Prognostic Significance of AKT/mTOR and MAPK Pathways and Antitumor Effect of mTOR Inhibitor in NF1-Related and Sporadic Malignant Peripheral Nerve Sheath Tumors

Malignant peripheral nerve sheath tumor (MPNST) is an uncommon soft tissue sarcoma with high metastatic potential and poor prognosis (1). Surgical resection is a curative treatment of choice for resectable MPNST, whereas no effective systemic therapy is currently available for unresectable or metastatic MPNST (2). The application of novel antitumor agents and precise prognostication are essential to improve the survival of patients with MPNST.

A genetic characteristic of MPNST is that approximately half of the cases develop from benign neurofibroma in neurofibromatosis type 1 (NF1) patients with NF1 gene deficiency (3). NF1 is a kind of tumor suppressor gene encoding neurofibromin, the dysfunction of which leads to multiple neurofibromas through Ras activation (4, 5). Ras activation, in turn, triggers the activation of 2 downstream pathways: the mitogen-activated protein kinase (MAPK) pathway and the AKT/mTOR pathway. The loss of neurofibromin activity is not sufficient for malignant transformation from neurofibroma to MPNST. This suggests that additional alterations in p14^ARF, p15^INK4b, p16^INK4a (6, 7), p53 (8), CHFR (9), CDK4, FOXM1 (10), hepatocyte growth factor (HGF; ref. 11), and MET (12) are necessary for the malignant transformation. However, the details of the molecular mechanism underlying the development of MPNST are poorly understood in NF1-related as well as sporadic cases.

The AKT/mTOR and MAPK pathways play important roles in modulating cellular functions in response to extracellular signals, such as growth factors and cytokines (13, 14). AKT is a serine/threonine kinase that is activated (phosphorylated) by phosphoinositide 3-kinase (PI3K) and that in turn activates (phosphorylates) the downstream molecules. mTOR is one of the downstream targets of AKT and is a key factor in the AKT/mTOR pathway. mTOR activates p70S6 kinase (p70S6K), directly or indirectly activates S6 ribosomal protein (S6RP), and inhibits 4E-binding protein 1 (4E-BP1). mTOR activation causes protein synthesis, which induces cell proliferation, survival, motility, invasion, and differentiation, and finally it can lead to cancer initiation and progression (15). The Ras/Raf/mitogen-activated protein kinase kinase (MEK)/extracellular signal–regulated kinase (ERK) pathway, also known as the MAPK pathway, also regulates a variety of cell functions including proliferation, growth, and survival (16). There is cross-talk between the AKT/mTOR and MAPK pathways; for example, Ras can activate the PI3K/AKT/mTOR pathway in addition to the Raf/MEK/ERK pathway; indeed, ERK can activate mTOR (17, 18).

The AKT/mTOR and MAPK pathways are known to be activated in some kinds of sarcomas (19–23), and several articles support that the AKT/mTOR and/or MAPK pathways, as well as their upstream receptor tyrosine kinases, are activated in MPNSTs (1, 24, 25). At least in patients with NF1, it is conceivable that Ras activation caused by Nf1 gene deficiency may induce the subsequent activation of the AKT/mTOR and Raf/MEK/ERK pathways. These pathways may play important roles in the initiation and progression of MPNST. Nonetheless, the clinicopathologic and prognostic significance of the activation status of the AKT/mTOR and MAPK pathways has not been investigated well in either NF1-related or sporadic MPNSTs.

Here, we conducted a large-scale clinicopathologic and prognostic analysis of the AKT/mTOR and MAPK pathways in a series of 135 MPNST clinical specimens. We then tested the antitumor activity of the mTOR inhibitor (Everolimus) on MPNST cell lines in vitro.

