The association of quality of life with potentially remediable disruptions of circadian sleep/activity rhythms in patients with advanced lung cancer

Patients participating in this investigation were recruited for a clinical trial that was conducted concurrently at Cancer Treatment Centers of America^® (CTCA) at Midwestern Regional Medical Center (MRMC) in Zion, Illinois and the WJB Dorn Veterans Medical Center (VAMC) in Columbia, South Carolina from June 2002 to April 2006. Patients between the ages of 18 and 80 who had a pathologically confirmed diagnosis of stage III or IV non-small cell lung cancer (NSCLC) and an Eastern Cooperative Oncology Group (ECOG) performance status of 0, 1, or 2 were eligible. Untreated patients and patients who had failed one prior chemotherapy treatment regimen were eligible. All patients signed an Informed Consent indicating that they were aware of the investigational nature of the study. The Institutional Review Boards at MRMC and VAMC approved the study. This current report is based only upon baseline data obtained at initial enrollment.

Our control database for actigraphic parameters is comprised of 3- to 7-day actigraphy measurements from 35 adults, aged 20 to 50 years, each having no known disease (Action-W2 database, AMI, NY, USA). EORTC QLQ-C30 Reference Population data is from a survey of 1,313 newly diagnosed lung cancer patients with stage III and IV disease,³² while the EORTC QLQ-C30 General Population data is from a survey of 1,965 randomly selected subjects in Norway aged eighteen to ninety-three years [33]. Ferrans/Powers QLI Health/Functioning General Population data is taken from 339 subjects, drawn randomly from a telephone directory representing urban, suburban, and rural areas in the Midwestern United States, who completed the QLI questionnaire (C. Ferrans, Feb 2009).

All patients agreed to complete their baseline quality of life questionnaires before undergoing their initial chemotherapy treatment cycle for this trial. At MRMC, actigraphy was performed in the inpatient setting immediately before (24-48 hours) and during their first chemotherapy cycle. VAMC actigraphy was obtained in the patients' domestic setting, prior to their initial cycle of chemotherapy.

The Ferrans and Powers Quality of Life Index (QLI) differs from the EORTC QLQ-C30 in that it elicits information quantifying each patient's satisfaction or dissatisfaction with various aspects of life and thus assesses the individual's feelings of emotional well-being and capacity to enjoy life. Consequently, this instrument evaluates how cancer has affected, both positively and negatively, the patient's well being in terms of their own personal values [22]. The QLI Cancer Version III consists of two parts. Part one requires the patient to describe satisfaction levels with 33 aspects of life. Part two requires the patient to rate the importance of the same 33 aspects of his or her life. The QLI produces an overall quality of life score and subscale scores on four specific domains: health and functioning, social and economic, psychological and spiritual, and family. The scores range from 0 to 30, with higher scores indicating greater satisfaction with life. Thus, the EORTC-QLQ C30 and the QLI provide distinct, unique, and complementary insights into different aspects of each patient's quality of life [34].

A watch-like wrist actigraph, worn on the non-dominant wrist, was used to record a patient's movement patterns continuously over a four to seven day span (Ambulatory Monitoring, Inc, AMI, Ardsley, NY USA). Internal motion sensors (accelerometers) capture patient movement data, measured as the number of accelerations per minute. These data are transferred to a computer for analysis to produce a report containing parameters of activity during each sleep and wake period, detailing their extent, timing, duration and other characteristics [35]. Actigraphy data were recorded in the hospital setting at MRMC in the four to seven days immediately prior to and during the first treatment, while these data were obtained in the patient's home for VAMC patients during the week prior to the first treatment.

Chronic Obstructive Pulmonary Disease (COPD) is common among lung cancer patients and, thereby, a potential confounding variable for this investigation of quality of life outcomes and circadian time structure. All VA patients were assessed clinically and with pulmonary functions for the presence of COPD. Its severity was graded according to the Spirometric Classification of COPD severity, by reference to percent of predicted forced expiratory volume in one second (FEV₁). Thirty to fifty percent of predicted FEV₁ is considered severe; moderate is 50 to 80%; and mild COPD is greater than 80% of predicted FEV_1. No such data were available for inpatients.

