Health-related quality of life questionnaires in lung cancer trials: a systematic literature review

Background Lung cancer is one of the leading causes of cancer deaths. Treatment goals are the relief of symptoms and the increase of overall survival. With the rising number of treatment alternatives, the need for comparable assessments of health-related quality of life (HRQoL) parameters grows. The aim of this paper was to identify and describe measurement instruments applied in lung cancer patients under drug therapy. Methods We conducted a systematic literature review at the beginning of 2011 using the electronic database Pubmed. Results A total of 43 studies were included in the review. About 17 different measurement instruments were identified, including 5 generic, 5 cancer-specific, 4 lung cancer-specific and 3 symptom-specific questionnaires. In 29 studies at least 2 instruments were used. In most cases these were cancer and lung cancer-specific ones. The most frequently used instruments are the EORTC QLQ-C30 and its lung cancer modules LC13 or LC17. Only 5 studies combined (lung) cancer-specific questionnaires with generic instruments. Conclusions The EORTC-C30 and EORTC-LC13 are the most frequently used health-related quality of life measurement instruments in pharmacological lung cancer trials.


Introduction
Lung cancer is one of the most frequent cancers in Germany. With more than 47,000 new cases in 2006, lung cancer ranks third among malignant tumors; with a 5-year survival rate of about 15% [1]. Symptoms include cough, coughing up blood, shortness of breath, chronic lung inflammation, chest pain, weakness or loss of appetite. Due to a long term symptom-free course of the disease and non-specific complaints at first, lung cancer in contrast to other tumours is often diagnosed at an advanced stage. Therefore, treatment goals for these patients are symptom relief and an increased overall survival [2]. At the same time therapies that improve survival time are often accompanied by burdensome (toxic) side effects.
Because of the increasing number of therapy lines and treatment alternatives, the declining differences in clinical effectiveness and cost of drugs, the importance of consistent and comparable health-related quality of life (HRQoL) parameters grows -both for medical and health economic evaluation. Up to today, their inclusion in clinical lung cancer trials is generally neglected [2][3][4].
The questionnaire-based measurement of HRQoL has become standard. To assess the HRQoL in patients with lung cancer, about 50 different instruments are available that directly address to the patient or apply to the practitioner [4]. In general, criteria used to distinguish the various instruments are the aggregated or disaggregated scores, ordinal measures or cardinal scales and the disease specificity (see Table 1) [5].
In addition to former research by Liu et al. [3], who reviewed and summarized HRQoL measures in kidney cancer, hepatocellular carcinoma, and leukemia, the aim of this systematic literature review is to investigate which questionnaires are applied in lung cancer patients treated with drugs. Using this approach, we examine whether the variety of possible lung cancer measurement instruments is also reflected by research practice.

Methods
Research and documentation were carried out in accordance with the guideline PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) [6]. We searched the database PubMed combining the following search terms: "Biological Therapy" (MeSH), "Chemicals and Drugs Category" (MeSH), "Drug Therapy" (MeSH), "Individualized Medicine" (MeSH), "Lung Neoplasms" (MeSH), "Outcome and Process Assessment (Health Care)" (MeSH Major Topic), "Quality of Life" (MeSH Major Topic), "Symptom Palliation" (Free text search, Major Topic). The quality of life associated key words were defined as major topic, to exclude articles that deal with the issue only as a secondary aspect. In addition, a manual search was carried out. In order to focus on current publications, the present review includes literature published in English and German language between 2001 and 2011. Titles, abstracts and full-texts of the identified studies were reviewed independently by three researchers. Exclusion criteria are documented in Figure 1. The identified studies were analyzed concerning the HRQoL results and used questionnaires.

Results
A total of 43 studies on the HRQoL measurement in lung cancer patients treated with drugs were identified (see Figure 1). The language restriction led only to small deviations from the overall number of identified studies.
Most of the identified literature presents results of phase III clinical trials (see Table 2). In 27 studies HRQoL is a primary endpoint. The majority of studies (n=38) includes patients with non-small cell lung cancer (NSCLC), mainly  [5].     The most frequently used instrument is the EORTC QLQ-C30 (n = 29), a general questionnaire designed for (self-or interviewer administration) use in cancer patient populations [50]. This 30-item multi-dimensional questionnaire is available in over 60 languages and contains 4 domains (functional and symptom scales, global quality of life, and single items) requested by 4-point Likert or visual analogue scales. Its lung cancer specific module LC13 (13 additional items) is also used widely.
The FACT questionnaires are less frequently used (n = 8). Here, the FACT-G (General) contains 27 multi-dimensional items (physical, emotional, social and functional well-being) scored on 4-point scales [51], available in more than 50 languages. The lung cancer questionnaire (FACT-L) is a combination of the FACT-G and disease specific items (in total 37). Further information about the questionnaires as well as comparisons of possible instruments for lung cancer patients can e.g. be found on PROQOLID [52], and in Liu et al. [3], Camps et al. [4] or Damm et al. [53]. In 29 of our identified studies (67%) at least two instruments were used, mostly cancer and lung cancer-specific ones (see Table 2). In 23 studies (53%) the EORTC QLQ-C30 in conjunction with the LC13 module was used. Five studies combined disease-specific (cancer or lung cancer) questionnaires with generic instruments; two times the EQ-5D was applied.

