1. INTRODUCTION

Vertebroplasty (VP) has gained in popularity as an effective treatment for painful osteoporotic vertebral fracture since its introduction by Galibert in 1987.1 Although the procedure is generally safe and effective, catastrophic complications such as post-VP infection (PVI) occur at a rate of 0.32% to 0.46%.2–4 Surgical management of PVI often requires an anterior approach to remove the cement, anterior fusion, and posterior instrumentation. This aggressive procedure is not well tolerated in aged and frail patients. The mortality rate of PVI is 10% to 33%.2,3 With less than 60 cases reported to date, the potential risk factors for PVI remain elusive.2

In primary vertebral osteomyelitis, hematogenous seeding from distant sites via the arterial route is the predominant pathway for pathogen infection of the vertebral column. Sites of infection that result in hematogenous spread include the respiratory tract, gastrointestinal tract, skin and soft tissue, and, most often, urinary tract infection (UTI).5 Thus, VP is contraindicated in patients with active infection.6 Asymptomatic bacteriuria (ASB), a condition that does not require treatment in most cases,7 is another potential concern when considering VP. While ASB is found to be a risk factor for prosthetic joint infection,8–10 the relationship between ASB and PVI is unknown.

This study aims to investigate the incidence of ASB among patients who underwent VP to treat painful osteoporotic vertebral fractures, identify the risk factors for PVI, and determine whether ASB is a risk factor for PVI.

2. METHODS 2.1. Study design and population

This study was approved by the institutional review board of our hospital (2020-02-013AC). A prospectively compiled database of patients who were admitted as VP candidates for the treatment of painful osteoporotic vertebral fractures between May 2015 and December 2019 was retrospectively analyzed to select the study cohort. Patients were excluded from the study if they had a follow-up of less than 12 months, an indwelling urinary catheter, previous VP, received VP for malignant metastasized vertebral fractures or multiple myeloma, or if VP was not performed. Patients who had a history malignancy and suffered from fragility-related vertebral fracture were not excluded. The patients’ demographic data, clinical characteristics, laboratory data, imaging findings, and biopsy results were collected from the database and medical records. The primary outcome was PVI at index level during follow-up.

2.2. Treatment protocol

All patients receiving conservative treatment for at least 6 weeks and still experiencing refractory back pain were considered as candidates for VP at our institute. The diagnosis of primary vertebral fracture was based on the following criteria: (1) magnetic resonance imaging (MRI) showing vertebral marrow edema in a T2-weighted image or STIR views; (2) air accumulation, vacuum phenomenon, or pseudarthrosis observed in the CT or MRI scan.

If CT or MRI scans were still ambiguous, a CT-guided biopsy was performed by radiologists to exclude the possibility of primary vertebral osteomyelitis.11 Biopsy samples were sent for culture of aerobic and anaerobic bacteria, tuberculosis, fungi, and permanent pathology sections. The diagnosis of vertebral osteomyelitis was always taken into careful consideration in case of any clinically suspicious observations before the VP procedure.

Urinalysis and urine culture were performed for all patients 1 week before admission. The urine was cultured using conventional methods in the microbiology laboratory, and all isolated microorganisms were identified using standard biochemical protocols. Urinalysis samples with more than 8 squamous epithelial cells per low-power field were excluded.12 A positive culture was defined as more than 1 bacterial species isolated in a quantitative count ≥105 colony-forming units/mL in a voided urine specimen. On admission, the patients were evaluated by the treating physicians for the symptoms of UTI, including urinary frequency, dysuria, gross hematuria, and suprapubic discomfort. Patients with a positive culture and symptoms of UTI were treated with antibiotics, and the VP was canceled. ASB was defined as a positive culture without symptoms. Whether to treat a patient with ASB with preoperative antibiotics was determined by the treating physician. Patients who were treated for ASB were given a 7-day course of oral empirical antibiotics according to in vitro susceptibility, starting from the date of admission. Due to the retrospective nature of this study, the ASB treatment protocol was neither mandatory nor consistent among the study cohort.

