[CANCER RESEARCH 35.1804 1808, July 1975)
The Binding of Carcinoembryonic
Fragments1
Antigen by Antibody and Its
James E. Morris,2 Marianne L. Egan, and Charles W. Todd
Department
of Immunology,
City of Hope National
Medical
Center, Duarte, California
SUMMARY
In order to assess the potency of antigenic fragments of
carcinoembryonic antigen (CEA) in the radioimmune assay,
it is necessary to know whether the high affinity of goat
anti-CEA antibody (which makes possible the detection of
as little as I0~" M CEA) is due to bivalent binding of the
CEA molecule. Immunoglobulin G and the F(ab')2 and
Fab' fragments derived from it were prepared from an
anti-CEA serum and tested for their ability to bind CEA.
Equivalent concentrations of binding sites of the bivalent
F(ab')2 and univalent Fab' fragments of anti-CEA were
identical to the immunoglobulin G fraction in the standard
inhibition curve. Fragments of CEA obtained by trypsin
digestion produced equivalent inhibition curves when tested
with either immunoglobulin G, F(ab')2, or Fab'. Thus,
increased avidity due to bivalent binding to a single antigen
molecule cannot be invoked to explain the sensitivity
observed in the CEA assay. This high sensitivity implicates
the protein rather than the carbohydrate as an important
part of the antigenic determinant(s) of CEA.
INTRODUCTION
CEA3 is a tumor-associated antigen originally described
by Gold and Freedman (10, 11). An initial investigation of
the distribution of CEA indicated it was found in the sera of
patients with adenocarcinoma of the digestive system and in
fetal gut tissue ( 19). In subsequent studies (see Ref. 21 for a
review), material that is either CEA or a cross-reacting
material was found in the sera of normal individuals and in
patients with a variety of other malignant and nonmalignant
diseases. In patients with nonmalignant diseases, the levels
detected are usually lower than those encountered in
patients with digestive cancer.
CEA is a glycoprotein with a molecular weight of 180,000
(17) which contains 45 to 57% carbohydrate and 35 to 46%
protein by weight (18). Structural studies have been under-
91010
taken to determine whether the antigenic determinants on
the tumor-associated molecules can be distinguished from
those found in other diseased and normal states. If the
determinants can be distinguished, it should be possible to
make tumor-specific antisera and thus greatly improve the
diagnostic value of the radioimmune assay for CEA. At the
present time, definitive proof concerning the nature of the
antigenic determinants on CEA is lacking.
Studies by Banjo et al.(\) have indicated that at least one
of the determinants is contained in the linkage of asparagine
to glucosamine. In these studies it was necessary to use very
large quantities of the antigenic fragments to obtain inhibi
tion in the radioimmune assay. We also have found that
antigenic fragments prepared by trypsin digestion of neuraminidase-treated, reduced, and alkylated CEA are less
effective inhibitors on a weight basis than the intact
molecule, although the decreased efficiency was not as large
as that observed by Banjo et al.
The finding that fragments were less effective inhibitors
than the intact molecule prompted us to consider the
possibility that bivalent binding of antibody to a single
molecule might be occurring. Such bivalent binding would
increase the sensitivity of the assay (functional affinity) with
respect to intact molecules over that which would obtain for
monovalent antigen fragments (intrinsic affinity) (13). To
determine whether bivalent binding of the antigen plays an
important role in the CEA radioimmune assay, IgG,4 as
well as F(ab')2 and Fab' fragments derived from it, were
prepared from an anti-CEA serum and tested for their
ability to bind CEA.
MATERIALS
AND METHODS
Preparation of CEA. CEA was obtained from liver
métastasesof colon adenocarcinomas, by the method
described by Coligan et al. (3). A radioimmune assay using
well-characterized CEA as a standard was used to follow
the purification (6, 8).
Preparation of Anti-CEA. CEA (10 /¿g)was dissolved in
1Supported in part by Grant IC-44 from the American Cancer Society
and Grant ÇA1263I from the National Cancer Institute, by Contract NOI 0.1 ml normal serum from goat Ace, mixed with 0.9 ml of
CB23877 with the National Cancer Institute, and by the Stuart L. Bernath 0.85% NaCl solution, and emulsified with an equal quantity
Cancer Research Fund.
of Freund's complete adjuvant.4 Three 10-/ig injections were
2 Present address: Biology Department. Battelle Pacific Northwest
given at monthly intervals, followed by a 1-jjg boost at 4.5
Laboratory. Richland, Wash. 99352.
