Article Text
Abstract
Objectives Inclusion of patient-reported outcomes (PROs) in cancer randomised controlled trials (RCTs) may be particularly important for older patients. The objectives of this systematic review were to quantify the frequency with which older patients are included in RCTs with PROs and to evaluate the quality of PRO reporting in those trials.
Methods All RCTs with PRO endpoints, published between January 2004 and February 2019, which included a patient sample with a mean/median age ≥70 years, were considered for this systematic review. The following cancer malignancies were considered: breast, colorectal, lung, prostate, gynaecological and bladder cancer.
Quality of PRO reporting was evaluated using the International Society for Quality of Life Research–PRO standards. Studies meeting at least two-thirds of these criteria were considered to have high-quality PRO reporting.
Results Of 649 RCTs identified with a PRO endpoint, only 72 (11.1%) included older patients. Of these, 35 trials (48.6%) were conducted in patients with metastatic/advanced disease. PROs were primary endpoints in 20 RCTs (27.8%). Overall survival was the most frequently reported clinical outcome in studies of patients with metastatic/advanced cancer (n=28, 80%). One-third of the RCTs (n=24, 33.3%) were considered to have high-quality PRO reporting. Overall, the largest prevalence of RCTs with high-quality PRO reporting was observed in prostate and colorectal cancers.
Conclusions Our review indicates not only that PRO–RCT-based studies in oncology rarely include older patients but also that completeness of PRO reporting of many of them is often suboptimal.
- older patients
- patient-reported outcomes
- quality of life
- survival
- cancer
- randomized-controlled trials
Statistics from Altmetric.com
Introduction
Due to the ageing of the society, the number of older persons is projected to more than double worldwide, rising from 962 million in 2017 to 2.1 billion in 2050.1 As cancer incidence increases with age, the number of older patients diagnosed with cancer is expected to increase over the next decades.2 3 Currently, 53.4% of new cancer cases in the USA are diagnosed in people aged 65 years or over, and the median age at cancer diagnosis is 66 years.4 However, despite the high incidence of cancer in older people, these patients are frequently under-represented in clinical trials evaluating new cancer treatments.5–8
Low recruitment of older patients in cancer clinical trials often reflects age-related exclusion criteria, such as comorbidity, reduced life expectancy, polypharmacy and its interaction with chemotherapy.9–11 Additionally, there is concern that disease symptoms and treatment toxicities may have a greater impact on the health-related quality of life (HRQOL) of older patients, limiting their functional independence and ability to carry out daily activities.12 13 Thus, when older patients are included in clinical trials, it becomes particularly important to assess outcomes such as physical, role and social functioning, and HRQOL,14 alongside the more traditional survival outcomes.2 12
Compared with younger patients, older people tend to weigh HRQOL as being more important than duration of life when making treatment decisions.15 However, some earlier research has suggested that HRQOL and other patient-reported outcomes (PROs) are rarely included in the design of randomised controlled trials (RCTs) that involve older patients with cancer.16 Also, previous reviews have shown that, regardless of the specific cancer populations enrolled, RCTs with a PRO endpoint are frequently poorly reported, thereby limiting availability of high-quality data that can robustly inform patient care.17–19
The main objectives of this review were to quantify the frequency with which older patients are included in cancer RCTs with PROs as primary or secondary endpoints and to evaluate the quality of the PRO results generated in those studies. The secondary objective was to summarise outcomes from the higher quality PRO studies.
Materials and methods
Criteria for trial inclusion
The analysis reported here was based on data collected in the Patient-Reported Outcomes Measurements Over Time In ONcology Registry (http://promotion.gimema.it).20 This registry includes a wide range of information mainly covering PRO assessment methodology, statistical analysis and outcomes from published cancer RCTs reports. Also, to maximise data quality, detailed information from each RCT is extracted by two independent reviewers and, in case of disagreement, a third senior reviewer is consulted for the final decision. If an RCT generates multiple publications (eg, a primary paper on clinical outcomes and a secondary paper on PROs), information from all related papers are considered. Data are collected and stored into a password-protected online platform: REDCap system.21 The studies included in this registry are identified using systematic literature searches with PubMed/MEDLINE, the Cochrane Library, PsycINFO and PsycARTICLES. Additional studies are identified through hand searching of the literature, and references of publications are checked to find relevant studies for inclusion. Details on search strategies used for each cancer site considered in this review are reported in online supplementary appendix A. RCTs published since January 2004 that enrolled at least 50 patients (combined arms) and employed at least one PRO measure are included in this registry. Studies published since 2004 are included as the quality of PRO reporting in RCTs was broadly known for the main cancer disease sites before that date.22 Studies comparing conventional medical treatments (eg, surgery, chemotherapy, radiotherapy or targeted therapies) are included. Studies on prevention and screening programmes, psychological interventions or complementary medicine are excluded. Only English articles are considered, while case reports and conference abstracts are excluded.
