Surgical excision and radiotherapy for brain metastasis from colorectal cancer: How frailty and comorbidity indices influence outcome

The incidence of brain metastasis (BM) from colorectal cancer (CRC) is increasing. This study aims to identify the clinical prognosticators and evaluate the prognostic validity of common comorbidity indices in patients with BM from CRC. This retrospective single‐center study analyzed 93 patients with BM from CRC who received surgical excision and/or radiotherapy. The clinical characteristics and prognostic indices including the 5‐item modified frailty index (mFI‐5) and prognostic nutritional index (PNI) were calculated from the collected patient data and analyzed. In this study, 66 (71.0%), 10 (10.8%), and 17 (18.3%) patients received whole‐brain radiotherapy (WBRT) alone, surgery alone, and surgery plus WBRT, respectively. The median survival of all patients was 3.98 months (IQR: 1.74–7.99). The 2‐ and 3‐year survival rates were 7.4% and 3.7%, respectively. Controlled primary tumor (p = 0.048), solitary BM (p = 0.001), surgery + radiation (p < 0.001), and greater PNI (p = 0.001) were independent predictors of favorable survival. In surgically treated patients, uncontrolled primary tumor (p = 0.006), presence of multiple BM (p < 0.001), and MFI‐5 ≥ 2 (p = 0.038) were independent prognosticators. For patients who received WBRT, the presence of two (p = 0.004) or multiple (p < 0.001) BM and PNI (p < 0.001) were independent survival predictors MFI‐5, multiple BM, and the status of the primary tumor were independent prognosticators for patients who underwent surgery for CRCBM. For patients who received WBRT, the PNI and the number of BM were independent survival predictors.


| INTRODUCTION
Brain metastasis (BM) from colorectal cancer (CRC) is an emerging challenge for neurosurgeons and neuro-oncologists because of the advances in therapeutic modalities and improved survival in patients with CRC. 1 The incidence of BM in patients with CRC was reported to be up to 13%, 1,2 and most patients had BM detected more than 1 year after the diagnosis of metastatic CRC. 3 Most patients with BM from CRC have unfavorable survival outcomes despite aggressive treatment strategies, including surgical excision, radiotherapy, radiosurgery, and systemic therapy. 3,4The reported median survival time after the diagnosis of BM varies from 2.5 to 87 months. 5,65][16][17][18] For instance, a higher MFI-5 score predicted worse early outcomes and 30-day readmissions following CRC surgeries. 19,20e PNI has been shown to be a prognostic indicator for survival in patients with BM from non-small cell lung cancer and glioblastoma. 21,22However, despite the growing literature supporting the validity of these comorbidity scores, studies specifically evaluating the comorbidity indices for predicting survival outcomes in patients with BM from CRC are lacking, and the prognostic values of these comorbidity indices in CRCBM patients remain unclear.
To address the clinical knowledge gap, this study aimed to identify the clinical prognosticators and evaluate the prognostic validity of five comorbidity indices in patients with BM from CRC in a singlecenter retrospective analysis.The knowledge gained in the present study may help clinicians to better identify high-risk patients who may benefit from additional clinical attention.

| METHODS
The present study is a retrospective single-center analysis of patients with BM from CRC.This study was approved by the institutional review board.Considering the retrospective design of the study, the requirement for informed consent was waived.Prognostic indices, including the MFI-5, PNI, systemic immuneinflammation index (SIII), neutrophil-to-lymphocyte ratio (NLR), and platelet-to-lymphocyte ratio (PLR) were calculated from the collected patient data.The extent of brain metastasis was evaluated using gadolinium-enhanced magnetic resonance imaging (MRI).
Categorical variables are presented as numbers and percentages.
Continuous variables are presented as mean ± standard deviation for parametric values, and median and interquartile range (IQR) for nonparametric values.Comparisons of categorical and continuous variables were performed using the χ 2 test and analysis of variance (ANOVA), respectively.Kaplan-Meier survival analysis was used to analyze OS, and the difference between the groups was calculated using the log-rank test.The Cox proportional hazards method was used to create the regression model and estimate hazard ratios.Variables with a p-value of <0.2 in the univariate Cox analysis were selected for further multivariate analysis of OS.Multicollinearity was checked, and none of the variables in the final regression model exhibited a variance inflation factor greater than 2. Statistical tests were conducted using MedCalc 19.7.2 (MedCalc Software, Ltd.).Statistical significance was set at p < 0.05.

