Tao Fan，HaiJun Zhao，XinGang Zhao，Cong Liang，YinQian Wang，QiFei Gai

Abstract Chiari I malformation has been shown to present different cerebrospinal fluid (CSF) flow patterns at the cranial-vertebral junction (CVJ). Posterior fossa decompression is the first-line treatment for symptomat- ic Chiari I malformation. However, there is still contro- versy on the indication and selection of decompression procedures. This research aims to investigate the clinical indications, outcomes, and complications of the decompression procedures as alternative treatments for Chiari I malformation, based on the different CSF flow patterns at the cranial-vertebral junction. In this study, 126 Chiari I malformation patients treated with the two decompression procedures were analyzed. According to the preoperative findings obtained by using cine phase-contrast MRI (cine PC-MRI), the abnormal CSF flow dynamics at the CVJ in Chiari I malformation was classified into three patterns. After a preoperative evalu- ation and an intraoperative ultrasound after craniecto- my, the two procedures were alternatively selected to treat the Chiari I malformation. The indication and selection of the two surgical procedures, as well as their outcomes and complications, are reported in detail in this work. Forty-eight patients underwent subdural decompression (SDD), and 78 received subarachnoid manipulation (SAM). Ninety patients were diagnosed as having Chiari I malformation with a syrinx. Two weeks after the operation, the modified Japanese Orthopedic Association (mJOA) scores increased from the preoper- ative value of 10.67 ± 1.61 to 12.74 ± 2.01 (P < 0.01).The mean duration of follow-up was 24.8 months; the mJOA scores increased from the postoperative value of 12.74 ± 2.01 to 12.79 ± 1.91 at the end of follow-up (P=0.48). More complications occurred in the patients who underwent SAM than in those who received SDD (SAM 11 of 78 (9.5%) vs SDD 2 of 48 (3.5%)). The abnormal CSF flow dynamics at the CVJ in Chiari I malformation can be classified into three patterns. A SAM procedure is more feasible in Chiari I malforma- tion (CM1) patients with pattern III CSF flow dynamics, whereas a SDD procedure is more suitable for CM1 patients with pattern I CSF flow dynamics. In CM1 patients with pattern II CSF flow dynamics, an intraop- erative ultrasound after craniectomy could play an important role in the selection of an effective decom-pression procedure.Keywords  Syrinx，Subdural　decompression，Subarachnoid decompression，preoperative CSF flow

Abbreviations

CM1 Chiari I malformation

CSF Cerebrospinal fluid

Cine PC-MRI Cine phase-contrast MRI

SPS Syringo-pleural shunting

mJOA Modified Japanese Orthopedic Association

CCOS Chicago Chiari Outcome Scale

Introduction

Chiari malformation was first described in 1891 as a caudal displacement of the cerebellar tonsils through the foramen magnum. A cerebellar tonsil location of more than 5 mm below the fora- men magnum has been widely regarded as the imaging standard in the diagnosis of Chiari I malformation. However, when the tonsillar descent is less than 5 mm, a diagnosis of Chiari I malforma- tion is doubtful. Cine phase-contrast magnetic resonance imaging (PC-MRI) is useful in detecting a cerebrospinal fluid (CSF) flow obstruction and

showing a clinically relevant herniation. Surgical techniques have long been used to treat this condi- tion. These procedures have one thing in common: They all use posterior fossa decompression to different extents.

The decompression method is still regarded as the optimum technique considering the related anatomic pathology. Generally, there are two main types of decompression procedure. The first is subdural decompression (SDD), in which a small craniectomy, a standard laminectomy of C1, and a Y-shaped dural opening are carried out while keep- ing the arach-noid membrane intact. The second type is subarachnoid manipulation (SAM), in which a small craniectomy, a standard laminecto- my of C1, subarachnoid cerebellar tonsillar coagu- lation and/or tonsillectomy, and adhesion lysis are done. However, there is little agreement on the selection of the type of procedure to be done. Cine PC-MRI has shown markedly disordered CSF flow dynamic at cranial-vertebral junction (CVJ) regu-larly. The disturbed CSF flow dynamics in Chiari I malformation may present different radiologic features and varying patterns.

In the present research, 126 Chiari I malfor-mation patients treated at a single center from 2008 to 2015 were studied. According to the preopera-tive findings obtained by using cine PC-MRI, the abnormal CSF flow dynamics at the CVJ was classified into three patterns. Based on these CSF flow patterns and the change in the CSF dynamics after a small window craniectomy, obtained by using an intraoperative ultra- sound, the decom-pression procedures were selected. The clinical characteristic CSF flow dynamic patterns of these three patterns were collected and analyzed. The indication and se- lection of the surgical proce- dures, as well as their outcomes and complications, are reported in detail in this work.

