Cavernous malformations (cms) of central nervous system are rare vascular abnormalities with an incidence of 0.5% in the general population (1, 2) and the genetic basis of cms has been well established (3). Familial cm typically presents with multifocal cms and/or a family history, which is caused by a loss of function mutation in one of three genes, including ccm1 (krit1), ccm2 (mgc4607), and ccm3 (pdcd10). The functions of these genes are all involved in signaling networks that maintain the integrity of the connections between neighboring vascular endothelial cells. Identification of at-risk individuals with cm through pedigree and additional genetic testing can provide counseling for individuals and family members (3).

Most cms occur in the intracranial structures, and very few in the spinal cord. Cms of the spinal cord (sccm) are rarer, accounting for approximately 5%–12% of all spinal cord vascular anomalies (4–6). Compared to intracranial cms, sccms are more aggressive because the spinal canal is less tolerant to space-occupying lesions. The onset time of sccms occurs primarily in the third and fourth decades of life, with a slightly more prominent incidence in women, and the hemorrhage rates are approximately 1.4%–6.8% per lesion and per year (7–9). The preoperative diagnosis of sscm is easier to make due to its typical appearance on magnetic resonance imaging (mri) (10).

Current therapeutic strategies favor the early removal of symptomatic sscm under the microscope once a diagnosis is established. In terms of surgical outcomes, reports indicated that 66% of patients experienced improvement, 28% showed stabilization, and only 50% achieved long-term control or improvement in pain (11). However, for asymptomatic cases diagnosed by chance, conservative strategies or surgical treatment remain controversial, often depending on the preference of the neurosurgeons for this option. At present, several issues are still under debate, such as the choice of conservative treatment and surgical treatment, timing of surgery, surgical approach, and outcomes of surgical resection of sccms.

In this retrospective study, we have detailed the epidemiology, clinical characteristics, surgical modalities and long-term outcomes of 98 patients with sccm treated at our neurosurgical center. We have also analyzed possible predictors associated with better outcomes, with the aim of providing neurosurgeons with additional information when choosing an optimal treatment strategy.

To date, the available data for patients with SCCM are still insufficient due to the low incidence of SCCM (4, 14, 15). The current debate focuses on two points: the natural history of SCCM and conservative vs. surgical management (4, 16). In previous studies, numerous authors have described a slight predominance of females in the SCCM, at a ratio of 1:1.1 (17, 18), with the highest incidence commonly occurring in the third and fourth decades of life (19). The mean age of the SCCMs in our series is 41.6 years, which is close to the results of previous studies (14, 20). However, the rate for men and women was 1:0.56, which was different from previous studies. The reason might be that the samples in our study were all surgical patients, selection bias was a contributing factor for this difference. Early SCCMs are usually asymptomatic or only transiently symptomatic with minor numbness or pain, and are often easily overlooked by patients due to memory bias, whereas some patients only have incidental detection of SCCMs on a systemic examination, resulting in a bias in the duration of symptoms. Thus, it is difficult to investigate the natural history of the SCCM. SCCM haemorrhage is defined as a hematoma found during surgery, including cases that were not found haemorrhage on preoperative MRI. It is well recognized that the common clinical feature of intramedullary CM is slowly progressive neurological deterioration, which appears to be associated with several minor hemorrhages (21). Symptoms caused by a small amount of bleeding may be extremely mild and the patient does not feel a noticeable change in symptoms. Moreover, the haemorrhage rate is mainly based on surgical cases, ignoring bleeding from CMs in conservatively treated patients. As a result, the annual haemorrhage rate of SCCMs may be significantly higher than actually reported. A series of previous studies reported annual hemorrhage rates of 1.6%–4.5% (20, 22, 23), and our study had a bleeding rate of 6%, slightly higher than reported. Badhiwala et al. reported that approximately 61% of patients had motor impairment, 58% had sensory impairment, 34% had pain, and 24% had bladder and/or bowel impairment (14). Of our 98 cases, 36% of patients experienced motor dysfunction, 54% experienced sensory impairment, 47% experienced pain, and bladder and/or bowel disorders accounted for approximately 16%. The reason for the discrepancy between these results and the previous literature may still be related to the finite sample size. The same is true for family history and multiple lesions. In our series, eight patients (8%) had a family history and four patients (4%) had multiple lesions, while Badhiwala's meta-analysis showed that 12% had a family history (14), and 17 cases (34%) were combined with intracranial CMs as reported by Mitha et al. (24). One possible reason is that not all patients and family members are willing to undergo a full MRI examination.

For asymptomatic or mild symptomatic SCCMs, however, which are particularly located deep in the ventral or lateral portions of the spinal cord, surgery can result in serious complications such as paralysis, and the choice of conservative treatment or surgical treatment is still another controversial. Kharkar et al. (9) reported on 14 patients with symptomatic SCCMs managed conservatively, and 71% (n = 10) were clinically stable at a mean follow-up of 80 months. Although these findings establish the benefits of conservative treatment, the number of cases is too small and additional samples are needed to further investigate the benefits and risks of conservative treatment. Numerous conservatively treated patients have asymptomatic or mild symptoms, and such patients tend to be inactive during clinical follow-up because their symptoms have little impact on daily life, resulting in a substantial lack of follow-up data. In addition, some patients frequently actively seek surgical treatment once SCCM is diagnosed, and neurosurgeons have a preference for surgical treatment. It has been reported that the majority (90%) of patients with SCCMs underwent surgical resection, while only 10% received conservative treatment (14). For these reasons, the number of conservatively treated patients is limited and the corresponding studies are few. As a result, the lack of research on the effects of conservative treatment may lead to a bias against the benefits of surgical treatment.

