Background Endovascular coiling (EVC) has been shown to yield superior clinical outcomes to surgical clipping (SC) in the treatment of ruptured cerebral aneurysms. The reasons for these differences remain obscure. We aimed to assess outcomes of EVC and SC relative to baseline physiological derangement.
Methods This was an exploratory analysis of prospectively collected trial data. Physiological derangement was assessed using the Acute Physiology and Chronic Health Evaluation II (APACHE II) scoring system. Other contributory variables such as age, World Federation of Neurosurgical Societies (WFNS) grade, and development of complications, including hydrocephalus and vasospasm, were included in the analysis. Clinical outcome was independently assessed at 90 days using the modified Rankin Scale (mRS). Hospital stay, ventilated days, and total norepinephrine dose were also used as secondary outcomes. Multivariate analysis was performed using binary logistic regression.
Results EVC was performed in 69 patients and SC in 66 patients. More profound physiological derangement (APACHE II score >15) was the strongest predictor of poor outcome in the overall cohort (OR 17.80, 95% CI 4.78 to 66.21, p<0.0001). For those with more deranged physiology (APACHE II score>15; 59 patients), WFNS grade ≥4 (OR 6.74, 1.43 to 31.75) and SC (OR 6.33, 1.27 to 31.38) were significant predictors of poor outcome (p<0.05). Favorable outcome (mRS 0–2) was seen in 11% of SC patients compared with 38% of EVC patients in this subgroup. SC patients had significantly increased total norepinephrine dose, ventilated days, and hospital stay (p<0.05).
Conclusions More profound physiological derangement at baseline is a strong predictor of eventual poor outcome, and outcomes for patients with more profound baseline physiological derangement may be improved if undergoing a coiling procedure.
Statistics from Altmetric.com
The International Subarachnoid Aneurysm Trial (ISAT) demonstrated an absolute 6.9% reduction in the rate of death or dependency at 1 year for patients treated with endovascular coiling (EVC) following aneurysmal subarachnoid hemorrhage (SAH).1 Subsequent systematic reviews2 ,3 of this and other prospective controlled studies have also concluded that EVC yields better clinical outcomes compared with surgical clipping (SC). The reasons for this difference have been debated in the literature and include rates and severity of cerebral vasospasm and delayed ischemia,4 ,5 degree of procedural brain tissue damage,6 and the rate of medical complications.7 Physiological derangement and medical comorbidities are also recognized as risk factors for poor outcome following aneurysmal SAH.8–11 However, assigning patients to either SC or EVC has traditionally been based on anatomical rather than physiological criteria, and there is little evidence regarding the interaction of treatment modality and presenting physiological derangement.
The Acute Physiology and Chronic Health Evaluation II (APACHE II) scoring system is widely used in many intensive care units and allows approximate prognostication based on a score from 0 to 71 for clinical and biochemical markers of cardiopulmonary, renal, hematological, and neurological function.12 Using this system, we aimed to investigate the difference in outcomes for EVC and SC patients relative to the degree of physiological derangement at baseline.
Study design and inclusion/exclusion criteria
This was an exploratory analysis of prospectively acquired clinical trial data obtained primarily for the assessment of clinical outcomes, and vasospasm severity and incidence, in patients treated with magnesium following aneurysmal SAH.13 The study was approved by the Northern Sydney and Central Coast Ethics Committee and written informed consent was obtained from each patient or legal surrogate.
Patients admitted between April 1, 2005 and February 1, 2010, presenting within 72 h of SAH confirmed by CT were included. Exclusion criteria included age <18 years, serum creatinine >200 µmol/L, if death was thought imminent within 72 h, if the patient had myasthenia gravis or was pregnant, or if vasospasm was present prior to inclusion in the study. The original study included 166 patients. This explorative analysis included only patients treated in the acute period with EVC or SC. Those patients who were angiographically negative were not included.
All patients were managed on the neurointensive care unit with routine insertion of arterial and central lines. After admission, digital subtraction angiography or CT angiography was performed, and the aneurysm responsible for the bleeding was occluded by operative or endovascular means using conventional techniques within 48 h. Decision to clip or coil was not necessarily based on aneurysm morphology but was dependent on the philosophy of the admitting neurosurgeon, some advocating aggressive surgical management with others regularly referring for EVC. At one of the centers participating in the study, there was limited endovascular availability. During the study period, young patients, those with multiple aneurysms accessible via a single surgical approach, or those with a hematoma requiring evacuation were primarily clipped.
