Arachnoiditis is one of most important factors in late complications of spinal cord injury patients. It leads to intra-medullar lesions such as syringomyelia cysts and microcysts, or extra-medullar lesions such as tethered cords and arachnoid cysts. All these pathologies were well described and differ in their clinical manifestations and in the forms in which they could be diagnosed. Abnormalities common to all the pathologies include arachnoiditis and altered cerebrospinal fluid (CSF) dynamics, which create the conditions for various pathologies, to appear in different ways in patients [1–11]. Sometimes these clinical alterations appear very slowly and insidiously, and could not be perceived by the patient. However, only 1/4 of all patients show clinical evidence of neurological deterioration, which can be revealed by magnetic resonance imaging (MRI) or conventional myelographic computed tomography.

For each of the situations referred above, various surgical procedures are available as described in assorted publications together with results [12, 13]. However, the underlying conditions can lead to relapse, as happens frequently in a reoperation for syringomyelia, in which it is common to find the shunt dislocated or obstructed by the intense adhesive reaction of the arachnoid. Notwithstanding cases of non-traumatic arachnoid cysts, neurosurgeons never limit surgery to the CSF input to the subarachnoid space below the cyst, but remove the arachnoid completely along with its adhesions that produced the cyst. Nevertheless, arachnoids cysts with traumatic origins must be treated in the same way [14, 5].

Many materials are currently available for fixing and stabilizing the spinal traumatic canal in patients who have instability after acute spinal cord injury (SCI). When such a patient who has been operated for stabilization needs a second operation, it has been verified that complete re-alignment of the canal is rarely obtained; it is incomplete in 72% of this group of patients [16, 17], and [18]. Therefore, an unknown percentage of patients will show another associated traumatic myelopathy that will lead to a ring-shaped dural fibrosis at the level of compression. These consequences cause compression of all sub-dural structures and provoke changes in the dynamics of CSF flow.

After previous surgical techniques had been evaluated, a new approach was proposed. The first two patients operated within this series presented with syringomyelia. Instead of conventional surgery, the medulla was completely freed of the fibrotic webs, bridges, and attachments that surrounded it. The dentate ligaments were cut to access the ventral face of the cord. With the medulla free and CSF circulation probably close to normal, it was decided to avoid myelotomy for shunting since there was no arachnoidal tissue to fix the shunt. The symptoms that motivated the surgery disappeared, and in control examinations, it was possible to evaluate the regression of the syrinx cysts.

Surgical operations took place at the Orthopaedic Hospital of Sant'Ana, Parede, after the complete approval of the Clinical Director and the President of the Scientific Committee of the Hospital, who is also the Orthopaedic Director. Before the procedure, full consent from the patients and assent from their relatives were obtained. Phase one of this work was initiated in October 2000 and completed in March 2003. After complete clinical data had been collected for 14 patients, it was decided to recruit another nine patients to the study; this was done between the middle of 2004 and March 2006.

The first two patients operated on had post-traumatic syringomyelia and the others were as follows: three patients with multiple microcystic lesions, one with a syrinx cyst, 3 with arachnoid cysts and 14 with tethered cords. Their initial symptoms were an increase in spasticity, sensory and motor deterioration and an increase of local and radicular pain.

This consisted of a laminectomy at four levels with the scar lesion in the centre of the surgical field. The dura mater was opened to make all the scar tissue visible, and the extremities were easily identified by the presence of the arachnoid cysts. When necessary, the laminectomy was expanded one level further above or below (six patients), until normal arachnoid tissue and spinal cord appeared. When the dura was opened, the first step was the opening of the arachnoid cysts situated rostral and caudal to the lesion in order to find the layer between the thickened arachnoid and the spinal cord. This layer was found on both sides, above and below. With arachnoiditis tissue fully attached to the dura, some traction could be applied to this thickened tissue, so that dissection could be started on all the dorsal face of the cord between the extremities of lesion above and below to the scar direction. At this point, of the major compression, the arachnoiditis tissue was removed fragment by fragment in order to avoid vascular or especially spinal cord lesions, although some fragments were left behind. The same procedure was performed on both lateral faces of the cord, moving it away from the roots. In one patient, both T-5 roots were cut to achieve a good dissection. The dentate ligaments were cut at all exposed levels. After removal of the fibrotic tissue from the cord posteriorly and laterally, the remainder, which was fully attached to dura mater, was removed. At this stage, it was possible to see that the spinal cord remained in very high tension owing to the remaining ventral adhesions. An anterior dissection was made as far from the mid-line as possible, then the spinal cord was pushed away to the mid-line from both sides with small and light movements in order to release it completely. Long tractions to the roots were avoided. When the spinal cord was finally released on its ventral face, the visible tension disappeared immediately and it was possible to move the cord from both sides. In some cases, to ensure that the spinal cord was completely released, a cotton Pattie was introduced at one side below the cord and removed from the other side. Any bone or disk fragments, or even small angulations of the canal, were removed or corrected, through a small incision parallel to the cord (five patients). This made it possible to re-establish the posterior wall of the fractured vertebra. Once the dura and spinal cord were completely free from adhesions, the dura mater was closed without augmentation duroplasty.

