Quality of life assessment as a predictor of survival in non-small cell lung cancer

Lung cancer is the most common cancer in the United States in terms of incidence and mortality with 219,440 new cases and 159,390 deaths in 2009 [1]. Patients with lung cancer experience a variety of distressing symptoms, many of which begin prior to diagnosis and continue throughout the course of the disease and its treatment, adversely affecting functional status and quality of life (QoL) [2, 3]. The vast majority of patients, especially those with advanced disease, do not have curative treatment options and therefore, the goal of therapy for such patients is prolongation of survival without negatively impacting QoL [4]. Unfortunately, differences in survival time across the spectrum of available treatments for advanced lung cancer are modest. Thus, treatment for lung cancer which offers the potential for prolonging patient survival must always be judged in the context of its effects on patient QoL [5‐8].

Since every patient's QoL is multidimensional, consisting of physical, functional, psychological, social, and spiritual domains [5], there is a growing consensus that treatment efficacy should be judged by its effects on both quantity and quality of life. This has led to the inclusion of QoL assessment as a primary endpoint in all types of clinical trials along with the traditional endpoints of tumor response and survival. There is general agreement in the medical and scientific research communities that patients are the best source of QoL information, especially when these patient-reported outcomes can be defined scientifically and measured with validated tools.

Extensive data in the literature have shown that QoL tools measuring the activities of daily life can predict survival in several different types of cancers independent of the extent of the disease and other clinical prognostic factors [9‐27]. These studies have used different QoL instruments which can be broadly classified into generic and targeted [28]. Generic QoL instruments tend to ask questions general enough for broad applicability, including for use among individuals in good health. On the other hand, targeted measures may focus on specific symptoms that are common to a particular cancer or to its treatment. There is no "gold standard" cancer QoL questionnaire available and the choice of selecting a QoL questionnaire for a particular study is governed for the most part by the research goals of that study.

The positive relationship between QoL and survival as described above has not been universally observed in the literature, with some studies founding no association between QoL and survival [29‐31]. As a result, there is no unanimity on the prognostic role of QoL in cancer. These conflicting and inconsistent results in the literature could perhaps reflect the different methodologies used to analyze the data (as an example by selecting different cut off values for variables), different patient populations investigated, different QoL questionnaires used, or alternatively, different selection of covariates to be included in the regression analysis [32]. Moreover, it is currently unclear why QoL may be prognostic, and it has been hypothesized that, in patients with micrometastatic disease, there is production of tumor factors that affect general health and that patients perceive this and report it in terms of poor QoL before it is clinically or radiologically apparent [33]. Therefore, preclinical progression of disease after curative treatment may be associated with worsening of QoL, which may be assessed with self-reported QoL measures [33]. It is possible that QoL presumably captures those aspects of disease severity that may not be apparent in the observer-rated performance status or tumor burden [34]. In the light of the above, we investigated whether QoL can predict survival in non-small cell lung cancer patients treated at a community hospital comprehensive cancer center.

Even though the QoL of patients is considered a critical endpoint in oncology that can also provide useful prognostic information to patients and clinicians, QoL continues to be evaluated infrequently in most clinic oncology practices apart from research studies. The barrier is not a lack of valid tools but rather the challenges inherent in incorporating QoL measurements into busy clinical practices [5]. This will continue as long as QoL assessment is regarded as peripheral to the goal of standard clinical cancer therapy. It is our contention, however, that QoL is as meaningful to patients as length of life, a situation which is coming to be appreciated by patients, advocates, insurance providers, and employers. Thus, there is a strong rationale to incorporate regular QoL assessment and management into all oncologic practice settings.

In the current study, we chose QLQ-C30 as a valid and reliable tool to assess patient QoL which is appropriate for assessing large, heterogeneous patient populations. It is relevant that this instrument concentrates on a patients' ability to fulfill the activities of daily living. While this emphasis has been regarded in research studies as an appropriate way to assess the effectiveness of new drugs or novel combinations of agents, cumulative experience from several decades of use of the QLQ-C30 also demonstrate the relevance of this information for clinical practitioners. Thus, results from the QLQ-C30 provide important information to guide treatment decisions based on balancing the need to control tumor progression with the need to know how treatment affects the patient's capacity to fulfill the activities of their daily life at work and in the home.

In this study, we found that global QoL as well as physical function were significant predictors of patient survival in the entire study population. The finding of physical function predicting survival in lung cancer has been reported by previous studies [11, 20, 39‐41]. For example, Ganz et al. reported a statistically significant relationship between initial patient-rated QoL scores and subsequent survival [40]. In another study by Herndon et al., increased pain, appetite loss, fatigue, lung carcinoma symptoms, poorer overall QoL, and poorer physical functioning predicted significantly poorer survival. However, only the pain score was retained in the final multivariate model [20]. Finally, a study conducted by Fielding and Wong found that better physical functioning and better appetite were predictive of longer survival after fully adjusting for sociodemographic and clinical factors in Chinese lung cancer patients [39].

