Perfusion-derived metrics to calculate cerebral infarct growth rate, “ultrafast progressor” phenotype, and neuroprotection

Abstract

Cerebral infarct growth rate (IGR) varies widely in ischemic stroke, and this has important clinical implications. In their recent article, Lin et al. explored IGR characteristics and treatment modification in so-called “ultrafast progressors”. We comment on the study’s methodology for calculating IGR and its interpretation, arguing that perfusion-derived metrics should probably not be adjusted for the time between symptom onset and imaging. Time-independent metrics may better characterize ultrafast progressors by avoiding assumptions about the linearity of infarct growth curves. These results could inform future studies, as ultrafast progressors might benefit the most from neuroprotection interventions.

Dear editor,

We read with interest the article by Lin et al., ¹ which explored the treatment effect of endovascular thrombectomy (EVT) in stroke patients with large CTP-estimated core (≥70 mL), and the interaction of infarct growth rate (IGR) with treatment outcomes in a multicenter, retrospective cohort. Consistent with recent trials, ² they observed a trend toward benefits from EVT, despite most patients experiencing poor functional outcomes at 3-month. ¹ Notably, they showed a high median IGR (44 mL/h) and a treatment-interaction effect of IGR, suggesting that patients with higher IGR, particularly “ultrafast progressors” (>70 mL/h), derived greater benefit from EVT (p-interaction <0.05). ¹ Conversely, the hypoperfusion intensity ratio (HIR), a perfusion metric indicating ischemic severity (severely hypoperfused tissue/total hypoperfused tissue volume), ³ was relatively low (median 0.42–0.43) and not associated with clinical outcomes in their patient sample.

This timely and interesting study provides valuable insights into infarct progression and its clinical relevance, particularly in light of recent positive large core EVT trials. Furthermore, these results could inform future studies on neuroprotective treatments (i.e., interventions to slow down or arrest infarct growth) since, theoretically, the higher IGR, the greater the potential benefit from neuroprotection (Figure 1). ⁴

Figure 1.

Ultrafast progressor phenotypes and neuroprotection benefits. The figure illustrates possible scenarios for various infarct growth patterns, ranging from logarithmic to linear, across the ultrafast to slow progressor spectrum. Neuroprotection aims to decelerate (to flatten) the infarct growth curve between T1 (time of neuroprotection administration) and T2 (time of reperfusion). The benefit is maximized in ultrafast progressors at a specific early interval point where the curve is steeper (the steeper the curve, the more infarct volume can be preserved per minute).

We would like to comment on the methodology used to calculate IGR, and the interpretations of the study results, which, we believe, merit further discussion.

IGR and stroke progression phenotypes

During acute stroke, viable, hypoperfused tissue (ischemic penumbra) is progressively transformed into infarction (ischemic core).^5,6 The rate of this transformation (the IGR) varies widely among patients, influenced primarily by perfusion levels and tissue resilience to ischemic injury. ⁷ In theory, serial imaging (ideally MRI to limit radiation) would be ideal to characterize infarct growth. ⁸ However, this approach is not practical due to limited MRI availability and patient compliance.

Therefore, in clinical routine, the two main approaches to estimating the IGR are: (i) quantifying infarct extension by directly visualizing the ischemic damage (e.g., hypodensity on non-contrast CT or diffusion restriction on MRI) and dividing by the time of symptom onset; or (ii) using perfusion-derived metrics (e.g., HIR) or collateral status assessment. ⁶ The authors of the study adopted the latter approach. By measuring the IGR, one can identify a spectrum of distinct clinical phenotypes, ranging from slow to ultrafast progressors. Lin and colleagues focused on ultrafast progressors in particular. ⁷

