We performed a retrospective study on surgically treated patients affected by BMs from diagnosed NSCLC. The patients were considered eligible for surgical treatment if they had a good preoperative performance status measured before surgery (Karnopsky performance status, KPS >50) and estimated overall survival of >3 months (according to the radiation therapy oncology group and the grade prognostic assessment rankings) [26,27]. The estimated target of the surgical procedure was the gross-total, near-total- or subtotal resection of the lesions. The molecular analysis of EGFR and ALK mutations was carried out on the brain lesion. Patients were included if, in the postoperative period, they could undergo an adjuvant chemo-radiotherapy and a follow-up program. We excluded all cases with incomplete or incorrect data on clinical, radiological, surgical, and follow-up records.
All patients underwent a general medical, neurological, and oncologic evaluation at admission. For all the included patients, we recorded patient-related variables such as sex, age, peri and post-operative KPS, clinical presentation, survival, antiepileptic prophylaxis and treatment, the incidence of postoperative seizures, and tumor- and surgery-related variables: number, location, and side of the lesions, tumor and edema volume, morphology, onset about the primary tumor and molecular profile (EGFR, ALK, and PD-L1). In particular, the specimens used in this study were examined for EGFR and ALK mutations. Immunohistochemistry with CDX-2, CK7, CK20, TTF-1 and Napsin-A expression was routinely performed in the Department of Neuropathology of our University Hospitals. All patients were submitted to a standardized and shared preoperative and operative Protocol reported in all studies referred to intracranial tumors [28-32].
Overall Survival (OS) was recorded in months; it was measured from the date of diagnosis to the date of death or the date of the last contact if alive. Clinical information was obtained by the respective institutions' digital database, whereas telephone interviews obtained OS data. A particular focus was centered on the performance status expressed as KPS results in a dichotomy data (> and < 70): such parameter was considered, as previously observed, as associated with Survival  to the presentation of BM  and in general as a protective factor when >70 [34-36]. In particular, it was recorded in three different moments:
- Before surgery at the time of diagnosis
- At 30 days after surgery (early post-operative evaluation and
- At the end of the adjuvant treatment (the moment of the last outpatient evaluation).
Data sources and Quantitative variables
The extent of resection (EOR) was determined by comparing the MR images obtained before surgery and the first early MRI after surgery, following RANO criteria. EOR was coded in a 3-step ordinal variable as reported elsewhere : Gross-Total Resection (GTR) <2 mm3 residual lesions; Near-Total Resection (NTR) (≧2 to <5 mm3), and Subtotal Resection (STR) (≧5 mm3).
In the case of GTR, “tumor progression” was defined as the first MRI scan demonstrating the presence of pathologically enhancing tissue characterized by an MRI pattern (mainly relying on Perfusion Weighted Imaging) inconsistent with a cerebral radiation injury (which is, in fact, a “pseudo-progression”). In incomplete resections (NTR/STR), a volumetric increase of the residual disease detected at the first postoperative MRI scan was considered disease progression. A close-range dedicated neuro-imaging follow-up program was routinely performed in our Institutions . This program included:
A standard early (maximum 24 hours after surgery) postoperative volumetric brain MRI.
At approximately one month from surgery (25-35 days), a volumetric brain MRI scan was repeated for a first step follow-up control and information for the radiation treatment planning. A volumetric brain MRI scan was performed every three months at the end of irradiation. We performed a complete outpatient clinical and neurological outpatient re-evaluation at every radiological reevaluation.
Size, statistics, and a potential source of Bias
The study size is given by the selection of the inclusion criteria. As previously stated, we addressed no missing data because incomplete records were an exclusion criterion. The sample was analyzed with SPSS v18 (SPSS Inc., Released 2009, PASW Statistics for Windows, Version 18.0, Chicago, Illinois, USA) to outline potential correlations between the investigated variables. Comparisons between nominal variables have been made with the Chi2 test. EOR, OS, and PFS mean, edema, and lesions volume, and their correlations with ALK and EGFR mutations were compared with One Way and Multivariate ANOVA analysis and Contrast analysis and Post-Hoc Tests. Kaplan-Meier survival analysis assessed survival. Continuous variables correlations have been investigated with Pearson’s Bivariate correlation. The threshold of statistical significance was considered p<.05. It is expected that when the underlying pathology is known and the BM is diagnosed based on radiological follow-up, the Pathological Anatomy laboratory does not carry out the molecular analysis of EGFR and ALK again to have a substantial loss of data. To avoid methodological biases, we, therefore, have to consider them as missing data rather than insert the molecular data found from other sources (e.g., the molecular data on the primary tumor)
The Institutional Review Board approved the informed consent of our Institution. Before the surgical procedure, all the patients gave informed, written explicit consent after appropriate information. Data reported in the study have been completely anonymized. No treatment randomization has been performed. This study is perfectly consistent with the Helsinki Declaration of Human Rights in Medical Research.