CPPopt targeting, a 4-hour COGiTATE-style loop

1. Three patient vignettes#

Vignette A. Canonical school-age severe TBI#

Liam, 12 years old, 40 kg. Severe TBI from a high-speed bicycle vs car. GCS 6 at scene, intubated, paralysed, frontal triple-bolt with ICP probe + PbtO2 + brain temperature placed on arrival. Bilateral frontal NIRS pads. Continuous arterial line. RASS −5, deep sedation. Day 2. Current state: ICP 18 mmHg (acceptable), MAP 75, CPP 57, PbtO2 18, PRx hovering at +0.18. ICM+ on the bedside computer has been collecting (CPP, PRx) pairs since hour 0; at hour 4 it reports a parabolic fit with R² 0.68 and CPPopt 71 mmHg, target band 66 to 76. The team must decide: lift CPP into the band by raising MAP with noradrenaline, or stay on the fixed CPP > 50 protocol the unit has used for a decade? 1 2

Vignette B. Toddler with severe TBI#

Riya, 3 years old, 14 kg. Severe TBI from a fall down a stairwell. GCS 5 at scene, frontal ICP bolt placed, no PbtO2 (the unit does not stock pediatric-sized PbtO2 probes). Day 1. ICP 22, MAP 70, CPP 48, PRx +0.30. The age-banded CPP floor for a 3-year-old is 50 mmHg; she is below it. The ICM+ buffer has only 90 minutes of data; CPPopt is not yet computable. The team starts a noradrenaline infusion at 0.05 mcg/kg/min to lift MAP, then begins second-line ICP measures (3% saline 3 mL/kg, head-of-bed 30 degrees, normothermia). By hour 4 they have a low-confidence CPPopt of 58 mmHg (R² 0.42, only 3 bins). They stay on the age-floor strategy until confidence builds: target CPP > 55 mmHg empirically. The toddler-specific point: CPPopt in young children may be lower than adult numbers; do not transpose adult thresholds. 3 4 5

Vignette C. Atypical: CPPopt looks beautiful but PbtO2 says no#

Hana, 15 years old. Severe TBI day 3, triple-bolt in. ICM+ reports CPPopt 75, band 70 to 80, current CPP 76, PRx +0.05. Looks perfect. But PbtO2 is 14 mmHg (low; target > 20). NIRS rSO2 is 60 / 61 (low end of normal). Haemoglobin 8.2 g/dL. PbtO2 in the ischaemic range despite a confident CPPopt is the canonical reason for the discordance triage: tissue oxygen depends on CPP plus haemoglobin plus arterial PO2 plus tissue diffusion plus metabolic demand. CPPopt is a perfusion-pressure target; it does not guarantee tissue oxygenation. Action: transfuse to Hb > 9, raise FiO2, deepen sedation to lower CMRO2, and only then raise CPP if PbtO2 stays low. The lesson: CPPopt narrows the perfusion target; PbtO2 verifies the tissue actually benefits. 6 7

2. The clinical question#

In a child with severe TBI, what is the right CPP target right now, for this brain, this hour? The classical answer was a fixed age-banded threshold (CPP > 40 for infants, > 50 for older children). The CPPopt answer is a continuously-recomputed band centred on the autoregulation U-curve vertex. The integration question: when do you abandon the fixed threshold and follow the band?

3. Pathophysiology refresher#

The autoregulation U-curve is the bedside operationalisation of the Lassen plateau. In a healthy brain, cerebral blood flow is constant across a wide range of CPP (roughly 60 to 150 mmHg in adults; narrower and lower in children). Below the lower limit (LLA), CBF falls passively with CPP (pressure-passive, ischaemic); above the upper limit (ULA), CBF rises passively (pressure-passive, hyperaemic with oedema risk). PRx, the moving-window correlation between MAP and ICP, is the most-used bedside index of autoregulation: negative or near-zero PRx indicates intact reactivity; positive PRx indicates pressure-passivity. 8 9 1

Why a U-curve? When CPP is below LLA, PRx is positive (vessels are maximally dilated and cannot dilate further). When CPP is above ULA, PRx is again positive (vessels are maximally constricted). In between, PRx is minimised at the perfusion pressure where autoregulatory reserve is greatest. Plot PRx against binned CPP over a 4-hour window and you typically see a U or J shape; the vertex is CPPopt. 1 10

Why 4 hours? Shorter windows give noisy fits (insufficient bin density); longer windows smear over physiological changes (CPPopt drifts with metabolic state, sedation depth, temperature). The 4-hour rolling window is the empirical compromise. 1 2

