A

btaairdcf©

K

1

npppts(a

3

m(c

0d

European Journal of Radiology 69 (2009) 289–295

Percutaneous vertebroplasty with a high-qualityrotational angiographic unit

Alessandro Pedicelli a,∗, Massimo Rollo a,1, Mariangela Piano a,1, Thomas J. Re a,1,Maria C. Cipriani b,1, Cesare Colosimo a,1, Lorenzo Bonomo a,1

a Department of Bioimaging and Radiological Sciences, Catholic University of Sacred Heart, Policl. A.Gemelli,l.go Gemelli 1, 00168 Rome, Italy

b Department of Gerontology, Catholic University of Sacred Heart, Policl. A.Gemelli, l.go Gemelli 1, 00168 Rome, Italy

Received 23 August 2007; received in revised form 13 October 2007; accepted 17 October 2007

bstract

We evaluated the reliability of a rotational angiographic unit (RA) with flat-panel detector as a single technique to guide percutaneous verte-roplasty (PVP) and for post-procedure assessment by 2D and 3D reformatted images. Fifty-five consecutive patients (104 vertebral bodies) werereated under RA fluoroscopy. Rotational acquisitions with 2D and 3D reconstruction were obtained in all patients for immediate post-proceduressessment. In complex cases, this technique was also used to evaluate the needle position during the procedure. All patients underwent CT scanfter the procedure. RA and CT findings were compared. In all cases, a safe trans-pedicular access and an accurate control of the bone-cementnjection were successfully performed with high-quality fluoroscopy, even at the thoracic levels and in case of vertebra plana. 2D and 3D rotationaleconstructions permitted CT-like images that clearly showed needle position and were similar to CT findings in depicting intrasomatic implant-

istribution. RA detected 40 cement leakages compared to 42 demonstrated by CT and showed overall 95% sensitivity and 100% specificityompared to CT for final post-procedure assessment. Our preliminary results suggest that high-quality RA is reliable and safe as a single techniqueor PVP guidance, control and post-procedure assessment. It permits fast and cost-effective procedures avoiding multi-modality imaging.

2007 Elsevier Ireland Ltd. All rights reserved.

gy; X

cbamc[pi

eywords: Spine; Vertebroplasty; 3D rotational imaging; Interventional radiolo

. Introduction

Since the first description by Galibert et al. [1], percuta-eous vertebroplasty (PVP) has become an established effectiverocedure for treatment of painful osteoporotic vertebral com-ression fractures, symptomatic vertebral haemangioma andainful vertebral body tumor (myeloma or metastases) refractoryo conservative therapy [2]. The image-guided procedure con-

ists in the percutaneous injection of polymethyl-methacrylatePMMA) through a needle into the vertebral body under localnesthesia and it is used to relieve pain and provide stability. PVP

∗ Corresponding author at: v.Trionfale 7130, 00135 Rome, Italy. Tel.: +39 060154799; fax: +39 06 35501928.

E-mail addresses: apedicelli@rm.unicatt.it (A. Pedicelli),rollo@rm.unicatt.it (M. Rollo), mariangela.piano@gmail.com

M. Piano), tomjre@gmail.com (T.J. Re), alexped@yahoo.com (M.C. Cipriani),olosimo@rm.unicatt.it (C. Colosimo), lbonomo@rm.unicatt.it (L. Bonomo).1 Tel.: +39 06 30154799; fax: +39 06 35501928.

fwis[

tanaa

720-048X/$ – see front matter © 2007 Elsevier Ireland Ltd. All rights reserved.oi:10.1016/j.ejrad.2007.10.017

-ray fluoroscopy; Spiral CT

an be performed using only single-plane fluoroscopy; however,iplane monitoring or CT-fluoroscopy or a combination of CTnd single-plane fluoroscopy are considered the best guidanceethods when more precise needle placement is required, espe-

ially in case of thoracic vertebrae, tumors and difficult situations2,3]. After the fluoroscopy-guided injection of PMMA, post-rocedural assessment is often needed for a precise evaluation ofmplant-distribution. Such assessment is particularly importantor the detection of extra-somatic cement leakages especiallyhen they occur in the spinal canal or vertebral foramen. CT

s considered the method of choice for exact final assessmentince fluoroscopic images could produce equivocal findings4].