Following clipping or coiling of the aneurysm, systolic blood pressure was maintained between 120 and 150 mm Hg. A nimodipine infusion of 20 µg/kg/h was commenced on arrival in the intensive care unit. All patients received nimodipine for 21 days, with at least 10 days of intravenous administration. Body temperature was aimed to be maintained below 37.5°C. Oxygen saturation was monitored continuously by pulse oximetry and maintained at >95%. Serum electrolytes were maintained within normal limits, and glucose levels were maintained at 6–10 mmol/L. Angiographic screening for vasospasm was performed routinely at days 5–7 or at an earlier time point if vasospasm was suspected. Patients were managed with adequate hydration, systolic blood pressure maintained between 140 and 160 mm Hg, and continued nimodipine infusion. The majority of patients with moderate–severe vasospasm (>25% arterial narrowing) were managed with a preemptive endovascular approach previously documented.14
Data collected and analysis
Baseline variables were collected, including demographic data, history of smoking, hypertension, use of statins, use of trial magnesium, previous SAH or stroke Glasgow Coma Scale (GCS) score, World Federation of Neurosurgical Societies (WFNS) grade, and APACHE II score. Aneurysm site, size, and presence of hydrocephalus were recorded through assessment of angiographic and CT imaging, respectively. Modified Rankin Scale (mRS) scores were obtained at 90 days by independent assessors. Favorable outcome was defined as mRS 0–2. Secondary outcome measures, including norepinephrine dose (mg), length of stay in the intensive care unit, ventilator time, and length of hospital stay were also assessed.
Statistical analysis was performed using Openstat statistical software. Non-parametric data (sex, WFNS grade, history of smoking, hypertension, SAH, stroke, magnesium or statin use, hydrocephalus, aneurysm location, and aneurysm securing modality) were compared using Fisher's exact test, and parametric data (age, length of stay in the intensive care unit (days), ventilation duration (days), hospital stay (days), norepinephrine dose) were compared using analysis of variance. Ordinal data (APACHE II score and GCS) were compared using the Mann–Whitney U test. Binary logistic regression was used for analysis of multiple variables, including age, gender, smoking history, vasospasm, hydrocephalus, aneurysm securing technique, APACHE II score, and WFNS grade with respect to patient outcome. Statistical significance was defined as p<0.05.
A total of 135 patients (69 EVC and 66 SC) were included in the analysis. Two patients who underwent SC were lost to follow-up and were therefore not included. Descriptive statistics are shown in table 1. There tended to be more patients with a prior history of previous stroke, statin use, and smoking among patients in the EVC group but this was not statistically significant. The magnitude and distribution of APACHE II and GCS scores was very similar between the EVC and SC groups (see figure 1). The incidence of moderate–severe and severe vasospasm was similar although there tended to be more endovascular interventions per treated patient in the SC group. Patients in the SC group as a whole received a significantly increased total norepinephrine dose and had a significantly longer hospital stay (p<0.05) (see table 2).
When APACHE II score was plotted against outcome for all patients (figure 2), a threshold of 15 was evident for the majority of outcomes clearly moving from favorable to unfavorable. Fifty-nine of 135 patients (43.7%) had an APACHE II score of >15. This value was close to the median score (13 for SC and 14 for EVC). Multivariate analysis (table 3) of the overall study cohort demonstrated that age ≥75 years (OR 9.9, 95% CI 1.62 to 60.60, p<0.05), WFNS grade ≥4 (OR 10.81, 95% CI 3.51 to 33.31, p<0.0001), and SC (OR 5.27, 95% CI 1.63 to 16.95, p<0.01) were significant predictors of poor outcome; 35/66 (53.03%) SC patients versus 44/69 (63.76%) of EVC patients achieved a favorable outcome. However, the greatest predictor of poor outcome was an APACHE II score of >15 (OR 17.80, 95% CI 4.78 to 66.21, p<0.0001). When WFNS grade 1–2 patients were individually investigated so as to minimize the neurological influence within the APACHE II score on outcome, on multivariate analysis, an APACHE II score of >15 was the strongest predictor of poor outcome (OR 11.89, 95% CI 3.08 to 45.07, p<0.001).
The study population was then dichotomized by APACHE II score of ≤15 or >15 to explore outcomes by modality relative to the degree of physiological derangement. Descriptive statistics for two groups are shown in table 4.
In the APACHE II ≤15 group, there was a significant increase in length of hospital stay in the SC group (p<0.05). On multivariate analysis, poor WFNS grade was the only significant predictor of poor outcome (OR 13.76, 95% CI 2.25 to 83.93, p<0.05). For patients with an APACHE II score of ≤15, there was no significant difference in outcomes between SC and EVC patients across all grades or within the good grade (WFNS 1–2) subgroup. There tended to be more severe vasospasm in the SC group but this was not significant. The distribution of favorable and unfavorable outcomes by baseline APACHE II score subgroupings (≤15 or >15) for each aneurysm securing modality is shown in table 5.
