An appraisal of laboratory models of androgenetic alopecia: A systematic review

Abstract Background Androgenetic alopecia (AGA) is the most common form of non‐scarring alopecia in humans. Several studies have used different laboratory models to study the pathogenesis and interventions for AGA. These study models have proved beneficial and have led to the approval of two drugs. However, the need to build on existing knowledge remains by examining the relevance of study models to the disease. Objective We sought to appraise laboratory or pre‐clinical models of AGA. Method We searched through databases (PubMed, ScienceDirect, Web of Science, World CAT, Scopus and Google Scholar) for articles on AGA‐related studies from 1942 to March 2019 with a focus on study models. Results The search rendered 101 studies after screening and deduplication. Several studies (70) used in vitro models, mostly consisting of two‐dimensional monolayer cells for experiments involving the characterization of androgen and 5‐alpha reductase (5AR) and inhibition thereof, the effects of dihydrotestosterone (DHT) and biomarker(s) of AGA. Twenty‐seven studies used in vivo models of mice and monkeys to investigate DHT synthesis, the expression and inhibition of 5AR and hair growth. Only four studies used AGA‐related or healthy excisional/punch biopsy explants as ex vivo models to study the action of 5AR inhibitors and AGA‐associated genes. No study used three‐dimensional [3‐D] organoids or organotypic human skin culture models. Conclusion We recommend clinically relevant laboratory models like human or patient‐derived 3‐D organoids or organotypic skin in AGA‐related studies. These models are closer to human scalp tissue and minimize the use of laboratory animals and could ultimately facilitate novel therapeutics.

though beneficial, however, have limitations since most are not fully representative of AGA. Also, available information on the molecular pathology and mechanism of AGA action is modest and thus hampers the progression of potential new treatments to human clinical trials. This problem may explain why only two drugs (Minoxidil, a vasodilator, and Finasteride, an alpha two reductase inhibitor) are currently approved and available for AGA treatment. Therefore, it is imperative to build upon existing models of AGA to facilitate the discovery of novel therapeutics.
This systematic review aims to appraise all AGArelated study models published to date for the testing of AGA interventions. However, not intended to evaluate the efficacy of AGA treatments.

| METHODS
The protocol for this study was registered and published in the International Prospective Register of Systematic Reviews (reg. number CRD42018107182; http://www.crd.york.ac.uk/PROSPERO).

| Search strategy
The Medical Search Headings terms used for the search were Androgenetic alopecia, Androgenic Alopecia, Hair Loss, and Male pattern baldness. Other search terms included In vitro, In vivo, Ex vivo, Finasteride, Minoxidil, Proliferation, Hair loss culture models and Xenografts, Tissue culture, Plant extracts, Hair shedding, Hair scalp biopsies, Induced alopecia, Cell differentiation, Cell lines, Primary cells, 2-D Models, 3-D Models, and Significant hair growth. The 'NOT' Boolean operator was used to exclude reviews, 'transplant studies', 'human studies', and 'clinical studies' (Table S1).

| Information sources
There was a search for articles on publication databases (PubMed, ScienceDirect, Web of Science, World CAT, Scopus and Google Scholar) from January 1942 until March 2019. A search strategy was developed for PubMed and adapted as per the requirements of other databases. A hand search in Google Scholar and Science Direct supplemented the search. The review also looked at unpublished articles from conferences and libraries (dissertations).

| Inclusion criteria
The inclusion criteria for the articles selected are as follows: preclinical laboratory studies and models (in vitro, in vivo and ex vivo) published in the English language. Our search excluded all articles involving human, clinical, diagnostic, drugs/phytochemicals efficacy and hair or cell transplant studies. Search results were filtered to include the research studies relevant to this review as per the pre-decided inclusion criteria.

| Data extraction and synthesis
We used a predesigned template adapted from Cochrane to populate the data extracted from the selected studies (Table S2). We analysed these data based on certain variables: the type of model (i.e., in vitro, ex vivo and in vivo), cell-types and techniques. We grouped research articles with similar variables.

| Quality assessment and risk of bias
The selected articles' quality was assessed by two authors (SN and AA) using a predesigned template validated by three independent assessors (Table S3). The quality assessors scored 10 similar questions, compared scoring results and any disagreements resolved through consensus. The articles were assessed against three questions, as follows: What's already known about this topic � Laboratory models are used to study the aetiology, pathogenesis and the efficacy of therapeutic interventions for AGA. These models, though beneficial, have only resulted in two FDA-approved drugs for the treatment of AGA, which may be partly due to the lack of disease-related models.

