Content uploaded by David T Yang
Author content
All content in this area was uploaded by David T Yang on Jan 04, 2016
Content may be subject to copyright.
Cancer Therapy: Preclinical
Development and Characterization of HPV-Positive and
HPV-Negative Head and Neck Squamous Cell Carcinoma
Tumorgrafts
Randall J. Kimple
1,6
, Paul M. Harari
1,6
, Alexandra D. Torres
1,2
, Robert Z. Yang
1,2
, Benjamin J. Soriano
3
,
Menggang Yu
4,6
, Eric A. Armstrong
1
, Grace C. Blitzer
1
, Molly A. Smith
1
, Laurel D. Lorenz
2
, Denis Lee
2
,
David T. Yang
3,6
, Timothy M. McCulloch
5,6
, Gregory K. Hartig
5,6
, and Paul F. Lambert
2,6
Abstract
Purpose: To develop a clinically relevant model system to study head and neck squamous cell carcinoma
(HNSCC), we have established and characterized a direct-from-patient tumorgraft model of human
papillomavirus (HPV)–positive and HPV-negative cancers.
Experimental Design: Patients with newly diagnosed or recurrent HNSCC were consented for donation
of tumor specimens. Surgically obtained tissue was implanted subcutaneously into immunodeficient mice.
During subsequent passages, both formalin-fixed/paraffin-embedded as well as flash-frozen tissues were
harvested. Tumors were analyzed for a variety of relevant tumor markers. Tumor growth rates and response
to radiation, cisplatin, or cetuximab were assessed and early passage cell strains were developed for rapid
testing of drug sensitivity.
Results: Tumorgrafts have been established in 22 of 26 patients to date. Significant diversity in tumorgraft
tumor differentiation was observed with good agreement in degree of differentiation between patient tumor
and tumorgraft (Kappa 0.72). Six tumorgrafts were HPV-positive on the basis of p16 staining. A strong
inverse correlation between tumorgraft p16 and p53 or Rb was identified (Spearman correlations P¼0.085
and P¼0.002, respectively). Significant growth inhibition of representative tumorgrafts was shown with
cisplatin, cetuximab, or radiation treatment delivered over a two-week period. Early passage cell strains
showed high consistency in response to cancer therapy between tumorgraft and cell strain.
Conclusions: We have established a robust human tumorgraft model system for investigating HPV-
positive and HPV-negative HNSCC. These tumorgrafts show strong correlation with the original tumor
specimens and provide a powerful resource for investigating mechanisms of therapeutic response as well as
preclinical testing. Clin Cancer Res; 19(4); 855–64. 2012 AACR.
Introduction
Head and neck squamous cell carcinoma (HNSCC) is the
sixth most common worldwide cancer with approximately
620,000 new diagnoses annually (1). The development of
these cancers has been traditionally associated with tobacco
and alcohol use. Evidence in recent years has identified an
etiologic role of human papillomavirus (HPV) in a sub-
stantial subset of patients with HNSCC, many of whom do
not have a strong history of tobacco and alcohol use (2).
Radiation alone or with concurrent chemotherapy is an
important component of therapy for patients with HNSCC.
Treatment can be quite toxic with significant acute and long-
term side effects. Retrospective and prospective analyses
confirm a striking difference in clinical outcome between
HPV-positive and HPV-negative HNSCC (3). However, at
present, there is no compelling data that therapy recom-
mendations can be guided on the basis of HPV status. To
facilitate the development of novel therapy approaches and
to better understand molecular mechanisms that underlie
therapeutic responses, robust preclinical model systems are
needed.
We sought to establish and validate a direct-from-
patient human tumorgraft model system of HNSCC that
could be utilized to identify molecular targets, validate
novel therapeutics, guide treatment recommendations,
and facilitate studies regarding biologic mechanisms of
treatment sensitivity and/or resistance. Such tumorgraft
model systems differ from traditional xenograft model
Authors' Affiliations:
1
Department of Human Oncology;
2
McArdle Lab-
oratory for Cancer Research and Department of Oncology;
3
Department of
Pathology;
4
Department of Biostatistics and Medical Informatics;
5
Depart-
ment of Surgery, Otolaryngology;
6
University of Wisconsin Carbone Can-
cer Center, University of Wisconsin, Madison, Wisconsin
Note: Supplementary data for this article are available at Clinical Cancer
Research Online (http://clincancerres.aacrjournals.org/).
Corresponding Author: Randall J. Kimple, Department of Human Oncol-
ogy, University of Wisconsin Comprehensive Cancer Center, 3107 WIMR,
1111 Highland Avenue, Madison, WI 53705. Phone: 608-265-9156; Fax:
608-263-9947;
E-mail: rkimple@humonc.wisc.edu
doi: 10.1158/1078-0432.CCR-12-2746
2012 American Association for Cancer Research.
Clinical
Cancer
Research
www.aacrjournals.org 855
systems in that the human tumors are grafted directly
from human subjects into immunodeficient mice, rather
than first being established as cell lines in tissue culture
and then being engrafted onto mice. The biologic rele-
vance of the traditional cell line-based xenograft model
system has come under increased scrutiny at various levels
including recent system wide analyses suggesting that
immortalized cell lines from a variety of tumor types
retain greater gene expression similarity to each other
than to their tissue of origin (4, 5). The tumorgraft model
system is increasingly recognized as a preclinical model
system that may provide greater biologic relevance to
human cancers at many levels including tumor patholo-
gy, growth, metastasis, disease outcome, and drug respon-
siveness (6–8).
Materials and Methods
Patient selection
Patients with newly diagnosed or recurrent HNSCC were
approached for possible tissue donation. Patients consent-
ing to participate in this IRB-approved protocol completed a
brief questionnaire collecting information regarding tobac-
co use, alcohol use, prior malignancy, sex, age, and prior
treatment received. At the time of surgery or staging biopsy,
a section of tumor was collected for research use making
sure not to compromise surgical margin or pathologic
assessment.
