A century-old marker of a filtration function, used as a marker of injury — the mismatch explains most of our diagnostic delay.
Koyner Ch 16 · Ronco Ch 9, 25
Five ways creatinine lies in the ICU
It lags
Creatinine-defined AKI is detected on average 2–3 days after the insult. Every nephrotoxin given in that window is given blind.
It is diluted
Volume of distribution expands with resuscitation. A patient +8 L can hold a "stable" creatinine while GFR halves.
It tracks muscle, not kidney
Production falls with sarcopenia, sepsis, immobility and liver failure. A creatinine of 0.6 in a cachectic ICU patient may be a GFR of 40.
It is secreted
10–15% of clearance is proximal tubular secretion via OCT2/MATE — block the transporter and the number rises with no change in filtration.
The fifth, and the most important
Creatinine reports function, never damage. It cannot distinguish a haemodynamic rise (SGLT2 inhibitor, RAAS blockade, effective diuresis) from acute tubular injury — and those two states deserve opposite management.
Steady state: GFR × SCr = creatinine generation
Non-steady state (the ICU): SCr is still rising toward its new plateau
KeGFR = (SCr₁ × eGFR₁) / SCrmean × [ 1 − (24 × ΔSCr) / (Δt × MaxΔSCr/day) ] MaxΔSCr per day ≈ creatinine production ÷ volume of distribution ≈ 1.5 mg/dL/day at GFR 0. If creatinine rises 0.8 mg/dL in 12 h, the instantaneous GFR is already near zero — the plateau value is 3 days away.
A single creatinine in a non-steady state is uninterpretable as an eGFR — every CKD-EPI, MDRD and Cockcroft–Gault equation assumes steady state
Kinetic eGFR converts the rate of rise into a real-time filtration estimate; useful for drug dosing and for arguing an early consult
A falling creatinine can equally mislead: dilution and reduced generation both lower it without recovery
Chen S. J Am Soc Nephrol. 2013;24:877–888 · Ronco Ch 25 · Koyner Ch 16
Drugs and assays that raise creatinine without touching GFR
Agent
Mechanism
Typical rise
What to do
Trimethoprim
Blocks OCT2/MATE tubular secretion of creatinine
0.2–0.4 mg/dL within days
Recheck cystatin C; do not stop a needed antibiotic reflexively
Cimetidine
Same transporter block (historically used to "correct" clearance)
0.2–0.4 mg/dL
Switch to famotidine if the number is confusing care
Cobicistat, ritonavir
MATE1 inhibition; pharmacokinetic boosters
~0.1–0.2 mg/dL, plateaus by week 4
Expect it; a further rise is real injury
Dolutegravir, bictegravir
OCT2 inhibition
~0.1–0.15 mg/dL, early and stable
Establish the new baseline, then trend from it
Flucytosine, cefoxitin, high ketones
Chromogen interference with the Jaffe assay
Variable, can be large
Ask the laboratory for an enzymatic assay
Creatine supplements, cooked meat load
Increased creatinine generation
0.1–0.3 mg/dL
Dietary history before the workup
Pitfall 1 — treating the assay instead of the patient
A creatinine that rises promptly, plateaus, and is unaccompanied by any change in urine output, sediment or electrolyte handling is a transporter or assay artefact until proven otherwise. Cystatin C settles it in one blood tube; a kidney biopsy and a cancelled antiretroviral do not.
Koyner Ch 16 · Ronco Ch 25 · Inker LA et al. N Engl J Med. 2021;385:1737–1749
Cystatin C and the 2021 race-free equations
13.3 kDa protein made by all nucleated cells at a near-constant rate; >99% filtered, then catabolised by the proximal tubule — so plasma level tracks GFR and urine level tracks proximal tubular injury
Independent of muscle mass: the right test in the cachectic, cirrhotic, paraplegic or amputee patient
Meta-analysis AUC for diagnosing AKI 0.89; rises roughly 12–24 h earlier than creatinine
Confounded by: corticosteroids (raise it), thyroid dysfunction (hyper raises, hypo lowers), smoking, obesity, high cell turnover and some inflammatory states
Cleared by high-flux and CRRT membranes — uninterpretable during kidney replacement therapy
Order it when
Creatinine and the clinical picture disagree
You must dose a narrow-therapeutic-index drug
The patient has no usable muscle mass
You suspect a secretion-blocking drug
eGFRcr-cys (CKD-EPI 2021) uses creatinine + cystatin C + age + sex — and no race coefficient The combined equation is more accurate than either marker alone (P30 ≈ 86–92% vs ~80% for creatinine alone) and is the preferred estimate whenever a decision hangs on the number.
Inker LA, Eneanya ND, Coresh J, et al. N Engl J Med. 2021;385:1737–1749 · Yong Z et al. Sci Rep. 2017;7:41012 · Ronco Ch 25 · Koyner Ch 16
02
Damage, function & the 2×2
Two independent axes — injury and filtration — and four clinically distinct quadrants that a creatinine alone collapses into two.
Koyner Ch 16–17 · Ronco Ch 24, 26–27
The damage × function matrix
Damage marker negative
Damage marker positive
Creatinine / UO normal
No AKI — the reassuring quadrant
Subclinical AKI — injury without a functional signal. Deceased-donor biopsies: about half of kidneys with histologic acute tubular injury had no creatinine-defined AKI
Clinical AKI with damage — true tubular injury. The quadrant our therapies were designed for
Why the left-lower box matters
In ADHF undergoing aggressive diuresis, NGAL and KIM-1 did not rise in patients whose creatinine went up — the "worsening renal function" was haemodynamic, not injury. Same story in SPRINT: intensive BP control raised creatinine and "CKD" incidence while YKL-40 and KIM-1 were lower.