One hundred and thirty-five paraffin-embedded MPNST specimens from 104 patients were retrieved from the file of soft tissue tumors registered in the Department of Anatomic Pathology, Pathologic Sciences, Kyushu University (Fukuoka, Japan) between 1964 and 2010. The specimens included 91 primary, 39 recurrent, and 5 metastatic tumors. Specimens of the primary tumor were not available from 13 patients who visited our hospital at the time of recurrence. The diagnosis of MPNST was made according to the latest edition of the World Health Organization classification (26) and the textbook of Enzinger and Weiss (27). Clinical details and follow-up information were obtained by reviewing medical charts. The clinicopathologic characteristics are summarized in Table 1. The 91 primary cases included 46 males and 45 females whose ages ranged from 1 to 88 years (median 47 years). In all, 37 patients were diagnosed with NF1 based on the NIH criteria (28). Tumors were located in the extremities in 39 cases (thigh 18, upper arm 13, lower leg 6, and forearm 2), in the trunk wall in 23 cases (back 9, buttock 7, chest wall 5, and abdominal wall 2), in the spinal or paraspinal region in 11 cases, in the head and neck in 9 cases, in the retroperitoneum in 7 cases, and in the visceral organs in 2 cases. All primary tumors were treated surgically, accompanied by adjuvant chemotherapy or irradiation in 4 cases. The histologic tumor grade was assigned as either low-grade or high-grade (29). As for staging of the primary tumors, the latest American Joint Committee on Cancer (AJCC) staging system was used (30). Follow-up information was available in 88 of 91 primary tumor cases. The median follow-up period after surgery was 26 months (range, 3–291 months).

This study was conducted in accordance with the principles embodied in the Declaration of Helsinki. The study was also approved by the Ethics Committee of Kyushu University (No.21-137) and conducted according to the Ethical Guidelines for Epidemiological Research enacted by the Japanese Government.

Immunohistochemical staining was conducted in the same way as described previously (19). Sections were pretreated with Target Retrieval Solution (Dako) in a microwave oven at 100°C for 20 minutes before being incubated with monoclonal antibodies of phosphorylated-AKT (p-AKT; Ser473, 1:50 dilution), p-mTOR (Ser2448, 1:50), p-S6RP (Ser235/236, 1:75), p-p70S6K (Thr389, 1:50), p-4E-BP1 (Thr37/46, 1:400), p-MEK1/2 (Ser221, 1:50), and p-ERK1/2 (Thr202/Tyr204, 1:100) at 4°C overnight. All the above antibodies were supplied by Cell Signaling Technology. The immune complex was detected with the DAKO EnVision Detection System (Dako). Coexistent endothelial cells were evaluated as a positive internal control. As a negative control, the primary antibody was omitted. Immunohistochemical results were judged by 2 investigators (M. Endo and Y. Oda), who were blinded to the clinical status of the patients. A consensus judgment was adopted as the proper immunohistochemical result. Positive staining for individual markers was evaluated on the basis of its staining intensity. When the tumor cells showed cytoplasmic and/or nuclear staining with equal to stronger intensity compared with that of the endothelial cells, the expression was considered positive by reference to the previously published articles (22, 31). The serial sections were also immunostained with anti-Ki-67 antibody (M 7240, 1:100; Dako) using the standard procedure. The Ki-67–labeling index was calculated as described previously (32).

Five pairs of snap-frozen samples from the tumor and the surrounding normal tissue were available from the collection of our department. Fresh samples were processed in the same way as the previous article (7).

Human MPNST cell lines, HS-Sch-2 (33), HS-PSS, and YST-1 (34) were provided by the RIKEN BRC through the National Bio-Resource Project of the Ministry of Education, Culture, Sports, Science, and Technology (MEXT), Japan. FU-SFT8611 (35) and FU-SFT9817 (35) were established by M. Aoki and H. Iwasaki, and FMS-1 (36) was established by M. Hakozaki. All cell lines except FMS-1 were derived from sporadic MPNSTs. HS-Sch-2 and HS-PSS cells were cultured in Dulbecco's modified Eagle's medium (DMEM); YST-1 and FMS-1 cells were maintained in RPMI-1640; FU-SFT-8611 and FU-SFT-9817 are cultured in DMEM/F-12. All of these media preparations were supplemented with 10% FBS plus penicillin and streptomycin. Everolimus (RAD001) was provided by Selleck Chemicals and diluted in dimethyl sulfoxide (DMSO).

Western blot analysis was conducted as previously described (7, 37) using monoclonal antibodies of p-AKT (Ser473, 1:500 dilution), AKT (1:500), p-mTOR (Ser2448, 1:500), mTOR(1:500), p-S6RP (Ser235/236, 1:500), S6RP (1:1,000), p-p70S6K (Thr389, 1:500), p-4E-BP1 (Thr37/46, 1:500), p-MEK1/2 (Ser221, 1:1,000), p-ERK1/2 (Thr202/Tyr204, 1:500), and ERK1/2 (1:500). All the above antibodies were provided by Cell Signaling Technology.

