The clinical and radiographic results of modern cementing
techniques for fixation of femoral components in total hip arthroplasty
have usually been excellent16,28-30.
However, several authors have described early loosening of femoral
components of various designs. Surface finish has been implicated
as a possible factor, and the results associated with use of the
Iowa stem (Zimmer, Warsaw, Indiana) have been reported to be particularly
sensitive to surface finish6,7,27,35.
This prosthesis, with the addition of a roughened, precoated finish,
was designed to reduce the stress on the cement mantle as well as
to provide greater offset than standard straight stems. Both surface finish
and stem geometry have been implicated as factors contributing to
early stem debonding leading to periprosthetic osteolysis and stem
failure.
Mohler et al.27 reported a
unique pattern of loosening of the Iowa stem whereby the implant
debonded from its cement mantle as manifested by an enlarging radiolucency
along the shoulder of the stem in zone I of Gruen et al.17. This debonding was followed by
progressive loosening of the prosthesis and the development of extensive
osteolysis in the proximal part of the femur and at the distal stem
tip. The authors noted that cement mantle deficiencies were common
and may have led to loosening. Mohler et al.27,
Callaghan et al.6, and Sporer
et al.35 suggested that the geometry
as well as the surface finish of the Iowa stem were primarily responsible for
this early pattern of loosening. They hypothesized that after debonding
occurs the cylindrical shape of the proximal part of the stem permits
the stem to rotate within its cement mantle, allowing the roughened
surface to abrade the polymethylmethacrylate and produce wear debris
particles, leading to osteolysis and eventual aseptic failure. However,
this early loosening has not been reported in other studies of roughened,
precoated stems of other geometries2,16,26,29,30,34.
In December 1989, the Iowa stem was introduced at our institution
for hips requiring greater offset as determined by templating of
preoperative radiographs. The purpose of this study was to report
the performance of the Iowa stem after a minimum follow-up of four
years.
In a nonrandomized study that began in December 1989, we evaluated
the results of the use of Iowa stems at our institution. The data,
including clinical scores and radiographs, were prospectively collected
at predetermined intervals. From December 1989 through October 1994,
102 Iowa femoral components were implanted in ninety-five patients
with an average age of sixty-nine years (range, forty to eighty-six
years). There were fifty-three men (fifty-six hips) and forty-two
women (forty-six hips). At the time of the review, sixteen patients
(seventeen hips) had died of unrelated causes less than forty-eight
months postoperatively and two patients (two hips) had been lost
to follow-up. All nineteen hips were well functioning and radiographically
stable at their last follow-up evaluation. One patient (one hip)
had incomplete radiographic follow-up but had a clinically well functioning
hip. Thus, the results of a complete clinical assessment and radiographic
evaluation were available for eighty-two hips in seventy-six patients.
The average duration of follow-up was sixty-five months (range,
forty-eight to 104 months).
The preoperative diagnosis was primary osteoarthrosis in eighty-nine
hips (87 percent), rheumatoid arthritis in four (4 percent), avascular
necrosis in seven (7 percent), and posttraumatic osteoarthrosis
in two (2 percent). In patients with disabling disease of the hip,
the indications for hybrid total hip arthroplasty (a cementless
acetabular component and a cemented femoral component) included
an age of more than sixty-five years or femoral bone stock that
was considered inadequate for cementless fixation. The Iowa stem
was used for hybrid total hip arthroplasty in patients who required
greater offset to restore the hip biomechanics, as determined by
preoperative templating, than was provided by the Harris Precoat
stem (Zimmer, Warsaw, Indiana), which was the stem that was normally
used during this time-interval at our institution.
The Iowa femoral components (Fig. 1) were made of forged cobalt-chromium-molybdenum
alloy, had a grit-blasted roughened surface finish (Ra roughness,
sixty to eighty microinches), and were precoated with a thin layer
of polymethylmethacrylate over the proximal one-third of the stem
during the manufacturing process. The stems came in two diameters
(16.5 and 18.5 millimeters) and were 140 millimeters long. The neck
angle used at our institution was 132 degrees with an offset of
forty to fifty-five millimeters, depending on the head and neck
prostheses that were selected. The components all had a collar and
were cemented with modern technique. This included the use of a
cement restrictor, a cement gun, vacuum-mixing, and cement pressurization.
