Article Text

Original research
Clinical features and outcomes of perimedullary arteriovenous fistulas: comparison between micro- and macro-type lesions
  1. Jiaxing Yu1,2,
  2. Shiju Zhang1,2,
  3. Lisong Bian3,
  4. Chuan He1,2,
  5. Ming Ye1,2,
  6. Guilin Li1,2,
  7. Peng Hu1,2,
  8. Liyong Sun1,2,
  9. Feng Ling1,2,
  10. Hongqi Zhang1,2,
  11. Tao Hong1,2
  1. 1Department of Neurosurgery, Xuanwu Hospital Capital Medical University, Beijing, China
  2. 2China International Neuroscience Institute (China-INI), Beijing, China
  3. 3Beijing Haidian Hospital, Beijing, China
  1. Correspondence to Dr Tao Hong; 2030921{at}qq.com; Dr Hongqi Zhang; xwzhanghq{at}163.com

Abstract

Background Although the angioarchitecture of perimedullary arteriovenous fistulas (PMAVFs) is straightforward, their size and blood flow are highly heterogeneous. This study aimed to evaluate the differences in clinical features and outcomes of PMAVFs based on lesion size and blood flow.

Methods 114 consecutive patients with PMAVFs from two institutes were retrospectively reviewed. The lesions were classified as either micro-PMAVFs (shunt point diameter <1 cm) or macro-PMAVFs (shunt point diameter ≥1 cm).

Results The patients with micro-PMAVFs were older at the first presentation (33.50 vs 13.50 years, p<0.001). Macro-PMAVFs were more commonly associated with spinal metameric arteriovenous shunts (6.9% vs 28.6%, p=0.003). Compared with the macro-PMAVFs, the micro-PMAVFs showed a significantly higher risk of gradual clinical deterioration after initial onset (73.6%/year vs 10.0%/year; HR 3.888, 95% CI 1.802 to 8.390, p=0.001). A total of 58.6% of the micro-PMAVFs were treated surgically, whereas 85.7% of the macro-PMAVFs were treated via endovascular approaches. Complete obliteration was 73.7% for the whole cohort, and was more common for the micro-PMAVFs than for the macro-PMAVFs (87.9% vs 58.9%, p=0.001). At the last follow-up, spinal function was significantly improved compared with the pretreatment status, and the rate of severe disability of patients with macro-PMAVFs was slightly but not significantly higher than that of patients with micro-PMAVFs (16.1% vs 8.6%, p=0.315)

Conclusions The clinical risks, treatment strategies and obliteration rates of PMAVFs differ based on their size and blood flow.

  • Spinal cord
  • Hemorrhage
  • Fistula

Data availability statement

Data are available upon reasonable request.

Statistics from Altmetric.com

WHAT IS ALREADY KNOWN ON THIS TOPIC?

  • Perimedullary arteriovenous fistulas (PMAVFs) are arteriovenous shunts (AVSs) that are superficially located on the spinal cord. The angioarchitecture of PMAVFs is straightforward, however, their size and blood flow are highly heterogeneous.

WHAT THIS STUDY ADDS?

  • Due to the rarity of these diseases, the clinical discrepancies of PMAVFs with different size and morphological characteristics have not been well recognized. We have reviewed the clinical course of the largest PMAVF cohort with consecutive patients and divided them into micro-PMAVFs (shunt point diameter < 1cm) and macro-PMAVFs (shunt point diameter≥1cm). The results indicated that the clinical risks, treatment strategies and obliteration rates of PMAVFs significantly differ based on their size and blood flow.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE AND/OR POLICY?

  • The higher natural clinical risk of micro-PMAVFs promotes the necessity of early clinical intervention. The complete obliteration of macro-PMAVFs is more difficult than that of micro-PMAVFs, and residual lesions could cause further spinal cord impairment. Therefore, a regular long-term follow-up is necessary.

