If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Corresponding author. Dittmar Böckler, MD, Department of Vascular and Endovascular Surgery, University of Heidelberg, Im Neuenheimer Feld 110, 69120 Heidelberg, Germany.
Spinal cord ischemia remains a devastating complication after thoracic aortic surgery. The aim of this study was to investigate the pathophysiology of spinal cord ischemia after thoracic aortic endografting and the role of intercostal artery blood supply for the spinal cord in a standardized animal model.
Methods
Female merino sheep were randomized to either I, open thoracotomy with cross-clamping of the descending aorta for 50 min (n=7), II, endograft implantation (TAG, WL Gore & Ass.), (n=6) or III open thoracotomy with clipping of all intercostal arteries (n=5) . CT-angiography was used to assess completion of surgical protocol and assess the fate of intercostal arteries. Tarloy score was used for daily neurological examination for up to 7 days post-operatively. Histological cross sections of the lumbar, thoracic and cervical spinal cords were scored for ischemic damage after stained with Hematoxylin-Eosin, Klüver-Barrrera and antibodies. Exact Kruskall-Wallis-Test was used for statistical assessment (p<0.05).
Results
Incidence of paraplegia was 100% in group I and 0% in group II (p=0.0004). When compared to the endovascular group, there was a higher rate of histological changes associated with spinal cord ischemia in the animals of the control group (p=0.0096). Group III animals showed no permanent neurological deficit and only 20% infarction rate (p=0.0318 compared to group I).
Conclusions
In sheep, incidence of histological and clinical ischemic injury of the spinal cord following endografting was very low. Complete thoracic aortic stent-grafting was feasible without permanent neurologic deficit. Following endovascular coverage or clipping of their origins, there is retrograde filling of the intercostal arteries which remain patent.
Following descending thoracic and thoracoabdominal aortic surgery, spinal cord ischemia with neurologic deficit (e.g. paraplegia) remains a devastating complication. Neurological injury after conventional open repair has been reported to occur in up to 40%.
Paraplegia rate after descending thoracic aortic exclusion in humans is clearly less, 1–2% in many series.
The pathophysiology of paraplegia is multifactorial and incompletely understood. Risk factors for postoperative paraplegia are: (1) the etiology and the extent of aortic injury, (2) duration of aortic cross-clamping, (3) the presence of other co-morbidities, (4) the urgency of the surgical intervention (e.g. elective, urgent or emergency), (5) intra-operative blood-loss and (6) peri-operative hypothermia. Paraplegia following endovascular repair of thoracic aortic pathologies (TEVAR) is observed in 0–12%.
Endovascular treatment of thoracic aortic pathologies holds the allure of reducing peri-procedural complications especially paraplegia rate. Therefore, the numbers of high risk patients who can be offered treatment for their disease is expanding.
The aim of this standardized experimental animal study was to analyze the incidence of paraplegia and spinal cord ischemia after endografting of the descending thoracic aorta. We investigated the correlation of clinical-neurological events (paralysis/paraplegia) with histopathological findings (cell death/infarction). In addition, we investigated the impact of clipping of the intercostal arteries on spinal cord perfusion.
Material and Methods
The study protocol was approved by the National Animal Research Committee. Eighteen female merino sheep, each weighing between 50–60 kg, were randomized into one of three groups.
Group I (n=7): open thoracotomy, cross-clamping of the entire descending thoracic aorta for 50 minutes
Group II (n=6): endograft implantation (TAG, W.L. Gore&Ass., Flagstaff, AZ)
Group III (n=5): open thoracotomy with clipping of all intercostals arteries.
Perioperative monitoring
All surgical procedures were conducted under general anesthesia, induction with Propofol 2 mg/kg and maintainance with 1.5% endtidal isoflurane in an air/oxygen mixture. Central venous pressure (CVP) readings were obtained from a catheter placed into the right jugular vein. Proximal mean arterial blood pressure was measured with an indwelling catheter placed into the carotid artery. Cerebrospinal fluid pressure (CSFP) was measured continuously through a percutaneously placed spinal catheter placed at level T10 of the spinal cord in all animals. Drainage was performed in order to maintain the CSFP below 20 mmHg, a value retrospectively defined to be normal in sheep. Neurological monitoring was undertaken with corticospinal motor evoked potential (D-wave, Ewacs Iom System 916, Inomed GmbH, Teningen, Germany). All procedures were performed under normothermic conditions (37 °C) measured by continuous rectal temperature.
