Patellar Fracture Risk Is Not Affected by Harvest Length During Bone–Patellar Tendon–Bone Harvest: A Biomechanical and Ultimate Load-to-Failure Analysis

Abstract

Purpose

To biomechanically assess the impact of different patellar defect sizes on the ultimate load to failure (ULTF) and fracture risk immediately after bone–patellar tendon–bone harvest.

Methods

Twelve fresh-frozen mid femur to mid tibia–fibula cadaveric specimens (8 female specimens) with an average age of 75 years (range, 64-88 years) were randomly assigned to a defect length of 20 or 25 mm. A triangular wedge bone block was harvested from the patella, and the defect was measured. The extensor mechanism was loaded until failure with the knee in 30° of flexion. ULTF and failure modes were recorded. Statistical analyses were performed between the 2 groups to identify significance (P < .05).

Results

Analyses were performed on 10 specimens because 2 were excluded. The average ULTF was not significantly different between the 20- and 25-mm groups (2,313 ± 863 N and 2,443 ± 839 N, respectively; P = .815) despite a significantly different percentage of total patellar defect length between the 2 groups (39.5% ± 3.9% and 49.1% ± 4.1%, respectively; P = .005). In 7 specimens, the first point of bony failure was at the patella, whereas in the remaining 3 specimens, it was distal to the patella.

Conclusions

The results of this biomechanical study show that up to a 25-mm-long patellar bone block harvest can be performed safely with a low risk of patellar fracture.

Clinical Relevance

There is concern for complications including an increased risk of patellar fracture at the harvest site after bone–patellar tendon–bone autograft harvest. The risk of this complication based on patellar harvest size is not well understood. This study provides biomechanical insights into assessing fracture risk of patellar harvest between 20 and 25 mm in length.

The bone–patellar tendon–bone (BPTB) autograft consists of a portion of the central aspect of the patellar tendon with its corresponding bone plugs from the patella and tibia. It has historically been referred to as the gold-standard graft for anterior cruciate ligament (ACL) reconstruction, mainly owing to its long-standing track record and widespread use.1, 2, 3, 4 This trend has largely continued, especially in the United States, where it remains the graft of choice for young recreational, high-level collegiate, and professional athletes.^5,6

Although graft choice should take into consideration revision rates and patient-reported outcomes, donor-site morbidity remains a similarly critical aspect of this discussion. Although outcomes with BPTB autograft are consistently good, there are complications associated with BPTB that are almost exclusively related to the graft harvesting technique. These harvest-related complications, particularly regarding the extensor mechanism, include patellar fracture, patellar tendon rupture, and patellar tendinitis, as shown in the existing literature.⁷ Although previous studies have quoted patellar fracture rates after BPTB harvest as low as 0.2% to 1.8%,^8,9 patellar fracture does pose a significantly devastating clinical scenario.^7,10

Despite the severity of this complication, research identifying the specific risk factors for patellar fracture has been limited.11, 12, 13 Sharkey et al.¹² conducted a biomechanical analysis that is closely aligned with our investigation, focusing on patellar strain and load to failure in human cadaveric knees after BPTB harvest. Our study focuses on analyzing patellar harvest size, which was not assessed in their study.

Recognizing the gap in the literature regarding fracture risk relative to the size of patellar harvest, we conducted a biomechanical evaluation. The purpose of this study was to biomechanically assess the impact of different patellar defect sizes on the ultimate load to failure (ULTF) and fracture risk immediately after BPTB harvest. We hypothesized that harvesting a larger portion of the patella would decrease the ULTF, increasing subsequent fracture risk.

Methods

Specimen Preparation

For this study, we obtained 12 fresh-frozen mid femur to mid tibia–fibula cadaveric specimens with no history of bone cancer or surgery from MedCure (Portland, OR). All specimens were thawed at room temperature for 24 hours prior to dissection and testing and were kept moist with regular sprays of saline solution. There were 8 female and 4 male specimens with an average age of 75 years (range, 64-88 years). By use of Sealed Envelope (London, England), half of the specimens were randomly assigned to the 20-mm defect group whereas the other half were assigned to the 25-mm defect group. Skin and subcutaneous tissue were removed. The fibula on each specimen was removed, and all soft tissue in the most distal 3 inches of the tibia on each specimen was removed. The specimens were then potted in 2-inch-diameter × 3-inch-long polyvinyl chloride pipe with Duz-All polymethyl methacrylate (PMMA) all-purpose self-cure acrylic repair material (Keystone Industries, Gibbstown, NJ). Each patella was digitally outlined using a MicroScribe 3D digitizer (Revware, Raleigh, NC) and imported into a computer-aided design program (Rhinoceros 7; TLM, Seattle, WA), in which the area and the length from the inferior pole to the superior pole of the patella were measured.

