Abstract
Background/Aim: Typically, prognostic scores predicting survival in patients with metastatic cancer are based on disease- and patient-related factors, such as extent of metastases, age and performance status. Patient-reported symptoms have been included less often. Our group has assessed all patients with the Edmonton Symptom Assessment System (ESAS, a one-sheet questionnaire addressing 11 major symptoms and wellbeing on a numeric scale of 0-10) before palliative radiotherapy (PRT) since 2012. Therefore, we were able to analyze the prognostic impact of baseline ESAS symptom severity. Patients and Methods: We performed a retrospective review of 102 patients treated with PRT between 2012 and 2015. All ESAS items were analyzed by two different methods, dichotomized by median score and by score <4 vs. ≥4. Uni- and multivariable analyses were performed to identify prognostic factors for survival, and from these a 4-tiered score was developed. Results: The most common tumor types were prostate, breast and non-small cell lung cancer, predominantly with distant metastases. Despite differences between the two methods of ESAS data handling, the final multivariable models were strikingly similar. Therefore, the better reproducible cut-off was chosen, i.e. a score ≥4. Multivariable analyses resulted in 4 significant prognostic factors, which contributed equally to the 4-tiered survival score (performance status, more than one cancer diagnosis, progressive disease outside the PRT target volume(s), ESAS appetite). Estimated median survival for different point sums was 24.5 months (0), 8.4 months (1), 4.7 months (2) and 3.0 months (3), p=0.0001. Conclusion: This score identified patients with different survival outcomes, including a good prognostic group with median survival of approximately 2 years. The results may be useful to inform PRT fractionation.
Routine assessment of patient-reported symptoms during the course of palliative cancer treatment provides fundamental information, e.g., regarding effectiveness of interventions, development of side effects and need for specialized palliative care consultation (1-3). For these reasons, the Edmonton symptom assessment system (ESAS) was introduced in our department and employed, e.g., during planning of palliative radiotherapy. ESAS is a short, one-sheet questionnaire addressing major symptoms and wellbeing on a numeric scale of 0-10 (4), which also has been adopted by others (5, 6). Previous research has suggested that patient-reported symptoms might complement survival prediction models, because patients with severe symptom burden had shorter survival than their counterparts with no or minimal symptoms (7, 8). Nevertheless, many prognostic scores rely solely on traditional variables, such as performance status (PS), cancer type and extent of metastases (9-14). We have recently shown (15) that ESAS scores were independent predictors of survival in multivariable analysis in a cohort of patients treated with palliative radiotherapy (n=102). Therefore, we performed a survival update for censored patients from the previous study (n=17) and developed a new prognostic model that also integrates the results of ESAS analyses. Thereby, we wanted to expand our ongoing efforts to develop prognostic models that support decision making for personalized palliative approaches (16-19).
Patients and Methods
We performed a retrospective analysis of 102 hospitalized and out-patients who started palliative radiotherapy at our department during the time period 2012-2015 and included patients with complete and incomplete treatment (5%). Radiotherapy typically consisted of daily 3- or 4-Gy fractions or a single dose of 8 Gy. The ESAS tool was administered by a registered oncology nurse immediately before physician consultation and imaging for treatment planning approximately one week before radiotherapy. The treating physician recorded the patients' medical history and Eastern Cooperative Oncology Group (ECOG) PS. All medical records were available in the hospital's electronic patient record (EPR) system. Statistical analysis was performed with IBM SPSS Statistics 24. In addition to all ESAS items, we analyzed a large number of baseline variables in univariate log-rank tests, where actuarial survival was calculated from the first day of radiotherapy. Those with p-value <0.1 were entered into a multivariable forward conditional Cox regression analysis. Survival was updated October 31, 2017. Eight patients were still alive with a median follow-up of 28 months (minimum 23 months). Date of death was entered in the remaining 94 patients. Based on the results of the multivariable analyses, a 4-tiered survival score was created. For each adverse prognostic feature with p-value <0.05, one point was assigned and the point sum determined the patients' affiliation to one of four prognostic strata. As a retrospective quality of care analysis, no approval from the Regional Committee for Medical and Health Research Ethics (REK Nord) was necessary. Similarly, no approval from the Norwegian Social Science Database (NSD) had to be obtained.
Baseline characteristics before palliative radiotherapy.
ESAS before palliative radiotherapy and its impact on survival.
Results
Most patients had prostate, lung or breast cancer with distant metastases. Table I shows additional baseline characteristics. Median survival was 6 months. The following items were associated with survival (p<0.1 in log-rank test): PS, primary tumor type (longer survival for prostate and breast cancer compared to others), history of more than one cancer diagnosis, lung metastases, pleural effusion and/or metastases, liver metastases, adrenal gland metastases, bone metastases, progressive disease outside the target volume(s), intravenous antibiotics and systemic therapy. The remaining parameters shown in Table I were not associated with survival (p≥0.1 in log-rank test). The ESAS items were analyzed in two different ways (dichotomized by median and proportion with symptom score ≥4). As shown in Table II, the method of analysis influenced the number of variables that were carried forward to multivariable analysis. The ESAS items included in both analyses were pain (not moving), dry mouth, appetite and overall wellbeing. In each case, worse symptoms were associated with shorter survival. All variables with univariate p<0.1 were entered into two different Cox regression models (dichotomized by median and score ≥4), whose results are shown in Table III. Except for appetite, all other ESAS items were not significant in the multivariable analyses. The results were almost identical, regardless of how ESAS data were handled (dichotomized by median or score ≥4).