With the present work we considered the possibility to avoidhe recourse to CT evaluating the reliability of the rotational

ngiographic unit (RA) with flat-panel detector as a single tech-ique to guide PVP and for post-procedure assessment by 2Dnd 3D reformatted images after reconstruction from rotationalcquisitions.

2 rnal o

2

dfq1efllo

iybtRwAaft2afTTimp3

ocacPtfwptawitagraRMo

ptt

wCeailfi

3

stpwd(nvtspiaw3oigiswmTmetastases from breast cancer had asymptomatic pulmonarycement micro-embolism during PVP; in another patient withosteoporotic vertebral fractures, a posterior foraminal cementleakage caused transient radicular pain that was resolved in about

90 A. Pedicelli et al. / European Jou

. Materials and methods

In our institution a RA (Allura Xper FD 20 with flat-paneletector, Philips Medical Systems, The Netherlands) is availableor vascular and extra-vascular interventional procedures. Highuality main features of the equipment are: imaging chain of 2k2;pixel = 154 �; flat detector with 5 × 106 px; detective quantum

fficiency (DQE) >65%; detector rotation at 90◦ in <3 s; pulseduoroscopy with 30 images/s; 1 mm Al equivalent filtration for

ow-energy X-ray radiations with auto-regulation on the basisf the patient constitution.

In this work 55 consecutive patients were included afternformed consent: 47 females and 8 males (mean age: 72.8ears; age range: 50–93 years) with osteoporotic vertebralody fracture (n = 41), myeloma (n = 4) and vertebral metas-ases (n = 10). On an outpatient basis, all patients underwentA-guided PVP, with complete post-procedure assessment bothith 2D/3D reconstructed rotational acquisitions and CT scans.total of n = 104 vertebrae were treated, localized between Th5

nd L5 (42 thoracic and 62 lumbar). All procedures were per-ormed by the same operator under local anesthesia by monorans-pedicular access. Mild conscious sedation was required in0 patients. Injection of PMMA (amount range: 2–4 mL) waslways guided by single-plane low-dose fluoroscopy and per-ormed with a 13 gauge needle for vertebrae between Th5 andh11 and with a 10 gauge needle for Th12 and lumbar levels.he 13 G needle was also used in case of difficult access like

n vertebra plana (n = 12 vertebrae), in severely compromisedetastatic vertebrae (n = 3 vertebrae) and in case of very small

edicles. In our series we measured the narrowest pedicle of.6 mm in size.

Rotational acquisitions with post-processing reconstructionf 2D and 3D reformatted pictures were obtained in complexases to evaluate the correct needle position through the pediclend to verify on the axial view that no crossing of its medialortex into the spinal canal had occurred before injecting theMMA. In all cases rotational acquisitions were obtained at

he end of the procedure in order to perform a final immediateollow-up assessment in the angio-suite. Rotational acquisitionsere always achieved with the patient remaining in the proneosition and holding his breath for a time of 8 s. Rotation ofhe C-arm covered 180◦, with the acquisition of 240 imagest a frame-rate of 30 images/s. A dedicated program (vert roll)as selected for acquisitions of the spine, with lowered kilovolt

mages compared to traditional angiographic rotational acquisi-ions, in order to optimize bone display. In about 5 s the completecquired series was transferred to a local workstation (Inte-ris 3D-RA) for post-processing. Finally, immediate automaticeconstruction of a 3D model with a matrix of 5123 was producednd different 2D (MultiPlanar Reformat, MPR) and 3D (Volumeendering, VR; Gradient Rendering, GR; Shaded Surface, SS oraximum Intensity Projection, MIP) display techniques were

btained.

Each patient also underwent a CT examination (LightSpeed

ro 4, General Electric; helical acquisition with a collima-ion of 1.5 mm) a few hours later or on the day following thereatment and post-processing multiplanar reconstructed images

Fqb

f Radiology 69 (2009) 289–295

ere obtained on a dedicated workstation. Complete RA andT data of the post-procedure study were independently blindvaluated by two radiologists assessing: presence and degree ofrtifacts and presence, site (posterior epidural, posterior foram-nal, paravertebral, intradiscal) and extension (thickness andength in mm) of any cement leakage. Respective findings werenally analyzed site-by-site.

