ORIGINAL ARTICLE

CELL CYCLE AND VIRAL AND IMMUNOLOGIC PROFILES OFHEAD AND NECK SQUAMOUS CELL CARCINOMA ASPREDICTABLE VARIABLES OF TUMOR PROGRESSION

Massimo Ritta, MD,1 Marco De Andrea, PhD,1,2 Michele Mondini, PhD,2,3 Jasenka Mazibrada, MD,1

Carlo Giordano, MD,4 Giancarlo Pecorari, MD,4 Massimiliano Garzaro, MD,4 Vincenzo Landolfo, MD,4

Marina Schena, MD,5 Luigi Chiusa, MD,6 Santo Landolfo, MD1

1 Department of Public Health and Microbiology, Medical School of Turin, Via Santena 9, 10126 Turin, Italy.E-mail: santo.landolfo@unito.it2 Department of Clinical and Experimental Medicine, Medical School of Novara, Novara, Italy3 NoToPharm srl, Via Ribes 5, 10010 Colleretto Giacosa (TO), Italy4 Department of Clinical Physiopathology, Medical School of Turin, Turin, Italy5 Department of Medical Oncology, Azienda Ospedaliera San Giovanni Battista, Turin, Italy6 Department of Biomedical Sciences and Human Oncology, Section of Pathology, Medical School of Turin,Turin, Italy

Accepted 18 July 2008Published online 15 December 2008 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/hed.20977

Abstract: Background. The aim of this study was to deter-

mine whether the aberrant expression of cell-cycle or immune-

response markers together with human papillomavirus (HPV)

positivity impacts patient survival in different head and neck

squamous cell carcinoma (HNSCC) subsets.

Methods. A total of 59 HNSCC specimens were analyzed for

expression of cell cycle and proliferation markers, and macro-

phage infiltration. HPV was evaluated by polymerase chain reac-

tion and DNA sequencing.

Results. The HPV presence in oropharynx carcinoma was

associated with survival advantage. Low Ki67 expression was

associated with favorable outcome in oropharynx and oral cavity

carcinoma. A more favorable outcome was associated with low

cyclin E expression in larynx carcinoma and with low p53

expression in squamous cell carcinoma (SCC) of the oral cavity.

A direct correlation between macrophage infiltration and

tumor proliferation index was observed irrespective of the tumor

subset.

Conclusions. The assessment of proliferation, viral, and

immunologic profiles may be crucial to finding beneficial treat-

ments for the different HNSCC subsets. VVC 2008 Wiley Periodi-

cals, Inc. Head Neck 31: 318–327, 2009

Keywords: head and neck squamous cell carcinoma; human

papillomavirus; proliferation index; p53; macrophage infiltration

Head and neck squamous cell carcinoma(HNSCC) is the eighth most common malignancyin the world, with approximately 620,000 patientsdiagnosed with cancer of the oral cavity, naso-pharynx, and larynx each year.1 The disease ischaracterized by local tumor aggressiveness,

Correspondence to: S. Landolfo

Contract grant sponsors: Compagnia di San Paolo, Regione Piemonte,MIUR, Fondazione CRT.

VVC 2008 Wiley Periodicals, Inc.

318 Cell Cycle and Viral and Immunologic Profiles of HNSCC HEAD & NECK—DOI 10.1002/hed March 2009

early recurrence, and a high frequency of secondprimary tumors.2

The majority of HNSCCs are differentiatedsquamous cell carcinomas (SCCs), which occur inthe oral cavity, oropharynx, hypopharynx, andlarynx. HNSCCs are etiologically heterogeneous,with 1 subset attributable primarily to humanpapillomavirus (HPV) infection and another toalcohol and tobacco use. Most HPV-associatedHNSCCs tend to occur in the oropharynx, withthe highest distribution in the tonsils. HPV16 isthe predominant genotype in head and necktumors, with different prevalences among the var-ious head and neck sites. These subsets are clini-cally and molecularly distinct, and these distinc-tions extend to patient prognosis. As the role ofHPV16 in HNSCC becomes better understood,these HPV16-related cancers are becomingincreasingly recognized as a biologically distinctsubgroup of HNSCC with a characteristic clinicalprofile. Indeed, the presence of HPV16 in HNSCChas been correlated with improved survival andmay serve as a useful biomarker for prognosis.3–8

In contrast, the clinical behavior of HPV-nega-tive HNSCC is heterogeneous, and the most fre-quent molecular alteration carried by patients withHPV-negative HNSCC is p53 mutation, which hasbeen correlated with a poor response to radiother-apy and chemotherapy.9 Thus, it is essential toinvestigate the molecular alterations in advancedHNSCCs thatmay help to identify patients who areat the greatest risk of progression.

