J Cancer Res Clin Oncol (2010) 136:1039–1048

DOI 10.1007/s00432-009-0749-3

ORIGINAL PAPER

Oral tongue squamous cell carcinoma: recurrent disease is associated with histopathologic risk score and young age

Marilena Vered · Dan Dayan · Alex Dobriyan · Ran Yahalom · Bruria Shalmon · Iris Barshack · Lev Bedrin · Yoav P. Talmi · Shlomo Taicher

Received: 2 March 2009 / Accepted: 9 December 2009 / Published online: 7 January 2010© Springer-Verlag 2010

AbstractPurpose To apply the Brandwein-Gensler et al.’s histo-pathologic risk score (RS) system and to evaluate its impacton locoregional recurrence and overall survival in a seriesof cases of oral tongue cancer, along with variables ofpatient age and margin status.Methods Sections of the resection specimens (N = 50)were submitted to a RS assignment of three components:the worst pattern of invasion, lymphocytic inWltration andperineural invasion. Risk scores of 0–2 were classiWed aslow-to-intermediate and RSs ¸3 were classiWed as highwith respect to recurrence and survival. Margins were con-sidered as “clean” if the tumor was ¸5 mm away fromthem, otherwise they were deWned as “positive”. Patients·60 years were considered “young” and those >60 years“old”. Kaplan–Meier survival analysis with univariate andCox multivariate regression model with stepwise forwardselection tests were used.

Results Univariate analysis showed that locoregionalrecurrence was negatively inXuenced by high RSs(P = 0.011), “young” age (P = 0.027) and positive margins(P = 0.027). Multivariate analysis revealed that the risk ofrecurrence was increased by high RSs (hazard ratio 11.14;P = 0.022) and “young” age (hazard ratio 3.41; P = 0.022).“Young” patients with high RSs had a higher frequency ofrecurrence rate compared to “young” patients with low-to-intermediate scores (P = 0.008) and “old” patients withlow-to-intermediate and high RSs (P = 0.012 andP = 0.011, respectively).Conclusions The histopathologic RS can serve to identifya subgroup of patients <60 years who have a high recur-rence rate of oral tongue cancer, irrespective of the marginstatus.

Keywords Tongue cancer · Histopathologic risk score · Age · Margin status · Recurrence

Introduction

A considerable number of histopathologic features havebeen proposed as key issues in predicting recurrence andsurvival in patients with oral squamous cell carcinoma(OSCC). Included are surgical margin status (Loree andStrong 1990; Spiro et al. 1999; Liao et al. 2008), tumor sizeas part of the TNM system (American Cancer Society2005), tumor thickness (Clark et al. 2006; Fukano et al.1997; Mohit-Tabatai et al. 1986; Shah and Lydiatt 1995;Spiro et al. 1986), tumor grade (Byers et al. 1998), andlymphovascular (Jones et al. 2009) and perineural invasion(PNI) (Fagan et al. 1998; Soo et al. 1986). When analyzedseparately, however, each of these factors yielded con-tradicting conclusions. A newer and more practical

M. Vered (&) · D. DayanDepartment of Oral Pathology and Oral Medicine, School of Dental Medicine, Tel Aviv University, 69978 Tel Aviv, Israele-mail: mvered@post.tau.ac.il; lmy@netvision.net.il

M. Vered · B. Shalmon · I. BarshackInstitute of Pathology, The Chaim Sheba Medical Center, Tel Hashomer, Israel

A. Dobriyan · R. Yahalom · S. TaicherDepartment of Oral and Maxillofacial Surgery, The Chaim Sheba Medical Center, Tel Hashomer, Israel

L. Bedrin · Y. P. TalmiDepartment of Otorhinolaryngology-Head and Neck Surgery, The Chaim Sheba Medical Center, Tel Hashomer, Israel

123

1040 J Cancer Res Clin Oncol (2010) 136:1039–1048

approach was that of multiparameter analyses, which com-bined diVerent histopathologic features (Martinez-Gimenoet al. 2005; Yuen et al. 2000) but it, too, was not free of dis-agreement.

One of the most challenging and compelling issues forboth clinicians and pathologists was the observation thatthere was still a recurrence rate of 25% even among small(T1 stage) OSCC with adequate resection margins: this wasunexpected when only the currently accepted clinical andhistopathologic parameters were considered Brandwein-Gensler et al. (2005). On these grounds, an updated,modiWed multiparameter analysis was proposed byBrandwein-Gensler et al. (2005). Their approach was basedon a three-tiered scoring system of three histopathologicalparameters consisting of worst pattern of invasion (WPOI),lymphocytic inWltrate (LI) and PNI, and the three valueswere added together to yield the patient’s risk score (RS).This has been shown to be signiWcantly predictive of recur-rence and overall survival (OS) in OSCC. That study alsoshowed an advantage of the histopathologic RS over thesurgical margin status in regard to recurrence and survival.Thus, according to Brandwein-Gensler et al. (2005) whensmall (T1) carcinomas with clean margins were re-assessedaccording to their proposed RS, all such cases fulWlled thecriteria for being at high risk for developing recurrentdisease.