Mice
Immunodeficient mice used for tumorgraft development
included male and female NOD scid gamma (NSG, Jackson
Laboratories) and Hsd:athymic Nude-Foxn1
nu
(Harlan Lab-
oratories). All mice were kept in the Association for Assess-
ment and Accreditation of Laboratory Animal Care
approved Wisconsin Institute for Medical Research Animal
Care Facility and studies with them were carried out in
accordance with an approved animal protocol.
Establishment of tumorgrafts
Tumor was transported directly from the operating room
to the laboratory in ice-cold culture media [Dulbecco’s
Modified Eagle Medium (DMEM) with 10% fetal bovine
serum (FBS), 1% penicillin/streptomycin, and 25 mg/mL
amphotericin] and minced in a culture dish to less than 1
mm
3
pieces under sterile conditions. Minced tumor pieces
were mixed 1:1 with reduced growth factor Matrigel (cat
#354230, BD Biosciences, Inc) and injected subcutaneously
into NSG mice with an 18 gauge needle as passage zero (P0).
Every effort was made to accomplish this transfer within 1
hour of tumor harvest from the surgical procedure. Subse-
quent passages were made in a similar fashion into either
NSG or athymic nude mice.
Cryopreservation of tumorgrafts was accomplished by
mixing minced tumor pieces with transport media supple-
mented with 10% dimethylsulfoxide (DMSO). Tumors
were frozen in controlled rate freezers (1C/minute) to
80C overnight and transferred to liquid nitrogen for
long-term storage. To thaw tumors, aliquots were warmed
to 37C in a heated water bath and tumor tissue was washed
twice in transport media (without DMSO) and immediately
implanted into mice.
Histology of primary and tumorgrafts
At each passage, a section of the tumor was reserved for
fixation in 10% neutral-buffered formalin and subsequent
embedding in paraffin blocks. Five-mm sections were cut
and hematoxylin and eosin (H&E) stains were conducted
on every 10
th
section. Each patient’s primary tumor was also
stained by H&E and slides imaged on an Olympus BX51
microscope (Olympus America, Inc). Comparisons
between primary tumor, first passage tumorgraft, and sub-
sequent tumorgraft passages were made by a surgical
pathologist on the basis of differentiation, keratinization,
and overall tumor architecture.
Assessment of human papillomavirus
Early passage (P0–P2) tissue at the time of harvest from
immunodeficient mice was snap frozen in liquid nitrogen.
Total genomic DNA and total RNA from this tissue were
isolated using the DNeasy Blood and Tissue Kit, the miR-
Neasy Mini Kit, and RNeasy minElute spin column
(Cat#69504, 217004, and 74204, respectively, from Qiagen
Inc). Multiple methods were used to assess for the presence
of HPV.
Quantitative real-time PCR (qRT-PCR) was carried out
on a BioRad CFX96. Briefly, total RNA was harvested
usingthemiRNeasywiththeMinEluteKitwithsnap-
frozen tissue samples from passage 0. cDNA was synthe-
sized using the iScript Reverse Transcription Supermix
Kit (Bio-Rad Laboratories) and 1,000 ng of total RNA.
qRT-PCR was carried out by using IQ Multiplex Powermix
with 10 ng cDNA per 10 mLreaction.GAPDH, HPV-16 E5,
E6, and E7 transcripts were detected using primers and
probes (Supplementary Table S1) purchased from Inte-
grated DNA Technologies, Inc. The thermocycler was
programmed for an initial 95C for 7 minutes followed
Translational Relevance
We describe and characterize a direct-from-patient
tumorgraft model system of human head and neck
cancer including tumors derived from both human
papillomavirus–associated cancers and tobacco-associ-
ated cancers. Comparisons of primary tumor and tumor-
grafts in addition to the significant diversity of tumor
morphology suggest that this resource represents an
outstanding platform for investigating novel therapeu-
tics and combinations of chemotherapy and/or radia-
tion. These tumorgrafts may prove to be a valuable
resource for optimizing therapy for specific subgroups
of patients with head and neck cancer and may shed light
into the molecular mechanisms underlying differential
therapeutic responses between HPV-positive and HPV-
negative head and neck cancer.
Kimple et al.
Clin Cancer Res; 19(4) February 15, 2013 Clinical Cancer Research
856
by 40 cycles of 94C for 15 seconds and 60Cfor30
seconds.
Total genomic DNA was used to probe for HPV DNA
using a nested PCR approach previously described (9).
Briefly, for both rounds of PCR, a final concentration of
1PCR buffer, 0.2 mmol/L dNTPs, 1.5 mmol/L MgCl
2
, and
0.2 U Taq was used. In the first round, 100 ng of purified
DNA and MY09 and MY11 (Supplementary Table S1)
primers (which detect multiple HPV subtypes) at 0.2
mmol/L were used. The thermocycler was programmed for
an initial 94C for 4 minutes followed by 40 cycles of 94C
for 15 seconds, 55C for 30 seconds, and 72C for 1 minute
with final extension at 72C for 5 minutes. Final products
from PCR reactions were run on a 1.5% agarose gel, stained
with ethidium bromide, and imaged.
Southern blot was conducted using 10 mg of BamHI
digested total cellular DNA separated on a 1.25% agarose
gel, transferred to Hybond Nþnylon membrane (Amer-
sham) and crosslinked. DNA probes were made by 50end
labeling 10 pmoles of HPV16-specific oligonucleotides
(Supplementary Table S1) in the presence of T4 polynucle-
otide kinase (New England Biolabs Inc) with [g-
32
P] ATP
(6,000 Ci/mmol) at 37C for 1.5 hours. The membrane was
prehybridized with Church hybridization buffer for 15
minutes at 52C followed by probe hybridization for 18
hours at 52C in a hybridization oven. Membrane was
washed with Church wash buffer, exposed to a storage
phosphor screen and scanned using a Typhoon 8610 imag-
ing system (Amersham).
