Title: Prognosis of cerebral cavernomas: on to treatment decisions
Abstract: In patients with non-traumatic intracerebral haemorrhage (ICH) establishment of whether the cause is a macrovascular lesion is important, because this might be treatable. In a study of 298 patients with ICH who were younger than 70 years,1van Asch CJ Velthuis BK Rinkel GJ et al.on behalf of the DIAGRAM InvestigatorsDiagnostic yield and accuracy of CT angiography, MR angiography, and digital subtraction angiography for detection of macrovascular causes of intracerebral haemorrhage: prospective, multicentre cohort study.BMJ. 2015; 351: h5762Crossref PubMed Scopus (54) Google Scholar a standardised diagnostic work-up of CT angiography within 48 h of onset, followed by MRI, magnetic resonance angiography, or digital subtraction angiography within 4–8 weeks identified 69 macrovascular lesions, ten of which were cerebral cavernous malformations (CCMs). The advent of MRI has enabled easy detection of CCMs, and nowadays a quarter of detected vascular malformations are CCMs.2Al-Shahi Salman R Bhattacharya JJ Currie DG et al.Prospective, population-based detection of intracranial vascular malformations in adults: the Scottish Intracranial Vascular Malformation Study (SIVMS).Stroke. 2003; 34: 1163-1169Crossref PubMed Scopus (300) Google Scholar However, whether risks of treatment of CCMs are counterbalanced by elimination of the risk of new or recurrent ICH depends on the risk of ICH and its clinical consequences. In a meta-analysis of individual patient data in The Lancet Neurology, Margaret Horne and colleagues3Horne MA Flemming KD Su I-C et al.for the Cerebral Cavernous Malformations Individual Patient Data Meta-analysis CollaboratorsClinical course of untreated cerebral cavernous malformations: a meta-analysis of individual patient data.Lancet Neurol. 2015; (published online Dec 1.)http://dx.doi.org/10.1016/S1474-4422(15)00303-8PubMed Google Scholar calculated the risks of symptomatic ICH for patients with an untreated CCM. In their comprehensive literature search, the investigators identified 22 publications that included 2957 patients. Data from 1337 patients (45%) could not be obtained, showing the challenges of undertaking individual patient data analyses. Therefore, the analysis was based on data from 1620 patients. During a median follow-up of 3·5 years, 204 patients had a symptomatic ICH, yielding a 5-year risk of 15·8% (95% CI 13·7–17·9). Location of CCM (brainstem vs other) and mode of presentation (ICH or focal neurological deficit [FND] vs other) were independently associated with ICH risk during follow-up (table). In 640 (40%) of 1620 patients, data were also available on the combined outcome of ICH or FND; findings were similar, but were less precise. Age, sex, and CCM multiplicity did not add independent prognostic information.TableKaplan-Meier estimates of 5-year risksBrainstem locationOther locationAll locationsICH (1620 patients at risk)ICH or FND presentation30·8%18·4%26·4%Other presentation8·0%3·8%4·3%All presentations27·7%8·2%15·8%ICH or FND (640 patients at risk)ICH or FND presentation51%22%35%Other presentation23%4%5%All presentations45%9%17%FND=focal neurological deficit. ICH=intracerebral haemorrhage. Open table in a new tab FND=focal neurological deficit. ICH=intracerebral haemorrhage. We commend the investigators for their ability to assemble data from seven cohorts, including two from Asia, without any missing data. 1159 (72%) of the patients presented clinically (ICH, FND, or seizure); in the other 461 (28%) patients the CCM was an incidental finding. Risk of bias of individual studies was low—one of the items checked when the investigators applied the PRISMA individual participant data statement.4Stewart LA Clarke M Rovers M et al.Preferred Reporting Items for Systematic Review and Meta-Analyses of individual participant data: the PRISMA-IPD Statement.JAMA. 2015; 313: 1657-1665Crossref PubMed Scopus (1083) Google Scholar The size of the dataset with 204 primary outcomes allowed robust estimation of the joint prognostic information of five preselected characteristics. Horne and colleagues3Horne MA Flemming KD Su I-C et al.for the Cerebral Cavernous Malformations Individual Patient Data Meta-analysis CollaboratorsClinical course of untreated cerebral cavernous malformations: a meta-analysis of individual patient data.Lancet Neurol. 