MARSHALL R. URIST
`Editor-in—Chief
`
`With the Assistance of
`
`THE ASSOCIATE EDITORS
`
`THE BOARD OF ADVISORY EDITORS
`
`THE BOARD OF CORRESPONDING EDITORS
`
`~59
`
`Number Two Hundred Three
`
`
`
`J.»B. LIPPINCOTT COMPANY
`
`Philadelphia
`
`Medtronic,
`
`Inc. MSD lOl3
`
`1
`
`

`
`2
`
`CLINICAL ORTHOPAEDICS AND RELATED RESEARCH
`GUIDELINES FOR AUTHORS
`
`CLINICAL ORTHOPAEDICS AND RELATED RESEARCH is designed for the publication of original articles olfering
`significant contributions to the advancement of orthopedic knowledge. Articles are accepted only for exclusive publication
`in CLINICAL ORTHOPAEDICS AND RELATED RESEARCH; published articles and their illustrations become the property
`of CLINICAL ORTI-IOPAEDICS AND RELATED RESEARCH.
`
`0 T0 COVER COSTS of postage and other overhead ex-
`penses, contributors should send a check for $25.00 (U.S.)
`with each unsolicited manuscript to CLINICAL ORTHOPAEDICS
`AND RELATED RESEARCH, UCLA Bone Research Laboratory,
`'- 1000 Veteran Avenue, Los Angeles, California 90024.
`0 ORGANIZE YOUR MANUSCRIPT as follows: Ab~
`stract; Introduction; Materials and Methods; Results; Dis-
`cussion.
`0 THE ACCURACY OF THE INFORMATION and of
`the citations of literature is the sole responsibility of the au-
`thor(s).
`w AUTHORS ARE NOT ENCOURAGED to submit ar-
`ticles in series, e.g., “Part I: Fractures of the Hip; Part II:
`Complications of Fractures of the Hip; Part III: etc.”
`0 THE ABSTRACT should consist of a one paragraph state-
`ment less than 200 words in length, stating the main conclusions
`without reference to any other part ofthe manuscript. The abstract
`must be independently understood for reproduction by libraries
`and other publications as a separate entity.
`0 ALPHABETIZE AND NUMBER REFERENCES. A
`bibliography of numbered references in alphabetical order
`should be attached to the manuscript. Each reference must
`be cited in the text by number. Follow exact style of references
`published in this copy of CLINICAL ORTHOPAEDICS AND RE-
`LATED RESEARCH. Each reference must list all co-authors.
`0 ALL TABULAR MATERIAL OR STATISTICAL
`DATA must be in the form of clear tables, placed one to a
`page, at the end of the manuscript. Tables must be titled,
`numbered, and cited in the text in numerical order.
`0 INCLUDE A SHEET OF LEGENDS DESCRIBING
`ILLUSTRATIONS. A separate sheet of typed legends (cap-
`tions) describing each illustration should be attached to the
`manuscript.
`0 SEND FOUR SETS OF BLACK AND WHITE GLOSSY
`PRINTS approximately 5" X 7" for all figures, for reviewers’ use.
`Original drawings and X-rays must be of professional quality
`and photographed as black and white glossy prints. Do not write
`on the face of illustrations; put as much information as possible
`in the legends. Ifarrows or letters are necessary, use a professional
`product such as I./etraset or Paratype. Attach a label to the backs
`of illustrations indicating “top,” figure number, and author’s
`name, or enter this information lightly with a pencil. Each figure
`must be cited in the appropriate place in the text in numerical
`order. Submit the minimum number of figures and tables nec-
`essary to highlight the article. The publisher reserves the right
`to charge for illustrations and tables in excess of one page of
`illustrations, including tabular material, per four pages ofprinted
`text.
`
`0 LETTERS OF PERMISSION from the publisher of
`any borrowed illustration or table should be enclosed with
`the manuscript. Acknowledgment of the sources must also
`be included at the end of the legend for borrowed illustrations
`and as a footnote to the table of any borrowed tabular material.
`
`- SEND THE ORIGINAL TYPESCRIPT AND TWO
`DUPLICATES. Manuscripts must be in English, typed dou-
`ble-spaced on the one side of 8*/2" X 11” paper with wide
`margins and pages numbered. Authors should retain copies
`of manuscripts and illustrations. Page 1 of the manuscript
`should be a cover (title) page listing the title of the article
`and authors’ names and degrees, with footnotes including
`complete addresses and professional afliliations.
