(CANCER RESEARCH 47, 936-942, February 15, 1987]

Evaluation of a Tetrazolium-based Semiautomated Colorimetrie Assay: Assessmentof Chemosensitivity Testing1

James Carmichael,2 William G. DeGraff, Adi F. Gazdar, John D. Minna, and James B. Mitchell

NCI-Navy Medical Oncology Branch [J. C., A. F. G., J. D. MJ and Kadiobiology Section, Radiation Oncology Branch [J. C., W. G. D., J. B. M.J, National CancerInstitute and Naval Hospital, Bethesda, Maryland 20814

ABSTRACT

Drug sensitivity assays were performed using a variation of a colorimetrie [3-{4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide(MTT)1 assay on V79, CHO-AuxBl, CH"C5, NCI-H460, and NCI-

H249 cell lines following optimization of experimental conditions foreach cell line. Results from this assay were compared with data assimilated simultaneously by clonogenic assay and by dye exclusion assay.Good correlation was observed using the CHO-AuxBl cell line and thepleiotropic drug-resistant mutant CH"C5, with similar degrees of relativeresistance observed with both the M l'I and clonogenic assays. Goodcorrelation was observed between the clonogenic and M l'I assays for I-

h drug exposures, although the MIT assay was more sensitive tovinblastine. In general, the clonogenic assay was more sensitive whencontinuous drug exposures were utilized, although this was primarilyrelated to the increased drug exposure time. While the use of the MITassay in drug sensitivity testing of primary tumor samples is limited,since contaminating normal cells may also reduce the tetrazolium, theM I'I assay can be semiautomated, and therefore it offers a valid, simple

method of assessing Chemosensitivity in established cell lines.

INTRODUCTION

Clonogenic assays (1) have been used extensively in theassessment of Chemosensitivity of both established tumor celllines (2) and freshly biopsied tumor tissue (3, 4). Low platingefficiencies, clumping artifacts, and assay duration are technicalproblems that limit the wide spread applicability of clonogenicassays on human tumor material (5-7). In an attempt to overcome the limitations of the clonogenic assay, a number of short-term assays have been developed, including dye exclusion techniques (8), initiated thymidine uptake (9), radiolabeled glucoseutilization (10), and automated image analysis of crystal violet-stained cells (11). However, there are conflicting views regarding the comparability of these assays to the clonogenic assay(12-15).

Recently, changes have been proposed in the National CancerInstitute's anticancer drug screening program. The first involves

the use of human tumor cell lines in a disease-oriented approach(16). Thus, a large panel of human lung cancer lines representing the major histológica! types (including non-small cell andsmall cell lung cancer) will be screened. The second involvesthe use of a colorimetrie assay based on the ability of live butnot dead tumor cells to reduce a tetrazolium-based compound(MTT3) to a blue forma/an product. Chemosensitivity assays

using tetrazolium dyes have been widely reported (17, 18) and

Received 4/25/86; revised 9/18/86; accepted 10/30/86.The costs of publication of this article were defrayed in part by the payment

of page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1The opinions or assertions contained herein are the private views of theauthors and are not to be construed as official or as reflecting the views of theDepartment of the Navy or the Department of Defense.

2To whom requests for reprints should be addressed, at Radiation OncologyBranch, National Cancer Institute, Bldg. 10, Room B3-B69, Bethesda, MD20892.

3The abbreviations used are: MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide (thiazolyl blue); DMSO, dimethyl sulfoxide; inMl. 50%reduction in colony number, absorbance, or tumor/duck RBC ratio; CHO,Chinese hamster ovary.

are in current use in Japan. With modification this assay isattractive in that it can be semiautomated (19), because theassay can be performed in 96-well plates, with MTT formazanproduction analyzed using a scanning multi well spectrotometer.Thus large numbers of samples can be analyzed simply andrapidly. Recently, further modification of this assay has beenreported (20). However, comparison of the MTT assay withother assays is essential prior to its use in large scale screeningstudies. Therefore, direct comparisons were made between theclonogenic, dye exclusion, and MTT assays utilizing both non-small cell and small cell lung cancer cell lines. In additionChinese hamster V79 cells and 2 Chinese hamster ovary celllines, the wild type AuxBl and its pleiotropic drug mutantCHRC5 (21), were also tested. These cell lines were assessed

for sensitivity to a range of cytotoxic drugs and to radiation,with the details of the radiation experiments described in thefollowing article (22). First we discovered the need to significantly modify the previously described MTT assay (19) to allowfor better solubilization of the formazan product. Using thismodification we were then able to show that the MTT assaygives results comparable to those of the clonogenic and dyeexclusion assays for lung cancer and Chinese hamster cells.Based on these studies it appears reasonable to use the MTTassay for new drug screening.

