Prognostic Significance of Activated AKT Expression in Soft-Tissue Sarcoma

Soft-tissue sarcomas (STS) comprise a heterogeneous group of mesenchymal tumors with wide spectrum in histologic features (1, 2). More than 50 histologic classifications and subclassifications have been proposed for STS, in which malignant grade of tumors is various even among tumors of the same histologic categories (1, 2). Therefore, precise categorization of malignant potential in each case is necessary for choosing appropriate therapeutic modalities for patients with STS (3, 4). Previously, histologic variables such as degree of cellularity, cellular pleomorphism, mitotic activity, degree of necrosis in the tumors, and pattern of tumor growth have been reported to be prognosticators for STS (5, 6). Mitotic activity usually reflects proliferative potential of tumors and is commonly used for characterization of STS, but interobserver and intraobserver discrepancy in mitotic counting has been reported (7). Therefore, more objective markers such as labeling index (LI) for Ki-67 (8–11), proliferating cell nuclear antigen (12), and argyrophilic nucleolar organizer region counting (13, 14) have been introduced in the estimation of proliferative activity.

AKT is a serine/threonine kinase activated by various growth factors, such as epidermal growth factor and platelet-derived growth factor, and plays a central role in tumorigenesis through promotion of cell survival by inhibiting apoptosis and mediating cell proliferation (15–18). There are sequential steps in AKT activation, including (a) constitutive phosphorylation of Thr⁴⁵⁰, (b) translocation to the membrane induced by growth factors, and (c) phosphorylation of Thr³⁰⁸ and Ser⁴⁷³ at cell membrane (19). Phosphorylation of Thr³⁰⁸ is necessary for the AKT activation, which reaches a maximum level by phosphorylation of Ser⁴⁷³ (19).

AKT activation represents the signal transduction induced by various growth factors including forkhead in rhabdomyosarcoma (FKHR; ref. 15). Therefore, estimation of AKT activation state could be useful for prediction of progression and prognosis of STS. In fact, several molecules involved in the AKT signaling pathway are reported to be dysregulated in a wide spectrum of human cancers (20, 21). As for STS, constitutive AKT activation was reported in the rhabdomyosarcoma cell lines (22) and clinical samples from Kaposi's sarcomas in AIDS patients (23).

Extracellular signal–regulated kinase 1 and 2 (ERK1/2) is a key regulatory enzyme engaged in cell proliferation through phosphorylation of various important factors (24, 25). ERK1/2 activates other protein kinases such as p90 ribosomal S6 kinase (RSK) 1, RSK2, and RSK3 (26, 27). The RSK proteins subsequently phosphorylate proteins which are involved in transcriptional activation, such as cyclic AMP response element binding protein, c-Fos, and serum response factors (26, 27). In addition, ERK1/2 activates histone 3 and high mobility group 14, which promote packaging of DNA into chromatin, resulting in increase of accessibility of transcription factors to DNA binding sites (26). Furthermore, ERK1/2 translocates to the nucleus and directly phosphorylates c-Jun and c-Fos and activates transcription factor 2 (26).

In the present study, the expression of AKT was immunohistochemically examined in 93 patients with STS, and its relationship with clinicopathologic factors and its prognostic significance were evaluated. The phosphorylation status of FKHR was examined and compared with the activation status of AKT. In addition, expression of activated ERK1/2 and the proliferation marker Ki 67 (7) in tumor cells were examined.

Patients. Ninety-three of adult patients with localized STS in the trunk and extremities were selected for the present study: they were treated at the Orthopedic Divisions of Osaka University Hospital, Osaka Center for Cancer and Cardiovascular Diseases, and Osaka National Hospital during the period from July 1992 to August 1999. There were 53 males and 40 females with ages ranging from 19 to 77 years (median, 57 years). Tumors were located in the upper extremities in 22 patients, in the trunk in 21, and in the lower extremities in 50. Wide local excision was done in 55 patients and amputation in 2. Thirteen patients did not receive wide local excision because of neurovascular involvement in seven patients, huge size of the tumor in five, and preoperative diagnosis of benign tumor in one. The remaining 23 patients received marginal or intralesional excision at other hospitals initially.

