Introduction

Spinal cord injury as a grievous neurological disease leads to paraplegia or tetraplegia with high rate of mortality and disability that mainly due to multifaceted complications including infections in bladder, renal failure, and cardiac and respiratory dysfunctions [1]. Based on the epidemiological survey, the annual incidence of SCI in North America is estimated to be between 24 and 77 per million people, or approximately 12,000 to 20,000 new cases per year. In addition, the total annual cost attributed to SCI in North America is about $14.5 billion, bringing a heavy burden to society in terms of healthcare costs [2, 3]. Those patients who suffer SCI will endure 4 misfortune phases based on pathophysiological classification system: acute (< 48 h), subacute (48 h to 14 days), intermediate (14 days to 6 months), and chronic (> 6 months). It is well known that the neurological impairment caused by primary injury (top 3 leading causes are vehicle accidents, violence, and accidental falls) is ineluctable; therefore, the rationale under current therapeutic strategy on SCI is acute-phase intervention preventing neural tissue from subsequent harmful cascade caused by the second injury and expecting vital impact on long-term function recovery [4].

Methylprednisolone (MPSS) as an initial therapy for SCI has been applied in clinic for more than 30 years. Recommendation developed by AOSpine expert panel in 2017 suggests a 24-h infusion of MPSS be offered to patients within 8 h of acute SCI as a treatment option. Although related corticosteroid clinical trials attract extreme criticisms over wound infection, gastrointestinal hemorrhages, sepsis, pulmonary embolism, severe pneumonia, and even death, MPSS until now is still an expedient strategy in the treatment of SCI [5,6,7]. With the understanding of SCI based on increasing clinical and basic researches, a plenty of key neuroprotective trials about pharmacological therapy (riluzole, magnesium, minocycline, GM-1, fibroblast growth factor, granulocyte colony-stimulating factor, hepatocyte growth factor) [8,9,10] and non-pharmacologic therapy (therapeutic hypothermia, CSF drainage) are underway. Furthermore, in terms of neuroregeneration, the specific clinical result of both drug therapeutics (Rho–ROCK inhibitor, anti-NOGO antibody) and nondrug therapeutics (spinal cord stimulation) are still in pending. To date, there are no standard and efficient treatments confidently available for SCI patients [11, 12].

Cell transplantation as a promising regimen is deemed to bring neural functional benefits following SCI. Candidate cell types including adipose-derived stem cells (ADSCs) [13, 14], bone mesenchymal stem cells (BMSCs), marrow mononuclear cells (MMCs), umbilical cord-derived mesenchymal stem cells (UCMSCs), olfactory ensheathing cells (OECs), autologous incubated macrophage (AIM), schwann cells (SCs), olfactory lamina propria (OLP), and fetal brain-derived neural stem/progenitor cell (FB-DNS/PCs) [15, 16] have an aptitude for providing trophic support, modulating the inflammatory response, regenerating lost neural circuits, and remyelinating denuded axons. Most basic researches and clinical trials also demonstrate that various cell types are generally feasible, but there still remains controversy exists due to significant heterogeneity, leaving us indeterminacy whether cell remedy possesses inherent merits in ameliorating patient’s prognosis. After retrieving and reviewing the literature, especially in systematic review and meta-analysis, there are 11 articles including 3 clinical (1 BMSC, 1 OEC, 1 stem cell) [17, 18] and 8 animal experiments (2 OEC, 1 BMSC, 3 stem cells, 1 SC, 1 NS/PC) [19,20,21,22,23,24,25,26,27]. However, most of the studies centering on a unitary cell type or self-control failure to provide us the comprehensive and detailed understanding of cell therapy, for example, whether different cell lines generate equivalent clinical outcome, what is the optimum time for cellular therapy, how many cells are suitable for the patient. Taken together, we aimed to probe into the issue mentioned above and provide evidence-based guidance for surgeons to make a better clinical decision by a meta-analysis.