Another important issue that recently became the focusof many investigations is the tendency of OSCC to developin patients younger than 60 years, which is the mean age forthis disease in the Western world (Conway et al. 2006;Muller et al. 2008). The incidence of the disease hasremained stable in patients ¸60 years for the last 40 years,but there has been a Wve- to sixfold increase in patients·45 years over the same period (Annertz et al. 2002).Opinions in the English language literature are controver-sial as to whether OSCC in young patients is characterizedby a more aggressive biological behavior than in oldpatients (Annertz et al. 2002; Garvello et al. 2007; Joneset al. 1989; Kuriakose et al. 1992; Popovitzer et al. 2004;Sarkaria and Harari 1994).

The present study was designed to further validate thehistopathologic RS system proposed by Brandwein-Gensleret al. (2005) and to apply it to a series of patients diagnosedwith oral cancer at a single site—the oral tongue, with spe-cial emphasis on patient age. The impact of RS along withthat of margin status and patient age on disease recurrenceand patient survival was examined.

Patients and methods

The medical records of all patients with oral tongue SCC(OTSCC) diagnosed, treated and followed-up at the Depart-

ment of Oral and Maxillofacial Surgery and the Departmentof Otorhinolaryngology-Head and Neck Surgery at theChaim Sheba Medical Center, Tel Hashomer, Israel,between 1981 and 2006, were retrieved for the presentstudy. Patients included were those who had undergone pri-mary resection and a concomitant neck dissection andreceived no prior treatment for their OTSCC. The studypopulation was further limited to those with a complete setof the permanent histopathologic slides of the resectionspecimens. This study was conducted with the approval ofthe medical center’s Helsinki Committee.

The reviewed clinical parameters included demographicdata, clinical tumor size (the “T” variable from the TNMstaging system), adjuvant radiotherapy (XRT) and patientoutcome. Patients were further classiWed as “young” if theywere <60 years and “old” if they were ¸60 years. ThiscutoV point of 60 years is the mean age for diagnosingOSCC (Conway et al. 2006; Muller et al. 2008). The clinicaloutcomes were measured by two endpoints: locoregionaldisease control expressed by locoregional recurrence (LR)and OS. Recurrence was calculated as the time between thedate of diagnosis and Wrst sign of treatment failure at theprimary tumor site, at the site of cervical metastases orboth. OS included patients alive and free of disease andpatients alive with disease at the last follow-up visit. Alldisease-free patients had a minimum of 18 months offollow-up.

Histopathological assessment

All the hematoxylin and eosin-stained slides of the resec-tion specimens were examined and scored according to thesystem established by Brandwein-Gensler et al. andincluded the WPOI, LI and PNI. The Wrst two parameterswere assessed at the tumor/host interface. Pattern of inva-sion of type 1 was allocated to a tumor with broad pushingborders; type 2 to a tumor with broad pushing “Wngers” orseparate large tumor islands; type 3 was assigned to a tumorwith invading islands >15 cells per island; type 4 to a tumorwith invading islands <15 cells per island, up to single cellinvasion; and type 5 was for tumor satellites of any sizewith ¸1 mm distance of intervening normal tissue (notWbrosis) at the tumor/host interface. The WPOI was takenas the highest score present, even if it was only focal. LIwas classiWed according to pattern 1, which was assignedwhen a continuous and dense rim of lymphoid tissue waspresent; pattern 2, patches of dense lymphoid tissue werepresent but was discontinuous along the interface; and pat-tern 3, which was assigned for limited or no lymphocyticresponse. PNI, deWned as carcinoma pursuing along orwithin a nerve, was classiWed as involving large nerves of¸1 mm in diameter or small nerves, <1 mm in diameter.

123

J Cancer Res Clin Oncol (2010) 136:1039–1048 1041

The score assignment to the WPOI parameter deWned types1, 2 and 3 as score 0, type 4 as score 1 and type 5 as score 3(brand). For LI, pattern 1 was assigned a score of 0; pattern2, a score of 1; and pattern 3, a score of 3. For PNI, when noPNI was found, the case was assigned a score of 0, smallnerves, a score of 1 and large nerves, a score of 3. The RSfor each case was calculated as the sum of the highestscores given to each of the parameters after examination ofall the slides. Figures 1, 2, and 3 exemplify how the assess-ment of the RS for three diVerent tumors was performed. ARS of “0” was considered as representing a low risk forrecurrence and good survival probability. RSs of 1 or 2indicated an intermediate risk for recurrence and survival,while RSs between 3 and 9 were indicative of a high riskfor recurrence and poor survival. The RSs were evaluated

by two expert pathologists (MV and BS) who wereunaware of the clinical outcomes.

