Low Positive Predictive Value Research Articles

Clinical performance of Bladder EpiCheck™ versus voided urine cytology for detecting recurrence of nonmuscle invasive bladder cancer: Systematic review and meta-analysis

BackgroundNonmuscle invasive bladder cancer (NMIBC) has a favorable prognosis but has high propensity for recurrence. Recent development in one of the urinary biomarker tests, Bladder EpiCheck™, offers a noninvasive and accurate method to detect NMIBC recurrence. In this study, we aimed to compare the diagnostic performance of Bladder EpiCheck™ with urine cytology to detect NMIBC recurrence. MethodsWe performed a systematic review search through PubMed, Embase, Cochrane Central Register of Controlled Trials, Web of Science, and Scopus from inception to July 2023. Diagnostic accuracy was defined by sensitivity, negative predictive value (NPV), specificity, and positive predictive value (PPV). ResultsA total of 6 studies involving 1588 patients were included. Bladder EpiCheck™ has a sensitivity and specificity of 0.81 (95% CI: 0.63–0.91; I2: 43%) and 0.87 (95% CI: 0.83–0.91; I2: 20%), respectively. On the other hand, urine cytology has a sensitivity and specificity of 0.63 (95% CI: 0.29–0.87; I2: 61%) and 0.97 (95% CI: 0.78–1.00; I2: 79%), respectively. EpiCheck™ has a higher NPV (0.94 (95% CI: 0.87–0.97) vs. 0.84 (95% CI: 0.80–0.87) though a lower PPV (0.62 (95% CI: 0.45–0.76) vs. 0.87 (95% CI: 0.56–0.97) than urine cytology. In our subgroup analysis, the sensitivity of Bladder EpiCheck™ for detecting high-grade tumors improved to 0.90 (95% CI: 0.83–0.94) while that for urine cytology improved to 0.72 (95% CI: 0.50–0.87). ConclusionBladder EpiCheck™ has a high sensitivity and NPV for detecting recurrence among patients with NMIBC.

Pediatric Rash Illness Outbreak with Initial Positive Measles Immunoglobulin M Antibody Test Results - American Samoa, March-July 2023.

On April 24, 2023, the American Samoa Department of Health (ASDoH) declared a public health emergency amid concern about a possible measles outbreak given low 2-dose vaccination coverage at the time. ASDoH had received two positive measles immunoglobulin (Ig) M test results after Flag Day festivities 1 week earlier from vaccinated children. ASDoH performed active case finding, took actions to mitigate transmission, and requested technical assistance from CDC. ASDoH implemented a vaccination campaign to improve suboptimal coverage. Confirmatory molecular testing of specimens from these initial persons under investigation (PUIs) was not possible, but subsequent testing of specimens from additional PUIs by Hawaii State Laboratories Division and CDC ruled out measles. In settings with low measles prevalence, measles antibody testing results have low positive predictive value and can lead to difficulties with interpreting results. Testing for additional pathogens revealed a variety of viruses known to cause common childhood viral exanthems. Both molecular and serologic testing should be performed for all suspected measles cases. To decrease the probability of false-positive IgM results, testing should be reserved for cases that meet the Council of State and Territorial Epidemiologists measles case definition, especially those in persons with no evidence of immunity and with a history of recent international travel. In addition, maintaining high measles vaccination coverage can prevent future outbreaks.

Assessing humoral immunity in daily practice: A retrospective study in a pediatric tertiary center

The evaluation of humoral immunity is usually performed through the assessment of serum immunoglobulin levels, vaccine titer responses, and B-cell enumeration and phenotyping. We performed a retrospective study assessing humoral immunity in 469 pediatric patients referred at the Sainte-Justine University Hospital Center. Almost half of the patients had at least one abnormal humoral immunological parameter at their evaluation, with low vaccine response titer to protein antigen being the most frequent. Fifteen patients (3.2 %) had a proven monogenic IEI, and 21 patients (4.5 %) required Ig replacement. Besides the infectious burden, hypoIgG remains the only parameter associated with Ig replacement therapy after the age of 6 years. Low antibody titers against conjugate vaccines had low sensitivity and positive predictive values for starting Ig replacement. Our study highlights the challenge of evaluating the humoral function in the pediatric population with suspected IEI with significant age and sex-dependent variability between parameters.

Flemish consensus statement on the prevention, diagnosis and treatment of urinary tract infections in older nursing home residents.

