MECHANISMAND SIGNIFICANCE OF THE THYMOLTURBIDITYTEST FORLIVER DISEASE

By HENRYG. KUNKELAND CHARLESL. HOAGLAND'(From the Hospital of the Rockefeller Institute for Medical Research, New York City)

(Received for publication April 7, 1947)

An increase in the amounts of the globulincomponents of the serum has long been recog-nized in advanced states of liver disease (1, 2, 3),although the significance of the alteration hasnever been defined. It has been the basis of theTakata-Ara, Weltmann and the formol-gel reac-tions which have been used for many years inthe diagnosis of liver disease. The nonspecificnature of these tests is now clear following thedemonstration that positive reactions are foundin any disease showing marked hyperglobulinemia(4, 5, 6). More recently, several new serum re-actions, which appear to depend on small changesin the proteins of the serum, have been used fordemonstrating liver disease. These include thecephalin flocculation, the colloidal gold and thethymol turbidity tests. The cephalin flocculationreaction has been studied in detail by Hanger andhis coworkers (7, 8) and has been found to be asensitive index of liver damage. The test may bepositive in patients with liver disease who shownormal serum protein values according to theusual methods of protein estimation and, con-versely, the serum of patients with marked hyper-globulinemia may show negative cephalin floccu-lation reactions. The exact serum protein con-stituent that is altered in liver disease and isresponsible for a positive cephalin flocculationtest has not been clearly established; recent workby Hanger (8) appears to implicate albumin inaddition to gamma globulin. The final solutionto the problem is hampered by the complexity ofthe cephalin flocculation reaction, a fact which isalso true of the colloidal gold reaction.

The technique of the thymol turbidity test ismuch simpler, however, and probably consists ofa direct precipitation of a protein appearing inliver disease by the addition of a thymol solution.It would seem, therefore, that a study of themechanism of this reaction and the protein com-

Deceased, August 2, 1946.

ponent concerned would be more likely to yieldclear-cut information regarding at least one ofthe proteins that appear in the blood stream dur-ing diseases of the liver. The present study wasan attempt to elucidate the mechanism of thereaction and to define its significance in terms ofclinical observations.

MATERIALS AND METHODS

The sera used in the study of the thymol turbidityreaction were selected from a group of 200 patients withinfectious hepatitis and 65 patients with other liver dis-orders admitted to the Out Patient Department of theHospital of the Rockefeller Institute.

Technique- of performing the thymol turbidity test.The thymol reagent was prepared as described by Mac-lagan (9). Slight variations of this method producedunsatisfactory results. In alkaline solution thymol issomewhat unstable and turbidity of the reagent oftenoccurs on standing. Exposure to air increases the tur-bidity of the solution and it was found important to keepthe thymol reagent tightly stoppered. As the solutionbecomes increasingly cloudy, its activity decreases and itis important that only clear or very slightly turbid solu-tions be used. If properly prepared, the thymol reagentis satisfactory for at least one month. Although the pHof the thymol reagent was slightly lower than that origi-nally stated by Maclagan, it proved to be satisfactory.

Three ml. of the thymol reagent were added to 0.05ml. of serum and the degree of turbidity measured inthe Coleman Jr. spectrophotometer at 650 mnu. This rep-resented a 1/60 dilution and corresponded to that orig-inally described by Maclagan. Figure 1 shows the tur-

bidity of sera from cases of liver disease at variousdilutions of reagent using saline dilutions as controls.It can be seen that the maximum turbidity was usuallyobtained at a 1/12 dilution and that differences in certainsera at the lower dilution might not be so marked at

the 1/60 dilution. Although the use of lower dilutionshas certain apparent advantages, all determinations dis-cussed in this paper were performed at Maclagan's stand-ard 1/60 dilution.

The degree of turbidity was compared with a BaSO4standard as described in a previous publication (10).This proved to be a satisfactory standard regardless oftube size or type of instrument used. The units of tur-

bidity corresponded to those originally' described byMaclagan using visual comparators.

1060

THYMOLTURBIDITY TEST IN LIVER DISEASE

30 I l Jdll l eal e

' o60 2 o4 0607 09 o

50a10,in

60

PN 40-

0 020 30 40 50 60 70 80 90 100BFamt thymol Peagent added

to one part zerum

FIG. 1. TURBIDITY READINGS AS MEASUREDIN THESPECTROPHOTOMETERAT VARIOUS DILUTIONS OF SERUMWITH THYMOLREAGENT

Estimation of the degree of flocculation of the thymolserum mixture was also carried out in a number ofpatients. The finding of Neefe (11) that in certainpatients, following an attack of infectious hepatitis, the24-hour flocculation with thymol reagent remained posi-.tive for a slightly longer period than the usual thymolturbidity reaction, was confirmed. In general, however,the estimation of flocculation proved to be a less sensitiveindex than turbidity determinations, and did not furnishquantitative results.

Electrophoretic analyses were made in diethylbarbituricacid buffer (.u = 0.1, pH = 8.6) by the method of Longs-worth (12). Determinations of lipids were carried outby the gasometric lipid carbon method of Van Slyke andFolch (13). Total and free cholesterol of the plasmawas determined by the method of Schoenheimer andSperry (14). Extraction of lipids from serum was doneby freezing in the presence of ether as described by Mc-Farlane (15). A modification of Hanger's method (7)was used for the determination of the cephalin floccula-tion reaction. Bromsulfalein retention was estimated bythe method of Rosenthal and White (16) modified forthe use of the Coleman Jr. spectrophotometer. Globulinwas determined electrophoretically and by fractionationwith NaSO, (17). In addition, quantitative measure-ments of globulin were obtained by a turbidometric tech-nique (18).