To assess for alternative HPV subtypes, an additional PCR
for the HPV E1 gene was carried out. Briefly, 100 ng of total
genomic DNA was amplified using the degenerate E1 pri-
mers (Supplementary Table S1) with the thermocycler
programmed for an initial 94C for 5 minutes followed by
40 cycles of 95C for 10 seconds, 50C for 10 seconds, and
72C for 30 seconds with final extension at 72C for 5
minutes. Final products were run on a 1.5% agarose gel,
stained with ethidium bromide, and imaged. Positive bands
were individually gel purified and sent for Sanger sequenc-
ing using the E1 primers. Sequences were annealed, base
discrepancies edited, and the resulting ideal sequence com-
pared via BLAST search.
Radiation and chemotherapy growth delay
Tumor growth rates and therapeutic response were mon-
itored by injecting athymic nude mice subcutaneously (n¼
12 per group) with tumors into bilateral flanks. Tumor
volume was monitored twice weekly by measurement
with Vernier calipers and calculated according to the equa-
tion V¼(p)/6 (large diameter) (small diameter)
2
.
When individual tumor volumes reached 200 mm
3
, mice
were stratified into treatment groups based on tumor size
such that each group had tumors with a similar range in size
as determined by Wilcoxon rank-sum test. Treatment com-
menced the next day with cisplatin (2 mg/kg), cetuximab
(0.2 mg), or vehicle control (0.95 normal saline) delivered
by intraperitoneal (IP) injection twice weekly. Radiation (2
Gy/fraction twice weekly) was delivered via an X-RAD 320
biologic irradiator (Precision X-Ray) using custom-
designed mouse jigs to immobilize animals and limit radi-
ation exposure to the tumors on the dorsal flanks. Time to
tumor quadrupling was calculated from the first day of
treatment. Curves were fit to an exponential growth equa-
tion and compared using the extra-sum-of-squares ftest
using Graphpad Prism v 5.0d.
Immunohistochemistry
The expression of p16 (BD Pharmingen catalog
#550834), p53, Rb, and EGFR was detected in histologic
sections of tumorgrafts by standard immunohistochemistry
(IHC). A one-step Gomori’s trichrome stain (ENG Scien-
tific) was used to develop 5 mparaffin sections of passage 0
tumorgrafts. Briefly, sections were deparaffinized, incubat-
ed in Bouins solution for 1 hour at 60C, washed in running
tap water, and the nuclei stained with Weigert’s iron hema-
toxylin for 15 minutes. Next, the slides were placed in
Gomori’s trichrome stain and incubated 20 minutes at
room temperature. Finally, the tumor sections were rinsed
with H
2
0, dehydrated, cleared, and coverslipped.
Images were acquired with a 20objective using an
Olympus BX51 fluorescent microscope and photographed
with a SPOT RT CCD camera (Diagnostic Instruments,
Inc.). A determination of positive versus negative was made
by a board certified pathologist based on cytoplasmic and/
or nuclear positivity more than 2þintensity in more than
70% of tumor cells (10). The intensity of p53 and Rb
staining was scored as follows: negative when less than
5% of tumor cells displayed staining; 1þwhen intensity
was mild; 2þ, moderate; 3þ, when intensity was equal to
the positive control; and 4þwhen intensity was greater than
the positive control (11).
Results
Establishment of tumorgrafts
A total of 37 patients with head and neck cancer have
been consented for tumor collection and establishment as
tumorgrafts. Twenty-six had tumor collected and implanted
subcutaneously into NSG mice as described in the Methods.
The remaining 9 patients did not undergo tumor collection
for the study because of limitations in tumor quantity, the
altered timing of surgery, and/or the availability of labora-
tory personnel for immediate tissue transfer. Average age of
donors was 61 (range 45–87). To date, 22 samples have
given rise to viable tumors (take rate: 85%). Molecular
characterization and assessment of HPV infection was car-
ried out for all tumors. Comprehensive tumor growth rate
assessment and therapeutic evaluation have been carried
out for 3 tumorgraft models to date: UW-SCC6, UW-
SCC14N, and UW-SCC22.
Clinical characteristics of the 26 patients who had tumor
collected and the relationship to tumor take rate are pre-
sented in Table 1. Briefly, by univariate analysis patient sex,
tobacco use, alcohol use, and T-stage did not correlate with
tumorgraft take rate. Successful tumorgraft take was also not
correlated with primary tumor differentiation or HPV sta-
tus. However, tumorgraft establishment was significantly
Head and Neck Cancer Tumorgrafts
www.aacrjournals.org Clin Cancer Res; 19(4) February 15, 2013 857
higher (100% vs. 64%, P¼0.02) from subjects with lymph
node metastases than from those without lymph node
metastases, regardless of whether the tumor biopsy was
taken from the primary tumor site or from a metastatic
lymph node.
Tumorgraft comparison with primary tumors
As shown in Table 2, tumorgrafts were established from
multiple subsites of the head and neck including orophar-
ynx (base of tongue and tonsil), oral cavity, and hypophar-
ynx. The 22 successful tumorgrafts analyzed were composed
of poorly differentiated tumor histology in 5 cases, mod-
erately differentiated in 9 cases, and well differentiated in 8
cases. Figure 1 shows representative images from 3 tumor-
grafts including H&E stain and p16 IHC of both the primary
tumor and early passage tumorgraft, trichrome stain, and
IHC of biomarkers p53, Rb, and EGFR. Strong retention of
overall tumor histology including cell morphology, stromal
component architecture, and the presence of cystic struc-
tures were observed between original patient tumor and
tumorgraft (Fig. 1) and in sections from multiple serial
passages of the same tumorgraft (Fig. 2A). In addition, good
agreement with regard to the degree of tumor differentia-
tion (i.e., poor, moderate, or well) was observed between
patient specimen and tumorgraft (Fig. 2B, unweighted
Kappa 0.72). To date, 17 tumors that were established in
mice (P0 generation) have been successfully passaged with
growth evident in the subsequent (i.e., P1) generation. Of
those tumors that have been passaged at least once (n¼22),
the mean time from initial tumorgraft implantation to first
passage was 115 days and to second passage (n¼17) was
109 days (Fig. 2C, P¼0.55). Eight tumors have been
passaged at least 4 times. Comparisons of differentiation
over multiple passages in a single tumorgraft have shown
remarkable stability (Supplementary Table S2) with a prob-
ability of change of differentiation estimated at 3% (one-
sided 95% CI: 15.8%).
