Outcomes of re-irradiation for brain recurrence after prophylactic or therapeutic whole-brain irradiation for small cell lung Cancer: a retrospective analysis

Background Small cell lung cancer (SCLC) can recur in the brain after whole-brain irradiation (WBI). We documented outcomes after treatment of such recurrences and sought predictors of local control and overall survival (OS). Materials and methods Eighty-five patients with SCLC and brain recurrence after prophylactic or therapeutic WBI in 1998–2015 were identified and data were extracted from the medical records. Survival was estimated with the Kaplan-Meier method, and univariate and multivariate Cox proportional hazards modeling was used to identify factors associated with OS or further brain progression. Results Brain recurrence was treated by stereotactic radiosurgery (SRS) in 33 patients (39%), repeat WBI in 14 (16%), chemotherapy-only in 16 (19%), and observation in 22 (26%). Median OS time after brain recurrence (OSrec) was 4.3 months for all patients; 6-month OSrec rates were 58% after SRS, 21% after repeat WBI, 50% after chemotherapy-only, and 5% after observation (P < 0.001). Inferior OSrec was associated with poor performance status (ECOG score ≥ 3) and uncontrolled extracranial disease. Superior OSrec was associated with receipt of ≥4 chemotherapy cycles before brain recurrence and receipt of chemotherapy, SRS, or repeat WBI afterward. Receipt of chemotherapy after brain recurrence correlated with brain progression. Conclusions Some patients with brain recurrence after WBI for SCLC can survive for extended periods with appropriate intervention, especially those with adequate performance status or controlled extracranial disease.


Background
Whole-brain irradiation (WBI) is an important component of treatment for small cell lung cancer (SCLC), whether given prophylactically or to treat known brain metastases. Because SCLC has a high propensity to seed in the brain, disease recurrence in the brain after WBI is not uncommon despite the radiosensitivity of the disease [1,2]. Treatment strategies for brain recurrence after initial therapy for SCLC are limited but include surgical resection, stereotactic radiosurgery (SRS), repeat WBI, and chemotherapy.
In one autopsy series, 65% of patients had evidence of brain metastases at autopsy, with the probability of brain metastases increasing with survival time to 80% after 2 years [3]. Extended survival times from the use of modern radiotherapy techniques [4,5] and the enhanced ability to detect small asymptomatic brain metastases [6] collectively increase the likelihood that more patients will manifest brain recurrence after initial WBI therapy for SCLC [6].
Repeat WBI for brain metastases from a variety of tumor types including lung cancer has been shown to be effective for symptom palliation [7][8][9][10][11]. A benefit from re-irradiating brain recurrences has also been found for patients with SCLC, although the median survival time after re-irradiation remains relatively short at 2 to 6 months [2,12,13]. In one study, patients with SCLC who developed brain metastases after prophylactic cranial irradiation (PCI) had effective symptom palliation from either repeat WBI or SRS [12]. Another series of patients with SCLC and brain relapse after WBI confirmed the feasibility of SRS as a salvage option [13,14].
Nevertheless, information on how best to treat SCLC that has recurred in the brain after WBI is still sparse. Considering the high prevalence and depressing prognosis associated with brain recurrence, identifying reliable prognostic factors would be helpful for choosing among the various available treatment strategies.
In this single-institution retrospective study, we evaluated patients with SCLC who developed brain recurrence after WBI, with the goals of assessing outcomes and identifying prognostic factors that could influence survival and brain control.

