positive mood was associated with shorter survival. A large study including 2733 cancer cases and 1548 deaths also showed no link between personality traits and cancer survival [34]. A systematic review of breast cancer studies found little consistent evidence that psychological coping styles play an important part in cancer survival or recurrence [35]. In light of the exploratory nature of our study that examined multiple relationships, it is also possible that our results represent a statistical artefact, and further study is needed to understand this finding. HRQoL is a multidimensional concept that provides a picture of patient status and well-being beyond clinical characteristics. Identification of prognostic HRQoL factors for survival could contribute to modification of treatment regimens and help identify patient groups for interventions [20]. The prognostic value of HRQoL may reflect patients’ experiences beyond the reach of conventional techniques (e.g. clinical characteristics).
T.M. Beer et al. / European Journal of Cancer 87 (2017) 21e29 A
27
QoL variable
HR (95% CI)
Physical well-being score Functional well-being score Emotional well-being Social well-being Prostate cancer subscale score Prostate cancer subscale pain-related score FACT Advanced Prostate Symptom Index (FAPSI) FACT-P Trial Outcome Index FACT-G total score FACT-P total score
0.91 0.99 0.98 0.99 1.01 0.95 0.93 0.92 0.94 0.93 0.7
0.8
0.9
1.0
1.1
(0.85-0.97) (0.95-1.04) (0.92-1.03) (0.95-1.04) (0.97-1.05) (0.92-0.99) (0.90-0.97) (0.88-0.96) (0.91-0.97) (0.90-0.97)
1.2
HR (95% CI) B
QoL variable
HR (95% CI)
Physical well-being score Functional well-being score Emotional well-being Social well-being Prostate cancer subscale score Prostate cancer subscale pain-related score FACT Advanced Prostate Symptom Index (FAPSI) FACT-P Trial Outcome Index FACT-G total score FACT-P total score
0.88 0.90 1.06 1.04 0.90 0.83 0.77 0.74 0.83 0.79 0.7
0.8
0.9
1.0
1.1
(0.82-0.94) (0.84-0.96) (0.99-1.14) (0.98-1.10) (0.85-0.95) (0.80-0.87) (0.74-0.81) (0.70-0.77) (0.79-0.86) (0.76-0.83)
1.2
HR (95% CI)
Fig. 4. The association between longitudinal PROs and rPFS (A) and OS (B) in a multivariate model in the PREVAIL trial. For rPFS, the model included the following covariates: treatment group, bone metastases at screening, presence of visceral disease at baseline and type of disease progression at study entry. For OS, the model included the following covariates: treatment group, bone metastases at screening, ECOG performance status at baseline, pain at baseline, time since diagnosis, presence of visceral disease at baseline and type of disease progression at study entry. CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; FACT-G, Functional Assessment of Cancer Therapy-General; FACT-P, Functional Assessment of Cancer Therapy-Prostate; HR, hazard ratio; OS, overall survival; PROs, patient-reported outcomes; QoL, quality of life; rPFS, radiographic progression-free survival.