2015; (published online Dec 1.)http://dx.doi.org/10.1016/S1474-4422(15)00303-8PubMed Google Scholar chose the five potential predictors well on the basis of previous publications and practicability, which is in accordance with recommendations for modern prognostic research.5Royston P Moons KG Altman DG Vergouwe Y Prognosis and prognostic research: developing a prognostic model.BMJ. 2009; 338: b604Crossref PubMed Scopus (769) Google Scholar This careful selection of predictors enhances applicability in clinical practice, in part because sophisticated laboratory tests are therefore not obligatory to allow prediction of risk. External validation of the model was not possible because all worldwide available data were already used for derivation—a common problem with prognostic modelling for rare diseases. The absolute risks of symptomatic ICH or focal deficits can now be compared with the risks of death, non-fatal ICH, or new or worse persistent FND after treatment of CCMs. These risks have been reviewed, and seem favourable compared with risks of new or recurrent ICH.6Poorthuis MHF Klijn CJM Algra A Rinkel GJE Al-Shahi Salman R Treatment of cerebral cavernous malformations: a systematic review and meta-regression analysis.J Neurol Neurosurg Psychiatry. 2014; 85: 1319-1323Crossref PubMed Scopus (43) Google Scholar The next challenge is to build a decision analytical model that compares risks of death, ICH, and FND with and without treatment to guide clinicians treating a patient with a CCM. Such a model would start with a risk prediction model similar to that developed for patients with unruptured intracranial aneurysms.7Greving JP Rinkel GJE Buskens E Algra A Cost-effectiveness of preventive treatment of intracranial aneurysms. New data and uncertainties.Neurology. 2009; 73: 258-265Crossref PubMed Scopus (62) Google Scholar That analysis identified uncertainties that needed further study and triggered the development of the PHASES score,8Greving JP Wermer MJH Brown Jr, RD et al.Development of the PHASES score for prediction of risk of rupture of intracranial aneurysms: a pooled analysis of six prospective cohort studies.Lancet Neurol. 2014; 13: 59-66Summary Full Text Full Text PDF PubMed Scopus (760) Google Scholar which was also based on a meta-analysis of individual patient data. A next step could be to combine the risk prediction of ICH with a new meta-analysis of individual patient data and prediction model to assess risks of treatment according to type of treatment, size and site of the CCM, and method of presentation of the patient. The decision model should also take into account the time dependency of the risks of ICH and FND if the CCM is left untreated, since Horne and colleagues3Horne MA Flemming KD Su I-C et al.for the Cerebral Cavernous Malformations Individual Patient Data Meta-analysis CollaboratorsClinical course of untreated cerebral cavernous malformations: a meta-analysis of individual patient data.Lancet Neurol. 2015; (published online Dec 1.)http://dx.doi.org/10.1016/S1474-4422(15)00303-8PubMed Google Scholar found a 3·5-times decreased risk of ICH from the first to the fifth year of follow-up in patients presenting with ICH or FND. Finally, for the decision to treat, the utility of awareness of an untreated incidental CCM should be assessed as another element. A modelling approach for patients with CCMs is important because randomised trials are not likely to be feasible in view of the low prevalence of CCM and the potentially strong a-priori beliefs on treatments, which might hinder the enrolment of sufficiently large numbers of patients, even in an international collaboration. This issue, allegedly, was one of the problems with the ARUBA trial,9Mohr JP Parides MK Stapf C et al.Medical management with or without interventional therapy for unruptured brain arteriovenous malformations (ARUBA): a multicentre, non-blinded, randomised trial.Lancet. 2014; 383: 614-621Summary Full Text Full Text PDF PubMed Scopus (797) Google Scholar, 10Stapf C Parides MK Moskowitz AJ Mohr JP Management of brain arteriovenous malformations—authors' reply.Lancet. 2014; 383: 1635-1636Summary Full Text Full Text PDF PubMed Scopus (7) Google Scholar which assessed medical management with or without interventional treatment for unruptured brain arteriovenous malformations—a disorder with some similarities with CCMs. Hence, meta-analysis of individual patient data from properly designed cohort studies of the risks of neurosurgery and stereotactic radiosurgery are warranted to provide elements for an appropriate decision model. We declare no competing interests. Clinical course of untreated cerebral cavernous malformations: a meta-analysis of individual patient dataMode of clinical presentation and CCM location are independently associated with ICH within 5 years of CCM diagnosis. These findings can inform decisions about CCM treatment. Full-Text PDF Open Access