MATERIALS AND METHODS

Cell Lines

Hamster Cell Lines. Chinese hamster lung fíbroblastV79 cells weregrown as a monolayer in Ham's F 12 medium (Gibco) supplemented

with 10% (v/v) fetal calf serum, penicillin, and streptomycin. Exponentially growing cultures were maintained in a humidified atmosphere of5% CO2-9S% air at 37'C, and under these conditions the plating

efficiency was 70-90% and the doubling time was 9-10 h. Two Chinesehamster ovary cell lines kindly provided by Dr. V. Ling were used:CHO-AuxBl and its pleiot topically drug-resistant mutant CHRCS (21).

These two lines were grown as monolayers under the same conditionsas for V79 cells, and plating efficiencies in both cell lines ranged from60 to 75%. Under these conditions doubling times were 13-15 h forCHO-AuxBl and 22 h for CHRC5.

Human Lung Cancer Cell Lines. Two human lung cancer cell lineswere used: NCI-H249, a small cell lung cancer, and NCI-H460, a largecell lung carcinoma cell line. Exponentially growing cultures of bothcell lines were grown in RI'M I 1640 (Gibco) supplemented with 10%

(v/v) fetal calf serum, penicillin, and streptomycin and were maintainedin a humidified atmosphere of 7% CO2-93% air at 37'C. NCI-H460

grew as a monolayer with a doubling time of 20-25 h and platingefficiency of 50-70%, whereas NCI-H249 grew as floating aggregates(23), with a doubling time of 85 h and cloning efficiency in soft agar of

MTT Assay

Single cell suspensions were obtained by mechanical disaggregationof the floating cell line and by trypsinization of monolayer cultures,with cell counts performed using an Elzone particle counter (ParticleData, Inc., Elmhurst, IL). The assay is dependent on the cellularreduction of MTT (Sigma Chemical Co., St. Louis, MO) by the

936

Research. on September 8, 2020. © 1987 American Association for Cancercancerres.aacrjournals.org Downloaded from

MTT ASSAY: CHEMOSENSITIVITY

mitochondria! dehydrogenase of viable cells to a blue formazan productwhich can be measured spectrophotometrically. Following appropriateincubation of cells, with or without drug, 0.1 mg (50 n\ of 2 mg/ml)MTT was added to each well and incubated at 37°Cfor a further 4 h

before processing as described below. To determine the relationship ofcell number to MTT formazan crystal formation increasing cell numbers from 100 to 100,000/well were plated, using V79, NCI-H460, andNCI-H249 cells. MTT was immediately added and plates were incubated for 4 h. Plates were then processed as described and absorbancewas determined. For cell growth studies, serially increasing cell numberswere plated in different columns across 96-well microtiter plates. Actualcell number was measured daily by removing cells from individual wellsat each cell concentration. Cell counts were performed using a paniclecounter, with all measurements carried out in triplicate. Identical plateswere processed daily using MTT, measuring absorbance with a scanningululimeli spectrophotometer (enzyme-linked immunosorbent assayreader; Biotek Instruments, Inc., Burlington, VT).

Growth rates varied between the cell lines used; therefore, the numberof cells plated was adjusted so that untreated cells were in exponentialgrowth phase at the time of harvest. An incubation time of 4 days wasselected for a number of reasons: (a) this allowed sufficient time fordrug-induced cell death to occur; (b) this enabled us to directly compareresults of this assay with data assimilated by dye exclusion assay, whichhas been widely used in this laboratory; and (c) by keeping the incubation time short, this obviated the need to refeed cells, maximizing thebenefit of the automation of the assay. In view of the rapid doublingtime of V79 cells, a 72-h incubation was considered optimal, becauselonger incubation times would have necessitated the seeding of lessthan 100 cells to ensure logarithmic growth at the end of the experiment.

Assays were performed using both 1-h and continuous exposure todrugs. For continuous exposure studies, equal numbers of cells wereinoculated into each well in 0.18 ml of culture medium, to which 0.02ml of lux concentrated drug or phosphate-buffered saline was added.For 1-h exposure, drugs and cells were mixed in 15 x 75-mm test tubes,and following incubation, cells were centrifugad at 250 x g for 5 min,washed twice with phosphate-buffered saline, and then plated directlyinto microtiter plates. Following incubation the media from platescontaining adherent cells was aspirated completely, whereas for thefloating cell line, NCI-H249, plates were centrifuged at 450 x g for 5min in a plate holder. The majority of the media was then aspirated,taking care not to disturb the formazan crystals, leaving approximately30 pi of media in each well. All results represent the average of aminimum of 8 wells. Surviving fraction or fractional absorbance wascalculated by the formula: mean of test sample •+•mean of untreated

sample. Additional controls consisted of media alone with no cells,with or without the various drugs.

