In patients with osteoporosis there is always a strong possibility that pedicle screws will loosen. This makes it difficult to select the appropriate osteoporotic patient for a spinal fusion. The purpose of this study was to determine the correlation between bone mineral density (BMD) and the magnitude of torque required to insert a pedicle screw. To accomplish this, 181 patients with degenerative disease of the lumbar spine were studied prospectively. Each underwent dual-energy x-ray absorptiometry (DEXA) and intra-operative measurement of the torque required to insert each pedicle screw. The levels of torque generated in patients with osteoporosis and osteopenia were significantly lower than those achieved in normal patients. Positive correlations were observed between BMD and T-value at the instrumented lumbar vertebrae, mean BMD and mean T-value of the lumbar vertebrae, and mean BMD and mean T-value of the proximal femur. The predictive torque (Nm) generated during pedicle screw insertion was [-0.127 + 1.62 × (BMD at the corresponding lumbar vertebrae)], as measured by linear regression analysis. The positive correlation between BMD and the maximum torque required to insert a pedicle screw suggests that pre-operative assessment of BMD may be useful in determining the ultimate strength of fixation of a device, as well as the number of levels that need to be fixed with pedicle screws in patients who are suspected of having osteoporosis.
Pedicle screws may be used when correcting deformity of the thoracic and lumbar spine.1 Biomechanical studies have shown that the pull-out strength of a pedicle screw is affected by bone mineral density (BMD),2,3 osteoporosis,4 cortical fixation,2,4 pedicle morphology,5 screw orientation6 and screw-thread area.7 Poor fixation can result in micromovement while fusion is progressing, fibrous encapsulation and loosening of the screws.8 In an ageing population there is a high incidence of degenerative spinal disease. Some of these patients will need surgery that includes pedicle screw fixation. If they are osteoporotic they are at increased risk of screw pull-out.9 To overcome this, many attempts have been made to enhance the initial fixation of the pedicle screw. These include the concomitant use of polymethylmethacrylate (PMMA) bone cement,10 the application of calcium phosphate11 or calcium sulphate ceramics to a pedicle,12 and the use of expandable pedicle screws.13 However, there remains some controversy as to whether the basic quality of the bone is the most crucial factor affecting the strength of fixation and stability.
Previous biomechanical studies have shown that pedicle screw fixation is strongly related to BMD14,15 and that the torque required to insert a screw is strongly associated with its pull-out strength.16,17 Nevertheless, it has not been possible to show a consistent relationship between the two.18 This may be due to differences in the mechanical response of trabecular bone to the forces generated by the insertion and removal of a screw.
The initial stability of a pedicle screw is critical. In an osteoporotic patient failure of a screw can lead to failure of fixation. If it were possible to predict the initial fixation strength of a pedicle screw it would help to plan post-operative management and to reduce the rate of failure. To date, no large-scale study has examined the relationship between the torque required to insert a pedicle screw and BMD in vivo. Nor have there been any studies that predict the torque required from the measured BMD.
This study examined the correlation between BMD and torque during insertion of a pedicle screw in patients designated for instrumented posterior fixation. Subgroup analysis of patients with osteoporosis or osteopenia was performed to identify the relationship between torque and BMD.
Patients and Methods
Between November 2006 and December 2009, 181 patients underwent dual-energy x-ray absorptiometry (DEXA) (Lunar; GE Healthcare, Madison, Wisconsin) in the three months before pedicle screw instrumentation of the lumbosacral spine. Each operation was carried out by a single surgeon (JHL, first author) for the management of degenerative spinal disease. Written informed consent was obtained for intra-operative measurement of the torque needed to insert a pedicle screw. The study was approved by the institutional review board at our hospital and was conducted prospectively.
In each case a conventional posterior approach to the lumbar spine was used: the muscles were elevated bilaterally and the laminae, foramina and facet joints were decompressed. The point of insertion of the screw was confirmed using a Kirschner (K-) wire and an image intensifier. A pilot hole was created with an awl and expanded using a T-shaped wrench, which increased the diameter of the hole to 2.5 mm. We used the same pedicle screws (Xia; Stryker Spine, Allendale, New Jersey) throughout the study. Screws 6.5 mm in diameter and 40 mm to 45 mm in length were used in the lumbar region, and screws measuring 7.5 mm in diameter and 40 mm in length were used in the sacral region. After inserting the screws we checked to ensure that they did not penetrate the anterior cortex of the vertebral body and that they were parallel to the longitudinal axis of the pedicle. The maximal torque generated during insertion was measured by attaching a connector, which had been specifically designed to ensure that it fitted into the head of the pedicle screw, to a digital measuring device (DTDK-N50EX; Kanon, Tsuru, Japan). The screw was then advanced at a speed of 90°/s to 180°/s (Fig. 1). Data from any cases in which post-operative plain radiography or CT showed that the screw was not located within the pedicle were excluded.