Methods and materials

Search strategy

To perform comprehensive retrieval strategy, we systematically searched relevant studies published in the electronic database including PubMed, Embase, Medline, Cochrane Library. Search terms were subjected to the following: “cell,” “spinal cord injury”, and “trial” with the Boolean logic operator “AND,” “NOT,” and “OR”. All studies were published before January 2018 without language and country restriction. Reference cited in the relevant literature and other articles in the meta-analysis were also reviewed (Table 1).

Table 1 Detailed search strategy
Full size table

Inclusion and exclusion criteria

The criteria for inclusion article were: (1) clinical trials; (2) containing control group; (3) that all experimental groups were treated with cell transplantation. The criteria for exclusion article were: (1) animal and cell experiment; (2) self-control study; (3) that experimental groups were treated without the cell. Two authors assessed the potentially eligible studies independently based on titles, abstracts, and full texts of the relevant references according to inclusion criteria and exclusion criteria. Any disagreement was discussed and resolved by a third independent author.

Data extraction

Study titles and abstracts were reviewed by two independent investigators to decide if they satisfied the inclusion criteria, and the full text of the included studies was searched for further analysis. The data were extracted by two authors independently using a purpose-designed form: first author and year, region, design, level, number of patients, gender, age, injury, treatment, cell types, cell numbers, duration, follow-up, and outcome. Seven quantitative data were extracted as follows: (1) ASIA score (2) ASIA motion score (3) ASIA sensory score (4) Barthel Index score (5) Residual urine volume (6) Rehabilitative time of automatic micturition. Disagreement between the two reviewers was settled by the third reviewer. If any disagreements existed, a third author was consulted for discussion until consensus was reached.

Quality assessment

The quality of the RCT and CCT studies was independently assessed by the two authors according to a six-item scale recommended by the Cochrane Back Review Group. A: random sequence generation; B: allocation concealment; C: blinding of outcome assessment; D: incomplete outcome data; E: selective reporting; F: other bias. Furthermore, the CCT studies were assessed according to MINORS scale. Every study was assessed by 2 independent researchers and judgment of every item. Any disagreement with respect to eligibility during the extraction was discussed and resolved.

Outcome indictor

In the present meta-analysis, outcome indicators are included: (1) ASIA score (2) ASIA motion score (3) ASIA sensory score (4) Barthel Index score (5) residual urine volume (6) rehabilitative time of automatic micturition (7) cell transplantation techniques (8) adverse events.

Statistical analysis

The risk ratio (RR) and the corresponding 95% confidence interval (CI) were used for the dichotomous outcomes, and the standardized mean difference (SMD) and 95% CI were assessed for the continuous outcomes. The Chi-square test and Higgins’s I2 test were used to evaluate the heterogeneity. A P value < 0.1 for the Chi-square test or I2 values exceeding 50% indicated substantial heterogeneity. A fixed-effect model was used if significant statistical heterogeneity was absent; otherwise, a random-effect model was applied. We used Review Manager Software (RevMan Version 5.3, The Cochrane Collaboration, Copenhagen, Denmark) to conduct the statistical analysis. Subgroup analyses were performed with stratification according to cell types, phase, and the different dosage of cells.

Cell types were performed by dividing research into stem cell and non-stem cell group; the time from injury to cell transplantation is < 14 days, 14–30 days, and > 30 days; the different dosage of cells is < 1 × 107, 1–5 × 107, 10–20 × 107, and > 20 × 107.

Results

Selection of studies

Flow chart for the inclusion of studies is shown in Fig. 1. The literature search initially yielded 6389 relevant trials from PubMed (N = 110), Embase (N = 157), Medline (N = 5260), the Cochrane Library (N = 2), Wanfang (N = 99), CNKI (N = 35), and SinoMed (N = 726). After we reviewed the titles and abstracts of all 6389 articles, 5914 trials were excluded. We continued to refine the 475 candidate studies under provident review. Then, 137 duplicate studies were excluded. 338 candidate studies remain to confirm further. Because of nonconformance with inclusion criteria, 316 articles were excluded. Finally, 22 studies with 964 patients were included (Fig. 1). We recorded the characteristics of the 22 included trials, as well as the details of the clinical outcome measurement (Tables 2 and 3).