Using the procedure described by Mosmann (19) with acid isopropylalcohol as solvent, we were unable to achieve solubilization of theformazan crystals, and minimal absorbance was detected at 570 nm. Amodification of this technique has recently been described (20). However, because it is proposed to use this assay for human tumor cell linescreening, this modification was far from ideal. Many of these cellsrequire serum for optimal growth and a number grow as floatingaggregates, not in monolayer culture. Various solvents were then tested,the best of which proved to be DMSO (Sigma) or mineral oil, althoughthe latter was successful only when all the media could be removedfrom the wells. Absorbance spectra for these solvents revealed absorbance peaks at 503 nm and 565 nm for DMSO and mineral oil,respectively. Because mineral oil is relatively nontoxic, it was used asthe solvent of choice for adherent cultures. Because all the mediumcould not be removed from floating cell cultures, DMSO was used tosolubilize the formazan crystals. The plates containing mineral oil wereallowed to stand overnight at 40°C,prior to reading on an enzyme-

linked immunosorbent assay reader at a wavelength of 570 nm. DMSOplates were placed on a plate shaker for 5 min and read immediately at540 nm.

Dye Exclusion Assay

These experiments were carried out using the procedure describedby Weisenthal et al. (8). All experiments were performed using contin

uous exposure to drugs, with triplicate assays used at all drug concentrations. Cells were collected as described for MTT assays and countedusing a particle counter. Optimal seeding concentrations, to obtainapproximately 5 x IO4cells at the time of harvest, were 500 cells/tubefor V79, 2,500 cells/tube for NCI-H460, and 10,000 cells/tube forNCI-H249. Appropriate cell numbers were then transferred into conicalpolypropylene tubes, with and without drug, and incubated at 37°Cfor

3 days in view of the rapid doubling time in experiments using V79cells, or 4 days for all other cell lines. Following incubation, 50,000aldehyde-fixed duck RBC were added to each tube and the cells wereconcentrated to a volume of 0.2 ml by centrifugation at 250 x g for 5min. The tubes were then vortexed, 0.2 ml of a 1% fast green-0.5%nigrosin solution was added, and tubes were again vortexed. After 10min of incubation, cell suspensions were vortexed and cytocentrifugedat 1,200 x g for 7 min. The cell discs were then fixed in methanol andcounters! ained with hematoxylin and eosin. Cells were counted at X400using a light microscope. The viable (hematoxylin-eosin-stained) tumorcell/duck RBC ratio was determined for each treatment group andcompared to control values.

Clonogenic Assay

Clonogenic assays were performed in standard fashion, using previously described techniques (3,24,25). Drug treatment was administeredeither by 1-h exposure, following which cells were rinsed free of drugusing phosphate-buffered saline prior to plating, or by continuousexposure, with triplicate cultures plated at each drug concentration.Adherent cell lines were plated directly onto IVIri dishes and incubatedwith or without drug for variable times, dependent on the growth rateof the cells. One hundred V79 cells were incubated for control plates,with higher cell numbers used for treated samples to give at least 50colonies and plates incubated for 6 days. Incubation times were 7 daysfor CHO cell lines and 10 days for NCI-H460 cells. For these cell linescontrol plates were inoculated with 250 cells/dish, with higher cellnumbers used for drug-treated samples. Colonies were fixed in metha-nohacetic acid (3:1), stained with crystal violet, and counted using amicroscope at x40. For all clonogenic assays colonies of greater than50 cells were counted, with cell inocula of 5 x IO4cells or less platedon 60-mm l'etri dishes and larger cell inocula plated on 101)nini l'etridishes. NCI-H249, the floating cell line, was suspended in 0.3% agarand plated onto a feeder layer of 0.6% agar, with the medium used inboth agar layers consisting of RPMI 1640 plus 10% (v/v) fetal calfserum. Experiments were performed using 3 ml for each agar layer in60-mm dishes and 8 ml for each agar layer in 100-mm dishes, with allplates incubated for 18 days.

Drugs

All drugs were obtained from commercial sources apart from mel-phalan which was obtained from the Drug Synthesis and ChemistryBranch of the National Cancer Institute. Drugs were prepared every 2months and stored at lOOOx concentrations at —70°C,apart from

melphalan which was prepared freshly for each experiment.

Study Design

All three assays were set up simultaneously for each cell line, usingthe same drug solutions. For 1-h drug exposure, cells were incubatedwith drug in a test tube, then washed free of drug, and plated directlyin both Pet ri dishes and 96-well plates. Survival or fractional absorbancewas expressed as a percentage of control values. For clonogenic studiesthis means that the number of colonies surviving a given treatmentwere compared to untreated control plates. For the MTT assay, absorbance levels from drug-treated cells were corrected against untreatedcontrol absorbance values. For the dye exclusion assay, the ratio ofviable tumor cells to duck RBC in treated specimens was comparedwith the same ratio in control tubes. The IDM was defined as 50%reduction in colony number in the clonogenic assay, 50% reduction ofabsorbance in the MTT assay, or 50% reduction in the tumor/duckRBC ratio using the dye exclusion assay.

Standard errors were calculated according to the method of Beving-ton (26) which takes into account the error for each experimental point

937

Research. on September 8, 2020. © 1987 American Association for Cancercancerres.aacrjournals.org Downloaded from

MTT ASSAY: CHEMOSENSITIVITY

and the error associated with control values. Error bars are shown onthe Figure when larger than the diameter of the symbol.

