Insertion Sites of the Anterior Cruciate Ligament

Studies agree, in broad terms, on where the anterior cruciate ligament (ACL) attaches to the femur and tibia. The femoral attachment lies on the posterior part of the medial surface of the lateral femoral condyle (the lateral wall of the intercondylar notch); its anterior margin abuts the lateral intercondylar ridge, and its posterior margin approaches the posterior articular cartilage margin of the lateral condyle. The tibial attachment lies in the anterior intercondylar area. Medially it follows the lateral slope of the medial intercondylar eminence and the cartilage margin of the medial plateau, anteriorly it is bounded by a bony ridge, and laterally and posterolaterally it borders the anterior horn of the lateral meniscus. From the anatomical measurements of Girgis et al. (1975) to histological, CT and three-dimensional MRI studies, the general location of the main fiber attachment and these relationships have been described consistently. The disagreements concern the margins of the attachment: whether the femoral direct insertion reaches the posterior cartilage margin, for example, and whether a small number of loose fibers cross the ridge.

The larger disagreements concern the size and center of the insertion. Published femoral attachment areas range from about 50 mm² to nearly 200 mm², and widths from about 5 mm to 11 mm. The bony tibial attachment measures about 100–110 mm²; with the portions attached to articular cartilage and the lateral meniscus included, it is about 150 mm². These several-fold differences arise mainly because studies measured different parts of the attachment: the whole attachment with its surface membrane and fan-like extension; the area where midsubstance fibers continue into bone after removal of surface tissue; the direct insertion defined histologically by a fibrocartilaginous transition zone; or the combined bony and cartilaginous attachment. Positional percentages also depend on the reference frame. For the same set of femoral insertion outlines, changing only the baseline and border placement of the quadrant grid shifted the group mean center coordinates by about 6 to 16 percentage points (Yahagi et al., 2018).

In the author's view, the normal ACL insertion is best understood as a dense zone formed by the direct insertion of midsubstance fibers, surrounded by a wider area of more loosely attached peripheral fibers. On the femur, the dense zone is a narrow band lying immediately behind the lateral intercondylar ridge, and the fan-like extension spreads mainly posteriorly and distally to the articular cartilage margin. On the tibia, the dense fibers are concentrated anteriorly and medially, aligned along the anterior ridge and the medial intercondylar eminence; laterally they are bounded by the anterior horn of the lateral meniscus, and the posterolateral area consists largely of fat and loose connective tissue. Any citation of insertion size or center position should state the structure measured, the reference frame and the method used to compute the center. Study means and their distributions describe study samples. They are not individual normal ranges, still less standards for judging whether a tunnel position is acceptable.

In this section, "whole attachment" refers to the macroscopically visible attachment including the fan-like extension. "Direct insertion" and "indirect insertion" are used only for studies with histological classification; macroscopically defined parts keep the names used by each study (midsubstance attachment, fan-like extension and so on). Bundle anatomy and observation conditions are discussed in Section 1.1, the contribution of each fiber region to stability in Sections 1.5 and 1.6, and the relationship between tunnel position and outcome in the chapters on reconstruction technique.

1. Extent and Histological Zones of the Femoral Attachment

Classic descriptions

Girgis et al. (1975) dissected 20 cadaveric knees and 24 fresh knees; in 29 knees the femur was split longitudinally to measure distances between the attachment and bony landmarks (p. 217, Table 1). The ACL attached to the posterior part of the medial surface of the lateral condyle as an arc (a segment of a circle), with an almost straight anterior margin, a convex posterior margin and a long axis tilted away from the vertical (p. 218); Fig. 2 gives a mean length of 23 mm. Odensten and Gillquist (1985), in 20 knees cleared of all non-collagenous tissue (donors aged 17–52 years, all accidental deaths), found an oval femoral attachment measuring 18±2 mm by 11±2 mm, with its center 15±3 mm from the junction of the posterior femoral shaft cortex and the proximal surface of the lateral condyle (p. 259, Fig. 3). These are among the few data from young adult specimens.

Colombet et al. (2006) measured the femoral attachment in 7 knees; the case-by-case data in Table 2 give a length of 18.3±2.3 mm and a width of 10.3±2.7 mm. The dimensions in the text do not match Table 2: the Results give 13.9±9.5 mm and 9.3±7.1 mm (p. 987). Recalculation from the seven individual values in Table 2 reproduces the means and SDs of Table 2; individual lengths range from 14.2 to 21.3 mm and cannot produce an SD of 9.5 mm. The statement in the Discussion that the femoral attachment is smaller than the tibial attachment (p. 991) rests on the erroneous text figures. By Table 2, the femoral attachment length (18.3 mm) exceeds the anteroposterior length of the tibial attachment (17.6 mm); only the width is smaller. Table 2 should be used for the dimensions.

Ferretti et al. (2007) outlined the attachment with a marking pen in 16 knees from 8 donors and measured it by three-dimensional laser scanning: total length 17.2±1.2 mm, width 9.9±0.8 mm at the AM–PL junction, and area 196.8±23.1 mm² (p. 1221, Table 1). The authors state that they measured the whole attachment including the surface fibrous membrane (p. 1223), and the area was calculated on the three-dimensional surface rather than as a planar projection. For both reasons, the values cannot be compared directly with studies that measured only the direct insertion or calculated planar areas.

Midsubstance attachment after removal of surface tissue

Mochizuki et al. (2006) divided the ACL into AM and PL portions in 10 embalmed knees and then carefully stripped the surface fibrous membrane from the midsubstance to the femoral attachment. They found two margins on the attachment: a broad margin including the surface membrane, spreading in a fan anteriorly and posteriorly, and a narrow margin where the midsubstance fibers continued directly into bone (p. 358, Fig. 2). In the narrow attachment, the AM portion measured 9.2±0.7 mm and the PL portion 6.0±0.8 mm in length, both 4.7±0.6 mm in width; an area of about 65 mm² was estimated from length and width, not measured (p. 360). A figure legend gives the wrong value: the legend of Fig. 5 gives the PL length as 4.7±0.6 mm, which does not match the 6.0±0.8 mm in the Abstract and Conclusion (p. 359). The Methods describe "Ten paired embalmed cadaveric knees (5 male and 5 female)" (p. 357), from which the number of donors and the pairing cannot be established. This paper stripped the broad attachment as the "surface membrane" (pp. 357, 359); the same group's 2014 study called the same region "fan-like extension fibres" adhering to bone, that is, fibers of the ACL itself. The definition of the tissue changed between the two papers.

Mochizuki et al. (2014), from the same group, examined changes during flexion and extension. Six fresh-frozen knees (3 male donors, 57–64 years) were used for dynamic observation and 22 embalmed knees for quantification. In extension, both the midsubstance fibers and the fan-like extension ran parallel to the notch roof. The fan-like extension fibers were thin and loose, adhered to bone and extended to the articular cartilage margin; tension applied to the midsubstance fibers appeared to be transmitted to the fan-like extension (p. 338). From 15°–30° of flexion, the midsubstance fibers changed direction relative to the femur while the fan-like extension stayed attached, and a fold formed between them that deepened beyond 60° of flexion. The authors cut along the fold at 60° of flexion and measured a midsubstance attachment of 50.8±12.6 mm², a fan-like extension of 91.4±23.7 mm² and a whole attachment of 142.2±24.4 mm² (p. 341, Table 1). Histologically the two fiber types could not be distinguished; the difference lay at the bony attachment, where the midsubstance fibers mostly had a cartilaginous transition zone and the fan-like extension fibers mostly attached to bone without one (pp. 340–342). The value of this study is that it reconciles descriptions of a broad attachment and of a narrow attachment: the authors argued that earlier studies had observed one or both margins (p. 343).

Histological direct insertion

Iwahashi et al. (2010) compared histology and CT in 8 knees from 4 embalmed donors (66–87 years): 4 knees were sectioned obliquely parallel to the notch roof and matched with CT, and 4 were sectioned sagittally (p. S14). Dense central collagen fibers passed into bone through uncalcified and calcified fibrocartilage, forming a direct insertion located in a crescent-shaped depression behind the lateral intercondylar ridge. The surrounding loose fibers attached without a cartilaginous transition, forming an indirect insertion that reached posteriorly to the articular cartilage margin and partly crossed the ridge to attach anterior to it (pp. S15–S17, Fig. 3). At three section levels, the histological width of the direct insertion was 7.8–9.5 mm, and the width of the CT depression differed from it by no more than 0.3 mm; the CT depression measured 17.4±0.9 mm in length, 8.0±0.5 mm in width and 128.3±10.5 mm² in area (pp. S18–S19, Tables 1 and 3).

The area comes from the CT depression: 128.3 mm² is the area of the depression measured on the CT model. Histology confirmed the correspondence between depression and direct insertion only at three section levels in 4 knees; the extent of the direct insertion was not traced point by point.
The posterior limit of the direct insertion is uncertain: The Abstract and Conclusion place the direct insertion between the ridge and the articular cartilage margin, but the legend of Fig. 3 describes a posterior band of loose fibers without cartilaginous transition in contact with the cartilage margin (p. S17). The width of this indirect band was not measured, so how far the direct insertion extends posteriorly cannot be determined.
Values at section level 3 are transposed between figure and table: Fig. 5 labels the histological width as 8.9±0.8 mm and the CT width as 8.6±0.5 mm; Table 1 and the text give the reverse (p. S18). If the range in Table 1 (8–10 mm, 4 knees) is exact, the combination 8.6±0.5 mm is also difficult to reconcile. Values at this level should not be cited.
The comparison with Ferretti et al. mixes measured structures: The authors compared their 128 mm² with the 196.8 mm² of Ferretti et al. and attributed the difference to soft tissue captured by laser scanning (pp. S19–S20). Ferretti et al. measured the whole attachment including the surface membrane, and Iwahashi et al. measured the depression containing the direct insertion; the two were never the same structure.
The tunnel recommendation goes beyond the study: The Discussion recommends placing the tunnel within the direct insertion (p. S20), citing earlier literature that the direct insertion provides the main mechanical connection; this study included no mechanical testing.

Sasaki et al. (2012) compared macroscopic and histological observations in 20 embalmed knees (56–78 years), also sectioned parallel to the notch roof and measured at 4 levels (pp. 1136–1137). On macroscopic observation, with the surface fibrous membrane retained, the attachment was oval in 16/20 knees, 17.7±2.7 mm long and 4.6±0.7 mm wide. Histologically, the direct insertion was 5.3±1.1 mm wide, bounded anteriorly by the lateral intercondylar ridge and posteriorly by a newly described "lateral intercondylar posterior ridge", with its posterior edge 4.4±0.5 mm from the posterior cartilage margin. An indirect insertion lay between the posterior ridge and the cartilage margin, and type I collagen staining showed posterior fibers continuous with the posterior cartilage (pp. 1138–1142, Tables 1 and 2). The distance from the lateral intercondylar ridge to the posterior cartilage margin was 10.1±1.3 mm, of which the direct insertion occupied roughly the anterior half. Iwahashi et al. measured a direct insertion 7.8–9.5 mm wide; Sasaki et al. measured about 5 mm. The former described the direct insertion as reaching the cartilage margin; the latter found an indirect zone of about 4 mm between the direct insertion and the cartilage margin. Both used elderly embalmed specimens and the same section orientation; Sasaki et al. additionally used polarized light and Alcian blue staining to identify fibrocartilage, so the criteria for the direct insertion were not identical. Which result is closer to the living knee cannot be determined from the available data.

The method for the SDs in the summary row is not stated: In Tables 1 and 2, the SDs in the "Mean" row are close to the dispersion of the four level means. For example, the four level means for the distance from the posterior edge of the direct insertion to the cartilage margin are 3.8, 4.5, 4.9 and 4.5 mm, with an SD of about 0.46 mm, and the table gives 0.5 mm; the SD of the four level means for direct insertion width is about 1.04 mm, and the table gives 1.1 mm. Between-specimen SDs within each level are 0.9–2.4 mm (pp. 1138, 1141, Tables 1 and 2). The paper does not state how the summary-row SDs were calculated, so the dispersion of 5.3±1.1 mm in the Abstract and of "about 4 mm" in the text cannot be taken as the variation among the 20 specimens.
The conclusion that the macroscopic attachment "corresponded to the direct insertion" holds only for width: In the same specimens, the posterior and anterior edges of the macroscopic attachment were 7.8 mm and 12.4 mm from the cartilage margin, compared with 4.4 mm and about 10 mm for the direct insertion (p. 1140); the authors acknowledge that the macroscopic attachment lies more anteriorly. The widths are similar (4.6 and 5.3 mm), but the positions differ by about 3 mm and overlap only in part. The word "corresponded" in the conclusion goes beyond the data.

Moulton et al. (2017) used younger specimens: 10 fresh-frozen knees, median age 56.5 years (33–63 years). Under scanning electron microscopy, the center of the attachment showed a four-layer direct insertion in all 10 knees, and the most posterior part showed a two-layer structure without fibrocartilage in all 10, with some fibers growing into the posterior cartilage margin. The interdigitation depth between fibers and bone was 387±81 μm at the direct insertion and 228±75 μm at the indirect insertion (pp. 167–169). Macroscopically, the fan-like extension lay entirely behind the lateral intercondylar ridge in all 10 knees (p. 167). Each specimen was sampled at only two sites, central and posterior, and the boundary between direct and indirect insertion was not mapped. The two sites within each specimen were compared with independent-samples t tests and Mann–Whitney tests, without accounting for pairing, although the difference in interdigitation depth is clear with or without pairing. The authors themselves describe their suggestion to adjust tunnel position according to graft and fixation as speculative (p. 170). The study was funded by the Steadman Philippon Research Institute.

Macroscopic division into midsubstance attachment and fan-like extension

Several later studies divided the attachment macroscopically along the fold or boundary between the midsubstance fibers and the fan-like extension. Iriuchishima et al. (2016), from the Nihon University group, measured a whole attachment of 102 mm², a midsubstance attachment of 56 mm² and a fan-like extension of 45 mm² in 14 embalmed knees (median age 82.5 years; medians, p. 256, Table 1). Suruga et al. (2017) measured 125±47, 67±21 and 59±31 mm² in 23 embalmed knees (median age 83 years), with a midsubstance attachment about 5 mm wide and 15 mm long (p. 1110, Tables 1 and 2). The two studies share the same ethics approval number (20-14) and the same age range (69–96 years); whether the specimens overlap is not stated. Morales-Avalos et al. (2022) used the same macroscopic fold in 81 embalmed Mexican knees: in the groups under 50 years, the direct attachment measured 104–122 mm² and the indirect attachment 17–19 mm²; in the groups over 50 years, 57–61 mm² and 47–56 mm² (p. 3409, Table 4).