For most patients with symptomatic SCCM, surgical removal remains an effective treatment to eliminate the associated lifetime risk of hemorrhage. Badhiwala et al. showed that symptoms improved in 51% of patients, remained the same in 38% and worsened in 11% during long-term follow-up (14). In our study, 83% patients (81/98) underwent long-term follow-up, which was within the acceptable range for data analysis. Of these, 42% showed improvement in symptoms, 51% remained stable and 7% deteriorated. This result compares unfavourably with previous data, but we found that in our study there was a higher proportion of patients with only pain and sensory disturbance, both of which were the most difficult to improve with surgery. The literature reports improvement rates of about 50% for postoperative pain and only 7% for sensory disturbances (25, 26). Bian et al. proposed that motor abnormalities had a more favorable clinical outcome than sensory symptoms (20). Park et al. showed that after complete resection, sensory deficits generally persisted during long-term follow-up (26). Our study also found that although there was no significant correlation between overall symptoms and prognosis in the multivariable regression analysis, there was a significant difference between the improved group and the unimproved group of sensory impairment (P < 0.05), indicating that sensory function was more difficult to recover than motor function. The literature suggests that the dorsal column is more susceptible to dorsal or dorsolateral myelotomy than the ventral or lateral motor fibers, which contributes to poor improvement in sensory deficits (26). Moreover, macrophages, which are abundant in hemosiderin in the post-surgery lesion, may additionally damage the sensory tract, causing the patient to suffer from paresthesia (26). Interestingly, in univariate analysis, we found that patients with severe symptoms (ASIA A-C) had a significantly better symptom improvement rate than patients with mild symptoms (ASIA D-E) (P < 0.001). Among the 14 patients with severe neurological dysfunction at follow-up, 12 patients (86%) showed better functional status, and two patients (14%) showed no improvement. The reason for this result may be related to the shorter duration of symptoms in patients with severe neurological disorders. We also found that patients with mild symptoms (Asia D-E) had a higher rate of symptom deterioration in the short postoperative period, but most of these patients recovered during long-term follow-up, only 6 patients had worsened, mainly manifested as numbness or increased pain without severe functional disability such as paralysis (Table 4). These data suggest that timely surgical resection could effectively improve the neurological status of patients with severe preoperative conditions. This is also consistent with the literature reports (25). These conclusions have important implications for treatment decisions. If the patient has only minor pain or sensory deficits that do not affect their quality of life, conservative treatment may be reasonable, as surgical outcomes for sensory deficits are unsatisfactory. However, if the sensory deficit is severe, aggressive surgical treatment should be recommended.

Based on long-term follow-up data, we analyzed the characteristics of the improved and unimproved groups. The data showed that the overall improvement rate was about 42% (34/81), which was close to the results of Badhiwala's systematic review (14). Reports suggest that the duration of symptoms is an important predictor of outcome. Cantore et al. qualitatively observed that patients with a longer clinical history had a worse prognosis (27). In recent years, a larger retrospective analysis has shown that patients with symptoms that last less than three months have higher rates of improvement than those with symptoms that last longer (28). A similar conclusion was found in our study, where the duration of symptoms was 4.6 ± 7.5 months of the improved group compared with 13.3 ± 14.3 months of the unimproved group (P < 0.0018), with a statistically significant difference. Based on these data, we do recommend that severe symptomatic SCCM should be treated surgically as early as possible, preferably within 3 months, which may lead to better clinical outcomes. Several studies have found that the severity of preoperative neurological injury is associated with poor outcomes (8). In contrast, patient age, sex, family history, multiple lesions, involved segments, and hemorrhage were not associated with prognosis (28).

We also found that the improvement rate of ventral or lateral deep SCCMs was lower than that of dorsal or superficial SCCMs, and the difference between the two groups was statistically significant (P = 0.011). The reason may be that ventral or lateral deep SCCMs are more difficult to remove, resulting in greater intra-operative damages to the spinal cord. Therefore, some authors have suggested that in the absence of symptoms or mild symptoms, close follow-up rather than surgical resection may be recommended for ventral or lateral deep lesions (18, 20) However, rigorous follow-up is necessary when choosing conservative treatments. They have also suggested considering a switch to surgical treatment when the MRI shows an increase in lesion size, especially if the surface of the spinal cord is raised in the posterior position (18, 20).

In recent years, with the development of microsurgical techniques and various positioning devices, some authors have become more and more in favor of resection of intramedullary lesions by hemilaminectomy (20, 29–33). We also analyzed the data of laminectomy and hemilaminectomy, and the rate of hemilaminectomy was 59% (48/81). There was no significant difference in symptom improvement between these two surgical procedures. This finding may be attributed to smaller iatrogenic spinal cord injury, milder postoperative reactions, and better spinal stability associated with minimally invasive microsurgery (30, 34–36).

The main limitation of this retrospective study is the inherent selection bias associated with the single-centre retrospective nature of the data and the relatively small sample size. This problem stems from the extreme scarcity of SCCM, limiting the ability to conduct prospective studies and/or randomized trials. Therefore, there is an inherent selection bias in this study that should be carefully considered when interpreting the results. In addition, all the patients have undergone surgery, however no data were available for conservatively treated patients, so future studies are needed to develop new therapeutic strategies to administer asymptomatic or mild symptomatic SCCMs.

For SCCM with severe symptoms, our results suggest that surgical therapy should be considered as the first choice. Early microsurgical resection might improve the prognosis of patients with neurological deficits. Since postoperative improvements in only pain and/or sensory deficits tend to be less satisfactory, conservative treatment also seems to be an potential option for these patients. Especially for CM located in the ventral or deep lateral spinal cord, surgical treatment should be more carefully considered. Hemilaminectomy and laminectomy are not prominent prognostic factors in our study. Therefore, minimally invasive hemilaminectomy may be preferred for this procedure in some situation.