Cryopreservation of tumorgrafts
One challenge of utilizing direct from patient tumorgrafts
is that tumor characteristics may change over time. While we
have not seen this to date, we have attempted to cryo-
preserve tumorgrafts to allow us to reanimate early passage
tumors at later time points. To date, 16 tumorgrafts have
been cryopreserved and 10 have been thawed and reim-
planted into NSG mice. At this time, 6 of 10 (60%) have
successfully grown additional tumor.
Assessment of HPV
In light of the causal association and prognostic signif-
icance of HPV infection in head and neck cancer, we
assessed the HPV status of all tumorgrafts. HPV status was
Table 1. Clinical characteristics of patients and corresponding frequency of tumorgraft establishment by
variable
Patient and tumor characteristics
Parameter Variable Patients (n) Tumor take rate (%) Significance
a
Age <60 13 85 NS
60 13 85
Sex Male 18 83 NS
Female 8 88
Tobacco use Nonsmoker 7 100 NS
Minimal use (<20 pack years) 1 100
>Minimal use 18 78
Alcohol use None/light 9 100 NS
Moderate 6 83
Heavy 11 73
T stage T1 or T2 12 83 NS
T3 or T4 14 86
Nodal status Node negative 11 64 P¼0.02
Node positive 15 100
Differentiation Well 11 82 NS
Moderate 11 82
Poor 4 100
HPV status Positive 8 100 NS
Negative 13 100
Testing not done 5 20
a
The frequency of tumor take compared via Fisher's exact test (age, sex, T-stage, nodal status, and HPV status) or Chi-squared test
(tobacco use, alcohol use, differentiation).
Kimple et al.
Clin Cancer Res; 19(4) February 15, 2013 Clinical Cancer Research
858
Table 2. Immunohistochemical characteristics and HPV status of tumorgraft at early passage
Tumorgraft Tumor site
Differentiation
of primary
Differentiation
of tumorgraft p53
a
Rb
b
Primary tumor
p16 status
(% of cells >¼
2þfor p16)
Tumorgraft
p16 status
(% of cells >¼
2þfor p16)
HPV-16 E6/E7
qRT-PCR
c
MY9/11
d
Southern
blot
d
Degenerate
E1 PCR
d
HPV subtype
from E1
sequencing
d
UW-SCC1P Base of Tongue Moderate Moderate þþ þþ positive (70%) positive (80%) þþþþ16
UW-SCC3P Tonsil Poor Poor þ þþþ positive (95%) negative (30%)
UW-SCC4P Floor of Mouth Well Well þþpositive (70%) positive (70%) þþþþ16
UW-SCC6P Tonsil Moderate Moderate þþpositive (90%) positive (90%) þþþþ16
UW-SCC10P Oral Tongue Moderate Poor þþ þþþ negative (0%) negative (30%)
UW-SCC12P Base of Tongue Well Well þþþ þþ positive (90%) negative (30%)
UW-SCC13P Oral Tongue Well Well þþþ þþþ negative (0%) negative (10%)
UW-SCC14N Base of Tongue Moderate Moderate þpositive (90%) negative (0%)
UW-SCC15P Floor of Mouth Poor Poor þþ þþ negative (5%) negative (0%) þþ16
UW-SCC17P Tonsil Poor Poor þþpositive (95%) positive (95%) þþþþ16
UW-SCC20P Floor of Mouth Well Moderate NE NE ND NE NE NE NE NE
UW-SCC22P Floor of Mouth Well Well þþþ þþþ negative (20%) negative (0%)
UW-SCC23P Floor of Mouth Well Well þþþ þþþ negative (10%) negative (0%)
UW-SCC24P Floor of Mouth Well Well þþþ þþ negative (0%) negative (20%)
UW-SCC25P Floor of Mouth Moderate Moderate þþþ negative (10%) positive (95%) þnegative
UW-SCC30P Floor of Mouth Moderate Well þþþ þþ negative (0%) negative (0%)
UW-SCC31P Hypopharynx Poor Poor þ þþþ negative (0%) negative (0%)
UW-SCC32P Buccal mucosa Well Well þþþ þþ negative (10%) negative (10%)
UW-SCC33P Supraglottis Moderate Moderate þþþ þþþ negative (40%) negative (50%)
UW-SCC34P Buccal mucosa Moderate Moderate þþnegative (0%) negative (10%)
UW-SCC35P Aveolar ridge Well Moderate þþ þþ negative (10%) negative (20%)
UW-SCC36P Tonsil Moderate Moderate þpositive (90%) positive (80%) þþ16
Abbreviations: ND, not done; NE, not evaluable.
a
Correlation between p16 and Rb and between primary tumor HPV and Rb (Spearman correlation P¼0.002 and 0.034, respectively).
b
Correlation between p16 and p53 and between primary tumor HPV and p53 (Spearman correlation P¼0.085 and 0.005, respectively).
c
Conducted on cDNA generated from mRNA.
d
Conducted on total genomic DNA.
Head and Neck Cancer Tumorgrafts
www.aacrjournals.org Clin Cancer Res; 19(4) February 15, 2013 859
assessed by p16 staining of the primary tumor (i.e., surgical
specimen) in all but 1 patient. The primary pathologic
specimens of 8 patients were HPV positive by p16 staining;
6 on the basis of at least 70% of cells having greater than
equal to 2þp16 staining, and 2 on p16 staining scored as
positive versus negative at an outside institution for which
slides were not available for quantification of p16 positivity
(Table 2).