Biomechanical Testing Setup

The quadriceps was isolated, separated from surrounding soft musculature, and lifted off the anterior femur using a Cobb elevator. The quadriceps was then placed in a clamp such that the bottom edge of the clamp sat 5 cm superior to the superior aspect of the patella. The specimens were placed into an 858 Mini Bionix II servohydraulic testing machine (MTS, Eden Prairie, MN) with the clamp attached to the actuator and the polyvinyl chloride pot secured in a base that prevented movement of the tibia in 6 df. This base was tilted, allowing 30° of knee flexion. We additionally chose 30° of flexion because this is where the patella engages with the trochlea fully and is most likely to fracture (unpublished data, Loudon et al., 2016).

Dry ice was used to freeze the clamp to the quadriceps. The actuator of the servohydraulic machine then pulled the quadriceps upward at a rate of 2 mm/s until failure. The force was recorded from the testing machine to measure the ULTF. The failure mode of each specimen was also recorded (Fig 1).

Fig 1.

Biomechanical testing setup.

Statistical analysis was performed using SPSS software (IBM, Armonk, NY). The means and standard deviations of all variables in each group were calculated, and 2-tailed t tests were conducted to generate P values to compare the mean values between the 2 groups. Significance was determined by P < .05.

Surgical Technique for BPTB Harvest

After exposure of the extensor mechanism, the patellar tendon was identified. The paratenon was sharply incised with a No. 10 blade. The paratenon was then carefully dissected off the patellar tendon to fully expose the anatomic borders of the tendon. The paratenon and prepatellar bursal tissue were similarly dissected off the patella to ensure full visualization of the harvest site. Next, a standard ruler was used to measure the width of the patellar tendon; the central aspect was then measured to a width of 10 mm. This was subsequently harvested sharply using a No. 10 blade in each specimen. Careful attention was taken during this step to ensure that we stayed exactly within the central 10 mm so that the corresponding patellar harvest site was located similarly within the center of the patella in the coronal plane. To ensure that we appropriately visualized the inferior pole of the patella (the lowest extent of the patellar harvest site), we then excised the corresponding harvested central 10 mm of tendon. Next, we used a ruler to mark the harvest site of the patellar bone block. All specimens were harvested for a width of 10 mm. The length of the harvest was randomized to either 20 or 25 mm, as noted earlier, which was marked with a sterile skin marker. Once the planned harvest size was marked, a 10-mm sagittal saw was used to harvest the bone block. The lateral edges of the block were cut first at a 45° angle relative to the flat surface of the patella. We paid careful attention to keep the oscillating blade contained within the planned borders of the harvest. We standardized the depth of the saw blade to 10 mm with a mark on the sagittal saw. Once we fully cut the lateral edges as described, the sagittal saw was used to complete the cut in the transverse plane at the most proximal aspect of the bone block (at either 20 or 25 mm). This was similarly performed for a depth of 10 mm. Finally, a 0.25-inch curved osteotome was used to deliver the harvest bone block off the patella in careful fashion (Fig 2). After the harvest was complete, we used the MicroScribe 3D digitizer to measure the following: length of harvest site, width of harvest site, and depth of harvest site (measured at the deepest, central aspect where the 45° cuts converged). We also measured patellar thickness with a digital caliper.

Fig 2.

Patellar defect.

Results

Analyses were performed on 10 specimens. One specimen in the 25-mm group was excluded because the knee was not flexed 30° during testing. In addition, 1 specimen in the 20-mm group, which had the smallest defect by length percentage, did not fail before our testing machine’s limits were reached at 4,900 N and was therefore excluded. All remaining data were within 3 standard deviations of their means. There were 4 female specimens in the 20-mm group, and there were 3 female specimens in the 25-mm group. The average defect lengths in the 20- and 25-mm groups were 19.8 ± 0.9 mm and 24.7 ± 0.5 mm, respectively (P < .001). This led to the defects taking up significantly different percentages of total patellar length (39.5% ± 3.9% and 49.1% ± 4.1%, respectively; P = .005). The specimens in the 20-mm group had a significantly higher age than those in the 25-mm group (78 ± 7 years vs 68 ± 4 years, P = .018). In addition, the 20-mm group had significantly shallower defects than the 25-mm group (11.5 ± 1.3 mm vs 13.4 mm ± 1.2, P = .048). Otherwise, average specimen donor weight, bone mass density, patellar area, patellar length, patellar thickness, and defect width were not significantly different between the groups (Table 1).

Table 1.