Immunological experiments were carried out by inject-ing rabbits with 5 to 8 mgm. of thymol protein every 2days for 8 injections. The antiserum was absorbed withnormal human serum. Precipitin tests were carried outby the technique of Swift, Wilson, and Lancefield (19).

EXPERIMENTAL

I. The relation of the thymol turbidity reaction tothe lipids in the serumIt was demonstrated by Maclagan (9) that the

precipitate resulting from the addition of thymolreagent to serum is high in cholesterol and phos-pholipids. Recant, Chargaff and Hanger (20)found that sera from cases of liver disease fromwhich lipids had been extracted with ether nolonger showed turbidity following the addition ofthymol reagent. These observations indicated thatlipids are an important factor in the thymol tur-bidity reaction. This work was confirmed andextended. Thymol was found to have a specialeffect on lipids in general. Any lipid emulsiontended to be broken up by the addition of athymol solution. Figure 2 shows the effect ofthymol in increasing the particle size of a lipidemulsion as viewed under the microscope. Whensuch emulsions were visualized with the nakedeye, an increase in turbidity accompanied thechange in particle size. This turbidity was purelythe result of the physical alteration in the lipidemulsion resulting from the addition of thymol.Lipemic sera from patients with liver disease,nephrosis, diabetes, and thyroid disease all showedan increase in turbidity upon the addition of thethymol reagent. This, however, was purely aphysical change in the lipid emulsion, since noprotein was precipitated as in the usual thymolturbidity reaction accompanying liver disease.The following example serves to illustrate thispoint.

Three-tenths ml. of lipemic serum from a patient withnephrosis was diluted 60 times with the thymol barbitalbuffer reagent. This gave an increased turbidity over

0 0o 00 0

000 00000 o

00 0

aL bFIG. 2. COMPARISONOF THE PARTICLE SIZE OF EQUAL

QUANTITIES OF A LIPID SUSPENSION IN (a) BARBITALBUFFER, (b) BARBITAL BUFFER PLUS THYMOL, ASVIEWED UNDERTHE MICROSCOPE

1061

0 0

0( 000-

HENRYG. KUNKEL AND CHARLESL. HOAGLAND

the serum with barbital buffer alone equivalent to 60units. The turbid mixture was then spun for 3 hoursin the centrifuge at 15,000 r.p.m. No sediment settled atthe bottom of the tube, but a white layer formed at thesurface leaving a clear solution underneath. Analysis ofthe surface layer showed that it contained 6 mgm. totallipid and 0.001 mgm. nitrogen. The material responsiblefor the turbidity had been brought to the surface of thesolution by high speed centrifugation and was found tocontain a negligible amount of protein. A similar ex-

periment carried out on the turbid material resulting fromthe addition of thymol reagent to clear serum from a case

of infectious hepatitis demonstrated that the cloudy mate-rial all settled to the bottom of the tube. Nitrogenanalysis of the precipitate showed that it contained 0.44mgm. of N. Whereas the turbidity of this serum hadbeen found to be equivalent to only 30 units, the materialresponsible for the turbidity consisted of a large amountof protein.

It appears, therefore, that the turbidity pro-

duced by the action of the thymol reagent on

lipemic sera from subjects without liver diseaseis due to an increase in the particle size of theprotein-free lipid suspension, while the turbidityproduced by the reagent in clear hepatitis serum

is due to the formation of a protein-lipid-thymolcomplex.

Since certain sera from cases of liver diseaseare lipemic, it was of some importance to find a

simple method of determining how much of theturbidity produced by the thymol reagent indi-cated a true reaction with precipitation of protein.When a thymol solution having the same pH as

the usual reagent, but with a high ionic strength,was added to clear serum from a patient withliver disease giving a positive reaction, no turbid-ity appeared. However, when this solution was

added to an artificial lipid emulsion or lipemicsera from patients with nephrosis, the turbidityproduced was the same as that caused by the lowionic strength reagent. The thymol altered thestate of the lipids regardless of the ionic strengthof the solution. The usual protein precipitationreaction occurred only with the low ionic strengththymol reagent. In evaluating the turbidity pro-duced in lipemic serum by Maclagan's thymolreagent, the amount of turbidity due to proteinprecipitation alone was obtained by using as blankin the photometer serum with the high ionicstrength thymol buffer (Table I).

The presence of lipids in the serum is also an

essential factor in the usual thymol turbidity reac-

tion, as indicated by the fact that positively react-ing sera after extraction with ether no longergive the reaction. The essential role of the lipidswas further borne out by the finding that surface-acting agents, such as various tweens, completelyinhibited the formation of any precipitate in hepa-titis serum to which thymol had been added.Once formed, the thymol precipitate also dis-solved readily on the addition of small amountsof tween 80.2 In other protein precipitation re-actions which depended purely on the naturalsolubility of the proteins involved, the presenceof tween actually enhanced the precipitation. Theaction of the tween was undoubtedly related toits effect on the state of the lipids involved inthe reaction.

