You are here

Article Text

Article menu

Download PDFPDF

Neurosurgery
Research paper
Clinical presentation, natural history and outcomes of intramedullary spinal cord cavernous malformations
  1. Anshit Goyal1,
  2. Lorenzo Rinaldo1,
  3. Redab Alkhataybeh1,
  4. Panagiotis Kerezoudis1,
  5. Mohammed Ali Alvi1,
  6. Kelly D Flemming2,
  7. Lindsy Williams2,
  8. Felix Diehn3,
  9. Mohamad Bydon1
  1. 1 Neurosurgery, Mayo Clinic, Rochester, New York, USA
  2. 2 Neurology, Mayo Clinic, Rochester, New York, USA
  3. 3 Radiology, Mayo Clinic, Rochester, New York, USA
  1. Correspondence to Dr Mohamad Bydon, Department of Neurologic Surgery, Mayo Clinic, Rochester, MN 55905, USA; bydon.mohamad{at}mayo.edu

Abstract

Objective There is a paucity of literature investigating the clinical course of patients with spinal intramedullary cavernous malformations (ISCMs). We present a large case series of ISCMs to describe clinical presentation, natural history and outcomes of both surgical and conservative management.

Methods We retrospectively reviewed the clinical course of patients diagnosed with ISCMs at our institution between 1995 and 2016. Haemorrhage was defined as clinical worsening in tandem with imaging changes visualised on follow-up MRI. Outcomes assessed included neurological status and annual haemorrhage rates.

Results A total of 107 patients met inclusion criteria. Follow-up data were available for 85 patients. While 21 (24.7%) patients underwent immediate surgical resection, 64 (75.3%) were initially managed conservatively. Among this latter group, 16 (25.0%) suffered a haemorrhage during follow-up and 11 (17.2%) required surgical resection due to interval bleeding or neurological worsening. The overall annual risk of haemorrhage was 5.5% per person year. The rate among patients who were symptomatic and asymptomatic on presentation was 9.5% and 0.8%, respectively. Median time to haemorrhage was 2.3 years (0.1–12.3). Univariate analysis identified higher ISCM size (p=0.024), history of prior haemorrhage (p=0.013) and presence of symptoms (p=0.003) as risk factors for subsequent haemorrhage. Multivariable proportional hazards analysis revealed presence of symptoms to be independently associated with haemorrhage during follow-up (HR 9.39, CI 1.86 to 170.8, p=0.013).

Conclusion Large, symptomatic ISCMs appear to be at increased risk for subsequent haemorrhage. Surgery may be considered in such lesions to prevent rebleeding and subsequent neurological worsening.

  • spinal cord
  • cavernoma
  • intramedullary
  • cavernous malformation
  • natural history
  • surgical resection
  • neurologic outcomes
  • hemorrhage
  • bleeding risk

https://doi.org/10.1136/jnnp-2018-319553

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    Cavernous malformations (CMs) are vascular malformations of the central nervous system, with a reported prevalence of 0.4%–0.6% in the general population.1–6 Most lesions are found intracranially within the supratentorial compartment.1 2 5 7–9 Intramedullary spinal CMs (ISCMs), by contrast, constitute a rare pathology, accounting for 5%–12% of all intraspinal vascular malformations.10–12

    With the advent of high-resolution MRI, these lesions are being more commonly detected in the general population, but overall prevalence still remains low.13 Given their overall rarity, in contrast to cerebral CMs, data on natural history, prognosis and outcomes following surgical or conservative management are limited for ISCMs and exist mostly in the form of case series and autopsy reports.14–22 Data also remain limited on the natural course of asymptomatic incidentally detected ISCMs. While surgical resection is considered the only definitive treatment for symptomatic lesions at this time,23 indications for surgery remain controversial on account of varied clinical outcomes.24 In this study, we aimed to describe the clinical presentation, natural history and outcomes of intramedullary spinal ISCMs through a retrospective review of patients diagnosed and treated at our institution. We also sought to identify predictors of favourable neurological outcomes and recurrent haemorrhage.

    Patients and methods

    Patient population

    Following Institutional Review Board (IRB) approval, we reviewed medical records of patients diagnosed with ISCMs between 1995 and 2016 at our institution. Patients were referred to our institution from local, national and international sources and underwent multidisciplinary evaluation by neurology and neurosurgery on presentation. Patients were considered to have an ISCM when (1) an intramedullary spinal cord lesion was identified as a CM by a board-certified neuroradiologist (FD) on MRI or (2) an intramedullary spinal cord lesion was histopathologically identified as a CM following surgical resection. Patients who underwent lesion resection prior to presentation at our institution were excluded from this study. Records were independently reviewed by three authors (AG, LR, RA) with any discrepancy resolved by discussion and further by the senior author (MB). Follow-up data were also obtained using chart and imaging review. This study was reported in accordance with the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines.25