Urinary tract infections (UTIs) are one of the most commonly reported infections in Belgian nursing home residents. In older adults, UTI diagnosis and management is complex, often leading to over-diagnosis and irrational antimicrobial use, stressing the need for a guideline approach. A consensus statement on the prevention, diagnosis and treatment of UTIs in older adults residing in nursing homes was developed in a collaborative effort between the Flemish Hospital Outbreak Support Teams, the Flemish Agency for Care and Health, the Association of the Flemish Coordinating and Advising General Practitioners, the Belgian Association of Urology, the Belgian Society for Gerontology and Geriatrics and PhD researchers based on a combination of clinical expertise, (inter)national guidelines and peer-reviewed studies. Optimizing fluid intake, appropriate toilet behaviour and posture, mobilization and local estrogen therapy in women are of proven value in UTI prevention, whereas the use of cranberry and probiotics is not to be advocated. The importance of avoiding bladder catheterization is stressed. In older nursing home residents, the diagnosis of UTIs remains challenging, mostly due to atypical systemic symptoms. A consensus diagnostic algorithm for UTI among residents with and without a urinary catheter was developed, including the presence of suggestive clinical symptoms and a positive urine culture. Urine dipsticks have a high negative but a low positive predictive value. C-reactive protein point-of-care testing is not recommended. Asymptomatic bacteriuria should not be screened for, in order to avoid unnecessary triggers for treatment. In cystitis, nitrofurantoin is the primary choice for treatment, with fosfomycin as an alternative; in prostatitis and uncomplicated pyelonephritis a fluoroquinolone is the advocated empirical antimicrobial.

Machine Learning Models for Predicting Dysphonia Following Anterior Cervical Discectomy and Fusion – A Swedish Registry Study

BACKGROUNDDysphonia is one of the more common complications following anterior cervical discectomy and fusion (ACDF). ACDF is the gold standard for treating degenerative cervical spine disorders, and identifying high-risk patients is therefore crucial. PURPOSEThis study aimed to evaluate different machine learning models to predict persistent dysphonia after ACDF. STUDY DESIGNA retrospective review of the nationwide Swedish spine registry (Swespine) PATIENT SAMPLEAll adults in the Swespine registry who underwent elective ACDF between 2006 and 2020. OUTCOME MEASURESThe primary outcome was self-reported dysphonia lasting at least one month after surgery. Predictive performance was assessed using discrimination and calibration metrics. METHODS: Patients with missing dysphonia data at the one-year follow-up were excluded. Data preprocessing involved one-hot encoding categorical variables, scaling continuous variables, and imputing missing values. Four machine learning models (logistic regression, random forest (RF), gradient boosting, K-nearest neighbor) were employed. The models were trained and tested using an 80:20 data split and 5-fold cross-validation, with performance metrics guiding the selection of the best model for predicting persistent dysphonia. RESULTSIn total, 2,708 were included in the study. Twelve key predictors were identified. Four machine learning models were tested, with the RF model achieving the best performance (AUC = 0.794). The most significant predictors across models included preoperative NDI, EQ5Dindex, preoperative neurology, number of operated levels, and use of a fusion cage. The RF model, chosen for its superior performance, showed high sensitivity and consistent accuracy, but a low specificity and positive predictive value. CONCLUSIONSIn this study, machine learning models were employed to identify predictors of persistent dysphonia following ACDF. Among the models tested, the RF classifier demonstrated superior performance, with an AUC value of 0.790. The RF model identified NDI, EQ5Dindex, and number of fused vertebrae as key variables. These findings underscore the potential of machine learning models in identifying patients at increased risk for dysphonia persisting for more than one month after surgery.

Assessment of Nonmass Lesions Detected with Screening Breast US Based on Mammographic Findings.

Background Breast nonmass lesions (NMLs) are observed at screening and diagnostic US. However, knowledge is limited on imaging features of NMLs at screening US. Purpose To identify features of NMLs at screening US that are suspicious for malignancy based on mammographic findings. Materials and Methods This retrospective, multicenter study included asymptomatic women who underwent screening US between January 2012 and December 2019. Eligible women had NMLs at US, concurrent screening mammography, and record of a final diagnosis. Logistic regression analyses were used to identify factors associated with malignancy. The diagnostic performance of each sonographic feature according to mammographic findings was calculated. A reader study was performed to assess interreader agreement for sonographic features. Results Among 993 NMLs in 993 patients (mean age, 50 years ± 9 [SD]), 914 (92.0%) were benign and 79 (8.0%) were malignant. Mean size was larger for malignant NMLs than for benign NMLs (2.6 cm ± 1.1 vs 1.9 cm ± 0.8; P < .001). In multivariable analysis, associated calcifications (odds ratio [OR], 21.6 [95% CI: 8.0, 58.2]; P < .001), posterior shadowing (OR, 6.9 [95% CI: 2.6, 18.4]; P < .001), segmental distribution (OR, 6.2 [95% CI: 2.7, 14.4]; P < .001), mixed echogenicity (OR, 5.0 [95% CI: 1.8, 14.0]; P < .001), and size (OR, 1.5 [95% CI: 1.1, 2.1]; P = .01) at US were associated with malignancy. Associated calcifications, posterior shadowing, segmental distribution, and mixed echogenicity showed positive predictive values (PPVs) of 44%, 22%, 22.9%, and 16.6%, respectively. Having a negative mammogram was associated with a lower malignancy rate (2.8% vs 28.8%) and lower PPVs for sonographic features (0.7%-10.4% vs 24%-55%) than having a positive mammogram. Interreader agreement for sonographic features was good to excellent (Fleiss κ 95% CI lower bound range, 0.63-0.81). Conclusion Calcifications, posterior shadowing, segmental distribution, and mixed echogenicity associated with NMLs can be considered suspicious features for malignancy at screening US. As malignancy rates and PPVs differ according to mammographic abnormalities, combined assessment is mandatory. © RSNA, 2024 Supplemental material is available for this article. See also the editorial by Grimm in this issue.

