Prevalence of Lewy pathology in essential tremor is twice as high as expected: A plausible explanation for the enhanced risk for Parkinson disease seen in essential tremor cases

To the Editor:

We read the letter from Shill et al regarding our recent paper characterizing Lewy pathology (LP) in 231 essential tremor (ET) brains in the Essential Tremor Centralized Brain Repository (1). We would like to use this opportunity to make several comments. First, they attach the term “incidental” as a label for LP. This label is fraught with problems because it is predicated on the assertion that these lesions have no identified clinical correlate (e.g. no parkinsonism on examination). They cite their own paper, published nearly 10 years ago, on the prevalence of “incidental Lewy body pathology” in normal elders (2). Careful inspection of that paper indicates that they do not specify how close to the time of death their subjects were followed. That paper even makes explicit note of the fact that some unspecified number of their subjects were not examined even to a time point that was a full 18 months prior to death (2). That is, a year and a half elapsed between the last neurological examination and death. Given the incidence of Parkinson disease (PD) as the second most common age-related neurodegenerative condition in the elderly, it is highly probable that clinical features of parkinsonism developed and were not detected in some unknown number of their study subjects, and by implication, that the LP observed in these individuals was not “incidental” but was a marker of early PD. Therefore, their assertions about the high prevalence of “incidental” LP in “normal” elders are problematic and their estimates of the burden of LP in the normal elderly are inflated.

In our article, we completely circumvented this thorny issue. For the reasons stated above, we completely avoid the use of the term “incidental Lewy bodies” (1). Rather, we focus on quantifying the prevalence and distribution of LP on postmortem brain examination of a group of ET cases who had no clinical parkinsonism at the time of study entry.

Second, Shill et al make the argument that we should have excluded ET cases who developed clinical PD (n = 6) or possible clinical PD (n = 8) during the course of the study, but this would not have been an appropriate methodological choice, as our goal was to determine the prevalence and distribution of LP on postmortem brain examination of a group of ET cases who had no clinical parkinsonism at the time of study entry. Thus, contrary to the assertion made by Shill et al, our estimate of the prevalence of LP should include these cases and was indeed 25.1% rather than the lower value of 20.3%.

Third, a query raised by Shill et al is the potential effect of Alzheimer-type pathology on our findings—that is, an issue of confounding due to co-occurring Alzheimer-type pathology. As written in our article (1), we considered the potential impact of Alzheimer-type pathology on the high prevalence of LP we observed. Even when we removed all ET cases with ABC categories of intermediate (n = 99) or high (n = 32), 22 of 100 (22.0%) ET cases had LP. Hence, the prevalence of LP remained high, even when taking this potential source of confounding into account.

Fourth, Shill et al selectively cite their own published work on LP, based on a modest sample size of 119 individuals presented in a single paper (2), and do so to the exclusion of the entirety of other published data. By contrast, we carefully reviewed, cited, and synthesized data from a full complement of 14 studies representing 6363 individuals, and then carefully derived a corrected value for LP in these autopsy series, which yielded a median value of 12.65% and a mean value of 14.5%. These values fall far short of the 25.1% we report in ET, being 73%–98% lower.

Fifth, Shill et al make passing note of the proportion of their ET cases with LP (22.1%), which interestingly is strikingly similar to that which we report (25.1%). As discussed, both repeatedly and in detail in the literature (3–5), problems with their diagnostic scheme reveal that many (likely two-thirds) of their “ET” cases likely had enhanced physiological tremor rather than ET. Hence, their study suffers significantly from diagnostic misclassification as the definition for ET did not exclude enhanced physiological tremor, which would falsely elevate the number of individuals labeled as ET (3–5). Their data on “ET” cannot be viewed as valid.

Finally, Shill et al exclusively draw support for their arguments by citing their own work rather than that of any others. This points to a lack of published data or wider support for the arguments they are attempting to put forward.

What is clear from our article is that a striking proportion—1 in 4—ET cases who entered a study without features of parkinsonism, ended up having LP on postmortem examination. When taken in the context of the entirety of published data on the prevalence of LP in normal elders, this value appears to be double the median value reported in studies of normal elders. As we note in our study, there is a compeling epidemiological literature demonstrating that prevalent ET is associated with increased odds and risk of PD and no methodologically robust data to support the contrary (6–11). In the 1 epidemiological study that was a prospective, longitudinal population-based study, ET cases were observed to have 4–5 times increased risk of developing incident PD than age-matched counterparts without ET (8). Hence, there is substantial evidence that ET is robust a risk factor for PD. The results of epidemiological studies are in keeping with what is observed in treatment settings by clinicians—a high proportion of ET cases seem to develop PD (12). The increased burden of LP observed in our ET cases is keeping with and provides an explanation for this enhanced risk of PD that is seen in ET cases. Future effort would best be spent on studying the associations between ET and LP and PD, as there is a strong and consistent biological signal that is becoming increasingly difficult to summarily sweep under the rug.

FUNDING

This study was supported by National Institutes of Health grant numbers NINDS R01 NS117745, R01 NS086736, and R01 NS088257.

CONFLICT OF INTEREST

The authors have no duality or conflicts of interest to declare.