[ROC (receiver operating characteristics) curve: principles and application in biology].

Laboratoire de biochimie, toxicologie cliniques, Hôpital d'Instruction des Armées du Val-de-Grâce, Paris, France.
Annales de biologie clinique (Impact Factor: 0.42). 01/2005; 63(2):145-54.
Source: PubMed

ABSTRACT Laboratory test's diagnostic performances are generally estimated by means of their sensibility, specificity and positive and negative predictive values. Unfortunately, these indices reflect only imperfectly the capacity of a test to correctly classify subjects into clinically relevant subgroups. The appeal to ROC (receiver operating characteristic) curve appears as a tool of choice for this evaluation. Used in the medical domain since the 60s, ROC curve is a graphic representation of the relation existing between the sensibility and the specificity of a test, calculated for all possible cut-off. It allows the determination and the comparison of the diagnostic performances of several tests. It is also used to consider the optimal cut-off of a test, by taking into account epidemiological and medical - economic data of the disease. Used in numerous medical domains, this statistical tool is easily accessible thanks to the development of computer softwares. This article exposes the principles of construction and exploitation of a ROC curve.

  • [Show abstract] [Hide abstract]
    ABSTRACT: Some forensic and clinical circumstances require knowledge of the frequency of drug use. Care of the patient, administrative, and legal consequences will be different if the subject is a regular or an occasional cannabis smoker. To this end, 11-nor-9-carboxy-Δ9-tetrahydrocannabinol (THCCOOH) has been proposed as a criterion to help to distinguish between these two groups of users. However, to date this indicator has not been adequately assessed under experimental conditions. We carried out a controlled administration study of smoked cannabis with a placebo. Cannabinoid levels were determined in whole blood using tandem mass spectrometry. Significantly high differences in THCCOOH concentrations were found between the two groups when measured during the screening visit, prior to the smoking session, and throughout the day of the experiment. Receiver operating characteristic (ROC) curves were determined and two threshold criteria were proposed in order to distinguish between these groups: a free THCCOOH concentration below 3 µg/L suggested an occasional consumption (≤ 1 joint/week) while a concentration higher than 40 µg/L corresponded to a heavy use (≥ 10 joints/month). These thresholds were tested and found to be consistent with previously published experimental data. The decision threshold of 40 µg/L could be a cut-off for possible disqualification for driving while under the influence of cannabis. A further medical assessment and follow-up would be necessary for the reissuing of a driving license once abstinence from cannabis has been demonstrated. A THCCOOH level below 3 µg/L would indicate that no medical assessment is required. Copyright © 2013 John Wiley & Sons, Ltd.
    Drug Testing and Analysis 10/2013; DOI:10.1002/dta.1581 · 2.82 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Background: Gastric cancer (GC) represents a major health burden worldwide often diagnosed at an advanced-stage. Biomarkers for screening and prevention of GC are missing. Changes in peripheral blood mitochondrial DNA (mtDNA) have emerged as potential preventive/diagnosis biomarker for cancer risk. We aimed to determine whether peripheral leucocytes mtDNA levels are associated with stages of the gastric carcinogenesis cascade. Methods: We measured mtDNA by quantitative real-time PCR assay in peripheral leucocytes of 28 non-atrophic gastritis patients (NAG), 74 GC patients and 48 matched-asymptomatic controls. In parallel, serological level of IL-8 was determined. Results: Mean mtDNA level was higher in GC patients (p-value=0.0095), compared to controls, with values >8.46 significantly associated with GC (OR=3.93). Three ranges of mtDNA values were identified, interval I: <2.0, interval II: 2.0 to 20 and interval III: >20. Interval I included mainly NAG cases and few GC samples and interval III corresponded almost exclusively to GC patients. All controls felt in interval II, together with some NAG and GC cases; IL-8 levels were significantly higher in GC patients (p-value <0.05) with levels >50 pg/ml observed exclusively in GC patients allowing to distinguish them within interval II. We validated mtDNA results in a second cohort of patients, confirming that mtDNA was significantly higher in GC than in patients with preneoplasia. Conclusions: Circulating levels of mtDNA and IL-8 constitute a potential biomarker for the early detection of GC. Impact: Our findings lead us to propose a new non-invasive method to detect patients with GC risk.
    Cancer Epidemiology Biomarkers & Prevention 08/2014; 23(11). DOI:10.1158/1055-9965.EPI-14-0471 · 4.32 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Backgrounds The objective of this study is to evaluate the contribution of laryngeal intraoperative nerve monitoring (IONM) during thyroidectomy in predicting postoperative laryngeal mobility. Methods Between 2009 and 2012, 127 patients underwent thyroidectomy during which 216 recurrent laryngeal nerves were stimulated with supra-threshold stimulations. Laryngeal mobility was examined through direct laryngoscopy. Statistical analysis was performed to determine specificity (SP), sensitivity (SE), negative predictive value (NPV), positive predictive value (PPV) and a threshold value in order to define a intraoperative diagnostic test. Results Nine patients had a unilateral laryngeal palsy. No bilateral laryngeal palsy was observed. The threshold value to assure the postoperative laryngeal mobility is 280 µV. For this value, SP is 94.06%, SE 100%, NPV 100%, PPV 47.83%. Conclusions Laryngeal IONM can predict a favorable outcome of laryngeal mobility in case where the response exceeds 280 µV. Under this value, the risk of palsy is about 50% suggesting a staged surgery. Head Neck, 2014
    Head & Neck 05/2014; DOI:10.1002/hed.23676 · 2.83 Impact Factor