All tumorgrafts were assessed for HPV status using 4
distinct tests: (i) expression of p16, a surrogate marker for
HPV infection, was assessed by IHC showing 6 of 22
tumorgrafts positive for p16 expression (Fig. 1); (ii) qRT-
PCR for HPV-16 E6 and E7 RNA, showing 4 of 22 tumor-
grafts positive by qRT-PCR (Table 2, Fig. 2D); (iii) PCR-
based detection was used to identify the presence of HPV
DNA from total genomic DNA using 2 different set of
degenerate primers known to detect multiple mucosotropic
HPV genotypes (9), showing 7 of 22 tumorgrafts positive by
HPV-specific PCR (Table 2, Supplementary Fig. S1A); and,
(iv) Southern blot of total genomic DNA for HPV-16 and
HPV-18, the 2 most frequently associated papillomaviruses,
was conducted using a procedure that can detect down to
0.1 copy per cell, showing 4 of 22 cases positive (Table 2,
Supplementary Fig. S1B).
In addition, samples from tumorgrafts that were pos-
itive on E1 degenerate PCR were assessed by Sanger
sequencing to determine the specific HPV subtype. All
showed near 100% identity with HPV-16. One sample,
UW-SCC25P was faintly positive upon PCR with the
MY9/11 primers and showed a positive doublet product
by PCR using primers specific for E1. However, upon
sequencing, these products, they showed no identity to
any HPV subtype.
Differences between p16 staining of the primary surgical
specimen and resultant tumorgraft were seen in 4 cases. In 3
of these cases, a different cutoff value for HPV positivity
would have led to agreement between patient and tumor-
graft. For example, UW-SCC3 and UW-SCC12 showed 90%
of cells p16 positive in the primary and 30% of cells p16
positive in the tumorgraft. Alternatively, UW-SCC25 was
scored as p16-negative in the primary (10% of cells p16
positive) and p16-positive in the tumorgraft (95% of cells
p16 positive). One case, UW-SCC14N had strong p16
staining in the primary tumor, isolated cells positive for
p16 in the lymph node and no p16 positive cells in the
tumorgraft derived from the lymph node. Those tumors
with 4 of more passages were assessed for p16 staining in
each passage and showed no alteration in p16 staining over
time (Supplementary Table S2).
Molecular markers—p53, Rb
Because of their known importance in HPV-associated
malignancy and their central role as tumor suppressor
proteins, we assessed the tumorgrafts for p53 and Rb
expression by IHC. A wide range of staining for both
markers was identified (Fig. 1 and Table 2) with a strong
inverse correlation between p16 and p53 and between p16
and Rb detection (Spearman correlations P¼0.03 and P<
0.001, respectively) as would be expected on the basis of the
mechanism of HPV oncogenesis.
Response to therapy
As an initial effort to assess the utility of tumorgrafts for
therapeutic response, tumorgrafts from 3 different patients
were tested in detail for response to radiation, cisplatin, and
cetuximab (Fig. 3A–C). Tumorgrafts from each patient
exhibited notably distinct growth patterns and response
profiles to these 3 treatments. UW-SCC14N, which derived
from a metastatic cervical lymph node, displayed the fastest
growth rate and limited response to radiation (Table 3). In
contrast, the 2 tumorgrafts derived from primary lesions,
UW-SCC6 and UW-SCC22, despite very different overall
growth profiles (Fig. 3), responded well to radiation, show-
ing a 2-fold increase in time to tumor quadrupling. Both of
UW-SCC6
Patient tumorPatient p16P0 tumorgraftTumorgraft p16TrichromeRbp53
UW-SCC14N UW-SCC22ABC
Figure 1. Histopathologic features of 3 patient tumors and corresponding
tumorgrafts. UW-SCC6 (A), UW-SCC14N (B), and UW-SCC22 (C).
Shown are photomicrografts of primary tumor H&E and p16 IHC top 2
rows in addition to tumorgraft passage 1 H&E, p16 IHC, Trichrome, Rb,
and p53. Overall, a strong correlation between Rb and p53 staining was
observed (Spearman correlation P¼0.04). Images at 200.
Kimple et al.
Clin Cancer Res; 19(4) February 15, 2013 Clinical Cancer Research
860
these tumorgrafts responded more briskly to cetuximab
than to cisplatin, which was not the case for UW-SCC14N.
The patient donating tissue for UW-SCC6 presented with
a T2N2BM0 oropharyngeal cancer and was treated with
radiation and concurrent cisplatin. The patient developed
lung metastases approximately 17 months after primary
treatment, but remained controlled at the site of the primary
and in the neck. UW-SCC14N presented with T1N2B oro-
pharyngeal cancer and underwent an initial neck dissection
followed by radiation and concurrent cetuximab chemo-
therapy. The patient’s disease is controlled 17 months after
initial therapy. UW-SCC22 presented with a T4aN1M0 oral
cavity tumor that was initially treated with surgical resec-
tion. Approximately 2 months after surgery, the tumor
recurred at both the site of the primary disease and in the
neck. The patient was treated with radiation and concurrent
cisplatin. One year later, the patient presented with locally
recurrent disease at which time the biopsy for tumorgraft
was taken. The patient died several months later of pro-
gressive disease.
Figure 2. A, well-differentiated
primary tumor along with
consistency between passages. B,
overall correlation between patient
and tumorgraft differentiation
showing good agreement
(unweighted Kappa ¼0.72, Std error
¼0.13). C, scatterplot depicting time
from implantation to passage for the
initial implantation (P0) and the
second (P1) and third (P2) passages.
One-way ANOVA, P¼0.28. D, qRT-
PCR with primers specific for HPV-
16 E6 and HPV-16 E7 RNA show high
correlation in Cq values between E6
and E7 (HPV-negative, circles;
HPVþ, triangles), Pearson r¼0.96,
P<0.0001.