Fig. 1
figure 1

Flow diagram for the meta-analysis method

Full size image
Table 2 Detailed of included studies
Full size table
Table 3 Characteristics of included studies
Full size table

Characteristics of the studies included in the meta-analysis

Characteristics of the included studies were given in Tables 2 and 3, including the following item.

Methodological study quality assessment

A summary of methodological domain assessment for each study is detailed in Fig. 2 and Table 4.

Fig. 2
figure 2

Risk of bias summary

Full size image
Table 4 Methodological quality of the CCT studies
Full size table

ASIA score

Pooled analysis of 15 studies [29,30,31,32,33,34, 38,39,40,41, 44, 46,47,48,49] indicated that cell group improved ASIA score significantly compared with control group. The overall effect of ASIA score was 2.54 (95% CI 1.74–3.71, P < 0.00001); because of low heterogeneity (P = 0.06, I2= 39%), a fixed model was used (Fig. 3). Then, subgroup analysis according to cell types, phase, and the number of the cells was performed to further explore the more valuable details and potential sources of heterogeneity. Although heterogeneity was existed within several subgroups, low-to-moderate heterogeneity within every trial is still considered valid.

Fig. 3
figure 3

Forest plot of ASIA meta-analysis

Full size image

In cell-type subgroup analysis, studies were divided into 2 groups, 12 for stem cell and 3 for non-stem cell. The ASIA score of stem cell group was 3.36 (95% CI 2.13–5.28, P < 0.00001) with heterogeneity (P = 0.60, I2= 0%), and non-stem cell group was 0.82 (95% CI 0.39–1.73, P = 0.61) with heterogeneity (P = 0.13, I2 = 51%). Interestingly, the result indicated that cell from different sources showed opposite result, stem cell appeared to be more superior to non-stem cell in meliorating ASIA score (Table 5).

Table 5 Results of subgroup analyses for ASIA score, motion score, sensation score, Barthel Index, and residual urinal
Full size table

Furthermore, the optimal time of intervention for the patient after SCI is a vital issue and still perplexed surgeon to date. Based on the interval time from injury to cell transplantation, the studies were divided into three groups, two for < 14 days group, 1 for 14–30 days group, and 12 for > 30 days group. The ASIA score of < 14 days group was 0.75 (95% CI 0.35–1.64, P = 0.48) with heterogeneity (P = 0.10, I2 = 63%), 14–30 days group was 3.03 (95% CI 0.89–10.34, P = 0.08), and > 30 days group was 3.47 (95% CI 2.14–5.63, P < 0.00001) with heterogeneity (P = 0.57, I2 = 0%). The result indicated that cell transplantation in the chronic phase is more suitable for the patient with SCI. However, in acute and subacute phase, cell therapy does not exhibit efficiency in treatment for SCI. Although only three studies were classified in the acute and subacute group, it could still provide a referable standard for doctors and researchers. The heterogeneity from the subgroup was not high or severe, and the result can be accepted (Table 5).

Cell number subgroup analysis: 3 for < 1 × 107 group, 5 for 1–5 × 107 group, 3 for 10–20 × 107 group, and 3 for > 20 × 107 group. The cell number subgroup of < 1 × 107 was 1.11 (95% CI 0.60–2.07, P = 0.73) with heterogeneity (P = 0.02, I2 = 74%), 1–5 × 107 group was 4.20 (95% CI 1.87–8.95, P = 0.0002) with heterogeneity (P = 0.24, I2 = 27%), 10–20 × 107 group was 5.61 (95% CI 1.57–19.99, P = 0.008) with heterogeneity (P = 0.57, I2 = 0%), and > 20 × 107 group was 1. 98 (95% CI 0.97–4.06, P = 0.06) with heterogeneity (P = 0.75, I2 = 0%). Notably, different from our previous experience that more cells may present better efficacy, the result showed that appropriate amount of cell numbers between 1–5 × 107 and 10–20 × 107 was preferable quantity for patient with SCI comparing with other two cell number groups (Table 5).