Confidence matters. The CPPopt algorithm computes a fit quality (R² of the parabola). With < 4 populated CPP bins or R² < 0.3, the fit is unreliable and the algorithm reports "no CPPopt available." This usually happens early in monitoring, after a sedation change, or when CPP has been held flat (no variability means no curve). The bedside response to "no CPPopt" is to fall back to the age-banded floor. 2 10

COGiTATE (Tas 2021) randomised 60 severe-adult-TBI patients to CPPopt-targeted care vs standard CPP > 60. The CPPopt-guided arm kept CPP within ±5 mmHg of CPPopt for 46.5% of monitored time (vs a 36% feasibility target) and spent less time below CPPopt than controls (19.1% vs 34.6%), with no increase in therapeutic intensity. It was a feasibility and safety trial, not powered for outcome, and did not show a mortality benefit. It remains the closest the field has to a controlled trial of the approach. 11

Pediatric evidence. In children with severe TBI, CPPopt is computable at the bedside and a deviation of CPP from CPPopt has been associated with worse outcome (Velle 2023, n=57). STARSHIP (Agrawal 2023) is the major multicentre pediatric autoregulation study. The pediatric evidence base is observational; no pediatric RCT exists yet. 5 12

The CPP dose hypothesis. Guiza 2015 (in adults) showed that the time-and-magnitude product of CPP outside the autoregulation band predicted outcome more strongly than mean CPP; the same analysis extended the framework to children. The implication: a brief CPP excursion outside the band is forgiven; sustained excursions are not. 13 10

4. The multimodal picture table#

The most useful pairings: CPPopt + PbtO2 (the gold-standard pair: pressure target + tissue verification), CPPopt + PRx + NIRS (the non-invasive trio when PbtO2 unavailable), and CPPopt + ORx (two autoregulation indices for cross-check).

5. Decision tree#

flowchart TD
  Start[Severe TBI, ICP probe in, ICM+ running] --> Wait[Wait 4 h for first CPPopt fit]
  Wait --> Conf{CPPopt confidence}
  Conf -->|R² < 0.3 or < 4 bins| Floor[Fall back to age-banded CPP floor]
  Conf -->|R² 0.3 to 0.5| Wait2[Watch; do not titrate yet]
  Conf -->|R² ≥ 0.5, ≥ 6 bins| Target[Adopt CPPopt band as target]
  Target --> Position{Current CPP vs band}
  Position -->|Below band| RaiseMAP[Raise MAP via noradrenaline; recheck PRx]
  Position -->|In band| Maintain[Maintain; recheck in 30 min]
  Position -->|Above band| LowerMAP[Lower MAP gently; reduce noradrenaline]
  RaiseMAP --> Reassess[Check PbtO2, NIRS, PRx in 15 min]
  Reassess -->|All improve or stable| Continue[Continue at new target]
  Reassess -->|PbtO2 falls or NIRS drops asymmetric| Revert[Revert; reconsider target]
  Floor --> Variability[Allow modest MAP variability to build buffer]
  Variability --> Wait

6. Step-by-step bedside actions#

1. Verify the multimodal hub is recording synchronously. ICM+ (or local equivalent) must receive ICP, MAP, and ideally PbtO2 at 50 to 100 Hz with timestamps aligned.
2. Allow the first 4-hour buffer to fill. Do not chase a CPP target during the buffer-fill; stay on the age-banded floor and let natural Mayer-wave variability populate the (CPP, PRx) cloud.
3. At hour 4 check confidence: R² and bin count. If R² ≥ 0.5 and bin count ≥ 6, adopt the band as the target.
4. Move CPP into the band gently. A 10 mmHg upward move over 30 minutes (titrate noradrenaline) is safer than a single bolus. Watch ICP for paradoxical rise (vasodilatory cascade if autoregulation is broken).
5. Recheck PbtO2 and NIRS at 15 minutes. Convergent improvement = the move was right. Divergent (PbtO2 falls, NIRS drops asymmetric) = stop, reconsider.
6. Allow modest CPP variation through the day. A flat CPP erases the curve. Mayer-wave variability of ±5 to 10 mmHg around the target is desirable; do not aggressively damp it.
7. Re-evaluate confidence every 30 to 60 minutes. Drug changes, sedation holds, fever, and seizures all shift the curve. A sudden loss of confidence is a signal to investigate, not to ignore.
8. When CPPopt drifts, treat the new band as the new target with 30-minute averaging. Do not chase minute-by-minute changes (this is noise, not signal).
9. Document CPPopt at every nursing handover with the band, current CPP, current PRx, and current PbtO2. The bedside team must understand the target.
10. Plan for sedation holds: pre-empt the loss of CPPopt during the daily neurological exam; revert to the age-banded floor for the duration; document and reconverge afterwards.

