Effect of timing, technique and molecular features on brain control with local therapies in oncogene-driven lung cancer

Background The improved efficacy of tyrosine kinase inhibitors (TKI) mandates reappraisal of local therapy (LT) for brain metastases (BM) of oncogene-driven non-small-cell lung cancer (NSCLC). Patients and methods This study included all epidermal growth factor receptor-mutated (EGFR+, n = 108) and anaplastic lymphoma kinase-rearranged (ALK+, n = 33) TKI-naive NSCLC patients diagnosed with BM in the Thoraxklinik Heidelberg between 2009 and 2019. Eighty-seven patients (62%) received early LT, while 54 (38%) received delayed (n = 34; 24%) or no LT (n = 20; 14%). LT comprised stereotactic (SRT; n = 40; 34%) or whole-brain radiotherapy (WBRT; n = 77; 66%), while neurosurgical resection was carried out in 19 cases. Results Median overall survival (OS) was 49.1 months for ALK+ and 19.5 months for EGFR+ patients (P = 0.001), with similar median intracranial progression-free survival (icPFS) (15.7 versus 14.0 months, respectively; P = 0.80). Despite the larger and more symptomatic BM (P < 0.001) of patients undergoing early LT, these experienced longer icPFS [hazard ratio (HR) 0.52; P = 0.024], but not OS (HR 1.63; P = 0.12), regardless of the radiotherapy technique (SRT versus WBRT) and number of lesions. High-risk oncogene variants, i.e. non-del19 EGFR mutations and ‘short’ EML4-ALK fusions (mainly variant 3, E6:A20), were associated with earlier intracranial progression (HR 2.97; P = 0.001). The longer icPFS with early LT was also evident in separate analyses of the EGFR+ and ALK+ subsets. Conclusions Despite preferential use for cases with poor prognostic factors, early LT prolongs the icPFS, but not OS, in TKI-treated EGFR+/ALK+ NSCLC. Considering the lack of survival benefit, and the neurocognitive effects of WBRT, patients presenting with polytopic BM may benefit from delaying radiotherapy, or from radiosurgery of multiple or selected lesions. For SRT candidates, the improved tumor control with earlier radiotherapy should be weighed against the potential toxicity and the enhanced intracranial activity of newer TKI. High-risk EGFR/ALK variants are associated with earlier intracranial failure and identify patients who could benefit from more aggressive management.


INTRODUCTION
Approximately 20% of non-small-cell lung cancers (NSCLC) harbor actionable driver mutations (dm-NSCLC), with important consequences for patient management and outcome. 1 Most frequent are mutations in the epidermal growth factor receptor gene (EGFR) occurring in 10%-15% of adenocarcinomas, followed by anaplastic lymphoma kinase gene (ALK) fusions in up to 5%. 2 Approximately 25% of these patients present with brain metastases (BM) at the time of diagnosis, which appear, at least in EGFR þ tumors, to be associated with a worse prognosis. [3][4][5] The strategies to treat central nervous system (CNS) involvement in dm-NSCLC are influenced by the increasing availability of targeted drugs with better CNS penetration and antitumor activity than conventional chemotherapy. [5][6][7] In particular, the next-generation tyrosine kinase inhibitors (TKI) osimertinib, alectinib, brigatinib, and lorlatinib achieve durable CNS responses in~80% of patients compared with 40%-60% for older targeted substances. [8][9][10][11] Besides, median overall survival (OS) currently exceeds 3 years for EGFR þ and 5 years for ALK þ NSCLC patients, 12,13 which increases the risk of long-term toxicity from local CNS treatments (LT). 14,15 Therefore, the timing and technique of LT for BM in dm-NSCLC are highly controversial. Conflicting results have been reported in previous retrospective analyses in EGFR þ NSCLC regarding the ability of LT to improve intracranial tumor control and OS, [16][17][18][19] while no similar study exists for ALK þ patients, and prospective data are lacking. Moreover, even though molecular tumor properties are increasingly recognized as crucial determinants of clinical outcome in both EGFR þ and ALK þ NSCLC, 4,20-22 their potential importance for management of brain disease remains unclear. The current study addresses these questions by analyzing consecutive patients treated over 10 years in a large European thoracic oncology center.

Study population, treatment, and data collection
This study was approved by the Heidelberg University ethics committee (S-172/2018 and S-145/2017) and carried out following institutional guidelines and the Declaration of Helsinki in its current version. Since this was a noninterventional, retrospective study, informed consent was obtained whenever possible, but was not required for every participant.