Descriptive statistics such as mean and standard error were computed for continuous variables and actigraphy endpoints to describe the average and variability of values across the population. Frequency and percentages were computed for qualitative factors such as sex and quality of life outcomes. Both parametric (Analysis of Variance) and non-parametric analyses were used to determine time of day differences among factor levels (SAS v 9.1, Cary, NC) and their shapes across (circadian) time of day. For four to seven days, an actigraphy watch recorded the number of accelerations per minute. These data were translated into polysomnographically-validated sleep/activity parameters through the Act Millenium and Action W2 software (Ambulatory Monitoring, Inc). Cosinor analysis was done on these sleep/activity patterns in order to assess the circadian characteristics of sleep and activity among these lung cancer patients. Cosinor analysis provides three standard parameters: mesor - the average activity over the 24-hr period, amplitude - peak to nadir difference and acrophase - the time of daily peak activity. We also computed measures of circadian rhythm robustness and day-to-day stability (circadian quotient = amplitude/mesor), peak activity (mesor + amplitude) and 24 hour autocorrelation for each patient. We further measured circadian indices of the daily timing of sleep and activity, such as the ratio of amount of activity during the day to nighttime activity, night/day sleep balance that is the amount of nighttime sleep relative to daytime sleep. Inpatient and outpatient differences were tested using t-test for quantitative characteristics while chi-square was used for categorical factors. The relationships between quality of life domain values and actigraphy measurements were measured using the Spearman rank correlation test.

84 patients were enrolled and consented at two sites. One site studied patients only in the outpatient home environment while the second site studied them only as inpatients. There were differences in demographic and clinical status in participants by site. All 42 outpatients were males while 19 of 42 inpatients were female. Outpatients were older on average with a mean age of 66 compared to the inpatients' mean age of 57 years (Table 1). More inpatients had Stage IV disease (74.4%) than outpatients (47.6%).

50% and 26% of inpatients and outpatients, respectively, were experiencing recurrence of disease after first line systemic treatment failure. The median time between terminating first line therapy and the administration of actigraphy in the ten outpatients was 51.6 months (range 7.9 to 78 months), but for inpatients it was 3.5 months (0.8 to 33.5 months).

Although all patients agreed to wear the actigraphs, twelve actigraphs were worn for less than 48 hours and/or had frequent missing observations due to instrument malfunction. Four patients failed to respond fully to the questionnaires, so we have full actigraphy and full questionnaire data for 68 of these 84 patients.

Both inpatients and outpatients reported lower scores for all EORTC-QLQ-C30 functional domains when compared to population-based controls (Table 2). The differences in domain scores between these lung cancer patients and population-based controls were large, ranging from 16 to nearly 30 points. Both inpatients and outpatients reported scores that were 20 to 30 points higher (worse) than the population based controls in the symptom items of fatigue, loss of appetite, insomnia, and pain. Our advanced lung cancer patients, as expected, also scored 10 to 20 points higher (worse) on these symptom items than the EORTC reference population of newly diagnosed lung cancer patients with stage III and IV disease. Despite various demographic and clinical differences between inpatients and outpatients, only two of 20 symptom categories, fatigue (p = 0.0187) and shortness of breath (p < 0.001), achieved statistical significance between our two patient groups, indicating the profound leveling effect of advanced non-small cell lung cancer vis a vis their most common and devastating symptoms.

Our patients' mean score of 16.23 in the Ferrans/Powers QLI health/functioning domain is nearly eight points lower (worse) than the population-based mean score. This is some two standard deviations below mean scores from a general population survey. Patients were very dissatisfied with their health. Interestingly, however, the other three QLI domains (social/economic, psychological/spiritual well being, and family) did not differ at all from the population-based scores (Table 3).

Actigraphic parameters of lung cancer patients were found to be grossly abnormal when compared to the healthy control population. Healthy individuals show a more robust 24-hour activity rhythm while lung cancer patients show a flatter curve (Figure 1). Cancer patients were 30% less active and napped at least three times longer than the controls during the putative wake/activity period (Table 4). During the nightly sleep span, lung cancer patients had significantly more and longer waking episodes than controls (Table 4). No gender differences were found.

Actigraphy characterizing the putative daily wake span and the overall circadian organization differed by site (Table 4, Figure 1). Outpatients were more active during the day and consolidated activity and sleep better compared to hospitalized inpatients. The sleep phase actigraphy parameters at both sites, however, were indistinguishable. These prominent site specific differences in actigraphy collection protocol required that all such data be analyzed by site. The most obvious ubiquitous and severe differences across the groups of normal controls and lung cancer patients include uniformly poor sleep among all lung cancer patients and decreasing daytime activity which is most severe, as might be expected, among hospitalized patients.

