TURKISH JOURNAL OF ONCOLOGY

Summary

OBJECTIVE
This study aimed to retrospectively evaluate overall survival (OS) of the patients with brain metastases (BM) from lung cancer who had been treated with whole-brain radiotherapy (WBRT) and gamma knife (GK) according to prognostic factors and prognostic score indexes.

METHODS
Ninety-five patients with brain metastases from lung cancer were retrospectively evaluated using age, sex, lung cancer histological type, extracranial metastases, primary tumor control, number of brain metastases, total brain metastases volume, brain metastasectomy, chemotherapy, EGFR mutation, EGFR-TKI therapy, Karnofsky Performance Status (KPS), Recursive Partitioning Analysis (RPA) Class, Basic Score for Brain Metastases (BS-BM), Graded Prognostic Assessment Index (DS-GPA) and Modified Lung-Specific between 2015 and 2018. Univariate analysis of OS was performed using the Kaplan?Meier method supplemented by the log-rank test. We also applied multivariate survival analysis using the Cox Regression Model.

RESULTS
The median OS for all patients with brain metastases from lung cancer was six months± SE: 0.807 (range: 1?42 months; 95% CI: 4.419?7.581) and one-year overall survival rate was 25.3%. The median OS was four months, four months, 12 months in the WBRT arm, the GK arm and the combined WBRT-GK arm, respectively (p=0.004). In multivariate analysis, treatment with WBRT?GK (p=0.030), brain metastasectomy (p=0.019), controlled primary tumor (p=0.004), chemotherapy (p=0.001) were significantly correlated with overall survival. BS-BM (p=0.033) was closely related to overall survival compared to other prognostic score indexes on the multivariate analysis.

CONCLUSION
The patients with BM benefited from WBRT and GK combined therapy. BS-BM for the survival of patients with BM from lung cancer is the most appropriate prognostic index.

Summary

Introduction

Lung cancer is the most common source of brain metastases.[1] The aims of treatment are palliation of neurological symptoms, maintenance of performance status, and local control of the metastatic disease. [2] Whole-brain radiotherapy (WBRT) is commonly used to improve neurological symptoms in patients with multiple brain metastases and control the disease within the brain.[1,3-6] Stereotactic radiosurgery is a good choice for oligometastatic brain lesions,[5,7,8] and a metastasectomy is also a treatment option. However, the selection of local treatment only or WBRT depends on the performance of the patient, number of brain metastases, and histology.[9] In patients with oligo brain metastases treated with radiosurgery or metastasectomy, the addition of WBRT to the treatment regimen has reduced intracranial recurrence and neurological mortality. The most important problem in patients with the addition of WBRT is neurocognitive function failure. However, functional independence and mean survival were not improved.[10] WBRT and radiosurgery are decreased in learning and memory function compared to radiosurgery alone.[11] WBRT and radiosurgery improved control of local and remote brain metastases compared to the radiosurgery alone. [12] The potential benefit of GK radiosurgery is the reduction of radiation to the surrounding normal brain parenchyma, which may thereby reduce neurological toxicities compared with WBRT [2,13] or which may improve local control when combined with WBRT.[5]

KPS score, primary lesion control, presence of extracranial metastases, presence of multiple metastases are important prognostic factors in the literature.[9,14-] In recent years, several prognostic scoring systems based on independent prognostic factors have been developed to evaluate pretreatment variable contributions, to choose the appropriate treatment for individual patients, and guide future research. The most widely used indices are RPA class, BS-BM, and GPA.[14,15,16] The addition of brain tumor volume to the lung-specific GPA index (Modified Lung-Specific GPA) better predicted the OS of patients with brain metastasis from lung cancer.[17] Clinical parameters used for prognostic indexes (RPA, BSBM, DS-GPA, and Modified Lung- Specific GPA) are shown in Table 1.

Table 1: Clinical parameters used for prognostic indexes (RPA, DS-GPA, and BS-BM)

In this study, retrospective examination of the patients with brain metastases from lung cancer who were treated with WBRT, GK, and combined WBRT? GK in a single center were reported. We tried to find out the appropriate prognostic indexes for all patients with BM who underwent GK, WBRT, and combined WBRT?GK.

Introduction

Methods

Study Design
This study included 95 patients with brain metastases from lung cancer who underwent total WBRT, GK radiosurgery, or combined WBRT?GK between 2015 and 2018. Seventy patients with non-small-cell lung cancer and twenty-five patients with small-cell lung cancer were included in our study.

