The ischemic penumbra: From concept to reality

You can see by reading that this issue has been known for decades and NO ONE IS WORKING ON A SOLUTION. That is how fucking bad the stroke medical world is.  NO STRATEGY, NO LEADERSHIP.

The ischemic penumbra: From concept to reality

Show all authors

Charlotte M Ermine, Andrew Bivard, Mark W Parsons*, ...

First Published December 1, 2020 Review Article 

https://doi.org/10.1177/1747493020975229

Article information

   

Abstract

The discovery that brain tissue could potentially be salvaged from ischaemia due to stroke, has led to major advances in the development of therapies for ischemic stroke. In this review, we detail the advances in the understanding of this area termed the ischaemic penumbra, from its discovery to the evolution of imaging techniques, and finally some of the treatments developed. Evolving from animal studies from the 70s and 80s and translated to clinical practice, the field of ischemic reperfusion therapy has largely been guided by an array of imaging techniques developed to positively identify the ischemic penumbra, including positron emission tomography, computed tomography and magnetic resonance imaging. More recently, numerous penumbral identification imaging studies have allowed for a better understanding of the progression of the ischaemic core at the expense of the penumbra, and identification of patients than can benefit from reperfusion therapies in the acute phase. Importantly, 40 years of critical imaging research on the ischaemic penumbra have allowed for considerable extension of the treatment time window and better patient selection for reperfusion therapy. The translation of the penumbra concept into routine clinical practice has shown that “tissue is at least as important as time.”

Keywords Stroke, ischemic stroke, penumbra, positron emission tomography, computed tomography scan, magnetic resonance imaging

Introduction

A major scientific discovery underpinning modern treatment for ischemic stroke was made four decades ago, with the identification of a region around the irreversibly injured core of infarction, where neurons were not functioning but could still be saved if perfusion was restored promptly. This region was coined the “ischemic penumbra”.¹ In this review, we will describe the evolution of our understanding of the ischemic penumbra from its discovery to the treatments currently available.

Ischemic penumbra: Discovery and definition

Animal studies have played a major role in understanding the evolution of brain tissue following ischemia. In 1973, Hossmann and Kleihues demonstrated following global ischemia in cats and monkeys that neuronal function could recover and survive an extended time past total oxygen depletion under suitable conditions.² In 1974, Symon et al. showed in a middle cerebral artery occlusion (MCAo) model of focal stroke in baboons, that in ischemic tissue, there were large variations in the reduction of cerebral blood flow (CBF), suggesting that brain tissue was being supplied with blood from more than one route (subsequently termed collaterals). This finding gave the first indication that a treatment post-stroke was possible, through salvaging the tissue at risk of infarction.³ In 1977, Astrup et al. using the same model showed that failure of electrical activity was not uniform across an ischemic region, suggesting that some tissue remained electrically active despite severe ischemia, and defined ischemic thresholds based on CBF levels.⁴ They identified three regions: (i) below 20 ml/100 g/min, where electrical function of the tissue is affected; (ii) below 15 ml/100 g/min, where electrical failure is complete; and (iii) below 5 ml/100 g/min, where release of extracellular K+ attests to impending cell death (Figure 1(a)). Furthermore, they showed that increasing CBF could restore evoked potential and normalize extracellular K+. Additional animal studies have confirmed the presence of salvageable tissue (subsequently called the penumbra) and have further refined the CBF threshold corresponding to the infarct core and to the ischemic penumbra^5–7 (Figure 1(b)). One of those studies by Jones et al. in 1981 used histological evidences correlated with electrode recording sites to validate the penumbra threshold of 20 ml/100 g/min. They further showed that the core threshold, in contrast to the penumbra threshold was time dependent, which was later confirmed in three recent clinical studies.^8–11 It is important to note that throughout this review, we will be referring to both infarct and ischemic core. These terms may not be strictly equivalent, as they differ from their mode of acquisition, namely histology and perfusion/diffusion/cerebral metabolic rate of oxygen (CMRO₂) imaging respectively.