Materials and method

Patient information

A total of 126 patients diagnosed with Chiari I malformation were surgically treated at our center from 2008 to 2015. The 126 patients’ information was obtained from Sanbo Brain Hospital. Informed consent was obtained from all patients prior to the study. All studies using the medical records of the patients were approved by the Institutional Review Board of Sanbo Brain Hospital. The medical records of the patients, as well as their symptoms, preoperative and postoperative radiographic find-ings, extent of surgery, outcomes, and complica-tions, were analyzed. Of the 126 patients, 62 were males and 64 were females; the average age was 38 (range, 9–64) years. The most common symptoms were pain, including headache, neck pain, and/or back pain; lower cranial nerve dysfunction, includ-ing oropharyngeal dysfunction, such as sleep apnea, snoring, and aspiration tendency; cerebellar symptoms, including gait instability and ataxia; and motor and sensory deficits. CSF flow patterns based on preoperative cine PC-MRI

An Achieva 1.5T scanner (Philips) was used in imaging. A two-dimensional cine phase-contrast sequence centered at the posterior fossa and CVJ was applied to access the CSF flow. The scan parameters were as follows: TR/TE of 24/9 ms (mi-nor variations in TR according to heart rate), flip angle of 20°, matrix of 256 ×192, field of view of 15×20cm, slice thick- nessof5mm,and velocity encoding of 15cm/s. All CSF flow pathways through the posterior fossa were imaged, including the central aqueduct, the foramen magnum, the

chamber of the fourth ventricle and its outlets, and the posterior fossa space behind the cerebellum and tonsils. The CSF flow patency was assessed and confirmed by a neuroradiologist based on the find- ings of the cine PC-MRI. Both the absence of and a decrease in flow signals were deemed as an” abnor-mal CSF flow.” The analysis focused on three loca-tions: the ventral space between the clivus and brain stem at the level of the foramen magnum (A), the central aqueduct and IV ventricle, including the foramen of Magendie (B), and the posterior fossa space behind the cerebellum and tonsils (C) (see Figs. 1, 2, 3, and 4).

Based on the obstructed locations, the abnor- mal CSF flow dynamics could be classified into three patterns: (1) CSF flow obstruction at the posterior fossa space behind the cerebellum and tonsils (C), (2) CSF flow obstruction both at the posterior fossa space behind the cerebellum and tonsils and at the IV ventricle and central aqueduct (B + C), and (3) CSF flow obstruction at the poste- rior fossa space behind the cerebellum and tonsils, at the IV ventricle and central aqueduct, and at the ventral space between the clivus and brain stem (A+ B+ C).The final diagnosis of abnormal CSF flow was established and confirmed by two or three neuroradiologists’ viewing the same imaging.

Surgical procedures

The standard posterior fossa decompression reported in the literature was adopted in our series.

Fig. 1 a T2-weighted MR scan of a 40-year-old female patient with a 5-year history of numbness and weakness of both upper extremities. The image demonstrates a Chiari I malformation with syrinx. Video 1, preoperative cine PC-MR shows CSF flow obstruction at the posterior fossa space behind the cerebellum and tonsils, at the IV ventricle and central aqueduct, and at the ventral space between the clivus and brainstem (A + B + C). Through the decompression window, ultrasound showed an inadequate CSF space with blockage of the CSF flow. She received SAM procedure. b, c The postoperative MRI 2 weeks and 1 year showed remarkable shrinkage of the syrinx. The mJOA score increased from preoperative value of 9 to 13 at the end of follow-up

The patient was in lateral position, with the head fixed with a Mayfield® skull clamp (Integra Life Sciences Corp., Cincinnati, OH, USA). In SDD, a small craniectomy, a standard laminectomy of C1, and a Y-shaped dural opening were carried out while keeping the arachnoid membrane intact. If a duraplasty was needed, a duraplasty was done by sticking a triangular artificial dura to the edge of the dura opening. SAM consisted of a midline durotomy after craniectomy, as previously men-tioned, with coagulation and/or subpial resection of the cerebellar tonsillar. In this procedure, the dura was closed primarily or sometimes duraplasty with a sutured artificial dura. The extent of resec- tion was ended when the tonsils were reduced after bipolar cauterization in the space above the fora-men magnum.

