Analysis of the baseline data from the APPLE Trial, a racially and ethnically diverse multicenter prospective cohort of pediatric SLE patients, reveals that both traditional and non-traditional risk factors are associated with increased CIMT in children with SLE. Traditional risk factors predictive of increased CIMT in this cohort include increased BMI, male sex and increasing age. Perhaps more intriguing is the identification of nontraditional risk factors including treatment with azathioprine and weight-adjusted prednisone dose.
The frequent and chronic use of glucocorticoids in children and adolescents with SLE likely plays a role in the prevalence of dyslipidemia and atherosclerosis, but this has not been well characterized to date. Some studies in adult SLE suggest an association between prednisone use and both increased total cholesterol levels and CIMT (32
). In contrast, other studies in adults with SLE suggest that those with carotid plaque received less aggressive immunosuppressive therapy (including less prednisone and cyclophosphamide) than those without plaque, suggesting a potentially protective effect of immunosuppressive therapy on the development of atherosclerosis, possibly due to better control of underlying inflammatory disease (6
). In the current study, moderate dose prednisone is associated with decreased mean-max CIMT, while higher or lower dose prednisone may predict increased mean-mean or mean-max CIMT, respectively. The demonstration of different relationships between prednisone dose and CIMT may suggest steroids are affecting more than one pathway in atherogenesis. At high doses, prednisone may increase CIMT by increasing traditional risk factors, such as cholesterol, LDL, and BMI. In contrast, with current understanding of the importance of inflammatory mechanisms in the pathogenesis of atherosclerosis (35
), lower dose prednisone may be associated with higher CIMT due the presence of ongoing active inflammatory disease. Given the cross-sectional nature of the data, it was not possible to assess the association between cumulative steroid dose and CIMT. However, this will be assessed in longitudinal analysis of the APPLE Trial. The differences in CIMT based on steroid dose may provide insight into previous conflicting results concerning the impact of oral prednisone on cardiovascular risk and suggest that there may be a ”therapeutic window” concerning cardiovascular toxicity in this population.
Other medications used in the treatment of SLE may impact the development of atherosclerosis. For example, cyclosporine contributes to dyslipidemia (36
), while hydroxychloroquine improves lipid profiles and is associated with decreased cardiovascular events (38
). Azathioprine, mycophenolate mofetil, and cyclophosphamide have not previously been shown to have significant impact on lipid profiles or atherogenesis (38
). A surprising finding in this study was that azathioprine was associated with increased CIMT, unlike the other immunosuppressants. This appears to be an effect independent of other variables including SLE disease activity and history of nephritis and not explained by collinearity with other variables. To further understand the relationship between azathioprine and CIMT, the 30 subjects taking azathioprine were compared to the 190 patients not taking azathioprine. As noted in , subjects on azathioprine were on average one year older (mean age 16.7 years vs. 15.6 years; p=0.027), had slightly longer disease duration (median duration 31.5 months vs. 21.5 months; p=0.005), and were more like to have a SLICC score > 0 (43.3% vs. 24.1%; p =0.027). Upon removing azathioprine from multivariable modeling, duration of SLE became statistically significant but SLICC score did not, suggesting a possible confounding effect due to relationship between azathioprine use and disease duration. However, when both variables (azathioprine and disease duration) were included in the multivariable modeling, only azathioprine remained significant suggesting azathioprine has a stronger association with CIMT. In contrast, age was a significant risk factor for higher CIMT even when azathioprine was included in the multivariable model for mean-max CIMT, suggesting both age and azathioprine have independent effects on CIMT. Given the relative affordability of azathioprine, additional analyses to explore whether azathioprine use reflected socioeconomic status revealed no correlation between the use of azathioprine and household income or parental education (data not shown).
Comparison of subjects taking azathioprine (N=30) and not taking azathioprine (N=190).
While the potential pro-atherogenic effects of azathioprine are not known, an interesting in vitro
observation using human umbilical vein endothelial cells showed that azathioprine and its active metabolite, 6-mercaptopurine, caused dose-dependent decreases in endothelial cell proliferation and altered endothelial nucleotide balance by reducing intracellular ATP and GTP (39
). It is possible that azathioprine has a direct deleterious impact on endothelial cell function and the capacity for endothelial repair in SLE. In addition, in vitro
studies suggest that azathioprine may increase homocysteine levels; however, we did not find a significant relationship between CIMT and homocysteine levels in this analysis (40
). It is interesting to note that azathioprine was found to be a contributing predictor of vascular events in a racially/ethnically diverse cohort of adults with SLE in the multicenter LUMINA (Lupus in Minorities: Nature Versus Nurture) cohort (41
). Additional study is necessary to further explore this potentially significant association.