Passage 1Passage 2Passage 3Passage 4Passage 5 Passage 0
WellModPoor
Well Mod
Primary
UW-SCC22
710
270
0
300
P = 0.28
40
35
30
25
20
20 25 30 35 40
HPV– UW-SCC3
UW-SCC36
UW-SCC14
UW-SCC12
UW-SCC4
UW-SCC17
UW-SCC6
UW-SCC1
Positive control
Cq for HPV-16 E7
Cq for HPV-16 E6
HPV+
Days until passage
200
100
0
P0 P1 P2
14
Poor
AB
C
D
Tumor graft
Head and Neck Cancer Tumorgrafts
www.aacrjournals.org Clin Cancer Res; 19(4) February 15, 2013 861
Finally, early passage cell strains were generated directly
from patient tumors if additional cells remained after the
initial tumorgraft implantation, or from subsequent tumor-
graft passages. To date, 6 cell strains have been generated
and passaged. No difference in primary tumor characteristic
seems to be predictive of cell strain development, although
the power to detect such a difference is severely limited by
the number of strains. The cell strains from UW-SCC14N
and UW-SCC24 were tested in proliferation assays and have
shown in vitro response to both cisplatin and cetuximab
(Fig. 3D). Interestingly, a very good response to cisplatin
was seen in UW-SCC14N, a tumor that also had shown a
good in vivo response to cisplatin alone (Fig. 3B).
Discussion
Over several decades, many investigators have relied
upon tumor cell line and xenograft model systems for
testing novel therapeutics. However, these systems carry
significant limitations based on adaptation to growth in
tissue culture including upregulation of survival genes,
alterations in multidrug resistance genes, and often greater
similarities to other cultured cells than to the primary
tumors they were originally intended to represent (4). In
addition, only 5 HPV-positive head and neck squamous
cancer cell lines have been described to date (12–16),
significantly limiting our ability to investigate differences
between HPV-positive and HPV-negative head and neck
cancers. The tumorgrafts and cell strains described in the
current study represent a promising system under develop-
ment by which to investigate molecular alterations under-
lying the growth behavior of head and neck cancers, to serve
as a preclinical model system for testing novel therapeutics
either alone or in combination with radiotherapy. While we
would hope that this model system could someday play a
useful role in the selection of optimal therapy for person-
alized medicine, the mean time required for tumorgraft
establishment (nearly 4 months) precludes the use for
initial therapeutic selection. However, Hidalgo and collea-
gues have successfully used tumorgrafts to identify effica-
cious therapies following initial therapeutic failure (7).
We commenced the current work anticipating that only a
small fraction of patient tumors would grow successfully as
tumorgrafts. However, the early success observed to date
with 22 of 26 tumorgrafts has been highly gratifying, and
may in part reflect the rapid transfer of tumor directly from
patient to mouse within 1 hour, the use of matrigel to
Ctrl
Cetuximab
Cisplatin
Radiation
Chemotherapy
2 Gy Radiation
0 50 100 150 200
0
600
1,200
1,800
0 20406080100
0
500
1,000
1,500
0102030
0
800
1,600
2,400
Time postimplantation (d)
ABC
D
-8 -7 -6 -5 -4
0
50
100
[Cisplatin, mol/L]
UW-SCC14N
UW-SCC24
0-10 -9 -8 -7 -6
[Cetuximab, mol/L]
*
*
*
*
*
#
*
*
#
Relative proliferation
(% no drug)
0 -9
Tumor volume (mm3)
Figure 3. Effects of radiation,
cetuximab, and cisplatin in 3
tumorgrafts: UW-SCC6 (A), UW-
SCC14N (B), and UW-SCC22
(C). Treatments, started when
subcutaneously growing tumors
reached a volume of approximately
200 mm
3
, were administered twice
weekly for 2 weeks as described in
the methods. Growth curves
plotting the mean (SEM) tumor
volume over time are shown. For
each group, n¼8–10 mice with
dual tumors began treatment.
,P<0.01;
#
,P¼not significant.
D, early passage cell strains
developed from UW-SCC24 and
UW-SCC14N were assessed for
proliferative potential in the
presence of indicated drugs. Cell
number was estimated by CCK8
assay and standardized to the
maximum relative value of vehicle
control.
Table 3. Time to tumorgraft quadrupling (d)
under control conditions and after radiation
(2 Gy twice weekly 4), cisplatin (0.2 mg/kg
twice weekly 4), or cetuximab (2 mg/kg twice
weekly 4)
Ctrl Radiation Cisplatin Cetuximab
UW-SCC6 44 79 91.5 123
UW-SCC14 13 13 25 17.0
UW-SCC22 55 125 125 181
% TGD
(Mean SD)
169 53 209 14 247 84
Abbreviation: TGD, tumor growth delay, the mean percent-
age increase in time to tumor quadrupling for each treatment
relative to control.
Kimple et al.
Clin Cancer Res; 19(4) February 15, 2013 Clinical Cancer Research
862
facilitate establishment of the tumorgraft and/or the use of
NSG mice, which are highly immunodeficient, as recipients.
Much akin to the diverse clinical presentations of cancer
patients, we have observed considerable diversity in tumor
differentiation, primary site location, lymph node status,
tobacco history, and HPV status (Tables 1 and 2) in the
tumorgrafts. In addition, these tumorgrafts have been estab-
lished from patients undergoing well-defined clinical treat-
ments for which detailed outcome data are being carefully
collected. The high tumor take rate in our study (85%)
provides preliminary confidence that the process of tumor-
grafting itself is not a significant selective pressure. The
increased take rate of tumorgrafts from patients with lymph
node metastases (regardless of the site of tissue) suggests
that intrinsic factors reflecting the biology of the individual
tumors may play a role in influencing tumorgraft take rates.
Perhaps pooled analysis from multiple disease sites may
identify critical molecular alterations associated with
tumorgraft take rates.
