patellar strain and patellofemoral contact after bone-patellar tendon-bone harvest for anterior...

2 5 6

Patellar Strain and Patellofemoral Contact After Bone - Patellar Tendon-Bone Harvest for Anterior Cruciate Ligament Reconstruction Neil A. Sharkey, PhD, Seth W. Donahue, BS, Tait S. Smith, BS, Brian K. Bay, PhD, Richard A. Marder, MD

ABSTRACT. Sharkey NA, Donahue SW, Smith TS, Bay BK, Marder RA. Patellar strain and patellofemoral contact after bone-pate l la r t endon-bone harvest for anterior cruciate liga- ment reconstruction. Arch Phys Med Rehabil 1997;78:256-63.

Objective: To characterize the morbific consequences of har- vesting a patellar tendon graft for use in reconstructing the anterior cruciate ligament (ACL) of the knee, specifcally, (1) to measure changes in patellar strain and patellofemoral contact due to graft harvest, (2) to evaluate the ability of bone-grafting the patellar defect to mitigate these effects, and (3) to character- ize failure of the extensor mechanism after harvest of a patellar tendon graft.

Design: Twenty-two cadaver knee joints were tested before and after harvest of a patellar tendon graft and after filling the patellar defect with polymethylmethacrylate to simulate a healed bone graft. Knees were positioned in 30 ° , 60 ° , and 90 ° flexion and loaded while measuring axial strain in the anterior patella and patellofemoral contact. Knees were then loaded to failure.

Results: Harvest of the graft produced increases in axial strain at all flexion angles. Filling the defect restored axial strain to normal values. Patellofemoral contact in the presence of a defect, either filled or empty, was not different from contact for intact patellae. Most knees failed by transpatellar fracture; mean extension moment at failure was 112.8Nm. The best predictors of failure were age and gender.

Conclusion: Patients undergoing ACL reconstruction with a patellar tendon graft are at increased risk of anterior knee pain and disruption of the extensor mechanism. Bone-grafting the patellar defect created by graft harvest can reduce these risks. Our findings underscore the importance of carefully controlled rehabilitation and suggest that i f an accelerated program of rehabilitation is anticipated, the patellar defect should be bone- grafted. Older patients, particularly women, are at increased risk of catastrophic failure of the knee extensor mechanism after ACL reconstruction using patellar tendon graft.

© 1997 by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabili- tation

T HE USE OF A U T O G E N O U S patellar tendon as a graft to reconstruct the torn anterior cruciate ligament (ACL) is

recognized as a highly successful procedure for stabilizing the

From the Orthopaedic Research Laboratories, University of California, Davis, School of Medicine, Sacramento, CA.

Submitted for publication June 3, 1996. Accepted in revised form August 30, 1996.

Supported by a grant from Interpore International, Irvine, CA. An organization with which one or more of the authors is associated has

received or will receive financial benefits from a commercial party having a direct financial interest in the results of the research supporting this article.

Reprint requests to Neil A. Sharkey, PhD, Orthopaedic Research Laboratories, Research Facility, Room 2000, 4815 Second Avenue, Sacramento, CA 95817.

© 1997 by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation

0003-9993/97/7803-405053.00/0

cruciate-deficient knee. However, notable morbidity has been described resulting from removal of the central third of the pa t e l l a tendon and its bony attachments. This includes patellar fracture, rupture of the extensor mechanism, contracture of the remaining pa t e l l a tendon inducing patella baja, as well as resid- ual quadriceps weakness, and pa te l l a subluxation. 1-11 In addi- tion, the onset of anterior knee pain after ACL reconstruction has been attributed to use of autogenous patellar tendon grafts, although this postoperative complication has been associated with other grafts as well. Bone-grafting the pa t e l l a defect has been advocated as a means of countering the potential adverse effects of removalJ z

Alterations in strength of the extensor apparatus, pa te l l a strain, and patellofemoral contact area and pressure have been assumed to be the mechanisms by which many of these unde- sired effects occur. Previous investigators have studied patello- femoral contact areas and pressures after removal of the patellar tendon only, without simultaneous harvest of the accompanying bone blocks from the patella and tibial tubercleJ TM Thus, the effects of a bony pa t e l l a defect were not examined in these studies. Pa te l l a surface strain after removal of a patellar block has been previously examined and shown to increase in magni- tude and alter direction. 15'16 To date, no study has analyzed the effects of removal of the patellar tendon and its bony attach- ments on patellofemoral contact areas, pressures, and patellar surface strains while loading the patellofemoral joint using physiologic moments. Furthermore, no study as yet has loaded specimens to failure, which would allow definition of the upper limits of permissible quadriceps loading during rehabilitation following reconstruction of the ACL.

This study was designed to measure the effect of harvesting a bone-pate l la r t endon-bone graft on patellofemoral mechan- ics and to measure the efficacy of filling the bony defect in the patella as a means of restoring normal patellofemoral mechanics after graft harvest.