Drug doses commonly used in CPP titration:

• Noradrenaline: 0.02 to 0.5 mcg/kg/min in children; titrate in 0.02 to 0.05 mcg/kg/min increments
• Vasopressin: 0.0001 to 0.002 units/kg/min as a second pressor when noradrenaline alone is insufficient
• Phenylephrine: 0.1 to 2 mcg/kg/min bolus or infusion (less commonly used in pediatrics)
• Fluid bolus: 5 to 10 mL/kg crystalloid; useful for absolute hypovolaemia, not as a chronic CPP-raiser

7. Management ladder and endpoints#

Success looks like: CPP > 80% of the time within the CPPopt band, PRx near zero, PbtO2 > 20, NIRS symmetric, ICP < 20, no new infarct on imaging, GCS recovery if survivable injury.

Failure looks like: refractory low PbtO2 despite optimal CPP (microvascular dysfunction), ICP rising paradoxically as CPP raised (lost autoregulation), or sustained CPP outside the band > 20% of the time (Guiza dose curve predicts worse outcome).

8. Variant subsections#

8.1 Severe TBI (the validated home)#

The original CPPopt setting. Aries 2012 (adult) showed retrospective association between % time within band and 6-month outcome. Depreitere 2014 extended this. COGiTATE (Tas 2021) is the first prospective controlled feasibility study. All this evidence is adult severe TBI; pediatric inference is by analogy plus Velle 2023 and STARSHIP. 1 10 11

8.2 SAH (emerging)#

In aneurysmal SAH, CPPopt computation is feasible but the U-curve is often distorted by vasospasm (vessels are constricted regardless of pressure). PRx may be flat through vasospasm even when DCI is developing (see TCD vs ICP vasospasm). Use CPPopt cautiously in the vasospasm window; pair with TCD and qEEG. 14

8.3 Pediatric severe TBI#

The PBTF 4th-edition guidelines (Kochanek 2019) endorse CPP-guided care with age-banded floors (50 mmHg for older children, 40 for infants and toddlers). CPPopt is mentioned as an emerging adjunct, not standard. Velle 2023 and STARSHIP (Agrawal 2023) are the pediatric autoregulation studies, and Smith 2023 reports a strong independent association between cerebrovascular pressure reactivity and outcome in children with severe TBI. 3 5 12 15

8.4 When CPPopt is unobtainable#

Common reasons CPPopt fails to compute: a flat CPP (no variability), too few CPP bins populated, R² < 0.3 (poor parabolic fit), a phase change in physiology (post-craniectomy, post-sedation change), or a software glitch. The fallback is always the age-banded floor: 50 mmHg for children > 5 years, 40 mmHg for younger. Document the fallback in the chart. 3

8.5 After decompressive craniectomy#

After DC, ICP often falls dramatically (effective volume increase). PRx behaviour changes (the cranium is now open, so the MAP-ICP correlation no longer reflects autoregulation in the same way). CPPopt may become uncomputable or may shift abruptly to a new band. The bedside team should re-establish baseline CPPopt over the first 4 to 6 hours post-DC and watch for the brain-bulge complication (sustained tissue herniation through the craniectomy defect). 16 17

8.6 Discordance: CPPopt and PbtO2 disagree#

Hana's vignette (C). CPPopt says one thing; PbtO2 says another. The differential: low haemoglobin (transfuse), low arterial PO2 (raise FiO2), high CMRO2 (deepen sedation, treat fever), microvascular dysfunction (no fix; supportive). The discordance is informative, not a defect; it forces you to broaden the diagnostic frame beyond perfusion pressure alone. See PRx vs COx discordance for a deeper treatment.

9. Multimodal integration matrix#

10. Worked alternative scenarios#

10.1 What if the patient is hyperaemic, not ischaemic?#

Liam at day 5: ICP 12, MAP 95, CPP 83, PRx +0.10. CPPopt 65, band 60 to 70. Current CPP is above the band. PbtO2 is 35 (high). NIRS rSO2 78%. This is hyperaemia; the response is to lower MAP gently (wean noradrenaline) into the band. Hyperaemic CPP carries oedema risk and is not benign just because the brain looks oxygenated. The CPPopt framework catches this where a fixed floor (CPP > 50) would have missed it.

