See related article by Carmichael et al., Cancer Res 1987;47:943–6.

Visit the Cancer Research 75th Anniversary timeline.

In this commentary, we highlight an article published in 1987, a product of collaboration between the National Cancer Institute-Navy Medical Oncology Branch and the Radiobiology Section of the Radiation Oncology Branch of the National Cancer Institute led by John D. Minna and James B. Mitchell (1). This article was selected by the Cancer Research Senior Editors and AACR Fellows as a highlight of the research published by Cancer Research over the past 75 years. Not only is this article highly cited, it has substantially altered the methods by which potential therapeutic anticancer agents are screened for efficacy. The authors sought to determine whether the MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenylte-trazolium bromide] assay and dye exclusion assays could be used to screen for radiation sensitizers or protectors instead of through the more time-consuming clonogenic assay. To fully appreciate the impact of this work, an understanding of the history of clonogenic assays is beneficial.

Although tumors are composed of a variety of cell types that play an important role in tumor maintenance and progression, it is evident that without the contribution of tumor cells driving local invasion and metastasis, cancer would not be a lethal disease. In the early days of surgical oncology and radiotherapy, radical local treatments were attempted to sterilize or remove all tumor cells to prevent tumor recurrence. An appreciation that the reproductive integrity of cancer cells was a critical characteristic for progression of tumors led to the notion that all tumor cells capable of reproduction must be removed or sterilized for therapy to be effective; however, methods for assessing reproductive integrity of tumor cells in vitro were limited.

In 1956, Theodore Puck and Philip Marcus published the first clonogenic assay of mammalian cells, in which HeLa cells were plated at low density and subjected to varying doses of irradiation (2, 3). The goal of Puck's study was to develop an assay that would allow an assessment of the fraction of an animal cell population that remains reproductively competent after a treatment, and thus allow assessment of agents that can alter the ability of a surviving cell to proliferate. This work demonstrated a high efficiency of colony formation from single HeLa cells and provided a unique mechanism to study the effects of drugs and radiation on the ability of tumor cell clones to repopulate after treatment. Indeed, use of the clonogenic assay provided a method to study the importance of a number of agents on tumor cell clonogenic capacity and provided an opportunity for mathematical modeling of response and quantitative comparisons between different exposures. These advancements eventually led to the appreciation of the oxygen effect (4), the importance of radiation quality (linear energy transfer), and the radiation dose rate effect (5), all of which have therapeutic implications in current clinical practice. In addition, the clonogenic assay provided a foundation for the testing of cell survival after exposure to cytotoxic agents in vitro (6).

For over 20 years, the clonogenic assay remained the gold standard for assessment of cancer cell response to radiation and other therapeutics. Despite the success of the clonogenic survival assay in predicting the anticancer efficacy of drugs, the limitations in the time required to complete the assay, and the requirement that tumor cell lines be capable of adherent growth and generation of distinguishable colonies in tissue culture led to interest in suitable alternatives. In the late 1970s and early 1980s, the use of shorter term in vitro assays promised to speed drug development by allowing a more rapid assessment of tumor cell response than could be obtained with clonogenic survival assays. These assays, including the MTT assay, were adaptable to 96-well culture plates, unlike classic clonogenic assays, and held the promise of the capacity for automation.

Although there was tremendous excitement about the possibility of using these more rapid screening assays at the time, the highlighted study cited several possible concerns of this approach. Perhaps most importantly, solid tumor cells that die after radiation exposure tend to undergo a mitotic-linked cell death, a phenomenon that may not occur until after several doublings. At lower radiation doses, the time between exposure and mitotic-linked death may require many additional divisions. The work discusses that during the latent period between radiation exposure and death, tumor cells exhibit intact structure and many aspects of metabolism. Thus, the use of assays that evaluate metabolism or proliferation in this early time frame after radiation exposure may not accurately determine whether cells are destined to die or have maintained reproductive integrity.