`
`0 SEND A LETTER OF TRANSMITTAL SIGNED BY
`ALL AUTHORS containing the following paragraph: “In
`consideration of CLINICAL ORTHOPAEDICS AND RELATED
`RESEARCH reviewing and editing my (our) submission, the
`author(s) undersigned hereby transfer(s), assign(s), or otherwise '
`convey(s) all copyright ownership to CLINICAL ORTHOPAEDICS
`AND RELATED RESEARCH, and represent(s) that he(they)
`own(s) all rights in the material submitted. This assignment
`is to take effect only in the event that such work is published
`in CLINICAL ORTHOPAEDICS AND RELATED RESEARCH.”
`
`0 CLINICAL ORTHOPAEDICS AND RELATED RE-
`SEARCH will expedite publication of selected manuscripts.
`To qualify for accelerated publication, the work must be judged
`to be of exceptional significance and immediate interest. Such
`manuscripts must be labelled “FOR EXPEDITED PUBLI-
`CATION,” contain the elements of a regular manuscript, and
`satisfy the criteria governing the publication of papers in
`CO&RR, but must not exceed 12 manuscript pages (excluding
`illustrations, tables, and references). The manuscript must be
`accompanied by a transmittal letter giving the reasons why
`the authors believe the work justifies expedited publication
`and the names of at least three knowledgeable surgeons who ‘
`will immediately return the manuscript with an impartial,
`detailed review. Except for minor revisions, such manuscripts
`will be either accepted or rejected within four weeks. The
`aim is to publish an expedited paper within three to six months
`after submission. The number of expedited publications is
`limited to one per issue.
`
`0 REPRINTS. The publisher will provide each senior au-
`thor, free ofcharge, with 50 reprints of his article. Ifadditional
`reprints are desired, they may be ordered on the order form .
`furnished with page proofs.
`
`Following are the Subjects of Symposia to be published in forthcoming numbers:
`
`Total Knee Arthroplasty; Long-Term Results in Chemonucleolysis
`
`Copyright © 1986, BY J. B. LIPPINCOTT COMPANY
`
`Library of Congress Catalog Card Number 53-7647
`
`CLINICAL ORTHOPAEDICS AND RELATED RESEARCH (ISSN 0009-921x) is published monthly by J. B. Lippincott Company, East Washington Square, Philadelphia.
`PA 19105. Printed in the U.S.A. © Copyright 1986 by J. B. Lippincott Company. All rights reserved. Second class postage paid at Hagerstown, MD, and at additional
`mailing oliices.
`Subscription information, orders or changes of address: (except Japan) 2350 Virginia Avenue, Hagerstown, MD 21740, or call l~800-638-3030; in Maryland, call
`collect 301-824-7300. In Japan, Contact Woodbell Scope, Mansui Bldg., 9-18, Kanda-Surugadai Z-chome, Chiyoda-lcu, Tokyo 101, Japan. In India, Nepal, Bangladesh,
`and Sri Lanka, contact Universal Subscription Agency Pvt. Ltd.. l0l—l02 Community Centre (F.F.) Saket, New Delhi—l 10017, India.
`Annual subscription rates: US. $145.00 individual, $195.00 institution; all other countries except Japan, India, Nepal, Bangladesh, and Sri Lanka, 8 172.00 individual,
`$222.00 institution. Resident rate, U.S.: $95.00, Canada $122.00. Single copies $25.00. Rates for airmail delivery available upon request. Copies will be replaced
`without charge if the publisher receives a request within 60 days of the mailing date in the U.S. or within 5 months in all other countries.
`POSTMASTER: Send address changes to CLINICAL ORTHOPAEDKB AND RELATED RESEARCH, 2350 Virginia Avenue, Hngerslown, MD 21740.