Paired comparisons ignored the pairing: Iriuchishima et al. (2016) compared center positions in the same 14 knees with and without the fan-like extension using the Mann–Whitney U test, a test for independent samples (p. 256), without using the paired structure of two tracings of the same knee. The Methods state that data are presented as medians and ranges, yet center positions are given with "±", and the statistic is not explained. The direction and magnitude of the center shift can be used as descriptive results.
The histological basis for the macroscopic boundary cites the wrong literature: Iriuchishima et al. (2016) and Suruga et al. (2017) both state that a significant correlation between the macroscopic and histological borders "has been previously reported". The references cited are, respectively, a triaxial accelerometer study in porcine knees (p. 258, ref. 24) and a biomechanical study comparing 11 o'clock and 10 o'clock femoral tunnel positions (p. 1111, ref. 4); neither concerns attachment histology. Among the available data, Mochizuki et al. (2014) described qualitatively in 6 knees how the bony attachment of the two fiber types differed, but reported no statistics correlating macroscopic and histological boundaries. The macroscopic divisions in the two studies can serve as descriptive data; they cannot be treated as histologically validated direct and indirect insertions.
Suruga et al. (2017) contains further numerical and citation errors: The AM fan-like extension is 27±11.5 mm² in the Abstract and 29±18 mm² in Table 1. Where the Discussion cites a study of selective fiber sectioning and resistance to anterior translation, the reference given (ref. 32) is an MRI study of cervical extensor muscles (p. 1111).
Morales-Avalos et al. (2022) contains multiple contradictions between tables and text: The text states that total attachment area does not change with age, yet Table 4 marks the difference between 145 and 123 mm² in men with P=0.0006; the text states that AM and PL areas are unrelated to age, yet every age comparison of the direct and indirect parts in Table 4 has P<0.0001 (pp. 3407–3409). Figures in the Discussion also differ from the tables: the direct attachment in young women is given as 114 mm², against 104 mm² in Tables 3 and 4. The component areas do not fully add up to the totals: in young men, in whom both bundles could be separated, the indirect attachment areas of AM and PL are 12 mm² each, 24 mm² in total, while the total indirect attachment area is 17 mm² (Tables 3 and 4). According to the statistical methods, age differences between groups were tested with the Mann–Whitney U test, and comparisons between sexes and between age groups with the chi-square test (p. 3406); how a chi-square test was applied to continuous morphometric values is not explained, and the limitations paragraph states that nonparametric tests were used. Table 3 marks several sex differences as nonsignificant, for example short axes of 3.12±0.72 and 4.92±0.85 mm in the group over 50 years; these labels are difficult to reconcile with the means and dispersion in the table and cannot be verified. The morphological category counts and the age-group areas in Table 4 can be cited as descriptive data; the conclusions about age and sex in the text should not be used.

The systematic review by Iriuchishima and Goto (2022) counted studies reporting the midsubstance attachment and fan-like extension (or direct and indirect insertions): of 57 studies, 16 addressed this distinction (Results). It pooled no size or position data, and it omitted the first author's own 2016 center-position study and the electron microscopy study of Moulton et al. (2017). The flow diagram labels 74 as records screened and 57 as full texts assessed, then excludes 17 and still includes 57, which is inconsistent with the 74−17=57 in the text. In Table 1, the citation numbers for 14 of the 16 primary studies point to the wrong references (for example, Mochizuki 2006 is given as ref. 33, which is a different paper). The review can serve only as a pointer, and the original data must be taken from each study. The review states that it was not registered and had no protocol.

The area ratio of fan-like extension to midsubstance attachment ranges from about 1:6, calculated from the young-group values in Table 4 of Morales-Avalos et al. (2022), to about 2:1 in Mochizuki et al. (2014). The boundaries were defined differently (a fold at 60° of flexion, an ink line in extension, a visible crease), and specimen age and preservation also differed, so these ratios cannot be pooled.

Mechanical data on the fan-like extension

Iwahashi et al., Sasaki et al. and Iriuchishima et al. all argued that the tunnel should reproduce the direct insertion or midsubstance attachment, on the grounds that the direct insertion provides the main mechanical connection. None of their studies tested this claim mechanically.

Kanto et al. (2025) tested the role of the posterior fan-like extension in 15 unpaired fresh-frozen human knees. Eight knees kept the attachment intact; in 7 knees, the connection between the midsubstance fibers and the fan-like extension was cut along the fold at 90° of flexion, leaving the fan-like extension itself on the bone (Methods; Limitations). Mean age was about 62.5 years in both groups; all soft tissue other than the ACL was removed, and anterior load was applied at 15° of flexion until failure. In the cut group, the posterior fan-like extension made up 35% of the whole attachment area. Failure load was 910.1±475.0 N in the intact group and 409.5±219.7 N in the cut group (P=0.031); 5/8 intact knees failed in the midsubstance, and 7/7 cut knees failed at the femoral attachment (Results). The Abstract gives the SD for the intact group as 473.3 N, slightly different from the Results. An independent-samples t test on the reported means and SDs gives P≈0.024, also not exactly the reported value, although this does not change the judgment at the 0.05 level.

No sham-cut control: The cut and intact groups differed in two respects: the posterior fan-like extension was no longer connected to the midsubstance fibers, and the attachment carried an additional incision. Failure at the attachment in 7/7 knees could equally be explained by stress concentration at the incision; the authors acknowledge that cutting other regions of the attachment might give similar results (Discussion). The experiment shows that attachment integrity affects failure strength; it cannot attribute the effect to the posterior fan-like extension alone.
Between-group comparison with large individual variation: The groups consisted of different knees, the coefficient of variation of failure load in the intact group was about 52%, and the methods of allocation and age matching were not described.

Sabzevari et al. (2020), from the same laboratory, peeled the posterior fan-like extension off the femur in 16 mature porcine knees. All 8 intact knees ended in fracture (7 femoral, 1 tibial), and all 8 resected knees failed at the femoral attachment (p. 1116). The reported ratio of 3599 N to 392 N compares fracture load with attachment failure load; the failure load of the intact ACL itself was never measured. Porcine anatomy and load values also cannot be converted directly to humans. Taken together, the two studies show that under destructive loading ex vivo, the integrity of the posterior part of the femoral attachment is related to failure load. Neither tested the role of the fan-like extension under physiological loads, and neither tested any reconstruction tunnel position. The statement in the Conclusion of Kanto et al. that a graft more closely reproducing this morphology may lower failure rates is speculation (Conclusion). Excluding the fan-like extension from the ACL attachment is at present a choice of definition, not an established functional conclusion.

Table 1. Structures measured and dimensions of the femoral attachment

Study	Specimens	Structure measured and method	Length (mm)	Width (mm)	Area (mm²)
Odensten and Gillquist (1985)	20 knees, 17–52 years	Whole attachment after removal of non-collagenous tissue, micrometer	18±2	11±2	—
Colombet et al. (2006)	7 knees, mean 75 years	Whole attachment, caliper (Table 2)	18.3±2.3	10.3±2.7	—
Edwards et al. (2008)	22 knees, mostly 60–75 years	Whole attachment, traced on photographs	14±2	7±1	—
Ferretti et al. (2007)	16 knees (8 donors), mean 75 years	Whole attachment including surface membrane, 3D laser scanning	17.2±1.2	9.9±0.8	196.8±23.1
Mochizuki et al. (2006)	10 knees, embalmed	Midsubstance attachment after removal of surface membrane	AM 9.2, PL 6.0	4.7	About 65 (estimated)
Mochizuki et al. (2014)	22 knees, embalmed	Division along the fold at 60° of flexion	—	—	Midsubstance 50.8; fan-like 91.4; whole 142.2
Iwahashi et al. (2010)	8 knees (4 donors), embalmed	CT depression confirmed by histology	17.4±0.9	8.0±0.5	128.3±10.5
Sasaki et al. (2012)	20 knees, embalmed	Histological direct insertion (width)	—	5.3	—
Iriuchishima et al. (2016)	14 knees, embalmed	Macroscopic division, medians	—	—	Whole 102; midsubstance 56; fan-like 45
Suruga et al. (2017)	23 knees, embalmed	Macroscopic division	Midsubstance 15±3	Midsubstance 5±1	Whole 125; midsubstance 67; fan-like 59

Key points

The structure measured determines the numbers: The whole attachment including surface membrane and fan-like extension is about 17–18 mm long and 10–11 mm wide; the midsubstance attachment after removal of surface tissue is about 5 mm wide; reported areas range from about 50 mm² to nearly 200 mm², mainly because of differences in the structure measured and the method.

The direct insertion lies immediately behind the lateral intercondylar ridge: Both histological studies found the direct insertion bounded anteriorly by the lateral intercondylar ridge and surrounded by loose fibers of the indirect insertion. They disagree on whether the direct insertion reaches the posterior cartilage margin (Iwahashi et al., 2010: width about 8 mm; Sasaki et al., 2012: width about 5 mm, posterior edge about 4 mm from the cartilage margin).

Macroscopic divisions are poorly validated: The histological correlation that two Nihon University studies cite for their macroscopic boundary rests on unrelated references; the conclusions of Morales-Avalos et al. (2022) on age and sex contradict their own tables, and the component areas do not fully add up to the totals.

The posterior connection of the attachment affects failure behavior: In an ex vivo study of human knees, cutting the connection between the posterior fan-like extension and the midsubstance fibers was associated with a failure load about 55% lower, and all failures moved to the femoral attachment (Kanto et al., 2025). Without a sham-cut control, the difference cannot be attributed entirely to the posterior fibers. The claim that only the direct insertion matters mechanically has no direct support.

2. Bony Landmarks on the Femoral Side

The femoral bony landmarks matter because the anterior limit of the direct insertion coincides with the lateral intercondylar ridge. This has histological support: Iwahashi et al. (2010) and Sasaki et al. (2012) both observed on sections that the direct insertion of dense fibers begins behind the ridge, and in the specimens of Mittendorfer et al. (2026) the ridge likewise lay at the anterior edge of the direct insertion. How common the ridge is, whether it runs the full length of the attachment, and whether it can be reliably identified on imaging or at arthroscopy are separate questions.

Lateral intercondylar ridge and lateral bifurcate ridge

The lateral intercondylar ridge is also known as the resident's ridge, a name given by Clancy; because the name had been used inconsistently in the literature, Ferretti et al. (2007) adopted the term lateral intercondylar ridge (p. 1222). This study had three samples: histological sections of 7 fetal knees at 19–22 weeks, arthroscopic observation in 60 patients undergoing ACL reconstruction, and 16 cadaveric knees from 8 donors (mean age 75 years). The lateral intercondylar ridge was found in all 60 patients and all 16 cadaveric knees. The lateral bifurcate ridge was found in 6/7 fetuses, 49/60 patients and 13/16 cadaveric knees; it measured 3.5±0.8 mm in length and separated only the anterior part of the AM and PL attachments. A change in slope between the AM and PL attachments was visible in all 16 cadaveric knees, at an angle of 27.7°±8.9° (pp. 1220–1222, Table 1).

"Constant presence" applies to qualified samples: All 60 patients had ACL ruptures, were operated on by one senior author and were examined after the stump had been cleared with a radiofrequency device (p. 1219); the observation that no ACL fibers attach anterior to the ridge also comes from these injured knees with debrided stumps. In one cadaveric knee the lateral intercondylar ridge extended only to the middle of the attachment (p. 1222), so the ridge ran the full length in 15/16. These figures show that the ridge was commonly visible in this study's samples; they are not a prevalence in the normal population.
Some statements go beyond the results: The Results open by stating that both ridges were found in all parts of the study (pp. 1219–1220), yet the fetal group reported only the lateral bifurcate ridge, and the lateral bifurcate ridge was absent in 1/7, 11/60 and 3/16 in the three groups. The Abstract and Table 1 also differ slightly (for example, a slope angle of 27.6°±8.8° against 27.7°±8.9°).

Iwahashi et al. (2010) measured the lateral intercondylar ridge on CT as 14.5±0.9 mm long and 0.8±0.1 mm high (p. S19, Table 3). The ridge is under 1 mm high and the CT slice thickness was 1 mm, so the precision of the height measurement is limited. Histological sections in the same study showed loose peripheral fibers crossing the ridge to attach anterior to it (p. S15). The statement that no ACL fibers at all lie anterior to the ridge is therefore too absolute; the accurate statement is that the direct insertion of dense fibers is bounded anteriorly by the ridge.

Bhattacharyya et al. (2018) examined the reliability of the ridge in 23 dry femora and 7 cadaveric femora with soft tissue preserved, and concluded that the lateral intercondylar ridge is "a consistent anatomical structure" (Abstract). In the dry femora, two investigators judged by eye that all 23 had "a definite bony ridge", without stating their criteria. In all 7 cadaveric specimens the ridge lay anterior to (above) the attachment, and the mean distance from the ridge to the midpoint of the inferior cartilage margin was 10.1 mm (9–11 mm) (Results 3.1, 3.2). For 23/23, the lower limit of the 95% confidence interval by the Wilson method is about 86%, which is not sufficient to show that the ridge is constantly present in the population.

The tunnel rule conflicts with the authors' own data: The authors propose that when the ridge is not visible, the upper edge of the tunnel should be no higher than 10 mm above the midpoint of the cartilage margin (Discussion). This limit is the mean of 7 specimens, and the measured range starts at 9 mm; in some specimens the rule would place the upper edge of the tunnel above the ridge, contradicting the authors' requirement that the upper edge of the tunnel always lie below the ridge.
Methods incompletely reported: Specimen age, sex, donor pairing and preservation were not reported. In the Methods, the source of the specimens reads "Department of Human Anatomy, XXXX", an anonymization placeholder that was never removed, and the source institution appears only in the Acknowledgments. The end point of the distance measurement is the ridge (LIR) in the Results and "the midpoint of the LIR" in the Discussion, and no SD is given.

