Sex Hormones Associate With Amyotrophic Lateral Sclerosis Risk and Survival

ABSTRACT

Amyotrophic lateral sclerosis (ALS) risk differs by sex and age, implicating sex hormones as potential modifiers. This study examined plasma levels of biologically active sex hormones and their association with ALS odds and survival in cases (females n = 131, males n = 189) and controls (females n = 138, males n = 150) from the University of Michigan Pranger ALS Clinic. Higher 11‐ketotestosterone levels were associated with increased ALS odds. In females, higher estrone, androstenedione, and 11‐hydroxyandrostenedione were associated with increased ALS odds, while elevated estrone and estradiol predicted shorter survival. These findings highlight the potential significance of sex hormones in ALS.

1. Introduction

Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disease resulting from genetic and environmental factors [1, 2]. ALS affects more males than females [3, 4]. This sex gap narrows around menopause, raising the possibility that sex hormones may influence ALS [5, 6]. The goal of the present study is to understand differences in androgens and estrogens between participants with ALS compared to controls and associations with ALS progression.

2. Methods

Cohort details are published [7]. Briefly, all patients with an El Escorial diagnosis of ALS and able to communicate in English seen at the University of Michigan Pranger ALS Clinic were invited to provide blood samples to the University of Michigan ALS Patient Biorepository. Control participant recruitment utilized population outreach to identify controls with a similar age and sex distribution to ALS cases. All participants provided informed consent (University of Michigan IRB approval HUM28826). Recorded participant information included age, sex, and race; further from ALS participants, the following information were recorded: onset segment, age at symptom onset, age at diagnosis, ALS family history, El Escorial criteria, and ALS Functional Rating Scale‐Revised (ALSFRS‐R) score. Blood samples were collected by peripheral venipuncture between January 02, 2019 and December 20, 2023, and plasma was isolated, aliquoted into cryovials, and stored at −80°C. Steroid sex hormone quantitation by liquid chromatography–tandem mass spectrometry was performed as published (Supplemental Methods) [8]. Supplemental Methods provides statistical method details. Briefly, participant demographics and sex hormone levels were summarized by groups. Sex hormones were correlated to age, ALSFRS‐R, and disease duration. Adjusted logistic regression and Cox regression models assessed ALS risk and survival, and several sensitivity analyses were performed. Benjamini‐Hochberg procedure adjusted p‐values for multiple comparisons and is denoted as p _adjusted.

3. Results

3.1. Participants and Samples

One hundred thirty‐one female and 189 male ALS and 138 female and 150 male control participant samples were analyzed (Table 1). Participants with ALS were older (Table S1).

TABLE 1.

Participant demographics by sex and case group.

	ALS female, N = 131 a	ALS male, N = 189 a	Control female, N = 138 a	Control male, N = 150 a	p b
Age at sample (years)	66.9 (59.9–73.3)	64.4 (57.9–72.7)	60.6 (54.6–65.8)	62.3 (56.9–69.7)	< 0.001
Race
Other	8 (6.1%)	8 (4.2%)	10 (7.2%)	7 (4.7%)	0.6
White or Caucasian	123 (94%)	181 (96%)	128 (93%)	143 (95%)
Age at diagnosis (years)	66.4 (59.0–72.4)	63.8 (56.7–71.9)			0.3
El Escorial criteria
Definite	58 (44%)	53 (28%)			0.011
Probable	42 (32%)	66 (35%)
Probable, lab supported	24 (18%)	55 (29%)
Possible	4 (3.1%)	13 (6.9%)
Suspected	3 (2.3%)	2 (1.1%)
Onset segment
Bulbar	42 (32%)	31 (17%)			< 0.001
Cervical	24 (18%)	80 (43%)
Lumbar	61 (47%)	70 (37%)
Respiratory	2 (1.5%)	3 (1.6%)
Thoracic	2 (1.5%)	3 (1.6%)
Family history of ALS	10 (7.9%)	20 (11%)			0.4
ALSFRS‐R	36 (31–40)	38 (33–42)			0.027
Time between symptom onset and diagnosis (years)	0.99 (0.69–1.59)	1.04 (0.56–1.90)			0.9
Time between diagnosis and sample (years)	0.35 (0.08–0.88)	0.37 (0.10–0.90)			0.5
Status
Censored	31 (24%)	64 (34%)			0.050
Observed death	100 (76%)	125 (66%)
Follow‐up time from diagnosis (years)	1.65 (1.01–2.75)	2.17 (1.33–3.27)			0.012
Censored	2.75 (1.59–4.53)	2.91 (2.10–4.16)			0.7
Observed death	1.47 (0.85–2.36)	1.90 (1.07–2.54)			0.059

Abbreviations: ALS, amyotrophic lateral sclerosis; ALSFRS‐R, ALS Functional Rating Scale‐Revised.

