Akt activation correlates with adverse outcome in tongue cancer

Authors


  • This study contains original data that have never been published in previous presentations, reports, or publications.

Abstract

BACKGROUND

Recent data have shown a significant association between phosphorylated-Akt (p-Akt) and failure of local disease control by radiation therapy in head and neck squamous carcinoma (HNSCC), and also that Akt activation correlates with histologic progression of HNSCC from premalignant lesions to invasive cancer. This study evaluated the role of Akt in previously untreated preneoplastic lesions of oral cavity and invasive tongue carcinoma on patient outcome and cancer development.

METHODS

PKB/Akt activation was assessed by immunohistochemistry using a phosphorylation state-specific antibody (Ser 473) in tongue cancer and preneoplastic specimens of oral cavity.

RESULTS

The expression of p-Akt was detected in 24 (46%) of the 52 available tongue cancer cases and in 10 (45%) of the 22 available preneoplastic lesions. In tongue cancer, with a median follow-up of 7.3 years, p-Akt was highly expressed in the cases that relapsed (15 of 17, 88%) or died of cancer (10 of 12, 83%). Disease-free survival was significantly shorter in cases with Akt expression (log rank test, P < 0.0001) independently of the stage and nodal status.

CONCLUSIONS

Expression of p-Akt correlated with worse outcome in patients with tongue cancer. This finding highlights the potential role of Akt as a prognostic marker and as a potential target for molecular therapeutics. Cancer 2005. © 2005 American Cancer Society.

The estimated incidence of new head and neck squamous cell carcinomas (HNSCC), including oral cavity, pharynx, and larynx, in the U.S. in 2004 is 38,530; of these, 7320 new cases of oral tongue cancers would be diagnosed in 2004, with 1700 deaths.1 HNSCC can be treated effectively with local therapy alone when it presents in early stage.2, 3 For the most common presentation of advanced AJCC Stages III and IV disease, surgery and radiation therapy result in 30% locoregional and distant relapse rate and chemotherapy, especially as part of concomitant treatment, and has conferred only a 4% absolute survival benefit at 2 and 5 years.4, 5 In particular, squamous cell carcinoma (SCC) of the tongue has increased in incidence and in patients with extracapsular spread overall; disease-specific survival rates were 30% and 50%, respectively.6, 7

To date, only clinicopathologic factors such as tumor size, nodal stage, and histologic characteristics of the tumor (close surgical margins, perineural invasion, lymphatic and vascular invasion) have been validated as prognostic factors in HNSCC.6 In terms of tumor biology, our group has previously demonstrated that overexpression of cyclin B1 and lack of FHIT and PTEN expression are associated with a more aggressive biological behavior of tongue SCC.8–10 The lipid phosphatase activity of PTEN is associated with down-regulated activities of multiple downstream components of the phosphatydilinositol-3-kinase (PI3K) pathway, including most notably the serine/threonine kinase AKT.11 The PI3K pathway has been targeted in efforts to identify new biological drugs because of its key role in vital cellular processes including survival, proliferation, migration, and differentiation.12 Akt activation has been found more frequent in the mucosa and the primary tumor site in patients with pharyngeal cancer.13 In vitro data have shown that the PI3K pathway is involved in Ras-induced radio-resistance and in the resistance to EGFR inhibition.14–16 In HNSCC, a significant association between PI3K activation by Akt phosphorylation and local control confirmed that PI3K activation might be a prognostic marker for response to radiation therapy.17