The tumor suppressor protein, p53, is a centralprocessing unit that receives specific signals,interprets combinatorial inputs, and determinesregulatory output. Signaling to p53 occurs byphysical interactions with protein complexes,which covalently modify specific amino acids ofp53, mediate subcellular localization of p53, tar-get p53 to specific chromatin sites, and/or dictatep53’s cellular concentration.10 These p53-regu-lated functions oppose tumor development andprogression, and their dysfunction in a tumor cellis generally caused by direct mutation of p53 ordisruption of its signaling interactions.11 In oppo-sition to p53, which acts as a brake in the cellcycle, is cyclin E, which is rate limiting for the G1/S-transition in the cell cycle of mammalian cells.Cyclin E binds to and activates cyclin-dependentkinase 2 (CDK2), initiating the processes requiredfor the G1-to-S-phase transition.12 The ability toenter S-phase requires cyclin E or cells will arrestat G1-phase, defining the important role of sub-strates, phosphorylated by cyclin E/CDK2, in pro-

moting DNA synthesis.13 Many investigationspoint to the relevance of cyclin E alteration inbreast cancer. The cyclin E gene is amplified insome breast cancer cell lines and it has been demon-strated that this amplification can result in several-fold overexpression of cyclin EmRNA that is consti-tutively expressed across all phases of the cell cycle.Such constitutive overexpression and activation ofcyclin E results in the deregulation of cell cycle pro-gression and chromosomal instability.11,14 Geneticabnormalities of components of the cell cycle, suchas p53 and p16/INK4A, are very common in mosttumor types, including HNSCC, and alteredexpression of cyclins and CDKs has also been iden-tified. However, information on cyclin E expressionin HNSCC has not been provided yet, hamperingthe possibility to exploit cyclin E as a marker inassociationwith tumor progression.15–17

Genetic and cell biology studies indicate thattumor growth is not just determined by accumula-tion of genetic alterations by malignant cancercells but also by tumor stroma. Numerous hostcells, such as inflammatory cells, endothelial cells,and fibroblasts, are recruited to and activated inthemicroenvironment of a developing tumor. Sub-sequent reciprocal interactions between thesestromal cells, their mediators, and geneticallyaltered ‘‘initiated’’ cells are indispensable for car-cinogenesis. Infiltration of leukocytes into theneoplastic microenvironment is a common featureof many epithelial malignancies. The attractedmonocytes differentiate into tumor-associatedmacrophages (TAMs) at the tumor site and arethought to produce various cytokines that pro-mote tumor progression.18,19 Consistent with thishypothesis, we have recently observed that thenumber of infiltrating macrophages was signifi-cantly associated with progression to malig-nancy.20

This study was undertaken to investigate theexpression pattern of cell cycle regulators, such asp53 and cyclin E, the proliferation activity by Ki-67 immunostaining, and HPV status in 59advanced HNSCC in an effort to identify factorswith potential clinical implications. The numberof TAMs present in the tumors was also semiquan-titatively determined.

PATIENTS AND METHODS

Patients and Samples Collection. Fifty-nine casesof primary SCCs of the head and neck region diag-nosed and treated at the Department of Otolarin-

Cell Cycle and Viral and Immunologic Profiles of HNSCC HEAD & NECK—DOI 10.1002/hed March 2009 319

gology and the Department of Clinical Oncology,San Giovanni Battista Hospital, Turin, between1995 and 2004 were included in the study. Theclinical and pathological staging and identifica-tion of anatomical sites of the lesions were basedon the Union Internationale Contre le CancerTNM classification of malignant tumors (2003).All patients underwent surgery as a firstapproach. As shown in Table 1, primary tumorssites were 22 SCCs in the oropharynx, 25 in theoral cavity, and 12 in the larynx. Six tumors werehistologically well differentiated, 27 were moder-ately differentiated, and 26 were poorly differenti-ated. Samples were grouped by TNM classifica-tion as follows: 10 in T1, 17 in T2, 14 in T3, 18 inT4; 14 in N0, 12 in N1, 24 in N2, 6 in N3, 3 in Nx;and 57 in M0, 2 in M1. The mean age of thepatients was 59 years (range, 37–75), the men towomen ratio was 48 to 11, and the follow-up inter-val ranged from 5 to 167months.

