The Knee 11 (2004) 197–201

Varusyvalgus alignment of the femur in total knee arthroplasty.

Can accuracy be improved by pre-operative CT scanning?
Vera Kinzela, Matthew Scaddana, Beverley Bradleyb, David Shakespearea,*
a
Department of Orthopaedic Surgery, Warwick Hospital, Warwick, UK
b
Department of Radiology, Warwick Hospital, Warwick, UK

Received 15 May 2003; received in revised form 27 May 2003; accepted 16 June 2003

Abstract

In a series of 80 consecutive knee replacements, pre-operative scout scans of the femur were taken and the angle between the
slope of the worn femur and the mechanical axis was measured. At surgery the angle of the distal femoral cut was adjusted to
make it perpendicular to the mechanical axis. Post-operative scans revealed that 97.5% of femoral components were implanted
within 28 of the target. The angle between a perfectly positioned virtual intra-medullary rod on the pre-operative films and the
mechanical axis showed marked variations between individuals indicating that the use of a fixed angular correction from an intra-
medullary rod is unreliable even under perfect circumstances. In practice it is subject to further inaccuracy due to variability in
the anatomy of the femur and rod placement. With the advent of robotically assisted surgery, the technical limits of conventional
instruments need to be explored in terms of improving accuracy for each individual patient. In terms of the coronal alignment of
the femur, this involves some degree of preplanning rather than relying on arbitrary targets based on incorrect assumptions.
䊚 2003 Elsevier B.V. All rights reserved.

Keywords: Total knee arthroplasty; Varus; Valgus

1. Introduction implants were within 38 of this target value and 81%

The aim in total knee arthroplasty is to place the
femoral component perpendicular to the mechanical axis
of the leg in the coronal plane. Failure to do so may
adversely affect implant survival w1–6x. Unfortunately
the exact position of the centre of the hip in three
dimensions remains extremely difficult to define in any
individual patient. Most instruments rely therefore on
indirect methods of obtaining correct alignment using
an intra-medullary rod. The choice of the angle of the
cut relative to the rod is often arbitrary. Extramedullary
guidance, even using a skin marker over the hip is
subject to rotational error. In our own previously report-
ed series of 362 knees w7x, we assumed that an intra-
medullary rod was in 58 of varus relative to the
mechanical axis of the leg in most individuals, as
demonstrated by Harding et al. w8x. Only 92% of
*Corresponding author. Department of Orthopaedics, Warwick
Hospital, Lakin Road, Warwick CV34 5BH, UK. Tel.yfax: q44-
1926-422-119.
E-mail address: davidshakespeare@uk-consultants.co.uk
(D. Shakespeare).
within 28 of target w8x. Although this compares favour-
ably with other series w9–11x, it is far from ideal and
presupposes that the 58 angle was correct in the first
place. Contributory factors to this inaccuracy include
valgus bowing of the distal femur, failure of the intra-
medullary rod to pass exactly up the femur and incon-
sistency of entry point into the femur w12,13x.
We set out to determine whether we could improve
this accuracy by using a pre-operative CT scout scan of
the femur. Direct measurement of the angle between the
slope of the distal femur and the mechanical axis would
allow appropriate correction at the time of surgery.
Although the slope of the natural undamaged joint line
is variable in most individuals, it usually amounts to 38
of varus relative to the mechanical axis. We anticipated
therefore that in medial disease, as bone is lost from the
medial condyle, little intra-operative correction would
be required. Conversely, in lateral disease with bone
loss from the lateral femoral condyle considerable cor-
rection would be required to cut the femur perpendicular
to the mechanical axis. In those knees with severe bi-

