The main findings of our study revealed that within the central mass of the infrapatellar fat pad, the vessels showed an arrangement of three layers, with the caliber of the vessels increasing from posterior to anterior, that is, towards the patellar tendon. Interconnections between these layers were established by numerous vascular channels. Together, they thus formed a functional unit. Except for the most superficial level, which was only found in the central portions of the infrapatellar fat pad or in the dorsodistal areas of the patellar tendon, both the intermediate and the deep vascular level provided vascular supply to the distal half of the patellar bone.
The vessels with the widest caliber were located in ultimate proximity to the posterior aspects of the patellar tendon and the patellar apex. The latter originated from the inferior genicular arteries and formed anastomoses within the fat pad. We, therefore, confirm the findings of Kohn et al. [8
] who already described this vascular configuration previously, the same applies to the polar vessels described by Scapinelli [11
]. However, our results showed that the transverse infrapatellar arteries established a complex branching pattern both medially and laterally. Especially noteworthy was the three dimensionality of the vascular architecture within the fat pad which had not been previously documented.
Our results revealed that, in the deep vascular layer, only one anastomosis supplied the inferior patellar pole directly by giving off small vessels. Nevertheless, it was found to be connected with the other anastomoses of the deep vascular layer ((3) and (4) in ).
As presented in , we found a proximal, a diagonal (running towards the patellar apex), a horizontal (which anastomosed with the contralateral artery next to the patellar apex), and a distal branch (which, in its further course, turned into a vessel of the intermediate vascular level). On both sides of the patella, we found an ascending artery which anastomosed with the descending branch given off by the ipsilateral superior genicular artery in about the middle third of the lateral and medial margins of the patella posterior to the retinacula. Prior to this anastomosis, both the descending and the ascending arteries gave off small branches in their entire course at regular intervals, which supplied the lateral and medial edges of the patellar bone. Based on our results, we agree with the findings of Björkström and Goldie [13
] that, besides the anterior vascular network [11
] and the infrapatellar anastomoses, further arteries located posteriorly to the retinacula supply the medial and lateral borders of the patella. Interestingly, arteries originating within the quadriceps tendon had been described by other authors [13
] but were not found to be present in our specimens. Instead of these vessels, in our specimens, an anastomosis was established by branches of the superior genicular arteries coursing next to the patellar base and posterior to the quadriceps tendon.
In the superomedial aspect of the patella, we found three parallel running branches originating from the superior medial genicular artery; one of them turned out to be the predominant vessel. After a short distance, the more proximally located branch joined the main branch. The main branch bifurcated next to the superomedial border of the patella. Whilst one branch ran to the anterior aspects of the quadriceps tendon and, therefore, was not available for further pursue, the other branch ran strictly vertical along the medial border of the patella. It soon gave off a small vessel, which followed a tortuous course directed towards the superior pole of the bone. The distally located flanking vessel joined the descending branch after the superior polar vessel had been given off. The branches of the lateral superior genicular artery showed a configuration which was very similar to that of the artery of the medial side. Both, the medial and the lateral superior polar vessels, formed an anastomosis in the superior aspect of the patella. Due to different techniques applied, this anastomosis together with its course had obviously been overlooked by other authors [10
The results of our work indicate the existence of a dual supply of the entire patellar circumference, both anterior and posterior to the retinacula. The anastomoses within the infrapatellar fat pad established a vascular framing of the whole knee cap (Figures and ).
In addition to the findings of Scapinelli [11
], besides the anterior peripatellar ring, an additional arterial supply reached the patella via infrapatellar fat pad and was not limited to its central parts but also found within its superior parts and the alar folds.
Interestingly, the amount of anastomoses found in the specimen which was treated by Sihler's method was not as impressive as the results of the microsurgical dissection, although the vascular architecture behaved very similar to the the aforementioned specimen. According to Slater et al. [17
], this may be due to the relatively high degree of retropatellar arthrosis this individual seemed to have suffered from.
Due to the small vessels under consideration as well as to a high amount of arteriosclerotic changes, angiographically we were only able to trace the main arteries contributing to the patellar supply.
Although numerous surgical interventions are known to jeopardize the vascular supply of the patella [7
], obviously not every intervention inevitably leads to vascular impairment. Due to its characteristic arterial supply, the patella can thus assume a certain degree of tolerance.
