Several medications, including statins, NSAIDs, anti-hypertensives, anti-diabetic medications, and beta-agonists were associated with older PD AAO in our cohort, while a smoking history and FH of PD were associated with younger PD AAO. A multiple regression model identified ABs, statins, and NSAIDs as strong independent predictors of older AAO of PD, suggesting that treatment with these medications may delay the onset of PD. On the other hand, smoking history was a strong independent predictor of younger AAO.

The question whether statins may affect the risk of developing PD has received conflicting answers from multiple epidemiological studies. Similar to our findings, a retrospective study of 419 PD patients showed that those treated with cholesterol-lowering medications had a delayed disease AAO of 9 years (63.6 versus 54.6 years) when compared with PD patients who were not on lipid-lowering treatment, in addition to elements suggestive of slower disease progression [10]. Of note, in our larger PD cohort AAO was further delayed, in absolute terms, to an age of 70.8 years. Putative neuroprotective mechanisms attributed to statins include inhibition of proinflammatory substances, microglia, and oxidative stress, as well as suppression of alpha-synuclein aggregation [11]. Most, but not all, epidemiological studies found the use of statins to be associated with a reduced risk of developing PD [3]. However, a randomized clinical trial examining potential disease modifying effects of simvastatin failed to demonstrate any effect over 24 months in patients with moderate PD [12].

To our knowledge, this is the first study showing a delayed PD AAO in subjects exposed to AB, who developed PD symptoms almost a full decade later than those never exposed. However, when ABs were initiated after the onset of PD, AAO did not differ from the group never exposed, suggesting that the cardiovascular indication for AB use did not influence PD AAO. These results add to a large, albeit controversial, body of research on the effect of adrenergic modulating drugs on PD risk. The beta-2 adrenoreceptor is a regulator of the alpha-synuclein gene, and it has been hypothesized that beta-2 adrenoreceptor antagonists increase expression of alpha-synuclein, thus promoting the development of PD, while beta-2 adrenoreceptor agonists decrease this expression and are thus protective [8]. In accordance with this theory, a decreased risk of PD with salbutamol use and an increased risk of PD with propranolol use was reported from a large population database [8]. Several studies casted doubt on these data, noting that beta-2-agonists are typically used in smoking-related pulmonary conditions, likely associated with a reduced risk to develop PD, and beta-2-antagonists—in particular propranolol – are frequently prescribed for tremor. Indeed, propranolol was prescribed after the onset of motor symptoms in over 90% of cases in our PD cohort, under an incorrect initial diagnosis of essential tremor, supporting the suggestion that it may not be truly associated with increased PD risk when adjusting for clinical indication [4]. Interestingly, other adrenergic blockers not used to treat tremor (i.e. carvedilol and metoprolol) were associated with a lower risk of developing PD [4]. Independent studies demonstrated that carvedilol was associated with a reduced PD risk when prescribed over a year prior to diagnosis [13], while other investigations have yielded conflicting results [14]. Although they also appeared to have a 5-year delayed PD AAO, very few patients on beta agonists were recorded in our cohort, preventing a meaningful statistical analysis.

NSAIDs, and most notably aspirin, have been associated with older AAO of PD, with stronger association noted with a higher dose of aspirin or a longer aspirin intake duration [2]. NSAIDs have been associated with a reduced PD risk in some studies, although data from other studies is controversial [7]. Neuroinflammation has been implicated as a potential mechanism, supported by alterations of inflammatory markers in serum and cerebrospinal fluid of PD patients, as well as neurohistological and neuroimaging evidence of inflammation [15].