Diagnostic validity and solute-corrected prevalence for hyponatremia and hypernatremia among 1,813,356 admissions

Abstract Background and hypothesis We aimed to evaluate the diagnostic validity of the International Classification of Diseases, 10th Revision (ICD-10) codes for hyponatremia and hypernatremia, using a database containing laboratory data. We also aimed to clarify whether corrections for blood glucose, triglyceride, and total protein may affect the prevalence and the diagnostic validity. Methods We retrospectively identified admissions with laboratory values using a Japanese hospital-based database. We calculated the sensitivity, specificity, and positive/negative predictive values of recorded ICD-10-based diagnoses of hyponatremia (E87.1) and hypernatremia (E87.2), using serum sodium measurements during hospitalization (&amp;lt;135 mmol/L and &amp;gt;145 mmol/L, respectively) as the reference standard. We also performed analyses with corrections of sodium concentrations for blood glucose, triglyceride, and total protein. Results We identified 1,813,356 hospitalizations, including 419,470 hyponatremic and 132,563 hypernatremic cases based on laboratory measurements and 18,378 hyponatremic and 2,950 hypernatremic cases based on ICD-10 codes. The sensitivity, specificity, positive predictive value, and negative predictive value of the ICD-10 codes were 4.1%, 99.9%, 92.5%, and 77.6%, respectively, for hyponatremia and 2.2%, &amp;gt;99.9%, 96.5%, and 92.8%, respectively, for hypernatremia. Corrections for blood glucose, triglyceride, and total protein did not largely alter diagnostic values, though prevalence changed especially after corrections for blood glucose and total protein. Conclusions The ICD-10 diagnostic codes showed low sensitivity, high specificity, and high positive predictive value for identifying hyponatremia and hypernatremia. Corrections for glucose or total protein did not affect diagnostic values but would be necessary for accurate prevalence calculation.

Intraoperative Hypotension Prediction: Current Methods, Controversies, and Research Outlook.

Intraoperative hypotension prediction has been increasingly emphasized due to its potential clinical value in reducing organ injury and the broad availability of large-scale patient datasets and powerful machine learning tools. Hypotension prediction methods can mitigate low blood pressure exposure time. However, they have yet to be convincingly demonstrated to improve objective outcomes; furthermore, they have recently become controversial. This review presents the current state of intraoperative hypotension prediction and makes recommendations on future research. We begin by overviewing the current hypotension prediction methods, which generally rely on the prevailing mean arterial pressure as one of the important input variables and typically show good sensitivity and specificity but low positive predictive value in forecasting near-term acute hypotensive events. We make specific suggestions on improving the definition of acute hypotensive events and evaluating hypotension prediction methods, along with general proposals on extending the methods to predict reduced blood flow and treatment effects. We present a start of a risk-benefit analysis of hypotension prediction methods in clinical practice. We conclude by coalescing this analysis with the current evidence to offer an outlook on prediction methods for intraoperative hypotension. A shift in research toward tailoring hypotension prediction methods to individual patients and pursuing methods to predict appropriate treatment in response to hypotension appear most promising to improve outcomes.

Utilizing Artificial Intelligence to Reconcile Indeterminate Lupus Anticoagulant Panel Results