Since the soluble lipids play an essential partin the thymol turbidity reaction, it seemed impor-tant to test the effect of the addition of variousconcentrations of lipid on the reaction. Becauseof the specific effect of thymol on lipid suspen-sions it was important to keep the lipids in theirmost soluble state. As a result, lipid was addedin the form of clear serum giving a negativethymol turbidity reaction. Four such sera werechosen containing varying amounts of lipid.When each of these sera was added to normalserum, the thymol turbidity reaction of the com-bination remained negative. However, whenadded to serum with a high gammaglobulin level

TABLE I

Comparison of the turbidity in units obtained in thespectrophotometer for various sera upon the addition of thethymol reagent, using as the zero control the same sera with(a) buffer alone, (b) high ionic strength buffer with thymol

Bfeat Buffer +Type of serum pH 7.6 and tH7y6 cad

Clear serum from patients with 26 26infectious hepatitis

Lipemic serum from patient with 52 0nephrosis

Lipemic serum from patient with 28 16infectious hepatitis

giving a positive thymol reaction, the turbidityof the combination was proportional to the amountof lipid in the added serum. This effect was morestrikingly brought out by first extracting the

2 Polyoxyethylene- sorbitan monooleate.

1062

THYMOLTURBIDITY TEST IN LIVER DISEASE

lipids from the serum of another patient with ahigh gamma globulin level in the serum. Thethymol turbidity reaction was reduced from 22to 9 units by this procedure. Table II shows the

TABLE II

Relative effect of ihe addition of 0.2 ml. of sera, differingonly in their lipid content, in restoring a positive thymolturbidity reaction to 0.2 ml. ether-extracted hepatitis serum

ThymolLipid Thymol turbidity of

content turbidity combination Protein in Lipid inof serum of serum of extracted precipitate precipitate

added added hepatitis serumand added serum

mgm. units units mgm. mgm.per cent

500 3 14 1.84 1.11900 4 27 2.58 2.46

2000 4 50 3.41 5.22

relative effectiveness of various sera in restoringa positive thymol turbidity reaction to this ex-

tracted hepatitis serum. It can be seen that thehigh lipid sera had a much greater effect thannormal serum. The major portion of the in-

400

C4

U

2200-V~~~~~~~R

7A

Concentration of lipid in -seum added (m9. 70)

FIG. 3. EFFECT OF THE ADDITION OF SERA CONTAIN-ING VARIABLE AMOUNTSOF LIPID ON THE TURBIDITY,LIPID CONTENT, AND PROTEIN CONTENTOF THE PRE-CIPITATE FORMEDIN EXTRACTEDHEPATITIS SERUMWITH

THYMOLREAGENT

creased turbidity was due to increased precipita-tion of lipid as shown by the protein and lipidanalyses of the precipitates. Figure 3 demon-strates more clearly the comparative effect ofvarious lipid concentrations on the turbidity, pro-tein content, and lipid content of the precipitateformed with extracted hepatitis serum. The re-sults were obtained in the same experiment illus-trated in Table II. It is evident that, althoughthere is increased precipitation of protein in thepresence of higher lipid concentrations, the majorportion of the increased turbidity is due to in-creased precipitation of lipid. In other words, theresulting turbidity reflected primarily the con-centration of lipid in the added serum. The re-lationship was so close that it was possible to usethis system as a rapid method of estimating theconcentration of lipid in an unknown serum.

The above data demonstrated the marked influ-ence of lipid concentration on the thymol turbidityreaction in a somewhat artificial system involvingthe addition of sera with variable lipid concentra-tions. In order to evaluate more clearly the effectof the concentration of lipids in sera on the reac-tion as it is usually applied, lipid analyses werecarried out simultaneously on the serum of pa-tients with liver disease and the specific precipitateresulting from the addition of thymol reagent.The amount of lipid in the precipitate varied from20 to 50 per cent and was directly proportional tothe concentration of lipid in the original serum(Figure 4). The turbidity that is usually meas-ured in the thymol turbidity reaction depends onboth the protein and lipid that are precipitated.Since the concentration of lipid in the precipitate isproportional to the concentration in the serum, itis clear that the thymol turbidity reaction actuallydetermines in part the concentration of lipid inthe serum of patients with liver disease.

Despite the fact that the level of the lipids inthe serum is one of the variables that is measuredin the thymol turbidity reaction, a number of pa-tients with liver disease other than infectious hepa-titis have been observed with high lipid levels inthe serum but with low or negative thymol tur-bidity test values. Fractionation of the total lipidin these cases into the cholesterol and phospho-lipid partitions did not reveal any specific effect.The addition of reactive protein, or serum contain-

1063

HENRYG. KUNKEL AND CHARLESL. HOAGLAND

40

IL

00

20 00 1000 1500 2000

Concentration of lipid in Serum (m9.76)

FIG. 4. THE RELATION OF THE LIPID CONTENT OF

SERUMFROM PATIENTS WITH LIVER DISEASE TO THE

LIPID CONTENT OF THE SPECIFIC PRECIPITATE FORMEDIN THE THYMOLTURBIDITY REACTION

ing reactive protein, always produced a positivereaction in proportion to the total lipid concentra-tion. The presence of the thymol-reacting pro-

tein was the essential factor while the level of thelipids affected only the intensity of the reaction.

II. Identification of the protein involved in thethymol turbidity reaction

Maclagan (9) analyzed the precipitate obtainedin a positive thymol turbidity reaction and foundthat it consisted of approximately 40 per centprotein. He suggested that this was probably a

gamma globulin because of its low solubility.Hanger and associates (20) were unable to obtainpositive thymol turbidity reactions by addinggammaglobulin electrophoretically separated fromhepatitis sera to normal sera and various lipidfractions. They concluded that the protein in-volved in the reaction wars probably an alpha or

beta globulin and not in the gamma globulinfraction.