The option of cell transplantation techniques is vital question for surgeon and patient. Subgroup analysis: 6 for posterior laminectomy + intraspinal cord cell injection group, 2 for intrathecal injection + intravenous injection group, and 5 for intrathecal injection group. The transplantation techniques subgroup of posterior laminectomy + intraspinal cord cell injection was 2.59 (95% CI 0.65–10.30, P = 0.18) with heterogeneity (P = 0.02, I2 = 64%), intrathecal injection + intravenous injection subgroup was 1.65 (95% CI 0.73–3.74, P = 0.23) with heterogeneity (P = 0.69, I2 = 0%), and intrathecal injection subgroup was 2.84 (95% CI 1.46–5.50, P = 0.002) with heterogeneity (P = 0.35, I2 = 9%). Notably, different from our previous experience that more cells may present better efficacy, the result showed that appropriate amount of cell numbers between 1–5 × 107 and 10–20 × 107 was preferable quantity for patient with SCI comparing with other two cell number group. The results showed that intrathecal injection appeared to be more superior to other two cell transplantation techniques (Table 5).

ASIA motion score

Pooled analysis of the 18 studies [28, 30,31,32,33, 35,36,37,38,39, 41,42,43,44,45,46,47,48] indicated that cellular treatment group significantly improved ASIA motion score compared with control group with heterogeneity (P < 0.00001, I2 = 76%). The overall change of ASIA motion score was 0.62 (95% CI 0.32–0.92, P < 0.0001) (Fig. 4). The overall effect exhibits that cell therapy improves the motion function comparing with control group. Subgroup analyses were conducted as mentioned above.

Fig. 4
figure 4

Forest plot of ASIA motion meta-analysis

Full size image

Cell-type subgroup analysis: 16 for stem cell and 2 for non-stem cell. The motion score of stem cell group was 0.69 (95% CI 0.38–0.99, P < 0.00001) with heterogeneity (P < 0.00001, I2 = 75%), and non-stem cell group was − 0.28 (95% CI − 0.89 to 0.33, P = 0.37) with heterogeneity (P = 0.31, I2 = 5%). The result consistent with ASIA score demonstrated that stem cell was superior to non-stem cell in motion improvement (Table 5).

Phase subgroup analysis: 2 for < 14 days group, 1 for 14–30 days group, and 15 for > 30 days group. The motion score of < 14 days group was 0.35 (95% CI −1.10 to 1.79, P = 0.64) with heterogeneity (P = 0.0003, I2 = 92%), 14–30 days group was 0.88 (95% CI 0.42–1.35), and > 30 days group was 0.64 (95% CI 0.30–0.98, P = 0.0002) with heterogeneity (P < 0.00001, I2 = 74%). The result showed that cell remedy in intermediate and chronic phase was superior to control group. However, in acute and subacute phase, there was no statistically difference comparing with control group (Table 5).

Cell number subgroup analysis: 4 for < 1 × 107 group, 6 for 1–5 × 107 group, 4 for 10–20 × 107 group, and 3 for > 20 × 107 group. The cell number subgroup of < 1 × 107 was 0.26 (95% CI − 0.36 to 0.88, P = 0.41) with heterogeneity (P = 0.008, I2 = 75%), 1–5 × 107 group was 0.64 (95% CI 0.11–1.16, P = 0.02) with heterogeneity (P = 0.0008, I2 = 76%), 10–20 × 107 group was 0.87 (95% CI 0.53–1.21, P < 0.00001) with heterogeneity (P = 0.32, I2 = 14%), and > 20 × 107 was 0.15 (95% CI − 0.22 to 0.52, P = 0.44) with heterogeneity (P = 0.57, I2 = 0%). The result identical to ASIA score showed that the interval of cell number between 1–5 × 107 and 10–20 × 107 demonstrated a better efficacy in treating SCI comparing with < 1 × 107 and > 20 × 107 group (Table 5).

