PTEN Mutation Located Only Outside Exons 5, 6, and 7 Is an Independent Predictor of Favorable Survival in Endometrial Carcinomas

Endometrial carcinoma is the most common malignancy of the female reproductive tract (1). Recently (2, 3, 4, 5, 6, 7), some of the genes contributing to development of endometrial carcinoma have been elucidated, including p53, KRAS, β-catenin, and PTEN. Among them, PTEN is the most frequently mutated gene ever reported in endometrial carcinoma (34–55%; Refs. 5, 6, 7). The tumor suppressor gene PTEN was identified on chromosome 10q23.3 that is homozygously deleted in many human malignancies (8, 9). Germ-line PTEN mutations have been detected in patients with autosomal dominant cancer predisposition syndromes: Cowden disease, Lhermitte-Duclos disease, and Bannayan-Zonana syndrome (10, 11). Somatic mutations and deletions of this gene have been reported in many types of sporadic tumors, including endometrial cancers (5, 6, 7), glioblastomas (8, 9), prostate cancers (12), and melanomas (13). The PTEN gene encodes a phosphatidylinositol phosphatase, and the product is reported to induce apoptosis and G₁ cell cycle arrest through antagonizing phosphatidylinositol 3′-kinase/Akt-mediated cell growth pathway (14). An aggressive phenotype in some types of tumors is reported to be associated with alteration in this gene (15, 16, 17). The prognostic significance of PTEN mutations in endometrial carcinoma is also beginning to be analyzed. A limited number of recent studies (18, 19) have shown that PTEN mutation is associated with favorable prognosis of patients with this disease. However, no studies have ever described the prognostic significance of mutated exons of PTEN. The current study investigated mutations of the PTEN gene and the prognostic significance of mutated exons in endometrial carcinomas from 67 Japanese patients. This is the first report presenting the prognostic impact of mutated exons in the PTEN gene. The relationship between PTEN mutation and clinicopathological characteristics was also examined in the present study.

Snap-frozen tumor samples were obtained from 67 Japanese patients who underwent treatment for primary endometrial carcinomas at the University of Tokyo Hospital between 1992 and 2000. All of the patients provided informed consent for the research use of their samples. Cellular DNA was extracted by a standard SDS-proteinase K procedure. The clinical status was known for all of the patients, with a mean follow-up duration of 47.2 months. The mean age of the 67 patients was 56 years ranging from 30 to 79 years. Staging of tumors, based on the FIGO² criteria, was as follows: 30 cases, stage I (T₁, N₀; tumor limited to corpus); 8 cases, stage II (T₂, N₀; tumor involving cervix but not extending outside uterus); 22 cases, stage III (T₃ or N₁; tumor extending outside uterus, including spread to vagina, but remaining within true pelvis or metastases to regional lymph nodes); and 7 cases, stage IV (T₄ or M₁; tumor invading bladder or bowel mucosa or distant metastases; Ref. 20). Histological classification was conducted according to the WHO criteria (21). The histological subtypes of tumors were endometrioid adenocarcinomas (59 cases), adenosquamous carcinomas (3 cases), clear cell carcinoma (1 case), squamous cell carcinoma (1 case), and mixed carcinomas (3 cases). Endometrioid adenocarcinomas were further subclassified according to their degree of histological differentiation into three grades (G₁, G₂, and G₃) using the criteria of FIGO classification (20).

The operative procedure included hysterectomy, bilateral salpingo-oophorectomy, and systematic aortic and pelvic lymph node dissection. Radical hysterectomy was performed on patients with positive findings on presurgical endocervical curettage, and total hysterectomy was performed on the remainder, who were negative for endocervical curettage. Postsurgically, patients with positive peritoneal cytology, adnexal/peritoneal involvement, or aortic node metastases were treated with combination chemotherapy, including cyclophosphamide, doxorubicin, and cisplatin. Whole-pelvis irradiation was indicated when at least one of the following factors existed: adnexal/peritoneal involvement, pelvic lymph node metastases, and deep muscular invasion (more than two-thirds depth in endometrioid G₁; more than one-third in other types or grades). Periaortic irradiation was administered to aortic node-positive patients.

