Application of classification tree and neural network algorithms to the identification of serological liver marker profiles for the diagnosis of hepatocellular carcinoma
Tcw. Poon et al., Application of classification tree and neural network algorithms to the identification of serological liver marker profiles for the diagnosis of hepatocellular carcinoma, ONCOL-BASEL, 61(4), 2001, pp. 275-283
Objective: Although many attempts have been made to identify tumour-specifi
c alpha -fetoprotein (AFP) glycoforms or other serological markers for the
diagnosis of hepatocellular carcinoma (HCC), none of the available markers
has, so far, shown satisfactory sensitivity and specificity. Here we aimed
to apply classification tree and neural network algorithms to interpret the
levels of multiple serological liver markers to improve overall specificit
y and sensitivity, particularly with a view to discriminating between liver
cirrhosis with and without HCC. Methods: We developed classification trees
and neural networks that identified serological liver marker profiles comp
rising AFP, alpha1-antitrypsin (A1AT), alpha2-macroglobulin (A2MG), thyroxi
ne-binding globulin (TBG), transferrin and albumin as well as sex and age,
which might permit the diagnosis of HCC. Data were collected from 65 HCC pa
tients, 51 patients with liver cirrhosis alone (LC) and 51 normal healthy s
ubjects. Results: The generated classification trees and neural networks sh
owed similar diagnostic values in differentiating HCC from LC. The classifi
cation trees identified AFP, A1AT and albumin as the most important classif
ication parameters, whereas the neural networks identified A2MG, AFP, A1AT
and albumin as the predominant factors. The classification logic of the cla
ssification trees indicated that more HCC cases could be identified among c
ases with slightly elevated AFP levels by using the serum levels of A1AT an
d albumin. The neural networks were also useful for the identification of t
he HCC cases when the AFP levels were below 500 ng/ml (p < 0.005). The neur
al networks could identify HCC cases with AFP levels within the normal rang
e, but the classification trees could not. By combining the conventional AF
P test and the neural networks, the overall diagnostic sensitivity for HCC
was significantly increased from 60.0 to 73.8% (p < 0.05) while maintaining
a high specificity (88.2%). The sensitivities for tumors of different size
s were similar. Conclusion: The neural network algorithm appeared to be mor
e powerful than the classification tree algorithm in the identification of
the distinctive serological liver marker profiles for the diagnosis of the
HCC subgroup without significant elevation in serum AFP levels. By incorpor
ating serological levels of other liver markers and including data from a l
arge number of patients and control subjects, it should prove possible to d
evelop a versatile neural network for early diagnosis of HCC. Copyright (C)
2001 S. Karger AG, Basel.