Tissue at Risk Is Overestimated in Perfusion-Weighted Imaging: MR Imaging in Acute Stroke Patients without Vessel Recanalization

Methods

Postprocessing

All imaging data sets were transferred into a 3D image format, and spatial alignment was performed as described elsewhere (14) to enable region of interest analysis across all modalities (ADC, PWI, T2) and time points (day 0, 1, 7). Regions of interest of ADC reduction were manually delineated and denoted as lesion at day 0 (L0). At day 1, the outer rim of the T2 lesion subtracted by L0 was defined as the lesion growth until day 1 (LG1). In the same way, lesion growth until day 7 (LG7) was depicted as the outer rim of the T2 lesion at day 7 subtracted by LG1 + L0. Finally, surviving tissue (ST7) was tissue appearing normal on T2 images at day 7 but initially included within the CBF perfusion abnormality at day 0 (Fig 1). These regions of interest were then transferred onto PWIs at day 0 to yield initial mean signal intensities and relative differences (rCBF, rCBV, rMTT, rTTP) by comparison with contralateral mirrored regions of interest.

• Download figure
• Open in new tab
• Download powerpoint

Fig 1.

Illustration of lesion definitions on a representative ADC image at day 0. Lesion on day 0 in black = ADC lesion at day 0; dark gray = lesion growth until day 1 (LG1); pale gray = lesion growth until day 7 (LG7); white = surviving tissue (ST7). LG1 and LG7 were found by surrounding follow-up T2 lesions at day 1 and 7 and transfer onto day 0 PWI images. ST7 is tissue initially showing CBF decrease but without T2 changes at day 7.

Results

Of the 19 enrolled patients, seven were female, mean age was 61 years (range, 27–79 years), and National Institute of Health Stroke Scale (NIHSS) was 12 (range, 4–22). At day 0, time from symptom onset to imaging was 2.8 hours (range, 1.7–4.7 hours), and 13 patients were treated by systemic thrombolysis with recombinant tissue plasminogen activator (rt-PA) immediately after MR imaging; however, complete vessel occlusion persisted at least until day 1 in all patients and until day 7 in 6. Recanalization at day 7 was less frequent in CTO than in MCA branch occlusions, despite less-frequent thrombolysis with rt-PA in the latter (Table 1). Recanalization at day 7 had no influence on lesion growth LG1 (P = .58) or LG7 (P = .95).

The overall median ADC lesion volume (L0) in all patients was 18.2 mL and increased to 39.4 and 43.8 mL on T2 images at days 1 and 7. The volume of surviving tissue (ST7) was about three times larger (148.5 mL) than the tissue volume finally progressing to infarction. Perfusion values in L0—as well as LG1, LG7, and ST7—are summarized in Table 2. The most severe reduction of rCBF and rCBV occurred in L0 and these values increased continuously to ST7. This results in an “onion-like” distribution of hypoperfusion with the most severe changes in the center of the ischemic territory (Fig 1). The same pattern of initial perfusion values was found in time parameters with a continuous decrease of relative and absolute prolongation of MTT and TTP from the center to the periphery. Pairwise testing between ST7 and all parameters revealed differences of ST7 versus LG0 or LG1. Only rCBV, however, showed a difference between ST7 and LG7 (P = .02). Thus, rCBV was the only candidate parameter distinguishing finally infarcted tissue from surviving brain at day 7. Estimation of a CBV threshold revealed a decrease to 0.82 compared with contralateral control regions with a sensitivity of 0.56 and specificity of 0.95.

Comparing the three most frequent occlusion types, lesion growth was larger in CTO than in M1/2 occlusions or MCA branch occlusions (Table 1, Fig 2), with the most prominent increase in CTO until day 7 (P < .01). Although the initial lesion volume in MCA branch occlusion was largest, lesion growth was very small in these peripheral occlusions despite failed recanalization at least at day 1.

• Download figure
• Open in new tab
• Download powerpoint

Fig 2.

Relative contribution of the three most frequent occlusion types to the total lesion volumes at day 0 (L0) or lesion growth LG1 and LG7. 100% is the sum of the lesions volumes of the three occlusion types.

Discussion

We determined initial lesion volumes and relative perfusion impairment on MR imaging at initial presentation of acute stroke patients. Because none of these patients showed recanalization at least up to day 1, reperfusion and salvage of severely hypoperfused tissue in the penumbra should not have occurred and tissue at risk should be expected to progress to infarction. This is the only setting in which an analysis of perfusion thresholds in the penumbra zone is valid and will not be confounded by recanalization. The most important question to be answered is whether a perfusion threshold exists between still viable tissue threatened to progress to infarction and surviving tissue. This perfusion threshold would define the real, potentially salvageable tissue and is equivalent to the threshold between LG7 and ST7.

