Circadian rhythm and aneurysmal subarachnoid hemorrhage: Is there an alarm clock for the rupture timing?

Study population

This study is based on our retrospective, monocentric database including all patients with SAH who had been admitted to our institution between January 2003 and June 2016. All consecutive cases with radiologically and clinically confirmed aneurysmal SAH were eligible for the study. SAH individuals with unknown ictus time and the cases when the time of aneurysm rupture could not be estimated within a 60-min time range (±1 h) were excluded from the final analysis.

For this study, the approval for data assessment was obtained by the institutional ethics committee (Ethikkomission, Fakultät für Medizin der Universität Duisburg-Essen, registration number 15-6331-BO) and it is registered in the German clinical trial register (Deutsches Register Klinischer Studien, unique identifier DRKS00008749).

Clinical management

All SAH patients admitted to our institution received initial medical care at our neurosurgical intensive care unit. Conservative measures including blood and intracranial pressure management, perpetuation of euvolemia, vasospasm prophylaxis by daily intake of oral nimodipine and daily transcranial Doppler ultrasound were performed according to the guidelines [12–14]. The bleeding source was detected by digital subtraction angiography and/or computed tomography (CT) angiography. To prevent a rebleeding event, the treatment of the ruptured aneurysm by coiling or clipping was performed within the first 24 h after admission. Invasive endovascular vasospasm treatment was performed in SAH individuals with delayed ischaemic neurological symptoms and/or angiographic proof of vasospasm (in unconscious individuals). Chronic hydrocephalus was treated by ventriculoperitoneal shunt placement. CT scans were performed routinely within 24 h after each interventional treatment and during weaning of an external ventricular drain placed for the treatment of acute hydrocephalus. Moreover, patients underwent additional CT scans if clinically indicated. The post-treatment clinical follow-up of SAH patients was routinely performed at our outpatient clinic.

Data management

Bleeding time was determined according to the documentation in the electronic medical records, and the ictus time was rounded to the closest hour. In addition, the following clinical, demographic and radiographic parameters were collected from the institutional observational SAH database for correlations with the ictus timing: age, sex, ethnicity, preexisting diseases and medications, initial clinical and radiological scoring of SAH (according to the World Federation of Neurosurgical Societies [WFNS] [15] and original Fisher scales [16] respectively), aneurysm characteristics (location and size), specific clinical events during intensive care treatment (aneurysm rebleeding and occurrence of new cerebral infarcts in follow-up CT scans) and parameters regarding the patients’ functional outcome 6 months after SAH (according to the modified Rankin scale [17]). The clinical conditions of the patients’ medical history were determined by the patients’ electronic medical records, their regularly used medication at admission and the anamnesis provided by the patients themselves, their next of kin and/or family doctor. In particular, alcohol or drug abuse was defined as the excessive use of alcohol or drugs like heroin, cocaine, marijuana etc., linked to past social or legal problems, or failure to fulfill important roles in everyday life due to one’s alcoholism and/or drug use [18]. Chronic painkiller consumption was defined as regular use of painkillers (non-opioids and opioids) for longer than 3 months because of chronic pain in the patients’ history. Patients’ age was dichotomized at 65 years; the WFNS and Fisher scales at admission were regarded as low (WFNS 1–3/Fisher 1–2) or high (WFNS 4–5/Fisher 3–4) grades. Modified Rankin scale >3 at 6 months’ follow-up was considered a poor outcome. A list of all parameters can be found in Table S1.

Statistical analysis

For statistical evaluation of the relationship between the ictus and other recorded parameters, the cases were grouped into eight major time intervals according to the documented bleeding time: 1–3 a.m.; 4–6 a.m.; 7–9 a.m.; 10–12 a.m.; 1–3 p.m.; 4–6 p.m.; 7–9 p.m.; and 10–12 p.m. The associations were analyzed in a univariate and multivariate manner. First, a Fisher exact test was performed for each parameter for every time interval and individual factor of influence. Then, significant results were included in the multivariable backward regression analysis to confirm independent associations. p values smaller than α = 5% were determined as statistically significant. Statistical analysis was performed using SPSS version 25.0 (SPSS Inc., IBM, USA).

Population characteristics

Of 995 consecutive SAH cases treated during the observational period, 213 patients were excluded from the analysis as the bleeding time could not be reliably detected from the clinical documentation. Therefore, our final study population consisted of 782 SAH patients (see Table 1 for baseline and outcome parameters of the final cohort). The comparison of characteristics of individuals included in and excluded from the final cohort showed a higher portion of individuals with less severe SAH (WFNS 1–3/Fisher 1–2) and delayed hospital admission amongst the cases which were excluded from the final analysis (see Table S2).