Supplemental material
Data collection
For the purpose of this review, we first selected from the registry all RCTs published up to February 2019 in breast, colorectal, lung, prostate, gynaecological and bladder cancers. These are the cancer sites with the highest incidence in the European Union in 2018.23 We then included only those RCTs whose patients had a mean or a median age of at least 70 years. When the mean/median age was not available for the overall sample but reported for individual treatment arms, we included those RCTs with mean/median age per study arm of at least 70 years. We chose this age threshold as it is commonly used as a chronological landmark in oncology clinical trials.13 24 Two reviewers were involved in screening the RCTs included for current analysis.
Type of information considered
Data considered from each RCT, for the purpose of this analysis, were basic trial characteristics; information about clinical and PRO endpoints; information about difference in overall survival (OS); and the quality of the PRO reporting, based on the International Society for Quality of Life Research (ISOQOL)-PRO checklist25 and the Consolidated Standards of Reporting Trials (CONSORT)–PRO extension26 (see further for more detail). Clinical endpoint classification was based on the categories used by Hamaker and colleagues16: OS, progression-free survival (PFS), toxicity, efficacy, healthcare utilisation, biological parameters and completion of treatment (online supplementary appendix B). Disease stage was classified according to the following four categories: only patients with metastatic/advanced cancer, only patients with non-metastatic/local cancer, both and unclear. If trials included patients with both metastatic/advanced and non-metastatic/local cancer, when possible, this was reclassified in either categories according to the most represented patient population.
Assessment of the quality of PRO reporting
We evaluated the quality of the PRO reporting using previously defined criteria27 28 based on the ISOQOL consensus-based recommendations for reporting PRO in publications of RCTs.25 This represents one of the most widely endorsed set of guidelines that formed the basis for the CONSORT–PRO extension.26 Briefly, the ISOQOL checklist consists of a set up to 28 issues that should be reported (in case PRO is a primary endpoint).
We defined an RCT as being of high quality, regarding the PRO assessment, if at least two-thirds of the ISOQOL checklist criteria were met. Further details on criteria definition have been previously reported.27 28 All analyses were performed with SAS V.9.4.
Results
A total of 950 papers were identified, which referred to 649 unique RCTs published in breast, colorectal, lung, prostate, gynaecological and bladder cancers. Of these, 577 did not meet our age cut-off criteria and the remaining 72 RCTs (11.1%) were included in our analysis.
Overview of study characteristics
Of the 72 RCTs identified, 39 (54.1%) were conducted in prostate cancer, 13 (18.1%) in lung cancer, 10 (13.9%) in colorectal cancer, 5 (6.9%) in breast cancer, 4 (5.6%) in bladder cancer and 1 (1.4%) in gynaecological cancer. In total, 35 trials (48.6%) were conducted in patients with metastatic/advanced disease. The majority of lung (n=10, 76.9%), colorectal (n=5, 50%) and prostate (n=19, 48.7%) studies enrolled patients with metastatic/advanced cancer, while the majority of breast (n=4, 80%) and all the bladder (n=4, 100%) studies enrolled patients with non-metastatic/local cancer.
Twenty-one RCTs (29.2%) published a secondary paper reporting additional PRO data. In 19 of these RCTs, PROs were included as secondary endpoints, and the average time between publication of the PRO paper and the primary trial publication was 23.4 months (range 0–50 months). PROs were a primary endpoint in 20 RCTs (27.8%) and a secondary endpoint in 52 RCTs (72.2%) (table 1).
Type of endpoints in RCTs in non-metastatic/local disease
Among the RCTs in patients with non-metastatic/local disease (n=27), PROs were a primary endpoint in 9 RCTs (33.3%) and a secondary endpoint in 18 RCTs (66.7%). Toxicity was evaluated in more than half of the trials (n=15, 55.6%), while survival endpoints were less frequently used: OS was an endpoint in 14 RCTs (51.9%) and PFS was an endpoint in 13 RCTs (48.2%) (table 2).