The obex of the IV ventricle was routinely visualized and explored. Adhesion lysis around the foramen of Magendie was routinely carried out. The decision to use either SDD or SAM was routinely made intraoperatively, before the dural opening was made. The CSF flow pattern obtained preoperatively by using cine PC- MRI was reeval- uated by examining the decompression window through an intraoperative ultrasound. The CSF flows at areas A, B, and C were reassessed, and the location of tonsillar herniation was reappraised based on the intraoperative ultrasound findings. When the intraoperative ultrasound showed an adequate CSF space without obstruction of the CSF flow at areas A, B, and C or tonsillar hernia- tion of less than the C1 level, SDD was carried out; otherwise, a dural opening was made and SAM was done.

Fig. 2 a T2-weighted MR scan of a 9-year-old male patient with an 8- month history of distending pain and numbness of his left limb. The image demonstrates an obvious syrinx. Video 2, preoperative cine PC- MR shows CSF flow obstruction both at the posterior fossa space behind the cerebellum and tonsils and at the IV ventricle and central aqueduct (B+ C). Through the decompression window, ultrasound showed an inadequate CSF space with blockage of the CSF flow. He received SAM procedure. b, c The postoperative MRI 1 week and 5 months later shows the resolution of the syrinx. The symptoms of his left limb gradually disappear. The mJOA score increased from preoperative value of 10 to 13 at the end of follow-up

Fig. 3 a T2-weighted MR scan of a 23-year-old female patient with a 2-year history of headache and neck pain. The image demonstrates a Chiari I malformation with a syrinx. Video 3, preoperative cine PC-MR shows CSF flow obstruction at the posterior fossa space behind the cerebellum and tonsils (C). Through the decompression window,ultrasound showed an adequate CSF space without blockage of the CSF flow. She received SDD procedure. b, c The postoperative MRI 2 weeks and 6 months showed gradual resolution of the syrinx. The headache and neck pain had vanished. The mJOA score increased from preoperative value of 11 to 13 at the end of follow-up

Patient evaluation and follow-up and statistical analysis

During follow-up, the patients underwent neurologic examination and imaging evaluation; the modified Japanese Orthopedic Association (mJOA) scores were used to determine the improvement in symptoms associated with a syrinx. The means ± standard deviation of the mJOA scores at the preoperative, postoperative, and latest follow-up time points were collected and analyzed by using a two-tailed Wilcoxon signed-rank test (SPSS17.0). The follow-up period ranged from 12 to 96 months, with an average of 24.8 months. Significance was set at P < 0.01. The 16-point Chicago Chiari Outcome Scale was also used to evaluate the result of the operation in all patients.

Fig. 4 a T2-weighted MR scan of a 36-year-old female patient with a 2-year history of chest zonesthesia, both upper extremities’ weakness, and hypaesthesia. She underwent syrinx-subarachnoid shunt surgery elsewhere. b One year later after operation, the syrinx slightly reduced, but the above symptoms aggravate gradually. Video 4, preoperative cine PC-MR shows CSF flow obstruction both at the posterior fossa space behind the cerebellum and tonsils and at the IV ventricle and central aqueduct (B + C). The adhesion of local arachnoid membrane was found (c). Meticulous release of adhesions (d). e The postoperative MRI 3 months later shows a reduction of the syrinx. The symptoms of the patient gradually improved. The mJOA score increased from preopera- tive value of 8 to 12 at the end of follow-up

Results

The 126 Chiari I malformation patients in this study could be divided into three groups according to the location(s) of the CSF flow obstruction based on the cine PC-MRI findings. Thirty-six patients were found to have CSF flow blockage at the posterior fossa space behind the cerebellum and tonsils (C) and were classified as pattern I. Forty-eight patients had CSF flow blockage at the posterior fossa space behind the cerebellum and tonsils (C) and at the IV ventricle and cerebral aqueduct (B); this group was classified as patternII. The remaining 42 patients were found to have CSF flow blockage at the posterior fossa space behind the cerebellum and tonsils, at the IV ventri- cle and central aqueduct, and at the ventral space between the clivus and brain stem (A+ B + C); this group was classified as pattern III.

Table 1 presents the general clinical data on the patients. In 90 (71.4%) patients diagnosed with Chiari I malformation, additional syringomyelia formation was found. Of the 126 patients, 48 (38%) underwent SDD and 78 (62%) received SAM. The intraoperative ultrasound reevaluation showed that two patients under pattern I had inadequate CSF flow before the dural opening; in comparison, the preoperative cine PC- MRI indicated that two patients under pattern III had adequate CSF flow.The intraoperative ultrasound also showed that most of the patients under pattern II (36 of 48) required SAM (Table 2).

