Welcome to a Rheumatologist's blog!Welcome to a Rheumatologist's blog!

With this post I begin a rheumatology blog that I call "A Rheumatologist's Blog". Its purpose is to present for discussion key issues in rheumatology. I welcome my colleagues and others to join in these discussions. You can do do this by joining the site and adding your comments.

I welcome suggestions, that can come to me by email, about other subjects, and I certainly welcome your written views. I hope this site can be an open forum for discussion.

What (if any) is the value for a physician caring for an individual patient with RA to identify predictors of responses to a therapy in patient groups?

Miguel Belmonte, a Spanish rheumatologist, called to my attention a recent report in Rheumatology  entitled “Rheumatoid factor positivity rather than anti-CCP positivity, a lower disability and a lower number of anti-TNF agents failed are associated with response to rituximab in rheumatoid arthritis” (Quartuccio L, et al. Rheumatology 2009; 48: 1557-9). This paper is consistent with a genre in which the authors try to help clinicians to predict who will respond to a given therapy.

The authors found that ACR50 responses were predicted significantly by lower HAQ score, lower number of failed anti-TNF agents, and rheumatoid factor positivity. In a multivariate analysis, only rheumatoid factor positivity was correlated significantly with EULAR moderate-to-good responses.

Several matters about this report are of concern to me:

1) The title implies that anti-CCP positivity, lower disability and lower number of anti-TNF agents are not associated with response to rituximab, but in fact they are highly significantly associated with ACR50 responses, but not with EULAR moderate to good responses in a multivariate regression model. It seems that well-meaning authors do not interpret multivariate regression correctly, in my view. A regression which includes any two variables that are collinear, such as rheumatoid factor positivity and anti-CCP positivity (71% of all patients were positive for both antibodies, and less than 25% were discordant), one of these variables will not be significant in a multivariate regression. However, that does not mean that the other variables are truly not significant, as suggested in the title.

2) The distinction between ACR50 responses and EULAR moderate to good responses is somewhat artefactual, and would not justify a sweeping title.

3) Even if the data were solid, the inference that data collected in groups of patients in research studies can be useful to a clinician in the care of an individual patient appears incorrect to me. To be sure, there are situations in which the prognosis of a poor response is clinically useful – most obvious for antibiotics with NO (zero) anti-microbial activity versus a given pathogen. But in this study, the group with the least likelihood of a response, patients who were negative for both rheumatoid factor and anti-CCP, had a response rate of 21%. Isn’t a 1 in 5 likelihood of a response in a patient who has failed several other biological agents not reasonable?

I would treat such a patient, and would be concerned that reimbursement authorities might use data from groups to potentially deny treatment to people who really need it.

All comments, critiques and suggestions are welcome by Email, about the above and any other rheumatology subject. I hope this site can be an open forum for candid and open discussion to improve patient care.

/* -1&&m>28){j=28+s;s='';if(j/g,'>');l[i].href='mailto:'+t.value}}catch(e){}}}catch(e){}})(document);/* ]]> */


Commenting: Dr. Frederick Wolfe
03/16/2010 - 12:45

Ted, I think your 1st point is quite correct. It is not a question of whether Rheumatoid Factor or CCP is the test we should use, but rather, how much better one test is than another. One way to have approached this question is to have formed a group of 4 categories and used the group in the regression analysis: RF(-) CCP(-), RF(+) CCP(-), RF(-) CCP(+), and RF(+) CCP(+). That would have allowed a better understanding of the contribution of these tests, and understand if they actually differed in their prognosis.

I don’t think that you are right when you doubt that group data can be used to guide individual decisions. In fact, the 1 in 5 likelihood that seems attractive to you comes from group data. Perhaps the issue is cost-effectiveness and exposure to adverse effects. Why would we not use a treatment that is free and without side effects? But that is not usually the case. Would you still recommend the treatment if the probability of success was 5%? 2?

Fred Wolfe

Commenting: Dr. Frederick Wolfe
03/16/2010 - 12:54

Ted, you have been the leading advocate of patient questionnaires in the US (probably in world). Yet it seems that rheumatologists don’t use questionnaires in patient care as often as you or I would like them to. Perhaps you might comment as to why this is the case – if you agree with me. I can’t think of practicing without patient questionnaires. What’s the problem?

Add comment