Mittendorfer et al. (2026) provide another set of figures. CT three-dimensional reconstruction of 20 fresh-frozen femora from 10 donors (7 women and 3 men, 63–92 years) detected the lateral intercondylar ridge in 17/20 (85.0%, 95% CI 64.0%–94.8%) and the lateral bifurcate ridge in 5/20 (25.0%); the corresponding PCL ridges on the medial wall were detected in 16/20 and 2/20 (p. 6, Table 1). The two observers were trained together beforehand and blinded to the data, with κ values from 0.773 (lateral intercondylar ridge) to 0.894 (lateral bifurcate ridge). Histologically, the lateral intercondylar ridge was seen in 16 of the 17 evaluable femora, and local bony prominences were visible at both the anterior and posterior edges of the direct insertion (pp. 6–9, Figs. 3 and 4, Table 2).

No case-by-case correspondence between CT and histology: The authors state that no specimen-level CT–histology registration or distance analysis was performed (p. 5). The CT detection rate of 85.0% and the histological rate of 94.1% come from partly different specimens, so the accuracy of CT in identifying the anterior and posterior edges of the direct insertion cannot be derived.
Bilateral specimens analyzed as independent: The 20 femora came from 10 donors; the authors acknowledge that no clustering model was used and describe the subgroup P values as exploratory (p. 5).

Extension of the ridge onto the notch roof

Gulan et al. (2025) measured 45 dry femora by optical three-dimensional scanning. The inclusion criteria required a clearly visible lateral intercondylar ridge (p. 2), so the study cannot estimate the prevalence of the ridge. The authors defined an "anterior cruciate ligament femoral insertion area" (ACL-FIA) bounded by the ridge, the cartilage margin of the lateral condyle and a connecting line of their own choosing, and then used a self-defined three-point plane to separate the notch roof from the lateral wall (pp. 2–3). The lateral intercondylar ridge measured 16.18±1.44 mm in total length, of which 5.02 mm (30.86%) lay on the roof; the inferred attachment area measured 134.19±34.17 mm², of which 31.55% lay on the roof (pp. 5–6, Tables 1 and 2).

The attachment area was inferred from bony landmarks, not from actual fiber attachment: The specimens were dry bones, without soft tissue or histological verification. The statement in the Abstract that about 30% of the ACL femoral attachment was confirmed to extend onto the roof actually means that, with the plane defined by the authors, about one third of the bone surface behind the ridge falls on the roof side. A study cited in the Discussion holds that the ridge coincides with the anterior margin of the ACL mainly in its proximal part, which is inconsistent with the authors' assumption that the entire ridge forms the anterior border of the attachment (p. 7).
The statistics cannot be reproduced: The correlation between the roof proportion of the ridge and the posterior notch width is given as r=0.51, P=0.0041 (p. 6); for 45 cases, r=0.51 corresponds to P≈0.0004, and 0.0041 is closer to the result for about 30 cases. The SD of the roof proportion of the inferred attachment area is 7.31% in the text and 7.51% in Table 2. The hypotheses also included a relationship between the attachment-area proportion and posterior notch width, for which no result is reported.

This study shows that the proximal segment of the lateral intercondylar ridge, together with the bone surface behind it, extends onto the notch roof, so that describing the femoral attachment arthroscopically as a lateral wall structure underestimates its extent toward the roof. How many fibers actually attach on the roof is a question for studies with the soft tissue preserved.

Bony prominence at the posterior edge of the direct insertion

Sasaki et al. (2012) observed on sections a bony prominence at the posterior edge of the direct insertion, named it the lateral intercondylar posterior ridge and proposed it as an arthroscopic landmark (pp. 1142–1143). The paper does not report in how many specimens it was present, its height or length, or whether it can be identified arthroscopically or on CT. The histological figures of Mittendorfer et al. (2026) also show a bony prominence at the posterior edge of the direct insertion, without naming or counting it. Sections in both studies show that a bony prominence can be present at the posterior edge of the direct insertion; whether it can be identified at arthroscopy or on imaging, and whether it helps localization, has not yet been tested.

Key points

The lateral intercondylar ridge marks the anterior limit of the direct insertion: Histological studies show that the direct insertion of dense fibers begins behind the ridge, and loose fibers can cross it.

Detection depends on the method of observation: Ferretti et al. (2007) identified the lateral intercondylar ridge arthroscopically in injured knees and in 16 cadaveric knees, in one of which it did not run the full length; CT three-dimensional reconstruction of fresh-frozen femora detected the ridge in 17/20, and failure to detect it on CT does not mean that it is anatomically absent (Mittendorfer et al., 2026). The lateral bifurcate ridge was identified in 5/20 to 13/16 across studies.

The ridge extends onto the notch roof: In a dry-bone study, about 31% of the length of the lateral intercondylar ridge lay on the roof side as defined by the authors (Gulan et al., 2025); this is a bony inference, not the proportion of actual fiber attachment.

A bony prominence can be present at the posterior edge of the direct insertion: The lateral intercondylar posterior ridge has histological support; its prevalence, arthroscopic visibility and value for localization have not been reported.

3. Methods for Expressing Femoral Insertion Position

Femoral insertion position has been expressed in several ways: the clock-face method, millimeter distances along the notch roof or the femoral shaft, percentages referenced to the posterior condylar circle, and the quadrant method. The quadrant method is the most widely used in imaging studies and reviews, and the one most often treated as a common standard.

Definition and classic values of the quadrant method

The quadrant method was introduced by Bernard and Hertel for standard lateral radiographs. The definition below follows the description and application of the method by Colombet et al. (2006) and Parkar et al. (2017). On a standard lateral image, a rectangle enclosing the lateral femoral condyle is drawn with the notch roof line (Blumensaat line) as its upper side. The deep–shallow direction is measured along the roof line from the posterior (deep) end, with the full length of the lateral condyle along the roof line as the denominator; the high–low direction is perpendicular to the roof line, measured from the roof line, with condylar height as the denominator (Parkar et al., 2017, pp. 2–3, Fig. 1; Colombet et al., 2006, p. 987, Fig. 6). Following the arthroscopic convention at 90° of flexion, the deep–shallow and high–low directions correspond respectively to the proximal–distal (posterior–anterior) and anterior–posterior (superior–inferior) directions in the anatomical position, which differ from the anatomical orientation of the condyle; the origin used by each study should be checked before its values are read.

The values most often treated as the classic femoral insertion position come from Bernard et al. (1997). Table 2 of the review by Parkar et al. lists them as 10 knees, 24.8% in the deep–shallow direction and 28.5% in the high–low direction. These figures are cited here from Parkar et al.; their specimens, traced structure and definition of center cannot be checked from that source.

Colombet et al. (2006) marked the centers of both bundles with beads in 7 unpaired fresh-frozen knees (mean age 75 years). On lateral radiographs, the AM center lay at 26.4%±2.6% in the deep–shallow direction and 25.3%±4.2% in the high–low direction, and the PL center at 32.3%±3.9% and 47.6%±6.5% (p. 988); Fig. 6 simplifies these to grid positions of 1/4, 1/4 and 1/3, 1/2 (p. 989). The "center" here is the parallel projection of the central fibers of each bundle onto the attachment surface, which the authors state explicitly is not the geometric center of the attachment area (pp. 986, 989).

The second direction is misnamed: The Results describe the direction of 25.3% and 47.6% as also "parallel to Blumensaat's line" (p. 988), but Fig. 6 shows it to be the high–low direction perpendicular to the roof line; the Methods do not give the denominator for this direction either.
The origin of the deep–shallow direction is inconsistent: The Results measure 26.4% from the deep point I′ (p. 988), whereas the Discussion describes it as measured from the most anterior point of the femoral condyle (point J′) (p. 991); Fig. 6 and the anatomy support measurement from the deep end.
"Intraobserver error" is in fact interobserver variation: Six surgeons each measured once, yet Tables 1 and 2 label the dispersion between them as "intraobserver error" (pp. 986–989), while the Methods, the Abstract and the radiographic results call it interobserver variation. The study had no repeated measurements by the same observer.
Commercial ties: The authors disclose a consulting relationship with Smith & Nephew Endoscopy, and the second author's fellowship was funded by the company (p. 984); the paper proposes that the data could be used to design instruments for double-bundle reconstruction (p. 992). Readers should know of this interest. It is not evidence of data falsification, and the anatomical measurements and the endorsement of double-bundle technique should be judged separately.

Grid placement changes the coordinates

Yahagi et al. (2018) tested the effect of the grid itself in 59 unpaired embalmed knees (median age 83 years). The authors first outlined the whole femoral insertion and computed its geometric center, then placed the grid in four ways on the same photograph: the grid baseline along Blumensaat's line either did or did not take account of a hill on the posterior part of the roof, and the lower and side borders were tangent either to the medial wall of the lateral condyle or to the articular cartilage surface (p. 457, Fig. 2). Of the 59 knees, the roof line was straight in 19, had a small posterior hill in 13 and a large hill in 27. In the 19 straight-roof knees, the group mean center lay at 33.7% (deep–shallow) and 47.6% (high–low) with the borders on bone, and at 39.4% and 39.4% with the borders on cartilage; in the 27 knees with a large hill, the group mean in the high–low direction varied from 50.4% to 34.4% across the four grids (p. 458). With the insertion outlines unchanged, the group means shifted by about 6 to 16 percentage points, a difference due entirely to the reference frame; individual knees did not necessarily shift by the same amount.

Significance judged in breach of the authors' own rule: The Methods define significance for pairwise comparisons as P<0.01 (p. 457), yet the Results call comparisons with P=0.027, 0.020 and 0.037 significant, and the figure legends switch to P<0.05 (p. 458). By the prespecified threshold, none of the pairwise comparisons in the small-hill group reaches statistical significance, and neither does the high–low comparison between grids 2 and 4 in the large-hill group. The coordinate differences between grids remain as descriptive results, and comparisons with P<0.01 can be used.

Iriuchishima et al. (2020), from the same group, reported that the insertion center in 40 knees with a roof hill was shallower than in 19 straight-roof knees (37.2% vs 33.7%, P=0.02), and on this basis recommended choosing the femoral tunnel according to roof morphology (Abstract).

Very probably the same specimens as Yahagi et al.: The number of knees, sex distribution, age and ethics approval number of these 59 knees are identical to those of Yahagi et al. The coordinates of the straight group (33.7%±4.7%, 47.6%±8.8%) match grid 1 of Yahagi et al., and the values for the hill group equal the case-weighted pooled values of grid 1 for the small-hill and large-hill groups of Yahagi et al. (p. 2455). The paper cites Yahagi et al. only for the grid method and does not state how the specimens of the two studies are related. Reviews and meta-analyses that treat them as independent samples may count the same knees twice.
The number of cases in the correlation analysis needs explanation: A hill apex exists only in the 40 knees with a hill, yet the paper states neither the actual number of cases in the correlation analysis nor how knees without a hill were handled. With r=0.28, P is about 0.08 for 40 cases and about 0.032 for 59 cases, the latter matching the reported value (p. 2455, Table 1). If the analysis included knees without a hill apex, the correlation is difficult to interpret.

Inclusion or exclusion of the fan-like extension

Iriuchishima et al. (2016) compared bundle centers in 14 knees with and without the fan-like extension (see Section 1): the AM center moved from 29% to 35% in the deep–shallow direction, the PL center from 37% to 43%, and the PL center in the high–low direction from 73% to 69% (p. 256, Table 2). With the fan-like extension excluded, the centers moved shallower and higher, consistent with the posterior and distal position of the fan-like extension. The "modified quadrant method" used in this study is not defined in the paper; a PL center of about 70% in the high–low direction differs greatly from the 47.6% of Colombet et al. and the roughly 49% pooled by Parkar et al., and its coordinates cannot be compared directly with those of radiographic quadrant studies.

Edwards et al. (2008) applied several references to the same 22 knees and obtained very different values. With the posterior condylar circle as reference, the AM center lay at 68% in the deep–shallow direction and 45% in the high–low direction; in an arthroscopic 16-square grid bordered by the cartilage margin, the AM center lay at 21% (shallow) and 24% (low) from the high–deep corner (p. 33). This grid is an arthroscopic modification of Bernard's radiographic method with borders on the cartilage margin, and its values are not interchangeable with those of the radiographic quadrant method.

The Abstract does not say that the origin was reversed: In the Abstract, the high–low coordinates of AM and PL are 55% and 62%, exactly 100 minus the 45% and 38% in the Results; the origin changed from the low end to the high end, whereas the deep–shallow coordinates in the same sentence were unchanged (pp. 29, 33). Citing the Abstract directly gives the false impression that PL lies higher than AM.
Clock-face position changes with the viewing direction: For the same attachment, viewing along the femoral shaft and viewing along the notch roof change the clock positions of AM and PL by about half an hour each (p. 33). The authors therefore stress control of flexion angle and viewing direction, a point with direct support from the data.

The "lateral wall clock" of Mochizuki et al. (2006) is yet another coordinate system: 1:40 and 3:10 are height fractions multiplied by 6, not angular clock positions in the usual sense (p. 359, Fig. 7), and they cannot be compared with the clock-face positions of other studies.

Quadrant coordinates on three-dimensional MRI

Nam et al. (2024) applied the quadrant method to three-dimensional MRI of the normal knees of 45 patients seen for contralateral knee injury (18–45 years; 38 men and 7 women). One author identified the attachment center on MRI, and the coordinates were read on simulated lateral images generated from the MRI data. The femoral attachment center lay at 31.6%±2.4% in the deep–shallow direction and 31.3%±3.5% in the high–low direction, with individual ranges of 26%–39% and 24%–40% (Results; Discussion). The value of the study is that it shows the spread between individuals with one method and one reader: 13 percentage points in the deep–shallow direction alone. The small SDs reflect a single-reader method without repeated measurements or a second observer; the attachment center was taken as the midpoint in two orthogonal planes rather than the geometric center of a three-dimensional outline; and there was no anatomical verification in the same knees. In Table 1, the male and female means in the high–low direction (31.7% and 31.0%), weighted by 38 and 7 cases, cannot yield the overall 31.3%.

Pooled values from reviews

Piefer et al. (2012) pooled femoral insertion studies published after 2000. They calculated unweighted means of the radiographic coordinates from 8 studies (whole ligament about 28.5% and 35.2%), projected them onto an idealized drawing of the femoral condyle and read along a line parallel to the femoral long axis: the center of the whole ligament lay at 43% of the proximal-to-distal length of the lateral notch wall, described as a percentage between cartilage margins that can be measured arthroscopically (pp. 874, 876, Fig. 3). A second rule placed the center anterior to the posterior articular cartilage margin by the tunnel radius plus 2.5 mm (p. 877).

