University of Groningen
Long-term cardiovascular effects of breast cancer treatment
Boerman, Liselotte
DOI:
10.33612/diss.116880323
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2020
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Boerman, L. (2020). Long-term cardiovascular effects of breast cancer treatment. University of Groningen. https://doi.org/10.33612/diss.116880323
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
Troponin and (NT-pro)BNP as predictors for the occurrence
of cardiac dysfunction during or after breast cancer
treatment: a systematic review
L.M. Boerman D. Brandenbarg C. van der Vlist P. van der Meer G.H. de Bock A. J. Berendsen M.Y. Berger Submitted
Chapter 4
Abstract
Purpose
To date, monitoring of cardiac function after cardiotoxic chemotherapy relied on imaging techniques. The aim of this review was to evaluate association between the biomarkers troponin and (NT-pro)BNP and cardiac dysfunction in adult women treated for breast cancer.
Patients and methods
PubMed and Embase were systematically searched. Studies that evaluated the association between troponin, (NT-pro)BNP during cardiotoxic therapy and cardiac dysfunction during and after treatment using echocardiography, magnetic resonance imaging, or multigated acquisition scans were included. Random effect models were applied to pool odds ratios (ORs).
Results
12 out of 691 identified studies were included, evaluating 742 women. Median follow-up ranged from 0–14 months. The weighted prevalence of cardiac dysfunction was 17.8% (range 0%–47%). Increased troponin or (NT-pro)BNP levels during treatment were not associated with cardiac dysfunction (OR 3.14, 95%CI 0.73–13.50 and 2.07, 95% CI 0.50-8.59, respectively).
Conclusion
There is insufficient evidence that either troponin or (NT-pro)BNP is associated with cardiac dysfunction during or after chemotherapy or radiotherapy for breast cancer. Studies with longer follow-up periods are needed to evaluate the predictive value of these biomarkers for cardiac dysfunction.
Systematic review troponin & (Nt-pro)BNP
59 Introduction
Each year approximately 1.7 million women are diagnosed with breast cancer worldwide, making it the most common type of cancer among women.1Fortunately, due to advances in treatment and early detection, survival rates have increased by 25% to a 10-years survival of 75%.8, 10, 11, 13, 117 Most women now receive chemotherapy, radiotherapy, or both.15-18 However, these therapies are associated with increased cardiovascular morbidity and mortality, which may present many years after treatment.73-75 On the one hand, after chemotherapy (mainly anthracyclines) and targeted therapy (trastuzumab) women have been shown to develop congestive heart failure significantly earlier than women who did not receive those therapies.34-37, 118 On the other hand, women who received radiotherapy have been shown to be at risk of cardiomyopathy and ischemic heart disease due to vascular damage.52
Cardiac dysfunction due to chemotherapy and/or radiotherapy for breast cancer may lead to congestive heart failure or cardiomyopathy at any stage from during treatment to more than 10 years after treatment has ended.35, 46The type of cardiac damage that occurs during initial treatment seems reversible, at least in the short term, possibly because agents like trastuzumab simply block the repair mechanism of the heart while being taken.64 By contrast, cardiac dysfunction that becomes clinically apparent during long-term follow-up is usually irreversible and is probably caused by oxidative stress and free radical damage to the cardiac muscle during chemotherapy and radiation treatment.64
Given the increasing number of long-term survivors of breast cancer, physicians should be aware of these potential irreversible effects on cardiac function. Indeed, early detection of cardiac dysfunction is important because timely treatment might prevent further deterioration.119-121 However, early detection is difficult because cardiac damage and dysfunction can be asymptomatic, or can manifest with non-specific symptoms like fatigue.39 To date, monitoring of cardiac dysfunction has relied on echocardiography, magnetic resonance imaging (MRI), and multigated acquisition (MUGA) scanning.122 In contrast to these costly and labor intensive methods biomarkers troponin, brain natriuretic peptide (BNP) and N-terminal pro-BNP (NT-proBNP) are widely available. Troponin assays are commonly used in the diagnosis of acute myocardial infarction123, while BNP and NT-proBNP are routinely used to screen for heart failure.124, 125 To date, experimental research has shown that exposure to anthracycline chemotherapy causes a dose-dependent reduction in cardiomyocyte contractility that correlates with the release of the cardiac-specific biomarker
troponin T.126 Furthermore, several studies have found that increased levels of BNP and NT-proBNP are negatively correlated with the left ventricular ejection fraction (LVEF) in patients receiving cardiotoxic treatments for breast cancer.127-131
Given that troponin, BNP or NT-proBNP levels may increase due to the cardiotoxic effects of chemotherapy and radiotherapy, they may be used as biomarkers to identify patients with cardiac damage after breast cancer treatment who might be at increased risk for cardiac dysfunction. In this systematic literature review and meta-analysis, we evaluated the association between increased value of the biomarkers troponin, BNP, NT-proBNP and the occurrence of cardiac dysfunction during or after chemotherapy and/or radiotherapy for breast cancer. In this study, (NT-pro)BNP is used to refer to studies of either BNP or NT-proBNP and troponin to studies of either Troponin I or T.
Material & Methods
Box 1 Search strategy
Design, selection criteria and data extraction
Searches were performed of the PubMed and Embase databases from inception to October 2016. The study was performed and is reported according to the guidance of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). The research protocol is available at the PROSPERO register [ID: CRD42014007195]. Search terms were based on MeSH terms and text related to breast cancer, troponin, (NT-pro)BNP, and cardiac dysfunction (Box 1). We also searched the reference lists of the included review articles and studies. Independently, three reviewers analyzed and identified studies for eligibility and extracted the relevant data (LMB, CvdV, and DB). Two reviewers assessed the methodological quality of the included articles (LMB and DB). Disagreements were resolved by consensus or arbitration with a third, independent reviewer (MYB or AJB).
For inclusion, studies were required to have a study population of women aged шϭϴ LJĞĂƌƐ ǁŚŽ ǁĞƌĞ ƚƌĞĂƚĞĚ ǁŝƚŚ ĐĂƌĚŝŽƚŽdžŝĐ ƚŚĞƌĂƉŝĞƐ ;ĂŶƚŚƌĂĐLJĐůŝŶĞƐ͕ ƚĂdžĂŶĞƐ, trastuzumab, or radiotherapy) for breast cancer, with no restrictions to care setting.
