Background
Hair loss affects up to 50% of both men and women throughout their lives, causing anxiety and disability that can have a significant effect on the patient’s quality of life. [
1,
2] The condition has been linked to an autoimmune disorder of the hair follicle, genetic background, hormones, medication, and psychological stress, which can alter the hair follicle cycle [
3,
4] Many studies have attempted to elucidate the pathogenesis of hair loss. However, the complex molecular interactions between the cells of the hair follicle have not been fully understood, and the exact cause of alopecia is still unknown.
Finasteride and minoxidil have been approved by the Food and Drug Administration of the United States (FDA, USA) to promote hair growth. However, the effectiveness of these drugs varies greatly among individuals, and they have unwanted side effects. Relatedly, many alopecia patients are concerned about the side effects associated with conventional therapies, and complementary and alternative medicine (CAM) has thus been suggested as a new treatment for alopecia. [
5] In particular, traditional Chinese medicine (TCM) is an important part of healthcare in East Asia, and it is commonly used to treat alopecia patients. [
5,
6]
Herbal medicine prescriptions consist of various herbal preparations. Thus, using the scientific method, researchers must investigate frequently used herbal combinations and categorize them. However, in the TCM literature, few narrative reviews have focused on herbs for alopecia treatment, [
6,
7] and the Chinese, Korean, and European research databases contain no studies that have classified herbs for alopecia treatment using statistical methods such as data mining.
Previous studies based on TCM pattern identification have shown that deficiency of liver and kidney (肝腎不足), deficiency of
qi and blood (氣血兩虛),
qi stagnation and blood stasis (氣滯血瘀), and blood-heat (血熱) are the main patterns linked to alopecia. [
6,
7] However, because a diverse range of herbal ingredients are used in TCM and because interactions between herbal medicine and the human body are complex, the mechanism underlying these TCM patterns is still unknown. [
8] Recently, statistical methods such as data mining have been applied to TCM research. However, to the best of our knowledge, no studies have used network analysis methods to assess herbal medicine used in hair loss treatment.
Therefore, this study aimed to identify—using association rule mining (ARM)—which herbal combinations are used frequently in hair loss treatment and to analyze the modular characteristics of these treatments using network analysis.
Discussion
We systematically searched literature for alopecia treatment formulas. The following herbs were frequently used in internal medicine: Polygonum multiflorum Thunb., Angelica sinensis (Oliv.) Dlels, Rehmannia glutinosa Libosch. (Prepared), and Ligusticum chuanxiong Hort. Conversely, Platycladus orientalis (L.) Franco, Angelica dahurica (Fisch. ex Hoffm.) Benth. et Hook.f., Vitex trifolia L. var. simplicifolia Cham., and Ligusticum chuanxiong Hort. were frequently used in external applications. Internal medicine and external application differed in terms of pharmacological efficacy and meridian tropism. Using the ARM method, the most frequently used two-herb combinations were (1) Polygonum multiflorum Thunb. and Angelica sinensis (Oliv.) Dlels, and (2) Ligusticum chuanxiong Hort. and Rehmannia glutinosa Libosch. (Prepared). Ligusticum chuanxiong Hort., Polygonum multiflorum Thunb., and Angelica sinensis (Oliv.) Dlels was the most frequently used three-herb combination. Using the network analysis method, we classified the herbs into three modules. The meridian entry (歸經) of many herbs in module 1 was “Liver,” whereas that in module 3 was “Stomach”.
The meridian tropism theory is important in traditional East Asian medicine—both pharmacologically and in clinical practice. [
74] According to meridian tropism theory, medicinal herbs have a certain
qi and flavor (氣味), and they exhibit curative effects on selected meridians. Therefore, depending on whether it belongs to the viscera or bowel group (臟腑), each medicinal herb is mainly used in a specific region. In other words, meridian tropism is a theory of the orientation of drug action. [
75] Several experimental studies have presented evidence for meridian tropism theory. [
74‐
76]
In the present study, the internal medicine group—
Polygonum multiflorum Thunb.,
Angelica sinensis (Oliv.) Dlels,
Rehmannia glutinosa Libosch. (Prepared),
Ligusticum chuanxiong Hort., and
Poria cocos (Schw.) Wolf—appeared in order. When we analyzed the top 10 medicinal herbs in the internal medicine group, every herb except for
Ligusticum chuanxiong Hort. belonged to the viscera meridian (臟), not the bowel meridian (腑). [
12] In the external application group—
Platycladus orientalis (L.) Franco,
Angelica dahurica (Fisch. ex Hoffm.) Benth. et Hook.f.,
Vitex trifolia L. var.
