To the Editor:

The recent study by Gorell et al.1 proposes a risk-reducing effect of tobacco smoking on PD because they found, as in a number of prior investigations, an inverse relationship between smoking and PD. Although the authors consider several possible explanations for the rarity of smoking in PD, they do not consider the most likely explanation for this well-known inverse association. Rather than interpret the findings as indicative of a “protective” effect of smoking on development of PD, the correct interpretation may be that PD is “protective” against being a smoker (e.g., see reference 2, which was not cited by the authors but has an almost identical title to that of their work). Thus, despite the pathoetiologic attractiveness of a protective effect of smoking, a likely explanation for the under-representation of smoking among PD patients is that the patients-to-be lose interest in smoking as a very early presymptomatic feature of their neurodegenerative condition.2

It is generally agreed that PD, like many other neurodegenerative conditions, develops over years, and only becomes noticeably symptomatic to afflicted persons and their families somewhere between 3 and 30 years into the neurodegenerative process.3,4 Because of the subtle subcortical mental and motor manifestations, the former of which may best be classified as decreased initiative or a type of bradyphrenia, some have proposed a “personality” predisposing to PD. However, back-extrapolations from clinical and pathologic data, as well as PET neuroimaging data, indicate that pathologic PD must precede symptomatic PD (with classic motor signs of tremor, shuffling gait, and bradykinesia) by some years.3,4

Gorell et al. report useful data regarding the temporal association of smoking cessation and development of PD. These data support an explanation in which early (preclinical) parkinsonism is marked by, inter alia, decreased desire to smoke. Namely, “distant quitters”—those who stopped smoking >20 years before “symptomatic PD”—show little difference in PD prevalence from those who never smoked. But the closer the examined interval to the onset of evident PD, the stronger the correlation with decreased smoking. As can be seen from the data in the authors’ table 1, the relative risk of PD was 1.0 (by definition) for nonsmokers, nearly the same (0.86) for distant quitting, 0.37 for recent quitting, and very much less (0.08) for currently heavy smokers. Reformulated, with some statistical caveats, this shows that for this group of PD patients and matched controls, only 4% of current heavy smokers, 13% of recent quitters, 26% of distant quitters, and 29% of “never smoked” had PD compared with an overall 24% of individuals who had PD. As the authors argue, this gradient could be expected for a protective effect of smoking on PD: namely, longer periods of smoking cause reduced likelihood of PD (although such a large difference in effect between persons smoking ∼35 versus ∼55 years would be surprising). However, the data are exactly what would be expected for an inhibiting effect of PD on smoking: namely, among those patients with “presymptomatic” PD, closer time frames to the onset of evident motoric symptoms are associated with reduced interest in and likelihood of persistent smoking. Even without such analysis, the social dynamics of tobacco addiction in this cohort5 with median age ∼70, who started smoking ∼1940 and were studied 1988–1992, are such that it is more plausible that a disease process (namely earliest PD) caused them to stop smoking, rather than postulate that about 90% of the PD patients (unlike others) who smoked just happened to quit their 30- to 50-year–long tobacco habits spontaneously in the 1970s and 1980s.

Therefore, I believe this well-performed case-control study does not provide strong evidence that smoking is biologically protective of PD, but rather is more consistent with earliest (preclinical) PD decreasing tobacco-smoking behavior.