Introduction

Systemic lupus erythematosus (SLE) is an autoimmune disease characterized by multiple organ involvement and the presence of autoantibodies. Delayed diagnosis and management can not only lead to organ damage but also increased mortality. Neuropsychiatric SLE (NPSLE) is not an uncommon manifestation with a prevalence rate of 14–95%, depending on the definition used. NPSLE can affect the central, peripheral, or autonomic nervous system, and can be either a focal or diffuse process, with severity ranging from mild to severe disabling conditions, resulting in significant morbidity and mortality. In general, a majority of NPSLE cases occur at the early stage of the disease, and NPSLE can be an initial manifestation of SLE occasionally [1].

Posterior reversible encephalopathy syndrome, or “PRES”, is an uncommon neurological disorder, characterized by hypertension, seizures, visual disturbance, and cognitive changes, with a unique brain imaging finding that is consistent with a reversible vasogenic edema at the posterior parietal or occipital lobes on computed tomography (CT) or magnetic resonance imaging (MRI) [2,3,4]. The disease was first described in 1996 by Hinchey et al., in patients who had eclampsia and acute hypertensive encephalopathy and received immunosuppressive agents [5]. During the past decade, PRES in patients with SLE has been reported increasingly as a case, or case series, and also reviewed in the literature from both western [6,7,8,9,10,11] and Asian countries [12,13,14,15,16,17]. Although many reports found that a majority of patients had active disease, particularly renal involvement, only a small number of studies were performed on a case-control study basis [9, 14]. Of interest, despite PRES being recognized increasingly, it is not included in the nomenclature of nervous system involvement in SLE [18].

This study aimed to determine the prevalence, clinical features, and outcomes of PRES in Thai patients with SLE, and also ascertain associated factors for the development of PRES by using a case-control study.

Materials and methods

The medical records from January 1986 to December 2016 of SLE patients in the Lupus Cohort of the Division of Rheumatology, Department of Internal Medicine, Chiang Mai University were reviewed. Those who had central nervous system (CNS) symptoms, and were diagnosed PRES according to an imaging study, were identified and included this study analysis. The diagnosis of SLE followed the 1997 American College of Rheumatology revised criteria for the classification of systemic lupus erythematosus [19]. Controls were selected from SLE patients in the same cohort, who had a hospital number that was close to and actually had SLE diagnosis within 5 years of the case. The ratio of case to control was 1:4. The clinical manifestations and laboratory findings in the controls, which corresponded to the PRES group in the same SLE disease period, were used for analysis. SLE disease activity was determined by the modified SLE disease activity index-2K (mSLEDAI-2K) [20].

All of the neuro-imaging studies were reviewed by two experienced neuro-radiologists: KO and SW. The neuro-imaging findings followed the description of Bartynski et al. [3], which was classified into (a) dominant parieto-occipital pattern, (b) holohemispheric watershed pattern, (c) superior frontal sulcus pattern, and (d) partial or asymmetric expression of the three patterns. The results were classified as complete resolution, partial resolution, no response, and progression of lesions in patients with a repeated neuro-imaging study. This study was approved by the Ethics Committee of the Faculty of Medicine, Chiang Mai University.

Statistical analysis

The STATA 13.0 computer software (Stata Corporation, Texas USA) was used for data processing and statistical analysis. Continuous data were expressed as mean ± standard deviation (SD) and categorical data as percent. The Student’s t test or Mann-Whitney U test was used to compare continuous data, and the Chi-square or Fisher’s exact test for comparing categorical data, where appropriate. Logistic regression analysis was used to determine associated factors for the development of PRES and reported as odds ratio (OR) and 95% confidence interval (95% CI). A p < 0.05 was considered statistically significant.