In an 8-year old patient, a thoracic laminotomy was performed to avoid further instability. In all the patients, the material used to stabilise the spinal canal was fully maintained.

Three patients abandoned the "follow-up" (patients 2,7,10 in Table 1). The remaining 20 had "follow-ups" ranging from 4 months to 66 months. There were no postoperative complications or temporary neurological deficits in any patient after surgery, and all patients with incomplete lesions maintained neurological functions below the injury site.

The results were evaluated from two points of view. The first was evaluation with MRI and computed tomography in the first and third years after surgery. The control MRI was compared with the previous examination. It was found that all subdural and extra-medullar lesions disappeared and CSF circulation was close to normal. The intra-medullar lesion, which presented as an unenhanced sagittal T1 weighted image of decreased signal intensity, was reduced in respect of sagittal size, but was present at the site of major compression. In T2 weighted images, a reduction of the medullar areas of hyperintensity was seen around the area of major compression. In the 3 cases of syrinx there was also complete regression of the cysts; in only 2 patients did some evidence of the location of the cysts remain. After three years, one of the patients showed spinal cord expansion or better-defined spinal cord walls in MRI images when compared with previous examinations.

The second criterion of evaluation was totally clinical, not only was neurological deterioration eradicated, but some neurological functions appeared below the level of the injury. Although neurological recovery had never been the reason for these operations, all patients subjected to this procedure showed some recovery of motor and sensory functions. In Table 1, the diagnoses show different pathologies with the same origins, arachnoiditis and altered CSF circulation, and all of them were treated by the same spinal cord untethering procedure. The 14 patients with tethered cord syndrome have an average follow up of 48.6 months and to date have suffered no relapses, which indicates a good result.

Altogether, the complete group of 20 patients with follow-up from four to 66 months continues to present good neurological recovery. According to the ASIA motor and sensory score scale the recovery averages were: Motor – 22 points, Touch – 18 points and Pinprick – 17 points (Table 2). In the same table, the percentage [see Additional file 2] presented was: Motor 20.6%, Touch 15.6% and Pinprick 14.4%. The Wilcoxon signed ranks test [Additional file 1] confirms P < 0.001 for all modalities.

There are surgical solutions for the various pathologies, that may appear (with some delay) after a SCI, and they are published with results in many series for each pathology. However, some percentage of relapses is normal and these are intimately linked to obstructions and dislocations of the CSF drainage system. Some authors have performed laminectomies in the area of the injury to effect neurological improvement. However, the results have not been statistically significant so this technique is rarely used. Nevertheless, some patients showed a measure of recovery, probably due to decompression of the intra-dural structures and some alteration in CSF dynamics [21, 22]. It becomes evident, experimentally and in a series of operated patients, that removal of the arachnoiditis may lead to some neurological recovery at the level of the injury and below it; this was confirmed in all the patients in this study [23, 25]. Falci [24], discussing surgical treatment of posttraumatic cystic and tethered spinal cords, described a group of patients operated for untethering, with one year follow up. The results achieved were: light touch – 2.38 points, pinprick – 3.88 points and motor score – 1.47 points. They were good results at the time, but completely different from the new alternative surgical procedure described here. Imaging techniques such as MRI and CT, which are available for the study of SCI patients, are insufficient to reveal the complete area of the medullar injury and especially the relationship established between the arachnoid, the pia and the dura after the accident and its subsequent consolidation. It is not possible to see either the fibrotic adhesions or the cysts that have formed, and it is often necessary to carry out a conventional myelographic tomography to achieve a more accurate diagnosis [26]. However, complete medullar release at the level of the injury, rostral, and caudal to it permits the extensive arachnoiditis existing in many patients to be removed and precludes the appearance of more attachments. The operation described has some advantages, because the responsible factors for obstructing the shunts and stents are eliminated. It is important to mention that completeness or incompleteness of injuries, the time between the acute traumatic event and this procedure, or the level and type of injury, were not used as selection criteria for this operation. In short, it is also important to note that this procedure has proved safe and without immediate or delayed adverse effects in 100% of the patients.

This surgical alternative, with complete removal of the arachnoiditis at the level of the injury and of the altered arachnoid on at least two rostral and caudal levels, allows the CSF to circulate as normally as possible, avoiding the relapses that are the reasons for many further surgical operations.

Myelotomies and the use of materials such as shunts and stents are avoided. In the three patients with syringomyelia, the symptoms disappeared post surgery and the control MRI scans showed regression of the cysts after one year and three years. In this particular pathology, with a follow up average of 56 months, there were no repeated relapses.

In all patients who in the T2 MRI scans showed signs of intra medullar hyper intensity, these images disappeared during the first year, and medullar expansion was visible at the end of the third year.

Any medullar atrophy makes this surgical procedure easier, and in no case were there temporary neurological deficits or other post-operative or delayed neurological deterioration.

Finally, a particularly significant conclusion is that all the patients with different levels and types of injuries showed motor and sensory recovery, with greatly improved qualities of life.