During the selection process of the cases for the presentstudy, special emphasis was given to the size of the sam-pling in order to obtain an accurate assessment of the surgi-cal margin status. We selected only T1 and T2 tumorswhich were blocked entirely. Larger tumors, T3–T4, had tobe sampled extensively, with multiple representative slidesfrom all margins of the specimen. Margin status was deW-ned as “clean” when the tumor was ¸5 mm away from theinked surgical margin. All the others (“close” = tumor<5 mm and “frankly positive” = tumor involving the surgi-cal margin) were considered as “positive”. This deWnitionfor the tumor-margin distance was aimed at all the aspectsof the resection specimen, both at the mucosal (epithelial)and submucosal (deep) margins. All 50 study patientsunderwent surgery with one exception of a patient whodenied surgery at the primary site. The RS analysis of that

Fig. 1 Example of a tumor assigned a risk score of 1. This was calcu-lated by summing the score assignment of the lymphocytic inWltrate(LI), worst pattern of invasion (WPOI) and perineural invasion (PNI).LI, which was dense and continuous, was given a score of 0 (a). WPOI,which was of type 4 pattern (inWltration of tumor islands <15 cells orsingle cell invasion), was given a score of 1 (b, arrows). No PNI wasfound in the tumor, therefore this variable was given a score of 0 (a bar1 mm, b bar 200 �m)

Fig. 2 The risk score given to the tumor was 3. WPOI had a type 2 pat-tern (broad pushing “Wngers”), which corresponded to a score of 0 (a).LI was given a score of 3 (limited lymphocytic inWltrate) (b). PNI wasassigned a score of 0 in the absence of perineural invasion (a bar500 �m, b bar 200 �m)

123

1042 J Cancer Res Clin Oncol (2010) 136:1039–1048

patient was performed on two large biopsy samples andyielded a high score of 5. That patient was treated by radi-cal neck dissection and XRT. In all, 34 patients were givenXRT, seven patients had combined XRT and chemotherapyand one patient was given only chemotherapy.

Statistical analysis

The Kaplan–Meier method was applied for survival analy-sis. The statistical signiWcance was evaluated using thelog-rank test. Univariate analysis was used to analyze theprognostic potential of the variables age, gender, T-stage,WPOI, LI, PNI, RS and margin status with respect to LRand OS. A Cox multivariate regression model with step-wise forward selection (likelihood ratio) was performed.Cox hazards ratios with 95% conWdence intervals and Pvalues were obtained for each covariate analyzed. Thecovariates analyzed included RSs (high vs. low-to-interme-

diate), age (·60 years vs. >60 years.), and margin status(“positive” vs. “clean”). In all analyses, P values <0.05were considered to be statistically signiWcant. Statisticalanalyses were performed by the SPSS software, version 15(Chicago, IL, USA).

Results

The strictly observed study inclusion criteria yielded a sam-ple of 50 patients.

General data (univariate analysis, Table 1)

Age. Mean 57.9 § 15.2 years (range 20–80 years); 24(48%) patients were classiWed as “young” and 26 (52%) as“old”.

Gender. Females: 24, males: 26.

Fig. 3 The risk score for the illustrated tumor was 9. LI was scored as 3 (no/limited lympho-cytic inWltrate) (a). WPOI was of type 5 pattern consisting of satel-lite tumor islands 2.7 mm away from main tumor (dashed line) (b). The deepest island of the main bulk of the tumor is indi-cated by a short arrow and the satellite tumor focus is indicated by a long arrow. The deepest island and the satellite tumor are illustrated at a higher magniWca-tion in c(1) and c(2), respec-tively. Therefore, WPOI was given a score 3. PNI involved nerve Wbers with a diameter of 1 mm (dashed line), and was scored as 3 (thin and thick arrows show intraneural and perineural involvement, respectively) (d); (a, b bar 1 mm; c, d bar 200 �m)

123

J Cancer Res Clin Oncol (2010) 136:1039–1048 1043

T-stage. T1: 25 (50%); T2: 17 (34%); T3: 3 (6%); T4: 5(10%). For statistical purposes, T3 and T4 tumors weregrouped.

Follow-up period. 3–230 months, mean 46.7 § 41.4months; during this period of time, 33 (66%) patients werefree of recurrence; 37 (74%) patients were disease-free; 10(20%) were alive with disease; and three patients (6%) diedof unrelated causes.

LR was found to be negatively inXuenced by “young”age (P = 0.027). Figure 4a shows that the 5-year LR was53% in the “young” versus 23.53% in the “old” group. Theanalyzed parameters had no statistically signiWcant impacton OS.

Histopathologic analysis (univariate analysis, Table 1)

Interobserver variations were non-signiWcant (P > 0.05).

Worst pattern of invasion. 47 (94%) patients had scoresof 0 and 1; 3 (6%) had a score of 3.

Lymphocytic inWltrate. 27 (53%) patients had scores of 0and 1 (grouped for statistical purposes), and 23 (47%)patients had a score of 3.

Perineural invasion. 11 (22%) patients had a score of 0,19 (38%) had a score of 1, and 20 (40%) patients had ascore of 3.

Risk score. 12 (24%) patients had low-to-intermediateRSs (0–2) and 38 (76%) patients had high RSs (¸3).

Margin status. 32 (65%) patients had “clean” marginsand 17 (35%) patients had “positive” margins.

