Objective Measurement and Characterization of Sleep Benefit in Parkinson's Disease

Abstract

Background

Sleep benefit (SB) in Parkinson's disease refers to improved motor symptoms upon waking despite an entire night without medications. Although it was first proposed 30 years ago, this phenomenon proved difficult to investigate, and its true prevalence and underlying mechanisms remain unclear.

Objective

This study aimed to identify and quantify SB through measurement of motor function using a validated smartphone application and to identify disease characteristics that predicted SB.

Methods

Ninety‐two patients recruited from 2 Movement Disorder Services were clinically assessed at home using a validated smartphone application. Each patient was tested in the on‐state, at the end of dose, and on waking (before medications) 3 times. Differences between the 3 states were used to determine the impact of sleep and levodopa on motor function. SB was considered to be a “measurable improvement in parkinsonism from the end of dose.”

Results

The morning waking motor function of 20 patients (22%) improved compared with the end‐of‐dose function, with 9 patients demonstrating superior function compared with their on‐state. No clinical features predicted SB. Although all participants subjectively reported motor fluctuations, only 35 patients (38%) demonstrated an objective improvement with levodopa. Patients who had SB more often demonstrated objective motor fluctuations compared with those who did not (65% vs. 31%; P = 0.008).

Conclusions

SB is a genuine motor phenomenon: 1 in 5 patients have a measurable improvement in motor function on waking. It remains questionable whether this improvement is a direct effect of sleep. Until its underlying mechanism is better understood, it is more appropriate to refer to this phenomenon as simply morning improvement or diurnal fluctuation of motor symptoms.

Despite an entire night without medications, some patients with Parkinson's disease (PD) report waking with improved motor function, equivalent to or better than the on‐medication state. It has been over 30 years since Marsden first described the phenomenon of “sleep benefit”1 (SB), but its characteristics and physiology remain poorly understood. SB reportedly occurs in up to 55% of patients, with improvement lasting up to several hours and also emerging after daytime naps.2, 3 Several physiological explanations have been hypothesized, with the concept of replenishment of dopaminergic storage in the nigral neuronal terminals during sleep being favored.1, 2, 4, 5

Efforts to confirm SB have used self‐reporting and structured questionnaires, but few studies have also included objective measures to quantify motor improvement.2, 4, 5, 6, 7, 8, 9, 10, 11 The number of studies examining SB is small, and the results are variable.2, 4, 5, 6, 7, 8, 9, 10, 11 No single clinical feature has been consistently identified as a predictor of this phenomenon.5, 9, 10, 12 Three small studies attempted to quantify SB,5, 9, 10 but the results were conflicting, and the correlations between subjective reporting and objective motor measurement of SB were inconsistent. These contradictions bring the very existence of motor improvement due to sleep into question.

We adopted a pragmatic model of home‐based testing using a validated, custom‐designed smartphone application that quantitatively measures motor function.13 In a large cohort of patients with PD, our objective was to: (1) identify and measure SB through quantitative motor assessment, and (2) determine whether any clinical characteristics could predict the occurrence of SB.

Patients and Methods

Overview

This study was approved by the Alfred and Royal Melbourne Hospitals’ Human Research and Ethics Committees. All eligible patients provided informed consent.

Patients and Data Collection

Consecutive patients who were subjectively experiencing medication‐related fluctuations were screened and recruited from 2 Movement Disorder Services. Patients fulfilled the UK Parkinson's Disease Society Brain Bank diagnostic criteria.14 Patients who were on monotherapy other than levodopa (l‐dopa) (i.e., dopamine agonists, amantadine, or monoamine oxidase inhibitors) or were receiving no pharmacotherapy were excluded. Other exclusion criteria were severe PD (Hoehn and Yahr Stages 4 and 5), device‐assisted therapies (apomorphine infusion, l‐dopa–carbidopa intestinal gel infusion, and deep brain stimulation), moderate cognitive impairment (a score <22 on the Montreal Cognitive Assessment15), and physical or visual disability deemed by the clinician to affect the patient's ability to use a smartphone.