The postoperative MRIs indicated a near-com- plete resolution or a significant shrinkage of the syrinx in 84.2% (32 of 38) of the patients who underwent SDD, compared with 88.5% (46 of 52) of the patients who received SAM. Two weeks after the operation, the mJOA scores increased from the preoperative value of 10.67 ± 1.61 to 12.74 ± 2.01(P < 0.01). The mean follow-up dura- tion was 24.8 months; mJOA scores showed no statistically significant difference between the two procedures (Table 3). According to the Chicago Chiari Outcome Scale, patients underwent SD mean scores of 14.70 to14.67 with the patients who underwent SAM.

Among the patients under pattern III, six cases underwent index surgery elsewhere, and the opera- tion carried out at our institution was revision surgery secondary to the progression of the syrinx size after the initial surgery and to worsening symptoms; all six patients underwent SAM. During the revision surgery, tedious sharp dissection and the meticulous release of adhesions were critical. With the successful restoration of the CSF flow, the syrinx showed a dramatic decrease in size in the MRI carried out 2 weeks after the surgery. Two of the patients under pattern III who did not show an obvious shrinkage of the syrinx after a reversion surgery of SAM underwent a subsequent syrin- go-pleural shunting (SPS) procedure, which was deemed necessary to drain the CSF flow out of the syrinx to improve the neurologic function.

leakage, meningiocele, any worsening of the preoperative neurological deficits, and any new postoperative neurological deficits are generally considered as surgical complications of Chiari I malformation. We did not see any of such neuro- logical deficits as posterior cranial nerve disorder,

transitory spinal accessory nerve injury, and cere- bellar or brain stem infarction in our study. The most common complications after posterior fossa decompression were fever and even CSF leaks at the surgery site. These complications occurred more often in the patients who underwent SAM (9.5%, 11 of 78) than in those who received SDD (3.5%, 2 of 48) (Table 3). In the case of cerebrospi- nal fluid leakage, we routinely set up a lumbar cistern drain for 5 to 7 days. No postoperative worsening of neuro- logic deficits was observed in our series.

Discussion

Previous studies have reported that the forma- tion of the syrinx is closely related to abnormal CSF flow dynamics at the CVJ. Twenty years ago, Oldfield et al. proposed that in patients with Chiari I malformation, each systolic pressure wave in the spinal cord may cause the tonsils to push inferiorly,causing elevated pressures in the spinal cord that force CSF into the central canal along perivascular and interstitial spaces, which lead to the syrinx. A cine PC-MRI assessment of the CSF flow may allow an objective measurement of the direct phys- iologic consequence of tonsillar ectopia, which may better represent the specific degree of overall hindbrain pathology and damage in the patient. It is very helpful to demonstrate a CSF flow obstruction as an indicator of a clinically relevant herniation.

Among patients with different degrees of tonsillar herniation, a wide degree of CSF flow pathology has been observed. A greater physiolog- ic consequence of tonsillar ectopia may manifest as greater abnormality in the hindbrain CSF flow. An extension of this concept would imply that increas- ing the physiologic consequence of tonsillar ecto- pia might first cause CSF flow obstruction dorsal to the cerebellum and tonsils and then further result in additional CSF flow obstruction at the central aqueduct and the ventral space between the clivus and brain stem. Thus, the CSF flow dynamics was detected in three areas: the ventral space between the clivus and brain stem at the level of the foramen magnum (A), the central aqueduct, including the foramen of Magendie (B), and the posterior fossa space behind the cerebellum and tonsils (C).

The CSF flow dynamics could be classified into three pat- terns: (1) obstruction at the posterior fossa space behind the cerebellum and tonsils (C), (2) obstruction both at the posterior fossa space behind the cerebellum and tonsils and at the IV ventricle and central aqueduct (B + C), and (3) obstruction at the posterior fossa space behind the cerebellum and tonsils, at the IV ventricle and central aqueduct, and at the ventral space between the clivus and brain stem (A + B + C). In the series studied, the syrinx was observed mainly in patients under patterns II and III. This finding further explained the fact that the formation of the syrinx was closely related to the obstruction of the CSF flow dynamics at the IV ventricle and cerebral aqueduct. Based on the three CSF flow patterns in Chiari I malformation, an effective decompression procedure could be selected to restore the CSF flow dynamics and further shrink the syrinx. It is important to recognize that the procedure used to treat the underlying etiology (e.g., structural collapse or a remarkable shrinkage of the syrinx) could ameliorate or at least stabilize the symptoms of the patient and offers the best opportunity to achieve long-term improvement.