    Data collection and outcome definitions

    Pertinent clinical variables such as demographics, neurological signs and symptoms, prior history of haemorrhage, family history of CMs, synchronous intracranial CMs, treatment offered (surgical vs conservative), use of anticoagulation and strenuous physical activity before bleeding were recorded and analysed for all eligible patients. Routine activities of daily living (ADLs) were considered non-strenuous while activities which require higher effort than ADLs such as running, jogging and weightlifting were considered as strenuous activities. The clinical course prior to presentation was classified into five types as described by Ogilvy et al:4 (1) Type A: discrete episodes of neurological deterioration with varying degrees of recovery between episodes, (2) Type B: slow progression of neurological decline, (3) Type C: acute onset of symptoms with rapid decline, (4) Type D: acute onset of mild symptoms with subsequent gradual decline lasting weeks to months and (v) Type E: asymptomatic incidentally detected lesions. Imaging characteristics such as size of CM, location along the spinal cord, predominant T1 and T2 MR signal were also recorded by a senior board-certified neuroradiologist (FD). Size was defined as the largest measurement in either dimension (anteroposterior, transverse and craniocaudal). Perilesional hemosiderin rim was not included in size measurements. Modified McCormick Score (MMcS) was used to grade neurological status at initial presentation, immediately prior to surgery, immediately following surgery and at last follow-up26 (table 1). Haemorrhage on follow-up was defined as neurological worsening in tandem with imaging evidence of acute bleeding. Figure 1 illustrates a representative example of an intramedullary CM as seen at initial presentation and following acute haemorrhage.

    Figure 1
    Figure 1

    Sagittal T2-weighted MR images show a cavernous malformation in the posterior part of the cervical cord at the C6 level as seen at (A) initial presentation and (B) following an acute haemorrhage.

    View this table:
    Table 1

    Modified McCormick Scale

    Management modalities

    Patients were managed conservatively or with surgical resection. Conservative management entailed serial MRIs with the interval at the discretion of the attending neurosurgeon or neurologist. In general, the interval between imaging studies was increased if the patient’s symptomatology remained stable and there was no change in the imaging appearance of the lesion. Imaging was typically obtained immediately after a patient reported neurological worsening. Patients treated with surgical resection were positioned prone on a standard operating table. Neuromonitoring was used at the discretion of the treating neurosurgeon. A focal laminectomy(s) was performed and a dural incision was made along the long axis of the thecal sac. A small myelotomy was made at the point where the cavernoma most closely approached the pial surface, typically evidenced by focal discoloration or hemosiderin deposition, and the lesion was removed in a piecemeal fashion. The dura was then closed in a water-tight fashion and the wound was closed in layers.

    Statistical analysis

    We performed descriptive statistics to assess baseline patient demographic and clinical characteristics. Continuous variables were compared using unpaired, two-tailed student’s t-test (parametric) or the Wilcoxon rank-sum test (non-parametric), while categorical variables were compared using the χ² test (parametric) or the Fisher’s exact test (non-parametric). Kaplan-Meier and cox proportional hazards methods were used for time to event analysis for haemorrhage on follow-up. For time to haemorrhage analysis, date of initial clinical presentation was used as the initial common starting point. Patients were censored at last known follow-up. Surgically managed patients were censored on the date of surgery. All p values were two-sided. Statistical significance was defined as p<0.05.

    Results

    Baseline characteristics and clinical presentation

    A total of 107 patients met inclusion criteria. Mean age at presentation was 49.6±17.3 years with 45% (n=48) being females. The majority of patients had an ISCM located in the thoracic spinal cord (59%, n=63). Synchronous intracranial lesions were present in 24.3% (n=26) of patients, while 10.3% (n=11) had a family history of CMs. The majority of patients (78.5%, n=85) were symptomatic at presentation while in the remainder, the lesion was discovered incidentally. Among symptomatic patients, a combination of motor/sensory/pain/sphincter symptoms was most common (n=23, 21.5%). Twenty patients (23%) complained of radicular/central pain at presentation and among them, sixteen patients reported pain as the predominant symptom. Only 11.2% (n=12) of patients had severe neurological deficit with grade IV/V on the MMcS. The majority (n=96, 89.7%) were ambulatory on presentation. 66.4% (n=71) had a history of haemorrhage prior to presentation. Among these patients, 32.4% (n=23) reported some degree of strenuous (n=10, 14.1%) or non-strenuous activity (n=13, 18.3%) immediately prior to onset of symptoms. The remainder of patients could not identify an event associated with symptom occurrence. Table 2 summarises baseline characteristics and clinical presentation of the cohort. In order to characterise the course of the disease better based on initial management, we further stratified the patients who received initial conservative management (n=64) and those who received upfront surgery (defined as time to surgery <3 months, n=21). Patients who received surgery at initial presentation were younger (Mean age: 52.5 vs 41.2, p=0.01), less likely to have a synchronous intracranial lesion (p=0.02), more likely to be symptomatic at presentation (p=0.02) and have had a prior haemorrhage in the cavernoma (p=0.03) (table 3).

