Cytology Reporting System for Lung Cancer from the Japan Lung Cancer Society and Japanese Society of Clinical Cytology: An Interobserver Reproducibility Study and Risk of Malignancy Evaluation on Cytology Specimens

Pathological classification of lung carcinoma is important for effective patient management. However, 70–80% of patients with non-small cell lung cancer are at unresectable advanced stages of the disease at the time of diagnosis [1], and classification is based on small biopsies and/or cytology in these cases. Currently, unified international nomenclatures for reporting gynecological [2], thyroid [3], urine [4, 5], salivary gland [6], effusion [7], and breast [8] cytology have been proposed. However, to the best of our knowledge, no widely accepted categorization system for respiratory cytology has been established to date. In most countries, there is no standardized reporting system, and each hospital uses its own reporting system for lung carcinoma. Most reporting systems used worldwide comprise three or four categories. The Papanicolaou Society of Cytopathology (PSC) has proposed and revised a pulmonary cytology specimen terminology and classification (Table 1) [9‒11]; it comprises the following six categories: (1) “non-diagnostic,” (2) “negative for malignancy,” (3) “atypical,” (4) “neoplastic, benign neoplasm, low-grade carcinoma,” (5) “suspicious for malignancy,” and (6) “positive for malignancy.”

The reporting system for lung carcinoma in Japan was published by the Japan Lung Cancer Society (JLCS) in 1978 and used widely in Japan (Table 2) [12, 13]. This classification comprises three steps. First, the samples are classified as adequate for diagnostic evaluation or inadequate due to processing and/or staining artifacts. Second, if the samples are adequate, they are classified into the following three categories: (1) “negative,” (2) “suspicious,” and (3) “positive.” The “negative” category is used when neither the atypical cells of malignant tumor nor atypical cells originating in lesions that are borderline between malignant and benign can be identified. The “suspicious” category is used when the atypical cells suspicious of malignant tumor or atypical cells originating in lesions that are borderline between malignant and benign can be identified. The “positive” category is used when malignant tumor cells are identified. Third, the pathological classification is suggested based on the cytomorphological findings of the samples if possible. This classification has been mainly proposed for the cytological diagnosis of sputum.

However, owing to recent advancements in the techniques of obtaining samples, such as brushing, transbronchial aspiration cytology (TBAC), and endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA), the JLCS and the Japanese Society of Clinical Cytology (JSCC) have proposed a revision of the second step of their reporting system (Table 3). The classification in the second step should have the following four categories instead of the abovementioned three categories: (1) “negative for malignancy,” (2) “atypical cells,” (3) “suspicious for malignancy,” and (4) “malignancy.” The “suspicious” category in the previous reporting system has been divided into two categories: “atypical cells” and “suspicious for malignancy.” The category of “atypical cells” is used when cytomorphological aberrations are greater than those of “negative for malignancy” but less than those of “suspicious for malignancy.” The category of “suspicious for malignancy” is used when the cytological features are suspicious but not diagnostic of carcinoma because of the insufficient features characteristic of carcinomas and/or insufficient number of atypical cells.

To confirm the utility of the four-tiered reporting system, we performed an interobserver reproducibility study using the reporting systems, i.e., the previous JLCS three-tiered reporting system, newly proposed four-tiered reporting system, and five-tiered reporting system, on cytological samples obtained from brushing, TBAC, and EBUS-TBNA.

We collected 92 cytological samples from three hospitals (original hospitals) of the co-authors (Kagawa University Hospital, Osaka Habikino Medical Center, and Tokyo Medical University Hospital). They were not consecutive cases but selected by cytotechnologists or cytopathologists of the original hospitals for the purpose of this interobserver reproducibility study. Cases with difficulty in differentiation between benign and malignancy were frequently selected (Fig. 1a–c). To evaluate the target cells on the provided glass slides from the original hospitals, cases with an “inadequate” category were excluded. The Papanicolaou-stained glass slides were mailed to a management cytotechnologist (A.T.) of the co-authors. Data on cytological diagnosis according to the three- and four-tiered reporting systems and the final clinicopathological diagnosis of each case at the original hospitals were sent to the management cytotechnologist. Subsequently, the glass slides were randomized and anonymized by the management cytotechnologist and sent to the seven hospitals of the co-authors, including the three original hospitals. No clinical information or imaging findings were provided to the observers. There was no marking of atypical cells on the glass slides. The cytotechnologists and cytopathologists of the seven hospitals classified the slides into each category by reviewing one Papanicolaou-stained slide per case according to the three-, four-, and five-tiered reporting systems. The cytotechnologists and cytopathologists of each hospital agreed on the diagnostic category for each case. The cytotechnologist had 7–36 years of experience, and the cytopathologist had 10–22 years of experience. The results of the evaluation of the category on each case based on the three reporting systems were sent to the management cytotechnologist.

