Title: Lung-molGPA in EGFR-mutated Adenocarcinoma: Prognostic Implications of Molecular Subtypes and Targeted Therapies

EGFR mutations are heterogenous but all carry the same weighting in the Lung-molGPA. The aim of this study was to elucidate the different prognostic implications of molecular subtypes and frontline TKIs in EGFR-mutated lung adenocarcinoma with synchronous brain metastases (BM) using the Lung-molGPA. Medical records were searched in hospital databases from 2011 to 2015. Patients with EGFR-mutated adenocarcinoma and brain metastases who received TKIs were included. The Kaplan-Meier method was used to estimate survival, and multivariate Cox proportional hazard models were used to estimate adjusted hazard ratios (HRs) and 95% confidence intervals (CIs). A total of 256 patients were included with a median overall survival (OS) of 17.2 months. In multivariate analysis of OS, only age (70 versus <70 years, HR:1.71, 95% CI:1.25-2.35, p<0.001), KPS (<70 versus 70, HR:1.71, 95% CI:1.26-2.31, p<0.001), and rare mutations (other versus exon 19 deletions, HR:1.78, 95% CI:1.04-3.05, p=0.037) remained statistically significant. In patients with a Lung-molGPA score 2.5, EGFR molecular subtypes had different median OS (exon 19 deletions versus Leu858Arg versus other, 18.8 vs 12.4 vs 12.1 months, p=0.021). In conclusion, different molecular subtypes treated with frontline TKIs have different prognostic implications in the Lung-molGPA. Further prospective studies are warranted to validate these findings.


Introduction
Brain metastases (BM) are common in patients with lung cancer, with rates of up to 30% at the first diagnosis, and results in shorter overall survival and increased morbidity [1]. The diagnosisspecific graded prognostic assessment (DS-GPA) tool is used to predict median survival. It is calculated using four prognostic factors including age, Karnofsky performance scale (KPS), extracranial metastases (ECM) and the number of BM, and the median survival of patients with lung cancer and BM according to the DS-GPA has been reported to range from 3.0 to 14.8 months [2]. Further, 50-60% of patients with EGFR-mutated lung adenocarcinoma develop BM, possibly due to longer survival, inferior CNS penetration of some EGFR-TKIs and/or actual predilection for BM. The Lung-molGPA index is an update of the DS-GPA that incorporates gene alteration data, and it has been shown to have a better prognostic ability [3]. However, whether frontline treatment and molecular subtypes such as exon 19 deletions, exon 21 substitutions (Leu858Arg, L858R), or other rare mutations have different prognostic implications is largely unknown in these patients. Therefore, we conducted this retrospective study of the prognostic implications of molecular subtypes and frontline TKIs on EGFR-mutated patients with BM to investigate this issue.

Results
Of the 256 patients with advanced EGFR-mutated lung adenocarcinoma and BM included in this study, 43.7% had exon 19 deletions, 49.2% had Leu858Arg, and 7.0% had other mutations. With regards to frontline treatment, 10.2%, 15.6% and 74.2% of the patients received afatinib, erlotinib or gefitinib, respectively. In addition, 9.4% of the patients received WBRT followed by TKIs as their initial treatment, 37.5% of the patients received WBRT and TKIs, and 53.1% of the patients received TKIs alone. Other baseline characteristics including sex, age, ECM, and number of BM are listed in Table S1. There were significant differences in baseline age (p<0.001), KPS (p<0.001), ECM (p<0.001), number of BM (p<0.001), timing of BM (p<0.001), EGFR mutation subtypes (p=0.040), and timing of WBRT (p<0.001) between those with a high (>2.5) and low (2.5) Lung-molGPA score, but not in sex   Reference Reference *Adjusted for sex, age, KPS, extracranial metastases, number of brain metastases, timing of brain metastases, EGFR mutation types, TKIs and timing of RT *Abbreviations: NSCLC, non-small-cell lung cancer; GPA, graded prognostic assessment; KPS, Karnofsky performance status; Del 19, exon 19 deletions; L858R, Leu858Arg; EGFR, epidermal growth factor receptor; TKI, tyrosine kinase inhibitors; RT, radiation therapy; WBRT, whole brain radiation therapy.

