Primary Culture of Squamous Head and Neck Cancer With and

Without 3T3 Fibroblasts and Effect of Clinical Tumor

Characteristics on Growth In Vitro

MELODY A. COBLEIGH, MD," JOHN L. KENNEDY, MD,* ALTON C. WONG, MD,* JAMES H. HILL, MD,§
KARIN M. LINDHOLM, BS,' JANE E. TIESENGA, MD,t RICHARD KIANG, BS,'
EDWARD L. APPLEBAUM, MD,t AND WILLIAM P. MCGUIRE, MD*

Twenty-one tumors from patients with squamous cell carcinoma of the head and neck (SCC H/N) were
cultured with and without 3T3 fibroblasts in order to determine whether, on the basis of improved tissue
culture medium, 3T3 cells could be deleted without altering growth and cloning efficiency. Thirty-five
additional primary SCC H/N specimens, cultured in the presence of 3T3 fibroblasts, were studied to
assess the effect of tumor differentiation, site of primary tumor, and site of specimen procurement on
growth. The authors conclude that 3T3 cells remain essential for optimal growth and cloning efficiency.
Also, 3T3 cells improved the number of successful cultures by 33% to 100%depending on the plating
density, and cloning efficiency was improved by 50%in the presence of 3T3 cells. Growth did not correlate
with tumor differentiation or site of origin of the tumor specimen. Culture of specimens from the primary
site resulted in growth significantly more frequently than culture of specimens obtained from metastatic
neck nodes.
Cancer 59:1732-1738, 1987.

S QUAMOUS CELL CARCINOMA of the head and neck

(SCC H/N) is a difficult tumor to culture from op-
erative specimens. Growth has been reported in 0% to
45% of specimens using a soft agar technique and uni-
formly low cloning efficiencies have been n0ted.l-j Growth
using surface culture techniques has occurred in 13 to 27
percent of specimen^.^,^
We became interested in a technique developed by
Rheinwald because of the high cloning efficiency observed
in malignant cell lines brought into cultures by this
means.7This technique utilizes 3T3 fibroblast feeder layers
and a complex tissue culture medium. The media used
in the original description of the technique included 20%
fetal calf serum (FCS), fortified Eagle Medium and hy-
drocortisone (0.4 pg/ml).8 A number of medium modi-
fications have been made since the technique was first
described (Table l).9 All of these modifications were based
upon observations made on cultures of human skin. Since
it is desirable to be able to grow SCC H/N without feeder
cells we wondered if the medium improvements had made
malignant squamous cell growth possible without 3T3
cell contact. The questions posed were two: In the presence
of the current complex tissue culture medium (1) do 3T3
fibroblasts still contribute significantly to colony formation
of SCC H/N primary cultures; and (2) do 3T3 fibroblasts
contribute significantly to cloning efficiency of SCC H/N
in primary culture? In 2 1 consecutive primary specimens
cultured with and without 3T3 fibroblasts we demonstrate
the continued need for 3T3 cells to obtain optimal growth
and cloning efficiencies of SCC H/N. We also describe
the relationship between tumor differentiation, site of pri-
mary tumor and site of specimen procurement (primary
site versus metastatic node), and growth of SCC H/N in
primary culture in the presence of 3T3 fibroblasts.

Methods
From the University of Illinois College of Medicine at Chicago, *De-
partment of Medicine, the ?Department of Otolaryngology-Head and Obtaining the Tumor Specimen
Neck Surgery,the $Department of Pathology, and the West Side Veterans'
Administration Medical Center, §Department of Surgery, Chicago, Il-
linois. Solid tumor specimens were harvested either in vivo
Supported by the Illinois Cancer Council and The University of Illinois directly from the tumor bed or from the surgical bloc
Foundation. immediately after tumor resection. These specimens were
Address for reprints: Melody A. Cobleigh, MD, Department of Med-
icine, P.O. Box 6998, Chicago, IL 60680.
placed immediately in cold McCoy's medium (Gibco,
Accepted for publication December 29, 1986. Grand Island, NY) supplemented with 10%heat-inacti-

1732
No. 10 IN VITRO GROWTHOF SCC HN
vated fetal calf serum (HI-FCS) (Gibco) and 1% penicillin/
streptomycin (Gibco) and transported expeditiously to the
laboratory. All H/N cancers studied arose from mucous
membrane.
