A and B, PDGFB expression was detected in the tumor cells and in the stroma. C, PDGFB was peripherally expressed in patient 3. D, PDGFB was focally expressed in patient 21 or diffusely expressed in DFSP. A-D, 3,3′-diaminobenzidine. Scale bars indicate 100 μm (A and B) and 1000 μm (C and D).
A and B, Patient 28: PDGFB (A) and PDGFRβ (B) were expressed in the same regions. C and D, Patient 23 PDGFB (C) was expressed peripherally, whereas PDGFRβ (D) was expressed centrally. A and C, 3-amino-9-ethylcarbazole; B and D, 3,3′-diaminobenzidine. Scale bar indicates 1000 μm (A-D).
A, Amplification products were detected using the COL1A2 exons 1, 5, 11, 17, and 23 and PDGFB exon 2 primers. B, The fusion protein was generated with a matched reading frame between COL1A2 exon 26 and PDGFB exon 2. bp indicates base pair.
A and B, Tumor cells expressed CD34 (A [3,3′-diaminobenzidine] and PDGFB (B [3-amino-9-ethylcarbazole]). Tumor cells at the margins expressed higher levels of PDGFB (B [arrowhead] and C), whereas nontumorous fibroblasts did not (B [asterisk] and D). C and D, 3-amino-9-ethylcarbazole. Scale bars indicate 1000 μm (A and B) and 100 μm (C and D).
eTable 1. Summary of Clinicopathologic Findings for 30 Patients With DFSP
eTable 2. Antigen Retrieval Procedures for Immunohistochemical Analysis
eTable 3. The Expression of PDGFB, PDGFRß, PDGFRa, CD34, Factor XIIIa, Nestin, Fibronectin, a-SMA, S-100 Protein, and Ki-67 in 30 DFSP and 48 Non-DFSP Mesenchymal Tumors
eTable 4.COL1A1, COL1A2, COL3A1, COL5A1, COL5A2, COL5A3, Elastin, Lamininß1, Fibronectin1, Fibrillin1, Fibrillin2, Fibulin5, Tenascin XB, and PDGFB Primers Used for Fusion Gene Analysis
eFigure 1. Pathological and Immunohistochemical Evaluation of Each DFSP Case
eFigure 2. PDGFB Expression in DFSP.
eFigure 3. PDGFB Expression in 48 Non-DFSP Mesenchymal Tumors Analyzed in This Study
eFigure 4. PDGFRa Expression in DFSP
eFigure 5. PDGFRa Expression in Dermatofibroma, Keloid, and Hypertrophic Scar Specimens
eFigure 6. Ring Chromosome Formation in DFSP Patient 9 Tumor Cells
eFigure 7. PDGFB Expression in Vascular Endothelial Cells
Nakamura I, Kariya Y, Okada E, Yasuda M, Matori S, Ishikawa O, Uezato H, Takahashi K. A Novel Chromosomal Translocation Associated With COL1A2-PDGFB Gene Fusion in Dermatofibrosarcoma ProtuberansPDGF Expression as a New Diagnostic Tool. JAMA Dermatol. 2015;151(12):1330-1337. doi:10.1001/jamadermatol.2015.2389
Dermatofibrosarcoma protuberans (DFSP) is a rare skin cancer that develops in the deep dermis to subcutaneous adipose tissues. A COL1A1-PDGFB gene fusion, leading to the constitutive expression of PDGFB, is the tumorigenic mechanism in most DFSP cases.
To evaluate the specificity of PDGFB expression as a diagnostic marker of DFSP and to determine whether other pathomechanisms (ie, gene fusions) exist in patients with DFSP without the COL1A1-PDGFB fusion gene.
Design, Setting, and Participants
All patients with DFSP registered in the pathologic database of the University of the Ryukyus from January 1, 1997, through December 31, 2013, and Gunma University from January 1, 1996, through December 31, 2011, were included in this analysis. Samples were obtained from 30 patients presenting with DFSP tumors. We examined the clinicopathologic characteristics and the expression of PDGFB, PDGFRβ, PDGFRα, CD34, nestin, factor XIIIa, fibronectin, α–smooth muscle actin, S-100 protein, and Ki-67 in 30 DFSP cases and 48 non-DFSP mesenchymal tumor cases by immunohistochemical analysis. We then analyzed tumor tissues for the presence of the COL1A1-PDGFB fusion gene. We also tested whether other genes enriched in fibroblasts formed fusion products with PDGFB by reverse transcription–polymerase chain reaction analysis, using gene-specific primers.