Circadian rhythm of SAH

As Figure 1 illustrates, there were two major time peaks for SAH occurrence in our study population, one in the morning between 7 and 9 a.m. (19.7% of SAH patients) and the other in the evening between 7 and 9 p.m. (17.1%). Especially low rates of SAH occurrence were observed in the night and very early morning hours, namely from 1 to 6 o’clock with a total of 67 cases of SAH (8.7%) (see Figure 1).

Impact of socio-demographic and clinical characteristics on bleeding time

Over 20 different factors were analyzed with regard to their influence on the circadian rhythm of SAH (see Table S1). However, only few showed a significant association with the time pattern alterations (see Figure 2). The circadian rhythm of SAH was influenced by socio-demographic (weekday, age, sex, ethnicity) and clinical conditions (smoking, alcohol/drug/painkiller consumption, hyperuricemia, hyperlipidemia). In contrast, common cardiovascular risk factors like hypertension, obesity, cardiac diseases and diabetes mellitus were not related to alterations in the bleeding time.

Regarding the SAH-related characteristics, a larger sack size of the ruptured aneurysm (>5 mm) was associated with a higher rate of aneurysm bleeding between 4 and 6 a.m. The initial clinical and radiographic severity of SAH was not associated with the bleeding time (see Table S3).

Multivariable analyses were performed with the significant factors from the univariate analyses within the pre-defined time intervals (see Table 2). Between 1 and 3 a.m., hyperlipidemia remained the only significant factor. Alteration of the circadian rhythm of SAH by age over 65 years was significant from 4 to 9 a.m. Hyperuricemia significantly impacted the SAH occurrence between 7 and 9 a.m. Most individual factors (hyperlipidemia, alcohol, drug and chronic painkiller consumption) were significant in the time interval from 1 to 3 p.m., whilst the female patients showed significantly higher rates of bleeding events between 4 and 6 p.m. In turn, preexisting hypothyroidism and an ictus on the weekend were less common for bleeding events between 7 and 9 p.m. Finally, SAH individuals with a medical history of chronic painkiller consumption showed higher rates of ictus between 10 and 12 p.m.

Association between bleeding time with complications and outcome of SAH patients

The occurrence of cerebral infarcts during SAH and patients’ outcome at discharge and 6 months’ follow-up did not show a statistical association with the bleeding timing. SAH patients who suffered an aneurysm rebleeding before treatment were less likely to experience the initial bleeding between 10 and 12 p.m. (see Table S3).

Circadian rhythm and the occurrence of SAH

The circadian rhythm in humans is a complex system initiated by light exposure and following a feedback loop between central and peripheral clocks through neurohumoral pathways. Disruption of the circadian rhythm is associated with various cardiovascular as well as metabolic diseases and events like hypertension, atherosclerosis, myocardial infarction, insulin resistance, ischaemic and hemorrhagic stroke [19–21]. The diurnal occurrence of SAH suggests it to be another result of an impaired circadian rhythm. According to our large monocentric series, SAH naturally exhibits a circadian rhythm of its own, peaking in the early morning (7–9 a.m.) and evening (7–9 p.m.).

To date, only few studies have focused on this specific topic [5–7, 9–11, 22, 23], and even fewer specifically analyzed the possible impact of certain clinical, demographic, ethnic and behavioral factors on the circadian rhythm of SAH [7, 9, 10]. In summary, all previously published studies, including our own, agree on the diurnal two-peaked incidence of SAH. According to our analysis, major alterations to this rhythm can be observed for the following parameters: weekday, age, sex, ethnicity, chronic consumption of alcohol and/or painkillers as well as metabolic conditions like hyperuricemia and hyperlipidemia.

As our study suggests, in the case of SAH occurring on weekends, the bleeding time pattern is shifted by some hours so that the morning peak is reached at 10–12 a.m., rather than 7–9 a.m., and the evening peak is less prominent than in the workday subgroup. This may be due to behavioral factors such as awakening later on weekends and less blood pressure raising activities and stress compared to workdays. Melatonin and glucocorticoids control nighttime sleepiness and daytime consciousness and activity. Glucocorticoids also enhance the synthesis of catecholamines which contribute to vasoconstriction, endothelial dysfunction and platelet activation as well as potentiate endothelial sensitivity to catecholamines [19, 21]. Almost in sync with the temporal pattern of SAH occurrence, cortisol secretion peaks between 7 and 8 a.m. and is lowest around 2 and 4 a.m. In accordance with the cortisol secretion, blood pressure rises in the morning, approximately 1 h prior to awakening, is maintained by a vascular tone during the daytime and decreases gradually late in the afternoon, reaching its trough at around 3 a.m. with a dip of 10%–20% of the initial blood pressure [19, 21, 24].