Head and neck SCC results from a multistep carcinogenesis process in which increasing degrees of mucosal changes and cellular atypia occur over large areas of the carcinogen-exposed upper aerodigestive tract epithelium.18 The process is associated with acquisition of a transformed phenotype19 and accumulation of specific molecular genetic events20–23; however, histopathologic evaluation remains the time-honored method in risk assessment of premalignant lesions. There is a clear need for better biologic models of risk. Along these lines, Akt activation was recently described as an early cellular response to carcinogen exposure and may be an important step in environmental carcinogenesis.24, 25 Moreover, evidence for Akt activation by nicotine and the tobacco-specific carcinogen 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK) leading to deregulation cell growth and apoptosis have been reported.24 Recent publications proposed that Akt activation, PI3K accumulation, and PTEN down-regulation detected in HNSCC could reflect the early biochemical effects of tobacco components such as nicotine. This hypothesis was based on the finding that histologically normal mucosa has been found to be positive for Akt2 amplification, p-Akt accumulation, and PTEN down-regulation, leading to the conclusion that a carcinogen-containing environment that initiated tumorigenesis also affected the nontransformed surrounding tissues, in accordance with the field cancerization theory.13 An additional observation of the same study was that Akt activation correlated with squamous cell carcinoma progression from normal epithelium to invasive cancer.25

These observations prompted us to explore the prognostic role of the PI3K pathway through Akt activation status in tongue cancer and its predictive impact on cancer development in head and neck tumorigenesis. Indeed, we observed frequent activation of Akt in preneoplastic lesions and invasive HNSCC and, most important, found a correlation between Akt activation and adverse outcome in tongue SCC.

MATERIALS AND METHODS

After approval from the Institutional Review Board, a waiver of informed consent was obtained for the conduct of this study.

Tongue Cancer

Patients diagnosed with squamous carcinoma of the tongue were retrospectively identified from the database of U.T.M.D. Anderson Cancer Center (Houston, TX) and from the clinical records of the Institute of Dental Sciences, University of Ancona (Ancona, Italy). Fifty-two cases (35 from U.T.M.D. Anderson Cancer Center and 17 from the University of Ancona) with previously untreated surgical excision performed between 1990 and 1997 had available follow-up data, resulting in eligibility for inclusion in the study. Baseline data obtained from patient charts included age, sex, histologic grade, tumor stage, nodal involvement, and survival (Table 1). All patients underwent surgical excision and the median follow-up was 7.3 years. Patients with lymph node involvement or bulky tumors (T3–T4) generally received standard postoperative radiotherapy and/or chemotherapy.

Table 1. Akt Activation in Tongue Cancer by Clinicopathological Characteristics of Patients
 United StatesItaly
P-Akt positiveP-Akt negativeP-Akt positiveP-Akt negative
  • P-Akt: phosphorylated-Akt; G: Grade; N: Node

  • a

    Fisher exact test.

  • b

    Chi-square test.

Total no. (%)16 (45.7)19 (54.3)8 (47)9 (53)
Age median (range)55 (38-78)69 (48-87)
Grading    
G181144
G26435
G32410
 P = 0.65aP = 0.8a
Gender    
Male131067
Female3922
 P = 0.15aP = 1.0a
Nodal status    
N011757
N+51232
 P = 0.06bP = 0.62a
Stage    
I0036
II8511
III3622
IV5820
 P = 0.44aP = 0.61a

Preneoplastic Lesions

The baseline biopsies from 22 participants with various degrees of oral cavity dysplasia in three chemoprevention trials at U.T.M.D. Anderson Cancer Center formed the materials for this study. Age, sex, histology, smoking status, site, and follow-up data were obtained from patient charts and case report forms of the clinical trials (Table 2). Cancer-free survival interval was calculated from the date of diagnosis to the date of cancer development.

Table 2. Akt Activation in Head and Neck Preneoplastic Lesions by Clinicopathological Characteristics of Patients
ParameterP-Akt negativeP-Akt positiveP value
  • P-Akt: phosphorylated-Akt.

  • a

    Cochran Armitage test.

  • b

    Fisher exact test.