The age of the patients ranged from 46 to 88 years; 48 of them were men and 42 were women. Most of the samples were obtained from bronchial brushing, but those from TBAC, bronchoalveolar lavage, washing of forceps, and EBUS-TBNA were also included. Two cases of sputum were excluded because the preservation of cells on the glass slides of sputum cytology was different from that of the samples obtained with the other techniques. The final clinicopathological diagnosis was made based on the pathological diagnosis of biopsy of the same case at the original hospitals. However, 8 cases were diagnosed based on cytology and clinical findings because biopsy was not performed. The final clinicopathological diagnoses of the cases at the original hospitals included 2 benign tumors, 50 malignant tumors, 32 inflammatory conditions, and 6 negative conditions. The 2 benign tumors comprised 1 hamartoma and 1 sclerosing pneumocytoma. The malignant tumors comprised 46 primary lung carcinomas (32 adenocarcinomas, 11 squamous cell carcinomas, and 3 small cell carcinomas) and 4 metastatic lung tumors. The inflammatory conditions comprised interstitial pneumonia, organizing pneumonia, bronchopneumonia, granulomatous disease, and non-specific inflammation. The biopsy in the negative conditions showed no significant change (Table 4).

The samples were classified according to the three-tiered reporting system of the JLCS, newly proposed four-tiered reporting system by the JLCS and JSCC, and five-tiered reporting system of the modified PSC published in 2016 [10]. Given that clinical information was not obtained from the cases and there were at least some cells on the slides in this study, the “non-diagnostic” category was excluded from the original classification of the PSC. The five-tiered reporting system of this study comprises the following categories: “negative for malignancy,” “atypical,” “neoplastic, benign neoplasm, low-grade carcinoma,” “suspicious for malignancy,” and “positive for malignancy.”

Agreement was regarded as unanimous if all the observers agreed and consensus if four or more of the observers agreed. Risk of malignancy (ROM) was defined for each category as the number of confirmed malignant cases divided by the total number of cases in the diagnostic category of each observer, and the mean ROM among the observers was calculated.

Statistical analysis was performed by the JMP 14 (SAS Institute Inc., Cary, NC, USA). The overall agreement was calculated according to the recommendation of Landis and Koch [14], with kappa (κ) <0.40 indicating poor agreement, κ = 0.40–0.75 fair agreement, and κ >0.75 excellent agreement. The kappa score was calculated by comparing the categories in each reporting system submitted by the seven hospitals.

Figure 2a and b show the relationship between the cytological and clinicopathological diagnoses at the original hospitals in the three- and four-tiered reporting systems. Negative conditions, inflammatory conditions, and benign tumors were combined together into negative cases.

In the three-tiered reporting system, malignant tumors were cytologically diagnosed as “positive” in 22 (44%) and “suspicious” in 28 (56%) cases. Clinicopathologically negative cases were diagnosed as “negative” in 20 (50%) and “suspicious” in 20 (50%) cases. Two benign tumors were cytologically diagnosed as “negative.”

In the four-tiered reporting system, malignant tumors were cytologically diagnosed as “malignant” in 22 (44%), “suspicious for malignancy” in 25 (50%), and “atypical cells” in 3 (6%) cases (cases 17, 71, and 92). There were no carcinoma cells on the Papanicolaou-stained glass slides of 2 cases with adenocarcinoma (cases 17 and 71) and 1 case with squamous cell carcinoma (case 92); however, there were some atypical cells observed on the glass slides. The original hospital diagnosed these cases as “atypical cells.” Clinicopathologically negative cases were diagnosed as “negative” in 20 (50%), “atypical cells” in 19 (47.5%), and “suspicious for malignancy” in 1 (2.5%) case (case 24). The Papanicolaou-stained glass slide of case 24 showed a cluster of atypical squamous cells with cannibalism (Fig. 1b). The original hospital diagnosed this case as “suspicious for malignancy.” Two cases of benign tumors were cytologically diagnosed as “negative.”

