Routinely collected data: a path to assess safety of human mesenchymal stromal cell transplantation in randomized controlled trials Authors: Amir Hossein Norooznezhad Published: 2025 ABSTRACT Human mesenchymal stromal cells (hMSCs) are currently at the center of interest in many randomized and non-randomized clinical trials. According to the data, the number of trials on hMSCs has increased rapidly over time. However, the safety of this treatment, despite some available evidence, remains questionable. Routinely collected data (RCD) has become a helpful approach for gathering clinical data, especially in clinical trials. This method of data collection has helped investigators overcome the limitations of randomized controlled trials (RCTs), such as ensuring long-term follow-ups. Herein, the potential role of RCD in investigating the safety of hMSCs in RCTs, particularly concerns about their possible pro-tumorigenic potential, is discussed. The patterns of recent trials in this field suggest high feasibility and the potential for using RCD for this purpose. Introduction Human mesenchymal stromal cells (hMSCs) have been used in various clinical trials for decades to treat different types of pathologies [1]. However, similar to any other treatment, this therapeutic option is not without complications and adverse events. Among the safety concerns that have been raised for intravenous application of hMSCs is the potential role of these cells in cancer pathogenesis. Different studies have reviewed various possible pathways through which MSCs might act as pro-tumorigenic agents [2–5]. Despite this, in December 2024, the United States Food and Drug Administration (FDA) approved allogeneic bone marrow-derived MSC treatment for steroid-refractory acute graft-versus-host disease (SR-aGVHD) [1]. However, it should be noted that for conditions such as SR-aGVHD with the overall six-month survival of less than 50% [6], the risk-benefit assessment could change the decision strategy. For other conditions, especially chronic ones with a low mortality rate, the safety question remains. Thus, researchers have begun investigating this possibility. Routinely collected data (RCD) Routinely collected data (RCD) has now become of interest in many studies, especially in randomized controlled trials (RCTs). This data could be collected from different sources such as registries, electronic health records (EHRs), or administrative claim data. Besides many advantages it offers for gathering the outcome of interest in more pragmatic RCTs, it also might be helpful to mitigate one of the most important drawbacks of RCTs: short follow-up durations [7]. In this case, the outcome of interest is the diagnosis of any type of cancer, which is a pragmatic outcome. Since RCD is known to be informative in answering the more pragmatic outcome questions [8], it seems that using these types of data, especially cancer registries, EHRs, and administrative claim data, could help the investigators in trials on MSCs easily follow their population of interest for a longer time. However, not only future studies, but also past studies could benefit from RCD to answer the question of whether treatment with MSCs increases the risk of cancer, or if this answer is positive, the mortality from the cancer. Current trends of mesenchymal stromal cell transplantation According to a systematic review, by July 2020, a total of 1,138 registered clinical trials on clinicaltrials.gov had been registered from different countries, exploring the possible therapeutic potentials of hMSCs. This systematic review categorized these trials across 17 different fields of medicine, such as traumatology, pneumology, neurology, cardiology, and immunology (the top five fields, representing 65.4% of all trials). Although the outcomes of each trial could vary depending on the pathology of interest and other factors (beyond the scope of this letter), the rapidly increasing number of trials over time might suggest a therapeutic potential of this treatment option in some fields and pathologies. As they showed, a notable number of the trials have been registered after 2005 [9] which increases the chance of finding individuals in the modern databases. Safety of mesenchymal stromal cell transplantation and RCD Thompson et al. investigated the safety of intravascular administration of MSCs in a systematic review and meta-analysis using 55 RCTs published since 2004 (search date: April 2019). Among those 55 RCTs, 49.1% (N = 27) of them had a sample size of more than 40 individuals, which seems a worthy cutoff for following them through available RCD. Moreover, of the 27 trials involving more than 40 patients published in the last 10 years, 70.4% (N = 19) were published during the past ten years (2014–2024), which increases the chance of data availability in RCD datasets. Additionally, it seems that all the trials were conducted in a single country, which makes follow-up using RCD more feasible due to the availability of a single within-country dataset for each RCT. Moreover, among the safety profiles assessed by Thompson et al. malignancy was also investigated among 34.5% (N = 19) of the total included RCTs. According to their results, the pooled effect (relative risk = 0.93, 95% confidence interval = 0.60–1.45, I² = 0%) showed no statistically significant difference between MSCs and placebo (or no treatment) regarding malignancy risk. However, upon reviewing their data, we found that the duration of follow-up varied among the studies, ranging from 1 day to 60 months, with a median of 24 months and an interquartile range of 6 to 24 months. Because ‘1 day’ seemed erroneous, it was excluded, adjusting the minimum follow-up to 3 months (the other values remained as they were) [10]. The calculations performed in this manuscript and by R Studio 4.4.1 (R Foundation for Statistical Computing). Additionally, since cancer could be considered a rare and sometimes slow-progressing event, a larger study population with a longer follow-up duration could increase the statistical power to detect the outcome of interest. Therefore, given the growing number of trials investigating the possible therapeutic effects of hMSCs across various medical fields on different diseases/disorders, addressing the limitations of short-term follow-up for such a critical potential adverse event remains essential. In these types of trials, the researchers could link to an RCD source, such as their national cancer registry or related EHR(s), and track their included cases over a specific time interval. As mentioned, the median follow-up duration for the malignancy assessments in the RCTs evaluated by Thompson et al. [10] was 12 months, and this potential adverse event was investigated in only 34.5% of RCTs included in their study. If all 55 included studies had utilized RCD to assess the malignancy as a serious element of the safety profile, they could have periodically updated their safety findings to arrive at a more definitive answer. Furthermore, it is not necessary to wait until the end of the follow-up period; investigators could publish interim reports, or even brief communications, periodically (e.g., every few years) to address concerns regarding malignancy. The use of RCD in evaluating cancer development following pharmacological interventions is well-established. One example in this field is the study by Vinogradova et al. which evaluated the risk of breast cancer following hormone replacement therapy (HRT). Their research, conducted in the United Kingdom, included 98,611 women primarily diagnosed with breast cancer and 457,498 controls. Utilizing the Research and Clinical Practice Research Datalink (CPRD) databases, the study assessed the risk associated with different types of HRT, adjusting for baseline characteristics such as demographic factors, alcohol consumption, smoking status, past medical history (including comorbidities), family history of breast cancer, and drug history [11]. Among the most important issues to be considered in follow-up studies is the cell isolation method used in the trials, which could yield heterogeneous, non-clonal stromal cultures. These contain a mixture of multipotent stem cells, committed progenitors, and differentiated cells [12, 13]. Conclusion Taken together, we strongly encourage investigators of previous RCTs on this topic as well as those planning to perform their future trials, to use routinely collected data from various sources such as their national cancer registries, insurance claims, and any other available source to update their safety profile of trial even by publishing it as a letter to the editor as a short follow-up report.