HRQoL reports may act as a marker for otherwiseundetected prognostic factors or as an indicator of the need to pay increased attention to patient symptoms and well-being [26]. In a multivariate analysis of PREVAIL data, the only baseline HRQoL variables that were significant predictors of rPFS were SWB and FACT-P TOI. The lower symptom burden and limited number of progression events may have precluded capture of a sufficiently broad range of patients to detect a relationship between HRQoL and rPFS. The study has several strengths, including the large population analysed, use of a validated instrument, high compliance with the questionnaire completion and the availability of mature and reliable time-to-event data (OS and rPFS). Importantly, we used models with baseline and time-dependent variables; previous studies in this area seem to have focused on baseline/pretreatment HRQoL only [7]. Potential study limitations include the absence of control for multiple comparisons; however, this may be considered acceptable for hypothesis-generating studies [36]. In addition, better
prognostic models may have been possible by making the model specification more sophisticated. However, our goal was to investigate the association between HRQoL domains and survival and disease progression and to compare baseline and change scores in predicting these outcomes. Other limitations include the exploratory, post hoc nature of the analyses, the selected population and the lack of a control for sociocultural effects that might affect HRQoL assessments. In conclusion, our exploratory analysis found that several baseline HRQoL domains were prognostic factors for rPFS and OS in patients with mCRPC. This association was maintained or strengthened over the disease course, indicating that changes in HRQoL are informative for expected survival. Moreover, these results indicate the prognostic value of PROs, in addition to that obtained from performance status, and provide quantitative estimates of these associations. Associations between HRQoL and clinical outcomes may help inform future decision-making and influence clinicians’ treatment approaches in mCRPC. Additional studies are required to confirm these observations and
28
T.M. Beer et al. / European Journal of Cancer 87 (2017) 21e29
demonstrate that therapy selection based on these variables helps patients.
[4]
Funding This study was funded by Astellas Pharma, Inc and Medivation, Inc (Medivation, Inc was acquired by Pfizer, Inc in September 2016), the co-developers of enzalutamide. The study concepts and design were proposed, developed and refined by the authors. Astellas was involved with the design, analysis and interpretation of the studies; Quintiles performed the analyses for this specific work. All authors participated in the writing, review and the decision to submit this manuscript for publication. Conflict of interest statement KS and CI are employees of QuintilesIMS, who received consulting fees from Astellas for the analysis and interpretation of data from the AFFIRM and PREVAIL trials. DP, SH and SN are employees of Astellas. TB received research funding from Astellas, Medivation and Janssen; engaged in consulting with Astellas and Janssen and participated in a certified nursing education programme supported by Astellas and Medivation. KM received fees for consultancy from Astellas, Medivation, Janssen and Bayer. BT was a principal investigator on the PREVAIL trial and an advisor for Astellas. DC reports personal fees from Astellas, outside the submitted work.
Acknowledgements The authors would like to thank Thomas Lavelle of Bioscript Science for assistance with writing and revising the draft manuscript, based on detailed discussion and feedback from all authors, and Stephanie Rippon and Rebecca Cocking of Complete HealthVizion for editorial assistance, all of which was funded by the study sponsors. Appendix A. Supplementary data
[5]
[6]
[7] [8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
Supplementary data related to this article can be found at https://doi.org/10.1016/j.ejca.2017.09.035.
