Although the genetic basis of androgenic alopecia has been clearly established, little is known about its non-genetic causes, such as environmental and lifestyle factors.
This study investigated blood and urine heavy metals concentrations, environmental exposure factors, personal behaviors, dietary intakes and the genotypes of related susceptibility genes in patients with androgenic alopecia (AGA).
Age, AGA level, residence area, work hours, sleep patterns, cigarette usage, alcohol consumption, betel nut usage, hair treatments, eating habits, body heavy metals concentrations and rs1998076, rs913063, rs1160312 and rs201571 SNP genotype data were collected from 354 men. Logistic regression analysis was performed to examine whether any of the factors displayed odds ratios (ORs) indicating association with moderate to severe AGA (≧IV). Subsequently, Hosmer-Lemeshow, Nagelkerke R2 and accuracy tests were conducted to help establish an optimal model.
Moderate to severe AGA was associated with the AA genotype of rs1160312 (22.50, 95% CI 3.99–126.83), blood vanadium concentration (0.02, 95% CI 0.01–0.04), and regular consumption of soy bean drinks (0.23, 95% CI 0.06–0.85), after adjustment for age. The results were corroborated by the Hosmer-Lemeshow test (P = 0.73), Nagelkerke R2 (0.59), accuracy test (0.816) and area under the curve (AUC; 0.90, 0.847–0.951) analysis.
Citation: Lai C-H, Chu N-F, Chang C-W, Wang S-L, Yang H-C, Chu C-M, et al. (2013) Androgenic Alopecia Is Associated with Less Dietary Soy, Higher Blood Vanadium and rs1160312 1 Polymorphism in Taiwanese Communities. PLoS ONE 8(12): e79789. https://doi.org/10.1371/journal.pone.0079789
Editor: Qinghua Sun, The Ohio State University, United States of America
Received: February 1, 2013; Accepted: September 25, 2013; Published: December 30, 2013
Copyright: © 2013 Lai et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This research was supported by Taichung County Government in Taiwan (5168-038). The interpretation and conclusions contained herein do not represent those of Bureau of Health, Taichung County or National Health Research Institutes. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
The incidence of androgenic alopecia (AGA) is increasing, while the age of onset of AGA continues to decrease. Studies have associated AGA with a variety of diseases, such as coronary heart disease , , , hypertension , prostate cancer , , and ischemic heart disease , and it is likely that AGA is a precursor symptom of these diseases. In 2007, Merck & Co. reported US$405 million in global sales of medical products related to AGA treatment, highlighting the tremendous social and economic impact of AGA ,. Moreover, AGA has important effects on mental health due to the changes in physical appearance that are caused by hair loss.
Many studies have been devoted to the genetic and androgen-related aspects of AGA , , ,. Based on a screening of 1025 blood samples from men aged 35 to 75, Richards et al. identified a baldness susceptibility gene that increases the risk of AGA six-fold; the variance explained by this allele was reported to be 13.7% . Nyholt et al.  reasoned that the major contributing factor to AGA is heredity, which accounts for 80% of the variance. However, the genetic aspect lacks specificity because an individual carrying a risk-associated allele will not suffer from AGA until he or she reaches a certain age. This indication that AGA risk alleles are modulated by age is consistent with the world-wide increase in the prevalence of AGA with age. Consequently, earlier onset AGA is associated with more severe characteristics .
To date, three AGA susceptibility genes have been identified: the AR gene on the X chromosome and two autosomal loci, 3q26  and 20p11 , . Richards et al. observed that variants in the 20p11 locus and the AR gene are common among Europeans and that men with at least one risk allele (20p11.22 or AR) at either locus have a seven-fold greater probability of developing AGA than those without either risk allele ; carriers of at least one risk allele accounted for one-seventh of all men in the study. Hillmer et al. also showed that the 20p11 locus is associated with early-onset AGA , .
In addition, Hillmer et al. discovered that DNA short tandem repeats on chromosome 3 (namely D3S3053, D3S1556 and D3S2425) are related to AGA . Chen et al  reported that the expression level of SRY increases with the severity of baldness. Therefore, we attempted to investigate two single nucleotide polymorphisms (SNPs) within the SRY gene.
Although a strong genetic basis for AGA has been established, little is known about its non-genetic causes, such as environmental and dietary factors. This study investigated the bodily heavy metals concentrations, dietary habits and genotypes of related susceptibility genes in patients with AGA.
It has been suggested that air pollution may lead to the over-accumulation of certain heavy metals in the scalp, resulting in hair loss ,. A study conducted in Lithuania reported that bald individuals had higher concentrations of lead, copper and cadmium and lower concentrations of zinc in their hair follicles than did individuals with normal hair . It has been proposed that lead may replace zinc in heme, while cadmium substitutes for zinc in metallothionein, and the combination of these losses of zinc likely cause alopecia .
Smoking also affects the development of AGA because the genotoxic compounds in cigarettes may damage the DNA in hair follicles and subsequently cause microvascular poisoning in hair papillae . Studies have established that a family history of AGA, the age of AGA onset (age ≤40 years old) and smoking are all correlated with AGA .
Despite the discovery of genes associated with this disorder, many factors contributing to the variable levels of AGA have yet to be elucidated. To date, no study has comprehensively examined many of the potential AGA-associated factors, such as dietary and body concentrations of heavy metals. Thus, we attempted to assess the association between the physiological concentrations of heavy metals, dietary factors and susceptibility genes in men with AGA.
Materials and Methods
The research subjects were men from 35 to 65 years of age who had lived in the districts of Dadu, Longjing and Shigang in Taichung for at least 5 consecutive years. Age and relocation time were used to divide individuals who met these criteria into three age groups (35–44, 45–54 and 55–64). Subsequently, 13 villages were selected as research sites based on their registered permanent residents; each village contained 40 individuals in each of the three age groups. A total of 1560 men were examined.
Consent letters and survey forms were mailed to the subjects from the sample list, and appointments were arranged at local health centers. The signed consent letters were collected at these appointments. A total of 354 men agreed to participate in the study. A physical examination, specimen collection and photographic documentation of the scalp were performed by nurses in these public health centers.
A matched case-control study was conducted with controls recruited from the communities in the previously defined districts of Dadu, Longjing and Shigang in Taichung of central Taiwan. Eligibility was restricted to locally registered residents who had lived in the selected areas for more than five consecutive years and who were between 35 and 64 years old.