The possible reasons why a history of smoking would predict a younger PD AAO in our cohort appear less intuitive. Notably, the percentage of subjects who had been active smokers was quite small, less than 5% of the total dataset, which may have biased the results. This particularly low prevalence may be explained by rates of smoking, which are substantially lower in California (about 12.3% in 2008) [16] than in Europe (about 27.2% in 2010) [17]. While smoking is generally associated with a reduced PD risk [2], the relationship of smoking with PD AAO is more controversial. One review found ten studies that showed a delay in AAO, two studies that showed an earlier AAO, and nine studies without a directional effect [2]. We also did not have information regarding smoking-pack-years, which could yield more differentiated results. We postulate that these patients may have already had a strong predisposition to developing PD, thus overcoming the possible protective effect of smoking. One fourth of smokers had a positive family history for PD, but the remaining did not have an identifiable risk factor. Only one patient on beta-agonists was a smoker. Another consideration is that smoking may be more common among younger age groups, and thus we are capturing younger subjects that smoked recently rather than older subjects who smoked over 10 years ago.

Patients being treated for T2D in one study developed PD about 7 years later (66.9 years) than patients who did not have T2D (60.7 years) or patients who developed diabetes after PD onset (60.6 years) [9]. AAO for PD patients who had been on anti-diabetic treatment longer than 7 years had a greater delay than those treated for 7 years or less [9]. In our analysis, out of the 122 patients on anti-diabetic medications at first encounter, we were unable to ascertain start time for 54 patients and thus none of them were included in the regression model, which may have reduced statistical power. Given the potential disease modifying effect of glucagon-like peptide receptor agonists in PD patients [18], the role of anti-diabetics should be explored further.

The medications associated with a delayed AAO in our study appear to have a few common links that may explain a disease modifying effect. First, many of them are antihypertensives, which may modulate the effects of hypertension as a risk factor for PD [19]. In addition, all medications that appear to modulate PD risk in this study have a direct or indirect effect on the sympathetic nervous system and neuroinflammatory response, which may have a prominent role in the pathogenesis of PD [20]. Sympathetic hyperactivity has a major influence on neuroinflammatory status through the inflammasome NLRP3–IL-1β axis, either as a trigger or a consequence of increased cytokine release, and these inflammatory markers have been reported as critical signaling molecules of immune activation in the central nervous system [20]. In addition to suppressing sympathetic overactivity and reducing (and potentially reversing) cardiac remodeling [21], AB have anti-inflammatory effects, as demonstrated by their attenuation of cytokine storm-induced lung injury in COVID-19 [22]. Similarly, statins have been shown to decrease sympathetic activity [23], as well as enhance anti-inflammatory and inhibit pro-inflammatory functions within microglia in vitro [24]. NSAIDs inhibit prostaglandin synthesis, which has been described to decrease plasma norepinephrine and thus reduce sympathetic function [25]. In preclinical studies, carnosine, which acts by suppressing sympathetic activity [26], slowed progression of motor deficits and alpha-synuclein deposition in the Thy1-aSyn mouse [27]. The role of the noradrenergic system in PD deserves further evaluation in human studies [28].

A major limitation of this study lies in its retrospective nature. Information regarding specific medication timing was obtained by chart review or by patient report, both of which may be inaccurate for multiple reasons including recall bias or lack of data regarding compliance. Some data points had to be excluded from the multiple regression analysis because we were unable to establish medication timing relative to development of PD symptoms. In addition, we were unable to precisely determine exposure time and dose effect. Small sample sizes also limit the interpretation of the data. Finally, since caffeine intake was not included in our standardized template, we could not correct risk modifying data for coffee exposure, a behavior previously associated to delayed PD AOO [2].

A final consideration is whether age may be a confounding factor, as older patients – who are more likely to develop PD—are also more likely to be on a number of these medications. Large surveys reported a similar proportion of patients within 18–65 years of age on AB as compared to those above 65 years of age [29]. On the other hand, different proportions of patients appear to be treated with cholesterol medications depending on age [30].

In conclusion, our study demonstrated that common medications including AB, statins, and NSAIDs are associated with a delayed onset of PD. Possible mechanisms modulating the risk of neurodegeneration include hypertension control, down-regulation of the sympathetic nervous system and anti-inflammatory actions. Research using a prospective study design in large cohorts is needed to examine these effects, as drug repurposing would have important advantages to PD patients and the healthcare system.