Type of endpoints in RCTs in metastatic/advanced disease
Among the studies conducted in patients with metastatic/advanced disease (n=35), PROs were a primary endpoint in 8 RCTs (22.9%) and a secondary endpoint in 27 RCTs (77.1%). In these 35 RCTs, OS was the most frequently reported (n=28, 80%), followed by PFS (n=26, 74.3%), toxicity (n=24, 68.6%) and efficacy (n=21, 60%) (table 2).
Overview of the methodological quality of PRO reporting
Twenty-four of the 72 studies (33.3%) were evaluated to have high-quality PRO reporting (see full list of references in online supplementary appendix C). Of these, 13 (54.2%) were conducted in patients with metastatic/advanced disease cancer. Overall, the largest prevalence of RCTs with high-quality PRO reporting was observed in prostate (n=18/39, 46.2%) and colorectal (n=4/10, 40%) cancers.
Summary of outcomes from higher quality PRO studies in non-metastatic/local disease
Of the non-metastatic/local disease RCTs with higher quality PRO reporting (n=11), four (36.4%) had a PRO as primary endpoint and seven (63.6%) had a PRO as secondary endpoint. The number of patients enrolled in these trials ranged from 98 to 1532. In four trials (36.4%), PROs favoured the experimental arm, and in five trials (45.5%), PROs did not differ between arms. OS was an endpoint in five trials (45.5%). In three of these trials, OS did not differ between study arms, and in one trial, OS was better in the experimental arm. In three trials, there were no significant differences observed between treatment arms in either OS or PROs. In one trial, OS did not differ, while PROs (role and social functioning) favoured the experimental arm. Another study reported improvement in OS in the experimental group (immediate androgen-deprivation therapy), but PROs (sexual activity, hormone treatment-related symptoms, hot flushes and breast symptoms) favoured the control group (delayed therapy). More details are reported in table 3.
Summary of outcomes from higher quality PRO studies in metastatic/advanced disease
Of the metastatic/advanced disease RCTs with higher quality PRO reporting (n=13), 2 (15.4%) had a PRO as a primary endpoint and 11 (84.6%) had a PRO as a secondary endpoint. The trial sample sizes ranged from 82 to 3040. In eight trials (61.5%), PROs were better for the experimental arm, and in four trials (30.8%), they did not differ between arms. OS was an endpoint in nine trials (69.2%), in seven (77.8%) of which survival was not different between arms and in only two (22.2%) of which survival improved in the experimental arm. In three of nine studies with OS as an endpoint (33.3%), both OS and PROs did not differ between arms, and in one study (11.1%), both favoured the experimental arm. In one study (11.1%), OS improved in the experimental group, but PROs did not differ between treatment arms, and in four RCTs (44.4%), PRO data favoured the experimental arm, while no differences in OS were detected between arms. Further details are reported in table 4.
Discussion
This comprehensive literature review examined publications extending over the past 15 years and found that, of 649 RCTs that included a PRO assessment, only 72 (11.1%) of the trials had a sample whose mean/median age was 70 years or greater. In only one-third of these 72 studies was the reporting of PROs considered as being of high quality. In view of the ageing population and predominance of cancer in older age groups that are often not studied or precluded from trials, this study shows that more research aimed at patients aged 70 years and older meeting higher PRO methodological standards is needed.
Among the majority of higher quality PRO studies, PRO data provided novel information that complemented traditional clinical endpoints. Of the 14 higher quality PRO studies evaluating OS, the majority (n=11, 78.6%) found no significant differences between arms in survival. Of these, in almost half of the cases, the PRO results favoured the experimental arm (n=5, 45.5%). For example, in a study conducted in patients with metastatic colorectal cancer, the addition of bevacizumab to fluorouracil and leucovorin was not found to provide OS benefits as compared with a placebo added to fluorouracil and leucovorin.29 However, time to deterioration in HRQOL outcomes was significantly longer in the bevacizumab arm compared with placebo.30
A recent US Food and Drug Administration analysis of trials supporting registration of new cancer therapies and conducted from 2005 to 2015 found that older people, in particular, those aged >75 years, were under-represented.31 Earlier, Lewis and colleagues5 found that only 32% of phase II and III cancer trials enrolled older patients, despite the fact that 61% of patients with incident cancers were older. Our results further extend these data by documenting that, even among RCTs that included PROs in their trial protocol, only few focused on older patient populations with cancer.