MATERIALS AND METHODS

Specimen Preparation and Loading Twenty-two nonembalmed knee joints obtained from human

cadavers were tested in this investigation; the mean age at the time of death was 72 years (range, 35 to 88 years). Cadavers were stored at 2°C; specimens were obtained within 48 hours of death. Twelve of the specimens were from women and 10 from men. Knees were evenly distributed between left and right but were nonpaired; each knee was from a separate donor. Spec- imens were procured by osteotomies performed approximately 20cm above and below the joint line and then frozen at - 4 ° C until preparation.

On the day of testing, specimens were thawed and the super- ficial soft tissues were removed from each knee leaving the extensor mechanism, capsule, and collateral ligaments intact. The patellar tendon was sectioned at the quadriceps aponeurosis, about 15cm above the superior pole of the patella. Using poly- methylmethacrylate, the proximal tibia was fixed into an alumi- num pot which was attached infefiorly to a telescoping cylinder.

Arch Phys Med Rehabil Vo178, March 1997

PATELLOFEMORAL BEHAVIOR AFTER HARVEST, Sharkey 257

~ial Strain Gages

Reference Holes

Fig 1. Illustration of anterior patellar surface showing the position of the two axial strain gauges in relation to the defect created by removal of the bone-patellar tendon graft. Probes were inserted into the superior reference holes during loading to determine the position of the pressure- sensitive film within the patellofemoral joint.

The distal end of this cylinder was connected to a platform which allowed free rotation in the sagittal plane. This plaform was fixed to the base of an Instron 1122 materials testing ma- chine." The overall length of the surrogate tibia from the knee joint line to the ankle joint line was 44cm. A similar telescoping aluminum cylinder was used to simulate the femoral shaft, ex- cept in this case the distal femur was attached to the cylinder using an intramedullary rod. The proximal end of this cylinder was connected to a load cell mounted on the crosshead of the Instron machine using a universal couple supplied by the manu- facturer. Total femoral length was 44cm. Using this experimen- tal arrangement (ie, the superiorly positioned universal couple representative of the hip joint and the inferiorly positioned rota- tional platform representative of the ankle), the knee could be set and held at a desired flexion angle simply by raising or lowering the Instron crosshead. Instantaneous center of rotation of the knee joint was not determined by the apparatus, but was regulated by the articular geometry and soft tissue constraints.

To simulate the action of the extensor mechanism (contrac- tion of the quadriceps muscle), a freeze clamp j7 was attached to the patellar tendon, 5 to 8cm proximal to the superior pole of the patella. A cable connected this clamp to a force-trans- ducer b in series with a computer-controlled linear stepper motoff mounted on the surrogate femoral shaft. The direction of quadri- ceps tension was aligned with the surrogate femur approximat- ing a physiologic Q-angle. A personal computer equipped with analog-to-digital hardware a was used for apparatus control and strain data acquisition. The experimental apparatus can repeti- tively deliver quadriceps loads exceeding 4kN and is capable of producing isometric extension moments that exceed average maximum voluntary isometric extension moments reported in the literature. 18-21

Measurement of Patellar Strain

Patellar strain in the axial direction was measured with two unidirectional strain gauges e located along the horizontal mid- line of the patella at about one third width intervals (fig 1).

The patellar tendon has a continuous insertion along the ante- rior patella; small axial gauges were chosen to minimize the removal of tendon necessary for gauge placement. Rectangular areas measuring 4 x 6mm were cleared of overlying tendon with a scalpel and small periosteal elevator, and the strain gauges were attached with cyanoacrylate per manufacturer in- structions. Gauges were positioned so that they would be located close to the superior comers of the bony defect created by harvesting the patellar tendon graft. Ten preliminary repeated

measurements of axially directed surface strain in the patella demonstrated a coefficient of variation of less than 3% using this measurement system.

Measurement of Patellofemoral Contact Area and Pressure

Patellofemoral contact was measured using both low range (2.5 to 10.0MPa) and medium range (10 to 50MPa) Fuji pres- sure-sensitive film. f Contact area and mean pressure were calcu- lated from the low-range imprint. Peak patellofemoral contact pressure was measured using low- or medium-range film de- pending on peak pressure magnitude.

Rectangles measuring 5 × 7cm of the two film layers were cut and sealed within two layers of polyethylene film with an impulse heat sealer, z Total thickness of the prepared film packets was .20mm for the low-range pressure film and .25mm for the medium-range film. During testing, film packets were inserted between the cartilaginous surfaces of the patella and femur by releasing the quadriceps freeze clamp from its cable and pulling the extensor mechanism anteriorly. Film position in relation to the patella was determined by drilling a pair of l-ram holes through the superior patella and inserting blunt tip probes which were pushed against the intra-articular film, leaving two refer- ence dots on the film. Experiments were conducted by activating the stepper motor attached to the quadriceps mechanism and loading the knees to a predetermined maximum isometric exten- sion moment. Knees were maintained at peak moment for 30 seconds, during which time reference points were marked on the film, and then unloaded. The entire loading sequence lasted 90 seconds. Although this is longer than most physiologic load- ing episodes, the 90-second interval was dictated by the speed of the apparatus and by the exposure requirements of the film. After testing, films were disassembled and laminated for preser- vation.