Simplex-P polymethylmethacrylate (Howmedica, Rutherford, New Jersey)
was used in all hips and was placed into the femoral canal in a
retrograde fashion. A distal centralizer was utilized in sixteen
hips. A modular twenty-eight-millimeter-diameter cobalt-chromium
femoral head with five available neck lengths was applied to the
Morse taper.
A hemispherical acetabular component made of titanium alloy with
a sintered titanium fiber-mesh porous coating for fixation was used
in all 102 hips. The Harris-Galante-II cup (Zimmer) was used in
the first ninety-four hips, and the Trilogy component (Zimmer) was
employed in the last eight. The acetabular components were inserted after
underreaming by one to two millimeters. In nearly all patients,
supplemental fixation was provided by one, two, or three 6.5-millimeter-diameter
acetabular bone screws.
Patients were followed in the office, according to a predetermined
protocol, at six weeks, at three months, at six months, and yearly
thereafter. At each visit, the patient was examined clinically and a
complete set of radiographs was made; these included an anteroposterior
radiograph of the pelvis, anteroposterior and lateral radiographs
of the hip, and a lateral radiograph of the acetabulum. The radiographs
were evaluated both qualitatively and quantitatively. The radiographs
made at six weeks postoperatively were used as the baseline. Two
independent observers scrutinized all radiographs of the femora
for radiolucent lines at the bone-cement and prosthesis-cement interfaces,
osteolysis, subsidence, migration, quality of the cement mantle,
and stem alignment (varus, valgus, or neutral). At each yearly visit,
an individual other than the operating surgeon determined a Harris
hip score19.
Qualitative evaluation of the acetabular component consisted
of analysis of the implant-bone interface for the presence and extent
of radiolucent lines on the anteroposterior radiograph of the pelvis.
The implant-bone interface was divided into five zones with a modification
of the method of DeLee and Charnley10,25.
The screws used to supplement the fixation of the acetabular cup
were evaluated closely for evidence of adjacent radiolucent lines,
breakage, or migration. Furthermore, the acetabular polyethylene was
analyzed for linear wear and the periacetabular bone was examined
for osteolysis. Comparisons were made among initial radiographs,
those made at all follow-up intervals, and the final follow-up radiographs.
Quantitative evaluation of the acetabular component was carried
out with a digitizing tablet (Sigma Scan; Jandel Scientific, Corte
Madera, California) as previously described25.
Definite loosening of the acetabular component was defined as more
than two millimeters of migration when compared with the position
on baseline radiographs. When radiolucent lines were present in
at least four zones and they were more than two millimeters wide
in at least one of the zones, the acetabular component was defined
as probably loose26.
To evaluate the femoral component, the proximal aspect of the
femur was divided into seven zones according to the method of Gruen
et al.17. Radiolucent lines at
the bone-cement and implant-cement interfaces were measured on both
the anteroposterior and the lateral radiographs of the proximal
part of the femur. Subsidence or a change in component position
was determined by previously described methods26.
The bone density of the calcar region of the proximal-medial cortex
of the femur was evaluated qualitatively. The diaphysis of the femur
was evaluated for any evidence of cortical hypertrophy.
Debonding of the implant from the cement mantle was defined as
a separation or a definite radiolucent line at the stem-cement interface
that was not initially present on postoperative radiographs. Osteolysis
was defined as a radiolucent area or line greater than three millimeters
in thickness and progressive in nature as compared with that seen
on the previous follow-up radiographs. Loosening of the femoral
component was defined with use of the criteria established by Harris
et al.20. The femoral component
was considered definitely loose if there was subsidence of more
than two millimeters, progressive radiolucency between the stem
and the cement, or fracture of the cement or the stem.
The cement mantles were graded as A, B, C, or D, according to
the criteria of Barrack et al.1,
on radiographs made within the first six months after implantation.