Keywords used were: “breast neoplasm”, “breast cancer”, “breast carcinoma”, “ductal
carcinoma”, “breast tumor”, “breast tumour”, “mammary neoplasm”, “mammary carcinoma”, “mammary cancer”, “cancer of the breast”, “pro-brain natriuretic peptide (1-76)”, “Natriuretic Peptide, Brain”, “Troponin”, “NT-pro BNP”, “pro-BNP”, “BNP”, “proBNP”, “cardiac biomarker”, “cardiotoxicity”, “cardiac toxicity”, “cardiac dysfunction” and “ ventricular dysfunction”
Box 1 Search strategy
Systematic review troponin & (Nt-pro)BNP
61
Studies had to address the target condition of cardiac dysfunction during or after treatment, assessed by echocardiography, MRI, or MUGA. The studies had to include measurements of troponin or (NT-pro)BNP at time of treatment. Data collection had to be prospective. We included studies with survivors of other types of cancer if separate analyses had been done for survivors of breast cancer.
Using a structured predefined data extraction form, we extracted information on the following: care setting, follow-up time, participants (age, diagnosis, and type of breast cancer treatment), drop-out rates (including the details), information about troponin and (NT-pro)BNP measurements (type, index tests, cutoff values, and measurement time points), and details of cardiac dysfunction measurements and outcomes (definitions of cardiac dysfunction as assessed by echocardiography, MRI, and MUGA and time of assessment (during or after treatment)).
Definitions
A positive biomarker was defined as one or more measurements above the cut-off value as used by the authors of the original study at the measurement points as depicted in table 1. For evaluating the occurrence of cardiac dysfunction the definition of the included study was followed (table 1). The treatment period was defined as the period from start of treatment until the end of cardio toxic treatment.
Quality assessment
To assess the methodological quality of the included articles, we used a checklist based on the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies of the National Institutes of Health (Table 2).132 Listed items were divided into five
categories: general, study population, type of breast cancer treatment, measured biomarkers, and outcome. Items could be answered “yes (1 point),” “no (0 points),” “cannot determine (CD),” “not applicable (NA),” and “not reported (NR).” The item was ignored if the outcome was marked NA, or grouped under “no” if marked NR or CD. The summary score of each study was then calculated and expressed as a percentage that was categorized into one of four categories: poor (0%–25%), fair (25%–50%), good (50%–75%), or excellent (75%–100%).
Analysis
For each study, the occurrence of cardiac dysfunction was calculated as a percentage of the total sample of women treated for breast cancer, weighted by study size. The relation between a biomarker test above the cut-off value and the occurrence of
cardiac dysfunction was evaluated by estimating odds ratio’s (ORs) and 95% confidence intervals (95% CIs). To meta-analyze the estimated results, Mantel–Haenszel random effects models were applied to calculate pooled estimates of the ORs. Overall statistical heterogeneity was tested by I2 tests. Results were presented stratified by moment of outcome assessment (during treatment only vs. during or after treatment) in forest plots. Potential sources of heterogeneity were evaluated. For outliers (identified by “eyeballing”), we evaluated whether bias or specific study characteristics could explain the results. All analyses were conducted using Review Manager (version 5.3. Copenhagen, The Cochrane Collaboration, 2014).
Results
Study characteristics and participants
The search yielded 691 unique studies (Figure 1), of which 12 met our inclusion criteria. These studies comprised 742 patients in total (range 9–251) with a mean age varying from 46 to 62 years (Table 1). In all but one study133, patients received anthracyclines.
In nine studies, all patients received trastuzumab133-141, in one study only 20% of
patients received traztuzumab treatment.142 Eight studies comprised women who also
received (neo)adjuvant radiotherapy.134, 135, 137-142 Two studies excluded patients who
had previously received chemotherapy or radiotherapy.139, 143 In three studies, patients
had received these therapies before baseline measurement in three studies.134, 136, 142
The other studies did not mention whether prior treatment was an exclusion criteria.133, 135, 137, 140, 141, 144, 145
The two reviewers initially agreed upon 75% of the items in the methodological quality list. The mean overall quality score for all included studies was 65% (range 38%– 88%).Three studies scored excellent134, 137, 139, seven studies scored good133, 135, 136,
141-143, 145, and two were of fair quality (Table 2).140, 144 The most prevalent shortcomings
were in reporting the participation rate of eligible persons and similarity of exposed and unexposed populations. Sample sizes were frequently not justified. Furthermore, blinding for the exposure state was a frequently occurring source of potential bias, as was appropriateness of analysis, and statistical adjustment for confounding variables.