References [1] Osoba D. Health-related quality of life and cancer clinical trials. Ther Adv Med Oncol 2011;3:57e71. [2] Cella D. Quality of life considerations in patients with advanced lung cancer. Semin Oncol 2004;31:16e20. [3] Cella D, Cappelleri JC, Bushmakin A, Charbonneau C, Li JZ, Kim ST, et al. Quality of life predicts progression-free survival in
[18]
[19]
patients with metastatic renal cell carcinoma treated with sunitinib versus interferon alfa. J Oncol Pract 2009;5:66e70. Maione P, Perrone F, Gallo C, Manzione L, Piantedosi F, Barbera S, et al. Pretreatment quality of life and functional status assessment significantly predict survival of elderly patients with advanced non-small-cell lung cancer receiving chemotherapy: a prognostic analysis of the multicenter Italian lung cancer in the elderly study. J Clin Oncol 2005;23:6865e72. Quinten C, Coens C, Mauer M, Comte S, Sprangers MA, Cleeland C, et al. Baseline quality of life as a prognostic indicator of survival: a meta-analysis of individual patient data from EORTC clinical trials. Lancet Oncol 2009;10:865e71. Scher HI, Morris MJ, Stadler WM, Higano C, Basch E, Fizazi K, et al. Trial design and objectives for castration-resistant prostate cancer: updated recommendations from the Prostate Cancer Clinical Trials Working Group 3. J Clin Oncol 2016;34:1402e18. Gupta D, Braun DP, Staren ED. Prognostic value of changes in quality of life scores in prostate cancer. BMC Urol 2013;13:32. Sullivan PW, Nelson JB, Mulani PM, Sleep D. Quality of life as a potential predictor for morbidity and mortality in patients with metastatic hormone-refractory prostate cancer. Qual Life Res 2006;15:1297e306. Nussbaum N, George DJ, Abernethy AP, Dolan CM, Oestreicher N, Flanders S, et al. Patient experience in the treatment of metastatic castration-resistant prostate cancer: state of the science. Prostate Cancer Prostatic Dis 2016;19:111e21. Scher HI, Fizazi K, Saad F, Taplin ME, Sternberg CN, Miller K, et al. Increased survival with enzalutamide in prostate cancer after chemotherapy. N Engl J Med 2012;367:1187e97. Beer TM, Armstrong AJ, Rathkopf DE, Loriot Y, Sternberg CN, Higano CS, et al. Enzalutamide in metastatic prostate cancer before chemotherapy. N Engl J Med 2014;371:424e33. Cella D, Nichol MB, Eton D, Nelson JB, Mulani P. Estimating clinically meaningful changes for the functional assessment of cancer therapy-prostate: results from a clinical trial of patients with metastatic hormone-refractory prostate cancer. Value Health 2009;12:124e9. Esper P, Mo F, Chodak G, Sinner M, Cella D, Pienta KJ. Measuring quality of life in men with prostate cancer using the functional assessment of cancer therapy-prostate instrument. Urology 1997;50:920e8. Victorson DE, Beaumont JL, Rosenbloom SK, Shevrin D, Cella D. Efficient assessment of the most important symptoms in advanced prostate cancer: the NCCN/FACT-P symptom index. Physco-oncology 2011;20:977e83. Yost KJ, Eton DT. Combining distribution- and anchor-based approaches to determine minimally important differences: the FACIT experience. Eval Health Prof 2005;28:172e91. Cella D, Ivanescu C, Holmstrom S, Bui CN, Spalding J, Fizazi K. Impact of enzalutamide on quality of life in men with metastatic castration-resistant prostate cancer after chemotherapy: additional analyses from the AFFIRM randomized clinical trial. Ann Oncol 2015;26:179e85. Loriot Y, Miller K, Sternberg CN, Fizazi K, de Bono JS, Chowdhury S, et al. Effect of enzalutamide on health-related quality of life, pain, and skeletal-related events in asymptomatic and minimally symptomatic, chemotherapy-naive patients with metastatic castration-resistant prostate cancer (PREVAIL): results from a randomised, phase 3 trial. Lancet Oncol 2015;16: 509e21. Basch E, Autio K, Ryan CJ, Mulders P, Shore N, Kheoh T, et al. Abiraterone acetate plus prednisone versus prednisone alone in chemotherapy-naive men with metastatic castration-resistant prostate cancer: patient-reported outcome results of a randomised phase 3 trial. Lancet Oncol 2013;14:1193e9. Fizazi K, Scher HI, Miller K, Basch E, Sternberg CN, Cella D, et al. Effect of enzalutamide on time to first skeletal-related event, pain, and quality of life in men with castration-resistant prostate