The category of C-2 includes either a thin cement mantle (less than
one millimeter) at any site or a defect in the mantle with metal in
direct contact with cortical bone. The mantle is also rated as C-2
if there is less than ten millimeters of cement between the stem
tip and the end of the cement plug.
Heterotopic ossification was graded according to the classification
of Brooker et al.4. A Kaplan-Meier21,32 survivorship analysis of all
102 hips was used to assess the life span of both the acetabular
and the femoral components and the development of femoral debonding
and osteolysis.
Clinical Results
For the eighty-two hips (seventy-six patients) with at least
forty-eight months of follow-up, the average preoperative Harris
hip score19 was 47 points (range,
16 to 69 points). At the latest follow-up examination, at an average
of sixty-five months (range, forty-eight to 104 months), the average
Harris hip score was 87 points (range, 24 to 100 points). Overall,
fifty-three (65 percent) had an excellent result (90 to 100 points),
sixteen (20 percent) had a good result (80 to 89 points), three (4
percent) had a fair result (70 to 79 points), and ten (12 percent)
had a poor result (less than 70 points).
Examination of the ten poor results (nine patients) demonstrated
that pain and limp were primarily responsible for the poor Harris
hip scores. The scores regarding walking supports revealed that two
of the nine patients in this group were unable to walk. This group
of nine patients included the one patient in the series who had
a periprosthetic infection, one of the patients with femoral osteolysis,
and the one patient who had a femoral revision without infection.
There were three reoperations that involved revision of one or
both prosthetic components. One was performed because of a late
deep hematogenous infection, which developed after the prosthesis
had been in situ fifteen months and which necessitated
removal of both the femoral and the acetabular component in a two-stage
revision. The second reoperation was performed in a patient who
had femoral loosening and subsequent development of a large osteolytic lesion
in zone II of Gruen et al.17 without
evidence of infection (Fig. 2-A and Fig. 2-B). The femoral component was revised
because of progressive symptoms and radiographic loosening with
osteolysis at fifty-three months after the index procedure; the
acetabular component was retained. The third reoperation was an
acetabular revision in a patient with recurrent dislocations secondary
to loss of abductor function from a trochanteric fracture. There
were no hips with aseptic loosening of the acetabular component,
and no revision operations were pending at the time of this review.
Complications
Postoperative complications included urinary tract infection
in three patients and urinary retention, a superficial wound infection
requiring intravenous antibiotics, and an ileus in one each. Atrial
fibrillation developed after the procedure in one patient, and one
patient had thrombosis of a femoral-popliteal bypass graft, requiring
thrombectomy and heparinization, on the second postoperative day.
No patient died, and there were no instances of excessive bleeding,
pulmonary emboli, or transient neurapraxias.
Late complications were seen in three patients, all described
in the section on revisions above.
Radiographic Results
Femoral Component
A one-millimeter-thick nonprogressive radiolucent line at the
prosthesis-cement interface, which was not present on the initial
postoperative radiographs, was seen in four hips at an average of
thirty-four months (Fig. 3-A and Fig. 3-B). One of these stems was revised,
as previously noted. All of these radiolucencies were present in Gruen
zone I. However, one radiolucent line was also seen in Gruen zone
VII. No hip had subsidence of the femoral stem (except for the hip
that had the stem revision, as previously mentioned), stem fracture,
or cement fracture.
Osteolysis was seen at the bone-cement interface in five hips
at an average of fifty-four months. One of the five hips had the
osteolysis in Gruen zone II and was revised as previously described.
Of the other four osteolytic lesions, one occurred in zone III;
two, in zone VI; and one, in zones V, VI, and VII. None of these
four hips were revised, and all four remained asymptomatic. Three
of the five hips had debonding in Gruen zone I, as described above.
The two hips that had isolated osteolysis in Gruen zone VI did not
have radiographic evidence of implant-cement radiolucencies elsewhere
to suggest debonding (Fig. 4-A and Fig. 4-B).
