Prevalent morphometric vertebral fractures as a risk factor for subsequent clinical vertebral fractures after short-fusion surgery in older Japanese women with degenerative spondylolisthesis

Article information

Asian Spine J. 2024;18(3):435-443

Publication date (electronic) : 2024 June 25

doi : https://doi.org/10.31616/asj.2023.0327

Yosuke Oishi¹, Eiichiro Nakamura², Keiji Muramatsu³, Masaaki Murase¹, Katsumi Doi¹, Yoshinori Takeuchi¹, Jun-ichi Hamawaki¹, Akinori Sakai²

¹Department of Orthopaedic Surgery, Hamawaki Orthopaedic Hospital, Hiroshima, Japan

²Department of Orthopaedic Surgery, University of Occupational and Environmental Health, Kitakyushu, Japan

³Department of Preventive Medicine and Community Health, University of Occupational and Environmental Health, Kitakyushu, Japan

Corresponding author: Yosuke Oishi, Department of Orthopaedic Surgery, Hamawaki Orthopaedic Hospital, 4-6-6 Otemachi, Naka-ku, Hiroshima 730-0051, Japan, Tel: +81-82-240-1611, Fax: +81-82-240-1122, E-mail: ikyoku@hamawaki.or.jp

Received 2023 October 6; Revised 2024 January 12; Accepted 2024 January 18.

Abstract

Introduction

Vertebral fractures (VFs) are the most common osteoporotic fractures [1] and are associated with long-term back pain, impaired function, and decreased health-related quality of life [2–5]. Subsequent VF is a complication of instrumented spinal fusion surgery [6–14]. Therefore, evaluating risk factors for subsequent VFs after fusion surgery is clinically important.

Previous studies have evaluated several risk factors for subsequent VFs after fusion surgery, such as bone fragility (older age, women, low bone mineral density [BMD], and medical comorbidities) [9–12] and increased mechanical loading on the vertebrae (global malalignment, long fusion, and fusion to the sacrum) [11–13]. Some population-based studies have reported the presence of prevalent morphometric VFs as an additional risk factor for osteoporotic VF occurrence in older women [15–19], independent of low BMD [20]. Horii et al. [21] demonstrated that participants with prevalent severe VFs (sVFs) had a higher frequency of subsequent VFs and sVFs than those without.

Prevalent VFs have been recognized as an indicator of bone fragility in reconstructive spinal surgery [22]. However, no studies have yet investigated whether prevalent morphometric VFs are a risk factor for subsequent VFs after short-fusion surgery in patients with degenerative lumbar diseases, including degenerative spondylolisthesis with instability [23]. Although degenerative spondylolisthesis tends to develop in older women, patients with degenerative spondylolisthesis have a different pathology than that in population-based studies.

Therefore, using the Kaplan-Meier method with propensity-score matching in a retrospective cohort study, we investigated whether the presence of prevalent morphometric VFs was a risk factor for subsequent clinical VFs in women aged ≥60 years who underwent short-fusion surgery for degenerative spondylolisthesis.

Materials and Methods

Patients

In this retrospective cohort study, consecutive medical records of women aged ≥60 years who presented to our institute with symptomatic neurological spinal canal stenosis caused by degenerative spondylolisthesis were retrospectively reviewed. Patients who underwent fusion surgery using instruments (posterolateral lumbar fusion [PLF] or posterior lumbar interbody fusion [PLIF]) within two segments of Hamawaki Orthopaedic Hospital from 2003 to 2017 were included (n=358). Generally, patients were examined, and additional radiographs were taken periodically. Patients who could be followed up for at least 3 years were included (n=276, 77%). Patients with coexisting degenerative scoliosis (Cobb angles on standing radiographs of ≥20°) were excluded. Furthermore, patients who had undergone lumbar fusion surgery, had a fracture after a major injury, or had an infection, rheumatoid arthritis, tumor, or any other disease that would induce pathological fractures were excluded. Finally, 237 patients (66%) were enrolled in the study, including 50 and 187 patients with prevalent morphometric VFs (VF [+] group) and nonprevalent morphometric VFs (VF [−] group), respectively. Among these, 40 and 80 patients in the VF (+) and VF (−) groups, respectively, were analyzed and matched by propensity score for age, follow-up duration, surgical procedure (PLF or PLIF), number of fused segments, body mass index (BMI), and number of patients treated for osteoporosis (caliper width: 0.2×standard deviation of propensity score) (Fig. 1).

Fig. 1

Patient selection flow chart.

