An effective treatment for acute respiratory distress syndrome (ARDS) remains elusive.
Preclinical studies have demonstrated therapeutic potential of multiple mesenchymal
stromal cell (MSC) and MSC-related therapies in the context of ARDS, including autologous,
allogenic, xenogenic, and induced pluripotent stem cell–derived MSCs; multipotent
adult progenitor cells; and MSC-derived extracellular vesicles and exosomes (1). These
cells, or cell-free products, have been sourced from various tissues, including bone
marrow, umbilical cord, perinatal tissue, and adipose. However, few preclinical studies,
and no clinical trials, have compared these head to head. Considerable debate persists
as to what is the most efficacious cell product. The situation is further complicated
when factors such as scalability, storage, delivery, dose, safety, and regulatory
compliance are considered.
In this issue of the Journal, Batchinsky and colleagues (pp. 1283–1292) use a porcine
model of smoke inhalation injury to compare the therapeutic efficacy of autologous
bone marrow aspirate concentrate (BMAC) (n = 10), allogenic bone marrow–derived MSCs
(n = 10), and sham injection (n = 12) (2). Their well-established model combined wood
smoke inhalation, targeting a carboxyhemoglobin concentration >70%, with a full-thickness
burn affecting 40% of the total body surface area, inducing moderate ARDS in the sham
group by 24 hours. Animals were supported with protocolized intensive care until death
or 72 hours. Autologous BMAC was harvested after injury, at 24 and 48 hours, and concentrated
at the bedside, using a point-of-care, automated, closed-loop apheresis device. Allogenic
MSCs were procured from bone marrow aspirate before the experiments and were expanded
by serial passage. Each animal group received a total of three treatments: immediately
after injury, at 24 hours, and at 48 hours using a pulmonary artery catheter.
The study showed that BMAC, and to a lesser extent allogenic MSCs, effectively delayed
the decline in PaO2
:Fi
O2
ratio associated with injury, postponing the onset of ARDS. Notably, 4 of the 10 animals
treated with BMAC did not meet ARDS criteria by 72 hours. The BMAC-treated group demonstrated
longer mean survival time and a reduced mortality rate compared with the sham and
allogeneic MSC groups. Furthermore, the BMAC-treated group had a lower incidence of
acute kidney injury; lower systemic IL-6, IL-8, and HMGB1 (high mobility group box
1) concentrations; and larger amounts of normally aerated lung tissue on computed
tomography imaging.
This was a complex and resource-intensive study with key strengths. The challenges
of successfully executing a large-animal model that requires clinical-level intensive
care and, at the same time, deeply characterizing the effect of the interventions
should not be underestimated. The study was clinically relevant by testing MSCs and
BMAC after injury occurred (albeit before the onset of ARDS).
Although the authors conclude that autologous MSCs (from BMAC) appear more potent
than allogenic MSCs, the situation is more nuanced. Allogenic cells in this study
had some protective effects, but the dose used was much lower than those used in previous
large-animal studies or clinical trials of ARDS (usually 1 million to 10 million cells/kg
compared with a total dose of approximately 0.4 million cells/kg in this study). The
relatively modest therapeutic effect in the allogenic MSC–treated group is therefore
unsurprising. It is also an oversimplification to describe autologous BMAC as “autologous
MSCs.” BMAC is an enriched mixture of mononucleated cells, platelets, cytokines, and
growth factors, many of which have immunomodulatory effects in ARDS (3, 4). MSCs account
for only ∼0.001–0.01% of the total mononuclear cell population of bone marrow (5).
The dose of MSCs in the BMAC in this study was not known, but rather a total white-cell
count was available. Notably, this was less than the dose of allogeneic MSCs in the
comparator group. It is plausible that some beneficial effects demonstrated by BMAC
in this study relate to the non-MSC components (Table 1).
Table 1.