Main Outcomes and Measures
We aimed to analyze tumor tissues for the presence of the COL1A1-PDGFB fusion gene to investigate expression of PDGFB in DFSP tumors.
PDGFB expression was detected in 28 (93%) of 30 patients with DFSP. PDGFB was not homogenously expressed in DFSP tumor cells, whereas CD34 and nestin were often expressed throughout the tumor mass. In 1 DFSP tumor, the COL1A1-PDGFB fusion gene was not detected even though PDGFB was expressed. We identified a novel COL1A2-PDGFB fusion gene in this tumor.
Conclusions and Relevance
Our findings indicate that PDGFB protein is expressed in most DFSP tumors and may be a useful diagnostic tool when used in conjunction with CD34 and nestin expression analysis. These PDGFB expression data, in addition to our discovery of a novel PDGF fusion gene, strongly support the concept that DFSP is a PDGFB-dependent tumor type.
Dermatofibrosarcoma protuberans (DFSP) is a slow-growing, moderately malignant mesenchymal tumor. Although distant metastases are rare, DFSP tumors may repeatedly regrow locally after removal.1Quiz Ref ID The typical storiform-type DFSP accounts for approximately half of all cases, although 8 other histologic subtypes exist.2,3 CD34 is a useful DFSP tumor marker for immunohistologic analysis4,5; however, CD34 expression alone is not sufficient to distinguish DFSP from other types of tumors. In fact, some DFSP tumors lack CD34 expression,6 whereas other tumor types, such as dermatofibroma, express CD34.7- 9 Nestin was also proposed as a marker for DFSP,10,11 along with the absence of factor XIIIa, α–smooth muscle actin (α-SMA), and S-100 protein. However, none of these markers is sufficiently specific to distinguish DFSP from other mesenchymal tumors.12
Quiz Ref IDThe tumorigenic mechanism of DFSP arises from a translocation of chromosomes 17 and 22, inducing the formation of a fusion gene between the COL1A1 (OMIM 120150) and PDGFB (OMIM 190040) genes.13 Constitutive expression of PDGFB drives the self-propagation of DFSP tumor cells. Although this COL1A1-PDGFB fusion gene is found in most patients with DFSP, it is absent in approximately 8% of patients,14 implying the existence of other DFSP pathomechanisms.
Human platelet-derived growth factor (PDGF) signaling involves 4 ligand genes: PDGFA-D and the 2 receptor genes PDGFRα and PDGFRβ. PDGF receptor (PDGFR) dimerization results in the expression of 3 active receptors: PDGFRαα, PDGFRαβ, and PDGFRββ. The PDGF ligands also dimerize; PDGF-AA and PDGF-CC are known to bind to PDGFRαα, whereas PDGF-BB binds to PDGFRββ.15
In this study, we performed an immunohistologic evaluation of PDGFB expression and a genetic analysis for fusion genes in 30 patients with DFSP. We confirmed that the PDGFB protein was expressed even in DSFP tumor cells lacking the COL1A1-PDGFB fusion gene, and we discovered a novel fusion gene in this tumor.
This study was approved by the ethics committees of the University of the Ryukyus and Gunma University. Written informed consent was obtained from patients whose tissues were analyzed for the presence of gene rearrangements in accordance with the ethics committee procedures of the respective institutions.
All patients with DFSP registered in the pathologic database of the University of the Ryukyus from January 1, 1997, through December 31, 2013, and Gunma University from January 1, 1996, through December 31, 2011, were included in this analysis. Samples were obtained from 30 patients presenting with DFSP tumors. Patient selection and patient or tumor clinicopathologic data are described in the eMethods and eTable 1 in the Supplement. A histologic analysis of DFSP tissues revealed that the storiform pattern accounted for most cases in this study (23 of 30 patients [77%]), followed by 2 sclerotic type, 2 Bednar tumors, 2 myoid type, and 1 mixed fibrosarcomatous and myoid type case(s). The control group included 48 patients with mesenchymal tumors and inflammatory diseases, consisting of 20 dermatofibroma, 6 keloid, 11 hypertrophic scar, 2 nodular fasciitis, 4 malignant fibrous histiocytoma (MFH), 2 fibrosarcoma, 2 leiomyosarcoma, and 1 low-grade fibromyxoid sarcoma case(s). All pathologic sections were reviewed by a dermatopathologist (Y.K.) at the University of the Ryukyus.