The femoral cement mantle was grade A in thirty-two hips (31
percent), grade B in nineteen (19 percent), grade C-1 in thirteen
(13 percent), grade C-2 in thirty-seven (36 percent), and grade
D in one (1 percent). The femoral stem was in neutral alignment
in fifty-two hips (51 percent), valgus in twenty-two (22 percent),
and varus in twenty-eight (27 percent). Varus or valgus alignment
usually produced a grade C-2 cement mantle as a result of the tip
of the stem nearing the femoral cortex (usually in zone III for
varus stems and in zone V for valgus stems).
The cement mantle was grade C-2 in all hips with debonding or
osteolysis. No hip with a grade-A, B, or C-1 cement mantle demonstrated
radiolucencies at the prosthesis-cement interface or evidence of
osteolysis. None of the sixteen stems that were inserted with a
distal centralizer had debonding or were associated with osteolysis
(p > 0.05). The single grade-D mantle was a result of a deficient cement
plug distal to the stem tip and was not associated with radiolucencies
or osteolysis.
Examination of the bone-cement interface demonstrated no linear
radiolucencies greater than one millimeter in width.
Review of the femoral radiographs revealed that there initially
had been complete contact between the collar and the calcar in fifty-four
hips (53 percent) and partial contact in forty-eight hips (47 percent).
At the time of the final follow-up, there was complete contact in
fifty hips (49 percent), partial contact in fifty hips (49 percent),
and no contact in two hips (2 percent). There was calcar resorption
or rounding at the time of the final follow-up in seventy-seven
hips; it was complete in five (5 percent) and partial in seventy-two
(71 percent). The density of the calcar was estimated to be decreased
in seventy-eight hips (76 percent), increased in one hip (1 percent),
and unchanged in twenty-three hips (23 percent). Hypertrophy of
the femoral cortex was seen in five hips (5 percent); it was primarily
found in Gruen zones III and V.
Radiographic review of the seventeen hips with less than forty-eight
months of follow-up showed no evidence of radiolucencies, debonding,
migration, osteolysis, or subsidence.
Acetabular Component
Linear polyethylene wear averaged 0.11 millimeter per year (maximum,
0.44 millimeter per year) over an average time-interval of sixty-five
months. The average inclination angle of the components was 41 degrees
(range, 21 to 68 degrees) of abduction. No screws broke, and only
two screws were associated with any radiolucency.
Peripheral, noncontinuous, nonprogressive radiolucent lines were
seen around fifty-two acetabular components (51 percent). All radiolucent
lines were less than one millimeter thick. No acetabular component
had a continuous or progressive radiolucent line. Periacetabular
osteolysis occurred in two hips, in zone A1 in one and in zone A2
in the other. Both lesions were small (less than one square centimeter
on radiographs). No cup migrated or was considered definitely loose.
Only two acetabular components were revised, one because of recurrent
dislocations and one because of infection, as mentioned above. There
were no acetabular revisions because of aseptic loosening.
Heterotopic bone formation was noted in forty-nine hips (48 percent).
It was categorized as class I in twenty-two (22 percent), class
II in fourteen (14 percent), and class III in thirteen (13 percent). No
hips demonstrated class-IV heterotopic bone formation, and no operations
were performed as a result of heterotopic ossification.
Survivorship Analysis
A Kaplan-Meier survivorship curve21,32 for
failure of either the femoral or the acetabular component necessitating
revision for any reason revealed that the chance of both implants
surviving at seven years was 96.5 percent (95 percent confidence
interval, 0.94 to 0.98). The probability of the femoral component
surviving at seven years was 97.6 percent (95 percent confidence
interval, 0.96 to 0.99) with revision for any reason as the end
point, 92.6 percent (95 percent confidence interval, 0.89 to 0.96)
with femoral osteolysis as the end point, and 90.6 percent (95 percent
confidence interval, 0.87 to 0.94) with femoral debonding, osteolysis,
or revision for any reason as the end point.
The purpose of this prospective study was to examine the performance
of 102 consecutive primary Iowa stems at our institution. We found
that the Iowa stem, with its cobra shape and increased offset, did
not perform as well as other cemented straight stems used at our
institution2,26. At an average
of 5.4 years, radiographic debonding developed at the stem-cement
interface in four (5 percent) of the eighty-two hips; one of them
was revised because of osteolysis and pain. Furthermore, although
the follow-up was short, femoral osteolysis had developed in five
hips, two of which did not have radiographic debonding.