In cell transplantation techniques subgroup analysis, the studies were divided into 3 groups, 4 for posterior laminectomy + intraspinal cord cell injection, 5 for intravenous injection + intrathecal injection, and 7 for intrathecal injection. The Motion score of posterior laminectomy + intraspinal cord cell injection group was 0.59 (95% CI − 0.31 to 1.49, P = 0. 20) with heterogeneity (P = 0.002, I2 = 80%), intravenous injection + intrathecal injection group was 0.87 (95% CI 0.14–1.60, P = 0.02) with heterogeneity (P = 0.0003, I2 = 81%), and intrathecal injection group was 0.46 (95% CI 0.06–0.86, P = 0.02) with heterogeneity (P < 0.006, I2 = 67%). The overall results showed similar to the results of motion that intrathecal injection and intravenous injection + intrathecal injection exhibited more favorable outcomes than posterior laminectomy + intraspinal cord cell injection (Table 5).

ASIA sensory score

Pooled analysis of the 18 studies [28, 30,31,32,33, 35,36,37,38,39, 41,42,43,44,45,46,47,48] indicated that cell transplantation group significantly improved ASIA sensory score compared with control group with heterogeneity (P < 0.00001, I2 = 81%). The overall change of ASIA sensory score was 0.77 (95% CI 0.42–1.12, P < 0.0001) (Fig. 5). The overall effect exhibited that cellular therapy was able to improve the sensory function of patients with SCI. Subgroup analysis was also performed.

Fig. 5
figure 5

Forest plot of ASIA sensation meta-analysis

Full size image

In cell-type subgroup analysis, the studies were divided into 2 groups, 16 for stem cell and 2 for non-stem cell. The sensory score of stem cell group was 0.87 (95% CI 0.54–1.19, P < 0.00001) with heterogeneity (P < 0.00001, I2= 77%), and non-stem cell group was -0.30 (95% CI − 1.51 to 0.90, P = 0.62) with heterogeneity (P = 0.15, I2 = 52%). The result also indicated that stem cell is superior to non-stem cell in sensory improvement similar to ASIA score and motion score (Table 5).

Phase subgroup analysis: 2 for < 14 days group, 1 for 14–30 days group, and 15 for > 30 days group. The sensory score of < 14 days group was 0.41 (95% CI −1.80 to 2.62, P = 0.72) with heterogeneity (P < 0.00001, I2 = 97%), 14–30 days group was 1.15 (95% CI 0.67 to 1.63, P < 0.00001), and > 30 days group was 0.79 (95% CI 0.44–1.13, P < 0.00001) with heterogeneity (P < 0.00001, I2 = 75%). The result showed that cell transplantation in intermediate or chronic phase was apt at treating SCI. However, control group was non-inferior to cell therapy in acute and subacute phase (Table 5).

Cell number subgroup analysis: 4 for < 1 × 107 group, 6 for 1–5 × 107 group, 4 for 10–20 × 107 group, and 3 for > 20 × 107 group. The cell number subgroup of < 1 × 107 was 0.39 (95% CI − 0.48 to 1.27, P = 0.38) with heterogeneity (P < 0.0001, I2 = 87%), 1–5 × 107 group was 0.71 (95% CI 0.22–1.19, P = 0.004) with heterogeneity (P = 0.004, I2 = 71%), 10–20 × 107 group was 1.62 (95% CI 1.18–2.07, P < 0.00001) with heterogeneity (P = 0.18, I2 = 38%), and > 20 × 107 was 0.08 (95% CI − 0.29–0.45, P = 0.65) with heterogeneity (P = 0.93, I2 = 0%). The result showed that the interval of cell number between 1–5 × 107 and 10–20 × 107 was superior quantity for treating SCI comparing with < 1 × 107 and > 20 × 107 group; this result was equal to ASIA score and motion score (Table 5).

