ABSTRACT
Objectives@#This study was conducted to develop a generalizable annotation tool for bilingual complex clinical text annotation, which led to the design and development of a clinical text annotation tool, ANNO. @*Methods@#We designed ANNO to enable human annotators to support the annotation of information in clinical documents efficiently and accurately. First, annotations for different classes (word or phrase types) can be tagged according to the type of word using the dictionary function. In addition, it is possible to evaluate and reconcile differences by comparing annotation results between human annotators. Moreover, if the regular expression set for each class is updated during annotation, it is automatically reflected in the new document. The regular expression set created by human annotators is designed such that a word tagged once is automatically labeled in new documents. @*Results@#Because ANNO is a Docker-based web application, users can use it freely without being subjected to dependency issues. Human annotators can share their annotation markups as regular expression sets with a dictionary structure, and they can cross-check their annotated corpora with each other. The dictionary-based regular expression sharing function, cross-check function for each annotator, and standardized input (Microsoft Excel) and output (extensible markup language [XML]) formats are the main features of ANNO. @*Conclusions@#With the growing need for massively annotated clinical data to support the development of machine learning models, we expect ANNO to be helpful to many researchers.
ABSTRACT
BACKGROUND@#Human leukocyte antigen (HLA) typing is important for transplant patients to prevent a severe mismatch reaction, and the result can also support the diagnosis of various disease or prediction of drug side effects. However, such secondary applications of HLA typing results are limited because they are typically provided in free-text format or PDFs on electronic medical records. We here propose a method to convert HLA genotype information stored in an unstructured format into a reusable structured format by extracting serotype/allele information.@*METHODS@#We queried HLA typing reports from the clinical data warehouse of Seoul National University Hospital (SUPPREME) from 2000 to 2018 as a rule-development data set (64,024 reports) and from the most recent year (6,181 reports) as a test set. We used a rule-based natural language approach using a Python regex function to extract the 1) number of patients in the report, 2) clinical characteristics such as indication of the HLA testing, and 3) precise HLA genotypes. The performance of the rules and codes was evaluated by comparison between the extracted results from the test set and a validation set generated by manual curation.@*RESULTS@#Among 11,287 reports for development set and 1,107 for the test set describing HLA typing for a single patient, iterative rule generation developed 124 extracting rules and 8 cleaning rules for HLA genotypes. Application of these rules extracted HLA genotypes with 0.892–0.999 precision and 0.795–0.998 recall for the five HLA genes. The precision and recall of the extracting rules for the number of patients in a report were 0.997 and 0.994 and those for the clinical variable extraction were 0.997 and 0.992, respectively. All extracted HLA alleles and serotypes were transformed according to formal HLA nomenclature by the cleaning rules.@*CONCLUSION@#The rule-based HLA genotype extraction method shows reliable accuracy. We believe that there are significant number of patients who takes profit when this under-used genetic information will be return to them.
ABSTRACT
BACKGROUND: Human leukocyte antigen (HLA) typing is important for transplant patients to prevent a severe mismatch reaction, and the result can also support the diagnosis of various disease or prediction of drug side effects. However, such secondary applications of HLA typing results are limited because they are typically provided in free-text format or PDFs on electronic medical records. We here propose a method to convert HLA genotype information stored in an unstructured format into a reusable structured format by extracting serotype/allele information.METHODS: We queried HLA typing reports from the clinical data warehouse of Seoul National University Hospital (SUPPREME) from 2000 to 2018 as a rule-development data set (64,024 reports) and from the most recent year (6,181 reports) as a test set. We used a rule-based natural language approach using a Python regex function to extract the 1) number of patients in the report, 2) clinical characteristics such as indication of the HLA testing, and 3) precise HLA genotypes. The performance of the rules and codes was evaluated by comparison between the extracted results from the test set and a validation set generated by manual curation.RESULTS: Among 11,287 reports for development set and 1,107 for the test set describing HLA typing for a single patient, iterative rule generation developed 124 extracting rules and 8 cleaning rules for HLA genotypes. Application of these rules extracted HLA genotypes with 0.892–0.999 precision and 0.795–0.998 recall for the five HLA genes. The precision and recall of the extracting rules for the number of patients in a report were 0.997 and 0.994 and those for the clinical variable extraction were 0.997 and 0.992, respectively. All extracted HLA alleles and serotypes were transformed according to formal HLA nomenclature by the cleaning rules.CONCLUSION: The rule-based HLA genotype extraction method shows reliable accuracy. We believe that there are significant number of patients who takes profit when this under-used genetic information will be return to them.