LR was negatively inXuenced by PNI of large nerveWbers (vs. small nerve Wbers, P = 0.038 and no PNI,P = 0.027), high RS (p = 0.011) and “positive” margin sta-tus (P = 0.027). Figure 4b shows that 48.25% patients withhigh RSs had recurrent disease in contrast to 8.24% patientswith low-to-intermediate RSs. The analyzed parametershad no statistically signiWcant impact on OS.

The impact of combined age-RS parameters on LR (Fig. 4c)

Given the signiWcant inXuence of age and RS on LR, ananalysis of these parameters in combination was per-formed. For these calculations, we divided the studygroup into four subgroups: “young” and low-to-intermedi-ate RS (scores 0–2); “young” and high RS (scores ¸3);“old” and low-to-intermediate RS; and, “old” and highRS. The most unfavorable impact on LR was found in the“young”, high RS subgroup, where 71.77% of the patientsexperienced recurrence 5 years after diagnosis, comparedto 29.41% in the “old”, high RS subgroup (P = 0.011),14.12% in the “young”, low-to-intermediate RS(P = 0.008) and 0% in the “old”, low-to-intermediate RSsubgroup (P = 0.012).

Multivariate analysis (Table 2)

On multivariate analysis, the high RSs (¸3) and “young”age (·60 years) were found to be unfavorable prognosticfactors for the 5-year LR. The relative risks of recurrencefor patients with high RSs was 11.14 (P = 0.022); “young”patients had a relative risk of 3.41 (P = 0.022). Theseresults are relative to patients with low-to-intermediate RSs(0–2) and “old” age (>60 years). Margin status was statisti-cally insigniWcant with respect to its impact on LR.

Distribution of patients according to margin status, tumor size (T), radiotherapy and recurrent disease (Table 3)

The Wnal pathological reports of the T1 lesions were that 18had “clean” margins and 7 had “positive” margins. Ten of

Table 1 Univariate analysis by risk parameters

All bold values are statistically signiWcanta One patient denied surgery

Parameter (N) LR OS

P value P value

Age 0.027 0.293

·60 years (24)

>60 years (26)

Gender 0.814 0.658

Females (24)

Males (26)

T-stage

T1 (25) versus T2 (17) 0.816 0.139

T1 (25) vs T3 + T4 (8) 0.905 0.159

T2 (17) versus T3 + T4 (8) 0.958 0.839

WPOI score

0 (5) versus 1 (42) 0.969 0.713

1 (42) versus 3 (3) 0.238 0.774

3 (3) versus 0 (5) 0.412 0.808

LI score 0.242 0.297

0 +1 (27)

3 (23)

PNI score

0 (11) versus 1 (19) 0.357 0.243

3 (20) versus 1 (19) 0.038 0.299

3 (20) versus 0 (11) 0.027 0.085

Risk score 0.011 0.087

¸3 (38)

0–2 (12)

Margin statusa 0.027 0.264

“Positive” (17)

“Clean” (32)

123

1044 J Cancer Res Clin Oncol (2010) 136:1039–1048

the 18 patients were not treated by XRT, and two (20%) ofthem developed recurrent disease. Among the eight patientstreated by XRT, one (12%) developed recurrent disease.Irrespective of margin status and XRT, all the patients withrecurrent disease (N = 7) in the T1 group had a high RS(¸4) and, with one exception, they were classiWed in the“young” age category.

Among the patients with T2 tumors, 12 had “clean” mar-gins and 4 had “positive” margins. Ten of the patients with“clean” margins were treated by adjuvant XRT, and 4 ofthem (40%) nevertheless had recurrent disease. Irrespectiveof margin status and XRT, all patients with recurrent dis-ease (N = 6) in the T2 group had a high RS (5 or 7). ThreeT2 patients were classiWed as “young” and three as “old”.

Table 2 Cox multivariate analysis for LR

All bold values are statistically signiWcant

Parameter P Hazard ratio

95% ConWdenceinterval

RS 0.022 1

Low-to-intermediate 11.147 1.426–87.123

High

Age

“Old” (>60 years) 0.022 1

“Young” (·60 years) 3.412 1.197–9.803

Margin status 0.57 – –

“Positive”

“Clean”

Fig. 4 Kaplan–Meier analysis for LR per a age (P = 0.027), b RS (P = 0.011), and c combined age-RS (“young”, RS ¸ 3 vs. “old”, RS ¸ 3,P = 0.011; “young”, RS ¸ 3 vs. “young”, RS < 3, P = 0.008; “young”, RS ¸ 3 vs. “old”, RS < 3, P = 0.012)

