We conducted a retrospective analysis of patient records. A total of 340 patients from one major German hospital with TGA diagnosed by a neurologist according to the criteria defined by Hodges and Warlow (12) were identified between January 1, 2011 and December 31, 2020. Patients who did not meet the diagnostic criteria were not included in the study. This study is an extension of the cohort from a previously reported study (10). Patients treated in 2011, 2012 and 2020 were added so that a full decade could be analyzed.

All patients were included with a discharge diagnosis of TGA during the relevant period. For the identification of possible recurrences, all subsequent inpatient and outpatient contacts were analyzed for these patients, even if the patients presented to the emergency department or outpatient department for other complaints. Evaluation of recurrences was also performed by a neurologist using the diagnostic criteria.

Patient characteristics were evaluated including demographics, cardiovascular risk factors and imaging findings [MRI or computed tomography (CT) scan, ultrasound, echocardiography].

Blood pressure values on admission were recorded and evaluated as single blood pressure values. Chronic hypertension was assumed if it was already indicated by patient history on admission or if antihypertensive medication was required because of hypertensive blood pressure levels during hospital stay and at discharge. Hypertensive blood pressure peaks on admission that normalized during the course and did not require sustained antihypertensive medication were not considered as chronic hypertension.

MRI scans were evaluated for signs of stroke and hippocampal diffusion-weighted imaging (DWI) lesions. The extent of cerebral microangiopathy was assessed using Fazekas' score (0–3) (13) reviewed by a neuroradiologist. The presence of cerebrovascular stenosis was assessed using ultrasound. Laboratory parameters [cholesterol level, glucose level, HbA1c, C-reactive protein (CRP)] were obtained from emergency room records. Echocardiography was evaluated for left ventricular ejection fraction and septal hypertrophy. Septal hypertrophy as a possible indicator for chronic hypertension has been defined as the presence of increased septal thickness (women > 9 mm, men > 10 mm) (14). In addition, CHA2DS2-VASc scores were determined for all patients at the time of discharge. This means that antihypertensive medication at discharge was considered as hypertension in the calculation of the CHA2DS2-VASc score.

Subsequent stays in our hospital, occurring as a result of TGA recurrence, were assessed. Our hospital operates the only department of neurology in town serving a catchment area of about 320,000 people. We therefore assumed that patients with TGA symptoms within this area most probably present to the emergency room of our institution.

To determine an appropriate recurrence rate, only patients from 2011 to 2015 were considered for this analysis, allowing for a minimum follow-up period of five years.

Data analysis was carried out using the Statistical Package for the Social Sciences (SPSS) version 25 (IBM®, International Business Machines Corporation, 2018). Descriptive statistics were displayed as mean ± standard deviation for continuous data and frequencies with percentages for categorical variables. Normal distribution was assessed via Shapiro–Wilk test with p < 0.05 indicating non-normal distribution, and homoscedasticity was assessed visually via q-q-plots.

The profile of cardiovascular risk factors was compared between TGA patients with and without recurrence by using parametric t-tests or non-parametric Mann–Whitney U-tests, depending on normal distribution. Patient groups (with versus without TGA recurrence) were compared using chi-square tests for categorical variables. A significance level of less than 0.05 in the two-sided test was assumed to be significant. In order to correct for alpha error accumulation in multiple testing, p-values were adjusted using the Bonferroni method (p_adj = p_obs^*k; where p_adj: adjusted p-value, p_obs: observed p-value and k = number of comparisons) (15).

In total, 340 TGA patients were included in this study. Twenty-four of them had suffered from a documented TGA recurrence and were hospitalized in our clinic. Mean age was 67.3 ± 10.5 years in TGA patients without recurrence and 63.6 ± 8.6 years in TGA patients with recurrence at onset of first episode (see Figure 1A). TGA patients with recurrence were therefore significantly younger than patients without recurrence (Mann–Whitney U = 2,797.500, z = 2.144, p = 0.032). No significant difference for gender was observed. Further demographic characteristics are presented in Table 1.

