Anterior knee pain has a number of underlying causes. Anterior tendon abnormalities include quadriceps and patellar tendon tear and degeneration [1, 2]. Chondromalacia patella is another possible cause of anterior knee pain [3]. Each of these abnormalities is effectively assessed and diagnosed on MR images. Abnormalities of the infrapatellar fat pad have also been described [4].

Three normal fat pads are located about the anterior knee: the quadriceps (anterior suprapatellar), the prefemoral (posterior suprapatellar or supratrochlear), and Hoffa (infrapatellar) fat pads [4–6]. Abnormal conditions that involve the Hoffa fat pad include trauma or inflammation to the infrapatellar plica, which may appear as abnormal fluid signal in the Hoffa fat pad [7]. In addition, impingement of the Hoffa fat pad, also known as Hoffa's disease, is characterized in the acute stage by an edematous and hemorrhagic-appearing infrapatellar fat pad often with mass effect [4].

In our practice, we noted that the quadriceps fat pad may show enlargement or mass effect on the adjacent suprapatellar recess. In addition, the quadriceps fat pad may appear to be of intermediate or fluid signal intensity. The purpose of this investigation was to characterize the MRI appearance of the quadriceps fat pad and to correlate these findings with other knee abnormalities, anatomic measurements of the extensor mechanism, and findings in the patient's history and at physical examination.

Institutional review board approval was obtained before the initiation of this investigation. Consecutive patients who had undergone MRI of the knee in April 2002 were identified. After we excluded patients with prior knee ligament reconstruction, prior extensor mechanism repair, or evidence on MRI of prior knee surgery, the study group consisted of 92 knee MR images from 84 patients. The patients were imaged using a dedicated extremity coil with a 1.5-T magnet (Signa, General Electric Medical Systems) and included sagittal intermediate T1-weighted proton-density–weighted spin–echo images (TR range/TE, 650–1,000/14; 3-mm slice thickness; 0.5-mm gap; 14-cm² field of view; 256 × 224 matrix; and 1 excitation) and axial, sagittal, and coronal proton density–weighted fatsaturation fast spin-echo images (3,016–4,500/16; 4-mm slice thickness; 1-mm gap; 14- to 16-cm² field of view; 256 × 192 matrix; and 3 excitations).

MR images were retrospectively reviewed by two fellowship-trained radiologists (experience, 8 and 9 years) with an opinion rendered by consensus. The reviewers were blinded with respect to patient history and clinical findings. With regard to the quadriceps fat pad, the presence of mass effect on the suprapatellar recess was recorded, evident by convex posterior contour. In addition, predominant quadriceps fat pad signal intensity was characterized as fat intensity, intermediate intensity (equal to muscle or hyaline cartilage), or fluid intensity on proton density–weighted fat-saturation MR images.

The two radiologists also gathered the following retrospective data by consensus: The signal intensity of the prefemoral fat pad and Hoffa fat pad was recorded as equal to fat, intermediate signal (equal to muscle or cartilage), or fluid signal on proton density–weighted fat-saturation MR images. Presence or absence of prepatellar edema (fluid signal on proton density–weighted MR images) was noted. The presence of the articular muscle was recorded, and maximal anteroposterior thickness, if present, was measured. The presence of medial synovial plica was recorded and was considered abnormal if greater than 2 mm thick or elongated between the patellar and femoral articulation. Synovitis (truncation or scalloping of the prefemoral fat pad, defects or displacement of the Hoffa fat pad, or nonvisualization or irregularity of the infrapatellar fat pad) was recorded [6]. The quadriceps tendon and patellar tendon were characterized as having full-thickness tear (tendon discontinuity and retraction), partial-thickness tear (intratendinous fluid signal but no complete disruption), or tendinosis (intratendinous intermediate signal equal to muscle). Patellar and trochlear chondromalacia was graded from 0 to 4 (0, normal; 1, signal abnormality without defect; 2, defect < 50% of cartilage thickness; 3, defect > 50% of thickness; 4, full-thickness defect) [8]. Joint fluid was also graded as physiologic (< 5 mm distention of the suprapatellar recess), small (5-mm to 1-cm distention), and large (> 1 cm distention).

