In May 2016, researchers at the now-defunct StemCells Inc., Newark, California, terminated a Phase 2 trial of neural stem cells for spinal cord injury due to lack of efficacy. Now, two papers in the February 14 Stem Cell Reports argue that this outcome might have been predicted with additional animal testing. Separate groups of researchers led by Aileen Anderson and Mathew Blurton-Jones, both at the University of California, Irvine, compared the performance of clinical and research-grade neural stem cells from StemCells Inc. in mouse models of spinal cord injury and Alzheimer’s disease, respectively. In both studies, the research-grade lines showed a treatment benefit but the clinical-grade cells did not. The findings may help explain the poor outcome in StemCells Inc.’s Phase 2 Pathway Study, and provide a lesson for ongoing or planned stem cell trials, the researchers say. They had collaborated with the company on preclinical testing. “No two stem cell lines are the same. To move forward with stem cell therapy, we need to be testing, preclinically, the exact cell line that will be going into patients,” Blurton-Jones told Alzforum.

Other scientists in the field agreed that variability of cell lines is a widespread problem, and they supported the idea of testing final cell lots in animals before taking them into people. At the same time, they noted that the limitations of animal models, as well as limitations in understanding exactly how stem cells are therapeutic, make it challenging to get definitive results from preclinical testing. “We don’t have clear criteria for the quality of the cells or the mechanism of action,” explained Mahendra Rao at the New York Stem Cell Foundation Research Institute, New York City.

StemCells Inc., in a written response to the papers, disagreed about the need to test clinical-grade cells in animal models. “The ultimate test of efficacy resides in the human setting,” wrote company lawyer Ken Stratton. Former StemCells Inc. scientists did not reply to requests for comment on this story. 

Safety Issue?

Clinical-grade stem cells from StemCells Inc. formed undifferentiated masses in the brain ventricles of some mice, proliferating and extending neurites into surrounding tissue (arrows). [Courtesy of Marsh et al., Stem Cell Reports.]

StemCells Inc. generated its human neural stem cell lines from donated fetal brains. The cells remain neural precursors in vitro, but once injected, mature into fully differentiated neurons and glia. In early preclinical work, Anderson and colleagues found that the cells promoted remyelination and motor recovery in mice after spinal cord injury (see Sep 2005 news). The company advanced the treatment to clinical trials, completing a Phase 1/2 study in 2015 in which they transplanted the cells into the spinal cords of 12 people with injuries in the thoracic segments of their spines. Company scientists reported favorable safety and efficacy results at conferences, but those findings have not been published.

For the Phase 2 Pathway Study, StemCells Inc. enrolled about 30 adults with cervical spinal cord injuries that had happened at least three months prior. This type of neck injury is more common than thoracic damage, and more challenging to treat. Company scientists injected clinical-grade human neural stem cells into the medulla of the cervical spinal cord. They assessed participants’ motor functions six, nine, and 12 months later using the international standards for neurological classification of spinal cord injury (ISNCSCI). The first six treated patients notched significant improvements at six months, but these tapered off toward baseline by 12 months. A six-month interim analysis of the remaining participants revealed only modest benefits in the treatment group that did not reach statistical significance. Scientists pulled the plug, concluding that the gains in the larger cohort were unlikely to reach the primary endpoint (see press release). In Aug 2016, StemCells Inc. was bought by the Israeli robotics company Microbot Medical Ltd. (See Nasdaq notice). 

Anderson’s new paper now provides a possible explanation for the trial failure. The researchers compared human “research-grade” cells versus clinical-grade cells that StemCells Inc. had supplied them. The researchers injected both types of cells into the cervical spinal cords of immunodeficient mice with contusions in that area. They saw stark differences. When transplanted two months after injury, the research-grade cell line modestly improved the animals’ gait, but the clinical-grade line did not. In fact, the more clinical-grade cells that engrafted and survived, the worse the animals walked. In an acute injury model, where cells were transplanted nine days after injury, research-grade cells enhanced performance on numerous motor tests, but clinical-grade cells once again did nothing. Both types of transplanted cell survived equally well, with the majority becoming oligodendrocytes. However, clinical-grade cells matured poorly, generating about half as many mature oligodendrocytes as did the research line. In a bizarre twist, Anderson notes in her paper that the company later told her that the cells they supplied to her were not the exact cell line used in the clinical trial.