The ictus peaks in younger SAH patients (<65 years) are more blunted compared to those of elderly patients. These results are similar to a recent retrospective study by Wu et al. who show a balanced curve of SAH occurrence in younger patients—in this case <55 years—and a dip at 4 a.m. [7]. Aging is associated with more comorbidities like atherosclerosis, hyperlipidemia, diabetes mellitus as well as sleep disorders like sleep episode fragmentation so the vasculature of elderly patients is more sensitive to environmental changes, for example blood pressure variation causing shear stress [25]. Endothelial dysfunction is promoted by impaired circadian rhythm, for example in sleep disorders or nightshift working [20, 26]. The endothelial system also follows a circadian rhythm by responding to neurohumoral signals with a different sensitivity as the day progresses. It is suggested that both endothelium-dependent and -independent relaxations are more remarkable at 3 a.m. [20, 21, 27].

Interestingly, the temporal distribution of SAH occurrence in patients with preexisting metabolic diseases like diabetes mellitus, cardiac diseases, hyperlipidemia, hyperuricemia and hypothyroidism shows an enhancement of the morning peak suggesting a damaged endothelium is more likely to decompensate with the morning surge. So, the morning surge with its rise in blood pressure falls into the vulnerable phase of the endothelium, which is assumingly why many cardiovascular events—such as SAH—occur in the early morning hours [19–21, 26, 28]. On the other hand, there was no significant rhythmicity in SAH occurrence in hypertensive or smoking patients in our study.

Male patients experienced an SAH with a lower prevalence between 4 and 6 p.m. than female patients in our cohort (6.3% vs. 12.7%). In the literature, there is only one study addressing this topic. Contrary to our results, female SAH patients had a lower prevalence around noon in that study [7]. In terms of hormonal differences between the sexes, estrogen plays an important role, especially in the protective effect on the vascular system by inhibiting artery calcification. Besides, there are behavioral and psychological differences between women and men which can trigger hormonal changes, for example in cortisol or catecholamine levels [26]. A study conducted on women with and without obesity demonstrates that cortisol levels remained higher in women with obesity after noon rather than a physiological drop in normal-weight women, and the cortisol response to food intake was higher [29].

Only 4.5% of our study cohort consisted of non-Caucasians. Although it is a small subgroup, interesting results could be identified with higher rates of SAH in this subgroup between 3 p.m. and 6 a.m. In particular in the time frame from 1 to 3 a.m., significantly more non-Caucasians experienced an SAH in comparison to Caucasian patients. The exact background of this association remains unclear. The impact of cultural factors, for example time of food intake, diet and behavioral differences in the circadian rhythm, has been widely discussed previously [30–34].

In contrast to the two-peaked trend of previous subgroups, the SAH occurrence in the subgroup of patients with chronic painkiller and/or alcohol consumption shows a major peak in the afternoon, between 1 and 3 p.m. Pathophysiologically, one can assume that patients of this subgroup tend to have shifted sleep–wake cycles and therefore an impaired circadian rhythm per se. Thus, the rhythmicity of SAH occurrence as seen above cannot be transferred fully in those cases. As our study is the first to analyze this subgroup of SAH patients in terms of rhythmicity, further investigations are necessary to verify or condemn these results.

Our findings might be helpful not only for better understanding of the mechanisms promoting aneurysm rupture but also for elaboration of specific recommendations for its prevention in individuals with unruptured aneurysms. In this context, many modifiable risk factors of aneurysm rupture have already been identified in the past decades, like smoking cessation and strict control of blood pressure. Whilst illustrating the circadian rhythm of bleeding time in this study, another important (and potentially modifiable) risk factor for the rupture event was identified. Our results showed associations between the timing of aneurysm rupture and specific socio-demographic, ethnic and behavioral patterns of the circadian rhythm in SAH patients. In particular, the moments of increased corticoid release and sympathetic activity may contribute to the rupture of an intracranial aneurysm. Therefore, the avoidance of strong fluctuations in cortisol levels (at least partially) by modification of behavioral patterns of patients with unruptured aneurysms might help to decrease the risk of SAH. As in many cases, this requires a multifactorial approach in accessing the individual risk profile with specific recommendations on the patients’ lifestyle with respect to the circadian rhythm and avoidance of stress peaks. Further research of the possible link between behavioral patterns, circadian rhythm and the timing of aneurysm rupture is mandatory.

Limitations

As this study is of a retrospective design, the accuracy of the collected data depends on the quality of documentation and recall of patients or their relatives in terms of rupture time. In particular, with regard to behavioral aspects and circumstances around the rupture time, there was no standardized questionnaire, so this information could only be retrospectively collected. Furthermore, approximately a fifth of the study population was excluded as the bleeding time could not be detected reliably. Excluded SAH patients differed significantly in terms of initial severity of SAH. This circumstance limits the generalizability of our results on SAH patients with lower initial WFNS/Fisher scores. For statistical purposes, the SAH timeline was divided into eight equal time intervals according to the major incidence peaks of bleeding events in our cohort. This approach might also impact the study results but allowed better statistical assessment (fewer subgroups with large sample size). Further studies of prospective and multicentric design are required to check these results.