Machine Learning-Based Sample Misidentification Error Detection in Clinical Laboratory Tests: A Retrospective Multicenter Study.

BACKGROUND: In clinical laboratories, the precision and sensitivity of autoverification technologies are crucial for ensuring reliable diagnostics. Conventional methods have limited sensitivity and applicability, making error detection challenging and reducing laboratory efficiency. This study introduces a machine learning (ML)-based autoverification technology to enhance tumor marker test error detection. METHODS: The effectiveness of various ML models was evaluated by analyzing a large data set of 397 751 for model training and internal validation and 215 339 for external validation. Sample misidentification was simulated by random shuffling error-free test results with a 1% error rate to achieve a real-world approximation. The ML models were developed with Bayesian optimization for tuning. Model validation was performed internally at the primary institution and externally at other institutions, comparing the ML models' performance with conventional delta check methods. RESULTS: Deep neural networks and extreme gradient boosting achieved an area under the receiver operating characteristic curve of 0.834 to 0.903, outperforming that of conventional methods (0.705 to 0.816). External validation by 3 independent laboratories showed that the balanced accuracy of the ML model ranged from 0.760 to 0.836, outperforming the balanced accuracy of 0.670 to 0.773 of the conventional models. CONCLUSIONS: This study addresses limitations regarding the sensitivity of current delta check methods for detection of sample misidentification errors and provides versatile models that mitigate the operational challenges faced by smaller laboratories. Our findings offer a pathway toward more efficient and reliable clinical laboratory testing.

Temporal Trends in Syphilis Incidence among Men with HIV in Busan, Korea, 2005-2022: A Retrospective Cohort Study.

We aimed to assess the temporal trends of incident syphilis and its associated risk factors among men with HIV (Human Immunodeficiency Virus) in Korea during the COVID-19 pandemic. We conducted a retrospective cohort study of men with HIV attending an HIV clinic in Korea between 2005 and 2022. Of 767 men with HIV, 499 were included and contributed 3220 person-years (PY) of the observation period. Eighty-two patients were diagnosed with incident syphilis, with an overall incidence of 2.55/100 PY (95% confidence interval [CI] 20.56-31.53). The incidence of syphilis per 100 PY gradually decreased from 2.43 (0.79-7.42) in 2005-2007 to 1.85 (1.08-3.17) in 2014-2016; however, it increased to 3.0 (1.99-4.53) in 2017-2019, and further to 3.33 (2.26-4.89) in 2020-2022. A multivariate analysis identified young age (≤30 years versus >50, adjusted HR 6.27, 95% CI 2.38-16.56, p < 0.001), treponemal test positive at baseline (2.33, 1.48-3.67, p < 0.001), men who have sex with men (2.36, 1.34-4.16, p = 0.003), and history of incarceration (2.62, 1.21-5.67, p = 0.015) as risk factors for incident syphilis. Recently, syphilis incidence in men with HIV has increased in Korea, especially in young patients and at-risk groups, highlighting the need for enhanced regular screening and targeted behavioral interventions among these populations.

The novel HLA-A*30:01:25 allele found in a Korean hematopoietic stem cell recipient.

The novel HLA-A*30:01:25 allele differs from HLA-A*30:01:01:01 by a single nucleotide substitution located within exon 2, codon 1.

Current Status of Flow Cytometric Immunophenotyping of Hematolymphoid Neoplasms in Korea.

Background: Flow cytometric immunophenotyping of hematolymphoid neoplasms (FCI-HLN) is essential for diagnosis, classification, and minimal residual disease (MRD) monitoring. FCI-HLN is typically performed using in-house protocols, raising the need for standardization. Therefore, we surveyed the current status of FCI-HLN in Korea to obtain fundamental data for quality improvement and standardization. Methods: Eight university hospitals actively conducting FCI-HLN participated in our survey. We analyzed responses to a questionnaire that included inquiries regarding test items, reagent antibodies (RAs), fluorophores, sample amounts (SAs), reagent antibody amounts (RAAs), acquisition cell number (ACN), isotype control (IC) usage, positive/negative criteria, and reporting. Results: Most hospitals used acute HLN, chronic HLN, plasma cell neoplasm (PCN), and MRD panels. The numbers of RAs were heterogeneous, with a maximum of 32, 26, 12, 14, and 10 antibodies used for acute HLN, chronic HLN, PCN, ALL-MRD, and multiple myeloma-MRD, respectively. The number of fluorophores ranged from 4 to 10. RAs, SAs, RAAs, and ACN were diverse. Most hospitals used a positive criterion of 20%, whereas one used 10% for acute and chronic HLN panels. Five hospitals used ICs for the negative criterion. Positive/negative assignments, percentages, and general opinions were commonly reported. In MRD reporting, the limit of detection and lower limit of quantification were included. Conclusions: This is the first comprehensive study on the current status of FCI-HLN in Korea, confirming the high heterogeneity and complexity of FCI-HLN practices. Standardization of FCI-HLN is urgently needed. The findings provide a reference for establishing standard FCI-HLN guidelines.