`
`2
`
`

`
`An internal Fixatorfor Posterior Application to
`
`Short Segments of the Thoracic, Lumbar,
`
`or Lumbosacral Spine
`
`Design and Testing
`
`MARTIN H. KRAG, M.D.,* BRUCE D. BEYNNON, M.S.,* MALCOLM H. POPE, PH.D.,*
`JOHN W. FRYMOYER, M.D.,* LARRY D. HAUGH, PH.D.,** AND
`DONALD L. WEAVER, M.D.’r
`
`A new spinal implant has been designed and bio-
`mechanical testing completed, intended for appli-
`cation to “short-segment" spinal defects such as
`disc degeneration, fracture, spondylolisthesis, or
`tumor. Major improvements over currently available a’
`devices include: (1) only 2-3 vertebrae are spanned,
`not 5-7 as with Harrington rods; (2) true three-
`dimensional fixation is achieved, preventing such
`problems as hook or rod dislocation; (3) three-di-
`mensional adjustment is easily accomplished, al-
`lowing fracture or spondylolisthesis reduction to be
`readily performed; (4) attachment to vertebrae is
`by means of transpedicular screws eliminating de-
`liberate encroachment into the spinal canal, such
`as Luque wires or Harrington hooks; (5) no special
`alignment between screws is needed (such as with
`holes or slots in a plate), allowing screw placement
`to fully conform to anatomic structures; and (6)
`laminectomy sites and lumbosacral junction are
`readily instrumented. Background investigations
`presented here for design of this device include: (1)
`CT-defined pedicle morphometry showing that
`screws may be larger than those currently used; (2)
`effect of pitch, minor diameter, and tooth profile on
`screw pull-out strength; (3) mechanical testing of a
`compact,
`three—dimensionally adjustable, strong,
`nonloosening articulating clamp; and (4) establish-
`
`From the University of Vermont, Burlington, Vermont.
`* Department of Orthopaedics and Rehabilitation.
`** Department of Math and Statistics.
`1' Department of Pathology.
`Supported in part by the National Institute of Handi-
`capped Research, Rehabilitation Center for Low Back
`Pain, Grant No. USOEG 008303001.
`Reprint requests to: Martin H. Krag, M.D., Department
`of Orthopaedics and Rehabilitation, College of Medicine,
`University of Vermont, Burlington, VT 05405.
`' Received: June 5, 1985.
`
`ing of the relationship between depth of penetration
`and strength of fixation of transpeduncular screws.
`
`Significant improvements remain possible
`in the design of surgical implants for posterior
`spinal stabilization, to be used for dealing with
`single motion-segment instability in the tho-
`racic, lumbar or lumbosacral level (such as
`fracture, spondylolisthesis, segmental
`insta-
`bility). This is true, even though major ad-
`vances have been made in this field in recent
`
`years,‘3"°9"“’62"16440 biomechanics have been
`increasingly applied to spine problems6* 1 7’2°’3“’-
`36, 37, 50, 53, 65, ea, 71, 73, 75, 31.33, as, 94, 96, 92, 99, 103, l06,_
`107, 118, 122, 132,133, 136, 137,141,149, 150, 153,154, 158,l59,_
`
`‘63*‘54=‘66 and a growing variety of spinal im-
`plants have become available.5"2"5"9’35’5 "5470-
`86, 87, 91, 92. 100, H0, H4, H5, H9, 121, 124, 126, 128,_
`l38,l39,l46.152
`
`Bearing in mind that a single, ideal implant
`is probably undefinable, the authors have de-
`signed a substantially-different posterior im-
`plant (for convenience, referred to as the Ver-
`mont Spinal Fixator or VSF). Presented here
`is the rationale for its design, new anatomic
`data related to it, biomechanical testing of it,
`and initial experience with its cadaver im-
`plantation.
`
`MINIMUM LENGTH OF
`SPINAL INVOLVEMENT
`
`When fusion is indicated for single-level in- 5'
`stability, only two vertebrae in principle need
`to be incorporated in the fusion mass. Even
`
`75
`
`3
`
`

`
`76
`
`Krag et al.
`
`in the case of instability from a severely-com-
`minuted fracture, only three vertebrae need
`be fused. From a mechanical viewpoint, a
`short fusion is not strengthened by adding to
`its length (a one-foot length of chain is as
`strong as a three-foot length of chain). From
`a biologic viewpoint, unnecessary disruption
`of normal tissue should certainly be avoided.
`Despite this, five, six, or even seven verte-
`brae are typically involved when using Har-
`rington distraction rods, the most commonly
`used implant for dealing with instability, used
`either in their standard configuration‘*“*8'2"22*
`30.32.}3.4-3.44,55,57,66.77,95,97.l l 1, I 30, l 35, I 43, I 56, l 60, l 65 or
`nlodificationsd l,16,l8,3 l ,34,38-40,42,45,48,65,72,
`lOO,l04,l34,i42,l44,l45,l64
`is related to the hiS_
`
`tory of Harrington rods,” which were devised
`for management of multiple-segment defor-
`mities, such as scoliosis, and then subsequently
`applied to single-level problems (instability,
`spondylolisthesis, fracture).
`However, mechanical characteristics of sin-
`gle-level problems are quite different, thus an
`implant specifically designed for this appli-
`cation is needed. Placement of Harrington
`hooks more closely together than five vertebrae
`fails to achieve adequate stabilization. This
`same dependency on five or more segments
`occurs with posterior plates and screws, such
`as Wi1liams,‘56‘”" Wilson,5°'84"°2 or Roy-
`Camille.6°’86*”“*‘26*“" One method of decreas-
`ing the fusion, length is the “rod long, fuse
`short” technique,4'2"°3"3“ in which the rods are
`removed after the graft is solid. Not only does
`this involve a second operation, but early
`evidence“’69 suggests that facet arthrosis may
`become a problem at the levels temporarily
`immobilized but not grafted.
`
`THREE—DlMENSIONAL FIXATION
`
`One major purpose of any fixation device
`is to increase the likelihood of achieving suc-
`cessful bone fusion. There is no data to suggest
`that motion in any particular direction is better
`or worse than" in any other, in terms of achiev-
`ing fusion. Thus, fully three—dimensiona1 (3-
`D) fixation becomes a logical objective. A sec-
`ond purpose of a fixation device is to limit
`
`Clinical Orthopaedics
`and Related Research
`
`intervertebral motion so as to prevent nerve
`root or spinal cord pressure. This may poten-
`tially be produced by any one of various dis-
`placements (either rotations or translations),
`so again 3-D fixation becomes the objective.
`Despite this, most current spinal posterior
`implants do not produce such 3-D fixation.
`This occurs for one or more of three reasons:
`
`(A) absence of a rigid attachment to the ver-
`tebra itself; (B) reliance upon soft-tissues to
`not stretch out or “creepg” or (C) absence of
`rigid attachment between components of the
`device.
`
`_ The most commonly used spinal implant,
`the Harrington rod system, has all three of the
`above characteristics. The hooks simply push
`(or pull) against their attachment sites, and
`they rely upon soft-tissue resistance2’3’8 as well
`as external trunk support (bracing) to produce
`sufficient vertebral-motion limitation. As a
`
`flattening in the lumbar
`lordosis
`result,
`spine59"°° or overdistraction at a fracture
`site2*3'8 can occur. Once displacement occurs,
`the hooks may detach from the rod. Modified
`hooks""“’66 or segmental wiring34*87’”“ rep-
`resent significant improvements, but flexion/
`extension can still occur by means of the lam-
`ina rotating within the hook or wire loop. This
`lack of 3-D fixation at the attachment sites of
`
`F
`
`Harrington hooks (or Luque wires, for that
`matter), is a part of the reason that these im-
`plants must span five, six, or even seven ver-
`tebrae for adequate stability.
`With posterior plates and screws, the screw
`does provide a rigid grip on the vertebra.
`However, the screw is not rigidly mechani-
`cally linked to the plate. Rather, sufficient
`:/
`compression between plate and underlying-
`bone is needed to prevent “toggle” of the screw
`in its plate hole‘33 or excessive shearing forces
`on the screw. The occurrence of this toggle is
`probably related to the fall-off in screw tension
`known to occur in viva. ‘O This issue is one of
`
`concern even when these plates are mounted
`onto broad surfaces such as the femur or
`
`tibia.“ The problem becomes even greater
`when the plates are attached to the spine where
`the bone bed is quite irregular.“°""3 This results
`in relatively small bone—plate contact areas
`
`:/
`
`4
`
`

`
`Number 203
`February, 1986
`
`and thus high contact pressures, increasing the
`likelihood of bending or shear loads being ap-
`plied to the screws. These plates, of course,
`overlay a significant portion of the surface
`normally used as a graft bed.‘63 As with the
`rods, here also, five vertebrae must be instru-
`mented for adequate control of significant in-
`stability. ‘24’ 126
`The Magerl external fixator
`obtain a rigid grip on each vertebra, does not
`rely upon soft tissues, and its components are
`rigidly linked together in three dimensions.
