A nomogram for predicting brain metastasis in patients with de novo stage IV breast cancer

Estimating the risk of brain metastasis for patients newly diagnosed with cancer

BACKGROUND

Brain metastases (BM) affect clinical management and prognosis but limited resources exist to estimate BM risk in newly diagnosed cancer patients. Additionally, guidelines for brain MRI screening are limited. We aimed to develop and validate models to predict risk of BM at diagnosis for the most common cancer types that spread to the brain.

METHODS

Breast cancer, melanoma, kidney cancer, colorectal cancer (CRC), small cell lung cancer (SCLC), and non-small cell lung cancer (NSCLC) data were extracted from the National Cancer Database to evaluate for the variables associated with the presence of BM at diagnosis. Multivariable logistic regression (LR) models were developed and performance was evaluated with Area Under the Receiver Operating Characteristic Curve (AUC) and random-split training and testing datasets. Nomograms and a Webtool were created for each cancer type.

RESULTS

We identify 4,828,305 patients from 2010-2018 (2,095,339 breast cancer, 472,611 melanoma, 407,627 kidney cancer, 627,090 CRC, 164,864 SCLC, and 1,060,774 NSCLC). The proportion of patients with BM at diagnosis is 0.3%, 1.5%, 1.3%, 0.3%, 16.0%, and 10.3% for breast cancer, melanoma, kidney cancer, CRC, SCLC, and NSCLC, respectively. The average AUC over 100 random splitting for the LR models is 0.9534 for breast cancer, 0.9420 for melanoma, 0.8785 for CRC, 0.9054 for kidney cancer, 0.7759 for NSCLC, and 0.6180 for SCLC.

CONCLUSIONS

We develop accurate models that predict the BM risk at diagnosis for multiple cancer types. The nomograms and Webtool may aid clinicians in considering brain MRI at the time of initial cancer diagnosis.

Development and external validation of a prediction model for brain metastases in patients with metastatic breast cancer

BACKGROUND

Breast cancer patients with brain metastasis (BM) have a poor prognosis. This study aims to identify the risk factors of BM in patients with metastatic breast cancer (MBC) and establish a competing risk model for predicting the risk of brain metastases at different time points along the course of disease.

METHODS

Patients with MBC admitted to the breast disease center of Peking University First Hospital from 2008 to 2019 were selected and retrospectively analyzed to establish a risk prediction model for brain metastases. Patients with MBC admitted to eight breast disease centers from 2015 to 2017 were selected for external validation of the competing risk model. The competing risk approach was used to estimate cumulative incidence. Univariate Fine-Gray competing risk regression, optimal subset regression, and LASSO Cox regression were used to screen potential predictors of brain metastases. Based on the results, a competing risk model for predicting brain metastases was established. The discrimination of the model was evaluated using AUC, Brier score, and C-index. The calibration was evaluated by the calibration curves. The model was assessed for clinical utility by decision curve analysis (DCA), as well as by comparing the cumulative incidence of brain metastases between groups with different predicted risks.

RESULTS

From 2008 to 2019, a total of 327 patients with MBC in the breast disease center of Peking University First Hospital were admitted into the training set for this study. Among them, 74 (22.6%) patients developed brain metastases. From 2015 to 2017, a total of 160 patients with MBC in eight breast disease centers were admitted into the validation set for this study. Among them, 26 (16.3%) patients developed brain metastases. BMI, age, histological type, breast cancer subtype, and extracranial metastasis pattern were included in the final competing risk model for BM. The C-index of the prediction model in the validation set was 0.695, and the AUCs for predicting the risk of brain metastases within 1, 3, and 5 years were 0.674, 0.670, and 0.729, respectively. Time-dependent DCA curves demonstrated a net benefit of the prediction model with thresholds of 9-26% and 13-40% when predicting the risk of brain metastases at 1 and 3 years, respectively. Significant differences were observed in the cumulative incidence of brain metastases between groups with different predicted risks (P < 0.05 by Gray's test).

CONCLUSIONS

In this study, a competing risk model for BM was innovatively established, with the multicenter data being used as an independent external validation set to confirm the predictive efficiency and universality of the model. The C-index, calibration curves, and DCA of the prediction model indicated good discrimination, calibration, and clinical utility, respectively. Considering the high risk of death in patients with metastatic breast cancer, the competing risk model of this study is more accurate in predicting the risk of brain metastases compared with the traditional Logistic and Cox regression models.