Radiographic assessment

Prevalent morphometric VFs were judged as grade ≥1 using the semiquantitative (SQ) method (grades 0–3) as described by Genant et al. [24], employing standing lateral lumbar and thoracic spine radiographs obtained preoperatively. Subsequent clinical VFs were clinically judged based on the complaint of acute lumbago and morphologically using radiographic findings corresponding to acute lumbago as follows: at least one vertebra was changed from SQ=0 to SQ ≥1, or a previous VF was changed to VF at a higher SQ grade than the previous latest radiographs at follow-up [25] (Table 1). A single VF was defined as one VF, whereas multiple VFs were defined as two or more VFs. Mild VF (mVF) and sVF were defined as the maximum grade of SQ=1 and SQ ≥2, respectively [21] (Table 1). Morphological evaluation using the SQ method was repeated for the same 50 patients over 1 month to allow for calculations of the intraclass correlation coefficient, and the kappa coefficient was calculated.

Table 1

Definition of each VF in this study

Results

The kappa coefficient for the morphological evaluation of the VFs using the SQ method was 0.838. Fig. 2 presents the two peaks of the prevalence of prevalent morphometric VFs: one at the thoracolumbar lesion (T11–L2) and the other at the lower lumbar lesion (L3–5). In the VF (+) group, 38 (76%), 12 (24%), 15 (30%), and 35 (70%) patients had a single VF, multiple VFs (Fig. 3A), mVF, and sVF (Fig. 3B), respectively. Overall, 66% of single and 83% of multiple VFs were sVFs.

Fig. 2

Prevalence (%) of prevalent morphometric vertebral fractures (VF) according to the site before matching.

Fig. 3

Prevalence (%) of prevalent morphometric vertebral fractures according to number (A) and grade (B) before matching. VF, vertebral fracture.

Subsequent clinical VFs were observed in 44 patients. Magnetic resonance (MR) imaging (MRT-203/P2, 1.5 T; Toshiba, Tokyo, Japan) was possible in 64% of the cases, showing low and high intensities on T1-weighted MR and short T1-inversion recovery images, respectively, thereby confirming fresh VF.

Before matching, the VF (+) group was comprised of significantly older patients at the time of surgery and had a shorter follow-up duration than the VF (−) group. However, no significant differences were observed between the two groups in terms of the surgical procedure (PLF or PLIF), number of fused segments, number of patients with laminectomy at the adjacent segment, patients with fusion to the sacrum, those treated for osteoporosis, those with DM, those with smoking history, BMI at the time of surgery, and LL (Table 2).

Table 2

Comparison of parameters between VF (+) and (−) groups before matching

No significant differences were found in any factor between the two groups after matching (Table 3). Little’s MCAR test showed that the hip BMD data and SVA were not MCAR before matching but became MCAR after matching. In addition, the integrated results from the 50 pooled datasets using multiple imputation methods showed no significant differences in hip BMD and SVA between the VF (+) and VF (−) groups before and after matching (Tables 2, 3).

Table 3

Comparison of parameters between VF (+) and (−) groups after matching

Among all patients, the Kaplan-Meier curves showed that the VF (+) group had a significantly higher incidence of subsequent clinical VFs after fusion surgery than the VF (−) group before matching (log-rank test, p<0.001) (Fig. 4A). Cox regression analysis for the presence of prevalent morphometric VFs and age showed that the presence of prevalent morphometric VFs was an independent risk factor for subsequent clinical VFs after short-fusion surgery (HR, 3.822; 95% CI, 2.107–6.932; p<0.001). Furthermore, Kaplan-Meier curves showed similar results after matching (log-rank test, p=0.013) (Fig. 4B).

Fig. 4

The Kaplan-Meier curves show that the incidence of subsequent clinical vertebral fractures (VF) was significantly higher in the VF (+) group than in the VF (−) group before (A) and after (B) matching (log-rank test, p<0.001 and p=0.013, respectively).

In both groups, most subsequent clinical VFs occurred at remote levels from the fusion site postoperatively. However, subsequent VFs occurred at the intrafusion level in the VF (+) group rather than in the VF (−) group before matching (Fig. 5). Subsequent sVF/VF (×100) values were 100% and 90% in the VF (+) and VF (−) groups, respectively.

Fig. 5

Levels of subsequent clinical vertebral fracture (VF) in the VF (+) (A) and VF (−) (B) groups.

No significant differences in subsequent clinical VFs were observed between the prevalent morphometric single VF and multiple VFs and between the prevalent morphometric mVF and sVF during the postoperative period before (Fig. 6A, B) and after matching. Furthermore, before matching, the prevalent single lower lumbar fracture group had a significantly higher frequency of subsequent VFs than the prevalent single thoracolumbar fracture group 20 months and 40 months after fusion surgery (Fig. 6C); however, no significant differences were observed at any period after matching.

Fig. 6

Incidence (%) of subsequent clinical vertebral fracture (VF) between prevalent single and multiple VFs (A), prevalent mild and severe VFs (B), and prevalent single lower lumbar and thoracolumbar VFs (C) at 20, 40, 60, and 80 months postoperatively before matching. NS, not significant.

