262

Predictive Values of Various Liver Function Tests with Respect to the Diagnosis of Liver Disease

MICHAEL SHEEHAN and PAT HAYTHORN

Good Samaritan Hospital & Medical Center, Portland, Oregon 97210

A prospective study of 181 patients suspected of hav- ing liver disease was carried out to determine the rela- tive effieiencies of serum bilirubin (total and direct), alkaline phosphatase (AP), gamma glutamyl transferase (GGT), alanine aminotransferase (ALT), and aspartate aminotransferase (AST) with respect to diagnosis. Liver biopsies, liver scans, abdominal ultrasound, and clinical parameters were also tabulated and used independently to evaluate the patient 's hepatic status and to determine the final diagnoses in each case. From the results of these tests for the 60 patients who were diagnosed as having liver disease, and the 87 patients who were felt to be free of liver disease, predictive values of the above tests were established. Data from this study suggests that while direct bilirubin is the most specific test, GGT is the most sensitive and has the fewest false negatives in the diagnosis of liver disease.

tests are given below, together with the limits (in paren- theses) used to classify these tests as abnormal in the present study. The somewhat higher limits in parentheses are a composite of the upper limit of normal for each test plus twice the respective analytic error, and were used to avoid classifying borderline values as positive: AP, men 78 (85) IU/1, women 61 (68) IU/ l ; ALT 17 (21) I U / I ; AST, men 30 (33) IU/1, women 21 (24) IU/1; GGT, men 38 (43), IU/1, women 36 (41) IU/1; total bilirubin, men 1.5 (1.8) mg/dl, women 1.2 (1.5) mg/dl; and direct bilirubin, 0.2 (0.4) mg/dl. In addition to the results for these tests, demographic data (age, sex) as well as results (when available) for HBs Ag testing, liver biopsies, liver scans, and abdominal ultrasound studies were also tabulated. Final diagnoses were determined in- dependently of our study and were abstracted from the patients ' charts after discharge.

A l though n u m e r o u s repor t s have appeared in the l i t e r a t u r e '~ c o m p a r i n g the sens i t iv i t i e s and d iagnos- t ic s i g n i f i c a n c e of va r ious l iver f u n c t i o n tests , very l i t t le has been published r ega rd ing the relat ive pre- dict ive va lues of these tes ts . Schmid t 1 and Keane 2 compared the sens i t iv i t i e s of some of the ro u t i ne as well as esoter ic tests , bu t i n s u f f i c i e n t da ta was p resen ted to al low ca l cu la t ion of the p red ic t ive va lues of these tes ts . Solberg et a l2 and also, Winkel et al2 have commented on the d iagnos t i c va lue of rou t ine l iver func t ion tests us ing d i s c r iminan t analysis , bu t again, predict ive values for these tests could not be de termined f rom the i r data.

We were i n t e r e s t ed in compa r ing the re la t ive pred ic t ive va lues and sens i t iv i t i e s of six se rum l iver f u n c t i o n t es t s : to ta l and d i rec t b i l i r ub in , GGT, ALT, AST, and AP. Speci f ica l ly , we were i n t e r e s t ed in knowing what the probabi l i ty was of f i n d i n g a p a t i e n t wi th l iver d isease wi th a no rma l to ta l bu t elevated direct b i l u rub in level, and if such a pa t i en t existed, what the values of the other l iver func- t ion tes ts would be. To a n s w e r these ques t ions , a p rospec t ive s tudy on p a t i e n t s wi th suspected l iver disease, c o m p a r i n g these tests , was ca r r i ed out.

MATERIALS AND METHODS

A total of 181 patients admitted as in-patients to Good Samari tan Hospital were studied. They were included if, after admitting lab work had been ordered, an additional request for one or more of the following serum tests was received in the laboratory: GGT, ALT, total or direct bilirubin. For each such request, a battery of all of the routine liver function tests (AP, ALT, AST, GGT, total and direct bilirubin) was then performed. Analyses for the enzymes were carried out on a GEMSAEC centri- fugal analyzer at 30 ° according to the following method- ologies: AP, method of Wilkinson et al.7; A LT, method of Henry et al2; AST, method of Karmeng; and GGT, method of Szasz ~°. Total and direct bilirubin assays were performed manually using the Jendrassik-Grof proce- dure ~1. Upper reference limits for our hospital for these

RESULTS

F o r the 181 pa t i en t s inves t iga ted , 115 had an eleva- t ion of a t leas t one of the six rou t ine l iver f u n c t i o n tests . Of these 115 cases, 60 had f ina l d iagnoses of l iver d iseases - - 16 cases of obs t r uc t i ve l iver dis- ease, 11 of a lcohol ic l iver disease, 7 wi th some form of m a l i g n a n c y i nvo l v i ng the liver, 15 wi th hepat i t i s , 2 wi th c i r rhos is , and 9 which were c o m b i n a t i o n s of two or more of the above c lass i f i ca t ions . Fo r the p a t i e n t s wi th l iver disease, 41 ou t of 60 had e levated AP levels, 54 out of 58 had e levated GGT levels, 39 out of 57 had e levated ALT levels, 38 ou t 51 had e levated AST levels, 27 out of 58 had e levated tota l b i l i r u b i n levels, and 28 out of 56 had e levated d i r ec t b i l i r u b i n levels. Of the p a t i e n t s for whom l iver dis- ease had been ru led out, 12 of 87 had e levated AP level, 14 out of 87 had e levated GGT levels, 11 ou t of 86 had e levated ALT levels, 6 out of 81 had ele- va ted AST levels, 2 ou t of 87 had e levated to ta l b i l i r u b i n levels, and 1 ou t of 84 had an e levated d i rec t b i l i r u b i n level. F r o m these data, posi t ive and nega t ive p red ic t ive va lues as well as sens i t iv - i t ies of the tes ts r e g a r d i n g the d i agnos i s of l iver d isease were de te rmined , as shown in Tab le 1. A g raph ica l compar i son of the re la t ive s ens i t i v i t i e s of to ta l ve r sus d i rec t b i l i r u b i n is seen in Fig. 1.