Comparison of Allogenic Mesenchymal Stromal Cell Products and Bone Marrow Aspirate
Concentrate
Allogenic MSC Preparations
Autologous BMAC
MSCs
Pure population: fixed number per dose
Variable percentage of total cell count
Platelets
—
Present (variable amount)
Leukocytes
—
Neutrophils, erythroblasts, lymphocytes, eosinophils, monocytes, and basophils (variable
amount)
Cytokines
—
IL-1Ra, IL-1β, IL-6, IL-8, GM-CSF, TNF-α variably reported
Growth factors
—
TGF-β, VEGF, PDGF, FGF-2/18, IGF-1
Other proteins
—
BMPs, osteoprotegerin reported in some but not all BMAC preparations
Potentially Immunogenic*
Yes
No
Persistence of cellular components
Rapidly cleared
Not yet known, but not foreign
Expansion/passage in vitro
†
Yes
No
Freeze–thaw cycle
‡
Usually (for clinical grade product)
No
Excipients
Usually DMSO
None
Exogenous animal product used in manufacture
Fetal bovine serum
None
Release criteria for clinical grade product
Known
Unknown
Potency assay
No standard assay
Unknown
Is suitability age dependent?
No
Likely
Definition of abbreviations: BMAC = bone marrow aspirate concentrate; BMP = bone morphogenic
protein; FGF = fibroblast growth factor; GM-CSF = granulocyte–macrophage colony-stimulating
factor; IGF-1 = insulin-like growth factor 1; MSC = mesenchymal stromal cell; PDGF = platelet-derived
growth factor; TGF-β = transforming growth factor-β; VEGF = vascular endothelial growth
factor.
*
Clinical studies have not shown clear anti-HLA antibody development, but long-term
studies are needed.
†
Serial passage linked to reduction in therapeutic efficacy.
‡
Freeze–thaw cycle may affect viability.
Although the present study supports further exploration of its therapeutic potential,
BMAC presents substantial challenges. The extent of MSC enrichment after concentration
in BMAC production varies according to site of aspiration, donor sex, and the device
used to perform concentration (6, 7). The concentration process may induce phenotypic
alterations within MSCs, modifying their in vivo actions (8), and in swine, the onset
of lung injury has been shown to render the MSCs retrieved from autologous bone marrow
less effective in terms of regenerative and immunomodulatory properties compared with
uninjured control animals (9). The issues of batch-to-batch variation and lack of
definitive potency assays for efficacy that have plagued the field of MSC therapy
are even more relevant for BMAC. Although BMAC offers the potential advantage of retrieving
MSCs that will not lose efficacy by undergoing repeated passage in vitro and avoids
the use of animal products such as fetal bovine serum or excipients such as DMSO,
it is difficult to understand how we might standardize BMAC composition at the bedside
for clinical trials and therapeutic use.
Furthermore, with increasing age, the cell yield from BMAC falls rapidly. A recent
study showed a greater than 67% reduction in mononuclear cell yield per milliliter
of marrow by 55–60 years of age compared with 19–20 years (6). In the LUNG SAFE (Large
Observational Study to Understand the Global Impact of Severe Acute Respiratory Failure)
study, the mean age of patients with ARDS was 61 years (95% confidence interval, 60–62 years)
(10). Even if therapeutically effective in other preclinical models of ARDS, clinical
trials may not be feasible if the substrate for autologous BMAC is insufficient in
most patients.
The heterogeneity of BMAC reflects the heterogeneity evident in the entirety of ARDS
cell therapy. Phenotypically different cell populations, obtained from diverse sources
using diverse protocols, and applied to a heterogeneous syndrome cannot be expected
to produce consistent outcomes, a fact evident in the results of recent clinical trials
in both coronavirus disease (COVID-19)–related and non–COVID-19–related ARDS (11–15).
So how can researchers address the heterogeneity, and can preclinical models contribute?
First, there should be a concerted effort to better characterize “MSCs” and their
related products used in experimentation, ideally beyond the minimal criteria required
to identify MSCs and including assays of function and relevant biological activity.
An example is tissue factor/CD142 (cluster of differentiation 142) expression, which
varies widely among MSCs but is implicated in the risk of thromboembolic complications
(16). Second, we need to develop an approach that allows comparison of different cell
products in a common model or a consistent product in alternative models. Given the
cost and complexity of high-fidelity models, such as the one described in the present
study, no single research group is likely to achieve this. The solution may be to
adopt protocols and collaborations that permit multicenter platform studies, as is
an evolving norm among clinical trials. The recent endeavor of the Stroke Preclinical
Assessment Network, which tested six treatment candidates across multiple models in
2,615 mice, serves as an exemplar (17).