Machine learning-based delta check method for detecting misidentification errors in tumor marker tests.

OBJECTIVES: Misidentification errors in tumor marker tests can lead to serious diagnostic and treatment errors. This study aims to develop a method for detecting these errors using a machine learning (ML)-based delta check approach, overcoming limitations of conventional methods. METHODS: We analyzed five tumor marker test results: alpha-fetoprotein (AFP), cancer antigen 19-9 (CA19-9), cancer antigen 125 (CA125), carcinoembryonic antigen (CEA), and prostate-specific antigen (PSA). A total of 246,261 records were used in the analysis. Of these, 179,929 records were used for model training and 66,332 records for performance evaluation. We developed a misidentification error detection model based on the random forest (RF) and deep neural network (DNN) methods. We performed an in silico simulation with 1 % random sample shuffling. The performance of the developed models was evaluated and compared to conventional delta check methods such as delta percent change (DPC), absolute DPC (absDPC), and reference change values (RCV). RESULTS: The DNN model outperformed the RF, DPC, absDPC, and RCV methods in detecting sample misidentification errors. It achieved balanced accuracies of 0.828, 0.842, 0.792, 0.818, and 0.833 for AFP, CA19-9, CA125, CEA, and PSA, respectively. Although the RF method performed better than DPC and absDPC, it showed similar or lower performance compared to RCV. CONCLUSIONS: Our research results demonstrate that an ML-based delta check method can more effectively detect sample misidentification errors compared to conventional delta check methods. In particular, the DNN model demonstrated superior and stable detection performance compared to the RF, DPC, absDPC, and RCV methods.

Initial Tumor Necrosis Factor-Alpha and Endothelial Activation Are Associated with Hemorrhagic Complications during Extracorporeal Membrane Oxygenation.

Studies on inflammatory markers, endothelial activation, and bleeding during extracorporeal membrane oxygenation (ECMO) are lacking. Blood samples were prospectively collected after ECMO initiation from 150 adult patients who underwent ECMO for respiratory failure between 2018 and 2021. After excluding patients who died early (within 48 h), 132 patients were finally included. Their tumor necrosis factor-alpha (TNF-α), tissue factor (TF), soluble thrombomodulin (sTM), and E-selectin levels were measured. A Cox proportional hazards regression model was used to estimate the hazard ratio for hemorrhagic complications during ECMO. The 132 patients were divided into hemorrhagic (n = 23, H group) and non-complication (n = 109, N group) groups. The sequential organ failure assessment score, hemoglobin level, and ECMO type were included as covariates in all Cox models to exclude the effects of clinical factors. After adjusting for these factors, initial TNF-α, TF, sTM, E-selectin, and activated protein C levels were significantly associated with hemorrhagic complications (all p < 0.001). TNF-α, TF, and E-selectin better predicted hemorrhagic complications than the model that included only the aforementioned clinical factors (clinical factors only (area under the curve [AUC]: 0.804), reference; TNF-α (AUC: 0.914); TF (AUC: 0.915); E-selectin (AUC: 0.869)). Conclusions: TNF-α levels were significantly predictive of hemorrhagic complications during ECMO.

Practical delta check limits for tumour markers in different clinical settings.

OBJECTIVES: Few studies have reported on delta checks for tumour markers, even though these markers are often evaluated serially. Therefore, this study aimed to establish a practical delta check limit in different clinical settings for five tumour markers: alpha-fetoprotein, cancer antigen 19-9, cancer antigen 125, carcinoembryonic antigen, and prostate-specific antigen. METHODS: Pairs of patients' results (current and previous) for five tumour markers between 2020 and 2021 were retrospectively collected from three university hospitals. The data were classified into three subgroups, namely: health check-up recipient (subgroup H), outpatient (subgroup O), and inpatient (subgroup I) clinics. The check limits of delta percent change (DPC), absolute DPC (absDPC), and reference change value (RCV) for each test were determined using the development set (the first 18 months, n=179,929) and then validated and simulated by applying the validation set (the last 6 months, n=66,332). RESULTS: The check limits of DPC and absDPC for most tests varied significantly among the subgroups. Likewise, the proportions of samples requiring further evaluation, calculated by excluding samples with both current and previous results within the reference intervals, were 0.2-2.9% (lower limit of DPC), 0.2-2.7% (upper limit of DPC), 0.3-5.6% (absDPC), and 0.8-35.3% (RCV99.9%). Furthermore, high negative predictive values >0.99 were observed in all subgroups in the in silico simulation. CONCLUSIONS: Using real-world data, we found that DPC was the most appropriate delta-check method for tumour markers. Moreover, Delta-check limits for tumour markers should be applied based on clinical settings.

The Discrepancy between Preoperative Tumor Markers and Imaging Outcomes in Predicting Ovarian Malignancy.

Preoperative tumor markers and imaging often differ in predicting whether an ovarian tumor is malignant. Therefore, we evaluated the correlation between the predictive values of imaging and tumor markers for diagnosing ovarian tumors, especially when there were discrepancies between the two. We enrolled 1047 patients with ovarian tumors. The predictive values and concordance rates between the preoperative risk of ovarian malignancy algorithm (ROMA) and imaging, including CT and MRI, were evaluated. Diagnoses of 561 CT (77.9%) and 322 MRI group (69.2%) participants were consistent with the ROMA. Among them, 96.4% of the CT (541/561) and 92.5% of the MRI (298/322) group predicted an accurate diagnosis. In contrast, 67.3% (101/150) of CT and 75.2% (100/133) of MRI cases accurately predicted the diagnosis in cases with discrepancies between ROMA and CT or MRI; a total of 32% (48/150) of the CT and 25.5% (34/133) of the MRI group showed an accurate ROMA diagnosis in cases with discrepancies between ROMA and imaging. In the event of a discrepancy between ROMA and imaging when ovarian tumor malignancy prediction, the question is which method should take precedence. This study demonstrates that MRI has the greatest diagnostic accuracy, followed by CT and ROMA. It is also important to understand underlying diseases and benign conditions and rare histopathologies of malignant tumors.

Sample collection for pre-transfusion crossmatching: Benefits of using an electronic identification system.

BACKGROUND: Transfusion of ABO blood group-mismatched blood or administration to the wrong recipient may result in fatal adverse events. To prevent these types of errors, various strategies have been employed. Recently, we developed a novel sample collection workflow for the pre-transfusion crossmatching test and patient recognition. This study aimed to analyse the usage of the new workflow and improvements in outcomes. METHODS: We analysed the number of crossmatching and wrong-patient errors among the blood transfusion cases during 3 years of data collection (from August 2018 to July 2021). From May 2021 to July 2021, the new workflow was implemented. Outcomes were calculated according to the department type, patient age and processing time. The sample processing time was defined as the time from placing the order to lab arrival. RESULTS: The new workflow utilisation increased from 50.7% to 80.3% and wrong-patient errors decreased annually. The new workflow was used for more adults (3001/3680 samples, 81.5%) than paediatric cases (345/522 samples, 65.5%; p < 0.001) and in general wards than in the emergency room or intensive care unit. The sample processing time differed according to ward type and timing of the request (day: 28.80, 2.43-3889.43 min, night: 3.36, 2.72-1671.47 min; p < 0.001). CONCLUSION: Wrong-patient errors were reduced without increasing sample-processing time after introducing the new workflow which included using an electronic identification system. The time needed for the blood processing differed according to the ward type, patient age, and timing of the request. Patient safety can be promoted by managing these factors and using an electronic identification system.

Effects of Different Centrifugation Protocols on the Detection of EGFR Mutations in Plasma Cell-Free DNA.

OBJECTIVES: Various preanalytical factors, including the collection tube, storage conditions, and centrifugation, affect the detection results of plasma cell-free DNA (cfDNA). We compared the effect of different centrifugation protocols on the detection of EGFR mutations in cfDNA. METHODS: We analyzed 117 plasma specimens from 110 patients with non-small cell lung cancer using the cobas EGFR Mutation Test v2 (Roche Diagnostics). We compared the identified EGFR mutations and semiquantitative index values from the 1- and 2-step centrifugation groups and confirmed the clinical impact of differences in the results after further high-speed centrifugation. RESULTS: We detected EGFR mutations in 44 (37.6%) and 47 (40.2%) samples that were centrifuged once and twice, respectively; the 2 groups showed an 89.7% (105/117) concordance and a strong correlation in their semiquantitative index values (r = 0.929). Among the 12 inconsistent result pairs, 9 samples of 2-step centrifugation (75%) were consistent with the results of a recent tissue biopsy. CONCLUSIONS: Additional high-speed centrifugation has been shown to increase the sensitivity of EGFR mutation detection in a commercial in vitro diagnostic real-time polymerase chain reaction device and is an optimal preanalytical factor for detecting low-allele frequency gene mutations using low concentrations of cfDNA.

