February 1983, Volume 33, Issue 2

Original Article

Computer Aided Clinical Discrimination Between Non-Cirrhotic Portal Fibrosis and Cirrhosis.

A.Koshy  ( Department of Hepatology, PGIME R, Chandigarh, India. )
Jayant V. Deshpande  ( Department of Mathematics, Punjab University, Chandigarh, india. )
C. M. Behl  ( Department of Mathematics, Punjab University, Chandigarh, india. )


Thirty eight patients with Non-Cirrhotic Portal Fibrosis (NCPF) and 64 cirrhotics were included in the study. The diagnosis was established in all cases with liver biopsy and splenoportography. Age, sex, duration of symptoms, number ot bleeds, ascites, hepatornegaly, splenomegaly, encephalopathy and presence of abdominal veins were used clinically to distinguish betweeii the diseases. Age, number of bleeds, liver enlargement, encephalopathy and presence of abdominal veins were not significantly diff­erent between the two diseases. Computer aided discriminant analysis was used to determine the relative importance and weightage to be given in discrimination. NCPF and cirrhosis could be discriminated with 82% accuracy using splenic size and ascitic score. Use of all significant parameters resulted in only a marginal additional discrimination of 1% (JPMA: 30, 33 1983).


Non-cirrhotic portal fibrosis is at present a well established disease entity (Datta and Bhagwat, 1981). Cirrhosis was clearly defined at the fifth Pan-American Congress in 1956 and recently modified by a working group of the World Health Organisation (Anthony et al., 1977). Several studies have brought out the clinical, biochemical, histological and haemodynamic differences between NCPF and cirrhosis (Sama et al., 1971). However no study has so far been attempted to determine the discriminant value of the clinical, biochemical or haemodynamic parameters. The present study demonstrates the discriminant values of the clinical parameters.

Material and Methods

Thirty eight consecutive proven cases of NCPF, sixty four consecutive proven cases of cirrhosis seen between 1979-1981 in the department of hepatology, PGI, Chandigarh were included for analysis. The diagnosis was confirmed after splenoportography and liver biopsy in all cases. The student ‘t’ test was used to determine significant differences in age, number of bleeds and liver enlargement in centimeters below the costal margin (cms BCM). The Chi square test was used to determine significant differences in presence or absence of ascites, encephalopathy and abdominal veins. The computer programme for discriminant analysis was derived as described by Anderson (1972). The ascites was graded and arbitrarily scored as shown in table 1.

If a person had a najor upper G.I. bleed Within a month of examination and had just demonstrable but definite shifting dullness it was scored 1. If a person had no major bleed and fluid thrill on examination, it was scored 5. If the latter person had been on daily diuretics for more than a week, it was then scored 6.


There was no significant difference in the age, number of bleeds and liver enlargement between NCPF and cirrhosis as shown in table II.

Neither was there any significant difference in encephalopathy and the presence of abdominal veins as shown in table III.

The sex, duration of symptoms, splenomegaly and ascitic score were significantly different between the two groups (Table IV).

The discriminant function (X factor) for splenic size, ascites and duration of symptoms are given in Table V

size and ascitic score alone. The use of all 8 parameters, including the sex improved the accuracy of prediction by only 1%. Using the two parameters, splenic size and ascitic score the cut off was + 0.00995 and all values greater indicated NCPF and all lesser values cirrhosis.


Computer aided discrimination is being increasingly used in medicine to fmd the discriminant functions of different parameters which help to distinguish diseases. It has been used to distinguish cholesterol stones from pigment stones radiologically (S chwartz et al., 1982), Discrirnin ant analysis between NCPF and cirrhosis based on haematologic and blood chemistry data was also attempted by Matsuda et al. (1979). We have found that application of such data to clinical features reveals that the extent of splenomegaly and development of ascites are the most distinguishing features between NCFFS and cirrhosis. Apparently there is a greater reticuloendothelial hyperplasia in the spleen in NCPF. This may heve etiopthogenetic implications.
The ascitic score rather than just the presence of absence of ascites was used because of the clinical observation of transient ascites in patients of both NCPF and extrahepatic portal vein obstru­ction. The one and three month cut off points have been arbitararily selected. However the score sor used appears to have very good discriminant function.
The use cf the scoring system may help identify NCPF in areas where they are infrequently encountered by alerting the physician to the possibility. It may also help identify the patients who are most likely to Withstand the stress cf an unselective shunt, such as a conventional Lino-renal shunt.


We thank Drs. V.Nanda and A.P.S. Narang  for help rendered. We also gratefully acknowledge the secretarial assistance of Mr.Pawan Kumar Aggarwal.


1. Anderson, T.W. ‘(1972) Classification into one of several multivariate normal populations, in an introduction to multivariate statistical analysis by New Delhi. Wiley Eastern, p.lSO.
2. Anthony, P.P,, Ishak, K.G., Nayak, N.C., Poulsen, H.E. and Scheuer, P.J. (1977) The morphology of cirrhosis; definition, nomenclature, and classification. Bull. WHO., 55:521.
3. Datta, D.V. and Bhagwat, A.G. (1981) Ideopathic portal hypertension- a review. Bull. P.G.L., 15:173.
4. Matsuda, N., Inone, Y., Takahashi, H. ef aL(1979) Determination and discrimination of portal congestive splenomegalies by multivariate analysis on hematological and blood chemistry data - a possibility of the first step screening. In 1979 Annual report. Ministry of Health and Welfare Research Committee for the study of ldeopathic Portal hypertension, Japan.
5. Sama, S.K.,Bhargava, S., Gopinath, N.,, Talwar, J.R., Nayak, N.C., Tandon, B.N.and Wig, K.L.,, (1971) Noncirrhotic portal fibrosis. Am.J. Med., 51:160.
6. Schwartz, S., Kressel, HY., Solway, R.D. ef aL (1982) Prediction of gall stone composition; Validation of a discriminant analysis modeL Gastroenterology, 82: 1244.

Journal of the Pakistan Medical Association has agreed to receive and publish manuscripts in accordance with the principles of the following committees: