This study was a prospective clinical feasibility study, primarily a safety assessment, conducted between April 2009 and March 2012. This study was approved by the Ethics Committee of Mie University Hospital. All participants provided informed consent.

Patients who were receiving diagnostic discography for suspected discogenic low back pain from April 2009 to March 2012 were recruited. Among 27 patients who received diagnostic discography, 14 were included in this study. Inclusion criteria for this study were being older than 18 years and having (1) chronic low back pain without leg pain for more than 3 months, (2) one or more lumbar discs (L3/L4 to L5/S1) with evidence of degenerative changes as per MRI (disc degeneration was defined as more than grade III via the Pfirrmann disc degeneration grade/classification system [15]), (3) maintenance of 50% or more of normal disc height, and (4) at least one symptomatic disc confirmed using standardized provocative discography and/or disc block. Exclusion criteria included abnormal neurological symptoms (e.g., radiculopathy) with lumbar spinal stenosis or spondylolisthesis and inflammatory arthritis (e.g., discitis).

Provocative discography and subsequent disc block were performed to diagnose discogenic low back pain as previously reported [16], with some modifications. Spinal needles (22 gauge, 150 mm; Hakko, Chikuma, Japan) were inserted into the center of the NP under fluoroscopy. Radiocontrast agent (Ominipaque 240, Daiichi-Sankyo, Tokyo, Japan) was injected into the disc until pain was provoked (maximum 2.0 mL). After concordant pain was evaluated, 1.0 mL of 2% lidocaine (AstraZeneca, Osaka, Japan) was injected. Two and 24 hours later, the extent of pain reduction was assessed. The patients with concordant pain at discography and/or improved back pain after the disc block were diagnosed with discogenic low back pain.

PRP isolation was performed using a sterile blood collection bag system (KARMI Blood Bag, Kawasumi Laboratories, Inc., Tokyo, Japan) at the Blood Transfusion Service of Mie University Hospital. Whole blood (200 mL) with an anti-coagulant was first centrifuged at 3,000 ×g for 15 minutes at room temperature to form a buffy coat (BC) layer containing platelets and leukocytes. The BC layer was resuspended in 20 mL of plasma and then centrifuged at 180 ×g for 15 minutes to separate the platelets from leukocytes and residual red blood cells. The resulting supernatant (PRP) was transferred to a storage bag (Fig. 1A). Autologous serum was prepared from 10 mL of coagulated whole blood by centrifuging for 10 minutes at 3,000 ×g. A mixture of autologous serum and 2% CaCl2 (Otsuka, Tokyo, Japan) was added to the PRP for clot (gel) formation. After incubation for 60 minutes at room temperature, the supernatant (PRP releasate) was isolated from the PRP gel by centrifugation (3,000 ×g for 5 minutes) (Fig. 1B). The samples were kept at −20℃ until used. The number of platelets and white blood cells (WBCs) of each whole blood and PRP fraction was counted using a hemocytometer. The levels of platelet-derived growth factor BB (PDGF-BB) in the PRP releasate and autologous serum were assessed by enzyme-linked immunosorbent assay kits (R&D systems, Minneapolis, MN, USA).

Intravenous antibiotics were administrated within 60 minutes before the injection procedure. The injection site was treated with local anesthetic (0.5% lidocaine). Under fluoroscopy, a 22 gauge, 150-mm spinal needle was inserted into the center of the targeted disc (Fig. 2). PRP releasate, 2 mL, was injected through a syringe filer (Millex GV Filter unit, Cat. #SLGV M33 RS, Millipore, Billerica, MA, USA).

After injecting PRP releasate, patients were hospitalized for 12 hours and were discharged a day later. Returning to normal, daily living activities was permitted three days after discharge. Physical activity, such as sports, was permitted four weeks after treatment according to the extent of pain reduction. To more accurately evaluate the preliminary efficacy of this treatment on pain, only the temporary use of nonsteroidal anti-inflammatory drugs (NSAIDs) was permitted for unbearable pain.

The efficacy of this treatment was assessed by a visual analog scale (VAS) [17] for back pain and the Roland-Morris Disability Questionnaire (RDQ) [18] for back pain-related disability at baseline and at 4, 8, 16, 24, 32, 40, and 48 weeks after the treatment. A neurological assessment, including motor strength, sensory function, and reflexes, was also performed at the same time points.

Lateral lumbar spine radiographs of each patient were taken centered on the L3 vertebrae in the standing position before and after treatment as well as every second month until the end of the study. The anterior and posterior heights and the depths of the intervertebral discs were measured, and disc height index (DHI) was calculated as previously reported [19]. The % DHI was calculated as the rate of change in DHI compared to the baseline [(DHI at follow-up–DHI at baseline)/DHI at baseline]×100%. The lumbar lordosis angle (the angle between the planes of the superior end plate of the first lumbar vertebra and the superior line of the sacrum) was also determined.

MRI analysis was performed before treatment and at three to four months (early stage) and 12 months (late stage) after treatment. MRI was performed using a 3.0-Tesla scanner (Achieva 3.0T, Philips, Amsterdam, Netherlands) with a SENSE-Spine coil (Philips). Sagittal T2-weighted (time to repeat, or TR, of 2,500 milliseconds; time to echo, or TE, of 90 milliseconds) fast spin-echo images were used to classify lumbar discs into five grades of degeneration using the Pfirrmann disc degeneration grading scheme [15].

Quantitative T2 mapping was performed using a multiecho spin-echo sequence in the sagittal plane. Scanning parameters were TR=2,500 milliseconds; TE=15 to 300 milliseconds (15 TEs); field of view=280 mm; slice thickness=4 mm; image matrix=512×512; number of excitations=1; and total scanning time=7 minutes and 20 seconds. To automatically segment and analyze regions of the NP usually unapparent in degenerate discs, a template-based method [20] was used. Briefly, MRI data from cadaveric human spines containing grade 1 discs (n=11) with visibly distinct NP regions were analyzed using Otsu's method [21] to threshold and segment NPs. Individual masks of the NP and whole disc were registered (rigid body with scaling) and averaged to create a single template. The template was applied to all discs by affine registration of the template to match the mask of the target discs. The registered template was applied to T2 maps to determine average T2 values of the NP and AF from each disc. To reduce variability between patients, all T2 values were normalized to the values at L3/4, which were usually much less degenerate compared to the injected levels. Among the 14 patients, quantitative T2-mapping analysis was performed in 11 patients who provided consent (early stage, n=4; late stage, n=1; and both early; and late stages, n=6).

The safety of this treatment was evaluated in terms of neurological changes; radiological examination, including changes in disc height; the lumbar lordosis angle; and the MRI T2 value. The presence or absence of adverse events associated with this treatment was also evaluated throughout the follow-up period.

The optimal sample size of this study was estimated to provide at least 80% power to detect the differences of VAS and RDQ between pre- and post-treatment at a 5% level of significance (with 0.8 effect size) using the G*Power 3 program [22]. Differences in VAS and RDQ scores, MRI T2-values, and radiographic L1–S angles were assessed for statistical significance using the paired t test. The significance of differences among mean radiograph measurements was analyzed by a 2-way repeated measures analysis of variance (ANOVA). The data are expressed as the mean±standard deviation of the mean (SD). Statistical analysis was performed using the StatView program (Abacus Concepts, Berkeley, CA, USA) with a significance level of p<0.05.

Data were analyzed from 14 patients (8 men and 6 women; mean age, 33.8 years; age range, 25–46 years). Fourteen discs were studied in this cohort; patient characteristics are summarized in Table 1. The average follow-up period was 10 months. All patients (100%) met the follow-up examination from baseline to 6 months. Ten patients (71.4%) were available for the follow-up examination at 8 and 10 months, and 9 patients (64.3%) were available for the final examination at 12 months. Between 6 and 8 months, four patients were lost to follow-up. Between 10 and 12 months, one patient was lost to follow-up due to relocation. The targeted discs were L4/L5 (in 11 cases) and L5/S1 (in 3 cases). The Pfirrmann disc degeneration grade was 3 in 12 patients and grade 4 in two patients. Thirteen patients received a single intradiscal injection of PRP releasate; however, one patient (#09) received a second injection 12 weeks after the first injection because of persistent low back pain. During the follow-up period, seven patients temporarily used NSAIDs (50%).