123

J Cancer Res Clin Oncol (2010) 136:1039–1048 1045

Discussion

The present sample of OTSCC cases (N = 50) was smallerthan the equivalent cohort examined by Brandwein-Gensleret al. (2005) (N = 85), but the main results were essentiallysimilar. In terms of margin status, in both their study andours, “clean” margins were deWned as a distance of ¸5 mmbetween tumor and the inked surgical margins of the speci-men. While we demonstrated an association between mar-gin status and LR, albeit only by univariate analysis,Brandwein-Gensler et al. (2005) found no association at all.The histopathologic interpretation of the terms “clean” and“positive” margins has been changing over the years, withsubsequent diVerent and often contradicting conclusions asto how this variable might aVect the probability of recurrentdisease and survival (Brandwein-Gensler et al. 2005; Liaoet al. 2008; Loree and Strong 1990; McMahon et al. 2003;Spiro et al. 1999). One feasible explanation for this incon-sistency may be supplied by a recent molecular study,which found chromosomal instability even in margins thatseemed to be “clean” microscopically, and in which the fol-low-up results showed recurrent disease in a considerablenumber of cases (O’Regan et al. 2008). According to theresults of the present study—which, on the whole, supportthose of Brandwein-Gensler et al. (2005)—it seems that theconcept of multiple clinical and histopathologic parametersshould prevail when assessing patient outcomes. Consider-ing only a single parameter, even one as important as mar-gin status, might not reXect the biological complexity of thetumors. This viewpoint is further supported by our observa-tion that patients with small OTSCC tumors (T1) that have“clean” margins, who were not given XRT according to thewidely accepted treatment protocols, experienced a 20%rate of recurrence, which was compatible with the 25%reported by Brandwein-Gensler et al. (2005). In both stud-ies, the association between high RSs and tumor recurrencein spite of “clean” margins was obvious. This further vali-dates the potential of the histopathologic RS to become animportant factor when considering administration of adju-vant radiation therapy. Interestingly, we found a recurrencerate of 12 and 40% for T1 and T2 tumors, respectively,among patients who received XRT. The tumors in all thesecases were assigned high RSs, which may suggest that this

system has an impact not only on the level of decision-mak-ing on adjuvant radiation therapy, but also on the need toenhance standard treatment protocols with novel anti-neo-plastic approaches.

An additional aspect to consider is the selection site ofthe oral tongue. The rationale for this stems from the obser-vation that it is known as being the most common oral sitefor squamous cell carcinoma (Regezi et al. 2008). Analo-gous to the diVerential histopathologic and clinical featuresof the potentially malignant oral lesions, it was assumedthat oral cancer sequelae may also be site-dependent. Thereare several studies which investigated this assumption, buttheir conclusions were contradictory (Bell et al. 2007;Sathyan et al. 2006). In the study by Brandwein-Gensleret al., recurrent disease was analyzed per oral site and itshowed a wide range, i.e., between 12 and 100%, amongdiVerent locations (Brandwein-Gensler et al. 2005). Thus,conclusions on the frequency of recurrence derived fromstudies that included multiple oral sites may not reXect anaccurate picture and could be the reason for the contradic-tory results, emphasizing the need for future studies to beconducted separately for diVerent oral sites. These shouldbe aimed to investigate whether anatomic diVerences mighthave an impact on tumor spread and recurrence.

Our results of both univariate and multivariate statisticalanalyses showed that age is an independent prognosticatorfor recurrence. We were prompted to examine this parameterin light of the growing concern about the four- to sixfoldincrease in the incidence of OSCC among patients youngerthan the typical average age of 60 years at diagnosis duringthe past four decades (Annertz et al. 2002; Muller et al.2008). Furthermore, the oral tongue was found by someauthors to be the most common site for the developmentof oral cancer (Conway et al. 2006; Muller et al. 2008).Although there is no speciWc consensus on how to deWnethe age range of “young” patients with OSCC, individualswho are ·45 years are generally considered appropriate tofulWll this criterion. Controversial results have beenreported in the English language literature on the diseaseoutcomes among these patients, with some arguing formore aggressive course and devastating outcomes(Garvello et al. 2007; Jones et al. 1989; Kuriakose et al.1992; Popovitzer et al. 2004; Sarkaria and Harari 1994),

Table 3 Distribution of patients according to margin sta-tus, tumor size (T), radiotherapy (XRT) and recurrence (Rec) (N = 49; one patient who denied surgery, developed recurrent disease)