Subjects were recruited from five villages within the Longjing and Dadu townships. The reference group was recruited from the two villages of the Shihgang Township, an area located in northeast Taichung County that is free of industrial pollution. Eligible subjects were registered local residents who had lived in the selected areas for more than five consecutive years and who were between 35 and 64 years old. A total of 1440 subjects were randomly selected from each age group (35, 45, and 55 years old), gender (male and female) and village, i.e., a total of 12 subgroups. This study focused on AGA in men. A total of 354 men agreed to participate in the study, for a response rate of 49.2% (354/720); 277 of the subjects were from the exposed area, and 77 were from the reference area.
This study was approved by the institutional review board at the Tri-Service General Hospital in Taipei, Taiwan. Study consent letters and questionnaires were mailed to the subjects from the sample list, and appointments were arranged at local health centers. All the participants signed informed consent waivers prior to study enrollment. The examination, specimen collection and photographic documentation of the scalp were performed at the public health centers.
Demographic and medical data.
At the beginning of the study, we mailed a self-administered questionnaire to the participants. The questionnaire gathered information about personal characteristics and lifestyle information, such as age, education, smoking habits, alcohol consumption, betel nut usage, work hours, sleep patterns, hair treatments, disease history, and eating habits. Subjects were asked to collect a first void urine sample using a 100 ml polyethylene container on the morning of their appointment. The subjects brought their survey questionnaires and urine samples to the public health centers. The intra-class correlation of dietary intake from 65 studies was 0.9 for two food frequency questionnaires (FFQ) administered 1 month apart, with correlations of 0.4 to 0.8 for various nutrients . The urine samples were labeled with the subject identification number, date, and time. The samples were transported in a cooler and stored at 4°C during shipping. Aliquots were prepared and stored in a −80°C freezer.
The blood samples were collected by venipuncture from the forearm in either (1) trace element k2 EDTA tubes (Becton Dickinson, Rutherford, New Jersey) to evaluate heavy metal concentrations or (2) EDTA tubes (Becton Dickinson, Rutherford, New Jersey) to extract DNA. The blood collection tubes were transported in a cooler and stored at 4°C during shipping to the laboratory. The blood samples were divided into aliquots for heavy metal analyses. For the DNA extraction, the EDTA-containing whole blood was centrifuged for 15 min, and the resulting buffy coat containing the white blood cells was subjected to genomic DNA extraction. SNP genotyping was performed using the ABI TaqMan® system. The degree of AGA was evaluated based on the Hamilton-Norwood scale by specialists at the public health centers.
Detection of susceptibility genes from blood samples
The keyword “androgenic alopecia” was used to search the Online Mendelian Inheritance in Man (OMIM) database for human studies, yielding three accession numbers: 109200, 300710 and 612421. These accession numbers were used in a literature search, which identified AGA-associated genes located on chromosomes X , , , 3  and 20 ,  (Table S1 in File SI). We selected the following 11 most relevant SNPs (Table S2 and S3 in File SI; criteria listed below): rs925391 (the polymorphism with the highest odds ratio), rs10521339 (the polymorphism with the lowest P value), rs6152 (a polymorphism reported independently by Hillmer et al. and Ellis et al.) and rs6625163 (the most recently identified locus, identified on the X chromosome by Richards et al. via a genome-wide screen) located within Xq12–Xq13; rs1158928 (the highest odds ratio) located within Xq11–Xq12; rs1998076 (the polymorphism with the highest odds ratio in the study by Hillmer et al.), rs1160312 (the polymorphism with the lowest P value reported by Richards et al.), rs201571 (the polymorphism reported by Hillmer et al. as associated with AGA and reported by Richards et al. to possess linkage disequilibrium with the rs1160312 locus) and rs913063 (used to verify an r2 of 1 between this locus and rs1160312) located within 20p11.1–20p11.2. No polymorphisms in rs925391, rs10521339, rs6625163, rs1158928, rs6152 or rs2534636 were found among Asians. However, five of the variants, rs1998076, rs913063, rs1160312, rs201571 and rs11575897, were found to be polymorphic among Asians. Thus, the latter five SNPs were used for the genotyping tests.
Measurements of the concentrations of heavy metals in the blood and urine
Inductively coupled plasma mass spectrometry (ICP-MS) (Agilent 7700× series ICP-MS, Agilent Technologies, Inc., Palo Alto, CA, USA) was used to measure the concentrations of heavy metals (including vanadium, manganese, cobalt, copper, zinc, arsenic, cadmium, lead, nickel, and chromium) in the blood and urine samples.
The recovery efficiencies from urine samples were determined by spiking a known quantity of a trace element (NIST SRM ®2670a) into a urine sample and following the same experimental procedure used for the treatment of urine samples. The recoveries were as follows: V, 102%; Mn, 103%; Co, 104%; Cu, 102%; Zn, 94%; As, 104%; Cd, 104%; Pb, 104%; Ni, 105%; and Cr, 99%. The blank tests for metals were performed using the same procedure used in the recovery efficiency tests but without adding the known standard solution. The limits of detection were as follows (in µg/L): V, 0.028; Mn, 0.027; Co, 0.004; Cu, 0.025; Zn, 0.075; As, 0.027; Cd, 0.017; Pb, 0.026; Ni, 0.032; and Cr, 0.037. At these limits, the signal-to-noise ratio was 3.
The recovery efficiencies from blood samples were determined by spiking a known quantity of trace elements (Seronorm™) into a blood sample and following the same experimental procedure used for the treatment of the blood samples. The recoveries were as follows: V, 104%; Mn, 101%; Co, 97%; Cu, 95%; Zn, 103%; As, 96%; Cd, 98%; Pb, 100%; Ni, 97%; and Cr, 104%. The limits of detection were as follows (in µg/L): V, 0.051; Mn, 0.018; Co, 0.006; Cu, 0.047; Zn, 0.032; As, 0.039; Cd, 0.013; Pb, 0.012; Ni, 0.019; and Cr, 0.032. At these limits, the signal-to-noise ratio was 3.
The analyses of the blanks, including field blanks and lab blanks, revealed no significant contamination (i.e., the ICP-MS integrated area was below the detection limit). All sample preparation and measurement steps were performed in a laminar flow cabinet.
Statistical analyses were performed using SPSS 17.0. Continuous variables are expressed as the means with standard errors. Categorical variables are expressed as percentages. The Bonferroni adjustment for multiple testing was performed using SISA  to control for a family-wise error rate of 0.05, which significant level is considered as 0.05/42 = 0.00114. The p-values in the tables are reported in scientific notation if too many digits were needed for evaluation and to address the issue of multiple testing.