Surgically resected specimens were fixed in 10% formalin and routinely processed for paraffin embedding. Histologic sections cut at 4-μm thickness were stained with H&E and reviewed by three pathologists (Y.T., Y.H., and K.A.). Periodic acid Schiff reaction, silver impregnation, and immunohistochemical procedures (avidin-biotin complex method) were done if necessary for diagnosis. Histologic classification as well as histologic factors such as mitotic count, cellularity, pleomorphism, necrosis, sclerotic change, and myxoid change of the tumors was evaluated. Distribution of histologic classification of the tumors is shown in Table 1.

Adjuvant chemotherapy was done in 47 patients (preoperative in 8, postoperative in 9, and both preoperative and postoperative in 30). Chemotherapeutic agents used were doxorubicin or its analogues alone in one patient; cisplatinum [cis-diammine-dichloroplatinum (CDDP)] or its derivative alone in one; ifosphamide alone in three; doxorubicin and CDDP in seven; doxorubicin and ifosphamide in seven; ifosphamide and etoposide in one; doxorubicin, CDDP, and ifosphamide in 14; doxorubicin, ifosphamide, and etoposide in one; doxorubicin, ifosphamide, and actinomycin D in one; ifosphamide, vindesine, doxorubicin, and dacarbazine (IFOVADIC) in three; IFOVADIC and CDDP in eight; and IFOVADIC, CDDP, and etoposide in one. Radiotherapy was done in 7 of 36 patients who received marginal or intralesional excision for the primary tumor and in 11 of 57 who received wide local excision or amputation.

Immunohistochemistry for phosphorylated AKT, phosphorylated ERK1/2, phosphorylated FKHR, and Ki-67. The specificity of anti–phosphorylated AKT (Thr³⁰⁸) (anti-p-AKT) polyclonal antibody (Cell Signaling Technology, Beverly, MA) was examined by Western blot analysis for the extracts from Jurkat cells, showing positive band after platelet-derived growth factor treatment (data not shown). The antibody preincubated with the antigen peptide was used in the negative control test and uniformly gave negative results.

Immunoperoxidase procedure (avidin-biotin complex method) was done on paraffin-embedded sections. Antigen retrieval was done with heating the sections in 10 mmol/L citrate buffer for 5 minutes. Anti-p-AKT (Thr³⁰⁸) polyclonal antibody, anti–phosphorylated p44/42 ERK1/2 (Thr²⁰²/Tyr²⁰⁴) (anti-p-ERK1/2) monoclonal antibody (Cell Signaling Technology), anti–phosphorylated FKHR (Ser²⁵⁶) polyclonal antibody (Cell Signaling Technology), and anti-Ki-67 (clone MIB-1) monoclonal antibody (Dako Cytomation A/S, Copenhagen, Denmark) were used as the primary antibodies at dilutions of 1:100, 1:100, 1:100, and 1:50, respectively. Sections were counterstained lightly with methyl green. For negative controls, immunohistochemistry without using the primary antibodies was done.

Tumor cells showed cytoplasmic and/or nuclear staining for p-AKT antibody. Intensity of staining was categorized as follows: level 1, weaker staining in either nucleus or cytoplasm of tumor cells compared with those in the endothelial cells; level 2, equal to or stronger staining in both the nucleus and cytoplasm. This categorization proved to be more significant than that with intranuclear or cytoplasmic staining alone. Immunohistochemistry with p-ERK1/2 revealed the cytoplasmic staining in the tumor cells. Staining intensity was categorized as follows: level 1, weaker staining than that in the endothelial cells; level 2, equal to or stronger staining. Staining intensity of phosphorylated FKHR (p-FKHR) in the cytoplasm of tumor cells was categorized in comparison with that in the endothelial cells as follows: level 1, weaker staining; level 2, equal to or stronger staining. For p-AKT, p-ERK1/2, and p-FKHR immunohistochemistry, cases showing level 1 and 2 staining in separate areas of the tumor were categorized as level 2. Cases with negative endothelial staining were regarded as poor antigen preservation and were not included in the analysis.