Disaggregation of the Tumor
The tumor specimen was carefully dissected free of gross
nontumor tissue. Any normal adjacent mucosa or intact
normal appearing epithelium was discarded. One gram
of tumor was cut with scalpels into 1-mm to 2-mm pieces
and suspended in a 25-ml trypsinizing flask containing
15 ml of an enzyme solution of DNase, 776 U/ml (Type
I from bovine pancreas, 2470 Kunitz U/mg solid, Sigma)
and collagenase, 800 U/ml (sterile-IA-S, 290 U/mg solid,
Sigma, St. Louis, MO). The diluent consisted of HBSS
with calcium, magnesium, and bicarbonate (Gibco), buf-
fered with 25 mmol HEPES. The flask was placed in an
incubator for 90 minutes on a magnetic stirrer at the low-
est possible speed. The incubator was maintained at 37”C,
and the internal atmosphere was humidified and supple-
mented with 5% carbon dioxide (C02).Subsequently the
specimen was washed twice in the diluent. Viability was
assessed by trypan blue exclusion.
Removal of Aggregates
Aggregates were removed by pouring the resuspended
specimen through a stainless steel wire mesh with a pore
size of 25 pm (Gilson Co., Malinta, OH). Filtrates then
were diluted to the final plating concentrations. Specimens
were not recentrifuged after filtration to prevent reaggre-
gation of cells.
Maintenance of 3T3 Fibroblasts
Swiss mouse embryo-derived 3T3 fibroblasts” were
purchased from the American Type Culture Collection
(Rockville, MD) and maintained in Dulbecco’s Modified
Eagle Medium (DMEM) (#320- 1965, Gibco) supple-
mented with 10% calf serum (CS). Stock cultures were
refed twice weekly and were subcultured by treating with
0.1%trypsin for 5 minutes at 37°C followed by vigorous
washing with DMEM/ 10% CS. Stock cultures were not
allowed to become confluent.
Preparation of 3T3 Feeder Layer
On the afternoon before specimen harvest 3T3 cells
were placed in 10 X 35 mm petri dishes with 2-mm grids
(Lux; Miles Laboratories, Naperville, IL) at one third
confluent density. Confluent density is 5.5 X 104/cm2,
thus 1.5 X lo5cells were placed in each 10 X 35 mm petri
dish (surface area 8 cm2). On the following morning after
fibroblasts had adhered to the dish, 3T3 cells were treated
with mitomycin C (Bristol, Syracuse, NY) at a concen-
- Cobleigh et al. 1733
TABLE1. Culture Conditions for Squamous Cells
Dulbecco’s Modified Eagle Medium
Ham’s nutrient F- I2 (25%)
Fetal calf serum (5%)
Hydrocortisone (0.4 rglml)
Epidermal growth factor (20 ng/ml)
Cholera toxin (10-10 M)
Transfemn ( 5 %/mu
Insulin ( 5 rg/ml)
Tri-iodothyronine (2 X 10-11 M)
Adenine (1.8 X 10-4 M)
Nutrient mixture F-12 (HAM) (25%)
Pen/strep (1%)
Amphotericin B (primary cultures only) (20 rg/ml)
Swiss mouse 3T3 fibroblasts
From Wu, 1982.9
tration of 4 pg/ml in DMEM/10% CS for 30 minutes at
37°C. This renders the cells incapable of reproduction.”
The mitomycin C was thoroughly washed from the plates
using several rinses of DMEM/10% CS, and 1 ml of squa-
mous cell culture medium was added preparatory to add-
ing the cancer specimen. Plates were returned to the in-
cubator.
Culture Medium
Squamous culture medium consisted of DMEM sup-
plemented with 25% Ham’s Nutrient F-12 (320-1765
Gibco), 5% FCS (Gibco), 100 U/ml penicillin and 100
pg/ml streptomycin (Gibco), 2.5 pg/ml amphotericin B
(Gibco), 0.4 pg/ml hydrocortisone (Calbiochem, San
Diego, CA), M cholera enterotoxin (Schwartz-
Mann, New York, NY), 5 pg/ml regular insulin (Eli Lilly,
Indianapolis, IN) and 2 X lo-” M triiodothyronine
(Sigma), 5 pg/ml transfenin (Sigma) and 1.8 X lop4 M
adenine (Sigma).