`Thus, this device represents a major advance
`in spinal implants. It does have the character-
`istics of any external fixator in that pin-track
`infections may occur and the posterior prom-
`inence of the device is an inconvenience. The
`
`89,90,l32,l63 does
`
`flexibility due to the unsupported span of the
`Schanz pins has been conjectured to provide
`a certain “shock—absorbing” quality. However,
`this same characteristic also requires supple-
`mental
`internal fixation in more unstable
`
`cases, limiting the case in which the pins may
`be applied percutaneously. In addition, this
`flexibility often requires the device to be used
`as either a distractor or a compressor, rather
`than as a fixator. Thus it relies upon soft tissues
`being sufficiently intact to prevent excessive
`creep.
`
`SPINAL CANAL AVOIDANCE AND
`SAFE IMPLANTATION
`
`into the spinal
`Deliberate encroachment
`canal is a routine part of most device implan-
`tation, either with hooks (e.g., Harrington,
`Weiss, Knodt) or wires (e.g., Luque) or both.
`Use of hooks alone has been quite safe, al-
`though problems do oecur,52'93 and there is
`the always-present risk of intraoperative errors
`when working within the spinal canal. The use
`of laminar wires has caused some major com-
`plications either during their placement67’”3~ ‘ 57
`or removal,9”°3 or when used in combination
`with Harrington hooks.'23“57
`All of these complications are related to de-
`liberate violation of the spinal canal necessary
`
`for implantation of these devices. Pedicle screw
`placement, however, does not require entrance
`into the canal. Extensive experience with the
`
`Internal Fixator for Posterior Application
`
`77
`
`safety of this method has been gathered, either
`with plates, external fixators, or facet joint fu-
`sions.‘‘‘'7‘‘‘' ‘2 The risks of screw placement too
`far medially or too far anteriorly do exist, but 2’
`modern image intensifiers allow good intra-
`operative visualization to guard against both
`these risks. A further safeguard to keeping the
`screw within the pedicle is the fact that the
`medial and inferior pedicle borders may
`be easily and safely palpated intraopera-
`tiVe1y_9I,92,I24
`Concerning the issue of anterior cortex en-
`gagement by the pedicle screw, it appears that
`this is unnecessary. Deliberate avoidance of
`anterior cortex is recommended by. Roy-
`Camille,'2"'2“ based on his extensive experi-
`ence in which screw bone interface loosening
`apparently is not a problem. Supportive evi-
`dence exists in the mechanical testing by La-
`which showed that engagement of »~’
`vaste,82'33
`the anterior cortex provides only slight addi-
`tional screw-pull-out strengthening. The issue
`of cortex engagement, as well as a number of
`other
`important aspects of pedicle screw
`placement, have been addressed by Zindrick
`et al., 156 using certain commercially-available
`screw types.
`
`USE OF SAFEST SURGICAL APPROACH
`
`The choice between anterior and posterior
`approach devices depends upon many factors,
`but certainly there are advantages to avoiding
`abdominal or thoracic cavity involvement. For
`cases in which the anterior approach is selected
`primarily in order to perform spinal canal de-
`compression, a variety of stabilizing devices
`are becoming available.‘2'35'5"5“’7°~"5*"9"2‘*
`‘24"26>‘28 These involve only a limited segment
`of spine and do not require spinal canal en-
`croachment. At least one of them” provides
`full three-dimensional rigid fixation, although
`anterior prominence and proximity to the
`aorta is an issue about which some concern
`exists.
`
`DEVICE REMOVAL
`
`Although Harrington rods are not routinely
`removed when accompanied by bone grafting
`
`5
`
`

`
`78
`
`Krag et al.
`
`along the entire length of the rod, their upper
`hooks or cut ends may cause sufficient irrita-
`tion to require removal. The “rod long, fuse
`short” technique“'2“63"“ described above does
`of course require device removal. The benefit
`to avoidance of a second operation for device
`removal is certainly obvious.