Smartphone Application

Home‐based motor assessment was carried out using a smartphone application that comprised 4 short tests: Timed Tapping Test (TTT), Rapid Alternating Movements (RAM), Tremor Tracker, and Cognitive Interference Test (CIT). This application measures hand dexterity and gives a reasonable indication of a patient's overall motor function comparable to that provided by Web‐based motor assessments.13, 16 It has been validated against the Movement Disorders Society‐sponsored revision of the Unified Parkinson's Disease Rating Scale motor examination (MDS‐UPDRS‐III)17 (r = 0.281–0.608; P < 0.0001) and the 2‐target tapping test12 (r = 0.339–0.72; P < 0.007). The repeatability of key test parameters was moderate to strong, with an intraclass correlation coefficient of 0.584 to 0.763 (P < 0.0001). MDS‐UPDRS‐III total score was best predicted by 4 key test parameters (TTT mean total score; RAM mean frequency of rotation; Tremor Tracker time [left hand, Part 4]; and CIT time [Set 1, color]; R² = 0.523; F[4,93] = 25.48; P < 0.0001). Predicted MDS‐UPDRS‐III scores calculated from this model correlated strongly with actual MDS‐UPDRS‐III scores (r = 0.686; P < 0.0001). Of the 4 key test parameters, the TTT mean total score performed best in terms of its correlation to the MDS‐UPDRS‐III total score (r = −0.608; P < 0.0001) and repeatability (intraclass correlation coefficient = 0.763; P < 0.001).13

Procedure

Clinical characteristics and demographics of the patients were documented. The l‐dopa equivalent daily dose was calculated according to published methods.18 Baseline assessments included on‐state MDS‐UPDRS‐III, the Montreal Cognitive Assessment, the Pittsburg Sleep Quality Scale (PSQI),19 and the Epworth Sleepiness Scale.19 Standardized demonstration of the smartphone application was provided to participants, including supervised completion of testing 3 times to confirm adequate application use. Patients completed testing at home in the on‐state, end‐of‐dose, and immediately upon waking in the morning before l‐dopa. On‐state was defined as 1 hour after receiving L‐dopa, and end‐of‐dose was defined as immediately before receiving l‐dopa (other than the first dose of the day). Testing was repeated 3 times in each state and was completed over a period of 3 to 5 days.

Statistical Analysis

All statistical analyses were performed using IBM SPSS version 22 (IBM Corporation, Armonk, NY). For each patient, the mean of 3 sets of test results from 4 key smartphone application test parameters (TTT mean total score; RAM mean frequency of rotation; Tremor Tracker time [left hand, Part 4]; and CIT time [Set 1, color]) from each state was calculated and used for analysis. Reliable changes in each of the 4 key test parameters were calculated with the reliable change criterion set at 80%.20 This cutoff was calibrated on the basis of the known “moderate” clinically important change on the UPDRS‐III (5 points).21 Based on this cutoff, the likelihood that measurement unreliability alone would account for a change in the result on repeated measurements in the same patient was less than 20%. For each test parameter, a change in performance greater than the reliable change was considered significant.13 Using a multiple regression model based on these 4 parameters, the predicted MDS‐UPDRS‐III score was calculated for each patient in the 3 different motor states.

Sleep effect was calculated as the difference in performance between “waking” state and end‐of‐dose function. The l‐dopa effect was determined as the difference in performance between on‐state and end‐of‐dose function. Significant improvement in motor function was defined as follows:

1. Reduction in predicted MDS‐UPDRS‐III score by 5 points or more; OR

2. Improvement on key test parameters meeting reliable change as follows:

   1. TTT‐Mean total score; OR

   2. Tremor Tacker time (left hand, Part 4) AND RAM mean frequency of rotation or CIT time (Set 1, color); OR

   3. RAM mean frequency of rotation AND CIT time (Set 1, color).

According to how waking motor function compared with on‐state and end‐of‐dose function, patients were categorized as follows:

1. Group A‐overnight motor deterioration;

2. Group B‐waking state function improved compared with end‐of‐dose function but was not as good as on‐state function;

3. Group C‐waking state function was as good as on‐state function; or

4. Group D‐waking state function was better than on‐state function.

Using Student t tests, Mann‐Whitney tests, and χ² statistics, demographics and clinical characteristics were analyzed. One‐way analyses of variance and Kruskal‐Wallis tests were performed to compare patient groups categorized by conditions of SB and l‐dopa responsiveness.