Results

Twenty-four of 1332 (1.80%) subjects in the cohort of SLE patients were found to have PRES. All of them were female. Four of these patients (16.67%) had recurrent episodes of PRES (2 repeat episodes in 2, and 3 in the other 2), giving a total of 30 PRES episodes. The mean ± SD age, age at SLE diagnosis, and age at the onset of PRES in all episodes were 29.78 ± 12.25, 25.02 ± 13.78, and 28.31 ± 12.61 years, respectively, which gave a mean duration of PRES development after SLE diagnosis of 3.29 ± 3.76 years. Ten (33.33%) episodes occurred within the first year of the disease. The PRES episodes led to SLE diagnosis in 2 cases, and PRES development within 1 month after SLE diagnosis in 3. Of those who had recurrent PRES, the mean duration of recurrent episodes was 0.69 ± 0.62 years after the previous episode.

Clinical features and serologic findings of SLE patients during PRES episodes are summarized in Table 1. All but one patient had high blood pressure at PRES onset, and the mean ± SD systolic blood pressure (BPs) and diastolic blood pressure (BPd) were 186.46 ± 29.21 and 116.42 ± 16.27 mmHg, respectively. Renal involvement, particularly active lupus nephritis, was a common finding seen in 26 episodes (86.67%). Renal biopsy was performed in 12 patients who had renal involvement, with lupus nephritis class III + IV, IV, and IV + V in 1 case, 10 cases, and 1 case, respectively. One patient had end-stage renal disease prior to developing PRES. Other clinical manifestations seen during 30 PRES episodes were fever in 11 (36.67%) episodes, thrombocytopenia in 10 (33.33%) [3 also had thrombotic microangiopathy: TMA], autoimmune hemolytic anemia (AIHA) in 7 (23.33%), skin rashes in 7 (23.33%), serositis in 3 (10.00%), leukopenia in 3 (10.00%), oral ulcer in 2 (6.67%), arthritis in 1 (3.33%), and pulmonary hemorrhage in 1 (3.33%) episode. Anti-dsDNA positive and low serum complement were seen in 14 of 20 (70.00%) and 8 of 23 (34.78%) episodes tested, respectively. Within 1 month prior to the onset of PRES, intravenous cyclophosphamide (IVCY) was given in 6 episodes [4 of them also received intravenous pulse methylprednisolone (IVMP)], and plasmapheresis was performed in one.

Table 1 Demographic data of 30 PRES episodes in 24 SLE patients
Full size table

Clinical features and outcome of PRES

Clinical, brain imaging study, treatment, and outcome of the 30 episodes of PRES are shown in Table 2. Seizure was the most common presenting symptom and seen in 28 of 30 episodes (93.33%), in which 3 (10.71%) had status epilepticus. Other symptoms included acute severe headache in 17 (56.67%) episodes, alteration of consciousness in 17 (56.67%), nausea and vomiting in 10 (33.33%), and blurred vision in 11 (36.67%), of which one had bilateral hemianopia and hemiparesis in 3 (10.00%). SLE disease activity determined by the mSLEDAI score at PRES onset was high (22.90 ± 7.17), even when the neurological component was excluded (mSLEDAI-N) (11.17 ± 4.62), indicating very active disease.

Table 2 Clinical, radiologic features, treatment, and outcome of 30 PRES episodes in 24 SLE patients
Full size table

Brain imaging study was performed in all of the episodes (computed tomography [CT] in 20 and magnetic resonance imaging [MRI] in 10) and the mean ± SD duration of brain imaging after the onset of symptoms was 2.15 ± 1.95 days. Parietal-occipital pattern, the most common brain lesion, was seen in 13 (43.33%) episodes, partial or asymmetric expression of the three patterns in 9 (30.00%) and holohemispheric watershed pattern in 7 (23.33%), and superior frontal sulcus pattern in 1 (3.33%) episode. Intra-cranial hemorrhage (ICH) was observed in 2 episodes. The pathological lesions involved the occipital lobes in 26 (86.67%) episodes, parietal lobes in 24 (80.00%), frontal lobes in 23 (76.67%), temporal lobes in 14 (46.67%), cerebellum in 13 (43.33%), basal ganglia in 7 (23.33%), brain stem in 6 (20.00%) and thalamus in 4 (13.33%), and bilaterally in 24 (80.00%), 20 (66.67%), 19 (63.33%), 11 (36.67%), 10 (33.33%), 7 (23.33%), 4 (13.33%), and 3 (10.00%) episodes, respectively.

