|Home | About | Journals | Submit | Contact Us | Français|
Information is limited regarding health-related quality of life (QOL) status of long term (greater than five years) lung cancer survivors (LTLCS). Obtaining knowledge about their QOL changes over time is a critical step towards improving poor and maintaining good QOL. The primary aim of this study was to conduct a seven-year longitudinal study in survivors of primary lung cancer that identified factors associated with either decline or improvement in QOL over time.
Between 1997 and 2003, 447 LTLCS were identified and followed through 2007 using validated questionnaires; data on overall QOL and specific symptoms were at two periods: short-term (less than three years) and long-term post diagnosis. The main analyses were of clinically significant changes (greater than 10%) and factors associated with overall QOL and symptom burden for each period and for changes over time.
Three hundred two (68%) underwent surgical resection only and 122 (27%) received surgical resection and radiation/chemotherapy. Recurrent or new lung malignancies were observed in 84 (19%) survivors. Significant decline or improvement in overall QOL over time were reported in 155 (35%) and 67 (15%) of 447 survivors, respectively. Among the 155 whose QOL declined, significantly worsened symptoms were fatigue (69%), pain (59%), dyspnea (58%), depressed appetite (49%), and coughing (42%). The symptom burden did not lessen among the 67 who reported improvement, suggesting survivors had adapted to their compromised physical condition.
LTLCS suffered substantial symptom burden that significantly impaired their QOL, indicating a need for targeted interventions to alleviate their symptoms.
Individuals who are alive more than five years after lung cancer diagnosis are referred to as long term lung cancer survivors (LTLCS).(1) In the U.S., approximately 26,000 individuals become LTLCS annually. According to the published literature regarding health-related quality of life (QOL), LTLCS do not experience the same QOL as their age-matched survivors of other cancers; they scored the lowest in health care utility among long term survivors of all cancers; and one in every four LTLCS is significantly restricted in physical ability or suffering from depressive symptoms.(1) Reasons for LTLCS’ degraded QOL are poorly characterized, particularly their persistently compromised lung capacity, treatment-related morbidity, and heavy symptom burden. Unfortunately, changes in QOL over time or the characteristics associated with these changes have rarely been evaluated in LTLCS. This study was designed to identify potential targets for future interventions that could be tailored to improve poor and maintain positive QOL among LTLCS with specific aims to characterize QOL over a 7-year period, to describe QOL changes over time, and to identify factors associated with progressively compromised or enhanced QOL.
The primary objectives were (1) to describe the overall QOL and main symptom burden of LTLCS and to examine changes overtime; and (2) to characterize QOL and symptom burden at two points periods, i.e., within 3 years and beyond 5 years post primary lung cancer diagnosis. Secondary objectives included (1) to identify factors associated with declining QOL measures over time; and (2) to evaluate the potential impact of missing data.
Since 1997, all patients with a pathologic diagnosis of primary lung cancer evaluated and treated at Mayo Clinic, Rochester, Minnesota, have been prospectively enrolled and followed for outcome research, using protocols approved by the Mayo Clinic Institutional Review Board, and all human participants have written informed consent.(2) Between January 1997 and May 2003, 6,252 new primary lung cancer cases were identified and pathologically confirmed. The follow-up process started within six months after diagnosis and then annually until patients’ death. More details have been described in our previous publications.(3, 4) By December 2007, 869 patients (14%) were verified as LTLCS who returned at least one follow-up questionnaire (any time post diagnosis). Six hundred sixty-nine (77.0%) answered a QOL questionnaire at least once beyond five years, and 447 (66.8%) have answered the questionnaire at least twice, one within three years and another beyond five years post lung cancer diagnosis, forming the basis for all study results.