    View this table:
    Table 2

    Patient characteristics on presentation*

    View this table:
    Table 3

    Comparison of patients receiving conservative management (n=64) and surgery (n=21) following initial evaluation

    Neurological outcomes and haemorrhage rates

    Follow-up outcomes data were available for 85 patients. While 64 patients were managed conservatively following initial presentation (natural history follow up-289.96 person years), 21 underwent immediate surgical resection (table 2). There were 16 patients (25%) who experienced a haemorrhage on follow-up, 11 (69%) of whom subsequently underwent surgical resection, bringing the total number of patients treated surgically to 32. Among conservatively managed patients, Table 4 highlights the differences between those who eventually required surgery (n=11) and those who continued to remain conservatively managed (n=53). The mean time to surgery was 358 days. Figure 2A illustrates the Kaplan-Meier curve for incident haemorrhage on follow-up (n=64). The median time to haemorrhage was 2.3 years. The overall annual haemorrhage rate in patients not requiring upfront surgery was calculated at 5.5% per person year. A higher annual haemorrhage rate was found in patients who were symptomatic at presentation (8.9%) and patients with a prior haemorrhage (9%). Follow-up haemorrhage rates were found to be lower in patients without any history of prior haemorrhage (2.1%) and in asymptomatic patients with incidentally detected lesions (0.8%). Figure 2B depicts Kaplan-Meier curves for haemorrhage in patients symptomatic and asymptomatic on presentation.

    Figure 2
    Figure 2

    Kaplan-Meier curves depicting (A) haemorrhage on follow-up from initial presentation in overall cohort (n=64) and (B) comparison of haemorrhage on follow-up between patients symptomatic and asymptomatic on presentation.

    View this table:
    Table 4

    Comparison of patients eventually receiving conservative management (n=53) and surgery (n=11) from the initial conservatively managed cohort (n=64)

    Among patients treated with surgery (n=32), 10 patients improved, 15 remained unchanged and 7 underwent neurological worsening as measured by MMcS compared with neurological baseline immediately prior to surgery. Gross total resection was achieved in 26 patients (93%) (extent of resection was not available in 4 patients). Table 5 highlights characteristics of patients who did (n=7) and did not (n=25) experience neurological worsening (according to MMcS measured at last follow-up) following surgical resection. No statistically significant differences were observed in terms of age, sex, time to surgery between the two groups. Neurological worsening was less frequently seen in patients with lesion size >1 cm (p=0.02). We found that while 13 patients exhibited worsening of neurological status immediately following surgery compared with preoperative MMcS, only 5 of these patients continued to exhibit worsened neurological function at late follow-up. However, while 19 patients had preserved neurological status immediately following surgery, 2 of these patients demonstrated worsening of MMcS at long-term follow-up. One of these patients experienced progressive lower extremity weakness due to spinal cord tethering while the other patient demonstrated worsened gait disturbance attributed to posterior column dysfunction although lower extremity muscle strength continued to be well preserved. Neurological outcomes tended to be comparable in the 52 conservatively managed patients who did not receive surgical intervention at any point, with only two patients (4%) experiencing worsening and the remaining patients demonstrating preserved (n=43, 82%) or improved (n=7, 14%) neurological function as per MMcS grade at last follow-up (compared with MMcS at baseline presentation).

    View this table:
    Table 5

    Comparison between patients who did and did not worsen following surgery‡

    We found that out of six patients who complained of preoperative pain, only one patient reported worsening of pain following surgical resection while two showed improvement and three reported no change in the level of pain. However, in the remainder of the surgically managed group, six patients developed new onset radicular/central pain that was absent prior to surgery. Overall, no statistical significance was noted between patients with and without neurological worsening at last follow-up in terms of baseline clinical characteristics and presentation (online supplementary table 1).

    Supplemental material

    Factors associated with haemorrhage on follow-up

    Natural history was evaluated for 64 patients who underwent initial conservative management following presentation (natural history—336.6 person years). Table 6 highlights clinical and radiological differences in patients with (n=16) and without haemorrhage (n=48) on follow-up. Patients who experienced a haemorrhage on follow-up tended to be symptomatic at initial presentation (p=0.027) and have experienced an overt clinical haemorrhage prior to presentation (p=0.040). A significantly higher proportion of patients in the haemorrhage group had lesion size >1 cm (p=0.023). No statistically significant differences were found between the two groups in terms of age, sex, location of lesion (cervical vs thoracic), family history of CMs, presence of synchronous intracranial lesions or duration of symptoms prior to presentation. Also, with the exception of size, no statistically significant differences were found between the two groups in radiological characteristics of the lesion at baseline presentation (predominant T1/T2 signal, presence of spinal cord oedema). On univariable Cox proportional hazards analysis, larger lesion size at presentation (p=0.02), prior history of haemorrhage (p=0.01), and presence of symptoms at baseline (p=0.003) were statistically significant predictors of haemorrhage on follow-up. Size of lesion >1 cm was also associated with haemorrhage (analysed categorically), although results were not statistically significant(p=0.06). Presence of symptoms was the only statistically significant predictor on multivariable proportional hazards analysis (p=0.013) (table 7).