Among inpatients we only found several statistically significant associations with actigraphy parameters. We found that the QLI Health/Functioning domain scores correlate with the 24-hour activity autocorrelation, which measures day to day stability of the peak and trough timings of the daily activity/sleep pattern (r = 0.34, p = 0.05). Inpatients with the most stable circadian time structures of the daily activity pattern express the highest level of satisfaction with their health. Those who are most disrupted have much worse satisfaction with their health. Two EORTC symptom items correlated strongly with activity circadian rhythm parameters; insomnia severity correlates negatively with 24-hour autocorrelation, the day-to-day reproducibility of peak and trough activity (r = -0.48, p = 0.003), and loss of appetite correlates with decreased peak daily activity (r = -0.41, p = 0.005) and Circadian Quotient, a measure of the strength of the circadian rhythm compared to noise or other high frequency rhythms (r = 0.4, p = 0.015). Circadian activity/sleep rhythm disruption is reflected in or caused by nocturnal insomnia and those patients with most decreased daily activity (ie. greatest daytime fatigue) have the poorest appetite, even in the hospital setting.

Among ambulatory outpatients with advanced lung cancer in their own homes, higher levels of daytime activity, as measured by the parameter "peak activity," is significantly associated with each of the five Power and Ferrans QLI domains: health/functioning(r = 0.51, p < 0.01), social/economic (r = 0.38, p = 0.048), psychological/spiritual (r = 0.45, p = 0.02), family (r = 0.45, p = 0.02), and overall QLI (r = 0.57, p < 0.01), see Figure 2 and Table 5. Robustness (largest peak to trough difference in daily activity or amplitude) of objectively measured daytime activity is reflective of all measured aspects of quality of life in the face of advanced lung cancer. The health/functioning domain also demonstrates a statistically significant positive relationship with 24-hour autocorrelation, or day-to-day circadian stability of peak and trough activity timings (r = 0.45, p = 0.02). Each Ferrans-Powers QLI domain exhibits a statistically significant relationship with the actigraphy parameter night-day sleep balance and quality of life scores in health/functioning (r = 0.39, p = 0.04), social/economic (r = 0.40, p < 0.04), psychological/spiritual (r = 0.45, p = 0.02), and family (r = 0.33, p = 0.10). The more robust the day/night difference between nocturnal and daytime activity levels, the better the patients score by each and every QLI measure.

Statistically significant correlations between EORTC QLQ C30 domains among outpatients were found. Global health exhibits a statistically significant positive association with circadian phase stability as reflected by the 24-hour autocorrelation (r = 0.53, p < 0.01), (see figure 2). The actigraphy parameter, night-day balance of time spent asleep and awake, shows a statistically significant association with the EORTC QLQ C30 domains role (r = 0.56, p < 0.01) and cognitive function (r = 0.45, p = 0.02).

Outpatient fatigue levels are associated with diminished robustness of the circadian quotient (r = -0.40, p = 0.04), rhythm quotient (r = -0.41, p = 0.03) and night-day balance of time spent asleep (r = -0.52, p < -.01). The more quantitatively robust the day-night activity/sleep measurement differences, the less fatigue these patients experience during each day. Rhythm quotient is also inversely associated with pain (r = -0.39, p = 0.04). The greater the pain, the less robust were day-night activity differences, objectively verifying that pain interferes with both sleep at night and activity during each day. Loss of appetite is likewise associated negatively with the actigraphy parameters night-day sleep balance (r = -0.47, p < 0.01). The more disordered the circadian organization, the more profound the cancer-associated loss of appetite.

Unlike the analysis for quality of life outcomes, there were no statistically significant associations found between actigraphy data and chronic COPD severity in this patient population with symptomatic advanced lung cancer.

Figure 3 shows the relationships among two measures of quality of life and circadian organization parameters. A robust circadian rhythm in activity/rest is associated with greater patient satisfaction with health/functioning, less fatigue, and better overall quality of life. These relationships, though present among all patients, are especially clear among outpatients where hospitalization does not mask usual circadian time structure.