CuIiure Conditions
Twenty specimens were cultured in triplicate over a 3
log range (5 X lo4, 5 X lo5, and 5 X lo6 trypan blue-
excluding cells per dish) in 2 ml of squamous culture me-
dium without 3T3 cells. Parallel cultures were established
in the presence of mitomycin C-treated 3T3 feeders. One
additional specimen was cultured only at a plating density
of 5.0 X lo4 with and without 3T3 cells due to an insuf-
ficient number of cells for the other concentrations. These
2 1 specimens constituted the comparison experiment and
an additional 15 specimens were cultured only with mi-
tomycin C-treated 3T3 cells at plating densities of l to 5
X lo6 cells per culture. These 15 specimens as well as the
20 specimens cultured at the 5 X lo6 density in the pres-
ence of 3T3 cells (35 in all) were used to assess the rela-
tionship between clonogenic growth and tumor differ-
entiation, primary tumor site, and site of origin of spec-
imen. Cultures were maintained in a 5% C 0 2 in air,
1734 CANCERMay 15 1987
FIG. 1 . Colony of SCC H/N. Colonies must contain at least 30 cells and demonstrate a closely knit cobblestone architecture (X200).
humidified, 37 "C incubator. Spent medium was replaced
twice weekly for 1 month.
Colony Counting and Definition
Whole-plate counts were performed twice weekly. A
colony was defined as a collection of 30 or more cells
growing in a closely knit cobblestone pattern on the sur-
face of the culture vessel (Fig. I). A specimen was con-
sidered to have grown if one or more colonies formed per
plate. Colony counting was discontinued when two col-
onies became confluent (Fig. 2) or if no colonies were
observed by day 28. If human fibroblasts became dense
they were thinned by incubating cultures in 0.2% ethyl-
enediamine tetraacetic acid (EDTA) for 30 seconds and
rinsing with squamous culture medium.
Histopathologic Grading of Tumor Specimens
Surgical pathology specimens were reviewed by a single
pathologist (J.L.K.). The degree of differentiation was
graded well, moderate, or poor. Keratinization, either with
keratin pearl formation or individual cell keratinization,
VOl. 59
were considered evidence of differentiation. All slides
showing tumor were examined. A tumor was considered
well differentiated if more than two-thirds of the area of
the tumor was keratinized, moderately differentiated if
two-thirds to one-third of the tumor was keratinized, and
poorly differentiated if less than one-third of the tumor
was keratinized. Nuclear pleomorphism, disorganized
growth pattern and random orientation of tumor cells
further characterized areas of poor differentiation.
Statistical Analysis
Growth with 3T3 cells was compared to growth without
3T3 cells at each plating density by chi-square. The num-
ber of colonies formed in successful cultures both with
and without 3T3 cells was compared using a paired, two-
tailed t test. The thirty-five 3T3-supported cultures (1 to
5 X lo6tumor cells per inoculum) were analyzed for the
relationship between tumor differentiation and growth
rate using a 3 X 2 contingency table. The primary site of
the tumor was recorded as well as the origin of the lab-
oratory specimen (metastaticneck lymph node or primary
No. 10 OF SCC HN
IN VITROGROWTH
FIG. 2. Colonies of SCC H/N have become confluent (arrow). Colony counting ceases when two or more colonies become confluent (X40).
site) in order to determine if either parameter affected
growth rate. This was done using the chi-square analysis.
Results
3T3 Support
The average number of cells per gram of tumor tissue
was 60 X lo6. The average viability was 66.7% f 3.3
(SEM). Comparison of growth from specimens cultured
with and without 3T3 cells revealed a consistent increase
in the number of successful cultures over a 3 log range of
plating densities (Fig. 3). At 5 X lo4 cells per dish 14%of
specimens grew without 3T3 support compared to 28%
with such support. These figures were 45% versus 60%
and 35% versus 65% for the 5 X lo5 and 5 X lo6 plating
densities, respectively. The P values for these differences
were less than 0.25, 0.25, and 0.05, respectively. Colony
growth was first observed after a median of 10.5 days.
Figure 4 shows a high-power view of a typical colony
stained with the Papanicolaou technique. Nineteen spec-
imens grew both with and without 3T3 support. All spec-
imens which grew without 3T3 cells also grew in the pres-
ence of 3T3 cells. Conversely, 12 specimens which grew
in the presence of 3T3 cells failed to grow in their absence.