Figure 1. Representation of the ischaemic thresholds from two studies: (a) Figure modified from Astrup et al. (1977), represents ischaemic thresholds for electrical failure and K+ release. (b) Cartoon modified from Jones et al. (1981), illustrating the CBF x Time interaction, where paralysis represents the ischaemic penumbra and infarction representing the core. Source: reproduced with permission from Astrup et al., 1977⁴ and Jones et al., 1981.⁷

The first definition of the ischemic penumbra was given by Astrup and Symon in 1981 as: “the region of reduced CBF with absent spontaneous or induced electrical potentials that still maintained ionic homeostasis and transmembrane electrical potentials”.¹² From there, it was suggested that the penumbra was ischemic tissue that was potentially reversible with timely restoration of perfusion, with the key point that treatment of stroke should focus on salvaging this tissue.¹³ CBF in the penumbral region is gradually increased as we move away from the core, which explains the progressive death of tissue with time, and therefore the possibility of successful treatment being time-dependent.^7,14–16 This was shown by positron emission tomography (PET) where CBF, CMRO₂, oxygen extraction fraction (OEF), and cerebral metabolic rate of glucose (CMRglc) were directly measured few hours post MCAo in baboons and later repeatedly measured over 24 h following MCAo in another study in cats.^15,17 This revealed initially a reduction in CBF and an increase of OEF in the ischemic core, and a decrease in CBF, CMRglc, CMRo2, and OEF in the periphery, with sequential studies showing that the core expands to the outer regions of the penumbra.¹⁵ This was confirmed one year later in a baboon model, where the volume of severely hypometabolic tissue measured via sequential PET was shown to be stable for the first few hours following insult, but then had enlarged at 24 h post-ischemia and continued to grow in subsequent days.¹⁶ Consistent with the penumbra concept, reperfusion at 6 h reversed this process.¹⁸

Thus, animal studies permitted the discovery of the existence of the ischemic penumbra, a tissue that was destined to die with time, but that did not immediately die in the first hours after vessel occlusion. These advances were crucial in identifying targets for rational treatment of stroke.

Identifying the ischemic penumbra: Evolution of imaging techniques

Following the discovery in animal studies of the ischemic penumbra, the next step was to document the existence of, and map the penumbra in man, as it was widely considered at that time that animal studies did not apply to human stroke. To work toward this goal, researchers investigated various methods of imaging the penumbra, from combined perfusion and metabolic imaging using PET to sole perfusion imaging including computed tomography (CT) and magnetic resonance (MR). When reviewing the historical evolution of imaging to define core and penumbra in humans, it is important to note that the penumbra threshold does not significantly vary with time after stroke onset.^19,20 However, the infarction (core) threshold is time-dependent and by definition has varied from study to study given the variability (and uncertainty) in timing of reperfusion historically. Indeed, only with thrombectomy (and hence known time and extent of reperfusion) it was established in human studies that earlier reperfusion can salvage more severely hypoperfused tissue. The current definitions of the ischemic penumbra according to each imaging techniques are summarized in Table 1.

Table 1. Summary of the definitions of the ischemic core and penumbra according to the different imaging techniques presented in this review

Table 1. Summary of the definitions of the ischemic core and penumbra according to the different imaging techniques presented in this review

View larger version

Positron emission tomography

PET is a quantitative imaging technique that uses radiotracers to measure regional cerebral blood volume (CBV), CBF, OEF, CMRglc, and CMRO₂, as well as regional binding of radioligands to neuroreceptors.²¹ These parameters allow the identification of the penumbra and its distinction from the core and from oligemia for instance, the moderately hypoperfused tissue that is functional and normally survives the insult (Figure 2(a)). The physiologic profiles of penumbral tissue after stroke have been characterized using CBF, OEF, and CMRO₂.^22–24

Figure 2. Identification of the ischaemic penumbra using different imaging techniques. (a) Using PET measurement of CBF, OEF and CMRO2, where the penumbra (red arrow) is identified by reduced CBF, increased OEF but relatively preserved CMRO₂ apart from a moderate reduction in the basal ganglia area. Image from Heiss et al. 2017¹⁰⁰. (b) Using DWI-PWI mismatch (image from Ebinger et al. 2009), the lesion from DWI is shown by the arrow, the perfusion deficit on PWI is shown in red and the ischeamic penumbra defined by the DWI/PWI mismatch is shown in blue¹⁰¹. (c) Using CTP (image from Bivard et al. 2013), which allows the creation of maps for CBF, CBF, MTT, TTP, Tmax and delay-time, to identify the infarction core in red and the ischaemic penumbra in green⁹. Source: reproduced with permission from Heiss and Weber, 2017.,¹⁰⁰ Ebinger et al., 2009,¹⁰¹ and Bivard et al., 2013.⁹ PET: positron emission tomography; CBF: cerebral blood flow; OEF: oxygen extraction fraction; CMRO2: cerebral metabolic rate of oxygen; DWI: diffusion-weighted imaging; PWI: perfusion-weighted imaging; CBV: cerebral blood volume; CBF: cerebral blood flow; MTT: mean transit time; TTP: time to peak; Tmax: time to maximum.

