Glycolipid and Hormonal Profiles in Young Men with Early-Onset Androgenetic Alopecia: A meta-analysis

Hormonal and metabolic abnormalities have been reported in men with early-onset androgenetic alopecia (AGA). Although this has been ascribed to the existence of a male polycystic ovary syndrome (PCOS)-equivalent, data on this topic are inconsistent and this syndrome has not been already acknowledged. To evaluate if, already before the age of 35 years, any difference occurs in the glycolipid and hormonal profiles and in the body weight in men with AGA compared to age-matched controls, we performed a comprehensive meta-analysis of all the available observational case-control studies of literature, using MEDLINE, Google Schoolar and Scopus databases. Among 10596 papers retrieved, seven studies were finally included, enrolling a total of 1009 participants. Our findings demonstrate that young men with AGA have a slightly but significantly worse glycolipid profile compared to controls and a hormonal pattern resembling those of women with PCOS, already before the age of 35 years. Therefore, early-onset AGA might represent a phenotypic sign of the male PCOS-equivalent. The acknowledgement of this syndrome would be of importance to prevent the long-term consequences on health in the affected men. The glycolipid profile and the body weight should be monitored in men with AGA starting from the second decade of life.

contacted the authors to know if they had the values of the entire cohort of patients with AGA, but we did not receive any answer. Finally, the study by Hirsso and colleagues 18 was not included since the group of control also consisted of men with I and II degree AGA.
Seven studies 9, 10, 19-23 met our inclusion criteria and were eligible to be included in this systematic review, with a total of 1009 participants: 522 were men with early-onset AGA and 487 were controls (Fig. 1). However, not all the studies included evaluated the same parameters. A higher number of articles investigating the metabolic profile was found compared to those investigating the hormonal status. We did not considered in the meta-analysis the total testosterone levels of the study by Banger and colleagues 22 and the SHBG levels of the study by Gonzalez-Gonzalez and colleagues 9 , since they were not comparable with those of the other studies. We unsuccessfully tried to get in touch with the authors.

Discussion
The results of this meta-analysis showed the presence of a slightly worse glycemic and lipid profile and of slightly higher BMI values in men with early-onset AGA. Moreover, although on a small number of patients and controls, we also observed significantly higher LH and DHEAS levels, and lower SHBG levels, a downward trend for the FSH values and an upward trend for the LH/FSH ratio. Interestingly, these abnormalities were already present in a group of young ( <35 years old) men.  A meta-analysis on a cohort of 4006 men showed a 2-fold higher risk of developing MetS in men with AGA 25 . On the contrary, other authors failed to find this association in cohorts of 116 27 , 1707 28 , 184 29 , and 120 30 men. However, Hirsso and colleagues described lower insulin-sensitivity in men with AGA 31 . Accordingly, Abicucu and colleagues reported a higher prevalence of insulin-resistance and MetS in men with AGA aged 20-50 years compared to age-matched controls 11 . Similar results have been reported also by other studies 19,32 .

Metabolic abnormalities
Male AGA has been associated with hypertension 33, 34 (odds ratio 2.195 33 ), atherosclerosis and higher intima-media thickness 17,35 , dyslipidemia, ischemic heart disease in men less than 50 years old 34 , and, in a cohort of 424 men younger than 45 years, with coronary heart disease 36 . In addition, data from two large epidemiologic studies, the Framingham study 37 and the NHANES I Epidemiologic Follow-up study 38 revealed a positive correlation between AGA and coronary heart disease in men.
In line with these evidences, the results of this meta-analysis support the hypothesis that men with early-onset AGA are at risk for metabolic impairment, having a worse (although still normal) metabolic profile already in the second decade of life compared to controls. Therefore, they should be evaluated for glycemic and lipid profiles and their BMI should be monitored early in life to prevent long-term metabolic and cardiovascular consequences on health.
Hormonal alterations and male PCOS-equivalent. Although based on a small cohort, the data of this meta-analysis suggest the presence of a hormonal pattern partially resembling that of women with PCOS.
The higher levels of DHEAS found in men with early-onset AGA compared to controls suggests that this hormone might play a role in the development of AGA. In fact, in the hair follicle, DHEAS is converted into molecules having higher androgen activity (testosterone, dihydrotestosterone) 39 . Moreover, these higher DHEAS levels may be involved in the pathogenesis of the observed slightly worse metabolic profile and might represent a biochemical feature of the male PCOS-equivalent. Indeed, obesity is associated with the hyperactivity of the hypothalamic-pituitary-adrenal (HPA) axis in men 40,41 . The hyperactivity of the adrenal glomerular, fascicular and reticular zones has been hypothesized to play a role in the development of insulin-resistance and obesity 42 . Conversely, free fat acids (FFA) (which obese subjects are exposed to) seem to stimulate the reticular zone of the adrenal cortex. In fact, in vitro studies on human cortical adrenal cells showed that the exposure to the oxidized fat acid EXODE increases basal and ACTH-stimulated synthesis of dehydroepiandrosterone (DHEA), while it does not have any effect on basal cortisol release. Moreover, ACTH-stimulated synthesis of cortisol decreases 43 .