    View this table:
    Table 6

    Comparison of patients with and without haemorrhage during follow-up

    View this table:
    Table 7

    Cox proportional hazards analysis identifying predictors of haemorrhage

    Discussion

    ISCMs constitute nearly 5%–12% of all spinal vascular malformations,1 12 24 though their true incidence is controversial. The natural history of these lesions is incompletely understood, as detailed in a recent review by Al-Shahi Salman et al,27 in contrast to cerebral CMs, for which consensus management guidelines are now available.28 Currently, most series on ISCMs report outcomes following surgical resection,23 29–34 while literature on natural history and outcomes following conservative management is more limited.35–37 In the present study, we report the outcomes of both conservatively and surgically managed patients, with the aim of better understanding the natural history of ISCMs and refining indications for surgical resection.

    Compared with cranial CMs, ISCMs tend to follow a more aggressive clinical course, presumably due to spinal cord’s lower tolerance for space-occupying lesions.32 Histologically, they consist of endothelial lined vascular spaces of varying sizes embedded in a connective tissue matrix.13 38 While there is often gliotic and hemosiderin-stained neural parenchyma around the lesion, there is no normal nervous tissue within the lesion itself, thus facilitating surgical resection.4 36 38 39 With regard to their location along the neuraxis, in our series, although the majority of lesions were located in the thoracic spinal cord (59%), a relatively large meta-analysis of 632 patients by Badhiwala et al revealed an almost equal distribution between the cervical and thoracic spinal cord regions.24 Consistent with prior reports, mean age of diagnosis corresponded to the 3rd-5th decade of life.24 29 30 33 35 A slightly higher male:female ratio (1.22:1) was noted in our series in terms of sex distribution which was similar to that described in the literature.24 36 37

    According to available literature, patients with ISCM tend to present with chronic progressive myelopathy.24 In our series, patients more commonly tended to present with either acute onset of neurological symptoms followed by a gradual decline (type D) or discrete episodes of neurological deterioration with varying amount of neurological recovery between episodes (type A).

    Natural history and predictors of haemorrhage

    Among 85 patients, initial conservative management was offered to 64 patients (total 336.6 person years of follow-up) and 16 patients incurred a haemorrhage on follow-up, leading to an annual haemorrhage rate of 5.5%, which is on the higher end of the spectrum of haemorrhage rates reported in the literature, ranging from 1.4% to 6.8%.22 24 29 30 34 36 40 In the largest meta-analysis of patient level data to date, including 632 patients across 40 studies, Badhiwala et al calculated an annual haemorrhage rate of 2.1%. One reason behind differing haemorrhage rates might be attributed to the variable definitions of haemorrhage across the literature.24 We considered haemorrhage to be consistent with new blood products on imaging in tandem with worsening of clinical symptoms.

    In addition, our series is the first to stratify rate of haemorrhage by patient clinical profile. Symptomatic patients and those with a prior history of haemorrhage had significantly higher annual haemorrhage rates (9.5% and 9.7%, respectively). In contrast, annual risk for haemorrhage was negligible for asymptomatic patients (0.8%). We found lesion size >1 cm, presence of symptoms and prior haemorrhage to be significant predictors for a haemorrhage on follow-up. This highlights that while oligosymptomatic patients may be managed conservatively, symptomatic patients with large lesions (>1 cm) may benefit from early surgery to prevent subsequent haemorrhage and neurological worsening, which is consistent with recommendations from existing case series of ISCMs.23 29–31 34 41 Recommendations from existing studies may be biased, however, given that most of them preferentially describe outcomes following surgical resection only. Our series, on the other hand, highlights a possible role for surgery in a carefully selected group of patients. While some may argue that overall annual haemorrhage rates might be low, the cumulative bleeding risk incurred might justify surgical resection, especially in younger patients.36 As mentioned earlier, we found evidence for a positive association between favourable postoperative neurological outcome and larger size of lesion (>1 cm).

    Neurological outcomes

    There were no significant baseline clinical differences between patients who did and did not worsen neurologically, which could be related to a small sample size (online supplementary table 1). Nonetheless, in the overall cohort, notable trends included a larger size of lesion and a higher probability of being symptomatic in patients who did worsen.