RESULTS

Modification of the MTT Assay to Give Better Solubili/ut ionof the MTT Formazan Product. The solubility of the MTT-reduced formazan product was poor in acid isopropyl alcoholas used in the method of Mosmann (19), while better absorptioncharacteristics were observed with mineral oil and DMSO assolvents (Fig. 1). From these experiments we chose 570 nm forreading the mineral oil solubilized product. DMSO solubiliza-tion gave an absorption peak (Xmax)at 503 nm. However, Amaxhas been shown to vary considerably for MTT formazan, dependent on the solvent used, and other factors such as theconcentration of metal ions in the solution (27). With floating

09

OJ

0.7

U 06

3 0.5

§ 0.4

03

02

0.1

00

MINERAL OIL ACID ISOPROPANOL

WAVELENGTH (nm)

Fig. 1. Visible absorption spectra of MTT formazan using an adherent cultureof V79 cells with dimethyl sulfoxide, mineral oil, and acid isopropyl alcohol assolvents, using equivalent concentrations of MTT.

cultures, when all the media was not removed, there was a shiftin the Amax,with maximal absorbance detected reproducibly at540 nm. Therefore, this wavelength was used for floating cultures, with DMSO as the solvent.

MTT Assay: Effect of Cell Number on MTT Formazan Production. The amount of formazan product generated and thenmeasured after solubilization in mineral oil or DMSO at 570or 540 nm is proportional to cell number, although absoluteabsorbance for a given cell number varies between cell lines(Fig. 2). These data were generated by plating cells at theinoculum noted on the :t-axis. Cells were then immediatelyincubated with MTT for 4 h, and the plates were then processed.There was good reproducibility between replicate wells withstandard errors «±10%.The growth rates of cell lines inmicrotiter wells were then determined using the MTT assay(Fig. 3). Individual wells were analyzed from each cell line, andcells were counted revealing doubling times of 10 h in V79cells, 26 h in NCI-H460 cells, and 85 h in NCI-H249 cells.These results are comparable to doubling times obtained bystandard methods in these cell lines (data not shown). For eachcell line optimal seeding concentrations were derived, to givewhere possible maximal absorbance while ensuring that cellsremained in exponential growth. The cell number chosen forV79 cells was 100 cells/well for a 3-day assay and for NCI-H460 2,500 and 10,000 cells/well using a 4-day assay. Metabolism of MTT was poor in NCI-H249 cells, with low absorbancevalues. However, because culture conditions such as cell densitycan significantly affect the results of clonogenic assays, we choseto keep seeding cell numbers similar between assays. Therefore,despite low absorbance values, the 10,000-NCI-H249 cells/wellsample was chosen as the optimal inoculum.

Assessment of Drug Resistance in CHO Pleiotropic MutantCells. Chemosensitivity was assessed by continuous exposure

Fig. 2. Absorbance as a function of cellnumber using V79, NCI-H460, and NCI-H249cells. Following plating, MTT was added immediately and cells were incubated for 4 h priorto processing. Absorbance was measured at awavelength of 540 nm using DMSO as solventand 570 nm using mineral oil as solvent.

NCI-H249

100 0 10 50 100

Cell Number/well ( x 103)

Fig. 3. Growth rate estimated by the meas-urement of absorbance following the MTT as-say in 3 cell lines, as a function of the initialcell number plated. Numbers in parentheses,number of cells per well.

l.t

! «

1.4

¡L»

í 1.0D<

.8

.«

.4

V79 NCI-H460 NCI-H249

>0 40 BO SO 10O

Time (hours)

) 80 100 120 140 190 O tt

Time (hours)

1S2 2SS

Time (hours)

938

Research. on September 8, 2020. © 1987 American Association for Cancercancerres.aacrjournals.org Downloaded from

MTT ASSAY: CHEMOSENSITIVITY

Fig. 4. Dose-response curves using ClioABI (AB) and CH"C5 (C5) cells followingcontinuous exposure to vinblastine, mel-phalan, cisplatin, and Adriamycin, using boththe MTT assay and clonogenic assay (Clon).

10J 10' 10'

Cisplatin (nM) Adriamycin (nM)

Table 1 IDX of 2 CHO cell lines, wild type AuxBl and its pleiotropic mutant CH'CS, following continuous exposure to various drugs using the clonogenic assay and

the MTT assay

Adriamycin(nM)Cell

LineAuxBl

(WT)CHO*CS

Resistance factorCA"355,000

140XMTT4510,000

220XVinblastine

(nM)CA12.5

65052xMTT9.5

43545XEtoposide

(nM)CA400

5,00037.5MTT450

10,00022xCisplatin

(nM)CA1,000

1,000IXMTT1,100

2,8002.5XMelphalan

(DM)CA50

4008xMTT28

300llx

* CA, clonogenic assay.