In cell transplantation techniques subgroup analysis, the studies were divided into 3 groups, 4 for posterior laminectomy + intraspinal cord cell injection, 5 for intravenous injection + intrathecal injection, and 7 for intrathecal injection. The sensation score of posterior laminectomy + intraspinal cord cell injection group was 0.67 (95% CI − 0.59 to 1.93, P = 0. 3) with heterogeneity (P < 0.00001, I2 = 89%), intravenous injection + intrathecal injection group was 0.75 (95% CI 0.27–1.23, P = 0.002) with heterogeneity (P = 0.05, I2 = 59%), and intrathecal injection group was 0.92 (95% CI 0.24–1.61, P = 0.008) with heterogeneity (P < 0.0001, I2 = 87%). The overall results showed that intrathecal injection and intravenous injection + intrathecal injection exhibited more favorable outcomes than posterior laminectomy + intraspinal cord cell injection (Table 5).

Barthel Index

Barthel Index was widely used for assessing the activities of daily living. The overall effect of the 8 studies [28, 30, 31, 36, 37, 41, 43, 48] showed that cell therapy group significantly improved Barthel Index score compared with control group with heterogeneity (P = 0.31, I2 = 15%). The overall change of Barthel Index score was 0.88 (95% CI 0.67–1.09, P < 0.00001) (Fig. 6). The result showed that cell therapy recuperated activity of daily living compared with control group. Potential sources of heterogeneity were explored by subgroup analysis in terms of cell types, phase, and the number of the cell.

Fig. 6
figure 6

Forest plot of Barthel Index score meta-analysis

Full size image

Cell-type subgroup analysis: all 8 for stem cell. The Barthel Index score of stem cell group was 0.88 (95% CI 0.67–1.09, P < 0.00001) with heterogeneity (P = 0.31, I2= 15%). The result showed that stem cell is able to improve the activity of daily living (Table 5).

In the phase subgroup analysis, only > 30 days group was classified because of lacking acute- and subacute-phase researches. The Barthel Index score of > 30 days group was 0.88 (95% CI 0.67–1.09, P < 0.00001) with heterogeneity (P = 0.31, I2 = 15%). The result showed that cell treatment in intermediate and chronic phase may be a right opportunity in treating SCI. (Table 5).

Cell number subgroup analysis: 1 for < 1 × 107 group, other 3 groups were categorized in 3 for 1–5 × 107 group, 1 for 10–20 × 107 group, and 2 for > 20 × 107 group. The cell number subgroup of < 1 × 107 group was 0.58 (95% CI 0.13–1.03, P = 0.01), 1–5 × 107 group was 1.04 (95% CI 0.73–1.35, P < 0.00001) with heterogeneity (P = 0.23, I2 = 33%), 10–20 × 107 group was 0.61 (95% CI − 0.21 to 1.43, P = 0.15), and > 20 × 107 was 0.71 (95% CI 0.17–1.24, P = 0.010) with heterogeneity (P = 0.50, I2 = 0%). The result showed that similar to ASIA score, sensation and motion score, 1–5 × 107 group can be an appropriate dosage in treating for SCI. However, opposite results in 10–20 × 107 and > 20 × 107 were likely on account of insufficient data and other bias. More studies on SCI were needed for assessing (Table 5).

