Table 2.
The Biomarker Surrogacy Evaluation Schema
| Criterion | Evaluation and Scores |
|---|---|
| Study design | 3 points |
| 0: Biologic plausibility and lower-quality clinical studies | Multiple clinical trials measured both proteinuria and recorded ESRD outcomes. These trials assessed the effects of different drugs, including RAAS inhibition, low-protein diet, and corticosteroids. |
| 1: At least two high-quality prospective observational studies | |
| 2: At least two high-quality adequately powered RCTs measuring S and T | |
| 3: At least three high-quality adequately powered RCTs measuring S and T | |
| Target outcome | 3 points |
| 0: Reversible disease-centered biomarker of harm | ESRD is a patient-centered clinical end point of irreversible organ morbidity. |
| 1: Irreversible disease-centered biomarker of harm | |
| 2: Patient-centered end point of reversible organ morbidity or burden of disease or clinical harm | |
| 3: Patient-centered clinical end point of irreversible organ morbidity or burden of disease or clinical harm or death | |
| Statistical strength of biomarker for target | 2.5 points |
| 0: Poor | In a meta-analysis of 21 randomized, controlled trials, the R2 from a weighted regression analysis was 0.46 (44). The same meta-analysis showed that the lower boundary of the 95% confidence interval intersects the horizontal axis at approximately 30% (44). This is the STEP. Under the assumption that the percentage change in albuminuria ranges between −70% and +20% (values derived from refs. 43 and 44), the STEP equals 0.33. Analyses of individual patient data of diabetic and nondiabetic CKD trials show in all trials that, within trials, the R2 exceeds 0.6 (45,54,64,65). The statistical strength, thus, ranges between good and excellent. |
| 1: Fair: R2trial≥0.2, STEP≥0.1, and R2ind≥0.2 | |
| 2: Good: R2trial≥0.4, STEP≥0.2, and R2ind≥0.4 | |
| 3: Excellent: R2trial≥0.6, STEP≥0.3, and R2ind≥0.6 | |
| Generalizability of biomarker-target relationship; clinical evidence across different risk populations and pharmacologic evidence across different drug class mechanisms | 3 points |
| 0: No clinical or pharmacologic evidence | Albuminuria predicts renal outcome across different populations and CKD disease etiologies. Therapy–induced short-term changes in albuminuria predict renal outcome across different populations, CKD disease etiologies, and interventions (RAAS intervention, intensive glucose control, anti-inflammatory drugs, or low-protein diet). |
| 1: Clinical or pharmacologic evidence | |
| 2: Clinical and pharmacologic evidence | |
| 3: Consistent clinical RCT and pharmacologic RCT evidence |
The Biomarker Surrogacy Evaluation Schema consists of four domains: (1) study design, (2) target outcome, (3) statistical strength, and (4) generalizability. The overall summed score ranges from 0 to 12. The score is converted to a surrogate/outcome level of evidence, with the most convincing (level 1) being a score of 12 and the least convincing (level 5) being a score of 0, 1, or 2. Intermediates are level 2 (scores 9–11), level 3 (scores 6–8), and level 4 (scores 3–5). RAAS, renin-angiotensin-aldosterone system; RCT, randomized controlled trial; S, surrogate (i.e., albuminuria); T, target outcome (i.e., ESRD); STEP, surrogate threshold effect proportion.