0968-0160/04/$ - see front matter 䊚 2003 Elsevier B.V. All rights reserved.
doi:10.1016/S0968-0160Ž03.00106-6
198 V. Kinzel et al. / The Knee 11 (2004) 197–201
compartmental disease we anticipated that the correction
would be unpredictable.
2. Materials and methods
2.1. Patient selection
Between May 2002 and March 2003 a consecutive
series of 80 patients undergoing primary knee replace-
ment were entered into the study.
2.2. Radiological evaluation
Pre- and post-operative CT scans of the entire femur
were performed. Correct rotational alignment for the
pre-operative scans was assessed by palpation of the
femoral epicondyles and ensuring that the epicondylar
axis was aligned parallel with the X-ray table. The first
20 patients were scanned both supine and prone to see
if this affected the measurements and 5 were scanned
in 108 of both internal and external rotation to determine
whether this influenced the measured slope of the joint
line as suggested by Lonner et al. w14x.
Post-operatively rotational alignment was assessed by
the symmetrical outline of the femoral component.
The images were both printed on a 43=35 cm2 film
and stored on optical disk. We measured the scans using
both the computer software and directly from the hard
copies. We felt the latter technique would be more
useful in the clinical setting. Using a transparent sheet
over the X-ray film, both the centre of the hip and the
midpoint between the femoral epicondyles were marked
to define the mechanical axis. The most distal projection
of both medial and lateral condyles was then marked
and the angle between the joint line and the mechanical
axis was measured directly. A magnifying glass was
used to ensure accurate placement of the marker points.
In order to avoid confusion with left and right knees the
angle subtended on the medial side of the knee was
measured in each case. The intra-operative correction
required to place the component correctly was calculated
from this. A varus correction was designated as positive
and a valgus correction as negative.
In order to investigate the angle between a perfectly
placed infra-medullary rod and the mechanical axis, a
virtual rod was laid over the pre-operative image along
the centre of the intra-medullary canal and the angle
was measured directly in all 80 cases.
2.3. Assessment of flexionyextension of the femoral
component
As an intra-medullary rod was not employed, we felt
it was important to record the position of the femoral
component in the sagittal plane. As part of the post-
operative assessment, all patients underwent a long
Fig. 1. The jig is fitted flush with the distal femur and secured by
passing the intra-medullary drill by 3–4 cm. Flexion and extension
were controlled by siting from the anterior femoral cortex.
lateral X-ray of the knee. Rotation of the component
was assessed relative to the posterior surface of the
distal femur as described by Rees et al. w15x.
2.4. Inter- and intra-observer error
A series of 10 scans were measured 3 times by 5
observers and the results were compared.
2.5. Surgical technique
All prostheses were Medial Pivot (Wright Medial
Technology, Arlington, TN) total knee replacements
inserted by or under the guidance of the senior author,
using distal cut first instrumentation.
Remaining articular cartilage was pared from the
distal femur as this was not visible on the CT scan, but
would have influenced the alignment of the jig on the
distal femur. The distal femoral cutting jig was then
applied to the end of the femur flush with the condyles.
In order not to incur the additional expense of new
instrumentation, the jig was fixed using the standard
intra-medullary drill, but this was inserted only far
enough to obtain secure fixation. The correct position
of the cutting block in terms of flexion and extension
was obtained by sliding a right-angled instrument along
the anterior surface of the femur and using this as a
siting post (Fig. 1). The appropriate correction was then
dialled into the cutting block and this was pinned to the
anterior surface of the femur (Fig. 2). After resecting
the distal femur, no further alterations were made to the
operative technique.
3. Results
3.1. Prone vs. supine scans
Seventeen of the 20 patients undergoing both prone
and supine scans had identical measurements. The
 V. Kinzel et al. / The Knee 11 (2004) 197–201 199

in Fig. 3. Those knees with the marked pre-operative

valgus deformity required the greatest varus correction.
On the post-operative scans, 78 (97.5%) of femoral
components were within 28 of the mechanical axis in
the coronal plane. The remaining 2 were only 38 out
(Fig. 4). This represents a much greater degree of
accuracy than we achieved using an intra-medullary rod.