Outcome measures and statistical analysis
The patients were classified into the following groups based on the mean BMD of the lumbar vertebrae: normal (non-osteoporotic) (T-value > -1.0), osteopenia (-2.5 < T-value ≤ -1.0), and osteoporosis (T-value ≤ -2.5). In these three groups the maximum torque generated during pedicle screw insertion was correlated with the BMD of the vertebrae into which the screws had been inserted, the mean BMD of all the lumbar vertebrae, and the mean BMD of the proximal femur using Pearson’s correlation coefficient.
In 111 patients who underwent three-dimensional (3D) CT post-operatively, the internal diameter of the pedicle was correlated with the torque generated by inserting the screw. The correlation between BMD and the predicted insertional torque was studied using linear regression analysis.
Two screws in the normal group and one screw in the osteoporotic group were excluded from the data as they had penetrated the walls of the pedicle.
Overall analysis of patients and prognostic value of BMD for torque generated during pedicle screw insertion
The mean torque generated during pedicle screw fixation was 1.63 Nm (sd 0.76) on the left side and 1.68 Nm (sd 0.84) on the right. The correlation coefficient (r) between the left and right sides was 0.704 (95% confidence interval (CI) 0.6458 to 0.7542, sample size = 337), indicating a strong positive correlation (p < 0.0001).
In patients who underwent post-operative 3D-CT (n = 111), the mean inner diameters of the right and left pedicles were 13.8 mm (sd 3.71) and 13.8 mm (sd 3.75), respectively (r = 0.954 (95% CI 0.9340 to 0.9684); p < 0.0001). The mean levels of torque generated during fixation in patients with osteoporosis (1.37 Nm (sd 0.73)) and those with osteopenia (1.48 Nm (sd 0.61)) were significantly lower than those in normal patients (2.08 Nm (sd 0.89)) (p < 0.0001 for both patient groups). There were positive correlations between BMD and T-value at the instrumented lumbar vertebrae, mean BMD and mean T-value at the lumbar vertebrae, BMD and mean T-value at the proximal femur, and maximum torque generated during pedicle screw fixation (Table I).
|Sample size||Correlation coefficient r (95% CI)||p-value|
|Lumbar BMD||349||0.492 (0.4078 to 0.5674)||< 0.0001|
|Lumbar T-value||349||0.521 (0.4400 to 0.5935)||< 0.0001|
|Lumbar mean BMD||680||0.316 (0.2470 to 0.3824)||< 0.0001|
|Lumbar mean T-value||674||0.499 (0.4399 to 0.5535)||< 0.0001|
|Femur BMD||670||0.445 (0.3817 to 0.5034)||< 0.0001|
|Femur T-value||670||0.420 (0.3559 to 0.4807)||< 0.0001|
Linear regression analysis of the prognostic value of BMD for the torque generated during pedicle screw fixation at the corresponding lumbar vertebrae was performed, and yielded the following regression formulae:
1. The predictive torque generated during fixation (Nm) = -0.127 + 1.62 × (the BMD at the corresponding lumbar vertebrae). F-ratio = 110.7, p < 0.001 (sample size = 349).
2. The predictive torque generated during fixation (Nm) = 0.94 + 0.882 × (the mean BMD in the lumbar vertebrae). F-ratio = 75.38, p < 0.001 (sample size = 680).
3. The predictive torque generated during fixation (Nm) = -0.32 + 2.41 × (the mean BMD in the proximal femur). F-ratio = 164.6, p < 0.001 (sample size = 670).
The correlation coefficient (r) between the torque generated during pedicle screw insertion and the inner diameter of the pedicles was 0.110, which was not statistically significant (p = 0.1).