Total no. (%)12 (55)10 (45) 
Age median (range)47 (33-75)65 (25-75) 
Dysplasia   
Mild4 (36%)7 (64%) 
Moderate2 (50%)2 (50%) 
Severe6 (86%)1 (14%)0.04a
Gender   
Male6 (55%)5 (45%) 
Female6 (55%)5(45%)1.00b
Smoking status   
Current3 (75%)1 (25%) 
Former6 (50%)6 (50%) 
Never smokers3 (50%)3 (50%)0.72a

Immunohistochemistry

A mouse polyclonal antiphospho (Ser 473) Akt antibody (Cell Signaling Technology, Beverly, MA) was used.

Unstained 4-μm tissue sections of formalin-fixed, paraffin-embedded specimens were deparaffinized, rehydrated, and steamed for 30 minutes in 10 mM citrate buffer (pH 6.0) to unmask the antigen. Immunohistochemistry was performed as previously described26 with the addition of anti-p-Akt antibody. Anti-p-Akt mouse polyclonal antibody was at 1:100 concentration in blocking buffer with 0.2% bovine serum albumin and 0.1% sodium azide solution (pH 7.6) at 4 °C overnight. The MDA-468 PTEN negative cell line with well-defined cytoplasmic positivity for p-Akt (Ser 473) was used as positive control by placing a tissue section of cell line pellets on the same slides as the tissue sections. As negative controls, we used different sections of each biopsy, which were treated with nonimmune rabbit serum without addition of the primary antibody.

Two squamous mucosa specimens from nonsmoking volunteers were also stained for p-Akt and some cellular cytoplasmic positivity (2–3%) was observed. Representative areas chosen by the dedicated pathologist were selected based on their localization and quality of the overall stained patterns for this estimation.

After visual evaluation by a head and neck pathologist (A.E.N.), two independent observers (E.M., V.P.) scored the slides without knowledge of clinical outcome of these cases. Cytoplasmic staining was considered positive when at least weak to moderate cytoplasmic staining was seen in over 15% of cells. The cut-off value of 15% represented the median value of labeling indices observed in our samples. Labeling index was defined as normal positive cells/total number of cells counted.

Statistical Analysis

The association between the clinicopathologic variables and immunohistochemical parameters were investigated using the Fisher exact test, chi-square test, and Cochran-Armitage trend test. Kaplan–Meier curves were fitted to estimate the probability of survival and cancer development.27 In tongue cancer, the endpoints for the outcome analysis were disease-free survival, defined as the time between diagnosis date and first relapse date or metastases date or cancer-related death date, and overall survival, defined as the time between diagnosis date and death or last follow-up date. In preneoplastic lesions the endpoint was time to cancer progression defined as the time between the date of diagnosis and the date of cancer development. The computations were carried out using the SAS (Cary, NC) software package and the statistical result was considered significant if P < 0.05.

RESULTS

Patient characteristics in cancer and preneoplastic lesions are respectively summarized in Tables 1 and 2. As noted in Table 1, there were no significant differences in patient characteristics distribution between the two countries except for stage (Fisher exact test, P < 0.0001).

Phosphorylated-Akt in Premalignant Lesions

In preneoplastic lesions, cytoplasmic p-Akt expression homogeneously distributed among all layers of the epithelium was found in 10 (45%) cases (Fig. 1c). p-Akt was more frequently expressed in mild and moderate dysplasia compared with severe dysplasia (Cochran Armitage test, P = 0.04, Table 2). No significant correlation was found between other clinicopathological characteristics (age, sex, site, and smoking). With a median follow-up of 11 months, 9 (41%) cases developed cancer and Akt activation was found in 5 (55%) of them, whereas 5 of 13 (38%) cases remained cancer-free (log rank test, P = 0.16) (Table 3; Fig. 2).

Figure 1.

Representative examples of p-Akt staining in tongue cancer: (a) p-Akt-positive, (b) p-Akt-negative; in preneoplastic lesions: (c) p-Akt-positive, (d) p-Akt-negative, and (e) in normal epithelium. Sections of each specimen were photographed at ×20 magnification.

