Open questions — Chronic kidney disease¶
Last curated: 2026-09-02 (revised during independent audit, then extended during a literature-deepening pass that widened the evidence base from 130 to 306 distinct records). Stable IDs are retained across future sweeps; answered questions should be closed explicitly rather than silently deleted.
Tier 1 questions could change practice and are designable now. Tier 2 questions require enabling measurement, longer follow-up or a prior answer.
Tier 1 — practice-changing and designable now¶
OQ-1. What is the optimal sequence and combination of the four disease-modifying classes in diabetic CKD?¶
RAS blockade, SGLT2 inhibition, finerenone and semaglutide each have kidney-outcome evidence, but lifetime combination gains are modelled rather than randomized (PMIDs: 11565518, 30990260, 33264825, 38785209, 37952217). Narrowed, not answered (audit 2026-09-02): CONFIDENCE randomized 779 people with type 2 diabetes and CKD to finerenone, empagliflozin or both, and combination therapy reduced uACR 29% more than finerenone alone and 32% more than empagliflozin alone at day 180 (PMID 40470996); a prespecified secondary analysis found combination therapy did not mitigate hyperkalaemia relative to finerenone alone (PMID 41493296). This covers two of four classes on a surrogate endpoint over six months. A pragmatic factorial or strategy trial powered for kidney failure and death, comparing sequence, discontinuation, potassium, cost and absolute benefit, remains undone.
OQ-2. Does finerenone improve hard kidney outcomes in non-diabetic CKD?¶
Largely answered (audit 2026-09-02). FIDELIO/FIDELITY established benefit in type 2 diabetes (PMIDs: 33264825, 35023547). FIND-CKD has since reported: in 1,584 adults with CKD and no diabetes, finerenone slowed total eGFR slope by 0.7 mL/min/1.73 m²/year (95% CI 0.3–1.1) over 32 months and lowered a composite kidney-or-cardiovascular outcome (HR 0.77, 95% CI 0.60–0.99), with hyperkalaemia in 17.0% versus 13.3% (PMID 42246672). The INFINITY pooled analysis of FIDELIO, FIGARO and FIND-CKD (14,574 participants) found a kidney composite HR of 0.76 (0.68–0.86), kidney failure alone 0.85 (0.74–0.99), and all-cause death 0.88 (0.79–0.99), consistent irrespective of glycaemic status and aetiology (PMID 42248158). What remains open: FIND-CKD's primary endpoint was eGFR slope, and its kidney-only composite crossed 1.0 (HR 0.78, 0.60–1.01), so a hard kidney-failure benefit in non-diabetic CKD is supported by pooling rather than demonstrated in a dedicated event-driven trial. The glomerular-disease signal is a prespecified exploratory subgroup of 903 participants (PMID 42246414), and the type 1 diabetes evidence is a 242-participant six-month albuminuria trial (PMID 41780000).
OQ-3. What is the absolute clinical-event benefit of SGLT2 inhibition in A1 CKD?¶
EMPA-KIDNEY showed marked relative slowing of chronic eGFR slope at uACR <30 mg/g, but slow progression yields fewer hard events and smaller absolute differences (PMID 38061371). Substantially narrowed (deepening pass 2026-09-02): a meta-analysis of 8 trials and 58,816 participants stratified by albuminuria found diabetes-specific hazard ratios for kidney disease progression similar above and below a uACR of 200 mg/g, with the absolute benefit larger at higher uACR simply because event rates are higher (PMID 41202026). What remains open is the absolute benefit specifically below 30 mg/g and the treatment horizon over which it accrues, which the 200 mg/g stratification does not resolve.
OQ-4. Which CKD screening strategy maximizes health while reducing inequity?¶
Cost-effectiveness depends on risk enrichment, uptake and treatment effects, while efficiency and equity may select different populations (PMIDs: 38213490, 38186904, 40227684). Pragmatic implementation trials should measure diagnosis communication and drug initiation, not screening yield alone.