The mean follow-up period of all TGA patients (N = 340) was 4.5 ± 2.7 years [Minimum: 15 days, Maximum: 9.98 years, Median: 4.57 years]. The mean latency to recurrence was 3.0 ± 2.1 years [Minimum: 32 days, Maximum: 7.7 years, Median: 2.60 years] after the first episode. The number of sustained TGA episodes varied from one to four episodes per patient. No significant difference for mean latency to recurrence for gender was observed (male 2.3 ± 2.0 years; female 3.4 ± 2.1 years; Mann–Whitney U = 89.000, z = 16.771, p = 0.215). The annual incidence rate for recurrence was 1.6% per year in the overall collective.

To determine an appropriate recurrence rate, only patients from 2011 to 2015 were considered for a subgroup analysis. These had at least a five-year follow-up period. The mean follow-up period of this subgroup (N = 160) was 7.0 ± 1.4 years [Minimum: 5.1 years, Maximum: 9.98 years, Median: 6.68 years]. During this period, 18 of 160 patients suffered a TGA recurrence corresponding to a recurrence rate of 11.3% (see Figure 2). The annual incidence rate of recurrence in the subgroup with at least a five-year follow-up period was also 1.6% per year.

There were no differences in vascular risk factors on admission (systolic and diastolic blood pressure, cholesterol, HbA1c, CRP, CHA₂DS₂-VASc score) between patient groups (isolated vs. recurrent TGA), as assessed by parametric t-test (in case of normal data distribution: diastolic blood pressure, cholesterol) or Mann–Whitney U-testing (in case of non-normal data distribution: other parameters) (see Table 1). Furthermore, no differences in cardiovascular comorbidity were detected with regard to chronic hypertension, septal hypertrophy in transthoracic echocardiography and hypercholesterolemia. Diabetes mellitus, atrial fibrillation, presence of cerebral stenosis, former stroke and left ventricular ejection fraction (LVEF) in transthoracic echocardiography each had small sample sizes. However, there was no significant difference here either (see Table 1).

About 278 of 316 TGA patients without recurrence and all 24 TGA patients with recurrence underwent MRI and were compared. There was no difference between both groups in the presence and number (unilateral versus bilateral) of hippocampal DWI lesions (see Table 1).

The extent of the cerebral microangiopathy was evaluated using the Fazekas' score in available MR images. Fazekas' scoring of available MR images revealed no cerebral microangiopathy (Fazekas' score 0) in 65.2% of TGA patients with recurrence (32.9% of TGA patients without recurrence), mild microangiopathy (score 1) in 21.7% (53.5%), moderate microangiopathy (score 2) in 8.7% (10.1%) and severe microangiopathy (score 3) in 4.3% (3.5%). Distribution of microangiopathic lesions using Fazekas' score in both groups (with versus without recurrence of TGA) is displayed in Figure 1B. Using Mann–Whitney U-test, the degree of microangiopathy (ranking from 0 = none to 3 = severe) was significantly lower in TGA patients with recurrence compared to TGA patients without recurrence (U = 2,326.000, z = 2.394, p = 0.017). The effect size according to Cohen was r = 0.25 corresponding to a small effect.

Based on an existing correlation between age and extent of cerebral microangiopathy in all patients (Pearson r 0.334, p < 0.001), a subgroup analysis of patients ≤ 70 years was performed (Mann–Whitney U-test of age between these groups: U = 1,771.500, z = 0.777, p = 0.437). Between groups with recurrent versus isolated TGA in patients aged ≤ 70 years, Fazekas' scoring of available MR images revealed no cerebral microangiopathy (Fazekas' score 0) in 70.0% of TGA patients with recurrence (41.7% of TGA patients without recurrence), mild microangiopathy (score 1) in 20.0% (43.9%), moderate microangiopathy (score 2) in 5.0% (7.1%) and severe microangiopathy (score 3) in 5.0% (2.0%). Using Mann–Whitney U-test, the degree of microangiopathy (ranking from 0 = none to 3 = severe) was significantly lower in TGA patients aged ≤ 70 years with recurrence compared to TGA patients without recurrence (U = 1,346.000, z = 2.045, p = 0.041). The effect size according to Cohen was r = 0.26 corresponding to a small effect. This analysis demonstrates a difference regardless of age.

In this subgroup of patients aged ≤ 70 years without microangiopathic changes on MRI (Fazekas' score 0) and at least 5 years of follow-up (N = 49), 12 patients suffered a recurrence within 5 years (24.5%).