“Clean” margins “Positive” margins

Rec Rec No Rec No Rec Rec Rec No Rec No Rec

+XRT ¡XRT +XRT ¡XRT +XRT ¡XRT +XRT ¡XRT

T1 (N = 25) 1 2 7 8 4 0 3 0

T2 (N = 16) 4 0 6 2 2 0 1 1

T3 (N = 3) 0 0 1 0 1 0 0 1

T4 (N = 5) 0 0 0 1 2 0 2 0

123

1046 J Cancer Res Clin Oncol (2010) 136:1039–1048

while others not Wnding signiWcant diVerences between“young” and “old” patients (Atula et al. 1996; Friendlanderet al. 1998; Goldstein and Irish 2005; Hart et al. 1999;Hyam et al. 2003; McMahon et al. 2003; Pitman et al.2000) and even reporting an improved clinical outcome in“young” compared to “old” patients (Annertz et al. 2002;Schantz and Yu 2002). For evaluating the parameter of age,we distributed the present series of patients into “young”and “old” using a cutoV of 60 years since we consider thatthis still represents the average age of the vast majority ofthe oral cancer patient population (Ries et al. 2008). Theoccurrence of the disease among young persons(<45 years), in spite of the alarming recent increase, is stillrare and does not exceed 5–7% (Muller et al. 2008; Shibo-ski et al. 2005). Our identiWcation of a subgroup of “young”patients with a signiWcantly higher frequency of recurrentdisease as determined by high histopathologic RSs (>3)may supply a feasible explanation for the observation that“young” and “old” patients may generally show similar dis-ease outcomes, although the disease is manifested by anexceptionally aggressive course in some “young” patients(Popovitzer et al. 2004). The possibility of diVerent patientproWles as a factor of age can be further explored fromaspects of molecular biology, etiology, comorbidity, andgender. A recent study on the alterations of p16 tumor sup-pressor gene and its corresponding protein product clearlyshowed that the nature of this gene mutation in the oraltongue carcinomas among young patients signiWcantlydiVered from that found in the base of the tongue in oldpatients (BergshoeV et al. 2008). Another contributing fac-tor to the proWle of the OSCC patient is the human papil-loma virus (HPV), in particular the high risk form HPV-16(Chaturvedi et al. 2008; D’Souza et al. 2007). It is of inter-est to note that the incidence of HPV-associated head andneck SCC appear to be dramatically increasing, while theclassic form (chronic exposure to tobacco and alcohol)appear to be decreasing in the United States, presumablydue to decreased smoking in conjunction with altered sex-ual practices (Weinberger et al. 2006). It is assumed thatpatients with HPV-associated head and neck SCC have abetter prognosis, because of the lack of the Weld canceriza-tion as in the classic form of SCC (Fakhry et al. 2008;Liang et al. 2008). However, in HPV-associated head andneck SCC the most common regions are the base of thetongue, the tonsillar region and the oropharynx (Soderberget al. 2008) and not the oral tongue—the focus of the cur-rent study. Furthermore, the incidence of HPV in oraltongue cancer in the “young” population (<45 year) wasfound to be low and is unlikely to constitute a signiWcantfactor in the rising trends of oral tongue cancer in this popu-lation (Liang et al. 2008). Comorbidity is a clinico-biologi-cal feature expected to aVect the course of the disease, thechoice of treatment and Wnally, the prognosis (Alho et al.

2007). Comorbid conditions were found to be a markedindependent risk factor for poor survival in head and neckSCC, including tongue cancer (Alho et al. 2007). Further-more, it was found in that study that the magnitude of thecomorbidity eVect was higher in “younger” (<65 years)than “older” patients. However, the main limitation inassessing comorbidity is the lack of a standardized indexfor head and neck cancer, in particular, for oral cancer(Reid et al. 2002; Ribeiro et al. 2000). With respect to thegender distribution of patients with tongue cancer, it is ofinterest to note that in spite of the fact that the male-to-female ratio is still generally estimated as 2:1 (Jamal et al.2009), an increased incidence among women has beenreported in several studies from diVerent parts of the world(Annertz et al. 2002; Pitman et al. 2000; Sathyan et al.2006). In these studies, a male-to-female ratio of 1.4:1 wasfound. Furthermore, there are a few studies, which reportedthat young women with tongue cancer outnumbered men(Jones et al. 1989; Popovitzer et al. 2004). The male-to-female ratio found in the present study (1.08:1) is wellwithin the range reported in previous works and may reXectan emerging change in the gender distribution. Gender-associated diVerences in tolerance to risk factors, hormonaland immunologic modiWcations have been suggested toexplain this change (Papageorge 2007). Consistent with thisnotion, some researchers examined the status of estrogenreceptors in cell lines of squamous cell carcinoma andfound a high level of expression of estrogen receptor-beta,whose function could be abrogated by an estrogen receptorantagonist (tamoxifen) (Ishida et al. 2007). However, cur-rently, there is no conclusive evidence to support tonguecancer in women as a separate entity. In light of these Wnd-ings, further studies to examine the multiparameter associa-tions between the RS system, age, and selected clinicalparameters and outcomes are needed and may ultimatelyaid in more appropriate management of these patients.

In summary, the present study applied a recently pro-posed histopathologic RS system on a series of patientsdiagnosed, treated and followed-up for squamous cell carci-noma of the oral tongue. This system can be routinelyapplied and is easily reproduced. It provides valuable initialprognostic information at a light microscopic level, prior tothe evaluation of the quality of the tumors on the molecularlevel. The main conclusion was that high RSs and “young”age are unfavorable prognostic factors for locoregional dis-ease control; in addition, using this, RS system enabled usto identify a subgroup of patients younger than 60 yearswho had a remarkably high frequency of recurrent disease.Furthermore, we showed that “clean” surgical margins andsmall tumor size (T1/T2) with high RSs were still associatedwith recurrence, irrespective of adjuvant XRT. Large-scalevalidation of the proposed RS system may well aid inobtaining better clinical results by considering modiWcations

123

J Cancer Res Clin Oncol (2010) 136:1039–1048 1047

in the established management approaches as well as newtreatment strategies.

Acknowledgments The authors would like to thank Ms. EstherEshkol for editorial assistance.

ConXict of interest statement We declare that we have no conXictof interest.