The gene-counting method was used to estimate the genotype frequencies of individual SNPs, which employs the standard EM algorithm for haplotype frequency estimation . Stata 8 and its associated macro programs were used to examine whether a SNP was in Hardy-Weinberg equilibrium , . Univariate logistic regression analysis was performed to analyze whether an allele of a SNP had an odds ratio indicating association with moderate to severe AGA (≧IV); subsequently, multivariable logistic regression was used to adjust the relevant confounding factors. In addition, the Hosmer-Lemeshow, explained variance (Nagelkerke R2) and accuracy tests were employed to help establish an optimal model.
Our study population included 60 men with moderate to severe AGA (≥IV). A total of 17.6% of the men aged 46–55 and 27% of men over 55 displayed moderate to severe AGA. These data indicate that the prevalence of AGA in men increased with age (Table 1). Additionally, our data indicated that for each increase of 1 year in age, the odds ratio of moderate to severe AGA was 1.08 (95% CI 1.04–1.13). The odds ratio of moderate to severe AGA for a positive family history was 8.57 (95% CI 3.35–25.77). Age and family history remained significant factors for moderate to severe AGA in our multivariate analysis.
Body concentrations of heavy metals and AGA
The geometric mean concentrations (µg/L) of ten heavy metals in the blood samples of participants with and without AGA are shown in Table 2. These concentration differences were statistically significant based on the univariate analysis.
In the multivariate analysis, after adjusting for age and family history, the odds ratios for the body concentrations of all tested heavy metals and between the two groups (with and without moderate to severe AGA) were statistically significant (P<0.00119). In addition, the correlation coefficients between the body concentrations of all ten metals were statistically significant. Therefore, the concentrations of heavy metals were subjected to further mutual adjustment, which showed that vanadium was the sole element with an odds ratio that reached 0.001 for moderate to severe AGA (P = 0.04).
The geometric mean urine concentrations (µg/L) of the heavy metals in the two groups (with and without AGA) are shown in Table 2. None of the concentration differences were statistically significant based on the univariate analysis. In the multivariate analysis, after adjustment for age and family history, none of the urine metals had an odds ratio indicating a statistically significant association with moderate to severe AGA. Likewise, after adjustment for age, family history and the concentrations of the ten metals, no other factor was found to have a statistically significant odds ratio for association with AGA.
Environmental, behaviors, dietary factors and moderate to severe AGA
The relationship between environmental factors, personal behaviors, dietary factors and moderate to severe AGA is illustrated in Table 3. In the univariate analysis, every increase of 1 hour of sleep generated an odds ratio of 0.69 for moderate to severe AGA (P<0.01). In contrast, poor sleep quality (inadequate sleep time, difficulty falling asleep or sleep interruption) and scalp abnormalities (itchy, oily or dry) had odds ratios of 1.96 (P = 0.02) and 2.51 (P<0.01), respectively, for the development of moderate to severe AGA. In addition, compared with subjects who rarely drink soy bean products, regular soy bean drinkers (at least 1–3 days per week) had an odds ratio of 0.04 (P = 0.04) for developing moderate to severe AGA.
After adjusting for age and family history, the odds ratios for sleep deficiency (fewer than 6 hours per day) and scalp abnormalities were 4.47 (P = 0.04) and 2.57 (P = 0.02), respectively. In the multivariate analysis, after adjusting for age, family history and area of residence, the odds ratios for sleep deficiency and scalp abnormalities were 4.30 (P = 0.04) and 3.02 (P = 0.01), respectively.
SNPs and AGA
The SNP analysis was conducted in 184 individuals who were randomly selected from the original pool of 354 subjects and their age-, residence area- and smoking status-matched controls (subjects exhibited moderate to severe AGA, and controls had no or only slight AGA). The matched variables sufficiently increased the statistical power with an efficient sample size.
Five loci (rs1998076, rs913063, rs1160312, rs201571 and rs11575897) were evaluated in the SNP genotyping tests. Primers for rs11575897 were generated using “Assay by Design” based on the two 500-bp fragments flanking the SNP. Unfortunately, these primers yielded poor results due to secondary structure formation or high GC content and failed to satisfy our quality control criteria; therefore, this SNP was excluded from the study. The genotype and allele frequencies of rs1998076, rs913063, rs1160312 and rs201571 in the two groups (AGA group and the control group) were in line with Hardy-Weinberg equilibrium (Table 4).
The rs1998076 locus is located at position 21828045 on chromosome 20 in an intergenic region. The wild-type allele is homozygous GG, while the mutant allele is homozygous AA. The GG genotype was considered to be the control. We found odds ratios of 0.28 (P = 0.06) for the AA genotype and 0.41 (P<0.01) for the AG (heterozygous) genotype, respectively, for developing moderate to severe AGA. In the dominant model, the GG genotype was used as the control, and the combined odds ratio of the AA and AG genotypes was 0.39 (P<0.01) for developing moderate to severe AGA. In the recessive model, the GG and GA genotypes were used as the control, and the odds ratio of the AA genotype was 0.45 (P<0.01) for developing moderate to severe AGA. Compared with the G allele, the A allele had an odds ratio of 0.52 (p<0.01) for developing moderate to severe AGA.
The rs913063 locus is located at position 21990418 of chromosome 20, upstream of the non-coding gene RP11-125P18.1. The wild-type genotype is CC, and the mutant genotype is AA. Using CC as the control group, we found that the AA and CA genotypes had odds ratios of 4.03 (P<0.01) and 0.72 (P = 0.37), respectively, for developing moderate to severe AGA. In the dominant model, the CC genotype was used as the control, and the combined odds ratio of the CA and AA genotypes was 1.17 (P = 0.63) for developing moderate to severe AGA. In the recessive model, the CC and CA genotypes were used as the controls, and the AA genotype was found to have an odds ratio of 4.68 (P<0.01) for developing moderate to severe AGA. Altogether, compared to the C allele, the A allele carried an odds ratio of 1.67 (P = 0.03) for developing moderate to severe AGA.
The rs1160312 locus is located at position 21998503 of chromosome 20, inside the non-coding gene RP11-125P18.1. The wild-type genotype is GG, while the mutant genotype is AA. Using the GG genotype as the control, we found that the AA and AG genotypes carried odds ratios of 3.93 (P<0.01) and 0.73 (P = 0.40), respectively, for developing moderate to severe AGA. In the dominant model, the GG genotype was used as the control, and the combined odds ratio of the AA and AG genotypes was 1.21 (P = 0.55) for developing moderate to severe AGA. In the recessive model, the GG and AG genotypes were used as the controls, and AA was found to harbor an odds ratio of 4.53 (P<0.01) for developing moderate to severe AGA. Hence, compared with the G allele, the A allele had an odds ratio of 1.71 (P = 0.02) for developing moderate to severe AGA.