ABSTRACT
BACKGROUND@#Human leukocyte antigen (HLA) typing is important for transplant patients to prevent a severe mismatch reaction, and the result can also support the diagnosis of various disease or prediction of drug side effects. However, such secondary applications of HLA typing results are limited because they are typically provided in free-text format or PDFs on electronic medical records. We here propose a method to convert HLA genotype information stored in an unstructured format into a reusable structured format by extracting serotype/allele information.@*METHODS@#We queried HLA typing reports from the clinical data warehouse of Seoul National University Hospital (SUPPREME) from 2000 to 2018 as a rule-development data set (64,024 reports) and from the most recent year (6,181 reports) as a test set. We used a rule-based natural language approach using a Python regex function to extract the 1) number of patients in the report, 2) clinical characteristics such as indication of the HLA testing, and 3) precise HLA genotypes. The performance of the rules and codes was evaluated by comparison between the extracted results from the test set and a validation set generated by manual curation.@*RESULTS@#Among 11,287 reports for development set and 1,107 for the test set describing HLA typing for a single patient, iterative rule generation developed 124 extracting rules and 8 cleaning rules for HLA genotypes. Application of these rules extracted HLA genotypes with 0.892–0.999 precision and 0.795–0.998 recall for the five HLA genes. The precision and recall of the extracting rules for the number of patients in a report were 0.997 and 0.994 and those for the clinical variable extraction were 0.997 and 0.992, respectively. All extracted HLA alleles and serotypes were transformed according to formal HLA nomenclature by the cleaning rules.@*CONCLUSION@#The rule-based HLA genotype extraction method shows reliable accuracy. We believe that there are significant number of patients who takes profit when this under-used genetic information will be return to them.
ABSTRACT
PURPOSE: The purpose of this study was to investigate the effectiveness and safety of spinal stereotactic radiosurgery (SRS) in treating spinal metastasis with epidural spinal cord compression (ESCC). MATERIALS AND METHODS: During 2013-2016, 149 regions of spinal metastasis in 105 patients treated with singlefraction (12-24 Gy) spinal SRS were reviewed. Cord compression of Bilsky grade 2 (with visible cerebrospinal fluid [CSF]) or 3 (no visible CSF) was defined as ESCC. Local progression (LP) and vertebral compression fracture (VCF) rates after SRS were evaluated using multivariate competing-risk regression analysis. RESULTS: The 1-year cumulative incidences of LP for Bilsky grades 0 (n=80), 1 (n=39), 2 (n=21), and 3 (n=9) were 3.0%, 8.4%, 0%, and 24.9%, respectively. Bilsky grade 2 ESCC did not significantly increase the LP rate (no LP for grade 2). The 1-year cumulative incidences of VCF for Bilsky grades 0, 1, 2, and 3 were 6.6%, 5.2%, 17.1%, and 12.1%, respectively. ESCC may increase VCF risk (subhazard ratio [SHR] for grade 2, 5.368; p=0.035; SHR for grade 3, 2.215; p=0.460). Complete or partial pain response rates after SRS were 79%, 78%, 53%, and 63% for Bilsky grades 0, 1, 2, and 3, respectively (p=0.008). No neurotoxicity of grade ≥ 3 was observed. CONCLUSION: Spinal SRS for spinal metastasis with Bilsky grade 2 ESCC did not increase the LP rate, was not associated with severe neurotoxicity, and showed moderate VCF and pain response rates. Bilsky grade 3 had a high LP rate.
