Impact of Glucose Trajectories on Outcomes After Intracerebral Hemorrhage: The ATACH-2 trial

This secondary analysis of the ATACH-2 trial demonstrates that early post-ICH hyperglycemia is significantly associated with poor functional outcomes, while late and persistent hyperglycemia are linked to worse survival rates, highlighting the prognostic value of continuous glucose monitoring and trajectory-based glycemic control strategies.

The Gist

Imagine your brain is a house that has just suffered a sudden, violent break-in (a stroke, specifically a brain bleed). The police (doctors) arrive to secure the scene, stop the bleeding, and manage the chaos. For a long time, doctors have known that if the "power grid" of the house (your blood sugar) is too high right when they arrive, things tend to go badly.

But this new study asks a different, more nuanced question: It's not just about the power surge at the moment of the break-in; it's about how the power grid behaves over the next three days.

Here is the story of the study, broken down into simple concepts:

The Setup: The ATACH-2 "House"

The researchers looked at data from 1,000 patients who had a brain bleed. They didn't just check their blood sugar once when they walked into the hospital. Instead, they checked it like a security camera at four specific times: Right now (0 hours), tomorrow morning (24h), the next day (48h), and the third day (72h).

They wanted to see the "story" of the sugar levels, not just a single snapshot.

The Four Characters (Glucose Trajectories)

Based on how the sugar levels moved over those three days, the researchers sorted the patients into four groups, like characters in a play:

1. The "Early Flash" (Early Hyperglycemia): These patients had a sugar spike right at the beginning (0h), but it went back to normal by day 3.
   • Analogy: Think of a car that revs its engine loudly when you first turn the key, but then idles smoothly.
2. The "Late Bloomer" (Late Hyperglycemia): These patients started with normal sugar, but it started creeping up and stayed high by day 3.
   • Analogy: A car that starts fine, but the engine starts overheating as the drive continues.
3. The "Steady Burn" (Persistent Hyperglycemia): These patients had high sugar from the start and it never went down.
   • Analogy: A car that is overheating from the moment you turn the key and stays that way the whole trip.
4. The "Smooth Sailer" (Persistent Normoglycemia): These patients kept their sugar levels normal the entire time.
   • Analogy: A car that runs perfectly cool and efficient from start to finish.

The Plot Twist: Different Problems for Different Groups

The study found that when the sugar went high mattered just as much as how high it went.

• The "Early Flash" Group: Even though their sugar went back to normal, having that initial spike was a bad sign for their long-term recovery. These patients were less likely to walk out of the hospital on their own two feet 90 days later.
  • The Metaphor: It's like a car that had a rough start; even if it cools down later, the initial jolt damaged the transmission, making the car less reliable in the long run.
• The "Late Bloomer" Group: These patients had the worst survival rates. They were the most likely to die within 90 days. They also had the highest rate of the brain bleed getting bigger (hematoma expansion).
  • The Metaphor: This is like a car that seems fine at first, but then the engine catches fire while you are driving. It's a delayed disaster that is very hard to stop.
• The "Smooth Sailors": This group had the best outcomes. They were the most likely to survive and the most likely to recover their independence.

Why Does This Happen? (The Science in Plain English)

The authors explain that high sugar is like pouring gasoline on a fire.

• Early High Sugar: This is often the body's "fight or flight" panic response to the trauma. It's a stress reaction.
• Late/Persistent High Sugar: This suggests the body is losing control. The brain is inflamed, and the high sugar makes that inflammation worse, causing more swelling and damage to the delicate brain tissue. It's like the fire department arrived, but the fuel keeps getting added to the fire for days.

The Big Takeaway

For a long time, doctors thought, "If the sugar is high when the patient arrives, we treat it. If it's normal, we're good."

This study says: That's not enough.

Just because the sugar is normal at the door doesn't mean it will stay that way. Just because it was high at the door doesn't mean it will stay high. Doctors need to watch the movie, not just the trailer.

• If the sugar spikes early, watch out for poor recovery.
• If the sugar spikes later, watch out for survival risks.

The Bottom Line

Managing blood sugar after a brain bleed isn't a "one-and-done" task. It requires continuous monitoring, like keeping an eye on a simmering pot. If you only check the pot once, you might miss the moment it boils over. This study suggests that keeping a steady eye on glucose levels for the first three days could help doctors predict who needs extra help and potentially save more lives.

Technical Summary

1. Problem Statement

Intracerebral hemorrhage (ICH) carries a high mortality rate (>30% at 30 days) and significant morbidity. While admission hyperglycemia is a known predictor of poor outcomes in ICH, relying on a single glucose measurement fails to capture the dynamic metabolic stress that occurs post-injury. Current guidelines recommend glucose monitoring but lack specific targets or protocols for managing glucose trajectories (changes over time). There is a gap in understanding how the temporal pattern of hyperglycemia (early, late, or persistent) specifically influences functional recovery, hematoma expansion, and survival in ICH patients.

