In our study, 14 (45.2%) of the 31 patients with SH and 11 (45.8%) of the 24 patients with TCH had RCVS. Over past decades, several studies have recruited patients with SH, focusing on the identification of SAH and sentinel headache (Table 4). A significant proportion of patients exhibited SAH and other secondary causes, but more than one-half of the patients in these studies did not receive a definite diagnosis. It has been postulated that RCVS accounts for most of these “benign TCH” cases . Our results show that about one-half of these patients exhibit RCVS. Due to a lack of similar studies, to the best of our knowledge, we compared our results with studies addressing related headache syndromes. In a study of 30 patients with HSA, 18 (60%) patients exhibited RCVS, 10 (33.3%) exhibited primary HSA and 2 (6.7%) exhibited other secondary causes (SAH and basilar artery dissection) . In another study of 21 patients with bath-related TCH, 13 (62%) patients exhibited RCVS and no other secondary causes were identified . The target populations in the above studies varied from that of our study: patients in the above studies were recruited mainly based on specific triggers (sexual activity and bathing, respectively), while our patients presented with SH or TCH, with or without triggers. In conclusion, our study provides direct evidence that RCVS is a common and likely under-recognized cause of SH and TCH.
Demographic data in our RCVS patients recaptured an obvious pattern reported in other studies with larger sample sizes: a majority of older female and a minority of younger male patients. In our study, around 80% of patients with RCVS were female, compared with 64.2% to 89.6% in previous studies [9, 11, 18]. Female RCVS patients were typically aged in their late forties or fifties, while male patients were aged in their thirties (women: 53.8 ± 5.8 years, men: 33.0 ± 4.2 years). The mean age of our female RCVS patients seemed slightly older than those in other studies (women: 44.2-49.7 years, men: 34.7-34.9 years) [9–11]. We propose that this difference is due to the absence of a third group of RCVS patients in our study: younger women in the postpartum state. This proposition is likely supported by the larger standard deviation in studies with higher ratios of postpartum female [9, 11].
The clinical profiles of our RCVS patients differed from previous studies in several aspects. First, our patients demonstrated more favorable outcomes. Only a few patients presented with subtle neurological symptoms, and no patients exhibited focal neurological signs or seizure throughout the course of the study. We did not observe any complications of cerebral infarction, intracerebral hemorrhage, or posterior reversible encephalopathy syndrome. Second, most of our RCVS patients were spontaneous (78.6%), i.e., without precipitants. There was no patient in the postpartum state. All patients with possible vasoactive substances used cough medicine and the percentage was probably over-estimated (please refer to Precipitating factors paragraph of Results). We encountered only one Vietnamese patient who reported use of ecstasy one month after childbirth but she was seen before the initiation of the study and thus not included . This was in contrast with the use of recreational drugs (especially cannabis) in French study or serotonergic drugs (selective serotonin reuptake inhibitors and triptans) in American study [9, 11]. In general, the above clinical profiles were similar to another Taiwanese study but differed from French and American cohorts [9–11]. The discrepancy may be attributed to the clinic-based settings of both Taiwanese studies, in comparison with emergency or inpatient settings of other studies, i.e. patients with complications were more likely to call upon emergency service and to be admitted. It is also possible that the benign course of our RCVS patients were related to the absence of postpartum state and cannabis use, as both have been associated with stroke and poorer outcome [20, 21]. However, another study from the same French group reported women and migraine history, instead of postpartum or vasoactive substances, were associated with intracranial hemorrhage in RCVS patients . Two recent retrospective studies of RCVS do not identify any precipitants as predictors of outcome. Of note, the target in one study was clinical worsening (rather than final outcome), while the other study was small in sample size (n = 10) [23, 24].
