Clinical translation of stem cell therapy for spinal cord injury still premature: results from a single-arm meta-analysis based on 62 clinical trials

According to the Global Burden of Diseases report, the age-standardized rate of spinal cord injury (SCI) is 13 (11 to 16) per 100,000 people, accounting for 27,042,505 cases (24,976,608 to 30,148,230) worldwide. In 2016 alone, there were 935,000 new cases of SCI [1]. A variety of treatment strategies, such as drugs (methylprednisolone, ganglioside), surgery (decompression and fusion surgery), and rehabilitation therapy (electrical stimulation, functional exercise), have been used widely in the clinic [2]. However, their efficacy is limited due to severe pathophysiological injuries after SCI, such as neuronal necrosis and apoptosis, ischemia-reperfusion injury, and the demyelination and degeneration of spinal axons. Currently, less than 1% of patients with SCI show complete recovery of neurological function [3, 4]. Therefore, new and more effective treatments are urgently required.

Decades of research have deepened our understanding of the mechanisms underlying injury, neural repair, and regeneration in SCI [5, 6]. Studies have shown that stem cells can protect and regenerate the injured spinal cord through neuroprotection, immunomodulation, axon sprouting and/or regeneration, neuronal relay formation, and myelin regeneration, among other mechanisms [4, 7]. Therefore, stem cell therapy has attracted great attention as a treatment strategy for SCI.

Stem cell therapy has shown great progress in preclinical research. Moreover, it has shown high therapeutic efficacy in other conditions, such as hematological malignancies, burns, and corneal transplants. Thus, it has attracted the attention of countless eager patients, clinicians, pharmaceutical companies, and the media [8‐10]. Several interested parties are eagerly calling for the clinical translation of stem cell therapy for SCI, leading to a large number of clinical trials for stem cells. To accelerate this process, findings from rodent models were directly validated in human patients instead of using a step-wise testing approach in rodent models, followed by canine and feline models, primate models, and finally humans. As a result, a vast majority of preclinical trials have been conducted in rodents, and there are fewer than 20 studies in large animal models, accounting for < 2% of all preclinical research [11]. On the contrary, the number of related clinical trials has exceeded 100. Subsequently, unproven stem cell therapies in which patients receive autologous transplants could be performed without sufficient evidence, supervision, or informed consent [12, 13].

More notably, owing to the explosive growth of basic and clinical research in the field of stem cell therapy, other important issues may be neglected in the interest of rapidly advancing the clinical translation of stem cell therapy. Of these, the first and most important issue is the safety of stem cells. Stem cell therapy can lead to tumor formation, inappropriate migration, secondary injury, infection, and other adverse effects (AEs) [14]. The second is the ethical concerns related to the source, extraction, and transformation of stem cells. For example, human embryonic stem cells (hESCs) are known to show effective outcomes. However, are these embryos a potential form of life or just cells for research? What are the legal implications of destroying embryos? Should hESCs be derived from excess gametes or blastocysts during in vitro fertilization, aborted fetuses, voluntarily donated germ cells, or embryos produced for research purposes? In addition, although umbilical cord mesenchymal stem cells (UCMSCs) can be derived from discarded neonatal umbilical cords, they are the private property of mothers and neonates and are genetic resources protected by the law. These cells may cause allergy, hemolysis, and chronic toxicity after transplantation. Furthermore, they may carry pathogens and genetic mutations. Induced pluripotent stem cells (iPSCs) have the ability of unlimited proliferation and multi-directional differentiation. However, undifferentiated iPSCs can develop teratomas due to gene insertion mutations and uncontrolled proliferation and differentiation, leading to ethics-related debates [15, 16]. Currently, one of the most pressing ethical issues is the growing number of medical institutions that offer unproven stem cell-based treatments [17]. Finally, there are regulatory controversies over stem cells, and whether these regulations should be strict or lenient is often debated. The US Food and Drug Administration (FDA) has established a strict three-level regulatory system of “regulations-regulation-guiding principles” that requires research on stem cells to be subject to FDA review. Recently, the FDA has further stepped up its regulatory and enforcement activity for stem cell therapy to ban clinics and companies that could undermine the health of the entire regenerative medicine industry. In contrast, stem cell regulation in Japan is relatively less strict. The regulation of stem cells in Japan is subject to different levels of review. There are three risk levels based on the source of cells, processing methods, and the scope of application. With regulatory reforms, stem cell products could receive conditional approval after positive clinical findings in only 10 patients [18]. Benefiting from its comprehensive legal system and special approval policies, Japan has become a global leader in technological research and the development of cell therapy products. In other countries, such as China and India, several problems surrounding the regulation of stem cells remain. The regulatory boards only propose the basic principles for the development and evaluation of cell therapy products, and there is no implementation of stratified supervision based on the characteristics and applications of stem cells [19]. In addition, the necessary issues for the successful clinical application of stem cells remain unaddressed, including the risk-benefit ratio, mode of transplant strategy, oversights, conflicts of interest, surgical innovation, informed consent, and patient vulnerability [13].

