New Criteria Reduce Inter-Observer Variability in Chronic Cerebrospinal Venous Insufficiency: A Case Control Study

Abstract

Background: The pathophysiological mechanism defined “chronic cerebrospinal venous insufficiency”
(CCSVI) diagnosed using Zamboni criteria has raised a heated debate about possible
correlations with several neurological disorders, but also on subjectivity of these ultrasonographic
criteria used for its diagnosis. Although in 2011 new criteria have been introduced to
reduce the high inter-observer variability only two studies were conducted according to the new
investigation protocol. Therefore, we wanted to verify the impact of the revised protocol and its
ability to meet the demand for reduction of the high heterogeneity in inter-observer agreement.
Patients and Methods: Between June 2010 and June 2014, 1020 subjects (693 MS patients
and 327 HCs) were prospectively screened for CCSVI by two investigators, blinded regarding
the observed subject. After exclusion of discordant cases between the two examiners, 630 patients
with clinically defined MS and 10 patients with CIS (clinically isolated syndrome) were
matched by gender (male 38.7%, male/female 248/392) and age (mean age 44.0 years, range
18.5-77.0 years) with 315 HCs (male 43.2%, male/female 136/179-mean age 46.5 years, range
19.8-79.9 years).
Results: The prevalence of CCSVI in MS subjects before the introduction of the new ECD
criteria (94.3%) was observed to be significantly reduced (83.4%) after their introduction
(p<0.001). In MS patients, the strength of inter-observer agreement changed from moderate
(k=0.532) to good (k=0.761) before/after the revision. B-mode analysis detected only 65.7% of
valvular defects. Its diagnostic accuracy was 88.6% (95% CI: 84.0%-93.2%), with a sensitivity
of 83.5% (95% CI: 75.2%-89.9%), a specificity of 96.0% (95% CI: 88.8%-99.2%), a positive
predictive value of 96.8% (95% CI: 91.0%-99.3%) and a negative predictive value of 80.0%
(95% CI: 70.3%-87.7%).
Conclusions: The new ECD criteria introduced by the revised protocol ensure, at this time, a
substantial reduction of the inter-observer variability. Under this perspective, M-mode analysis
is essential for its ability to identify the valvular abnormalities frequently not detectable by Bmode
analysis.