Abstract Lupus anticoagulant (LA) is one laboratory criterion for diagnosing antiphospholipid antibody syndrome (APS), a condition associated with hypercoagulability, pregnancy complications, and death. LA diagnosis requires a complex panel, often performed at a specialized laboratory. Ruling out false positive or transient LAs requires ≥ two positive tests ≥ 12 weeks apart. Most LA tests are negative, with small subsets resulting positive or indeterminate. Indeterminates leave clinicians with limited understanding of appropriate clinical follow-up. Thus, the International Society on Thrombosis and Hemostasis discourages weak resulting nomenclature such as “indeterminate.” Our study used artificial intelligence to predict LA indeterminate results as positive or negative, identifying patients requiring further testing. This IRB approved study reviewed nearly 4000 LA tests performed from 2019-2023. Parameters included CBC, thromboelastography, renal/hepatic function assays, coagulation tests, antiphospholipid antibodies and LA tests. LA test components included screening tests (APTT, Dilute Russel Viper Venom Test (DRVVT) screen), mixing studies (1:1 APTT mix, DRVVT mix), phospholipid-dependent confirmation (DRVVT, hexagonal phase, platelet neutralization), and factor VIII. A positive LA test fulfilled 4 criteria: at least one positive screening test, mixing study and phospholipid-dependent assay and factor VIII inhibitor exclusion. An LA panel meeting 2-3 criteria resulted indeterminate. Studies meeting &amp;lt;2 criteria were negative. This analysis was performed using Machine Intelligence Learning Optimizer (MILO) platform which displays the performance of multiple models. MILO was trained with two classifications: positive and negative cases. Data was split into training (100 training positives; 100 negatives) and testing (primary and secondary validation) sets (100 positives; 1388 negatives). LA panels affected by drug interactions (9.4%) or containing incomplete datasets (21%) were excluded. Sensitivity, specificity, negative predictive value (NPV) and positive predictive value (PPV) determined model efficacy. 2903 negative (84%), 318 positive (9.1%), and 246 indeterminate (7.1%) LA results were included from 3467 patients. Only 84 of the 246 indeterminate patients (34%) were retested. Of these, 28 remained indeterminate, 43 became negative and 13 resulted positive. The best performing MILO models were the K-nearest neighbor (KNN) and logistic regression. The ability of these models to determine which indeterminate cases would become positive or negative was evaluated. KNN displayed the highest specificity (0.88) and PPV (0.57). All models had a high NPV and specificity due to substantive exposure to negative cases for the training and testing. Exclusion of some parameters did not improve the performance of the models. However, even the best model (various KNN models) had low sensitivity and PPV due to a small pool of positive training/testing cases and a much smaller group of indeterminate LA with subsequent positive results. Creating models accurately predicting indeterminate cases that may become positive enables appropriate clinical follow-up. Thus, improving model accuracy will require a larger sample size of positive indeterminates with follow-up.

Validation and correlation of high-sensitive troponin I and troponin T in the emergency department

BackgroundTroponin elevation is frequently observed in various scenarios in the Emergency Department (ED), yet there is a paucity of studies investigating simultaneously measured high-sensitivity cardiac troponin T (hs-cTnT) and troponin I (hs-cTnI) within a diverse cohort in a clinical setting.MethodsAll patients who underwent troponin testing at a single center were eligible for this study. Only patients with simultaneous samples with hs-cTnI (Siemens) and hs-cTnT (Roche) were included, regardless of chief complaint.ResultsAnalysis of 1987 samples from 1134 patients showed a significant correlation between hs-cTnT and hs-cTnI (r = 0.86, p < 0.01). Of these samples, 65% exceeded the upper reference limit (URL) for hs-cTnT, and 30% for hs-cTnI with 39% who exhibited elevated hs-cTnT levels alongside normal hs-cTnI levels. The area under the curve (AUC) for acute myocardial infarction (AMI) for the index visit was 0.80 (95% CI; 0.75–0.85) for hs-cTnT and 0.87 (95% CI; 0.83–0.91) for hs-cTnI. Sensitivity and specificity were 91% and 39% for hs-cTnT, and 80% and 80% for hs-cTnI. Positive predictive value (PPV) and negative predictive value (NPV) was 9.3% and 98.5% for hs-cTnT respectively, corresponding for hs-cTnI was 21.3% and 98.3% respectively. Hazard ratios for 1-year mortality were 1.52 (95% CI; 1.40–1.66) for hs-cTnT and 1.26 (95% CI; 1.18–1.34) for hs-cTnI.ConclusionElevated troponins above the URL were very common in this diverse cohort, particularly for hs-cTnT, which was twice as frequent compared to hs-cTnI, resulting in low specificity and PPV for AMI.

FibroScan-AST score for diagnosing fibrotic MASH: A systematic review and meta-analysis of diagnostic test accuracy studies.

