Clinical Context: Brain metastases (BM) from lung carcinoma account for 50% of all brain metastases, and stereotactic radiosurgery (SRS) is the preferred treatment for limited BM. However, predicting which lesions will respond versus progress after SRS remains challenging, particularly at 6-12 months.
Study Goal: This retrospective study evaluated whether early post-SRS dynamic contrast-enhanced MRI (DCE-MRI) and dynamic susceptibility contrast MRI (DSC-MRI) parameters could predict midterm (6-12 month) treatment response.
Key Finding: Early post-SRS K-trans (the vascular permeability transfer constant from DCE-MRI) was the best predictor of midterm response, with absolute post-SRS K-trans achieving AUC 0.75 and 78% accuracy.
Population: 26 patients with 54 lung carcinoma BMs (NSCLC 67%, SCLC 33%) underwent SRS with pre- and post-SRS perfusion MRI at 4-8 weeks, with midterm follow-up at 6-12 months.
K-trans (DCE-MRI): K-trans (volume transfer constant) measures the rate at which gadolinium contrast moves from plasma into extravascular extracellular space, reflecting vascular permeability and blood-brain barrier integrity. Post-radiation decreases in K-trans indicate vascular normalization after successful treatment.
Ve (Extravascular Extracellular Space): Ve represents the fraction of tissue occupied by extracellular space; it relates to tumor cellularity and the extracellular matrix content, with changes after SRS potentially reflecting tumor cell death and matrix remodeling.
Vp (Plasma Volume Fraction): Vp reflects intravascular blood volume density and may serve as a marker of tumor vascularity, complementing K-trans in characterizing the vascular microenvironment of brain metastases.
nCBV (DSC-MRI): Normalized cerebral blood volume (nCBV), derived from DSC perfusion imaging, estimates tumor blood volume through contrast susceptibility effects, providing information about tumor-induced angiogenesis and vascular remodeling.
SRS Protocol: Patients were treated with Leksell Gamma Knife or Elekta Versa HD. Single-fraction SRS was delivered in 70% of lesions (median 20 Gy), with 30% receiving fractionated SRS (20-30 Gy total) based on lesion size and location.
MRI Protocol: DCE-MRI and DSC-MRI were performed on a 1.5T scanner at baseline (pre-SRS) and repeated at 4-8 weeks post-SRS. The Extended Tofts Linear Model was applied to DCE data to derive K-trans, Ve, and Vp maps.
Response Classification: Midterm outcomes at 6-12 months were classified per RANO-BM criteria as responders (complete response, partial response, or stable disease) or non-responders (progressive disease).
ROI Analysis: ROIs were manually drawn on the enhancing lesion by an experienced neuroradiologist with 7 years of specialized experience, with a blinded second reviewer ensuring conformity to enhancing lesion boundaries.
Response Distribution: 40 of 54 BMs (74%) were responders (including 32 with complete or partial response), while 14 (26%) were non-responders. Progressive lesions showed a mean 5.5-fold volume increase, while responders demonstrated 60% mean volume reduction.
K-trans as Best Predictor: Responders showed significantly lower absolute post-SRS K-trans (median 0.015 vs. 0.035 min-1; p = 0.005) and a greater proportional decrease from baseline (-27% vs. +13%; p = 0.017) compared to non-responders.
Ve Significance: Post-SRS Ve was significantly lower in responders (p = 0.009), consistent with reduced extracellular space volume reflecting tumor cell loss after effective radiosurgery.
DSC-MRI Limitations: Neither dynamic change in nCBV nor post-SRS nCBV alone significantly predicted midterm response; however, combining post-SRS nCBV with K-trans slightly improved performance (AUC 0.76 vs. 0.75 for K-trans alone).
Early Salvage Intervention: Identifying non-responders at the 4-8 week post-SRS MRI (rather than waiting for the 6-month scan showing clear progression) could enable earlier consideration of additional SRS, WBRT, or systemic therapy changes.
Differentiating Progression from Radiation Necrosis: K-trans elevation in non-responders reflects ongoing blood-brain barrier disruption from viable tumor, while radiation necrosis (a benign post-treatment effect) tends to show different vascular permeability patterns.
Personalized Follow-Up: Patients with high post-SRS K-trans could be scheduled for more frequent follow-up imaging and early multidisciplinary review, while those with low K-trans (predicted responders) could have standard surveillance intervals.
Integration with Clinical Factors: Combining K-trans with clinical variables (histology, extracranial disease status, NSCLC vs. SCLC) may further improve midterm response prediction beyond imaging parameters alone.
Small Sample Size: With only 26 patients and 54 BMs, this study is underpowered for robust statistical conclusions. Confidence intervals around AUC 0.75 are wide, and results should be considered preliminary hypothesis-generating findings.
Single-Center Retrospective Design: Results from one center's 1.5T scanner with specific acquisition parameters may not generalize to different field strengths, scanner manufacturers, or perfusion analysis software.
Systemic Treatment Heterogeneity: While a 10-day washout period before and after SRS was mandated, varying systemic treatments (chemotherapy, targeted therapy, immunotherapy) between patients could confound perfusion parameter changes.
Larger Prospective Studies Needed: A multicenter prospective study with standardized MRI protocols, larger cohorts, and predefined clinical decision algorithms based on early K-trans would be required to establish K-trans as a clinically actionable predictive biomarker.