Methods
After approval of this study by the appropriate institutional review board, we identified 85 patients with SCLC and brain recurrence after WBI, given either with therapeutic or prophylactic intent, from 1998 through 2015 at a single tertiary cancer center. Management strategies for those brain recurrences (SRS, repeat WBI, chemotherapy, or observation) were extracted from the patients' medical records, as was the following demographic and clinical information: factors at the initial diagnosis [sex, race, initial disease stage (limited vs. extensive), the presence and number of brain metastases at initial diagnosis (≤4 vs. > 4), intent of the first WBI (prophylactic vs. therapeutic), number of chemotherapy cycles before brain recurrences (< 4 vs. ≥4), and use and dose of thoracic radiation therapy (< 45 Gy vs. ≥45 Gy)]; factors at the brain recurrence [age, ECOG performance status (PS), symptoms and number of brain lesions (≤4 vs. > 4), the presence and status of extracranial disease (controlled vs. uncontrolled), and time from the initial WBI to the brain recurrence (< 6 months vs. ≥6 months)]; and factors after brain recurrence [treatment strategy (SRS, repeat WBI, chemotherapy only, or observation)]. Patients were typically given steroids, especially if they had symptoms, immediately upon diagnosis of brain recurrence. Patients with brain recurrence managed with observation were followed with best supportive care, including drugs such as steroids for symptom control.
The primary endpoint of this study was overall survival after brain recurrence (OSrec), which was defined as the time from the date of brain recurrence to the date of death. Local control in the brain after brain recurrence (BCrec), evaluated radiologically or clinically, was defined as the time from the date of brain recurrence to the date of detection of further brain recurrence. Patients lost to follow-up were censored at the date of the last follow-up. Progression was defined as lesions that persisted after SRS, repeat WBI, or chemotherapy or, in the observation group, as brain recurrence. The Kaplan-Meier method was used to assess survival and progression probabilities, and log-rank tests were used to compare Kaplan-Meier estimates of event rates between potential predictors. Univariate and multivariate Cox proportional hazards models were used to assess associations between factors and survival or further disease progression in the brain, and those factors with a P value < 0.1 were used to construct the final multivariate Cox model. All analyses were two-sided, and significance was set at P < 0.05. Statistical analyses were done with SPSS version 24 (SPSS, Chicago, IL, USA).
Brain recurrence was managed with SRS in 33 patients (39%), with repeat WBI in 14 (16%), with chemotherapy only in 16 (19%), and with observation (and best supportive care) in the remaining 22 patients (26%). Two of the patients in the SRS group had received SRS after craniotomy owing to the size of the lesion. One patient in the repeat WBI group had local fractionated radiotherapy rather than WBI. Of the 33 patients in the SRS group, 27 (82%) had ≤4 brain lesions at recurrence; by contrast, 9 patients (64%) in the repeat WBI group and 13 patients (59%) in the observation group had > 4 lesions. More than a third of patients in the SRS group also received additional chemotherapy, but only one patient in the repeat WBI group and no one in the observation group received further chemotherapy.
Nineteen patients in the SRS group (58%) had further brain failure, and 5 of those patients received a second SRS; 2 patients who received a second SRS had a third recurrence, which was treated with salvage WBI; and 1 was treated with salvage WBI. Thirteen patients in the chemotherapy-only group (81%) had further brain failure; 1 was treated with SRS, 1 was treated with SRS and had a second recurrence also treated with SRS, 2 were treated with SRS and had a second recurrence treated with salvage WBI, and 3 were treated with salvage WBI. Ten patients in the repeat-WBI group (71%) had further brain failure, of whom 1 had a craniotomy and the other 9 received best supportive care. The median total dose of the first course of WBI was 30 Gy (range, 20-36 Gy), given in a median 3 Gy per fraction (range, 2-3 Gy). Most patients given repeat WBI received 20 Gy in ten 2-Gy fractions. For patients who received SRS, the prescribed dose was 12-20 Gy per lesion depending on their size and location.
As for factors influencing OSrec, univariate analysis showed that worse OSrec was associated with poor PS (≥3) at recurrence, the presence of symptomatic brain lesions, and the presence of uncontrolled extracranial disease; better OSrec was associated with receipt of ≥4 chemotherapy cycles before brain recurrence, receipt of chemotherapy after brain recurrence, receipt of repeat WBI or SRS for brain recurrence, and an interval of ≥6 months from the end of the first WBI to the brain recurrence. Univariate analysis for factors associated with BCrec showed that worse BCrec was associated with poor PS (≥3), > 4 brain lesions at recurrence, symptomatic brain lesions, and uncontrolled extracranial disease; better BCrec was associated with receipt of repeat WBI or SRS for the brain recurrence. OSrec and BCrec did not differ for patients with limitedversus extensive-stage disease at initial diagnosis (Table 2).
Ten of the 85 patients in our study lived for more than 1 year after the first brain recurrence (Table 4). Notably, 8 of those 10 patients did not have liver or bone involvement at initial diagnosis. Moreover, Patients 5 and 7, who received temozolomide (without re-irradiation) for the initial recurrence later underwent SRS for a subsequent brain recurrence. Therefore all 10 of these patients had received SRS, whether for the initial or subsequent brain recurrence.