Data Collection
Data of 95 patients were retrospectively collected and evaluated regarding the clinical characteristics, including age, sex, WBRT treatment, GK treatment, histological type of lung cancer, extracranial metastases, primary tumor control, number of brain metastases, total brain metastases volume, brain metastasectomy, EGFR mutation, EGFR-TKI therapy, Karnofsky Performance Status (KPS), RPA Class, BS-BM, DS- GPA, and Modified Lung-Specific GPA. Prognostic indexes, such as RPA Class, BS-BM, DS-GP, Modified Lung-Specific GPA, were applied to patients with brain metastases. These criteria were chosen in accordance with previous studies that identified significant predictors of survival in patients with brain metastases.[9,14,16-18]

Study Procedures WBRT
Patients with multiple brain metastases were chosen for the treatment with WBRT. Patients were immobilised in a supine position within a thermoplastic mask. The brain was contoured as a clinical target volume (CTV) until the foramen magnum. The CTV was equal to the PTV. All BM, optic nerves, brainstem, eyes and lenses were contoured. Patients were positioned with a mask. The use of a planning CT was mandatory with a slice thickness of ?5 mm. WBRT was performed with 6-MV photons from a Siemens Artiste linear accelerator. The daily prescription dose is 2.5 and 3 Gy prescribed at the ICRU reference point.

Study Procedures Radiosurgery
The patients with <3 cm and 1-3 brain metastases were chosen for treatment with gamma knife. Patients were immobilised in a supine position with a stereotactic fixation system using an invasive frame. A planning CT scan with ≤2 mm thick contiguous slices (preferable CT slice thickness=1 mm) will be fused to a contrastenhanced stereotactic MRI scan. BMs were contoured as a CTV to PTV margin used. Radiosurgery was performed with Elekta Leksell GK machine. Doses ranged from 15 to 25 Gy.

Patient Follow Up, Salvage Therapy
The metastatic brain lesions of the patients in this study were followed by magnetic resonance imaging (MRI). OS was referenced from the day the diagnosis of brain metastases was confirmed by MRI. Intracerebral failure was diagnosed with MRI. The exact frequency and number of MRIs following irradiation were unavailable because the anonymized database used did not include these data. In general, the follow-up schedule after therapy included MRI every three months, whereas MRI was performed only in cases of new or progressive symptoms in most patients undergoing WBRT or GK. All patients with brain metastases from lung cancer were treated after being evaluated by the Neurosurgery and Radiation Oncology Departments. Salvage SRS (stereotactic radio-surgery) was added to a treatment option for recurrent BM after the failure of WBRT.[19] After SRS, salvage WBRT was added to reduce intracranial relapses and neurologic deaths.[20] Patients who underwent only Gamma Knife treatment, only WBRT treatment, salvage treatment accepted in GK arm, in WBRT arm, in the combine treatment arm (WBRT-GK arm), respectively. We accepted patients who underwent salvage treatment due to progression after WBRT treatment or SRS treatment combined WBRT-GK arm.

Statistical Analysis
Univariate analysis of OS was performed using the Kaplan? Meier method supplemented by the log-rank test to find out the factors.[21] We also applied multivariate survival analysis using the Cox Regression Model. Only the factors that exhibited statistical significance in univariate analysis were included in a multivariate analysis that utilized Cox proportional hazards regression tests. All tests were two-tailed, and a p-value <0.05 was considered significant. The statistical analyses were reviewed by medical statistician staff in our medical faculty.

Methods

Results

Patients Characteristics
This study included 95 patients with brain metastases from lung cancer. Of 95 patients, 70 (73.7%) and 25 (26.3%) had non-small-cell cancer and small-cell ca, respectively. Of 95 patients, 28 (29.4%), 36 (37.9%), 6 (6.4%) and 25 (26.3%) had squamous cell ca, adeno ca, nsclc (other type) and small-cell ca, respectively. Among these patients, 89 (93.7%) and six (6.3%) were men and women, respectively. The median age was 61 years (min: 37 years, max: 83 years). The number of patients underwent WBRT, GK radiosurgery, and combined WBRT?GK was 27 (28.1%), 39 (40.6%), and 29 (30.2%), respectively. In our study, 83 of 95 patients died during the study period. The median OS for all patients with brain metastases from lung cancer was six months±SE: 0.807 (range: 1-42 months; 95% CI: 4.419?7.581) and one-year overall survival rate was 25.3%.