Following the approval of the first agent for the management of metabolic dysfunction-associated steatohepatitis (MASH), identification of patients with fibrotic MASH (MASH with NAS≥4 and fibrosis stage≥2) is crucial. We assessed the performance of FibroScan-aspartate aminotransferase (AST) score (FAST) for ruling in/out fibrotic MASH. We searched Medline, Cochrane Library, Web of Science, Scopus, and gray literature sources up to January 11, 2024. Studies were eligible if they assessed the accuracy of FAST score for the detection of fibrotic MASH using biopsy as the reference standard at previously reported thresholds (FAST≥0.67 for ruling-in and ≤0.35 for ruling-out fibrotic MASH). We calculated pooled sensitivity and specificity estimates for FAST thresholds alongside 95% confidence intervals following bivariate random- effects models. We assessed the certainty of evidence using the Grading of Recommendations Assessment, Development and Evaluation framework. We included 16 studies with 8838 participants. A FAST score≥0.67 yielded a pooled specificity of 0.87 (0.82-0.90) while a FAST score≤0.35 yielded a summary sensitivity of 0.88 (0.83-0.91). At a prevalence of 30%, the positive predictive value for ruling-in fibrotic MASH was 60% while the negative predictive value for ruling-out the target condition was 91%. AST levels, cirrhosis prevalence, and number of pathologists reviewing biopsies were sources of heterogeneity among studies. The certainty of evidence was low to very low. FAST score can be used as a triage test for ruling out fibrotic MASH. Nevertheless, its low positive predictive value necessitates sequential testing for ruling-in fibrotic MASH.

Expert consensus on the clinical application strategy of NIPT2.0, a new-generation non-invasive prenatal screening technology

The new-generation non-invasive prenatal screening technology (NIPT2.0) is a new method successfully realized in recent years based on high-throughput sequencing to synchronously and accurately detect fetal chromosomal aneuploidies, microdeletion/microduplication syndromes and dominantly inherited monogenic disorders. NIPT2.0 can circumvent the shortcomings of previous non-invasive prenatal screening techniques (NIPT and NIPT Plus) including incapability to detect fetal monogenic disorders, insufficient accuracy of detection and low positive predictive values for certain chromosomal abnormalities (in particular trisomy 13, sex chromosomal abnormalities, and small-segment microdeletions and microduplication syndromes). How to apply NIPT2.0 reasonably and normatively to maximize its clinical value has become an issue which requires clarification. The Reproductive Health Branch of the Chinese Maternal and Child Health Care Association and the Genetic Diagnosis Branch of the Genetics Society of China have organized experts to fully discuss and jointly drafted this consensus, which has put forwards suggestions over the clinical application strategy for NIPT2.0, including the scope of application, target disease, pre-test consultation, clinical application pathway, post-test genetic counseling and intervention, quality control and limitations, for the reference by peers, with a view to standardize its application and provide better clinical service.

Artificial intelligence-enhanced detection of subclinical coronary artery disease in athletes: diagnostic performance and limitations.

This study evaluates the diagnostic performance of artificial intelligence (AI)-based coronary computed tomography angiography (CCTA) for detecting coronary artery disease (CAD) and assessing fractional flow reserve (FFR) in asymptomatic male marathon runners. We prospectively recruited 100 asymptomatic male marathon runners over the age of 45 for CAD screening. CCTA was analyzed using AI models (CorEx and Spimed-AI) on a local server. The models focused on detecting significant CAD (≥ 50% diameter stenosis, CAD-RADS 3, 4, or 5) and distinguishing hemodynamically significant stenosis (FFR ≤ 0.8) from non-significant stenosis (FFR > 0.8). Statistical analysis included sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy. The AI model demonstrated high sensitivity, with 91.2% for any CAD and 100% for significant CAD, and high NPV, with 92.7% for any CAD and 100% for significant CAD. The diagnostic accuracy was 73.4% for any CAD and 90.4% for significant CAD. However, the PPV was lower, particularly for significant CAD (25.0%), indicating a higher incidence of false positives. AI-enhanced CCTA is a valuable non-invasive tool for detecting CAD in asymptomatic, low-risk populations. The AI model exhibited high sensitivity and NPV, particularly for identifying significant stenosis, reinforcing its potential role in screening. However, limitations such as a lower PPV and overestimation of disease indicate that further refinement of AI algorithms is needed to improve specificity. Despite these challenges, AI-based CCTA offers significant promise when integrated with clinical expertise, enhancing diagnostic accuracy and guiding patient management in low-risk groups.

7771 Surveillance Role in Differentiated Thyroid Cancer Through Ultrasensitive Thyroglobulin IRMA with High Negative Prediction Value

