Percutaneous autologous bone marrow concentrate for knee osteoarthritis: patient-reported outcomes and progenitor cell content

Utilizing ultrasound or fluoroscopic guidance, bone marrow was aspirated from the posterior superior iliac spine into heparinized syringes using a small-volume, multi-site technique [16, 17]. The autologous bone marrow aspirate (BMA) (60 to 120 mL in total) was manually processed into BMC (1.5 to 6.2 mL) by trained laboratory personnel, as previously detailed [18]. A component of the processing quality assurance (QA) protocol requires a small fraction of BMC (< 0.2 mL) to be set aside for cellular analysis and subsequent cryopreservation. An automated haematology analyzer (ABX Micros 60, Horiba Medical, France) was used to obtain a complete blood count for each BMC sample, and the white blood cell parameter was used as a representative measurement for the concentration of total nucleated cells [19]. The remaining portion of reserved BMC was cryopreserved at a concentration of ten million cells per mL using cryopreservation medium containing 30% fetal bovine serum (FBS) and 5% dimethyl sulfoxide (DMSO) by volume and a controlled rate freezing process [20].

All patients were encouraged to cease the use of non-steroidal anti-inflammatory drugs two weeks prior to and several weeks following BMC treatment. Under sterile conditions, patients received an intra-articular or intra-articular plus intraosseous injection(s) of BMC into the affected joint space via imaging guidance. Additionally, based on the patient’s clinical presentation and imaging findings, they may have received further injections into the supporting structures (i.e., ligaments, tendons, meniscus) if these structures were also diseased, damaged, or injured. Further, intra-articular injections of prolotherapy and concentrated leukocyte poor PRP were performed two to four days prior to (prolotherapy) and following (PRP) BMC treatment [21]. Patients with suspected ligamentous instability based on clinical indications were fitted for a hinged unilateral unloader knee brace or a patellar stabilizer brace (Breg, Inc., Carlsbad, CA, USA) and instructed to wear the brace during weight bearing activity for four weeks. If patients experienced substantial post-procedural pain, opioid rescue medication was prescribed for up to five days. Patients were advised to avoid activities that caused a worsening of pain throughout their stepwise rehabilitation protocol, which began with rest and household/community ambulation. Progression of physical activities consisted of pool or low impact exercise, followed by walking, resistance training and jogging, and ultimately advancing to full functional activity. While face-to-face physical therapy sessions were encouraged, they were not required.

Cryopreserved BMC samples were removed from cryogenic storage and rapidly warmed at 37°C for two minutes. Thawed BMC samples were promptly diluted tenfold in pre-warmed complete culture medium (CCM) containing 10% FBS and 1 ng per mL human fibroblast growth factor, counted using an automated cell counter (TC20, BioRad, Hercules, CA, USA) and trypan blue, and directly plated within standard six well tissue culture plates at two separate cell densities, 10×10³ and 30×10³ viable nucleated cells per cm². Following 72 hours of culture, non-adherent cells were removed by washing, and the remaining adherent cells were maintained at 37°C and 5% CO₂ with biweekly replacement of CCM.

After a 14-day culture period, the plates were washed and stained for colonies using crystal violet in methanol. A CFU-F was defined as any colony greater than 1 mm in diameter and containing a minimum of 100 cells. All colonies were counted by two independent observers. To determine CFU-F frequency, colony counts were averaged across wells and divided by the number of plated cells per well, and the CFU-F concentration was calculated by multiplying the CFU-F frequency of the cryopreserved sample by the nucleated cell concentration, as measured prior to cryopreservation [20].

Patient-reported outcomes were compared over time by fitting a mixed-effects model with Geisser–Greenhouse correction and Tukey’s multiple comparison tests. Patients reporting overall favorable outcomes (responders) to autologous BMC therapy were characterized as meeting or exceeding both the minimal clinically important difference (MCID) of nine points for LEFS and 40% improvement for SANE at the 6-month follow-up (the 12-month follow-up was used for three patients not reporting 6-month functional outcomes), based on previous studies [13, 22, 23]. Patient demographics and BMC cellularity were compared between responders and non-responders using unpaired t tests with Welch’s correction. A receiver operating characteristics (ROC) curve analysis was used to determine the concentration of CFU-F for best differentiating responders from non-responders [24]. Reported mean outcomes of knees treated both above and below the identified target CFU-F concentration were compared against the LEFS MCID and SANE > 40% thresholds using one-sample t tests. Results were considered significant at P < 0.05. All statistical analyses were performed using GraphPad Prism 9 (GraphPad Software, La Jolla, CA, USA).

