Pancreatic ductal adenocarcinoma (PDAC) has a poor prognosis (5-year overall survival rate, <10%).¹ However, PDAC ≤10 mm has a good prognosis (5-year survival rate, >80%).² High-grade pancreatic intraepithelial neoplasia (HG-PanIN) represents early-stage precancerous lesions,^3,4 with a high risk of progression to invasive carcinoma.^4-6 PDAC ≤10 mm and HG-PanIN are attractive targets because of their good prognosis.^5-7 However, the detection rate of these lesions, particularly HG-PanIN, is extremely low,^2,6,8 owing to their microscopic size and absence of mass-forming features.⁴ Although indirect morphological changes, such as stenosis and main pancreatic duct (MPD) dilation, have been used to detect PDAC ≤10 mm and HG-PanIN,^7,9,10 they have also been observed in benign pancreatic diseases.¹¹ A large-scale screening study reported that endoscopic ultrasound (EUS) could detect more early pancreatic lesions, such as small PDAC and high-grade dysplasia, than magnetic resonance imaging (MRI).¹²

However, validation of specific EUS findings suggestive of PDAC ≤10 mm or HG-PanIN is lacking.⁹ Herein, we analyzed the EUS findings for the differential diagnosis of HG-PanIN, PDAC ≤10 mm, and benign stenosis, and examined the interobserver reliability of EUS.

1. Ethics

The study protocol was approved by the Institutional Review Board of Oita San-ai Medical Center (IRB number: 020007K) and was registered in the UMIN Clinical Trials Registry (UMIN000044789). The requirement for written informed consent was waived because of the retrospective nature of the study. The work was carried out in compliance with the Ethical Principles for Medical Research Involving Human Subjects outlined in the Helsinki Declaration in 1975 (revised in 2000). All authors had access to the study data and reviewed and approved the final manuscript.

2. Eligibility

We obtained data of patients from January 2014 to January 2021 from four participating general hospitals. Among patients undergoing EUS, those with pancreatic duct irregularity without a clear mass on EUS and those with a pancreatic tumor ≤10 mm measured on EUS with a distinct mass were included. Patients with pancreatic tumors other than PDAC were diagnosed by EUS-guided fine needle aspiration/histological diagnosis from resected specimens/clinical follow-up examination. Patients diagnosed with high-grade dysplasia and carcinoma derived from intraductal papillary mucinous neoplasms from resected specimens were excluded, because of their different nature.

3. HG-PanIN

All included HG-PanIN cases without invasive carcinoma were diagnosed using resected specimens by experienced pathologists specializing in pancreatic pathology, in addition to pathologists at each hospital. HG-PanIN was defined, according to the World Health Organization,^3,4 as flat or papillary/micropapillary/cribriform with severe nuclear atypia, loss of polarity, macronucleoli, or abnormal mitotic figures.

4. PDAC ≤10 mm

As diameter measurement by EUS might be affected by each endosonographer, the maximum diameter of invasive PDAC ≤10 mm was confirmed using histopathological specimens for accurate and observational evaluation. Similarly, patients with PDAC diagnosed by only clinical follow-up examination and those who were treated with preoperative chemotherapy were also excluded. HG-PanIN and PDAC ≤10 mm were considered lesions requiring surgical resection and were collectively defined as high-risk pancreatic lesions.

5. Benign stenosis

Focal areas of MPD stenosis suspected to be HG-PanIN or small invasive PDAC based on characteristic imaging findings (e.g., ductal irregularities, including focal stenosis/dilation) were evaluated preoperatively by EUS-guided fine needle aspiration/pancreatic juice cytology, to confirm the diagnosis histologically.^9,13 Lesions without evidence of HG-PanIN or invasive PDAC on one or more cytological examinations with stable imaging findings over a follow-up period of >36 months were defined as benign stenosis. Resected lesions that were subsequently confirmed to be benign were also defined as benign stenosis. Patients with autoimmune pancreatitis and/or MPD stenosis due to pancreatic stones were also excluded, because their accurate evaluation of EUS findings was complicated by halation.

6. Endoscopic ultrasound

GF-UCT260/GF-UE260 echoendoscopes (Olympus Medical Systems, Tokyo, Japan) paired with various ultrasound processors (ProSound SSD α-10/F75/ARIETTA [Aloka, Tokyo, Japan]; EU-ME2 [Olympus Medical Systems]) were used. Images were analyzed using EUS videos of the pancreatic lesions; therefore, patients whose EUS videos were inadequately recorded were also excluded.