The nine exons of the PTEN gene were amplified using 11 intron-based primer pairs. Exons 5 and 8 divided were amplified with two overlapping primer pairs. Primers for the exons were as follows: exon 1, CAG AAG AAG CCC CGC CAC CAG and AGA GGA GCA GCC GCA GAA ATG; exon 2, TGA CCA CCT TTT ATT ACT CC and TAC GGT AAG CCA AAA AAT GA; exon 3, ATA TTC TCT GAA AAG CTC TGG and TTA ATC GGT TTA GGA ATA CAA; exon 4, TTC AGG CAA TGT TTG TTA and CTT TAT GCA ATA CTT TTT CCT A; exon 5, AGT TTG TAT GCA ACA TTT CTA A and TTC CAG CTT TAC AGT GAA TTG (first pair) and GAC CAA TGG CTA AGT GAA GAT and AGC AAC TAT CTT TAA AAC CTG T (second pair); exon 6, TTG GCT TCT CTT TTT TTT CTG and ACA TGG AAG GAT GAG AAT TTC; exon 7, ACA GAA TCC ATA TTT CGT GTA and TAA TGT CTC ACC AAT GCC A; exon 8, ACA CAT CAC ATA CAT ACA AGT C and GTG CAG ATA ATG ACA AGG AAT A (first pair) and TTA AAT ATG TCA TTT CAT TTC TTT TTC and CTT TGT CTT TAT TTG CTT TGT (second pair); and exon 9, AAG ATG AGT CAT ATT TGT GGG T and GAC ACA ATG TCC TAT TGC CAT (6, 22). Each 20-μl reaction mixture for PTEN amplification contained 100 ng of genomic DNA, 250 μm of each deoxynucleotide triphosphate, 1 × PCR Buffer II (Perkin-Elmer), 2.5 mm MgCl₂, 0–10% DMSO, 0.5 unit of AmpliTaq Gold (Perkin-Elmer), and 1 μm of each primer. PCR amplifications were performed in a Perkin-Elmer Cetus 9600 thermocycler with denaturation at 94°C for 10 min, followed by 35 cycles of 94°C for 30 s, 44–55°C for 30 s, 72°C for 30 s, and final extension at 72°C for 10 min.

PCR products were diluted 1:1 with loading buffer (95% formamide, 0.05% xylene cyanol), denatured for 5 min at 95°C, and then cooled on ice. Five μl of each sample were electrophoresed using precast 12.5% acrylamide gels, buffer strips, and GenePhor electrophoresis unit (Pharmacia Biotech). Silver staining was then performed using Hoefer automated gel stainer (Pharmacia Biotech) with PlusOne DNA silver staining kit (0.2% silver nitrate, 0.7% benzene sulfonic acid; Pharmacia Biotech). Aberrant bands revealed by single-strand conformational polymorphism analysis were excised from the gel and amplified under the PCR conditions described above. The products were purified with a Suprec-01 (TaKaRa) and submitted to the Sawady Technology (Tokyo, Japan) for direct sequencing.

Survival curves were constructed using the Kaplan-Meier method, and differences were tested using the log-rank statistic (23). The Cox proportional hazard model was used for the univariate and multivariate analyses (24). The association of variables was evaluated by the χ² test or the Fisher’s exact test, where appropriate. In statistical analyses, patients with multiple PTEN mutations were dealt with as follows: e.g., a patient with mutations in exon 5 and 8 was grouped as “mutant in exon 5, 6, or 7” and a patient with mutations in exon 3 and 8 was grouped as “mutant only outside exons 5–7.”

We screened the PTEN gene in 67 endometrial cancers and found 47 mutations in 37 (55%) of these specimens (Table 1). Eight cases had two mutations and one case had three mutations in this gene. There were 27 frameshift mutations, 13 missense mutations, six nonsense mutations, and one deletion involving a splicing donor site. All of the 13 missense mutations were confirmed by the absence of alterations in corresponding constitutional DNA sequences. Mutations were distributed in the following manner: 18 mutations in exon 8, nine mutations in exon 7, seven mutations in exon 5, five mutations in exon 6, five mutations in exon 3, two mutations in exon 1, and one mutation in exon 2 (Fig. 1). No mutations were detected in exon 4 and 9. High frequencies of frameshifts and mutations in exon 8 were consistent with other published observations (5, 6, 7, 18, 19) in endometrial carcinomas.

