TacroGuard · Clinical evidence pack
Hard cases, one dependable answer each.
A curated compendium for transplant physicians: real, challenging tacrolimus cases run through TacroGuard — each shown as the clinician’s pain-point and the system’s deterministic output. The cases were chosen to test the tool at its limits, where trust is actually earned. Companion to the TacroGuard clinician training module.
How to read each verdict✓ cross-check agreed two independent methods concur — act with confidence
⚠ flagged — verify methods diverge — confirm before prescribing
🛑 hard stop contraindicated interaction — verify before dosing
⚠ out of population / low confidence the tool declares its own limits
TG-06Starting dose — nothing back yet
TG-N4Green path — early phase on track
TG-D7Green path — stable maintenance
TG-PNDeterminism — same case, same answer
TG-K1Opposing forces that self-resolve
TG-KDTwo infections pulling opposite ways
TG-VVOpposing Tier-1 drugs + hard stop
TG-I1Cross-check catches an overshoot
TG-I2Interaction propping up the level
TG-STCritical, multi-flag — loudest about limits
TG-SCSepsis raises the target, not the dose
TG-PDOut of the validated population
TG-R1Real case — early expresser, impaired graft
TG-R2Real case — delayed graft function
TG-R3Real case — late maintenance, non-expresser
TG-MrKPrototype — the decision that goes wrong
TG-06Starting dose — nothing back yet
45 y M · 65 kgKidney · Day 0Genotype not resultedNo prior levels
The pain-point“First prescription. No trough back, genotype not resulted yet. What do I safely start?”
TacroGuard6.5 mg/daypopulation-based starting regimen — explicitly provisional
Target 8–12 ng/mL
Starting estimateResubmit after first trough
Why it earns trustWith no individual data it gives a guideline-based starting dose and says it is provisional, then asks for one trough to switch to an individualised recommendation — no false precision on day zero.
TG-N4Green path — early phase on track
52 y M · 68 kgKidney · Day 13Recent troughs below target
The pain-point“Two weeks out, trough sitting below target on the current dose — nudge up by how much?”
TacroGuard6.5 mg/dayindividualised from his own recent levels
Target 8–12 ng/mL
✓ independent cross-check agreed
Why it earns trustTwo independent methods produced the same recommendation — a built-in agreement check that lets you act on the number with confidence.
TG-D7Green path — stable maintenance
52 y F · 68 kgKidney · Day 133Once-daily formulationStable troughs
The pain-point“Stable at four months on once-daily dosing — confirm I’m neither over- nor under-exposing.”
TacroGuard4.0 mg/daymaintenance-phase target, individualised
Target 5–8 ng/mL
✓ cross-check agreed exactly
Why it earns trustBoth independent methods landed on the identical maintenance dose — maximal confidence to hold steady rather than tinker.
TG-PNDeterminism — same case, same answer
52 y F · 68 kgKidney · Day 135Independently re-submitted
The pain-point“Would a different clinician, entering the same patient, get the same recommendation?”
TacroGuard4.0 mg/dayidentical inputs, independently entered
Target 5–8 ng/mL
✓ reproducible — operator-independent
Why it earns trustAn independently submitted, identical clinical picture returned the identical dose. The recommendation is reproducible — not a matter of who typed it in.
TG-K1Opposing forces that self-resolve
38 y M · 60 kgKidney · CYP3A5 expresserRifampicin on boardBK viraemia active
The pain-point“BK says lower the level; rifampicin has already crashed it. Do I go up or down?”
TacroGuard8.0 mg/daytwo opposing directives reconciled
Target 3–5 ng/mL
BK — target loweredEnzyme inducer on board✓ cross-check agreed
Why it earns trustIt names the tension out loud — BK wants less exposure while the inducer has already driven levels low — and directs any further reduction to the co-drugs, not tacrolimus, guarding against rebound when the inducer stops.
TG-KDTwo infections pulling opposite ways
50 y F · 65 kgKidney · Day 72BK + CMV both activeDiltiazem on board
The pain-point“BK wants the level down, CMV wants it up — which infection wins the target?”
TacroGuard2.5 mg/daygraft-threat prioritised
Target 3–5 ng/mL
BK — target loweredCMV — also activeDiltiazem interaction✓ cross-check agreed
Why it earns trustIt prioritises the graft-threatening BK for the target and states the correct order of operations: treat CMV with antivirals, not by raising tacrolimus.
TG-VVOpposing Tier-1 drugs + hard stop
58 y F · 68 kgLiver · Day 29Strong-expresser genotypeAzole inhibitor + inducer
The pain-point“A potent inhibitor AND an inducer, on a strong-expresser genotype — three arrows, opposite ways.”
TacroGuard8.5 mg/dayflagged for verification — not autopilot
Target 8–12 ng/mL
🛑 Hard stop — inhibitor⚠ cross-check flagged — verify
Why it earns trustIt refuses to autopilot: the inhibitor raises a hard-stop banner and the cross-check flags the instability, telling you to verify before dosing rather than trust one number.
TG-I1Cross-check catches an overshoot
45 y M · 70 kgKidney · Day 20Immunoassay TDMMMF-associated diarrhoea
The pain-point“Persistently low troughs on a reasonable dose — just push higher?”
