A total of 2104 results were retrieved from the database search. Fourteen papers fulfilled the selection criteria and were included in this study. The process of a study selection is presented in Fig. 1.
Among a total of 340 patients, 213 had fmoAA, 60 had alopecia totalis (AT) and 67 had alopecia universalis (AU). The male to female gender ratio was 155:138 (Lee et al.  and Gadzhigoroeva  did not provide gender distributions). The mean age of patients was 26.7 years (range 1–80, SD 9.15, 95% CI 21–32.4; excluding missing data in Gupta et al.  and Lee et al. ). Separating the paediatric population from the adults was not feasible because of the unavailability of data.
In the papers selected, the primary severity of the disease was assessed according to the SALT scale , the percentage of scalp involvement, and the size of lesions (in centimetres). In the majority of articles, a “good response” was defined as either a regrowth of more than 50% or as a cosmetically acceptable effect.
Cyclosporine administration was exclusively oral while an intravenous and oral administration was used in the case of systemic corticosteroids (Tables 1, 2).
Assessment of Treatment Efficacy
The mean response rate of the whole group at the end of treatment was 65.00% (221/340; range 25–100%). The overall hair regrowth rates regardless of type of therapy were higher in the group with fmoAA (124/165; mean 75.15%; range 40–100%) than in AT (30/46; mean 65.22%; range 25–100%) or AU (24/52; mean 46.15%; range 25–100%). Articles with missing response rates in alopecia areata were not included in the statistical analysis (applies to Jang et al. , Shapiro et al.  and Lai et al. ).
Assessment of Corticosteroid-Combined Therapy vs. Monotherapy
Efficacy of the combined treatment was higher in comparison to CsA in monotherapy (152/219; mean 69.41%; range 0–80% vs. 69/121; mean 57.02%; range 6.67–100%). Articles with missing information about response rates in each subtype of alopecia areata were not included in the statistical analysis (applies to Jang et al. , Shapiro et al.  and Lai et al. ). Successful treatment in monotherapy was characterised by low heterogeneity in comparison to combined therapy (Figs. 2a, 3a).
Methylprednisolone was the most commonly used corticosteroid, administered mainly orally (Shaheedi-Dadras et al.  only used the intravenous route), with a mean response rate at the end of therapy of 67.76% (124/183; range 0–80%). The combined treatment with methylprednisolone was significantly more effective in comparison to CsA monotherapy (124/183; mean 67.76%; range 0–80% vs. 69/121; mean 57.02%; range 6.67–100%).
Mean Time of Treatment/Time to Effect
The duration of the treatment ranged from 2 to 36 months. The mean time of the treatment (data not extractable in two studies: Teshima et al. , Kim et al. ) was 6.75 months (SD 4.36, 95% CI 4.28–9.22).
Time to effect varied from 0.69 to 5.8 months. From the available data from seven studies (which included 180 patients) time to effect was 2.45 months (SD 1.75, 95% CI 1.15–3.75).
Assessment of Recurrence
The total recurrence rate was 47.40% (73/154, range 0–100%). CsA therapy with systemic corticosteroids was characterized by significantly lower recurrence rates in comparison to the CsA monotherapy algorithm (39/108; mean 36.11%; range 0–80% and 34/46; mean 73.91%, range 6.67–100%, respectively). Heterogeneity of both groups was high (Figs. 2b, 3b). Methylprednisolone combined therapy also showed smaller relapse rate compared to monotherapy (37/100; mean 37.0%; range 0–80% vs. 34/46; mean 73.91%; range 6.67–100%).
Assessment of Side Effects
Data regarding the side effects were incomplete and the analysis was conducted on the basis of the available information (Lee et al.  and Gadzhigoroeva  mentioned the exact side effects, yet did not provide the number of patients affected). If the occurrence of exact adverse effect was mentioned in the article but the number of patients was not provided, we decided to use a ‘≥’ sign instead of ‘=’ next to the countable sum to indicate the presence of this adverse effect. Side effects were observed in 36.76% patients (125/340). Gastrointestinal problems (n ≥ 28), hypertrichosis (n ≥ 20) and hypertension (n ≥ 9) were the most common side effects of therapy, regardless of the therapy regimen. Other symptoms included dyslipidaemia (n ≥ 8), headaches (n ≥ 7), oedema (n ≥ 6), weight gain (n = 6), acneiform eruption (n = 5), hirsutism (n = 5), musculoskeletal complaints (n = 3), respiratory disorders (n = 3), menstrual abnormalities (n = 2), weakness (n = 2), abnormal liver function tests (n ≥ 1), haematuria (n = 1), hyperbilirubinemia (n = 1), gingival hyperplasia (n = 1), paraesthesia (n = 1), urinary tract infection (n = 1), pruritus (n = 1) and ophthalmologic disorders (n = 1). Seven patients (7/340, 2.06%) were withdrawn from the treatment because of unacceptable adverse effects, mostly presenting themselves with hypertrichosis (n ≥ 3) and hypertension (n ≥ 2), as well as menstrual abnormalities, generalized oedema, abnormal results of the liver function tests or abnormal lipid levels.
Assessment of Dosage
A mean dose of 5.1 mg/kg/day (SD 1.01, 95% CI 4.21–5.98 mg/kg/day) in monotherapy was reduced to 3.29 mg/kg/day (SD 1.01, 95% CI 2.48–4.1 mg/kg/day) when systemic corticosteroids were added (the value calculated disregarding the works of Jang et al. , Lee et al.  and Kim et al.  because of a different dosage unit).
Regression test for funnel plot asymmetry revealed the presence of publication bias on an outcome defined as successful treatment (p = 0.047) (Fig. 4a). Bias did not exist for combined therapies (p = 0.086) (Fig. 4b).