Abstract Disclosure: J. Jang: None. H. Kim: None. S. Ha: None. K. Jung: None. G. Jung: None. S. Cho: None. D. Park: None. G. Cheon: None. Y. Park: None. TSH-stimulated serum thyroglobulin (s-Tg) is a well-established marker for monitoring differentiated thyroid cancer (DTC) following remnant ablation therapy. The 2nd-generation immunoradiometric assay for Tg (ultrasensitive Tg, uTg) was developed to detect Tg with higher sensitivity compared with the 1st generation (conventional Tg, cTg). This study aimed to compare efficacy of both methods, cTg and uTg, to determine whether unstimulated Tg (us-Tg) could predict s-Tg levels. Paired serum samples were collected from 274 DTC patients before (us-Tg) and after (s-Tg) TSH stimulation for 1st radioiodine therapy after surgery (1st Tg) or both 1st therapy and 2nd therapy/diagnostic test (1st and 2nd Tg, respectively). Both cTg (Dynotest Tg-plus, BRAHMS Gmbh, Germany, LOD 0.1 ng/mL and functional sensitivity (FS) 0.2 ng/mL) and uTg (RIAKEY, Shinjin Medics, South Korea; LOD 0.01 ng/mL and FS 0.06 ng/mL) were measured. For analysis, samples with inadequate TSH stimulation (&amp;lt;30 μIU/mL) or with anti-thyroglobulin antibody levels &amp;gt; 60 U/mL were excluded. Ultimately, the included samples consisted of 296 for us-Tg and 352 for s-Tg.. Among these, paired samples for us-Tg and s-Tg were available in 260 cases (189 for 1st Tg and 71 for 2nd Tg). Comparing all Tg values, cTg and uTg levels had a correlation (Intraclass correlation coefficient r=0.752, P &amp;lt; 0.001), with no significant difference between us-Tg (r=0.803, P &amp;lt; 0.001) and s-Tg (r= 0.752, P &amp;lt; 0.001). The optimal cut-off level for predicting s-cTg at 1 ng/mL was 0.12 ng/mL in uTg or 0.16 ng/mL in cTg tests using AUC-ROC curve analysis. Considering the FS for cTg, diagnostic performance was assessed with a cut-off level of 0.2 ng/mL for cTg and 0.16 ng/mL for uTg. For s-cTg positivity as ≥ 1 ng/mL, the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of uTg were 72.1%, 59.8%. 52.0%, and 78.0%, respectively, while for cTg, they were 33.3%, 95.3%, 78.4%, and 73.9%, respectively. Similar trends were observed for s-cTg ≥ 2 ng/mL (69.0%, 71.4%, 35.0%, and 91.2% for uTg, respectively; 49.0%, 93.8%, 64.9%, and 88.7% for cTg, respectively). In essence, uTg demonstrated higher sensitivity than cTg, while having lower specificity and PPV, with a similar NPV. Of the 71 patients with paired 2nd Tg levels, 7 (9.8%) had biochemical disease (BCD) by s-cTg (1 ng/mL), whereas only 1 (1.4%) showed BCD using us-cTg ≥ 0.2 ng/mL. Conversely, 40 (56.0%) were classified as BCD using us-uTg ≥ 0.16 ng/mL. During the 6.48 years of follow-up, 4 patients developed structural recurrence; all had positive us-Tg levels with uTg tests, while only 1 patient demonstrated positivity with cTg tests. Although, the ultrasensitive Tg is sensitive in predicting positive stimulated Tg and recurrence, the specificity is very low resulting high frequency of BCD cases without clinical significance. Therefore, the clinical utility of uTg appears to be limited. Presentation: 6/2/2024

B-320 Understanding CA 72-4: Exploring False Positives in Clinical Context

Abstract Background CA 72-4, initially identified in 1981 as an antigen reactive to murine antibodies, is a glycoprotein expressed on the cell surface of certain carcinoma cells. Its production is elevated in various cancers, including ovarian, pancreatic, colorectal, and gastric cancer. However, CA 72-4 serum concentration may rise in non-cancerous conditions, leading to potential misdiagnosis and a low positive predictive value (PPV), particularly in asymptomatic patients. Our aim in this study was to describe CA 72-4 false positive values and assess the pathological conditions and pharmacological treatments associated with this. We also intended to compare other tumour markers status in false and true positive results of CA 72-4. Methods A retrospective analysis form pseudo-anonymized database was conducted using data from 1290 laboratory analysis from different patients taking a blood test at our laboratory hospital from 2017 to 2019. CA 72-4 and other tumour markers were analysed by ECLIA (Roche Diagnostics Gmbh. Mannheim, Germany). Results Among 1290 CA 72-4 laboratory tests in different patients, 204 were considered positive because they were above the reference interval (&amp;lt;7 U/mL). Within this subgroup, 44 of the 204 were clearly classified as false positives, as patients were not diagnosed with any form of neoplastic process. 45% of cases were between 7-14 U/mL (group 1), 30% were between 14-28 U/mL (group 2), and finally 25% of patients were above 28 U/mL ranging from 35.1 to 240.4 U/mL (group 3). Interestingly, CA19-9, CA 125 and CEA were measured in 37 of the false positive cases and it was found that 59.5% had at least another tumour marker above the reference range. In 103 true positive cases it was found that 88% of patients had at least one tumour marker above the normal range. Patients exhibiting false positive results for CA 72-4 demonstrated a wide spectrum of pathological conditions, with pulmonary disease, inflammation, infection, and heart and kidney disease emerging as the most prevalent. Group 1 patients most common underlying pathologies were pulmonary disease in 25%, heart disease in 25% and either viral or bacterial infection in 60% of patients. Group 2 patients most common pathologies were heart disease in 42%, pulmonary disease in 33% and infection in 25% of patients. In group 3 patients, renal disease was present in 45% of patients, heart disease was found in 45% of patients and inflammation/sepsis was found in 36% of patients. When looking into pharmacological treatment, 86.3% of patients were taking at least one of previously described treatments increasing CA 72-4 in serum; proton pump inhibitors, corticosteroids, antiplatelet drugs and colchicine. Drug treatment proportion was very similar across the different groups. Conclusions CA72-4 false positive cases are sometimes in conjunction with other false positive tumour markers but not in the same proportion as true positives. In addition, most false positive patients can be explained by pharmacological treatments. This research may help in the understanding and identification of potential factors influencing tumor marker levels and thus contribute to the interpretation of results.