The finding of global QoL predicting survival in lung cancer has also been reported by previous studies [15, 20, 22, 25, 41‐43]. For example, Langendijk et al. examined the significance of pretreatment QoL and symptom scores in lung cancer patients treated with high dose radiotherapy and found the EORTC global QoL score as the strongest prognostic factor [22]. Montazeri et al. showed that pre-diagnosis EORTC global QoL was a significant predictor of survival in lung cancer [25]. Dharma-Wardene et al. found that baseline FACT-G total score is a statistically significant predictor of survival in patients with advanced lung cancer [15]. Brown et al. found EORTC global QoL, role functioning, fatigue, appetite loss, and constipation as prognostic indicators of survival in patients with NSCLC at 12 weeks [42]. Maione et al. found that only pretreatment EORTC global QoL and instrumental activities of daily living scores were significant predictors for survival in elderly patients with advanced non-small-cell lung cancer treated with chemotherapy [43].

In the previously published studies mentioned above, the most commonly adjusted variables were age, gender, socioeconomic status, performance status, various cell counts, metastasis, treatment types, cancer stage, histology subtypes, sites involved, and weight loss. However, it is worth mentioning that some of these studies failed to provide adequate controls for confounding, for example, there is no information on the extent of the disease or whether the tumor size shrank in response to therapy [15, 22, 32]. In most of these studies, it is not possible to determine whether QoL independently predicted survival or simply reflected the progress of the underlying disease. For example, it has been suggested that physical functioning might be a surrogate marker for an unrecognized biological prognostic indicator, so a causal association between physical functioning and survival time should not be inferred [28].

It is to be noted that in our study, several QoL function and symptom scales that were predictive of survival on univariate analysis, lost their statistical significance after the basic demographic and clinical factors were adjusted for. Only physical and global function retained their statistical significance in multivariate modeling. Collectively, these results demonstrate that after the potential confounding factors are adjusted for in the multivariate analysis, the prognostic effects of QoL either disappear or are mitigated significantly. Thus, before making any claims regarding the independent prognostic effects of QoL in cancer, adequate statistical adjustment of covariates needs to be achieved. Finally, our finding on global function should be interpreted with caution because the global function score likely reflects the contribution of the physical function score.

We also found that physical function, nausea/vomiting, insomnia, and diarrhea in newly diagnosed patients predicted survival in multivariate analysis. Similarly, in previously treated patients, physical function was predictive of survival in multivariate analysis. For both of these clinical settings, it is easy to appreciate the negative impact these attributes would exert on the ability of a patient to function. While the current study does not elucidate causative relationships between these functions and patient survival, the importance of measuring these functions in patients and attempting to address their effects therapeutically should be obvious. The finding of poorer physical, role and social functioning scores and fatigue, nausea/vomiting, pain, dyspnea and appetite loss symptom scores in previously treated patients as compared to newly diagnosed patients is not surprising given that previously treated patients have more advanced disease and are well into their course of illness as compared to newly diagnosed patients.

The results of this study have important implications for both clinical and research practices in an integrative oncologic care setting. They suggest that health care professionals should evaluate baseline QoL in all patients and take this into consideration when planning treatment. QoL should also be assessed regularly during treatment and appropriate intervention taken to improve QoL when indicated. QoL was initially designed to be used as a clinical indicator. Self-reported QoL assessment using validated questionnaires plays a very important role in acquiring and monitoring information on patient's functional and symptom status throughout the course of treatment. In the absence of concrete information on the prognostic role of QoL in cancer survival, its utility as a clinical indicator should not be jeopardized. In fact, information about QoL in lung cancer patients undergoing treatment can provide healthcare providers with a perspective on post-treatment recovery, including the positive aspects of long-term care, as well as anticipated problems [3].

Although this study presents interesting findings and additional questions for further inquiry, several limitations require acknowledgment. Our study, because of its retrospective nature, relies on data not primarily meant for research. As a result, we could not control for some additional factors in our analyses that could influence survival such as medical comorbidities, performance status, socioeconomic factors, support system, exercise, insurance coverage, employment and educational level. It has been suggested that QoL data might be markers of the socioeconomic status of cancer patients. For instance, cancer patients with higher social class would have a better QoL, and consequently those who report a better QoL at baseline assessment may live longer [28]. As suggested by Fielding and Wong, full adjustment is needed for factors affecting outcome in order to demonstrate any independent predictive power of QoL [39]. The patient cohort was limited only to those patients who were English speakers and therefore is not representative of the complete spectrum of lung cancer patients. A majority of our patients had advanced stage disease at presentation. As a result, generalizability of the study findings to NSCLC patients with early stage disease might be questionable. However, we have no reasons to believe that patients with early stage disease will have different findings. Moreover, this study is not able to reveal causative relationships between any QoL element and patient survival. Rather, QoL was found to act as a surrogate marker for otherwise undetected prognostic factors [11]. We did not control for the multiple comparisons made in this study, but this is acceptable for hypothesis-generating studies [26]. Finally, we did not evaluate the prognostic significance of changes in QoL scores during the entire duration of treatment.

Nevertheless, this study also has several strengths, including a large sample size; complete data for all QLQ-C30 subscales for the entire study sample; high compliance with completion of the questionnaire; the use of a valid and reliable QoL instrument; comprehensive documented clinical parameters in nearly all patients; and availability of mature and reliable survival data.

This study suggests that baseline global QoL and physical function scale provide useful prognostic information in non-small cell lung cancer patients.