For immunocytochemical staining, cultured cells on culture slide were washed with PBS, fixed in methanol/acetone for 20 minutes at 4°C and washed with PBS again. The endogenous peroxidase activity was blocked with methanol containing 0.5% hydrogen peroxide for 20 minutes. Then sections were treated at 4°C overnight with primary antibodies for each p-AKT, p-mTOR, p-S6RP, p-p70S6K, p-4E-BP1, p-MEK1/2, and p-ERK1/2 at the same dilution as used in immunohistochemistry. Subsequent reactions were carried out with the DAKO EnVision Detection System. A positive reaction was visualized with H₂O₂ containing 3,3′-diaminobenzidine as chromogen, and the sections were counterstained with hematoxylin.

MPNST cell lines HS-Sch-2, HS-PSS, YST-1, FU-SFT8611, FU-SFT9817, and FMS-1 were plated on 96-well plates at a concentration of 1,000 cells per well in serum-containing growth medium. Cells were treated with carrier alone (0.1% DMSO) or Everolimus (RAD001, Selleck Chemicals) with the indicated concentrations for 72 hours. Viability was assessed by WST-8 assay using the Cell Counting Kit 8 (CCK-8, Dojindo Molecular Technologies) according to the manufacturer's instructions and the previous article (38). The absorbance at 450 nm was measured by a microplate reader (Model 680 Microplate Reader, Bio-Rad Laboratories). All experiments were done in quadruplicate and repeated 3 times.

Wound-healing assay was conducted using MPNST cell lines HS-Sch-2, HS-PSS, FU-SFT8611, FU-SFT9817, and FMS-1. Confluent cell monolayers in 6-well plates were wounded by scraping with a micropipette tip. The cells were washed and then cultured in complete media containing the noted reagents and 30 nmol/L of Everolimus or carrier alone. The sizes of the scratches were imaged immediately (0 hours) and at 6 hours with a microscope (BZ-8000, Keyence). Each assay was conducted in triplicated and repeated 4 times.

Invasion assays were conducted with MPNST cell lines HS-Sch-2, HS-PSS, YST-1, FU-SFT8611, FU-SFT9817, and FMS-1 using the 24-well Biocoat Matrigel invasion chamber (BD Biosciences) according to the manufacturer's protocol. Briefly, cells were seeded into upper chamber at 1 × 10⁵ per chamber in serum-free media. Outer wells were filled with media containing 5% FBS. The chambers were treated with 30 nmol/L of Everolimus or carrier alone. Cells were incubated at 37°C with 5% carbon dioxide for 22 hours, and then noninvading cells were removed by wiping with a cotton swab. Cells that had migrated through the filter and adhered to its lower surface were fixed and stained using the Diff-Quik Kit (Sysmex), as described previously (39). The number of invading cells on membrane was counted in five microscopic fields (×400). Each assay was conducted in triplicate and repeated 3 times. Data are expressed as the percentage of invasion through the Matrigel matrix and membrane relative to the migration through the control membrane according to the manufacturer's manual.

Chi-square test or, when appropriate, Fisher exact test was used to evaluate the association between 2 variables. Overall survival was adopted as the endpoint for survival analyses. Survival curves were calculated with the Kaplan–Meier method, and the differences were compared by the log-rank test. Cox proportional hazards regression analysis was conducted to estimate the HRs for positive risk factors for death. In vitro data were analyzed by Student t test. Statistical significance was defined as P < 0.05. Data analysis was conducted with the JMP statistical software package (version 9.0.2; SAS Institute Inc.).

MPNST cells showed cytoplasmic and/or nuclear staining for p-AKT, p-mTOR, p-S6RP, p-p70S6K, p-4E-BP1, p-MEK1/2, and p-ERK1/2 antibodies (Fig. 1). In a case of MPNST arising from neurofibroma in a patient with NF1, positivity for each factor in the AKT/mTOR pathway showed a clear contrast between MPNST and the benign neurofibroma area: positive staining in the MPNST area and, in contrast, negative staining in the neurofibroma area (Supplementary Fig. S1).