Survival Analysis
In univariate analysis (Kaplan-Meier, log-rank test), the median OS of patients with brain metastases from lung cancer was significantly associated with treatment combined WBRT-GK, the presence of brain metastasectomy controlled primary tumor, the absence of extracranial metastases, chemotherapy, KPS score, RPA class, BS-BM, and Modified Lung-Specific GPA (p<0.05). The median OS of patients with brain metastases from lung cancer was not statistically significant with the age, sex, histological type of lung cancer, number of brain metastases, the tumor volume of brain metastases, EGFR mutation state, EGFR-TKI therapy, DS-GPA (p>0.05). Clinical characteristics and results of the univariate analysis of the OS of patients with brain metastasis from lung cancer are presented in Table 2. The median OS was four months, four months, 12 months in the WBRT arm, the GK arm, the WBRTGK arm, respectively (p=0.004) (Fig. 1). In our study, 26 of the 35 patients who underwent surgery, 57 of the 60 patients who did not undergo surgery, died. The median OS was four months and 12 months in patients who did not undergo surgery and the patients who underwent surgery, respectively (p=0.004). The median OS was significantly associated with the controlled primary tumor and the absence of extracranial metastases (p<0.05). The median OS was four months, 20 months in the absence of primary tumor control, in the presence of primary tumor control, respectively (p=0.000). The median OS was 12 months in patients without extracranial metastases, whereas the median OS was four months in patients with extracranial metastases (p=0.035). The median OS was seven and three months in patients who received chemotherapy and who received no chemotherapy, respectively (p=0.002).

Table 2: Clinical characteristics and results of the univariate analysis (Kaplan-Meier, log-rank test) of OS of patients with brain metastasis from lung cancer

Fig 1: Kaplan-Meier analysis of OS according to treatment arms.

KPS, RPA class, BS-BM and Modified Lung-Specific GPA were all closely related to prognosis in our study on univariate anal¬ysis (Kaplan-Meier, log-rank test). The median OS was not statistically significant with DS-GPA (Fig. 2). The median OS was eight and four months in patients with KPS ?70 and lower KPS, respectively (p=0.005). The median OS was 20 months, eight months, four months in patients with RPA Class 1, RPA Class 2, RPA Class 3, respectively (p=0.003). The median OS was three months, seven months, 12 months, 20 months in patients with BS-BM 0, BS-BM 1, BS-BM 2, BS-BM 3, respectively (p=0.001). The median OS was four months, six months, 12 months in patients with DS-GPA 0-1, DS-GPA 1.5-2.5, DS-GPA 3-3.5-4, respectively (p=0.057). The median OS was four months, seven months, 12 months in patients with Modified Lung-Specific GPA?1, Modified Lung- Specific 13, respectively (p=0.027).

Fig 2: Kaplan-Meier analysis of OS according to each of the prognostic score indexes.

In univariate analysis, we found that patients with brain metastases from lung cancer with age ?60 years, small-cell ca, the absence of brain metastasectomy, controlled primary tumor, extracranial metastases, the brain metastases ?5, the total tumor volume >4 cm3, receiving chemotherapy had better overall survival in the combined WBRT-GK treatment arm according to treatment modality arms. However, patients with KPS ≥70, KPS ?60, RPA Class 2-3, BS-BM 0, BS-BM 1, DS-GPA 0-1, DS-GPA 1.5-2.5, Modified Lung GPA ≤1 were significantly correlated with overall survival in the combined WBRT?GK treatment arm according to treatment modality arms on univariate analysis (p<0.05). Results of the univariate analysis (Kaplan- Meier, log-rank test) of overall survival of the prognostic factors in patients with brain metastasis from lung cancer to treatment modality arms are presented in Table 3. For patients with brain metastases from small cell ca, the median OS was four months, two month, 18 months in the WBRT, the GK arm, the combined WBRT?GK the treatment arm, respectively (p=0.000). The median OS was not statistically significant for patients with brain metastasectomy from non-smallcell according to the treatment arms (p>0.05). In the absence of brain metastasectomy, the median OS was four months in the WBRT arm and GK arm, and the median OS was eight months in the combined WBRT? GK arm (p=0.026). The median OS was not statistically significant for patients who underwent brain metastasectomy according to the treatment arms (p>0.05). Patients with extracranial metastases had the highest median OS of 11 months in combined WBRT?GK arm (p=0.003). Patients with controlled primary tumor had the highest median OS of 27 months in combined WBRT?GK arm (p=0.021). Among patients with oligo brain metastases, the median OS was not statistically significant according to the treatment modality arms (p>0.05). For the patients with five or more brain metastases, the median OS was four months, one month in the WBRT, the GK arm, respectively, and 11 months in the combined WBRT?GK arm according to the treatment modality arms (p=0.014). The median OS was four months in the WBRT arm and GK arm, and the median OS was 12 months in the combined WBRT?GK arm for patients with brain metastases volume >4 cm3, respectively (p=0.009). In patients receiving chemotherapy treatment, the combined WBRT-GK arm had higher median survival with 12 months than other treatment arms (p=0.032).