Surgical experiments
In group I, the animals were placed in the right lateral decubitus position. A left antero-lateral thoracotomy was performed between the 3rd–7th intercostal space. The thoracic aorta was mobilized at the level of the innominate artery and the diaphragm. The aortic crossclamp was applied distal to the origin of the innominate artery (in sheep the left common carotid and subclavian arteries originate from the innominate artery) and proximal to the diaphragm for a total of 50 minutes. In group II, the animals were placed into the supine position. Endograft deployment was achieved via the iliac artery using fluoroscopic control: Three overlapping stent-grafts were placed into each animal to cover the entire thoracic aorta from the innominate artery to the celiac trunk (TAG, 26–200 mm, W.L. Gore & Ass. Inc., Flagstaff, AZ). Intraoperative angiograms confirmed correct placements of the endografts.
In group III, positioning, surgical approach and chest closure was identical to group I. The paired intercostal arteries of the thoracic aorta were identified and clipped sequentially (Ligaclip ERCA, Ethicon Endo-Surgery Inc., OH USA) without aortic cross-clamping.
Radiological assessment
Pre- and post-operatively, all animals were examined radiographically under general anesthesia using a multislice computed tomography angiography (MS-CTA) using ECG-trigger controlled for heart rate variability. A standard scanner (Aquilion 16, Toshiba, Tokyo, Japan) performed the CT protocol with parameters published in a web supplement. For pre-operative planning of the required endograft size, the aortic diameter was determined in the anchoring zones using a center line algorithm. Multiplanar reconstructions (MPR) on a workstation (Vitrea 2, Vital Images Inc., Plymouth, MN, USA, software 3.5) determined the opening diameter of all intercostal artery pairs. In the post-operative CT analysis (CTA), length of the instrumented aortic segment as well as the correct placement of the stent grafts were documented for group II animals. Pre and post-operatively, contrast enhancement of the intercostal arteries and the rami dorsales were evaluated in all groups.
Postoperative phase: Neurological function and extraction of spinal cord segments
All animals were extubated postoperatively. Buprenorphine (0,005 mg/kg, i.m.) was used for post-operative analgesia as needed. Post-operative neurological examination was performed daily by an independent veterinarian in all animals. To assess neurological function, the Tarlov score
was used as follows: 0=paraplegia, 1=light limb movements, 2=good limb movements, but not able to stand, 3=able to stand, but not able to walk, 4=normal function. In order to assess immediate postoperative (IND) and delayed onset neurological deficits (DND), animals were sacrificed under general anesthesia (Ketamine 10 mg/kg and Xylazine) 0.2 mg/kg) on the seventh postoperative day by the intravenous injection of T61® (1.0 ml/kg, Intervet GmbH, Unterschleißheim, Germany). Spinal cord was harvested and fixed in 4% paraformaldehyde before staining.
Histological and immunohistochemical staining
Transverse slices of the cord, spaced 5 cm apart, were obtained from lumbar, thoracic and cervical segments and embedded in paraffin. Slices (4 μm) were stained with hematoxylin-eosin (HE) or Klüver-Barrera (KB) using standard laboratory procedures. Immunohistochemical staining was performed on deparaffinized sections. For antigen retrieval, sections were placed in heated citrate buffer and microwaved at 600 W for 20 min. After cooling and washing in phosphate buffered saline (PBS) they were blocked with 5% porcine serum dissolved in PBS. Sections were incubated with a human monoclonal anti-MAP2 antibody (MAP=microtubule associated protein 2), Clone HM-2 (Sigma, Seelze, Germany, 1:200 dilution) at room temperature for 2 hours, a biotinylated secondary linker antibody (Super Sensitive Multi Link, BioGenex) and an alkaline phosphatase conjugated streptavidin tertiary antibody (BioGenex, San Ramon CA, USA), according to the manufacturers instructions, each time followed by washing with PBS. A colour substrate (Fast Red Substrate System, DAKO, Glostrup, Denmark) was applied until a colour reaction was detectable by light microscopy.
