Factors predictive of corticosteroid psychosis in patients with systemic lupus erythematosus

Patients and methods.

Between January 1999 and June 2001, SLE patients in our unit who were treated with corticosteroids for the first time or who received an augmented dose of the drug were prospectively followed. All patients fulfilled at least four of the American College of Rheumatology criteria for SLE.8 Disease characteristics, steroid regimens, psychiatric history, and various biochemical parameters at the time of corticosteroid treatment were recorded. The patients were evaluated at regular intervals in our rheumatology clinics for their disease status and the occurrence of overt psychiatric symptoms.

Corticosteroid-induced psychiatric disturbances were defined as new symptoms that appeared temporally within 8 weeks of institution or augmentation of steroids and resolved completely by a reduction in steroid dosage without additional immunosuppressive agents. Other causes such as infection and metabolic derangement had to be excluded. Psychiatrists diagnosed the psychiatric events using the Diagnostic and Statistical Manual of Mental Disorders (4th edition) criteria. Psychosis referred to disturbance in the perception of reality characterized by delusions or hallucinations, or both, without insight and in the absence of delirium. Mania was classified under mood disorder with manic features and was characterized by predominantly and persistently elevated, expansive, or irritable mood. The disturbance had to be severe enough to cause significant distress or social impairment.

Results.

One hundred twenty-six courses of corticosteroids were administered to 92 SLE patients. Table 1 shows the clinical information of the patients. Six psychiatric events occurred in six patients, giving an overall incidence of 4.8% for corticosteroid-induced psychiatric adverse events. The incidence was not different between patients with first-time steroid use and those with dosage augmentation (3.7% vs 6.8%; p = 0.42). Table 2 summarizes the nature of psychiatric events and investigation results of the patients who developed psychiatric disturbances to corticosteroid use. Three had psychosis, whereas three developed mania.

At the time of psychiatric events, three patients were first-time steroid users (daily prednisone dose range, 30 to 45 mg), and three had dosage augmentation (mean increase in daily prednisone dose from baseline, 25.8 mg [range, 15 to 32.5 mg]). All patients were hypoalbuminemic, and none had neuropsychiatric symptoms before steroid treatment. All events developed within 3 weeks (mean, 12.2 days; range, 1 to 21 days) of steroid administration, and in five of the six episodes, symptoms resolved completely after steroid dosage reduction. In one patient, an additional 8-week course of phenothiazine was given. Psychosis did not recur after phenothiazine discontinuation. The mean daily prednisone dose in these six patients was reduced from 40.0 mg to 18.3 mg (54% reduction).

Neurologic investigations were unremarkable for all patients except for mild elevation of CSF protein and cerebral atrophy in one patient. CSF immunoglobulin (Ig) G oligoclonal bands were negative in the three patients who had the test done. MRI of the brain performed in three patients did not reveal evidence of active CNS lupus or infection.

Table 3 shows the clinical characteristics of the patients and steroid regimens with regard to whether psychiatric events had developed. Univariate predictors of psychiatric events were history of anxiety disorders, low C3 levels, and hypoalbuminemia. A family history of psychiatric disorders and lower creatinine levels also reached borderline significance. After entry of these five covariates into the multivariate regression model, only hypoalbuminemia remained an independent predictor for psychiatric events (hazard ratio, 0.80 per g/dL [95% CI, 0.60 to 0.97]; p = 0.03).

Discussion.

In patients with SLE, corticosteroid-induced psychiatric events may be difficult to distinguish from neuropsychiatric manifestations of the SLE. The antiribosomal P antibody is specific for SLE and has been shown to be associated with disease-related psychosis.9 However, the low sensitivity limits its clinical usefulness for the diagnosis of lupus psychosis. Thus, a temporal relationship between beginning corticosteroids and psychiatric events, and resolution of symptoms after dosage reduction or controlled withdrawal of corticosteroids remain important to suggest a treatment-related problem.

In a study of SLE, psychosis was described in 57% of patients who were given large doses of prednisone (100 to 500 mg/d).6 Another survey reported a 30% incidence of psychosis in SLE patients receiving high-dose prednisone (60 to 100 mg/d).7 Because a much lower prednisone dose was prescribed to our SLE patients and a detailed psychiatric and cognitive assessment was unavailable for apparently asymptomatic patients, the incidence of corticosteroid-induced psychiatric disturbances may have been underestimated.

There is some evidence for a relationship between corticosteroid dose and psychiatric reactions.2 The incidence of psychiatric reactions in hospitalized patients receiving 40 mg prednisone per day or less, 41 to 80 mg/d, or greater than 80 mg/d was 1.3%, 4.6%, and 18.4%, respectively. Another study of patients with alopecia areata also reported a significant relationship between prednisone dosage and emotional lability.4 However, psychosis, depression, and behavioral changes may occur in patients receiving low-dose or alternate-day steroid therapy.1,5,7⇓⇓ Moreover, the threshold steroid dose for psychiatric disturbances is highly variable among different persons.

In our study, a relationship between prednisone dose and psychiatric adverse events could not be demonstrated. This might be partially explained by the fact that most (>97%) patients received a daily prednisone dose of less than 1.0 mg/kg; therefore, a dose relationship was difficult to establish. In patients who developed psychiatric events, the daily prednisone dose varied widely, from 0.57 to 1.05 mg/kg. One patient developed hypomanic symptoms 24 hours after an increase in daily prednisone dose of 15 mg. The time taken for the development of psychiatric events was also variable, ranging from 1 to 21 days.

Apart from steroid dosage, other factors such as age, sex, premorbid personality, history of psychiatric disorders, or corticosteroid-induced psychiatric side effects were not consistently found to predispose to psychiatric disturbances during each given course of corticosteroid.1 One of our patients had a history of steroid psychosis, but she did not have recurrence of the problem when prednisone was increased again to two-thirds of the previous dosage (at which psychosis occurred). A history of anxiety disorders and active SLE (low C3) were associated with psychiatric adverse effects, but significance could not be reached after multivariate adjustment. Nevertheless, close monitoring is required for SLE patients receiving corticosteroids.

Finally, we demonstrated that corticosteroid-induced psychiatric adverse effects were independently associated with hypoalbuminemia. This finding is consistent with a report that described steroid psychosis in three hypoalbuminemic SLE patients.3 Two had not developed psychiatric symptoms when they received similar dosage of steroids at periods of higher albumin levels. Another study also reported a higher incidence of corticosteroid-related adverse events in patients with serum albumin levels less than 25 g/dL.10 In a hypoalbuminemic state, the free fraction of corticosteroid was increased, which might cause more CNS side effects. However, the exact mechanisms by which hypoalbuminemia may lead to steroid psychosis remain speculative because at high doses, corticosteroid-binding globulin is more important than albumin in the binding of prednisone. Close observation for psychiatric disturbances is necessary for hypoalbuminemic SLE patients receiving corticosteroid therapy, regardless of the dosage and route of administration.