The severity of hypoperfusion decreased from the center of the lesion defined by initial ADC decrease (L0) to the penumbra zone (LG1 and LG7) and finally to surviving tissue in all relative values of CBF, CBV, MTT, and TTP. This is consistent with pathophysiologic assumptions and the penumbra concept (4, 15). A considerable overlap exists in all perfusion values, and only rCBV was found to differentiate between LG7 and ST7. A CBV threshold was estimated with a moderate sensitivity for detecting surviving tissue beyond a CBV of 0.82 compared with contralateral control values, but with a high specificity for infarction below 0.82. This supports the clinical observation of frequent lesion growth in patients without recanalization and definite infarction of regions with visible CBV reduction (6). The slight increase of rCBV in ST7 results from an CBV increase in oligemic regions of the affected arterial territory due to leptomeningeal collateral flow (16) and vasodilatation (4, 17, 18).

Our results are not comparable to previous studies, because of inclusion of patients with failed recanalization only. Similar perfusion values within regions of initial diffusion restriction (L0) (4, 19) or regions progressing to infarction (LG1 and LG7) were, however, reported (5, 20–24). Contrary to our results, some recent studies found rCBF thresholds (20) and MTT (21) more useful to differentiate between tissue progressing to infarction and surviving tissue. In both studies, no spatial normalization of initial and follow-up imaging into a standardized 3D space was performed, also correcting for asymmetrical brain morphology. More importantly, previous studies did not consider whether recanalization had occurred and the time point of imaging was later (20) or the outer rim of hypoperfused tissue was defined by the MTT (20, 21), which is known to overestimate penumbral tissue (25, 26). Because differences also exist in patient inclusion criteria, time points of MR imaging, therapy, and methodology of image data evaluation, a generally accepted consent on perfusion thresholds is not in sight. Regions of visible CBV reduction predict the final infarct with high specificity, but sensitivity for lesion growth may be low, resulting in misinterpretation and underestimation of the final infarct. By contrast, the time parameters TTP and MTT bear the risk of frequent overestimation. Therefore, we suggest outlining perfusion lesions on CBF images until further refinements of the singular value decomposition have been evaluated (25).

Although the assessment of hypoperfused tissue by CBF was more conservative compared with others who used perfusion time parameters MTT or TTP (5, 21, 27), we found a large volume of initially hypoperfused tissue not progressing to infarction. In line with previous findings (16), lesion growth was dependent on the site of the vessel occlusion (27), with the greatest progression in CTO until day 1 and especially day 7. Because M1/2 occlusions experienced moderate progression until day 1, the lesion growth in MCA branch occlusions was very small (28) even though they showed the largest initial ADC lesion. Obviously, the penumbra defined by ADC decrease (as the core lesion) and CBF reduction (the outer border) is very small in branch occlusions, but is this simple penumbra concept valuable? Studies on the normalization of ADC indicate the salvage of even severe ADC decrease (13), especially in more distal occlusions in case of recanalization at early time points (29). Thrombolysis should not be withheld in patients with MCA branch occlusions even if there is no PWI/DWI mismatch, because there might be a chance of ADC recovery. Nevertheless, our results indicate that the penumbra is very large in CTO and may persist for a period of days, so more-aggressive approaches to recanalize CTOs are necessary, because thrombolysis is often ineffective (16, 30, 31). Decision making and further evaluation of penumbral thresholds need to consider the vessel status, because routine brain parenchymal imaging by CT (early signs of ischemia) or MR imaging (PWI/DWI mismatch) alone may not be sufficient.

Some limitations of the present study must be discussed. First, the number of patients included is low, especially after differentiation of occlusion types, although 189 candidates for thrombolysis have been screened by MR imaging within 6 hours of the onset of symptoms. Persisting occlusion without any region of reperfusion on follow-up imaging was rare. This led to the inclusion of patients showing recanalization until day 7, which may have influenced LG7 perfusion values at day 0. On the other hand, the volume of infarct growth between day 1 and 7 (LG7) was relative small and independent from recanalization. Recanalization at day 7 was improper to influence initial perfusion values at day 0 directly. Furthermore, only insignificant differences in recanalization rates were found between the occlusion types (Table 1) and the larger LG7 volume in CTO could not be attributed to less frequent recanalization. Second, one may argue that lesion growth may be dependent on anatomic prerequisites with more proximal occlusions naturally supplying a greater territory; however, relative lesion growth LG1 + LG7/L0 + ST7 was also different: CTO, 42.8%; M1/2, 12.1%; and MCA branch, 14.9%.

Conclusion

Perfusion thresholds of the real penumbra are difficult to determine on initial MR imaging and only reduced CBV was found to be highly specific for final infarction. If recanalization fails, the occlusion type is predictive of lesion growth and is an important variable for therapeutic conclusions. More aggressive treatment regimens such as mechanical clot retrieval or fragmentation may be considered in selected cases of carotid T occlusions beyond the accepted time limits of 3 hours and when the actual ADC lesion volume is still small.