In a recent review, Marandino and colleagues32 found that, of 446 RCTs conducted between 2012 and 2016, only 236 included a PRO endpoint. In our review, we found that, in the same period, 21 RCTs with a PRO endpoint included older patients. Although it is difficult to make a direct comparison between our results and those of Marandino and colleagues32 due to differences in the methodology used to identify the studies, our data suggest that fewer RCTs that include an evaluation of PROs are conducted in older patient populations.
In 2011, the European Organisation for Research and Treatment of Cancer (EORTC) Elderly Task Force indicated that the inclusion of HRQOL, as composite endpoint (ie, in combination with efficacy endpoint) or primary endpoint, in clinical trials involving older people was one of its priorities.33 Our results suggest that this recommendation has not been followed unfortunately, as the proportion of RCTs with a PRO as primary endpoint included in our review actually also declined after 2011 (36.6% in RCTs published up to 2011 vs 16.1% in RCTs published after 2011). A recent analysis of the endpoints used in ongoing trials conducted in haematological malignancies found that, even in trials developed for older patients, less than one-fifth included a PRO measure.16
Although there is no consensus on how to best evaluate HRQOL in older patients with cancer, a specific questionnaire developed by the EORTC (ie, the EORTC Quality of Life Questionnaire - Elderly Cancer Patients Module QLQ-ELD14)34 has been available since 2013 and could facilitate implementation of HRQOL in this setting. However, inspection of the US National Institutes of Health clinical trial registry (ww.clinicaltrials.gov) indicates that, among the 5192 cancer trials recruiting older patients registered from 2014 onward and currently ongoing, only 1045 (20.1%) included a PRO measure and only 8 used a measure specifically designed for older patients, such as the EORTC QLQ-ELD14.34
In our systematic review, we found that, not only PRO–RCT-based studies do rarely include older patients with cancer but also methodological rigorousness of many of them is often suboptimal. Indeed, in two-thirds of the studies, the minimum level of adherence to the well-established international quality standards25 was not reached. One of the reasons for the poor methodological quality of PRO reporting might be the limited space dedicated to PROs in the primary publication, which may result in the under-reporting of PRO information. Marandino and colleagues found that, even when PRO results are presented in the primary paper, space dedicated in the Results section is less than 10%, and that secondary PRO papers are likely to be published 2 or 3 years after the primary publication.32 Among the studies included in our review with a PRO as secondary endpoint, only 36.7% dedicated a secondary paper on PRO outcomes, and the average time to secondary publication was 2 years.
Our study has several limitations that should be noted. First, we included only trials investigating pharmacological, surgical or medical interventions, and excluded studies on prevention, psychological interventions or complementary medicine. This may have excluded some trials conducted specifically on older patients with cancer. Second, we did not perform a meta-analysis on the quantitative data obtained. However, given the heterogeneity of PRO measures used across all studies and large variability in data reporting across studies, we did not make any attempt to perform such analysis. Also, although we used a comprehensive searching strategy, it is possible that some RCTs with a PRO endpoint might have been missed. Lastly, we selected RCTs where the median or mean age was at least 70 years. This selection criterion permitted inclusion of studies that recruited patients with a wide age range, also those younger than 70 years.
Our work also has key strengths. To the best of our knowledge, we have provided the first evidence-based data, from a large sample of RCTs conducted across a wide range of solid tumours, on the frequency of inclusion of older patients with cancer in trials with a PRO endpoint. Also, evaluation of the PRO methodological quality was based on the highest quality standards endorsed by major scientific societies.25 26 Finally, our review covered a large time window, which extends over the last 15 years, and reported very up to date studies being published until early 2019.
In conclusion, our analysis indicates that, among cancer RCTs including PROs, the proportion of those conducted in older patients is low. Investigators should minimise barriers that may exclude older people from participating in trials, incorporate PRO outcomes that may be particularly relevant to this population and ensure that the methodology used to collect and analyse PROs meets high-quality standards.
References
Footnotes
Contributors Study concepts: FE and FS. Study design: FE and FS. Data acquisition: FE and FS. Data analysis and interpretation: all authors. Statistical analysis: FC, FE and FS. Manuscript preparation: FE and FS. Manuscript editing: all authors. Manuscript review: all authors.
Funding This work was supported by the Gruppo Italiano Malattie Ematologiche dell’Adulto.
Competing interests None declared.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.