Figure 2.
Glycemic profile and body mass index in young men with androgenetic alopecia compared to controls. Men with early-onset androgenetic alopecia younger than thirty-five years old showed significantly higher insulin serum levels, HOMA index, a marker of insulin-resistance, and body mass index values compared to age-matched controls. Insulin was expressed in µUI/ml; Body mass index was expressed in Kg/m 2 .
Accordingly, the in vivo exposure to high doses of FFA increases the production of adrenal androgens (DHEA and androstenedione) in humans 44 . Finally, the treatment with pioglitazone, an insulin-sensitizer, showed to decrease ACTH-stimulated serum DHEAS levels in adult female resus monkeys 45 , thus showing the stimulatory effect of insulin-resistance on the reticular cortex function in the animal model.
These evidences suggest that the increase in DHEAS serum levels observed in men with early-onset AGA might play a role in the increase of insulin, HOMA-index, BMI. On the contrary, also obesity and insulin-resistance may be effective in increasing DHEAS levels.
The lower levels of SHBG found in men with early-onset AGA may represent a sign of metabolic imbalance. In fact, low levels of SHBG have been proposed as a marker of insulin-resistance and hyperglycemia/DM II in patients with AGA 13 . Along this line, insulin has shown to inhibit the SHBG synthesis in the liver 46 . These lower levels might represent another biochemical sign of the male PCOS-equivalent.
There is a great amount of evidence showing the association between AGA and metabolic-cardiovascular impairment. The present meta-analytic study demonstrates that a worse (but still normal) glycolipid profile already exist before the age of 35 years in men with AGA compared to controls. This association might be due to the existence of the male PCOS-equivalent. As a matter of facts, although only few studies evaluated the hormone profile in men with early-onset AGA since now, the evidence coming from such researches suggest the presence of a hormonal profile resembling that of women with PCOS in these men.  . Hormonal profile in young men with androgenetic alopecia compared to controls. Men with earlyonset androgenetic alopecia younger than thirty-five years old showed significantly higher luteinizing hormone (LH) and dehydroepiandrosterone sulphate (DHEAS), and lower sex hormone binding globulin (SHBG) serum levels compared to controls. A downward trend for serum follicle-stimulating hormone (FSH) and an upward trend for the LH/FSH ratio was found. No difference in testosterone levels was observed among the two groups. Both LH and FSH were expressed in mIU/ml, DHEAS in µg/ml, SHBG in nmol/l, total testosterone in ng/ml. of about the 70% and it is inherited thought an oligogenic mechanism 49 . The genetic background responsible for the susceptibility to the PCOS may be inherited also by men, and, therefore, we speculate that the early-onset AGA could represent a phenotypic sign of the male PCOS-equivalent, a complex syndrome with a metabolic background. This would explain the glycolipid and hormonal profile found in men with early-onset AGA, and also the association between AGA and cardiovascular diseases. The acknowledgement of this syndrome would be of importance to prevent the long term consequences on health of these men.
Limits of the study. The major limit of this study was the small cohort, especially for the evaluation of the hormonal profile. This was also due to the decision to include only studies on men younger than 35 years. We used this strict cut-off to evaluate if any difference in the glycolipid profile in men with early-onset AGA compared to controls could be already found early in life.
In addition, five among the seven studies evaluated in this meta-analysis did not include BMI-matched controls. This may represent a bias since both glycemic and lipid profiles might primarily relate to the higher BMI found in men with early-onset AGA compared to controls. However, the BMI-matched case-control studies 9, 10 both reported a significantly higher HOMA-index in men with early-onset AGA compared to controls, thus showing that this finding is independent from BMI.