We used rabbit anti-PDGFB polyclonal antiserum (Abcam), which reacts with the wild-type PDGFB protein and the PDGFB protein originating from the fusion gene. Additional details regarding antibodies and immunohistologic findings are given in the eMethods and eTable 2 in the Supplement. Each patient sample was scored according to the percentage of stained tumor cells. Tumors were also scored based on the distribution of tumor cells within patient samples: diffuse, focal, central dominance, or peripheral dominance. Because only a limited number of slides were available for some patient samples, cells expressing PDGFRα, nestin, and fibronectin were not quantified in all cases (Table and eTable 3 in the Supplement).
RNA was extracted and reverse transcribed as described in the eMethods in the Supplement. To detect the presence of COL1A1-PDGFB fusion transcripts, reverse transcription–polymerase chain reaction (RT-PCR) was performed using COL1A1 or COL1A2 (OMIN 120160)-specific forward primers (eMethods and eTable 4 in the Supplement) and a PDGFB reverse primer or using COL3A1 (OMIM 120180), COL5A1 (OMIM 120215), COL5A2 (OMIM 120190), COL5A3 (OMIM 120216), elastin (OMIM 130160), lamininβ1 (OMIM 150240), fibronectin1 (OMIM 135600), fibrillin1 (OMIM 134797), fibrillin2 (OMIM 612570), fibulin 5 (OMIM 604580), and tenascin XB (OMIM 600985)–specific forward primers and a PDGFB reverse primer. Complementary DNAs were amplified and analyzed as described in the eMethods in the Supplement.
We assessed and scored the expression levels and localization of PDGFB, PDGFRβ, PDGFRα, CD34, factor XIIIa, nestin, fibronectin, α-SMA, S-100 protein, and Ki-67 in tumor tissue from 30 patients with DFSP and 48 patients with other mesenchymal tumor types or inflammatory diseases (Table and eTable 3 and eFigure 1 in the Supplement).
PDGFB expression was evaluated in the cell membrane and cytoplasm of tumor cells (Figure 1A and B). Quiz Ref IDImmunohistologic analysis revealed that PDGFB was expressed in tumor tissue from 28 (93%) of the 30 patients with DFSP (Table and eTable 3 in the Supplement); however, PDGFB expression was patchy and not evenly distributed within a single tumor mass (Figure 1C and D). In most patients (26 of 28 [93%]), PDGFB was only expressed in a subset of cells in the DFSP tumor mass, whereas in the remaining 2 patients, PDGFB was expressed diffusely throughout the tumor (Table and eFigure 2A in the Supplement). Of the 2 patients in which PDGFB was not expressed (patients 5 and 18), patient 5 presented with a CD34-positive, S-100 protein–negative Bednar tumor (eFigure 2B in the Supplement), and although patient 18 presented as a storiform-pattern DFSP, it lacked CD34 expression (eFigure 2C in the Supplement). RNA samples could not be obtained from these 2 patients. Therefore, it was not possible to search for fusion genes or other genetic evidence. In contrast, PDGFB was expressed in tumor cells from only 1 of the 48 control cases—a single dermatofibroma (eFigure 3A-J in the Supplement).
All DFSP tumors expressed PDGFRβ. Similar to PDGFB, PDGFRβ expression varied within tumors from just a few cells to all tumor cells and was not homogenous within the tumor mass. A comparison of PDGFB and PDGFRβ expression in tumor tissue did not reveal a consistent trend in most patients. In some DFSP samples, PDGFRβ was coexpressed with PDGFB in the same tumor cells (Figure 2A and B), whereas in others, PDGFRβ expression was low in tumor cells with high PDGFB expression (Figure 2C and D). All dermatofibromas also expressed PDGFRβ. In contrast, PDGFRα was expressed in only 8 of the 30 patients with DFSP (eFigure 4 in the Supplement). A higher proportion of dermatofibromas (14 of 19 [74%]) exhibited focal expression of PDGFRα (eFigure 5 in the Supplement).