Content-related results of the identified literature
Because of the different study populations and treatment regimes it is not possible to compare all the different studies in terms of HRQoL (see Table 2). However we tried to arrange some groups of HRQoL findings.
A majority of the included trials comparing various agent regimes shows no significant differences between treatment arms [13,15,21,29,31,41,44,47]. Another group of studies report cautious assumptions of HRQoL improvements [18,22,37] [20,34,49]. Cella et al. and Natale et al. reported on HRQoL improvements after the administration of gefitinib, compared to baseline in heavily pretreated patients and the correlation of these improvements to the tumour response [16,35].
In terms of erlotinib, Lilenbaum et al. could not show significant improvements in progression-free survival, median survival, and HRQoL in comparison to standard chemotherapy [27]. Bezjak et al. showed significant improvements in HRQoL, if erlotinib was given as second line treatment after chemotherapy [11].
Three studies investigated the application of cytostatic agents in elderly patients. Bianco et al. showed improvements in HRQoL for gemcitabine as a single agent therapy [12], Hensing et al. demonstrated that the application of carboplatin/paclitaxel has no significantly different impact on HRQoL between younger (<70 years) and elderly patients [24]. LeCaer et al. showed stable HRQoL values during docetaxel monotherapy [25]. Movsas et al. and Sarna et al. reported no significant HRQoL differences between treatment regimes in combination with or without amifostine [33,40].
A last group of studies showed influences on HRQoL values, e.g. age and baseline quality of life (QoL) [Bozcuk et al.], the cognitive function [Chen], as well as clinical parameters Morita et al. [14,17,32].

Discussion
The present study continues the work of previous reviews like the one of Liu et al. for the indication of lung cancer [3].
HRQoL measurement obtains a twofold meaning in the field of lung cancer medication. This is due to the often severe (toxic) therapeutic side effects, but also because of the high demand for symptom palliation. However, the measurement of HRQoL in respective trials is still not consistent and barely comparable [2][3][4].
By far the most frequently used questionnaire is the EORTC QLQ-C30 in conjunction with the lung cancerspecific module LC13. In comparison, even other lung cancer specific instruments like the FACT-L and the lung cancer symptom scale (LCSS) are only used in relatively few studies [4]. The same applies to the generic instruments. Especially the EQ-5D, which is relevant for health economic evaluations, is rarely used. However, it turns out that, besides the dominant EORTC instruments, a broad portfolio of other questionnaires is applied in different varieties and combinations. This also includes highly sensitive symptom-specific questionnaires. The comparability of these study results thereby is restricted.
A comparison with further literature shows that our results are e.g. in line with Liu et al. and also with Salvo et al. [3,54]. The latter conducted a literature review, published in 2009, searching for quality of life measurement instruments in cancer patients receiving palliative radiotherapy for symptomatic lung cancer. The authors also concluded that EORTC QLQ-C30 was the most commonly used questionnaire (in 13 of 20 trials). Of those 13 studies, 8 also used the LC13 module. The FACT-L was used in only 2 of the included 20 trials. Salvo et al. furthermore considered that HRQoL was of low priority as an endpoint and that measures created for lung cancer patients were underused.
Montazeri et al. identified the EORTC QLQ-C30 and the EORTC QLQ-LC13 as the most sophisticated questionnaires, compared to 50 other instruments to measure the HRQoL in lung cancer patients [55]. This might be one reason for their common use. Another might be the specific symptoms of the lung cancer disease. Moreover, the disease-specific, ordinal scaled profile instruments allow a separate evaluation of therapeutic effects for various dimensions. However, for the comparing of costs and effectiveness in health economic evaluations, an index value is required.
Many of the identified questionnaires allow for the aggregation into one index value, but this is by simply summarising the dimension values. Thus, it is based on an arbitrary weighting of individual aspects of quality of life. Such a value is met with the criticism of being barely interpretable and informative. In addition, the ordinal scaling of the items is inadequate for the economic evaluation of treatment alternatives. Here it is crucial, whether significant improvements in HRQoL can be achieved in the situation of relatively expensive therapy. The EQ-5D is the only identified questionnaire in this study that measures the quality of life in a cardinal index value and thus meets the requirements of cost-utility analyzes. However, as a generic instrument, the EQ-5D is limited by the disadvantage of a relatively low sensitivity in the measurement of small changes in quality of life. This is probably one reason for its relatively rare use in lung cancer studies.
Nevertheless, there are ways in which the identified questionnaires could be used in economic evaluations. One possibility is to transform the most widely used questionnaires EORTC QLQ-C30 and QLQ-LC13 into preference-based versions, similar to the development of the SF-36. This was already considered by the EORTC in the early 90s [50].
In what follows, we highlight some limitations with respect to our research strategy and the results of literature reviews in general. One major limitation of literature reviews is the publication bias. We only considered published full-text study results, no short reports, no conference presentations or study protocols. A comparison with the database clinicaltrials.gov might be a more complete research approach.
Furthermore the restriction of the publication time period might also be a limitation. However, we wanted to analyse the actual situation. In addition, the developments in the fields of HRQoL questionnaires could lead to problems of interpretation.
Our literature search was performed in the Pubmed database only. This is another limitation of our work. Nevertheless, we tried to limit this bias by conducting an additional manual search.

Conclusions
The cancer-specific EORTC QLQ-C30 with the lung cancer-specific LC13 module are the dominant instruments in HRQoL measurement in lung cancer studies. Besides these instruments, a broad spectrum of other questionnaires is used in different varieties and combinations. Only a small number of studies used generic instruments like the EQ-5D.