Introduction

Perimedullary arteriovenous fistulas (PMAVFs) are arteriovenous shunts (AVSs) that are superficially located on the spinal cord. They are supplied by primary pial branches of the anterior/posterior spinal cord arteries and drain directly into the spinal perimedullary venous system.1 2 A nidus structure is not found in these lesions.3 4

Although the angioarchitecture of PMAVFs is straightforward, their size and blood flow are highly heterogeneous. Both low-flow shunts with a single shunt point and moderately dilated drainage, and high-flow shunts with multiple shunt points and extremely dilated draining veins, have been reported in the literature.1 2 5 Previous clinical evidence has indicated that lesion size and blood flow may be related to the spectrum of pathophysiological mechanisms of PMAVFs (eg, venous congestion, bleeding or spinal cord compression), which can result in diverse clinical characteristics.2 3 6 Additionally, when developing treatment strategies for specific PMAVF patients, these parameters are also important considerations.4–8 Therefore, the optimal classification of PMAVFs should take the lesion size and blood flow into account.9–11 However, PMAVFs comprise only approximately 30% of all intradural spinal cord AVSs,12 and few studies have evaluated their clinical features and treatment outcomes.2 4 5 13 The clinical discrepancies of PMAVFs with different sizes and blood flow have not been well recognized.

In the present study, we reviewed the clinical course of a PMAVF cohort with consecutive patients from two institutes to report our treatment experiences and evaluate the effect of the lesion size and blood flow on the clinical features and outcomes of PMAVFs.

Methods

Patients

We reviewed consecutive patients with PMAVFs who were admitted to the Xuanwu Hospital and the Beijing Haidian Hospital between January 2007 and December 2018. The patients were included if they had been surgically or endovascularly managed at these two institutes. Patients who received their first invasive treatment (eg, microsurgery and endovascular embolization) at other institutes were not eligible for this study, and patients with concurrent tethered cords, spinal tumors or any other kind of conditions that impaired neurological function were also not included. Patients without long-term follow-up data or complete digital subtraction angiography (DSA) were excluded. The study was reviewed and approved by the local ethics committee, and the requirement for informed consent from the patients was waived given the study’s retrospective nature.

Radiological evaluation

Baseline clinical characteristics, including age of onset, sex, symptoms, treatment and follow-up data, were derived from the spinal vascular disease database of our institute. Angioarchitecture features were determined from Digital Imaging and Communications in Medicine (DICOM) DSA. Based on the previous classification of PMAVF,9–11 the lesions were divided into micro-PMAVFs and macro-PMAVFs by senior interventional neuroradiologists (figure 1). We defined micro-PMAVFs as lesions with a maximum shunt point diameter (including the venous pouch on the shunt point; if there are multiple shunt points, the diameters of all shunt points are added) of <1 cm, whereas macro-PMAVFs were defined as lesions with a maximum shunt point diameter of ≥1 cm.

Figure 1

Illustrations of both PMAVFs. One patient with micro-PMAVF (A–C) had weakness of both lower limbs and difficulty in urination and defecation for 4 months. Sagittal T2-weighted MRI (A) showed perimedullary flow voids (white arrow) signal and edema of the conus medullaris. DSA (B and C) showed that a micro-PMAVF (star) was supplied by PSA (black arrow) and drained via the perimedullary vein (double black arrows). The other patient had lumbosacral pain, urinary incontinence and weakness of both lower limbs for 1 month, and was then diagnosed as having macro-PMAVFs (D–I). Sagittal T2-weighted MRI (D) showed flow voids (white arrow) compressing the conus medullaris. DSA of ASA (E) and PSAs (F and G) showed a macro-PMAVF with multiple fistula points (stars) which were located at T10, T12 and L1. A dilated venous pouch was located at T12. DSA of the fourth left lumbar artery (H) and left internal iliac artery (I) showed two arteriovenous shunts of nerve root, which revealed the lesions were spinal metameric arteriovenous shunts. The black arrow indicates a feeding artery and double black arrows indicate draining veins. ASA, anterior spinal artery; DSA, digital subtraction angiography; PMAVF, perimedullary arteriovenous fistula; PSA, posterior spinal artery.