In this study, CD34 was expressed in the tumors of 29 (97%) of the 30 patients with DFSP. Unlike PDGFB and PDGFRβ, CD34 expression was diffuse throughout the tumor tissue. In the tumors of 25 patients (83%), CD34 was expressed in more than 51% of tumor cells (Table and eTable 3 in the Supplement). Patient 18, the only patient in whom CD34 expression was completely absent, presented with a storiform-pattern DFSP that was negative for PDGFB expression. In 48 control samples, CD34 expression was detected in 2 patients with dermatofibroma. However, in those 2 samples, CD34-positive cells were not distributed throughout the tumor mass.
Nestin was originally discovered as a type VI intermediate filament, and in terms of pathologic diagnosis, its expression has been reported in several brain tumors, such as glioblastoma and more recently in DFSP.10,11,16 Nestin was detected in the tumors of 29 (97%) of the 30 patients with DFSP. It was often diffusely distributed in tumor tissue, being present in 51% (or greater) of tumor cells in 19 (63%) of 30 patients. Its expression was absent only in patient 5, who had a Bednar tumor. Although this specimen was positive for CD34, it did not express PDGFB. Among patients with dermatofibroma, nestin expression was observed in 3 (16%) of 19 (eTable 3 in the Supplement).
Factor XIIIa was expressed in the tumors of 20 (67%) of 30 patients with DFSP. There was no correlation between factor XIIIa expression and either the histologic type of DFSP or the clinical data. Expression of α-SMA was found in the tumors of 6 (20%) of the 30 patients with DFSP. Three cases coincided with myoid areas, and the other 3 cases were storiform DFSP. Expression of S-100 protein was not observed in any of the 30 patients with DFSP, whereas the S-100 protein was expressed in 13 (65%) of 20 dermatofibromas. The presence of fibronectin was recently reported to prolong the proliferative effects of PDGF.17 Fibronectin was expressed in the tumors of 16 (57%) of 28 patients with DFSP. A comparison of areas of fibronectin expression with areas of high expression of PDGFB, PDGFRβ, and Ki-67 within DFSP tissues did not reveal a consistent trend. Fibronectin was also highly expressed in fibroblastic tumors, such as dermatofibromas, true keloids, and hypertrophic scars.
Ki-67–positive tumor cells accounted for at least 5% of total tumor cells in 20 (67%) of 30 patients with DFSP. In most patients, the proportion of Ki-67–positive cells was 5% to 25%, or less, and there was no correlation between the histologic localization of tumor cells (ie, dermis or adipose tissue localization) and regions of high Ki-67 expression. No correlation was found between the rate of Ki-67 expression and the various histologic types.
We next investigated the correlation between regions of PDGFB, PDGFRβ, PDGFRα, nestin, and fibronectin expression and the Ki-67 expression index in DFSP tumor tissues. No consistent correlation was found between PDGFB localization and the frequency of Ki-67–positive cells. Similar to PDGFB, there was no association between the localization of PDGFRβ, PDGFRα, nestin, or fibronectin and the frequency of Ki-67–positive cells. Seventeen of 20 patients with dermatofibroma had a Ki-67–positive index of 1% to 5%, whereas almost all patients with MFH and fibrosarcoma had a Ki-67–positive index of 26% or higher. The frequency of Ki-67–positive cells in DFSP was 5% to 25% (1+) in most patients, which was lower than that of patients with MFH and fibrosarcoma (eTable 3 in the Supplement).
RNA was extracted from tumor cells in 13 patients with DFSP, and RT-PCR was used to test for the presence of the COL1A1-PDGFB fusion gene. Quiz Ref IDThe fusion gene was present in 11 (85%) of the 13 patients analyzed. In these 11 patients, we determined that exon 5, 9 (2 cases), 11, 25 (3 cases), 32, 33, 41, or 45, located at the 5′ of the COL1A1 gene, had fused, in frame, with the PDGFB gene. Immunohistologic staining revealed that all patients with DFSP who had these COL1A1-PDGFB fusion genes expressed PDGFB. However, the 2 patients (patients 9 and 13) who did not possess the COL1A1-PDGFB fusion gene also expressed PDGFB protein.