The 43% comes from a schematic drawing: No knee was measured by this method, and there was no arthroscopic validation. The paper is inconsistent about the end points of the 43%: the Abstract and Results refer to the arthroscopically visible osteochondral junction, whereas the Discussion acknowledges that the measurement was taken along the bony lateral wall and did not account for articular cartilage, which is not visible on radiographs (p. 876).
The source of the 2.5 mm: In Table 2, the posterior margin distances of four studies are 4 mm, 3.5 mm, 2.5 mm and the statement by Iwahashi et al. that the attachment extends to the posterior cartilage margin. The mean is 2.5 mm only if that statement is counted as 0 mm (p. 877). Iwahashi et al. measured the direct insertion; the other studies measured the macroscopic whole attachment.
Colombet's erroneous figures were transcribed: Table 3 and Fig. 4 use the erroneous dimensions from the text of Colombet et al., 13.9±9.5 mm × 9.3±7.1 mm (pp. 878–879).

Spalding et al. (2014) pointed out in a letter that on lateral projection the outline of the femoral condyle extends beyond the articular cartilage margin, particularly distally, so that the projected full length differs from the distance between the cartilage margins of the lateral wall seen at arthroscopy; the same center can give 43% or 50% depending on the reference (pp. 538–539, Fig. 2). The main criticism is valid, and the explanation in the Discussion of Piefer et al. supports it. The letter also overstates: it refers to "a radiologically validated figure of 43%", although Piefer et al. performed no radiographic measurement. The 50% in the letter is taken from another three-dimensional CT study and illustrated with one CT image of a drilled tunnel; it too serves only to show the difference produced by the choice of reference and cannot be taken as a surgical target for all individuals. In their reply, the authors accepted that the letter raised reasonable points, while noting that CT shows only bone, that the position of the cartilage margin can only be inferred, and that they doubted whether the end points marked in the letter's figure corresponded to the osteochondral junction seen at arthroscopy (p. 539). The letter's authors disclosed support from Zimmer UK, Arthrex and Smith & Nephew, respectively; the original review declared no relevant conflicts of interest.

The systematic review by Parkar et al. (2017) included 16 studies. On the femoral side, 13 studies with 218 knees gave weighted means of 29% (deep–shallow) and 35% (high–low) and weighted medians of 26% and 34%; on the tibial side, 10 studies with 300 knees gave a weighted mean of 42% in the anteroposterior direction (pp. 2, 6–7, Tables 2 and 3). When a primary study reported only the two bundle centers, the authors took their arithmetic mean as the center of the whole ligament (p. 5). This calculation was not weighted by bundle area and is not equivalent to a center computed from a tracing of the whole attachment. The Bernard et al. (1997) values included in the review can likewise be cited only through it.

The distribution of study means is presented as an individual normal range: The authors calculated weighted 5th and 95th percentiles from study means, giving 24%–37% in the femoral deep–shallow direction, 28%–43% in the high–low direction and 39%–46% in the tibial anteroposterior direction, and recommended using them to judge whether postoperative tunnel placement is "in or out of the anatomic range" (Abstract; pp. 8–9, Fig. 3). These percentiles describe how the means of 13 or 10 studies are distributed, which is a different matter from the distribution of individual knees. Within single studies, the SDs in the high–low direction in the groups of Yahagi et al. are about 6%–9%, and the individual range in the deep–shallow direction in the 45 knees of Nam et al. is 26%–39%; variation between individuals adds to the variation between studies. There is no basis for treating 24%–37% as the range within which 90% of normal knees fall, or for judging a given tunnel acceptable on that basis. The authors themselves acknowledge in the Discussion that the data are means from small studies and that the separation of the two bundles may be biased (pp. 10, 12). The upper limit in the high–low direction also depends on a value that needs explanation. In Table 2, Takahashi et al. (2006) contributed only two bundle centers (26.9% and 53.2% in the high–low direction); by the authors' stated two-bundle averaging rule, the whole-ligament coordinate should be 40.05%, yet the table lists 42.6%, exactly the 95th percentile in the high–low direction (pp. 5–6). The authors do not state that this entry came from another source, so the 43% upper limit rests on a value that does not match their own calculation rule.
Data entry problems: The tibial table lists Stäubli and Rauschning (1994) as 10 knees and 44.0%, whereas the center in that study's 10 cadaveric knees lay at 41.2%; 44% is the result from 35 subjects examined by MR arthrography and the tunnel position recommended by the authors (Stäubli and Rauschning, pp. 138, 141–143). The sample size for the two tibial bundles is 118 knees in the title of Table 3 and 106 knees in the text (pp. 7–8). These problems have little effect on the weighted means, but they show that samples and measured structures were not matched study by study in the summary tables.

These pooled values show where the centers reported by different studies cluster, and they draw attention to differences in reference. They are not normal values for individual anatomy, and they are not surgical targets validated against clinical outcome.

Key points

For the same insertions, coordinates change with the reference frame: Changing only the baseline and border placement of the quadrant grid shifted the group mean centers by about 6 to 16 percentage points (Yahagi et al., 2018); including or excluding the fan-like extension moved the bundle centers by about 4 to 6 percentage points (Iriuchishima et al., 2016). Percentages from different grids or with different origins cannot be compared directly.

Classic values are mostly second-hand or read off a drawing: The 24.8% and 28.5% of Bernard et al. (1997) are cited from Parkar et al. (2017); the 43% of Piefer et al. (2012) is a coordinate read after projecting radiographic means onto a schematic drawing and cannot be converted directly into an arthroscopic percentage between cartilage margins.

Review percentiles are not individual normal ranges: The 24%–37%, 28%–43% and 39%–46% that Parkar et al. (2017) calculated from study means describe variation between studies and cannot be used to judge whether an individual tunnel is acceptable.

Reporting problems in several studies: Colombet et al. (2006) misname a direction and are inconsistent about its origin; Yahagi et al. (2018) call comparisons significant that did not meet their own threshold; Iriuchishima et al. (2020) very probably used the same specimens as Yahagi et al. without saying so; the Abstract of Edwards et al. (2008) reverses the origin of the high–low direction.

4. Morphology of the Tibial Attachment

Classic descriptions and the whole attachment

Girgis et al. (1975) recorded that the ACL attaches to a broad depressed area anterior and lateral to the medial intercondylar eminence, with some fibers attaching to the base of the eminence and a distinct slip extending to the anterior horn of the lateral meniscus (p. 218). In the 19 specimens in Table 2, the mean attachment length was 29.3 mm and the anterior margin lay 15.2 mm from the anterior articular margin of the tibia; in 9 specimens the attachment wrapped posteriorly around the medial intercondylar eminence and blended with the posterior horn of the lateral meniscus, and in 2 specimens additional fibers attached to the anterior horn of the medial meniscus (pp. 218–220, Table 2). The text reports the connection to the anterior horn of the lateral meniscus in all 44 knees, but lists attachment measurements for only 19.

Odensten and Gillquist (1985) found the tibial attachment in 20 young adult knees to be an oval elongated anteroposteriorly, 17±3 mm long and 11±2 mm wide, with its center 23±4 mm from the anterior medial meniscocapsular junction (p. 259, Fig. 4). The text also gives a distance of 7±1 mm from the center to the plane of the anterior margin of the medial tibial articular surface, but Fig. 4 draws these 7 mm as a mediolateral distance; this often-cited number should be read together with the inconsistency.

Edwards et al. (2007) measured 55 frozen specimens (32 isolated tibial plateaus and 23 intact knees, mostly 65–80 years): the attachment was 18±2 mm (11–23 mm) long anteroposteriorly and 9±2 mm (7–14 mm) wide mediolaterally; in 47 of the 55, the AM and PL bundles were arranged mediolaterally, and in 8 the arrangement was closer to anteroposterior (p. 1417). Colombet et al. (2006) measured 17.6±2.1 mm anteroposteriorly and 12.7±2.8 mm mediolaterally in 7 knees (p. 987, Table 1). All of these describe the macroscopic whole attachment, including all fibers visible after removal of the synovium.

C-shape and duck-foot

Siebold et al. (2015) sectioned the ACL in extension in 20 knees (6 fresh-frozen and 14 paraffin-treated; median age 78 years), avoided flexion during cutting and sectioning, and exposed the tibial insertion layer by layer (p. 3137). The part where the midsubstance fibers continued into the tibia was C-shaped, running from beside the medial intercondylar eminence around to the front of the anterior root of the lateral meniscus; it was 12.6 mm long and 3.3 mm thick, with an area of 31.4±7.2 mm². The center of the C was the bony attachment of the lateral meniscus anterior root, covered by fat; the authors saw no fibers attaching centrally or posterolaterally and called the posterior fibers posteromedial fibers (pp. 3138–3139, Table 2). They called the C-shaped part the "direct" insertion and the fibers fanning anteriorly to the anterior edge of the tibial plateau the "indirect" insertion (79.6±12.7 mm²); together these formed a duck-foot bony footprint (110.9±14.7 mm², p. 3139). In most specimens, the outer fibers of the anterior and posterior horns of the lateral meniscus blended with the C-shaped insertion like a belt (p. 3138, Fig. 2).

"Direct" and "indirect" are macroscopic labels only: The authors state that a histological study was in progress (p. 3141). The same group's subsequent histological study (Oka et al., 2016) called this labeling into question; see below.
Values misplaced in Table 3: The midsubstance width of 11.9 mm and thickness of 3.5 mm listed for the present study in Table 3 are in fact the values 5 mm from the tibial insertion in Table 2, and 37.0 mm² has no source anywhere in the paper (p. 3140); the length column of Table 3 actually contains thickness. The midsubstance values should be taken from Table 2 (9.9 mm, 3.9 mm, 38.7 mm²). The formal erratum (PMID 25149644) corrected only the order of the authors' names.
The conditions of observation limit the conclusions: All dissections were performed by one person, and all observations were "reconfirmed by the coauthors, who watched and assisted the dissections"; there was no independent or blinded reading, and the preparation of the 14 "paraffined" knees is not defined. The fat pad between the lateral meniscus anterior root and the ACL was removed during dissection (p. 3137). Quiles et al. (2018) pointed out that the area where classic descriptions place posterolateral fibers may have been removed, or judged to be fat, at this step (p. 10). The conclusion that there are no posterolateral fibers is affected by this step, which the paper does not discuss.

Histological examination by the same group

Oka et al. (2016) made serial sagittal sections of 6 undissected fresh-frozen knees (56–72 years, 3 pairs), divided from medial to lateral into four blocks, T1–T4 (p. 748). The insertion showed the typical four-layer structure of a direct insertion. Its anterior margin was bounded by a bony anterior ridge, which the authors considered to be the classic Parsons' knob; its medial margin followed the ridge on the anterior part of the medial intercondylar eminence and met the articular cartilage of the medial plateau. The anteroposterior length of the insertion decreased from medial to lateral, from 10.8±1.1 mm in T1 to 6.2±1.1 mm in T4. Behind the lateral half of the insertion, between it and the posterior eminence, lay synovium, fat and vessels, without attaching fibers (pp. 749–751, Table 1). On this basis the authors wrote that, histologically, the C-shape might be only the cross section of the midsubstance ligament above the actual direct insertion, and the duck-foot footprint might be the histological direct insertion (p. 751).

The direct observation from the sections is that the bony surface of the insertion showed a layered fibrocartilaginous transition in every block, that the anterior margin arose from the anterior ridge, and that the transition layer was thickest medially and thinned laterally. The authors' interpretation of the C-shape and duck-foot uses "might" twice and is a speculation based on the sections. Even so, the result does not fit the macroscopic labeling of the anterior fanning fibers as an "indirect" insertion by Siebold et al.; the same group's histology did not support the division into a C-shaped direct insertion and an indirect remainder. Whether the C-shape corresponds to the insertion itself or to the ligament cross section above it cannot be determined, because the study had no axial sections.

Figures in the text do not match the table: The Discussion states that the longest insertion was 10.2 mm in the most medial section and contrasts this with the 14–30 mm of macroscopic studies (p. 751); in Table 1, the mean for T1 is 10.8 mm with a maximum of 12.1 mm, and the maximum for T2 is 13.7 mm. The ligament thickness at T4 is 2.7 mm in the text and 2.9 mm in Table 1.
Statistical handling insufficiently described: T1–T4 are four planes of the same knee, and the 6 knees came from 3 pairs. The Methods state only "analysis of variance (ANOVA), followed by post hoc testing with a Tukey procedure" (p. 749), without saying whether the multiple planes of each knee were treated as repeated measures or how the two sides of the same donor were handled. The difference between T1 and T4 is large and its direction is unaffected, but the reported P values cannot be verified.
Sagittal sections only: The C-shape and duck-foot are axial shapes, and the study had no axial sections; the narrow region near the lateral intercondylar eminence may not have been sampled, so the conclusion that there are no posterolateral fibers rests on a limited set of planes. No funding or conflict-of-interest statement was found in the paper.

L-shaped dense fiber insertion

Kusano et al. (2017) used 12 embalmed knees (78–95 years, 10 women): 6 were sectioned coronally, parallel to the course of the lateral meniscus anterior horn, and matched with CT, and 6 were sectioned sagittally (Sections 2.1–2.5). Dense ACL fibers inserted directly on the top and lateral slope of the medial intercondylar ridge (medial intercondylar eminence). Anterior to the anterior horn, the insertion was 10.2±0.6 mm wide; at the level of the anterior horn, the ACL and the anterior horn divided the slope equally, with dense fiber insertion lengths of about 5.4–5.8 mm each; posterior to the anterior horn, all 6 knees had only loose connective tissue between the ACL fibers and the bone (Results 3.2, Table 2). Laterally, the ACL fibers covered the anterior horn without blending with it (Section 4). Where the two met on the slope, the inclination of the bone surface changed; on CT three-dimensional reconstruction this appeared as an anteroposterior ridge, which the authors called the central intercondylar ridge (Sections 3.3–3.4). The resulting dense fiber insertion was L-shaped: wide anteriorly and narrowing posteriorly along the medial side of the anterior horn (Fig. 6c).

The L-shape is an interpolation: Each knee had only 4 coronal levels, and the L-shape was assembled from 3 of them together with the sagittal sections of another group of 6 knees; Fig. 6c is a schematic of a single specimen. No areas, individual outlines or observer agreement data were reported.
Some figures are inconsistent: The insertion length at sagittal level 5 is 10.0 mm in the text and 10.2 mm in Table 3. In Table 4, the distance from the ridge apex to the point of inclination change at level 2 is 5.7±0.3 mm with a range of 4.8–6.2 mm; if "±" denotes the SD of 6 cases, this combination is arithmetically impossible.
The recommendation for double-bundle or triple-bundle reconstruction comes from morphological reasoning and is linked to the senior author's own technique (Section 4); this study did not test it.