Table 3: Overall survival of the univariate analysis (Kaplan-Meier, log-rank test) of patients with brain metastases from lung cancer according to treatment modality arms

Patients with KPS ≥70, KPS ≤60, RPA Class 2-3, BS-BM 0, BS-BM 1, DS-GPA 0-1, DS-GPA 1.5-2.5, Modified Lung GPA≤1 had the highest median overall survival in the combined WBRT?GK treatment arm according to treatment modality arms on univariate analysis (p<0.05).

The multivariate analysis demonstrated that treatment with WBRT-GK (p=0.030), brain metastasectomy (p=0.019), controlled primary tumor (p=0.004), chemotherapy treatment (p=0.001) were significantly correlated with overall survival. Results of the multivariate analysis (the Cox Regression Model) of the overall survival of the prognostic factors are presented in Table 4. In multivariate analysis, BS-BM (p=0.033) was closely related to the overall survival among the class of prognostic score indexes in our study. In multivariate analysis, BS-BM was a better predictor for the survival of patients with BM from lung cancer. However, KPS score, RPA class, DS-GPA and Modified Lung-Specific GPA were not statistically significant with overall survival between patients with BM from lung cancer on multivariate analysis. Results of the multivariate analysis of overall survival according to prognostic score indexes are presented in Table 4.

Table 4: Results of the multivariate analysis (Cox- regression test) of overall survival of the prognostic factors and prognostic score indexes of patients with brain metastases from lung cancer

Results

Discussion

In our study, including 95 patients with brain metastases from lung cancer, the median OS was not statistically significant between patients with small-cell ca and nonsmall- cell ca (p>0.05). The median OS of patients with brain metastases from small-cell ca was four months, two months, 18 months in the WBRT, the GK arm, the combined WBRT-GK arm, respectively (p=0.000). A retrospective study by Wegner et al. suggested that patients with SCLC who underwent a combined WBRT with radiosurgical boost had improved outcomes over patients who received either WBRT or SRS alone.[22] In our study, similar to the study of Wegner et al. combined radiotherapy and radiosurgery for small cell ca increases survival. In cases of wild-type EGFR and ALK NSCLC, there are few effective systemic options, and therefore, WBRT may have a more prominent role. Despite the current trend of preferring SRS alone, we need to carefully consider the important role of WBRT, particularly in patients with BM from NSCLC who have a favorable prognosis.[23] In our study, there was not statistically significance between OS of patients with EGFR mutation state, EGFR-TKI therapy.

In the study of Bowden, the presence of multiple brain metastases is a negative predictor of worse prognosis.[24] In other studies, the number of metastatic lesions of 1-2 vs 3-4 vs. ≥5 did not approach significance. The importance of the number of brain metastases vs. the total tumor volume in predicting OS remains under debate.[9,25,26] We found that the number of brain metastases and total tumor volume were statistically significantly only according to treatment modality arms on univariate analyses. In our study, the median OS was highest in the combined WBRT-GK arm for patients with brain metastases volume >4 cm3 (p=0.009). Several studies showed that the tumor volume was statistically significantly correlated with the OS.[2,27,28] Total tumor volume should be examined more closely in future studies. In our study, the median survival was the highest for patients with ≥5 BM in the combined WBRT?GK arm (p=0.014). Gamma Knife salvage therapy may be applied with WBRT in patients with ≥5 BM.[29] Considering the survival benefit of the combination of WBRT and SRS, hippocampus sparing-WBRT studies are required to reduce the cognitive effects of WBRT.[30]

In our study, patients who underwent brain metastasectomy had better the median survival than the patents who did not undergo brain metastasectomy, which was statistically significant. The combined WBRT-GK treatment improved the overall survival of the patients who did not undergo brain metastasectomy. The combined WBRT-GK treatment was not related with the median OS in patients who underwent brain metastasectomy. In two trials [31,32], a survival benefit was reported for patients underwent the combined brain metastasectomy+WBRT.