The lesion locations were categorized as cervical cord (from C1 to C7), thoracic cord (from T1 to T9), and thoracolumbar cord (from T10 to the tip of the conus medullaris). The spinal canal was equally divided into three longitudinal sections on the lateral view of the angiograph, based on which the relationship between the shunt point and the spinal cord is defined as ventral, lateral and dorsal, respectively. The aneurysm was characterized as a localized outward bulging of the vascular wall on the feeding artery, shunt point or drainage origination.14 15 Spinal metameric AVSs (SMAVSs) were defined as multiple AVSs, and non-spinal cord AVSs (eg, nerve root, dura, bone, muscles and skin) were located in the same metameric region as the spinal cord AVSs.4 16 17 Poor drainage was defined as main drainage vein stenosis of >50% or a ratio of the diameter of the main draining vein to the main feeding artery of <0.5. The treatment results were classified as complete or partial obliteration based on post-treatment DSA. Complete obliteration was defined as having a negative status on the angiogram.

Treatment

Transarterial embolization is a prioritized treatment at our institute.13 For this procedure, a femoral artery access was used to place a guiding catheter in the segmental artery and to inject embolic agent (n-butylcyanoacrylate (NBCA) or Glubran 2) using an appropriate microcatheter. The key step to achieve curative embolization is the occlusion of the draining veins.18 For extremely high-flow lesions with expanded drainage channels, coils could be used to reduce the blood flow and then create a turbulence to facilitate polymerization after the injection of the liquid embolic agent.5

When embolization is deemed technically unfeasible due to small or tortuous feeding arteries, microsurgical resection should be considered. All microsurgeries were performed through a dorsal approach. Motor-evoked potential and somatosensory-evoked potential monitoring were performed for all surgical patients. Since 2013, intraoperative DSA and methylene blue angiography have been used at our institute to facilitate the recognition of the angioarchitecture as well as residual AVSs during surgery.13

Clinical evaluation

The follow-up times at our institution were at discharge and at 1 month, 6 months, and yearly intervals through direct interviews or telephone contact, which were performed by clinicians who were not directly involved in the treatment of these patients.12 According to the modified Aminoff and Logue scale (mALS; online supplemental table 1),19 spinal function was divided into slight disability (mALS ≤2), moderate disability (3≤mALS≤5), and severe disability (6≤mALS≤11). The onset pattern was divided into acute and gradual. An acute onset was defined as an increase in the mALS of ≥1 point within 24 hours or a severe sudden spinal pain of ≥5 on the numerical rating scale (0–10). If the mALS score at the latest follow-up was stable or decreased compared with the pretreatment score, the prognosis was defined as favorable; otherwise, we defined it as a poor prognosis. An increase in the mALS of ≥1 point was defined as clinical deterioration. Clinical deterioration that occurred within 2 weeks after treatment and was sustained for more than 6 months or until death was defined as a permanent complication. Clinical deterioration that occurred beyond more than 2 weeks after treatment was defined as long-term deterioration.

In the analysis of clinical risk after onset, the inception was defined as the first onset, and the endpoint included clinical deterioration or invasive treatment (if no deterioration occurred). The clinical deterioration in this part of the analysis was further divided into acute and gradual, using the same definition for acute versus gradual onset patterns. When assessing the clinical risk after interventional treatment, the endpoint was defined as the first time of long-term clinical deterioration or the last follow-up (ie, no deterioration occurred), and the inception was determined as the latest treatment of that endpoint.

Statistical analysis

The differences between the groups were evaluated using the χ2 test for the categorical variables and the Wilcoxon rank sum test for the continuous variables. The multivariate models included variables that were significant at p≤0.1 in the univariate analysis for the outcome of interest. The annual risk of clinical deterioration was calculated as the number of patients with deterioration during the follow-up divided by person-years of follow-up. The risk of clinical deterioration was estimated using the Kaplan-Meier product-limit method, and the resulting curves were compared using the log-rank test. The Cox proportional hazards model was used to estimate the significance of several variables in predicting the relative risk (hazard ratio (HR)) of clinical deterioration. Statistical analyses were performed using SPSS software (version 25.0) and GraphPad Prism (version 7.0). All statistical tests were two-sided, and p values ≤0.05 were considered statistically significant.