OQ-5. Did race-free eGFR change referral, drug eligibility and transplant access?¶
The equation and task-force recommendation are established (PMIDs: 34554658, 34563581). Partly answered, and the answer points in two directions (deepening pass 2026-09-02): implementation reclassified 45.8% of Black adults with CKD to more advanced stages and 44.0% of non-Black adults to less severe ones in a 1.5-million-adult US health system (PMID 39100866); it unmasked a high-risk subgroup that the race-adjusted equation had concealed (PMID 36069064); and in a predominantly White European population it raised eGFR by a median 3.9 mL/min/1.73 m² and reclassified 36.2% of G3a–G5 patients upward without changing prognostic accuracy (PMID 35689668). Against that, simulation on 459 Black patients showed 13–22% more falling below common oncology trial eligibility floors (PMID 36606692), and only 32% of Black kidney transplant candidates received the mandated waiting-time modification, with wide centre-level variation (PMIDs: 37488677, 40327843). What remains open: no study has measured the net effect on any patient-important outcome, and no study has measured whether cystatin C access widened or narrowed disparities.
OQ-6. Can potassium-binder enablement preserve cardiorenal benefit long term?¶
AMBER improved spironolactone persistence at 12 weeks and meta-analysis supports RAASi optimization, but kidney-failure and cardiovascular-outcome effects are unknown (PMIDs: 31533906, 40542996).
OQ-7. What causal effect does conservative kidney management have relative to dialysis in older adults?¶
Reviews show strong selection and heterogeneous survival, quality-of-life and utilization results (PMIDs: 28538218, 34507554, 35285915, 41363177). A target-trial emulation with frailty, preference and longitudinal symptom data is feasible even if randomization is not.
OQ-8. Does routine symptom monitoring with feedback improve life participation?¶
IPOS-Renal is validated and symptom-feedback delivery is feasible, but clinical benefit remains uncertain (PMIDs: 29729346, 37993776). A cluster trial should use fatigue and life participation as co-primary patient outcomes (PMIDs: 29551585, 40569671).
OQ-9. Which IgA-nephropathy sequence best converts proteinuria response into preserved GFR?¶
NefIgArd and PROTECT have two-year outcome data (PMIDs: 37591292, 37931634), while FSGS demonstrated that superior proteinuria remission need not improve eGFR slope (PMID 37921461). The surrogate half is now answered for one mechanism (deepening pass 2026-09-02): iptacopan reduced 9-month uPCR by 38.3% (PMID 39453772) and at 24 months halved the annualized eGFR slope (−3.10 versus −6.12 mL/min/1.73 m²/year; difference 3.02, 95% CI 2.02–4.01) and reduced a composite kidney-failure endpoint from 33.5% to 21.4% (HR 0.57, 0.40–0.81) in 477 patients (PMID 41910396). Atrasentan's 2.5-year result was more equivocal — eGFR difference 2.4 (−0.1 to 4.8; p=0.057) at week 136 with a significant total slope difference of 1.4/year (0.5–2.3) (PMID 42242268) — and ravulizumab has phase 2 proteinuria data only (PMID 39455063). The sequence question is untouched: at least six agents now have phase 3 proteinuria evidence or trials in IgA nephropathy and none of the trials is comparative.
OQ-10. Can dialysis preparation reduce crisis starts without moving dialysis earlier?¶
IDEAL rejects eGFR-only early initiation but shows symptoms drove 75.9% of late-arm participants to start above the target (PMID 20581422). Risk-triggered education and preparation could be randomized without mandating earlier treatment.
OQ-21. Should the eGFR threshold defining CKD vary with age?¶
Age-adapted criteria (75/60/45 mL/min/1.73 m² below 40, 40–64, and 65 or over) removed 72,703 people from an incident CKD cohort, of whom 54,342 (75%) were aged 65 or over with eGFR 45–59 and A1 albuminuria, a 5-year kidney-failure risk of ≤0.12%, and a death risk exceeding kidney-failure risk by 69- to 935-fold across age bands (PMID 34459844). The symmetric claim is that a fixed threshold under-diagnoses young adults, in whom risk begins rising above 60 (PMID 38411155). KDIGO retains one threshold (PMID 38490803), defensibly, because albuminuria already captures much of the risk and because risk is continuous (PMID 37787795). What would settle it: no trial has randomized a diagnostic threshold, and the decisive outcome — net harm from labelling versus net benefit from earlier protective therapy — has never been measured directly. A stepped-wedge implementation of age-adapted reporting with prescribing, referral, and patient-reported outcomes is designable now.