SDD and SAM are generally accepted as the effective surgical procedures in the treatment of Chiari I malformation, and their effects have been confirmed by clinical practice. Pure craniectomy alone does not achieve adequate decompression, as confirmed by intraoperative ultrasonography; thus,the effect of such operation is limited. After a small window craniectomy, SDD and SAM can effective- ly restore the blocked CSF flow at the CVJ while expanding the volume and size of the posterior fossa. The two procedures have been reported to be effective and have been previously compared in the literature. Romero and Pereira found that 9 of 10 patients improved with SDD and thus concluded that SDD was effective in Chiari I malformation (CM1) cases associated with a syrinx. Another

previous study reported on 22 adult patients, with the conclusion that SAM may improve the symp- toms in Chiari I malformation. Similar improve- ments in the syrinx without need for decompres-sion were observed in 15 infantile cases of Chiari I malformation with tonsillar reduction, which underwent a combination of resection and cauteri- zation. In an older study, seven of eight patients who underwent SAM showed “very good” clinical outcomes. Thus far, no author has applied these two methods in selected Chiari I malformation patients.

In the present series, the decompression procedure was chosen according to the different CSF flow dynamics at the CVJ. After a small window craniectomy, an intraoperative ultrasound was done to detect subdural space. If the CSF flow was ameliorated in areas A, B, and C, SDD was carried out directly. For the patients under pattern I and for some cases under pattern II, an SDD proce- dure was deemed more suitable. However, for the rest of the patients under patterns II and III, SAM was needed to resect the herniated cerebellar tonsils and/or release the arachnoid membrane adhesion to restore the cerebrospinal fluid outflow at the CVJ. Two patients showed no change in the syrinx after the SAM procedure. When the CSF flow restoration fails, we suggest the application of CSF-diverting procedures, such as SPS, to shrink the syrinx, as presented in our previous research.

In this study, SDD and SAM were found to be associated with favorable results on clinical signs and symptoms, and no meaningful difference between the two procedures was observed. Howev- er, SAM showed a higher rate of complications compared with SDD. Postoperative MRIs showed a near complete resolution or a significant shrinkage of the syrinx in 32 of 38 (84.2%) patients who underwent SDD, compared with 46 of 52 (88.5%)patients who received SAM. The statistical analysis of the mJOA scores showed no statistically significant difference between the two procedures. However, more complications occurred in patients who underwent SAM than in those who received SDD.

Some authors recommend cerebellar tonsil resection, whereas others do not. Stanko et al. Suggested that tonsillar cautery might provide an extra benefit in the resolution of the syrinx compared with bone-only decompression alone or in combination with dural opening. Yilmaz et al. reported that in cases of tonsillar descent over the C-1 arch, duraplasty could play an important role in shrinking the syrinx and improving the clinical outcome. Both decompression procedures (SDD and SAM) could be effective. Our study suggests that SDD and SAM may both provide an extra benefit in the resolution of the syrinx and the improvement of the clinical outcome. SDD may lead to lesser complications but may have an inadequate decompression effect, whereas SAM may have a more sufficient decompression effect but may lead to more complications. Based on the different cerebrospinal fluid flow patterns at the CVJ, as obtained by cine PC-MRI and intraoperative ultrasound, either SDD or SAM can be used to effectively restore the CSF flow and shrink the syrinx at lower costs and with lesser risks.

Conclusion

The CSF flow dynamics in Chiari I malformation could be classified into three patterns: (1) obstruction at the posterior fossa space behind the cerebellum and tonsils, (2) obstruction both at the posterior fossa space behind the cerebellum and tonsils and at the IV ventricle and central aqueduct, and (3) obstruction at the posterior fossa space behind the cerebellumand tonsils, at the IV ventricle and central aqueduct, and at the ventral space between the clivus and brain stem. An SAM procedure is more feasible in patients under pattern III, whereas an SDD procedure is more suitable for patients under pattern I. In pattern II patients, the intraoperative ultrasound findings could play an important role in the selection of the procedure.

Author contribution Tao Fan, HaiJun Zhao,XinGang Zhao, Cong Liang, YinQian Wang, QiFeiGai:

Conception and design of the study,acquisition and interpretation of data, drafting theCompliance with ethical standards Financial support This study was funded by Construction Project of National Clinical Key Specialties of People’s Republic of China [Ministry of Health of People’s Republic of China 873(2011)] and the Capital Health Research and Development of Special 2014-2-8011. And the corresponding author Tao Fan received the support of those funding.

Conflict of interest The authors declare that they have no conflict of interest.

Ethical approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. For this type of study, formal consent is not required. This article does not contain any studies with human participants performed by any of the authors.

Informed consent Informed consent was obtained from all individualparticipants included in the study.