Correlation between bone density at the lumbar vertebrae and that at the proximal femur
The correlation coefficient (r) between the BMD of the lumbar vertebrae into which the pedicle screws had been inserted and the mean BMD of the proximal femur was 0.566 (95% CI 0.4899 to 0.6333, sample size = 348; p < 0.0001). The correlation coefficient (r) between the mean BMD of the lumbar vertebrae and the mean BMD of the proximal femur was 0.425 (95% CI 0.3614 to 0.4848, sample size = 679; p < 0.0001). The correlation coefficient (r) between the T-value of the instrumented lumbar vertebrae and the mean T-value of the proximal femur was 0.526 (95% CI 0.4452 to 0.5978, sample size = 348; p < 0.0001). The correlation coefficient (r) between the mean T-value of the lumbar vertebrae and the mean T-value of the proximal femur was 0.638 (95% CI 0.5911 to 0.6806, sample size = 679; p < 0.0001).
Treatment outcomes in patients with osteoporosis
Patients with osteoporosis were defined as having a mean T-value ≤ -2.5 at the lumbar vertebrae or proximal femur. The mean torque generated during pedicle screw insertion in these patients was 1.37 Nm (sd 0.729) (sample size = 143). There was no linear relationship between the mean BMD values of the lumbar vertebrae and those of the proximal femur. The BMD and T-value of the instrumented lumbar vertebrae and the mean BMD of the lumbar vertebrae were all positively correlated with the torque generated during screw insertion. In addition, the mean T-value at the lumbar vertebrae showed a weak positive correlation with the torque generated during screw insertion (Table II).
|Sample size||Correlation coefficient r (95% CI)||p-value|
|Lumbar BMD||63||0.446 (0.2225 to 0.6245)||0.0003|
|Lumbar T-value||63||0.436 (0.2109 to 0.6170)||0.0004|
|Lumbar mean BMD||139||0.312 (0.1532 to 0.4547)||0.0002|
|Lumbar mean T-value||139||0.276 (0.1149 to 0.4232)||0.001|
|Femur BMD||139||-0.082 (-0.2447 to 0.0862)||0.3401|
|Femur T-value||139||-0.002 (-0.1685 to 0.1645)||0.981|
Treatment outcomes in patients with osteopenia
Patients with osteopenia were defined as having a mean T-value > -2.5 and ≤ -1 at the lumbar vertebrae or proximal femur. The mean torque generated during screw insertion was 1.48 Nm (sd 0.615) (sample size = 303). A weak positive correlation was seen between the mean BMD of the lumbar vertebrae and that of the proximal femur (r = 0.278 (95% CI 0.1309 to 0.4133); p = 0.0003). The BMD and T-value of the instrumented lumbar vertebrae, the mean T-value of the lumbar vertebrae, and the mean BMD and mean T-value of the proximal femur all showed weak positive correlations with the torque generated during screw insertion (Table III).
|Sample size||Correlation coefficient r (95% CI)||p-value|
|Lumbar BMD||167||0.184 (0.0334 to 0.3271)||0.0171|
|Lumbar T-value||167||0.243 (0.0950 to 0.3811)||0.0015|
|Lumbar mean BMD||299||0.057 (-0.0568 to 0.1694)||0.3257|
|Lumbar mean T-value||299||0.176 (0.0636 to 0.2835)||0.0023|
|Femur BMD||295||0.161 (0.0475 to 0.2700)||0.0056|
|Femur T-value||295||0.140 (0.0265 to 0.2505)||0.0159|
Treatment outcomes in normal patients
Patients with normal BMD were defined as having a mean T-value > -1 at the lumbar vertebrae or proximal femur. The mean torque generated during screw insertion was 2.08 Nm (sd 0.887) (sample size = 244). There was a strong positive correlation between the mean BMD of the lumbar vertebrae and that of the proximal femur (r = 0.7130 (95% CI 0.6447 to 0.7700); p < 0.0001). The BMD and T-value in the instrumented vertebrae, the mean BMD and T-value of the lumbar vertebrae, and the BMD and T-value of the proximal femur all were positively correlated with the torque generated during screw insertion (Table IV).
|Sample size||Correlation coefficient r (95% CI)||p-value|
|Lumbar BMD||117||0.467 (0.3113 to 0.597)||< 0.0001|
|Lumbar T-value||117||0.456 (0.2930 to 0.5843)||< 0.0001|
|Lumbar mean BMD||236||0.532 (0.4337 to 0.6177)||< 0.0001|
|Lumbar mean T-value||236||0.527 (0.4279 to 0.6132)||< 0.0001|
|Femur BMD||236||0.472 (0.3658 to 0.5652)||< 0.0001|
|Femur T-value||236||0.452 (0.3438 to 0.5478)||< 0.0001|
Several studies have reported a positive correlation between bone quality, as measured by BMD, and the strength of fixation using a pedicle screw. However, most of these studies were conducted using cadavers or large animals. We report the first prospective study of a large group of patients in a clinical setting. The results are of clinical significance, as on subgroup analysis there was some variation between the osteoporotic, osteopenic and normal patients.