Why the right-upper box matters
Pooled prospective cohorts: NGAL-positive / creatinine-negative patients had higher rates of kidney replacement therapy and death than NGAL-negative / creatinine-negative patients. Subclinical injury is not a laboratory curiosity.
Haase M et al. J Am Coll Cardiol. 2011;57:1752–1761 · Ahmad T et al. Circulation. 2018;137:2016–2028 · Moledina DG et al. Am J Kidney Dis. 2017;70:807–816 · Koyner Ch 16
The damage biomarker menu — biology, sample, window
Marker
Biology
Sample
Peak after insult
Signal strength
NGAL (25 kDa)
Lipocalin upregulated in thick ascending limb and collecting duct
Urine or plasma
2–3 h, detectable by 3 h
Plasma AUC 0.80 for AKI progression (TRIBE-AKI)
KIM-1 (38.7 kDa)
Transmembrane protein on dedifferentiated proximal tubule
Urine
12 h, persists ~48 h
Best for established ischaemic injury; poor for early warning
IL-18 (22 kDa)
Caspase-1–dependent proinflammatory cytokine
Urine
6 h
Separates ATN from prerenal azotaemia; inflammation-confounded
Approved in Japan; strongest data in cardiac surgery
[TIMP-2]·[IGFBP7]
G1 cell-cycle arrest — "kidney stress", not yet necrosis
Urine
4–12 h
FDA-cleared 2014; AUC 0.80 for stage 2–3 AKI in 12 h
Cystatin C (urine)
Filtered then normally fully reabsorbed — its appearance means proximal tubular failure
Urine
12–24 h
Dual function/damage reader
Ronco Ch 26–27 · Krawczeski CD et al. J Am Coll Cardiol. 2011 · Koyner JL et al. J Am Soc Nephrol. 2012;23:905–914 · Koyner Ch 16
NGAL — the most studied, the most confounded
Where it earns its keep
Paediatric cardiac surgery: urinary NGAL rose ≥50-fold at 2 h after bypass and preceded the creatinine rise by ≥24 h in every child who developed AKI
TRIBE-AKI adults: highest tertile of plasma NGAL carried an odds ratio 7.7 (95% CI 2.6–22.5) for AKI progression
Emergency department: separates transient from sustained AKI in >600 patients
Higher levels at the time of AKI predict 3-year mortality
Where it fails
NGAL is also a neutrophil granule protein — sepsis, pneumonia, UTI and any systemic inflammation raise it with no tubular injury
CKD raises the baseline; pyuria raises the urine value
Plasma and urine assays are not interchangeable and are not standardised across laboratories
No trial has shown that acting on an NGAL result improves any patient outcome
How to use it if your laboratory offers it
Read NGAL as a trend within one patient against a known insult time, in a population with a high pre-test probability. A single value in an undifferentiated septic patient carries almost no information.
Mishra J et al. Lancet. 2005;365:1231–1238 · Koyner JL et al. J Am Soc Nephrol. 2012;23:905–914 · Nickolas TL et al. J Am Coll Cardiol. 2012;59:246–255 · Ronco Ch 26
Cell-cycle arrest: why [TIMP-2]·[IGFBP7] is different
The idea
Damage markers report cells that are already injured. Cell-cycle arrest markers report cells that have stopped dividing to protect themselves — stress before structural damage.
The mechanism
ROS, PAMPs and DAMPs push tubular cells into the cell cycle; IGFBP7 induces p53/p21 and TIMP-2 induces p27, which block CyclD-CDK4 and CyclE-CDK2 and impose a transient G1/S arrest.
Why arrest helps
It prevents division with damaged DNA and conserves ATP during a bioenergetic crisis. In sepsis models the G1:S ratio rises within 6 h, subsides by 24 h, and functional recovery follows at 72 h.
Why it is measurable
Both proteins are secreted into the tubular lumen and appear in urine, so a single voided or catheter sample reads the whole nephron population.
The division of labour
IGFBP7 performs better in surgical patients; TIMP-2 performs better in sepsis. The product of the two was chosen as the panel.
Kashani K et al. Crit Care. 2013;17:R25 · Ronco Ch 27 · Koyner Ch 16
Sapphire — the derivation, and the two numbers you must remember
SAPPHIRE · Crit Care 2013;17:R25
Discovery in 522 critically ill adults without AKI at enrolment; >340 candidate proteins screened. Validation in 728 patients across 35 centres in North America and Europe, all within 24 h of ICU admission, KDIGO stage 2–3 excluded. Endpoint: moderate-to-severe AKI within 12 h. [TIMP-2]·[IGFBP7] AUC 0.80 — beating NGAL, KIM-1, IL-18 and every other marker tested, and improving every clinical model it was added to.
≤ 0.3(ng/mL)²/1000 — sensitivity 92%, specificity 46%. A negative test genuinely rules out imminent stage 2–3 AKI
0.3 – 2.0Kidney stress present. Positive predictive value ~25% — monitor, de-risk, do not act irreversibly
> 2.0Specificity 90%; positive predictive value ~50%. Roughly a 7-fold higher risk than a negative test
Report units are (ng/mL)² / 1000 — a printed "0.3" is 300 (ng/mL)² Assay CV 10% at the 0.3 cut-off; limit of quantitation 0.002. Interference at albumin >1250 mg/L, conjugated bilirubin >72 mg/L, methylene blue >0.49 mg/L. Stable 6 h at room temperature, 24 h refrigerated.