For every animal, sections from high lumbar, lower lumbar and thoracic cord were assessed using an Olympus BH-2 light microscope and camera (Olympus Europe GmbH, Hamburg, Germany). Cervical, thoracic and lumbar cords were evaluated at 4 different levels. Cervical cord sections served as internal controls. Ischemic damage was scored on HE and KB stained slices using a modified score described by Meylaerts et al.:
0=no changes, 1=no infarction, but hypoxic changes of neurons (homogenized and/or swollen or shrunken neurons, eosinophilia in HE, paleness in KB exceeding changes caused by perimortal ischemia as observed in cervical slices), 2=small infarction (1/3 of the gray matter), 3=moderate infarction (1/3 to ½ of the gray matter), 4=large infarction (>1/2 of the gray matter), 5=complete infarction of the gray area.
Statistics
Exact Kruskall-Wallis-Test was used to identify differences of ordinal data between animal group I, II and III. Statxact V.3.1 (Cytel Boston, MA, USA) software was used for biometric calculations. A p-value<0.05 was defined to be statistically significant.
Results
All three groups were comparable with respect to body temperature, hemodynamic stability, fluid requirements and urine output.
Clinical-neurological outcomes
All 18 animals were assessed neurologically daily for a maximum of 7 days after surgery. Intraoperative spinal fluid drainage was performed in 2 out of 18 animals of the group I (No. 3: 10 ml, No. 4: 12 ml) in order to maintain normal CSF pressure, below 20 mm Hg (in sheep).
In group I, the paraplegia rate was 100% (n=7, score 0). Due to irreversible and complete paraplegia of all sheep in group I, mercy killing was performed on day 1 in these animals. The postoperative Tarlov scores of the animals are summarized in Table 1. One group I animal died naturally due to pulmonary insufficiency on the day of surgery, but outside working hours so that the spinal cord could not be retrieved promptly. Complete paraplegia (score 0) had been documented prior to its demise.
Table 1Daily assessment of neurological outcome by documentation of Tarlov scores in all groups
Group
Animal Nr.
day 1
day 2
day 3
day 4
day 5
day 6
day 7
I
2
0
3
0
4
0
10
0
11
0
15
0
20
0
II
5
4
4
4
4
died
6
2
4
4
4
4
4
4
7
4
4
4
4
4
4
4
8
4
4
4
4
4
4
4
9
4
4
4
4
4
4
4
19
2
3
3
3
4
4
4
III
12
4
4
4
4
4
4
4
13
4
4
4
4
4
4
4
14
3
4
4
4
4
4
4
16
4
4
4
4
4
4
4
17
4
4
4
4
4
4
4
0=paraplegia, 1=light limb movements, 2=good movements, but not able to stand, 3=able to stand, but not able to walk, 4=normal function.
In group II, each animal received 2–4 (mean=3) endoprostheses to cover the entire descending thoracic aorta. Paraplegia rate after 7 days was 0% (n=6, score 4). However, two animals (No. 6, 19) suffered from temporary early neurologic deficit (33%). They were not able to stand (Tarlov score 2) on day 1 but recovered on day 2 and 5 post-operatively respectively. Neither animal had elevated CSFP, so that no drainage occurred. One sheep (No. 5) died on postoperative day 4 due to retrograde dissection and pericardial tamponade, with a Tarlov score of 4.
In group III, animal No. 14 was unable to walk on the first postoperative day, but recovered quickly to normal motor function. All other animals had Tarlov scores of 4 in the 7 day observation period.
There was significant difference in the neurological outcome between animals of group I and II (p=0.0004) as well as between animals of group I and III (p=0.0026). No statistical difference was observed between animals of group II and III (p=0.6914) (Table 3).
Histological assessment
Clinical and histological results are summarized in Table 3 and Table 4. Histological examination of the spinal cord was performed in 15 eligible specimens (Group I:4, II:6, III:5). All scores are documented in Table 2. The highest ischemic score was used for the calculation of significance. All lumbar and low thoracic specimens examined in group I showed complete infarction (score 5), in accordance with postoperative paraplegia. In contrast, in the endovascular group (II), 3 animals out of 6 had no spinal cord ischemia, while 2 animals showed hypoxic cell changes in a proximal lumbar slice (score 1), one of which (No. 19) also showed motor dysfunction. One animal (No. 7) had a small circumscribed infarction (<1/3 of gray matter, score 2) in the most proximal lumbar section with no clinical deficit. Histopathologically significant differences between group I and II were seen in all lumbar and thoracic spinal cord sections examined. High lumbar and low thoracic segments in group I seemed to have a higher incidence of ischemic change. Amongst the 5 animals in group III, two had histological evidence of ischemia, consistent with the clinically noted neurological dysfunction in animal No. 14. There were no significant differences in the histological specimens of group II and III animals (p= 0.84). In all groups, cervical cord slices used for individual cross control were without ischemic changes (score 0).