QOL measures were chosen based on their relevance to general health and QOL, psychometric properties among cancer survivors, ease of administration, and low respondent burden. The Lung Cancer Symptom Scale (LCSS) is a disease- and site-specific symptom burden assessment focusing on the symptom and function of patients with lung cancer.(5, 6) The LCSS consists of two parts, only the patient scale was used in this study. The patient scales consist of nine items and are worded (and abbreviated in the analysis) as follows: (1) How is your appetite (appetite)? (2) How much fatigue do you have (fatigue)? (3) How much coughing do you have (cough)? (4) How much shortness of breath do you have (dyspnea)? (5) How much blood do you see in your sputum (hemoptysis)? (6) How much pain do you have (pain)? (7) How bad are your symptoms from lung cancer (lung cancer symptom distress)? (8) How much has your illness affected your ability to carry out normal activities (illness affecting normal activities)? (9) How would you rate your quality of life today (overall QOL)? The intensity of patient responses is measured by visual analog scales.(6, 7) In an earlier study, internal consistency was high (Cronbach α = 0.89).(8)
Information abstracted from participants’ medical records included demographics, lung cancer histology, staging, anatomical location, and treatment. Tobacco history information included current or previous use, duration, average amount of cigarettes smoked per day, and years since the participant quit smoking. For participants who received medical care outside Mayo Clinic, copies of the relevant medical records were acquired and reviewed. Further details regarding pathology and treatment classifications have been previously published.(2, 3, 9) To avoid misclassification, we reported the occurrence of disease progression, recurrent disease, and second primary lung cancer together.(4)
Overall QOL (item 9 of LCSS) and symptoms (items 1–8) were assessed as scales varying from 0 (worst) to 100 (best) and as a binary variable. Normative data for overall QOL in healthy individuals establish 80 as the mean and 20 as the standard deviation (SD); one-half SD has been conventionally used to define the normative range, that is, mean ± ½ SD.(10–14) A clinically important decline or improvement is defined as a greater than 10-point or 10% change; poor QOL was defined as less than 50 on the 100-point scale.(15) Profile analysis to model the longitudinal trend over time via repeated measures analysis of variance was used.(16)
Overall QOL was also assessed at two time points by non-parametric tests: short-term (within three years of diagnosis) and long-term (over five years post diagnosis) follow-up, for each time period and for changes over the two. A clinically important decline or improvement was tested by chi-square or Fisher’s exact tests.(17) Rank transformed QOL scores were used but remarkably consistent with the results obtained by using the original scores; therefore, models using original scores were presented. Both multiple logistic and linear regression models were developed to determine which covariates were associated with poor (binary) and worse (a continous variable) overall QOL, respectively, via a stepwise procedure for covariate selection. For secondary objectives, comparing patients’ average QOL between short-term and long-term responses, a standard Wilcoxon test has 80% power to detect a difference of 0.25 standard deviation based on 447 patients. Since the extent of surgical resection, e.g., lobectomy versus pneumonectomy may impact patients‘ QOL measures, we have evaluated a subset of patients who had only surgery as treatment.
We compared the following three groups: patients who responded at both time points (n=447), patients who only answered a short-term or a long-term (n=422) questionnaire, and patients who never responded (n=309), with regard to the following variables: age at diagnosis, sex, race/ethnicity, education, smoking status, TNM stage, histology, and primary treatment. Supplementary Table S1 provides basic descriptive information of the three subgroups of survivors. The impact of these differing characteristics on QOL results were evaluated.
Among the 447 LTLCS, the mean age at the time of initial lung cancer diagnosis was 65 years in men (n=223) and 63 years in women (n=224), p=0.01. Table 1 presents more detailed demographic information by gender with regard to race and ethnicity (greater than 90% of the total were Caucasian), education level (57% had greater than 12 years of education), marital status (79% married), employment (42% employed) status, and cigarette smoke exposure history including second-hand smoking (82% former or current smokers). Additional information regarding tumor histology, lung cancer stage, comorbidities, and treatment are provided in Supplementary Tables S2 and S3. Recurrent lung cancer or new lung malignancies (disease-bearing) were observed in 84 (19%) survivors (Supplementary Table S2). Three hundred two patients (68%) undertook surgical resection only, 122 (27%) received combinational treatment (surgery plus radiation or chemotherapy), and 23 (5%) had non-surgical treatment (Supplementary Table S3).