The rs201571 locus is located at position 21961514 of chromosome 20, within an intergenic region. The wild-type genotype is CC, while the mutant genotype is TT. Using the CC genotype as the control, we found that the TT and CT genotypes harbored odds ratios of 4.24 (P<0.01) and 0.82 (P = 0.59), respectively, for developing moderate to severe AGA. In the recessive model, the CC and CT genotypes were used as the control, and TT was found to have an odds ratio of 1.87 (P<0.01) for developing moderate to severe AGA. Therefore, compared with the C allele, the T allele had an odds ratio of 1.87 (p<0.01) for developing moderate to severe AGA.
AGA and models of the related factors
In addition to comparing the genotypes, residence area and smoking status, other factors with significant odds ratios for the development of moderate to severe AGA were subjected to logistic regression analysis (Table 5). These factors included blood concentrations of vanadium, the AA genotype of rs1160312, sleep deficiency (fewer than 6 hours per day) and frequent consumption of soy bean drinks (3 days per week), which had odds ratios (represented here with their respective 95% confidence intervals) of 17.67 (5.48–57.00), 0.981 (0.972–0.991), 10.75 (3.12–37.03), 6.99 (1.29–37.87) and 0.23 (0.08–0.69), respectively, for moderate to severe AGA. The Hosmer-Lemeshow test revealed that the goodness of fit was not statistically significant (P = 0.50). The explained variance (Nagelkerke R2) was 0.64, the accuracy was 0.861 and the AUC was 0.926 (95% CI: 0.839–0.962). The interactions between the SNPs and the environmental or personal behavioral risk factors for AGA were not statistically significant in these models.
This study found that the prevalence of moderate to severe AGA in men in our study areas is 17% (60/354), higher than previously reported . The prevalence of AGA in men increased with age. Age, family history, poor sleep quality, scalp abnormalities (itchy, oily or dry), and drinking soy bean products regularly were still important factors for moderate to severe AGA. The body concentrations of all tested heavy metals and the genotypes of the rs1998076, rs913063, rs1160312 and rs201571 polymorphisms were all correlated to the development of moderate to severe AGA.
AGA and physiological concentrations of heavy metals
In 1978, vanadium was found to be an essential trace element in humans and animals . The necessity of vanadium has been confirmed by the World Health Organization . Most often found in seafood, vanadium is involved in a variety of biological processes, including hematopoiesis, the maintenance of blood pressure, growth promotion, the maintenance of cholesterol levels, and the stimulation of receptors and other enzymes that phosphorylate insulin and regulate its biological activity. In sugar metabolism, vanadium mainly facilitates the entry of glucose into cells as a hypoglycemic agent. Vanadium deficiency can lead to many characteristics, including increased cholesterol, anemia, myocardial weakness and diabetes . Furthermore, studies have shown that hypertension, which is itself closely associated with high cholesterol, and impaired glucose tolerance (or Type II diabetes) are both associated with the development of AGA ,. Therefore, the role of vanadium in humans requires further investigation. Vanadium deficiency in poultry has been shown to cause incomplete feather coverage or apparent slower feather growth .
Interestingly, there is a significant discrepancy between this study and one previously performed by Naginiene et al. , who reported that the hair, blood and urine of bald individuals contain increased concentrations of lead, copper and cadmium but decreased concentrations of zinc. Our data showed that individuals with moderate to severe AGA had decreased concentrations of lead, copper, cadmium and zinc in their blood and increased concentrations of lead, cadmium and zinc in their urine (the copper concentration was similar to that of the control). One possible cause for these differences is the different racial backgrounds of the subjects in the two studies. Additionally, the sampling methods were different; Naginiene et al.  used children as the control group and recruited adult men and women for the bald group, while this study used adult men with and without AGA in the bald and control groups. Due to these methodological differences, it is difficult to directly compare the two studies. Lastly, we found a significant correlation between blood vanadium concentration and protection against moderate to severe AGA that has not been reported previously; Naginiene et al. did not examine vanadium concentrations.
Dietary factors, environmental factors, personal behaviors and AGA
Although Su and Chen found an association between smoking and AGA , our results are consistent with other previous studies , ,  that failed to reveal a significant correlation between cigarette smoking and AGA. This inconsistency may be caused by differences in the sampling methods and subject demographics. The present study did not exclude smoking as a risk factor, and the measurements included a detailed record of smoking frequency per day and age of smoking commencement. Nevertheless, to avoid complications based on smoking, we balanced the ratios of smokers and non-smokers in both the control and AGA groups in this study.
Our data on environmental factors, personal behaviors and dietary intake indicated that sleep deprivation, oily scalp and the frequent consumption of soy bean drinks are correlated with moderate to severe AGA. A previous study revealed that insufficient sleep is a risk factor for sebaceous gland diseases, including seborrheic dermatitis, acne, AGA and rosacea . Oily scalp as a risk factor for moderate to severe AGA can be explained as follows: Oily skin results from robust secretion from the sebaceous glands, which are controlled by androgens. This increased secretion in turn causes changes to the hair cycle and subsequently aggravates the manifestation of AGA , , . Our data suggest that sleeping for fewer than 6 hours each day increases the risk of AGA. However, because this was a case-control study, we were unable to deduce a cause and effect relationship or identify the specific mechanisms underlying this correlation. Further investigation is needed to address these issues.
We found that frequent soy bean drink consumption is protective against moderate to severe AGA. Soy bean drinks are rich in isoflavones, and the isoflavone metabolite equol displays high levels of antioxidant , ,  and estrogen-like activities , , , . There are many reports that isoflavone , , antioxidants  and estrogens , , , , , , , , , , , ,  are protective against alopecia. The previous reports found that orally administered soymetide-4 (MITL), a soy-derived immunostimulating peptide from soy bean beta-conglycinin alpha' subunit, suppressed the alopecia induced by the anti-cancer drug etoposide , , , , . Hypothetically, soy oil compounds may act to modify alopecia susceptibility by modulating estrogen-dependent mechanisms or inflammatory activity. Further studies are needed to explore the connections between isoflavone, equol and AGA.
Genes and AGA
Regarding the relationship between chromosome 20 and AGA, this report is consistent with other relevant studies , . Hillmer et al. showed that the risk-associated SNP rs2180439[T] is in linkage disequilibrium with rs1998076[G]. This is consistent with our findings that rs1998076[A] is a protective SNP, while rs202571[T] is a risk-associated SNP. The study by Richards et al. revealed that the r2 between rs1160312 and rs913063 is 1, indicating complete linkage disequilibrium. In other words, the information concerning rs913063 can be obtained by examining rs1160312. Although this linkage was reported previously, we performed genotyping to confirm this hypothesis. We found that both rs1160312[A] and rs913063[A] are risk-associated SNPs, corroborating the previous studies by Hillmer et al. and Richards et al. , 
Both rs1998076 and rs201571 are located in intergenic regions, while both rs1160312 and rs913063 are within the non-coding gene RP11-125P18.1. No transcription has been reported in any of these regions. However, previous studies have suggested that these SNPs might interact with PAX1 , , which is highly expressed in the scalp. Although PAX1 is outside of the LD region, its expression pattern indicates that it affects AGA, possibly due to altering the expression of other loci in this LD region.