OQ-22. Does correcting metabolic acidosis slow CKD progression?¶
This is unanswered rather than answered negatively. An open-label trial of sodium bicarbonate against standard care in 740 patients reported creatinine doubling in 6.6% versus 17.0% and death in 3.1% versus 6.8% over 36 months (PMID 31598912); a blinded placebo-controlled trial of an acid binder in 1,480 patients found a hazard ratio of 0.99 (0.8–1.2) but achieved a between-arm bicarbonate separation of only about 1 mEq/L (PMID 38261535). No adequately powered blinded trial has yet achieved a large, durable bicarbonate separation. A trial delivering oral alkali at UBI doses with blinded endpoint adjudication would resolve it, and the fruit-and-vegetable comparison (PMID 30995657) suggests the delivery route may itself matter for cardiovascular risk markers.
OQ-23. Do aldosterone synthase inhibitors improve on mineralocorticoid receptor blockade?¶
Two event-driven trials totalling about 16,000 participants are recruiting on background SGLT2 inhibition — vicadrostat plus empagliflozin (NCT06531824, n=11,000) and baxdrostat plus dapagliflozin (NCT06742723, n=5,000) — with non-interchangeable primary composites. The mechanistic question is whether suppressing aldosterone synthesis rather than blocking its receptor delivers equal or greater kidney protection with a different hyperkalaemia profile, given that finerenone produced mild hyperkalaemia in 21.4% versus 9.2% under protocol-scheduled monitoring (PMID 34732509). Neither trial reports before 2028.
OQ-24. Is anti-inflammatory therapy effective in CKD?¶
CANTOS provided randomized evidence in a prespecified CKD subgroup of 1,875 participants: canakinumab reduced major adverse cardiovascular events (HR 0.82, 0.68–1.00), with the largest benefit in those achieving on-treatment hs-CRP below 2 mg/L (HR 0.68, 0.53–0.86) (PMID 29793629). No dedicated CKD trial followed for seven years. ZEUS now randomizes 6,376 participants with atherosclerotic cardiovascular disease, CKD (mean eGFR 44.5 mL/min/1.73 m²) and hs-CRP ≥2 mg/L to ziltivekimab or placebo, with 3-point MACE as primary outcome (PMID 41369941). Whether IL-6 inhibition affects kidney outcomes as well as cardiovascular ones is not the primary question of that trial.
OQ-25. Should sickle cell trait be communicated as a kidney disease risk factor?¶
In REGARDS, sickle cell trait carried an ESRD hazard ratio of 2.03 (1.44–2.84) — similar in magnitude to the APOL1 high-risk genotype in the same cohort (1.77, 1.31–2.38) — while haemoglobin C trait carried no excess risk (PMID 28280138). Sickle cell trait is routinely communicated as a carrier state without health consequences. What is missing is a prospective evaluation of whether disclosure with kidney-directed screening changes detection, treatment or outcome, and whether it causes harm through labelling.
Tier 2 — enabling science or longer horizon required¶
OQ-11. Which biological components of creatinine–cystatin C discordance are clinically actionable?¶
The combined equation is more accurate overall (PMID 34554658), but whether discordance phenotypes improve dosing or prognosis beyond direct variables remains unresolved.
OQ-12. Can fibrosis markers identify reversible rather than fixed progression?¶
Fibrosis is a common final pathway (PMID 18161745), but a marker that distinguishes active, treatment-responsive injury from scar has not been validated against outcomes.
OQ-13. How should AKI episodes update long-term CKD risk?¶
Current risk tools are dominated by eGFR and uACR (PMIDs: 26757465, 36857500). A dynamic model needs standardized AKI severity, recovery and competing-risk integration; AKI management itself remains outside this condition.
OQ-14. What outcome validates phosphate or calcification as a treatment target?¶
EVOLVE was neutral on its unadjusted primary endpoint and calcification trials remain heterogeneous (PMIDs: 23121374, 35232774). A validated causal surrogate is still missing.
OQ-15. Who benefits from dietary protein restriction enough to justify burden?¶
MDRD and later meta-analyses leave modest, adherence-sensitive estimates (PMIDs: 8114857, 10541304, 30403710). Biomarker-defined or rapidly progressive subgroups may differ.