Attachment to articular cartilage and the lateral meniscus

Muro et al. (2022) studied 11 knees from 6 donors (52–74 years): 10 knees for macroscopic dissection and area measurement, and 1 knee for serial sections at 0.2 mm intervals and three-dimensional reconstruction (Methods). After the ACL fibers were removed bundle by bundle and the attachment margins marked, the meniscus and articular cartilage were removed; the medial marking line disappeared with the cartilage, showing that the medial part attached to articular cartilage. The whole attachment measured 151.9±37.2 mm², of which 102.6±27.5 mm² (67.5%) was on bone, 40.9±13.6 mm² (26.9%) on articular cartilage, 6.5±4.6 mm² (4.3%) on the lateral meniscus and 1.9±5.2 mm² (1.3%) on the transverse ligament (Table 1). The cartilaginous attachment also passed through a fibrocartilaginous layer; serial sections showed a few lateral ACL fibers gradually diverging and finally merging with the superficial lateral fibers of the anterior horn of the lateral meniscus (Results).

The main significance of this study lies in the structure measured: the tibial attachment area of about 110 mm² reported earlier corresponds roughly to the bony part, and including the attachment on cartilage brings it to about 150 mm². The authors state that the roughly 27% cartilaginous attachment refers to area and does not represent the proportion of fibers or tension (Discussion).

Functional conclusions were not tested: Statements that the ACL–lateral meniscus complex forms "a highly stable hoop structure" and that the cartilage attachment is "an important attachment for ACL" have no support from mechanical testing; the histological findings come from a single knee and cannot give a frequency.
Observations differ from those of Kusano et al.: In serial sections of 1 knee, Muro et al. saw a few lateral ACL fibers gradually diverging and finally merging with the superficial fibers of the lateral meniscus anterior horn, which shows that such interlacing can exist; in coronal sections of 6 knees, Kusano et al. saw ACL fibers covering the anterior horn without merging with it. The two studies differed in section orientation and sampling, and the negative result of the latter cannot refute the observation of the former; the frequency and mechanical significance of the interlacing are unknown. The blending of outer fibers in most specimens described by Siebold et al. was a macroscopic observation.
The breakdown in the Abstract is incomplete: The Abstract and text list only bone, cartilage and meniscus, which sum to 150.0 mm²; the transverse ligament appears only in Table 1, without a definition or the number of knees in which it was found.

Muro et al. (2024) analyzed the attachment outlines of 25 knees from 17 donors (8 men and 9 women). The Methods give "12 men and 13 women", which are numbers of knees rather than donors, and the mean age of 77.8 years was also calculated per knee (p. 683, Table 1). The attachment comprised bone and cartilage, excluding the part interlacing with the meniscus. Two observers classified the shapes as oval in 6/9, C-shaped in 11/7 and triangular in 8/9 (counted from Table 1, for the two observers respectively); all 4 disagreements involved the C-shape. Principal component analysis showed no clear separation between types, and shape varied continuously (pp. 685–686). Of the 8 donors with both knees included, 4 had left and right knees classified as different shapes.

The claim that anterior horn position determines attachment shape was not tested statistically: The Abstract states that the position of the lateral meniscus anterior horn "influences" ACL attachment shape, and the authors recommend predicting attachment shape intraoperatively from the position of the anterior horn (Abstract; p. 689). The paper contains no statistical analysis of the relationship between shape and anterior horn position, only a trend on a scatter plot; a rough test of the Table 1 data suggests at most a moderate association, mainly between oval shape and a more anterior horn. The accuracy of prediction was not assessed.
The sample may overlap with the 2022 study: The two studies come from the same institution with the same ethics approval number (M2018-243). Table 1 of the 2024 study contains a group of donors (3 men and 3 women, 52–74 years, mean 65.3 years, 10 knees in total) whose characteristics match those of the 2022 donors exactly; the paper does not mention any overlap, and the two studies should not be treated as independent samples.

Three-dimensional reconstruction of the bone surface

Quiles et al. (2018) reconstructed 14 embalmed tibial plateaus (8 donors, 16–76 years) by stereophotogrammetry, after cauterizing the attachment outline onto the bone surface along the soft tissue margin. All attachments were duck-foot shaped, wide anteriorly and narrow posteriorly, straight medially and concave laterally, with a narrow fiber-free strip posteriorly; the three-dimensional surface area was 145.2 mm² (81–207.1 mm²), with AM and PL portions of 84.3 and 60.9 mm² (pp. 6–7). The study supports the classic AM–PL arrangement and contradicts the conclusion of Siebold et al. that there are no posterolateral fibers.

Inconsistent shape counts: The Results give 8 triangular and 5 oval (13 in total); the Discussion gives 9 triangular (64%) with the rest oval (pp. 6, 10).
Circular argument for the bundle boundary: The two bundles were not marked separately on the bone surface. The authors transferred inter-bundle distances from drawings onto the model, found that they roughly matched a groove on the bone surface, and then used the groove as evidence of the boundary between the bundles (pp. 7, 11).
Citation and comparison errors: Of the 3 references supporting a sentence in the Discussion on the limitations of earlier landmarks, two are studies of the posterior cruciate ligament and one concerns software for classifying mitochondrial DNA haplogroups (p. 12). The comparison with Edwards et al. confuses the over-the-back ridge with the anterior margin of the PCL; the distance from the ACL center to the posterior eminence area measured by Quiles et al. themselves, 17.2 mm, is close to the 15 mm of Edwards et al. (p. 11).
A single examiner and no reliability data; 6 pairs of left and right tibiae treated as independent samples; specimens with Outerbridge grade IV osteoarthritis included.

Age and morphology

Morales-Avalos et al. (2021) compared the macroscopic shape of the tibial attachment in 64 embalmed Mexican knees divided at 50 years of age: 23/27 in the group under 50 years were oval and 31/37 in the group over 50 years were C-shaped, with about 11%–15% triangular in each group (pp. 925–926, Table 1). This is a cross-sectional comparison between different individuals. It can show differences in shape between age groups, but the change within individuals implied by the title, that the footprint "changes with ageing from oval/elliptical to C-shaped", goes beyond the study design. Only the shape of the stump outline after transection was recorded, without areas, dimensions or position measurements, and fibers were not distinguished from synovium histologically. The age of the female group under 50 years is given as 27±5.1 years with a range of 21–49 years (13 cases), which is arithmetically impossible; with this range, the smallest possible SD is about 6.6 years (p. 925). The degeneration of the posterolateral bundle with age proposed by the authors has no measurement behind it.

These data still have value. The specimens of Siebold et al., Oka et al. and Kusano et al. had median or mean ages of 68–83 years, while the young adult specimens of Odensten and Gillquist were reported as oval. Whether the C-shape description applies to young patients cannot be established from the available data. In the 3T three-dimensional MRI study of Scheffler et al. (2018), the tibial attachment was C-shaped in 19, oval in 9 and double-C in 2 of 30 patients seen for other knee problems (median age 31 years) (Results). These are shape readings on MRI, without anatomical comparison or reliability statistics.

Key points

Shape names refer to different structures: Duck-foot or oval describes the whole attachment on bone; the C-shape is the description by Siebold et al. of where the midsubstance fibers reach the tibia, which Oka et al. speculated may lie above the direct insertion; the L-shape describes the insertion area of dense fibers. The three are not mutually exclusive.

Labeling the C-shaped part as the direct insertion lacks histological support: The direct and indirect insertions of Siebold et al. (2015) are macroscopic labels only; the same group's sections showed a fibrocartilaginous transition on the bony surface of every block, and the authors speculated that the duck-foot footprint may correspond to the histological direct insertion (Oka et al., 2016).

The attachment is not confined to bone: Including articular cartilage and the lateral meniscus, the tibial attachment measures about 152 mm², of which about 103 mm² is on bone (Muro et al., 2022). Serial sections of a single specimen showed ACL fibers interlacing with the lateral meniscus anterior horn, whereas coronal sections of another 6 knees did not; the frequency of interlacing is unknown.

Whether posterolateral fibers are found depends on sampling: Siebold et al. and Oka et al. found no posterolateral fibers, whereas Quiles et al. measured a PL portion of about 61 mm²; Kusano et al. found no dense fibers posterior to the anterior horn. The disagreement arises partly from whether the fat beside the anterior root was removed and from the sectioning planes.

Mostly elderly specimens: The C-shape description comes mostly from specimens over 68 years of age, and young adult specimens in cadaveric studies were mostly oval; in an MRI study, 19 of 30 subjects with a median age of 31 years were classified as C-shaped (Scheffler et al., 2018). The age-group differences come from cross-sectional comparisons, and reading criteria differ between imaging and macroscopic observation.

5. Landmarks and Positional References on the Tibia

There are more landmarks around the tibial attachment than on the femoral side, but none has been confirmed repeatedly by both histological and imaging studies in the way the lateral intercondylar ridge has. Medially the attachment is bounded by the medial intercondylar eminence and its lateral slope; anteriorly there is an anterior ridge; laterally lies the anterior horn of the lateral meniscus; and posteriorly lie the over-the-back ridge (also called the retro-eminence ridge) and the posterior cruciate ligament. Studies have measured distances from different landmarks as origins, so the same attachment center is described by several sets of numbers.

Medial intercondylar eminence and anterior ridge

Agreement is best for the medial border. The sections of Oka et al. (2016) show the medial margin of the attachment following the ridge on the anterior part of the medial intercondylar eminence and meeting the articular cartilage of the medial plateau; Kusano et al. (2017) saw the dense fibers of both the anterior and posterior parts converge on the top of the medial intercondylar ridge; and Muro et al. (2022) found that some medial fibers attach directly to cartilage. Edwards et al. (2007) measured the AM and PL centers 5±1 mm and 4±1 mm lateral to the lateral edge of the medial intercondylar eminence, and in that sample neither distance correlated significantly with plateau width (p. 1418).

Yonetani et al. (2019) carried this relationship into young living subjects. In CT three-dimensional reconstructions of the contralateral knees of 34 patients undergoing ACL reconstruction (15–46 years), the tibial attachments of the ACL and of the lateral meniscus anterior horn were outlined using the bony landmarks established by Kusano et al. The two shared the lateral slope of the medial intercondylar ridge, which the central intercondylar ridge roughly bisected; the ACL attachment was 14.5±1.9 mm in sagittal length, 12.7 mm at its widest anteriorly and about 7–8 mm at the more posterior levels, occupying 51%–57% of the slope (Sections 3.1–3.2, Supplementary Table S2). The attachment was boot-shaped and lay against the roughly rectangular anterior horn attachment lateral to it.

The attachment was inferred from bony outlines: This is a construction from bony landmarks on CT, without verification at the fiber level; the correspondence between bony landmarks and attachment comes only from the 6 elderly embalmed knees of Kusano et al. The authors state that differences between examiners were excluded but provide no reliability data.
Text and supplementary data disagree: The four SDs and the level 2 mean of the anterior horn attachment width differ between the text and Supplementary Table S2; the overall sagittal length of 14.5 mm does not match the 14.29 mm obtained by weighting the male and female values (14.9 and 13.6 mm; 18 and 16 cases).
"Proportional to individual body size" comes mainly from the sex difference: In the pooled sample, attachment width correlated with height and weight; within men it correlated with plateau size; within women, no significant correlation with any body-size measure was found in this sample (Supplementary Tables S3, S4). The Discussion states that attachment size in men was unrelated to weight, but S3 shows a significant correlation. The statement in the Discussion that the width of the insertion "is five or six millimeters for females" is below the female means in S2 (6.4–6.8 mm), and the recommendation of double tunnels or rectangular tunnels derived from it has not been tested.
The formal erratum (PMID 31255526) states only that the author list was incorrect and corrects no data.

The anterior ridge forms the anterior border of the attachment. Oka et al. saw the anterior margin arise from a bony anterior ridge in all sections and regarded it as the classic Parsons' knob; the three-dimensional CT of Kusano et al. likewise showed an anterior bony ridge, and Quiles et al. (2018) saw this prominence in all 14 tibiae (p. 6).

Anterior root of the lateral meniscus

The anterior root of the lateral meniscus is the soft tissue landmark most closely related to the ACL. Siebold et al. (2015) placed it at the center of the C-shaped insertion; Kusano et al. described the dense ACL fibers as inserting broadly anterior to the root, sharing the slope with it at its level and disappearing posterior to it; and Muro et al. (2024) attempted to predict attachment shape from the position of the root.

Dimitriou et al. (2021a) used three-dimensional MRI to compare 90 patients with ACL rupture and 90 controls matched for sex, age and body mass index, measuring the attachment center relative to the posteromedial point of the lateral meniscus anterior root and the apex of the medial tibial spine (pp. 807–808). The attachment center lay about 12 mm anterior and about 5 mm lateral to the apex of the medial tibial spine, with no difference between groups; relative to the anterior root, the distances differed significantly between groups (for example, 9.2±3.3 and 13.3±4.3 mm mediolaterally) (Table 1). The authors concluded that both are reliable intraoperative landmarks and proposed a "safe zone" 2.6 mm medial to the anterior root (Abstract; p. 811).

"Intraoperative" reliability was not tested: These reference points are points on MRI models, without any arthroscopic observation. The near-zero bias in the Bland–Altman analysis follows from the way the corrected position was calculated and does not indicate good agreement; converted to millimeters, the limits of agreement are about ±3.4 to ±9 mm, comparable to or larger than the radius of a tunnel.
The reference point itself moves between groups: Measured in the plateau coordinate system, the attachment centers of the two groups did not differ; measured from the anterior root, they did, which indicates that the root reference point lay differently in the rupture group (about 3.6 mm more medially, by Table 1). This is not discussed in relation to the root's role as one of the "reliable intraoperative landmarks". The "safe zone" was calculated from the rupture group only; calculated from the control group, it would be about 4.7 mm.
The millimeter values and percentages for anteroposterior position in Table 1 are mutually incompatible: By the paper's definition, the normalized position is the distance from a common origin divided by plateau depth (p. 808), so the anteroposterior distance in each knee equals its percentage multiplied by its plateau depth. Under this relationship, the plateau depth of 45.3±4.6 mm (p. 809) and the normalized position of 39.7%±5.7% reported for the rupture group are compatible only with a mean anteroposterior distance of about 17.7–18.2 mm, whereas the table gives 13.5 mm; for the controls, 45.4±4.5 mm and 38.3%±4.9% correspond to about 17.2–17.6 mm, whereas the table gives 12.9 mm (p. 810, Table 1). These ranges already allow for the largest possible correlation between the two quantities, and rounding cannot close a gap of more than 4 mm; the other millimeter values and percentages in the table broadly correspond. The paper's data cannot show whether the error lies in the millimeter value, the percentage or its definition. The millimeter values and percentages in this row cannot be converted into each other, and any citation should state this.
Very probably the same cohort as the 2020 study: The two studies share the same ethics number, and in both the rupture group comprises 90 patients (55 men and 35 women, 16–45 years) with 90 matched controls (Dimitriou et al., 2020); the papers do not state how the two are related.