Results

Baseline characteristics

A total of 114 patients with PMAVFs were included. The baseline characteristics are shown in table 1. Of the patients studied, micro-PMAVFs were found in 58 patients (50.9%), whereas macro-PMAVFs were found in 56 patients (49.1%). Both types of lesions showed a slight but non-significant male predominance. Patients with macro-PMAVFs were significantly younger at first presentation than patients with micro-PMAVFs (median age 13.50 vs 33.50 years, p<0.001). Compared with the patients with micro-PMAVFs, those with macro-PMAVFs were more likely to have SMAVSs (6.9% vs 28.6%, p=0.003), and aneurysms were also more common (50.0% vs 98.2%, p<0.001).

Table 1

Clinical features and outcomes

Symptoms

Acute versus gradual onset was observed in 61 (53.5%) versus 53 patients (46.5%) for the whole cohort. There was no statistically significant difference between patients with micro- and macro-PMAVFs (p=0.460). The most common symptoms in the patients with PMAVFs were dyskinesia (92 of 114 patients, 80.7%), followed by sensory disturbance (75 of 114 patients, 65.8%) and sphincter dysfunction (70 of 114 patients, 61.4%). Until the initiation of treatments, 42 patients (36.9%) had slight disability, 39 (34.2%) had moderate disability, and 33 (28.9%) had severe disability. Statistical analysis showed that the severity of spinal cord dysfunction was comparable between the two groups (p=0.290), but the rates of both dyskinesia and sphincter dysfunction of patients with PMAVF located on the thoracic and thoracolumbar cord were significantly higher than those with a cervical PMAVF (dyskinesia, 50.0% vs 88.0%, p<0.0001; sphincter dysfunction, 13.6% vs 72.8%, p<0.0001, respectively) (online supplemental table 2). In this cohort, all six patients with micro-PMAVFs located at C1-C3 segments showed subarachnoid hemorrhage manifestation and all had an AVF-related aneurysm.

Clinical risks after onset

During the observational period of 209.74 patient-years for the clinical risk analysis after onset, a total of 45 patients experienced a gradual clinical deterioration, yielding an annual gradual deterioration rate of 21.5%. Both log-rank and Cox multivariate analyses indicated that the patients with micro-PMAVFs had a significantly higher risk of gradual clinical deterioration than those with macro-PMAVFs (73.6%/year vs 10.0%/year; HR 3.888, 95% CI 1.802 to 8.390, p=0.001) (figure 2A, online supplemental table 3).

Figure 2

Kaplan-Meier curves demonstrated the micro-PMAVFs had a significantly higher risk of gradual clinical deterioration before clinical intervention (A), but the risk of acute clinical deterioration of the two types were similar (B). PMAVF, perimedullary arteriovenous fistula.

When acute clinical deterioration was used as the endpoint, a total of 20 patients suffered from acute deterioration during the observation period of 223.88 patient-years, which yielded an annual acute deterioration rate of 8.9%. The micro- and macro-type lesions showed comparable risks for the acute event (7.8%/year vs 9.4%/year, p=0.242) (figure 2B, online supplemental table 4).

Treatment

Thirty-seven patients (32.5%) underwent microsurgery, 66 (57.9%) underwent embolization, and 11 (9.6%) underwent combined treatment modalities. The clinical outcomes are shown in table 1. Patients with micro-PMAVFs more frequently received microsurgery (58.6% vs 25.0%, p<0.001), whereas patients with macro-PMAVFs more commonly received endovascular embolization (50.0% vs 85.7%, p<0.001). Complete obliteration was 73.7% for the whole cohort, 63.6% after endovascular embolization, 89.2% after microsurgery, and 81.8% after combined treatment modalities. The complete obliteration rate of micro-PMAVFs was significantly higher than that of the macro-type lesions (87.9% vs 58.9%, p=0.001), but multivariate logistic analysis indicated that spinal metameric AVSs were the only independent risk factor for incurable treatment (45.0% vs 79.8%; OR 4.985, 95% CI 1.692 to 14.684, p=0.004) (online supplemental table 5).

In this cohort, 11 of 114 patients (9.6%) experienced permanent complications, and there was no significant difference between the micro- and macro-type lesions (10.3% vs 8.9%, p=1.000). The permanent complication rate was 10.4% (5/48 patients) after microsurgery and 7.8% (6/77 patients) after embolization. Among the patients who suffered from embolization-related complications, three had intraoperative subarachnoid hemorrhage, one had intraoperative intramedullary hemorrhage because the venous pouch ruptured on the side close to the spinal cord (figure 3), and one had anterior spinal artery embolization by reflux of embolic agent; we could not determine the exact mechanism for the remaining patients.