Table 1 summarizes the correlation between immunoreactivity for each phosphorylated protein and each clinicopathologic variable. p-AKT, p-mTOR, p-S6RP, p-p70S6K, p-4E-BP1, p-MEK1/2, and p-ERK1/2 expressions were positive in 53 (58.2%), 43 (47.3%), 49 (53.8%), 52 (57.1%), 57 (62.6%), 85 (93.4%), and 74 (81.3%) of the 91 primary tumors, respectively. Among clinical variables, deep tumor location was associated with positive immunoreactivity for p-AKT (P = 0.034), p-mTOR (P = 0.0053), p-S6RP (P = 0.0076), p-4E-BP1 (P = 0.035), and p-ERK1/2 (P = 0.0024), respectively. There was no significant difference in positivity between NF1-related and sporadic MPNSTs. Among pathologic variables, positive expression of p-AKT was associated with a high Ki-67–labeling index (more than 30%). Positivity for p-p70S6K was correlated with rhabdomyoblastic differentiation. Also, positive staining for p-4E-BP1 and p-ERK1/2 was frequently observed in the subgroup with high mitotic counts (more than 10/10 HPF) and high histologic grade.

For each factor in the AKT/mTOR pathway, the correlation between immunostainings is summarized in Supplementary Table S1. There were significant correlations in positivity between the immunostainings for p-AKT, p-mTOR, p-S6RP, p-p70S6K, and p-4E-BP1.

Immunoreactivity for each antibody was confirmed by Western blot analysis (Supplementary Fig. S2). The expression levels in Western blot analysis corresponded closely to the levels in immunohistochemistry. p-AKT, AKT, p-mTOR, mTOR, and p-p70S6K expressions were higher in tumor samples than in the corresponding normal tissue in both NF1-related and sporadic MPNSTs, whereas this tendency was not apparent for p-MEK1/2 and p-ERK1/2.

Among the factors of the AKT/mTOR and MAPK pathways we investigated, positivity for p-AKT (P = 0.045), p-mTOR (P = 0.0039), and p-S6RP (P = 0.026) each showed a significant association with overall survival in univariate prognostic analysis (Table 2; Fig. 2A–C). Combinatorial prognostic analysis with p-AKT and p-mTOR expressions showed that positive staining for both p-AKT and p-mTOR meant a worse prognosis than if either or neither of them was positive (Fig. 2D). Furthermore, positive staining for more than 2 factors among p-AKT, p-mTOR, and p-S6RP indicated worse prognosis than the others (Fig. 2E). Meanwhile, the expressions of p-p70S6K and p-4E-BP1, which are downstream of the AKT/mTOR pathway, and expressions of p-MEK1/2 and p-ERK1/2, which are in the MAPK pathway, were not associated with overall survival (Table 2 and Supplementary Fig. S3). About NF1-related versus sporadic MPNST, the former showed worse prognosis than the latter (P = 0.029). Each of the following clinicopathologic variables was also associated with poor prognosis: tumor location in deep tissue, large tumor size (more than 5 cm), adjuvant therapy treatment, present tumor necrosis, high Ki-67–labeling index (more than 30%), epithelioid feature, high histologic grade, and advanced AJCC stage (stage II or higher).

A multivariate analysis was conducted with each of p-AKT, p-mTOR, and p-S6RP and with clinicopathologic variables that were related to poor prognosis in the univariate analysis except for tumor depth, tumor size, and histologic grade (Table 3 and Supplementary Table S2). Those 3 factors were excluded because AJCC stage was derived from them. The multivariate analysis revealed that each p-mTOR- and p-S6RP-positive expressions and high Ki-67–labeling index were the independent poor prognostic factors for overall survival (Table 3).

On the basis of the clinicopathologic and prognostic analyses showing the importance of AKT/mTOR/S6RP pathway activation to the survival of patients with MPNST, we chose Everolimus as an inhibitor for an in vitro experiment. Everolimus dose-dependently inhibited cell proliferation for all MPNST cell lines derived from both NF1-related and sporadic MPNSTs (Fig. 3A). When compared with no drug controls, decreased p-S6RP and p-p70S6K expressions by Everolimus [30 nmol/L, a dose achievable in humans by oral administration (40)] were confirmed by Western blotting (Supplementary Fig. S4) and immunocytochemistry (Supplementary Fig. S5). Everolimus administration showed no obvious effects on p-AKT, p-mTOR, p-MEK, and p-ERK1/2 expressions. A wound-healing assay showed a decrease in motility of all MPNST cell lines with Everolimus treatment (Fig. 3B). In a Matrigel invasion assay, Everolimus caused a decrease in the invasion of all MPNST cell lines except for FU-SFT9817, which was one of the sporadic MPNST-derived cell lines (Fig. 3C).