The mean scores and standard deviations (SD) at seven time points, i.e., 1 to 7 years post lung cancer diagnosis and the trend for each scale, varied from 81±20 (year-2) to 75±21 (year-7). After adjusting for multiple factors listed in Table 1 and Supplementary Tables S2–S3, the mean scores remained virtually the same for all scales, but the SD was reduced by 55–67% (data not shown). Symptom subscales that consistently scored low throughout the seven years were fatigue (scores varying 57–67) and dyspnea (59–69); subscales that consistently scored higher than 80 were lung cancer symptom distress (86–88) and hemoptysis in particular (95–97). As shown in Figure 1, changing mean overall QOL from 81 in year-2 to 75 in year-6 did not reach the level of clinically important decline; however, specific scales that met clinically important decline over time included pain, fatigue, and dyspnea: Particularly, fatigue and dyspnea dropped to under 60, a level indicative of serious impairment.
In order to assess the impact on QOL by the extent of surgical resection of primary lung cancer, mean scores of overall and specific symptom scales were compared between patients who received lobectomy (73.5%) and who received more extensive resections (15.6%, bilobectomy, pneumonectomy, completion pneumonectomy, and sleeve lobectomy). We observed somewhat worse mean scores for several scales, particularly fatigue, coughing, and short of breath; however, none were over 10-point of 10% difference.
The clinically important decline and improvement in overall QOL over time were reported in 35% and 15%, respectively, of the 447 survivors (Supplementary Figure S1A). No striking difference was observed between the 363 disease-free survivors and the 84 disease-bearing survivors regarding the proportions with clinically important change over time (Supplementary Figures S1B-S1C). Table 2 presents change over time of overall QOL and the subscales among all 447 survivors. The clinically important decline in scores was primarily confined within the patient’s assessment of symptoms; specifically, we observed 15 points worse in fatigue (in 47% of survivors), 14 points worse in pain (43%), and 12 points worse in dyspnea (44%).
Among the 155 LTLCS who reported a clinically important decline in overall QOL, five symptom scales dropped in larger magnitude and higher percentage: fatigue (24-point decline in 69% of the 155), pain (23-point decline in 59%), dyspnea (20-point decline in 58%), appetite (11-point decline in 49%), and coughing (12-point decline in 42%); six of the eight subscales had 10 points or more decline. In sharp contrast, among the 67 survivors who reported a significant improvement (by 19-point increase) in overall QOL, no specific symptom or functional domain increased by greater than 10 points; moreover, the six symptom subscales had minimal or no increase or even numerically declined (pain, coughing, and dyspnea). Figure 2 illustrates the contrasting changes over time of subscales among individuals with a clinically important decline in overall QOL (left panel) and individuals with a clinically important improvement in overall QOL.
Participants’ characteristics listed in Table 1 and Supplementary Tables S2–S3 were included in the analyses for both follow-up periods, side by side (Table 3). Under a multiple logistic regression model, factors significantly associated with a poor overall QOL at short-term follow-up were male gender, squamous cell carcinoma, and poor fatigue and pain scores; at long-term follow-up they were older age, lung cancer progression or recurrence, and poor scores in fatigue, dyspnea, and pain. Similar results were found under the alternative multiple linear regression models (Supplementary Table S4).
The same analyses were performed to identify patients who had significant change over time: contributing factors to a clinically significant decline in overall QOL scores were more than 16 years of education and clinically significant declines in fatigue and pain (Supplementary Table S5, middle column); whereas, additional factors associated with an improved overall QOL (Supplementary Table S5, right column) were sex, co-morbid conditions of non-lung diseases (e.g., cardiovascular, cerebral, other malignant diseases, and diabetes), and a clinically significant improvement in fatigue, dyspnea, and pain scales.