Table 2Histopathological analysis of cervical, thoracic and lumbar spinal cord segments scored for ischemic changes
Group
Animal Nr.
Spinal Cord Level
Lumbar
Thoracic
Cervical
I
2
5
5
0
4
5
5
0
10
5
5
0
15
5
4
0
II
5
0
0
0
6
0
0
0
7
2
0
0
8
0
0
0
9
1
0
0
19
1
0
0
III
12
0
0
0
13
1
0
0
14
2
0
0
16
0
0
0
17
0
0
0
0=no changes, 1=no infarction, but hypoxic changes of neurons (homogenized and/or swollen or shrunken neurons, eosinophilia in HE, paleness in KB exceeding changes caused by mortality ischemia as observed in cervical slices) 2=small infarction (1/3 of the gray matter), 3=moderate infarction (1/3 to ½ of the gray matter), 4=large infarction (>1/2 of the gray matter), 5=complete infarction of the gray area.
Postoperative contrast enhanced CT-angiography was completed in 18 animals to confirm stent grafting of the total descending aorta and complete clipping of all intercostal arteries (Fig. 1). In addition, retrograde blood flow to the intercostal arteries was preserved in group II and III animals despite endoluminal occlusion and clipping at their aortic origin, respectively. There was no quantitative difference in retrograde flow in these both groups. In group I, animals postoperative CT-angiography revealed open intercostal arteries. The spinal cord feeding branches of the intercostal arteries were patent in all animals as shown in the CT-angiograms (Fig. 2).
Fig. 1Left: 3D-Volume rendering computed tomography (CT) reconstruction demonstrating 3 endografts covering the descending thoracic aorta in group II. Right: sagital view of CT demonstrating successful clipping of intercostals arteries. Note retrograde perfusion in the cranial three vessels.
This sheep model is the first experimental model analyzing spinal cord ischemic injury after endovascular descending thoracic aortic surgery under standardized conditions. We demonstrated that endovascular treatment, compared to aortic clamping, was associated with a significantly lower incidence of paraplegia (p=0.0004) as well as spinal cord ischemia (p=0.0096). In addition, we showed that, endovascular repair of the entire descending thoracic aorta in sheep is feasible, with moderate incidence of transient paraparesis. The fate of the intercostal arteries after stent grafting or clipping was patency with retrograde perfusion.
In addition to common used staining techniques [hematoxylin-eosin (HE), Klüver-Barrera (KB)] immunohistochemical staining (MAP) of spinal cord ischemia was performed and evaluated for the first time in this study (Fig. 3, Fig. 4). Several authors
Acute focal ischemia-induced alterations in MAP2 immunostaining: description of temporal changes and utilization as a marker for volumetric assessment of acute brain injury.
demonstrated that cytoskeletal damage occurs early in transient or global ischemia and that loss of MAP2 immunostaining was a sensitive and reliable marker for the extent of neuronal damage.
Fig. 3Microscopic findings in lumbar spinal cord sections of group I animals compared to control animals. A, C: section of group II/III animal shows regular morphology with intact gray matter and clearly discernible neurons (A: HE, C: KB, magnification 20x). B, D: Section of group I animal with severe fresh hypoxic damage to gray matter. The low magnigfication shows no discernible neurons and distinct oedema (B: HE, D: KB, magnification 20x). E, F: Comparison of immunohistochemical staining of neuronal cytoskeleton associated protein MAP2 as early and sensitive indicator of hypoxic neuronal changes. E: In control animal, neuronal cytoskeleton is clearly visible in gray matter (magnification 20x). F: Complete loss of cytoskeletal MAP2 staining in group I animal (20x).