The highlighted work described the evaluation of the MTT assay and a dye exclusion assay as an alternative to classical clonogenic assays. Using two cancer cell lines and Chinese hamster V79 cells, multiple variations in experimental conditions were evaluated to determine the effects on the correlation of MTT and dye exclusion to the clonogenic survival assay. The characteristics of the survival curves generated with the three assays after irradiation varied from that obtained with clonogenic assays, depending on a number of factors, including plating density and duration of the assay. Critically, shorter incubation times prior to MTT assay resulted in invalid results, as did increasing confluence. Despite these limitations, at optimal plating density and incubation times, the dye exclusion assay and MTT assay gave comparable results to the clonogenic assay.

Additional considerations for using these short-term assays as surrogates for clonogenic assays described in this work relate to the requirement to maintain the assay for several cell divisions. The plating density of tumor cells used for short-term assays must be optimized to maintain exponential growth during the course of the assay. Thus, controls must often be plated at lower densities to prevent overgrowth at the late time points necessitated by the need to follow cells through multiple divisions. The extended duration of culture in turn requires media changes or a method to account for evaporation, which may impact central wells in a different fashion than those at the edge of the plate. For cells that grow in suspension, removing and replacing media to account for evaporation carries with it the risk of loss of cells.

Perhaps most importantly, this work clearly demonstrated a substantial variation between cell lines in the optimal conditions for assay and clearly demonstrated the complexity of utilizing these assays for screening: insufficient time to MTT assay resulting in insufficient doublings to observe mitotic-linked cell death and insufficient time to exhaust residual dehydrogenase activity in cells destined to die. For the dye exclusion assay, the plating density was found to be critical to the results obtained. Taken together, these results suggested that these assays might be an excellent alternative for screening potential sensitizers and protectors in the setting of a clear understanding of the doubling time for each cell line and a knowledge of the optimal seeding density for each line that allows exponential growth through the duration of the assay.

This study has been widely cited and used as a blueprint for developing screening assays for chemotherapeutics, agents targeting signal transduction, and for testing radiation modifiers (7, 8). Prior work had concluded that these potentially automated assays did not correlate with clonogenic assays for the testing of chemotherapeutics; however, the impact of factors such as plating density and assay time had not been explored or defined. By providing a clear biologic rationale for considering these variables and demonstrating the importance of considering the characteristics of each cell line evaluated in these assays, the work opened a new era for screening new therapeutic agents. Indeed, the use of these assays or similar assays in evaluating new anticancer therapeutic agents is ubiquitous.

Although this work was published prior to the era of agents targeting individual molecules and signal transduction, the conclusions set forth were prescient. The principle that a knowledge of cell line doubling times and optimal plating density in the conditions of the assay becomes even more critical in this current age of drug development. Agents targeting signal transduction pathways or metabolism may alter the rate of proliferation in tumor cells through a variety of causes, including cell-cycle block. These changes in cell cycling should be considered when optimizing MTT and similar assays, especially in the context of combinations of therapies that may require multiple doublings to observe a mitotic-linked cell death. Thus, the caution urged by the authors in 1987 takes on new importance in an era when agents may have substantial effects on cycling or metabolism in a reversible manner, one that does not impact the clonogenic capacity after drug withdrawal.

The dangers of not optimizing the conditions of the assay are that agents with efficacy may be falsely labeled as ineffective in screening studies. In general, reducing the time until assay and plating at too great a density resulted in a decrease in the effects observed from similar treatments compared with clonogenic assays in this study. Likely, this relates to the concerns identified above, such as an insufficient time to manifest mitotic-linked death or an insufficient time to observe a reduction in dehydrogenase activity. With careful consideration of the characteristics of the cells used, agents tested, and assay employed, these errors can be minimized.

Although the use of assays such as the MTT assay has become ubiquitous in the screening of chemotherapeutic agents, it is important to note that the clonogenic assay remains the gold standard in the evaluation of candidate radiation modifiers in vitro (9, 10). Indeed, the senior author of the highlighted work continues to favor use of clonogenic assays in identifying and characterizing radiation sensitizers (11, 12). This work has clearly revolutionized the field of screening for drugs and agents targeting cell signaling with anticancer efficacy. Although the work has been used to justify the widespread use of these assays, the authors' cautions remain just as important today as when they were when first published. It is critical to remember that short-term assays do not directly measure viability. Thus, if reproductive integrity is an important endpoint, confirmatory clonogenic assays should be considered. Furthermore, even if reproductive integrity is not a critical experimental endpoint, confounding factors must always be considered as experimental results are reviewed from these short-term assays.