Systematic review troponin & (Nt-pro)BNP
63
Records identified through database searching (n = 870) Scre eni n g Included E lig ib ility Identi fi cati o
n Additional records identified through other sources (n=0)
Records after duplicates removed (n = 691)
Records screened (n = 691)
Records excluded based on title and abstract
(n = 485)
Full-text articles assessed for eligibility
(n = 115)
Full-text articles excluded No reference standard (n = 13) No index test (n = 24)
No separate analysis for breast cancer patients (n = 10)
No analysis relating biomarker and cardiac function (n = 10)
No information on prognostic value available (n = 4)
Unable to construct a 2 × 2 table (n =31) No prognostic research (n = 11) Studies included in quantitative synthesis (meta-analysis) (n = 12) No full text available (n = 91)
Figure 2 Flow diagram for study inclusion
S creening Included E lig ibility Ide nt if ic at ion
4
Table 1 Characterist ics of t h e included studies Outcom e M e a sur e ECHO a n d M R I Mea su ri n g poi nts: sam e t im e po int s as bi om ar k e rs f o r ECHO , and at base line and 12 m o . for MRI ECHO Measuri ng poi n ts for ca rd iot o xi cit y : baseli ne, a n d every 3 mo . until th e end o f trastu mab at 15 mo . (NT-p ro)B NP NT -p ro BNP Cutoff v a lu e : >35 pm ol/l Mea su ri n g poi nts for bi om ark er: Baseline ( b efor e st ar ti ng treatm ent wi th anthracycl ines), be fo re and at 3, 6, 9, and 12 m o . af te r st a rt ing tra stu zu m a b N T -proBNP Cu to ff: ? Measuri ng p o in ts fo r bi omarker: baseli ne, an d
every 3 mo. until t
he e nd of trastu mab at 15 mo . Troponi n TnT Cutoff v a lu e : >0.01 ng /m L Mea su ri n g poi nts: Baseline (be for e st ar t treatm ent wi th anthracycl ines) , be fo re , and at 3, 6, 9, and 12 m o . af te r st ar ti ng tra stu zu m a b Tn T Cu to ff val u e: hsTnT > 5.5 pg/mL Measuri ng poi n ts for bi omarker : baseli ne, an d every 3 mo . until t he en d of trastumab at 15 mo . Defi ni ti on of C ardiotoxi ci ty Decline in LVE F of at l e as t 10% be low 55% wit h ac com pany ing si g n s o r sy mp to ms o f CHF , ne cess it a ti ng di sc ont inua tion
of the drug reduction of >s &
шϱй Popul ati o n Inclusi on: Fe mal e patients who r eceiv ed adjuv ant t rast u zu mab in th e se ttin g of H E R2 overexpress ing BC . Excl usi o n : E lig ib le : E = u n kn ow n Included : E = 42 M e an a g e : 47 ± 9 y Fol low-up d u rati on afte r tr e a tm e n t: 0 mo. Lo st to f o llo w -u p : no t rep o rted Inclus ion:
HER2-p os iti v e b re a st ca nc er and sched ule d to re ceive adjuvant chemo th e rap y includ ing anth racycline s,
taxanes and trastu
zumab E x clusi o n: -E lig ible : Eс Un kn own In cl u d ed : Eс 20 Setti ng and T reat m ent S e tting : consecutive di agno se d B C pat ient s wit h HER 2 overexpress ion a t Universit y hospi tal C a nada BC t reat m ent : E p irubicin (88%) Doxorubicin (12%), Tr ast u zu mab (100%) R a diot he ra p y ( 98%) S e tti ng: Breast
cancer women at medical u
n ivers ity hospital BC treatment: An th racycli nes (100%) Tax a nes (100%) Tras tu zumab (100%) Radioth e rap y (63%) D esi gn P rospective coh ort s tu d y Pro spective cohort stu d y Author Fal lah-R ad, 2011 135 Katsurad a, 2014 141
Systematic review troponin & (Nt-pro)BNP 65 ECHO Mea su ri n g poi nts for cardiotoxi ci ty: Baseline (befor e any t h e rapy), af te r anthracycl ines (3 mo.), 6, 9,12 and 15 mo. af te r base line ECHO /MR I M e a suri n g poi nts for ca rdi o to xi ci ty : Baseline (befor e treatm ent with tr ast u zu mab); 3 m o nt hs , 6 m o nt hs , and 12 m o nt hs af te r di agno si s NT -p ro BNP Cutoff v a lu e : >125 pg /m L M e a suri n g poi nts for bi om ark er : Baseline, and af te r 3 and 6 mo. BNP Cutoff: > 100 pg /m L M e a suri n g poi nts for bi om ark er: Baseline (befor e treatm ent with tr ast u zu mab); 3 m o nt hs , 6 m o nt hs , and 12 m o nt hs af te r di agno si s TnI Cutoff v a lu e : >0.03 ng /m L Mea su ri n g poi nts for bi om ark er: Baseline, and af te r 3 and 6 mo . TnI Cutoff v a lu e : >0.3 ng/ m L Mea su ri n g poi nts for bi om ark er: Baseline (befor e treatm ent with tr ast u zu mab); 3 m o nt hs , 6 m o nt hs , and 12 m o nt hs af te r di agno si s C a rdiom yopat hy wi th
decreased LVEF, a reduction
of LVEF o f LVEF >5% t o < 55% wit h sy mp to ms o f he ar t f a il ur e or an asym pt om ati c reduction of >s &ш ϭϬйƚ Ž <55% (CR E C) LVEF < 55% Inclusi on: - WĂƚ ŝĞŶƚ ƐшϭϴLJ Ž Ĩ ĂŐ Ğ - Newly diag n osed with h u m a n ep id erm a l g row th factor recept o r 2-overexpress ing BC and scheduled t o rece ive adjuv ant t h e rapy inc lud ing ant h ra cycl ine s, t axan es, and tr ast u zu mab Excl usi o n : P a
tients who had a basel