An additional reviewer retrospectively measured various structures on the MR images. On a mid-sagittal image of the patella, the total length of the patella was measured from its most superior to most inferior extent. The patellar tendon length and patellar articular cartilage sagittal length were also measured. The femoral sulcus angle (angle of the anterior trochlear groove) was measured on the axial sequence at a space approximately 10 mm above the distal femur. In addition, the maximal anteroposterior thickness of the quadriceps fat pad was measured and recorded.

Each of the MR images was prospectively interpreted, and the findings were reported by one of six musculoskeletal fellowship-trained radiologists as part of their daily clinical assignment. One reviewer who did not prospectively interpret the MR images reviewed the MRI reports retrospectively and recorded information concerning the meniscus (no tear, equivocal tear, definite tear), anterior cruciate ligament (normal, partial-thickness tear, full-thickness tear), posterior cruciate ligament (normal, abnormal), and medial and lateral collateral ligaments (normal, abnormal). This information was recorded without knowledge of the clinical history.

One reviewer then retrospectively evaluated the referring physician's notes for clinical indicators of anterior or patellofemoral knee pain as determined by history and physical examination. The clinical history and physical examinations were categorized as unequivocally positive for anterior or patellofemoral knee pain or other (to include other knee pain or cases in which it was unclear if the patient had anterior knee pain).

The data pertaining to the quadriceps fat pad (mass effect, signal intensity, and anteroposterior thickness) were compared with all other data using the chi-square or Student's t test to determine any statistically significant association.

The 92 knee MRI examinations were from 84 patients (42 men, 42 women). The average age was 46 years (range, 20–74 years). With regard to the quadriceps fat pad on MRI, mass effect on the suprapatellar recess was noted in 12% (11/92) (Fig. 1A, 1B, 1C). The average anteroposterior measurement of the quadriceps fat pad from the 92 knees was 7.5 mm (range, 4–12.2 mm). The average size of the quadriceps fat pad having mass effect was 8.1 mm (range, 7.4–12.2 mm), which was significantly different from the average size of 6.9 mm (range, 4–10.5 mm) without mass effect (p = 0.0006). No significant relationship was seen between patient age and mass effect (p = 0.20) or fat pad size (p = 0.96).

On MR images, 46% (42/92) of the quadriceps fat pads had signal intensity equal to fat, 51% (47/92) had intermediate signal, and 3% (3/92) had fluid signal on proton density–weighted fat-saturation sequences. Of the 11 quadriceps fat pads with mass effect on the suprapatellar recess, 91% (10/11) had signal intensity greater than fat (9/11 were intermediate signal and 1/11 were fluid signal). In the remaining 81 cases without mass effect, 49% (40/81) of the quadriceps fat pads had signal intensity greater than fat (38/81 had intermediate signal and 2/81 had fluid signal). Of those quadriceps fat pads with fat signal, the average size was 7.0 mm. Those of intermediate signal had an average size of 7.8 mm, and those with fluid signal had an average size of 8.5 mm. Quadriceps fat pad mass effect on the suprapatellar recess was associated with quadriceps fat pad signal intensity (χ² = 7.19, p = 0.0274). If one assumes that there is a progression of signal intensity of the quadriceps fat pad from fat to intermediate signal to fluid signal, then the proportion of individuals with mass effect increases as fat pad signal changes progress to fluid signal (χ² = 7.18, p = 0.0074). No significant relationship was found between quadriceps fat pad signal intensity and patient age (p = 0.45).

With regard to the other fat pads about the knee, the prefemoral fat pad was equal to fat signal in 99% (91/92), intermediate signal in 0% (0/92), and fluid signal in 1% (1/92) on proton density–weighted MRI sequences. The Hoffa fat pad was equal to fat in 89% (82/92), intermediate signal in 3% (3/92), and fluid signal in 8% (7/92) (Fig. 2). There were no significant associations between quadriceps fat pad mass effect and the signal intensity of the prefemoral fat pad (χ² = 0.1358, p = 0.7125) or the Hoffa fat pad (χ² = 0.4430, p = 0.8013).