Subject(s)
Humans , Cerebrospinal Fluid , Disease Progression , Fractures, Compression , Incidence , Neoplasm Metastasis , Radiosurgery , Spinal Cord Compression , SpineABSTRACT
PURPOSE: The purpose of this study was to investigate the non-inferiority of omitting radiotherapy (RT) after breast-conserving surgery (BCS) for hormone receptor (HR)‒positive T1N0 breast cancer in elderly women. MATERIALS AND METHODS: From 2004 to 2014, HR-positive T1N0 breast cancer patients aged 50 years or older and receiving BCS were retrieved from the Surveillance, Epidemiology, and End RESULTS: 18 database. After propensity score matching between the no-RT and RT groups, univariate and multivariate analyses were performed. Identified prognostic factors were used to stratify the risk groups. In each risk group, 10-year cancer-specific survival (CSS) rates were compared between the no-RT and RT groups. RESULTS: After propensity score matching, the numbers of patients in the no-RT and RT groups were both 18,586. For patients who satisfied both a tumor size of 1-10 mm and a tumor grade of 1-2, omitting RT did not decrease the CSS rate at any age group, ranging from ≥ 50 to ≥ 85 years; for patients aged ≥ 50 years, the 10-year CSS rates in the no-RT and RT groups were 97.2% and 96.8%, respectively (adjusted hazard ratio, 0.862; p=0.312). However, for patients with a tumor size of 11-20 mm or tumor grade of 3-4, RT significantly increased the CSS rate irrespective of age. CONCLUSION: RT after BCS for HR-positive T1N0 breast cancer in elderly women might be omitted without causing a decrease in the CSS rate, but only in patients who satisfy both a small tumor size (≤ 10 mm) and low tumor grade (1-2).
Subject(s)
Aged , Female , Humans , Breast Neoplasms , Breast , Epidemiology , Mastectomy, Segmental , Multivariate Analysis , Propensity Score , Radiotherapy , Radiotherapy, Adjuvant , Receptors, Estrogen , Receptors, ProgesteroneABSTRACT
BACKGROUND: The purpose of this study was to evaluate the treatment outcome of our optic nerve sheath meningioma (ONSM) case series in terms of preventing tumor growth and preserving vision in ONSM patients. METHODS: Between July 2003 and March 2015, 1,398 patients with intracranial meningioma were diagnosed at Seoul National University Bundang Hospital. Among them, only 13 patients (0.93%) were diagnosed with ONSM and enrolled in the present study. Tumor volume changes of ONSM patients and their visual acuity were evaluated before and after treatments. RESULTS: The median follow-up time was 50 months (range, 12–133 months). Visual acuity was evaluated in 12 of 13 patients, and visual acuity was found to be preserved in 9 of 12 patients (75%). Tumor volume was reduced in all patients. The tumor control rate was 100% in the present study. The difference in tumor volume between pretreatment and last follow-up was statistically significant (p=0.015). CONCLUSION: Intensity-modulated radiotherapy (IMRT) and gamma knife radiosurgery (GKS) could maintain visual acuity and stabilize tumor volume in ONSM patients, suggesting that IMRT and GKS may be effective therapies for ONSM. However, which treatment is the more effective modality must be confirmed by prospective studies and longer-term follow-up.
Subject(s)
Humans , Follow-Up Studies , Meningioma , Optic Nerve , Prospective Studies , Radiosurgery , Radiotherapy, Intensity-Modulated , Seoul , Treatment Outcome , Tumor Burden , Visual AcuityABSTRACT
PURPOSE: The optimal indications for omitting adjuvant radiation therapy (RT) after breast-conserving surgery are still controversial in ductal carcinoma in situ (DCIS) of the breast. The purpose of this study was to validate the role of postoperative RT in DCIS patients aged ≤50 years and with tumor margin widths of <1 cm, both of which have been proven to be high-risk features for recurrence in cohorts not receiving RT. METHODS: Using two multicenter retrospective studies on DCIS, a pooled analysis was performed among patients aged ≤50 years and with margin widths < 1 cm. All patients underwent breast-conserving surgery. Two hundred thirty-two patients received postoperative RT, while 54 did not. The median follow-up period was 77 months (range, 2–190 months) and 70 months (range, 5–166 months) in the patients who received RT and those who did not, respectively. RESULTS: The patients who received RT had larger tumors (p < 0.001), higher nuclear grade (p < 0.001), closer margin width (p < 0.001), and negative estrogen receptor expression (p=0.010) compared with those who did not receive RT. During the follow-up period, there were 17 ipsilateral breast tumor recurrences (IBTRs) as follows: invasive carcinoma in 10 patients and DCIS in seven. In the univariate analysis, the treatment with RT and human epidermal growth factor receptor 2 (HER2) status were significant risk factors for IBTR. The 7-year IBTR rates with and without postoperative RT were 3.6% and 13.1%, respectively (p=0.008). HER2-positive tumors had a higher IBTR rate than the HER2-negative tumors (7-year rate, 13.6% vs. 3.9%; p=0.003). CONCLUSION: Postoperative RT following breast-conserving surgery significantly reduced the 7-year IBTR rate in the DCIS patients aged ≤50 years and with margin widths < 1 cm. HER2 positivity was associated with increased IBTR in these patients.