3The abbreviations used are: CEA, carcinoembryonic antigen; IgG. months. The serum used in the present study (Ace 13) was
immunoglobulin G: F(ab')2, bivalent immunoglobulin obtained by pepsin
digestion; Fab', monovalent immunoglobulin obtained by reduced and
alkylated F(ab')2.
Received December 9. 1974; accepted April 3. 1975.
1804
4The nomenclature used for goat immunoglobulins and their fragments
follows that recommended by a committee of the World Health Organiza
tion (20).
CANCER
RESEARCH
VOL. 35
Downloaded from cancerres.aacrjournals.org on June 17, 2017. © 1975 American Association for Cancer Research.
Binding of CEA
obtained
5 months after the initial immunization.
Preparation of Horse Anti-Goat IgG. Horse anti-goat
IgG was prepared as previously described (8).
Purification of IgG. IgG was prepared from an anti-CEA
serum by the method of Levy and Sober (14).
Pepsin Digestion. The purified goat IgG was treated with
pepsin (2 times crystallized; Worthington Biochemical
Corp., Freehold, N. J.) essentially as described by Nisonoff
and Rivers (15). Separation of the undigested IgG, F(ab')2,
and other digestion products was accomplished on a 3- x
182-cm Bio-Gel P-200 column (Bio-Rad Laboratories,
Richmond, Calif.).
Reduction of F(ab')2 Fragment. F(ab')2 was dialyzed
against 0.2 M Tris-HCl, pH 8.0 (Trisma; Sigma Chemical
Co., St. Louis, Mo.). The sample was incubated 1.5 hr at
room temperature in the presence of 2 mM dithiothreitol
(Calbiochem, La Jolla, Calif.) under an atmosphere of N2.
An equimolar amount of iodoacetamide was added, and
alkylation was allowed to proceed for 2 hr at 4°in the dark.
The reduced and alkylated protein was dialyzed against
Tns-NaCl buffer, pH 8.0, and applied to a Bio-Gel P-200
column (3 x 182 cm) to separate the unreduced F(ab')2
from the reduced and alkylated Fab'.
Protein Concentration. Samples of IgG, F(ab')2, and Fab'
were exhaustively dialyzed against 0.1 M ammonium bicar
bonate, lyophilized, and dried to constant weight. Each
preparation (10 to 15 mg) was brought up to 10 ml, and the
absorbance was read at 280 nm to determine the extinction
coefficient.
Analytical Ultracentrifugation. Ultracentrifugation was
performed in 0.0175 MNaCl:0.0175 Mphosphate, pH 7.2, at
20°in a Spinco Model E centrifuge at 59,790 rpm.
Immunoelectrophoresis. Immunoelectrophoresis was
performed by the method described by Scheidegger (16).
Radioiodinations. CEA samples were labeled with 126I,
and goat IgG with 131I,using the chloramine T procedure
(8). Purified goat IgG, F(ab')2, and Fab' were labeled with
125I,by the same procedure.
Precipitation of '"I-labeled Goat IgG, 125I-labeled
llab )„and ' l-lahek-d Fab' by Horse Anti-Goat IgG.
Increasing quantities of the radioiodinated IgG or its
fragments were added to a constant amount (50 /il of 1:40)
of normal goat serum and checked for 100% precipitation
by horse anti-goat IgG after incubation for 1 hr at 37°and
for 15 min at -4°.
Radioimmune Assay for Anti-CEA Activity. The tripleisotope double-antibody assay for CEA, as described by
Egan et al. (6, 8), was used to determine the binding of
'"•-labeledCEA by IgG F(ab')2, and Fab'. An equal molar
concentration of binding sites of each antibody component,
IgG, F(ab')2, or Fab', was added to a series of tubes
containing increasing amounts of CEA (0.5 to 262 ng) and a
constant amount (0.5 ng) of 125I-labeled CEA. To achieve
an equal molar concentration of binding sites, the molar
concentration of univalent Fab' added was twice that of the
bivalent IgG and F(ab')2. Normal goat IgG was added in an
appropriate quantity (50 p\ of 1:40 normal goat serum) to
achieve equivalence with the horse anti-goat IgG used to
precipitate the antigen-antibody complexes. A mixture of
0.075 M NaCl:0.075 M phosphate, pH 7.2, and 1 mg gelatin
JULY
per ml was added to bring the volume of each tube to 260 ^1.