Fig. 2,A shows the Kaplan-Meier survival curves for both groups of patients, PTEN mutant versus PTEN wild type. Mutation of the PTEN gene was associated with a trend toward better survival, although the data were not statistically significant (P = 0.32; Fig. 2,A). To examine the association between survival and mutated region of the PTEN gene, we compared survival rates based on mutated exons. The 4-year survival rate for each mutated exon was as follows: exon 1, 2, and 3, 100%; exon 8, 92%; exon 6, 80%; exon 5, 69%; and exon 7, 48%. The Kaplan-Meier method revealed that mutation only outside exons 5–7 correlated with a trend toward better survival and that the survival rate for the group with mutations in exon 5, 6, or 7 was almost the same as that for the group with wild-type PTEN (5-year survival rates for patients with PTEN mutations only outside exons 5–7, in exon 5, 6, or 7, and with wild-type PTEN were 93%, 65%, and 66%, respectively; Fig. 2,B). Fig. 2,C shows the survival curves for the group with mutations only outside exons 5–7 and the remaining patients’ group. Mutation only outside exons 5–7 was found to be associated with significantly better survival (P = 0.04 by log-rank test; Fig. 2,C). We also compared survival rates in the subset of advanced-stage disease (stage III/IV). Mutation only outside exons 5–7 in advanced cases was associated with a trend toward better survival, although this could not reach statistical significance (5-year survival rates for patients with mutation only outside exons 5–7 and the remaining patients were 83% and 45%, respectively; P = 0.13 by log-rank test; Fig. 2 D).

Next, univariate analysis was performed for various favorable prognostic factors including age (≤60 versus >60), histological grade (G₁ versus G_2–3/nonendometrioid), FIGO stage (I/II versus III/IV), status of the PTEN gene, and PTEN mutation only outside exons 5–7 (Table 2). G₁, stage I/II, and PTEN mutation only outside exons 5–7 were found to be significant indicators of increased survival (P = 0.001, 0.0007, and 0.02, respectively; Table 2). Subsequently, multivariate analysis was conducted using the three favorable prognostic factors, which were shown to be significant in the univariate analysis. G₁, stage I/II, and PTEN mutation only outside exons 5–7 were found to be significant and independent predictors for improved survival (P = 0.0006, 0.004, and 0.004, respectively; Table 2).

Table 3 shows the relationship between PTEN mutation and clinicopathological features in 67 endometrial carcinomas. G₁ tumors tended to be less frequently mutated in the PTEN gene (χ² for trend, 2.9; P = 0.09). In addition, PTEN mutation tended to occur more frequently in patients who were postmenopausal and had low BMI (χ² for trend = 1.8; P = 0.18 and 2.7; 0.10, respectively). However, these data were not statistically significant. There was no relationship between PTEN mutation and age, gravidity/parity, diabetes mellitus, peritoneal cytology, muscular invasion, cervical/adnexal/peritoneal involvement, lymph nodes metastases, or FIGO stage. We further evaluated the association between clinicopathological features and mutation only outside exons 5–7 of PTEN, which was found to be a significant prognostic factor in the survival analysis. However, no significant association was observed. Variables of histology (endometrioid versus nonendometrioid), grade (G₁ versus G_2–3), and the expression of estrogen or progesterone receptors (positive versus negative) were also examined, but no correlation with PTEN mutation in any exon or mutation only outside exons 5–7 was found (data not shown).

PTEN mutation was detected in 37 of 67 (55%) endometrial carcinomas from Japanese patients, and their mutation frequency and distribution were consistent with published findings (5, 6) on Caucasian patients. First, we compared survival rates based on the PTEN status. PTEN mutation was found to be associated with a trend toward better survival in keeping with previous studies (18, 19), although this was not statistically significant. Further comparison of survival rates according to mutated exons revealed that mutation only outside exons 5–7 was associated with significantly favorable prognosis. Subsequent univariate and multivariate survival analyses indicated that mutation only outside exons 5–7 was an independent and significant predictor of favorable survival. Next, we examined the relationship between mutation only outside exons 5–7 and clinicopathological features, but no significant correlation was observed. These findings suggest that mutation only outside exons 5–7 of PTEN might represent a molecular predictor of favorable survival, independent of clinical and pathological characteristics of tumors.

As mentioned above, the patients were treated according to the status of the following factors: peritoneal cytology, muscular invasion, cervical/adnexal/peritoneal involvement, and aortic/pelvic lymph node metastases. There were no differences in these factors between the group with mutations only outside exons 5–7 and the remaining patients (Table 3). Therefore, the therapeutic modalities could not contribute to the different outcome. In fact, there were no differences in the number of patients treated postsurgically with chemotherapy or radiotherapy between the group with mutations only outside exons 5–7 and the remainder (chemotherapy, 6 of 18 versus 22 of 49; P = 0.4; radiotherapy, 10 of 18 versus 23 of 49; P = 0.5; chemotherapy or radiotherapy, 10 of 18 versus 32 of 49; P = 0.5; χ² test).