`
`AVOIDANCE OF TRANSCUTANEOUS
`COMPONENTS
`
`Preliminary experience with an external
`spinal fixator is growing.89‘92 This device has
`certain unique and useful advantages, but it
`does present the risk of pin-track infections,
`which limits the length of time the device can
`remain in place. Furthermore, the external
`components appear to be somewhat Cumber-
`some, requiring a special brace and mattress.
`In order to simultaneously satisfy all six of
`the above criterion, the authors approach has
`been to devise an “internal fixator” that (1)
`could be adjusted to span as short as a single
`motion segment, (2) provides rigid 3-D fixa-
`tion by attachment through the pedicle into
`the vertebral body, (3) does not violate the
`spinal canal,
`(4) utilizes the posterior ap-
`proach, (5) does not require removal, and (6)
`is completely internalized.
`
`DESIGN APPROACH
`
`In attempting to satisfy these six criteria,
`five major design issues were identified for
`further investigation.
`
`IN VIVO LOADS
`
`Little data are available concerning the loads
`which the implant needs to support. Schlaepfer
`and co-workers’32*’63 have presented their re-
`sults using an external spinal fixator as a load
`transducer, but the data are not fully three-
`dimensional and do not allow full separation
`of the various load components. This impor-
`tant work, however, does suggest that the im-
`plant is very largely “shielded” from bending
`loads by the trunk extensor muscles. Other
`
`in, vivo measurements have been made
`using strain gages on Harrington rods in
`
`Clinical Orthopaedics
`and Related Research
`
`humans‘°5*‘2°*‘5' and in sheep,‘°6 or on Dwyer
`cables in dogs, ‘36 but these are diflicult to con-
`vert into loads acting at the site of instability.
`“Free-body analysis” estimates of in vivo loads
`are only as good as the estimates upon which
`they are based, and do not deal at all with load
`sharing between vertebrae and muscles.
`Thus, for present purposes, the best answer
`to the question “how strong should the im-
`plant be?” comes from the empiric clinical ex-
`perience that has been accumulated in five
`areas. First, posterior plates are typically
`attached'24"2""27 using 3.5-4.5-mm cortical
`screws. Although ideally these screws are pro-
`tected from all but pure tensile loads, in prac-
`tice this seems to be quite unlikely, especially
`for the screws at the ends of the plates. As
`noted earlier, these screws are almost surely
`exposed to some shearing and bending loads.
`Despite this, screw breakage has not been re-
`ported as a significant problem.‘24"26 Thus,
`whatever the in viva loads actually are, 3.5-
`4.5-mm screws are strong enough to prevent
`breakage.
`in vivo bending of the plates
`Secondly,
`themselves has not been reported to be a
`problem. Mechanical testing in vilr083'83 has
`shown that plastic deformation of the Roy-
`Camille plates occurs at only 1 1.3-nm (8.3 ft-
`lbs). For comparison, this is even weaker than
`the bending strength (14.7 nm or 10.85 ft—lbs)
`of the 5-mm portion of the Schanz pins used
`in the external spinal fixator (see below). Thus,
`in viva bending loads taken by the plate must r’
`be less than 11.3 nm.
`
`Thirdly, Cyron ‘and co-workers” have
`shown in vitro that spondylolysis can be pro-
`duced with a mean moment of 35 nm for L5
`vertebrae and 28 nm for L, vertebrae. These
`must represent upper limits to in vivo mo-
`ments, since spondylolysis does not routinely
`develop after spinal injuries, even with com-
`plete paraplegia in which trunk muscle de-
`nervation may occur.
`Fourthly, significant experience has been
`reported for facet joint fusions with a screw
`placed obliquely across the facet joint in con-
`junction with posterior bone graft for various
`nontraumatic conditions. Boucher” encoun-
`
`6
`
`

`
`Number 203
`February, 1986
`
`Internal Fixator for Posterior Application
`
`79
`
`tered only two broken screws out of a total of
`482. Out of 44 L5-S, fusions, King” does not
`describe any breakages. In the 150 patients of
`Pennal er al.,‘ '2 only one screw broke. In the
`first two studies, screw diameters were not
`specified, but were probably approximately ‘/3
`in. In the last study, screw minor diameter was
`1/3 in.