ine LVEF < 50% E lig ib le : E = 99 Included : E = 81 M e an a g e : 50 ± 10 y Fol low-up d u rati on aft e r tr eat m ent : 0m o Inclusi on: Newly diag n osed wit h HER 2-overexpress in g BC Excl usi o n : A rtific ia l pa cem a ker im pl ant a tion, c h ro n ic ar rh y thmia su ch as at ri al fi br illat ion, c h ro ni c k idne y di se ase ( e GFR <30m L/ m in/1.7 3m 2), m e tastatic cancer Elig ib le : Eс unk o wn Included : Eс 16 (included in anal ysi s 9) M e an a g e : 62. 3 ± 12.6 y e ar s Fol low-up d u rati on aft e r tr eat m ent : u n kn ow n Lost t o fol low-u p : 7 S e tting : consecutive pat ient s ne wl y di agno se d wit h B C at a universi ty hos pi tal in Ja pa n BC T reat m ent : Doxorubicin (88%) Ep ir ubicin (12%) Taxan es (100%) Tr ast u zu mab ( 100%) R a diot he ra p y ( 60%) S e tting : C o nsecutive newly diag n osed H E R2 overexpress ive BC at universi ty hospi tal Japan BC t reat m ent : Ant h ra cyclin es (66.7%) Taxan es (33.3%) Tr ast u zu mab ( 100%) R a diot he ra p y ( 55.6%) P rospective coh ort s tu d y P rospective coh ort s tu d y Ky, 201 3 139 Nakano, 2016 140
4
ECHO/M UGA M e a suri n g poi nts for ca rdi o to xi ci ty : Baseline ECHO/M UGA at 1 m o . bef o re t o 1 m o . af te r tr ast u zu mab-initia tion. Repeated ev ery 3–4 mo. dur ing treatm ent (unclear whe th e r EC HO or MUGA ) MUGA and ECH O M e a suri n g poi nts for ca rdi o to xi ci ty : base line , af te r 4 cycles a n d a fter com pletion o f treatm en t both MUGA and ECH O BNP Cutoff: шϮϬϬƉŐ ͬŵ > M e a suri n g poi nts for bi om ark er: Baseline (before treatm ent);
every 3 wk during treatm
ent, up to 12 mo . NT -p ro BNP Cutoff v a lu e : 12 5 ng/L M e a suri n g poi nts for bi om ark er : be fo re , at 1 and 7 days af te r th e firs t tra stu zu m a b inf u sion, and at the end of each cycle cTnI Cutoff v a lu e : шϬ͘ϭ ng/ m L Measur ing points f o r bi om ar k e r: Bas e li n e (befo re tr eat m ent ); eve ry 3 wk duri n g tr eat m ent , up to 12 m o TnI Cutoff v a lu e : >500,000 n g /m L M e a suri n g poi nts for bi om ark er : be fo re , at 1 and 7 days af te r th e f ir st tra stu zu m a b inf u sion, and at t h e end of each c ycle
Decrease in ejection fract
ion of 15% o r mor e from baseline o r to a value be low 50% Se vere heart fa il ur e (Com m o n toxicit y cri ter ia g rade 3; NYH A functiona l cla ss II to IV) Inclusi on: F e m a le patients ag ed >18 y -F irst diag n osis of HE R2 -p ositive pr im ar y B C ( st a ges 1–3) -D eem ed elig ible for tr ast u zu mab adjuvant t h er apy Excl usi o n : m iss ing l a bor a to ry values or i n suff icien t tes ting for LVEF Elig ib le : Eс 54 Included : Eс 49 M e an a g e : 55. 41 y ( p at ient s wit h a no rm al LVEF) and 59.58 y (pat ient s wi th a de c reased LVEF) Fol low-up d u rati on aft e r tr eat m ent : 0 mo. Lost t o fol low-u p : 0 Inclusi on: Ő Ğш ϭϴLJ ͕ĚŝĂ ŐŶ Ž ƐĞ Ě with H E R2 -p ositive m e ta static or local ly advan ced BC , suitable for tr eatm
ent with pac
litaxel and trastuz u m a b, and an E a stern C ooperativ e Oncol o g y Group perf o rm ance sta tus o f 0 to 2. Excl usi o n : T reatm
ent with any
inv e st ig at iona l dr ug wi th in 30 days be fo re t h e s tar t o f t h e st udy, r a diot h e ra py wi th in 4 wk of e n ro ll m e nt , ser ious uncontrol led C N S m e tastases, LVEF < 40% on MUG A , o r sym p tom a
tic heart fai
lure NY H A fu nct iona l c lass III or IV. Pat ient s S e tting : ne wly di agno sed pr im ar y BC patients at a cancer cl in ic in th e USA BC t reat m ent : A n thracyclin es (44%) Taxan es (85%) Tr ast u zu mab ( 100%) Ra dioth e ra p y (37%) S e tting : P a tients at the Univers ity ho sp it al G roni n gen (Nether lands) BC t reat m ent : E p ir ubicin (40%) Doxorubicin (6 7 % ) Pac lit axe l ( 100%) Tr as tu zu mab (100%) P rospective coh ort s tu d y P rospective Coho rt st udy On itilo , 2012 137 Peri k, 2006 136
Systematic review troponin & (Nt-pro)BNP 67 ECHO M e a suri n g poi nts for ca rdi o to xi ci ty : A t baseline and 12 week s and 24 week s, and every th ree m o nt hs thereaf ter BNP Cutoff: > 100 pg /m L M e a suri n g poi nts for bi om ark er: A t baseline, and every oth er cy cle for up to 6 tim e po int s (12,18, 24,3 0 week s) cTnI Cutoff v a lu e : > 0.0 6 ng/ m L M e a suri n g poi nts for bi om ark er : A t baseline, and every oth er cycle f o r up t o 6 tim e points (12,18, 24,3 0 week s) a sig n ificant asym pt om atic LVE F decline o f 10% t o 15% t o less than ϱϬйŽƌ ш 16% (f ro m base line ) or experience d clinica l he ar t fa ilu re abno rm al labor a to ry t e st s (ne u tr oph il coun t <1.5 × 10 9/L, pl at ele ts <100 × 10 9/L, se ru m to ta l bi lir ub in >1.5 × ULN , an ALT or AST m o re t h an 2.5 × ULN [ > 5.
0 × ULN in case of liver
m e ta st ases], al k a li ne phosp h at ase m o re t h an 2.5 × ULN [> 4.0× ULN in cas e o f l iver or bo ne m e ta st ases], or s e ru m cr eat in ine m o re t h an 1.5 × ULN ). E lig ib le p a tie n ts : Eс 17 Included p a ti ents : Eс 15 M e an a g e : 52 ,4 y Fol low-up du rati on after tr eat m ent : 0 mo. Lost to fol lo w-up: 2 (b oth d u e to on col o g ica l rea son s) Inclusi on: H E R2 -p ositive breas t cancer with m e tasta tic d isease, Ő Ğш ϭϴLJ ͕ĂŶ ĂƐƚ Ğƌ Ŷ C ooperative On colog y Group pe rf o rm ance st a tus o f 0 t o 1, m e asur able o r nonmeasur able di se ase, z e ro t o o n e p rior treatm ent, ad eq ua te o rg a n ĨƵ ŶĐƚ ŝŽŶ͕ďĂƐĞ ůŝŶĞ >s &Ž Ĩ шϱϬй by e cho wi th in 4 we ek s be fo re enrollm ent Excl usi o n : a history of p rior ca rd ia c m o rb id it ies wit h in 12 m o nt hs ( u nst a ble angina, m yocardial infar ction, cong e stive hear t fai lure, S e tting : Metastatic breast cance r patients a t a cancer centre in the USA BC t reat m ent : A n thracyclin es (0%) Taxan es (100%) Tr ast u zu mab ( 100%) Ra dioth e ra p y (?%) P rospective coh ort s tu d y Yu, 2016 133
