This month’s eJournal club concerns biomarkers of AKI. Therehas been considerable interest in developing novel biomarkers of AKI and CKD
and a lot of effort has been focused on novel biomarkers such as NGAL, KIM-1
and IL-18. The TRIBE-AKI consortium has published a number of well-designed
studies investigating novel and traditional biomarkers. The reason that these
biomarkers are desired is because creatinine elevations tend to occur
relatively late in the course of AKI and therefore, the sense is that this is
too late to initiate therapies which may prevent progression of AKI. However,
the results of these biomarker studies have been relatively disappointing and
they have not yet entered regular clinical practice.
This month in CJASN, a study was published looking at
the performance of post-operative albuminuria as a biomarker of AKI in patients
following cardiac surgery. The highest quintile of albuminuria was associated
with a RR of 2.97 for AKI relative to the lowest quintile. While this appears
good and the AUC for a model including albuminuria to predict AKI was 0.81, the
majority of patients with albuminuria did not develop AKI and the model missed
a significant number of patients with AKI. Still, when you combine this with
other studies showing that the urinalysis is an excellent predictor of outcome
in patients with AKI, older biomarkers are not looking so bad after all.
Perhaps we will be able to come up with a combination of biomarkers which will
allow us to better predict those patients at greater risk of AKI. To me, it
seems that the bigger issue is low sensitivity rather than low specificity. I
would rather have a model which will allow me to rapidly rule out those who
will not develop AKI than one that will misclassify patients into a low risk
group.
the performance of post-operative albuminuria as a biomarker of AKI in patients
following cardiac surgery. The highest quintile of albuminuria was associated
with a RR of 2.97 for AKI relative to the lowest quintile. While this appears
good and the AUC for a model including albuminuria to predict AKI was 0.81, the
majority of patients with albuminuria did not develop AKI and the model missed
a significant number of patients with AKI. Still, when you combine this with
other studies showing that the urinalysis is an excellent predictor of outcome
in patients with AKI, older biomarkers are not looking so bad after all.
Perhaps we will be able to come up with a combination of biomarkers which will
allow us to better predict those patients at greater risk of AKI. To me, it
seems that the bigger issue is low sensitivity rather than low specificity. I
would rather have a model which will allow me to rapidly rule out those who
will not develop AKI than one that will misclassify patients into a low risk
group.
It was interesting in this study that ACR was not a good
predictor of AKI – the absolute level of albumin performed better. This is at odds
with the majority of studies which suggest that albumin should be corrected for
creatinine level. This is possibly due to the large variation in creatinine
generation in patients in the ICU – although relatively constant under normal
circumstances, the amount of creatinine produced daily changes rapidly in sick
patients – as was evidenced by a recent study of creatinine excretion in
patients on CRRT.
predictor of AKI – the absolute level of albumin performed better. This is at odds
with the majority of studies which suggest that albumin should be corrected for
creatinine level. This is possibly due to the large variation in creatinine
generation in patients in the ICU – although relatively constant under normal
circumstances, the amount of creatinine produced daily changes rapidly in sick
patients – as was evidenced by a recent study of creatinine excretion in
patients on CRRT.