The Prognostic Value of Spontaneous Apoptosis, Bax,Bcl-2, and p53 in Oral Squamous Cell Carcinoma ofthe Tongue

Xin Xie, M.D.1

Ole Petter F. Clausen, M.D.2

Paula De Angelis, Ph.D.2

Morten Boysen, M.D.1

1 Department of Otolaryngology, The National Hos-pital, University of Oslo, Oslo, Norway.

2 Department of Pathology, The National Hospital,University of Oslo, Oslo, Norway.

Supported by the Norwegian Cancer Society andthe Anders Jahre Foundation.

Address for reprints: Morten Boysen, M.D., Depart-ment of Otolaryngology, The National Hospital,0027 Oslo, Norway.

Received January 11, 1999; revision receivedMarch 29, 1999; accepted March 29, 1999.

BACKGROUND. Bax, Bcl-2, and p53 proteins are involved in the regulation of

apoptosis and have been reported to correlate with prognosis in several tumor

types.

METHODS. Bax, Bcl-2, p53, and the level of spontaneous apoptosis were evaluated

in formalin fixed, paraffin embedded pretreatment specimens from 85 T1– 4 squa-

mous cell carcinomas (SCCs) of the tongue by immunohistochemical methods.

The percentage of apoptotic cells labeled by the terminal deoxynucleotidyl trans-

ferase (TdT)–mediated dUTP labeling (TUNEL) method was expressed as an apo-

ptotic index (AI). For Bax and Bcl-2 evaluation, the fraction of tumor cells stained

and the staining intensities were given scores that were added together, resulting

in a final score. p53 immunostaining was expressed as a percentage of positive

cells.

RESULTS. High AI was significantly associated with high Bax expression (P 5

0.0122) and highly differentiated tumors (P 5 0.0062). No correlation was found

between AI and Bcl-2 expression. There was no correlation between p53 positivity

and any of the other apoptosis-related parameters. Whereas low AI scores and low

Bax expression correlated significantly with poor prognosis (P 5 0.0053 and P 5

0.0012, respectively), a low Bcl-2 expression was associated with a favorable clinical

outcome (P 5 0.0262). Patients with a high Bcl-2/Bax expression ratio had a

significantly poorer prognosis than those with a low ratio (P , 0.0001). Multivariate

analysis revealed that Bax expression, the Bcl-2/Bax expression ratio, and the T and

N classifications were significantly independent prognostic variables. The Bcl-2/

Bax expression ratio was the strongest independent prognostic parameter.

CONCLUSIONS. AI, individual Bax and Bcl-2 expression, and particularly the Bcl-2/

Bax expression ratio have prognostic value in SCC of the tongue. Cancer 1999;86:

913–20. © 1999 American Cancer Society.

KEYWORDS: apoptosis, Bax, Bcl-2, p53, prognosis, tongue squamous cell carci-noma.

Tongue cancer is the most frequently encountered tumor of the oralcavity; there are 60 new cases per year in Norway.1 As compared

with other cancers of the oral cavity, tongue cancer has a greaterpropensity for lymph node metastasis, with incidence rates rangingfrom 15% to 75%, depending on the extent of the primary tumor. Theoverall disease specific survival ranges from 70% for patients withearly stage, lymph node negative disease to 30% for those with ad-vanced lesions.1 In clinical practice, the planning of treatment andprediction of prognosis for patients with oral squamous cell carci-noma (SCC) of the tongue are mainly based on the TNM classifica-tion. This system, however, does not predict the clinical courses of

913

© 1999 American Cancer Society

individual patients and does not give any informationregarding the biologic characteristics of a tumor. It istherefore important to look for new objective prognos-tic factors that might add information about the ag-gressiveness of tumors. Studies of the significance ofapoptosis or programmed cell death represent a bio-logically important field of research that has attractedgreat attention in recent years.