Subject(s)
Humans , Age Factors , Breast , Breast Neoplasms , Carcinoma, Ductal , Carcinoma, Intraductal, Noninfiltrating , Cohort Studies , Estrogens , Follow-Up Studies , Mastectomy, Segmental , Radiation Oncology , Radiotherapy , ErbB Receptors , Recurrence , Retrospective Studies , Risk FactorsABSTRACT
PURPOSE: To evaluate the diagnostic accuracy of computed tomography (CT), magnetic resonance imaging (MRI), and positron emission tomography/computed tomography (PET/CT) in predicting pelvic lymph node (LN) metastases in patients with cervical cancer. MATERIALS AND METHODS: From January 2009 to March 2015, 114 patients with FIGO stage IA1-IIB uterine cervical cancer who underwent hysterectomy with pelvic lymphadenectomy and took CT, MRI, and PET/CT before surgery were enrolled in this study. The criteria for LN metastases were a LN diameter ≥1.0 cm and/or the presence of central necrosis on CT, a LN diameter ≥1.0 cm on MRI, and a focally increased FDG uptake on PET/CT. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy for pelvic LN metastases were estimated. RESULTS: The sensitivity, specificity, PPV, NPV, and accuracy for detection of pelvic LN metastases were 51.4%, 85.9%, 41.3%, 90.1%, and 80.3% for CT; 24.3%, 96.3%, 56.3%, 86.8%, and 84.6% for MRI; and 48.6%, 89.5%, 47.4%, 90.0%, and 82.9% for PET/CT, respectively. The sensitivity of PET/CT and CT was higher than that of MRI (p=0.004 and p= 0.013, respectively). The specificity of MRI was higher than those of PET/CT and CT (p=0.002 and p=0.001, respectively). The difference of specificity between PET/CT and CT was not statistically significant (p=0.167). CONCLUSION: These results indicate that preoperative CT, MRI, and PET/CT showed low to moderate sensitivity and PPV, and moderate to high specificity, NPV, and accuracy. More efforts are necessary to improve sensitivity of imaging modalities in order to predict pelvic LN metastases.
Subject(s)
Humans , Electrons , Hysterectomy , Lymph Node Excision , Lymph Nodes , Magnetic Resonance Imaging , Necrosis , Neoplasm Metastasis , Positron-Emission Tomography , Positron Emission Tomography Computed Tomography , Sensitivity and Specificity , Uterine Cervical NeoplasmsABSTRACT
PURPOSE: The purpose of this study is to evaluate the treatment outcomes of adjuvant radiotherapy using vaginal brachytherapy (VB) with a lower dose per fraction and/or external beam radiotherapy (EBRT) following surgery for patients with stage I endometrial carcinoma. MATERIALS AND METHODS: The subjects were 43 patients with the International Federation of Gynecology and Obstetrics (FIGO) stage I endometrial cancer who underwent adjuvant radiotherapy following surgery between March 2000 and April 2014. Of these, 25 received postoperative VB alone, while 18 received postoperative EBRT to the whole pelvis; 3 of these were treated with EBRT plus VB. The median EBRT dose was 50.0 Gy (45.0–50.4 Gy) and the VB dose was 24 Gy in 6 fractions. Tumor dose was prescribed at a depth of 5 mm from the cylinder surface and delivered twice per week. RESULTS: The median follow-up period for all patients was 57 months (range, 9 to 188 months). Five-year disease-free survival (DFS) and overall survival (OS) for all patients were 92.5% and 95.3%, respectively. Adjuvant radiotherapy was performed according to risk factors and stage IB, grade 3 and lymphovascular invasion were observed more frequently in the EBRT group. Five-year DFS for EBRT and VB alone were 88.1% and 96.0%, respectively (p = 0.42), and 5-year OS for EBRT and VB alone were 94.4% and 96%, respectively (p = 0.38). There was no locoregional recurrence in any patient. Two patients who received EBRT and 1 patient who received VB alone developed distant metastatic disease. Two patients who received EBRT had severe complications, one each of grade 3 gastrointestinal complication and pelvic bone insufficiency fracture. CONCLUSION: Adjuvant radiotherapy achieved high DFS and OS with acceptable toxicity in stage I endometrial cancer. VB (with a lower dose per fraction) may be a viable option for selected patients with early-stage endometrial cancer following surgery.