To increase measurement accuracy, films were analyzed by a three-stage process of digitization, filtering, and measure- ment--techniques previously established in our laboratoriesY Films were digitized at a spatial resolution of 2.95 pixels per millimeter and eight-bit gray-scale resolution using a flatbed scanner, h Contrast and brightness were adjusted to yield a con- tinuous gray-scale histogram that spanned the entire range of the film from no signal to fully saturated. A Macintosh IIcx personal computer and IMAGE software (public domain; devel- oped by Wayne Rasband, National Institutes of Health) was used to conduct image filtering and measurement. Images were enhanced for measurement with two digital filters, a background subtraction filter (2D rolling bail with a radius of 50 pixels) to reduce variations in scanning brightness followed by convolu- tion filter to spatially smooth the film signal. Filtering reduces the grainy appearance typically seen in the images due to the microcapsule nature of the film. Manufacturer's instructions recommend averaging film color density measurements over an area of at least 2mm 2 to prevent image grain from adversely influencing data. Convolution filtering achieves this signal aver- aging function in a single operation for the entire image.

Pressure magnitudes were determined using calibration curves constructed in our laboratory. In-house calibration curves were produced from pressure images made using a delrin cylin- der mounted in the Instron machine. These curves, though not appreciably different than curves supplied by the manufacture, were used because they were produced using the same prepara- tion techniques, timing, and laboratory conditions that were used during actual experimentation. In addition, constructing in-house curves verified proper film function. Preliminary stud- ies of repeated measurements using our system yielded coeffi- cients of variation for contact area, mean pressure, and peak

Arch Phys Med Rehabil Vol 78, March 1997

2 5 8 PATELLOFEMORAL BEHAVIOR AFTER HARVEST, Sharkey

1200

1000

800

600

400

200

0

90 Degrees Flexion , / / / / / "

Z • • • / / / / / • • / 60 Degrees Flexion

.< / ~ ~ ~ 30 Degrees Flexion

.4 ~ ~ - - -- -- Lateral Axial Strain I 1

~ . . . I . . ~ 1 . . . . [ . . . . [ . . . . I I

0 10 20 30 40 50 60 70 80

Moment (Nm)

Fig 2. Line plots of mean axial surface strain for intact patella versus knee extension moment, Increasing the extension moment about the knee by simulated quadriceps contraction produced corresponding in- creases in axial strain, The rate of strain increase, as defined by the slope of the strain versus moment plots, increased significantly with increasing knee f lexion (p < .0001; n = 13). Strains recorded at the medial and lateral gauge sites were not significantly different,

pressure less than 3%, 5%, and 10%, respectively. Variance between known applied loads and measured loads (the integral of an entire pressure distribution map) was 4%.

Digitized images of the entire articular surface of each patella were collected using a Pulnix CDC video camera) Each digi- tized pressure pattern image was overlaid with its corresponding digitized articular surface image using IMAGE software. Proper alignment was achieved by matching the drill holes on the patella surface image to the dots on the film image (produced with the blunt tip inserted through the drill holes during testing). The combined images were then split by two vertical lines into equal thirds based on articular width. The medial region encompassed the area from approximately the center of the medial facet to the medial articular edge, the central region corresponded roughly to the area between the centers of the medial and lateral facets, and the lateral region encompassed the area from the lateral facet to the lateral cartilaginous edge.

Patellofemoral contact force for the entire joint, as well as for each of the three regions, was calculated from area and mean pressure data. Total articular surface area, articular width, and articular height were measured for each patella. The vertical distances from the most inferior point on the articular surface of the patellae to the center of contact pressure was measured for each image. Contact positions were normalized for each patella and expressed as a percentage of patella height.

Experimental Design and Analysis After preparation and mounting, each knee was sequentially

loaded in 30 °, 60 °, and 90 ° flexion by simulated isometric quad- riceps contraction. The knees were progressively loaded at each position until reaching half the reported maximum isometric flexion moment as produced by voluntary quadriceps contrac- tionJ g2l The maximum values used to calculate these moments were 92Nm, 142Nm, and 105Nm for 30 °, 60 °, and 90 °, respec- tively, and have been used to determine load magnitudes in other investigations of knee biomechanJcs. 23-25 Preliminary ex- periments using peak moments from the literature or two thirds of these peak moments repeatedly fractured patellae in the trans- verse plane after graft procurement. Therefore, simulated quad- riceps contraction was halted at extension moments of 46Nm at 30 °, 71Nm at 60 °, and 52Nm at 90 °. The vertical component

of the extension moment was measured during loading with the load cell located in the crosshead of the Instron. Axial patellar surface strain (as detected by the medial and lateral gauges) and quadriceps force (from the load cell in series with the extensor mechanism) were digitized and recorded at each 10- N increase in the vertical component of the extension moment.