Apoptosis is a genetically regulated cell death in-volved in the deletion of cells in normal as well asmalignant tissues.2,3 Recently, the bcl-2 family of pro-teins, including Bcl-2, Bcl-xL, Bcl-xS, Mcl-1, Bax, andBad, have been shown to play an important role in theregulation of apoptosis.4 Whereas the Bcl-2, Bcl-xL andMcl-1 proteins appear to inhibit apoptosis, Bax, Bad,and Bcl-xS proteins apparently promote apoptosis.4 –7

It has been shown that the proteins of the bcl-2 genefamily heterodimerize and homodimerize with eachother, and the relative proportions of these dimersmay determine whether or not a cell becomes apopto-tic.7–9

Overexpression of Bcl-2 has been reported to pro-tect tumor cells from apoptosis,9,10 whereas increasedBax expression promotes apoptosis induced by cyto-toxic drugs and radiation.11–13 Conflicting results havebeen presented with regard to the correlation betweenBcl-2 and Bax expression and prognosis.14 –20 An ele-vated Bcl-2/Bax mRNA expression ratio has beenshown to correlate with poor clinical outcome in lowgrade urinary bladder carcinoma.21

Loss of TP53 tumor suppressor gene function dueto mutation represents the most common geneticevent known in human cancer.22 Wild-type p53 con-tributes to tumor suppression through at least twomechanisms in response to DNA damage, i.e., arrest ofcell proliferation and induction of apoptosis.23–25 p53may induce apoptotic cell death by down-regulatingBcl-2 and up-regulating Bax expression, which hasbeen suggested to determine the survival or death ofcells following an apoptotic stimulus.26,27 Deregula-tion of the apoptotic pathway due to loss of p53 func-tion probably contributes to treatment failures in pa-tients with cancer by making malignant cells moreresistant to chemotherapy and radiotherapy.28 Thereare conflicting results with regard to p53 as a prognos-tic marker in HNSCC. Some studies showed no corre-lation,29,30 whereas others demonstrated a significantcorrelation between p53 expression and progno-sis.31,32

In this study, we have investigated the level ofspontaneous apoptosis and expression of the apopto-sis-related proteins Bax, Bcl-2, and p53 in a series oftongue SCC. Our intention was to investigate possible

correlations among these apoptotic markers and theirpossible association with prognosis.

MATERIALS AND METHODSFormalin fixed, paraffin embedded pretreatment tu-mor material was available from 85 patients who wererandomly selected from among a total of 300 patientstreated with curative intent for tongue SCC between1983 and 1992. There were 34 women and 51 men,ranging in age from 31 to 88 years (mean, 63 years).Clinical findings, treatment, and follow-up were re-corded prospectively. None of the patients were lost tofollow-up. Only 3 of 14 patients who died of diseasesother than tongue SCC had an observation timeshorter than 3 years (1.7, 2.3, and 2.8 years). Therewere 53 T1–2 and 32 T3– 4 tumors.

Five patients with small primary tumors only un-derwent surgery. Twenty-five patients received com-bined treatment with radiotherapy (60 –70 gray [Gy])followed by surgery, and 19 patients received com-bined surgery and radiotherapy (50 – 60 Gy). Thirty-sixpatients who had inoperable tumors, were unsuitablecandidates for surgery, or refused surgery receivedradiotherapy alone (70 – 80 Gy). Elective radiotherapyof the neck comprised 50 Gy, whereas patients withneck metastases received 60 Gy followed by surgerywhenever possible. Seventeen of the 53 patients withT1–2 tumors and 23 of the 32 patients with T3– 4tumors died of their disease.

Histopathology and Immunohistochemical StainingSerial sections 5 mm thick were cut from the tissueblocks and mounted on gelatin-coated slides. Onesection was stained with hematoxylin and eosin forverification of the original histopathologic diagnosisand for estimation of the degree of histologic differ-entiation. One of the authors (O.P.F.C.) graded thedifferentiation as follows: well differentiated, 1; mod-erately, 2; poorly, 3; and undifferentiated, 4.

Bax and Bcl-2 ImmunohistochemistryStainingBcl-2 and Bax immunostaining and the evaluationwere performed by methods previously described.19 Inbrief, the sections were deparaffinized by 2 washes inxylene for 5 minutes each and then dehydrated inabsolute ethanol. The sections were incubated in 3%(volume of solute per volume of solvent) hydrogenperoxide in methanol (45 seconds) to block endoge-nous peroxidase, followed by incubation with 95% and70% ethanol (15 seconds each), distilled water (1minute), and phosphate-buffered saline (PBS) (5 min-utes). The sections were heated in a pressure cookerfor 5 minutes in 10 mM citric acid buffer (pH 6.0),