Blurton-Jones and colleagues found analogous results in AD model mice. Previously, they had reported that the research-grade neural stem cells enhanced hippocampal synapse formation and memory in 3xTg mice that were immunosuppressed with drugs (see Ager et al., 2015). In their new study, Blurton-Jones and colleagues tested a StemCells Inc. neural cell line made under Good Manufacturing Practice (GMP) conditions. This is a quality control requirement for therapies used in human trials. The initial batch of GMP cells sent by the company survived poorly in culture and could not be used, so the researchers requested a younger batch of GMP cells and cultured them in their lab. Thus, these clinical-grade cells were not identical to those used in patients, either.

Nonetheless, the authors wanted to test the long-term effects of stem-cell transplantation. Because pharmaceutical immunosuppression only works for about three months before a graft is rejected, first author Samuel Marsh transplanted the cells into the hippocampus of an immunodeficient 5xFAD mouse previously developed by their lab (see Feb 2016 news). The GMP cells migrated throughout the hippocampus and into neighboring subcortical and cortical regions as well as research-grade cells had in 3xTg mice, and nearly all survived five months later. However, treated mice performed no better than controls in the Morris water maze and Y maze, which test hippocampal-dependent learning and working memory, respectively. Synaptic density of treated mice actually declined.

Moreover, the transplanted neural stem cells differentiated poorly, with nearly all still expressing immature neuronal and oligodendrocyte markers five months later. In about one-quarter of the mice, the researchers found clusters of undifferentiated stem cells in the brain ventricles. About 4 percent of these cells were dividing, and in several cases, cells or neurites from the cluster penetrated through ventricle walls and infiltrated into the striatum (see image above). Injected stem cells are expected to stop dividing and to differentiate into mature cell types. Dividing cells are a concern because in theory, they could give rise to cancer. While a veterinary pathologist found no evidence of cancer, a neuropathologist noted the clusters shared some features in common with neurocytomas, a form of benign intraventricular tumor. The findings “warrant caution and long-term patient monitoring,” the authors suggested. In response, StemCells Inc. wrote, “After 10 years of clinical trial experience involving HuCNS-SC transplantation in more than 50 subjects (with clinical follow-up extending to five years post-transplant), absolutely no safety concerns regarding the HuCNS-SC cells have been identified.”

Why do clinical-grade neural stem cells work less well than research-grade cells in mice? There is no way to know for sure, because the production processes for the clinical-grade cells were not shared with the academic researchers, Anderson told Alzforum. However, she noted that reduced efficacy after scale-up has hampered other types of stem cells. Researchers attribute some failed trials of mesenchymal stem cells (MSCs) to mass production issues (see Galipeau, 2013; Chinnadurai et al., 2015). MSCs are used in most current AD stem cell trials. In some cases, cells need time in culture to recover after various manipulations, Anderson said. Factors such as thawing or dividing lines too many times can also weaken cells. Anderson believes companies need to take the time to test the effects of all these production decisions before moving to trials.

Clive Svendsen at Cedars-Sinai Medical Center, Los Angeles, agreed. “The most likely conclusion one can draw from the data is that there was a flaw in their manufacturing process,” he told Alzforum. Svendsen directs the Regenerative Medicine Institute at Cedars-Sinai and is involved in stem cell trials for several neurodegenerative diseases.

The Food and Drug Administration guidance on cell therapy products recommends preclinical testing of the final product that will go into patients, but does not require it. “In certain cases, due to the species-specific nature of the clinical product … testing of an analogous product may be a suitable alternative,” the guidance states. However, in practice, FDA consultants now strongly advise researchers to test their final lot in animals, particularly for safety issues, Svendsen noted. “The FDA has become much more sophisticated in the last five years with regard to stem cell therapies,” Svendsen said. Other researchers criticized the leeway the agency allows for researchers to use different cell lines in preclinical and clinical work. “I think that’s a mistake,” Blurton-Jones said.