Anti-hypertensive and anti-epileptic drugs were prescribed in 24 and 23 episodes, respectively. The mean ± SD duration of anti-epileptic drug treatment was 32.56 ± 27.55 days. High-dose corticosteroids (8 of which were IVMP) were given in all but one episode (96.67%). IVCY was given in 8 (26.67%) episodes to control lupus activity, mainly for lupus nephritis. Rituximab was given in 2 episodes and plasma exchange in 2 to patients with TMA. All of the patients had clinical improvement, e.g., return to normal consciousness, disappearance of headache, and blurred vision with mean ± SD duration of 1.65 ± 1.50 days.

A repeat brain imaging study was performed after the first one in 18 patients with mean ± SD duration of 13.46 ± 7.35 days. The brain imaging studies showed complete resolution of the lesions in 9 (50.00%) episodes, partial resolution of the lesions in 6 (33.33%), no improvement in 1 (5.56%) episode, and progressive lesions in 2 (11.11%) episodes.

Associated factors of PRES

Demographic data and co-morbidities at the first episode of PRES in 24 SLE patients were compared with 96 SLE controls, as shown in Table 3. Both groups had comparable age, age of SLE at onset, duration of SLE, and number of SLE classification criteria. Co-morbidities including diabetes mellitus, dyslipidemia, history of neurological symptoms (e.g. stroke and seizures), chronic kidney disease (creatinine > 1.5 mg/dL), past or current alcohol consumption and smoking, thalassemia, history of anti-phospholipid antibody syndrome, or past history of vasculitis were comparable in both groups, except for hypertension, which was more common in the PRES group (41.67 vs. 19.79%, p = 0.025). One patient was a hepatitis B virus carrier and 2 had hypothyroidism; with both being in the control group.

Table 3 Demographic data of SLE patients among cases with PRES and controls
Full size table

In order to identify the associated factors of PRES, clinical manifestations, and laboratory findings of SLE, patients with PRES (SLE-PRES), at their last visit prior to developing PRES were used for analysis, and compared with their controls, in order to identify its associated factors. As two patients developed PRES at the time of SLE diagnosis, and 3 with known SLE were referred from other hospitals after the onset of seizures, information on their last visit prior to PRES onset was not available. Therefore, only 19 patients with clinical and laboratory data prior to the onset of PRES remained for analysis. Clinical features and serological abnormalities, according to the 1997 ACR classification criteria, SLE disease activity and blood pressure among the SLE patients with PRES and their controls are shown in Table 4. Overall, cumulative clinical manifestations and serological findings were comparable in both groups, except that patients with PRES had more significant hematological (84.21 vs. 56.57%, p = 0.026) and renal involvement (89.47 vs. 43.42%, p < 0.001) than their controls. Among those with hematological involvement, SLE patients with PRES also had more significant autoimmune hemolytic anemia [AIHA] (68.42 vs. 25.00%, p < 0.00) and thrombocytopenia (26.31 vs. 9.21%, p = 0.045).

Table 4 Cumulative clinical features and serological findings according to the 1997 ACR classification criteria for SLE, SLE disease activity and blood pressure, and laboratory findings at last visit prior to onset of PRES among 19 SLE with PRES patients and 76 controls
Full size table

When comparing the last visit prior to PRES onset with their controls, SLE patients with PRES had significantly higher SLE disease activity, as determined by the mSLEDAI score (10.74 ± 3.48 vs. 3.99 ± 4.18, p < 0.001), and a proportion of those had high disease activity [mSLEDAI > 10] (52.63 vs. 7.89%). The mSLEDAI score was much higher among the PRES group at the time of PRES onset than at their last visit prior to the onset of PRES (24.63 ± 6.74 vs. 10.74 ± 3.48, p = < 0.001). However, when the neurological component of the mSLEDAI score (mSLEDAI-N) was excluded, the mSLEDAI score in SLE patients with PRES did not increase significantly (12.42 ± 4.04 vs. 10.74 ± 3.48, p = 0.072), indicating that the increased mSLEDAI score at PRES onset mainly came from the neurological component.