. Results

High-quality fluoroscopy allowed in all cases to perform aafe trans-pedicular access, even at the thoracic levels. No cor-ical penetration with extra-pedicular or extra-somatic needlerogress occurred. Also difficult cases like vertebrae planaeere safely approached with precise introduction of the nee-le through the pedicle and inside the collapsed vertebral bodyFig. 1). When rotational acquisitions were used to evaluate theeedle position (n = 3 vertebrae with a small pedicle and n = 3ertebrae with not clear visualization of the pedicle borders),he reformatted images obtained after post-processing recon-truction clearly showed the course of the needle through theedicle with no significant metal artifacts deteriorating the def-nition of the vertebral pedicle or body cortical borders (Fig. 2and b). The best visualization of the needle and its relationshipith the vertebral structures was achieved with 2D MPR andD MIP techniques. High spatial and contrast resolution of flu-roscopy always provided accurate control of the subsequentnjection of PMMA with a neat delineation of the implant mar-ins in case of homogeneous and compact distribution withinntravertebral clefts (Fig. 3a) and a detailed visualization of theponge-like cement-distribution in case of osteoporotic bodiesith no intravertebral clefts or lysis (Fig. 3b). In our series noajor complications occurred during or after the procedures.wo minor complications were reported: one patient with bone

ig. 1. An 80-year-old female with osteoporotic compression fractures. High-uality fluoroscopy permits a safe trans-pedicular access within the collapsedody of a vertebra plana (L1).

A. Pedicelli et al. / European Journal of Radiology 69 (2009) 289–295 291

Fig. 2. A 75-year-old female with osteoporotic compression fractures. Fluo-roscopy (a) does not permit a good visualization of the pedicle borders duringnrc

1Bai

eoTvlr

Fig. 3. In (a), a 82-year-old male with osteoporotic compression fracture ofL4: after treatment, fluoroscopy shows the compact distribution of the cementicc

twdCcrrtsma

4

eedle insertion in L5. The 2D MPR images with an axial view obtained afterotational acquisition allow a CT-like visualization of the correct trans-pedicularourse of the needle (b) during the percutaneous access.

month by oral administration of oxycodone and paracetamol.esides these two cases, a total of 40 asymptomatic cement leak-ges (n = 8 posterior epidural, n = 20 paravertebral, and n = 12ntradiscal) occurred during the procedures.

Rotational reformatted images correctly revealed the pres-nce and extension of 40 leaks (Fig. 4a–f) compared to a totalf 42 leakages shown by CT as the gold standard technique.

wo false negatives reported with RA were attributed to theery small size of the leakage, not visible at lower spatial reso-ution of the reformatted images in comparison with axial andeformatted CT images. Sensitivity and specificity for detec-

mpfh

nside a large intravertebral cleft. In (b), a 76-year-old female with osteoporoticompression fractures: fluoroscopy shows the sponge-like distribution of theement during injection in the L3 body.

ion of cement leakages with reconstructed rotational imagesere 95% and 100%, respectively. The intrasomatic implant-istribution displayed by RA reformatted images was similar toT findings in all cases (Fig. 5a and b). The best visualization ofement-distribution and extra-somatic leaks was obtained afterotational acquisitions by integration of 2D MPR and 3D MIPeconstructed images (Fig. 6). RA reformatted images resultedo be more sensitive to motion artifacts compared to CT. In oureries these artifacts occurred in 9 patients but they never causedisinterpretation of images for the final assessment as resulted

fter site-by-site analysis of RA and CT findings.

. Discussion

PVP is characterized by a very low complication rate in the

ajority of reports [2–8], but a small number of serious com-

lications are described [9–10] and all authors agree about theact that such events can be minimized by using appropriateigh-quality guiding systems.

292 A. Pedicelli et al. / European Journal of Radiology 69 (2009) 289–295

Fig. 4. A 75-year-old female with osteoporotic compression fractures: after treatment, 2D MPR axial images from rotational acquisition (a) show a posterior epiduralleakage through the basivertebral veins (arrow). RA definition of leakage site and rate is comparable to CT findings (b; arrow). RA MPR axial images at another level(c) in the same patient show a very small posterior leakage (arrow), as confirmed at CT (d; arrow). A 70-year-old female with osteoporotic compression fractures:after treatment, 2D MPR RA coronal images (e) show site and rate of an intradiscal leakage (arrow), comparable to CT findings (f; arrow).