Immunohistochemistry. Serial sections (2 lmthick) were cut from each selected block and dew-axed. Antigen unmasking was performed bymicrowaving in a conventional pressure cooker, inwhich the slides were placed for 30 minutes in a10 mM citrate buffer at pH 6.0. To abolish endoge-

nous peroxidase activity, sections were immersedfor 10 minutes in 3% hydrogen peroxide solution,buffered in phosphate-buffered saline (PBS) 13 atpH 7.3, and then incubated sequentially with pro-tein blocking agent (UltraTech HRP Streptavidin-Biotin Universal Detection System; Immunotech,Marseille, France) to reduce nonspecific binding.The slides were then incubated with the primarymonoclonal antibodies (mAb) (mouse) for 1.5hours at room temperature in a humid chamber.The biotinylated secondary antibody was applied,followed by incubation with streptavidin–horse-radish peroxidase complex (Immunotech). As anegative control, a separate set of slides was incu-bated with PBS 13 instead of primary mousemAbs. The immunologic reactions were developedat room temperature with 3,30-diamino-benzidinetetrahydrochloride (DAB) solution (Roche, Mann-heim, Germany), counterstained with Mayer’s he-matoxylin, dehydrated, and finally mounted withEUKITT (Bioptica, Milan, Italy). The followingantisera were used: (1) Ki67 (clone MIB-1, work-ing dilution 1:100), (2) CD68 (clone PG-M1, 1:50dilution) provided by DAKO Cytomation, Den-mark; (3) p53 (Santa Cruz Biotechnology, SantaCruz, CA; clone DO-1, working dilution 1:5000);(4) cyclin E (Santa Cruz Biotechnology; cloneHE12, working dilution 1:200).

Interpretation of Immunohistochemical Staining.

The intensity of immunohistochemical stainingwas evaluated in 5 areas of the slide sections. Posi-tive cells were stained dark brown. Ki67, cyclin E,and p53 expression was evaluated by determiningthe percentage of squamous cancer cells showingnuclear immunoreactivity, with cutoff values thatdiffered depending on the type of antibody used.Inflammatory cells positive for CD68 immuno-staining showed cytoplasmic reactivity within thetumor mass. Immunoreactivity was referred to aslow or high. CD68 staining was detected as cyto-plasmic immunoreactivity in the inflammatorycells within the tumormass.

HPV Testing. The presence and typing of HPVDNAwas assayed by highly sensitive polymerasechain reaction (PCR) and sequencing. DNA wasextracted from 15-lm paraffin-embedded tissuesections using a commercial extraction kit (NucleoSpin Tissue, Machery-Nagel, Germany), accord-ing to the manufacturer’s specified protocol.DNA integrity was confirmed by amplificationof the b-globin gene. To increase the sensitivityof HPV detection, nested PCR assays were

Table 1. Characteristics of the head and neck squamous cell

carcinoma (HNSCC) patients included in the study.

Characteristics

Oropharynx,

no. (%)*Oral cavity,

no. (%)*Larynx,

no. (%)*

Number of patients 22 25 12

Median age 59 y 59 y 58 y

Men/women 20/2 16/9 12/0

T classification

T1 6 (27.2) 3 (12) 1 (8.3)

T2 10 (45.5) 4 (16) 3 (25)

T3 2 (9.1) 7 (28) 5 (41.7)

T4 4 (18.2) 11 (44) 3 (25)

N classification

N0 5 (22.7) 6 (24) 3 (25)

N1 6 (27.3) 3 (12) 3 (25)

N2 7 (31.8) 12 (48) 5 (41.7)

N3 2 (9.1) 3 (12) 1 (8.3)

Nx 2 (9.1) 1 (4) 0

M classification

M0 22 (100) 25 (100) 10 (83.3)

M1 0 0 2 (16.7)

Histological grade

Well differentiated 1 (4.5) 4 (16) 1 (8.3)

Moderately

differentiated

8 (36.4) 12 (48) 7 (58.3)

Poorly

differentiated

13 (59.1) 9 (36) 4 (33.3)

*Number of patients (percentage of patients) from each anatomic site.

320 Cell Cycle and Viral and Immunologic Profiles of HNSCC HEAD & NECK—DOI 10.1002/hed March 2009

performed using MY09/MY11 as the outer, andGP51/GP61 as the inner primers.21 A 50 lL reac-tion mixture consisted of 1 lM of each primer, 300ng of the extracted sample, 13 Taq Buffer (10 mMTris-HCl, pH 8.3, 50 mM KCl, 1 mM MgCl2, and0.1% gelatin), 200 lM of each of the 4 dNTPs, and1 unit of Taq DNA Polymerase (Sigma, St. Louis,MO). The PCR products were verified by directsequencing with the GP51/GP61 primers on aDNA sequencer (PRIMM, Milan, Italy). HPV typewas determined on the basis of >90% homologywith HPV sequences deposited in GenBank usingthe BLAST network service at NCBI (http://www.ncbi.nlm.nih.gov/blast/Blast.cgi).

The negative controls were samples withwater replacing target DNA in the reaction mix-ture. The positive control was DNA from cervicalcancer cells positive for high-risk HPV (Casky).Standard precautions concerning spatial separa-tion of pre- and post-PCR steps, aliquoting ofreagents, and single use of scalpels for processingtissue specimens were strictly followed.