Results

Ninety‐two patients with PD were recruited with mean age of 65 years (range, 38–87 years) and a mean disease duration of 8.7 years (range, 2–24 years). The mean on‐state MDS‐UPDRS‐III total score was 20 (range, 1–47), and the mean Hoehn and Yahr stage was 2 (range, stages 1–3). The mean l‐dopa equivalent daily dose (LEDD) was 930 mg daily (range, 150–3046 mg daily), and the mean treatment duration was 6.3 years (range, 0–21 years). No patients took any dopaminergic medications overnight. The mean time between daytime l‐dopa doses was 4.3 hours (range, 3–7 hours), whereas the mean time from last evening dose to the first morning dose of l‐dopa dose was significantly longer at 11 hours (range, 6–17 hours; P < 0.0001).

Although all participants subjectively reported medication‐related motor fluctuations (MFs), only 35 patients (38%) demonstrated objective improvement on l‐dopa according to our definitions. This group was considered to have objective MFs. All 35 patients had improved predicted MDS‐UPDRS‐III or TTT mean total scores, and 16 of those 35 patients (48%) demonstrated improvements on both measures. The mean reduction in the predicted MDS‐UPDRS‐III score was 9 (range, 5–23), representing a 38% reduction from the mean end‐of‐dose score of 24. The mean improvement in the TTT mean total score was 18 (range, 13–31), representing a 22% improvement from the mean end‐of‐dose score of 82.

Thirty‐five patients (38%) had overnight motor deterioration (Group A). The waking function of 5 patients (5%) improved from end‐of‐dose function but was not as good as on‐state function (Group B). Forty‐three patients (47%) demonstrated motor function on waking similar to on‐state function (Group C). Of this group, 11 had objective MFs, and their end‐of‐dose function improved to the same degree as with l‐dopa. The other 32 patients did not have objective MFs, and their waking motor function was no different from their end‐of‐dose/on‐state function. The waking function of 9 patients (10%) was superior to their on‐state function (Group D). Of these, 2 patients had objective MFs. The other 7 patients without objective MFs improved from their invariable on‐state/end‐of‐dose function. Therefore, in total, 57 of 92 patients (62%) had either improved or stable motor function on waking. Of these, 20 of 92 patients (22%; 11 from Group C and 9 from Group D) had improved function on waking compared with end‐of‐dose function, and their waking motor function was similar or superior to their on‐state function. This group was considered to have objective SB (Fig. 1). All but 2 of 20 patients (10%) improved on both predicted mean MDS‐UPDRS‐III and TTT total scores. The mean reduction in the predicted MDS‐UPDRS‐III score was 7 (range, 5–12), representing a 30% reduction from the mean end‐of‐dose score of 23. The mean improvement in the TTT mean total score was 17 (range, 13–27), representing a 20% improvement from the mean end‐of‐dose score of 86. Patients with SB had objective MFs more often than those without SB (65% vs. 31%; P = 0.008).

Figure 1.

The proportion of patients with different degrees of objective sleep benefit and motor fluctuations is illustrated.

No clinical characteristics differentiated patients with and without objective SB. Although more patients with objective SB snored (33% vs. 5%; P = 0.011), this was not reflected on other measures of sleep quality (PSQI and Epworth sleepiness scale). The LEDD, the receipt of long‐acting dopaminergic medications, and the mean time from last evening to first morning l‐dopa dose also did not differ between patients with and without objective SB (Table 1).

Table 1.

Demographics and clinical characteristics of patients with and without objective sleep benefit