VP was performed under local anesthesia by specialty-trained spine surgeons. Preoperative prophylactic parenteral antibiotics were not routinely administered. Through a 2–3 mm incision, a Jamshidi needle was inserted percutaneously into the pedicle under C-arm imaging guidance, through which polymethylmethacrylate cement was inserted.13,14 The patient was discharged on the same day or the next day if no acute complications or discomfort occurred postoperatively. Postoperative bracing was prescribed for 3 months. All patients received outpatient clinic follow-up at 1 week, 4 weeks, 8 weeks, 12 weeks, 6 months, and 12 months after surgery. A plain radiograph of the spine was taken at each follow-up.

Of the 758 patients who were admitted for VP in our hospital during the study period, 42 did not meet the inclusion criteria (15 lost to follow-up, 12 for a history of previous VP, 7 for malignant metastasized vertebral fractures, 4 for multiple myeloma, and 4 who did not undergo VP due to active UTI), leaving 716 patients in the final cohort (Fig. 1). In the total cohort, 509 patients were female (71.1%), and the mean age was 78.6 ± 9.6 years (range, 63–106). No patient had an indwelling urinary catheter when admitted for VP (Table 1).

Table 1 - Patient demographics of the PVI group Patient number Age/sex Comorbidities BMI (kg/m2) Smoking Level Urine culture Treatment for ASB Causative organism of PVI Pathology from biopsy Time to PVI (weeks) 1 85/M HTN 22.3 No T12 No growth – No biopsy No biopsy 6 2 75/F DM, HTN 24.4 No L2, L5 Escherichia coli No No growth Osteomyelitis 4 3 84/M DM, HTN, HCC, gout 27.4 Yes T12 Proteus mirabilis No No biopsy No biopsy 2 4 79/F HTN, Parkinsonism 26.2 No T8, T12 Escherichia coli No Staphylococcus aureus Chronic necrosis 2 5 85/F HTN, colon cancer 27.7 No L1 No growth – No growth Chronic abscess 3 6 87/M HTN 32.1 Yes L1 No growth – Proteus mirabilis Osteomyelitis 5 7 88/M HTN, RCC 33.3 Yes L1 Enterobacter cloacae Yes Cutibacterium acnes Osteomyelitis 2 8 78/F HTN 19.9 No T10 No growth – No growth Osteomyelitis 4 9 82/F HTN, tuberculosis 19.9 No T12, L1 No growth – No growth Osteomyelitis 9

BMI = body mass index; DM = diabetes mellitus; HCC = hepatic cell carcinoma; HTN = hypertension; PVI = post-VP infection; RCC = renal cell carcinoma.

Fig. 1:

Number of patients enrolled, excluded, and included in data analysis.

2.3. Diagnosis of post-VP infection

Post-VP infection was diagnosed based on either of the following assessments: (1) pathology: CT-guided vertebral biopsy at the index level revealing osteomyelitis with infiltration of neutrophils or lymphocytes, or granulomatous change in TB cases15,16; (2) microbiology: biopsy culture yielding growth of bacteria or tubercle bacillus; (3) a combination of clinical findings, laboratory data, and imaging: refractory back pain or fever after the procedure, elevated CRP (normal range, <0.5 mg/dL), and a change in MRI signal in the involved vertebral body at index level with postcontrast enhancement of bone marrow or abscess margins.16,17 Patients were divided accordingly into two groups: PVI and non-PVI groups.

2.4. Statistical analysis

All data analysis and calculations were performed using IBM SPSS Statistics for Windows, version 25.0 (IBM, Armonk, NY). Data were compared between patients with and without PVI. Independent-sample t-tests were used to compare continuous variables, while chi-squared or Fisher’s exact tests were used to compare categorical variables. To identify risk factors for infection after VP, the variables that differed significantly between groups were initially evaluated using univariate logistic regression analysis. Those significantly associated with PVI at p ≤ 0.10 in univariate analysis were entered into the multivariate logistic regression model. The statistical power of each significant variable was analyzed. Potential modifier effects of the variables were also studied. The results are expressed as odds ratios (ORs) with 95% confidence intervals (CIs). A 2-tailed p value <0.05 was considered statistically significant.