Data From Tested Knee Specimens

Female specimens, n	20-mm-Long Defect (n = 5)	25-mm-Long Defect (n = 5)	P Value
	4	3
Age, yr	78 ± 7	62 ± 4	.018∗
Specimen donor weight, lb	189 ± 85	194 ± 52	.910
BMD, g/cm2	1.12 ± 0.28	1.22 ± 0.32	.613
Patellar area, cm2	19.1 ± 4.8	18.5 ± 2.6	.811
Patellar thickness, mm	19.6 ± 2.7	21.8 ± 1.9	.159
Patellar length, mm	50.5 ± 5.8	50.6 ± 4.5	.993
Defect length, mm	19.8 ± 0.9	24.7 ± 0.5	<.001∗
Defect length %	39.5 ± 3.9	49.1 ± 4.1	.005∗
Defect width, mm	10.8 ± 1.6	11.3 ± 1.5	.575
Defect depth, mm	11.5 ± 1.3	13.4 ± 1.2	.048∗
Ultimate load to failure, N	2,313 ± 863	2,443 ± 839	.815

NOTE. Data are presented as mean ± standard deviation unless otherwise indicated.

BMD, bone mineral density.

^∗Statistically significant (P < .05).

The average ULTF for the 20-mm group was 2,313 ± 863 N, and the average ULTF for the 25-mm group was 2,443 ± 839 N, with no statistically significantly difference between groups (P = .815). Among the 10 tested specimens that failed and were not excluded, the first point of bony failure was at the patella in 7 specimens, with 6 occurring as avulsion fractures involving the patellar tendon and 1 occurring as a hemi-transverse fracture from the medial edge of patella to the defect. The remaining 3 specimens did not fail at the patella but rather at the tibia. It is worthy to note that many specimens appeared to sustain both intra-articular ligamentous failure and bony failure, presumably occurring at either the ACL or posterior cruciate ligament, but it was difficult to determine which type of failure occurred first. Therefore, we recorded the first point of bony failure for all failed specimens (Table 2).

Table 2.

Modes of Failure of Tested Specimens

Mode of Failure	Specimens in 20-mm Group, n	Specimens in 25-mm Group, n
Patellar avulsion fracture	3	3
Hemi-transverse patellar fracture	1	0
Distal fracture	1	2
Exceeded machine limits	1	0
Excluded for not meeting protocol	0	1

Discussion

The most important finding of this study was that no significant difference was observed in the ULTF when comparing the 20- and 25-mm patellar defect groups. Although we understand that defect depth and width play a role in the risk of patellar fracture in relation to volume of the remaining patella, the findings of this study suggest that harvesting a longer BPTB graft, ranging up to a length of 25 mm, can be conducted safely because it shows no inferiority in the ULTF compared with the shorter, 20-mm-long defect. These results challenge the limitations established by Christen and Jakob⁸ because the grafts with the 25-mm harvest length performed similarly to those with the 20-mm harvest length.

Historically, patellar tendon grafts were the most common grafts used in ACL reconstruction because they typically possess more static stability¹⁴ and require fewer revision operations than hamstring grafts.¹⁵ Previous retrospective studies have shown failure rates via patellar fracture after BPTB harvest ranging from 0.2% to 1.8%.^8,9 Currently, quadriceps tendon grafts are gaining popularity because they are associated with less anterior knee pain.¹⁶ The patellar fracture risk remains high for quadriceps tendon graft with patellar bone block, ranging from 3.5% to 8.8%.^17,18 Although the patellar fracture risk is low, fractures can occur either intraoperatively or postoperatively and fracture management ranges from nonoperative management to internal fixation.^8,17,19, 20, 21, 22, 23, 24, 25, 26 Both Fu et al.¹⁷ and Stein et al.²⁶ observed mechanisms of fracture other than graft harvesting and direct trauma, such as fractures sustained from indirect trauma and during strength testing at follow-up visits. Graft harvesting itself can lead to increased fracture risk through imprecise saw cuts, spreading osteotomies, and the use of a patellar bone block that is too large; however, improvement in the current technology has decreased the risk of intraoperative patellar fracture during the harvesting process.⁸ Lee et al.²² and Milankov et al.²³ further observed graft site failure via patellar tendon rupture; however, the incidence was less than the already low incidence of patellar fractures seen in large populations of ACL reconstructions. Although the incidence of patellar fracture does not appear to lead to a difference in subjective outcome scores when assessed at regular intervals postoperatively, the impact of such a complication can certainly alter the overall recovery process.^17,21,24,26

Certain attempts have been made to mitigate the risk of subsequent patellar fractures in patients after ACL reconstruction. One particular method that has been evaluated more recently involves the use of bone grafting to fill the empty patellar space created by harvesting the BPTB graft.^12,27, 28, 29 Sharkey et al.¹² performed a biomechanical study looking at the use of PMMA as a bone graft and found restoration of axial strain to normal values. In a retrospective review by Kato et al.²⁸ that specifically assessed the use of a β-tricalcium bone block and its impact on protrusion and anterior knee symptoms postoperatively, only 2.4% of patients experienced subsequent patellar fracture with the presence of a bone block. In addition, a retrospective review by Shichman et al.²⁹ analyzed patient-reported outcomes and complication rates between patients who received demineralized bone matrix bone graft had significantly lower rates of anterior knee pain in additon to fracture rates after harvest.