Table 2 IDy, following continuous exposure to various drugs using the clonogenic assay, the dye exclusion assay, and the MTT assay

Adriamycin (nM) Vinblastine (nM) Etoposide (HM) Cisplatin (HM)

CelllineV79

NC1-H460NCI-H249CA«40

1038DEA64

10120MTT60 8100CA9

96.4DEA3

1.512.4MTT2 1.514.4CA300

150250DEA500

2002000MTT500

4002000CA2000

2502100DEA1650

5001500MTT1100

4502750

" CA, clonogenic assay; DEA, dye exclusion assay.

to 5 drugs, doxorubicin (Adriamycin), vinblastine, cisplatin,melphalan, and etoposide, in 2 CHO lines, the wild-typeAuxBl, and its pleiotropic mutant CHRC5. The latter was made

resistant to colchicine and has been shown to exhibit a typicalcross-resistance pattern to many drugs (21). Clonogenic andMTT assays were carried out simultaneously in these cells andthe results are illustrated in Fig. 4, with IDsoS shown in Table1. As can be seen in Fig. 4, a high degree of resistance toAdriamycin and vinblastine is exhibited by the CHRC5 cell line,

with the relative degree of resistance between cell lines similarusing both assays. There is little cross-resistance to cisplatin,and this is exemplified by both assays. Some cross-resistanceto melphalan is exhibited with quantitatively similar resultsobtained with both assays. The clonogenic assay was moresensitive in all experiments, probably relating to the longerduration of drug exposure with this assay. The relative degreeof resistance observed for each drug using the clonogenic technique was maintained using the MTT assay.

Comparison of Assays Using Continuous Drug Exposure. Thethree cell lines (V79, NCI-H460, and NCI-H249) were used tocompare all three assays following continuous exposure to the

single agents Adriamycin, vinblastine, etoposide, and cisplatin.Each drug was tested at a minimum of three concentrations,and the IDsos were determined as shown in Table 2. Foradherent cell lines, excellent correlation was observed for If),,,sfollowing continuous exposure to cisplatin, Adriamycin, andetoposide, although nonclonogenic assays appeared more sensitive to vinblastine.

With cisplatin good correlation was observed for all cell linesin the three assays. Nonclonogenic assays gave higher IDM,sforAdriamycin and etoposide, using the floating cell line NCI-H249, but it should be noted that drug exposure time was 18days for the clonogenic assay and 4 days for nonclonogenicassays. Dose-response curves were generated for all 4 drugsusing the 3 assays with representative examples illustrated inFig. 5, which shows that reasonable correlation is observedbetween the MTT and clonogenic assays. The clonogenic assayappeared to be more sensitive, although allowance should bemade for increased drug exposure time in the clonogenic assays,particularly those using NCI-H249 cells.

Comparison of Assays following 1-h Drug Exposure. Comparisons were made of the MTT assay and clonogenic assay in the

939

Research. on September 8, 2020. © 1987 American Association for Cancercancerres.aacrjournals.org Downloaded from

MTT ASSAY: CHEMOSENSITIVITY

Fig. 5. Dose-responsecurves using threeassaymethodsfollowingcontinuousexposureto drugs.A, responseof V79 cellsto continuousAdriamycinexposure;B, responseof NCI-II4i.il cells to continuouscisplatinexposure;C, responseof NCI-H249 cellsto continuousvinblastineexposure.

V79Dy* Exclusion

20 40 60 eo 100Adriamycin(nM)

NCI-H460

«elusion

MTTClonog.nlc

NCI-H249

Dy* Exclusion

500 1000 1500 2000Cisplatin (nM)

2500 20 40 60Vinblaslin«(nM)

assessment of chemosensitivity following a 1-h drug exposureto Adriamycin, vinblastine, and cisplatin. The first experimentwas performed in V79 cells and the results are illustrated inFig. 6. As can be seen, the assays correlated extremely well forAdriamycin and cisplatin, although the MTT assay was markedly more sensitive to vinblastine. Further studies were thenperformed using Adriamycin against the 2 human lung cancercell lines, showing good correlation between the clonogenic andMTT assays, as shown in Fig. 7. Increased sensitivity to vinblastine in the MTT assay was also seen using NCI-H460 cells

as illustrated in Fig. 7.Potential Limitations of the MTT Assay. The MTT assay is

a growth assay and, therefore, does not distinguish betweencytostatic and cytocidal effects. The effect of cytostatic agentson the MTT assay was assessed with the use of isoleucine-depleted culture medium, which is known to block cells in cellcycle, causing a d arrest (28, 29). V79 cells were incubated inisoleucine-free medium for IS h, following which isoleucinewas added and cells were allowed to proliferate for a further 48h. The MTT assay was then performed and the results werecompared with those for control cells which had been grown incomplete medium throughout. Significant impairment ofgrowth was observed in isoleucine-depleted cells. In contrast,15 h of isoleucine depletion did not affect the results of aclonogenic assay.