Of 701 primary hybrid total hip arthroplasties that were performed
at our institution during the review period, 484 involved use of
a Harris Precoat component, which is another roughened, precoated
stem but with less offset. The Harris Precoat stem had a grit-blasted
surface finish with an Ra roughness of sixty to ninety microinches
and an offset of thirty-three to forty-three millimeters, depending
on the prosthetic head and neck selected. The Iowa stem used in
this study had a surface finish with an Ra roughness of sixty to
eighty microinches and available offsets of forty to fifty-five millimeters.
A review of the results of 153 total hip arthroplasties performed
with the Harris Precoat stem at our institution, and followed for
a similar duration of five to eight years, showed only a 1 percent
failure rate and no osteolysis adjacent to any nonrevised stem26. This comparison is very appropriate,
as the two groups were identical with regard to patient selection,
surgical skill, surface finish, precoating, and duration of follow-up.
The main differences were stem geometry and offset. Furthermore,
the ten-year survival rate of the Harris Precoat stem, with revision,
aseptic loosening, or osteolysis as the end point, was 98.4 percent
(95 percent confidence interval, 0.97 to 1.00)2.
The overall rate of failure and development of osteolysis associated
with the Iowa stem in this series was higher than that associated
with other contemporary designs of cemented stems. Goldberg et al.16 reported on 125 consecutive hybrid
total hip arthroplasties and noted that only one femoral stem was
revised because of mechanical loosening and only one other stem
was radiographically loose after an average of 8.6 years of follow-up.
Similarly, Oishi et al.30 found
a 1 percent rate of failure (loosening or revision) of the femoral
component six to eight years after 100 hybrid total hip arthroplasties.
Berger et al.2, reviewing the
results of 150 consecutive hybrid total hip arthroplasties, reported
a ten-year survival rate of the stem of 98.4 percent (95 percent
confidence interval, 0.97 to 1.00). Lastly, Mulroy et al.29, in a review of the results of 162
hybrid total hip arthroplasties, reported a 2 percent rate of femoral revision
at an average of fifteen years.
The 90.6 percent rate of survival of the Iowa stem at seven years
in our study, with revision, debonding, or osteolysis used as the
end point, was similar to the performance of this stem reported
by other authors. In a study of 131 hybrid total hip arthroplasties
with the Iowa femoral component, Callaghan et al.6 reported
that the rate of radiographic aseptic loosening of the femoral component
was 7 percent (nine hips) at eight to nine years. Of these nine hips,
eight had been revised because of aseptic loosening, as evidenced
by stem debonding, at an average of 6.25 years postoperatively.
Six of these revised hips demonstrated femoral osteolysis. The rate
of aseptic loosening of the stem in the present series was very
similar (5 percent).
Several authors have attributed the stem's poor survival rate
to its surface finish6,27,35.
Sporer et al.35 hypothesized that
the rougher surface finish of the grit-blasted precoated stem may
produce more wear debris from the cement mantle as a result of its
abrasive nature compared with the bead-blasted nonprecoated stem.
Such debris may lead to third-body wear, and possibly an increase
in polyethylene debris, and therefore may lead to greater and earlier
debris-induced osteolysis and loosening. However, this surface finish
has performed well at our institution2,26,
as demonstrated by the 98.4 percent rate of survival of a different
cemented grit-blasted precoated stem at ten years. Other centers
have reported similar results. Brown and Lachiewicz5 reported that, five to nine years
after arthroplasty with a precoated femoral component in 119 consecutive
hips, only two stems had definite loosening and none demonstrated
debonding. Therefore, we believe that it was not the surface finish
but rather the stem's geometry, shape, and increased offset and
the effect that these had on cementing technique and strain that
were responsible for the Iowa component's poor performance.
A number of studies have recently suggested that a thin cement
mantle (less than two millimeters in thickness) may contribute to
loosening and failure of cemented femoral components12,22,36. Mulroy et al.29 showed that a thin cement mantle
(less than one millimeter thick) at one point or more was a significant
predictor (p < 0.05) of radiographic loosening of the femoral
component. Reductions in axial and shear strain of the cement have
been noted with thicker cement mantles13,14.