Kashani K, Al-Khafaji A, Ardiles T, et al. Crit Care. 2013;17:R25 · Ronco Ch 27
Topaz and Opal — validation, and the limits of validation
TOPAZ · Am J Respir Crit Care Med 2014;189:932
420 patients, 23 US centres, enrolled within 24 h of ICU admission; AKI within 12 h adjudicated by three blinded experts. Adding urinary [TIMP-2]·[IGFBP7] at the 0.3 cut-off lifted the clinical model from AUC 0.70 (0.63–0.76) to 0.86 (0.80–0.90). Patients above 0.3 had roughly a 7-fold higher risk of moderate-severe AKI.
OPAL · Nephrol Dial Transpl 2014;29:2054
154 patients at six US sites, run on the commercial NephroCheck platform locally rather than a central ELISA. Confirmed the two Sapphire thresholds: sensitivity 89% at 0.3 and specificity 90% at 2.0. This is the study that made the cut-offs usable at a bedside analyser.
Pitfall 2 — using the test outside the population it was built in
Every one of these cohorts excluded patients already in KDIGO stage 2–3 and enrolled within 24 h of ICU admission. The test answers one question: "will this adult (≥21 y) critically ill patient develop stage 2–3 AKI in the next 12 hours?" It does not diagnose established AKI, does not stage it, does not identify the cause, and performs poorly when the pre-test probability is low. Sending it on a stable ward patient generates cost and anxiety, not information.
Bihorac A et al. Am J Respir Crit Care Med. 2014;189:932–939 · Hoste EAJ et al. Nephrol Dial Transplant. 2014;29:2054–2061 · Ronco Ch 27
Timing is the whole game
Marker
Draw it
Question it answers
Do not draw it when
[TIMP-2]·[IGFBP7]
0–12 h after a defined insult (ICU admission, bypass, laparotomy)
Will stage 2–3 AKI appear in 12 h?
Stage 2–3 AKI already present
NGAL
2–6 h after insult, then serially
Is there tubular injury now?
Undifferentiated sepsis with no insult time
IL-18 / L-FABP
6 h
Injury vs pure haemodynamics
Systemic inflammatory states (IL-18)
KIM-1
12–48 h
Did injury actually occur?
You need an early warning
Cystatin C (plasma)
12–24 h; serially
What is the real GFR?
On CRRT; on steroids; thyroid disease
Creatinine
Every day, forever
Staging, trials, outcomes — the anchor
Never; just never read it alone
Key point
Damage markers rise and fall on the clock of the insult, not the clock of the ward round. A biomarker drawn 48 h after cardiopulmonary bypass has missed its own window and will read negative in a genuinely injured kidney.
Krawczeski CD et al. J Am Coll Cardiol. 2011 · Ronco Ch 26–27 · Koyner Ch 16
Order with intent — biomarker stewardship
01
Pre-testName the insult and its time zero. No insult, no test.
→
02
Pre-commitWrite down what a high and a low result will each change, before you send it.
→
03
PairAdd a functional test (FST) to a damage marker — different axes, additive information.
→
04
DocumentRisk class, actions taken, recheck interval. An uninterpreted result is a wasted one.
Actions a high result should trigger
Stop or dose-adjust every nephrotoxin; review contrast plans
Avoid hyperglycaemia; hold ACEi/ARB for 48 h post-operatively
Early nephrology involvement and a KRT plan on paper
Actions it must never trigger
Starting kidney replacement therapy — the test predicts, it does not prescribe (see Lecture 08 on timing)
A diuretic given "to protect the kidney"
Cancelling an indicated contrast study
A kidney biopsy without an independent indication
Ostermann M et al. JAMA Netw Open. 2020;3:e2019209 (ADQI 23) · Koyner Ch 8, 16
The newer entrants: proenkephalin and CCL14
Proenkephalin A 119–159 (penKid)
A ~4.9 kDa stable fragment of the enkephalin precursor; freely filtered, not protein-bound, not produced by inflammation
Behaves as a functional marker: tracks measured GFR in real time and is not confounded by sepsis-driven cytokine release the way cystatin C-era markers were feared to be
In the Kid-SSS analysis of AdrenOSS-1, an elevated admission penKid identified septic patients who went on to major adverse kidney events — before any creatinine change
Not yet FDA-cleared for routine US practice; assay availability and thresholds still vary by cohort
Urinary CCL14
A monocyte-recruiting chemokine; answers a different and harder question — who with established stage 2–3 AKI will stay there
RUBY: predicted persistent severe AKI (stage 3 lasting ≥72 h) with AUC 0.83 (0.78–0.87), outperforming every marker in the comparison set
Two thresholds mirroring the NephroCheck logic: a low-value rule-out (~1.3 ng/mL) and a high-value rule-in (~13 ng/mL)
Clinical use case: rationing KRT resources and framing goals-of-care conversations, not preventing AKI
Hollinger A et al. Kidney Int Rep. 2018;3:1424–1433 (Kid-SSS) · Hoste E et al. Intensive Care Med. 2020;46:943–953 (RUBY) · Koyner Ch 17
Why biomarkers alone changed nothing — and what did
A test improves outcomes only when it is welded to an intervention that works. Two decades of biomarker studies measured discrimination; the trials that mattered measured what happened when someone acted on the number.
PrevAKI · Intensive Care Med 2017;43:1551
276 cardiac surgery patients randomised only if post-operative [TIMP-2]·[IGFBP7] >0.3. Intervention = the KDIGO bundle (avoid nephrotoxins, stop ACEi/ARB 48 h, tight glucose, avoid contrast, PiCCO-guided haemodynamic optimisation to SVV <11, CI >3, MAP >65). AKI at 72 h 55.1% vs 71.1% (p = 0.004); stage 2–3 AKI 29.7% vs 44.9% (p = 0.009). No mortality difference.