Figure 3

One patient with macro-PMAVF experienced weakness of both lower limbs and difficulty in urination and defecation for 3 months (mALS=5). Sagittal T2-weighted MRI (A) showed perimedullary flow voids (white arrow) compressing the spinal cord at T10-T11 level. Anteroposterior view of left T11 intercostal artery angiogram (B) showed that a macro-PMAVF with a venous pouch (double arrows) was supplied by PSA (black arrow) and drained via the perimedullary vein. During the endovascular operation (C), the coil accidentally untwisted (arrow), and it could neither be further coiled nor withdrawn. Therefore, we had to push the remaining parts of coil into the segment artery (double arrows). Although the shunt point and terminal part of the feeding artery was almost intact, the post-operation DSA was negative (D). The patient suffered severe back pain 1 hour after the endovascular procedures; sagittal T2-weighted MRI (E) demonstrated thrombus formation in the venous pouch (double arrows), intramedullary hemorrhage (black arrow) and worsened spinal cord edema (white arrow). DSA, digital subtraction angiography; mALS, modified Aminoff and Logue scale; PMAVF, perimedullary arteriovenous fistula; PSA, posterior spinal artery.

Long-term clinical deterioration after partial obliteration

For the patients who received partial obliteration, seven experienced long-term clinical deterioration events (including two patients with subarachnoid hemorrhage, one with intramedullary hemorrhage and four with non-hemorrhagic, gradual deterioration) during the 115.07 patient-years of follow-up, yielding an overall annual long-term clinical deterioration rate of 6.1% (online supplemental table 6). After partial obliteration, the patients with micro-PMAVFs showed a higher deterioration risk than those with macro-type lesions (10.4%/year vs 4.6%/year), but the log-rank test revealed that the difference was not significant (p=0.284).

Clinical outcomes

The median follow-up period was 49.40 months (IQR 22.75–72.33 months). At the last follow-up, 70 patients (61.4%) had slight disability, 30 (26.3%) had moderate disability, and 14 (12.3%) had severe disability. Spinal function was significantly improved compared with the pretreatment status (p<0.0001; online supplemental table 7). The rate of severe disability of patients with macro-PMAVFs was slightly but not significantly higher than that of patients with micro-PMAVFs (16.1% vs 8.6%, p=0.315). A favorable prognosis was achieved in 106 patients (93.0%), and the rate was similar for patients with micro-PMAVFs and macro-PMAVFs (94.8% vs 91.1%, p=0.486).

Discussion

In this study, we reviewed the clinical data of the largest PMAVF cohort reported to date. The results showed that the clinical features, treatment, and clinical outcomes of PMAVFs differed based on the size and blood flow of the lesion.

The difference in clinical risk before treatment for the two types of PMAVFs was remarkable. Our findings confirmed that micro-PMAVFs were associated with a significantly higher risk of gradual clinical deterioration despite their small size. We believe this discrepancy was a result of the specific pathophysiological mechanisms of the two lesion types.2 3 6 20 The small size of the micro-type lesions could hardly cause a significant mass effect on nerve tissues. Thus, their most common pathophysiological mechanism should be venous congestion myelopathy.7 21 The significantly older onset age of those with micro-PMAVFs supports this hypothesis. Previous studies indicated that the radicular veins of spinal cord would be occluded with age due to fibrosis.7 22 Therefore, during the early ages, excess blood flow through PMAVFs could not cause neurological deficits. The sufficient outlets of the spinal cord venous system may provide adequate compensation until the number of radicular veins decreases with age.22 Most macro-PMAVFs were found in children or young adults; during this stage, the radicular veins are abundant.7 However, extremely dilated venous pouches can lead to symptoms of spinal cord or nerve root compression.2

Previous studies have shown that AVSs with venous congestion myelopathy, such as spinal dural arteriovenous fistulas and paravertebral AVSs with intradural drainage, are usually associated with fast progressive neurological deficits.20 23 24 The onset of symptoms related to spinal venous congestion may represent a decompensation of the spinal venous system, and persistent blood flow from the shunts would cause progressive spinal cord damage. Therefore, clinical intervention for micro-PMAVFs is urgently needed.