All knees were tested intact, retested after harvesting a bone- patellar tendon-bone graft using standard surgical technique, and retested again after filling the bony defect in the patella with polymethymethacrylate. The patellar defect made by re- moving the bone plug was roughly rectangular in shape and 5ram deep. Defect width occupied the central third of the infe- rior patella and defect length extended from the inferior pole to the horizontal midline. The defect space was subsequently filled by compressing doughy polymethylmethacrylate into the void and allowing it to cure for at least 10 minutes after hard- ening.

Each knee was loaded three times for each position and condi- tion, first without pressure-sensitive film, then with low-range pressure-sensitive film within the patellofemoral joint, and fi- nally with medium-range film in the joint space. Preliminary repeated experiments on an intact knee demonstrated consistent contact pressure measurements regardless of testing sequence or number of trials. Therefore, for experimental consistency, the order of testing was fixed rather than randomized. Strain measurements were not affected by the presence of the pressure- sensitive film; these measurements were averaged over the three trials. After completing the strain, deformation, and contact pressure experiments each knee was loaded to failure while in 60 ° flexion; maximum quadriceps load, maximum extension moments, and mode of failure were recorded.

After all testing was completed, each patella was excised from surrounding tissue; fragmented patellae were reassembled. Using a caliper, patellar height, width, and thickness were mea- sured at their greatest point to a tenth of a millimeter. Articular cartilage was characterized using a modification of the Outer- bridge classification. Specimens with cartilage erosions or ex- posed subchondral bone (Outerbridge IV) were excluded from the study prior to testing. Articular cartilage was otherwise noted to be normal, fragmented and fissured affecting less than 25% of the patella (mild degeneration), or fragmented and fis- sured but affecting more than 25% of the patella (moderate degeneration). Finally, a digitized image of the articular surface of each reassembled patella was collected as previously de- scribed.

Axially directed surface strain in the medial and lateral pa- tella, mean axial strain (averaged medial and lateral strains), contact area, mean pressure, peak pressure, and force for the central, lateral, and medial regions of the patellofemoral joint, vertical position of the contact center of pressure, quadriceps load at failure, and extension moment at failure were the depen- dent variables used to assess the impact of bone-patel lar ten- don-bone graft harvest, and subsequent defect backfilling, on patellofemoral mechanics. All strain and contact parameters measured at peak moments were compiled and analyzed using repeated measures two-way analyses of variance (ANOVA). Flexion angle (30 °, 60 °, 90 °) and patellar condition (intact, de- fect, filled defect) were the independent factors. One-way re- peated-measures analysis of variance (ANOVA) tests were run to examine the effect of patellar condition at each of the three flexion angles. Analyses of variance were also conducted to examine whether articular degeneration correlated with changes in patellar strain or patellofemoral contact characteristics. AN- OVA tests demonstrating significant differences (p < .05) were followed with the Tukey Honestly Significantly Different (HSD) method of individual comparisons, to establish which

Arch Phys Med Rehabil Vol 78, March 1997

PATELLOFEMORAL BEHAVIOR AFTER HARVEST, Sharkey 259

Fig 3. Bar graphs demonstrating the changes in patellar strain in- duced by harvesting a mid-third bone-patellar tendon graft and the effects of filling the bone de- fect with polymethylmethacry- late. An asterisk indicates a sig- nificant difference relative to the intact state (p < .05; n = 13; error bars = 1 standard deviation),

300P

2501

200~ =

.~ 150,

100'

3000-

2500

2000-

1500-

1000-

500-

0

30 60 90 Flexion Angle (degrees)

3000 12500 Lateral Axial Strain

2000-

1500-

10002

500

0 30 60

Flexion Angle (degrees)

Mean Axial Strain

¢ :g

3O I

60 Flexion Angle (degrees)

9O

• Intact

[ ] Open Defect

[ ] Filled Defect

oiil 90

groups were different. In addition, to decrease likelihood of a Type I error when multiple comparisons are made, a p value of < .01 was taken as significant.

Backward stepwise linear regressions were run to determine if patellar strain could be predicted using a combination of patellofemoral contact parameters. Stepwise regressions were also used to examine the predictability of failure using age, sex, and patellar dimensional data. A p value of .05 or less was considered significant.

RESULTS

Five of the 22 specimens failed after graft removal while being loaded in 60 ° flexion. They were, by necessity, excluded from strain and patellofemoral contact analyses but included in the failure analyses. Complete sets of patellofemoral contact and dimensional data were collected from the remaining 17 specimens. The decision to measure axial strains was made after four specimens had already been tested for joint contact; therefore, axial strain was measured in only the last 13 knees.

