Prevalence of Propionibacterium acnes in Intervertebral Discs of Patients Undergoing Lumbar Microdiscectomy: A Prospective Cross-Sectional Study

Background The relationship between intervertebral disc degeneration and chronic infection by Propionibacterium acnes is controversial with contradictory evidence available in the literature. Previous studies investigating these relationships were under-powered and fraught with methodical differences; moreover, they have not taken into consideration P. acnes’ ability to form biofilms or attempted to quantitate the bioburden with regard to determining bacterial counts/genome equivalents as criteria to differentiate true infection from contamination. The aim of this prospective cross-sectional study was to determine the prevalence of P. acnes in patients undergoing lumbar disc microdiscectomy. Methods and Findings The sample consisted of 290 adult patients undergoing lumbar microdiscectomy for symptomatic lumbar disc herniation. An intraoperative biopsy and pre-operative clinical data were taken in all cases. One biopsy fragment was homogenized and used for quantitative anaerobic culture and a second was frozen and used for real-time PCR-based quantification of P. acnes genomes. P. acnes was identified in 115 cases (40%), coagulase-negative staphylococci in 31 cases (11%) and alpha-hemolytic streptococci in 8 cases (3%). P. acnes counts ranged from 100 to 9000 CFU/ml with a median of 400 CFU/ml. The prevalence of intervertebral discs with abundant P. acnes (≥ 1x103 CFU/ml) was 11% (39 cases). There was significant correlation between the bacterial counts obtained by culture and the number of P. acnes genomes detected by real-time PCR (r = 0.4363, p<0.0001). Conclusions In a large series of patients, the prevalence of discs with abundant P. acnes was 11%. We believe, disc tissue homogenization releases P. acnes from the biofilm so that they can then potentially be cultured, reducing the rate of false-negative cultures. Further, quantification study revealing significant bioburden based on both culture and real-time PCR minimize the likelihood that observed findings are due to contamination and supports the hypothesis P. acnes acts as a pathogen in these cases of degenerative disc disease.

aim of this prospective cross-sectional study was to determine the prevalence of P. acnes in patients undergoing lumbar disc microdiscectomy.

Methods and Findings
The sample consisted of 290 adult patients undergoing lumbar microdiscectomy for symptomatic lumbar disc herniation. An intraoperative biopsy and pre-operative clinical data were taken in all cases. One biopsy fragment was homogenized and used for quantitative anaerobic culture and a second was frozen and used for real-time PCR-based quantification of P. acnes genomes. P. acnes was identified in 115 cases (40%), coagulase-negative staphylococci in 31 cases (11%) and alpha-hemolytic streptococci in 8 cases (3%). P. acnes counts ranged from 100 to 9000 CFU/ml with a median of 400 CFU/ml. The prevalence of intervertebral discs with abundant P. acnes ( 1x10 3 CFU/ml) was 11% (39 cases). There was significant correlation between the bacterial counts obtained by culture and the number of P. acnes genomes detected by real-time PCR (r = 0.4363, p<0.0001).
The pioneering study published in 2001 by Stirling et al. in Lancet, reported that 53% of the patients with severe sciatica had Gram-positive anaerobic microorganisms in their disc tissue of which 84% were P. acnes [12]. Considering the potential clinical consequences of a P. acnes infection as an etiological factor in degenerative disc disease, these findings are indicative of potentially significant similarities to the discovery of Helicobacter pylori involvement in pathogenesis of peptic ulcers and the shift it led to in the way peptic ulcers are treated [13]. Subsequently, several independent studies have demonstrated the presence of infected extruded disc tissue from first-time disc herniations, with P. acnes being the most commonly involved organism [14][15][16][17][18]. According to a meta-analysis of these studies from 2015, P. acnes was the most prevalent bacteria in disc tissue, with the pooled estimate of the proportion with positive samples being 34% [19]. More recently, the first experimental evidence was provided showing that P. acnes has the ability to induce disc degeneration in a rabbit model [20]. Other studies, however, reported very low or zero prevalence of bacterial infection in samples from patients with lumbar disc herniation, quoting contamination as the predominant source [21][22][23].
Contributing to this lack of consensus, significant methodical differences exist among studies. The most notable of which is the lack of laboratory procedures that take into account a possible biofilm mode of growth associated with P. acnes infection. This translates into inconsistencies in regard to tissue homogenization procedures to break up biofilm aggregates, leading to negative culture results. As a consequence, P. acnes positivity was likely underestimated in some of the previous studies [21,23]. None of the studies implemented quantitative culture or real-time PCR-based quantification of P. acnes genomes. We believe that quantitation of the bacterial bioburden is an important criterion that can contribute to discrimination between truly infected and skin-contaminated cases. Thus, a lack of quantitative methods to identify P. acnes could have led to overestimating the positivity rates in previous studies. These methodical weaknesses together with the small sample sizes (only one study enrolled more than 100 patients for bacterial culture [23]) have left these studies underpowered, with limited applicability of their results towards clinical decision-making or any definitive conclusions.
In this prospective cross-sectional study on the currently largest series of patients undergoing microdiscectomy for lumbar disc herniation, we implemented disc tissue homogenization, quantitative microbiologic culture, real-time PCR-based detection of P. acnes genomes, and procedural DNA contamination controls to determine P. acnes counts in disc tissue samples and evaluated the prevalence of intervertebral discs with an abundance of P. acnes.

Study design and patients
Between May 2015 and December 2015, adults scheduled for microdiscectomy for symptomatic lumbar disc herniation at University Hospital Brno (Brno, Czech Republic) and St. Anne's University Hospital (Brno, Czech Republic) were prospectively screened for potential participation in this cross-sectional study. Inclusion criteria included: lumbar or lumbosacral radiculopathy with or without sensory deficits but with either a matching, clinically relevant motor deficit in correlating lumbar or sacral nerve root distributions (see imaging criteria) or with radicular pain (sciatica or femoralgia) that was intractable by conservative means; matching physical examination findings including positive straight leg raise test, dermatomal sensory deficits, myotomal motor deficits and/or a diminished deep tendon reflexes; current magnetic resonance imaging or computed tomography imaging of the lumbosacral spine showing a free nucleus pulposus sequestration or a disc herniation / protrusion in a distribution correlating with the clinically affected nerve roots and with the physical examination. Exclusion criteria included: coexistent infection or immunologically compromised conditions; corticosteroid or antibiotics use in the month before surgery; trauma; unknown radiographic mass; diagnosis of inflammatory arthritis or other rheumatologic diseases. The following epidemiological and clinical data were collected: Gender, age, intervertebral segment involved, type of herniation, prevalence of previous spinal surgeries, prior epidural steroid injections, and development of post-operative discitis. A written informed consent was obtained from each patient. The study was approved by the Institutional Review Boards of University Hospital Brno and St. Anne's University Hospital Brno.

Collection of intraoperative samples
The surgical site was scrubbed with triple preparation of povidone iodine and draped using standard sterile technique. Standard perioperative antibiotics were given before the skin incision in all cases. Cefazolin was the standard antibiotic given in most cases. In penicillin-allergic patients, either vancomycin or clindamycin was administered. The precise location for the skin incision was guided by intraoperative fluoroscopy and a posterior midline approach using sharp dissection and electrocautery was performed. After placement of a self-retaining retractor (Caspar type) and under an operating microscope, ligamentum flavum was resected as required by means of Penfield dissectors and Kerrison rongeurs. The disc herniation was exposed by gentle retraction of the traversing nerve root and then removed together with the remnant loose fragments of the nucleus pulposus in the disc space near the annular defect. All tissue samples were handled in such a way as to minimize their contamination, retained in a closed sterile sample cup, and then passed off to the field for labeling and transport to the Department of Microbiology at St. Anne's University hospital (Brno, Czech Republic), where the disc tissues samples from both centers were further analyzed. The sample sizes were approximately 3x3x5-10x5x5 mm and not measured accurately as we attempted to obtain as much material as the possible from the surgical specimen. Samples were not frozen prior to processing and culture was established within 2 to 4 hours post-surgery.

Microbiological culture
Fresh disc tissue samples were cut into smaller fragments using a sterile, individually packaged, gamma-irradiated scalpel; and, a sterile, gamma-irradiated petri dish. One of these fragments was placed into a 2 mL microcentrifuge DNA-free tube and stored at -80°C until processed for P. acnes DNA analysis. Tissue processing and homogenization was carried out in a sterile pestle and mortar with sterile quartz sand (size particle 0.1-0.5 mm; Penta, Czech Republic) and saline solution in aseptic conditions, in a class 2 biological safety cabinet. The homogenized tissue samples were inoculated onto Wilkins Chalgren Anaerobic Agar (Hi Media Laboratories, India) with 7% of sheep's blood and vitamin K. Inoculated plates were incubated anaerobically (80% nitrogen, 10% CO 2 and 10% H 2 ) in an Anaerobic Work Station (Ruskinn Technology, UK) at 37°C for 14 days and assessed for bacterial growth. The quantity of bacteria in the sample was expressed as colony forming units (CFU) in 1 ml of the homogenate. Identification of bacteria was carried out biochemically using the Rapid ANA II System (Remel, USA).

Quantitative real-time PCR
Frozen tissue samples were thawed (median wet weight was 130 mg, range 20-180 mg), cut into small fragments and transferred to a sterile 2 mL microcentrifuge tube by use of newly opened sterile sets of needle, scalpel and tweezers. Small fragments of the tissue samples were further suspended in 500 μl of ATL buffer (Qiagen, Germany) with 50 μl of proteinase K (20 mg/ml) (Qiagen) and digested at 56°C and 650 rpm in a thermomixer overnight. To each set of the samples that were processed in parallel, we used a tube with sterile water as a laboratory contamination control to follow the entire laboratory process from digestion of the tissue and DNA isolation to real-time PCR analysis. DNA was extracted by use of the QIAamp UCP Pathogen Mini Kit (Qiagen) as described in the manufacturer's instructions. Concentration of DNA were measured spectrophotometrically using a Nanodrop 2000 (Thermofischer, USA); or with fluorescent dye and a Qubit 3.0 fluorometer (Life Technologies, USA) for samples with DNA concentrations less than 5ng/μl. A previously described real-time PCR assay was performed using primers to amplify a 131-bp region of the 16S rRNA gene of P. acnes: forward primer 5'-GCGTGAGTGACGGTAATGGGTA -3', reverse primer 5'-TTCCGACGCGAT CAACCA-3' and TaqMan probe 5'-AGCGTTGTCCGGATTTATTGGGCG-3' [24]. The 15-μl PCR reaction mixture contained 6,75 μl of DNA sample, 5 pmol of each primer and 2 pmol of TaqMan probe, and 1X TaqMan Gene Expression Master Mix (Life Technologies, USA). The QuantStudio 12K Flex system (Life Technologies, USA) was used with the thermal cycling profile of 50°C for 2 min, 95°C for 10 min and 50 cycles of 95°C for 15 s and 60°C for 1 min. The P. acnes genome equivalents in samples were estimated with an internal standard curve prepared with five replicates of six concentrations (10-10 6 copies) of synthesized P. acnes amplicon (131 bp) (Integrated DNA Technologies, USA). Laboratory contamination controls described above and PCR negative controls were included in every PCR reaction. Assays were done in duplicate for each sample, and the mean number of the 16S rRNA gene copies was calculated. To eliminate laboratory contamination, 16S rRNA counts detected in laboratory contamination control were subtracted from the copies number in the tissue samples. The number of bacterial genomes in each sample was finally calculated using the known number of copies of the 16S rRNA operon (3 copies/cell) in P. acnes [4] and represented as the number of bacterial genomes in 500 ng of total DNA extracted from the disc tissue sample. Human ß-globin gene was included as an internal control to allow assessment of the specimen quality and the nucleic acid extraction as well as the inhibition amplification process as described previously [24].

Data analysis
A description of identified bacterial frequencies was performed and the relationship between presence/counts of P. acnes and clinical parameters was evaluated. Categorical variables were analyzed using a two-sided Fisher's exact test, continuous variables (genome counts, age) with the non-parametric Mann-Whitney U test. Spearman's correlation was used to test the association between bacterial counts reached in the same samples by different methods. The significance level was set at 0.05. The statistical software used was Prism 5 (GraphPad Software Inc., USA).

Patient characteristics
We prospectively enrolled 290 patients (171 males and 119 females) with an average age of 47 ± 13 years. One hundred ninety-two cases were enrolled at University Hospital Brno, ninety eight cases at St. Anne's University Hospital. The degenerative intervertebral disc segments were L2/L3 in 8 (3%), L3/L4 in 21 (7%), L4/L5 in 137 (47%) and L5/S1 in 124 (43%) cases. For 8 patients who underwent a double-level disk surgery, only the caudal disk was selected to avoid a statistical bias. Disc herniation types included disc protrusion in 23 (8%), disc extrusion in 126 (44%), and disc sequestration in 141 (48%) patients. Medical history of previous spinal surgery was noted in 51 (17%) cases, prior epidural steroid injection was received by 14 (5%) patients. None of the patients developed clinically evident post-operative discitis. There were no significant differences in patient characteristics between the two recruitment centers regarding age, gender and disease-specific clinical data which were collected and used in our study.

Results of bacterial culture
Bacteria were identified in 130 of 290 disc tissues samples, the other 160 biopsies resulted in no bacterial colonies. The following bacteria were isolated: P. acnes in 115 cases (40%), coagulasenegative staphylococci (CoNS) in 31 cases (11%) and alpha-hemolytic streptococci (AHS) in 8 cases (3%) (Fig 1A). Twenty-four patients (8%) had two microorganisms present in the disc. No patients had more than two types of bacteria isolated. We observed significant variability in P. acnes counts among positive disc samples ranging from 100 to 9000 CFU/ml (median 400 CFU/ml).
A threshold was defined as 1x10 3 CFU/ml (75 th percentile) for the discs with abundant P. acnes. Prevalence of the disc samples with abundant P. acnes was 11% (39 cases); in the remaining 29% (76 cases) of positive disc samples P. acnes was not abundant (< 1x10 3 CFU/ml).
Propionibacterium acnes genome counts in disc tissue samples DNA was successfully extracted from all samples included in the study with concentration ranging from 2 to 45 ng/μl (median 13 ng/μl). The mean Ct (threshold cycle) value for the internal control, human ß-globin gene, was 24.03 ± 1.81. Eighteen cases of the 290 cases were excluded from statistical evaluation due to negative of ß-globin gene (Ct > 35). The mean P. acnes genomes count in the laboratory contamination control samples was 4 ± 4 genomes. P. acnes genomes were undetectable in only 13 (4%) cases. The number of P. acnes genomes in the 259 P. acnes-positive disc tissues ranged from 2 to 5831 with a median of 260 genomes per 500 ng of total DNA. There were significant differences in the number of P. acnes genomes detected in the samples with abundant P. acnes by culture (1039 ± 244 genomes), when compared to samples with non-abundant P. acnes (448 ± 98 genomes) and samples negative by culture (415 ± 57 genomes) (for both P < 0.0001, Fig 1C). We also found significant Pearson correlation between CFU/ml and P. acnes genome counts detected by real-time PCR in patients positive by culture (r = 0.4363, p<0.0001). There was no difference in the number of P. acnes genomes between disc tissue samples with non-abundant P. acnes and samples negative by culture (p = 0.6973).

Relationships of Propionibacterium acnes and patient characteristics
The average age of patients having discs with abundant P. acnes by culture was significantly lower (42 versus 48 years; p = 0.0086). There were no significant differences when comparing the groups of patients (i.e., abundant P. acnes, non-abundant P. acnes, or negative cultures) in gender, prevalence of previous spinal surgery, prevalence of prior epidural steroid injection, distribution of herniation type, or affected intervertebral level (summarized in Table 1). With the exception of the age, there were no significant associations found between P. acnes prevalence and clinical characteristics including the situation in which all patients with a positive culture (abundant + non-abundant) were considered positive for statistical evaluation or the P. acnes genome counts were compared between groups (all p > 0.05).

Discussion
Traditionally, mechanical, degenerative (age-related), traumatic and genetics-based theories have been proposed as possible etiologies of degenerative disc disease [25], but evidence has emerged over the last decade showing that low-grade infections, mostly by P. acnes, playing a potential causative role in disc degeneration and resultant disc-herniation and radicular pain [11,19]. However, the potential relationship between intervertebral disc degeneration and chronic infection by low-virulence P. acnes remains controversial, with contradictory evidence in the literature. We suggest that for accurate detection of P. acnes, it is critical to anticipate its existence in its biofilm form [5,26,27] and to use quantitative methods to determine bacteria counts. Unfortunately, this approach has not been utilized in any of the published studies in this field. In the current study, we homogenized disc samples to disrupt biofilm aggregates (which if undisrupted often lead to negative culture results [27]) to eliminate false negativity, and used quantitative anaerobic culture and real-time PCR to determine bacterial counts/ genomes as one of the criteria to eliminate potential contamination and false positivity.
We prospectively enrolled 290 patients undergoing microdiscectomy for lumbar disc herniation, which is the largest series of patients to date, and examined their intraoperative disc biopsies by quantitative culture and real-time PCR. We isolated three bacterial species by culture: P. acnes, CoNS and AHS in 40%, 11% and 3% of cases, respectively. Our data are in agreement with a majority of the previous reports describing P. acnes and CoNS as the most frequently isolated by anaerobic culture [12,[14][15][16][17]. Overall the P. acnes positivity rate observed in our study is similar to the results of Stirling et al. [12], Arndt et al. [16] and Albert et al. [17], who observed 44%, 35% and 40% positivity rate, respectively. Analogically to our results, 5%, 14% and 31% prevalence of CoNS was described in three others studies [12,14,16].
We used standard surgical care protocols, including the use of prophylactic perioperative antibiotics (cefazolin), thus ensuring patient safety while conducting our research. We reasoned that cephalosporin prophylactic antibiotics would not affect the research results because these antibiotics do not penetrate into the intervertebral discs and because P. acnes biofilm within the disc is antibiotic resistant at the typically used doses [28]. In two protocols, prophylactic antibiotics were administered after collection of the biopsies to avoid inhibition of any bacteria in the sample [17], or were not administered at all [22]. In the first case, the positivity rate of P. acnes was identical with our study (40% in both), in the second study P. acnes positivity was only 4%, indicating that prophylactic cefazoline does not have any inhibitory effects on the P. acnes within the disc.
None of the previous studies determined P. acnes counts in the discs and distribution of the bacteria counts within their case series. We described the range of P. acnes counts obtained by culture and observed an 11% prevalence of intervertebral discs with abundant P. acnes ( 1x10 3 CFU/ml). We suggest, that this quantity-adjusted prevalence is less biased by contamination than overall positivity. Similar prevalence of the P. acnes positive disc samples were observed by Agarwal et al. [15], Coscia et al. [14] and Zhou et al. [18] describing 13%, 19% and 24% positivity rate, respectively. As quantitative data from other studies are not available, and differences in the positivity rates differ significantly (from 4 to 44%), it is likely that some authors evaluated weak positivity as contamination and others as positivity. However, there are also studies describing low or zero P. acnes positivity of discs in their case series [21][22][23]. For instance, P. acnes was not isolated from any of 30 disc samples in the study by Ben-Galim et al. [21], who implemented inoculation of cultures directly in the operating theatre to minimize the risk of contamination. Authors of this study suggest that the high prevalence of P. acnes observed in other studies mainly reflects contamination. We speculate that this unusual approach did not provide for sufficient homogenization of the tissue and disruption of P. acnes biofilms, which combined accounts for their negative culture results. Similarly, in the large-scale study on patients with chronic lower back pain, disc fragments were cultured directly without tissue processing or homogenization, and P. acnes were isolated in only 2 of 313 cases [23,29].
We implemented a real-time PCR assay and laboratory contamination controls to accurately quantify P. acnes genomes directly in the disc tissue samples. The numbers of P. acnes genomes in the discs, which were P. acnes abundant by culture, were significantly higher when compared to non-abundant or P. acnes negative discs. There was no difference in the numbers of P. acnes genomes between the discs with non-abundant P. acnes and negative samples by culture, supporting a potential contamination origin of the disc samples with non-abundant P. acnes and accuracy of the empirical threshold ( 1x10 3 CFU/ml) for the samples with abundant P. acnes. Only in 13 (4%) cases, were no P. acnes genomes detected. This is in agreement with outputs of large-scale studies based on next-generation sequencing of P. acnes DNA in a variety of clinical specimens concluding that when highly sensitive molecular methods are employed, P. acnes DNA could be detected in practically all sample types [30]. Moreover, it was shown that P. acnes DNA is a frequent contaminant of commercial Taq polymerases and PCR solutions [31], highlighting the necessity of quantitative information to distinguish between contamination and positivity [32]. Until now, only one team focused on direct detection of P. acnes DNA in disc tissue by qualitative end-point PCR [17]. Others used PCR only for verification of P. acnes DNA in purified colonies [18]. Albert et al. observed consistent results of culture and PCR methods; and, surprisingly, the cases negative by culture (60%) were negative by PCR [17], which could be explained by the decreased number of cycles (only 35) in their assay, probably to artificially decrease the sensitivity of their method.
We did not observe any significant association between P. acnes positivity and prevalence of previous spinal surgery and prior epidural steroid injection indicating endogenous origin of P. acnes and their role in degenerative disc disease more than their iatrogenic origin through disc contamination in the previous spinal surgeries. P. acnes positivity rates were independent of gender, type of herniation, and affected spinal segment. The average age of patients having discs with abundant P. acnes were lower (42 versus 48 years). We hypothesize, that P. acnes infection could accelerate regular mechanical, age-related, degenerative process in affected individuals and result in the necessity of earlier surgical intervention.
Also of note is the two most common commensals, P. acnes and coagulase-negative staphylococci, were also the two most common bacteria found in the disc tissue samples, which reflects the compelling need to gain an ability to discriminate between a contamination and an infection.
This study has several limitations. Aerobic cultures were not performed given disc tissue originates in a notably anaerobic environment. As we consider tissue homogenization a critical step of the experimental protocol, more standardized homogenization methods than mortar and pestle should be implemented. We assume that the biofilm form of P. acnes is typically present within the disc, but do not provide direct evidence of P. acnes biofilms from the disc tissue by microscopic methods. However, in the field of orthopedic infections, there is strong evidence of P. acnes forming a biofilm [26,27]. As disc tissue from age-matched healthy individuals is not available, the identification of P. acnes in biopsies from comparable healthy individuals cannot be ruled out. Further, the lack of control cultures cannot exclude the possibility of contamination from the laboratory, operating room, surgical field, or skin. With exception of one [22], most of the previous studies that included negative controls proved that they are largely sterile [12,14,18]. Since positive endogenous controls do not necessarily connote contamination of the disc sample, we do not believe that endogenous controls are the optimal solution to identify contaminated disc samples. In our study, P. acnes counts and the number of P. acnes genomes were implemented as one of the criteria to eliminate potential contamination. The final answer with regard to distinguishing endogenous P. acnes infection of the disc from skin contamination will probably utilize comparative genomics to differentiate P. acnes genomes from the skin and disc tissue from a given individual [33,34].
In conclusion, to our knowledge this is the first study that provides quantitative information about P. acnes in disc tissue samples in the largest series of patients with lumbar disc herniation-we demonstrated an 11% prevalence of the disc herniation samples with an abundance of P. acnes. Because "finding abundant microorganism in organisms suffering from the disease" is one of the most important of Koch's postulates, in judging the relationship between a specific microbe and a specific disease, we believe that our findings support the theory that P. acnes is involved in the pathogenesis of at least a subset of patients with degenerative disc disease. Future studies are needed to evaluate relationship between P. acnes infection and clinical outcomes of these patients.