    In a large retrospective review by Zhang et al, the authors noted similar neurological outcomes following surgical and conservative management.36 In our cohort, neurological outcomes tended to be comparable in 50 out of 52 conservatively managed patients (with available MMcS grade at last follow-up) who did not receive surgical intervention at any point. However, the equivocal outcomes must be viewed in the larger context of treatment selection bias between the two approaches, as symptomatic patients were more likely to receive surgery. As evidenced in table 3, patients receiving surgery upfront were more likely to be younger, symptomatic and have a solitary lesion in the spinal cord that had previously haemorrhaged. Conservative management was offered to 64 patients at initial presentation, with 11 of these patients ultimately receiving surgery due to subsequent haemorrhage and worsening. However, conservative management seems to be a reasonable strategy for incidentally detected lesions given the low haemorrhage rates observed in our cohort. Among patients receiving surgery, 7 out of 32 patients had worse MMcS grade at last follow-up (compared with immediate preoperative baseline). A total of 13 patients exhibited neurological worsening in the immediate postoperative period, but eight patients returned to their preoperative baseline at last follow-up. It is important to point out that two patients worsened on follow-up despite having stable neurological status immediately following surgery, due to progressive spinal cord tethering in one patient and posterior column dysfunction in another. Following surgery, however, resection of smaller lesions (<1 cm) was associated with a greater risk for worsening of neurological deficit at last follow-up (table 5). We hypothesise that this might be explained by more extensive spinal cord dissection required to resect smaller lesions, as opposed to a lesser degree of parenchymal manipulation required for resection of larger lesions rising to the pial surface. Similar findings have been described by Badhiwala et al, in a meta-analysis of individual patient data.24 In addition to lesion size >1 cm, the authors noted time to surgery <3 months, surgical technique via hemilaminectomy, gross total resection and predominantly motor symptoms to be predictive of an improved neurological outcome following surgery. Presence of preoperative sensory/pain symptoms was suggested to be predictive of poor postoperative neurological recovery. Similarly, in our series, a shorter time to surgery was observed in our cohort in patients with preserved neurological function postoperatively, although it was not statistically significant.

    Pain control also remains a contentious issue in patients with spinal cord cavernomas, with most authors remaining sceptical about the efficacy of lesion resection to improve pain. In the current series, out of six patients who presented with pain as the predominant symptom prior to surgery, five patients had at least some improvement in pain while one patient experienced worsening of pain intensity. Moreover, six patients developed new onset pain that was absent preoperatively. In a series by Kharkar et al, four patients having pain predominant presentation underwent surgery.42 While one patient reported complete pain resolution, the others had limited or no improvement. Deutsch et al reported a series of five patients with ISCMs presenting predominantly with pain, with all patients reporting some degree of pain resolution following surgery.33 In one of the largest cohorts (n=23) of surgically managed patients presenting with pain, Kim et al found pain improvement in 78% of patients in the immediate postoperative period, but sustained pain relief after 1 year was only present in 52% of patients.43 The reason behind such conflicting outcomes remains unclear. Individual-level variation in host response (inflammation, gliosis and subsequent scarring) to preoperative blood products and subsequent parenchymal/lesion manipulation at the time of surgery could potentially explain the inconsistency in observed pain outcomes.

    Limitations

    There were numerous limitations to this study. First, it was retrospective observational by design and based on single-institutional data. Though this represents a large series of ISCMs, the relatively small sample size increases the likelihood of type I error in our analysis. Given the relatively limited information on the natural history of ISCMs, this was deemed justifiable. Future, prospective studies will be needed to confirm our results. Second, although records were viewed retrospectively by two independent reviewers, the quality of data regarding prior haemorrhage and clinical course was limited by patient recall bias. We tried to compensate for this using prespecified exposure and outcome definitions (see Methods section). Imaging was reviewed by a board-certified neuroradiologist, and the width of hemosiderin rim on MRI was excluded from measurements to record the size most accurately. Third, we lost 20% of our patients to follow-up. Fifth, only 75% of the 85 patients received conservative management at initial presentation, which presents a significant selection bias to treatment. Nevertheless, this remains one of the largest series on spinal intramedullary CMs to date, including patients undergoing both conservative and surgical management, providing plausible hypotheses that can be perhaps tested with larger studies in the future. In our series, we also attempted to identify the nature of physical activity preceding symptom onset, and although we were confronted with substantial missing data in this regard, a history of preceding trauma or strenuous activity could be discerned from chart review in nearly 40% (14/35) of remaining patients. This observation brings to attention a previously undescribed potential risk factor (for acute neurological worsening due to haemorrhage) which might be worthy of exploration in future studies.