Sixty-five patients (29 females and 36 males) aged 62.2 ± 8.2 years with moderate to severe knee OA (16.7% II, 47.2% III, 36.1% IV) were treated non-surgically using image guided, percutaneous injections of autologous BMC (Fig. 1). Seven patients were injected bilaterally for a total of 72 treated knees. Depending on the presence of bone marrow lesions on diagnostic imaging, a subset of patients (27 of 72 knees) received intra-articular as well as intraosseous BMC injections, as determined by the treating physician. An average volume of 2.9 ± 1.3 mL of BMC, containing 440 ± 155 million nucleated cells per mL, was injected into the index knee(s). The CFU-F frequency of cryopreserved BMC samples ranged from 4 to 105 CFU-Fs per million nucleated cells with an average of 43.3 ± 23.9 (0.00433% ± 0.00239%). After accounting for the range of nucleated cell concentrations measured within the BMC samples, the average concentration of colony-forming cells was 18.6×10³ ± 11.2×10³ CFU-F per mL of BMC, ranging from 925 to 57.1×10³. A summary of patient demographics, BMC cellularity, and reported outcomes is presented in Table 1.

Patients treated with autologous BMC reported a considerable and sustained reduction in pain, gain in function, and overall percent perceived improvement (Fig. 2). There were no differences in reported outcomes (P > 0.05) between knees receiving intra-articular injections and knees receiving intra-articular plus intraosseous injections (Supplementary Fig. 1). Consequently, all outcomes were analyzed as a single cohort. Baseline pain levels (NPS) decreased from 4.1 ± 2.1 to 2.1 ± 2.0 and 2.0 ± 1.8 at the six month and 12-month follow-ups, respectively (Fig. 2A), while the knee functional scores (LEFS) increased from 45.4 ± 14.0 to 59.4 ± 14.3 and 59.1 ± 14.3 over the same period (Fig. 2B). Similarly, reported percent improvement (SANE) increased from 34.8% ± 30.9% at one month to 55.4% ± 34.4% at 12 months (Fig. 2C). Patient-reported outcomes (LEFS, NPS, SANE) were significantly improved (P < 0.001) over baseline (or 1 month for SANE) at all subsequent follow-ups.

While baseline and six month SANE data were collected from every patient, follow-up rates varied at other time points, ranging from 96% (69 of 72 knees) at the one month follow-up to 85% (61 of 72 knees) at the 12-month follow-up. A majority of patients (44 or 72 knees) reported in engaging in some form of post-treatment physical therapy. Moreover, ten patients reported receiving an additional injection, in the form of PRP (6), prolotherapy (2), concentrated serum proteins (1), and Hoffa’s fat pad hydrodissection (1), within the 12-month period following BMC treatment.

A majority of patients (66%, 43 of 65) reported meeting or exceeding the LEFS MCID (9 points) and SANE threshold (> 40%) with respect to one or more treated knees at the six month follow-up and thus were considered responders. However, a subset of knees (36%, 26 of 72) failed to respond to autologous BMC treatment, and approximately one-third of the non-responding knees (9 of 26) reported deteriorating function (ΔLEFS ≤ 0) and no improvement (SANE ≤ 0) at the six month follow-up. Injectate cellularity, patient age, and BMI were compared between responding and non-responding patient cohorts (Fig. 3). The mean CFU-F concentration, frequency, and nucleated cell concentration were significantly greater (P = 0.004, 0.030, 0.046) within the BMC of responding knees (Fig. 3A–C), yet no differences (P > 0.05) were observed between responding knees and non-responding knees with respect to BMC volume, biological age, or BMI (Fig. 3D–F). Further, no differences in average knee OA severity grades were observed between responding and non-responding knees (P > 0.05). High interpatient variability in CFU-F size, density, and frequency were observed in both groups (Fig. 3G–H). A summary of patient demographics, BMC cellularity, and reported outcomes from responders and non-responders is presented in Table 2.

Patient-reported outcomes were compared against the CFU-F concentrations of their respective autologous BMC treatments (Fig. 4). Scatterplots of CFU-F concentration versus percent improvement and change in LEFS at both 6- and 12-month follow-up time points indicate patients having higher concentrations of colony-forming cells generally report improved outcomes (Fig. 4A–B). The receiver operating characteristic (ROC) curve (AUC = 0.677, P = 0.013) revealed a maximum sensitivity and specificity of 63.0% and 76.9%, respectively, at a CFU-F concentration of 18×10³ CFU-F per mL of BMC (Supplementary Fig. 2). When separated into two cohorts based on this CFU-F concentration cutoff, knees treated with autologous BMC having more than 18×10³ CFU-F per mL reported outcomes, on average, that were significantly greater than the MCID for LEFS (6-month P = 0.004, 12-month P = 0.025) and SANE (6-month P < 0.001, 12-month P = 0.005) threshold (Fig. 4C–D). In contrast, patients treated with BMC having fewer than 18×10³ CFU-F per mL reported average outcomes that failed to exceed the LEFS MCID and SANE threshold.