A systematic review of the association between the age of onset of spinal bulbar muscular atrophy (Kennedy's disease) and the length of CAG repeats in the androgen receptor gene

Introduction Spinal bulbar muscular atrophy (SBMA) is an X-linked recessive motor neuron disorder caused by the presence of ≥38 CAG repeats in the androgen receptor gene. Existing literature indicates a relationship between CAG repeat number and the onset age of some motor symptoms of SBMA. This review explores the effect of larger versus shorter CAG repeats on the age of weakness onset in male SBMA patients. Methods Three databases (October 2021; MEDLINE, SCOPUS, and Web of Science), Cambridge University Press, and Annals of Neurology were searched. 514 articles were initially identified, of which 13 were included for qualitative synthesis. Results Eleven of the thirteen articles identified a statistically significant inverse correlation between CAG repeat length and age of weakness onset in SBMA. Five studies indicated that SBMA patients with between 35 and 37 CAG repeats had an older age of weakness onset than patients with over 40 CAG repeats. The minimum number of CAG repeats associated with weakness was in the mid-to-late thirties. Conclusion Identification of a relationship between CAG repeat number and age of weakness may enable earlier detection and intervention for SBMA. In the future, studies should use interviews, chart reviews, and standardized scoring methods to reduce effects of retrospective bias.


Introduction
Spinal bulbar muscular atrophy (SBMA), also known as Kennedy's disease, is an X-linked recessive motor neuron disorder associated with >38 CAG repeats in the androgen receptor (AR) gene [1], that affects approximately 1/40,000 men (although prevalence is greater in certain locations, such as the Vaasa region of Finland) [2].Increased CAG expansion induces toxic polyQ-AR fragments, resulting in the disruption of transcription, mitochondrial function, protein cycling, cell signaling pathways and autophagy [1,3].The average age of onset of SBMA is in the third or fourth decade and, although female carriers display some symptoms of the condition such as distal motor deficits, cramping and/ or fasciculations [1], men are more severely affected by the condition [4] with symptoms including dysarthria, dysphagia, gynecomastia, cramping, fasciculations and tremor [1].
Generally, SBMA follows a gradual progression, starting with postural tremors of the upper limb in the early thirties [5].In addition to the onset of dysarthria and dysphagia in the forties, lower limb motor deficits also develop, eventually requiring the use of walking aids by the fifth decade [5].Finally, patients become wheelchair-bound by, or during, their sixth decade [5].Patients with early-stage involvement of bulbar musculature often die from recurrent aspiration pneumonia in their fifties [6].
Although it is known that the age of onset of non-motor manifestations of SBMA is independent of CAG repeat length, primary literature indicates that there is also a clear negative correlation between the age of onset of motor manifestations and CAG repeat number (for review see [4]).Although several studies [5][6][7][8][9][10][11][12][13][14][15][16][17] have pondered the relationship between CAG repeat number and the age of the most frequent symptom of SBMA, weakness [18], to date, no systematic reviews have specifically focused on the possible connection between these two factors.As such, herein we seek to address this deficit by exploring the effect of a greater number of CAG repeats on the age of weakness onset in comparison to a lower number of CAG repeats in male SBMA patients (between the ages of 11 and 83) who did not suffer from other major chronic illnesses.
Given that very little is known about the etiology of SBMA, probably due to its relatively low prevalence in the population, and that current treatments for SBMA only facilitate symptom management, men with the condition should be identified as soon as possible such that treatment of symptoms and counselling can begin sooner [4].As such, any addition to the SBMA-associated literature that increases understanding of the disease is timely.Primary literature was reviewed with the following objectives borne in mind: 1. To determine if the association between CAG repeat number and age of onset of symptoms of SBMA is significant.2. Determine what minimum and maximum number of CAG repeats are associated with weakness onset in SBMA.