Five patellae had completely normal articular surfaces, 6 were

Superior

30 Degrees 60 ~ ¢ ~ s 90 ~recs

Medial Lateral

1Rferillr

Fig 4. Schematic representation of the patellar articular surface demon- strating the mean position and distribution of patellofemoral contact at 30 °, 60 °, and 90 ° knee flexion. These parameters were unaffected by mid- third bone-patellar tendon graft harvest.

classified as mildly degenerative, and 6 demonstrated moderate degeneration. The average height, width, and depth of the bone defect created by harvesting the patellar tendon was 19.3, 9.5, and 5.7mm, respectively, constituting 46.5 _+ 4.2% of patellar height, 21.1 _+ 2.4% of width, and 24.9 _+ 5.8% of depth.

The average quadriceps forces required to achieve the maxi- mum extension moments used in these experiments (or one half of those reported for young healthy subjects) were 867 _+ 130N at 30 °, 1,679 + 256N at 60 °, and 1,788N _+ 174N at 90 ° knee flexion.

Patellar Strain

Axial strain at both the medial and lateral locations increased linearly with increasing extension moment (fig 2), at all flexion angles regardless of patellar condition. However, each flexion angle produced a significantly different relationship between extension moment and axial patellar strain as defined by the slope of the moment-strain curves (p < .0001, fig 2). In addition, the peak strains recorded at 30 ° knee flexion were significantly different than those recorded at 60 ° and 90 ° (p < .0001).

Harvesting the autograft significantly increased the medial patellar strain by 43% at 30 ° (p < .006), 32% at 60 ° (p < .002), and 43% at 90 ° (p < .0001). Filling the defect with poymethylmethacrylate restored the medial strains back to their original levels at 30 ° and 60°; peak strains remained significantly different than those for the intact patellae at 90 ° (p < .0001, fig 3). The axially directed lateral strain in the patellae demon- strated similar but less pronounced behavior (fig 3). Significant changes in lateral strain due to graft harvest were seen only at 90 ° flexion with an empty defect, where it increased from 1,097 _+ 579#e to 1,304 _+ 581#e (/9 < .004). Harvesting the graft produced significant increases in mean axial strain at all flexion angles (p < .003). Filling the defect produced mean axial strains

Arch Phys Med Rehabil Vol 78, March 1997

2 6 0 PATELLOFEMORAL BEHAVIOR AFTER HARVEST, Sharkey

'go 1 600

500-

~ 4 0 0 - g

300- <

200-

100-

0-

Contact Area

30 60 Flexion Angle (degrees)

20-

1 8 -

1 6 -

1 4 -

212-- P 10-

8-

!1 6

Mean Pressure

n 30

• Intact

[ ] Open Defect

[ ] Filled Defect

H I I

60 Flexion Angle (degrees)

n 90

g

30 60 Flexion Angle (degrees)

90

20 t 18 P e a k Pressure

60 Flexion Angle(degrees)

90

Fig 5. Bar graphs of mean patel- Iofemoral contact parameters. Harvest of a mid-third bone-pa- tellar tendon graft had no mea- surable effect on patellofemoral mechanics (n = 17; error bars = 1 standard deviation).

significantly less than those measured with an empty defect at 30 ° (p < .02) and 90 ° flexion (p < .001). Average axial strains measured with the defect filled were not significantly different than average axial strains measured for intact patellae (fig 3). Cartilage condition had no effect on the medial, lateral, or aver- aged axial strain.

Patellofemoral Contact

As knee flexion angle increased from 30 ° to 90 °, patellofem- oral contact area increased and shifted superiorly on the patella (fig 4). Mean contact pressure and peak contact pressure also increased significantly with increasing flexion angle. However, the condition of the anterior patella (intact, with graft defect, or with graft defect filled) did not affect patellofemoral contact (fig 5).

The amount of articular degeneration did not significantly alter peak pressure; however, at 90 ° there was a trend toward greater peak pressure with increasing degeneration (p < .068). At this flexion angle, peak pressure for intact patellae ranged from 9.81 _+ 1.49MPa for patellae with normal cartilage up to 13.23 _+ 3.27MPa for moderately degenerative patellae.

Contact force, the product of contact area and mean pressure, magnified the effects of flexion angle (fig 5). Force distribution across the patellae was concentrated within the central third; the mean contact force for all flexion angles in this region was 738 + 550N, compared to a lateral contact force of 354 _+ 341N and a medial force of 482 _+ 390N. Graft harvest or articular degeneration did not change the magnitude or distribution of pateUofemoral force transmission.

Relationships Between Patella Strain and Patellofemoral Contact

Backward stepwise linear regression analyses, done on data from intact patellae, revealed weak but significant relationships

between patellar strain and patellofemoral contact. Significant interrelationships were noted for axial strain in the lateral side (R 2 = .74, p < .001) and medial side (R 2 = .50, p < .001) of the patella. Mean axial strain at the horizontal midline was related to a linear combination of total patellofemoral contact force, contact force in the central region of the patella, and peak patellofemoral contact pressure (R 2 = .639, p < .001).

