Abstract Archives of the RSNA, 2014
SSM18-02
Combined Use of Apparent Diffusion Coefficient and Cerebral Spinal Fluid Biomarkers Improves Sensitivity and Specificity of Diagnosing Primary Central Nervous System Lymphoma
Scientific Papers
Presented on December 3, 2014
Presented as part of SSM18: Neuroradiology (Neuro-Oncology)
Ramon Francisco Barajas MD, Abstract Co-Author: Nothing to Disclose
James Rubenstein, Abstract Co-Author: Nothing to Disclose
Marc Christopher Mabray MD, Presenter: Nothing to Disclose
Javier Villanueva Meyer MD, Abstract Co-Author: Nothing to Disclose
Soonmee Cha MD, Abstract Co-Author: Nothing to Disclose
The diagnosis of primary central nervous system lymphoma (PCNSL) using morphologic MR imaging can be challenging given its lack of sensitivity and specificity. Therefore, we aimed to determine if the combined use of MRI derived apparent diffusion coefficient (ADC) and cerebral spinal fluid (CSF) biomarkers (chemokine ligand (CXCL) 13 and interleukin (IL) 10) improves the diagnostic capabilities of PCNSL in a cohort of patients with enhancing brain lesions.
89 patients with newly diagnosed enhancing brain lesions (45 PCNSL, 21 metastatic tumor, 14 Glioblastoma, and 9 acute demyelinating disease) detected by MR imaging underwent CSF sampling in this HIPPA compliant CHR approved study. CSF samples provided standardized CXCL-13 and IL-10 concentrations (pg/ml). Co-registered ADC maps (GE Functool v4.4) allowed for placement of regions of interest about the enhancing lesions. Lesion minimum, 25th percentile, and mean relative ADC (rADC) values were calculated as a ratio of white matter values. Student’s T-test compared differences between groups. Linear regression was performed with Pearson correlation. Receiver operating characteristic analysis provided threshold values. P-value less than 0.05 equated statistical significance.
rADC values were significantly decreased and CSF biomarkers increased within PCNSL (rADCmin= 0.72, rADC25th= 0.91, rADCmean= 0.98, CXCL-13= 3382, IL-10= 1773) when compared to other enhancing brain lesions (rADCmin= 1.15, rADC25th= 1.39, rADCmean= 1.36, CXCL-13= 113, IL-10= 5.93; P <0.01). rADC minimum, 25th percentile, and mean values demonstrated a significant correlation with CXCL-13 alone (R> 0.32, P< 0.05). CXCL-13 value >340 provided a sensitivity and specificity of 89% and 75% for the diagnosis of PCNSL. Mean rADC threshold value of 1.2 provided the best diagnostic capability (sensitivity 95%, specificity 64%). A multi-parametric diagnostic model using CXCL-13 and rADC provided a sensitivity and specificity of 100% and 94%.
The combined use of MRI derived ADC and CSF derived CXCL-13 biomarkers improved the diagnostic capability of PCNSL within this cohort of patients.
The MRI and CSF derived mulitparametric diagnostic model utilized in this study could negate the need for tissue sampling prior to the implementation of medical therapy in patients with PCNSL.
Barajas, R,
Rubenstein, J,
Mabray, M,
Villanueva Meyer, J,
Cha, S,
Combined Use of Apparent Diffusion Coefficient and Cerebral Spinal Fluid Biomarkers Improves Sensitivity and Specificity of Diagnosing Primary Central Nervous System Lymphoma. Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL.
http://archive.rsna.org/2014/14008997.html