In the present study, every stem that debonded or was associated
with osteolysis had a grade-C-2 cement mantle. The design and geometry
of the Iowa stem involves a cylindrical shape distal to a proximal
cobra shape with increased femoral neck offset in comparison with
that of standard straight stems. Thus, insertion of a stem with
increased lateralization and offset into the proximal part of the femur
during cementing can lead to a suboptimal cement mantle. First,
because of the stem's cobra shape and offset, there is a tendency
to insert the stem initially by way of a varus position. This creates
an inadequate cement mantle in the area just distal to the greater
trochanter. As a result, there can be very thin cement or almost
a complete lack of cement in Gruen zone I or II, or both. We found seven
hips with a grade-C-2 cement mantle as a result of such deficiencies.
In addition, because of the stem's geometry and shape, there
was a greater tendency to position the stem in valgus or varus malalignment
rather than in a neutral position. Malpositioning was noted in 49
percent of the hips in our series, with twenty-eight hips in varus
and twenty-two hips in valgus. This malpositioning can also result
in a thin (grade-C-2) cement mantle with a resultant increase in
cement strain13. The deficiency
was localized to Gruen zone III in varus hips and to Gruen zone
V in valgus hips. Ebramzadeh et al.12 showed
that progressive loosening, cement fracture, and radiolucent lines
at the stem-cement or bone-cement interface are more likely to develop when
a stem is placed in more than 5 degrees of varus. A distal centralizer
was used in only sixteen hips in our study, and neither debonding
nor osteolysis developed in any of those hips. The use of this device
reduced the tendency to place the stem in varus or valgus alignment
and improved the cement mantle (none were grade C-2). The quality
of the cement mantle improved, particularly on the lateral radiographs.
Distal centralization has been shown to prevent malpositioning and
to significantly reduce suboptimal cement mantles (p < 0.0001)3,15,18.
The Iowa stem's shape may have also influenced the transmission
of load to the surrounding cement mantle as well as the development
of osteolysis. The Harris Precoat stem has a rectangular shape with
rounded edges, whereas the Iowa stem has a cylindrical shape. Mohler
et al.27 postulated that the cylindrical
shape allows the Iowa stem to rotate within its cement mantle after debonding
has occurred, leading to increased wear debris, osteolysis, and
eventual failure. Several authors have noted that a flat-sided femoral
stem provides more torsional resistance and lower peak tensile stresses
in the proximal part of the cement mantle than a round femoral stem8,24. Thus, the design of the Iowa
stem may predispose to early debonding as a result of this increase
in cement stress.
Finally, the increased offset of the Iowa stem may have contributed
to the higher debonding and failure rates in comparison with those
of standard straight stems. An increase in offset has several advantages,
including increasing the abductor moment arm of the hip. Consequently,
there is a decrease in the necessary abductor force about the hip
and therefore a reduction in the resultant force across the hip.
A concern, however, is the potential increase in the bending moment
on the prosthesis23,37. An increase
in the cement mantle strain may lead to debonding or failure of
the implant. Medial and lateral cement strains have been shown to
increase in the proximal, middle, and distal parts of the cement
mantle as a result of increased offset9.
Chang et al.8 demonstrated that
increasing neck length, and hence offset, increases shear stresses
at the cement interface by 24 percent.
In conclusion, our experience with the Iowa femoral stem demonstrated
relatively early debonding, loosening, and osteolysis. All of the
stems that were radiographically loose, debonded, or associated
with osteolysis had a deficient cement mantle. When the femoral
component is cemented, emphasis must be placed on technique to produce
an optimal mantle. This may involve the use of proximal and distal
centralizers as well as a rasp-to-stem ratio that allows for an
adequate mantle of at least one millimeter in thickness2,11,29. We stopped using the Iowa
stem but have not abandoned roughened and precoated stems in general.
We believe that the design of the stem - that is, its geometry and
increased offset - and the effect that it had on the quality and
strain of the cement mantle, not the stem's surface finish or precoat,
was the critical element in the failure mechanism.