REGULAR CONTENT
Final ID
549
Type
Original Scientific Research-Oral or Pos
Authors
M Yuhasz1, E Morris1, J Merola2, N Chaudhary1, S Sigal3, J Gross1, E Aaltonen1
Institutions
1New York University School of Medicine, New York, NY, 2Yale School of Medicine, New Haven, CT, 3Albert Einstein School of Medicine, Bronx, NY
Purpose
To compare MELD and MELD-Na scoring systems and determine qualitative risk thresholds for predicting early mortality after transjugular intrahepatic portosystemic shunt (TIPS) creation.
Materials & Methods
Model for End-Stage Liver Disease (MELD) and MELD-Na scores were calculated for 141 patients (70.8% male, mean age 57.5 years) prior to TIPS creation at a single institution between 2011-2016 and retrospective medical record review identified 30-, 90-, and 180-day post-procedure mortality. Analysis of MELD/MELD-Na scores and 30-, 90-, and 180-day mortality was performed with logistic regression, receiver operating characteristic (ROC) curve C-statistics, and cut point calculations.
Results
TIPS were created in 141 patients (13.5% acute variceal bleed, 24.8% history of variceal bleed, 53.2% refractory ascites, 8.5% portal vein thrombosis) with subsequent 30-, 90-, and 180- day mortality rates of 7.8%, 14.2%, and 18.0% respectively. Higher MELD and MELD-Na scores were associated with acute variceal bleed (p<0.001) and increased mortality (p<0.001 for all time points). Logistic regression demonstrated a significant relationship between MELD/MELD-Na and mortality (p<0.001 for all) with score increases of one point corresponding to increased odds of death ratios ranging from 1.19-1.29 for MELD and 1.19-1.26 for MELD-Na. ROC analysis demonstrated no significant difference between MELD (0.808) and MELD-Na (0.784) C-statistics (p=0.220-0.773). The optimal threshold for MELD of 19.9 (sensitivity 73%, specificity 91%) corresponds with 39% (above threshold) vs 3% (below threshold) 30-day mortality. The optimal threshold for MELD-Na of 22.9 (sensitivity 82%, specificity 85%) corresponds with 30% vs 3% 30-day mortality.
Conclusions
MELD-Na is equally effective as MELD in predicting early post-TIPS mortality. While additional validation is required to corroborate these conclusions, MELD-Na scores may be utilized for future TIPS prognostication. We suggest a modified MELD threshold of 20 and establishment of a MELD-Na threshold of 23 for elective procedures, scores below which interventional radiologists can confidently deliver care for patients in need with low associated post-procedure mortality rates.
Final ID
549
Type
Original Scientific Research-Oral or Pos
Authors
M Yuhasz1, E Morris1, J Merola2, N Chaudhary1, S Sigal3, J Gross1, E Aaltonen1
Institutions
1New York University School of Medicine, New York, NY, 2Yale School of Medicine, New Haven, CT, 3Albert Einstein School of Medicine, Bronx, NY
Purpose
To compare MELD and MELD-Na scoring systems and determine qualitative risk thresholds for predicting early mortality after transjugular intrahepatic portosystemic shunt (TIPS) creation.
Materials & Methods
Model for End-Stage Liver Disease (MELD) and MELD-Na scores were calculated for 141 patients (70.8% male, mean age 57.5 years) prior to TIPS creation at a single institution between 2011-2016 and retrospective medical record review identified 30-, 90-, and 180-day post-procedure mortality. Analysis of MELD/MELD-Na scores and 30-, 90-, and 180-day mortality was performed with logistic regression, receiver operating characteristic (ROC) curve C-statistics, and cut point calculations.
Results
TIPS were created in 141 patients (13.5% acute variceal bleed, 24.8% history of variceal bleed, 53.2% refractory ascites, 8.5% portal vein thrombosis) with subsequent 30-, 90-, and 180- day mortality rates of 7.8%, 14.2%, and 18.0% respectively. Higher MELD and MELD-Na scores were associated with acute variceal bleed (p<0.001) and increased mortality (p<0.001 for all time points). Logistic regression demonstrated a significant relationship between MELD/MELD-Na and mortality (p<0.001 for all) with score increases of one point corresponding to increased odds of death ratios ranging from 1.19-1.29 for MELD and 1.19-1.26 for MELD-Na. ROC analysis demonstrated no significant difference between MELD (0.808) and MELD-Na (0.784) C-statistics (p=0.220-0.773). The optimal threshold for MELD of 19.9 (sensitivity 73%, specificity 91%) corresponds with 39% (above threshold) vs 3% (below threshold) 30-day mortality. The optimal threshold for MELD-Na of 22.9 (sensitivity 82%, specificity 85%) corresponds with 30% vs 3% 30-day mortality.
Conclusions
MELD-Na is equally effective as MELD in predicting early post-TIPS mortality. While additional validation is required to corroborate these conclusions, MELD-Na scores may be utilized for future TIPS prognostication. We suggest a modified MELD threshold of 20 and establishment of a MELD-Na threshold of 23 for elective procedures, scores below which interventional radiologists can confidently deliver care for patients in need with low associated post-procedure mortality rates.