With regard to measurements of anatomic structures, the average patellar length was 41.2 mm (range, 32.8–51.4 mm), the average patellar articular length was 30.6 mm (range, 21.5–37.9 mm), and the average patellar tendon length was 45.7 mm (range, 30.4–60.9 mm). The anterior femoral sulcus angle was 117.8° (range, 100.5–137.1°). When we evaluated the presence or absence of quadriceps fat pad mass effect, no significant differences were found in the measurements of patellar length (p = 0.4164), patellar articular length (p = 0.5859), patellar tendon length (p = 0.7587), and anterior femoral sulcus angle (p = 0.8350).

With regard to joint effusion and synovitis, joint fluid was considered physiologic or absent in 39% (36/92), small in 37% (34/92), and large in 24% (22/92). Synovitis was considered present in 11% (10/92) and absent in 89% (82/92). Prepatellar edema was present in 98% (90/92) (Fig. 2) and absent in 2% (2/92). There were no significant associations between quadriceps fat pad mass effect and joint effusion (χ² = 4.8530, p = 0.0883), synovitis (χ² = 0.0404, p = 0.8408), or prepatellar edema (χ² = 2.7803, p = 0.3534).

The quadriceps tendon showed normal appearance in 60% (55/92), tendinosis in 40% (37/92), partial-thickness tear in 0% (0/92), and full-thickness tear in 0% (0/92). The patellar tendon showed normal appearance in 67% (62/92), tendinosis in 32% (29/92), partial-thickness tear in 1% (1/92), and full-thickness tear in 0% (0/92). There were no significant associations between quadriceps fat pad mass effect and quadriceps tendon findings (χ² = 0.8613, p = 0.3534) or patellar tendon findings (χ² = 0.2573, p = 0.8793). The articular muscle was visualized in 80% (74/92), ranging from 1- to 8-mm anteroposterior thickness.

In the assessment of chondromalacia, the medial facet of the patella was normal in 53% (49/92), whereas 15% (14/92) showed grade 1 chondromalacia; 9% (8/92), grade 2; 12% (11/92), grade 3; and 11% (10/92) showed grade 4 chondromalacia on MRI.

With regard to the lateral facet of the patella, 40.2% (37/92) were normal; 28.3% (26/92) showed grade 1 chondromalacia; 9.8% (9/92), grade 2; 5.4% (5/92), grade 3; and 16.3% (15/92) showed grade 4 chondromalacia on MRI. With regard to the trochlea of the femur, 66% (61/92) were normal; 8% (7/92) showed grade 1 chondromalacia; 10% (9/92), grade 2; 4% (4/92), grade 3; and 12% (11/92) showed grade 4 chondromalacia on MRI. There were no significant associations between quadriceps fat pad mass effect and chondromalacia of the trochlea (χ² = 4.3879, p = 0.3561), medial facet (χ² = 4.9683, p = 0.2906), or lateral facet (χ² = 3.9645, p = 0.4108). A medial plica was identified in 43% (40/92) and was considered abnormally thick (> 2 mm) or elongated in 3% (3/92) with no significant association with quadriceps fat pad mass effect (χ² = 0.7966, p = 0.6714).

Retrospective review of MRI reports showed medial meniscus abnormalities in 51% (47/92) (equivocal tear in 1/92 and definite tear in 46/92), lateral meniscus abnormalities in 29% (27/92) (equivocal tear in 3/92 and definite tear in 24/92), and no significant association with quadriceps fat pad mass effect (χ² = 1.1416, p = 0.5651 and χ² = 1.5917, p = 0.4512, respectively). With regard to knee ligaments, the anterior cruciate ligament was abnormal in 17% (16/92) (partial-thickness tear in 2/92 and full-thickness tear in 14/92), the posterior cruciate ligament was abnormal in 3% (3/92), the medial collateral ligament was abnormal in 23% (21/92), and the lateral collateral ligament was abnormal in 3% (3/92). There was no association between quadriceps fat pad mass effect and anterior cruciate ligament (χ² = 0.3410, p = 0.8433), posterior cruciate ligament (χ² = 0.4166, p = 0.5187), medial collateral ligament (χ² = 1.3235, p = 0.254), or lateral collateral ligament findings (χ² = 0.4166, p = 0.5181).