The interobserver agreement was fair in the three reporting systems. However, the agreement was better in the three-tiered reporting system (κ = 0.50) than in the four- (κ = 0.45) and five-tiered (κ = 0.45) reporting systems. After combining the “suspicious for malignancy” and “malignancy” categories in the four- and five-tiered reporting systems, the interobserver agreement in the four- (κ = 0.52) and five-tiered (κ = 0.53) systems improved.

Figure 3 shows the distribution of the number of concordant observers in each system. No significant difference was observed among them. Consensus agreement in the three-, four-, and five-tiered reporting systems was 88.9, 80.0, and 80.0%, respectively. Figure 4 shows the distribution of the number of concordant observers in the four- and five-tiered reporting systems after the “suspicious for malignancy” and “malignancy” categories were combined. Consensus agreement in the four- and five-tiered reporting system was 92.2 and 92.2%, respectively.

Among the seven observers, consensus agreement was obtained in 72 of 90 cases (80.0%) and unanimous agreement in 32 of 90 cases (35.6%) in the four-tiered reporting system. Figure 5 shows the number of cytological agreements among 7 observers in the four-tiered reporting system depending on the clinicopathological diagnosis at the original hospitals. Negative conditions, inflammatory conditions, and benign tumors were combined together into negative cases. Consensus agreement was obtained in 77.5% of negative cases and in 80% of malignant tumors. Unanimous agreement was obtained in 37.5% of negative cases and in 34.0% of malignant tumors. Unanimous agreement was obtained in 2 benign tumors.

We evaluated the agreement of cytological diagnosis of each observer with the final clinicopathological diagnosis at the original hospitals. “Suspicious for malignancy” and “malignancy” categories were regarded as agreement in malignant tumors, and the “negative for malignancy” and “atypical cells” categories were regarded as agreement in negative cases. Consensus agreement was obtained in 73 of 90 cases (81.14%) and unanimous agreement in 46 of 90 cases (51.1%). Figure 6 shows the number of agreements of cytological diagnosis of each observer with the final clinicopathological diagnosis at the original hospitals in the four-tiered reporting system depending on the status of the diseases. Negative conditions, inflammatory conditions, and benign tumors were combined together into negative cases. Consensus agreement was obtained in 80.0% of negative cases and in 82.0% of malignant tumors. Unanimous agreement was obtained in 52.5% of negative cases, 50.0% of malignant tumors, and 2 benign tumors. There were 4 cases in which the number of agreements of cytological diagnosis of each observer with the final clinicopathological diagnosis was zero. Three cases were malignant tumors and 1 case was a negative case. The types of malignant tumors observed were 2 adenocarcinomas (cases 17 and 71) and 1 squamous cell carcinoma (case 92). Because there were no malignant cells on the glass slides, no observer diagnosed these cases as “suspicious for malignancy” or “malignant.” The Papanicolaou-stained glass slide of 1 negative case (case 24) showed a cluster of atypical squamous cells with cannibalism (Fig. 1b), and 5 observers diagnosed this case as “malignant,” whereas 2 diagnosed this as “suspicious for malignancy.”

The average ROM among the 7 observers in the four-tiered reporting system was 19.3% for the “negative for malignancy” category, 45.6% for the “atypical cells” category, 74.7% for the “suspicious for malignancy” category, and 88.1% for the “malignancy” category (Table 5). The ROM for the “inadequate” category could not be obtained because cases in the “inadequate” category were excluded from this analysis.

This study showed that the interobserver agreement of the four-tiered reporting system was fair (κ = 0.45). When the “suspicious for malignancy” and “malignancy” categories were combined, the interobserver agreement improved further (κ = 0.52). The interobserver reproducibility was excellent in benign tumors. However, when atypia was observed on the specimens in the slides, the interobserver agreement was lower than that in the benign tumors. The agreement of cytological diagnosis with the final clinicopathological diagnosis was better in benign tumors but fair in other conditions.

In this study, 90 respiratory cases were collected from three hospitals and retrospectively reclassified based on the three-, four-, and five-tiered reporting systems. The associated ROM for each category in the four-tiered reporting system was also analyzed (Table 5). The ROM in this study for “negative for malignancy” (19.3%) was lower than that reported by previous reports [10, 15]. This can be explained by the method of the sample collection. The samples were not from consecutive cases but collected for this reproducibility study. The original hospitals tended to submit definitely “negative” category cases for slides with negative findings, and the number of cases with malignant neoplasm showing a negative finding on the slides were few. The ROM of “atypical cells,” “suspicious for malignancy,” and “malignancy” here was comparable with other ROM in the PSC [10, 15]. There may be a ROM for the “inadequate” category; however, this could not be obtained because cases with an “inadequate” category were excluded in this study.

