Article Text

Through the looking glass of rheumatoid arthritis to study inflammation and high-density lipoprotein
  1. Katherine P Liao
  1. Correspondence to Dr Katherine P Liao, Division of Rheumatology, Immunology, and Allergy, Harvard Medical School, Brigham and Women's Hospital, 60 Fenwood Road, Boston, MA 02115, USA; kliao{at}

Statistics from

Destruction of bone and soft tissue in rheumatoid arthritis (RA) occurs as a result of unchecked systemic inflammation. While swollen and deformed joints are easy to see, less obvious to the naked eye is the impact of this same inflammatory process on the cardiovascular (CV) system. Cardiovascular disease (CVD) is the leading cause of death in patients with RA. The mortality rate from CVD in RA is approximately 1.5 times that of individuals from the general population with the same age, sex and CV risk factors.1 The excess CV risk has been attributed to inflammation; however, the pathways and mechanisms linking inflammation to CV events are less clear. Lipids, specifically low-density lipoprotein cholesterol (LDL-C), are considered part of the causal pathway for atherosclerosis and CVD in the general population. However, the established relationships between LDL-C and CV risk are altered in RA. As expected, elevated levels of LDL-C are associated with increased CV risk in RA; however, low LDL-C is also associated with increased CV risk.2 Subjects with RA who experience a reduction in inflammation, considered beneficial for CV risk, have concomitant increases in LDL-C,3 suggesting the opposite effect of increased CV risk. Post hoc studies of lipids from randomised controlled trials of RA therapies have shown that increases in LDL-C occur across different classes of RA therapies.4 Together, these data demonstrate that the relationship between LDL-C, inflammation and increased CV risk, observed in the general population, becomes uncoupled in RA. A similar uncoupling was observed with total cholesterol (TC). Interestingly, the departure from established relationships was not as apparent when examining high-density lipoprotein (HDL-C) levels and CV risk in RA.

Recent studies have highlighted that HDL function is an independent predictor of incident CV events, even after adjusting for HDL-C levels.5 This HDL function, as measured by cholesterol efflux capacity, is considered to be stable in the general population. While the expected relationship of higher HDL-C levels and reduced CV risk is preserved in RA, cholesterol efflux capacity is impaired in RA subjects. Furthermore, a reduction in inflammation is associated with significant improvements in cholesterol efflux capacity,3 suggesting that HDL dysfunction may explain some of the excess CV risk in RA. Studying HDL function is a focus of the study presented by O'Neill et al6 in this journal.

Using RA as a model to study the relationship between inflammation and lipids is ideal for several reasons. First, subjects with RA have higher absolute levels of inflammation than the general population. In this study, the median C-reactive protein (CRP) was 22.7 mg/L in RA compared with 1.0 mg/L in healthy controls. Second, since individuals with RA have higher levels of inflammation, they also experience larger swings in levels of inflammation through response to therapy or during RA flares. An illustrative example of CRP levels for an individual in the first year of RA diagnosis is shown in figure 1. A fluctuation of 1.0 mg/L, with values ranging from 1.0 mg to 2.0 mg/L, shown in the control patient is barely noticeable compared with changes in CRP of ≥10 mg/L typically observed during an RA flare. The larger magnitude of inflammation and magnitude of change provide the power to detect associations that may not otherwise be observed in the general population. However, we know that inflammation, even in the range of 3–10 mg/L, has a significant impact on CV risk in the general population.

Figure 1

Illustrative example showing typical C-reactive protein (CRP) levels for a patient with rheumatoid arthritis (RA) in the first year after diagnosis, reflecting successful treatment and one RA flare, compared with CRP levels observed in a patient without inflammatory disease (CRP range from 1.0 to 2.0 mg/L).

Examining the typical course of RA through a different lens, RA flares and subsequent treatment can serve as an opportunity to study lipids in a natural experiment. Lipid biochemical parameters could be studied in a model of elevated inflammation during a flare, and the same parameters could be compared in the same subject when inflammation is controlled with RA therapy. This is precisely the paradigm used in the study presented by O'Neill et al.