Variable	Objective Sleep Benefit: No./Total No. (%)		P value
	No	Yes
No. of patients	72/92 (78)	20/92 (22)	–
Men	36/72 (50)	11/20 (55)	0.802
Right handed	67/72 (93)	18/20 (90)	0.643
Dominant side worst	25/72 (35)	10/20 (50)	0.298
Age: Mean ± SD, y	64.7 ± 9.8	65.6 ± 7.9	0.693
Age at symptom onset: Mean ± SD, y	56.2 ± 10.1	56.3 ± 8.3	0.952
Time of disease from diagnosis: Mean ± SD, y	8.5 ± 5	9.3 ± 3.7	0.487
Time on treatment: Mean ± SD, y	6.2 ± 4.6	6.9 ± 3.9	0.526
LEDD: Mean ± SD, mg	908 ± 553	1008 ± 480	0.432
Time from waking to first dose medication: Mean ± SD, min	66 ± 81	41 ± 44	0.065
Medications other than standard l‐dopa
Controlled‐release l‐dopa after 9 pm	28/72 (39)	8/20 (40)	1.000
Antidepressants	27/72 (38)	7/20 (35)	1.000
Antipsychotics	3/72 (4)	3/20 (15)	0.114
Dopamine agonists	41/72 (57)	11/20 (55)	1.000
Long‐acting dopamine agonists	31/72 (43)	7/20 (35)	0.621
Monoamine oxidase inhibitors	7/72 (10)	1/20 (5)	0.681
Entacapone	23/72 (32)	10/20 (50)	0.188
Amantadine	6/72 (8)	3/20 (15)	0.402
Anticholinergic	2/72 (3)	0/20 (0)	1.000
PSQI total score: Mean ± SD	7 ± 4	7 ± 3	0.997
Time to fall asleep: Mean ± SD, min	21 ± 17	19 ± 12	0.505
Sleep per night: Mean ± SD h	6.6 ± 1.4	6.3 ± 1.5	0.568
ESS total score: Mean ± SD	9.2 ± 5.2	10.2 ± 3.5	0.311
MOCA total score: Mean ± SD	27 ± 2	28 ± 1	0.123
On‐state MDS‐UPDRS‐III total score: Mean ± SD	19 ± 9	21 ± 11	0.492
Total “midline” signs: Mean ± SD	3 ± 3	3 ± 2	0.691
Nocturnal symptoms
Bed immobility	19/72 (26)	5/20 (25)	1.000
Symptoms of REM sleep behavior disorder	24/72 (33)	8/20 (40)	0.603
Restless legs syndrome	25/72 (35)	7/20 (35)	1.000
Obstructive sleep apnea	4/72 (6)	2/20 (10)	0.608
Snoring most nights	24/72 (33)	1/20 (5)	0.011
Vivid nightmares	7/72 (10)	4/20 (20)	0.246
Nocturia	47/72 (65)	17/20 (85)	0.106
Motor complications
Dyskinesia	31/72 (43)	6/20 (30)	0.318
Unpredictable off periods	4/72 (6)	1/20 (5)	1.000
Dystonia	29/72 (40)	5/20 (25)	0.296
Autonomic symptoms
Constipation	40/72 (56)	16/20 (80)	0.069
Bladder disturbances	43/72 (60)	12/20 (60)	1.000
Postural dizziness	30/72 (42)	7/20 (35)	0.797
Others
Impulse‐control behaviors	14/72 (20)	3/20 (15)	0.756
Visual hallucinations	26/72 (36)	8/20 (40)	0.797
Sensory symptoms	14/72 (20)	0/20 (0)	0.035

SD, standard deviation; LEDD, l‐dopa equivalent daily dose; PSQI, Pittsburg Sleep Quality Scale; ESS, Epworth Sleepiness Scale; MOCA, Montreal Cognitive Assessment; MDS‐UPDRS‐III, Movement Disorders Society‐sponsored revision of the Unified Parkinson's Disease Rating Scale motor examination; REM, rapid eye movement.

Clinical characteristics were further analyzed with patients categorized into 4 groups based on their status of objective MFs and SB (MF−/SB−, MF−/SB+, MF+/SB+, MF+/SB−). A 1‐way between‐groups analysis of variance was performed, and there was a significant difference in the PSQI total score (F[3,91] = 2.8; P = 0.045) and the PSQI time to fall asleep (F[3,91] = 3.8; P = 0.012). Despite reaching statistical significance, post‐hoc comparisons using the Tukey honest significant difference test indicated no difference in PSQI total scores between the groups. Whereas, for PSQI time to fall sleep, the mean score for MF−/SB− (17 ± 12 minutes) was significantly shorter compared with that for MF+/SB− (30 ± 22 minutes; P = 0.01), but the effect size (0.12) was only medium.22 On the Kruskal‐Wallis test, the presence of sensory symptoms (P = 0.041) differed significantly across the 4 patient groups. However, further pairwise comparison of groups using Mann‐Whitney test failed to demonstrate any statistical difference.