Residual urinal

Pooled analysis of the 4 studies [30, 31, 39, 43], all of which were for stem cell, demonstrated that cell therapy group significantly improved residual urinal compared with control group with heterogeneity (P = 0.24, I2 = 29%). The overall change of residual urinal was 1.00 (95% CI 0.61–1.38, P < 0.00001) (Fig. 7). The results indicate that cellular therapy is able to improve bladder function in patients with SCI (Table 5).

Fig. 7
figure 7

Forest plot of residual urine meta-analysis

Full size image

Rehabilitative time of automatic micturition

Overall analysis of the 3 studies [30, 31, 33], all of which were for stem cell, indicated that cell transplantation group significantly improved days of the rehabilitative time of automatic micturition compared with control group with heterogeneity (P = 0.73, I2 = 0%). The overall change of days of automatic urine was -0.67 (95%CI − 1.10 to − 0.23, P = 0.002) (Fig. 8). The result exhibited that stem cell remedy improved the rehabilitative time of automatic micturition (Table 5).

Fig. 8
figure 8

Forest plot of rehabilitative time of automatic micturition meta-analysis

Full size image

Adverse events

We evaluated the adverse event in cell transplantation group including 13 items: mortality, sepsis or wound infection, meningeal irritation, cardiopulmonary events, neurologic pain, neurologic deterioration, tumor, urinary system event, febrile reaction, headache, constipation, rigidity or spasticity, and abdominal distension. The top 5 adverse events were febrile reaction (11.5%), neurologic pain (11.3%), headache (2.6%), neurologic deterioration (2.4%), and rigidity or spasticity (1.6%). The results were given in (Table 6).

Table 6 Adverse events’ estimate
Full size table

Discussion

We have conducted an overall and systematic review of controlled clinical studies on the efficacy and safety of cell transplantation for the patient with SCI. Our results demonstrated that cell transplantation might be a meritorious method for SCI. Remarkably, in subgroup analyses, the stem cell makes a better performance than the non-stem cell in ASIA score, motion, and sensation score. Moreover, stem cell showed more efficacy than control group in Barthel Index, residental urinal, and rehabilitative time of automatic micturition. In the aspect of opportunity for delivering, > 30 day was a preferable stage in treating SCI. Paralleled results were observed in ASIA score, motion, and sensation score, Barthel Index. Than, results in ASIA score, motion, and sensation score convinced us to consider that the interval of cell dosage between 1–5 × 107 and 10–20 × 107 was the potent quantity for the patient with SCI comparing with < 1 × 107 and > 20 × 107 groups, as well as implied that both low and high dosages might gain no benefits for the patient with SCI. Furthermore, although all above results exhibited that intrathecal injection and intravenous injection + intrathecal injection showed more superior to posterior laminectomy + intraspinal cord cell injection, we noticed that the high heterogeneity existed in forest plot. So, a sensitivity analysis was also conducted, in which 1 study at a time was removed and the others analyzed to estimate whether the results could have been affected markedly by a single study. We found some interesting results. When we removed “Lammertse DP et al. 2012,” the results were reversed in posterior laminectomy + intraspinal cord cell injection group and then showed more superior to other 2 techniques (Supplements 13–15). Furthermore, similar reversed results were observed in sensation and motion sensitivity analysis group. We further analyzed the reason why “Lammertse DP et al. 2012” article made total reversed. First, cell-type-autologous incubated macrophage. Second, complete spinal cord injury. Thirdly, the conclusion of Lammertse DP`s research is the results failed to show a significant difference between the autologous incubated macrophages and control groups. Although forest plot exhibited relatively fine trend results in 3 cell transplantation techniques, we were inclined to choose minimally invasive techniques, which based on 3 reasons (1) the overall condition of patients with SCI (2) the patient`s tolerance to posterior laminectomy + intraspinal cord cell injection surgery, and (3) the patient`s capability of recovery after posterior laminectomy + intraspinal cord cell injection surgery. In terms of safety, although there is no standard definition for “safety,” in this article, we considered “safety” as (1) mild symptoms recovered within few days via treatment or self-healing, (2) or very low serious adverse events rate (such as mortality, infection) after receiving cell transplantation (3) or very low adverse events rate in long-term follow-up. Cell transplantation did not occur to serious adverse events; we found that although a higher degree of febrile reaction (11.5%) and neurological pain (11.3%) given in Table 6, above two symptoms were transient and self-healing in a few days and did not increase the meaningful mortality or morbidity. And then other adverse events such as meningeal irritation, headache, constipation, and rigidity or spasticity were also eliminated by symptomatic treatment or self-healing within few days. Only Nirmeen et al. [32] and Lammertse et al. [38] reported serious adverse events including mortality and infection. Notably, there were few adverse events in long-term follow-up exist in the 22 included studies. Top 5 adverse events were febrile reaction, neurologic pain, headache, neurologic deterioration, and rigidity or spasticity.