Patient management must correlate with the cytological reports of the lung after cytological evaluation. Thus, we propose the following management options (Table 5): (a) if the specimen is considered to be “inadequate,” the reasons for the inadequacy should be documented in the report and re-examination should be strongly recommended; (b) even if cytological diagnosis is “negative for malignancy,” clinical follow-up is needed because these patients have been in an unusual situation where cytological evaluation of the disease is necessary; (c) when cytological diagnosis is “atypical cells,” clinical follow-up is needed and re-examination is recommended; (d) when cytological diagnosis is “suspicious for malignancy,” further examination, including biopsy, is recommended, and when resection of the lung is scheduled to treat the patient, frozen section of the tumor by partial resection of the lung should be performed to confirm that the tumor is malignant; and (e) when the cytological diagnosis is “malignancy,” therapy should be provided as soon as possible.

The criteria for diagnosing atypical cells were defined using four cytomorphological features: nuclear cytoplasmic ratio, hyperchromasia, irregular nuclear membrane, and coarse chromatin in the Paris system for reporting urinary cytology [5]. For the reporting system for respiratory cytology, we do not have any objective criteria for “atypical cells” and “suspicious for malignancy.” We have used the following criteria for “atypical cells” and “suspicious for malignancy” in the reporting system for respiratory cytology. “Atypical cells” are characterized by three-dimensional clusters, large and polygonal cells, enlarged nuclei, high N/C ratios, irregular macronucleoli, smooth and regular nuclear membranes, occasional preservation of cell polarity, and degenerative smudgy nuclei with absent or little nuclear detail. Smears are often hypocellular, and mitoses and single atypical cells are absent or rare. Significant background inflammation may be present [16]. We must make efforts to minimize the reporting rate of “atypical cells” as much as possible. By doing so, we can help guide clinicians toward an appropriate management strategy and reduce patient anxiety related to an indeterminate diagnosis. When malignant cytological features are insufficiently developed or are expressed as an insufficient number of cells for the definitive diagnosis of malignancy, assignment to the category “suspicious for malignancy” is appropriate. Specimens with small flat sheets with anisonucleosis, nuclear groove, and/or cytoplasmic nuclear pseudoinclusion, which are suspicious for adenocarcinoma in situ, are assigned to “suspicious for malignancy.” There are different cytological techniques to obtain respiratory lesions, such as bronchial brushing, TBAC, EBUS-TBNA, and computed tomography-guided fine-needle aspiration. There may be different cytological features related to different sampling techniques; however, the differences have not been reported in the literature [9, 16]. We cannot comment on this problem because most of the samples addressed here were obtained via bronchial brushing. We recognize that the preservation of cells on the glass slides for sputum cytology was different from that of the samples obtained using other techniques.

Cases identified as “suspicious” in the three-tiered reporting system were further classified into “atypical cells” or “suspicious for malignancy” in the four-tiered reporting system. Most of the “suspicious” cases in malignant tumors in the three-tiered reporting system were moved to “suspicious for malignancy” in the four-tiered reporting system. Most of the “suspicious” cases in negative conditions in the three-tiered reporting system were moved to “atypical cells” in the four-tiered reporting system. The four-tiered reporting system proposes more precise information in patient management after cytological diagnosis than the three-tiered reporting system. In the negative conditions, one identified as “suspicious” in the three-tiered reporting system was moved to “suspicious for malignancy” in the four-tiered reporting system (case 24). The Papanicolaou-stained glass slide of this case showed a cluster of atypical squamous cells with cannibalism (Fig. 1b). Most observers suspected this as squamous cell carcinoma. However, given that only one cluster was observed on a glass slide of Papanicolaou staining, tumor cells were not identified with transbronchial biopsy and follow-up of chest roentgenogram and clinical findings did not show any sign of malignancy, indicating that this case must have been overdiagnosed by the observers. Equivocal lung cytology is a recurring problem in every cytology laboratory and is, thus, a diagnostic challenge even for highly experienced cytopathologists and cytotechnologists [17].