Several factors can affect the insertional torque of a pedicle screw, including variations in the point of insertion, the type of instrument used, the size of the instrument used to enlarge the screw track, and the design of the screwdriver handle used for measurement. In this study, the screw track was enlarged by one surgeon using the same instrument and screwdriver handle, which minimised potential errors of measurement. However, there remain some limitations to this study. First, we did not make a prospective assessment of sample size; even so, it was quite large compared with those of other in vivo studies19,20 and we were able to achieve statistically significant results. Secondly, as we only measured final torque for each screw we could not correlate the differences in initial torque and middle torque or with BMD. However, the stability of the pedicle screw is determined mainly by the final torque, not the initial or middle torque, and this is probably the most clinically significant. Thirdly, BMD, as measured by DEXA, cannot represent the bone quality of the lumbar spine exactly, as it is a two-dimensional imaging technique and degenerative change in the vertebra or calcification of the soft tissues, including the adjacent vascular structures, may overestimate its true value. However, DEXA is still the most commonly used diagnostic tool in the field of osteoporosis, and the insertional torque data derived from this study can be applied clinically.
The torque generated during pedicle screw insertion showed positive correlations with the BMD of the vertebrae into which the screws had been inserted, T-value, mean T-value at the lumbar vertebrae, and mean BMD and T-value at the proximal femur. Furthermore, the mean magnitude of torque in the osteoporotic patients was significantly lower than that in non-osteoporotic patients. However, there was no significant difference in the torque generated during screw insertion between patients with osteoporosis and those with osteopenia. These results differ from those of other in vitro reports. Moreover, although the diameter of the pedicle has been reported to be one of the factors affecting insertional torque,16 we found no correlation between the two in this study. We suggest that the trabecular bone inside the pedicle greatly affects the insertional torque, and the relatively larger diameter of the pedicle compared to that of the pedicle screws (6.5 mm to 7.5 mm) meant that there was no correlation between the diameter of the pedicle and insertional torque. Trabecular bone will also have different mechanical properties to that of cadaveric bone or a sawbones model.
In this study, although the torque generated during pedicle screw insertion in osteoporotic patients correlated strongly with the BMD and T-value of the relevant vertebrae as well as the mean BMD of the lumbar vertebrae, there was no correlation with those of the proximal femur. This could be due to the lack of correlation between the BMD of the lumbar vertebrae and that at the proximal femur in patients with osteoporosis. There was a weak positive correlation between the mean BMD of the lumbar vertebrae and that of the proximal femur in patients with osteopenia but no correlation between mean BMD of the lumbar vertebrae and that of the proximal femur in patients with osteoporosis. The correlation coefficient (r) between BMD and the T-value of the vertebrae into which screws had been inserted was significantly higher than that between the mean BMD and mean T-value of the lumbar vertebrae. This suggests that the BMD of the vertebrae into which the screws are inserted is of greater significance than the mean value of the lumbar vertebrae.
Overall, in patients with osteoporosis the BMD of the instrumented vertebrae was strongly and positively correlated with the maximum torque generated during screw insertion, which suggests that pre-operative assessment of BMD is essential. The maximum torque generated in patients with osteopenia or osteoporosis could be predicted by a regression formula using the BMD of the relevant vertebrae, the mean BMD of the lumbar vertebrae, or that of the proximal femur.
In spinal surgery, torque-measuring devices are not routinely used in the operating theatre. Instead, the surgeon judges the insertional torque by determining how much force is needed to insert the screw.17 However, this judgement is likely to be inaccurate. The measurement of insertional torque with a torque gauge in patients with osteoporosis and osteopenia can help to plan the extent of fusion and the duration of post-operative immobilisation.