The complete obliteration rate of this cohort was favorable. Without a doubt, compared with nidus-type spinal cord AVSs, the treatment for PMAVFs is more facile given their simple angioarchitecture.11 25 A previous study showed that the complete obliteration rate of PMAVFs was significantly higher than that of nidus-type spinal cord AVSs.13 Even so, based on our findings, the size and blood flow also played a noticeable role in the treatment modalities and clinical outcomes. Similar to other institutions,3 4 26 we preferred microsurgery for micro-type lesions because their feeding arteries were usually too small and tortuous for microcatheters to get close enough to the shunt points. For the macro-type lesions, the microsurgical procedure was challenging due to the limited surgical field and the extremely engorged vessels, especially if the shunts were ventrally located. Therefore, endovascular embolization is the first line of treatment.

The key step in achieving complete embolization of an AVS is to obliterate the origination of draining veins.18 With the development of embolic agents and microcatheters, the goal of delivering embolic materials to the drainage vein becomes easier. However, for macro-PMAVFs with extremely dilated venous pouches, the venous pouch may only be partially embolized by coils to avoid mass effects. In this case, it should be noted that the terminal of the feeding artery should also be completely occluded. One patient in our cohort suffered from intramedullary hemorrhage due to a residual feeding artery after partial embolization of the venous pouch (figure 3).

Consistent with other institutes,1 27 we found that complete obliteration of macro-type lesions is more difficult than micro-type PMAVFs. For incurable macro-type lesions, our results indicated that partial treatment might delay the progression of the disease, and 87.0% of patients with residual macro-PMAVF (data not shown) obtained a favorable prognosis at the last follow-up. However, the follow-up data showed that residual shunts could still cause spinal cord damage, which might lead to a worse long-term prognosis. Therefore, we suggest annual MRI follow-up for patients with residual lesions. Additionally, efforts to achieve a cure through multiple endovascular treatments or combined surgical and endovascular procedures should be encouraged.

Limitation

The main limitations of the study are the retrospective design and the potential bias when considering that our cohort included referral patterns. The second limitation is that we could not accurately assess the hemorrhagic risk of PMAVFs in the current analysis. Most PMAVF hemorrhage events could be subarachnoid hemorrhages, which require timely examination for detection. Since our departments are referral centers that draw patients from the whole country, most of the hemorrhagic events of the cohort occurred before admission and lacked timely radiological examination. Therefore, for an objective evaluation, we decided to use clinical manifestations to assess the clinical features of the disease.

In conclusion, the clinical risks, treatment strategies and outcomes of PMAVFs differ based on their size and blood flow. The higher gradual deterioration risk of micro-PMAVFs promotes the necessity of early clinical intervention. Current treatment modalities, including microsurgery and embolization, can achieve favorable clinical outcomes. However, the complete obliteration of macro-PMAVFs is more difficult than that of micro-PMAVFs, and residual lesions could cause further spinal cord impairment, which may lead to a worse long-term prognosis.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Patient consent for publication

Ethics approval

This work was approved by our institutional ethics committee (Xuanwu Hospital, No.2016032).

Acknowledgments

We want to thank all the patients for their participation in this research.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • JY, SZ and LB are joint first authors.

  • HZ and TH contributed equally.

  • JY, SZ and LB contributed equally.

  • Contributors In addition to the guarantors of this work (Tao Hong and Hongqi Zhang), all authors were involved and made substantial contributions to the conception or design of the work, or the acquisition, analysis, or interpretation of the data; drafting the work or revising it critically for important intellectual content; final approval of the version published; and agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

  • Funding This work was supported by the National Natural Science Foundation of China (81971113, 81971104, 81671202, 82122020), Beijing Municipal Science and Technology Commission with grant D161100003816001, Beijing Municipal Education Commission with grant CIT&TCD201904095 and Beijing Municipal Administration of Hospitals with grant DFL20180801 and QML20190802.

  • Competing interests None declared.

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

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

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.