Pitfalls when using CT perfusion metrics to determine IGR

The authors calculated the IGR by dividing the CTP-estimated core (volume with relative cerebral blood flow [rCBF] ≤30%) by the time from symptom onset to imaging. First, it is important to note that ischemic “core” on baseline CT perfusion imaging is not necessarily “dead tissue” but rather tissue with a very high (but not 100%) likelihood of irreversible tissue damage even if fast, near-complete reperfusion is achieved. This is why previously, the term “sit-UV” – “severely ischemic tissue of unknown viability” has been suggested to describe the “CTP core”. ⁹ Second, if intravenous thrombolysis is administered, it may promote thrombus fragmentation and migration, which changes the overall volume of tissue at risk and ischemic core and thus may confound any perfusion-based IGR measurement attempt. Third, while the approach the authors took (rCBF <30% volume/time = IGR) is widely used, it remains unclear whether perfusion status dynamically changes over time to reflect infarct growth in ischemic stroke patients. ⁶ Evidence from stroke animal models suggests otherwise, showing that perfusion status remains relatively stable as infarct grows over time. ⁵ Additionally, ischemic core is often overestimated and underestimated in the early and late windows, respectively, suggesting that the accuracy of particular CTP core thresholds varies at different time points. ¹⁰ Therefore, perfusion status may serve as a surrogate for IGR (i.e., infarct volume divided by time), and adjusting it over time introduces conceptual inconsistencies (i.e., infarct volume divided by time, divided again by time). Indeed, HIR, another perfusion-derived metric explored by the authors, is not time-adjusted when used to estimate IGR.

Prognostic value and treatment effect of IGR

Dividing the CTP-estimated core by time results in two major problems: (i) it may lead to an overestimation of the IGR in early presenters, and (ii) it closely links IGR to the time of onset—early presenters will inherently exhibit a higher IGR. As a result, it could be that the so-called “ultrafast progressors” may simply be “ultra-early presenters”.

When interpreting the study results through this lens, the results might simply show that patients presenting earlier have better outcomes and greater treatment benefits than those presenting later. The determining factor might not be IGR (ultrafast vs. non-ultrafast progressors) but time (early vs. late presenters). This would also align with well-established evidence that time elapsed from stroke onset to reperfusion is the most critical determinant of outcomes in patients who achieve successful recanalization. ¹¹

Hypoperfusion intensity ratio (HIR) as a measure of IGR

As explained by Lin and colleagues, HIR represents the ratio between severely hypoperfused tissue to total hypoperfused tissue, calculated as Time-to-maximum (Tmax)>10-second volume/Tmax > 6-second or delay time (DT) >6-second volume/DT > 2-second volume. It is widely used to classify progressor phenotypes in ischemic stroke,^3,12 whereby higher HIR indicates a larger proportion of severely hypoperfused tissue and, thus, a higher IGR. Most studies define fast progressors as patients with an HIR ≥0.5. ⁶

Lin et al. reported a relatively low HIR in their study (median = 0.43–0.44), despite CTP-estimated core comprising more than half of the total hypoperfused tissue (EVT group: median core 113 mL, median penumbra 74 mL; no-EVT group: 95.3 mL and 91 mL). ¹ Brain tissue volumes with severe hypoperfusion should be rather extensive in large core patients, such as those in the study sample, and thus, the median HIR would be expected to be substantially higher than 0.50. The rather low median IGR in the study is somewhat surprising and raises questions about the generalizability of its results regarding HIR, especially since no association between HIR with outcomes or treatment effect was observed, contradicting most prior research.

Ultrafast progressors have non-linear IGR patterns

Finally, as discussed by the authors, the precise dynamics of IGR in ischemic stroke remain unclear. Emerging evidence suggests a linear growth pattern in slow progressors, progressively transitioning to a logarithmic trajectory in fast progressors, where infarct progression is most rapid at onset and gradually slows over time (Figure 1).^5,6 Therefore, we believe that infarct growth in ultrafast progressors is unlikely to follow a linear trajectory.

Perfusion-derived metrics, that are independent of time, such as HIR, are not constrained by assumptions about infarct growth curves. As such, they could better characterize ultrafast progressors.