What's new or what does this study add?
� This systematic review summarizes published evidence of laboratory models used in AGArelated research and revealed the scarcity of studies that use human or patient-derived 3-D tissue culture models. These models are more suitable and more physiologically representative than 2-D cell culture and animal models for pre-clinical research. Thus, this article could guide the selection of appropriate study models for future research, and thus facilitate research translation.
1. Is the model used in the study appropriate or relevant (i.e., the use of patients' or control direct tissues or biopsies, patient-derived primary or immortalized two dimensional [2-D] or three dimensional [3-D] cells/cell lines)? Studies that used patient-related or human tissue instead of animal tissue as models had high scores. 2. Do the techniques and methods of analysis used in the study adequately address the research question? 3. Are the results from the research study validated by more than one research technique and are the results reproducible (i.e., the number of independent experiments)? Articles with results confirmed by three or more techniques, and a minimum of three independent experiments, scored higher than studies that used two or one methods.
The maximum score per question was five points.

| Study selection
Our initial search rendered 1228 related articles after removing duplicates. A further review of the full-text articles based on set eligibility criteria excluded 1128 and an additional two with the reason (spheroids but unrelated to AGA). Three more research papers retrieved from an updated hand search supplement the remaining 98 articles, thus bringing the entire selected research articles to 101. These articles included studies involving 70 in vitro, 4 ex vivo and 27 in vivo models ( Figure 1).

| Quality assessment of included studies
The overall results show a relatively high percentage (62%) of articles with a middle-risk score ( Figure 3). The evaluation of these articles revealed that 75% used appropriate models, while 25% of the studies having a high-risk score on the reliability of the models. The models used are monolayer (single or co-culture) of immortalized cell lines, human skin tissue, or mice/mice xenografts, or stump-tailed macaques. The articles had clear objectives or research questions, with methods and procedures that precisely addressed these objectives. About 38% of the selected studies had not validated experimental results with more than one technique.

| DISCUSSION
This systematic review, which sought to appraise models used in preclinical AGA research, rendered three groups of models: in vitro (n ¼ 70), ex vivo (n ¼ 4) and in vivo (n ¼ 27).

| In vitro 2-D models
The use of in vitro models is fundamental to all biomedical studies. The advancements made in studying cells, bacteria and viruses are often the first strides into understanding in vivo conditions. The in vitro models from included studies were 2-D monolayer (single or coculture cells) of immortalized or primary epithelial and mesenchymal cell lines. Immortalized cell lines are often used in research in the place of primary cells because they are cost-effective, easy to use and provide an unlimited supply of material. These cells can also grow indefinitely in culture and bypass ethical concerns associated with animal and human tissue use. However, immortalized cell lines undergo significant mutations to become immortal, which can alter cellular physiology. Also, genetic changes can occur over multiple passages, leading to phenotypic differences among isolates and unreproducible experimental data. Examples of immortalized cells used in the reviewed articles were human prostate cancer (PC3) cells, human keratinocytes (HaCaTs), prostate adenocarcinoma (LNCaP) cells, human sebaceous gland (SZ95 sebocytes) cells, human melanoma cells, human embryonic kidney 293 cells and the Shionogi carcinoma-115 cells. 4,6,10,43,[56][57][58][59][60][61][62][63][64][65] Although the use of these epithelial cells provided useful data to study potential AGA therapeutics, they are not related to this disorder. Therefore, some studies used DPCs. 43,45,48,53 Scalp DPCs are specialized mesenchymal cells of tissues affected by AGA that exist in the dermal papilla located at the basal layer of hair follicles. These cells play a pivotal role in the hair cycle. These cells were used as 2-D in vitro models in androgen-based and 5AR kinetics and inhibition assays to study the paradox of DHT, gene expression and AGA markers in two culture systems. The first system involved introducing growth factors (from keratinocytes-conditioned growth medium with or without medium containing fibroblast growth factor) to cultured DPCs. The second system included the co-culture of DPCs with effector cells (e.g., HaCaTs). Although these 2-D models are instrumental in enhancing the understanding of AGA's molecular pathology and mechanism, there have been limitations associated with these 2-D models. For instance, 2-D models do not accurately recapitulate the tissue architecture and cellular mechanisms present in a whole skin, impeding advancement towards effective AGA therapeutics.

