Abstract
Background/Aim: For treatment of brain metastases, a patient's survival prognosis should be considered. Existing survival scores appear complex and require complete tumor staging. For many patients, a faster and simpler tool would be helpful. Patients and Methods: This retrospective study investigated the prognostic value of the number of pre-treatment symptoms plus eight other factors on survival of patients irradiated for brain metastases. Other factors included whole-brain radiotherapy (WBRT) regimen, age, gender, performance score, primary tumor type, number of brain metastases, extracranial metastases, and interval between cancer diagnosis and WBRT. Results: The number of symptoms (p=0.002) and all other factors were significantly associated with survival on univariate analyses. On multivariate analysis, all factors but the number of symptoms (p=0.47) and primary tumor type (p=0.48) were significant. Conclusion: Since the number of symptoms was not an independent predictor of survival, it cannot replace existing scoring tools and may only serve for orientation.
Whole-brain radiotherapy (WBRT), either alone or in combination with local therapies, is the most common treatment for brain metastases worldwide (1). For patients with multiple cerebral lesions and those with a limited performance status or severe comorbidities, WBRT alone is frequently used. Common WBRT regimens include 5×4 Gy over 1 week, 10×3 Gy over 2 weeks and 20×2 Gy over 4 weeks (1). A shorter program is preferred for patients with poor survival prognoses in order to give them as much time as possible with their family and friends, and enable them to get important things done before they die. Administration of a short course of WBRT for these patients is supported by the fact that 5×4 Gy was shown not to be inferior to 10×3 Gy regarding survival and intracerebral control in patients with multiple brain metastases (2). In contrast, patients with more favorable survival prognoses benefit from longer-course WBRT including total doses beyond 30 Gy and doses per faction of less than 3 Gy in terms of improved outcomes (survival and intracerebral control) with less neurocognitive decline (3, 4). Thus, prediction of a patient's remaining lifetime appears crucial for selecting the optimal treatment. This aspect is already well-recognized and scoring instruments facilitating the prediction of individual survival prognoses have been developed. Major instruments for patients receiving WBRT include recursive partitioning analysis (RPA) classification, graded prognostic assessment (GPA) classification, Rades Score and WBRT-30 score (5-8). All of these tools were based on four prognostic factors, including the presence or not of extracranial metastases. To use these instruments properly, i.e. to be able to state whether extracranial metastases are present, complete tumor staging is required, but often not possible. Complete staging can be time consuming and would delay the start of treatment. Moreover, it may be burdensome, particularly for patients with poor performance status and significant comorbidities. Thus, particularly for this latter group, a very simple scoring tool would be desirable for estimating their survival prognoses quickly and without significant effort. Such a tool could be based on the symptoms caused by the brain metastases, since patient history and physical examination are mandatory and available for all cases. The present study investigated the potential prognostic value of the number of pre-treatment symptoms for survival in a large cohort of patients irradiated for brain metastases. It was assumed that fewer symptoms were associated with a more favorable survival prognosis.
Patients and Methods
In a retrospective cohort study of patients who were treated with WBRT alone for brain metastases between 1994 and 2017, the number of pre-treatment symptoms (prior to WBRT and administration of corticosteroids) was evaluated for a potential association with survival. Data of patients obtained from an existing anonymized database were supplemented by data from patients treated at two German institutions between 2010 and 2017. To be eligible for this study, patients must have received WBRT alone and clinical symptoms must have been recorded in the patient files.
