*2.2. Data Analysis*

A number of qualitative or quantitative assessment variables (e.g., hair density, diameter, results of pull test, total number and frequency of PRP treatments, total amount of PRP per treatment, mean age of each group) were retrieved from each published clinical trial included in this meta-analysis. Data are expressed as means ± SD. The normality of the data samples was verified with the Kolmogorov-Smirnov, D'Agostino-Person and Shapiro-Wilk tests; normally distributed data were compared with Student's *t* test, and data sets that did not pass normality tests with its non-parametric versions: Mann-Whitney for independent data sets and Wilcoxon signed rank test for paired data. Most variables included in analysis were normally distributed, therefore we used directly Student's *t* test for independent samples (two-tailed), except for the absolute change in hair density, where we used its non-parametric variant, the non-directional Mann-Whitney test. We also performed linear correlation and regression analysis, indicating the values of Pearson's product-momentum correlation coefficient *r*, its statistical significance *p*, and the linear regression function. Statistical significance was assessed using a critical level *p* = 0.05. Forrest plots were generated with the clinical trials meta-analysis software Review Manager (RevMan) Version 5.4.1 The Cochrane Collaboration, 2020 [32].

#### **3. Results**

The literature search performed according to the criteria described within the Methods section resulted in the identification of 15 clinical trials, including studies on the effects of PRP treatment of AGA suitable for inclusion in our meta-analysis. The PRISMA flowchart describing the successive steps of the selection procedure is shown in Figure 1, while Table 1 presents the main features of the selected clinical trials. Most of them (86.7%) were randomized, 46.7% were double-blind, and 26.7% were single-blind. We found marked differences regarding the study design: 2 studies applied PRP/placebo on half-scalp and 3 studies on individual areas of the scalp in same patients, resulting in better case-control matching. Two studies included separate PRP and negative control (placebo) groups, while two other studies comprised two groups with different PRP administration protocols, differing in timing or preparation quality; three studies compared effects of PRP alone with those of PRP combined with minoxidil, or different PRP/minoxidil combinations, or other combinations of synthetic growth factors similar to those retrieved in PRP [16].

**Figure 1.** PRISMA flowchart showing the algorithm used for selection of the studies included in the current meta-analysis.




#### **Table 1.** *Cont*.

The inclusion and exclusion criteria used within the clinical trials selected for analysis are listed in Table S1. Most studies used the Norwood-Hamilton scale for assessment of male patients with AGA and the Ludwig scale for female patients. Exclusion criteria were also largely similar among studies, including the use of other topical or systemic hair growth medications, other causes of alopecia, other general diseases (endocrine, inflammatory, autoimmune, neoplasia, platelet and other bleeding disorders), and risk factors such as alcohol use or smoking.

There were also important differences in the PRP preparation methods, as shown in Table S2. The amount of peripheral blood used for one PRP treatment ranged between 9 and 60 mL, centrifugation was done as a single-step or two-steps, and calcium chloride was the most frequently used agent for α-granule release by platelet fractions. The most widely used monitoring methods were digital photography and phototrichogram analysis using the standard TrichoScan method.

Figure 2 presents the absolute and relative changes in average hair density for *n* = 17 study groups included in this meta-analysis. The mean hair density within these groups varied from an initial value of 141.9 ± 108.2 hairs/cm<sup>2</sup> (mean ± SD) to 177.5 ± 129.7 hairs/cm<sup>2</sup> at the last evaluation, a statistically significant increase as proved by a two-tailed Wilcoxon signed rank test (*p* = 0.0004). The same study groups were analyzed for effects on hair density using a Forrest plot shown in Figure 3; the main difference between baseline and post-therapy levels was 36.84 hairs/cm<sup>2</sup> (95% confidence interval 22.63–51.06).