The effect of strong reducing agents on the interpretation ofthe MTT assay was also studied. Cysteamine (0.1, 1.0, and 10HIM)was added to individual wells together with MTT. Rapidreduction of MTT was observed in the solution, although nocrystals were formed. Significant absorption was observed at540 nm, although following aspiration and resolubilization inDMSO, no absorption was detectable at this wavelength. Whena similar experiment was performed using V79 cells, cysteaminereduction did not interfere with the cellular reduction of MTT,and following resolubilization absorbance values between V79

controls and cysteamine-treated cells were similar.Many compounds, such as Adriamycin, are known to exhibit

absorption peaks in the 500-600-nm range, although aspirationof drug-containing medium from the wells resulted in no effecton absorbance readings. However, these factors can cause problems with floating cell lines, because all the medium is notremoved prior to solubilization of the dye. Alteration of theoxidation-reduction balance of the cell by preincubation withagents that deplete intracellular glutathione levels, such as A-ethylmaleimide, diamide, and buthionine sulfoximine, did notaffect the cellular reduction of MTT.

DISCUSSION

This study confirms, as shown in Fig. 2, that the MTT assayshows good correlation between spectrophotometric absorbanceand cell number as described previously (19). Using thismethod, estimated doubling times for all three cell lines werequantitatively similar to those obtained using standard techniques. However, using the published method for the MTTassay (19), considerable difficulty was encountered in resol ubi 1-izing the formazan crystals. An absorbance spectrum of tormazan solubili/ed by acid isopropyl alcohol showed a wideindistinct peak, which is greatly inferior to the absorbance peaksobserved using mineral oil or DMSO as solvents. Greatersolubility and absorbance readings were seen using the latter 2solvents. Optimal results were achieved by measuring formazancrystals solubili/ed in mineral oil at a wavelength of 570 nmand DMSO-solubilized crystals at 540 nm.

It should be stressed that with the MTT assay, optimalconditions should be elucidated for each cell line, in terms ofboth cell number plated and assay duration. It is essential thatsufficient time is allowed for cell death and loss of dehydrogen-ase activity to occur and that control cells are in exponentialgrowth at the time the assay is processed. Because the IDâ„¢is

10

Fig. 6. Dose-responsecurves using V79cellsfollowinga 1-h exposureO.M) to Adria- Qmycin (left), vinblastine(middle), or cisplatin -^3 •»_*(rigHl).Clonogenicassaycurvesand MTT as- U "trsaycurvesare depicted. fjj "u.;"

fe

I

. 1O •

\-—<J>••—Clonogenic

^Clonogenic

-Ì

O 1 2 3 4 5 6 O S 10 15 20 25 30 O 20 40 60 80 100 120

940

Research. on September 8, 2020. © 1987 American Association for Cancercancerres.aacrjournals.org Downloaded from

Fig. 7. Dose-response curves using NCI-H460 and NCI-H249 cells following a 1-hexposure to vinblastine or Adriamycin, usingboth the MTT assay and clonogenic assay(Clon).

MTT ASSAY: CHEMOSENSITIVITY

NCI-H460 NCI-H480 HCI-H24»

2 10°10''IO"'10"

1C'—

+ _f,f_f_t_J=f^_t-410Clonrl

MTTIx:-•

,' -•IQ-0

101 IO2 10* 104—I10.J

Vinblastine (nM)

10' 10s IO1 10* 10'

Adriamycin(nM)

io2 io1Adriamycin (nM)

io4

defined as 50% reduction in absorbance compared to controlvalues, if control cells are allowed to plateau, but treated cellscontinue to grow, the ID50 for that drug could be seriouslyoverestimated. Likewise, with treated cells time should be allowed for cell death and loss of dehydrogenase activity; otherwise overestimation of ID50s may occur. Similar conditionsapply to the dye exclusion assay, where optimal seeding concentrations and assay duration are important factors in obtaining satisfactory results.

Using the MTT and dye exclusion assays, good correlationwith the clonogenic assay was achieved in the chemosensitivitytesting of a range of cytotoxic drugs. Differences in relativechemosensitivity between cell lines were maintained. For continuous drug exposure both nonclonogenic assays exhibitedsimilar dose-response curves, whereas clonogenic assays appeared more sensitive. However, stronger correlation betweenassays was observed using 1-h drug exposures, although differences are noted for certain drugs as exemplified by the vinblastine data. This drug has cytostatic in addition to cytocidalproperties, and as a result more activity is observed with theMTT assay. It should be stressed that this assay does notdifferentiate between cytostatic and cytocidal effects on cells.As has been reported previously (20), when medium containingserum or phenol red is used in this assay, high backgroundvalues are observed. However, maximal absorbance for manyhuman tumor cell lines is approximately 1.0 absorbance unit,limiting the range of detectability of this assay. Therefore, theMTT assay is limited to the detection of 1 log of cell kill, as isconfirmed by Fig. 2. Likewise, using the dye exclusion assay,when significant degrees of cell killing are observed, the numberof viable tumor cells is low, with reading errors correspondinglyhigh. These results agree with previously published data, claiming good correlation between dye exclusion assays and clonogenic assays (14, 15). This particular dye exclusion assay hasthe advantage over other short-term assays, in that tumor cellscan be distinguished from nontumorous cells by differentialstaining techniques, which is beneficial in the testing of primarytumor samples. The dye exclusion assay, however, is highlylabor intensive, particularly at the time of reading. The MTTassay, in contrast, offers an excellent opportunity for the rapidtesting of large numbers of drugs with good reproducibility,particularly in adherent cell lines, where the standard deviationof absorbance readings approximates ±5%.However, the assayis less optimal for the testing of floating cell lines. Standarddeviations of ±15%were observed in absorbance readings offloating cell cultures. The increased variability in absorbancewith floating cultures may be caused by variation in residualvolume following aspiration of medium, prior to solubilizationof the MTT formazan. This problem could be resolved by theproduction of a tetrazolium salt, which on reduction forms awater-soluble formazan dye, allowing microtiter plates to beread directly.