Based on recent reports of imaging findings of HG-PanIN and small PDAC,^7-10,12 the following six EUS findings were evaluated for distinguishing HG-PanIN, PDAC ≤10 mm, and benign stenosis: (1) MPD stenosis (defined as a MPD diameter ≤1 mm with an upstream MPD diameter relative to the stenosis >1 mm) (Fig. 1A); (2) upstream MPD dilation (defined as an MPD diameter upstream of the pancreatic lesion ≥2 mm) (Fig. 1A); (3) localized hypoechoic areas surrounding MPD irregularities (focal stenosis/dilation), further classified as light (mottled hypoechoic areas without clear demarcation from the surrounding pancreatic parenchyma) (Fig. 1B) or hypoechoic areas (round hypoechoic areas with clear demarcation from the surrounding pancreatic parenchyma) (Fig. 1C); (4) branch pancreatic duct (BPD) dilation (adjacent to the pancreatic lesion) (Fig. 1D); (5) prominent lobular segmentation relative to the upstream or downstream pancreas; and (6) localized parenchymal atrophy compared to the upstream or downstream pancreas.

Figure 1. Endoscopic ultrasound findings for differentiating high-grade pancreatic intraepithelial neoplasia, pancreatic ductal adenocarcinoma ≤10 mm, and benign stenosis. (A) Upstream dilated main pancreatic duct (MPD) and downstream MPD (right and left arrowhead) with stenosis (between arrows). (B) Mottled hypoechoic areas (encircled by rounds) around MPD stenosis. (C) Round hypoechoic areas (surrounded by rounds). (D) Branch pancreatic duct dilation (white-round lumen) surrounding the MPD.

7. Image analysis

EUS findings for distinguishing HG-PanIN, PDAC ≤10 mm, and benign stenosis were analyzed with consensus (first reading) by two endoscopists (R.S. and K.Y. [Kentaro]) with >10 years of experience in EUS diagnosis of early PDAC. The sensitivity and specificity of each EUS finding in the differential diagnosis of high-risk pancreatic lesions and benign stenosis were also examined. Subsequently, two endoscopists (J.N. and H.A.) with >10 and 2 years of experience in EUS diagnosis of early PDAC, respectively, were recruited for the blinded study. They independently evaluated randomly sorted EUS videos without access to clinical data or imaging findings. They were only provided with information about lesion location (head/body/tail), following which the six aforementioned EUS findings were evaluated for all lesions. After the registration of blind review results, detailed EUS image interpretations of each blinded reviewer were validated by first consensus reading reviewers.

8. Study outcomes

The primary outcome was to determine significant EUS findings for distinguishing HG-PanIN, PDAC ≤10 mm, and benign stenosis. The secondary outcomes were (1) to determine the diagnostic performance of each EUS finding for distinguishing high-risk pancreatic lesions from benign stenosis and (2) to evaluate the diagnostic concordance of each EUS finding between the first reading and two blinded reviews.

9. Statistical analyses

Patient characteristics are presented as means±standard deviations, depending on the normality of the distribution. Continuous variables were expressed as means±standard deviations and compared using the analysis of variance. Categorical data were compared using the chi-square test or Fisher exact test. Multivariate logistic regression analysis was performed to determine the most significant EUS finding, using significant EUS findings from the univariate analysis. The sensitivity and specificity of each EUS finding in distinguishing HG-PanIN and PDAC ≤10 mm from benign stenosis were evaluated using two-way contingency tables and receiver operating characteristic curves. The area under the curve was calculated. Interobserver agreement for EUS findings was calculated using the Fleiss kappa for multiple readers. Kappa values were interpreted as follows: <0.00, poor; 0.00 to 0.20, slight; 0.21 to 0.40, fair; 0.41 to 0.60, moderate; 0.61 to 0.80, substantial; and 0.81 to 1.00, almost perfect agreement.¹⁴ All statistical analyses were conducted using SPSS version 28.0 (IBM Corp., Armonk, NY, USA). A p-value <0.05 was considered statistically significant.

1. Patient characteristics

The inclusion and diagnostic criteria for pancreatic lesions are presented in Fig. 2. Among 168 patients with MPD irregularities without a clear mass on EUS and 87 patients with pancreatic tumor with a clear mass, HG-PanIN, PDAC <10 mm, and benign stenosis were diagnosed in 29, 11 and 20 patients, respectively. All HG-PanIN and PDAC <10 mm cases were diagnosed using resected specimens. None of the patients who were diagnosed as having PDAC >10 mm by EUS were found to have a final tumor size of <10 mm on histological evaluation after resection.

