Quality of life in a real-world cohort of advanced breast cancer patients: a study of the SONABRE Registry

Advanced breast cancer (ABC), defined as metastatic breast cancer (stage IV), is a major cause of death among women worldwide [1]. Despite improving outcome of patients with ABC due to the introduction of new treatment agents and strategies, the disease remains largely incurable [2‐4]. Thus, treatment focuses on both quality- and prolongation of life. Previous studies have shown that the quality of life (QoL) of these patients is positively associated with response to antitumor treatment, time to progression, and survival, but negatively associated with toxicity [5‐8]. It is therefore essential to report on QoL in addition to clinical parameters to determine the health benefit of a new treatment.

There are different questionnaires suitable for evaluation of QoL among patients with ABC ranging from disease specific tools such as the European Organization for Research and Treatment for Cancer Quality of Life Questionnaire (EORTC QLQ-C30) to more generic instruments that allow comparison of QoL across diseases. Here we use the European Quality of Life Five-Dimensions questionnaire (EQ-5D) to describe QoL and estimate a single summary index value rated on a scale from 0 (dead) to 1 (full health) [9]. The EQ-5D is the most widely used generic instrument to obtain such an utility score that reflects the overall health-state of participants according to the preferences of the general population of a country or region. In several European countries, this standardized questionnaire is preferred as a key component in cost-utility analysis [10, 11], and recently it was highlighted to be the most commonly cited multi-attribute utility instrument [12]. It has been used in clinical trials as well as observational studies for different types and stages of treatment of malignant diseases [13].

Despite the increased awareness of evaluation of QoL in addition to clinical outcomes, QoL and health-utility are underreported and not uniformly addressed in most breast cancer studies [8, 14‐17]. Gaining insights into the QoL and the factors influencing this outcome is especially important for advanced stages of breast cancer, as the goal for these patients does comprise not only prolonging life, but also optimizing the QoL, considering treatments given in this disease stage are of palliative intend. Understanding which factors influence the QoL of these patients promotes individualized high-quality care. The primary aim of our analysis was to evaluate QoL by means of the EQ-5D in patients with ABC in a real-world Dutch cohort. In addition, we reported differences in QoL between subgroups of patients based on age, comorbidity, tumor-, and treatment characteristics, and assessed the association of duration of metastatic disease and time to death with QoL.

This cross-sectional study evaluated the QoL in real-life of 92 ABC patients on different treatments (including no treatment). The median EQ-5D utility score was 0.691 (IQR 0.244), and differed significantly between subgroups based on age and the presence of comorbidity to both a statistically significant (p < 0.05) as well as a clinically relevant (difference in utilities > 0.03) extend, to the detriment of patients aged ≥ 65 years and patients with any comorbidity. Interestingly, we noticed a weak but significant positive correlation between the observed QoL and a patients’ remaining survival time (r = 0.260, p = 0.0252). Patients with a few months to live reported lower QoL as compared to those with a longer time to live, and this association was most clearly observed in the patients aged < 65 years (r = 0.340, p = 0.0169).

In clinical practice, it is difficult to determine the prognosis of an individual patient accurately. The decision on whether to start a new treatment line in case of progression is based on the doctor’s experience and perception and the patient’s preferences. The doctor’s perception of the patient’s wellbeing is reflected by the performance status ascribed to the general functioning of the patient. Indeed, performance status has shown to be related to treatment duration [36, 37] and overall survival [37‐41] of patients with ABC within real-world studies. The observation that QoL can also be of assistance in guiding these decisions is interesting and should be further investigated in prospective studies.

Previous real-world studies among breast cancer patients using the EQ-5D report similar to slightly higher utility scores (median ranging from 0.64–0.82) [19, 20, 22, 24, 26, 42, 43] compared to our median (0.69). These differences in utility scores can partly be explained by the fact that most of these studies [22, 24, 26, 43] only included patients that were actively treated for their disease, leaving out the more vulnerable patients receiving supportive care. Additionally, the majority of these real-world studies [19, 20, 24, 42] also included patients with early breast cancer for which it can be expected that the QoL will be better due to less disease-related symptom burden. Furthermore, the mean age of patients within these trials varied, where studies with a lower mean age generally reported better utility scores; e.g., the study by Kim et al. reported a mean utility of 0.82 in a population with a median age of 49.3 years [43].