simplicifolia Cham.,
Ligusticum chuanxiong Hort., and
Aconitum carmichaelii Debx.—appeared in order. Six herbs in external application group belong to the bowel meridian. [
12] These differences in meridian tropism between internal and external medicine may be associated with the drug absorption pathways or medicinal guide herb (引經藥). [
77] The category of each medicinal herb also differed. In the internal medicine group, six herbs belonged to the tonifying and replenishing medicinal category. However, in the external application group, three herbs were exterior-releasing medicinals, and three were blood-activating and stasis-dispelling medicinals (Table
1).
We identified frequently used two-herb and three-herb set combinations (Tables
2 &
3).
Polygonum multiflorum Thunb.,
Angelica sinensis (Oliv.) Dlels,
Ligusticum chuanxiong Hort. and
Rehmannia glutinosa Libosch. (Prepared) are the main herbs used in alopecia treatment. The six two-herb combinations of the four main herbs were the top six combinations of two-herb sets. These four main herbs were also important in the three-herb sets. However, the three-herb combination of
Polygonum multiflorum Thunb.,
Ligusticum chuanxiong Hort., and
Rehmannia glutinosa Libosch. (Prepared) occupied the relatively low 8th place.
Paeonia lactiflora Pall. did not appear in the two-herb sets, but it was frequently observed in the three-herbs sets, indicating that this herb is used as an adjunct in alopecia treatment.
Interestingly, the lift value of the
Ligustrum lucidum Ait. and
Eclipta prostrata L. combination was higher than the frequency and support values, and these herbs often appeared together with
Polygonum multiflorum Thunb. or
Angelica sinensis (Oliv.) Dlels. That said, all the medicinal herbs appeared frequently. In contrast, in the case of
Ligustrum lucidum Ait. and
Eclipta prostrata L., the lift value was higher than the frequency of each medicinal herb, indicating that
Ligustrum lucidum Ait. and
Eclipta prostrata L. are usually prescribed together. The herbal formula name of the
Ligustrum lucidum Ait. and
Eclipta prostrata L. combination is Yijihwan (二至丸). It has antioxidant activity and has been prescribed for hair loss in clinical practice. [
78] Among the three-herb sets, the lift value of the
Ligustrum lucidum Ait.,
Eclipta prostrata L., and
Polygonum multiflorum Thunb. combination was also relatively high.
When we use the ARM method, the number of herbs that comprise each herb set should be determined in advance. For this reason we only identified frequently used two-herb and three-herb sets (Tables
2 &
3), and we used network analysis to assess the relationships of all medicinal herbs used to treat hair loss, regardless of the number of herbs in the set (Figs.
2 &
3). We reviewed previous literature regarding pattern identification in alopecia. [
6,
7,
79‐
81] Blood heat engendering wind (血熱生風), blood stasis due to
qi stagnation (氣滯血瘀), dual deficiency of
qi and blood (氣血兩虛), liver-kidney depletion (肝腎不足), and spleen-stomach dampness-heat (脾胃濕熱) were the major pattern identifications in alopecia. Pathology was classified in terms of the viscera and bowels (臟腑) theory or the
qi and blood (氣血) theory.
According to our network analysis, Module 1 herbs affect the “Liver” meridian more and seem to tonify
qi and blood. [
82] Module 3 herbs belong to the “Stomach” meridian more and seem to help digestion and absorption. Module 2 herbs seem to act on body surfaces, and they tend to be used externally, although further research is needed in this regard. These modules were similar to the traditional pattern identification framework derived from alopecia literature reviews. [
6,
7,
79‐
81] In the present study, we reconfirmed the that Module 1 comprises tonifying “Liver” and “Kidney” strategies, and that Module 3 belongs more to the “Stomach” meridian, indicating that treatment of digestion and absorption are important in alopecia treatment.