Figure 2. Inclusion and diagnostic criteria for HG-PanIN, PDAC ≤10 mm, and benign stenosis.
EUS, endoscopic ultrasound; AIP, autoimmune pancreatitis; MPD, main pancreatic duct; PDAC, pancreatic ductal adenocarcinoma; EUS-FNAB, EUS-guided fine needle aspiration biopsy; HG-PanIN, high-grade pancreatic intraepithelial neoplasia.

Eleven patients were diagnosed with PDAC <10 mm (median tumor diameter, 8 mm; range, 2 to 10 mm). Four patients underwent surgical resection of MPD irregularities without a clear mass on EUS and other radiological examinations. Eight patients with benign stenosis were diagnosed by histological examination of the resected specimens (low-grade PanIN [n=5]; chronic infla mmatory changes [n=3]); the remaining 12 patients were diagnosed after >3 years of follow-up and negative pancreatic juice cytology (Table 1).

Table 1 . Clinical Characteristics of Patients with Each Pancreatic Lesion.

Characteristic	HG-PanIN (n=29)	PDAC ≤10 mm (n=11)	Benign stenosis (n=20)	p-value
Age, yr	70.3±7.5	67.0±12.0	68.8±9.5	0.51
Male sex	19 (65.5)	5 (45.5)	14 (70.0)	0.38
Body mass index, kg/m2	22.1±2.0	22.6±5.8	22.2±3.3	0.94
Smoking status (current or former)	17 (58.6)	4 (36.4)	9 (45.0)	0.39
Heavy alcohol consumption (>37.5 g/day)	8 (27.6)	0	4 (20.0)	0.15
Diabetes mellitus	12 (41.4)	4 (36.4)	4 (20.0)	0.29
Chronic pancreatitis	1 (3.4)	0	0	0.58
Family history of pancreatic cancer	1 (3.4)	0	1 (5.0)	0.76
Elevated tumor marker (CEA and/or CA19-9)	1 (3.4)	1 (9.1)	2 (10.0)	0.62
Elevated amylase	4 (13.8)	3 (27.3)	4 (20.0)	0.60
Location				0.57
Head	5 (17.2)	2 (18.2)	5 (25.0)
Body & tail	24 (82.8)	9 (81.8)	15 (75.0)
Method of final diagnosis				<0.001
Histological analysis of resected specimen	29 (100.0)	11 (100.0)	8 (40.0)
Pancreatic juice cytology and 3 yr of follow-up	0	0	12 (60.0)

Data are presented as mean±SD or number (%)..

HG-PanIN, high-grade pancreatic intraepithelial neoplasia; PDAC, pancreatic ductal adenocarcinoma; CEA, carcinoembryonic antigen; CA19-9, carbohydrate antigen 19-9..

There were no significant differences in clinical characteristics between patients with HG-PanIN, PDAC ≤10 mm, and benign stenosis.

2. Specific EUS findings in distinguishing the three types of lesions

Table 2 summarizes the EUS findings of HG-PanIN, PDAC ≤10 mm, and benign stenosis. In the comparison of HG-PanIN and benign stenosis, only hypoechoic areas surrounding MPD irregularities were significant for distinguishing HG-PanIN from benign stenosis (82.8% vs 15.0%, respectively: odds ratio [OR], 27.2; 95% confidence interval [CI], 5.7 to 129.5; p<0.001). In the comparison of PDAC <10 mm and benign stenosis, MPD irregularities (90.9% vs 15.0%, respectively: OR, 56.7; 95% CI, 5.2 to 621.0; p<0.001) and BPD dilation (90.9% vs 50.0%, respectively: OR, 10.0; 95% CI, 1.1 to 93.4; p=0.02) were significantly more frequent in PDAC ≤10 mm than in benign stenosis. These two EUS findings were analyzed in a multivariate analysis. The presence of hypoechoic areas surrounding MPD irregularities was most significant for distinguishing PDAC <10 mm from benign stenosis (OR, 46.7; 95% CI, 3.9 to 557.1; p=0.002).

Table 2 . Specific EUS Findings in the Differential Diagnosis of HG-PanIN, PDAC ≤10 mm, and Benign Stenosis.