Destructive Testing and Failure Analysis

Fourteen of the 17 knees subjected to destructive testing (after nondestructive testing with the patella defect filled) failed, and three knees exceeded the limits of the apparatus (quadriceps force greater than 4,200N) without apparent damage. The mean quadriceps force at failure for the 14 knees was 3,042 _+ 738N, and the mean extension moment at failure was 129.4 _+ 31.7Nm. For the five knees that failed during the course of nondestructive loading, mean quadriceps force at failure was 1,764 +_ 323N, and mean extension moment at failure was 66.2 _+ 10.7Nm. All of these knees failed in 60 ° flexion; four failed during testing with an empty defect, and one failed after the defect was filled. Combining data for the 19 failed knees yielded an average quadriceps load at failure equal to 2,706 + 866N with a corre- sponding mean extension moment of 112.8 _+ 39.6Nm. Fifteen of the 19 specimens failed by transpatella fracture, with the fracture line tending to originate at the superior corners of the void created by graft harvest (fig 6). Two knees failed by mid- substance rupture of the remaining patellar tendon, one failed at the patellar insertion site, and one at the tibial insertion site.

The 14 knees that failed during destructive testing were used for stepwise linear regression analysis (fig 7). The most signifi- cant (p = .0043) and predictive (R 2 = .76) regression equation incorporated age and sex as independent variables and is given below.

Arch Phys Med Rehabi! Vol 78, March 1997

PATELLOFEMORAL BEHAVIOR AFTER HARVEST, Sharkey 261

Fig 6. Transverse fracture of the patellar originating at the superior ex- tent of the graft harvest site and exposing the articular surface of the femur. This mode of extensor mechanism failure after patellar tendon- bone graft harvesting occurred in 15 of 19 subjected to destructive test- ing.

Extension Moment at Failure (Nm)

= 299.3 - 1.8 (Age) - 47.6 (Sex).

For the sex variable, male was set equal to 0 and female equal to 1.

DISCUSSION

Rehabilitation following reconstruction of the ACL of the knee using patellar tendon graft can be complicated by the residual morbidity of the graft harvest. This includes anterior knee pain, possibly due to altered patellofemoral mechanics, and weakening of the extensor mechanism with risk of tendon or patellar fracture if the rehabilitation program is too aggres- sive. HI In this study we investigated the biomechanical basis for these problems and evaluated the utility of bone-grafting the patellar defect as a means of mitigating the effects of graft harvest.

We found that the patellar defect created by harvesting a patellar tendon graft caused significant changes in axial patellar strain. No associated changes in patellofemoral contact were detected, however. Normal patellar strain patterns were partially restored by fiUing the defect with polymethylmethacrylate to

simulate a healed bone graft. Forceful simulated quadriceps contraction induced extensor mechanism failure, primarily by transpatellar fracture, at extension moments well below those measured in maximum voluntary quadriceps contraction in healthy subjects. 18-2~

We were limited by availability to using elderly cadavers, whereas ACL rupture typically occurs in younger patients. The repeated-measures design of the study, with each specimen serv- ing as its own control, makes extrapolation to a younger patient population reasonable. It is also germane that postinjury disuse results in rapid loss of muscle strength and bone density in the injured leg, rendering it more comparable to the elderly specimens we studied. Sievanen et a] 26 measured a 22.4% loss of bone mineral density in the patella and a 50% loss of exten- sion muscle strength in the injured leg of a 26-year-old woman athlete at 15 weeks after ACL rupture and reconstruction. At 1 year postinjury, muscle strength in the injured leg had returned to the preinjury level, but bone mineral density of the patella was still reduced by 7%.

Steen and associates 16 measured anterior patellar surface strains before and after bone-tendon harvest, noting small but significant changes in strain magnitude after harvest. Similarly, Friis et al27 using thermoplastic techniques, found increased stresses in the patella adjacent to the defect created by harvest. These studies used similar experimental designs in which fresh cadaver knees were mounted at specific flexion angles and loaded by tension through the quadriceps tendon near its patellar insertion. However, the quadriceps forces and resultant exten- sion moments used for these investigations were less than one third of those used in the present study and only one sixth of potential physiologic force.

We observed a return toward normal axial patellar strains following filling of the patellar defect with polymethylmeth- acrylate to simulate a healed bone graft, suggesting that grafting the defect as recommended by Daluga et al~2 at least partially restores stress-distributing mechanisms of the patella. Fatigue damage of cyclically loaded bone is directly related to both the number of cycles and the strain magnitudeZ*'29; by reducing

200 _

~ 150_

i 100.

50.