BigpAK · Ann Surg 2018;267:1013
121 patients after major abdominal surgery, same biomarker trigger, a lighter bundle. Any-stage AKI 32% vs 48% (p = 0.07, primary endpoint not met); stage 2–3 AKI 7% vs 20% (p = 0.04), with shorter ICU and hospital stay.
PrevAKI-Multicentre · Anesth Analg 2021;133:292
Multicentre replication in biomarker-positive cardiac surgery patients: moderate-to-severe AKI 14.0% vs 23.9% with the bundle, without a change in any-stage AKI, KRT or mortality. The effect is real but modest, and it belongs to the bundle — the biomarker only selects who gets it.
Meersch M et al. Intensive Care Med. 2017;43:1551–1561 · Göcze I et al. Ann Surg. 2018;267:1013–1020 · Zarbock A et al. Anesth Analg. 2021;133:292–302 · Koyner Ch 8
03
Stress-testing the kidney
Cardiology stresses the heart before it fails; we can stress the glomerulus with protein and the tubule with furosemide — and the cheap test wins.
Ronco Ch 9 · Koyner Ch 16
Renal functional reserve — a normal GFR is not a safe GFR
Healthy subjects raise GFR measurably 1–2 h after an acute protein load of 1–1.2 g/kg (red meat is the classic stimulus; IV amino acids do the same)
RFR = maxGFR − baseline GFR. Mechanism: filtered amino acids resensitise the macula densa, releasing NO and prostaglandins → afferent vasodilation and nephron recruitment
RFR falls with age, nephron loss, hyperfiltration states and pregnancy — in each trimester baseline GFR climbs while maxGFR does not, so reserve is progressively spent
Creatinine stays normal until roughly 50% of nephron mass is gone: the living kidney donor with a "normal" eGFR has already lost half of the reserve
Losing reserve also means losing autoregulatory range — this is the physiology behind the NSAID- and volume-sensitivity of the CKD patient (Lecture 02)
Why it is not routine
Cardiologists stress-test because they can act on a positive result. We have had no equivalent intervention — so the test stayed in the laboratory. That may change: reduced RFR is the most plausible mechanism by which "fully recovered" AKI survivors go on to develop CKD (Lecture 09).
Bosch JP et al. Am J Med. 1983;75:943 · Fliser D et al. Kidney Int. 1993 · Ronco Ch 9
The furosemide stress test — the protocol, exactly
1
Select — KDIGO stage 1 or 2 AKI, and the patient must be euvolemic. Correct hypovolaemia first; the test is unsafe and uninterpretable in a volume-depleted patient.
2
Dose — furosemide IV 1.0 mg/kg if loop-diuretic naïve, 1.5 mg/kg if the patient has had a loop diuretic in the preceding week.
3
Replace — match urine losses mL-for-mL with isotonic crystalloid over the 2 hours. This is the step most often skipped and the one that turns the test into an iatrogenic insult.
4
Measure — hourly urine output for 2 hours via an accurate catheter.
5
Read — <200 mL over 2 hours = a failed FST (equivalently, <100 mL in either hour). Passing means the tubule can still secrete an organic acid and still respond to NKCC2 blockade.
Furosemide is not filtered — it is albumin-bound and secreted into the lumen by proximal hOAT transporters,
then blocks Na-K-2Cl in the thick ascending limb. So a failed FST means either the proximal tubule cannot secrete, or the loop cannot respond, or both. It is a direct functional interrogation of the nephron — not a surrogate.
Chawla LS, Davison DL, Brasha-Mitchell E, et al. Crit Care. 2013;17:R207 · Ronco Ch 9
The FST beats the biomarker panel — and the two together beat both
0.87AUC (SE 0.05) for progression to KDIGO stage 3 within 14 days
87 / 84% sensitivity / specificity at the 200 mL-in-2-h cut-off
0.86AUC for the need for kidney replacement therapy
0.90AUC for progression when the FST is applied to biomarker-positive patients
The headline result
In the Koyner follow-up of the Chawla cohort, no individual damage biomarker reached an AUC of 0.80 for AKI progression, while the FST reached 0.87 — and rose to 0.90 (progression) and 0.91 (KRT) when used in patients who were already biomarker-positive. A €5 ampoule of furosemide out-predicts the panel; the panel's value is in choosing who to stress-test.
Pitfall — the FST is a diagnostic test, not a therapeutic one
A patient who passes has not been "treated". A patient who fails must not then receive escalating furosemide infusions to chase urine — the diuretic does not change the natural history of AKI (Lecture 03), and the failed test is a prompt to plan for KRT, not to double the dose. Also avoid the FST in overt hypovolaemia, decompensated cirrhosis with tense ascites, and sulfa-anaphylaxis history.
Koyner JL, Davison DL, Brasha-Mitchell E, et al. J Am Soc Nephrol. 2015;26:2023–2031 · Ronco Ch 26 · Koyner Ch 16
AKI sub-phenotypes: one syndrome, at least two diseases
Latent class analysis of two critical-care cohorts found two reproducible AKI sub-phenotypes with different biology and very different outcomes
AKI-SP2 (hyperinflammatory) — higher soluble TNF receptor-1, angiopoietin-2/angiopoietin-1 ratio and ICAM-1; markedly higher mortality
AKI-SP1 (hypoinflammatory) — endothelial and inflammatory markers near normal; better survival despite an identical creatinine
In the VASST cohort there was a treatment-by-sub-phenotype interaction: vasopressin versus norepinephrine did not affect the two groups equally — the first real signal that AKI trials may have been diluted by lumping
Practical implication today: none of these markers is orderable. The lesson is interpretive — stop treating "AKI" as one exposure
Why this matters for every negative AKI trial you have read
If half of an enrolled population has haemodynamic AKI with an intact tubule and half has inflammatory tubular injury, a therapy that works well in one can be invisible in the pooled result. Enrichment — by biomarker, by sub-phenotype, or by a failed FST — is how the next generation of trials will be designed.