Statistical Analysis. A 1-way analysis of variance(ANOVA) with Bonferroni post test was used fordetermining the statistical significance of differ-ences between the disease groups for Ki67, p53,cyclin E, and CD 68 expression.

Correlations were obtained using the Pearsontest when it could be assumed that data weresampled from Gaussian populations, and theSpearman test for nonparametric correlations. Alltests were 2-tailed and a probability value p < .05was considered statistically significant.

We estimated survival curves using theKaplan–Meier method, and compared them usinga 2-sided log-rank test. Analyses were carried outwith GraphPad Prism 4.03 software (GraphPadSoftware, SanDiego, CA; www.graphpad.com).

RESULTS

HPV DNA Detection. The overall frequency ofHPV infection was 50% in the oropharynx SCC,36% in the oral cavity SCC, and 58.3% in the lar-ynx SCC. These percentages, although slightlyhigher, are still in the range reported by otherinvestigators.4,5,22 HPV type was identified asHPV16 in all tumors, with the sole exception of 2laryngeal carcinomas that were positive forHPV6. HPV infection did not correlate with tumordifferentiation, node status, or with the other clin-ical features evaluated (Tables 1 and 2).

Immunohistochemical Studies. The expression ofKi67, cyclin E, and p53 as proliferation and cellcycle markers, and of CD68 as macrophage infil-tration marker, were examined by immunohisto-chemistry in a series of clinical samples ofHNSCC. The results of the immunohistochemicalassays are summarized in Table 2 and illustratedin Figure 1. In normal epithelia, Ki67-positivecells were mostly located in the basal cell layer,whereas in the tumor mass they were distributedin all levels of the epithelium. Remarkably, in low-grade SCCs, positive Ki67 staining was detectedonly in cells at the periphery of the tumor cellnests (Figure 1, panel A), such that the central

Table 2. Evaluation of the biological parameters referring to tumor anatomic sites.

Primary tumor sites

p valueOropharynx, no. (%)* Oral cavity, no. (%)* Larynx, no. (%)*

Number of patients 22 25 12

HPV infection 11 (50) 9 (36) 7 (58.3)

Ki67 (cutoff: 25%)

Low (%) 3 (13.7) 1 (4) 3 (25)

High (%) 19 (86.3) 24 (96) 9 (75) .1716

Cyclin E (cutoff: 20%)

Low (%) 8 (36.4) 8 (32) 6 (50)

High (%) 14 (63.6) 17 (68) 6 (50) .5666

p53 (cutoff: 15%)

Low (%) 8 (36.4) 9 (36) 2 (16.7)

High (%) 14 (63.6) 16 (64) 10 (83.3) .4347

CD68

Low (%) 9 (40.9) 9 (36) 8 (66.7)

High (%) 13 (59.1) 16 (64) 4 (33.3) .1983

*Number of patients (percentage of patients) from each anatomic site.

Cell Cycle and Viral and Immunologic Profiles of HNSCC HEAD & NECK—DOI 10.1002/hed March 2009 321

FIGURE 1. Immunohistochemical expression of Ki67, cyclin E, p53, and CD68 in low-expressing (left panels) or high-expressing (right

panels) head and neck squamous cell carcinoma (HNSCC). Immunohistochemical staining shows positive cells in brown color,

counterstained with hematoxylin (original magnification 3200). [Color figure can be viewed in the online issue, which is available at

www.interscience.wiley.com.]

322 Cell Cycle and Viral and Immunologic Profiles of HNSCC HEAD & NECK—DOI 10.1002/hed March 2009

keratinizing areas and adjacent tumor cells werenegative. Conversely, the staining was more dif-fuse in high-grade carcinomas (Figure 1, panel B).Overall, the proliferative activity, as defined byKi67 immunostaining, was significantly related totumor differentiation and histological diagnosis.Three of 22 (13.7%) oropharyngeal SCCs, 1 of25 (4%) oral SCCs, and 3 of 12 (25%) laryngealSCCs were considered negative. For Ki67 immu-nostaining, a significant association with a partic-ular anatomical site was not observed (p 5 .1716)(Table 2).

A different immunophenotype pattern wasobserved with the othermarkers employed. CyclinE immunostaining was considered positive whencyclin E expression was equal or greater than20% in all 3 anatomical sites. According to thiscutoff value, 14 of 22 samples (63.6%) in the oro-pharynx, 17 of 25 (68%) in the oral cavity, and 6 of12 in the larynx (50%) were above the cutoff of20% and turned out to be positive. Thus, no signif-icant correlation between cyclin E expression anda particular anatomical site was found (p5 .5666)(Table 2).

p53 immunostaining was considered positivewhen its expression was 15% or more in all ana-tomical sites (Figure 1, panels E and F). Based onthis cutoff value, 14 of 22 (63.6%) oropharyngealSCC and 16 of 25 (64%) oral SCC samples werepositive. In contrast, 10 of 12 (83.3%) laryngealSCC samples were positive for p53 immunostain-ing. No significant correlation between p53expression and a particular anatomical site wasfound (p5 .4347) (Table 2).