For those patients with an APACHE II score of >15, there tended to be more vasospasm and poorer grade patients in the SC group but this did not reach significance. There were a significantly greater number of older patients (≥75 years of age) in the EVC group (p<0.05). On multivariate analysis, WFNS grade ≥4 (OR 6.74, 1.43 to 31.75) and SC (OR 6.33, 1.27 to 31.38) were the only significant predictors of poor outcome (p<0.05). In this population, 11% (3/27) of SC patients achieved a favorable outcome compared with 37.5% (12/32) of patients in EVC group. For good grade patients (WFNS 1–2) with an APACHE II score of >15, favorable outcome for SC was 3/10 (30%) and for EVC 9/16 (56.3%). SC patients also demonstrated significantly increased total norepinephrine dose, ventilated days, and hospital stay (p<0.05) relative to EVC patients. As grouped APACHE II scores increased, these outcome measures and the percentage favorable outcome (mRS 0–2) diverged for SC relative to EVC patients (table 6, figure 3).
The APACHE II scoring system is well established in intensive care practice. The system allows approximate prognostication based on a score from 0 to 71 for clinical and biochemical markers of cardiopulmonary, renal, hematological, and neurological function.12 A breakdown of the scoring system is displayed in table 7. A number of observational studies have related other physiological scores to outcomes in SAH patients.9 ,10 We aimed to identify whether the physiological derangement expressed by the APACHE II score had implications for differences in clinical outcome following either EVC or SC.
To our knowledge, this is the first study that has investigated the impact of a dichotomized APACHE II score in SAH. We have demonstrated that more significant physiological derangement, defined as a dichotomized APACHE II score of >15 at presentation, is a strong independent predictor of poor outcome in the SAH population as a whole. We did not investigate the impact of specific physiological variables within the score but it has previously been shown that hypoxemia, metabolic acidosis, hyperglycemia, and cardiovascular instability are the dominant factors that influence eventual outcome.8
While clinical outcomes did not differ by treatment modality in the less physiologically deranged population (APACHE II ≤15), for those with more profound physiological derangement (APACHE II >15) there was a significant difference in clinical outcome by treatment modality. Within this population there tended to be more poor grade (WFNS ≥4) patients in the SC group, but on multivariate analysis both poor grade and SC were predictors of poor outcome. Furthermore, functional outcome differences by treatment modality for the good grade subgroup (WFNS grades 1–2) tended to significance (perhaps not reaching significance as the study was underpowered), suggesting an effect independent of presenting grade. In keeping with this, outcomes in the EVC group were improved despite having a significantly greater number of older patients.
SC was associated with a marked increase in inotropic support (total norepinephrine dose), and hospital and ventilated days, suggesting a more profound physiological impact. This may have been confounded by a higher incidence of vasospasm in the SC group but the incidence of severe vasospasm was similar, and on multivariate analysis, clipping and poor WFNS grade were significant predictors of poor outcome of similar magnitude in these patients, whereas vasospasm was not. Functional outcomes for SC and EVC patients appeared to diverge with increasing APACHE II grouped scores along with degree of inotropic support, time ventilated, and hospital stay. The findings do suggest that the impact of the clipping procedure relative to a coiling procedure on deranged physiology may result in worse clinical outcomes.
These findings may give a clue as to why EVC is associated with better outcomes following treatment of aneurysmal SAH. Nevertheless, a clear underlying reason for these differences is uncertain. In one study, surgical and endovascular aneurysm therapies were associated with similar risks of cardiac injury and dysfunction after SAH.15 Rates of postoperative pulmonary edema were also similar when the treatment modalities were compared.16 Craniotomy and SC is associated with longer procedural times but whether it results in a greater systemic inflammatory response (which is known to complicate SAH)17 remains to be demonstrated. Immunodepression is also known to complicate SAH 18 and predisposes to pneumonia but the impact of aneurysm securing procedure has not been studied. Could a more minimally invasive aneurysm securing procedure limit these processes in some patients?