We have reported four factors that were significantly different between the 447 study participants and the potential study subjects who were eligible but were not included due to non-participation or missing data (Supplementary Table S1). Specifically, the study sample was enriched with highly-educated never smokers who had resected adenocarcinoma or carcinoid or salivary gland tumors, all with better overall QOL scores. The impact of these four factors on overall scores of QOL (Supplementary Table S6) demonstrated a 4–12 point difference across subgroups, no matter if the overall QOL was taken at the first response or an average of all responses for each subject. Given the characteristics of sudy participants who in general reported a better overall QOL, the eligible subjects who were not included in this study were likely to have an unfavorable overall QOL; therefore, our results may under-report the true magnitude of compromised QOL in LTCLS.
Health related quality of life has been identified by cancer survivors to be as important as survival length (quantity).(18, 19) This study is the first to prospectively characterize QOL and symptom burden in a large and recent cohort of LTLCS (N=447) who were alive beyond five years.(1) Based on self-reported QOL assessments between two time periods, significant decline in overall QOL was observed in 35% of the LTLCS, and significant improvement in overall QOL was observed in only 15% of the survivors. Interestingly, a similar percentage of decline was observed in both disease-free and disease-bearing survivors, suggesting that cancer recurrence and progression are not the only factors that impact QOL in LTLCS. Among the survivors with degraded overall QOL, almost all specific symptom scales exceeded a 10% reduction or clinically important decline: pain, fatigue, coughing, dyspnea, and appetite. Therefore, QOL of long-term lung cancer survivors showed substantial deficits relative to other cancer survivor populations, as evidenced by a concise overview below.
Investigations involving many tumor types have identified how factors unrelated to the disease can impact cancer survivors’ overall QOL, particularly surrounding the time of treatment and shortly thereafter.(20–26) There were a small number of tumor types where survivors reported improved QOL through finding meaning and purpose in life since being diagnosed with cancer. Specifically, long-term survivors of breast cancer or lymphoma reported comparable or better QOL than their healthy age-matched controls, which has been summarized in multiple studies involving approximately 4,000 adult cancer survivors.(27) Ferrell and colleagues(20) studied 687 cancer survivors and reported that all QOL components were positively impacted. Ashing-Giwa and colleagues(21) reported findings from 278 breast cancer survivors, showing comparable or better QOL than healthy age-matched controls. Ganz and colleagues(22) also reported high levels of functioning and QOL in 914 breast cancer survivors. Mosconi and colleagues(23) demonstrated that the QOL among 1,772 long-term breast and colon cancer survivors appears to be the same as healthy populations. For breast cancer patients who have a successful outcome in terms of reconstructive surgery, they are virtually indistinguishable from healthy people.
In contrast, outcome research in lung cancer has been focused mainly on short-term survival and QOL. The lack of published LTLC survivorship research has only recently been recognized as an important deficit. A cross-sectional study of QOL in LTLCS has shown that objective measures of demographic and clinical factors alone can not explain a large amount of variance in the total QOL scores, while self-reported distressed mood was the most important predictor of QOL.(24) A prospective longitudinal approach is required to consider the changes and effectively evaluate the multiple factors that affect QOL in LTLCS, which was the design and aims of the current study. We observed that the overall QOL score within three years of diagnosis was around 80, which is equivalent to healthy populations,(25) then decreased to 75 after four years post diagnosis. This is consistent with the concept that even among the best performers among LTLCS, QOL declines over time and warrants more attention. Symptoms related to higher death rates were observed during years 1–4 after diagnosis;(28, 29) our current study results highlighted the deteriorating QOL associated with these symptoms. Specifically, our results show that most symptom scales declined over time except hemoptysis, lung cancer symptom distress, and illness affecting normal activities. This is consistent with the concept that symptoms related to treatment dissipate over time; however, other symptoms particularly fatigue and dyspnea dropped to below 60, indicative of serious impairment and demands further understanding the physical and biological mechanisms, which in turn could lead to targeted interventions to improve QOL. Such targeted interventions, which are to be identified and evaluated, should start early in the course of surviving lung cancer, as evidenced by our independent analyses of symptom clusters within 5 years post diagnosis).(28, 29)
Noted were the small subset of survivors (15%) in our study who reported an improved QOL overtime. However, when looking beyond the overall QOL and into the specific symptom scales, coughing, dyspnea, and pain actually decreased 4, 3, and 2 points, respectively. This paradoxical phenomenon suggests that QOL of survivors may be affected by a myriad of factors, which may reflect differently in a subjective evaluation of QOL and on observed health status. Discrepancies seen in measures of self-reported versus observed levels of physical functioning and objective health status or well being has led to the recognition of response shift, i.e., changes in internal standard of reference, values, and reconceptualization of the target construct in QOL studies.(26, 27)
Our study also found that surgical resections that are more extensive than lobectomy, such as bi-lobectomy and pneumonectomy, may negatively impact patients’ symptom burdens and QOL; however, the negative impact from major resections, although on average not reaching the threshold of clinical significance, were present in both short-term and long-term follow-up periods.
