Multi-modality treatment approach for paediatric AVMs with quality-of-life outcome measures

Despite the potentially devastating and permanently disabling effects of paediatric arteriovenous malformations (pAVMs), there is a paucity of studies reporting long-term quality-of-life (QoL) outcomes in AVM patients. We aim to evaluate the management strategies for paediatric intracranial pAVMs in the UK and long-term QoL outcomes using a validated paediatric quality-of-life outcome measure. In this single-centre case-series, we retrospectively reviewed a prospectively maintained database of all paediatric patients (i.e. 0–18 years old) with intracranial AVMs, who were managed at Alder Hey Children’s Hospital from July 2007 to December 2021. We also collected the PedsQL 4.0 score for these patients as a measure of QoL. Fifty-two AVMs were included in our analysis. Forty (80%) were ruptured, 8 (16%) required emergency intervention, 17 (35%) required elective surgery, 15 (30%) underwent endovascular embolisation, and 15 (30%) patients underwent stereotactic radiosurgery. There was an 88% overall obliteration rate. Two (4%) pAVMs rebled, and there were no mortalities. Overall, the mean time from diagnosis to definitive treatment was 144 days (median 119; range 0–586). QoL outcomes were collected for 26 (51%) patients. Ruptured pAVM presentation was associated with worse QoL (p = 0.0008). Location impacted psychosocial scores significantly (71.4, 56.9, and 46.6 for right supratentorial, left supratentorial, and infratentorial, respectively; p = 0.04). This study shows a staged multi-modality treatment approach to pAVMs is safe and effective, with superior obliteration rates with surgery alone. QoL scores are impacted by AVM presentation and location regardless of treatment modality.


Introduction
Intracranial arteriovenous malformation (AVM) rupture is the leading cause of intracranial haemorrhage (ICH) in childhood [1]. Children comprise between 10 and 20% of all AVM presentations [2,3]. The resulting functional and cognitive deficits can impose a considerable socioeconomic burden on the patient (and their family) and adversely affect their long-term quality-of-life (QoL). AVM rupture is associated with a 30-50% risk of significant or permanent neurological deficit and a 5-10% risk to life [1].
Considering a 2-4% annual risk of intracranial AVM rupture, which increases to 4-6% in the period immediately following AVM rupture, paediatric patients have a significant cumulative lifetime risk of ICH [1]. Thus, active treatment of childhood AVMs is advocated for suitable AVMs to prevent the occurrence of a future disabling or fatal haemorrhage although consensus guidelines do not exist [1]. Complete microsurgical excision is considered the gold standard definitive treatment for AVMs, and correlates to higher rates of complete obliteration on digital subtraction angiography (DSA) and lower rates of recurrence [4]. The length of follow-up required for treated paediatric AVMs (pAVMs) is widely debated due to significantly varying rates of AVM recurrence reported between studies (0.6-13%) [4]. However, higher rates of AVM recurrence are noted in the paediatric population compared to adults (9.5% vs 2.7%) [5].
Several pAVM case-series studies report often neurological outcomes following AVM treatment using a variety of neurological, functional, and cognitive scales, which are not validated in the paediatric population. Moreover, despite the potentially devastating and permanently disabling effects of childhood AVMs, there is a paucity of studies reporting long-term QoL outcomes in AVM patients.
The primary aim of this single-centre retrospective caseseries is to evaluate the management strategies for paediatric intracranial AVMs at our institution. Our secondary aim is to measure long-term QoL outcomes using a specific paediatric quality-of-life outcome measure.

Study methodology
In this single-centre case-series, we retrospectively reviewed all paediatric patients with intracranial AVMs, who were managed at our institution from July 2007 to December 2021. We identified all cases from a prospectively maintained operative database in conjunction with clinic and angiography lists.

Inclusion and exclusion
Patients aged between 0 and 18 years old with a diagnosis of an intracranial AVM confirmed on imaging (i.e. magnetic resonance angiography [MRA] or digital subtraction angiography [DSA]) were included in this study. Multiple AVMs in one patient were treated as separate entities. Patients ≥ 18 years old at diagnosis, those with an arteriovenous fistula, vein of Galen malformation, and spinal AVM without an intracranial AVM were excluded from this study.