Agents which alter the oxidation-reduction balance of thecell by glutathione depletion did not affect the results of theassay. Likewise, strong reducing agents cause reduction of MTTin solution but do not affect the interpretation of the assaywhen all the supernatant is removed prior to solubilization ofthe MTT formazan crystals. For these reasons difficulties remain in the handling of floating cell cultures using this technique.

In these studies cell lines with vastly different growth characteristics were used. Doubling times varied from 10 to 85 h,and although the majority of cells grew as monolayers, thesmall cell lung cancer line grew as floating aggregates. Forchemosensitivity assays utilizing nonclonogenic techniques,there are many potential factors that may interfere with theinterpretation of any results. Theoretically, for cell lines ofdiffering growth rates, the relative contributions of in situ celldeath, cell cycle delay, and inhibition of proliferation could bedifferent and could explain observed differences in chemosensitivity seen in the continuous drug exposure experiments.However, the experiments utilizing 1-h incubation times showsthese factors to be relatively unimportant, with drug exposuretime being the most influential variable. Optimal assay durationfor human lines should be a minimum of 4 days to allow forcell death and loss of dehydrogenase activity with a maximumof 7 days to obviate the necessity of refeeding of cultures.

It is obviously appreciated that drug exposure, as measuredby area under the curve calculations, varied between assayswhen continuous drug exposure was used. The claim is notmade that similar drug concentrations are effective in the MTTassay but that similar dose-response curves are observed. Likewise, differences in sensitivity between cell lines are detectableand are preserved to a similar degree. Accepting the limitationsdescribed, the semiautomated MTT assay offers a valid, rapid,and simple method to assess chemosensitivity in cell lines.However, the value of this assay in the testing of primary tumorsamples may be limited because normal cell contaminants mayalso reduce the dye.

ACKNOWLEDGMENTS

The authors wish to thank Dr. Angelo Russo for his helpful technicalsuggestions and Janita Coen for typing the manuscript.

REFERENCES

1. Puck, T. T., and Marcus, P. I. A rapid method of viable cell titration andcolony production with HeLa cells in tissue culture: the use of x-irradiatedcells to supply conditioning factors. Proc. Nati. Acad. Sci. USA, 41: 432-437, 1955.

2. Hill, It. T. Use of continuous human tumor cell lines to evaluate drugs byclonogenic assays. In: P. P. Dendy and B. T. Hill (eds.), Human Tumor DrugSensitivity Testing in vitro, pp. 129-146. New York: Academic Press, 1983.

3. Hamburger, A. W., and Salmon, S. E. Primary bioassay of human tumorstem cells. Science (Wash. DC), 197:461-463, 1977.

4. Courtenay. V. D., and Mills, J. An in vitro colony assay for human tumors

941

Research. on September 8, 2020. © 1987 American Association for Cancercancerres.aacrjournals.org Downloaded from

MTT ASSAY: CHEMOSENSITIVITY

grown in immune-suppressed mice and treated in vivo with cytotoxic agents.Br. J. Cancer, 37:261-268,1978.

5. Bertoncello, I., Bradley, T. R., Campbell, J. J., et al. Limitations of the clonalagar assay for the assessment of primary human ovarian tumor biopsies. Br.J. Cancer, 45: 803-811,1982.

6. Alley, M. ( '.. and Lieber, M. M. Improved optical detection of colony

enlargement and drug cytotoxicity in primary soft agar cultures of humansolid tumor cells. Br. J. Cancer, 49: 225-233, 1984.

7. Selby, P., Buick, R. N., and Tannock, I. A critical appraisal of the "humantumor stem cell assay." N. Engl. J. Med., JO«:129-134,1983.

8. Weisenthal, L. M., Marsden, J. A., Dill, P. I... and Macaluso, C. k. A noveldye exclusion method for testing in vitro chemosensitivity of human tumors.Cancer Res., 43:749-757, 1983.

9. Twentyman, P. R., Walls, G. A., and Wright, K. A. The response of tumorcells to radiation and cytotoxic drugs—a comparison of clonogenic andisotope uptake assays. Br. J. Cancer, 50:625-631, 1984.