- Cobleigh et al. 1735
Mean cloning efficiencies were calculated for each in-
oculum, with and without 3T3 support. Specimenswhich
did not grow at any inoculum, with or without 3T3 sup-
port were not included in these calculations. The mean
cloning efficiencies of specimens cultured in the absence
of 3T3 cells were 1.64 X lop4%,6.93 X and 6.4
X for 5 X lo6, 5 X lo5 and 5 X lo4 inoculums,
respectively. The mean cloning efficiencies of specimens
cultured in the presence of 3T3 cells were 4.01 X lop4%,
1.3 X lop3%,and 1.14 X lo-’%, respectively. Statistical
differences for cultures with versus without 3T3 support
were P < 0.001, P < 0.06, and P < 0.05 for 5 X lo6, 5
X lo5,and 5 X lo4 inoculums, respectively (paired t test).
Clinical Correlations
The primary site, specimen source, degree of tumor
differentiation and in vitro growth status of tumor in the
35 patients assessed are shown in Table 2. Colony growth
was observed in 24/35 (69%)of specimens. No correlation
between site of primary tumor or tumor differentiation
and growth was apparent. The source of the culture spec-
imen did make a difference. Five specimens were obtained
from neck nodes and only one of those grew compared
1736 CANCERM a y 1 5 1987
PERCENT OF SPECIMENS WHICH GREW.
=W r-
55-
50-
45-
40-
3s-
30-
25-
m-
15 -
10 -
5 -
-
n-
5.ooo.m y30. a00
PLATING DENSITY
with 30 specimensobtained from the primary site of which
23 grew (P< 0.005). The mean cloning efficiency for the
entire group of specimens was 7.5 x
Discussion
Fibroblasts can be cultivated in vitro with ease. Epi-
thelial cells are more fastidious in their growth require-
ments and are frequently overgrown by stromal fibro-
blasts. Culture environments which inhibit fibroblast
growth such as the human tumor colony forming assay12
provide a selective advantage for cells which are fibroblast-
independent (such as ovarian cancer cells) but not for
cells which are relatively or absolutely fibroblast-depen-
dent and/or anchorage dependent (such as squamous
cancer cells).
Squamous cell carcinoma of the head and neck is an
anchorage-dependent tumor. In our experience, using a
separation vehicle with 25-pm pores to attain a relatively
pure single cell suspension, none of 51 specimens was
cultured successfully in the human tumor colony forming
assay.’ On the other hand, using the same separation ve-
hicle, we cultured 69% of 35 specimens successfully in an
assay developed by Rheinwald and Green.8The problem
of fibroblast overgrowth of epithelial components which
are fibroblast-dependent can be circumvented by using
mitomycin C-treated xenogeneic fibroblastswhich support
epithelial growth while inhibiting autologous fibroblast
growth.
In 1974, while isolating epithelial cell lines from por-
tions of mouse teratoma Rheinwald and Green found that
one line grew only when cocultivated with fibroblast^.'^
Gamma-irradiated 3T3 cells sufficed as the feeder layer
for this line and the same feeder cell population was sub-
VOl. 59
FIG.3. This graph illustrates the
effect of 3T3 fibroblasts on growth
of SCC H/N. Three plating densities
with and without 3T3 cells are shown
on the abscissa. The ordinate depicts
the percent of specimens which
formed colonies. Twenty specimens
were studied at the 5,000,000 and
500,000 inoculum; 2 1 at the 50,000
inocolum.
sequently used successfully to support growth of human
skin and of human SCC H/N.7 The original culture me-
dium was simple, consisting of fortified Eagle medium,
hydrocortisone, and FCS8 In serial publications this me-
dium was supplemented with a purine base, a variety of
hormones and hormone inducers, amino acids, and an-
tibiotics (Table l).9 Each modification was made because
it resulted in improved growth of human keratino-
cytes.8.9.14- I6
Cell lines have been established from SCC H/N spec-
imens using 3T3 feeder layers7but the medium used con-
tained only fortified Eagle Medium, FCS, and hydro-
cortisone. To be able to grow SCC H/N without feeder
cells is important because medium components are then
better defined. Therefore we wondered if, in the evolution
of the current culture medium, 3T3 cells had become
unnecessary for support of malignant squamous cell
growth. Our results show that 3T3 fibroblasts contribute
importantly to both the number of successful cultures and
cloning efficiency of SCC H/N despite improvements in
culture medium. Although it is possible to confuse small
SCC H/N colonies with normal squamous colonies under
inverted microscopy, it is highly unlikely that this occurred
with important frequency in our study. Cancer specimens
were dissected free of grossly normal epithelium, assuring
the invasive nature of the cells plated. Ten consecutive
specimens which formed colonies were cytospun and
stained before culture in order to assure the malignant
nature of the preparation. Eight specimens contained no
squames. Two contained 1 squam each. Furthermore,
specimens which formed colonies were randomly replated
in culture vessels which contained microscope slides.