Predicting Fungemia in the ICU: Unveiling the Value of Weekly Fungal Surveillance and Yeast Colonisation Monitoring.

Fungemia remains a major threat in intensive care units (ICUs), with high mortality rates despite advances in diagnostics and treatment. Colonisation by yeasts is an independent risk factor for fungemia; however, its predictive utility requires further research. In this 8-year study, we analysed 38,017 samples from 3206 patients and 171 fungemia episodes as part of a weekly fungal surveillance programme. We evaluated species-specific colonisation patterns, the predictive value of the Colonisation Index (CI) and Corrected Colonisation Index (CCI), and candidemia risks associated with different yeast species and anatomical site colonisation. Our results showed that C. auris, N. glabratus, and C. parapsilosis colonisation increased with longer hospital stays (0.8% to 11.55%, 8.13% to 16.8%, and 1.93% to 5.14%, respectively). The CI and CCI had low discriminatory power (AUROC 67% and 66%). Colonisation by any yeast genera demonstrated high sensitivity (98.32%) and negative predictive value (NPV) (95.90%) but low specificity and positive predictive value (PPV) (23.90% and 6.64%). Tracheal and urine cultures had the highest PPV (15.64% and 12.91%), while inguinal cultures had the highest NPV (98.60%). C. auris (12.32%) and C. parapsilosis (5.5%) were associated with a higher fungemia risk (log-rank < 0.001). These findings support the use of weekly surveillance to better stratify the fungemia risk and optimise antifungal use in ICUs.

Identifying optimal serum 1,3-β-D-Glucan cut-off for diagnosing Pneumocystis Jirovecii Pneumonia in non-HIV patients in the intensive care unit

BackgroundSerum (1,3)-β-D-glucan (BDG) detection for diagnosis of Pneumocystis jirovecii pneumonia (PJP) in non-human immunodeficiency virus (HIV) immunocompromised patients lacks intensive care unit (ICU)-specific data. We aimed to assess its performance and determine the optimal cutoff for PJP in ICU population.MethodsThis retrospective study included critically ill non-HIV immunocompromised patients admitted to a medical ICU with suspected pneumonia, undergoing simultaneous microbiological testing for P. jirovecii on lower respiratory tract specimens and serum BDG. Confounders affecting BDG positivity were explored by multivariable logistic regression. Optimal cut-offs were derived from Youden’s index for the entire cohort and subgroups stratified by confounders. Diagnostic performance of serum BDG was estimated at different cutoffs.ResultsOf 400 patients included, 42% were diagnosed with PJP and 58.3% had positive serum BDG. Serum BDG’s area under the receiver operating characteristic curve was 0.90 (0.87–0.93). At manufacturer’s 150 pg/ml cut-off, serum BDG had high sensitivity and negative predictive value (94%), but low specificity and positive predictive value (67%). Confounders associated with a positive serum BDG in PJP diagnosis included IVIG infusion within 3 days (odds ratio [OR] 9.24; 95% confidence interval [CI] 4.09–20.88, p < 0.001), other invasive fungal infections (OR 4.46; 95% CI 2.10–9.49, p < 0.001) and gram-negative bacteremia (OR 29.02; 95% CI 9.03–93.23, p < 0.001). The application of optimal BDG cut-off values determined by Youden’s index (252 pg/ml, 390 pg/ml, and 202 pg/ml) specific for all patients and subgroups with or without confounders improved the specificity (79%, 74%, and 88%) and corresponding PPV (75%, 65%, and 85%), while maintaining reasonable sensitivity and NPV.ConclusionsTailoring serum BDG cutoff specific to PJP and incorporating consideration of confounders could enhance serum BDG’s diagnostic performance in the ICU settings.