Over-the-back ridge and the posterior cruciate ligament

Edwards et al. (2007) compared several anteroposterior references, and dispersion was smallest with the over-the-back ridge as origin: the PL center lay 10±1 mm anterior to it, the AM center 17±2 mm and the center of the whole attachment 15±2 mm (p. 1417). Colombet et al. (2006) measured the AM center 17.5±1.9 mm from the same ridge (Table 1).

"Most reproducible" means smallest SD: Edwards et al. made no repeated measurements or comparisons among observers. "Most reproducible" means that the SD and range of distances across specimens were smallest with the over-the-back ridge as origin, which does not show that the landmark is easy to identify at arthroscopy.
Sample sizes for individual measurements are not given: The Abstract states that 55 specimens were measured. Working back from the reported correlation coefficients and P values, most correlation analyses of the attachment and bundle centers correspond to about 23 specimens (possibly the 23 intact knees), and only the analysis of attachment length corresponds to 55. This back-calculation is the present author's own; Edwards et al. do not report the sample size for each analysis.
The range of the attachment center is impossible: The attachment center was 35±5 mm (26–57 mm) from the posterior tibial margin, corresponding to 64%±5% (53%–74%) of maximum anteroposterior depth (p. 1417). Maximum plateau depth does not exceed 64 mm, so 57 mm corresponds to at least 89%, outside the authors' own percentage range.

Quiles et al. (2018) found that the over-the-back ridge was not constant in their sample and was partly seen in degenerate tibiae (p. 6); they measured a straight-line distance of 23.1 mm from the ACL center to the anterior margin of the PCL (p. 11).

Anteroposterior percentages

Percentages for the tibial attachment center likewise depend on the reference. Stäubli and Rauschning (1994) measured along the midsagittal line from the anterior tibial margin, with the sagittal diameter as denominator. The center lay at 41.2% in 10 cadaveric knees and at 43.3% in 5 cryosectioned specimens, and on MR arthrography of 35 subjects with intact ACLs at 44.1% in men and 43.7% in women; the anterior margin of the attachment lay at about 24.6%–28.3% and the posterior margin at about 56.9%–62.1% (pp. 141–143). On this basis the authors recommended placing the tunnel center at 44% of the sagittal diameter.

A single midsagittal line: The attachment lies on the sloping eminence and partly away from the midline, so its anterior and posterior limits along the midsagittal line cannot represent the whole attachment; the title refers to an "attachment area", yet no area or mediolateral data are given. The three samples were independent, with no cross-method comparison in the same knees.
The conclusion contradicts the text, and the recommended value is not a result: The first sentence of the conclusion states that the study supports placing the tunnel center in the most anterior part of the attachment (p. 144), whereas the Discussion warns that doing so causes impingement of the graft on the notch roof (p. 143). The 44.3% recommended in the Discussion appears in none of the results, and the recommended anterior margin is given as 23.4% in one place and 24.6% in another.

Colombet et al. (2006) measured along the Amis–Jakob line (parallel to the plateau, with the anterior and posterior cortices as 0 and 100%) and found the AM center at 36% and the PL center at 52% (p. 988). Edwards et al. (2007), on axial photographs of the plateau referenced to the posterior condylar line, found the attachment center at about 64% of maximum anteroposterior depth from the posterior margin (p. 1417), or about 36% measured from the front, which differs from the 41%–44% that Stäubli and Rauschning obtained on lateral views along the midsagittal line. The two differ in projection, origin and denominator, and their numbers cannot be compared directly.

Babaoğlu et al. (2026) had the largest MRI sample: 636 knees in 545 people (15–80 years). Measured parallel to the plateau with the midpoint of the tibial tubercle as the anterior reference, the attachment midpoint lay at about 40.8% of the anteroposterior reference distance (median; interquartile range about 36.7%–45.1%), with no difference between the sexes (p. 9, Table 3). The value is close to the 41%–44% of Stäubli and Rauschning, but the origin is the tibial tubercle rather than the anterior plateau margin, and the two are not equivalent.

Sex counts are knee counts: The Abstract and Results give "338 women, 298 men", which sum to 636, the number of knees; the numbers of men and women among the 545 people are not reported (pp. 1, 8).
Inconsistent definitions of the reference and the measurement point: The anterior reference is the midpoint of the tibial tubercle in the Abstract, the anterior margin of the proximal tibia in the Methods and the anterior tibial margin in the list of abbreviations (pp. 1, 5, 16); if numerator and denominator start from different points, the percentage is not a proportion of a single line segment. The attachment "midpoint" was measured on the first slice of the 3 mm images to show the attachment, a slice closer to the margin of the attachment than to its center.
Both knees treated as independent: Both knees of 91 bilateral subjects were counted separately, and the statistical tests did not account for within-person correlation (pp. 3, 7–8).
The text contradicts the table: The text and conclusion state that the coronal attachment percentage was slightly and significantly higher in right knees; Table 4, however, gives 44.03% for left and 43.82% for right knees, P=0.122; a coronal angle listed in the Discussion as differing significantly between sides has P=0.633 in Table 4 (pp. 9–10, 15).
The inference about injury lacks support: The conclusion suggests that a larger sagittal angle and shorter anteroposterior distance in women "may represent a structural substrate contributing to the higher ACL injury rates reported in females" (Abstract); the study followed up no injuries.

The 42% pooled by Parkar et al. (2017) from 10 studies (p. 7) likewise mixes results from lateral radiographs, CT, MRI and direct measurement, and its percentiles cannot serve as an individual normal range (see Section 3).

Key points

The medial border is the most consistent: The medial margin of the attachment follows the lateral slope of the medial intercondylar eminence, and some fibers attach to the cartilage of the medial plateau; CT in young adults shows the ACL and the lateral meniscus anterior horn sharing this slope (Yonetani et al., 2019), but this is a bony inference without verification at the fiber level.

The anterior root has limited precision as a reference: In an MRI study, localization relative to the anterior root had a dispersion comparable to the radius of a tunnel, and the root reference point lay differently in ruptured and control knees (Dimitriou et al., 2021a); the millimeter values and percentages for anteroposterior position in Table 1 of that study are mutually incompatible by its own definition and cannot be converted into each other.

Anteroposterior percentages change with the method: Measurement on lateral views along the midsagittal line gives 41%–44% (Stäubli and Rauschning, 1994), and axial measurement on the plateau about 36% (Edwards et al., 2007). Projection, origin and denominator should be checked before values are compared.

The over-the-back ridge gives the smallest dispersion: Distances measured from the over-the-back ridge have the smallest SD (Edwards et al., 2007), but repeated measurements between observers are lacking, and the ridge is not constant in some specimens (Quiles et al., 2018).

6. Individual Variation

Size and shape

Individual variation in attachment size is marked even within one study using one method. In the 22 knees of Edwards et al. (2008), the femoral attachment was 8–18 mm long and 6–10 mm wide (p. 32); in the 22 knees of Mochizuki et al. (2014), the whole attachment area ranged from 106.5 to 188.8 mm², and the proportion occupied by the fan-like extension from 47% to 80% (Table 1); in Iriuchishima et al. (2016), the whole attachment area ranged from 72 to 199 mm² (Table 1). Individuals differ by a factor of nearly two.

Data from young adults come mainly from imaging. Lin et al. (2023) analyzed 50 young adults with healthy knees (21–44 years, 86% men) on three-dimensional MRI: the femoral attachment was elongated oval in 33, triangular in 12 and double teardrop in 5; the elongated oval attachments measured 58±22 mm² (23–120 mm²) in area, 14±3 mm in length and 5±2 mm in width (pp. 5517–5520, Table 2). The outlines were traced on MRI by two authors who classified them by consensus, without anatomical comparison; area was measured on a single oblique plane and cannot be compared directly with cadaveric areas. The double teardrop is an outline category on imaging and does not show that these 5 subjects had two separately enveloped anatomical bundles.

Direction reversed: The Abstract and Results describe the lower center in the triangular and double teardrop groups as "higher". By the authors' own convention (68% for the "low tear" region, TT-L, and 41% for the "high tear" region, TT-H), and by Fig. 5 and the Discussion, a larger percentage means a lower position, so these two groups were in fact lower (pp. 5514, 5518, Table 3).
The description of bony indices does not match Table 1: The text states that the double teardrop group had "a larger α angle and lateral femoral condyle index"; in Table 1, the α angle in this group is 42.8°, smaller than the 53°–55° of the other two groups, and the lateral femoral condyle index does not differ among the three groups, the larger value belonging to a different index (p. 5517). The speculation in the Discussion that such individuals are more prone to ACL injury therefore has no basis.
Two regions of the same knee compared as independent groups: The upper and lower parts of the double teardrop come from the same 5 knees, yet they were entered with the other groups into a one-way ANOVA (Table 3).

In the 3T three-dimensional MRI study of Scheffler et al. (2018), the femoral attachment was band-shaped in all 30 patients seen for other knee problems, 14 mm long and 5 mm wide, with an area of 54 mm² (Results). There were no observer agreement statistics; the range of femoral attachment length is 8–19 mm in the Abstract and 10–19 mm in the Results; and the paper repeats the erroneous 13.9 mm from the text of Colombet et al. Lin et al. state that no previous study had assessed femoral attachment shape on three-dimensional MRI, yet they themselves cite this study by Scheffler et al. (pp. 5515, 5518–5519).

Dimitriou et al. (2021b) described variation of the femoral attachment center in 95 patients with ACL rupture (16–45 years, MRI within 1 month of injury): relative to the center of the lateral condyle sphere, the center lay between 1.8 mm and 12.3 mm posterior and between 7.7 mm distal and 4.8 mm proximal (Section 3.2).

Data from ruptured knees called "normative": The authors propose these values as population reference values (Section 4), yet all subjects had ACL ruptures, and the same group's 2020 study reported that the femoral attachment center in ruptured knees differs from that in intact knees (Dimitriou et al., 2020). The paper states that data from these patients had been used in another study (Section 2.1); its rupture group resembles that of the 2020 study in number, sex distribution and age range and has the same ethics number, so the two may overlap substantially.
Only ranges reported: The Methods promise means, SDs and interquartile ranges, but the Results give only minimum and maximum values; with 95 cases, the 6.5% in the Results cannot be a proportion of a whole number of cases.

Of the 8 bilateral donors in Muro et al. (2024), 4 had left and right tibial attachments classified as different shapes (Table 1). The two sides of one person can differ, and inferring attachment shape from the contralateral knee alone is unreliable.

Age

Morales-Avalos et al. (2022) reported in 81 embalmed Mexican knees that the femoral attachment was semicircular in 33/36 in the group under 50 years and flat and band-shaped in 43/45 in the group over 50 years; the posterior edge of the direct attachment lay about 2 mm from the posterior cartilage margin in the younger group and about 8 mm in the older group, P<0.0001 (pp. 3406–3409, Tables 1 and 4). The corresponding tibial study is discussed in Section 4. Both studies are cross-sectional comparisons between individuals of different ages. They can show that the two age groups differ in attachment shape, but the change with age within individuals implied by their titles would require longitudinal data. The area divisions and sex comparisons in the femoral study contain the table contradictions described in Section 1; the shape counts and the broad difference between age groups can be used. The young adult specimens of Odensten and Gillquist (1985) had an oval femoral attachment of 18 mm × 11 mm, consistent with the semicircular description in the younger group of Morales-Avalos et al. MRI data are inconsistent: Scheffler et al. (2018) classified all 30 femoral attachments in subjects with a median age of 31 years as band-shaped, while the young adults of Lin et al. (2023) were mostly elongated oval. Imaging outlines and macroscopic cadaveric observation use different reading criteria and should not be compared directly.

Sex

The available data do not reliably show whether the relative position of the attachment differs by sex.

Dimitriou et al. (2020) used three-dimensional MRI to compare 90 patients with ACL rupture and 90 matched controls (55 men and 35 women in each group). They concluded that attachment position did not differ by sex and should not be regarded as a reason for the higher rate of ACL rupture in women (Abstract).

The sex comparison of femoral anteroposterior position in intact knees cannot be verified: In intact knees, the anteroposterior position of the femoral attachment was 23.5%±5.0% in women and 26.3%±4.1% in men (35 and 55 cases), and the paper reports P=0.30 (pp. 4–5, Table 1). The paper states only that two-way ANOVA was used, without stating whether each P value refers to a main effect, an interaction or a within-group comparison, or how the error term and multiple comparisons were handled. An approximate check of the tabulated data gives P below 0.01 whether the two groups are compared independently or within groups using the pooled error of all four groups; the tibial P values in the same table do agree with such a check. The source of P=0.30 cannot be verified from the paper, and the conclusion that "no sex-specific differences exist" lacks a verifiable basis for femoral anteroposterior position.
The direction of the comparison between rupture and intact groups is self-contradictory: The proximal–distal values of the femoral attachment are smaller in ruptured knees (14.2%–14.6%) than in intact knees (18.6%–19.7%), and all values are reported as "proximal to the FEA" (flexion–extension axis), yet the Abstract states that ruptured knees are "more proximal" (pp. 1, 4).
The controls are not healthy knees: Controls were patients with a noncontact knee injury whose MRI showed no ACL rupture, and the attachment in the rupture group was traced on a torn ligament, so the two groups cannot be pooled as a normal reference. The authors infer that the more posterior femoral attachment in the rupture group is a predisposing factor for injury (Abstract); the comparison was made after injury and cannot establish which came first.

In the tibial study of Zhang et al. (2023), the sex difference in millimeters was about 2 mm in the Chinese group and about 3.1 mm in the white group (Abstract); after normalization to plateau size, the differences shrank to about 1–2 percentage points, and the normalized difference in the white group was not statistically significant (Tables 2 and 3). In Yonetani et al. (2019), tibial attachment width correlated with plateau size in men, whereas in women no correlation with body-size measures was found in this sample (see Section 5). Several studies agree that absolute attachment dimensions and distances vary with body size; whether relative position differs by sex cannot be concluded from the current data.