Among the top 10 herbs in the internal medication group, none belonged to Module 2. All herbs except for
Poria cocos (Schw.) Wolf and
Glycyrrhiza uralensis Fisch. belonged to Module 1 and the “Liver” meridian (Table
1). All herbs except for
Ligusticum chuanxiong Hort. belonged to the viscera (臟) group and not the bowel (腑) group. In contrast, of the top 10 herbs in the external application group, five belonged to module 2. They also affected bowel meridians such as the “Stomach,” “Large intestine,” and “Urinary bladder.” Thus, it may be that Module 2 is associated with external application, but further study will be needed, as we only conducted network analysis on the internal medicine group. Many of herbs in Module 3 belong more to the “Stomach” meridian. However, there were not module 3 medicinal herb in the internal and external groups top 10 herbs, with the exception of
Poria cocos (Schw.) Wolf and
Glycyrrhiza uralensis Fisch., indicating that treatment of digestion and absorption, which are related to Module 3, may be an adjunctive strategy in traditional Asian medicine. However, further research is needed in this regard.
Additional analysis was conducted on the top 20 medicinal herbs of each module (Table
4). In Module 1, 13 herbs were tonifying and replenishing medicinals (補益藥), mostly oriented towards the “Liver” and “Kidney” meridians. Therefore, Module 1 herbs are characterized as tonifying the “Liver” and “Kidney” meridians. Five dampness-draining diuretic medicinals (利水滲濕藥) and five
Qi-regulating medicinals (理氣藥) occupy half of Module 3. Most of these were oriented towards the “Spleen” and “Stomach” meridians. Therefore, Module 3 herbs are related to digestive function.
Module 2 comprised five exterior-releasing medicinals (解表藥), three interior-warming medicinals (溫裏藥), and three blood-activating and stasis-dispelling medicinals (活血祛瘀藥). [
12,
13] Thus, Module 2 was apparently associated with excretion and divergence., Presumably, Module 2 herbs act on the body surface or are external medicines, although further research is needed in this regard.
Existing studies on pattern identification have taken a top-down theoretical approach. In contrast, the present research adopted a practical, bottom-up approach based on formulas that are prescribed in clinical practice. We conducted this novel approach to pattern identification by carrying out a network analysis of medicinal herbs used in alopecia treatment. We rediscovered the classical pattern identification of alopecia treatment, and we suggest that clinicians adopt a “Liver” or “Stomach”-oriented approach to alopecia treatment.
The current research had several strengths. To our knowledge, this was the first study that used bioinformatics methods and searched Chinese, English, and Korean databases to assess which medicinal herbs have been used to treat alopecia. We adopted a practical network analysis approach based on formulas that are frequently used in clinical practice, rather than a theoretical/literature approach. Using this method, we explored the frequency, combination patterns, and meridian tropism of medicinal herbs used in alopecia treatment. We also classified herbs into three modules, confirming the value of classical pattern identification and the meridian tropism theory. Moreover, we explored the pathology of alopecia from the perspective of traditional east Asian medicine.
Our data mining methodology, which employed ARM and network analysis, also had several strengths. Firstly, in the ARM method, the number of herbs comprising the combination must be determined in advance. To overcome such shortcomings, we used network analysis to look at the overall combination pattern of medicinal herbs without limiting the number of herbs in the combination. Secondly, previous top-down research based on ancient literature has offered hypotheses about the pattern identification category of alopecia. In contrast, our bottom-up study categorized herbs into three modules based on the combination patterns of the formula. Lastly, previous research was limited in that it could only “qualitatively” interpret the characteristics of medicinal herbs or formulas used in alopecia treatment. We overcame this limitation by extracting significant “quantitative” characteristics using the permutation test.
Our research also had several limitations. The present study was based on the frequency of formulas used in clinical practice and literature. For this reason, we could not evaluate new candidate medicinal herbs emerging from recent clinical/experimental studies, neither could we reflect the importance of medicinal herb dose in each formula. Relatedly, we did not evaluate the clinical effectiveness of each formula in our study. Further clinical/experimental studies are needed to assess whether the classifications derived from our research have real meaning. Meridian tropism theory is controversial and may not accurately reflect the characteristics of each medicinal herb. Finally, we did not analyze external medicine, and the heterogeneity within Module 2 was not completely resolved.
The present research raises several indications for future research. We should analyze externally applied alopecia treatments, and we need to conduct a proof-of-concept study to corroborate our research. Using network pharmacologic analysis of medicinal herbs in each module, a hair loss mechanism could be identified based on meridian tropism theory (traditional medicine theory). Such studies may also indicate the pharmacological mechanism of hair loss treatment (western medicine theory). Multi-component, multi-target concepts are essential in herbal medicine pharmacology. Thus, we could propose new research methodology based on the techniques used in the present study. This methodology could be utilized to develop new hair loss drugs from natural products.