Included were all EGFR þ /ALK þ NSCLC patients diagnosed with BM at our institution between 2009 and 2019. Patient characteristics and treatment details were systematically collected from the medical records. 23 Histological diagnosis and molecular profiling of NSCLC were carried out at the Institute of Pathology, Heidelberg University Hospital, according to the criteria of the current WHO Classification (2015) for lung cancer. 24 Next-generation sequencing (NGS) was carried out on a semiconductor-based platform (Ther-moFisher Scientific, Waltham, MA) with custom panels covering 38-42 genes considered relevant for lung cancer biology, which included all EGFR exons, and TP53 exons 4-10 (DNA-based), as well as known ALK fusion variants (RNAbased), as published. 25 The decision for early (i.e. within 30 days of TKI start) versus delayed LT was made by our Multidisciplinary Thoracic Oncology Tumor Board (MTB) considering various parameters, for example, the presence or absence of symptoms (with preferential use of early LT for symptomatic patients), or the location of brain lesions (with preferential use of early LT for infratentorial lesions). The decision for stereotactic radiotherapy (SRT) versus whole-brain radiotherapy (WBRT) was also made by our MTB, which recommended SRT for 4 metastases until 2016 (in accordance with the guidelines of the German Society for Radio-Oncology), 26 and for 10 metastases thereafter. For SRT, either stereotactic radiosurgery (SRS) or hypofractionated stereotactic radiotherapy (HFSRT), the CyberKnife M6 system or a linear accelerator (LINAC) adapted for radiosurgery were used. Target volume delineation and treatment planning for SRT were based on highresolution contrast-enhanced computed tomography (CT), as well as magnetic resonance imaging (MRI) scans and carried out as previously described. 27,28 Employed safety margins depended on treatment technique and ranged between 1 mm for CyberKnife and 2-3 mm for LINAC-based radiosurgery. The target volume for post-operative SRT encompassed the resection cavity with a safety margin of 3-4 mm. Post-operative cavities were treated with HFSRT, as were lesions with a diameter larger than 3 cm. SRT doses ranged between 18 and 20 Gy margin dose with prescription to the enclosing 70% isodose for CyberKnife treatments and 80% isodose for LINAC-based radiosurgery. Doses for HFSRT ranged between 30 and 35 Gy in 6-7 fractions. WBRT was delivered using conventional techniques at typically 30 Gy in 10 fractions, five times a week. Alternate dose regimens of 35 Gy in 14 fractions or 40-42 Gy in 20-21 fractions were used in eight cases. Follow-up consisted of regular high-resolution cranial MRI and/or contrast-enhanced CT scans. Intracranial tumor status was assessed by neuroradiologists according to the criteria for Response Assessment in Neuro-Oncology for brain metastases (RANO-BM). 29

Statistical analysis
Descriptive statistics for baseline variables included mean (with standard deviation, SD) and/or median values (with range or interquartile range, IQR) for continuous variables, and absolute or relative frequencies for categorical variables. Follow-up time was calculated using the reverse KaplaneMeier (KM) method. 30 OS was calculated from BM diagnosis to the date of death or last follow-up. Intracranial progression-free survival (icPFS) was calculated from BM diagnosis to last imaging follow-up or radiologic progression. OS and icPFS were analyzed according to KM. The prognostic influence of baseline characteristics on OS and icPFS was analyzed using proportional hazards Cox regression. Baseline variables were systematically examined for their relationship with clinical endpoints in order to uncover potential confounders. Multivariable modeling included parameters showing significant associations with outcome in univariable analysis, and those of special clinical interest. Variable selection was verified with a component-wise gradient boosting algorithm to optimize the C-index 31,32 using the R-package mboost with a step length of 0.0002 and initial number of iterations of 2000 for OS and 1000 for icPFS, due to different numbers of patients and events. Since this is a retrospective exploratory data analysis, P values are of descriptive nature. Statistical analyses were carried out with the R software (v.3.6.2; R Core Team, 2019) and SPSS v24 (IBM, Armonk, NY).