3. RESULTS

Of the 716 patients in the final cohort, nine (1.26%) were diagnosed with PVI occurring at a mean time of 4.1 ± 2.3 weeks (range, 2–9 weeks) after the index VP. Seven patients were diagnosed based on pathological and microbiological findings, and 2 were diagnosed based on typical clinical presentation and image features (Table 1). Patients in the PVI group were older (p = 0.03), had a higher incidence of malignancy (p = 0.01), and had a history of smoking (p = 0.01). No significant difference was observed in any of the other demographic characteristics or clinical features. Additional information is shown in Table 2.

Table 2 - Patient demographic data Overall (n = 716) PVI group (n = 9) Non-PVI group (n = 707) p Sex (n [%]) 0.30  Male 207 (28.9%) 4 (44.4%) 203 (28.7%)  Female 509 (72.1%) 5 (55.6%) 504 (71.3%) Age (years) 78.6 ± 9.6 82.6 ± 4.4 78.5 ± 9.7 0.03* BMI (kg/m2) 23.59 ± 4.11 (14–35.8) 25.91 ± 4.82 (19.9–33.3) 23.56 ± 4.1 (14–35.8) 0.09 Smoking (n [%]) 41 (5.7%) 3 (33.3%) 38 (5.4%) 0.01* Medical history (n [%])  Hypertension 401 (56.0%) 8 (88.9%) 393 (55.6%) 0.046*  Diabetes mellitus 152 (21.2%) 2 (22.2%) 150 (21.2%) 1.00  Malignancy 37 (5.2%) 3 (33.3%) 34 (4.8%) 0.01*  ESRD 20 (2.8%) 0 (0%) 20 (2.8%) 1.00  Involved segment (n) 820 12 808 1.00 Thoracic spine (n [%]) 338 (41.2%) 5 (41.6%) 333 (41.2%) 1.00  T6 4 0 4  T7 11 0 11  T8 20 1 19  T9 19 0 19  T10 26 1 25  T11 58 0 58  T12 200 3 197 Lumbar spine (n [%]) 482 (58.8%) 7 (58.3%) 475 (58.8%)  L1 230 4 226  L2 105 2 103  L3 85 0 85  L4 47 0 47  L5 15 1 14 Number of VP segment (n [%])  Single segment 620 (86.6%) 6 (33.3%) 614 (86.8%) 0.11  Multiple segments 96 (13.4%) 3 (66.7%) 93 (13.2%) 0.14 Time from injury to VP (weeks) 9.8 ± 0.4 (1–78) 9.8 ± 8.1 (1–26) 9.9 ± 11.3 (1–78) 0.99 Length of hospitalization (days) 4.2 ± 0.3 (1–45) 8.2 ± 14.3 (2–45) 4.4 ± 7.4 (1–18) 0.45 Follow-up (months) 20.1 ± 9.4 (0–40) 16.9 ± 9.4 (0–34) 20.3 ± 9.3 (6–40) 0.28

Data are presented as the mean ± standard deviation (range) if not otherwise specified.

*Statistically significant.

BMI = body mass index; ESRD = end-stage renal disease; PVI = post-VP infection.

Comparison of the laboratory test results between patients in the PVI and non-PVI groups was shown in Table 3. No significant difference in the preoperative CRP levels or the proportion of patients with abnormal CRP level was observed between the two groups. Preoperative urinalysis showed that abnormal urinalysis was present in 55.6% (5/9) of the PVI patients and 30.5% (251/707) of the non-PVI patients (p = 0.21). While squamous epithelial cells were found in 596 of the 716 urine samples (83%), none of the samples presented more than 8 cells per low-power field. Culture of the urine specimens yielded positive growth and met the diagnosis of ASB in 101 patients (14.1%), with 4 patients in the PVI group (44.4%; 4/9) and 97 patients in the non-PVI group (13.7%; 97/707). The fraction of patients with ASB was significantly higher in the PVI group than in the non-PVI group (p = 0.03). The types of microorganisms isolated from urine in each group are compared in Table 4.

Table 3 - Preoperative lab test results Characteristic Overall (n = 716) PVI group (n = 9) Non-PVI group (n = 707) p CRP on admission (mg/L) 0.89 ± 2.12 3.87 ± 9.47 0.88 ± 1.87 0.40  >0.5 mg/L (n [%]) 229 (32.0%) 3 (33.3%) 226 (32.0%) 1.00  >1.0 mg/L (n [%]) 129 (18.0%) 2 (22.2%) 127 (18.0%) 0.62 Abnormal Urinalysis (n [%]) 256 (35.7%) 5 (55.6%) 251 (35.5%) 0.21  Nitrite positive 109 (15.2%) 3 (33.3%) 106 (15.0%) 0.14  Leukoesterase positive 191 (26.7%) 3 (33.3%) 188 (26.6%) 0.71  Pyuria 243 (33.9%) 4 (44.4%) 239 (33.8%) 0.50  Bacteriuria 232 (32.4%) 3 (33.3%) 229 (31.1%) 1.00 ASB (n [%]) 101 (14.1%) 4 (44.4%) 97 (13.7%) 0.03*  Treated ASB 24 (23.8%) 1 (25.0%) 23 (23.7%) 1.00  Untreated ASB 77 (76.2%) 3 (75.0%) 74 (76.3%) 1.00

Data are presented as the mean ± standard deviation if not otherwise specified.

*Statistically significant.

ASB = asymptomatic bacteriuria; CRP = C-reactive protein; PVI = post-VP infection.

Table 4 - Microorganisms isolated from preoperative urine cultures Isolated species PVI group (n = 4) Non-PVI group (n = 97) p Gram positive 1 (25.0%) 18 (18.6%) 0.57  Staphylococcus aureus 4 (4.1%)  Enterococcus spp. 1 (25.0%) 14 (14.4%) Gram negative 3 (75.0%) 73 (75.3%) 1.00  Klebsiella pneumoniae 1 (25.0%) 12 (12.4%)  Escherichia coli 2 (50.0%) 42 (43.3%)  Morganella morganii 3 (3.1%)  Pseudomonas aeruginosa 4 (4.1%)  Proteus mirabilis 6 (6.2%)  Citrobacter koseri 3 (3.1%)  Gardnerella vaginalis 3 (3.1%) Fungi 6 (6.2%) 1.00  Candida albicans 6 (6.2%)

Data are presented as number (%).

Detailed demographic data for patients in the PVI group are shown in Table 1. The microorganisms found in urine differed from those cultured from the infected vertebrae, indicating that no case of PVI was caused by a pathogen present in the preoperative urine. Of the 4 patients with preoperative ASB in the PVI group, only one was treated before VP. The fraction of patients receiving treatment for ASB did not differ significantly between the PVI and non-PVI groups (25% vs. 23.7%, respectively).

Univariate logistic regression analysis identified multiple risk factors for PVI, including smoking, hypertension, a history of malignancy, and ASB. Multivariable logistic regression analysis revealed that independent risk factors for PVI include smoking (OR, 16.26; 95% CI, 2.58–102.65; p < 0.01), a history of malignancy (OR, 7.27; 95% CI, 1.31–40.31; p = 0.02), and ASB (OR, 5.61; 95% CI, 1.14–27.66; p = 0.03). A post hoc statistical power calculation using the group size (n= 716), the incidence of the PVI (1.26%), and an alpha value of 0.05, all ORs of the four significant factor in univariate logistic regression yields a power of 100%. However, accounting for risk variables using multivariable logistic regression, hypertension was not an independent predictor of the occurrence of PVI (Table 5).