The use of PET in the early 1980s allowed the identification of a hypoperfused region called “misery perfusion”, with preserved CMRO₂ and increased OEF.^25–27 Using quantitative voxel-based mapping of these variables, this region was later validated as the ischemic penumbra. The quantitative regional measurements of CMRO₂ and of CBF were found to accurately distinguish viable from non-viable tissue. The penumbra threshold, which separates the penumbra from the oligemia, was found to be around 20 ml/100 g/min²⁸ and the threshold for core, which separates the penumbra from the core, has been found to sit around 8 ml/100 g/min, which were carried out generally > 3 h from onset. However, lower thresholds characterize the core at earlier timepoints, as it was in the animal studies. For instance, Heiss et al. showed that the volume of tissue with CBF < 12 ml/100 g/min was still salvageable within 3 h from onset, and the volume of salvaged penumbra correlated with neurological improvement (shown earlier by Furlan et al.).²⁹ Baron and his group used a cohort of 30 patients with first ever middle cerebral artery stroke and performed PET within 18 h of onset and at one month, with co-registration of follow-up infarct from one-month CT scan. They replicated in man the Astrup and Symon model with identification of three distinct tissue types: the core, the ischemic penumbra, and the oligemia.^28,30,31 Using their paradigm and applying the classical threshold of CMRO₂ for irreversible damage set at 1.4 ml/100 g/min, they determined the penumbra CBF threshold in man and successfully mapped the penumbra.³²

The identification of the penumbra using PET was further refined to include three criteria: (1) have misery perfusion, which is defined with a high OEF value, and a partially preserved CMRO₂; (2) an undetermined outcome of the affected tissue, toward necrosis or survival; and (3) clinical correlation, namely initial neurological deficit proportional to volume of (core + penumbra), and neurological recovery at one month correlated with the volume of surviving penumbra.³³ These two clinical correlations are of utmost importance, as they indicate that (i) the volume of core expresses the already irrecuperable neurological deficit, regardless of any sort of therapy including recanalization; and (ii) targeting the penumbra with appropriate therapy, namely timely recanalization, is the key to improve the outcome of acute stroke patients. Baron’s group further identified four PET patterns of increasing severity that characterized the ischemic brain: (1) an isolated increase in CBV, maintaining the CBF; (2) an increase in OEF in response to the reduction of CBF but with a maintained CMRO₂, which corresponds to the oligemia; (3) a marked increase in OEF in regions with reduced CBF and CMRO₂, maintaining tissue metabolism, which corresponds to the ischemic penumbra; and (4) very low CBF and CMRO₂ with variable OEF, which corresponds to the ischemic core.³³ Furthermore, while the presence of extensive penumbra was not associated with functional outcome (reflecting the uncertain fate of the penumbra), the pattern of extensive core invariably predicts poor functional outcome, and the pattern of extensive hyperperfusion invariably predicts excellent spontaneous outcome.³¹ This key finding led the authors to advocate the use of core/penumbra imaging to triage stroke patients for individualized management and treatment.^31,34 Finally, PET studies documented that substantial volumes of penumbra persisted up to 18 h³² and perhaps even beyond,³⁵ suggesting that delayed treatment targeting the penumbra, including recanalization, could be considered in some patients selected based on core/penumbra imaging.³⁶

Additionally, PET can also map the binding of specific radioligands to neuroreceptors, which proved useful in identifying the ischemic penumbra. For example, ¹¹C-Flumazenil (FMZ), which labels the benzodiazepine receptor, has been successfully used to differentiate infarcted from penumbral tissue within 3 h of stroke onset.^37–39 FMZ binding was reduced to < 3.4 times the mean value of normal tissue for tissue that did not benefit from reperfusion, thus allowing prediction of irreversible tissue damage. Another radiotracer, ¹⁸F-fluoromisonidazole (FMISO), which irreversibly binds to hypoxic cells, was validated in a rat model of MCAo for use in PET imaging to define penumbral tissue.^40,41 FMISO was able to identify hypoxic tissue in the first 48 h following stroke onset.^42,43 However, a recent pilot study has suggested that FMISO is unable to differentiate the penumbra from the (still hypoxic) core.⁴⁴