^a Median (Q1–Q3); n (%).

^b Kruskal‐Wallis rank sum test; Pearson's Chi‐squared test.

3.2. Steroid Analyses and Associations

We measured traditional sex hormones, including estrone, estradiol, testosterone, and androstenedione, as well as adrenal‐derived 11‐ketotestosterone and 11‐hydroxyandrostenedione in participant samples (Table S2, Figure S1). Estradiol was frequently below detection limits in females and males and lower in female ALS versus controls. Testosterone was higher in males. In females and males, 11‐ketotestosterone was higher in ALS versus controls; 11‐hydroxyandrostenedione levels were higher in ALS females versus controls. Estrone and estradiol decreased with age in control females, and androstenedione, but not testosterone, decreased with age for all groups except female ALS cases (Figure S2). Most sex hormone precursor–product pairs were correlated (Figure S3).

11‐Ketotestosterone was associated with ALS (odds ratio [OR] = 1.76, 95% confidence interval [CI], 1.44–2.14, p _adjusted < 0.001) (Figure 1A) and remained significant when stratified by sex. Estrone (OR 1.32, 95% CI 1.04–1.68, p _adjusted = 0.040), androstenedione (OR 1.36, 95% CI 1.04–1.76, p _adjusted = 0.040), and 11‐hydroxyandrostenedione (OR 1.36, 95% CI 1.04–1.78, p _adjusted = 0.040) were significantly associated with ALS in females. No other sex hormones were associated with ALS. Matched (Table S3, Figure S4) and postmenopausal age female (Figure S5) sensitivity analyses were consistent with the main model, although estrone, androstenedione, and 11‐hydroxyandrostenedione were not significant in the female matched risk analysis. However, the female‐age‐greater‐than‐65‐year model showed androstenedione and 11‐hydroxyandrostenedione, and the age‐greater‐than‐55‐year and age‐greater‐that‐60‐year models showed estrone, testosterone, androstenedione, and 11‐hydroxyandrostenedione as significant associative factors with ALS status despite the smaller sample sizes.

FIGURE 1.

Steroid risk and survival analyses. (A) Covariate‐adjusted logistic regression analysis where each individual steroid marker is linked to ALS risk. Steroid markers are log‐transformed and standardized. The analysis considers three groups: overall (320 ALS and 288 controls), female (131 ALS and 138 controls), and male (189 ALS and 150 controls). The adjustment covariates include age at sample for all groups and sex for the overall group. (B) Covariate‐adjusted Cox regression analysis relates each individual steroid marker to years since diagnosis. Steroid levels are log‐transformed and standardized. The models are adjusted for age at diagnosis, family history of ALS, diagnostic ALSFRS‐R score, time between symptom onset and diagnosis, diagnostic El Escorial criteria (definite vs. non‐definite), onset segment (bulbar vs. non‐bulbar). The analysis considers three groups: overall (n = 311), female (126), and male (185). In the overall group analysis, sex is included as an additional covariate. OR, odds ratio corresponding to one standard deviation increase in log‐transformed steroid markers; CI, confidence interval; Q‐value (BH), p‐values adjusted for multiple comparison using the Benjamini‐Hochberg method; HR, hazard ratio corresponding to one standard deviation increase in log‐transformed steroid markers. Separate corrections for multiple comparisons are performed for each group. p‐Values and Q‐values less than 0.05 are marked in bold. Q‐values less than 0.05 are highlighted in red.

In ALS‐only analysis, there were modestly significant associations with disease duration for estrone in males (R = 0.14, p‐value = 0.018), 11‐ketotestosterone in females (R = −0.25, p‐value < 0.001), and 11‐hydroxyandrostenedione in females (R = −0.27, p‐value < 0.001) (Figure S6). There were modestly significant correlations with ALSFRS‐R scores for testosterone in males (R = 0.21, p‐value = 0.002), androstenedione in females (R = 0.29, p‐value < 0.001) and males (R = 0.15, p‐value = 0.031), and 11‐hydroxyandrostenedione in females (R = 0.23, p‐value = 0.033) (Figure S7). Covariate‐adjusted Cox proportional hazard model showed in females that higher estrone (hazard ratio [HR] = 1.48, 95% CI 1.15–1.91, p _adjusted = 0.008) and estradiol (HR = 1.47, 95% CI 1.15–1.88, p _adjusted = 0.008) (Figure 1B, Figure S8)—but no other sex hormones—were associated with survival. Survival sensitivity analyses showed similar effect sizes for females in all postmenopausal age categories (Figure S9) and those without estradiol use (Figure S10). Finally, internal validation analysis showed internal reproducibility and robustness of these findings (Table S4).

4. Discussion

Here, higher levels of the sex hormone 11‐ketotestosterone in males and females and high levels of androstenedione, 11‐hydroxyandrostenedione, and estrone in females are associated with ALS case status. Further, higher levels of estrone and estradiol in females are associated with poorer ALS survival. To our knowledge, these associations are novel. In males, the testes‐derived testosterone is the major circulating androgen. Thus, the contribution of 11‐ketotestosterone to total androgen supply in males is small because testosterone concentrations in men are 10 times higher than 11‐ketotestosterone [9, 10]. However, in females, 11‐ketotestosterone, which is a metabolite of the purely adrenal‐derived 11‐hydroxyandrostenedione (Figure S11), is the major circulating and bioactive androgen. It is typically about 10% higher than testosterone and does not show an age‐related decline [9, 10]. Produced in the adrenal gland, 11‐ketotestosterone is the downstream product of the biologically inactive dehydroepiandrosterone (DHEA) and a metabolite of purely adrenal‐derived 11‐hydroxyandrostenedione [9]. Thus, it is also unsurprising that in females 11‐hydroxyandrostenedione was a significant finding. Additionally, in females, estrone and androstenedione are associated with greater ALS odds. Estrone is the major postmenopausal estrogen formed via adrenal androstenedione [11]. This is the largest study, to our knowledge, of sex hormone assessments in ALS and is the first to measure these adrenal‐derived androgens and traditional androgens and estrogens in ALS.

Few prior studies have measured sex hormones in ALS cohorts; a small study of 35 ALS participants (25 males, 10 females) and 57 controls (38 males, 17 females) measured dehydroepiandrosterone sulfate (DHEAS), total and free testosterone, and estradiol from serum and found decreased levels of free testosterone in both male and female ALS participants [12]. However, this study used direct radioimmunoassays, which are unreliable, particularly for free testosterone, compared to mass spectrometry [13]. Another small study of 64 males showed that 60% of ALS participants had lower total testosterone levels than age‐matched healthy males; further, higher testosterone levels were associated with a higher ALSFRS‐R [14]. A study of 11 ALS participants compared to 10 controls showed higher total and free (by calculation with sex hormone binding globulin [SHBG]) testosterone levels in female ALS cases versus controls [15]. Another study found high DHEAS relative to age‐matched normal ranges in 27 females but not in 32 males with ALS [16]. Our study differs from these by simultaneously assessing traditional androgens and estrogens and adrenal‐derived 11‐oxygenated androgens using mass spectrometry. DHEA, DHEAS, 11‐hydroxyandrostenedione, and 11‐ketotestosterone are all adrenocorticotropic hormone (ACTH)–dependent adrenal androgens, and measures of DHEA and DHEAS correlate with those of 11‐ketotestesterone [17]. The emerging theme from these studies is that higher androgens are an ALS risk factor in females, which reflects the lower prevalence of ALS in females. Our study is the first to show this association for 11‐ketotestosterone, the major bioactive androgen in females. We also found higher testosterone in ALS females, but the CI for this association included 1.

Analyses also showed higher estrone, androstenedione, and 11‐hydroxyandrostenedione associated with greater ALS risk in females, and higher estrone and estradiol associated with poorer ALS survival in females, which have not been assessed in prior studies to our knowledge. Several studies report on ALS risk. One European study found female users of oral contraception pills were at a reduced ALS risk, with the risk decreasing as the duration of therapy increased [5]; and hormone replacement therapy was associated with a reduced ALS risk in the Netherlands [5]. A separate study of females in the Netherlands (209 ALS, 672 controls) showed that a longer reproductive window (therefore longer estrogen exposure) was associated with a decreased ALS rate [6]. Alternatively, this negative effect of estrogen on ALS survival could be age‐related, as a Medicare Part D prescription claims study in persons greater than 66 years old showed that females using tamoxifen (which blocks estrogen) had a reduced ALS risk at 1 and 3 years, with an OR of 0.62 and 0.61, respectively [18]. In this case, estrogens may have differing effects based on menopause status.

Given the detrimental role of inflammation in ALS [7], it is notable that these androgens have overall pro‐inflammatory effects [19]. Importantly, in females, the pro‐inflammatory effects of androgens oppose the anti‐inflammatory actions of estradiol [11]. Additionally, although estrone was previously considered only as a precursor to estradiol, studies from the past 10 years have demonstrated pro‐inflammatory effects of estrone as a ligand for estrogen receptor alpha [11]. Additionally, as androstenedione and 11‐hydroxyandrostenedione are precursors of androgens, their association with ALS risk may likewise be related to inflammatory mechanisms [19]. Potentially supporting this anti‐inflammatory effect of estrogen are studies of oral contraception and hormone replacement therapy discussed above.

This study has limitations. Samples were not collected fasting nor at uniform times to minimize effects of diurnal variation, as this is difficult to achieve in large cohorts. While we could not account for sex hormone changes during the ovarian cycle, most female participants were postmenopausal, and sensitivity analyses controlling for menopause were consistent with the overall population. Unlike DHEA, DHEAS, and traditional androgens, 11‐ketotestosterone does not decline with age, making this androgen the relevant biomarker for androgen action in older females [20]. Given that the female estrone, androstenedione, and 11‐hydroxyandrostenedione risk results were not seen in the matched cohort analysis, future cohort studies would be warranted. Finally, this study is cross‐sectional and associative and therefore does not prove causality, meaning future longitudinal studies are necessary to determine the temporal changes in these hormone levels. This study also has strengths, including a large cohort size, comprehensive measurement of biologically active sex hormones, and strong internal validity.

In conclusion, we find that higher 11‐ketotestosterone is associated with a greater ALS risk, particularly in females, and higher estrone, androstenedione, and 11‐hydroxyandrostenedione are also associated with ALS risk in females. Additionally, in females, higher estrone and estradiol are associated with a poorer ALS survival. These findings may have important implications for our understanding of factors related to ALS risk and survival, prevention strategies, and early intervention.

Author Contributions

Conceptualization: S.A.G. and B.J.M. Funding acquisition: S.A.G., B.J.M., and R.J.A. Methodology: S.A.G. and R.J.A. Data curation: D.G.S. Formal analysis: S.A.G., D.‐G.J., and J.P. Writing – original draft preparation: S.A.G. and D.‐G.J. Writing – review and editing: All authors. S.A.G. is the guarantor of the study.

Funding

This work was supported by CDC/ATSDR (R01TS000339), U.S. Department of Defense (AL200064), Eric and Linda Novak, James and Margaret Hiller, Robert A. Epstein and Joan M. Chernoff‐Epstein Emerging Scholar Fund, Stanford Morris ALS Research Fund, National Institutes of Health (K23ES027221, R01ES030049, R01NS120926, R01NS127188).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supplemental Methods: acn370281‐sup‐0001‐Supinfo.docx.

Figure S1: acn370281‐sup‐0002‐FigureS1.docx.

Table S1: acn370281‐sup‐0003‐TableS1.docx.

Table S2: acn370281‐sup‐0004‐TableS2.xlsx.

Table S3: acn370281‐sup‐0005‐TableS3.docx.

Table S4: acn370281‐sup‐0006‐TableS4.xlsx.

Funding Statement

This work was funded by CDC/ATSDR grant R01TS000339; U.S. Department of Defense grant AL200064; Eric and Linda Novak ; James and Margaret Hiller ; Robert A. Epstein and Joan M. Chernoff‐Epstein Emerging Scholar Fund ; Stanford Morris ALS Research Fund ; National Institutes of Health grants K23ES027221, R01ES030049, R01NS120926, and R01NS127188; University of Michigan .

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.