EUS finding	No. (%)			p-value (HG-PanIN vs benign stenosis)	p-value (PDAC ≤10 mm vs benign stenosis)	p-value (HG-PanIN vs PDAC ≤10 mm)
	HG-PanIN (n=29)	PDAC ≤10 mm (n=11)	Benign stenosis (n=20)
MPD stenosis	27 (93.1)	11 (100.0)	20 (100.0)	0.23	NA	0.37
Upstream MPD dilation	23 (79.3)	11 (100.0)	15 (75.0)	0.72	0.07	0.10
Hypoechoic areas surrounding MPD irregularities	24 (82.8)	10 (90.9)	3 (15.0)	<0.001	<0.001	0.52
Mottled hypoechoic areas without demarcation	23 (79.3)	2 (18.2)	3 (15.0)	<0.001	0.82	<0.001
Round hypoechoic areas with demarcation	1 (3.4)	8 (72.7)	0	0.40	<0.001	<0.001
BPD dilation	22 (75.9)	10 (90.9)	10 (50.0)	0.06	0.02	0.29
Prominent lobular segmentation	7 (24.1)	3 (27.3)	3 (15.0)	0.44	0.40	0.84
Localized parenchymal atrophy	6 (20.7)	1 (9.1)	2 (10.0)	0.32	0.94	0.39

EUS, endoscopic ultrasound; HG-PanIN, high-grade pancreatic intraepithelial neoplasia; PDAC, pancreatic ductal adenocarcinoma; MPD, main pancreatic duct; BPD, branch pancreatic duct; NA, not assessed..

In the subgroup analysis, mottled hypoechoic areas were more frequently observed in HG-PanIN than in PDAC <10 mm (79.3% vs 18.9%, respectively: OR, 17.3; 95% CI, 2.9 to 101.9; p<0.001). Conversely, round hypoechoic areas were more frequently observed in PDAC <10 mm than in HG-PanIN (72.7% vs 3.4%, respectively: OR, 74.7; 95% CI, 6.8 to 819.5; p<0.001).

The sensitivity of hypoechoic areas surrounding MPD irregularities and BPD dilation for distinguishing high-risk pancreatic lesions from benign stenosis was 85.0% and 80.0%, respectively; the specificity was 85.0% and 50.0%, respectively (Table 3). The interobserver agreement was moderate (κ=0.51 and κ=0.42, respectively).

Table 3 . Sensitivity, Specificity, and Reliability of EUS Findings in Differentiating High-Risk Pancreatic Lesions from Benign Stenosis.

EUS finding	Sensitivity, %	Specificity, %	AUC (95% CI)	Fleiss kappa (95% CI)	Interobserver agreement of 3 reading results*
MPD stenosis	97.5	-	0.49 (0.33 to 0.64)	0.61 (0.47 to 0.76)	Substantial
MPD dilation	82.5	25.0	0.54 (0.38 to 0.70)	0.22 (0.07 to 0.36)	Fair
Hypoechoic areas surrounding MPD irregularities	85.0	85.0	0.85 (0.74 to 0.96)	0.51 (0.36 to 0.66)	Moderate
BPD dilation	80.0	50.0	0.65 (0.50 to 0.80)	0.42 (0.28 to 0.57)	Moderate
Prominent lobular segmentation	25.0	85.0	0.55 (0.40 to 0.70)	0.04 (–0.10 to 0.19)	Slight
Localized parenchymal atrophy	50.0	82.5	0.54 (0.39 to 0.69)	0.15 (0.06 to 0.30)	Slight

EUS, endoscopic ultrasound; AUC, area under the curve; CI, confidence interval; MPD, main pancreatic duct; BPD, branch pancreatic duct..

*First reading and two blinded reviews..

Among the six EUS findings evaluated in this study, the presence of hypoechoic areas surrounding MPD irregularities was the most significant predictor of HG-PanIN or PDAC ≤10 mm. This finding showed high sensitivity and specificity for the diagnosis of high-risk pancreatic lesions. Further stratification of the hypoechoic areas into mottled areas without demarcation or round areas with demarcation was useful in distinguishing HG-PanIN from small PDAC. The presence of BPD dilation adjacent to the pancreatic lesions was also useful for their differential diagnosis and should be noted by the endosonographer. To our knowledge, this is the first comparative study to identify specific EUS findings associated with HG-PanIN, PDAC ≤10 mm, and benign stenosis.