| Demographics
A total of 93 patients were included in this study.The mean age of the patients was 63.1 ± 10.9, with 46 (49.5%) patients being male.
Regarding the primary tumor, 57 (61.3%) patients had a rectosigmoid origin, and the primary CRC was under control in 37 (39.8)patients.KRAS mutation status was examined in 57 (61.3%) patients, among whom 25 (26.9%) and 32 (34.4%) had wild type and mutant KRAS, respectively.Prior to the diagnosis of BM, 79 (84.9%) patients received at least one cycle of chemotherapy.Comparing patients with different BM treatment, patients who received surgical treatment had lower MFI-5 score ( p = 0.031) and had higher percentage of postoperative chemotherapy (p = 0.015) (Table 1).

| The status of BM at diagnosis
Among the 93 patients, 52, 11, and 30 patients had single, two, and multiple (≥3) BM detected by gadolinium-enhanced MRI at the time of BM diagnosis, respectively.Infratentorial involvement was observed in 51 (54.8%) patients.For the treatment of BM, 66 (71.0%), 10 (10.8%), and 17 (18.3%)patients received WBRT alone, surgery alone, and surgery plus WBRT, respectively.Fifty-one (54.8%) patients received at least one cycle of chemotherapy after diagnosis of BM.The value of the prognostic indices calculated from the last available clinical and laboratory data prior to BM treatment are shown in Table 1.
Patients who received surgery + WBRT had significantly longer OS compared to patients who received surgery alone and WBRT alone ( p < 0.001) (Table 1).Patients with controlled primary tumors had better survival outcomes compared to those with uncontrolled primary CRC (9.63 vs. 2.66 months; p < 0.001) (Figure 1A).An increased number of BM was associated with worse outcomes (7.66

| Survival prognosticators
Univariate Cox regression showed that patients with multiple BM and greater MFI-5 scores were associated with worse survival, whereas patients with controlled primary tumors, who received surgery + WBRT, and those with a greater PNI were associated with better survival (Table 2).Multivariate Cox regression revealed that a controlled primary tumor ( p = 0.048), multiple BM ( p = 0.001), receiving surgery + WBRT for BM ( p < 0.001), and greater PNI ( p = 0.001) were independent factors that predicted survival (Table 2).The RPA class, SIII, NLR, and PLR were not significantly associated with OS.

| Subgroup analyses for surgically treated and non-surgically treated patients
Subgroup analyses were performed for patients who underwent surgery (surgery and surgery + WBRT) or radiation therapy (WBRT) as the primary treatment for BM.For surgically treated patients, uncontrolled primary tumor (p = 0.006), presence of multiple BM (p < 0.001), and MFI-5 ≥ 2 (p = 0.038) were independent poor prognosticators of OS (Table 3).For patients who received WBRT, the presence of two ( p = 0.004) or multiple ( p < 0.001) BM and PNI (p < 0.001) were independent predictors of OS (Table 4).