Region and ethnicity

Two studies from the same group compared Chinese and white subjects. Zhang et al. (2023) reported that both the anteroposterior and the mediolateral distance of the tibial attachment center from an anteromedial plateau origin were smaller in the Chinese group (17.1±2.3 vs 20.0±3.4 mm; 34.2±2.4 vs 37.4±3.6 mm), differing by about 3.8 and 2.7 percentage points after normalization (Table 1). Zhang et al. (2024) reported that the femoral attachment center in the Chinese group was about 1.5 mm more posterior and about 1.1 mm more proximal than in the white group (Abstract).

Ethnicity is confounded with scanning site, equipment and year: Neither study defines how ethnicity was determined. The 2024 study states that the MRI scans of the white group came from a Swiss collaborating center in 2015–2017, with Philips equipment and 1 mm slice sequences, and those of the Chinese group from Guangzhou in 2019–2022, with Siemens equipment and 0.5 mm slice sequences (p. 217); the 2023 study, however, states that both groups were health-check subjects examined in 2019–2022, all scanned on a Siemens Skyra (p. 2). If the white groups of the two studies are the same, one of these descriptions must be wrong; on the 2024 description, the so-called ethnic difference cannot be separated from differences in site, equipment and sequence.
Inconsistent figures: The Abstract of the 2023 study gives the highest probability of injury to the lateral meniscus anterior root as 30%, and the text and conclusion as 70%; two P values in the text belong to other rows of the table (pp. 1, 6–7, 9).

The meta-analysis by Wang et al. (2025) included 50 studies with 1652 knees, divided by region of publication into Asian and Western groups of 25 studies each. It reported the femoral attachment center at 35.2% from the posterior margin of the lateral condyle in the Asian group and 27.3% in the Western group, and at 39.4% and 33.0% from Blumensaat's line; the femoral attachment area was smaller in the Asian group (96.3 vs 126.8 mm²), and tibial attachment position and area did not differ (Abstract; p. 5, Table 2). On this basis the authors recommend that, in reconstruction for Asian patients, the femoral tunnel be placed about 3.5 mm shallower and 1.4 mm lower than in Western patients (p. 7).

Regional grouping is not ethnicity: Studies were grouped by study location and specimen source (p. 3); the Western group also included Australian studies and the Mexican specimens of Morales-Avalos et al. (Table 1).
Different measurements were pooled: In the area analysis, the Asian group combined the whole attachment including the fan-like extension (125 mm² in Suruga et al.), the CT depression containing the histological direct insertion (128.3 mm² in Iwahashi et al.) and MRI tracings on a single oblique plane (Lin et al.), while the Western group included a three-dimensional surface area that included the surface membrane (196.8 mm² in Ferretti et al.). The position analysis combined lateral radiographs, photographs, three-dimensional CT and MRI with different grid placements, and it included Edwards et al. (2008), even though they state explicitly that their grid differs from the quadrant method. As shown in Section 3, grid placement alone can shift the group mean center in the same knees by about 6 to 16 percentage points. In the femoral position analyses, I² was 91.0%–98.4% in each group (p. 5, Fig. 3), and the authors attributed this heterogeneity to natural variation between individuals (p. 8). Individual variation is already reflected in the SD of each study; I² expresses differences between studies, which may arise from the structure measured and the reference frame or from real population differences, and this meta-analysis did not distinguish between them.
Data in the forest plots do not match the original studies: In the area analysis in Fig. 5A, Suruga et al. are entered as 30 knees and 125±47 mm², whereas the original study had 23 knees and 125 mm² is the whole attachment including the fan-like extension; Morales-Avalos et al. are entered as 88 knees and 136±22 mm², but the original study analyzed 81 knees, and 136 mm² is exactly the unweighted mean of the two subgroups under 50 years; the 142.2 mm² of Mochizuki et al. is assigned to 6 knees, although the original study quantified 22 knees and used 6 for dynamic observation; Lin et al. are entered as 50 knees and 55±10.1 mm², against original subgroup values of 58±22, 47±18 and 68±30 mm². In Fig. 5B, Muro et al. are entered as 11 knees and 143.5±27.5 mm²; the original study quantified 10 knees, 143.5 mm² is the sum of the bony 102.6 mm² and the cartilaginous 40.9 mm², and 27.5 is the SD of the bony component alone (p. 6, Fig. 5). In the position analyses, the whole-ligament femoral and tibial centers of Colombet et al. (Figs. 3 and 4) were obtained by averaging two bundle centers, and the tibial values of Edwards et al. (2007) were likewise averages of two bundles, although Edwards et al. reported the center of the whole attachment directly.
Other reporting problems: In the flow diagram, 46890 records from PubMed and 6620 from Web of Science became 6878 after a duplicate-removal step (Fig. 1). If both numbers are counts of records without internal duplicates within each database, combining the two databases and removing only duplicates could not leave fewer than 46890 records. The diagram does not state whether the initial numbers are search hits or deduplicated records, or whether several searches were combined, so the process shown cannot explain this large reduction; whether the cause is duplication within databases, a typographical error or an unreported screening step cannot be determined from the paper. The Discussion gives the Asian femoral deep–shallow coordinate as 32.5%, which does not match the 35.2% in the Abstract and Table 2 (p. 5). Missing SDs in primary studies were imputed from "similar studies", and the text does not state which studies received imputed values or how they were chosen (p. 4).
Opposite in direction to a primary study: The contemporaneous comparison by Zhang et al. (2024) found the femoral attachment more posterior and more proximal in the Chinese group, the opposite direction to this meta-analysis.

The available data do not support recommendations to adjust femoral or tibial tunnel position by ethnicity.

Anomalous attachments

Norman et al. (2025) reported a 16-year-old girl in whom arthroscopy found no ACL tissue at the normal tibial attachment, with a thin ligament-like band running from the lateral wall of the notch to the anterior horn of the lateral meniscus. Preoperative MRI had been read as an ACL tear; the authors performed meniscal repair and quadriceps tendon reconstruction, and the patient was asymptomatic 8 months after surgery (Case Report). The authors could not determine whether this was a congenital anomaly or abnormal healing after an old injury. The paper describes the patient as "without obvious skeletal deformities", yet arthroscopy showed marked hypoplasia of the tibial intercondylar eminence, which was not assessed on imaging. A case report can show only that such an attachment exists; it cannot estimate its frequency.

Of the 19 specimens of Girgis et al. (1975), 2 had additional ACL fibers attaching to the anterior horn of the medial meniscus (footnote to Table 2), and a few of the 10 knees of Muro et al. (2022) had attachment to the transverse ligament (Table 1). These frequencies come from small samples and show only that the attachment is not always confined to bone and the lateral meniscus.

Key points

Individual variation approaches a factor of two: Within single studies, femoral attachment length, area and fan-like extension proportion can differ by a factor of nearly two; a standard position expressed as a mean cannot represent the individual.

Age-group differences come from cross-sectional comparisons: In cadaveric studies, elderly specimens mostly show a band-shaped femoral attachment and a C-shaped tibial attachment, and young adult specimens mostly a semicircular or oval shape; MRI results are inconsistent. No longitudinal data show whether these differences represent change with age within individuals.

Sex differences are unresolved: Absolute dimensions vary with body size; in the sex comparison of relative position, the P value for femoral anteroposterior position in Dimitriou et al. (2020) cannot be verified from their tabulated data.

Evidence for ethnic comparisons is insufficient: Comparisons of Chinese and white subjects are confounded with scanning site, equipment and year; the meta-analysis grouped by region of publication pooled different structures and reference frames, and its conclusion runs opposite to a primary study, so it should not be used to adjust tunnel position.

7. Interpreting Data on the Normal Attachment and Its Location

On the femur, histological studies show the dense midsubstance fibers anchored by direct insertion in the depression behind the lateral intercondylar ridge; two studies of elderly embalmed specimens measured its width at about 8 mm and about 5 mm, and the CT depression is about 17 mm long. The surrounding looser fibers attach indirectly, spreading mainly posteriorly and distally to the articular cartilage margin, and a few cross the ridge. The lateral intercondylar ridge was visible in most specimens across studies and marks the anterior limit of the direct insertion; CT of elderly fresh-frozen femora failed to detect it in 3/20, and macroscopic observation included a case in which it did not run the full length of the attachment. The two histological studies disagree on the position of the posterior edge of the direct insertion. Cutting the connection between the posterior fan-like extension and the midsubstance fibers lowers failure load ex vivo; because there was no sham-cut control, there are still no data on how much tension the posterior fibers carry under physiological load.

On the tibia, sectional and dissection studies show the anterior margin of the attachment arising from the anterior ridge; medially it follows the lateral slope of the medial intercondylar eminence, some fibers attach to the articular cartilage of the medial plateau, and laterally it borders the anterior horn of the lateral meniscus. In the 6 knees of Kusano et al., there was only loose connective tissue posterior to the anterior horn, and in the sections of Oka et al., synovium, fat and vessels lay behind the lateral half of the attachment. The macroscopic whole attachment is mostly duck-foot or oval, wide anteriorly and narrow posteriorly; the C-shape is the description by Siebold et al. of where the midsubstance fibers reach the bone, and Oka et al. speculated that it may lie above the direct insertion. Interlacing of ACL fibers with the lateral meniscus anterior horn has appeared in serial sections of at least 1 knee, with unknown frequency; whether fibers attach in the posterolateral area depends on whether the fat beside the anterior root was removed and on the sectioning plane.

Numbers are easier to misread than spatial relationships. An attachment area, width or center coordinate reported by a study is meaningful only for its own measured structure, reference frame and definition of center. Quadrant percentages change with grid borders, origin and inclusion of the fan-like extension; tibial anteroposterior percentages change with projection and denominator; and a whole-ligament center obtained by averaging two bundle centers differs from a center computed from a tracing of the whole attachment. Judgments on postoperative imaging that tunnel placement is "in or out of the anatomic range" cannot be based on percentiles calculated from study means in a review.

The populations studied are also skewed. Most anatomical studies of the femoral and tibial attachments used embalmed specimens over 60 years of age, often over 80. Data from young adults come mainly from the 20 knees of accident victims in Odensten and Gillquist (1985) and from several CT and MRI studies, in which attachment outlines were inferred from images without anatomical or histological verification in the same knees. The band-shaped femoral attachment and C-shaped tibial attachment common in elderly specimens should be applied to young patients with caution.

Anatomical studies often propose a tunnel position, a tunnel shape or a choice between single-bundle and double-bundle reconstruction in their Discussion. These recommendations come from morphological reasoning; which tunnel position improves stability and clinical outcome is a question for the biomechanical and clinical studies discussed in the chapters on reconstruction technique.

Key points

Spatial relationships are reliable; numbers depend on method: The femoral direct insertion lies immediately behind the lateral intercondylar ridge, and the dense tibial fibers lie anteromedially, bounded laterally by the anterior horn of the lateral meniscus; attachment dimensions and center coordinates should be cited together with the structure measured, the reference frame and the definition of center.

Study means are not individual standards: Review percentiles and the means and SDs of single studies cannot be used directly to judge whether an individual knee is normal or whether a tunnel is acceptable.

Data come mainly from elderly specimens: Attachment data for young adults come mostly from imaging inference without anatomical verification.

Anatomy is not a surgical target: Tunnel recommendations in anatomical studies are inferences; their effect on outcome must be tested in reconstruction studies.