MRI, EUS, and computed tomography (CT) have been used in surveillance studies of PDAC;^8,12,15-17 however, EUS is likely to have the highest diagnostic ability for detecting PDAC,¹³ and is endorsed by the consensus guideline.¹⁸ However, the detection rates of PDAC ≤10 mm and HG-PanIN are suboptimal.^2,4,9,12

Chronic focal pancreatitis-like changes surrounding low- and HG-PanIN and small PDAC can occur due to micro-obstructive pancreatitis in the BPD, leading to acinar cell loss, lobulocentric parenchymal atrophy, fibrosis, and fatty replacement.^7,10,19-22 These changes can be detected as indirect imaging findings, such as MPD stenosis/dilation,^7,9,10,23 BPD dilation,^9,10,23-26 localized pancreatic parenchymal atrophy,^9,10,27 and hypoechoic areas surrounding MPD irregularities.^9,10,23 Morphological changes in the MPD can be detected by MRI and CT, and are considered triggers for suspecting HG-PanIN or small PDAC during screening.^8,9,12 However, these changes are unspecific and can also be detected in patients with benign pancreatic diseases.¹¹ EUS can detect focal pancreatic abnormalities in more detail than MRI and CT.^12,24 EUS can also detect hypoechoic areas surrounding MPD irregularities, reflecting histopathological changes associated with high-risk pancreatic lesions. Hypoechoic areas surrounding MPD irregularities may reflect the different echogenicity of the pancreatic parenchyma, which is affected by the refraction of ultrasound waves by pancreatic acini.¹⁰ Several studies have reported the importance of hypoechoic areas surrounding MPD irregularities as indirect evidence of HG-PanIN.^10,23 These hypoechoic areas were detected in the majority of cases by EUS before surgical resection.

In this study, most patients with HG-PanIN and PDAC ≤10 mm had discrete focal hypoechoic areas on EUS, which could be distinguished from benign stenosis. In addition, most HG-PanIN cases could be distinguished from PDACs ≤10 mm by detailed EUS examination. Mottled hypoechoic areas surrounding the MPD may reflect a local decrease in echogenicity due to the replacement of acinar cells by fibrosis around multifocal PanIN lesions.^9,10 In contrast, round hypoechoic areas may reflect the tumor itself, as well as the surrounding fibrosis.^9,10 Representative EUS images and pathological findings are presented in Fig. 3. In addition, microdilations of the BPD, identified on EUS as tiny hypoechoic or anechoic nodules, may become more prominent as hypoechoic areas surrounding MPD irregularities.²⁸ Moreover, mottled hypoechoic areas may become more conspicuous as PanIN progresses.^9,26

Figure 3. Representative endoscopic ultrasound imaging and pathological findings of high-grade pancreatic intraepithelial neoplasia (HG-PanIN) and pancreatic ductal adenocarcinoma ≤10 mm cases. (A-D) A high-grade PanIN case. (A) Mottled hypoechoic areas (surrounded by rounds) were detected between upstream and downstream main pancreatic duct (between the white arrows). (B) Hematoxylin-and-eosin-stained specimens showed decreased acinar cells and the appearance of remarkable fibrosis around HG-PanIN lesions. (C, D) Moderate and strong magnification revealed pancreatic epithelial neoplasia of cytoarchitectural atypia ranging from low grade to high grade. The micropapillary architecture of HG-PanIN, loss of polarity, enlargement and irregular nuclei was shown (C: ×4, D: ×20). (E-G) A 10 mm-invasive ductal adenocarcinoma case. (E) Endoscopic ultrasound detected round hypoechoic areas in the pancreatic parenchyma. (F) Hematoxylin-and-eosin-stained specimens showed reduced acinar cells and desmoplastic stroma around invasive adenocarcinoma. (G) Moderate magnification showed moderately differentiated invasive ductal adenocarcinoma with cribriform structures and irregular glandular ducts (×4).

In this study, HG-PanINs were considered high-risk pancreatic lesions requiring surgical intervention, similar to invasive PDAC. Therefore, the presence of hypoechoic areas surrounding MPD irregularities was useful in distinguishing these lesions from benign stenosis. In addition, most cases of PDAC ≤10 mm could be distinguished from HG-PanIN based on the finding of round hypoechoic areas surrounding MPD irregularities. However, a few cases of PDAC ≤10 mm had mottled hypoechoic areas or no hypoechoic areas and were difficult to distinguish from HG-PanIN. In addition, especially mottled hypoechoic areas did not have a clear demarcation from the surrounding pancreatic parenchyma and the hypoechoic area widespread around the pancreatic duct irregularity. Therefore, it was difficult to determine the cutoff size of the hypoechoic area based on our data.