Bhatraju PK, Zelnick LR, Herting J, et al. Am J Respir Crit Care Med. 2019;199:863–872 · Koyner Ch 16
04
Risk scores & electronic alerts
Scores are excellent at describing populations and mediocre at deciding for individuals — know which job you are asking them to do.
AUROC 0.74–0.81; best used to decide who gets a biomarker
Thakar CV et al. J Am Soc Nephrol. 2005;16:162–168 · Mehta RH et al. Circulation. 2006;114:2208–2216 · Wijeysundera DN et al. JAMA. 2007;297:1801–1809 · Mehran R et al. J Am Coll Cardiol. 2004;44:1393–1399 · Basu RK et al. Clin J Am Soc Nephrol. 2014;9:654–662 · Koyner Ch 7
ICU severity scores: what they are for, and what they are not
Mortality prediction models
APACHE II — 12 physiological variables + age + chronic health, worst value in the first 24 h. Still the most widely used
APACHE IV — maximum 286 points; score 56 ≈ 10% predicted mortality, 132 ≈ 90%. Takes ~37 minutes to compute by hand
SAPS III — 20 variables in three blocks, region-specific equations
In large external validations discrimination is moderate (c ≈ 0.80–0.83) but calibration consistently degrades and needs local recalibration
Organ dysfunction scores
SOFA — six systems, worst value each day; the only one that includes a treatment variable (vasopressor dose). An acute rise of ≥2 points defines organ dysfunction in Sepsis-3
SOFA >15 has been associated with ~90% mortality; a rise over the first 48 h with >50% mortality, a fall with ~27%
MODS and LODS — similar discrimination, less used
The renal SOFA sub-score is creatinine- and urine-output-based and therefore inherits every limitation from section 01
Pitfall — the self-fulfilling score
These models were built to compare units, not to triage patients. AKI-specific mortality models perform worse than general ICU models on external validation (most AUROC <0.7). Quoting a predicted mortality at the bedside risks a prophecy that fulfils itself.
Knaus WA et al. Crit Care Med. 1985;13:818 · Vincent JL et al. Intensive Care Med. 1996;22:707 · Harrison DA et al. Crit Care Med. 2007;35:1091 · Koyner Ch 7 · Ronco Ch 6
Electronic AKI alerts: detection without a pathway does nothing
Tackling AKI · stepped-wedge cluster RCT, 5 UK hospitals, 24,059 AKI episodes
E-alert + care bundle + education. No change in the primary outcome of 30-day mortality. Improved AKI recognition, shorter AKI duration and shorter length of stay — a process win, not a survival win.
Randomised alert trials · Lancet 2015 / BMJ 2021
Automated creatinine-triggered alerts delivered in isolation did not improve creatinine trajectory, dialysis, or death; in the multicentre replication the signal in some sites ran the wrong way. An alert without an owner and an action set is noise added to an already noisy ICU.
What this means for your hospital project
Build the response first — who is paged, what they must do in the next hour, and how it is audited. Then attach the trigger. The ICE-AKI and NINJA programmes worked because the alert was the smallest part of the intervention.
Selby NM et al. J Am Soc Nephrol. 2019;30:505–515 · Wilson FP et al. Lancet. 2015;385:1966–1974 · Wilson FP et al. BMJ. 2021;372:m4786 · Koyner Ch 8
05
Imaging the kidney and the venous side
Ultrasound answers binary questions superbly and continuous questions badly — ask it the right kind of question.
Koyner Ch 18 · Ronco Ch 33–35
Renal ultrasound: five questions, in this order
Is there obstruction? Confluent, arborising anechoic spaces within the hyperechoic sinus fat — graded mild / moderate / severe
Is this kidney chronically diseased? Length <9 cm, cortical thinning, parenchyma hyperechoic to liver — CKD, and it changes the whole prognosis
Is the parenchyma abnormal? Echogenicity is compared to liver (right) and spleen (left); it is subjective and operator-dependent, so describe rather than score
Is there a structural lesion? Cysts, masses, stones, perinephric collections, renal vein thrombus
What does the bladder show? A distended bladder makes the upper tracts irrelevant until it is drained; bilateral ureteric jets on colour Doppler argue against, but do not exclude, obstruction
Bladder volume ≈ 0.52 × length × width × height (cm, scanned in two planes) A "blocked catheter" and a neurogenic bladder are the two cheapest AKI cures in the hospital. Scan before you consult.
Pitfall — hydronephrosis is neither sensitive nor specific
False positives: parapelvic cysts (discrete, with posterior acoustic enhancement), prominent vasculature (use Doppler), brisk diuresis, diabetes insipidus, pregnancy, the denervated transplant. False negatives: obstruction in the first hours, encasing retroperitoneal malignancy or fibrosis, severe volume depletion, and the already-stented ureter. A normal scan in a patient with a compelling story buys a repeat scan or a CT, not reassurance.