The mean ± SD BPs and BPd, as well as the proportion of patients with uncontrolled blood pressure prior to the onset of PRES, were comparable in both groups. When comparing the prior to onset of PRES with the time of PRES onset, the latter had a significant increase in BPs (185.79 ± 22.41 vs. 127.05 ± 12.77 mmHg, p < 0.001) and BPd (114.89 ± 11.22 vs. 78.32 ± 11.40 mmHg, p < 0.001).

Laboratory findings at the last visit before the onset of PRES among SLE patients with PRES and their controls are shown in Table 4 . SLE patients with PRES had a significantly lower mean hemoglobin level (9.10 ± 1.87 vs. 11.33 ± 1.90 gm/dL, p < 0.001), higher proportion of those with hemoglobin < 10 mg/dL (68.42 vs. 22.36%, p < 0.001), higher degree of proteinuria determined by urine protein/creatinine ratio (UPCR) [in gm protein/gm creatinine] (2.98 ± 2.15 vs. 1.13 ± 1.84, p < 0.001), higher proportion of those with UPCR > 1.0 (89.47 vs. 30.26%, p < 0.001), higher serum creatinine (1.61 ± 0.89 vs. 0.87 ± 0.53 mg/dL, p < 0.001), higher proportion of those with serum creatinine > 1.5 mg/dL (42.10 vs. 5.26%, p < 0.001), lower mean serum albumin (2.87 ± 0.69 gm/dL vs. 3.44 ± 0.78 gm/dL, p = 0.001), higher proportion of those with serum albumin < 3.0 gm/dL (57.89 vs. 19.74, p = 0.002), lower mean complement C3 level (389.75 ± 130.12 vs. 673.54 ± 338.90 mg/dL, p = 0.026), and higher proportion of those with a low complement C3 level (84.61 vs. 46.15%, p = 0.039) when compared to the controls. There was no significant difference in proportion of the number of patients with a positive test for serum anti-dsDNA antibodies in either group.

Table 5 shows the treatment received prior to the onset of PRES among SLE patients with PRES and those without. SLE patients with PRES received a significantly higher dose of daily prednisolone (33.16 ± 18.80 vs. 16.25 ± 14.20 mg/day, p < 0.001), and had a higher proportion of those receiving prednisolone of more than 15 mg/day (73.68 vs. 34.21%, p = 0.002), but a significantly lower proportion of those receiving hydroxychloroquine (15.78 vs. 47.36%, p = 0.012). There was no significant difference in the number of patients receiving immunosuppressive drugs or their type and dosage.

Table 5 Treatment received prior to onset of PRES among 19 SLE patients with PRES and 76 controls
Full size table

Differences in demographic data, cumulative clinical features, and laboratory findings were preceded by multiple logistic regression analysis, in order to determine factors that might predict the development of PRES (Table 6). Only recent proteinuria UPCR > 1.00 (OR 15.72, 95% CI 3.12–79.21, p = 0.001) and recent hemoglobin < 10 gm/dL (OR 5.12, 95% CI 1.37–19.15, p = 0.015) were found to be the associated factors in the development of PRES.

Table 6 Risk factors of PRES among SLE patients
Full size table

Mortality rate

With a mean ± SD duration of 1.64 ± 1.74 years follow-up, 7 patients (29.17%) in the PRES group died. Four deaths (57.14%), or 16.67% of PRES cases, occurred within 3 months after PRES onset. Causes of death were pneumonia in 5 cases, urinary tract infection in 1 case, and pulmonary hemorrhage in 1. The mean ± SD mSLEDAI and mSLEDAI-N score at the time of PRES onset prior to death was 17.25 ± 2.87 and 7.25 ± 5.06, respectively. Only 8 patients (8.33%) in the control group died, with a mean duration of 2.03 ± 2.00 years follow-up (p = 0.015).