Improving Detection of Hospital Adverse Events Using Machine Learning on Real-World Narrative EMR Data

ObjectiveAdministrative data often underrepresents hospital adverse events (AEs) due to limitations in International Classification of Diseases (ICD) coding. By leveraging electronic medical records (EMRs), we aim to mitigate these discrepancies and enhance the precision of healthcare surveillance and performance evaluations. To this end, we have developed a machine learning (ML)-based approach that utilizes EMR text data to detect common AEs. MethodsWe sampled adult admissions from four Calgary hospitals (2017 - 2022). Registered nurses assessed charts for 17 AEs, and the results were used as reference standard. We compared two AE detection methods: the standard ICD-based method following Canadian guidelines, and our ML algorithm applied to EMR narratives. Sensitivity, positive predictive value (PPV), negative predictive value (NPV), and specificity for both methods were calculated and compared against the reference standard. ResultsWe analyzed 9,566 patients, of whom 1,506 were identified with AEs. Of the 17 AEs, the sensitivity in ICD-coded data ranged from 0-37%, and in EMRs, it was between 75-100%. Both showed low PPV (0-50% ICD vs.1-34% EMR). ICD data had high specificity ranging from 99-100% and NPV (99%-100%), while EMRs had specificity between 68-94% and an NPV of 100%. Conclusion and ImplicationsML significantly enhances sensitivity for AE detection compared to ICD-10-CA coding, despite both methods experiencing low PPV due to imbalances in EMR data. This marked improvement in sensitivity highlights ML's potential to transform AE surveillance and reporting, promising significant advancements in patient safety and healthcare quality by enabling more accurate and comprehensive identification of AEs.

Validity of self-reported Helicobacter pylori eradication treatment from questionnaire and interview surveys of the JPHC-NEXT study: comparison with prescription history from insurance claims data.

We aimed to evaluate the validity of self-administered questionnaire surveys and face-to-face interview surveys for the detection of Helicobacter pylori eradication therapy. Participants were a cohort, aged 40-74 years, living in three different locations of Japan, who took part in the baseline survey (2011-2012) of the Japan Public Health Center-based Prospective Study for the Next Generation (JPHC-NEXT). Five years after the baseline survey, a questionnaire and interview survey were independently conducted to determine the history of Helicobacter pylori eradication treatment over the 5-year period. Prescription of Helicobacter pylori eradication medications in national insurance claims data from the baseline survey to the 5-year survey was used as a reference standard. In total, 15,760 questionnaire surveys and 8,006 interview surveys were included in the analysis. There were 3,471 respondents to the questionnaire and 2,398 respondents to the interview who reported having received Helicobacter pylori eradication treatment within the past five years. Comparison of the questionnaire survey to national insurance claims data showed a sensitivity of 95.1% (2213/2328), specificity of 90.6% (12174/13432), positive predictive value of 63.8% (2213/3471), negative predictive value of 99.1% (12174/12289), and Cohen's Kappa value of 0.71. Respective values of the interview survey were 94.4% (1694/1795), 88.7% (5507/6211), 70.6% (1694/2398), 98.2% (5507/5608), and 0.74. Both the questionnaire and the interview showed high sensitivity, high specificity, and good agreement with the insurance claim prescriptions data. Some participants may have received eradication treatment without going through the public insurance claim database, resulting in a low positive predictive value.

Unsupervised Pattern Analysis to Differentiate Multiple Sclerosis Phenotypes Using Principal Component Analysis on Various MRI Sequences.

Background: Multiple sclerosis (MS) has two main phenotypes: relapse-remitting MS (RRMS) and progressive MS (PMS), distinguished by disability profiles and treatment response. Differentiating them using conventional MRI is challenging. Objective: This study explores the use of scaled subprofile modelling using principal component analysis (SSM/PCA) on MRI data to distinguish between MS phenotypes. Methods: MRI scans were performed on patients with RRMS (n = 30) and patients with PMS (n = 20), using the standard sequences T1w, T2w, T2w-FLAIR, and the myelin-sensitive sequences magnetisation transfer (MT) ratio (MTR), quantitative MT (qMT), inhomogeneous MT ratio (ihMTR), and quantitative inhomogeneous MT (qihMT). Results: SSM/PCA analysis of qihMT images best differentiated PMS from RRMS, with the highest specificity (87%) and positive predictive value (PPV) (83%), but a lower sensitivity (67%) and negative predictive value (NPV) (72%). Conversely, T1w data analysis showed the highest sensitivity (93%) and NPV (89%), with a lower PPV (67%) and specificity (53%). Phenotype classification agreement between T1w and qihMT was observed in 57% of patients. In the subset with concordant classifications, the sensitivity, specificity, PPV, and NPV were 100%, 88%, 90%, and 100%, respectively. Conclusions: SSM/PCA on MRI data revealed distinctive patterns for MS phenotypes. Optimal discrimination occurred with qihMT and T1w sequences, with qihMT identifying PMS and T1w identifying RRMS. When qihMT and T1w analyses align, MS phenotype prediction improves.