Overall, we have observed strong histologic stability
across serial passage of a single tumorgraft. While additional
comparisons over multiple passages are needed to confirm
phenotypic stability, our experience to date suggests high
intrapatient fidelity in terms of tumor differentiation and
p16 status. An exception to this pattern is the change
observed in HPV-status in select patients. For example,
UW-SCC14N is p16 positive in the patient’s primary tumor,
but negative for all tests in the tumorgraft. On the other
hand, UW-SCC25 is p16 negative in the patient, but p16
positive in the tumorgraft. There are several possible
explanations for these differences. It may simply reflect a
difference in the percentage of cells staining p16-positive
suggesting a potential selection bias in our model system.
Alternatively, we and others have previously described p16-
positive, HPV-negative tumors (10, 17, 18). It is unclear in
our model whether the loss of p16 expression represents
loss of episomal HPV-genomes, false-positive testing, alter-
native molecular pathway activation, or coincident tumor
development. In most cases, p16 status and degenerate PCR
were in agreement, those cases with discrepancies may
represent false-positive results as a third test for high- and
low-risk HPV using degenerate primers failed to detect a
product in these cases. This seems to predominantly reflect
differences in the percentage of cells staining p16-positive,
thus may not represent a true difference in biology. We and
others have previously described both p16-negative but
HPV-positive cases, as well as p16-positive but HPV negative
cancers (10, 17, 18). We did, however, identify the expected
correlation between HPV-positivity and low p53 and low
Rb (Table 2). Interestingly, not all groups confirm this
expected correlation between HPV-positive HNC and p53
expression intensity (18–20) suggesting that there may be
variation in expression of E6 and consequently incomplete
p53 degradation; alternatively, mutations in p53 may be
present in a subset of HPVþHNC patients resulting in
variable p53 expression.
Preclinical validation of therapeutic targets and response
profiling remains an expensive and time-intensive process.
There is considerable concern that human cancer cell lines,
either in vitro or as tumor xenografts, often show limited
ability to predict patient response to cancer therapy (21).
This may reflect the tremendous selection pressure required
to grow human tumor cells in artificial tissue culture systems
with adherence to plastic ware and/or reliance on culture
media. The primary tumorgraft system described in this
report reflects a systematic effort to more faithfully preserve
molecular, genetic, architectural, and treatment response
characteristics of the original human tumor specimen.
These HNSCC tumorgrafts may prove useful not only for
the investigation of radiation and chemotherapy response
profiles but also to uncover distinctions between HPVþand
HPVtumors with regard to growth characteristics and
response to conventional as well as new molecular thera-
pies. Only systematic investigation over time will confirm if
these tumorgraft model systems can prove consistently
more faithful and predictive of true clinical response and
outcome. A potential limitation of this model system is the
use of immunodeficient mice necessary to enable tumor-
graft growth. It has been suggested that the presence of an
intact immune response in HPVþHNC results in improved
tumor control (22). However, the published data also
suggest that cytotoxic therapies can result in tumor control
even in the absence of an intact immune system.
An important component of the current studies is the
development of cell strains. These nonimmortalized early
passage cells may more faithfully represent patient response
patterns; to date they show a high correlation with tumor-
graft response in our hands (Fig. 3). Successful establish-
ment of cell strains may provide a less expensive and more
efficient system in which to evaluate or screen therapeutic
regimens. While additional validation is necessary, our
results thus far suggest that cell strains may provide a
powerful adjunct to the human tumorgrafts.
In conclusion, we have described a large panel of human
head and neck cancer squamous cell carcinoma tumorgrafts
that reflect both HPV-positive and HPV-negative tumors.
The tumorgrafts display a spectrum of differentiation typical
of clinical histopathologic specimens, and a high degree of
consistency between original patient tumor and tumorgraft.
The combined use of early passage cell strains and tumor-
grafts provides a powerful system for investigating novel
therapeutics, combination therapies, and for testing
hypotheses of mechanisms of therapeutic response and
resistance in human head and neck cancer.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: R.J. Kimple, P. Harari, R.Z. Yang, P.F. Lambert
Development of methodology: R.J. Kimple, P. Harari, R.Z. Yang, E.A.
Armstrong
Acquisition of data (provided animals, acquired and managed patients,
provided facilities, etc.): R.J. Kimple, P. Harari, R.Z. Yang, E.A. Armstrong,
G.C. Blitzer, M.A. Smith, L.D. Lorenz, D. Lee, D. Yang, T.M. McCulloch, G.K.
Hartig
Analysis and interpretation of data (e.g., statistical analysis, biosta-
tistics, computational analysis): R.J. Kimple, R.Z. Yang, B.J. Soriano, M.
Yu, L.D. Lorenz, P.F. Lambert
Head and Neck Cancer Tumorgrafts
www.aacrjournals.org Clin Cancer Res; 19(4) February 15, 2013 863
Writing, review, and/or revision of the manuscript: R.J. Kimple, P.
Harari, R.Z. Yang, L.D. Lorenz, D. Yang, T.M. McCulloch, P.F. Lambert
Administrative, technical, or material support (i.e., reporting or orga-
nizing data, constructing databases): R.J. Kimple, A.D. Torres, R.Z. Yang,
M.A. Smith, D. Yang, G.K. Hartig, P.F. Lambert
Study supervision: P. Harari, P.F. Lambert
Grant Support
This study was supported by UWCCC pilot grant to P. Lambert and P.
Harari and by R01 CA 113448-01(P. Harari), R01 DE017315 (P. Lam-
bert), and U01 CA141583 (P. Lambert). R. Kimple is supported by the
UW Kaye Fellowship in Head and Neck Cancer Research, K99
CA160639-01, Radiological Society of North American Research Fellow
Grant, and AACR/Bristol Myers Squibb Fellowship in Clinical Cancer
Research.
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 in accordance with 18 U.S.C. Section 1734 solely to indicate
this fact.
Received August 22, 2012; revised November 8, 2012; accepted November
26, 2012; published OnlineFirst December 18, 2012.