Fig. 4Higher magnification (40x) of the slices shown in Fig. 3 C (control animal) and D (group I animal). A: Note granular appearance of neurons (Nissl substance) and prominent nucleoli in slices of control animal. B: In slices of group I animal, neurons are pale, swollen and homogenized, indicating fresh hypoxic injury.
The pathophysiology of spinal cord ischemia and consecutive paraplegia in aortic surgery is complex and multifactorial: reduced spinal cord perfusion pressure (cross-clamping), reperfusion injury and atheromatous microembolism following manipulation of spinal arteries arising from diseased aortic segments are discussed.
In open repair, several strategies to prevent spinal cord ischemic injury have been developed, the most common being distal aortic perfusion, reimplantation of intercostal arteries,
Epidural cooling for the prevention of ischemid injury to the spinal cord during aortic occlusion in a rabbit model: determination of the optimal temperature.
Cerebrospinal fluid drainage and distal aortic perfusion: reducing neurological complications in repair of thoracoabdominal aortic aneurysms types II and III.
Despite these efforts to prevent spinal cord ischemia after open repair, neurologic deficit is still of main concern in thoracic and thoracoabdominal aneurysms from 1–2% in TAA, up to 10% in type I to up to 40% in type II TAAA repairs.
Neurologic deficit in patients at high risk with thoracoabdominal aortic aneurysms: the role of cerebral spinal fluid drainage and distal aortic perfusion.
Despite the lack of long-term results, endovascular aortic repair has become the first choice of therapy for thoracic aortic pathologies in many centres. TEVAR, while less invasive than open repair, is still associated with signs of spinal cord ischemia in 0–12% of all cases.
should be regarded with caution due to different diseased aortic segments, different underlying aortic pathologies (aneurysm, trauma, dissections) and various comorbidities. According to Svensson
postoperative paraplegia is determined by three factors: the duration and degree of ischemia, the presence or absence of reperfusion and the presence or absence of mediator-associated reperfusion injury. In addition to these altered flow dynamics, inflammatory mediators, caused by temporary tissue ischemia, are released with reperfusion.
No aortic occlusion, no proximal hypertension and no distal hypoperfusion with secondary autoregulative disturbances apply to TEVAR.
2.
Distal aortic perfusion is not interrupted in TEVAR. Thus the perfusion of lumbar arteries feeding the spinal cord is not compromised.
3.
Furthermore, there is no reperfusion injury following TEVAR, since intercostals arteries cannot be re-implantated.
Permanent restoration of spinal cord blood flow by re-implantation of critical segmental arteries is generally recommended and so called “essential” in open surgery.
Influence of preservation or perfusion of intraoperatively identified spinal cord blood supply on spinal motor evoked potentials and paraplegia after aortic surgery.
However, to date no prospective randomized study has proved significant risk reduction of paraplegia by this adjunct surgical procedure. It is conceivable, that potential cytotoxic damage is enhanced by reperfusion or re-implanted intercostal arteries. In our study, complete occlusion of all thoracic intercostal arteries resulted in no permanent neurological deficit in the group II and III animals (Table 1, Table 2). This is in contrast to previous clinical studies of open repair, in which the risk of a postoperative neurological complications increased with the number of occluded segmental arteries.
Therefore, our results challenge this protective measure. In addition, endovascular coverage of intercostal arteries does not necessarily result in complete interruption of blood flow in these arteries. As documented by CT-angiography, the intercostal arteries stay patent and show retrograde collateral blood flow (Fig. 2). Recruitment of collateral blood supply from other regions to the spinal cord becomes more apparent. This is confirmed by a study of Griepp et al.,
Looking for the artery of Adamkiewicz: a quest to minimize paraplegia after operations for aneurysms of the descending thoracic and thoracoabdominal aorta.
in which intercostals were not reattached in any patients and an overall paraplegia rate of 2% was achieved. It is conceivable that instead of re-implantation of intercostal arteries during open repair, clipping before aortic clamping might be a beneficial adjunct in order to reduce aortic clamp-time to a minimum.
Greenberg et al. compared total endograft length in patients with or without neurological compromise and showed a significant correlation of length with the paraplegia rate.
In our experiment the entire descending thoracic aorta was covered by stent grafts in group II animals, without causing permanent neurological deficits. Clinically, this may eventually allow a more aggressive endovascular therapy with multiple stent graft implantation in patients with either extensive thoracic or thoracoabdominal aortic aneurysms. It even maybe possible to apply this technique to cases of type B dissection by closing multiple entries through the descending thoracic aorta, to increase false lumen thrombosis rate.