Discussion
Available options for patients with disease recurrence in the brain after WBI for SCLC are limited, but our findings suggest that some patients can expect better survival when those recurrences are treated with therapeutic intent by using SRS, repeat WBI, or chemotherapy, especially patients with adequate PS or controlled extracranial disease.
Despite the radiosensitivity of SCLC, brain recurrence after WBI is not uncommon [1,2]. In one meta-analysis One patient received local (not whole-brain) radiation to tumor site using conventional radiotherapy Abbreviations: SRS stereotactic radiosurgery, WBI whole-brain irradiation, ECOG Eastern Cooperative Oncology Group of 7 trials comparing PCI with no PCI for patients with limited-stage SCLC, 27% of all patients had disease recurrence in the brain [1]. Another retrospective series revealed brain recurrence in 15% of patients with SCLC after WBI; all of those recurrences were treated with either repeat WBI (12%) or SRS (3%) [2].
In the current study, we found that both PS and extracranial disease control were important contributors to survival after brain recurrence. Although these results were not entirely consistent with those of other studies (perhaps because of patient heterogeneity), some other studies also found these factors to be important for survival in such cases. Among several studies of prognostic factors and outcomes after re-irradiation with WBI for various malignancies including lung cancer [7][8][9][10][11], one showed that repeat WBI relieved symptoms in 71% of treated patients, and even though the median survival time after re-irradiation was short at 4 months, the 1-year survival rate for patients with no extracranial metastases was better at 14% versus only 4% for those with extracranial metastases [7]. A similar study also found better prognosis among patients with stable extracranial disease at re-irradiation [8]. In addition to extracranial disease control, high Karnofsky performance score (KPS) [9][10][11], the absence of severe symptoms [10], and longer interval between the two courses of WBI [11] have also been correlated with improved survival in other studies.
A few studies have also examined the feasibility of SRS for brain metastases after WBI for various cancers, including lung [15][16][17]. In one such study, salvage SRS produced a median survival time of 8.4 months overall, and was longer for patients with a single brain metastasis than for those with multiple brain metastases (12.0 vs. 7.9 months) [15]. Another study found that high KPS, controlled extracranial metastases, and a good response to re-irradiation all had positive effects on survival [16]. The importance of KPS was also confirmed in another study, along with neurologic function [17].
Although available data are limited, we did find a few studies that addressed treatment for brain recurrences after WBI in SCLC. One study reported a benefit from brain retreatment with WBI or SRS among patients with SCLC and post-PCI brain metastases; in that study, 40% of symptomatic patients achieved symptom palliation A B C Fig. 1 Kaplan-Meier plot of survival after brain failure after wholebrain irradiation for small cell lung cancer by (a) Eastern Cooperative Oncology Group performance status score (ECOG PS); (b) extracranial disease control; or (c) form of treatment for brain recurrence. Survival after brain failure was significantly better for patients with superior PS, extracranial disease control, and intervention than for those with poor PS, uncontrolled extracranial disease, and observation/best supportive care. SRS, stereotactic radiosurgery; WBI, whole-brain irradiation Table 2 Univariate analysis of factors affecting survival and further brain control after recurrence in the brain after therapeutic or prophylactic irradiation after re-irradiation, and the median survival times were 3 months after repeat-WBI and 5 months after SRS. That study further showed that patients with higher KPS (≥50) had significantly improved survival after re-irradiation, a finding that is comparable to ours [12]. The importance of PS in survival was also confirmed in a study that used salvage SRS after WBI; other factors found to be significant were tumor volume and the interval between SCLC diagnosis and the appearance of brain metastases [14]. Another retrospective study assessing the efficacy of salvage SRS after WBI for SCLC showed that although the median OS time for all patients was 5.9 months, 24% of patients survived for at least 12 months, and patients with no evidence of extracranial disease had Table 2 Univariate analysis of factors affecting survival and further brain control after recurrence in the brain after therapeutic or prophylactic irradiation (Continued)  Abbreviations: ECOG Eastern Cooperative Oncology Group, WBI whole-brain irradiation, G-knife Gamma knife therapy significantly better survival than did patients with stable or progressive extracranial disease [13].
Our findings also showed that patients who received ≥4 cycles of chemotherapy before brain recurrence had better survival, perhaps because of a lower systemic disease burden. We also found that patients who received chemotherapy after brain recurrence had better survival and better brain control than those who did not.
Although several new chemotherapy agents have been evaluated in phase II trials for second-line treatment of SCLC, to date they have not resulted in changes to standard practice, which is currently platinum plus etoposide for first-line therapy and topotecan for second-line therapy [18]. Further improvements in systemic therapy, such as targeted agents and immune checkpoint inhibitors, are eagerly awaited. Table 4 Characteristics of the ten patients who lived more than 1 year after brain recurrence We found that all 10 of the patients who lived for more than 1 year after brain recurrence received SRS, whether for initial or subsequent brain recurrence, a finding that highlights the importance of local treatment to improve outcome [19,20].
The graded prognostic assessment (GPA) is advocated as a reliable and useful tool for estimating survival and for selecting appropriate treatment strategies for patients with brain metastases [21]. The original GPA score included age, KPS, number of brain lesions, and the presence or absence of extracranial metastases and has been validated for its prognostic potential in lung cancer [22,23]. Although we also identified PS and extracranial disease control as significant prognostic factors, we could not calculate GPA in our patient population because they not only had recurrent disease but they also tended to have a more extensive systemic disease burden than the patients included in the development of the GPA analysis.
This study had some limitations that should be considered in interpreting our results, chief among them being its retrospective nature, broad patient heterogeneity, and relatively small sample size. We cannot rule out the possibility that our findings could reflect selection bias. Thus our results should be considered exploratory and ideally should be validated in a larger, prospective cooperative group study.

Conclusions
The prognosis for patients with SCLC that recurs in the brain after WBI is still rather dismal, but survival in some cases, particularly those with adequate PS and controlled extracranial disease, can be extended if the recurrence is managed therapeutically.