We hypothesized that a gene other than COL1A1 may have fused with PDGFB to induce PDGFB expression. To explore this possibility, we performed RT-PCR for selected genes that were highly expressed in fibroblasts by creating gene-specific forward primers and a PDGFB-specific reverse primer to amplify potential fusion genes. Of interest, we amplified a novel fusion gene between COL1A2 and PDGFB in one patient (Table, patient 9). This gene resulted from an in-frame fusion between COL1A2 exon 26 and PDGFB exon 2 (Figure 3A and B). A chromosomal karyotype analysis revealed the presence of a supernumerary ring gene that is also present in DFSP with COL1A1-PDGFB translocations (eFigure 6 in the Supplement).
On histopathologic analysis, the DFSP in patient 9 exhibited the classic storiform pattern, and CD34 expression was evenly distributed throughout the tumor mass (Figure 4A). Tumor cells at the margins expressed increased levels of PDGFB, whereas nontumorous fibroblasts in the tumor vicinity did not express PDGFB (Figure 4B-D). Nestin was expressed in a subset of the tumor cells. PDGFRβ was diffusely expressed in the tumor. In patient 13, however, the amount of RNA extracted from the tumor tissue was insufficient to search for a fusion gene other than COL1A1.
Our results indicate that the analysis of PDGFB, nestin, and CD34 expression is effective for pathologic diagnosis of DFSP. Our discovery of a novel COL1A2-PDGFB fusion gene also verified that constitutive expression of PDGFB is the fundamental pathomechanism for DFSP tumorigenesis.
In our analysis, PDGFB was expressed in 28 (93%) of 30 patients with DFSP and was absent in 47 (98%) of 48 patients with non-DFSP mesenchymal tumors. Thus, PDGFB is a highly specific marker for DFSP, with a limited range of expression of all mesenchymal tumors studied. We found PDGFB expression in only one patient with dermatofibroma. In support of these results, previous studies18,19 also determined that PDGFB is expressed in a small number of patients with dermatofibroma and MFH.
CD34 was expressed in 97% of patients with DFSP and was absent in 96% of patients with non-DFSP tumors. In the 2 patients with DFSP (patients 5 and 18) in whom PDGFB was not expressed, CD34 expression was also absent in one of them (patient 18). In patient 18, DFSP had been diagnosed based on tumor morphologic findings; however, its mechanism of tumorigenesis and tissue origin may have differed from other DFSPs. In the 30 patients with DFSP analyzed, the expressions of PDGFB, CD34, and nestin were of almost equivalent sensitivity. However, whereas CD34 and nestin expressions tended to be diffuse, PDGFB was unevenly expressed within the same tumor tissues (Figure 1C and D). The CD34 and nestin double-positive cells are likely to be remnant DFSP cells at the surgical margin.
In a previous study18 that included a small number of patients with DFSP, only 1 in 6 patients with DFSP is positive for PDGFB, unlike the high rate observed in our study in which a distinct PDGFB antibody was used from our current analysis. In fact, to our knowledge,ours is the only study to report PDGFB expression in a large number of DFSP cases. Because PDGFB expression is not uniform within most DSFP tumors, it is possible that determining PDGFB expression from analysis of a small biopsy sample, representing only part of the tumor structure, will reduce the positive rate. From this perspective, CD34 expression analysis may be more useful for pathologic diagnosis. In tumor regions with low PDGFRβ and PDGFB expression, autocrine or paracrine PDGF signaling may have caused these proteins to become internalized.15 Because PDGFB would be internalized after binding PDGFRβ, only a small fraction of DFSP tumor cells expressed PDGFB, even though all tumor cells contained the fusion gene. Quiz Ref IDNestin was reported to be diffusely expressed in tumor cells in 94% to 100% of patients with DFSP, whereas in dermatofibroma, nestin was only expressed in specific regions of the tumor mass in 0% to 13% of patients.10,11 In our analysis, nestin was expressed in 29 (97%) of 30 patients with DFSP, and in most cases, it was diffusely and homogenously expressed, similar to CD34. Taken together, the combinatorial expression of CD34, nestin, and PDGFB may enable more precise diagnosis of DFSP tumors. We found that there were 7 cases of local recurrence among 30 patients with DFSP, including 4 patients who were 1% to 25% (1+) positive for PDGFB, 3 patients who were 26% to 50% (2+) positive for PDGFB, and no patients with tumor-related death. There was no correlation between PDGF expression and recurrence.