OQ-16. Can a plant-predominant diet reduce acidosis without increasing clinically important hyperkalaemia?¶
Mechanistic and review evidence is promising, while potassium bioavailability and additive exposure complicate conventional restriction (PMIDs: 32528189, 32775988, 37610407). A hard-outcome diet trial remains difficult.
OQ-17. Which genomic diagnoses change adult CKD management enough to justify broad testing?¶
Diagnostic utility is established in selected biopsy and family-history populations (PMIDs: 32723786, 34515170), but downstream treatment, family and cost outcomes need prospective study.
OQ-18. What is the minimum clinically important difference for CKD fatigue across stages?¶
Fatigue is a core outcome but may not behave identically in non-dialysis CKD, dialysis and transplant populations (PMIDs: 29551585, 40599823).
OQ-19. Can eGFR slope be used without being misled by acute haemodynamic effects?¶
DAPA-CKD and EMPA-KIDNEY separate acute dips from chronic slopes (PMIDs: 34619108, 38061371). Cross-class validation needs consistent estimands and off-treatment follow-up.
OQ-20. How much of endemic CKD of unknown aetiology is heat, toxin, infection or social exposure?¶
Reviews document inconsistent case definitions and multiple correlated exposures (PMIDs: 37403003, 33116757). Harmonized prospective cohorts with exposure measurement are prerequisite.
OQ-26. Why do urate and hydration associations fail to replicate as causal effects?¶
Three interventions with strong observational support, plausible mechanism and demonstrated target engagement produced null results: allopurinol in CKD stage 3–4 (eGFR difference −0.10 mL/min/1.73 m²/year, PMID 32579811), allopurinol in type 1 diabetes with iohexol-measured GFR (0.001 mL/min/1.73 m², PMID 32579810), and coached water intake raising urine volume by 0.6 L/day (eGFR difference −0.3, PMID 29801012). Understanding what these three exposures share — probably reverse causation through GFR itself determining the measured exposure — would improve the prior on other observationally-derived targets.
OQ-27. How should pregnancy be represented in CKD prognostic models?¶
Pregnancy in CKD stages 3–5 produced an eGFR fall of 4.5 mL/min/1.73 m² against a pre-pregnancy annual decline of 1.8, equivalent to 1.7, 2.1 and 4.9 years of progression at stages 3a, 3b and 4–5 (PMID 33313680). In the opposite direction, preeclampsia carried a subsequent hazard of 3.23 (1.64–6.36) for eGFR below 60 and 3.60 (2.38–5.44) for albuminuria, yet only about a quarter of affected people had post-partum kidney testing (PMID 37516302). No prognostic model contains a pregnancy term, and no care pathway routinely follows kidney function after preeclampsia.
OQ-28. Does anything modify the delivery constraint?¶
The therapeutic layer now supports a 37% relative reduction in kidney disease progression (PMID 36351458), while the median global supply of nephrologists is 11.75 per million population, with Africa at 1.12 and South Asia at 1.81, and more than half of 167 surveyed countries reporting nephrologist shortages (PMID 39235198). Microsimulation across 31 countries projects CKD costs rising 9.3% against a 5.8% population rise by 2027 at current diagnosis rates (PMID 41141496). Whether task-shifting, protocolised primary-care prescribing or algorithmic detection can close any part of this gap has not been tested against clinical endpoints.