Novel models by machine learning to predict prognosis of breast cancer brain metastases

BACKGROUND

Breast cancer brain metastases (BCBM) are the most fatal, with limited survival in all breast cancer distant metastases. These patients are deemed to be incurable. Thus, survival time is their foremost concern. However, there is a lack of accurate prediction models in the clinic. What's more, primary surgery for BCBM patients is still controversial.

METHODS

The data used for analysis in this study was obtained from the SEER database (2010-2019). We made a COX regression analysis to identify prognostic factors of BCBM patients. Through cross-validation, we constructed XGBoost models to predict survival in patients with BCBM. Meanwhile, a BCBM cohort from our hospital was used to validate our models. We also investigated the prognosis of patients treated with surgery or not, using propensity score matching and K-M survival analysis. Our results were further validated by subgroup COX analysis in patients with different molecular subtypes.

RESULTS

The XGBoost models we created had high precision and correctness, and they were the most accurate models to predict the survival of BCBM patients (6-month AUC = 0.824, 1-year AUC = 0.813, 2-year AUC = 0.800 and 3-year survival AUC = 0.803). Moreover, the models still exhibited good performance in an externally independent dataset (6-month: AUC = 0.820; 1-year: AUC = 0.732; 2-year: AUC = 0.795; 3-year: AUC = 0.936). Then we used Shiny-Web tool to make our models be easily used from website. Interestingly, we found that the BCBM patients with an annual income of over USD$70,000 had better BCSS (HR = 0.523, 95%CI 0.273-0.999, P < 0.05) than those with less than USD$40,000. The results showed that in all distant metastasis sites, only lung metastasis was an independent poor prognostic factor for patients with BCBM (OS: HR = 1.606, 95%CI 1.157-2.230, P < 0.01; BCSS: HR = 1.698, 95%CI 1.219-2.365, P < 0.01), while bone, liver, distant lymph nodes and other metastases were not. We also found that surgical treatment significantly improved both OS and BCSS in BCBM patients with the HER2 + molecular subtypes and was beneficial to OS of the HR-/HER2- subtype. In contrast, surgery could not help BCBM patients with HR + /HER2- subtype improve their prognosis (OS: HR = 0.887, 95%CI 0.608-1.293, P = 0.510; BCSS: HR = 0.909, 95%CI 0.604-1.368, P = 0.630).

CONCLUSION

We analyzed the clinical features of BCBM patients and constructed 4 machine-learning prognostic models to predict their survival. Our validation results indicate that these models should be highly reproducible in patients with BCBM. We also identified potential prognostic factors for BCBM patients and suggested that primary surgery might improve the survival of BCBM patients with HER2 + and triple-negative subtypes.

HER2-low breast cancer could be associated with an increased risk of brain metastasis

PURPOSE

The HER2-low breast cancer is a newly recognized entity with the clinical characteristics is yet to be defined. We hypothesized that HER2-low breast cancer could lead to an increased rate of brain metastases in patients with localized breast cancer. We tested this hypothesis in a large cohort of breast cancer patients with long follow-up.

METHODS

We included 2686 adult breast cancer patients followed up in Hacettepe University Cancer Center. Patients with 1 + positive HER2 expression and 2 + HER2 expression with a negative FISH were categorized as HER2-low disease. We evaluated the brain metastasis risk with binary logistic regression analyses and reported odds ratios (OR) with 95% confidence intervals (CI).

RESULTS

During a median 95.4 (IQR 72.6-123.1) month follow-up, 184 patients developed brain metastasis (6.9%). The brain metastases were developed in 5.1% of the patients with HER2-negative disease, 8.5% of the patients with HER2-low disease, and 10.1% of the patients with HER2-positive disease. A multivariable binary logistic regression model demonstrated an increased risk of brain metastasis in patients with HER2-low disease (OR: 1.611, 95% CI 1.055-2.460, p = 0.027) and in HER2-positive patients (OR: 1.837, 95% CI 1.308-2.580, p < 0.001). Additionally, HR + -HER2-low disease was associated with a decreased DFS compared to HR + -HER2-negative disease (p = 0.008).

CONCLUSION

In this study, we observed an increased risk of brain metastasis in localized breast cancer patients with HER2-low disease. We think that a high level of vigilance and a low threshold for brain imaging could benefit HER2-low breast cancer patients similar to the patients with HER-positive disease.