PET imaging is a very powerful research tool for mapping the penumbral region and the core, and the original PET studies laid the platform for the advances we see now in routine acute stroke imaging and treatment. However, PET has limitations, including the complexity of execution, relatively poor spatial resolution, the cost, limited access, and the long time necessary to produce the tracers with a cyclotron and dedicated hot chemistry due to the short half-lives of the positron emitters, and perform the measurements, making the use of PET in acute clinical settings impractical and is rarely used clinically today.

Single photon emission computer tomography

Single photon emission computer tomography (SPECT) uses gamma rays to detect radiotracers injected, most commonly ^99m-Technetium hexamethylpropyleneamineoxime (^99mTc-HMPAO), and renders a 3D mapping of cerebral perfusion. Its advantages over PET are the ease of production and transport of the tracers, and its disadvantages the inaccurate quantification and poorer spatial resolution due to scattering.

The ischemic penumbra was identified with SPECT within 3–6 h of stroke insult, as the region with a signal tracer of 40–70% that of the contralateral side.^21,45 The final infarct size was also accurately predicted by a SPECT imaging within 6 h of symptoms.^46–48 However, while the penumbra was identifiable using SPECT technology, early studies found that SPECT imaging of regional hypoperfusion failed to show any advantages in prediction of stroke outcome over the initial clinical evaluation.^49,50 Nevertheless, one study found that SPECT imaging within 3–6 h of stroke onset using ^99mTc-HMPAO was useful to identify patients at high risk of symptomatic hemorrhagic transformation before reperfusion treatment.⁵¹ Later studies did show that ^99mTc-HMPAO-SPECT was capable of predicting clinical outcome, and the use of ^99mTc-ECD, another radiotracer, was useful in distinguishing transient ischemic attacks from stroke and patients with massive infarction.^52–54 However, it became clear that using CBF mapping alone with SPECT was insufficient to distinguish the penumbra from the core. Accordingly, it was subsequently reported that the combination of SPECT imaging for detection of hypoperfused volumes around the core and magnetic resonance imaging (MRI) for detection of infarcted tissue was able to predict infarct growth and clinical outcome.⁵⁵

SPECT is cheaper and more available than PET; however, it is not used in routine clinical care due to limitations, including the limited availability of the radiotracers, the length of data acquisition, the difficulties in data analysis, and the coarse spatial resolution.

Perfusion weighted and diffusion weighted MRI

MRI is widely used in acute stroke in clinical practice, including to image the penumbra. This technique applies powerful magnetic fields to identify molecular signatures, notably from hydrogen. Various images can be generated, including: (i) structural T1- and T2-weighted; (ii) MR angiography of the intracranial vessels; (iii) T2* imaging, which allows the detection of recent or old hemorrhages; (iv) diffusion-weighted imaging (DWI), which uses the movement of hydrogen in water molecules to generate maps of the apparent diffusion coefficient (ADC); and (v) perfusion-weighted imaging (PWI), which is a bolus tracking technique that uses the injection of a contrast agent (generally gadolinium), to produce CBV, CBF, and other perfusion maps.⁵⁶ These acquisitions only take few minutes, but are highly sensitive to patient motion, and MRI in general requires extensive pre-screening for safety making them less ideal in the acute setting.

The interest in MRI for acute stroke dates from the development of DWI. The principles of DWI relate to the reduction in ATPase activity that occurs almost immediately after onset of severe ischemia, which then causes a redistribution of water from the extracellular space to the intracellular space, this leads to “restricted diffusion” and reduction in the ADC, which correlates with irreversibly damaged tissue. An early DWI study in a rat model of MCAo used the comparison of T2-weighted images and DWI for early detection of ischemia, with the assumption that hyperintensity on DWI = irreversible injury (now termed ischemic core).⁵⁷ A few years later, it was found that the comparison of PWI and DWI was a better predictor of ischemic penumbral tissue than conventional T1 and T2 images.⁵⁸ It was reported that PWI within 6 h of stroke onset had a sensitivity of 95% and a specificity of 100% in detecting the salvageable tissue and that intermediate ADC values corresponded to penumbral tissue.^59,60 However, it become clear that the DWI lesion did not always reflect only ischemic core. DWI lesions have been shown to be reversible with early reperfusion, although in a fraction of cases, this early reversal might be temporary with the tissue ultimately becoming infarcted.^61–65 Early work with PWI suggested that the ischemic penumbra was shown to correspond to the region with a mean increase of 73% in mean transit time (MTT) of the gadolinium bolus and with a 29% increase in relative cerebral blood volume (rCBV),⁶⁶ although others suggested relative CBF was more accurate.⁶⁷