(a) Electrophoretic analysis of the thymol pre-cipitate. Studies of -the precipitate obtained withthymol reagent were hampered by the large amountof lipid that was present. Turbid solutions were

always obtained when attempts were made todissolve the precipitate and it was impossible toobtain an electrophoretic pattern. Extraction of

the lipids from the precipitate in the cold withether alone, ether and alcohol, or acetone alone,always resulted in denaturation of the proteins inthe precipitate. When removed from serum theproteins were apparently less resistant to the ac-tion of organic solvents.

Attempts were also made to dissolve the pre-cipitate resulting from the thymol reaction in nor-mal serum and observe the change in the electro-phoretic pattern. The high lipid concentration,however, still interfered and no conclusive re-sults could be obtained. The difficulty was finallyovercome by the use of tween 80, a strong emulsi-fying agent. A 1 per cent concentration of thismaterial did not affect the electrophoretic patternof normal serum. A clear solution suitable forelectrophoretic determinations was obtained inthe following manner.

Two hundred ten ml. of thymol reagent were addedto 15 ml. of very reactive hepatitis serum (38 units) at00 C. The precipitate was collected by centrifugation inthe cold and suspended in 10 ml. barbital buffer con-taining 1 per cent tween at pH 8.5 and A = 0.1. This wasthen dialyzed in the cold against barbital buffer at thesame pH and ionic strength. A small amount of sedi-ment that still remained was thrown down by centrifuga-tion and the resulting supernate was quite clear.

Figure 5 shows the electrophoretic pattern ofthe protein solution obtained in the above manner.The sharp peak (b), representing almost the en-tire amount of protein present, had a mobility of1.6 x 10-5 which places it in the gammaglobulinfraction but with an unusually rapid mobility. A

(a) (b) (b) (a)

FIG. 5. ASCENDINGAND DESCENDINGELECTROPHORETICPATTERNS OF THE PROTEIN COMPONENTOF THE PRE-CIPITATE FORMEDIN THE THYMOLTURBIDITY REACTION

Peak (a) = f globulin (/A= 2.83 x 10-5); peak (b)= Y globulin (.a = 1.60 X 10-5).

1064

THYMOLTURBIDITY TEST IN LIVER DISEASE

small definite beta globulin peak (a) is also visible.In a second experiment with material from anotherpatient, lipid and tween were removed by sub-jecting the protein to repeated precipitations fromlarge volumes of solution by lowering the salt con-centration. The final preparation contained nolipid and the electrophoretic pattern showed asingle peak with the mobility of a gammaglobulin.Further experiments with the thymol-precipitatedprotein in the absence of tween indicated thatwhen all of the lipid was removed by centrifuga-tion at 12,000 r.p.m. at high salt concentrations,the remaining protein migrated extremely slowly.In 1 experiment the mobility was 0.5 x 10-5.It appeared as if the more lipid that was removed,the more slowly the remaining protein migrated.

(b) Effect of the addition of gammaglobulin tonormal serum. The importance of the gammaglobulin fraction was borne out in a series of ex-periments with gamma globulin separated elec-trophoretically from various types of sera. Con-siderable difficulty was encountered in adjustingthe protein concentration, the ionic strength, andthe pH of each fraction to exactly the same levelbecause of the small and variable amount of gammaglobulin that was obtained by electrophoretic sepa-ration. Slight variation in these fractions causedmarked differences in their effect on the thymolturbidity reaction of serum and, as a result, thismethod of studying the protein component respon-sible for the reaction was not completely satisfac-tory. However, it was possible to demonstratethat gamma globulin fractions from certain serademonstrating a very high thymol turbidity reac-tion, although negative alone, produced positivereactions with thymol reagent when added tonormal serum. The gammaglobulin precipitatedwith thymol reagent only in the presence of lipid.The amount of turbidity produced was propor-tional to the concentration of lipid in the serum towhich the gamma globulin was added. No tur-bidity was produced with serum in the presenceof a small amount of tween.

That the gammaglobulin fraction is importantin the thymol reaction as it is usually carried outwas also indicated by experiments where this frac-tion was removed from 2 positive sera electro-phoretically. The sera, after being brought totheir previous state in all respects, except that the

gammaglobulin was absent, now gave a negativereaction.

Comparison of gamma globulin preparationsfrom various sera in respect to their ability to in-duce a positive reaction in the presence of lipidwas attempted (Table III). Four mgm. of

TABLE III

Comparative effect of equal concentrations of gamma glob-ulin, obtained from various sources, on the thymol turbidityreaction of a high lipid serum

Thymol turbidityThymol of gammaglobulinSource of gamma turbidity on addition of

globulin of serum serum containing950 mgm.

per cent lipid

units units1. Hepatitis serum 40 262. Hepatitis serum 31 283. Normal serum 3 164. Normal serum 2 205. Cirrhosis serum 4 216. Multiple myeloma serum 3 8

gamma globulin were added to each 0.1 ml. ofserum used. The gammaglobulin obtained fromthe serum showing the pattern illustrated inFigure 6a was slightly more active in inducing apositive thymol reaction than was that obtainedfrom normal serum. It also appeared to be slightlymore active than was gamma globulin obtainedfrom the serum of a patient with cirrhosis of theliver which showed an increase in the gammaglobulin fraction but a negative thymol turbidityreaction. A definite and clear-cut difference inactivity was obtained when the hepatitis gammaglobulin was compared with gammaglobulin fromthe serum of a patient with multiple myeloma. Thelatter serum had given a negative thymol turbidityreaction. Since all the gamma globulin prepara-tions had some activity in the presence of lipid, itwas impossible to draw very definite conclusions.Certainly, the activity of the gammaglobulin prep-arations was not proportional to the activity ofthe sera from which they were obtained.

(c) The change in the electrophoretic patternof hepatitis serum follouning removal of the thymolprotein. In order to obtain further informationabout the protein or protein complex that is pre-cipitated with thymol reagent, electrophoreticanalyses were carried out on 3 highly active hepa-titis sera before and after removal of the thymol

1065

HENRYG. KUNKELAND CHARLESL. HOAGLAND

a

FIG. 6. ASCENDING ELECTROPHORETICPATTERNS BEFORE AND AFTER REMOVALOFTHE THYMOLPRECIPITATE IN 3 PATIENTS DEMONSTRATINGVERY HIGH VALUES FORTHE THYMOLTURBIDITY TEST

The dotted lines indicate the areas which were altered by the precipitation.

precipitate. The thymol precipitate-free serumwas obtained in the following manner.

To 7 ml. of hepatitis serum, 98 ml. of thymol reagentwere added. The precipitate was permitted to settle for30 minutes and then removed by centrifugation. Theclear supernatant solution was dialyzed against salineuntil all traces of thymol were removed. The volumeof solution was brought down by evaporation in thedialysis bag placed in front of an electric fan. Thisprocedure was continued until the protein content of thesolution was equal to that in the original serum minusthe quantity in the thymol precipitate. Electrophoreticanalyses were then carried out after the usual adjustmentto a pH of 8.6 and /A= 0.1 by dialysis against barbitalbuffer.

By superimposing the tracings obtained beforeand after removal of the thymol precipitate (Fig-'ure 6), the differences in the electrophoretic pat-terns of the 3 sera are clearly visible. The pre-dominant alteration was in the beta globulin frac-tion. The question arises as to whether the changein this fraction is due solely to the removal oflipid from the serum as a result of the thymol pre-cipitatiofn. A decrease in the beta globulin isknown to occur with extraction of lipids from theserum by means of organic solvents. Calculationby planimetry of the protein lost from the gammaglobulin peak showed some variation in the as-cending and descending patterns of the same se-rum. In addition, the amount of protein lost bythe precipitation was small in comparison with

the total protein, less than 8 per cent in all 3 cases.It was, therefore, difficult to obtain exact evidenceas to how much of the protein precipitated camefrom the gamma globulin fraction. The figuresobtained ranged between 25 and 45 per cent withan average of 37 per cent for the ascending anddescending patterns of the 3 sera. These resultsindicated a portion of the precipitated protein camefrom the gamma globulin fraction but that thelargest portion came from the beta fraction.

(d) Immunological results. In order to find outmore definitely whether the globulin precipitatedin the thymol turbidity reaction is an abnormalprotein or just an increased amount of normal pro-tein in the serum, antibodies to the protein illus-trated in Figure 5 were obtained in rabbits. Theantiserum reacted strongly with normal serum, andno definite difference could be obtained with seracontaining large amounts of gamma globulin orsera giving a very positive thymol turbidity reac-tion. Absorption of the antiserum with normalserum did not aid in demonstrating a differencebetween normal and hepatitis serum.

(e) Effect of albumin on the thymol turbidityreaction. In view of the known effect of albumin inincreasing the solubility of proteins in general andthe specific effect of albumin on the cephalin floc-culation reaction (8), studies were carried out todetermine its effect on the thymol turbidity reac-tion. Figure 7 demonstrates the fall in the values

1Q66

THYMOLTURBIDITY TEST IN LIVER DISEASE

for the thymol turbidity test in the serum of a pa-tient with hepatitis upon the addition of increasingamounts of concentrated human albumin. Thealbumin concentration of the serum had to be al-most doubled before much change occurred.

in the body, albumin may have a specific effect onthe reaction not encountered in the in vitro ex-periment. The level of albumin in the body, there-fore, is another factor influencing the intensity ofthe thymol turbidity reaction.

1~20,a-toc

15

10

5 10 15 20Albumin added (gm 7%)

FIG. 7. THE EFFECT OF THE ADDITION OF VARIOUSAMOUNTSOF HUMANALBUMIN ON THE THYMOLTUR-BIDITY REACTION IN THE SERUMOF A PATIENT WITHINFECTIOUS HEPATITIS

III. Clinical observations on the effect of the pro-tein and lipid components of the serum on thethymol turbidity reaction during the course ofacute infectious hepatitis.In order to demonstrate further the influence of

the lipid level of the serum on the reaction understudy, lipid determinations were carried out on

TABLE IV

Average lipid concentration in the serum of patients withacute infectious hepatitis showing high thyMol turbidity testvalues compared with those showing low values

Group25 30

No. ofpatients

Thymolturbidity

test values

Averagetotal lipid

concentration

A 6 above 25B 6 below 15

2 groups of patients during the early icteric phaseof acute infectious hepatitis (Table IV). The

In vivo studies with concentrated solutions ofhuman serum albumin in patients with diseasesof the liver have-shown a definite decrease in theintensity of the thymol turbidity reaction of the se-

rum following intensive therapy with this material.The decrease in the thymol turbidity reaction couldbe accounted for partly on the basis of an increasein plasma volume and on the in vitro effect of al-bumin described in the previous paragraph. How-ever, in certain patients the fall was too great tobe explained in such a manner. An example ofsuch an effect was seen in the case of a 12-year-oldboy with very severe acute infectious hepatitis ac-

companied by edema and ascites. The administra-tion of three 25-gram units of albumin produced a

fall in the thymol turbidity reaction from 27 to 7units with an increase of only 1 gram per cent inthe serum albumin level. The change in plasmavolume in this patient was approximately 10 per

cent. The cephalin flocculation reaction showedno change during this period. The.thymol tur-bidity reaction returned to its original level within6 days following cessation of therapy. Apparently,

3.0 3.5Globulin (3m. per cent)

4.0

FIG. 8. RELATION BETWEENTHE ELEVATION OF THE

THYMOLTURBIDITY TEST AND THE GLOBULIN LEVEL OF

THE SERUMDURING THE CONVALESCENTPERIOD OF ACUTEINFECTIOUS HEPATITIS

\

I I

1067

HENRYG. KUNKELAND CHARLESL. HOAGLAND

30

TION INIILY &NDBEWE0HMLTRIIYADTLGOUIEEMNTOSDRNH--TER4PERIODINARELAPSEOFINFECTIOUSHEPATITIS

4 No12 24 56 48 60 72 84 96 108 120 152144 156 168 180192 204

Day of disecuse

FIG. 9. ILLUSTRATION OF THlE CLOSE PARALLELISM BETWEEN THYMOL TURBIDITY AND LIPID DETERMINA-TIONS INITIALLY, AND BETWEEN TIIYMOL TURBIDITY AND TOTAL GLOBULIN DETEMINATIONS DURING THELATER PERIOD IN A RELAPSE 'OF INFECTIOUS HEPATITIS

severity of the illness in the 2 groups was ap-proximately equal and they differed only in theintensity of the thymol turbidity reaction. Noneof the patients showed more than slight elevationof the globulin level. It can be seen that the aver-age lipid concentration in the sera of patients withhigh values for the thymol turbidity test was muchgreater than in the group showing low values.

During the convalescent period of infectioushepatitis the influence of the lipid level becameless pronounced and values for the thymol turbiditytest paralleled the globulin level of the serum to aclose degree. Figure 8 demonstrates this rela-tion in a group of patients during convalescenceat a time when the lipid changes were minimal.

The comparative influence of the protein andlipid components on the thymol turbidity reactionduring the course of infectious hepatitis may alsobe seen from Figure 9. This chart illustrates theclose parallelism of values obtained by simultane-ous determinations of the thymol turbidity reac-tion and plasma lipids in demonstrating a delayedrise during the course of a recurrence of acuteinfectious hepatitis. The total globulin levelshowed an even more delayed elevation and dur-ing the late convalescent period the thymol tur-bidity reaction followed the pattern of the globulinsvery closely. Electrophoretic determinations dem-onstrated that the rise in total globulin was al-

most entirely due to an increase in the gammaglobulin fraction.

DISCUSSION

The experiments that have been described makeit clear that there are 2 main components in serumthat are important in the thymol turbidity reac-tion: (a) the protein component, (b) the lipidcomponent. Electrophoretic analyses have indi-cated that these 2 components are not separate butare intimately related. The entire question re-volves about the difficult problem of loose protein-lipid combinations that migrate in the beta globulinfraction of serum. The ease with which thesecombinations are broken makes it difficult to studythe lipid-protein complexes individually. The ob-servations reported have shown a slower electro-phoretic migration of the protein as the lipid isremoved and suggest the possibility that somebeta globulins may represent light combinationsof lipid and gamma globulin.

In view of the specific effect of thymol on lipidsin general, it seems logical to suppose that dilutionof serum with a thymol solution would tend tothrow lipid-protein complexes out of solution.The protein is carried along by the effect of thymolin decreasing the solubility of the lipids. In ad-dition, the low ionic strength buffer used in thereagent will sometimes precipitate gammaglobulin

1068.

THYMOLTURBIDITY TEST IN LIVER DISEASE

from hepatitis serum (18) in the absence of thy-mol. Such a buffer solution has been used byWolff for detecting the globulin elevation in ma-laria serum (21). The effect of the thymol inincreasing the precipitate appears to be attributableto the action of thymol in decreasing the emulsifica-tion of lipids. The inhibitory effect of emulsifyingagents on the thymol turbidity reaction also indi-cated that the protein precipitation was enhancedby the effect of thymol on the lipids in the serum.The reaction, therefore, probably results from acombination of the effect of thymol in decreasingthe dispersion of the lipid component, and theeffect of dilution with low ionic-strength buffer indecreasing the solubility of the globulin component,in such a way as to precipitate the protein-lipidcombination.

The fact that the lipids of the serum play agreater role in the thymol turbidity reaction thanjust forming a part of the protein-lipid-beta glob-ulin complex essential to the reaction, is indicatedby the observations which demonstrated that theper cent lipid in the thymol precipitate is propor-tional to the total lipid in the serum of patientswith infectious hepatitis. In addition, the experi-ments with mixtures of 2 different sera showedthat a high lipid serum, although negative by it-self, markedly affected the intensity of the re-action of the combination.