It has been reported that the AR gene on the X chromosome is strongly associated with male AGA , , . However, there are no polymorphisms in this gene among Asian populations and also in rs6152 of the present study. In addition, there are two SNPs within SRY . The lack of AGA-associated SNP polymorphisms among Asian populations may underlie the lower frequency of AGA in this ethnic group. Sehgal et al. reported that a gradual shift in the type of AGA from the earlier types (II and III) to more severe types (VI) occurs significantly with increasing age . Pathomvanich et al. reported that the prevalence of AGA Norwood III–VII was 38.52% and significantly increased with age in a study of 1124 men . The lack of polymorphisms in AGA-associated SNPs among Asians might suggest a role for DNA methylation which results in changes in gene expression. This could explain why age is correlated with the prevalence of AGA , , , . Further studies are needed to explore the connection between the lack of polymorphisms, DNA methylation and AGA among Asians. Further studies are also needed to explore the connection between the lacks of polymorphisms, DNA methylation and AGA.
Alopecia is a complex skin disorder observed in individuals whose conscious experience of distress is often absent and may be precipitated by environmental events, not simply the influence of inherited factors . The contribution of genetic factors to alopecia is strong, but environmental factors, such as environmental stress, still play an important role, and the genetics of alopecia are consistent with a polygenic additive mode of inheritance . This study examined whether any evidence for an environmental component to the risk for AGA exists; our results indicated that the genes and environmental factors studied account for a proportion of the risk of AGA. One limitation of this study is that AGA may be affected by many genes that have not yet been identified; twin studies have shown that the AGA condition is heritable, and a family history of AGA has been included in our analysis for this reason. However, there are still many findings , , , , , , , , , , , , , , , , , , ,  that suggest that heritable and epigenetic ,  differences also play a role in alopecia; this may replace the classical discussion of the roles of genetic and environmental factors in alopecia.
The environmental factors examined have been adequately reported, and the basis and the mechanistic explanations provided were congruent with previous findings , , , , , , , , , . The cross-sectional nature of the measurements of blood metals and personal behaviors (for example, sleep patterns , , ,  and oily scalp , , , , , , ) limit our ability to identify causal relationships; the time of onset of moderate to severe hair loss will need more investigation in future work.
Finally, we made an effort to adjust our independent variables in multiple logistic regression analyses. This adjustment identified several significant factors. Additional research will be required to investigate the effect of epigenetic changes on alopecia. Our epidemiological analyses are based on a sample that is representative of the Taiwanese population. Cases and controls were defined by AGA level, which was evaluated based on the Hamilton-Norwood scale by dermatological specialists at the Public Health Centers. The use of the Hamilton-Norwood scale ensured that baldness was accurately assessed according to international criteria , , , , , , , .
The optimal model
It has been reported that genes only contribute to 13.7% of the explained variance in AGA . Thus, in this study, family history, the concentration of vanadium in blood, the AA genotype of rs1160312 and the regular consumption of soy bean drinks (3 days per week) were examined in a logistic regression analysis. The results showed that these factors contributed to 59% of the explained variance of AGA. Furthermore, using ROC curve analysis, we found that rs1160312 is unable to completely account for all AGA-associated genetic factors and thus cannot replace the influence of family history on AGA risk.
Supporting tables. Table S1. SNP loci reported to be associated with AGA. Table S2. AGA-associated SNP loci investigated in this study. Table S3. Genotype details.
The interpretation and conclusions contained herein do not represent the official opinions of the Bureau of Health of Taichung County or the National Health Research Institutes.
Conceived and designed the experiments: CMC CHL NFC. Performed the experiments: CTC MHL. Analyzed the data: CMC CHL CTC MHL HCY WCC SLS YCC. Contributed reagents/materials/analysis tools: CMC CHL SLW CWC KHC WMW SHL. Wrote the paper: CMC CHL CWC HCY CTC MHL KHC.
- 1. Giltay EJ, Toorians AW, Sarabdjitsingh AR, de Vries NA, Gooren LJ (2004) Established risk factors for coronary heart disease are unrelated to androgen-induced baldness in female-to-male transsexuals. J Endocrinol 180: 107–112.
- 2. Lotufo PA, Chae CU, Ajani UA, Hennekens CH, Manson JE (2000) Male pattern baldness and coronary heart disease: the Physicians' Health Study. Arch Intern Med 160: 165–171.
- 3. Matilainen VA, Makinen PK, Keinanen-Kiukaanniemi SM (2001) Early onset of androgenetic alopecia associated with early severe coronary heart disease: a population-based, case-control study. J Cardiovasc Risk 8: 147–151.
- 4. Ahouansou S, Le Toumelin P, Crickx B, Descamps V (2007) Association of androgenetic alopecia and hypertension. Eur J Dermatol 17: 220–222.
- 5. Demark-Wahnefried W, Schildkraut JM (2001) Correspondence re: E. Hawk, et al., Male pattern baldness and clinical prostate cancer in the epidemiologic follow-up of the First National Health and Nutrition Examination Survey. Cancer Epidemiol.Biomark. Prev., 9: 523–527, 2000. Cancer Epidemiol Biomarkers Prev 10: 415–416.
- 6. Hawk E, Breslow RA, Graubard BI (2000) Male pattern baldness and clinical prostate cancer in the epidemiologic follow-up of the first National Health and Nutrition Examination Survey. Cancer Epidemiol Biomarkers Prev 9: 523–527.
- 7. Ford ES, Freedman DS, Byers T (1996) Baldness and ischemic heart disease in a national sample of men. Am J Epidemiol 143: 651–657.
- 8. Richards JB, Yuan X, Geller F, Waterworth D, Bataille V, et al. (2008) Male-pattern baldness susceptibility locus at 20p11. Nat Genet 40: 1282–1284.
- 9. Reynolds FD, Darden WR (1974) Constructing Life Style and Psychographics. Chicago: American Marketing Association: 74–87.
- 10. Hillmer AM, Brockschmidt FF, Hanneken S, Eigelshoven S, Steffens M, et al. (2008) Susceptibility variants for male-pattern baldness on chromosome 20p11. Nat Genet 40: 1279–1281.