Table 3. Cancer Development by P-Akt Expression in Head and Neck Preneoplastic Lesions
P-AktCancer developmentTotalP value
No (%)Yes (%)
  • P-Akt: phosphorylated-Akt.

  • a

    Log-rank test.

Loss8 (67)4 (33)12 
Retention5 (50)5 (50)100.16a
Figure 2.

Cancer-free survival by Akt activation in head and neck preneoplastic lesions.

Phosphorylated-Akt in Tongue Cancer

Twenty-four (46%) samples were scored positive; of the remaining 28 negative cases, 15 showed complete absence of staining. The large majority of the positive samples showed cytoplasmic staining and only a few cells showed nuclear staining. In cancer samples, intensive cytoplasmic positivity was noted at the periphery of tumor clusters (Fig. 1a). We observed different patterns of Akt activation; the most frequently observed pattern was positive staining in the cytoplasm of the invasive front of the squamous cell carcinoma. No correlation was found between clinicopathological characteristics (age, sex, grading, stage, lymph node invasion) and p-Akt expression (Table 1).

Phosphorylated-Akt Predicts Outcome in Tongue Cancer

In multivariate analysis, we stratified by country and adjusted for stage, nodal status, age, gender, and histology, providing evaluation of the impact of these factors on survival.

With a median follow-up of 7.3 years, 29 (56%) of the 52 patients died. p-Akt was highly expressed in 14 of these 29 (48%) cases. Overall survival was not significantly affected by sex, stage, histology, or p-Akt status; however, age and nodal status did affect overall survival significantly (Table 4). Twelve of the 29 deaths (41%) were cancer related and 10 (83%) of these 12 cases showed Akt activation (Fig. 3). Seventeen (33%) of the 52 patients relapsed (10 locally and 7 with distant metastasis) and/or died of cancer (12 cancer-related deaths). p-Akt was highly expressed in 15 (88%) of these 17 cases. Disease-free survival was significantly shorter in cases with Akt activation (log rank test, P < 0.0001) (Fig. 4a–c). Through multivariate analysis, p-Akt emerged as the sole adverse prognostic factor for disease-free survival, superseding the effect of nodal status (Table 4).

Table 4. Multivariate Cox Model in Survival Analysis
CovariateHazard ratio (95% CI)P value
  1. CI: confidence interval; P-Akt: phosphorylated-Akt.

Overall survival
Age (continuous variable)1.06 (1.02, 1.10)0.001
Nodal status (positive vs. negative)2.44 (1.06, 5.58)0.04
P-Akt1.83 (0.82, 4.09)0.14
Disease-free survival
P-Akt10.69 (2.43, 46.95)0.02
Figure 3.

Local relapses and distant metastasis in tongue cancer by Akt activation status.

Figure 4.

Akt activation in tongue cancer: disease-free survival curve by (a) stage, by (b) Akt activation, and by (c) nodal status and p-Akt activation. E = events (relapse or metastasis or cancer death), N = total number.

DISCUSSION

Activation of the PI3K/Akt pathway plays a pivotal role in cell survival, proliferation, migration, and differentiation that underlie the biology of human cancer.12 The most important downstream effector of PI3K is PKB/Akt serine/threonine kinase, whose constitutive activity promotes cellular survival and resistance to chemotherapy and γ-irradiation.14 Recently published data confirmed the central role of Akt activation in the mucosa and the primary tumor site of patients with pharyngeal cancer.13 However, there is little information on the role of Akt activation as a prognostic factor in oral cancer specimens. Our study is the first to explore the predictive significance of p-Akt expression in the phenotypic progression of squamous head and neck tumorigenesis. Our results show that Akt activation is a significant prognostic factor for a shorter disease-free survival in patients with squamous carcinoma of the tongue, independent of stage and nodal status. The pattern of staining was mostly cytoplasmic, with a few samples also demonstrating nuclear staining. Exclusive cytoplasmic staining was observed at the invasive front of SCC, in accordance with data recently published.25 Scant cytoplasmic staining in a few cells in basal and parabasal layers was occasionally detected in normal epithelium adjacent to the tumor and in the normal epithelium of two squamous mucosa specimens from nonsmoking volunteers (2–3%). On the basis of this finding, the median labeling index value of 15% was considered appropriate as a cut-off for determination of positive staining. A similar cut-off was used in recently published studies.17, 27 In fact, some studies have reported p-Akt positivity in the majority of specimens based on the presence of any positivity, including nuclear staining.17, 27 In our study, using the median labeling index observed in our specimens as the cut-off value, 46% of the samples showed p-Akt expression.