| DISCUSSION
Patients with BM from CRC have heterogeneous characteristics and their management strategies can be complex.Optimal risk stratification strategies for these patients are limited.The present study evaluated the prognostic validity of five comorbidity indices in patients with BM from CRC.Our results showed that MFI-5, along with the presence of multiple BM and the status of the primary tumor, were independent prognosticators for patients who underwent surgical resection of CRCBM.Whereas for patients who received WBRT for CRCBM, PNI and the number of BM were independent predictors of survival.
Patients with BM represent a diverse population. 23In state-ofthe-art practice, BM are considered a special site of metastatic cancer rather than a single disease entity. 24In this study, we specifically focused on BM from CRC, since it is an emerging condition due to the advances in therapeutic modalities and improved survival in patients with CRC.Given the diverse patient characteristics, risk stratification for these patients is pivotal and can provide valuable information to facilitate the identification of high-risk patients who may benefit from additional clinical attention and help clinicians in patient counseling and making decisions regarding treatment.Therefore, we aimed to identify patient, CRC, BM, and comorbidity-related factors that could predict patient outcomes.
We investigated the validity of five prognostic indices, including the MFI-5, PNI, SIII, NLR, and PLR in patients with BM from CRC, and our results demonstrated that a greater PNI was an independent predictor of favorable survival.PNI is a nutrition-immune parameter based on serum albumin level and lymphocyte count. 17Hypoalbuminemia can result in dysregulated immune function, abnormal activation of systemic inflammation, and loss of circulating drug efficacy. 25Albumin binds to endogenous ligands and the albumin-drug complex serves as a drug reservoir that can enhance drug biodistribution and bioavailability.6][27] Likewise, lymphocytes are involved in anti-tumoral immunity, and lymphocyte dysfunction or imbalance is associated with cancer progression. 28Previous studies have shown that pretreatment PNI predicted the outcome in patients with CRC, 29,30  Besides the comorbidity indices, our analyses also revealed that the control of the primary CRC and the presence of multiple BM were independent factors that predicted survival.The impact of coexisting extracranial metastases (ECM) and the number of BM at the time of the diagnosis of BM have been reported and were shown to be associated with worse survival. 1,8,10,23,31,32Thurmaier and colleagues reported that the pattern of ECM and the control of primary tumor were also relevant to survival, with patients with concurrent liver and lung metastasis demonstrating the worst outcomes. 23In our results, although ECM was not associated with survival, we showed that patients who had controlled primary tumor had better survival.This finding might suggest that for CRC patients with BM and ECM, as long as the primary tumor and ECM were controlled, resection of BM could potentially improve survival.Additionally, the association between the number of BM detected by MRI at the initial diagnosis and outcomes has also been described in the literature. 8,10,31,32 should be noted that the presence of multiple metastases may influence the decision-making regarding local treatment for BM, although multiple BM are not a contraindication for metastasectomy. 33The primary treatment modality selected for BM was significantly associated with patient outcome.][9][10] Likewise, our results also showed that patients who underwent surgery + WBRT survived longer compared with those treated non-operatively.These findings further demonstrated that patients with BM indeed represent a heterogeneous population, and the effect of potential treatment allocation bias could not be clearly assessed if they were evaluated together. 23 minimize the bias in patient selection and the survival difference between surgery and WBRT, and to further evaluate the validity of the comorbidity indices in each patient population, we performed subgroup analyses in patients who received surgery and WBRT as primary treatment for BM.In the surgery subgroup, MFI-5, but not PNI, T A B L E 3 Subgroup analysis of patients receiving surgical excision of BM. was an independent prognosticator of OS.In the overall analysis, there was indeed an association between a higher MFI-5 score and worse prognosis in the univariate analysis and a similar trend in the multivariate analysis.A probable explanation for this discrepancy is that for patients who underwent surgery, frailty and medical comorbidities may have a greater impact on the perioperative outcomes.
5][36] Moreover, associations between patient frailty and survival outcomes after surgery for BM have also been reported in the literature. 36,37 the WBRT subgroup, the results were similar to that of the overall analysis, with the number of BM and PNI shown to be independent survival prognosticators.Nonetheless, compared with the overall analysis, there was a slight difference in the WBRT subgroup results, which is the effect of the number of BM.In the overall analysis, only the presence of multiple (≥3) BM was associated with worse survival, whereas in the WBRT subgroup, both two and multiple BM were associated with worse survival.Given that patients with one or two BM may be more likely to receive surgical resection and thereby have a better outcome compared with those with multiple BM, it is probable that the survival difference between patients with one or two BM might be less in the surgical or overall cohort.In contrast, in the non-surgical subgroup, since an increasing number of BM represents a more aggressive disease status, the number of BM may be more likely to have a dose-dependent effect on survival. 8,10,23,31e findings of the present study should be interpreted in light of some limitations.First, this study was a single-center study with a limited number of patients.Second, although we reported the KRAS mutation status in the majority of our patients, BRAF V600E mutation status was only available in a small group of patients and the number of these patients was insufficient for analysis.Future studies are required to evaluate the impact of these genes potentially relevant to the survival of CRC BM patients.Third, although surgery and WBRT were compared, the effect of stereotactic radiosurgery (SRS) was not evaluated due to unavailability of SRS in our institution.Furthermore, the clonal evolution of metastatic CRC was reported and might have an impact on patient outcomes, but this information was unavailable in our patients because most patients who had a recurrence of BM did not receive a second tissue proof.Since stereotactic radiosurgery T A B L E 4 Subgroup analysis of patients receiving radiotherapy for BM.
Patients with histologically proven colorectal adenocarcinoma and a histological or radiological diagnosis of BM between January 2010 and December 2021 were included.Patients with nonadenocarcinoma CRC, leptomeningeal carcinomatosis, or secondary malignancies were excluded.All patients had at least 6 months of follow-up.The OS after BM was calculated from the date of radiographic documentation of BM to the date of the patient's death or the last follow-up investigation.Data for all patients were compiled from electronic medical records and clinical notes.Primary treatments for CRC BM were categorized as the following: 1. Patients who underwent surgery and subsequent postoperative whole-brain radiotherapy (WBRT) were assigned as surgery + WBRT group.2. Patients who received surgery alone without postoperative WBRT were assigned as surgery group.3. Patients who received WBRT alone assigned as WBRT group.Steroids were prescribed according to patients' symptom in all groups.Patients who did not received either surgery or WBRT for BM were excluded.

T A B L E 1
Patient characteristics.

F I G U R E 1
Kaplan-Meier survival analysis curve for OS.The survival curves between patients with controlled and uncontrolled primary CRC (A), different number of BM (B), treatment of BM (C), and MFI-5 scores (D).