Discussion

In a large study using a validated, objective measurement of motor performance immediately upon waking, we demonstrated that SB is a real and quantifiable phenomenon. Almost two‐thirds of patients with PD did not show the expected deterioration in motor function upon waking in the morning. One in 5 patients experienced a significant improvement in motor performance upon waking compared with end‐of‐dose function, and some had superior motor function compared with their best on‐state function. Apart from measurable medication‐related MFs, no single clinical factor was identified that could predict SB.

In contrast to previous studies in which motor function was compared before and after sleep,9, 10 we chose end‐of‐dose function and on‐state function as references. Logically, the morning waking state should be similar to or worse than end‐of‐dose given an entire night without l‐dopa. Therefore, any improvement compared with end‐of‐dose function should be considered significant and due to changes brought about from sleep. SB has traditionally been defined as motor improvement from sleep equivalent to or better than that in a medication‐induced on‐state.3 This provides a clear reference point for patients, particularly when SB was ascertained through self‐reporting. However, advances in ambulatory motor assessment allow the identification of a significant but milder degree of objective SB, which otherwise would have been excluded by this strict definition.

We were able to broadly identify 3 categories of patients. The first demonstrated clear overnight motor deterioration compared with end‐of‐dose function. The second had minimal objective MFs throughout the day, with stable motor function even upon waking (nonfluctuators). Although this may have resulted from limited test capability (e.g., sensitivity, reliability, and repeatability), the margin of error in our smartphone application is probably small enough that any MFs not identified are likely to be minor. Therefore, what this observation highlights is that a degree of discrepancy does exist between subjective experience and objective measurement of motor function, because all participants reported medication‐related, subjective change. Despite a lack of deterioration on waking, this category of patients was not considered to have SB. The third category either improved their waking morning motor function from an invariable baseline or had objective MFs, and their waking state improved by varying degrees compared with their end‐of‐dose state. We propose that patients within this third category who improved their waking morning motor function by the same extent as or superior to their on‐state function constitute the population with true SB.

The mechanism of SB remains a mystery. It could be considered a spontaneous rather than drug‐induced form of MFs.7 In contrast to normal individuals, whose finger tapping speed is greatest in the afternoon and early evening,23 treatment‐naïve patients with de novo PD performed better in the morning.24 Therefore, it appears that SB is naturally occurring and is related to the intrinsic disease state of dopamine deficiency. If SB is indeed a form of spontaneous MF, then reviewing the metabolism of nigrostriatal dopamine and the possible pathophysiology of MFs may provide clues about its underlying mechanisms. The overall l‐dopa response is determined by long‐duration response (LDR) and short‐duration response (SDR). An SDR establishes and decays quickly; its duration shortens with disease progression and may subsequently cause rebound worsening.25, 26 LDR builds up and decays over many days, and its magnitude is inversely proportional to disease severity. LDR account for up to one‐half of overall l‐dopa responses.27 Shortening of SDR decline in endogenous dopamine synthesis, and reduction of LDR lead to MFs.26 Alterations of l‐dopa pharmacokinetics28 cannot adequately explain MFs, because l‐dopa clearance remains unchanged even with disease progression.29 The underlying mechanism of MFs may be in the presynaptic regulation of vesicular dopamine release through the loss of (e.g., desensitization or downregulation) presynaptic D2 autoreceptor function, leading to excessive synaptic dopamine and resultant increased turnover.28, 30

The mechanism of LDR is unknown, and, analogous to SB, a theory of capacity to replenish dopamine storage had been put forward.26 It is conceivable that the 2 phenomena may be parallel or indeed may exist on the same spectrum. In the absence of long‐acting dopaminergic medications, waking motor function is determined by endogenously derived dopamine and LDR. Because LDR does not change over the short term, SB must be caused by enhanced, endogenously derived dopamine through an unknown mechanism. Consider dopamine‐responsive dystonia, a pure biochemical model of dopamine deficiency without nigrostriatal neurodegeneration. Its characteristic diurnal fluctuation indicates that patients are indeed able to somehow endogenously “replenish” their nigrostriatal system during sleep.31 Thus, the mechanism of SB in PD probably also has a presynaptic location.

