Page 27 - Cardiac testing for coronary artery disease in potential kidney transplant recipients

ability was also observed in the DSE results, which may be due to transthoracic echocardiography was also found to offer high speci-
unevenness in local expertise to interpret test results across differ- ficity (95% to 96%) but low sensitivity (31% to 33%). Mitral
ent studies. annular calcification on echocardiography was studied in the same
Significant heterogeneity was present, which could not be ex- population (Sharma 2005) and this had higher sensitivity (61%)
plained by differences in reference threshold and partial verifica- at the expense of lower specificity (72%). The marked variability
tion.Clearly, otherfactorsmayhave contributedtothe clinical het- in sensitivity and specificity of resting ECG confirms that it has
erogeneity in the results. These include differences in study popu- no role in triaging patients for CAD. Notwithstanding the limita-
lation characteristics (such as prevalence of chest pain, prevalence tions posed by few numbers of studies and participants presented,
of diabetes) and test application (diagnostic test threshold, criteria EBCT and calcium scoring methods also appeared to have limited
for positive test, choice of stress agent and stress protocol, and op- utility in evaluating the cardiac health of potential kidney trans-
erator variability). Limited data from the small numbers of studies plant recipients. This is reflected in the fact that the optimal test
and participants meant that we were unable to perform subgroup performance of EBCT in the only study identified (Rosario 2010)
analyses of the effect of DM and prevalence of angina and IHD was a calcium score of 1330.72, which is higher than the usual
on diagnostic performance. Other differences across studies may threshold used in the general population. There is also a theoret-
also have played a role. One possible factor was sex of the partic- ical disadvantage of calcium scoring methods in potential kidney
ipants. One study (Gowdak 2010) showed that among patients transplant recipients due to the increased prevalence of arterial
with diabetes, MPS test performance was influenced by the sex of calcification in patients with CKD, arising from metabolic bone
participants; sensitivity was lower in women (females 56%; males disease. Although published studies were not identified in this re-
65%). Accuracy data based on sex was not reported in any of the view, other tests that might be expected to have limited application
included studies. Hence, we were unable to determine if the sex in the pre-transplant setting for patients with CKD include CT
of the participant influenced diagnostic accuracy. coronary angiography (exposure to nephrotoxic IV contrast that
Generally, methodological quality was poorly reported. Method- could adversely affect any residual kidney function) and magnetic
ological quality scoring was based on published reports and addi- resonance imaging (MRI) or angiography (risk of gadolinium in-
tional data provided from correspondence with study authors. Un- duced nephrogenic systemic fibrosis).
clearreportingof certainmethodological issuesmaynotnecessarily
indicate poor study design; restrictions imposed by journal word
limits, or editing, may have precluded reporting all QUADAS Strengths and weaknesses of the review
items. Several methodological quality items were reported less fre-
A strength of this review was the sensitive electronic search strategy
quently than others. These included blinding of reference tests (7/
developed that identified both published and unpublished stud-
25 not reported), blinding of index tests (8/25 not reported), and
ies. Our search strategy excluded search filters for diagnostic terms
acceptable delay between tests (12/25 not reported). In addition
because they have limited utility (Leeflang 2006; Ritchie 2007).
to the studies where blinding of reference and index tests was un-
Other strengths included our analytic approach of combining re-
certain, 3/25 studies reported no blinding of the reference stan-
sults from studies with similar methodological characteristics and
dard; one study reported no blinding of the index test. Therefore,
applying the HSROC model to conduct our analysis. The hier-
lack of blinding may have affected our results; the overall effect of
archical modelling strategy accounted for sampling variability in
unblinded reporting of reference and index tests is generally leads
estimates of sensitivity and specificity (and their correlations) in
to overestimation of diagnostic accuracy (Leeflang 2006).
each study when estimating the random effects. This resulted in
We did not find any studies that investigated cardiopulmonary ex-
accuracy estimates that provided better assessments of underlying
ercise testing, CT coronary angiography, magnetic resonance an-
common log odds ratios (Macaskill 2003). To ensure that findings
giography or cardiac magnetic resonance imaging. Fewer than five
were generalisable, we included only studies that investigated only
studies were found for each of EBCT, ECG, conventional echocar-
potential kidney transplant recipients. We excluded studies that
diography, exercise ventriculography, DSF and CIMT. This pre-
enrolled participants with ESKD because it could be reasonably
cluded any further meaningful comparisons other than that be-
anticipated that inclusion of unselected dialysis patients would
tween DSE and MPS. DSF and exercise ventriculography are sel-
modify expected differences in underlying prevalence of CAD,
dom used for CAD screening. Nevertheless, results from studies
and the presence and severity of other comorbidities, as well as
identified for this review (DSF: sensitivity 78%, specificity 66%;
differences in clinical rationales for testing. By concentrating on
exercise ventriculography: sensitivity 50%, specificity 67%) sug-
potential transplant candidates our findings may not be general-
gest that neither DSF nor exercise ventriculography were likely
isable to dialysis or CKD patients who would not benefit from
to be superior to DSE or MPS. EST appeared to have offer high
transplantation. Our vigilance in contacting authors to obtain data
specificity (91%) but poor sensitivity (36%) in the one study
missing or not reported in studies was rewarded by a satisfying
that included a sufficient number of participants (Sharma 2005).
number of responses.
Resting wall motion abnormality detected on traditional resting
Significant heterogeneity was present among studies that investi-
Cardiac testing for coronary artery disease in potential kidney transplant recipients (Review) 25
Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.