The “perfusion-diffusion mismatch” concept was coined by Warach et al. and immediately became popular.⁵⁸ This concept, derived from PET, dictates that salvageable tissue corresponds to the difference between the smaller diffusion lesion and the larger perfusion deficit (Figure 2(b)).^68–70 Consistent with previous animal and human work summarized above, mismatch incidence decreases with time, from 75% at 6 h to 44% at 18 h post-stroke onset.⁷¹ While these techniques have been extensively used to detect the ischemic penumbra, it was later suggested that the PWI/DWI mismatch region could be much larger than the true penumbra.⁷² In fact the rim of the perfusion lesion reflected the oligemic tissue seen on PET (due to excellent collateral blood supply) rather than the penumbra, which makes the PWI/DWI mismatch less accurate than originally described.^73–75 Further efforts were made to make the PWI lesion more specific for penumbra and core by directly validating PWI against PET studies of CBF,^76,77 but also by improving perfusion algorithms,^78,79 as well as by applying stricter and validated perfusion thresholds.⁸⁰ Currently, a time to maximum (Tmax) of > 6 s is most accurate in delineating the penumbra from the core.^41,81

More recently, some research groups have focused on the use of susceptibility-weighted imaging (SWI), which maps the differences in magnetic susceptibility of deoxygenated blood, blood products, iron, and calcium. Using this technique, the positive DWI/SWI mismatch has also been shown to represent severely hypoxic tissue, consistent with the ischemic penumbra, although again, this is not widely used due to complex and lengthy MRI sequences.^82,83

Arterial spin labeling (ASL) is another perfusion MRI technique that can quantify CBF without the injection of a contrast agent. The accuracy of ASL at detecting the infarcted regions was shown on a small subset of patients with acute ischemic stroke, where the hypoperfused regions obtained from ALS were compared to those obtained from dynamic susceptibility contrast on PWI and found to be consistent.⁸⁴ Furthermore, a later study by Bivard et al. determined the ALS–CBF threshold (set at 40%) that accurately identified the penumbral tissue through ASL–DWI mismatch, and was shown to be specific and sensitive in comparison to PWI–DWI mismatch.⁸⁵ The advantage of using ASL for ischemic penumbra identification is that it allows for rapid and non-invasive quantitative measurement of CBF but spatial resolution is relatively poor and there are problems with loss of signal if there is long delay between labeling of the flowing blood and arrival to ischemic tissue, due to large vessel occlusion.

While MRI techniques are powerful in detecting acute ischemia (and hemorrhage), it is relatively time consuming, not always easily accessed quickly from the emergency room, and has several contraindications (e.g. metallic inserts such as pacemakers or claustrophobia). Nonetheless, it is used for routine hyperacute clinical imaging (prior to treatment decisions) in many centers worldwide, particularly in Europe.

Computed tomography

There are different uses of CT in acute ischemic stroke, including non-contrast computed tomography (NCCT), CT angiography, and computed tomography perfusion (CTP). NCCT can identify recent hemorrhages and may distinguish ischemic tissue (so called early ischemic signs) as well as reveal non-stroke conditions (e.g. brain tumors) and for this reason is widely used in the acute stroke setting, but it cannot directly identify the penumbra.⁸⁶ Overall NCCT was shown to have poor sensitivity to ischemia, as compared to DWI.⁸⁷ NCCT can display two types of ischemic changes: parenchymal attenuation and focal swelling. Although NCCT does not differentiate specifically core and penumbra, a comparison with CTP has shown that the parenchymal attenuation likely corresponds to the core and the isolated focal swelling, likely corresponds to the penumbra, as it is a region with elevated CBV.^86,88