4
ECHO Me a suri n g poi nts for ca rdi o to xi ci ty : Baseline (befor e tra stu zu m a b treatm en t), ever y 3 m o . dur ing tra stu zu m a b treatm en t, every 3 m o . dur ing t h e f ir st yea r afte r di sc ont inua tion, and every 6 m o . T h ereafter ( in case of loss to fo llow -up or oncolog ic d e ath, last eva lua tion = fi nal measur em ent ) - cTnI Cutoff v a lu e : >0/ 08 ng/ m l M e a suri n g poi nts for bi om ark er : TnI be fo re and so on af te r e a ch tra stu zu m a b cycle (ea ch patient and ea ch cardiolog ic ch eck af ter com pletion o f th e th era p y) LVEF decrease of >10 uni ts f ro m base line , asso ciat e d wi th a decline be low th e norm a l lim it of 50% uncontrol led ven tricu la r arrhythm ias), prior pertu zum a b, Őƌ ĂĚĞш ϮŶĞ Ƶƌ ŽƉ at hy E lig ib le : Eс u n kn ow n Included : Eс 69 M e di an a g e : 53 ( ra ng e 26 -84) Fol low-up d u rati on aft e r tr eat m ent : m e di an 21 m o nt hs (r ang e 2-38); t reat m ent dur a tion 30 week s Los t to fo llow -up : 2 Inclusi on: W o m e n wit h B C unde rg oing t rast u zu mab therapy (ea rly-stag e and advanced or m e ta st at ic B C, wi th H E R2 overexpres sion) Excl usi o n : A g e < 18 y o r > 75 y , ischem ic and va lvu la r hea rt di se ase, LVEF < 55% be fo re treatm en t, severe hyper tension, liĨ ĞĞ džƉĞ Đƚ ĂŶĐLJ чϭϮǁŬ ͕Ž ƌ abno rm al r e nal o r hepa tic functions. Elig ib le p a tie n ts : Eс u n kn ow n Included p a ti ents : Eс 251 M e an a g e : 50 ± 10 y Fol low-up d u rati on aft e r tr eat m ent : m e di an 14 m o . (r ang e 1– 79 mo.) Lost t o fol low-u p : 66 (62 died of onc olog ic reas on) S e tting : C o nsecutive di agno se d B C patients a t a cancer center in Italy BC t reat m ent : E p ir ubicin (31%) Doxorubicin (48%) Taxan es (7%) Tr ast u zu mab ( 100%) R a diot he ra p y ( 35%) P rospective coh ort s tu d y C ardinal e, 2010 134
Systematic review troponin & (Nt-pro)BNP 69 ECHO / MUGA Mea su ri n g poi nts for cardiotoxi ci ty: baseline (b efore treatm en t) and 1 y e ar af te r chem otherapy both MUGA and ECHO ECHO M e a suri n g poi nts for cardiotoxi ci ty: Before radioth e ra p y , im m e diately after ra diot he ra p y , 8 and 12 m o . af te r ra dioth e ra p y - - cTnT Cutoff v a lu e : >0.0 1 ng/ m L M e a suri n g poi nts for bi om ark er : B e fo re and 72 ho ur s after each cycl e of chem otherapy (eac h pat ien t 4–6 cycles ) TnI Cutoff v a lu e : >0.1 3 ng/ m L M e a suri n g poi nts for bi om ark er : first and last day of ra dioth e ra p y Final LVEF <50% or an abso lut e dr op of > 10% in the LVE F S ig n ificant decrease in conventio nal p a ra m e ters for syst olic or diastoli c function or stra in Inclusi on: patients treated wi th epiru b icin -cont a in in g adjuvant c h em o the ra py fo r st ag e 2 or 3 B C who have underg o ne either m o dified radica l m a stectom y or br ea st-saving surg ery. Excl usi o n : Pat ient s wit h m e ta st asis, ischem ic or val v ular hear t diseas e , any ty pe o f car d io m yopat hy , hear t f a il ur e , base line LVE F < 50%, and unc o nt rolle d hy pe rt ens ion. E lig ib le p a tie n ts : Eс u n kn ow n Included p a ti ents : Eс 45 M e an a g e : 48 ± 8 y Fol low-up d u rati on aft e r t reat m ent : 12 mo. Lost t o fol low -u p : 4 (2 due to oncolog ical r e asons) Inclusi on: P a tients with hist olog ical ly proven earl y BC req u ir ing adjuvant R T , a nd fulfi lling the foll ow ing crite ria : prio r surg ery (lum pectom y or m a stect om y with or without ax illary dissec tion) and pri o r adjuv ant ant h ra cy c and t axan e-b a sed chem otherapy. Excl usi o n : P rior RT with inclusion of t h e he ar t i n t h e R T f ie lds ; histor y o f se ri o u s cardiac i lln e ss , i n c luding c o ngest ive h e ar t fa il ur e or c a rd iom yopat hy ; and po or echog enicity. E lig ib le p a tie n ts : Eс 75 Included p a ti ents : Eс 56 M e an a g e : rig h t sided BC : 5 2 ± 7 y Le ft -s id e d B C : 54 ± 8 y Fol low-up d u rati on aft e r t reat m ent : 14 mo . Lost t o fol low -u p : unclear, sam p les av ai lable for 56 out o f 75 pat ien ts S e tting : BC patients a t the Uni ver si ty ho sp it al Tur k e y BC t reat m ent : E p irubicin (100%) S e tting : C o nsecutive pat ient s d iagno se d w ith BC at th e Universit y hospi tal Belg ium BC t reat m ent : E p ir ubicin (91%) Doxorubicin (4%) Pe g ylated liposom al doxorubicin (5%) Taxan es (100%) Tr ast u zu mab ( 20%) Ra dioth e ra p y (100%) P rospective coh ort s tu d y Pro spective cohort study Erdi m , 2009 144 Erv en , 2012 142
4