Primary tumor macrophage content has beendemonstrated to be a strong predictor of tumoraggressiveness in HNSCC. To assess if macro-phage infiltration could affect clinical outcome, weexamined the expression of the macrophagemarker CD68. All sample tissues studied con-tained cells positive for the macrophage markerCD68. TAMs were predominantly located in thetumor stroma, although they were also observedbetween tumor cells. TAMs were distributed het-erogeneously in the tumor tissues and were oftennumerous in advanced-stage carcinomas. Stainedmacrophages were also observed in normal tis-sues surrounding the tumor mass, only in thestroma, and their number was lower. The CD68immunophenotype results were classified as highin 13 of 22 (59.1%) and in 16 of 25 (64%) tumorsderived from the oropharynx and oral cavity,respectively. In contrast, only 4 of 12 (33.3%)laryngeal SCCs were classified as high, indicating

an inverse correlation between macrophage infil-tration and this particular anatomical site. How-ever, this association was not significant (p 5.1983) (Table 2).

Next, potential associations between the pres-ence/absence of HPV and expression of cell cyclemarkers were examined. Among the 59 tumors,there was a strong association between HPV pres-ence and reduced Ki67 expression (p5 .0026) (Ta-ble 3). All tumors that were HPV negative weremore likely to have upregulated Ki67 (32 of 32,100%), whereas of the 27 HPV-positive tumors, 20tumors had upregulated Ki67 (74%) and 7 haddownregulated Ki67 (26%) expression. In con-trast, there was no association between the pres-ence/absence of HPVand cyclin E and p53 expres-sion, away from the tumor site.

Statistical Analysis. Patients with oropharyngealcarcinomas testing HPV16-positive had a betterclinical outcome than those testing HPV-negative.The Kaplan–Meier analysis showed that patientswith HPV16-positive tumors had significantlyimproved overall survival compared with thosewith HPV-negative tumors (p5 .0211) (see Figure2). In contrast, in carcinomas of the oral cavityand larynx, HPV positivity seemed to be a nega-tive prognostic factor, although the significance ofthe Kaplan–Meier analysis was low (p 5 .2802and p5 .5439, respectively) (data not shown).

Ki67 expression in carcinomas derived fromthe oropharynx tended to be inversely correlatedwith prognosis, i.e., patients with cancer express-ing Ki67 at a level< 25%were found to have a sig-nificantly better clinical outcome than those withKi67 at expression levels � 25% (p 5 .0188)(Figure 3, panel A). A very favorable outcome wasobserved in the HPV-positive oropharynx subset

Table 3. Histological markers in head and neck squamous cell

carcinoma (HNSCC) patients grouped by HPV status.

HPV2, no. (%)* HPV1, no. (%)*Fisher’s

exact test

Ki67

<25% 0 7 (11.9)

�25% 32 (54.2) 20 (33.9) p 5 .0026

Cyclin E

<20% 10 (17) 12 (20.3)

�20% 22 (37.3) 15 (25.4) p 5 .4182

p53

<15% 11 (18.6) 7 (11.9)

�15% 21 (35.6) 20 (33.9) p 5 .5758

*Number of patients (percentage of patients).

Cell Cycle and Viral and Immunologic Profiles of HNSCC HEAD & NECK—DOI 10.1002/hed March 2009 323

in association with a Ki67 immunostaining scorebelow the cutoff value of 25%. The survival benefitobserved in these patients, amounting to a 2.48

relative mortality reduction at 5 years, occurredirrespective of tumor stage. In contrast, all HPV-negative patients displayed an immunostainingscore above 25% with a survival curve that didnot overcome 5 years. Similarly, an immunostain-ing score below 25% in SCC of oral cavity showeda strong correlation with a better follow-up (p 5.0240) (Figure 3, panel B). In contrast, no associa-tion between Ki67 positivity below 25% and a sur-vival benefit in patients with carcinoma arising inthe larynx was found (p 5 .2467) (data notshown).

A similar investigation of cyclin E expressionyielded different results. HPV-positive SCCsturned out to have a similar level of cyclin Eexpression when compared with HPV-negativeSCC (p 5 .4182) (Table 3), irrespective of tumorsite. Low cyclin E expression (<20%) was notassociated with increased overall patient survivalin either the oropharynx or the oral cavity (p 5.5308 and p 5 .5594, respectively) (data notshown). In contrast, cyclin E expression below20% in SCC of the larynx showed a strong associa-

FIGURE 2. Cumulative prognostic value of human papillomavirus

(HPV) positivity for patients with squamous cell carcinoma (SCC)

of the oropharynx, expressed as probability of overall survival.