After an overnight incubation at 37°,the normal goat IgG
and anti-CEA IgG or its fragments were precipitated with
horse anti-goat IgG for 1 hr at 37°and for 15 min at -4°.
Normal goat IgG labeled with 131Iwas included in trace
amounts (1 fig) to check for complete precipitation of the
goat IgG by horse anti-goat IgG. 22Na served as a volume
marker for residual supernatant (12) and was used to
calculate the percentage radioactivity in the supernatant and
precipitate.
CEA Fragments. CEA (10.02 mg) was incubated with 50
units (1 unit = 0.625 ¿ig)of neuraminidase (Vibrio cholera;
General Biochemicals, Chagrin Falls, Ohio) in 0.05 M
acetate buffer containing 0.1% CaCl2 for 24 hr at 37°.The
solution was dialyzed for 2 days versus deionized water and
then versus \ M Tris, pH 8.1, and 6 mM EDTA for 1 day.
After reduction with 10 mMdithiothreitol in a 9 Murea:0. l M
Tris buffer, pH 8.1, containing 6 mM EDTA for 2 hr at 37°,
the sample was alkylated with 50 /¿Ci(8.55 /IM) iodo[l14C]acetamide (0.031 mCi/mg) for 30 min at 4°,and for 1
more hr after sufficient unlabeled iodocetamide was added
to bring the final concentration to 20 mM. The reduced and
alkylated CEA was separated from the other reactants on a
Bio-Gel P2 column equilibrated with 50 mM NH4HCO3.
The reduced and alkylated CEA, 2.3 mg, was incubated in
deionized water in a pH-stat with three 50-/ig aliquots of
trypsin [which had been treated with L-(l-tosylamido2-phenyl)ethylchloromethyl ketone; Worthington Biochem
ical Corp.] until there was no additional uptake of base.
After lyophilization, the trypsin-digested material was dis
solved in 10 mM acetic acid and chromatographed on a
Sephadex G-50 column, 117 x 1.5 cm, which was equili
brated with 10 mM acetic acid.
Chemicals and Radiochemicals. Unless specifically stated,
all chemicals were Baker Analyzed (Baker Chemical Co.,
Phillipsburg, N.J.). Iodo[l-14C]acetamide (0.031 mCi/mg),
126I(>350 mCi/ml), 131I(500 to 750 mCi/ml), and 22Na
(carrier free) were purchased from New England Nuclear,
Boston, Mass.
RESULTS
Preparation of IgG, F(ab')2, and Fab'. All preparations,
goat IgG, F(ab')2, and Fab', gave a single line in immunoelectrophoresis
with horse anti-goat IgG and rabbit antiwhole goat serum. All samples moved as a single peak in the
ultracentrifuge
with the following s20 w values: IgG, 6.6 S;
F(ab')2, 4.5 S; and Fab', 3.2 S.
Adjustment of Equal Concentrations of Antibody Sites.
The extinction coefficients at 280 nm (EJ'j.m) determined
as described in "Materials and Methods" were 13.2, 13.6,
and 13.3, for IgG, F(ab')2, and Fab', respectively. These
extinction coefficients were used to adjust each of the
solutions to an equal concentration
of antibody-binding
sites in preparation
for the radioimmune
assay. It was
established by radioimmune assay that the horse anti-goat
IgG precipitated the IgG, F(ab')2, and Fab' with equal
efficiency.
Determination of Relative Binding of CEA by IgG,
1975
Downloaded from cancerres.aacrjournals.org on June 17, 2017. © 1975 American Association for Cancer Research.
1805
J. E. Morris et al.
F(ab')2, and Fab'. Since the antigenic valence of CEA is
unknown, it is impossible to determine a precise dissociation
constant for the anti-CEA preparations. However, by
comparing the amount of CEA that is required to inhibit
50% of the binding of a constant amount of 125I-labeled
CEA by equivalent molar quantities of antibody binding
sites, it is possible to determine relative binding efficiencies
for the antibody preparations. Chart 1 illustrates the curves
obtained when increasing amounts of CEA are added to 0.5
ng '"¡-labeled CEA and 1.01 pmoles IgG, 1.02 pmoles
F(ab')2, or 2.01 pmoles Fab'. The concentration of CEA
required to cause 50% inhibition of binding is within the
same order of magnitude for all 3 preparations; 1.0-10~10 to
1.5-IO-10 mole/liter (Table 1).