Symptoms included seizures, dysarthria, motor deficits, sensory deficits, neurocognitive deficits, headache, nausea/vomiting, vertigo, vision disturbance, and hearing problems. The number of symptoms was grouped as 0, 1, 2, or 3 or more symptoms. In addition to the number of symptoms, eight other factors were analyzed for survival including dose-fractionation of WBRT (5×4 Gy vs. 10×3 Gy vs. 20×2 Gy), age (≤62 vs. ≥63 years, median=63 years), gender, Eastern Cooperative Oncology Group performance score (ECOG-PS) (0-1 vs. 2 vs. 3-4), primary tumor type [breast cancer vs. non-small cell lung cancer vs. small cell lung cancer vs. renal cell carcinoma vs. malignant melanoma vs. cancer of unknown primary (CUP) vs. gastrointestinal cancers vs. other tumors], number of brain metastases (1 vs. 2-3 vs. ≥4), extracranial metastases (no vs. yes) and interval between cancer diagnosis and WBRT (≤6 vs. ≥7 months, median=7 months). Distributions of these factors are shown in Table I.
It was planned to follow-up patients until death or for at least 6 months. They were censored at the time of death or last follow-up visit. The study was approved by the Ethics Committee of the University of Lübeck (reference: 19-103A). Univariate analyses were performed with the Kaplan–Meier method supplemented by the log-rank test. The Bonferroni correction for multiple comparisons was used to adjust the p-values derived from the univariate analysis. Since nine factors were investigated, p-values of less than 0.0056 were considered significant representing an alpha-level of less than 0.05. Significant factors were analyzed for independence and to minimize the influence of confounding variables in a multivariate analysis (Cox proportional hazards model).
Results
A total of 1,907 patients, 1,354 from an existing anonymized database and 553 additional patients, met the eligibility criteria. On univariate analysis (Table II), improved survival was significantly associated with a lower number of pre-treatment symptoms (p=0.002). Since survival rates of patients with 0, 1 and 2 symptoms were similar, they were subsequently combined into one group. Survival rates of patients with 0-2 symptoms at 3, 6, 9 and 12 months were 50%, 32%, 24% and 18%, respectively, and significantly better than in patients with 3 or more symptoms (p<0.001).
On multivariate analysis, the number of pre-treatment symptoms (p=0.82) was not a significant factor (Table III). In an additional multivariate analysis, the different grouping of pre-treatment symptoms (0-2 vs. ≥3) did also not achieve significance (hazard ratio=1.02, 95% confidence interval=0.95-1.10, p=0.54).
Distribution of number of pre-treatment symptoms and other investigated factors.
Discussion
Despite recent developments favoring the treatment with neurosurgery or radiosurgery alone, a considerable number of patients with brain metastases are still candidates for WBRT (1). When aiming to provide an individual treatment including the WBRT regimen appropriately addressing the patient's personal situation and needs, the patient's remaining lifetime should be regarded.
For most patients with a short survival time, short-course WBRT with 5×4 Gy is the preferred regimen. In a previous study of 442 patients treated with WBRT alone for more than three brain metastases, 5×4 Gy was not inferior to 10×3 Gy with respect to survival and intracerebral control (2). Grade 3 radiation-related acute toxicity rates were 9% and 6%, respectively (p=0.35). For selected patients with very limited survival prognoses, best supportive care (BSC) including corticosteroids without WBRT may be considered. In a randomized trial of patients with brain metastases from non-small cell lung cancer and very poor prognoses, BSC including dexamethasone was not inferior to BSC combined with 5×4 Gy of WBRT (9). In the WBRT group, more episodes of drowsiness, hair loss, nausea, and dry or itchy scalp were recorded. However, serious adverse events were not more common. The difference between mean quality-adjusted life-years was only 4.7 days, i.e. 41.7 days in the BSC alone group and 46.4 days in the BSC plus WBRT group. Thus, WBRT did not provide a significant benefit for this group of patients.
Results of the univariate analyses of survival at 3, 6, 9 and 12 months.