Dots not yet connected¶
| # | Dot A | Dot B | The missing junction | Powers |
|---|---|---|---|---|
| D1 | Four drug classes reduce kidney outcomes (PMIDs: 11565518, 30990260, 33264825, 38785209) | Combination benefit is modelled (PMID 37952217) | Randomized sequence and full-regimen effectiveness | OQ-1 |
| D2 | Race-free eGFR changes classification (PMID 34554658) | Screening access is unequal (PMID 40227684) | Downstream diagnosis, referral and treatment by equity stratum | OQ-4, OQ-5 |
| D3 | KFRE predicts kidney failure (PMID 26757465) | CKM decisions depend on competing death and burden (PMID 35285915) | One calibrated decision model for both pathways | OQ-7, OQ-10 |
| D4 | Low-uACR EMPA-KIDNEY slope benefit (PMID 38061371) | Hard events accrue slowly | Absolute event benefit and treatment horizon in A1 CKD | OQ-3, OQ-19 |
| D5 | Binders enable RAAS/MRA use (PMIDs: 31533906, 40542996) | Outcome trials establish drug benefit | Whether enablement recovers that benefit | OQ-6 |
| D6 | Patients prioritize fatigue and life participation (PMIDs: 29551585, 40569671) | Symptom feedback is feasible (PMID 37993776) | Effect of measurement-feedback on lived outcomes | OQ-8, OQ-18 |
| D7 | Proteinuria is treatment-responsive in FSGS (PMID 37921461) | eGFR slope was neutral | Surrogate validation by disease and drug class | OQ-9, OQ-19 |
| D8 | Plant diets lower acid load conceptually (PMID 32528189) | Hyperkalaemia limits therapy (PMID 40542996) | Food-source-specific potassium strategy with clinical outcomes | OQ-16 |
| D9 | Genomics can resolve diagnosis (PMID 32723786) | Cause-specific drugs are emerging (PMIDs: 23121377, 37591292) | Prospective genotype-to-treatment utility | OQ-17 |
| D10 | IDEAL rejects threshold-only dialysis start (PMID 20581422) | KFRE can trigger preparation (PMID 26757465) | Preparation without treatment acceleration | OQ-10 |
| D11 | CKD-MBD markers track risk (PMID 28646995) | Clinical trials of calcification modification are weak (PMID 35232774) | Causal marker and patient-important endpoint | OQ-14 |
| D12 | Endemic CKDu clusters with heat and work (PMID 33116757) | Global CKD burden models aggregate causes (PMID 32061315) | Exposure-specific burden and prevention estimate | OQ-20 |
| D13 | Proteinuria reduction converts to kidney failure benefit in IgA nephropathy (PMID 41910396) | The identical class of endpoint did not convert in FSGS (PMID 37921461) | Disease- and mechanism-specific surrogate validation before regulatory reliance | OQ-9, OQ-19 |
| D14 | Ambient heat is associated with eGFR loss in treated CKD across 21 countries (PMID 38580424) | Global burden models attribute CKD DALYs to glucose, BMI and blood pressure only (PMID 41213283) | Climate exposure as a quantified term in CKD burden projection | OQ-20, OQ-28 |
| D15 | Sickle cell trait carries an ESRD hazard similar to APOL1 high-risk genotype (PMID 28280138) | APOL1 genotype is entering living-donor evaluation (PMID 42329639) | Whether trait status should enter the same counselling and screening pathways | OQ-25, OQ-17 |
| D16 | Preeclampsia triples the subsequent hazard of eGFR below 60 (PMID 37516302) | Only ~26% receive post-partum kidney testing (PMID 37516302) | A post-partum kidney surveillance pathway with measured uptake and yield | OQ-27 |
| D17 | Incident coronary calcification predicts atherosclerotic events; progression within calcified arteries does not (PMID 39154888) | Every CKD-MBD intervention trial targeted the biochemistry rather than the transition to calcification (PMIDs: 23121374, 34003226) | An intervention trial keyed to the zero-to-nonzero transition rather than to phosphate or PTH | OQ-14 |
| D18 | Frailty predicts the outcomes that compete with kidney failure (PMID 33218914) | KFRE contains no frailty term and resists every attempted enrichment (PMIDs: 36857500, 26787778) | A single model returning kidney-failure risk and competing-mortality risk on the same scale | OQ-7, OQ-10 |
| D19 | Discontinuing RAASi after hyperkalaemia is associated with higher mortality and dialysis initiation (PMID 35085685) | Potassium binders demonstrably prevent discontinuation (PMID 35900838) | A trial randomizing binder-enabled continuation to clinical endpoints | OQ-6 |
| D20 | Non-diabetic kidney disease is found in 58.8% of biopsied patients with diabetes and carries 2.56-fold lower ESKD risk (PMID 42034202) | Biopsy is performed rarely in diabetes | A prospective study of a lowered biopsy threshold with treatment and outcome follow-up | OQ-17 |