CV Risk in RA: Is a New Measurement Tool the Answer? Save

A new expanded cardiovascular (CV) risk calculator for use in patients with rheumatoid arthritis (RA) has been developed to take into account disease-specific factors that can worsen risk, researchers reported.

Compared with a base risk model limited to traditional cardiovascular risk factors, the expanded model showed significantly greater discrimination, with a change in the c-statistic (area under the receiver operating characteristic curve) from 0.7261 to 0.7609,P=0.0104, according to Daniel H. Solomon, MD, of Brigham and Women's Hospital in Boston, and colleagues.

And in a version of the expanded model that used the Framingham Risk Score categories for 10-year risk of a CV event, risk was correctly classified in 17% more patients than with the base model, the researchers reported in the Augus tArthritis & Rheumatology.

"Risk prediction scores (or rules) provide an important tool for clinicians and patients to move epidemiologic observations into the clinic," they wrote. However, studies have suggested that commonly used prediction tools such as the Framingham Risk Score and the Reynolds Risk Score underestimate the risk in patients with RA, who have a higher likelihood of CV events for various reasons including systemic inflammation and medication use.

Therefore, the researchers developed and internally validated a new tool based on data from the Consortium of Rheumatology Researchers of North America (CORRONA) cohort, a U.S. registry that includes 40,000 patients.

"This is an important contribution to the assessment of cardiovascular disease in these patients," said Eric Matteson, MD, chair of rheumatology at the Mayo Clinic in Rochester, Minn., who was not involved in the project.

It also highlights a potential gap in patient care. "I think the topic [of CV risk assessment] is moderately interesting to most rheumatologists. I say moderately as most are aware of the CV risk, yet few do much about it," said John J. Cush, MD, director of clinical rheumatology at Baylor Research Institute in Dallas, who also was not involved.

"Only a minority of rheumatologists monitor lipids, and fewer will manage abnormal values or manage hypertension and blood pressure issues," Cush toldMedPage Today.

The full study cohort of 23,605 patients was divided into a derivation cohort, consisting of two-thirds of the patients, and a validation cohort, made up by the remaining one-third.

At study entry, patients' potential risk predictors were assessed, including the traditional risk factors of age, gender, hypertension, diabetes, dyslipidemia, and smoking.

Disease-specific variables thought to possibly influence risk also were obtained, including disease activity as measured on the Clinical Disease Activity Index (CDAI), disability according to the Health Assessment Questionnaire disability index (HAQ-DI), disease duration, serologic status, joint erosions and subcutaneous nodules, and use of medications including steroids, methotrexate, and nonsteroidal anti-inflammatory drugs.

In the derivation cohort, a regression model initially included only the base (traditional) risk factors, with all other variables then being added. Those found to be significant were included in the fully adjusted expanded model, which was then tested in the validation cohort.

A net reclassification index was calculated according to the categories used in the Framingham Risk Score, with 10-year risk of having a CV event of less than 5%, 5-10%, 10-20%, and more than 20%, and also according to the cutoff of 7.5% recommended more recently by the American College of Cardiology/American Heart Association (ACC/AHA).

Disease-specific variables that were included in the final model were:

• Moderate or high disease activity versus low or remission on the CDAI, HR 1.31 (95% CI 1.07-1.61)
• Moderate or high disability versus low or none on the HAQ-DI, HR 1.18 (95% CI 0.95-1.46)
• Daily prednisone use versus non-use, HR 1.61 (95% CI 1.33-1.95)
• Disease duration longer than 10 years, HR 1.43 (95% CI 1.18-1.73)

Patients' mean age was 57, and most were women. A total of 29% were hypertensive, 15% were smokers, 15% had dyslipidemia, and 7% had diabetes.

During follow-up of 2.9 years, the incidence rates were 2.5 per 1,000 person-years for myocardial infarction, 3 per 1,000 for stroke, and 1 per 1,000 for cardiovascular-related death.

Using the ACC/AHA cutoff, the risk was classified correctly in 10% more patients than the base model, according to Solomon and colleagues.

The accuracy of the model also was assessed according to the Akaike's information criterion test of the quality of fit of the model, and the Hosmer-Lemeshow goodness-of-fit test, and both showed positive results.

The researchers then offered two hypothetical examples of risk calculation. One was a 55-year-old woman who has had rheumatoid arthritis for more than 10 years, has hypertension, has moderate disease activity and a HAQ-DI above 0.5, and is being treated with steroids. She does not have diabetes or dyslipidemia, and is not a smoker.

Her 10-year probability, according to the expanded risk tool, is 9.3%. "This exceeds the current ACC/AHA recommended threshold for starting a statin, which is a 10-year risk probability of 7.5%," the researchers wrote.

The second example was a 50-year-old man whose disease duration is less than 10 years, who has hypertension, low disease activity, and a HAQ-DI above 0.5, but does not have diabetes or dyslipidemia and is not a smoker.

His probability was only 5%, and therefore did not reach the statin threshold.

In an editorial accompanying the article, Deborah P.M. Symmons, MD, of the Arthritis Research UK Center for Epidemiology at the University of Manchester in England, argued, "it is unclear whether there is a need for disease-specific [atherosclerotic cardiovascular disease] risk calculators for patients with RA. We already advocate minimizing disease activity in all RA patients. This in itself should reduce CV risk."

She continued, "We may serve our patients better by assessing [traditional risk factors] in the context of proven general population risk calculators ... using readily available measures rather than requiring additional tests (such as measures of disease activity and autoantibody tests) before the risk of [atherosclerotic cardiovascular disease] can be calculated and addressed."

Matteson also advised some caution about the expanded risk tool. "As with all new tools, external validation will be important to confirm its validity and generalizability before applying it clinically. In the daily care of our patients, the focus should remain on optimal management of disease and risk factors to reduce the risk of cardiovascular disease."

Solomon and colleagues disclosued financial relationships with CORRONA, Amgen, Pfizer, Lilly, AstraZeneca, Celegene, Novartis, Bristol-Myers Squibb, Crescendo, AbbVie, UCB and Janssen.

This piece is brought to RheumNow readers by our friends at MedPage Today, where it was originally published on July 31, 2015.