This study employed a systematic review and meta-analysis methodology, adhering to the guidelines outlined in the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) checklist.

The research question was created using the PICO format: among patients with androgenetic alopecia, does autologous platelet-rich plasma, compared to a placebo, result in hair density increase in randomized clinical trials?

To be included in the analysis, studies needed to meet the following eligibility criteria. Firstly, only full-length original articles reporting on randomized clinical trials were considered. The intervention of interest was the administration of Platelet-Rich Plasma (PRP) injections. PRP can either be activated or non-activated. The use of PRP in combination with the platelet activator/agonist allows for the release of multiple growth factors and differentiation factors upon platelet activation, which can enhance the process.30 Activation can be done by adding a platelet activator/agonist as topical bovine thrombin or 10% calcium chloride to the PRP solution right before injection. The control group consisted of individuals who received a placebo. This is to control PRP efficacy. Comparison with other standard therapies, such as minoxidil or 5-alpha-reductase inhibitors, could contaminate data and provide non-representative results. If PRP doesn’t show benefit compared to placebo, there is no point in comparing it to other therapies. Our main outcome of interest was changes, either positive or negative, in hair density. It refers to changes in the number of hairs per unit area (cm2) on the skin surface. Changes in hair diameter were secondary outcomes.

The target population comprised patients aged between 15 and 63 years diagnosed with AGA, as is the most likely population to be affected by androgenetic alopecia, who would potentially search for and benefit from PRP interventions. Secondly, women and men can be included, as androgenetic alopecia can affect both genders. While it’s more prevalent in men, women can also experience this condition. Characterization by the Norwood I‒VII and Ludwig I‒III classifications. Some studies use inter-patient and others use intra-patient comparison; in this review, both could be included. Studies with a sample size of fewer than ten patients, those lacking information on the PRP processing method, or studies that subjectively assessed increased hair density were excluded.

Furthermore, patients with chronic diseases, individuals on anticoagulant therapy, pregnant women, those with malignancies, and those utilizing complementary therapies were also excluded. Additionally, studies focusing on other types of alopecia, such as alopecia areata, cicatricial alopecia, or lichen planus, were not considered. There were no limitations regarding the PRP processing method.

The authors conducted a comprehensive search until April 1, 2023, in seven databases. These databases included MEDLINE (through PubMed), Virtual Health Library (BVS), EMBASE, Cochrane Library, Web of Science, Scopus, and MedRxiv (a source of grey literature, encompassing pre-print studies). The selection of these databases aimed to capture a wide range of relevant studies for inclusion in the review and meta-analysis. References from studies cited in this review writing process were also screened and eligible for inclusion.

For the search in electronic databases, relevant Medical Subject Headings (MeSH) terms and keywords were utilized. These terms were adapted based on the specific database. The following terms were employed to ensure a comprehensive search: “Hair regrowth”, “Hair restoration”, “Hair regeneration”, “Hair Growth”, “Alopecia”, “Baldness”, “Hair Loss”, “Hair Losses”, “Loss, Hair”, “Losses, Hair”, “Alopecia, Male Pattern”, “Male Pattern Alopecia”, “Baldness, Male Pattern”, “Male Pattern Baldness”, “Female Pattern Baldness”, “Baldness, Female Pattern”, “Androgenetic Alopecia”, “Pattern Baldness”, “Baldness”, “Pattern”, “Androgenic Alopecia”, “Alopecia, Androgenic”, “Alopecias, Androgenic”, “Pseudopelade”, “Alopecia Cicatrisata”, “Alopecia Cicatrisatas”, “Platelet-Rich Plasma”, “Plasma, Platelet-Rich”, “PRP”, “thrombocyte rich plasma”, and “Randomized Clinical Trial (Publication Type)”. To improve the sensitivity for identifying randomized clinical trials, recommendations from McGill University were followed.31 Specific search strategies for each database utilized in this study can be found in the supplementary file.

Two independent reviewers conducted the assessment of titles and abstracts for all identified clinical trial reports. Blinding was ensured, with each reviewer unaware of the decisions made by the other. When possible, studies that met the inclusion criteria were selected for further evaluation. In cases of disagreement between reviewers, a third reviewer was involved to make the final decision regarding study inclusion or exclusion. The Rayyan software was utilized to facilitate the selection process, particularly for reading and screening titles and abstracts.

Each reviewer independently performed data extraction without access to the other reviewer's extraction. The following data were extracted from the selected studies: author, year, location, comparator, study design, sample size, follow-up time, method of PRP preparation, capillary density and diameter measurements, and funding sources. A predefined Excel spreadsheet was used to record the extracted data for each reviewer.

Risk of bias assessment was conducted using the RoB-2 tool provided by Cochrane. This tool assesses possible biases in five domains, including randomization, intended intervention, missing data, outcome measurement, and outcome reporting.32 The GRADE method was employed to assess the strength of the recommendation for the intervention.33

Included studies were grouped based on the characteristics provided in Table 1. The extracted data were qualitatively synthesized and recorded in a dedicated spreadsheet. For studies that provided the mean and standard deviation of the outcomes of interest, a random-effects meta-analysis was conducted using the restricted maximum likelihood method. When these data were absent, authors were contacted. When authors didn’t answer, if provided sufficient data for estimating the mean and standard deviation, the study would be included in the meta-analysis. The effect size and 95% Confidence Intervals (95% CI) were measured by the mean difference. Statistical heterogeneity was assessed using the I² statistic, with values below 25% considered low, between 25% and 50% considered moderate, and above 50% considered high. Subgroup analysis and meta-regression were performed based on the use of a pre-application activator, sex, study design, risk of bias and spin method following the aforementioned meta-analytic model. Statistical package STATA® version 16 was used for data analysis. Funnel plots followed by Egger’s test were done to assess asymmetry and publication bias.

Gressenberger P:34 The Interquartile Range (IQR) was calculated from the reported range as Range/4. This IQR value was then used to estimate the standard deviation according to the empirical rule relationship of IQR/1.349. The median was used as an initial point estimate for the mean. From these estimates, the variance was determined as the square of the standard deviation. Finally, a 95% CI was placed around the estimated mean using a t-distribution, with degrees of freedom based on the sample size. This process allows approximate quantification of the mean and variation based on limited summary statistics when the full dataset is not available for direct calculation.

Chuah SY:35 Hair density values at 6 months were calculated by taking the baseline values and adding or subtracting reported difference values. Standard deviations were also adjusted to account for variability in the different groups. Specifically, the baseline means, and standard deviation were used as a starting point. Then, the difference in means was either added or subtracted based on whether the difference was positive or negative. The difference standard deviation was also added or subtracted to the baseline standard deviation. This allowed the full range of values to be determined while incorporating the variability represented by the CI. Standard Deviation (SD) was calculated by: SD = √(Baseline SD + Difference SD).

Dubin DP:36 The baseline range was used to estimate the standard deviation for each group according to the formula of range/4 * (sample size/(sample size - 1)). The mean difference from the baseline was then added to the original baseline mean to calculate the estimated post-treatment mean. The baseline standard deviation was assumed to apply after treatment. Finally, a 95% CI was placed around each estimated post-treatment mean using the t-distribution, with degrees of freedom based on the sample size and estimated standard deviation. This process allowed the estimation of key metrics after the intervention based solely on summary baseline data and mean treatment differences reported for each experimental condition.

Rodrigues BL:37 The authors extracted data from Fig. 1 with WebPlotDigitizer (Ankit Rohatgi, version 4.6, 2022). Then, the mean and standard deviation were calculated based on Luo et al. and Shi et al. works to estimate mean and standard deviation from the sample size, minimum value, first interquartile, median, third interquartile, and maximum.38,39

Fig. 1.

Prisma flow diagram.

(0.34MB).

The authors couldn’t estimate or retrieve data from Siah TW et al., therefore this report has only been included in qualitative synthesis, without participation in the meta-analysis model.40

The research protocol was registered at PROSPERO (International Prospective Register of Systematic Reviews) under the registration number CRD42023407334.

Following a comprehensive database search, 545 records were identified. Among them, the authors removed 227 duplicates, leaving 318 papers for full-text screening. After careful reference screening, 10 registers were also added. Of these, 26 papers were potentially eligible for inclusion. All reports were retrieved. Ultimately, 14 of them were included in the qualitative synthesis.34–37,40–49 Among the included studies, 13 provided sufficient data for quantitative synthesis through meta-analysis. It’s important to note that some studies had mean and standard deviation estimated as explained in the methods section. Supplementary Material contains the excluded reports list. Fig. 1 presents the PRISMA flow diagram.