10. Von Hoff, D. D., Forseth, B., and Warfei, L. E. Use of a radiometrie systemto screen for antineoplastic agents: correlation with a human tumor cloningsystem. Cancer Res., «.-4032-4038, 1985.

11. Fräser,L. B., Spitzer, G., Ajani, J. A., Brock, W. A., Lukeman, J., l'ai hak,S., Tomasovic, B., Thielvoldt, D., Williams, M., Vines, A., and Tofilon, P.Drug and radiation sensitivity measurements of successful primary mono-layer culturing of human tumor cells using cell-adhesive matrix and supplemented medium. Cancer Res., 46:1263-1274, 1986.

12. Roper, P. R., and Drewinko, B. Comparison of «ivitro methods to determinedrug-induced lethality. Cancer Res., 36: 2182-2188, 1976.

13. Twentyman, P. R. Experimental chemotherapy studies: intercomparison ofassays. Br. J. Cancer, 41:279-287, 1980.

14. Weisenthal, L. M., Dill, P. L., Kurnick, N. B., et tu. Comparison of dyeexclusion assays with a clonogenic assay in the determination of drug inducedcytotoxicity. Cancer Res., 43: 258-264, 1983.

15. Bosanquet, A. G., Bird, M. C., and Gilby, E. D. In vitro determination oftumor chemosensitivity in haematological malignancies. Haematol. Oncol.,3:1-10,1985.

16. The Cancer Letter Vol. 10, No. 41, 1984.17. Black, M. M., and Speer, F. D. Further observations on the effects of cancer

chemotherapeutic agents in the in vitro dehydrogenase activity of cancertissue. J. Nati. Cancer Inst., 14:1147-1158,1954.

18. Kundo, T., and Ohkubo, K. In vitro test for prediction of side effects ofcarcinostatic agents, dann, 58:349-354, 1967.

19. Mosmann, T. Rapid colorimetrie assay for cellular growth and survival:application to proliferation and cytotoxicity assays. J. Inumino!. Methods,65: 55-63, 1983.

20. Denizot, F., and Lang, R. Rapid colorimetrie assay for cell growth andsurvival. Modifications to this tetrazolium dye procedure giving improvedsensitivity and reliability. J. Immunol. Methods, 89: 271-277, 1986.

21. Ling, V., Kartner, N., Sudo, T., Sim ¡mnitch, L., and Riordan, J. R. Multi-drug-resistance phenotype in Chinese hamster ovary cells. Cancer Treat.Rep., 67:869-874,1983.

22. Carmichael, J., DeGraff, W. B., Gazdar, A. F., Minna, J. D., and Mitchell,J. B. Evaluation of a tetrazolium-based semiautomatic colorimetrie assay:assessment of radiosensitivity. Cancer Res., 47:943-946, 1987.

23. Carney, D. N., Gazdar, A. F., Bepler, G., Guccion, J. G., Marangos, P. J.,Moody, T. W., Zweig, M. H., and Minna, J. D. Establishment and identification of small cell lung cancer cell lines having classic and variant features.Cancer Res., 45:2913-2923, 1985.

24. Puck, T. T., Morkovin, D., Marcus, P. I., and Cieciura, S. J. Action of X-rays on mammalian cells. II. Survival curves of cells from normal humantissues. J. Exp. Med., ¡06:485-500,1957.

25. Von Hoff, D. D., Clark, G. M., Stogdill, B. J., Sarodsy, M. F., O'Brien, M.

T., Casper, J. T., Mattox, D. E., Page, C. P., Cruz, A. B., and Sandbach, J.F. Prospective clinical trial of a human tumor-cloning system. Cancer Res.,43:1926-1931, 1983.

26. Bevington, P. R. Data Reduction and Error Analysis for the Physical Sciences, 71 pp. New York: McGraw-Hill Book Co., 1969.

27. Alunan, F. P. Tetrazolium salts and forma/ans. IK Progress in Histochem-¡stryand Cytochemistry, p. 29. Stuttgart, Germany: Gustav Fischer Verlag,1976.

28. Tobey, R. A., and Key, K. D. Isoleucine-mediated regulation of genomeregulation in various mammalian cell lines. Cancer Res., 31:46-51,1971.

29. Tobey, R. A. Production and characterization of mammalian cells reversiblyarrested in d by growth in isoleucine-deficient medium. Methods Cell Biol.,(5:67-112, 1973.

942

Research. on September 8, 2020. © 1987 American Association for Cancercancerres.aacrjournals.org Downloaded from

1987;47:936-942. Cancer Res   James Carmichael, William G. DeGraff, Adi F. Gazdar, et al.   Assay: Assessment of Chemosensitivity TestingEvaluation of a Tetrazolium-based Semiautomated Colorimetric

  Updated version

  http://cancerres.aacrjournals.org/content/47/4/936

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/47/4/936To request permission to re-use all or part of this article, use this link

Research. on September 8, 2020. © 1987 American Association for Cancercancerres.aacrjournals.org Downloaded from