References

Alho OP, Hannula K, Luokkala A, Teppo H, Koivunen P, Kantola S(2007) DiVerential prognostic impact of comorbidity in head andneck cancer. Head Neck 29:913–918

American Cancer Society (2005) American Cancer Society. Cancerfacts and Wgures 2005. http://www.cancer.org/downloads/STT/CAFF2005f4PWSecured.pdf

Annertz K, Anderson H, Bjorklund A et al (2002) Incidence and sur-vival of squamous cell carcinoma of the tongue in Scandinavia,with special reference to young adults. Int J Cancer 101:95–99

Atula S, Grenman R, Laippala P, Syrjanen S (1996) Cancer of thetongue in patients younger than 40 years. A distinct entity? ArchOtolaryngol Head Neck Surg 122:1313–1319

Bell RB, Kademani D, Homer L, Dierks EJ, Potter BE (2007) Tonguecancer: is there a diVerence in survival compared with other sub-sites in the oral cavity? J Oral Maxillofac Surg 65:229–236

BergshoeV VE, Hopman AH, Zwijnenberg IR et al (2008) Chromo-some instability in resection margins predicts recurrence of oralsquamous cell carcinoma. J Pathol 215:347–348

Brandwein-Gensler M, Teixeira MS, Lewins CM et al (2005) Oralsquamous cell carcinoma. Histologic risk assessment, but notmargin status, is strongly predictive of local disease-free andoverall survival. Am J Surg Pathol 29:167–178

Byers RM, El-Naggar AK, Lee YY et al (1998) Can we detect or pre-dict the presence of occult nodal metastases in patients with squa-mous carcinoma of the oral tongue? Head Neck 20:138–144

Chaturvedi AK, Engels EA, Anderson WF, Gillison ML (2008) Inci-dence trends for human papillomavirus-related and–unrelatedoral squamous cell carcinomas in the United States. J Clin Oncol26:612–619

Clark JR, Naranjo N, Franklin JH, de Almeida J, Gullane PJ (2006)Established prognostic variables in N0 oral carcinoma. Otolaryn-gol Head Neck Surg 135:748–753

Conway DI, Stockton DL, Warnakulasuriya KA, Ogden G, Macpher-son LM (2006) Incidence of oral and oropharyngeal cancer inUnited Kingdom (1990–1999)—recent trends and regional varia-tion. Oral Oncol 42:586–592

D’Souza G, Kreimer AR, Viscidi R et al (2007) Case-control study ofhuman papillomavirus and oropharyngeal cancer. N Engl J Med356:1944–1956

Fagan JJ, Collins B, Barnes L, D’Amico F, Myers EN, Johnson JT(1998) Perineural invasion in squamous cell carcinoma of thehead and neck. Arch Otolaryngol Head Neck Surg 124:637–640

Fakhry C, Westra WH, Li S et al (2008) Improved survival of patientswith human papillomavirus-positive head and neck squamous cellcarcinoma in a prospective clinical trial. J Natl Cancer Inst100:261–269

Friendlander PL, Schantz SP, Shaha AR, Yu G, Shah JP (1998) Squa-mous cell carcinoma of the tongue in young patients: a matched-pair analysis. Head Neck 20:363–368

Fukano H, Matsuura H, Hasegawa Y, Nakamura S (1997) Depth ofinvasion as a predictive factor for cervical lymph node metastasisin tongue carcinoma. Head Neck 19:205–210

Garvello W, SpreaWco R, Gaini RM (2007) Oral tongue cancer inyoung patients: a matched analysis. Oral Oncol 43:894–897

Goldstein DP, Irish JC (2005) Head and neck squamous cell carcinomain the young patients. Curr Opin Otolaryngol Head Neck Surg13:207–211

Hart AK, Karakla DW, Pitman KT, Adams JF (1999) Oral and oropha-ryngeal squamous cell carcinoma in young adults: a report on 13cases and review of the literature. Otolaryngol Head Neck Surg120:828–833

Hyam DM, Conway RC, Sathiyaseelan Y et al (2003) Tongue cancer:do patients younger than 40 do worse? Aust Dent J 48:50–54

Ishida H, Wada K, Masuda T et al (2007) Critical role of estrogenreceptor on anoikis and invasion of squamous cell carcinoma.Cancer Sci 98:636–643

Jamal A, Siegel R, Ward E, Hao Y, Xu J, Thun MJ (2009) Cancer sta-tistics. CA Cancer J Clin. doi:10.3322/caac.20006

Jones JB, Lampe HB, Cheung HW (1989) Carcinoma of the tongue inyoung patients. J Otolaryngol 18:105–108

Jones HB, Sykes A, Bayman N et al (2009) The impact of lymphovascu-lar invasion on survival in oral carcinoma. Oral Oncol 45:10–15

Kuriakose M, Sankaranarayanan M, Nair MK et al (1992) Comparisonof oral squamous cell carcinoma in younger and older patients inIndia. Eur J Cancer B Oral Oncol B 28:113–120

Liang XH, Lewis J, Foote R, Smith D, Kademani D (2008) Prevalenceand signiWcance of human papillomavirus in oral tongue cancer:the Mayo Clinic experience. J Oral Maxillofac Surg 66:1875–1880

