On 2016 Dec 20, David Keller commented:

Effects of Rheumatoid Arthritis, and Its Treatments, on Parkinson Disease

Sung and colleagues derived two new and independent hypotheses from this study. First, they observed an inverse association between rheumatoid arthritis (RA) and the risk of subsequent development of Parkinson disease (PD), consistent with the hypothesis that RA is protective against PD. Second, they observed that the risk of developing PD was reduced even more in RA patients treated with biological disease-modifying anti-rheumatic drugs (DMARDs) but not in patients treated without DMARDs or with non-biological DMARDs. [1] This led to their second hypothesis, that "biologic DMARDs appear to further reduce the PD risk" in RA patients (more so than treatment with non-biological DMARDs, or no DMARDs), suggesting a possible "role of biologic DMARDs in PD treatment".

In summary, the authors of this study propose two new hypotheses to fully explain their results:

Hypothesis #1: Rheumatoid Arthritis disease is protective against the onset of Parkinson disease.

Hypothesis #2: Biological DMARDS, as treatment for RA, confer additional protection against the onset of PD.

Sung and colleagues point out that Hypothesis #1 contradicts "the hypothesis that chronic inflammation in RA may increase the risk of developing PD", citing a recent study that, similarly, "identified an inverse association between PD and systemic lupus erythematosus", and another confirmatory study that "reported a 30% reduction in the risk of developing PD in patients with RA and systemic involvement" [Sung's references #23 and #24]. Sung mentioned that RA patients are more likely to take NSAIDs, and that certain NSAIDs have been found to be protective against PD, but Sung claims that the association of RA with protection from PD withstood controlling for NSAID use. In a separate comment, I will address the errors I believe Sung and colleagues made when correcting their data for NSAID use, and how those errors could falsely support Hypothesis #1, above.

However, suppose hypothesis #1 is true. The association of additional reduction of PD incidence with the use of biological DMARDs might not be due to their having an intrinsic neuroprotective effect, but, rather, to the fact that they are reserved for use in the most severe or refractory cases of RA. The increased level of RA disease activity and severity associated with the use of biological DMARDs could be the cause for the additional decreased risk of PD. The inability to distinguish whether the additional protection from PD was due to neuroprotective benefits of biological DMARDs, or due to the more severe RA which caused biological DMARDs to be "indicated" (medically needed), is an example of confounding by "indication bias".

Biological DMARDs have likely been compared with non-biological DMARDs in randomized clinical trials for treatment of rheumatoid arthritis. If these trials included data on the new onset of PD, they could be pooled in a meta-analysis to test Sung's Hypothesis #2.

Reference

Sung YF, Liu FC, Lin CC, et al. Reduced risk of Parkinson disease in patients with rheumatoid arthritis: A nationwide population-based study. Mayo Clin Proc. 2016;91(10):1346-1353.

On 2016 Dec 19, David Keller commented:

Only ibuprofen is associated with reduced PD risk - controlling for use of any NSAID introduces error

The following quote is from the above paper by Sung [1]:

"Previous studies [2,3] have reported that nonaspirin NSAIDs, particularly ibuprofen, may be associated with a reduced risk of developing Parkinson disease. However, after controlling for all comorbidities and NSAID use, patients with RA still exhibited a reduced risk of PD compared with patients without RA..." However, the two cited studies actually demonstrated that ibuprofen use is associated with significantly reduced risk of PD, but other commonly-used NSAIDs are not.

In a landmark paper published in 2011, but not cited by Sung, Gao and colleagues published a new observational trial, plus a comprehensive meta-analysis, which concluded that ibuprofen, but not other NSAIDs, is associated with a significant 38% reduction in risk for PD [4]. Therefore, the bolded phrase should be corrected to read: "ibuprofen, but not other NSAIDs, is associated with a significantly reduced risk of developing PD".

Sung and colleagues controlled for NSAID use, but not separately for ibuprofen use; their Table 3 presents 11 baseline variables, and the adjusted HR of each. NSAIDs are discussed together as a single group, exhibiting a small but significant 9% protective effect against PD, the result of diluting the larger 38% protection associated with ibuprofen with the lack of significant protection associated with other NSAIDs.