Undoubtedly, despite several challenges, the initial success of stem cells in animal experiments and patients is indeed a cause for celebration and inspiration. It has brought immense hope of recovery for countless patients with SCI. However, experts must remain aware that there is still a lot of work to be done. Although systematic reviews of stem cell therapy for SCI have been published previously, they have had certain problems and limitations [20, 21]. Most of the current studies are single-arm, early-stage clinical trials with the main purpose of evaluating the safety of stem cells. However, published systematic reviews have only included randomized controlled trials, focusing on the effectiveness of stem cells, resulting in a limited number of included studies. Therefore, such systematic reviews do not fully analyze the existing data. Furthermore, they do not address the most important aspect, i.e., the safety of stem cell therapy and the feasibility of its translation from laboratory research to clinical practice. As such, it is unclear if we now have enough evidence to support the immediate clinical translation of stem cell therapy. Therefore, we performed a single-arm meta-analysis that included all the published clinical data and did not exclude studies based on the type of clinical research. Accordingly, we systematically evaluated the benefits and risks of stem cell therapy in patients with SCI in order to examine the feasibility of its clinical translation.

This systematic review and meta-analysis followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [22]. Although the study was not registered with the PROSPERO database or other comparable databases, we verified that no similar study has been registered before undertaking this study.

In recent years, the number of clinical trials based on stem cells has increased tremendously. Globally, there are now thousands of registered trials claiming to use “stem cells” in experimental treatments [85]. Thus, it appears that stem cell therapy has a well-established and strong clinical value. However, in reality, although some progress has been made, the clinical application of stem cells is still in its early stage. At present, clinical research on stem cells mainly includes phase I clinical trials, case series, and case reports. High-quality randomized controlled trials are lacking, and even simple controlled trials are few in number. Therefore, it is difficult to assess the efficacy of stem cells via head-to-head comparisons through meta-analysis [86]. In addition, although differences in cell types, sources, culture conditions, patient age, the degree of SCI, and other factors can make comparisons between studies challenging, these comparisons are still necessary [87]. Therefore, we analyzed the current evidence through a single-arm meta-analysis. Our findings showed that 48.9% of patients could benefit from stem cell therapy, i.e., 48.9% of patients showed at least one grade of improvement in the ASIA score after stem cell therapy. However, this only represents a slight improvement in sensory and motor function, far from the expected requirements for walking or daily activities. It is worth noting that the ASIA score-based assessment of sensory and motor function relies on subjective evaluation by the assessor and the patient, reducing result reliability to a certain extent [88]. Although this observed rate of 48.9% is lower than the rate of 78.57% obtained by Srivastava et al. in a randomized controlled trial with 200 patients [69] and the rate of 70% observed by Cheng et al. [31], given that small functional improvements are also extremely important for patient survival and well-being, stem cell therapy for SCI remains promising. Meanwhile, an increasing number of patients believe that the recovery of intestinal and bladder functions is as or more important than walking. Although our results show that the intestinal and bladder function improves in nearly half of all patients, these benefits only represent slight improvements in bladder fullness, urination impulse, and defecation (without sphincter control), which do not quite meet patient expectations [83].

Indeed, it is difficult to recruit enough SCI patients for clinical trials due to factors such as injury severity and patient age and physical condition. For cell-based trials, patients with SCI and grade A ASIA scores are recruited to avoid further damage. However, recovery in these patients may be low, and estimates of stem cell effectiveness may be inaccurate due to imprecisions in the measurement of subjective outcomes. Moreover, ASIA scale scores cannot fully reflect the severity of the pathology, and patients with the same ASIA grade may show a wide range of injury severity. Although the 48.8% rate of effectiveness appears encouraging, studies have shown that up to 70% of patients with complete cervical SCI can recover at least one spinal cord level within 1 year after injury [89]. Furthermore, 33% of patients with thoracic SCI have been found to show improvements of at least one grade [90]. Such spontaneous recovery in cases of SCI makes it difficult to confirm the efficacy of stem cells. Given that most trials lack a control group, it is not possible to rule out the contribution of spinal cord decompression or natural recovery to the therapeutic improvements observed after stem cell transplantation. Hence, stem cells cannot solely be credited with the therapeutic effects [90, 91]. Therefore, the actual therapeutic effects of the stem cells need to be further explored in standardized controlled trials according to the relevant guidelines.