This study has some limitations. In sheep, lumbar arteries play a more important role in collateral circulation to the spinal cord than in humans. Stent grafts were placed in healthy, non-aneurysmal and non-atherosclerotic aortas. Direct comparison to open repair is not possible since there was no thoracic graft sewn in. Clamp and sew technique (Group I) does certainly not correspond to operations with widely used left-left perfusion. Whether paraplegia will result from 50minutes of aortic cross clamping also is determined by other factors, such as the individual variation in collateral blood supply to the spinal cord and body temperature. For logistical reasons, a monitoring period of 7 days was used record delayed neurological deficit (DND). However, neurological injury can be detected as late as 30 days after TEVAR in humans.
Reversal of twice-delayed neurologic deficits with cerebrospinal fluid drainage after thoracoabdominal aneurysm repair: a case report and plea for a national database collection.
Similarly, it was not feasible to continue measuring the cerebrospinal pressure once the operation was concluded. Further future experiments are planned and will investigate paraplegia and spinal cord ischemia with shorter clamping times, with protective adjuncts such as temporary distal aortic perfusion during cross clamping and clipping of segmental arteries before clamping.
Cross-clamping of the thoracic aorta. Influence of aortic shunts, laminectomy, papaverine, calcium channel blocker, allopurinol, and superoxide dismutase on spinal cord blood flow and paraplegia in baboons.
In sheep, both TEVAR and intercostal artery ligation can result in spinal cord ischemia, but the degree of ischemia appears to be significantly less than after 50 minutes of thoracic aortic cross-clamping.
2.
Following endovascular coverage or clipping of their origins, intercostals arteries fill retrogradely and stay patent.
3.
Aortic cross-clamping remains the decisive factor in the pathophysiology of spinal cord ischemia. The role of antegrade intercostal artery perfusion might be overestimated.
4.
Further experiments comparing the endovascular group to aortic cross clamping with adjuncts of distal aortic perfusion and intercostal ligation are necessary.
Acknowledgment
This study was funded by research grants from the German Society for Vascular Surgery, by W.L. Gore & Ass (Flagstaff, AZ, USA) and Medtronic Europe (Tolochenaz, Switzerland). The authors declare that the sponsoring did not influence the procurement of medical products. We would like to thank Martina Jochim
Acute focal ischemia-induced alterations in MAP2 immunostaining: description of temporal changes and utilization as a marker for volumetric assessment of acute brain injury.
Epidural cooling for the prevention of ischemid injury to the spinal cord during aortic occlusion in a rabbit model: determination of the optimal temperature.
Cerebrospinal fluid drainage and distal aortic perfusion: reducing neurological complications in repair of thoracoabdominal aortic aneurysms types II and III.
Neurologic deficit in patients at high risk with thoracoabdominal aortic aneurysms: the role of cerebral spinal fluid drainage and distal aortic perfusion.
Cross-clamping of the thoracic aorta. Influence of aortic shunts, laminectomy, papaverine, calcium channel blocker, allopurinol, and superoxide dismutase on spinal cord blood flow and paraplegia in baboons.
Influence of preservation or perfusion of intraoperatively identified spinal cord blood supply on spinal motor evoked potentials and paraplegia after aortic surgery.
Looking for the artery of Adamkiewicz: a quest to minimize paraplegia after operations for aneurysms of the descending thoracic and thoracoabdominal aorta.
Reversal of twice-delayed neurologic deficits with cerebrospinal fluid drainage after thoracoabdominal aneurysm repair: a case report and plea for a national database collection.
To submit a comment for a journal article, please use the space above and note the following:
We will review submitted comments as soon as possible, striving for within two business days.
This forum is intended for constructive dialogue. Comments that are commercial or promotional in nature, pertain to specific medical cases, are not relevant to the article for which they have been submitted, or are otherwise inappropriate will not be posted.
We require that commenters identify themselves with names and affiliations.
00356-5&id=gr1.jpg){kind=link}
00356-5&id=gr2.jpg){kind=link}
00356-5&id=gr3.jpg){kind=link}
00356-5&id=gr4.jpg){kind=link}