O

• • R 2= 0.74

. . . . I . . . . I . . . . I . . . . [ 50 100 150 200

Measured Fai lure M o m e n t (Nm)

Fig 7. Scatter plots comparing the experimentally determined extension moments at which catastrophic failure of the post-graft harvest exten- sor mechanism occurred, with the failure extension moments predicted using linear regression with age and sex as the independent variables.

Arch Phys Med Rehabil Vol 78, March 1997

2 6 2 PATELLOFEMORAL BEHAVIOR AFTER HARVEST, Sharkey

strains, grafting would reduce fatigue damage and thereby de- crease the risk of fracture. The timing of clinical failures of the extensor mechanism (3 to 9 months after ACL reconstruction) is consistent with weakening of the patella by the accumulation of fatigue damage superimposed on disuse osteopenia. One of us (RAM) has noted that on knees treated with patellar tendon graft reconstruction for an ACL rupture, the patellar defect re- mains palpable for up to 3 years after surgery, indicating that spontaneous refilling of the ungrafted patellar defect occurs very slowly if at all. These observations suggest that the patellar defect should probably be bone-grafted at the initial surgery.

Our pateUofemoral contact data for intact patellae agree well with data from previous investigations. 14'a3-25 Increased patello- femoral loading at a given angle of flexion is accommodated predominately by increases in joint contact pressure, and much less so by increases in joint contact area. The migration of articular contact from the distal portion of the patella proximally with increasing flexion angle that we observed has also been well documented by several investigators. 23'2~'3°'3~

The relative reduction in ultimate strength of the knee exten- sor mechanism after graft harvest, as compared to intact knees, could not be directly measured. Preliminary attempts to induce rupture or patellar fracture in intact extensor mechanisms were unsuccessful either because their strength exceeded the abilities of the experimental apparatus (4,200N of quadriceps force) or because fractures occurred at the interfaces between the tibia or femur and the mounting fixtures. To circumvent this problem we compared the extension moments at failure to normal knee extension moments measured in other studies ~8-2~ and found that following graft harvest knees failed at moments of about 80% of those produced by maximum voluntary isometric contraction at 60 ° knee flexion. This comparison should be viewed with caution, given that the maximum extension moment values were measured in young healthy individuals while the mean age for our specimens was 72 years. Postinjury loss of strength would place a young subject with ACL rupture between these two extremes. Despite these comparative limitations, the extension moments and quadriceps forces at which failure occurred in our postoperative preparations represent a substantial reduction in the ultimate strength of the extensor mechanism even with the patellar defect filled. The patellar fractures that occurred during failure of our specimens were similar to those observed clini- cally by Bonatus and Alexander 2 and by McCarroll s and empha- size the need for careful control of rehabilitation following mid-third patellar tendon reconstructions of the ACL. Daluga L2 recommended filling the defect to protect against these failures, a recommendation supported by our strain measurements.

Acknowledgment: This work was funded by Interpore Interna- tional. The authors wish to thank Timothy Moseley and Robert Kincaid for their technical assistance.

References 1. Bonamo JJ, Krinick RM, Sporn AA. Rupture of the patellar liga-

ment after use of its central third for anterior cruciate reconstruction. J Bone Joint Surg 1984;66-A:1294-7.

2. Bonatus TJ, Alexander AH. Patellar fracture and avulsion of the patellar ligament complicating arthroscopic anterior cruciate liga- ment reconstruction. Orthop Rev 1991;20:770-4.

3. DeLee JC, Craviotto DF. Rupture of the quadriceps tendon after a central third patellar tendon anterior cruciate reconstruction. Am J Sports Med 1991; 19:415-6.

4. Graf B, Uhr F. Complications of intra-articular anterior cruciate reconstruction. Clin Sports Med 1988;7:835-48.

5. Hughston JC. Complications of anterior cruciate ligament surgery. Orthop Clin North Am 1985; 16:237-40.

6. Langan P, Fontanetta AP. Rupture of the patellar tendon after use of its central third. Orthop Rev 1987; 16:317-21.

7. Marder RA, Rasldnd JR, Carroll M. Prospective evaluation of arthroscopically assisted anerior cruciate ligament reconstruction: patellar tendon versus semitendinosus and gracilis tendons. Am J Sports Med 1991; 19:478-84.

8. McCarroll JR. Fracture of the patella during a golf swing following reconstruction of the anterior cruciate ligament: a case report. Am J Sports Med 1983; 11:26-7.

9. Paulos LE, Rosenburg RD, Drawbert J. Infrapatellar contracture syndrome: an unrecognized cause of knee stiffness with patella entrapment and patella infera. Am J Sports Med 1987; 15:331-41.

10. Sachs RA, Daniel DM, Stone ML, Garfein RF:. Patellofemoral prob- lems after anterior cruciate ligament reconstruction. Am J Sports Med 1989; 17:760-5.

11. Tibone JE, Antich TJ. A biomechanical analysis of anterior cruciate ligament reconstruction with the patellar tendon: a two year fol- lowup. Am J Sports Med 1988; 16:332-5.