Presynaptic dopamine is packed in vesicles by Type 2 vesicular monoamine transporter (VMAT2). In response to action potential, vesicles release their contents into the synaptic cleft. Some dopamine is metabolized, while some is retrieved through plasma membrane dopamine transporters (DATs).28 Dopaminergic neurotransmission has an underlying diurnal variation. The extracellular dopamine level rises in the morning, peaks in the early afternoon, and drops in the latter one‐half of the day to reach a trough in the early hours of the morning.32 Motor symptoms and l‐dopa responsiveness mirror this pattern.33 A recent study using DAT knockout mice suggested that this natural diurnal variation in extracellular dopaminergic tone is governed by DAT and is unrelated to fluctuation in dopaminergic neuron firing rates.32 Although there is a loss of DAT in PD, how α‐synuclein affects the kinetics of remaining DAT is unclear. Therefore, “stored dopaminergic capacity” in SB and LDR may lie in alterations of dopamine reuptake, packaging, and storage in the disease state.

If LDR and SB are parallel phenomena, then they should be universally present in all patients with PD. If so, then why is SB only clinically evident in some? The presence of objective MFs facilitates the detection of any improvement in waking motor function with reference to end‐of‐dose function. In patients without objective MFs, SB can only be detected when there is an improvement beyond steady baseline motor function. Therefore, we propose that SB may be a dynamic phenomenon evident only during certain narrow windows in the disease course. In treatment‐naïve patients whose motor function purely reflects endogenous dopamine production, SB logically should be evident, and this had indeed been observed.24 In early disease where LDR is robust and motor control is stable, any SB may be easily concealed. With the onset of MFs and decay in LDR, SB may then be unmasked. With disease progression, further neurodegeneration may lead to a decline in SB paralleling the decline in LDR. Thus, the overall l‐dopa level on waking may be insufficient to trigger an antiparkinsonian effect, leading to loss of SB.

Coining the term “sleep benefit” to describe this diurnal fluctuation of motor function implies a direct relationship to sleep, which may not necessarily be the case. Previous studies suggested that SB in PD may be associated with shorter sleep time,8 more sleep disruption, and more frequent night‐time awakenings.2, 9 Consistent with these findings, our data suggest that both sleep quality and sleep duration have little bearing on the occurrence of SB. Although several studies have suggested that motor benefit might occur after daytime naps, this conclusion is based solely on subjective reporting.2, 34, 35 Until its underlying mechanism is understood better, it is more appropriate to refer to this phenomenon as simply morning improvement or diurnal fluctuation of motor symptoms.

Although this study benefits from a validated and mobile method of motor function assessment, several limitations should be highlighted. The study population is relatively selected, with patients typically in the early to middle disease stages (Hoehn and Yahr stages 2 and 3), hence results may not be generalizable. We adopted an unsupervised, home‐based test approach using mobile technology, which enabled testing a large number of patients without consuming substantial resources. This is at the expense of test reliability and repeatability, which not surprisingly, are inferior compared with traditional labor‐intensive, face‐to‐face assessment36 and other resource‐demanding forms of continuous motor monitoring.37 Because our smartphone application has performance comparable to that of Web‐based motor assessment16 and reasonable reliability thresholds, we believe the margin of error is acceptable for such a simple, portable test. Although more extensive direct clinical testing would have been ideal, the amount of time and resources required would render the same study infeasible. The incorporation of polysomnography will be valuable as a measurement of sleep quality based on self‐reporting and may compromise the detection of any sleep‐related factors.

SB in PD is a genuine, measurable phenomenon, with one in five patients improving by varying degree compared to end‐of‐dose, and some even waking up better than their on‐state. SB should be thought of as a form of intrinsic MF. Given its frequency of occurrence and magnitude, it is a clinically meaningful phenomenon and should be considered when implementing dopaminergic treatment and when motor function is measured in clinical studies.

Author Roles: 1. Research Project: A. Conception, B. Organization, C. Execution; 2. Statistical Analysis: A. Design, B. Execution, C. Review and Critique; 3. Manuscript Preparation: A. Writing the First Draft, B. Review and Critique.

W.L.: 1A, 1B, 1C, 2A, 2B, 3A

A.H.E.: 1A, 1B, 2C, 3B

D.R.W.: 1A, 1B, 2C, 3B

Disclosures

Ethical Compliance Statement: We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this work is consistent with those guidelines.

Funding Sources and Conflict of Interest: No specific funding was received for this work. David R. Williams and Andrew H. Evans declare intellectual property rights to the Parkinson's Disease Assessment Application.

Financial Disclosures for the previous 12 months: The authors report no sources of funding and no conflicts of interest.