Koyner Ch 18 · Faubel S et al. Clin J Am Soc Nephrol. 2014;9:382–394 · O'Neill WC. Clin J Am Soc Nephrol. 2014;9:373–381
Renal resistive index — a real signal buried in confounders
RRI = (peak systolic velocity − end-diastolic velocity) / peak systolic velocity Measure in an interlobar artery near the mesorenal column; report the average of three cycles in each kidney.
Normal <0.70. In one comparative series, acute interstitial nephritis ran ~0.75 vs ~0.58 for acute glomerular disease; ATN, vasculitis and hepatorenal syndrome sit >0.78–0.82
Baseline RRI <0.68 after bypass, shock, trauma or sepsis identified patients who did not develop AKI; baseline >0.77–0.78 identified those who needed KRT within 1–5 days
A falling RRI toward a low-resistance waveform predicts recovery, and it falls before the creatinine does
Intra-abdominal hypertension, high PEEP, vasopressor dose
Operator angle, depth and gain settings
Which is why RRI is a trend within a patient, never a threshold across patients
Ronco Ch 33 · Koyner Ch 18 · Darmon M et al. Intensive Care Med. 2011;37:68–76
VExUS — grading venous congestion 0 to 3
Vein
Normal
Mild
Severe
Hepatic vein
S wave larger than D wave
S wave smaller than D but still below baseline
S wave reversed (above baseline)
Portal vein
Continuous, pulsatility <30%
Pulsatility 30–49%
Pulsatility ≥50%
Intrarenal vein
Continuous
Biphasic (systolic + diastolic components)
Monophasic, diastolic only
0IVC <2 cm — no congestion, regardless of the venous waveforms
1IVC ≥2 cm with only mild abnormalities
2IVC ≥2 cm with one severe pattern
3IVC ≥2 cm with two or more severe patterns
Why the nephrologist should own this
In the derivation cardiac-surgery cohort, VExUS grade 3 carried a hazard ratio of 3.69 (95% CI 1.65–8.24) for subsequent AKI — a stronger association than CVP alone. A grade-3 kidney needs decongestion, not perfusion pressure: giving fluid or raising MAP here makes it worse (see Lecture 02). Grades are for hypervolaemia; VExUS says nothing useful in a hypovolaemic or a fluid-naïve patient.
Beaubien-Souligny W, Rola P, Haycock K, et al. Ultrasound J. 2020;12:16 · Koyner Ch 18 · Ronco Ch 33
The rest of the volume examination — and when to leave ultrasound
Lung
B-lines: vertical reverberation artefacts that abolish A-lines. One per field is normal; >2 per field is pathological and the count scales with extravascular lung water
Diffuse and homogeneous → cardiogenic oedema; patchy with thickened pleura and subpleural consolidations → ARDS
In dialysis patients B-lines beat the physical examination for overload and fall dynamically with ultrafiltration
IVC
Measured 2 cm below the diaphragm, confirmed by hepatic vein continuity
Useful only at the extremes: flat and collapsing, or plethoric and fixed
Invalidated by positive-pressure ventilation, RV failure, pulmonary hypertension, tamponade and raised intra-abdominal pressure
Beyond ultrasound
Non-contrast CT — stones, retroperitoneal disease, an equivocal ultrasound in a large patient
Contrast-enhanced ultrasound — regional cortical perfusion with no nephrotoxicity; promising, no standard protocol
MR / nuclear — rarely justifies moving an unstable ICU patient; gadolinium raises the NSF question in eGFR <30
Read the pattern, not the number
Plethoric IVC + diffuse B-lines + monophasic renal venous flow is a congested kidney. Flat IVC + A-lines + consolidation is a septic, under-filled kidney. Neither conclusion depends on any single measurement being right.
Koyner Ch 18 · Lichtenstein DA, Mezière GA. Chest. 2008;134:117–125 (BLUE) · Zoccali C et al. J Am Soc Nephrol. 2013;24:639–646 · Ronco Ch 34–35
06
Contrast-associated AKI
The most over-feared diagnosis in the hospital — and the fear itself now causes measurable harm.
Koyner Ch 44 · Ronco Ch 35
What the control groups did to contrast nephropathy
Reported incidence is entirely definition-dependent: in CKD stage 3–4, a ≥25% creatinine rise occurred in 13.2% after non-coronary angiography, 8.5% after coronary angiography and 6.5% after contrast CT. ICU series report 11.5–19%
Then someone scanned the control group. In CKD patients, AKI after non-contrast CT was 8.8% — versus 9.7% with iodixanol and 9.9% with iohexol
Meta-analysis of 13 studies: risk of AKI after contrast-enhanced versus non-enhanced imaging, RR 0.79 (95% CI 0.62–1.02)
Propensity-matched cohort of 6,877 ICU patients: no difference in AKI with contrast, OR 0.88 (0.75–1.05)
These are observational and residually confounded — but they establish that most of what we called contrast nephropathy was the background noise of critical illness
The association that remains
CA-AKI, when it occurs, is associated with in-hospital death (7.1% vs 1.1% after PCI), a roughly 3-fold higher 1-year risk of death/stroke/MI/ESKD, ~2 extra hospital days and faster CKD progression. Causality is unproven — CA-AKI may simply mark the patient with less reserve and more instability.
Weisbord SD et al. · Bruce RJ et al. AJR. 2009;192:711 · McDonald JS et al. Radiology. 2013;267:119–128 · McCullough PA et al. · Koyner Ch 44
Who is genuinely at risk — and who is not
Patient
Real risk?