Study eligibility
Inclusion criteria (see Table B.1, Appendix B) enabled the identification of as many articles as possible that addressed the relationship between CAG repeat number and the specific age of onset of weakness in SBMA patients.Selected studies must have at least some male SBMA patients, while female-only studies were excluded.Only studies that examined SBMA patients (with or without female carriers) that had no other major health conditions such as a concomitant diagnosis of schizophrenia, were included.Studies that involved patients with health conditions that commonly co-occur with SBMA (e.g.type II diabetes) were included.Additionally, the age profile of participants had to range from 11 to 83, which was the most common age range encountered in our literature search.
Any form of peer-reviewed, primary research study (e.g.case control study, etc.), was included if it was written in English and did not include animal subjects.However, reviews, commentaries and/or editorials were excluded.Both freely available texts and articles that could only be purchased [7] were included if they were accessible online.Given the rare nature of the disease and the relative dearth of available literature pertaining to it, no date range was applied to searches.

Study identification
Searches were conducted between September and October 2021, by an independent reviewer.PubMed was subsequently searched using the following combined terms that were designed to address three concepts: "spinal bulbar muscular atrophy", "age of onset of muscle weakness", and "CAG repeat", which yielded 116 results with additional searches in SCOPUS and Web of Science databases, utilizing variations of the same three concepts, which generated 390, and 3 results, respectively.Individual searches were conducted in Cambridge University Press, and the Annals of Neurology: Wiley Online Library which yielded six results (see Appendix A for list of search terms, and Fig. 1 for the search process).

Study selection
As SBMA has not been extensively investigated, we sought to find as many eligible articles as possible such that search filters were not applied.Our searches yielded a total of 515 results (Fig. 1).These were uploaded to EndNote, where duplicates were removed both using Endnote software, and manually during title sorting.Overall, 346 titles from databases were excluded, with five articles identified from individual journal searches also being excluded (Fig. 1).Titles that mentioned SBMA and clinical signs or symptoms and/or pathophysiology, as well as those that mentioned a relationship between CAG repeats and the onset of muscle weakness in male SBMA patients (with or without a female carrier population being part of the study) were included.Given the scarcity of data, all studies, irrespective of the site of onset of weakness (e.g.facial, bulbar or spinal) in the patients analyzed, were considered.Ultimately, 91 articles passed the initial title screening process (Fig. 1 and Table B.2 which is in Appendix B).Articles were then Fig. 1.Flow chart of screening, and criteria for inclusion and exclusion of articles [8].

D.J. Bellai and M.G. Rae
further screened based on their abstract and full text, with 78 articles being further excluded.In all, 13 articles met our inclusion criteria and passed screening (Fig. 1).

Critical appraisal
As all reviewed studies were quantitative in nature, they were graded using the EBL Critical Appraisal Checklist [20] (see Appendix C), which provides an overall score for a study based upon the validity of its population, data collection, study design, and results (see Table B.3).For a study to be designated as valid, it must have an overall validity score of 75%.As shown in Table B.3, four papers failed this test [9,13,17,19] with overall scores of 67% (Doyu et al., 1992), 71% (Fu et al., 2013), 72% (Song et al., 2015) and 62% (Tomik et al., 2006), primarily due to poor population validity, but also due to poor results validity in the cases of Doyu et al. (1992) and Song et al. (2015) [9,17].