**Figure 2.** Effects of PRP treatments on absolute and percentage change in hair density within the clinical trials selected for meta-analysis. (**a**) absolute changes in hair density (hairs/cm<sup>2</sup> ) at the end of evaluation period. (**b**) percentage changes in hair density relative to initial values. The legend is the same for both graphs. Data for experimental groups where only PRP treatment was applied are marked with squares, those where minoxidil treatment was added are marked with hollow star symbols.


**Figure 3.** Forrest plot showing the effects of PRP treatment on hair density in selected study groups.

We have also assessed the correlations between the total number of PRP treatments and the absolute or relative change in hair density: both were statistically non-significant (r = 0.176, *p* = 0.47 for absolute and r = −0.115, *p* = 0.64 for relative changes, Figure 4). However, the percentage change in hair density was significantly correlated with the frequency of PRP treatments (r = 0.50, *p* = 0.03), but the absolute change in hair density was not (r = 0.135, *p* = 0.58) (Figure 5). We also found no significant correlation between the amount of PRP administered per treatment and the absolute (r = −0.069, *p* = 0.78) or relative change in hair density (r = −0.215, *p* = 0.38), while the mean ages of the treatment groups were significantly correlated with the absolute (r = −0.44, *p* = 0.07) and relative (r = −0.56, *p* = 0.016) change in hair density (Figure 6).

**Figure 5.** Correlations between the frequency of PRP treatments and absolute or percentage change in hair density within the clinical trials selected for meta-analysis. (**a**) lack of significant correlation of absolute changes in hair density (hairs/cm<sup>2</sup> ) at the end of evaluation period. (**b**) significant correlation with percentage changes in hair density relative to initial values (same legend for both graphs). The regression lines are plotted in red, and the corresponding equations, values of correlation coefficients *r* and their probabilities *p* are marked on each graph.

**Figure 6.** Correlations between the mean age of experimental groups and absolute or percentage change in hair density within the clinical trials selected for meta-analysis. (**a**) almost significant correlation with the absolute changes in hair density (hairs/cm<sup>2</sup> ) at the end of evaluation period. (**b**) significant correlation with percentage changes in hair density relative to initial values (same legend for both graphs). The regression lines are plotted in red, and the corresponding equations, the values of the correlation coefficients *r,* and their probabilities *p* are marked on each graph.

We verified if the sex composition of study groups exerted an effect on initial values, absolute and relative changes in hair density, using data exposed in Supplementary Table S3. Thus, the initial hair density, the absolute and the relative (%) change in hair density did not show significant differences between study groups formed exclusively from males vs. groups of females or with mixed gender composition.

Although relatively few studies of those included in the meta-analysis provided relevant data, we assessed the effects of PRP therapy on the mean hair diameter and the results of the pull test. Therefore, for n = 8 study groups the mean hair diameter increased by 15.67 µm after therapy (95% confidence interval 9.77–21.57) (Figure 7), while the results of the pull test decreased on average by 5.32 (95% confidence interval 2.84–7.80) for n = 5 study groups (Figure 8).

8 of the 15 studies included in the analysis (53.3%) reported use of patient selfassessment questionnaires, and 5 of them (33.3%) used physician assessments. Patient self-assessment comprised most often the degree of satisfaction following PRP therapy, usually at several time points, and divided on four levels (e.g., highly satisfied, satisfied, dissatisfied, highly dissatisfied). One study reported the percentage improvement after therapy on a 5-level scale [44], and another one on a 1-10 scale [45]. One self-assessment questionnaire included supplementary items such as evaluation of results (Yes, No, Unsure/maybe), recommendation to other patients (Yes, No, Maybe), and motivation to continue (Yes, No, Maybe) [38]. Physician assessments were based on analysis of global photographs of the scalp or four-level satisfaction questionnaires.


**Figure 7.** Forrest plot showing the effects of PRP treatment on mean hair diameter in selected study groups.


**Figure 8.** Forrest plot showing effects of PRP treatment on results of pull test in selected study groups.