Thus, the HR of 0.91 used by Sung to control for the use of any NSAID systematically under-corrects for the protection from PD for patients who took ibuprofen, while systematically over-correcting when the NSAID used was not ibuprofen. To eliminate these systematic errors, the study data should be reanalyzed, and corrected specifically for ibuprofen use, rather than for the use of any NSAID.

Until this correction is made, it is unclear how much of the apparent protection associated with RA disease or with biological DMARDs was actually attributable to the use of ibuprofen.

In an unpublished reply to these arguments, Sung's group wrote: "[Keller's] criticism focuses on the issue whether [any] non-aspirin NSAID or ibuprofen only, has the truly protective effect against the development of PD". Sung and colleagues agreed that "ibuprofen was associated with decreased risk of PD, but not aspirin or other NSAIDs" and concluded that "ibuprofen use should be considered as an important covariable in future correlational research in PD." [5]

References

1: Sung YF, Liu FC, Lin CC, Lee JT, Yang FC, Chou YC, Lin CL, Kao CH, Lo HY, Yang TY. Reduced Risk of Parkinson Disease in Patients With Rheumatoid Arthritis: A Nationwide Population-Based Study. Mayo Clin Proc. 2016 Oct;91(10):1346-1353. doi: 10.1016/j.mayocp.2016.06.023. PubMed PMID:27712633.

2: Chen, H., Jacobs, E., Schwarzschild, M.A. et al, Nonsteroidal antiinflammatory drug use and the risk for Parkinson's disease. Ann Neurol. 2005;58:963–967.

3: Rees, K., Stowe, R., Patel, S. et al, Non-steroidal anti-inflammatory drugs as disease-modifying agents for Parkinson's disease: evidence from observational studies. Cochrane Database Syst Rev. 2011;:CD008454.

4: Gao X, Chen H, Schwarzschild MA, Ascherio A. Use of ibuprofen and risk of Parkinson disease. Neurology. 2011 Mar 8;76(10):863-9. doi:10.1212/WNL.0b013e31820f2d79. PubMed PMID: 21368281; PubMed Central PMCID: PMC3059148.

5: Sung YF, Lin CL, Kao CH, and Yang TY. Reply to Keller's unpublished letter to Mayo Clinic Proceedings. Received by email on November 22, 2016.

On 2016 Dec 19, David Keller commented:

Only ibuprofen is associated with reduced PD risk - controlling for use of any NSAID introduces error

The following quote is from the above paper by Sung [1]:

"Previous studies [2,3] have reported that nonaspirin NSAIDs, particularly ibuprofen, may be associated with a reduced risk of developing Parkinson disease. However, after controlling for all comorbidities and NSAID use, patients with RA still exhibited a reduced risk of PD compared with patients without RA..." However, the two cited studies actually demonstrated that ibuprofen use is associated with significantly reduced risk of PD, but other commonly-used NSAIDs are not.

In a landmark paper published in 2011, but not cited by Sung, Gao and colleagues published a new observational trial, plus a comprehensive meta-analysis, which concluded that ibuprofen, but not other NSAIDs, is associated with a significant 38% reduction in risk for PD [4]. Therefore, the bolded phrase should be corrected to read: "ibuprofen, but not other NSAIDs, is associated with a significantly reduced risk of developing PD".

Sung and colleagues controlled for NSAID use, but not separately for ibuprofen use; their Table 3 presents 11 baseline variables, and the adjusted HR of each. NSAIDs are discussed together as a single group, exhibiting a small but significant 9% protective effect against PD, the result of diluting the larger 38% protection associated with ibuprofen with the lack of significant protection associated with other NSAIDs.

Thus, the HR of 0.91 used by Sung to control for the use of any NSAID systematically under-corrects for the protection from PD for patients who took ibuprofen, while systematically over-correcting when the NSAID used was not ibuprofen. To eliminate these systematic errors, the study data should be reanalyzed, and corrected specifically for ibuprofen use, rather than for the use of any NSAID.

Until this correction is made, it is unclear how much of the apparent protection associated with RA disease or with biological DMARDs was actually attributable to the use of ibuprofen.