Discussion

In this study, Kaplan-Meier analysis demonstrated that the VF (+) group had a significantly higher incidence of subsequent clinical VFs than the VF (−) group, and Cox regression analysis showed that the presence of prevalent morphometric VFs was an independent risk factor for subsequent clinical VFs before matching. Kaplan-Meier analysis yielded similar results after matching.

Fusion surgery induces excessive mechanical stress on the adjacent vertebrae. Previous studies have shown that BMD around the fusion site decreased at some point after fusion surgery [26,27]. Moreover, population-based studies have demonstrated that the presence of prevalent morphometric VFs carries a risk of subsequent VFs, independent of BMD [15–20]. Our results revealed that the presence of prevalent morphometric VFs may be a risk factor for subsequent clinical VFs in older women who underwent short-fusion surgery for degenerative spondylolisthesis. In addition, subsequent VFs occurred at intrafusion levels in the VF (+) group, potentially worsening the clinical outcomes because of the destruction of the fusion site.

A few studies have reported the risk factors for subsequent VFs in patients who underwent short-fusion surgery for degenerative lumbar diseases. Toyone et al. [9] demonstrated that the incidence of subsequent VFs was higher in postmenopausal women than in men after fusion surgery, particularly at the level adjacent to the fusion site. Nakahashi et al. [10] indicated that older women aged ≥70 years had a higher incidence of subsequent VFs than younger women after fusion surgery. However, none of these studies analyzed patients according to the presence or absence of prevalent VFs. Kao et al. [11] showed that subsequent VFs occurred in 31.8% and 22.7% of patients with and without prevalent VFs, respectively, for an average of 24.9 months after fusion surgery; however, the study included long fusion cases and did not directly compare the incidence of subsequent VFs with and without prevalent VFs.

Our results showed no significant differences in the number of fused segments or the number of patients with fusion to the sacrum, LL, or SVA. Thus, the presence of prevalent VFs may indicate not only increased mechanical loading on the vertebrae but also increased bone fragility. Moreover, no significant differences in age, sex, or BMD were found between the two groups. Consequently, the presence of prevalent VFs was a risk factor indicating bone fragility, independent of older age, female sex, and low BMD.

No significant differences were observed in the subsequent clinical VFs and sVFs between the prevalent morphometric mVF and sVF groups. This result differs from that of a recent population-based study in which the incidence of both VFs and sVFs tended to increase sequentially according to the prevalent VF (−), mVF, and sVF groups [21]. This implies that the higher rates of prevalent VFs in population-based studies indicate increased bone fragility, although this was not observed in the present study.

In this study, although the sample size was relatively small, the prevalent single lower lumbar VF group had a significantly higher frequency of subsequent clinical VFs than the prevalent single thoracolumbar VF group in the early period after fusion surgery, but before matching. A Japanese multicenter study indicated that patients with prevalent symptomatic lower lumbar VFs might have a higher bone fragility incidence than those with prevalent thoracolumbar VFs postoperatively [6]. However, Kaplan-Meier analysis, excluding prevalent lower lumbar VFs, also showed that before matching, the VF (+) group had a significantly higher incidence of subsequent clinical VFs than the VF (−) group (log-rank test, p=0.007), and no significant differences were found in other parameters between the two groups. Therefore, prevalent morphometric VFs were a risk factor for subsequent clinical VFs, even if the prevalent VFs were not near the fusion site.

Regarding the prevention of subsequent VFs, several studies have demonstrated a greater reduction in subsequent VFs with TP use than with no treatment or BP use [7,28,29]. Therefore, the use of a potent osteoporosis medication may have a preventive effect after short-fusion surgery for degenerative lumbar disease with prevalent VFs.

This study had some limitations. First, the sample size was small, and the study was a retrospective cohort. However, the prevalence of VFs in patients with degenerative spondylolisthesis treated with fusion surgery is limited, and the detection of subsequent clinical VFs requires time. Second, in analyzing the number, grade, and site of prevalent VFs, the sample size decreased because the VF (+) group was further subdivided. Third, hip BMD and SVA data were missing, and these data were not MCAR before matching. Because determining whether these data were MCAR was difficult, the multiple imputation method was employed to address this issue, even though these data were MCAR after matching. Furthermore, because of the large amount of data with missing values, a more reliable multiple imputation method was used instead of a complete case analysis.

Conclusions

The presence of prevalent morphometric VFs may be a risk factor for subsequent clinical VFs in older women who underwent short-fusion surgery for degenerative spondylolisthesis. Furthermore, the use of a potent osteoporosis medication should be recommended if fusion surgery is performed on patients with degenerative spondylolisthesis with prevalent morphometric VFs.

Article information Continued