Retrospective review of clinical notes indicated the presence of defined anterior knee pain by history in 18% (17/92) and at physical examination in 15% (14/92). There was a statistically significant association between anterior knee pain at physical examination and the presence of quadriceps fat pad mass effect on the suprapatellar recess (χ² = 8.76, p = 0.0031). In the 14 patients with anterior knee pain at physical examination, the average anteroposterior quadriceps fat pad thickness was 8.1 mm (range, 5.6–12.2 mm), and mass effect of the quadriceps fat pad was present in 36% (5/14). In the 78 patients without defined anterior knee pain at physical examination, the average anteroposterior quadriceps fat pad thickness was 7.4 mm (range, 4–10.5 mm), and mass effect of the quadriceps fat pad was present in 8% (6/78). Anterior knee pain at physical examination was also significantly associated with an abnormal medial collateral ligament (χ² = 4.83, p = 0.0031) and anterior knee pain by history (χ² = 22.76, p < 0.0001). There were no other statistically significant associations between anterior knee pain at physical examination or by history and other data.

The results of this study show that mass effect of the quadriceps fat pad on the suprapatellar recess was identified in 12% (11/92) of consecutive knee MRI examinations, and the quadriceps fat pad was of intermediate or fluid signal intensity in 54% (50/92) on proton density–weighted images with fat saturation. The finding of quadriceps fat pad mass effect on the suprapatellar recess was significantly associated with anterior knee pain at physical examination. In addition, 100% (5/5) of patients with quadriceps fat pad mass effect and anterior knee pain had fat pad signal intensity greater than fat and equal to muscle or hyaline cartilage on proton density–weighted images with fat saturation. As quadriceps fat pad signal increased to intermediate signal and fluid signal on fluid-sensitive sequences, mass effect was more likely to be present.

The quadriceps or suprapatellar fat pad is a normal fat pad, positioned between the distal quadriceps tendon anteriorly and the suprapatellar recess posteriorly. It normally measures 6 mm (range, 4–8 mm) in women and 7 mm (range, 5–9 mm) in men [5]. The suprapatellar recess, an extension of the knee joint, does not possess a capsule. Therefore, the posterior surface of the quadriceps fat pad and a segment of the distal quadriceps tendon are lined with synovium [5]. These synovial-lined surfaces articulate with the trochlea during knee flexion; this process increases congruency of the extensor mechanism [5]. An additional anterior knee structure, the articular muscle, is found deep relative to the quadriceps muscle, is present in all individuals, and is routinely visualized on MR images [9]. It originates from the femur as one to seven muscle bundles and inserts on the suprapatellar recess, where it applies tension to the suprapatellar recess during knee extension, protecting the relatively redundant suprapatellar recess from entrapment between the femur and the patella [9].

The criterion for quadriceps fat pad mass effect on the suprapatellar recess in our study was a posterior convex border. Our results show that this subjective assessment is useful because of a significant association between mass effect and measured anteroposterior thickness. However, the range (5.6–12.2 mm) of quadriceps fat pad thickness in patients with anterior knee pain at physical examination did overlap the range (4–10.5 mm) of those patients without anterior pain and the data (4–9 mm) of earlier studies [5]. Therefore, the subjective assessment of the posterior fat pad surface for abnormal convexity is the most effective method of diagnosis of quadriceps fat pad enlargement because this abnormal convexity correlated with anterior knee pain.

