Abstract
BACKGROUND AND PURPOSE: Although diffusion-weighted MR imaging is a powerful tool for evaluating brain ischemia, histopathologic correlates of temporal diffusion changes in cerebral hypoxia/ischemia have not been extensively examined. Diffusion-weighted MR imaging was used to evaluate the relationship between the time course of apparent diffusion coefficient (ADC) changes and the histopathologic findings in cerebral hypoxia/ischemia.
METHODS: Thirty 3-week-old rats were subjected to either a 15-, 30-, or 60-minute hypoxic/ischemic insult (unilateral common carotid artery ligation and exposure to 8% oxygen), during and after which diffusion- and T2-weighted MR imaging was performed. Each animal was killed 48 hours or 6 hours after the insult, and fixed sections of the parietal cortex were examined by light microscopy. Ten other (control) rats were subjected to only unilateral common carotid artery ligation or hypoxia.
RESULTS: The experimental rats showed three patterns of ADC change, depending on the duration of the hypoxic/ischemic insult: transient (15-minute), biphasic (15-, 30-, or 60-minute), and persistent (60-minute) ADC reduction patterns. The transient ADC reduction pattern (reduction during the insult and recovery after resuscitation) was associated with selective neuronal death. The biphasic and persistent ADC reduction patterns (transient recovery and no recovery after resuscitation, respectively) were associated with cerebral infarction.
CONCLUSION: Different temporal patterns of ADC change are associated with different histopathologic findings. Although the clinical manifestations of these different histopathologic presentations are not yet defined, this study indicates that sequential diffusion studies are a potentially powerful tool in the evaluation of hypoxic/ischemic brain injury.
Cerebral hypoxia/ischemia is a major cause of perinatal brain damage (1). Recently, diffusion-weighted MR imaging has been used for the clinical evaluation of hypoxia/ischemia in neonates (2, 3). This technique enables physicians to detect ischemic brain damage in its very early phase, when ischemic tissue may potentially be responsive to clinical management (4–8). Furthermore, by combining a set of diffusion-weighted MR images obtained under conditions of varying gradient strength, quantitative studies of the apparent diffusion coefficient (ADC) can be obtained (5, 9).
Diffusion-weighted MR imaging has been used to evaluate the pathophysiology of cerebral hypoxia/ischemia and the efficacy of cerebroprotective therapies (10–15). To the best of our knowledge, however, the relationship between the time course of ADC changes and the histopathologic presentation is not known. Thus, it is not clear whether diffusion-weighted MR imaging can be predictive of the histopathologic findings in cerebral hypoxia/ischemia. In this study, we subjected young rats to various durations of cerebral hypoxia/ischemia and compared the ADC changes with the histopathologic appearance. Our purpose was to obtain repeated diffusion-weighted images for the quantitative characterization of diffusion for up to 48 hours after a hypoxic/ischemic insult and to correlate changes in signal intensity on these images with histopathologic findings at 48 hours.
Methods
Animal Preparation
The animal experiments were performed in accordance with our institutional guidelines for animal research. In all, 40 3-week-old male Sprague-Dawley rats with an average body weight of 47 ± 5 g (mean ± SD) were used. Anesthesia was induced with 5% isoflurane and then maintained with 1.5% isoflurane in a mixture of 70% N2O, 30% O2 throughout the surgical procedure. Thirty-five of these rats underwent left common carotid artery (CCA) ligation via a ventral transverse cervical incision; the artery was isolated, separated from contiguous structures, ligated in two places, and severed between the ligatures. The other five rats underwent a sham operation that involved just cervical incision and separation of the artery from contiguous structures on the left side. Body temperature was measured using a rectal probe, and the entire surgical procedure lasted less than 5 minutes.
Experimental Protocol
The 40 rats were divided into three groups: group A rats (n = 5) underwent unilateral CCA ligation but not hypoxia, group B rats (n = 5) underwent the sham operation and hypoxia, and group C rats (n = 30) underwent both unilateral CCA ligation and hypoxia. Baseline diffusion-weighted MR imaging was performed, after which the rats in groups B and C were subjected to hypoxia by reducing the inspired oxygen concentration to 8% by N2 dilution. Diffusion-weighted MR imaging was repeated during the period of hypoxia, and the ADC changes in the ipsilateral cerebral cortex were monitored using ADC maps that were generated as soon as each data acquisition was complete.
The rats in group C were randomly assigned to three groups based on the duration of the hypoxic/ischemic insult: rats in groups C-1 (n = 10), C-2 (n = 10), and C-3 (n = 10) were subjected to 15-, 30-, and 60-minute hypoxic/ischemic insult, respectively. All the rats were resuscitated by increasing the oxygen concentration to 30%. Diffusion-weighted MR imaging was repeated 60 minutes after starting resuscitation, and the rats were returned to their cages after they awoke from anesthesia. Diffusion-weighted MR imaging was performed again 48 hours after the hypoxic/ischemic insult, except for rats that showed severe clinical impairment, who were examined by MR imaging 6 hours after the insult.
