Faktorska analiza povezanosti inflamatornih, lipidnih, srčanih i bubrežnih biomarkera sa Klasifikacijom dugoročnog 30-godišnjeg kardiovaskularnog rizika

Abstract

U kliničkoj praksi koristi se nekoliko skorova za procenu kratkoročnog (10-godišnjeg) rizika od pojave različitih oblika kardiovaskularnih bolesti (KVB) koji se zasnivaju na multivarijabilnim regre- sionim jednačinama izvedenim iz rezultata praćenja različitih kohortnih grupa. Međutim, pošto je starost promenljiva kojoj se dodeljuje najveći broj poena u modelima 10-godišnjeg rizika, mnoge osobe sa zna čajnim opterećenjem faktorima rizika imaju kratko- ročni rizik daleko ispod granice koja uslovljava inten- zivan tretman, iako njihov dugoročni (30-godišnji) rizik može biti značajan. Takođe, drugi biomarkeri mogu da identifikuju osobe sa većim kardiovasku- larnim rizikom od onog izračunatog primenom skoro- va kratkoročnog rizika. Cilj rada bio je da se analizira priroda uticaja ispitivanih biomarkera na kardiovas- kularni rizik i njihovo grupisanje, kao i povezanost dobijenih faktora sa kategorizacijom 30-godišnjeg rizika faktorskom analizom. Pomoću interaktivnog kalkulatora »30-year risk of cardiovascular disease« izračunavan je dugoročni 30-godišnji rizik za pojavu »kompletne« KVB (sve manifestacije KVB) i »teške« KVB (potencijalno fatalne komplikacije KVB). Analiza glavnih komponenti je korišćena za ispitivanje grupisanja markera inflamacije [visoko-osetljivi C- reaktivni protein (hsCRP), serumski amiloid A (SAA), fibrinogen, a1-kiseli glikoprotein (A1AGP), haptoglo- bin, C3 i C4 komponente komplementa], metabo-lizma lipida [non-HDL i LDL holesterol, trigliceridi, apolipoprotein A-I (apo A-I), apolipoprotein B (apo B), lipoprotein (a) (Lp(a))], bubrežne [kreatinin, mokraćna kiselina, cistatin C (Cys-C)] i srčane funk- cije [N-terminalni pro-natriuretički peptid tip B (NT- proBNP), visoko-osetljivi srčani troponin T (hs-cTnT)], dobijenih analizom uzoraka seruma 242 zdrave oso- be. Faktorskom analizom identifikovano je 5 klastera, kojima je objašnjeno je 67,4% ukupne varijacije, ras- poređene na sledeći način 1) 29,7% »sistemska infla- macija« (hsCRP, fibrinogen, SAA, A1AGP, haptoglo- bin, C3 i C4 komponenta komplementa); 2) 12,5% »aterogena dislipidemija« (LDL i non-HDL holesterol, apo B i trigliceridi); 3) 11,0% »kardiorenalni faktor« (kreatinin, mokraćna kiselina, Cys-C i hs-cTnT); 4) 7,6% »hemodinamski faktor« (NT-proBNP) i 5) 6,7% »lipoproteinski faktor« [apo A-I, Lp(a)]. Prediktivne vrednosti u proceni 30-godišnjeg rizika za »komplet- nu KVB« i »tešku KVB« su bile značajne za četiri fak- tora (OR 1,892–5,590; P<0,0001 i OR 2,183– 5,931; P<0,0001, redom), a »hemodinamski fak- tor« nije imao statistički značajan prediktivni potenci- jal za vrednosti iznad optimalnih/normalnih za odgo- varajući pol i starost (P>0,05). Površine ispod ROC krivih (AUC) modela sa pet faktora u predikciji povećanog 30-godišnjeg rizika za »kompletnu KVB« i »tešku KVB« iznosile su 0,881 i 0,888, redom, i nisu bile statistički značajno različite od multivarijabilnog logističkog modela od 18 polaznih parametara (0,892 i 0,901; P>0,05; redom). Sistemska inflama- cija, aterogena dislipidemija, kardiorenalna funkcija i lipoproteinski status nezavisno doprinose dugo- ročnom, 30-godišnjem riziku iznad normalnog/opti- malnog kako za ozbiljne komplikacije KVB, tako i za sve vrste kardiovaskularnih komplikacija.

Several risk score algorithms for short-term (10- year) cardiovascular risk assessment based on multi- variable regression equations derived from different cohorts are being used in clinical practice. However, since the age is variable with the strongest influence on short-term risk, many individuals with moderate increase of other traditional risk factors would have a 10-year risk below cutoff for intensive treatment, but a significant long-term (30-year) risk. Also, other bio- markers might identify persons with higher actual cardiovascular risk compared with calculated using short-term risk scores. The aim of this study was to analyze the nature of influence of examined biomark- ers on cardiovascular risk and their clustering, as well as relations of identified factors with long-term 30- year risk categorization, using factor analysis. Interactive calculator »30-year risk of cardiovascular disease« was used for long-term 30-year risk calcula- tion, for both »full CVD« (all manifestations of cardio- vascular disease) and »hard CVD« (serious manifesta- tions of CVD). Principal component analysis was used to investigate clustering of markers of inflammation [high sensitivity C-reactive protein (hsCRP), serum amyloid A (SAA), fibrinogen, a1-acid glycoprotein (A1AGP), haptoglobin, C3 and C4 complement components], lipid metabolism [non-HDL and LDL cholesterol, triglycerides, apolipoprotein A-I (apo A- I), apolipoprotein B (apo B), lipoprotein (a) (Lp(a))], renal [creatinine, uric acid, cystatin C (Cys-C)] and cardiac function [N-terminal pro-natriuretic peptide type B (NT-proBNP), high sensitivity cardiac troponin T (hs-cTnT)], obtained from 242 apparently healthy individuals. Factor analysis identified five clusters, which explained 67.4% of the total variance distrib- uted as follows: 1) 29.7% »systemic inflammation« (hsCRP, fibrinogen, SAA, A1AGP, haptoglobin, C3, C4); 2) 12.5% »atherogenic dyslipidemia«, (LDL and non-HDL cholesterol, apo B, triglycerides); 3) 11.0% »cardiorenal factor« (creatinine, uric acid, Cys-C, hs- cTnT); 4) 7.6% »hemodynamic factor« (NT-proBNP); and 5) 6.7% »lipoprotein factor« [apo A-I, Lp(a)]. When estimating 30-year risk from both »full CVD« and »hard CVD«, predictive values were significant for four factors (OR 1.892–5.590, P<0.0001 and OR 2.183–5.931, P<0.0001, respectively), and »hemodynamic factor« had no statistical significance in predicting potential for values above optimal/nor- mal for corresponding gender and age (P>0.05). The areas under the receiver operating characteristic curves (AUCs) of the five factor model in predicting increased 30-year risk for »full CVD« and »hard CVD« were 0.881 and 0.888, respectively, which were not statistically significantly different from AUCs of the multivariable logistic model of 18 original parameters (0.892 and 0.901, P>0.05, respectively). Long- term, 30-year risk above normal/optimal for hard CVD complications and for all kinds of cardiovascular complications was independently contributed by sys- temic inflammation, atherogenic dyslipidemia, car- diorenal function and lipoprotein status.