What to do
eGFR ≥45, stable, IV contrast CT
Effectively none
Scan. No prophylaxis, no delay, no post-scan creatinine ritual
eGFR 30–44, stable
Small
Scan. Isotonic saline is reasonable but AMACING found no benefit; check creatinine at 48–96 h
eGFR <30, or AKI in evolution
Yes — the principal patient-related risk factor
Lowest contrast volume, isotonic saline, stop NSAIDs, discuss alternatives with radiology — then scan if indicated
Intra-arterial, first-pass renal exposure
Yes — higher than IV for the same dose
Applies to coronary and aortic angiography; minimise volume and repeat exposures
Volume depleted, hypotensive, on IABP
Yes — and it is modifiable
Resuscitate first if the clock allows; this matters more than the contrast agent
Repeat contrast within 48–72 h
Yes
Consolidate imaging; ask whether the second study changes management
Diabetes with normal kidney function
No independent risk
Diabetes amplifies risk only when CKD is present
ESKD on maintenance dialysis
No residual function to lose
Scan freely. No urgent dialysis "to remove contrast" — prophylactic haemodialysis is potentially harmful
Koyner Ch 44 · Davenport MS et al. Radiology. 2020;294:660–668 (ACR/NKF consensus)
Prophylaxis: what two large trials abolished
PRESERVE · N Engl J Med 2018;378:603
4,993 patients with CKD undergoing non-emergent angiography, 2×2 factorial. Oral N-acetylcysteine 600 mg twice daily for 5 days vs placebo: 90-day death, dialysis or persistent kidney impairment OR 1.02 (0.78–1.33); CA-AKI OR 1.06 (0.87–1.28). IV isotonic sodium bicarbonate vs isotonic saline: composite OR 0.93 (0.72–1.22); CA-AKI OR 1.16 (0.96–1.41). Both interventions dead.
AMACING · Lancet 2017;389:1312
660 high-risk patients with eGFR 30–59 randomised to prophylactic IV hydration or no prophylaxis before elective contrast. CA-AKI 2.6% with no prophylaxis vs 2.7% with hydration — non-inferior, cheaper, and with fewer complications of fluid administration.
What survives
Isotonic saline in genuinely high-risk patients: 1 mL/kg/h for 12 h before, during and after; or 3 mL/kg/h for 1 h pre, 1–1.5 mL/kg/h during, 4–6 h after
Lowest volume of low- or iso-osmolal contrast (they are equivalent to each other; both beat high-osmolal)
Hold NSAIDs until CA-AKI has been excluded
Recheck creatinine at 48–96 h in at-risk patients
What does not
N-acetylcysteine, sodium bicarbonate, statins started for this purpose
Prophylactic haemodialysis; CRRT data are conflicting and insufficient
Stopping ACEi/ARB or diuretics routinely — the data do not support it
POSEIDON showed fluid can be given safely even with a raised LVEDP, so "heart failure" is not an automatic exemption
Weisbord SD et al. N Engl J Med. 2018;378:603–614 · Nijssen EC et al. Lancet. 2017;389:1312–1322 · Brar SS et al. Lancet. 2014;383:1814 (POSEIDON) · Koyner Ch 44
"Renalism": the arithmetic nobody does out loud
The harm we can measure
In >57,000 patients with acute myocardial infarction, those with CKD were about 50% less likely to undergo coronary angiography
Observational data show a survival advantage for CKD patients who do undergo angiography and revascularisation
Untreated pulmonary embolism carries a mortality around 25–30%; an aortic dissection missed for six hours is often fatal
The comparator is not "AKI vs no AKI" — it is a small, mostly transient creatinine rise against a missed lethal diagnosis
The consult you should actually give
"Yes, scan. Here is how we de-risk it."
Name the fluid, the volume limit and the recheck interval in the note
Correct hypotension and hypovolaemia first if the clock permits — that is the modifiable risk factor
Refuse only when an equally diagnostic non-contrast alternative genuinely exists
Key point
The nephrologist's job in the contrast conversation is to de-risk, not to block. Every avoided study has a cost that never appears in the kidney outcome literature, because the patient who died of the undiagnosed condition was never counted as a contrast complication.
Chertow GM et al. J Am Soc Nephrol. 2004;15:2462 (renalism) · Koyner Ch 44 · Davenport MS et al. Radiology. 2020;294:660–668
07
Cases & wrap-up
Two consults where the tests we have just discussed genuinely change the plan — and one where they must not.
Koyner Ch 8, 16, 18, 44
Case 1 · Post-cardiac surgery, hour 4
71 M, redo mitral valve replacement, 168 min on cardiopulmonary bypass, EF 40%, insulin-requiring diabetes, pre-operative creatinine 1.4 mg/dL (eGFR 48). The ICU asks whether they should "send a NephroCheck".
Fix the gradientMAP 68 − CVP 14 = RPP 54 mmHg. Decongest or raise MAP — decide which, then reassess
→
04
Re-measureCreatinine and urine output q12h; repeat biomarker only if it will change the plan again
The one-line summary
A high [TIMP-2]·[IGFBP7] earns this patient an hour of attention, not a dialysis catheter. In PrevAKI that hour was worth an absolute reduction of about 15 percentage points in AKI at 72 h.
Meersch M et al. Intensive Care Med. 2017;43:1551–1561 · Zarbock A et al. Anesth Analg. 2021;133:292–302 · Koyner Ch 8, 39
Case 2 · Septic shock, day 3 — the creatinine has stalled
58 F, urosepsis, off vasopressors since yesterday, cumulative balance +6.5 L. Creatinine rose 0.8 → 2.1 mg/dL over 48 h and has now been flat for 24 h. The team wants to know whether she is recovering or heading for dialysis.
Name the VExUS grade. Then decide: furosemide stress test now, or not yet?