We conducted a large-scale and detailed clinicopathologic and prognostic assessment about AKT/mTOR and MAPK pathways in MPNSTs. To the best of our knowledge, our univariate prognostic analysis provided the first evidence that the activation of AKT, mTOR, and S6RP was associated with poor overall survival in MPNSTs. In addition, multivariate analysis identified p-mTOR and p-S6RP expressions as independent prognostic factors. These findings mean that, among AKT, mTOR, S6RP, p70S6K, 4E-BP1, MEK1/2, and ERK1/2, AKT/mTOR/S6RP activation is the most important process influencing prognosis in MPNSTs (Fig. 4). On the basis of these findings, in terms of AKT/mTOR and MAPK signaling pathways, the inhibition of S6RP activation by mTOR inhibitor seems to be one of the most ideal treatment targets for clinically aggressive MPNST, in which systemic therapy is more likely to be needed. Everolimus is an orally bioavailable derivative of rapamycin and forms a complex with the FK binding protein complex (FKBP-12), which binds with high affinity to mTOR (40). The Everolimus–mTOR interaction then causes the inactivation of p70S6K and S6RP. The advantage of Everolimus among AKT/mTOR/S6RP inhibitors is that it is already being used safely for patients with cancer in clinical settings (40). Our in vitro study showed that Everolimus had an antitumor effect on MPNST cell lines at a clinically available concentration. A single use of Everolimus inhibited not only cell proliferation but also cell motility and cell invasion on cell lines regardless of whether it was derived from sporadic or NF1-related MPNST. Also, several researchers have studied in vitro antitumor effect of mTOR inhibitors including Everolimus on MPNST cell lines (24, 41, 42). Johansson and colleagues reported that Everolimus (RAD001) induced apoptosis and inhibited cell proliferation on MPNST cell lines (41). Johannessen and colleagues revealed that rapamycin suppressed cyclin D1 expression, which led to cell-cycle arrest and inhibition of cell proliferation on MPNST cell lines (42). In their article, they also showed that angiogenesis was inhibited by rapamycin (42). Zou and colleagues described that cell proliferation of MPNST cell lines was inhibited by rapamycin (24). In addition, our study yielded the first evidence that mTOR inhibitor decreased cell motility and invasion on both NF1-related and sporadic MPNST cell lines, and also provided additional information on antiproliferation activity of Everolimus. In summary, mTOR inhibitors have been both characterized in the literature as leading to cell-cycle arrest by suppressing cyclin D1 expression, inducing apoptosis, and inhibiting cell proliferation and angiogenesis, and shown to repress cell motility and invasion in our study. Recently, it was suggested that the mTOR inhibitor can induce paradoxical AKT activation due to a negative feedback loop between p70S6K and the insulin-like receptor substrate-1 (IRS1; refs. 24, 43) and that it also can enhance MAPK pathway signaling (44), which leads to resistance to the mTOR inhibitor. Our in vitro study showed no obvious gain in the expression levels of p-AKT, p-MEK1/2, and p-ERK1/2 after 30 nmol/L of Everolimus was administered. On the basis of the findings of our in vitro study, it is reasonable to say that a single use of Everolimus is a potential novel drug therapy for sporadic and NF1-related MPNSTs. However, further in vivo and in vitro investigations in combination with other kinase inhibitors or cytotoxic agents are encouraged because the mTOR inhibitor might show stronger antitumor activity when administered in combination (24, 41, 45, 46).

This study revealed that AKT/mTOR/S6RP activation was not only suitable as a treatment target, but also useful for prognostication. Combinatorial evaluation of p-AKT and p-mTOR expressions or p-AKT, p-mTOR, and p-S6RP expressions enables us to estimate the duration of patient survival precisely. It is easy to apply this prognosis prediction method to routine pathologic examinations, because it is based on immunohistochemistry of formalin-fixed, paraffin-embedded specimens. It is reasonable to conclude that p-AKT, p-mTOR, and p-S6RP expressions are useful prognostic indicators of MPNST, either individually or in combination.