The mean platelet count of PRP was approximately 3.7 times greater than that of whole blood (whole blood, [242.7±45.3]×10³ platelets/µL; PRP, [907.1±1,039.3]×10³ platelets/µL). The mean WBC count of PRP was about 1/230 of whole blood (whole blood, [6.27±1.16]×10³ cells/µL; PRP, [0.20±0.31]×10³ cells/µL). The average level of PDGF-BB in the PRP releasate was approximately 2.1 times higher than that in autologous serum (PDGF-BB [ng/µL], autologous serum, 3.39±1.67; PRP, 7.12±3.11; p<0.01).

More than 50% reduction of LBP, as evaluated by VAS scores, was observed in 71% (10/14) of patients within four weeks after PRP-releasate injection; this was generally maintained throughout the observation period. However, LBP returned in two patients (#02 and #07) (Fig. 3A). The mean pain score (VAS) significantly decreased after injection of PRP releasate (baseline, 7.5±1.3 [n=14]; 4 weeks, 3.1±2.5 [n=14]; 8 weeks, 3.2±2.0 [n=14]; 16 weeks, 3.4±1.9 [n=14]; 24 weeks, 3.2±2.4 [n=14]; 32 weeks, 3.0±1.9 [n=10]; 40 weeks, 2.8±2.6 [n=10]; 48 weeks, 2.9±2.8 [n=9]; all p<0.01 vs. baseline) (Fig. 3B).

Improvement in physical disability scores (RDQ scores) was relatively better than that in VAS scores. Particularly, 79% of patients (11/14) showed a significant reduction (more than 50%) in RDQ scores four weeks after PRP-releasate injection. This was maintained for 48 weeks after treatment. However, RDQ scores returned to higher levels in the same two patients whose VAS scores rebounded (patients #02 and #07) (Fig. 4A). The mean RDQ score significantly decreased after injection (baseline, 12.6±4.1 [n=14]; 4 weeks, 5.1±5.2 [n=14]; 8 weeks, 5.2±5.5 [n=14]; 16 weeks, 4.7±4.5 [n=14]; 24 weeks, 3.6±4.5 [n=14]; 32 weeks, 4.0±3.0 [n=10]; 40 weeks, 3.8±5.3 [n=10]; 48 weeks, 2.8±3.9 [n=9]; all p<0.01 vs. baseline) (Fig. 4B).

For all patients, their lumbar lateral radiographs showed no significant progression of disc height narrowing and ossification following the injection of PRP releasate into the targeted disc (Fig. 5). The % DHI of PRP releasate-injected discs and the control (L3/L4) discs had similar changes during the follow-up period (p=0.9) (Fig. 6A). Analysis of variance revealed no significant interaction between disc treatment and time point (p=0.9). In addition, the angle of lumbar lordosis (L1–S angle) did not change significantly during the evaluation period (Fig. 6B).

Sagittal T2 maps suggested no significant changes in the normalized T2 values of the NPs between baseline (Fig. 7A) and the follow-ups (early stage shown) (Fig. 7B). At baseline, the mean normalized T2 value of the NP at the treated level was 0.56±0.1. Little change was seen at the early stage (0.53±0.06), or the late stage (0.56±0.10), using repeated measurement ANOVA (p=0.68) (Fig. 7C). The mean normalized T2 value of the AF showed a similar trend with no significant difference before and after treatment (baseline, 1.01±0.06; early stage, 1.01±0.15; late stage, 0.94±0.09; p=0.35, repeated measurement ANOVA) (Fig. 7D).

Two patients exhibited leg numbness at 1 month (#04) and 6 months (#03) post-treatment; however, this symptom disappeared within 7 days. None of the patients exhibited neurological deterioration or symptoms of discitis. No other adverse events associated with treatment were observed. Two patients (#02 and #07) experienced the return of low back pain, albeit not worse than that at baseline.