While the extent to which stem cell therapy can benefit patients is unclear, the poor design and implementation of clinical trials also hinder the clinical applications of stem cells. Randomized controlled and double-blind human trials including placebo groups provide the most accurate and reliable data and are superior to observational studies or case reports [92]. However, most of the current studies are observational studies, case series, and so on. Small sample size and poor quality are also common problems in clinical trials [93]. Although the International Campaign for Cures of Spinal Cord Injury Paralysis (ICCP) has established a series of guidelines and standards for the design of clinical trials of SCI [92], 54.84% of the studies we examined were not registered on the clinical trial platform in advance. Hence, it was impossible to determine whether they reported all the results in accordance with the protocols and without any bias. The selective reporting of research results can lead to publication bias, thereby affecting the reliability of the conclusions garnered from a systematic review, or even leading to the opposite conclusion in some cases. Moreover, blinding is important to ensure the authenticity of clinical trial results. However, only 27.42% of studies included used blinding protocols. Failure to perform blinded evaluations may lead to the exaggeration of the actual effect and false-positive results [94]. Most of the included studies were phase I clinical trials, which are typically aimed at assessing the safety of stem cells. However, all studies explored and reported the efficacy of stem cells as the primary outcome measure, and ignored the reporting of AEs. Therefore, the safety of stem cells could be overestimated. At the same time, 35.48% of the studies followed patients for less than 1 year, which may have been insufficient to detect some AEs, such as tumor formation and inappropriate migration. Therefore, future clinical studies should extend the follow-up period as much as possible to fully explore the safety of stem cells. In addition, 79.03% of the studies did not pre-estimate the required number of patients to be enrolled in the trial or did not statistically analyze their results, reducing the likelihood of meaningful results. Most clinical trials do not have accurate and consistent inclusion and exclusion criteria for patients, leading to large differences in baseline characteristics, such as the location, type, injury duration, and severity of SCI. Moreover, the sources of stem cells and transplantation doses, timing, and routes also vary. This leads to study heterogeneity in meta-analyses and reduces the value of their findings in guiding future clinical trials. Hence, there is an urgent need to establish more stringent testing standards for stem cell therapy in SCI in order to standardize the design of clinical trials.

Patient safety should always be the top priority. The safety and AEs of stem cell therapy are mainly related to the inherent characteristics of the transplanted stem cells and the transplantation procedure. No serious AEs such as tumor formation were found in any of the studies we reviewed. Accordingly, these studies claimed that stem cell therapy was safe. However, it must be noted that the absence of serious AEs does not guarantee the safety of stem cell therapy. The 62 included studies reported a total of 28 AEs, including those affecting the nervous system, musculoskeletal system, digestive system, and cardiovascular system. Most of these AEs were mild, and the patients recovered completely after medical interventions. However, it is currently too soon to assume that stem cell therapy is safe, because this could further motivate those with blind confidence in stem cell therapy, which is not conducive to the scientific evaluation of its efficacy and safety. In addition, a study by Aspinall et al. showed that only 30% of clinical studies adequately reported various AEs during the clinical trial [95]. Hence, most studies may have ignored or even covered up AEs during the trial to “improve” the reported safety of stem cells. The meta-analysis by Zhao et al., which was based on control experiments, found a higher incidence of neuropathic pain and fever after stem cell therapy [96]. This was consistent with our findings, which indicated that the incidence of neuropathic pain, abnormal feeling, muscle spasms, vomiting, and urinary tract infection is higher than 20% in patients with SCI undergoing stem cell therapy. However, we reported a higher incidence of AEs than Zhao et al. because we enrolled more clinical trials, including single-arm studies. As a result, our sources of evidence were more comprehensive, and the results are more realistic. Although the observed AEs may not be related to the characteristics of the stem cells themselves, they could be affected by the transplantation technique, the severity of SCI, and the patients’ physical conditions. For example, neuropathic pain is mainly related to the use of large-caliber spinal needles during transplantation [97]. However, the observation of multiple AEs has greatly hindered the clinical applications of stem cells. In addition, after the subgroup analysis of stem cells from different sources, we found that mesenchymal stem cells such as BMSCs and UCMSCs—which were previously considered to have high safety—were associated with the highest rate of AEs among all the stem cells included in the current analysis [98]. This could be because the number of studies on BMSCs is the largest (61.29%, 38/62), and these studies also tend to have a large sample size. Hence, the AEs related to BMSCs are more apparent. In contrast, there was only one study on CBSCs. This study included 25 patients who were followed up for only 8 months. Hence, the AEs of CBSCs may be under-recognized. The possible AEs caused by stem cells can be explored properly only by reasonably calculating the required sample size, including patients who were clearly diagnosed with SCI, and ensuring a sufficient follow-up period. Hence, future studies should strongly focus on improving and standardizing these aspects.