T.M. Beer et al. / European Journal of Cancer 87 (2017) 21e29
[20]
[21]
[22]
[23]
[24]
[25]
[26]
cancer: results from the randomised, phase 3 AFFIRM trial. Lancet Oncol 2014;15:1147e56. Harland S, Staffurth J, Molina A, Hao Y, Gagnon DD, Sternberg CN, et al. Effect of abiraterone acetate treatment on the quality of life of patients with metastatic castration-resistant prostate cancer after failure of docetaxel chemotherapy. Eur J Cancer 2013;49:3648e57. Nilsson S, Cislo P, Sartor O, Vogelzang NJ, Coleman RE, O’Sullivan JM, et al. Patient-reported quality-of-life analysis of radium-223 dichloride from the phase III ALSYMPCA study. Ann Oncol 2016;27:868e74. Parker C, Zhan L, Cislo P, Reuning-Scherer J, Vogelzang NJ, Nilsson S, et al. Effect of radium-223 dichloride (Ra-223) on hospitalisation: an analysis from the phase 3 randomised Alpharadin in Symptomatic Prostate Cancer Patients (ALSYMPCA) trial. Eur J Cancer 2017;71:1e6. ¨ zgu¨roglu M, Hansen S, Kocak I, Bahl A, Oudard S, Tombal B, O et al. Impact of cabazitaxel on 2-year survival and palliation of tumour-related pain in men with metastatic castration-resistant prostate cancer treated in the TROPIC trial. Ann Oncol 2013; 24:2402e8. Halabi S, Vogelzang NJ, Kornblith AB, Ou SS, Kantoff PW, Dawson NA, et al. Pain predicts overall survival in men with metastatic castration-refractory prostate cancer. J Clin Oncol 2008;26:2544e9. Collette L, van Andel G, Bottomley A, Oosterhof GO, Albrecht W, de Reijke TM, et al. Is baseline quality of life useful for predicting survival with hormone-refractory prostate cancer? A pooled analysis of three studies of the European Organisation for Research and Treatment of Cancer Genitourinary Group. J Clin Oncol 2004;22:3877e85. Sadetsky N, Hubbard A, Carroll PR, Satariano W. Predictive value of serial measurements of quality of life on all-cause mortality in prostate cancer patients: data from CaPSURE (cancer of the prostate strategic urologic research endeavor) database. Qual Life Res 2009;18:1019e27.
29
[27] Lyyra TM, Leskinen E, Jylha¨ M, Heikkinen E. Self-rated health and mortality in older men and women: a time-dependent covariate analysis. Arch Gerontol Geriatr 2009;48:14e8. [28] Traina S, Li T, Johnson K, Ho KF, Molina A, Cella D. Is there a relationship between patient-reported outcomes (Pros) and clinical outcomes in metastatic castration-resistant prostate cancer (Mcrpc) post-docetaxel? Value Health 2015;18:A470. [29] Traina S, Li T, Johnson K, Ho KF, Molina A, Cella D. Analysis of the relationship between patient-reported outcomes (Pros) and clinical outcomes in metastatic castration-resistant prostate cancer (Mcrpc) patients without prior chemotherapy. Value Health 2015;18:A335e6. [30] Holt-Lunstad J, Smith TB, Layton JB. Social relationships and mortality risk: a meta-analytic review. PLoS Med 2010;7: e1000316. [31] Lamers SM, Bolier L, Westerhof GJ, Smit F, Bohlmeijer ET. The impact of emotional well-being on long-term recovery and survival in physical illness: a meta-analysis. J Behav Med 2012;35:538e47. [32] Coyne JC, Pajak TF, Harris J, Konski A, Movsas B, Ang K, et al. Emotional well-being does not predict survival in head and neck cancer patients: a Radiation Therapy Oncology Group study. Cancer 2007;110:2568e75. [33] Brown JE, Butow PN, Culjak G, Coates AS, Dunn SM. Psychosocial predictors of outcome: time to relapse and survival in patients with early stage melanoma. Br J Cancer 2000;83:1448e53. [34] Nakaya N, Bidstrup PE, Saito-Nakaya K, Frederiksen K, Koskenvuo M, Pukkala E, et al. Personality traits and cancer risk and survival based on Finnish and Swedish registry data. Am J Epidemiol 2010;172:377e85. [35] Petticrew M, Bell R, Hunter D. Influence of psychological coping on survival and recurrence in people with cancer: systematic review. BMJ 2002;325:1066. [36] Fayers PM, Machin D. Exploring longitudinal data. Quality of life: the assessment, analysis and interpretation of patientreported outcomes. 2nd ed. John Wiley & Sons, Ltd; 2007. p. 305e27.