We also investigated the role of Akt activation in preneoplastic lesions in the oral cavity in an effort to confirm recent in vitro findings that Akt activation is an early cellular response to nicotine and other tobacco-specific carcinogens exposure.24 Akt activation in bronchial dysplasia28 and pharmacologic inhibition of premalignant and malignant human bronchial epithelium growth by PI3K inhibitors29 support this notion. How this activation contributes to the tumorigenic potential of the lesions remains to be explored. In the setting of HNSCC, Akt activation has been correlated with squamous cell progression from normal epithelium to dysplasia and then to infiltrating carcinoma,25 and with concomitant PI3K accumulation and PTEN down-regulation, reflects an early biochemical effect in response to tobacco components such as nicotine.13 We observed a much higher percentage of Akt positivity in mildly dysplastic lesions as compared with moderate and severely dysplastic ones (Table 2). This might be due to the small sample of cases with severe dysplasia examined and also to the possibility that p-Akt might play a role in the initial stage of malignant transformation, but might not be necessary for acquisition of invasive properties by the premalignant cells, as has also been observed in lung tumorigenesis28 and in sporadic colon carcinogenesis.30 Finally, it is possible that although only a small subset of severely dysplastic lesions express p-Akt, these harbor the clones of cells that subsequently expand to give rise to cancer. In addition, if we consider all dysplasia as a continuous spectrum of changes, the overall frequency of p-Akt-positive is 45%, virtually identical to the frequency of p-Akt-positive in cancer. It is now becoming increasingly clear that the separation in terms of risk for malignant transformation between mild/moderate and severe dysplasia is actually not as important as the degree of genetic abnormalities or aneuploidy that these lesions harbor.31 As such, it is not surprising that we do not observe a stepwise increasing accumulation of p-Akt expression along the increasing degrees of dysplasia, as would be traditionally expected. In support of a role of the PI3K/Akt pathway activation in HNSCC biology is the finding also that in surgical margins of HNSCC patients overexpression of functionally active eIF4E, a downstream effector of the Akt/mTOR signaling pathway, has been found to be an independent risk factor for recurrence32 and was correlated with Akt activation; therefore, Akt activation might represent a valid predictive marker in the definition of the risk to develop second primary tumors (SPTs) and early recurrences in HNSCC.

The predictive role of Akt in HNSCC suggests that targeting Akt might be a useful strategy for therapy in this disease. Indeed, there are several lines of evidence suggesting that successful targeting of PI3K/Akt might lead to better therapeutic outcomes, especially in combination with EGFR inhibitors15, 33 or in enhancing chemotherapeutic efficacy.34

In conclusion, our findings suggest that targeting Akt might be of interest as part of a multimodality therapy in HNSCC. A larger cohort with a longer follow-up of preneoplastic lesions is needed to better define the role of Akt in tumorigenesis and integrate it in a risk model for cancer development.

Acknowledgements

The authors thank Dr. Gaetano De Rosa, Department of Pathology of University of Naples Federico II, for his role in collecting the Italian tongue cancer specimens, and Reuben Lotan for kindly providing the HNSCC cell lines used for this study.

Ancillary