The review included 431 participants, with 319 subjects undergoing PRP intervention and 310 individuals receiving a placebo. Intra-subject participants were considered in both groups. Most randomized clinical trials were conducted after 2019 and utilized inter-subject comparisons. Half used pre-application activators. Studies included patients with all degrees of alopecia. Table 1 summarizes the included studies' characteristics.

The majority of the studies used the trichoscan device and patient photos to aid in the analysis and measurement of capillary density as an outcome. Regarding PRP application, most studies administered monthly injections at a rate of 0.1‒0.2 mL/cm2, although Tawfik AA et al. achieved better results with weekly injections.41 The method of obtaining and applying PPR is presented for each study in Table 2.

Most of the included studies demonstrated a low risk of bias. However, Siah TW et al. exhibited a high risk of bias due to the lack of explanation regarding the randomization process.40 Similarly, Rodrigues BL et al. presented some risks in domains 1 and 4 related to randomization and outcome measurement, respectively.37 Qu Q, Chuah SY, and Toama MA also presented some concerns related to randomization, as no information from concealment was provided.35,48,49Fig. 2 demonstrates all results from RoB-2 risk of bias.

Fig. 2.

Rob-2 overall risk of bias.

(0.33MB).

Among the fourteen studies included in the synthesis, thirteen demonstrated improved hair density compared to the pretreatment baseline. However, Gressenberger P et al. reported worsening hair conditions without a significant difference between the groups.34 Around 57% of the studies showed a statistically significant difference in hair density between treated and placebo groups.

The meta-analysis showed a mean difference of 27.55 hairs/cm2 between the groups, with 95% CI ranging from 14.04–41.06. The forest plot is presented in Fig. 3.

Fig. 3.

Hair density forest plot.

(0.5MB).

The I2 statistic indicated high heterogeneity among the studies, with a value of 95.99% and Q(12) = 279.84. This result was statistically significant (p < 0.00).

In subgroup analysis, reports that used active PRP demonstrated a slightly lower mean difference of 25.25 hairs/cm2, with 95% CI ranging from 4.89 to 45.61, I² of 96.78%, p < 0.05, and Q(6) = 225.89. At the same time, studies using inactive PRP showed a similar mean difference of 29.79 hairs/cm2, with a 95% CI ranging from 11.16 to 48.41, I2 of 85.98%, p < 0.05, and Q(5) = 35.55.

Subgroup analysis of PRP preparation related to the spin method was also conducted. Both single and double-spin preparations were effective in increasing hair density (p < 0.05). Subgroup analysis was also performed considering study design, inter or intra-subject. Inter-subject comparisons presented a mean difference of 33.72 hairs/cm2, I2 = 78.19, 95% CI from 14.04 to 49.68 (p < 0.05), and Q(7) = 44.20. Intra-subject presented worse results, with a mean difference of 19.20 hairs/cm2, I2 = 98.46%, 95% CI from −4.02 to 42.43 (p = 0.10), and Q(4) = 209.61. The last subgroup was gender-related studies. Only two studies with only females presented sufficient data for meta-analysis, which limits data pooling. Male-only studies had a mean difference of 39.12, I² = 53.65%, 95% CI from 24.94 to 53.30 (p = 0.06), and Q(5) = 10.64. Mixed studies had a mean difference of 7.39, I2 = 73.01%, 95% CI from 1.57 to 13.22 (p < 0.05), and Q(4) = 18.96. High risk of bias presented mean hair density increase = 7.34 n/cm2, I2 = 79.77%, 95% CI from 0.96 to 13.73, and Q(3) = 18.87. Low risk of bias mean increase was 39.98, I2 = 77.49%, 95% CI from 25.09 to 54.87, and Q(8) = 41.08. Both p < 0.00. Subgroup-exclusive Forest plots are available in the Supplementary Material.

Five studies provided sufficient data for the meta-analysis. Fig. 4 presents the forest plot.

Fig. 4.

Hair diameter meta-analysis.