Oliveira et al. reported more punctate hippocampal hyperintensities in patients with recurrent TGA (n = 5) versus single TGA (N = 0) in a total of 34 patients with available MRI data (p = 0.001) (7). In our study, there was no difference between patients with versus without subsequent recurrence in the presence and number (unilateral vs. bilateral) of hippocampal DWI lesions. This result is consistent with other previous studies (8, 27, 28).

A recent study by our group showed an association between acute hypertension and TGA occurrence (10). Herein, TGA patients were less likely to suffer from chronic hypertension compared to acute stroke patients. This was reflected by lower levels of consecutive hypertensive disorders, such as the extent of cerebral microangiopathy and septal hypertrophy in transthoracic echocardiography. These observations are substantiated by evidence of a lower vascular mortality risk in TGA patients compared with TIA patients (11). It was hypothesized that acute hypertensive peaks may trigger TGA episodes, especially in patients who are not adapted to chronic hypertension (10). These observations are well in line with our current results. The risk of recurrence is increased in patients who (1) are younger and (2) have less sequelae of chronic hypertension such as cerebral microangiopathy secondary to wall hypertrophy in arterioles. These patients might be more prone to TGA recurrence as they are less well adapted to episodes of acutely increased blood pressure. The predictor “lack of adaptation to hypertensive blood pressure episodes” persists longer in younger patients and is supported by the absence of microangiopathic lesions, leaving these patients at persistently increased risk for recurrent TGA. It is unclear, however, whether the elevated blood pressure values are a cause or consequence of a TGA episode. One could argue that the elevated acute blood pressure values are a consequence of emotional stress. Alternatively, TGA patients have significantly higher acute blood pressure values on admission compared with acute stroke patients (10), whereas stroke patients may have greater stress.

In clinical practice, patients frequently ask if they will have to face TGA episodes ever again. Our data may assist in patient guidance by informing about the presumptive recurrence risk. Patients with an age up to 70 years and absence of cerebral microangiopathy are subjected to an approximately 25% risk of recurrence within a median latency of four years. Therapeutically, the avoidance of blood pressure peaks would be advisable but cannot be proven by our data. Interestingly, a case report of successful prophylaxis of recurrent coital TGA with metoprolol supports this theory (31). Further prospective studies are required in order to address this issue.

Because of the retrospective design, it is possible that not all patients with TGA recurrence were included in our study, since no telephone survey was conducted using recurrence questionnaires. However, our hospital is a tertiary hospital operating the only neurology department including a large Stroke Unit for a catchment area of more than 320,000 inhabitants. TGA patients are typically admitted to our hospital due to presumptive stroke. Therefore, the vast majority of TGA patients in our area should have been treated in our department and subsequently included in the study.

Due to the retrospective study design, the exact time of TGA symptom onset and the resulting time interval between symptom onset and the MRI examination could not be clearly ascertained in all cases. Unfortunately, many of reported TGA predictors, such as personal and family history of migraine, personal history of depression, positive family history of dementia and a personal history of head injury, as well as several conditions under which TGA occurred, such as Valsalva maneuver, medication and sexual intercourse, could not be examined in our retrospective study design in particular because these conditions were not systematically recorded in patient records. This is a weakness of the study due to the retrospective study design. In a prospective study approach, these would be relevant parameters to collect. Due to the lack of a systematic survey of the duration of amnesia as well as limited MRI rates, a rare disease such as transient amnestic epilepsy could not be disclosed in all cases. We believe that this very rare differential diagnosis could be adequately delineated by inpatient workup in a neurology department. Patients with transient epileptic amnesia did not receive a diagnosis of TGA. Because of the lack of systematic follow-up, we have to consider that the older participants had a higher probability of dying within the next 5 years compared to younger participants and therefore did not experience another TGA episode (resulting in an association of younger age and TGA recurrence). It is also possible that some patients with recurrence did not seek neurologic help if it resembled a previous TGA. Here, the behavior in seeking help could possibly differ between younger and older patients. Another limitation of our study is the overall small sample size of TGA recurrence. This leads to a limitation of possible calculations, e.g., association of diabetes and risk of recurrence. The small sample size questions the generalizability of the data.