In an unpublished reply to these arguments, Sung's group wrote: "[Keller's] criticism focuses on the issue whether [any] non-aspirin NSAID or ibuprofen only, has the truly protective effect against the development of PD". Sung and colleagues agreed that "ibuprofen was associated with decreased risk of PD, but not aspirin or other NSAIDs" and concluded that "ibuprofen use should be considered as an important covariable in future correlational research in PD." [5]

References

1: Sung YF, Liu FC, Lin CC, Lee JT, Yang FC, Chou YC, Lin CL, Kao CH, Lo HY, Yang TY. Reduced Risk of Parkinson Disease in Patients With Rheumatoid Arthritis: A Nationwide Population-Based Study. Mayo Clin Proc. 2016 Oct;91(10):1346-1353. doi: 10.1016/j.mayocp.2016.06.023. PubMed PMID:27712633.

2: Chen, H., Jacobs, E., Schwarzschild, M.A. et al, Nonsteroidal antiinflammatory drug use and the risk for Parkinson's disease. Ann Neurol. 2005;58:963–967.

3: Rees, K., Stowe, R., Patel, S. et al, Non-steroidal anti-inflammatory drugs as disease-modifying agents for Parkinson's disease: evidence from observational studies. Cochrane Database Syst Rev. 2011;:CD008454.

4: Gao X, Chen H, Schwarzschild MA, Ascherio A. Use of ibuprofen and risk of Parkinson disease. Neurology. 2011 Mar 8;76(10):863-9. doi:10.1212/WNL.0b013e31820f2d79. PubMed PMID: 21368281; PubMed Central PMCID: PMC3059148.

5: Sung YF, Lin CL, Kao CH, and Yang TY. Reply to Keller's unpublished letter to Mayo Clinic Proceedings. Received by email on November 22, 2016.

On 2016 Dec 20, David Keller commented:

Effects of Rheumatoid Arthritis, and Its Treatments, on Parkinson Disease

Sung and colleagues derived two new and independent hypotheses from this study. First, they observed an inverse association between rheumatoid arthritis (RA) and the risk of subsequent development of Parkinson disease (PD), consistent with the hypothesis that RA is protective against PD. Second, they observed that the risk of developing PD was reduced even more in RA patients treated with biological disease-modifying anti-rheumatic drugs (DMARDs) but not in patients treated without DMARDs or with non-biological DMARDs. [1] This led to their second hypothesis, that "biologic DMARDs appear to further reduce the PD risk" in RA patients (more so than treatment with non-biological DMARDs, or no DMARDs), suggesting a possible "role of biologic DMARDs in PD treatment".

In summary, the authors of this study propose two new hypotheses to fully explain their results:

Hypothesis #1: Rheumatoid Arthritis disease is protective against the onset of Parkinson disease.

Hypothesis #2: Biological DMARDS, as treatment for RA, confer additional protection against the onset of PD.

Sung and colleagues point out that Hypothesis #1 contradicts "the hypothesis that chronic inflammation in RA may increase the risk of developing PD", citing a recent study that, similarly, "identified an inverse association between PD and systemic lupus erythematosus", and another confirmatory study that "reported a 30% reduction in the risk of developing PD in patients with RA and systemic involvement" [Sung's references #23 and #24]. Sung mentioned that RA patients are more likely to take NSAIDs, and that certain NSAIDs have been found to be protective against PD, but Sung claims that the association of RA with protection from PD withstood controlling for NSAID use. In a separate comment, I will address the errors I believe Sung and colleagues made when correcting their data for NSAID use, and how those errors could falsely support Hypothesis #1, above.

However, suppose hypothesis #1 is true. The association of additional reduction of PD incidence with the use of biological DMARDs might not be due to their having an intrinsic neuroprotective effect, but, rather, to the fact that they are reserved for use in the most severe or refractory cases of RA. The increased level of RA disease activity and severity associated with the use of biological DMARDs could be the cause for the additional decreased risk of PD. The inability to distinguish whether the additional protection from PD was due to neuroprotective benefits of biological DMARDs, or due to the more severe RA which caused biological DMARDs to be "indicated" (medically needed), is an example of confounding by "indication bias".

Biological DMARDs have likely been compared with non-biological DMARDs in randomized clinical trials for treatment of rheumatoid arthritis. If these trials included data on the new onset of PD, they could be pooled in a meta-analysis to test Sung's Hypothesis #2.

Reference

Sung YF, Liu FC, Lin CC, et al. Reduced risk of Parkinson disease in patients with rheumatoid arthritis: A nationwide population-based study. Mayo Clin Proc. 2016;91(10):1346-1353.