New prediction model for brain metastasis in germ cell tumours

Researchers have identified 5 variables that can predict brain metastasis in germ cell tumours (GCTs). Clinicians can use the resulting prediction model to identify high-risk patients.

The presence of brain metastasis is a negative prognostic factor in patients who have a GCT; Dr Ryan Ashkar (Indiana University Simon Comprehensive Cancer Centre, Indiana, USA) and his team sought to identify factors to establish a model to predict patients at risk for developing brain metastasis [1]. To this end, 2,291 patients from a prospectively maintained database were identified who had been treated for testicular cancer at Indiana University between January 1990 and September 2017.

Patients were separated into 2 groups based on the presence (n=154; 6.7%) or absence (n=2,137; 93.3%) of brain metastasis. Kaplan-Meier analysis demonstrated a 2-year progression-free survival of 17% versus 65% (P<0.001) in those with brain metastasis versus those without. The same method showed an overall survival rate of 62% in patients with brain metastasis compared with 91% (P<0.001) in those without.

The data was then divided into datasets for training and validation of the model with equal numbers of events in each dataset. Logistic regression was used to identify a predictive model for the presence of brain metastasis. The resulting model identified weighted variables that can be used to predict progression to brain metastasis:

• age of ≥40 at time of diagnosis (1 point);
• b-hCG level ≥5,000 prior to chemotherapy (1 point);
• the presence of bone or pulmonary metastases, with pulmonary metastases further subdivided into those <3 cm or ≥3 cm:
  
  
  
  • bone metastasis (1 point);
  • pulmonary metastasis <3 cm (2 points);
  • pulmonary metastasis ≥3 cm (3 points); and
• histology demonstrating a predominance of choriocarcinomatous cells (2 points).

The total number of points yields a corresponding probability of progression to brain metastasis (see Figure). 

Figure: Probability of brain metastasis, as characterised by the prediction model [1]



Dr Ashkar concluded that this model can predict the occurrence of brain metastasis in patients with metastatic GCTs and recommend its use by clinicians to identify patients at risk of developing brain metastasis.

1. Ashkar R. Prediction model for brain metastasis (BM) in patients with metastatic germ-cell tumours (mGCT) accounting for size of pulmonary metastases. Abstract 378, ASCO Genitourinary Cancers Symposium, 11–13 February 2021.

 

Copyright ©2021 Medicom Medical Publishers



Posted on 24 March, 202118 June, 2021