Factors found to have a significant association with QoL within previous observational studies for patients with breast cancer of various stages, including ABC, were age, [19, 20] fatigue, [19, 25, 26] financial difficulties, [19] pain, [19, 22, 25] body image, [25] comorbidities, [20],performance status, [23] type of therapy, [19‐24] number of lines of therapy, [21, 24] location of metastases, [21, 22, 25], and disease progression [23, 26‐28]. However, these associations were inconsistent, possibly due to differences in questionnaires used (EQ-5D, EORTC QLQ-C30 or -BR23, Patient Care Monitor, FACT-B), population’s health-related preferences, cultural dissimilarities, and methodology used in the valuation process [44]. Additionally, not all studies investigated the same variables, and within the investigated variables, impact varied between studies. This inconsistency in significance of the relation between the mentioned factors and the experienced QoL might be due to the lack of standardized methods, in combination with the observational nature of the studies.

Here, we found no association between the reported QoL and duration of metastatic disease and the number of prior lines of systemic therapy (data not shown). This stability of QoL with increasing number of treatment lines could be due to a correct selection of suitable patients for further therapy by their oncologists. Mostly, patients who are fit enough will be treated with further lines of treatment. Alternatively, the gradually increased therapeutic possibilities, the increased duration and amount of response to treatment resulting in better symptom control, could be an explanation of the lack of association between QoL and number of treatment lines.

The decision-making process on whether or not to continue treatment is also influenced by culture. Studies on EQ-5D value sets indicated that population-specific beliefs about health can contribute to differences in valuing a specific dimension of the EQ-5D as more or less important [44, 45]. Within a simulation study, Dutch respondents ascribed less weight than UK respondents to most dimensions of the EQ-5D, with the exception of the anxiety/depression dimension [45]. Another study comparing different value sets for the EQ-5D also found the Dutch value set was the only one out of 14 sets ascribing a worse health-state to a depressed patient compared to a patient with pain [44]. Several other studies among healthy as well as specific disease populations indicated that using different country-specific value sets produces substantially different results, and that despite a high level of correlation these tariffs cannot be used interchangeably [46‐54]. More research is needed regarding the international transferability of utilities to ensure the economic evaluations underlying decision making are reliable and applicable to the intended population.

Thus, due to the large variation in methods used to assess QoL in breast cancer patients, [8, 14‐16] there is a need for a more standardized approach. Therefore, the European guideline on the treatment of ABC urges the development of specific tools to evaluate QoL in ABC patients with attention to solid methodology [55]. The goal is to find a patient-centered way to measure QoL which incorporates the most relevant factors for patients, physicians, and decision makers with regard to drug approval and reimbursement. An important step forward is the central registration of patient-reported outcome data. For this purpose the EORTC Quality of Life Group developed a dedicated registry (PROMOTION) to identify, track, and analyze information about patient-reported outcomes (PRO), including QoL, of cancer patients included in randomized clinical trials since 2004 [56]. This database contains information regarding the assessment methodology, statistical design, and reported clinical and PRO results.

Our study is a cross-sectional study, and as a result changes in QoL during the disease course could not be investigated. Unfortunately, multivariate regression models were not performed due to the limited sample size of some subgroups and the skewness of the data. We intentionally mainly used descriptive statistics and (non)parametric tests, as regression models would be of limited usefulness in this case. Using a larger sample size and adopting multivariate regression models might be an informative step for future research to investigate a cause-effect relationship more profoundly. Furthermore, we only included patients that visited our outpatient clinic which could have led to an overestimation of the QoL scores. Finally, we did not have data on performance score and other possible relevant factors, such as toxicity. While highlighting the potential value of QoL in a similar role to performance status, such as in guiding patient decision making, correlation between the EQ-5D values and performance status assessments would have been worth mentioning. If highly correlated, using the EQ-5D could provide the advantage of covering more dimensions of QoL over the assessment of performance status. Conversely, performance score assessment is routinely done and might provide a quick and easy tool to guide decision-making. Strong points of our study are on the use of the preferred and validated generic instrument (EQ-5D), and that fact that our study is of important relevance for the ABC field, especially since it contains health-state utility data from routine clinical practice, which may be considered more representative data for guidance of decision-making than data from clinical trials.

In conclusion, within this real-world cross-sectional study, QoL was significantly associated with age, comorbidity, and remaining survival duration. The observation of a lower QoL in ABC patients, possibly indicating the last period of life, may assist clinical decision-making on timing of cessation of systemic antitumor therapy.