3.5. The virtual rod

The alignment of a perfectly placed virtual rod relative

to the mechanical axis on the pre-operative scans shows
a high degree of individual variation (Fig. 5). Under
ideal circumstances in 90% of knees the alignment is
within 28 of the mechanical axis. This is highly unlikely
to be reproduced under clinical conditions because of
variability in placement of the rod.

3.6. Flexionyextension of the femoral component

The mean angle of flexion relative to the posterior

femoral surface was y1.58 (S.D. y3.7; range 58 to
y158). Initially there was some tendency to place the
component in slight extension, which was easily cor-
rected in later cases by securing the jig in slight flexion
relative to the anterior femur.
Fig. 2. The angle of the distal femoral cutting block is adjusted
according to the predetermined CT scan.

remaining 3 differed by only 18. A two-tailed Student’s

t-test revealed no significant differences between the
two methods of measurement (Ps0.19). Thereafter,
patients underwent only supine scans as these were
technically easier to perform.

3.2. Observer variability

Inter-observer errors were investigated using a two-

tailed Student’s t-test. Measurements for each observer
were compared against the overall average measure-
ments. The average difference of 0.0005 between
observers was very small and was found to be not
significant (Ps0.88). Intra-observer variation was rare
and was never greater than 18.

3.3. Computer vs. hard copy assessment

There was no significant difference between the read-

ings taken from the computer and those from the printed
films.

3.4. Accuracy of the technique

Of the 80 patients studied, 57 (71%) patients had

predominately medial disease and 23 had lateral disease. Fig. 3. Pre- and post-operative CT scout scans. The mechanical axis
The angular correction required for all knees is shown and the slope of the joint line and femoral component are marked.
200 V. Kinzel et al. / The Knee 11 (2004) 197–201

4. Discussion

Surgical robots are now undergoing clinical assess-

ment for use in knee replacement surgery w16,17x. Their
ultimate role will depend on their ability to allow the
components to be implanted with a high degree of
accuracy for each individual patient. Placement of the
femoral component in the coronal plane using conven-
tional instruments has proved most challenging because
of our inability to define the centre of the hip. Few
surgeons perform pre-operative planning. Most rely on
the assumption that there is fixed relationship in every
case between an intra-medullary rod and the mechanical
axis of the leg.
Our study has shown that this assumption is clearly
wrong. Not only is there wide variability between a
perfectly placed virtual rod, but even when an arbitrary
target is set, clinical variables dictate that only 81% are Fig. 5. Alignment of the femoral component relative to the mechanical
implanted within 28 of it w7x. axis in 80 knee replacements.
We feel we have demonstrated that using a 10-min
scan with minimum X-ray exposure a high degree of Although we anticipated that the technique might
accuracy can be achieved in femoral placement in the prove unsuitable in severely damaged knees, this proved
coronal plane. The technique also eliminates the need not to be the case. The correction required was less
for full insertion of an intra-medullary rod with the predictable as expected. The use of magnification over-
attendant risk of fat embolism. Placement of the femoral came the problem of defining the slope of the distal
component in the sagittal plane is straightforward when femur in the presence of osteophytes and overlying bone
referenced from the anterior femoral cortex and our shadows.
results lie well within the range reported by Faris et al. The advent of robots has stimulated our research into
w18x to have no effect on the range of movement of the improving the accuracy of mechanical instruments. The
knee. aim is to perform accurate surgery for each individual
For the purposes of the study we employed our patient rather than using arbitrary methods based on
standard instruments for the Medial Pivot Knee albeit false assumptions. Ultimately the knee should not only
used in a novel way. The technique could be adapted be perfectly aligned but also perfectly balanced. For
easily to other instrument systems. We have now devel- those sceptical about the need for improved accuracy it
oped a new jig which can be applied without the intra- is worth considering the geometry in an average knee.
medullary drill and which allows incremental 18 One degree of inaccuracy is the equivalent of 1-mm
corrections. bone thickness. Put another way, if the cut is 28 out, the