Discussion

A 1.80% prevalence of PRES in SLE patients was identified in this study, indicating that PRES is an uncommon manifestation in SLE. The prevalence of PRES in this study was in the range of 0.69% reported from Taiwan [16] and 2.02% from Korea [14]. The prevalence in study and other reports from Asia [14, 16] might have been much higher than that reported from western countries. Although PRES in SLE have been reported from western countries, its prevalence was not able to be determined [6,7,8, 10, 11]. Three hospitals in Toronto, of which one also was a large lupus clinic, reported only 7 cases of PRES in SLE identified over 2.5 years [7]. Interestingly, 6 of these patients were Asian ethnics. The high prevalence of PRES in SLE among the Asian population might be related to more severity in SLE patients, as well as the high prevalence of lupus nephritis among Asian populations [21, 22]. It should be noted that one-third of the PRES episodes in this study occurred within the first year of their SLE diagnosis, and PRES led to the diagnosis of SLE in two patients, thus indicating that PRES tended to occur at the early stage of SLE. The onset of PRES in the early stage of SLE also has been recognized [7]. Furthermore, a majority of the patients also had active nephritis at the time of PRES onset.

The clinical features of PRES, including acute onset of hypertension, headache, nausea/vomiting, blurred vision, and seizures were common in this study, and similar to those previously reported [6,7,8,9,10,11,12,13,14, 16, 17]. However, only 42.85% of patients had a seizure in Barber’s series [7], which was lower than in others. Acute severe headache and seizures were the result of a sudden increase in intra-cranial pressure by abruptly increased blood pressure. Blurred vision was related to the involvement of lesions that involved posterior circulation of the brain, as confirmed by CT or MRI scan in almost 90% of the episodes [23]. The outcome of treatment in terms of neurological deficit in the present study was rather good, as the visual symptoms and seizures as well as neurological deficit recovered within a few days after blood pressure were controlled. This was confirmed by the improvement or disappearance of lesions in repeated brain imaging studies of approximately 83.33% of the episodes in the present study; however, 3 (16.67%) episodes showed no improvement or progressive lesions. This prevalence was similar to that recently reviewed by Ferreira et al. [23]. The rapid resolution of the imaging studies supported the capillary leakage from systemic hypertension and the effects of vascular endothelium as the cause of PRES [5].

Recurrent episodes of PRES occurred in 20% of the patients in this study, which was slightly higher than the 15.53% in that of Lai et al. [16]. Patients who had recurrent episodes in this study had very active lupus nephritis and poorly controlled hypertension. In addition, ICH was observed in 2 (6.67%) episodes in this study, which was lower than the 15.38% in that of Lai et al. [16], and the 26.92% in a recent review by Ferreira et al. [23]. Despite ICH being identified, neurological symptoms in these 2 patients improved shortly after their blood pressure was under control. Hypoalbuminemia (< 20 gm/L) and thrombocytopenia (< 30 × 109/L) were found to be risk factors associated with ICH [16]. However, the patients who had 2 episodes of ICH in this study did not have hypoalbuminemia or thrombocytopenia.

Treatment of PRES in SLE includes the use of anti-epileptic and anti-hypertensive drugs. High-dose corticosteroids and immunosuppressive drugs are used according to the activity and severity of organ involvement in SLE. It should be noted that patients in the PRES group had a tendency to receive more cyclophosphamide and mycophenolate mofetil, which might reflect more in this group on active disease and nephritis, and a higher proportion of patients in the control group received more hydroxychloroquine, which might reflect less severity of the disease, as determined by the mSLEDAI score prior to the onset of PRES.

The development of PRES shortly after IVMP and IVCP has been previously mentioned [24,25,26], and 6 patients in the present study had been given intravenous IVCY (4 of them also received IVMP) within 1 month prior to developing PRES. High-dose glucocorticoids might enhance abnormal vascular tone and the cytotoxic effect of cyclophosphamide on the vascular endothelium has been proposed as the mechanism [25, 26]. However, it is difficult to assume that PRES was attributed to these 2 drugs, as SLE patients who developed PRES tended to have active disease and received many medications. In addition, the number of reported PRES cases in SLE patients seems to be very small when compared with the number of SLE patients that have been given IVMP and IVCY. Therefore, the role of these 2 medications in the development of PRES needs further studies.