Inhibition by Trypsin CEA Fragments. Three peaks of
l4C-labeled trypsin-digested material were obtained on a
Sephadex G-50 column (Chart 2). All 3 fractions eluted
after 125I-labeled CEA. Only the higher-molecular-weight
peak contained significant CEA activity.
The fragments of CEA produced by trypsin digestion of
neuraminidase-treated, -reduced and -alkylated CEA were
10 times less effective than the intact molecule on a weight
basis in inhibiting the reaction of 125I-labeled CEA and
anti-CEA (Chart 3). Although the fragments were less
effective as inhibitors, the trypsin digestion destroyed very
few, if any, antigenic determinants, since the fragments
inhibited 85% of the reaction between anti-CEA and intact
125I-labeled CEA. Chart 3 also illustrates that there was no
/
•¿
10
ng CEA
Chart I. Inhibition curves obtained with increasing quantities of cold
CEA added to tubes containing 12SI-labeled CEA (0.5 ng), '"[-labeled goat
IgG (1 pg). 50 ¡Aof 1:40 normal goat serum containing 2 p.Mcombining
sites of goal IgG anti-CEA (O). F(ab')2 (•),or Fab' (A). The incubation
and precipitation were performed as described in "Materials
and Meth
ods."
1806
Table 1
Amount of CEA required to inhibit 50% of the binding of I2sl-labeled
CEA by anti-CEA and its fragments
12M-labeled CEA (0.5 ng) was incubated with increasing quantities of
CEA (0.5 to 262 ng) and equimolar binding sites of IgG, F(ab')2, and Fab'.
The radioimmune
assay was completed as described in "Materials
and
Methods."
Antibody
Amount required to inhibit 50%
CEA/260
^1°4.85.46.8Mole/liter1.0
101.2
x 1010-'°1.5
x
x 10-'°
IgGF(ab')2Fab'ng
" Volume
of assay.
difference in the amount of fragments needed for 50%
inhibition, whether IgG, F(ab')2, or Fab' was used as the
antibody.
DISCUSSION
The increased avidity resulting from bivalent binding of 2
determinants on 1 antigen molecule has been described in a
mathematical model by Crothers and Metzger (5). If 2
antigenic binding sites are sufficiently close to one another,
each site may be occupied by 1 of the 2 combining sites on a
single antibody molecule. Although 1 liaison may dissoci
ate, the antibody-combining site and the antigenic determi
nant will remain in close proximity as long as the 2nd
liaison holds. This will greatly enhance the probability of
reestablishing the ruptured liaison. The net effect may lead
to an apparent increase in the binding affinity of several
orders of magnitude. Hornick and Karush ( 13) have termed
the resultant binding affinity "functional affinity" to distin
guish it from the intrinsic affinity of each individual
antibody-combining site.
In studies on the isolation of the antigenic fragments of
CEA, it is necessary to know whether bivalent binding of the
anti-CEA to a single CEA molecule occurs. Univalent
antigen fragments would appear to be inefficient inhibitors,
compared with the intact molecule, in a system in which
bivalent binding contributes significantly to the dissociation
constant. Thus, it would be possible to be easily misled into
thinking an isolated fragment was not an important anti
genic determinant if it did not inhibit with the same
efficiency as the intact molecule.
Since the valence of CEA is not known at the present
time, it is impossible to determine a precise dissociation
constant for the CEA anti-CEA system. However, calcula
tion of the concentration of free antigen molecules at 50%
inhibition with intact IgG, F(ab')2, or Fab' makes it possible
to compare the relative dissociation constants for these
molecules. The relative dissociation constants were found to
be within 50% of one another, indicating that there is no
difference in the functional and intrinsic dissociation con
stants exhibited by the anti-CEA in the radioimmune assay.
These results would preclude the possibility that the
reduction in the capacity of the fragments to inhibit could be
CANCER
RESEARCH
VOL. 35
Downloaded from cancerres.aacrjournals.org on June 17, 2017. © 1975 American Association for Cancer Research.