For Ki-67 immunohistochemistry, cells showing intranuclear staining were judged as Ki-67 positive. The Ki-67-positive cells among 200 tumor cells were counted and the percentage was shown as the Ki-67 LI. Cases were divided into two groups: level 1, Ki-67 LI < 20%; level 2, Ki-67 LI ≥ 20%.

Follow-up. Follow-up period for survivors ranged from 10 to 132 months (median, 65 months). Actuarial overall and disease-free 5-year survival rates were 79.5% and 40.5%, respectively. Distant metastasis was found in 28 (30.1%) patients, of whom 17 (60.7%) died due to tumor. Local recurrence occurred in 45 (48.4%) patients, of whom 10 died due to tumor. Twenty-two of these 45 patients were initially admitted to other hospitals and received marginal excision; thereafter they were admitted to Osaka University Hospital, Osaka Center for Cancer and Cardiovascular Diseases with recurrent STS. Rate of local recurrence by types of surgery was 0% for amputation, 21.8% for wide local excision, 91.2% for marginal excision, and 100% for intralesional excision.

Statistics. Statistical analysis was done with the JMP software (SAS Institute, Inc., Cary, NC). Correlation among each clinicopathologic factors was evaluated by χ² test. Overall and disease-free survival rates were calculated by the Kaplan-Meier method (28) and difference in survival curves was analyzed by the log-rank test. Independent prognostic factors were analyzed by the Cox proportional hazards regression model in a stepwise manner (29). P < 0.05 was considered to be statistically significant.

p-AKT, p-ERK1/2, and p-FKHR expressions in STS. Tumor cells in 12 cases showed a constant level 2 staining for p-AKT (Fig. 1B) whereas 14 cases showed combined level 1 and 2 staining. As a total, 26 cases (28.0%) were regarded as level 2 p-AKT expression. The remaining 67 cases (72.0%) showed constant level 1 staining throughout the tumors (Fig. 1A). Eighty-seven (93.5%) and 6 (6.5%) cases showed level 1 and level 2 staining for p-ERK1/2, respectively (Fig. 1D and E). p-FKHR immunohistochemistry was done in 60 cases: 52 (83.3%) and 8 (16.7%) cases showed level 1 and level 2 staining, respectively (Fig. 1G and H). Immunohistochemistry without primary antibodies gave no staining. Correlation between p-AKT and p-ERK1/2 expressions and other clinicopathologic factors was shown in Table 2. Tumors with level 2 p-AKT expression showed the higher ratio of level 2 p-ERK1/2 and p-FKHR expressions (P < 0.05 and P < 0.01), and cases with level 2 p-ERK1/2 expression showed the higher ratio of huge tumor size (> 10 cm; P < 0.05). No significant correlation was observed among other factors.

Univariate and multivariate analyses for prognostic factors of STS. Patients with level 1 p-AKT expression showed a significantly better 5-year survival rate (85.6%) than those with level 2 expression (63.2%; P < 0.05; Fig. 2). Univariate analysis revealed that mitotic count, cellularity, necrosis, Ki-67 LI, and p-AKT and p-ERK1/2 expressions are prognostically significant for overall survival, and p-AKT and p-ERK1/2 expressions for recurrence-free survival (Table 3). Correlation between expression level of p-AKT and p-ERK1/2 and recurrence-free survival became far more distinct among patients who received the wide local resection and amputation (Fig. 3).

Multivariate analysis with factors proven to be significant in the univariate analysis revealed Ki-67 LI and p-AKT expression to be independent prognosticators for overall survival, and p-AKT expression for recurrence-free survival (Table 4).