The lung reporting system by the PSC evaluates specimen adequacy [9, 11]. “Satisfactory for evaluation” is used when a specimen yields a material representative of a specific pathologic entity, whereas “unsatisfactory for evaluation” is used when the specimen cannot be evaluated [11]. The PSC also includes the “non-diagnostic” category. It is defined as a specimen that comprises only normal ciliated respiratory epithelium, non-ciliated bronchiolar epithelium, pneumocytes, and/or macrophages when a distinct nodule, cavity, or mass is identified on imaging. Specimens in which cells are obscured by clotted blood or other preparation artifacts are also assigned to this category. For diagnosis of the “non-diagnostic” category of the PSC, radiological information is needed. However, no clinical or radiological information was provided to the observers in our study. Without this information, it is impossible to determine whether the epithelial cells were derived from benign tumors or normal lung. The reporting system for lung carcinoma proposed by the JLCS and JSCC evaluates specimen adequacy. “Adequate” is used when abundant cellular materials are present, whereas “inadequate” is used when abundant cellular material is absent or artifacts obscuring cell identity are present. Specimens that cannot be assessed due to excessive blood or mucus are categorized as “inadequate” in the reporting system used by the JLCS and JSCC. This is due to the same patient management used for these two categories; therefore, repeat cytological sampling with other techniques and/or biopsy should be performed. Even if cytological sampling is “inadequate,” the technical billing charge can be covered by medical insurance in Japan. If the sample contains abnormal cells, no matter how few, the specimen is considered “adequate.” Our reporting system has been designed for disease diagnoses in patients with pulmonary nodule as well as in patients with radiographically diffuse shadow, such as inflammatory conditions and diffuse malignant neoplasm.

In the four-tiered reporting system, “negative for malignancy” may comprise several specific inflammatory and degenerative processes, including bacterial pneumonia, viral pneumonia, interstitial pneumonia, fungal infection, mycobacterial infection, abscess, eosinophilic pneumonia, granuloma, pulmonary fibrosis, emphysema, and bronchial asthma. Benign tumors (e.g., hamartoma and sclerosing pneumocytoma) are also categorized as “negative for malignancy.” “Atypical cells” comprises conditions such as radiation pneumonitis, chemotherapy, pulmonary infarct, oxygen therapy, environmental toxin, and ARDS. Cytomorphological aberrations are greater than those recognized due to inflammation or repair but lesser than those assigned to “suspicious for malignancy.” Given that the respiratory system is often affected by infection or environmental insults, “atypical cells” is frequently encountered on cytological examination in daily practice. Inflammatory changes were cytologically diagnosed as “atypical cells” frequently in this study. Atypical cells obtained via bronchial brushing in a patient with acute interstitial pneumonia (case 80) revealed the presence of large nuclei and prominent nucleoli. Three out of 7 observers evaluated this case as “malignant” (Fig. 1c). While these observations have been noted in patients with florid inflammatory changes, the possibility of a malignant tumor cannot be completely excluded. Clinical follow-up and further examination are recommended for such patients if necessary.

Two cases of benign tumor were reported in our study: hamartoma and sclerosing pneumocytoma. Both these samples were obtained with bronchial brushing. Cytological evaluation of these cases was “negative for malignancy” by all the observers in the four- and five-tiered reporting systems, although the five-tiered reporting system has the “neoplastic, benign neoplasm, low-grade carcinoma” category. The sample of the case with hamartoma did not include a chondromatous component and provide clue for the diagnosis of hamartoma. The diagnosis of benign tumor in cytological samples is difficult without radiological information. The tumor cells from sclerosing pneumocytoma may also be difficult to differentiate from normal pneumocyte epithelium in the lung. The PSC system includes the “neoplastic, benign neoplasm, low-grade carcinoma” category. However, we were unable to determine if the tumor cells were definitely derived from sclerosing pneumocytoma or hamartoma and not from normal alveolar tissue because no cytological atypia was observed in the sclerosing pneumocytoma or hamartoma. For these reasons, we did not include the “benign tumor” category in our newly proposed four-tiered reporting system. A definitive diagnosis of benign tumor requires a constellation of findings that includes clinical history and radiographic, cytological, and biopsy findings.