DISCUSSION
The optimal timing and technique of LT in dm-NSCLC with BM are subjects of ongoing debate. Our study shows that early LT improves icPFS, but not OS of EGFR þ and ALK þ patients, independent of the RT technique, i.e. WBRT versus SRT (Tables 2 and 3). Therefore, the previously postulated special effect of WBRT on icPFS due to the eradication of diffuse micrometastases 33,34 might be less relevant in dm-NSCLC patients, in which the superior brain efficacy of TKI compared with chemotherapy additionally contributes to intracranial tumor control. Along the same lines, the presence of 4 versus >4 brain lesions did not influence OS or icPFS in our study either, supporting the notion that the degree of initial intracranial spread is not crucial. 35 An important exception was patients with a solitary BM, who have been highlighted as a particularly favorable subgroup in previous reports and had longer icPFS in our study, as well (Table 2). 36 Considering the lack of efficacy advantage for WBRT over SRT, and its deleterious effect on cognitive function and quality of life, 14,15,37,38 WBRT should be avoided whenever possible, and delayed LT might be reasonable when application of SRT is technically limited due to a very large number of lesions. Of note, while use of SRT was restricted to patients with 1-3 metastases in the past, it is currently being extended with equivalent OS results for up to 10 BM, and its role for >10 BM is also being reappraised. 35,[39][40][41] Emerging individualized concepts for extensive, polytopic brain involvement include SRT for the symptomatic, largest or fastest progressing lesions, in analogy to the concept of oligoprogression, 42 and control of the remaining BM by CNS-active systemic therapies. However, the exact workflows and patient selection criteria for application remain to be defined, ideally in the form of prospective clinical trials. When WBRT is inevitable, hippocampal avoidance in combination with memantine can also reduce cognitive impairment. 43,44 It is worth noting that the addition of EGFR TKI to WBRT does not further increase neurotoxicity according to a systematic review. 45 Another important issue is whether upfront LT could be altogether omitted in favor of TKI-only treatment. Prospective evidence to support this strategy is available only for asymptomatic BM from EGFR þ /ALK þ NSCLC, and only with first-line use of the newer TKI osimertinib, alectinib, brigatinib, or lorlatinib, which show very good intracranial efficacy. [8][9][10][11] For symptomatic BM, surgery and/or radiotherapy are generally preferable in order to quickly reduce mass effects to the healthy brain and alleviate neurological symptoms. 6 Also, for EGFR þ patients with BM receiving first-/second-generation EGFR inhibitors, a meta-analysis of 1086 patients from seven studies showed that upfront RT in addition to TKI resulted in a better icPFS and OS compared with TKI only, especially if the number of BMs was limited. 16 Similar results were also observed in a second meta-analysis of 24 studies including 2810 EGFR þ patients: RT plus first-/ second-generation TKI resulted in a better response, longer OS, and longer icPFS than TKI monotherapy. 46 However, other investigators have reported different results, for example, that the OS benefit from a combined upfront RT/ TKI approach is restricted to patients with 1-4 BM only according to a retrospective analysis of 176 EGFR þ cases, 17 or that upfront WBRT/TKI does not improve OS, but only  icPFS, and this only for patients with >3 BM. 19 In our study, early LT improved icPFS, but not OS in EGFR þ and ALK þ NSCLC with BM, regardless of the number of lesions (4 versus >4) and the applied technique (WBRT versus SRT, Tables 2 and 3). Summarizing the available evidence, prolongation of icPFS from a combined upfront RT/TKI treatment of BM in dm-NSCLC appears to be reproducible, but the potential OS benefit remains controversial and becomes less likely, as more potent EGFR (osimertinib) and ALK inhibitors (alectinib, brigatinib, lorlatinib) enter the first-line setting. Prospective data are urgently needed to clarify the exact conditions under which upfront RT might be safely omitted in favor of TKI-only treatment with these newer compounds, as well as the neurocognitive side-effects and influence on quality of life for each strategy. Such a randomized phase II study of EGFR þ patients is currently ongoing (OUTRUN, NCT03497767). For ALK þ NSCLC, data on the effect of early versus delayed LT on patient survival are scarce. Despite the lack of OS benefit, early brain radiotherapy prolonged icPFS in our study and should therefore be considered for crizotinibtreated patients, which comprised the majority (n ¼ 28/ 33; 85%) of our ALK þ cases. 47,48 The improved intracranial control when adding RT to crizotinib was also evident in a retrospective analysis of patients enrolled in the PROFILE 1005 and 1007 studies. 49 At the mechanistic level, the tumoricidal effect of radiotherapy not only complements the weak activity of crizotinib, which achieves a brain response rate of only 30%-50% as monotherapy, 49,50 but also increases permeability of the blood-brain barrier, as could be shown in pharmacokinetic studies. 51 On the other hand, newer ALK inhibitors show intracranial response rates comparable to these of radiotherapy, i.e.~80% (44), which could obviate benefit from additional RT, as suggested by a small retrospective series of patients treated with brainpenetrant TKI presented recently. 52 Real-world data are not yet mature for this comparison, because the time since first-line approval of alectinib and brigatinib (December 2018 and May 2020, respectively in Europe) is still shorter than their median first-line PFS (>2 years). Whether cranial RT can safely be deferred for newly diagnosed ALK þ NSCLC patients with BM receiving highly brain-active TKI, remains unclear and will need to be addressed in future studies.