`
`Fifthly, the external spinal fixator9"92”32"33
`utilizes 6—mm Schanz pins thinned down to
`5—mm along their anterior 6-cm. These pins,
`of course, are fully exposed to all the loads
`taken by the fixator. Breakage or bending“ of
`these pins has not been reported. This should
`probably not be surprising, since the bending
`strength (load needed to produce plastic de-
`formation) of the 5-mm portion of pin is 14.7
`nm per pin or 29.4 nm per pair. Thus it can
`be seen empirically that 5 mm certainly seems
`to be strong enough. If an even larger size
`could be used, the margin of safety would only
`increase.
`
`VERTEBRAL MORPHOMETRIC
`CONSTRAINTS
`
`The pedicle seems to be the strongest site
`accessible posteriorly through which to obtain
`a three-dimensionally rigid “grip” onto the
`vertebra. Certainly, no other site with this
`property seems to have been proposed. The
`limiting factor to the size of the screw that can
`be placed from posteriorly through the pedicle
`into the vertebral body is the mediolateral
`width of the pedicle. Saillantm has reported
`certain important data from cadavers, but
`these data have certain drawbacks. Firstly, only
`average values were given and not the ranges
`or standard deviations. Secondly, bone-screw
`path length was only reported for a purely sag-
`ittal screw placement: other screw placement
`angles may be preferable9"97' and would alter
`this length. Thirdly, only the pedicle diameter
`perpendicular to the pedicle axis was reported:
`for screws placed at any other direction than
`along the pedicle axis, an effectively smaller
`pedicle diameter may be present (Fig. 1D). Fi-
`nally, data were obtained from cadavers alone,
`without any radiographic correlation, the latter
`
`
`
`FIGS. lA~lD. Construction lines used to obtain ‘
`measurements from vertebral CT scans. (A) Ver-
`tebral body length from anterior cortex to line A,
`pedicle length from line A to posterior cortex or to
`line B (“facet corrected”). (B) Pedicle axis angle
`measured from sagittal plane (anteromedial posi-
`tive), pedicle diameter along a perpendicular to axis.
`(C) Screw-path length (or chord length) from an-
`terior to posterior cortex at 0°, 5°, 10°, or l5° pos-
`terolaterally from the sagittal plane. (D) Pedicle di-
`ameter at 0° and 15°: note that bone Contact points
`do not fall along a common perpendicular to the
`axis.
`
`being more appropriate to the clinical situa-
`tion. Thus, a morphometric study addressing
`these issues was undertaken. The major find-
`ings are presented below, and full details are
`reported elsewhere.8°
`
`PEDICLE SCREW DESIGN
`
`Bone—~screw interface strength is commonly
`the limiting factor in the overall strength of a
`stabilizing implant, at least over the first few
`days or weeks (fatigue of metal or resorption
`of bone may become a problem later on).
`Some testing of mechanical characteristics of
`pedicle screws has been performed. Lavaste32’83
`compared
`various
`commercially-available
`screws in pull-out tests. Zindrick er al. ‘66 com-
`pared certain commercially-available screws,
`and also the effects of various details of screw
`
`placement technique.
`Optimizing pedicle screw pull-out strength
`requires a systematic study in which various
`screw design features are varied systematically.
`This has not previously been reported for ped-
`
`7
`
`

`
`80
`
`Krag et al.
`
`icle screws. Furthermore, despite aifairly large
`literature characterizing the pull-out strength
`of various screw designs in limb bones, it does
`not appear that a systematic study has been
`_ done that independently varies the different
`screw design features such as tooth profile,
`pitch, and minor diameter. Bechto17 compared
`pu1l—out strength from dog limb bones of
`screws with one each of eight different tooth
`profiles, but the minor diameters were un-
`specified and various pitches were used in such
`a way as to not allow the effect of this variable
`to be isolated. Koranyi er a[.75 reported equal
`pull-out strengths for both “V” toothed Sher-
`man screws and buttress-toothed AO screws,
`using dog or cattle femora. However, neither
`major nor minor diameters were specified al-
`though tooth heights were the same. Lyon er
`a[.,88 Nunamaker and Perren, ‘O9 and Schatzker
`et al. ‘3‘ each studied various groups of different
`commercially-available screws, but the indi-
`vidual
`effects of pitch, major diameter,
`and minor diameter could not be segregated,
`since these parameters were not systematically
`varied.