The cause of the quadriceps fat pad abnormalities in this study is unknown. However, infrapatellar fat pad enlargement and edema from trauma and impingement have been described in Hoffa's disease [4]. Therefore, one analogous theory is that of quadriceps fat pad impingement. However, none of the patients with enlargement of the quadriceps fat pad had direct trauma or overuse. Two of the five enlarged quadriceps fat pads in patients with anterior knee pain occurred bilaterally in one patient. This finding could suggest a developmental cause related to the anatomy of the extensor mechanism or possibly abnormal mechanics. However, we found no association between quadriceps fat pad enlargement and patellar length, patellar tendon length, patellar articular length, and femoral sulcus angle. In addition, no association was found between thickness of the articular muscle and quadriceps fat pad enlargement. The finding that anterior knee pain at physical examination was significantly associated with quadriceps fat pad enlargement and medial collateral ligament abnormality may suggest a potential biomechanical cause if indeed these two processes are related. However, quadriceps fat pad enlargement was not significantly associated with medial collateral ligament abnormality on MR images.

On further review of the patients with anterior knee pain found at physical examination and quadriceps fat pad enlargement seen on MR images, we attempted to find a common underlying factor such as a patient's body habitus or occupation, especially given bilateral involvement in one patient. It has been shown that at high knee flexion angles, patellofemoral contact is at the proximal patellar pole [10] and the suprapatellar fat pad articulates with the trochlea [5]. We theorized that excessive knee flexion at high angles may be a cause; however, we found no evidence for such association. One patient was a 38-year-old male computer programmer (320 lb [145 kg]), one was a 47-year-old male nurse (232 lb [105 kg]), one (with bilateral involvement) was a 41-year-old female custodial engineer (148 lb [67 kg]), and one was a 32-year-old man with a desk job (175 lb [79 kg]). Because of the small number of patients with anterior knee pain and quadriceps fat pad enlargement, no conclusions can be drawn from these clinical data.

Another potential theory for quadriceps fat pad enlargement is that these changes are secondary to adjacent knee abnormalities, such as chondromalacia, tendon abnormalities, or synovitis. However, no association suggested this cause. One final theory of quadriceps fat pad enlargement is that of a primary or intrinsic cause. This would have to be proven with histologic or pathologic evaluation, which was not available in this study. IV gadolinium compound was not administered in our patients; we might have gained further information by this addition.

Because the exact cause of quadriceps fat pad enlargement associated with anterior knee pain at physical examination is not known, a treatment for this problem is also not known. All our patients with quadriceps fat pad enlargement and anterior knee pain at physical examination were prescribed physical therapy. In addition, one patient had intraarticular corticosteroid injection followed by three Synvisc injections (hylan G-F 20, Wyeth-Ayerst Pharmaceuticals), and another had an intraarticular injection of some unknown medication. Each has shown mild improvement of his or her symptoms. It is not known if direct quadriceps fat pad injection or surgical resection has any benefits because neither was performed in our patients. Symptom improvement after corticosteroid injection of quadriceps fat pad would indicate that the fat pad is the source of the symptoms, rather than simply representing a related imaging sign.

In a recent article, Cothran et al. [7] described the MRI findings of curvilinear or globular fluid signal in the Hoffa fat pad in five patients. They found that this abnormality represented injury or inflammation of the infrapatellar plica (or ligamentum mucosum) and can be a potential cause for knee pain or hemarthrosis [7]. Incidentally, we found a similar abnormality in 8% (7/92) of our patients (Fig. 2). Anterior knee pain revealed by history or at physical examination was not associated with this finding in our study population. Prepatellar edema was present in 98% (90/92) of patients (Fig. 2). Although the cause of this finding is unclear, the high prevalence and lack of significant correlation with anterior knee pain suggest that it may be a physiologic phenomenon related to knee movement or mechanics.

We acknowledge limitations to this study, which was retrospective without IV gadolinium administered to our patients. In addition, surgical proof of the data gathered from MRI reports was not obtained. Another limitation is lack of histologic or pathologic data regarding quadriceps fat pad abnormalities. Last, assessment of quadriceps fat pad characteristics on MR images in control subjects with normal findings was not possible because each patient had a clinical indication for knee MRI.

In summary, mass effect of the quadriceps fat pad on the suprapatellar recess on MRI has a prevalence of 12% and is significantly associated with intermediate or fluid signal intensity of the quadriceps fat pad and anterior knee pain. Further studies are needed to determine the precise cause and treatment for this abnormality.

Address correspondence to J. Jacobson.