Common non-neurological complications of SAH include anemia, hypertension, cardiac arrhythmia, fever, electrolyte changes, pulmonary edema, pneumonia, hepatic dysfunction, renal dysfunction, and thrombocytopenia.19 Pneumonia, sepsis, fever, anemia, and hyperglycemia independently predict poor outcome and death.20 Medical complications are linked to severity of presenting grade.19 Additionally, a large Canadian multicentre study demonstrated that SC patients more commonly suffered medical complications, such as urinary tract infection, pneumonia, cardiorespiratory arrest, and seizures, and that these complications were linked to poor outcome.7 Our finding that SC patients with more profound physiological derangement (APACHE II >15) were noted to have significantly increased total norepinephrine dose, days ventilated, and hospital stay is consistent with this finding.
Many authors have attempted to explain the differences in outcome between patients undergoing EVC and SC through assessing the rates of cerebral vasospasm or delayed ischemia. Although there is some controversy regarding this, the general consensus is that that clipping results in more vasospasm and delayed ischemia.4 ,5 Naidech et al have also demonstrated an association between baseline physiological derangement and rates of infarction on CT.21 We demonstrated a significantly increased incidence of moderate–severe vasospasm in the SC population with more deranged physiology (APACHE II >15) but not in the SC population with an APACHE II score of ≤15, which suggests that a possible synergistic effect may exist, with both physiological derangement and craniotomy/clipping combining to result in a greater risk of vasospasm. However, the rate of vasospasm in the coiled population with an APACHE II score ≤15 was higher than in the population with an APACHE II score >15, suggesting that physiological derangement did not necessarily contribute to the development of vasospasm. Furthermore, we did not demonstrate a difference in the rate of severe vasospasm between treatment modalities. Interestingly, vasospasm failed to represent an independent predictor of poor outcome but the majority of patients in this study were managed using a paradigm employing angiographic screening and preemptive hypertensive and endovascular management on identification of significant angiographic vasospasm,15 which may limit its clinical impact.
The results of this study have implications for daily practice: the choice of whether to clip or coil an aneurysm has traditionally been made on the basis of morphological criteria, such as aneurysm location, size, and neck anatomy. We suggest, on the basis of these findings, that as well as anatomical criteria it is worth considering the physiological condition of the patient as one of the primary factors in assigning treatment modality. For example, is it preferable to clip a complex aneurysm in a medically unwell patient or would it be preferable to treat with partial coiling of the dome22 in order that the patient can be brought back for retreatment (using stents, flow diversion, or surgery) following recovery rather than undergo an acute clipping procedure? This approach has not been answered by this study but certainly warrants future investigation.
This was a small explorative analysis, and although the data suggest some interesting findings, it is necessary to explore these concepts in a larger population. The data were not collected specifically to answer the impact of physiological derangement relative to aneurysm securing procedure on outcome. Dichotomizing APACHE II score based on the available data introduces bias, and the assignment of treatment modality was not randomized so there is an inherent selection bias with respect to this also. The procedural complication rate was not specifically recorded. There were more posterior circulation aneurysms in the coiling group, and Fisher grade was not taken into account; it is plausible that the results were therefore biased by patients needing surgical hematoma evacuation being selected into the clipping group. However, in a recent prospective study of 381 consecutive SAH patients, hematoma evacuation was a reason for SC in only 3% of patients.23 Furthermore, in this study, SC was a significant predictor of poor outcome of similar magnitude to poor WFNS grade (indicative of acute neurological injury) on multivariate analysis, and there also tended to be more favorable outcomes in the good grade (WFNS 1–2) subgroup with more deranged physiology (APACHE II >15) when coiled rather than clipped, suggesting an interaction between treatment modality and physiological derangement independent of the acute neurological injury.
An additional criticism of this study is that the short follow-up period of 90 days is likely to show a poorer outcome when assessing a more invasive intervention, such as a craniotomy. Ideally, if a future study were to be constructed to explore the interaction between deranged physiology, treatment modality, and outcome, a longer follow-up period would be needed. On the other hand, it could be argued that this is precisely the point; patients undergoing a more invasive procedure may require more inotropic support, more time ventilated, and more time in hospital. They could therefore take much longer to recover and are therefore more at risk of nosocomial infection and other medical complications. The results do therefore give an interesting insight into the role that physiological assessment could have in shaping the design of future prospective investigations.
In summary, we have demonstrated that in this population, more marked physiological derangement at baseline is a strong predictor of eventual poor outcome. These preliminary results also suggest that clinical outcomes for patients with more severe baseline physiological derangement could possibly be improved if undergoing a coiling rather than a clipping procedure but this should be investigated using a prospective approach in the future.
Contributors All authors contributed to the manuscript and study.
Competing interests None.
Ethics approval The study was approved by the Northern Sydney and Central Coast Ethics Committee.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Parties interested in data sharing may contact Celia Bradford.
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.