`
`In order to design a pedicle screw with op-
`timal bone—metal interface strength, the au-
`thors undertook the study reported here (re-
`ported in more detail elsewhere"), using var-
`ious combinations of pitch, minor diameter,
`and tooth profile, for each of various major
`diameters. Pull-out testing was utilized, since
`this load type would be most sensitive to
`thread-design variations. Of course, pure pull-
`out loads alone are not likely to occur in vivo,
`since additional kinds of loads (bending,
`shearing) would usually occur simultaneously.
`
`ARTICULATING CLAMP
`
`Some sort of mechanism is needed to rigidly
`link together the four pedicle screws after they
`are placed into the vertebra above and the ver-
`
`tebra below the site of instability. The four
`most important design objectives were felt to
`be adjustability, strength, compactness, and
`security. Adjustability in all three dimensions
`was sought, since this would simplify pedicle
`
`Clinical Orthopaedics
`and Related Research
`
`screw insertion: no special alignment between
`the screws would need to be maintained during
`their insertion. Adjustability in 3-D also allows
`the reduction (in the case of fractures or spon-
`dylolisthesis) to be “unconstrained” and can
`be performed in a controlled fashion with the
`fixator already in place (but before tightening
`the locking mechanism). The strength of this
`articulating clamp should exceed that of the
`pedicle screw, so as not to become the limiting
`factor to overall implant strength. Compact-
`ness is obviously important for comfort and
`for normal muscle function. Finally, “secu-
`rity” means that the likelihood for loosening
`‘be extremely low.
`.
`To prevent loosening, the threads which the
`clamp bolt engage inside the rod clamp are of
`a special pattern known as Spiralock (Kaynar:
`A MicroDot Company, Fullerton, California).
`This “state-of-the-art” thread is primarily used
`in aircraft and other critical high—vibration
`applications. Other advantages of this thread
`besides security against loosening include (1)
`it may be repeatedly tightened, loosened, and
`retightened without degradation; (2) it has a
`much better distribution of loads along the
`engaged bolt threads compared to standard
`threads; and (3) a separate locking nut is not
`needed.
`
`In order to satisfy all four of these criteria,
`a series of clamping systems were designed and
`tested by mock-up cadaver implantation. The
`final articulating clamp type is that shown in
`Figures 2 and 3. Note that fine stepwise ad-
`justability exists for rotation about
`the x
`(transverse) axis, in increments of 6°, since
`there are 60 radially-arranged teeth on the
`“face gear” on the head of the bone screw (only
`36 teeth are shown in a preliminary version,
`shown in Figures 2 and 3). Infinite adjusta-
`bility exists for longitudinal axis rotation (Ry)
`and longitudinal
`translation or
`lengthen-
`ing (Ty).
`
`DEPTH OF SCREW PENETRATION
`. INTO VERTEBRA
`
`How close to the anterior cortex should the
`
`tip of the pedicle screw be placed? The greater
`
`8
`
`

`
`Number 203
`February. 1986
`
`Internal Fixator for Posterior Application
`
`81
`
`the depth of penetration (Fig. 4), the more se-
`cure the screw “grip” on bone, but the greater
`the risk of cortical breakthrough and damage
`to aorta or other structures.
`
`Magerl” recommends placement of the
`screw tip just into but not through the anterior
`cortex. Direct testing of various depths of
`placement, however, does not appear to have
`been done. Roy-Camille,”4'”°"27 on the con-
`trary, recommends avoidance of anterior cor-
`tex engagement. His clinical reports do not
`describe in detail the depth of penetration ac-
`tually used, but illustrative roentgenograms
`show penetration of approximately 50%—60%
`depth. Screw loosening has not been reported
`as a significant problem. Mechanical testing
`in vitro by Lavaste32'83 suggested that anterior
`cortex engagement did not add significantly
`to the pull-out strength of the screws. It appears
`from this experience that even a close ap-
`proach to the anterior cortex is not necessary.
`Because of these conflicting recommenda-
`tions concerning depth of screw penetration,
`specific investigation of this issue seemed in-
`dicated and is reported here.
`
`METHODS
`
`VERTEBRAL MORPHOMETRY
`
`A retrospective review was performed of com-
`puted tomography (CT) scans of 91 vertebrae from
`
`
`
`FIG. 2. Articulating clamp assembly sh