References
1. Ferlay J, Shin HR, Bray F, Forman D, Mathers C, Parkin DM. Estimates
of worldwide burden of cancer in 2008: GLOBOCAN 2008. Int J Cancer
2010;127:2893–917.
2. Gillison ML, Koch WM, Capone RB, Spafford M, Westra WH, Wu L,
et al. Evidence for a causal association between human papillom avirus
and a subset of head and neck cancers. J Natl Cancer Inst 2000;92:
709–20.
3. Lassen P. The role of Human papillomavirus in head and neck cancer
and the impact on radiotherapy outcome. Radiother Oncol 2010;95:
371–80.
4. Gillet JP, Calcagno AM, Varma S, Marino M, Green LJ, Vora MI, et al.
Redefining the relevance of established cancer cell lines to the study of
mechanisms of clinical anti-cancer drug resistance. Proc Natl Acad Sci
U S A 2011;108:18708–13.
5. Uva P, Lahm A, Sbardellati A, Grigoriadis A, Tutt A, de Rinaldis E.
Comparative membranome expression analysis in primary tumors and
derived cell lines. PLoS One 2010;5:e11742.
6. DeRose YS, Wang G, Lin YC, Bernard PS, Buys SS, Ebbert MT, et al.
Tumor grafts derived from women with breast cancer authentically
reflect tumor pathology, growth, metastasis and disease outcomes.
Nat Med 2011;17:1514–20.
7. Hidalgo M, Bruckheimer E, Rajeshkumar NV, Garrido-Laguna I, De
Oliveira E, Rubio-Viqueira B, et al. A pilot clinical study of treatment
guided by personalized tumorgrafts in patients with advanced cancer.
Mol Cancer Ther 2011;10:1311–6.
8. Sivanand S, Pena-Llopis S, Zhao H, Kucejova B, Spence P, Pavia-
Jimenez A, et al. A validated tumorgraft model reveals activity of
dovitinib against renal cell carcinoma. Sci Transl Med 2012;4:
137ra75.
9. Snijders PJ, van den Brule AJ, Schrijnemakers HF, Snow G, Meijer CJ,
Walboomers JM. The use of general primers in the polymerase chain
reaction permits the detection of a broad spectrum of human papil-
lomavirus genotypes. J Gen Virol 1990;71 (Pt 1):173–81.
10. Schache AG, Liloglou T, Risk JM, Filia A, Jones TM, Sheard J, et al.
Evaluation of human papilloma virus diagnostic testing in oropharyn-
geal squamous cell carcinoma: sensitivity, specificity, and prognostic
discrimination. Clin Cancer Res 2011;17:6262–71.
11. Grant SW, Kyshtoobayeva AS, Kurosaki T, Jakowatz J, Fruehauf JP.
Mutant p53 correlates with reduced expression of thrombospondin-1,
increased angiogenesis, and metastatic progression in melanoma.
Cancer Detect Prev 1998;22:185–94.
12. Tang AL, Hauff SJ, Owen JH, Graham MP, Czerwinski MJ, Park JJ,
et al. UM-SCC-104: a new human papillomavirus-16-positive cancer
stem cell-containing head and neck squamous cell carcinoma cell line.
Head Neck 2012;34:1480–91.
13. Zhao M, Sano D, Pickering CR, Jasser SA, Henderson YC, Clayman
GL, et al. Assembly and initial characterization of a panel of 85
genomically validated cell lines from diverse head and neck tumor
sites. Clin Cancer Res 2011;17:7248–64.
14. Ferris RL, Martinez I, Sirianni N, Wang J, Lopez-Albaitero A, Gollin SM,
et al. Human papillomavirus-16 associated squamous cell carcinoma
of the head and neck (SCCHN): a natural disease model provides
insights into viral carcinogenesis. Eur J Cancer 2005;41:807–15.
15. Gwosdz C, Balz V, Scheckenbach K, Bier H. p53, p63 and p73
expression in squamous cell carcinomas of the head and neck and
their response to cisplatin exposure. Adv Otorhinolaryngol 2005;62:
58–71.
16. Steenbergen RD, Hermsen MA, Walboomers JM, Joenje H, Arwert F,
Meijer CJ, et al. Integrated human papillomavirus type 16 and loss of
heterozygosity at 11q22 and 18q21 in an oral carcinoma and its
derivative cell line. Cancer Res 1995;55:5465–71.
17. El-Naggar AK, Westra WH. p16 expression as a surrogate marker for
HPV-related oropharyngeal carcinoma: a guide for interpretative rel-
evance and consistency. Head Neck 2012;34:459–61.
18. Harris SL, Thorne LB, Seaman WT, Hayes DN, Couch ME, Kimple RJ.
Association of p16(INK4a) overexpression with improved outcomes in
young patients with squamous cell cancers of the oral tongue. Head
Neck 2011;33:1622–7.
19. Kumar B, Cordell KG, Lee JS, Worden FP, Prince ME, Tran HH, et al.
EGFR, p16, HPV Titer, Bcl-xL and p53, sex, and smoking as indicators
of response to therapy and survival in oropharyngeal cancer. J Clin
Oncol 2008;26:3128–37.
20. Fury MG, Drobnjak M, Sima CS, Asher M, Shah J, Lee N, et al. Tissue
microarray evidence of association between p16 and phosphorylated
eIF4E in tonsillar squamous cell carcinoma. Head Neck 2011;33:
1340–5.
21. de Bono JS, Ashworth A. Translating cancer research into targeted
therapeutics. Nature 2010;467:543–9.
22. Spanos WC, Nowicki P, Lee DW, Hoover A, Hostager B, Gupta A, et al.
Immune response during therapy with cisplatin or radiation for human
papillomavirus-related head and neck cancer. Arch Otolaryngol Head
Neck Surg 2009;135:1137–46.
Kimple et al.
Clin Cancer Res; 19(4) February 15, 2013 Clinical Cancer Research
864