Another potential problem was that StemCells Inc. generated cell lines from numerous different fetal samples, creating large variability between lines, Svendsen noted. All of these different lines were labeled identically as HuCNS-SC cells. Anderson advocates that companies should be required to disclose specific cell line reference numbers as part of collaborations with academic scientists and clinical trials, similar to the lot numbers used with chemicals. She also believes that companies would benefit from greater transparency and sharing of data with preclinical researchers. She noted that StemCells Inc. collaborated more closely with academic scientists on the thoracic spinal cord trial than on the cervical one, and obtained better results. “There has to be better communication between clinicians, companies, and basic scientists, so that the final design has the best chance of success,” she told Alzforum.

New and ongoing stem cell trials may avoid some of the pitfalls that tripped up StemCells Inc., Svendsen believes. For his upcoming Phase 1 trial of neural stem cells for ALS, he tested both safety and efficacy of the final product in mice. In addition, his work and that of Neuralstem Inc., Rockville, Maryland, uses cells expanded from a single fetal source, lowering variability. Svendsen and colleagues maintain a bank of frozen cells ready to grow up for future trials. Neuralstem recently ran a Phase 1 study of neural stem cells for spinal cord injury. An ongoing Phase 2 trial of mesenchymal stem cells for AD, sponsored by Stemedica Cell Technologies, San Diego, also uses cells from a single source, the company’s Alexei Lukashev told Alzforum. Stemedica’s clinical-grade cells are cultured in the same way as research-grade, he added.

Currently, only a handful of trials are testing stem cells for spinal cord injury or AD. Researchers worry that the failure of StemCells Inc. will discourage investment in this area. Rao noted that some companies have obtained FDA approval for trials but have been unable to raise enough funding to run them.

“Neural and other stem cell populations still have tremendous potential for the field. Stem cell trials are in early days—we have to do enough of them to learn how to do them well,” Anderson noted.—Madolyn Bowman Rogers


No Available Comments

Make a Comment

To make a comment you must login or register.


News Citations

  1. Neural Stem Cells Strum the Right Cord
  2. Bon Appétit: Endogenous Antibodies Prod Microglia to Eat Aβ Deposits

Research Models Citations

  1. 3xTg
  2. 5xFAD (B6SJL)

Paper Citations

  1. . Human neural stem cells improve cognition and promote synaptic growth in two complementary transgenic models of Alzheimer's disease and neuronal loss. Hippocampus. 2014 Dec 19; PubMed.
  2. . The mesenchymal stromal cells dilemma--does a negative phase III trial of random donor mesenchymal stromal cells in steroid-resistant graft-versus-host disease represent a death knell or a bump in the road?. Cytotherapy. 2013 Jan;15(1):2-8. PubMed.
  3. . Challenges in animal modelling of mesenchymal stromal cell therapy for inflammatory bowel disease. World J Gastroenterol. 2015 Apr 28;21(16):4779-87. PubMed.

External Citations

  1. Phase 2 trial
  2. Phase 1/2 study
  3. press release
  4. Nasdaq notice
  5. Food and Drug Administration guidance
  6. Phase 1 trial
  7. Phase 1 study
  8. Phase 2 trial

Further Reading

Primary Papers

  1. . Preclinical Efficacy Failure of Human CNS-Derived Stem Cells for Use in the Pathway Study of Cervical Spinal Cord Injury. Stem Cell Reports. 2017 Feb 14;8(2):249-263. PubMed.
  2. . HuCNS-SC Human NSCs Fail to Differentiate, Form Ectopic Clusters, and Provide No Cognitive Benefits in a Transgenic Model of Alzheimer's Disease. Stem Cell Reports. 2017 Feb 14;8(2):235-248. PubMed.
  3. . Reaction from StemCells, Inc. to Two Papers in Stem Cell Reports on the Efficacy of Human NSCs in Mouse Models of Alzheimer's Disease and Spinal Cord Injury. Stem Cell Reports. 2017 Feb 14;8(2):194-195. PubMed.
  4. . Response to StemCells Inc. Stem Cell Reports. 2017 Feb 14;8(2):195-197. PubMed.