The rats in group B (n = 5) were resuscitated 30 minutes after the onset of hypoxia, and were examined by diffusion-weighted MR imaging according to the above protocol. The rats in group A underwent diffusion-weighted MR imaging 1, 2, and 48 hours after unilateral CCA ligation.
Each rat was killed by perfusion of fixative immediately after the final MR imaging examination.
MR Imaging
MR imaging was carried out using a 4.7-T superconducting system with a 33-cm horizontal bore magnet and a 65-mT/m maximum gradient capability (Unity INOVA, Varian, Palo Alto, CA). A quadrature coil with an internal diameter of 8 cm was tuned to 200 MHz for radiofrequency excitation and MR signal reception. After the surgical procedure, each rat was placed in a supine position and body temperature was maintained at 37° ± 0.3°C by a circulating water heating pad placed around the body. Each rat was artificially ventilated with 1.5% isoflurane in a mixture of 70% N2O and 30% O2, immobilized with injections of pancuronium bromide (100 mg/kg), and placed in an MR-compatible stereotaxic frame to prevent motion artifacts. The inspired oxygen (FIO2) and expiratory carbon dioxide (PCO2) concentrations were monitored continuously throughout the experiment, and the tidal volume was adjusted to produce PCO2 of 35 to 40 mm Hg.
Diffusion-weighted MR imaging was performed using a multisection, spin-echo sequence with parameters of 1500/80/1 (TR/TE/excitations), a matrix of 128 × 64, a field of view of 30 × 30 mm, and a section thickness of 2 mm without an intersection gap. The diffusion gradients (duration [δ], 30.5 milliseconds; separation time [Δ], 36.7 milliseconds; and gradient strength [G], 0 or 26 mT/m) were applied along the three orthogonal directions (x-, y-, and z-axes). The resulting values for the gradient factor b were 0 or 1200 s/mm2, and the acquisition time for one set of diffusion-weighted MR images was 6.4 minutes.
Data Analysis
All analyses were performed using a SunSparc 10 workstation (Sun Microsystems, Mountain View, CA) and image analyzing software (XDS, Davis Bioengineering, St. Louis, MO). ADC maps were plotted on a pixel-by-pixel basis using the equation (9) where S0 and S1 are the signals of the two diffusion-weighted MR images that represent the average of three orthogonal planes (ie, a trace of diffusion tensor, and b0 and b1 = 0 and 1200 s/mm2, respectively). A section 1.8 mm posterior to the bregma was chosen, and regions of interest (ROIs) for the ADC measurements were drawn on the image of the ipsilateral parietal cortex. All data are expressed as the mean ± SD.
Histologic Examination
Immediately after the final MR imaging acquisition procedure, each animal, which was under deep anesthesia, was removed from the MR unit and its brain was fixed by perfusion with a solution of 4% paraformaldehyde in phosphate-buffered saline via an indwelling left ventricular catheter for up to 30 minutes. The brain was then removed, and sequential 2-mm-thick slices corresponding to the MR images were cut and placed in cooled fixative. A slice corresponding to the ADC map 1.8 mm posterior to the bregma was chosen and was prepared and stained with hematoxylin-eosin for light microscopic examination.
Statistical Analysis
ADC values at certain time points were compared with the baseline ADC values for each group by using Student's paired t-test. Differences with P values of less than .05 were considered significant.
Results
Time Course of ADC Changes
The mean baseline ADC value in the cerebral cortex for all 40 rats was 6.97 ± 0.4 × 10−4 mm2/s. The control rats (groups A and B) showed neither abnormal signal intensities on T2-weighted images nor ADC changes throughout the period of the experiments. Furthermore, the contralateral parietal cortices of the experimental rats (group C) showed neither abnormal signal intensities on T2-weighted images nor ADC changes throughout the period of the experiments.