Conclusive remarks
The results of this meta-analysis showed the presence of slightly worse glycemic and lipid profiles in men with early-onset AGA. This finding is in line with the strong association between early-onset AGA and MetS or CVDs. Hence, monitoring the glycemic and lipid profiles and the BMI in men with early-onset AGA should be suggested to early detect any potential metabolic impairment.
The mechanism by which early-onset AGA is associated with metabolic and cardiovascular abnormalities is still not known. Based also on the hormonal findings, we hypothesize that these men may display the male PCOS-equivalent. However, the presence of hormonal alterations in these men need to be further investigated.

Methods
Sources. This study was performed using the MOOSE Guidelines for Meta-analyses and systematic reviews of observational studies. Data were independently extracted by A.E.C. and R.C. A systematic search was performed though the MEDLINE, Google Scholar, and Scopus databases, from each database inception to April 30, 2017.  Table 1. Summary of the studies included and their quality assessment a . Abbreviations: AGA = androgenetic alopecia; BMI = body mass index; CRP = C-reactive protein; DHEAS = dehydroepiandrosterone sulphate; ESR = erythrocyte sedimentation rate; FAI = free androgen index; FBS = fasting blood sugar; FIRI = fasting insulin resistance index; HDL-C = HDL cholesterol; LDL-C = LDL cholesterol; SHBG = sex hormone binding globulin; total-C = total cholesterol; VLDL = very low-density lipoproteins; WC = waist circumference; WHR = waist to hip ratio. Values in square brackets represents the standard deviation, whereas those in round brackets represents the mean difference. a The quality assessment of individual studies was based on the Ottawa-Newcastle scale 52 . Possible scores are 0 to 4 asterisks for selection, 0 to 2 for comparability, 0 to 3 for exposure or outcome. One asterisk indicates the lowest score, 4 asterisks the highest score. Studies with a score < 5 were considered to have high risk of bias, between 5 and 7 moderate risk, > 7 low risk of bias. b The degree of androgenetic alopecia was evaluated though the Hamilton-Norwood Scale 50, 51 .
Our search strategy was based on the following key words: "male androgenetic alopecia", "male premature baldness", "insulin", "insulin-resistance" and "polycystic ovary syndrome male equivalent". Additional manual searches were made using the reference lists of relevant studies. No language restriction was used for any literature search.
The authors were contacted for missing data.
Study selection. Information on the year of publication, country, continent, gender, study design, mean age of cases and controls was collected. Studies which met the following inclusion criteria were included in the meta-analysis: 1. Design: observational case-control studies on men with (cases) and without (controls) early-onset AGA, evaluated through the Hamilton-Norwood scale 50, 51 . 2. Age of cases and controls lower than 35 years. 3. No treatment with finasteride or any other medication. 4. no major comorbidities (including thyroid, pituitary or adrenal disorders, liver or kidney failure) in both cases and controls. 5. No treatment with weigh loss or insulin sensitizer drugs, glucocorticoids, androgens and antiandrogens in both cases and controls. 6. No other types of alopecia apart from the androgenetic one.
Studies that did not meet these criteria were excluded. The quality assessment of the studies included in the present meta-analysis was performed with the Ottawa-Newcastle scale, which evaluates three separate domains that refer to selection of study groups, comparability of groups and ascertainment of outcomes 52 . The maximum score is 9. Studies with a score < 5 were considered to have high risk of bias, between 5 and 7 moderate risk, > 7 low risk of bias ( Table 1).
The Cochran-Q and I 2 statistics were used for the assessment of statistical heterogeneity. Specifically, statistical heterogeneity was tested using the chi-square test. If I 2 ≤ 50%, the variation of the studies was considered to be homogenous, the fixed effect model was adopted. If I 2 > 50%, there was significant heterogeneity between studies, the random effects model was used. All P values ≤ 0.05 were considered statistically significant. The analysis was performed using RevMan software v. 5.3 (Cochrane Collaboration, Oxford, UK). For each outcome, the standard MD with the 95% CI was calculated.