914 CANCER September 15, 1999 / Volume 86 / Number 6

followed by rinsing in lukewarm tap water. Next, theywere placed in TBS (Tris-buffered saline, pH 7.8) for 5minutes, and then blocked in TNK buffer (100 mMTris 7.6 –7.8, 550 mM sodium chloride, 10 mM potas-sium chloride), which contained 2% (weight of soluteper volume of solution) bovine serum albumin (BSA),0.1% Triton X100, and 1% normal goat serum. Therabbit antihuman Bax and Bcl-2 polyclonal antibodies(Santa Cruz Biotechnology, Santa Cruz, CA; 1:20 dilu-tion of 100 mg/mL stock made up in TNK buffer) wereadded and the sections were incubated overnight in ahumidified chamber at 4°C. The sections were thenwashed once with PBS and incubated for 1 hour atroom temperature in a humidified chamber with bio-tinylated goat antirabbit antibody (1:500) made up inTNK buffer, followed by washing with PBS, incubationfor 30 minutes at room temperature with streptavidinhorseradish-peroxidase (1:20) made up in TNK buffer,then in a development solution containing 0.06% dia-minobenzidine (DAB) and 0.1% (volume of solute pervolume of solvent) hydrogen peroxide made up inTNK buffer (without goat serum, BSA, and TritonX100). The sections were finally counterstained withhematoxylin and mounted.

EvaluationThe sections were examined microscopically by two ofthe authors in conjunction (X.X. and O.P.F.C.) andscored according to the fraction of stained tumor cellsand the staining intensity. The scores for the percent-ages of positive tumor cells and their staining intensitywere added together to obtain a final score with 11classes ranging from 0 to 5 (Table 1). Sections withoutprimary antibodies were used as negative controls.Striated muscular tissue and/or normal epitheliumpresent in the sections served as internal controls forstaining intensity and were given a score of 1. Theintensity of the immunostaining was sometimes het-erogeneous. We classified the intensity according to

the dominant intensity, having examined the wholetumor area available in each section.19 Occasional dis-agreements regarding the classification were dis-cussed and a consensus reached.

Detection of Apoptotic Cells in Tumor SectionsStainingA terminal deoxynucleotidyl transferase (TdT)–medi-ated dUTP labeling (TUNEL) method was used for thedetection of apoptotic cells. For this purpose, we usedthe APOP Tag Plus In Situ Apoptosis Detection Kit(peroxidase) (Oncor, Gaithersburg, MD). The stainingwas performed according to the manufacturer’s pro-cedure. In brief, deparaffinized sections were treatedwith proteinase K (20 mg/mL) for 15 minutes at roomtemperature prior to using TdT to label the 39-OHends of DNA with digoxigenin-labeled nucleotides (1hour incubation at 37°C). The sections were thentreated with antidigoxigenin antibody peroxidase con-jugate for 30 minutes at room temperature, stainedwith DAB for 5 minutes to produce the characteristicbrown color of positive cells, counterstained with he-matoxylin, and mounted. Sections included in the kitwere stained and served as positive controls.

EvaluationMicroscopic evaluation was performed on 10 –15 fieldsusing a 3100 oil immersion lens. The first field wassubjectively selected and subsequent fields were sys-tematically selected roughly proportional to the over-all size of the tumor area. Cells were defined as apo-ptotic when the whole nuclear area was labeled orwhen occasional labeled globular bodies (apoptoticbodies) could be observed in the cytoplasm. The levelsof spontaneous apoptosis were expressed as an apo-ptotic index (AI), which was the percentage of apopto-tic events from at least 1000 tumor cells counted ineach section.

p53 ImmunohistochemistryStainingAfter dewaxing in xylene, the sections were dehy-drated in ethanol and rinsed in distilled water. Forantigen retrieval the slides were incubated in 10 mMcitric acid buffer (pH 5 6.3) in a microwave oven(750 W) for 5 minutes 35. After cooling for 30 min-utes at room temperature, the sections were washedin Tris buffer for 5 minutes and incubated at roomtemperature with the mouse monoclonal antibodyDO7 (Novocastra Laboratories Ltd., Newcastle UponTyne, UK) at 1:100 dilution. One positive and onenegative control (without the primary antibody)were included for each batch of immunostainedsections. Staining was performed with the conven-

TABLE 1Classification Used for the Evaluation of ImmunohistochemicalStaining of Bax and Bcl-2, and Calculation of Expression Score

% of cells stained ScoreStainingintensity Score

0% 0 Negative 01–30% 1 Weak 0.531–70% 2 Moderate 171–100% 3 Intense 1.5

Very intense 2

The scores for the percentages of cells stained and the staining intensity were added together, resulting

in 11 classes ranging from 0 to 5.

Apoptosis in Tongue Carcinoma/Xie et al. 915

tional alkaline phosphatase–antialkaline phospha-tase (APAAP) method.