Fig. 4. Angular correction required in 80 knees to cut the distal femur perpendicular to the mechanical axis.
 V. Kinzel et al. / The Knee 11 (2004) 197–201
gap between the bone ends varies by 2 mm, the
difference between component thicknesses. There is an
inextricable link between accuracy of bone cuts and
ligament balance. In the past many medial ligaments
have been released to make up for poor bone alignment.
A released ligament is rarely a balanced ligament
throughout the range of movement. If the bone cuts are
correct, the ligaments will more often than not balance
themselves.
Acknowledgments
We are grateful to Wright Medical who provided
funding for the CT scans taken during this study and to
the staff of the CT unit at Warwick Hospital for the
high quality of their work.
References
w1x Moreland JR. Intramedullary vs. extramedullary total knee
instrumentation. Controversies of total knee replacement. New
York: Lippincott–Raven, 1991.
w2x Moreland JR. Mechanics of failure of total knee arthroplasty.
Clin Orthop 1988;226:49.
w3x Harvey IA, Manning MP, Sampath SA, et al. Alignment of
total knee arthroplasty. The relationship to radiolucency around
the tibial component. Med Eng Phys 1995;17:182.
w4x Wasielewski RC, Galante JO, Leighty RM, et al. Wear patterns
on retrieved polythene tibial inserts and their relationship to
technical considerations during total knee arthroplasty. Clin
Orthop 1994;299:31.
w5x Ritter MA, Fans PM, Keating EM. Postoperative alignment of
total knee replacement: its effect on survival. Clin Orthop
1994;299:153.
201
w6x Tew M, Waugh W. Tibiofemoral alignment and the results of
knee replacement. J Bone Joint Surg Part B 1985;67:551.
w7x Lam L, Shakespeare DT. Varusyvalgus alignment of the fem-
oral component in total knee arthroplasty. The Knee
2003;10:237 –241.
w8x Harding U, Crawford RW, Mclardy-Smith P, Murray DW. The
importance of femoral intramedullary entry point in knee
arthroplasty. Knee 1999;6:207 –210.
w9x Cates HE, Ritter MA, Keating EM, Faris PM. Intramedullary
versus extramedullary femoral alignment systems in total knee
replacement. Clin Orthop 1993;286:32 –39.
w10x Engh GA, Peterson TL. Comparative experience with intra-
medullary and extramedullary alignment in total knee arthro-
plasty. J Arthroplasty 1990;5:1 –8.
w11x Teter KE, Bregman D, Colwell CW Jr.. The efficacy of intra-
medullary femoral alignment in total knee replacement. Clin
Orthop 1995;December(321):117 –121.
w12x Jeffery JA. Accuracy of intra-medullary femoral alignment in
total knee replacement: intra-operative assessment of alignment
rod position. Knee 1999;6:211 –215.
w13x Novotny J, Gonzalez MH, Amirouche FM, Li YC. Geometric
analysis of potential error in using femoral intramedullary
guides in total knee arthroplasty. J Arthroplasty
2001;16(5):641 –647.
w14x Lonner JH, Laird MT, Stuchin SA. Effect of rotation and knee
flexion on radiographic alignment in total knee arthroplasties.
Clin Orthop 1996;331:102 –106.
w15x Rees JL, Price AJ, Beard DJ, Robinson BJ, Murray DW.
Defining the femoral axis on lateral knee fluoroscopy. The
Knee 2002;9:65 –68.
w16x Stulberg S, Loan P, Sarin V. Computer Assisted Navigation in
Total Knee Replacement. Results of an initial experience in
thirty-five patients. J Bone Joint Surg 2002;84(A):90 –97.
w17x Sikorski JM, Chauhan S. Computer Assisted Orthopaedic
Surgery. Do we need CAOS. J Bone Joint Surg 2003;85(B):
319 –323.
w18x Faris PM, Ritter MA, Keating EM. Sagittal plane positioning
of the femoral component in total knee arthroplasty. J Arthro-
plasty 1988;3:355 –358.