Chemotherapy has a limited role in the treatment of brain metastases. Several studies have reported that some patients might benefit from aggressive therapy, including surgery, radiotherapy and chemotherapy. [33] Kim et al. analyzed retrospectively the outcome of chemotherapy only, upfront whole brain radiotherapy or stereotactic radiosurgery in NSCLC patients with asymptomatic brain metastases. There was no significant difference in OS among three groups, but a subset analysis of 110 patients suggested a potential role of systemic chemotherapy alone or upfront SRS followed by chemotherapy.[34] In another study, a combination of local therapies and systemic chemotherapy shown to increase survival in NSCLC patients with brain metastases.[35] In a multicenter phase II study by Galletta et al. analyzed the association of combination with cisplatin, fotemustine, and whole brain radiotherapy, but this scheme does not represent a therapeutic option for patients with NSCLC.[36] Similar to these reports, we showed that patients with brain metastases from lung cancer had better survival when chemotherapy was used as a treatment option. Since we have a small number of patients groups as a limitation, we look forward to the new reports assessing the efficacy of chemotherapy for brain metastases from lung cancer.

Similar to other studies, we demonstrated overall survival was better in patients with KPS ≥ 70 on univariate analysis.[37,6] Ji et al. showed that overall survival was increased in patients with favorable KPS on univariate analysis and multivariate analyses.[38] In our study, OS was increased in patients with favorable KPS, control of the extracranial disease, controlled primary tumor as reported by Gao.[39] Our series also matches with previous reports that have demonstrated improved median OS with better control of the extracranial disease.[13,35] The OS was better in the study of 294 cases with controlled primary tumor, similar to our study.[25] However, another study reported that the OS improved in patients with controlled primary tumor and previous metastasectomy.[40] RPA Class 1 (Class 1, median survival 20 months) was different from the original data because the number of our patients in RPA class 1 was small. In the first RTOG study by Gaspar, the median OS of patients with RPA Class 1 was 7.1 months.[14] In another study, including 445 cases by Gaspar, median OS was 6.2 months for RPA 1, 3.8 months for RPA 2.[41] In the study of 528 cases by Nieder, the RPA class received similar results, but the number of patients in RPA Class 1 was very small.[42] In the study of 110 patients by Lorenzoni et al., median OS was 27.6 months for RPA Class 1, 10.7 months for RPA Class 2, similar to our study.[15] In our multivariate analysis, BS-BM (p=0.033) was a better predicted the survival of BM. Lorenzoni reported that median OS was undefined for BS-BM 3 (55% at 32 months) and was 13.1 months for BS-BM 2. However, the median OS was 20 months for BS-BM 3, and 12 months for BS-BM 2 in our study.[15] Similar to our multivariate analysis, Villa et al. reported that the median OS improved for BS-BM (p<0.001). In this prospective study, prognostic indexes as RPA, and BS-BM were prognostically relevant in BM patients similar to in our univariate analysis.[43] In other retrospective study of 335 patients by Ji et al., RPA and BSBM were statistically significant for OS, similar to our study.[38] In the study of 1960 cases with BM by Sperduto, GPA was an appropriate prognostic index.[16] In our study, DS-GPA was not statistically significant for OS, while it was statistically significant in other studies. [15,43,38] Modified Lung-Specific GPA improved the OS of patients with brain metastasis from lung cancer, similar to our study.[17]

Limitations of the Study
Our study has several limitations. The retrospective nature of this study and the small number of patients are the major limitations.

Discussion

Conclusion

In multivariate analysis, we found that control of primary tumor, combined treatment with WBRT-GK, brain metastasectomy and chemotherapy were statistically significant for overall survival of patients with BM. In univariate analysis, prognostic indexes as KPS, RPA class, BS-BM, and Modified Lung-Specific GPA assessed could predict patient prognosis, demonstrating the reliability and clinical relevance of these scores. However, only BS-BM was a better predictor for the survival of patients with BM according to prognostic score indexes on multivariate analysis. Our data suggest that BS-BM is the most appropriate prognostic index.

Peer-review: Externally peer-reviewed.

Conflict of Interest: None declared.

Ethics Committee Approval: Approval from the research ethics board was obtained from Pamukkale University Ethics Committee.

Financial Support: None declared.
Authorship contributions: Concept - H.S., P.B.B.; Design - H.S., P.B.B.; Supervision - H.S., P.B.B.; Funding - None; Materials - H.S., P.B.B.; Data collection and/or processing -H.S., P.B.B.; Data analysis and/or interpretation - H.S., P.B.B.; Literature search - H.S., P.B.B.; Writing - H.S., P.B.B.; Critical review - H.S., P.B.B.

Conclusion