TacroGuard12.5 mg/daycross-check disagreed — verify first
Target 8–12 ng/mL
⚠ cross-check flagged — verifyCheck assay & adherence
Why it earns trustThe two methods diverged, so instead of a confident jump it flags — check sampling time, assay type and adherence (the diarrhoea) before escalating. It catches exactly where a naive calculation would overshoot.
TG-I2Interaction propping up the level
38 y F · 42 kgKidney · Day 34Low body weightDiltiazem (moderate inhibitor)
The pain-point“Trough already top-of-range, and diltiazem is on board holding it up — do I reduce?”
TacroGuard4.0 mg/daycompound interaction surfaced, decision left to you
Target 6–10 ng/mL
Diltiazem interaction✓ cross-check agreed
Why it earns trustIt surfaces that the inhibitor may be propping up the level and asks you to weigh that before accepting the dose — the tool flags, the clinician decides.
TG-STCritical, multi-flag — loudest about limits
44 y M · 78 kgLiver · Day 21Child-Pugh CCRRT + deteriorating sepsisCMV + azole inhibitor
The pain-point“A crashing liver recipient on CRRT, septic, on an azole, troughs swinging wildly — I need help but I can’t trust a black box here.”
TacroGuard2.5 mg/dayengine down-weights itself and stops
Target 10–14 ng/mL
🛑 Hard stop — azoleCMV — target raisedCRRT + sepsis⚠ LOW confidence
Why it earns trustIn the messiest case it is loudest about its limits: a safety override, a hard-stop, a raised target for CMV, and an explicit LOW-CONFIDENCE label — it points you to frequent TDM, not to itself.
TG-SCSepsis raises the target, not the dose
54 y F · 60 kgKidney · Day 88CRRTWorsening sepsis(synthetic case)
The pain-point“Septic and deteriorating — drop immunosuppression to fight infection, or hold to protect the graft?”
TacroGuard4.5 mg/daytarget raised, dose held; flagged to re-check
Target 8–12 ng/mL
Target raised — sepsisCRRT⚠ cross-check flagged — verify
Why it earns trustSepsis raises rejection risk, so it lifts the target to keep exposure adequate rather than reflexively cutting the dose — and flags for a fresh trough given the instability. (Independently-constructed synthetic case.)
TG-PDOut of the validated population
6 y M · 20 kgKidney · Day 43PaediatricExpresser genotype
The pain-point“A six-year-old — but adult dosing tools weren’t built for him.”
TacroGuard3.5 mg/dayoutside validated population — declared
Target 6–10 ng/mL
⚠ Paediatric — out of population⚠ cross-check flagged
Why it earns trustThe large cross-check divergence is the tool announcing the patient is outside its adult-validated range — a declared boundary, not a confident number it has no right to give.
TG-R1Real case — early expresser, impaired graft
38 y M · KidneyDay 3 (early)CYP3A5 expresser *1/*1Cr 3.8 — impaired
The pain-point“Post-op day 3: an expresser with a slow early graft (creatinine 3.8). Dose up for the genotype, or hold for the kidney?”
TacroGuard11.0 mg/dayindividualised from the day-3 trough
Target 8–12 ng/mL
Expresser *1/*1✓ cross-check agreed
Why it earns trustA real, de-identified early-graft case: the engine fits the dose to his own handling from the first trough and holds the early-phase target through the genotype-vs-renal tension — cross-check concordant.
TG-R2Real case — delayed graft function
52 y F · KidneyDay 14CYP3A5 expresser *1/*3Cr 4.8 — DGF
The pain-point“Two weeks out, trough 8.2 on 5 mg but creatinine 4.8 — delayed graft function. Push to target, or is tacrolimus contributing?”
TacroGuard7.5 mg/dayindividualised; cross-check flagged
Target 8–12 ng/mL
Expresser *1/*3⚠ cross-check flagged — verify
Why it earns trustA real, de-identified delayed-graft-function case: with the impaired renal covariate the two methods diverge, so the engine flags for verification (sampling time, the DGF picture) rather than a confident push — the honest behaviour on an unstable early graft.
TG-R3Real case — late maintenance, non-expresser
45 y M · KidneyDay 420 (>1 yr)CYP3A5 non-expresser *3/*3Cr 1.8 · stable
The pain-point“Over a year out, a stable non-expresser on 3 mg — is any change warranted, and to what target?”
TacroGuard2.5 mg/daylate-phase minimisation target
Target 3–6 ng/mL
Non-expresser *3/*3✓ cross-check agreed
Why it earns trustA real, de-identified long-term case: the engine applies the late-phase minimisation target (3–6) — not the early band — and trims the dose accordingly, cross-check concordant. (This case surfaced, and validated the fix for, the day→phase mapping.)
TG-MrKPrototype — the decision that goes wrong
45 y M · KidneyPOD 90CYP3A5 expresserCr 1.4→1.9 · trough 13
The pain-point“Creatinine climbing and trough 13 — rejection (go up) or toxicity (come down)? Backwards, and the injury accelerates.”
TacroGuard5.5 mg/day · reduce, not increasehigh trough + rising creatinine = toxicity pattern
Target 6–10 ng/mL
Trough 13 → supratherapeutic🔴 CNI toxicity pattern
Why it earns trustThe prototype teaching case (see the Mr K one-pager): a high trough with a rising creatinine is toxicity, not rejection, so the engine reduces and warns against the increase reflex — the clinical heart of the tool.