Patient demographics
Demographic data including sex and age were recorded. We recorded the following clinical data: main presenting symptom (i.e. haemorrhage, seizure, headache, incidental or other); significant medical history (i.e. a diagnosis of HHT, associated flow aneurysm, concomitant AV fistula, or multiple AVMs); Spetzler-Martin grade (SMG) of the AVM [6]; AVM size, location, drainage; pre-treatment Glasgow Coma Score (GCS); time from diagnosis to treatment [7]; type of treatment; post-operative complications; and ongoing problems on follow-up.

Treatment
Definitive AVM treatment was defined as either surgical AVM excision, endovascular embolisation (EE), SRS, any combination of these modalities, or conservative management. Emergency evacuation of haematoma, aneurysm coiling (without AVM embolisation attempt), insertion of external ventricular drain (EVD), insertion of ventriculoperitoneal shunt (VPS), and insertion of intracranial pressure monitor (ICPM) were classified as adjuvant procedures.

Outcome measures
The primary outcome measures were complete obliteration confirmed on imaging (i.e. MRA or DSA) and AVM-related mortality. Secondary outcome measures were ICH secondary to AVM rupture whilst awaiting treatment, and AVM recurrence following complete obliteration seen on imaging. The QoL outcome measure utilised was the PedsQL 4.0 SF15, a valid, standardised, generic, self-reporting assessment tool to measure health-related QoL in children and adolescents on a scale from 0 to 100. Results are expressed as a total summary score, as well as physical and psychosocial health summary scores [8].

Results
We identified 59 patients with AVMs managed at Alder Hey Children's Hospital between June 2007 to March 2021. Eight patients were excluded from the analysis due to insufficient data. Thus, 51 patients with 52 AVMs were included in our analysis.
Patient demographics are summarised in Table 1. Our cohort comprised of 30 (59%) males and 21 (41%) females with a mean age of 10 years old (range 0-16). A significant associated medical history was noted in 6 (11%) patients; these included 3 (6%) concomitant AV fistulas, and 3 (6%) patients with a history of HHT. Notable AVM features were as follows: 4 (8%) patients had multiple AVMs and 10 (19%) patients had associated flow aneurysms. One patient with multiple AVMs had sufficiently detailed records for both AVMs to be analysed as separate entities.

Emergency surgical treatment
Eight (16%) patients required emergency intervention, which was defined as definitive intervention within < 48 h of AVM diagnosis (see Table 2). These patients presented with a reduced GCS, requiring an emergency craniotomy and haematoma evacuation with simultaneous AVM excision. One (12.5%) patient underwent an intra-operative angiogram. All 8 AVMs were completely obliterated and remained obliterated on followup. The mean follow-up period was 3.7 years (range 2.2-6.9), and 2 (25%) patients were still having ongoing follow-up.

Elective surgical treatment
Elective treatment was defined as definitive AVM treatment > 48 h following initial diagnosis. Seventeen (35%) AVMs were electively surgically excised (see Table 3). Fourteen (82%) AVMs were ruptured on presentation, and the remaining 3 (18%) AVMs were unruptured. One (6%) of these AVMs ruptured a day before scheduled SRS treatment; hence, this patient was treated as an emergency in another hospital. The SMG was known for 13 patients; a median of SMG of 3 (range 2-4) was found in this group.
Surgery was the last step in multimodal management for 2 AVMs. In 3 cases, elective surgery was unsuccessful; thus, these AVMs underwent further SRS (n = 2; 4%) or EE (n = 1; 2%). Complete obliteration was achieved in 14 (82%) elective surgery patients, while a further 2 (12%) incompletely obliterated AVMs were successfully obliterated following further SRS or EE. One (2%) residual AVM remained. There was no AVM recurrence in a mean follow-up period of 2.5 years (range 0.2-7.0). An early rebleed occurred prior to treatment.

Primary EE
Fifteen children underwent primary EE as either standalone treatment or as part of a multistage and/or multimodal AVM treatment strategy (see Table 4). Twelve (71%) were ruptured on presentation. The SMG was known for 13 patients; a median SMG of 3 (range 1 4) was found in this group. The mean follow-up was 3 years (range 0.1-13). Complete obliteration was achieved in 11 (65%) of this cohort, although most required multistage or multimodal treatment.

