Measuring GFR: a systematic review

Background: No comprehensive systematic review of the accuracy of glomerular filtration rate (GFR) measurement methods using renal inulin clearance as reference has been published.

Study design: Systematic review with meta-analysis of cross-sectional diagnostic studies.

Setting & population: Published original studies and systematic reviews in any population.

Selection criteria for studies: Index and reference measurements conducted within 48 hours; at least 15 participants studied; GFR markers measured in plasma or urine; plasma clearance calculation algorithm verified in another study; tubular secretion of creatinine had not been blocked by medicines.

Index tests: Endogenous creatinine clearance; renal or plasma clearance of chromium 51-labeled ethylenediaminetetraacetic acid (51Cr-EDTA), diethylenetriaminepentaacetic acid (DTPA), iohexol, and iothalamate; and plasma clearance of inulin.

Reference test: Renal inulin clearance measured under continuous inulin infusion and urine collection.

Results: Mean bias <10%, median bias <5%, the proportion of errors in the index measurements that did not exceed 30% (P30) ≥80%, and P10 ≥50% were set as requirements for sufficient accuracy. Based on the GRADE (Grading of Recommendations Assessment, Development and Evaluation) approach, the quality of evidence across studies was rated for each index method. Renal clearance of iothalamate measured GFR with sufficient accuracy (strong evidence). Renal and plasma clearance of 51Cr-EDTA and plasma clearance of iohexol were sufficiently accurate to measure GFR (moderately strong evidence). Renal clearance of DTPA, renal clearance of iohexol, and plasma clearance of inulin had sufficient accuracy (limited evidence). Endogenous creatinine clearance was an inaccurate method (strong evidence), as was plasma clearance of DTPA (limited evidence). The evidence to determine the accuracy of plasma iothalamate clearance was insufficient. With the exception of plasma clearance of inulin, only renal clearance methods had P30 >90%.

Limitations: The included studies were few and most were old and small, which may limit generalizability. Requirements for sufficient accuracy may depend on clinical setting.

Conclusions: At least moderately strong evidence suggests that renal clearance of 51Cr-EDTA or iothalamate and plasma clearance of 51Cr-EDTA or iohexol are sufficiently accurate methods to measure GFR.