Discussion
This study is the first study to address the prognostic implications of molecular subtypes and frontline TKIs on EGFR-mutated BM stratified by Lung-molGPA score. This cohort further validated the usefulness of the Lung-molGPA index in an East Asian population. The median OS in this EGFRmutated lung cancer cohort was 17.2 months, which is slightly longer than the OS reported in pure EGFR-mutated or mixed lung cancer studies [3,[5][6][7][8]. This may be because all of the patients in this cohort received at least one TKI, which may have provided both systematic and intra-cranial disease control [9,10]. In addition, the number of BM (>4 versus 1-4, HR: 1.25, 95% CI: 0.93-1.68, p=0.137) and ECM (present versus absence, HR: 1.18, 95% CI: 0.82-1.69, p=0.383) did not have prognostic value when TKIs were applied in a frontline setting. Although the updated Lung-molGPA was useful in prognostic stratification in these patients, it could not reflect differences in molecular subtypes or frontline TKI treatment. Patients with other rare mutations had a worse OS after adjusting for other factors and frontline TKIs, which reflects the relatively poor efficacy of first-and second-generation TKIs in patients with these mutations [11]. We further stratified the cohort into those with high (>2.5) and low (2.5) Lung-molGPA scores arbitrarily to balance the number of patients, and found different prognostic implications of molecular subtypes and frontline TKI treatment. In the patients with a low Lung-molGPA score, inherent molecular subtypes were more associated with the prognosis, but not the frontline TKI. In addition, the patients with different molecular subtypes had distinct survival benefits with frontline TKIs [12][13][14]. In those with a high Lung-molGPA score, which deemed to be with better prognosis, frontline TKI and inherent molecular subtypes were not associated with different prognostic implications. It seemed to be that the differences within inherent molecular subtypes were diluted with a high Lung-molGPA score. However, whether this observation can be applied to third generation TKIs or osimertinib is not known. Osimertinib has been reported to show remarkable intra-cranial responses and median survival up to 15.2 months in these patients [15]. Whether palliative radiation to BMs is beneficial in EGFR-mutated lung cancer is under intense debate, with rapid advancements in newer TKIs being able to penetrate the CNS [10,11]. In our cohort, only two patients (1.2%) received stereotactic radiosurgery to BM with restrict reimbursement policy, and the timing to WBRT was not related to prognosis. Thus, it seems reasonable to keep WBRT as a last resort when BM worsen under frontline TKIs, since it is associated with a risk of cognitive dysfunction and could further deteriorate the quality of life in these patients. However, further prospective studies are needed to elucidate this issue. In addition, the survival of patients with other rare EGFR mutations and BM has not been addressed before, since these patient comprise less than 5% of all cases of EGFR-mutated non-small-cell lung carcinoma. In this cohort, 77.8% of the patients were treated with frontline gefitinib, and the median OS was 12.2 months.
There are several limitations to this study. In retrospective studies, assessments of local control and intra-cranial progression-free survival can be unreliable, and thus we chose OS as an alternative but most clinically meaningful endpoint. In addition, different survival outcomes were noted with the different frontline TKIs, but such difference was diminished after multivariable analysis. Our finding should be deemed as hypothesis-generating, and larger prospective study empowered to address this issue is eagerly awaiting. In addition, frontline osimertinib is not available with the reimbursement policy in Taiwan. However, this data could still be valuable as real-world evidence Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 31 July 2020 doi:10.20944/preprints202007.0739.v1 since many parts of this world do not have osimertinib for frontline use, partly because of the costeffectiveness issues.

Materials and Methods
We conducted this retrospective hospital-based cohort study at Chang-Gung Memorial Hospital in Taiwan Kaplan-Meier curves were used to estimate OS, and the log-rank test was used to compare times to events between groups. Multivariate analyses and stratified analyses were performed using Cox proportional hazards regression models (hazard ratio, HR). All reported p values were two-tailed, and a p value of 0.05 was considered to indicate statistical significance. Adjustments were made for multiple comparisons. All analyses were performed using SAS version 9.4 (SAS Inc., Cary, NC).

Conclusions
The Lung-molGPA can predict the prognosis in EGFR-mutated adenocarcinoma, and different molecular subtypes and frontline TKI treatment have distinct prognostic implications in these patients. Further studies are warranted to verify these findings. Abbreviations: NSCLC, non-small-cell lung cancer; GPA, graded prognostic assessment; KPS, Karnofsky performance status; Del 19, exon 19 deletions; L858R, Leu858Arg; EGFR, epidermal growth factor receptor; TKI, tyrosine kinase inhibitors; RT, radiation therapy; WBRT, whole brain radiation therapy.