These slides were alcohol fixed and papanicolaou stained.
All colonies exhibited morphologic features of malignant
No. 10 IN VITROGROWTHOF SCC HN
FIG.4. Papanicolaou stain of a colony of SCC H/N. Notice the prominent, multiple nucleoli, frequent mitotic figures, large nuclear-to-cytoplasmic
ratio, hyperchromatic nuclei, and anisocytosis (XSOO).
squamous cells, including greater than ten mitotic figures
per ten lOOX fields, large nuclear to cytoplasmic ratios
and prominent, usually multiple nucleoli (Fig. 4).
Growth of SCC H/N has been correlated with site of
specimen procurement and tumor differentiation in a soft
agar a ~ s a y . ~ , ’However,
~”’ these variables have not been
analyzed for primary culture using surface culture tech-
niques. We found that specimens obtained from primary
tumor sites grew significantlymore often than those from
metastatic lymph nodes (77% versus 20%).Whether this
reflects an intrinsic biologic property of the primary tumor
or the possibility that lymph node specimens contain fewer
tumor cells per gram of tissue is a question which deserves
further study. We observed no correlation between tumor
differentiation or location of the primary tumor and
growth in vitro.
Krause6 observed no correlation between the site of
tumor origin, extent of disease or degree of tumor differ-
entiation and surface growth. They reported that every
cell line established was derived from a rapidly growing,
recurrent tumor. Possibly the culture system they used is
best able to propagate the most aggressive tumors since
- Cobleigh et al. 1737
lines were established from three of 194 tumors. Using
the Rheinwald technique 69%of our specimens grew in
primary culture and as a result it is unlikely that a
subgroup of highly aggressive tumors was selected. Fur-
thermore we studied the correlation between clinical tu-
mor characteristics and growth in primary culture, not
growth of cell lines.
The cloning efficiencies reported here are lower than
those reported for cultured cell lines. This is to be expected
since the primary specimen is not a pure tumor cell pop-
ulation but rather consists of fibroblasts, lymphocytes,
histiocytes, plasma cells, monocytes, granulocytes, mast
cells, and tumor cells which vary proportionally from pa-
tient to patient. Furthermore, colony counting ceased
when two colonies became confluent. This frequently oc-
curred when new colonies were still forming. Thus, the
figures reported here are underestimatesof the true cloning
efficiency. This problem is resolvable with use of larger
culture plates. For primary cultures we recommend be-
ginning with a plating density of lo5 or lo6 cells since
only 1/2 1 specimens cultured grew at the 1O4 plating den-
sity and not at the other densities.
1738 CANCERMay 15 1987 VOl. 59

TABLE2. Characteristics of Tumor Specimens primary tumor site than from a metastatic lymph node;
Cultured on 3T3 Cells (4) There is no correlation between growth and tumor
Tumor Specimen differentiation or site of origin of the cancer; and ( 5 ) This
Specimen Primary site Source differentiation grew technique represents a significant improvement in the
1 Oropharynx Primary Well Yes
culture of SCC H/N when compared with techniques
2 Larynx Primary Mod Yes which use semisolid medium. Sixty-nine percent of 35
3 Tongue Primary Mod Yes specimens cultured grew in this system. Each of these
4 Larynx Primary Mod Yes
5 Nasopharynx Primary Mod Yes
conclusions represents new information important for
6 Tonsil Primary Well No other investigators using the technique for the study of
7 Larynx Primary Mod No SCC H/N.
8 Base of tongue Primary Mod Yes
9 Tonsil Primary Mod Yes REFERENCES
10 Larynx Primary Mod Yes
I1 Larynx Node Mod No I . Archibald SD, Browman GP, Holmes M, Jellie S. Current limi-