Results of the multivariable Cox regression analyses, endpoint overall survival.
Based on the 4 significant variables in Table III and an ESAS cut-off ≥4, a prognostic score was calculated. The resulting point sum ranged between 0 (favorable prognosis) and 3 (unfavorable prognosis). None of the patients had a point sum of 4. Estimated median survival for different point sums was 24.5 months (0 points), 8.4 months (1 point), 4.7 months (2 points) and 3.0 months (3 points), p=0.0001 (Figure 1). The 1-year survival rates were 77, 36, 11 and 4%, respectively (2-year: 62, 7, 8, 0%). The best prognostic group included patients with non-small cell lung cancer (NSCLC) and brain metastases, NSCLC and bone metastases, prostate cancer and bone metastases, breast cancer with bone metastases, kidney cancer with bone metastases, kidney cancer with multiple organ metastases, and one 85-year old patient with organ-confined bladder cancer unfit for radical radiotherapy. As required for a point sum of 0, all patients in the best prognostic group had ECOG PS 0-1, only one cancer diagnosis, no progressive lesions outside the radiotherapy target volume(s) and ESAS appetite score 0-3.
Discussion
The present study was performed as an extension and update of a previous one (15) and examined the impact of patient-reported symptoms on survival after palliative radiotherapy. The survival update reduced the number of censored patients from 17 to 8 and increased their median follow-up to 28 months. In univariate analyses dichotomized by median, pain while not moving, dyspnoea, dry mouth, appetite, constipation and overall wellbeing were associated with survival, identical to the previous study. However, as a new finding, we discovered that using an ESAS cut-off ≥4 rather than median changed the results of these univariate analyses. The cut-off ≥4 has also been advocated previously (20) and may be used to triage patients with relatively severe symptom burden to different palliative measures. Compared to the population median, which might vary from one study to another, a fixed cut-off is reproducible. Regardless of how ESAS data were handled, the results of multivariable analyses were similar and suggested that relations between ESAS variables, PS and other baseline characteristics existed. However, the ESAS item appetite was still associated with survival and therefore became part of the 4-tiered survival score. From a scientific point of view, our study confirms previous results regarding the usefulness of patient-reported symptoms as predictors of survival (7, 8). It would be interesting to examine whether or not appetite provides identical prognostic information as weight loss or cachexia.
Actuarial survival based on a prognostic score that included 4 variables. Estimated median survival for different point sums was 24.5 months (0 points, group 0), 8.4 months (1 point, group 1), 4.7 months (2 points, group 2) and 3.0 months (3 points, group 3), p=0.0001 (log-rank test pooled over all strata). Number of patients: 13, 28, 36, 25.
The survival score identified 4 groups of patients with different prognosis (Figure 1). It was striking to see that the best prognostic group survived for a median of approximately 2 years. In all other groups, 2-year survival rates were <10%. Therefore, the score might be applicable mainly to identify patients who should be managed with the most effective strategies available, rather than with low-dose short-course radiotherapy. While the latter often improves symptoms, it is not effective enough in terms of local tumor control (21-23). Given that palliative radiotherapy is not a one-size-fits-all approach, assigning the right patient to the right fractionation regimen is very important (24, 25). As required for a point sum of 0, all patients in the best prognostic group had ECOG PS 0-1, only one cancer diagnosis, no progressive lesions outside the radiotherapy target volume(s) and ESAS appetite score 0-3. We suggest that these patients should be considered for radiotherapy regimens that provide better local control than those with low biologically-equivalent dose.
When interpreting our results, the following drawbacks must be taken into account. In retrospective studies, selection bias and incomplete recording of data might complicate the analyses and their interpretation. Since the study size and consequently statistical power was limited, we allowed prognostic factors with p-value <0.1 in univariate log-rank tests to enter the multivariable model. Thus, larger confirmatory studies are warranted. Our attempt to develop a survival score should be considered a proof of principle rather than a definitive score that already is recommended for clinical use. Our intention is to point out that models, which are solely based on traditional variables such as PS, age and primary tumor type, likely can be refined and strengthened by including additional information.
Conclusion
Reduced appetite and other baseline variables predicted significantly shorter survival. The proposed score identified 4 groups with different prognosis and should be examined in additional independent cohorts. Validated scores may be useful to inform PRT fractionation.
Acknowledgements
None. No funding to disclose.
Footnotes
Conflicts of Interest
The Authors declare that they have no competing interests.
- Received January 5, 2018.
- Revision received January 19, 2018.
- Accepted January 23, 2018.
- Copyright© 2018, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved
References
In this issue
Jump to section
Related Articles
Cited By...
- Symptom Burden in Patients Treated With Palliative Radiotherapy Before and During the COVID-19 Pandemic
- Symptom Burden in Patients With Oligometastases at the Start of Palliative Radiotherapy
- Symptom Burden in Patients With Reduced Performance Status at the Start of Palliative Radiotherapy
- Quality of Life Following Stereotactic Body Radiotherapy Versus Three-Dimensional Conformal Radiotherapy for Vertebral Metastases: Secondary Analysis of an Exploratory Phase II Randomized Trial
- Quality of Life and Radiation-induced Late Toxicity Following Intensity-modulated Versus Three-dimensional Conformal Radiotherapy for Patients with Spinal Bone Metastases: Results of a Randomized Trial