    Directions for future investigations

    Given our study limitations, future prospective studies with a larger sample size are needed. This can be challenging given the rare incidence of these lesions. On the other hand, it may be argued that this calls for the need for multi-institutional collaborative efforts to pool data on outcomes and natural history. In today’s era for outcomes based management, building disease registries is one example of such an effort that can bring us one step closer to forming consensus guidelines. The Angioma Alliance (http://www.angiomaalliance.org) is a commendable example to pool available data and provide public awareness for patients with CMs. Literature reviews combining individual patient data (eg, Badhiwala et al 24) can be useful to address gaps in knowledge that are otherwise left unanswered by single-institutional studies. The above efforts were instrumental in describing systematic consensus guidelines for management of cerebral CMs.28 Propensity matching can be a useful technique to compensate for treatment selection bias and provide a fair comparison of outcomes between surgical and conservative management. In addition to haemorrhage risk and neurological function, quality of life and pain outcomes remain important concerns for patients with spinal cord CMs and more studies reporting such outcomes would also be welcome investigations for the future.

    Conclusion

    Intramedullary CMs are rare vascular malformations of the spinal cord. Symptomatic patients, patients with a large lesion size and prior history of haemorrhage may be at higher risk for bleeding. While surgical intervention may be deferred for asymptomatic patients, lesion resection in carefully selected symptomatic patients may prevent haemorrhage and halt subsequent worsening of neurological function. Given the paucity of existing data, larger studies and collaborative efforts to pool data on outcomes and natural history may provide basis for more definitive evidence-based management.