In patients with more favorable survival prognoses, intracerebral control and survival are also important endpoints. These patients are not uncommon, and a considerable number receive WBRT without local treatment. For example, in the study performed to create the RPA classification, all patients received WBRT without local treatment. The 236 patients of RPA class 1 (representing 20% of the entire cohort) had a median survival time of 7.1 months (5). In the study creating the GPA classification, 9% (N=102) of patients had a median survival of 11.0 months, and an additional 16% (N=168) of patients had a median survival of 6.9 months (6). Patients with longer expected survival were reported to achieve better intracerebral control and survival with longer-course WBRT including total doses >30 Gy. A retrospective study of 184 patients with favorable prognoses compared 10×3 Gy and 20×2 Gy (4). Survival rates at 1 and 2 years were 50% and 14%, respectively, after 10×3 Gy, and 61% and 30%, respectively, after 20×2 Gy. Intracerebral control rates at 1 and 2 years were 28% and 11%, respectively, after 10×3 Gy, and 44% and 20%, respectively, after 20×2 Gy.
Results of the multivariate Cox proportional hazards model.
It becomes obvious that precise estimation of a patient's prognosis is important to optimally tailor the treatment. Several major scoring tools addressing this aspect exist (5-8). However, they are relatively complex and require complete tumor staging prior to the start of treatment which may result in a considerable treatment delay. Such delay is generally not acceptable for patients with brain metastases requiring relief of their debilitating symptoms. Thus, a very simple and fast tool would be preferable, particularly for patients with a poor performance status and severe brain symptoms. This study aimed to provide such a tool, simply based on the number of significant pre-treatment symptoms. In a large cohort of 1,907 patients treated with WBRT alone, 10 major pre-treatment symptoms were recorded. The majority of patients presented with one (42%) or two (33%) symptoms. Fewer than 10% were symptom-free, and 16% of patients had three or more symptoms. On univariate analysis, the number of symptoms was significantly inversely associated with survival. However, on multivariate analysis, significance was not achieved. Thus, the number of symptoms was not an independent predictor of survival and appeared not useful for the creation of a survival score. As the survival rates of patients with 0-2 symptoms were almost identical at 3 and 6 months and still similar at 9 and 12 months, we performed an additional analysis to investigate the potential prognostic value of the number of symptoms and compared 0-2 to ≥3 symptoms. However, again the results were significant on univariate but not on multivariate analysis. When interpreting these results, one should take into account the limitations of this study, mainly its retrospective design. Retrospective studies always bear the risk of hidden selection biases. Another limitation of the study is that it has not yet been externally validated. Since all patients included in this study received WBRT alone, the results cannot be generalized to other treatments for brain metastases including neurosurgical resection alone, radiosurgery alone and such local therapies combined with WBRT.
Since the number of symptoms was not significant on multivariate analyses, we feel that it cannot replace existing scoring tools designed to predict the survival prognoses of patients irradiated for brain metastases. It may only serve for orientation. One may consider re-evaluation of the existing scoring tools after the exclusion of visceral metastases as a predictive factor to avoid the need for complete tumor staging, which often results in a delay of treatment. This may also be considered for diagnosis-specific scoring tools (10-15).
In conclusion, the number of symptoms was not an independent predictor of survival in patients irradiated for brain metastases and, therefore, cannot replace existing tools. It may only serve for orientation. Since existing tools are based on presence/absence of visceral metastases, they require complete tumor staging. Such staging may result in treatment delay, which is often not acceptable, particularly for patients requiring urgent relief of symptoms. Existing tools may be re-evaluated after exclusion of visceral metastases as an essential component in order to hopefully provide a fast-to-use survival score based on independent prognostic factors.
Footnotes
Authors' Contributions
D.R., H.C.H., S.J. and S.E.S. participated in the design of the study. D.R., H.C.H., L.D. and S.J. provided data. D.R. and S.E.S. performed the analyses and the interpretation of the data. D.R. and S.E.S. drafted the article, which was reviewed and approved in its final form by all Authors.
Conflicts of Interest
On behalf of all Authors, the corresponding Author states that there is no conflict of interest related to this study.
- Received May 28, 2019.
- Revision received July 4, 2019.
- Accepted July 5, 2019.
- Copyright© 2019, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved