J. clin. Path. (1962), 15, 221

The determination of cholesterol in serum

by persulphuric acid oxidationR. N. BEALE AND D. CROFT

From the Institute of Clinical Pathology and Medical Research, DepartmentofPublic Health, Sydney, New South Wales

SYNOPSIS A method for the determination of total cholesterol in serum is described. Optimumconditions are found for the isolation of the steroid free from extraneous chromogens, and a new

sensitive colour reagent is introduced. The procedure is convenient for routine application, the finalcolour being light-stable and reasonably time-stable at room temperature. Comparisons are madewith one other method and with another colour reagent. It is concluded that, for accurate results,treatment of the serum with alcoholic alkali followed by extraction with a hydrophobic solvent isessential. The choice of the colour-developing reagent is then no longer dependent on specificityfor cholesterol, but becomes inter alia a matter of ease of operation, and of sensitivity, stability, andreproducibility of the final colour. These points are discussed in relation to other methods.

The problem of determining accurately the totalcholesterol content of blood serum may be resolvedinto two main parts, namely, the quantitative isola-tion of the cholesterol in a pure state and measure-ment of the isolated steroid by means of convenientdeveloping reagents which lead to a sufficientlysensitive and stable colour.The first part of the problem has been attacked in

many ways by numerous workers, for example, byextraction with one or more organic solvents withconcomitant precipitation of proteins (e.g., Bloor,1916; Sackett, 1925); by similar extraction followedby precipitation of the cholesterol in the hydrolysedor unhydrolysed extract with digitonin (Sperry andWebb, 1950); by treatment with ethanolic potassiumhydroxide followed by extraction (Abell, Levy,Brodie, and Kendall, 1952; Trinder, 1952; Andersonand Keys, 1956; Brown, 1959); or by combinationsof these. Others (e.g., Zlatkis, Zak, and Boyle, 1953)have avoided isolation and added colour reagentsdirectly to the serum.The second part of the problem has also received

much attention and since 1953 a number of methodshave been published which are improvements onearlier techniques of colour development employingthe Liebermann-Burchard reaction, the short-comings of which are well documented.The present work has led to a colorimetric method

which incorporates the best features of those pre-

Received for publication 5 October 1961.

viously described, reduces operations to a minimum,and introduces the use of persulphuric acid oxidationin an acetic-sulphuric acid medium. The colourproduced is completely stable to light and sufficientlyso to time, and the method has good sensitivityrelative to others.

REAGENTS

Chemicals of A.R. quality should be used where possible.

1 ALCOHOL Absolute or 95% ethanol is suitable.

2 6M POTASSIUM HYDROXIDE IN METHANOL Dissolve84 g. potassium hydroxide pellets in about 150 ml.methanol, cool and dilute to 200 ml. This reagent neednot be filtered. It keeps for long periods in the refrigerator.

3 M POTASSIUM HYDROXIDE, ALCOHOLIC Dilute reagent(2) 1 in 5 with reagent (1). This is best done in smallquantities as it deteriorates if ordinary ethanol isemployed. The reagent need not be filtered.

M POTASSIUM HYDROXIDE IN PURIFIED ALCOHOL This isan alternative to the preceding reagent. Ninety-five to100% ethanol is purified as follows. Add about 0-5 g.2: 4-dinitrophenylhydrazine to each litre of alcoholfollowed by a few drops of concentrated sulphuric acid.The mixture is then refluxed for four to five hours (pre-ferably with a gentle stream of nitrogen bubbling throughthe liquid), and left sealed from air overnight. Thefollowing day, the liquid is distilled with an efficient anti-splash head or a fractionating column to avoid carry-over of the hydrazine derivative. A small first fraction is

221

on April 6, 2020 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.15.3.221 on 1 May 1962. D

ownloaded from

R. N. Beale and D. Croft

rejected and about 800% of the remainder collected;nitrogen is again bubbled during distillation. The productis aldehyde-free. Dissolve 35 g. potassium hydroxidepellets in aldehyde-free alcohol, cool, and dilute to 500 ml.with the alcohol. This reagent is fairly stable and willkeep at least two months in the cold. Reject it whensigns of yellowing appear.

4 PETROLEUM SPIRIT, BOILING-RANGE 60 TO 80'C.

5 ACETIC ACID Use glacial acetic acid; no purificationis necessary.

6 PERSULPHURIC ACID REAGENT Dissolve 20 mg.potassium persulphate in 1 ml. cold water and add rapidly100 ml. chilled concentrated sulphuric acid. Mixthoroughly. This reagent will keep for about threeweeks in the refrigerator where it should be stored whennot in use, and is rejected when the colorimeter readingfor a 600 mg. % cholesterol standard is markedly lowerthan with fresh reagent.

7 STANDARD CHOLESTEROL Dissolve 10 mg. purecholesterol (e.g., Townson and Mercer's, chemically pure)in ice-cold petroleum spirit, and dilute to 100 ml. withcold solvent. Store in the refrigerator when not in use.Check the purity of the cholesterol by comparison withan authentic specimen, as some samples contain extran-eous substances which are difficult to remove by recrystal-lization. Purity is checked by developing the colour witheach sample and comparing the visible spectra, or thereadings at about 498 mu if a filter instrument is used.

Aliquots of the standard solution are pipetted intotest-tubes and the solvent evaporated in a water-bath at80 to 100°C. by blowing a gentle stream of air into thetube which is kept agitated. Nitrogen, instead of air,does not appear to be necessary. The dry standards arestable indefinitely, so that if desired a stock of these maybe prepared.

METHOD

Measure 0-15 ml. serum into a glass-stoppered tube ofabout 20 ml. capacity. Then add 3 ml. M potassiumhydroxide, mix and heat in a 60'C. water-bath for 15minutes. Cool the tube to room temperature and add 3 ml.water and 6 ml. light petroleum. Shake vigorously for30 seconds to one minute and stand the tube for a moment.A clean separation occurs almost instantaneously. Afterthe phases have separated, withdraw 2 ml. of the petrollayer and pipette into an ordinary rimmed test-tube.Evaporate the petrol as described previously. Theresidue is stable and may be stored indefinitely at thisstage. For colour development add 5 ml. glacial aceticacid and allow the tube to stand for about five minuteswith occasional shaking to dissolve the cholesterolresidue. Then run 2-3 ml. persulphuric acid reagentdown the wall of the sloped tube so as to form a layerbeneath the acetic acid solution; allow a minute fordrainage. Then shake the tube vigorously for about half aminute and transfer it without delay to a 37°C. water-bath where it is allowed to incubate for five to 10 minutes.

Cool the tube to room temperature in water avoidingvigorous agitation and transfer the product to a colori-meter tube or spectrophotometer cell. Read against ablank made by mixing 5 ml. acetic acid and 2-3 ml.persulphuric acid reagent. The product has maximumabsorbance at 498 mi; if a colorimeter is to be used afilter similar to the Ilford 623 is suitable.For calibration measure 1, 2, and 3 ml. cholesterol

standard into rimmed test-tubes, evaporate the solventand then proceed as for the colour development of theserum residue. These standards are equivalent to seracontaining 200, 400, and 600 mg. % cholesterol. A cali-bration graph may be used, but twice weekly checks witha duplicate standard are advisable. If preferred, watermay be used as a blank in setting up the calibration andfor future analyses.

EXPERIMENTAL

PREPARATION OF THE SERUM FOR COLOUR DEVELOP-MENT In view of the variety of methods used forliberating cholesterol and its esters from serumconstituents, we attempted initially an assessmentof the various procedures for isolation such as thosementioned previously. Our objective was to discoverone which would separate the cholesterol either assuch or mixed with its serum esters, would give aquantitative recovery of the total cholesterol, andwould lead to an isolated product free of extraneouschromogens; in addition it was desirable that anycoloured substances such as bilirubin or haemo-globin should be left behind during the isolationprocedure. The criterion used was that based on anexamination of the visible absorption spectra of thecolours derived from treated serum and from purecholesterol. If the ratio of the absorbances at aseries of corresponding wavelengths remained con-stant, then it could be concluded either that thecholesterol had been obtained in a pure state (ortogether with esters which gave the same absorptionspectra on a molar basis as free cholesterol) or thatthe method of colour development was specificfor cholesterol, or both. Purity of the isolatedcholesterol was gauged by colour development ofdried extracts and standards with a modifiedLiebermann-Burchard reagent (Anderson and Keys),with the reagents of Zlatkis et al. as modified byHenly (1957) but without purification of the aceticacid, and also with our reagents.The methods of extraction which were investigated

involved addition of acetic acid and centrifugation;addition of acetic and trichloracetic acids followedby centrifugation; hot and cold extraction withvarious solvents or mixtures of solvents; and lastlytreatment with aqueous, methanolic or ethanolicpotassium hydroxide, or with mixtures of these, atvarious temperatures and for varying times, followedby petrol extraction.

222

on April 6, 2020 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.15.3.221 on 1 May 1962. D

ownloaded from

The determination of cholesterol in serum by persulphuric acid oxidation

Examination of the spectra of the final productsfrom the first three types of extraction procedureshowed lack of correspondence with the spectrumfrom pure cholesterol; in all cases extraneousabsorption was observed giving falsely high values,so that such methods were rejected as unsatisfactoryfor accurate work. Since the result was much thesame for any of the three sets of colour reagents usedit appears that none of these methods of colourdevelopment is specific for cholesterol; contributionof non-reacting colour from the serum itself is afurther possible source of positive error. Thesefindings are supported by the work of Morris (1959)who carried out a painstaking comparison of sixmethods. If the work of Schoenheimer and Sperry(1934) and of Sperry and Webb (1950) is taken as abase-line for comparative purposes it is apparentfrom Morris's data that the methods of Pearson,Stem, and McGavack (1953), Zlatkis et al. (with orwithout purified acetic acid), and Sackett give meanvalues significantly higher; furthermore, acid extrac-tants lead to higher values than neutral organicextractants.

Attention was therefore directed to treatment ofthe serum with alkali and petrol, and by this meanssatisfactory results could be achieved under optimumconditions (the optima not being particularly critical,as will be described). It was found possible to pro-duce spectra from the extracts which were virtuallyidentical with those from pure cholesterol (Fig. 1).Since the colour reagents mentioned above are notspecific, it must be concluded from this that bytreatment with alcoholic potassium hydroxide andextraction with petrol the cholesterol is isolatedfrom serum in almost pure form, the impuritieshaving negligible chromogenic potentialities with atleast some colour reagents. This finding confirms

0-4

z 03z

0U)0-2 -ID

z 0.-I3J

400 425 450 475 500 550 600WAVELENGTH (mp.)

FIG. 1. Absorption spectrum of colour from a 270 mg. %cholesterol standard (full curve) and the same from a

processed serum, showing freedom from extraneousabsorption (broken curve).5

the work of Abell et al. who showed by counter-current distribution that the cholesterol obtained bytheir procedure is 99% pure.A number of variations in detail for alcoholic

alkali treatment have been published. Trinder heatsto boiling with 0-5 N potassium hydroxide and dis-continues as soon as steady boiling occurs, whileothers (e.g., Anderson and Keys; Sobel, Goodman,and Blau, 1951; and Brown) heat at 370, 450, or65°C. for 90 minutes or less. Little specific investi-gation of the variables such as alkali concentration,temperature and time of heating, and nature of sol-vent seems to have been published, although accord-ing to Bloor, Pelkan, and Allen (1922) heating withstrong alkali is dangerous, a contention which seemsto have passed unchallenged subsequently. Some ofthe parameters were explored, therefore, in order todetermine the optimum conditions for isolation. Thecolour reagent described under 'Method' was used.

Time of heating at 60°C. Treatment with Mpotassium hydroxide in alcohol (95 % or more)gave 55% maximum absorption at 498 miL for animmediate extraction, 94% after two minutes, andmaximum colour for between eight and 16 minutes'extraction; very slightly lower values were obtainedat 24 and 30 minutes. The best spectral purityrelative to a standard was obtained after 15 to 20minutes' heating and this period is recommended.

Concentration of alkali Aliquots of a serumwere incubated at 60°C. for 15 minutes with 0-5, 1, 2,and 4 M alcoholic potash and processed as described.No significant difference between any of the re-sultant spectra was observed. M potassium hydro-xide was chosen to allow for any possible weakeningof the solution with time, due to carbonate for-mation. At 60°C., therefore, considerable stability ofcholesterol in the presence of alkali is indicated.

Heating temperature Treatment with 0-5 Malcoholic alkali at the boiling point was the onlyother temperature variant investigated. UnderTrinder's conditions, described previously, a some-what lower absorption at 498 m,u was obtained com-pared with the present method. If, however, themixture was refluxed for two to eight minutes, aresult not significantly different from 15 minutes'treatment at 600C. was obtained. A comparisonwith Trinder's method is described later.

Variation ofsolvent When aqueous or methanolicpotash of any concentration was used, poor re-coveries of serum cholesterol were obtained. On theother hand addition to ethanolic potassium hydro-xide of water up to 16% v/v or of methanol up to17% v/v had no effect on the final result.

THE COLOUR DEVELOPING REAGENT Many colourreactions for cholesterol have been described,

223

on April 6, 2020 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.15.3.221 on 1 May 1962. D

ownloaded from

R. N. Beale and D. Croft

mostly employing a combination of two or more ofthe solvents acetic acid, acetic anhydride, sulphuricacid, chloroform, and more recently, ethylenedichloride and acetyl chloride. Reaction is carriedout in the presence of an oxidizing agent or agentssuch as sulphuric acid, ferric chloride, or perchloricacid. The most sensitive so far described for analy-tical purposes is that of Brown who modifiedTrinder's procedure by replacing sulphuric acid withperchloric acid. Both these reactions use ethylenedichloride which is toxic, and acetyl chloride, thevapour of which is volatile and hydrolyses readily inair to acetic and hydrochloric acids. The purity ofethylene dichloride appears to be critical for coloursensitivity and stability in Brown's method. Whenwe used laboratory reagent grade material which hadno processing other than redistillation,with rejectionof the first and last runnings, the colour fadedobviously, a drop of 25% being observed in twohours when the solution stood in diffuse daylight.In direct sunlight the colour faded to yellow in 25minutes. When the solvent was purified and driedas recommended and clean, dry apparatus used,with precautions to exclude moisture, considerableimprovements in stability were obtained for bothsera and standards. After 30 minutes in the darkonly about 2% of the colour faded; after an equi-valent time in diffuse daylight, a drop in colour ofabout 4% was found; the blank was unchanged.The colour was decidedly unstable to sunlight, 10minutes' irradiation causing a decrease of 6 to 10%

in the colorimeter reading.Ferric chloride oxidation in acetic-sulphuric acid

(Zlatkis et al., 1953) is an amenable reaction forroutine purposes if used in conjunction with theserum treatment described under 'Method'. It has theadvantage of employing a stable oxidizing colourreagent (if modified by incorporating the ferricchloride in the acetic acid rather than the sulphuricacid) but the disadvantage of producing a colourwhich is markedly unstable to light. After 15 minutesin bright diffuse daylight at room temperature thesolution gave a spectrum which showed a drop of9 %0 in maximum absorbance at 560 m,u whencompared with the dark-developed spectrum. After11 minutes in direct bright daylight the originalspectrum changed markedly and a brown colour wasformed. If the colour is developed in the light, it hasquite a different shade to a duplicate developed indarkness. On the other hand, the time stability ofthe colour in darkness is good, little change beingobserved over several hours. Another disadvantage,as pointed out by Morris, is the tendency forstationary air bubbles to form in the viscous acidmixture. These are difficult to remove except bycentrifugation and interfere during colorimetry,

giving erroneously high readings. At lower sulphuricacid concentrations less viscous media are producedbut there is considerable loss of sensitivity.

In the present work we investigated a number ofcombinations of solvents such as those referred toearlier, these being used in the presence or absence ofan oxidant. Of the solvents, mixtures of acetic andsulphuric acids were considered to be the most con-venient for several reasons; they are relativelyinexpensive and readily available in large quantitiesin most laboratories; they may be used under someconditions, such as the present ones, without time-consuming purification; they are relatively non-volatile and they give a medium with good potentia-lity for sensitive colour reactions. In conjunctionwith such a solvent mixture a wide range of oxidantswas examined and those were selected which pro-duced the most intense colour. Three gave acceptableresults, namely, ferric ion, periodate, and persulphate.Ferric salts were excluded by reason of the lightinstability and viscosity of the coloured product;periodate was somewhat inferior to persulphatesince over-oxidation occurred and colour faded.Persulphate was therefore chosen since it was freefrom these disadvantages.

Variation in concentration of constituents Havingdetermined the optimum proportions of acetic acid,sulphuric acid, and persulphate, we varied each inturn from the optimum. A deficiency of sulphuricacid led to considerable loss of sensitivity and anexcess gave too viscous a reagent and too high aninitial oxidizing temperature. The optimum desirablein order to avoid both these factors was found to be5 volumes of acetic acid to 2 to 24 volumes ofsulphuric acid. These proportions gave maximumcolour sensitivity at 498 m,u and a medium which,under the conditions used, did not often occludeair bubbles unless handled too vigorously aftercooling to room temperature. The concentration ofpersulphate was not critical; decreasing concentra-tion led to decrease in sensitivity and a differentabsorption spectrum, while excess persulphate led toover-oxidation and loss of colour. The optimumrange for a fresh reagent was found to be 10 to 25 mg.per 100 ml. sulphuric acid.

Conditions for colour development The markedrise in temperature on mixing the persulphuric andacetic acid layers assists in the rapid development ofan orange-red colour in the presence of cholesterol.The process is then completed by maintaining themixture above room temperature for a time. At 60°C.no more than two to three minutes is required butthis temperature causes small changes in the spectrumof the coloured product. Incubation at 37°C. for fiveto 10 minutes is as effective in producing maximumcolour and during this time no significant change in

224

on April 6, 2020 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.15.3.221 on 1 May 1962. D

ownloaded from

The determination of cholesterol in serum by persulphuric acid oxidation

the spectrum is observed, the absorption at 498 m,ubeing constant. The lower temperature is used,therefore, and the solutions are then ready forcolorimetry after cooling to room temperature.

Stability of colour The colour formed is insen-sitive to light: after 10 minutes' irradiation in brightsunlight outside the laboratory no significantchange in the spectrum was observed. The sameresult was found after 90 minutes' irradiation in acombination of direct sunlight and bright daylight.After a further 17 hours in diffuse daylight and dark-ness at about 20°C., the spectrum showed somechange, there being a drop in absorbance at 498 m,tof 4 to 5% and an increase of the same order in theregion of 430 m,u. The time of reading is not critical,therefore, but for the most accurate results measure-ments are best made within a few hours of colourdevelopment.

Obedience to Beer's law Cholesterol standardsup to a serum equivalent of 600 mg. % (up to 0 3 mg.)were developed and the absorbances at 498 m,u (ortheir colorimeter equivalent) were measured onvarious instruments. The readings on a moderatelyhigh resolution recording spectrophotometer showedobedience to Beer's law up to at least the highestlevel measured. The calibration graph is non-linearat the higher level for most colorimeters for theusual reason that light scatter in their optical systemslowers readings at the top end of the scale. It isprobably advisable therefore to confine colorimeterreadings to no more than 70% full-scale deflection.Within this range linearity is maintained for anaverage instrument.

RESULTS

RECOVERY OF CHOLESTEROL Since no way could beenvisaged of incorporating cholesterol into serumwhich would represent an approach to the true stateof distribution, it was not considered meaningful toadd cholesterol to serum in order to carry outrecovery experiments in the usual fashion. Instead,several dried standards were processed as for serumin order to discover whether losses occurred duringalkali treatment and solvent extraction or throughchemical breakdown of cholesterol. The meanrecovery of pure steroid was quantitative (100±2 %).This is not to say that those from sera would be assatisfactory, but other studies would be required todecide this point. In the present work we used themethod of alkali treatment which appeared to givethe maximum and most consistent recovery ofserum cholesterol and compared this with themethod of Trinder and with a modification of that ofZlatkis et al. In addition, the colour reagentsdeveloped by Trinder and by us were employed in

combination with both methods of alkali treatment.The results for a number of sera determined in theselatter ways are shown in Table I; all analyses werecarried out in duplicate and the individual meanstaken for comparative purposes.

TABLE I

COMPARISON OF SERUM CHOLESTEROL LEVELS OBTAINEDUSING COMBINATIONS OF TWO ALKALI TREATMENTS AND

TWO COLOUR REAGENTS

Serum Trinder Colour Reagent Present Colour ReagentNo.

Trinder Present Trinder PresentIsolation Isolation Isolation Isolation(mg. %) (mg. %) (mg. %) (mg. %)

263228155309230326328339359354

324141116216204203

14183

350138121223

280338

319134129218

35013713122627626427134215197

A number of points of interest emerge from TableI when the mean values for corresponding groups ofsera are compared. Either colour reagent gives, forthe same alkali treatment, means in close accord.When the two procedures for isolating the cholesterolare compared, however (using either colour reagent),the mean serum level with Trinder's treatment isabout 90% of that using the present one.

In Table 11 are shown the comparative results forsera analysed by the present procedure and by amodification of Henly's development of the Zlatkismethod. No purification of the acetic acid A.R.was carried out and the cholesterol was isolated asdescribed in this paper. The ferric chloride wasincorporated in the acetic acid as recommended byHenly. Procedures somewhat similar to the Zlatkis-Henly modification have been described recently byMann (1961) and by Connerty, Briggs, and Eaton(1961), all of whom, like Zlatkis et al., use ferric ion-sulphuric acid-acetic acid reagents. For each samplereferred to in Table II, four determinations werecarried out using a fresh aliquot of serum for eachof the four analyses. In the modified Zlatkis-Henlymethod, any bubbles in the coloured product wereremoved by light centrifugation of the colorimetertubes. Colour development was carried out in a darkcupboard and subsequent operations were per-formed as soon as possible in subdued light. As isapparent from Table II, no significant differencebetween the two sets of values was found.

225

on April 6, 2020 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.15.3.221 on 1 May 1962. D

ownloaded from

R. N. Beale and D. Croft

TABLE II CHOLESTEROL ESTERS The spectra of the colouredCOMPARISON OF PRESENT METHOD WITH A COMBINED derivatives from some esters of cholesterol wereMODIFICATION OF THAT OF ZLATKIS et al. (1953) AND OF measured before and after alkali treatment under the

HENLY (1957) described conditions. In all cases the spectra had the

Serum No. Present Zlatkis-Henly Percentage same shape with a maximum at 498 mu. However,(mg. %) (mg. %) Difference the band intensities (or the absorbance at 498 m,t)

818 166 166 167 166 0 varied slightly from that found for cholesterol.166 165 Small differences were observed for oleate (97±20%

952 162 161 159 164 +19 of cholesterol), acetate (104+2% of cholesterol), and159 169 for stearate (98 ±2-0% of cholesterol), the compari-

965 198 197 176 180 -8-6 sons being made on a molar basis. It would appear196 184 that, with the method described, the presence of even

988 206 207 203 211 4 1-9 considerable amounts of esters in the final extracts

024 181 178 173 176 -11 would have little effect on the values found for174 179 serum levels.

152 150 -1-3147208 203 -3-8197323 320 +0-6316109 110 +58110213 216 +1 9218

Mean difference -0-3%

REPRODUCIBILITY This was assessed for the presentmethod by processing 12 pairs of standards and 13pairs of test sera and applying the statistic

(S.E.M.)2 2Nwhere S.E.M. = standard error of measurement

a percentage difference between one

value of the pair and its meanN = number of pairs

The S.E.M. was ± 1-0%, so that the extreme con-

fidence limits for a single test are about ± 3 0%; thatis to say, in 99 7% of cases the measured value shouldbe no more than ± 3 % away from the 'true' value.Morris found Trinder's method to be the most repro-ducible of those he investigated, the S.E.M. being± 1 2 %; our value for Trinder's method was in closeagreement with this, being + 1P4%. The reproduci-bility of the present method therefore correspondsclosely with his, From Table II it appears that ferricchloride oxidation gives results not quite so reprodu-cible as those above, the S.E.M. being ± 1-5%. UsingZlatkis' original method, Morris found a S.E.M.of ± 1-6% -for unpurified acetic acid. As describedunder 'Experimental', the method of isolation is suchthat, with the present colour reagents, no interferingchromogens remain; neither is there the problem ofpossible variations in chromogenic power of differentesters ifcomplete hydrolysis is assumed. With deliber-ately haemolysed or highly icteric sera no interferencesdue to haem compounds or bilirubin were observed.

SENSITIVITY The most sensitive method so farreported is that of Brown, which he states to have8-2 times the sensitivity of the Liebermann-Burchardprocedure described by Abell et al. On the same

basis Trinder's method gives a factor of 4'6 and thatof Zlatkis et al. is 7-3. The present method is inter-mediate between those of Zlatkis et al. and Trinder inthis respect, having a factor of 6-2 when equal con-

centrations of cholesterol in the final product are

compared. In practice, all these more sensitiveprocedures give good colorimeter readings for quitesmall amounts of serum. For a 200 mg.% cholesterolstandard (100 ,ug. cholesterol) our end product has a

linear absorbance of about 0-31 at 498 m, for a 1 cm.light path with a reagent blank in the comparisoncell. Thus, as little as 30 ,ug. cholesterol can readilybe measured.

DISCUSSION

It emerges that the main requisite in determiningcholesterol is its quantitative liberation from theserum in a sufficiently pure condition, so that, withany chromofacient, the colour is due to cholesterolalone. This becomes necessary since, as may bededuced, for example, from Morris's data and as

appears from the present work, most colour reagentsare non-specific for cholesterol. Thus direct methodsfail to meet the situation; procedures involving onlyacid or solvent treatment of serum are inadequatealso, since extracted chromogenic substances, some-

times in considerable quantity, again contaminatethe cholesterol. When the Liebermann-Burchardreaction was applied to solvent-extracted and alsoto alkali-treated serum, only a small but significantdifference was found between the spectra of theresulting colours, so that the Liebermann-Burchardreagent (as modified by Abell or Anderson and Keys)approaches nearest to specificity for cholesterol inthe presence of solvent extractable contaminants.

950

563

025

576

599

150154209213321314104104216208

152

211

318

104

212

226

on April 6, 2020 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.15.3.221 on 1 May 1962. D

ownloaded from

The determination of cholesterol in serum by persulphuric acid oxidation

However, even if acceptable in this respect, the well-known limitations of the reaction in other respectssuggest the need for improved methods employ-ing more sensitive and convenient reagents andproducing more stable resultants.

Since it is difficult to ascertain whether or notrecoveries from sera are quantitative, the presentapproach has been to obtain maximum samplecolour together with correspondence of spectra fromsera and pure cholesterol. Such criteria could besatisfied by treatment under the conditions corre-sponding or approximating to those described under'Method'. When this treatment is combined withextraction by a hydrophobic solvent virtual freedomfrom extraneous chromogens is achievable with thereagents described under 'Method'.

In our hands, when Trinder's method of treatmentwas closely followed, apparently low recoveries ofcholesterol resulted; although if the alkali treatmentwas continued at the boiling point for as little as twominutes, close correspondence of serum levels wasobtained by both methods.

Little appears to be known concerning the natureof cholesterol binding in serum, but the purpose ofalkali treatment is sometimes stated to be that ofester hydrolysis. It would appear, however, aspointed out by Abell, that a more important functionof such procedures is to liberate the cholesterol andits esters from their complexes and combinations inserum, saponification being a concomitant process;otherwise it is difficult to account for the findings inTable I, as it has been shown that with our reagentsat least some esters have approximately the equivalentchromogenic power of cholesterol. A minimum timeappears necessary to complete this process ofliberation; thus, for example, Anderson and Keysuse 2% alcoholic potash at 37°C. for 90 minutes.Brown employs 2 5% alkali at 65°C. for 30 to 60minutes, while in the present method we find Malkali at 60°C. for 15 minutes to be adequate; thelatter conditions would be approximately equivalentto two minutes' refluxing at 90°C., if one assumes adoubling of reaction rate for each 100 rise in tem-perature. This may explain our lower results withTrinder's method since less than two minutes'refluxing is allowed. Thus, provided that treatmenttime exceeds an irreducible minimum, the conditionsof alkaline digestion are by no means critical andcomparable values are then obtainable with any oneof several colour reagents.

All of the latter so far discussed have been shownto be non-specific for cholesterol, but with extractionprocedures such as that described, interferingchromogens are of negligible proportions. Thechoice of reagent can subsequently be resolved intoquestions of sensitivity, reproducibility, ease of

handling and preparation, and time- and light-stability. In this latter respect the colour reagentdescribed under 'Method' is the only one of thosewe have examined which gives a product which isquite stable to light and at the same time does notnecessarily involve special purifications of solvents.No comparison between the present method and

a digitonin precipitation method has been attempted,but at first sight evidence could be adduced that suchmethods lead to low values for serum cholesterollevels. Thus Anderson and Keys find that their K5(alkali treatment) method gives values some 3%higher than the digitonin procedure of Foldes andWilson (1950). Since, as appears from the presentwork, this cannot be attributed to non-specificity ofthe K5 colour reagent, their alternative explanationof losses due to slight solubility of cholesteroldigitonide would seem possible. On the other hand,Morris obtains values 1 % lower by Anderson andKeys' than by Sperry and Webb's method. However,his values for Trinder's method, which would beexpected to agree with Anderson and Keys', aresignificantly higher than either the above-mentionedones by 6 to 7 %. It would be expected, therefore,from this evidence that our values also would behigher than those obtained using a digitonin pro-cedure by at least this amount. Assessment is mademore difficult by the fact that when Trinder com-pares his method with that of Sperry and Webb, heobtains very close agreement for the mean of 10 sera.One possible explanation would be that Morris usedslightly longer periods of alkali treatment than didTrinder, which, as we have shown, could lead tovalues up to 10% higher. Further critical com-parisons of the various methods are indicated.

This work is published with the approval of the Directorof State Health Services, Department of Public Health,New South Wales.

REFERENCES

Abell, L. L., Levy, B. B., Brodie, B. B., and Kendall, F. E. (1952).J. biol. Chem., 195, 357.

Anderson, J. T., and Keys, A. (1956). Clin. Chem., 2, 145.Bloor, W. R. (1916). J. biol. Chem., 24, 227.-, Pelkan, K. F., and Allen, D. M. (1922). Ibid., 52, 191.Brown, W. D. (1959). Aust. J. exp. Biol., 37, 523.Connerty, H. V., Briggs, A. R., and Eaton, E. H. (1961). Clin. Chem.,

7, 37.Foldes, F. F., and Wilson, B. C. (1950). Analyt. Chem., 22, 1210.Henly, A. A. (1957). Analyst, 82, 286.Mann, G. V. (1961). Clin. Chem., 7, No. 3, p. 275.Morris, T. G. (1959). J. clin. Path., 12, 518.Pearson, S., Stern, S., and McGavack, T. H. (1953). Analyt. Chem.,

25, 813.Sackett, G. E. (1925). J. biol. Chem., 64, 203.Schoenheimer, R., and Sperry, W. M. (1934). Ibid., 106, 745.Sobel, A. E., Goodman, J., and Blau, M. (1951). Analyt. Chem., 23,

516.Sperry, W. M., and Webb, M. (1950). J. biol. Chem., 187, 97.Trinder, P. (1952). Analyst, 77, 321.Zlatkis, A., Zak, B., and Boyle, A. J. (1953). J. Lab. clin. Med., 41, 486.

227

on April 6, 2020 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.15.3.221 on 1 May 1962. D

ownloaded from