12. Daluga D, Johnson C, Bach BR. Primary bone grafting following graft procurement for anterior cruciate ligament insufficiency. J Arthro Rel Surg 1990;6:205-8.

13. D'agata SD, Pearsall AW, Reider B, Draganich LF. An in vitro analysis of patellofemoral contact areas and pressures following procurement of the central one-third patellar tendon. Am J Sports Med 1993;21:212-9.

14. Eilerman M, Thomas J, Marsalka D. The effects of harvesting the central one-third of the patellar tendon on patellofemoral contact pressure. Am J Sports Med 1992;20:738-41.

15. Goldstein SA, Coale E, Weiss AP, Grossnickle M, Meller B, Mat- thews LS. Patellar surface strain. J Orthop P(es 1986;4:372-7.

16. Steen H, Tseng KF, Carpenter JE, Goldstein SA. Patella surface strain changes with removal of bone block fo r ACL graft [abstract]. Trans Orthop Res Soc 1994; 19:607.

17. Sharkey NA, Smith TS, Lundmark DC. Freeze clamping musculo- tendinous junctions for in-vitro simulation of joint mechanics. J Biomech 1995;28:631-8.

18. Haffajee D, Moritz U, Svantesson G. Isometric knee extension strength as a function of joint angle, muscle length and motor unit activity. Acta Orthop Scand 1972;43:138-47.

19. Lieb FJ, Perry J. Quadriceps function. An electromyographic study under isometric conditions. J Bone Joint Surg 1971;53-A:749-58.

20. Lindahl O, Movin A, Ringqvist I. Knee extension. Measurement of the isometric force in different positions of the knee-joint. Acta Orthop Scand 1969;40:79-85.

21. Smidt GL. Biomechanical analysis of knee flexion and extension. J Biomech 1973;6:79-92.

22. Olson SA, Bay B, Chapman MW, Sharkey NA. Biomechanical consequences of fracture and repair of the posterior wall of the acetabulum. J Bone Joint Surg 1995;77-A:1184-92.

23. Huberti HH, Hayes WC. Patellofemoral contact pressures. The in- fluence of Q-angle and tendofemoraI contact. J Bone Joint Surg 1984;66-A:715-24.

24. Huberti HH, Hayes WC. Contact pressures in chondromalacia patel- lae and the effects of capsular reconstructive procedures. J Orthop Res 1988;6:499-508.

25. Marder RA, Swanson TV, Sharkey NA, Duwellius PJ. Effects of partial patellectomy and reattachment of the patellar tendon on the patellofemoral contact areas and pressures. J Bone Joint Surg 1993; 75-A:35-45.

26. Sievanen H, Kannus P, Heinonen A, Oja P, Vuori I. Bone mineral density and muscle strength of lower extremities after long-term strength training, subsequent knee ligament injury and rehabilita- tion. Bone 1994; 15:85-90.

27. Friis EA, Cooke FW, McQueen DA, Henning CE. Effect of bone block removal and patellar prosthesis on stresses in the human patella. Am J Sports Med 1994;22:696-701.

28. Caler WE, Carter DR. Bone creep-fatigue damage accumulation. J Biomech 1989;22:625-35.

29. Carter DR, Caler WE. A cumulative damage model for bone frac- ture. J Orthop Res 1985;3:84-90.

30. Goodfellow J, Hungerford DS, Zindel M. Patello-femoral joint me- chanics and pathology, functional anatomy of the patello-femoral joint. J Bone Joint Surg 1976;58-B:287-90.

31. Ficat RP, Hungerford DS. Disorders of the patellofemoral joint. Baltimore: Williams and Wilkins, 1977.

Arch Phys Med Rehabil Vol 78, March 1997

PATELLOFEMORAL BEHAVIOR AFTER HARVEST, Sharkey 2 6 3

Suppliers a. Istron Corporation, 100 Royall Street, Canton, MA 02021. b. Model LCF-500; Omega Engineering, Inc., PO Box 2669, One

Omega Drive, Stamford, CT 06907-0047. c. Model 52-355; Industrial Devices Corporation, 64 Digital Drive,

Novato, CA 94949-5704. d. Series 500 measurement and control system; Keithley Metrabyte/

Asyst/DAC Data Acquisition Division, 440 Myles Standish Boule- vard, Taunton, MA 02780.

e. Model CAE-06-125UN-350; Measurements Group, Inc., PO Box 27777, Raleigh, NC 27611.

f. Itochu International Inc., 335 Madison Avenue, New York, NY 10017.

g. American International Electric Company, 9830 Bell Ranch Drive, Santa Fe Springs, CA 90670.

h. Abaton Scan 300/GS; Everex Systems, Inc., 48431 Milmont Drive, Fremont, CA 94538.

i. Pulnix American, Inc., 1330 Orleans Drive, Sunnyvale, CA 94089.

Arch Phys Med Rehabil Vol 78, March 1997