The effect of gamma globulin in acting withnormal serum to give a positive thymol turbidityreaction suggests a similarity to the cephalin floc-culation reaction where such an effect is also seen(8). One of the main points of dissimilarity be-tween the 2 reactions that has been cited (20) isthat extraction of the lipids from hepatitis serumdoes not affect the cephalin flocculation reactionas it does the thymol reaction. This, however, can-not be considered a fundamental difference be-cause, in the cephalin flocculation reaction, thecephalin-cholesterol reagent used supplies amplelipid to an extracted serum. The thymol alters thestate of the lipids of the serum so that a lipidglobulin complex precipitates from the serum ofcertain patients with liver involvement. In thecephalin flocculation reaction, a lipid suspensionis added to the serum so that a lipid globulincomplex precipitates. The difference in the 2tests is mainly in the manner in which the lipid

globulin precipitation is brought about. Clinically,the 2 tests are related, although occasional markeddifferences do occur (11, 22 to 24). One of themain discrepancies is seen in the earlier develop-ment of a positive cephalin flocculation reactionduring the pre-icteric stage of acute infectioushepatitis. A possible explanation may lie in thefact that the lipids show a delayed rise in thisdisease, and, therefore, the effect of increasedlipids on the thymol reaction would not be present.

An increased concentration of gamma globulinwas found electrophoretically in all sera showinga positive thymol turbidity reaction. Most of thepositive sera but not all also showed increasedamounts of beta globulin. There is difficulty inexplaining negative reactions in certain patientswith cirrhosis of the liver who have markedly in-creased gammaglobulin levels, high beta globulinconcentrations, high total lipid values and, in ad-dition, low plasma albumin. This should be theideal situation for a very positive thymol turbidityreaction. The evidence for -a qualitative differencein the increased globulin is not sufficiently con-clusive to explain such a case entirely, especiallyin view of the negative results obtained in immuni-zation experiments. Similar discrepancies havebeen noted in the cephalin flocculation reaction andHanger and his associates (8) have obtained evi-dence that albumin plays an important role. Thisproblem requires further study.

In the course of studies on patients with varioustypes of liver disease associated with marked hy-pergammaglobulinemia, it was noted that most ofthem had equally high thymol turbidity values:approximately 35 units. This was also true ofsera from patients with kala azar and schistosomi-asis with marked hypergammaglobulinemia. Thisrepresented a protein precipitation of approxi-mately 0.8 gram, a very small portion of the gammaglobulin of these sera. This was not true of pa-tients with multiple myeloma. The gammaglob-ulin elevation in this disease appears to be verydifferent. A negative thymol turbidity reactionwas always obtained. It appeared as if therewas a maximal amount of protein that could beprecipitated. A possible explanation for this phe-nomenon may reside in the fact that all these serashowed approximately the same amount of lipidand, therefore, only a limited amount of lipid was

1069

HENRYG. KUNKELAND CHARLESL. HOAGLAND

available .to precipitate with globulin in the pres-ence of thymol reagent. More protein readily pre-cipitated if additional lipid was added.

The close relation between the level of the lipidsin serum and the intensity of the thymol turbidityreaction must be kept in mind in evaluating ob-servations by this test because of the frequent oc-currence of elevation of the plasma lipids in liverdisease. The dependence of the test on 2 variablefactors, the protein and lipid components, is some-what of a disadvantage to a clear analysis of thesignificance of the reaction. This may be thereason for the lack of correlation between the in-tensity of the reaction and the severity of illnessobserved in a clinical study of infectious hepa-titis (22).

Serial determinations of the thymol turbiditytest during the course of infectious hepatitis haverevealed that values for the test show a delayedrise following the onset of the disease and a pro-longed elevation during convalescence (25). Therelative effect of the lipid and globulin componentsof the reaction upon this pattern was demon-strated. The values for the thymol turbidity testclosely paralleled the plasma lipids during theacute phase of the disease and the total globulinlevel during convalescence. The prolonged highvalues for the thymol turbidity test following in-fectious hepatitis were due to persistent elevationof the serum globulins.. These observations weresubstantiated by electrophoretic patterns takenduring the course of this disease. During theearly period of elevation of values for the thymolturbidity test the dominant abnormality was anincrease in the electrophoretic beta globulin. Thiswas a reflection of the elevated plasma lipids.Later the predominant aberration was an increasein the gamma globulin, which remained elevatedas long as the thymol turbidity reaction remainedpositive. Thus, it is apparent that the thymolturbidity test has a different significance duringdifferent stages of a single disease such as acuteinfectious hepatitis.

Determination of the lipid changes alone is nota sensitive index of acute liver damage, and thereis no simple specific method of determining thesmall increases in globulin following such damage.The thymol turbidity test in reflecting the com-bination of these aberrations is a more sensitive

index of liver injury than either one alone. Sincethe essential factor in the. mechanism of the reac-tion is the increase in the globulin component fol-lowing acute liver injury, investigations of thesignificance of the reaction really call for an un-derstanding of the corresponding hyperglobu-linemia.

The serum globulins are known to show a de-layed rise in several diseases; this has been con-sidered to be a reflection of the development ofantibodies. Recent studies on typhus (26), how-ever, have demonstrated that complement fixationtests and the Weil-Felix reaction did not parallelthe alteration in globulin. The explanation forthe change in liver disease is also very obscure.Liver biopsies obtained from the patients withmarked hyperglobulinemia have revealed markedevidence of regeneration of liver tissue, suggestingthat globulin elevation may be correlated with thehealing process.

SUMMARY

1. The turbidity produced by the thymol reagentof Maclagan in the serum of patients with infec-tious hepatitis is shown to depend on the presenceboth of lipids and of abnormal lipid protein com-plexes migrating in the beta globulin fraction ofthe serum. The gammaglobulin fraction of serumalso plays an important role in the reaction. Therelative importance of the different componentsin the reaction varies with different sera. De-velopment of the turbidity is prevented if thelipids are kept in solution by the addition of atween or are extracted with ether, or if the gammaglobulin is removed.