Objective 1: To determine if the association between CAG repeat number and age of onset of symptoms of SBMA is significant
All thirteen eligible studies examined the relationship between CAG repeat expansion number and the age of onset of weakness in SBMA patients [5,7,[9][10][11][12][13][14][15][16][17][18][19] and could be divided into two groups: i) studies that used a scoring system to assess weakness and other signs, and, ii) studies that used informal patient interviews and/or chart reviews only.Two studies used a combination of methods [5,14].In the first group, scoring systems included: the activity of daily living scores [9,10,14,17], the Limb Norris score, Norris Bulbar score and ALS functional rating scale-revised [12], the Neuropathy Impairment Score [18], an unspecified standard scoring system [11], and a scale to grade weakness [7].The second group of studies used either patient interviews [7] or both chart reviews and interviews [13][14][15]19], or chart reviews only [16] to screen symptoms and/or signs of SBMA.
With regard to assessing CAG repeat number and the age of onset of weakness in SBMA patients (see Table 1, and Table D.1 in Appendix D), the available results indicated that there were three main outcomes: 1.No significant relationship identified either due to identifying a statistically insignificant inverse correlation [19], or a significant correlation without an r value [7], or no correlation [13]; 2. Identification of either a significant moderate inverse relationship between CAG repeats and the age of onset of weakness [5,9,10,14,[16][17][18], or a strong inverse correlation [11,15] ; and 3. Identification of a relationship based upon phenotype, as with Suzuki et al. (2007), where the number of CAG repeats was compared with weakness onset in the motor-dominant phenotype [12].Note that group 2 also contains results obtained from Lee et al. (2005), which claimed a "weak" correlation (r = − 0.4) [14], as the correlation they identified is classified as "moderate" based on their own published r value (e.g., ± 0.4 to ±0.6 is moderate, based upon accepted r value classification [21])., although all eligible studies did report the number of CAG repeats, they differed based upon the way in which said data was presented such that they can be organized into three main groups: 1.Five studies that listed the ages of participants alongside the minimum and maximum number of recorded CAG repeat number in a retrievable form [7,9,10,14,18]; 2. Seven studies that listed CAG repeat number and age ranges [5,11,13,[15][16][17]19]; and, 3.One study that involved a unique experimental design [12].
In the first group, the article by Sinnreich et al. (2004) reported the shortest number of CAG repeats at 36 (weakness onset at age 46), ranging up to 55 (weakness onset at 38), with the cut-off for consideration being >35 repeats [18].Mariotti et al. (2000) also reported a minimum CAG repeat number in the thirties, ranging from 39 to 50, with the patient with the fewest repeats experiencing weakness at age 50, and the two patients with the most repeats experiencing weakness at ages 35 and 31 respectivelywhile indicating that a normal CAG repeat range was between 12 and 30, and that in SBMA it could be from 39 to 62 [10].Of these five studies, three reported a minimum number of CAG repeats ≥40: Doyu et al. (1992), who reported a range of 40-55, noting that the patient with the highest number of repeats experienced weakness onset at age 25, forty-five years before the patient with the lowest number of repeats (at age 70) [9]; Shimada et al. (1995), who identified a slightly later age-at-weakness-onset (age 49) with a shorter repeat of 41 compared with a patient with the greatest CAG repeat number, 52, and the onset of weakness at age 44 [7]; and, lastly, Lee et al. (2005) who reported the highest number of minimum CAG repeats, 45, and a maximum repeat number of 54, with an age-at-weakness-onset of 55 and 39 respectfully [12].
In the second group of studies, the articles by mean age-at-weakness-onset of 44.4 ± 12.0 (25.0-75.0)years [11].Three of the seven studies reported a minimum number of CAG repeats of exactly 40.Specifically, Sperfeld et al. (2002) proposed that the threshold of normal CAG repeat number is 38 and reported a range of 40-42 CAG repeats [15].Tomik et al. (2006) identified a CAG repeat number range of 45-52, claiming that the normal number of CAG repeats was between 5 and 33, but that for SBMA, it was between 40 and 62 [13].Atsuta et al. (2006) only reported a CAG repeat number range of 40-57 for SBMA, with a mean number of 46.6 ± 3.5 [5].Finally for this group, the two studies with the highest recorded minimum CAG repeat number were Ni et al. (2015), with an average number of CAG repeats of 48.6 ± 3.5 (42-61), and an average age of weakness onset of 44.2 ± 10.2 (24-71) years [16], and Fu et al. (2013), who found a CAG repeat range of 42-53 (mean 47 ± 3) [19].The latter group proposed a normal CAG repeat number range between 14 and 32, with >40 repeats being reported as abnormal [19].
One study, by Suzuki et al. (2007), differed from both aforementioned groups.This study subdivided patients with a CAG repeat number < 47 (short repeats) from those with a repeat number ≥ 47 and then conducted electromyographic and nerve conduction studies on both groups separately [12].They reported a mean CAG repeat number of 47.8 ± 3.1, with a range of 41-57 CAG repeats [12].