Stratification of patients with Ki-67 LI and p-AKT and p-ERK1/2 expressions. Cases were stratified in combination with Ki-67 LI and p-AKT and p-ERK1/2 expressions. They were classified into four groups according to the number of level 2 categories of the three factors: group A (without level 2 category for all of Ki-67 LI and p-AKT and p-ERK1/2 expressions), group B (one level 2 category), group C (two level 2 categories), and group D (three level 2 categories). The 5-year survival rates of patients with group A, B, C, and D STS were 55.0%, 37.0%, 0.0%, and 0.0% for disease-free survival (P < 0.001) and 93.9%, 79.3%, 39.6%, and 50.0% for overall survival (P < 0.0001), respectively (Fig. 4).

Five-year overall survival rate (79.5%) in the present series is similar to that in the recent reports from Western countries (3, 4). In addition, patients' characteristics such as gender, age, and tumor location in the present series were similar to those in our previous study (13) and reports from Japan (30) and other Asian (31) and Western countries (3, 4). As for histologic subtypes, however, there was a marked difference between the previous (13) and the present series. Especially, frequency of malignant fibrous histiocytoma among all STS in the present series (26%) showed a marked decrease compared with our previous study (37%). This is due to the change of criteria for diagnosis of malignant fibrous histiocytoma; tumors showing pleomorphic and poorly differentiated histology with no findings indicative of specific tissue differentiation tended to be classified as malignant fibrous histiocytoma previously on the H&E-stained sections. At present, specific differentiation of tumor cells was carefully searched with aid of immunohistochemichal analysis; then the number of cases diagnosed as leiomyosarcomas, liposarcomas, or malignant neurogenic sarcomas increases (32) whereas the distribution of histologic variables, such as mitotic count, cellularity, and necrosis, is almost similar between the previous (13) and the present series.

Both AKT and ERK are major targets downstream of various oncoproteins such as growth factor receptor tyrosine kinases, Ras and Raf (18–20, 24, 25). AKT activation works antiapoptotically through prevention of the release of cytochrome c from mitochondria, inactivating forkhead transcription factors, such as FKHR, and proapoptotic factors, such BAD and procaspase 9, and activating IκB kinase (33). AKT progresses cell cycle through stabilization of cyclin D expression and inhibition of p27^Kip1 activity (33). Furthermore, AKT promotes tumor cell invasiveness through induction of matrix metalloproteinase production or promotion of angiogenesis by increasing the activation of endothelial nitric oxide synthase (33). In the present analysis, p-AKT expression correlated with p-FKHR expression, but not with factors involved in the proliferation such as mitotic count and Ki-67 LI. Taken together, antiapoptotic signal, but not proliferative stimulation, might contribute to the poor prognosis of STS showing AKT activation.

In the present analysis, p-AKT expression in the cytoplasm was a prognostic marker for both disease-free and overall survival. Patients with strong intranuclear p-AKT expression showed worse prognosis compared with those with weaker expression, although the difference was not significant. Prognosis was worst in patients with strong p-AKT expression in both the nucleus and cytoplasm. Thus, this categorization was chosen in the present study. Importance of intranuclear localization of p-AKT in the phosphorylation of transcription factors for genes that have opposing effect on cell death and promote cell growth has been reported (34).

The influence of p-AKT and p-ERK1/2 expressions on the recurrence of tumors was more marked in patients receiving wide local excision and amputation. Recently, the local recurrence of STS was reported to be an indicator of tumor aggressiveness and poor prognosis, particularly in patients receiving adequate resection (35). In the present analysis, p-AKT and p-ERK1/2 expressions proved to be predictors of local recurrence; thus, these could be used as markers for aggressiveness of STS. Multivariate analysis revealed Ki-67 LI and p-AKT expression to be independent prognosticators for overall survival of STS. Ki-67 antigen is expressed in cells in the cell cycle except at G₀ phase, and thus is used as a marker of cell proliferation (8). Combination of p-AKT and p-ERK1/2 expressions and Ki-67 LI proved to be a useful tool in the stratification of cases with STS.

In conclusion, stratification of STS cases with combination of p-AKT expression and Ki-67 LI predicts the biological behavior of tumors. p-AKT expression as determined by immunohistochemistry could be a new prognosticator for STS.