According to the study by Lee et al. (25), PTEN crystal structure shows that PTEN consists of a phosphatase domain and a C2 domain, which are both important for tumor suppressor function. Exons 5 and 6 contain WPD loop, P loop, and TI loop, which are making the active site pocket of the phosphatase domain (Fig. 1; Ref. 25). Exon 7 contains CBR3 loop, which is thought to play a central role in phospholipid membrane binding of the C2 domain (Fig. 1; Ref. 25). In fact, mutation on the WPD, P, or TI loop of recombinant PTEN protein reduces its phosphatase activity compared with wild-type PTEN, and mutation on the CBR3 loop reduces affinity for membranes in vitro (25). According to the review of PTEN mutational spectra by Ali et al. (26), most mutations in exons 5 and 6 are nonsense mutations located around WPD, P, or TI loop, and most mutations in exon 7 are frameshift or nonsense mutations. It is possible that mutations involving the WPD, P, TI, or CBR3 loop disrupt the tumor suppressor function of PTEN more completely and thus contribute to the development of tumor with a more virulent phenotype. In the present study, however, the survival rate of patients with mutation in exon 5, 6, or 7 was similar to that of patients with wild-type PTEN. Now, we are analyzing the alterations of other tumor suppressor genes and oncogenes in the PTEN wild-type tumors to identify the reason for this observation.

Although data presented here indicated that mutation in exon 8 was associated with better survival, Lee et al. (25) have shown that mutation on the cα2 element in exon 8 also reduces the in vitro membrane affinity of PTEN, and most alterations in exon 8 are reported to be frameshift mutations that result in truncation of the cα2 element (Fig. 1; Refs. 25, 26). According to Lee et al. (25), however, the cells expressing PTEN mutant of the phosphatase active site show extensive proliferation similar to the control cells, whereas CBR3 and cα2 mutants show intermediate growth suppressive activities. Most mutations in exon 7 are reportedly frameshift or nonsense mutations (26). Those mutations in exon 7 result in truncation of both the CBR3 loop in exon 7 and the cα2 element in exon 8 (Fig. 1). It is possible that loss of both the CBR3 and cα2 elements in the C2 domain may be required to disrupt the tumor suppressor function to the extent comparable with mutation of the phosphatase active site. This may explain why mutations in exons 7 and 8 showed different prognostic significance in the present study.

The current study investigated the relationship between PTEN mutation and clinicopathological features but failed to find any correlation between PTEN mutation and estrogen-related features. It is believed that there are two different pathogenetic types of endometrial carcinomas: estrogen-dependent type I and estrogen-independent type II (27). Type I tumors, most of which are low-grade endometrioid adenocarcinomas histologically, tend to occur in younger premenopausal women and have a favorable prognosis. Exposure to unopposed estrogen, either endogenous or exogenous, is thought to contribute to the development of type I tumors. Type II is composed of high-grade serous adenocarcinomas or clear cell carcinomas histologically. In contrast with type I, type II is likely to occur in older postmenopausal women and has a poor prognosis. A few recent studies (18, 19) have indicated that PTEN mutation is associated with low-grade endometrioid histology and favorable prognosis of patients. These previous findings suggest that mutation in this gene could be involved in development of type I tumors. However, the present study failed to find any correlation between PTEN mutation and features of type I tumors, i.e., endometrioid histology, G₁, and premenopause, or risk factors related to prolonged unopposed stimulation with endogenous estrogen, i.e., nulliparity, high BMI, and diabetes mellitus. History of estrogen replacement therapy without progestins, unopposed exogenous estrogen, did not correlate with PTEN mutation either (data not shown). These findings suggest that PTEN alteration may be contributing to carcinogenesis in the endometrium through an estrogen-independent pathway. However, larger epidemiological studies and further molecular research are needed to clarify this issue.

In conclusion, our data suggest that mutation located only outside exons 5–7 of PTEN represents a molecular predictor of favorable survival, independent of tumor spread or biological characteristics of tumors. Screening PTEN mutations in endometrial carcinoma could provide useful prognostic and therapeutic information in addition to what we can obtain from clinical and pathological data; e.g., the decision whether to perform aggressive resection of metastatic tumor or to offer chemotherapeutic regimens for tumor dormancy in cases of advanced stage may be affected by the PTEN mutation site. We believe that additional extensive studies on PTEN genetic alterations in endometrial carcinoma will help clarify the tumor suppressor function of PTEN and also give great benefit to patients with this disease.