A. Pedicelli et al. / European Journal o

Fig. 5. A 78-year-old female with osteoporotic compression fractures.After treatment, the post-procedure assessment of the intrasomatic implant-dc

maaaaalogccwad

it

flpvTrawpsaanpemaiaotacrrddv

aihmtca

smao[

3tsMmoib

istribution by 2D MPR images obtained after rotational acquisition (a) isomparable to CT findings (b).

As the majority of the Authors, we consider fluoroscopy theethod of choice as a guiding system, since it is available in

lmost all centers and the majority of interventional radiologistsre trained with it. Furthermore it is easy to use, cost-effectivend reliable as an imaging-based guide for most cases needingPVP procedure. The availability of a high-quality equipment

llows the operator to guide the procedure in almost all thoracicevels and in all lumbar levels. Although procedures performedn upper thoracic and cervical levels in many cases need CTuidance for precise needle insertion and reduction of compli-ation [2,3,11], treatment of vertebrae at these levels is not as

ommon as the treatment of the rest of the spine. In these casese regard the combination of CT and a C-arm fluoroscopic unit

s the method of choice for guidance and control of the proce-ure. Even if this combination is complex, it is easy to organize

ltia

f Radiology 69 (2009) 289–295 293

n most centers when it is not used systematically and is limitedo a small number of cases.

Hoffmann et al. and Trumm et al. [3,7] propose biplaneuoroscopy, CT-fluoroscopy or a combination of CT and single-lane fluoroscopy as the best guiding methods for the exactisualization of the needle position and cement-distribution.he availability of CT-fluoroscopy is of advantage to provide

eal-time accurate insertion of the needle, but for most centressystematic employment of the CT equipment can interfereith the daily diagnostic load. Moreover, CT-fluoroscopy isenalized by a very high average radiation dose compared toingle-plane fluoroscopy, higher radiation exposure to the oper-tor and a small working spaces that may increase contaminationnd post-intervention infection risks for the patient [11,12];evertheless, it needs to be mentioned, that using either single-lane fluoroscopy and CT-fluoroscopy the associated radiationxposure is also strongly user dependent. New developmentsight assist to further reduce the radiation exposure for patients

nd treating physicians. Finally, CT-fluoroscopy, if not specif-cally equipped, is confined to an axial two-dimensional viewnd does not provide a real-time control of all possible sitesf cement leakages especially of the intradiscal ones. In allhe cases studied in our work, single-plane fluoroscopy withvailable rapid rotation of the C-arm (<3 s) provided accurateontrol of intrasomatic cement-distribution and of all sites atisk for cement leakage by multiple views. Multiple views areequired since anterior, posterior and intradiscal leakages areetectable on lateral views, whereas the frontal view more easilyetects leakages in segmental veins often missed on the lateraliew [13].

CT is useful for the intercostovertebral or posterolateralpproach at the thoracic level when the pedicle is too small, butt is associated to a higher risk of pneumothorax or paraspinalematoma [2,3]. In our series, high resolution fluoroscopy per-itted in all cases to perform a safe trans-pedicular access even at

he highest thoracic level (Th5) treated, avoiding the risk of suchomplications. In a few cases rotational reconstructed imagesided in guiding the introduction of the needle.

For post-procedure assessment CT is considered the goldtandard technique [4,14] since fluoroscopy and radiographs caniss or underestimate cement leakages, especially in critical

reas like into the spinal canal and in the neural foramina wherenly 7% of the leakages are correctly identified on radiographs13].

Our results in 55 patients show the validity of RA with 2D andD reformatted images from rotational acquisitions as an advan-ageous tool for PVP post-procedure assessment. In our opinionome key points should be identified: integration of MPR and 3D

IP images provides the best information. In fact, RA 2D refor-atted images are similar to CT-slabs and MIP images among

ther 3D methods allow the best cement visualization and def-nition with resulting a better distinction between PMMA andone structures. RA reconstructed images are characterized by a

ower spatial resolution compared to CT, with a matrix of 5123 ofhe 3D model, but are sufficiently accurate in the evaluation ofmplant-distribution and cement leakage detection. Rotationalcquisition appears to be more sensitive to motion-artifact but

294 A. Pedicelli et al. / European Journal of Radiology 69 (2009) 289–295

Fig. 6. A 67-year-old male with osteoporotic compression fractures. After treatment, the quality of 2D MPR images from rotational acquisition (a) is decreasedb ng pod T (c

ip

gtsa

laainw3bofHfadi

foibrPaceeinastts

y motion artifacts and thus doubtful about the exact localization of a seemiemonstration of a posterior epidural cement leakage (arrow) as confirmed by C

f a 8 s-apnea can be obtained from a cooperative patient, it isossible to minimize this problem.