Binding of CEA
Churl 2. Chromatography
of CEA fragments.
The CEA that had been alkylated with iodo[l"Cjacetamide
and digested with trypsin was ap
plied to a 117- x 1.5-cm Sephadex G-50 column and
eluted with 10 mvi acetic acid. •¿,
14C cpm/100^1;
1200
BOO
A, CEA activity. The void volume was measured
with '"l-labeled CEA. and the total elution volume
was measured with 22Na.
50
60
FRACTION
90
80
70-
60-
so2010-
Log ng Inhibitor
Chart 3. Inhibition curves obtained
to 920 /ig) of CEA fragments obtained
dase-treated.
reduced, and alkylated
'"I-labeled
CEA. l ¿iggoat '"I-labeled
serum containing 2 pvi combining sites
by adding increasing quantities (46
by trypsin digestion of neuraminiCEA. Tubes contained 0.5 ng
IgG. and 50 p I 1:40 normal goat
of IgG (O). F(ab')2 (•).and Fab'
(A).
attributed to univalency of the antigen. An alternate
explanation would be that only a small proportion of the
fragments are in the proper conformation at any given time
to fit completely into the antigen-combining site. If lack of
proper conformation is the correct explanation for the
decreased efficiency of fragments to act as inhibitors in the
assay, this strongly implicates the protein in the major
antigenic determinant(s) of CEA.
These experiments make it possible to rule out bivalent
binding as a factor contributing to the high dissociation
constant of those antibodies playing the major role in the
radioimmune assay. Populations of antibodies with lower
affinities and directed toward other determinants are proba
bly also present but contribute relatively little. A simple
approximation of the dissociation constant of the highest
JULY
70
BO
NUMBER
affinity antibody measurable in the system on the basis of
the assumption that CEA is a monovalent antigen, can be
made. We routinely obtain antisera in goats that detect
fmoles of CEA in 0.2-ml samples by radioimmune assay (6,
8). This indicates a dissociation constant for the highest
affinity antibody present of I0~n mole/liter assuming
monovalency of the CEA. The dissociation constant will fall
by only 1 order of magnitude for each 10-fold increase in the
valency of CEA; thus the dissociation constant will still be
high, even if CEA is not functionally monovalent in the
radioimmune assay. This high dissociation constant of the
antibody present in anti-CEA sera would implicate the
protein as important in the antigenic determinant(s) mea
sured in the radioimmune assay, since it appears that
antibodies directed against carbohydrate antigenic determi
nants do not attain affinities as high as those directed
against peptide antigenic determinants (9).
Since bivalent binding to a single antigen molecule does
not play a role in the radioimmune assay for CEA, the
antigenic determinants are probably not repeating units of
closely spaced carbohydrate units. This conclusion is sup
ported by periodate oxidation studies (2, 4, 7) in which
antigenic activity survives extensive carbohydrate destruc
tion.
These results strongly suggest that the major antigenic
determinant(s) of CEA measured in the radioimmune assay
resides in the protein portion.
ACKNOWLEDGMENTS
The technical assistance of Vickie Vanik and William C. Schnute, Jr., is
gratefully acknowledged.
REFERENCES
I. Banjo, C., Gold, P., Gehrke, C. W., Freedman, S. O., and Krupey, J.
Preparation and Isolation of Immunologically
Active Glycopeptides
from Carcinoembryonic
Antigen (CEA). Intern. J. Cancer, 13:
151 163, 1974.
1975
Downloaded from cancerres.aacrjournals.org on June 17, 2017. © 1975 American Association for Cancer Research.
1807
J. E. Morris et al.
2. Coligan, J. E., Egan, M. L., Guyer, R. L., Schnute, W. C., Jr., and
Todd, C. W. Structural Studies on the Carcinoembryonic Antigen.
Ann. N. Y. Acad. Sci., in press.
3. Coligan, J. E., Lautenschleger, J. T., Egan, m. 1., and Todd, C. W.
Isolation and Characterization of Carcinoembryonic Antigen. Immunochemistry, 9: 377 386, 1972.
4. Coligan, J. C., and Todd, C. W. Structural Studies on Carcinoembry
onic Antigen (CEA): Periodate Oxidation. Biochemistry, 14: 805
810, 1975.