Table 5 - Univariate and multivariate logistic regression analysis of PVI data Univariate logistic regression Multivariate logistic regression Variables OR (95% CI) p OR (95% CI) p Age 1.05 (0.97 to 1.14) 0.21 Smoking 8.79 (2.12 to 36.51) <0.01* 16.26 (2.58–102.65) <0.01* Hypertension 6.39 (0.80 to 51.38) 0.08 8.28 (0.95–72.32) 0.06 History of malignancy 9.06 (2.18 to 37.79) <0.01* 7.27 (1.31–40.31) 0.02* Asymptomatic bacteriuria 5.03 (1.33 to 19.06) 0.02* 5.61 (1.14–27.66) 0.03*

*Statistically significant.

CI = confidence interval; OR = odds ratio.

4. DISCUSSION

The overall prevalence of ASB in our patient cohort was 14.1% in patients who underwent VP. We found that 44.4% of the patients with PVI had preoperative ASB. Further analysis revealed that the likelihood of developing PVI was 5.61 times higher in patients with ASB than in those without. No case of PVI was caused by pathogens present in the pre-VP urine culture. We observed that smoking habit and a history of malignancy were also risk factors for PVI.

We observed a PVI rate of 1.26% (9/716), which is higher than the rates of 0.32% to 0.46% reported in previous studies.2–4 This difference in findings possibly resulted from our determination of PVI based on not only pathology and microbiological data but also a clinical diagnosis supported by MRI features. No standard diagnostic criteria are reported in the literature.2,3,18 Previous studies have described cases with radiographic characteristics typical of PVI but failed to identify the causative organisms or acquire positive pathology results.4,19 Among previous studies, the rate of identification of organisms causing PVI is 50% to 78%.2–4 In our study, two of the nine patients were diagnosed based on typical radiologic findings and clinical presentation (cases 1 and 3 in Table 4).

No clear risk factors for PVI have been identified in previous studies, likely due in part to its very low incidence. Previously identified risk factors for complications and mortality following VP include American Society of Anesthesiologists physical status class 4, history of COPD, kidney disease, or disseminated cancer.20,21 These results are consistent with our finding that the incidence of malignancies was higher among patients with PVI than without. This finding likely reflects the greater comorbidity burdens and weakened immunity resulting from cancer or related therapy. Smoking is a known risk factor for surgical site infection after spinal surgery,22 a finding consistent with our observations. Studies have reported that several compounds present in cigarette smoking have immunosuppressive effects that include impairment of innate defenses against pathogens, modulation of antigen presentation, and promotion of inflammatory processes.23,24

The preoperative management of ASB in specific orthopedic procedures has received much attention.8–10 The prevalence of ASB increases with age, occurring in about 20% of women and 10% of men over 80 years of age.25 The rate in long-term care facilities is even higher, ranging from 25% to 50% in women and 15% to 35% in men.26 In our study cohort, the prevalence of ASB was 14.1%, and patients with ASB were 5.61 times more likely to have PVI than were those without. Several studies report coincident UTI at the time of VP followed by a subsequent surgical infection.27–29 Abdelrahman et al3 reported that of the six cases of VP infection investigated in their study, five presented with UTI as a comorbidity. Since the method used to diagnose UTI was not clearly defined in these studies, we assume that these patients might have had ASB. The symptoms of UTI can be clinically subtle and difficult to differentiate by physicians. Nevertheless, our study shows that a positive preoperative urinary culture, even without symptoms, is a risk factor for PVI.

Possible explanations for the increased risk of PVI in patients with ASB include direct contamination, hematogenous spread, and ASB serving as a marker of poor immunity. Wound contamination and hematogenous spread were highly suspected in our series, since the majority of infections occurred within 6 weeks after surgery (8 of 9). However, in the two patients with positive urine and surgical site cultures, the bacterial species responsible for the VP and urine infections differed. Previous studies of coincident UTI at the time of VP followed by infection did not report the species identified in urine cultures.3,27–29 Studies on ASB before arthroplasty have also noted differences in culture results between urine and periprosthetic infections.8,9