The evolution of CT scanners to multidetector allowed for imaging of the whole brain rapidly and the development of CTP. CTP uses an iodinated contrast agent and X-ray to measure CBV, CBF, MTT, and is comparable to PWI. Thanks to the semiquantitative CBV and CBF maps generated, CTP is very sensitive in identifying the core; however, it is not as specific for the differentiation of core and penumbra.^89–91 Two CTP parameters are sensitive to penumbral identification, Tmax and delay time (DT). Indeed similar to PWI, a Tmax > 6 s can estimate hypoperfused tissue with CBF < 20 ml/100 g/min, i.e. the ischemic penumbra.⁹² A DT of > 2 s was also shown to accurately represent the penumbra, and when associated with CBF < 40% was shown to represent the core, but subsequently DT > 3 s and core < 30% has been shown to be more specific (Figure 2(c)).^8,9 A caveat is the time-dependence of the core threshold, which was well known from animal studies⁷ but has been documented in man only recently,^10,11,93,94 pointing to the need to adjust the core threshold to time elapsed since stroke onset if a more accurate picture of the physiological situation is desired. To objectively select patients who can benefit from thrombolysis and fasten the process, automated measurement of MTT, Tmax, and CBF were developed, through a mathematical processing called deconvolution, with several variations.⁹⁵

Xenon-enhanced CT is another CT technique that was used in the early days of CT imaging, and requires the inhalation of xenon. The CBF is measured as positively proportional to the absorption of xenon by the tissue. This measurement of CBF was proven accurate in baboons and was further capable of delineating the ischemic penumbra.^96–98 However, this technique has a huge limitation with the use of xenon, which induces side effects, notably sedation.

MR and CT perfusion have similar accuracy in identifying key perfusion thresholds such as Tmax and DT, although CBF and CBV are less comparable.⁹⁹ However, MRI has the clear advantage to CT in that it uses a different modality (DWI) to measure core, whereas CT relies on perfusion measures such as CBF or CBV. Nevertheless, due to the wide availability of CT scanners, CT currently is the typical assessment of acute stroke for decision-making in most centers around the world.

Salvaging the ischemic penumbra: Currently available treatments

The aim of ischemic stroke therapy is to reperfuse the penumbra and salvage as much brain tissue as possible to unsure a better clinical outcome. Currently, intravenous injection of tissue plasminogen activator (tPA) is the gold standard medical treatment for ischemic stroke. Another highly successful reperfusion treatment is mechanical thrombectomy for large vessel occlusion. For both, earlier treatment increases the chance of benefit (the “time is brain” mantra) but it has become increasingly apparent that to focus on stroke onset time to guide decisions is oversimplistic. Some patients have little to gain from reperfusion treatment early after stroke onset and others stand to gain many hours later. This relates to size of core and penumbra, and, is underpinned by the collateral supply to the tissue.^102,103 The comparison of standard clinical predictors (i.e. onset-to-treatment time) with CTP imaging measurements, such as ischaemic core and penumbral volumes, showed that imaging parameters (especially infarct core volume) were better predictors of good or bad clinical outcome following thrombolytic treatment than time to treatment, and, that CTP improves the identification of patients who can benefit from tPA from those who are less likely to^106,107.

Several clinical trials have shown the clinical benefits of salvaging the penumbra in patients suffering from stroke and further identified imaging parameters that could predict the outcome of a given patient from thrombolytic therapies.^31,104,105 The comparison of standard clinical predictors, such as onset-to-treatment time, with CTP imaging measurements, such as ischemic core and penumbral volumes, showed that imaging parameters (especially infarct core volume) were better predictors of good or bad clinical outcome following thrombolytic treatment than time to treatment, and, that CTP improves the identification of patients who benefit from tPA from those who are less likely to.^106,107