However, if a medial parapatellar approach is combined with a lateral release, this may lead to an insufficient supply of the patella and subsequently increase the incidence of bone necrosis [18
] and patellar fractures [19
]. Preservation of the superior lateral genicular artery during lateral release as well as preservation of the infrapatellar fat pad is considered inter alia as reliable options in order to avoid patellar fracture resulting from total knee arthroplasty [20
]. The superior lateral genicular artery may easily be detected in the subsynovial layer 1-2
cm distal to the inferior margin of the vastus lateralis muscle [21
]. When arthroscopically performing a lateral release, Vialle et al. emphasized selective hemostasis of the superior lateral vascular pedicle and visualization of the inferior lateral vascular pedicle in order to minimize the risk of hemarthrosis which had been reported to occur in 10 to 18% of cases [22
Nicholls et al. compared the reduction of blood flow to both the medial and lateral access but did not find any significant difference. From this, they concluded that both the medial and lateral arteries were involved equally in the vascular supply of the patella. Blood flow was found to be reduced to 53% following a medial approach and 27% after a lateral approach. The supply of the patellar tendon was not affected by both approaches. As the infrapatellar fat pad was preserved in all patients, due to its high amount of anastomoses, great importance is attributed to this structure on behalf of the vascular supply of the patellar tendon [23
]. Significant shortening of the patellar tendon after resection of the infrapatellar fat pad was presented in two articles [24
]. As reported by Takatoku et al., postsurgical scar formation after resection of the infrapatellar fat pad led to abnormal shortening of the patellar tendon as well as deforming of the patella in growing rabbits. Moreover, preservation of the infrapatellar fat pad seems to prevent early degeneration of the articular surface of the patellofemoral joint [26
]. Sanchis-Alfonso et al. observed the healing process of the patellar tendon after patellar tendon autograft and concluded that the infrapatellar fat pad as well as the paratenon played an important role in the healing process of the patellar tendon [27
Based on a study in monkeys, Ogata et al. reported that performing a medial parapatellar approach which included a partial resection of the medial portion of the fat pad reduced the blood flow to the patella to 65% of the control animals. If the fat pad was completely removed, the blood flow decreased further to 49% of the control value. Total fat pad removal combined with a lateral release reduced blood flow to a dramatic low amount of 17% of the control value [28
Another procedure which might endanger the patellar supply is the still controversially discussed need for patellar resurfacing. Proponents of resurfacing have justified this procedure with a lower incidence of revision surgery due to persistent anterior knee pain [29
]. In addition, others have argued that forgoing patellar resurfacing has reduced the risk of patella fractures and component loosening [30
Eversion of the patella is typically performed to optimize the operation field. However, as reported by Hasegawa et al., patellar eversion has led to a significant reduction of the patellar blood flow [31
]. Concerning patellar denervation performed as a part of open knee surgery, in another study, no significant differences in outcome between denervated and nondenervated patellae could be detected during a 2-year followup period [32
There are several reasons why we recommend preserving the infrapatellar fat pad especially during total knee arthroplasty: first we believe, taking our results into consideration, that the feeding arteries provided via Hoffa's fat pad play an important role in patella viability, as they may compensate the loss of other feeding arteries caused by invasive access routes. Second, shortening of the patellar tendon may arise with fat pad resection [24
]. Third, fat pad resection has an influence on the biomechanics of the knee joint [33
], as it seems to stabilize the knee joint, especially during extremes of knee motion (flexion angles of less than 20° and greater than 100°) [6
However, if parts of the fat pad have to be resected, we emphasize sparing at least those parts of the fat pad which are located in ultimate proximity to the patellar borders. As far as the central mass of the fat pad is concerned, the fatty tissue next to the patellar tendon should be spared as the strongest anastomoses have been found to be present in this segment which, in turn, are also an integrative part of the arterial supply of the patellar tendon [34
Weaknesses of our study include the small number of cases, so that no reliable conclusions can be drawn with respect to variations of the vascular anatomy within the infrapatellar fat pad (of Hoffa). Counting from another publication, the number of anastomoses varies depending on the degree of retropatellar arthrosis [17
]. Furthermore, definite conclusions about the percentage of blood supply which reaches the patella via infrapatellar fat pad cannot be drawn. Further studies are needed in order to better estimate to which extent the anastomoses within Hoffa's fat pad are in a position to compensate the loss of other feeding arteries.