Our study showed that one-third of lupus patients developed serious infections during follow-up. Common bacteria, such as E. coli, Staphylococcus and S. pneumoniae were the most frequent isolates. Several clinical, immunological and therapeutic variables showed association with the risk of suffering major infections. Only three variables, however, had a significant independent effect: lung disease, dose of prednisone and treatment with antimalarials at the study point, the latter showing a strongly protective effect.
Clinical features predisposing SLE patients to infection are not well established. Immunosuppressive drugs have increased the frequency of infections in only a small proportion of studies [4, 8, 9]. Treatment with steroids has been identified as a risk factor in several studies [4, 6, 8, 9], but not in all series [2, 3, 5, 7]. Renal disease, despite being associated with infections in the univariate analysis in some series, has not retained the statistical significance in the multivariant analysis [5, 6]. Only the series of the Hopkins Lupus Cohort found an independent effect of serum creatinine on the rate of hospitalizations due to infection [7]. Likewise, low complement levels have been significant independent predictors in three studies [7, 8, 15]. A SLEDAI score higher than 12 at diagnosis has also been found to increase the risk of infection in one recent study [15].
It is thus a difficult task to identify lupus patients at high risk of serious infectious complications. The inclusion in some studies of patients with both minor and major infections [3, 6, 9] complicates the interpretation of results. The introduction of ever variables and at the time of infection variables, depending on the studies, also adds noise to the conclusions. In addition, many variables (lupus activity, renal disease, low complement, treatment with steroids and immunosuppressive drugs) are intimately related with each other in clinical practice, and thus their relative weight and interactions are difficult to ascertain, even with multivariate analysis.
Our study could somewhat clarify the clinical profile of patients prone to suffering serious infections with the potential to influence the prognosis of SLE. The nested case–control design, matched for age and time to event, reduces the possibility of an immortal treatment bias [12]. We analyzed clinical and immunological variables that could have an effect on immunity (leukopenia, lupus nephritis, drug therapy, complement levels), could reflect lupus activity (SLEDAI score, anti-DNA antibodies, lupus nephritis, lung disease, leukopenia, thrombocytopenia, drug therapy) and/or have known prognostic significance (sex, aPL, lupus nephritis, lung involvement, leukopenia, thrombocytopenia, early damage). Treatment variables were recorded in close temporal relationship with the study point.
In the present study, the prednisone dose at the time of the event had a facilitating effect on infections, in line with data coming from other series [4, 6, 8, 9]. It is noteworthy that the median dose of patients with major infections was only 7.5 mg/day. According to the results of the logistic regression, each increase of 10 mg/day prednisone multiplied by 11 the risk of suffering a serious infection. Every effort should therefore be made to limit both the dose and the time of exposure to steroids of lupus patients, also taking into account the close relation of irreversible damage and prednisone use [16]. On the contrary, no individual immunosuppressive drug or treatment with any of them increased the risk for serious infections. This apparent paradox could be explained by the low numbers of patients on cyclophosphamide, which is usually given in our unit as low-dose pulses, with a lesser frequency of infectious complications [17]. We also excluded minor infections, such as nondisseminated herpes zoster and minor urinary tract infections, that have been associated with immunosuppressive drugs like azathioprine, cyclophosphamide and methotrexate [18, 19]. The association between lung disease and infection in our cohort could be explained by the frequent occurrence of serious respiratory infections in patients with restrictive disease (six out of 14 patients). Noteworthy, complement levels or the SLEDAI score did not modify the risk of major infection of our patients [15].
The most relevant finding of the present study, however, was that patients taking antimalarials were 16 times less likely to suffer a major infection. From the pharmacological point of view, this effect is not surprising. Originally synthesized as antiparasitic agents, these drugs have a wide range of antimicrobial effects [20]. In vitro activity against bacteria (Tropheryma whipplei, S. aureus, Legionella pneumophila, Francisella tularensis, Mycobacterium spp., Salmonella typhi, E. coli, Borrelia burgdorferi), fungi (Histoplasma capsulatum, Cryptococcus neoformans, Aspergillus fumigatus) and viruses (including human immunodeficiency virus) has been reported [20]. Antimalarials interfere with invasion and internalization of E. coli into host cells [21]. Clinical studies have established the combination doxycycline plus hydroxychloroquine (HCQ) as the standard treatment for chronic Q fever [20]. Likewise, combination treatment with HCQ, OH carbamide and didanosine reduces the human immunodeficiency virus load, having been proposed as a cheaper alternative for the treatment of AIDS [22].
The antibacterial effects of antimalarials are mediated by pH-dependent iron deprivation and by increasing the phagolysosomal pH, both inhibiting the growth of intracellular organisms [20]. Likewise, the antiviral effects are related to increasing the lysosomal pH, which disrupts hydrolases and inhibit the post-translational modification of newly synthesized proteins [23].
Two previous studies in patients with lupus suggest the protective effect of antimalarials against infection. Sisó and colleagues, in a retrospective cohort study of patients with lupus nephritis, found a lower frequency of infections among those previously treated with antimalarials [24]. Bultink and colleagues, in a study designed to analyze the effect of the deficiency of functional mannose-binding lectin, found that treatment with HCQ was protective of major infections, with an adjusted OR (OR = 0.05, 95% CI = 0.01 to 0.23) very similar to that obtained in the present study [5]. The authors attributed this association, however, to the presumed lesser severity of lupus in patients taking HCQ. In addition, in 1996 Podrebarac and colleagues reported two cases of P. carinii pneumonia in patients with SLE treated with high-dose steroids shortly after discontinuation of HCQ [25]. We believe that our study confirms the antimicrobial effects of antimalarials, which can be actually an additional reason for the increased survival seen in SLE patients taking these drugs [24, 26, 27].
Our study has the main limitation of the observational design – the specific treatment of each patient was therefore not randomized, but rather decided on clinical grounds. In fact, patients given antimalarials were younger and had less severe disease, with a lower frequency of nephritis, leukopenia and early irreversible damage. When these adjustment variables were added to the regression model, however, the protective effect of antimalarials on the occurrence of major infections remained unchanged.