SRS
Fifteen (30%) patients underwent SRS as either a standalone treatment or as part of a multistage and/or multimodal AVM treatment strategy (see Table 5). Overall SRS was used 19 times in this cohort. Ten (67%) were ruptured on presentation. The SMG was known for 13 patients; a median SMG  Table 6 demonstrates the treatment sequence and overall obliteration rates for all combinations of AVM management in our patients. Twenty-five (50%) of our patients underwent either elective or acute surgical excision of their AVM, of which 22 (88%) were successful (i.e. complete obliteration). In 2 (4%) cases, surgery was the first stage of multimodal management, and in another 3 (6%) cases, surgery was the last step in multimodal management. Of these, there were 2 (4%) AVMs, which had failed previous SRS and EE, and only 1 of these was amenable to surgery. The remaining AVM rebled shortly after initial presentation and was treated with partial endovascular coiling of the aneurysm segment, followed by one dose of SRS, and subsequently underwent surgical intervention with partial AVM excision, followed by another dose of SRS. This patient is having ongoing followup and awaiting confirmation of complete obliteration following most recent dose of SRS; he is approximately 2 years post-SRS follow-up.
Procedure-related complications were as follows: 1 (4%) instance of osteomyelitis of a bone flap following a craniotomy for surgical AVM excision. SRS was associated with 1 (5%) new cyst development, 1 (5%) case of radionecrosis requiring surgical excision, and 1 (5%) case of delayed oedema causing seizures. EE was associated with 1 case of sagittal sinus thrombosis and new mild neurological deficit, prolonging the patient's hospital stay.
Two (4%) AVMs ruptured whilst awaiting treatment; 1 rebleed occurred shortly after presentation. The other occurred in an untreatable AVM causing 1 (2%) mortality. There were no instances of AVM recurrence.

Timing of treatment
Overall, the mean time from diagnosis to definitive treatment was 144 days (median 119; range 0-586). For each treatment modality subgroup, the mean time from diagnosis to definitive treatment were as follows: 123 days (median; range 7-586) for elective surgery, 135 days (median; range 21-327) for EE, and 291 days (75-586) for SRS.

Time to treatment
Patients must be referred to an external centre for SRS treatment at our institution, which probably delays definitive

Obliteration
Pezeshkpour et al. recently published a meta-analysis showing that AVMobliteration rates vary greatly between treatment strategies, centres, andcountries [12]. Overall obliteration rates of the pooled datawere 69.8% with a 95% confidence interval of 62.9-75.9% [12]. Our study had an 88% overall obliterationrate. Other UK centres with similar pAVM cohorts report 39.5-83.0% completeobliteration rates [10,13]. We noted thatShtaya et al. who favoured surgical excision reported a much higherobliteration rate compared to Bal et al. who favoured EE [13]. At ourinstitution, patients with surgical excision as part of their treatmentstrategy had a significantly higher rate of AVM obliteration (96%; p = 0.001)compared to patients undergoing EE and SRS alone (65% and 47%, respectively).Nevertheless, it is important to highlight that EE is rarely a stand-alonetreatment for paediatric AVMs and is usually an indispensable adjuvant tosurgery and/or SRS, particularly in treatment strategies for higher grade AVMin keeping with our study findings [14]. Emergency AVM excision and electiveAVM excision did not  change the percentage of AVM obliteration (100% and 85%) between the cohorts, suggesting that AVM characteristics and treatment modalityas opposed to urgency of treatment influence obliteration rates.