The 10 rats subjected to 15 minutes of hypoxia/ischemia (group C-1) showed two patterns of ADC reduction in the ipsilateral parietal cortex: transient (n = 8) and biphasic (n = 2). The transient ADC reduction pattern included acute ADC reduction during the hypoxic/ischemic insult (4.97 ± 0.40 × 10−4 mm2/s), recovery to the baseline ADC value 60 minutes after resuscitation (7.01 ± 0.48 × 10−4 mm2/s), and no further alteration 48 hours after the insult (6.99 ± 0.4 × 10−4 mm2/s). The ADC value during the hypoxic/ischemic insult was significantly lower than the baseline ADC value (P < .05), but the ADC values 60 minutes after resuscitation and 48 hours after hypoxic/ischemic insult were not significantly different from the baseline ADC value. In these eight rats, no abnormal signal intensity was observed on T2-weighted images throughout the experiment (Fig 1). The biphasic ADC reduction pattern included acute ADC reduction during the hypoxic/ischemic insult (3.97 ± 0.21 × 10−4 mm2/s), transient ADC recovery 60 minutes after resuscitation (6.98 ± 0.32 × 10−4 mm2/s), and secondary ADC reduction 48 hours after the insult (4.38 ± 0.12 × 10−4 mm2/s). The ADC values during and 48 hours after hypoxic/ischemic insult were significantly lower than the baseline ADC value (both, P < .05); but the ADC value 60 minutes after resuscitation was not significantly different from the baseline ADC value. In these two rats, T2-weighted images showed high signal intensity in the affected regions 48 hours after the insult.
All 10 rats subjected to 30 minutes of hypoxia/ischemia (group C-2) showed the biphasic ADC reduction pattern. The ADC values were 4.03 ± 0.43 × 10−4 mm2/s during the hypoxic/ischemic insult (significantly lower than the baseline ADC value; P < .05), 6.96 ± 0.45 × 10−4 mm2/s 60 minutes after resuscitation (no significant difference from the baseline ADC value), and 4.74 ± 0.63 × 10−4 mm2/s 48 hours after the insult (significantly lower than the baseline ADC value; P < .05). T2-weighted images showed high signal intensity in the affected region 48 hours after the insult (Fig 2).
The 10 rats subjected to 60 minutes of hypoxia/ischemia (group C-3) showed two patterns of ADC reduction: biphasic (n = 4) or persistent (n = 6). The ADC values of the four rats that showed the biphasic ADC reduction pattern were 4.31 ± 0.15 × 10−4 mm2/s during the hypoxic/ischemic insult (significantly lower than the baseline ADC value; P < .05), 7.06 ± 0.39 × 10−4 mm2/s 60 minutes after resuscitation (no significant difference from the baseline ADC value), and 4.25 ± 0.69 × 10−4 mm2/s 48 hours after the insult (significantly lower than the baseline ADC value; P < .05). The T2-weighted images of these four rats showed high signal intensity in the affected region 48 hours after the insult. The other six rats showed the persistent ADC reduction pattern: an ADC decrease during the hypoxic/ischemic insult (3.93 ± 0.28 × 10−4 mm2/s) and a further decrease 60 minutes after resuscitation (2.98 ± 0.39 × 10−4 mm2/s). The ADC values during the insult and 60 minutes after resuscitation were significantly lower than the baseline ADC value (both, P < .05). Each of these six rats showed severe clinical impairment after they awoke from anesthesia and they died within 24 hours of the hypoxic/ischemic insult. High signal intensities on T2-weighted images and sustained ADC decreases were seen 6 hours after the insult in these six rats. At that time, large and midline shifts, caused by swelling of the ipsilateral cerebral hemisphere, were observed in the affected regions (Fig 3).
The mean ADC values at the end of the hypoxic/ischemic insult were 4.97 ± 0.40 × 10−4 mm2/s, 4.08 ± 0.37 × 10−4 mm2/s, and 3.93 ± 0.28 × 10−4 mm2/s in the rats with the transient, biphasic, and persistent ADC reduction patterns, respectively. The mean ADC value at the end of the hypoxic/ischemic insult in the rats with the transient ADC reduction pattern was significantly higher than that in the rats with the biphasic or permanent ADC reduction patterns (P < .01), whereas there was no significant difference in ADC values between the rats with the biphasic or permanent ADC reduction patterns.
Relationship between ADC Changes and Histopathologic Findings
The time course of the ADC changes and the histopathologic findings of each group are summarized in the Table. The control rats showed no ADC changes (n = 10) and their histopathologic examinations showed that neurons were intact and that the cell density was well preserved (Fig 4A), whereas the rats with the transient ADC reduction pattern (n = 8) invariably showed neuronal loss with ischemic changes (cytoplasmic eosinophilia and pyknotic nuclei), which indicated selective neuronal death. In these rats, pan-necrosis was not found, but more than half of the neurons showed selective neuronal death (Fig 4B). Rats with the biphasic ADC reduction pattern (n = 16) showed pan-cellular necrosis, indicative of cerebral infarction on histopathologic examination (Fig 4C); and the rats with the persistent ADC reduction pattern (n = 6) showed a large hemispheric infarction, as evidenced by extensive neuronal pyknosis and severe neuropilar microvacuolation 6 hours after the hypoxic/ischemic insult (Fig 4D).