Creatinine clearance may be increased in settings of glomerular hyperfiltration, such as proteinuria or hypertension (31
). Although patients with significant renal insufficiency or nephrotic syndrome were excluded from the APPLE Trial, many participants had proteinuria and hypertension. With univariable analysis, both proteinuria and creatinine clearance were associated with CIMT, but only creatinine clearance remained significant in the multivariable model. It is possible that creatinine clearance and proteinuria are correlated; however, longitudinal study of this cohort will be necessary to further evaluate this effect.
In this study, CIMT was assessed using two different approaches—mean-mean common CIMT and mean-max CIMT with a correlation of 0.78 between the two outcomes, suggesting that the two outcomes are correlated but are not perfectly linearly related. The two approaches differ in the segments of the carotid artery measured, as well as how the thickness is assessed (average vs. maximal thickness of the measured segments). Both have been used in prior clinical trials and review of the literature reveals no clear superiority of one method over the other in predicting atherosclerotic risk (43
). In multiple regression models, we identified factors (sex and medication use) that were significantly associated with both outcomes, and others that were significantly associated with only one outcome (age and Lp(a) were related to mean-max CIMT only, while BMI and creatinine clearance were related to mean-mean common CIMT only). Segment-specific risk factor associations have been reported previously in IMT studies and are generally considered to reflect the influence on focal thickening of local phenomenon such as sheer stress and arterial remodeling (45
). Interestingly, Espeland et al. (47
) reported segment-specific risk factor associations in coronary artery disease cases, but not in controls of similar age. The observation that obesity is more closely related to common CIMT than to bifurcation or internal CIMT is consistent with prior reports in adults (46
). Associations between estimated creatinine clearance and common CIMT have previously been reported in children with chronic renal failure (48
) and in young adults with myocardial infarction (49
). Reports from children and adolescents with growth hormone deficiency (50
) and from adult women with SLE and cardiovascular disease (51
) have also documented significant increases for both Lp(a) and common CIMT relative to normal controls. Segment-specific differences in associations between CIMT and either creatinine clearance or Lp(a) have not been reported previously.
There are several limitations to this study. The sample size was only powered to detect difference in the primary outcome of the APPLE Trial – progression rate of CIMT. Thus it may not have sufficient power to identify all meaningful clinical associations of risk factors with baseline CIMT. However, this is the largest cohort of pediatric SLE patients ever examined for cardiovascular risk and the potential relationships reported here are hypothesis generating and can be confirmed by future studies. In addition, as in any clinic trial, there is an inherent selection bias such that the APPLE cohort may not be completely representative of pediatric lupus patients as a whole. For example, obese patients were more likely to be excluded due the inability to obtain high quality CIMT images. Patients with nephrotic range proteinuria, very elevated cholesterol or renal insufficiency, also excluded from the APPLE study, are known to have increased CIMT (5
), These subsets of pediatric SLE patients may have increased CIMT and increased lipid abnormalities and different relationships between the risk factors and CIMT. Thus, their exclusion may affect the generalizability of the reported findings.
The presented models only explain approximately 21% of the variation in CIMT suggesting other measured or unmeasured risk factors not identified in the current study. Cross sectional analyses of the baseline data also limits the ability to draw conclusions about causation between risk factors and CIMT outcomes. Once the APPLE Trial is complete, longitudinal analysis will provide more meaningful assessment of the impact of glucocorticoids, azathioprine and other factors on CIMT. Unfortunately, the APPLE study does not include matched normal controls. Because of the inherent variation in CIMT measurement techniques, the absolute thickness can not be compared between studies. Therefore, there is no potential to compare the CIMT results in this cohort with a healthy population.
In summary, traditional as well as non-traditional risk factors predict increased CIMT in pediatric patients with SLE. Of interest, treatment with azathioprine was associated with increased CIMT and the relationship with weight-adjusted oral prednisone dose may not be linear. Future longitudinal analysis of data from the APPLE Trial will further clarify the role of medications such as azathioprine and corticosteroids, as well as other potential risk factors of CIMT progression. Understanding risk factors in the development of atherosclerosis in children and adolescents with SLE will inform evidence-based treatment recommendations to avoid irreversible cardiovascular damage.