Rebleed and recurrence
There were 2 (4%) rebleeds in our series, which may have been related togreater AVM complexity; the bleed occurring shortly after presentation requiredall three treatment modalities, and the other was inoperable. Additionally, theliterature suggests AVM presentation, treatment timing, and modality and influencesthe rebleed risk.Gross et al. found significantly lower post-operativemorbidity in surgically treated haemorrhagic AVM compared to unruptured AVMs [1]. An overall annual rebleed rate of 2.71 ± 1.32%was reported by Blauwblomme et al., although there was a high rate ofrebleeding of 3.88 ± 1.39% in the first year following presentation [15]. Similarly, Darsautet al. reported a 4% rebleed rate occurring prior to first treatment [14]. Pezeshkpour  During a mean follow-up period of 3.3 years, there were no instances ofAVM recurrence in our patients, likely related to our high rate of completeobliteration. AVM recurrence varies between centres (0-13%) and can occur aslate as at 6-9 years, raising questions around adequacy of resection,post-operative imaging, and appropriate follow-up (i.e. duration and strategy) [5,18]. Jimenez et al.showed that a significantly higher percentage of patients not followed up withappropriate imaging presented with further rupture of a recurrent or residualAVM, causing further potentially preventable neurological deficit (13% vs 57%) [19]. Our centre's practise is to perform a DSA 5 yearsafter final treatment to confirm complete obliteration with no recurrence.Should this occur before the patient's 16th birthday, we obtain an additionalDSA at 16 years old.

Patient reported outcomes
Most pAVM studies report neurological outcomes though few studies report QoL outcomes [20]. PedsQL (unlike mRS) is a reliable tool for use in children [8,21]. Furthermore, it has been validated in paediatric brain tumours, ischaemic stroke, and traumatic brain injury [8,21]. Most papers report the functional (i.e. neurological) outcome in children using the mRS (validated for use in adults). There are a few studies reporting educational outcomes in pAVMs patients. Van Essen et al. found no difference in educational outcomes between children with brain AVMs and the general population [22]. Contrary to our series, Abecassis et al. found no difference in PedsQL score between ruptured and unruptured presentations; this was probably due to the small (n = 26) cohort size, as their result graph shows a clustering of higher PedsQL scores for the haemorrhagic presentation [21]. In keeping with our results, Abecassis et al. reported that parietal AVMs had the least negative impact on QoL, while infratentorial AVMs had the most negative impact on QoL (not statistically significant) [21]. Nevertheless, our cohort showed a significantly worse psychosocial score for infratentorial location versus right-sided supratentorial AVMs (p = 0.04). There may also be an association between worse scores for left-sided AVMs compared to right-sided AVMs, though not statistically significant (p = 0.28). Additionally, Graph 1 supports the role of SMG in providing insights into future QoL. Given that most people are left hemisphere dominant, a reduction in psychosocial scores in this group may be expected. However, our findings in the cerebellar group are interesting and add to the growing body of knowledge around the role of the cerebellum in cognition and emotion. Traditionally, the cerebellum is understood for its role in motor planning and coordination; however, these results indicate that the cerebellum also plays a large part in emotional control and social skills, as described by Bodranghein et al. [23]. The total PedsQL score also correlated with the location of eloquent brain (p = 0.2), although this was not statistically significant. Nevertheless, it is established that the posterior fossa is a much smaller and tighter space than the supratentorial brain; hence, there is little room to compensate for a mass (e.g. large AVM or haematoma).

Limitations and future direction
Our findings are limited by retrospective data collection at a single centre. Prospective data collection from a national UK database of pAVM patients would significantly improve reliability of the results, aid in the development of national pAVM management guidance, and enable comparison of pre-and post-treatment function and accurate correlation of neurological outcomes with QoL outcomes. Furthermore, increased reporting of long-term QoL following AVM treatment using valid outcome measures will inform AVM patient and parent counselling with regards to long-term expectations for the varying treatment modalities in the context of individual AVM characteristics .

Conclusion
This study shows a staged multi-modality treatment approach to pAVMs is safe and effective, with superior obliteration rates with surgery alone. There was a low re-bleed rate in the group (n = 2; 4%). QoL scores are impacted by AVM presentation (mean ruptured 59.0, mean unruptured 81.6; p = 0.0008) and location (mean psycho-social scores of 71.4, 56.9, and 46.6 for right supratentorial, left supratentorial, and infratentorial, respectively) regardless of treatment modality.
Author contribution Natasha Aziz: data collection, analysis, wrote paper; John C. Duddy: data collection, analysis, wrote and edited parts of paper; Daniel S. and Dawn H.: data collection, reviewed and edit paper; Anil Israni, Mani Puthuran, Arun Chandran: reviewed and edited paper; Conor Mallucci: study supervisor.