The main advantage of RA with respect to traditional angio-raphic equipments is obviously the possibility of performingherapy control immediately after the cement injection in theame angio-suite with reconstructed rotational acquisitions [4],voiding the additional charges of a CT examination.

Nowadays RA is considered a valid tool in daily vascu-ar interventional procedures after considerable improvementschieved by technical developments [15,16] and since its firstpplications in the seventies [17]. Most of all it is used fornterventional neuroradiology as a useful technique for plan-ing intracranial aneurysms embolization, providing 3D modelshich can be analyzed by multiple views. Initial reports aboutD RA application for extra-vascular interventions have alsoeen published [18]. To our knowledge at present there is onlyne preliminary report evaluating the possible application of RAor PVP in a patients population [4]: this work, performed byodek-Wuerz et al., focuses on image reconstruction methods

or PVP therapy control in 20 patients. Verlaan et al. defined theccuracy of reconstructed mid-sagittal images from a 3D RAataset for the evaluation of spine specimens treated by minimalnvasive procedures [19].

atle

sterior leakage (arrow). Integration with 3D MIP images (b) permits a clear; arrow).

In some cases, post-processed acquisitions could be also help-ul in the pretherapeutic work-up of the patients when evaluationf the bone structure is needed and no preliminary CT imagings available. Cortical defects, vertebral osteolysis or intraverte-ral clefts identified by a vacuum sign can be visualized witheconstructed images, and can predict the distribution of theMMA [20,21]. In case of small pedicles, planning of the bestpproach to the vertebral body and the choice of needle caliberan be carried out with axial and coronal views which show thexact course of the pedicle and allow measurement of its diam-ter size. In difficult cases we utilized post-processed rotationalmages to assess the correct trans-pedicular introduction of theeedle by CT-like axial views. However, this technical trick waspplied to a few number of procedures and we have no compari-on with CT. Although the images were helpful for the guidance,his represents a limitation of our work for the demonstration ofhe accuracy of RA 2D or 3D reformatted images as a guidingystem for the needle insertion.

It is reported that the radiation exposure to the patient with

single rotational acquisition seems to be reduced compared

o a standard CT study [22–24], but these preliminary pub-ished results need further comparative investigations on spinexaminations.

rnal o

aceriatcta

A

Dc

R

[

[

[

[

[

[

[

[

[

[

[

[

[

A. Pedicelli et al. / European Jou

In conclusion, our study showed that high-quality RA is reli-ble and safe when used as the only technique for guidance,ontrol and post-procedural assessment of PVP. Avoiding themployment of different imaging modalities it can provide aeduction of time and costs. High-quality fluoroscopy as a guid-ng and control system reduces the risks related to needle accessnd cement injection. Detection of cement leakage with a rota-ional image acquisition achieved a good specificity–sensitivityompared to CT, regarded as the gold standard. Therefore, rota-ional acquisition is a useful supplementary tool to classic PVPnd may contribute to patient safety.

cknowledgements

The authors thank Mr. Fernando De Meo, Charge Nurse of theept. of Gerontology, for his praiseworthy professional medical

are coordination and assistance to patients.

eferences

[1] Galibert P, Deramond H, Rosat P, et al. Preliminary note on the treatment ofvertebral angioma by percutaneous acrylic vertebroplasty. Neurochirurgie1987;33(2):166–8.

[2] Gangi A, Sabharwal T, Irani FG, et al. Quality assurance guidelines forpercutaneous vertebroplasty. Cardiovasc Intervent Radiol 2006;29:173–8.

[3] Hoffmann RT, Jakobs TF, Trumm C, et al. Vertebroplasty in the treatmentof osteoporotic vertebral body fracture. Eur Radiol 2007;(February) [Epubahead of print].

