CT imaging predicts decompensation, mortality in patients with cirrhosis

Liver surface nodularity measurements from routine computed tomography images in adult patients with cirrhosis accurately predicted liver decompensation and overall survival, according to data presented at the 2015 American Roentgen Ray Society Annual Meeting.

“It is commonly known that the liver surface becomes more nodular or bumpy as cirrhosis progresses, but there have been no successful methods for noninvasively measuring liver surface nodularity, until now,” Andrew D. Smith, MD, PhD, director of radiology research, University of Mississippi Medical Center, told Healio.com/Hepatology. “[My team and I] have developed methods to noninvasively measure liver surface nodularity from computed tomography images.” Smith and colleagues analyzed data retrieved from an electronic medical record search of adult patients with cirrhosis, liver CT imaging performed between January 2006 and May 2012 and available MELD scores within 6 months of imaging.

Andrew D. Smith

A baseline cohort of 858 patients with compensated and decompensated cirrhosis and a follow-up cohort of 346 patients with compensated cirrhosis were included in the final analysis. The researchers used liver surface nodularity measurements from the patients’ CT images and serum liver function tests in patients with cirrhosis to predict rate of liver decompensation and mortality, according to Smith.

Results showed patients in the baseline cohort with decompensated liver disease had a higher liver surface nodularity score compared with patients in the follow-up cohort with compensated cirrhosis (3.83 ± 1.16 vs. 2.82 ± 0.74; P < .001).

In a multivariate logistic model, the nodularity and MELD scores were independently associated with decompensated cirrhosis at baseline (P < .001 for both). In a risk model that combined nodularity and MELD scores of the baseline cohort, the median survival rate of mild risk (4.69 years) patients was greater compared to moderate (2.76 years), severe (1.41 years) and critical risk patients (0.08 years; P < .001).

Among the follow-up cohort, 137 decompensated events were observed. Nodularity score and MELD score were found to be associated with liver decompensation (HR = 1.26 per unit increase in nodularity score, P = .041; HR = 1.05 per unit increase in MELD score, P = .006) and mortality (HR = 1.3 per unit increase in nodularity score, P = .014; HR = 1.05 per unit increase in MELD score, P = .001).

Patients in the follow-up cohort had a greater median time to decompensation if they had a nodularity score less than 2.5 compared with patients with a score of at least 3.5 (6.14 years vs. 2.26 years), respectively.

“Advantages of measuring liver surface nodularity from CT images include: noninvasive quantitative technique, ability to make measurements on previously acquired CT images, applicability to any CT scanner or manufacturer, wide availability and frequent use of CT imaging in cirrhosis, rapid processing time, no requirement for intravenous contrast, no need for specialized image acquisition techniques and few measurement failures,” Smith said.

“Having the ability to predict these events will lead to changes in clinical management, and it may even be possible to evaluate treatment response over time.” – by Melinda Stevens

Reference:

Branch C. Abstract #5008. Presented at: ARRS Annual Meeting; April 19-24, 2015; Toronto.

Disclosure: Smith reports having a patent related to methods for quantifying liver surface nodularity and is the president and chief executive officer of Liver Nodularity LLC.