Another inherent drawback of QOL studies is survival bias, coupled with non-participation bias. These biases are encountered in a study such as this one due to the severity of the illness (i.e., lung cancer in our study). Missing data and non-participation may affect the validity of the study results. A critical issue is whether respondents and non-respondents differ systematically and whether investigators are fully aware of the potential biases. In our study, we have carefully evaluated chararteristics (Supplementary Table S1) and identified the magnitude of point changes in QOL associated with each potential confounding factor (Supplementary Table S6). A consensus was a potential underestimation of compromised QOL among LTLCS because missing data and non-participation was significantly over-represented in subgroups with more severe disease and higher mortality. This non random missing phenomenon has been well-addressed in the previous studies, either focusing on symptom patterns during the first year after lung cancer diangosis,(30–33), or in evaluating symptom clusters up to 5 years post diagnosis.(28, 29).
Although the LCSS is a well-recognized QOL tool, other scales have been available; in particular, specific scales for anxiety or depression such as HADS-D, HADS-A, PHQ-9 and CES-D (34–36) might have been useful to more precisely capture the emotional health of these patients and better discriminate the mental health status of disease-free survivors from disease-bearing survivors. These more mood focused psycho-social tools have been recently evaluated in subgroups of patients and data should be mature in analyses.
To conclude, self-reported QOL of LTLCS showed substantial deficits; affected domains were mainly in symptom burden, indicating a demand for targeted interventions for improved QOL. There is a need to identify and intervene with subgroups of LTLCS who are at an elevated risk of diminished QOL soon after receiving a lung cancer diagnosis and undergoing cancer treatment. Further understanding is needed as to the mechanisms of progression of symptoms and loss of function. Accurate knowledge including risk estimation for medical events is critical to the long-term management of lung cancer survivors as well as to evaluate treatment strategies.
Supplementary Figure 1A. Change in overall QOL over time for 447 long-term survivors
Supplementary Figure 1B. Change in overall QOL over time for 363 disease-free survivors
Supplementary Figure 1C. Change in overall QOL over time for 84 disease-bearing survivors
Supplementary Table 1. Comparison of characteristics of study participants and non-participants
Supplementary Table 2. Lung cancer features and comorbidities of 447 long-term survivors
Supplementary Table 3. Lung cancer treatment modalities of 447 long-term survivors
Supplementary Table 4. Factors associated with overall QOL at two time points (linear regression).
Supplementary Table 5. Factors associated with a change over time in overall QOL (logistic regression)
Supplementary Table 6. Comparison of characteristics that potentially bias overall QOL
This work was supported by research grants CA77118, CA80127, CA84354, and CA115857 from the U.S. National Institutes of Health.
We thank Susan Ernst, M.A., (Mayo Clinic) for technical support in the manuscript preparation and Mary E. Johnson (Mayo Clinic) for her input on various stages of this work.