| In vivo models
Twenty-seven studies used either mice or rats and monkeys as in vivo models, the most common being F I G U R E 1 A PRISMA diagram illustrating the search, screening and assessment strategy of articles reviewed, and qualitative synthesis of the selected articles C57BL/6, Kunming mice and male Sprague Dawley rats, male Wister/ST rats, C57BL/6NCrSlc strain mice, C3H/ He strain mice, B6CBACF1/J female mice, male Swiss albino and human scalp hair grafted into nude mice (i.e., human scalp hair xenografts). 21,46,[50][51][52]64,91,92 These rodents are used in general as models in medical research due to genetic, biological and behavioural similarities to humans. More so, they are useful in replicating many human disorders.
The stump-tailed macaque monkey was the only primate-related in vivo model used in the reviewed studies. 10,88,90 Stump-tailed macaque was the preferred choice because they possess hereditary balding characteristics similar, in many respects, to those of AGA in humans. However, macaques are found only in Asia, and the cost, risk and near extinction compromise the use of this model. Hence, the number of studies using these monkeys has diminished over the years. Animal models are considered suitable for studying human diseases as their biological milieu resembles human homoeostatic conditions. However, the use of animals in research is subject to strict ethical guidelines and often laden by the high human translational failures, perhaps due to inadequate mimicry of human pathophysiology. Thus, 3-D ex vivo models are fast becoming more prominent in understanding AGA.

| Ex vivo models
Only four of the included studies were ex vivo, using either diseased (AGA) or normal excisional and punch biopsy explants (whole skin organ cultures). 6,23,60,67 These models were used to study the action of 5AR inhibitors and the role of genes in AGA. Skin explants can maintain the cutaneous structure of the skin and, hence, allowing for studies, for example, that evaluate the effects of 5AR inhibitors on tissue morphology and gene expression. Although ex vivo organ culture is an easy-to-use and relatively cheap model, its use may be hampered by human ethical consideration and the availability of skin organ donors leading to insufficient skin specimens.    method evaluation because it reveals the specificity, accuracy, precision and, ultimately, the data's reliability. Validation is also essential to demonstrate that the model and technique used is suitable for the intended use. Many of the included studies (76%) were not exhaustive, with only 24% using three or more techniques to validate their results. Therefore, there is a need for continuity in work already done and the development of models with more relevant technologies.

| Risk of bias
Assessing the risk of bias of studies contained in the body of evidence is foundational and central to all systematic reviews to improve the transparency, consistency and scientific rigour of the research. Bias refers to factors that can confound the overall observations and conclusions of a study, which may lead to inappropriate recommendations. These discrepancies can result in wasted resources and loss of opportunities to discover effective therapies.

| LIMITATIONS
None of the included studies used 3-D models of AGA. The use of 3-D AGA models (e.g., organ culture from either punch biopsies, skin tissues from patients [healthy or diseased] or cells isolated from tissue samples) has an advantage over 2-D models. These models allow experimentation with human diseased tissues under controlled conditions than would be challenging to achieve in 2-D models, which may prove unachievable with in vivo models, thus allowing for a more detailed cellular and molecular characterization. Interestingly, 2-D models can be used to create 3-D models. The 3-D models can either be organoid cultures or be organotypic skin cultures. Organoids are in vitro-derived 3-D cell aggregates derived from primary tissue, which possess similar composition and architecture to primary tissue, and exhibit organ functionality. On the other hand, organotypic skin cultures use primary human cells, and cell culture inserts to recapitulate the stratified epidermal architecture of the skin to replicate the normal anatomy and physiology. This approach, especially with whole skin punch biopsies, allows for the in vitro maintenance of skin appendages, for example, hair follicles. Therefore, organotypic skin cultures may be instrumental in preclinical AGA research to validate the mechanisms of diseases and test the novel therapeutics.

| CONCLUSION
There is substantial evidence on the use of various models in the study of many diseases, including AGA and the discovery of effective targeted therapies. AGA models to date have helped in the understanding of this disorder. Howbeit, most of these models are not ideal representatives of AGA in humans. Only four studies  used whole human skin biopsies from patients (diseased and healthy), a closer approximation of AGA in humans. However, none of the studies used 3-D organoids or organotypic skin culture models. Therefore, there is a need to improve these models with a focus on the use of diseased and healthy biopsies or cell lines isolated from these biopsies. These 3-D models allow experimentation closer to humans, which may prove difficult or impossible with in vivo models while minimizing harm to animals. It is also vital to validate experiments with a minimum of three techniques. Finally, 3-D experimental data should ultimately be tested in human clinical trials to achieve formidable progress towards fast and effective treatment solutions for AGA. A good translation of AGA preclinical research into human trials will ensure the discovery of new and more effective drugs.