O'Neill et al present data from a small but important randomised controlled trial examining the link between inflammation and HDL function. In their study, 18 RA subjects with active and erosive disease were randomised to receive either infliximab, a tumour necrosis factor inhibitor (TNFi) or placebo. All subjects were already on the first-line therapy for RA, methotrexate (MTX). Based on the baseline CRP levels and DAS28, the RA subjects would be classified as having an RA flare (figure 1). Subjects were followed for 54 weeks for the main analysis. Lipids, including TC, HDL-C, LDL-C, triglycerides and glucose were measured at baseline and at 54 weeks. As expected, subjects who received infliximab in addition to MTX experienced a significant reduction in CRP from 26.8 to 2.3 mg/L. Consistent with previous studies, subjects who received infliximab also had increases in TC (24.4%), HDL-C (30.0%) and LDL-C (25.8%).

The properties of HDL assayed in this study include cholesterol efflux and anti-oxidation, measured using cholesterol efflux capacity, nitric oxide (NO) bioavailability, superoxide (SO) production and paraoxonase-1 (PON-1) activity. At 54 weeks, subjects who received infliximab (n=11, had a significant increase in NO bioavailability and a reduction in SO production. While no significant change was observed for PON-1 activity and cholesterol efflux capacity, the mean values were overall improved at 54 weeks. In comparison, subjects in the placebo group (n=7) at 54 weeks had a significant reduction in SO production and no changes in the other three measurements.

Based on these data, the authors postulated that anti-inflammatory treatments can improve the antioxidant and endothelial protective properties of HDL. Furthermore, that the lack of significant changes in cholesterol efflux capacity before and after therapy suggests that cholesterol efflux may be less sensitive to changes in inflammation. One potential reason for the lack of association not discussed by the authors is variation in response to therapy. In a recent study examining net cholesterol efflux in RA subjects starting biological therapies, no significant difference was observed in cholesterol efflux in subjects before and after 6 months of therapy.7 However, a significant inverse association was observed between the change in RA disease activity and change in cholesterol efflux capacity. These findings corroborate a previous study which found that a reduction in CRP was associated with significant improvement in cholesterol efflux capacity.3 It would have been interesting to know the correlations between CRP or DAS28 and measures of HDL function in the study by O'Neill et al. Since not all RA subjects respond to infliximab, using CRP or DAS28 can allow for a more direct measure of the impact of changes in inflammation on HDL function. Sample size may also be an issue since the study was designed to detect a change in NO bioavailability and not cholesterol efflux capacity. Lastly, as mentioned by the authors, the measurement of cholesterol efflux can vary and differences in methods can lead to variation in findings.

In the concluding paragraphs, the authors discuss that MTX monotherapy alone may not be sufficient immune suppression to reduce CV risk in the general population. One hypothesis, currently being tested in separate randomised controlled trials, is whether reducing inflammation reduces CV risk. CIRT8 and CANTOS9 are testing this hypothesis in the general population, while TARGET10 is testing this hypothesis in RA. TARGET is also designed to compare CV risk reduction across different therapies. If we believe that the majority of CV benefit is derived from controlling inflammation, then it is possible that improvements in HDL function would also be seen in MTX responders, which were not part of this study. MTX is the first-line agent for RA because it is effective in controlling inflammation, with a subset individuals achieving remission on MTX alone. Knowing TNFis may have additional benefits, requires a comparison between TNFi and MTX-treated subjects, controlling for changes in inflammation. This study included subjects with established RA, who were not responding to MTX monotherapy. Based on the design of the study, it is not possible to disentangle whether the improvements in HDL function were from reducing inflammation, adding TNFi or both. If the main benefit for CV risk reduction is derived from reducing inflammation, then several options for anti-inflammatory agents with a favourable safety profile are available. Moreover, since the magnitude of inflammation in the general population is less than RA, potent immunosuppressive agents may not be required.

Finally, I agree with the authors' closing remarks that studying HDL function may provide insight into why recent therapies which raise HDL-C levels did not show CV benefit. What we have learned from RA is that in the setting of inflammation, routine lipid levels are suboptimal markers of CV risk. Studies in RA are underway to examine whether advanced lipoprotein measures such as HDL function and particle size may correlate better with CV risk than routine lipids levels alone. It is likely that for RA and other inflammatory diseases, we will need measures beyond routine lipids to accurately assess CV risk. Findings from these studies can serve as a magnifying glass to understand the impact of inflammation on lipids and CV risk, which in turn may inform the aspects of CV risk stemming from inflammation in the general population.

View Abstract


  • Contributors KPL drafted the editorial and takes full responsibility for its content.

  • Funding KPL is funded by the K08 AR 060257, R01 HL127118, and the Harold and Duval Bowen Fund.

  • Competing interests None declared.

  • Provenance and peer review Commissioned; externally peer reviewed.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles