Jeffrey Sparks, MD, and Kazuki Yoshida, MD, on the Impact of Passive Smoking on Development of RA

Drs Jeffrey Sparks and Kazuki Yoshida discuss their research into the impact of passive smoking throughout the life course of patients who develop rheumatoid arthritis in adulthood.

Jeffrey Sparks, MD, is a rheumatologist and research scientist at Brigham and Women's Hospital and an assistant professor of medicine at Harvard Medical School in Boston, Massachusetts. Kazuki Yoshida, MD, is a rheumatologist and research scientist at Brigham and Women's Hospital.

Rebecca Mashaw:  Good afternoon, everyone. Welcome to another podcast from the Rheumatology & Arthritis Learning Network. I'm your host, Rebecca Mashaw. I'm delighted to have here today, Drs. Kazuki Yoshida and Jeffrey Sparks, who are both at Brigham and Women's Hospital in Boston and Harvard Medical School.

They're going to be talking about a study they conducted about the risk of developing rheumatoid arthritis in adulthood of patients who have been exposed to passive smoking throughout their life course. I'm going to turn it over to the experts now and let them talk to us about this very interesting subject.

Dr. Jeffrey Sparks:  My name is Jeff Sparks, one of the rheumatologists and investigators here at Brigham and Women's Hospital. Dr. Kazuki Yoshida and I teamed up on this very interesting study, and we're happy to chat more about it. Kazuki, I wonder if you could just tell me a bit about the background and reason that we did this study?

Dr. Kazuki Yoshida:  Jeff thanks for the introduction, and Rebecca for the opportunity. I'm Kazuki Yoshida. I'm an associate rheumatologist at the Brigham.

In this study, titled passive smoking throughout the life course and the risk of incidental rheumatoid arthritis, in adult for the man and women, we examined the potential direct link of passive smoking exposures, and incidentally development of rheumatoid arthritis among women using the Nurses' Health Study II.

Specifically, we studied three types of passive smoking exposures. One is maternal smoking during pregnancy, meaning that pregnancy of participants for the study. Then the second exposure we studied was the childhood exposure to parental smoking. Then the third type of passive smoking was adult passive smoking exposure through living with a smoker after age 18.

Then we studied development of rheumatoid arthritis overall, as well as by seropositive. Presence of rheumatoid factor and/or CCP antibody or absence of it.

Dr. Sparks:  Now, I'll add that there's been a lot of intense interest in the mucosal paradigm of RA pathogenesis, whereby inflammation at the mucosal surfaces, including the lungs, might be a trigger to developing antibodies or other events that could precede the clinical onset of disease by many years even.

Dr. Yoshida:  Important background here is that the personal smoking of participants themselves are strongly associated with development of rheumatoid arthritis. It is one of the most well-established environmental risk factors of rheumatoid arthritis development.

However, the past data on passive smoking, being exposed to smoking through passive exposure, is less well-established. The question we wanted to ask in this study was whether passive smoking exposure is directly linked to development of rheumatoid arthritis when we account for the personal smoking exposure of these participants in the Nurses' Health Study.

Dr. Sparks:  Before we get to the results, maybe we could talk a bit about the Nurses' Health Study II and about some of the strengths that really made this analysis possible.

Dr. Yoshida:  Nurses' Health Study II is a longitudinal cohort of almost 100,000 participants who were female nurses at the time of enrollment, which started in 1989 and still going on to the present day.

The strengths of the cohort we are using, Nurses' Health Study II, is that it's a longitudinal cohort that has been collecting information every 2 years since they started the cohort in 1989. It has strong data both on the exposure of passive smoking as well as development of rheumatoid arthritis.

On the passive smoking side of the information, we have 3 types of exposure—maternal smoking during pregnancy, childhood parental smoking, and adult passive smoking. These were their exposures of interest. The important thing that we had to factor in is the personal smoking of these participants.

In the Nurses' Health Study II, we had a smoking recorded as passive smoking every 2 years for these participants, now enabling us to control for this important known risk factor. Then, on the outcome side of that study, we wanted to examine the development of rheumatoid arthritis.

The Nurses' Health Study II cohort the rheumatoid arthritis, when it was reported by the participant, the participant then received an additional questionnaire, and then we requested additional medical records that were reviewed for the confirmation of diagnosis of rheumatoid arthritis as well as seropositivity.

Dr. Sparks:  Just to expound on that, the sample size is really large. This one was 91,000 women, I think, and they're followed throughout their entire adulthood. This survey extended some of the exposures back even to when they're in the uteroo, childhood, late adolescence, early adulthood, and then all the way up to these women, now at retirement age.

Then confirming the cases of RA, not just by self-report, but actually tracking on their medical records to make sure that they meet research classification for RA and also getting the date of diagnosis and the antibodies that are sent during clinical follow-up.

There's very few data sets that are around that could interrogate something that happens throughout the entire life course to understand how it might affect adulthood onset rheumatoid arthritis. It's been really fun analyzing this, and Kazuki has taken it to definitely different levels. Kazuki, can you tell us the high points about the results? What did we find?

Dr. Yoshida:  Among the 3 exposures we examined, passive smoking exposures, we found that childhood parental smoking was directly associated with adult incidental rheumatoid arthritis, specifically, seropositive rheumatoid arthritis. That association of potential direct impact in terms of hazard ratio was a 75% increase. That was the key takeaway from our study.

In comparison, the maternal smoking during pregnancy was not associated with adult incidental rheumatoid arthritis study. In the examination of potential direct impact of adult passive smoking, we did see some trend toward association, but we not find a significant direct association between adult passive smoking and incidental rheumatoid arthritis study.

Dr. Sparks:  For those who are really interested, Kazuki is a Japanese-trained rheumatologist, but also came here to get a PhD in both epidemiology and biostatistics at the Harvard School of Public Health, so a real content expert.

We did something called life course epidemiology that also handled how the variables can sometimes confound, but also sometimes mediate, the relationships that we were looking for. Kazuki, that's my interpretation as a physician-scientist that delves into clinical research, but maybe you can give a little more detail about some of the statistical analyses that made this analysis so special.

Dr. Yoshida:  We adopted the life course epidemiology analysis framework. Also, this framework tries to analyze the impact of earlier life exposure when we have to consider later life events that may be confound or mediate association of the earlier exposure to later life outcome. In specific case, our earlier life exposure of interest was, let's say, childhood parental smoking.

Then there was a certain level of personal smoking update later in their life, then our outcome of interest in incidental rheumatoid arthritis was apparently happening in adult life. What we wanted to specifically study was the potential direct association of early-life passive smoking with incidental rheumatoid arthritis.

To account for that personal smoking update, which is known to be higher among children who are exposed to passive smoking, we used the marginal structure model approach to estimate what is called a control direct effect, effect of earlier life exposure directly linked to outcome when we control for the later life exposure that was partly acting as a mediator.

Dr. Sparks:  Is it fair to say that the association that we found of childhood smoking with seropositive RA in adulthood is not explained by the participants' personal smoking?

Dr. Yoshida:  To the extent, our analysis was able to do. What we did was actually to do that control direct effect margin structure modal analysis, as well as a regular time-varying covariate modeling.

Then, interestingly, the potential direct link was only statistically significant in the direct effect marginal structure model, where at the regular time-varying covariate model did not find the significant association, although there was a trend. In terms of the strength of data, our data, not the Nurses' Health Study II cohort, has a very detailed information about lifestyle factors.

For example, diet, BMI, and other information that may be linked to increased risk of rheumatoid arthritis so that extent of adjustment through statistical modeling was very strong with this data. Although that said, this is still an observational study. Hopefully, we don't have a randomized study...of passive smoking exposure, but that limitation of the observational study that we can only adjust for things we measured and had information on still remains.

Dr. Sparks:  Are there other limitations of the analysis?

Dr. Yoshida:  The Nurses' Health Study is a very rich data set, having information on various aspects of life on this participant, but the selection of participants is a little specific. These were female nurses at the time of enrollment.

Then the reason for choosing these participants is that they are very reliable in terms of reporting their conditions and other lifestyle factors to the study with a greater than 90% follow-up rate. That also restricts the potential scope of the study in terms of generalizability that a very specific subset of the population. Although in terms of US geographic region, it's very varied across the United States.

Dr. Sparks:  How about the implications for the findings? What do you think of research related to seeing patients? How does this help us?

Dr. Yoshida:  In terms of the potential biological interpretation, the interesting part here is that potential mucosal exposure to smoke through passive smoking early in childhood may be an early precursor of immune dysregulation which might further develop into an autoimmune condition later in life so that the potential interpretation of finding about that direct impact.

That's probably the most important part in terms of biological implications. In terms of public health importance, smoking is probably a known risk factor for multiple things. Also, smoking being bad may not be novel by itself.

The important thing here is that it is not just the personal smoking acting on these people themselves, but that passive smoking exposure may then impact the life of their offspring, of these smokers who may later on in their respective life have autoimmune condition, potentially triggered by earlier passive smoking exposure.

The public health implication is, it's not just about smokers themselves, but could be their offspring who might be affected by smoking.

Dr. Sparks:  To add some more speculation, to me, it sounds like this could be a first trigger, perhaps in some people who have genetics that might predispose them to RA. This statistical framework model is very well for this because, unlike the traditional way of just adjusting for personal smoking, this allows second hits to be part of the causal framework.

You could certainly see something related to maybe passive smoking as a first trigger, maybe active smoking as a second trigger, maybe it's environmental factors, maybe it's stress or infection such as respiratory infections.

It is something that's intriguing, that something that happens so early in life could predispose something that affects our immune system that develops clinically years down the line. Even though we only studied rheumatoid arthritis, certainly antibodies seem to be very important in the pathogenesis of many autoimmune diseases—obviously, lupus being another one that comes to mind. So you do wonder how this could be a framework for how antibodies might be produced at mucosal surfaces in the lungs many years prior to disease onset, which is certainly been shown in many other diseases.

Dr. Yoshida:  That's exactly right. Then the analysis framework of life course epidemiology is general. It doesn't have to be passive smoking or active smoking that are involved that is variable.

It's pretty much useful for any situations where there is early life exposure, and then later life, secondary exposure, which complicates the analysis of the first earlier life exposure. We could potentially use this for other rheumatic diseases, as well as other areas of research.

Dr. Sparks:  Great. I think you hit on the future directions. Certainly, you wonder about many other types of exposures such as BMI, or different food intakes. This life course epidemiology has a flexible framework, particularly for chronic diseases that have an insidious onset like this.

Kazuki, you're just down the hall. We could chat about this for hours. We should probably wrap up now, but it's been a pleasure working on this study with you. I'm looking forward to our next collaborations.

Dr. Yoshida:  I'm looking forward to that, too.

Dr. Sparks:  Thanks again.

Dr. Yoshida:  Thank you. Thanks, everyone for listening in.