- 11. Hillmer AM, Hanneken S, Ritzmann S, Becker T, Freudenberg J, et al. (2005) Genetic variation in the human androgen receptor gene is the major determinant of common early-onset androgenetic alopecia. Am J Hum Genet 77: 140–148.
- 12. Hamilton JB (1959) A male pattern baldness in wattled starlings resembling the condition in man. Ann N Y Acad Sci 83: 429–447.
- 13. Nyholt DR, Gillespie NA, Heath AC, Martin NG (2003) Genetic basis of male pattern baldness. J Invest Dermatol 121: 1561–1564.
- 14. Su LH, Chen TH (2007) Association of androgenetic alopecia with smoking and its prevalence among Asian men: a community-based survey. Arch Dermatol 143: 1401–1406.
- 15. Hillmer AM, Flaquer A, Hanneken S, Eigelshoven S, Kortum AK, et al. (2008) Genome-wide scan and fine-mapping linkage study of androgenetic alopecia reveals a locus on chromosome 3q26. Am J Hum Genet 82: 737–743.
- 16. Chen W, Yang CC, Tsai RY, Liao CY, Yen YT, et al. (2007) Expression of sex-determining genes in the scalp of men with androgenetic alopecia. Dermatology 214: 199–204.
- 17. Gryboś R, Zagrodzki P, Krośniak ŁŁ, M., Szklarzewicz J, Gołaś J, et al. (2005) Level and Relationship of Elements in Scalp Hair of Males: Effect of Air Pollution and Smoking Habits. Polish journal of Environmmental Studies 14: 35–40.
- 18. Naginiene R, Abdrachmanovas O, Kregzdyte R, Rsyselis S (2002) Investigation of heavy metals in people with alopecia. Trace Elements Electrolytes 19: 87–90.
- 19. Skalnyj A (1999) Microelemental disease(diagnostics and treatment). Secientific World, Moscow 96.
- 20. Trueb RM (2002) Molecular mechanisms of androgenetic alopecia. Exp Gerontol 37: 981–990.
- 21. Willett WC, editor (1998) Implications of total energy intake for epidemiologic analyses. Nutritional epidemiology. 2nd ed. New York: Oxford University Press; 1998. p. 273–301. 2nd ed. New York: Oxford University Press. 273–301 p.
- 22. Ellis JA, Scurrah KJ, Cobb JE, Zaloumis SG, Duncan AE, et al. (2007) Baldness and the androgen receptor: the AR polyglycine repeat polymorphism does not confer susceptibility to androgenetic alopecia. Hum Genet 121: 451–457.
- 23. Ellis JA, Stebbing M, Harrap SB (2001) Polymorphism of the androgen receptor gene is associated with male pattern baldness. J Invest Dermatol 116: 452–455.
- 24. SISA: Simple Interactive Statistical Analysis. Available: http://www.quantitativeskills.com/sisa/calculations/bonfer.htm Accessed: 3 Jul 2013
- 25. Zhao JH (2004) 2LD, GENECOUNTING and HAP: Computer programs for linkage disequilibrium analysis. Bioinformatics 20: 1325–1326.
- 26. Biostatistical Resources Stata Programs. Available: http://www.biostat-resources.com/stata/index.htm Accessed: 4 Nov 2013
- 27. Cantley LC Jr, Cantley LG, Josephson L (1978) A characterization of vanadate interactions with the (Na,K)-ATPase. Mechanistic and regulatory implications. J Biol Chem 253: 7361–7368.
- 28. Xia M (2003) The Biochemical and Physiological Action of Trace Elements. Studies of Trace Elements and Health 20: 41–44.
- 29. Starka L, Duskova M, Cermakova I, Vrbikova J, Hill M (2005) Premature androgenic alopecia and insulin resistance. Male equivalent of polycystic ovary syndrome? Endocr Regul 39: 127–131.
- 30. Zhou JH (2006) Trace elements- Vanadiumin in biological research. Feed Industry 27: 59–62.
- 31. Gonzalez-Gonzalez JG, Mancillas-Adame LG, Fernandez-Reyes M, Gomez-Flores M, Lavalle-Gonzalez FJ, et al. (2009) Androgenetic alopecia and insulin resistance in young men. Clin Endocrinol (Oxf) 71: 494–499.
- 32. Severi G, Sinclair R, Hopper JL, English DR, McCredie MR, et al. (2003) Androgenetic alopecia in men aged 40–69 years: prevalence and risk factors. Br J Dermatol 149: 1207–1213.
- 33. Shahar E, Heiss G, Rosamond WD, Szklo M (2008) Baldness and myocardial infarction in men: the atherosclerosis risk in communities study. Am J Epidemiol 167: 676–683.
- 34. Zhang H, Liao W, Chao W, Chen Q, Zeng H, et al. (2008) Risk factors for sebaceous gland diseases and their relationship to gastrointestinal dysfunction in Han adolescents. J Dermatol 35: 555–561.
- 35. Essah PA, Wickham EP 3rd, Nunley JR, Nestler JE (2006) Dermatology of androgen-related disorders. Clin Dermatol 24: 289–298.
- 36. Rosenfield RL (2005) Hirsutism and the variable response of the pilosebaceous unit to androgen. J Investig Dermatol Symp Proc 10: 205–208.
- 37. Zouboulis CC, Degitz K (2004) Androgen action on human skin – from basic research to clinical significance. Exp Dermatol 13 Suppl 4: 5–10.
- 38. Kang HJ, Ansbacher R, Hammoud MM (2002) Use of alternative and complementary medicine in menopause. Int J Gynaecol Obstet 79: 195–207.
- 39. Turner R, Baron T, Wolffram S, Minihane AM, Cassidy A, et al. (2004) Effect of circulating forms of soy isoflavones on the oxidation of low density lipoprotein. Free Radic Res 38: 209–216.
- 40. Wei H, Cai Q, Rahn RO (1996) Inhibition of UV light- and Fenton reaction-induced oxidative DNA damage by the soybean isoflavone genistein. Carcinogenesis 17: 73–77.
- 41. Knight DC, Eden JA (1996) A review of the clinical effects of phytoestrogens. Obstet Gynecol 87: 897–904.
- 42. Messina M (2000) Soyfoods and soybean phyto-oestrogens (isoflavones) as possible alternatives to hormone replacement therapy (HRT). Eur J Cancer 36 Suppl 4: S71–72.
- 43. Cabeza M, Bratoeff E, Heuze I, Ramirez E, Sanchez M, et al. (2003) Effect of beta-sitosterol as inhibitor of 5 alpha-reductase in hamster prostate. Proc West Pharmacol Soc 46: 153–155.