1 Cheung WY , LenkeLG, LukKD. Prediction of scoliosis correction with thoracic segmental pedicle screw constructs using fulcrum bending radiographs. Spine2010;35:557–561. Google Scholar
2 Soshi S , ShibaR, KondoH, MurotaK. An experimental study on transpedicular screw fixation in relation to osteoporosis of the lumbar spine. Spine1991;16:1335–1341. Google Scholar
3 Pfeiffer M , GilbertsonLG, GoelVK, et al.Effect of specimen fixation method on pullout tests of pedicle screws. Spine1996;21:1037–1044. Google Scholar
4 Zindrick MR , WiltseLL, WidellEH, et al.A biomechanical study of intrapeduncular screw fixation in the lumbosacral spine. Clin Orthop1986;203:99–112. Google Scholar
5 Lill CA , SchneiderE, GoldhahnJ, JaslemannA, ZeifangF. Mechanical performance of cylindrical and dual core pedicle screws in calf and human vertebrae. Arch Ortho Trauma Surg2010;126:686–694. Google Scholar
6 Zehnder S , BledsoeJG, PuryearA. The effects of screw orientation in severely osteoporotic bone: a comparison with locked plating. Clin Biomech (Bristol, Avon)2009;24:589–594. Google Scholar
7 Abshire BB , McLainRF, ValdevitA, KambicHE. Characteristics of pullout failure in conical and cylindrical pedicle screws after full insertion and back-out. Spine J2001;1:408–414. Google Scholar
8 Law M , TencerAF, AndersonPA. Caudo-cephalad loading of pedicle screws-mechanisms of loosening and methods of augmentation. Spine1993;18:2438–2443. Google Scholar
9 Burval DJ , McLainRF, MilksR, InceogluS. Primary pedicle screw augmentation in osteoporotic lumbar vertebrae-biomechanical analysis of pedicle fixation strength. Spine2007;32:1077–1083. Google Scholar
10 Cook SD , SalkeldSL, StanleyT, FacianeA, MillerSD. Biomechanical study of pedicle screw fixation in severely osteoporotic bone. Spine J2004;4:402–408. Google Scholar
11 Hashemi A , BednarD, ZiadaS. Pullout strength of pedicle screws augmented with particulate calcium phosphate: an experimental study. Spine J2009;9:404–410. Google Scholar
12 Liu D , LeiW, WuZX, et al.Augmentation of pedicle screw stability with calcium sulfate cement in osteoporotic sheep biomechanical and screw-bone interfacial evaluation. J Spinal Disord Tech2011;24:235–241. Google Scholar
13 Cook SD , BarberaJ, RubiM, SalkeldSL, WhitecloudTS 3rd. Lumbosacral fixation using expandable pedicle screws. an alternative in reoperation and osteoporosis. Spine J2001;1:109–114. Google Scholar
14 Halvorson TL , KelleyLA, ThomasKA, WhitecloudTS 3rd, CookSD. Effects of bone mineral density on pedicle screw fixation. Spine1994;19:2415–2420. Google Scholar
15 Okuyama KE , AbeE, SuzukiT, et al.Influence of bone mineral density on pedicle screw fixation: a study of pedicle screw fixation augmenting posterior lumbar interbody fusion in elderly patients. Spine J2001;1:402–407. Google Scholar
16 Zdeblick TA , KunzDN, CookeME, McCabeR. Pedicle screw pullout strength: correlation with insertional torque. Spine1993;18:1673–1676. Google Scholar
17 Siddiqui AA , BlakemoreME, TarziI. Experimental analysis of screw hold as judged by operators v pullout strength. Injury2005;36:55–59. Google Scholar
18 Inceoglu S , FerraraL, McLainRF. Pedicle screw fixation strength: pullout versus insertional torque. Spine J2004;4:513–518. Google Scholar
19 Ozawa T , TakahashiK, YamagataM, et al.Insertional torque of the lumbar pedicle screw during surgery. J Orthop Sci2005;10:133–136. Google Scholar
20 Mizuno K , ShinomiyaK, NakaiO, ShindoS, OtaniK. Intraoperative insertion torque of lumbar pedicle screw and postoperative radiographic evaluation: short-term observation. J Orthop Sci2005;10:137–144. Google Scholar
This work was supported by a grant-in-Aid from the SMG-SNU Boramae Medical Center.
No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article.
Supplementary material. Scattergraphs showing A1) the regression equations for expected torque by bone mineral density (BMD) and A2) the correlations between T-value and insertion torque, each for the corresponding lumbar vertebra, the mean lumbar result and the proximal femur, and A3) the correlation between BMD and torque at the corresponding lumbar vertebra for osteoporotic, osteopenic and normal patients, are available with the electronic version of this article on our website www.jbjs.org.uk