Among the many safety issues surrounding stem cell transplantation, tumorigenesis is much more worrying than the fever and neuropathic pain caused by mild immune or allergic reactions [56, 79]. Stem cell products derived from hESCs and iPSCs confer the highest risk of tumorigenesis due to the presence of residual undifferentiated stem cells, malignantly transformed cells/mutations, and genetic instability. In addition, oncogenic activation due to the expression of foreign genes (e.g., various growth factors) and insertional mutagenesis of genetically modified viral vectors (e.g., retroviruses and lentiviruses) increases the risk of tumorigenicity and oncogenicity in stem cells [99]. For example, after iPSCs are treated with retroviruses, the retroviruses can get reactivated, resulting in tumor formation [100]. Although the tumorigenic potential of adult stem cells is thought to be lower than that of iPSC and hESCs, there have been reports of brain tumor formation after the injection of fetal NSCs [101]. Therefore, 71–74% of patients are “definitely” or “probably” not interested in participating in stem cell trials, and the possible risk of cancer is the greatest deterrent for these patients [46].

Currently, there is no global consensus on risk assessment strategies for the tumorigenicity and oncogenicity of stem cells. According to FDA guidelines, the risk of tumorigenesis can be decreased mainly by controlling the level of undifferentiated ESCs or other cellular impurities in hESC-derived cell products. The FDA guidelines also focus on the selection of preclinical animal species and the predictive ability of animal models to assess whether implanted cells form tumors at the transplant site and examine the malignant transformation of host cells and implanted cells in various tissues and organs. In fact, factors such as genetic and epigenetic variations in cell culture can increase the risk of tumorigenesis. Therefore, strategies to reduce the risk of tumors should be adopted during the development of high-risk stem cell products. Corresponding tumorigenicity testing methods can be established based on the quality and properties of candidate stem cells, the expected patient population, and tumorigenicity risk assessments. In preclinical studies, nude mice are usually used to study tumorigenicity in vivo, residual undifferentiated cells and transformed cells are used to study tumorigenicity in vitro, and server combined immune deficiency (SCID) mice are used to study tumorigenicity over the long term. No serious AEs like tumorigenesis have been reported in clinical trials so far, which could be due to the short follow-up period, because most of the studies (79%, 49/62) followed up patients for less than 2 years. Furthermore, the early termination of a trial is unfavorable for observing serious AEs such as tumor formation, especially because post-transplant ectopic growth can take up to 8 years [102, 103].

Although the success of stem cell therapy in preclinical studies has laid a good foundation for clinical research, its clinical translation has not been smooth. The number of new phase I and II clinical trials continued to increase from 2006 to 2012 but has since stagnated and even declined in 2018. The main reason is that the efficacy of stem cells is far less than expected [104]. First, animal studies attempt to minimize the variables in the experiment (e.g., baseline characteristics of the animals and the extent and location of lesions). However, patients with SCI show great heterogeneity (e.g., severity and location of injury, complications, age, sex, and rehabilitation training). Hence, the treatment efficacy observed in patients is often much lower than that observed in animals. Second, animal experiments often use the thoracic SCI model to study hindlimb motor recovery and related neural circuit changes. However, patients usually become quadriplegic due to cervical spine injury. In translational research, the extent and location of the lesion strongly influence the efficacy of the treatment. Some treatments yield beneficial effects only in subjects with specific lesion types [105]. Third, there are great differences in the inclusion and exclusion criteria among clinically recruited patients, and the location, severity, and timing of injury also differ. Therefore, it is difficult to include a homogenous group of patients even in high-quality randomized controlled trials, which makes the interpretation of treatment efficacy complicated and inaccurate. Fourth, the therapeutic effects observed in animal models are usually statistically significant. In clinical trials, the evaluation of functional results is more difficult because the small functional improvements generated by stem cells at the anatomical/histological level are difficult to detect. Finally, a careful investigation of the time window, dose, and route of stem cell transplantation is not a routine practice in animal studies, although these are major clinical problems that must be addressed in human studies [105].