EvaluationTen systematically chosen fields were evaluated bymeans of a 3100 oil immersion lens and the percent-age of positive tumor nuclei was obtained. Specimenswith less than 5% positive nuclei were considerednegative. The assessment was performed by one of theauthors (X.X.), and the results were based on the eval-uation of at least 700 tumor nuclei in each section.

For all parameters, areas with pronounced inflam-mation, necrosis, or artificial damage were avoided.The sections were assessed without knowledge of theclinical outcome.

ReproducibilityFour months after the initial evaluation, eight sectionswere randomly selected to test the reproducibility ofeach of the parameters measured.

StatisticsThe data were stored and analyzed by means of SASversion 6.12 software (SAS Institute, Cary, NC). TheStudent t test was used for comparison of continuousdata and the Mann–Whitney–Wilcoxon test was usedto compare categoric data. The Pearson test was usedto test the correlation between parameters. The logrank test was used to screen for the prognostic value ofclinical and apoptosis-related parameters in relationto disease specific survival, and Kaplan–Meier plotswere used to illustrate the effect of selected variableson the disease specific survival. A case was censoredwhen death resulted from causes other than the orig-inal tumor or when there was no evidence of theoriginal tumor at the last follow-up consultation. Coxproportional hazards regression analysis was per-formed by the stepwise procedure. P values ,0.05were considered statistically significant. Reproducibil-ity was tested by means of the least-square regressionanalysis.

RESULTSBax and Bcl-2 immunoreactivity were found in 83tumors (97%). In cases in which identical areas of thetumor could be identified, Bax and Bcl-2 expressionwere often observed in the same area of the tumor.Apoptotic cells labeled by the TUNEL method werescattered throughout the tumor area. Many of thepositive cells showed apoptosis by morphology, butsome cells were apparently preapoptotic because nomorphological characteristics of apoptosis were ob-served. Positive p53 immunostaining was found in 52of the 85 tumors (61%). Table 2 presents the overall

mean values, standard deviations, and ranges for pa-rameters evaluated.

No correlation was found between Bcl-2 and Baxexpression. High AI correlated with high Bax expres-sion (P 5 0.0122) and highly differentiated tumors(P 5 0.0062), whereas no correlation was found be-tween the AI and Bcl-2 expression. No correlationswere found between p53 positivity and other apoptot-ic-related parameters, but tumors showing p53 posi-tivity lower than the mean value tended to have highBax expression (P 5 0.076). There was no correlationbetween apoptosis-related parameters and either T orN classification.

Patients with high Bax expression had better dis-ease specific survival compared with those who hadlow Bax expression (P 5 0.0012, Fig. 1A). For Bcl-2expression the opposite correlation was found, withsignificantly poorer disease specific survival for pa-tients with high Bcl-2 expression (P 5 0.0262, Fig. 1B).High AI was associated with a favorable clinical out-come compared with low AI (P 5 0.0053, Fig. 1C).There was no difference between p53 positive andnegative tumors regarding the disease specific survival(P 5 0.495). However, patients with tumors with p53positivity exceeding the mean value (.30%) tended tohave a poor prognosis when compared with those whohad tumors with p53 positivity lower than mean value(P 5 0.0576). Patients with a Bcl-2/Bax expressionratio .1 had an unfavorable clinical outcome com-pared with those who had ratios #1 (P , 0.0001,Fig. 1D).

Table 3 summarizes the univariate analyses forclinical and immunohistochemical variables for dis-ease specific survival using the overall mean values ascutoff levels. All parameters, except p53 positivity,showed significant correlations. Multivariate analysisrevealed that the T and N classifications, Bax expres-sion, and Bcl-2/Bax expression ratio were significantand independent variables. The Bcl-2/Bax expressionratio was the strongest independent prognostic pa-rameter, with a relative risk of 4.57 (Table 3).

The reproducibility of the apoptotic-related pa-

TABLE 2Mean, Standard Deviation, and Range for Apoptosis-RelatedParameters in 85 Tongue Squamous Cell Carcinomas

Parameters Mean 6 SD Range

Bcl-2 3.5 6 1.0 0–4.5Bax 3.0 6 1.0 0–5AI 1% 6 0.6% 0.1–3.4%p53 pos. 30% 6 29% 0–88%

SD: standard deviation; AI: apoptotic index; pos: positivity.