Thus, our data clearly reveal a close association between the time course of ADC changes and the histopathologic findings in the ipsilateral parietal cortex after hypoxic/ischemic insult.
Discussion
We exposed young rats to unilateral CCA ligation and hypoxia, which is an established model of cerebral hypoxia/ischemia (16). Using this model, we compared the time course of ADC changes with the histopathologic findings in cerebral hypoxia/ischemia. Acute ADC reduction during the early phase of ischemia has been reported to be associated with cell swelling resulting from ischemia-induced energy failure (4, 8). Recent reports on experimental and human brain ischemia have shown that early reperfusion can reverse acute ADC reductions and that diffusion-weighted MR imaging delineates not only irreversible but also potentially reversible ischemic brain changes (17–19). In this regard, we found three time-course patterns of ADC changes, which depended on the duration of the hypoxic/ischemic insult: the transient, biphasic, and persistent ADC reduction patterns.
The severity of insult in this hypoxia/ischemia model has been modified by varying the duration of hypoxia (20), but a large variation in the extent of brain injury is known to occur even after the same duration of insult (21). This variation may be due to interindividual variability in susceptibility to hypoxia. We found different ADC reduction patterns and histopathologic presentations among the rats subjected to the same hypoxic/ischemic insult. Our data, however, clearly show a close relationship between the time course pattern of ADC changes and the histopathologic findings in cerebral hypoxia/ischemia.
Transient ADC reduction was seen in eight of 10 rats subjected to hypoxia/ischemia for 15 minutes. As expected from previous reports (17, 18), no cerebral infarction was found in rats with the transient ADC reduction pattern; however, neurons in the ipsilateral parietal cortices of these rats consistently showed selective neuronal death 48 hours after the hypoxic/ischemic insult. This finding is important because such neuronal injury can cause neurodevelopmental disabilities, depending on the number of injured neurons. Neuronal death after cerebral hypoxia/ischemia has been reported to present as two different modes of cell death: necrosis and apoptosis (22). Necrosis predominates in more intense forms of ischemic damage, whereas apoptosis occurs in milder forms of ischemic damage (23). Delayed neuronal death is also a well-known phenomenon after a brief period of brain ischemia (24). In this study, we did not look for apoptosis or histopathologic changes earlier than 48 hours after the hypoxic/ischemic insult unless the rat showed severe clinical impairment, and we did not determine when or how the neuronal damage occurred in the rats with the transient ADC reduction pattern. Our data, however, show that ADC normalization after a short period of hypoxic/ischemic insult does not always correlate with neuronal salvage. Therefore, we speculate that once ADC reduction is caused by cerebral hypoxia/ischemia, not only resuscitation but also additional cerebroprotective therapy may be effective to reduce selective neuronal death.
The biphasic ADC reduction pattern was seen in two of 10 rats subjected to 15 minutes of hypoxia/ischemia, in all 10 of the rats subjected to 30 minutes of hypoxia/ischemia, and in four of the 10 rats subjected to 60 minutes of hypoxia/ischemia. Secondary ADC reduction is a well-known occurrence (11–13), and this appeared to be analogous to the biphasic energy failure that has been measured by phosphorous nuclear MR spectroscopy in cerebral hypoxia/ischemia (25–27). All the rats with the biphasic ADC reduction pattern showed pan-cellular necrosis, which is indicative of cerebral infarction. The persistent ADC reduction pattern, seen in six of the 10 rats subjected to 60 minutes of hypoxia/ischemia, was associated with massive infarction with a midline shift and early death of the animal. The rats with biphasic or persistent ADC reduction patterns showed high signal intensity on T2-weighted images in the affected regions 48 hours and 6 hours after the insult, respectively. These findings demonstrate that high signal intensity on T2-weighted images is indicative of irreversible brain damage in cerebral hypoxia/ischemia.
Conclusion
We found three different time course patterns of ADC changes, depending on the duration of the hypoxic/ischemic insult: the transient, biphasic, and persistent ADC reduction patterns. The transient pattern correlated with selective neuronal death, whereas the biphasic and persistent patterns were associated with cerebral infarction that finally showed high signal intensity on T2-weighted images. Although the clinical manifestations of these different histopathologic findings are not yet defined, this study indicates that sequential diffusion studies are a potentially powerful tool in the evaluation of hypoxic/ischemic brain injury.
Acknowledgments
We are grateful for the helpful contributions and support of Prof. K. Hirakawa during this project.
Footnotes
↵1 Address reprint requests to Naoyuki Miyasaka, MD, Department of Obstetrics and Gynecology, School of Medicine, Tokyo Medical and Dental University, Yushima 1-5-45, Bunkyo-ku, Tokyo 113-8519, Japan.
References
- Received April 1, 1999.
- Copyright © American Society of Neuroradiology