In our study, patients with RCVS were less likely to have short headache duration < 1 hour (Table 3). The duration of RCVS ranges from 5 minutes to 36 hours with a mean of 5 hours in one study, and a median of 3 hours in another study [9, 10]. The duration of primary headaches vary in a wide span, by the definition of ICHD-2 . For primary TCH, the duration range is 1 hour to 10 days and for primary exertional headache, the duration range is 5 minutes to 48 hours. Nevertheless, a study of 596 adolescent patients with primary exertional headache reported 467 (78.4%) patients had headache duration < 1 hour . The duration of primary cough headache (1 second to 30 minutes) is significantly shorter than the duration of RCVS; however, only two (11.8%) of the non-RCVS patients in our study reported cough as the trigger (Table 3). Another study of HSA showed no difference in headache characteristics, including duration, between patients with primary HSA and RCVS . In general, fewer patients with RCVS had short headache duration (< 1 hour) compared to non-RCVS patients; however, the headache duration may vary significantly in the latter group.
Among the various triggers, bathing (including showering and water exposure) was more frequently associated with RCVS in both SH and TCH cohorts. It has been reported that RCVS was noted in 13 (62%) of 21 patients with bath-related TCH . In our study, all patients reported bathing as a trigger were in the RCVS group (100%, n = 6). Bathing is unique among RCVS triggers, in that it is not associated with the Valsalva maneuver or emotion. In contrast to bathing, exertion was associated with non-RCVS diagnoses, but the significance of this association was borderline, and only in the SH sample (SH: P = 0.048; TCH: P = 0.098). Further studies are necessary to validate the associations between different triggers and RCVS.
Although TCH was first used to describe the headache associated with an unruptured aneurysm , only one (3.1%) patient with SAH was noted in our study. The studies focused on SAH and sentinel headache on patients with SH showed that 6.2-25% of them exhibited SAH, except for the earliest study, which identified 71.4% of patients exhibiting SAH (Table 4) [3–5, 26]. Compared with a very modest decline of SAH incidence over recent decades, the dramatic decrease of SAH ratios in these studies was confusing . We proposed that the low ratio of SAH in our patients may be due to the following reasons. First, the definition of SH, unlike that of TCH, was not well-determined and varied with each study (Table 4). Second, previous studies either recruited patients from an emergency department only or from both emergency and outpatient clinics, while our study was strictly clinic-based. Given the low rates of patients receiving lumbar puncture and catheter angiography in our study, the possibility of missed SAH may not be excluded. Nevertheless, based on the high sensitivity of the modern CT to detect SAH and CTA/MRA to detect aneurysm and the fact that no patients developed SAH during at least 6 months of follow-up, the contribution of missed SAH to the low SAH percentage in our study may be minor [17, 28, 29].
This study has limitations. First, the sample size is relatively small and further study with a larger population is needed. Second, as stated above, the definition of SH and the clinical setting were not consistent across studies and the ratios of RCVS in these patients may change accordingly. In this study, we included patients with a new SH of possible vascular origins while excluded those with typical clinical presentations suggesting other primary or secondary headaches. This may carry a potential risk of missing SAH or RCVS patients presenting with typical features of other headaches. We also excluded patients with recurrent SH or TCH. Interpretation and generalization of the data should be handled with caution. Third, the percentage of patients receiving catheter angiography was low. Although CTA and MRA have been widely accepted for detection of vasoconstriction in patients with RCVS, catheter angiography is still the gold standard [7, 18]. Besides, vasoconstriction has been reported to elude primary detection, and serial repetition may be necessary [9, 18]. Therefore, the percentage of RCVS may be underestimated. This observation did not change—and may perhaps strengthen—our conclusion that RCVS is a common cause of SH and TCH. The low angiography rate may also contribute to the result that arterial dissection in patients with RCVS was not observed . Fourth, the percentage of RCVS patients receiving lumbar puncture was also low (SH: 21.4%, TCH: 18.2%). As stated above, we could not exclude the possibility of missed SAH, but the chance may be low. Fifth, several rare causes of TCH were not excluded properly. Without MR venography, cerebral venous sinus thrombosis may be ignored, but the neuroradiologist denied any related findings by CT or MRI in all patients. Without routine lumbar puncture, meningitis may be undetected; however, none of the patients exhibited fever or neck stiffness. We did not screen pheochromocytoma, though no patients exhibited uncontrolled hypertension. Altogether, these diseases were rare in patients with isolated TCH, and we were convinced that the ratio of RCVS may not change significantly .