916 CANCER September 15, 1999 / Volume 86 / Number 6

rameters was acceptable, with values for least squareregression analysis (R2) for Bax, Bcl-2, AI, and p53 of0.82, 0.77, 0.62, and 0.98, respectively.

DISCUSSIONHigh Bax expression correlated significantly withhigh values for the AI in the current study, whereasno correlation was found between the levels of ex-pression of Bcl-2 and AI. Bax expression emerged asthe most important single, i.e., not composite, prog-nostic marker, whereas high expression was associ-ated with good prognosis when compared with lowexpression. This observation concurs with the re-sults of our previous study of glottic SCC.19 Our

study further revealed that high expression of Bcl-2was significantly associated with poor prognosis andlow expression with good prognosis. Previous stud-ies of Bcl-2 expression in head and neck carcinomashave shown that both positive and negative tumorshave been associated with good prognosis.15,17 Con-flicting results have also been demonstrated for sev-eral other tumor types. In breast and esophagealcarcinomas, Bcl-2 expression correlated with goodprognosis.14,16The opposite correlation was found inprostate and testicular carcinomas,33,34 whereas nosuch correlation was demonstrated in colorectalcarcinomas.35 Overexpression of the Bax protein hasbeen associated with favorable outcome in breast

FIGURE 1. Disease free survival is

shown according to (A) Bax expression

(P 5 0.0012), (B) Bcl-2 expression (P 5

0.0262), (C) apoptotic index (AI) (P 5

0.0053), and (D) Bcl-2/Bax expression

ratio (P , 0.000l) for 85 patients with

oral squamous cell carcinoma of the

tongue.

TABLE 3Univariate and Multivariate Analyses for Clinical Parameters, Apoptotic Index, Bax Expression, Bcl-2 Expression, and p53 Positivity in Relationto Disease Free Survival for 85 Patients with Tongue Squamous Cell Carcinoma

Prognosticvariables

Clinicalgroups/meancutoff levels

Survivors(n 5 45)

Nonsurvivors(n 5 40)

Log rank CoxregressionP value

Riskratiox2 P value

T class. T1–2/T3–4 36/9 17/23 12.31 0.0004 0.0338 1.305N class. N0/N1 30/15 15/25 9.06 0.0026 0.0429 2.140Stage I–II/III–IV 27/18 11/29 10.26 0.0014Tumor differen. ,2/$2 26/19 21/19 0.17 0.6759AI ,1%/$1% 15/30 27/13 7.76 0.0053Bax ,3/$3 11/34 23/17 10.53 0.0012 0.0197 0.455Bcl-2 ,3.5/$3.5 24/21 8/32 4.94 0.0262p53 pos. ,30%/$30% 29/16 17/23 3.60 0.0576Bcl-2/Bax #1/.1 33/12 8/32 24.43 ,0.0001 0.0004 4.570

Class.: classification; differen.: differentiation; pos.: positivity; AI: apoptotic index.

Apoptosis in Tongue Carcinoma/Xie et al. 917

adenocarcinoma and glottic SCC18,19 and poor prog-nosis in ovarian carcinomas,20 whereas no associa-tion between Bax expression and prognosis wasfound in a study of gastric carcinomas.36 The con-tradictory results concerning prognosis and expres-sion of Bax and Bcl-2 proteins, for tumors of thehead and neck as well as tumors at other sites,indicate that induction of apoptosis by proteins ofthe bcl-2 gene family is very complex and that theinfluence of individual proteins may vary from tu-mor to tumor. Differences in staining techniquesand evaluation may also account for some of thediscrepancies observed.

Accumulated evidence suggests that the pro-teins of the bcl-2 gene family may interact with eachother as homodimers and heterodimers, and thattheir relative proportions regulate the process ofapoptosis.7–9 We therefore analyzed the possiblecorrelation between two of these proteins in deter-mining clinical outcome. Interestingly, we foundthat the ratio of Bcl-2/Bax expression appeared tobe the best variable in predicting disease specificsurvival in tongue SCC. To our knowledge, this is thefirst time such a correlation has been demonstrated.This result is in agreement with a study of low gradeurinary bladder tumors using a Bcl-2/Bax mRNAexpression ratio.21 Overexpression of Bax has beenobserved to promote apoptosis by increasing thesusceptibility to anticancer drugs and radiation,11–13

whereas overexpression of Bcl-2 permitted cells tosurvive such influences.9,10 Because 80 of 85 pa-tients included in this material were treated by ra-diotherapy alone or by radiotherapy combined withsurgery, the prognostic significance of high Baxand/or low Bcl-2 expression may be interpreted asan indicator of response to radiotherapy in tongueSCC.