[4] Hodek-Wuerz R, Martin JB, Wilhelm K, et al. Percutaneous vertebroplasty:preliminary experiences with rotational acquisitions and 3D reconstruc-tions for therapy control. Cardiovasc Intervent Radiol 2006;29:862–5.

[5] Layton KF, Thielen KR, Koch CA, et al. Vertebroplasty, first 1000 levelsof a single center: evaluation of the outcomes and complications. AJNR2007;28(4):683–9.

[6] Baumann C, Fuchs H, Kiwit J, et al. Complications in percutaneous verte-broplasty associated with puncture or cement leakage. Cardiovasc InterventRadiol 2007;30(2):161–8.

[7] Trumm CG, Jakobs TF, Zech CJ, et al. Vertebroplasty in the treatment ofback pain. Radiologe 2006;46(6):495–505.

[8] Anselmetti GC, Corgnier A, Debernardi F, Regge D. Treatment of painfulcompression vertebral fractures with vertebroplasty: results and complica-tions. Radiol Med 2005;110(3):262–72.

[

[

f Radiology 69 (2009) 289–295 295

[9] Lim KJ, Yoon SZ, Jeon YS, et al. An intraatrial thrombus and pulmonarythromboembolism as a late complication of percutaneous vertebroplasty.Anesth Analg 2007;104(4):924–6.

10] Wu CC, Lin MH, Yang SH, et al. Surgical removal of extravasated epiduraland neuroforaminal polymethylmethacrylate after percutaneous vertebro-plasty in the thoracic spine. Eur Spine J 2006;(October) [Epub ahead ofprint].

11] Kim JH, Park KS, Yi S, et al. Real-time CT fluoroscopy (CTF)-guided verte-broplasty in osteoporotic spine fractures. Yonsei Med J 2005;46(5):635–42.

12] Neeman Z, Dromi SA, Sarin S, et al. CT fluoroscopy shielding: decreasesin scattered radiation for the patient and operator. J Vasc Interv Radiol2006;17(12):1999–2004.

13] Yeom JS, Kim WJ, Choy WS, et al. Leakage of cement in percutaneoustranspedicular vertebroplasty for painful osteoporotic compression frac-tures. J Bone Joint Surg Br 2003;85(1):83–9.

14] Hierholzer J, Fuchs H, Westphalen K, et al. Percutaneous vertebroplasty—the role of osseous phlebography. Rofo 2005;177(3):386–92.

15] Anxionnat R, Bracard S, Ducrocq X, et al. Intracranial aneurysms: clinicalvalue of 3D digital subtraction angiography in the therapeutic decision andendovascular treatment. Radiology 2001;218:799–808.

16] Van den Berg JC, Moll FL. Three-dimensional rotational angiographyin peripheral endovascular interventions. J Endovasc Ther 2003;10(3):595–600.

17] Cornelis G, Bellet A, van Eygen B, et al. Rotational multiple sequenceroentgenography of intracranial aneurysms. Acta Radiol Diagn (Stockh)1972;13:74–6.

18] Uchida M, Abe T, Nishimura K, et al. Rotational three-dimensionalcholangiography—initial clinical experience. Radiat Med 2002;20(4):213–5.

19] Verlaan JJ, van de Kraats EB, van Walsum T, et al. Three-dimensionalrotational X-ray imaging for spine surgery: a quantitative validation studycomparing reconstructed images with corresponding anatomical sections.Spine 2005;30(5):556–61.

20] Cotten A, Dewatre F, Cortet B, et al. Percutaneous vertebroplasty for oste-olytic metastases and myeloma: effects of the percentage of lesion fillingand the leakage of methyl methacrylate at clinical follow-up. Radiology1996;200(2):525–30.

21] Hiwatashi A, Ohgiya Y, Kakimoto N, et al. Cement leakage during vertebro-plasty can be predicted on preoperative MRI. AJR 2007;188(4):1089–93.

22] Bridcut RR, Murphy E, Workman A, et al. Patient dose from 3D rotationalneurovascular studies. Br J Radiol 2007;(January) [Epub ahead of print].

23] Linsenmaier U, Rock C, Euler E, et al. Three-dimensional CT with a mod-ified C-arm image intensifier: feasibility. Radiology 2002;224(1):286–92.

24] Racadio JM, Fricke BL, Jones BV, et al. Three-dimensional rota-tional angiography of neurovascular lesions in pediatric patients. AJR2006;186(1):75–84.