12 Pyriform sinus Node Well No tations of the in vitro clonogenic assay in squamous cell carcinoma of
13 Tonsil Node Well Yes the head and neck (Abstr 164). Presented at the International Conference
14 Hypopharynx Primary Mod Yes on Head and Neck Cancer, July 1984. (in press).
15 Tonsil Primary Mod Yes 2. Cobleigh MA, Gallagher PA, Hill JH, Applebaum EL, McGuire
16 Base of tongue Primary Poor Yes WP. Growth of human squamous head and neck cancer in vitro. Am J
17 Hypopharynx Primary Well Yes Pathol 1984; 115:397-402.
18 Larynx Primary Mod Yes 3. Johns ME. The clonal assay of head and neck tumor cells: Results
19 Tonsil Primary Mod Yes and clinical correlations. Laryngoscope 1982; (Suppl) 92: 1-26.
20 Larynx Primary Mod Yes 4. Mattox DE, Von Hoff DD. In vitro stem cell assay in head and
21 Oropharynx Node Mod No neck squamous carcinoma. Am J Surg 1980; 140527-530.
22 Larynx Primary Well Yes 5. Rupniak HT, Hill BT. The poor cloning ability in agar of human
23 Base of tongue Primary Well Yes tumor cells from biopsies of primary tumors. Cell Biol Znt Rep 1980; 4:
24 Tongue Primary Well No 479-486.
25 Larynx Primary Well Yes 6. Krause CJ, Carey TE, Ott RW, Hurbis C, McClatchey KD, Regezi
26 Oropharynx Primary Well Yes JA. Human squamous cell carcinoma: Establishment and characteriza-
27 Larynx Primary Well Yes tion of new permanent cell lines. Arch Otolaryngol 1981; 107:703-710.
28 Tongue Primary Well No 7. Rheinwald JG, Beckett MA. Tumorigenic keratinocyte lines re-
29 Floor of Primary Well Yes quiring anchorage and fibroblast support cultured from human squamous
mouth cell carcinomas. Cancer Res 1981; 41:1657-1663.
30 Larynx Primary Poor Yes 8. Rheinwald JG, Green H. Serial cultivation of strains of human
31 Larynx Node Poor No epidermal keratinocytes: The formation of keratinizing colonies from
32 Larynx Primary Mod No single cells. Cell 1975; 6:331-344.
33 Pyriform sinus Primary Mod Yes 9. Wu YJ, Parker LM, Binder NE et al. The mesothelial keratins: A
34 Floor of Primary Well No new family of cytoskeletal proteins identified in cultured mesothelial
mouth cells and non-keratinizing epithelia. Cell 1982; 3 1:693-703.
35 Base of tongue Primary Mod No 10. Todaro GJ, Green H. Quantitative studies of the growth of mouse
embryo cells in culture and their development into established lines. J
Well: well-differentiated;Mod: moderately differentiated; Poor: poorly Cell Biol 1963; 17:299-313.
differentiated. 11. Rheinwald JG. Serial Cultivation of normal human epidermal
keratinocytes. Meth Cell Biol 1980 21A:229-254.
12. Salmon SE, Hamburger AW. Quantitation of differential sensi-
Conclusion tivity of human-tumor stem cells to anti-cancer drugs. N Engl J Med
1978; 298: 1322-1 327.
We conclude that (1) 3T3 fibroblasts continue to sig- 13. Rheinwald JG. Green H. Formation of a keratinizing eDithelium
nificantly enhance th successful primary culture of SCC in culture by a cloned cell line derived from a teratoma. &i 1975; 6:
3 17-330.
H/N at high inoculums in spite of improvements in cul- 14. Green H. Cyclic AMP in relation to proliferation ofthe epidermal
ture medium. Enhancement also was observed at lower cell: A new view. Cell 1978; 15:801-811.
inoculums but statistical significance was not reached, 15. Rheinwald JG, Green H. Epidermal growth factor and the mul-
tiplication of cultured human epidermal keratinocytes. Nature 1977;
probably due to the small numbers studied; (2) 3T3 cells 265~421-424.
enhance cloning efficiency of cultures which are destined 16. Watt FW,Green H. Involucrin synthesis is correlated with cell
to grow without them in spite of improvements in culture size in human epidermal cultures. J Cell Biol 1981; 90:738-742.
17. Kish JA, Crissman JD, Haas C et al. Parameters for predicting
medium; (3) Primary growth of SCC H/N in this assay is growth of squamous head and neck tumors in the human stem cell assay
more likely to occur if the specimen is taken from the (HTSCA) (Abstr). Proc Am Assoc Cancer Res 1983; 24:3.