Liao C-T, Chang JT-C, Wang H-M et al (2008) Analysis of risk factorsof predictive tumor control in oral cavity cancer. Ann Surg Oncol15:915–922

Loree TR, Strong EW (1990) SigniWcance of positive margins in oralcavity squamous carcinoma. Am J Surg 160:410–414

Martinez-Gimeno C, Molinero AP, Castro V, Sastre MJ, Castro EE,Aguirre-Jaime A (2005) Prospective validation of the Martinez-Gimeno clinicopathologic scoring system (MGSS) for evaluatingrisk of cervical lymph node metastases of squamous cell carci-noma of the oral cavity. Head Neck 27:320–325

McMahon J, O’Brien CJ, Pathak I et al (2003) InXuence of conditionof surgical margins on local recurrence and disease-speciWc sur-vival in oral and oropharyngeal cancer. Br J Oral Maxillofac Surg41:224–231

Mohit-Tabatai MA, Sobel HJ, Rush BF, Mashberg A (1986) Relationof thickness of Xoor of mouth stage I and II cancers to regionalmetastasis. Am J Surg 152:351–353

Muller S, Pan Y, Li R, Chi AC (2008) Changing trends in oral squamouscell carcinoma with particular reference to young patients: 1971–2006. The Emory University experience. Head Neck Pathol 2:60–66

O’Regan EM, Toner ME, Finn SP et al (2008) p16/INK4A genetic andepigenetic proWles diVer in relation to age and site in head andneck squamous cell carcinomas. Hum Pathol 39:452–458

Papageorge MB (2007) Etiology of oral cancer in the young patient: istongue cancer becoming the other cancer in women? Oral Max-illofac Surg Clin N Am 19:163–171

Pitman KT, Johnson JT, Wagner RL, Myers EN (2000) Cancer of thetongue in patients younger than forty. Head Neck 22:297–302

Popovitzer A, Shpitzer T, Bahar G, Marshak G, Ulanovski D, Feinmes-ser R (2004) Squamous cell carcinoma of the oral tongue in youngpatients. Laryngoscope 114:915–917

Regezi JA, Sciubba JJ, Jordan RC (2008) Oral pathology. Clinicalpathologic correlations. Saunders, China

Reid BC, Alberg AJ, Klassen AC et al (2002) A comparison of threecomorbidity indexes in a head and neck cancer population. OralOncol 38:187–194

Ribeiro KC, Kowalski LP, Latorre MR (2000) Impact of comorbidity,symptoms, and patients’ characteristics on the prognosis of oralcarcinomas. Arch Otolaryngol Head Neck Surg 126:1079–1085

123

1048 J Cancer Res Clin Oncol (2010) 136:1039–1048

Ries LA, Melbert D, Krapcho M et al (2008) SEER Cancer StatisticsReview, 1975–2005, National Cancer Institute. Bethesda, MD,Cancer of the oral cavity and pharynx (Invasive). http://seer.can-cer.gov/csr/1975_2005. Accessed Nov 2007

Sarkaria NJ, Harari PM (1994) Oral tongue in young adults less than40 years of age: rationale for aggressive therapy. Head Neck16:107–111

Sathyan KM, Sailasree R, Jayasurya R et al (2006) Carcinoma of tongueand the buccal mucosa represent diVerent biological subentities ofthe oral carcinoma. J Cancer Res Clin Oncol 132:601–609

Schantz SP, Yu GP (2002) Head and neck cancer incidence trends inyoung Americans, 1973–1997, with a special analysis for tonguecarcinoma. Arch Otolaryngol Head Neck Surg 128:268–274

Shah JP, Lydiatt W (1995) Treatment of the cancer of head and neck.CA Cancer J Clin 45:352–368

Shiboski CH, Schmidt BL, Jordan RC (2005) Tongue and tonsil carci-noma. Increasing trends in the U.S. population ages 20–44 years.Cancer 103:1843–1849

Soderberg C, Perez DS, Ukpo OC et al (2008) DiVerential loss ofexpression of common fragile site genes between oral tongue and

oropharyngeal squamous cell carcinomas. Cytogenet GenomeRes 121:201–210

Soo KC, Carter RL, O’Brien CJ, Bar L, Bliss JM, Shaw HJ (1986)Prognostic implications of perineural spread in squamous carci-nomas of the head and neck. Laryngoscope 96:1145–1148

Spiro RH, Huvos AG, Wong GY, Spiro JD, Gnecco CA, Strong EW(1986) Predictive value of tumor thickness in squamous carci-noma conWned to the tongue and Xoor of the mouth. Am J Surg152:345–350

Spiro RH, Guillamondegui O Jr, Paulino AF, Huvos AG (1999) Patternof invasion and margin assessment in patients with oral tonguecancer. Head Neck 21:408–413

Weinberger PM, Yu Z, HaVty BG et al (2006) Molecular classiWcationidentiWes a subset of favorable prognosis. J Clin Oncol 24:736–747

Yuen AP, Lam KY, Wei WI et al (2000) A comparison of the prognos-tic signiWcance of tumor diameter, length, width, thickness, area,volume, and clinicopathological features of oral tongue carci-noma. Am J Surg 180:139–143

123