Evidence for paper review

Paper summary (from Krochmal et al., 2017/2018)

The authors review urinary peptidomics (primarily CE‑MS based) for kidney-disease biomarker discovery, patient stratification, drug-response prediction, and improving translational value of animal models. They present CKD273 as the most advanced classifier (discovered in 609 individuals; validated multicenter) and discuss protease inference (PROTEASIX/TopFIND/PROSPER) for generating mechanistic hypotheses (e.g., MMP‑2/MMP‑9 downregulation in diabetic nephropathy validated by zymography in mice). The review notes reproducibility improvements in CE‑MS but highlights regulatory and clinical-endpoint barriers to implementation and discloses industry links (Mosaiques Diagnostics). Review: Urinary peptidomics (Krochmal et al.)

Methodological strengths & weaknesses (evidence-linked)

• Strength — sensitive, top-down peptidomics: CE‑MS separates endogenous small peptides without digestion, enabling detection of native PTMs and low‑MW peptides relevant to kidney biology CE‑MS method strengths
• Weakness — sample handling & inter-lab variability: authors and methods reviews both highlight need for standardized pre-analytics and blinded multicenter validation to avoid false positives and overfitting CE‑MS preanalytical variability
• Bias/risk of COI: lead author affiliations include Mosaiques Diagnostics; the review discloses that Mischak is founder/co-owner and Krochmal employee — a nontrivial potential influence on framing and emphasis (CKD273 commercialization pathways discussed) COI disclosure in review

Key claims vs supporting evidence (critical)

1. CKD273 prognostic utility: claimed strong — supported by discovery (n=609) and multicenter validations (PRIORITY ancillary and multicenter validation showing high AUCs reported). Evidence: discovery and several validation studies show improved prognostic accuracy over albuminuria/eGFR; FDA Letter of Support for CKD273 early-phase use reported by authors. Caveat: many validation datasets come from groups involved in CKD273 development — independent external replication remains limited. CKD273 evidence summary
2. Protease inference to suggest drug targets: authors present PROTEASIX predictions and validation (MMP‑2/9 decreased in DN mouse). This is a promising path from peptidome → upstream biology but inference depends on prediction tool assumptions and peptide sequence coverage; only limited protease validations so far. Protease inference + validation
3. Animal model translation: authors show peptidomics-based 'humanized readout' improving translatability (21 ortholog peptides in DN mouse → classifier with good sensitivity/specificity), but other models (ZDF rats) showed limited CKD comparability — translation is model-dependent and not universal. Translatability evidence

Primary blindspots, limitations and possible biases

• Conflict of interest and commercialization bias: Mosaiques involvement is explicit — increases risk of emphasis on CKD273 readiness; independent replication and data sharing safeguards are essential to mitigate vendor-driven framing COI disclosure
• Selective validation and circularity risk: many validations use cohorts generated/curated by overlapping groups; risk of optimistic performance estimates unless large independent, pre-registered multicenter validations are performed (the PRIORITY trial helps, but independent funder-led replications remain limited). See CE‑MS review on the necessity of blinded, large validations Validation & standardization caution
• Limited public raw-data sharing: the review cites many datasets but does not supply centralized raw-data accessions; this impedes fully independent reanalysis and meta-analyses (authors point to a Human Urinary Proteome database but raw deposition practices vary), limiting reproducibility assessment Data availability note

Practical takeaways for researchers & drug developers

• Use urinary peptidomics (CE‑MS) for early biomarker discovery and molecular stratification — especially to enrich trial populations (risk stratification) and reduce N needed for trials, but require prespecified analytic plans and independent validation cohorts to avoid inflation of effect sizes Clinical trial enrichment note
• Translate via humanized classifiers back‑translated to animal models to improve model selection and increase confidence in target biology — but always confirm mechanisms in orthogonal assays (e.g., protease assays, zymography, tissue expression) Humanized readouts for translation
• Adopt open-data deposition and blinded, multi-center validations to address COI and biases before clinical deployment.

Concrete reproducibility checklist (recommended)

1. Pre-register biomarker discovery & validation plan (cohort definitions, endpoints, statistical tests).
2. Deposit raw CE‑MS files and metadata (sample collection, storage, containers, processing) to a public repository.
3. Use independent blinded test cohorts assembled by external groups (no overlapping sample collection teams).
4. Provide QC metrics: internal standard peptides, normalization peptides, inter-run CVs, peptide identification FDRs.
5. Complement peptidomic predictions with orthogonal validation (tissue proteomics, activity assays, functional models).

What would change my view (falsification tests)

• Large, independent, prospective multicenter trials showing CKD273 fails to predict CKD progression after proper preregistration and blinded assessment would falsify current clinical claims.
• Independent reanalysis of raw CE‑MS data revealing major pre-analytic confounders (e.g., container effects, undetected hematuria) driving classifier performance would undercut conclusions.
• Failure to replicate protease predictions (e.g., MMP changes) in human kidney tissue or independent assays would downgrade mechanistic claims.

Quick references (key sources cited in review)

• Primary review: Krochmal et al., Urinary peptidomics in kidney disease and drug research (2017/2018) — core claims, CKD273, protease inference and translational examples. Krochmal et al., review
• CE‑MS methods & standardization review: detailed workflow, limits, and standardization needs. CE‑MS technical review