Interest in developing screening assays for drugs that may provide the capacity for prediction of efficacy in in vivo models, or preferably in patients with cancer, remains an area of active research. An appreciation of the complexity of the tumor microenvironment, matrix interactions, and the importance of cell–cell interactions has led to the development of complex model systems for screening. If effective at truly predicting clinical efficacy, these assays may supplant MTT and similar short-term assays as methods of choice for screening candidate compounds. Nevertheless, the lessons taught by Carmichael and colleagues (1) should remind us to carefully consider and optimize the many variables resident in these screening experiments to enhance our ability to obtain meaningful results.

No potential conflicts of interest were disclosed.

This research was supported by the Intramural Research Program of the NIH, NCI.

1.
Carmichael
J
,
DeGraff
WG
,
Gazdar
AF
,
Minna
JD
,
Mitchell
JB
. 
Evaluation of a tetrazolium-based semiautomated colorimetric assay: assessment of radiosensitivity
.
Cancer Res
1987
;
47
:
943
6
.
2.
Puck
TT
,
Marcus
PI
. 
A rapid method for viable cell titration and clone production with hela cells in tissue culture: the use of X-Irradiated cells to supply conditioning factors
.
Proc Natl Acad Sci U S A
1955
;
41
:
432
7
.
3.
Puck
TT
,
Marcus
PI
,
Cieciura
SJ
. 
Clonal growth of mammalian cells in vitro; growth characteristics of colonies from single HeLa cells with and without a feeder layer
.
J Exp Med
1956
;
103
:
273
83
.
4.
Gray
LH
,
Chase
HB
,
Deschner
EE
,
Hunt
JW
,
Scott
OC
. 
The influence of oxygen and peroxides on the response of mammalian cells and tissues to ionizing radiations
.
Prog Nucl Energy 6 Biol Sci
1959
;
2
:
69
81
.
5.
Hall
EJ
,
Bedford
JS
. 
Dose rate: Its effect on the survival of hela cells irradiated with gamma rays
.
Radiat Res
1964
;
22
:
305
15
.
6.
Roper
PR
,
Drewinko
B
. 
Comparison of in vitro methods to determine drug-induced cell lethality
.
Cancer Res
1976
;
36
:
2182
8
.
7.
Liu
Q
,
Wang
M
,
Kern
AM
,
Khaled
S
,
Han
J
,
Yeap
BY
, et al
Adapting a drug screening platform to discover associations of molecular targeted radiosensitizers with genomic biomarkers
.
Mol Cancer Res
2015
;
13
:
713
20
.
8.
Shoemaker
RH
. 
The NCI60 human tumour cell line anticancer drug screen
.
Nat Rev Cancer
2006
;
6
:
813
23
.
9.
Citrin
DE
,
Mitchell
JB
. 
Altering the response to radiation: sensitizers and protectors
.
Semin Oncol
2014
;
41
:
848
59
.
10.
Kahn
J
,
Tofilon
PJ
,
Camphausen
K
. 
Preclinical models in radiation oncology
.
Radiat Oncol
2012
;
7
:
223
.
11.
Mitchell
JB
,
Choudhuri
R
,
Fabre
K
,
Sowers
AL
,
Citrin
D
,
Zabludoff
SD
, et al
In vitro and in vivo radiation sensitization of human tumor cells by a novel checkpoint kinase inhibitor, AZD7762
.
Clin Cancer Res
2010
;
16
:
2076
84
.
12.
Leiker
AJ
,
DeGraff
W
,
Choudhuri
R
,
Sowers
AL
,
Thetford
A
,
Cook
JA
, et al
Radiation enhancement of head and neck squamous cell carcinoma by the dual PI3K/mTOR inhibitor PF-05212384
.
Clin Cancer Res
2015
;
21
:
2792
801
.