Comparison of serum and urinary biomarker panels with albumin/creatinine ratio in the prediction of renal function decline in type 1 diabetes
DIABETOLOGIA | JANUARY 08, 2020
Colombo M, McGurnaghan SJ, Blackbourn LAK, Dalton RN, Dunger D, Bell S, Petrie JR, Green F, MacRury S, McKnight JA, Chalmers J, Collier A, McKeigue PM and Colhoun HM.
The key findings from this study across the range of CKD stages were that serum biomarkers improve the prediction of future eGFR and progression to <30 ml min−1 [1.73 m]−2 beyond baseline eGFR. As in our past studies [11, 21, 22], a large number of the biomarkers evaluated showed highly significant associations with eGFR and its decline. However, almost all of the predictive information could be obtained using just a few of these intercorrelated biomarkers. Here, we found KIM-1 measured by SIMOA assay and either CD27 or TNFR1 to contain most of the predictive information. CD27 and TNFR1 are very highly correlated (r = 0.80) and are both members of the TNF superfamily and therefore likely interchangeable as predictive biomarkers. Beyond these, there is some predictive information in syndecan-1, clusterin or α-1-microglobulin.
An important emphasis of the study reported here was a comparison of biomarkers with the performance of the widely accepted measurement ACR. In this regard, we found that a parsimonious panel of serum biomarkers performed better than ACR for predicting eGFR and eGFR progression to <30 ml min−1 [1.73 m]−2. Indeed, serum biomarkers contained almost double the prediction information (measured in bits) than ACR for progression to <30 ml min−1 [1.73 m]−2. The serum biomarkers were predictive of final eGFR in those with starting eGFR above 90 ml min−1 [1.73 m]−2, suggesting that they rise before ACR does and may be more sensitive to early renal damage. Of note, the urinary biomarkers did not consistently outperform ACR.
These data challenge the place of ACR in our clinical management of people with diabetes. Most clinical guidelines now recommend annual measurement of ACR in people with diabetes. Our data suggest that measuring just KIM-1 and TNFR1 or CD27 in the same sample might obviate the need for doing a urine sample collection and then measuring ACR. Collecting urine samples is time intensive in outpatient clinics and many people miss out on annual screening because they do not bring or cannot produce a urine sample. As we note above, while in Scotland we achieve high rates of annual eGFR measurement, ACR capture rates are much lower . Additional differences in reporting rates for ACR than eGFR exist elsewhere, e.g. in the National Diabetes Audit in England and Wales and in reports from Denmark [23, 24]. Clearly, further validation is required before advocating replacing ACR. Longitudinal analyses of repeat measurements of these biomarkers, for example, are needed to establish their variability and thresholds for specific risk levels warranting clinical action (akin to the microalbuminuria thresholds).
Replication in different cohorts across the range of eGFR, using a variety of assay methods, and testing for prediction of ESRD are also needed. Moreover, at the present time these biomarkers are much more expensive to assay than ACR and are not routinely available on many commonly used clinically certified assay platforms. Some may argue that ACR is also measured because it is a biomarker of more widespread vascular disease and mortality independently of renal function: this should also be shown for candidate replacement serum biomarkers. Nonetheless, at the very least our data suggest that further studies to evaluate replacing urinary ACR with serum biomarkers are warranted, given the logistic and predictive advantages they may offer.
Serum biomarkers might be useful on top of ACR rather than instead of it. In regard to the latter, it is important to note that these biomarkers also added to the prediction of final eGFR or progression even when ACR was included in the model. Furthermore, they also predicted progression to <30 ml min−1 [1.73 m]−2 in those normoalbuminuric or microalbuminuric at baseline, and did so independently of ACR. This suggests they may be useful even when ACR has already been measured and found not to be in the macroalbuminuric range. Once macroalbuminuria has been detected, however, they do not improve prediction further. Such people with macroalbuminuria would in any case be managed uniformly as a high-risk group, and it is questionable whether any further risk discrimination among them would alter clinical practice.
Our study has used eGFR and its progression to <30 and <45 ml min−1 [1.73 m]−2 as endpoints. Of course, as we recently reviewed, creatinine-based eGFR is itself an imperfect measure of underlying GFR, especially at early stages of renal function decline . Ideally, biomarker studies would also evaluate prediction of harder endpoints such as ESRD. However, to evaluate prediction from early stages of renal function decline to ESRD requires large unselected cohorts of people with type 1 diabetes to be followed for many years for sufficient ESRD endpoints to accrue. Such studies are logistically challenging and usually have very few ESRD endpoints in those with eGFR >60 ml min−1 [1.73 m]−2 at baseline [25, 26]. Furthermore, it requires that we generalise from samples taken several decades ago to the contemporary state of diabetes. Therefore, studies such as ours with intermediate endpoints are also needed. It should be noted that the imprecision with which eGFR quantifies underlying GFR means that power to detect biomarker associations is reduced rather than false positive associations being detected. Our study will accrue ESRD endpoints as we continue follow-up, eventually allowing associations with this endpoint to be confirmed. Further analyses will also attempt to disentangle the contribution of prediction of intervening acute renal failure events to overall prediction of decline as events accrue. Another limitation of our study is that we were not able to assess incremental prediction on top of the validated kidney failure risk equation , as calcium and phosphate data were not available in most participants.
A key strength of our study was the deliberate use of samples from a wide range of starting eGFR. Another key strength of our study is the use of advanced statistical methods that avoid over-optimistic assessments of prediction. These include cross-validation and use of penalised models that account for the high number of analytes being evaluated and can better handle correlations between predictors.
In summary, a parsimonious panel of serum biomarkers measurable along with creatinine may outperform ACR for predicting renal disease progression from early CKD stages in type 1 diabetes, and with further development might obviate the need for urine testing.
Share this page