A previous study reported a heterogeneous pattern of multifocal atrophy accompanying multifocal PanIN on EUS.¹⁹ However, in the present study, the extent of prominent lobular segmentation of the parenchyma and localized parenchymal atrophy surrounding HG-PanIN and PDAC ≤10 mm, compared to those of the upstream and downstream pancreatic parenchyma, were not particularly useful in distinguishing these lesions from benign stenosis. We found MPD stenosis with upstream MPD dilation, surrounding BPD dilation, and mottled hypoechoic areas without demarcation of the stenosis to be typical EUS findings of HG-PanIN (Fig. 4A). Conversely, MPD stenosis with upstream MPD dilation, surrounding BPD dilation, and round hypoechoic areas with demarcation of the stenosis may be typical EUS findings of PDAC ≤10 mm (Fig. 4B).

Figure 4. Schema of the specific endoscopic ultrasound findings of typical HG-PanIN and PDAC ≤10 mm cases. (A) HG-PanIN showed MPD stenosis (between the white arrows) with downstream MPD (white circle), upstream dilated MPD (white square), BPD dilation (white square), and mottled hypoechoic areas (mottled shading). (B) PDAC ≤10 mm showed MPD and BPD changes and round hypoechoic areas.
HG-PanIN, high-grade pancreatic intraepithelial neoplasia; PDAC, pancreatic ductal adenocarcinoma; MPD, main pancreatic duct; BPD, branch pancreatic duct.

Reliability is an important factor for EUS-diagnosed lesions. Furthermore, there may be high interobserver variability in chronic pancreatitis features.^9,29 In the present study, the reliability of EUS findings was evaluated with fair to substantial interobserver agreement for MPD stenosis, MPD dilation, hypoechoic areas surrounding MPD irregularities, and BPD dilation. However, the findings of prominent lobular segmentation and localized atrophy around lesions were not reliable. Focal parenchymal atrophy may be evaluated more accurately in combination with other imaging modalities, such as contrast-enhanced CT.^9,27

The diagnostic rate of HG-PanIN and PDAC ≤10 mm may be further improved by evaluating other imaging findings, especially specific EUS findings. A comprehensive clinical decision support system based on histopathological or cytological examination and specific EUS findings is required.

This study had some limitations. First, the relatively small sample size and use of different EUS equipment may cause some bias. Second, the reliability of important EUS findings was moderate, not high. The endosonographer learning curve may improve reliability. Third, it is also possible that some cases of benign stenosis (as defined in this study) may progress to HG-PanIN or invasive PDAC during a longer follow-up period.³⁰ In addition, considering that the tumor size might increase between EUS examination and surgery, and that some patients might not undergo surgery, more patients might be included. Prospective studies with more cases under uniform conditions are needed. Nevertheless, this is the first study to investigate EUS findings distinguishing between HG-PanIN, PDAC ≤10 mm, and benign stenosis. The reliability of the EUS findings was assessed by analyzing EUS videos instead of EUS images.

In conclusion, the presence of hypoechoic areas surrounding MPD irregularities on EUS may help to distinguish HG-PanIN, PDAC ≤10 mm, and benign stenosis with relatively high sensitivity and specificity and moderate reliability.

No potential conflict of interest relevant to this article was reported.

Study concept and design: R.S., Kentaro Yamao, R.M., J.N. Data acquisition: R.S., Kentaro Yamao, R.M., J.N., Kenji Yamao. Data analysis and interpretation: R.S., Kentaro Yamao, R.M., J.N. Statistical analysis: R.S., H.N. Administrative, technical, or material support: H.A., H.N., Kazuhiro Mizukami, Kenji Yamao. Drafting of the manuscript: R.S., Kentaro Yamao, R.M., J.N., H.A., H.N., Kazuhiro Mizukami, Kenji Yamao, V.B. Critical revision of the manuscript for important intellectual content: R.S., Kentaro Yamao, R.M., J.N., V.B., Y.A., Kazunari Murakami. Study supervision: R.S., Kentaro Yamao, Kazuhiro Mizukami, Kazunari Murakami. All authors read and approved the final manuscript.