Clinical trials have used the different imaging mentioned above to visualize the core and penumbra and to extend the time windows for reperfusion treatment. DEFUSE, an observational study involving 74 patients and EPITHET, a randomized controlled trial, used PWI/DWI mismatch to examine the time window for tPA to 6 h. They suggested (although not definitively due to too lenient perfusion thresholds) that tPA improved clinical outcome and salvaged the penumbra between 3 and 6 h of stroke onset.^108,109 Desmoteplase, a thrombolytic agent derived from bat saliva, was investigated for its therapeutic potential following ischemic stroke. However, mostly due to methodological problems with perfusion CT, the phase 3 trial (DIAS-2) failed to show a benefit for desmoteplase in the 3–9 h window.¹¹⁰ The aforementioned trials from the early 2000s suffered from lack of standardization of core and penumbral assessments. The development of automated core and penumbral volumetric software (including RAPID (iSchemaView, Menlo Park, CA, USA)) contributed to the success of later trials using PWI/DWI and/or CTP patient selection. These trials, including DEFUSE 3, DAWN, EXTEND, EXTEND IA, and SWIFT PRIME, are directly derived from the historical core/penumbra concepts, with perfusion CT (and some MR) being the dominant selection modality in these ground-breaking trials. These trials have proven that perfusion imaging/core mismatch (or a clinical-core mismatch variant seen in DAWN) is efficient at selecting patient more likely to respond to reperfusion therapy.^111–116 This is particularly so in the late time window trials (DAWN, DEFUSE 3, and EXTEND) where the original concept of the ischemic penumbra to select patients for therapy at late timepoints has been definitively proven. In the earlier time window studies (EXTEND IA, SWIFT PRIME), there is still a view that the use of perfusion imaging to select a more treatment responsive subgroup of patients (which clearly occurred in these trials) may lead to a proportion of patients who still may benefit being excluded.^114,115

Compared to most of the early window thrombectomy RCTs, the DEFUSE 3 study showed a larger absolute benefit of thrombectomy beyond 6 h (up to 16 h), which might be due to their better patient selection (e.g. the presence of penumbra).¹¹² However, it should also be noted that in this trial, the control group had quite low rates of good outcome. In a similar vein, the DAWN trial enrolled patients with large vessel occlusion who had a mismatch between infarct size and clinical deficit (an alternative way to look at the penumbra) and showed a very large absolute benefit of thrombectomy treatment in the 6–24 h window.¹¹¹

Thrombolysis trials have also used perfusion imaging to accurately identify the penumbra and extend the treatment time window. The EXTEND trial compared the clinical outcomes of patients who received alteplase or placebo between 4.5 and 9 h of symptom onset or with wake-up stroke (< 9 h from the mid-point in time from going to bed and awakening with symptoms). EXTEND predominantly used CTP mismatch selection (< 20% selected with MRI) and showed a significant benefit in mismatch patients treated with alteplase.¹¹³ A subsequent pooled analysis of the alteplase trials using the perfusion imaging core/penumbra selection approach (EXTEND, ECASS IV, and EPITHET) showed that the treatment benefit of alteplase compared to placebo was seen exclusively in the patients who fulfilled target mismatch criteria by automated volumetric analysis.^109,113,117 Notably, EPITHET and ECASS IV did not use an automated volumetric approach of core and penumbra. Other trials have used the core/penumbra selection approach with CTP to test another thrombolytic, namely tenecteplase. Indeed, the first study to use the “dual target” selection approach (large vessel occlusion + core/penumbra mismatch) was in a phase 2B trial in 2012. This study showed greater reperfusion and improved clinical outcomes compared to alteplase.¹¹⁸ A subsequent study using a similar dual target selection approach (EXTEND IA TNK part 1) showed that tenecteplase lead to superior early recanalization rates and better clinical outcomes than did alteplase.¹¹⁶ Notably, other studies of tenecteplase but not using the dual target approach (vessel occlusion + penumbra/core mismatch) have been less successful in showing superiority of tenecteplase to alteplase.^119,120

The bottom-line from these revolutionary penumbral imaging selection studies, is while patients should be treated as quickly as possible, patients with a favorable imaging profile (penumbra/core mismatch) have good collaterals and slow infarct growth. Such patients can achieve excellent outcomes from reperfusion therapy up to 24 h after stroke onset.

Conclusion

The ischemic penumbra has been in the center of ischemic stroke research for the last 40 years. Acute imaging of the penumbra is a critical step toward selection of patients that can best benefit from penumbral-salvaging reperfusion therapies. While the imaging techniques have evolved through the years, there is still no imaging gold standard for core or penumbra. This is because they are in fact imaging surrogates for the cellular processes identified in the original experimental studies. Currently the choice between CT and MRI is made based on availability, feasibility, and time pressure. Routine penumbral imaging is widely used around the world, although there is still no wide consensus about its necessity in the early time window. Nonetheless, core/penumbral imaging has been crucial to extending the time window for reperfusion therapy. It has become clear, with translation of the experimental concept of the penumbra into its routine imaging in clinical practice, that “tissue is at least as important as time.”