Subject(s)
Female , Humans , Brachytherapy , Disease-Free Survival , Endometrial Neoplasms , Follow-Up Studies , Fractures, Stress , Gynecology , Obstetrics , Pelvic Bones , Pelvis , Radiotherapy , Radiotherapy, Adjuvant , Radiotherapy, Conformal , Recurrence , Risk FactorsABSTRACT
PURPOSE: To analyze prognostic factors for locoregional recurrence (LRR), distant metastasis (DM), and overall survival (OS) in cervical cancer patients who underwent radical hysterectomy followed by postoperative radiotherapy (PORT) in a single institute. MATERIALS AND METHODS: Clinicopathologic data of 135 patients with clinical stage IA2 to IIA2 cervical cancer treated with PORT from 2001 to 2012 were reviewed, retrospectively. Postoperative parametrial resection margin (PRM) and vaginal resection margin (VRM) were investigated separately. The median treatment dosage of external beam radiotherapy (EBRT) to the whole pelvis was 50.4 Gy in 1.8 Gy/fraction. High-dose-rate vaginal brachytherapy after EBRT was given to patients with positive or close VRMs. Concurrent platinum-based chemoradiotherapy (CCRT) was administered to 73 patients with positive resection margin, lymph node (LN) metastasis, or direct extension of parametrium. Kaplan-Meier method and log-rank test were used for analyzing LRR, DM, and OS; Cox regression was applied to analyze prognostic factors. RESULTS: The 5-year disease-free survival was 79% and 5-year OS was 91%. In univariate analysis, positive or close PRM, LN metastasis, direct extension of parametrium, lymphovascular invasion, histology of adenocarcinoma, and chemotherapy were related with more DM and poor OS. In multivariate analysis, PRM and LN metastasis remained independent prognostic factors for OS. CONCLUSION: PORT after radical hysterectomy in uterine cervical cancer showed excellent OS in this study. Positive or close PRM after radical hysterectomy in uterine cervical cancer correlates with poor prognosis even with CCRT. Therefore, additional treatments to improve local control such as radiation boosting need to be considered.
Subject(s)
Humans , Adenocarcinoma , Brachytherapy , Chemoradiotherapy , Disease-Free Survival , Drug Therapy , Hysterectomy , Lymph Nodes , Multivariate Analysis , Neoplasm Metastasis , Pelvis , Prognosis , Radiotherapy , Recurrence , Retrospective Studies , Uterine Cervical NeoplasmsABSTRACT
PURPOSE: To analyze prognostic factors for locoregional recurrence (LRR), distant metastasis (DM), and overall survival (OS) in cervical cancer patients who underwent radical hysterectomy followed by postoperative radiotherapy (PORT) in a single institute. MATERIALS AND METHODS: Clinicopathologic data of 135 patients with clinical stage IA2 to IIA2 cervical cancer treated with PORT from 2001 to 2012 were reviewed, retrospectively. Postoperative parametrial resection margin (PRM) and vaginal resection margin (VRM) were investigated separately. The median treatment dosage of external beam radiotherapy (EBRT) to the whole pelvis was 50.4 Gy in 1.8 Gy/fraction. High-dose-rate vaginal brachytherapy after EBRT was given to patients with positive or close VRMs. Concurrent platinum-based chemoradiotherapy (CCRT) was administered to 73 patients with positive resection margin, lymph node (LN) metastasis, or direct extension of parametrium. Kaplan-Meier method and log-rank test were used for analyzing LRR, DM, and OS; Cox regression was applied to analyze prognostic factors. RESULTS: The 5-year disease-free survival was 79% and 5-year OS was 91%. In univariate analysis, positive or close PRM, LN metastasis, direct extension of parametrium, lymphovascular invasion, histology of adenocarcinoma, and chemotherapy were related with more DM and poor OS. In multivariate analysis, PRM and LN metastasis remained independent prognostic factors for OS. CONCLUSION: PORT after radical hysterectomy in uterine cervical cancer showed excellent OS in this study. Positive or close PRM after radical hysterectomy in uterine cervical cancer correlates with poor prognosis even with CCRT. Therefore, additional treatments to improve local control such as radiation boosting need to be considered.