    References

    1. ↵
      2. Gross BA ,
      3. Lin N ,
      4. Du R , et al
      . The natural history of intracranial cavernous malformations. Neurosurg Focus 2011;26.doi:10.3171/2011.3.FOCUS1165
    2. ↵
      2. Del Curling O ,
      3. Kelly DL ,
      4. Elster AD , et al
      . An analysis of the natural history of cavernous angiomas. J Neurosurg 1991;6:702–8.doi:10.3171/jns.1991.75.5.0702
      OpenUrl
    3. ↵
      2. Kim DS ,
      3. Park YG ,
      4. Choi JU , et al
      . An analysis of the natural history of cavernous malformations. Surg Neurol 1997;48:9–17. discussion 17–8.doi:10.1016/S0090-3019(96)00425-9
      OpenUrlCrossRefPubMedWeb of Science
    4. ↵
      2. Ogilvy CS ,
      3. Louis DN ,
      4. Ojemann RG
      . Intramedullary cavernous angiomas of the spinal cord: clinical presentation, pathological features, and surgical management. Neurosurgery 1992;31:219–29. discussion 229–30.
      OpenUrlCrossRefPubMedWeb of Science
    5. ↵
      2. Robinson JR ,
      3. Awad IA ,
      4. Little JR
      . Natural history of the cavernous angioma. J Neurosurg 1991;41:709–14.doi:10.3171/jns.1991.75.5.0709
      OpenUrl
    6. ↵
      2. Sarwar M ,
      3. McCormick WF
      . Case report and review. Arch Neurol 1978;35:323–5.
      OpenUrlCrossRefPubMedWeb of Science
    7. ↵
      2. Cantu C ,
      3. Murillo-Bonilla L ,
      4. Arauz A , et al
      . Predictive factors for intracerebral hemorrhage in patients with cavernous angiomas. Neurol Res 2005;27:314–8.doi:10.1179/016164105X39914
      OpenUrlCrossRefPubMed
    8. ↵
      2. Kivelev J ,
      3. Laakso A ,
      4. Niemelä M , et al
      . A proposed grading system of brain and spinal cavernomas. Neurosurgery 2011;69:807–14. discussion 813–4.doi:10.1227/NEU.0b013e31821ffbb5
      OpenUrlCrossRefPubMed
    9. ↵
      2. Voigt K ,
      3. Yaşargil MG
      . Cerebral cavernous haemangiomas or cavernomas. Incidence, pathology, localization, diagnosis, clinical features and treatment. Review of the literature and report of an unusual case. Neurochirurgia 1976;19:59–68.doi:10.1055/s-0028-1090391
      OpenUrlPubMed
    10. ↵
      2. Cosgrove GR ,
      3. Bertrand G ,
      4. Fontaine S , et al
      . Cavernous angiomas of the spinal cord. J Neurosurg 1988;131:31–6.doi:10.3171/jns.1988.68.1.0031
      OpenUrl
    11. ↵
      2. Gross BA ,
      3. Du R ,
      4. Popp AJ , et al
      . Intramedullary spinal cord cavernous malformations. Neurosurg Focus 2010;40.doi:10.3171/2010.6.FOCUS10144
    12. ↵
      2. Jellinger K
      . Pathology of spinal vascular malformations and vascular tumors. In: Spinal angiomas. Springer Berlin Heidelberg, 1978: 18–44.
    13. ↵
      2. Lanzino G
      . Cavernous malformations of the brain and spinal cord. Thieme, 2011.
    14. ↵
      2. Barrena MR ,
      3. Guelbenzu S ,
      4. Mayayo-Sinues E , et al
      . A clinico radiological study of patients with cavernous angiomas of the spinal cord. Rev Neurol 2003;37:1–7.
      OpenUrlPubMed
    15. ↵
      2. Bucciero A ,
      3. del Basso de Caro M ,
      4. Carangelo B , et al
      . Intramedullary cavernoma: a case report and review of the literature. Acta Neurol 1994;16:162–9.
      OpenUrl
    16. ↵
      2. Hadlich R
      . Ein fall von tumor cavernosus des Rückenmarks MIT besonderer Berücksichtigung Der neueren Theorien über die Genese des Cavernoms. Virchows Arch path Anat 1903;172:429–45.doi:10.1007/BF01993932
      OpenUrl
    17. ↵
      2. Lee KS ,
      3. Spetzler RF
      . Spinal cord cavernous malformation in a patient with familial intracranial cavernous malformations. Neurosurgery 1990;26:877–80.doi:10.1227/00006123-199005000-00025
      OpenUrlCrossRefPubMed
    18. ↵
      2. Pagni CA ,
      3. Canavero S ,
      4. Forni M
      . Report of a cavernoma of the cauda equina and review of the literature. Surg Neurol 1990;33:124–31.doi:10.1016/0090-3019(90)90021-G
      OpenUrlCrossRefPubMed
    19. ↵
      2. Roncaroli F ,
      3. Scheithauer BW ,
      4. Krauss WE
      . Capillary hemangioma of the spinal cord. Report of four cases. J Neurosurg 2000;93(1 Suppl):148–51.
      OpenUrlCrossRefPubMedWeb of Science
    20. ↵
      2. Santoro A ,
      3. Piccirilli M ,
      4. Frati A , et al
      . Intramedullary spinal cord cavernous malformations: report of ten new cases. Neurosurg Rev 2004;27:93–8.doi:10.1007/s10143-003-0302-9
      OpenUrlPubMed
    21. ↵
      2. Zentner J ,
      3. Hassler W ,
      4. Gawehn J , et al
      . Intramedullary cavernous angiomas. Surg Neurol 1989;31:64–8.doi:10.1016/0090-3019(89)90220-6
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      2. Zevgaridis D ,
      3. Medele RJ ,
      4. Hamburger C , et al
      . Cavernous haemangiomas of the spinal cord. A review of 117 cases. Acta Neurochir 1999;141:237–45.doi:10.1007/s007010050293
      OpenUrlCrossRefPubMedWeb of Science
    23. ↵
      2. Mitha AP ,
      3. Turner JD ,
      4. Abla AA , et al
      . Outcomes following resection of intramedullary spinal cord cavernous malformations: a 25-year experience. J Neurosurg Spine 2011;14:605–11.doi:10.3171/2011.1.SPINE10454
      OpenUrlPubMed
    24. ↵
      2. Badhiwala JH ,
      3. Farrokhyar F ,
      4. Alhazzani W , et al
      . Surgical outcomes and natural history of intramedullary spinal cord cavernous malformations: a single-center series and meta-analysis of individual patient data. J Neurosurg Spine 2014;26:662–76.doi:10.3171/2014.6.SPINE13949
      OpenUrl
    25. ↵
      STROBE statement: home. Available: https://www.strobe-statement.org/index.php?id=strobe-home [Accessed 1 Apr 2018].