Prognostic Impact of Peripheral Blood T-Cell Subsets at the Time of Diagnosis on Survival in Patients with Diffuse Large B-Cell Lymphoma.

INTRODUCTION: The effects of lymphocyte subtypes, including helper (Th), natural killer (NK), and regulatory (Treg) cells, and other T-cell subtypes on treatment outcomes in diffuse large B-cell lymphoma (DLBCL) patients are not clearly established. METHODS: Among 151 consecutive patients diagnosed with DLBCL, we collected peripheral blood samples at diagnosis from 91 patients who received at least 1 cycle of R-CHOP (cyclophosphamide, doxorubicin, vincristine, and prednisone plus rituximab) chemotherapy and analyzed lymphocyte subsets by flow cytometry. RESULTS: DLBCL patients had a higher proportion of CD4+CD25+ Treg (p < 0.001) and lower absolute lymphocyte count than those of healthy controls. Lymphopenia at diagnosis was associated with advanced-stage disease (p = 0.001), a high-intermediate/high-risk International Prognostic Index (IPI) (p < 0.001), and older age (p = 0.060). High-intermediate/high-risk IPI, high proportion of CD3+CD4+ Th cells, and extranodal site ≥2 correlated with unfavorable prognostic factors for survival. High proportion of Th cells was associated with fewer cytotoxic T cells and NK cells at the time of diagnosis. CONCLUSION: This study showed an association between circulating lymphocyte subsets including Th cells, Tregs, and NK cells and clinical outcomes in DLBCL; however, further confirmation is needed via prospective trials.

A novel HLA-A*24:460 allele identified in a hematopoietic stem cell transplantation donor and recipient.

HLA-A*24:460 differs from HLA-A*24:02 by one nucleotide in codon 285.

Clinical and Genetic Characteristics of BRCA1/2 Mutation in Korean Ovarian Cancer Patients: A Multicenter Study and Literature Review.

PURPOSE: We investigated the clinical relevance and spectrum of BRCA1/2 mutations in Korean ovarian cancer (KoOC) patients. MATERIALS AND METHODS: Two hundred seventy-nine KoOC patients were enrolled from three university hospitals between 2012 and 2017. Their peripheral blood samples were obtained for BRCA1/2 mutation analysis by direct sequencing. Clinicopathological characteristics were retrospectively reviewed, and spectrum analyses of BRCA1/2 mutation were assessed by systematic literature review. RESULTS: Frequency of BRCA1/2 mutations was 16.5% in KoOC patients. BRCA1/2 mutations were significantly associated with family history of breast/ovarian cancer (p<0.001), serous histology (p=0.044), and advanced International Federation of Gynecology and Obstetrics (FIGO) stage (III/IV, p=0.018) but not with early age-of-onset (age < 50, p=0.729). Literature review of BRCA1/2 mutations in KoOC patients found 111 (55 distinct) mutations with high proportion of Korean-specific mutations (24/55, 43.6%). Comparing the spectrum of BRCA1/2 mutation between KoOC and Korean breast cancer (KoBC) patients, the ratio of BRCA1-to-BRCA2 mutations was different, with BRCA1 (78.4%) being predominant in KoOC and BRCA2 in KoBC (59.2%). The most common mutation also differed between the two (c.3627insA of BRCA1 in KoOC and c.7480C>T of BRCA2 in KoBC). CONCLUSION: The clinical relevance of BRCA1/2 mutations in KoOC patients was confirmed but that of early age-of-onset was not. Possible inconsistency in the ratio of BRCA1-to-BRCA2 mutations and the most common mutation between KoOC and KoBC may probably suggest presence of mutation sequence-associated penetrance tendency in hereditary Korean breast and ovarian cancer. These data may provide insights for optimal genetic counseling and prophylactic treatment for at-risk relatives of KoOC patients.

Comparison of Cerebrospinal Fluid Cytokine Levels in Children of Enteroviral Meningitis With Versus Without Pleocytosis.

In viral meningitis, proinflammatory cytokines were detected at higher levels in the cerebrospinal fluid (CSF) and might play an important role in the inflammatory process. Our goal was to compare the cytokine profiles in the CSF of children of enteroviral meningitis (EVM) with versus without CSF pleocytosis. In total, 158 patients were enrolled in this prospective cohort study and were classified as EVM (group-A, n = 101), nonenteroviral aseptic meningitis (group-B, n = 27), and control (group-C, n = 30) groups. Of the 101 children with EVM, 71 had CSF pleocytosis (group-A1) and 30 had CSF nonpleocytosis (group-A2). Fifteen cytokines/chemokines in the CSF were measured simultaneously by immunoassay. Significant differences were found in interleukin (IL)-2, IL-6, and IL-8 levels in the CSF across the 3 groups, with the highest levels in group-A, followed by group-B and group-C. The levels of IL-1ß, IL-2, IL-6, IL8, IL-10, interferon-γ, and tumor necrosis factor-α were significantly higher in the CSF of group-A1 than in that of group-A2. Group-A2 was significantly younger than group-A1 (3.4 ± 2.8 years versus 5.5 ± 3.2 years, P = 0.016). Significant differences between CSF pleocytosis and nonpleocytosis in EVM appear to be associated with distinct levels of CSF cytokines.

Increasing use of therapeutic apheresis as a liver-saving modality.

INTRODUCTION: Therapeutic plasma exchange (TPE) is used for temporary support of liver function in patients presenting with early graft dysfunction after liver transplantation (LT) or liver surgery. We analyzed the effect of therapeutic apheresis on patients with liver disease. METHODS: Between January 2011 and August 2016, 93 apheresis procedures were performed for 26 patients at our institution. Anti-ABO isoagglutination immunoglobulin (Ig) M titer was checked using a type A and type B 3% red blood cell (RBC) suspension in saline with two-fold serial dilutions of patient serum. Anti-ABO isoagglutination IgG titer was checked by a type A and B 0.8% RBC suspension using a low-ionic strength/Coombs card. RESULTS: ABO-incompatible (ABOi) LT was the most common (n=10, 38.5%) indication for apheresis; early graft dysfunction after LT (n=8, 30.7%) was the second most common. Median initial IgM and IgG anti-ABO titers for ABOi LT recipients were 1:16 (range, 1:8-1:128) and 1:48 (range, 1:8-1:2048). We performed preoperative TPE in 10 recipients (median number of sessions, 1.5; range, 1-11). Among patients with early graft dysfunction, those who underwent living donor LT had better survival (4/4; 100%) than those who underwent nonliving donor LT (0/3; 0%). Patients who underwent living donor LT first and then additional LT also survived after three TPE sessions. CONCLUSION: Therapeutic apheresis is associated with a good survival rate and is essential for liver support in patients with early graft dysfunction after LT or posthepatectomy liver failure and during preparation for ABOi LT.

The risk factors for herpes zoster in bortezomib treatment in patients with multiple myeloma.

BACKGROUND: Bortezomib has significant activity in treating multiple myeloma (MM). The risk of herpes zoster (HZ) has been reported to increase significantly with bortezomib treatment, but the predisposing factors for HZ are not clear. This study is a retrospective analysis of the relevant risk factors for HZ in Korean MM patients treated with bortezomib. METHODS: Sixty-six patients with refractory or relapsed MM who underwent chemotherapy with bortezomib were included in the study. Prophylactic antiviral drugs were not used for treatment. The following parameters were reviewed: age, gender, stage and type of MM, extent of previous treatment, history of HZ, duration from the time of diagnosis to the time of bortezomib treatment initiation, and absolute lymphocyte counts (ALC) at the time of bortezomib treatment initiation. RESULTS: The incidence of HZ was 16.7%. There were no intergroup differences between the HZ-positive and the HZ-negative groups with regard to a history of HZ, number of previous treatments, and exposure to steroids before bortezomib treatment. The median duration from the time of MM diagnosis to the time of bortezomib treatment initiation in the HZ-positive group was significantly shorter than that in the HZ-negative group. The median ALC at the time of bortezomib initiation in the HZ-positive group was significantly lower than that in the HZ-negative group. CONCLUSION: Bortezomib itself might act as a risk factor for HZ by inhibiting cell-mediated immunity, and patients with low ALC at the time of bortezomib treatment initiation were at greater risk of HZ during bortezomib treatment.