(0.29MB).

Results revealed a mean difference of 2.02 μm between treated and placebo groups, with 95% CI ranging from −0.85 μm to 4.88 μm, and I2 = 77.11% (p = 0.02).

Hair diameter couldn’t have publication bias assessed, as there were only 5 studies, and the Cochrane Handbook suggests a minimum of 10 studies.50 The funnel plot generated after hair density meta-analysis is available in Fig. 5. Following, the authors conducted Egger’s test, beta1 = 0.83, SE of beta1 = 0.816, z = 1.01, Prob >|z| = 0.3115.

Fig. 5.

Funnel plot.

(0.11MB).

Based on these results, the coefficient beta1 is not statistically significant at the 5% level since the p-value is greater than 0.05. The standard error of beta1 is also quite high relative to the coefficient value. Although Egger’s test didn’t identify publication bias, funnel plot asymmetry results indicate a clear presence of publication bias. Supplementary Material presents specific subgroups funnel plots.

The GRADEPro GDT analysis rated the evidence for PRP on hair density as low-quality evidence (++). This rating was based on a low overall risk of bias, high heterogeneity, direct outcomes, publication bias, and high precision in the results. As for hair diameter, it was assigned a very low-quality evidence rating (++). This rating was based on several factors, including a severe overall risk of bias, high heterogeneity, serious inconsistency, lack of direct outcomes, and imprecision in the results. Assessment can be seen in Table 3.

Results from meta-regression can be found in Table 4.

R-squared measures the proportion of total variation in the outcome variable that is explained by the predictors in the model. In this case, the value of 0.00% suggests that the predictors included in the model explain very little of the total variation.

The Wald Chi-Square test is used to assess the overall significance of the meta-regression model. In this case, the Chi-Square statistic is 2.68, and the p-value is 0.6134. Since the p-value is greater than the typical significance level of 0.05, the authors fail to reject the null hypothesis, indicating that the overall model is not statistically significant.

The action of PRP in treating AGA is primarily attributed to the growth factors and cytokines present in the platelet concentrate. These bioactive substances play a crucial role in stimulating anti-inflammatory, angiogenic, and proliferative pathways.52 The key factors associated with hair growth include Vascular Endothelial Growth Factor (VEGF), which is upregulated during the anagen phase and plays a role in determining the follicular size and inducing perifollicular angiogenesis; Glial Cell Line-Derived Neurotrophic Factor (GDNF), which suppresses the transition to the catagen phase and exhibits a decline in expression over time during the anagen phase; Insulin-like Growth Factor 1 (IGF-1), which synergistically works with GDNF to promote the proliferation of basal Ki67+ keratinocytes and prolong the anagen phase; and Fibroblast Growth Factor 2 (FGF-2), which promotes dermal papilla proliferation and can be enhanced by adenosine application or in conjunction with nanoparticles.53–58

Considering the importance of these factors, pre-application regulation of them can be crucial in achieving optimal results. Two studies included in this review examined the role of growth factors in hair regrowth. Siah TW et al. demonstrated a significant association between GDNF and hair density. Additionally, they observed a tendency towards a correlation between VEGF and FGF-2, although this finding may have been limited by the sample size.40 Rodrigues BL et al., although not establishing an explicit correlation between Epidermal Growth Factor (EGF), Platelet-Derived Growth Factor (PDGF), VEGF, and aesthetic outcomes, demonstrated an increase in hair density and the presence of anagen phase follicles, which was not observed in the group treated with saline injection alone.37

These findings further support the hypothesis of the efficacy of platelet concentrate in promoting hair growth. Moreover, Gentile P et al., although not directly studying growth factors, also observed beneficial effects following PRP application, including epidermal thickening, increased perifollicular vascularization, and an increase in ki67+ keratinocytes, thereby reinforcing the action of PRP.46

Qu Q, in his analysis, provided important information on alternative routes after studying mice pelage. He suggests that PRP treatment can activate β-Catenin, PDGF, and AKT signaling pathways, coupled with suppression of p53 levels to induce hair growth.48 This provides new insights into future research directions that could lead to a better understanding of hair growth with PRP.