Discussion of results
CAG repeat number may account for approximately 60% of clinical heterogeneity in SBMA patients, with genetic, environmental and epigenetic factors also likely to play some role in influencing disease progression [1].However, the contribution of these relative unknowns aside, determining the correlation between the most commonlymentioned SBMA symptom, weakness, and the primary clinical marker for the condition, CAG repeat number on the AR gene, might potentially uncover a strong indicator of disease onset that could enable early SBMA detection and, thereafter, management.As such, this review is important and timely given that no other review to date has specifically investigated a possible connection between the age of onset of weakness in SBMA and CAG repeat number.
All of the studies utilized for this review addressed our first objective, with 11 of the 13 indicating that a significant inverse correlation did indeed exist between CAG repeat number and onset of SBMA weakness [5,7,[9][10][11][15][16][17][18] (Table 1).With regard to this objective, our review presented two major issues.Firstly, a surprising number of articles failed screening.Although the study by Doyu et al. (1992) only just failed our initial screening due to poor population validity (56%), data collection validity (63%) and results validity (67%) (overall score 67%) (Table B.3, Appendix B) [9], we felt that the use of disability scores to assess clinical severity in this paper, combined with the limited number of studies available, mitigated against its exclusion.Notably, a single study did find that shorter CAG repeat number (<47) was associated with an older onset of weakness [12].Interestingly, of the three studies that failed to find a strong, or even a moderate, statistically significant correlation between CAG repeat number and age of onset of weakness, two had poor external validity [13,19].For example, the overall validity for the article by Tomik et al. (2006) was only 62%, primarily due to its poor population validity (33%; Table B.3, Appendix B), resulting mainly from the lack of agematched healthy controls to compare EMG results against, as well as the small sample size (11 SBMA patients and three carriers, see Table D.1 in Appendix D).Similarly, the study by Fu et al. (2013), also had a relatively low overall validity score of 71%, also mostly due to its poor external validity (50%) (Table B.3) due to its small population size and retrospective nature [19].Although the third study that did not find a statistically significant correlation, by Lee et al. (2005), did pass screening validity with an overall validity score of 78% it also had a low ) due to sample size issues and a lack of accountability for confounding factors [17].
Our review of included articles also revealed a second major issue: the need for larger scale studies to better characterize SBMA.Indeed, Fu et al. (2013) suggested that any future characterization of the functional progression, and the clinical and electrophysiological aspects of SBMA, be conducted using larger scale prospective studies [19].Similar sentiments were expressed by Song et al. (2015) and Lee et al. (2005) [14,17].
With regard to our second objective, the majority of studies presented both CAG repeat number data and ages of onset of weakness, such that these data could be compared across studies to determine the minimum and maximum number of repeats at age of weakness onset in SBMA [7,9,10,14,18], apart from two articles [11,17].Of the five studies that addressed the objective, CAG repeat number ranged from the mid-thirties to the mid-fifties, coinciding with the accepted marker of ≥38 repeats used to indicate the presence of SBMA [1], with all five studies claiming that patients with larger repeat numbers had an earlier onset of weakness [7,9,10,14,18].There has been some data collected with regard to the effect of the number of CAG repeats, with median lengths varying in various racial groups (as reported by the Neuromuscular Disease Center at Washington University): White Caucasian (21− 22), African American (19)(20), Hispanic (23) and Asian (22− 23) [22].They state the CAG repeat range for SBMA to be 40-68 (closely approximating the ranges reported in this review) while claiming that the CAG repeat number is associated with earlier and more severe disease onset but not specific clinical features [22].