MUGA Mea su ri n g poi nts for cardiotoxi ci ty: Baseline (befor e start t reatm ent), at 2, 6, 9, and 1 8 mo . ECHO/M UGA Mea su ri n g poi nts for cardiotoxi ci ty: Baseline (im m e diately b e fore th e fi rs t cycle o f chem otherapy) and af te r th e 8 th, 16 th, a n d ( 2 4 th (last ) adm ini st ra tion of chem otherapy. E v ery 3 m o . thereaf ter an d MUGA if LVE F de cr eased > 20% - - TnI Cutoff v a lu e : “Det ect abl e ” i .e., шϬ͘ϬϲŶŐ ͬŵ >ĂŶĚ шϬ͘ϬϰŶŐ ͬŵ > M e a suri n g poi nts for bi om ark er : Ever y 2 wk dur ing treatm en t cTnT Cutoff v a lu e : шϬ͘ϭ U/l Me a suri n g poi nts for bi om ark er : Imm e diately bef ore and 4h af te r chem otherapy adm ini st ra tion Sy mp to ma t ic or asym pt o ma tic dec line in LVEF> 10% t o <50% Abnorm a l LVEF o r diastoli c dysfunct ion Inclusi on: early BC which overexpress e d HER-2/ne u, r e g a rd le ss o f nodal st a tus o r tum o r size. E C G and norm a l LVE F by Dh'; шϱϬйͿ͘ Excl usi o n : P a tients with se rious m e dical illnesse s in cluding uns table a n g ina , m yocar d ia l inf a rc tion, and CHF. A n abso lut e ne ut ro ph il c o un ƚ фϭϬϬϬ ͬʅ> ͖ Ɖů Ăƚ ĞůĞ ƚ ĐŽƵŶƚ фϭϬϬϬϬϬͬ ʅ>͖ ĂďŶŽ ƌŵ Ăů bi lir ub in and t ransam inase s >2.5 × ULN . P a tien ts trea ted w ith C Y P 3 A4 induc e rs/i nhi b it o rs and/or t reat m e n t with drug s that m a y prolong the QT c. P a tients with g rade 3 QT c. E lig ib le : Eс Unk no wn Included : Eс 95 M e di an a g e : 46 ( ra ng e 28– 7 3) Fol low-u p durati on after tr eat m en t: 4 mo .Lost to fol lo w -up : (46 st opped treatm en t) Inclusi on: patients affect ed by HER- 2-ne g a ti ve adv a nced/met ast a ti c B C, not previousl y trea ted Excl usi o n :-patients young er t h an 18 and older than 75 y -h aving a hist o ry of ca rdiovascu lar dis ease, drug -u ncontrolled sys tem ic hyper ten sion, LVEF < 50% or c h ro ni c r e nal f a ilur e E lig ib le : Eс Unk no wn Included ͗Eс 20 M e di an a g e : 56 y ( rang e 34–75 y ) Fol low-up d u rati on aft e r t reat m ent : 14 mo. (24 wk t reat m ent ; medi an f oll ow -up o f 20 m o .) Lost t o fol low -u p : 0 S e tting : ne wly di agno se d pr im ar y B C patients a t a cancer cl inic in th e USA BC t reat m ent : Doxorubicin (100%) Pa clitaxel (100%) Tr as tu zu mab (100%) Lapat ini b (100%) Ra dioth e ra p y (100%) Setting : wom e n wit h chem otherapy-naïve m e tastatic BC at a g e neral ho sp it al in Ital y BC t reat m ent : E p ir ubicin (100%) Taxan es (100%) Tr ast u zu mab ( 0%) R a diot he ra p y ( 0%) P rospective coho rt s tudy P rospective coh ort s tu d y M orri s, 2011 145 Ni sti co, 2007 143
Systematic review troponin & (Nt-pro)BNP
71
Abbreviations Table 1: BC, breast cancer; BMI, BMI, body mass index; ALT, alanine aminotransferase; AST, aspartate aminotransferase; CAD, coronary artery disease; CHF, congestive heart failure; CNS, central nervous system; CREC, Cardiac Review and Evaluation Committee; ECG, electrocardiogram; ECHO, echocardiography; Hs CRP, high-sensitivity C-reactive protein; LVEF, left ventricular ejection fraction; MRI, magnetic resonance imaging; MUGA, multigated acquisition scan; NYHA, New York Heart Association; RT, radiotherapy; ULN, upper limit of normal; TnI, troponin I; TnT, troponin T
Table 2 Quality assessment
Criteria Fal lah-rad, 2011 135 Ky, 2013 139 Oni til o, 2012 137 Per ik , 2006 136 Car d in ale, 2010 134 Er di m , 20 09 144 Er ven, 2012 14 2 Kat su ra da, 2014 141 Mor ri s, 2011 145 Nak a no, 2016 140 Nistico, 200 7 143 Yu, 201 6 133 General
1. Was the research question or objective in this paper clearly stated? And was it in concordance to our research question?
Y Y Y Y Y Y No Y Y No No No
Study population/patients 2. Was the study population clearly specified and defined? (At least mentioned: age of patients; type of cancer therapy given, time since cancer therapy, and setting)
Y Y Y No Y No Y Y Y Y Y Y
3. Was the participation rate of eligible persons at least 50%?
NR Y No CD NR NR NR NR NR NR CD NR 4. Were all the subjects selected or
recruited from the same or similar populations? Were inclusion and exclusion criteria for being in the study pre-specified and applied uniformly to all participants?
CD Y Y CD Y CD Y CD CD CD CD CD
5. Were sample size justification, power description, or variance and effect estimates provided?
No Y NR Y No No No No No NA No No
6. Was loss to follow-up after baseline 20% or less?
NR Y Y Y Y Y No Y No No Y Y Exposure to cardiotoxic treatment
7. Where details of treatment described? (At least mentioned: cumulative dose of chemotherapy, type of chemotherapy, radiation therapy, left/right sided radiation therapy, and trastuzumab treatment.)
Y Y No Y Y CD No No Y No Y Y
Prognostic markers
8. For the analyses in this paper, were the biomarkers (troponin and/or (NT-pro)BNP) measured prior to the outcome being measured?
Y Y Y Y Y Y Y Y Y Y Y Y
9. Was the prognostic marker
assessed more than once over time? Y Y Y Y Y Y Y Y Y Y Y Y 10. Were the prognostic markers
(independent variables) clearly defined, valid, reliable, and implemented consistently across all study participants?
Y Y Y Y Y Y Y Y Y Y Y Y
Outcome
11. Was the timeframe sufficient so that one could reasonably expect to see an association between exposure and outcome if it existed? Follow-up was at least until the end of treatment with cardiotoxic therapies (anthracyclines, taxanes,
trastuzumab, and radiotherapy).
Y Y Y Y Y Y Y Y Y Y Y Y
12. Were the outcome measures (dependent variables) clearly defined, valid, reliable, and implemented consistently across all study participants?