FIGURE 3. Cumulative prognostic value of the indicated markers for patients affected by squamous cell carcinoma (SCC) of the oro-

pharynx (A), oral cavity (B, D), and larynx (C), expressed as probability of overall survival. Only significant survival curves are shown,

as other biomarkers had no statistically detectable impact on survival.

324 Cell Cycle and Viral and Immunologic Profiles of HNSCC HEAD & NECK—DOI 10.1002/hed March 2009

tion with a more favorable outcome (5-year sur-vival, p5 .0046) (Figure 3, panel C).

p53 nuclear staining was rare in normal epi-thelium, where its expression was generally re-stricted to isolated basal and parabasal epithelialcells. In testing the prognostic role of p53 status,significant results were obtained when SCC oforal cavity were investigated. Kaplan–Meier anal-ysis showed a median survival time of 57.5months for p53 expression below the cutoff pointof 15%, versus a median survival time of

16 months for p53 above 15%. The difference wasstatistically significant (p 5 .0236) (Figure 3,panel D). No such correlation of p53 expressionwas demonstrated with oropharynx- and larynx-derived tumors (p 5 .2600 and p 5 .8673, respec-tively) (data not shown).

Finally, when a comparative analysis of theexpression rate for the above cell proliferationmarkers was attempted, an inverse correlationbetween p53 and cyclin E expression was found (p5 .0421, r 5 2.2655) (Figure 4, panel A), inde-pendent of the tumor site. Moreover, in accordwith previous results, when a correlation betweenthe macrophage infiltration as measured by CD68staining and the tumor proliferation index asmeasured by Ki67 staining was evaluated, adirect association between the 2 markers’ expres-sion was observed. The median macrophagecounts were significantly higher in those tumorsexpressing Ki67 above the cutoff point (>25%) in-dependently of the anatomical site, suggestingthat the number of infiltrating macrophages wassignificantly associated with progression to malig-nancy (Figure 4, panel B).

DISCUSSION

HNSCCs, including tumors derived from orophar-ynx, larynx, and oral cavity, are etiologically het-erogeneous with 1 subset derived from the oro-pharynx attributable primarily to HPV infectionand another to alcohol and tobacco abuse. All sub-sets are clinically and molecularly distinct, andthese distinctions extend to patient prognosis,making it difficult to assess their malignancy andpredict the outcome of treatment. Molecularmarkers defining certain genotypes and pheno-types, and representing tumor subgroups withmore homogenous behavior must thus be foundfor the different HNSCC types.

In this study, we have attempted to identifydifferent molecular markers, including HPVDNA, to be used as prognostic factors for each tu-mor subgroup. Consistent with results reportedby other investigators, our studies demonstratethat HPV-positive oropharyngeal cancers com-prise a distinct molecular and pathologic diseaseentity that is causally associated with HPV infec-tion and has a markedly improved prognosis. Weexamined the prevalence of HPV in SCC of theoropharynx, larynx, and oral cavity. In addition,the clinical outcome of the patients was reviewedand correlated with the presence of HPV, the pro-

FIGURE 4. (A) Correlation between p53 and cyclin E expres-

sion in head and neck squamous cell carcinoma (HNSCC) (p 5.0421, r 5 2.2655). The line represents the calculated regres-

sion line. (B) Correlation of immunohistochemical expression of

CD68 with Ki67. The levels of CD68 expression showed posi-

tive correlation with Ki67 immunostaining (p 5 .0128).

Cell Cycle and Viral and Immunologic Profiles of HNSCC HEAD & NECK—DOI 10.1002/hed March 2009 325

liferation index of tumor cells (Ki67 and cyclin E),and p53 expression rate.

The Kaplan–Meier analysis showed thatpatients with HPV-positive oropharyngeal carci-nomas had a better overall survival than thosewith HPV-negative carcinomas. Such an associa-tion has not been found in patients harboringSCCs of both the oral cavity and larynx. Themechanisms underlying the better clinical out-come of HPV-positive oropharynx SCC remainunexplained. A few studies have concluded thatthe reason for a better prognosis could beexplained by an enhanced radiosensitivity ofthese HPV-positive SCC in comparison to HPV-negative tumors.1,23,24 It was suggested that theinteraction between E6 and p53 does not result inits full functional abrogation as compared withmutated TP53 showing p53 overexpression. Thiswould justify an increased radiosensitivity sus-tained by a functioning p53 protein.25 Whateverthe explanation, HPV-positive oropharyngealSCC may constitute a distinct molecular andpathologic disease entity that is causally associ-ated with HPV infection and has a markedlyimproved prognosis.