Future considerations
We believe that any future SBMA studies investigating a possible connection between CAG repeat number and age of weakness onset should consider the potential effects of mosaicism as this may help to explain atypical presentations of SBMA (i.e. a patient with 26 CAG repeats and symptoms of SBMA) [13].Larger sample sizes would also be desirable in order to enable more accurate measurement of the decline in motor function in SBMA patients over time, a claim repeated by multiple articles (Table D.1, Appendix D).Furthermore, age-matched controls and/or controls with other muscular atrophies should be included to provide appropriate baseline comparison data during electromyographic or nerve conduction studies.Finally, one could also mitigate against error due to the use of a variety of methods to assess function in SBMA patients (complicating inter-study comparisons) by minimizing recollection bias when reporting symptoms by combining patient interview data, chart reviews and standardized scoring systems.

Limitations
One limitation of this review was its language restriction (English) such that sixteen non-English SBMA articles did not pass screening (Fig. 1).Another limitation was the inclusion of articles that were only available online, which meant that two potentially useful studies were excluded from our analysis (Fig. 1).In addition, many of the articles that were included were of Asian origin, which could arguably have introduced an ethnic genotype bias, potentially impacting the external validity of the study.However, one must note that SBMA is a rare illness, such that few institutions are likely to have the resources available to justify its study.
Finally, four articles that failed initial screening [9,13,17,19] were included in this review.Studies that used only chart reviews and/or patient interviews, but not an objective scoring system to assess symptoms [7,[13][14][15][16]19], are subject to recollection bias, which may have impacted the recorded age at weakness onset and the correlation of interest.

Conclusion
Weakness is the most commonly complained of symptom among SBMA patients [1].This review of the relevant SBMA literature has shown that the majority of studies investigating this facet of SBMA pathophysiology have demonstrated an inverse correlation between the number of CAG repeats and the age of weakness onset in SBMA patients.Furthermore, most of the aforementioned studies have proposed that the minimum number of CAG repeats associated with SBMA-linked weakness falls within at least the mid-to-late thirties range or greater.By exploring and revealing the relationship between this reliable marker of SBMA and a common clinical sign of the condition as we have done here, this review not only adds to the relatively scant literature on SBMA, but also, significantly, may facilitate earlier detection of the condition and thereby facilitate its prompt treatment.

Declaration of competing interest
None.This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Table 1
Summary of studies that are included in this review article.

Table 1 (
continued ) 3. Inverse correlation between creatine kinase level and disease duration, and age at examination (p = 0.019 and p = 0.004, respectively).Additionally, all nerve conduction findings, except the (continued on next page) D.J. Bellai and M.G.Rae 1.3.3.2.Objective 2: To determine what minimum and maximum number of CAG repeats are associated with weakness onset in SBMA As shown in Table 1 (and Table D.1 in Appendix D) Song et al. (2015)2018)senbohm et al. (2018)both reported the lowest minimum CAG repeat number of the group at 39[11,17].However, their reported age ranges and average CAG repeat number differed, withSong et al. (2015)reporting a mean of 44 CAG repeats (39-55) and an average age-at-onset of symptoms of SBMA of 44.5(20.0-71. 0) years [17], while Rosenbohm et al. (2018) reported a mean length of 46.2 ± 3.3 (39.0-55.0),and a

Table 1
(continued ) D.J.Bellaiand M.G.Rae population validity score (60%; Table 1 and Table D.1) stemming from its small sample size [14].Other articles that failed screening included Song et al. (2015), who reported only a moderate correlation between CAG repeat length and SBMA weakness (overall score 72%; Table 1 and Table D.1), derivative of both poor population and results validity (both 50%

Table B . 3
Validity score for quantitative studies using the EBL critical appraisal checklist.

for section validity: (YþNþU¼T)
If Y/T <75% or if N+U/T > 25% then you can safely conclude that the section identifies significant omissions and that the study's validity is questionable.It is important to look at the overall validity as well as section validity.

for overall validity: (YþNþU¼T)
If Y/T ≥75% or if N+U/T ≤ 25% then you can safely conclude that the study is valid.