5. Crothers, D. M., and Metzger, H. The Influence of Polyvalency on the
Binding Properties of Antibodies. Immunochemistry, 9: 341 351,
1972.
6. Egan, M. Carcinoembryonic Antigen. In: R. Nakamura (ed.), Immunopathology: Clinical Laboratory Concepts and Methods, pp.
564 580. Boston: Little, Brown and Co., 1974.
7. Egan, M. L., Coligan, J. E., Morris, J. E.. Schnute, W. C., Jr., and
Todd, C. W. Antigenic Determinants on Carcinoembryonic Antigen:
Chemical and Immunological Studies. Proceedings. Eleventh Intern.
Cancer Congress, in press.
8. Egan, M. L., Lautenschleger, J. T., Coligan, J. E., and Todd, C. W.
Radioimmune Assay of Carcinoembryonic Antigen. Immunochemis
try, 9: 289 299, 1972.
9. Eisen, H. N. Antibody-Antigen Reactions, in: B. D. Davis, R.
Dulbecco, H. N. Eisen. H. S. Ginsberg, and W. B. Wood (eds.).
Microbiology, pp. 376 377. New York: Harper & Row, Inc., 1968.
10. Gold, P., and Greedman, S. O. Demonstration of Tumor Specific
Antigens in Human Colonie Carcinomata by Immunological Toler
ance and Absorption Technique. J. Exptl. Med., 121:439 462, 1965.
11. Gold, P., and Freedman, S. O. Specific Carcinoembryonic Antigens of
1808
the Human
Digestive System. J. Exptl. Med., 122: 467 481, 1965.
12. Gotschlich, E. C. A Simplification of the Radioactive Antigen Binding
Test by a Double Label Technique. J. Immunol., 107.-910 9] I, 1971.
13. Hornick. C. L.. and Karush, F. Antibody Affinity III: The Role of
Multivalence. Immunochemistry, 9: 325 340. 1972.
14. Levy, H. B., and Sober, H. A. A Simple Chromatographie Method for
Preparation of Gamma Globulin. Proc. Soc. Exptl. Biol. Med.. 103:
250-252, I960.
15. Nisonoff, A., and Rivers. M. M. Recombination of a Mixture of
Univalent Antibody Fragments of Different Specificity. Arch. Biochem. Biophys.. 93: 460 462, 1961.
16. Scheidegger, J. J. Une Micro-méthodede l'Immunoelectrophorese.
Intern. Arch. Allergy, 7: 103 110, 1955.
17. Slayter, H. and Coligan. J. E. Electron Microscopy and Physical
Characterization of the Carcinoembryonic Antigen. Biochemistry, in
press.
18. Terry, W. D., Henkart, P. A., Coligan, J. E., and Todd, C. W.
Carcinoembryonic Antigen: Characterization and Clinical Applica
tions. Transplant. Rev., 20: 100-129, 1974.
19. Thomson, D. M. P., Krupey, J.. Freedman, S. O., and Gold, P. The
Radioimmune Assay of Circulating Carcinoembryonic Antigen of the
Human Digestive System. Proc. Nati. Acad. Sei. U. S.. 64: 161 167.
1969.
20. World Health Organization. Nomenclature for Human Immunoglobulins. Bull. WHO, 30: 447 450, 1964.
21. Zamcheck, N., and Kupchik, H. Z. The Interdependence of Clinical
Investigations and Methodological Development in the Early Evolu
tion of Assays for Carcinoembryonic Antigen. Cancer Res., 34:
2131-2136, 1974.
CANCER RESEARCH
VOL. 35
Downloaded from cancerres.aacrjournals.org on June 17, 2017. © 1975 American Association for Cancer Research.
The Binding of Carcinoembryonic Antigen by Antibody and Its
Fragments
James E. Morris, Marianne L. Egan and Charles W. Todd
Cancer Res 1975;35:1804-1808.
Updated version
E-mail alerts
Reprints and
Subscriptions
Permissions
Access the most recent version of this article at:
http://cancerres.aacrjournals.org/content/35/7/1804
Sign up to receive free email-alerts related to this article or journal.
To order reprints of this article or to subscribe to the journal, contact the AACR Publications
Department at [email protected].
To request permission to re-use all or part of this article, contact the AACR Publications
Department at [email protected].
Downloaded from cancerres.aacrjournals.org on June 17, 2017. © 1975 American Association for Cancer Research.