A different pattern in association with theprognostic markers employed emerged whenHNSCCs of different anatomical sites were ana-lyzed. First, the presence of HPV16 in oral cavityand laryngeal SCCs did not correlate with a betterclinical outcome. Second, low p53 expression inthe oral cavity was found to be a predictive factorfor good prognosis. At 96 months, 62% of patientswith p53 expression <15% were still alive,whereas none of the patients with p53 expression>15%were alive at 24months (p5 .0236).

A high proliferation rate has been correlatedwith aggressive behavior of tumors at various an-atomical sites. In HNSCCs, several data suggestthat cell proliferation indices are reliable and re-producible indicators of tumor aggressiveness.26

Consistent with previous findings, we observedthat SCC of the oropharynx and oral cavity withlower proliferation index displayed a better over-all survival (p 5 .0188 and p 5 .0240, respec-tively). This correlation was not found for laryn-geal SCC (p5 .2467).

Although elevated cell proliferation activity is1 of the most remarkable features of HNSCC,abnormalities in cell cycle regulation have yet tobe thoroughly investigated in this type of tumor.HNSCC tissue microarray analysis demonstratedmultiple alterations at various checkpoints of cellcycle progression with a median overall survival

and time-to-progression longer in patients withcyclin A-expressing tumors. Moreover, positiveexpression of cyclin E in tumors was also associ-ated with an increased median time-to-progres-sion.15 However, no correlation of these expres-sionmarkers with a particular tumor location wasobserved. In this study, no significant correlationbetween cyclin E expression and a particular ana-tomical site was found irrespectively from thepresence of HPV. However, low cyclin E expres-sion levels (<20%) in SCC of the larynx showed astrong association with a more favorable outcome(5-year survival, p5 .0046), suggesting that cyclinE is a good predictor of patient response in thisparticular anatomical site. The cyclin E gene hasbeen found to be amplified and the cyclin E proteinconstitutively expressed in several breast cancercell lines.27,28 Consistent with our conclusion, ele-vated levels of cell cycle protein cyclin E assessedby immunohistochemistry analysis have beenassociated with a poor prognosis following breastcancer and malignant ovarian germ celltumors.29,30 Moreover, high levels of cyclin E asmeasured by western blotting were the strongestindependent factor in predicting survival follow-ing breast cancer.31

Infiltration of leukocytes into the neoplasticmicroenvironment is a common feature of manyepithelial malignancies. Extensive analysis ofhuman tumor samples has revealed that abun-dance of innate immune cells, in particular macro-phages, correlates with angiogenesis and poorprognosis. In line with these findings, when a cor-relation between macrophage infiltration asmeasured by CD68 staining and tumor prolifera-tion index as measured by Ki67 staining was eval-uated, a direct association between the 2 markers’expression was observed, suggesting that thenumber of infiltrating macrophages in HNSCCwas significantly associated with progression tomalignancy.

REFERENCES

1. Li W, Thompson CH, O’Brien CJ, et al. Human papillo-mavirus positivity predicts favourable outcome for squa-mous carcinoma of the tonsil. Int J Cancer 2003;106:553–558.

2. Chen R, Aaltonen LM, Vaheri A. Human papillomavirustype 16 in head and neck carcinogenesis. Rev Med Virol2005;15:351–363.

3. Azzimonti B, Pagano M, Mondini M, et al. Altered pat-terns of the interferon-inducible gene IFI16 expressionin head and neck squamous cell carcinoma: immunohis-tochemical study including correlation with retinoblas-toma protein, human papillomavirus infection and prolif-eration index. Histopathology 2004;45:560–572.

326 Cell Cycle and Viral and Immunologic Profiles of HNSCC HEAD & NECK—DOI 10.1002/hed March 2009

4. Gillison ML, Koch WM, Capone RB, et al. Evidence for acausal association between human papillomavirus and asubset of head and neck cancers. J Natl Cancer Inst2000;92:709–720.

5. Ragin CC, Taioli E. Survival of squamous cell carcinomaof the head and neck in relation to human papillomavi-rus infection: review and meta-analysis. Int J Cancer2007;121:1813–1820.

6. Begum S, Gillison ML, Nicol TL, Westra WH. Detectionof human papillomavirus-16 in fine-needle aspirates todetermine tumor origin in patients with metastatic squa-mous cell carcinoma of the head and neck. Clin CancerRes 2007;13:1186–1191.

7. GillisonML, LowyDR. A causal role for human papillomavi-rus in head and neck cancer. Lancet 2004;363:1488–1489.

8. D’souza G, Kreimer AR, Viscidi R, et al. Case-controlstudy of human papillomavirus and oropharyngeal can-cer. N Engl J Med 2007;356:1944–1956.

9. Perrone F, Suardi S, Pastore E, et al. Molecular andcytogenetic subgroups of oropharyngeal squamous cellcarcinoma. Clin Cancer Res 2006;12:6643–6651.

