Elmo's Heidi again? :-)

Alan

I've read a report about a Spanish team doing computer evaluation of CT
scans that had good accuracy (but it's very new, it was presented today
at the EASL 2005 conference that's ongoing now in Paris):

FIBROSIS ASSESMENT IN HEPATITIS C BY COMPUTER-PERFORMED OPTICAL ANALYSIS
OF CT OF THE LIVER: A NOVEL NON-INVASIVE AND USEFUL METHOD
[...] Results: OS-WMF correlates with fibrosis (r=0.67;p<0.001),
increasing with F-stage: F0=0.77±1.08; F1=1.19±1.14; F2=1.64±1.99;
F3=2.88±0.61; F4=3.34±0.36 (ANOVA, p<0.0001). Mean fibrosis of the
cohort was 1.7±1.3 by histopathological analyses and 1.8±1.3 by OS-WMF,
p=ns. Since OS-WMF was 1.27+1.13 in F0-F2 and 3.09+0.55 in F3-F4,
p<0.0001, it discriminates patients with F3-F4: if OS-WMF<2, advanced
fibrosis can be excluded (NPV=97.9%) and if OS-WMF>3 the possibility of
showing F3-F4 is high (PPV=90%). However, one third of patients with
fibrosis F0-F1 showed OS-WMF>2.  Conclusions: Computer-performed optical
analysis of images from conventional CT of the liver is a reliable
method, able to measure hepatic fibrosis in patients with hepatitis C.
Weighted mean value and distribution of fibrosis in the liver can be
quantified. Major discrepancies with histopathological results are found
in patients with higher fibrosis stage. Whether this is a bias of
proposed method or an error due to sampling limitations of liver biopsy
requires further analysis
http://www.kenes.com/easl2005sci/program/CopyFile.asp?FileName=599.doc

From the same conference:

ARE CURRENT FIBROSIS SERUM MARKERS REALLY AN ALTERNATIVE TO LIVER BIOPSY
IN CHRONIC HEPATITIS C?

We recently described a score of fibrosis combining PIIINP and MMP-1,
which are involved in fibrogenesis and fibrolysis respectively. The aim
of this study was to evaluate its diagnostic accuracy compared to that
of a panel of other markers including prothrombin time (PT), hyaluronate
(HA), Fibrotest (FT), APRI and Forns' score in chronic hepatitis C (CHC)
patients Methods: One hundred and eighty CHC patients seen in our centre
for a pre-therapeutic liver biopsy were prospectively included. Liver
fibrosis was assessed using the METAVIR index. Sinusoidal fibrosis,
steatosis and iron load were quantified. Biochemical measurements were
performed on serums collected the day of the biopsy. Results: Median
length of liver biopsies was 22 mm. Patients were classified as F0:n=15,
F1:n=74, F2:n=40, F3:n=26, F4:n=24. Overall diagnostic accuracy of the 4
combined scores was better than that observed with PT and HA alone.
AUROC ranged from 0.79 to 0.84 for discriminating F0/F1 vs F2/F3/F4 and
from 0.81 to 0.88 for discriminating F0/F1/F2 vs F3/F4 (NS). Accuracy
was not altered by the length of biopsies or by the presence of
sinusoidal fibrosis, steatosis or iron in liver. Applying the lower
cut-off described for each score, METAVIR fibrosis greater than F1 could
be ruled out in about one third of patients with NPV ranging from 82 to
84%. The higher cut-off selected about 20% of patients whom extensive
fibrosis (F3F4) was confirmed with a PPV ranging from 89 to 93%.
Simultaneous use of at least 2 scores improved NPV up to 100% for the
diagnosis of F0F1, but decreased the number of selected patients.
Commercial application of FT proposes an estimated fibrosis METAVIR.
Discordances of at least 1 point were observed in 52% and of 2 points in
22% of patients. The only parameter associated with discordance was FT
between 0.20 and 0.80. Conclusion: Our results confirm that PIIINP/MMP-1
provides relevant information on liver fibrosis with a good accuracy,
comparable to that observed with FT, APRI and Forns' score. However none
of these scores can reliably give and estimated METAVIR index in each
patient. Combination of several scores could improve their accuracy.
http://www.kenes.com/easl2005sci/program/CopyFile.asp?FileName=577.doc

Thomas