TABLE 1

PREDICTIVE VALUES OF VARIOUS LIVER FUNCTION TESTS WITH RESPECT TO DIAGNOSIS OF LIVER DISEASE

Predictive Predictive Test Value (-t-) Value ( - ) Sensitivity

GGT . . . . . . . . . . . . . . 79 95 93 AST . . . . . . . . . . . . . . 86 85 74 ALT . . . . . . . . . . . . . . 78 81 68 AP . . . . . . . . . . . . . . . 77 80 68 Direct bilirubin . . . . . 97 75 50 Total bilirubin . . . . . . 93 73 47

100

10.0 ¢ -

t . . . i

~o 1.0 • - . _ l - ~

: . ':1 :

o •

. e

1.o lOP D i rec t Bil irubin

mg/dJ

100

Figure 1 - - Diwtriate plot o/ direct bilir~tbin versus total bili.rubin for 58 patie~tts with liver disease. Each point represe~ts o~e patie~t. Dotted lilies indicate ~tppcr limits of normol.

DISCUSSION

In answer to our f irst question, Fig. 1 shows that there were 4 patients with liver disease who had normal total with elevated direct bilirubin concentra- tions. Of the population of patients with liver disease, this represents 6.7% ; and for the entire study popula- tion, 2.2%. For these patients, 3 out of 4 had slight elevations of AP, whereas all 4 had marked elevations of GGT, ALT, and AST. On the other hand, for pa- tients with elevated levels of GGT, AP or ALT, 19, 15, and 13 respectively had normal direct bilirubin levels. Thus, in answer to our second question, an isolated increased direct bilirubin level would be improbable in cases of liver disease, at least in the initial phases of the diseases. In our experience, we have seen this only in cases of prolonged stasis of blood in the liver and in the recovery stages of various liver diseases. Even in biopsy or surgically proven liver disease, the direct bilirubin assay was relatively insensitive when compared to GGT. Despite these drawbacks, it has been our experience that this test is still used primarily as a screen rather than a con- f irming test or for following therapy.

263

As shown in Table 1, GGT proved to be the most sensitive test. Somewhat unexpected, however, was the finding that the AST assay was more sensitive than the ALT (others 1'2 have shown the reverse to be true). Not unexpectedly, total and direct bilirubin assays were found to be the least sensitive of all of the routine liver function tests. Although there were some differences, the ranking shown in Table 1 were mostly independent of the type of liver disease. For patients with hepatitis, the AP assay replaced the total bilirubin assay as the least sensitive test. Other- wise, the rankings were the same.

Several workers ~2"" have commented on the diagnos- tic utility of the AST/ALT ratio. De Ritis et al. ~ used this ratio to discriminate between viral hepatitis and obstructive liver disease (the former having a ratio around 0.5 and the latter around 1.0). In our study, this ratio was found to be both insehsitive and non-specific, in that 25 per cent of the patients with obstructive liver disease had AST/ALT ratios of less than 0.6, while for approximately 30 per cent of the cases of viral hepatitis this ratio was near unity. In our experience, this ratio has not proven itself to be a very reliable diagnostic aid.

REFERENCES

1. Schmidt, E., in: Evaluation of Liver Function: A Multifaceted Approach to Clinical Diagnosis, L.M. Demers and L.M. Shaw, ed., Urban and Schwarben- berg, Baltimore, 1978, p. 79.

2. Keane, P.M., Garcia, L., Gupta, R.N., Walker, W.H.C., Clin Biochem., 6, 41, 1973.

3. Solberg, H.E., Skrede, S., Blomhoff, J.P., Sca~l. J. Clbu Lab. Invest., 35, 713, 1975.

4. Winkel, P., Ramsoe, K., Lyngbye, J., Tygstrup, N., Clbz. Chem. 21, 71, 1975.

5. Skrede, S., Blomhoff, JP, Elgio, K, Gjone, E, Scan. J. Gastroent., 8, Suppl. 19, 37, 1973.

6. Rosalki, S.B., Rau, D., Clin. Chim. Acts , 39, 41, 1972. 7. Wilkinson, J.H., Boutwell, J.H., Winsten, S., Clin.

Chem. 15, 487, 1969. 8. Henry, R.H., Chiamori, N., Golup, O.H., Berkman, S.,

Am. J. Clin. Path., 34, 381, 1960. 9. Karmen, A., Wroblewski, F., Ladue, J.E., J. Clin.

10. 11.

12.

13.

14.

i~vest., 34, 125, 1955. Szasz, G., Cli~z. Chem., 15, Jendrassik, L., Grof, P., 1938. De Ritis, F., Coltorti, M., 1972.

124, 1969. Biochem. Ztsch., 297, 81,

Giusti, G., Lancet, 1, 685,

WaUerstedt, S., Olsson, R., Waldenstrom, J., Acta Med. Scand., 195, 227, 1974. Goldberg, D.M., Ellis, G., Ward, A.M., Cli~. Chim. Acts , 72, 379, 1976.