    26. ↵
      2. Klekamp J
      . Treatment of intramedullary tumors: analysis of surgical morbidity and long-term results. J Neurosurg Spine 2013;56:12–26.doi:10.3171/2013.3.SPINE121063
      OpenUrl
    27. ↵
      2. Al-Shahi Salman R ,
      3. Kitchen N ,
      4. Thomson J , et al
      . Top ten research priorities for brain and spine cavernous malformations. Lancet Neurol 2016;15:354–5.doi:10.1016/S1474-4422(16)00039-9
      OpenUrl
    28. ↵
      2. Akers A ,
      3. Al-Shahi Salman R ,
      4. A Awad I , et al
      . Synopsis of guidelines for the clinical management of cerebral cavernous malformations: consensus recommendations based on systematic literature review by the angioma alliance scientific Advisory Board clinical experts panel. Neurosurgery 2017;80:665–80.doi:10.1093/neuros/nyx091
      OpenUrl
    29. ↵
      2. Bian LG ,
      3. Bertalanffy H ,
      4. Sun QF , et al
      . Intramedullary cavernous malformations: clinical features and surgical technique via hemilaminectomy. Clin Neurol Neurosurg 2009;111:511–7.doi:10.1016/j.clineuro.2009.02.003
      OpenUrlPubMed
    30. ↵
      2. Sandalcioglu IE ,
      3. Wiedemayer H ,
      4. Gasser T , et al
      . Intramedullary spinal cord cavernous malformations: clinical features and risk of hemorrhage. Neurosurg Rev 2003;26:253–6.doi:10.1007/s10143-003-0260-2
      OpenUrlPubMed
    31. ↵
      2. Maslehaty H ,
      3. Barth H ,
      4. Petridis AK , et al
      . Symptomatic spinal cavernous malformations: indication for microsurgical treatment and outcome. Eur Spine J 2011;20:1765–70.doi:10.1007/s00586-011-1898-z
      OpenUrlPubMed
    32. ↵
      2. Deutsch H ,
      3. Jallo GI ,
      4. Faktorovich A , et al
      . Spinal intramedullary cavernoma: clinical presentation and surgical outcome. J Neurosurg 2000;74:65–70.doi:10.3171/spi.2000.93.1.0065
      OpenUrl
    33. ↵
      2. Deutsch H
      . Pain outcomes after surgery in patients with intramedullary spinal cord cavernous malformations. Neurosurg Focus 2010;93.doi:10.3171/2010.6.FOCUS10108
    34. ↵
      2. Aoyama T ,
      3. Hida K ,
      4. Houkin K
      . Intramedullary cavernous angiomas of the spinal cord: clinical characteristics of 13 lesions. Neurol. Med. Chir.(Tokyo) 2011;51:561–6.doi:10.2176/nmc.51.561
      OpenUrl
    35. ↵
      2. Steiger H-J ,
      3. Turowski B ,
      4. Hänggi D
      . Prognostic factors for the outcome of surgical and conservative treatment of symptomatic spinal cord cavernous malformations: a review of a series of 20 patients. Neurosurg Focus 2010;56.doi:10.3171/2010.6.FOCUS10123
    36. ↵
      2. Zhang L ,
      3. Yang W ,
      4. Jia W , et al
      . Comparison of outcome between surgical and conservative management of symptomatic spinal cord cavernous malformations. Neurosurgery 2016;78:552–61.doi:10.1227/NEU.0000000000001075
      OpenUrl
    37. ↵
      2. Liang J-T ,
      3. Bao Y-H ,
      4. Zhang H-Q , et al
      . Management and prognosis of symptomatic patients with intramedullary spinal cord cavernoma: clinical article. J Neurosurg Spine 2011;15:447–56.
      OpenUrlPubMed
    38. ↵
      2. Clatterbuck RE et al
      . Ultrastructural and immunocytochemical evidence that an incompetent blood-brain barrier is related to the pathophysiology of cavernous malformations. J Neurol Neurosurg Psychiatry 2001;71:188–92.doi:10.1136/jnnp.71.2.188
      OpenUrlAbstract/FREE Full Text
    39. ↵
      2. Rigamonti D ,
      3. Johnson PC ,
      4. Spetzler RF , et al
      . Cavernous malformations and capillary telangiectasia: a spectrum within a single pathological entity. Neurosurgery 1991;28:60–4.doi:10.1227/00006123-199101000-00010
      OpenUrlCrossRefPubMedWeb of Science
    40. ↵
      2. Furuya K ,
      3. Sasaki, T ,
      4. Suzuki I , et al
      . Intramedullary angiographically occult vascular malformations of the spinal cord. Neurosurgery 1996;39:1123–32. discussion 1131–2.doi:10.1097/00006123-199612000-00011
      OpenUrlCrossRefPubMed
    41. ↵
      2. Tong X ,
      3. Deng X ,
      4. Li H , et al
      . Clinical presentation and surgical outcome of intramedullary spinal cord cavernous malformations. J Neurosurg Spine 2012;74:308–14.doi:10.3171/2011.11.SPINE11536
      OpenUrl
    42. ↵
      2. Kharkar S ,
      3. Shuck J ,
      4. Conway J , et al
      . The natural history of conservatively managed symptomatic intramedullary spinal cord cavernomas. Neurosurgery 2007;60:865–72. discussion 865–72.doi:10.1227/01.NEU.0000255437.36742.15
      OpenUrlPubMed
    43. ↵
      2. Kim LJ ,
      3. Klopfenstein JD ,
      4. Zabramski JM , et al
      . Analysis of pain resolution after surgical resection of intramedullary spinal cord cavernous malformations. Neurosurgery 2006;58:106–11. discussion 106–11.doi:10.1227/01.NEU.0000192161.95893.D7
      OpenUrlPubMed

    Footnotes

    • AG and LR contributed equally.

    • Contributors AG, LR: conceptualisation and design, data collection, analysis and drafting of manuscript. RA: data collection. PK, MAA, KDF, LW: reviewing and revising original draft. FD: data collection, reviewing and revising original draft. MB: study supervision, reviewing and revising original draft.

    • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

    • Competing interests None declared.

    • Patient consent for publication Not required.

    • Ethics approval Institutional Review Board (IRB) approval was obtained (IRB #15-006838).

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Data sharing statement All available data provided in the manuscript and as supplement.