Two answers required — grade, then action
Case 2 — the two tests that answered the question
VExUS grade 3
IVC 2.4 cm + portal pulsatility 55% + monophasic intrarenal venous flow = two severe patterns
The renal problem is back-pressure, not forward flow: MAP 74 − CVP 15 gives a renal perfusion pressure of 59 mmHg
Therapy is decongestion. Fluid or a higher MAP target here makes the kidney worse
Furosemide stress test
Furosemide 1.0 mg/kg IV (loop-naïve), urine counted hourly for 2 h
≥200 mL → tubule intact: proceed with a loop infusion and tolerate a modest creatinine rise during effective decongestion
<200 mL → AUC 0.87 for progression to stage 3, 0.86 for KRT: secure access, brief the family, stop escalating the diuretic
The trap in this case
A creatinine that has been flat for 24 h feels like recovery. It is equally consistent with a new steady state at a much lower GFR, or with reduced creatinine generation in a catabolic, immobile patient. Neither the number nor its slope can distinguish these — only a functional test and the urine output can.
Chawla LS et al. Crit Care. 2013;17:R207 · Koyner JL et al. J Am Soc Nephrol. 2015;26:2023–2031 · Beaubien-Souligny W et al. Ultrasound J. 2020;12:16
Quick poll
A 63-year-old admitted 30 hours ago with community-acquired pneumonia. Creatinine 1.1 (baseline 1.0), urine output 0.9 mL/kg/h, no vasopressors, no contrast planned. The night registrar sent a [TIMP-2]·[IGFBP7]: it comes back at 0.9.
A. Start prophylactic KRT planning · B. Give a 1 L fluid bolus · C. Deploy the KDIGO bundle and repeat the biomarker in 12 h · D. Note that the test should not have been sent, interpret it in the light of the low pre-test probability, and manage the pneumonia
Hands up — then name the property of the test that makes your answer correct
Key takeaways
Creatinine lags 2–3 days, is diluted by resuscitation, tracks muscle mass, and is raised by trimethoprim, cimetidine, cobicistat and dolutegravir without any change in GFR — cystatin C and the CKD-EPI 2021 combined equation settle the argument.
Think in the damage × function 2×2: subclinical AKI (damage without a creatinine rise) carries real excess KRT and mortality, and haemodynamic AKI (creatinine rise without damage) does not.
[TIMP-2]·[IGFBP7] answers exactly one question — stage 2–3 AKI within 12 h — with cut-offs 0.3 (sensitivity 92%) and 2.0 (specificity 90%) from Sapphire, validated in Topaz and Opal, and useless outside that population.
No biomarker has improved an outcome on its own; PrevAKI, PrevAKI-MC and BigpAK improved outcomes by using the biomarker to select who receives the KDIGO care bundle.
The furosemide stress test (1.0–1.5 mg/kg, <200 mL in 2 h = failure) reaches AUC 0.87 for progression to stage 3 and out-predicts every damage biomarker — and rises to 0.90 when combined with them.
Ultrasound answers binary questions: obstruction, chronicity, and VExUS grade. Contrast is safe in eGFR ≥30; skip N-acetylcysteine and bicarbonate (PRESERVE), skip routine prophylactic hydration in eGFR 30–59 (AMACING), and never let renalism delay a study that could save the patient.
References & further reading
Where to go deeper
Kashani K, Al-Khafaji A, Ardiles T, et al. Discovery and validation of cell cycle arrest biomarkers in human acute kidney injury (Sapphire). Crit Care. 2013;17:R25.
Bihorac A, Chawla LS, Shaw AD, et al. Validation of cell-cycle arrest biomarkers for acute kidney injury using clinical adjudication (Topaz). Am J Respir Crit Care Med. 2014;189:932–939.
Chawla LS, Davison DL, Brasha-Mitchell E, et al. Development and standardization of a furosemide stress test to predict the severity of acute kidney injury. Crit Care. 2013;17:R207. — and Koyner JL, Davison DL, Brasha-Mitchell E, et al. Furosemide stress test and biomarkers for the prediction of AKI severity. J Am Soc Nephrol. 2015;26:2023–2031.
Meersch M, Schmidt C, Hoffmeier A, et al. Prevention of cardiac surgery-associated AKI by implementing the KDIGO guidelines in high-risk patients identified by biomarkers: the PrevAKI randomised controlled trial. Intensive Care Med. 2017;43:1551–1561.
Inker LA, Eneanya ND, Coresh J, et al. New creatinine- and cystatin C-based equations to estimate GFR without race. N Engl J Med. 2021;385:1737–1749.
Bhatraju PK, Zelnick LR, Herting J, et al. Identification of acute kidney injury subphenotypes with differing molecular signatures and responses to vasopressin therapy. Am J Respir Crit Care Med. 2019;199:863–872.
Weisbord SD, Gallagher M, Jneid H, et al. Outcomes after angiography with sodium bicarbonate and acetylcysteine (PRESERVE). N Engl J Med. 2018;378:603–614. — and Nijssen EC, Rennenberg RJ, Nelemans PJ, et al. Prophylactic hydration to protect renal function from intravascular iodinated contrast material (AMACING). Lancet. 2017;389:1312–1322.
Beaubien-Souligny W, Rola P, Haycock K, et al. Quantifying systemic congestion with point-of-care ultrasound: the VExUS grading system. Ultrasound J. 2020;12:16. · Koyner JL, Topf JM, Lerma EV, eds. Handbook of Critical Care Nephrology. Wolters Kluwer; 2021 (Ch 7, 8, 16–18, 39, 44).
Critical Care Nephrology · Two-Week Intensive
Thank you
Questions & discussion — bring one test you will stop ordering and one you will start.