The results of this study also showed that lowscores for AI correlated with poor prognosis. This is incontrast to the results of a study of laryngeal carcino-mas in which no correlation was found between AIand prognosis.37 For other types of tumors, such ascervical carcinomas38 and breast carcinomas,39 highscores for AI correlated with poor prognosis, whereasopposite results have been reported for colorectal tu-mors.40,41 These contradictory findings suggest thatthe prognostic significance of apoptosis may vary be-tween tumor types. It should be emphasized that thein situ technique for detection of apoptosis is notentirely specific, because overlap between apoptoticand necrotic cell death has been reported.42 This mayexplain why some morphologically apoptotic cells donot stain, which may result in the underestimation ofAI in some cases.

The p53 protein may suppress tumor growththrough induction of apoptosis and growth ar-rest.23–25 In response to DNA damage, wild-type p53protein may activate the transcription of proteinssuch as p21/waf1 and gadd45, to arrest cells in theG1-phase of the cell cycle and repair the DNA dam-age.43,44 p53 has been suggested to induce apoptosisthrough regulation of Bax and Bcl-2 gene expres-sion.26,27 In this study, 61% of all tumors expresseddetectable levels of p53. It is assumed that accumu-lation of p53 protein represents TP53 dysfunction,because missense mutations of TP53 stabilize theprotein by increasing its half-life considerably com-pared with the wild-type protein,45 thus allowingdetection only of mutated protein under the condi-tions used. The accumulated p53 protein that wasdetected in this study was therefore assumed to bestabilized by mutation. However, it cannot be ruledout that accumulation of wild-type p53 protein tosome extent also may accumulate by complexing toother proteins.46 The DO7 antibody was used in thisstudy because this antibody has been shown to besuperior to others in detecting p53 accumulation informalin fixed material.47 We found no correlationbetween p53 accumulation and other apoptoticvariables. However, there was a tendency that tu-mors with relatively low p53 expression (,30% pos-itive nuclei) expressed higher levels of Bax and thatthis group had a favorable prognosis.

Our study suggests that the levels of spontaneousapoptosis, Bax expression, and Bcl-2 expression haveprognostic value in SCC of the tongue, and that theBcl-2/Bax expression ratio is the strongest prognosticvariable regarding disease specific survival.

REFERENCES1. Næss Å. Incidence and survival of Norwegian head and neck

cancer patients in six five-year periods from 1960 to 1989.Output from the database at the Norwegian Cancer Registry.Oslo: The Norwegian Cancer Registry, 1991.

2. Searle J, Collins DJ, Harmon B, Kerr JF. The spontaneousoccurrence of apoptosis in squamous carcinomas of theuterine cervix. Pathology 1973;5:163–9.

3. Wyllie AH. The biology of cell death in tumours. AnticancerRes 1985;5:131– 6.

4. Reed JC. Bcl-2 and the regulation of programmed cell death.J Cell Biol 1994;124:1– 6.

5. Hockenbery DM. Bcl-2 in cancer, development and apopto-sis. J Cell Sci 1994;18:51–5.

6. Boise LH, Gottschalk AR, Quintans J, Thompson CB. Bcl-2and Bcl-2–related proteins in apoptosis regulation. Curr TopMicrobiol Immunol 1995;200:107-21.

7. Oltvai ZN, Milliman CL, Korsmeyer SJ. Bcl-2 heterodimer-izes in vivo with a conserved homolog, Bax, that acceleratesprogrammed cell death. Cell 1993;74:609 –19.

8. Oltvai ZN, Korsmeyer SJ. Checkpoints of dueling dimers foildeath wishes. Cell 1994;79:189 –92.

918 CANCER September 15, 1999 / Volume 86 / Number 6

9. Reed JC. Bcl-2: prevention of apoptosis as a mechanism ofdrug resistance. Hematol Oncol Clin North Am 1995;9:451–73.

10. Miyashita T, Reed JC. Bcl-2 gene transfer increases relativeresistance of S49.1 and WEH17.2 lymphoid cells to cell deathand DNA fragmentation induced by glucocorticoids andmultiple chemotherapeutic drugs. Cancer Res 1992;52:5407–11.

11. Wagener C, Bargou RC, Daniel PT, Bommert K, Mapara MY,Royer HD, et al. Induction of the death-promoting genebax-alpha sensitizes cultured breast cancer cells to drug-induced apoptosis. Int J Cancer 1996;67:138 – 41.