Even with the numerous retrospective studies assessing patellar fracture risk, there is a scarcity of biomechanical literature evaluating the effects of quadriceps loading on the maximum load to failure in specimens with patellar defects due to BPTB graft harvest. The first biomechanical study, to our knowledge, was conducted by Sharkey et al.,¹² who evaluated differences in knee flexion and mean extension moments at the time of failure to determine the true mechanisms of failure after ACL reconstruction. They showed that patellar strain, a critical factor in the risk of fracture, could be significantly mitigated by filling the harvest defect with PMMA, suggesting a potential preventive measure for patellar fracture.¹² Most of their knees failed by transverse patellar fracture (78.9%), whereas the other knees failed at the midsubstance of the remaining patellar tendon (10.5%), patellar insertion site (5.3%), or tibial insertion site (5.3%).¹² No failures were seen at the tibia, as in our study. However, in their study, different flexion angles were tested and all knees had failed at 60° of flexion, whereas our model was set to 30°, which may account for the differences seen in our study. Other similar biomechanical studies previously conducted include a study by Wilder et al.¹³ that looked at the patellar fracture rate among previous ACL reconstructions but was performed in the post–total knee arthroplasty setting. Another study biomechanically assessed patellar fracture risk depending on bone plug shape but not according to bone block size.¹¹

After ACL reconstruction, patients initiate isometric quadriceps loading in the immediate postoperative phase.³⁰ The incidence of patellar fractures is generally observed to occur between 8 and 12 weeks postoperatively.^26,31 These fractures are frequently a consequence of indirect trauma, such as abrupt quadriceps activation during rehabilitation protocols.³² Biomechanical analyses have reported the ULTF of BPTB grafts to range from 1,581 to 1,784 N whereas the ULTF of native ACLs is 2,160 N.33, 34, 35, 36, 37 The findings of our study indicate that both BPTB autografts and native ACLs are likely to experience mechanical failure prior to the force magnitude required to induce patellar fracture in the immediate postoperative period. However, given the advanced age and likely reduced bone density of our cadaveric specimens, it is unlikely that patients with similar bone characteristics would generate sufficient force to cause such a fracture during typical activities. These results underscore the importance of tailoring postoperative recommendations and rehabilitation strategies to the individual, considering factors such as age, bone density, and activity level to mitigate fracture risk after BPTB harvest in ACL reconstruction.

Limitations

Several limitations were present during this study. First, this study was not prospectively powered. Purchasing enough cadavers to appropriately power the study was cost prohibitive; therefore, the number of cadavers in each group was decided based on prior cadaveric biomechanical literature. Because of this, there a risk of a β error. Additionally, given the biomechanical nature of the study and the use of cadaveric knees, replication of real-world conditions that may generate patellar fracture was approximated to the best of our ability. However, this model is not likely to truly replicate real-world conditions in terms of the injury mechanism and the typical patient population affected by ACL injury. Furthermore, the generalizability of this type of study is limited because only one injury mechanism was tested but individual patients may have variations in movement patterns, which may impact the true load to failure of a patient. Regarding our setup for quadriceps load application, there was limited visualization of the intracapsular ligaments (ACL and posterior cruciate ligament) as load was applied. One could speculate that these structures failed before failure occurred at either the patella or tibia given that previous studies have shown ACL failure at lower loads than those in our study. Moreover, the primary focus of the study was bony failure, so any ligamentous failure was not considered for further evaluation. Finally, a significant difference in defect depth was found between the groups. This was unintentional and a result of small deviations in the harvest process but could have affected the results because it was not factored into the analysis.

Conclusions

The results of this biomechanical study show that up to a 25-mm-long patellar bone block harvest can be performed safely with a low risk of patellar fracture.

Funding

This study was funded by the Arthrex Sports Medicine Fellowship Grant.

Disclosures

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: R.A.A. reports board membership with ConMed; reports a consulting or advisory relationship with ConMed; and owns equity or stocks in ConMed. K.J.C. reports that financial support was provided by Arthrex; reports board membership with American Orthopaedic Society for Sports Medicine; and reports a consulting or advisory relationship with Arthrex. All other authors (M.J.P., R.N., R.P., T.K., A.D., E.O., C.M.E.) declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.