10. Bode AM, Dong Z. Post-translational modification of p53in tumorigenesis. Nat Rev Cancer 2004;4:793–805.

11. Barton MC, Akli S, Keyomarsi K. Deregulation of cyclinE meets dysfunction in p53: closing the escape hatch onbreast cancer. J Cell Physiol 2006;209:686–694.

12. Ohtsubo M, Theodoras AM, Schumacher J, Roberts JM,Pagano M. Human cyclin E, a nuclear protein essentialfor the G1-to-S phase transition. Mol Cell Biol 1995;15:2612–2624.

13. Koff A, Giordano A, Desai D, et al. Formation and acti-vation of a cyclin E-cdk2 complex during the G1 phase ofthe human cell cycle. Science 1992;257:1689–1694.

14. Akli S, Van Pelt CS, Bui T, et al. Overexpression of thelow molecular weight cyclin E in transgenic mice inducesmetastatic mammary carcinomas through the disruptionof the ARF-p53 pathway. Cancer Res 2007;67:7212–7222.

15. Rodriguez-Pinilla M, Rodriguez-Peralto JL, Hitt R, et al.Cyclin A as a predictive factor for chemotherapy res-ponse in advanced head and neck cancer. Clin CancerRes 2004;10:8486–8492.

16. Califano J, Westra WH, Meininger G, Corio R, KochWM, Sidransky D. Genetic progression and clonal rela-tionship of recurrent premalignant head and necklesions. Clin Cancer Res 2000;6:347–352.

17. Okami K, Reed AL, Cairns P, et al. Cyclin D1 amplifica-tion is independent of p16 inactivation in head and necksquamous cell carcinoma. Oncogene 1999;18:3541–3545.

18. Mantovani A, Sica A, Sozzani S, Allavena P, Vecchi A,Locati M. The chemokine system in diverse forms of

macrophage activation and polarization. Trends Immu-nol 2004;25:677–686.

19. Martinez FO, Gordon S, Locati M, Mantovani A. Tran-scriptional profiling of the human monocyte-to-macro-phage differentiation and polarization: new moleculesand patterns of gene expression. J Immunol 2006;177:7303–7311.

20. Mazibrada J, Ritta M, Mondini M, et al. Interactionbetween inflammation and angiogenesis during differentstages of cervical carcinogenesis. Gynecol Oncol 2008;108:112–120.

21. Haws ALF, He Q, Rady PL, et al. Nested PCR with thePGMY09/11 and GP5(1)/6(1) primer sets improves det-ection of HPV DNA in cervical samples. J Virol Methods2004;122:87–93.

22. Syrjanen S. Human papillomavirus (HPV) in head andneck cancer. J Clin Virol 2005;32(Suppl 1):S59–S66.

23. Mellin H, Friesland S, Lewensohn R, Dalianis T, Munck-Wikland E. Human papillomavirus (HPV) DNA in tonsil-lar cancer: clinical correlates, risk of relapse, and sur-vival. Int J Cancer 2000;89:300–304.

24. De Petrini M, Ritta M, Schena M, et al. Head and necksquamous cell carcinoma: role of the human papillomavi-rus in tumour progression. New Microbiol 2006;29:25–33.

25. Sisk EA, Soltys SG, Zhu S, Fisher SG, Carey TE, Brad-ford CR. Human papillomavirus and p53 mutational sta-tus as prognostic factors in head and neck carcinoma.Head Neck 2002;24:841–849.

26. Pich A, Chiusa L, Navone R. Prognostic relevance of cellproliferation in head and neck tumors. Ann Oncol2004;15:1319–1329.

27. Keyomarsi K, Conte D Jr, Toyofuku W, Fox MP. Deregu-lation of cyclin E in breast cancer. Oncogene 1995;11:941–950.

28. Keyomarsi K, Pardee AB. Redundant cyclin overexpres-sion and gene amplification in breast cancer cells. ProcNatl Acad Sci U S A 1993;90:1112–1116.

29. Chappuis PO, Donato E, Goffin JR, et al. Cyclin Eexpression in breast cancer: predicting germline BRCA1mutations, prognosis and response to treatment. AnnOncol 2005;16:735–742.

30. Potemski P, Kusinska R, Watala C, Pluciennik E, Bed-narek AK, Kordek R. Cyclin E expression in breast can-cer correlates with negative steroid receptor status,HER2 expression, tumor grade and proliferation. J ExpClin Cancer Res 2006;25:59–64.

31. Keyomarsi K, Tucker SL, Buchholz TA, et al. Cyclin Eand survival in patients with breast cancer. N Engl JMed 2002;347:1566–1575.

Cell Cycle and Viral and Immunologic Profiles of HNSCC HEAD & NECK—DOI 10.1002/hed March 2009 327