Comparing with previously published clinical meta-analysis

There are 3 clinical studies on meta-analysis. Li et al. [17] included 24 studies concerning of 594 patients treated with BMSC transplantation and provided that AIS improvement rate was analyzed in favor of BMSCs 6.13 (95% CI, 3.0–12.51; P < 0.001). The application of cell transplantation numbers between 107 and 108 groups seemed to be more beneficial than 106 and 109 (P < 0.05 for all groups). Our results were similar to conclude that stem cell seemed to show more potent efficacy with cell number of 107 and 108 maybe an optimum choice in clinical. Li et al. [18] applied OECs to treat chronic SCI patient; 11 articles that included 1219 patients were selected for review. Total AIS improvement rate, which was 39.0% (95% CI 28.1–51.1%), and also showed improvement in ASIA motor score and ASIA sensation score. We draw the semblable results in the effect, but lower mortality in our research. Fan et al. [19] included 10 studies comprising 377 patients. Three different origins of stem cell including BMCs, MNCs, and UCMSCs that were in treating for SCI were reviewed; the results showed that stem cell significantly improved AIS grading rate 2.95 (95% CI 1.64–5.29, P = 0.0003) and ameliorated lower-limb light touch score 3.43 (95% CI 0.01–6.86, P = 0.05) and lower-limb pinprick score 3.93 (95% CI 0.74–7.12, P = 0.02). However, it did not significantly improve motor score 1.89 (95% CI − 0.25 to 4.03, P = 0.08) and activities of daily living score 1.12 (95% CI − 1.17 to 4.04, P = 0.45). Intriguingly, our research showed that applying stem cell could bring improvement in motion score and activities of daily living score.

Strengths and limitations

To date, this is the first attempt to summarize the efficacy and safety of cell transplantation therapy for SCI. First, different from the previously published clinical meta-analysis, our up-to-date article retrieval yielded 21 eligible clinical controlled studies. It provided more high-level literature from origin and generated more credible results by evidence-based medicine analysis. Second, we also compared cell types to explore which cell lines bring more merits to the patient and provide researchers and surgeons an overall view of cellular therapy on SCI. Finally, we obtained more detailed information by conducting subgroup analyses to make further understanding in treatment for SCI.

This systematic review had some limitations. First, although we included studies with the controlled group, there are 8 studies without randomized method; this may lower the strength of evidence in our research. Second, some trials did not provide the detailed processes of cell transplantation, cell detection, and cell state, and we did not pursue these data to complete the analysis of the trials. Finally, socio-ethics factors, regional healthcare policy, and the medical condition might influence the outcome of cell transplantation.

Conclusions

Our systematic review with meta-analyses concluded that cell transplantation appears to be safe and has substantial beneficial effects in patients with SCI. Moreover, stem cell remains predominant in the treatment of SCI and provide more efficacy than other cells. Notably, treating SCI with stem cell, the dosage of cells between 1–5 × 107 and 10–20 × 107, in intermediate or chronic phase, may bring more advantage to SCI patients.