Our study has some limitations. First, given that clinical information or imaging findings were not provided to the observers, cytological diagnosis was made only based on the Papanicolaou-stained glass slides. This method of cytological evaluation is different from that in cytological diagnosis in daily practice. If clinical information and imaging findings were provided to the observers, cytological agreement among the observers would improve. Second, the samples were not consecutive cases but selected by cytotechnologists or cytopathologists of the original hospitals for the purpose of this interobserver reproducibility study. Especially, cases with difficulty in differentiation between benign and malignancy were included. This selection bias made the cytological diagnosis in this study more difficult than the routine cytological diagnosis in daily practice. A study using the four-tiered reporting system in the routine samples is warranted to confirm our results. Third, TBACs or EBUS-TBNAs, which make a significant proportion of cases in diagnostic practice, are clearly under-represented in this study. However, this bias is due to the sample collection method of this study. We have intentionally collected cytological samples via bronchial brushing because this technique is often performed in Japan for the diagnosis of abnormal findings in the chest roentgenogram. Further study using samples obtained via TBAC or EBUS-TBNA is needed to generalize the results of our study. Fourth, some slides had few atypical cells with artifact due to drying of the sample. Given that there are few target cells, we did not exclude these samples from the study. It was difficult to make a cytological diagnosis based on these slides, and agreement among the observers was poor. Fifth, the Papanicolaou staining procedure is slightly different among the three hospitals that provided the slides. Consequently, the observers reviewed the slides with different colorization of the cytoplasm and nuclei of the target cells. Sixth, only 2 benign tumors were identified in this study. Although the interobserver reproducibility of the cytological diagnosis of benign tumors was 100%, there is a need to evaluate further cases to conclude that interobserver reproducibility is excellent in benign tumors.

In conclusion, the reporting system proposed by the JLCS and JSCC was designed to enhance the communication between clinicians and pathologists and among different institutions. Given that this system is a revision of the reporting system for lung carcinoma in Japan by the JLCS, we evaluate the adequacy of the samples before categorization. Specimens that cannot be assessed due to excessive blood or mucus and those of the “unsatisfactory evaluation” category of the PSC classification are combined together into the “inadequate” category in the JLCS and JSCC classification. The four-tiered reporting system is relatively simple and applicable to cytological diagnosis of any respiratory diseases. Each category of the four-tiered reporting system has different ROM and provides appropriate information concerning patient management. A study using the four-tiered reporting system in routine samples is warranted to confirm our results. We propose establishing an international classification for respiratory cytology, harmonizing the reporting systems proposed by different countries. Although we propose the use of the four-tiered reporting system as a classification for respiratory cytology, international reporting systems in other organs do not adopt the two-step evaluation system. Therefore, the “inadequate” category could be one of the categories in the one-step evaluation system in an international reporting system for lung carcinoma that will be provided in the near future.

The authors thank Prof. Robert Y. Osamura, Prof. Fernando Schmitt, Prof. Lukas Bubendorf, and Dr. Lan Chen for inspiring us to establish a new cytological reporting system for lung carcinoma. The authors also thank Dr. Satoru Shimizu for giving us advice on statistical analysis of the data in this interobserver reproducibility study.

The study protocol has been approved by the ethics committee of Tokyo Women’s Medical University (approval No. 4873, approval date: July 26, 2018) and the ethics committee of all participating facilities.

The authors have no conflicts of interest to declare.

The Japan Lung Cancer Society and the Japanese Society of Clinical Cytology paid the expense of the meeting, postal fee of the glass slides, and fee for the English editing service.

Study concept and design: K. Hiroshima, A. Yoshizawa, A. Takenaka, R. Haba, K. Kawahara, Y. Minami, H. Kakinuma, Y. Shibuki, S. Miyake, and Y. Satoh. Providing glass slides: R. Haba, K. Kawahara, S. Miyake, K. Kajio, K. Miyamoto, J. Matsubayashi, and T. Nagao. Collecting and management of glass slides: A. Ta­kenaka. Reviewing the glass slides: A. Takenaka, R. Haba, K. Ka­wahara, H. Kakinuma, Y. Shibuki, S. Miyake, K. Kajio, K. Miyamoto, M. Nagatomo, S. Nishimura, M. Mano, J. Matsubayashi, N. Motoi, T. Nagao, S. Nakatsuka, and T. Yoshida. Analysis and interpretation of data: K. Hiroshima, A. Yoshizawa, A. Takenaka, and Y. Satoh. Drafting of the manuscript: K. Hiroshima. Critical revision of the manuscript: A. Yoshizawa, R. Haba, K. Kawahara, Y. Minami, and Y. Satoh.