Neinvazivni dijagnostički model dijagnostike anemije u mijelodisplastičnom sindromu

master's thesis

• Title (english): Noninvasive diagnostic model for diagnosis of myelodysplastic syndrome related anemia
• Author: Antonio Hanžek
• Mentor: Slobodanka Ostojić-Kolonić (mentor)
• Committee member: Tajana Filipec Kanižaj (predsjednik povjerenstva)
• Committee member: Jasenka Markeljević (član povjerenstva)
• Committee member: Slobodanka Ostojić-Kolonić (član povjerenstva)
• Granter: University of Zagreb; School of Medicine; (Department of Internal Medicine); Zagreb
• Defense date and country: 2019-07-12, Croatia
• Scientific / art field, discipline and subdiscipline: BIOMEDICINE AND HEALTHCARE; Clinical Medical Sciences; Internal Medicine

Abstract

Aspiracija koštane srži predstavlja zlatni standard za postavljanje dijagnoze mijelodisplastičnog sindroma (MDS). U svrhu izbjegavanja spomenutog invazivnog dijagnostičkog postupka, razvijen je neinvazivni dijagnostički model temeljem lako dostupnih laboratorijskih parametara. Skupina pacijenata s MDS-om (n=53) uspoređena je s kontrolnom skupinom (n=121). Za kontrolnu skupinu odabrani su pacijenti s Ne-Hodgkin limfomom B i T-imunofenotipa, multiplim mijelomom, Waldenströmovom makroglobulinemijom, leukemijom vlasastih stanica i aplastičnom anemijom. Od ukupno analiziranih 15 parametara, u sklopu multivarijatne logističke regresije, njih 3 pokazalo se statistički značajnima u razlikovanju MDS skupine od kontrolne skupine: prosječni volumen eritrocita (p <.0001), apsolutni broj neutrofila (p<.0001) i prosječni volumen trombocita (p=.0036). Širina distribucije volumena eritrocita pokazao se kao statistički granično značajan laboratorijski parametar (p=0,0740) te je stoga ostavljen u konačnom modelu. Površina ispod ROC (engl. receiver operating characteristic) krivulje iznosi 0,9567. Pomoću dobivenog modela koristeći formulu logističke regresije izračunata je vjerojatnost (P) postojanja MDS-a u svakog pojedinog pacijenta. Vrijednost vjerojatnosti 0,3 odabrana je kao granična vrijednost s osjetljivošću (86,79%), specifičnošću (86,78%), pozitivnom prediktivnom vrijednošću (74,19%) i negativnom prediktivnom vrijednošću (93,75%). Pacijenti s P vrijednošću iznad definirane granične (P=0,3) kategorizirani su kao „vjerojatni MDS“, a pacijenti s P vrijednošću ispod granične kategorizirani su kao „vjerojatno nije MDS“. P vrijednosti povezane su s izračunatim rizikom pri dijagnozi MDS-a; najviše pogrešno svrstanih pacijenata u sklopu interne validacije modela bilo je u skupini niskog rizika, njih 4 (30,8%).

Abstract (english)

Bone marrow aspiration is a gold standard for the diagnosis of myelodysplastic syndrome (MDS). For the purpose of avoiding this invasive diagnostic procedure, new diagnostic model based on easily available laboratory parameters was developed. Group of patients with MDS (n=53) was compared with the control group (n=121). For the control group, patients with B and T-cell non-Hodgkin's lymphoma, multiple myeloma, Waldenstrom's macroglobulinemia, hairy cell leukemia and aplastic anemia were selected. Using multivariate logistic regression, of 15 parameters analysed in total, 3 of them were significantly different between groups: mean corpuscular volume (p <.0001), absolute neutrophil count (p<.0001) and mean platelet volume (p=.0036). Red cell distribution width turned out to be borderline statistically significant laboratory parameter (p=0,0740) and therefore was included in the final model. The area under the ROC curve (AUC) was 0,9567. By using a developed model, probability (P) of having an MDS was calculated for each patient. Probability of 0,3 was chosen as a cut-off value with a sensitivity (86,79%), specificity (86,78%), positive predictive value (74,19%) and negative predictive value (93,75%). Patients with P value above cut-off value (P=0,3) were categorized as „probably MDS“, while patients with P value below cut-off were labeled as „probably not MDS“. P values were compared with MDS risk calculated at diagnosis; most of inaccurately classified patients within internal validation of model were in low risk group, 4 of them (30,4 %).

• Keywords: MDS; multivarijatna logistička regresija; neinvazivna dijagnostika
• Keywords (english): MDS; multivariate logistic regression; noninvasive diagnosis
• Language: croatian
• URN:NBN: urn:nbn:hr:105:626534
• Study programme: Title: Medicine; Study programme type: university; Study level: integrated undergraduate and graduate; Academic / professional title: doktor/doktorica medicine (doktor/doktorica medicine)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2020-06-12 08:56:53