Y Y No Y Y CD No CD Y CD CD Y
13. Was the outcome assessed more than once over time?
Y Y Y Y Y No Y Y Y Y Y Y
14. Were the outcome assessors blinded to the biomarker status of participants?
NR Y Y NR Y NR NR Y Y NR NR NR
15. Was appropriate analytic analysis performed and presented? Hazard ratio/Odds ratio?
No Y Y No Y No NA No No No NA No
16. For exposures that can vary in amount or level, did the study examine different levels of the exposure as related to the outcome (e.g., categories of exposure, or exposure measured as continuous variable)? (Corrected/stratified for cumulative dose of
chemotherapy/trastuzumab, or left-sided radiation.)
No No Y NA Y No CD No NA No NA No
17. Were key potential confounding variables measured and adjusted statistically for their impact on the relationship between exposure(s) and outcome(s)?(age for [NT-pro]BNP, cumulative dose
chemo/Herceptin/radiation, left-sided radiation, age at time of diagnosis, multivariate model.)
NA No CD NA Y NA NA NA NA NA NA NA
Total scorea + ++ ++ + ++ -/+ + + + -/+ + +
CD, cannot determine; NA, not applicable; NR, not reported; [NT-pro]BNP: brain natriuretic peptide (BNP) or the N-terminal of BNP (NT-proBNP); Y, yes
Systematic review troponin & (Nt-pro)BNP
73
Prognostic value of troponin
In all 12 included studies, troponin was evaluated as a biomarker to evaluate the occurrence of cardiac dysfunction: troponin I in eight studies and troponin T in four studies (Table 1). The median prevalence of troponin levels above the cut-off value was 14% (range 0%–67%), which did not vary by of the type (I or T) or of the cut-off value used. The weighted prevalence of cardiac dysfunction was 17.8% (range 0%–47%), and was independent of the method by which it was assessed (echocardiography, MUGA, or MRI). In three smaller studies, no cardiac dysfunction was observed.140, 142, 143 Median
follow-up periods in the 12 included studies ranged from 0 to 14 months.
The pooled OR for cardiac dysfunction in patients with elevated troponin was 3.14 (95% CI 0.73–13.50) when compared with the odds for patients with normal troponin levels; however, the I2 was 85%, indicating statistically significant
heterogeneity (Figure 2). Due to the small number of studies, no subgroup analysis was performed by follow-up duration. One study was identified as an outlier.134 The
heterogeneity decreased to 46% after it was excluded, and the pooled OR decreased to 1.72 (95%CI 0.66-4.45). This outlier study of Cardinale and coworkers was the largest study (n = 251) with the longest follow-up after end of trastuzumab therapy (median 14 and range 1–79 months)134, and showed an association between troponin and
cardiotoxicity (OR 33.34, 95%CI 13.29–78.71). This study did not differ from the other included studies by the type of breast cancer treatment, cutoff value of the biomarker, choice of reference standards, prevalence of cardiac dysfunction, or bias in our quality assessment.
Two studies 134, 139 reported Hazard Ratios, indicating that cardiac dysfunction typically
manifested significantly earlier after a positive troponin test result compared with a negative troponin test result (HR 1.36, 95% CI 1.07–1.73; HR 17.6, 95% CI 8.85–35; respectively).
Figure 2 All studies investigating troponin as a prognostic biomarker for the occurrence of cardiac dysfunction following cardiotoxic therapies for breast cancer
a FU (follow-up) is defined as the median number of months after the end of treatment b estimate of OR
Prognostic value of NT-pro BNP
Seven studies, which included 279 patients in total, investigated (NT-pro)BNP as a biomarker for predicting cardiac dysfunction in patients receiving cardiotoxic breast cancer therapies. NT-proBNP was used in four studies135, 136, 139, 141 and BNP was used in
three.133, 137, 140 One study used a cutoff value of 297 pg/mL135, another did not mention
the cutoff used141, and the remainder used cutoff values of 100–200 pg/mL.133, 136, 137,
139, 140 In six studies, patients were followed until the end of treatment, but one study133
had a longer follow-up period (Table 1). The weighted prevalence of cardiac dysfunction was 21.7% (range 3%–47%) in these seven studies. Ten cardiotoxicity events were observed in the study by Fallah-Rad, but there were no NT-proBNP levels above the cut-off value.135 Unfortunately, two studies lacked key information: Ky et. al.
did not present the number of events for women with increased and normal
NT-Prevalence/ cum. Incidence (%)
FU* Measurement
Biomarker OR (95%CI) HR (95%CI) Odds Ratio IV, Random, 95% CI Events Total Events Total
Outcome assessed only during treatment
Fallah-Rad 2011 (40) 0 0 10 32 31.3% 0 3 weeks after AC &
during trastuzumab N.E.
-Katsurada 2014 (46) 7 9 2 10 47.4% 0 During AC and during
trastuzmab 14 (1.54 -127.23) -Ky 2013 (44) 11 26 12 52 29.5% 0 just after AC &
during trastuzumab 2.44 (0.89-6.72) 1.36 (1.07-1.73) Nakano 2016 (45) 0 0 0 9 0,00% 0 During AC and during
trastuzmab N.E.
-Onitilio 2012 (42) 0 0 14 49 28.6% 0 Unclear, probably
during trastzumab N.E.
-Perik 2006 (41) 0 1 4 15 25% 0 During trastuzmab &
paclitaxel 0.85 (0.03-25.05)
-Outcome during and after treatment Cardinale 2010 (39) 26 36 16 215 16.7% 14 Unclear, during trastuzmab and follow-up 33.34 (13.29-78.71) 17.6 (8.85-35)
Erdim 2009 (49) 9 26 4 15 31.7% 12Before and 72h after
AC 1.46 (0.36-5.91)
-Erven 2012 (47) 0 0 0 56 0% 14
First and last day of RT(Unclear time-interval between CT first day of RT)
N.E.
-Morris 2011 (50) 6 64 6 31 12.6% 4 During AC and during
trastuzmab 0.43 (0.13-1.47)
-Nistico 2007 (48) 0 0 0 20 0% 20Before and after AC
and P N.E.