References

[1]
Babaoğlu E, Bamaç B, Memişoğlu K. Morphometric Evaluation of Anterior Cruciate Ligament Orientation and Tibial Footprint Location Using Magnetic Resonance Imaging. Diagnostics (Basel). 2026;16(5):748. PMID 41828024; DOI 10.3390/diagnostics16050748.
[2]
Bernard M, Hertel P, Hornung H, Cierpinski T. Femoral insertion of the ACL. Radiographic quadrant method. Am J Knee Surg. 1997;10(1):14-21; discussion 21-2. PMID 9051173. Values in the text are cited from Parkar et al. (2017).
[3]
Bhattacharyya R, Ker A, Fogg Q, Spencer SJ, Joseph J. Lateral Intercondylar Ridge: Is it a reliable landmark for femoral ACL insertion?: An anatomical study. Int J Surg. 2018;50:55-59. PMID 29253556; DOI 10.1016/j.ijsu.2017.12.013.
[4]
Colombet P, Robinson J, Christel P, Franceschi JP, Djian P, Bellier G, Sbihi A. Morphology of anterior cruciate ligament attachments for anatomic reconstruction: a cadaveric dissection and radiographic study. Arthroscopy. 2006;22(9):984-992. PMID 16952729; DOI 10.1016/j.arthro.2006.04.102.
[5]
Dimitriou D, Wang Z, Zou D, Helmy N, Tsai TY. Do Sex-Specific Differences Exist in ACL Attachment Location? An MRI-Based 3-Dimensional Topographic Analysis. Orthop J Sports Med. 2020;8(11):2325967120964477. PMID 33294470; DOI 10.1177/2325967120964477.
[6]
Dimitriou D, Zou D, Wang Z, Tsai TY, Helmy N. Anterior root of lateral meniscus and medial tibial spine are reliable intraoperative landmarks for the tibial footprint of anterior cruciate ligament. Knee Surg Sports Traumatol Arthrosc. 2021;29(3):806-813. PMID 32419045; DOI 10.1007/s00167-020-06018-0. Cited in the text as Dimitriou et al. (2021a).
[7]
Dimitriou D, Cheng R, Yang Y, Baumgaertner B, Helmy N, Tsai TY. High variability in anterior cruciate ligament femoral footprint: Implications for anatomical anterior cruciate ligament reconstruction. Knee. 2021;30:141-147. PMID 33930701; DOI 10.1016/j.knee.2021.01.015. Cited in the text as Dimitriou et al. (2021b).
[8]
Edwards A, Bull AM, Amis AA. The attachments of the anteromedial and posterolateral fibre bundles of the anterior cruciate ligament: Part 1: tibial attachment. Knee Surg Sports Traumatol Arthrosc. 2007;15(12):1414-1421. PMID 17934717; DOI 10.1007/s00167-007-0417-6.
[9]
Edwards A, Bull AM, Amis AA. The attachments of the anteromedial and posterolateral fibre bundles of the anterior cruciate ligament. Part 2: femoral attachment. Knee Surg Sports Traumatol Arthrosc. 2008;16(1):29-36. PMID 17957351; DOI 10.1007/s00167-007-0410-0.
[10]
Ferretti M, Ekdahl M, Shen W, Fu FH. Osseous landmarks of the femoral attachment of the anterior cruciate ligament: an anatomic study. Arthroscopy. 2007;23(11):1218-1225. PMID 17986410; DOI 10.1016/j.arthro.2007.09.008.
[11]
Girgis FG, Marshall JL, Monajem A. The cruciate ligaments of the knee joint. Anatomical, functional and experimental analysis. Clin Orthop Relat Res. 1975;(106):216-231. PMID 1126079; DOI 10.1097/00003086-197501000-00033.
[12]
Gulan L, Kovačević M, Žauhar G, Jurdana H, Gulan G. The forgotten third: understanding the ACL femoral insertion and lateral intercondylar ridge extension onto the notch roof. Surg Radiol Anat. 2025;47(1):118. PMID 40195153; DOI 10.1007/s00276-025-03628-6.
[13]
Iriuchishima T, Ryu K, Aizawa S, Fu FH. The difference in centre position in the ACL femoral footprint inclusive and exclusive of the fan-like extension fibres. Knee Surg Sports Traumatol Arthrosc. 2016;24(1):254-259. PMID 25308156; DOI 10.1007/s00167-014-3373-y.
[14]
Iriuchishima T, Suruga M, Yahagi Y, Iwama G, Aizawa S, Fu FH. The location of the femoral ACL footprint center is different depending on the Blumensaat's line morphology. Knee Surg Sports Traumatol Arthrosc. 2020;28(8):2453-2457. PMID 32514842; DOI 10.1007/s00167-020-06087-1.
[15]
Iriuchishima T, Goto B. Systematic Review of Cadaveric Studies on Anterior Cruciate Ligament Anatomy Focusing on the Mid-substance Insertion and Fan-like Extension Fibers. Indian J Orthop. 2022;56(9):1525-1532. PMID 36052387; DOI 10.1007/s43465-022-00695-4.
[16]
Iwahashi T, Shino K, Nakata K, Otsubo H, Suzuki T, Amano H, Nakamura N. Direct anterior cruciate ligament insertion to the femur assessed by histology and 3-dimensional volume-rendered computed tomography. Arthroscopy. 2010;26(9 Suppl):S13-S20. PMID 20667684; DOI 10.1016/j.arthro.2010.01.023.
[17]
Kanto R, Smolinski MP, Fayed AM, Linde MA, Musahl V, Fu FH, Smolinski P. Posterior fan-like extension of the anterior cruciate ligament femoral insertion increases ligament failure strength. J ISAKOS. 2025;14:100939. PMID 40783135; DOI 10.1016/j.jisako.2025.100939.
[18]
Kusano M, Yonetani Y, Mae T, Nakata K, Yoshikawa H, Shino K. Tibial insertions of the anterior cruciate ligament and the anterior horn of the lateral meniscus: A histological and computed tomographic study. Knee. 2017;24(4):782-791. PMID 28559005; DOI 10.1016/j.knee.2017.04.014.
[19]
Lin J, Zhang S, Xin E, Liang M, Yang L, Chen J. Anterior cruciate ligament femoral footprint is oblong-ovate, triangular, or two-tears shaped in healthy young adults: three-dimensional MRI analysis. Knee Surg Sports Traumatol Arthrosc. 2023;31(12):5514-5523. PMID 37828405; DOI 10.1007/s00167-023-07606-6.
[20]
Mittendorfer JM, Duezguen Z, Horak L, Gahleitner A, Mandler EM, Hirtler L. Consistency of Subcortical Osseous Structures in Femoral Cruciate Ligament Attachment Sites: A Combined 3D-CT and Histological Anatomical Study. J Pers Med. 2026;16(7):352. PMID 42506079; DOI 10.3390/jpm16070352.
[21]
Mochizuki T, Muneta T, Nagase T, Shirasawa S, Akita KI, Sekiya I. Cadaveric knee observation study for describing anatomic femoral tunnel placement for two-bundle anterior cruciate ligament reconstruction. Arthroscopy. 2006;22(4):356-361. PMID 16581446; DOI 10.1016/j.arthro.2005.09.020.
[22]
Mochizuki T, Fujishiro H, Nimura A, Mahakkanukrauh P, Yasuda K, Muneta T, Akita K. Anatomic and histologic analysis of the mid-substance and fan-like extension fibres of the anterior cruciate ligament during knee motion, with special reference to the femoral attachment. Knee Surg Sports Traumatol Arthrosc. 2014;22(2):336-344. PMID 23344119; DOI 10.1007/s00167-013-2404-4.
[23]
Morales-Avalos R, Castillo-Escobedo TA, Elizondo-Omaña RE, Del Carmen Theriot-Giron M, Perelli S, Guzmán-López S, Peña-Martínez VM, Vílchez-Cavazos F, Monllau JC. The morphology of the tibial footprint of the anterior cruciate ligament changes with ageing from oval/elliptical to C-shaped. Knee Surg Sports Traumatol Arthrosc. 2021;29(3):922-930. PMID 32385559; DOI 10.1007/s00167-020-06049-7.
[24]
Morales-Avalos R, Perelli S, Vilchez-Cavazos F, Castillo-Escobedo T, Peña-Martínez VM, Elizondo-Omaña R, Guzmán-López S, Padilla-Medina JR, Monllau JC. The morphology of the femoral footprint of the anterior cruciate ligament changes with aging from a large semicircular shape to a small flat ribbon-like shape. Knee Surg Sports Traumatol Arthrosc. 2022;30(10):3402-3413. PMID 35318508; DOI 10.1007/s00167-022-06935-2.
[25]
Moulton SG, Steineman BD, Haut Donahue TL, Fontboté CA, Cram TR, LaPrade RF. Direct versus indirect ACL femoral attachment fibres and their implications on ACL graft placement. Knee Surg Sports Traumatol Arthrosc. 2017;25(1):165-171. PMID 27295056; DOI 10.1007/s00167-016-4188-9.
[26]
Muro S, Kim J, Tsukada S, Akita K. Significance of the broad non-bony attachments of the anterior cruciate ligament on the tibial side. Sci Rep. 2022;12(1):6844. PMID 35477722; DOI 10.1038/s41598-022-10806-8.
[27]
Muro S, Kim J, Nimura A, Tsukada S, Akita K. Morphometric Analysis of the Tibial Attachment Shape of the Anterior Cruciate Ligament and Its Relationship With the Location of the Anterior Horn of the Lateral Meniscus. Am J Sports Med. 2024;52(3):682-690. PMID 38284162; DOI 10.1177/03635465231219978.
[28]
Nam VT, Năng VSQ, Hiếu PT, Minh HN, Quỳnh PB, Dũng TT, Toàn DĐ. Location of the Anatomic Footprint Centers of the Anterior Cruciate Ligament Determined by Quadrant Method on Three-Dimensional Magnetic Resonance Imaging. Indian J Orthop. 2024;58(11):1650-1656. PMID 39539322; DOI 10.1007/s43465-024-01234-z.
[29]
Norman M, Zehner K, Gardner E. Aberrant Insertion of the Anterior Cruciate Ligament on the Lateral Meniscus: A Case Report. JBJS Case Connect. 2025;15(1):e24.00395. PMID 39854511; DOI 10.2106/JBJS.CC.24.00395.
[30]
Odensten M, Gillquist J. Functional anatomy of the anterior cruciate ligament and a rationale for reconstruction. J Bone Joint Surg Am. 1985;67(2):257-262. PMID 3968118; DOI 10.2106/00004623-198567020-00012.
[31]
Oka S, Schuhmacher P, Brehmer A, Traut U, Kirsch J, Siebold R. Histological analysis of the tibial anterior cruciate ligament insertion. Knee Surg Sports Traumatol Arthrosc. 2016;24(3):747-753. PMID 26685684; DOI 10.1007/s00167-015-3924-x.
[32]
Parkar AP, Adriaensen MEAPM, Vindfeld S, Solheim E. The Anatomic Centers of the Femoral and Tibial Insertions of the Anterior Cruciate Ligament: A Systematic Review of Imaging and Cadaveric Studies Reporting Normal Center Locations. Am J Sports Med. 2017;45(9):2180-2188. PMID 27899355; DOI 10.1177/0363546516673984. Page numbers refer to the authors' archived manuscript.
[33]
Piefer JW, Pflugner TR, Hwang MD, Lubowitz JH. Anterior cruciate ligament femoral footprint anatomy: systematic review of the 21st century literature. Arthroscopy. 2012;28(6):872-881. PMID 22301358; DOI 10.1016/j.arthro.2011.11.026.
[34]
Quiles C, Constantino JA, Gañán Y, Macías D, Quiles M. Stereophotogrammetric surface anatomy of the anterior cruciate ligament's tibial footprint: Precise osseous structure and distances to arthroscopically-relevant landmarks. Knee. 2018;25(4):531-544. PMID 29705076; DOI 10.1016/j.knee.2018.03.016. Page numbers refer to the Article in Press version.
[35]
Sabzevari S, Shaikh H, Marshall B, Zhu J, Linde MA, Smolinski P, Fu FH. The femoral posterior fan-like extension of the ACL insertion increases the failure load. Knee Surg Sports Traumatol Arthrosc. 2020;28(4):1113-1118. PMID 31642946; DOI 10.1007/s00167-019-05753-3.
[36]
Sasaki N, Ishibashi Y, Tsuda E, Yamamoto Y, Maeda S, Mizukami H, Toh S, Yagihashi S, Tonosaki Y. The femoral insertion of the anterior cruciate ligament: discrepancy between macroscopic and histological observations. Arthroscopy. 2012;28(8):1135-1146. PMID 22440794; DOI 10.1016/j.arthro.2011.12.021.
[37]
Scheffler SU, Maschewski K, Becker R, Asbach P. In-vivo three-dimensional MR imaging of the intact anterior cruciate ligament shows a variable insertion pattern of the femoral and tibial footprints. Knee Surg Sports Traumatol Arthrosc. 2018;26(12):3667-3672. PMID 29728742; DOI 10.1007/s00167-018-4939-x.
[38]
Siebold R, Schuhmacher P, Fernandez F, Śmigielski R, Fink C, Brehmer A, Kirsch J. Flat midsubstance of the anterior cruciate ligament with tibial "C"-shaped insertion site. Knee Surg Sports Traumatol Arthrosc. 2015;23(11):3136-3142. PMID 24841941; DOI 10.1007/s00167-014-3058-6. See the erratum in the next entry.
[39]
Siebold R, Schuhmacher P, Fernandez F, Śmigielski R, Fink C, Brehmer A, Kirsch J. Erratum to: Flat midsubstance of the anterior cruciate ligament with tibial "C"-shaped insertion site. Knee Surg Sports Traumatol Arthrosc. 2016;24(9):3046. PMID 25149644; DOI 10.1007/s00167-014-3230-z.
[40]
Spalding T, Robinson J, Getgood A. Concerns over interpretation of data in "anterior cruciate ligament femoral footprint anatomy: systematic review of the 21st century literature". Arthroscopy. 2014;30(5):538-539. PMID 24725311; DOI 10.1016/j.arthro.2014.02.004.
[41]
Stäubli HU, Rauschning W. Tibial attachment area of the anterior cruciate ligament in the extended knee position. Anatomy and cryosections in vitro complemented by magnetic resonance arthrography in vivo. Knee Surg Sports Traumatol Arthrosc. 1994;2(3):138-146. PMID 7584195; DOI 10.1007/BF01467915.
[42]
Suruga M, Horaguchi T, Iriuchishima T, Yahagi Y, Iwama G, Tokuhashi Y, Aizawa S. Morphological size evaluation of the mid-substance insertion areas and the fan-like extension fibers in the femoral ACL footprint. Arch Orthop Trauma Surg. 2017;137(8):1107-1113. PMID 28593582; DOI 10.1007/s00402-017-2726-7.
[43]
Takahashi M, Doi M, Abe M, Suzuki D, Nagano A. Anatomical study of the femoral and tibial insertions of the anteromedial and posterolateral bundles of human anterior cruciate ligament. Am J Sports Med. 2006;34(5):787-792. PMID 16452272; DOI 10.1177/0363546505282625. The text refers only to the entry of its values in Table 2 of Parkar et al. (2017).
[44]
Wang C, Mok YR, Wang M, Ma X, Hui JHP, Lee DYH. Comparative Meta-Analysis of ACL Footprint Anatomy Between Asian and Western Populations. Orthop J Sports Med. 2025;13(12):23259671251397407. PMID 41426266; DOI 10.1177/23259671251397407.
[45]
Yahagi Y, Iriuchishima T, Horaguchi T, Suruga M, Tokuhashi Y, Aizawa S. The importance of Blumensaat's line morphology for accurate femoral ACL footprint evaluation using the quadrant method. Knee Surg Sports Traumatol Arthrosc. 2018;26(2):455-461. PMID 28283721; DOI 10.1007/s00167-017-4501-2.
[46]
Yonetani Y, Kusano M, Tsujii A, Kinugasa K, Hamada M, Shino K. Tibial insertion of the anterior cruciate ligament and anterior horn of the lateral meniscus share the lateral slope of the medial intercondylar ridge: A computed tomography study in a young, healthy population. Knee. 2019;26(3):612-618. PMID 31078391; DOI 10.1016/j.knee.2019.04.009. See the erratum in the next entry.
[47]
Yonetani Y, Kusano M, Tsujii A, Kinugasa K, Hamada M, Shino K. Erratum to "Tibial insertion of the anterior cruciate ligament and anterior horn of the lateral meniscus share the lateral slope of the medial intercondylar ridge: A computed tomography study in a young, healthy population" [Knee 26 (2019) 612-618]. Knee. 2019;26(4):951. PMID 31255526; DOI 10.1016/j.knee.2019.06.006.
[48]
Zhang L, Li C, Zhang J, Zou D, Dimitriou D, Xing X, Tsai TY, Li P. Significant race and gender differences in anterior cruciate ligament tibial footprint location: a 3D-based analysis. J Orthop Traumatol. 2023;24(1):33. PMID 37389687; DOI 10.1186/s10195-023-00710-w.
[49]
Zhang L, Huang T, Li C, Xing X, Zou D, Dimitriou D, Tsai TY, Li P. Race and Gender Differences in Anterior Cruciate Ligament Femoral Footprint Location and Orientation: A 3D-MRI Study. Orthop Surg. 2024;16(1):216-226. PMID 37953405; DOI 10.1111/os.13918.