2. In lipemic sera the thymol reagent causes anonspecific increase in the turbidity due to in-crease in particle size of the lipid globules. Cor-rection for this effect can be made by using a photo-metric blank prepared from serum and a thymolreagent containing such a high buffer concentra-tion that precipitation of the globulin componentis prevented.

3. Immunological studies did not reveal evidencethat the protein concerned in the reaction wasabnormal.

4. A possible similarity between the mechanismof the thymol turbidity and cephalin flocculationreactions was discussed.

1070

THYMOLTURBIDITY TEST IN LIVER DISEASE

5. In patients with acute infectious hepatitisvalues for the thymol turbidity test were found toparallel alterations in serum lipids initially and al-terations in gammaglobulin during late convales-cence.

BIBLIOGRAPHY

1. Filinski, W., L'Augmentation de taux de la globulindans le serum du sang comme resultat de l'insuffi-sance hepatique. Presse Med., 1922, 30, 236.

2. Wiener, H. J., and Wiener, R. E., Plasma proteins.Arch. Int. Med., 1930, 46, 236.

3. Foley, E. F., Keeton, R. W., Kendrick, A. B., andDarling, D., Alterations of serum protein as anindex of hepatic failure. Arch. Int. Med., 1937,60, 64.

4. Cassiano P., Applicazioni della reazione di Takata inpsichiatria. Schizofrenie, 1935, 4, 151.

5. Magath, J. B., The Takata-Ara test of liver func-tion. Proc. Staff Meet., Mayo Clin., 1935, 10,493.

6. Levinson, S. A., Klein, R. I., and Rosenblum, P.,The Weltmann serum coagulation reaction; Pre-liminary report. J. Lab. & Clin. Med., 1937, 23,53.

7. Kabat, E. A., Hanger, F. M., Moore, D. H., andLand6w, H., The relation of 'cephalin flocculationand colloidal gold reactions to the serum proteins.J. Clin. Invest., 1943, 22, 563.

8. Moore, D. B., Pierson, P. S., Hanger, F. M., andMoore, D. H., Mechanism of the positive cephalin-cholesterol flocculation reaction in hepatitis. J.Clin. Invest., 1945, 24, 292.

9. Maclagan, N. F., The thymol turbidity test as anindicator of liver dysfunction. Brit. J. Exper.Path., 1944, 25, 234.

10. Shank, R. E., and Hoagland, C. L., A modified methodfor the quantitative determination of thymol tur-bidity reaction of serum. J. Biol. Chem., 1946,162, 133.

11. Neefe, J. R., Results of hepatic tests in chronichepatitis without jaundice; correlation with theclinical course and liver biopsy findings. Gastro-enterology, 1946, 7, 1.

12. Longsworth, L. G., Recent advances in the study ofproteins by electrophoresis. Chem. Rev., 1942,30, 323.

13. Van Slyke, D. D., and Folch, J., Manometric carbondetermination. J. Biol. Chem., 1940, 136, 509.

14. Schoenheimer, R., and Sperry, W.M., A micromethod

for the determination of free and combined choles-terol. J. Biol. Chem., 1934, 106, 745.

15. McFarlane, A. S., Behavior of the lipoids in humanserum. Nature, 1942, 149, 439.

16. Rosenthal, S. M., and White, E. C., Clinical applica-tion of the bromsulphalein test for hepatic function.J. A. M. A., 1925, 84, 1112.

17. Howe, P. E., The use of sodium sulfate as the glob-ulin precipitant in the determination of proteins inblood. J. Biol. Chem., 1921, 49, 93.

18. Kunkel, H. G., The use of a serum dilution techniquefor the rapid and quantitative estimation of ele-vated globulin with special reference to liver dis-ease. To be published.

19. Swift, H. F., Wilson, A. T., and Lancefield, R. C.,Typing group A hemolytic streptococci by Mprecipitin reactions in capillary pipettes. J. Ex-per. Med., 1943, 78, 127.

20. Recant, L., Chargaff, E., and Hanger, F., Compari-son of the cephalin-cholesterol flocculation withthe thymol turbidity test. Proc. Soc. Exper. Biol.and Med., 1945, 60, 245.

21. Wolff, E. K., Buffer precipitation test (B.P.T.) formalaria. Tr. Roy. Soc. Trop. Med. & Hyg., 1939,32, 707.

%

22. Kunkel, H. G., Value and limitations of the thymolturbidity test as an index of liver disease. Am. J.Med., In Press.

23. Havens, W. Paul, Jr., and Marck, R. E., A com-parison of the cephalin-cholesterol flocculation andthymol turbidity tests in patients with experi-mentally induced infectious hepatitis. J. Clin. In-vest., 1946, 25, 816.

24. Watson, C. J., and Rappaport, E. M., A comparisonof the results obtained with Hanger cephalin-cholesterol flocculation test and the Maclaganthymol turbidity test in patients with liver dis-ease. J. Lab. and Clin. Med., 1945, 30, 983.

25. Kunkel, H. G., and Hoagland, C. L., Persistence ofelevated values for the thymol turbidity test follow-ing infectious hepatitis. Proc. Soc. Exper. Biol. andMed., 1946, 62, 258.

26. Tierney, N. A., and Yeomans, A., Metabolic studiesin louse-borne typhus. Observations on serum'electrolyte pattern, serum protein partition, andnitrogen balance. J. Clin. Invest., 1946, 25, 822.

ACKNOWLEDGMENT

The authors are indebted to Dr. T. Shedlovsky andDr. D. A. MacInnes for their generosity in making avail-able the electrophoretic equipment used in this study.

1071