To our knowledge, our study is the first to highlight the impact of molecular tumor characteristics on intracranial disease control. While non-del19 EGFR mutations, 'short' EML4-ALK fusions (mainly variant 3), and presence of TP53 co-mutations have all been linked to earlier systemic treatment failure in both EGFR þ and ALK þ NSCLC, [53][54][55][56] our results show that high-risk oncogene variants are more important that TP53 status for intracranial disease control ( Table 2 and Supplementary Figure S1, available at https:// doi.org/10.1016/j.esmoop.2021.100161). This is in accordance with the lack of association between TP53 mutations and brain involvement in both EGFR þ and ALK þ NSCLC, while unfavorable oncogene variants increase metastatic potential (EML4-ALK V3) and/or decrease TKI sensitivity (EML4-ALK V3 and non-del19 EGFR mutations). 57-59 6 months] for EGFR þ patients with early local therapy (LT) versus 11.5 months (95% CI 8.8-14.2 months) for EGFR þ patients with delayed LT (P ¼ 0.267 in univariable testing, and P ¼ 0.027 in multivariable testing, Table 3). (B) Median icPFS was 36.0 months (CI not available due to the low number of events) for ALK þ patients with early LT versus 9.4 months (95% CI 5.5-13.2 months) for ALK þ patients with delayed LT (P ¼ 0.002 in univariable and P ¼ 0.045 in multivariable testing, Table 3). Therefore, patients with these molecular alterations constitute a higher 'brain risk' population, which would probably benefit more from closer radiologic monitoring as well as earlier and more aggressive LT.
The main limitations of our study are its retrospective character and relatively small number of patients, especially with the rare ALK þ disease. Given the heterogeneity of our cohort (Table 1), potential confounders were controlled by inclusion in multivariable modeling (Table 2). For the interpretation of our results, it is important to consider that the decisions for WBRT versus SRT, as well as for early versus delayed LT are frequently linked to the presence of unfavorable BM characteristics, 60,61 such as larger and/or polytopic lesions, more symptomatic and/or requiring steroids, which was also evident in our cohort (Table 1). Despite this poor prognostic profile of BM, early LT could improve intracranial disease control, and thus also prevent earlier BM-related death, which highlights the exquisite antitumor potency of radiotherapy. Nonetheless, the fact that this prolongation of icPFS did not translate to a longer OS ( Table 2) suggests that adverse biologic characteristics of tumors causing aggressive brain involvement and necessitating early LT, presumably facilitate progression at other (extracranial) sites, which then becomes the limiting factor for survival. The main strengths of our study are the homogeneous management of our patients, who were consecutive, TKI-naive, and treated at the same large tertiary cancer center; the standardized molecular profiling with combined DNA/RNA NGS also carried out for all patients in the same institution; the systematic in-house MRIbased follow-up; as well as a dedicated clinical registry, which ensured consistency of data capture and processing. 23 Furthermore, the present study is to our knowledge the first real-world analysis to include a reasonably sized ALK þ NSCLC cohort evaluable for OS, and the first to systematically examine the effect of molecular tumor characteristics on intracranial disease control.
In summary, a combined TKI/early LT strategy in EGFR þ and ALK þ NSCLC with BM improves icPFS but not OS, regardless of the radiotherapy technique (SRT or WBRT) and number of brain lesions. This could be linked to the generally more adverse prognostic profile of patients chosen for the early LT strategy in the clinical routine. Considering the lack of OS benefit and the toxicities of WBRT compared with SRT, decisions about the timing and technique of radiotherapy in dm-NSCLC should be individualized based on the patient's life expectancy. In the particular case of polytopic BM early-on, WBRT should be avoided by delaying RT under MRI surveillance or by SRT of multiple BM. High-risk oncogene variants, i.e. non-del19 EGFR mutations and EML4-ALK V3, confer earlier intracranial failure and identify patients who could benefit from more aggressive surveillance and treatment strategies.