-Yu 2016 (38) 0 3 2 64 3% ±14 During trastuzmab 3.57 (0.14-89.53)
-3.14 (0.73- 13.50) Total
heterogeneity: Tau²=2.93; Chi²=39.03, df=6 (P<0.00001); I²=85%
Normal troponin Increased
troponin
Test for overall effect: Z=1.54 (P=0.12)
Systematic review troponin & (Nt-pro)BNP
75
proBNP levels separately139, while Katsurada et al. did not present either the numbers
of women with increased NT-proBNP levels or the cutoff levels used.141The pooled OR
for cardiac dysfunction in patients with elevated (NTpro)BNP was 2.07 (95% CI 0.50-8.59) (figure 3). Heterogeneity was low (I2=0%). No subgroup analyses were performed
due to the low number of studies. Finally, although one study estimated the time to occurrence of cardiac dysfunction during treatment, it failed to show any association with a positive (NT-pro)BNP (HR: 0.89; 95% CI: 0.59–1.35).139
Figure 3 All studies investigating (NTpro)BNP as a prognostic biomarker for the occurrence of cardiac dysfunction following cardiotoxic therapies for breast cancer
a FU (follow-up) is defined as the median number of months after the end of treatment; † NR: Not reported in the article; (NT-pro)BNP: brain natriuretic peptide (BNP) or the N-terminal of BNP (NT-pro)BNP
Discussion
To our knowledge, this is the first systematic review evaluating the association of commonly available cardiac biomarkers with cardiac dysfunction in women treated for breast cancer. We found a prevalence of 17.4% (range 0-47.4%) for cardiac dysfunction in women treated for breast cancer. This is high in comparison to prevalence reported in women without a history of breast cancer. In a recent study examining cardiac dysfunction in 350 women who survived breast cancer and 350 women without cancer, cardiac dysfunction was defined as a LVEF<54%. In this study prevalences of 15.3% and 7% respectively, were reported.146
In this review, we found insufficient evidence of an association between troponin or (NTpro)BNP measured during treatment and cardiac dysfunction. When time to occurrence of cardiac dysfunction was taken into account, cardiac dysfunction
manifested earlier in women with positive troponin results during treatment than in women with normal troponin results.134, 139 However, there were few events and
confidence intervals were wide. Because of the small number of studies, small number of patients, and short follow-up durations, the results should be interpreted with caution.
In the included studies, the prevalence of raised troponin levels (median 14%, range 0%–67%) was remarkably high when compared with the prevalence in the general population (2.5%).147 Because the release of troponin is related to cell
apoptosis and reduced cardiac muscle contractility147, increased levels may indicate
damage to the cardiac muscle during treatment that could lead to cardiac dysfunction. However, in this meta-analysis we did not show a statistically significant relationship between increased troponin levels and cardiac dysfunction, possibly due to the short follow-up of the included studies. (NT-pro)BNP has been associated with increased wall tension in the heart.148 Given that (NT-pro)BNP levels were increased in 0%–80% of the
women in the included studies, it is possible that increased wall tension is present without resulting in clinically demonstrable cardiac dysfunction.
Specific biomarker kinetics might have influenced the results. In the included studies, a positive troponin or (NT-pro)BNP was defined as the presence of at least one increased measurement. However, in the study by Cardinale et al., it was reported that troponin was prognostic for a reduction in LVEF and poor cardiac outcome in patients with persistent (>1 month) increased troponin values.149 While a single increased value
might be either due to reversible damage or to measurement error, a more prolonged elevation may be associated with irreversible cardiomyocyte damage.150 It is possible
that the rate of troponin increase during treatment151, or the relative increase in
troponin from baseline to the end of treatment, has more prognostic value.142 For
(NT-pro)BNP, only persistently increased levels during treatment152 or a lower cutoff
point153 might be associated with cardiac dysfunction.
All included studies measured LVEF as primary endpoint for cardiotoxicity, even though a decreased LVEF is a late sign because of the significant functional reserve of the heart. It could be that women with increased troponin or (NT-pro)BNP levels had pre-clinical cardiac dysfunction for which the used imaging techniques (echocardiography, MRI, or MUGA) were insufficiently sensitive. Strain measurements that are currently in development may be able to detect dysfunction caused by damage from cardiotoxic therapies at an earlier stage.140, 142, 143
Not all possible risk factors for cardiac dysfunction were considered in the included studies. Factors that may also be considered are the cumulative anthracycline
Systematic review troponin & (Nt-pro)BNP
77
dose, smoking, and other cardiovascular risk factors.69 The included studies gave
insufficient information for adjusted analysis. Therefore the lack of an association found, should be interpreted with caution.
Strengths and limitations
A strength of this systematic review was the extensive literature search, which minimized the risk of missing relevant publications. To avoid missing studies, we did not use a prognostic filter to identify prognostic studies only.154 In addition, no relation
was observed between quality and the prognostic value of either biomarker. However, there were some important limitations. To date, for example, it has proven difficult to standardize LVEF thresholds, and the classification for cardiotoxicity remains a topic of debate.67 In this review, differences in the definitions of cardiac dysfunction between
studies used were revealed. This may have led to under- or overestimation of the prognostic value of the biomarkers. Unfortunately, we also could not investigate the effect of different cardiotoxic treatments on the association between biomarker levels and cardiotoxicity, because this was not analyzed in the included studies. Furthermore, we could not perform a separate analysis of women with either chemotherapy or radiotherapy as most studies included only women with both treatments. Thus, we cannot comment specifically on the value of the cardiac biomarkers in predicting the occurrence of radiation-induced heart disease.
Furthermore, the ORs calculated in eight of the studies did not take renal disease into account, while another four studies excluded patients with renal disease.134, 136, 140, 143
This is especially important because moderate reductions in the estimated glomerular filtration rate can increase serum troponin155 or (NT-pro)BNP156 levels.
Conclusion
Recognition of cardiac dysfunction is difficult in women treated for breast cancer, not least because symptoms may overlap with other late symptoms of cancer. Therefore, it is important to know who will develop cardiac dysfunction in the long-term. Based on current evidence, this review and analysis showed that the occurrence of cardiac dysfunction cannot be predicted by troponin or (NT-pro)BNP. However, because of the small number of studies, small number of patients, and short follow-up durations, the results should be interpreted with caution. Given the multifactorial nature of cardiovascular disease, it seems only prudent to include additional risk factors in any clinical prediction model. Moving forward, more robust studies with both serial measures and lower thresholds of biomarkers, as well as longer follow-up of cardiac
function, are needed to evaluate the predictive value of these markers in a clinical prediction model.
Acknowledgements
We thank Dr Robert Sykes (www.doctored.org.uk) for his editorial assistance.
Disclosure