- 44. Harada N, Okajima K (2009) Effects of capsaicin and isoflavone on blood pressure and serum levels of insulin-like growth factor-I in normotensive and hypertensive volunteers with alopecia. Biosci Biotechnol Biochem 73: 1456–1459.
- 45. Harada N, Okajima K, Arai M, Kurihara H, Nakagata N (2007) Administration of capsaicin and isoflavone promotes hair growth by increasing insulin-like growth factor-I production in mice and in humans with alopecia. Growth Horm IGF Res 17: 408–415.
- 46. Koca R, Armutcu F, Altinyazar C, Gurel A (2005) Evaluation of lipid peroxidation, oxidant/antioxidant status, and serum nitric oxide levels in alopecia areata. Med Sci Monit 11: CR296–299.
- 47. Adenuga P, Summers P, Bergfeld W (2012) Hair regrowth in a male patient with extensive androgenetic alopecia on estrogen therapy. J Am Acad Dermatol 67: e121–123.
- 48. Belezos NK (1965) Local estrogen and ultraviolet irradiation in the treatment of total alopecia (areata). Dermatologica 131: 304–308.
- 49. Endres HJ (1980) [Therapy of alopecia of different etiologies using an estrogen- and corticoid-containing topical preparation]. Z Hautkr 55: 14–18.
- 50. Georgala S, Katoulis AC, Georgala C, Moussatou V, Bozi E, et al. (2004) Topical estrogen therapy for androgenetic alopecia in menopausal females. Dermatology 208: 178–179.
- 51. Gusarova AS (1968) [Dynamics of estrogen excretion in patients with seborrheal alopecia under the effect of external application of sex hormones]. Vestn Dermatol Venerol 42: 40–44.
- 52. McElwee KJ, Niiyama S, Freyschmidt-Paul P, Wenzel E, Kissling S, et al. (2003) Dietary soy oil content and soy-derived phytoestrogen genistein increase resistance to alopecia areata onset in C3H/HeJ mice. Exp Dermatol 12: 30–36.
- 53. Nomiyama T, Arakawa A, Hattori S, Konishi K, Takenaka H, et al. (2013) Intractable diffuse alopecia caused by multifactorial side-effects in treatment of acute lymphocytic leukemia: connection to iatrogenic failure of estrogen secretion. Pediatr Dermatol 30: 105–108.
- 54. Ohnemus U, Unalan M, Handjiski B, Paus R (2004) Topical estrogen accelerates hair regrowth in mice after chemotherapy-induced alopecia by favoring the dystrophic catagen response pathway to damage. J Invest Dermatol 122: 7–13.
- 55. Orfanos CE, Wustner H (1975) [Penetration and side effects of local estrogen application in alopecia androgenetica]. Hautarzt 26: 367–369.
- 56. Rubisz-Brzezinska J, Zych F (1971) [Studies on the excretion of estrogen, gonadotropins, 17-ketosteroids and 17-hydroxycorticosteroids in women with diffuse alopecia]. Przegl Dermatol 58: 413–420.
- 57. Wallace ML, Smoller BR (1998) Estrogen and progesterone receptors in androgenic alopecia versus alopecia areata. Am J Dermatopathol 20: 160–163.
- 58. Wustner H, Orfanos CE (1974) [Alopecia androgenetica and its local treatment with estrogen- and corticosteroid externa]. Z Hautkr 49: 879–888.
- 59. Spitzer RR, Phillips PH (1946) Alopecia in rats fed certain soybean oil meal rations. Proc Soc Exp Biol Med 63: 10–13.
- 60. Tsuruki T, Takahata K, Yoshikawa M (2005) Anti-alopecia mechanisms of soymetide-4, an immunostimulating peptide derived from soy beta-conglycinin. Peptides 26: 707–711.
- 61. Tsuruki T, Takahata K, Yoshikawa M (2004) A soy-derived immunostimulating peptide inhibits etoposide-induced alopecia in neonatal rats. J Invest Dermatol 122: 848–850.
- 62. McElwee KJ, Hoffmann R, Freyschmidt-Paul P, Wenzel E, Kissling S, et al. (2002) Resistance to alopecia areata in C3H/HeJ mice is associated with increased expression of regulatory cytokines and a failure to recruit CD4+ and CD8+ cells. J Invest Dermatol 119: 1426–1433.
- 63. Sehgal VN, Kak R, Aggarwal A, Srivastava G, Rajput P (2007) Male pattern androgenetic alopecia in an Indian context: a perspective study. J Eur Acad Dermatol Venereol 21: 473–479.
- 64. Pathomvanich D, Pongratananukul S, Thienthaworn P, Manoshai S (2002) A random study of Asian male androgenetic alopecia in Bangkok, Thailand. Dermatol Surg 28: 804–807.
- 65. Hirsso P, Rajala U, Hiltunen L, Jokelainen J, Keinanen-Kiukaanniemi S, et al. (2007) Obesity and low-grade inflammation among young Finnish men with early-onset alopecia. Dermatology 214: 125–129.
- 66. Bernal Gonzalez C, Fernandez Salas C, Martinez S, Ezquieta Zubicaray B (2006) [Premature androgenetic alopecia in adult male with nonclassic 21-OH deficiency. A novel nonsense CYP21A2 mutation (Y336X) in 2 affected siblings]. Med Clin (Barc) 127: 617–621.
- 67. Diaz-Atienza F, Gurpegui M (2011) Environmental stress but not subjective distress in children or adolescents with alopecia areata. J Psychosom Res 71: 102–107.
- 68. Yang S, Yang J, Liu JB, Wang HY, Yang Q, et al. (2004) The genetic epidemiology of alopecia areata in China. Br J Dermatol 151: 16–23.
- 69. Koyama T, Kobayashi K, Wakisaka N, Hirayama N, Konishi S, et al. (2013) Eleven pairs of Japanese male twins suggest the role of epigenetic differences in androgenetic alopecia. Eur J Dermatol 23: 113–115.
- 70. Gatherwright J, Liu MT, Gliniak C, Totonchi A, Guyuron B (2012) The contribution of endogenous and exogenous factors to female alopecia: a study of identical twins. Plast Reconstr Surg 130: 1219–1226.
- 71. Kuldeep C, Singhal H, Khare AK, Mittal A, Gupta LK, et al. (2011) Randomized comparison of topical betamethasone valerate foam, intralesional triamcinolone acetonide and tacrolimus ointment in management of localized alopecia areata. Int J Trichology 3: 20–24.
- 72. Rodriguez TA, Fernandes KE, Dresser KL, Duvic M (2010) Concordance rate of alopecia areata in identical twins supports both genetic and environmental factors. J Am Acad Dermatol 62: 525–527.