The optimal technique for preparing PRP remains a subject of ongoing debate and investigation. In an effort to identify the most effective preparation method, Stevens J et al. conducted a review of the available literature.59 Their findings suggested several recommendations, including the use of subdermal bolus injections to minimize discomfort, the incorporation of an activator, monthly treatment intervals, and preparation using the single-spin method. These suggestions aimed to enhance treatment outcomes by optimizing the PRP preparation process.

In contrast with this, our meta-analysis revealed that the use of a pre-application activator suggested slightly lower improvements in hair density when compared to non-active. These findings indicate that pre-activating PRP prior to application may decrease its efficacy in promoting hair regrowth. Despite this, the difference found is very small. Therefore, the authors recommend that future studies explore the duality between pre-activation or not of PRP as a preparation method.

In contrast, the Tawfik AA study employed a different approach, with more frequent injections at weekly intervals using the double-spin method in women with alopecia.41 Remarkably, this study demonstrated exceptional results, showing sustained hair growth even after 6 months of treatment. Similarly, Singh S et al. achieved significant outcomes by utilizing the double-spin method while excluding Platelet-Poor Plasma (PPP) and adopting less frequent monthly applications.43 These findings suggest that shorter intervals between injections and the utilization of the double-spin method may contribute to more effective hair density improvement. Despite these data, our subgroup meta-analysis of the spin method revealed greater differences with double spin. However, the difference was small between both preparations.

Study design is of utmost importance. It’s interesting to note that our subgroup meta-analysis considering Intra/Inter-subject comparisons presented better results with inter-subject designs. This may be related to PRP spread in the scalp, affecting placebo areas in intra-subject comparisons. Accordingly, research conducted with this design may suggest smaller effects.

Our last subgroup meta-analysis was related to gender, in order to explain the high heterogeneity present. Although it was capable of reducing heterogeneity, the findings were not statistically significant. Additionally, the authors couldn’t ascertain individual-by-individual changes, limiting our analysis capability. More studies should be conducted exclusively on men and women for this issue to be answered.

Considering these findings, it’s apparent that various preparation methods for PRP can yield positive results. Further research is warranted to determine the optimal preparation technique for PRP in hair regrowth treatments. Factors such as injection frequency, activation protocols, and the specific centrifugation method employed may all influence the efficacy of PRP treatment. Future studies should aim to compare different preparation methods and evaluate their impact on hair regrowth outcomes to establish more precise guidelines for clinical practice.

Globally, the treatment approach for AGA often involves the use of 5-alpha reductase inhibitors, such as finasteride and dutasteride, as well as topical solutions containing minoxidil, an adenosine 5'-triphosphate-sensitive potassium channel opener.4,60–62 Adil A and Godwin M conducted a systematic review and meta-analysis comparing the effects of these medications against placebo.63 The analysis revealed mean differences in hair density compared to placebo as follows: finasteride 1 mg = 18.37 hairs/cm2, minoxidil 5% = 14.94 hairs/cm2, and 2% minoxidil = 8.11 hairs/cm2. Notably, finasteride exhibited high heterogeneity (I2 = 91%), while minoxidil demonstrated negligible heterogeneity (I2 = 0%).

In the present review, PRP therapy showed a mean difference of 27.55 hairs/cm2 compared to placebo, which is higher than the improvement observed with finasteride. However, both PRP and finasteride exhibited high heterogeneity. This improvement is clinically relevant, as PRP provides a notable increase in hair density without the requirement for daily use and the risk of undesired collateral effects. In addition, the PRP mechanism of action differs from other hair treatments, meaning it could be used in a combined therapy, enhancing hair growth results even more. These findings are consistent with the results reported by Gentile P et al., who conducted a systematic review comparing PRP therapy to conventional treatments, as well as with the findings of Singh S et al. and Pakhomova EE, who compared PRP to minoxidil and combined therapies.43,64,65 The evidence suggests that PRP therapy is more effective than minoxidil alone, and combination therapy may offer even greater efficacy. This suggests that PRP could be a viable treatment option for AGA.

In summary, PRP therapy has shown promising results in improving hair density in AGA patients. It appears to be a favorable treatment option when compared to finasteride and may provide enhanced efficacy when used in combination with other therapies. These findings support the growing body of evidence highlighting the potential of PRP as a valuable therapeutic approach for managing AGA.