12. Sakakura C, Sweeney EA, Shirahama T, Igarashi Y, HakomoriS, Nakatani H, et al. Overexpression of bax sensitizes humanbreast cancer MCF-7 cells to radiation-induced apoptosis.Int J Cancer 1996;67:101–5.

13. Kitada S, Krajewski S, Miyashita T, Krajewska M, Reed JC.Gamma-radiation induces upregulation of Bax proteinand apoptosis in radiosensitive cells in vivo. Oncogene1996;12:187–92.

14. Lipponen P, Pietilainen T, Kosma VM, Aaltomaa S, Eskeli-nen M, Syrjanen K. Apoptosis suppressing protein bcl-2 isexpressed in well-differentiated breast carcinomas withfavourable prognosis. J Pathol 1995;177:49 –55.

15. Wilson GD, Grover R, Richman PI, Daley FM, Saunders MI,Dische S. Bcl-2 expression correlates with favourable out-come in head and neck cancer treated by accelerated radio-therapy. Anticancer Res 1996;16:2403– 8.

16. Ohbu M, Saegusa M, Kobayashi N, Tsukamoto H, Mieno H,Kakita A, et al. Expression of bcl-2 protein in esophagealsquamous cell carcinomas and its association with lymphnode metastasis. Cancer 1997;79:1287–93.

17. Friedman M, Grey P, Venkatesan TK, Bloch I, Chawla P,Caldarelli DD, et al. Prognostic significance of Bcl-2 expres-sion in localized squamous cell carcinoma of the head andneck. Ann Otol Rhinol Laryngol 1997;106:445–50.

18. Krajewski S, Blomqvist C, Franssila K, Krajewska M, Wase-nius VM, Niskanen E, et al. Reduced expression of proapop-totic gene Bax is associated with poor response rates tocombination chemotherapy and shorter survival in womenwith metastatic breast adenocarcinoma. Cancer Res 1995;55:4471– 8.

19. Xie X, Clausen OPF, De Angelis P, Boysen M. Bax expressionhas prognostic significance that is enhanced when com-bined with AgNOR counts in glottic carcinomas. Br J Cancer1998;78:100 –5.

20. Marx D, Binder C, Meden H, Lenthe T, Ziemek T, Hidde-mann T, et al. Differential expression of apoptosis associ-ated genes bax and bcl-2 in ovarian cancer. Anticancer Res1997;17:2233– 40.

21. Gazzaniga P, Gradilone A, Vercillo R, Gandini O, Silvestri I,Napolitano M, et al. Bcl-2/bax mRNA expression ratio asprognostic factor in low-grade urinary bladder cancer. Int JCancer 1996;69:100 – 4.

22. Hollstein M, Sidransky D, Vogelstein B, Harris CC. p53 mu-tations in human cancers. Science 1991;253:49 –53.

23. Hartwell L. Defects in a cell cycle checkpoint may be re-sponsible for the genomic instability of cancer cells. Cell1992;71:543– 6.

24. Yonish-Rouach E, Resnitzky D, Lotem J, Sachs L, Kimchi A,Oren M. Wild-type p53 induces apoptosis of myeloid leu-kaemic cells that is inhibited by interleukin-6. Nature 1991;352:345–7.

25. Attardi LD, Lowe SW, Brugarolas J, Jacks T. Transcriptionalactivation by p53, but not induction of the p21 gene, isessential for oncogene-mediated apoptosis. EMBO J 1996;15:3693–701.

26. Miyashita T, Krajewski S, Krajewska M, Wang HG, Lin HK,Liebermann DA, et al. Tumor suppressor p53 is a regulatorof bcl-2 and bax gene expression in vitro and in vivo. On-cogene 1994;9:1799 – 805.

27. Selvakumaran M, Lin HK, Miyashita T, Wang HG, KrajewskiS, Reed JC, et al. Immediate early up-regulation of baxexpression by p53 but not TGFb1: a paradigm for distinctapoptotic pathways. Oncogene 1994;9:1791– 8.

28. Lowe SW, Ruley HE, Jacks T, Housman DE. p53-dependentapoptosis modulates the cytotoxicity of anticancer agents.Cell 1993;74:957– 67.

29. Frank JL, Bur ME, Garb JL, Kay S, Ware JL, Sismanis A, etal. p53 tumor suppressor oncogene expression in squa-mous cell carcinoma of the hypopharynx. Cancer 1994;73:181– 6.