The mortality rate of 29.17% in PRES patients in this study was significantly higher than that of the controls. The 16.67% of deaths that occurred within 3 months after PRES onset also was also high. It should be noted that the major causes of death were not related directly to PRES. The mortality rate was reportedly 30.00% from South Africa [27], 26.92% from Taiwan [16], 13.33% from Korea [14], 7.69% from India [13], and 4.76–6.25% from Mexico [9, 28]. However, there was no mortality reported from Canada [7], USA [6], or Malaysia [12]. Whether the high mortality rate was related to ethnicity or the healthcare system was not clear. In this study, the high mortality rate might have been related to a very active disease and the presence of extra-renal manifestations, as determined by the very high mean ± SD mSLEDAI score of 22.90 ± 7.17 during PRES episodes and even 11.17 ± 4.62 when the neurological component was excluded (mSLEDAI-N). A high SLEDAI score ≥ 18 without the neurological component (SLEDAI-N) has been associated with increased mortality [16]. The mSLEDAI-N in patients in this study was lower than that reported by Lai et al. [16], as 4 points from the complement and the dsDNA components were not included in the scores.

This study found that anemia (hemoglobin < 10 gm/dL) and proteinuria (UPCR > 1.0) were associated factors of PRES. The presence of active nephritis during PRES episodes has been well recognized. However, in a multicenter case-controlled study involving 48 cases of PRES and 96 controls, Merayo-Chalico et al. [9] found that hypertension, renal dysfunction, lymphopenia, SLEDAI score of ≥ 6, and younger age at onset were independent associated factors of PRES. In another case-controlled study involving 15 cases of PRES and 48 controls, Jung et al. [14] found that only renal insufficiency (creatinine ≥ 1.5 mg/dL) was an independent risk factor of PRES. To the authors’ knowledge, the presence of anemia found in this study has never been described as an independent associated factor of PRES in SLE. Further studies need to confirm this finding.

The strength of this study was due to it being a controlled study. The control subjects were recruited from the same SLE cohort, had a hospital number that was close to the case, and SLE diagnosis within 5 years to the case, which ensured that the cases and controls had similar lupus healthcare. Despite many studies having reported that active lupus nephritis is a risk factor for PRES in SLE, only a few of them were performed on a case-control basis [9, 14]. However, there are still some limitations in this study. The medical records were reviewed retrospectively, therefore, some clinical and laboratory data might be missing. In addition, the treatment of SLE and PRES depended on discretion of the consultant rheumatologists. Not all patients had received an MRI brain imaging study either initially or for follow-up, mainly due to the reimbursement in healthcare policy for the patients, which made the precise location of the lesions and follow-up of the disease difficult. However, all brain imaging in this study was reviewed by neuro-radiologists, who were well experienced in reading neuro-imaging. Therefore, errors in interpretation or misclassification of the lesions would likely be minimal. The small number of cases might be an issue and have some impact on the statistical analysis. However, the 30 PRES episodes among 24 patients from a single center in this study should not be considered as small when compared with the 48 PRES episodes among 43 patients from the multicenter study of Merayo-Chalico et al. [9]. Lastly, the Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index (SLICC-DI) was not determined in this study. It might be of interest to see if the SLICC-DI was associated with the mortality rate in this study. The presence of high SLICC damage score has been shown to associate with high mortality [29].

Conclusion

This study described the clinical characteristics and associated factors of PRES in Thai patients with SLE. PRES usually occurred in those who had active disease, particularly active renal involvement. The clinical features, brain imaging study, and outcomes of PRES in Thai patients with SLE were similar to those described previously. The presence of high-degree proteinuria (UPCR > 1.0), which reflected underlying active nephritis and anemia (recent hemoglobin < 10 gm/dL), was found to be an independent associated factor of PRES. The presence of anemia as an associated factor of PRES needs to be confirmed.