2. Methodology

• Study Design: A secondary analysis of the ATACH-2 trial (Antihypertensive Treatment of Acute Cerebral Hemorrhage II), a prospective, multicenter, randomized trial involving 1,000 patients with acute ICH presenting within 4.5 hours of symptom onset.
• Data Source: Data was retrieved from the NINDS biorepository. The study was exempt from IRB review.
• Glucose Trajectory Classification:
  • Timepoints: Blood glucose levels were measured at 0h (baseline), 24h, 48h, and 72h.
  • Definition: Hyperglycemia was defined as $\ge$140 mg/dL.
  • Modeling: A Linear Mixed Effects (LME) model was employed with fixed effects for time and random intercepts/slopes for individual patients.
  • Grouping: Using patient-specific estimates from the LME model, patients were categorized into four trajectory groups based on predicted glucose levels at 0h and 72h:
    1. Early Hyperglycemia: High at 0h, normal at 72h.
    2. Late Hyperglycemia: Normal at 0h, high at 72h.
    3. Persistent Hyperglycemia: High at both 0h and 72h.
    4. Persistent Normoglycemia: Normal at both timepoints (Reference group).
• Outcomes:
  • Primary: 90-day functional outcome (modified Rankin Scale [mRS] 0–2 = good; 3–6 = poor).
  • Secondary: 90-day mortality, hematoma expansion ($\ge$33% increase), and perihematomal edema expansion ($\ge$40% increase).
• Statistical Analysis:
  • Univariate analysis (Chi-square, Mann-Whitney U).
  • Survival analysis using Kaplan-Meier curves and log-rank tests.
  • Multivariable logistic regression (for functional outcomes) and Cox proportional hazards regression (for mortality), adjusted for demographics, comorbidities, stroke severity (NIHSS), and treatment arm.
  • Subgroup analyses were performed for diabetic vs. non-diabetic patients and treatment arms.

3. Key Contributions

• Trajectory-Based Approach: Moves beyond static admission glucose to characterize the dynamic evolution of glucose levels over the first 72 hours using advanced statistical modeling (LME).
• Differential Impact: Demonstrates that the timing of hyperglycemia matters. Early hyperglycemia is distinctively linked to functional disability, while late/persistent hyperglycemia is more strongly linked to mortality and hematoma expansion.
• Diabetes Stratification: Highlights that the prognostic value of stress-induced hyperglycemia differs significantly between diabetic and non-diabetic populations, with non-diabetics showing stronger associations between hyperglycemia trajectories and poor outcomes.

4. Results

• Cohort Distribution: Of 1,000 patients (median age 62, 38% female):
  • Persistent Normoglycemia: 63.5% (635 patients)
  • Persistent Hyperglycemia: 22.5% (225 patients)
  • Early Hyperglycemia: 8.1% (81 patients)
  • Late Hyperglycemia: 5.9% (59 patients)
• Functional Outcomes (mRS 0–2 at 90 days):
  • Persistent Normoglycemia: 45.8% favorable outcome.
  • Hyperglycemic Groups: 30.9% (Early), 30.5% (Late), and 32.0% (Persistent) favorable outcomes ($p < 0.001$).
  • Multivariable Analysis: Early hyperglycemia remained significantly associated with poor functional outcomes after adjustment (OR 2.27, 95% CI 1.10–4.68, $p=0.026$).
• Survival and Mortality:
  • Late hyperglycemia had the lowest survival probability (86%) and the highest rate of hematoma expansion (35.3%, $p=0.029$).
  • In Cox regression, Late and Persistent hyperglycemia groups had significantly higher hazard ratios for 90-day mortality compared to the normoglycemic group.
• Subgroup Findings:
  • Non-Diabetics: The association between early/persistent hyperglycemia and poor functional outcomes remained significant.
  • Diabetics: No statistically significant association was found between hyperglycemia trajectories and outcomes, likely due to smaller sample size (18.6% of cohort) and potential cellular adaptation to chronic hyperglycemia.

5. Significance and Clinical Implications

• Risk Stratification: A single admission glucose measurement is insufficient for prognostication. Clinicians should monitor glucose trajectories to identify patients at risk for specific adverse events (e.g., early hyperglycemia predicts functional disability; late hyperglycemia predicts mortality).
• Pathophysiological Insight: The findings suggest distinct mechanisms:
  • Early hyperglycemia may reflect an acute sympathetic stress response (catecholamine surge) that correlates with initial injury severity.
  • Late/Persistent hyperglycemia may indicate ongoing metabolic derangement, contributing to secondary brain injury, edema, and hematoma expansion.
• Therapeutic Window: The data supports the need for continuous glucose monitoring and targeted interventions during the subacute phase (24–72 hours) of ICH, rather than just at admission.
• Future Directions: The authors call for prospective trials to determine if maintaining normoglycemia specifically during the 24–72 hour window can mitigate secondary brain injury and improve recovery, noting that previous trials (like SHINE) in ischemic stroke did not show benefits, suggesting ICH may require different management protocols.

Limitations Noted: The study is a secondary analysis (not powered for glucose outcomes), lacks standardized hyperglycemia treatment protocols across centers, excludes severe ICH cases (low GCS), and relies on intermittent rather than continuous glucose monitoring.