10.2 What if CPPopt drifts upward through the day?#

CPPopt at hour 4: 70. Hour 12: 75. Hour 20: 82. The upward drift over 16 hours signals a shift in the brain's autoregulation envelope, often reflecting evolving cerebral oedema, rising metabolic demand from fever, or weaning of sedation. The action: investigate (temperature, sedation, fluid balance, ICP trend), and follow the band. Do not anchor on the initial CPPopt; the value is the current 4-hour vertex, not the day-1 number. 1

10.3 What if you cannot compute CPPopt at all?#

A bedside unit without ICM+, or a patient on V-V ECMO with damped arterial waveforms. The fallback is the age-banded CPP floor (PBTF guideline) plus the principle: vary MAP modestly through the day to learn the patient's response, watch ICP for paradoxical rises (lost autoregulation marker), use NIRS and (if available) PbtO2 as bedside surrogates. The Liam protocol degrades gracefully to a "fixed floor plus surveillance" strategy when the CPPopt machine is off. 3

11. Outcome data#

• Aries 2012: in 327 adult severe TBI patients, increased deviation of mean CPP from CPPopt was associated with worse 6-month outcome (mortality and unfavourable Glasgow Outcome Scale). 1
• Depreitere 2014: the "ICP dose" concept extended to CPP: time-and-magnitude product outside the autoregulation band predicts outcome. 10
• Guiza 2015: dose-response curves for ICP and CPP across hundreds of patients confirm thresholds and time exposure as outcome predictors. 13
• COGiTATE (Tas 2021): 60 adult severe TBI patients randomised to CPPopt vs CPP > 60. Feasibility met (46.5% of CPPopt time in band vs a 36% target), safety established; not powered for outcome, and no mortality benefit shown. 11 21
• Donnelly 2017: MAPopt as the non-invasive analogue of CPPopt; feasible in patients without ICP. 22
• Velle 2023: pediatric severe TBI; CPPopt is computable and deviation of CPP from CPPopt is associated with outcome. 5
• Velle 2023: pediatric CPPopt study (n=57); CPPopt-targeted care is computable in selected children with severe TBI. 5

12. Pitfalls#

• Acting on low-confidence fits. R² < 0.5 or < 4 populated bins = do not titrate; wait for confidence or fall back to floor.
• Flat CPP destroys the data. Aggressive single-target therapy that holds CPP constant prevents the algorithm from finding the curve. Allow modest variability.
• Chasing noise. Minute-by-minute CPPopt changes are within the noise floor; use 30-minute averaging or 1-hour smoothing for titration decisions.
• Transposing adult thresholds. Pediatric CPPopt is often lower than adult values; do not anchor to "CPP > 70" because that is the adult ICU norm.
• Forgetting PbtO2 verification. A confident CPPopt with low PbtO2 (Vignette C) means the perfusion target is not sufficient. Transfuse, oxygenate, deepen sedation before raising CPP further.
• Ignoring the dose curve. Brief excursions outside the band are tolerated; sustained excursions (> 20% of time) predict worse outcome. Document time-in-band as a quality metric.
• Aggressive pressor escalation. Noradrenaline > 0.5 mcg/kg/min carries cardiac and splanchnic risk; consider whether the patient is salvageable rather than escalating indefinitely.
• Loss of CPPopt during exams. Sedation holds, intubation changes, and seizures all break the curve. Pre-empt and document the fall-back to age-floor; do not oscillate the pressor during the gap.

13. Pediatric considerations#

14. Combine with#

• PRx and autoregulation foundations: the mechanism behind the U-curve.
• ICP monitoring: the upstream requirement for PRx and CPPopt.
• PbtO2: the gold-standard tissue verifier.
• NIRS: the non-invasive tissue surrogate.
• TCD / Mx: non-invasive autoregulation cross-validation.
• Osmotherapy ICP NIRS: when CPPopt cannot be achieved with pressors alone.
• PbtO2-CPP titration: the convergent-discordant decision tree.
• PRx vs COx discordance: when two autoregulation indices disagree.
• Discordance triage: the general framework for multimodal disagreement.

15. Evidence summary and recent literature (2022 to 2025)#

Foundational#

Recent literature (2022 to 2025)#

• Tas 2021 COGiTATE: the prospective feasibility RCT in 60 adult severe TBI patients. Feasibility met; not powered for outcome and no mortality benefit shown; the trial that legitimises CPPopt-guided care for further study. 11 21
• Pediatric CPPopt arc: early feasibility (Tas 2021), association of CPP-CPPopt deviation with outcome (Velle 2023), and the prospective STARSHIP multicentre cohort. 4 5 12
• Laws 2022 pediatric MMM review: places CPPopt among specialist-centre modalities for centres with the required signal processing; emphasises the importance of PbtO2 verification. 26
• Smith 2023: pediatric cerebrovascular pressure reactivity has a strong, independent association with outcome in severe TBI (n=196). 15
• Hawthorne 2014 ICP review: still-relevant context on the ICP dose hypothesis and how it primes CPP-dose thinking. 27
• Chesnut 2012 best-practice: the landmark BEST-TRIP trial of ICP-guided vs symptom-guided care (Bolivia / Ecuador); informs the wider question of how aggressively to titrate to MMM-derived targets. 28

16. Self-check#