In conclusion, the progress in clinical trials for stem cells has been exciting. However, most studies are in the early phase I/II stage, and clinical data are still being collected. It is too early to confirm that stem cells exert enormous curative effects. There are many differences and ambiguities in the selection of patients, cell types, timing of intervention, and doses and routes of stem cell transplantation in different clinical trials [75]. Therefore, close cooperation between preclinical and clinical aspects is warranted. Improving the safety, efficacy, and reproducibility of trials; identifying optimal transplantation parameters; strictly evaluating the benefits and risks of stem cell therapy; and strengthening supervision methods are urgent needs in this field [106, 107].

We estimated the efficacy of stem cell therapy and examined the rate of associated AEs to objectively and comprehensively analyze the feasibility of clinical translation, which is key for guiding future research. To our knowledge, this study is the first to perform such an analysis.

Nevertheless, there are some undeniable limitations to this study. First, there were great differences in the baseline characteristics of patients, types of stem cells, and timing, dose, and route of transplantation among the studies. Hence, we were unable to conduct further subgroup analysis and obtain more information. Second, real stem cells show self-renewal and can differentiate into various types of cells. In practical clinical research, the definition of stem cells is not strict. Bone marrow mononuclear cells and neural progenitor cells are also sometimes used as stem cells [9, 108]. In the absence of a clear definition, we included studies that claimed to examine “stem cells,” but some of these studies may have actually used stem cell-like cells. Third, we included only studies in English and did not examine grey literature and conference abstracts. As a result, there may have been some language bias, and some important findings may have been left out. Finally, although we developed a protocol and conducted the study in strict accordance with this protocol, we did not register it in PROSPERO or other comparable databases in advance, which may have led to reporting bias.

In order to better highlight the purpose and value of this study, we compared it with previously published studies. A total of 8 systematic reviews/meta-analyses (SR/MA) of stem cells for SCI were searched [20, 21, 109‐114], as detailed in Table 2. Although safety is paramount for the clinical application of any therapy, we found that the published SR/MA focused on efficacy indicators while largely ignoring the safety outcomes of stem cell therapy. Although 7 of the 8 studies reported AEs, their reporting was not comprehensive. The AEs reported in these studies mainly included fever, headache, and neuropathic pain, but our study reported 28 kinds of AEs caused by stem cell transplantation, of which neuropathic pain, abnormal feeling, muscle spasms, vomiting, and urinary tract infections showed an incidence of > 20%. We calculated the incidence of various AEs and highlighted the common types of AEs. In addition, the types of studies we included were more extensive, the number of studies was larger, and the data extracted were richer. Therefore, our study can provide abundant data on the safety of stem cells and can guide future research.

To facilitate the translation of stem cell therapy from the bench to the bedside, stem cells must be fully proven to be safe and effective in animal and clinical studies. The effectiveness of stem cells is very clear in rodent models. However, our systematic review of clinical trials found that although stem cells have great potential in improving neurological function in patients with SCI, stem cell transplantation could cause 28 kinds of AEs. Some of these AEs may be potentially serious and cannot be ignored, even if they have not been detected. Current clinical trials have small samples, low quality, and lack control groups. Thus, they cannot demonstrate the safety of stem cell therapy completely. Therefore, the possible AEs caused by stem cell transplantation need to be properly explored in preclinical large animal models and early clinical studies. Furthermore, optimal transplantation conditions and parameters need to be identified to improve the therapeutic effects of stem cells. At the same time, experts in the field must pay attention to safety concerns, as the proven safety of stem cell therapy will be paramount for its clinical adoption. Our findings show that the current evidence is insufficient to advocate for the widespread use of stem cell therapy for SCI and warn against its rapid clinical translation. Until well-designed animal experiments and high-quality clinical studies are performed, the introduction of stem cell therapy into the clinic should be slow and performed with caution.