This is a large-scale clinicopathologic and prognostic study using whole tumor specimens. We used whole tumor specimens instead of tissue microarrays because phosphorylated proteins in the AKT/mTOR and MAPK pathways sometimes show heterogeneous staining. The clinicopathologic analysis revealed that the AKT/mTOR and MAPK pathways were more frequently activated in deeply located MPNSTs than in superficial ones. A deeply situated tumor is often difficult to resect completely and is one of the factors in poor prognosis for MPNSTs. This finding implies that a deeply situated MPNST should be a good candidate for an AKT/mTOR inhibitor, because elevated p-AKT was associated with greater sensitivity for mTOR inhibitor in an endometrial cancer model (47). 4E-BP1 and ERK1/2 activation was associated with both high-mitotic counts and high-histologic grade. This is reasonable because 4E-BP1 and ERK1/2 are thought to regulate cell proliferation closely. About a clinicopathologic study about portions of the AKT/mTOR and MAPK pathways, a few tissue microarray-based expression studies with a significant number of MPNST tumor samples have been reported (1, 24). The authors of those studies reported 91% positivity for p-MEK1/2, which is very consistent with our result (93.4%). They had the same results as ours that p-MEK1/2 expression failed to show prognostic significance. Also, their study showed that the AKT/mTOR pathway was less frequently activated than the MAPK pathway in MPNSTs; this was consistent with our result. They did not make survival analysis with AKT/mTOR expression data; however, they found that higher p-AKT expression in metastatic lesions compared with localized cases (24). Considering all the valuable data about the AKT/mTOR and MAPK pathways in clinical MPNST samples, in MPNSTs the MAPK pathway is more frequently activated than the AKT/mTOR pathway. However, the AKT/mTOR pathway was much more involved than the MAPK pathway in the malignant clinical behavior of MPNST. This implies that MAPK pathway activation, possibly induced by NF1 gene deficiency, is associated with the formation of neurofibroma and the initiation of MPNST. Meanwhile, the AKT-mTOR pathway plays an important role in the progression of MPNST. This hypothesis is supported by basic research on the relationship between Ras/Raf/ERK hyperactivation and dedifferentiation of Schwann cells, from which neurofibroma and MPNST are thought to originate (48).

Interestingly, there is no significant difference between NF1-related and sporadic MPNST clinical samples in the activation status of the AKT/mTOR and MAPK pathways. We could not find a clear difference in the expression levels of phosphorylated proteins between cell lines derived from NF1-related (FMS-1) and sporadic (FU-SFT8611, FU-SFT9817, HS-PSS, HS-sch-2, and YST-1) MPNSTs. About the pathophysiology of MPNST, the NF1-related MPNST has the NF1 gene deficiency, leading to RAS activation, whereas little is known about sporadic MPNST. Recent studies have revealed that many sporadic MPNSTs also display alterations in the NF1 gene (49, 50). That may explain why the AKT/mTOR and MAPK pathways are frequently activated in sporadic cases as well as in NF1-related cases.

In summary, we have clarified that mTOR inhibition is one of the ideal treatment options for both NF1-related and sporadic MPSNTs through a large-scale clinicopathologic and prognostic assessment of the AKT/mTOR and MAPK pathways and through an in vitro study using Everolimus on MPNST cell lines. Further in vivo and in vitro investigations are warranted to evaluate the efficacy of the mTOR inhibitor, either alone or in combination, in MPNST.

No potential conflicts of interest were disclosed.

Conception and design: M. Endo, Y. Dobashi, Y. Oda

Development of methodology: M. Endo, N. Setsu, K.-I. Iida, Y. Matsumoto, Y. Dobashi, Y. Oda

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Endo, H. Yamamoto, N. Setsu, T. Ishii, Y. Matsumoto, M. Aoki, H. Iwasaki, K. Nishiyama

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Endo, Y. Dobashi, Y. Oda

Writing, review, and/or revision of the manuscript: M. Endo, Y. Iwamoto, Y. Oda

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Endo, H. Yamamoto, K. Kohashi, Y. Takahashi, Y. Matsumoto, M. Hakozaki, H. Iwasaki, K. Nishiyama, Y. Iwamoto, Y. Oda

Study supervision: K. Kohashi, Y. Takahashi, Y. Dobashi, Y. Iwamoto, Y. Oda

The authors thank J. Kishimoto, Center for Clinical and Translational Research (CCTR), Kyushu University, for the excellent advice on the statistical analysis; N. Tateishi and K. Matsuda for the dedicated assistance; and the Research Support Center, Graduate School of Medical Sciences, Kyushu University for the technical support. We also thank KN International for revising the English used in this article.

Y. Oda is supported by a Grant-in-Aid for Scientific Research [(B); no. 21390107] from the Japan Society for the Promotion of Science (JSPS).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.