30. Nylander K, Stenling R, Gustafsson H, Zackrisson B, Roos G.p53 expression and cell proliferation in squamous cell car-cinomas of head and neck. Cancer 1995;75:87–93.

31. Caminero MJ, Nunez F, Suarez C, Ablanedo P, Riera JR,Dominguez F. Detection of p53 protein in oropharyngealcarcinoma: prognostic implications. Arch Otolaryngol HeadNeck Surg 1996;122:769 –72.

32. Hirvikoski P, Kumpulainen E, Virtaniemi J, Johansson R,Haapasalo H, Marin S, et al. p53 expression and cell prolif-eration as prognostic factors in laryngeal squamous cellcarcinoma. J Clin Oncol 1997;15:3111–20.

33. Keshgegian AA, Johnston E, Cnaan A. Bcl-2 oncoproteinpositivity and high MIB-1 (Ki-67) proliferative rate are inde-pendent predictive markers for recurrence in prostate car-cinoma. Am J Clin Pathol 1998;110:443–9.

34. Eid H, Gulyas M, Geczi L, Bodrogi I, Institoris E, Bak M.Expression of bcl-2 in testicular carcinoma: correlation withtumor progression and MDR1/Pgp. Cancer 1998;83:331– 6.

35. Tollenaar RA, van Krieken JH, van Slooten HJ, Bruinvels DJ,Nelemans KM, van den Broek LJ, et al. Immunohistochem-ical detection of p53 and Bcl-2 in colorectal carcinoma: noevidence for prognostic significance. Br J Cancer 1998;77:1842–7.

36. Koshida Y, Saegusa M, Okayasu I. Apoptosis, cell prolifera-tion and expression of Bcl-2 and Bax in gastric carcinomas:immunohistochemical and clinicopathological study. Br JCancer 1997;75:367–73.

37. Hagedorn HG, Tubel J, Wiest I, Nerlich AG. In situ apoptoticand proliferation index in laryngeal squamous cell carcino-mas. Anal Cell Pathol 1998;16:177– 84.

38. Levine EL, Renehan A, Gossiel R, Davidson SE, Roberts SA,Chadwick C, et al. Apoptosis, intrinsic radiosensitivity andprediction of radiotherapy response in cervical carcinoma.Radiother Oncol 1995;37:1–9.

39. Zhang GJ, Kimijima I, Abe R, Watanabe T, Kanno M, Hara K,et al. Apoptotic index correlates to bcl-2 and p53 proteinexpression, histological grade and prognosis in invasivebreast cancers. Anticancer Res 1998;18:1989 –98.

40. Langlois NE, Lamb J, Eremin O, Heys SD. Apoptosis incolorectal carcinoma occurring in patients aged 45 yearsand under: relationship to prognosis, mitosis, and immuno-histochemical demonstration of p53, c-myc and bcl-2 pro-tein products. J Pathol 1997;182:392–7.

Apoptosis in Tongue Carcinoma/Xie et al. 919

41. Sugamura K, Makino M, Kaibara N. Apoptosis as a prognos-tic factor in colorectal carcinoma. Surg Today 1998;28:145–50.

42. Harrison DJ. Counting apoptosis— why and how? J ClinPathol (Clin Mol Pathol) 1996;49:M245– 6.

43. Xiong Y, Hannon GJ, Zhang H, Casso D, Kobayashi R, BeachD. P21 is universal inhibitor of cyclin kinases. Nature 1993;366:701– 4.

44. Smith ML, Chen IT, Zhan Q, Bae I, Chen CY, Gilmer TM, etal. Interaction of the p53-regulated protein GADD45 withproliferating cell nuclear antigen. Science 1994;266:1376 – 80.

45. Finlay CA, Hinds PW, Tan TH, Eliyahu D, Oren M, Levine AJ.Activating mutations for transformation by p53 produce agene product that forms an hsc70-p53 complex with analtered half-life. Mol Cell Biol 1988;8:531–9.

46. Hall PA, McKee PH, Menage HDP, Dover R, Lane DP. Highlevels of p53 proteins in UV-irradiated normal human skin.Oncogene 1993; 8:203–7.

47. Baas IO, Mulder JW, Offerhaus GJ, Vogelstein B, HamiltonSR. An evaluation of six antibodies for immunohistochem-istry of mutant p53 gene product in archival colorectal neo-plasms. J Pathol 1994;172:5–12.

920 CANCER September 15, 1999 / Volume 86 / Number 6