A Phase I Dose-Escalation Clinical Trial to Assess the Safety and Efficacy of Umbilical Cord-Derived Mesenchymal Stromal Cells in Knee Osteoarthritis

Major Points and Findings:

This study from Santiago, Chile tested umbilical cord-derived mesenchymal stromal cells (UC-MSCs) at three doses in knee osteoarthritis, on the assumption that a larger number of cells might work better. It did not turn out that way. The highest dose caused so much pain and swelling after injection that the investigators stopped recruiting to it. The two lower doses were better tolerated and were the ones that showed improvement at six months.

The paper has two parts, a preclinical mouse experiment and a 40-patient phase I trial. In mice the higher dose protected cartilage better. In patients it did not. The human trial had no placebo group, so it shows which dose was better tolerated and cannot show whether the treatment works.

Aim:

To find out whether UC-MSCs (a product called Cellistem, made by the university spin-off Cells for Cells) show a dose-response effect in osteoarthritis, first in mice and then in patients, with safety as the primary endpoint of the clinical trial.

Methods:

Mouse study: Osteoarthritis was induced in mice by injecting collagenase into the knee. Groups of 10 mice then received 200,000 cells (“high”) or 50,000 cells (“low”) into the joint on days 7 and 14. On day 42 the joints were examined by micro-CT and by histology using the OARSI cartilage score.

Clinical trial: The human study was a phase I dose-escalation trial at the University of Los Andes Clinical Center, recruiting from March to May 2019. Patients entered three sequential dose cohorts with no placebo group. The injecting surgeon and the outcome assessors were blinded to the dose.

Sixty people were screened and 40 enrolled. To be included, patients had to be aged 30 to 75, have had knee pain every day for at least 3 months, report pain of 40 mm or more on a 100 mm visual analogue scale, and have Kellgren-Lawrence grade 1 to 3 on X-ray. The exclusion criteria were:

• Meniscal rupture
• Symptomatic osteoarthritis in both knees
• Valgus deformity of more than 10 degrees or varus of more than 5 degrees
• Disease of the hip or spine, infection, secondary arthritis, previous malignancy
• Steroid or hyaluronic acid injection into that knee in the past 6 months

Those enrolled were on average 53 to 58 years old and about half were women. Roughly two thirds had grade II disease and one third grade III.

The cells were allogeneic UC-MSCs used at passage 5. Release required viability above 80%, standard MSC surface markers, sterility, endotoxin at or below 0.5 EU/mL and no clumps. The batch was chosen from three donors for the highest secretion of thrombospondin-2, a cartilage-protective factor. Each patient had one injection of 3 mL (saline with 5% AB plasma) in a masked syringe:

• Low dose: 2 million cells (n = 16)
• Medium dose: 20 million cells (n = 16)
• High dose: 80 million cells (n = 8; recruitment stopped early after an interim analysis showed more adverse events)

Patients were told to avoid physical activity for 5 days, and painkiller use was recorded for the first week. Follow-up was at 1, 4, 12 and 24 weeks with WOMAC and pain VAS, plus knee MRI scored with WORMS by two blinded radiologists at baseline and 6 months.

Results:

In the mice, both doses improved bone mineral density compared with untreated osteoarthritic mice. On histology only the high dose helped. The mean cartilage damage score was 4.5 against 12.5 in untreated mice at the medial tibia and 12.5 against 23 at the lateral tibia. The low dose was no different from no treatment.

Safety: No serious adverse events, joint infections, hospital admissions, disability or tumours occurred in the patients, and MRI at 6 months showed no safety signal. Side effects of the injection rose in step with the dose:

• Clinically significant pain (above 40 mm and lasting more than 72 hours): 6 of 16 (37.5%) at low dose, 11 of 16 (68.7%) at medium dose, 8 of 8 (100%) at high dose
• Synovitis or joint effusion lasting about a week: 0, 1 (6.2%) and 3 of 8 (37.5%)
• Fever: 1 patient, in the low-dose group
• Painkiller use in the first week: 31% at low dose against more than 80% in the other two groups

Efficacy: At 6 months the low and medium dose groups had significantly lower WOMAC pain, function and total scores and lower VAS pain than at baseline. There were no significant differences between the three dose groups at the end of follow-up. The paper shows these results only as graphs and gives no numerical follow-up scores, so the size of the improvement cannot be quoted here. Baseline total WOMAC was 36.9 (low), 26.4 (medium) and 38.6 (high).

MRI showed no significant change in cartilage or any other feature. Mean WORMS went from 47.8 to 49.1 (low), 39.4 to 46.8 (medium) and 44.3 to 41.8 (high), with P values of 0.84 to 0.95.

The published abstract and the full text differ in places. The abstract says 100% of high-dose patients had “injection-related swelling”. In the full text and Table 2, 100% had clinically significant pain and 37.5% had synovitis or effusion. The abstract says all doses gave significant improvement, whereas the results section reports this for the low and medium doses. The abstract and the methods also give different ClinicalTrials.gov numbers (NCT03810521 and NCT02580695). This summary follows the full text.

Conclusions:

The authors conclude that intra-articular Cellistem is safe, that choosing the dose is critical, and that the low and medium doses performed best. They propose that a high cell number may itself provoke inflammation in the joint.

The authors acknowledge their limitations. There were few patients per group, so the low-dose result needs confirming in a larger trial. There was also no control group, which the authors say is “critical” to confirm that the symptom improvement was due to the cells. With only before-and-after comparisons over 6 months, the efficacy findings should be treated as preliminary.

Several authors receive stipends from Cells for Cells, which makes the product. One is its Chief Scientific Officer and holds related patents.

Background Information:

The mouse results did not predict the results in patients. The authors write that the dose-response seen in mice “did not correlate with the observed outcomes in patients”, and use this to argue that dose-finding has to be done in humans.

Other trials cited in the paper point the same way. In a study by Günay and colleagues, 3 patients developed mild effusions after 100 million UC-MSCs. In an adipose MSC trial by Pers and colleagues, the lowest dose gave the clearest improvement in pain and function.

The same group’s previous randomized phase I/II study used 20 million cells, the medium dose here, and reported no severe adverse events and improved pain and function at one year.