- 73. Rodriguez TA, Duvic M (2008) Onset of alopecia areata after Epstein-Barr virus infectious mononucleosis. J Am Acad Dermatol 59: 137–139.
- 74. Ellis CN, Brown MF, Voorhees JJ (2002) Sulfasalazine for alopecia areata. J Am Acad Dermatol 46: 541–544.
- 75. Jackow C, Puffer N, Hordinsky M, Nelson J, Tarrand J, et al. (1998) Alopecia areata and cytomegalovirus infection in twins: genes versus environment? J Am Acad Dermatol 38: 418–425.
- 76. Dogra D, Sood A, Khaitan BK (1996) Alopecia areata in identical twins. Indian J Dermatol Venereol Leprol 62: 199.
- 77. Scerri L, Pace JL (1992) Identical twins with identical alopecia areata. J Am Acad Dermatol 27: 766–767.
- 78. Roenigk HH Jr, Kuruvilla S (1987) Topical minoxidil for male pattern alopecia in two sets of twins. Cutis 39: 329.
- 79. Sequeiros J, Sack GH Jr (1985) Linear skin atrophy, scarring alopecia, anonychia, and tongue lesion: a “new” syndrome? Am J Med Genet 21: 669–680.
- 80. Mitchell AJ, Krull EA (1984) Alopecia areata: pathogenesis and treatment. J Am Acad Dermatol 11: 763–775.
- 81. Cole GW, Herzlinger D (1984) Alopecia universalis in identical twins. Int J Dermatol 23: 283.
- 82. Gol'dshtein LM, Chipizhenko VA (1978) [Familial alopecia areata]. Vestn Dermatol Venerol 36–38.
- 83. Dmitrienko LP, Shakhnes IE (1977) [Familial alopecia in twins]. Vestn Dermatol Venerol 59–61.
- 84. Mamelok AE, Weidman AI, Zion LS (1956) Alopecia areata occurring simultaneously in identical twins. AMA Arch Derm 74: 424–426.
- 85. Gedda L, Testa I, Benigni A (1954) [Concordant congenital alopecia, achromotrichia and transverse palmar line in dizygotic triplets (two monozygotic males and one female)]. Acta Genet Med Gemellol (Roma) 3: 117–132.
- 86. Fischer HR (1953) [Alopecia areata in uniovular twins]. Z Haut Geschlechtskr 15: 178–179.
- 87. Hendren OS (1949) Identical alopecia areata in identical twins. Arch Derm Syphilol 60: 793–795.
- 88. Omens DV, Omens HD (1946) Alopecia areata in twins. Arch Derm Syphilol 53: 193.
- 89. Zhao M, Liang G, Wu X, Wang S, Zhang P, et al. (2012) Abnormal epigenetic modifications in peripheral blood mononuclear cells from patients with alopecia areata. Br J Dermatol 166: 226–273.
- 90. Chen CC, Chuong CM (2012) Multi-layered environmental regulation on the homeostasis of stem cells: the saga of hair growth and alopecia. J Dermatol Sci 66: 3–11.
- 91. Kyei A, Bergfeld WF, Piliang M, Summers P (2011) Medical and environmental risk factors for the development of central centrifugal cicatricial alopecia: a population study. Arch Dermatol 147: 909–914.
- 92. Bechard A, Meagher R, Mason G (2011) Environmental enrichment reduces the likelihood of alopecia in adult C57BL/6J mice. J Am Assoc Lab Anim Sci 50: 171–174.
- 93. Fleck M (1951) [Alopecia in children due to adverse environmental factors]. Dermatol Wochenschr 123: 98–99.
- 94. Alfani S, Antinone V, Mozzetta A, Di Pietro C, Mazzanti C, et al. (2012) Psychological status of patients with alopecia areata. Acta Derm Venereol 92: 304–306.
- 95. Hardcastle NJ, Tunbridge AJ, Shum KW, Dockrell DH, Green ST (2005) Alopecia in association with severe seborrhoeic dermatitis following combination antiretroviral therapy for acute retroviral syndrome. J Eur Acad Dermatol Venereol 19: 631–633.
- 96. Hay IC, Jamieson M, Ormerod AD (1998) Randomized trial of aromatherapy. Successful treatment for alopecia areata. Arch Dermatol 134: 1349–1352.
- 97. Gilhar A, Pillar T, Assay B, David M (1992) Failure of passive transfer of serum from patients with alopecia areata and alopecia universalis to inhibit hair growth in transplants of human scalp skin grafted on to nude mice. Br J Dermatol 126: 166–171.
- 98. Gilhar A, Pillar T, Etzioni A (1990) Topical cyclosporine in male pattern alopecia. J Am Acad Dermatol 22: 251–253.
- 99. Kim GW, Park JM, Chin HW, Ko HC, Kim MB, et al. (2012) Comparative analysis of the use of complementary and alternative medicine by Korean patients with androgenetic alopecia, atopic dermatitis and psoriasis. J Eur Acad Dermatol Venereol
- 100. George AO (1990) Androgenetic alopecia–is contact dermatitis an accelerating factor? Contact Dermatitis 22: 112.
- 101. Mumcuoglu C, Ekmekci TR, Ucak S (2011) The investigation of insulin resistance and metabolic syndrome in male patients with early-onset androgenetic alopecia. Eur J Dermatol 21: 79–82.
- 102. Acibucu F, Kayatas M, Candan F (2010) The association of insulin resistance and metabolic syndrome in early androgenetic alopecia. Singapore Med J 51: 931–936.
- 103. Cremers RG, Aben KK, Vermeulen SH, den Heijer M, van Oort IM, et al. (2010) Androgenic alopecia is not useful as an indicator of men at high risk of prostate cancer. Eur J Cancer 46: 3294–3299.
- 104. Krupa Shankar D, Chakravarthi M, Shilpakar R (2009) Male androgenetic alopecia: population-based study in 1,005 subjects. Int J Trichology 1: 131–133.
- 105. Dogramaci AC, Balci DD, Balci A, Karazincir S, Savas N, et al. (2009) Is androgenetic alopecia a risk for atherosclerosis? J Eur Acad Dermatol Venereol 23: 673–677.
- 106. Camacho FM, Garcia-Hernandez MJ, Fernandez-Crehuet JL (2008) Value of hormonal levels in patients with male androgenetic alopecia treated with finasteride: better response in patients under 26 years old. Br J Dermatol 158: 1121–1124.
- 107. Kim BJ, Kim JY, Eun HC, Kwon OS, Kim MN, et al. (2006) Androgenetic alopecia in adolescents: a report of 43 cases. J Dermatol 33: 696–699.