In this study, we discovered a novel fusion gene associated with PDGFB. Because a detailed analysis revealed that the COL1A1-PDGFB fusion gene was not detectable in approximately 8% of patients with DFSP,14 we hypothesized that these patients may have a fusion gene other than COL1A1-PDGFB. In patient 9, a karyotypic analysis revealed the presence of chromosomal rings, and an immunohistologic analysis revealed that the PDGFB protein was expressed in the tumor mass; however, the conventional COL1A1-PDGFB fusion gene was not detected by RT-PCR. We then selected genes that exhibited high expression in fibroblasts for RT-PCR amplification of putative PDGFB fusion genes. Indeed, we were able to detect a novel fusion gene between the COL1A2 and PDGFB genes in tumor tissue from 1 patient with DFSP (patient 9).
The human PDGFB gene generates a 241 amino acid protein encoded by 7 exons. Only exons 4 and 5 of the PDGFB gene encode the mature protein, whereas exon 1 encodes the signal sequence and inhibitory control region; exons 2 and 3, the N-terminal sites that are cleaved in the Golgi apparatus; exon 6, a C-terminal domain that is cleaved during protein maturation to its secreted form; and exon 7, encoding a nontranslated region. On the basis of the PDGFB gene structure, even if a portion of the N-terminal fusion gene, such as COL1A1 or COL1A2, is transcribed, the fusion protein on the N-terminal side will be cleaved intracellularly during the PDGFB maturation process.20- 22 PDGFB is mostly expressed in the membrane-bound form; however, a small fraction of PDGFB is secreted extracellularly. PDGFB in the stroma seldom binds to the PDGFR and is retained in the stroma without clearance through internalization. All DFSP fusion genes reported to date reveal that the COL1A1 gene is fused with intron 1 of the PDGFB gene and the fusion protein with COL1A1 starts from the exon 2 region of PDGFB. The novel COL1A2-PDGFB fusion gene discovered in patient 9 similarly bound with intron 1 of PDGFB. Therefore, it is likely that the DFSP tumorigenesis mechanism involves the powerful promoter activity of the COL1A1 or COL1A2 genes, contributing to constitutive expression of PDGFB.
In our analysis, we found that patients with fusion genes between PDGFB and either COL1A1 or COL1A2 did not possess DFSP tumors with specific clinical or histopathologic characteristics, but instead the tumors could only be categorized as “typical” DFSP. In fact, in the 3 patients with DFSP who possessed the same COL1A1 exon 25 fusion gene (patients 2, 11, and 14), the tumors were all of different histologic types (a myoid, storiform, and Bednar tumor, respectively). In patient 13, no COL1A1-PDGFB fusion gene was detected. It is also possible that the expression of PDGFB is caused by an unidentified gene other than COL1A. Our immunohistologic analysis suggested that DFSP tumors may arise from a mesenchymal cell type expressing CD34, nestin, and PDGFRβ rather than from the malignant transformation of dermal fibroblasts.
Accepted for Publication: June 15, 2015.
Corresponding Author: Kenzo Takahashi, MD, PhD, Department of Dermatology, Graduate School of Medicine, University of the Ryukyus, 207 Uehara, Nishihara-cho, Okinawa 903-0215, Japan (email@example.com).
Published Online: September 2, 2015. doi:10.1001/jamadermatol.2015.2389.
Author Contributions: Drs Nakamura and Takahashi had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Nakamura, Kariya, Yasuda, Matori, Ishikawa, Uezato, Takahashi.
Acquisition, analysis, or interpretation of data: Nakamura, Kariya, Okada, Matori, Takahashi.
Drafting of the manuscript: Nakamura, Kariya, Yasuda, Matori, Takahashi.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Nakamura, Kariya, Matori, Takahashi.
Obtained funding: Matori, Takahashi.
Administrative, technical, or material support: Nakamura, Kariya, Okada, Yasuda, Matori, Takahashi.
Study supervision: Ishikawa, Uezato, Takahashi.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported in part by a Grant-in-Aid for Young Scientists (B) 25860960 from the Ministry of Education, Culture, Sports, Science, and Technology of Japan (Drs Matori and Takahashi).
Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and the decision to submit the manuscript for publication.
Additional Contributions: Ayako Nakamura, BS, and Ritsuko Tokumon provided technical support for this study. No compensation was provided.